Protein Design with Fluoroprolines: 4,4-Difluoroproline Does Not Eliminate the Rate-Limiting Step of Thioredoxin Folding.

C4 -substituted fluoroprolines (4R)-fluoroproline ((4R)-Flp) and (4S)-fluoroproline ((4S)-Flp) have been used in protein engineering to enhance the thermodynamic stability of peptides and proteins. The electron-withdrawing effect of fluorine can bias the pucker of the pyrrolidine ring, influence the conformational preference of the preceding peptide bond, and can accelerate the cis/trans prolyl peptide bond isomerisation by diminishing its double bond character. The role of 4,4-difluoroproline (Dfp) in the acceleration of the refolding rate of globular proteins bearing a proline (Pro) residue in the cis conformation in the native state remains elusive. Moreover, the impact of Dfp on the thermodynamic stability and bioactivity of globular proteins has been seldom described. In this study, we show that the incorporation of Dfp caused a redox state dependent and position dependent destabilisation of the thioredoxin (Trx) fold, while the catalytic activities of the modified proteins remained unchanged. The Pro to Dfp substitution at the conserved cisPro76 in the thioredoxin variant Trx1P did not elicited acceleration of the rate-limiting trans-to-cis isomerization of the Ile75-Pro76 peptide bond. Our results show that pucker preferences in the context of a tertiary structure could play a major role in protein folding, thus overtaking the rules determined for cis/trans isomerisation barriers determined in model peptides.

Introduction

Proline displays unique structural features among the proteinogenic amino acids. Two of the main chain atoms of Pro, Cα and nitrogen, are constricted within the pyrrolidine ring, thus conferring conformational restriction on the phi angle in the peptide bond. In folded proteins, although the trans conformation of peptide bonds is still slightly favoured between any residue and the following proline (Xaa‐Pro), the steric hindrance of the pyrrolidine ring is not fully alleviated by the adoption of the trans conformation. Thus, cis and trans conformations of Xaa‐Pro peptide bonds are almost isoenergetic. In the cellular environment, the trans to cis isomerisation of prolyl peptide bonds where Pro is in the cis conformation in the native state is a slow process in the time range of several minutes. In nature, peptidyl‐prolyl cis‐trans isomerases (PPIases) catalyse the formation of the correct peptide conformation by diminishing the energy barrier between the two conformational states. In living organisms, the switching between cis and trans conformation catalysed by PPIases can act as a molecular timer that regulates the timing of key biological processes. Trans to cis isomerisation is often the rate limiting folding step in the in vitro refolding process for proteins displaying a Pro residue in the cis conformation in the native state. The pyrrolidine ring of Pro generally adopts either the exo or the endo pucker conformation. In the exo pucker the C4 atom points towards the opposite side of the carbonyl group of Pro, while in the endo pucker the C4 atom and the Pro carbonyl group are on the same side of the plane. The analysis of a set of high resolution crystal structures of proteins revealed that 89 % of Pro residues involved in a cis peptide bond displays the endo pucker, while there is no defined preference for one conformation over the other when Pro is found in a trans peptide bond.

Substitutions on the pyrrolidine ring with electron‐withdrawing or sterically hindered chemical groups can have an influence both on the puckering preference and on the cis/trans equilibrium. In the context of globular proteins, (2S; 4R)‐fluoroproline ((4R)‐Flp) and (2S; 4S)‐fluoroproline ((4S)‐Flp) have been employed for the fine tuning of protein thermodynamic stability, folding, and bioactivity. Increase in the thermodynamic stability of peptides and proteins substituted with 4‐fluoroprolines have been commonly attributed to the biasing of the ring pucker elicited by the stereochemistry of the electron‐withdrawing fluorine accompanied by favourable pre‐organization effects on the main chain. (4R)‐Flp has been shown to promote the exo pucker and to stabilise the trans peptide bond conformation, while (4S)‐Flp favours the endo pucker and increases the stability of the cis‐peptide bond. In addition, the presence of an electron‐withdrawing element at this position decreases the double bond character of the peptide bond, thus facilitating the cis/trans isomerization process by diminishing the rotational barrier between the two conformations.

The impact of the incorporation of (2S)‐4,4‐difluoroproline (Dfp) on the thermodynamic stability and folding kinetics of globular proteins has been by far less investigated. Similar to Pro, Dfp does not seem to have a strong conformational preference, but due to the presence of two electron‐withdrawing fluorine atoms it can enhance the cis/trans isomerization kinetics to a larger extent than (4R)‐ and (4S)‐Flp, by diminishing the double bond character of the peptide bond. Therefore, it has been suggested that Dfp might prove useful for enhancing the folding rates of proteins where cis/trans isomerisation around the prolyl peptide bond is the rate limiting step. Differently from hydrocarbons in Pro, the fluorocarbon bonds on Dfp display an inverted polarity at the C4, thus creating a negatively charged surface. Moreover, Dfp is slightly more lipophilic than Pro. Therefore, Dfp might be also employed to modulate hydrophobic effects in proteins without altering the native conformational preference of the protein.

In previous studies, we have incorporated (4R)‐ and (4S)‐Flp into an E. coli thioredoxin variant which contains cisPro76 as the only proline residue (Trx1P), whereas the trans Pro residues at positions 34, 40, 64, 68 were mutated to Ala.[ , ] E. coli thioredoxin is regarded as the prototype of the family of thioredoxin‐like oxidoreductases. It displays a conserved cis prolyl peptide bond at Pro76 (essential for Trx bioactivity and thermodynamic stability) buried in the hydrophobic interior in close proximity to the catalytic disulfide bond between Cys32 and Cys35 (Figure 1). We have shown that the incorporation of (4S)‐Flp into Trx1P (Trx‐4S‐Flp) accelerated the formation of the native cis‐peptide bond in the context of the tertiary structure by nine‐fold relative to the parent protein. In the unfolded state, (4S)‐Flp also accelerated the trans‐to‐cis reaction by one order of magnitude. The incorporation of (4R)‐Flp at cis‐Pro76 into Trx1P (Trx‐4R‐Flp) elicited no changes in the refolding kinetics of the protein, while a two‐fold acceleration of the trans‐to‐cis reaction was reported in the unfolded state (Table S1 Supporting Information). In this work, we aimed at investigating the effect of Dfp, on the thermodynamic stability and bioactivity of Trx as well as on the kinetics of prolyl cis/trans isomerization in the context of the intact tertiary structure and in the unfolded state of Trx1P. We reasoned that incorporating a 4‐fluorinated Pro analogue devoid of a strong conformational preference would let us dissect the effects arising from the ring pucker preference and from the electron‐withdrawing effect of fluorine.

Figure 1

Three‐dimensional structure of E. coli thioredoxin (PDB: 2TRX). In the black square is shown the local structure around the catalytic disulfide bond with cisPro76 in close proximity (surface polarity is colour‐coded from negative charge (red) to positive charge (blue)).

Results and Discussion

The fluorinated protein variant Trx1Dfp, substituted at cisPro76, was produced in M9 minimal medium using the proline auxotrophic cell strain CAG18515 transformed with an additional plasmid for overexpression of an engineered E. coli prolyl‐tRNA‐synthetase, ProRS (C443G), to increase Dfp incorporation. After purification by ion exchange chromatography, the yields for the fluorinated protein variant were 5.4 mg/L of bacterial culture. The identity of the isolated protein and the incorporation yields for Dfp (93 %) were confirmed by mass spectrometry (Supporting Information, Figure S1).

At first, we determined the thermodynamic stability of the oxidised and reduced forms of Trx1Dfp by measuring the unfolding/refolding equilibrium transitions at different guanidinium chloride (GdmCl) concentrations. The free energy of folding (ΔG0) for the oxidized and reduced forms of Trx1P were −38.0±1.3 kJ/mol and −29.2±1.7 kJ/mol respectively and were in agreement within experimental error with data previously reported by our group under identical conditions. The thermodynamic stability of the oxidised form of the Trx1Dfp variant (Trx1Dfpox) was significantly decreased (−29.1±1.2 kJ/mol) in comparison to the parent protein, while the ΔG0 of the reduced form (Trx1Dfpred) was almost identical to that of Trx1P (Figure 2 and Table 1). Therefore, Trx1Dfpox and Trx1Dfpred displayed roughly the same thermodynamic stability. This is surprising because Trx is an oxidoreductase acting as a strong cytoplasmic reductant and its oxidized form in nature is more stable than the reduced one. The less negative ΔG0 of Trx1Dfpox might arise from the presence of the two electronegative substituents on the C4 on the Pro ring, as the pKa value of Cys32 might decrease, thus leading to a destabilisation of the oxidised form of Trx1Dfp at pH 7.0.

Figure 2

GdmCl‐dependent unfolding‐refolding equilibria of the oxidised (blue) and reduced (red) forms of Trx1P and Trx1Dfp in 50 mM MOPS⋅NaOH at pH 7.0 and 25 °C, determined by following the signal change at 220 nm by CD. The unfolding (closed symbols) and refolding (open symbols) equilibria were evaluated according to the two‐state model of folding and normalized (solid lines).

Table 1

Thermodynamic parameters for the Trx variants determined by unfolding/refolding equilibrium transitions in GdmCl (pH 7.0, 25 °C).

	ΔG0 ox [kJ/mol]	ΔΔG0 ox/red [kJ/mol]	meq [kJ/(mol M)]	Transition midpoint (M GdmCl)
Trx1Pox	−38.0+1.3	8.8	14.2+0.9	2.68
Trx1Pred	−29.2+1.7		11.7+0.4	2.55
Trx1Dfpox	−29.1+.1.2	−0.7	10.1+0.6	2.88
Trx1Dfpred	−29.8+1.8		10.3+0.4	2.89
Trx2Pox	−36.7+1.7	6.6	14.2+1.1	2.58
Trx2Pred	−30.1+1.2		15.2+1.2	1.98
Trx2Dfpox	−30.0+1.1	0.9	11.4+1.2	2.63
Trx2Dfpred	−29.1+1.3		13.0+1.0	2.24

[a] A negative value of ΔΔG0 ox/red means that the reduced form of the Trx variant is more stable than the oxidised form.

Next, we determined the redox potential of Trx1Dfp, calculated from the deduced equilibrium constants with wild‐type Trx (E0’=‐270 mV) as previously described (Equations 1 and 2 and Figure S2 in the Supporting Information). The calculated redox potential for Trx1Dfp (E0’=‐234 mV) was almost identical to that of the parent protein. Thus, the loss in thermodynamic stability of Trx1Dfpox did not translate into a more oxidising redox potential.

In order to verify if the destabilising effect of Dfp on the oxidised form of the Trx variant is position dependent, we incorporated a second Dfp at Pro34 into an E. coli Trx variant (Trx2P) containing the Pro residues cisPro76 and transPro34. Pro34 is located between the two catalytic cysteines at the N‐terminal end of α helix 1 and it is solvent exposed, while Pro76 is completely buried inside the tertiary structure (Figure 1).

The yields for the purified protein Trx2Dfp were 3.4 mg/L of bacterial culture and the incorporation yields for Dfp at positions 34 and 76 were ∼89 % (Supporting Information, Figure S3). The free energy of folding for Trx2Dfp, determined from the unfolding/refolding equilibrium transitions, showed that no additional destabilisation occurred upon incorporation of the second Dfp residue (Figure 2 and Table 1). In fact, Trx2Dfpox was 6.7 kJ/mol less stable than the parent protein Trx2Pox, while both proteins displayed a similar thermodynamic stability for the reduced form.

Next, we investigated the ability of thioredoxin reductase (TrxR) to recognize the Trx variants. In the cellular environment, Trx receives reducing equivalents from NADPH through TrxR, which is the only protein capable of keeping Trx in its active reduced state. E. coli TrxR activity was determined by monitoring the increase in absorbance at 412 nm that follows the reduction of 5,5‐dithio‐bis‐(2‐nitrobenzoic acid) (DTNB) by reduced Trx. The initial velocities were plotted against different substrate (Trx) concentrations and evaluated according to Michaelis‐Menten kinetics (Figure 3, left). The KM values for Trx1P and Trx2P were 3.0 μM and 3.5 μM respectively.

Figure 3

Functional properties of Trx1Dfp and Trx2Dfp compared to their parent proteins Trx1P and Trx2P. Left: Kinetic analyses of Trx variants as substrates of thioredoxin reductase (TrxR) in 50 mM MOPS⋅NaOH at pH 7.0 and 25 °C. TrxR activity was determined by monitoring the change in absorbance at 412 nm during the initial 60 seconds of the reaction. Initial velocities were plotted against Trx concentrations and data fitted according to Michaelis‐Menten kinetics (solid lines). Right: Reactivities of the active‐site cysteine pairs in Trx1Dfp and Trx2Dfp compared to Trx1P and Trx2P determined by following the increase in fluorescence at 350 nm (λex=280 nm) that accompanies the reduction of oxidized Trx variants (2.0 μM) by DTTred (20 nM) at pH 7.0 and 25 °C) (solid lines). Data are an average of three independent measurements.

The Pro76 to Dfp replacement in Trx1Dfp increased the KM value by 3.1 fold (KM=9.5 μM) in comparison to the parent protein, while the additional incorporation of Dfp at Pro34 in Trx2Dfp resulted in a 3.7 fold increase of the KM value (KM=13 μM) in respect to the parent protein Trx2P. To determine whether Dfp has an effect on the catalytic activities of the Trx variants, we assessed their ability to act as a reductant of the disulfide bonds of bovine insulin by reduced DTT at different catalyst (Trx) concentrations. The reactions were monitored by following the increase in optical density at 650 nm resulting from aggregation of the reduced insulin B chain (Figure S4 in the Supporting Information). The specific insulin reductase activity of all Trx variants was in the range 4.43 ⋅ 10−2–5.68 ⋅ 10−2 min−1 μM−1 and was comparable to that of wild‐type Trx (5.12 ⋅ 10−2 min−1 μM−1). The fluorinated protein variants were ∼1.2 fold less active than their parent proteins, thus it can be concluded that Dfp has a negligible effect on the catalytic activity of Trx. The reactivity of the disulfide bonds of all variants assessed by the increase in the fluorescence emission at 350 nm (λex=280 nm) upon reduction of the catalytic disulfides with DTT, showed a 1.5 fold higher reactivity for the fluorinated variants in respect to their parent proteins (Figure 3, right). Therefore, Dfp proved to have only a slightly negative impact on the interaction between Trx and TrxR, but did not affect the catalytic activity or the reactivity of the active‐site disulfide bonds of the fluorinated variants. It is possible that Dfp causes slight changes in the local tertiary structure, in particular at position 76 in the loop facing the active site, thus making it a slightly worse substrate for TrxR.

Next we investigated the influence of Dfp on the kinetics of proline cis/trans isomerization in the context of the intact tertiary structure and in the unfolded state of Trx1Dfp. We have previously shown that in the unfolded state, 95 % of Trx1P molecules present a non‐native trans peptide bond at Pro76 (U ), while 5 % display the native cis peptide bond (U ). The U molecules fold very fast to the native state (N ), while the 95 % U collapse into a long‐lived intermediate (I ) that displays an intact tertiary structure with a buried non‐native trans Pro76. The very slow conversion (t1/2=∼100 min) of I into N is the bottleneck of the folding pathway of Trx1P (Scheme 1).

Scheme 1

Simplified refolding pathway of Trx1P (adapted from Roderer et al.).

We performed interrupted refolding experiments (N‐tests) to determine both the percentage of U and U molecules in the unfolded state and the folding kinetics of the I ‐to‐N reaction. Briefly, Trx1P and Trx1Dfp were completely unfolded in GdmCl (4.0 M) at 25 °C and pH 7.0 to reach U /U equilibrium. The proteins were then let refold at 25 °C by rapid dilution (1 : 20) with MOPS⋅NaOH (50 mM, pH 7.0).

Finally, at different refolding times, protein samples were unfolded again at 25 °C by addition of GdmCl (Trx1P: 3.4 M final concentration, Trx1Dfp: 3.2 M final concentration). This final unfolding step was monitored by following the change in CD signal intensity at 220 nm and the change in the amplitude was plotted against the refolding time (Figure 4). The rates obtained from the N‐tests for the I ‐to‐N reaction for Trx1Dfp and the parent protein were almost identical, giving a half‐time of 147±21 min and 124±15 min respectively (Figure 4 and Table 2). The half‐times were also confirmed by fluorescence spectroscopy within the experimental error (Trx1Dfp t1/2=143±15 min; Trx1P t1/2=128±13 min). In fact, the I ‐to‐N folding reaction can be monitored by following the decrease in the fluorescence emission at 345 nm (λex=280 nm) when I is converted to N (Figure S5 in the Supporting Information). Our results showed that in the context of the tertiary structure of the Trx1P variant, Dfp behaves as (4R)‐Flp and Pro and has no effect on the kinetics of the trans to cis isomerisation reaction.

Figure 4

Kinetics of formation of N from I during refolding of Trx1P and Trx1Dfp, recorded by interrupted refolding experiments (N‐tests). After complete unfolding in 4.0 M GdmCl, at 25 °C and pH 7.0 the proteins were then let refold at 25 °C by rapid dilution (1 : 20) with MOPS⋅NaOH (50 mM, pH 7.0). After different refolding times, the protein samples were unfolded again at 25 °C by addition of GdmCl. This last unfolding step was monitored by following the change in CD signal intensity at 220 nm and the change in the amplitude was plotted against the refolding time. Data are an average of three independent measurements.

Table 2

Kinetic folding parameters of Trx1P and Trx1Dfp.

	Trx1P	Trx1Dfp
trans ‐to‐cis isomerization in the context of tertiary structure
k of I trans →N cis [s−1][a]	9.28+0.5×10−5	7.86+0.7×10−5
t1/2 of I trans →N cis (min)[a]	124+15	147+21
Ile75‐Pro76 cis/trans equilibrium in the unfolded state
K cis/trans [b]	0.038+0.009	0.049+0.007
k app [s−1][c]	2.38+0.48×10−2	5.32+0.49×10−2
k trans to cis [s−1][d]	9.04+0.45×10−4	2.61+0.21×10−3
k cis to trans [s−1][d]	2.29+0.59×10−2	5.06+0.81×10−2

[a] Values determined by N‐tests (Figure 4). [b] Calculated from the fraction of fast folders (Figure 4). [c] Values determined by U‐tests (Figure 5). [d] Calculated from k and K .

‐to‐

k of I →N [s−1][a]

t1/2 of I →N (min)[a]

Ile75‐Pro76 equilibrium in the unfolded state

K [b]

k [s−1][c]

k [s−1][d]

k [s−1][d]

Figure 5

Kinetics of the attainment of the cis/trans equilibrium of the Ile75‐Pro76 (Trx1P) and Ile75‐Dfp76 (Trx1Dfp) measured by interrupted unfolding experiments (U‐tests). Trx variants were completely unfolded in GdmCl and after different incubation times, proteins were refolded by 50‐fold dilution with MOPS⋅NaOH. Refolding was measured by following the decrease in tryptophan fluorescence. The amplitudes of the slow refolding phases (proportional to the fraction of Utrans molecules with a trans 75–76 peptide bond in the unfolded state prior to refolding) were plotted against unfolding time. Data are an average of three independent measurements.

The trans to cis isomerisation around Tyr47‐Pro48 in the context of a structured intermediate with Pro48 in trans is also the rate‐limiting step in the in vitro refolding of the barstar protein. Upon incorporation of Dfp into the pseudo wild‐type barstar, containing cisPro48 as single Pro residue, a 3.6 fold increase of the I ‐to‐N folding rate was observed. A similar acceleration (3.5 fold) was reported upon incorporation of (4S)‐Flp, while (4R)‐Flp at cisPro48 induced no changes in the refolding kinetics relative to the parent protein. As previously mentioned, the I ‐to‐N isomerisation around Ile75‐Pro76 in Trx1P occurs in the context of a fully intact tertiary structure with Pro76 completely buried in the protein interior. In barstar, Pro48 is located at the surface and is largely solvent exposed. These results might suggest that differences in the local polar interactions created by fluorocarbons in the protein core could affect the refolding rates, as these can be influenced by environmental factors. Another plausible explanation might involve the pucker preference of the fluorinated prolines. Bioinformatics analysis has shown that in the thioredoxin fold, there is an almost unique preference for the endo pucker of the conserved cisPro in the context of the tertiary structure. In fact, cisPro76 in Trx1P also adopts the endo pucker and a well‐defined endo pucker was maintained upon incorporation of (4S)‐Flp, while this was less pronounced upon incorporation of (4R)‐Flp which favours the exo conformation. Thus, we could speculate that no acceleration of the isomerisation rate in Trx1Dfp and Trx‐(4R)‐Flp relative to the parent protein is because neither Dfp nor (4R)‐Flp favour the adoption of the endo pucker. It should be mentioned that when Dfp was incorporated at cisPro32 in human β2‐microglobulin, the trans to cis prolyl bond isomerisation was no longer the rate limiting step for the folding of this protein. The authors suggested that this behaviour can be attributed to the strong electron‐withdrawing effect of the two fluorine atoms on the Pro ring, as Dfp displays the lowest energy barrier for cis/trans isomerization in comparison to Pro and its 4‐monofluorinated analogues. Our results show that this explanation does not apply in the context of the thioredoxin fold. It could be speculated that as Pro 32, which displays an endo pucker in the crystal structure of β2‐microglobulin, is located in a flexible loop, the biasing of the pucker might play a minor role. Another possible reason that might explain why Dfp did not accelerate the trans to cis isomerisation rate in Trx1P is the presence of additional energetic costs associated with the protein molecule adopting the trans‐amide conformation at position 75–76 in the long‐lived intermediate. It might be speculated that the pucker exerts an influence on the stability of the intermediate itself. In this context, the energetic costs for reaching the native state become identical for Pro and Dfp containing variants.

Next, we investigated the attainment of the cis/trans equilibrium of the 75–76 peptide bond in the unfolded state of Trx1Dfp by interrupted unfolding experiments (U‐tests).[ , ] Briefly, the native protein was unfolded in 5.0 M GdmCl and after different incubation times was then refolded at 25 °C by rapid dilution (1 : 50) with MOPS⋅NaOH (50 mM, pH 7.0). The refolding reactions were monitored following the decrease in fluorescence at 345 nm (λ ex=280 nm) and the refolding amplitudes were plotted against unfolding time and yielded the apparent rate constant (k app =k +k ) of the attainment of the U U equilibrium (Figure 5, Table 2). The trans to cis reaction (k ) for Trx1Dfp was accelerated almost 3‐fold in comparison to Trx1P while the rate constant for the cis to trans reaction (k ) was ∼2‐fold increased. The increase of both microscopic rate constants k and k for Trx1Dfp in the unfolded state is slightly lower than the acceleration observed for the model peptides NAc‐Dfp‐OMe and Suc‐Ala‐Ser‐Dfp‐Phe‐pNA in comparison to their correspondent non‐fluorinated peptide (3.5‐5 fold acceleration). The amount of U molecules for Trx1P determined from the intercepts with the y‐axis (N‐tests; Figure 3) was 3.7±0.9 % and was in accordance with our previous studies within the experimental error.[ , ] Trx1Dfp displayed a similar population (4.7±0.7 %) of U molecules. Therefore, the Pro to Dfp replacement in Trx1P did not significantly change the K equilibrium (Table 2). The very modest increase in U molecules in the unfolded state is in agreement with data reported about the population of cis conformers in the unfolded state of model peptides where Pro is replaced by Dfp. We have previously shown that (4S)‐Flp at position 76 in Trx‐(4S)‐Flp accelerated the rate for the trans to cis isomerisation reaction in the unfolded state by one order of magnitude, while the population of U molecules was increased to 14 %. Similarly to the results obtained for Trx1Dfp, the incorporation of (4R)‐Flp into Trx‐(4R)‐Flp led to a 2‐fold increase of the isomerization rate constant k and to a slight increase of U molecules in the unfolded state.

Conclusion

In conclusion, we have shown that the incorporation of Dfp into the thioredoxin variants Trx1P and Trx2P causes a significant destabilisation of the oxidised form of the fluorinated proteins. The destabilisation was position dependent, suggesting that the substitution at cisPro76 is less well tolerated, likely due to local polarity effects in the hydrophobic protein interior. Further, we proved that the incorporation of Dfp had only minor deleterious effects on the recognition of the fluorinated protein variants Trx1Dfp and Trx2Dfp by TrxR, while the catalytic properties of the variants remained unaffected. Finally, we showed that the replacement of cisPro76 with Dfp did not eliminate the bottleneck of the thioredoxin refolding reaction nor had an impact on the attainment of the cis/trans equilibrium in the unfolded state. Our results suggest that the pucker effect in the context of tertiary structures might prevail over cis/trans isomerisation rates measured for the free proline analogues or in the context of short model peptides.

Experimental Section

Equilibrium unfolding and refolding transitions

For the unfolding equilibrium transitions, the various Trx variants were incubated at different GdmCl concentrations overnight at 25 °C, in 50 mM MOPS⋅NaOH, pH 7.0. The Trx concentration was 17 μM in all samples. Samples of reduced Trx variants also contained 5 mM DTT. For the refolding equilibrium, the Trx variants were first fully unfolded in 6.0 M GdmCl overnight at 25 °C. MOPS⋅NaOH, pH 7.0 was then added, containing different concentrations of GdmCl. These solutions were then incubated at 25 °C overnight, to allow the samples to attain equilibrium. The CD signal of each sample at 220 nm was recorded for 1 min and averaged, on a JASCO J‐810 Circular Dichroism Spectrophotometer. The average CD signals were then plotted against the GdmCl concentrations. The data were fitted according to the two state model of folding and normalized (Equation 1.

where mn and mu are the “slopes” of the pre‐ and post‐transition baselines; [D] is the denaturant concentration; S 0 and S 0 are the signals of native and unfolded protein at 0 M denaturant; ΔG0 is the free energy of folding at 0 M denaturant; and m is the linear dependence of ΔG0 on [D]. The values of mn and mu were set to 0 for the fitting.

Interrupted refolding experiments (N‐tests)

Trx variants were unfolded overnight in GdmCl (4.0 M) at 25 °C and pH 7.0 to attain U /U equilibrium, and subsequently refolded at 25 °C by rapid dilution (1 : 20) with MOPS⋅NaOH (50 mM, pH 7.0) to a final GdmCl concentration of 0.2 M. After different refolding times, samples were unfolded again at 25 °C by addition of GdmCl (Trx1P: 3.4 M final concentration, Trx1Dfp: 3.2 M final concentration). The final Trx concentration was 17 μM. Unfolding was monitored by the change in CD signal intensity at 220 nm on a JASCO J‐810 Circular Dichroism Spectrophotometer. The amplitudes of the unfolding traces, deduced from monoexponential fits with the initial and final signals as open parameters, were plotted against refolding time. Extrapolation of the mono‐exponential fit to zero refolding time yielded the respective fraction of U .

Interrupted unfolding experiments (U‐tests)

Native Trx variants (N , 200 μM) were rapidly unfolded in GdmCl (5.0 M) with MOPS⋅NaOH (50 mM, pH 7.0) at 25 °C, incubated for different times under these conditions, and then refolded at 25 °C by rapid dilution (1 : 50) with MOPS⋅NaOH (50 mM, pH 7.0) to a final GdmCl concentration of 0.1 M. The final Trx concentration was 4 μM. The refolding reactions were recorded as the decrease in fluorescence at 345 nm (λ ex=280 nm), on a Cary Eclipse Fluorescence Spectrophotometer. Refolding kinetics were fitted mono‐exponentially with initial and final signal intensity as open parameters. The refolding amplitudes were plotted against unfolding time and yielded the apparent rate constant (k app) of the attainment of the U /U equilibrium.

Trx variants as a substrate for bacterial TrxR

Experiments were carried out using a quartz cuvette and a Cary‐50 UV‐Visible spectrophotometer, at 25 °C, in 50 mM MOPS⋅NaOH, pH 7.0, 0.4 mM EDTA. Recombinant E. coli TrxR was incubated with DTNB. Solutions of various concentrations of the Trx variants were added to the TrxR/DTNB solution and the reaction was started by addition of NADPH. The final concentrations were as follows; TrxR 33 nM, Trx 0.1–300 μM, DTNB 3.3 mM and NADPH 0.8 mM. TrxR activity was determined by monitoring the change in absorbance at 412 nm during the initial 60 seconds of the reaction. Slopes were converted into initial velocities using the extinction coefficient ϵ412nm=28300 M−1 cm−1, corresponding to the formation of two TNB molecules per catalytic cycle. Initial velocities were plotted against Trx concentrations and data fitted according to Michaelis‐Menten kinetics.

Reduction Kinetics of Trx variants by DTT

Equilibrium fluorescence spectra of the Trx variants were recorded between 320 and 400 nm, with excitation at 280 nm, using a Cary Eclipse Fluorescence Spectrophotometer. The reduction of the oxidized form of the Trx variants was measured by addition of reduced DTT with a final concentration of 2 μM Trx and 20 nM DTT in 100 mM NaH2PO4‐NaOH, pH 7.0. The sample was excited at 280 nm and the increase in emission was recorded at 350 nm. The monoexponential fitting was used to calculate the rate of reduction.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.