Unusual Dynamics of Ligand Binding to the Heme Domain of the

Article pubs.acs.org/JPCB

Unusual Dynamics of Ligand Binding to the Heme Domain of the Bacterial CO Sensor Protein RcoM‑2 Latifa Bouzhir-Sima,†,∥ Roberto Motterlini,‡,§ Julia Gross,† Marten H. Vos,*,† and Ursula Liebl† †

LOB, Ecole Polytechnique, CNRS, INSERM, Université Paris-Saclay, 91128 Palaiseau Cedex, France Faculté de Médicine, Université Paris-Est, Créteil 94000, France § INSERM, U955, Equipe 12, Créteil 94000, France ‡

ABSTRACT: The aerobic Gram-negative bacterium Burkholderia xenovorans expresses two highly homologous carbon monoxide (CO)-responsive transcriptional regulators, RcoM-1 and RcoM-2, which display extraordinarily high CO affinities, even under oxygenic conditions. To gain insight into the origin and perspectives of this feature, we characterized the ligand-binding properties of the N-terminal, heme-binding Per/Arnt/Sim sensor domain of RcoM-2 by time-resolved spectroscopy. We show that upon photodissociation of the heme−ligand bond, CO geminately rebinds to the heme with picosecond time constants and more than 99% rebinding yield, an unprecedented property of native heme proteins. Remarkably, the rebinding kinetics speeds up when the protein motions are slowed by cooling or solvent viscosity. This indicates that the origin of the observed efficient rebinding is a protein-imposed CO configuration in the heme pocket that is highly favorable for binding, a feature strongly in contrast to that of hemoglobins. The binding of CO to the ferrous heme from the solvent requires dissociation of the methionine axial heme ligand. From the kinetics of ligand binding and the extreme stability of the CO complex, we deduce that the dissociation constant for CO is lower than 100 pM. Finally, we show that when the ferric complex is exposed to CO gas or a COreleasing molecule under oxygenic conditions formation of the ferrous carbonyl complex can occur on a time scale of minutes in the presence of a redox mediator. These findings pave the way for possible applications of the RcoM-2 heme domain as a CO sensor and/or scavenger.

■

INTRODUCTION Heme-based gas sensor proteins play important roles in physiological signal transmission pathways and, in particular, in the adaptation of metabolic functions to environmental conditions.1,2 In these systems, the binding to or release of small gaseous ligands, such as O2, NO, or CO, from the heme cofactor leads to alterations in associated functions, such as binding to DNA and catalysis of biochemical reactions. These induce, directly (in heme-based transcriptional regulators) or indirectly, modifications in the protein expression levels. Until recently, most attention was devoted to proteins sensing molecular oxygen or nitric oxide. With the recognition of CO as an important signaling agent in both eukaryotic and prokaryotic organisms over the last decade, the attention has also turned to CO-sensing proteins. Although CO can act as a competitive inhibitor for numerous heme-based enzymes, to date only a limited number of COspecific sensor proteins are known. In eukaryotic organisms, CO is a breakdown product of heme, produced by heme oxygenases.3−5 Eukaryotic sensors include the CO-regulated transcription factor, NPAS2, that is involved in regulation of the circadian rhythm,6 containing two heme-binding Per/Arnt/Sim (PAS) domains. © XXXX American Chemical Society

Some prokaryotic organisms can metabolize environmentally produced CO and employ CO sensors to control this metabolism. Only two types of bacterial heme-based CO-specific sensors have been unambiguously identified. The best characterized is CooA, a homodimeric transcriptional activator that in the presence of CO promotes the transcription of genes involved in anaerobic CO metabolism in Rhodospirillum rubrum.7,8 Incidentally, CooA is also the only heme-based sensor protein for which the structure of the holoprotein has been identified in the CO-free form, which is inactive for DNA binding;9 in this system, the heme is linked to the N-terminus. More recently, two CO-responsive transcriptional regulators, heme proteins RcoM-1 and RcoM-2 (regulator of CO metabolism), were identified in the aerobic bacterium Burkholderia xenovorans LB400, which is involved in the regulation of aerobic, and probably also anaerobic, CO oxidation.10 Here, a b-type heme is incorporated into a PAS domain, as is the case in a number of other heme-based sensor proteins, including the prototypical bacterial oxygen sensor proteins FixL and EcDos11 as well as the eukaryotic protein Received: August 12, 2016 Revised: September 22, 2016

A

DOI: 10.1021/acs.jpcb.6b08160 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

turer’s instructions and using the following primers: M104H: forward: 5′-TTCCCCGCCTCCTGTAGCGCATATGATTAATATCCCGGATCG-3′; reverse: 5′-CGATCCGGGATATTAATCATATGCGCTACAGGAGGCGGGGAA-3′; M104I: forward: 5′-CCGCCTCCTGTAGCGATAATGATTAATATCCCGGA-3′; reverse: 5′-TCCGGGATATTAATCATTATCGCTACAGGAGGCGG-3′; M105I: forward: 5′CCTCCTGTAGCGATGATAATTAATATCCCGGATCG-3′; reverse: 5′-CGATCCGGGATATTAATTATCATCGCTACAGGAGG-3′; M104I/M105I: forward: 5′-GCCTCCTGTAGCGATAATAATTAATATCCCGGATCGT-3′; reverse: 5′-ACGATCCGGGATATTAATTATTATCGCTACAGGAGGC-3′. The purified proteins were suspended in 50 mM Tris−HCl buffer (pH 8.0), 150 mM NaCl, and unless otherwise indicated, all experiments were performed in this buffer. Purified RcoMH-2 was predominantly in the ferrous CObound form. To obtain the ferric form, it was illuminated for ∼12 h with white light originating from a halogen lamp equipped with an optical fiber at an intensity of ∼0.15 W/cm2, at 4 °C in an airexposed vial. To obtain the ferrous unliganded form, the protein was degassed and reduced with a slight excess of dithionite. To obtain the CO-bound form, the ferrous protein was exposed to 1 atm CO in the cell or vial headspace. The ferrous NO form was fully formed within 5 min upon exposure of the ferrous protein to 0.1 atm NO in the headspace. The ferric NO form was fully formed within 1 min upon exposure of the ferric protein to 0.1 atm NO in the headspace. On the time scale of several hours, a fraction of this form was subsequently transformed to the ferrous NO form. All steady-state and ultrafast spectroscopy experiments were performed in 1 mm cells that could be gas-tight sealed with rubber stoppers. Steady-state spectra were obtained using a Shimadzu UV−vis 1700 spectrometer. Ultrafast transient absorption experiments at a 500 Hz repetition rate, with a 570 nm pump pulse and a continuum probe pulse, were performed as in ref 22, at a protein concentration of ∼50 μM. Stopped-flow experiments to monitor CO binding to the ferrous protein were performed on a Biologic SFM 300 instrument, equipped with a JM Tidas diode detector. The optical path length of the measuring cell was 0.8 mm. The instrument was extensively flushed with gaseous nitrogen prior to use. After mixing, the protein concentration was ∼25 μM. The binding of CO to the protein under aerobic conditions on the time scale of minutes to weeks was followed by recording the full spectra of solutions of the protein (∼20 μM) at various delay times after exposure to either 1 atm CO gas, a mixture of 0.1 atm CO and 0.9 atm air, or different concentrations of the CO-releasing molecule [Mn(CO) 4 {S 2 CNMe(CH2CO2H)}]23,24 (CORM-401). In some of these experiments, the redox mediator N,N,N′,N′-tetramethyl-p-phenylenediamine (TMPD) was added. Experiments were performed at room temperature, unless indicated otherwise.

NPAS2. In RcoM proteins, this sensor domain is linked to a LytTR DNA-binding domain. RcoM-1 and RcoM-2 are highly homologous (88% sequence identity) and were first described by Roberts and co-workers,10 with RcoM-2 being the best characterized to date.12,13 Although the structure of the heme-containing PAS sensor domain is still unknown, both heme coordination and environment display substantial differences with FixL-like PAS domains, pointing at different signal transmission mechanisms within the heme domain. Like that in EcDos, in the absence of external ligands, the heme cofactor in RcoM-2 is six-coordinated in both the ferrous and ferric states, with a conserved histidine as a proximal ligand. However, the distal heme ligand is a cysteine in the ferric form of RcoM-213 (H2O in EcDos14) and a methionine (Met104) in the ferrous form in RcoM-1; in the PAS fold, this methionine is not aligned with the methionine ligand in EcDos.10,15 In addition, whereas arginine in the distal pocket is thought to play a crucial functional role in FixL16−18 (and EcDos19), presumably no arginine is present in the distal pocket of RcoM proteins.10 A unique property of RcoM proteins is that they can be purified in the ferrous CO-bound form after standard aerobic expression in Escherichia coli. A non-CO-liganded form can then be obtained only after prolonged strong illumination under aerobic conditions.10 Thus, the protein has an extremely high affinity for CO, estimated at least in the low nanomolar range from titration in RcoM-1.20 By contrast, an O2-bound form cannot be observed stably or even intermittently. 10 Indeed, upon illumination of the CO-bound form under air, the Fe(II) form, in which the distal methionine replaced CO, is partly accumulated and then transforms to the Fe(III) state in hours,10 implying that any O2 binds extremely slowly, and considerably slower than the autoxidation rate. The properties of the protein, acting as a highly efficient CO scavenger, virtually without interference of molecular oxygen, consistent with CO-specific sensing under aerobic conditions, are most intriguing. Yet, the underlying kinetic properties of the interaction of the heme with ligands have not been determined yet, presumably due to the difficulty in dissociating CO from the protein. Here, focusing on the PAS heme-sensor domain, termed RcoMH-2, we set out to kinetically characterize the interaction with ligands under both anaerobic and aerobic conditions. We use ultrafast spectroscopy as an appropriate tool to study the ligand dynamics within the sensor domain21 as well as stopped-flow spectroscopy approaches to study bimolecular ligand binding. We uncover highly unusual kinetic properties, including a quasi-unity yield of CO rebinding to the heme on the picosecond time scale. These characteristics make RcoMH-2 a unique and convenient model system to study the intrinsic physical−chemical properties of heme−CO bond formation. Furthermore, our findings will be discussed in terms of the molecular function of the protein and of potential applications as an oxygen-compatible CO scavenger.

■

■

EXPERIMENTAL METHODS The heme-containing RcoMH-2 PAS sensor domain was heterologously expressed in E. coli BL21 DE3 from a pQE80L expression vector using 1 mM isopropyl-β-D-thiogalactopyranoside. RcoMH-2 (amino acids 1−160, expected MW 17.27 kDa) was purified by affinity chromatography on Ni-TED columns (Protino Ni-TED, Macherey Nagel) and eluted with 50 mM NaH2PO4, 300 mM NaCl, and 250 mM imidazole, at pH 8.0, followed by imidazol removal on Econo-Pac columns (Bio-Rad). Site-directed mutants were constructed using the QuikChange II site-directed mutagenesis kit (Agilent), following the manufac-

RESULTS Steady-State Absorption Spectroscopy. Figure 1 shows the absorption spectra of wild-type (WT) RcoMH-2 under different conditions. The purified heme-sensor domain and the subsequently prepared CO-bound form have virtually identical spectra, with maxima at 422, 540, and 570 nm, which are typical of the Fe(II)−CO complex.12 This implies that purified RcoMH-2 is predominantly in the CO-bound form, as was previously reported for full-length RcoM-110 but different from that reported for the full-length RcoM-2 protein, which could be purified directly in the B

DOI: 10.1021/acs.jpcb.6b08160 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 1. Absorption spectra of WT RcoMH-2, subsequently after purification, after prolonged illumination, after degassing and reduction with sodium dithionite, and after exposure to 1 atm CO. The Q-band region is shown on an expanded scale for clarity.

ferric state.12 The origin of this difference remains to be determined but may (a) reflect easier access and/or higher affinity of CO to the isolated heme domain or (b) be related to different E. coli growth conditions in both expression systems (cf. ref 20 on purification of RcoM-1 in this context). The Fe(II) state has the typical features of a hexacoordinate heme with intrinsic protein ligands, with maxima at 426, 530, and 561 nm, similar to those of full RcoM-2, in which the ligands are thought to be His74 and Met104 by analogy to RcoM-1. The Fe(III) form, obtained by extended illumination of the purified protein, has a broad maximum at 354 nm as well as a maximum at 534 nm and a shoulder at 560 nm in the α band; both features point to a hexacoordinate histidine−cysteine complex.13,25 The Soret band is relatively broad and has a maximum at 414 nm; it is blue-shifted with respect to that for the reported full-length RcoM-2,13 suggesting that in the Fe(III) form of the sensor domain a substantial heme fraction is five-coordinated.12 To investigate the effect of the potential heme ligand Met104 and its neighboring residue Met105 on the coordination properties of heme, we generated the variants M104H, M104I, M105I, and double mutant M104I/M105I. As in the case of the WT, all mutants were purified predominantly in the CO-bound state. The CO from the M104I mutant protein could only be partially photodissociated even upon prolonged illumination. This finding can be understood in terms of diminished competition between CO and an intrinsic ligand for binding to the heme, indicating that Met104 is the intrinsic distal ligand in the ferrous form, as that in the case of full-length RcoM-1.10,15 However, the steady-state absorption spectra of M104H, M105I, and M104I/M105I (not shown) are all very similar to those of the WT for the photodissociated ferric form and the six-coordinate ferrous form. This implies that, unlike in mutants of full-length RcoM-1 in which Met104 is modified to Leu,10,15 residues other than Met104 can also ligate to the heme in our RcoMH-2 constructs, in which Met104 is modified to Ile. The spectra of the CO-bound form were also very similar to those of WT for all mutants studied. CO Rebinding. Figure 2A shows the transient absorption spectra after dissociation of CO from the RcoMH-2-CO complex on the picosecond and nanosecond time scales. These spectra, with minima at 421 nm and maxima at 438 nm, are typical fivecoordinate minus six-coordinate (CO-bound) spectra for b-type

Figure 2. (A) Transient spectra at different delay times after dissociation of CO from the fully reduced and CO-bound RcoMH-2 complex. (B) Kinetics of CO rebinding, monitored at 438 nm. The red line is a fit. The inset compares the kinetics of the purified complex and the complex exposed to a CO atmosphere (black) to that of the complex prepared by prolonged illumination and subsequent reduction and exposure to CO (red).

hemes. After initial photophysical processes in the first few picoseconds, the shape of the spectra does not change but, remarkably, the amplitude of the transient spectrum fully (>99%) vanishes within a few nanoseconds. This implies that CO almost completely rebinds and that the yield of escape from the protein is extremely low. Such a highly efficient heme−CO geminate recombination has not been reported previously for any other heme protein (cf. ref 26). The corresponding CO rebinding kinetics (Figure 2B) can be fit with two exponentials of ∼170 ps (65%) and ∼500 ps (35%). Given the proximity of these time constants, this does not necessarily imply that rebinding occurs in two distinct phases but could also reflect a distribution of rate constants. Very similar results were obtained with samples of the complex directly as purified (inset, Figure 2B), in which the heme is also mainly in the CO-bound form but which contains no CO added in solution. Yet, whereas >99% rebinding is also observed here, the slower rebinding phases are somewhat suppressed. This feature can be explained by hypothesizing that the slower CO rebinding phases are correlated with those proteins in which the probability of CO escape is the highest and that are therefore more likely to be COC

DOI: 10.1021/acs.jpcb.6b08160 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B depleted prior to the experiment. This proposal is further supported by the finding that the amplitude of the dissociation signal decreases during the experiment with as-prepared RcoMH2 (not shown), due to the slow rebinding with any CO in the solution (see below). Generally, the CO recombination kinetics depends on the competition between CO escape from the heme pocket and its rebinding to the heme. In the case of RcoMH-2, the former process occurs with a rate orders of magnitude lower than that of the latter. This offers the unique opportunity to study the thermodynamic properties of the rebinding process without complications arising from the escape process. The RcoMH-2 sensor domain was found to be stable up to above 50 °C. Figure 3A shows the CO rebinding kinetics in the 8−50 °C temperature range. Remarkably, the kinetics slow down upon raising the temperature. Analysis in terms of a two-exponential fit indicate that the relative amplitude of the phases is almost temperatureindependent and that both rates diminish with temperature (inset Figure 3A), both corresponding to formal negative activation barriers of ∼70 meV (Figure 3B). Thus, slowing the motions of the protein environment allows the dissociated CO to reassociate faster with the heme cofactor. This conclusion is also supported by the observed viscosity dependence: in the presence of glycerol, the rebinding also accelerates (Figure 3C). In reduced unliganded RcoM-2, Met104 has been identified as the distal axial ligand of the heme iron.12 The effect of mutation of this residue and the adjacent Met105 residue to isoleucine, which cannot act as an axial ligand to the heme iron, on the CO recombination kinetics was investigated (Figure 4). The kinetics of the M104I, M104H, and M105I single variants and the M104I/ M105I double mutant were all highly similar to that of the WT, indicating little change in the distal heme pockets of the CObound forms. This interpretation implies that upon CO binding the distal ligand is displaced and moves out of the heme pocket. We note that whereas the M104I and M105I single mutants display similar biexponential kinetics as WT, those of the double mutant and the M104H mutant have a significantly broader distribution of rates (three rate constants, 80, 280 ps, and 1.4 ns are required to fit the latter). This finding suggests that these latter relatively stronger modifications induce a broader distribution of configurations of the distal heme pocket. NO Rebinding. Despite its function as a heme domain of a CO-activated transcriptional regulator, RcoMH-2 can also bind NO, in the ferrous12 as well as the ferric (Figure 5A, inset) state. The latter spectrum, which has not been reported previously for an RcoM protein, displays maxima at 416, 531, and 564 nm, similar to that of other Fe(III)−His−NO complexes.27−29 Figure 5 shows the transient spectra observed upon NO photodissociation and the associated NO recombination kinetics. In both cases, transient spectra reflecting formation of a fivecoordinate heme are observed. As is also observed in other heme proteins, the transient spectra display additional spectral redshifts that decay on the time scale of a few picoseconds, reflecting heme photophysical processes.26 Complete (>99%) recombination with the dissociated NO occurs on the picosecond time scale for both ferrous and ferric RcoMH-2. Such extensive NO recombination (in contrast to the extensive CO recombination reported above) is not unusual in heme proteins26 and is fully consistent with a closed heme pocket configuration. CO Binding from Solution. As binding of CO from solution to ferrous RcoMH-2 heme cannot be studied by flash photolysis because of the vanishingly low quantum yield of CO escape to the solvent, we employed stopped-flow spectroscopy. Spectral

Figure 3. (A) Temperature-dependent CO recombination in WT RcoMH-2. Inset: Parameters of two-exponential fits. (B) Arrhenius plots of the rate constant. The slopes correspond to activation energies of −75 ± 12 (k1) and −70 ± 20 (k2) meV. (C) Kinetics at 20 °C in buffer (black) and 50% glycerol (red).

changes solely associated with the replacement of the distal methionine by CO were observed, with no indication of a significant population of spectrally distinct intermediate species (Figure 6A, inset). In this case, the observed CO binding rate, kobs, can, in principle, be described by a hyperbolic curve as (cf. ref 30) kobs = D

k −Mk CO[CO] kM + k CO[CO]

(1) DOI: 10.1021/acs.jpcb.6b08160 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

the rate is expected to increase linearly with [CO], with an apparent bimolecular rate of kCO/KM, and at high [CO], it is expected to reach a limiting rate of k−M. We found that the kinetics, observed on the millisecond-tosecond time scale, depends on the CO concentration, but it is multiphasic (Figure 6A). It can be reasonably described by a biexponential function, with a fast component (∼30%, [CO]independent at least above 0.1 mM), with a rate of ∼20 s−1, and a slow component that approaches a limit at the highest CO concentration (0.5 mM) (Figure 6B). As a relatively simple way to account for such kinetic complexities, these findings can be described in terms of two different populations. Here, in ∼30% of the complexes, CO binding is limited by thermal dissociation of the distal methionine and its subsequent displacement out of the heme pocket (k−M) at 20 s−1 and kCO/KM > 105 M−1 s−1. In ∼70% of the complexes, a different methionine-displaced configuration is reached at a slower rate (∼5 s−1) in which CO binding is slower (kCO/KM ∼ 15 × 103 M−1 s−1). These findings imply that at low [CO] the effective bimolecular CO association rate exceeds 104 M−1 s−1. The dissociation rate could not be directly determined, but it must be very low in view of the state in which the protein is isolated. Under oxygenic conditions, at room temperature, purified RcoMH-2 remains in the CO-bound state for days; therefore, we estimate that koff ≪

Figure 4. Comparison of CO recombination in WT and mutant RcoMH2. The inset represents the same on a log absorption axis.

where k−M is the rate of dissociation and displacement of the distal methionine, kM is the rate of association of the distal methionine (KM = kM/k−M ≫ 1, as the protein is initially predominantly sixcoordinated), and kCO is the bimolecular rate constant for the binding of CO to the five-coordinate heme. Here, at low [CO],

Figure 5. NO recombination with ferric (A, B) and ferrous (C, D) WT RcoMH-2. The transient spectra (A, C) show bleaching of the NO-liganded forms and induced absorption of the five-coordinate unliganded forms, with maxima at ∼382 nm (ferric) and ∼438 nm (ferrous). In both cases, NO rebinds completely in a multiphasic manner; the fits in the kinetics have time constants of 3.6 ps (40%) and 37 ps (60%) (ferric; B) and 5.8 ps (95%) and 130 ps (5%) (ferrous; D), respectively. The inset of (A) represents the steady-state spectrum of the ferric NO complex. E

DOI: 10.1021/acs.jpcb.6b08160 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 7. Formation of the Fe(II)−CO complex from ferric RcoMH-2 (∼20 μM) at 37 °C, monitored by absorption spectroscopy. (A) Spectra at different delay times (in min) after addition of 100 μM CO (equilibration with 10% gaseous CO/90% air). The samples were not illuminated between the spectra. (B) Kinetics of Fe(II)−CO under different conditions. The lines correspond to fits to the exponential curves.

Figure 6. Kinetics of binding of CO to reduced RcoMH-2, as determined by stopped-flow spectroscopy. (A) kinetics after mixing at various [CO] and transient spectra with respect to the spectrum directly after mixing (inset). (B) Results of fits to biexponential kinetics as a function of [CO]. Note that the rates are on a logarithmic scale. The curve corresponds to eq 1, with k−M = 5 s−1 and (kCO/KM) = 15 × 103 M−1 s−1.

so by quasi-irreversibly binding CO during intervals in which the heme adopts the intrinsically short-lived ferrous state. We also investigated how CO delivered by CO-releasing molecules (CORMs) is captured by RcoMH-2. CORM-401 is a recently developed water-soluble tetracarbonyl compound that releases small quantities of CO in solution but easily transfers up to ∼3 equiv of CO per CORM-401 to CO-binding targets.23 Indeed, we observed the formation of the RcoMH-2−CO complex in the presence of this compound (Figure 7B). Interestingly, this process occurred substantially faster (∼7 h) than that upon equilibration with CO gas (∼5 days, see above). This finding suggests that CORM-401, in addition to its COreleasing abilities, also has a modest reducing effect, thus facilitating CO uptake by RcoMH-2. In the presence of 100 μM TMPD, the process speeds up to ∼21 min. Remarkably, in this case, CO binding occurs slower than than in the presence of CO gas and is independent of the CORM-401 concentration. We suggest that in this case the CO transfer process is rate-limited by the formation of an RcoMH-2−CORM-401 complex.

10−6 s−1. Taken together, a conservative lower limit for the affinity of CO for RcoMH-2 is 1010 M−1. CO Binding in the Oxidized Form. The RcoMH-2 heme domain is purified, under aerobic conditions, predominantly in the ferrous Fe(II)−CO form. This finding (suggesting that heme has a relatively high redox potential) raises the question of whether this complex can also be formed starting from the ferric Fe(III) form and in the presence of O2. The data in Figure 7 show that this is possible. Upon equilibration of the ferric complex with 1 atm gaseous CO, the Fe(II)−CO complex is slowly formed, at 37 °C, with a time constant of ∼5 days. As CO can bind only to the reduced ferrous heme, this reaction presumably occurs via a transient population of the unliganded ferrous state. The reaction speed was found to be insensitive to the presence of O2, and under conditions of excess CO, it was found to be virtually [CO]independent (Figure 6B), indicating that the frequency of redox change is rate-limiting. Indeed, upon addition of the redox mediator TMPD (100 μM), which facilitates transient reduction of the heme while maintaining a predominantly Fe(III) complex in the absence of CO, the formation of the Fe(II)−CO complex is accelerated by ∼3 orders of magnitude to ∼11 min. Altogether, these results indicate that because of its high CO affinity RcoMH2 can form a CO complex even in the presence of oxygen when the unliganded complex is almost exclusively in the ferric form. It does

■

DISCUSSION In the present work, we have investigated the highly unusual ligand-binding and -release properties of the CO-sensor domain of the transcriptional regulator RcoM-2 by determining the kinetics, spanning ∼16 orders of magnitude in time. We found that dissociation of NO and CO from the sensor heme leads to virtually complete geminate heme−ligand rebinding within a few hundred picoseconds (NO) or a few nanoseconds (CO), with an extremely low yield of escape of the gaseous ligand from the F

DOI: 10.1021/acs.jpcb.6b08160 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

heme normal, rendering the binding probability likely more sensitive to orientational fluctuations. We suggest that our finding of substantial slowing of the heme−CO binding with increasing temperature in a highly reactive system implies that the photodissociated CO is initially trapped in a near-perpendicular orientation within a heme pocket, leaving little orientational freedom. Modest acceleration of CO rebinding with decreasing temperature has also been reported by Benabbas et al. for CooA, at a lower temperature range (200−300 K).34 These authors interpreted this effect as a temperature-dependent distribution of heme configurations affecting the heme−CO reactivity. For our results on RcomH-2, the above interpretation of the effect of temperature in terms of protein motions influencing the CO configuration is supported by the observed changes upon raising the viscosity, which also arrests protein motions. Binding of CO from the solvent to the reduced heme is limited, at high CO concentration, by methionine dissociation and protein rearrangement processes occurring on the millisecond time scale, as seen in many other six-coordinate heme proteins. We found (Figure 6) that the binding kinetics is heterogeneous in RcoMH2, but the effective bimolecular association constant is always >104 M−1 s−1; in a substantial fraction (∼30%), it exceeds 105 M−1 s−1. In view of the extremely low off rate, we determined a conservative lower limit for the affinity of 1010 M−1 (Kd sub-100 pM). This lower boundary well exceeds the usual range for heme proteins, which extends from ∼105 to 109 M−1. In particular, the CO sensor CooA has a relatively low affinity (∼5 × 105 M−1 38), notwithstanding the substantial geminate CO recombination.22,34,35 Some truncated hemoglobins, in which geminate recombination also occurs,26,32 display a CO affinity of the order of 109 M−1,39 with that of Nostoc commune being the highest, determined at 4 × 109 M−1, 40 due to its high ligand-association rate. It should further be emphasized that the CO affinity of RcoMH-2 is orders of magnitude higher than that of the PAS domains of the sensor proteins FixL (∼105 M−1 41) and EcDos (effectively ∼3 × 106 M−1 42). The affinity of RcoMH-2 for CO is exceptionally high. The present lack of structural information prohibits a detailed discussion of the molecular origin of this feature. However, the properties of the distal heme pocket are clearly distinct from those of other heme-containing PAS domains. If similar properties would hold for the full-length RcoM-2 protein, then the high affinity and extremely low dissociation rate constant would raise questions about its functioning as a CO sensor. The sensor would be activated in the presence of low concentrations of CO, but it is unclear how it could be subsequently deactivated in the absence of CO. However, the presence of the DNA-binding domain, with which allosteric interactions necessarily do exist, as well as other cellular constituents (including target DNA) interacting with RcoM proteins, may influence the effective ligand interaction parameters of the heme domain. For instance, in CooA, DNA binding has been shown to alter the heme−CO rebinding kinetics in a manner that suggests rigidification of the heme environment.34 The CO-binding properties of full-length RcoM proteins will be investigated in future work. In addition to its physiological significance, the unusual properties of RcoMH-2 may be exploited in applications related to CO signaling and toxicity. For instance, RcoMH-2 is a very efficient biocompatible scavenger of CO, even under oxygenic conditions (Figure 7). Interestingly, a water-soluble porphyrin complex, hemoCD (Kd CO ∼50 pM), has been recently

complex. Whereas such properties are not unusual for the highaffinity ligand NO,31 escape yields lower than 1%, as we determined for the RcoMH-2−CO complex, have not been reported earlier for native ligand-binding heme protein complexes. Escape yields of 2 × 109 s−1 and efficiency >99%, we deduce that the rate of CO escape from the RcoMH-2 heme pocket is