Artificial Iron Proteins: Modeling the Active Sites in Non-Heme Dioxygenases

Abstract

An important class of non-heme dioxygenases contain a conserved Fe binding site that consists of a 2-His-1-carboxylate facial triad. Results from structural biology show that in the resting state, these proteins are six-coordinate with aqua ligands occupying the remaining three coordination sites. We have utilized biotin-streptavidin (Sav) technology to design new artificial Fe proteins (ArMs) that have many of the same structural features found within active sites of these non-heme dioxygenases. A Sav variant was isolated that contains the S₁₁₂E mutation which installed a carboxylate side-chain in the appropriate position to bind to a synthetic Fe^II complex confined within Sav. Structural studies using X-ray diffraction (XRD) methods revealed a facial triad binding site that is composed of two N-donors from the biotinylated ligand and the monodentate coordination of the carboxylate from S₁₁₂E. Two aqua ligands complete the primary coordination sphere of the Fe^II center with both involved in hydrogen bond networks within Sav. The corresponding Fe^III protein was also prepared and structurally characterized to find a six-coordinate complex with two exogenous acetato ligands. The Fe^III protein was further shown to bind an exogenous azido ligand through replacement of one acetato ligand. Spectroscopic studies of the ArMs in solution support the results found by XRD.

Synopsis and Graphical Abstract

Biological O₂ activation is often accomplished with dioxygenases that contain non-heme iron centers with a common 2-His-1-carboxylate endogenous binding site for iron. We have engineered an artificial protein to model this site using biotin-streptavidin technology. Structural and spectroscopic studies support formation of an Fe(II) artificial protein with similar features as those in the native proteins, including hydrogen bonds within the secondary coordination sphere. This work highlights the utility of embedding metal complexes with protein hosts.

Introduction

Mononuclear non-heme Fe oxygenases are enzymes that catalyze a wide range of oxidative transformations using dioxygen as the terminal oxidant.^1–7 Many of these enzymes share a common Fe binding site that is composed of side chains from two histidine and either a glutamate or aspartate amino acid residues.^8–10 This endogenous binding site coordinates an Fe center in a facial manner and is referred to as the 2-His-1-carboxylate facial triad.¹¹ Information from structural biology indicates that in the FeII resting state of these proteins, and in the absence of any exogenous cofactors, the remaining coordination sites are occupied by water molecules (Figure 1).^12–14 A common feature of these structures is the presence of an intramolecular hydrogen bond (H-bond) between one of the aqua ligands and the carbonyl group of the coordinated carboxylate ligand.

Figure 1.

Fe binding site in cephalosporin synthase showing 2-His-1-carboxylate facial triad (A, PDB: 1RXF) and ChemDraw version of the active site (B).

Attempts to model this type of coordination environment include synthetic systems that catalyze the epoxidation and/or cis-dihydroxylation of olefins with H₂O₂ and O₂,^15,16 and the intramolecular arene hydroxylation of benzoylformate with O₂.^17–19 However, it has proven challenging to develop synthetic complexes whose coordination spheres resemble those found within these proteins. For instance, reported complexes that emulate the facial triad have been derived from sterically hindered N-donors and terphenylcarboxylate ligands²⁰ and tripodal ligands^21–30 but lack the aqua ligands found within protein active sites. In addition, these systems do not replicate the control over the secondary coordination spheres that is seen in metalloproteins.

We have utilized a complementary approach to develop bio-relevant Fe active sites that combine synthetic inorganic chemistry and protein engineering to design artificial metalloproteins (ArMs).^31–37 The approach capitalizes on biotin-streptavidin (Sav) technology to predictably insert biotinylated synthetic complexes into variants of Sav.^38–41 We reasoned that these protein hosts could provide ligands within the primary coordination sphere and simultaneously influence the secondary coordination sphere to better emulate natural active sites. Moreover, these constructs would be stable in aqueous solution allowing for water to act as a ligand, as is found in natural metalloproteins. Our previous work has shown how the primary coordination sphere can be manipulated to produce ArMs that model the properties in cupredoxins through engineering an endogenous thiolate ligand into Sav that forms a strong covalent bond to the Cu centers.⁴² Moreover, we have demonstrated that Sav can regulate the secondary coordination sphere around a Cu–hydroperoxido complex using intramolecular H-bonding networks.⁴³ In this report, we describe the design and development of new artificial Fe proteins that simulate the 2-His-1-carboxylate facial triad active site in non-heme monooxygenases. Both primary and secondary coordination spheres of the Fe centers are influenced by the Sav host to produce the new ArMs in which intramolecular H-bonds play a key structural role.

Results and Discussion

Design Concepts, Preparations, and Solution Properties of the Fe ArMs.

Previous work in our group found that binding of an endogenous ligand from Sav to a synthetic metallocofactor is possible. Our work on Type 1 Cu sites showed how Cu–S_thiolate coordination could be achieved by using a relatively short ethylene linker between the biotin and synthetic complex that positioned the Cu center close to the residue at position 112.⁴² We reasoned that Fe complexes prepared from the similar construct, bis(2-pyridylmethyl)amine (Figure 2: biot-et-dpa), may also place the metal center near this residue. Introduction of a S₁₁₂E mutation in Sav would produce a variant that could then promote Fe–O_carboxylate bond formation. In addition, the Sav host was re-engineered to incorporate two additional mutations, K₁₂₁A/E₁₀₁Q, as described previously.⁴⁴ The parent protein host is referred to as 2xm-Sav and the specific protein host used to develop mononuclear Fe sites is denoted 2xm-S₁₁₂E-Sav.

Figure 2.

Biotinylated ligand used in this study.

The biotinylated biot-et-dpa construct was prepared via a 3-step route from dpa (Scheme S1) using a modified literature procedure.⁴² Formation of the Fe^II complex was achieved by treating biot-et-dpa with FeBr₂ in CH₃CN under inert atmosphere to afford [Fe^II(biot-et-dpa)(OH₂)₂]Br₂. Synthesis of the Fe^III complex was accomplished by allowing biot-et-dpa to react with FeCl₃·6H₂O in EtOH to yield [Fe^III(biot-et-dpa)(OH₂)₃]Cl₃. The ArMs containing the Fe^II complex, [Fe^II(biot-et-dpa(OH₂)₂(κ¹-O_E112)⊂2xm-S₁₁₂E-Sav] (1, eq 1), and the corresponding Fe^III

$$\left[{\text{Fe}}^{\text{II}}(\text{biot-et-dpa}){\left({\text{OH}}_2\right)}_2\right]{\text{Br}}_2\underset{\underset{{\text{N}}_{\text{2}},\text{r}\mathrm{t}}{\text{phosphate buffer}}}{\overset{2{\text{xm-S}}_{112}\text{E-Sav}}{\to }}\underset{(1)}{\left[{\text{Fe}}^{\text{II}}(\text{biot-et-dpa}){\left({\text{OH}}_2\right)}_2\left({\text{K}}^1{\text{-O}}_{{\text{E}}_{112}}\right)\subset 2{\text{xm-S}}_{112}\text{E-Sav}\right]}$$ (1)

complex, [Fe^III(biot-et-dpa(X)₂(κ¹-O_E112)⊂2xm-S₁₁₂E-Sav] (2, eq 2), were prepared by incubating a water solution of the complexes with 2xm-S₁₁₂E-Sav at pH 8 in phosphate buffer (50 mM) or at pH 6 in acetate buffer (100 mM). The 2xm-S₁₁₂E-Sav variant is a homotetramer and to ensure complete binding of the Fe complexes a 2-(4’-hydroxyazo-benzene)benzoic acid (HABA) assay was performed on each ArM.⁴⁵ These studies indicated 4:1 ratio of biotinylated Fe complex to protein host for each ArM that confirmed complete occupancy within the Sav homotetramer (Figure S1).

$$\left[{\text{Fe}}^{\text{III}}(\text{biot-et-dpa}){\left({\text{OH}}_2\right)}_3\right]{\text{Cl}}_3\underset{\underset{\text{rt}}{\text{acetate buffer}}}{\overset{2{\text{xm-S}}_{112}\text{E-Sav}}{\to }}\underset{\underset{{\text{X=OH}}_{\text{2}},{\text{OAc}}^{-}}{(2)}}{\left[{\text{Fe}}^{\text{III}}(\text{biot-et-dpa}){(\text{X})}_2\left({\text{k}}^1{\text{-O}}_{{\text{E}}_{112}}\right)\subset 2{\text{xm-S}}_{112}\text{E-Sav}\right]}$$ (2)

Electronic absorbance and EPR spectroscopies were used to characterize 1 and 2 in solution. For 1, one feature at λ_max (ε_M, M⁻¹cm⁻¹) = 375 (700) nm was observed with the band arising from a pyridine-to-FeII charge transfer transition (Figure S2A). For 2, three absorbance bands were found at λ_max (ε_M, M⁻¹cm²²¹²¹) = 350 (2500), 485 (220), and 650 (50) nm (Figure S2B). The parallel-mode, X-band EPR spectrum for 1 displayed a broad signal with a g-value at 8.8 that is consistent with a high spin Fe^II species with an S = 2 spin ground state (Figure S3A). The perpendicular-mode, X-band EPR spectrum of 2 revealed weak features with g-values at 9.1 and 4.3 which are consistent with the presence of high-spin Fe^III species in a rhombic coordination geometry (Figure S3B). Further details on the magnetic properties of these proteins will be described in a subsequent report.

Structural Characterization of 1.

A significant advantage of utilizing biotin-Sav technology in the design of ArMs is the strong tendency of Sav to form crystals, which gives us the ability to obtain molecular structures using X-ray diffraction (XRD) methods. In general, the high affinity of biotin for Sav allows for in crystallo preparation of Fe ArMs by incubation of apo-Sav crystals with a solution of biotinlyated Fe complexes. For biotinylated complexes, this soaking method should generate single crystals of Sav that contain one complex per subunit. An added benefit of this approach is that we can explore chemistry performed within a single crystal that can be compared to similar processes in solution.

Single crystals of 1 were prepared by soaking crystals of apo-2xm-S₁₁₂E-Sav with [Fe^II(biot-et-dpa)(OH₂)₂]Br₂. Its structure was solved to a resolution of 1.47 Å to reveal a mononuclear Fe complex immobilized within each subunit in which the Fe center is coordinated by O-atoms from the carboxylate group of S₁₁₂E bound in a κ¹ fashion and two aqua ligands (Figure 3, Tables 1 & 2, Tables S1 & S2). Nitrogen donors N1 and N2 from dpa also bind to the Fe center with Fe–N1 and Fe–N2 bond lengths of 1.94 Å and 2.32 Å. Our structure showed two conformations of the pyridine ring containing N3 (Figure 3A). The major complex (1a) was modelled to 70% occupancy and it was found that this pyridine ring is not coordinated (unbound state); instead N3 is H-bonded to the aqua ligand of O3. The resultant complex is 5-coordinate with a square pyramidal geometry and a τ-value of 0.05 (Figure 3B). The minor complex (1b) that was modelled to 30% occupancy is 6-coordinate in which N3 is bound to the Fe center (bound state) with an Fe–N3 bond length of 2.21 Å and N2–Fe–N3 and N3–Fe–O3 bond angles of 107° and 79° (Figure 3C). The carboxylate group at S₁₁₂E also has two conformations. In this case, the major confirmation placed O1 in a position to coordinate to the Fe center and was modelled to 80% occupancy (Figure S4A). The minor species has the carboxylate not coordinated to the Fe center and has a 20% occupancy (Figure S4A).

Figure 3.

Overlay of the molecular structures of 1a (grey) and 1b (black) (A, PDB: 6UIY). The individual molecular structures of 1a (B) and 1b (C) and a partial space-filling representation of 1a highlighting the π-stacking interaction of the biotinylated Fe^II complex that includes one of its pyridine rings with W_120’ from the neighboring subunit (D, dark grey). Fe ions are colored in orange.

Table 1.

Selected bond lengths (Å) and angles (˚) for the ArMs from XRD measurements.

Bond lengths and angles	1a	1b	2	2–N3
Fe–O1	2.16	2.16	2.11	2.35
Fe–O2	2.17	2.17	2.17	2.00
Fe–O3	2.22	2.22	2.15	–
Fe–N1	1.94	2.31	2.15	2.27
Fe–N2	2.32	2.36	2.23	2.52
Fe–N3	–	2.21	2.18	2.27
Fe–N4	–	–	–	2.15
N1–Fe–O2	167	167	167	169
N1–Fe–N2	79	67	71	79
N2–Fe–O3	170	164	165	–
O2–Fe–O3	91	91	96	–
O1–Fe–N1	100	91	90	95
O1–Fe–N2	82	78	79	75
O1–Fe–N3	–	166	164	146
O1–Fe–O2	91	91	96	87
O1–Fe–O3	92	92	95	–
N1–Fe–N3	–	79	76	83
N2–Fe–N3	–	107	104	71
N3–Fe–O2	–	101	100	89
N3–Fe–O3	–	79	78	–
O1–Fe–N4	–	–	–	110
O2–Fe–N4	–	–	–	89
N1–Fe–N4	–	–	–	101
N2–Fe–N4	–	–	–	175
N3–Fe–N4	–	–	–	103
Fe–N4–N5	–	–	–	120

Table 2.

Selected H-bonds (Å) for the ArMs from XRD measurements.

Distances	1	2	2–N3
N3···O3	3.51	–	–
O3···Wat22	2.54	–	–
Wat22···N49	3.00	–	–
Wat22···Wat191	3.05	–	–
Wat191···Wat190	2.34	–	–
Wat190···O4	3.12	–	–
O4···O2	2.71	–	–
O6···N49	–	2.97	–
N4···Wat1	–	–	2.88
Wat1···N49	–	–	3.17

The molecular structure of 1a revealed an Fe site with a facial arrangement of two N-atom donors from dpa and the O-atom from the residue of S₁₁₂E that is similar to the 2-His-1-carboxylate facial triad observed in all non-heme monooxygenases. In addition, the κ¹-coordination of the glutamate residue in 1a and 1b leaves O4 positioned to H-bond with the aqua ligand of O2 as gauged by the O4…O2 distance of 2.71 Å. This type of monodentate carboxylate coordination to Fe with an additional intramolecular H-bond is seen in several native non-heme enzymes, such as deacetoxy-cephalosporin C synthase (Figure 1A),¹² phenylalanine hydroxylase binary enzyme (PDB: 1J8U),⁴⁶ and soybean lipoxygenase (PDB: 1JNQ).¹⁴ These enzymes have O…O distances that range between 2.5 and 3.6 Å. The structures of the Fe^II complex in 1 and Fe^II sites in natural enzymes also share the coordination of multiple coordinated water molecules to their Fe^II centers.

The 2xm-S₁₁₂E-Sav host in 1 also promotes other non-covalent interactions with the artificial Fe cofactor. For instance, the aqua ligand with O2 has an additional H-bond that involves a structural water molecule (Wat22) that is also H-bonded to the residue of N₄₉ and the backbone carbonyl of A₈₆. This aqua ligand thus participates in two H-bonds which could account for the slightly longer Fe–O3 bond length of 2.22 Å when compared to that found for the Fe–O2 bond (2.17 Å). These water molecules are also part of a larger H-bonding network that extends from N3 to the glutamate 112 side-chain and help in stabilizing its coordination to the Fe center (Figures 3B, C). Furthermore, the non-coordinating pyridine ring participates in an additional π-stacking interaction with the residue of W₁₂₀ from the adjacent subunit with a centroid_py– centroid_trp distance of 4.86 Å that likely contributes to the stabilization of this conformation (Figure 3D).

In synthetic complexes, dpa binds solely as a tridenate ligand to Fe^II/III centers. In fact, analysis of all the structures in the Cambridge Structural Database⁴⁷ for complexes containing Fe(dpa) fragments (over 300 entries) found that both pyridine N-atoms are bound to the Fe center in the crystalline phase. Confinement of [Fe^II(biot-et-dpa)(OH₂)₂]Br₂ within 2xm-S₁₁₂E-Sav appears to promote the unusual bidentate coordination of dpa. The trapping of this Fe^II species with a non-coordinating pyridine ligand highlights the importance of the local environment around a metal center and how it can influence the structure of the embedded metal complex.

Structural Characterization of biot-et-dpa ⊂ 2xm-S₁₁₂E-Sav.

The unusual molecular structure of the Fe complexes in 1 prompted us to examine the structure of 2xm-S₁₁₂E-Sav containing just biot-et-dpa. We were interested in determining how the dpa ligand was positioned within the protein host in the absence of the Fe center and to examine if any structural differences occurred upon coordination to Fe. The structure of biot-et-dpa ⊂ 2xm-S₁₁₂E-Sav was solved to a resolution of 1.35 Å and both the dpa ligand and carboxylate group at position 112 were modelled with single conformations at 100% occupancy (Figure 4A, Tables S1 & S2). Both N1 and N2 are centered near the carboxylate group at position 112 and the pyridine ring containing N3 adopts a nearly identical confirmation to what is found in 1a, including the same π-stacking interaction with W120 (Figure 4C). Additional non-covalent interactions are present in the form of an extended H-bonding network that includes N1 and N3, a series of water molecules, and the side chains of N₄₉ and S₁₁₂E (Figure 4A). Atom N3 thus forms a H-bond as in 1a but with a structural water molecule rather than an aqua ligand. Notice that the H-bonding network also links N3 to the carboxylate group at S₁₁₂E which influences its orientation.

Figure 4.

The molecule structure of biot-et-dpa ⊂ 2xm-S₁₁₂E-Sav (A, PDB: 6UIU). The 2F₀-F_c electron density map (grey, contoured at 1σ) is highlighted in B with same labelling as in A. An overlay of biot-et-dpa ⊂ 2xm-S₁₁₂E-Sav (black) and 1a (grey) (C) that shows the H-bonding networks associated with biot-et-dpa ⊂ 2xm-S₁₁₂E-Sav (black dashes) and 1a (grey dashes). The number schemes in B & C are the same as in A.

The structure found of biot-et-dpa ⊂ 2xm-S₁₁₂E-Sav provide insights on how the protein host influences both the primary and secondary coordination spheres in 1 (Figures 4A, B). The glutamate and pyridine with N3 have two different conformations in the structure of 1 (Figure S4A). For the glutamate, the minor state in 1 has the carboxylate occupying the same position as that observed in biot-et-dpa ⊂ 2xm-S₁₁₂E-Sav that places O-atoms greater than 3.0 Å from the Fe center. In its bound state, the carboxylate group rotates toward the dpa ligand enabling O1 to coordinate to the Fe center. Moreover, the placement of unbound pyridine in 1a is only displaced by 0.6 Å from the position occupied by the same ring in biot-et-dpa ⊂ 2xm-S₁₁₂E-Sav. This change appears to be enough to promote intramolecular H-bonding with the aqua ligand of O3 but not enough to coordinate to the Fe center (Figure 4C). For the bound state in 1b, there is a significant change in the position of the pyridylmethyl arm that results in N3 moving over 2.0 Å in order to coordinate to the Fe_II center. There is also an extended H-bonding network like the one found in 1 that includes N3, several water molecules, and the S112E sidechain (Figure 4C).

Structural Characterization of 2.

To evaluate the effects of Fe oxidation on the molecular structure of the active site, we prepared single crystals of 2 by soaking crystals of apo 2xm-S₁₁₂E-Sav with [Fe^III(biot-et-dpa)(OH₂)₃]Cl₃. Analysis of crystals that diffracted to a 1.40 Å resolution again revealed the formation of a six-coordinate mononuclear Fe complex within each subunit of Sav (Figure 5, Tables 1 & 2, Tables S3 & S4). The primary coordination sphere is composed of an N₃O₃ donor set with an O-atom donor from the residue of S₁₁₂E that is bound in a similar κ¹ manner as observed in 1. The other O-atom donors are from acetate ions (from the crystallization condition) that also bind as monodentate ligands and occupy the same coordination sites as the aqua ligands in 1. The dpa ligand binds facially to the Fe^III center and all three N-donors are coordinated. Notice that the structural water molecule (Wat22) that was found in 1 is not present in 2 which allows O6 of the acetato ligand to H-bond with the amide residue of N₄₉, with a N₄₉…O6 distance of 2.97 Å.

Figure 5.

The molecule structure of 2 (A, PDB: 6UIO) and an overlay of the structure collected using XFEL (B, carbons colored in teal, PDB: 6US6). The number scheme in B is the same as in A.

To evaluate if photoreduction was occurring on the crystals of 2 and the extent of change that would cause on the resultant structure, XRD data was also collected using an X-ray free electron laser (XFEL). This technique utilizes femtosecond X-ray pulses to collect XRD data in a serial fashion with each diffraction image obtained from a new crystal.^48–51 Detection occurs before the diffusion of radicals or solvated electrons that often causes X-ray-induced changes in protein structures. Therefore, this method has provided structural data for metal centers in a non-damaged (that is, not photoreduced) state for several metalloproteins.^48–51 We were able to obtain a molecular structure of 2 that was resolved to 1.50 Å (Tables S3 & S4). The structures of 2 obtained from frozen crystals by XRD and room temperature crystals with the XFEL produced the molecular structures that were identical (Figure S4D) with an average difference in the Fe– N/O bond lengths of 0.03 Å. Notice that there is a slight difference in the orientations of the ligands in the two structures (Figure 5B) which could be caused by the different temperatures that were used for data collection. Nevertheless, these results suggest that the structure of 2 obtained using synchrotron radiation at cryogenic temperatures had only limited radiation damage.

X-ray Absorption Spectroscopy.

The properties of the Fe center in 1 were further probed using X-ray absorption spectroscopy (XAS) to gain a better understanding of the structure of protein-embedded complex in solution (see SI, Figures S5–S7). X-ray absorption near edge structure (XANES) analysis provided an energy edge for 1 at 7122.4 eV (Figure S7), which is consistent with an Fe^II center.^52–54 Extended X-ray absorption fine structure (EXAFS) analysis revealed that for 1 there are five O/N scatterers at a distance of 2.15 Å (Table 3, Tables S5-S6). Two additional shells from the Fe center were fit to four carbon scatterers at 3.01 Å and one at 2.63 Å. The longer distance carbon scatterers could include the carbonyl C-atom of S₁₁₂E side-chain which is found at Fe…C distance of 3.12 Å by XRD. The shorter distance Fe…C distance of 2.60 Å matched that found by XRD for the Fe center to the methylene carbon of the unbound pyridylmethyl group. Taken together, these data suggest that the predominate species in solution for 1 is also the unbound state in which one pyridine ring of dpa is not coordinated (Figure 3B).

Table 3.

Comparison of metrical parameters for 1 obtained by XAS and XRD.

Bond	No. of Scatterers	XRD (Å)	XRD (Å)
Average Fe–O/N	5	2.15	2.16
Average Fe···C	4	3.01	3.01
Fe···Cmethylene	1	2.63	2.60

Azide Binding to 2.

We chose to examine the binding of azide ions to 2 in order to investigate how exogenous ligands bind to the Fe center. In solution studies, 2 was treated with a 100-fold excess of NaN₃ in acetate buffer pH 6 to afford a new spectrum with a band at a λ_max (ε_M, M⁻¹cm⁻¹) = 405 (3000) nm that is assigned to an azido-to-Fe^III charge transfer transition (Figure 6A, S8).⁵⁵ Parallel in crystallo studies were performed that involved soaking crystals of 2 in a 100 mM solution of NaN3 to afford 2–N₃, whose structure was obtained to a resolution of a 1.85 Å (Figure 6B,C). Analysis of the molecular structure of 2–N₃ revealed a monomeric Fe complex where the dpa ligand, the carboxylate side chain of S₁₁₂E, and one acetato ligand containing O3 are coordinated in the same positions as found in 2 (Figures 6B, C). However, the second acetato ligand is absent. In its place is electron density that we have modelled as a terminal azido ligand with an Fe–N4 bond length of 2.15 Å. The azido ligand has an Fe–N4–N5 bond angle of 120.1° that is comparable to bond angles found in previously reported Fe^III–N₃ complexes.⁵⁶ The presence of the azido ligand in 2–N3 causes an 0.29 Å elongation of Fe–N2 bond distance compared to what is observed in 2, which is consistent with an azido ligand having a stronger trans-influence than an acetate ion. The structural water molecule Wat1is modelled at 100% occupancy and is part of an H-bonding network that includes the azido ligand. The H-bonding network extends from the proximal N4 of the azido ligand and includes Wat1, and the amide residue of N₄₉ with N4·Wat1 and Wat1·N₄₉ distances of 2.88 and 3.17 Å, respectively (Figure 6B).

Figure 6.

Electronic absorbance spectra of 2 (black) and 2–N₃ (grey) recorded at 4°C in 100 mM acetate buffer pH 6 (A). Molecular structure of 2–N₃ (B, PDB: 6UIZ), and the 2F₀-F_c electron density map (grey, contoured at 1σ), F₀-F_c omit map (green, contoured at 3σ) and anomalous difference density (red, contoured at 5 σ) are shown in C. The numbering scheme in C is the same as in B.

Summary and Conclusions

The results from these studies on our new Fe proteins further illustrates the versatility of using biotin-Sav technology to engineer ArMs. Our approach leverages the proper placement of synthetic complexes proximal to endogenous functional groups to produce new biomimetic constructs. We utilized knowledge gained from our previous work on Cu proteins, which revealed that an ethylene linker between a metal complex and the biotin positioned the Cu center near the side-chain at position 112. Following this design concept, the Sav protein was re-engineered to produce the host 2xm-S₁₁₂E-Sav which included a glutamate at position 112. When matched with the biotinylated complexes, [Fe^II(biot-et-dpa) (OH₂)₂]Br₂ and [Fe^III(biot-et-dpa)(OH₂)₃]Cl₃, ArMs were produced with properties that resemble those found in mononuclear non-heme iron enzymes. The positioning of the artificial cofactors within 2xm-S₁₁₂E-Sav allow for monodentate coordination of the carboxylate side-chain in an identical manner as found in the active sites of Fe monooxygenases. Moreover, the unusual structural finding in 1a, in which one pyridine group of the dpa ligand is not coordinated, produced an Fe binding that is similar to the 2-His-1-carboxylate facial triad found in the natural metalloproteins. The observation that only one pyridine group is coordinated to the Fe center in 1a is unusual and has not been structurally observed in other Fe complexes with dpa. We attribute this difference to the confinement of the complex within a protein host that help regulate the secondary coordination sphere around the Fe complexes. In fact, there are several non-covalent interactions to the complexes that include those with a pyridine ring of the dpa ligand that stabilizes the unbound state found in 1a. An additional H-bond was found between an aqua ligand and the noncoordinating O-atom of a glutamate side-chain, an interaction that is found in many non-heme Fe enzymes.

The structure of 2 also revealed a κ¹–coordination to the FeIII center of the carboxylate at position 112. In 2, the Fe complex is 6-coordinate and the aqua ligands have been replaced with acetato ligands to produce a primary coordination sphere that contains three monodentate carboxylate ligands. Again, H-bonds involving the carboxylate ligands appear to help stabilize this unusual coordination environment. Substitution of the one of the acetato ligands was achieved by treating 2 with excess azide ion to produce 2–N₃. The structure of 2–N₃, along with those of 1 and 2, also illustrate the importance of the amide side-chain of N49 in maintaining H-bonding networks around the complexes. Taken together, our findings highlight how the synergistic design of Sav and an artificial metallocofactor allow for the control of both the primary and secondary coordination spheres of a confined metal complex within a protein host.

Experimental

General Methods.

All commercially available reagents were obtained of the highest purity and used as received. Acetonitrile (CH3CN), Ethanol (CH₃CH₂OH) and diethylether were degassed with argon and dried by vacuum filtration through activated alumina according to the procedure by Grubbs.⁵⁷ Triethylamine was distilled from KOH. Thin-layer chromatography (TLC) was performed on Merck 60 F254 aluminum-backed silica plates or Merck 60 F254 glass-backed basic aluminum oxide plates. Eluted plates were visualized using UV light. Silica or basic alumina gel chromatography was performed with the indicated solvent system using Fisher reagent silica gel 60 (230–400 mesh) or Sigma reagent Brockmann 1 basic aluminum oxide 58 (150 mesh). Biotin pentafluorophenol ester (biot-PFP)⁵⁸ and di-(2-picolyl)amine⁵⁹ were prepared according to literature procedures.

2-(2-(bis(pyridin-2-ylmethyl)amino)ethyl)isoindoline-1,3-dione (1).⁴²

Dpa (2.03 g, 10.0 mmol), bromoethylphthalimide (2.93 g, 11.0 mmol), KI (0.339 g, 0.002 mol) and K₂CO₃ (5.6 g, 41 mmol) were dissolved in 60 mL acetonitrile and refluxed for 24 hours. The solution was cooled to room temperature, filtered, and reduced to dryness. The maroon-brown residue was dissolved in 30 mL dichloromethane and washed with 3 × 30 mL aqueous NaHCO₃ (with a small addition of brine) and 2 × 30 mL water. The maroon-brown organic layer was acidified with 20 mL 12 mM HCl dissolved in 10 mL water and washed with 5 × 30 mL dichloromethane. The aqueous layer was carefully neutralized with solid NaHCO₃ and extracted with 4 × 30 mL dichloromethane. The solution was dried with MgSO₄, filtered and reduced to dryness to yield a red-brown oil. The crude product was purified via column chromatography with silica gel and methanol:dichloromethane:Et₃N (9:90:1) or with basic alumina and methanol:dichloromethane (10:90) as the eluent to yield pure 1 as a yellow oil (1.68 g, 44%–50%). ¹H (500 mHz, CDCl₃) δ 8.44 (d, 2H), 7.82 (dd, 2H), 7.74 (dd, 2H) 7.42 (t, 2H), 7.34 (d, 2H), 7.06 (t, 2H), 3.85 (m, 6H), 2.86 (t, 2H). MS (ESI, MeOH) m/z calcd C₂₂H₂₀N₄O₂ [M + (Na⁺)] 395.15, found 395.09.

N,N-bis(pyridin-2-ylmethyl)ethane-1,2-diamine (2).⁴²

Hydrazine monohydrate (1.10 mL, 0.020 mol) and 1 (1.7 g, 5.0 mmol) were dissolved in 60 mL ethanol and refluxed under N₂ for 3 hours. The phthalhydrazide byproduct precipitated as a white solid after 15 minutes of reflux. The solution was filtered to remove the phthalhydrazide and washed with 3 × 5 mL chloroform. The solution was reduced to dryness and the yellow oily residue was dissolved in 40 mL chloroform and 40 mL of 1 M NaOH. The aqueous layer was extracted with 3 × 40 mL chloroform, dried over MgSO₄, filtered, and reduced to dryness. The product 2 was recovered as a yellow oil (0.80 g, 77%). ¹H (500 mHz, CDCl₃) δ 8.53 (d, 2H), 7.66 (t, 2H), 7.50 (d, 2H) 7.15 (t, 2H), 3.85 (s, 4H), 2.80 (t, 2H), 2.66 (t, 2H). MS (ESI, MeOH) m/z calcd C₁₄H₁₈N₄ [M + (H⁺)] 243.16, found 243.11.

Biot-ethyl-dpa (3).⁴²

A solution of biot-PFP (1.30 g, 3.01 mmol), 2 (0.85 g, 3.0 mmol), and triethylamine (0.39 g, 3.0 mmol) in 20 mL DMF was allowed to stir overnight. The DMF was removed under vacuum to yield a sticky tan residue. The residue was triterated with diethylether until a free-flowing solid formed (3–7 days). The light tan solid was filtered, washed with diethylether, and dried under vacuum (1.49 g, 91%). The solid 3 was stored under an inert atmosphere. ¹H (500 mHz, DMSO) δ 8.44 (d, 2H), 7.75 (br, 1H), 7.71 (t, 2H) 7.50 (d, 2H), 7.21 (t, 2H), 6.40 (s, 1H), 6.33 (s, 1H), 4.27 (t, 1H), 4.07 (t, 1H), 3.72 (s, 4H), 3.16 (q, 2H), 3.04 (q, 1 H), 2.78 (dd, 1 H), 2.55 (s, 1H), 2.46 (t, 2H), 2.02 (t, 2H), 1.45 (m, 6H). MS (ESI, MeOH) m/z calcd C₂₄H₃₂N₆O₂S [M + (Na⁺)] 491.22, found 491.17.

Preparation of Metal Complexes

[Fe^II(biot-ethyl-dpa)(OH₂)₂]Br₂.⁶⁰

[Fe^II(Biot-ethyl-DPA)(OH₂)₂]Br₂ was prepared by addition of 3 (53 mg, 0.24 mmol) in 10 mL acetonitrile to FeBr₂. A pale yellow solid immediately precipitated from the solution. The suspension was allowed to stir under N₂ for 30 mins, after which the pale-yellow solid was collected via filtration, washed with CH₃CN:diethyl ether (1:1), and dried under vacuum (0.125 g, 75%). The solid was stored under an inert atmosphere. HR-MS (ESI, 1% DMF:MeCN) m/z calcd for C₂₄H₃₂FeN₆O₂SBr [M – (Br–)] 603.08, found 603.08. Elem. Anal. Calcd for (C₂₄H₃₆N₆SO₄FeBr₂): C, 40.02; H, 5.04; N, 11.67. Found: C, 39.67; H, 4.52; N, 11.38.

[Fe^III(biot-ethyl-dpa)(OH₂)₃]Cl₃.⁶¹

[Fe^III(Biot-ethyl-DPA)(OH₂)₃]Cl₃ was prepared by addition of 3 (0.09 g, 0.2 mmol) in 3 mL ethanol to FeCl₃·6H₂O (0.10 g, 0.38 mmol) in 3 mL ethanol. A yellow solid immediately precipitated from the solution. The suspension was allowed to stir under N₂ for 15 minutes, after which the yellow solid was isolated via filtration. The solid was washed with 10 mL chilled ethanol:ether (4:1) and dried under vacuum (0.10 g, 86%). The solid was stored under an inert atmosphere. HR-MS (ESI, MeOH) m/z calcd for C₂₄H₃₂Cl₂FeN₆O₂S [M – (Cl–)] 594.10, found 594.08. Elem. Anal. Calcd for (C₂₄H₃₈N₆SO₅FeCl₃): C, 42.09; H, 5.59; N, 12.27. Found: C, 42.28; H, 5.12; N, 12.23.

Analytical and Spectroscopic Methods.

HABA Titrations.

To 2.4 mL of 8 μM Sav in 200 mM sodium phosphate buffer at pH 7 was added 300 μL of a 10 mM 2-(4΄-hydroxyazobenzene)benzoic acid (HABA) in 200 mM phosphate buffer pH 7. After 5 min equilibration, the absorbance at 506 nm was recorded. A solution of 1 mM Fe complex in nanopure water was added in 4–20 μL portions until approximately 4 equivalents had been added. The absorbance at 506 nm was recorded until no further changes in intensity were observed.

Electronic Absorption Studies.

A solution of lyophilized protein (50–250 μM) was prepared in nanopure water. Four equivalents of Fe complex (0.2–1 mM) in nanopure water were added to the protein solution. Samples were diluted to their final concentrations in a final volume of 500 uL containing 50 mM potassium phosphate or 100 mM sodium acetate buffer at the indicated pH.

EPR Studies.

A solution of protein (500 μM) was prepared in nanopure water. Four equivalents of Fe complex (2 mM) in nanopure water were added to the protein solution. Samples were diluted to their final concentrations in a final volume of 300 uL containing 50 mM potassium phosphate or 100 mM sodium acetate buffer at the indicated pH. The sample was transferred to an EPR tube, frozen at 77 K in liquid nitrogen and run at 10 K.

XAS Studies.

A solution of protein (750 μM) was prepared in nanopure water. Four equivalents of Fe complex (3 mM) in nanopure water were added to the protein solution. Samples were diluted to their final concentrations in a final volume of 250 uL with nanopure water. The sample was then poured in liquid ethane to freeze and packed as a solid into a pre-cooled XAS sample holder.

Protein Crystallization

Crystallization of [Fe^II(biot-et-dpa(OH₂)₂(κ¹-O_E112)⊂2xm-S₁₁₂E-Sav].

Apo-Sav protein was crystallized by sitting drop vapor diffusion method under an inert atmosphere. Diffraction quality crystals were grown at room temperature by mixing 3.5 μL of protein solution (26 mg/mL lyophilized protein in water) and 1.5 μL of crystallization buffer (2.0 ammonium sulfate, 0.1 M sodium acetate, pH 4). The droplet was equilibrated against a reservoir solution of 100 uL crystallization buffer. Single crystals of Sav were prepared by soaking apo-crystals in a soaking buffer (2.6 ammonium sulfate, 0.1 M sodium acetate, pH 8) with a 10 mM stock solution of [Fe^II(Biot-et-DPA)(OH₂)₂]Br₂ in nanopure water (9 μL crystallization buffer, 1 μL [Fe^II(biot-et-dpa)(OH₂)₂]Br₂) overnight. After the soaking, crystals were transferred to cryo-protectant for 1 min (30% glycerol in crystallization buffer) and shock-frozen in liquid nitrogen.

Crystallization of [Fe^III(biot-et-dpa(OAc)₂(κ¹-O_E112)⊂2xm-S₁₁₂E-Sav].

Apo-Sav protein was crystallized by sitting drop vapor diffusion method. Diffraction quality crystals were grown at room temperature by mixing 3.5 μL of protein solution (26 mg/mL lyophilized protein in water) and 1.5 μL of crystallization buffer (2.0 ammonium sulfate, 0.1 M sodium acetate, pH 4). The droplet was equilibrated against a reservoir solution of 100 uL crystallization buffer. Single crystals of Sav were prepared by soaking apo-crystals in a soaking buffer (2.6 ammonium sulfate, 0.1 M sodium acetate, pH 6) with a 10 mM stock solution of [Fe^III(biot-ethyl-dpa)(OH₂)₃]Cl₃ in nanopure water (9 μL crystallization buffer, 1 μL [Fe^III(biot-ethyl-dpa)(OH₂)₃]Cl₃) overnight. After the soaking, crystals were transferred to cryo-protectant for 1 min (30% glycerol in crystallization buffer) and shock-frozen in liquid nitrogen. Crystals were prepared in the same manner for data collected using XFEL.

Crystallization of biot-et-dpa⊂ 2xm-S112E-Sav.

Apo-Sav protein was crystallized by sitting drop vapor diffusion method. Diffraction quality crystals were grown at room temperature by mixing 3.5 μL of protein solution (26 mg/mL lyophilized protein in water) and 1.5 μL of crystallization buffer (2.0 ammonium sulfate, 0.1 M sodium acetate, pH 4). The droplet was equilibrated against a reservoir solution of 100 uL crystallization buffer. Single crystals of Sav were prepared by soaking apo-crystals in a soaking buffer (2.6 ammonium sulfate, 0.1 M sodium acetate, pH 6) with a 10 mM stock solution of biot-et-dpa in nanopure water (9 μL crystallization buffer, 1 μL biot-et-dpa) overnight. After the soaking, crystals were transferred to cryo-protectant for 1 min (30% glycerol in crystallization buffer) and shock-frozen in liquid nitrogen.

Crystallization of [Fe^III(biot-ethyl-dpa)(OAc)(N₃)(κ¹-O_E)⊂ 2xm-S₁₁₂E-Sav].

Apo-Sav protein was crystallized by sitting drop vapor diffusion method. Diffraction quality crystals were grown at room temperature by mixing 3.5 μL of protein solution (26 mg/mL lyophilized protein in water) and 1.5 μL of crystallization buffer (2.0 ammonium sulfate, 0.1 M sodium acetate, pH 4). The droplet was equilibrated against a reservoir solution of 100 uL crystallization buffer. Single crystals of Sav were prepared by soaking apo-crystals in a soaking buffer (2.6 ammonium sulfate, 0.1 M sodium acetate, pH 6) with a 10 mM stock solution of [Fe^III(biot-ethyl-dpa)(OH₂)₃]Cl₃ in nanopure water (9 μL crystallization buffer, 1 μL [Fe^III(biot-ethyl-dpa)(OH₂)₃]Cl₃) overnight. After this initial soaking, crystals were again soaked in a soaking buffer (2.6 ammonium sulfate, 0.1 M sodium acetate, pH 6) with a 1 M stock solution of NaN₃ in nanopure water for 5–10 mins (9 μL crystallization buffer, 1 μL NaN₃). After the second soaking, the crystals were transferred to cryo-protectant for 1 minute (30% glycerol in crystallization buffer) and shock-frozen in liquid nitrogen.

Physical Methods

Instrumentation.

¹H, ¹³C, and ¹⁹F NMR spectra were recorded at 500/600, 125, and 400 MHz, respectively. ¹H NMR spectra were reported in ppm on the δ scale and referenced to tetramethylsilane or solvent residual. The data are presented as follows: chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, quin = quintet, m = multiplet, br = broad), and integration. Mass spectra were measured on a MicroMass AutoSpec E, a MicroMass Analytical 7070E, or a MicroMass LCT Electrospray instrument. Electronic absorbance spectra were recorded with a Cary 50 or 8453 Agilent UV-vis spectrophotometer. X-band (9.64 GHz) EPR spectra were recorded on a Bruker spectrometer equipped with Oxford liquid helium cryostats. The quantification of all signals is relative to a CuEDTA spin standard. The concentration of the standard was derived from an atomic absorption standard (Aldrich). For all instruments, the microwave frequency was calibrated with a frequency counter and the magnetic field with an NMR gaussmeter. A modulation frequency of 100 kHz was used for all EPR spectra. The EPR simulation software (SpinCount) was written by our collaborating author Michael P. Hendrich.⁶²