Inverse solvent isotope effects arising from substrate triggering in the factor inhibiting HIF (FIH-1)

Abstract

Oxygen homeostasis plays a critical role in angiogenesis, erythropoiesis and cell metabolism. Oxygen homeostasis is set by the hypoxia inducible factor-1α (HIF-1α) pathway, which is controlled by Factor Inhibiting HIF-1α (FIH). FIH is a non-heme Fe(II), α-ketoglutarate dependent dioxygenase that inhibits HIF-1α by hydroxylating the C-terminal transactivation domain (CTAD) of HIF-1α at HIF-Asn⁸⁰³. A tight coupling between CTAD binding and O₂-activation is essential for hypoxia sensing, making changes in the coordination geometry of Fe(II) upon CTAD encounter a crucial feature of this enzyme. Although the consensus chemical mechanism for FIH proposes that CTAD binding triggers O₂-activation by causing the Fe(II) cofactor to release an aquo ligand, experimental evidence of this point has been absent. More broadly, this proposed coordination change at Fe(II) has not been observed during steady-state turnover in any αKG oxygenase to date. In this manuscript, solvent isotope effects (SIEs) were used as a direct mechanistic probe of substrate triggered aquo release in FIH, as inverse SIEs (SIE < 1) are signatures for pre-equilibrium aquo release from metal ions. Our mechanistic studies of FIH have revealed inverse solvent isotope effects in the steady-state rate constants at limiting concentrations of CTAD or αKG: ^D2Ok_cat/K_M(CTAD) = 0.40 ± 0.07, ^D2Ok_cat/K_M(αKG) = 0.32 ± 0.08, providing direct evidence for aquo release during steady-state turnover. Furthermore, the SIE at saturating concentrations of CTAD and αKG was inverse, ^D2Ok_cat = 0.51 ± 0.07, indicating that aquo release occurs after CTAD binds. The inverse kinetic SIEs observed in the steady state for FIH can be explained by a strong Fe-OH₂ bond. The stable Fe-OH₂ bond plays an important part in FIH’s regulatory role over O₂ homeostasis in humans, and points toward a strategy for tightly coupling O₂-activation with CTAD hydroxylation that relies on substrate triggering.

Introduction

Oxygen homeostasis is essential for proper cellular function in humans and is tightly controlled by a small group of non-heme Fe(II), alpha-ketoglutarate (αKG)-dependent dioxygenases. Oxygen-dependent regulation of the transcriptional coactivator hypoxia inducible factor-1α (HIF-1α) maintains O₂ homeostasis as HIF-1α controls more than 100 genes, directing processes such as angiogenesis, glycolysis and erythropoeisis.(1–3) The Fe(II)/αKG-dependent dioxygenase ‘factor inhibiting HIF-1α’ (FIH-1 or FIH) inhibits HIF-1α in the presence of sufficient [O₂] by hydroxylating the C-terminal activation domain (CTAD) of HIF-1α at HIF-Asn⁸⁰³.(4–7) This hydroxylation prevents recruitment of the CREB binding protein, effectively down regulating HIF-1α dependent gene expression in response to normal or elevated [O₂].

The consensus mechanism for Fe(II)/αKG-dependent dioxygenases proposes that binding of primary substrate triggers the Fe(II) to bind O₂, due to aquo release opening a coordination site for O₂ (Scheme 1). Substrate triggering is a central tenet of the consensus mechanism, as it can explain the relative O₂-reactivity of αKG oxygenases. (11)(49) Subsequent O₂ activation allows for oxidative decarboxylation of αKG and the formation of a highly reactive ferryl intermediate. H-atom abstraction and ensuing ^•OH rebound effectively hydroxylates the primary substrate. As sensing [O₂] requires that O₂-activation only occur when HIF-1α is bound, substrate triggering is the key to O₂-sensing by FIH.

Scheme 1.

Consensus chemical mechanism for FIH.

Various experimental techniques have been used to test the consensus mechanism of the Fe(II)/αKG-dependent dioxygenases. Extensive spectroscopic studies on the Fe(II)/αKG-dependent dioxygenase clavaminate synthase (CAS) have provided insight to the coordination change due to aquo release induced by primary substrate. MCD and CD studies of (Fe + αKG)CAS indicate the Fe(II) is predominantly 6 coordinate. Upon addition of primary substrate (S) the active site iron of (Fe + αKG + S)CAS adopts a 5 coordinate square-pyramidal geometry (8–10) due to release of the aquo ligand. Stopped-flow and freeze-quench techniques have been used to study the Fe(II)/αKG-dependent dioxygenase, taurine dioxygenase (TauD). Loss of the aquo ligand was observed under single-turnover conditions, and these studies identified a Fe(IV)=O intermediate as the active oxidant in TauD.(11–13) However these studies did not address aquo release under multiple turnover conditions leaving open the possibility that aquo release is not a recurring step in the steady-state.

To date, neither in depth spectroscopic studies of the coordination changes proposed upon substrate binding, nor mechanistic probes for substrate triggering have been reported for FIH, making substrate triggering in this enzyme largely speculative. The X-ray crystallographic data for FIH bound to various substrates is ambiguous with respect to the coordination geometry of Fe(II), making it difficult to draw conclusions about the link between substrate binding and aquo release. Structures of (Fe + αKG)FIH are reported as containing either 5-coordinate(15) or 6-coordinate(16) Fe(II), however the crystal structure of (Fe + αKG + CTAD)FIH shows a 5-coordinate Fe(II),(15) suggesting that aquo release may partially occur prior to CTAD binding. EPR spectroscopy of Co(II)-substituted (Co + αKG)FIH revealed a mixture of 5-coordinate and 6-coordinate Co(II), further suggesting that aquo release may occur prior to CTAD binding.(17, 18) Crystal structures of both (Fe + αKG + Notch)FIH and (Fe + αKG + TNKS2)FIH were refined with an axial aquo ligand present, suggesting that aquo release occurs after these substrates bind.(19, 20) Direct mechanistic probes of aquo release during turnover would provide a crucial view into the mechanism of this enzyme, as well as providing insight into substrate triggering by this broad class of enzymes.

We propose that the extensive H-bonding network surrounding the active site of FIH constitutes a second coordination sphere, which transduces CTAD binding into aquo release. Steady-state kinetics of FIH point mutants indicate that removal of hydrogen-bonds from several residues surrounding the Fe(II) impairs catalytic efficiency(14), suggesting that substrate binding causes alterations to those hydrogen bonds. FIH has been shown to have a tightly coupled oxidative and reductive half reaction, avoiding the formation of reactive oxygen species during turnover.(21) Point mutations that perturb this H-bonding network significantly decrease FIH activity and lead to uncoupled αKG consumption.(14) Based on FIH crystal structures and mechanistic data for point mutants, we hypothesized that aquo release occurs on every turnover during the steady state.

To test this hypothesis, we applied mechanistic probes that are sensitive to those steps preceding O₂-activation. Specifically, we focused on the steps involving αKG and CTAD binding, as these steps lead to substrate triggering: the aquo release and priming of the Fe(II) for O₂ binding/activation. Steady-state kinetic assays were performed to identify the preferred binding order for CTAD and αKG. Solvent isotope effects (SIEs) on steady-state rate constants were measured to test for the release of aquo ligands from the Fe(II) cofactor. When the rate constants are larger in D₂O-containing buffers than for H₂O-containing buffer, the kinetics SIEs are ‘inverse’ and are diagnostic for pre-equilibrium release of aquo ligands from metal cofactors. We observed inverse SIEs on steady-state rate constants, providing direct evidence for the pre-equilibrium release of 1, 2 and 3 aquo ligands from the Fe(II) under different conditions. Our results establish two very important points. First, CTAD-binding to (Fe + αKG)FIH leads to release of 1 aquo ligand, suggesting that the overall strategy to control hydroxylation in FIH relies on substrate triggered changes in hydrogen-bonding to release the aquo ligand. Second, the rate-limiting step in the steady-state precedes or coincides with the first irreversible step – this is likely to be decarboxylation (Scheme 1), meaning that FIH does not accumulate oxidized intermediates. Both observations point to an overall strategy for O₂-activation that relies on careful control over substrate triggering in FIH.

Experimental Procedures

Materials

All buffers and reagents were purchased from commercial vendors and were not further purified, with the exception of the CTAD peptide. The CTAD peptide corresponding to the C-terminal activation domain of human HIF-1α, (HIF-1α^788–826) contained a Cys⁸⁰⁰ → Ala point mutation(21): DESGLPQLTSYDAEVNAPIQGSRNLLQGEELLRALDQVN. The CTAD peptide was purchased as a desalted peptide from EZBiolab (Carmel, IN, USA) with free N and C-termini. RP-HPLC utilizing a gradient from 30% acetonitrile/0.1% trifluoroacetic acid (TFA) to 95% acetonitrile/0.1% TFA was used to obtain CTAD of >95% purity.

Protein Expression and Purification

FIH was overexpressed in E. coli with an N-terminal His₆ tag and purified as previously described.(18) Briefly, His₆-FIH was separated from cell lysate via Ni-NTA column chromatography and the affinity tag was then cleaved with thrombin. Three additional residues, from the fusion protein (NH₂-GlySerHis-) remained on the N-terminus following thrombin cleavage. Cleaved FIH was collected as flow through from a Ni-NTA column and then incubated with EDTA to removed metal. Dimeric FIH was obtained with size-exclusion chromatography using Sephadex G-75 resin and 50 mM NaCl 50 mM Tris pH 8.00 as the running buffer. Purified FIH was buffer exchanged into 50 mM HEPES pH 7.00. The molecular weight was confirmed by QSTAR TOF-MS (expected, 40.566 kDa; observed, 40.574 kDa), while the purity (>95%) was assessed by SDS-PAGE.

Steady-state kinetics assays

All assays were performed at 37.0°C with saturating concentrations of FeSO₄ (50 μM) and ascorbate (2 mM), and ambient O₂ concentration. Assays in which CTAD was the varied substrate (15 μM – 250 μM) utilized saturating αKG (500 μM) unless specified otherwise. Assays with αKG as the varied substrate (5 μM – 200 μM) utilized a fixed CTAD concentration of either 39 μM (~1/2K_M(CTAD)) or 306 μM (~4xK_M(CTAD)). Assay reagents were mixed and incubated for two minutes at 37.0°C in microcentrigue tubes. Then the reaction was initiated with the addition of enzyme ([E]_T = 0.5 μM). Reaction aliquots (5 μL) were quenched with 75% acetonitrile/0.2% TFA (20 μL) saturated with 3,5-dimethoxy-4-hydroxycinnamic acid and analyzed for peptide hydroxylation using a Bruker Daltonics Omniflex MALDI-TOF-MS. The mole fraction of product, χ_CTAD-OH, was determined by the relative intensities of hydroxylated CTAD (CTAD^OH, 4271 m/z) and unhydroxylated CTAD (4255 m/z). Initial rates were determined from 5–7 quenched timepoints. The steady-state rate constants, k_cat and k_cat/K_M, were obtained by nonlinear least-squares fitting of initial rate data (0 to ~15% fractional conversion) to the Michaelis-Menten equation.

Viscosity Assays

Steady-state assays were performed in 50 mM HEPES pH 7.00, using sucrose as the viscosogen to test for the rate limitation of diffusional steps. Initial rates were measured as described above, using buffers containing 10% (η_rel=1.3), 18% (η_rel=1.8) and 25% (η_rel=2.8) sucrose solutions (w/w).(22)

Solvent Isotope Assays

Deuterium oxide (D, 99.9%) was purchased from Cambridge Isotope Laboratories (Andover, MA, USA) and used as received. The pD was determined by pre-soaking the pH meter in D₂O for ten minutes and then adding 0.4 to the meter reading of the D₂O solution of interest (pD = pH_read + 0.4).(23) Steady-state assays were performed as described above. All reagent stocks used in the steady-state assays in D₂O, were prepared using D₂O. Working FIH stock solutions were made by diluting high concentration stocks from H₂O into D₂O containing 50mM HEPES pD 7.00. Assays were performed in 50 mM HEPES, pD 7.00, with a final D₂O percentage of 97%. SIEs were calculated from the ratio of rate constants observed in buffers containing H₂O or D₂O; e.g. ^D2Ok_cat = k_cat(H2O)/k_cat(D2O).

Coupling Ratio

The extent of coupling between FIH’s two half reactions in D₂O was determined from monitoring the succinate and CTAD^OH concentrations throughout a reaction. Reactions containing αKG (500 μM), FeSO₄ (50 μM), CTAD (200 μM) and FIH (5 μM) were performed in 50 mM Tris pD 7.00, and analyzed similarly to previously reported procedures.(14, 24) A Hamilton PRP-X300 anion exclusion column was used to separate the succinate produced from the quenched reactions. UV detection at 210 nm was used to determine the succinate concentration. Using aliquots from the same quenched assay, a Bruker Daltonics Omniflex MALDI-TOF-MS was used to determine the CTAD^OH concentration. The coupling ratio (C) was determined by fitting [succinate] as a linear function of [CTAD^OH] for multiple quench points, where C = [succinate]/[CTAD^OH].

Results

Solvent isotope effects

Solvent isotope effects (SIEs) were measured to test for aquo release from the Fe(II) as part of substrate triggering during turnover in FIH. By varying the degree of saturation with respect to αKG and CTAD, the SIE on kinetic constants reporting on different microscopic steps were accessed. SIEs arising from the pre-equilibirum release of 1, 2, and 3 aquo ligands were predicted from the consensus mechanism, depending on the assay conditions (Scheme 1). As the limiting SIEs are multiples of the proton fractionation factor (Φ) for the O-L bond (L = H or D)(25), they are quite distinct for the number of aquo ligands released.

The fractionation factor for L₂O (where L = H or D) is inverse when water is bonded to another Lewis acid, leading to a tendency for D₂O to accumulate in bulk solvent. For example, Φ_O-L = 0.69 for L₃O⁺ (26–28) due to zero-point energy differences. The fractionation factor, Φ_O-L, for aquo release from Co(II) in two carbonic anhydrase isozymes has been found to be between 0.72 and 0.90,(25, 29) agreeing closely with the fractionation factor for $\text{Co}{({\text{H}}_2\text{O})}_6^{2+}$ in solution, (Φ_O-L = 0.73).(30) As the fractionation factor for the Fe²⁺-OH₂ in FIH is unknown, we estimate Φ = 0.70 as is commonly done for fractionation of water from other M-OH₂ ⇌ M + OH₂ equilibria.(30) As Φ is reported on a per-bond basis, and are multiplicative, ^D2OK_eq for the release of aquo ligands are diagnostic for the numbers of H₂O released from Fe: one (^D2OK_eq = Φ² = 0.49), two (^D2OK_eq = Φ⁴ = 0.24) and three (^D2OK_eq = Φ⁶ = 0.12).

The equilibrium fractionation factors will lead to inverse kinetic SIEs on different kinetic constants provided that the aquo release is in a pre-equilibrium prior to a subsequent rate-limiting step. As shown in the Appendix, k_cat and k_cat/K_M(CTAD) can report on one aquo release – leading to the prediction that the limiting values for ^D2Ok_cat ~ ^D2O(k_cat/K_M(CTAD)) = 0.49. In contrast, the limiting values for ^D2O(k_cat/KM_(αKG)) will report on the release of 2 or 3 aquo ligands depending on the degree to which FIH is saturated with CTAD. It is important to note that our assays used ambient [O₂] (220μM), which is approximately 2xK_M(O2),(31) making our kinetic constants apparent.

The rate constants obtained under conditions of saturating αKG, ambient [O₂] and varied [CTAD] are k_cat and k_cat/K_M(CTAD). Under these conditions, k_cat reports on all steps after CTAD binding up through product release, and is predicted to involve the release of one aquo ligand.(Scheme 1) The initial rate assays in D₂O exhibited increased rates when compared to similar assays in H₂O, indicating an inverse SIE on both k_cat and k_cat/K_M(CTAD). We observed ^D2Ok_cat = 0.51 ± 0.07, in good agreement with the limiting SIE (0.49) for the pre-equilibrium release of one aquo ligand (Figure 1). Similarly, we observed ^D2O(k_cat/K_M(CTAD)) = 0.40 ± 0.07, which is also in good agreement with one aquo release prior to a rate-limiting step.

Fig. 1.

Steady-state kinetics of FIH in H₂O (●) and 97% D₂O (■) buffers. FIH (0.5 μM), ascorbate (2 mM), αKG (500 μM), FeSO₄ (50 μM) and CTAD (0–250 μM) were in 50mM HEPES pL 7.00.

Due to the sequential ordered mechanism (see below), the apparent k_cat/K_M(αKG) encompasses distinct steps depending on the fixed concentration of CTAD. Since k_cat/K_M(αKG) reports on steps from the encounter with αKG up through the first irreversible step, high or low fixed concentrations of CTAD were used to isolate different microscopic steps. CTAD binding is kinetically irreversible at high CTAD concentrations, meaning that k_cat/K_{M(αKG)High[CTAD]} reports on only those steps between αKG encounter and CTAD binding, encompassing the release of two aquo ligands. The observed SIE of 0.32 ± 0.08 on k_cat/K_{M(αKG)High[CTAD]} was in reasonable agreement with the limiting value (0.24) expected for the release of two aquo ligands. In contrast, CTAD binding is kinetically reversible at sub-saturating CTAD concentrations, making a subsequent step (thought to be O₂ activation) the first irreversible step. Thus, k_cat/K_{M(αKG)Low[CTAD]} reports on all steps between αKG binding and O₂ activation, which encompasses the release of three aquo ligands in the consensus mechanism. The observed SIE of 0.11 ± 0.03 for k_cat/K_{M(αKG)Low[CTAD]} agreed closely with the limiting value (0.12) expected for the release of three aquo ligands prior to a rate-limiting step (Figure 2).

Fig. 2.

Steady-state kinetics of FIH for αKG at 37μM CTAD and 306μM CTAD, in H₂O (●,■) and D₂O(○,□)at 37.0°C. FIH (0.5μM), ascorbate (2 mM), αKG (0 – 210 μM), FeSO₄ (50 μM) and CTAD (37 μM and 306 μM) were in 50 mM HEPES pL 7.00.

Validation of Observed SIEs

A series of control experiments were completed to ensure that the SIEs arose from aquo release. Control assays showed the steady-state rate constants, k_cat and k_cat/K_M(CTAD), were pH independent between pH 6.50 and pH 8.00 at fixed ionic strength (I = 120 mM). Furthermore, steady-state assays at pH 7.00 showed that k_cat and k_cat/K_M(CTAD) were independent of ionic strength (Figure S1). Viscosity experiments were completed as a control for the increased relative viscosity of D₂O, as solvent viscosity can affect the rate of diffusional steps, or of conformational changes. The binding order of αKG and CTAD was also investigated to define the chemical steps reported on by k_cat/K_M(CTAD) and k_cat/K_M(αKG), as these rate constants proved different microscopic steps within the chemical mechanism reporting the release of up to three aquo ligands.

FIH’s activity increased approximately two fold in D₂O, prompting us to check the coupling between the oxidative and reductive half-reactions in D₂O. As the oxidative half reaction produces succinate, and the reductive half-reaction produces CTAD^OH, the [succinate]/[CTAD^OH] ratio is equal to the coupling between these half reactions (C=[succinate]/[CTAD^OH]). Previously we showed tight coupling between succinate production and CTAD hydroxylation in H₂O.(14) Here we observed that FIH’s two half reactions remained tightly coupled in D₂O, with C equal to unity within experimental uncertainty(C = 1.1 ± 0.1) (Table 2).

Table 2.

Coupling of [Succinate] and [CTAD^OH] for FIH. 50 mM Tris pL 7.00, at 37.0°C.

	Ca
D2O	1.1 ± 0.1b
H2O	0.98 ± 0.03c

^a $C=\frac{\mathit{moles}\mathit{succinate}}{\mathit{moles}{\mathit{CTAD}}^{OH}}$

^bReactions contained αKG (500 μM), FeSO₄ (50 μM), CTAD (200 μM) and FIH (5 μM) in 50 mM Tris pD 7.00

^cReference (14)

Diffusional steps are not rate-limiting

Steady-state rate constants were measured as a function of relative viscosity to test for diffusional steps that might contribute to the observed rate constants and SIEs, as the viscosity of D₂O is greater than that of H₂O. Initial rate data were collected in 50 mM HEPES, pH 7.00, with sucrose as the viscosogen, and fitted to the Michaelis-Menten equation. Analysis of k_cat, k_cat/K_M(CTAD) and k_cat/K_M(αKG) as a function of relative viscosity indicated that each of the steady-state rate constants were independent of solvent viscosity (Figure S2).

The normalized regression plot showed that relative viscosity had no effect on any rate constant. Thus, neither diffusional encounter, relevant for k_cat/K_M, nor product release, relevant for k_cat, are rate-limiting. If diffusional encounter were rate limiting, k_cat/K_M would decrease as the relative solvent viscosity increased.(32, 33) Furthermore, the lack of a viscosity effect indicates FIH does not undergo a viscosity-sensitive conformational change, which has been shown to obscure SIEs for some enzymes.(34, 35)

Sequential Order: αKG binds before CTAD

The consensus mechanism for αKG oxygenases is a sequential ordered model that, when applied to FIH, predicts αKG binds prior to CTAD. Although steady-state analyses of several other αKG-dependent oxygenases have been shown to follow this binding order,(36–38) we tested this sequential binding model for FIH due to the interesting SIEs that we observed under varied [CTAD]. As the binding order of αKG and CTAD affects the microscopic steps defined by the steady-state kinetic parameters, this is critical to our interpretations of the SIEs.

Four different fixed CTAD concentrations, ranging from ½K_M(CTAD) to 4xK_M(CTAD) were chosen for steady state kinetic assays where the αKG concentration was the varied substrate. In close agreement with previously reported values,(14, 31) the K_M(αKG) remained constant at 20 ± 2μM for all CTAD concentrations. The regression plot of k_cat/K_M(αKG) as a function of [CTAD] passed through the origin, as expected for ordered sequential binding of αKG prior to CTAD (Figure 3).(39)

Fig. 3.

Regression plot showing k_cat/K_M(αKG) as a function of CTAD concentration. Inset: Steady-state kinetics of FIH with αKG as varied substrate, at different fixed CTAD concentrations at 37.0°C. FIH (0.5 μM), ascorbate (2 mM), αKG (0–210 μM), FeSO₄ (50 μM) and CTAD (39 – 306 μM) were in 50mM HEPES pH 7.00.

Discussion

O₂ sensing by FIH is critical to cellular growth and development making close consonance between [O₂] and the FIH-catalyzed hydroxylation of CTAD central to O₂-homeostasis. An additional benefit is that controlled O₂ activation by FIH would avoid ROS production (21), preventing anomalous oxidations. Although it has been shown that decarboxylation of αKG and hydroxylation of CTAD by FIH is tightly coupled,(14) the mechanistic strategy used by FIH to ensure that O₂ activation leads to CTAD hydroxylation with high fidelity remains unclear. Our overall hypothesis is the hydrogen bonding from the second coordination sphere ensures that the Fe(II) in (Fe + αKG)FIH remains 6-coordinate, and that CTAD binding alters these hydrogen bonds leading to O₂ activation – the substrate triggering model. CTAD binding could trigger FIH by affecting any one of the microscopic steps in the overall chemical mechanism, so long as the rate of O₂-activation were increased upon CTAD binding. Our observation of inverse SIEs on k_cat/K_M(CTAD) and k_cat/KM_(αKG) indicate that the rate-limiting step in the steady state follows aquo release, and is either coincident with or precedes decarboxylation (Scheme 1). This points towards a strategy for tightly coupling O₂-activation with CTAD hydroxylation that relies on substrate triggering.

O₂-activation is rate-limiting in FIH

Inverse kinetic SIEs (SIE < 1) are diagnostic for aquo release from metal ions, as the vast majority of SIEs are normal (SIE > 1).(29) Two other causes of inverse kinetic SIEs, the involvement of a CysS- nucleophile in catalysis, and a conformational change, were ruled out for FIH. In the former case, there are no active site CysSH residues; in the latter case, we verified that viscosity did not affect the rate of turnover.

Inverse kinetic SIEs have been reported for several metalloenzymes, including a number of hydrolases, carbonic anhydrase, and a point mutant of SLO.(29, 40–43) In each case, the inverse kinetic SIEs indicated that aquo release was a pre-equilibrium step prior to a subsequent rate limiting step. Our observation of inverse kinetic SIEs corresponding to multiple aquo release steps is the first such report, to the best of our knowledge. The inverse SIEs arise from an unfavorable equilibrium for aquo release, combined with a slow O₂ activation step.

The consensus mechanism predicts that both k_cat and k_cat/K_M(CTAD) encompass one aquo release for a limiting predicted SIE = 0.49. The experimentally observed kinetic SIEs correlate nicely with the theoretical value expected for the equilibrium release of one aquo ligand, as ^D2Ok_cat = 0.51 ± 0.07 and ^D2Ok_cat/K_M(CTAD) = 0.40 ± 0.07.(Table 1) A kinetic model with separate microscopic steps for αKG, CTAD and O₂ binding, as well as the final aquo release, was constructed from the consensus chemical mechanism and used to analyze the observed SIEs (Scheme 2). As k_cat encompasses all steps with the exception of diffusional encounter with substrates, a separate, reversible microscopic step for aquo release was needed, as the ^D2Ok_cat < 1 indicated that aquo release must be separate from CTAD encounter.

Table 1.

Solvent Isotope Effects for FIH. 50 mM HEPES pL 7.00, at 37.0°C.

	H2O	D2O	Experimental SIE	Theoretical SIE	# of Aquo Ligands
kcat (min−1)a	30 ± 2.5	59 ± 2.0	0.51 ± 0.07	0.49	1
kcat/KMCTAD (μM−1 min−1)a	0.43 ± 0.10	1.09 ± 0.11	0.40 ± 0.07	0.49	1
kcat/KMαKG(High[CTAD]) (μM−1 min−1)b	1.94 ± 0.13	6.05 ± 1.37	0.32 ± 0.08	0.24	2
kcat/KMαKG(Low[CTAD]) (μM−1 min−1)c	0.29 ± 0.07	2.60 ± 0.34	0.11 ± 0.03	0.12	3

^aCTAD as the varied substrate; FIH (0.5μM), ascorbate (2 mM), αKG (500 μM), FeSO₄ (50 μM), and CTAD (0 – 250 μM)

^bαKG as the varied substrate; FIH (0.5μM), ascorbate (2 mM), αKG (5 – 200 μM), FeSO₄ (50 μM), and CTAD (306 μM)

^cαKG as the varied substrate;FIH (0.5μM), ascorbate (2 mM), αKG (5 – 200 μM), FeSO₄ (50 μM), and CTAD (37 μM)

Scheme 2.

Proposed kinetic mechanism for FIH (S = CTAD).

The similarity in the SIEs for k_cat and k_cat/K_M(CTAD) strongly suggests that the overall rate-limiting step in FIH turnover is common to both rate constants. As k_cat/K_M(CTAD) encompasses steps between CTAD encounter and the first irreversible step, the common step has to be the O₂-activation step (Scheme1). This step is typically viewed as the initial attack on the 2-oxo group of αKG which leads to oxidative decarboxylation. This is noteworthy as that this predicts that FIH will not accumulate the ferryl intermediate during turnover, in contrast to TauD.(12) Slow O₂-activation can account for the tight coupling between decarboxylation and hydroxylation seen for FIH(14) as hydroxylation must be very fast relative to the decarboxylation step. Thus, we conclude that FIH likely adopts a strategy to ensure tight coupling between O₂-binding and -activation in which O₂-activation is rate-limiting.

We analyzed the observed kinetic SIEs using language predominately developed by Northrop and Cleland (44, 45), in which observed kinetic SIEs are a function of “commitments to catalysis” and the equilibrium and kinetic isotope effects (^D2OK_qq and ^D2Ok₅) on the aquo release step (k₅). The commitments to catalysis, C_f and C_r, are ratios of microscopic rate constants that describe the tendency of an enzyme to go forward or backward through the isotopically sensitive step. Full expressions are provided in the Appendix. In the case of ^D2Ok_cat/K_M(CTAD), this expression takes the form below. The primary virtue of such a presentation is that it permits us to focus on specific segments of the kinetic mechanism. As equilibrium isotope effects reflect fractionation factors, the equilibrium SIE on aquo release is predicted to be ~ 0.5 (^D2OK₅ = 0.49).

$${}^{D2O}\left(\frac{k_{\mathit{cat}}}{K_{M(\mathit{CTAD})}}\right)=\frac{{{}^{D2O}k}_5+C_f+C_r{{}^{D2O}K}_5}{1+C_f+C_r}$$ (1)

It is clear that the only way for ^D2Ok_cat/K_M(CTAD) = ^D2OK_eq is for the reverse commitment to be very large (C_r ≫ C_f). Reverse commitment (C_r) is the kinetic competition between H₂O and O₂ for the triggered form of enzyme – when C_r is large, it means that H₂O is the preferred ligand (Appendix – Eqn. A10). Furthermore, the C_f/C_r ratio is, to a first approximation, the equilibrium for aquo release – when C_f/C_r ≪1, it means that K₅ is much less than one and that the position of this equilibrium favors the aquo-on state. Similar analysis of the commitment factors for ^D2Ok_cat further supports the conclusion that aquo release is disfavored thermodynamically, despite a slightly different SIE expression and forward commitment on k_cat (C_vf) (Appendix).

Substrate triggering is faster than O₂-activation

Steps involved in substrate triggering were accessed by measuring the kinetic SIE with sub-saturating [αKG]. As k_cat/K_M(αKG) includes steps between diffusional encounter of αKG with FIH, up through and including the subsequent irreversible step, CTAD binding and substrate triggering may be observable on this kinetic constant. Under conditions of high [CTAD], only the two aquo ligands released upon αKG binding may be observed, as saturating FIH with CTAD makes CTAD binding kinetically irreversible for k_cat/K_{M(αKG)High[CTAD]}. In contrast, subsaturating [CTAD] makes it possible to access release of all three aquo ligands on k_cat/K_{M(αKG)Low[CTAD]}, provided that a later step be rate-limiting.

The observed SIE for k_cat/K_{M(αKG)High[CTAD]} (SIE = 0.32 ± 0.08) correlates well with the pre-equilibrium release of two aquo ligands (Table 1), as expected for kinetically irreversible CTAD binding. The key observation supporting rapid substrate triggering was the SIE on k_cat/K_{M(αKG)Low[CTAD]} (SIE = 0.11 ± 0.03), as this indicated the pre-equilibrium release of all three aquo ligands prior to a subsequent rate-limiting step. This establishes that steps associated with CTAD binding and substrate triggering are not slow under these conditions, and further points to O₂-activation as the rate-limiting step.

Hydrogen bonding from the second coordination sphere

Turnover in FIH and other αKG oxygenases depends on substrate triggering – minimally the release of the aquo ligand from the (Fe + αKG + Substr) enzyme form – prior to O₂ binding. What is striking about the kinetics of FIH are the inverse kinetic SIEs, implicating slow O₂-activation in this enzyme. In the case of FIH, substrate triggering leads to rate-limiting O₂-activation, meaning that the intermediate which is expected to accumulate in the steady-state is a relatively innocuous Fe²⁺ center. In contrast, TauD exhibits SIEs of unity(46), and the partially rate-limiting steps include product release and H-atom transfer by the [Fe^IVO]²⁺ intermediate.(11) For TauD, substrate triggering leads to rapid O₂-activation, and the partial accumulation of a powerful oxidant, [Fe^IVO]²⁺.(12) It is as if TauD were built for speed, but FIH built for fidelity. As the ligands to the Fe(II) are identical in these two enzymes, the reasons for such disparate strategies for the oxidation reaction must lie beyond the primary coordination sphere.

We attribute the difference in oxidation strategies between these enzymes to their second coordination spheres. The most striking difference between their second coordination spheres is in the hydrogen bonding between the facial triad Asp²⁰¹ ligand and the axial aquo ligand. In FIH, the remote O-atom of Asp²⁰¹ forms a 2.8 Å hydrogen bond to the bound aquo ligand in (Fe + αKG)FIH (Figure 4) – once CTAD binds, two new hydrogen bonds form to the Asp²⁰¹ which appear to serve to partially stimulate aquo release.(16, 17) The extensive hydrogen bonding to this aquo ligand likely serves to stabilize the Fe-OH₂ bond, which would explain the large reverse commitments to catalysis (C_r) and the inverse SIEs. Such high affinity for the aquo ligand could serve to throttle back the oxidation reaction such that O₂-activation only occurs when CTAD is present. We note a certain similarity to the hydrogen bonding network found in PHD2, which leads to an inverse ^D2Ok_cat (47) and the absence of any accumulating intermediates in pre-steady state experiments.(24)

Fig. 4.

FIH active site with the 2.8 Å Asp²⁰¹-aquo hydrogen bond. (PDBID: 3P3P)

TauD is a significant contrast to FIH, structurally and kinetically. The facial triad Asp of TauD is rotated away from the aquo ligand, and cannot form a hydrogen bond to the bound aquo ligand. This leads to a weak Fe²⁺-OH₂ bond strength(48), and TauD readily activates O₂ in the absence of the substrate taurine.(49, 50) The second coordination sphere of TauD favors aquo-release, which can lead to rapid O₂-activation – unfortunately, this leads to a reduction in fidelity.

Conclusion

Inverse kinetic SIEs on k_cat/K_M and k_cat limit the possible rate-limiting step for FIH to a step between aquo release and decarboxylation (Scheme 1). This points to a step very early in the catalytic cycle, such as O₂ activation, as the likely rate-limiting step. Such a strategy would aid in FIH’s regulatory role over O₂ homeostasis in humans, as it would lead to direct transduction of intracellular O₂ levels into a readable signal, hydroxylated HIF-Asn⁸⁰³. This tight control over O₂-activation strongly suggests that oxidized intermediates will not accumulate in FIH under normal turnover conditions, and may reflect the demands of a regulatory function in O₂-sensing by tightly correlating O₂-activation with substrate hydroxylation.

Abbreviations

HIF

hypoxia inducible factor-1α

FIH

factor-inhibiting HIF

αKG

α-ketoglutarate

CTAD

C-terminal transactivation domain

CREB

cAMP response element-binding protein

CAS

clavaminate synthase

MCD

magnetic circular dichroism

CD

circular dichroism

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

SIE

solvent isotope effect

ROS

reactive oxygen species

PHD2

prolyl hydroxylase dioxygenase 2

TauD

taurine dioxygenase

SLO

soybean lipoxygenase

Appendix

As there are multiple isotope sensitive steps for the kinetic mechanism of FIH (Scheme 2), algebraic expressions for each kinetic parameter were derived using the net rate method of Tian.(51) The expression for ^D2Ok_cat takes the form:

$${}^{D2O}(k_{\mathit{cat}})=\frac{{{}^{D2O}k}_5+C_{Vf}+C_r{{}^{D2O}K}_5+{{}^{D2O}k}_{15}\left(\frac{k_5}{k_{15}}\right)}{1+C_{Vf}+C_r}$$ (A1)

$$C_{Vf}=k_5\left[\frac{k_8+k_9}{k_7{k}_9[O_2]}+\frac{1}{k_9}+\frac{1}{k_{11}}+\frac{1}{k_{15}}\right]$$ (A2)

$$C_r=k_6\left[\frac{1}{k_7[O_2]}+\frac{k_8}{k_7{k}_9[O_2]}\right]$$ (A3)

The parameters C_Vf and C_r are the forward and reverse commitments to catalysis. ^D2Ok₅ is the kinetic SIE for water release, ^D2OK₅ is the equilibrium SIE on aquo release and ^D2Ok₁₅ the kinetic SIE on water re-binding. The equations derived for the SIEs for k_cat/K_M(CTAD), k_cat/K_{M(αKG)High[CTAD]}, k_cat/K_{M(αKG)Low[CTAD]} take similar form and are shown below.

$${}^{D2O}\left(\frac{k_{\mathit{cat}}}{K_{M(\alpha KG)[\mathit{HighCTAD}]}}\right)=\frac{{{}^{D2O}k}_1\left(\frac{k_3[\mathit{CTAD}]}{k_2}\right)+{{}^{D2O}K}_1}{1+\frac{k_3[\mathit{CTAD}]}{k_2}}$$ (A4)

$${}^{D2O}\left(\frac{k_{\mathit{cat}}}{K_{M(\alpha KG)[\mathit{LowCTAD}]}}\right)=\frac{{{}^{D2O}k}_1\left(\frac{k_3[\mathit{CTAD}]k_5}{k_2{k}_4}\right)+{{}^{D2O}K}_1\left(\frac{k_5}{k_4}\right)+{{}^{D2O}k}_5{{}^{D2O}K}_1+C_r{{}^{D2O}K}_1{{}^{D2O}K}_5}{1+C_f+C_r}$$ (A5)

$$C_f=k_5\left(\frac{1}{k_4}+\frac{k_3}{k_2{k}_4}\right)$$ (A6)

$$C_r=k_6\left[\frac{1}{k_7[O_2]}+\frac{k_8}{k_7{k}_9[O_2]}\right]$$ (A7)

$${}^{D2O}\left(\frac{k_{\mathit{cat}}}{K_{M(\mathit{CTAD})}}\right)=\frac{{{}^{D2O}k}_5+C_f+C_r{{}^{D2O}K}_5}{1+C_f+C_r}$$ (A8)

$$C_f=\frac{k_5}{k_4}$$ (A9)

$$C_r=k_6\left[\frac{1}{k_7[O_2]}+\frac{k_8}{k_7{k}_9[O_2]}\right]$$ (A10)