Fosfomycin Biosynthesis via Transient Cytidylylation of 2

Jul 20, 2017 - Gliotoxin Biosynthesis: Structure, Mechanism, and Metal ... The biosynthesis of gliotoxin, a virulence factor of the human pathogenic f...
0 downloads 0 Views 2MB Size
Articles pubs.acs.org/acschemicalbiology

Fosfomycin Biosynthesis via Transient Cytidylylation of 2‑Hydroxyethylphosphonate by the Bifunctional Fom1 Enzyme Su-Hee Cho,†,∥ Seung-Young Kim,‡,∥ Takeo Tomita,†,∥ Taro Shiraishi,† Jin-Soo Park,† Shusuke Sato,§ Fumitaka Kudo,§ Tadashi Eguchi,§ Nobutaka Funa,‡ Makoto Nishiyama,† and Tomohisa Kuzuyama*,† †

Biotechnology Research Center, The University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan Department of Food Science and Biotechnology, University of Shizuoka, 52-1 Yada, Suruga-ku, Shizuoka 422-8526, Japan § Department of Chemistry, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8551, Japan ‡

S Supporting Information *

ABSTRACT: Fosfomycin is a wide-spectrum phosphonate antibiotic that is used clinically to treat cystitis, tympanitis, etc. Its biosynthesis starts with the formation of a carbon−phosphorus bond catalyzed by the phosphoenolpyruvate phosphomutase Fom1. We identified an additional cytidylyltransferase (CyTase) domain at the Fom1 N-terminus in addition to the phosphoenolpyruvate phosphomutase domain at the Fom1 C-terminus. Here, we demonstrate that Fom1 is bifunctional and that the Fom1 CyTase domain catalyzes the cytidylylation of the 2-hydroxyethylphosphonate (HEP) intermediate to produce cytidylyl-HEP. On the basis of this new function of Fom1, we propose a revised fosfomycin biosynthetic pathway that involves the transient CMPconjugated intermediate. The identification of a biosynthetic mechanism via such transient cytidylylation of a biosynthetic intermediate fundamentally advances the understanding of phosphonate biosynthesis in nature. The crystal structure of the cytidylyl-HEP-bound CyTase domain provides a basis for the substrate specificity and reveals unique catalytic elements not found in other members of the CyTase family.

the Fom3 enzyme with HEP yielded an unidentified product in a SAM-, sodium dithionite-, and methylcobalamin-dependent manner.15 In the final step, HPP is converted to fosfomycin by the nonheme-iron peroxidase Fom4.16−19 Thus, the biosynthetic mechanism that links the HEP and HPP intermediates remains to be elucidated. To gain insight into the biosynthetic route leading to HPP from HEP, we re-examined the fosfomycin biosynthetic gene cluster ( fom) in Streptomyces wedmorensis.5 As a consequence, we found that Fom1 has an extra cytidylyltransferase (CyTase) domain in its N-terminus, in addition to its C-terminal PEP phosphomutase domain (Figure 1b). This domain is also found in the N-terminus of the PEP phosphomutase encoded by the gene ( fom1) from another fosfomycin-producing bacterium, S. f radiae. By contrast, neither BcpB9 nor FrbD,10 both of which encode PEP phosphomutases that catalyze the first step of the biosynthesis of bialaphos and FR-900098, respectively, has this extra domain. CyTase belongs to a family of transferases that specifically transfer a phosphorus-containing nucleotide group to another substrate (known as nucleotidyltransferases).20 For

Fosfomycin is a natural phosphonate antibiotic discovered in the broth from the soil-dwelling bacterium Streptomyces f radiae1 and shows broad-spectrum activities against Gram-positive and Gram-negative bacteria.2 Therefore, fosfomycin is used to treat cystitis, tympanitis, etc. Although fosfomycin has a simple structure (Figure 1a),3 its biosynthetic pathway has not been fully elucidated. To date, the pathway has been proposed to involve five sequential reactions catalyzed by the enzymes Fom1, Fom2, FomC, Fom3, and Fom4 (Figure 1a).4−7 The biosynthetic pathway begins with the conversion of phosphoenolpyruvate (PEP) to phosphonopyruvate (PnPy). This conversion is catalyzed by PEP phosphomutase, an enzyme common to the biosynthesis of all natural phosphonates including the herbicide bialaphos8,9 and the antibiotic FR900098.10 In fosfomycin biosynthesis, Fom1 catalyzes the first step to form PnPy, which is then decarboxylated to phosphonoacetaldehyde (PnAA) by the PnPy decarboxylase Fom2.11 PnAA is then reduced to 2-hydroxyethylphosphonate (HEP) by the dehydrogenase FomC.12 Next, the methylcobalamin-dependent radical-S-adenosyl-L-methionine (SAM) enzyme, Fom3, is thought to catalyze stereospecific methylation at C2 of HEP to form 2S-hydroxypropylphosphonate (HPP).13,14 However, its enzymatic activity has so far remained elusive; preliminary evidence has only shown that incubation of © XXXX American Chemical Society

Received: May 18, 2017 Accepted: July 12, 2017

A

DOI: 10.1021/acschembio.7b00419 ACS Chem. Biol. XXXX, XXX, XXX−XXX

Articles

ACS Chemical Biology

Figure 1. Fosfomycin biosynthesis. (a) Biosynthetic pathway of fosfomycin. The revised steps that we propose here are framed. The hypothetical steps are shown as dotted arrows. The fosfomycin biosynthetic gene cluster (fom) is also shown. FomA and FomB proteins are responsible for the bacteria’s self-resistance.22−24 FomD is a DUF (Domain of Unknown Function) 402 domain-containing protein. (b) The domain organization of Fom1. aa, amino acids.

example, choline-phosphate cytidylyltransferase (EC 2.7.7.15; CCT), which is responsible for regulating phosphatidylcholine content in cell membranes, catalyzes the formation of a CDPcholine conjugate using CTP and choline phosphate as the substrates.21 The existence of the N-terminal CyTase domain in Fom1 has been overlooked because Fom1 has been considered as a PEP phosphomutase that solely catalyzes the conversion of PEP to PnPy.5−7,11 This finding prompted us to elucidate the biochemical function of the CyTase domain of Fom1. Here, we demonstrate that the Fom1 CyTase domain is essential for the biosynthesis of fosfomycin and catalyzes cytidylyl transfer to the phosphonate group of HEP to produce a novel CMPconjugated phosphonate intermediate. On the basis of the new function of Fom1, we propose a revised biosynthetic pathway of fosfomycin, which involves the transient CMPconjugated intermediate (Figure 1a). Furthermore, we determined the crystal structure of the HEP−CMP-bound form of the Fom1 CyTase domain, which provides a basis for the substrate specificity and verifies that HEP−CMP is a biosynthetic intermediate in fosfomycin biosynthesis.

Figure 2. Production of fosfomycin by S. wedomorensis and mutants. LC/MS analyses (extracted ion count chromatograms at m/z 137.0009) of extracts from the S. wedomorensis wild-type and mutant strains. The Δfom1 and Δfom2 mutants were individually incubated in the absence and presence of 5.0 mM HEP or 5.0 mM racemic HPP.



RESULTS In Vivo Function of Fom1. To investigate the function of the Fom1 CyTase domain in fosfomycin biosynthesis, we inactivated the fom1 gene in S. wedmorensis (Supporting Information Figure 1). The resultant deletion mutant, Δfom1, and the Δfom2 mutant (NP-7)11 were grown, and fosfomycin production was analyzed by LC/MS. This LC/MS analysis revealed that both mutants were unable to produce fosfomycin (Figure 2). When HPP was added to the culture medium, fosfomycin production by both mutants was recovered, whereas HEP supplementation resulted in recovery of the fosfomycin production in Δfom2 but not in Δfom1. These results indicated that the fom1 gene product not only catalyzes the conversion of PEP to PnPy (through the PEP phosphomutase reaction) but is also essential for an unknown conversion that produces HPP from HEP. Thus, we hypothesized that Fom1 is bifunctional and that its CyTase domain may catalyze cytidylyl transfer to HEP to yield an unknown intermediate that leads to the formation of HPP.

Characterization of the CyTase domain. To test our hypothesis, we overexpressed the Fom1 CyTase domain only as well as the full-length Fom1 in Escherichia coli as N-terminal His8-tagged proteins and purified the recombinant proteins to near homogeneity (Supporting Information Figure 2). Recombinant full-length Fom1 was incubated with HEP in the presence of CTP, and the reaction mixture was subjected to HPLC analysis. The analysis revealed the formation of a new product concomitant with almost complete consumption of CTP (Figure 3a). The CyTase domain only also produced this compound in a comparable amount. The product was then purified from the reaction mixture and identified as a CMPconjugated HEP (cytidylyl−HEP or HEP−CMP) by HR/MS and extensive NMR spectral analyses (Figure 3b, Supporting Information Table 1 and Figures 3−8). These results demonstrate that Fom1 is bifunctional and that the Fom1 CyTase domain catalyzes cytidylylation of the HEP intermediate to produce HEP−CMP (Figure 3b). It should be B

DOI: 10.1021/acschembio.7b00419 ACS Chem. Biol. XXXX, XXX, XXX−XXX

Articles

ACS Chemical Biology

using CTP as a substrate, though it also reacts with ATP to some extent (Supporting Information Figure 10). Crystal Structure of the CyTase Domain. To probe the molecular basis of the function of the Fom1 CyTase domain, we conducted a crystallographic analysis. We first attempted to crystallize the full-length Fom1 protein, but we could not screen crystallization conditions because the recombinant protein readily aggregated in the buffer. Therefore, we decided to perform a crystallographic analysis of the Fom1 CyTase domain. The crystal structure of the CyTase domain was determined by the SeMet MAD method, and the native CyTase domain structure in complex with the product, HEP−CMP, was refined at 1.93 Å resolution (Supporting Information Table 2). There was one monomer in the crystallographic asymmetric unit (Figure 4a). Models were built from Gln2 to Ile127. The CyTase domain adopts an α/β nucleotide binding fold composed of a twisted five-stranded parallel β-sheet flanked by five helices, a structure that is found in many other α/β nucleotide binding fold proteins (a family of the nucleotidyltransferase α/β phosphodiesterases), including class I aminoacyl-tRNA synthetases.20 Using the PDBeFold server (http:// www.ebi.ac.uk/msd-srv/ssm), we found that the CyTase domain exhibits high structural similarity with CTP:glycerol3-phosphate cytidylyltransferase (CGT) from Bacillus subtilis (BsCGT; PDB code 1N1D, a 1.37 Å Cα RMSD over 106 aligned residues),25 CGT from Staphylococcus aureus (PDB code 2B7L, a 1.52 Å Cα RMSD over 105 aligned residues),26 and CTP:phosphocholine cytidylyltransferase from the rat (ratCCT; PDB code 3HL4, a 1.93 Å Cα RMSD over 116 aligned residues).27 Analysis using the PDBePISA server (http://www.ebi.ac.uk/pdbe/pisa) showed that the CyTase domain forms a homodimer with a buried surface area of 714 Å2. The dimer interface was formed with the helices α1 and α3 and the loop leading to α1 (Figure 4b). BsCGT and ratCCT also form homodimers in a manner similar to that of the Fom1 CyTase domain. HEP−CMP Binding Characteristics. The product of the CyTase-catalyzed reaction, HEP−CMP, was bound at the active site pocket of the Fom1 CyTase domain (Figure 4c and d), although we added the substrates, HEP and CTP, and a cofactor, Mg2+ ions, to the protein solution used for crystallization. We assume that the added substrates were converted to HEP−CMP via the CyTase-catalyzed reaction and

Figure 3. Functional characterization of the CyTase domain of Fom1. (a) LC profiles for the enzymatic conversion of CTP and HEP by the full-length Fom1 or the CyTase domain of Fom1. An unknown product (*) was purified from the reaction mixture and identified as HEP−CMP. (b) Reaction catalyzed by the CyTase domain of Fom1. The CyTase domain condenses HEP and CTP to produce HEP− CMP and a pyrophosphate coproduct.

noted that a single enzyme catalyzes both the first-step and fourth-step reactions in a biosynthetic pathway of a natural product. An investigation into whether the CyTase reaction is dependent on a metal ion for the CyTase reaction revealed that it requires a divalent metal ion, such as Mg2+, Ni2+, Fe2+, Mn2+, or Co2+, as a cofactor (Supporting Information Figure 9). Notably, a much greater amount of the product was observed when Co2+ was used as the metal ion, suggesting that the Fom1 CyTase domain exhibits its highest activity when using Co2+ as a cofactor. An examination of the nucleotide specificity demonstrated that Fom1 exhibits the highest activity when

Figure 4. Structure of the CyTase domain of Fom1. (a) CyTase domain in the asymmetric unit of the crystal. The secondary structure elements are shown as a ribbon model. (b) The dimeric structure of the CyTase domain. One subunit is in green, and the other is in white. The portions that form the dimer interface of the subunits are in yellow. (c) An omit difference (Fo − Fc) electron density map contoured at the 3.5-σ level for the bound HEP−CMP. (d) HEP−CMP-binding site. A stereoview of this figure is shown in Supporting Information Figure 11. C

DOI: 10.1021/acschembio.7b00419 ACS Chem. Biol. XXXX, XXX, XXX−XXX

Articles

ACS Chemical Biology

Fom3 may catalyze the methylation of HEP−CMP to form HPP−CMP. Very recently, we found that Fom3 actually methylates HEP−CMP in a SAM-dependent manner to produce HPP−CMP.28 Presumably, an unidentified nucleotide hydrolase-like enzyme then catalyzes the hydrolytic breakdown of HPP−CMP to produce HPP and CMP. Thus, we propose a revised fosfomycin biosynthetic pathway that includes the transient CMP-conjugated intermediates (Figure 1a). Such transient cytidylylation of a biosynthetic intermediate has been reported in the biosynthesis of the phosphonate antibiotic FR900098.29 In FR-900098 biosynthesis, FrbH, which is composed of an N-terminal cytidylyltransferase domain and a C-terminal decarboxylase domain, catalyzes the cytidylylation of the biosynthetic intermediate, 2-amino-4-phosphonobutyrate, to form CMP−2-amino-4-phosphonobutyrate, followed by the subsequent decarboxylation to form CMP−3-aminobutylphosphonate. After several reaction steps, the nucleotide hydrolase FrbI ultimately catalyzes the hydrolysis of CMP− FR-900098 to produce FR-900098 and CMP. Another precedent involving transient cytidylylation of a biosynthetic intermediate is the 2-C-methylerythritol 4-phosphate (MEP) pathway for isoprene unit biosynthesis.30 In this pathway, MEP reacts with CTP through MEP cytidylyltransferase,31,32 and the resulting CMP-conjugated intermediate is further phosphorylated at the C2 position by a kinase. Cyclization then occurs, with the loss of CMP and formation of 2-C-methylerythritol 2,4-cyclodiphosphate. However, both the precedents of transient cytidylylation are apparently different from the case of Fom1, because Fom1 is a bifunctional enzyme and catalyzes both the first-step and fourth-step reactions in a biosynthetic pathway. Distribution of Chimeric Enzymes Similar to Fom1. Of the phosphonate biosynthetic gene clusters, a bifunctional CyTase-fused PEP phosphomutase is found only in the fosfomycin biosynthetic gene cluster from Streptomyces strains. By contrast, a BLAST search using full-length Fom1 as a query revealed that more than 100 types of bacteria, including proteobacteria, bacteroidetes, firmicutes, and cyanobacteria, possess bifunctional homologues (Supporting Information Table 3). These bacteria are not known to produce fosfomycin, and they lack the other genes required for fosfomycin biosynthesis. This fact suggests that Fom1 homologues are widely distributed in nature and possess as-yet unidentified function(s). Functional analysis of these homologues may lead to the discovery of novel natural products and chemistry. Unique Features of the Active Site. CyTase family enzymes all catalyze the direct attack of a nucleophilic phosphoryl at the α-phosphate of CTP to displace pyrophosphate and to create a nucleotidyl phosphoester.20 A divalent cation such as Co2+ might act as a Lewis acid that favors the nucleophilic attack on the CTP α-phosphorus atom and that stabilizes the pyrophosphate leaving group, by analogy to the suggested function of the Mg2+ ion in Methanocaldococcus jannaschii nicotinamide mononucleotide adenylyltransferase, a member of the nucleotidyltransferase α/β phosphodiesterases.33 Therefore, the Fom1 CyTase domain structure in complex with HEP−CMP alone determined in this study might reflect a state after releasing the pyrophosphate coproduct complexed with a divalent cation. Lee et al. have proposed three structural elements that catalyze analogous reactions of the CyTase family enzymes.27 Here, we describe the active site of the Fom1 CyTase domain (Figure 5a, Supporting Information Figure 14a) by comparing

that an excess quantity of the resulting HEP−CMP conveniently remained at the active site. Many residues of the CyTase domain surround HEP−CMP (Figure 4d, Supporting Information Figure 11). Of these residues, the two fingerprint sequences of the cytidylyltransferase family (the HXGH and RTEGIST(T/S) motifs), which are primarily responsible for the recognition of CTP, are included (Supporting Information Figure 12).25,27 Notably, 16HPGH in Fom1 is an exact match for the first conserved sequence, HXGH, whereas 123YTPGISST in Fom1 is somewhat different from the second conserved sequence, RTEGIST(T/S). The most significant difference is that the first arginine residue of the second conserved sequence, which binds to the γ-phosphate of CTP and stacks with a cytosine base, is replaced with Tyr123 in Fom1. The cytosine moiety of CMP forms π−π stacking with Tyr123 and forms a contact with Gly18 on the opposite side (Figure 4d, Supporting Information Figures 11, 13a). The moiety is recognized in a specific manner. The 4-amino group forms a hydrogen bond with main-chain carbonyls of Thr124 and Ile127. The N-3 atom forms a hydrogen bond with the main-chain amide of Thr124, and the 2-keto group forms a hydrogen bond with a water molecule, which binds to Thr124, Pro122, and Asn21. These hydrogen bonds ensure specific recognition of the cytosine moiety of CTP and the exclusion of other nucleotides, which is consistent with the nucleotide specificity experimental results (Supporting Information Figure 10). Asp94 and the main chain of Gly92 recognize the ribose of the CMP moiety (Figure 4d). Ser11 and Lys46 contact the αphosphate of HEP−CMP (Figure 4d), suggesting that these residues might be involved in forming this α-phosphate of CTP into a pentavalent phosphate, which is susceptible to electrophilic attack by the phosphonate of HEP. Ser11, Tyr45, Lys46, His91, and Gln100 contact the β*-phosphonate of HEP−CMP, suggesting that these residues are involved in the precise positioning of the nucleophilic phosphonate of HEP. Leu37, Leu75, and Trp95, which constitute the active site pocket wall, form hydrophobic contacts with the HEP moiety (Figure 4d, Supporting Information Figures 11, 13b). Tyr77, His91, and Gln100 specifically recognize the distal hydroxyl group of HEP with a hydrogen bond network, suggesting these residues are crucial for the specific recognition of HEP. Thus, these HEP− CMP binding characteristics in the crystal structure of the HEP−CMP-bound CyTase domain verify that HEP−CMP is a biosynthetic intermediate of fosfomycin.



DISCUSSION This study presents the identification of the CyTase domain of Fom1 as an enzyme that catalyzes the transient CMPmodification of a fosfomycin biosynthesis intermediate and provides structural evidence for the function of the CyTase domain. We propose that the Fom1 CyTase domain possesses a unique active site arrangement and is therefore a new subtype of the cytidylyltransferase family. It should be noted that Fom1 catalyzes a typical cytidylyl transfer reaction despite its unusual active site arrangement. Revised Fosfomycin Biosynthetic Pathway. Until the present study, HEP has been thought to be converted to HPP by the SAM-dependent methyltransferase Fom3 (Figure 1a).14,15 Here, we unequivocally demonstrated that the Fom1 CyTase domain condenses HEP and CTP to produce HEP− CMP. Taking this result into consideration, we hypothesized that the true substrate of Fom3 may be HEP−CMP and that D

DOI: 10.1021/acschembio.7b00419 ACS Chem. Biol. XXXX, XXX, XXX−XXX

Articles

ACS Chemical Biology

Figure 5. Comparison of active sites between the CyTase domain of Fom1 and the related enzymes. (a) CyTase domain of Fom1 bound with HEP−CMP. (b) BsCGT bound with CDP-glycerol. (c) ratCCT bound with CDP-choline. In each figure, the bound ligands and their α- and β*phosphorus atoms are in white and orange, respectively. The residues directly coordinate the phosphonate and phosphates of the bound products are shown as stick models. The residues belonging to the loop between βA and α1 including the HPGH motif is in cyan. Tyr45 and Lys46 on the loop between α2 and α3 are in magenta. His91 and Gln100 on the loop between βD and α5 are in yellow. Tyr123 on the loop following βE including the YTPGISST motif is in green. Analogous residues present in BsCGT and ratCCT are shown with the same color coding schemes. Stereoviews of this figure are shown in Supporting Information Figure 14.

BsCGT, Lys44 contacts the β*-phosphate of CDP-glycerol and Lys46 contacts the α-phosphate and β*-phosphate.35 Substitution of Lys44 or Lys46 with alanine reduced the apparent Vmax by a factor of approximately 10 and increased the Km for glycerol-3-phosphate by 125-fold.35 This result suggests that each lysine plays a role in the reaction. In ratCCT, these residues are replaced with Phe121 and Gly123, respectively. The replaced residues are not involved in the contact with CDP-choline. Alternatively, Lys122 contacts α-phosphate and β*-phosphate of CDP-choline. Substitution of Lys122 with arginine reduced the kcat by 1000-fold and increased the Km for phosphocholine by 100-fold,36 suggesting that Lys122 is a key element for the reaction in ratCCT. In Fom1, Lys46, which corresponds to Lys122 in ratCCT, contacts the α-phosphate and β*-phosphonate of HEP−CMP in a similar manner to Lys122 in ratCCT. In ratCCT, Tyr173 forms an additional contact with the β*-phosphate of CDP-choline. By contrast, Gln96 in Fom1, which corresponds to Tyr173 in ratCCT, does not form any contacts with HEP-CMP. Alternatively, Tyr45 in Fom1 directs into the active site and contacts the β*phosphonate of HEP−CMP. Therefore, a Tyr45/Lys46 pair is presumed to play an important role in the Fom1 reaction. His91 and Gln100 in Fom1 also make contact with the ββ*phosphonate of HEP−CMP. These residues and Tyr77 also make contact with the distal hydroxyl group of the HEP moiety, suggesting that these three residues may partially be involved in the substrate specificity of the Fom1 CyTase domain. In summary, although the CyTase domain of Fom1 shares the same protein fold and a similar catalytic mechanism with other members of the CyTase family, the residues involved in the mechanism are, in part, unique compared with other members.

it with the well-characterized and structure-determined CyTase homologues, BsCGT (Figure 5b, Supporting Information Figure 14b)25 and ratCCT (Figure 5c, Supporting Information Figure 14c).27 The Fom1 CyTase domain possesses similar but apparently different structural elements that bind the reaction products from those of BsCGT and ratCCT (Supporting Information Figure 12). The first element is found at the α1 helix and the loop leading to it, which contain the 16HPGH motif. The last histidine (His19) in this motif contacts an exooxygen of the α-phosphate of the bound nucleotide and thus may participate in transition state stabilization. The second element resides in the loop following the βE strand. This element is known as the RTEGIST(S/T) motif in CyTase family enzymes.25,27 The boldface positions indicate the residues that form direct hydrogen bonds with the cytosine base. The first arginine in this motif contacts the γ-phosphate of CTP in BsCGT.25 Replacement of the arginine residue of BsCGT has been reported to significantly decrease the kcat but does not affect Km for CTP,34 suggesting its direct involvement in catalysis. Of this motif in the Fom1 CyTase domain, the residues involved in cytosine recognition are conserved as Thr124 and Ile127 and actually form hydrogen bonds with the cytosine base. Intriguingly, the first arginine of this motif is replaced with Tyr123 in Fom1. The phenolic hydroxyl group of Tyr123 might form a contact with the γ-phosphate of CTP to stabilize the binding and could participate in catalysis, similar to the arginine in BsCGT. However, the significance of the tyrosine residue for catalysis is unclear at present. The third element is the residues that directly coordinate αphosphate and β*-phosphonate of the bound product; main chain NHs of Ser11 and Ala12 on the loop between βA and α1 (cyan), the side chains of Tyr45 and Lys46 on the loop between α2 and α3 (magenta), and the side chains of His91 and Gln100 on the loop between βD and α5 (yellow). This element varies among family members. For example, backbone atoms of Ser11/Ala12 in Fom1, Thr9/Phe10 in BsCGT and Ile84/Phe85 in ratCCT coordinate the α-phosphate of each bound product. Notably, the side chains of Thr9 of BsCGT and Ile84 of ratCCT do not form contacts with the α-phosphate, whereas the side chain hydroxyl group of Ser11 of Fom1 forms a hydrogen bond with the exo-oxygen of the α-phosphate of the product. This additional hydrogen bond might improve the stabilization of the α-phosphate in the transition state. In



CONCLUSIONS The biochemical analyses of Fom1 and the crystal structure of the HEP−CMP-bound complex of the Fom1 CyTase domain revealed a fosfomycin biosynthetic pathway involving the transient cytidylylation of HEP. The crystal structure also revealed the substrate specificities for HEP and CTP and unique catalytic elements of the Fom1 CyTase domain. These results help to clarify the complete fosfomycin biosynthetic pathway. E

DOI: 10.1021/acschembio.7b00419 ACS Chem. Biol. XXXX, XXX, XXX−XXX

Articles

ACS Chemical Biology

■ ■

(6) Seto, H., and Kuzuyama, T. (1999) Bioactive natural products with carbon-phosphorus bonds and their biosynthesis. Nat. Prod. Rep. 16, 589−596. (7) Woodyer, R. D., Shao, Z., Thomas, P. M., Kelleher, N. L., Blodgett, J. A., Metcalf, W. W., van der Donk, W. A., and Zhao, H. (2006) Heterologous production of fosfomycin and identification of the minimal biosynthetic gene cluster. Chem. Biol. 13, 1171−1182. (8) Seidel, H. M., Freeman, S., Seto, H., and Knowles, J. R. (1988) Phosphonate biosynthesis: isolation of the enzyme responsible for the formation of a carbon-phosphorus bond. Nature 335, 457−458. (9) Hidaka, T., Mori, M., Imai, S., Hara, O., Nagaoka, K., and Seto, H. (1989) Studies on the biosynthesis of bialaphos (SF-1293). 9. Biochemical mechanism of C-P bond formation in bialaphos: discovery of phosphoenolpyruvate phosphomutase which catalyzes the formation of phosphonopyruvate from phosphoenolpyruvate. J. Antibiot. 42, 491−494. (10) Eliot, A. C., Griffin, B. M., Thomas, P. M., Johannes, T. W., Kelleher, N. L., Zhao, H., and Metcalf, W. W. (2008) Cloning, expression, and biochemical characterization of Streptomyces rubellomurinus genes required for biosynthesis of antimalarial compound FR900098. Chem. Biol. 15, 765−770. (11) Hidaka, T., Iwakura, H., Imai, S., and Seto, H. (1992) Studies on the biosynthesis of fosfomycin. 3. Detection of phosphoenol-pyruvate phosphomutase activity in a fosfomycin high-producing strain of Streptomyces wedmorensis and characterization of its blocked mutant NP-7. J. Antibiot. 45, 1008−1010. (12) Shao, Z., Blodgett, J. A., Circello, B. T., Eliot, A. C., Woodyer, R., Li, G., van der Donk, W. A., Metcalf, W. W., and Zhao, H. (2008) Biosynthesis of 2-hydroxyethylphosphonate, an unexpected intermediate common to multiple phosphonate biosynthetic pathways. J. Biol. Chem. 283, 23161−23168. (13) Kuzuyama, T., Hidaka, T., Kamigiri, K., Imai, S., and Seto, H. (1992) Studies on the biosynthesis of fosfomycin. 4. The biosynthetic origin of the methyl group of fosfomycin. J. Antibiot. 45, 1812−1814. (14) Woodyer, R., Li, G., Zhao, H., and van der Donk, W. W. (2007) New insight into the mechanism of methyl transfer during the biosynthesis of fosfomycin. Chem. Commun., 359−361. (15) Allen, K. D., and Wang, S. C. (2014) Initial characterization of Fom3 from Streptomyces wedmorensis: The methyltransferase in fosfomycin biosynthesis. Arch. Biochem. Biophys. 543, 67−73. (16) Seto, H., Hidaka, T., Kuzuyama, T., Shibahara, S., Usui, T., Sakanaka, O., and Imai, S. (1991) Studies on the biosynthesis of fosfomycin. 2. Conversion of 2-hydroxypropylphosphonic acid to fosfomycin by blocked mutants of Streptomyces wedmorensis. J. Antibiot. 44, 1286−1288. (17) Liu, P.-H., Murakami, K., Seki, T., He, X., Yeung, S. M., Kuzuyama, T., Seto, H., and Liu, H.-W. (2001) Protein purification and function assignment of the epoxidase catalyzing the formation of fosfomycin. J. Am. Chem. Soc. 123, 4619−4620. (18) Zhao, Z., Liu, P.-H., Murakami, K., Kuzuyama, T., Seto, H., and Liu, H.-W. (2002) Mechanistic studies of HPP epoxidase: Configuration of the substrate governs its enzymatic fate. Angew. Chem., Int. Ed. 41, 4529−4532. (19) Higgins, L. J., Yan, F., Liu, P.-H., Liu, H.-W., and Drennan, C. L. (2005) Structural insight into antibiotic fosfomycin biosynthesis by a mononuclear iron enzyme. Nature 437, 838−844. (20) Bork, P., Holm, L., Koonin, E. V., and Sander, C. (1995) The cytidylyltransferase superfamily: identification of the nucleotidebinding site and fold prediction. Proteins: Struct., Funct., Genet. 22, 259−266. (21) Weinhold, P. A., Rounsifer, M. E., and Feldman, D. A. (1986) The purification and characterization of CTP:phosphorylcholine cytidylyltransferase from rat liver. J. Biol. Chem. 261, 5104−5110. (22) Kuzuyama, T., Kobayashi, S., O’Hara, K., Hidaka, T., and Seto, H. (1996) Fosfomycin monophosphate and fosfomycin diphosphate, two inactivated fosfomycin derivatives formed by gene products of fomA and fomB from a fosfomycin producing organism Streptomyces wedmorensis. J. Antibiot. 49, 502−504.

METHODS

All methods used in this study are available in the Supporting Information.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acschembio.7b00419. Methods and supplementary figures, tables, and notes (PDF) Accession Codes

Structural data are available in the RCSB PDB database under the accession number 5X3D.



AUTHOR INFORMATION

Corresponding Author

*Biotechnology Research Center, The University of Tokyo, 11-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan. E-mail: utkuz@ mail.ecc.u-tokyo.ac.jp. ORCID

Tadashi Eguchi: 0000-0002-7830-7104 Makoto Nishiyama: 0000-0001-8143-8052 Tomohisa Kuzuyama: 0000-0002-7221-5858 Author Contributions ∥

These authors contributed equally. S.-Y.K. constructed the Δfom1 mutant. S.-H.C., S.-Y.K., T.S., J.-S.P., and S.S. performed biochemical analyses. S.-H.C. prepared the crystals. S.-H.C. and T.T. performed the crystallographic analysis. S.-H.C., T.T., and T.K. analyzed the 3D structure. S.-H.C., S.-Y.K., T.T., T.S., F.K., T.E., N.F., M.N., and T.K. planned the experiments. S.-H.C., S.Y.K., T.T., T.S., and T.K. wrote the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by JSPS KAKENHI (Grant no. 25892023 to S.-Y.K., Grant no. 16H06451 to T.E. and Grant no. 16H06453 to T.K.) and JSPS A3 Foresight Program (Grant no. 16822333). This work was performed under the approval of the Photon Factory Program Advisory Committee (Proposal numbers: 2014R-10, 2015R-17, and 2016R-13).



REFERENCES

(1) Rogers, T. O., and Birnbaum, J. (1974) Biosynthesis of fosfomycin by Streptomyces f radiae. Antimicrob. Agents Chemother. 5, 121−132. (2) Reeves, D. S. (1994) Fosfomycin trometamol. J. Antimicrob. Chemother. 34, 853−858. (3) White, R. F., Birnbaum, J., Meyer, R. T., ten Broeke, J., Chemerda, J. M., and Demain, A. L. (1971) Microbial epoxidation of cis-propenylphosphonic to (−)-cis-1,2-epoxypropylphosphonic acid. Appl. Microbiol. 22, 55−60. (4) Kuzuyama, T., Hidaka, T., Imai, S., and Seto, H. (1993) Studies on the biosynthesis of fosfomycin. V. Cloning of genes for fosfomycin biosynthesis. J. Antibiot. 46, 1478−1480. (5) Hidaka, T., Goda, M., Kuzuyama, T., Takei, N., Hidaka, M., and Seto, H. (1995) Cloning and nucleotide sequence of fosfomycin biosynthetic genes of Streptomyces wedmorensis. Mol. Gen. Genet. 249, 274−280. F

DOI: 10.1021/acschembio.7b00419 ACS Chem. Biol. XXXX, XXX, XXX−XXX

Articles

ACS Chemical Biology (23) Kobayashi, S., Kuzuyama, T., and Seto, H. (2000) Characterization of fomA and fomB gene products from Streptomyces wedmorensis that confer fosfomycin resistance to Escherichia coli. Antimicrob. Agents Chemother. 44, 647−650. (24) Pakhomova, S., Bartlett, S. G., Augustus, A., Kuzuyama, T., and Newcomer, M. E. (2008) Crystal structure of fosfomycin resistance kinase FomA from Streptomyces wedmorensis. J. Biol. Chem. 283, 28518−28526. (25) Weber, C. H., Park, Y.-S., Sanker, S., Kent, C., and Ludwig, M. L. (1999) A prototypical cytidylyltransferase: CTP:glycerol-3-phosphate cytidylyltransferase from Bacillus subtilis. Structure 7, 1113−1124. (26) Fong, D. H., Yim, V. C., D’Elia, M. A., Brown, E. D., and Berghuis, A. M. (2006) Crystal structure of CTP:glycerol-3-phosphate cytidylyltransferase from Staphylococcus aureus: examination of structural basis for kinetic mechanism. Biochim. Biophys. Acta, Proteins Proteomics 1764, 63−69. (27) Lee, J., Johnson, J., Ding, Z., Paetzel, M., and Cornell, R. B. (2009) Crystal structure of a mammalian CTP: phosphocholine cytidylyltransferase catalytic domain reveals novel active site residues within a highly conserved nucleotidyltransferase fold. J. Biol. Chem. 284, 33535−33548. (28) Sato, S., Kudo, F., Kim, S.-Y., Kuzuyama, T., and Eguchi, T. (2017) Methylcobalamin-dependent radical SAM C-methyltransferase Fom3 recognizes cytidylyl-2-hydroxyethylphosphonate and catalyzes the nonstereoselective C-methylation in fosfomycin biosynthesis. Biochemistry 56, 3519−3522. (29) Johannes, T. W., DeSieno, M. A., Griffin, B. M., Thomas, P. M., Kelleher, N. L., Metcalf, W. W., and Zhao, H. (2010) Deciphering the late biosynthetic steps of antimalarial compound FR-900098. Chem. Biol. 17, 57−64. (30) Kuzuyama, T., and Seto, H. (2003) Diversity of the biosynthesis of the isoprene units. Nat. Prod. Rep. 20, 171−183. (31) Kuzuyama, T., Takagi, M., Kaneda, K., Dairi, T., and Seto, H. (2000) Formation of 4-(cytidine 5′-diphospho)-2-C-methyl-D-erythritol from 2-C-methyl-D-erythritol 4-phosphate by 2-C-methyl-Derythritol 4-phosphate cytidylyltransferase, a new enzyme in the nonmevalonate pathway. Tetrahedron Lett. 41, 703−706. (32) Rohdich, F., Wungsintaweekul, J., Fellermeier, M., Sagner, S., Herz, S., Kis, K., Eisenreich, W., Bacher, A., and Zenk, M. H. (1999) Cytidine 5′-triphosphate-dependent biosynthesis of isoprenoids: YgbP protein of Escherichia coli catalyzes the formation of 4-diphosphocytidyl-2-C-methylerythritol. Proc. Natl. Acad. Sci. U. S. A. 96, 11758− 11763. (33) D’Angelo, I., Raffaelli, N., Dabusti, V., Lorenzi, T., Magni, G., and Rizzi, M. (2000) Structure of nicotinamide mononucleotide adenylyltransferase: a key enzyme in NAD+ biosynthesis. Structure 8, 993−1004. (34) Park, Y.-S., Gee, P., Sanker, S., Schurter, E. J., Zuiderweg, E. R., and Kent, C. (1997) Identification of functional conserved residues of CTP:glycerol-3-phosphate cytidylyltransferase. Role of histidines in the conserved HXGH in catalysis. J. Biol. Chem. 272, 15161−15166. (35) Pattridge, K. A., Weber, C. H., Friesen, J. A., Sanker, S., Kent, C., and Ludwig, M. L. (2003) Glycerol-3-phosphate cytidylyltransferase. Structural changes induced by binding of CDP-glycerol and the role of lysine residues in catalysis. J. Biol. Chem. 278, 51863−51871. (36) Helmink, B. A., Braker, J. D., Kent, C., and Friesen, J. A. (2003) Identification of lysine 122 and arginine 196 as important functional residues of rat CTP:phosphocholine cytidylyltransferase alpha. Biochemistry 42, 5043−5051.

G

DOI: 10.1021/acschembio.7b00419 ACS Chem. Biol. XXXX, XXX, XXX−XXX