Commentary


Nature Chemical Biology 4, 152 - 156 (2008)
doi:10.1038/nchembio0308-152

Expanding chemical biology of 2-oxoglutarate oxygenases

Christoph Loenarz1 & Christopher J Schofield1

  1. Christoph Loenarz and Christopher J. Schofield are in the Chemistry Research Laboratory and the Oxford Centre for Integrative Systems Biology, Mansfield Road, Oxford OX1 3TA, UK. e-mail: christopher.schofield@chem.ox.ac.uk


Beyond established roles in collagen biosynthesis, hypoxic signaling and fatty acid metabolism, recent reports have now revealed roles for human 2-oxoglutarate–dependent oxygenases in histone and nucleic acid demethylation and in signaling protein hydroxylation. The emerging role of these oxygenases in enabling a multiplicity of histone modifications has some analogy with their role in enabling structural diversity in secondary metabolism.


In 1967, Hutton et al. reported an oxygenase-catalyzed reaction that requires 2-oxoglutarate (2OG) while working with collagen prolyl hydroxylase (CPH)1. Since this discovery it has been found that 2OG oxygenases are a large superfamily that is involved in a wide range of biological roles (Fig. 1). In almost all cases 2OG oxygenases couple the two-electron oxidation of substrate to the oxidative decarboxylation of 2OG to give succinate and carbon dioxide. All identified 2OG oxygenases use Fe(II) as a cofactor, and some such as CPH require ascorbate for optimal activity2, 3. Crystallographic studies have revealed a double-stranded beta-helix (or jelly roll) fold that supports a highly but not universally conserved Fe(II)-binding His-Xxx-Asp/Glu...His triad motif (Fig. 2). 2OG-binding residues are less well conserved and are characteristic of different subfamilies. Substrate-binding residues vary and may involve mobile elements surrounding the active site4.

Figure 1: Known and proposed roles for human 2OG oxygenases with outline catalytic cycle.

Figure 1 : Known and proposed roles for human 2OG oxygenases with outline catalytic cycle.

See main text for identified roles of individual 2OG oxygenases; the requirement for O2, 2OG, Fe(II) and the coproducts succinate and CO2 are omitted for clarity. Color code: red, incorporated oxygen; blue, methyl residue acted upon. CPH, collagen proline (4R)-hydroxylase; P3H, proline (3S)-hydroxylase; LH, lysine hydroxylase; TMLH, trimethyllysine hydroxylase; GBBH, gamma-butyrobetaine hydroxylase; PhyH/PAHX, phytanoyl-CoA hydroxylase; PHD1–PHD3, prolyl hydroxylase domain–containing proteins 1–3; FIH, factor inhibiting HIF; EGFH/AAH, epidermal growth factor hydroxylase; hABH2/3, AlkB homologs 2 and 3; FTO, fat mass and obesity associated; JMJD, jumonji domain–containing; R, methyl or hydrogen.

Full size image (84 KB)

Figure 2: Overall folds and active sites of human 2OG oxygenases.

Figure 2 : Overall folds and active sites of human 2OG oxygenases.

(a) FIH, the human asparaginyl hydroxylase involved in the oxygen sensing mechanism, with HIF-1alpha Asn803 (ref. 8; Protein Data Bank (PDB) code 1H2K). (b) Human JMJD2A, which is involved in demethylation of methyllysyl histone residues26 (Ni(II) substituting for Fe(II); PDB code 2OQ6). (c) AsnO, the asparagine hydroxylase from Streptomyces coelicolor A3(2) involved in secondary metabolism29 (PDB code 2OG7). Color codes: dark blue, DSBH fold; light blue, helix; white, loop; green, substrate/product; gold, Fe(II)-binding residues. NOG, N-oxalylglycine (2OG analog).

Full size image (66 KB)

The range of different types of reactions that the 2OG oxygenases are able to catalyze is among the most diverse of any enzyme family5. Diversity is also an established feature of the secondary metabolism in plants and microorganisms in which 2OG oxygenases catalyze hydroxylations as well as a wide range of other oxidative reactions involving desaturation, ring closure, epimerization and chlorination. Oxidases that do not use 2OG as cosubstrate but are structurally homologous to the 2OG oxygenases occur in the unique bicyclization reaction involved in penicillin biosynthesis from a tripeptide and in the formation of the plant signaling molecule ethylene from 1-aminocyclopropane-1-carboxylic acid2, 3. In contrast, in animals the range of identified chemistry catalyzed by 2OG oxygenases is limited to hydroxylation and demethylation initiated by hydroxylation.

After the pioneering work on CPH, 2OG oxygenases were shown to carry out other roles in animals, from fatty acid metabolism6, 7 to hypoxic signaling3, 8. However, sequence analyses suggest that in humans there are >60 2OG oxygenases; for many there remains little or no functional knowledge. Given the wide range of chemical transformations catalyzed by these enzymes in plants and microorganisms, it is likely that elucidating the functions of the remaining human 2OG oxygenases will reveal new biological roles and potentially an expanded range of chemical reactions. Here we describe the known biochemical functions of human 2OG oxygenases, focusing on newly identified roles.

Biosynthesis, metabolism and O2 sensing

A number of biological roles for this enzyme family have already been defined. Important post-translational modifications to collagen are known to be catalyzed by two types of 2OG oxygenases. CPH catalyzes the trans prolyl-4-hydroxylation of Xxx-Pro-Gly motifs within collagen (Fig. 1), a modification that stabilizes the collagen triple helix via operation of the gauche effect9. CPH is an alpha2beta2 heterodimer with the beta-subunit identical to protein disulfide isomerase. CPH activity is strongly dependent on ascorbate (vitamin C), a lack of which in the human diet is known to result in scurvy, involving breakdown of connective tissue. Collagen also undergoes prolyl-3-hydroxylation catalyzed by prolyl-3-hydroxylases (or leprecans), which requires prior collagen 4-hydroxylation. In contrast to 4-hydroxylation, prolyl-3-hydroxylation destabilizes the collagen triple helix, possibly allowing other modifications to occur. Hydroxylation at C5 of lysyl residues in collagen can serve as an anchor point for glycosylation. One form of human lysyl hydroxylase is bifunctional, also having glucosyltransferase activity whereas the other forms only have hydroxylase activity. Mutations to the genes encoding collagen hydroxylase also result in connective tissue disease10.

2OG oxygenases have two distinct roles in fatty acid metabolism. Trimethyllysine hydroxylase (TMLH) and gamma-butyrobetaine hydroxylase (GBBH) catalyze the first and last steps, respectively, in carnitine biosynthesis from Nepsilon-trimethyllysine7 (Fig. 1). Second, phytanoyl coenzyme A hydroxylase (PhyH or PAHX) catalyzes an important step in the alpha-oxidation degradation pathway for phytanic acid6. Phytanic acid, derived from the side chain of chlorophyll, cannot be processed through the normal fatty acid beta-oxidation pathway because it has a beta-methyl group; hence an initial alpha-oxidation pathway occurs in the peroxisome to remove one carbon. Following formation of phytanoyl-CoA as a mixture of C2 epimers, PhyH catalyzes C2 hydroxylation to give both threo stereoisomers of hydroxyphytanoyl-CoA (Fig. 1). Lesions to the gene encoding PhyH lead to Refsum disease resulting from an accumulation of phytanic acid.

The alpha,beta-heterodimeric hypoxia-inducible transcription factor (HIF) enables a complex and coordinated response to hypoxia involving an array of more than 100 genes, including those encoding for glycolysis enzymes and proteins involved in regulating blood cell levels and blood vessel formation8. Both the activity and degradation rate of the HIF-alpha subunit are regulated by its post-translational hydroxylation as catalyzed by 2OG oxygenases (Fig. 1). The requirement of oxygen for these 2OG oxygenases as a cosubstrate enables animals to respond to hypoxia. Hydroxylation of the HIF-alpha subunit at either of two conserved prolyl residues by prolyl hydroxylases (PHD) enables it to bind to the von Hippel-Lindau protein elongin C/B complex, which targets HIF-alpha for ubiquitin tagging and proteasomal degradation. In addition, hydroxylation by asparaginyl hydroxylases (factor inhibiting HIF or FIH) of an asparaginyl residue in the C-terminal transcriptional activation domain of HIF-alpha (Fig. 2a) reduces its interaction with the transcriptional coactivator protein p300.

FIH is not the only human asparaginyl hydroxylase. After the identification of beta-hydroxylated asparaginyl and aspartyl residues in epidermal growth factor (EGF) domains, including in blood coagulation factors, an EGF hydroxylase was isolated11. Sequence and mutagenesis analyses suggest that the EGF hydroxylase may not have the typical His-Xxx-Asp/Glu...His Fe(II) binding motif. Unlike FIH, the EGF hydroxylase accepts both aspartyl and asparaginyl residues as substrates; hydroxylation occurs to give the (2S,3R) products, in contrast to FIH, which gives the (2S,3S) product (Fig. 1). The biological role of the EGF hydroxylase is unknown, though it may be involved in signaling and its loss in mice promotes developmental defects and tumor growth11.

Hydroxylation of ankyrin repeats

Recent efforts have focused on finding alternative substrates for the HIF hydroxylases and have provided evidence that the hydroxylation of cytoplasmic proteins is an important, relatively unrecognized post-translational modification. Although several potential substrates for the HIF prolyl hydroxylases have been identified, none as yet has been reported to be an endogenous hydroxylation substrate8. In contrast, recent work has provided strong evidence that the single human form of FIH has at least several substrates other than HIF. Interaction screens identified ankyrin repeat domain (ARD)-containing proteins from the NFkappaB signaling system (inflammatory response) as FIH binding partners12. Assays in vitro and in cells demonstrated that conserved asparaginyl residues from ARD proteins were FIH substrates, including at endogenous levels (Fig. 1). FIH also catalyzes asparaginyl hydroxylation within ARDs of endogenous Notch receptors13. Further evidence for ubiquitous FIH-mediated ARD hydroxylation comes from the report that ASB4 (ankyrin repeat and SOCS box protein 4) is hydroxylated by FIH. It is proposed that ASB4 hydroxylation enables it to target proteins for ubiquitin-mediated degradation in an oxygen-dependent fashion in an analogous manner to HIF-alpha prolyl hydroxylation14.

ARD proteins from both NFkappaB and Notch systems were found to compete with HIF-alpha for FIH-dependent hydroxylation and in most cases were actually better substrates than HIF-alpha in vitro. Structures of Notch peptides bound to FIH imply that significant conformational changes are required in order for the ARD fold to bind productively at the FIH active site. Asparaginyl hydroxylation decreased ARD binding to FIH; the hydroxylation status of ARD proteins that are FIH accessible may regulate the amount of FIH available to hydroxylate HIF-alpha. It is now important to determine whether ARD hydroxylation has a common function within different systems—for example, protein stabilization—or whether, like post-translational modifications such as prolyl hydroxylation and phosphorylation, it has different roles in different contexts.

From DNA repair to obesity

The formation of 5-methylcytosine from cytosine is a well-characterized epigenetic modification to DNA. Other types of DNA methylation have been normally perceived to be toxic. In Escherichia coli the 2OG oxygenase AlkB catalyzes repair of N-alkylated single-stranded DNA caused by alkylating agents15. The AlkB reaction proceeds via methyl hydroxylation followed by fragmentation with concomitant release of formaldehyde, which is similar to the proposed histone demethylase mechanism. 1-Methyladenine, 3-methylcytosine and (less efficiently) 1-methylguanine and 3-methylthymine are demethylated by AlkB catalysis. AlkB also catalyzes dealkylation of DNA alkylated with larger groups, including etheno DNA modifications produced by lipid peroxidation adducts.

Eight human AlkB homologs (ABH1–ABH8) have been proposed; so far two (ABH2 and ABH3) are reported to have DNA demethylation activity15 (Fig. 1). In contrast to ABH3 and AlkB, ABH2 prefers double-stranded DNA substrates. ABH2 and ABH3 also have been reported to display activity with methylated RNA. ABH8 has a predicted RNA-binding motif and a methyltransferase domain. ABHs are ubiquitously expressed in human tissues; ABH1, ABH3, ABH4, ABH6 and ABH7 are localized in both the cytoplasm and nuclei, whereas ABH8 is located only in the cytoplasm. Mice lacking either or both of the Alkbh2 or Alkbh3 genes are viable. Without exposure to alkylating agents, Alkbh2 null mice but not Alkbh3 null mice accumulate 1-methyladenine in their livers, which is consistent with a physiological role for ABH2 in DNA repair15, 16. This does not exclude other roles, including the possibility that ABHs act in an oxygen-dependent signal regulation.

One unexpected link between 2OG oxygenase–dependent nucleic acid demethylation and physiology has come from obesity research. Recent studies have revealed a strong association between common human variants in the first intron of the fat mass– and obesity-associated FTO gene and obesity in both children and adults, with 16% of some populations being homozygous for the risk allele17. As adults, these individuals weigh about 3 kg more than those homozygous for the low-risk allele, due to increased fat mass. Bioinformatic analysis of the predicted FTO sequence led to the proposal that it is a 2OG oxygenase related to the ABHs18. Mouse FTO localizes to the nucleus and catalyzes the demethylation of N-methylated DNA, much like the ABHs18 (Fig. 1). Under the reported assay conditions, FTO had a different selectivity to ABH2 and ABH3; it preferentially catalyzes demethylation of 3-methylthymine and (less efficiently) 1-methyladenine and 3-methylcytosine in single-stranded DNA. FTO mRNA levels are high in the brain and are reported to be regulated by feeding and fasting, which is consistent with a role in regulating feeding or metabolism. Establishing whether the demethylation activity of FTO is linked to increased fat mass is now an important objective.

Dynamic histone modifications

Eukaryotic DNA is condensed in the form of nucleosomes comprising 146 base pairs folded around a core consisting of two copies of four histone proteins (H2A, H2B, H3 and H4). Covalent modifications to the tails of these histones regulate the organization and availability of the DNA in nucleosomes. The N-terminal tails of the core histone proteins are modified in a combinatorial and dynamic manner by modifications including acetylation, methylation, formylation and phosphorylation (Fig. 3a). Histone lysyl residues can be specifically recognized in their methylated state by Royal family (chromo, tudor, MBT domains) and plant homeo domain–containing proteins19. Although certain types of modifications appear to be associated with specific biological functions (that is, lysyl acetylation and arginyl methylation mostly activate transcription), it seems unlikely that there is a simple conserved code. Rather, an array of modifications likely enables a sophisticated regulatory system. The number of possible combinations of modifications coupled to the problem of quantitative analyses makes deciphering the roles of individual combinations challenging. Of all residues, lysine is the most extensively modified: acetylated, formylated, biotinylated, Nepsilon-monomethylated, dimethylated, trimethylated and ubiquitin-modified forms have been identified.

Figure 3: The combinatorial and dynamic nature of covalent modifications to the histone H3 N terminus, highlighting the role of 2OG oxygenases, and their role in introducing diversity into peptide-derived secondary metabolites.

Figure 3 : The combinatorial and dynamic nature of covalent modifications to the histone H3 N terminus, highlighting the role of 2OG oxygenases, and their role in introducing diversity into peptide-derived secondary metabolites.

(a) Identified modifications to part of the N terminus of histone H3 and their interplay19, 20. Me, methylation (blue); Ac, acetylation; For, formylation; Cit, citrulline residue; P, phosphorylation. Arrowheads and bars, positive (green) and negative (red) effects, respectively, of one modification on occurrence of another. (b) Human 2OG oxygenases reported to modify histone N termini and residue specificity3, 19, 20, 22, 24, 30. FBXL10/11, F-box and leucine-rich repeat proteins 10 and 11; JHDM2, jumonji histone demethylase 2 type; JMJD2/3/6, jumonji domain–containing protein 2, 3 and 6 type; JARID1, jumonji (A+T)-rich interactive domain 1 type; UTX, ubiquitously transcribed tetratricopeptide repeat gene X chromosome linked; sym and asym, symmetric and asymmetric. (c) Diversity-enhancing role of 2OG oxygenases in the daptomycin-like calcium-dependent antibiotic29 CDA3 and etamycin4 biosynthesis. AsnO, asparagine hydroxylase (see Fig. 2c); P4H, proline (4R)-hydroxylase from Streptomyces griseoviridus P8648.

Full size image (114 KB)

The first identified histone lysyl demethylase (LSD1) is an amine oxidase; though it is a mono- and dimethyllysyl demethylase, its mechanism precludes demethylation of trimethylated lysyl histone substrates. Tsukada et al. demonstrated that a 2OG oxygenase from the jumonji domain–containing (JmjC) family is capable of catalyzing the demethylation of tri-, di- and monomethylated lysyl histone peptides20, 21 (Fig. 1). This finding led a series of reports describing other 2OG oxygenases as histone lysyl demethylases (Fig. 3b). In humans, about 30 JmjC (usually multidomain) proteins have been identified20 and grouped into seven distinct subfamilies: JHDM1, JHDM2, JMJD2, PHF2, PHF8, jumonji (A+T)-rich interactive domain (JARID), ubiquitously transcribed tetratricopeptide repeat X/Y-linked (UTX/UTY) and JmjC-domain only. Assignment of physiological roles for the JmjC demethylases is presently lagging behind biochemical and cell biology studies; however, some are connected to endocrine regulation or cancer. Missense mutations to JARID1C in people with X-linked mental retardation causes reduced H3K4me3 demethylation activity, leading to proposed roles in neuronal survival and dendritic development22.

Chang et al. recently reported that JMJD6 (previously and misleadingly termed the phosphatidylserine receptor23) is a 2OG demethylase acting on histone methylarginyl residues24 (Fig. 1). This work further extends the scope of 2OG oxygenase chemistry in humans and is biologically important because it is the first description of a methylarginyl demethylase. Although peptidylarginyl deiminase 4 (PADI4) can act on methylated arginyl residues in histones, PADI4 is not a true demethylase as it catalyzes a hydrolysis reaction resulting in a citrulline residue product. Using antibody precipitation coupled to mass spectrometric analyses and cell-based assays, JMJD6 was shown to catalyze demethylation of specific histone dimethylated and monomethylated arginyl residues (H3R2 and H4R3, Fig. 3b). Further work is required to define the substrate selectivity of JMJD6—in terms of its preferred histone sequences and also to investigate whether JMJD6 catalyzes oxidations other than demethylation. Evidence from mice studies in which JMJD6 was deleted or reduced suggests it has an important developmental role, possibly involving its methylarginyl demethylation activity25.

Important objectives within the field of molecular epigenetics are to determine how substrate specificity and discrimination between different methylation states is achieved at the biochemical level and to correlate this data with that from cellular and higher levels. Structural biology is starting to provide some insights into how selectivity for the histone 2OG oxygenases is achieved—for example, with enzyme–substrate complexes26, 27 (Fig. 2b)—but much further work is required, including on how dynamic interactions are related, in order to work toward a molecular understanding of the role of oxygenases (and other enzymes) in epigenetic regulation.

More than hydroxylation?

The recent advances in understanding the biological roles of 2OG oxygenases, along with the many human enzymes in this family with no known function, suggest that much about the biology of human 2OG oxygenases awaits discovery. The plethora of complex post-translational modifications to histones has some analogy with the biosynthesis of secondary metabolites. Most secondary metabolites are oligomers in which structural diversity is introduced by the use of varied precursors, by modification of oligomers and by branched and overlapping pathways. 2OG oxygenases are involved in examples of all of these diversifying mechanisms. For example, in secondary metabolites formed through nonribosomal peptide synthetase (NRPS)-catalyzed oligomerization, 2OG oxygenases enable diversity by modifying amino acid precursors, most commonly by hydroxylation, and by modification of the peptide products of NRPS catalysis. Hydroxylation of alkyl side chains of >10 amino acids occurs in the biosynthesis of antibiotic (Figs. 2c and 3c), antitumor and antifungal natural products. Hydroxylation can also enable further transformations including macrolactone formation, glycosylation and additional oxidation (see, for example, ref. 28). 2OG oxygenases use the highly reactive Fe(IV) species5 to enable catalysis of more unusual oxidative modifications to peptides; for example in the ring-forming reactions in beta-lactam antibiotic biosynthesis4. So far there is no evidence that the exotic oxidative modifications catalyzed by 2OG oxygenases in secondary metabolism occur in the dynamic array of post-translational modification of histones. Complex oxidative covalent post-translational modifications involving lysyl (and hydroxylysyl) and other residues in metazoans do occur in the cross-linking of extracellular proteins such as collagen and elastin. The prevalence of 2OG oxygenase–catalyzed reactions in humans hints at the possibility that more complex transformations, possibly involving imine and carbonyl chemistry, may await discovery in the post-translational modifications of histones and other proteins.



Top

Acknowledgments

We thank the Biotechnology and Biological Sciences Research Council, the Wellcome Trust and the Rhodes Trust (C.L.) for funding.

Competing interests statement:

The authors declare  competing financial interests.

Top

References

  1. Hutton, J.J. Jr., Kaplan, A. & Udenfriend, S. Arch. Biochem. Biophys. 121, 384–391 (1967). | Article | PubMed | ChemPort |
  2. Costas, M., Mehn, M.P., Jensen, M.P. & Que, L. Chem. Rev. 104, 939–986 (2004). | Article | PubMed | ISI | ChemPort |
  3. Ozer, A. & Bruick, R.K. Nat. Chem. Biol. 3, 144–153 (2007). | Article | PubMed | ChemPort |
  4. Clifton, I.J. et al. J. Inorg. Biochem. 100, 644–669 (2006). | Article | PubMed | ISI | ChemPort |
  5. Flashman, E. & Schofield, C.J. Nat. Chem. Biol. 3, 86–87 (2007). | Article | PubMed | ChemPort |
  6. Schofield, C.J. & McDonough, M.A. Biochem. Soc. Trans. 35, 870–875 (2007). | Article | PubMed | ChemPort |
  7. Vaz, F.M. & Wanders, R.J. Biochem. J. 361, 417–429 (2002). | Article | PubMed | ISI | ChemPort |
  8. Coleman, M.L. & Ratcliffe, P.J. Essays Biochem. 43, 1–15 (2007). | Article | PubMed | ChemPort |
  9. Jenkins, C.L. & Raines, R.T. Nat. Prod. Rep. 19, 49–59 (2002). | Article | PubMed | ChemPort |
  10. Myllyharju, J. & Kivirikko, K.I. Trends Genet. 20, 33–43 (2004). | Article | PubMed | ISI | ChemPort |
  11. Dinchuk, J.E. et al. J. Biol. Chem. 277, 12970–12977 (2002). | Article | PubMed | ISI | ChemPort |
  12. Cockman, M.E. et al. Proc. Natl. Acad. Sci. USA 103, 14767–14772 (2006). | Article | PubMed | ChemPort |
  13. Coleman, M.L. et al. J. Biol. Chem. 282, 24027–24038 (2007). | Article | PubMed | ChemPort |
  14. Ferguson, J.E. III et al. Mol. Cell. Biol. 27, 6407–6419 (2007). | Article | PubMed | ChemPort |
  15. Sedgwick, B., Bates, P.A., Paik, J., Jacobs, S.C. & Lindahl, T. DNA Repair (Amst.) 6, 429–442 (2007). | Article | PubMed | ChemPort |
  16. Ringvoll, J. et al. EMBO J. 25, 2189–2198 (2006). | Article | PubMed | ChemPort |
  17. Frayling, T.M. et al. Science 316, 889–894 (2007). | Article | PubMed | ISI | ChemPort |
  18. Gerken, T. et al. Science 318, 1469–1472 (2007). | Article | PubMed | ChemPort |
  19. Kouzarides, T. Cell 128, 693–705 (2007). | Article | PubMed | ISI | ChemPort |
  20. Klose, R.J. & Zhang, Y. Nat. Rev. Mol. Cell Biol. 8, 307–318 (2007). | Article | PubMed | ISI | ChemPort |
  21. Tsukada, Y.-I. et al. Nature 439, 811–816 (2006). | Article | PubMed | ISI | ChemPort |
  22. Shi, Y. Nat. Rev. Genet. 8, 829–833 (2007). | Article | PubMed | ChemPort |
  23. Cikala, M. et al. BMC Cell Biol. 5, 26 (2004). | Article | PubMed |
  24. Chang, B., Chen, Y., Zhao, Y. & Bruick, R.K. Science 318, 444–447 (2007). | Article | PubMed | ChemPort |
  25. Bose, J. et al. J. Biol. 3, 15 (2004). | Article | PubMed |
  26. Ng, S.S. et al. Nature 448, 87–91 (2007). | Article | PubMed | ISI | ChemPort |
  27. Chen, Z. et al. Proc. Natl. Acad. Sci. USA 104, 10818–10823 (2007). | Article | PubMed | ChemPort |
  28. Chen, H. et al. Biochemistry 40, 11651–11659 (2001). | Article | PubMed | ChemPort |
  29. Strieker, M., Kopp, F., Mahlert, C., Essen, L.-O. & Marahiel, M.A. ACS Chem. Biol. 2, 187–196 (2007). | Article | PubMed | ChemPort |
  30. Agger, K. et al. Nature 449, 731–734 (2007). | Article | PubMed | ChemPort |

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated.

NEWS AND VIEWS

The most versatile of all reactive intermediates?

Nature Chemical Biology News and Views (01 Feb 2007)