ESI-MS Reveals Mechanism of O₂ Sensing.

To study the mechanism of O₂ sensing by FNR, optimized ionization conditions for detection of the monomeric rather than dimeric form of [4Fe-4S] FNR were used, so that intermediate species could be clearly and unambiguously identified (Fig. 2B). The addition of O₂ (under pseudofirst-order reaction conditions where the [O₂]:[4Fe-4S] ratio was ~10) resulted in the formation of a complex series of overlapping peaks (Fig. 2B). These peaks can be subdivided into two distinct groups corresponding to protein-bound cluster fragments (29,700–29,850 Da) and sulfur adducts of apo-FNR (29,550–29,700 Da). The cluster-bound fragments include the [2Fe-2S] end product of the cluster conversion reaction (at 29,720 Da) along with species with masses that correspond to [2Fe-3S] (29,752 Da) and [2Fe-4S] (29,784 Da) forms, which most likely represent singly and doubly persulfide-coordinated [2Fe-2S] clusters, respectively, as recently shown by resonance Raman (16) (3 and 4). In addition, a peak corresponding to a [3Fe-4S] form that is a well-characterized intermediate of the cluster conversion process (13, 14) was observed at 29,843 Da (Table 1). Each of these species has been previously observed in solution using various spectroscopic methods. The spectra also contained a peak at 29,811 Da consistent with the presence of a [3Fe-3S] cluster form (Table 1), which may represent a previously undetected intermediate in the conversion process, in which one of two sulfides that are eventually lost from the original [4Fe-4S] cluster has been ejected.

To investigate further these intermediates of cluster conversion, a ³⁴S-substituted form of [4Fe-4S] FNR was generated (16) The deconvoluted spectrum contained a major peak at 29,905 Da, +8 Da relative to the predicted mass of natural abundance [4Fe-4S] FNR (Fig. 3A). The predicted increase in mass on substitution of all sulfides is +7.6 Da (calculated from natural isotope abundance). Exposure of the ³⁴S-labeled [4Fe-4S] FNR sample to O₂ resulted in cluster conversion as described above, except that the peaks corresponding to the cluster breakdown products were mass shifted as follows: [3Fe-4S] by +8 Da, [2Fe-2S] by +4 Da, [2Fe-3S] by +6 Da, and [2Fe-4S] by +8 Da (Fig. 3 A and B). The isotope substitution data also showed that the sulfur adducts of apo-FNR observed after reaction of [4Fe-4S] FNR with O₂ are derived from the cluster (Fig. 3C) and that oxygen adducts of FNR are not formed (Fig. S2). Overall, the isotope substitution data provide unambiguous confirmation of the assignments for the cluster conversion intermediates and products.

Open in a separate window

Fig. 3.

Mass shifts observed for the FNR [4Fe-4S] cluster, conversion intermediates, and cluster products on ³⁴S substitution of cluster sulfides. (A) Deconvoluted mass spectra of natural abundance sulfur [4Fe-4S] S24F FNR and cluster conversion intermediates (black lines) and the equivalent ³⁴S-substituted forms (red lines) as indicated. Dotted lines represent Gaussian fits of the MS data. (B) The same as in A except that spectra represent cluster conversion products as indicated. (C) The same as in A except that spectra show the apo-FNR peak and the first persulfide adduct. Predicted mass shifts for the assigned species are indicated.

To investigate the kinetic behavior of these species, [4Fe-4S] FNR was combined with excess dissolved O₂ and continuously infused into the ESI source of the instrument. Fig. 2D shows the deconvoluted spectra obtained during the reaction time course over 120 min. Identical measurements of FNR that was not exposed to O₂ before infusion showed that the changes observed were entirely caused by O₂ exposure. At the earliest time points, a series of overlapping peaks identical in mass to those in Fig. 2B was observed. Peaks corresponding to the protein-bound cluster fragments (29,700–29,850 Da) increased in intensity, reaching a maximum ~15–30 min post-O₂ exposure. During this period, the [4Fe-4S] peak remained the most abundant species before decaying away along with the protein-bound cluster fragment peaks at later time points. In contrast, peaks corresponding to sulfur adducts of apo-FNR continued to rise to a maximum ~30–50 min postexposure (Fig. 2D).

Abundances of the different cluster fragments and apoprotein species in Fig. 2D were plotted as a function of time. In MS experiments, abundances are reported relative to the most abundant species (arbitrarily set to 100%). For FNR experiments, the most abundant species over most of the time course was the [4Fe-4S] cluster, and therefore, these data fail to provide any information on the rate of [4Fe-4S] cluster decay. To enable global analysis to be performed, absorbance spectroscopy was, therefore, used to measure (at 406 nm) the decay of the cluster under conditions identical to those used for MS experiments. The absorbance data were converted to relative abundance (with the starting point set to 100%) for direct comparison with the MS data. Global analysis of multiple (n = 4) mass spectrometric kinetic datasets leads to the reaction scheme shown in Fig. 4A. This scheme was able to model the formation and/or decay of the peak intensities corresponding to the previously characterized [4Fe-4S], [3Fe-4S], [2Fe-2S], and apo-FNR species (Fig. 4B), indicating that the [3Fe-4S] cluster is the first intermediate formed (maximizing before 20 min) followed by [2Fe-2S] and apo-FNR. The temporal nature of [3Fe-3S], which maximized at ~20 min (Fig. 4C), is consistent with an intermediate in the [4Fe-4S] to [2Fe-2S] cluster conversion pathway. However, both the [2Fe-4S] and [2Fe-3S] species (persulfide-coordinating forms of [2Fe-2S]) also reach a maximum at ~20 min before the [2Fe-2S] cluster (~30 min) (Fig. 4 D and E, respectively). The data show that [2Fe-3S] is not formed from [2Fe-2S] or the reverse; rather, [3Fe-3S] clusters can give rise to both [2Fe-3S] and [2Fe-2S] clusters, and therefore, this part of the reaction represents a branch point in an otherwise largely linear reaction mechanism. Similarly, the [2Fe-4S] cluster must form from [3Fe-4S] (and not [3Fe-3S]), and therefore, the formation of this species represents a second branch point (Fig. 4A).

Open in a separate window

Fig. 4.

Mechanism of [4Fe-4S] cluster conversion. (A) Reaction scheme used to simulate the kinetic dependence native MS data. (B) Plots of relative abundances of [4Fe-4S] cluster (A_{406 nm}; black squares), [3Fe-4S] (yellow triangles), [2Fe-2S] cluster (red diamonds), and apo (white circles) species as a function of time after exposure to excess O₂. Global fitting to the experimental data using the reaction scheme depicted in A is shown as solid lines. (C–E) Plots of relative abundances of the [3Fe-3S] cluster (yellow triangles in C), the [2Fe-4S] cluster (red diamonds in D), and the [2Fe-3S] cluster (red diamonds in E). Global fits to the experimental data are shown as solid lines. Dashed lines in C–E show the responses of the [4Fe-4S], [3Fe-4S], and [2Fe-2S] clusters for easy comparison. Error bars show SEs for average MS datasets (n = 4). The global fitting model was initiated with 100% relative abundance of [4Fe-4S] clusters.

The time dependence of intensity in the apoprotein region indicates that apo-FNR was converted to sulfane-containing species, apo(S)_n, with essentially full conversion at ~80 min (Fig. 4B). Indeed, the global fit was dependent on this being part of the mechanistic scheme (Fig. 4A). Similarly, the [2Fe-3S] and [2Fe-4S] clusters, which both contain persulfides, give rise to apo(S)_n species after the disassembly of their ligated [2Fe-2S] cluster as shown in Fig. 4A. An important feature of the data (both from MS and absorbance spectroscopy) is that the rate at which the apoprotein is formed from [2Fe-2S] in ammonium acetate buffer is greater than that previously reported for other buffer systems (13, 14). The poor stability of [2Fe-2S] species under these conditions means that they decay away in the MS spectrum rather than accumulating.

Overall, the model describes the behavior of the main [4Fe-4S], [3Fe-4S], and [2Fe-2S] cluster species, previously identified by spectroscopy, and the appearance of the [3Fe-3S], [2Fe-4S], [2Fe-3S], and apo-FNR species, all observed here simultaneously by MS (Fig. 4 B–E). We note that the model indicates that direct conversion of the [4Fe-4S] cluster to the [3Fe-3S] species does not occur. Thus, loss of a single iron seems to be an obligatory first step in the cluster conversion process, with loss of sulfide following that step.

Table 2 shows the observed rate constants required to describe the reaction scheme depicted in Fig. 4A. The first reaction, corresponding to the conversion of the [4Fe-4S] cluster into the [3Fe-4S] cluster, has an observed rate constant (k_obs1) of 0.02 min⁻¹ (0.003 s⁻¹), implying that it is rate limiting as previously observed (13, 14). Division of k_obs1 by the O₂ concentration (122 μM) used here provides an estimate of the apparent second-order rate constant, k = 25 M⁻¹ s⁻¹. This value is lower than the previously reported value of 80 M⁻¹ s⁻¹ for S24F FNR (28). It is known that the [4Fe-4S] to [2Fe-2S] cluster conversion is influenced by the nature of the buffer environment, in this case ammonium acetate.

Kinetics of Persulfide Adduct Formation Determined by LC-MS.

LC-MS, in which protein samples are in denaturing solvent, can be used to follow the formation of sulfane adducts during Fe-S cluster reactions, because these adducts are covalent species that survive protein unfolding (29). LC-MS was applied to samples of FNR equivalent to those used for native MS experiments. The reaction was quenched at specific time points to halt the reaction and inhibit any subsequent sulfur exchange, so that the time evolution of sulfane adducts could be correlated with the native MS data and used to monitor cluster sulfide oxidation. In the absence of O₂, the major peak at 29,547 Da corresponded to apo-FNR without sulfur adducts (Fig. 5). Shoulder peaks at +32, +64, and +96 Da are caused by the addition of one, two, and three sulfane sulfur atoms, respectively (29).

Open in a separate window

Fig. 5.

Tracking sulfide oxidation by LC-MS and native MS. (A) Deconvoluted LC-MS spectra before (black line) and after (red line) exposure to excess O₂ reveal the formation of persulfide adducts. The broad feature at 29,900 Da in the anaerobic spectrum corresponds to poorly resolved cluster fragments. (B) Survey plot showing the formation of persulfide adducts during the reaction time course. (C) Abbreviated reaction scheme based on that in Fig. 3A used to simulate the kinetic behavior of FNR persulfide species. (D) Plots of relative abundances of the apo-S (blue squares), apo-SS (gray triangles), and apo-SSS (yellow circles) species detected in native MS experiments. Global fitting to the experimental data using the reaction scheme depicted in Fig. 4C is shown as solid lines. Dashed lines show the responses of the [4Fe-4S] (black), [2Fe-2S] (red), and apo (gray) species for comparison. Error bars show SEs for average MS datasets (n = 4). The global fitting model was initiated with 100% relative abundance of [4Fe-4S] clusters.

The addition of O₂ resulted in a 2-Da shift in the apo-FNR to 29,545 Da, consistent with the presence of a single disulfide bond. In addition, there was an increased abundance of peaks corresponding to the sulfane adducts as previously observed (16) (Fig. 5). As can be seen from Fig. 5B, the single or double sulfane adducts of FNR become the major species with increasing time. The time dependence of their appearance shows that sulfide oxidation occurs simultaneously with cluster conversion. From the native MS experiments above, it was suggested that the [2Fe-3S] and [2Fe-4S] species give rise to the single and double sulfane forms of apo-FNR, respectively, after the disassembly of the persulfide-ligated [2Fe-2S] cluster. The temporal appearance of the single and double sulfane adducts, maximizing at ~30–40 min, supports this hypothesis (Fig. S3).

Open in a separate window

Fig. S3.

Tracking sulfide oxidation by LC-MS. Plots of relative abundances of apo-S (gray circles), apo-SS (red circles), and apo-SSS (blue circles) detected by LC-MS after exposure to excess O₂. Global fits to the experimental data using the reaction scheme depicted in Fig. 5C are shown as solid lines. The predicted response of the apo species is shown in yellow (note the similarity to the apo species’ behavior in Fig. 4B). The global fitting model was initiated with 100% relative abundance of [4Fe-4S] clusters.

The apo(S)_n species detected by LC-MS result from not only the presence of apo(S)_n species in solution but also, denaturation of cluster-bound sulfane adducts present at the point of quenching, and these adducts cannot be distinguished in the LC-MS experiment. However, because the [2Fe-2S] forms are not stable in the ammonium acetate buffer of the native MS experiment, the kinetic profiles of formation of apo(S)_n species in the LC-MS are similar to those in the native MS experiment, and the LC-MS data could be readily modeled by extending the initial mechanistic model (Fig. 5C) to reflect the proposed fate of the [2Fe-2S] clusters (Fig. S3). An important feature of the extended model is that apo(S)_{n = 1–4} species are in equilibrium with each other, implying that sulfide is readily mobilized from one species to another, presumably via disulfide exchange. The extended model, initiated with [4Fe-4S] FNR at 100% relative abundance, describes the appearance of the apo(S)_{n = 1–3} adducts detected via LC-MS and importantly, was also able to model the formation of apo(S)_n species detected by native MS (Fig. 5D). The observed rate constants obtained from the experimental data fits (Fig. 5D and Fig. S3) are given in Table 3.

Mass spectrometry.

For native ESI-MS, an aliquot of S24F FNR was exchanged into buffer E (250 mM ammonium acetate, pH 6.7) using Zeba spin (Thermo Scientific) or midi-PD10 (GE Healthcare) desalting columns; the volume of the eluent was increased to 2 mL, and the concentration of [4Fe-4S] S24F FNR was determined via absorbance at 406 nm (see below). An aliquot of the sample was combined with an aliquot (1.6 mL) of buffer E containing dissolved atmospheric O₂ (168 μM O₂; final concentration) in an anaerobic cuvette to give ~14 μM [4Fe-4S] S24F FNR. Buffer E lacking dissolved O₂ was used for control experiments. The sample was immediately loaded into a 1-mL gas-tight syringe (Hamilton) and infused directly using a syringe pump (0.3 mL/h) into the ESI source of a Bruker micrOTOF-QIII mass spectrometer (Bruker Daltonics) operating in the positive ion mode. The ESI-TOF was calibrated using ESI-L Low Concentration Tuning Mix (Agilent Technologies). Before the introduction of sample, the gas-tight syringe (Hamilton) and associated PEEK tubing (Upchurch Scientific) were flushed with 5 mL anaerobic buffer E. The oxygen permeability of PEEK tubing is 14 mL/250 cm² (1 atm/25 °C) over 24 h (Upchurch Scientific). MS data were acquired over the m/z range of 700–3,500 continuously for 120 min, with acquisition controlled using Bruker oTOF Control software and parameters as follows: dry gas flow of 4 L/min, nebulizer gas pressure of 0.8 Bar, dry gas at 180 °C, capillary voltage of 4,500 V, offset of 500 V, ion energy of 5 eV, collision radio frequency of 200 Vpp, and collision cell energy of 10 eV. Optimization of experimental conditions for the transmission of dimeric species (2,500–5,000 m/z) was achieved by increasing the equivalent of the cone voltage (in-source collision-induced dissociation of 75 eV), reducing the collision cell energy (4 eV), and increasing the collision RF (1,500 Vpp) (34).

For LC-MS, an aliquot of S24F FNR was exchanged into buffer E (120 μL; ~252 μM [4Fe-4S]) using a Zeba spin column, and the volume of the eluent increased to 500 μL. For O₂ reactivity measurements, the sample was combined with an aliquot (1.6 mL) of oxygenated buffer E (~14 μM [4Fe-4S], 168 μM O₂; final concentration) in an anaerobic cuvette and allowed to react. Samples (50 μL) were removed at varying time points and diluted to ~1.5 μM (final concentration) with an aliquot (406 μL) of quenching solution [an aqueous mixture of 0.8 mM EDTA, 0.7% (vol/vol) formic acid, 2% (vol/vol) acetonitrile]. For experiments probing the reversibility of adduct formation before and after reaction with O₂, samples were treated with 20 mM DTT. Samples were sealed, removed from the anaerobic cabinet, and injected (1 µL) onto a ProSwift reversed phase RP-1S column (4.6 × 50 mm; Thermo Scientific) at 25 °C using an UltiMate 3000 HPLC system (Dionex).

Gradient elution was performed at a flow rate of 0.2 mL/min using a linear gradient (15 min) from 2 to 100% (vol/vol) acetonitrile and 0.1% (vol/vol) formic acid. The eluent was continuously infused into a Bruker micrOTOF-QIII mass spectrometer running Hystar (Bruker Daltonics) using positive mode ESI. The mass spectrometer was calibrated with ESI-L tuning mix (Agilent Technologies). MS acquisition parameters were as follows: dry gas flow of 8 L/min, nebulizer gas pressure of 1.8 Bar, dry gas at 240 °C, capillary voltage of 4,500 V, offset of 500 V, and collision RF of 650 Vpp.

Processing and analysis of MS experimental data were carried out using Compass DataAnalysis, version 4.1 (Bruker Daltonik). Neutral mass spectra were generated using the ESI Compass, version 1.3 Maximum Entropy deconvolution algorithm over a mass range of 29,000–31,000 Da for the monomer and 59,000–60,500 Da for the dimer. For kinetic modeling, to clearly resolve overlapping peaks, multiple Gaussian functions were fitted to the experimental data using a least squares regression function in Origin 8 (Microcal) (34). Exact masses are reported from peak centroids representing the isotope average neutral mass. For apoproteins, these masses are derived from m/z spectra, for which peaks correspond to [M + nH]ⁿ⁺/n. For cluster-containing proteins, where the cluster contributes charge, peaks correspond to [M + FeS^x+ + (n − x)H]ⁿ⁺/n, where M is the molecular mass of the protein, FeS is the mass of the iron-sulfur cluster of x+ charge, H is the mass of the proton, and n is the total charge. In the expression, the x+ charge of the cluster offsets the number of protons required to achieve the observed charge state (n+) (26). Predicted masses are given as the isotope average of the neutral protein or protein complex in which cofactor binding is expected to be charge compensated (35). For time-resolved MS intensity data, kinetic schemes were modeled using Dynafit 4 (BioKin Ltd) (36).

We thank the University of East Anglia for funding the purchase of the electrospray ionization mass spectrometer instrument. This work was supported by Biotechnology and Biological Sciences Research Council Grant BB/L007673/1 (to J.C.C., A.J.T., and N.E.L.B.).