Role of tryptophan residues of Erv1: Trp⁹⁵ and Trp¹⁸³ are important for its folding and oxidase function

Erv1 (essential for respiration and viability 1) is an essential component of the mitochondrial import and assembly (MIA) pathway, playing a critical role during import and oxidative folding of the mitochondrial intermembrane space (IMS) proteins [1–3]. In the MIA pathway, Mia40 acts as a thiol oxidoreductase and import receptor, interacting directly with the substrates of the pathway {e.g. Tim9 (translocase of inner membrane 9), Tim10, Cox17 (cytochrome oxidase 17) and Ccs1 (copper chaperone for superoxide dismutase 1)}. Erv1 is a FAD-dependent thiol oxidase belonging to the single domain ERV/ALR (augmenter of liver regeneration) oxidase family. It functions downstream of Mia40, catalysing re-oxidation of the reduced Mia40 and transferring electrons (via FAD) to molecular oxygen and/or cytochrome c [4–10].

Saccharomyces cerevisiae Erv1 has a total of 189 amino acids including a highly conserved (among ERV/ALR family) FAD-binding domain of approximately 100 amino acids at the C-terminus (Figure 1). There are six conserved cysteine residues forming three pairs of disulfide bonds, with four of the cysteine residues located in the FAD-binding domain. Several structures of the FAD-binding domain of ERV/ALR proteins have been published [10–15], in which they are all crystallized as head-to-tail homodimers. Each subunit contains a four-helical bundle (H1–H4) that harbours the FAD cofactor and an additional (fifth) single-turn helix (Figure 1). The FAD-binding domain acts as a catalytic core (also called catalytic domain) containing a CXXC redox centre disulfide (Cys¹³⁰–Cys¹³³), located proximal to the isoalloxazine ring of FAD cofactor and a C-terminal CX₁₆C structural disulfide. The shuttle disulfide (Cys³⁰–Cys³³) of Erv1 is located in the non-conserved N-terminal domain. It accepts the electrons from reduced Mia40 and transfers them to the active-site disulphide (C¹³⁰-C¹³³) and then in turn to the cofactor FAD and cytochrome c and/or molecular oxygen [4,16–19]. Erv1 and ALR employ a similar catalytic mechanism which involves several intermediate states, including a shuttle disulfide reduced state, S→FAD charge-transfer complexes and an FADH₂-Erv1 state, during their catalytic cycle [17,20–22]. Whilst the shuttle and the active-site disulfides are functionally essential, the conserved CX₁₆C disulfide (C¹⁵⁹-C¹⁷⁶) plays an important role in stabilizing the folding of Erv1 [6,23].

Tryptophan plays important roles in protein stability despite its scarcity in proteins. Erv1 has six tryptophan residues that are all located in the C-terminal FAD-binding domain making the domain tryptophan-rich. Among them Trp⁹⁵, Trp¹⁵⁷ and Trp¹⁸³ are highly conserved across the ERV/ALR family; Trp¹⁸⁷ is partially conserved; whereas Trp¹³² and Trp¹⁷⁹ are not conserved (Figure 1a). Trp⁹⁵, Trp¹³² and Trp¹⁵⁷ are located in H1, H3 and H4 respectively; Trp¹⁷⁹ and Trp¹⁸³ are in the fifth single-turn helix (H5) and Trp¹⁸⁷ is located at the very end of C-terminus, the third residue from the end (Figure 1a). Interestingly, the highly conserved Trp¹⁸³ is next to the highly conserved residue Arg¹⁸², whose mutation in the human homologue ALR caused diseases. In 2009, Di Fonzo et al. [24] reported the first disease-associated mutant of ALR, in which a single conserved arginine to histidine substitution (ALR^R194H) caused an autosomal recessive myopathy. Structural and functional study of the corresponding Erv1^R182H mutant showed that Arg¹⁸² plays an important role in the folding and FAD binding of Erv1 [25]. Although the oligomerization state of Erv1 in vivo is unclear, our recent in vitro study showed that whilst the purified full-length Erv1 formed a stable tetramer, the Erv1^R182H mutant was a homodimer under the same experimental condition [25]. Visualization of the structures of Erv1 [11] indicates that both Trp ⁹⁵ and Trp¹⁸³ side chains are involved in stabilizing the cofactor binding. Firstly, Trp⁹⁵ forms an H-bond with the OH group of the ribitol moiety of FAD and Trp¹⁸³ forms an H-bond with the nitrogen of adenine moiety of FAD using its side chain, suggesting a role similar to Arg¹⁸² in stabilizing FAD binding. Secondly, both Trp⁹⁵ and Trp¹⁸³ seem to play an important part in forming a series of π stackings with the cofactor FAD, involving the isoalloxazine ring of FAD, side chains of Trp⁹⁵, Tyr¹²⁸, His¹⁶², Phe¹⁷⁴, Trp¹⁸³ and the adenine ring of FAD (Figure 1c). In particular, the invariable Trp⁹⁵ of ERV/ALR and Ero1 proteins locates in the centre of the aromatic stacking and is sandwiched between the isoalloxazine ring of FAD and the invariable His¹⁶² [11,13,26,27]. Though an important role can be assumed for some of the tryptophan residues, no experimental study on the importance of the tryptophan residues and the π stackings in ERV/ALR enzymes’ structure and function has been reported.

In the present study, we investigated the structural and functional roles of all six tryptophan residues of Erv1 with a focus on Trp⁹⁵, using biochemical and biophysical methods, as well as computational and yeast genetic approaches. First, a set of six single tryptophan-to-phenylalanine yeast mutant strains were generated and their effects on cell viability were tested at various temperatures. Then, the tryptophan mutants were expressed and purified from Escherichia coli. Their effects on Erv1 folding, FAD-binding, stability and oxidase activity were studied using size exclusion chromatography, spectroscopic (absorption, CD and FAD fluorescence) and oxygen consumption analyses. Our results show that except Erv1^W183F, all mutants were purified as a tetramer like the wild-type (WT) protein, whereas Erv1^W183F was a dimer. Erv1^W95F has the strongest effect on the stability and hence the function of Erv1, which is followed by Erv1^W183F. Though the other mutants also result in observable effects on the folding and stability of Erv1, there is no obvious defect in the oxidase activity of the protein, including the highly conserved Trp¹⁵⁷ mutant. Finally, computational analysis was performed to provide insight into the FAD-binding energy landscape at the atomic level, which assisted our understanding of the molecular mechanism by which Erv1^W95F causes the functional defect of Erv1.

A construct encoding C-terminally LE(H)₆-tagged full-length, WT Erv1 cloned into E. coli expression vector pET24a(+) (Novagen) using the Nde1 and Xho1 restriction sites was used as DNA template [28] to generate all mutant constructs for protein purification work in the present study. For yeast studies, ERV1 together with endogenous promoter and terminator regions was inserted into the centromeric plasmid pRS414 as previously described [17] and was used as a DNA template for mutagenesis. Tryptophan-to-phenylalanine mutations were introduced by PCR using overlapping primer pairs containing the desired mutation point. Plasmid DNA constructs with the correct mutation was verified by DNA sequencing.

WT Erv1 and tryptophan mutants were expressed in E. coli strain Rosetta-gami™ 2 (Novagen) and purified as previously described [17,25]. Briefly, protein was expressed in the presence of 0.5 mM IPTG and 10 μM FAD at 16 °C for 16–20 h. Induced cell pellets resuspended in buffer A (150 mM NaCl, 50 mM Tris/HCl, pH 7.4) containing 5 mM imidazole, 50 μM FAD and one tablet of EDTA-free protease inhibitor cocktail (Roche) were lysed by sonication on ice. The supernatant fraction containing LE(H)₆-tagged Erv1 was bound to 2–3 ml of Ni²⁺-charged histidine. Bind resin (Novagen) pre-equilibrated with binding buffer. Following a washing step with 20–30 ml of wash buffer (buffer A plus 20 mM imidazole), the protein was eluted with 4–6 ml of elution buffer (buffer A containing 500 mM imidazole). FAD (100 μM) was added to eluted proteins before storage at −80°C until further use. The affinity-purified proteins were further separated for use in in vitro studies by gel filtration chromatography using a Superdex 200 (or Superdex 75) 10/30 column connected to an ÄKTA-FPLC system (GE Healthcare) at 4°C in BAE (150 mM NaCl, 50 mM Tris/HCl, 1 mM EDTA, pH 7.4).

Absorption spectra were recorded from 250 to 700 nm, at 1 nm intervals, in a 1 cm path-length quartz cuvette using a Cary 300 Bio UV-Visible spectrophotometer (Varian Ltd.). Measurement of the molar extinction coefficient was done in BAE as previously described [6,7]. The calculated molar extinction coefficients are summarized in Table 1. The FAD content of the WT and mutant Erv1 was calculated based on the eqn (1), an extinction coefficient for free FAD (fFAD) of 11.3 mM⁻¹ cm⁻¹ and the absorption spectrum of Erv1 or mutants in BAE plus 1% SDS.

ε₂₈₀ is the extinction coefficient of the apoprotein; 42.54 mM⁻¹ cm⁻¹for the WT and 36.85 mM⁻¹ cm⁻¹ for the single tryptophan-to-phenylalanine mutants of Erv1. A₂₈₀ and A₄₅₀ are absorbance at 280 and 450 nm respectively. The ratio of ε₂₈₀/ε₄₅₀ of fFAD in BAE plus 1% SDS determined in the present study is 1.82.

CD analysis was done using a Chirascan CD spectrometer (Applied Photophysics Ltd.) and either a 1 mm (far UV) or 5 mm (near UV) path-length quartz cuvette. Each spectrum represents an average of four independent scans with the spectra for buffer alone subtracted. Thermal denaturation was measured at 222 nm in 1°C intervals over 5°C–90°C, with a temperature increase of 1°C/min. Melting temperatures (T_m) were calculated by obtaining the first derivative of each profile and finding the maximum.

Fluorescence measurements were recorded using a Cary Eclipse fluorescence spectrophotometer (Varian Ltd.) in a 1 cm × 0.2 cm path-length quartz cuvette. Thermal denaturation was followed by the increase in FAD fluorescence at 535 nm after excitation at 455 nm, in 1°C intervals between 5°C–90°C, with a temperature increase of 1°C/min.

Erv1 oxidase activity was measured using a Clark-type oxygen electrode (Hansatech Instrument Ltd) in a 0.5 ml of reaction volume at 25°C or 37°C in BAE as previously described [6]. Data analysis of the oxygen consumption profile and the calculation of the reaction slope were performed using the Microcal™ Origin™ statistical software package.

A ∆erv1 knockout strain containing WT ERV1 in a URA3 plasmid [24] was co-transformed with WT or mutant ERV1 on TRP1 plasmid pRS414. For counter-selection tests, the strains were spotted on 5-fluoroorotic acid (FOA) plates and incubated for up to 2–3 days at 25°C, 30°C or 37°C.

All experiments were carried out in BAE (150 mM NaCl, 50 mM Tris/HCl, 1 mM EDTA, pH 7.4) unless specifically stated. For multi-angle laser light-scattering, protein samples were applied to a Superdex 200 10/30 gel filtration column (GE Healthcare) running with buffer A. Proteins eluting from the column passed through an in-line DAWN EOS laser photometer set at 682 nm and an Optilab rEX refractor. To calculate the weight-averaged molecular mass, the light scattering intensity and eluent refractive index were analysed using ASTRA version 4.8 software.

MD simulations of the WT and Erv1^W95F mutant were performed using AMBER12 [29] with the protein modelled by the AMBER ff99SB force field [30] and the substrate FAD represented using the parameters and partial charges developed by Asada et al. [31]. The initial protein co-ordinate was taken from the crystal structure of the C-terminal core domain (CTD) of Erv1 from Saccharomyces cerevisiae at 2.0 Å (1 Å=0.1 nm) resolution (PDB code: 4E0H) [11]. In the crystal structure, the CTD of Erv1 exists as a homodimer, with each subunit formed of a four-helix bundle that accommodates FAD binding. All crystallographic waters were removed whereas hydrogens were added using the LEAP program within AMBER12. The protein–FAD complex consisting of 206 amino acid residues (103 residues in each subunit) and two FADs was solvated in a rectangular TIP3P water box with at least 8 Å between the edge of the box and the protein. Chloride ions were added to neutralize the overall change of the system. Following minimization, the system was heated to 298 K, equilibrated for 200 ps under the constant pressure condition and the production simulation was performed for 1 ns.

The MM-GBSA approach [32,33] implemented in AMBER12 was used to study the FAD binding energy to Erv1 WT (ΔE_WT) and Erv1^W95F mutant (ΔE_W95F). A total of 500 frames from the 1-ns MD trajectory were used. The MM–GBSA free energy was obtained from a combination of the gas phase molecular mechanics energy term (E_MM=E_ele + E_vdw + E_int) and the solvation free energy term (G_solvation=G_polar + G_nonpolar) without the solute entropy term. The difference in the energy of FAD binding in the WT and W95F was derived by ΔΔE=ΔE_W95F − ΔE_WT.

To investigate the requirement of individual tryptophan residues of Erv1 in its function in vivo, we generated six tryptophan mutant yeast strains by expressing plasmid-encoded erv1 single tryptophan-to-phenylalanine mutants, in an ERV1 knockout strain ∆erv1, isogenic to the WT Trp³⁰³, as described previously [24]. Then, effects of the mutation on the cell viability were checked by spot-testing at 25°C, 30°C and 37°C (Figure 2). Upon counter-selection of WT ERV1 (on URA3-containing plasmid), the Erv1^W95F mutant was temperature sensitive at both 30°C and 37°C, Erv1^W183F grew normally at 25°C and 30°C, but slowly at 37°C, whereas the other four mutants showed no obvious growth defect at all three temperatures. This result confirmed that Trp⁹⁵ and to a lesser extent Trp¹⁸³ play important roles in the function of Erv1 in vivo. The other four tryptophan residues, including the highly conserved Trp¹⁵⁷, were not important for cell viability under these experimental conditions.

To understand the structural and functional roles of tryptophan residues of Erv1, six single tryptophan-to-phenylalanine mutant constructs of Erv1 were generated and expressed in E. coli. All mutants were successfully purified using the same method as that used for the WT protein, as described previously [6,17]. Gel filtration chromatography analysis showed that apart from Erv1^W183F all other five mutants eluted in the same or similar position as the WT protein and thus existed mainly in a tetramer form like the WT Erv1 [17,25] (Figure 3a; Supplementary Figure S1a). Elution of Erv1^W183F was delayed and the molecular mass of Erv1^W183F was determined using light scattering analysis to be 44±3 kDa (Supplementary Figure S1b), suggesting that Erv1^W183F formed a dimer (monomer Erv1 is 22 kDa).

Though there is a difference in oligomerization state, all six mutants displayed the same yellowish colour as the WT protein and showed very similar overall UV-vis absorption spectra profiles (Figure 3b). The visible absorption maximum λ_max was shifted most by Erv1^W95F, from 460 nm (for WT) to 458 nm (Table 1). The result suggests that FAD cofactor binding was not significantly affected by the mutation. Consistently, similar extinction coefficients were obtained as shown in Table 1, which were determined by adding of 1% SDS to release FAD as described previously [6,7]. Moreover, FAD content of each purified mutant was determined (see methods). Apart from Erv1^W183F all the other five mutants had ∼95% FAD content, very similar compared with the WT Erv1 and thus one FAD per monomer protein. However, the FAD content of Erv1^W183F decreased by more than 20%, indicating that Trp¹⁸³ and possibly also the tetramer formation play a role in stabilizing cofactor binding.

Next, far and near UV CD analyses were carried out to investigate the effects of the tryptophan mutants on the folding of Erv1. The spectral profiles were compared with that of the WT protein in relative intensity since they were different in their FAD content and thus holoprotein concentration. The far UV CD spectra (Figure 4a) were normalized at 208 nm. Overall, all the spectra showed characteristics of highly α-helical conformations, but with some differences in the region from 215 to 240 nm, especially for Erv1^W95F and Erv1^W183F. Subtraction of the Erv1^W183F spectrum from that of the WT gave a bell-shaped peak centred at 230 nm, characteristic of strong aromatic–aromatic interactions (Figure 4b). In terms of the differences between the WT and Erv1^W95F spectra, apart from a positive peak at ∼230 nm, there was also a negative peak at ∼220 nm suggesting loss of some α-helical structure in Erv1^W95F. Similarly, clear differences in the near UV CD spectra between the WT and these two mutants (Erv1^W95F and Erv1^W183F) were observed (Figure 4c). The result showed that the peak of Erv1 near UV CD at 285 nm was contributed mainly by Trp⁹⁵, whereas Trp¹⁸³ seemed to contribute to the peak at 295 nm significantly. Furthermore, the facts that the near UV CD spectrum of Erv1^W183F was closely similar to that of Erv1^R182H and that they both formed a dimer (rather than tetramer) indicated that the signal at 295 nm may be associated with formation of an Erv1tetramer.

Taken together, these results showed that both Trp⁹⁵ and Trp¹⁸³ are structurally important in folding and FAD binding of Erv1. Whereas Erv1^W95F affected the secondary and tertiary structure of Erv1 without affecting its oligomerization state, Erv1^W183F disrupted not only the oligomeric state, but also the secondary structure and tertiary aromatic–aromatic interactions and decreased FAD content under the experimental conditions.

It has been shown that FAD release and thermal unfolding of Erv1 are a co-operative process [25]. Thus, we investigated how each tryptophan mutation affects the thermal stability of the protein based on FAD fluorescence intensity change. Firstly, FAD fluorescence spectra of the WT and six mutants and the same concentration of fFAD were recorded at 5°C. As shown in Figure 5(a), whereas W95F displayed ∼20% fFAD intensity, negligible fluorescence intensity was detected for the WT and other mutants due to Erv1 binding quenching FAD fluorescence. Secondly, temperature-dependence of FAD fluorescence intensity change and thus the cofactor release was measured (Figure 5b) and the results were summarized in Table 1. Apart from W132F, which showed the same midpoints of thermal denaturation (T_m) of 64°C as the WT Erv1, all other five mutants displayed decreased thermal stability. W95F is the most unstable mutant with a T_m of 41°C, which is 23°C lower than that of the WT protein. For W183F, T_m of 51°C, a decrease of ∼13°C was observed (Table 1). W95F had the strongest effect on the stability of Erv1, followed by W183F, W157F and the non-conserved tryptophan mutants W179F, W187F, whereas W132F showed no obvious effect on the thermal stability of Erv1 (Figure 5b; Table 1). This order is not surprising, as the first three residues (W95F, W183F and W157F) are highly conserved whereas the last three are not (Figure 1). Similar T_m values were determined for the mutants based on CD intensity change at 222 nm (Table 1), showing that Erv1 unfolding and cofactor release was a coupled process.

To gain more understanding of the reversibility of the cofactor binding, the FAD fluorescence spectra of the proteins after thermal denaturation were re-recorded at 5°C (Figure 5c). There was little change to WT Erv1 and fFAD following renaturation. However, clear FAD fluorescence intensity increases were observed for all six mutants (Figures 5c and 5d), showing that there was a defect in re-folding and/or cofactor binding in the mutants after thermal denaturation. In summary, consistent with the conclusion above, temperature-dependent studies showed Erv1^W95F and Erv1^W183F had the strongest effect on the stability of Erv1 and FAD-binding, though all six mutants displayed decreased thermal stability of Erv1.

To understand the importance of the tryptophan residues in Erv1 function, the oxidase activity of the WT and mutants was analysed using oxygen consumption assay with 5 mM TCEP [tris-(2-carboxyethyl) phosphine] as substrate as established previously [25]. The relative activity of each mutant was calculated based on their initial oxygen consumption rate compared with that of the WT Erv1 (Table 2; Figure 6a). At 25°C and 30°C, only Erv1^W95F showed a clear functional defect, with ∼50% activity of the WT Erv1 (Table 2). At 37°C, both Erv1^W95F and Erv1^W183F displayed a significant functional defect; which Erv1^W95F showed ∼9% and Erv1^W183F had ∼53% activity of the WT Erv1 respectively (Figure 6a). Under the same conditions, the other four mutants, including the highly conserved Trp¹⁵⁷ mutant, showed no obvious functional defect and Erv1^W187F was slightly more active than the WT Erv1 at all three temperatures. These results are consistent with our in vivo findings and provide a good explanation for the temperature sensitive growth phenotypes we observed (Figure 2).

Our previous study showed that the Erv1^R182H, a disease related mutant, has stronger effects on folding and FAD-binding of the catalytic reaction intermediates of Erv1 than the oxidized steady state enzyme [25]. The cofactor was released readily from Erv1^R182H during its catalytic cycle leading to complete inactivation of the enzyme. In the present study, although only ∼30% of the oxidized Erv1^W95F and 10% of Erv1^W183F were denatured at 37°C (Figure 5b), the activity loss was higher, ∼90% and 50% respectively. Thus, we reasoned that similar to Erv1^R182H, Erv1^W95F and Erv1^W183F had stronger destabilization effects on folding and FAD-binding of the reaction intermediates than on the initial oxidized Erv1. To test this hypothesis, the oxygen consumption experiment was carried out in the presence of fFAD. As expected, the oxidase activities of these mutants were recovered by addition of fFAD and in a fFAD concentration dependent manner (Figure 6b). Moreover, no obvious differences in the stability of the oxidized enzymes were observed based on thermal denaturation studies. T_m of 66±1°C for the WT and 46±2°C for the W95F mutant were obtained in the presence of 10 μM fFAD, which are the same as those determined in the absence of fFAD (Table 1).

To understand the effects of W95F mutant on the stability of FAD-binding further, MD simulations (See ‘Experimental’) of the WT and Erv1^W95F mutant were performed based on the crystal structure of the CTD of Erv1 at 2.0 Å resolution (PDB code: 4E0H) [11]. The overall FAD-binding energies of the WT Erv1 [−71.1±6.0 kcal/mol (1 cal ≡ 4.184 J)] and Erv1^W95F (−73.7±4.7 kcal/mol), as calculated by the MM-GBSA method, are similar. When the binding energy is decomposed into the contribution of individual residues, key residues that contribute to FAD-binding are identified as shown in Figures 7(a) and 7(b). The residues include Gly⁹¹, Arg⁹², Trp⁹⁵, Tyr¹²⁸, Cys¹³³, Phe¹³⁷, Cys¹⁵⁹, His¹⁶², Val¹⁶⁵, Lys¹⁷¹, Phe¹⁷⁴, Arg¹⁸² and Trp¹⁸³. Consistently, these key residues are all located in proximity to and interacting with the isoalloxazine ring, ribitol moiety and adenosine moiety of FAD respectively (Supplementary Figure S1). However, apparent differences in the contribution of individual residues between the WT and the W95F mutant were observed. The residues with significant binding energy changes (ΔΔE > ±0.5 kcal/mol) are Val⁸⁷, Arg⁹², Trp/Phe⁹⁵, Tyr¹²⁸, Trp¹³², Cys¹³³ and Lys¹⁷¹ (Figure 7c) and interestingly they are all positioned in close proximity to the isoalloxazine ring of FAD (Figure 7d). The result indicates that Trp⁹⁵ plays a key role in stabilizing the isoalloxazine ring in its positioning within the binding site and how it interacts with key residues, such as Cys¹³³ for which the binding energy to FAD is changed by 1.55 kcal/mol in Erv1^W95F (Figure 7c).

In the present report, we investigated the structural and functional roles of all six tryptophan residues of Erv1, using yeast genetic, biochemical and computational methods, with focus on the highly conserved Trp⁹⁵. We showed that all the tryptophan mutants apart from W132F, had an impact on the thermal stability of Erv1 as evidenced by their decreased Tm (Figure 5; Table 1). In terms of function, the three non-conserved tryptophan residues (Trp¹³², Trp¹⁷⁹ and Trp¹⁸⁷) and Trp¹⁵⁷, a highly conserved residue in the ERV/ALR sub-family, seem not to be essential for Erv1 function under normal cell growth or our experimental conditions (Figure 1; Table 2). However, both Trp⁹⁵ and Trp¹⁸³ play an important role in stabilizing the folding and FAD-binding and thus the function of Erv1.

Mutagenesis studies of several FAD/FMN-dependent enzymes suggested that aromatic residues forming π-stacking interactions with the isoalloxazine ring of FAD/FMN help stabilize formation of the anionic flavin hydroquinone intermediate [34–36]. Similarly, in our study, mutating Trp⁹⁵ into phenylalanine affects the π–π repulsive interactions between the ring systems, which may modulate the catalytic steps for the formation of the S→FAD charge–transfer complexes. Our experimental results showed that Erv1^W95F has the strongest effect on the stability and oxidase activity of Erv1. The functional defect of Erv1^W95F is largely due to the destabilizing effects on the catalytic reaction intermediates since its activity can be recovered and sustained by addition of fFAD in the reaction. This observation is consistent with the result of our computational analysis, suggesting that Trp⁹⁵ is crucial in stabilizing the isoalloxazine ring to interact with Cys¹³³, hence the redox centre disulfide. Consequently, Erv1^W95F mutation causes a strong functional defect of Erv1 and impaired cell growth at both 30°C and 37°C.

Trp¹⁸³, on the other hand, is not directly involved in the π-stacking network and only forms an H-bond with the nitrogen of adenine moiety of FAD. This serves to maintain the orientation of the adenine ring (FAD is in a bent conformation with the isoalloxazine and the adenine ring buried in the active site and sandwiched between Trp⁹⁵ and His¹⁶²). Therefore, mutating Trp¹⁸³ to phenylalanine is not as severe as W95F in terms of its effect on Erv1 oxidase function (Table 2). However, Erv1^W183F showed a significant effect on the oligomerization state of Erv1, which leads to a clear functional and growth defect at 37°C but not at 30°C. This result is the same as that caused by Erv1^R182H, a disease related mutant, forming a dimer under the same experimental conditions [25]. The fact that both Arg¹⁸² and Trp¹⁸³ are part of the fifth single turn α-helix (H5) of Erv1 shows that this region is important in stabilizing the tetramer formation. Furthermore, Arg¹⁸² may form intermolecular hydrogen bonds to stabilize the tetramer formation, which may be assisted by a typical side-chain cation–π interaction with Trp¹⁸³. Since tryptophan is more favoured to interact with cationic side chains than phenylalanine [37], we hypothesize that mutation of Trp¹⁸³ disrupts the cation–π interaction with Arg¹⁸², causing disruption of the intermolecular interactions mediated by Arg¹⁸² and subsequent dissociation of the tetrameric structure of Erv1. Such a molecular mechanism of oligomerization formation has been observed previously [38]. Together with the observation that purified W183F mutant showed a decreased FAD content than the WT and other tryptophan mutants (Table 1), our results suggest that tetramer formation plays an important role in stabilizing the cofactor binding by Erv1 through the cation–π interaction between Arg¹⁸² and Trp¹⁸³ and Trp¹⁸³’s contribution in maintaining the π–π stacking network.

More generally, the aromatic ring stacking is conserved not only in ERV/ALR family, but all FAD-dependent thiol oxidases, including Ero1 (a thiol oxidase of the endoplasmic reticulum) that has no sequence homology with ERV/ALR enzymes [26,27,39]. Thus, it is tenable that the extensive π–π stacking around FAD is the key to create a rigid active site to receive electrons from the active-site disulfide (Cys¹³⁰–Cys¹³³) and pass electrons to molecular oxygen. Taken together, the current study contributes to our understanding of how thiol oxidases use FAD to catalyse the disulfide bond formation and broadly, provide insights into the function and mechanism of the structurally diverse flavinenzymes.

Qi Wang, Swee Kim Ang, Efrain Ceh-Pavia and Hui Lu performed experiments and analysed the data. Jiayun Pang performed computational analyses. Hui Lu and Jiayun Pang wrote the paper with help, contribution and feedback from all other authors.

We thank Marjorie Howard of the faculty biomolecular analysis core facility for help with light scattering measurement, John Milward and Abid Javed for initial preliminary work, Michael Spiller for advice on yeast genetics and Andy Munro and Derren Heyes for helpful comments on the manuscript.

This work was supported by the Biotechnology and Biological Sciences Research Council [grant number BB/H017208]; the Leverhulme Trust [grant number F/00120/CB]; the UK National Service for Computational Chemistry Software (to J.P.); and the UK High-End Computing Consortium for Biomolecular Simulation (to J.P.).

ALR

augmenter of liver regeneration

Erv

essential for respiration and viability

fFAD

free FAD

FOA

5-fluoroorotic acid

MIA

mitochondrial import and assembly

TCEP

tris-(2-carboxyethyl) phosphine

WT

wild-type