FEBS Letters 587 (2013) 2851–2859

journal homepage: www.FEBSLetters.org

Co-evolution of HAD and hotdog-fold thioesterase domain function in the menaquinone-pathway fusion proteins BF1314 and PG1653 ⇑ ⇑ Min Wang a, Feng Song a,1, Rui Wu b,2, Karen N. Allen b, , Patrick S. Mariano a, Debra Dunaway-Mariano a, a Department of Chemistry & Chemical Biology, University of New Mexico, Albuquerque, NM 87131, USA b Department of Chemistry, Boston University, Boston, MA 02215, USA article info abstract

Article history: The function of a Bacteroidetes menaquinone biosynthetic pathway fusion protein comprised of an Received 8 June 2013 N-terminal haloacid dehalogenase (HAD) family domain and a C-terminal hotdog-fold family Accepted 2 July 2013 domain is described. Whereas the thioesterase domain efficiently catalyzes 1,4-dihydroxy- Available online 10 July 2013 napthoyl-CoA hydrolysis, an intermediate step in the menaquinone pathway, the HAD domain is devoid of catalytic activity. In some Bacteroidetes a homologous, catalytically active 1,4-dihydroxy- Edited by Athel Cornish-Bowden napthoyl-CoA thioesterase replaces the fusion protein. Following the gene fusion event, sequence divergence resulted in a HAD domain that functions solely as the oligomerization domain of an otherwise inactive thioesterase domain. Keywords: Haloacid dehalogenase Ó 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. 14-Dihroxynapthoyl-coenzyme A evolution Enzyme superfamily Bacteroides fragilis Porphyromonas gingivalis

1. Introduction Historically, the Escherichia coli menaquinone pathway and the structures and mechanisms of the pathway have been the Menaquinone-7, also known as vitamin K2, is a -soluble focus of investigation [15–25]. Nevertheless the menaquinone molecule that, in humans, plays a crucial role in blood coagulation, pathways of other bacteria have been characterized [26–28], some bone metabolism, and calcification of arteries [1–6]. Bacteria of the of which employ different biosynthetic strategies [29]. In the first large intestine synthesize menaquinone-7 (hereafter simplified to step of the upper pathway (Fig. 1a), chorismate, derived from the ‘‘menaquinone’’) for use as an electron transporter in anaerobic shikimate pathway, is converted to isochorismate by MenF, an respiration [7,8]. Humans do not synthesize menaquinone but ac- isochorismate synthase. In the second step 2-succinyl-5-eno- quire it from their diet and intestinal flora [9]. Because the menqui- lpyruvyl-6-hydroxy-3-cyclohexene-1-carboxylate is formed by none biosynthetic pathway is restricted to bacteria, the enzymes the action of the thiamine-dependent enzyme MenD. MenH, a that function in this pathway are potential targets for the develop- member of the a,b-fold family, has recently been shown ment of novel antibiotics [10–14]. Thus, the elucidation of the to catalyze the elimination of the pyruvyl-substituent from this chemical steps of the menaquinone pathway and the catalytic intermediate. The product 2-succinyl-6-hydroxy-2,4-cyclohexadi- mechanisms of the enzyme catalysts are important objectives. ene-1-carboxylate, is converted through the elimination of H2O to 2-succinylbenzoate as catalyzed by MenC, a member of the enolase superfamily. The succinyl side chain of 2-succinylbenzoate is acti- Abbreviations: CoA, coenzyme A; HAD, haloacid dehalogenase; IPTG, isopropyl- vated for cyclization through the action of the ligase MenE that cat- 1-thio-b-D-galactopyranoside; MES, 2-(N-morpholino)ethanesulfonate; HEPES, N- (2-hydroxyethyl)piperazine-N0-2-ethanesulfonate alyzes the transformation of the carboxylate group to the coenzyme ⇑ Corresponding authors. Fax: +1 617 353 6466 (K.N. Allen), fax: +1 505 277 2609 A (CoA) thioester. The naphthoquinol skeleton of 1,4-dihydroxy- (D. Dunaway-Mariano). napthoyl-CoA is generated via a cyclization reaction catalyzed by E-mail addresses: [email protected] (K.N. Allen), [email protected] (D. Dunaway- 1,4-dihydroxynaphthoyl-CoA synthase, MenB, a member of the Mariano). crotonase family. The genes encoding these six enzymes in E. coli are 1 Present address: 3054 E Cornwallis Road, Syngenta Crop Protection, LLC, Research Triangle Park, NC 27709, USA. co-located (Fig. 1B). The 1,4-dihydroxynapthoyl-CoA is then hydro- 2 Present address: Immunogen. Inc., 830 Winter St, Waltham, MA 02451, USA. lyzed by a thioesterase of the hotdog-fold family [30–32]. In the

0014-5793/$36.00 Ó 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.febslet.2013.07.009 2852 M. Wang et al. / FEBS Letters 587 (2013) 2851–2859

Fig. 1. A: The chemical steps of the E. coli menaquinone biosynthetic pathway. B: The menaquinone biosynthetic pathway gene cluster in E. coli K-12 and in B. fragilis. The E. coli genes menf, mend, menh, menb, menc, and mene encode isochorismate synthase, 2-succinyl-5-enolpyruvyl-6-hydroxy-3-cyclohexene-1-carboxylate synthase, 2- succinyl-6-hydroxy-3-cyclohexene-1-carboxylate synthase, 2-succinylbenzoate synthase, 2-succinylbenzoate:CoA ligase, and napthoate synthase, respectively. The B. fragilis gene cluster is annotated based on the sequence homology to the E. coli K-12 genes. The two ‘‘extra genes’’ BF1314 and BF1317 encode the fusion protein and a hypothetical protein, respectfully.

lower pathway, menaquinone is generated from 1,4-dihydroxy- phosphatase domain as an oligomerization domain. In addition, napthoate in two steps. First, the polyprenyl unit from octaprenyl- through comprehensive bioinformatic analysis we track the two pyrophosphate is added in a reaction catalyzed by MenA [33]. domains throughout Bacteroidetes to provide insight into the path- Second, the napthyl ring is methylated by reaction with S-adeno- way for evolution of the fusion protein. sylmethionine as catalyzed by MenG [34]. Our interest in the biochemical basis for adaptation by bacteria 2. Materials and methods that colonize the human internal cavity lead us to a comparative analysis of the meniquinone gene clusters among the genomes of 2.1. Materials Bacteroidetes species. These species constitute a large fraction of the human gut bacterial community. Homologs to all E. coli upper Unless otherwise stated, all chemicals were obtained from Sig- pathway genes were found in these species with the exception of ma–Aldrich. Primers, T4 DNA ligase, and restriction enzymes were MenH. Most significant was the discovery of an additional gene purchased from Invitrogen, Pfu and Pfu Turbo polymerases were that co-locates with the meniquinone gene cluster in a large num- from Stratagene, the pET-14a, and pET23a, pET28a and pET28b vec- ber of the Bacteroidetes species (Fig. 1B). This gene encodes a fu- tor kits were from Novagen, and the Geneclean Spin and the Qiaprep sion protein consisting of an N-terminal domain that is Spin Miniprep kits were from Qiagen. Genomic DNA from Bacteroi- homologous to the class C2B haloacid dehalogenase (HAD) enzyme des thetaiotaomicron VPI-5482 (ATCC-29148D), Porphyromonas superfamily and a C-terminal domain that is homol- gingivalis W83, Bacteroides fragilis and Bacteroides vulgatus were ogous to type AB thioesterases of the hotdog-fold enzyme super- from ATCC. DEAE Sepharose was from Amersham Biosciences. family. Our first objective was to demonstrate that the hotdog- Butyryl-CoA, isobutyryl-CoA, hexanoyl-CoA, octanoyl-CoA, deca- fold domain catalyzes the hydrolysis of the menaquinone pathway noyl-CoA, myristoryl-CoA and benzoyl-CoA were purchased from 1,4-dihydroxynapthoyl-CoA to 1,4-dihydroxynapthoate, and by Sigma–Aldrich. The phosphate and anhydrides were pur- doing so complete the set of pathway catalysts. Our second objec- chased from Sigma–Aldrich. The 4-hydroxybenzoyl-CoA and 3- tive was to determine the function of the appended HAD domain hydroxyphenylacetyl-CoA were synthesized according to published and its relationship to the menaquinone pathway. Our approach procedures [37–40]. Napthoyl-CoA and 1,4-dihydroxynapthoyl- to this problem represents the merging of the kinetic analyses en- CoA were synthesized as detailed in Supporting Information. abled by Michaelis and Menton [35] (see [36] for a translation) 100 years ago and the bioinformatic methods that have become 2.2. Gene cloning and expression and protein purification available in the genomic era. Herein, we provide the results from steady-state kinetic analy- The genes encoding the BT4699 (NCBI accession NP_813610, ses of the targeted gene product derived from two different Bacter- UniProt accession Q89YN2), PG1653 (NCBI accession NP_905773, oidetes species, which establish the identity of the menaquinone UniProt accession Q7MU91), BF1314 (NCBI accession YP_098598, pathway thioesterase and which define the role of the HAD UniProt accession Q5LFS7) and BVU2420 (NCBI accession M. Wang et al. / FEBS Letters 587 (2013) 2851–2859 2853

YP_001299701, UniProt accession A6L315) were amplified from was calibrated using the gel-filtration low molecular weight and the genomic DNA derived from B. thetaiotaomicron, P. gingivalis high molecular weight calibration kits from Amersham W83, B. fragilis and B. vulgatus, respectively, using PCR techniques Biosciences. in combination with commercial primers that contain the endo- cleavage sites Nde I and Xho I. The PCR products were 2.5. Steady-state kinetic assays purified and then subcloned into a pET-28a vector, which adds a

His6 tag to the N-terminus of each protein. Plasmid DNA was puri- The rate of enzyme-catalyzed hydrolysis of p-nitrophenyl phos- fied using a Qiaprep Spin Miniprep Kit. The identity of the sub- phate (PNPP) was determined at 25 or 37 °C by monitoring the in- cloned genes was demonstrated by full-length DNA sequencing crease in solution absorbance at 410 nm (De = 18.4 mM1 cm1). (University of New Mexico). The 0.5 ml assay mixtures contained 50 mM HEPES (pH 7.0) and

Transformed E. coli BL21 (DE3) competent cells were grown at 5 mM MgCl2. Hydrolysis reactions of all other phosphate esters 37 °C for 6 h in 3 L of LB media containing 50 lg/ml of kanamycin were monitored using the Enzcheck phosphate assay kit (Invitro- with shaking at 220 rpm. The cells were induced with 0.2 mM iso- gen). The enzyme-catalyzed phosphate release in solutions con- propyl-1-thio-b-D-galactopyranoside (IPTG) with shaking at taining 20 mM Tris–HCl, 2 mM MgCl2, and 2 mM sodium azide 170 rpm for 12 h at 20 °C. The cells were harvested by centrifuga- (pH 7.5) was monitored at 25 °C by measuring the increase in tion (6500 rpm for 15 min at 4 °C), resuspended in 300 ml of lysis absorbance at 360 nm. buffer (50 mM N-(2-hydroxyethyl)piperazine-N0-2-ethanesulfo- Thioesterase activity was measured using a DTNB coupled as- nate (HEPES), 300 mM NaCl, 10 mM imidazole, pH 8.0) and lysed say. The reaction solutions (1 ml) contained specified concentra- by 2 passages through a French Press cell (12,000 psi). The cell ly- tions of substrate and 1 mM DTNB in 0.15 M KCl and 50 mM sate was centrifuged at 20,000 rpm for 60 min at 4 °C. The BT4699 HEPES (pH 7.5). The progress of the reaction was monitored at proved to be insoluble and was not pursued further. The respective 412 nm. BF1314, PG1653, and BVU2420-containing supernatants were sep- Initial velocities of the reactions were measured as a function of arately chromatographed on Ni-affinity columns. The columns substrate concentrations (0.5- to 10-fold Km). The initial velocity were washed with 200 ml of lysis buffer followed by 200 ml of data were fitted using Eq. (1) and the computer program KinetA- wash buffer (50 mM HEPES, 300 mM NaCl, 20 mM imidazole, pH syst to define Vmax and Km. 8.0) before eluting the His6-tagged protein with 100 ml of elution v ¼ V max½S=ð½SþKmÞð1Þ buffer (50 mM HEPES, 300 mM NaCl, 250 mM imidazole, pH 8.0). The protein containing fractions were concentrated by using a where v is the initial velocity, Vmax the maximum velocity, [S] the 10 kDa Macrosep centricon (Pall Filtron) and then dialyzed against substrate concentration, Km the Michaelis constant. The kcat values

50 mM HEPES, 300 mM NaCl (pH 7.5) at 4 °C. The His6-tag was were calculated from the ratios of Vmax and enzyme concentration. cleaved from BVU2420 to enhance its solubility at lower salt con- The pH dependence of the catalysis of glycerophosphate hydroly- centration (viz. 50 mM NaCl). Accordingly, 30 ll of thrombin sis by the BF1314 HAD domain was examined by measuring the kcat (500 U/ml) were added to the concentrated protein solution at and Km at pH 5–7.5. Reaction buffers contained various concentra- room temperature (25 °C) and the resulting solution was incubated tions of glycerophosphate, 2 mM MgCl2, 2 mM sodium azide, overnight and then passed through a Ni affinity column. The eluate 50 mM acetate and 50 mM 2-(N-morpholino)ethanesulfonate was dialyzed at 4 °C in 50 mM HEPES and 50 mM NaCl (pH 7.5). (MES) for pH 5.0–5.5, 50 mM Bis–Tris for pH 6.0–6.5, and 50 mM The protein concentration was determined by using the Brad- HEPES for pH 7.0–7.5. The hydrolysis reactions were monitored as ford assay and also, by measuring the solution absorption at described above using the Enzcheck phosphate assay kit at 25 °C. 280 nm. Protein purity was verified by SDS–PAGE to be greater than 95% with final yields as follows: BF1314 (1.2 mg/g wet cells), 3. Results and discussion PG1653 (1.7 mg/g wet cells) and BVU2420 (5.0 mg/g wet cells). 3.1. Production of the fusion protein of the Bacteroidetes menaquinone 2.3. Preparation of PG1653 truncation mutants pathway

The HAD domain and the thioesterase domain of the fusion pro- Fusion proteins, also known as chimeric proteins, are created by tein PG1653 were prepared as truncation constructs using Quick- a gene mutation that joins two protein-encoding genes to form a change site-directed mutagenesis. To generate the HAD domain single transcriptional unit. Provided that the full coding region of the codon for Val271 was mutated to the stop codon TAA. Oligonu- each gene is preserved, the translation product is likely to assume cleotide primers 50-CGA ATA TGA AGC CTA ACC TTT CTC CGT AGA- a fold that conserves the native structures and functions of the two 30 (forward), and 50-TCT ACG GAG AAA GGT TAG GCT TCA TAT TCG- parent proteins, thus forming a single-chain, bifunctional protein. 30 (reverse) were used in the PCR reaction. To generate the thioes- Bifunctional enzymes that catalyze two (not necessarily succes- terase domain the Pro272 codon was mutated to a start codon. Oli- sive) steps of a metabolic pathway are common to bacteria because gonucleotide primers 50-TAT GAA GCC CAT ATG TTC TCC GTA GAA- the genes encoding the pathway enzymes are often clustered on 30 (forward) and 50-ATG GGA AAA AAG CTT TTA CCG CTG TTC AGG- the bacterial genome. 30 (reverse) were used in the PCR reactions. The respective PCR The fusion protein of the Bacteroidetes menaquinone pathway products were cloned into the PET-28b vector, which was used consists of an N-terminal HAD phosphatase domain and a C-termi- to transform competent E. coli BL21 (DE3) cells as described in nal type AB hotdog-fold thioesterase domain. The amino-acid se- the previous section. Purification of the truncation mutants was quences of the fusion proteins identified via a BLAST search of carried out as described for the wild-type PG1653. the NCBI gene databank is shown in Fig. SI1. Our initial hypothesis was that the thioesterase domain catalyzes the hydrolysis of the 2.4. Native molecular weight determination 1,4-dihydroxynapthoyl-CoA, and that the HAD domain catalyzes inorganic pyrophosphate hydrolysis. Inorganic pyrophosphate is Chromatography was performed using a ÄKTA P-920 FPLC sys- generated by the 2-succinylbenzoate-CoA synthase (MenE) of the tem equipped with a Hiprep 16/60 Sephacryl S-200 HR column (GE upper pathway and by 1,4-dihydroxynapthoate-octaprenyltrans- Healthcare). The size exclusion column was equilibrated with ferase (MenA) of the lower pathway. In a previous study we had 50 mM HEPES (pH 7.5), 500 mM NaCl, and 1 mM DTT. The column shown that the HAD protein BT2127 of B. thetaiotaomicron is a 2854 M. Wang et al. / FEBS Letters 587 (2013) 2851–2859 pyrophosphatase and, based on its gene context, posited that it 3.2. Catalytic activity of the hotdog-fold domain in the fusion protein drives the coupled reactions catalyzed by a bifunctional L-fucose kinase/L-fucose 1-phosphate guanyltransferase forward via hydro- A search of the NCBI genebank using the program BLAST re- lysis of the pyrophosphate product [41]. By analogy, the reaction vealed that the closest homologs to the fusion protein hotdog-fold catalyzed by the HAD domain of the Bacteroidetes menaquinone domains belong to the hotdog-fold PaaI/YdiI family. Members of pathway fusion protein might be coupled with the reactions cata- this family are type AB thioesterases, structurally represented lyzed by MenE and/or MenA. In order to test the proposed func- (Fig. 2A) by E. coli ydiI (38% sequence identity; PDB accession code tions of the fusion proteins, the encoding genes from two species 1VH5). The hotdog-fold domains of each of the Bacteroidetes fu- of Bacteroides (thetaiotaomicron BT4699 and fragilis BF1314; sion proteins conserve the catalytic Glu (Glu333 using BF1314 82.7% pairwise sequence identity) along with an encoding gene numbering) residue that corresponds to Glu63 in the E. coli ydiI from a more distant Bacteroidetes species, Porphyromonas gingiva- and Glu58 in the B. vulgatus BVU2420 (see sequence logo in lis PG1653 (43.0% and 42.1% pairwise sequence identity, respec- Fig. 3 and sequence alignments in Figs SI1 and SI2). To demonstrate tively) were cloned for overexpression in E. coli. The B. that these proteins are indeed bona fide thioesterases, they were thetaiotaomicron BT4699 was found to have limited solubility in evaluated as catalysts for aroyl- and acyl-CoA hydrolysis by the His6-tagged and native forms and was therefore not pursued measuring the steady-state rate constants kcat and Km for a panel further. PG1653 and BF1314 were purified to homogeneity and of thioesters (Table 1). It is evident from these results that aliphatic subjected to steady-state kinetic analysis for thioesterase and acyl-CoA thioesters (C2–C14) are not viable physiological sub- phosphatase activity, as was the thioesterase domain homolog strates. In contrast, the three classes of aromatic ring-containing BVU2420 from B. vulgatus. thioesters (R–C(@O)SCoA where R = phenyl; R = napthyl; R =

Fig. 2. A: The structure of the hotdog-fold thioesterase ydiI from E. coli (PDB accession # 1VH5) as represented by the dimer (subunits shown in wheat and green). The two catalytic sites are highlighted by the red and blue colored stick representations of the key catalytic residues. B: The structure of the HAD phosphatase BT3352 (PDB accession 3NIW) as represented by the monomer (catalytic domain is colored wheat and the cap domain green). The catalytic residues are shown as individually colored sticks. C: Depiction of the threading model (calculated using Phyre [55]) of BF1314 based on the experimental coordinates of BT3352 (PDB 3NIW; 47% sequence identity) depicting the active site between the catalytic Rossmann-fold core domain (green) and substrate specificity cap domain (wheat) with conserved catalytic and supporting residues shown as sticks and Mg2+ cofactor as a magenta sphere. Residue numbering for BF1314 with the corresponding residue for BT3352 is in parenthesis. M. Wang et al. / FEBS Letters 587 (2013) 2851–2859 2855

CH2-phenyl) were found to be active substrates. Notably, BVU2420, clusters (Fig. 1B) provides the biological context for the thioester- BF1314 and PG1653 displayed a clear preference for the napthyl ase-catalyzed 1,4-dihydroxynapthoyl-CoA hydrolysis. ring-containing thioester substrate. The respective kcat/Km values (1 106 M1 s1) measured for the 1,4-dihydroxynapthoyl-CoA 3.3. The HAD domain in the fusion protein does not act as a thioester are sufficiently high as to suggest that it is the physiolog- phosphatase ical substrate. Furthermore, the observation that the 1,4-dihydrox- ynapthoyl-CoA kcat/Km values are roughly two orders of magnitude A BLAST search revealed that the closest homologs to the fusion -CoA suggests specialization, based on ring hydroxyl group recog- protein HAD domains belong to the cap type C2B structural class nition, for targeting 1,4-dihydroxynapthoyl-CoA. The presence of (InterPro IPR006379) of the HAD enzyme superfamily and within the encoding genes in the respective menaquinone pathway gene this class, the Cof family (InterPro IPR000150). Members of the

Fig. 3. The sequence logo generated from the alignment of the fusion proteins using the server at http://weblogo.berkeley.edu/logo.cgi. Catalytic residues marked with red asterisks. 2856 M. Wang et al. / FEBS Letters 587 (2013) 2851–2859

C2B class, structurally represented by B. thetaiotaomicron BT3352 with Phe in PG1653 was observed for each of sequences derived (50% sequence identity with the HAD domain of BT4699) (see from three different P. gingivalis strains (Fig. SI3), and thus the Fig. 2B), are typically found to be promiscuous sugar phosphate substitution cannot be attributed to a sequencing error. phosphatases (for examples see [42–44]). Examination of the Because PG1653 lacks the Asp acid/base we questioned its alignment of the sequences of BT3352 with the HAD domains of ability to catalyze phosphate hydrolysis. However, we ex- BF1314 and BT4699 (Fig. SI3), and threading models of the HAD pected to observe BF1314 phosphohydrolase activity given that it domains of BF1314 (Fig. 2C) and BT4699 (not shown), indicate that possesses a full set of catalytic residues, and that HAD phosphohy- the catalytic sites of the BF1314 and BT4699 HAD domains are in- drolases display some degree of substrate promiscuity when tact. Specifically, using the numbering for BF1314 the Asp9 nucle- tested against a panel of phosphate ester and phosphate anhy- ophile, Asp11 acid/base, Mg2+ cofactor-binding Asp200 and Asp224 dride metabolites [48]. Our initial hunch, that the HAD domain and the substrate phosphate-binding residues Lys197 and Ser43 of the fusion protein might function as a pyrophosphatase (vide are present as is the Arg46 which positions the Asp11 acid/base supra), was pursued by testing BF1314 and PG1653-catalyzed residue. These residues are strictly conserved among the fusion hydrolysis of inorganic pyrophosphate; however, no activity was proteins (see sequence logo in Fig. 3). However, a notable excep- detected. A second possibility, that these enzymes support the tion is the substitution of a Phe11 for Asp11 acid/base observed lower menaquinone pathway by catalyzing the hydrolysis of for the PG1653 (see sequence alignment in Fig. SI3). Generally, farnesyl-pyrophosphate (possible regulatory function) and/or amino-acid replacement of the Asp acid/base residue in HAD farnesyl-monophosphate (possible housekeeping function), was superfamily phosphatases ablates catalysis of organophosphate also dismissed because neither compound is dephosphorylated hydrolysis (for examples see [42,44–47]). The replacement of Asp by BF1314 or PG1653.

Table 1 Steady-state kinetic constants for PG1653, BF1314 and BVU2420-catalyzed thioester hydrolysis measured at pH 7.5 and 25 °C. See Section 2 for details.

1 1 1 Substrate Enzyme kcat (s ) km (lM) Km/kcat (M s ) Butyryl-CoA PG1653 <104 –– BF1314 <104 –– BVU2420 (3.5 ± 0.1) 102 (6.9 ± 0.2) 102 5.1 101 Isobutyryl-CoA PG1653 <104 –– BF1314 <104 –– BVU2420 (2.2 ± 0.1) 102 (6.1 ± 0.4) 102 3.6 101 Hexanoyl-CoA PG1653 <104 –– BF1314 <104 –– BVU2420 (4.2 ± 0.2) 102 (3.3 ± 0.3) 103 1.3 101 Octanoyl-CoA PG1653 <104 –– BF1314 <104 –– BVU2420 (3.2 ± 0.3) 103 (1.7 ± 0.1) 104 1.9 101 Decanoyl-CoA PG1653 <104 –– BF1314 <104 –– BVU2420 <104 –– Myristoryl-CoA PG1653 <104 –– BF1314 <104 –– BVU2420 <104 –– 1,4-Dihydroxynapthoyl-CoA PG1653 (1.0 ± 0.1) 101 4.0 ± 0.3 2.5 106 BF1314 (1.2 ± 0.1) 101 7.0 ± 0.5 1.7 106 BVU2420 (1.3 ± 0.1) 101 (1.5 ± 0.1) 101 8.7 105 Napthoyl-CoA PG1653 6.9 ± 0.2 (1.7 ± 0.2) 102 4.0 104 BF1314 6.4 ± 0.3 (2.3 ± 0.4) 102 2.9 104 BVU2420 4.6 ± 0.2 (1.8 ± 0.2) 102 2.5 104 4-Hydroxybenzoyl-CoA PG1653 4.6 ± 0.1 (8.0 ± 0.6) 102 5.8 103 BF1314 1.1 ± 0.1 (2.3 ± 0.2) 102 4.7 103 BVU2420 1.2 ± 0.1 (3.3 ± 0.2) 101 3.6 104 Benzoyl-CoA PG1653 (3.0 ± 0.1) 102 (5.8 ± 0.5) 101 5.2 102 BF1314 (2.4 ± 0.2) 102 (8.4 ± 0.8) 101 2.8 102 BVU2420 (2.8 ± 0.1) 102 (3.1 ± 0.4) 101 9.0 102 3-Hydroxyphenylacetyl-CoA PG1653 (3.1 ± 0.2) 101 (1.7 ± 0.3) 102 2.0 103 BF1314 (1.9 ± 0.1) 101 (4.6 ± 0.4) 101 4.1 103 BVU2420 (3.4 ± 0.1) 102 (8.1 ± 0.6) 102 4.2 101

Table 2

Steady-state kinetic constants for BF1314 and engineered BF1314 HAD domain-catalyzed hydrolysis measured in the presence of 1 mM MgCl2 at pH 7.5 and 25 °C. See Section 2 for details.

Substrate BF1314 BF1314-HAD Domain

1 1 1 1 1 1 kcat (min ) Km (mM) kcat/Km (M s ) kcat (min ) Km (mM) kcat/Km (M s ) p-Nitrophenylphosphate <1 103 –– <1 103 –– Carbamoylphosphate 1.3 ± 0.1 0.44 ± 0.03 54 2.5 ± 0.02 0.97 ± 0.08 42 Imidodiphosphate 12.6 ± 0.6 5.7 ± 0.5 37 23 ± 1 7.4 ± 0.5 51 Acetylphosphate 0.48 ± 0.02 0.51 ± 0.04 16 1.2 ± 0.1 0.87 ± 0.06 23 3-Glycerophosphate <1 103 – – 0.66 ± 0.06 0.063 ± 0.002 170 3-Glycerophosphatea – – – 6.0 ± 0.6 0.087 ± 0.007 1100

a Measured at pH 5.5. M. Wang et al. / FEBS Letters 587 (2013) 2851–2859 2857

To continue the search for a substrate, BF1314 and PG1653 dent of the HAD domain. Accordingly, the BF1314 gene was engi- (wild-type and the Phe11Asp variant) were tested against of a vari- neered to express the thioesterase domain as a standalone ety of phosphate esters and phosphate anhydrides. Compounds protein (residues 274–410). The engineered thioesterase domain that showed no detectable substrate activity included adenosine- protein proved to be soluble and stable during purification. Yet, 50-triphosphate, adenosine-50-diphosphate, adenosine-50-mono- when tested for 1,4-dihydroxynapthoyl-CoA thioesterase activity 0 4 1 phosphate, uridine-5 -monophosphate, erythrose-4-phosphate, it was catalytically inactive (viz. kcat <1 10 s ). Thus, the thio- fructose-1-phosphate, glucose-6-phosphate, a-glucose-1-phos- domain requires the HAD domain for catalytic function. phate, fructose-1,6-bisphosphate, thiamine phosphate, threonine Our next step was to explore the structural basis for this phosphate, serine phosphate, ribose-5-phosphate, arabinose-5- requirement. phosphate, 2-keto-3-deoxy-6-phosphogluconate, 6-phosphoglu- conic acid, glyceraldehyde-3-phosphate, 3-phosphoglycerate, 3.5. Effect of HAD and thioesterase domains on oligomerization state 2-phosphoglycolate, phosphoenolpyruvate and p-nitrophenyl- phosphate. BF1314, but not PG1653, displayed a very low level of To gain insight into the interaction between the HAD and thio- activity toward three substrates that are relatively susceptible to esterase domains within the fusion protein, the ability of the engi- spontaneous hydrolysis in solution, namely carbamoylphosphate, neered BF1314 HAD domain and the BF1314 thioesterase domain imidodiphosphate and acetylphosphate (respective kcat/Km values to form a stable complex was tested by gel filtration chromatogra- of 54 M1 s1,37M1 s1 and 16 M1 s1; Table 2). phy. In addition, the ability of the engineered BF1314 HAD domain The lack of phosphatase activity observed for BF1314 and (residues 1–290) to restore catalytic activity to the engineered PG1653 stands in contrast to the phosphatase activity previously BF1314 thioesterase domain was evaluated by steady-state kinetic reported for the homologous standalone HAD phosphatase analysis. The 1:1 mixture of the two domains failed to catalyze the BT3352, which was used to generate the threading model of hydrolysis of 1,4-dihydroxynapthoyl-CoA (SI Table 1). In addition, BF1314 shown in Fig. 2C. BT3352 is highly efficient in catalysis of the same mixture cleanly separated into a HAD domain fraction hydrolytic dephosphorylation of C4 and C5 sugar phosphates, with and a thioesterase domain fraction on a gel-filtration column the highest activity observed towards erythrose-4 phosphate (kcat/ (Fig. SI5). These findings indicate that the two domains do not form 5 1 1 Km = 1.3 10 M s ) [49]. Why then is BF1314 inactive? Is the a stable complex. However, based on their elution profiles we association of the HAD domain with the thioesterase domain in learned that the engineered HAD domain exists as a tetramer in some way inhibiting the phosphatase activity? solution (monomer mass: 30 kDa; observed mass: 130 kDa) yet To pursue these questions we determined if the HAD domain the engineered thioesterase domain is a monomer in solution (cal- phosphatase activity could be restored by removal of the thioester- culated monomer mass: 17 kDa; observed mass: 15 kDa). Be- ase domain. cause the active site of the hotdog-fold thioesterase is formed at the interface of two associated subunits (Fig. 2A) the minimal bio- 3.4. Catalytic properties of the HAD domain and thioesterase domains logical unit (i.e., catalytically active species) is a dimer. Most typi- of BF1314 cally the hotdog-fold thioesterase exists as a tetramer (viz. dimer of dimer) [51,52], as is also found to be the case with the tetrameric The BF1314 gene was engineered to express the N-terminal (native) BF1314 fusion protein (monomer mass: 47 kDa; observed HAD domain (residues 1–290) as a standalone protein. The HAD mass: 190 kDa determined by gel filtration chromatography). Gi- domain C-terminus region was identified from the alignment of ven that the engineered BF1314 HAD domain is also a tetramer we the sequences of the Bacteroides fusion proteins with the corre- surmised that the HAD domain serves as an oligomerization do- sponding standalone HAD phosphatase homologs. The engineered main. A priori, by reducing the entropy loss for thioesterase do- BF1314 HAD domain proved to be soluble and stable. However, main association, the strong association of the HAD domains when screened against the substrate library, the high level of cat- could augment weak association of the thioesterase domains. alytic activity observed for BT3352 was not found (Table 2). Never- And, by taking on this role, the HAD domain might ultimately lose theless, a new activity was observed, namely the hydrolysis of the catalytic function. The absence of an evolutionary selection pres- phosphate ester, glycerophosphate. At pH 7.5 and 25 °C catalysis sure for catalytic function is indicated by the replacement of an 1 1 occurs with low efficiency (kcat/Km = 170 M s ), yet, when mea- essential catalytic residue, namely the Asp acid/base in PG1653 sured at pH 5.5, the efficiency is significantly higher (kcat/ (vide supra). 1 1 Km = 1100 M s ) (pH/rate profile shown in Fig. SI4). Thus, the catalytic site is functional, but only with this one phosphate ester. 3.6. Bioinformatic analysis of sequence divergence, gene duplication, Based on these findings we concluded that the engineered and biological range BF1314 HAD domain is in fact a phosphatase, albeit an inefficient one. Given the importance of cap domain-catalytic domain associ- The B. fragilis BF1314 sequence was used as query in a BLAST ation for catalytic turnover, and dissociation for substrate binding search of protein sequences deposited in the NCBI databank. The and product release [50], a likely scenario is that coordinated do- biological range of the fusion protein is limited to the phyla Bacter- main movement does not occur in the fusion proteins. At the oidetes and more specifically, to certain species of Bacteroides (cac- two extremes, the cap and catalytic domains might remain in a cae, cellulosilyticus, clarus, coprosuis, eggerthii, fragilis, finegoldii, partially closed conformation thereby limiting access of the sub- fluxus, helcogenes, intestinalis, ovatus, stercoris, uniformis, faecis, nor- strate to the catalytic site or alternatively, the two domains might dii, oleiciplenus, salyersiae, thetaiotaomicron and xylanisolvens) remain in an open conformation thereby preventing active site (80% sequence identity), Parabacteroides (merdae, johnsonii, gold- desolvation. Because the removal of the thioesterase domain re- steinii, distasonis and sp. D13) (75% sequence identity), Paraprev- stores a modest level of catalytic activity, albeit restricted to a otella (clara, xylaniphila) (73% sequence identity), Tannerella small substrate, we assume that it is the attached thioesterase do- (forsythia) (73% sequence identity), Capnocytophaga (sp oral toxin main that is perturbing cap function in the native enzyme. 329) (73% sequence identity) and Porphyromonas (gingivalis, uenonis, If the HAD domain of the fusion protein does not function as a and asaccharolytica)(40% sequence identity). Except for P. gingivalis, catalyst, then what function might preserve its evolutionary part- T. forsythia and Capnocytophaga which are members of the red nership with the thioesterase domain? To address this question, complex of periodontal pathogens, these fusion protein-producing the thioesterase domain was tested for catalytic activity indepen- organisms are found in the upper or lower mammalian gastrointestinal 2858 M. Wang et al. / FEBS Letters 587 (2013) 2851–2859 tract. All are facultative anaerobes, which in the absence of oxygen uti- disruption of conformational coupling between the cap and core lize menaquinone as the electron acceptor during energy metabolism. domains. Some of the sequenced genomes of species that possess the fu- We posit that following the fusion of the two functionally unre- sion protein have been fully annotated, enabling inspection of gene lated genes, the HAD and thioesterase domains co-evolved result- neighborhoods. In the case of the human gut bacteria B. fragilis and ing in the loss of phosphohydrolase catalytic function in the HAD thetaiotaomicron and P. distasonis and uenonis the fusion protein domain and an acquired dependence of the thioesterase domain gene is adjacent to the cluster of menb-f genes. In contrast, the on its HAD domain partner for oligomerization. In some Bacteroi- location of the fusion-protein genes is distant from the menb-f gene detes species, a homologous yet independent thioesterase cata- cluster in the genome of the pig gut bacterium B. helcogenes. The lyzes 1,4-dihydroxynapthoyl-CoA hydrolysis in the menaquinone distant location of the fusion protein gene from the menb-f gene pathway, thus indicating that the fusion protein is the vestige of cluster is also observed for the human bacteria Porphyromonas gin- a random mutation that linked the two encoding genes. Gene fu- givalis, asaccharolytica and uenonis as well as in Paraprevotella sions can confer many advantages. These include co-regulation of xylaniphila and Parabacteroides sp. D13. gene expression for enzymes in a pathway, the insurance of the We anticipated that Bacteroidetes/Chlorobi species, which do stoichiometry of the gene products [53] or the regulation of subcel- not possess the fusion protein, yet do contain the genes encoding lular localization [54]. However, in the case of the fusion of the the menaquinone pathway enzymes would necessarily have a HAD domain and thioesterase domain a very different outcome re- standalone thioesterase to function in place of the fusion protein. sults, in which an independent thioesterase (exemplified by Indeed, thioesterases sharing 49–56% sequence identity with the BVU2420) becomes structurally dependent on a HAD phosphatase BF1314 fusion protein thioesterase domain were found to be en- (exemplified by BT3352) for oligomerization. The persistence of coded by the genomes of Bacteroides species coprocola, coprophilus, the fusion protein is likely due to the requirement for catalysis of dorei, plebeius, salantitronis and vulgatus as well by the genomes of 1,4-dihydroxynapthoyl-CoA hydrolysis coupled with the absence the distant Bacteriodetes/Chlorobi species Cytophaga hutchinsonii, of a second copy of the 1,4-dihydroxynapthoyl-CoA thioesterase Dokdonia donghaensis, Bizonia argentinensis, Flavobacteriales bacterium gene. Unlike the HAD phosphatase we found no evidence for dupli- ALC-1, Gramella forsetii, Kordia algicida, Lacinutrix sp. 5H-3-7-4, cation of the thioesterase gene. Leeuwenhoekiella blandensis, Muricauda ruestringensis, Paludibacter propionicigenes WB4, Zobellia galactanivorans, Zunongwangia pro- Acknowledgement funda and all represented Prevotella species. These latter thioester- ases share 34–48% sequence identity with the BF1314 fusion This work was supported by National Institutes of Health Grant protein thioesterase domain. Notably, the thioesterase-encoding RO1 GM28688 (to D.D.-M.). genes are typically co-located with menaquinone pathway genes, consistent with the participation of the thioesterases in the respec- Appendix A. Supplementary data tive pathways. We characterized one such thioesterase from B. vulgatus, BVU2420, to show that it catalyzes the hydrolysis of Supplementary data associated with this article can be found, 1,4-dihydroxynapthoyl-CoA with high efficiency (kcat/Km =1 in the online version, at http://dx.doi.org/10.1016/j.febslet. 6 1 1 10 M s ; Table 2) (sequence alignment is shown in Fig. SI2) 2013.07.009. and is thus an ortholog. A majority of the sequenced genomes that encode the stand- References alone thioesterases also encode close homologs to the BF1314 fu- sion protein HAD domain (>40% sequence identity). Intriguingly, [1] Ferland, G. (2012) Vitamin K, an emerging nutrient in brain function. a significant number of the bacteria that possess the fusion protein BioFactors 38, 151–157. [2] Shearer, M.J. and Newman, P. (2008) Metabolism and cell biology of vitamin K. also possess a standalone HAD phosphatase with >45% sequence Thromb. Haemost. 100, 530–547. identity (Bacteroides caccae, clarus, fragilis, finegoldii, helcogenes, [3] Shearer, M.J., Fu, X. and Booth, S.L. (2012) Vitamin K nutrition metabolism, and thetaiotaomicron and xylanisolvens, Parabacteroides sp D13, Parap- requirements: current concepts and future research. Adv. Nutr: Int. Rev. J. 3, 182–195. [4] Bugel, S. (2008) Vitamin K and bone health in adult humans in vitamin K, revotella xylaniphila and Capnocytophaga sp oral toxin 329). In con- Elsevier Academic Press Inc, San Diego, pp. 393–416, Elsevier Academic Press trast, we found no examples of bacteria that possess a standalone Inc, San Diego. thioesterase homolog in addition to the HAD-thioesterase fusion [5] Shea, M.K. and Holden, R.M. (2012) Vitamin K status and vascular calcification: evidence from observational and clinical studies. Adv. Nutr: Int. Rev. J. 3, 158– protein. 165. [6] Ferland, G. (2012) Vitamin K and the nervous system: an overview of its actions. Adv. Nutr. 3, 204–212. 3.7. Conclusions [7] Nowicka, B. and Kruk, J. (2010) Occurrence, and function of isoprenoid quinones. Biochim. Biophys. Acta (BBA) – Bioenerg. 1797, 1587–1605. [8] Sharma, P., Teixeira de Mattos, M.J., Hellingwerf, K.J. and Bekker, M. (2012) On The HAD-thioesterase fusion proteins, examined here (PG1653 the function of the various quinone species in Escherichia coli. FEBS J. 279, and BF1314) show highly efficient thioesterase activity toward 3364–3373. [9] Bentley, R. and Meganathan, R. (1982) Biosynthesis of vitamin K the substrate 1,4-dihydroxynapthoyl-CoA. Together with this (menaquinone) in bacteria. Microbiol. Rev. 46, 241–280. in vitro activity, gene neighborhood analysis places these enzymes [10] Kurosu, M. and Begari, E. (2010) Vitamin K2 in electron transport system: are firmly in the menaquinone pathway. However, the HAD domain of enzymes involved in vitamin K2 biosynthesis promising drug targets? the fusion protein is devoid of catalytic activity and most likely Molecules 15, 1531–1553. [11] Li, X., Liu, N., Zhang, H., Knudson, S.E., Li, H.J., Lai, C.T., Simmerling, C., Slayden, takes on a scaffolding function. BT3352, a homologue with high se- R.A. and Tonge, P.J. (2011) CoA adducts of 4-oxo-4-phenylbut-2-enoates: quence identity (35.8–47.1%) to the HAD domain of the fusion pro- inhibitors of MenB from the M. tuberculosis menaquinone biosynthesis tein, efficiently catalyzes the hydrolysis of small sugar phosphates pathway. ACS Med. Chem. Lett. 2, 818–823. [12] Lu, X., Zhou, R., Sharma, I., Li, X., Kumar, G., Swaminathan, S., Tonge, P.J. and [49]. Because a comparative model shows that the active sites of Tan, D.S. (2012) Stable analogues of OSB-AMP: potent inhibitors of MenE, the the HAD domain of the fusion proteins are intact (with the excep- o-succinylbenzoate-CoA synthetase from bacterial menaquinone biosynthesis. tion of the catalytic acid/base in PG1653), the lack of phosphohy- ChemBioChem 13, 129–136. [13] Lu, X., Zhang, H., Tonge, P.J. and Tan, D.S. (2008) Mechanism-based inhibitors drolase activity can be attributed to changes over the entire of MenE, an acyl-CoA synthetase involved in bacterial menaquinone scaffold that effect catalysis and perhaps more specifically, to biosynthesis. Bioorg. Med. Chem. Lett. 18, 5963–5966. M. Wang et al. / FEBS Letters 587 (2013) 2851–2859 2859

[14] Debnath, J., Siricilla, S., Wan, B., Crick, D.C., Lenaerts, A.J., Franzblau, S.G. and [35] Michaelis, L. and Menten, M.L. (1913) Die Kinetik der Invertinwirkung. Kurosu, M. (2012) Discovery of selective menaquinone biosynthesis inhibitors Biochem. Z. 49, 333–369. against Mycobacterium tuberculosis. J. Med. Chem. 55, 3739–3755. [36] Johnson, K.A. and Goody, R.S. (2011) The original michaelis constant: [15] Heide, L., Arendt, S. and Leistner, E. (1982) Enzymatic synthesis, translation of the 1913 Michaelis–Menten Paper. Biochemistry 50, 8264– characterization, and metabolism of the coenzyme A ester of o- 8269. succinylbenzoic acid, an intermediate in menaquinone (vitamin K2) [37] Zhuang, Z., Song, F., Zhao, H., Li, L., Cao, J., Eisenstein, E., Herzberg, O. and biosynthesis. J. Biol. Chem. 257, 7396–7400. Dunaway-Mariano, D. (2008) Divergence of function in the hot dog fold [16] Meganathan, R. and Bentley, R. (1983) Thiamine pyrophosphate requirement enzyme superfamily: the bacterial thioesterase YciA. Biochemistry 47, 2789– for o-succinylbenzoic acid synthesis in Escherichia coli and evidence for an 2796. intermediate. J. Bacteriol. 153, 739–746. [38] Song, F., Zhuang, Z., Finci, L., Dunaway-Mariano, D., Kniewel, R., Buglino, J.A., [17] Palaniappan, C., Sharma, V., Hudspeth, M.E. and Meganathan, R. (1992) Solorzano, V., Wu, J. and Lima, C.D. (2006) Structure, function, and mechanism Menaquinone (vitamin K2) biosynthesis: evidence that the Escherichia coli of the phenylacetate pathway hot dog-fold thioesterase PaaI. J. Biol. Chem. menD gene encodes both 2-succinyl-6-hydroxy-2,4-cyclohexadiene-1- 281, 11028–11038. synthase and alpha-ketoglutarate decarboxylase activities. J. [39] Luo, L., Taylor, K.L., Xiang, H., Wei, Y., Zhang, W. and Dunaway-Mariano, D. Bacteriol. 174, 8111–8118. (2001) Role of active site binding interactions in 4-chlorobenzoyl-coenzyme A [18] Sharma, V., Meganathan, R. and Hudspeth, M.E. (1993) Menaquinone (vitamin dehalogenase catalysis. Biochemistry 40, 15684–15692. K2) biosynthesis: cloning, nucleotide sequence, and expression of the menC [40] Merkel, S.M., Eberhard, A.E., Gibson, J. and Harwood, C.S. (1989) Involvement gene from Escherichia coli. J. Bacteriol. 175, 4917–4921. of coenzyme A thioesters in anaerobic metabolism of 4-hydroxybenzoate by [19] Daruwala, R., Kwon, O., Meganathan, R. and Hudspeth, M.E. (1996) A new Rhodopseudomonas palustris. J. Bacteriol. 171, 1–7. isochorismate synthase specifically involved in menaquinone (vitamin K2) [41] Huang, H., Patskovsky, Y., Toro, R., Farelli, J.D., Pandya, C., Almo, S.C., Allen, K.N. biosynthesis encoded by the menF gene. FEMS Microbiol. Lett. 140, 159–163. and Dunaway-Mariano, D. (2011) Divergence of structure and function in the [20] Meganathan, R. (2001) Biosynthesis of menaquinone (vitamin K2) and haloacid dehalogenase enzyme superfamily: Bacteroides thetaiotaomicron ubiquinone (coenzyme Q): a perspective on enzymatic mechanisms. Vitam. BT2127 is an inorganic pyrophosphatase. Biochemistry 50, 8937–8949. Horm. 61, 173–218. [42] Lu, Z., Dunaway-Mariano, D. and Allen, K.N. (2011) The X-ray crystallographic [21] Klenchin, V.A., Taylor Ringia, E.A., Gerlt, J.A. and Rayment, I. (2003) Evolution structure and specificity profile of HAD superfamily phosphohydrolase of enzymatic activity in the enolase superfamily: structural and mutagenic BT1666: comparison of paralogous functions in B. thetaiotaomicron. studies of the mechanism of the reaction catalyzed by o-succinylbenzoate Proteins: Struct., Funct., Bioinf. 79, 3099–3107. synthase from Escherichia coli. Biochemistry 42, 14427–14433. [43] Lu, Z., Dunaway-Mariano, D. and Allen, K.N. (2005) HAD superfamily [22] Jiang, M., Cao, Y., Guo, Z.-F., Chen, M., Chen, X. and Guo, Z. (2007) phosphotransferase substrate diversification: structure and function analysis Menaquinone biosynthesis in Escherichia coli: identification of 2-succinyl-5- of HAD subclass IIB sugar phosphatase BT4131. Biochemistry 44, 8684–8696. enolpyruvyl-6-hydroxy-3-cyclohexene-1-carboxylate as a novel intermediate [44] Roberts, A., Lee, S.-Y., McCullagh, E., Silversmith, R.E. and Wemmer, D.E. (2005) and re-evaluation of MenD activity. Biochemistry 46, 10979–10989. Ybiv from Escherichia coli K12 is a HAD phosphatase. Proteins: Struct., Funct., [23] Jiang, M., Chen, X., Guo, Z.-F., Cao, Y., Chen, M. and Guo, Z. (2008) Identification Bioinf. 58, 790–801. and characterization of (1R,6R)-2-succinyl-6-hydroxy-2,4-cyclohexadiene-1- [45] Lu, Z., Dunaway-Mariano, D. and Allen, K.N. (2008) The catalytic scaffold of the carboxylate synthase in the menaquinone biosynthesis of Escherichia coli. haloalkanoic acid dehalogenase enzyme superfamily acts as a mold for the Biochemistry 47, 3426–3434. trigonal bipyramidal transition state. Proc. Natl. Acad. Sci. USA 105, 5687– [24] Li, H.-J., Li, X., Liu, N., Zhang, H., Truglio, J.J., Mishra, S., Kisker, C., Garcia-Diaz, 5692. M. and Tonge, P.J. (2011) Mechanism of the intramolecular Claisen [46] Nguyen, H.H., Wang, L., Huang, H., Peisach, E., Dunaway-Mariano, D. and Allen, condensation reaction catalyzed by MenB, a crotonase superfamily member. K.N. (2010) Structural determinants of substrate recognition in the HAD Biochemistry 50, 9532–9544. superfamily member D-glycero-D-manno-heptose-1,7-bisphosphate [25] Johnston, J.M., Jiang, M., Guo, Z. and Baker, E.N. (2013) Crystal sructures of phosphatase (GmhB). Biochemistry 49, 1082–1092. E. coli native MenH and two active site mutants. PLoS One 8, e61325. [47] Walldén, K., Rinaldo-Matthis, A., Ruzzenente, B., Rampazzo, C., Bianchi, V. and [26] Meganathan, R., Bentley, R. and Taber, H. (1981) Identification of Bacillus Nordlund, P. (2007) Crystal structures of human and murine subtilis men mutants which lack O-succinylbenzoyl-coenzyme A synthetase deoxyribonucleotidases: insights into recognition of substrates and and dihydroxynaphthoate synthase. J. Bacteriol. 145, 328–332. nucleotide analogues. Biochemistry 46, 13809–13818. [27] Rowland, B., Hill, K., Miller, P., Driscoll, J. and Taber, H. (1995) Structural [48] Kuznetsova, E., Proudfoot, M., Gonzalez, C.F., Brown, G., Omelchenko, M.V., organization of a Bacillus subtilis operon encoding menaquinone biosynthetic Borozan, I., Carmel, L., Wolf, Y.I., Mori, H., Savchenko, A.V., Arrowsmith, C.H., enzymes. Gene 167, 105–109. Koonin, E.V., Edwards, A.M. and Yakunin, A.F. (2006) Genome-wide analysis of [28] Dairi, T. (2012) in: Chapter Six Menaquinone Biosyntheses in Microorganisms substrate specificities of the Escherichia coli haloacid dehalogenase-like in Methods in Enzymology (David, A.H., Ed.), pp. 107–122, Academic Press. phosphatase family. J. Biol. Chem. 281, 36149–36161. [29] Dairi, T. (2009) An alternative menaquinone biosynthetic pathway operating [49] Wang, M. (2011) Evolution of structure and function among hotdog-fold in microorganisms: an attractive target for drug discovery to pathogenic thioesterase and HAD family phosphates, University of New Mexico. Helicobacter and Chlamydia strains. J. Antibiot. 62, 347–352. [50] Allen, K.N. and Dunaway-Mariano, D. (2009) Markers of fitness in a successful [30] Widhalm, J.R., van Oostende, C., Furt, F. and Basset, G.J. (2009) A dedicated enzyme superfamily. Curr. Opin. Struct. Biol. 19, 658–665. thioesterase of the Hotdog-fold family is required for the biosynthesis of the [51] Thoden, J.B., Holden, H.M., Zhuang, Z. and Dunaway-Mariano, D. (2002) X-ray naphthoquinone ring of vitamin K1. Proc. Natl. Acad. Sci. USA 106, 5599–5603. crystallographic analyses of inhibitor and substrate complexes of wild-type [31] Latham, J.A. (2012) Structure to function: case studies of hotdog-fold and mutant 4-hydroxybenzoyl-CoA thioesterase. J. Biol. Chem. 277, 27468– superfamily thioesterases from Escherichia coli, University of New Mexico. 27476. [32] Chen, M., Ma, X., Chen, X., Jiang, M., Song, H. and Guo, Z. (2013) Identification [52] Thoden, J.B., Zhuang, Z., Dunaway-Mariano, D. and Holden, H.M. (2003) The of a hotdog fold thioesterase involved in the biosynthesis of menaquinone in structure of 4-hydroxybenzoyl-CoA thioesterase from Arthrobacter sp. strain Escherichia coli. J. Bacteriol. 195, 2768–2775. SU. J. Biol. Chem. 278, 43709–43716. [33] Suvarna, K., Stevenson, D., Meganathan, R. and Hudspeth, M.E.S. (1998) [53] Mccarthy, A.D. and Hardie, D.G. (1984) Fatty-acid synthase – an example of Menaquinone (vitamin K2) biosynthesis: localization and characterization of protein evolution by gene fusion. Trends Biochem. Sci. 9, 60–63. the menA gene from Escherichia coli. J. Bacteriol. 180, 2782–2787. [54] Long, M. (2000) A new function evolved from gene fusion. Genome Res. 10, [34] Koike-Takeshita, A., Koyama, T. and Ogura, K. (1997) Identification of a novel 1655–1657. gene cluster participating in menaquinone (vitamin K2) biosynthesis. Cloning [55] Kelley, L.A. and Sternberg, M.J. (2009) Protein structure prediction on the and sequence determination of the 2-heptaprenyl-1,4-naphthoquinone Web: a case study using the Phyre server. Nat. Protoc. 4, 363–371. methyltransferase gene of Bacillus stearothermophilus. J. Biol. Chem. 272, 12380–12383.