LsfA reduces with high efficiency a wide range of hydroperoxides
Initially, we explored the reaction between reduced LsfA and peroxynitrite by the HRP competition assay (25). As expected, increasing amounts of LsfA inhibited compound I formation (Fig. 1A), which resulted in a rate constant of (0.8 ± 0.14) x107 M− 1.s− 1 (Table I). As a second independent approach, we assessed the initial rate of peroxynitrite decay at increasing LsfA concentrations (Fig. 1B) (26), yielding a rate constant of (1.9 ± 0.5) x107 M− 1.s− 1 at pH 7.4 and 25°C (Table I).
The reaction between reduced LsfA and H2O2 was also investigated by the HRP competition assay (25). Again, at a fixed HRP concentration, increasing amounts of LsfA inhibited HRP compound I generation (Fig. 1C). Furthermore, we characterized this reaction by assessing redox-dependent changes in the intrinsic fluorescence of LsfA. Two phases were detected: an initial rapid decay followed by a subsequent gradual fluorescence increase (Fig. 1D). The kobs values were determined by fitting the first phase to first exponential decays, which were dependent on H2O2 concentration (Fig. 1D, insert). These kobs values were attributed to the oxidation step, as previously outlined for other Prxs (27–31). Lastly, in a third independent approach, we employed a competition assay between H2O2 and peroxynitrite for LsfA (32, 33), (Fig. 1E). Of note, the rate constant for the reaction between reduced LsfA with H2O2 determined by three independent assays yielded values around 3 x 107 M− 1.s− 1 (Table I).
Regarding the second phase of the fluorimetric profile (Fig. 1F), kobs values were obtained by fitting the experimental data to exponential increases, which were dose-dependent at H2O2 concentrations higher than 50 µM. The linear fit of the last yielded a slope that corresponds to the rate constant of hyperoxidation (302 ± 7) M− 1s− 1 at pH 7.4 and 25°C (Fig. S1 and Table 1). Noteworthy, hyperoxidation of CP was confirmed by western blotting (Fig. S2). At this point, we do not understand the reasons why kobs were independent of H2O2 concentrations up to 50 µM (Fig. S1 and Supplementary text).
We also examined the oxidation of LsfA by organic hydroperoxides (tert-butyl hydroperoxide, t-BOOH). We explored this reaction by the competition between peroxynitrite and t-BOOH for LsfA (Fig. 1G) and by the fluorimetric assay, following an approach similar to the one described for the oxidation of LsfA by H2O2 (Fig. 1H). By both assays, the rate constants were in the 106 M− 1.s− 1 range, one order of magnitude slower than the oxidations of LsfA by H2O2 and peroxynitrite (Table I). The second-order rate constant for the hyperoxidation of LsfA by t-BOOH was also determined (Fig. 1I) and again confirmed by western blotting (Fig. S2).
Table I summarizes all the data regarding the oxidation/hyperoxidation of LsfA obtained herein. The koxidation/khyperoidation is around 105 for H2O2 and 104 for t-BOOH. For a matter of comparison, human Prx2, Prx3 and AhpE from Mycobacterium tuberculosis display a koxidation/khyperoidation ratio of approximately 103 (35, 36). Prx6 from Plasmodium falciparum (PfPrx6) has a koxidation/khyperoidation ratio of 105 and 103 for H2O2 and t-BOOH, respectively (29). Although data on kinetics of hyperoxidation of Prxs is still limited, these data indicate that Prx6 enzymes (mostly 1-Cys Prx) are more resistant to hyperoxidation than AhpC/Prx1 homologs (mostly typical 2-Cys Prx).
LsfA is specifically reduced by ascorbate
After establishing that LsfA is relatively promiscuous regarding the oxidizing substrates, we investigated its reductive pathway. Previously, we described that LsfA is reduced by ascorbate with a rate constant in the 103 M− 1 s− 1 range at pH 7.2 and 25°C (19). Here, we described the expression and purification of different thiol-based systems from P. aeruginosa to assess their capacity to reduce LsfA. Initially, we showed that the reconstituted thioredoxin-dependent system efficiently reduced 5,5′-dithiobis (2-nitrobenzoic acid) (DTNB), an artificial disulfide (Fig. S3). Then, using a standard coupled assay, we observed that thioredoxin reductase B2 (PA14_53290) together with thioredoxin (Trx (PA14_11340) or TrxA (PA14_69200) failed to support LsfA peroxidase activity (Fig. S4 A and B). Likewise, we did not observe LsfA peroxidase activity in the presence of glutathione alone or with monothiolic glutaredoxin (PA14_18650) (Fig. S4 C). Consequently, LsfA exhibited significant resistance to the reduction by distinct thiol-based systems (Fig. S4) and ascorbate remained as the only reducing substrate described so far (19). We cannot exclude, however, that other reductant might also support the peroxidase activity of LsfA.
Structural Characterization
We elucidated two crystallographic structures in reduced (CP-S−, PDBid = 6P0W) and sulfonic acid (CP-SO3H, PDBid = 7KUU) states at 2.4 and 2.0 Ǻ resolution, respectively (Fig. 2) that gave us structural insights on the LsfA substrate specificity. These structures represent the first described structures of Prxs from P. aeruginosa. The two LsfA structures appeared as antiparallel dimers (Fig. 2A), exhibiting the typical arrangement for proteins belonging to the Prx6 subgroup (type B dimer) and displaying the thioredoxin fold (Fig. 2B) (13). The overall shapes of the two structures are highly similar (Fig. 2A), differing only in the active site (Fig. 2C and D). In both cases, aside from the fully conserved catalytic triad (Thr42; CP; Arg122), His 37 is in close proximity to CP, which is present in the α helix 5, (Fig. 2C and D), a typical feature for Prx6 enzymes. The oligomerization of LsfA in the native, reduced, oxidized and hyperoxidized states is always dimeric, as determined by SAXS analyses and size exclusion chromatography (Fig.S5 and S6). The indirect fourier transform (IFT) modeling curves for all samples are similar indicating that the treatments did not lead to important changes in the protein structure (Fig. S6). The obtained P(r) curve revealed an overall maximum size of ~ 65Å, indicating a globular shape, corroborating with the radius of gyration and molecular weight (Table S1).
A search on the Protein Data Bank - PDB (https://www.rcsb.org/), using DALI server (http://ekhidna2.biocenter.helsinki.fi/dali/) and LsfA (6P0W) as a template (37), returned structures of nine different proteins, all belonging to the Prx6 subgroup. These Prx6 enzymes share high amino acid sequence similarity (Fig. S7), mainly at the PVCTTE motif (11).
As expected, the overall shapes of Prx6 enzymes are highly similar, all presenting a hydrophobic patch surrounding their active sites near their CP (Fig. S8). Additionally, the electrostatic surfaces around the active sites are distinct among the analyzed Prx6s in the reduced state (Fig. 3A). LsfA (6P0W), TkPrx (6IU0), PhPrx (3W6G) and ApPrx (3A5W) present a highly positively charged surface nearby the active sites, while HsPrx6 (5B6M) is only weakly positive around the active site. These structural differences may reflect distinct affinities for substrates and/or inhibitors.
An investigation into the LsfA active site revealed a close interaction between CP and His37 in both structures (6P0W and 7KUU). His37 is highly conserved among Prx6 members and its interaction with CP was highlighted before through the description of a hypervalent state (38). In Anabaena sp., the conserved His residue was described as presenting high mobility (39). Accordingly, in HsPrx6 the positions occupied by this His residue are dependent on the oxidative state (40). Comparing the available Prx6 structures in the reduced state, we observed that the side chain of His37 in LsfA (6P0W) was the only one positioned in a distinct configuration (Fig. 3B and C). In contrast, His 37 in the sulfonic acid structure of LsfA (7KUU) presented a similar configuration to His residues of the other Prx6 structures (Fig. 3C). Finally, it was recently proposed that this conserved His residue would protect CP from hyperoxidation (41), which is consistent with the high koxidation/khyperoidation described here for LsfA (Table I).
We also attempted to gain structural insights into the interaction between LsfA and ascorbate, as this is the only biological reducing substrate described so far. Hence, we performed unbiased computer simulations and observed that ascorbate docked into the LsfA active site (Fig. S9A), which is consistent with the positively charged surface nearby the active sites (Fig. 3A). A recurrent interaction was observed between Thr120 residue with ascorbate through hydrogen bonding. Hydrophobic interactions were also observed, involving Pro38, Thr42, Pro43, Val44, and the Arg122 (the catalytic arginine) residues (Fig. S9B and C).
Cellular assays
The roles of LsfA in the P. aeruginosa response to oxidative insults were investigated by assessing a ΔlsfA strain, whose lsfA gene was deleted (24). Following incubation with 3-ATZ (a catalase inhibitor), the ΔlsfA strain exhibited heightened sensitivity compared to the wild-type (WT) strain when exposed to paraquat, a compound that generates a flux of H2O2 through spontaneous superoxide dismutation (Fig. 4A). Additionally, the ΔlsfA was significantly more susceptible to bolus addition of H2O2 (Fig. 4B) as previously described (24). Moreover, the ΔlsfA strain appeared to be more sensitive to SIN-1 (a peroxynitrite generator) compared to the WT strain (Fig. 4A), although this result did not achieve statistical significance. Noteworthy, other bacterial peroxidases, such as Ohr (9), can reduce peroxynitrite with similar efficiency. Surprisingly, ΔlsfA exhibited increased resistance to t-BOOH (Fig. 4B). This unexpected result might be attributable to a compensatory effect, possibly involving the induction of one of the several peroxidases in P. aeruginosa. Compensatory effects among antioxidant systems are well-documented and can lead to mutant strains more resistant to oxidants than the wild-type counterparts (42–44).
Utilizing the Hyper7 probe (45), the H2O2 levels were monitored in real-time, without disrupting the cellular compartmentalization. Of note, this is the first study employing genetically encoded probes in P. aeruginosa. The basal level of the Hyper7 signal (A488/400 excitation ratio) was higher in the ΔlsfA (0.3) than in the wild type (0.2) strain, indicating that the mutant strain is in a constitutively more oxidative state (Fig. 4C and D, blue line). After H2O2 exposure, the Hyper7 signal was higher in the ΔlsfA strain in comparison with WT for all concentrations tested, consistent with the notion that LsfA outcompeted Hyper7 for H2O2 reduction in the wild-type strain (Fig. 4C and D). Additionally, the recovery of the Hyper7 fluorescence to the basal state was delayed in the ΔlsfA strain (Fig. 4C and D). Therefore, the cellular results presented so far indicated that LsfA plays a major role in the response of P. aeruginosa to H2O2.
Given the little information available regarding ascorbate metabolism in P. aeruginosa, we investigated the involvement of ascorbate in the P. aeruginosa response to H2O2. As E. coli and Vibrio cholerae can use ascorbate as a sole carbon source (46, 47), we explored whether the same phenomenon could occur in P. aeruginosa. However, at high ascorbate concentrations, we observed only a slight bacterial growth when this molecule was used as the sole carbon source (Fig. S10). Subsequently, we verified that ascorbate (up to 10 mM) did not impair bacterial growth in a M9 media containing glucose (Fig. S11).
Finally, we investigated whether ascorbate could support the peroxidase activity of LsfA in cellular systems, taking advantage of Hyper7 (45). In this regard, redox-dependent changes in Hyper7 fluorescence were monitored in WT (Fig. 5A and C) or ΔlsfA (Fig. 5B and D), and the cells were exposed to 4mM H2O2 (first dotted line) followed by ascorbate (second dotted line). Remarkably, ascorbate accelerated the recovery of Hyper7 fluorescence only in WT (Fig. 5A and C) but not in ΔlsfA (Fig. 5B and D) cells. This data supports the notion that P. aeruginosa can internalize ascorbate and use it to support LsfA peroxidase activity.