Vol. 42, No. 6APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Dec. 1981, p. 1074-10840099-2240/81/121074-1 1$02.00/0

Kinetic Explanation for Accumulation of Nitrite, Nitric Oxide,and Nitrous Oxide During Bacterial Denitrificationt

MICHAEL R. BETLACH't AND JAMES M. TIEDJE" "

Department ofMicrobiology and Public Health,' and Department of Crop and Soil Sciences,2Michigan State University, East Lansing, Michigan 48824

Received 20 April 1981/Accepted 25 August 1981

The kinetics of denitrification and the causes of nitrite and nitrous oxideaccumulation were examined in resting cell suspensions of three denitrifiers. AnAlcaligenes species and a Pseudomonas fluorescens isolate characteristicallyaccumulated nitrite when reducing nitrate; a Flavobacterium isolate did not.Nitrate did not inhibit nitrite reduction in cultures grown with tungstate toprevent formation of an active nitrate reductase; rather, accumulation of nitriteseemed to depend on the relative rates of nitrate and nitrite reduction. Eachisolate rapidly reduced nitrous oxide even when nitrate or nitrite had beenincluded in the incubation mixture. Nitrate also did not inhibit nitrous oxidereduction in Alcaligenes odorans, an organism incapable of nitrate reduction.Thus, added nitrate or nitrite does not always cause nitrous oxide accumulation,as has often been reported for denitrifying soils. All strains produced smallamounts of nitric oxide during denitrification in a pattern suggesting that nitricoxide was also under kinetic control similar to that of nitrite and nitrous oxide.Apparent Km values for nitrate and nitrite reduction were 15 MM or less for eachisolate. The Km value for nitrous oxide reduction by Flavobacterium sp. was 0.5pM. Numerical solutions to a mathematical model of denitrification based on

Michaelis-Menten kinetics showed that differences in reduction rates of thenitrogenous compounds were sufficient to account for the observed patterns ofnitrite, nitric oxide, and nitrous oxide accumulation. Addition of oxygen inhibitedgas production from `3N03 by Alcaligenes sp. and P. fluorescens, but it did notreduce gas production by Flavobacterium sp. However, all three isolates producedhigher ratios of nitrous oxide to dinitrogen as the oxygen tension increased.Inclusion of oxygen in the model as a nonspecific inhibitor of each step indenitrification resulted in decreased gas production but increased ratios of nitrousoxide to dinitrogen, as observed experimentally. The simplicity of this kineticmodel of denitrification and its ability to unify disparate observations shouldmake the model a useful guide in research on the physiology of denitrifierresponse to environmental effectors.

Denitrification is the reduction of nitratethrough the intermediates nitrite, perhaps nitricoxide, and nitrous oxide to form dinitrogen gas(7, 14, 24, 31, 39). Nitrate-amended soils anddenitrifying bacterial cultures often accumulatenitrite and nitrous oxide but not nitric oxide.Renewed interest in accumulation of nitrite andnitrous oxide during denitrification reflects con-cern for their roles in nitrosamine formation anddestruction of stratospheric ozone, respectively(5).

t Journal article no. 9274 of the Michigan AgriculturalExperiment Station.

: Present address: Extraterrestrial Research Division,Ames Research Center, National Aeronautics and Space Ad-ministration, Moffett Field, CA 94035.

Many factors affect the accumulation of de-nitrification intermediates, but there have beenfew attempts to develop a unifying explanation.We report here on a kinetic mechanism whichcould explain a number of the observations sum-marized below. Nitrite accumulation may resultfrom a lag in synthesis of nitrite reductase (20,34, 37) or from nitrate inhibition of nitric oxidereductase (25). Nitrate and nitrite amendmentsenhance production of nitrous oxide (2, 8, 29,30), perhaps by inhibiting nitrous oxide reduc-tase or by acting as preferred electron acceptors.Net nitrous oxide production in soils varies withtime after onset of anaerobiosis (9, 23, 29, 30),possibly owing to a lag in synthesis of nitrousoxide reductase (9). Aerobic soils are sources of

1074

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

KINETICS OF BACTERIAL DENITRIFICATION 1075

rather than sinks for nitrous oxide (10, 11, 12,23), suggesting that nitrous oxide reduction ismore sensitive to oxygen inhibition than werethe other steps in denitrification, but nitrousoxide production during nitrification may ac-count for such observations as well (3).Measurement of denitrification rates and ni-

trous oxide production in soils is complicated bythe presence of heterogeneous communities ofdenitrifying organisms (13) and by the chemicaland physical complexity of soils (14). For in-stance, lack of suitable carbon sources (18, 38)and slow diffusion of nitrate to the sites of de-nitrification (26, 28) may alter not only the rateof denitrification but also the estimates of itskinetic parameters. In situ rate measurementsintegrate the effects of these extrinsic factors ondenitrification and so are suitable for estimatesof global nitrogen flux. However, such integratedmeasurements mask the basic dynamics of thedenitrification process.To avoid the complex and uncertain effect of

soils on denitrification, we examined causes foraccumulation of nitrite and nitrous oxide in purecultures of denitrifiers. Our results suggest thatthe transient accumulation and steady-statepools of intermediates can in many cases bedetermined by reaction rates rather than anyspecific inhibition of a step in the denitrificationsequence. We developed a model of denitrifica-tion based on Michaelis-Menten kinetics to dem-onstrate how different rates of reduction andgeneral inhibition of denitrification, e.g., by 02,could lead to the observed patterns of accumu-lation. The kinetic model should be a useful toolwith which to examine accumulation of denitri-fication intermediates in other systems and inresponse to other environmental variables.

MATERIALS AND METHODSOrganisms. An Alcaligenes species, isolate 17

(ATCC 33513); a Flavobacterium species, isolate 175(ATCC 33514); and a strain of Pseudomonas fluores-cens biotype II, isolate 72 (ATCC 33512) were selectedfrom a collection of numerically dominant denitrifiersfrom world soils (13). The isolates had been charac-terized previously (T. N. Gamble, M.S. thesis, Michi-gan State University, East Lansing, 1976). Alcaligenesodorans was obtained from the American Type Cul-ture Collection, Rockville, Md. (no. 15554). This spe-cies, wich does not reduce nitrate, readily denitrifies ifsupplied with nitrite (27). It was used to determinewhether nitrate directly inhibited nitrous oxide reduc-tion.Media. Organisms were grown on nutrient broth

(Difco Laboratories, Detroit, Mich.) containing 10 mMKNO3. For A. odorans, 5 mM KNO2 was used insteadof KNO3. Seed cultures (50-ml volumes) were grownfor 24 h at 30°C and then used to inoculate 450 ml ofnitrate broth in 500-ml Erlenmeyer flasks. A rubber

stopper was wired in place on each flask, and thecultures were incubated at 30°C for 24 h. This proce-dure allowed cultures to use residual oxygen in theflask while shifting to anaerobic metabolism. Nitrateand nitrite were not detected in the spent media.When desired, organisms incapable of nitrate reduc-tion were obtained by substituting 5 mM KNO2 forKNO3 and including 10 mM Na2WO4 in the growthmedium to prevent formation of an active nitratereductase (32).

Resting cell suspensions. Cells were harvestedby centrifugation, washed twice in nutrient broth con-taining 200 ,tg of chloramphenicol ml-', and resus-pended in nutrient broth containing chloramphenicol.The chloramphenicol concentration was sufficient toprevent growth of the least-sensitive organism, P.fluorescens. Stock cell suspensions were refrigeratedor kept on ice and used within 12 h after harvest.

Nitrate and nitrite reduction. A stock cell sus-pension was diluted before use to give desired rates ofreduction. The diluted cell suspension, 18 ml, wasplaced in a 125-ml flask containing a stirring bar. Alarge serum stopper was placed over the top of theflask, and the headspace gases were replaced threetimes with argon. Slight overpressure was maintainedin the flask to minimize oxygen entry. The cell suspen-sion was preincubated for 5 min at room temperature(25°C). Two milliliters of nitrate or nitrite solutionwas injected through the serum stopper to start thereaction. At regular intervals, the flask was inverted,and a sample was removed by syringe. A volume ofargon equal to that of the sample was injected tomaintain overpressure. Cells were removed by filtra-tion through 0.4 yM polycarbonate filters with glassfiber prefilters (Nucleopore Corp., Pleasanton, Calif.).

Nitrite was determined by diazotization (1). Nitratewas determined with Szechrome NAS reagent andSzechrome NB reagent (Polysciences Inc., Warring-ton, Penn.), which form stable diphenylamine andbenzidine complexes, respectively, with nitrate in con-centrated mineral acids. Samples were treated with0.1 volume of 5% (wt/vol) sulfamic acid to removenitrite before nitrate determination. For some experi-ments (see Fig. 1 and 2), nitrate and nitrite weredetermined with a Technicon Autoanalyzer equippedwith a cadmium reduction column.NO and N20 measurements. A sealed, continuous

gas-circulating system (15) connected to a Perkin El-mer 910 gas chromatograph equipped with a PorapakQ column and 'Ni electron capture detector was usedto monitor accumulation and reduction of gaseousintermediates during denitrification. The incubationvessel was a 125-ml flask equipped with a side injectionport. Nitrate or nitrite was added to 50 ml of nutrientbroth in the flask, which then was flushed with argonfor 5 min. A known volume of nitrous oxide was addedand allowed to equilibrate for 20 min. The reactionwas started by adding a volume of cell suspensionsufficient to give a final absorbance at 600 nm of 0.1.NO and N20 concentrations were calculated with adigital integrator; all concentrations were corrected forchanges in volume resulting from removal of gas andliquid samples.

Liquid samples for nitrate and nitrite analyses were

VOL. 42, 1981

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

1076 BETLACH AND TIEDJE

withdrawn from the flask through an 18-gauge spinaltap needle connected to a 1-ml syringe through a one-way valve. Samples were immediately placed in theacidic colorimetric reagents, which stopped denitrifi-cation.Progress curve analysis for estimation of Km

values. The Km values for nitrite and nitrous oxidereduction were estimated by aalyzing progress curvesin which the disappearance of substrate was moni-tored. The integrated Michaelis-Menten equation fora single-substrate irreversible reaction was fitted byan iterative method, using the Newton-Raphson pro-cedure (6, 22) and Wilkinson's procedure (36) to esti-mate the standard error.Oxygen effect on nitrous oxide reduction. The

effect of oxygen on nitrous oxide reduction was inves-tigated by using '3NO3- generated at the MichiganState University sector-focused cyclotron and purifiedof '3N gases as previously described (35). A 5-ml vol-ume of cell suspension (absorbance at 600 nm, 0.2) waspipetted into Hungate-type screw-capped tubes(Bellco Glass, Inc., Vineland, N.J.). The tubes werecapped, and the atmosphere was replaced with helium.Immediately before nitrate addition, a volume of gaswas removed from each tube and replaced with anequal volume of air. One milliliter of a 30-,uM '3Nsolution was injected into the tube, which was thenplaced horizontally on a rotary shaker set at 250 rpm.During incubation, headspace gas was sampled by asyringe and analyzed by a gas chromatograph-propor-tional counter system described previously (35). Ra-dioactive counts in the gas peaks were corrected forbackground and '3N decay, and the peak areas andcentroids were calculated with FASTFIT, a programwritten by R. B. Firestone, based on statistical proce-dures used to analyze nuclear spectra.Mathematical model of denitrification. To see

whether the results we obtained could be accountedfor by simple kinetic processes, we developed a deni-trification model for computer simulation. We as-sumed that denitrification took place according to thesequence: N03 -+ N02 -- NO -* N20 -+ N2. Eachreductive step was modeled by Michaelis-Menten ki-netics:

Vmnax -SiVi =K S (1)

K., + Si

where the subscript i refers to the appropriate con-stants and concentration for each substrate in thesequence. Rather than derive the exact solution forthis four-reaction sequence, we approximated solu-tions by numerical integration, using Euler's method.The model was not intended to predict denitrificationin resting cell cultures but rather to demonstrate howsimple kinetics could lead to the behavior observed.For this reason, we deliberately used dimensionlessunits for substrate concentration and time in the fig-ures presenting results of the simulations.To examine what effect oxygen may have if it is a

general inhibitor of each reductive step in the denitri-fication pathway, we modified the model so that:

Vmax Svi = P Kma+5, (2)

K.1 + Si

In this equation, P represents the fraction of the bulksolution that was anaerobic at any given time. Allreaction rates took place at 100 P percent of the rateunder completely anaerobic conditions. P was heldconstant throughout the simulation for each oxygenconcentration. Computer program listings for this sim-ulation model, the progress curve analysis, and FAST-FIT are included in M. R. Betlach's Ph.D. dissertation,Michigan State University, East Lansing, 1979, D.A.no. 80-06083.

RESULTS

Patterns of nitrite accumulation. The ex-tent of nitrite accumulation in resting cell sus-pensions varied from organism to organism. Fla-vobacterium sp. produced almost undetectablequantities of nitrite during nitrate reduction un-der the conditions used here (Fig. 1). In contrast,P. fluorescens (Fig. 2) and Alcaligenes sp. pro-duced much larger quantities of nitrite.

Nitrite accumulation by P. fluorescens andAlcaligenes sp. might have resulted either fromnitrate inhibition of nitrite reduction or from adifference in the rates of nitrate and nitrite re-duction. Comparison of the rates of nitrate andnitrite reduction for P. fluorescens (Table 1)supported the latter explanation: the rate ofnitrite accumulation observed was not signifi-cantly different from that calculated as the dif-

C

-c-

-J

, CD -

C~D-

z

a:

6-

CD -

CD

LLJ

z

0-

Flavobacteri um

20 40 60

TIME( MIN)FIG. 1. Lack of nitrite accumulation during deni-

trification by Flavobacterium sp. (160 ug of proteinml-'). The initial nitrate concentration was 0.5 mM.

APPL. ENVIRON. MICROBIOL.

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

KINETICS OF BACTERIAL DENITRIFICATION 1077

CDD

II.

z

w-

ILlo

CD

cE cw-0X

crDO

_e'z7

LL C_

1- O

P. fluores

N03

NO

20

TIME ( MIFIG. 2. Pattern of nitrite accum

nitrification by P. fluorescens (160;The initial nitrate concentration u

TABLE 1. Calculated and measuraccumulation during denitrit

fluorescens (160 ig ofprotein mtdata shown in Fig.

Time interProcess for rate es

mate (mir

Nitrate reduction 5-25

Nitrite reduction 30-50

Nitrite accumulationCalculated'Measured 5-25

min-1, respectively), but cells grown with nitritecens and tungstate were unable to reduce nitrate (0.35

+ 0.30 ,uM min-'; rate ± to.o5S&). Nitrate concen-trations as high as 8 mM had no effect on therate of nitrite reduction in Flavobacterium sp.

P. fluorescens (Table 2). Hence, nitrate did notdirectly inhibit nitrite reduction in either orga-nism.

Since nitrite accumulation appeared to be dueto different rates of nitrate and nitrite reduction,

)2 we used a series of computer simulations to testwhether the aforementioned model of denitrifi-cation could generate behavior similar to thatobserved (Fig. 3). The Km values for nitratereductase and nitrite reductase were the same inboth simulations, but the relative maximal rateof reduction, Vmazi, varied. When the Vmax fornitrate reduction was set to one-fifth that fornitrite reduction, little nitrite accumulated (Fig.3a), as had occurred with Flavobacterium sp.(Fig. 1). When the rate of nitrate reduction wasfive times that of nitrite reduction, nitrite accu-

§ | mulated in almost stoichiometric amounts be-40 60 fore being reduced (Fig. 3b), thus following a

I N) pattern similar to that observed for P. fluores-cens (Fig. 2). Thus, the kinetic explanation forulation during de- nitrite accumulation is consistent with the ex-

ig ofprotein ml'). perimental results.

vas 0.5 mM. Nitrate and nitrite inmbition of nitrousdreooxide reduction. To test whether nitrate or

ed rates of nitrite nitrite inhibited nitrous oxide reduction in any

fication by P. of the isolates, monitored reduction of added

'. nitrous oxide in suspensions exposed to differentconcentrations of nitrate or nitrite. The nitrous

tval Rate (AM oxide concentration should have increased ifniimn)a either ion inhibited reduction. Addition of ni-

trate to Flavobacterium suspensions neither af-19.1 ± 2.28 fected the rate of nitrous oxide reduction nor

10.7 ± 0.85 resulted in nitrous oxide accumulation (Fig. 4).

8.4 ± 3.1310.3 ± 1.35

"Rates determined by linear regression; confidencelimits are to.o5sb; 3 df.

'Difference between measured rates of nitrate re-duction and nitrite reduction.

ference between rates of nitrate and nitrite re-duction. Furthermore, if nitrate had inhibitednitrite reduction by acting on the nitrite reduc-tase itself, such inhibition should have beenobserved in cells incapable of nitrate reduction.We obtained such suspensions by growing cul-tures anaerobically in medium containing nitriteand tungstate. Cells grown with nitrite alonereduced nitrate as rapidly as those grown withnitrate (29.0 7.5 ,uM min-' and 29.3 + 4.6 ,uM

TABLE 2. Effects of nitrate concentration on therate of nitrite reduction by Flavobacterium sp. andP. fluorescens lacking nitrate reductase activityaNitrate Nitrite reduction rate (uxM min-Y)concn(mM) Flavobacterium sp. P. fluorescens

0 17.2 ± 1.44 (4) 9.0 ± 0.60 (6)1.43 19.1 ± 3.01 (3) 8.8 ± 0.24 (6)3.57 14.5 ± 1.59 (5) 8.6 ± 0.36 (6)7.14 17.7 ± 2.80 (5) 10.0 ± 0.35 (6)

a Cells were grown in nutrient broth containing 5mM KNO2 and 5 mM Na2WO4. The initial nitriteconcentration was 0.125 mM in the assay system.Flavobacterium sp. had 32 tcg of protein ml-' and P.fluorescens had 160 t,g of protein mn-'.

b Rates were determined by linear regression; con-fidence limits are to.O05sb; numbers in parentheses aredegrees of freedom.

VOL. 42, 1981

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

1078 BETLACH AND TIEDJE

dium at the end of either experiment, indicatingthat both substrates had been reduced to dini-trogen.

CD

CDo

oD3-j

CD

LLJ (DC)

x

CD

oa:~Hn

-0 5 10 15 20 25

T I M EFIG. 3. Pattern of nitrite accumulation as a func-

tion of the rates of nitrate and nitrite reduction in amodel of denitrification. The Km for nitrate reductionwas 1, the Km for nitrite reduction was 2, and theinitial nitrate concentration was 40. The maximumvelocities for nitrate and nitrite reduction were 2 and10, respectively, for (a) and 10 and 2, respectively, for(b).

Inclusion of 0.28 mM nitrite also did not affectthe rate of nitrous oxide reduction (data notshown). Nitrous oxide reduction by the Alcali-genes isolate and P. fluorescens also was notinhibited by addition of nitrate or nitrite atconcentrations similar to those given above. Fur-thermore, the low steady-state concentrations ofN20 (see below) were independent of theamount of nitrate or nitrite added.

Since A. odorans reduces nitrite but not ni-trate (23), it can be considered physiologicallysimilar to cells grown anaerobically in the pres-

ence of tungstate to inhibit nitrate reductase.Neither nitrate nor nitrite affected the rate ofnitrous oxide reduction in this organism (Table3). Net reduction of nitrous oxide began imme-diately after addition of cells to nitrate broth,but not until 25 and 47 min after cells wereadded to broth containing 0.2 and 0.5mM nitrite,respectively. This apparent lag can be attributedto a period during which the rate of nitritereduction to nitrous oxide matched the rate ofnitrous oxide reduction to nitrogen. Neither ni-trite nor nitrous oxide was detected in the me-

Flavobact erium

NOJ3(mM)

0.00 00.36 X1.43 03.57 +

20 40 60I0

TIME( MIN)FIG. 4. Failure of various nitrate concentrations

to inhibit nitrous oxide reduction by Flavobacteriumsp. (32 jig ofprotein ml-'). The initial nitrous oxideamount was 820 nmol of N20-N (100ppm, vol/vol, ingas phase). In this and the following figures, amountsof nitrogenous oxides as Nper flask are used ratherthan concentrations in solution to simplify compari-son between gaseous and ionic species.

TABLE 3. Effect of nitrate concentration and nitriteconcentration on the rate of nitrous oxide reduction

by A. odoransaFirst-order rate constant, kConcn (mM) (min- 1)b

Nitrate0 0.147 ± 0.0201 0.153 ± 0.04410 0.169 ± 0.025100 0.158 ± 0.016

Nitrite0 0.143 ± 0.0190.2 0.154 ± 0.0360.5 0.149 ± 0.020

a The initial nitrous oxide concentration was ap-proximately 100 ppm (100,ul l-'), and the cell concen-tration was 32 jig of protein ml-'.

b Rate constants were determined by linear regres-sion on the transformation: ln(N20), = ln(N20)0- kt;confidence limits are to.o5sb; 3 df.

CD-"1-

CD

Hy-c:CD-

LUJuCD

uCD

CDo-

:zCD

---

Hz CD-

LLJC.)

CD

CD

APPL. ENVIRON. MICROBIOL.

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

KINETICS OF BACTERIAL DENITRIFICATION

Nitric oxide and nitrous oxide accumu-lation. When nitrate or nitrite was included inthe medium along with 820 nmol of N20, eachisolate still reduced the nitrous oxide concentra-tion to a constant value. A low nitrous oxideconcentration was maintained until all ionic spe-cies had been depleted (Fig. 5 and 6). Nitrousoxide added after the steady state had becomeestablished also was reduced rapidly until itsconcentration reached the previous value (Fig.6). Maintenance of a relatively constant N20concentration even after perturbation was evi-dence that the steps in denitrification were inkinetic equilibrium.

All three organisms accumulated nitric oxideshortly after cell suspensions had been added toincubation vessels containing nitrate or nitrite.The concentration of nitric oxide also becameconstant until all the nitrate or nitrite had beenused. No nitric oxide was detected in nitritebroth before addition of the cell suspensions, nordid the amount produced correlate with theamount of nitrite present (cf. Fig. 5 and 6).Hence, nitric oxide production could not be at-tributed to chemical decomposition of nitrite.The transient appearance of low amounts of

nitric oxide and nitrous oxide during denitrifi-

CDo) a

L i Flavobacterium

U BvNOCD

CDU)

NO2CD

CD- N 0 A

-J-

d: 2CDz 5

0 50 100 1 50TIME ( MIN )

FIG. 5. Pattern of nitrite and gas accumulationduring denitrification by Flavobacterium sp. (32 Itgofprotein ml-'). The initial amounts were 86 ,umol ofKNO3 (1.72 mM) and 820 nmol of N20-N (N20, 120ppm [vol/vol]).

C)C\J

Co-J

CD

CD

C-)

CD

coLUJ

o -

C)-

I0

P. fluorescens

50 100TIME (MIN)

FIG. 6. Pattern of nitrite and gas accumulationduring denitrification by P. fluorescens (32 jig ofprotein ml-'). The figure shows the return of nitrousoxide to a steady-state value after perturbation. Ini-tial amounts were 20.5 pnol ofKNO3 (0.40 mM) and820 nmol of N20-N. At 95 min (arrow), 410 nmol ofN20-N was added to the flask.

cation and the rapid return of nitrous oxide tosteady-state amounts after perturbation sug-gested again that simple kinetics could haveaccounted for the behavior observed. Additionof these two steps to the denitrification modelresulted in the simulation results presented inFig. 7. The high initial N20 concentration wasrapidly reduced to a steady-state value, whichwas also reestablished after more N20 had beenadded. Such relaxation kinetics were similar tothe pattern observed for P. fluorescens (Fig. 6).In the simulation, nitric oxide also appeared asan intermediate which was maintained at rela-tively constant amounts until the more-oxidizedcompounds had been depleted. The qualitativebehavior ofNO and N20 in this mode was similarto the experimental results obtained for all fourbacterial species.

Effect of oxygen on nitrous oxide accu-mulation. We used '3NO3 to investigate theeffect of oxygen on nitrous oxide production andconsumption. At higher oxygen concentrations,all three isolates produced a higher percentageof nitrous oxide in the gases released than thatproduced at lowered oxygen concentrations (Fig.

VOL. 42, 1981 1079

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

1080 BETLACH AND TIEDJE

2:CD

NX-H- N0° fgal:

Li5CD

CD

- N02

^I N2~~~N0 added b

2:~~~~~~CDo- k N20

5 10 15 20 25

CE~ ~ ~ TM

tial N03 concentration was 40; the initial N2O-Nconcentration was 4. At the point indicated by thearrow, the N20-N concentration was increased by 3.For reduction ofN03-, N02-, NO, and N20-N, the Kmvalues were 1, 2, 2, and 2, respectively; the Vm..c valueswere 2, 4, 8, and 6, respectively.

8a and 9a; data for Alcaligenes sp. not shown).After additional incubation, cell suspensions hadreduced percentages of nitrous oxide, yet therelationship between the percentage of nitrousoxide and oxygen concentration remained thesame (Fig. 9; 20 min). Higher oxygen concentra-tions inhibited denitrification by P. fluorescens(Fig. 8b) and Alcaligenes sp. However, no de-crease in total gas production was observed inseveral experiments with Flavobacterium sp.(Fig. 9b). Nitrous oxide reduction in this orga-nism appeared to be more sensitive to oxygenthan were the other steps in denitrification.Could general inhibition of denitrification by

oxygen have caused increased production of ni-trous oxide relative to dinitrogen as oxygen ten-sion increased, or was nitrous oxide reductionmore sensitive to oxygen than the other reduc-tive steps? As a preliminary approach to solvingthis problem, we modified the denitrificationmodel to include a term for oxygen inhibition ofeach step in the pathway, as described in equa-

tion 2. The contribution of N20 to total gasproduction was calculated at various times foreach of the four values for P. As the systembecame more aerobic the percentage of N20produced increased (Fig. lOa), as observed forall three isolates. In addition, the time-depend-ent decrease in N20 contribution at differentoxygen tensions resembled the behavior ob-served in the bacterial cultures (Fig. 9a). Asexpected, the total amount of gas produced de-creased at higher oxygen tensions (Fig. lOb).Km values for nitrate, nitrite, and nitrous

oxide reduction. Km values for nitrite reduc-tion were determined by analysis of progresscurves. The values for Alcaligenes sp., Flavo-bacterium sp., and P. fluorescens were similar:12.9 ± 2.6, 5.6 + 1.6, and 5.5 ± 1.1 jiM(Km ± standard error), respectively.

It was not possible to obtain Km estimnates fornitrate reduction directly, owing to the limitedsensitivity and precision of the colorimetric pro-cedures used. Since rates of nitrate reductionobserved at concentrations greater than 15 ,uMwere linear, the Km values for nitrate reductionought to have been at least as low as those fornitrite reduction. Recent results of Edwards andTiedje confirm that nitrate Km values are in the

0(0)

.-10

0

0C\'z

0CY)

0

F-z:D0

(0)0v-

(0

CY)

00 5 10

OXYGEN (°/o)FIG. 8. Effects of increasing oxygen concentration

on composition and amount ofgases produced by P.fluorescens (65 yg ofprotein ml-') 10 min after addi-tion of '3NO3 . The initial N03- concentration was 5pLM. (a) Percentage ofgaseous products (N20 and N2)as N20; (b) total 13N in N20 and N2 per 1-ml sample.

APPL. ENVIRON. MICROBIOL.

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

KINETICS OF BACTERIAL DENITRIFICATION 1081

0-

-00

0C\7Z

(I)H-zD0

0

0

0

Q

I Flavobacteriuma /

10 min

20 min

_ b ° 20 min

I10 min

0 5

OXYGEN (o/%)FIG. 9. Effects of increasing oxygen concentration

on composition and amount of gases produced byFlavobacterium sp. (65 jAg ofprotein ml-) 10 and 20min after addition of 3NO3-. The initial N03 con-centration was 5pM. (a) Percentage ofgaseous prod-ucts (N20 and N2) as N20; (b) total '3N in N20 andN2 per 1-ml sample.

range of 5 to 10 AM (R. M. Edwards and J. M.Tiedje, Abstr. Annu. Meet. Am. Soc. Microbiol.1981, N48, p. 181).The estimated Km value for nitrous oxide re-

duction by Flavobacterium sp. was 440 + 11 nM(1.92 + 0.048 Pa). This value was determinedfrom experiments starting at several initial par-tial pressures of nitrous oxide. Data from a typ-ical experiment are shown in Fig. 11. To ensurethat the biological rate of N20 reduction wasmeasured rather than the rate of N20 transferfrom the gas to the liquid phase, we increasedthe transfer rate by sparging gas through thecontinously stirred medium. In addition, the rateof N20 reduction was reduced by decreasing celldensity to an absorbance of 0.005 and the liquidvolume to 25 ml.

If phase transfer had limited the initial rate ofnitrous oxide reduction, we should have ob-tained a biphasic exponential curve: the firstportion of the curve would represent the rate oftransfer into the liquid phase, and the secondportion would represent the rate of biologicalreduction when transfer was sufficiently rapid tosupport the reduction rate. If the biological rateof reduction had always been greater than the

rate of gas transfer, there should have been onlyone exponential curve, reflecting diffusion ofN20 into the aqueous phase. Also, if the iterativeprocedure for the progress curve analysis wouldconverge in the latter case, the Km value ob-tained should have been at least as high as theinitial nitrous oxide concentration. We did notobserve a single or a biphasic exponential curve(Fig. 11), nor did we obtain a Km value as highas the initial N20 concentration. Thus we areconfident that only the biological process wasmeasured in these experiments.

DISCUSSIONThe results presented above permit us to rule

out direct inhibition of nitrite reduction by ni-trate as a mechanism of nitrite accumulation:the rates of nitrite reduction by P. fluorescenswere similar in the presence and absence ofnitrate (Table 1), and nitrate did not inhibitnitrite reduction in tungstate-grown bacterialacking a functional nitrate reductase (Table 3).

1 Apparently, different rates of reduction of theintermediates of denitrification caused their ac-cumulation, rather than any specific inhibitorymechanism. The kinetic model of denitrification

LO_Cl

I

0

0N

z

z+0

z

U)cli

CDCD

CD

a) Model / 5 min

10min

_ b)

10 min_W T 1 ] |~~5 m

0.2 0.4 0.6 0.8 1.0OI.0P( REROBIC)

FIG. 10. Effects of increasing oxygen concentra-tion on composition and amount of gases producedin a model of denitrification 5 and 10 min after N03addition. The initial N03- concentration was 40. Therate constants are as specified for Fig. 7. (a) Percent-age of gaseous products (20 and N2) as N20; (b)total amount ofN20 and N2 produced.

VOL. 42, 1981

awll-

"I

E

M

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

CELL SUSPENSION ADDED be the actual inhibitor of N20 reduction, sinceN20 accumulation is greater with nitrite thanwith nitrate. However, the experiments reportedhere showed that neither nitrate nor nitrite atconcentrations comparable to those used byFirestone et al. (8) had any effect on the rate ofnitrous oxide reduction in pure cultures of den-itrifiers. The explanation for the discrepancybetween our results and the previous results forsoils may be due to one or both of the following.We used cultures harvested in late exponentialphase of growth which probably had more N20reductase relative to the previous enzymes inthe sequence than is the case in soils (9). Sec-ondly, the more-oxidized N-oxides may be pref-erentially reduced when the supply of electrondonor is limited. Kohl et al. (18) and Yoshinariet al. (38) found that addition of glucose to soilincreased the denitrification rate and also af-fected the Km estimated for denitrification. Useofmore-oxidized electron acceptors when carbonwas limiting could have prevented reduction of

-,, g | , | nitrous oxide, an effect difficult to distinguish

_ d20O 0 2 ° 4 ° 6 0 from direct inhibition of the nitrous oxide reduc-tase. Use of a rich medium in the experiments

T I ME (M I N) reported here prevented such carbon limitation.11. Progress curve used to determine the Km As noted, carbon availability could affect the

for nitrous oxide reduction by Flavobacterium estimation ofKm values for denitrification in soil.2 concentration in the gas phase at start of Phillips et al. (26) and Reddy et al. (28) recentlyation was 72ppm (vol/vol), and cell concentra- demonstrated that the physical structure of soilfter addition (arrow) was 1.6,tug ofprotein ml-1. also may influence the kinetics of denitrification.

They concluded that the typical procedureition 1) was constructed to demonstrate which uses flooded soil to determine the rate ofsimple kinetics may cause the patterns of nitrate reduction actually measures the rate ofe accumulation observed. Varying the rel- nitrate diffusion from the overlying solution intorates of nitrate and nitrite reduction in the the soil matrix. Such mass transfer resistance1 determined the extent of nitrite accumu- would result in apparent first-order denitrifica-i (Fig. 3). In the bacterial cultures, the tion rates at nitrate concentrations higher thanmal velocity of nitrate or nitrite reduction those required to saturate enzymes in a systemi have depended not only on substrate con- not limited by diffusion and consequently wouldation but also on the number of reductase give higher Km estimates (33). Both carbon lim-cules present. Variations in the extent of itation and effects of diffusion may explain whye accumulation by P. fluorescens-com- Km values for denitrification in soils: 0.23 mMFig. 2 and 6-suggest that the relative (17), 0.29 and 3.5 mM (18), and 0.13 and 1.2 mMnts of nitrate and nitrite reductase were (38), are so much higher than the values weonstant. Differences may reflect both the obtained for nitrate and nitrite reduction byat which the cultures were harvested rela- pure cultures, which were <0.015 mM.;o the time they became anaerobic as well Diffusion limitations may also have affectedrresponding lags in synthesis of nitrite re- measurements of the Km value for nitrous oxidese relative to nitrate reductase, as Williams reduction. The value we obtained for Flavobac-reported (37). terium sp., 0.5 tIM, was much smaller than thosetrous oxide frequently accumulates in soils reported previously: 5 ,uM for a lysate of Para-lg experimental studies of denitrification. coccus denitrificans (19), 30 to 60 ,uM for Pseu-ratio of N20 to N2 may be increased by domonas denitrificans (21), less than 100 ,uMIg nitrate to soils (2, 8, 29, 30), suggesting for P. aeruginosa (31), and 0.7 to 1.0 mM for anitrate specifically inhibits nitrous oxide soil (38). We had to use very low cell concentra-tion. Firestone et al. (8) postulated that tions and rapid mixing to ensure that our Kme produced during nitrate reduction may measurement was not affected by diffusion of

D

cSJ

(D

-

CJcl-l:z

FIGvaluesp. N2incubction a)

(equahow Enitrit4ativemodelationmaxiIwoulc

centrimolenitritpareamounot c

time -

tive tas co]ductaet al.

NitdurinThe i

addinthat:reducnitrit4

1082 BETLACH AND TIEDJE APPL. ENVIRON. MICROBIOL.

OD

CD

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

KINETICS OF BACTERIAL DENITRIFICATION 1083

N20 across the gas-liquid interface. Thoughlower than previous estimates, the Km value forFlavobacterium sp. was still 50 times higherthan the N20 concentration in solutions equili-brated with the atmosphere.

In the three isolates we used, the concentra-tion of nitrous oxide appeared to be under ki-netic control. Such control was especially evi-dent when the nitrous oxide concentration rap-idly returned to a steady-state value after addi-tion of more N20 (Fig. 6). The slight accumula-tion during denitrification of nitric oxide sug-gested that it, like nitrite and nitrous oxide, wasalso under kinetic control. The increase in NOafter addition of the substrate and its depletionafter nitrate and nitrite were consumed was thepattern expected ifNO had been an intermediateor in equilibrium with an enzyme-bound inter-mediate in the denitrification pathway. Use ofan electron capture detector and a system de-signed to exclude oxygen may account for ourbeing able to readily detect NO during denitri-fication and other investigators not being able todo so. The kinetic model of denitrification rep-resented by equation 1 could account for thetransient appearance of nitric oxide and nitrousoxide during denitrification as well as the accu-mulation of nitrite (Fig. 7).

Increased oxygen in soils enhances the ratioof N20 to N2 produced during denitrification.However, oxygen also inhibits denitrificationand represses synthesis of the denitrifying en-zymes (5, 14, 24). Such general inhibition wasapparent in the decrease in total gas productionby Alcaligenes sp. and P. fluorescens exposedto increasing oxygen concentrations (Fig. 8). Gasproduction by Flavobacterium was not inhibitedby oxygen (Fig. 9), although the proportion ofgas as N20 increased with increased oxygen.Focht (10) developed a zero-order kinetic modelof denitrification based on data from Nommik(23), in which increased production of N20 athigher soil aeration results from oxygen inhibi-tion ofN20 reduction greater than that of nitratereduction. Since gas production by Flavobacter-ium sp. was not decreased by oxygen, this orga-nism appears to correspond to Focht's model.Nitrous oxide reductases of the other two orga-nisms might also have been more sensitive tooxygen than were the enzymes catalyzing nitrousoxide formation, but the decrease in gas produc-tion at increased oxygen concentrations sug-gested that general inhibition of denitrificationby oxygen may cause similar behavior. Resultsof the model in which a general oxygen inhibi-tion term, P, was included gave results (Fig. 10)qualitatively similar to those obtained for P.fluorescens (Fig. 8).

An untested prediction of the model repre-sented by equation 2 is that the percentage ofnitrite relative to the more-reduced nitrogencompounds should also increase with increasedoxygen concentration. Furthermore, since themathematics of the model would remain un-changed, any factor which inhibited all steps indenitrification equally should elicit a responsesimilar to that observed with increased oxygenconcentration. Recently, Keeney et al. (16) re-ported that soils incubated at 5 and 15°C pro-duced a higher proportion of gas as N20 ratherthan N2 when compared with soils incubated at25°C. Total gas production was lower at thelower temperatures. Data presented by Bollagand Kurek (4) showed that the effects of pesti-cide metabolites on denitrification were similarto those of decreased temperature: in the pres-ence of pesticide metabolites, more nitrite andnitrous oxide are produced during denitrifica-tion, which takes place at a reduced rate com-pared with that of unamended controls. Thetime course data presented in both papers showthat the effects of temperature and pesticidemetabolites are less pronounced with longer in-cubation, as we found for the oxygen effect onN20, both by the bacteria (Fig. 9) and in themodel (Fig. 10).The kinetic model presented in this paper has

the advantage ofbeing simple without sacrificingany power to unify disparate observations re-garding accumulation of intermediates duringdenitrification. We developed the model basedon work with pure cultures, and we were able touse it for isolating the physiology of the denitri-fiers from the limitations imposed upon them bycomplex environments. There is no less a needto investigate how the physicochemical environ-ment modifies this physiological potential (e.g.,by controlling substrate availability through lim-ited diffusion). We do not intend the hypothesisthat accumulation of intermediates reflects dif-ferent rates of reduction to obscure phenotypicdifferences among denitrifying bacteria (e.g., 02on Flavobacterium spp.), nor is it meant todiscount the possibility of specific inhibition insome cases. However, we do feel that the kineticmodel can be a useful guide in research on thephysiology and ecology of denitrification.

ACKNOWLEDGMENTSWe thank Henry Kaspar, Mary Firestone, and Richard

Firestone for their cooperation and advice on this work.This work was supported by grants DEB-77-1923 and PHY-

79-24017 from the National Science Foundation and U.S.Department of Agriculture Regional Research project NE-39.

ADDENDUM IN PROOFThe isolate termed Alcaligenes in this paper has

VOL. 42, 1981

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from

1084 BETLACH AND TIEDJE

recently been classified as Pseudomonas alcaligenesby the American Type Culture Collection. We did notoriginally consider this identification, since P. alca-ligenes is reported not to denitrify in Bergey's Manual(8th ed., p. 219).

LITERATURE CITED

1. Black, C. A. 1965. Methods of soil analysis. Part 2,Chemical and microbiological methods, p. 1179-1237.American Society of Agronomy, Inc., Madison, Wisc.

2. Blackmer, A. M., and J. M. Bremner. 1978. Inhibitoryeffect of nitrate on reduction of N20 to N2 by soilmicroorganisms. Soil Biol. Biochem. 10:187-191.

3. Blackmer, A. M., J. M. Bremner, and E. L. Schmidt.1980. Production of nitrous oxide by ammonia-oxidizingchemoautotrophic microorganisms in soil. Appl. Envi-ron. Microbiol. 40:1060-1066.

4. Bollag, J. M., and E. J. Kurek. 1980. Nitrite and nitrousoxide accumulation during denitrification in the pres-ence of pesticide derivatives. Appl. Environ. Microbiol.39:845-849.

5. Delwiche, C. C., and B. A. Bryan. 1976. Denitrification.Annu. Rev. Microbiol. 30:241-262.

6. Duggleby, R. G., and J. F. Morrison. 1977. The analysisof progress curves for enzyme-catalyzed reactions bynon-linear regression. Biochim. Biophys. Acta 481:297-312.

7. Firestone, M. K., R. B. Firestone, and J. M. Tiedje.1979. Nitric oxide as an intermediate in denitrification:evidence from nitrogen-13 isotope exchange. Biochem.Biophys. Res. Commun. 91:10-16.

8. Firestone, M. K., M. S. Smith, R. B. Firestone, and J.M. Tiedje. 1979. The influence of nitrate, nitrite, andoxygen on the composition of the gaseous products ofdenitrification in soil. Soil Sci. Soc. Am. J. 43:1140-1144.

9. Firestone, M. K., and J. M. Tiedje. 1979. Temporalchange in nitrous oxide and dinitrogen from denitrifi-cation following onset of anaerobiosis. Appl. Environ.Microbiol. 38:673-679.

10. Focht, D. D. 1974. The effect of temperature, pH, andaeration on the production of nitrous oxide and gaseousnitrogen-a zero-order kinetic model. Soil Sci. 118:173-179.

11. Freney, J. R., 0. T. Denmead, and J. R. Simpson.1978. Soil as a source or sink for atmospheric nitrousoxide. Nature (London) 273:530-532.

12. Freney, J. R., 0. T. Denmead, and J. R. Simpson.1979. Nitrous oxide emission from soils at low moisturecontents. Soil Biol. Biochem. 11:167-173.

13. Gamble, T. N., M. R. Betlach, and J. M. Tiedje. 1978.Commonality of numerically dominant denitrifying bac-teria from world soils. Appl. Environ. Microbiol. 33:926-939.

14. Garcia, J. L. 1975. La denitrification dans les sols. Bull.Inst. Pasteur Paris 73:167-193.

15. Kaspar, H. F., and J. M. Tiedje. 1980. Response ofelectron-capture detector to hydrogen, oxygen, nitro-gen, carbon dioxide, nitric oxide and nitrous oxide. J.Chromatogr. 193:142-147.

16. Keeney, D. R., I. R. Fillery, and G. P. Marx. 1979.Effect of temperature on the gaseous nitrogen productsof denitrification in a silt loam soil. Soil Sci. Soc. Am.J. 43:1124-1128.

17. Klemedtsson, L., B. H. Svensson, T. Lindberg, andT. Rosswall. 1977. The use of acetylene inhibition ofnitrous oxide reductase in quantifying denitrification insoils. Swed. J. Agric. Res. 7:179-185.

18. Kohl, D. H., F. Vithayathil, P. Whitlow, G. Shearer,and S. H. Chien. 1976. Denitrification kinetics in soilsystems: the significance of good fits of data to mathe-

APPL. ENVIRON. MICROBIOL.

matical forms. Soil Sci. Soc. Am. J. 40:249-253.19. Kristjansson, J. K., and T. C. Hollocher. 1980. First

practical assay for soluble nitrous oxide reductase ofdenitrifying bacteria and a partial kinetic characteriza-tion. J. Biol. Chem. 255:704-707.

20. Lam, Y., and D. J. D. Nicholas. 1969. Aerobic andanaerobic respiration in Micrococcus denitrificans.Biochirn. Biophys. Acta 172:450-461.

21. Matsubara, T., and T. Mori. 1968. Studies on denitrifi-cation. IX. Nitrous oxide, its production and reductionto nitrogen. J. Biochem. (Tokyo) 64:863-871.

22. Nimmo, I. A., and G. L. Atkins. 1974. A comparison oftwo methods for fitting the integrated Michaelis-Men-ten equation. Biochem. J. 141:913-914.

23. Nommik, H. 1956. Investigations on denitrification in soil.Acta Agric. Scand. 6:195-228.

24. Payne, W. J. 1973. Reduction of nitrogenous oxides bymicroorganisms. Bacteriol. Rev. 37:409-452.

25. Payne, W. J., and P. S. Riley. 1969. Suppression bynitrate of enzymatic reduction of nitric oxide. Proc. Soc.Exp. Biol. Med. 132:258-260.

26. Phillips, R. E., K. R. Reddy, and W. H. Patrick, Jr.1978. The role of nitrate diffusion in determining theorder and rate of denitrification in flooded soil. II.Theoretical analysis and interpretation. Soil Sci. Soc.Am. J. 42:272-278.

27. Pichinoty, F., M. Veron, M. Mandel, M. Durand, C.Job, and J. L. Garcia. 1978. Etude physiologique ettaxonomique du genre Alcaligenes: A. denitrificans, A.odorans et A. faecalis. Can. J. Microbiol. 24:743-753.

28. Reddy, K. R., W. H. Patrick, Jr., and R. E. Phillips.1978. The role of nitrate diffusion in determining theorder and rate of denitrification in flooded soil. I. Ex-perimental results. Soil Sci. Soc. Am. J. 42:268-272.

29. Ryden, J. C., L. J. Lund, and D. D. Focht. 1979. Directmeasurement of denitrification loss from soils. I. Labo-ratory evaluation ofacetylene inhibition ofnitrous oxidereduction. Soil Sci. Soc. Am. J. 43:104-110.

30. Ryden, J. C., L. J. Lund, J. Letey, and D. D. Focht.1979. Direct measurement of denitrification loss fromsoils. II. Development and application of field methods.Soil Sci. Soc. Am. J. 43:110-118.

31. St. John, R. T., and T. C. Hollocher. 1977. Nitrogen 15tracer studies on the pathway of denitrification in Pseu-domonas aeruginosa. J. Biol. Chem. 252:212-218.

32. Scott, R. H., G. T. Sperl, and J. A. DeMoss. 1979. Invitro incorporation of molybdate into demolybdopro-teins in Escherichia coli. J. Bacteriol. 137:719-726.

33. Shieh, W. K. 1979. Theoretical analysis of the effect ofmass-transfer resistances on the Lineweaver-Burk plot.Biotechnol. Bioeng. 21:302-304.

34. Swain, H. M., H. J. Somerville, and J. A. Cole. 1978.Denitrification during growth of Pseudomonas aerugi-nosa on octane. J. Gen. Microbiol. 107:103-112.

35. Tiedje, J. M., R. B. Firestone, M. K. Firestone, M. R.Betlach, M. S. Smith, and W. H. Caskey. 1979.Methods for the production and use of nitrogen-13 instudies of denitrification. Soil Sci. Soc. Am. J. 43:708-715.

36. Wilkinson, N. G. 1961. Statistical estimations in enzymekinetics. Biochem. J. 80:324-332.

37. Williams, D. R., J. J. Rowe, P. Romero, and R. G.Eagon. 1978. Denitrifying Pseudomonas aeruginosa:some parameters of growth and active transport. Appl.Environ. Microbiol. 367:257-263.

38. Yoshinari, T., R. Hynes, and R. Knowles. 1976. Acet-ylene inhibition of nitrous oxide reduction and meas-urement of denitrification and nitrogen fixation in soil.Soil Biol. Biochem. 8:1-7.

39. Zumft, W. G., and J. Cardenas. 1979. The inorganicbiochemistry of nitrogen bioenergetic processes. Natur-wissenschaften 667:81-88.

on October 16, 2020 by guest

http://aem.asm

.org/D

ownloaded from