Rational design of electron/proton transfer mechanisms in the exoelectrogenic bacteria Geobacter sulfurreducens.

The redox potential values of cytochromes can be modulated by the protonation/deprotonation of neighbor groups (redox-Bohr effect), a mechanism that permits the proteins to couple electron/proton transfer. In the respiratory chains, this effect is particularly relevant if observed in the physiological pH range, as it may contribute to the electrochemical gradient for ATP synthesis. A constitutively produced family of five triheme cytochromes (PpcA-E) from the bacterium Geobacter sulfurreducens plays a crucial role in extracellular electron transfer, a hallmark that permits this bacterium to be explored for several biotechnological applications. Two members of this family (PpcA and PpcD) couple electron/proton transfer in the physiological pH range, a feature not shared with PpcB and PpcE. That ability is crucial for G. sulfurreducens' growth in Fe(III)-reducing habitats since extra contributors to the electrochemical gradient are needed. It was postulated that the redox-Bohr effect is determined by the nature of residue 6, a leucine in PpcA/PpcD and a phenylalanine in PpcB/PpcE. To confirm this hypothesis, Phe6 was replaced by leucine in PpcB and PpcE. The functional properties of these mutants were investigated by NMR and UV-visible spectroscopy to assess their capability to couple electron/proton transfer in the physiological pH range. The results obtained showed that the mutants have an increased redox-Bohr effect and are now capable of coupling electron/proton transfer. This confirms the determinant role of the nature of residue 6 in the modulation of the redox-Bohr effect in this family of cytochromes, opening routes to engineer Geobacter cells with improved biomass production.


Introduction
modulated by the solution's pH, a feature designated redox-Bohr effect which enables the protein to couple electron and proton transfer (17). Thus, in addition to the cytochrome's role in bridging the electron transfer between the inner and the outer membrane, the modulation of the hemes' redox potential values by the pH endows the PpcA-family cytochromes with the necessary properties to contribute to the electrochemical membrane gradient responsible for ATP generation when Geobacter utilizes extracellular electron acceptors (16,17). Indeed, modelling of G. sulfurreducens' growth in Fe(III)-reducing conditions by Mahadevan and co-workers (18) has shown that, due to the lack of H + consumption in the cytoplasmic side of the inner membrane, the bacteria's observed growth rate is only explained by considering an electrochemical gradient for ATP production generated by cytoplasmic and periplasmic sources. The natural abundance of the PpcA family members, their functional working redox potential ranges which are well-centered with the electrochemical responses of G. sulfurreducens biofilms' (19), and their ability to couple electron-proton transfer make the PpcA-family proteins the best candidates to contribute to the necessary electrochemical gradient when the bacteria switches from soluble to insoluble terminal electron acceptors (17,18).
The ability of a cytochrome to couple electron and proton transfer is only relevant if observed within the physiological range for the bacteria's pH growth (9,17). The data obtained from the detailed thermodynamic characterization of the PpcA-family members showed that PpcA and PpcD have a higher redox-Bohr effect compared to PpcB and PpcE (9). In addition, PpcA and PpcD have a well-defined and preferential electron/proton transfer mechanism in the physiological pH range (7)(8), whereas such preference was not verified in PpcB and PpcE (17). Considering the high amino acid identity of these protein sequences, the differences observed in the cytochrome's behavior may be attributed to the specific role played by some amino acids in the modulation of their functional properties. The redox-Bohr center has been attributed to propionate 13 of heme IV (P 13 IV ) in the PpcA cytochrome family (9, modulation of the redox-Bohr effect in this family of cytochromes, two mutants were constructed in the present work: PpcBF6L and PpcEF6L. The complete thermodynamic characterization of these two mutants using NMR and UV-visible spectroscopy was conducted to clarify the functional impact of this mutation in the electron/proton transfer mechanisms.

Cell growth and protein purification
PpcBF6L and PpcEF6L from G. sulfurreducens were produced and purified as previously described (20). Briefly, E. coli BL21 (DE3) cells containing the plasmid pEC86 encoding for the cytochrome c maturation gene cluster ccmABCDEFGH and a chloramphenicol resistance gene, were co-transformed with the plasmid containing either the gene for PpcBF6L or PpcEF6L, which carries an ampicillin resistance gene. Cells were grown at 30 °C in 2xYT media, supplemented with 34 μg/mL chloramphenicol and 100 μg/mL ampicillin, to an OD 600 of approximately 1.5. Protein expression was then induced with 10 μM of isopropyl β-D-thiogalactoside (IPTG) and the cell cultures were grown overnight at 30 °C. Following the overnight incubation, the cells were harvested by centrifugation at 6400 xg for 20 minutes. The periplasmic fraction was obtained after incubating the cells with lysis buffer (100 mM Tris-HCl (pH 8.0), 0.5 mM EDTA, 20% sucrose and 0.5 mg/ml lysozyme) for 15 minutes. This fraction was recovered by centrifugation at 14700 xg, at 4 °C for 20 minutes followed by ultracentrifugation at 225000 xg, at 4 °C for 1h. The obtained supernatant from this last centrifugation step was dialyzed against 2 x 4.5 L of 10 mM Tris-HCl (pH 8.0) and loaded onto the ion-exchange purification columns 2x5 mL Bio-Scale™ Mini UNOsphere™ S cartridges (BioRad), equilibrated with the same buffer. The protein was eluted with a sodium chloride gradient (0-300 mM) and the obtained fraction was concentrated and injected in a Superdex 75 molecular exclusion column (GE Healthcare), equilibrated with 100 mM sodium phosphate buffer (pH 8.0). Protein purity was evaluated by Coomassie stained SDS-PAGE.

NMR Studies on PpcBF6L and PpcEF6L
The structural and functional studies conducted in PpcBF6L and PpcEF6L mutants matched the conditions used for the studies of the respective wild-type proteins (9). Regarding sample preparation, for the assignment of the hemes in the reduced state, PpcBF6L and PpcEF6L samples of 1 mM were prepared in 80 mM phosphate buffer with NaCl (250mM final ionic strength), in 2  for PpcBF6L and 6 to 9 for PpcEF6L. The slightly different pH ranges allowed us to proper map the chemical shift dependence of the selected signals in the NMR spectra and the concomitant determination of the pK a values of the redox-Bohr center in the two proteins (see below in the subsection Thermodynamic model). Samples were firstly reduced using the procedure previously described and then, for obtaining their partial oxidation, hydrogen from the reduced sample was flushed out with argon followed by the addition of controlled amounts of air to the NMR tube with a Hamilton syringe.
All NMR spectra were recorded in Bruker Avance III 600 spectrometer equipped with a triple resonance cryoprobe (TCI) at 288 K. For the assignment of the heme substituents in the fully reduced state, 2D 1 H, 1 H-NOESY spectra were acquired with a mixing time of 80 ms, collecting 2048 (t 2 ) x 256 (t 1 ) data points to cover a sweep width of 8.4 kHz, with 160 scans per increment. For the thermodynamic experiments, at each pH value a series of 2D 1 H, 1 H-EXchange SpectroscopY (EXSY) experiments, with the sample poised at different degrees of oxidation, were conducted to unambiguously map the oxidation of the individual hemes throughout the redox titrations. All the 2D 1 H, 1 H-EXSY NMR experiments were acquired with a mixing time of 25 ms and with 2048 (t 2 ) x 256 (t 1 ) data points to cover a sweep width of 27.5 kHz, with at least 256 scans per increment. 1D 1 H NMR spectra were obtained before and after each 2D NMR spectrum to check for any changes in the oxidation state of the sample during the 2D NMR experiment acquisition.
The assignment of the NMR signals was obtained as described for the wild-type protein (9,21,22).
The proton chemical shifts were reported relative to DSS at 0 ppm. The NMR spectra were processed using TOPSPIN software (Bruker Biospin, Karlsruhe, Germany) and analyzed with Sparky Software (TD Goddard and DG Kneller, Sparky 3, University of California, San Francisco, USA).

Redox titrations of the mutated cytochromes followed by UV-visible spectroscopy
The redox titrations followed by UV-Visible spectroscopy of the mutant proteins were carried out inside an anaerobic LABstar glove box (MBraun) with argon circulation and oxygen levels kept under 0.5 ppm, at pH 7.0 and 8.0 (288 K), as previously described for the wild-type proteins (9). Solutions containing 18 μM of protein were prepared in 80 mM phosphate buffer with NaCl (250 mM final ionic

Thermodynamic model
The thermodynamic characterization of PpcB and PpcE mutant proteins has followed the same methodology employed for the functional characterization of the respective wild-type proteins (9) which will be briefly described it in this section for the sake of completeness (for a more detailed explanation see reference (23)).
In solution, the triheme cytochromes can adopt eight different microscopic oxidation stages which are grouped according to the number of oxidized hemes into four macroscopic oxidation stages, linked by three successive one-electron oxidation steps (see Figure S1) (12). Furthermore, the existence of a redox-Bohr center group which may be protonated of deprotonated in each microstate leads to a total of 16 microstates. The complete thermodynamic characterization of a triheme cytochrome requires the determination of both the individual heme oxidation fractions and the total reduced protein fraction (23). The former is obtained by the monitorization of the individual heme oxidation profiles using 2D 1 H, 1 H-EXSY NMR experiments, since the NMR paramagnetic shifts of a particular heme are proportional to its degree of oxidation. The latter is obtained through redox titrations followed by UV-Visible spectroscopy. It is important to stress that the individual heme NMR signals can be discriminated through this spectroscopic technique since the interconversion between microstates within the same oxidation stage (intramolecular electron exchange) is fast on the NMR time scale and the interconversion between microstates of different oxidation stages (intermolecular electron exchange) is slow. Thus, by following the methodology employed for the wild-type protein, the individual heme The chemical shifts of the heme methyl groups in the oxidized state are essentially determined by the geometry of the heme axial histidine ring planes, which differ considerably for PpcB and PpcE (22).
Consequently, the dispersion of the signals is different in the two proteins (i.e., the same heme methyl does not have the same chemical shift in the two proteins). However, since only one methyl per heme is necessary to properly probe the oxidation profile of its heme, and their adequateness was previously validated (9), signals in less spectral crowded regions were chosen to increase the number of experimental points and minimize the errors associated to the thermodynamic parameters. The NMR was then fitted, together with the data from UV-Visible redox titrations performed at pH 7 and 8, to the thermodynamic model which describes three redox centers and one redox-Bohr center (9). The NMR experimental uncertainty was evaluated from the line width of each NMR signal at half height and an uncertainty of 3% was given to the UV-Visible data points of the total optical signal.

Results and Discussion
The high amino acid sequence identity of the PpcA-family members, combined with their different functional properties, makes these proteins excellent models for studying the impact of specific amino acids in the modulation of their properties. The present study focuses on residue 6, located in the vicinity of the heme IV to which the redox-Bohr center was assigned within the PpcA-family members (9,16,21). The nature of residue 6 has been proposed to regulate the ability of these proteins to perform electron/proton transfer, a mechanism that might contribute to the electrochemical gradient across the inner membrane. It was observed that PpcA/PpcD can couple electron/proton transfer and have a leucine in position 6, whereas PpcB/PpcE are unable to do it and possess a phenylalanine in this position. Consequently, the phenylalanine in position 6 was replaced by a leucine in PpcB and PpcE and the impact of this mutation in the functional properties of the mutants was investigated.

Impact of the mutations in the global fold and heme core of the proteins
The PpcA family of triheme cytochromes possesses a highly compact arrangement of the three heme groups, with an average of 24 residues per heme. Due to the small size of the proteins and the heme iron's low-spin state, the NMR spectra are extremely well-resolved, and the heme groups' signals cover very typical regions of the spectra. For this reason, any disturbance in the polypeptide chain will To evaluate the effect of the mutation in residue 6 in the functional mechanism of PpcBF6L and PpcEF6L mutants, the thermodynamic characterization of both proteins was pursued in experimental conditions that matched those previously used for the wild-type proteins. The oxidation of each heme through the different oxidation stages was monitored in the pH range 5.4 to 8.9 for PpcBF6L and 6 to 9 for PpcEF6L using 2D 1  Because the hemes in these proteins are spatially close, their redox potential values are modulated by the redox interactions with neighboring hemes. These heme-heme redox interactions are strongly dependent on the heme iron distances and on the local dielectric constants (27). Similarly, to the wildtype proteins, the largest redox interactions in the mutants correspond to the closest pairs of hemes (hemes I-III and hemes III-IV - Table 1). The positive or negative values of the redox interactions indicate that the oxidation of a particular heme restricts or facilitates the oxidation of its neighbor, respectively.
Given the conserved architecture of the heme core, the smaller redox interaction values in the mutants are caused by the perturbation on the hydrophobic aromatic path at the vicinity of the heme groups, which may impact the electron transfer between hemes. In fact, aromatic groups have been shown to assist electron transfer to heme groups both within the same protein (28) and between electron partners (29). In PpcB and PpcE this path is formed by three (Phe 6 , Phe 15 , Phe 41 ) and four (Phe 6 , Phe 15 , Phe 41 , Tyr 45 ) residues, respectively ( Figure 1). Compared to PpcBF6L, the perturbation on the Thus, in addition to the perturbation of the hydrophobic aromatic path, the inclusion of a leucine residue at position 6 may also affect the H-bond network at the heme core, which further contributes to the variations in local dielectric constants.
Another feature that emerges from the thermodynamic parameters is the more negative values of the redox-Bohr interactions between the hemes and redox-Bohr center ( Table 1)

Physiological impact of the mutations in the oxidation order of the hemes
In both mutants, in the fully reduced and protonated state, the first heme to oxidize is the one with lowest redox potential -heme III -followed by heme I and then heme IV, indicating that the order of oxidation of the hemes in the mutants is conserved compared to the wild-type proteins ( Table 1). As the individual heme redox potentials are modulated by redox and redox-Bohr interactions, the redox potentials at a given pH usually differ from those observed for the fully reduced and protonated proteins. The redox behavior of PpcB and PpcE was discussed previously at the physiological pH of G.
sufurreducens' growth (9) and, for consistency, in the present work, the analysis of the heme oxidation profile of the mutants was conducted in the same conditions ( Figure 5A). As expected, due to changes in the network of redox and redox-Bohr interactions, the redox potential values of the hemes in both mutants were affected, especially the ones of PpcEF6L. However, their working redox potential range is not significantly affected ( Figure 5). In fact, in PpcBF6L, the difference between the redox potential values of the hemes spans 28 mV, which is very similar to the one of its wild-type protein (25 mV). In the case of PpcEF6L, this difference is more expressive -48 mV in the wild-type protein versus 58 mV in the mutant -which is in line with the impact of the mutation we observed in the individual redox potentials in the fully reduced and protonated state ( Table 1). The redox curves are steeper in the mutants, an effect that is particularly notorious also in PpcEF6L (cf. dashed and solid lines in Figure 5A) and is explained by the smaller value of the redox interactions in the mutants. Therefore, the oxidation profile of each heme is less affected by the oxidation of its neighbors, yielding a curve with a slope closer to a pure Nernst redox curve. Nevertheless, and despite these changes, the order of oxidation of the hemes is still the same as the wild-type proteins (III-I-IV), highlighting the retention of the redox properties of these proteins in the respective mutant.

The redox-Bohr effect impacts the functional mechanism of the cytochromes
As discussed in the previous sections, the replacement of Phe 6 by a leucine in PpcB and PpcE caused a significant variation in their redox-Bohr properties. In fact, in both cases, the deprotonation energy of the redox-Bohr center increased in such a way that the pK red values for both mutants increased compared to the wild-type proteins ( Table 2). The analysis of the pK red and pK ox values clearly shows an increase in the ΔpK a (pK red -pK ox ) of both proteins. In fact, PpcBF6L and PpcEF6L present a ΔpK a value 0.4 and 0.9 units higher than those of the respective wild-type proteins. These results are consistent with those obtained for PpcAL6F, in which the introduction of an aromatic amino acid decreased the ΔpK a of the mutant protein 0.8 units relatively to the wild-type protein (Table 3). Dantas and co-workers (20) have suggested that this variation in the ΔpK a value could be associated with the heme core's aromatic amino acid content. PpcA and PpcD are the two proteins which present highest ΔpK a values (2.0 and 1.8, respectively) and possess two aromatic amino acids, whereas PpcB and PpcE have, respectively, three and four aromatic amino acids in the heme core and a ΔpK a value of 1.1 and 0.3 ( Figure 1 and Table 3). ΔpK a value of PpcEF6L, now with three aromatic residues in the heme core, is nearly coincident with the ΔpK a value of wild-type PpcB, which holds three aromatic amino acids in the heme core (see Tables 2   and 3). Thus, the content of aromatic amino acids seems to directly impact the ΔpK a values. These values are calculated based on the redox interaction energies between the heme groups and the redox-Bohr center and are higher (more negative values) in the mutants than in the wild-type proteins (see  Tables 1 and 2). Considering this, most likely the network of π-π interactions established by more than two aromatic amino acids in the heme core weakens the interactions between the hemes and the redox-Bohr center, yielding smaller energies for this parameter.
The wider range of pK a values observed for PpcBF6L and PpcEF6L spans more appropriately over the physiological pH range of G. sulfurreducens, hinting that these two proteins could, now, effectively couple electron and proton transfer. Considering this, it was important to analyze the relative contribution of the 16 possible microstates along the redox cycle of the wild-type and mutant proteins at pH 7.5 ( Figure 6). A well-defined pathway is favored when one microstate clearly has higher molar fraction than that of another microstate within the same oxidation stage, thus favoring the directionality of electron transfer events. In the case of wild-type PpcB and PpcE, at stage 0 there are two populations with nearly equivalent molar fractions -P 0 and P 0H (cf. Figure 6A and 6C). This will not favor the directionality of subsequent electron transfer events since several microstates will coexist in solution in the subsequent redox stages. In contrast, in PpcBF6L and PpcEF6L ( Figure 6B and 6D), the molar faction of the P 0H population is clearly dominant over that of P 0 , which will confer directionality to the subsequent electron transfer events since most of the cytochrome molecules are in the same redox microstate. In PpcBF6L, the oxidation stages 0 and 1 are dominated by microstates P 0H and P 3H ; however, the dominant curve of oxidation stage 1 is intercepted by the curve corresponding to microstate P 134 and, thus, this microstate dominates the last oxidation stage ( Figure 6B). Therefore, the preferential route of PpcBF6L for electron transfer, in which proton transfer is coupled to a two-electron transfer event between oxidation stages 1 and 3, is defined as follows: P 0H  P 3H  P 134 . It is worth noting that in the wild-type cytochrome all the oxidation stages are dominated by deprotonated microstates (see dashed lines in Figure 6A), and, in contrast, in PpcBF6L the electron transfer is now coupled to deprotonation of the redox-Bohr center, i.e. e -/H + transfer is now observed (see transition of solid to dashed lines in Figure 6B). Compared to the wild-type protein, in PpcEF6L the molar fraction of the protonated microstate P 0H is also substantially higher than that of its deprotonated form (P 0 ), as it is the molar fraction of the deprotonated microstate P 134 versus that of its protonated form (P 134H ) ( Figure 6C and 6D). Consequently, similarly to PpcBF6L, in PpcEF6L a preferential route for electron transfer containing an e -/H + transfer step is favored: P 0H  P 3H  P 13  P 134 .
The analysis of the most relevant microstates at pH 7.5 indicates that the substitution of Phe 6 by leucine favors a preferential e -/H + transfer pathway, endowing PpcBF6L and PpcEF6L with the necessary properties to couple the e -/H + transfer, a feature not shared with the respective wild-type cytochromes.
However, fluctuations in G. sufurreducens' pH and redox environment are inevitable. Thus, the determination of the most relevant microstates for the mutants and wild-type cytochromes in a wider pH range, including the whole physiological pH range of Geobacter, will permit a better evaluation of the physiological impact of the mutation in PpcB and PpcE (Figure 7). The analysis of Figure 7 shows that PpcB can perform e -/H + transfer in the pH range 6.4 to 7.5 (see colored regions in Figure 7), although at discrete redox potential ranges, making it an unsuitable contributor to the periplasmic electrochemical gradient in the case of fluctuations in the pH and redox potential in G. sufurreducens' environment. In contrast, PpcBF6L is able to perform e -/H + transfer continuously in a wider pH (6.8 to 8.2) and redox (approximately from -170 mV to -90 mV) ranges, which perfectly matches the physiological pH range of G. sufurreducens' growth. Thus, in this interval, the protein is never either fully reduced or fully oxidized, which permits it to effectively couple e -/H + transfer and contribute to G. sufurreducens' electrochemical gradient. The impact of the mutation is even more notorious for PpcE. In fact, the wild-type protein can

Conclusions and Implications
In        Table 1 for wild-type (panels A and C) and mutant (panels B and D) cytochromes. The solid and dashed lines indicate, respectively, the protonated and deprotonated microstates (see Figure S1).
The curves of the most relevant microstates are labeled.  Table 1 -Thermodynamic parameters determined for PpcBF6L and PpcEF6L in the fully reduced and protonated form. The energies are reported in meV and standard errors are given in parentheses. The oxidation energies of the hemes and the deprotonating energy of the redox-Bohr center are represented in the diagonal in bold. The redox (heme-heme) and redox-Bohr (heme-proton) interaction energies are represented in the off-diagonal energy values. The thermodynamic parameters of wild-type PpcB and PpcE from G. sulfurreducens (Gs) are also indicated for comparison. , following the nomenclature for the pairwise interacting centers model (23). For comparison, the macroscopic pK a values of the cytochromes PpcB and PpcE from G. sulfurreducens (Gs) are also indicated.

Protein
Oxidation stage ΔpK a pK red pK ox PpcBF6L Gs (this work) 8 Table 3 -Impact of the aromatic amino acid content in the redox-Bohr properties of proteins. The ΔpK a values were taken from Table 2 and the pH range for electron/proton (e -/H + ) transfer was determined with the parameters indicated in Table 1. PpcA Gs (9) 2 2 6.5 -8.5