Functional Dissection of the Multi-Domain Di-Heme Cytochrome c 550 from Thermus thermophilus

In bacteria, oxidation of sulfite to sulfate, the most common strategy for sulfite detoxification, is mainly accomplished by the molybdenum-containing sulfite:acceptor oxidoreductases (SORs). Bacterial SORs are very diverse proteins; they can exist as monomers or homodimers of their core subunit, as well as heterodimers with an additional cytochrome c subunit. We have previously described the homodimeric SOR from Thermus thermophilus HB8 (SORTTHB8), identified its physiological electron acceptor, cytochrome c 550, and demonstrated the key role of the latter in coupling sulfite oxidation to aerobic respiration. Herein, the role of this di-heme cytochrome c was further investigated. The cytochrome was shown to be composed of two conformationally independent domains, each containing one heme moiety. Each domain was separately cloned, expressed in E. coli and purified to homogeneity. Stopped-flow experiments showed that: i) the N-terminal domain is the only one accepting electrons from SORTTHB8; ii) the N- and C-terminal domains are in rapid redox equilibrium and iii) both domains are able to transfer electrons further to cytochrome c 552, the physiological substrate of the ba 3 and caa 3 terminal oxidases. These findings show that cytochrome c 550 functions as a electron shuttle, without working as an electron wire with one heme acting as the electron entry and the other as the electron exit site. Although contribution of the cytochrome c 550 C-terminal domain to T. thermophilus sulfur respiration seems to be dispensable, we suggest that di-heme composition of the cytochrome physiologically enables storage of the two electrons generated from sulfite oxidation, thereof ensuring efficient contribution of sulfite detoxification to the respiratory chain-mediated energy generation.


Introduction
In addition to its natural occurrence in the environment, sulfite is an extremely important intermediary in sulfur metabolism, arising from a variety of reactions both in prokaryotes and eukaryotes [1][2][3]. The nucleophilicity and strong reducing capacity of sulfite account for its high toxicity. In the cell it can react with disulfide bonds causing protein inactivation and DNA damage. Although some microorganisms use sulfite as the sole electron/energy source [4,5], accumulation of sulfite in the cell generally leads to massive damage, so that both prokaryotic and eukaryotic cells require efficient sulfite detoxification systems.
The most common strategy for sulfite detoxification in Bacteria and Archaea involves oxidation to sulfate, accomplished either directly or indirectly via the adenosine 59-phosphosulfate reductase pathway [2,6]. Molybdenum-containing sulfite:acceptor oxidoreductases (SORs) catalyze the direct oxidation of sulfite to sulfate. They have been identified in mammals [7], birds [8], plants [9] and prokaryotes [10]. Two types of SORs have been identified to date: i) the sulfite oxidases (EC 1.8.3.1) that are able to utilize O 2 as a direct electron acceptor, but also ferricyanide and sometimes cytochrome c, and ii) the sulfite dehydrogenases (EC 1.8.2.1) unable to transfer electrons to O 2 . The first bacterial SOR was discovered almost half a century ago [11]. Although, since then, SORs have been shown to be widely distributed among bacteria, their exact physiological role is still elusive.
Compared to the vertebrate and plant enzymes, bacterial SORs are structurally much more diverse. The protein core consists of a molybdenum binding site and a dimerization domain. The enzymes can exist as monomers or homodimers of the core structure, as well as heterodimers with an additional cytochrome c subunit [10,[12][13][14][15][16]. The SORs containing both the molybdenum cofactor and the additional cytochrome c subunit have been classified as Group 1 SORs, while members of the Group 2 contain only the molybdenum cofactor and are called 'atypical' SORs [17]. The name 'atypical' arises from the fact that most of these enzymes, if not all, cannot efficiently use horse heart cytochrome c as substrate and display higher activities when assayed with the artificial electron acceptor ferricyanide. To date, Group 1 includes only the cytochrome c-containing SOR isolated from Starkeya novella [10]. The SOR from Campylobacter jejuni, though originally defined as a two-subunit protein acting similarly to the S. novella enzyme, likely belongs to Group 2 SORs, as the molybdenum-and the heme-containing subunits do not co-purify [18]. Similarly, all the other bacterial SORs so far described fall into Group 2 [12][13][14][15][16], thus calling for a revision of the term 'atypical'.
With the exception of the enzyme from Deinococcus radiodurans, all characterized SORs were interestingly found to be encoded upstream their putative physiological electron acceptors (c-type cytochromes or other redox proteins [19]). Consistently, the genes coding for the c-type cytochromes identified as electron acceptors for SORs from S. novella [10], C. jejuni [18], Sinorhizobium meliloti [20] and Thermus thermophilus [16] were all found downstream the relative SOR-encoding gene. These c-type cytochromes are also very diverse. They differ in size and heme content and this is an additional feature contributing to the complexity of SORs.
Several attempts have been made to elucidate how SORmediated sulfite oxidation is integrated in cell metabolism. It is postulated that Group 1 SORs are directly linked to the respiratory chain via their natural electron acceptor cytochromes [20]. In S. novella, cytochrome c 550 [10] was tentatively suggested to enable the association between sulfite oxidation and aerobic respiration, based on the notion that cytochromes c are natural substrates for cytochrome c oxidases [20]; the hypothesis however, remains to be tested as yet. Similarly, in C. jejuni electrons from sulfite oxidation were proposed to enter the respiratory chain downstream the bc 1 complex via the natural substrate of cb oxidase [18]. Although such a scenario seems plausible, also in this case the exact electron transfer pathway and the redox proteins involved have not been identified. Only cell extracts, and not purified proteins, were used in these experiments and, indeed, one cannot exclude the involvement of additional, unidentified electron shuttles.
Recently, the connection to the respiratory chain was demonstrated for Group 2 SORs from S. meliloti [20] and T. thermophilus [16], whose natural electron acceptors (cytochrome c Smc04048 and cytochrome c 550 , respectively) have been identified. In the former study [20], however, experiments were performed using isolated cell membranes and the association of the electron acceptor of SorT, cytochrome c Smc04048, with cytochrome oxidases remains to be proven. On the contrary, in the latter study on T. thermophilus the complete electron transfer pathway linking sulfite oxidation to oxygen reduction was unveiled [16]. Accordingly, the electrons generated upon sulfite oxidation by SOR TTHB8 are transferred to the natural electron acceptor of the enzyme, cytochrome c 550 , and from here to cytochrome c 552 , the physiological electron donor of the two terminal cytochrome c oxidases, ba 3 and caa 3 .
Here, the role of T. thermophilus cytochrome c 550 in coupling sulfite oxidation to cell respiration has been further investigated.

Materials and Methods
Purification of Cytochrome c 552 and the Cytochrome c Oxidases ba 3

and caa 3
Native ba 3 -and caa 3 -type cytochrome c oxidases were isolated from T. thermophilus HB8 cells according to previously published procedures [21,22]. Native cytochrome c 552 was purified according to Soulimane and co-authors [23]. Purified proteins were concentrated by ultrafiltration, fast frozen in liquid nitrogen and stored at 280uC.

Expression and Purification of SOR TTHB8 and Cytochrome c 550
Expression and purification of the proteins were conducted as described previously [16].

Determination of Cytochromes and Cytochrome Oxidases Concentration
UV/vis absorption spectra were recorded with a Perkin Elmer Lambda 5 spectrophotometer. Concentration of the proteins was obtained from the dithionite reduced-minus oxidized spectra using the following extinction coefficients: e = 18000 M 21 cm 21 (l = 550 nm) for cytochrome c 550 ; e = 21000 M 21 cm 21 (l = 552 nm) for cyt c 552 ; e = 6300 M 21 cm 21 (l = 613 nm) for ba 3 oxidase and e = 24000 M 21 cm 21 (l = 604 nm) for caa 3 oxidase. The concentration of SOR TTHB8 was determined using e = 67350 M 21 cm 21 (l = 280 nm).

Limited Proteolysis
Cytochrome c 550 was subjected to limited proteolysis by trypsin (Sigma-Aldrich) for 90 minutes at 37uC in 25 mM Tris-HCl pH 8.2, at a cytochrome:trypsin mass ratio of 100:1. To separate the proteolytic products, the reaction mixture was diluted with water and loaded on a FractogelH TMAE 650(S) column (Merck, Germany) equilibrated with 5 mM Tris-HCl pH 8.2. The flowthrough was collected, the column washed with the equilibration buffer and finally bound proteins were eluted with the same buffer containing 150 mM NaCl.

Construction of the Cytochrome c 550 Domains Expression Plasmids
The sequence encoding the mature N-terminal domain of cytochrome c 550 (c 550 [N]) was amplified from T. thermophilus HB8 genomic DNA by PCR using the primers 59-ATCTGAC-CATGGCTCAGACCACCCTCCCCGAG-39, containing NcoI restriction site (underlined), and 59-CAGTGACTCGAGT-CAGGCAGGGGTCTCCTGGGCTG-39, containing XhoI restriction site (underlined). Similarly, the sequence encoding the mature C-terminal domain of cytochrome c 550 (c 550 [C]) was amplified using the primers 59-ATCTGACCATGGCTCC-CAAAACGGGAGCCCAGGTCTAC-39, containing NcoI restriction site (underlined), and 59-CAGTGACTCGAG TCATGGCAGTTTGAGGCCTTGGCGGAG-39, containing XhoI restriction site (underlined). The products were NcoI and XhoI cloned into the expression vector pET22b+ (Invitrogen) to yield the pET22bC550N and pET22bC550C vectors. These constructs permit the expression of the recombinant domains in E. coli, fused to the pelB leader sequence for an optimal translocation to the periplasmic space.

Expression and Purification of the Recombinant Cytochrome c 550 N-terminal Domain (c 550 [N])
The BL21(DE3) E. coli strain was co-transformed with the pET22bC550N and pEC86 vectors [24], the latter containing the cytochrome maturation gene cluster necessary for the production of cytochrome c in E. coli under aerobic conditions [25]. The recombinant c 550 [N] was produced by growing the cells in LB medium containing ampicillin (100 mg/ml) and chloramphenicol (34 mg/ml) at 37uC for 24 h under shaking, and without protein expression inducers. Periplasmic proteins were prepared from fresh biomass. Cells were washed in PBS buffer (20 mM phosphate, 135 mM NaCl, 1 mM KCl, pH 7.4) and spun down at 80006g for 20 min at 4uC. The pellet was resuspended in 100 mM Tris-HCl pH 8 buffer containing 0.75 M sucrose. Osmotic shock was induced by slowly adding 2 volumes of ice chilled 1 mM EDTA. Following 10 min incubation at room temperature, spheroplasts were prepared by incubation with 1 mg/ml lysozyme for 45 min at room temperature under gentle shaking. Following the addition of 25 mM MgCl 2 , and 50 mg/ml DNaseI to reduce the viscosity of the extract, intact spheroplasts were removed by centrifugation at 80006g for 10 min at 4uC. The supernatant containing the recombinant domain was extensively dialyzed against 5 mM Tris-HCl pH 8.0 and then loaded on a FractogelH TMAE 650(S) (Merck, Germany) column equilibrated at 4uC with the same buffer. The protein was eluted with a gradient of NaCl (0-250 mM) and the fractions containing the protein were pooled, concentrated and desalted using a PD10 column (GE Healthcare, Germany) equilibrated with 5 mM Tris-HCl pH 8.0. The eluate was then loaded on a CaptoQ XL anion exchange column (GE Healthcare, Germany) equilibrated with the same buffer and eluted with a gradient of NaCl (0-150 mM). Fractions containing the domain were pooled, concentrated and finally purified by gel filtration on a Superdex 75 column (GE Healthcare, Germany) at 4uC with 5 mM Tris-HCl pH 8.0 buffer containing 150 mM NaCl. The isolated protein was concentrated by ultrafiltration, fast frozen in liquid nitrogen and stored at 280uC.

Expression and Purification of the Recombinant Cytochrome c 550 C-terminal Domain (c 550 [C])
The C-terminal domain was expressed and initially extracted as described above for the N-terminal domain. The supernatant containing the c 550 [C] domain was extensively dialyzed against 5 mM Tris-acetate buffer pH 6.0 and then loaded on a CM SepharoseH (Merck, Germany) column equilibrated at 4uC with the same buffer. The protein was eluted with a gradient of NaCl (0-250 mM) and the fractions containing the protein were pooled, concentrated and finally purified by gel filtration on a Superdex 75 column (GE Healthcare, Germany) at 4uC with 5 mM Tris-HCl pH 8.0 buffer containing 150 mM NaCl. The last step was repeated twice. The isolated protein was concentrated by ultrafiltration, fast frozen in liquid nitrogen and stored at 280uC.

Analytical Size Exclusion Chromatography
Analysis of the association of cytochrome c 550 domains was carried out by analytical size exclusion chromatography (SEC). 1.3610 28 moles of each domain were incubated for one hour at room temperature in 10 mM Tris-HCl, 50 mM NaCl, pH 8.0 before injection on a Superdex S75 10/30 column at a flow-rate of 0.5 ml/min. Elution profiles were recorded at 280 nm. The cytochrome full-length was analysed in a similar way.

Electron Transfer activity
Stopped-flow experiments were carried out with a thermostated instrument (DX.17 MV, Applied Photophysics, Leatherhead, UK), equipped with a 1-cm pathlength observation chamber. Reactions were investigated by monitoring the absorption changes at selected wavelengths. When necessary, ionic strength was adjusted by addition of KCl and the buffer was degassed with vacuum/N 2 cycles. Data were analyzed using the software MATLAB (The Mathworks, South Natick, MA).
Reduction by SOR TTHB8 of the cytochrome c 550 and its domains was investigated anaerobically at 45uC in 100 mM Tris-HCl pH 8.0 buffer containing 0.1 mM EDTA. To prevent inhibition of SOR TTHB8 resulting from prolonged incubation of the enzyme with a large excess of sulfite, in these experiments the stopped-flow instrument was used in the sequential mixing mode. Typically, 2 mM SOR TTHB8 was pre-mixed with 4 mM sulfite and after 500 ms further mixed with increasing amounts of oxidized cytochrome c 550 , c 550 [N], c 550 [C] or a 1:1 mixture of the two domains. The reduction of the cytochrome or its domains was monitored at 418 nm or, in the case of a too high signal in the Soret region, at 555 nm. The turnover rates (TN) of reduction of cytochrome c 550 and its domains were calculated by dividing the concentration (expressed in mM) of the c 550 sample reduced at t = t K by the half time of the reaction and the concentration of SOR TTHB8 in the experiment (typically 0.5 mM after mixing).
The kinetics of electron transfer between ascorbate-reduced cyt c 550 (or its domains) and cyt c 552 was assayed anaerobically at 4uC. This low temperature was chosen to slow-down and thus better resolve in time the reactions. Experiments were carried out in 5 mM Bis-Tris pH 7.0 buffer under non-pseudo-first order conditions, i.e., at comparable concentrations of the two proteins. Observed rate constants (k obs ) were therefore obtained by fitting the experimental time courses to the equations described in [26] for the analysis of bimolecular reactions assayed under secondorder conditions.
The oxidation of cytochrome c 550 and its domains by ba 3 -or

Results
The Cytochrome c 550 is Composed of Two Conformationally Independent Domains The recently identified periplasmic di-heme cytochrome c 550 [16] as a whole shows poor similarity to known proteins. However, when analyzed separately, its N-terminal domain sequence containing one heme binding site presents a high homology to the subunit B of SOR from C. jejuni [18], whereas the C-terminal domain, containing the other heme binding site, exhibits a high sequence identity with c 552 from T. thermophilus HB8. This led to the hypothesis that cytochrome c 550 is likely organized in two distinct domains, each one possibly presenting an independent fold and a heme cofactor, with distinct roles in mediating electron transfer between SOR TTHB8 and the respiratory chain, through cytochrome c 552 [16].
Consistently, according to the DomPred server [27] cytochrome c 550 consists of two domains, with a predicted boundary at residue 107 and a proline rich region (91-106 aa) likely representing a flexible inter-domain linker [28] ( Figure 1A). While globular proteins, due to their native rigid structure, are typically resistant to proteolysis under physiological conditions, flexible inter-domain linkers can be substrates for proteases. This makes the limited proteolysis approach suitable to confirm the multi-domain organization of a protein [29]. Based on this notion, cytochrome c 550 was subjected to limited proteolysis using trypsin. Overall, the cytochrome presents 19 putative cleavage sites for this protease ( Figure 1A). Among these sites, the one in position K105 is of particular interest, as it is located within the predicted linker, being therefore potentially more accessible to the protease. On this basis, limited proteolysis of the recombinant cytochrome c 550 is expected to yield the N-and C-terminal domains as the major cleavage products.
As c 550 [N] and c 550 [C] have significantly different calculated isoelectric points (5 and 7.97, respectively), the two domains are expected to have opposite net charges at pH 7, being easily separable by ion exchange (IEX) chromatography. The products of the limited proteolysis of cytochrome c 550 were therefore directly subjected to analytical IEX chromatography using the basic anionic exchanger FractogelH EMD TMAE (S) and subsequently analyzed by SDS-PAGE. As expected, two major products of limited proteolysis were obtained in addition to the band corresponding to non-digested cytochrome ( Figure 1B . This is most likely due to the presence of multiple trypsin cleavage sites within the protein, especially those located at the termini of the domains, more easily accessed by the protease ( Figure 1A). The limited proteolysis did not disturb the core of the domains or the heme binding, as the two generated fragments exhibited UV-Vis spectroscopic properties identical to those of the individually expressed domains (see below). This indicates that the two proteolytic fragments correspond to the two domains , each of them being independently folded and associated with one heme cofactor.
The use of recombinantly produced, isolated polypeptidic domains has proven to be a valuable approach to investigate intra-and intermolecular electron transfer between redox centers in multi-domain proteins with largely overlapping spectral properties [30,31]. Therefore, to further confirm the existence of two independent domains in cytochrome c 550 and to investigate their function, the N-and C-terminal parts of the protein were individually produced in E. coli. The recombinant domains contained the PelB signal sequence to promote their translocation to the periplasmic space and fragments of the flexible linker in order to enhance the stability of both domains ( Figure 1A Figure 2B). This shows that the splitting of the signal does not arise from the presence of two hemes. Instead, it is the result of the transition between the ground state and two or more excited states close in energy. Overall, the spectral properties confirmed that one heme cofactor was successfully incorporated in each domain during recombinant expression in E. coli.  Figure 2C). All together the results presented above indicate that cytochrome c 550 folds into two independent domains with distinct properties, each carrying a single heme group. These features may have implications with regard to the electron transfer activity of the cytochrome.

The N-terminal Domain of Cytochrome c 550 Accepts the Electrons from SOR TTHB8
We have previously shown that cytochrome c 550 is the physiological electron acceptor of the sulfite:cytochrome c oxidoreductase encoded by the ttha1326 gene in T. thermophilus HB8 [16]. Based on the high similarity between c 550 [N] and the SorB subunit of the sulfite:cytochrome c oxidoreductase from C. jejuni, it is very likely that this domain acts as an electron acceptor for SOR TTHB8 . The 3D model of SOR TTHB8 was automatically built by means of the SWISS-MODEL server [32][33][34], using as a template the only available structure for a microbial SOR, i.e., the one of the S. novella enzyme (PDB ID: 2c9x, segment A; 33% identity to SOR TTHB8 ) ( Figure 3A). In S. novella SOR, the formation of complementary electrostatic surfaces at the interface of the SorA and SorB subunit of sulfite:cytochrome c oxidoreductase has been observed [35]. Similarly, a positively charged surface area surrounding the pocket cradling the molybdopterin cofactor is present also in the SOR from T. thermophilus HB8 (Figure 3A), while the c 550 [N] presents an overall negative surface charge as shown by IEX chromatography. It is, therefore, reasonable to assume that c 550 [N] and SOR TTHB8 present analogies to typical SORs both in terms of function and type of interactions.
In order to test which domain of cytochrome c 550 preferably interacts with SOR TTHB8 , the electron transfer between the latter enzyme and the recombinant c  were mixed anaerobically with oxidized c 552 and the reaction was followed at 4uC at ionic strengths ranging from 2 to 152 mM. The reaction was studied under non pseudo-first order conditions and, accordingly, the experimental traces were fitted following the analysis described elsewhere [26]. From the results presented in Figure 5AB it can be concluded that, despite the low temperature,  Table 1 shows the estimated forward (k F ) and reverse (k R ) rate constants of the reaction measured at the same ionic strength (12 mM). The observed differences in k F and k R are not significant due to the rather high experimental error in those measurements, partly arising from the large optical overlap among the investigated proteins. Based on the results, we conclude that both c 550 [N] and c 550 [C] exchange electrons with cytochrome c 552 at rates similar to those previously measured with the fulllength cytochrome c 550 [16]. Interestingly, the reaction between c 550 [N] and c 552 displays a similar ionic strength dependence to the full-length protein, whereas the dependence is less pronounced in the case of c 550 [C] ( Figure 5C). This suggests that electrostatic forces play an important role in the interaction between c 550 [N] and c 552 , while apolar interactions may be involved in molecular recognition between c 550 [C] and c 552 .
In order to test this possibility, the 3D model of c 550 [C] was automatically built by means of the SWISS-MODEL server [32][33][34], using the X-ray structure of cytochrome c 552 from T. thermophilus HB8 (PDB ID: 1c52; ,50% identity) as template ( Figure 3BC). The heme cleft in the very well described cytochrome c 552 is surrounded by a hydrophobic belt ( Figure 3C) consisting of residues G13, C14, F26, V68, M69 and F72; this patch of residues likely participates in the interaction between the cytochrome and the terminal oxidase ba 3 [36]. Interestingly, the model structure of c 550 [C] clearly shows that such hydrophobic residues (G11, C12, F24, V66, M67 and F70) are structurally conserved around the cleft ( Figure 3B). It is therefore likely that molecular recognition between c 550 [C] and c 552 is also mediated by this hydrophobic patch, in line with the modest ionic strength dependence reported in Figure 5C. When assayed separately, c 550 [N], similarly to full-length c 550 , is very slowly oxidized by either ba 3 or caa 3 oxidase, whereas c 550 [C] is quickly oxidized by either of the two oxidases ( Figure 6A,B). Considering the high sequence similarity between c 550 [C] and c 552 and the structurally conserved hydrophobic belt ( Figure 3BC) likely participating in molecular recognition between c 552 and ba 3oxidase, this finding is perhaps not unexpected. On the other hand, we have previously shown (and confirmed here) that the electron transfer between full-length c 550 and ba 3 or caa 3 is not efficient, unless mediated by c 552 acting as an electron shuttle [16]. Therefore, we suggest that the relatively fast direct electron transfer between c 550 [C] and terminal oxidases here documented is of no physiological value. If the conserved hydrophobic patch mentioned above is involved in mediating electron transfer from c 550 [C], in the full-length protein such a patch should be not accessible by large molecules, like ba 3 or caa 3 oxidase, though possibly still allowing interaction with much smaller molecules such as cytochrome c 552 .
Interestingly, a 1:1 mixture of ascorbate-reduced c 550 [N] and c 550 [C] is also quickly and fully oxidized by ba 3 or caa 3 ( Figure 6A,B), which further confirms that the two domains of c 550 are in relatively fast redox equilibrium. Expectedly, in the presence of 10 nM oxidized c 552 , both c 550 [N] and c 550 [C] are promptly oxidized by ba 3 (Figure 6C,D) or caa 3 (data not shown), as previously observed with the full-length protein [16]. This further supports a fast electron transfer between either of the two domains of cytochrome c 550 and cytochrome c 552 .

Discussion
While sulfite oxidizing enzymes (SOEs) in vertebrates and humans have been studied for over 40 years, significant progress in studying bacterial SOEs has only been made in the last decade. Bacterial SOEs are extremely diverse in terms of structure and oxidizing proteins. Discovery of novel SOEs is being constantly reported, but in most cases the relative electron acceptor was not identified. Hence, usually the link to the respiratory chain is not shown [12][13][14]18,20]. The complexity of those pathways is a real challenge for understanding how SOEs participate in cell metabolism and energy production. Recently, the complete sulfite oxidation pathway from T. thermophilus has been described [16]; it was shown that the electrons generated during sulfite oxidation are injected into the respiratory chain at the level of cytochromes c 552 , via the di-heme cytochrome c 550. Here we have undertaken a detailed investigation of the role played by the novel cytochrome c 550 in linking sulfite oxidation to cell respiration.
Sequence analysis, limited proteolysis and individual expression of the N-and C-terminal regions of the cytochrome confirmed that overall the protein consists of two independent domains, each associated with one c-type heme. At physiological pH, the protein is most likely characterized by an asymmetric charge distribution: it has a negatively charged N-terminal domain and a positively charged C-terminal one. The detailed analysis of other di-heme ctype cytochromes revealed several features shared by c 4 -type cytochromes. Cytochromes c 4 are periplasmic or membranebound members of class I cytochromes c with a molecular mass of ,20 kDa found in a variety of bacteria [37,38]. Analysis of the characterized cytochromes c 4 from Vibrio cholerae [39], Pseudoalteromonas haloplanktis [40], Pseudomonas aeruginosa, Azotobacter vinelandii [41], Acidithiobacillus ferrooxidans [42] and Pseudomonas stutzeri [43,44], and the high resolution X-ray structures of the last two [45,46] show that, similarly to T. thermophilus c 550 , these di-heme proteins are formed by two domains connected with a flexible, ,10 aa long linker.
While T. thermophilus c 550 exhibits only ,15% identity to cytochromes c 4 , the latter cytochromes show relatively high sequence similarities not only one each other, but also between the two domains of the same cytochrome (though to a minor extent). This led to the hypothesis that cytochromes c 4 result from duplication of a common ancestral gene [45,46]. Given the low sequence conservation typically observed among multi-heme  cytochromes c [47], it would be tempting to postulate close relationships between T. thermophilus c 550 and cytochromes c 4 . In the case of cytochrome c 550 , however, the fusion of two distantly related cytochromes rather than gene duplication appears to be more likely; the domains indeed exhibit both remarkably different sequences and sizes (with c 550 [N] and c 550 [C] representing 1/3 and 2/3 of the full-length protein, respectively). It seems very likely that c 550 [C] originates from duplication and divergence of cytochrome c 552 [48]. Similarly, fusion of two distantly related cytochromes has been proposed also for the di-heme cytochrome c subunit of the flavocytochrome c sulfide dehydrogenase from Chromatium vinosum. This subunit exhibits ,15% identity to both cytochromes c 4 [49] and T. thermophilus cytochrome c 550 , while showing a fold very similar to that of cytochrome c 4 [50].
Despite the low sequence similarity, the dipolar nature of T. thermophilus cytochrome c 550 (resulting from an asymmetric charge distribution between the two domains) has been also reported for the cytochromes c 4 isolated from P. stutzeri and A. ferrooxidans [43,45,46]. It has been proposed that the dipolar nature of the cytochrome c 4 from P. stutzeri is important for the interaction of the cytochrome with its redox partners. The positively charged Cterminal domain is proposed to interact with the negatively charged pocket of cytochrome c oxidase, while negatively charged N-terminal domain with a reductase [45]. However, despite the extensive characterization of P. stutzeri cytochrome c 4 (see references in [44]), its specific physiological function remains to be established as yet. Conversely, analysis of the cytochrome c 4 from A. ferrooxidans showed the crucial role of a negatively charged residue E121 localized on the overall positively charged Cterminal domain in recognition of its electron donor, rusticyanin, while the Y63 localized in the negatively charged N-terminal domain seems to be responsible for the electron transfer from c 4 to cytochrome c oxidase [51].  The results presented here clearly show that the N-(but not the C-) terminal domain of T. thermophilus cytochrome c 550 binds to and accepts electrons from SOR TTHB8 . Based on the 3D model generated for SOR TTHB8 and IEX chromatography experiments on c 550 [N], recognition between these two proteins most likely involves electrostatic interactions between the positively charged area surrounding the molybdopterin cofactor in SOR TTHB8 and the overall negative charge of c 550 [N]. Electrostatic interactions have been also proposed to drive the reaction of SorT from S. meliloti with its negatively charged natural electron acceptor, cytochrome Smc04048 [20]. The importance of electrostatic interactions between SOR TTHB8 and c 550 [N] is further supported by the presence of a conserved arginine residue (R50) in the SOR TTHB8 in a position that has been identified as crucial for electron transfer in SorA from S. novella (R55) [35,52]. Functionally, therefore, the c 550 [N] resembles the heme subunit of a classical Group 1 SOR, and the full-length cytochrome c 550 resembles the cytochrome c subunit of the flavocytochrome c sulfide dehydrogenase from Chromatium vinosum, where the N-terminal part of the di-heme cytochrome c tightly interacts with the flavin-containing enzymatically active subunit of the protein.
Our results show that the two domains of cytochrome c 550 can rapidly exchange electrons once the c 550 [N] has been reduced by SOR TTHB8 . Furthermore, both domains are also able to reduce cytochrome c 552 , the molecule connecting sulfite oxidation directly to the respiratory chain. The c 550 [C] most likely binds c 552 via hydrophobic interactions and electron transfer is therefore only modestly affected by the ionic strength. Hydrophobic interactions have also been proposed to be involved in the reaction of cytochrome c 4 from A. ferrooxidans [46,51] with its electron acceptor. Moreover they have also been shown to be particularly important in T. thermophilus [30] and in thermophilic organisms in general, as electrostatic interactions are weakened at high temperatures [53].
The finding that both domains of T. thermophilus c 550 can pass electrons to c 552 rules out the possibility that the protein acts as a wire, where one heme serves as the entrance and the other as the exit site for electrons. Our results show that in principle the pathway could be functional with only c 550 [N], as this domain is able by itself to shuttle electrons from SOR TTHB8 to c 552 . The c 550 [C] seems, therefore, dispensable for c 550 to fulfill its function. However, the presence of a second heme center in rapid redox equilibrium with both c 550 [N] and c 552 enables cytochrome c 550 to assist a two-electron transfer process, similarly to the di-heme subunit of flavocytochrome c sulfide dehydrogenase from C. vinosum [50]. This confers to cytochrome c 550 the ability to store the two electrons generated during oxidation of a single sulfite molecule and inject them into the terminal oxidases of the respiratory chain via cytochrome c 552 , thereof ensuring an efficient coupling between sulfite oxidation and the respiration-mediated energy production.