Catalytic mechanism and protein engineering of copper-containing nitrite reductase

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

Chapter

8

Reorgani

zati

on

Energy

of

the

Type-1

El

ectron

Transfer

Si

te

of

Copper-contai

ni

ng

Ni

tri

te

Reductase

Lowered

by

i

ts

M ethi

oni

ne

Li

gand

This chapter is to be submitted.

Abstract

Introduction

The rate of electron transfer between two redox sites depends on the reorganization energy (λ), the difference in their midpoint potentials (ǻG0), and their electronic coupling (41). To allow for biological electron transfer at high rates, λ should be close to the driving force -ǻG0. After iron, copper is the most frequently used metal in nature to facilitate electron transfer in enzymes. W hile synthetic copper complexes have a much higher λ than iron complexes (2.4 eV for Cu-phenanthroline (235) compared to 0.5 – 1.1 eV for various Fe complexes (236)) , the reorganization energies of biological copper-containing electron transfer sites are similar to those of iron sites (0.6 eV for various c-type cytochromes (237), 0.4 eV for a CuA site (238), 0.6-0.8 eV for a type-1 copper site, (46,

239, 240). Here we investigated how the protein manages to lower the reorganization energy of copper in a type-1 site.

Type-1 copper sites are found in small electron transferring proteins (cupredoxins) that serve in electron transfer between larger redox-enzymes, or in enzymes where they assist in internal electron transfer. The enzymes that contain a type-1 copper site can be involved in respiration (nitrite reductase), or the oxidation of a range of compounds (laccase, ceruloplasmin). In a type-1 site, two histidines and one cysteine bind the copper; these ligands are conserved in all known type-1 sites. Furthermore, often one or two additional ligands bind to the copper. These weaker axial ligands can be methionine, glutamine, or a backbone carbonyl oxygen. The oxidized type-1 site has strong absorption bands around 600 nm (and sometimes 460 nm) which are used in this investigation to follow the reduction of the type-1 site.

As a model to study the electron transfer reaction we used the copper-containing nitrite reductase (NiR) from Alcaligenes faecalis S-6. The physiological role of copper-containing nitrite reductase is the dissimilatory reduction of nitrite (NO2- + 2H+ + e-→ NO

+ H2O) (56); the reverse reaction is catalyzed with similar rates (chapter 3). NiR is a

Here we investigate the effect of the removal of the axial methionine in M150G and M150T and the subsequent binding of Met62 in NiR M150G on the reorganization energy of the type-1 site by studying their reduction by synthetic electron donors and by pulse-radiolysis. The results show that removal of the axial methionine increases the reorganization energy by 0.3 eV and that binding of Met62 keeps the reorganization energy at the level of the wt NiR. Control experiments excluded that changes in electronic coupling were significant. X-ray diffraction results shows that the structural differences between the reduced and oxidized type-1 site of NiR M150G, with Met62 bound, are similar to those between the oxidized and reduced native type-1 site.

Materials and Methods

Electron Transfer and Reorganization Energy

Native NiR and its variants were prepared as described (chapter 7). For the stopped-flow experiments an applied photophysics SX.18MV, equipped with oxygen-impermeable flow lines and set at a temperature of 25 °C, was used. High purity argon was used to deoxygenate solutions. Prior to use remaining traces of oxygen were removed by scrubbing the gas stream with a solution containing 0.5 mM methylviologen, 50 µM proflavin, 75 mM EDTA and 50 mM of phosphate pH 6.0 (241). Butyl rubber and PEEK (Rheodyne) were used in the experimental set-up to protect the gas stream and anaerobic solutions from oxygen. We prepared the Fe(II)EDTA [Fe(II) ethylenediaminetetraacetate] and the Fe(II)HEDTA [Fe(II) N-(2-Hydroxyethyl)ethylenediamine-N,N',N'-triacetate] dilution series (always >10 x excess when reacted) at constant ionic strength (I = 0.2 M with sodium phosphate, pH 7.0 (242)]. To test the anaerobic conditions, the reduction of azurin by Fe(II)EDTA was measured and the results agreed with previous data (we found 1.48 × 104 M-1s-1 which is in agreement with the original value of 1.30 × 104 M-1s-1(242)). Furthermore, no coloration of the Fe(II)EDTA and Fe(II)HEDTA stocks was observed (when the solutions were exposed to air a brown color rapidly appeared, coloration is the result of oxidation of the Fe(II) to Fe(III)).

The visible absorption of the type-1 site of NiR was used to study its reduction by ferrous EDTA in a stopped-flow experiment (scheme 1, in which T1OX;T2OX is the NiR with both the type-1 and the type-2 site in the Cu(II) state, T1RED;T2OX is NiR with the type-1 site in the Cu(I) state, etcetera, Keq= [T1OX;T2RED]/[T1RED;T2OX] = k2/k-2).

T1OX;T2OX T1RED;T2OX T1OX;T2RED T1RED;T2RED k1'[Fe(II)EDTA] Keq k3'[Fe(II)EDTA]

(1) Under the assumption that electrons enter via the type-1 site, the equations that describe the total absorbance (due to the species T1OX;t2OX and T1OX;T2RED) versus time are derived in the supplementary material. Equation 2 shows the absorbance of the type-1 site versus time [A(t)] (A0, absorbance upon mixing; k1 = k1’ × [Fe], k3 = k3’× [Fe], [Fe] is the

concentration of reductant; KR = 1/Keq). ¸ ¸ ¸ ¸ ¸ ¹ · ¨ ¨ ¨ ¨ ¨ © § °¿ ° ¾ ½ °¯ ° ® ­ − ° ° ¿ ° ° ¾ ½ ° ° ¯ ° ° ® ­ + − + + = + − − − _{R 3}t t 1 R R t 1 0 e e e A A(t) K k k k K k k K k 1 1 3 1 1 1 (2)

If the first and second reduction step have the same rate (k1 = k3), then equation 2 simplifies

to equation 3.

{

}

Equations 2 and 3 both predict that at least two exponentials are expected (unless Keq = 1

and k1 = k3), and that the rates of these exponentials should be linearly related to the

reductant concentration. If k1= k3 and Keq 1, then the ratio of the observed rates equals KR

+ 1 and the ratio of the observed amplitudes equals KR-1. If in equation 3 Keq < 1, then both

amplitudes have the same sign; if Keq > 1, then the amplitudes have opposite sign (the result

of the build-up of the T1OX;T2RED intermediate and the concomitant decay of the T1OX;T2OX form). To analyze stopped-flow data with these equations a routine written for this specific purpose under Igor Pro (WaveMetrics, Inc.) was employed (available at request).

The data were analyzed with Marcus theory (41). Equation 4 was derived assuming that the only parameters which vary due to the mutations are driving force and reorganization energy (see supplementary material). Control experiments for this assumption are discussed.

In equation 4, kRM is the second order rate of reduction of the type-1 site of the mutant

protein by the reductant; kRW is the same for the wild-type enzyme; ∆Ȝ is the reorganization

energy of the type-1 site of the mutant (ȜMutant_{) minus the reorganization energy of the}

type-1 site of the wt NiR (ȜWT_{). In equation 5, F is the Faraday constant, R is the gas constant, T}

the absolute temperature, and EM is the midpoint potential of the type-1 site.

[

] [

]

[

] [

]

The reorganization energies of the mutant type-1 sites were obtained from pulse-radiolysis data as described (45, 46).

Crystal Structure

M150G crystals were grown at room temperature by the hanging drop vapor diffusion method, as described previously (chapter 7). The mother liquor consisted of 6-10 % PEG6000, 10 mM sodium acetate, pH 4.5, 2 mM zinc acetate, and 2 mM CuSO4.

Crystals were moved to mother liquor without CuSO4 for 30 minutes to remove excess

copper, then to copper-free mother liquor containing 1 M acetamide for 10-20 minutes, then to Buffer A (copper-free mother liquor containing 1 M acetamide and 20 mM ascorbic acid) until they became colorless. Lastly, crystals were transferred into Buffer A plus glycerol as a cryoprotectant. Native NiR crystals were grown by vapor diffusion in mother liquor containing 6-10% PEG4000 and 0.1 M sodium acetate, pH 4.0. Reduced crystals were obtained by soaking in mother liquor supplemented with 20 mM ascorbic acid until the crystals went from green to colorless, and then in mother liquor supplemented with glycerol as a cryoprotectant. Reduced M150G-acetamide and reduced native NiR crystals were looped into a cryostream (Oxford Cryo Systems) for diffraction studies using a MAR345 detector and Rigaku RU-300 x-ray generator. Reduced, acetamide-soaked M150G NiR and reduced native NiR diffracted to better than 1.9 Å resolution and diffraction data was processed with DENZO (192).

Reduced, acetamide-soaked M150G and native NiR crystals contain the NiR trimer in the asymmetric unit of space group P212121. For reduced M150G-AcM, a 1.4 Å

The copper ligand geometry and positions of the copper atoms were not restrained throughout the refinement. In reduced M150G-AcM, chain A begins with Thr5 and ends with Gly339 while chains B and C begin with Ala4 and end with Gly339. In reduced native NiR, all chains begin with Ala4 and end with Gly339. Statistics of data processing and structure refinement are presented in Table 1.

RESULTS

Reduction of NiR followed by stopped-flow

To determine the reorganization energy of the modified type-1 sites we followed the reduction of wt, M150T, and M150G by Fe(II)EDTA and Fe(II)HEDTA by monitoring the absorption at 600 nm or 460 nm versus time. This experiment was not carried out for M150H since its reduction potential (104 ± 5 mV versus NHE) is so close to that of the reductants (120 mV versus NHE for Fe(II)EDTA and 47 mV versus NHE for Fe(II)HEDTA) that reverse electron transfer would affect the kinetics. When wt NiR was reduced, with either Fe(II)EDTA or Fe(II)HEDTA, traces were observed as depicted in Figure 1A. The predictions described for scheme 1 in the materials and methods section were tested by fitting the data with simple exponentials and comparing the observed rates and amplitudes with those expected (see materials and methods) if scheme 1 applies and if k1 = k3. Attempts to fit the reduction of native NiR with a single exponential produced

systematic deviations between fit and data (results not shown); a double exponential fit of the reduction of native NiR did yield amplitudes of opposite sign, showing that Keq > 1.

When fitted with equation 2 (assuming different values for k1 and k3), then the obtained

values for k1 and k3 were identical within error. The data for native NiR could be fitted

equally well (Figure 1A) when assuming that k1 and k3 were identical (equation 3).

Simulations of the absorbance of T1OX_;T2OX _{and T1}OX_;T2RED _{versus time confirmed the}

build-up of a T1OX_;T2RED _{intermediate (Figure 1A, inset).}

For the M150T and M150G variants of NiR double exponentials with only positive amplitudes were observed (Figure 1B) meaning that Keq < 1. The reduction of

time (s) 0 20 40 60 80 100 a b s o rp ti o n a t 5 8 9 n m 0.00 0.01 0.02 0.03 time (s) 0 5 10 15 20 25 a b s o rp ti o n a t 6 0 0 n m 0.00 0.01 0.02 0.03 A B 0 20 40 60 80 100 0.00 0.01 0.02 0.03 0 5 10 15 20 25 0.00 0.01 0.02 0.03

Table 1: Crystallographic Data Collection and Refinement Statistics for reduced M150G-Acetamide and reduced wt NiR

Structure Reduced M150G-AcM NiR Reduced native NiR Cell dimensions (Å) a = 61.0 b = 102.0 c = 146.4 a = 61.5 b = 102.4 c = 145.7 Resolution (Å) 1.90 (1.97-1.90)A _{1.85 (1.92-1.85)} R-merge 0.071 (0.400) 0.116 (0.413) {I}/{σ(I)} 15.6 (2.49) 12.3 (4.7) Completeness (%) 94.7 (90.1) 91.5 (100) Unique reflections 68966 (6488) 72738 (7855) Working R-factor 0.175 0.150 Free R-factor 0.220 0.180 R.m.s.d. bond length (Å) 0.014 0.010 Overall B-factor (Å2_{) 27.6 17.3} Water molecules 869 1106 PDB entry code 2B08

A _{Values in parenthesis are for the highest resolution shell. {I}/{σ(I)} is the average intensity divided by the}

average estimated error in intensity.

Figure 1: Reduction of NiR by Fe(II)EDTA

For M150T it was not possible to fit all parameters at once since Keq was difficult

to fit due to the low amplitude of the slower phase. To determine the second order rate constants, we fixed 1/Keq at 30, which produced satisfactory fits (Figure 1B). Values of

1/Keq of 15 or 60 were clearly too low or too high (not shown), and were therefore quoted

as the error margins (Table 2). All of the data could be fit excellently with equation 3 (in which k1 = k3), and the dependence of the rates on reductant concentration was always

linear (Figure 2).

From the known reduction potentials of the T1 sites (caption Table 2) and the values of Keq obtained by fitting the traces with equation 3, the values for the reduction

potential of the type-2 site could be calculated (Table 2). Within experimental error they proved to be equal (average 234 ± 4 mV versus NHE).

The bimolecular rate of reduction of wt NiR by Fe(II)HEDTA and Fe(II)EDTA was not significantly influenced by the presence of acetamide (Table 2). The bimolecular rate constants for the reduction of M150T, M150G, wt NiR and their known reduction potentials (Table 2) were used to calculate the increase in reorganization energy of the type-1 site (ǻȜ) with equation 4. Both for Mtype-150G and for Mtype-150T, and both with Fe(II)EDTA and Fe(II)HEDTA as the electron donors, ǻȜ amounts to 310 ± 40 meV (Table 2). When acetamide was added to NiR M150G the rate of reduction decreased to slightly below that of the wt NiR (Table 2). Since for M150G with acetamide bound the midpoint potential is < 225 mV versus NHE (chapter 7), an upper-limit for ∆λ was obtained amounting to ǻȜ < 70 ± 20 meV.

Internal Electron Transfer in NiR

Figure 2: Second order rate versus reductant concentration

The plots rate constants obtained from fitting the data to equation 3 versus Fe(II)HEDTA; open triangles, NiR M150T; open circles, NiR M150G; closed circles, NiR wt.

Table 2: Kinetic and thermodynamic constants determined from the reduction of NiR variants with inorganic donors

NiR variant Reductant acetamide (mM) rate of reduction (M-1 _s-1₎ Keq EM of the type-2 site (mV) B ∆Ȝ (eV) WT FeHEDTA 0 51.5 ± 0.2 1.5 ± 0.5 224 ± 14 0 WT FeHEDTA 500 55.6 ± 0.6 1.7 ± 0.6 226 ± 14 -0.02 ± 0.04 WT FeHEDTA 750 56.9 ± 0.4 1.8 ± 0.3 228 ± 11 -0.02 ± 0.04 M150T FeHEDTA 0 144.3 ± 1.3 0.033 C _{253± 23 0.30 ± 0.04} M150G FeHEDTA 0 83.2 ± 2.5 ND ND 0.31 ± 0.05 M150G FeHEDTA 500 45.7 ± 0.6 0.13 ± 0.04 225 ± 12 <0.07 M150G FeHEDTA 750 51.6 ± 0.7 0.24 ± 0.03 234 ± 9 <0.05 WT A _{FeEDTA 0 119 ± 3 2.5 ± 0.7 236 ± 12 0} M150G FeEDTA 0 185 ± 2 0.083± 0.031 251 ± 14 0.32 ± 0.03 M150G FeEDTA 500 98.7 ± 1.2 0.22 ± 0.04 239 ± 9 <0.09 M150G FeEDTA 750 97.9 ± 1.5 0.25 ± 0.02 235 ± 7 <0.09

A_{In the presence of 750 mM acetamide traces identical to those without acetamide were obtained.} B _{This was calculated with the know type-1 site reduction potentials (chapter 7) E}

M wt = 213 ± 5

mV vs NHE, EM M150G = 312 ± 5 mV, EM M150T = 340 ± 5 mV, EM M150G in the presence of

500 mM acetamide = 278 ± 5 mV versus NHE, EM M150G in the presence of 750 mM acetamide

= 271 ± 5 mV versus NHE. C _{This value is in between .017 and 0.067. For experimental details see}

Materials and Methods. ND, not determined.

Figure 3: Time resolved absorption changes

Reduction (left panel) and reoxidation (right panel) of the type-1 site in M150G NiR at 460 nm (protein concentration 64 µM, pH 7.0, optical path length 3 cm, pulse width 0.5 µs).

Figure 4: Arrhenius plots

Table 3: Rate constants and activation parameters for the internal electron equilibration obtained by pulse-radiolysis Variant k298 (s-1) -∆G0 (meV) ∆H≠(kJ Mol-1₎ ∆S ≠_{(J K}-1

mol-1₎ λTotal (eV) λT1 (eV) ǻȜT1 (eV)

WT 2480 ± 160 -21 16.9 ± 2.2 -123 ± 8 1.12 ± 0.13 0.57 ± 0.26 0

M150G 380 ± 30 78 24.8 ± 2.4 -113 ± 7 1.39 ± 0.11 1.11 ± 0.22 0.54 ± 0.22

M150H 2350 ± 170 -129 20.5 ± 4.5 -112 ± 15 1.29 ± 0.14 0.91 ± 0.29 0.34 ± 0.29

M150T 690 ± 40 106 ND ND 1.25 ± 0.10 0.83 ± 0.20 0.26 ± 0.20

The values for M150T were determined at 275 K instead of at 298 K. ND, not determined; ȜTotal, average

reorganization energy of type-1 and type-2 site. ȜT1, reorganization energy of the type-1 site.

Figure 5: Stereo views of M150G-AcM and native NiR.

(A) Superposition of oxidized (light gray) and reduced (dark gray) M150G-AcM. Residues near the type-1 site are labeled (AcM; acetamide). (B) Superposition of oxidized

Crystal structures of the reduced NiR

Examination of all three monomers in the crystal structure of reduced M150G-AcM shows that the Met62 side-chain is displaced by acetamide and coordinates to the type-1 copper at a position that partly coincides with that of Met150 in native NiR (Figure 5). The A chain of the structure was used for detailed comparison with oxidized acetamide-bound M150G NiR (chapter 7). Torsional changes in the Met62 side-chain of reduced M150G-AcM equal those in the oxidized structure. The Ȥ1, Ȥ2, and Ȥ3 angles change by

117°, 44°, and 20° respectively, compared to the Met62 side-chain of reduced native NiR. Additionally, the backbone ij angle shifts 17°. The combined torsional changes result in movement of the Met62 Sδ by roughly 4.4 Å to lie 2.28 Å from the type-1 copper atom.

The acetamide molecule forms hydrogen bonds with two buried water molecules near the type-1 site (not shown).

Comparison of reduced M150G-acetamide with oxidized M150G-acetamide reveals small differences in coordination geometry (Table 4). Specifically, the Met62 Sδ-Cu bond

is consistently shorter (§ 0.1 Å) and the Met62 Sδ shifts § 0.3 Å such that the

Met62-Cu-His95 angle changes by 9° and the Met62-Cu-His145 angle changes by 5°. The conformation of Met62 differs slightly between the reduced and oxidized structures (r.m.s. deviation 0.16 Å). In contrast, the bound acetamide is well ordered (B-factor 36 Å2) and is located at nearly the same position in either oxidation state (Figure 5).

Density is observed for additional metal atoms near the type-1 copper sites in each chain. In chains B and C, copper atoms at 30% and 40% occupancy are refined ~2 Å from the type-1 site copper atom. These dinuclear copper sites have been observed previously in the structure of reduced D98N NiR (84), and in native NiR structures when crystals are reduced with ascorbate in the presence of excess copper (unpublished results). Previous attempts to obtain reduced, acetamide-bound M150G NiR with excess free copper (2 mM) in the mother liquor and soaking solutions resulted in all three type-1 sites in the asymmetric unit containing substantial dinuclear copper binding, with the Met62 Sδ

occupying a position that overlaps that of Met150 in the dinuclear site of reduced D98N NiR (data not shown). In chain A, a copper atom at 30 % occupancy (B-factor 37 Å2_{) is}

found on the surface of the NiR near the type-1 site (modeled 2.1 Å from the Nε2 atom of

Table 4: Metal ligand geometry of type-1 sites in native Nitrite Reductase and M150G A Oxidized native NiR Reduced Native NiR Oxidized M150G acetamide Reduced M150G acetamide Distances (Å) Axial – Cu 2.48 2.41 2.37 2.28 95 – Cu 2.07 2.06 2.13 2.18 136 – Cu 2.22 2.21 2.24 2.27 145 – Cu 2.06 2.18 2.04 2.12 Cu – NSN B _{0.64 0.66 0.63 0.70} Angles (°) 136 – Cu – 95 129 130 126 119 136 – Cu – Axial 106 109 100 100 Axial – Cu – 95 89 90 95 104 Axial – Cu – 145 133 131 129 124 θC _{64 64 71 78}

A _{The numbers 95, 136, and 145 in the left column refer to the N}

δ of His95, the Sγ of Cys136, and the Nδ of

His145. Axial refers to the Sδ of Met150 in the wt NiR and the Sδ of Met62 in the M150G structures. Sigma

values (standard deviations determined from average values of three monomers in the asymmetric unit) amount to less than 5 % for bond angles and less than 3 % for bond distances. The data of reduced M150G acetamide are those of the A chain which appeared to have least formation of a dinuclear site B _{This is the distance}

between the Cu atom and the NSN plane determined by the ligand atoms of residues H95/Cys136/His145. C_θ

is the dihedral angle between the planes through Cu-136-Axial and the plane through Cu-95-145.

Reduced native NiR was solved to higher resolution (1.85 Å) than previously (83). The observed differences at the type-1 site between reduced wt NiR and the oxidized structure are minimal (Figure 5, Table 4). Comparison of the ligand geometry over the three monomers in the asymmetric unit reveals a consistent lengthening of the His145-Cu bond in the reduced protein. This change is largely due to a shift in the copper position of roughly 0.1 Å in line with the His145-Cu bond.

Discussion

Reduction of NiR by Fe(II)EDTA/Fe(II)HEDTA

The stopped-flow data were analyzed with equation 3, which presumes that electrons enter via the type-1 site, followed by a rapid establishment of the redox-equilibrium between type-1 and type-2 site. Since Fe(II)HEDTA and FE(II)EDTA are far larger than the physiological substrate (NO2-), it is indeed not expected that these enter the

narrow tunnel (77, 83, 139) to the type-2 site cavity, as the SO2- radical can do in

confirmed by the rates measured by pulse-radiolysis (Table 3); these rates are much faster than the rates of reduction of the type-1 site by Fe(II)EDTA/Fe(II)HEDTA.

Earlier studies of the reduction of NiR by dithionite focussed on the different reducing species (S2O42-/SO2-.) and the resulting kinetics (130, 243). Here, with

Fe(II)EDTA and Fe(II)HEDTA, no complications occur, which makes it easier to study the effects of mutations on the kinetic parameters of the redox-sites. Furthermore, since the reductants do not bind to the type-2 site (like SO2-. does (243)), the reduction potential of

the type-2 site could be measured. A further advantage of the presented approach is that non-linear regression analysis could be used, which produces more accurate results (244) than when the data first have to be linearized.

The reduction potential found for the type-2 site (234 ± 4 mV versus NHE) is within the range of midpoint potentials of other nitrite reductases (172 mV to 310 mV for other NiRs (99, 111)). The midpoint potentials for the type-2 sites of all variants are within error identical to the average of 234 mV in agreement with the earlier finding that mutations in the type-1 site do affect the type-2 site midpoint potential (45). The successful use of equation 3 in calculating these midpoint potentials again confirms the consistency of the present analysis. Our data indicate that the rate of the second reduction step (the reduction of T1OX;T2RED) is identical to the rate of the first reduction step. This is similar to what is observed for A. xylosoxidans NiR (243) where the rate of reduction of the type-2 site is independent of the redox-state of the type-1 site.

The method to determine ∆Ȝ

For the estimation of ∆λ from stopped-flow experiments, equation 4 was used, which was derived under the assumption that the mutations only alter reduction potential and reorganization energy and not the electronic coupling. Sources of error (41) could be that large work terms come into play or that there exist mechanistic differences between the cross-reactions with the native NiR and the mutant NiRs. We have assumed that work terms and mechanisms are identical for the mutant and the wt proteins because the mutations are not located on the protein surface and do not introduce charges. As a control experiment for these assumptions we used two different electron donors, Fe(II)HEDTA1- _and

Fe(II)EDTA2-_{, which differ by coordination and net charge. They gave the same ∆λ values}

In a second series of experiments, we estimated the reorganization energies of the type-1 sites by pulse-radiolysis. Also with this established technique an increased Ȝ was found, within error equal to the values obtained by stopped-flow (Table 3). Thus, both techniques report a substantial increase in Ȝ due to the mutations.

Reorganization energy and the methionine ligand

The reorganization energy is the energy needed to displace the nuclei from their equilibrium positions in the reduced state to their preferred positions in the oxidized state without changing the redox state, and vice versa (41). The reorganization energy can be divided into inner-sphere (displacing atoms belonging to the redox-center) and outer-sphere (reorienting surrounding atoms) contributions. The outer-sphere reorganization energy is mainly due to the reorientation of the solvent, and is effectively decreased by burying the redox-center in the protein. Although an increase in the contribution of outer-sphere reorganization energy to ǻȜ in M150T and M150G cannot be excluded, it is unlikely to be significant as both mutants have an identical ǻȜ (Table 2), while an identical increase in solvent accessibility is not expected with these mutations.

The inner-sphere reorganization energy (Ȝi) can be calculated (41) from equation

6, which sums the products of all the changing bond lengths (∆R) and their force constants (k) for the oxidized and reduced state.

¦ ∆ + = j ox_j red_j j red j ox j i R k k k k ₂ λ (6)

Equation 6 shows that a redox-protein can lower its Ȝi by a) smaller bond-length changes

between oxidized and reduced states; b) lower force-constants for these bonds (186, 245-247). Computational work (8, 9, 198, 214, 248-250) suggested that the presence of a methionine in a type-1 site may result in smaller bond length changes and low force constants that reduce Ȝi by hundreds of meV. Thus, the computational results are in the

same order of magnitude as our experiments that show that the binding of methionine lowers the reorganization energy of the type-1 site by 0.3 eV.

When the methionine ligand lowers the reorganization energy of the type-1 copper site, its presence results in faster electron transfer. Glutamine as an axial ligand may have the same function; replacing it by methionine lowers the reorganization energy by only 0.1 eV (251). It can be calculated (41) that the 0.3 eV lower reorganization energy affects the rate of electron transfer more than the 100 mV lowering of the reduction potential of a type-1 site brought about by the presence of a methionine (2type-14). For a CuA site it was recently

mV, from which one could speculate that also in a CuA site the methionine is more relevant

in lowering the reorganization energy. Thus, it appears that the methionine effects electron transfer both via reorganization energy and via a shift in reduction potential.

Comparison of oxidized and reduced native NiR shows that the type-1 copper site changes little upon reduction, consistent with other type-1 sites (232). Similarly, the coordination geometry of the type-1 site of M150G-AcM is hardly altered upon reduction. Furthermore, the acetamide molecule is retained and does not shift significantly chain A of reduced M150G-AcM, which is not perturbed by the formation of a dinuclear copper site. Structural changes in the copper site are limited to a slight reorganization of Met62. As in native protein, the overall geometry of the type-1 site does not change dramatically between oxidized and reduced states of M150G-AcM, allowing for efficient electron transfer function in this enzyme (equation 6).

Conclusions

A method with simple reductants was used to study the ǻȜ that resulted from protein engineering. At the same time, the midpoint potential of the type-2 site of NiR from A. faecalis S-6 was determined at 234 ± 4 mV. The reliability of the protocol was checked by an independent determination of ǻȜ by pulse radiolysis experiments. Crystal structures showed that the conformation of the type-1 site is hardly altered when the NiR M150G with an allosteric ligand in place is reduced. The same is observed when going from oxidized to reduced wt NiR. Furthermore, it makes no significant difference for the reorganization energy whether Met150 in wt NiR or Met62 in NiR M150G binds to the copper. The binding of a methionine ligand in a type-1 copper site lowers its reorganization energy by 0.3 eV.

Supplementary Material

Expected Kinetics for the reaction

AoxBox AredBox AoxBred AredBredk1'[Fe(II)EDTA] Keq k3'[Fe(II)EDTA]

A and B are here defined as the type-1 and the type-2 site. Thus the assumption is that the type-1 site accepts electrons and the type-2 site not. The assumption that electron transfer is irreversible is justified since the reduction potentials of FeEDTA (120 mV versus NHE) and FeHEDTA (47 mV versus NHE) are much lower than that of the used NiR variants and these reagents were also in large excess.

Some short-hand definitions Keq= [AoxBred]/[AredBox] = k2/k-2

KR = 1/Keq

k1= k1’[FeEDTA]

k3= k3’[FeEDTA]

Standard mass equations

1) ABTotal = [AoxBox] + [AoxBred] + [AredBox] + [AredBred]

2) ABTotal t=0 = [AoxBox] t=0

First absorbing species:

The first relevant species (which absorbs) is AoxBox: 3) d[AoxBox]/dt = - k1 [AoxBox]

Therefore its concentration changes over time according to: 4) [AoxBox](t) = [AoxBox]t=0× e-k1× t = ABTotal × e-k1 × t

Second absorbing species

For the second absorbing species (AoxBred) we can assume fast equilibrium with AredBox since values measured for k2 and k3 (using pulse-radiolysis) are much faster than k1 and k3.

5) [AoxBred]/([AredBox) + [AoxBred]) = [AoxBred]/(KR× [AoxBred] + [AoxBred]) = 1/(KR+ 1)

6) then d[AoxBred]/dt = +k1[AoxBox]/(KR+1) – k3[AoxBred]

then (43)

Total absorbing species

8) [AoxBox](t) + [AoxBred](t) = (ABTotal× e(-k1×t) )+ (ABTotal× [k1/(KR+1)] / [k3-(k1/(KR+1))] ×

[e(-k1 t/(KR+1)) – e(-k4×t)

After separating rates and amplitudes it shows that three exponentials are expected

9) [AoxBox](t) + [AoxBred](t) = {ABTotal× e(-k1×t) } + {ABTotal× [k1/(KR+1)] / [k3-(k1/(KR+1))] ×

e(-k1 t/(K

R+1)) } – {ABTotal× [k1/(KR+1)] / [k3-(k1/(KR+1))] × e(-k3×t)}

Absorbance versus time with three or two exponentials

Assuming that the extinction coefficient is 2900 mM-1 _cm-1 _{for both AoxBred and AoxBox then}

10) AbsTotal _{= 2900 × {AoxBred + AoxBox}}

11) absorbance (t) = V + W + X 12) V = AbsTotal t=0 × 1 × e(-k1×t) 13) W = AbsTotal t=0 × [k1/(KR+1)] / [k3-(k1/(KR+1))] × e(-k1 t/(KR+1)) 14) X = AbsTotal t=0 × [k1/(KR+1)] / [k3-(k1/(KR+1))] × – e(-k3×t)

If k1 = k3 then only two exponentials expected

15) absorbance (t) = Y + Z 16) Y (=V + X) = AbsTotal t=0 × [1 - k/(KR+1)] / [k-(k/(KR+1))]× e(-k’×t) 17) Z (=W) = AbsTotal t=0 × [k/(KR+1)] / [k-(k/(KR+1))] × e(-k t/(KR+1)) Since 18) [k/(KR+1)] / [k-(k/(KR+1))] = k / [k×(KR+1) –k] = 1/(KR+1-1) = 1 / KR = Keq Therefore 19) Y = AbsTotal t=0 × [1- Keq ] × e(-k×t) 20) Z = AbsTotal t=0 × Keq × e(-k× t/(KR + 1))

Equations 10-20 reveal that

1. A fast equilibrium (k2/k3) influences the observed rate constants because one of its components

(AredBox) does not absorb.

2. Double exponentials are expected, unless Keq= 1 and k1 = k3.

3. Both rates should increase linearly with [FeEDTA]. 4. If Keq < 1 both amplitudes have a positive value.

5. If Keq > 1 one of the amplitudes has a negative value.

6. If k1= k3, then the ratio of the observed rates equals KR+ 1.

Conversion of second order rates to ∆λ

Considering a reductor R that can reduce both A and B, with all species in redox-equilibrium with each other at the end, then the cross-reactions, the electron-self -exchange reactions, and the equilibria are of the form:

Rred Rox Rox Rred ] [Aox][Bred ] [Ared][Box Bred Rox Box Rred Ared Rox Aox Rred RR AB RB RA +   →  + = +   →  + +   →  + k k k K (etcetera)

With the cross relation given by Marcus and Sutin (41), the rate of the electron-self-exchange rate of a mutant protein over that of the wt protein can be calculated.

1) kRA = (kRRkAAKRAfRA)0.5×WRA therefore 2) kRA2= kRRkAAKRAfRA × WRA2 therefore 3) kAA = kRA2/(kRRKRAfRA × WRA2) Since 4) kRB = (kRRkBBKRbfRB)0.5×WRB

In the same way

5) kAA / kBB = [kRA2/(kRRKRAfRA × WRA2)] / [kRB2/(kRRKRBfRB × WRB2)]

Now the work terms (W) for reduction of protein A and B will be identical if it concerns proteins with the same surface characteristics, like mutations below the surface. The same is true for f (which equals unity when the second order rate constants are far less than diffusion controlled).

6) kAA/ kBB = [(kRA)2/(KRA)]/ [(kRB)2/(KRB)]= [(kRA)2×KRB]/ [(kRB)2× KRA]

The second-order rate constants are equal to the true electron transfer rate constant (ket) divided by the

dissociation constant for the protein-protein complex. It can be assumed that this dissociation constant is equal for both protein variants as long as their surfaces do not change. Therefore:

7) ketAA/ ketBB = [(kRA)2×KRB]/ [(kRB)2×KRA] Since 8) KRB/KRA= ([Ared][Rox])/([Aox][Rred])/([Bred][Rox])/([Box][Rred]) = ([Ared][Box])/([Aox][Bred]) = KBA Thus 9) ketAA/ ketBB = (kRA)2×KBA/(kRB)2

The electronic term HAB2 can be eliminated under the assumption that it is identical for wt and mutant

NiR, thus the path of electron transfer should not be affected by the mutation. Therefore, since ∆G0’ = 0 for an electron self exchange reaction, and with ȜAA= ȜBB + ∆Ȝ

11) ∆Ȝ = -4RT × ln[(ketAA/ ketBB)×((ȜBB+∆Ȝ)/ȜBB)0.5]

The term ((ȜBB_+∆Ȝ)/ȜBB₎0.5 _{is negligible. In our calculation, using a Ȝ of the wt site of 0.8 eV, the ∆∆λ}

was < 0.02 eV. Therefore 12) ∆Ȝ = -4RT ln[ketAA/ ketBB]

After substituting in equation 9 13) ∆Ȝ = -4RT ln[kRA

2

×KBA/ kRB 2