Temperature-Dependent Reactivity of a Non-heme Fe^III(OH)(SR) Complex: Relevance to Isopenicillin N Synthase

Abstract

Non-heme iron complexes with cis-Fe^III(OH)(SAr/OAr) coordination were isolated and examined for their reactivity with a tertiary carbon radical. The sulfur-ligated complex shows a temperature dependence on ·OH versus ArS transfer, whereas the oxygen-ligated complex does not. These results provide the first working model for C−S bond formation in isopenicillin N synthase and indicate that kinetic control may be a key factor in the selectivity of non-heme iron “rebound” processes.

Isopenicillin N synthase (IPNS) belongs to a class of nonheme iron enzymes that utilize Fe^II/O₂ in the absence of a cosubstrate to catalyze the oxidation of L-Δ-(α-aminoadipoyl)-L-cysteinyl-D-valine (ACV) tripeptide to isopenicillin (IPN) (Scheme 1).¹⁻³ IPN gets further processed to form antibiotics such as penicillin and cephalosporins.⁴⁻⁶ The biosynthesis of IPN is divided into two major steps: (a) formation of the β-lactam ring via Fe^III−OO^•− and Fe^III−OOH intermediates and (b) closure of the thiazolidine ring involving C−S bond formation from the reaction of a tertiary carbon radical (R·) with an Fe^III(OH)(SR) intermediate.⁷⁻¹¹ Substrate probes have shown that other products besides the native thiazolidine compound can be formed (e.g., S-oxygenates,¹⁰ thioacids,¹² and ring-expanded¹³ or hydroxylated¹⁴ products). These studies suggest that the substrate structure and orientation are important in determining the outcome of the reaction between R· and Fe^III(OH)(SR). Computational studies showed that sulfur transfer is kinetically favored over hydroxylation, supporting the observed selectivity of this step with the native substrate.^11,15 However, experimental studies that directly examine the C−S bond formation step are absent.

Scheme 1.

IPNS Reactivity and Model

The inherent factors that may contribute to the selectivity of sulfuryl versus hydroxyl transfer in the absence of a protein pocket have not been examined previously. Similar selectivity arises for halogen versus hydroxyl transfer in the non-heme iron halogenases, and the properties that control the selectivity for halogenation are still under debate.¹⁶⁻²⁶ A few synthetic non-heme iron catalysts have shown some promise toward selective halogenation and related processes,²⁷⁻³⁴ but the principles for designing a selective halogenation catalyst are not well-understood. Similarly, there are no reports describing the analogous Fe^III(OH)(SR) species, and no studies to date have shown selective sulfuryl over hydroxyl transfer mediated by a non-heme iron complex.

We previously developed a ligand with H-bonding groups (BNPA^Ph2O⁻) that allowed us to isolate Fe^III(OH)(X) (X = OTf, Cl, Br) complexes and examine their reactivity toward carbon radicals.^35,36 Herein we report the first structurally characterized Fe^III(OH)(SAr) complex, prepared from BNPA^Ph2O⁻, and describe its reactivity toward a tertiary carbon radical. The phenolate analogue, Fe^III(OH)(OAr), was also prepared and examined for comparative reactivity. A temperature-dependent switch in hydroxyl versus sulfur transfer is seen for the arylthiolate analogue.

Complexes Fe^II(BNPA^Ph2O)(SPh^p-NO2) (1) and Fe^II(BNPA^Ph2O)(OPh^p-NO2) (2) were synthesized by the addition of the appropriate sodium thiophenolate or phenolate salt to Fe^II(BNPA^Ph2O)(OTf).³⁶ The crystal structures (Figure 1) revealed five-coordinate iron(II) complexes with the thiophenolate/phenolate ligand bound in place of OTf⁻. Bond distances are typical for high-spin (hs) iron(II).³⁷⁻³⁹

Figure 1.

Displacement ellipsoid plots (50% probability level) for 1 and 2 at 110(2) K. Hydrogen atoms (except for N−H) have been omitted for clarity.

The reactions of 1 and 2 with dry excess O₂ in THF at 23 °C (Scheme 2) led to the ferric complexes Fe^III(BNPA^Ph₂O)- (OH)(SPh^p-NO2) (3) and Fe^III(BNPA^Ph2O)(OH)(OPh^p-NO₂) (4), respectively, which were characterized by single-crystal X-ray diffraction (XRD). The crystal structures (Figure 2) reveal six-coordinate complexes with the thiophenolate or phenolate ligand bound in an equatorial position and a terminal hydroxide ligand occupying the axial H-bonded site, as observed previously.^35,36 The Fe^III−OH distances of 1.9034(18) and 1.908(2) Å in 3 and 4, respectively, are similar to those in other terminal Fe^III(OH) complexes.^38,40−48 In contrast, the Fe^III−S bond length of 2.4483(8) Å in 3 is longer than the few other non-heme high-spin Fe^III−SAr distances previously reported (2.35(1)−2.41(2) Å).⁴⁹⁻⁵² However, the analogous Fe^III−OAr distance of 2.000(3) Å in 4 is within the typical range (1.93−2.00 Å).⁵³⁻⁵⁵

Scheme 2.

Synthesis of 3 and 4

Figure 2.

Displacement ellipsoid plots (50% probability level) for 3 and 4 at 110(2) K. Hydrogen atoms (except for N−H and O−H) have been omitted for clarity.

Comparison of 3 and 4 shows that the phenolate group is trans to the alkoxide in 4, whereas the thiophenolate group is trans to a pyridine donor in 3. In addition, there is no π−π stacking between the phenolate ring and any of the pyridine rings in 4, while there is π−π stacking⁵⁶ between the thiophenolate group and the pyridine ring containing N(4) in 3. This interaction is characterized by a centroid-to-centroid distance of 3.6 Å and an angle of 9.3° between the least-squares planes of the two aromatic rings.⁵⁷

A density functional theory (DFT) calculation gave Fe−S = 2.484 Å for 3, reproducing the elongated Fe−S bond length observed in the crystal structure. A comparison of the structure of 3 with an optimized geometry (QM/MM) for the proposed cis-Fe^III(OH)(SR) intermediate in IPNS¹⁵ reveals a resemblance between the Fe^III−S and Fe^III−OH bond lengths of 2.37 and 1.87 Å, respectively, for IPNS, and those of 3. Thus, complex 3 is, to our knowledge, the first synthetic model of the proposed cis-Fe^III(OH)(SR) intermediate in IPNS.

The ¹H NMR spectra of 1 and 2 show relatively sharp paramagnetically shifted peaks from 90 to −10 ppm indicative of hs (S = 2) iron(II). In comparison, the spectra of 3 and 4 exhibit broad resonances from 80 to 10 ppm, as expected for a hs (S = ⁵/₂) Fe^III species. Zero-field Mössbauer spectroscopy of ⁵⁷Fe-enriched 1 and 2 shows sharp quadrupole doublets with parameters Δ = 0.94 mm s⁻¹ and |ΔE_Q| = 2.87 mm s⁻¹ for 1 and Δ = 1.03 mm s⁻¹ and |ΔE_Q| = 2.76 mm s⁻¹ for 2. Mössbauer analysis of ⁵⁷Fe-enriched 3 and 4 revealed broad quadrupole doublets with parameters Δ = 0.42 mm s⁻¹ and |ΔE_Q| = 0.96 mm s⁻¹ for 3 and Δ = 0.47 mm s⁻¹ and |ΔE_Q| = 1.01 mm s⁻¹ for 4. Such broadening for similar ferric complexes is known and can be explained by the population of an intermediate relaxation regime.^35,36,58,59

The complex Fe^II(BNPA^Ph2O)(OH) (5) was also prepared for comparison by adding OH⁻ to Fe^II(BNPA^Ph2O)(OTf). Crystallization of 5 as orange blocks came from a reaction with LiOH and gave the structure shown in Figure 3. There is a lithium ion bound between OH⁻ and O1 and coordinated by OTf⁻ and THF, leading to the formula 5·Li(OTf)(THF). There is also an additional H-bond between OH⁻ and OTf⁻ that stabilizes the structure.

Figure 3.

Displacement ellipsoid plot (50% probability level) for 5· Li(OTf)(THF) at 110(2) K. Hydrogen atoms (except for N−H and O−H) are omitted for clarity.

¹H NMR spectroscopy of 5 prepared in situ from ⁿBu₄NOH or from crystals of 5·Li(OTf)(THF) gave nearly identical spectra (Figures S9 and S10). Mössbauer spectroscopy on ⁵⁷Fe-5 synthesized from either ⁿBu₄NOH or LiOH in 2-MeTHF showed identical isomer shifts but different quadrupole splittings (Table S4). These trends for 5 versus the lithium adduct are reproduced by DFT calculations and support the coordination of Li⁺ in solution.

The reaction of 3 with the substituted triphenylmethyl radical (p-OMe-C₆H₄)₃C· in toluene/THF at 23 °C was then examined. Triarylmethyl radicals are relatively stable and have been used recently by us and others to examine their reactivity with M−X (X = O, N, halide) bonds.^{43,60−66,70} Analysis by ¹H NMR spectroscopy showed the complete conversion of the ferric complex 3 into the ferrous thiolate complex 1, consistent with selective hydroxyl transfer over sulfur transfer from the iron complex to the carbon radical (Scheme 3). The alcohol product (p-OMe-C₆H₄)₃COH was also identified in the ¹H NMR spectrum, and no evidence for formation of the thioether (p-OMe-C₆H₄)₃CSAr was detected. Mössbauer spectroscopy provided additional corroborating data for the selectivity of ·OH transfer. The reaction of isotopically enriched ⁵⁷Fe-3 revealed a sharp quadrupole doublet with Δ = 0.96 mm s⁻¹ and |ΔE_Q| = 2.83 mm s⁻¹, corresponding to the Fe^II(thiolate) complex 1 (Figure 4).

Scheme 3.

Reactions of 3 with Tertiary Carbon Radical

Figure 4.

Zero-field ⁵⁷Fe Mössbauer spectra (80 K) of (a) 1, (b) 3, and (c) 3 + (p-OMe-C₆H₄)₃C· at 23 °C and (d) 5, (e) 3, and (f) 3 + (p-OMe-C₆H₄)₃C· at −35 °C. Isomer shift (Δ) and quadrupole splitting (|ΔE_Q|) values are given in mm s⁻¹.

However, lowering the reaction temperature causes a dramatic shift in the product distribution. Addition of 1 equiv of (p-OMe-C₆H₄)₃C· to 3 at −35 °C for 1 h in THF/ toluene leads to the formation of Fe^II(OH) complex 5 instead of Fe^II(SPh^p-NO2) complex 1, as observed by ¹H NMR spectroscopy. Corresponding analysis by Mössbauer spectroscopy (Figure 4) reveals a sharp quadrupole doublet with Δ = 1.00 mm s⁻¹ and |ΔE_Q| = 2.38 mm s⁻¹, which is a close match to the spectrum of 5. Taken together, the data show that sulfur transfer preferentially occurs over hydroxyl transfer at −35 °C (Scheme 3).

To examine the generality of these reactions, the p-CF₃-substituted complexes Fe^II(BNPA^Ph2O)(SAr^p-CF3) (6) and Fe^III(BNPA^Ph2O)(OH)(SAr^p-CF3) (7) were prepared (see the Supporting Information). Reaction of 7 with (p-OMe-C₆H₄)₃C· at −35 °C leads to formation of the Fe^II(OH) product 5, the same selectivity as seen for 3. The formation of (p-OMe-C₆H₄)₃CSPh^p-CF3 (80%) was also confirmed by ¹H NMR and ¹⁹F NMR spectroscopy.

To examine the influence of temperature in more detail, the reaction of 3 and (p-OMe-C₆H₄)₃C· was carried out between 23 and −35 °C at intervals of 10 °C. The ¹H NMR spectrum for the reaction at 23 °C (Figure S45) shows only the presence of the Fe^II(SAr) complex, the product expected from selective · OH transfer. A second product begins to appear at −5 °C, as evidenced by new peaks at 84.8, 63.8, and 56.1 ppm. These peaks correspond to the Fe^II(OH) complex produced from ArS· transfer to the carbon radical. As the reaction temperature is further lowered, the ArS· transfer pathway becomes more favorable and is the dominant pathway by −25 °C. These data are consistent with a switch in mechanism that is dependent on the reaction temperature.

The phenoxide analogue Fe^III(OH)(OPh^p-NO2) (4) was reacted with the same tertiary carbon radical (p-OMe-C₆H₄)₃C· to examine the reactivity of an O donor versus an S donor. In contrast to the sulfur analogue, only ·OH transfer was seen by Mössbauer spectroscopy (Figure 5) and NMR spectroscopy at both 23 and −35 °C. The relative reaction rates of 3 and 4 with (p-OMe-C₆H₄)₃C· were assessed through competition experiments in which a 1:1 mixture of 3 and 4 was reacted with (p-OMe-C₆H₄)₃C· (1 equiv) at either 23 or −35 °C. The reaction at 23 °C led to the formation of only the Fe^II(OAr) product 2 along with unreacted 3, showing that · OH transfer is significantly faster from 4 than from 3.

Figure 5.

Zero-field ⁵⁷Fe Mössbauer spectra (80 K) of (top) 2, (middle) 4, and (bottom) 4 + (p-OMe-C₆H₄)₃C· at 23 °C. Isomer shift (Δ) and quadrupole splitting (|ΔE_Q|) values are given in mm s⁻¹.

However, reaction at −35 °C led to a small amount of (p-OMe-C₆H₄)₃C−SPh^p-NO2 (by TLC) and 5 (by ¹H NMR), together with the major product 2. These results show that ArS· transfer from 3 to the carbon radical has become competitive with ·OH transfer from 4 but still remains slower overall. We conclude that the relative rates of rebound follow the trend k_OH(4) > k_OH(3) at 23 °C and k_OH(4) > k_SAr(3) at −35 °C. The results from the reactions with (p-OMe-C₆H₄)₃C· are summarized in Scheme 4.

Scheme 4.

Reaction of 4 with Tertiary Carbon Radical

In summary, a new series of iron(II) and iron(III) complexes are described. The iron(III) complexes provide a platform to examine the competition between cis-ligated OH versus SAr/OAr groups in reactions with carbon radicals. Reaction of Fe^III(OH)(SAr) with (p-OMe-C₆H₄)₃C· at 23 °C gives (p-OMe-C₆H₄)₃COH. However, the same reaction at −35 °C leads to the thioether (p-OMe-C₆H₄)₃CSAr. In contrast, Fe^III(OH)(OAr) produces only (p-OMe-C₆H₄)₃COH independent of temperature.

Transfer of ·OH from either 3 or 4 is likely thermodynamically favored because of the relative strength of the C−OH bond being formed, compared with the alternative C−OAr or C−SAr bonds.⁶⁷ Consistent with the expected thermodynamic trend in bond strengths, complexes 3 and 4 produce only (p-OMe-C₆H₄)₃COH at 23 °C. However, the same reaction for 3 at −35 °C preferentially produces (p-OMe-C₆H₄)₃CSAr. The equatorial Fe−S bond in 3 is significantly elongated and therefore weakened, resulting in a lower kinetic barrier for sulfur transfer.

The kinetic versus thermodynamic pathways are illustrated in the qualitative reaction coordinate diagram in Figure S46. This analysis implies that formation of the thioether is reversible at 23 °C, which is supported by the reductive cleavage of thioether bonds.^68,69 The Fe^III−OAr bond in 4, on the other hand, does not show any significant elongation, which is consistent with the lack of phenoxyl transfer.

Complex 3 is, to our knowledge, the first synthetic model of the ferric hydroxothiolate intermediate in IPNS. The overall reactivity of 3 is similar to that revealed by calculations on IPNS, which indicate that sulfur transfer is kinetically favored whereas hydroxylation is thermodynamically controlled.¹⁵ These results show that the inherent electronic and structural features of a non-heme Fe center can significantly influence the outcome of the rebound step without contributions from an enzyme pocket or substrate orientation effects.