Nitrogen Isotopic Fractionation and l80 Exchange in Relation to the Mechanism of Denitrification of Nitrite by Pseudomonas stutzeri*

Two types of mechanisms for the enzymatic reduc- tion of NO; to N2O have been proposed. In one, two NO; ions are reduced in parallel, with the nitrogen- nitrogen bond formed from reduced intermediates. In the second, the two NO; ions enter the reaction sequen- tially, with the nitrogen of at least one of the two having a valence of 3+ when the nitrogen-nitrogen bond is formed. Our objective was to distinguish between these two types of mechanism. Toward that end, the exchange of ‘*O from H20 to NO; and the overall nitrogen isotopic fractionation factor (Bob.) were meas- ured. The rate of exchange of oxygen from H20 to NO;, resulting from a protonation-dehydration step preceding reductive events in both mechanisms, was less than 10% of the rate of denitrification at both low and high [NO;]. The value of Bobs was 1.010 f 0.001 and 1.020 f 0.001 at low and high [NO;], respectively. Expressions for Bob., as a function of the measured rate of entry of oxygen from Ha0 into NO;, were derived for both types of mechanism. The measured depend- ence of on substrate concentration, as constrained by the “0 exchange data, is inconsistent with the first type of mechanism, but consistent with the second type. Thus, by combining nitrogen isotopic fractionation and “0 exchange data, we rule out any mechanism in Pseu- domonas stutzeri in which NO; ions are reduced in parallel, with the nitrogen-nitrogen bond being formed from reduced intermediates.

NzO production. In one version, NO is an enzyme-bound intermediate (1); and in another, NO is a free intermediate (2,3). The first version was proposed by Garber and Hollocher (1) and is shown diagrammatically as Mechanism I in Fig. 1. In this proposed mechanism, NO; is activated by protonationdehydration to form E .NO+, and the nitrosyl is reduced by two electrons to E .NO-. Nitroxyl is then released from the enzyme and reacts with another free nitroxyl and hydrogen ions to form NzO and HzO nonenzymatically.
In the second type of mechanism, nitrogen-nitrogen bond formation and reduction occur in a linear, rather than parallel, path from NO; to NzO. At least one of the nitrogen atoms of NOC has a 3+ valence when the nitrogen-nitrogen bond is made. An example is the mechanism proposed by Averill and Tiedje (4) (shown as Mechanism I1 in Fig. 2). In this mechanism, as in Mechanism I, the enzyme-bound NO; is dehydrated to form E . NO+. At this point, the mechanisms diverge. In Mechanism 11, a second NO1 makes a nucleophilic attack on E . NO+, resulting in the formation of E .Nz03, which is ultimately reduced to NzO. Whereas Averill and co-workers (5) have subsequently modified their original proposal (see Averill and Tiedje (4)), to take into account more recent experimental results (6, 7),' the essential feature (from the point of view of our paper) remains, namely, two NO; ions are added to the reaction sequentially, rather than in parallel, with at least one nitrogen atom having a 3+ valence. In this mechanism, the nitrogen atoms of both NO; ions have a 3+ valence when the nitrogen-nitrogen bond is made.
The first reductive step in both mechanisms (the step with rate constant 126 in Mechanism I and 127 in Mechanism 11) and steps beyond are represented as unidirectional. The consequences, for the interpretation of the data reported here, of the assumption of unidirectionality of these steps being partially or wholly incorrect is addressed under "Discussion." The reversible dehydration step(s) (see, for example, the second step of both mechanisms in Figs. 1 and 2) allows for enzymatically catalyzed exchange of "0 between water and an intermediate of the reaction (e.g. E.NO+), leading to the possibility of incorporation into product, NzO, and substrate, NO;. Such "0 incorporation into NzO by three species of denitrifying microorganisms and into NO; by two species Garber and Hollocher (6) showed positional equivalence of nitrogen in 14,15N20 produced by concomitant reduction of I6NO; and "NO. This was interpreted as favoring NO-as precursor of N20. However, Aerssens et al. (5) pointed out that cis-hyponitrite (an intermediate not explicitly included in the originally proposed mechanism of Averill and Tiedje (4)) is an equally plausible intermediate if the nitrogen atom attached to the metal center of the enzyme exchanges rapidly with the other nitrogen atom, as they consider likely based on an analogous system (8). In the original proposal of Averill and Tiedje (4), the step from E.NO+ + NO; + E.N203 was irreversible. The nitrogen isotopic fractionation data of Bryan et al. (7) require that such a step, if it exists, be reversible (see "Discussion"). In the latest published version of their mechanism, Aerssens et al. (5) show this step as reversible. intermediates; and R and R, represent reduced and oxidized forms of the proximal reductant, a cytochrome, at the end of an electron transport chain, respectively. Two electrons must be added to E. Nz03 to transform it to NZOf, and two electrons must also be added to E . N20%-to produce NzO, whereas R represents the reduced form of a cytochrome, a one-electron carrier. This representation is not meant to imply coupling. Rather, each of these two "stepsn (steps 7 and 9) subsumes two one-electron reductions. In the mechanism originally proposed by Averill and Tiedje (4), step 5 was shown as irreversible. In a later paper, Averill and co-workers (5) show this step as reversible.
was experimentally demonstrated by Garber and Hollocher (1). Aerssens et al. ( 5 ) showed that the extent of enzymatic exchange of "0 from HzO to N20 decreased with increasing NO; concentration. The investigators interpreted this finding as favoring Mechanism I1 over I. On the other hand, Garber et al. (9) observed isotopic equilibrium between "N and 14N in N 2 0 produced in the presence of denitrifying bacteria and equimolar concentrations of "NO; plus unlabeled N20:-, one of the intermediates of Mechanism 11. This was interpreted as showing that NzO$ cannot be an intermediate of denitrification. We regard this interpretation as inconclusive because of the high background of nonenzymatic evolution of N20 in cell-free controls which resulted from decomposition of NzO%-. In addition, their data are not completely internally consistent? If the finding of Garber et al. (9) and its interpre-For example, the 16N abundance of NzO produced when P. stutzeri was incubated with "NO; and "N20%-was lower than would be expected from the amount of NzO produced by biological denitrification (as opposed to the amount produced by chemical decomposition of NzOf-). This requires a higher rate of NzO? decomposition in the presence of P. stutzeri than in its absence. However, when P. stutzeri was incubated with NzO%-(no NO; present), less N20 was formed than in controls without cells. Thus, decomposition of NZO%does not seem to be enhanced by the presence of the bacteria. Despite this and other inconsistencies, the data of Garber et al. (9) cast enough doubt on Nz0;-as an intermediate in the denitrification of NO; that we consider the burden now to rest on those who propose enzyme-bound N20f as an intermediate in biological denitrification to demonstrate that it is a suitable substrate for the enzyme. tation are verified, it would appear that Mechanism 11, as proposed, cannot be correct. However, it does not rule out all mechanisms of the second type, namely, sequential addition of two NO; ions to the reaction. Goretski and Hollocher (3) have presented more recent evidence in favor of the first kind of mechanism. For more recent evidence in favor of Mechanism 11, see Weeg-Aerssens et al. (10).
Chien et al. (11) and Mariotti et al. (12) found that the overall nitrogen isotope effect (Bobs) associated with denitrification of NOi in soils decreased when carbonaceous materials were added. Bryan et al. (7) reported a similar finding using pure cultures and cell-free extracts of P. stutzeri. In these competitive experiments: Bryan et al. found that Bob decreased with the concentration of the ultimate electron donor (succinate in these experiments) and increased with the concentration of NO;. Whether the concentration of NO, or the ultimate electron donor was varied, the relationship between Bobs and the velocity of the reaction was linear.
The same relationship between velocity (normalized to Vma) and Bob was seen in whole cells and cell-free extracts.
These isotopic fractionation results bear on the mechanism of denitrification in the following way. The value of Bob is established by intrinsic isotopic fractionation factors associated with individual steps within the overall reaction, by the relative rates of steps in the forward uerszm reverse directions, and by the mechanism of the overall reaction. Since the magnitude of intrinsic isotope effects is very nearly constant: the only way in which Bob can vary is for the relative rates of forward and reverse reactions to vary. It can therefore be concluded that in P. stutzeri the concentrations of both NO; and the ultimate electron donor affect the relative rates of forward and reverse reactions. With additional information concerning the magnitude and direction of changes in the relative rates of forward and reverse steps of the reaction in response to changes in NO; and succinate concentration, the concomitant variation in Bob can be used to provide information about the reaction mechanism. This paper reports enzymatically catalyzed exchange of "0 between H20 and NO; and the nitrogen isotope effect in denitrifying cultures of P. stutzeri measured in the same experiment. The objective was to determine the rate of return of enzyme-bound nitrogen intermediates to the free NOT pool compared to the overall rate of disappearance of NO; in order to determine whether one or both types of mechanism described above are consistent with the nitrogen isotopic fractionation data, as constrained by the '"0 exchange data.

MATERIALS AND METHODS
Preparation of Resting Cultures-Cultures of P. stutzeri (JM300) were grown anaerobically in defined medium (100 mM NaN03, 25 mM sodium succinate, 18.7 mM NH4C1, 0.8 mM MgSO4, 5 mM K2HP0,, 5 mM KHzP04, 0.36 mM FeSO,, and 2 ml/liter trace elements). The trace element solution consisted of 125 mg of EDTA, 154 mg of MnSO,. HzO, 10 mg of CuS04.5H20,24.5 mg of Co(NOd2. 6H20, 17.7 mg of Naa40,. 10H20, and 100 mg of NazMoOl. 2Hz0 in 100 ml of HzO. The pH of the growth medium was 6.8. Cells were harvested at midlog phase by centrifugation; washed three times with In competitive isotopic fractionation experiments, both heavy and light isotopes (usually at natural abundance) are present in the same reaction vessel and are thus exposed to the same enzyme concentration. This type of experiment is the method of choice when measuring isotopic fractionation involving heavy atoms (atoms heavier than hydrogen).
Isotopic fractionation factors are a function of temperature. However, in the physiological temperature range, this variation is quite small. 100 mM MOPS,6 0.8 mM MgS04, 5 mM KzHP04, 5 mM KHzP04 (pH 6.8); and resuspended in washing medium to a concentration 10 times that at harvest.
Incubation Conditions-Cells were incubated anaerobically (under helium) in a medium consisting of 5 mM KzHP04, 5 mM KHzP04 (pH 6.8), 0.8 mM MgSO,, 25 mM sodium succinate, and the indicated concentrations of NO; (at natural abundance) in "0-enriched (200%0 '"0 or 0.0408 atom % excess '"0) HzO. This low '"0 enrichment was selected in order that the "N and "0 analyses could be done on identical samples incubated in parallel. (About 0.8 mg of nitrogen is required for analysis. Therefore, for incubations at low NO; concentrations, a large volume was used. For example, when the concentration of NO; was 0.107 mM, the volume of the incubation mixtures was 1000 ml. High l8O enrichments would have been prohibitively expensive.) The time course of the reaction was followed by measuring the concentration of NO; colorimetrically (13). The reaction was stopped when about half of the added NO; had been consumed by adding 5 ml of 50% (w/w) NaOH/liter. NaOH was added at time 0 to obtain control samples. The initial concentration of NOT in control samples was half that in experimental samples, so that the final NO; concentration was approximately the same in control and experimental samples.
Isolation of Nitrogen from Residual NO; for Analysis-The incubation mixture was concentrated to a volume of about 100 ml by heating on a hot plate in a fume hood, care being taken to avoid bringing the samples to dryness. NH:, part of which arises from the breakdown of compounds such as amides as a result of heating under alkaline conditions, was completely removed by steam distillation. The nitrogen of the residual NO; was then reduced to NH: and collected by steam distillation (14). After titration to determine the quantity of nitrogen, acidification with H2SOl, and concentration to about 1 ml, the (NH4)zS04 was placed into a Sprinson-Rittenberg tube, frozen in liquid Nz, and evacuated three times prior to generation of Nz with NaOBr, as described previously (15).
Preparation of Samples for Analysis of '"0 in Residual N0;"NOT was reacted with azide to produce NzO, the Nz0 being collected cryogenically for "0 analysis. The overall reaction is: HNOz + HN3 + NzO + Nz + HZO. This reaction requires mildly acidic conditions. Acidic conditions also promote the competing chemical exchange of '"0 between Hz0 and NO;. Therefore, conditions must be adjusted so as to cause the rate of production of N20 by reaction with NB to be very much greater then chemical isotopic exchange. Initial attempts to accomplish this were based on the method described by Garber and Hollocher (1). The method was modified to accommodate the very much larger volumes used in our experiments (10-100 ml of sample in our experiments, compared to 0.1 ml in the experiments of Garber and Hollocher). The scaled-up version of this procedure resulted in 60-100% chemical exchange during the reaction with azide. Considerable effort was required to work out conditions which would result in an acceptable level of chemical isotopic exchange during reduction of NOT to N20 using azide. The crucial features required under our experimental conditions (low NO; concentrations and large volumes) are: (i) high azide concentration (16), (ii) high halide concentration (16), (iii) moderate acidity, and (iv) anaerobicity. The method used is summarized below.
After stopping the reaction with NaOH, as noted above, the incubation mixtures were neutralized with 1 M KHzPO4 to avoid loci of low pH. (Chemical isotopic exchange was unacceptably high when samples were neutralized with HzSOI, HCl, or H3P04.) After neutralization, these mixtures were repeatedly degassed under vacuum and backfilled with helium. An azide solution (3.5 M) containing 1.66 M NaCl was likewise repeatedly evacuated and backfilled with helium. Similarly treated 1 M HC1 was added to the azide solution (2.16 parts HCl to 1 part azide solution), and the mixture was briefly evacuated and backfilled with helium. (Acidified azide solutions cannot be exhaustively evacuated because under acidic conditions, substantial quantities of HN3 (an explosive) would be lost, resulting in an increase of pH.) For samples containing the higher NO; concentration, 10 ml of the acidified azide mixture was placed into an evacuated reaction vessel by syringe. The flask was further evacuated (briefly) prior to injection of an equal volume of sample (10 ml). For samples containing the lower NO; concentration, 100 ml of the acidified azide solution was poured into an evacuated vessel which was immediately evacuated again. An equal volume of sample (100 ml) was forced into the vessel from a separatory funnel with helium. The final pH of the reaction The abbreviation used is: MOPS, 4-morpholinepropanesulfonic acid. mixture was 4.5. The reaction was carried out for 10 min at room temperature; and then the product, NzO, was flushed out of the vessel with helium into a U-tube submerged in liquid Nz for 10 or 20 min for high and low NO; concentrations, respectively. Two NaOH traps were placed between the reaction vessel and the pump and between the reaction vessel and the U-tube to convert any HN3, which may have evolved from the reaction vessel, to NaN3. (Efficient trapping and conversion of HN3 is extremely important from the standpoint of safety, especially when working with large volumes of concentrated azide under acidic conditions.) The NaOH traps were kept inside steel sleeves to contain any explosion which might occur due to inefficient conversion of HN3 to NaN3. The cold trap between the NaOH traps and the pump was also placed inside a steel sleeve during thawing. In addition to these safety measures, the entire system of reaction vessel and traps was housed behind a 12-mm-thick plastic shield, and waste azide was alkalinized. (We can attest from personal experience that this concern for safety is essential. Fortunately, the explosion which occurred in our laboratory during the course of our experiments did no damage.) The U-tube containing the product of the azide reaction was evacuated while still in liquid Nz in order to remove helium and other extraneous gases. T h NzO was then cryogenically transferred to a second tube containing 2 ml of 50% (w/w) NaOH and allowed to equilibrate overnight at room temperature. This step is necessary to ensure removal of the last trace of COZ, which grossly interferes with analysis of "0 in N20, when using an isotope ratio mass spectrometer (as required for precise measurement of the low '"0 enrichment we used). After three successive cryogenic transfers, the NzO was analyzed for "0 abundance.
Isotope Analysis-The "N abundance of NO generated from NH: and the l80 abundance of NzO produced by reduction of NO; by Ng were measured in a dual collector, dual inlet mass spectrometer (VG Micromass 602E). ' ' N abundances were determined by comparing the ratio (29 m/e + 28 m/e) of the sample to that of a working standard of NZ generated from (NH4)zS04 whose "N abundance relative to atmospheric Ne was measured by A. Mariotti (Institut National de la Recherche Agonomique, Versailles, France). '"0 abundances were measured by comparing the ratio 46 m/e + 44 m/e to that of an "0 working standard. The working standard for '"0 measurements was tank NzO whose '"0 abundance compared to standard mean ocean water was measured by comparing it to N20 generated from NO; (with Ng) in standard Hz0 of known '"0 abundance under conditions in which complete '"0 exchange between Hz0 and NO; was achieved.
Conditions for complete '"0 exchange were to adjust the NO; solution to pH 11 and then to return the pH to 4.0 with HzSO4 and to carry out the reaction as described above, but in the absence of halides.
The "0 abundance of the standard Hz0 was measured by Tom Anderson (University of Illinois). The '"0 abundance of Hz0 in the medium was determined by comparing the "0 abundance of COz equilibrated with medium uersus standard HzO.
Isotope abundances were expressed as 6 values (%o enrichment of the heavy isotope), i.e.

H H
where H and L refer to the concentration of molecules bearing the heavier and lighter isotope, respectively, and spl and std refer to experimental sample and standard, respectively.
The overall nitrogen isotopic fractionation factor (&b) was calculated from an approximate relationship derived by Mariotti et al.
where the subscripts s,t = 0 and s,t = t refer to the substrate at zero time and at the time the incubation was stopped, respectively, and f is the fraction of substrate remaining at t = t.
The percent enzymatically catalyzed '"0 exchange between HzO and NO; was calculated as where the subscripts spl, cont, and Hz0 refer to the experimental sample, the control, and HzO in the incubation medium, respectively. Fig. 3 shows the time course of disappearance of NO; during incubation with resting P. stutzeri cells as measured by colorimetric assay. Four replicate incubations were done using two initial concentrations of NO; (0.107 and 2.29 mM). The concentration of cells was adjusted so that the time required for disappearance of half of the NO; was approximately the same for the two NO; concentrations. (For the high NO; concentration, the cell concentration was 6.4 times higher than for the low NO; concentration). In two replicate incubations at each initial NO; concentration, the water in the incubation medium contained "0. After stopping the reaction, these four samples were used to determine the "0 abundance of the residual NO;, whereas the other four samples were used to determine the abundance of the residual NO;. NO; disappearance was linear during the period of the incubation. Velocities, calculated from the slope of the regressions, were 88.1 and 242 nmol (min-mg of cell, dry weight)" for the low and high NO; concentrations, respectively.

RESULTS
The amount of NO; consumed between time 0 and the time that the reaction was stopped was determined by distillation techniques. Since NO; consumption was linear with time, it was possible to calculate the velocity of the reaction from this  Table I). These calculated velocities are close to the values calculated from the slope of the regressions of Fig. 3. Bryan et al. (7) previously reported that denitrification of NO; by P. stutzeri followed Michaelis-Menten kinetics. The velocities calculated from the data shown in Table I are most consistent with values for     mM (the concentrations used in the experiment of Table 11), respectively. This prediction is in close agreement with the values calculated from results reported here, namely, 1.010 and 1.020 (Table 11). Table I11 shows the results of "0 measurements for replicate incubations carried out in HZl80 (b"0 !z 200%0). Control samples (samples to which NaOH was added to the incubation medium before the cells) were not paired to experimental samples. For this reason, the "0 abundances of NO; in two control samples at each NO; concentration were averaged, with the mean being subtracted from the mean of the experimental samples to determine the "0 enrichment due to biological activity. The "0 enrichment of the control samples resulted from chemical exchange which was about 3 times greater at the higher NO; concentration than at the lower.
At the low NO; concentration, the mean "0 enrichment of residual NO; due to biological activity was 10.6 f 2.4%0 (61s0,1 -6180mnt~). Since the "0 enrichment of the incubation medium was 198%0, 5.3 f 1.2% of the oxygen atoms in the NO; were enzymatically exchanged with Hz0 in the medium. At the high NO; concentration, the "0 enrichment of residual NO; due to enzyme-catalyzed exchange was 9.5 f l.1%0, that is, 5.3 & 0.6% of the oxygen atoms in the NO; were exchanged with HzO as a result of enzymatic activity. Despite the fact that the reaction was carried to a high extent of reaction (about 50%), very little enzymatically catalyzed oxygen exchange occurred.
The rate of enzyme-catalyzed entry of oxygen atoms into substrate (NO;) from water (to be called u.sW) can be calculated as follows. The quantity of oxygen in NO; derived from HzO (to be called q(t)) changes with time and is determined by the rate of entry of oxygen atoms into the NO; pool from HzO (u.sW) and the rate at which oxygen atoms derived from HzO leave the NO; pool (to be called u1sw). The relationship between q(t) and these two rates may be represented as a firstorder differential equation: dq(t)/dt = vasw -u1sW. Solution of this differential equation will allow us to estimate a value for uesW, which is equal to kz[E.NO;] (see Figs. 1 and 2). Under steady-state conditions, kz[E. NO;] is constant. Therefore, vasw is a (presently unknown) constant. At any instant, ulsW is determined by the overall velocity of the disappearance of oxygen from the NO; pool via denitrification (uolrs) and the proportion of the total oxygen of the NO; pool (T) that was derived from HzO: ulsw(t) = uol~![q(t)/T(t)]. The concentration of NO; was shown to decrease linearly with time (Fig. 3), that  is, uolf~ was constant. Therefore, T(t) = iu o d , where i is the initial quantity of NO;.
With the above information, the differential equation given above can be written more explicitly as follows.
The integrated form of this equation, found by the method described by Leighton (18), is as follows.
Our measurement of the exchange of "0 between Hz0 and NO; allows us to calculate q at time t', the time the incubation was terminated; q(t') = the fractional exchange of "0 from HzO to NO; times the total quantity of NO; in the medium at time t'. Substituting this value and the measured values of i, t', and uolf~ into this equation permits us to calculate vas, at t'. But since u.sW is constant with time, the value at t' is its value at all times. Using values in Tables I and 111, we calculated uesW and U.S~/UOI~S. Table IV shows that the rate of enzymatically catalyzed entry of oxygen from Hz0 to the NO; pool is about 5-8% of the overall rate of oxygen atom loss. The work of other investigators (1, 5 ) with P. stutzeri has shown that the initial protonation-dehydration step is highly reversible. It is therefore reasonable to conclude that enzyme-substrate dissociation is extremely slow in this organism.

DISCUSSION
One possible explanation for the observation that Bobs increases with substrate concentration is that entry of substrate into the cell is diffusion-limited at low substrate concentration and by an isotopically sensitive enzymatic reaction at high concentration. In this case, as NO; concentration approached 0, Bob would approach approximately 1.00 since there is little isotopic fractionation associated with the diffusion of an ion in aqueous solution. As substrate concentration approached infinity, the isotope effect associated with the rate-limiting enzymatic reaction would be fully expressed. Such a set of circumstances would result in the observed variation of Bobs with NO; concentration. This simple explanation does not account for the variation in bob associated with denitrification of NO; in P. stutzeri because the variation in Bobs was also observed in cell-free extracts, and the relationship between Bob and velocity (normalized to maximum velocity) was the same in whole cells and cell-free extracts (7), that is, the rate of entry of NO; into the cell was fast compared to the enzymatic reaction even at low substrate concentrations. It can be concluded that the isotope effect for the enzymatic reaction varies with substrate concentration in the case of NO; denitrification by P. stutzeri. The mechanistic significance of this variable isotope effect is discussed below.
Mechanism I-The small degree of incorporation of oxygen from HzO into NO; at both high and low [NO;] means that either (or both) step 2 (dissociation of the enzyme-substrate complex) or (and) step 4 (hydration-deprotonation) must be very slow compared to the forward reactions with which they compete, irrespective of [NO;]. Because of the substantial degree of incorporation of oxygen atoms from HzO into NzO in P. stutzeri observed by others (1, 5 ) , it seems most likely that the rate of enzyme substrate dissociation rather than hydration-deprotonation limits the rate of entry of oxygen atoms into NO;. However, it seems prudent to examine both cases. If the low rate of entry of oxygen into NO; from HzO resulted from rate limitation by enzyme-substrate dissociation  ]. For example, increasing the concentration of NO; causes an increase in the ratio of oxidized to reduced species in the electron transport chain. Therefore, the ratio of the forward to the reverse rate of reaction of E. NO+ decreases with [NO;]. However, this also cannot account for the observed variation of Bobs with [NO;] because, in the case that the rate of hydration/protonation limits the rate of entry of oxygen atoms from H20 into NO;, the isotopic preference of the subsequent step 5 (the reduction of E. NO;) would be only slightly expressed (19). The conclusion from the above argument is that, whereas Bobs might be significantly greater than 1.00, there is no way that Bobs could have differed substantially at the two NO; concentrations used in the experiments reported in this paper. Whereas we believe most readers will find the above discussion convincing, some have suggested that the small observed incorporation of oxygen into NO; from HzO is large enough to allow sufficient variation in the expression intrinsic isotope effects associated with steps beyond step 3 to account for the observed variation of Bobs with NO; concentration. A more quantitative discussion is required to meet that objection to our conclusion; namely, that the data of this paper are inconsistent with the operation of Mechanism I in P. stutzeri under the conditions of our experiment. Such a discussion is based on the equations derived in the Miniprint,'where expressions for Bobs for the two mechanisms Portions of this paper (including Figs. S1 and S2 and Equations 1-57) are presented in miniprint at the end of this paper. Miniprint -are derived. The derivation for Mechanism I is based on the assumption that step 5 is unidirectional. (If step 5 is unidirectional, it is irrelevant to the derivation of an expression for Bobs whether subsequent steps are or are not unidirectional since isotopic fractionation associated with steps beyond the first unidirectional step following entry of substrate is not expressed in @&s. ) The assumption of unidirectionality of step 5 is based on the strongly exergonic (=+320 mV) nature of the half-reaction for the one-electron reduction of NO+ to NO (20) as well as the exergonic nature of the half-reaction for the oxidation of the stoichiometric electron donor used in the experiments here, succinate (=+320 mV for its two-electron oxidation to fumarate). If the assumption that step 5 is partially or wholly incorrect, then the effect of the error would be that isotopic fractionation associated with later steps, up to and including the first unidirectional step (but not beyond), would be folded into the isotope effect for step 5 (p5).   ]. The remaining question is whether the error in the measured value of uesw/u0lfS is large enough to encompass values of N which could lead to plausible values of the intrinsic isotope effects consistent with the observed variation in Bobs. Examination of this question is described below.
Because isotopic fractionation is not expected to be associated with binding of NO; to the enzyme or with enzyme-NO; dissociation, and p2 may be taken to be approximately equal to 1.000. Therefore, the equation for @ob8 for Mechanism I may be rewritten as the following. ,,itrite) and that 0 < N < 1. To meet these constraints, it was necessary to assign values to (u,sw/ U O I~S ) I~~~,~~~~ that were lower than the experimental mean and to assign values to (UeSw/UOlfS)highnitrite that were higher than the experimental mean. The combinations used encompassed 2 standard errors below the mean for (uesw/uol~)~ow nitrite and 2 standard errors above the mean for (UeSw/UOlfS)higbnitrite. For Mechanism I to account for the experimentally observed variation of Bobs with [NO;], it is necessary that the parameter N also vary with [NO;]. Fig. 4 shows the difference in values of N at low and high [NO;] over the range of possible values of P calculated from these assigned values of uesw/uolfS using Equation 35 of the Miniprint. As might be expected, the difference in values of N for the two NO; concentrations increases with deviation of u~s~/ u~~~ from the measured mean.
Values of p3 and p5/p4 were calculated from the measured values of bobs at the two NO; concentrations and the equation for pobs given above using the same assigned combinations of values for u~s~/ u~~~s at the low and high NO; concentrations used for Fig. 4. The most plausible values for intrinsic isotopic fractionation factors of the pobs equation were (predictably) obtained when the assigned values of u,sw~uolfS resulted in the largest difference in values of N for the two NO; concentrations; namely, ( v~s~/ u o~~s ) I~~ nitrite -2 standard errors from the measured mean, and (U&/UOlfS)high nitrite = +2 standard errors from the measured mean. Fig. 5 shows the values of N as a function of P for the two calculated NO; concentrations using these assigned values of u~s~/ u~~~, and Fig. 6 shows values of p3 and &/p4 for the same assigned values of u.s,.,/uo~~s. Fig. 6 shows that the most plausible values for p3 and p5/p4 occur in the limit as P + 0. (In the other limit, P + 1, p3 and p5/p4 drop out and p -+ 1 pobs = 01. lim That is, in the limit P +. 1, Bobs is invariant and equal to about 1.000.) In the limit P + 0, N may be calculated from Equation 34 of the Miniprint.
where a' = 1, Using this expression together with the expression for Pob, we can examine the consistency of experimental results with predictions from Mechanism I1 in the two limits P + 1 and P "-* 0. Given the assumption that PI and & are 1.000, in the limit P 1, Step 2 limits the entry of oxygen from Hz0 into NO;.
' From Ref. 6  in the full expression for bobs cancels. In the opposite limit, P + 0 ( i e . kz >> k3), it does not. However, it is possible to substitute a measured value for this term by using ] increases. Thus, in either limit P + 0 or P + 1, Mechanism I1 is consistent with the constraints of our isotopic fractionation and "0 exchange data. The above analysis is based on derivations from a model in which the first reductive step is taken to be unidirectional. To the degree that this assumption is incorrect, the isotope effect of subsequent isotopically sensitive steps (up to and including the first unidirectional step) will be folded into p7.
This has the effect of extending even further the range of values of p7 which would satisfy the derived equation given the constraints of the isotopic fractionation and "0 exchange data. Since the demands of the equation derived from Mechanism I1 already fall within plausible values of the intrinsic isotope effects, it is of no particular consequence that the range of plausible values may be wider than it would be if p7 reflected only events associated with the first reductive step.
We conclude, on the basis of the above analysis: that the nitrogen isotopic fractionation data in combination with "0 exchange data are inconsistent with Mechanism I and are consistent with Mechanism 11. Two issues need to be addressed concerning the viability of Mechanism 11. First, it is well known that NO can serve as sole substrate in P. stutzeri and other denitrifying microorganisms for the enzymatic reduction to NzO (e.g. Refs. 2, 5 ,  26, and 27), albeit at a reduced rate. Aerssens et al. ( 5 ) pointed out that this need not be taken as evidence against Mechanism 11. They proposed that NO may be oxidized to NO+ and then converted to NO; by way of steps 4 and 2 (Fig. 2). This NO; would then be available for entry into the pathway at step 5. Data reported in this paper show that the rate of back-reaction of E . NO+ to free NO; is very slow in P. stutzeri compared to the overall rate of reduction (40%). However, it may be fast enough to support the suggestion of Aerssens et al. (5). Garber and Hollocher (27) reported that the rate of NO reduction was less than 25% of the rate of NO; reduction in P. stutzeri.
If Mechanism I1 is correct, only half of the nitrogen in the product (NzO) would need to be derived from NO; in the presence of NO as the sole substrate, the other half being derived from the oxidation of NO to E .NO+. Thus, the rate of reduction of NO relative to the rate reduction of NO; (<25%) observed by Garber and Hollocher is reasonably consistent with a 10% return of enzyme-bound nitrogen to the free NO; pool. Moreover, oxidation of E .NO to E .NO+ might cause the concentration of E . NO+ to be increased relative to its concentration in the absence of NO. Mass action might well increase the rate of the reverse reaction, yielding more NOT for entry into the reaction at step 5 than would be expected on the basis of the rate of formation of NO; from E . NO+ in the absence of NO. (On the other hand, the net oxidation of NO to E .NO+ would have to take place in the reducing environment responsible for the production of NzO.) It is also possible that NO is reduced in a separate enzymatic process. An enzyme apparently distinct from nitrite reductase which reduces NO has been reported in at least one denitrifying microorganism (28).
The second issue arises from the observation that the major product of reduction of NO; by cell-free extracts and partially purified nitrite reductase of most denitrifying organisms is NO (29). Moreover, in whole cell experiments, Goretski and Hollocher (3) succeeded in trapping NO with hemoglobin, indicating that at least some fraction' of NO; is reduced by a pathway in which NO is a free intermediate, yet NO is not an intermediate in Mechanism 11. Averill and Tiedje (4) proposed two competing pathways of reaction in the forward direction of the intermediate, E.NO+ namely, attack on this intermediate by NO; followed by reduction uersus decomposition to NO (i.e. Fez+. NO+ -+ Fe3+. NO + NO + Fe3+). They considered the first pathway to predominate in uiuo, but pointed out that disruption of the cell and partial purification of the enzyme could so alter the environment of the enzyme that the ratio of the rates of the two pathways could well be inverted. A second possible competition for an intermediate of Mechanism I1 which would yield NO is the decomposition of E.Nz03. The chemical precedent for this suggestion" is the dissociation of free N203 to NO + NOz (30). In the context of denitrification, the NZ03 would be bound to the enzyme by interacting with Fez+. The liberation of NO would leave behind Fe2+.NOz which would undergo internal oxidationreduction to form Fe3+-NO;. After dissociation of NO; from Fe3+, the Fe3+ would then be reduced to Fez+ by the appropriate redox reagent. Again, the relative rates of the two competing pathways could well be dramatically affected by cell disruption and enzyme purification. The possibility of the existence of forward pathways competing for the same intermediate and yielding NO is supported by observations by Kim and Hollocher (31) on NO; reduction by nitrite reductase purified from Pseudomonas aeruginosa. Most of the reductive product was NO, but 2-5% of the product was NzO, even though no reduction occurred when NO; was replaced by NO. The data presented in this paper are inconsistent with a mechanism of denitrification of NOT in which two NOT ions are reduced prior to nitrogen-nitrogen bond formation. Our data are consistent with a mechanism in which two NO; ions enter the reaction sequentially, with the nitrogen of at least one of the two having a valence of 3+ when the nitrogennitrogen bond is made. It would be of interest to determine the generality of the results reported here since it seems probable that the enzyme-nitrosyl complex has two fates, namely, to react with NO; to form N203 versus reduction to NO. If the first fate is predominant, then the main mode of production of NzO is a mechanism like that proposed by Averill and Tiedje (4), Mechanism 11. However, it is easy to imagine that, in other organisms or under different conditions, the second fate might predominate. In those cases, the main path to N20 would be by way of a mechanism like that proposed by Payne (2) or Garber and Hollocher (1).