Human Oxygenase Variants Employing a Single Protein FeII Ligand Are Catalytically Active

Abstract Aspartate/asparagine‐β‐hydroxylase (AspH) is a human 2‐oxoglutarate (2OG) and FeII oxygenase that catalyses C3 hydroxylations of aspartate/asparagine residues of epidermal growth factor‐like domains (EGFDs). Unusually, AspH employs two histidine residues to chelate FeII rather than the typical triad of two histidine and one glutamate/aspartate residue. We report kinetic, inhibition, and crystallographic studies concerning human AspH variants in which either of its FeII binding histidine residues are substituted for alanine. Both the H725A and, in particular, the H679A AspH variants retain substantial catalytic activity. Crystal structures clearly reveal metal‐ligation by only a single protein histidine ligand. The results have implications for the functional assignment of 2OG oxygenases and for the design of non‐protein biomimetic catalysts.


Introduction
2-Oxoglutarate (2OG) and Fe II dependent oxygenase superfamily enzymes are widely distributed and catalyse ar ange of oxidations,t ypically hydroxylations,t hough they catalyse other reactions including hydrogenations,d esaturations,f ragmentations,and ring formations. [1] In humans,they have important physiological roles,f or example in the hypoxic response and epigenetics. [2] Most, but not all, of their reactions comprise two electron substrate oxidations coupled to conversion of 2OG to succinate and CO 2 .T hey have aconserved distorted double-stranded b-helix core fold which supports an active site with as ingle Fe II .T he consensus mechanism for 2OG oxygenases involves an ordered sequential process in which 2OG,then substrate,then O 2 bind at the active site. [3] Structural studies have shown many 2OG oxygenases employ at riad of conserved His,A sp or Glu, His residues (the HXD/E…H motif) to ligate the active site Fe II ,leaving three coordination sites for 2OG,then O 2 to bind to the Fe II ,the former via its oxalyl group. [1a, 4] Aspartate/asparagine-b-hydroxylase (AspH) is ah uman 2OG oxygenase that catalyses C3 hydroxylations of Asp/Asn residues in epidermal growth factor like domains (EGFDs; Figure 1a) [5] -AspH is ap roposed target for cancer treatment. [6] AspH is highly unusual amongst structurally characterised 2OG dependent hydroxylases in employing only two residues,H 679 and H725, to ligate its Fe II (Figure 1b). [7] In this regard, it is related to the 2OG dependent halogenases, where the absence of the Asp/Glu residue of the HXD/E…H motif is used to enable Fe II halide binding. [1b, 4] Here we report the unexpected observation that AspH variants with asingle Fe II histidine ligand retain substantial catalytic activity.T he results have implications for studying the bioinformatical functional assignment of 2OG oxygenases,w hich have Figure 1. AspH catalysesatypical 2OG oxygenase reaction but has an atypical Fe II binding mode, employing two rather than the typical three protein residues. a) The AspH reaction;b)view of the AspH Fe II binding site (H679 and H725 complex the active site metal ion with Mn substituting for Fe ;P DB:6 YYW) [8] compared with that of c) a human 2OG oxygenasewith an Fe II binding triad (i.e. H199, H279, and D201;PDB:1H2L), [9] that is, the asparagineresidue hydroxylase, factor inhibitinghypoxia-inducible transcription factor HIF-a (FIH). extensively relied on the HXD/ E…H triad, and the design of non-protein biomimetic catalysts.

Results and Discussion
Catalytically inactive variants of Fe II ligating residues of the 2OG oxygenases are commonly used in controls during functional assignment work involving cellular studies,i ncluding for AspH. [10] As part of our work on AspH, we therefore made the H679A and H725A variants of our previously used human AspH construct (His 6 -AspH 315-758 ). [7] Therequisite mutations were made by standard methods and the variants were purified to near homogeneity by Ni II -affinity and sizeexclusion chromatography (> 95 % pure by SDS-PAGEa nd MS analysis;Supporting Figure S1). Initially,w ec onducted turnover assays for the AspH variants alongside wildtype (wt) AspH under reported assay conditions,a nalysing hydroxylation of as ynthetic peptide substrate,h FX-EGFD1 86-124 -4Ser (Supporting Figure S2a), [7] which mimics the requisite disulfide isomer of the EGFD1 of the reported AspH substrate human coagulation factor X (hFX). [11] Unexpectedly,t he results revealed that, whilst the variants were less active than wt AspH, they both retained substantial activity:i np articular the H679A variant manifested % 50 %s ubstrate turnover under conditions where wt AspH gave > 95 %t urnover (30 minutes incubation);t he H725A variant was less active (< 5%), but clear evidence for substrate hydroxylation was observed by mass spectrometry (MS) (Supporting Figure S2b). We then carried out more detailed kinetic analyses using our previously reported solid phase extraction coupled to MS (SPE-MS) assay,monitoring substrate depletion/product formation (+ 16 Da), [12] under optimized buffer conditions (Supporting Figure S3). Apparent Michaelis constants (K app m )a nd turnover numbers (k app cat )o fw tA spH for Fe II and 2OG are in good agreement with reported data, which were determined using ad ifferent synthetic cyclic peptide substrate. [12] Thek app cat -values for the H679A AspH variant are marginally lower than those of wt AspH (Table 1), assuming approximately full enzyme activity as suggested by active site titration of wt AspH. [12] By contrast, the k app cat -values for the H725A AspH variant are approximately half those for H679A AspH (Table 1), in agreement with the observed lower substrate hydroxylation in our initial turnover assays (Supporting Figure S2b).
Interestingly,the H679A AspH K app m -values for Fe II ,either in the presence or absence of L-ascorbate (LAA), and the hFX-EGFD1 86-124 -4Ser substrate peptide are,w ithin experimental error, in the range of those of wt AspH (Table 1) and those reported for other 2OG oxygenases. [13] As for wt AspH, [12] the presence of LAA reduces the K app m -value for Fe II , but LAA is not arequirement for productive catalysis by the AspH variants (Supporting Figure S7). Notably,t he H679A and wt AspH K app m -values for 2OG differ substantially,t he former is about two orders of magnitude higher than the latter ( Table 1, entry 3). Thes lightly different assay conditions (0.2 mMH 679A AspH in the absence of NaCl vs.0 .1 mMw t AspH and 50 mM NaCl in the reaction buffer;S upporting Figures S4-S6) likely do not account for this substantial discrepancy.T hough higher than the 2OG K app m -values for wt AspH and other human 2OG oxygenases, [14] the 2OG K app m concentration for H679A is in the range of 2OG concentrations reported in cells (up to > 1mM [15] )and lower than the reported 2OG K app m -value for g-butyrobetaine hydroxylase ( % 150-470 mM). [16] Despite the difference in the K app m -values for 2OG,t he results with H679A AspH imply that the presence of only as ingle polar Fe II -coordinating residue does not preclude efficient hydroxylation-at least in the case of this variant under conditions when 2OG is not limiting. They also reveal the importance of the Fe II chelating histidines in productive 2OG binding.
TheH 725A AspH substitution appears to have am ore pronounced effect than the H679A substitution on the K app m -values for Fe II ,w hich are approximately twofold and threefold higher than those of wt and H679A AspH, respectively (Table 1, entries 1a nd 2). Strikingly,f or H725A AspH, the K app m -value for 2OG is % 200 times greater than that of wt AspH and about double that of H679A AspH (Table 1, entry 3). By contrast, the H725A AspH K m -value for the hFX-EGFD1 86-124 -4Ser substrate peptide is in the range of that of H679A AspH (Table 1, entry 4).
Thek cat /K m -values (specificity constants) highlight the large discrepancies between wt AspH and the two variants with respect to 2OG (Table 1). TheH679A and H725A AspH  Figure S2a).
[f ]Apparent substrate inhibition combinedw ith low detectability of the peptide at low concentrations prevented determination.
k cat /K m -values for 2OG are substantially smaller than that of wt AspH, whereas the k cat /K m -values for Fe II are,within error, similar. NMR turnover assays in the absence of substrate show that H679A and H725A AspH-catalysed oxidative decarboxylation of 2OG is highly coupled to substrate oxidation, demonstrating that the discrepancyi sn ot due to substrate-uncoupled 2OG oxidation (Supporting Figure S8). We further investigated the effects of the His to Ala substitutions on the interactions of Fe II and 2OG with AspH by determining half-maximum inhibitory concentrations (IC 50 -values) of reported 2OG oxygenase inhibitors comprising metal ions and 2OG analogues (Supporting Information Section 2). Mn, Co,N i, and Zn ions are reported wt AspH inhibitors [17] and also inhibit both AspH variants ( Table 2, entries 1-4). In agreement with results for wt AspH, most efficient AspH variant inhibition by the investigated metal ions was observed for Zn II ,w ith Mn II being the least potent. Thereduced potencyofthe metal ions for AspH variant over wt AspH inhibition in part reflects the higher Fe II concentration in the AspH variant inhibition assays (4 mMf or H679A AspH and 10 mMf or H725A AspH vs.2mMf or wt AspH;S upporting Information Section 2). By contrast and consistent with the kinetic analyses,t he inhibition results for the 2OG analogues reveal clear differences for the variants compared to wt AspH. The2 OG-competitive broad-spectrum 2OG oxygenase inhibitor N-oxalylglycine (NOG) potently inhibits wt AspH, [7,17] but does not apparently inhibit either AspH variant (Table 2, entry 5). This discrepancy possibly in part reflects the higher 2OG concentration used with the AspH variant, compared to the wt AspH inhibition assays (110 mMf or H679A AspH and 215 mMf or H725A AspH vs.3 mMf or wt AspH;S upporting Information Section 2). Another broad-spectrum 2OG oxygenase inhibitor, pyridine-2,4-dicarboxylic acid (2,4-PDCA), inhibits the AspH variants,a lbeit less efficiently compared to wt AspH ( Table 2, entry 6). Thep otency of 2,4-PDCAa ppears to decrease with increasing 2OG assay concentration;however, for wt AspH, it has been reported that 2,4-PDCAm aintains its potencya th igh 2OG concentration (IC 50 % 0.1 mMa t 200 mM2 OG vs.I C 50 % 0.03 mMa t3mM2 OG), [18] suggesting that varying the AspH Fe II binding ligands has apronounced effect on the stability of the AspH:Fe II :2OG/2OG analogue complex.
To investigate the structural basis of the kinetic differences between wt AspH and its variants,t he latter were crystallised in the presence of Mn II and 2OG or NOG,with or without the synthetic hFX-EGFD1 86-124 -4Ser substrate peptide.T he structures were solved by molecular replacement using reported structures (PDB ID:5 JTC [17] or 5JQY [7] )a s search models.T he AspH variants crystallised in one of two forms in the P2 1 2 1 2 1 space group (1.6-2.7 resolution; Supporting Figures S10-S20 and Supporting Table S1), in accord with reported wt AspH structures. [7,8,17] Thec rystallisation conditions were found to determine the nature of the AspH variant structures.W hen H679A AspH was crystallised with Mn II and NOG,b ut without the hFX-EGFD1 86-124 -4Ser substrate,ametal ion complex was not obtained. Comparison with aw tAspH structure reveals that the substitution of H679 by alanine neither substantially changes the overall AspH fold nor the conformation of key active site residues engaged in Fe II /2OG binding in wt AspH (Supporting Figure S11). Instead of the anticipated NOG,an acetate from the buffer was present at the active site,with its carboxylate positioned to interact with R735 (2.6 and 3.0 ) and S668 (2.4 )-these residues normally interact with the 2OG C5 carboxylate (Figure 1b). Several active site water molecules,b ut no metal ions,w ere observed, suggesting impaired metal ion binding capacity in the absence of the H679 imidazole (Figure 2a).
To obtain am etal-bound H679A AspH structure,w e added the hFX-EGFD1 86-124 -4Ser substrate to the Mn II /NOG containing crystallisation mixture aiming to stabilise metal ion binding, potentially by additional metal ion coordination with the substrate aspartyl side chain carboxylate,asobserved in some cases for wt AspH:Mn:ligand:hFX-EGFD1 86-124 -4Ser structures, [7,8,17] or by sterically hindering the release of the metal ion and/or 2OG/NOG.C lear electron density for EGFD1 86-124 -4Ser was observed in the resultant H679A AspH structure,w ith the AspH and substrate conformations being similar to those reported (Supporting Figures S14-S20); however, neither aM ni on nor NOG were bound in the active site (Figure 2b). Instead, as observed before with acetate,aformate ion from the buffer was positioned to interact with R735 (2.7 and 2.9 )a nd S668 (2.5 ) ( Figure 2b). Similar results were obtained when H725A AspH was crystallised in the presence of Mn II ,2 OG,a nd hFX-EGFD1 86-124 -Ser,w ith ap ropionate ion being positioned to interact with R735 and S668, though the propionate was observed in two conformations (Figure 2c).
These crystallographic observations correlate with the increased 2OG K m -values of the AspH variants compared to wt [17] 0.03 AE 0.01 H679A 0.13 AE 0.01 H725A 11.2 AE 1.4 [a] Mean of two independent runs (n = 2; mean AE SD). H679A/H725A AspH inhibition assays were performed as described in the Supporting Information Section 2u sing 0.1 mMA spH variant and 4.0 mMhFX-EGFD1 86-124 -4Ser (Supporting Figure S2a);the assays were of good quality as high S/N ratios and Z'-factors [19] (> 0.5 for each plate) manifest (Supporting Figure S9).

Angewandte Chemie
Forschungsartikel wt AspH ( Table 1) and lack of inhibition by NOG (Table 2, entry 5);t hus,u nder conditions where wt AspH crystallises with metal and 2OG/NOG (and, in some cases,h FX-EGFD1 86-124 -Ser), [7,8,17] the latter are outcompeted for binding by abundant carboxylic acids in the crystallisation buffer. The crystallographic studies also unambiguously reveal the presence of only one potential Fe II -binding histidine residue for the H679A and H725A AspH variants (Figure 2a-c). We worked to obtain crystallisation conditions in the absence of carboxylic acids in the crystallisation buffer and obtained aH 679A AspH structure in complex with Mn II , NOG,a nd hFX-EGFD1 86-124 -4Ser (Supporting Figure S18). Thes tructure revealed clear density for the side chains of H725 and A679 (Figure 2d). Superimposition of views of analogous H679A and wt AspH:Mn:NOG:hFX-EGFD1 86-124 -4Ser structures (PDB ID:5 JQY) [7] reveals no substantial alterations in the conformations of H725 or of residues directly engaged in NOG binding (i.e.S668, R688, H690, and R735), or of the Mn ion position (Figure 2f).
In the H679A AspH:Mn:NOG:hFX-EGFD1 86-124 -4Ser complex, the Mn ion is clearly ligated by only one proteinbound ligand, that is,H 725 with the Nt-imidazole-Mn distance being 2.2 ,t he same as in wt AspH, and the A679 methyl carbon-Mn distance being 5.8 (compared to 2.3 for the wt AspH H679 Nt-imidazole-Mn distance;F igure 2d and 3). [7] Water w2 occupies the same coordination site as the single coordinating water in the analogous wt AspH structure ( Figure 3). Strikingly,water w3 occupies the coordination site of the Nt-imidazole nitrogen of H679 in the wt AspH structure (Mn-w3 distance:2 .2 ;F igure 3), apparently compensating for the H679 to alanine mutation. Water w1 coordinates to Mn at ap osition trans to the Nt-imidazole nitrogen of H725, that is,atthe position where O 2 is predicted to bind, [8] in am anner which does not change the relative alignment of the substrate backbone with respect to the reported wt AspH:Mn:NOG:hFX-EGFD1 86-124 -4Ser structure.Asingle conformation for the substrate D103 hFX side chain is observed in the H679A AspH:Mn:NOG:hFX-EGFD1 86-124 -4Ser structure,a pparently positioning its C3 methylene for stereoselective oxidation during catalysis. However,t wo conformations are observed for the D103 hFX side chain carboxylate in the wt AspH:Mn:NOG:hFX-EGFD1 86-124 -4Ser structure as well as in other AspH:Mn:ligand:substrate structures,i ncluding those complexed with 2OG ( Figure 3). [7,8] In one conformation, one of the D103 hFX carboxylate oxygen atoms is positioned to coordinate the Mn ion (2.9 ), that is,itoccupies aposition close to the position where water w1 complexes the Mn ion in the H679A AspH:Mn:NOG:hFX-EGFD1 86-124 -4Ser structure.I nt he wt AspH:Mn:NOG:hFX-EGFD1 86-124 -4Ser structure,t he other D103 hFX side chain carboxylate conformer adopts as imilar conformation as exclusively observed for the D103 hFX side chain carboxylate in the H679A AspH:Mn:NOG:hFX-EGFD1 86-124 -4Ser structure.

Conclusion
Thec ombined solution studies presented here,c learly demonstrate that the H679A and H725A AspH variants retain substantial catalytic activity compared to wt AspH (Table 1a nd Supporting Figure S2). Thec rystallographic analyses provide clear evidence for the presence of as ingle protein (histidine) ligand in each variant. Previous mutation studies with bovine AspH reported no activity for the H675A bovine AspH variant (corresponding to human H679A AspH by sequence alignment [7] ), with reduced activity for the H675D/E variants,which can coordinate metals via their side chains. [20] Surprisingly,t he kinetic analyses revealed comparatively small effects of the H679A and H725A mutations on the  [7] Angewandte Chemie Forschungsartikel K app m -value for Fe II ,that is,the values for wt and H679A AspH are similar and approximately three times lower than the value for H725A AspH (Table 1). Thei nvolvement of other (second sphere) active site residues in Fe II binding is likely, but there is no evidence for direct Fe II chelation by other residues,though this and/or the involvement of the substrate in Fe II binding cannot be entirely ruled out during catalysis. By contrast, the effects of the histidine substitutions on the K app m -value for 2OG are much more substantial, that is,t he values for H679A AspH ( % 109 mM) and H725A AspH ( % 211 mM) are more than two orders of magnitude higher than the value for wt AspH ( % 1.1 mM; Table 1, entry 3). In wt AspH, the H679 imidazole coordinates Fe II trans to the 2OG C1 carboxylate (Figure 1b and Supporting Figure S21). [8] The replacement of this interaction by aw ater (or hydroxide) in H679A AspH may alter backdonation of electron density from the Fe ion to the ligand during 2OG binding or in intermediates during catalysis,r esulting in altered kinetic parameters,i ncluding an increased H679A AspH K app m -value for 2OG.T he H725 imidazole that coordinates Fe II opposite to the proposed O 2 -coordination site (Supporting Figure S21) may similarly regulate the stability of O 2 (and 2OG) derived intermediates,i ncluding the Fe III -2OG-superoxide species in catalysis.
Theobservation of substantial catalytic activity for a2OG oxygenase with as ingle protein-based Fe II ligand has implications for functional assignment work on the superfamily. Theresults show care should be taken in the use of Fe II ligand variants in negative controls in cellular and in vivo work. Studies with isolated enzyme/substrate(s) should be carried out (note that the K app m -values for the AspH variants investigated are in the range of cellular 2OG concentrations [15] ). Work with FIH has shown that substitution of the Asp-residue of the HXD…H motif for Gly,b ut not Ala or Glu, enables retention of some activity (note that AspH also has an HXG…H motif), further supporting the need for experimental determination of al ack of activity. [21] Secondly,b ioinformatics searches for 2OG oxygenases commonly look for DSBH fold proteins with the consensus HXD/E…H motif (or an HXA/G…H motif in the case of 2OG dependent halogenases [4] ). Thek nowledge that they can operate with two histidines as in the case of AspH or only one histidine ligand, substantially expands the set of potential 2OG oxygenases.Indeed, anumber of potential 2OG oxygenases have atypical Fe II binding residues,i ncluding,i nh umans,P HD finger protein 2( PHF2) [22] and hairless; [23] of particular interest is the naturally occurring human AspH variant of unknown function, aspartate b-hydroxylase domain-containing protein 1(AspHD1), which sequence analysis predict has only the distal His of the typical HXD/E…H motif as Fe II ligand, the proximal His (i.e.c orresponding to H679 in wt AspH) is substituted for an Arg residue (Supporting Figure S22). These enzymes are the subject of ongoing studies.
Many attempts to make non-protein biomimetic oxidation catalysts of non-heme Fe II oxygenases have employed 3o r more Lewis basic metal ion ligands. [24] Ther esults presented here suggest that catalysts with 2l igands,o re ven 1l igand, should be viable-we appreciate holding the metal ion on the catalyst will be achallenge-this might be achieved by use of suitable cosubstrates (the use of cosubstrates other than 2OG has been reported to retain the catalytic activity of human 2OG oxygenases [8,25] )orbyuse of metal trapping by ahydrophobic barrier as shown for other types of metal catalysts.