HAGH

HAGH catalyzes a key step in the detoxification of reactive and toxic 2-oxoaldehydes, including methylglyoxal ^[12]. 2-Oxoaldehydes react with DNA and proteins, resulting in aberrant postoligomerization modifications, which are associated with aging by causing DNA instability and alteration of protein function ^[13].

The well-characterized glyoxalase system converts the toxic 2-oxoaldehydes to less-reactive metabolites ^[13] and comprises two key enzymes: Glyoxalase I and HAGH ^[13]. Glyoxalase I, which is not an MBLf protein, catalyzes the first step; that is isomerization of a nonenzymatically formed hemithioacetal adduct of the ubiquitous reducing agent glutathione and the toxic 2-oxoaldehyde, to give a (S)-2-hydroxyacylglutathione thioester ^[14]. HAGH then catalyzes hydrolysis of the thioester to give a 2-hydroxycarboxylate product (e.g., lactate from methylglyoxal) and regenerate glutathione (Figure 2A) ^[13].

Open in a separate window

Figure 2

Known Reactions Catalyzed by Human Metallo β-Lactamase Fold (hMBLf) Enzymes (and Cytidine Monophospho-N-Acetylneuraminic Acid Hydroxylase). (A) Group 1, glyoxalase II, and human ethylmalonic encephalopathy protein 1 (hETHE1); (B) Group 2: DNA crosslink repair 1 (DCR1 or SNM1A, B, and C) enzymes; (C) Group 3: N-acyl-phosphatidylethanolamine-hydrolyzing phospholipase D (NAPE-PLD) and an enzyme present in most if not all mammals except humans, cytidine monophospho-N-acetylneuraminic acid hydroxylase (CMAH). Substrate atoms directly involved in enzyme catalysis are in red. Note that not all hMBLs have assigned functions.

Following the determination of a crystal structure for HAGH [Protein Data Bank (PDB) ID:1XM8] from Arabidopsis thaliana, which revealed a di-zinc binding site ^[15], human HAGH structures have been reported, including in complex with glutathione and the substrate analog S-(N-hydroxy-N-bromophenylcarbamoyl)glutathione (PDB ID: 1QH3/5); this is one of the few MBL structures providing detailed insights into substrate binding ^[16] (Figure 3A). As with other MBLf superfamily members, HAGH can use metal ions other than zinc, including iron and manganese ions ^[17]. Of the characterized hMBLf proteins, the HAGH reaction is perhaps the most closely related to that of classical MBLs. The proposed HAGH mechanism involves nucleophilic attack of water onto the substrate carbonyl to give a negatively charged tetrahedral intermediate that is stabilized by a zinc ion, followed by C–S bond cleavage (Figure 4A). In this regard, it is notable that, in solution, the proposed rate-limiting step in HAGH catalysis and, at least in some cases of β-lactam hydrolysis, is breakdown of the tetrahedral intermediate, whereas in acyclic amide hydrolysis, it is (normally) formation of the tetrahedral intermediate. Interestingly, some β-lactam antibiotics hydrolyzing MBLs can catalyze thioester hydrolysis in a mechanism related to that of HAGH ^[9], and thiols are an established class of MBL inhibitors ^{18, 19}.

Open in a separate window

Figure 3

Overall Folds and Active Sites of Human Metallo β-Lactamase Fold (hMBLf) Enzymes. (A) Glyoxalase II family hMBLs: glyoxalase II [Protein Data Bank (PDB) ID: 1QH5], LACTB2 (PDB ID: 4AD9), ETHE1 (PDB ID: 4CHL). (B) DNA/RNA-interacting hMBLs: SNM1A (PDB ID: 4B87), SNM1B (PDB ID: 5AHO), CPSF73 (PDB ID: 2I7T) and ELAC1 (PDB ID: 3ZWF). (C) Group 3 hMBLs: NAPE-PLD (PDB ID: 4QN9). The crystallographic studies reveal how the conserved MBLf supports different types of reaction. Given the varied reaction types catalyzed by MBLf enzymes, the metal-coordination chemistry for the different enzymes appears to be similar; however, differences are such that the development of selective inhibitors should be possible. Note the conservation of some residues in binding the ‘second’ zinc in enzymes that only bind one metal (e.g., ETHE1) and that work on the accurate definition of metal use by nonheme metallo-proteins is limited by availability of methods for quantifying metal binding by proteins at endogenous levels. It should also be noted that the use of different metals can alter the catalytic profile of the same MBLf enzyme ^[113]. The MBL domain is in blue (β-sheets) and cyan (α-helices); β-CPSF-Artemis-SNM1-Pso2 (β-CASP) domains are in green. N- and C-terminal residues are in red. Zinc and iron ions are in pink and orange, respectively. Sulfate and phosphate ions are in green. Substrates and inhibitors, if present, are in green. The standard BBL MBL numbering system is used ^[103], with the active site motif number in parentheses (see main text). Abbreviations: 3PE, 1,2-diacyl-sn-glycero-3-phosphoethanolamine [note that, in (C), part of the 3PE molecule is omitted to show active site interactions]; HBPC-GSH, S-(N-hydroxy-N-bromophenylcarbamoyl) glutathione; TA, L(+)-tartaric acid. Note that there are few high-quality full-length (i.e., including noncatalytic domains) enzyme–substrate and/or product structures reported for nonclassical MBLs; product complex structures are reported for classical MBLs ^[114].

Open in a separate window

Figure 4

Outline Mechanism for Reactions Catalyzed by Human Metallo β-Lactamase Fold (hMBLf) Enzymes. (A) Outline of the Glyoxalase II mechanism, showing the proposed tetrahedral intermediate ^[16]. (B) Proposed ethylmalonic encephalopathy protein 1 (ETHE1) mechanism; initial iron complexation of GSSH thiol and GSSH cysteinyl-glycine amide nitrogen to the iron (II) is followed by binding of O₂, then oxidation of the ligated disulfide ^{10, 30}. (C) Possible outline mechanism for the SNM1-type nucleases.

ETHE1

ETHE1 has a key role in the mitochondrial metabolism of the toxic (and smelly) small molecule H₂S. In mitochondria, ETHE1 catalyzes the oxygen-dependent conversion of the persulfide (GSSH) formed from glutathione and H₂S to give glutathione and persulfite ^[10] (Figure 2A). Mutations in the ETHE1 gene correlate with the autosomal recessive metabolic disease ethylmalonic encephalopathy (EE), which leads to death in infancy ^{20, 21, 22}. To date, 16 clinically observed mutations have been identified in ETHE1, at least some of which are likely inactivating. Substitutions at the conserved Arg163, located on a loop in close proximity to the active site, are proposed to cause reduced protein stability and alteration of the redox potential of the iron center ^[23]. However, some missense mutations occurring at highly conserved residues lead to normal or reduced levels of active ETHE1 ^{10, 20, 24, 25}.

ETHE1 is overexpressed in hepatocellular carcinoma cells and is proposed to abrogate apoptosis mediated by the transcription factor p53, thereby enhancing cancer progression ^{26, 27}. Two possible mechanisms have been proposed for this process. ETHE1 is proposed to act as a nuclear-cytoplasmic shuttling protein that binds the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) protein RelA/p65 in the nucleus and promotes its transport to the cytoplasm. Given that the nuclear localization of NF-κB is essential in p53-mediated apoptosis, the ETHE1-dependent sequestration of RelA/p65 in the cytoplasm results in failure of p53-mediated cell death ^[26]. Alternatively, it is proposed that ETHE1 may reduce p53-regulated transcription of proapoptotic genes by interacting with histone deacetylase 1 (HDAC1) and thereby promotes p53 deacetylation ^[27]. Given the roles of ETHE1 in thiol metabolism and the role of metal ions in both HDAC and hMBLf catalysis, it is notable that HDAC inhibitors are often metal chelators, including appropriately functionalized thiols and hydroxamic acids. These classes of HDAC inhibitor are known classes of ‘classical’ MBL inhibitor ^[28]. It is possible that the biological activities of such inhibitors, including effects on cell proliferation, arise from inhibition of hMBLf proteins (including ETHE1) in addition to HDACs.

From a mechanistic perspective, ETHE1 is remarkable. Biochemical and structural studies reveal that it binds a single Fe (III) ion, instead of Zn (II), into its active site ^{10, 29}. The human ETHE1 structure (PDB ID: 4CHL) ^[30] is closely related to that of ETHE1 from A. thaliana and reveals a dimer with a single iron ion at each active site ^[30] (Figure 3A). Interestingly, in both the A. thaliana and human ETHE1 structures, an oxidized cysteine residue is present close to the active site, although the role of this modification in unknown ^{29, 30}. In a manner closely reminiscent of the nonheme iron oxygenases of the 2-oxoglutarate oxygenase and other superfamilies, ETHE1 uses a 2-His-carboxylate site motif to complex its nonheme iron. The proposed ETHE1 mechanism is related, at least in some aspects, to that of the key enzyme in penicillin biosynthesis [i.e., isopenicillin N synthase (IPNS) ^[31]], and involves initial complexation of the GSSH thiol and cysteinyl-glycine amide nitrogen to the iron followed by binding of O₂ and oxidation of ligated sulfur or disulfide ^{10, 30} (Figure 4B). Further work is required to characterize the proposed intermediates in ETHE1 catalysis.

PNKD

PNKD exists in four different isoforms (due to alternate splicing), of which isoforms 1, 3, and 4 contain an MBLf domain, whereas isoform 2 does not ^[32]. The biochemical role and mechanism of PNKDs are unknown. The PNKD MBLf domain has high sequence similarity with HAGH, and PNKD is proposed as a Glyoxalase II-type enzyme ^[33]. Mutations of the PNKD gene correlate with paroxysmal nonkinesigenic dyskinesia or Dystonia type 8 (DYT8), which is characterized by dystonia, chorea, and athetosis (i.e., involuntary muscle contractions) ^{33, 34, 35}. The observation of dysfunction in dopamine signaling (i.e., increased dopamine turnover) in the basal ganglia of PNKD-1-knockout mice suggests that the dopamine pathway is involved in paroxysmal nonkinesigenic dyskinesia ^{33, 36}. In support of this, mutant PNKD-expressing mice display alterations in exocytosis, suggesting that PNKD is involved in modulation of neurotransmitter release at the nigrostriatal dopaminergic terminals ^[37]. PNKD is present in muscle tissues including skeletal muscle and myocardial myofibrils, where it interacts with sarcomeric proteins, indicating a role for it as a myofibrillogenesis regulator ^{38, 39, 40}. PNKD2 stimulates fiber formation and, indirectly, activates the focal adhesion kinase/(Ser/Thr) kinase (FAK/Akt) signaling pathway to promote cell migration and proliferation ^[41]. PNKD2 is overexpressed in some cancer cells (e.g., ovarian) and is proposed to be involved in tumor cell proliferation, migration, and adhesion ^[42]. Although the biochemical roles of PNKD have not yet been identified, collectively these studies suggest that PNKD is involved in multiple cellular pathways, and its malfunction has pleiotropic consequences.

The other Glyoxalase II hMBLf subfamily proteins are of unassigned function. Sequence analyses suggest that MBLAC1/2, HAGHL. and LACTB2 bind two metal ions [as supported by crystallographic analyses for LACTB2 (PDB ID: 4AD9)] and may be hydrolases (Figure 3A).

DNA Crosslink Repair 1A and 1B Enzymes (DCR1A and 1B or SNM1A and 1B)

DNA crosslink repair 1A (SNM1A or DCR1A) and DNA crosslink repair 1B (SNM1B, DCR1B, or Apollo) are - exonucleases that are activated on DNA damage. SNM1A/B are important in interstrand crosslink (ICL) repair; in particular, SNM1A has a striking ability to resect past site-specific crosslinks in vitro ^[48]. Cells depleted in SNM1A/B show increases in sensitivity to ICL-inducing agents, such as cisplatin and mitomycin C ^[49]. Mutations of the SNM1A and SNM1B MBLf domains cause loss of nuclease activity, revealing the crucial role of the MBLf domain in repair ^{49, 50}. Cells depleted of SNM1A accumulate double-strand (ds) breaks (DSB) as a consequence of replication fork breakage. These observations indicate a role for SNM1A in the predominant, replication-coupled, ICL repair pathway. In support of this, in mammalian cells, SNM1A promotes G₁ cell cycle checkpoint arrest after exposure to ionizing radiation ^{51, 52}.

SNM1B has an established role in telomere maintenance via interaction with the telomere-associated factor (TRF2) ^[50]. SNM1B is involved in the generation of 3′-overhang structures at telomeres, protecting them from deleterious repair by nonhomologous end-joining (NHEJ) ^[53]. Similarly to SNM1A, depletion of SNM1B sensitizes cells to ICLs as well as to other genotoxins, such as ionizing radiation and cisplatin ^{54, 55, 56}.

DNA Crosslink Repair 1C (DCR1C, SNM1C, or Artemis)

SNM1C has a key role in one of the major pathways for repair of DSB in mammalian cells: the NHEJ pathway ^[47]. SNM1C is also involved in antibody and T cell diversification as a consequence of its role in V(D)J recombination, a specialized somatic recombination process that is essential for adaptive immunity ^[57]. SNM1C is a 5′-3′ exonuclease acting on both single-strand (ss) and ds DNA and, uniquely within the DNAase MBLf family, is also an endonuclease that acts on DNA hairpin structures generated during V(D)J recombination and on DNA overhangs during NHEJ repair of ds breaks. The endonuclease activity of SNM1C is promoted by its phosphorylation and complexation with DNA-dependent protein kinases ^{58, 59}. Mutations of the SNM1C MBLf domain correlate with radiosensitive severe combined immunodeficiency syndrome (RS-SCID), likely due to impaired V(D)J recombination ^{57, 59} and defects in repair of ionizing-radiation-induced ds breaks. The C-terminal region of SNM1C directly interacts with the binding domain of DNA Ligase IV, a unique mammalian ligase involved in the final step of ds break repair by NHEJ ^[60]. Inhibition of the SNM1C/Ligase IV complex formation is of medicinal interest, because decreasing the efficiency of NHEJ repair in tumors might synergize with chemotherapeutic agents and radiotherapy ^[60]. SNM1C may also have a role in cell cycle regulation via direct interaction with the Cul4A-DDB1 E3 ubiquitin ligase complex, which is required for the degradation of the cyclin-dependent kinase protein inhibitor p27 during the transition from G₁ to S phase ^[61].

ELAC 1 and 2 Proteins (ELAC1/2)

ELAC1/2 are nucleases that mediate 3′-tRNA processing; both belong to the RNAse Z family and contain two active site zinc ions ^[62]. Human ELAC1 is significantly smaller than ELAC2 (363 versus 826 residues), and is similar to the C-terminal MBL domain-containing region of ELAC2; thus, ELAC1 is likely to have evolved as a consequence of gene duplication ^[63]. ELAC2 displays activity as a 3′-end tRNA-processing nuclease involved in the maturation of mitochondrial (mt) tRNA ^{64, 65}. Mutations in the ELAC2 gene correlate with accumulation of unprocessed (mt) tRNAs, leading to impaired mitochondrial translation, which is associated with hypertrophic cardiomyopathy, an often lethal inborn pathology ^[66]. Similar to ELAC2, ELAC1 has 3′-tRNA-processing activity and appears to have endoribonuclease activity on unstructured RNA ^[67]. Although they have similar biochemical functions, the ELAC proteins have different subcellular localizations. ELAC1 localizes in the cytoplasm, whereas ELAC2 localizes in both the nucleus and mitochondria ^[68]. The cytoplasmic localization of ELAC1 suggests that it is responsible for degradation of unstructured RNA ^[68]. By contrast, ELAC2 reportedly interacts with the -tubulin complex in the centrosome during cell division, suggesting a possible role as a cell cycle regulator. Such a role could explain the correlation between mutations in ELAC2 and tumor onset, and other links between polymorphism of ELAC2 and prostate cancer development ^{69, 70, 71, 72}.

The Cleavage and Polyadenylation Specificity Factor hMBLf Proteins

Structural Studies on hMBLf Nucleases

At present, crystal structures of four of the nuclease subfamily of hMBLf proteins have been reported, for SNM1A, SNM1B, CPSF73, and ELAC1; all of which bind one or two zinc ions in their active site (Figure 3B). Crystallographic analyses show a single zinc ion in the SNM1A active site (PDB ID: 4B87), but two zinc ions in the SNM1B active site (PDB ID: 5AHO), as observed for ELAC1 (PDB: 3ZWF) and CPSF73 (PDB ID: 2I7T) ^[74]. More detailed biochemical analyses are required to define the biologically relevant metal usages. The CPSF73 and SNM1A/B structures display high overall similarity, in part because they have both MBL and β-CASP domains (Figure 3B); in each case, both domains are required for catalysis, consistent with the similarity in the reactions catalyzed by each (Figure 2B). By contrast, in ELAC1, the αββα core MBLf structure represents the entire ELAC1 protein, consistent with the different reactions catalyzed by ELAC1 ^[67]. Despite the emerging structural information, the mechanistic details of the hMBLf nucleases remain to be determined. By analogy with the MBLs, an outline mechanism of action for the SNM1 enzymes is shown in Figure 4C: a zinc ion-activated water mediates nucleophilic attack on the phosphate group, which is also activated and/or bound to a zinc ion.

Cytidine Monophospho-N-Acetylneuraminic Acid Hydroxylase

Sialic acids occur at the termini of sugar chains on the surface-exposed portion of plasma membrane proteins and have important roles in ligand–receptor, cell–cell, and cell–pathogen interactions. The most common forms of sialic acid in animals are N-acetylneuraminic acid (Neu5Ac) and its hydroxylated form, N-glycolylneuraminic acid (Neu5Gc) (Figure 2C). In all mammals except humans, cytidine monophospho-N-acetylneuraminic acid hydroxylase (CMAH) catalyzes the addition of an oxygen atom onto the methyl group of the acetyl group of CMP-Neu5Ac to give Neu5Gc ^[89]. In humans, CMAH is inactivated by deletion of exon 6, which would encode the hydroxylase active site. The relevant mutation occurred more than 0.5 million years ago, before the Neanderthal divergence ^{90, 91}. In humans, Neu5Gc is detected at low levels, likely derived from food intake ^[89]. The loss of catalytically active CMAH in humans is proposed to be associated with the acquisition of either resistance or susceptibility to pathogens that preferentially recognize Neu5Gc or Neu5Ac ^{92, 93}.

There is emerging evidence that the catalytically inactive human ‘CMAH-like’ protein has biological roles. It is reported to be overproduced on the surface of adult stem cells and to be involved in activation of Wnt signaling ^[91]. Overexpression of adult stem cell-specific CMAH leads to an increased cellular uptake of exogenous Neu5Gc, suggesting a role for CMAH in Neu5Gc transport ^[94]. CMAH is also proposed to be involved in glucose metabolism. CMAH^–/– knockout mice manifest an obese phenotype with impaired glucose tolerance and pancreatic β cell dysfunction ^[95]. The absence of functional CMAH in humans is proposed to be involved in our susceptibility to progressive β cell degeneration and, thus, insulin resistance and type 2 diabetes mellitus ^[95].

From a biochemical perspective, the mechanism by which (nonhuman) CMAH catalyzes hydroxylation of an unactivated acetyl methyl group is interesting but as yet no structure of CMAH has been reported. Similar to ETHE1, CMAH is likely an oxygenase. However, unlike ETHE1, which apparently only requires iron as cofactor, native murine CMAH activity is stimulated by NADH, cytochrome b5, and NADH-cytochrome b5 reductase ^{96, 97}.

We thank all our colleagues and collaborators who have worked on metallo β lactamase fold enzymes. We apologize for incomplete citation due to space limitations. The authors received grants from: Medical Research Council (MRC)/Canadian grant G1100135 (I.P.), Cancer Research UK, (C.J.S.), MRC MR/L007665/1 (J.B., S.Y.L, P.J.M, and C.J.S.) and the Biotechnology and Biological Sciences Research Council (C.J.S).