FIH (Factor inhibiting hypoxia-inducible factor), an asparaginyl β-hydroxylase belonging to the super-family of 2-oxoglutarate and Fe(II)-dependent dioxygenases, catalyses hydroxylation of Asn-803 of hypoxia-inducible factor, a transcription factor that regulates the mammalian hypoxic response. Only one other asparaginyl β-hydroxylase, which catalyses hydroxylation of both aspartyl and asparaginyl residues in EGF (epidermal growth factor)-like domains, has been characterized. In the light of recent crystal structures of FIH, we compare FIH with the EGFH (EGF β-hydroxylase) and putative asparagine/asparaginyl hydroxylases. Sequence analyses imply that EGFH does not contain the HXD/E iron-binding motif characteristic of most of the 2-oxoglutarate oxygenases.
The activity of HIF (hypoxia-inducible factor), a transcription factor that allows mammalian cells to respond to hypoxia by enabling the expression of an array of genes, is blocked by its post-translational hydroxylation. Two types of hydroxylation have been identified in human HIF-α: at two prolines located in oxygen-dependent degradation domains and at an asparaginyl residue in a transactivation domain. Prolyl-4-hydroxylation is mediated by prolyl-4-hydroxylase enzymes [PHDs (prolyl hydroxylase domains) or EGLNs] and asparaginyl hydroxylation is mediated by FIH (factor-inhibiting HIF). Both the PHDs and FIH have been identified as members of the superfamily of 2OG (2-oxoglutarate) and Fe(II)-dependent dioxygenases. Hydroxylation of either Pro-402 or Pro-564 causes degradation of HIF-1α via the proteasome pathway and hydroxylation at Asn-803 blocks the interaction between HIF and the transcriptional co-activator p300. The active signalling state of HIF arises as HIF-α molecules synthesized during periods of hypoxia are not subjected to modification by either FIH or the PHDs, and so escape both degradation and inactivation. On recovery of normoxia, the HIF hydroxylases restore the inactive signalling state. Because of their role in controlling the activity of HIF, these hydroxylases are implicated as cellular oxygen sensors (see [1–4] for reviews).
The 2OG oxygenases couple the oxidation of their prime substrate with that of 2OG, which is converted into succinate and carbon dioxide (see [5,6] for reviews). Crystal structures are available for FIH [7,8] and a FIH/HIF-1α 775–826 complex . These structures reveal that FIH is homodimeric and possesses the DSBH (double-stranded β-helix or jelly roll) core-motif, comprising eight β-strands, which is characteristic of the 2OG dioxygenases. Recent studies of FIH have also shown that the length of the HIF substrate  and the sequence surrounding the site of hydroxylation  are important in determining catalytic efficiency. Furthermore, the homodimeric form of FIH appears to be necessary for productive-substrate binding and catalysis ; a single-point mutation (L340R, Leu340→Arg) was found to be sufficient to disrupt the dimer interface and to ablate catalysis. Sequence analyses not only imply that FIH is one of a distinct subfamily of 2OG oxygenases, but is also part of the JmjC family of transcription factors and putative transcription factors. The family is characterized by the presence of a DSBH and it has been proposed that a significant number of its members may be hydroxylases involved in signalling .
Possible mechanisms for the reversal of asparaginyl hydroxylation
To date, no evidence has been reported for the direct reversal of HIF-α hydroxylation. The HIF hydroxylases themselves almost certainly cannot catalyse the direct reduction of hydroxylated residues, at least via a process involving reformation of dioxygen, but further modification of HIF-α to effect a net reversal is possible. Phosphorylation of HIF-1α at Thr-796 was shown to be necessary for HIF-mediated transcription  and may serve to prevent HIF asparaginyl hydroxylation by FIH . Additionally, it may be that further modification at hydroxylated Asn-803 also occurs. One chemically plausible modification is the elimination of water via the action of a (reversible) dehydratase to form a dehydroasparagine residue. This may then be reduced by, for example, an NADPH enzyme. A precedent for this proposal comes from the observation that a serine racemase homologue from Saccharomyces cerevisiae has L-threo-3-hydroxyaspartate dehydratase activity . Another possible means of reversal or negation of the effects of hydroxylation is the retro-aldol reaction of β-hydroxyasparagine to give 2-oxoacetamide and a glycine residue. Such a process has precedent with a D-3-hydroxyaspartate aldolase, recently isolated from Paracoccus denitrificans . In the case of further modification of 4-hydroxyprolyl residues, ‘reversal’ by direct reduction or via formation of dehydroproline, is possible but has not been observed. In this case, a retro-aldol reaction is unlikely on chemical grounds.
Other asparaginyl hydroxylases and asparagine hydroxylases
FIH is not the only mammalian enzyme known to catalyse hydroxylation of asparaginyl residues. Following the identification of β-hydroxylated asparagine and aspartate residues in EGF (epidermal growth factor) domains of proteins including the coagulation factors VII, IX and X , a human aspartyl (asparaginyl) β-hydroxylase was identified [EGFH (EGF β-hydroxylase)]. Site-directed muatgenesis studies have identified at least one histidine residue involved in iron binding . Unlike FIH, the EGF hydroxylase accepts both asparaginyl and aspartyl residues as substrates for hydroxylation. The physiological role of EGF hydroxylation is yet to be determined; it has been proposed that it is involved in Notch-pathway signalling and there is evidence that lack of it may promote tumour formation in mice . Whereas FIH catalyses β-hydroxylation to give the threo-stereoisomer (i.e. the 2S, 3S product) at Asn-803, EGFH catalyses erythro-β-hydroxylation (i.e. to give the 2S, 3R product) of both aspartyl and asparaginyl residues  (Scheme 1).
(a) Stoichiometry of a 2OG oxygenase catalysed hydroxylation; stereochemistry of the reactions catalysed by FIH (b), EGFH (c), proposed reaction for AsnO  and the CAS catalysed reaction (d)
Oxidation of amino acids and peptides are common themes in the biosynthesis of both peptide- and non-peptide-based secondary metabolites produced by microorganisms. Oxidation can occur at the free amino acid level, such as proline to trans-4-hydroxy-L-proline (2S, 4R) as catalysed by proline-4-hydroxylase from Streptomyces griseoviridus . Subsequently, the oxidized amino acid can be incorporated into a peptide-derived secondary metabolite via the action of a non-ribosomal peptide synthetase. Alternatively, oxidation can occur after incorporation of the amino acids into an intermediate. In contrast with known examples of side-chain oxidation by 2-oxoglutarate oxygenases in higher organisms, in microorganisms, the oxidation reactions are not limited to hydroxylation. Various unusual oxidation reactions have been observed including the remarkable oxidative ring closures involved in the biosynthesis of the β-lactam antibiotics (see  for a review).
β-Hydroxyasparaginyl and β-hydroxyaspartyl residues have been observed in a number of non-ribosome-derived peptide antibiotics. The biosynthetic gene cluster for the calcium-dependent lipopeptide antibiotics from Streptomyces coelicolor, which contains a threo-β-hydroxy-D-asparagine (2R, 3S) residue, has recently been analysed by Hojati et al. . It contains an open reading frame (asnO) that displays sequence similarity with the 2OG oxygenase CAS (clavaminic acid synthase) . CAS catalyses three reactions during the biosynthesis of clavulanic acid (a clinically important β-lactamase inhibitor), one of which involves β-hydroxylation of an arginine derivative. The apparent involvement of β-hydroxylated aspartyl and asparaginyl residues in calcium binding by both EGF domains and calcium-dependent antibiotics is striking especially as in both cases they are biosynthesized by the action of 2OG oxygenases. CAS, FIH and, in likelihood, AsnO form part of a set of 2OG oxygenases characterized by the presence of an insert between the fourth and fifth strands of their DSBH core . Although previous alignments involving the C-terminal region of EGFH have demonstrated similarity with procollagen prolyl and lysyl hydroxylases , it has not been clear as to whether EGFH contains a DSBH core. Sequence alignments produced using the threading method of 3DPSSM  (Figure 1) clearly indicate the presence of a DSBH in EGFH and unexpectedly reveal that EGFH shows similarity with bacterial proline-3-hydroxylase from Streptomyces sp. . However, the multidomain EGFH does not contain the C-terminal helical domain of proline-3-hydroxylase; but it does contain a tetratricopeptide domain N-terminal to the DSBH. Such tetratricopeptide domains are found in JmjC proteins such as the ubiquitously transcribed X and Y chromosome tetratricopeptide repeat proteins (UTX and UTY). Sequence alignments also support mutagenesis studies implying that His-675 acts as an iron ligand for EGFH. Similarly, by analogy with proline-3-hydroxylase and other 2OG oxygenases, it would be expected that His-722 (Figure 1, arrow 3) also acts as an iron ligand and that Arg-733 (Figure 1, arrow 4) is involved in binding the 5-carboxylate of 2OG. An unusual feature of EGFH is the lack of an HXD motif on the second strand of the DSBH (Figure 1, arrows 1 and 2). This is interesting because, to date, all 2OG oxygenases and many other non-haem oxygenases contain a facial triad of 2-His, 1-carboxylate metal-binding residues . It may be that EGFH has only two iron ligands or that the third iron ligand comes from a different part of the structure to structurally characterized 2OG oxygenases. It is possible that the HXD/E facial triad of iron-binding residues is not universally present in the 2OG oxygenases.
Partial sequence alignment of the C-terminal domain of EGFH with proline-3-hydroxylase
Genes: Regulation, Processing and Interference: A Focus Topic at BioScience2004, held at SECC Glasgow, U.K., 18–22 July 2004. Edited by I. McEwan (Aberdeen, U.K.), B. White (Glasgow, U.K.), S. Graham (Glasgow, U.K.), S. Roberts (Manchester, U.K.), A. Sharrocks (Manchester, U.K.), D. Black (Organon, U.K.), S. Newbury (Oxford, U.K.), J. Sayers (Sheffield, U.K.) and A. Lloyd (University College London, U.K.).
We thank the BBSRC, Wellcome Trust and EU for funding and all our colleagues for their efforts on hydroxylases.