2-Oxoglutarate oxygenases are inhibited by a range of transition metals

Abstract

2-Oxoglutarate oxygenases are inhibited by a range of transition metals, as exemplified by studies on human histone demethylases and prolyl hydroxylase domain 2 (PHD2 or EGLN1). The biological effects associated with 2-oxoglutarate oxygenase inhibition may result from inhibition of more than one enzyme and by mechanisms in addition to simple competition with the Fe(II) cofactor.

---

The 2-oxoglutarate (2OG) oxygenases are a diverse superfamily of Fe(II) dependent enzymes which couple the oxidative decarboxylation of 2OG to the oxidation of protein, lipid, nucleotide or small molecule substrates.¹ All members of the superfamily for which crystal structures are available contain a characteristic Fe(II)-binding His-Glu/Asp-His triad.² Compared to the heme-bound iron in the cytochrome dependent redox enzymes, 2OG oxygenases have a relatively weak affinity for Fe(II), which means that they can be inhibited by other transition metals capable of binding to the His-Glu/Asp-His triad. The activity of many 2OG oxygenases in vitro is stimulated by the addition of ascorbate or other reducing agents.³

Most well-characterised 2OG oxygenases have been shown to be inhibited by at least one transition metal. Examples of 2OG oxygenases inhibited by transition metals include human collagen lysyl hydroxylase (Zn(II)),⁴ the hypoxia inducible factor (HIF) prolyl and asparaginyl hydroxylases (Ni(II), Cd(II), Mn(II) and Co(II)),⁵ as well as bacterial enzymes such as taurine dioxygenase (Ni(II) and Co(II)).⁶ The in vivo effects of certain transition metals have been attributed to the inhibition of 2OG oxygenases. Co(II) ions are known to cause cellular and physiological effects that mimic the hypoxic response in animals.⁷ This effect is proposed to be due to inhibition of the HIF hydroxylases. Recently, 2OG dependent histone demethylases have been shown to be inhibited by transition metals. Changes in global histone methylation levels have also been ascribed to transition metal inhibition of these enzymes.^8–13 The ability of divalent transition metals to compete with active-site Fe(II) in the 2OG oxygenases has also been exploited to form catalytically inert enzyme–metal complexes for crystallographic and spectroscopic analyses.²

Here we report an in vitro investigation on the inhibition of the human JMJD2 histone demethylases (JMJD2A, a histone lysine demethylase which demethylates histone H3K9me3 and H3K36me3, and the highly related demethylase JMJD2E, which demethylates only H3K9me3)^14,15 and, for comparison, the human HIF prolyl hydroxylase PHD2, by divalent metal ions [Mg(II), Ca(II), Mn(II), Co(II), Ni(II), Cu(II), Zn(II) and Cd(II)].‡

Most of the metals tested against PHD2 were relatively weak inhibitors under the incubation conditions, consistent with the relatively high affinity of PHD2 for Fe(II).^16,17 However, Mn(II) was a relatively potent PHD2 inhibitor (IC₅₀ = 21 μM), consistent with data which suggest that it is an effective structural stabilizer of PHD2.¹⁸ Zn(II) and Cd(II) also inhibit PHD2, and previously have been observed to compete with Fe(II) at the active site of this enzyme.¹⁷

The JMJD2 histone demethylases JMJD2A and JMJD2E were inhibited by the divalent transition metals tested, with varying degrees of potency, under standard assay conditions. The divalent alkaline earth metals tested did not inhibit (data not shown). The IC₅₀ values for the divalent transition metals (Table 1) are in the same range as the Fe(II) concentration required for half-maximal activity (10.6 μM, data not shown) (with the exception of Mn(II), which is a weak inhibitor, and Cu(II), which is a potent inhibitor), suggesting that these enzymes have a similar affinity for these metals to that of Fe(II). In general, the histone demethylases appear to be more strongly inhibited by the tested transition metals (with much lower IC₅₀s) than PHD2, including Ni(II) and Co(II). However, Mn(II) displays much lower levels of inhibition of the histone demethylases (IC₅₀ = 101.2 μM and 224 μM for JMJD2A and JMJD2E, respectively) than of PHD2. Zn(II) and Cd(II) display similar inhibition profiles against both of the demethylases screened, likely competing with Fe(II) for binding to the active site (Table 1).

Table 1 IC₅₀ values (μM) for transition metal inhibition of JMJD2A (2 μM), JMJD2E (2 μM), and PHD2 (2 μM) showing the effect of increasing [Fe(II)]. [Ascorbate] = 100 μM. Copurifying iron was not removed from the enzymes prior to incubation. Enzyme, ascorbate and Fe(II) were mixed immediately prior to addition of other transition metals. Data represent the results of three replicate experiments, where the standard error of the mean of log(IC₅₀) was <10%

[Fe(II)]	JMJD2A		JMJD2E		PHD2
	10 μM	50 μM	10 μM	50 μM	20 μM	100 μM
Mn(II)	101.2	N/D	224	N/D	21	37.6
Co(II)	5.7	4.5	9.4	6.3	48.3	296.1
Ni(II)	9.4	5.2	15.6	12.5	185.2	211.2
Cu(II)	0.45	0.7	0.46	0.48	6.6	8.3
Zn(II)	7.8	6.9	13	11.5	9.3	10.7
Cd(II)	7.2	5.5	15.1	12.4	57.4	58

---

We then studied the effects of varying the Fe(II) concentration on the degree of inhibition observed. When relatively low levels of ascorbate (100 μM) were employed, as in the initially used conditions, increasing the Fe(II) concentration (from 10 μM to 50 μM) did not lead to an increase in the IC₅₀ values for the transition metals Co(II), Ni(II), Zn(II) and Cd(II) (Table 1). The lack of an increase in IC₅₀ values may be due (in part) to ascorbate degradation, both by iron present in solution¹⁹ and by the transition metals.²⁰ Fe(II) oxidizes to Fe(III) in oxygenated HEPES buffer at pH 7.5, and Fe(III) has been shown to degrade dehydroascorbate in aqueous solution.¹⁹ In turn, decreased ascorbate levels could lead to depletion of Fe(II) in solution (due to oxidation to Fe(III)). This effect would result in an apparent increase in the potency of inhibition by transition metals at higher iron concentrations, because Fe(III) does not bind to the active site of 2OG oxygenases (unpublished data). To test whether the transition metals may cause inhibition of the enzymes via an ascorbate mediated process, we carried out studies with JMJD2A at higher ascorbate concentration (500 μM).

We found that with 500 μM ascorbate in the incubations, the IC₅₀ values (for JMJD2A) increased when [Fe(II)] increased from 10 to 50 μM (Table 2). The increase in IC₅₀ values is significantly greater for Zn(II), Cd(II) and Mn(II) than for Ni(II) and Co(II), consistent with the previously reported ability of Ni(II) and Co(II) to degrade ascorbate in solution.²⁰

Table 2 IC₅₀ values for metal ion inhibition of JMJD2A at increased ascorbate concentration (500 μM). An increase in IC₅₀ values is observed on increasing [Fe(II)], implying competition for the active site

IC50	10 μM Fe(II)	50 μM Fe(II)
Mn(II)	264.7	425.9
Co(II)	5.3	6.6
Ni(II)	10.3	11.7
Cu(II)	0.5	2.1
Zn(II)	9.3	12.8
Cd(II)	10.2	16.5

---

These results are interesting in the light of cellular studies suggesting that at least some of the effects of Ni(II) and Co(II) in mimicking hypoxia may be due in part to the ability of Ni(II) and Co(II) to degrade ascorbate in vivo, leading to decreased activity of PHD2.²⁰ In support of this, it has been shown that Cr(VI), which does not inhibit PHD2 by binding to the active site, still inhibits enzymatic activity by depletion of cellular ascorbate levels.²¹

Our currently used recombinant forms of JMJD2A and JMJD2E contain an N-terminal His₆Tag (sequence MHHHHHHSSGVDLGTENLYFQS) that enables the production of stable and active enzymes. We found that addition of a peptide with a HisTag sequence (HHHHHHSSG) at 5 μM increased the IC₅₀ values for Ni(II) (from 10.1 to 17.1 μM) and Co(II) (from 5.6 to 8.9 μM), but not for Cd(II) (values of 7.0 and 7.1 (from 10.1 to 17.15 μM) were observed with and without the peptide), for JMJD2A (at 2 μM). These observations demonstrate that the HisTag can interfere with the assay results and that metal binding sites other than at the active site to potentially affect inhibition studies. The latter is relevant to wild type forms of both the JMJD2 histone demethylases and PHD2, all of which contain zinc as well as Fe(II) binding sites.^17,22

It should also be noted that Cu(II) inhibited both JMJD2A and JMJD2E at sub-stoichiometric levels, implying that its mechanism of inhibition was not merely competition for the active site. Other studies on the effects of Cu(II) in biological systems have shown that it is capable of generating reactive oxygen species (ROS) capable of protein damage.²³ Inhibition assays where superoxide dismutase was added to the reaction mixture (at 2 μM concentration) showed an increase in the IC₅₀ for Cu(II) from 0.5 μM to 3 μM, consistent with this mechanism; however, the possibility of Cu(II) binding to the superoxide dismutase active site cannot be ruled out as a reason for the increased IC₅₀.

In conclusion, our results, together with those previously reported^20,21,24 imply that the inhibition of 2OG oxygenases in cells does not necessarily occur solely by simple competition with Fe(II) for active site binding, though this may be important. Possible alternative mechanisms of inhibition of activity in cells include (i) degradation of ascorbate/redox-sensitive cofactors, (ii) oxidative damage to the enzyme (as demonstrated for some 2OG oxygenases/related enzymes)²⁵ and (iii) binding to metal binding sites away from the active site (both full length PHD2¹⁷ and some histone demethylases²² contain zinc binding sites). Importantly, the results demonstrate that it is likely that most 2OG oxygenases will be inhibited by transition metals, though to different extents. Thus, the assignment of the cellular and physiological effects of particular transition metals to the inhibition of specific enzymes, without direct evidence for this mechanism in cells, should be treated with caution. It is therefore interesting that array analyses reveal substantial, but imperfect, concord between the effect of hypoxia, Co(II) ions and a non-specific 2OG oxygenase inhibitor in human cells.⁷ It is possible that the partial hypoxia mimicry of Co(II) ions and certain non-specific inhibitors in terms of the regulation of expression emerges from multiple interactions with 2OG oxygenases and other molecules, of which inhibition of the HIF hydroxylases by competition with Fe(II) is one important component. We note that an analogous argument may apply to the regulation of expression by oxygen itself.

Notes and references

1. R. P. Hausinger, Crit. Rev. Biochem. Mol. Biol., 2004, 39, 21.
2. I. J. Clifton, M. A. McDonough, D. Ehrismann, N. J. Kershaw, N. Granatino and C. J. Schofield, J. Inorg. Biochem., 2006, 100, 644.
3. R. E. Saari and R. P. Hausinger, Biochemistry, 1998, 37, 3035.
4. U. Puistola, T. M. Turpeenniemi-Hujanen, R. Myllylä and K. I. Kivirikko, Biochim. Biophys. Acta, Enzymol., 1980, 611, 51.
5. M. Hirsila, P. Koivunen, L. Xu, T. Seeley, K. I. Kivirikko and J. Myllyharju, Faseb. J., 2005, 19, 1308.
6. E. Kalliri, P. K. Grzyska and R. P. Hausinger, Biochem. Biophys. Res. Commun., 2005, 338, 191.
7. A. Vengellur, J. M. Phillips, J. B. Hogenesch and J. J. LaPres, Physiol. Genomics, 2005, 22, 308.
8. H. Chen and M. Costa, BioMetals, 2009, 22, 191.
9. H. Chen, N. C. Giri, R. Zhang, K. Yamane, Y. Zhang, M. Maroney and M. Costa, J. Biol. Chem., 2010, 285, 7374.
10. H. Chen, Q. Ke, T. Kluz, Y. Yan and M. Costa, Mol. Cell. Biol., 2006, 26, 3728.
11. Q. Li, Q. Ke and M. Costa, Carcinogenesis, 2009, 30, 1243.
12. H. Sun, X. Zhou, H. Chen, Q. Li and M. Costa, Toxicol. Appl. Pharmacol., 2009, 237, 258.
13. X. Zhou, Q. Li, A. Arita, H. Sun and M. Costa, Toxicol. Appl. Pharmacol., 2009, 236, 78.
14. R. J. Klose and Y. Zhang, Nat. Rev. Mol. Cell Biol., 2007, 8, 307.
15. N. R. Rose, S. S. Ng, J. Mecinovic, B. M. Lienard, S. H. Bello, Z. Sun, M. A. McDonough, U. Oppermann and C. J. Schofield, J. Med. Chem., 2008, 51, 7053.
16. L. A. McNeill, E. Flashman, M. R. Buck, K. S. Hewitson, I. J. Clifton, G. Jeschke, T. D. Claridge, D. Ehrismann, N. J. Oldham and C. J. Schofield, Mol. BioSyst., 2005, 1, 321.
17. J. Mecinovic, R. Chowdhury, B. M. Lienard, E. Flashman, M. R. Buck, N. J. Oldham and C. J. Schofield, ChemMedChem, 2008, 3, 569.
18. R. Chowdhury, M. A. McDonough, J. Mecinovic, C. Loenarz, E. Flashman, K. S. Hewitson, C. Domene and C. J. Schofield, Structure, 2009, 17, 981.
19. J. Mecinovic, R. B. Hamed and C. J. Schofield, Angew. Chem., Int. Ed., 2009, 48, 2796.
20. K. Salnikow, S. P. Donald, R. K. Bruick, A. Zhitkovich, J. M. Phang and K. S. Kasprzak, J. Biol. Chem., 2004, 279, 40337.
21. M. Kaczmarek, O. A. Timofeeva, A. Karaczyn, A. Malyguine, K. S. Kasprzak and K. Salnikow, Free Radical Biol. Med., 2007, 42, 1246.
22. Z. Chen, J. Zang, J. Whetstine, X. Hong, F. Davrazou, T. G. Kutateladze, M. Simpson, Q. Mao, C. H. Pan, S. Dai, J. Hagman, K. Hansen, Y. Shi and G. Zhang, Cell, 2006, 125, 691.
23. R. Munday, C. M. Munday and C. C. Winterbourn, Free Radical Biol. Med., 2004, 36, 757.
24. M. Kaczmarek, R. E. Cachau, I. A. Topol, K. S. Kasprzak, A. Ghio and K. Salnikow, Toxicol. Sci., 2008, 107, 394.
25. J. N. Barlow, Z. Zhang, P. John, J. E. Baldwin and C. J. Schofield, Biochemistry, 1997, 36, 3563.
26. E. Flashman, E. A. L. Bagg, R. Chowdhury, J. Mecinovic, C. Loenarz, M. A. McDonough, K. S. Hewitson and C. J. Schofield, J. Biol. Chem., 2007, 283, 3808.

---

Footnotes

† Present address: Department of Chemistry and Chemical Biology, Harvard University, 12 Oxford St, Cambridge, MA, 02138, USA.

‡ JMJD2A, JMJD2E and PHD2 were expressed in E. coli as His₆-tagged proteins, and purified as described.^15,26 Peptide substrates (ARKme3-STGGK for demethylases and DLDLEMLAPYIPMDDDFQL for PHD2) and the HisTag peptide (HHHHHHSSG) were synthesised as C-terminal amides on a CSBio automated peptide synthesiser, and purified by RP-HPLC, as described.¹⁵ Typical assay mixtures consisted of 2OG, peptide substrate, enzyme, sodium ascorbate, ferrous ammonium sulfate, and transition metal inhibitor, with concentrations varying as appropriate. Assays were initiated by addition of 2OG and peptide, and were quenched with MeOH (1 : 1 v/v), before cocrystallising assay mixture with MALDI-TOF matrix (α-cyano-4-hydroxycinnamic acid) for MS analysis of products, as described.¹⁵ For the experiments to assess the affect of the His-tag, JMJD2A (2 μM) was incubated in the presence and absence of HHHHHHSSG peptide (5 μM). Serial dilutions of Ni(II), Co(II) and Cd(II); Fe(II) (10 μM) were employed with ascorbate (100 μM), 2OG (50 μM) and K9me₃ peptide substrate (10 μM) using a 30 min incubation at 37 °C (final volume 20 μL); the reaction was quenched with 20 μL methanol.

---