Increased turnover under limiting [O₂] by the Thr³⁸⁷→Ala variant of the HIF-prolyl hydroxylase PHD2

Abstract

PHD2 is a 2-oxoglutarate, non-heme Fe²⁺ dependent oxygenase that senses O₂ levels in human cells by hydroxylating two prolyl residues in the oxygen dependent degradation domain (ODD) of HIF1α. Identifying the active site contacts that determine the rate of reaction under limiting O₂ is crucial for understanding how these enzymes sense pO₂, and may suggest methods for chemically altering hypoxia responses. A hydrogen bonding network extends from the Fe(II) cofactor through ordered waters to the Thr³⁸⁷ residue in the second coordination sphere. Here we tested the impact of the sidechain of Thr³⁸⁷ on the reactivity of PHD2 toward O₂ through a combination of point mutagenesis, steady state kinetic experiments and {FeNO}⁷ EPR spectroscopy. The steady state kinetic parameters for Thr³⁸⁷→Asn were very similar to those of WT-PHD2, but k_cat and k_cat/K_M(O2) for Thr³⁸⁷→Ala were increased by roughly 15-fold. X-band EPR spectroscopy of the {FeNO}⁷ centers of the (Fe+NO+2OG) enzyme forms showed the presence of a more rhombic line shape in Thr³⁸⁷→Ala than seen for WT-PHD2, indicating an altered conformation for bound gas in this variant. Here we show that the sidechain of residue Thr³⁸⁷ plays a significant role in determining the rate of turnover by PHD2 at low [O₂].

Graphical abstract

Introduction

HIF prolyl-4-hydroxylase 2 (PHD2) is a non heme Fe(II), 2-oxoglutarate (2OG) dependent oxygenase that serves as the primary O₂ sensor in human cells^1,2. PHD2 regulates the transcriptional activity of the hypoxia-inducible factor-1α (HIF-1α) transcription factor by hydroxylating Pro⁴⁰² and Pro⁵⁶⁴ within the oxygen dependent degradation domain (ODD) of HIF-1α^1,3–5. As HIF-1α hydroxylation underlies angiogenesis and the balance between aerobic/anaerobic metabolism, PHD2 plays a crucial role in health conditions such as ischemia, anemia, and cancer^6,7. Identifying the structural features of PHD2 that determine the reaction rate under conditions of limiting [O₂] could point the way to altering cellular hypoxia sensing.

The (Fe+2OG)PHD2 form of enzyme contains an Fe(II) cofactor coordinated by a His₂Asp facial triad, a bidentate 2OG ligand, and an aquo ligand. PHD2 is thought to follow the consensus chemical mechanism for 2OG oxygenases (Scheme 1).^8–10 When substrate ODD binds near the cofactor, the aquo ligand is released in analogy to other 2OG oxygenases, permitting O₂ to bind at the Fe for subsequent chemistry. Oxidative decarboxylation generates the ferryl (FeO)²⁺, an intermediate that has been observed in related enzymes,^8,9,11,12 which hydroxylates the Pro target residue to complete the chemical steps of turnover (Scheme 1). Notably, PHD2 appears to be rate-limited by a step associated with O₂ binding or oxidative decarboxylation, as the (FeO)²⁺ intermediate did not accumulate in the pre-steady state for PHD2,¹⁰ and PHD2 exhibited an inverse solvent kinetic isotope effect (KSIE).^13,14 This is in contrast to other 2OG oxygenases that accumulate the (FeO)²⁺ in the pre-steady state, for which product release or hydroxylation appear to be rate-limiting step.^8,9,11,12

Scheme 1.

Consensus chemical mechanism for PHD2.

The connection between substrate binding and O₂ activation is crucial to understanding the chemistry of this class of enzyme. A recent study revealed a conserved pattern of second-sphere contacts in enzymes structurally related to PHD2¹⁵ which may be functionally significant. The X-ray crystal structure of (Fe+2OG)PHD2¹⁶ showed a hydrogen bonding network connecting the aquo ligand and 2OG to the sidechain of Thr³⁸⁷ (Figure 1). Thr³⁸⁷ is located on the second β strand, at a position which forms a conserved H-bond in many 2OG oxygenases.¹⁵ As O₂ activation is integral to the function of PHD2 and other 2OG oxygenases, the disposition of Thr³⁸⁷ may be crucial to gas binding and reactivity.

Figure 1.

PHD2 active site (PDB ID: 3OUJ).¹⁶ Waters (red sphere), distances (Å).

Here we tested the role of second sphere contacts from Thr³⁸⁷ on PHD2 hydroxylation chemistry through steady state kinetics and {FeNO}⁷ EPR spectroscopy. Steady-state kinetic parameters for the Thr³⁸⁷→Asn variant, which conserved the hydrogen bonding capacity at this position, were very similar to those of WT-PHD2. Remarkably, the macroscopic rate constants for the Thr³⁸⁷→Ala variant were 15 times larger than for WT-PHD2, which is attributed to faster O₂ activation. Substrate inhibition was observed for the Thr³⁸⁷→Ala variant, which implicates product release as a slow step in this variant – a corollary is that other substrate inhibition in other 2OG oxygenases likely reflects slow succinate release¹⁷. These results suggest that the second sphere contacts from Thr³⁸⁷ slow turnover such that decarboxylation chemistry is rate-limiting in WT-PHD2.

Materials and Methods

Materials

All chemicals were purchased from commercial vendors, and used without purification. The sequences of the peptide substrates were derived from the native HIF-1α^556–574 (CODD) sequence. The sequence of the CODD peptide was DLDLEALAP⁵⁶⁴YIPADDDFQL. The underlined residues were changed from the native sequences as follows: CODD (M561A, M568A) to avoid methionine oxidation. The CODD peptide (99 % purity) was purchased from GL Biochem LTD (Shanghai).

Protein expression and purification

Recombinant human PHD2_178–426 and its variants, Thr³⁸⁷→Ala and Thr³⁸⁷→Asn were expressed as a C-terminal GST fusion protein from E. coli BL21(DE3) cells using a pGEX-4T-1 vector (Stratagene) and purified as previously described^14,18. Briefly, PHD2-GST was purified using an affinity column (GSTrap, GE Bioscience), then the GST-tag was cleaved with restriction grade thrombin for 16 hours at 4 °C. The thrombin was removed with a Hitrap Benzamidine column (GE Bioscience), then PHD2 was treated with 50 mM EDTA overnight to remove metals. PHD2 was buffer exchanged with 50 mM HEPES pH 7.50 and stored at −20 °C. The mass of the WT-PHD2 and the variants were determined by a QStar-XL hybrid quadrupole-TOF mass spectrometer (Applied Biosystems). WT-PHD2 (27786.6 Da calculated, 27784.7 Da observed), Thr³⁸⁷→Ala (27755.2 Da calculated, 27756.2 Da observed), and Thr³⁸⁷→Asn (27799.6 Da calculated, 27798.2 Da observed).

Steady state kinetic assays

Steady-state kinetic constants were obtained from initial rates were measured under saturating concentrations of (NH₄)₂Fe(SO₄)₂ (10 µM) and ascorbic acid (1 mM) in 50 mM HEPES buffer pH 7.00 at 37.0 °C. MALDI-TOF was used to measure the production of of hydroxylated product CODD^OH ((M+O+Na⁺): 2174 m/z calc., 2172 m/z obs.) from the substrate peptide CODD ((M+Na⁺): 2158 m/z calc., 2156 m/z obs.). Substrate inhibition constants were obtained by fitting the initial-rate data to Eq. 1, in which the initial rate (v_o) is a function of the maximal rate (v_max), the Michaelis constant (K_M) and the inhibition constant (K_i) for the substrate (S).

$${\upsilon }_0=\frac{V_{\mathit{\text{max}}}}{1+\frac{K_M}{[S]}+\frac{[S]}{K_i}}$$ (Equation 1)

For assays in which [2OG] was varied, the concentration of CODD (10 µM) was fixed at saturating levels. Reactions were quenched at different time points in saturated 4-α-cyano hydroxycinnamic acid dissolved in 75 % acetonitrile and 0.2 % trifluoroacetic acid. Quenched reactions were spotted onto a MALDI target plate and analyzed using a Bruker Daltonics Omniflex MALDI-TOF mass spectrometer. For assays in which the [CODD] varied, the concentration of 2OG (100 µM) was fixed at saturating.

Assays in which O₂ was the varied substrate used saturating concentrations of (NH₄)₂Fe(SO₄)₂ (10 µM), 2OG (100 µM), CODD (10 µM) and ascorbic acid (1 mM) in 50 mM HEPES pH 7.00 at 37 °C. The concentration of dissolved O₂ was controlled by varying the ratio of N₂ and O₂ gas flowing through separate flow meters through a gas washer into an Atmosbag (Sigma-Aldrich), with the dissolved O₂ concentration monitored by a YSI model 5300 biological oxygen monitor. Reactions were quenched and analyzed as for other steady state reactions.

2OG Binding

The dissociation constant (K_D) of 2OG was determined by measuring the quenching of the intrinsic tryptophan fluorescence of PHD2 upon binding 2OG. Samples were excited at 295 nm and the emission measured at 330 nm using a PTI spectrofluorometer (PTI-QM-4 2005SE). A cuvette containing PHD2 (1 µM), and MnSO₄ (20 µM) in 50 mM HEPES pH 7.00 was titrated with aliquots of 2OG (500 µM); the K_D was obtained by fitting the data to equation 2, in which the observed fluorescence intensity (I) is normalized to the intensity in the absence of 2OG (I₀) and the presence of saturating [2OG] (I_f).

$$\frac{I-I_0}{I_f-I_0}=\frac{([PHD2]+[2OG]+K_D)-\sqrt{{([PHD2]+[2OG]+K_D)}^2-(4[PHD2][2OG])}}{2[PHD2]}$$ (Equation 2)

X-band EPR of {FeNO}⁷

Anaerobic samples were prepared a glovebox ([O₂] < 1 ppm) for analysis. (NH₄)₂FeSO₄ was brought into the glovebox as a solid and dissolved using degassed H₂O. The DEANO stocks were prepared in 10 mM NaOH in the glovebox and the concentration verified by its published extinction coefficient and characteristic UV absorbance at 250 nm.^19,20 A 100 µL EPR sample was prepared anaerobically and contained PHD2 (0.10 mM), (NH₄)₂FeSO₄, (0.10 mM), 2OG (0.50mM), CODD (100 µM) and DEANO (0.5 mM) in 50 mM HEPES pH 7.00 at room temperature (23°C). The sample was aged for 20 minutes to allow NO release from DEANO, then flash frozen in liquid nitrogen. EPR spectra were collected using a Bruker Elexsys E-500 EPR equipped with a DM4116 cavity and a Bruker ER 4118CF-O LHe/LN₂ cryostat at 9.624 GHz frequency, 2.0 mW power, 10 G modulation amplitude, 100 GHz modulation frequency, 163 ms time constant, 4 K. As the electronic structure of the S = 3/2 ferrous-nitrosyl center is highly axial (E/D < 0.1), the observed EPR resonances (g_x,y,z) were related to the rhombicity of the zero field splitting (E/D) using an effective spin Hamiltonian (Eq. 3).²¹

$$g_x=g_0[2-\raisebox{1ex}{$3E$}\!\left/ \!\raisebox{-1ex}{$D$}\right.-\raisebox{1ex}{$3$}\!\left/ \!\raisebox{-1ex}{$2$}\right.{(\raisebox{1ex}{$E$}\!\left/ \!\raisebox{-1ex}{$D$}\right.)}^2]$$ (Eq. 3a)

$$g_y=g_0[2+\raisebox{1ex}{$3E$}\!\left/ \!\raisebox{-1ex}{$D$}\right.-\raisebox{1ex}{$3$}\!\left/ \!\raisebox{-1ex}{$2$}\right.{(\raisebox{1ex}{$E$}\!\left/ \!\raisebox{-1ex}{$D$}\right.)}^2]$$ (Eq. 3b)

$$g_z=g_0[1-3{(\raisebox{1ex}{$E$}\!\left/ \!\raisebox{-1ex}{$D$}\right.)}^2]$$ (Eq. 3c)

Results and Discussion

Steady state kinetics with varied [2OG]

The structure of PHD2 showed a hydrogen bond network connecting 2OG, the aquo ligand, and Thr³⁸⁷ (Figure 1). As a hydrogen bonding residue is structurally conserved at this position in other 2OG oxygenases,^13,15 we hypothesized that this played an important role during turnover. To test the effect of this contact on steps between 2OG binding and ODD binding, initial rate data were measured as a function of varied [2OG] using saturating concentrations of Fe²⁺ (10 µM), CODD (10 µM), and ascorbate (1 mM); [O₂] was fixed at ambient levels (217 uM). WT-PHD2 exhibited simple saturation kinetics over the range of 1 – 50 µM 2OG, with values of k_cat (2.3 min⁻¹) and k_cat/K_M(2OG) (2.7 µM⁻¹min⁻¹) similar to those previously reported^{14,18,22–24}. Similarly, the conservative Thr³⁸⁷→Asn variant exhibited saturation kinetics over the tested concentration range, with modestly decreased rate constants relative to WT-PHD2 (Table 1).

Table 1.

Steady state kinetics of WT-PHD2 and variants with varied [2OG].^a

Enzyme	kcat min−1	kcat/KM(2OG) µM−1min−1	KI(2OG) µM	KM(2OG), µM	KD(2OG)b µM
WT-PHD2	2.3 ± 0.1	2.7 ± 0.4	>2.6×104	0.9 ± 0.2	0.7 ± 0.2
Thr387→Ala	32.9 ± 1.4	2.7 ± 0.3	2150 ± 340	12.0 ± 1.6	1.2 ± 0.2
Thr387→Asn	1.0 ± 0.1	1.5 ± 0.2	>5×104	0.7 ± 0.1	0.6 ± 0.2

^a. Reactions contained (NH₄)₂Fe(SO₄)₂ (10 µM), ascorbic acid (1 mM), 2OG (1–3000 µM), and CODD (10 µM) in 50 mM HEPES pH 7.00, 37°C, ambient [O₂](217 µM).

^b. Fluorescence titrations: PHD2 (1.1 µM), MnSO₄ (20 µM) in 50 mM HEPES pH 7.00 titrated with 2OG (500 µM).

The initial rate data for the Thr³⁸⁷→Ala variant was a significant contrast to that of WT-PHD2. While Thr³⁸⁷→Ala exhibited simple saturation kinetics for [2OG] < 250 µM, this variant exhibited ~50% activity at 3000 µM 2OG (Figure 2). Fitting the initial rate data to a simple substrate inhibition model led to parameters of k_cat (32.9 min⁻¹), k_cat/K_M(2OG) (2.7 µM⁻¹min⁻¹), and K_I(2OG) (2200 µM). The high value for k_cat indicated that Thr³⁸⁷→Ala performed chemistry more rapidly than WT-PHD2, but the observed substrate inhibition suggested that slow release of one of the products could limit turnover.

Figure 2.

Steady state kinetics for PHD2 variants with varied [2OG]. Reaction mixtures included (NH₄)₂Fe(SO₄)₂ (10 µM), ascorbic acid (1 mM), 2OG (1–3000 µM), and CODD (10 µM) in 50 mM HEPES pH 7.00, 37.0 °C, ambient [O₂] (217 µM).

The steady state rate constant k_cat/K_M(2OG) reflects those steps between 2OG binding and the subsequent irreversible step, which is CODD binding under conditions of saturating [CODD]. As this rate constant is essentially unchanged upon mutation of Thr³⁸⁷, contacts from this residue do not affect these steps. In contrast, k_cat reflects non-diffusional steps, and was greatly affected by mutation of Thr³⁸⁷. While k_cat was unchanged in the conservative Thr³⁸⁷→Asn variant, this kinetic parameter increased ~ 15 fold for the Thr³⁸⁷→Ala variant, suggesting that a chemical step was affected by the Thr³⁸⁷ variants.

The substrate inhibition observed for Thr³⁸⁷→Ala prompted us to test the other variants for this effect. WT-PHD2 retained >95% activity even at extremely high levels of 2OG; fits of this data to the substrate inhibition model were unsatisfactory (K_I(2OG) >26 mM) indicating that PHD2 did not undergo substrate inhibition. Thr³⁸⁷→Asn similarly showed minimal inhibition at extremely high 2OG concentration (K_I(2OG) >50 mM). The absence of significant substrate inhibition observed for WT-PHD2 was consistent with product release being rapid, consistent with the prior reports that implicated a step early in catalysis as being rate-limiting for WT-PHD2¹⁰.

Substrate inhibition could arise from many possible mechanisms in which substrate binding leads to an enzyme form of lower reactivity. Inhibition by 2OG could be due to 2OG binding to the Fe²⁺ cofactor in a geometry that prevented O₂ from reacting, or by binding to the Fe²⁺ when an equivalent of succinate or 2OG was already present. Although our kinetics data and 2OG binding data cannot distinguish among these possibilities, a unified model that accounts for the observed kinetics (see below) and substrate inhibition found for Thr³⁸⁷→Ala is one in which chemistry is much faster than succinate release, such that the (Fe+Succ)PHD2 form of enzyme accumulates in the steady state for this variant. Inhibition by 2OG or by CODD would arise from these substrates binding to the (Fe+Succ)PHD2 form of enzyme, thereby preventing the enzyme from adopting the proper cofactor orientation to enter into catalysis.

The binding affinity for 2OG was measured for each of the variants by intrinsic tryptophan fluorescence quenching (supplemental material). Binding curves were fit to a single binding site model using Eq. 2, which indicated a similar avidity toward 2OG (K_D ~ 1 µM) for each variant. Attempts to fit a second 2OG binding event were not successful, possibly due to the large difference in magnitude between K_M(2OG) and K_I(2OG) for each of the variants.

Steady state kinetics with varied [CODD]

The impact of Thr³⁸⁷ variants on the chemical steps of turnover was tested by measuring steady state kinetics with CODD as the varied substrate. Initial rates of CODD hydroxylation were measured as a function of varied CODD concentration in 50 mM HEPES (pH 7.00, 37.0 °C), with saturating concentrations of Fe²⁺ (10 µM), 2OG (100 µM), and ascorbate (1 mM); the [O₂] was fixed at the ambient level (217 µM). As O₂ was not saturating, k_cat/K_M(CODD) will reflect steps spanning CODD binding up through the irreversible chemistry involved with oxidative decarboxylation. Steady state kinetic constants, k_cat, and k_cat/K_M(CODD), were determined by fitting the data to the Michaelis-Menten equation (Figure 3). The rate constants for WT-PHD2 were in good agreement with prior reports^14,22,25. While the conservative Thr³⁸⁷→Asn mutation led to rate constants that were modestly decreased from those of WT-PHD2, those for the Thr³⁸⁷→Ala variant were ~ 10 fold greater than those of WT-PHD2, indicating that removal of the hydrogen bond from Thr³⁸⁷ greatly impacted chemical steps. The rate constants for the Thr³⁸⁷→Ala also exhibited pronounced substrate inhibition at elevated CODD concentrations (Figure 3).

Figure 3.

Steady state kinetics data with varied [CODD]. Reactions included (NH₄)₂Fe(SO₄)₂ (10 µM), ascorbic acid (1 mM), 2OG (100 µM), and CODD (2–40 µM) in 50 mM HEPES pH 7.00, 37.0 °C, ambient [O₂](217 µM).

The initial rate data for Thr³⁸⁷→Ala was fit in two ways in order to adequately treat the inhibition. The first fit used the Michaelis-Menten equation to fit the initial rate data for [CODD] < 10 µM, leading to fitted parameters of k_cat = 33 min⁻¹ and k_cat/K_M(CODD) = 13 µM⁻¹min⁻¹. These kinetic constants were remarkable because they suggested that the active site in WT-PHD2 was arranged to slow turnover.

The second fit used the substrate inhibition model and resulted in k_cat = 44 min⁻¹, k_cat/K_M(CODD) = 11 µM⁻¹min⁻¹, and K_I = 38 µM (Table 2). This second fit reproduced the data well, suggesting that the peptide substrate (CODD) was a stronger inhibitor than 2OG toward the Thr³⁸⁷→Ala variant. As for 2OG inhibition, inhibition by the CODD substrate could arise from many possible mechanisms in which CODD binding would lead to a low-activity enzyme form. A simple model that accounts for the kinetic changes for the Thr³⁸⁷→Ala variant is one in which succinate release is slow, as this could lead to either 2OG or CODD binding to the unreactive (Fe+succ)PHD2 enzyme form. The substrate inhibition by 2OG and CODD implicate a step after O₂ binding/activation as rate-limiting in the Thr³⁸⁷→Ala variant.

Table 2.

Steady state kinetic constants, k_cat and k_cat/K_M(CODD) with varied [CODD].^a

Enzyme	kcat, min−1	kcat/KM(CODD), µM−1min−1	KI, µM	KM, µM
WT-PHD2	1.9 ± 0.1	2.4 ± 0.2	-	0.8 ± 0.1
Thr387→Ala	44 ± 3	11 ± 1	38 ± 6	4.1 ± 0.7
Thr387→Asn	0.9 ± 0.1	0.7 ± 0.1	-	1.4 ± 0.2

^a. Reactions contained (NH₄)₂Fe(SO₄)₂ (10 µM), ascorbic acid (1 mM), 2OG (100 µM), and CODD (2 – 50 µM) in 50 mM HEPES pH 7.00, 37.0 °C, ambient [O₂](217 µM).

Steady state kinetics with varied [O₂]

The increased rates of turnover for the Thr³⁸⁷→Ala variant suggested that steps involved in O₂ activation were faster than for WT-PHD2. The steady-state kinetics as a function of varied O₂ concentration were measured to isolate steps between O₂ binding and oxidative decarboxylation. The initial rate data showed simple saturation kinetics for all but the Thr³⁸⁷→Ala variant, and was fit to the Michaelis-Menten equation (Figure 4). Both WT-PHD2 and the conservative Thr³⁸⁷→Asn variant exhibited sluggish reactivity toward O₂, with the kinetic parameters for WT-PHD2 k_cat/K_M(O2) = 0.015 µM⁻¹min-1 (~ 0.25 × 10⁻⁹ M⁻¹ sec⁻¹) and K_M(O2) = 530 µM consistent with prior reports^28,29. The high K_M(O2) has been rationalized as being necessary for hypoxia sensing, as PHD2 activity would be proportionate to physiologically relevant levels of [O₂].³⁰ The Thr³⁸⁷→Ala variant is a contrast to this, as k_cat/K_M(O2) = 0.3 µM⁻¹ min⁻¹, a 20-fold increase in reaction rate over that of WT-PHD2 which indicates that step(s) involving O₂ activation have markedly increased in rate for this variant. The Thr³⁸⁷→Ala variant also exhibited saturation kinetics up to 220 µM O₂, but with an activity that was only ~25% of maximal at elevated O₂ concentrations (Fig. 4). Attempts at fitting this activity curve to a variety of inhibition models led to unsatisfactory fits, suggesting that the decreased activity at high [O₂] may reflect enzyme inactivation, as is seen for several other 2OG oxygenases.^31–34

Figure 4.

Steady state kinetics as a function of varied [O₂]. Reaction mixtures included (NH₄)₂Fe(SO₄)₂ (10 µM), 2OG (100 µM), CODD (10 µM), and ascorbic acid (1 mM) in 50 mM HEPES pH 7.00. (open circles) WT-PHD2, (closed circles) Thr³⁸⁷→Asn, (triangles) Thr³⁸⁷→Ala.

As our experimental conditions used saturating levels of 2OG and CODD, the macroscopic rate constant (k_cat/K_M(O2)) reports only on steps between O₂ binding and the subsequent irreversible step, which in the consensus mechanism for PHD2 is oxidative decarboxylation. The increased values for k_cat and k_cat/K_M(O2) observed for Thr³⁸⁷→Ala point definitively to an increased rate for oxidative decarboxylation in this variant over that observed for WT-PHD2.

The kinetic parameters for the Thr³⁸⁷→Ala variant are remarkable for two reasons. First, the increased k_cat/K_M(O2) and related mechanistic data indicate that chemistry has become faster as a result of this point mutation, underscoring the enormous influence over oxidative decarboxylation exerted by 2°-sphere contacts in PHD2. This implies that O₂-activation is not just a function of the redox potential at the Fe²⁺ cofactor, but will include the polar environment near the 2OG and the distal O-atom of bound O₂. Second, the Thr³⁸⁷→Ala variant has a lower K_M(O2) than found for WT-PHD2, making for a more sensitive activity response to changes in [O₂], particularly over the physiological range of [O₂] < 50 µM. This opens the possibility of using enzyme delivery³⁵ or engineering approaches to tailor the hypoxia response within cells.

EPR spectra of (Fe²⁺NO) PHD2

According to the consensus mechanism, PHD2 must bind O₂ such that the distal O-atom can attack the 2-oxo position of 2OG. We propose that the Thr³⁸⁷ of WT-PHD2 favors a specific gas binding geometry such that the bound O₂ is not positioned optimally for reaction with 2OG. As NO binds to Fe²⁺ in non-heme enzymes to form a spectroscopically accessible {FeNO}⁷ with S = 3/2, and adopts a bent geometry similar to that of bound O₂^27,36,37,) we used EPR lineshape analysis to assess the impact of Thr³⁸⁷ on bound gas geometry.

The EPR lineshape of {FeNO}⁷ centers in 2OG oxygenases and related enzymes is dominated by the large zero-field splitting,^36–38 with EPR features near g_eff ~ 2 and g_eff ~ 4 that are sensitive reporters of the rhombic zero-field splitting parameter (E/D).²¹ Rhombic zero-field splitting can be directly calculated by use of Eq. 3, relating the observed g_eff values to the E/D ratio. As the zero-field splitting arises from admixture of the low-lying excited states into the ground state, the E/D ratio ultimately reports on the geometry of the {FeNO}⁷ spin center. For example, the EPR lineshape of the {FeNO}⁷ center is very sensitive to the Fe-NO bond angle in sterically constrained models complexes³⁹ as well as equatorial bonding in non-heme proteins^21,37 due to the splitting of the g_eff ~ 4 features caused by rhombic zero-field splitting.

Ferrous nitrosyl complexes of WT-PHD2 and the Thr³⁸⁷→Ala variant were prepared by anaerobically generating NO in the presence of the (Fe+2OG)PHD2/CODD form of each enzyme. The X-band EPR spectra were collected at 4 K in a liquid helium flow cryostat, showing a complex lineshape near g_eff = 4 for both samples (Figure 5). The {FeNO}⁷ center of WT-PHD2 exhibited intense features at g_eff = 4.08, 3.96, which arose from an S=3/2 spin with a highly axial zero-field splitting (E/D = 0.011); as well as a minor feature at g_eff ~ 4.3 that likely arose from adventitious Fe³⁺. In contrast, the {FeNO}⁷ center of Thr³⁸⁷→Ala was more complex, with two S = 3/2 species evident. One species was very similar to that observed for WT-PHD2, with g_eff = 4.09, 3.94 (E/D = 0.014); the second center was much more rhombic, g_eff = 4.33, 3.75 (E/D = 0.48).

Figure 5.

X-band EPR of (Fe²⁺+NO+2OG)PHD2/CODD. PHD2 (100 µM), CODD (100 µM), (NH₄)₂FeSO₄, (100 µM), 2OG (500 µM), and DEANO (500 µM) in 50 mM HEPES pH 7.00. 9.624 GHz frequency; 2.0 mW power; 10 G, 100 GHz modulation, 4 K.

Although it is not possible to ascribe specific geometries to each of these spin centers, the Thr³⁸⁷→Ala mutation lead to a much more heterogeneous {FeNO}⁷ center than found in WT-PHD2. This likely arose from the steric interactions between the bound NO and contacts with the active site environment. Coupled with the observed increase in activity for Thr³⁸⁷→Ala, this suggests that the geometry of bound gas is correlated with O₂ activation such that WT-PHD2 positions the bound gas in a non-optimal orientation for oxidative decarboxylation. In the consensus mechanism O₂ binds as a putative (Fe³⁺ O₂^—) adduct, which then attacks the C-2 keto position of 2OG to initiate oxidative decarboxylation. Should the distal O-atom be stabilized by an electric dipole interaction with the hydroxyl group of Thr³⁸⁷, it could be less nucleophilic in WT-PHD2. Alternately, Thr³⁸⁷ could cause the distal O-atom to be oriented away from the C-2 keto position, thereby lowering the reactivity of the putative superoxide in WT-PHD2.

Implications for hypoxia sensing

Why is turnover so slow for WT-PHD2, and how could removing the hydrogen bonds from Thr³⁸⁷ increase the kinetic parameters for O₂ activation? The answers to both of these questions are speculative at this point, but considering the structural and mechanistic data for PHD2 within the context of the physiological role of this enzyme points to the intriguing possibility that the naturally occurring residue is part of the machinery that limits maximal turnover.

As the physiological role of PHD2 is to turn off gene expression, it is very likely that this enzyme fulfills its role while turning over on the minutes timescale – this lack of evolutionary pressure to increase reaction rates may lead to an active site optimized for something other than rapid turnover. Single turnover experiments indicated that hydroxylated CODD product formed on the minutes timescale (k_obs = 0.018 s⁻¹, 5°C) without any accumulating intermediates.¹⁰ As this rate constant is virtually identical to that observed in the steady state (k_cat = 0.03 s⁻¹, 37 °C)¹⁴, it is very likely that oxidative decarboxylation is rate-limiting in WT-PHD2. This is unusual for 2OG-dependent oxygenases, but may reflect the O₂-sensing from the function of PHD2.

The implication of this work are two-fold. First, contacts within the active site control the O₂ activation rate in 2OG oxygenases. Second, PHD2 can be engineered into a more active form than found in nature, in particular with respect to the reactivity under low [O₂] conditions. Although this was achieved by point mutagenesis in the current report, one can envision the use of small molecules to target the distal hydrogen-bonding in PHD2. Achieving increased turnover under hypoxic conditions would lead to altered cellular hypoxia sensing.

Table 3.

Steady state kinetic parameters using O₂ as the varied substrate.

Enzyme	kcat, min−1	kcat/KM, µM−1min−1	KM, µM
WT-PHD2	8 ± 0.7	0.015 ± 0.001	530 ± 90
Thr387→Ala	29 ± 3	0.3 ± 0.1	100 ± 20
Thr387→Asn	3.8 ± 0.5	0.005 ± 0.001	760 ± 170

[Fe²⁺] = 10 µM, [2OG] = 100 µM, [CODD] = 10 µM, 37.0°C, pH 7.00.

Abbreviations

2OG

2-oxo-glutarate, α-ketoglutarate

AtsK

Alkylsulfatase

FIH

factor-inhibiting HIF

HAT

hydrogen atom transfer

HEPES

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

HIF

hypoxia inducible factor-1α

MALDI-TOF-MS

Matrix assisted laser desorption ionization-time of flight-mass spectrometry

PHD2

prolyl hydroxylase domain 2

TauD

taurine dioxygenase