H2180 Isotope Exchange Studies on the Mechanism of Reduction of Nitric Oxide and Nitrite to Nitrous Oxide by Denitrifying Bacteria EVIDENCE FOR AN ELECTROPHILIC NITROSYL DURING REDUCTION OF NITRIC OXIDE*

fluorescens was a soil isolate in Characterization of copper-type or cytochrome cdl reductases in these has been classification of eutrophus ATCC as a copper NiR-containing organism dithiocarbamate

Reduction of NO and NO; by whole cells of eight strains of denitrifying bacteria known to contain either heme cdl or copper-containing nitrite reductases (NiRs) has been examined in the presence of HZl80. All organisms containing heme cd, NiRs exhibited relatively large extents of exchange between NO; and Hz"O (39-loo%), as monitored by the "0 content of product NzO. Organisms containing copper NiRs gave highly variable results, with Achromobacter cycloclastes and Pseudomonas aureofaciens exhibiting no "0 incorporation and Rhodopseudomonas sphaeroides and Alcaligenes entrophus exhibiting complete exchange between NO; and HZ"0. Organisms containing heme cd, NiRs exhibited significant but lower levels of exchange between NO and H2"0 than between NO; and Hz"O, while organisms containing copper NiRs gave significantly higher amounts of "0 incorporation than observed for the heme cdl organisms. These results demonstrate the existence of an NOderived species capable of undergoing 0-atom exchange with H 2 1 8 0 during the reduction of NO. Trapping experiments with l6NO, l4N;, and crude extracts of R . sphaeroides support the electrophilic nature of this intermediate and suggest its formulation as an enzyme nitrosyl, E-NO+, analogous to that observed during reduction of NO;.
The observation of lower levels of "0 incorporation with NO; than with NO as substrate for A. cycloclastes and P . aureofaciens indicates that, for these organisms at least, a sequential pathway involving free NO as an intermediate is significantly less important than a direct pathway in which NzO is formed via reaction of two NO; ions on a single enzyme.

* This research was supported by National Science Foundation
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
11 To whom correspondence should be addressed.
The mechanism of microbial denitrification remains a controversial subject, despite a wealth of detailed studies on both intact bacteria and isolated enzymes (1)(2)(3)(4). It is now generally accepted that denitrifying bacteria possess a nitric oxide reductase activity that is distinct from the nitrite reductase (NiR)' activity. The latter are typically soluble enzymes that are rather easily purified and have been shown to be of two distinct types: a cytochrome cdl-containing dimer of -60-kDa subunits and a copper-containing enzyme that is more variable in both subunit size and degree of oligomerization (1,5). The membrane-bound nature of the nitric oxide reductase activities has hindered their purification and characterization, but recently nitric oxide reductases have been purified to apparent homogeneity from two organisms (6-8) and shown to contain both heme b and c prosthetic groups (7,8).
Virtually all workers in the field now agree that at least a significant portion of the total nitrogen flux occurs via a stepwise pathway with NO as an intermediate (Equation 1, where NOR is nitric oxide reductase), rather than via the direct pathway previously proposed (Equation 2), in which two NO; ions are reduced to N20 on a single enzyme (NiR) (9).
Indeed, quantitative studies of NO levels during denitrification by several denitrifiers have been interpreted as indicating that only the former pathway (Equation 1) is operative and that NO is a free obligatory intermediate in denitrification (10)(11)(12)(13). This conclusion is consistent with the observed lack of reduction of NO by isolated NiRs, with the fact that most isolated NiRs produce predominantly NO upon reduction of nitrite, and with the fact that mutants lacking either the heme cdl (14) or copper NiRs2 are still capable of reducing NO. It fails, however, to account for the following observations. (i) Purified NiRs do, in at least some cases, produce significant amounts of NzO that cannot be attributed to chemical reduc- we demonstrate via comparison of the amount of "0 incorporated into NzO from either NO; or NO that, in certain bacteria at least, reduction of NO; may not proceed entirely according to the stepwise pathway shown in Equation 1.

MATERIALS AND METHODS
Bacterial Strains-Pseudomonas aeruginosa PAOl was from B. W.
Holloway, Monash University, Clayton, Australia; Rhodopseudomonas sphaeroides forma sp. denitrificans was from T. Satoh; and Pseudomonas stutzeri JM 300 was from J. Ingraham. Pseudomonas fluorescens AK-15 was a soil isolate obtained in this lab~ratory.~ The rest of the strains used were from ATCC. Characterization of coppertype or cytochrome cdl nitrite reductases in some of these strains has been described (22); classification of Alcaligenes eutrophus ATCC 17699 as a copper NiR-containing organism was performed using N,N-dimethyl dithiocarbamate as described (22).
Sample Preparation-Cultures were grown anaerobically in 3% tryptic soy broth (Sigma) containing 0.15% potassium nitrate in 110ml serum bottles. P. aeruginosa PAOl was grown overnight at 37 "C, and the rest of the cultures were grown at 30 "C. Cells were harvested, washed, and resuspended in tryptic soy broth in an 8-ml serum bottle. These bottles were made anaerobic by flushing with argon, and nitrite or NO was added to start the reaction; suspensions were maintained at room temperature. 100 p1 of 10 N NaOH was added to stop the reaction and absorb CO,. Crude extracts were prepared by sonication followed by centrifugation at 10,000 X g for 30 min to remove cell debris.
Isotopes and Their Analy~is-~~NO was prepared by mixing 1 ml of 100 mM &SO4, 1 ml of 100 mM KI, and 1 ml of 299 mM Na"N02 (99.9 atom I5N%) in a 25-ml serum bottle (23). Gas chromatography/ mass spectrometry measurements and the calculation of the extent of IHO exchange were performed as described (19, 23).

RESULTS AND DISCUSSION
Exchange with HZ"O during Nitrite Reduction-The results obtained for four denitrifiers known to contain heme cdl NiRs a n d four denitrifiers known to contain copper NiRs are presented in Table I (18,19,23-25), organisms containing such enzymes exhibited relatively large amounts of "0 incorporation from HZ1'0 (extents of exchange ranging from 39 to 80%) during reduction of NO; to NzO. In fact, strain 2 exhibited significantly more than 50% exchange. This is significant because our original hypothesis regarding the direct pathway (9) (Equa-:' A. Arunakumari and J. M. Tiedje, unpublished results. tion 2) postulated that the N-N bond was formed by attack of a second NO; upon an electrophilic nitrosyl intermediate, E-NO+, derived from the first nitrite, which is known to undergo facile "0 exchange by a reversible hydrationldehydration process (23-25). Shearer and Kohl (21) have shown that the NiR from P. stutzeri JM300 is a "sticky" enzyme and that NO; is committed to reduction once bound. Hence, "0labeled NO; does not accumulate and "0 incorporations of >50% are not consistent with the direct pathway (Equation 2), unless the NiR in organism 2 differs substantially from that in P. stutzeri JM 300 (organism 3).
The organisms containing copper NiRs exhibited significantly different behavior in two cases (6 and 7 ) , where essentially no "0 incorporation into NzO product was observed.
This result is consistent with previous work on the Achromobacter cycloclastes system, which showed undetectable amounts of "0 exchange (26  Table I, substantial amounts of "0 were also observed in NzO produced by reduction of NO. The extent of "0 incorporation observed with NO tended to be lower than that observed with NO; for the same organism (except for 6 and 7, see below) but was well above background levels for all but one case. The four organisms with copper NiRs gave  I11 Nitrosyl transfer from "NO to 14Ni in reaction containing H2"0 and R. sphaeroides crude extracts Reaction mixtures contained 2 pmol of "NO and 1.0 mM NaN3; all other conditions were as in Table 11. Reactions were stopped after 30 min. Percent nitrosation was calculated based on the ratio of 14,15N20 and total amount of nitrous oxide produced. Data are for duplicate experiments at 1 mM azide, but similar results were obtained with duplicates run at both 2.5 and 5 mM azide as well. significantly higher extents of "0 exchange (30-84%) than did those known to contain a heme cdl NiR (4-19%), but the origin of this difference is unclear since an NO reductase has yet to be purified from any of the former. The data in Table I, column 2, demonstrate the presence of an NO-derived species capable of undergoing 0-atom exchange with H2"0 during the reduction of NO, which would not be expected apriori to proceed via an electrophilic species. The overall reaction can be written as

Isotope
indicating that water is produced during the reaction (presumably by protonation and dehydration of a hyponitrite level species containing two N atoms, such as N20;-), suggesting a possible route for "0 incorporation if the final dehydration step were reversible. As a control experiment, cells of Pseudomonas aureofaciens were grown on NO:, suspended in medium containing 10% H2"0, and incubated anaerobically for 5 h at room temperature with 0.1 ml of N 2 0 (8.8 pmol) in an 8-ml bottle. The "0 content of the N20 was measured and did not differ from the natural abundance. Thus, the observed '"0 incorporation during reduction of NO must occur prior to reduction to the NzO level.
This conclusion is also supported by the data in Table 11, which demonstrate that for R. sphaeroides at least the extent of exchange with H2"0 decreased as the concentration of electron donor/mediator was increased. This suggests strongly that '"0 exchange occurs via a relatively oxidized nitrogen intermediate.
Trapping with Azide during NO Reduction-If an electrophilic NO-derived species is indeed present during reduction of NO, it might be expected to react with nucleophiles other than H,'"O. For example, N: and NH,OH have been reported to trap the electrophilic nitrosyl produced by the heme cdl NiRs (19, 20, 23-25). Thus, crude extracts of R. sphaeroides were treated with 15N0 in the presence of 14N: and H2180; the amounts of the various isotopically labeled forms of N 2 0 formed are given in Table 111. It is clear that substantial (-14%) amounts of a nitrosation product 14,15N20 were observed even at the relatively low azide concentration used (1 mM) and that both the nitrosation product and NO reduction product exhibited comparable "0 incorporation.
Zmplications for the Mechanism of NO Reduction-The Hz"O exchange and N; trapping results presented above strongly imply the presence of an electrophilic mononitrogen intermediate during reduction of NO. The simplest such species is an enzyme nitrosyl, E-NO', analogous to that observed for the heme cdl NiRs. This is a most unexpected result, since it is not obvious why an oxidized NO species should be an intermediate in its reduction to N20. We note that similar results observed earlier for P. stutzeri J M 300 in the presence of NO and H2"0 or "NO and I4NH2OH (19) are also completely consistent with the results reported here (although they were initially attributed to an NO complex of the heme cd, NiR in this organism).
Two possible tentative explanations for the data are shown in Schemes 1 and 2. In Scheme 1, the "0 exchange and nucleophilic trapping occur via a species that is not on the catalytic pathway for NO reduction but is rather on an oxidized "shunt." Given the redox potentials for synthetic heme nitrosyls (27)(28)(29), it might not be surprising if electron transfer to another center on the enzyme or elsewhere were to generate a transient oxidized species that, as shown, has nothing to do with catalysis. The other extreme is represented in Scheme 2, in which the E-NO' species is an obligatory intermediate, possibly reacting via a hypothetical enzymebound NO; with a second E-NO' in a fashion analogous to that postulated by us earlier for the reduction of NO; by NiR (9). Available data do not permit us to distinguish between these alternatives or the many possible variants thereof.
Zmplications for the Pathway of Denitrification-Examination of the data in columns 1 and 2 of Table I reveals that, in most cases, our results are fully compatible with NO as an obligatory intermediate in denitrification, i.e. the sequential pathway shown in Equation 1. That is, for organisms 1-5 and 8, the extent of "0 incorporation observed for NO; as substrate is greater than that observed for NO, reflecting the fact that for these organisms "0 exchange can occur at either the NO; + NO or NO + N 2 0 steps. These data do not, however, constitute proof that the bulk of the nitrogen flux proceeds via Equation 1 rather than Equation 2, although they are not inconsistent with this view. In contrast, the data for organisms 6 and 7 (A. cycloclastes and P. aureofaciens) are not compatible with the view that NO is an obligatory free intermediate, because the amount of IRO exchange observed with NO; as substrate is far less than with NO. (If the situation shown in Equation 1 were to obtain, the amount of "0 incorporated into N20 derived from NO; would always have to be at least equal to that in N20 derived from NO.) Thus, for these organisms at least, the sequential pathway of Equation 1 appears to be significantly less important than a direct pathway as indicated in Equation 2.