Photoacoustic Calorimetry Studies of O2-Sensing FixL and (R200, I209) Variants from Sinorhizobium meliloti Reveal Conformational Changes Coupled to Ligand Photodissociation from the Heme-PAS Domain

FixL is an oxygen-sensing heme-PAS protein that regulates nitrogen fixation in the root nodules of plants. In this paper, we present the first photothermal studies of the full-length wild-type FixL protein from Sinorhizobium meliloti and the first thermodynamic profile of a full-length heme-PAS protein. Photoacoustic calorimetry studies reveal a quadriphasic relaxation for SmFixL*WT and the five variant proteins (SmFixL*R200H, SmFixL*R200Q, SmFixL*R200E, SmFixL*R200A, and SmFixL*I209M) with four intermediates from <20 ns to ∼1.5 μs associated with the photodissociation of CO from the heme. The altered thermodynamic profiles of the full-length SmFixL* variant proteins confirm that the conserved heme domain residues R200 and I209 are important for signal transduction. In contrast, the truncated heme domain, SmFixLH128–264, shows only a single, fast monophasic relaxation at <50 ns associated with the fast disruption of a salt bridge and release of CO to the solvent, suggesting that the full-length protein is necessary to observe the conformational changes that propagate the signal from the heme domain to the kinase domain.


■ INTRODUCTION
A large number of heme proteins have now been discovered in nature that sense O 2 , NO, or CO. 1−8 These proteins, called heme-based gas-sensing proteins, are found in all kingdoms of life and regulate many important biological processes, such as blood pressure, circadian rhythm, nitrogen fixation, chemotaxis, and photosynthesis. 1−20 The FixL protein was the first oxygen-sensing heme protein discovered in nature and has been an important paradigm for the structure and function of the heme-PAS family of gas-sensing heme proteins. 1−24 FixL from B. japonicum has a second N-terminal PAS domain instead of an N-terminal membrane anchor.The kinase domain of SmFixL contains a histidine285 (His291 in BjFixL) that is autophosphorylated by ATP (Figure 1).Under hypoxic conditions in the root nodules, O 2 is released from the heme of dimeric SmFixL, activating the histidine kinase domain, which transfers a phosphate group from histidine285 to an aspartate54 residue on the dimeric response regulator SmFixJ.−17 The O 2 analogs NO and CO can also bind to the heme of reduced, deoxySmFixL and inhibit the kinase domain, but to a lesser extent than O 2 . 25−37 The heme domains of SmFixL and BjFixL are ∼53% identical in an amino acid sequence based on UniProt, while the residues in the heme pocket are ∼70% similar.−37 The

Biochemistry
BjFixLH 140−270 and SmFixLH 127−260 heme domains are dimeric in solution, analogous to the full-length FixL proteins. 38,39−36,40−44  , there is a rearrangement of the hydrogenbonding network between the heme 6,7 propionates and the conserved Arg206 (Arg200 in SmFixL) and His 214 amino acid residues due to the flattening of the heme after binding of the ligand. 27,28The BjFixLArg206Ala variant resulted in a significant reduction of the kinase inhibition by O 2 , suggesting that this conserved arginine residue is important for signal transduction. 27,28Balland et al. showed that the conserved Arg220 from BjFixL (Arg214 in SmFixL) is important in the oxygen binding mechanism, as it interacts with O 2 bound to the heme iron. 40The displacement of Arg220 or the coordination of a strong field ligand to the heme iron, however, is not responsible for the conformational changes but only the formation of a strong hydrogen-bonding network between Arg220 and the ligand O 2 . 40On the other hand, Miyatake et al. have observed that the hydrophobic residues in the distal heme pocket, which likely interact with O 2 , play an important role in signal transduction between the heme and kinase domains. 26Thermodynamic profiles for CO photodissociation from the truncated heme domain of BjFixLH 140−270 by Miksovska et al. demonstrated a biphasic relaxation. 41The first phase involved a contraction of the solvent associated with a change in charge distribution after the reorganization of the salt bridge between Glu182 and Arg227 or a possible reorientation of Arg206 after photodissociation of CO from the heme.The second phase with a lifetime of 150 ns was attributed to an expansion due to the ligand being released to the solvent. 41Ayers and Moffat, and Cusanovich and Meyer, suggested that the signal originated from the heme domain can be propagated through the linker to the kinase domain by quaternary structural changes via a distortion of the β-sheet based on time-resolved crystallographic studies of the BjFixLH heme domain upon CO photolysis. 42,43Finally, Reynolds et al. demonstrated the importance of the conserved, proximal Arg200 in the stabilization of the kinase inhibition related to the oxy form of SmFixL. 44They showed a clear relation between the H-bond of Arg200 and the 6-propionate heme group that stabilized the inactive form of SmFixL upon O 2 binding to the heme.
Although FixL and other heme-based gas-sensing proteins have been proposed to transduce their signals through conformational changes coupled to ligand binding or dissociation, the details of these conformational changes are still being determined. 1−48 In this report, PAC studies are used to generate the thermodynamic profiles for the CO ligand escaping from the heme of the following proteins: the truncated heme domain SmFixLH 128−264 , SmFixL* wild-type with the heme, coiled-coil linker, and histidine kinase domains, but with the N-terminal membrane anchor removed for solubility, and four different SmFixL*variants R200A(Alanine), R200Q(Glutamine), R200E-(Glutamate), and R200H(Histidine).These SmFixL* variant proteins, each prepared in pH 7.8 Tris buffer, are presented in order to have a range of polarities and H-bond abilities to compare to those of wild-type SmFixL*.Another SmFixL* variant, I209M(Methionine), was also studied as Mukai et al. have demonstrated that the conformational changes associated with Ile209 and Ile210 are involved in the kinase activity of SmFixL. 49The current paper describes the first thermodynamic study of a full-length FixL protein with the heme-PAS, coiled−coil linker, and kinase domains and the first photothermal study of a full-length heme-PAS protein.

Site-Directed Mutagenesis of SmFixL*. Wild-type
SmFixL* and SmFixL* variant proteins were prepared as previously described. 44rotein Expression and Purification of Truncated Heme Domains, SmFixLH 128−264 and SmFixL*.The truncated heme domain SmFixLH 128−264 was expressed, purified, and characterized as previously described. 50Wildtype and variant SmFixL* with the N-terminal heme-PAS, the coiled-coil linker, and the kinase domain but without the Nterminal membrane linker (AA's 1−126) was expressed in Escherichia coli and purified as previously described. 44Protein samples were exchanged into the appropriate buffer, 20 mM Tris-HCl, 0.1 M NaCl, pH 7.8, using either a 3 or 50 mL Millipore Stirred Ultrafiltration cell under compressed nitrogen and stored in aliquots at −80 °C.
Protein Characterization.Wild-type and variant SmFixL* proteins were separated on 12 or 15% PAGE Gold Precast Gels from Lonza to assess purity and visualized with Coomassie Brilliant Blue (Biorad) staining using Colorburst (Sigma-Aldrich) electrophoresis markers.The heme content of each batch of purified protein was assayed by the standard pyridine hemochromagen assay using the extinction coefficient of the reduced pyridine complex (34.4 mM −1 cm −1 ) at 557 nm. 51SmFixLH 128−264 and SmFixL* proteins were also characterized by MALDI-TOF mass spectrometry to confirm that the proteins were intact and pure. 44,50ample Preparation for PAC Measurements.Samples for PAC were prepared by diluting SmFixL* and SmFixLH 128−264 proteins into a buffer containing 20 mM Tris (pH 8.0).The deoxy form of the protein was formed by placing the oxy form of each SmFixL protein in a quartz optical cuvette that was then sealed with a septum cap and purged with Argon gas.A fresh dithionite solution was added from a buffered stock solution to obtain a final concentration of ∼6 μM.The CO-bound form was obtained by saturating solutions of the deoxy SmFixL* of SmFixLH 128−264 with CO, resulting in a final solution CO concentration of 1 mM (1 atm pressure).The protein concentration for PAC samples was ∼5 μM and that for transient absorption was ∼9 μM.
PAC Methods.PAC measurements were performed by placing a 1 cm × 1 cm quartz cuvette containing 1.0 mL of a sample in a temperature-controlled sample holder (Quantum Northwest) housing a Panametrics V103 transducer.Contact between the cuvette and the detector was facilitated by a thin layer of vacuum grease.Photodissociation of CO was achieved with a 532 nm laser pulse (Continuum Minilite I frequency double Q-switched Nd:YAG laser, 6 ns pulse, < 80 μJ).The acoustic signal was amplified with an ultrasonic preamp (Panametrics) and recorded using an NI 5102 oscilloscope (15 MHz) controlled by VirtualBench software (National Instrument).The PAC data were analyzed using the multiple temperature method, in which sample and calorimetric reference acoustic traces are obtained as a function of temperature.The ratio of the amplitudes of the sample and reference acoustic signals (φ) was then plotted versus C p ρ/β according to where ϕ is the ratio of the acoustic amplitudes for the sample and reference (i.e., ϕ = {A S /A R }), Q is the heat released to the solvent, β is the coefficient of thermal expansion of the solvent (K −1 ), C p is the heat capacity (cal g −1 K −1 ), ρ is the density (g mL −1 ), and ΔV represents conformational/electrostriction contributions to the solution volume change.A plot of ϕE hν versus C p ρ/β (varied by changing the solution temperature) gives a straight line with a slope equal to ΔV and an intercept equal to the released heat (Q).Subtracting Q from E hν and dividing by the quantum yield gives ΔH (i.e., ΔH = {E hν − Q}/Φ) for processes occurring faster than the time resolution of the instrument (<20 ns).The Q/Φ values for subsequent kinetic processes represent −ΔH for that step (i.e., heat released).For kinetic events occurring slower than the response time of the piezoelectric crystal (between ∼20 ns and ∼20 μs), the resulting sample acoustic wave is shifted in frequency relative to the reference waveform.In order to extract the relevant Q i , ΔV, and rate constants (k i ) corresponding to the observed kinetic processes, the sample acoustic wave, E(t) obs , is treated as a convolution of an instrument response function, T(t), and time-dependent heat generating function, H(t), according to where In practice, the instrument response function T(t) is taken to be the calorimetric reference waveform.The amplitudes, ϕ i , and lifetimes, t i , of the resolvable kinetic processes were extracted by using a simplex parameter estimation algorithm within software developed in our laboratory.Subtracting Q P (obtained from the <20 ns phase) from E hν and scaling to the reaction quantum yield (Φ) gave the reaction enthalpy (ΔH P ) for the initial (prompt) phase of the reaction, while Q i = −ΔH i for each additional step resolved in the deconvolution.The corresponding ΔV values for the SmFixL* proteins and the truncated SmFixLH 128−264 heme domain were obtained from the slopes of the plots described in eq 1 (with the ΔV for the prompt phase being the slope/Φ).The number of kinetic phases is determined through the evaluation of multiparameter fits.Parameter sets for kinetic phases are added until convergence is reached in the χ 2 values, residuals, and autocorrelation.An example is displayed in the Supporting Information.
Transient Absorption Methods.Transient absorption (TA) experiments were performed by monitoring the change in intensity of light from a Xe arc lamp (Oriel) emerging from the sample followed by passage through a 1/4 m single monochromator equipped with an Oriel R928 photomultiplier tube.The signal was amplified using a home-built preamplifier (1 MHz bandwidth) followed by a Stanford Instruments SR445A 350 MHz postamplifier.The signal was digitized using a Tektronix TDS7404 4 GHz digital oscilloscope.The sample was excited with the second harmonic of a continuum Leopard I Q-switched mode-locked Nd:YAG laser (<20 ps, 20 mJ/ pulse, 20 Hz).Signal traces are the average of 20 laser pulses.

■ RESULTS
Optical Absorption Spectroscopy.The UV−vis spectra of SmFixL*WT and R200Q in the as-isolated, the reduced, the deoxy Fe(II)state, and the CO-bound states are shown in Figure 3.
The Soret and Q-band wavelengths for the wild-type SmFixL*, the R200 variants, the I209M variants, and the truncated heme domain, SmFixLH 128−264 , are summarized in Table 1.The SmFixL* wild-type and variant proteins (R200A, R200Q, R200E, and R200H) have similar optical spectra, regardless of the change in residue.The optical spectra of the as-isolated SmFixL*WT and R200 variant proteins have Soret bands at ∼418 nm and Q-band at ∼543 and 577 nm indicative of the oxy state.In addition, the as-isolated form of the R200 Biochemistry variants (R200Q/H/A/E) has shoulders at ∼395 nm characteristic of the oxidized, Fe(III) state, which is shown in Figure 3, for R200Q.These R200 variants were previously shown to autoxidize in air during aerobic purification. 44The SmFixL*WT and the R200 variant proteins can be reduced to the reduced, Fe(II) state with sodium dithionite, resulting in Soret bands blue-shifted to ∼434 nm with a broad Q-band at ∼566 nm.The binding of CO to the reduced, deoxy Fe(II) SmFixL*WT and R200 variant enzymes results in Soret bands at ∼424 nm and Q bands at ∼542 and ∼572 nm (Figure 3).The fact that the optical absorption spectra of the various forms of SmFixL*WT and R200 variants (R200A, R200Q, R200E, and R200H) are independent of the nature of the variations associated with a range of polarities and H-bond abilities indicate that the salt bridge between the heme-6propionate and R200 within the distal pocket near the heme group does not have a significant impact on the electronic structure of the heme group.In addition, the truncated heme domain, SmFixLH 128−264 , does not significantly alter the electronic properties of the optical spectra (Figure 4 and Table 1).
Interestingly, while the SmFixL*I209M variant has similar visible spectra in the reduced, Fe(II) and CO-bound states as SmFixL*WT, the as-isolated SmFixL*I209M variant has a Soret band at ∼434 nm and a broad Q-band centered at ∼557 nm that is indicative of the reduced, Fe(II)state (Supporting Information Figure S1 and Table 1).Thus, as-isolated SmFixL*I209M does not appear to form the same O 2 complex as SmFixL*WT.It is possible that there is a small amount of the oxy state of SmFixL*I209M that is not readily observable by UV−vis spectrosocopy or that some reduced, Fe(II) SmFixL*I209M forms an oxy complex that is rapidly autooxidized to the Fe(III) state, since we see a small shoulder at ∼395 nm that is indicative of oxidized, Fe(III)SmFixL*.The

Biochemistry
intense 418 nm Soret peak that is characteristic of the oxy state is not observed, while the CO-bound state of SmFixL*I209M is similar to that of SmFixL*WT(CO) with a peak at 424 nm (Table 1).It appears that while the I209M variant readily binds CO, the change to a larger, more polarizable methionine sulfur group significantly lowers the affinity of reduced Fe(II) SmFixL*I209M for O 2 .This suggests that the conserved isoleucine 209 residue in the distal pocket of SmFixL* plays an important role in O 2 binding to the heme iron.We are in the process of further characterizing this interesting SmFix-L*I209M variant protein.
Transient Absorption.Photolysis of CO from SmFixL*WT(CO) and the five different variants (R200A, R200Q, R200E, R200H, and I209M) probed at 450 nm results in the formation of a five coordinate high-spin heme complex, which decays back to the preflash CO-bound complex (Supporting Information Figure S2) with monophasic relaxation kinetics.Table 2  The fact that a frequency shift is observed between the sample and reference acoustic signals indicates that kinetic events occur between <20 ns and ∼20 μs.The deconvolution of the acoustic wave between the sample and the reference results in four kinetic phases with average lifetimes of <20 ns (prompt phase), ∼190 ns, ∼512 ns, and ∼1.5 μs.The reaction volume and enthalpy changes were calculated using an Φ of 0.86 determined by Rodgers et al. 52 A plot of φE hν versus C p ρ/ β for the four observed intermediates is shown in Figure 7 for full-length SmFixL*WT.
The calculated ΔH and ΔV values for each intermediate associated with CO photodissociation from SmFixL*WT, the R200 and I209M variants, and the truncated heme domains of SmFixLH and BjFixLH are shown in Table 3.The additional plots of φE hν versus C p ρ/β for the SmFixL* variants are displayed in the Supporting Information.

■ DISCUSSION
Previous studies using time-resolved crystallography demonstrated that within 1 μs after CO photolysis, the truncated heme domain, BjFixLH 141−270 , has relaxed to a conformation of the protein, which is identical to the deoxy Fe(II) state. 30hese studies also demonstrated that the transmission of the signal after photodissociation of CO is not restricted to a single region surrounding the heme but to an ensemble of regions, including the FG loop and the β-sheet distal of the heme involving the movement of Arg206.Photodissociation of CO from BjFixLH also results in the doming and displacement of the heme, and the collapse of the hydrophobic residues Leu 236, Ile 215, and Ile 238 in the distal pocket in order to fill the space left after CO leaves the heme pocket.Furthermore, the displacement of the heme-6-propionate group and the FG loop residues Pro212, His213, and Ile216 is notable, as are the conformational changes in the proximal histidine and the Fα-    Biochemistry solvent, following a perturbation to the salt bridge between Glu182 and Arg227. 41The corresponding slow phase with a lifetime τ ∼ 150 ns and a ΔH of ∼6 kcal mol −1 and ΔV of ∼5 mL mol −1 was associated with the escape of the CO molecule to the solvent. 41In contrast, the thermodynamics of CO release from SmFixLH 128−264 is more similar to photolysis from the more truncated heme domain BjFixLH 151−256 with ΔH ∼ 5 kcal mol −1 and ΔV ∼ 25 mL mol −1 (Table 3).The truncated BjFixLH 151−256 heme domain has an additional 11 amino acid residues deleted from the N-terminus and 14 amino acid residues deleted from the C-terminus.Miksovska et al. concluded that the truncation of these additional N-terminal and C-terminal amino acids from the heme-PAS domain of BjFixLH 151−256 induces changes associated with the protein surface that accelerate the release of the ligand from the protein and/or change the salt-bridge interactions. 41The results obtained for the variation of volume are similar between BjFixLH 151−256 and SmFixLH 128−264 , but the variation in enthalpy is nearly 2-fold lower, which may arise from the fact that the salt bridge does not involve the same amino acids for the different proteins.
Full-length SmFixL*.Photolysis of CO from the heme of SmFixL*WT with the heme, coiled-coil linker and kinase domains, results in four intermediates between <20 ns and ∼2.0 μs with lifetimes of <20 ns (prompt phase), ∼ 190 ns, ∼ 512 ns, and ∼1.5 μs.With regard to the prompt phase (<∼20 ns), the volume and enthalpies are distinct from those observed for the truncated heme domain, SmFixLH 128−264 , with a ΔH of ∼ -9 kcal mol −1 for the full-length SmFixL*WT protein versus ΔH of ∼9 kcal mol −1 for the SmFixLH 128−264 heme domain and ΔV of ∼10 mL mol −1 for the full-length SmFixL*WT protein versus ΔV of ∼21 mL mol −1 for SmFixLH 128−264 , suggesting that the initial response of the heme domain to ligand release is linked to conformational changes involving both the coiled-coil linker and kinase domains.−60 The time-resolved transient grating studies assigned the ∼180 ns lifetime to the migration of the photocleaved CO to a hydrophobic Xe binding pocket near the heme iron made up of distal residues Leu29, Ile107, and Val68. 60−60 It is possible that CO migration occurs within the heme domain of SmFixL* with the conserved hydrophobic residues of the distal pocket (Ile210, Val232, Ile209, and Leu230), forming an initial transient CO binding site (Figure 8).
The initial migration of CO from the heme to the hydrophobic pocket would then be followed by the migration of CO through this cavity to the bulk solvent.The endothermic nature of this ∼190 ns CO migration phase indicates that this is largely entropically driven, while the corresponding CO release to the bulk solvent at ∼512 ns is an enthalpy-driven event that likely involves protein reorganization.The relatively small volume change for CO release to the solvent (∼+35 mL mol −1 ) is likely due to the corresponding  4) at ∼1.5 μs, which could result in a (b) righthanded supercoiled rotation of the histidine kinase domain of SmFixL*, similar to the one observed at ∼2 μs for the chimeric YF1 protein that contains the histidine kinase domains of BjFixL. 64In step 5, the right-handed supercoiled rotation could cause a rearrangement of the histidine kinase domains of SmFixL*, as observed in the chimeric YF1 protein after 250 ms, to form kinase active, deoxy Fe(II)SmFixL*. 64−17 In step 6, additional CO can be added to reform kinase inactive SmFixL*(CO) through additional conformational changes.

Biochemistry
entry of a water molecule to the pocket (∼ − 18 mL mol −1 ).The fact that these phases which are assigned here to heme domain CO release events only occur in the full-length SmFixL* protein, and not with the truncated heme domain SmFixLH 128−264 , suggests that the coiled-coil linker region connecting the heme domain to the kinase domain may be essential to regulating ligand binding and release and couple these events to histidine kinase domain conformational changes.Recent structural and site-directed mutagenesis studies of the full-length FixL 2 -FixJ 2 dimer from B. japonicum showed that the coiled-coil linker domain is important for signal transduction between the heme-PAS sensor and the histidine kinase domains. 32Changing 11 of 12 different amino acids in the coiled-coil linker of full-length BjFixL resulted in significantly impaired kinase activity. 32−63 The slowest phase of CO photodissociation in SmFixL* occurs on a time scale similar to that reported for O 2 photodissociation obtained from time-resolved UV resonance Raman spectroscopy (1.5 μs from PAC versus 3 μs from UV resonance Raman). 36Interestingly, the UV resonance Raman study does not show such a phase for CO photodissociation.It is possible that signal transmission from the heme domain may involve different pathways depending on the ligand type.The UV resonance Raman probed the Tyr8a band of Tyr201 that forms a hydrogen bond with Glu234.Upon photodissociation of O 2 , this H-bond is disrupted, which contributes, in part, to the signal migration. 36The 1.5 μs slow phase of SmFixL* also correlates with recent time-resolved solution X-ray scattering experiments with the chimeric YF1 protein that contains a LOV photosensor PAS domain, a coiled-coil linker, and the histidine kinase domain of FixL from B. japonicum that is kinase active in the dark. 64They found that the coiled-coil linker and histidine kinase domain of dimeric YF1 undergo a left-handed supercoiled rotation that is complete 2 μs after blue light excitation, which triggers a rearrangement of the histidine kinase domains after 250 ms and results in kinase inactivation. 64In contrast, a H22P variant of YF1 is kinase inactive in the dark state but becomes kinase active after exposure to blue light with inverted sign polarity, suggesting that the H22P YF1 variant instead undergoes a right-handed supercoiled rotation which activates the kinase domains. 64,65If SmFixL behaves in a manner similar to that of the H22P YF1 variant protein, we would predict that the loss of CO from the heme of SmFixL would result in a right-handed supercoiled rotation activating the histidine kinase domains.Our PAC experiments suggest that the initial conformational changes we observe upon CO photolysis from the heme domain of SmFixL* are propagated to the histidine kinase domain through the coiled-coil linker domain to form a similar ∼1.5 μs intermediate.More studies are needed to determine whether FixL undergoes the same supercoiled rotation and histidine kinase rearrangement as the chimeric YF1 protein and whether this regulates the histidine kinase activity of FixL.
SmFixL*R200 and I209 Variants.The corresponding thermodynamics associated with SmFixL*WT and the R200A, R200Q, R200E, R200H, and I209M variants all display four kinetic phases with lifetimes similar to that observed for the SmFixL*WT protein (Table 3).As can be seen from Table 3, all full-length SmFixL* variant proteins display exothermic prompt phases (τ < 20 ns), followed by endothermic ligand migration phases (τ avg ∼ 104 ns), exothermic ligand release phases (τ avg ∼ 417 ns), and endothermic conformational change phases (τ avg ∼ 1.49 μs).However, the ΔH values vary significantly among the protein variants.The initial phases for the R200A and R200E have similar ΔH values to WT, while the R200Q, R200H and I209M variants all have significantly lower values.This suggests minimal perturbation to the hydrogen-bonding network surrounding the heme-6-propionate network for the R200A and R200E variants, but some disruption occurs for the R200Q and R200H variants.The thermodynamics associated with the variants further demonstrate that the R200 and heme-6-propionate H-bond network may also play a role in stabilizing the transient CO binding to the distal hydrophobic pocket with the most pronounced perturbation from the I209M variant, as might be expected as this residue contributes to the putative transient CO binding pocket.Interestingly, the R200A, R200Q, and I209M variants have much lower ΔH values, indicating that a more entropically driven process drives the conformational response, while the opposite is true for the R200E and R200H variants.
The corresponding volume changes are also widely variable between the variants and the WT protein.Volume changes largely arise from protein−solvent interactions and are most influenced by changes in the overall surface charge.The magnitude of molar volume changes is dependent largely on the extent of changes in surface charge exposure.In the case of the SmFixL* variants, the R200A, R200Q, and R200E variants give rise to increase in ΔV upon initial CO photolysis, which is likely due to surface charge reduction (i.e., reducing solvent− residue electrostriction) and is consistent with the modest changes in the ΔH values (i.e., greater electrostriction results in greater charge stability and more negative ΔH values).In contrast, the R200H and I209M variants have lower ΔV values, indicating more exposed charge and more negative ΔH values.The ΔV associated with migration of CO within the protein interior for the R200Q and R200A variants is similar to that of WT, while the remaining variants have less negative ΔV values consistent with overall charge reduction.The CO release process gives rise to ΔV values similar for all variants except for R200A.Interestingly, the longer time scale conformational change gives ΔV values similar to WT (R200A and I209M) and values more positive than WT (R200E, R200H, and R200Q).These trends also track well with the observed ΔH values.
Although the time-resolved thermodynamics presented here lack the atomic level detail required for detailed mechanistic analysis, they do provide important insights into the mechanism of FixL signaling.The data presented here for full-length SmFixLWT* and five variant proteins are the first to demonstrate multiphasic response to ligand release from the heme domain with four intermediates observed from <20 ns to ∼1.5 μs.These studies also provide important thermodynamic boundaries for future studies of the signal transduction mechanism of FixL and other heme-PAS proteins.

Biochemistry
SmFixL*R200H, and SmFixL*I209M) with four intermediates from <20 ns to ∼1.5 μs associated with the photodissociation of CO from the heme.The thermodynamic profiles of the R200 and I209M variant proteins confirm that these conserved residues are important in the transmission of the signal and that changing either residue results in altered intermediates, which may cause a different transmitted signal.The results for the truncated heme domain SmFixLH 128−264 instead show a monophasic relaxation at <50 ns associated with a fast disruption of the salt bridge and release of CO to the solvent, suggesting that the full-length protein is necessary to observe the conformational changes that are transmitted from the heme domain through the coiled-coil linker to the kinase domain.These are the first studies to observe multiple conformational changes in a full-length FixL protein and are a model for the ubiquitous heme-PAS and histidine kinase families of signaling proteins.
Additional figures including the electronic spectrum of as-isolated SmFixL*I209M, the transient absorption data for CO recombination, the photoacoustic waves of SmFixL* variant proteins and the plots of (S/R)* Ehν versus C p ρ/β for CO photolysis f rom SmFixL* variants and a fit of the SmFixLWT* PAC data (PDF)

Figure 1 .
Figure 1.Schematic representation of the domain structures of full-length and truncated versions of the FixL protein.(A) Full-length FixL from S. meliloti containing the N-terminal membrane anchor, the N-terminal heme-PAS domain, the coiled linker, and the histidine kinase domains.(B) Full-length SmFixL* 127−505 which contains the heme-PAS domain, the coiled-coil linker, and the kinase domain but is missing the N-terminal membrane anchor for solubility and instead has an N-terminal extension: (N-TMITPSLAAGR(127−505)-C). 20(C) The truncated heme domain of SmFixLH 128−264 .(D) Full-length BjFixL from B. japonicum containing the PAS-A and heme-PAS-B domains, the coiled-coil linker, and the histidine kinase domain.(E) The truncated heme domain BjFixLH 140−270 .(F) The response regulator FixJ from S. meliloti is phosphorylated by deoxy SmFixLWT and deoxy SmFixL*.(G) FixJ from B. japonicum is phosphorylated by deoxy BjFixLWT.
summarizes the rate constants for CO rebinding to the previously reported truncated heme domains of BjFixLH 140−270 , and BjFixLH 151−256 , and the fulllength SmFixL*WT and the five variants R200A, R200Q, R200E, R200H, and I209M.The truncated heme domain BjFixLH 140−270 had a CO rebinding rate constant equal to 10.2 ± 0.3 s −1 , while the shorter truncated heme domain, BjFixLH 151−256 , was equal to 17.3 ± 0.1 s −1 . 41Full-length SmFixL*WT with the heme and kinase domain and the five different variants had rate constants between 33 and 41 s −1 .The results show a faster CO rebinding rate constant for fulllength SmFixL*WT and the R200 variants compared to the one observed for the truncated heme domains of BjFixLH 140−270 and BjFixLH 151−256 .These results indicate that the full-length SmFixL* protein accelerated the rebinding of CO, and they also demonstrate that a change at R200 does not significantly affect the rebinding of CO to the heme relative to SmFixL*WT.The rate constant associated with CO rebinding to SmFixL*I209M is slightly slower than SmFixL*WT, which may be due to the fact that the methionine sulfur group could interact with the CO molecule, compared to the alkyl group in isoleucine, which has a steric repulsion with CO.It is also possible that differences in the hydrogen-bonding networks in the heme pockets of full-length SmFixL* and the truncated heme domains of BjFixLH could affect the CO rebinding rates observed.Photoacoustic Calorimetry.Figures 5 and 6 display representative PAC traces for CO bound to SmFixL*WT and SmFixLH 128−264 and the calorimetric reference compound Fe4SP obtained in 20 mM Tris buffer, pH 8.0.

Figure 5 .
Figure 5. Overlay of the acoustic waves for the photolysis of CO from SmFixL*WT (red solid line) and the reference Fe(III)4SP (blue solid line).

Figure 6 .
Figure 6.Overlay of the acoustic waves for the photolysis of CO from the truncated heme domain SmFixLH 124−264 (red solid line) and reference Fe(III)4SP (blue solid line).
helix backbone atoms of the H and I β-strands with Leu 236 and Val 253 on the surface of the protein.Structural similarities between BjFixL and SmFixL suggest that these conformational changes may also occur after the photo cleavage of CO from the heme of SmFixL.The overall ΔV and ΔH values for each phase are related to (1) a change in overall charge distribution on the protein (i.e., change in net protein dipole leading to solvent reorganization), (2) formation of one or more salt-bridge interactions (the release of electrostricted water molecules upon salt-bridge formation results in volume increases), and/or (3) an increase in the solvent accessible van der Waals volume of the protein immediately upon photolysis.Truncated SmFixLH 128−264 Heme Domain.Photolysis of CO from the truncated SmFixLH 128−264 heme domain results in a monophasic relaxation after the photodissociation of CO with a ΔH of ∼9 kcal mol −1 and ΔV of ∼22 mL mol −1 .CO photolysis of the heme domain of SmFixLH 128−264 produces thermodynamics distinct from those from the heme domain of FixL from B. japonicum (BjFixLH 140−270 ), which display a biphasic relaxation after the photodissociation of CO. 41 For BjFixLH 140−270 , the fast phase (ΔH of −1.4 kcal mol −1 and ΔV of 14 mL mol −1 ) was associated with a reorganization of the

a
Lifetime values are reported for the 20°C deconvolutions b Miksovska et al.41

Figure 8 .
Figure 8. Proposed mechanism for the escape of CO from the heme of SmFixL*(CO) after photodisssociation.CO is photolyzed from the heme of SmFixL*(CO) in <20 ns in step (1) to form transient intermediate 1, which is the prompt phase.In step (2), CO migration occurs within the heme domain of SmFixL* with the conserved hydrophobic residues of the distal heme pocket (Ile210, Val232, Ile209, and Leu230) potentially forming an initial transient CO binding site, Intermediate 2, after ∼190 ns.In step (3), CO would then migrate from the initial heme domain transient binding site to the bulk solvent to form intermediate 3 at ∼512 ns.The initial conformational changes that are propagated from the heme domain to the coiled-coil linker domain in steps 1 to 3 result in the formation of intermediate 4 in step (4) at ∼1.5 μs, which could result in a (b) righthanded supercoiled rotation of the histidine kinase domain of SmFixL*, similar to the one observed at ∼2 μs for the chimeric YF1 protein that contains the histidine kinase domains of BjFixL.64In step 5, the right-handed supercoiled rotation could cause a rearrangement of the histidine kinase domains of SmFixL*, as observed in the chimeric YF1 protein after 250 ms, to form kinase active, deoxy Fe(II)SmFixL*.64 Kinase active, deoxy Fe(II)SmFixL* can phosphorylate SmFixJ, the response regulator, which triggers the expression of FixK2, the transcriptional activator of the nitrogen fixation genes.13−17In step 6, additional CO can be added to reform kinase inactive SmFixL*(CO) through additional conformational changes.