Evidence That Two Enzyme-derived Histidine Ligands Are Sufficient for Iron Binding and Catalysis by Factor Inhibiting HIF (FIH)*

A 2-His-1-carboxylate triad of iron binding residues is present in many non-heme iron oxygenases including the Fe(II) and 2-oxoglutarate (2OG)-dependent dioxygenases. Three variants (D201A, D201E, and D201G) of the iron binding Asp-201 residue of an asparaginyl hydroxylase, factor inhibiting HIF (FIH), were made and analyzed. FIH-D201A and FIH-D201E did not catalyze asparaginyl hydroxylation, but in the presence of a reducing agent, they displayed enhanced 2OG turnover when compared with wild-type FIH. Turnover of 2OG by FIH-D201A was significantly stimulated by the addition of HIF-1α786–826 peptide. Like FIH-D201A and D201E, the D201G variant enhanced 2OG turnover but rather unexpectedly catalyzed asparaginyl hydroxylation. Crystal structures of the FIH-D201A and D201G variants in complex with Fe(II)/Zn(II), 2OG, and HIF-1α786–826/788–806 implied that only two FIH-based residues (His-199 and His-279) are required for metal binding. The results indicate that variation of 2OG-dependent dioxygenase iron-ligating residues as a means of functional assignment should be treated with caution. The results are of mechanistic interest in the light of recent biochemical and structural analyses of non-heme iron and 2OG-dependent halogenases that are similar to the FIH-D201A/G variants in that they use only two His-residues to ligate iron.

The 2-His-1-carboxylate iron binding triad (HX(D/E) . . . H) is an extensively conserved feature of the active site of the nonheme iron-dependent oxygenases and oxidases, including the 2OG 6 -dependent dioxygenase family (2OG oxygenases). Crystallographic and spectroscopic studies imply that the enzyme⅐Fe⅐2OG intermediate has an octahedral metal coordination, with three of the coordination sites occupied by the 2-oxo-acid of 2OG and a water molecule. In some cases, binding of substrate to this complex has been shown to induce loss of the iron bound water so enabling dioxygen binding. Oxidative decarboxylation of 2OG results in the formation of a ferryl species (Fe(VI)ϭO) that effects oxidation of the substrate/cosubstrate with the regeneration of Fe(II) at the active site (for review, see Refs. [1][2][3]. In addition, spectroscopic studies have demonstrated that interaction of the non-iron-ligated oxygen atom of the carboxylate side chain of the metal binding Asp/Glu residue with the iron bound water molecule has a role in oxygen activation (4).
Hydroxylation is the most common reaction catalyzed by the 2OG oxygenases, but family members also catalyze other oxidative reactions including demethylations, desaturations, epoxidations, and rearrangements. Related enzymes, which do not employ 2OG as a co-substrate, have been shown to catalyze oxidative fragmentations and cyclization reactions (for review, see Refs. 2 and 5). Recently, two 2OG oxygenases have been reported to catalyze oxidative halogenations (6,7); the structure of one of these halogenases, SyrB2, revealed that the iron was coordinated by only two enzyme-derived His residues (8). The carboxylate-bearing residue found in all prior 2OG oxygenase structures was substituted by an Ala in SyrB2 apparently leaving sufficient space for a chloride ion to fill the vacant iron coordination site.
Studies on the iron cofactor binding residues of FIH have supported the assignment of FIH as a 2OG oxygenase and enabled identification of the iron binding residues as His-199, Asp-201, and His-279. Lando et al. (12) demonstrated that substitution of either His-199 or Asp-201 to alanine residues ablated FIH activity in cells. Crystallographic analyses confirmed the assignment of the iron binding residues (16 -18). A structure of dimeric FIH in complex with Fe(II), 2OG, and HIF-1␣ 786 -826 revealed that the sidechain carboxylate of FIH Asp-201 not only coordinates the iron via one oxygen, but that its other oxygen is positioned to accept a hydrogen bond from an Fe(II)-bound water molecule; upon substrate binding, the interaction of the water is apparently weakened because Asp-201 is repositioned to form a hydrogen bond with the backbone amide nitrogen of HIF-1␣ Asn-803 as observed in the crystal structures ( Fig. 2) (17).
We report here that although FIH-D201A and FIH-D201E do not catalyze HIF-1␣ hydroxylation, under appropriate conditions, these enzymes do catalyze enhanced 2OG turnover when compared with the wild-type FIH. Surprisingly, the FIH-D201G mutant catalyzes asparaginyl hydroxylation at comparable levels to wild-type FIH. Crystal structures of FIH-D201A⅐Fe(II)⅐2OG⅐HIF-1␣ 786 -826 and D201G⅐Zn(II)⅐2OG⅐ HIF-1␣ 788 -806 reveal that the only FIH-based ligands required to bind iron are His-199 and His-279. Together with the recent studies on 2OG halogenases (6, 7) and spectroscopic analyses (4), our results indicate that there is more flexibility in the iron coordination chemistry of 2OG oxygenases than previously thought.
Radiochemical Assays-Enzyme assays measured the release of [ 14 C]CO 2 from 1-[ 14 C]2OG (11). FIH variants (11.5 M) were incubated with HIF-1␣ 786 -826 (57.5 M) in the presence of 80 M Fe(II), 160 M 2OG, 4 mM ascorbate, and 1 mM dithiothreitol (DTT) in 50 mM Tris-HCl, pH 7.5 at 37°C for 30 min. The final reaction volume was 100 l. Under these conditions, ϳ50% of the HIF-1␣ substrate is hydroxylated by wild-type FIH. Reactions were quenched by the addition of 200 l of methanol followed by a 30-min incubation on ice. Ascorbate analogs or alternative substrates (Sigma-Aldrich) were dissolved in Tris-HCl buffer (50 mM, pH 7.5) and tested in the 1-[ 14 C]2OG decarboxylation assay as reported above. For time course data, reactions were stopped at specific time points by quenching as described above. Assays monitoring oxygen depletion were carried out as described (19).
Hydroxylation Assays Using MALDI MS HIF-1␣ 788 -806 peptide substrate was obtained from Peptide Protein Research Ltd. (Fareham, UK). Reactions were carried out using the same conditions as for the radiochemical assays, except unlabeled 2OG was used and the final 2OG (500 M) and Fe(II) (400 M) concentrations were increased. Incubations in the presence of sodium carbonate, sodium formate, sodium acetate, sodium propionate, or sodium azide (1, 5, 10, 20, 50, 100, or 200 mM) were carried out. After incubation (30 min), trifluoroacetic acid (0.1% final concentration) was added, and the reaction mixture was diluted five times with double distilled water. For MS analyses, the quenched reaction mixtures (1 l) were mixed with ␣-cyano-4-hydroxycinnamic acid (1 l) and spotted onto a target plate. MALDI-time-of-flight spectra in the negative ion mode were then recorded for each sample.
Crystallography-The crystallization conditions for both FIH-D201A and FIH-D201G were similar to those described for wild-type FIH (17). The variants were crystallized (in a Belle Laboratories glove box under anaerobic conditions for FIH-D201A and aerobic conditions for FIH-D201G) by hanging drop vapor diffusion at 18°C using 2-l drops (1 l of protein solution:1 l of well solution) against a well solution of 0.1 M HEPES, pH 7.5, 1.6 M (NH 4 ) 2 SO 4 , and 6% polyethylene glycol 400. The FIH-D201A or FIH-D201G protein solution contained 14 or 25 mg/ml, respectively, in 50 mM Tris-HCl, pH 7.5, with 1 mM HIF-1␣ 786 -826 or 0.5 mM HIF-1␣ 788 -806 peptide, 1 mM FeSO 4 , or ZnCl 2 and 2 mM 2OG (added under anaerobic conditions for FIH-D201A). Crystals were fully grown within 8 days. The crystals were removed from the glove box, transferred into a well solution containing 24% glycerol that was prepared in the glove box and immediately retrieved using a nylon loop, frozen by rapid plunging into liquid nitrogen, and stored under liquid nitrogen until data collection. The entire process of removing the crystals from the glove box through freezing took less than 5 min to minimize the diffusion of oxygen into the crystal-containing solution. Crystal-containing loops were mounted on a goniometer in a cryo-stream at 100 K. For FIH-D201A, 153 images of 1°oscillation, 15 s per image with a crystal to detector distance of 275 mm, were collected at the European Synchrotron Radiation Facility (ESRF) BM14 tuned to wavelength of 0.95372 Å using a MarMosiaic 225 CCD detector. For FIH-D201G, 360 images of 0.5°oscillation, 1 s per image, and a crystal to detector distance of 294 mm were used. Data were collected at Diamond I04 tuned to a wavelength of 0.9696 Å using an ADSC Quantum 315 CCD detector. Data were processed using HKL2000 (20) (supplemental Table S1).
To follow the refinement progress, a subset of 5% of the data was withheld from refinement throughout to obtain an R free value. The coordinates from PDB entry 1H2N (17) were used as the starting model for one cycle of rigid body refinement followed by cycles of maximum likelihoodsimulated annealing with bulk solvent correction using CNS version 1.1 (21) and model rebuilding using COOT version 0.26 (22) until convergence of R and R free . Water molecules were placed based on 2F o Ϫ F c electron density peaks that were within hydrogen-bonding distance of protein donors or acceptors. The structure was analyzed for problem areas between refinement cycles using PROCHECK (23) and WHATIF (24). The structures have been deposited in the Research Collaboratory for Structural Bioinformatics (RCSB) PDB as PDB IDs 2ILM (FIH-D201A) and 3D8C (FIH-D201G).

RESULTS
Biochemical Characterization of FIH D201A-Consistent with the previous cell-based results (12), purified FIH-D201A did not catalyze hydroxylation of HIF peptide substrate (HIF-1␣ 788 -806 ) in vitro, FIGURE 2. Insights from crystal structures of FIH-D201A and FIH-D201G. a, stereo view of the iron binding site of the FIH-D201A⅐Fe(II)⅐2OG⅐HIF-1␣ 786 -826 complex. Experimental electron density OMIT map (F o Ϫ F c ) contoured to 5 represented as blue mesh (electron density is carved out around the residues and ligands displayed for clarity). The unanticipated electron density adjacent to the iron was provisionally modeled as (bi)carbonate (see "Results" for discussion). b, comparison of the wild-type FIH⅐Fe(II)⅐2OG⅐HIF-1␣ 786 -826 complex (PDB ID 1H2L) with the FIH-D201A⅐Fe(II)⅐2OG⅐HIF-1␣ 786 -826 complex (wild-type FIH (blue) in complex with HIF substrate (cyan) and FIH-D201A (yellow) in complex with HIF substrate (magenta)). This figure emphasizes several important interactions between wild-type FIH and HIF-1␣ that are lost in the FIH-D201A complex: wild-type FIH Asp-201 and HIF-1␣ Asn-803 main chain nitrogen (yellow dash), wild-type FIH Gln-239 and HIF-1␣ Asn-803 side chain, and wild-type FIH Trp-296 and HIF-1␣ Val-802. Note the presence of the assigned sulfate ion (orange and red) in the FIH-D201A structure apparently replacing the carboxylate of the HIF-1␣ Glu-801 in the wild-type FIH⅐Fe(II)⅐2OG⅐HIF-1␣ 786 -826 complex, which is also observed in uncomplexed wild-type FIH structure (PDB 1H2N, not shown). c, stereo views from the crystal structures of 2OG-dependent halogenase, SyrB2 (pink) superimposed on the FIH-D201A variant (yellow). The FIH-D201A variant shares the same HXA . . . H motif as SyrB2 (PDB ID 2FCT); however, the FIH-D201A apparently does not provide enough space for a chloride ion to complete octahedral coordination to the Fe(II), which could explain why FIH-D201A does not have halogenase activity toward HIF-1␣ under our assay conditions. Distances between the FIH-D201A Ala-201 C␤ methyl group and Fe(II) in each of the structures are shown as black dashed lines to emphasize this point. d, stereo view ball-and-stick representation of the FIH-D201G⅐Zn(II)⅐2OG⅐HIF-1␣ 786 -826 complex metal binding site. Experimental electron density 2F o Ϫ F c contoured to 1.0 represented as blue mesh (electron density is carved out around the residues and ligands displayed for clarity). FIH and 2OG are colored green, HIF-1␣ 786 -826 is colored yellow, Zn(II) is colored as a gray sphere, and the Zn(II) bound water is colored as a red sphere. Wat, water. as demonstrated by MS analyses (Fig. 3). However, using an assay in which the amount of [ 14 C]CO 2 produced from 1-[ 14 C]2OG is used to measure activity, significant stimulation of 2OG turnover by the FIH-D201A variant was observed under standard assay conditions in the presence of ascorbate (supplemental Fig. S1). For FIH-D201A, the extent of 2OG decarboxylation was greater than for wild-type FIH, with all of the 2OG in the reaction mixture being consumed in the FIH-D201A assay. Under the same conditions, wild-type FIH catalyzed hydroxylation of ϳ50% of the HIF-1␣ substrate and decarboxylated only ϳ15% of the 2OG present (Fig. 4). Attempts to restore the wild-type hydroxylase activity with the FIH-D201A mutant by the use of small molecules (formate, acetate, propionate, carbonate, and azide) in assays with HIF-1␣ 788 -806 were unsuccessful (supplemental Fig. S3).
Ascorbate and HIF-1␣ Dependence-Ascorbate has relatively little effect on wild-type FIH activity. Unexpectedly, when using HIF-1␣ 786 -826 as a substrate for FIH-D201A, it was found that the addition of L-ascorbate was a requirement for enhanced 2OG turnover, stimulating activity ϳ10-fold. Furthermore, the 2OG turnover activity of FIH-D201A was reduced (ϳ10-fold) in the absence of HIF-1␣ 786 -826 substrate, implying that both ascorbate and HIF-1␣ 786 -826 substrate are required for efficient 2OG decarboxylation (Fig. 4) despite the absence of HIF-1␣ 788 -806 hydroxylation. For both wild-type FIH and FIH-D201A, activity was reduced in the absence of added Fe(II).
Dimerization Dependence-The FIH-L340R substitution has been shown to convert the FIH homodimer into a predominantly monomeric form that shows ϳ20% of the wild-type FIH 2OG turnover activity but is inactive with respect to HIF hydroxylation (26). To investigate whether the homodimeric form of FIH-D201A was required for the uncoupled 2OG turnover, the double mutant FIH-D201A/L340R was prepared and analyzed. FIH-D201A/L340R did not stimulate 2OG turnover in the presence of L-ascorbate and/or HIF-1␣, indicating that, as for wild-type FIH, the homodimeric form of FIH-D201A is required for catalysis (Fig. 4).
Further evidence for the selectivity of reducing agent came from the observation that DTT effectively substituted for L-ascorbate (ϳ95% activity), but dithionite, ␤-mercaptoethanol, and tris-(2-carboxyethyl)phosphine hydrochloride all stimulated activity to less than 10% (ϳ3, 9, and Ͻ1%, respectively) of that observed for L-ascorbate and DTT. Since phenol and its derivatives have been used both as alternative substrates and as competitive inhibitors of the copper-dependent ascorbate oxidase (27), phenol and selected phenolic derivatives (4-chlorophenol, 4-nirophenol, 4-cyanophenol, 4-tertbutylphenol, 4-methylphenol, 1,2-dihydroxybenzene, 4-nitrocatechol) were tested both as ascorbate substitutes and as inhibitors of FIH-D201A. None of the phenolic compounds could replace ascorbate as stimulators of FIH-D201A activity or were found to be potent inhibitors.
FIH-D201E-Given that the structurally characterized 2OGdependent dioxygenases (JMJD2A and clavaminic acid synthase) have a glutamic acid residue functioning as the carboxylate iron binding ligand (28,29), the FIH-D201E variant was investigated. The FIH-D201E had low 2OG turnover activity in the absence of ascorbate, and MS analyses showed that this mutant did not support HIF-1␣ 788 -806 hydroxylation. However, FIH-D201E did catalyze the turnover of 2OG (to a lesser extent than FIH-D201A), which was stimulated by the addition of L-ascorbate and GST-HIF-1␣ 786 -826 (Fig. 4).
FIH-D201G and FIH-D201G/ W296R-To further probe the role of the carboxylate residue of the facial triad in FIH, a glycyl residue was substituted for Asp-201. The FIH-D201G mutant behaved similarly to FIH-D201A in that 2OG decarboxylation was highly dependent on the presence of L-ascorbate. 2OG turnover by FIH-D201G was less dependent on GST-HIF-1␣ 786 -826 when compared with FIH-D201A (Fig. 4). An unanticipated result was that, in contrast to FIH-D201A and FIH-D201E, FIH-D201G supported HIF-1␣ 788 -806 hydroxylation (as shown by MS assays) at levels comparable with wild-type FIH (Fig. 3). A FIH-D201G/W296R double variant was also generated because an arginine occupies the equivalent position as the FIH Trp-296 at the active site of the SyrB2 2OG-dependent halogenase (8). Catalysis of HIF-1␣ 788 -806 hydroxylation by the double variant was not detected by MS analyses (Fig. 3), possibly because the substitution of the hydrophobic tryptophan with a polar/charged arginine probably disrupts the interaction with Val-802 of HIF1␣ (17). For the double variant, 2OG decarboxylation was observed, but to a lesser extent than that observed with the D201A variant. 2OG turnover was highly dependent on the presence of L-ascorbate but was less dependent on the presence of HIF-1␣ 786 -826 when compared with D201A (Fig.  4). As for FIH-D201A and FIH-D201G, activity of FIH-D201G/ W296R was stimulated by the addition of Fe(II).
Crystal Structures of FIH-D201A and FIH-D201G in Complex with HIF-1␣ Substrate-To examine the mode of metal coordination in FIH-D201A and FIH-D201G, crystallographic analyses were performed. Attempts to crystallize the D201E mutant failed, so modeling of this variant was carried out (supplemental Fig. S2).
FIH-D201A crystals were grown in the presence of Fe(II), 2OG, and HIF-1␣ 786 -826 under anaerobic conditions by the hanging drop vapor diffusion method. Synchrotron data were collected to 2.3 Å resolution and refined to a final R factor of 21.0% and R free of 26.7%. FIH-D201G crystals were grown in the presence of Zn(II), 2OG, and HIF-1␣ 788 -806 by the hanging drop vapor diffusion method. Synchrotron data were collected to 2.1 Å resolution and refined to a final R factor of 21.8% and R free of 25.6%.
In both FIH-D201 variant structures, the active site metal is coordinated by only two enzyme-derived residues (His-199 and His-279) that adopt near identical conformations to those in the wild-type FIH structures. The main-chain atoms of residue 201 overlay with those of the wild-type structure (Fig. 2, a and  b). In both variant structures, the 2OG is coordinated to the Fe(II) in a similar manner to wild-type FIH with its 1-carboxylate group binding trans to His-199. However, the electron density map of the FIH-D201A⅐Fe(II)⅐2OG⅐HIF-1␣ 786 -826 structure indicated an additional planar metal ligand, possibly comprised of four atoms bound to Fe(II) in a bidentate manner that apparently distorts the typical octahedral metal coordination geometry. This apparent ligand, provisionally assigned as a carbonate or bicarbonate ion, is possibly derived from CO 2 , generated by 2OG (30), but other possibilities including acetate or a HIF-1␣-derived species cannot be ruled out. One of the two assigned Fe(II)-ligated oxygen atoms of the (bi)carbonate approximately bisects the position of the two carboxylate oxygens of Asp-201 in wild-type FIH and has an apparently longer coordination bond length (2.5 Å) than the other Fe(II)-coordinating atoms (Յ 2.4 Å).
In the FIH-D201A⅐Fe(II)⅐2OG⅐HIF-1␣ 786 -826 structure, there was reasonable occupancy for residues 794 -799 of HIF-1␣ in site 1, and comparison with the wild-type FIH⅐Fe(II)⅐2OG⅐HIF-1␣ 786 -826 complex reveals that this region of the HIF-1␣ peptide adopts a similar conformation in both structures (Fig. 2d). This region of HIF-1␣ binds close to the FIH dimer interface formed by the two C-terminal helices (26). However, HIF-1␣ residues (800 -806 in PDB ID 1H2L) near the iron binding site in the FIH-D201A⅐Fe(II)⅐2OG⅐HIF-1␣ 786 -826 structure are probably disordered, as observed by broken residual difference density that did not refine well, even when defined with a low occupancy.
Overall, the crystallographic results imply that only two enzyme-derived iron-coordinating ligands are required for 2OG turnover and hydroxylation. Although it is possible that the differences may result from crystallographic artifacts, the different binding modes observed for HIF-1␣ 786 -826 in the FIH-D201A and the WT FIH/FIH-D201G correlate with the lack of hydroxylation activity observed for FIH-D201A. The non-productive binding of HIF-1␣ 786 -826 in FIH-D201A crystals may be related to the apparent metal ligand observed for this variant because it is in position to interfere with substrate binding.

DISCUSSION
Although we cannot entirely exclude the possibility of a rearrangement in solution to provide a third protein based Fe(II) ligand, the biochemical and crystallographic results presented here provide evidence that only two His-residues are required for FIH-D201A/G to bind iron and 2OG to catalyze the oxidative decarboxylation of 2OG to succinate and carbon dioxide.
In contrast to the wild-type and FIH-D201G HIF substrate complex structures, density for HIF-1␣ Asn-803 and the surrounding residues (800 -806) was not observed adjacent to the iron center in the FIH-D201A substrate complex. Although we cannot be certain that this reflects the situation in solution, the presence of a methyl side chain at FIH residue-201 must hinder productive binding. The difference in the hydroxylation activity between the FIH-D201A and FIH-D201G cannot solely result from steric effects because in wild-type FIH residue-201 is an aspartate. It is possible that the presence of a glycine, rather than an alanine residue, enables complexation of a water molecule to the iron that can hydrogen-bond to HIF-1␣ Asn-803 in a analogous manner to that observed for Asp-201 in the wildtype FIH-HIF structural complex.
The FIH-D201A mutant resembles the SyrB2 2OG-dependent chlorinating enzyme, of which the iron is also coordinated by two His-residues with an Ala residue in a similar position as residue 201 in FIH (7). During catalysis by SyrB2, a halide ion replaces the iron coordination site occupied by the side-chain carboxylate of Asp-201 in FIH (Fig. 2c) (8). Attempted chlorination of the HIF-1␣ substrate by FIH-D201A in assays supplemented with sodium chloride was unsuccessful. Structural comparison of FIH-D201A with SyrB2 suggests that this may in part be due to a lack of space for the chloride ion to coordinate Fe(II) (Fig. 2c). Although our attempted halogenation experiments were negative, the knowledge that FIH-D201G only requires two enzyme-derived iron ligands opens the way for studies aimed at further modifying FIH to enable the halogenation of unactivated C-H bonds in proteins.
In light of the discovery of the 2OG chlorinating enzymes (7,8) and the results described here for the FIH-D201A mutant, consideration should be given to the possibility of an Fe(II)coordinating "two-histidine dyad" in searches for 2OG oxygenases in bioinformatic-based approaches. Other variants of the facial triad may also exist; modeling of the JmjC domain of the fission yeast protein Epe1, thought to be a 2OG-dependent hydroxylase (31), using an alignment with FIH, indicates that the second His residue of the facial triad is replaced by Tyr (32). Thus the predicted iron-coordinating residues in Epe1 are His, Glu, and Tyr (HXE . . . Y). Sequence analyses also imply that the HX(D/E) . . . H facial triad may not be conserved in other 2OGdependent dioxygenases, including the epidermal growth factor aspartyl/asparaginyl hydroxylase (33).
The FIH-D201A/E variants do not catalyze HIF-1␣ asparaginyl hydroxylation, but do catalyze decarboxylation of 2OG, and at a higher rate than that observed for the wildtype FIH. However, unlike the hydroxylation reaction catalyzed by wild-type FIH with HIF-1␣ C-terminal transactivation domain, the oxidation of 2OG by both FIH-D201G/A mutants was very dependent on the presence of a reducing agent, such as ascorbate or DTT. In the case of FIH-D201G, 2OG turnover was coupled, at least partially, to substrate hydroxylation. However, in contrast to wild-type FIH, for FIH-D201G, both 2OG decarboxylation in the absence of substrate and HIF substrate hydroxylation were ascorbatedependent. This dependence resembles that of some other 2OG oxygenases (see e.g. Refs. 34 and 35), including collagen prolyl hydroxylase, the activity of which is well known to be stimulated by ascorbate.
It is reasonable to propose that the first part of the catalytic cycles for both the wild-type FIH and the FIH-D201X mutants is common, i.e. 2OG is oxidized to produce succinate, carbon dioxide, and an Fe(VI)ϭO intermediate (36,37). Because HIF-1␣ is not hydroxylated by the FIH-D201A/E mutants, regeneration of Fe(II) from the higher oxidation state intermediates is required to complete their catalytic cycle. The requirement for a reducing agent may in part rationalize the increased dependence of these FIH variants on ascorbate when compared with WT FIH. It is also possible that, in addition to completing some catalytic cycles, ascorbate affects the reactivity of intermediate complexes.
Substitutions of iron binding residues or other active site residues of 2OG-dependent dioxygenases have been reported (38,39) to (partially) decouple 2OG and prime substrate oxidation. However, the increased 2OG turnover by the D201A/E/G variants of FIH is unusual. In FIH, Asp-201 is proposed to play a role both in binding iron and in accepting a hydrogen bond from HIF Asn-803, the substrate residue that is hydroxylated. A third possible role for Asp-201 is in hindering binding of oxygen prior to that of HIF substrate. The non-iron-ligating side-chain carboxylate oxygen may stabilize binding of a water molecule to the five coordinate FIH⅐Fe⅐2OG complex, giving a relatively stable six-coordinate complex (as proposed for other 2OG oxygenases (4)). On substrate binding, a five-coordinate complex is formed that can react with oxygen and in which the non-ligating oxygen of Asp-201 hydrogen bonds with Asn-803 of HIF. This mechanism could prevent FIH-mediated non-productive 2OG turnover when HIF substrate is not available. Substitutions of FIH Asp-201 may result in a weakened interaction between the water molecule and Fe(II) in the FIH-D201A/E⅐Fe(II)⅐2OG complexes, making reaction with oxygen easier and causing the high levels of 2OG turnover observed for these variants in the absence of HIF.
Significantly, despite the fact the methyl side chain of Ala-201 can neither covalently bind the iron nor hydrogen-bond to the substrate, the 2OG turnover activity of FIH-D201A was stimulated by the presence of HIF-1␣ 786 -826 . Since the HIF-1␣ 786 -826 was not itself hydroxylated by FIH-D201A, the results imply that the oxidation of 2OG is catalyzed by a hybrid complex comprised of FIH-D201A and HIF-1␣ 786 -826 . It is possible that the selectivity of this hybrid complex could be engineered by synthetic modifications to the HIF-1␣ 786 -826 component. The HIF-1␣ 786 -826 -stimulating effect was seen to a much lesser extent with the D201E FIH mutant, possibly because this mutant can still bind iron via a triad of FIH-derived residues; the larger glutamyl side chain may also limit space available for substrate (peptide or ascorbate) binding.
The increased 2OG turnover activity observed for FIH-D201A in the presence of HIF-1␣ 786 -826 substrate may in part be because the HIF-1␣ substrate modifies the FIH active site to strengthen Fe(II) binding. However, the structural studies suggest that such an effect may be more complicated than strengthening of iron binding via a direct complexation. Although the detailed mechanism by which HIF-1␣ peptides stimulate 2OG turnover by FIH-D201A is complicated, the available evidence implies that it may also include binding interactions away from the immediate vicinity of the Fe(II) including those at, or close to, the dimer interface.