• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of embojLink to Publisher's site
EMBO J. Jul 8, 2009; 28(13): 1953–1964.
Published online Jun 4, 2009. doi:  10.1038/emboj.2009.146
PMCID: PMC2711183

How the MccB bacterial ancestor of ubiquitin E1 initiates biosynthesis of the microcin C7 antibiotic


The 39-kDa Escherichia coli enzyme MccB catalyses a remarkable posttranslational modification of the MccA heptapeptide during the biosynthesis of microcin C7 (MccC7), a ‘Trojan horse' antibiotic. The approximately 260-residue C-terminal region of MccB is homologous to ubiquitin-like protein (UBL) activating enzyme (E1) adenylation domains. Accordingly, MccB-catalysed C-terminal MccA-acyl-adenylation is reminiscent of the E1-catalysed activation reaction. However, unlike E1 substrates, which are UBLs with a C-terminal di-glycine sequence, MccB's substrate, MccA, is a short peptide with an essential C-terminal Asn. Furthermore, after an intramolecular rearrangement of MccA-acyl-adenylate, MccB catalyses a second, unique reaction, producing a stable phosphoramidate-linked analogue of acyl-adenylated aspartic acid. We report six-crystal structures of MccB in apo, substrate-, intermediate-, and inhibitor-bound forms. Structural and kinetic analyses reveal a novel-peptide clamping mechanism for MccB binding to heptapeptide substrates and a dynamic-active site for catalysing dual adenosine triphosphate-consuming reactions. The results provide insight into how a distinctive member of the E1 superfamily carries out two-step activation for generating the peptidyl-antibiotic MccC7.

Keywords: antibiotic, MccB, MccC7, microcin, ubiquitin activating enzyme


To survive in a competitive milieu, bacteria generate a variety of antimicrobial agents that are toxic to rival species. Many clinically important antibiotics, such as penicillin, actinomycin, and vancomycin, contain peptide scaffolds that are generated by bacterial non-ribosomal peptide synthetases (Mootz et al, 2002). Other antibiotics contain ribosomally synthesized peptides, including the low-molecular weight microcins (Nolan and Walsh, 2008). Microcins are posttranslationally modified peptides that inhibit growth of competing Gram-negative bacteria, such as Escherichia, Salmonella, and Enterobacter, at nanomolar concentrations (Duquesne et al, 2007). Microcin C7 (MccC7, Figure 1A), a heptapeptide with an N-formylated methionine, an aminopropyl moiety, and a C-terminal phosphoramidate linkage to adenosine (Guijarro et al, 1995), is produced by Escherichia coli to eradicate competitor strains through a ‘Trojan horse' mechanism. After import into target cells, the peptide framework of MccC7 is cleaved by non-specific peptidases to release a toxic adenylated-aspartic-acid mimic that targets aspartyl-tRNA synthetase, and thereby inhibits protein synthesis (Metlitskaya et al, 2006).

Figure 1
The MccB reaction. (A) Structure of MccC7. (B) Mechanism for the MccB reaction. Only the C-terminus of the MccA heptapeptide is shown for clarity. (C) List of MccB structures presented. (D) Chemical structures of the peptides used in this study.

Understanding the biosynthetic pathways of natural products, such as MccC7, allows us to gain insights into enzyme mechanism, provides opportunities for developing new antibiotics, and gives new prospects for protein design aimed at generating novel biologically active compounds. MccC7 is generated by two posttranslational modifications of a precursor ribosomal heptapeptide, MccA, featuring the sequence MRTGNAN. Posttranslational steps in MccC7 biosynthesis involve the conversion of the C-terminal Asn7 to an Asp amide, in which the nitrogen is linked by a phosphoramidate bond to adenosine monophosphate (AMP), followed by aminopropylation of one of the phosphate oxygens. The migration of the Asn7 carboxamido nitrogen and the N–P bond-forming step are catalysed in a Mg2+-dependent manner by the enzyme MccB (Figure 1B) (Roush et al, 2008). First, in a conventional acyl-adenylation reaction that consumes one adenosine triphosphate (ATP) molecule, MccB catalyses adenylation of the C-terminus of MccA. The resulting MccA-acyl-adenylate then undergoes intramolecular rearrangement, during which AMP is expelled by the carboxamido nitrogen of Asn7 as a peptidyl-succinimide intermediate is formed. Next, MccB catalyses an unusual adenylation of the succinimide: the succinimidyl nitrogen attacks the α-phosphate of a second ATP molecule. This links the AMP moiety to the peptide terminus through an N–P bond. The succinimide ring is hydrolysed by regiospecific water-mediated opening to yield the peptidyl-acyl-N-P-adenosine group, which is the Trojan horse reagent. Unlike acyl-adenylate complexes, which are hydrolytically labile-mixed acyl-phosphoric anhydrides, the phosphoramidate-linked analogue of aspartic acyl-adenylate (Figure 1B) is hydrolytically stable, allowing persistence throughout the remainder of the MccC7 biosynthetic pathway (aminopropylation, export by the producing cell and import by a sensitive bacterial cell) and ultimate toxic inhibition of aspartyl-tRNA synthetase and protein synthesis in competing bacteria.

The 351-residue MccB has two discernible regions of sequence. The N-terminal (~90 residues) region is not detectably homologous to known structures. The C-terminal (~260 residues) region shares homology with the adenylation domain portion of ubiquitin-like (UBL) protein-activating enzymes, also called E1s, which initiate UBL conjugation (Supplementary Figures 1 and 2). MccB-catalysed adenylation of the MccA C-terminus is reminiscent of the first E1-catalysed reaction, which is the C-terminal adenylation of UBL substrates. However, the substrates of MccB and E1s are markedly divergent. Although MccB catalyses C-terminal transformation of a peptide destined for secretion to combat other organisms, substrates of E1s are proteins that serve as sulphur carriers during biosynthetic pathways, or as covalent protein modifiers (Hochstrasser, 2000, 2009).

The sequence similarity between MccB and UBL-activating enzymes raises several questions about common and divergent aspects of the MccB mechanism. First, how does MccB recognize a short-peptide substrate, when all other family members have a [gt-or-equal, slanted]8 kDa UBL protein substrate? Second, why does MccB catalyse adenylation of a peptide harbouring a C-terminal Asn or succinimide, rather than the Gly-Gly sequence found at the C-termini of UBLs? Third, how can MccB catalyse two successive adenylation reactions? To address these questions, we performed structural, biophysical, and biochemical analyses of MccB–MccA–nucleotide interactions.

Results and discussion

Overall structure of MccB

We determined crystal structures of MccB separately and in complex with the following ligands: (1) MccA (first-peptide substrate), (2) MgATP (nucleotide substrate and divalent metal required for catalysis), (3) MccA and α,β-methylene ATP (AMPCPP) (first-peptide substrate and non-hydrolysable ATP analogue), (4) MccA–succinimide (second-peptide substrate), and (5) MccA–N7isoAsn (inhibitor) (Roush et al, 2008) (Figure 1C and D; Table I; Supplementary Figure 3). All asymmetric units contain two nearly identical MccB homodimers, with chains designated as A–B or C–D. Thus, the six structures presented here contain a total of 12 homodimers and 24 protomers. The 12 A–B and C–D homodimers superimpose with a Cα root mean square deviation (RMSD) of 76 Å. Owing to this structural similarity, our discussion focuses on chains A–B. The MccB homodimer has an elongated shape. Each homodimer has two parallel, two-domain ‘active units'. Each ‘active unit' contains an adenylation domain and a globular domain not found in other E1 enzymes. These domains are generated from parts of both protomers in the homodimer (Figure 2).

Figure 2
Overall structure of MccB. The MccB–succinimide complex structure is shown, with MgATP modelled into the adenlyation-active site after superposition of MccB–succinimide and MccB–MgATP structures. The two protomers in the MccB homodimer ...
Table 1
Data collection and refinement statistics

For the adenylation domain, the two parallel halves comprise residues (1) A104-A260, A287-A351, B90-B103 and (2) B104-B260, B287-B351, A90-A103. Each half includes a Rossmann-fold ATP-binding cleft. A five-stranded β-sheet, stabilized by an adjacent zinc-binding site, forms a concave peptide-binding surface. Both halves, individually and together, superimpose with the homologous regions of the bacterial UBL-activating enzymes MoeB and ThiF with Cα RMSDs of 1.5 Å over half and 1.7 Å over the dimer (Lake et al, 2001; Duda et al, 2005), and with the eukaryotic E1s UBA1, NAE1-UBA3, and SAE1-UBA2 with Cα RMSDs of 1.8–1.9 Å over half and 2.1 Å over the dimer (Walden et al, 2003b; Lois and Lima, 2005; Lee and Schindelin, 2008).

The unique MccB domain is formed by the coupling of two separated regions of the MccB sequence: the N-terminal residues 1–87 from one monomer and residues 261–287 from the opposite protomer within the homodimeric assembly (Supplementary Figure 1). This latter region corresponds to an E1 enzyme ‘crossover loop', which crosses over the C-terminal tail from the associated UBL (Lake et al, 2001; Walden et al, 2003a; Lois and Lima, 2005; Lee and Schindelin, 2008). In MccB, the ‘crossover loop' is incorporated into the MccB-specific domain, which adopts a structure loosely related to a winged-helix fold. In our peptide co-crystal structures, numerous contacts are observed between this domain and the associated peptides. Thus, we refer to this domain as the ‘peptide clamp', owing to its interactions with MccA.

The adenylation and peptide clamp domains pack together in a manner resembling a clamshell. In the overall assembly of an ‘active unit', each domain would represent half the shell, which together generate a large central groove that cradles the ATP and peptide substrates (Figure 2).

Anchoring MccA through the unique peptide clamp

Detailed insights into heptapeptide recognition come from superimposing the MccB–MccA and MccB–succinimide structures. MccA extends over MccB's adenylation domain β-sheet with its C-terminus approaching the ATP-binding site for adenylation (Figure 3A). The peptides bind in a groove formed by the adenylation domain β-sheet cleft on one side, and the peptide clamp on the other, burying an average of 940 A2 from MccA (85% of the total surface area) and ~ 600 A2 from MccB.

Figure 3
The unique MccB peptide clamp domain required for activity. (A) Close-up view of the MccB-binding site for peptide substrates. Structural superposition of MccB–MccA, (only MccA shown), MccB–succinimide, and MccB–MgATP (only MgATP ...

To test the functional importance of the peptide clamp domain, we performed enzymatic assays on mutant versions of MccB. Deleting the MccB N-terminal extension (ΔNTE), which comprises the bulk of the peptide clamp domain, completely eliminates the activity (Figure 3B; Supplementary Figure 4). To test whether a ΔNTE mutant has an intact adenylation domain assembly, we assayed homodimerization by analytical ultracentrifugation (AUC). As the AUC data indicate dimerization, reflecting proper folding (Supplementary Figure 5), we tested whether adding a 10-kDa protein fragment corresponding to MccB's N-terminal 87 residues (NTE) could restore any activity to the ΔNTE mutant in trans. Approximately 20% of the wild-type (WT) activity was restored (Figure 3B; Table II). These results support the importance of the peptide clamp domain, and indicate that the adenylation and peptide clamp domains work together optimally when expressed as a single unit.

Table 2
MccB mutants kinetics

Certain MccB residues from both the clamp and adenylation domains contact the peptides in all the co-crystal structures (Figure 4). Two such residues from the clamp domain are K10 and E26. MccB's K10 directly contacts the carbonyl oxygen from G4 and the T3 side chain from MccA and the succinimide. This latter interaction may have a function in constraining the sequences allowed at MccA position 3 during MccC7 biosynthesis (Kazakov et al, 2007). The MccB–K10A mutant has little effect on catalytic rate, but has an increase in the Km of MccA (Table II), which further shows its importance for peptide binding.

Figure 4
MccB binding to MccA. Stereo panels showing contacts between MccB and the MccA peptide ligands. The two protomers in the MccB homodimer are shown in cartoon representation in pink and magenta. MccB side chains and peptides are depicted as sticks, and ...

Residues in the adenylation domain form a pocket, in which the N-terminus of the peptides binds. The peptide N-terminal Met1 adopts a similar conformation in all the complexes, and is encircled by a deep hydrophobic channel generated by residues I243, V245, W326, and H333. The carbonyl of the peptide Met1 participates in a hydrogen bonding network with R322 and Q335 (Figure 4). In the absence of peptide (as in the ATP-bound structure), R322 and Q335 are bound to the solvent.

Past Met1, the next four peptide residues follow related trajectories towards the adenylation-active site. Although none of these structurally represents a catalytic intermediate, the C-terminus of the succinimide peptide comes close to approaching the adenylation-active site. In the crystals, the C-terminus of the peptide substrate MccA occupies several locations, none of which is as close to the adenylation-active site as the succinimide. A different backbone conformation is observed for the MccA–N7isoAsn inhibitor, which is a subtle variant of the first-peptide substrate. Residues Thr3-Asn5 of the inhibitor are ‘looped out' by one residue relative to the corresponding substrate structures, such that the Ala6-isoAsn7 backbone aligns with Asn5-Ala6 from the succinimide peptide. Thus, the C-terminus of MccA–N7isoAsn does not come as close to the adenylation-active site as the succinimide peptide. The conformational variability observed at the C-termini of the peptides will be discussed below.

Comparison of the peptide-MccB structures with UBLs complexed with their activating enzymes (Lake et al, 2001; Walden et al, 2003a; Lois and Lima, 2005; Lehmann et al, 2006; Huang et al, 2007; Lee and Schindelin, 2008) reveals that the MccA-derived peptides resemble the last seven residues of E1-bound UBLs (Supplementary Figure 6). Thus, it seems that the unique MccB peptide clamping mechanisms direct substrates to an E1-like active site.

Adenylation-active site

ATP binds in a cleft centred around the MccB Gly-Cys-Gly-Gly-Ile-Gly nucleotide-binding motif, near the C-terminal end of the MccA-derived peptides. Many features of MccB's ATP-binding site resemble those of MoeB/ThiF and E1s, with both protomers in the MccB dimer contributing to the ATP-binding site (Figure 5A; Supplementary Figure 7) (Lake et al, 2001; Walden et al, 2003a; Duda et al, 2005; Lois and Lima, 2005; Huang et al, 2007). A hydrophobic pocket (L121, I194, A213, L219) binds ATP's adenine. Polar residues (N154, R157, Q158, K170) contact ATP's ribose and phosphates. An Arg finger (94) is provided by the opposite monomer in the complex. Mutation of the Arg finger to alanine (R94A) has no significant effect on the kcat, but increases the Km for both MccA and ATP (Table II).

Figure 5
A common adenylation-active site for MccA and MccA–succinimide. (A) Close-up views of the MccB adenylation-active site is shown in stereo. The MccB–MgATP structure is shown, with the succinimide intermediate modelled after structural superposition ...

Mg2+ is coordinated by the three ATP phosphates and D214. D214 is absolutely conserved as an Asp in all UBL-activating enzymes (Supplementary Figure 2). The corresponding Asp residue was shown earlier to be essential for catalysis by the UBL-activating enzymes E. coli MoeB and the human NEDD8 E1 (Lake et al, 2001; Walden et al, 2003b). Furthermore, mutation of the corresponding Asp in the ubiquitin E1, UBA1, had pleiotropic effects, including increasing the Km for ATP and ubiquitin and decreasing the kcat for many steps in ubiquitination (Tokgoz et al, 2006). The earlier studies on UBA1 support a dual function of the Asp in binding MgATP and transition state stabilization (Tokgoz et al, 2006). The MccB residue D214 is essential for catalysis; we were unable to detect conversion of MccA to succinimide or product by the D214A mutant. (Figure 5B). However, coincubation of the D214A and ΔNTE mutants with MccA yields activity similar to mixing the isolated NTE with the ΔNTE mutant (Supplementary Figure 8). The former result could either be explained by complementation in trans or by a low concentration of mixed dimers in which ΔNTE would contribute the catalytic D214, and the full-length version harbouring the D214A mutation would contribute a peptide clamp to form one complete-active site per heterodimer. In either case, the finding of limited activity indicates the importance of the adenylation-active site working in concert with the peptide clamp domain.

Unique features of the MccB-active site

In addition to the shared properties with E1s, the MccB Y239 stands out as being unique in the adenylation catalytic site. Y239 contacts both the ATP α-phosphate and the heptapeptide substrate C-terminus (Figure 5A). Y239 is located on the opposite side of MccA from the Mg2+-coordinating D214. Together, Y239 and D214 seem to form a bridge. The MccA and succinimide peptide substrates would need to pass into a channel under the D214–Y239 bridge for adenylation. Y239 has an important catalytic role, as illustrated by the ~five-fold increase in Km for both ATP and MccA and >600-fold decrease in kcat in the Y239A mutant (Table II; Figure 5C).

Notably, the MccA C-terminus is atypical among substrates of UBL-activating enzymes as not terminating in the sequence Gly-Gly. Structural comparison reveals that UBL penultimate Glys are oriented by a key E1 Arg, which is lacking and substituted with I220 in MccB. MccB's aforementioned unique Y239 may also prevent Gly-Gly recognition. The corresponding E1 residue is a small Ala or Thr, which allows proper orientation of a UBL C-terminal Gly-Gly. In contrast, the MccB adenylation-active site is slightly displaced as compared with UBL-activating enzymes, with Y239 altering the conformation of the MccA-contacting β-strand13, and obscuring the path taken by a UBL Gly-Gly (Supplementary Figure 9). The channel leading under the D214–Y239 bridge and towards ATP is lined with backbone hydroxyls, which may also favour binding to MccA's C-terminal Asn, and prevent binding of closely related peptides as observed for MccA-N7D (Roush et al, 2008).

Successive adenylation reactions at a common-active site

A striking feature of the MccB mechanism is the usage of two ATP molecules per catalytic cycle: one for the adenylation of MccA and the second for the formation of the N–P bond through the adenylation of the succinimide intermediate (Roush et al, 2008). The mutants D214A and Y239A have significant reductions in overall turnover (Figure 5B and C); we were unable to detect any product formation by D214A starting from the heptapeptide substrate MccA (Figure 5B). The structural data suggests that there is one common-active site for MccA and the succinimide intermediate, so we therefore evaluated the processing for the synthetic peptidyl-succinimide intermediate on the second half reaction with our most deleterious point mutants, D214A and Y239A. The detection limit for high-performance liquid chromatography (HPLC) assays was 10 pmol of MccB product, which amounted to a rate of 0.01% of WT enzyme over a 24-h time period. The conversion of the succinimide intermediate to product, representing the second ATP consumption step, occurs 40 times faster than the overall reaction starting with MccA (Roush et al, 2008). Therefore, assaying the mutants with the succinimide intermediate is more sensitive. On incubation of WT MccB (0.1 μM), or the D214A or Y239A mutants at 150-fold higher enzyme concentrations (15 μM) to magnify sensitivity with 100 μM peptidyl-succinimide, product formation was detected (Figure 5D). D214A was ~150-fold slower than WT, and Y239A was ~600-fold slower than WT MccB in this second half reaction.

Therefore, the D214A and Y239A mutations markedly affect both overall turnover and the peptidyl-succinimide to acyl-phosphoramidate product step. As the peptidyl-succinimide does not accumulate in assays of D214A and Y239A starting with MccA (data not shown), we intuit that both the first and the second ATP-using steps are slowed down more than two orders of magnitude and that the formation of the heptapeptidyl-AMP, the cyclization to the succinimide, and the rearrangement to the final N–P-linked product occur in the same active site.

Conformational variability of MccB, MccA-derived heptapeptides, and nucleotide

Comparing the different structures reveals that all three components—MccB, MccA, and nucleotide—display conformational variability (Figure 6). MccB conformational flexibility occurs both at a global and a local level. Superposition of the adenylation domains of the different structures reveals that the adenylation and peptide clamp domains can adopt a range of orientations with respect to each other, opening and closing as for two halves of a clamshell (Figure 6A). Despite this flexibility, there are many connections between the two domains. As an example, N241 projects away from the MccB adenylation domain to coordinate with the location of peptide clamp domain. This link has a function in the reaction as substitution with Ala decreases kcat five-fold and increases Km ~six-fold for both MccA and ATP (Supplementary Tables I–III).

Figure 6
Conformational flexibility of MccB, MccA, and nucleotide substrates. (A) Structures of free MccB and the MccB–succinimide complex are shown superimposed over the adenylation domain, with the two protomers in each structure coloured in dark cyan/light ...

There is also extensive local flexibility at the adenylation-active site. Side-chain positions of ATP-binding residues, such as R94, and a key catalytic residue, the Mg2+-coordinating D214, vary considerably between structures (Figure 6B). The nucleotide ligands also adopt a range of conformations within the active site. Although the adenine and ribose moieties superimpose well (RMSD ~0.5 Å), the relative locations of the three phosphates in AMPCPP differ from ATP, and between distinct AMPCPP molecules in the asymmetric unit (Figure 6C).

The peptide structures also show striking conformational variability. In each MccA complex, the C-terminus is not visible for at least one peptide in the asymmetric unit (see Materials and methods). When the C-terminus is present, it is not directed towards the adenylation active site. Thus, the Cα positions of the succinimide and MccA's C-terminus in structures with or without AMPCPP differ by ~8 Å (Figure 6D). The ability of MccB to bind substrate peptides with a range of conformations, anchored by their N-terminus and backbone carbonyls, likely explains why in vivo substitutions for MccA's residues 2, 4–6 are tolerated for the production of a complete MccC7 variant, but few are accepted for T3 (Kazakov et al, 2007).

The peptide C-terminal flexibility was unexpected and may be a unique feature of the MccB–MccA system when compared with UBL-activating enzymes. For each of the five structurally determined UBL-activating enzyme complexes, there are 2–21 independent UBL models that superimpose well at their C-termini (Lake et al, 2001; Walden et al, 2003a; Lois and Lima, 2005; Lehmann et al, 2006; Huang et al, 2007; Lee and Schindelin, 2008; Souphron et al, 2008), and they also superimpose well for the complexes from different pathways. Further studies will be necessary to reveal whether UBL C-termini also display such conformational flexibility in complexes with their activating enzymes, or whether this is a unique feature of the MccB–MccA system.

E1-like protein that functions without forming a thioester or an acyl-disulfide intermediate

UBL-activating enzymes generally catalyse multiple reactions. After UBL adenylation, subsequent reactions often involve formation of a covalent UBL~enzyme complex, such as a thioester or acyl-disulfide intermediate linked through an activating enzyme thiol (Ciechanover et al, 1981; Haas and Rose, 1982; Haas et al, 1982; Haas and Siepmann, 1997; Johnson et al, 1997; Xi et al, 2001; Bohnsack and Haas, 2003; Pickart and Eddins, 2004), although neither E. coli MoeB nor ThiF from some organisms require such a covalent enzyme–UBL intermediate (Leimkuhler et al, 2001; Park et al, 2003). To explore the possibility of an acyl-S-enzyme intermediate for MccB, all cysteines not involved in zinc binding (Cys residues 44, 68, 123, 227 and 302) were mutated to alanines (Supplementary Figure 10A). No MccB Cys is essential for catalysis. One Cys, C123, is located in the Gly-Cys-Gly-Gly-Ile-Gly Walker A motif (Walker et al, 1982), and thus seems to have a structural function in establishing the nucleotide-binding site (Figure 7). In other UBL-activating enzymes, the residue corresponding to C123 is either a Leu or a Val, so, therefore, the C123V and C123L mutants were considered. All Cys mutants had similar activity profiles to the WT enzyme (Supplementary Figure 10B) with the exception of C123L and C123V. Conversion of C123 to bulkier, hydrophobic residues caused the succinimide intermediate to be present at higher levels than in reactions with the WT enzyme. Mutation of C123 to a bulkier Leu impaired activity (Figure 7B; Table II). Consistent with the structurally important role of C123, iodoacetamide is inactivating, likely due to an indirect effect on the structure of the adenylation-active site (Figure 7; Supplementary Figure 10). Thus, unlike E1s and the prokaryotic UBL-activating enzyme ThiF (Xi et al, 2001), MccB function does not require thiol-mediated catalysis. Intramolecular capture of the nascent peptidyl-AMP is the only kinetically significant outcome.

Figure 7
A structural role for the only non-zinc-binding Cys important for MccB activity. (A) Close-up view of the MccB structure showing the location of C123 in the Gly-Cys-Gly-Gly-Ile-Gly Walker A nucleotide-binding motif of MccB. (B) MccB C123L and C123V mutants ...

Prokaryotic variations on E1–UBL themes

Prokaryotic UBL–E1-like complexes, ThiS–ThiF and MoaD–MoeB, have long been recognized for their roles in sulphur incorporation into thiazole and molybdopterin, respectively (Rajagopalan, 1997; Begley et al, 1999). Both ThiS and MoaD structurally resemble UBLs, and are activated by ThiF or MoeB-catalysed adenylation of their C-terminal Gly-Gly sequences. ThiF and MoeB consist exclusively of an adenylation domain (Lake et al, 2001; Duda et al, 2005; Lehmann et al, 2006), and thus represent minimal modules harbouring UBL recognition and adenylation activities.

The structures of MccB reveal four key features that functionally distinguish its unique properties relative to ThiF, MoeB, and all other UBL-activating enzyme family members. The first two define MccB as a modifier of peptide, and not of a UBL: (1) the ~100-residue NTE and (2) an extended crossover loop sequence together form the peptide clamp domain. The second two define MccB as NOT recognizing a C-terminal Gly-Gly motif: (3) the lack of a conserved Arg at MccB's position I220 and (4) the D214–Y239 bridge (Y239 being unique to MccB). This latter characteristic may also contribute to MccB's selectivity for a C-terminal Asn. Interestingly, these features define a large clade of the UBL-activating enzyme family. Database searches revealed that numerous bacterial species have proteins sharing all these hallmark features of MccB (Severinov et al, 2007). Furthermore, potential paralogues of MccA and downstream components of the MccC7 biosynthetic pathway have also been identified in these same organisms. Notably, relative to the active site sequences, the anticipated peptide-binding residues are poorly conserved among the MccB homologues, and the MccA paralogues also display a range of sequences, while maintaining a C-terminal Asn. Although these enzymes remain uncharacterized, it seems likely that the MccB-like enzymes from many bacteria use a range of starting peptide sequences to generate phosphoramidate-linked analogues of aspartic acid acyl-adenylate.

A recent study identified a novel Mycobacterial UBL, Pup, which like ubiquitin becomes C-terminally conjugated to a target's Lys through an isopeptide linkage (Pearce et al, 2008). Similar to MccA, Pup's C-terminal residue is not a Gly, but rather a Gln that seems to undergo deamidation during conjugation. No E1 has been identified for Pup. Thus, it seems that prokaryotes use a wide array of variations of UBLs, E1s, or both not only for sulphur incorporation in biosynthetic pathways, but also for a range of posttranslational modifications of proteins (e.g. pupylation) and of peptides for conversion to antibiotics.

Implications for MccB-catalysed production of a phosphoramidate-linked analogue of aspartic acid acyl-adenylate

Our structural and biochemical data yielded several insights into a unique member of the UBL-activating enzyme family. MccB can use peptide substrates by virtue of a novel and essential peptide clamp domain. The clamp works in concert with the opposing surface, which is the adenylation domain. Despite similarities to the adenylation-active sites of UBL-activating enzymes, MccB catalyses two adenylation reactions: first on a C-terminal Asn and subsequently on a succinimide moiety to form an N–P bond. Both substrates are directed towards the active site for attack on the ATP α-phosphate and consequent adenylation. The D214A and Y239A mutants are significantly impaired in converting both the first- and second-peptide substrates to product; a single mutation hinders both adenylation events. In our crystal structures, these residues are poised near the ATP α-phosphate to position the substrate C-termini for adenylation. Therefore, it seems that each adenylation occurs within the same active site.

Although none of the structurally observed peptide C-termini are positioned for catalysis, a catalytically relevant model for the final reaction can be generated by subtle substrate peptide rearrangements. Rotation about the succinimide C–N bond ~180° reduces the catalytic gap within 4 Å (Figure 8). The model is consistent with a mechanism, whereby the MccB Y239 and the Mg2+ ion coordinated by MccB's D214 orient the ATP α-phosphate. On the basis of their relative locations to the α-phosphate, we speculate that D214 and Y239 may form a bridge that positions the MccA C-terminus for the initial reaction. With the Mg2+ ion relieving electrostatic repulsion, the MccA C-terminal carboxylate oxygen can first attack the ATP α-phosphate. The Mg2+ ion and side chains from MccB R157, Q158, and K170 are poised to stabilize a developing negative charge on the ATP β-phosphate in the anticipated transient pentacovalent phosphate species, allowing resolution to MccA-peptidyl-adenylate and inorganic pyrophosphate.

Figure 8
Rotation of the succinimide about the C–N bond may position the intermediate for catalysis. An ~180° rotation about the structurally observed succinimide group allows proximity to the ATP α-phosphate, allowing a model for ...

Subsequent steps in this antibiotic biosynthetic mechanism are exclusive to MccB, and are not shared by UBL-activating enzymes. The MccB structures reveal features that can accommodate the unique proceeding chemical steps on the nascent peptidyl-AMP. The structurally observed local flexibility at the adenylation-active site likely facilitates having multiple catalytic events occur there after the initial adenylation reaction. These events include the intramolecular closure of the Asn7 carboxamido nitrogen on the –CO-AMP moiety and collapse of the tetrahedral adduct to expel AMP to generate the succinimide intermediate. This may be released from the enzyme or, given the peptide clamp domain, may react with a second molecule of ATP before dissociation. Exogenous peptidyl-succinimide is chemically and kinetically competent as an intermediate for MccB conversion to the N-P-AMP final product, so that the ‘off' and ‘on' steps can be fast relative to the second adenylation.

The most remarkable chemical transformation catalysed by MccB is the subsequent transfer of AMP to the succinimido nitrogen. We propose that placement of the peptidyl-succinimide in the active site, secured by MccB's unique peptide clamp, orients the succinimide nitrogen to attack another molecule of ATP that has diffused into the active site. Thus, the peptide clamp domain would allow two adenylation reactions to be performed sequentially: the first with a typical peptide-C-terminal carboxylate oxygen as nucleophile and the second with the succinimide nitrogen as nucleophile. In the proposed mechanism, two canonically weak nucleophiles within the substrate framework are competent. The first is the entropically favoured intramolecular cyclization of the Asn7 β-carboxamide N on Cα of the peptidyl-AMP nascent product. The second is the intermolecular reaction of the cyclic succinimido nitrogen of Pα of another bound ATP, which creates the unusual N–P linkage that is the hydrolytically stable centrepiece of the antibiotic maturation logic.

Materials and methods

Purification of MccB proteins for activity assays and crystallography

For activity assays, MccB and variants were prepared as described earlier (Supplementary data) (Roush et al, 2008). Preparation of MccB truncation mutants included a thrombin cleavage step. For crystallographic studies, MccB with a C-terminal His6-tag and a cleavable, N-terminal His6-tag were purified by Ni2+ affinity, anion exchange, and gel filtration chromatography in 20 mM Tris pH 8.0, 150 mM NaCl, and 5 mM DTT. Preparation of MccB lacking the N-terminal His6-tag included a thrombin cleavage step after affinity purification. MccB was concentrated to 25–28 mg/ml by ultrafiltration at 4°C, aliquoted, flash frozen, and stored at −80°C. Peptides for crystallography were synthesized and reverse-phase HPLC purified by the Hartwell Center for Bioinformatics and Biotechnology at St Jude.

Crystallization and data collection

Crystals of MccB-His6 were grown by hanging drop vapour diffusion at 4°C in 100 mM Tris pH 8.0, 2.1 M NaCl2, cryoprotected in the mother liquor supplemented with increasing concentrations of glycerol in four steps from 10 to 40% (v/v), and flash frozen in liquid nitrogen for data collection. These crystals form in the space group P212121 with two MccB homodimers in the asymmetric unit. The MccB-His6 dimers packed so that the C-terminal His6-tag of protomers A and C passed under the crossover loops of adjacent protomers D and B, respectively. To prevent a His6-tag from interfering with the crossover loop for all complex structures, MccB produced with a cleavable and removed N-terminal His6-tag was crystallized. Screens produced crystals in several conditions at 4 and 18°C. Crystals in space group P21 were grown at 18°C by microseeding, with two MccB homodimers per asymmetric unit. The well condition varied depending on the complex; the precipitant was 24–26% pentaerythritol ethoxylate (15/4 EO/OH, Hampton Research). Additional components of each crystallization condition are as follows: MccB(MccA–N7isoAsn): 50 mM Tris pH 8.0, 50 mM (NH4)2SO4, 10 mM MccA–N7isoAsn; MccB(MccA): 50 mM Bis–Tris pH 6.5, 50 mM (NH4)2SO4, 10 mM SeMet-MccA; MccB(succinimide): 50 mM Bis–Tris pH 6.5, 50 mM (NH4)2SO4, 10 mM succinimidyl peptide; MccB(ATP): 50 mM Na Hepes pH 7.5, 100 mM MgSO4, 25 mM ATP; MccB(MccA–AMPCPP): 50 mM Na Hepes pH 7.5, 100 mM MgSO4, 9 mM α,β-methyleneadenosine 5′-triphosphate (AMPCPP) (SIGMA, catalogue #M6517), 9 mM SeMet-MccA.

Pentaerythritol ethoxylate was a suitable cryoprotectant, so crystals were flash frozen after harvesting directly from the drop. For MccB(MccA–N7isoAsn) and MccB(succinimide), the crystals were soaked in 30% pentaerythritol ethoxylate supplemented with the buffer and additives at the same concentration as in the well, and a higher concentration of peptide ligand, 60 and 20 mM, respectively.

Reflection data were collected at beamlines 8.2.1 and 8.2.2 at The Advanced Light Source and SERCAT 22-ID, 22-BM, and NECAT 24-ID at The Advanced Photon Source and were processed with HKL2000 (Otwinowski and Minor, 1997).

Structure determination, refinement, and analysis

The crystal structure of MccB-His6 (Apo) was determined to 2.8 Å by three-wavelength Zinc-MAD (Table I). Four Zn2+ sites were located and refined using the program SOLVE (Terwilliger and Berendzen, 1999). The final Z-score and figure of merit were 30.0 and 0.28, respectively. Density modification of phases from SOLVE was performed with RESOLVE (Terwilliger, 2000), yielding maps with continuous electron density from the N- to C-termini. The apo model was built manually using the ThiF model as a guide (PDB 1ZFN; Duda et al, 2005) in COOT (Emsley and Cowtan, 2004) and refined against the peak dataset using restrained refinement in REFMAC5 (Murshudov et al, 1997), with 5% of the reflections set aside as the test set. Placement of side chains was verified with simulated-annealing composite-omit maps calculated with central nervous system (CNS) (Brunger et al, 1998).

MccB(MccA–N7isoAsn) was determined by molecular replacement with PHASER (McCoy et al, 2007) using the apo dimers A–B and C–D, lacking crossover loops as independent search models. Subsequent complex structures were determined by rigid-body refinement with REFMAC5 using the MccB(MccA–N7isoAsn) structure lacking the peptide, solvent, and crossover loops. Each asymmetric unit can contain four peptides (two for each dimer), and in all structures the N-terminal Met or SeMet was observed in FoFc maps contoured at 3 σ. In addition, the anomalous signal from SeMet-MccA was used to verify the location of the N-terminus of the peptide; 10 σ peaks for the SeMet were observed in anomalous maps (Supplementary Figure 3A). Density extended beyond the N-terminal Met or SeMet for a minimum of one peptide within the asymmetric unit, so that each structure contains at least one peptide modelled from the N- to C-termini. Peptides that were not built completely are truncated. For the nucleotide-bound structures, all four cofactors are modelled. There seems to be a high propensity for anion binding to this region, as additional electron density in the MccB–MccA–N7isoAsn structure localizes to the same relative location as ATP's α-phosphate. We modelled this density as sulphate from NH4(SO4)2 in the crystallization solution. Complex structures were refined with REFMAC5 (Murshudov et al, 1997) using TLS or CNS (Brunger et al, 1998) with the same set of reflections set aside as the test set to avoid biasing the Rfree. For analysis, all A–B and C–D dimers were superimposed with MultiProt (Shatsky et al, 2004).

MccB activity assays

Peptides were synthesized through solid-phase peptide synthesis by staff at the Biopolymers Facility, Department of Biochemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA. Their purity and identities were confirmed by HPLC/LCMS. MccA and MccA–N7isoAsn have the sequences MRTGNAN and MRTGNAisoN, respectively. The succinimidyl peptide (MRTGNA-succinimide) was synthesized as described earlier (Novoa et al, 1986; Roush et al, 2008).

MccB was incubated with MccA (250 μM) in reaction buffer (75 mM Tris–HCl pH 8.0, 5 mM MgCl2, 2.5 mM TCEP, 5 mM ATP). Reactions were quenched by addition of equal volume of 0.6% TFA/H2O. Product formation was monitored by analytical HPLC using a gradient of 0–40% B in 8 min with a flow rate of 4 ml/min (solvent A: 0.1% TFA in H2O; solvent B: acetonitrile). For a list of all mutants tested, see Supplementary Table I.

Iodoacetamide inactivation of MccB

MccB (5 μM) was incubated in buffer (75 mM Tris–HCl, 5 mM MgCl2) with or without iodoacetamide (5 mM) at 37°C for 1 h. Unreacted iodoacetamide was quenched by the addition of 10 mM dithiothreitol. Product formation was monitored after incubation with ATP (5 mM) and MccA (250 μM) at room temperature for 1.5 h.

Kinetic characterization of the MccB-catalysed reaction

The detection limit for HPLC assays was 10 pmol of MccB product, which amounted to a rate of 0.01% of WT enzyme over a 24-h time period. Kinetic investigations were conducted in buffer (75 mM Tris–HCl pH 8.0, 5 mM MgCl2, 2.5 mM TCEP). The 100-μl reactions were quenched with equal volume of 0.6% TFA/H2O, and product formation was monitored by analytical HPLC. Product peaks (260 nm) were integrated and the resulting values were converted to molecules formed by using an ATP standard curve (62.5–500 μM ATP). Data were fit to the Lineweaver–Burk equation in Microsoft Excel and reported values are the mean±standard deviation of at least three independent trials. For the range of concentrations tested, see Supplementary Tables II and III.

Accession codes

The coordinates and structure factors have been deposited in the RCSB with the following accession codes: 3H5A Apo, 3H57 MccB(MgATP), 3H9G MccB(MccA-N7isoAsn), 3H9Q MccB(MccA), 3H9J MccB(MccA-AMPCPP), 3H5R MccB(Succinimide).

Supplementary Material

Supplementary Information


We thank Elizabeth Nolan and Michael Fischbach for helpful discussions. This work was supported in part by ALSAC (American Syrian Lebanese Associated Charities), grants from the NIH (RO1GM49338 to CTW, R01GM069530 to BAS, F32CA136121 to CAR, P30CA021765 to St Jude Cancer Center), and by the Howard Hughes Medical Institute. BAS is an Investigator of the Howard Hughes Medical Institute. The Advanced Light Source is supported by the Director, Office of Science, Office of Basic Energy Sciences, the US Department of Energy under Contract No. DE-AC02-05CH11231, and the Advanced Photon Source under Contract No. W-31-109-Eng-38.


  • Begley TP, Xi J, Kinsland C, Taylor S, McLafferty F (1999) The enzymology of sulfur activation during thiamin and biotin biosynthesis. Curr Opin Chem Biol 3: 623–629 [PubMed]
  • Bohnsack RN, Haas AL (2003) Conservation in the mechanism of Nedd8 activation by the human AppBp1-Uba3 heterodimer. J Biol Chem 278: 26823–26830 [PubMed]
  • Brunger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve RW, Jiang JS, Kuszewski J, Nilges M, Pannu NS, Read RJ, Rice LM, Simonson T, Warren GL (1998) Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr D Biol Crystallogr 54: 905–921 [PubMed]
  • Ciechanover A, Heller H, Katz-Etzion R, Hershko A (1981) Activation of the heat-stable polypeptide of the ATP-dependent proteolytic system. Proc Natl Acad Sci USA 78: 761–765 [PMC free article] [PubMed]
  • Duda DM, Walden H, Sfondouris J, Schulman BA (2005) Structural analysis of Escherichia coli ThiF. J Mol Biol 349: 774–786 [PubMed]
  • Duquesne S, Petit V, Peduzzi J, Rebuffat S (2007) Structural and functional diversity of microcins, gene-encoded antibacterial peptides from enterobacteria. J Mol Microbiol Biotechnol 13: 200–209 [PubMed]
  • Emsley P, Cowtan K (2004) Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr 60: 2126–2132 [PubMed]
  • Guijarro JI, Gonzalez-Pastor JE, Baleux F, San Millan JL, Castilla MA, Rico M, Moreno F, Delepierre M (1995) Chemical structure and translation inhibition studies of the antibiotic microcin C7.J Biol Chem 270: 23520–23532 [PubMed]
  • Haas AL, Rose IA (1982) The mechanism of ubiquitin activating enzyme. A kinetic and equilibrium analysis. J Biol Chem 257: 10329–10337 [PubMed]
  • Haas AL, Siepmann TJ (1997) Pathways of ubiquitin conjugation. FASEB J 11: 1257–1268 [PubMed]
  • Haas AL, Warms JV, Hershko A, Rose IA (1982) Ubiquitin-activating enzyme. Mechanism and role in protein-ubiquitin conjugation. J Biol Chem 257: 2543–2548 [PubMed]
  • Hochstrasser M (2000) Evolution and function of ubiquitin-like protein-conjugation systems. Nat Cell Biol 2: E153–E157 [PubMed]
  • Hochstrasser M (2009) Origin and function of ubiquitin-like proteins. Nature 458: 422–429 [PMC free article] [PubMed]
  • Huang DT, Hunt HW, Zhuang M, Ohi MD, Holton JM, Schulman BA (2007) Basis for a ubiquitin-like protein thioester switch toggling E1-E2 affinity. Nature 445: 394–398 [PMC free article] [PubMed]
  • Johnson ES, Schwienhorst I, Dohmen RJ, Blobel G (1997) The ubiquitin-like protein Smt3p is activated for conjugation to other proteins by an Aos1p/Uba2p heterodimer. EMBO J 16: 5509–5519 [PMC free article] [PubMed]
  • Kazakov T, Metlitskaya A, Severinov K (2007) Amino acid residues required for maturation, cell uptake, and processing of translation inhibitor microcin C. J Bacteriol 189: 2114–2118 [PMC free article] [PubMed]
  • Lake MW, Wuebbens MM, Rajagopalan KV, Schindelin H (2001) Mechanism of ubiquitin activation revealed by the structure of a bacterial MoeB-MoaD complex. Nature 414: 325–329 [PubMed]
  • Lee I, Schindelin H (2008) Structural insights into E1-catalyzed ubiquitin activation and transfer to conjugating enzymes. Cell 134: 268–278 [PubMed]
  • Lehmann C, Begley TP, Ealick SE (2006) Structure of the Escherichia coli ThiS-ThiF complex, a key component of the sulfur transfer system in thiamin biosynthesis. Biochemistry 45: 11–19 [PMC free article] [PubMed]
  • Leimkuhler S, Wuebbens MM, Rajagopalan KV (2001) Characterization of Escherichia coli MoeB and its involvement in the activation of molybdopterin synthase for the biosynthesis of the molybdenum cofactor. J Biol Chem 276: 34695–34701 [PubMed]
  • Lois LM, Lima CD (2005) Structures of the SUMO E1 provide mechanistic insights into SUMO activation and E2 recruitment to E1. EMBO J 24: 439–451 [PMC free article] [PubMed]
  • McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC, Read RJ (2007) Phaser crystallographic software. J Appl Cryst 40: 17 [PMC free article] [PubMed]
  • Metlitskaya A, Kazakov T, Kommer A, Pavlova O, Praetorius-Ibba M, Ibba M, Krasheninnikov I, Kolb V, Khmel I, Severinov K (2006) Aspartyl-tRNA synthetase is the target of peptide nucleotide antibiotic Microcin C. J Biol Chem 281: 18033–18042 [PubMed]
  • Mootz HD, Schwarzer D, Marahiel MA (2002) Ways of assembling complex natural products on modular nonribosomal peptide synthetases. Chembiochem 3: 490–504 [PubMed]
  • Murshudov GN, Vagin AA, Dodson EJ (1997) Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr 53: 240–255 [PubMed]
  • Nolan EM, Walsh CT (2008) Investigations of the MceIJ-catalyzed posttranslational modification of the microcin E492 C-terminus: linkage of ribosomal and nonribosomal peptides to form ‘trojan horse' antibiotics. Biochemistry 47: 9289–9299 [PubMed]
  • Novoa MA, Diaz-Guerra L, San Millan JL, Moreno F (1986) Cloning and mapping of the genetic determinants for microcin C7 production and immunity. J Bacteriol 168: 1384–1391 [PMC free article] [PubMed]
  • Otwinowski Z, Minor W (1997) Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol 276: 307–326
  • Park JH, Dorrestein PC, Zhai H, Kinsland C, McLafferty FW, Begley TP (2003) Biosynthesis of the thiazole moiety of thiamin pyrophosphate (vitamin B1). Biochemistry 42: 12430–12438 [PubMed]
  • Pearce MJ, Mintseris J, Ferreyra J, Gygi SP, Darwin KH (2008) Ubiquitin-like protein involved in the proteasome pathway of Mycobacterium tuberculosis. Science 322: 1104–1107 [PMC free article] [PubMed]
  • Pickart CM, Eddins MJ (2004) Ubiquitin: structures, functions, mechanisms. Biochim Biophys Acta 1695: 55–72 [PubMed]
  • Rajagopalan KV (1997) Biosynthesis and processing of the molybdenum cofactors. Biochem Soc Trans 25: 757–761 [PubMed]
  • Roush RF, Nolan EM, Lohr F, Walsh CT (2008) Maturation of an Escherichia coli ribosomal peptide antibiotic by ATP-consuming N-P bond formation in microcin C7. J Am Chem Soc 130: 3603–3609 [PubMed]
  • Severinov K, Semenova E, Kazakov A, Kazakov T, Gelfand MS (2007) Low-molecular-weight post-translationally modified microcins. Mol Microbiol 65: 1380–1394 [PubMed]
  • Shatsky M, Nussinov R, Wolfson HJ (2004) A method for simultaneous alignment of multiple protein structures. Proteins 56: 143–156 [PubMed]
  • Souphron J, Waddell MB, Paydar A, Tokgoz-Gromley Z, Roussel MF, Schulman BA (2008) Structural dissection of a gating mechanism preventing misactivation of ubiquitin by NEDD8's E1. Biochemistry 47: 8961–8969 [PMC free article] [PubMed]
  • Terwilliger TC (2000) Maximum-likelihood density modification. Crystallogr D Biol Crystallogr 56: 965–972 [PMC free article] [PubMed]
  • Terwilliger TC, Berendzen J (1999) Automated MAD and MIR structure solution. Crystallogr D Biol Crystallogr 55 (Part 4): 849–861 [PMC free article] [PubMed]
  • Tokgoz Z, Bohnsack RN, Haas AL (2006) Pleiotropic effects of ATP.Mg2+ binding in the catalytic cycle of ubiquitin-activating enzyme. J Biol Chem 281: 14729–14737 [PubMed]
  • Walden H, Podgorski MS, Huang DT, Miller DW, Howard RJ, Minor DL Jr, Holton JM, Schulman BA (2003a) The structure of the APPBP1-UBA3-NEDD8-ATP complex reveals the basis for selective ubiquitin-like protein activation by an E1. Mol Cell 12: 1427–1437 [PubMed]
  • Walden H, Podgorski MS, Schulman BA (2003b) Insights into the ubiquitin transfer cascade from the structure of the activating enzyme for NEDD8. Nature 422: 330–334 [PubMed]
  • Walker JE, Saraste M, Runswick MJ, Gay NJ (1982) Distantly related sequences in the alpha- and beta-subunits of ATP synthase, myosin, kinases and other ATP-requiring enzymes and a common nucleotide binding fold. EMBO J 1: 945–951 [PMC free article] [PubMed]
  • Xi J, Ge Y, Kinsland C, McLafferty FW, Begley TP (2001) Biosynthesis of the thiazole moiety of thiamin in Escherichia coli: identification of an acyldisulfide-linked protein—protein conjugate that is functionally analogous to the ubiquitin/E1 complex. Proc Natl Acad Sci USA 98: 8513–8518 [PMC free article] [PubMed]

Articles from The EMBO Journal are provided here courtesy of The European Molecular Biology Organization
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...


  • Compound
    PubChem Compound links
  • MedGen
    Related information in MedGen
  • Pathways + GO
    Pathways + GO
    Pathways, annotations and biological systems (BioSystems) that cite the current article.
  • Protein
    Published protein sequences
  • PubMed
    PubMed citations for these articles
  • Structure
    Published 3D structures
  • Substance
    PubChem Substance links
  • Taxonomy
    Related taxonomy entry
  • Taxonomy Tree
    Taxonomy Tree

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...