Introduction

The conversion of methane to methanol is of significant interest for technologies aimed at gas-to-liquid fuel conversion, and for the sequestration of a greenhouse gas, alongside the desire to understand the fundamental chemistry of strong C–H bond activation. Nature facilitates this challenging chemical reaction with two different methane monooxygenase enzymes [1]. The most common type is the copper-containing particulate methane monooxygenase (pMMO) [2,3,4], which is a topic of intense research, in particular for the assignment of the active site and its structure [5,6,7,8,9,10,11]. In copper limited environments [2, 12], methanotrophic bacteria express the iron-containing soluble methane monooxygenase (sMMO), which has a well characterized non-heme carboxylate-bridged di-iron active site [13, 14].

The active site of sMMO is located within the hydroxylase protein (MMOH) and forms various catalytic intermediates that have been spectroscopically identified [13, 15,16,17,18,19,20,21,22,23,24,25]. An abbreviated catalytic mechanism of MMOH is shown in Fig. 1. For the critical intermediate of interest, MMOHQ, the O–O bond of the preceding peroxo intermediate (MMOHP) has been cleaved and the resultant intermediate is able to activate the strong 105 kcal/mol C–H bond of methane, allowing for its subsequent conversion to methanol. To our knowledge, MMOHQ is the only di-iron(IV) intermediate that has been identified in Nature [17, 19, 20].

Fig. 1
figure 1

Abbreviated catalytic scheme of sMMO with corresponding iron oxidation states, local spin states (Sloc) and cluster ground spin state (Stot) for MMOHox, MMOHred, and MMOHQ. The structure of MMOHQ is drawn as the proposed ‘open-core’ structure [23, 24], although a closed Fe2(µ-O)2 core is also proposed [21, 26]

Full size image

The exact geometric structure of the MMOHQ intermediate has been hotly debated for over two decades [14, 17, 18, 21, 23,24,25,26,27]. Spectroscopic characterizations of MMOHQ have included vibrational studies (transient resonant Raman and nuclear resonance vibrational scattering, NRVS), which have also been used to argue for a “closed-core” bis-µ-oxo structure [21, 25]. However, high-resolution Fe K-edge X-ray absorption (HERFD XAS) characterization and pre-edge analysis of MMOHQ in comparison with biomimetic models support an “open core” structure [23]. Recent HERFD extended X-ray absorption fine structure (EXAFS) measurements of RFQ samples of sMMO were fit with a long Fe–Fe distance of ~ 3.3 Å and showed no evidence for the initially reported 2.46 Å di-iron distance that would have supported a closed-core structure [24]. The short Fe–Fe distance in the previous 1997 EXAFS report [18] was shown to arise from metallic iron background scattering contributions, which are suppressed in the HERFD EXAFS [24]. Because the RFQ samples in the HERFD EXAFS study contained primarily MMOHQ (~ 50%) and MMOHred (40%), the fitted distance is weighted average of these contributions. As the di-iron distance in MMOHred is established to be ~ 3.2 Å [28], the longer fitted distance of the RFQ samples therefore suggests a longer di-iron distance for MMOHQ of ~ 3.4 Å, consistent with an open core structure [24]. Further, a recent quantum mechanics-molecular mechanics (QM/MM) computational study of MMOHQ has shown that the computed spectroscopic signatures of an open core structure are consistent with the Mössbauer, Fe HERFD pre-edge XAS, EXAFS di-iron distance measurements, and resonance Raman [29]. These QM/MM computational studies, however, have not yet been extended to the recently reported NRVS spectrum of MMOHQ [25].

While the geometric structure of MMOHQ remains controversial, its electronic structure is also key to understanding catalytic activity and predicting spectroscopic signatures. The initial spectroscopic characterization and identification of MMOHQ included its characteristic optical absorption at 430 nm and a single observed Mössbauer quadruple doublet in the case of MMOHQ trapped using the enzyme from Methylosinus trichosporium OB3b [16, 19, 20]. The single Mössbauer doublet from two irons was interpreted to indicate that the iron sites have similar coordination environments, thereby yielding overlapping doublets. The isomer shift of the observed doublet is entirely consistent with FeIV oxidation state, however, the Mössbauer characterization of MMOHQ remains inconclusive for the assigned local spin states of the iron sites. Low temperature and applied magnetic field Mössbauer spectroscopy of MMOHQ determined that the total ground spin state of the cluster is Stot = 0, meaning that the FeIV ions are antiferromagnetically coupled, where iron sites must have local spin states (Sloc) of intermediate-spin (Sloc = 1) or high-spin (Sloc = 2) [20]. As originally stated by Münck, the isomer shift and quadrupole splittings of MMOHQ are consistent with both Sloc = 1 and Sloc = 2 local spin states; the applied field measurement only serves to measure the coupling strength between the two sites [20]. Non-diamagnetic cluster spin states allow for clear local spin-state assignments by Mössbauer spectroscopy. For other mononuclear FeIV synthetic chemical and biological centers, applied field Mössbauer measurement has been essential to differentiate between intermediate and high-spin iron [30,31,32,33,34]. For example, in the case of the Stot = 1/2 FeIII–FeIV cluster intermediate (X) of ribonucleotide reductase (RNR) [35], the paramagnetic nature of the cluster allowed iron hyperfine and local spin states to be determined by applied field measurements.

Attempts to understand or model both the local and cluster spin states of MMOHQ through biomimetic chemistry have also been challenging [33, 36,37,38,39]. Different local and cluster spin states have been identified in these models, including those with Sloc = 1 or 2 and ferro- or antiferromagnetic coupling of the iron sites. A sampling of both mono and di-nuclear biomimetic models and their Mössbauer parameters are listed in Table 1. The observed Mössbauer isomer shifts and quadrupole splittings for both Sloc = 1 and 2 FeIV sites span a fairly wide range around what is observed for MMOHQ, making a direct assignment of a local spin state from these parameters alone prohibitive.

Table 1 Mössbauer parameters of various FeIV sites
Full size table

Perhaps the primary support for the now more commonly accepted Sloc = 2 spin state of MMOHQ arises in part from analogy to other non-heme iron chemistry [39]. Enzymatic non-heme FeIV sites favor local high-spin states, in part due to their weak-field ligand environments [30], similar to that present in MMOH. Furthermore, the enhanced reactivity of high-spin S = 2 FeIV sites to perform hydrogen atom transfer reactions, as seen in other forms of non-heme biological catalysis and biomimetic chemistry [31, 33, 34, 38, 48, 49], is often used to infer a spin-state assignment for MMOHQ. Cryo-reduction experiments of MMOHQ have also revealed an electronic structure similar in character to RNR-X, suggesting locally high-spin FeIV sites [50]. Previous computational studies of MMOHQ also favor Sloc = 2 iron spin states [29, 51,52,53]. Ultimately, these chemical analogies or other studies have fallen short of a direct experimental characterization of the local spin states in MMOHQ, motivating an investigation of the electronic structure by means of an alternative spectroscopic approach.

For 3d transition metals, Kβ (3p → 1s) X-ray emission spectroscopy (XES) has been well demonstrated to be excellent reporter of electronic structure due to 3p–3d exchange in the final state [54,55,56,57,58]. Kβ mainlines have previously been demonstrated to both report and correlate to oxidation state and spin state. As a marker of oxidation state, the maximum of the Kβ mainline, the Kβ1,3 feature, generally shifts to higher energy with increased oxidation-state [54]. On the other hand, the intensity of the Kβ’ feature, at lower energy, correlates with the number of unpaired electrons, thereby allowing for spin-state assignments. For instance, for ferrous complexes, the presence or absence of the Kβ’ has been used as a marker of high-spin (S = 2) or low-spin (S = 0) electronic configurations [57,58,59,60].

While these rule-of-thumb trends may hold true in idealized systematic studies, covalency is shown to have profound influences on the shape and energies of Kβ mainlines [56, 58]. Increasing metal–ligand covalency delocalizes metal 3d orbital character onto the ligands and thus decreases the metal 3p-3d exchange integrals, which determine the final state splitting in Kβ spectra [56]. Ultimately, this relationship predicts that expected shifts of the Kβ1,3 to higher energy with increased oxidation-state may be counteracted by increased metal–ligand covalency, making some interpretations across an oxidation state series difficult. However, these challenges generally do not hinder the ability of Kβ XES to probe local spin states and distinguish between spin states in samples of known oxidation-state and similar covalency [54, 56,57,58, 61, 62]. Generally, Fe Kβ XES is not sensitive to the magnetic spin coupling (J) between the iron centers in bimetallic or larger clusters; therefore, the bulk XES measurement reports on the average local spin state of all iron sites [63,64,65,66]. This makes the approach attractive to possibly probe the local spin states in diamagnetic clusters such as MMOHQ and distinguish between its local spin-state assignments.

Experimental

The sMMO proteins 57Fe-enriched MMOH and MMOB were purified according to protocols described recently in the literature [22, 67]. The Fe Kβ mainline XES spectra were collected on frozen solution samples of MMOHred, MMOHox and freeze-quenched samples (MMOH-RFQ) to trap MMOHQ of the same samples described in a previously published HERFD EXAFS study [24]. The MMOH-RFQ samples were prepared as previously described by freezing the reaction mixture in precooled sample cell of super-cooled liquid nitrogen (-199 °C) [24].

57Fe Mössbauer spectroscopy was used to quantitate the three components, MMOHred, MMOHox and MMOHQ, in the MMOH-RFQ sample. As previously reported [24], the frozen solution samples had 46% MMOHQ, 39.5% MMOHred, and 14.5% MMOHox.

The S = 1, FeIV = O complex ([FeIV(O)(2PyN2Q)](PF6)2 (2PyN2Q = 1,1-di(pyridin-2-yl)-N,N-bis(quinolin-2-ylmethyl)methanamine)) was synthesized and isolated following published procedures [44, 68]. Solid samples for XES data collection were prepared in 1.0 mm Al spacers and sealed with 38 µm thick Kapton tape.

The Fe Kβ XES data were collected at beam line ID-26 of the European Synchrotron Radiation Facility (ESRF) operating at 6 GeV and 200 mA. All samples were measured in a liquid helium cryostat operating at 20 K. The energy of the incident beam was selected using either a Si(111) or Si(311) double crystal monochromator, each of which was calibrated by setting the first inflection of an iron foil to 7111.2 eV. The incident monochromator was then set to an excitation energy of 7800 eV to non-resonantly excite the sample. The XES spectra were collected with a 1 m radius Johann spectrometer equipped with five Ge(620) spherically bent analyzer crystals and calibrated by the scanning of elastic scattering lines. A collected reference Kβ XES spectrum of Fe2O3 is displayed in Fig. S1. The measured resolution of the elastic scattering line was ~ 1.5 eV (fwhm) which includes the contribution of the upstream Si(111) monochromator. For the synthetic solid sample, an avalanche photodiode (APD) detector was used, while the more dilute protein samples required the use of an energy resolving Ketek detector to further improve the S/N of the emission spectra. For the XES collection of the protein, the incident beam was attenuated to a maximum estimated flux of 4 × 1012 photons/s within a spot size of 1.2 mm (w) × 0.1 mm (v) to further reduce radiation damage. Maximum sample exposure dwell times were determined through the evaluation of repeated fast X-ray near edge absorption (XANES) scans (5–60 s) to determine total acceptable photon doses [24]. Collection of the protein XES was completed in small segments (~ 12 eV wide, 50–61 points) of the entire XES spectra (7.025–7.08 keV) with 2 eV overlap per energy segment (10–11 data points). Each segment was collected on a fresh sample spot, limiting the total sample spot exposure time. The maximum dwell times per data point were: MMOHred, 1.25 secs; MMOHox 0.40 secs; MMOHQ, 0.15 secs. This limited the sample spot exposure times to a maximum of 76.25 secs (MMOHred), 24.4 secs (MMOHox) and 15.25 secs (MMOHQ). This collection process allowed for improved signal-to-noise to be acquired at each data point, and to mitigate beam induced damage. All data segments were splined together through the averaging of their overlapping components. A sample of the process is shown in Fig. S2.

All Kβ XES spectra were normalized to a unit area of 1 over the energy range of 7025 to 7080 eV. The XES spectrum of pure MMOHQ was calculated by subtracting the MMOHred and MMOHox components from MMOH-RFQ at the ratios determined by Mössbauer and renormalizing to a unit area of 1. The first moments reported are determined over the described energy ranges by the following equation,

$${M}_{1}=\frac{{\sum }_{i}\left({E}_{i}{I}_{i}\right)}{{\sum }_{i}{I}_{i}}$$

where Ei and Ii are the energy and intensities, respectively, at data point i.

Results

Samples of MMOHred, MMOHox and rapid-freeze-quenched (RFQ) samples of MMOHQ were prepared as previously described [24]. The Fe Kβ XES spectra of the three prepared samples of sMMO (MMOHred/ox/RFQ) are shown in Fig. 2a. No significant differences in the Kβ1,3 maxima are apparent. One does, however, observe slight intensity differences in the Kβ’ region (at ~ 7045 eV) following a trend of MMOHred > MMOHox > MMOH-RFQ. As the RFQ sample is a mixture of all three different components and therefore three oxidation states, an assignment of the convoluted spectrum is challenging. The individual components of the RFQ samples were quantitated by 57Fe Mössbauer spectroscopy, with an estimated 46% yield of MMOHQ [24]. The 57Fe Mössbauer quantification of each component in the MMOH-RFQ sample allows for the ‘pure’ MMOHQ spectrum to be calculated as previously established in the HERFD XAS studies of sMMO and other non-heme di-iron proteins [23, 69].

Fig. 2
figure 2

a Fe Kβ mainline XES of MMOHred, MMOHox and MMOH-RFQ samples of sMMO. b The MMOHQ spectrum is determined by subtraction of the MMOHred and MMOHox components from the MMOH-RFQ spectrum. c The MMOHQ XES is compared to that of an S = 1 FeIV = O

Full size image

The ‘pure’ MMOHQ mainline spectrum is shown in Fig. 2b, overlaid with MMOHred and MMOHox. Here, one can clearly see that the Kβ1,3 peak position does not shift for the various oxidation states FeII2/FeIII2/FeIV2 of sMMO. Often, the first moment of the Kβ1,3 is used rather than the peak maximum to determine energy shifts because the first moment has been shown to have more sensitivity to small changes [54, 70,71,72]. However, the first moment analysis of the three spectra, Table 2, does not reveal significant energy shifts either, even though it was previously shown that all three of these samples exhibit clearly different XAS pre-edge and rising edge energies (see Figure S5 in Ref. [24]), consistent with their anticipated oxidation states and previous HERFD-XAS characterization [23].

Table 2 Key Fe Kβ mainline parameters
Full size table

While the Kβ1,3 may not exhibit clear differences with oxidation-state changes, the Kβ’ intensity of the three states clearly varies, Fig. 2b. The Kβ’ feature of MMOHred appears slightly more intense than MMOHox. This is in agreement with the subtle intensity differences observed in other high-spin FeII versus FeIII mainlines, such as [Fe(H2O)6]2+/3+ [58]. MMOHQ has a clear Kβ’ feature, indicating that there are unpaired d electrons (Sloc > 0). The first moment of the Kβ’ of MMOHQ feature is at approximately the same energy as the more reduced iron species. In fact, the Kβ’ features for each of the three MMOH states appear at approximately the same energies, with only the intensities of the Kβ’ features varying, making the Δ(Kβ1,3– Kβ’) splitting approximately the same for each species, Table 2.

The mainline of MMOHQ may be the result of two local FeIV spin states: high-spin S = 2, or intermediate-spin S = 1. To determine the spin state, the mainline is directly compared to the FeIV S = 1 mainline from an FeIV = O complex [Fe(O)2PyN2Q)]2+, Scheme 1. Two immediate differences are observed in the spectra plotted in Fig. 2c. First, the Kβ1,3 energy for the FeIV = O S = 1 model complex is clearly shifted to lower energy by approximately 1 eV compared to MMOHQ and second, the Kβ’ region for the FeIV = O model is significantly less intense than that of MMOHQ. Furthermore, the same first moment analysis of the Kβ’ of the FeIV = O mainline reveals an + 1 eV compared to MMOHQ, decreasing the Δ(Kβ1,3– Kβ’) splitting by ~ 2 eV, comparatively.

Scheme 1
scheme 1

FeIV = O; [Fe(O)(2PyN2Q)]2+

Full size image

With respect to the Kβ’ intensity of MMOHQ, one may refer to the fingerprinting analyses performed for ferrous ion sites [57,58,59,60]. The presence of the Kβ’ in the high-spin S = 2 and its absence in low-spin S = 0 centers is used as a spin-state diagnostic tool. For MMOHQ, the clearly greater intensity of its Kβ’ feature and larger Δ(Kβ1,3– Kβ’) splitting compared to that of the established S = 1 FeIV = O in Fig. 2c strongly suggests higher local spin states for the FeIV ions of MMOHQ.

For the Δ(Kβ1,3– Kβ’) splitting, it has been previously demonstrated that the energetic splitting is proportional to the nominal spin of the metal center, but these energy shifts may be further modulated by covalency [56]. However, the quantification of the covalency in the various states of MMOH is difficult to estimate. One may naturally consider a potential di-iron(IV) open core structure that contains terminal oxos to be more covalent than the MMOHox and MMOHred states. However, the previous Mn Kβ mainline characterization of the step-wise deprotonation of high-spin MnIV S = 2 Mn2(µ-OH)2 dimers to the Mn2(µ-O) cores exhibited only minimal shifts (~ 0.5 eV) in the Kβ1,3 energies [64]. This suggests that when comparing potential oxo vs hydroxo ligation to the iron sites, the covalency changes will result in only minor energetic perturbations and that local iron spin state should be the dominant contribution to the Kβ mainline energies. Our analysis and comparison of the sMMO Kβ mainline lines clearly supports the conclusion that the local spin states of the iron sites of MMOHQ are S = 2. Furthermore, the Kβ mainline analysis of MMOHQ as an Sloc = 2 is very consistent with the Sloc = 2 and Sloc = 5/2 spin states of the MMOHred and MMOHox, respectively [13, 73].

Discussion

While previous Mössbauer experiments of MMOHQ provide evidence for isomer shifts in clear agreement with FeIV ions, the antiferromagnetic coupling and Stot = 0 cluster spin state do not yield conclusive information of the local spin states of the iron sites. Here, we have used Fe Kβ XES to probe the local spin state of MMOHQ. The ΔKβ splitting and the Kβ’ intensity of MMOHQ are excellent indicators of local high-spin S = 2 iron sites that antiferromagnetically couple to form the Stot = 0 cluster.

The Kβ mainline of MMOHQ has distinct features that also resemble other assigned S = 2 FeIV centers [58, 74]. In a recent study of the Fe Kβ signatures of various compounds, the Kβ emission spectra of the iron perovskites LaFeIIIO3 and SrFeIVO3 were reported [58]. In LaFeIIIO3, a typical S = 5/2 ferric mainline is observed, whereas the high-spin S = 2 FeIV mainline of SrFeIVO3 exhibits a similar Kβ1,3 mainline maxima, but a slightly decreased Kβ’ intensity, similar to that observed for MMOHQ. However, we do note that this perovskite sample does not have the same molecular and/or electronic properties as the non-heme carboxylate-bridged di-iron center of sMMO, and so only qualitative comparisons are made.

Perhaps the most closely related structure to MMOHQ is an RNR intermediate. The more widely studied di-iron RNR forms a high-valent FeIII–FeIV intermediate termed ‘X,’ that has a Stot = 1/2 cluster spin state as described earlier. Similarly, the hetero-metallic Mn–Fe class I-c of RNRs also forms a high-valent intermediate MnIV–FeIV intermediate [75]. For the d4 Fe center of the MnIV–FeIV RNR, the Fe Kβ mainline does not exhibit any clear energy shifts of the Kβ1,3 for the local FeII, FeIII and FeIV centers, and the FeIV Kβ’ exhibits a decreased intensity relative to the lower Fe oxidation-state intermediates [74]. The Fe Kβ XES of the S = 2 FeIV center from Mn–Fe RNR has similar intensity to what we have observed for MMOHQ (Figs. S3 and S4), lending additional support to the idea that measured Kβ XES of MMOHQ reflects two S = 2 FeIV ions and not a beam damaged product yielding high-spin S = 5/2 ferric sites, which would further increase Kβ’ intensity and ΔKβ splitting.[25, 50] Furthermore, the ΔKβ splitting observed in RNR is similar to that seen here in MMOHQ, Fig. S4. It is important again to note the clear presence of a Kβ’ feature in the S = 2 FeIV RNR compared to the lack of a well-resolved Kβ’ in the S = 1 FeIV = O model reported here (Fig. S4), offering further support that that the Kβ mainline observed for MMOHQ is diagnostic of an S = 2 local spin states at the iron atoms.

Conclusion and outlook

In summary, the high-spin FeII, FeIII and FeIV sites of sMMO do not exhibit oxidation-state-dependent energy shifts by Fe Kβ XES. Modest differences are observed in the Kβ’ due to spin state. Notably, the ΔKβ(Kβ1,3-Kβ’) splitting of sMMO remains fairly constant, which is very consistent with high-spin states for MMOHox, MMOHred, and MMOHQ. Direct comparison of the MMOHQ mainline to a known S = 1 FeIV complex has shown dramatic spectral differences due to changes in the apparent spin state. This approach, in line with Fe Kβ XES spin state fingerprinting for ferrous and ferric iron, clearly distinguishes between the intermediate S = 1 spin of the model complex studied here and the apparent Sloc = 2 of MMOHQ. The Kβ XES experiment offers a clear experimental approach to determine local spin state in ambiguous systems, where magnetic spin coupling may make such determinations challenging for other experimental methods.

In the realm of X-ray spectroscopic studies, XES has become a viable tool to monitor spin state and/or oxidation states in both static samples and in situ or operando studies. Some of the clearest utility of Fe Kβ XES has been to discriminate between low- versus high-spin iron species [56,57,58,59,60]. The lack of significant shifts in Kβ1,3 mainline energies for the different oxidation states of MMOH intermediates precludes one from attempting more advanced X-ray spectroscopic techniques, such as “site selective” XAS or EXAFS for further enhancement of the MMOHQ signal in the mixed sample through the means of energy selective detection [76, 77]. The mainline spectra reported here will be valuable in future studies of sMMO. With the previous success of using Mn Kβ emission to fingerprint oxidation state during in situ measurements [70, 78], one hopes to be able to utilize the Fe Kβ XES to fingerprint, monitor and validate electronic changes during in situ studies. Such XES approaches have been highlighted in X-ray free electron laser (XFEL) protein diffraction studies to monitor a catalytic site’s oxidation-state [79, 80]. The present results on the Fe Kβ mainlines of sMMO, however, demonstrate the potential difficulty of employing such methods for monitoring oxidation-state during in situ studies, where the Kβ1,3 energy does not offer great sensitivity but the Kβ’ itself offers spin state (correlated to the oxidation-state) information through its clear intensity changes. Given the inherently lower intensity of the Kβ’ feature, using this feature to follow spin state will require data collection with high signal-to-noise to ensure accurate assignments. We do note that the Kα (2p → 1 s) emission spectrum is an order of magnitude more intense than the Kβ emission, allowing for easier measurements. Kα XES does not offer the same degree of electronic sensitivity, and extraction of oxidation or spin-state information is challenging [54, 58]. The three states of sMMO do not exhibit significant energy differences in their Kα spectra (Figure S6 in ref [24].) and more recent Kα XES of MMOHred and MMOHox collected during XFEL crystallography experiments exhibits only subtle changes in the linewidths and shapes [81]. While more challenging to collect, Kβ XES appears to offer the most sensitivity to spin state and the best potential fingerprint for formation of the di-iron(IV) (Sloc = 2) cluster.

Fe Kβ XES of sMMO has proven to be a valuable technique to probe the local spin states of the di-iron site in the catalytic cluster. Understanding and verifying that MMOHQ is, in fact, Sloc = 2 is important to both understanding and proposing mechanisms for reactivity. Further, the ability to clearly distinguish S = 1 from S = 2 iron sites via Kβ XES has potential for experimentally assessing the role of two-state reactivity in a wide range of high-valent FeIV–oxo complexes [48, 82]. Hence, the present results deepen our understanding of the electronic structure of MMOHQ, while providing experimental fingerprints to enable the study of a wide range of high-valent iron species in both molecular models and enzymes.