Catalytic mechanism and protein engineering of copper-containing nitrite reductase

(1)

nitrite reductase

Wijma, Hein Jakob

Citation

Wijma, H. J. (2006, February 9). Catalytic mechanism and protein engineering of

copper-containing nitrite reductase. Retrieved from https://hdl.handle.net/1887/4302

Version:

Corrected Publisher’s Version

License:

Licence agreement concerning inclusion of doctoral thesis in the

_{Institutional Repository of the University of Leiden}

Downloaded from:

https://hdl.handle.net/1887/4302

(2)

Chapter

4

A

Random-Sequenti

al

M echani

sm

for

Ni

tri

te

Bi

ndi

ng

and

Acti

ve

Si

te

Reducti

on

i

n

Copper-Contai

ni

ng

Ni

tri

te

This chapter is to be submitted.

Abstract

Copper-containing nitrite reductase contains per monomer a type-1 and a type-2 Cu-site. Electrons enter through type-1 and are shuttled to the type-2 active site where they are used for the enzymatic conversion of nitrite into nitric oxide. The enzyme kinetics have been studied at saturating electron donor concentrations. The kinetics as a function of pH and [NO2-] could be understood assuming thatthe free (i.e. the fourth) coordination position

at the type-2 copper can be occupied by water, hydroxide, or nitrite, and that the rate of internalelectron transfer (from type-1 to type-2) decreases according to H2O > NO2-> OH-.

Depending on pH,therefore,binding of nitrite willslow down or enhance internalelectron transfer leading to substrate inhibition or activation: at low [NO2-] reduction of the type-2

site precedes nitrite binding, at high [NO2-] the reverse occurs. Consequently, the st

(3)

Introduction

Copper-containing nitrite reductase (NiR) is one of the enzymes of the denitrification pathway (56). Denitrification globally recycles fixed nitrogen (NO3-, NO2-)

to the atmosphere (N2) with NO and N2O being formed as byproducts; they act as ozon

scavenger and as greenhouse gas, respectively (57, 58). Cu-containing nitrite reductases are found in bacteria, archaea, and fungi (56, 59, 129). In pathogens, Cu-containing NiR is known to enhance the resistance against human sera in Neisseria gonorrhoeae (62) and allows Neisseria meningitidis to respire on nitrite under the microaerobic conditions encountered during disease in man (64). Furthermore, there is an interest in applying NiR in amperometric biosensors to selectively monitor toxic nitrite in natural waters and waste streams (65-67).

NiR is a homotrimer, in which each subunit contains a type-1 copper site that accepts electrons from a physiological electron donor and transfers them to a type-2 copper site that is located deeper inside the enzyme (77, 78, 81). The type-2 copper is coordinated by 3 histidines and forms the active site together with a water network, an aspartate, and a histidine. The latter two residues hydrogen bond to the nitrite and donate protons for the enzymatic reaction (83-87). In the absence of nitrite a water molecule occupies the fourth ligand position. Deprotonation of this water molecule occurs between pH 6 and pH 7 (87) and results in OH- as the fourth ligand (135). NiR catalyzes the reduction of nitrite bidirectionally (NO2- + e- + 2H+ NO + H2O) (chapter 3).

An important mechanistic question currently under debate (56) is whether nitrite binds to the oxidised type-2 Cu first after which the electron is transferred to the type-2 site (54, 83, 129, 131) or whether the type-2 site is first reduced after which the nitrite binds to the Cu (69, 82, 126, 170). Also the possibility has been considered that the two routes operate in parallel (69, 126). Key observations indicating that nitrite binds first to the oxidized Cu are that the reduced type-2 site in NiR prefers a low occupancy of the fourth coordination position (54, 83, 131) and that the pre-reduced NiR only sluggishly reacts with nitrite (54, 130). On the other hand, in model complexes for the type-2 site, the copper is reduced prior to nitrite binding (127, 128, 171) and the nitrite dissociation constant of the oxidised type-2 site in NiR is more than ten-fold higher (82) than its Michaelis constant (KM).

(4)

to type-2 site is rate-limiting. Below pH 6.5, the catalytic activity diminished at higher nitrite concentrations in agreement with electron transfer being slower to the nitrite bound type-2 site than to the water bound type-2 site. Above pH 6.5, substrate-activation like kinetics were observed, in agreement with electron transfer to the nitrite bound type-2 site being faster than electron transfer to the hydroxyl bound type-2 site. W hen NiR was immobilised on a rotating disk electrode, minimal background activity was observed, which allowed for accurate measurements of the catalytic activity versus nitrite concentration over a wide pH range. To study the effect of slower electron transfer between type-1 and type-2 site, NiR M150T was used, which has a type-1 site with a 125 mV higher midpoint potential and a 0.3 eV higher reorganization energy (unpublished results)). These properties slow down electron transfer to the type-2 site 50-fold. The conclusion is that NiR employs both pathways (i.e., a random-sequential mechanism), the relative importance of each of them depending on pH and nitrite concentration.

Materials and Methods

Material

W ild-type NiR (from Alcaligenes faecalis S-6) and NiR M150T were prepared as described in chapter 3 and 7. The Cu-content of native NiR and NiR M150T used for electrode experiments was 1.7 Cu per monomer, as determined by the bicinchoninic acid method (148). For assays in solution, native NiR with a Cu-content of 2.0 per monomer was used.

Activity Assays in Solution

Unless reported otherwise, the buffer was 25 mM malate-MES-HEPES (MMH). The pH was adjusted with NaOH. Reported pH values are those of the assay buffer and were determined at the temperature of the assay. Nitrite reductase activity assays using saturating concentrations of reduced pseudoazurin (400 µM >> KM) as the electron donor

were carried out as described (chapter 3) at 25 °C. The activity was measured by monitoring the increase of the absorption at 590 nm, typical of oxidized pseudoazurin (ε590nm = 2900 M-1 cm-1 (76)). A linear increase of absorption in time (providing real initial

rates) was observed in all cases.

(5)

(172). The catalytic activity was calculated from the decline in absorption of reduced benzyl viologen (ε603nm = 14,500 M-1 cm-1 (173)) after addition of NiR, corrected for

baseline drift. The decrease in absorption was linear in time after addition of NiR, and stayed virtually linear until > 95 % of the reduced viologen had been oxidised.

The activity with benzyl viologen as an electron donor was larger than the activity determined from the loss of nitrite (results not shown). The reason for this discrepancy is the non-enzymatic reduction of the produced nitric oxide to nitrous oxide or ammonia by the viologen (56). In some assays also methyl viologen can be used instead of benzyl viologen. The activities with a viologen as the electron donor, as quantified by nitrite disappearance, are similar to the activities with pseudoazurin as the electron donor (for Alcaligenes faecalis S-6 NiR at pH 7 the kcat of nitrite reduction is 338 s-1 with methyl

viologen (chapter 6) and 392 s-1 with pseudoazurin (chapter 3) as the electron donor).

Activity assays on a rotating disk electrode

Activity assays employing an pyrolytic graphite edge (PGE) electrode were carried out in an anaerobic cell (Metrohm) provided with an AUTOLAB rotating disk electrode (Eco Chemie), a platinum counter electrode, and an isothermal calomel electrode. During the experiments, the cell was kept anaerobic by flushing the head space with argon. The PGE electrode was home-built and had a surface area of 4 mm2. The epoxy (Araldite CY1300:HY1300 mixed 3:1 w/w) was from Ciba/Robnor. The PGE was supplied by GE Advanced Ceramics. For NiR immobilisation, the PGE electrode was polished thoroughly on a polishing cloth (Buehler) covered with 6 micron diamond, sonicated in H2O, washed

with buffer, and exposed to NiR (> 100 µM) for approximately 30 seconds, rinsed with H2O, and inserted in the buffer. After sealing the cell, the electrode was cycled between 560

and -140 mV versus NHE (Normal Hydrogen Electrode) to check that oxygen was absent. Observed currents were corrected for background currents observed in the absence of nitrite.

(6)

For the reliable determination of enzymatic constants by PFV, the rotation rates should be sufficiently fast to prevent that the transport of substrate to the electrode or the build-up of product near the electrode becomes rate-limiting (164). Varying the rate of rotation between 500 and 9000 rpm did not influence the catalytic current, also not at the lowest nitrite concentrations. Thus, it appears that the catalytically active coverage of NiR on the electrodes is sufficiently low so as not to alter the substrate or product concentration near the surface of the electrode during turn-over. Also the low currents that we observe (approximately 300 nA) are in agreement with a low catalytically active coverage, that was estimated as follows. For assays of NiR immobilised on an electrode, the turnover rate (kt)

is linearly related to the catalytic current (it) by equation 1

it = - FAΓkt (1)

in which A is the surface area of the electrode, Γ is the density of catalytically active enzymes per unit surface , F is the Faraday constant, and a negative sign corresponds with a reduction. With kcat = 1500 s-1 (Figure 1) the enzymatic coverage can be calculated via

(7)

Cu2+ N N N Cu2+ N N N NO2 -k_-1 k₁[S] H2O Cu1+ N N N H2O Cu1+ N N N k2 k_-2 Cu1+ N N N NO2 -k₃ k_-3 k5 k_-5 P k₆ k_-4 k₄[S] Type-1RED Type-1OX Type-1RED Type-1OX B upper route A lower route

Scheme 1. Proposed catalytic mechanism of NiR. The catalytic and productrelease step k6is in factreversible (chapter 3), which is notmodelled since in the currentexperiments the

build-up ofthe productnitric oxide is prevented.The nitrite is depicted as deprotonated butit may also be protonated in the catalytic cycle.

Modeling

The experimental results (vide infra) will be analyzed on the basis of the reactions presented in scheme 1. Depicted are the type-2 site and the various changes it may undergo during turn-over. In the starting configuration the oxidized Cu(II) carries water as the fourth ligand. Replacement of the water (step 1, k1 and k-1) by nitrite and subsequent electron

transfer from the type-1 site to the type-2 site (step 3, k3 and k-3) leads to the reduced nitrite

(8)

via the lower route (step 2 and 4). At higher nitrite concentration, the nitrite binding will become faster than electron transfer to the H2O/OH- bound type-2 site, and catalysis will

follow the upper route (via step 1 and 3).

For understanding the effect of this random-sequential mechanism on the steady-state kinetics, it is instructive to look at scheme 1 under two simplifying conditions. When the reaction occurs either via the upper or the lower route and when substrate binding, reduction, and isomerisation steps are fast relative to the catalytic step 6, it can be shown (175) that the standard Henri-Michaelis-Menten (HMM) behaviour is expected for both routes with the turnover rate is given by

kt(S) = kcat × [S] / (KM + [S]) (2).

Under these conditions, the kcat and the KM depend on the equilibrium constants (Ki=ki/k-i,

Kdox=1/K1, Kdred = 1/K4) and the rate of the catalytic step (k6) according to equation 3 and 4

for the lower route

kcat = k6 (3)

KM = Kdred(1 + K2 + K2K5) / K2 (4)

and according to equation 5 and 6 for the upper route.

kcat = k6K3/(K3 + 1) (5)

KM = Kdox/(K3+1) (6).

These equations show that both routes have Michaelis constants that are proportional to Kdox and Kdred, respectively.

Relaxing the two simplifying conditions mentioned abover, equation 7 obtains (see supplementary materials for derivation and details)

kt(S) = (kcatA[S] + kcatB[S]2/KMB) / (KMA+ [S] + [S]2/KMB) (7).

Equation 7 also applies when the isomerization (step 5) is left out from scheme 1. kcatA and

KMA, and kcatB and KMB are primarily associate with the lower (A) and upper route (B) in

scheme 1, respectively. Furthermore, when step 6 is not rate limiting we find that (see Table S1 in supplementary material; the numbers in this table are examples) the ratio of kcatB/kcatA approximately equals the ratio of the electron transfer rates k3 and k2. Thus, the

random-sequential mechanism is expected to result in a second order dependence of catalytic activity on substrate concentration in which KMA is a measure for Kdred, KMB is a

measure for Kdox, and kcatB/kcatA reports on the ratio of electron transfer rates with nitrite

(9)

Results

Deviations from Henri-Michaelis-Menten Kinetics by NiR

We started by measuring the catalytic activity (kt) of nitrite reductase versus pH

with pseudoazurin as the electron donor at saturating concentrations. The peak of maximum activity was observed around pH 5.6 at 500 µM nitrite while at 5 mM nitrite it occurred at pH 6.0 (Figure 1A). Interestingly, the maximum activity was 25 % higher with 500 µM nitrite than with 5 mM nitrite. To investigate this phenomenon further, we assayed the activity as a function of nitrite concentration. At pH 4.9, the dependence of activity on nitrite concentration showed an initial maximum followed by a steady decrease at higher nitrite concentrations (Figure 1B) that could be fitted to equation 7 resulting in kcatB/kcatA <

0.4.

To investigate the possible dependence of these kinetics on the electron donor benzyl viologen was used instead of pseudoazurin, again at saturating concentrations. Benzyl viologen is a synthetic dye that is a far stronger reductor (EM = -350 mV versus

NHE ) than pseudoazurin (EM = 295 mV at pH 6 (chapter 3)). Furthermore, benzyl viologen

is expected to reduce NiR via a non-specific collisional mechanism while pseudoazurin forms a specific complex with NiR (79, 80, 100); and we wanted to exclude that the complex formation might affect the catalytic properties. Also with benzyl viologen as the electron donor, the catalytic rate diminished at higher nitrite concentrations (Figure 1C).

Finally, in a separate experiment the catalytic activity of NiR immobilized on a rotating disk electrode was measured by observing the reductive current (Figure 2A). The activity displayed the same dependence on nitrite concentration (Figure 2B) as observed before in the bulk experiments, both at electrode potentials of -70 and +110 mV versus NHE. The latter observation shows that the measured kinetics do not depend on the kinetics of the electron transfer between electrode and type-1 site. If the experiment was started at high nitrite concentration, and the solution was diluted, indeed an increase in catalytic current was observed (results not shown). Also the M150T mutant of NiR, the use of which is intended to probe the effect of slower electron transfer from type-1 to type-2 site, displayed a decrease in icat at higher nitrite concentrations (Figure S1 of supplementary

(10)

Figure 1: Dependence of catalytic activity on pH and nitrite concentration

(A) activity versus pH with pseudoazurin as the electron donor. The filled circles are in the presence of 5 mM nitrite, the open circles are in the presence of 500 µM nitrite. The lines merely connect the points, they are not a fit. The buffer was 25 mM Malate-MES-HEPES set at pH with NaOH; the temperature was 25 °C. (B) Activity versus nitrite concentration with pseudoazurin as the electron donor at pH 4.9. The solid line is a fit of the data points to equation 7 yielding kcatA = 2114 ± 426 s-1, kcatB < 690 s-1, KMA = 98 ± 29 µM, KMB = 1.07 ±

0.51 mM. (C) Activity versus nitrite concentration with benzyl viologen as the electron donor at pH 5.4. The solid line is a least-squares fit to equation 7 resulting in kcatB/kcatA= 0.29 ±

(11)

While the kinetic data below pH 6.5 could be satisfactorily analyzed with the use of equation 7, also at pH > 6.5 the latter equation gave a better it of the data than the use of equation 2 which is based on HMM kinetics. This is apparent in figure 3 where especially at higher nitrite concentration the best fits according to equation 2 deviate significantly from the experimental points. To check whether the observed extra activity at high nitrite concentration was not caused by a co-purified contaminant (such as proteolytically degraded enzyme or redox-active metal ions) we used two independent batches of NiR under otherwise identical conditions (Figure 3A). Both batches gave the same deviations from HMM kinetics (Figure 3A) and, when fitted to the HMM equation, produced the same Michaelis constants (47 ± 6 µM and 48 ± 5 µM). This we take as evidence that the deviation from HMM kinetics is characteristic of the intact NiR, and is not caused by a contaminant.

Table 1: Effect of pH on kinetic constants of NiR in solution assays electron donor pH kcatB_/kcatA

KMA_(µM) _KMB_(µM) pseudoazurin 4.9 < 0.41 98 ± 29 1070 ± 510 BV-SDT 4.6 < 0.22 124 ± 22 389 ± 97 BV-SDT 5.4 0.29 ± 0.21 184 ± 37 4470 ± 3510 BV-SDT 6.4 > 1 152 ± 22 > 10000 BV 7.1 > 1 79 ± 12 > 5000

The reported errors are standard deviations. BV-SDT, benzyl viologen reduced in situ by sodium dithionite; BV, reduced benzyl viologen was prepared electrochemically. See Materials and Methods for experimental details.

Figure 2. Kinetic constants measured by protein film voltammetry

(A)Reductive current versus time upon subsequent nitrite additions. After the electrode was kept at -70 mV versus NHE for 200 seconds, the background current was stable and the nitrite concentration was increased stepwise. (B) The baseline-subtracted reductive current versus concentration of added nitrite at pH 5.7. The solid line is a fit to equation 7 and yielded kCATB/kCATA = 0.25 ± 0.02, KMA = 43 ± 2 µM, KMB = 2961 ± 273 µM. The buffer was 25

mM Malate-MES-HEPES at 0 °C. For experimental details see materials and methods.

(12)

NaNO2 (mM) 0 5 10 15 20 i (t n A ) 80 60 40 20 0 it / NaNO2 (pA M -1₎ 0.0 0.5 1.0 1.5 2.0 It (n A ) 0 20 40 60 80 NaNO2 (mM) 0 1 2 3 4 5 kt ( a .u .) 0.0 0.1 0.2 0.3 kt / NaNO2 (a.u.) 0 100 200 300 400 kt ( a .u .) 0.0 0.1 0.2 0.3 0.4 A B

Figure 3. Deviations from Henri-Michaelis-Menten kinetics at high pH.

(A) Activity versus nitrite concentration of two different batches of native NiR immobilised on a PGE electrode. Closed and open circles represent the different batches which are fitted to the Henri-Michaelis-Menten equation (equation 1) by a continuous and a broken line respectively. The two fits resulted in KM values of 47 ± 6 µM and 48 ± 5 µM (at pH 7.47). The

inset shows the Eadie-Hofstee plot in which all points should be at a straight line if the Henri-Michaelis-Menten equation was obeyed. The y-axis in the Eadie-Hofstee plot is activity, the x-axis is activity divided by substrate concentration. (B) Activity versus nitrite concentration at pH 7.14 with benzyl viologen as electron donor. Shown is a fit to the Henri-Michaelis-Menten equation 2 (thin solid line) and to equation 7 (thick line, kcatB/kcatA > 1.3, KMA = 79 ±

12 µM, KM B

> 5000 µM, within the range of accessible nitrite concentrations (< 5 mM) no saturation of the rate occured meaning only lower estimates of kcatB/kcatA and KMB are

(13)

With benzyl viologen as electron donor, identical deviations from HMM occurred reproducibly above pH 6 (Figure 3B; the thin line is a fit according to HMM kinetics, Table 1). An Eadie-Hofstee plot, in which deviations from HMM kinetics appear as deviations from linearity, showed the same, non-linear, trends for assays on the electrode and in solution (insets in Figure 3A and B). It was possible, however, to fit the activity versus nitrite concentration to equation 7 (Figure 3B thick line, Table 1). Also for NiR M150T a substrate activation like phenomenon occurred (results not shown). Thus, over the measured pH range the data is best fit assuming a random-sequential mechanism according to equation 7. Since kcatB< kcatA below pH 6.5 and kcatB > kcatA above (Table 1), it appears

that electron transfer to the type-2 site is slowed down by nitrite replacing the water below pH 6.5, while it is becomes faster above pH 6.5 where the nitrite replaces a hydroxyl ion. Kinetic constants versus pH

At the rotating disk electrode, a far lower background activity was observed than in solution assays, which made it possible to accurately measure KMA, KMB, and the ratio of

kcatB/kcatA over a wide pH range (Figure 4). The individual trends of kcatA and kcatB versus pH

could not be determined accurately since the currents depended on the enzymatic coverage, which was only reproducible within 50 % between different protein films (see Figure 4A). Still, a plot of icatA (equation 1) versus pH determined for wt NiR at +110 mV versus NHE

(Figure 4A, closed circles) agrees with the activity versus pH at 500 µM nitrite when pseudoazurin is used as the electron donor (Figure 1A). While the datapoints at -70 mV exhibit a larger spread still a similar trend is observed in this case as well. Above pH 9 the catalytic currents for wt NiR were too low to be observed, while below pH 4 the higher background activity and the lower catalytic currents of NiR prevented measurements. For NiR M150T the observed currents were far lower (Figure 4A), and as a result the observable range was limited from pH 4.5 to 7.5.

The kcatB/kcatA was approximately 0.3 below pH 6.5 (Figure 4B) while above pH 7

kcatB/kcatA was >1 (results not shown, exact values could not be obtained since KMB was

above the experimentally acceptable range of nitrite concentrations). Below pH 6.0 in bulk solution (Table 1), and for M150T on the electrode, most data points for kcatB/kcatA were less

(14)

For M150T the KMA was 2-10 fold lower than for wt NiR (Figure 4C). A plot of

KMB versus pH (Figure 4D) revealed a slope of 1 for both wt NiR and NiR M150T. For wt,

there was no significant difference between the KMB values at -70 mV and at 110 mV. The

values of M150T were two times lower than those of the wt NiR at the same electrode potential. The logarithmic plot of the solution data of KMB versus pH also displays a slope

of approximately 1 (Table 1).

Figure 4. Kinetic constants of electrode immobilized NiR versus pH

Open circles denote the kinetic constants of wt NiR determined on a PGE electrode held at -70 mV versus NHE, closed circles denote those determined at + 110 mV versus NHE, and gray-filled squares denote the datapoints of NiR M150T determined at -70 mV versus NHE (no datapoints shown in panel B since icatB was difficult to de determine due to the low

currents). The thin line in panel D is a linear fit of log KMB versus pH of the datapoints at

+110 mV, the dashed line is a fit to the datapoints at -70 mV for wt, and the thick line is a fit to the datapoints of M150T. In panel B and D no datapoints are shown above pH 7 since above this pH no saturation is observed and thus no data could be obtained. (A) icatA versus

pH; (B) kcatB/kcatA versus pH; (C) KMA versus pH; (D) KMB versus pH

(15)

Discussion

Random-sequential mechanism

The simplest explanation for our data is obtained when using a random-sequential mechanism that accounts for nitrite binding and type-2 site reduction (scheme 1). At low nitrite concentration, reduction of the type-2 site is faster than nitrite binding to the oxidized site (lower route), while at high nitrite concentrations the binding is faster and the enzyme is driven into the upper route. Below pH 6.5, we find that the upper route is slower than the lower route (indicated by kcatB/kcatA < 1) since the rate-limiting step of electron transfer to

the type-2 site is slower with nitrite bound than with H2O bound, in agreement with

literature reports that electron transfer from the type-1 to the type-2 site is slower in the presence of nitrite (99, 114, 136, 176) and that electron transfer is a rate-limiting step for turnover of the NiR (87, 115, 136). Above pH 6.5, the upper route is faster than the lower route (kcatB/kcatA > 1) because of the replacement of the Cu-bound H2O at the type-2 site by

an OH- (87, 135) and because electron transfer to the type-2 site is faster with nitrite bound than with OH- (Table 1, Figure 3). Also, the absolute magnitude and pH dependence of KMB, which corresponds to the nitrite binding step in scheme 1 (step 1), agrees with the Kdox

for nitrite (vide infra), consistent with the random-sequential mechanism. Below we consider a number of alternatives to explain our data.

According to an ordered mechanism electron transfer from type-1 to type-2 occurs to either the H2O/OH--bound or the nitrite bound type-2 site. Such a sharp division is not

observed in pulse-radiolysis experiments (99, 114, 136, 176). Secondly, with assays employing a strong and fast electron donor, like benzyl viologen or methyl viologen, it is expected that at low nitrite concentrations a large proportion of the NiR will have both the type-1 and the type-2 site reduced. Still, with viologens as the electron donor, the KMA

(Table 1) and kcat values of NiR (see Materials and Methods) are similar to those found in

the presence of the physiological electron donor which is only able to reduce the type-1 site. This observation would mean that nitrite can bind to the reduced type-2 site and that the ‘nitrite-binds-prior-to-reduction’ ordered mechanism does not obtain. Assuming a random-sequential mechanism also obviates the need to postulate the existence of a sensor loop between type-1 and type-2 site (54, 177).

(16)

Against a ‘reduction-before-nitrite-binding’ ordered mechanism (thereby supporting a ‘nitrite-binding-prior-to-reduction’ ordered mechanism) it has been argued that reduction of type-2 site before nitrite binding may result in a three-coordinated inactive form (54). We have data (chapter 5) indicating that after reduction of the type-2 site a slow reversible isomerization to a reduced inactive form may occur, indeed (scheme 1, step 5). Since this step is slow compared to step 4 at nitrite concentrations above Kdred, however, the

possible occurrence of this state is still compatible with a random-sequential mechanism. The effect of pH and nitrite concentration on catalysis

Our data on the pH dependence of the catalytic activity of NiR (Figure 1) are in agreement with literature data (116). Also our KMA value at pH 7 (79 ± 12 µM) with benzyl

viologen as the electron donor agrees with the value of 74 µM determined by Kakutani (116) with methyl viologen as the electron donor.

In this work, for the first time the effect of nitrite concentration on the activity at low pH is studied. An all determining factor is the logarithmic dependence of KMB on pH,

the affinity constant for the upper route. Our data show that below pH 6 the KMB becomes

low enough so that at typical assay conditions (at ≅ 1 mM nitrite concentration) catalysis via the upper route begins to prevail, resulting in a drop in activity (indicated by kcatB <

kcatA, Figure 4B). Above pH 6, the KMB is so high that in typical activity assays most

catalysis occurs via the lower route. Thus, from pH 6 to pH 5 the loss of activity observed in plots of activity versus pH (72, 82, 86, 87, 116), can be assigned to the increasing prevalence of the slower upper route. Below pH 5 the kcatA and kcatBdrop dramatically and

at pH < 4 end up below the detection limit (Figure 4A). The simplest explanation for this is that the type-1 to type-2 electron transfer rate decreases below this pH, as it does in related nitrite reductases (136).

A key feature of scheme 1, in agreement with literature data (99, 114, 136, 176) is that the rate-limiting electron transfer rate between type-1 and type-2 site depends on the ligand that is bound to the type-2 site. The rates decrease in the order H2O > nitrite > OH-.

The slower electron transfer to the type-2 site with nitrite/OH- bound may be due to a lower midpoint potential of the type-2 site and/or a higher reorganization energy (41) than with H2O as a ligand. In pulse-radiolysis experiments indeed a 100 mV decrease in type-2 site

midpoint potential is found going from pH 6 to pH 8 (136).

The up to 10 times lower values of KMA for NiR M150T illustrate the dependence

of this parameter on the electron transfer kinetics. With slower electron transfer (unpublished results), KM will decrease since saturation will occur at lower nitrite

(17)

pronounced for KMA (Figure 4C) than for KMB (Figure 4D) while KMA also follows a

complicated behavior versus pH (for a simple protonation/deprotonation equilibrium slopes equal to 1, 0, or -1 are expected in a logarithmic plot). From Table S1 (supplementary materials) it appears that KMA varies much stronger with various kinetic parameters in the

system than KMB which, therefore, appears a better probe for the nitrite binding than KMA.

The linear dependence of log KMB on pH suggests that either a chemical group in

NiR or nitrite itself needs to be protonated before catalysis and, since no leveling off is observed, that the pKa is < 4.5. The corresponding acid could be the nitrite (pKa = 3.25), the active site aspartate (usual pKa range in proteins (43) is 2 - 5.5) that is known to hydrogen bond to the bound nitrite (83, 84, 87), or the active site His255 (usual range for histidine is 5-8). The linear dependence of log KMB on pH is similar to what has been found

(82) for the Kdox for the type-2 site of Alcaligenes xylosoxidans NiR (in frozen samples < 30

µM at pH 5.2, 350 µM at pH 7.5). EPR experiments on Alcaligenes faecalis S-6 NiR as a function of pH (Fittipaldi, Wijma, unpublished results) have shown that the Kdox for the

type-2 site (§1 mM at pH 6, §10 mM at pH 7 in frozen samples) is indeed approximately equal to the KMB (see Figure 4D).

Conclusions

A random-sequential mechanism is in agreement with the steady-state kinetics observed for wt and M150T NiR with all tested electron donors. At low nitrite concentration, electron transfer occurs prior to nitrite binding while at higher nitrite concentration nitrite binding occurs first. The velocity of the rate-limiting electron transfer step between the type-1 and the type-2 site depends on the ligand that is bound at the type-2 site and decreases in the order H2O > nitrite > OH-. The KMB approximately equals the Kdox

(18)

Supplementary material of chapter 4

Random Sequential Mechanism, steady-state Rate equations

For a catalytic mechanism according to scheme S1, it can be calculated with the determinants method (179) that the turnover rate of the enzyme (kt) depends on substrate

concentration (S) with a quadratic term in S (equation S1-S5). The determinant method is similar to the King-Altman method and originates from the same kind of vector algebra. We used the determinant method since it allowed us to do all calculations on a computer. kt(S) = (kcatA[S] + kcatB[S]2/KMB) / (KMA+ [S] + [S]2/KMB) (S1) kcatA = (S2) k-5k6 (k2k4k-1 + k2k4k3+ k1k-2k3) / (k1k-5k-2k-3+ k-5k6k1k-2+ k2k-5k4k3+ k-5k4k-1k6+ k2k-5k4k-1+ k1k-5k-2k-4 + k1k5k3k-4 + k-5k4k-1k-3 + k-5k4k3k6 + k1k-5k-2k3 + k2k-5k4k-3 + k1k-5k3k-4) KMA = (S3) (k2k-5k3k-4 + k2k5k3k-4 + k2k-5k-1k6 + k-5k-2k3k6 + k2k-5k-1k-3 + k2k5k3k6 + k2k5k-1k-4 + k2k-5k-1k-4 + k2k5k-1k-3 +k2k-5k3k6 + k-5k-2k3k-4 + k-5k-2k-1k-3 + k-5k-2k-1k6 + k-5k-2k-1k-4 + k2k5k-1k6) / (k1k -5k-2k-3 + k1k-5k-2k6 + k2k-5k4k3 + k-5k4k-1k6 + k2k-5k4k-1 + k1k-5k-2k-4 + k1k5k3k-4 + k-5k4k-1k-3 + k-5k4k3k6 + k1k-5k-2k3 + k2k-5k4k-3 + k1k-5k3k-4) kcatB = k6k3 / (k3 + k6 + k-3) (S4) KMB = (S5) (k1k-5k-2k-3 + k1k-5k-2k6 + k2k-5k4k3 + k-5k4k-1k6 + k2k-5k4k-1 + k1k-5k-2k-4 + k1k5k3k-4 + k-5k4k-1k-3 + k-5k4k3k6 + k1k-5k-2k3 + k2k-5k4k-3 + k1k-5k3k-4) / (k1k-5k4k3 + k1k-5k4k6 + k1k-5k4k-3)

Without rearrangement step

(19)

kcatA = (S7) (k2k4k-1k6 + k2k6k4k3 + k1k-2k6k3) / (k4k3k6 + k1k3k-4 + k2k4k-3 + k2k4k-1 + k2k4k3 + k1k-2k3 + k4k -1k6 + k4k-1k-3 + k1k-2k-4 + k1k-2k6 + k1k-2k-3) KMA = (S8) (k2k3k6 + k2k3k-4 + k-2k-1k6 + k-2k-1k-3 + k-2k-1k-4 + k-2k3k6 + k2k-1k6 + k2k-1k-3 + k2k-1k-4 + k-2k3k -4) / (k4k3k6 + k1k3k-4 + k2k4k-3 + k2k4k-1 + k2k4k3 + k1k-2k3 + k4k-1k6 + k4k-1k-3 + k1k-2k-4 + k1k-2k6 + k1k-2k-3) kcatB = (k6k3) / ( k3 + k6 + k-3) (S9) KMB = (S10) (k4k3k6 + k1k3k-4 + k2k4k-3 + k2k4k-1 + k2k4k3 + k1k-2k3 + k4k-1k6 + k4k-1k-3 + k1k-2k-4 + k1k-2k6 + k1k-2k-3) / (k1k4k3 + k1k4k6 + k1k4k-3) Cu2+ N N N Cu2+ N N N HNO2 k_-1 k1[S] H2O Cu1+ N N N H₂O Cu1+ N N N k2 k_-2 Cu1+ N N N HNO2 k3 k-3 k5 k_-5 k-4 k₄[S] P k6

Scheme S1: Random-Sequential Mechanism.

(20)

Table S1: Simulations of the Random-Sequential Mechanism of Scheme S1 Altered parameters A _{Resulting constants}B Description k2 (s-1₎ k3 (s-1₎ k5 (s-1₎ k-5 (s-1₎ k6 (s-1₎ kcatB_/ kcatA k3/k2 KMA (µM) Kdred (µM) KMB (µM) Kdox (µM) initial parameters 1000 250 0.1 0.1 5000 0.31 0.25 27 10 156 100 1000 250 0.1 0.1 500 0.55 0.25 26 10 278 100 slower catalytic step 1000 250 0.1 0.1 50 0.93 0.25 26 10 475 100 1000 250 1 0.1 5000 0.31 0.25 107 10 160 100 1000 250 10 0.1 5000 0.40 0.25 717 10 203 100 1000 250 10 1 5000 0.31 0.25 107 10 160 100 faster isomerisation step 1000 250 10 10 5000 0.31 0.25 27 10 156 100 100 250 0.1 0.1 5000 2.24 2.5 13 10 134 100 slowdown of ET step 2 C 10 250 0.1 0.1 5000 8.80 25 11 10 132 100 100 25 0.1 0.1 5000 0.28 0.25 13 10 116 100 slowdown of both ET steps C 10 2.5 0.1 0.1 5000 0.27 0.25 11 10 112 100 k1 (M-1 s-1 ) k-1 (s-1₎ k4 (M-1 s-1 ) k-4 (s-1₎ initial parameters 1× 107 ₁₀₀₀ _{1 × 10}8 ₁₀₀₀ _0.31 _0.25 ₂₇ ₁₀ ₁₅₆ ₁₀₀ 1× 107 ₁₀₀₀ 1 × 107 ₁₀₀ _0.41 _0.25 ₁₄₈ ₁₀ ₂₄₆ ₁₀₀ 1× 107 ₁₀₀₀ 1 × 109 10000 0.29 0.25 7 10 147 100 1 × 106 ₁₀₀ _{1 × 10}8 ₁₀₀₀ _0.30 _0.25 ₂₉ ₁₀ ₄₁₆ ₁₀₀ decrease and increase of on and off ratesD

(21)

Substrate Inhibition/Activation

k

_CAT1

E

ES E + P

K

_S1

ES

₂

k

CAT

ES + P

2

K

_S2

Scheme S2: Substrate inhibition/activation

Equation S1 also describes the effect of substrate inhibition or activation due to the binding of a second nitrite. For the substrate inhibition/activation scheme S2, the rate is also determined by equation S1. If all substrate binding steps are fast relative to the catalytic step, one can show with rapid-equilibrium treatment (175) that the kcatA refers to kcat1, and

KMA = KS1, kcatB to kcat2 and KMB to kS2.

NaNO2 (mΜ) 0 2 4 6 8 10 iCA T ( -n A ) 12 10 8 6 4 2 0