Catalytic Thr or Ser Residue Modulates Structural Switches in 2-Cys Peroxiredoxin by Distinct Mechanisms

Typical 2-Cys Peroxiredoxins (2-Cys Prxs) reduce hydroperoxides with extraordinary rates due to an active site composed of a catalytic triad, containing a peroxidatic cysteine (CP), an Arg, and a Thr (or Ser). 2-Cys Prx are involved in processes such as cancer; neurodegeneration and host-pathogen interactions. During catalysis, 2-Cys Prxs switch between decamers and dimers. Analysis of 2-Cys Prx structures in the fully folded (but not locally unfolded) form revealed a highly conserved, non-conventional hydrogen bond (CH-π) between the catalytic triad Thr of a dimer with an aromatic residue of an adjacent dimer. In contrast, structures of 2-Cys Prxs with a Ser in place of the Thr do not display this CH-π bond. Chromatographic and structural data indicate that the Thr (but not Ser) destabilizes the decamer structure in the oxidized state probably through steric hindrance. As a general trend, mutations in a yeast 2-Cys Prx (Tsa1) favoring the dimeric state also displayed a decreased catalytic activity. Remarkably, yeast naturally contains Thr-Ser variants (Tsa1 and Tsa2, respectively) with distinct oligomeric stabilities in their disulfide states.

Typical 2-Cys peroxiredoxins (2-Cys Prx) from the Prx1/AhpC sub-group comprise a large class of thiol-specific antioxidant enzymes capable of reducing hydroperoxides with high efficiency (k = 10 6 -10 8 M −1 s −1 ) and specificity [1][2][3][4][5] . The reactivity of 2-Cys Prx toward peroxides is considered extraordinary as they react one to ten million times faster than free Cys 6 . Due to their catalytic properties these enzymes are considered H 2 O 2 sensors, which is probably related with the association of 2-Cys Prx with processes such as tumor suppression, neuronal differentiation and cardiovascular disease (reviewed in refs 7,8). 2-Cys Prxs make use of a fully conserved Cys residue, known as the peroxidatic cysteine (C P ), for the reduction of hydroperoxides, which results in the formation of a sulfenic acid (C P -SOH) (Fig. 1a,b). The basic unit of 2-Cys Prxs is a homodimer (α 2 ) and during the catalytic cycle a second Cys residue, the so-called resolving cysteine (C R ) reacts with C P -SOH of the adjacent monomer (Fig. 1b,c) to form an intermolecular disulfide bond (reviewed in refs 8 and 9). A partial unfolding of the α -helix containing C P is required for the intermolecular disulfide formation and thioredoxin (Trx) is commonly the electron donor to reduce the disulfide [7][8][9] (Fig. 1a-c). Therefore, during catalysis, 2-Cys Prx enzymes assume two structural states: fully folded (FF) and locally unfolded (LU). Alternatively, C P -SOH can further react with additional hydroperoxide molecules giving rise to hyperoxidized states such as sulfinic (C P -SO 2 H − ) and sulfonic (C P -SO 3 H − ) acids ( Fig. 1b-d). These hyperoxidized forms cannot be reduced by Trx, however in eukaryotic organisms C P -SO 2 H can be reduced back to C P -SOH by an ATP dependent sulfiredoxin 8 . (a) C P in the fully folded (FF) form (S − p ) displays high reactivity towards hydroperoxides and stabilizes the decamer state. This quaternary structure corresponds to a strong decamer. (b) The oxidation product (sulfenic acid -S p OH), also a strong decamer, and can have two outcomes: condensation or hyperoxidation. (c) The intermolecular disulfide is formed after a local unfolding process, because the two Cys residues are far away in the FF state, which destabilizes the decamer. In this work, we show that the extra Cγ in the side chain of catalytic Thr (Thr 44 in Tsa1) contributes to decamer destabilization. The reduced form of typical 2-Cys Prx, which favors the strong decamer, is regenerated by Trx. The disulfide form of 2-Cys Prx can switch from a weak metastable decamer (c) to dimers (e). Alternatively, the sulfenic acid of 2-Cys Prx (b) can be hyperoxidized (d) by further oxidation of C P to S p O 2 H by a second molecule of hydroperoxide. This process can be reversed by Srx, with ATP consumption.
Bank. The analysis of the yeast Tsa1 crystallographic structure 31 in the reduced form revealed that (α 2 ) 5 decamer is stabilized by several non-covalent interactions at the dimer-dimer interface (Fig. 2a-c). Among these interactions, the Cγ atom of Thr44 acts as H-bond donor to the π -system of Tyr77 in the adjacent dimer (Fig. 2d). This CH-π bond is a non-conventional H bond that plays important contributions in protein structure and function (reviewed in refs 32,33). In the Tsa1 decameric structure, all five dimer-dimer interfaces have two CH-π H bonds between Thr44 and Tyr77 residues (Fig. 2e), and involves residues that are part of the so-called region I (Thr44) and region II (Tyr77), which are complementary at the dimer-dimer interface 21 .
Remarkably, the CH-π H bonds between the catalytic Thr and the Tyr of the adjacent dimer are fully conserved among all analyzed structures of 2-Cys Prx (Prx1/AhpC) in the FF state (See Supplementary Fig. 1 and Fig. 2d and Supplementary Table 1). In some cases, the Tyr is substituted by a Phe residue, which can also accept H bonds through electrons of its π -system (See Supplementary Fig. 1 and Supplementary Fig. 2a,c,d).
In the LU state, the orientation between Thr44 and Tyr77 residues changes, and consequently the CH-π H bond is disrupted (Fig. 2e,f; See Supplementary Fig. 2b). As the microenvironment around the active site Thr is highly crowded (See Supplementary Fig. 3a,b,d-g), structural rearrangements of the neighboring residues during FF → LU transition are likely required to allow Thr44 to swing (Fig. 2e,f), probably provoking disruption of other non-covalent interactions at the dimer-dimer interface.
In contrast, the microenvironment around the active site Ser (equivalent to Thr44 in Tsa1) in AhpC from Amphibacillus xylanus (See Supplementary Fig. 3c) and in Tsa2 from Saccharomyces cerevisiae (See Supplementary  Fig. 3h-k) is less crowded. Ser45 in the decameric structure of Tsa2 (which shares 86% of identity with Tsa1) appears as distinct rotamers in distinct chains, indicating high flexibility of this residue 34 . Remarkably, the CH-π hydrogen bond was not observed between Ser45 and Tyr78 (Tyr77 in Tsa1) in any of the Ser45 rotamers in Tsa2. Additionally, some of the chains from the Tsa2 decamer are in the FF state, while others are in the LU state. In fact, analysis of B factors in the α -helix containing the C P of the decameric Tsa1 and Tsa2 structures revealed that the thermal variation of Ser45 in Tsa2 is remarkably higher than the corresponding value for Thr44 in Tsa1 (Fig. 2g,h).
Therefore, it is possible that the conserved Thr/Ser not only participates in catalysis 10,12 , but is also involved in modulating the oligomeric state of the enzyme. For instance, while Thr44 Cγ atom of Tsa1 interacts with the neighboring dimer, the Oγ atom has a polar interaction with the Sγ atom of C P, which is further stabilized by polar contacts with the guanidine group of Arg123 (Fig. 2d,e; See Supplementary Fig. 2d). Furthermore, the catalytic triad of 2-Cys Prx is located near the dimer-dimer interface delimited by Tyr77 (Fig. 2d). Taken together, Thr44 provides a physical interaction between the catalytic center and the dimer-dimer interface, and indicates that the presence of Thr or Ser in the catalytic triad can influence the equilibrium among 2-Cys Prx quaternary structures.
Structural characterization of Tsa1 Thr44 mutants. Based on the structural considerations described above, we hypothesized that substitution of Thr44 in Tsa1 by other residues would affect decamer stability. Since the theoretical molecular weight of Tsa1 homodimers (~48 kDa) in solution is easily distinguishable from Tsa1 decamers (~240 kDa), we employed size exclusion chromatography (SEC) to evaluate the effects of Thr44 substitutions on Tsa1 quaternary structure. Under reducing conditions, Tsa1 WT , Tsa1 T44A , and Tsa1 T44S eluted as decamers ( Fig. 3a-c, solid lines), whereas Tsa1 T44V eluted as a mixture of dimers and decamers, with the dimeric forms being clearly more predominant (Fig. 3d, solid line). As expected, the oxidized (disulfide) samples favored dissociation of decamers 21 as compared with the reduced forms (Fig. 3, dashed lines). The exception was Tsa1 T44S , which migrated exclusively as decamers in both the oxidized and reduced forms (Fig. 3c). In contrast, oxidized Tsa1 T44V migrated mainly as dimer (Fig. 3d, dashed line). With intermediate behavior, comparable amounts of dimers and decamers of Tsa1 WT and Tsa1 T44A were detected in oxidizing conditions ( Fig. 3a,b, dashed lines). In summary, two of the Thr mutations provoked dramatic effects on the quaternary structure, with Tsa1 T44S favoring decamer stabilization and Tsa1 T44V favoring dimer stabilization.
Kinetic characterization of Tsa1 Thr44 mutants. Next, we examined the effects of the mutations on Tsa1 reactivity towards hydroperoxides using the Trx system coupled assay. Tsa1 WT and Tsa1 T44S presented catalytic efficiencies towards H 2 O 2 in the 10 4 M −1 s −1 range (See Supplementary Fig. 5a,c), similar to the previously obtained value 35 . On the other hand, Tsa1 T44A presented a slight decrease in activity (See Supplementary Fig. 5b), whereas the catalytic efficiency of Tsa1 T44V for H 2 O 2 decreased by two orders of magnitude and the K M value (520 μ M) increased by two orders of magnitude (See Supplementary Fig. 5d). All enzymatic parameters are summarized in Supplementary Table 2. The results for Tsa1 T44V are consistent with a previous study, which showed that the catalytic efficiency of human PrxV, with a Thr for Val mutation, also decreased the catalytic efficiency by a factor of one thousand 13 . All kinetic parameters were determined under conditions where Trx (at 2 μ M) was not limiting the catalysis (See Supplementary Fig. 5e).
In the case of cumene hydroperoxide (CHP) reduction, Tsa1 WT and Tsa1 T44S also presented similar enzymatic parameters (See Supplementary Fig. 5a-c, insets). In contrast to H 2 O 2 reduction, Tsa1 T44A presented a very high K M (300 μ M) for CHP and consequently a four-fold decrease in the catalytic efficiency (See Supplementary Fig. 5b, inset). In the case of the Tsa1 T44V mutant, an even more drastic drop in the reduction of CHP was observed (See Supplementary Fig. 5d, inset), precluding the determination of the kinetic parameters (Supplementary Table 2). Similarly, a 2-Cys Prx from L. donovani (TXNPx) also displayed negligible peroxidase activity towards organic peroxides when the homologous Thr residue was mutated to Val (TXNPx T49V ) 13 . These results indicate that Val mutations are not well tolerated at this position, and that eliminating the hydroxyl group from the side chain impairs the catalytic activity of 2-Cys Prxs. The factors underlying why organic peroxide reduction is more inhibited than H 2 O 2 reduction are not clear, and may be related to distinct effects of amino acid side chains on Scientific RepoRts | 6:33133 | DOI: 10.1038/srep33133 Cγ perform a CH-π bond with the Tyr77 of the adjacent dimer, while the Oγ of Thr44 side chain is involved in polar interactions with C P Sγ , which, in turn, is stabilized by the guanidine group of the Arg123. For (c,d), the carbon atoms are represented by colors similar to the polypeptide chain to which they belong, whereas N = blue, O = red and S = orange. For (e,f), the crystallographic structures of human Prx4 in the FF state ((e) 3TKP) and in the LU state ((f) 3TJB). The residues of catalytic triad and the conserved Tyr/Phe are represented by spheres and atoms are colored as follows: C = yellow, gray or light blue, according to the cartoon representation of the chain; N = blue; O = red; S = gold. (g) Wire representations of Thr/Ser region of yeast Tsa1 and Tsa2 (pdb codes 3SBC and 5DVB, respectively). The normalized B-factor was obtained by dividing the B-factor value of each Cα residue by the B-factor average value of all the Cα residues (normalized B-factor = B-factor of the residue/average B-factor of all residues). The wire thickness and color (blue = low to red = high) denotes the higher B factors for Tsa2 as compared to Tsa1. (h) Graphical detail of B-factor variation by amino acid in the portion containing the Thr or Ser from Tsa1 (blue curves) or Tsa2 (red curves), respectively. In the crystallographic structure of Tsa1, all chains are in the FF state while in the Tsa2 chains a mixture of FF, LU and intermediate states is found. Therefore, Tsa2 chains in FF state (C, D and G) and the respective chains of the Tsa1 were compared.
Scientific RepoRts | 6:33133 | DOI: 10.1038/srep33133 stabilization of the respective transition states with H 2 O 2 and CHP; and/or with the release of the corresponding leaving groups.
We also evaluated the effects of the Thr mutations by non-reducing SDS-PAGE. Reduced proteins were oxidized with three equivalents of H 2 O 2 or CHP and the disappearance of monomers was monitored. Among the three mutants, only Tsa1 T44V presented a significant delay in disulfide formation for both H 2 O 2 and CHP treatments (See Supplementary Fig. 6). For oxidations of 2-Cys Prx by H 2 O 2 , a similar trend in the reactivities as those observed in the Trx system coupled assay: Tsa1 WT ~ Tsa1 T44S > Tsa1 T44A ≫ Tsa1 T44V , taking into account that the SDS PAGE assay has a low resolution in time, being unable to discriminate moderate from very fast rates. For oxidation by CHP, similar tendency was also observed, but Tsa1 T44V was oxidized in seconds (See Supplementary Fig. 6), whereas in the Trx system coupled assay, no NADPH consumption was detected (See Supplementary Fig. 5d). These apparent inconsistences are possibly related with differences in the principles underlying these two assays.
In the case of the Trx system coupled assay, a complete turnover of the enzyme is required for NADPH consumption. Mutations affecting any of the three reactions (hydroperoxide reduction; condensation; disulfide reduction) or structural transitions (described in Fig. 1) might impair or even block the turnover of the enzyme. In contrast, only the oxidation and condensation reactions are measured in the non-reducing SDS-PAGE assay.
Determination of apparent second order rate constants for the reaction between Tsa1 proteins and hydroperoxides. The interpretation of the results obtained by the Trx system -coupled assay is not always straightforward. Among other reasons this is complicated by the facts that the overall assay requires three different enzymes, involving various consecutive thiol-disulfide exchange reactions (among 2-Cys Prx, Trx and Trx reductase 1) and that catalytic cycle of 2-Cys Prx enzymes itself is complex, comprising three reactions (mentioned above) and conformational changes (LU-FF transitions; dimer-decamer switches) 9 . Therefore, direct rate constants between hydroperoxides and reduced Tsa1 proteins were determined by taking advantage of the changes in their intrinsic fluorescence upon oxidation (Fig. 4a). Under pseudo first-order conditions (excess peroxide), a two phase kinetic profile was observed for Tsa1 WT (Fig. 4a,b). The first rapid phase is accompanied by a drop in fluorescence (first 5 ms), after an apparent 2 ms lag time, which coincides with the dead time of the instrument. The second phase is slower and is accompanied by a fluorescence increase.
Recently, a similar kinetic profile was described for a bacterial 2-Cys Prx (StAhpC) 36 . In order to fit the experimental data, the authors proposed a three phase kinetic model. In the first step, H 2 O 2 would reversibly bind to the StAhpC catalytic site, forming an enzyme -substrate complex, followed by C P peroxidation (which was associated to the fast fluorescence decay) and resolution (which was associated to second fluorescence increase phase). However, in the case of Tsa1 and under our experimental conditions the rise in fluorescence (second slower phase) could not be satisfactorily fitted and therefore was not further analyzed here. Therefore, Tsa1 and StAhpC display different fluorimetric properties, which might be due to distinct structural features of Trp residues (See Supplementary Fig. 7) or due to the fact that Tsa1 is sensitive to hyperoxidation, whereas StAhpC is resistant 37 .
Nevertheless, the rate of the initial rapid fluorescence decay increased linearly with peroxide concentrations (Fig. 4b,c), indicating that this phase is associated with C P oxidation. To test this assumption, we determined what would be the rate constant of C P oxidation under pseudo first-order conditions, fitting the fast decay in the fluorescence to a single first-order exponential equation or the whole biphasic time course to a two-exponential function (two consecutive first-order processes) for Tsa1 WT . Using either approach, the rate constant for the Tsa1 WT oxidation by H 2 O 2 was close to 5 × 10 7 M −1 s −1 (Fig. 4c). Overall, this fluorimetric approach yielded second order rate constants for the Tsa1 WT oxidation by H 2 O 2 and CHP in the 10 6 -10 7 M −1 s −1 range (Table 1), which agrees well with values previously determined by independent methods 3 . These convergences validated the fluorimetric method and made us confident to apply the same approach to the remaining Tsa1 mutants. Tsa1 T44S mutant was also oxidized by H 2 O 2 and CHP with extraordinary rates but the fluorimetric changes during the reaction was distinct from Tsa1 WT , showing only a fast fluorescence decay associated with C P oxidation (See Supplementary  Fig. 8). A decrease in reactivity was observed for both Tsa1 T44A and Tsa1 T44V enzymes (See Supplementary Fig. 8). Once again the Tsa1 T44V mutant was far less reactive (Table 1), while the Tsa1 T44A mutant was still quite active, one Enzymes were previously reduced with 20 mM DTT for 30 minutes at room temperature, and excess DTT was removed by gel filtration. Total intrinsic fluorescence decays upon oxidation by hydroperoxides were followed (λ ex = 280 nm; λ em > 310 nm using a cutoff filter) at 25 °C. The assays were performed in 40 mM phosphate buffer (pH 7.4). (c) The k obs values were obtained by fitting the stopped-flow data to single exponentials; second order rate constant was calculated from the slope of k obs plot versus oxidant concentration.
Scientific RepoRts | 6:33133 | DOI: 10.1038/srep33133 thousand fold faster than the free amino acid cysteine 6 , although at the same time, one thousand fold slower than the Tsa1 WT . On the other hand, the corresponding rate constant for the oxidation of Tsa1 T44V could not be determined due to its very low reactivity (Supplementary Fig. 8). The fluorimetric traces for Tsa1 T44V correlated quite well with the non-reducing SDS-PAGE assay (See Supplementary Fig. 6), but not with the Trx system coupled assay possibly by reasons related to the distinct principles of these three assays, as discussed above.
It is worth mentioning that the values obtained for catalytic efficiencies in steady state kinetics (k cat /K M ) were consistently in the range of 10 3 -10 4 M −1 s −1 , thus typically a thousand fold lower, than that obtained using fluorescence spectroscopy. This is probably due to other reactions and/or structural rearrangements limiting the overall turnover under steady-state conditions as mentioned above. It is worth mentioning that reduction of yeast TrxR1 by Trx1 and Trx2 are in order of 10 7 M −1 s −1 38 , reinforcing the hypothesis that structural rearrangements limit the reaction velocity. Nevertheless, the same trend was observed regardless of the approach (Trx coupled; non-reducing SDS-PAGE or fluorimetric assay) employed. In the case of Tsa1 T44V , this mutation decreased the ability of Tsa1 to reduce hydroperoxides by several orders of magnitude, whereas the Tsa1 T44A mutation induced a less pronounced effect. These results indicate the importance of the polar hydroxyl group in the catalytic mechanism.

Relationship between quaternary structures and catalytic activity of 2-Cys Prx.
The results presented so far indicated that distinct residues present at position 44 of Tsa1 play critical roles in both decamer stabilization and peroxidase activity. Yeast Tsa2 (86% of amino acid identity with Tsa1) was also analyzed, since this enzyme naturally presents a Ser in the position equivalent to Thr44 in Tsa1 3 . Similar to the Tsa1 T44S mutant (Fig. 3c), Tsa2 migrated exclusively as decamers independent of the enzyme redox state (Fig. 5a), further supporting the hypothesis that Ser in place of Thr stabilizes the decameric state of 2-Cys Prxs. In addition, Tsa2 reduced H 2 O 2 and CHP with extraordinary rates (Fig. 5b, Table 1; see Supplementary Fig. 9a,b).
As mentioned previously, the aromatic residue, Tyr77, also participates in the non-conventional CH-π H bond with Thr44. Therefore, the Tsa1 Y77A mutant was also produced, overexpressed, and purified. This mutant protein always eluted as dimers (Fig. 5c) and the peroxidase activity of Tsa1 Y77A was greatly reduced ( Fig. 5d; see Supplementary Fig. 9c,e). Tyr77 performs multiple interactions besides the non-conventional CH-π hydrogen bond with Thr44 in the dimer-dimer interface (See Supplementary Fig. 3d-g) and its replacement by alanine may disrupt various forces that stabilize the decamer. Substitution of Tyr77 residue that plays no role on transition state stabilization 11,12 , but that dramatically affected the ability of Tsa1 to reduce peroxides suggested that the decameric quaternary structure is important to activity. Furthermore, Tsa1 S78D mutant was also generated based on the information that in StAhpC and in HsPrx4, similar substitutions provoked decamer destabilization 4,25 . Thr77 in StAhpC (Ser78 in Tsa1) is located in the decameric interface of the enzyme but does not have direct interactions with any of the active site residues. Substitution of Thr77 by Asp in StAhpC precluded decamer formation 4 . As predicted, Tsa1 S78D also eluted as dimers, independent of the redox state (Fig. 5e) and the peroxidase activity was only residual in the Trx system coupled assay (Fig. 5f; see Supplementary Fig. 9d,f). Actually, the Tsa1 S78D mutation resulted in a more pronounced loss of peroxidase activity than what was reported for StAhpC T77D 4 .
As an independent and confirmatory assay, non-reducing SDS PAGE revealed that rates of disulfide formation in Tsa1 Y77A and Tsa1 S78D were extremely slow (See Supplementary Fig. 9g,h), precluding the determination of the rate constants values by available methods (Table 1). In contrast, Tsa2 displayed extraordinary second order rate constants for reduction of H 2 O 2 and CHP (Table 1).
Since Thr44 provides a physical interaction between the catalytic center and the dimer-dimer interface it is intuitive a relationship between activity and oligomerization. However, the correlation between these two processes is not perfect, probably reflecting the fact that other factors interferes with the peroxidase activity and also with decamer stabilization. Nevertheless, if we analyze the more pronounced effects in the oligomeric state, such as Tsa2/Tsa1 T44S (decamers with high activity) versus Tsa1 T44V /Tsa1 Y77A /Tsa1 S78D (dimers with residual activity) the interplay between activity and oligomerization is clear (Fig. 5). These results further indicate a positive trend between decamer structure and peroxidase activity, although other factors such as pH and peroxiredoxin concentration also affect the enzymatic activity of 2-Cys Prx 8,9,18-30 .

Discussion
The equilibrium between dimers and decamers for 2-Cys Prx is dependent on C P redox state among other factors 8,9,[18][19][20][21][22][23][24][25][26][27][28][29][30] , with oxidation to disulfides typically favoring dimers. Therefore, such a quaternary structure switch may be considered as part of the 2-Cys Prx catalytic cycle 9 (Fig. 1). Here, we present evidences that the presence of a Thr or Ser in the catalytic triad (Thr44 in Tsa1; Ser45 in Tsa2) plays a critical structural role in the stabilization or destabilization of 2-Cys Prx decamers upon disulfide formation ( Fig. 1; Supplementary Fig. 10). In proteins containing a Thr in the catalytic triad, oxidation leads to decamer dissociation into dimers 9 (Fig. 1c-e), while with proteins containing a Ser, at the analogous position, such dissociation is not observed. This is an interesting structural finding by itself and may indicate different non-redundant or complementary enzyme functions. Through a comparative structural analysis of 2-Cys Prx enzymes, it became clear that although most residues in the dimer-dimer interface retain similar conformation between the FF and LU states, the conserved Thr residue (Thr44 in Tsa1) swings considerably (Fig. 2e,f; See Supplementary Fig. 2a,b). This reorientation was observed in all of the analyzed decamer structures and is associated with the disruption of the Thr-Tyr (or Phe) CH-π bond in the LU state. Moreover, we observed that the decameric structure was prevalent (compare Ser44 versus Thr44 and Ala44 versus Val44 in Fig. 3) for protein containing residues with smaller side chains, consistent with a steric factor underlying the decamer stability. In fact, the microenvironment around the active site Thr is more crowded (See Supplementary Fig. 3) and significant structural rearrangements of neighboring residues are likely required to allow Thr44 to swing during the FF-LU transition. Possibly, the extra Cγ in the Thr side chain (in comparison with Ser side chain) may create steric effects with nearby residues, thus disfavoring the decameric state during the transition from the FF to LU states. In fact, a weakly associated decamer (in the LU state) has been proposed as an intermediate between the FF decamer and the LU dimer 21 (Fig. 1c) and other intermediate oligomeric structures have also been observed 14,19 . On the other hand, it is possible that smaller side chains of residues at position 44 (for example, the active site Ser in AhpC from A. xylanus and Tsa2 from S. cerevisiae, Supplementary Fig. 3) might allow for smoother FF-LU transitions, by avoiding major structural perturbations at the dimer-dimer interface.
Other mutations in the dimer-dimer interface (Thr77 residue of StAhpC and Thr118 in Prx4) were reported to provoke major changes in the stability of decamers 4,25 . However, crystallographic structures of these proteins did not reveal major alterations in the dimer-dimer interface. In most of these structures, the 2-Cys Prx enzymes are in the LU state with an intermolecular disulfide bond and with the active site Thr (Thr43 in StAhpC) also swung in comparison to the FF state. Our structural analysis of StAhpC and Prx4 mutants indicated that subtle modifications in the dimer-dimer interface may induce major changes in the dimer-decamer transitions at lower protein concentrations.
The presence of a hydroxyl group in the side chain is also important for the stabilization of decamers (compare Ser44 versus Ala44 and Thr44 versus Val44 in Fig. 3). This is likely related to the fact that the O atom in the side chain of Thr residues performs distinct polar interactions in the FF and LU states, in both cases possibly these interactions help to stabilize 2-Cys Prx in the decameric form (See Supplementary Fig. 11). In contrast, the few available structures of 2-Cys Prx in the dimeric form show that the active site Thr of Prx1 C83S from R. norvegicus possesses polar interactions distinct from the FF and LU decamers 20 . The hydroxyl group of the residue at position 44 is clearly important for the peroxidase activity (please compare the second order rate constants for Ser44 and Thr44 versus Ala44 and Val44, respectively in Table 1). Given the positive tendency between 2-Cys Prx decameric structure and activity, it is possible that Thr and certainly Ser residues stabilize the decamer, thereby allowing for optimal orientation of active site residues for peroxide reduction.
However, these associations between decamer and activity end after C P oxidation. For 2-Cys Prx containing Thr, this residue undergo a re-orientation during the FF-LU transition, which may disrupt the non-conventional CH-π hydrogen bond and perhaps various other neighboring interactions in the dimer-dimer interface, inducing decamer to dimer transition. In contrast, for 2-Cys Prx enzymes containing Ser, the FF-LU transition is smoother and the decameric oxidized structure is maintained throughout the catalytic cycle. It is tempting to speculate that the slow rise in fluorescence observed for the Tsa1 WT is associated with this decamer dissociation (Fig. 4a,b), since it was not observed for the Tsa1 T44S (See Supplementary Fig. 8) or for Tsa2 (Fig. 5b). If true, this would provide an opportunity to access the kinetics of decamer dissociation. However, at this moment, the mechanism behind of the slow rise in fluorescence during Tsa1 oxidation is still under investigation.
2-Cys Prx enzymes display complex structural and biochemical properties, involving distinct quaternary structures and oxidation states. Distinct post-translational modifications (hyperoxidation, glutathionylation and phosphorylation) in 2-Cys Prx also influence the quaternary structures they adopt and also their biological activities, indicating that mechanisms were positively selected for though evolution [26][27][28] . It is quite possible that these structural transitions among dimer, decamer and even higher molecular weight species relate to cellular functions, i.e. as peroxidase, chaperone, binding partner, enzyme activator and/or redox sensor (reviewed in ref. 18).
Our contribution here is to show that the active site Thr/Ser is also involved in modulating the stabilities of distinct 2-Cys Prx quaternary structures. The relationship between peroxidase activity and oligomerization is complex but disruption of dimer-dimer interactions in the reduced state is associated with a decreased catalytic efficiency with possible physiological significance as 2-Cys Prxs are considered the H 2 O 2 sensors 7,8,10 .
Protein expression and purification. Single colonies of E. coli BL21 (DE3) strain containing the pET-15b/tsa1, pET15b/tsa1 T44A , pET15b/tsa1 T44S , pET15b/tsa1 T44V , pET15b/tsa1 Y77A , pET15b/tsa1 S78D or pPROEx/tsa2 plasmids were inoculated in LB medium (20 ml) containing 0.1 mg ampicillin/mL and grown overnight at 37 °C with shaking at 250 rpm. The following day the cultures were transferred to 1 L of fresh LB medium, and cultured further until OD 600 reached 0.6-0.8. The overexpression of all the proteins was induced by the addition of 0.3 mM isopropyl-1-thio-β -D-galactopyranoside (IPTG) at 37 °C for 3hs with shaking at 250 rpm in an orbital shaker. Cells were harvested by centrifugation and cell pellets were resuspended in start buffer (50 mM sodium phosphate buffer, pH 7.4, 500 mM NaCl, 20 mM imidazole and 2 mM phenylmethyl sulfonyl fluoride -PMSF) and disrupted by sonication. The cell extracts were kept in ice during streptomycin sulfate (1%) treatment for 20 min. The supernatants were clarified by centrifugation, homogenized by filtration and purified by immobilized metal affinity chromatography (IMAC) using HisTrap column (GE Healthcare). Imidazole was removed by gel filtration using PD10 desalting column (GE Healthcare) and the purity of recombinant proteins was verified by SDS-PAGE.