Recombinant His-tagged HiPIP-41, CytC-18, and CytC-78 derived from Ferrovum sp. PN-J47-F6 were produced, purified and concentrated as described previously (Ullrich et al., 2023).

If not stated otherwise all experiments were conducted at 25°C in sodium citrate phosphate buffer (41.4 mM Na₂HPO₄, 79.3 mM citric acid, pH 3.1). The experimental set-up is shown in Supplementary Figure 1.

Structural models of the mature wildtype protein sequences of HiPIP-41 (WP_067495359), CytC-18 (WP_067493319), and CytC-78 (WP_229347545) were predicted using the D-i-Tasser suite (Deep-learning based Iterative Threading ASSEmbly Refinement; Yang and Zhang, 2015; Zheng et al., 2022, 2023). N-terminal signal peptides were predicted using SignalP5.0 (Almagro Armenteros et al., 2019) and omitted from the protein sequence. Structural models of the HiPIP of Rhodocyclus tenuis (1Isu), Cyc₄₁ of Acidithiobacillus ferrooxidans (1h1o) and c₄ of Pseudomonas strutzeri (1 m70) were retrieved from the Protein Database (PDB). The structural model of Hip of Acidithiobacillus ferridurans (UniProt ID: Q93MF8) was retrieved from the AlphaFold protein Structure Database (Jumper et al., 2021). Structural models were visualized using Chimera 1.15 (Pettersen et al., 2004). Superimposition of the HiPIP structural models was achieved using MatchMaker implemented in Chimera 1.15 (Meng et al., 2006). Hydrophobic contacts and hydrogen bonds at the domain:domain interfaces of c₄ cytochromes and within the cofactor binding pockets were predicted by LigPlot⁺ (Wallace et al., 1995; Laskowski and Swindells, 2011) and amended by manual inspection of the structural models.

The electrochemical characterization of the recombinant Ferrovum-derived redox proteins HiPIP-41, CytC-18 and CytC-78 aimed to determine the E_m-values of their cofactors. The different size of the native redox proteins and the different nature of their cofactors required different approaches for their electrochemical characterization.

The [4Fe-4S]-cofactor in HiPIPs is covalently bound by four highly conserved cysteine residues and is stabilized in a hydrophobic binding pocket with aromatic residues playing a fundamental role in electron transfer and E_m modulation (Agarwal et al., 1995; Iwagami et al., 1995; Bian et al., 1996; Parisini et al., 1999; Liu et al., 2014). Our previous analysis of the homology-based structure of HiPIP-41 has already suggested that the [4Fe-4S]-cofactor is surrounded by a high number of hydrophobic and aromatic residues (Ullrich et al., 2023). Thus, the new structural model of HiPIP-41 based on a combined approach of homology modeling and a deep learning algorithm was compared to the AlphaFold-predicted structure of Hip of the acidophile A. ferrooxidans and the experimentally solved structure of the HiPIP of R. tenuis (PDB: 1Isu). Hip and HiPIP 1Isu were chosen for this comparative approach because they have already been electrochemically characterized (Table 1). Our analysis focused on conserved and unique structural features with special regard to residue candidates involved in tuning the E_m-values.

Hydrophobic contacts between the protein and the cofactor in the three HiPIPs were predicted using LigPlot⁺ with subsequent evaluation by manual inspection of their structural models. The [4Fe-4S]-cofactor in HiPIP-41 is surrounded by 10 hydrophobic residues of which five are aromatic (Figures 4A,B). The high number of hydrophobic contacts completely shields the cofactor from the surrounding solvent (Figures 4C,D). However, some of these residues not only contribute to the hydrophobic character of the binding pocket but also to the surface properties of HiPIP-41. The hydroxyl group of Tyr-54, for example, is oriented toward the surface while the aromatic ring shields the cofactor (Figure 4C). The polar sidechain of the Asn-27 is similarly oriented toward the surface while the CH₂-group is positioned toward the protein core (Figure 4D). In the case of Tyr-43 and Phe-58 the oxygen moieties of their peptide bonds are pointing toward the surface (Figure 4C).

The cofactor in HiPIP 1Isu is surrounded by a similar number of hydrophobic (10) and aromatic residues (5) (Supplementary Figures 3A,B) while the ten hydrophobic contacts to the Hip cofactor include only four aromatic residues (Supplementary Figures 3C,D). Similarly to HiPIP-41, the cofactors of Hip and HiPIP 1Isu are completely buried within the protein cores and are not accessible for the surrounding solvent (Supplementary Figures 3E,F). The superimposition of all three structural models shows the high structural similarity of all HiPIPs where HiPIP-41 and HiPIP 1Isu even share the short N-terminal α-helix (Supplementary Figure 4). The high proline content of Hip was reported to interfere with secondary structural elements (Nouailler et al., 2006), which might explain the slightly lower structural similarity to the other two HiPIPs. Despite the generally low sequence similarities of less than 40% between the three HiPIPs, many of the residues with hydrophobic contacts to the cofactor were identified at identical positions (Supplementary Figures 4A–C). A remarkable exception presents position 53 (1Isu numbering), where HiPIP-41 harbors an additional cysteine residue while HiPIP 1Isu and Hip have glycine residues. Two other positions are also noteworthy, because only one of the three HiPIPs possesses an aromatic residue while the other two harbor aliphatic residues at the same position: the first is Phe-10 in HiPIP 1Isu and the second is Tyr-43 in HiPIP-41. Otherwise four of five aromatic residues with hydrophobic contacts to the cofactor are highly conserved between all three structures (Supplementary Figure 4C).

However, the only clearly distinguishing feature of the two acidophile-derived HiPIPs Hip and HiPIP-41 present the two additional cysteine residues not involved in cofactor coordination (Supplementary Figure 4D). While Cys-52 and Cys-84 of Hip were found to form a structure stabilizing disulfide bond (Nouailler et al., 2006), the role of Cys-50 and Cys-53 of HiPIP-41 has not yet been elucidated (Ullrich et al., 2023). The orientation of Cys-53 toward the cofactor in HiPIP-41 together with its restricted presence in Ferrovum spp. HiPIPs (Ullrich et al., 2023) seems to be a rather striking. Still, it remains so far unclear whether these additional cysteines might contribute to the at least 240 mV more positive E_m in the HiPIPs of the acidophiles.

The electrochemical characterization of CytC-18 and CytC-78 showed (i) that the potential window for the full redox transition is smaller for CytC-78 which corresponds with the smaller ΔE_m of its two heme centers and (ii) that the hysteresis between the oxidative and reductive potential sweeps is smaller in CytC-18. Hence, although the E_m-values of the two Ferrovum-derived cytochromes are quite similar, both are characterized by an individual redox behavior. Earlier studies suggested that both, the heme-to-heme positioning and the heme:protein contacts affect the intramolecular electron transfer (Kadziola and Larsen, 1997; Abergel et al., 2003; de March et al., 2015). Aiming to collect evidence for correlations between structure and properties of CytC-18 and CytC-78, we focused our analyses on their hydrogen bond networks at the domain:domain interface, hydrophobic contacts in the heme-binding pockets and the solvent accessibility of their heme centers. Moreover, these analyses provided the basis for the subsequent comparison to available structures of the homologous c₄ cytochromes of the neutrophile P. strutzeri (PDB: 1 m70; P.s. c₄) and of Cyc₄₁ of the acidophile A. ferrooxidans (PDB: 1h1o; A.f. Cyc₄₁) which have also been electrochemically characterized.

The spectroelectrochemical characterization of CytC-18 and CytC-78 indicates that the full redox transition occurs stepwise via two redox transitions at different potentials. These lower and higher potential redox transitions are attributable to the two heme centers covalently bound in the N-terminal (heme-1) and the C-terminal domain (heme-2) of the cytochromes. In CytC-18, the ΔE_m between the lower and higher potential heme is 140 mV while it is 91 mV in CytC-78. With exception of P.a. c₄ (Carpenter et al., 2020) and Cyt c₄ of P. haloplanktis (Di Rocco et al., 2008) the two heme centers in homologous cytochromes were also characterized by individual E_m-values (Leitch et al., 1985; Cavazza et al., 1996; Giudici-Orticoni et al., 2000; Malarte et al., 2005; Chang et al., 2010; Carpenter et al., 2020). The two sequential redox transitions resulted in a double-sigmoidal shape of the absorption potential curves of CytC-18 and CytC-78 which was more clearly defined in case of CytC-18. Theoretically, this curve shape reflects the modeled redox behavior of a two-center redox protein with non-interacting E_m-values (Catarino and Turner, 2001; Chi et al., 2010). However, for both cytochromes the lower potential redox transition only corresponded to approx. 25% of the total absorption change between the fully reduced and fully oxidized state while the higher potential redox transition corresponded to 75%. This observation suggests that the redox state of one heme center affects the E_m of the other heme reflecting an anti-cooperative redox interaction (Zickermann et al., 1995) between the two heme centers in CytC-18 and CytC-78. Similar observations were also reported between hemes a and a₃ in the cytochrome c oxidase of Paracoccus denitrificans (Gorbikova et al., 2006), hemes b and c₁ in the P. denitrificans bc₁ complex (Covian et al., 2007) or hemes b₅₅₈ and b₅₉₅ of the E. coli cytochrome bd oxidase (Bloch et al., 2009).

The analysis of the heme-heme geometry and the Fe-Fe distances highlighted a another common structural feature of CytC-18, CytC-78, A. f. Cyc₄₁, and P.s. c₄. Both heme centers are located in the same plane and in face-to-face orientation according the classification of de March et al. (2015). The spatial separation of the hemes in individual domains together with their face-to-face orientation leads to long distances Fe-Fe distances of 18.2 Å (CytC-18) and 19.2 Å (P.s. c₄). This heme geometry affects the mode of electron transfer between the two redox centers. In the cytochrome c oxidase (Tan et al., 2004) or the decaheme outer membrane cytochromes MrtC and OmcA of Shewanella oneidensis (Tikhonova and Popov, 2014) the close proximity of the redox centers allows the rapid electron transfer via electron tunneling. Since this mode requires distances between the redox centers of less than 14 Å (Page et al., 1999), the electron transfer in CytC-18, CytC-78, A. f. Cyc₄₁ and P.s. c₄ is more likely to be realized via proton-coupled electron transfer (Huynh and Meyer, 2007). The slightly tilted angle of their hemes together with the face-to-face geometry results in the close proximity of the respective A-ring proprionates at the domain:domain interfaces. The four proprionate groups in each cytochrome are involved in a complex hydrogen bond network involving numerous residues of both domains. The experimentally solved structures of P.s. c₄ and A.f. Cyc₄₁ also included water molecules at the domain:domain interface which are also likely to be part of this hydrogen bond network. These hydrogen bond networks in all four c₄ cytochromes present the pre-requisite of the proton-coupled electron transfer in a pre-association phase by minimizing the proton tunneling distance (Huynh and Meyer, 2007).

The constitution of the hydrogen bond network at the domain:domain interface might affect the redox kinetics of the cytochromes (Crowley and Ubbink, 2003). Although we did not determine kinetics parameters in the present study our electrochemical characterization of the Ferrovum-derived cytochromes strongly suggests that CytC-18 and CytC-78 harbor individual redox behaviors as was indicated by the different potential windows necessary for the full redox transition and the different degree of hysteresis between the oxidative and the reductive potential sweeps. The overall constitution of hydrogen bond network is very similar in CytC-18 and CytC-78. However, the most prominent differences present the acidic aspartate and glutamate residues in proximity of the heme-1 proprionate groups in CytC-18. They are expected to be fully protonated at pH 3.1 and are thus uncharged but also potential donors and acceptors of hydrogen bonds. In contrast to that, the hydrogen bond network of the heme-1 proprionate groups in CytC-78 involves an additional arginine residue which is not only a potential hydrogen bonding partner but also carries a positive charge at the guanidino group. Apart from the influence of the hydrogen bond network on the interdomain electron transfer also the dynamics of the domain:domain association is thought to play a central role for the redox activity of a protein (Camacho et al., 1999; Crowley and Ubbink, 2003). Indeed we observed differences in the nature of contacts at the domain:domain interfaces in CytC-18 and CytC-78. While the domain interface of CytC-18 involves more hydrogen bonds than the interface in CytC-78, the situation is vice versa with respect to the number of hydrophobic contacts. Since hydrophobic contacts act in longer range than hydrogen bonds (Israelachvili and Pashley, 1982; Onofrio et al., 2014) the different nature of interdomain contacts in the two Ferrovum-derived cytochrome is probably affecting the association dynamics and thereby the intramolecular electron transfer kinetics.

In their comparative study on structure and function correlations involving 33 heme-binding proteins with experimentally solved structures Smith et al. (2010) did not identify clear connections between the cytochrome’s E_m and specific structural features. They have instead postulated that a combination of various structural features influence the E_m. We focused our structural comparison therefore on hydrophobic contacts in the heme binding pockets and heme solvent accessibility, because they have already been identified as a central E_m influencing factor in other redox proteins (Di Rocco et al., 2008, 2011; de March et al., 2015; Hosseinzadeh et al., 2016; Carpenter et al., 2020). The structural comparison of CytC-18, CytC-78, A.f. Cyc₄₁, and P.s. c₄ revealed that the heme centers in A.f. Cyc₄₁ formed the highest number of hydrophobic contacts to their binding pockets. The total number of hydrophobic contacts in the other three cytochromes was similar, but CytC-18 and CytC-78 harbored the largest fraction of aromatic residues among these contacts. During the purification and concentration procedure of CytC-78 and CytC-18 we indeed observed a long stability of the fully reduced state of CytC-78 or partly reduced state of CytC-18 (Ullrich et al., 2023), respectively, while P.s. c₄ was observed to quickly oxidize during the crystallization process in absence of an chemical oxidant (Kadziola and Larsen, 1997). Thus, the higher number of hydrophobic contacts or aromatic residues in proximity of the heme center might indeed result in the more effective shielding of the heme centers in the acidophile-derived cytochromes as was suggested by their structural models. However, the domain:domain interface, and thereby the hydrogen bond network involving the heme proprionates and residues of both domains, appears to be equally accessible for solvent water molecules since the number of water molecules at the interface of A.t. Cyc₄₁ (7) and P.s. c₄ (9) was similar.

Besides the above discussed contacts in the heme-binding pocket, other structural aspects were discussed as influencing factors of the E_m, such as non-planar distortions of the heme porphyrin ring that have also been identified to effect the heme’s redox properties by influencing the E_m, transition dipoles and axial ligand affinity (Shelnutt et al., 1998; Smith et al., 2010). In case of A.f. Cyc₄₁ and P.s. c₄ no distortion of the porphyrin ring structure has been reported (Kadziola and Larsen, 1997; Abergel et al., 2003; Carpenter et al., 2020). An overall structural similarity of CytC-18 and CytC-78 to A.f. Cyc₄₁ and P.s. c₄ leads us assume that the heme centers in the Ferrovum-derived cytochromes do also not have any structural distortions. Furthermore, mutant studies demonstrated that the presence of polar or charged residues in close proximity to the heme iron decreased the E_m. The disruption of a hydrogen bond to the histidine axial ligand in myoglobin increased its E_m (Bhagi-Damodaran et al., 2014). This effect was explained by the increase of the positive character of the Nε-atom of the histidine axial ligand and the thereby reduced electron donating ability of its imidazole ring toward the heme iron (Valentine et al., 1979). The exchange of aliphatic residues providing hydrophobic contacts to the porphyrin ring system by acidic or polar residues similarly resulted in the E_m becoming up to 200 mV more negative than in the wildtype myoglobin (Varadarajan et al., 1989). Specific electrostatic contacts between the protein chain and the heme center indeed present interesting starting points for future studies. These will also require experimental structure determination because electrostatic interactions require a close proximity of the partners, and thus a high-resolution structure of the protein is needed to study them.

The E_m of HiPIP-41 is approx. 240 mV more positive than that of HiPIP 1Isu of R. tenuis and similar to the E_m of Hip of A. ferridurans. The covalent binding of the [4Fe-4S]-cofactor by four highly conserved cysteines resulting in its tight shielding from the solvent is acknowledged to cause the far more positive E_m in comparison to ferredoxins (Carter et al., 1972; Dey et al., 2007). Moreover, fewer hydrogen bonds between peptide bonds of the protein backbone and the cysteine sulfur atoms were detected in HiPIPs in contrast to ferredoxins (Backes et al., 1991), resulting in a lower polarity of the cofactor environment in HiPIPs (Heering et al., 1995). Beside the similar electrostatic contacts in the cofactor environment, the total number of hydrophobic contacts in the cofactor binding pocket was also similar in the structural models of HiPIP-41, HiPIP 1Isu and Hip, resulting in the anticipated complete burial of their cofactors and the effective shielding from the surrounding solvent. Moreover, the superimposition of the three structural models showed a high structural similarity and the conserved positions of the hydrophobic contacts. Since aromatic residues are anticipated to contribute to the cofactor stability by supporting its tight shielding from the protein’s surrounding solvent (Agarwal et al., 1995; Iwagami et al., 1995), their high level of conservation was not surprising. However, with respect to non-aromatic residues only their position within the protein sequence and structure were conserved but not the residues themselves. In this context, we did not identify any correlations between the size or the degree of hydrophobicity of the hydrophobic contacts and the E_m.

However, the most remarkable structural feature of HiPIP-41 and Hip present the two additional cysteine residues. In Hip, Cys-52 and Cys-84 form a disulfide bridge which was assumed to contribute to the acid stability of the protein (Nouailler et al., 2006). In HiPIP-41, Cys-50 and Cys-55 are located in the core of the protein and Cys-53 is oriented toward the cofactor. These two additional cysteines were found to be highly conserved in the HiPIPs of Ferrovum spp., but they are not found in homologous proteins in neutrophilic relatives (Ullrich et al., 2023). Since cysteine residues were classified as very hydrophobic due to their very frequent location within a protein’s hydrophobic core (Janin, 1979; Rose et al., 1985), we propose that they might indeed contribute to the hydrophobic character of the surrounding of the [4Fe-4S]-cofactor and thereby modulating the E_m in the acidophile-derived HiPIPs. The HiPIP Iro from A. ferrooxidans with a very positive E_m of 630 mV (Yamanaka and Fukumori, 1995), however, does also not contain any additional cysteine residues (Nouailler et al., 2006).

Apart from the solvent accessibility of the [4Fe-4S]-cofactor and the very hydrophobic character of its binding pocket, also the polarity of the cluster environment (Heering et al., 1995; Stephens et al., 1996), the general surface charge of the protein (Banci et al., 1995; Stephens et al., 1996; Capozzi et al., 1998) and the position of polar patches on the protein’s surface (Babini et al., 1998; Parisini et al., 1999) have been discussed as E_m modulating factors. The surface of HiPIP 1Isu is characterized by the presence of patches with both negative and positive electrostatic potential caused by acidic, basic and polar residues being exposed to the surface (Rayment et al., 1992). In contrast, the surface of HiPIP-41 is characterized by large patches of neutral surface electrostatic potential or patches with slightly positive electrostatic potentials (Ullrich et al., 2023). With respect to missing data on the surface charge of Hip it is difficult to evaluate its potential influencing effect of the overall surface charge on the E_m. An earlier study of Babini et al. (1998) has indeed demonstrated how the mutation of a surface exposed histidine influences the E_m and the redox kinetics of the HiPIP of Chromatium vinosum. Residues Tyr-54 and Asn-27 in HiPIP-41 might present candidates for similar studies since they are in contact with both, the cofactor-binding pocket and the surface of the protein.

Based on our structural analyses and comparisons, we propose that the N-terminal hemes (hemes-1) in CytC-18 and CytC-78 are the higher potential hemes and the C-terminal hemes (hemes-2) the lower potential hemes. In CytC-18, heme-1 is more tightly surrounded by aliphatic and aromatic residues resulting in its deep burial within the protein core and an efficient shielding from the surrounding solvent. Similar observations were reported for the N-terminal heme center of Cyc₄₁ (Abergel et al., 2003) and the C-terminal heme center of P.s. c₄ (Nissum et al., 1997; Carpenter et al., 2020), respectively. Moreover, we assume that the tighter shielding of the N-terminal heme center in CytC-18 in comparison to the C-terminal heme results in the observed larger ΔE_m by increasing the E_m of the higher potential heme in comparison to the lower potential heme. The heme binding-pockets in the N- and C-terminal domains in CytC-78 have a more similar appearance which we assume contributes to the smaller ΔE_m of its hemes. Still, the overall similar number and nature of hydrophobic contacts in the heme-binding pockets of both Ferrovum-derived cytochromes lead us to infer that the higher potential heme is associated with the N-terminal domain in both cytochromes.

The different location of the higher and lower potential hemes in the acidophile-derived c₄ cytochromes CytC-18, CytC-78, and A. f. Cyc₄₁ in contrast to P.s. c₄ might reflect the different organization of the electron transfer chains. While CytC-18, CytC-78, and Cyc₄₁ accept electrons from a soluble redox protein with a very positive E_m (HiPIP-41 or rusticyanin) to transfer them further downhill to the cytochrome c oxidase in the inner membrane (Giudici-Orticoni et al., 1999; Ullrich et al., 2023), P.s. c₄ shuttles electrons between inner membrane complexes III and IV (Smith et al., 1981; Kadziola and Larsen, 1997). The C-terminal domain of P.s. c₄ was proposed to interact with the cytochrome c oxidase (Iwata et al., 1995; Tsukihara et al., 1995) suggesting that the higher potential heme transfers electrons further downhill to the terminal oxidase. A mutant study of Cyc₄₁ variants (Malarte et al., 2005) and a computational docking study (Jiang et al., 2021) draw a similar conclusion for the interaction of Cyc₄₁ in the electron transfer chain in A. ferrooxidans, suggesting that the C-terminal lower potential heme accepts electrons from rusticyanin while the N-terminal higher potential heme transfers them downhill to the aa₃-type cytochrome c oxidase. Expanding this conclusion to CytC-18 and CytC-78, we propose that the C-terminal domains with the lower potential hemes interact with HiPIP-41 in the electron transfer chain of Ferrovum sp. PN-J47-F6.

With respect to the electron acceptors of CytC-18 and CytC-78 only educated guesses are possible at this stage of our current research. In A. ferrooxidans, Cyc₄₁ (Cyc₁) has a slightly more positive E_m than CycA1 (Cavazza et al., 1996; Giudici-Orticoni et al., 2000). While Cyc₄₁ transfers electrons downhill to the aa₃-type cytochrome c oxidase, CycA1 transfers a smaller fraction of electrons uphill to the bc₁ complex (Bird et al., 2011). The encoding gene of Cyc₄₁ is localized directly downstream of the gene encoding the outer membrane monoheme cytochrome Cyc2 (Valdés et al., 2008). In Ferrovum sp. PN-J47-F6, the gene coding CytC-18 or its homolog in Ferrovum spp. is localized directly downstream of the Cyc2-like encoding gene (Ullrich et al., 2018). Based on the genetic organization of its encoding gene and its slightly more positive E_m, we propose that CytC-18 transfers electrons downhill to the ccb₃-type cytochrome c oxidase while CytC-78 interacts with the bc₁ complex within the uphill branch of the electron transfer chain.