Divergence of substrate specificity and function in the Escherichia coli hotdog-fold thioesterase paralogs YdiI and YbdB.

The work described in this paper, and its companion paper (Wu, R., Latham, J. A., Chen, D., Farelli, J., Zhao, H., Matthews, K. Allen, K. N., and Dunaway-Mariano, D. (2014) Structure and Catalysis in the Escherichia coli Hotdog-fold Thioesterase Paralogs YdiI and YbdB. Biochemistry, DOI: 10.1021/bi500334v), focuses on the evolution of a pair of paralogous hotdog-fold superfamily thioesterases of E. coli, YbdB and YdiI, which share a high level of sequence identity but perform different biological functions (viz., proofreader of 2,3-dihydroxybenzoyl-holoEntB in the enterobactin biosynthetic pathway and catalyst of the 1,4-dihydoxynapthoyl-CoA hydrolysis step in the menaquinone biosynthetic pathway, respectively). In vitro substrate activity screening of a library of thioester metabolites showed that YbdB displays high activity with benzoyl-holoEntB and benzoyl-CoA substrates, marginal activity with acyl-CoA thioesters, and no activity with 1,4-dihydoxynapthoyl-CoA. YdiI, on the other hand, showed a high level of activity with its physiological substrate, significant activity toward a wide range of acyl-CoA thioesters, and minimal activity toward benzoyl-holoEntB. These results were interpreted as evidence for substrate promiscuity that facilitates YbdB and YdiI evolvability, and divergence in substrate preference, which correlates with their assumed biological function. YdiI support of the menaquinone biosynthetic pathway was confirmed by demonstrating reduced anaerobic growth of the E. coli ydiI-knockout mutant (vs wild-type E. coli) on glucose in the presence of the electron acceptor fumarate. Bioinformatic analysis revealed that a small biological range exists for YbdB orthologs (i.e., limited to Enterobacteriales) relative to that of YdiI orthologs. The divergence in YbdB and YdiI substrate specificity detailed in this paper set the stage for their structural analyses reported in the companion paper.

The physiological roles of cellular thioesterases are centered on the catalyzed hydrolysis of thioester metabolites to their corresponding organic acid and thiol components. Thioesters are prevalent in the cell wherein they serve as activated forms of organic acids in a wide range of anabolic and catabolic chemical pathways. Thioesterases act on the thioester precursors,[3] intermediates,[4] or products[5] of such pathways; alternatively, they perform supportive roles as proofreaders,[6] housekeepers,[7,8] or regulators.[9,10]

Thioesterases belong primarily to the hotdog-fold or α,β-hydrolase-fold protein superfamilies.[11,12] The hotdog-fold thioesterases possess a characteristic core structure consisting of a 5-turn α-helix cradled by a curved, 5-stranded antiparallel β-sheet.[13] Dimerization forms an elongated β-sheet and two active sites located at opposite ends of the subunit interface. The topology of the substrate-binding site restricts catalytic activity to thioesters in which the thiol unit is coenzyme A (CoA) or a phosphopantetheine-functionalized acyl carrier protein (holoACP).[14] In contrast, the substrate range covered by the thioesterase family of the α,β-hydrolase-fold protein superfamily also includes cysteine-linked thioesters (e.g., palmitoylated proteins).[15]

The largest family within the hotdog-fold protein superfamily is comprised of thioesterases, principally because of the large demand for thioester hydrolysis in the cell. The prevalence of the hotdog-fold thioesterase family, in particular, suggests high evolvability, meaning that the hotdog-fold scaffold readily takes on new functions. It is well-known that substrate promiscuity is a key factor in the rapid gain of novel biological function (for a recent review of this topic, see ref (16)). The intrinsic promiscuity of a hotdog-fold thioesterase can be revealed through determination of its substrate specificity profile by in vitro substrate activity screening. Herein, we examine the substrate specificities of a pair of hotdog-fold thioesterase family paralogs,aEscherichia coli YbdB and YdiI, to discover that although they are intrinsically promiscuous, they display high catalytic efficiency for, and discrimination between, their respective physiological substrates (viz., aberrant aroyl-holoEntBb thioesters sometimes formed as dead-end adducts during enterobactin biosynthesis[1,6,17] and 1,4-dihydroxynapthoyl-CoA formed as an intermediate of the menaquinone biosynthetic pathway[4,18−20]). In addition, we used bioinformatic methods to identify and map the biological ranges of YbdB and YdiI orthologs and site-directed mutagenesis to evaluate potential sequence markers. In the companion paper,[2] we report on the structure and mechanism E. coli YbdB and YdiI.

Materials and Methods

The restriction enzymes, T4 DNA ligase, oligonucleotide primers, and the competent E. coli BL21(DE3) cells were purchased from Invitrogen. Pfu Turbo and Deep Vent DNA polymerases were purchased from Strategene. The cloning vectors were from Novagen. DNA sequencing was performed by the DNA Sequencing Facility of the University of New Mexico. Acetyl-CoA, benzoyl-CoA, propanoyl-CoA, hexanoyl-CoA, lauroyl-CoA, myristoyl-CoA, palmitoyl-CoA, and oleoyl-CoA were purchased from Sigma. The thioester substrates 4-hydroxybenzyol-CoA, 3-hydroxybenzoyl-CoA, 1,4-dihydroxynapthoyl-CoA, 3-hydroxyphenylacetyl-CoA, and coumaroyl-CoA were synthesized as previously reported.[18,21]E. coli strains JW1676 (ΔydiI::kanr) and BW25113 (wild-type) of the Keio collection were obtained from Yale University.[22] The engineered E. coli strain DK574, carrying the plasmid pJT93 expressing the E. coli AcpS transferase gene, under tac-promoter control, was a kind gift from Dr. John Cronan of the University of Illinois. The holoACP (UniProt accession code P0A6A8) purified from this strain was converted to benzoyl-holoACP by using the chemical procedure reported in ref (23). The molecular mass and purity of the isolated adduct were verified by ES-MS analysis.

Growth Curve Measurements for Wild-Type and ydiI-Knockout E. coli Strains

Aerobic growth curves were carried out in sterile vented flasks (Nalgene) each containing 50 mL of M9 minimal media supplemented with 4% glucose as the sole carbon source, and with or without added kanamycin. An aliquot of a liquid culture of JW1676 (ydiI-knockout strain) or BW25113 (wild-type strain) E. coli cells grown overnight in LB broth was added to the media to an initial A600 ∼ 0.01. Cultures were incubated at 37 °C with orbital shaking at 180 rpm. The culture A600 was determined at 1 h intervals for 14 h. Anaerobic growth curves were measured in a similar fashion using 50 mL of M9 minimal media supplemented with double the amount of phosphate and with 4% glucose plus 4% fumarate. The sterile flasks were capped with sterilized stoppers and purged for 2 min with N2 gas passed through a sterile 0.2 μm in-line filter. Aliquots were removed by syringe, hourly over a 12 h period, for A600 determination.

Cloning, Expression, and Purification of C-terminus His-Tagged YbdB and YdiI

The genes encoding YbdB (UniProt accession code P0A8Y8) and YdiI (UniProt accession code P77781) were cloned by using a PCR-based strategy in which genomic DNA prepared from E. coli strain K12 (substrain W3110) was used as template, commercial oligonucleotides as the primers, and Pfu Turbo as the DNA polymerase. The NdeI and XhoI-treated PCR product was ligated, using T4 DNA ligase, to NdeI/XhoI sites of pET-23a (Novabiochem) to give the plasmids, ybdB-His/pET-23a and ydiI-His/pET-23a. The resulting clones were used to transform competent E. coli BL21 (DE3) cells (Novagen). The cells were grown at 37 °C in LB medium containing 50 μg/mL of ampicillin. The cell culture was induced with 0.4 mM isopropyl-β-d-galactopyranoside (IPTG) once the optical density had reached 0.6 (OD at 600 nm). Following an overnight induction at 20 °C, the cells were harvested by centrifugation at 5000g for 15 min. The cell pellet was suspended in 150 mL of lysis buffer (20 mM Tris·HCl, 10 mM imidazole, 1 mM DTT, and 300 mM NaCl, pH 7.5) and passed through a French pressure cell at 1200 psi before centrifugation at 48 000g for 1 h at 4 °C. The supernatant was loaded onto a Ni-NTA column (2 cm × 11 cm), pre-equilibrated with lysis buffer. The column was washed at a flow rate of 0.5 mL/min with 150 mL of lysis buffer followed by 100 mL of wash buffer (20 mM Tris·HCl, 300 mM NaCl, 50 mM imidazole at pH 7.5). The His6-tagged proteins were eluted with 90 mL of elution buffer (20 mM Tris·HCl, 300 mM NaCl, 250 mM imidazole at pH 7.5). The protein fractions were analyzed by SDS-PAGE before pooling, concentrating, and dialyzing (50 mM Tris·HCl, 100 mM NaCl at pH 7.5). Aliquots of the protein solutions were frozen for storage at −80 °C. Yield: ∼15 mg of His6-YbdB/g wet cell paste and ∼10 mg of His6-YdiI/g wet cell paste.

Preparation of YbdB and YdiI Site-Directed Mutants

Site-directed mutagenesis was carried out using the QuikChange PCR strategy (Stratgene) and the ydiI-His/pET-23a or ybdB-His/pET-23a plasmid as template with commercial primers and Pfu Turbo as the polymerase. The sequence of the mutated gene was confirmed by DNA sequencing. The recombinant mutant plasmids were used to transform competent E. coli BL21 Star (DE3) cells. The His6-tagged mutant proteins YbdB M68V and YdiI V68M and F50A were purified to >90% homogeneity (as determined by SDS-PAGE analysis) as described above in a yield of ∼10–25 mg protein/g wet cell paste.

Determination of Steady-State Kinetic Constants

Thioesterase activity was measured using a 5,5′-dithio-bis(2-nitrobenzoic acid) (DTNB) coupled assay. Reactions were monitored at 412 nm (Δε = 13.6 mM–1·cm–1) using a Beckman 640U spectrometer. Reactions were carried out 25 °C with 0.5 mL solutions containing 50 mM K+HEPES (pH 7.5), 1 mM DNTB, an optimal concentration of thioesterase, and varying concentrations of the substrate (0.5Km to 5Km). The catalyzed hydrolysis of 4-HB-CoA in 50 mM K+HEPES (pH 7.5) was directly monitored at 300 nm (Δε = 11.8 mM–1·cm–1). The initial velocity data, measured as a function of substrate concentration, were analyzed using Enzyme Kinetics v 1.4 and eq 1:where v is initial velocity, Vmax is maximum velocity, [S] is substrate concentration, and Km is the Michaelis constant. The kcat was calculated from Vmax/[E] where [E] is the total enzyme concentration as determined by the Bradford method.[24]

Determination of YbdB and YdiI pH-Rate Profiles for Catalyzed 4-HB-CoA Thioester Hydrolysis

The initial velocities of YbdB and YdiI-catalyzed hydrolysis of 4-HB-CoA, at varying concentration (in the range of 0.5Km to 5Km), were measured at 25 °C (vide supra). The pH values of the reaction solutions were maintained using 50 mM MES, HEPES, TAPS, or CAPSO. Control reactions, in which the enzyme was preincubated in the reaction buffer and then assayed at pH 7.5, were carried out to detect possible inactivation at the pH extremes.

Bioinformatic Analysis

BLAST searches of the sequenced genomes deposited in NCBI were carried out using the Genomics Groups BLAST server (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi). The E. coli K12 YbdB and YdiI sequences were used as queries, as were the sequences of the E. coli enterobactin and menaquinone biosynthetic pathway enzymes. Protein sequence alignments were generated using COBALT (www.ncbi.nlm.nih.gov/tools/cobalt/)[25] and displayed in ESPrit3 (http://espript.ibcp.fr/ESPript/ESPript/index.php).[26] Pathway gene context was determined by browsing gene neighborhoods in PATRIC (http://patric.vbi.vt.edu/) and EnsemblBacteria (http://bacteria.ensembl.org/index.html). Homologs of the E. coli YbdB and YdiI were identified for each bacterial phylum by carrying out BLAST searches of deposited genomes, one genus at a time. E. coli YdiI and YbdB homologs having >35% sequence identity for >80% sequence coverage when aligned pairwise with the YdiI sequence are listed in Table SI1 of the Supporting Information. The genomes of the order Enterobacteriales were examined one species at a time using E. coli YbdB or YdiI as query in parallel BLAST searches. We found one, two, or zero closely related sequence homologs (viz., >50% sequence identity for >85% coverage) for each of the 92 genomes. To distinguish between YbdB or YdiI orthologs, we compared pairwise sequence identities (viz., homolog vs E. coli YbdB and homolog vs E. coli YdiI; higher sequence identity associated with the orthologous pair) and we checked for evidence of encoded enterobactin and menquinone pathway enzymes by carrying out BLAST searches using the sequences of the respective pathway enzymes as queries.

Results and Discussion

pH Dependence of YbdB and YdiI Catalytic Efficiency

The variation in kcat and kcat/Km values for YdiI and YbdB-catalyzed thioester hydrolysis as a function of reaction solution pH was measured for the purpose of defining the optimal pH range for catalysis. 4-Hydroxybenzoyl-CoA (4-HB-CoA) was selected to serve as the substrate in the pH rate profile determinations because it is a highly active substrate for both enzymes (see Table 1), and its reaction can be monitored directly and continuously by measuring the decrease in solution absorbance (at 300 nm) associated with the cleavage of its para-hydroxyphenyl-conjugated thioester group.[27] As illustrated in Figure 1, the kcat and kcat/Km values measured for YbdB are maximal at or near neutral pH, whereas both drop at acidic and at basic pH. This trend was also noted by Guo and co-workers for YbdB-catalyzed 2-hydroxybenzoyl-CoA and 2,3-dihydroxybenzoyl-CoA hydrolysis.[1] The YdiI kcat/Km value decreases below pH 6.5 and above 8.5, whereas the kcat value is relatively constant above pH 6.5 but then decreases with decreasing pH.

Table 1

Steady-State Kinetic Parameters of YbdB- and YdiI-Catalyzed Hydrolysis of Various Acyl-CoA, Aryl-CoA, Aryl-Holo-ACP, or Aryl-holo-EntB Substrates at pH 7.5 and 25 °Ca

	YdiI			YbdBb
substrateb	kcat (s–1)	Km (μM)	kcat/Km (M–1 s–1)	kcat (s–1)	Km (μM)	kcat/Km (M–1 s–1)
acetyl-CoA	<1 × 10–4	NDd	NDd	(4.4 ± 0.2) × 10–3b	800 ± 90	5.5
propionyl-CoA	(2.1 ± 0.1) × 10–1	120 ± 10	1.7 × 103	(1.3 ± 0.1) × 10–2b	400 ± 40	3.1 × 101
β-methylcrotonyl-CoA	(5.0 ± 0.2) × 10–1	69.4 ± 0.4	7.3 × 103	NDd	NDd	NDd
β-methylmalonyl-CoA	(6.7 ± 0.3) × 10–1	115 ± 7	5.8 × 103	NDd	NDd	NDd
hexanoyl-CoA	(3.0 ± 0.1) × 10–1	21 ± 1	1.4 × 104	(1.4 ± 0.1) × 10–1b	260 ± 20	5.2 × 102
decanoyl-CoA	NDd	NDd	NDd	(2.7 ± 0.1) × 10–2b	45 ± 2	5.4 × 102
lauroyl-CoA	(7.4 ± 0.1) × 10–1	2.2 ± 0.1	3.3 × 105	(2.8 ± 0.1) × 10–2b	44 ± 2	6.2 × 102
myristoyl-CoA	(6.2 ± 0.1) × 10–1	1.5 ± 0.2	4.1 × 105	(7.8 ± 0.3) × 10–2	11 ± 1	7.1 × 103
palmitoyl-CoA	(5.8 ± 0.1) × 10–1	1.9 ± 0.1	3.0 × 105	(8.5 ± 0.3) × 10–2b	55 ± 9	1.5 × 103
oleoyl-CoA	(1.2 ± 0.1) × 10–1	1.3 ± 0.1	9.2 × 104	(3.0 ± 0.2) × 10–2	13 ± 2	2.3 × 103
benzoyl-CoA	17.7 ± 0.7	25 ± 3	7.1 × 105	2.2 ± 0.2	12 ± 1	1.8 × 105
4-HB-CoA	5.2 ± 0.2	9 ± 1	5.8 × 105	1.6 ± 0.1b	21 ± 2	7.6 × 104
3-HB-CoA	NDd	NDd	NDd	1.2 ± 0.01b	37 ± 1	3.4 × 104
1,4-DHN-CoA	1.6 ± 0.1	8 ± 1	2.0 × 105	(9.3 ± 0.2) × 10–3	17 ± 1	5.5 × 102
3-HPA-CoA	NDd	NDd	NDd	2.1 ± 0.5b	37 ± 1	5.7 × 104
coumaroyl-CoA	8.4 ± 0.2	30 ± 2	2.8 × 105	(8.2 ± 0.2) × 10–1	10 ± 1	8.2 × 104
2,4-DHB-EntB	3.6 × 10–3	200 ± 20	1.8 × 101	3.7 ± 0.1b	25 ± 1	1.4 × 105
2,3-DHB-EntB	NDd	NDd	NDd	2.8 ± 0.1b	15 ± 1	1.8 × 105
lauroyl-EntB	NDd	NDd	NDd	(1.0 ± 0.01) × 10–1b	32 ± 2	6.3 × 102
benzoyl-ACP	(8.3 ± 0.7) × 10–2	54 ± 5	1.5 × 103	(1.3 ± 0.1) × 10–2	60 ± 10	2.2 × 102

See Materials and Methods for details.

Kinetic constants are from ref (17).

Abbreviations: HB-CoA, hydroxybenzoyl-CoA; DHN-CoA, dihydroxynapthoyl-CoA; HPA-CoA, hydroxyphenylacetyl-CoA; DHB-EntB, dihydroxybenzoyl-EntB.

ND stands for not determined.

Figure 1

Plots of (A) log kcat or (B) log kcat/Km measured for YdiI (○) and YbdB (●) catalyzed hydrolysis of 4-hydroxbenzoyl-CoA at 25 °C.

The pH-dependences of YdiI and YbdB catalysis are similar but not identical. The kinetic experiments reported below were carried out at a solution pH of 7.5 because both enzymes are fully active and stable at this pH, and the DTNB-based assay for continuous monitoring via coupled reaction with the CoA thiolate anion could be used in the determination of the substrate specificity profiles (see below).

YbdB and YdiI Substrate Specificity Profiles

The steady-state rate constants kcat and Km were determined for YbdB- and YdiI-catalyzed hydrolysis of a structurally diverse library of thioesters for the purpose of evaluating selectivity toward the thiol moiety (CoA vs holoACP)[2] and toward the acyl or aroyl moiety (see Chart 1 for chemical structures). The CoA thioesters represent various classes of metabolites, differing in the size, shape, and polarity of the acyl/aroyl group. The acyl carrier protein (ACP)-based thioesters were used to test recognition of the holoACP of the E. coli fatty acid synthetic (FAS) pathway[28] and the holoACP domain of EntB of the E. coli enterobactin synthetic pathway[29] (Figure 2A).

Chart 1

Structures of Representative Thioesters Tested in the YbdB and YdiI Substrate Activity Screen

Figure 2

Summary of the E. coli enterobactin biosynthestic pathway. (A) Ordering of the clustered genes encoding the steps of the biosynthetic pathway (blue) and the genes encoding proteins involved in enterobactin transport and function (black). The proofreading hotdog-fold thioesterase YbdB gene is colored red. The chemical steps of the biosynthetic pathway catalyzed by the enzymes isochorismate synthase (EntC), bifunctional isochorismate lyase/aryl carrier protein (EntB), 2,3-dihydro-2,3-dihydroxybenzoate dehydrogenase (EntA), phosphopantetheinyltransferase component of entobacterin synthase multienzyme complex (EntD), enterobactin synthase component F (EntF), and 2,3 dihydroxybenzoate-AMP ligase (EntE). (B) Depiction of the mischarging of EntB catalyzed by EntD or holoEntB catalyzed EntE, and the EntB-regenerating thioester hydrolysis reaction (rescue) catalyzed by YbdB (aka EntH). In the figure, “R” can represent any organic group.

The physiological substrate for YbdB (aka EntH) is mischarged holoEntB,[1,6,17] which in principle can be formed by the phosphopantetheinyl transferase (EntD)-catalyzed reaction of an acyl-CoA or aroyl-CoA (in place of CoA) with EntB, or by the ATP-dependent holoEntB-catalyzed aroylation with an aberrant (hydroxy)benzoatec in place of the native substrate 2,3-dihydroxybenzoate[1] (Figure 2B). In our previous study,[17] YbdB was shown to be highly active toward the enterobactin pathway intermediate 2,3-dihydroxybenzoyl-holoEntB (kcat = 2.8 s–1, kcat/Km= 1.8 × 105 M–1 s–1) and toward the ring hydroxyl positional isomer 2,4-dihydroxybenzoyl-holoEntB (kcat = 3.7 s–1, kcat/Km= 1.4 × 105 M–1 s–1), whereas the activity observed toward lauryl-holoEntB was found to be significantly lower (kcat = 0.1 s–1, kcat/Km= 6.3 × 102 M–1 s–1). Thus, the proofreading function of YbdB appears to be primarily directed at stalled aroyl-holoEntB adducts. We also showed that benzoyl, 3-hydroxybenzoyl, and 4-hydroxybenzoyl-CoA are highly active YbdB substrates (kcat values are 2.2, 1.2, and 1.6 s–1, and kcat/Km values are 1.8 × 105, 3.4 × 104, and 7.6 × 104 M–1 s–1)[17] as are 3-hydroxyphenylacetyl-CoA and coumaryl-CoA (kcat values are 2.1 and 0.82 s–1, and kcat/Km values are 5.7 × 104 and 8.4 × 104 M–1 s–1) (Table 1). In contrast, the small acyl-CoA thioesters tested (acetyl-, propionyl-, β-methylcrotonyl-, and β-methylmalonyl-CoA) showed minimal or no detectable substrate activity with YbdB, and the short-to-medium chain length (C6-C12) fatty acyl-CoA thioesters displayed low turnover rates (kcat range 0.03–0.1 s–1) and low kcat/Km values ((5–6) × 102 M–1 s–1) (Table 1). The longer chain (C14–C18) fatty acyl-CoA thioesters also displayed low turnover rates (0.03–0.09 s–1) but had modestly improved kcat/Km values ((2–7) × 103 M–1 s–1) (Table 1). Together the results indicate that YbdB recognizes CoA, as well as holoEntB, as the thioester thiol moiety, and that aroyl- and phenyl-substituted acyl moieties are targeted whereas purely aliphatic acyl moieties are not. Hydroxylation of the substrate benzoyl ring does not appear to have a significant impact on activity.c

Given that YbdB is compatible with both CoA and holoEntB as the substrate thioester moiety, we were curious to learn whether YbdB can distinguish between different cellular ACPs. Thus, the benzoyl adduct of the holoACP which functions in E. coli fatty acid synthesis was prepared for testing. In contrast to the cases of benzoyl-CoA and (hydroxy)benzoyl-holoEntB, benzoyl-holoACP (kcat = 0.013 s–1, kcat/Km= 2.2 × 102 M–1 s–1) proved to be a very poor substrate (Table 1), thereby indicating that the YbdB substrate binding site is selective for the ACP domain of EntB.

As will be detailed in the section that follows, the physiological substrate of E. coli YdiI is the menquinone pathway intermediate 1,4-dihydroxynapthoyl-CoA[18,20] (Figure 3). We were therefore particularly interested in comparing the catalytic efficiency of YbdB and YdiI with this substrate. Whereas the YdiI activity with 1,4-dihydroxynapthoyl-CoA proved to be high (kcat = 1.6 s–1, kcat/Km = 2.0 × 105 M–1 s–1), the YbdB activity was quite low (kcat = 0.009 s–1, kcat/Km = 5.5 × 102 M–1 s–1). Conversely, YdiI proved to be an ineffective catalyst for the hydrolysis of the holoEntB adducts 2,4-dihydroxybenzoyl-holoEntB, 2,3-dihydroxybenzoyl-holoEntB, and lauryl-holoEntB. Only 2,4-dihydroxybenzoyl-holoEntB was hydrolyzed, and this took place at a very slow rate (kcat = 0.0036 s–1, kcat/Km = 1.8 × 101 M–1 s–1). YdiI catalyzed the hydrolysis of the benzoyl-holoACP (FAS) (kcat = 0.083 s–1, kcat/Km = 1.5 × 103 M–1 s–1), yet at an efficiency that is two orders of magnitude lower than that observed with the corresponding CoA thioester, benzoyl-CoA (kcat = 17.7 s–1, kcat/Km = 7.1 × 105 M–1 s–1).

Figure 3

Menaquinone biosynthetic pathway. (A) The ordering of the clustered genes in the genome of Teredinibacter turnerae T7901 is shown to illustrate the colocation of the ydiI with the pathway genes which is found in some bacterial species but does not occur in E. coli. (B) Reaction steps of the E. coli pathway catalyzed by upper pathway enzymes isochorismate synthase (MenF), 2-succinyl-5-enolpyruvyl-6-hydroxy-3-cyclohexene-1-carboxylate synthase (MenD), 2-succinyl-6-hydroxy-2, 4-cyclohexadiene-1-carboxylate synthase (MenH), o-succinylbenzoate synthase (MenC), o-succinylbenzoic acid:CoA ligase (MenE), naphthoate synthase (MenB), 1,4-dihydroxy-2-naphthoyl-CoA thioesterase (YdiI), and lower pathway enzymes 1,4-dihydroxy-2-naphthoate octaprenyltransferase (MenA) and ubiquinone/menaquinone biosynthesis methyltransferase (MenG).

Next, we tested the level of promiscuity that YdiI exhibits toward aroyl-CoA and acyl-CoA thioesters. Although YdiI was not active toward acetyl-CoA, a modest level of activity was observed with propionyl-CoA (kcat = 0.21 s–1 and kcat/Km = 1.7 × 103 M–1 s–1) and the small, polar acyl-CoA thioesters tested, namely, β-methylcrotonyl-CoA and β-methylmalonyl-CoA (kcat = 0.50 s–1 and 0.67 s–1; kcat/Km = 7.3 × 103 M–1 s–1 and 5.8 × 103 M–1 s–1, respectively) (Table 1). The YdiI kcat/Km values measured for the short-to-long chain fatty acyl-CoA thioesters (C6-C18) (0.14–4.1 × 105 M–1 s–1), on the other hand, were observed to be in the same range as the kcat/Km values measured for coumaroyl-CoA (2.8 × 105 M–1 s–1), benzoyl-CoA (7.1 × 105 M–1 s–1), and 4-hydroxybenzoyl-CoA (5.8 × 105 M–1 s–1). However, the YdiI kcat values determined for the fatty acyl-CoA thioesters (range of 0.12–0.74 s–1) are significantly lower than the kcat values measured for the phenyl ring-containing substrates (range of 5.2–17.7 s–1). A similar trend is observed with the YbdB kcat values (Table 1). This finding suggests that productive binding is more likely to occur with an aroyl-thioester substrate than an acyl-thioester substrate. The structural basis for this discrimination is provided in the companion paper.[2]

In summary, the comparison of the YbdB and YdiI substrate specificity profiles reveals that both thioesterases are (i) promiscuous, (ii) most active with benzoyl- and phenylacyl-based thioester substrates, and (iii) significantly more active with a CoA-based thioester substrate than with the corresponding FAS holoACP-based substrate. On the other hand, YbdB is highly active with EntB-based thioester substrates (kcat/Km ∼ 1 × 105 M–1 s–1), whereas YdiI is not (kcat/Km < 20 M–1 s–1), and conversely, YdiI is highly active with 1,4-dihydroxynapthoyl-CoA (kcat/Km = 2.0 × 105 M–1 s–1) and fatty acyl-CoA thioester substrates (kcat/Km ∼ 1 × 105 M–1 s–1), whereas YbdB is not (kcat/Km = 5.5 × 102 M–1 s–1 and kcat/Km ∼ 1 × 103 M–1 s–1, respectively).

YdiI Biological Function

The findings from the gene neighborhood analysis reported in the section which follows revealed that only in some taxomic groups of bacteria the YdiI/YbdB homolog gene is colocated with all (see, e.g., Figure 2A which depicts the ydiI gene neighborhood within the genome of Teredinibacter turnerae T7901) or a partial set (e.g., species of Bacteroides(19)) of the genes encoding the menaquinone pathway enzymes. In E. coli, for example, ydiI is not colocated with the genes that support menaquinone synthesis. To demonstrate that YdiI is an essential catalyst for menaquinone synthesis in E. coli, the ydiI gene knockout mutant was subjected to two lines of investigation. First, Guo et al. showed that napthoquinone production in this mutant was significantly reduced as compared to the parent strain.[18] Second, in the present work, we measured the growth curves for wild-type E. coli cells and for the ydiI gene knockout mutant cells under conditions requiring anaerobic respiration. By serving as an electron transporter, menaquinone supports anaerobic respiration in facultative anaerobes such as E. coli. Precedent for our experiment is provided by previous work wherein it was shown that the disruption of menaquinone synthesis in E. coli via mutation of pathway gene menB or menD had no apparent impact on cell growth under aerobic conditions, whereas these mutant strains did not grow on glucose in the presence of fumarate, the ultimate electron acceptor, under anaerobic conditions.[30] As shown in Figure 4, the growth curves measured for wild-type E. coli K12 and the ydiI-knockout mutant grown on glucose under aerobic conditions are essentially identical. On the other hand, the comparison of the growth curves measured for wild-type E. coli K12 and the ydiI-knockout mutant grown anaerobically in the presence fumarate indicates that growth of the mutant cells is significantly inhibited (Figure 4). Thus, as in the case of the menaquinone pathway genes menB and menD, ydiI is required for growth under oxygen-limited conditions. Indeed, the NCBI GEO Profiles for YdiI reflect a significant increase observed in the expression of ydiI in E. coli cells grown under oxygen-limited conditions.[31]

Figure 4

Comparison of the growth curves measured for E. coli wild-type (black circle) and the E. coli gene knockout mutant ΔydiI (red circle) on glucose as the carbon source under aerobic conditions and for E. coli wild-type (black down triangle) and the E. coli gene-knockout mutant ΔydiI (red down triangle) on glucose in the presence of fumarate under anaerobic conditions.

Together with the high kcat/Km value of 2 × 105 M–1 s–1 measured for YdiI-catalyzed conversion of 1,4-dihydroxynapthoyl-CoA to 1,4-dihydroxynapthoate, these findings provide convincing evidence that YdiI mediates an essential step of the menaquinone biosynthetic pathway of E. coli.

Bioinformatic Analysis of the Divergence of the YbdB and YdiI

Ortholog Tracking

To identify homologs to E. coli YdiI and YbdB, BLAST searches of bacterial proteomes were carried out. In Table 2, we list the number of species per phyla and the percent sequence identity shared with the E. coli YdiI (see Table SI1 of the Supporting Information for an expanded list that gives statistics for each bacterial species). We discovered that the YdiI/YbdB homologs (as defined by sharing 35% or greater sequence identity) are spread throughout evolutionarily diverse bacterial phyla (viz., Acintobacteria, Bacteroidetes/Chlorobi, Chlamydiae, Firmicutes, and Proteobactia) and that only the genomes of certain species of the genera of the family Enterobacteriaceae (order Enterobacteriales; class Gammaproteobacteria) encode two such homologs (see Table 3).

Table 2

Summary of the Number of Species That Encode a YbdB/YdiI Homolog and the Percent Sequence Identity (% SI) Shared with the E. coli YdiI Sequencea

phylum/class	no. of species with homolog	range of % SI
Actinobacteria	29	41–53%
Bacteroidetes	82	36–64%
	28 fusionb
Chlamydiae	1	51%
Chlorobi	11	45–49%
Chloroflexi	3	41–53%
Deinococci	12	40–52%
Firmicutes/Bacilli	70	37–55%
Firmicutes/Clostridia	7	45–50%
Proteobacteria/Alpha	3	47–50%
Proteobacteria/Beta	28	46–56%
Proteobacteria/Delta	5	35–56%
Proteobacteria/Gamma	297 + 40c	37–99%

See Table SI1 of Supporting Information for a listing of the individual species and protein accession codes.

These homologs are fusion proteins which possess an N-terminal haloalkanoic acid dehalogenase (HAD) superfamily domain and a C-terminal YdiI-like domain.[19]

The number of species that possess a second YbdB/YdiI homolog.

Table 3

Findings from the BLAST searches of the 173 Enterobacteriales Genomes Deposited in the NCBI Database Using the E. coli YbdB and YdiI Protein Sequences as Queriesa

genusbc	species	no.	thioesterase(s)	Ent genes	Men genes	residue 68
Brenneria	sp. EniD312	1	YdiI (68) [YbdB (60)]	BFX	DBE	Val
Citrobacter	koseri	2	YbdB (93) and YdiI (91)	BFD	DBE	Met and Val
Citrobacter	rodentium	2	YbdB (94) and YdiI (89)	BFD	DBE	Met and Val
Citrobacter	sp. 30_2	2	YbdB (92) and YdiI (90)	BFD	DBE	Met and Val
Citrobacter	youngae	2	YbdB (94) and YdiI (86)	BFD	DBE	Met and Val
Cronobacter	turcensis	2	YbdB (81) and YdiI (76)	BFD	DBE	Met and Val
Cronobacter	sakazaki	2	YbdB (82) and YdiI (76)	BFD	DBE	Met and Val
Dickeya	dadantii	2	YbdB (57)c and YdiI (64)	BFX	DBE	Ile and Val
Dickeya	zeae	2	YbdB (57)c and YdiI (64)	BFX	DBE	Ile and Val
Edwardsilla	icttaluri	1	YdiI (71) [YbdB(60)]	XXX	DBE	Leu
Edwardsilla	tarda	1	YdiI (71) [YbdB (60)]	XXX	DBE	Leu
Enterobacter	asburiae	2	YbdB (88) and YdiI (81)	BFD	DBE	Met and Val
Enterobacter	sp. R4-368, 638	2	YbdB (91) and YdiI (82)	BFD	DBE	Met and Val
Enterobacter	cloacae	2	YbdB (88) and YdiI (81)	BFD	DBE	Met and Val
Enterobacter	hormaechei	2	YbdB (89) and YdiI (80)	BFD	DBE	Met and Val
Enterobacter	lignolyticus	2	YbdB (89) and YdiI (78)	BFD	DBE	Met and Val
Enterobacter	cancerogenus	2	YbdB (89) and YdiI (79)	BFD	DBE	Met and Val
Enterobacter	aerongenes	2	YbdB (85) and YdiI (79)	BFD	DBE	Met and Val
Enterobacteriaceae	bacterium FGI 57	2	YbdB (92) and YdiI (76)	BFD	DBE	Met and Met
Enterobacteriaceae	bacterium 9254FAA	1	YdiI (76) [YbdB (60)]	XXX	DBE	Leu
Erwinia	billiingiae Eb661	1	YdiI (67) [YbdB (60)]	XXX	DBE	Met
Escherichia	albertii	2	YbdB (99) and YdiI (93)	BFD	DBE	Met and Ile
Escherichia	coli	2	YbdB (100) and YdiI (100)	BFD	DBE	Met and Val
Escherichia	fergusonii	2	YbdB (100) and YdiI (86)	BFD	DBE	Met and Val
Klebsilla	oxytoca	2	YbdB (86) and YdiI (82)	BFD	DBE	Met and Val
Klebsilla	pneumoniae	2	YbdB (92) and YdiI (82)	BFD	DBE	Met and Val
Morganella	morganii	1	YdiI (68) [YbdB (60)]	XXX	DBE	Val
Pantoea	ananatis LMG 5342	1	YbdB (62) nor YdiI (62)	XXX	XXX	Val
Pantoea	vagans C91	1	YbdB (59) nor YdiI (63)	BFX	XXX	Ile
Pantoea	sp. At-9B	1	YbdB (61) nor YdiI (63)	BFX	XXX	Ile
Pectobacterium	carotovorum	1	YdiI (70) [YbdB (59)]	BfXd	DBE	Val
Pectobacterium	wasabiae	1	YdiI (69) [YbdB (59)]	BfXd	DBE	Val
Pectobacterium	atrospsepticum	1	YdiI (69) [YbdB (59)]	BfXd	DBE	Val
Pectobacterium	sp. SCC3193	1	YdiI (69) [YbdB (59)]	BfXd	DBE	Val
Photorhabdus	asymbiotica	1	YdiI (70) [YbdB (56)]	BXX	DBE	Leu
Photorhabdus	luminescens	1	YdiI (70) [YbdB (57)]	BXX	DBE	Leu
Proteus	mirabilis	1	YdiI (71) [YbdB (1)]	XXX	DBE	Ile
Proteus	penneri	1	YdiI (72) [YbdB (59)]	XXX	DBE	Val
Providencia	alcalifaciens	1	YdiI (76) [YbdB (57)]	XXX	DBE	Leu
Providencia	rettgeri	1	YdiI (71) [YbdB (58)]	XXX	DBE	Ile
Providencia	rustigianii	1	YdiI (71) [YbdB (57)]	XXX	DBE	Ile
Providencia	stuartii	1	YdiI (71) [YbdB (57)]	XXX	DBE	Ile
Rahnella	aquatilis	1	YdiI (69) [YbdB (56)]	XXX	DBE	Met
Rahnella	sp. Y9602	1	YdiI (69) [YbdB (56)]	XXX	DBE	Met
Raoultella	ornithinolytica	2	YbdB (85) and YdiI (82)	BFD	DBE	Met and Val
Samonella	enterica	2	YbdB (92) and YdiI (90)	BFD	DBE	Met and Val
Samonella	bongori	2	YbdB (91) and YdiI (86)	BFD	DBE	Met and Val
Serratia	marcescens	1	YdiI (72) [YbdB (59)]	BFX	DBE	Met
Serratia	odorifera	1	YdiI (74) [YbdB (59)]	BFX	DBE	Met
Serratia	plymuthica	1	YdiI (73) [YbdB (59)]	BFX	DBE	Met
Serratia	proteamaculans	1	YdiI (74) [YbdB (59)]	BFX	DBE	Met
Serratia	symbiotica	1	YdiI (67) nor YbdB (54)	BXX	DBX	Ile
Serratia	sp. AS12	1	YdiI (73) [YbdB (59)]	BFX	DBE	Met
Shigella	sp. D9	2	YdiI (100) and YbdB (99)	BFX	DBE	Met and Val
Shigella	flexneri	2	YdiI (99) and YbdB (99)	BFD	DBE	Met and Val
Shigella	dysenteriae	2	YdiI (99) and YbdB (99)	BFD	DBE	Met and Val
Shigella boydii	boYdiI Sb227	2	YdiI (99) and YbdB (100)	BFD	DBE	Met and Val
Shigella	sonnei 53G	2	YdiI (99) and YbdB (99)	BFD	DBE	Met and Val
Shimwellia	blattae	2	YdiI (73) and YbdB (74)	BFX	DBE	Met and Val
Xenorhabdus	bovienii	1	YdiI (65) [YbdB (54)]	BFX	DBE	Leu
Xenorhabdus	nematophila	1	YdiI (72) [YbdB (59)]	XXX	DBE	Phe
Yersinia	aldovae	1	YdiI (68) nor YbdB (63)	XXX	DXE	Met
Yersinia	bercovieri	1	YdiI (68) nor YbdB (63)	XXX	DXE	Met
Yersinia	enterocolitica	1	YdiI (68) [YbdB (63)]	XXX	DBE	Met
Yersinia	frederiksenii	1	YdiI (65) nor YbdB (62)	BFX	DXE	Met
Yersinia	intermedia	1	YdiI (69) [YbdB (60)]	XXX	DBE	Met
Yersinia	kristensenii	1	YdiI (68) nor YbdB (62)	BFX	DXE	Met
Yersinia	mollaretii	1	YdiI (70) nor YbdB (61)	XXX	DXE	Met
Yersinia	pestis KIM10+	1	YdiI (70) [YbdB (62)]	XXX	DBE	Met

Listed under the column heading “no.” is the number of YbdB and YdiI sequence homologs found having sequence identities >50% over >85% coverage. Reported under the column heading “thioesterases” are the sequence identities determined for the single homologs aligned separately with the E. coli YbdB and YdiI, and listed within the respective parentheses. In cases of two homologs, the reported sequence identity reflects that determined using the most homologous E. coli sequence. Ortholog assignment is indicated by bold font. The term “nor” is used to indicate that function not known, and ortholog assignment is not made. The column headings “Ent genes” and “Men genes” denote the presence of probable orthologs to the three diagnostic enterobactin pathway genes (entB, entF, and entD), represented as B, F, D, and to the three diagnostic menaquinone pathway genes (menD, menB, and menE), represented as D, B, E, respectively. An absent gene is represented using “X”. The probable orthologs were identified by BLAST searches of the individual proteomes using the respective E. coli pathway enzyme sequences as query, coupled with the criteria of >40% pairwise sequence identity for >85% coverage. See text for description of residue 68. See Table SI2 of the Supporting Information for protein accession codes.

Bacterial species found not to possess YbdB/YdiI homolog genes include Buchnera aphidicola str. Sg (Schizaphis graminum); Blochmanniachromaiodes, pennsylvanicus, and vafer; Hamiltonella defensa; Morganella endobia; Riesia pediculicola; Erwina amylovora, chrysanthemi, pyrifoliae, sp. Ejp617, and tasmaniensis; Klebsilla variicola, sp. 1 1 55, sp. 4 1 44FAA, sp. KTE92, sp. MS 92–3, and sp. OBRC7; Sodalis glossinidius; Wigglesworthia glossinidia; Yersinia pseudotuberculosis, rukeri, massiliensis; Yersinia pestis biovar microtus str. 91001, medievalis str. harbin 35.

Posited to function as proofreader in in chrysobactin biosynthetic pathway

The E. coli EntF shares ∼40% identity with this homolog labeled “f”, which is a component of the biosynthetic pathway leading to the nonribosomal peptide pyoverdine.

To assign biological function to the Ybdb/YdiI homologs, we carried out additional BLAST searches using as queries the E. coli MenB, MenD, and MenE sequences to detect the menaquinone pathway, and the E. coli EntB, EntD, and EntF sequences to detect the enterobactin pathway.d The occurrence of the enterobactin pathway enzymes was restricted to certain species of the Enterobacteriaceae family, each of which possesses two Ybdb/YdiI homologs (see Table 3). On the other hand, the menaquinone pathway enzymes were found to be encoded by the genomes of most, yet not all, of the species that encode one or two Ybdb/YdiI homologs (Table SI1 of the Supporting Information and Table 3).

Pursuant of our original goal to correlate divergence of YbdB/YdiI homolog function with the divergence of structure, we now take an in-depth look at the homologs within the Enterobacteriaceae family. In Table 3, we list the sequence identities of each YbdB/YdiI homolog with the E. coli YdiI and YbdB. Homologs, which share greater identity with the E. coli YdiI, than with the E. coli YbdB, are viewed as closer in lineage to the former than to the latter. In Table 3, we also list the “score cards” for the presence versus absence of the diagnostic enzymes of the menaquinone and enterobactin pathways. In all but a few species (described below), we found that either all three enzymes of the pathway are present or none are, and based on this information we inferred the biological function of the homolog. With only a few exceptions, which are addressed below, the predicted lineage and biological function support the assignment of each homolog as ortholog to E. coli YdiI versus YbdB.

Specifically, the single YbdB/YdiI homolog found in species of the genera Edwardsilla, Enterobacteriacea (sp. B9254FAA only), Monganella, Proteus, Providencia, Rahnella, Photorhabudus, and Yershina (pestis KIM10+, intermedia, and enterocoltica only) has the highest sequence identity with the E. coli YdiI, and it is accompanied by the menaquinone pathway enzymes and not the enterobactin pathway enzymes. These particular homologs are assigned as E. coli YdiI orthologs (indicated in Table 3 by use of the bold font). Two YbdB/YdiI homologs are found in species of the genera Citrobacter, Cronobacter, Eschericia, Samonella, Raoultella, Shigella, Shimwella, Enterobacteriacea, Enterobacter, and Klebsilla as are the respective sets of three diagnostic enzymes of the menaquinone and enterobactin pathways. In each case, the YdiI ortholog and YbdB ortholog can be clearly distinguished on the basis of the relative sequence identities with the E. coli paralogs.

Exceptions to the two scenarios presented above are attributed to the absence of one or more of the diagnostic menaquinone or enterobactin pathway enzymes. The genome of Pantoea ananatis LMG 5342, for instance, encodes a single YbdB/YdiI homolog, which shares equal (viz., 62%) sequence identity with the E. coli YdiI and YbdB, yet the genome does not encode the enzymes of the menaquinone or enterobactin pathway. The lineage and biological function are thus undefined. The genome of Serratia symbiotica, on the other hand, encodes pathway enzymes EntB, MenD, and MenB only, and a single YbdB/YdiI homolog, which shares 67% sequence identity with E. coli YdiI versus 54% with E. coli YbdB. In contrast, the other species Serratia listed in Table 3 possess all three (diagnostic) menaquinone pathway enzymes, yet only homologs to the enterobactin pathway enzymes EntB and EntF. The single YbdB/YdiI homolog produced in these species shares greater sequence identity with the E. coli YdiI (72–74%) than with the YbdB (54%). These data support the ortholog assignment YdiI for Serratia species marcescens, odorifera, plymuthica, proteamaculans, sp. AS12, and sp. D9; however, the function of the homolog from Serratia symbiotica remains undefined. The conservation of the EntB and EntF homologs among these species raises the possibility of the existence of an alternate siderophore biosynthetic pathway, which does not employ a YbdB-like housekeeper (i.e., EntB proofreader). An analogous scenario exists for two species Xenorhabdus bovienii, which possesses homologs to MenD, MenB, and MenE (hence, the menaquinone pathway) as well as to EntB and EntF (suggestive of a possible pathway leading to an alternate siderophore), and Xenorhabdus nematophila, which conserves the three diagnostic menaquinone pathway enzymes, but lacks the EntB and EntF homologs. As with Xenorhabdus bovienii, Brenneria sp. EniD312 possesses a single YdiI/YbdB homolog, which is most closely related in sequence to that of the E. coli YdiI (vs YbdB), and the three diagnostic menaquinone pathway enzymes plus the homologs to the E. coli EntB and EntF. Whereas we do not know the identity of the pathways that the EntB and EntF homologs of Serratia, Xenorhabdus, and Brenneria serve, the biosynthetic pathway leading to the nonribosomal peptide pyoverdine is the likely home of the EntB and EntF homologs of Pectobacterium. The represented species of Pectobacterium possess the menaquinone pathway in which the single YdiI/YbdB homolog, which is most closely related in sequence to that of the E. coli YdiI (vs YbdB), is predicted to function. In summary, the YdiI/YbdB homologs of Serratia, Xenorhabdus, and Brenneria are assigned as YdiI orthologs.

Notably, the two representative species of the genus Dickeya synthesize chrysobactin[32] in place of enterobactin. The two siderophores are structurally similar; both contain 2,3-dihydoxybenzoate and l-serine modules. Chrysobactin biosynthesis utilizes homologs to the enterobactin pathway synthases EntB and EntF, and we might expect that one of the two encoded YdiI/YbdB homologs has a proofreading role in chrysobactin biosynthesis similar to that performed by the E. coli YbdB in support of enterobactin biosynthesis. This YdiI/YbdB homolog is significantly more distant from the E. coli YbdB (57% sequence identity) than are the assigned YbdB orthologs (>80% sequence identity) (Table 3). The other YdiI/YbdB homolog of Dickeya shares 64% sequence identity with the E. coli YdiI and is predicted to function in the encoded menaquinone pathway, consistent with its assignment as an ortholog to the E. coli YdiI.

The species of the genus Yersinia show the greatest variation with regard to their utilization of the YdiI/YbdB homolog. For instance, we did not find a homolog encoded by the genomes of Yersinia pseudotuberculosis, rukeri, and massiliensis, or by the genomes of Yersinia pestis biovar microtus str. 91001 and medievalis str. harbin 35. On the other hand, the YdiI/YbdB homolog, which is most closely matched with the E. coli YdiI (68–70% sequence identity), coexists with the menaquinone pathway in Yersinia pestis KIM10+, intermedia, and enterocoltica and is thus assigned as a YdiI ortholog. The YdiI/YbdB homologs of Yersinia aldovae, bercovieri, and mollaretii are missing MenB, and all three enterobactin pathway enzymes. Yersinia frederikseni and kristensenii are missing MenB and EntD. The higher sequence identity of these homologs with the E. coli YdiI (vs YbdB) suggests their lineage; however, their functions are not defined.

Gene Context

The genes encoding the proteins involved in the synthesis of 2,3-dihydroxybenzoate, the assembly of enterobactin, as well as its transport and cleavage are clustered on the E. coli genome along with ybdB (see Figure 2). The ybdB orthologs of other species of Enterobacteriaceae (Table 3) are also clustered with enterobactin pathway genes (see Figures SI1 and SI3 of the Supporting Information). Interestingly, the genes encoding the chrysobactin pathway proteins in Dickeya dadantii (and zea) are likewise clustered with the gene encoding the homolog to the YbdB of the enterobactin pathway (Figure SI3).

Oddly, the genes encoding the menaquinone pathway in E. coli are not colocated with ydiI (Figure 5). Instead, located downstream of the YdiI gene is a gene (ydiJ in E. coli) which encodes a large (1018 amino acids) multidomain protein annotated in EcoGene as a FAD-linked, 4Fe-4S cluster-containing oxidoreductase of unknown function. Located upstream is the suf operon, the protein products of which function in Fe-S cluster assembly under conditions of iron starvation or oxidative stress[33] (Figure 5). This gene context is largely conserved within Enterobacteriaceae;e however, outside this taxonomic group, the gene context is highly varied, and numerous examples can be found where YdiI and the menaquinone pathway genes are colocated (see Figure 2 and Figures SI2 and SI4 of the Supporting Information).

Figure 5

E. coliydiI gene neighborhood (top) and the menaquinone pathway gene cluster (bottom). The arrows indicate the direction of gene transcription.

YbdB and YdiI Sequence Markers

Based on the respective three-dimensional structures reported in the companion paper,[2] we identified amino acids 15, 68, and 71 as points of divergence in the respective E. coli YbdB and YdiI active site regions that accommodate the substrate aryl/alkyl group. For E. coli YbdB versus YdiI, the residues are Thr15 versus Met15, Met68 versus Val68, and Phe71 versus Tyr71, respectively. To determine if these residues are conserved, the amino acid sequences of the YbdB orthologs derived from Enterobacteriaceae species (Table 3) were aligned, as were the YdiI orthologs. The resulting multiple sequence alignments are shown in Figures SI5 and SI6 of the Supporting Information. The YbdB sequence alignment reveals high conservation of Met68 and Thr15 and conservative replacement of Phe71 (with Tyr or Trp). Two outliers are the YbdB homologs of the Dickeya zeae and dadanti, which are presumed to function in the chrysobactin biosynthetic pathway.[32] These homologs share 60% pairwise sequence identity with the E. coli YbdB and possess Met15, Val68, and Trp71. A third outlier is the YbdB homolog of Shimwellia blattae (74% pairwise sequence identity with E. coli YbdB), as it possesses Met15, Met68, and Phe71. The residue at position 15 in YdiI is variable (the E. coli YdiI Met15 is replaced with polar residues as well as nonpolar residues), the E. coli YbdB Val68 is conservatively replaced with Met, Leu, or Ile, and the Tyr71 is stringently conserved. Notably, the strict conservation of YdiI Tyr71 is not required for efficient catalysis as revealed by the high (wild-type level) activity of the YdiI Y71A mutant (Table 4).

Table 4

Steady-State Kinetic Constants for Mutant YdiI- and YbdB-Catalyzed Hydrolysis of Benzoyl-CoA, Lauroyl-CoA, and 1,4-Dihydroxynapthoyl-CoA Measured at pH 7.5 and 25 °C

	benzoyl-CoA			1,4-dihydroxynapthoyl-CoA			lauroyl-CoA
	kcat (s–1)	Km (μM)	kcat/Km (M–1 s–1)	kcat (s–1)	Km (μM)	kcat/Km (M–1 s–1)	kcat (s–1)	Km (μM)	kcat/Km (M–1 s–1)
YbdB
WTa	2.2 ± 0.1	12 ± 1	1.8 × 105	0.0093 ± 0.002	16 ± 2	5.8 × 102	0.028 ± 0.002	44 ± 2	6.4 × 102
M68V	1.8 ± 0.1	24 ± 2	7.5 × 104	0.16 ± 0.02	5.9 ± 0.6	2.7 × 104	0.046 ± 0.004	10.4 ± 0.3	4.5 × 103
YdiI
WTa	18 ± 1	25 ± 3	7.2 × 105	1.6 ± 0.1	8 ± 1	2.0 × 105	0.74 ± 0.01	2.2 ± 0.2	3.4 × 105
V68M	28 ± 1	27 ± 2	1.0 × 106	0.8 ± 0.1	2.0 ± 0.2	4.0 × 105	0.77 ± 0.03	28 ± 2	2.8 × 104
Y71A	28 ± 1	61 ± 4	4.6 × 105	NDb	NDb	NDb	0.95 ± 0.02	3.4 ± 0.2	2.8 × 105

WT stands for wild-type.

ND stands for not determined.

Despite some variation in the identity of the amino acid at position 68 in YdiI, based on structural considerations, we posit that this position is most closely linked to substrate binding. Specifically, the structures of the E. coli paralogs (see the companion paper[2]) show that Met68 in YbdB and Val68 in YdiI are located at the back of the alkyl/aryl binding pocket. To examine the importance of these residues to substrate recognition, the “residue-swapped” site-directed mutants YbdB M68V and YdiI V68M were prepared, and subjected to steady-state kinetic analysis (see Table 4 for the kinetic constants). The substrates tested with these mutants are benzoyl-CoA (highly active substrate for both YbdB and YdiI), 1,4-dihydroxynapthoyl-CoA (highly active substrate for YdiI but not YbdB), and lauroyl-CoA (highly active substrate for YdiI but not YbdB). The kcat/Km values measured for YbdB M68V and YdiI V68M catalyzed-hydrolysis of benzoyl-CoA are essentially the same as those measured for the wild-type enzymes. In contrast, the kcat/Km value measured for YbdB M68V with 1,4-dihydroxynapthoyl-CoA serving as substrate (2.7 × 104 M–1 s–1) is 47-fold larger than that measured with wild-type YbdB. The kcat/Km value measured for YdiI V68M with 1,4-dihydroxynapthoyl-CoA serving as substrate is essentially unchanged. With lauroyl-CoA serving as substrate, the kcat/Km determined for YdiI V68M is reduced ∼10-fold in value whereas the kcat/Km value measured for YbdB M68V (4.5 × 103 M–1 s–1) is increased ∼10-fold. Thus, the comparatively lower catalytic efficiency of YbdB with the larger alky- and aryl-substituted substrates can be explained, at least in part, by the steric constraints imposed by the Met68.

Conclusion

We have shown that the biological range of YbdB is restricted to species of the family Enterobacteriaceae, whereas that of YdiI extends well beyond this, to other divisions of Proteobacteria as well as to other phyla (see Figure 6). The high overall sequence identity between E. coli YbdB and YdiI (59%) indicates that they are paralogs. Whereas YdiI is found in the vast majority of represented Enterobacteriaceae species, either alone or in combination with YbdB, the occurrence of YbdB is more restricted and coincides with that of YdiI. Indeed, whereas the siderophore enterobactin is unique to Enterobacteriaceae species, the electron acceptor menaquinone is required by a wide range of facultative anaerobes, including most species of Enterobacteriaceae. Thus, a reasonable scenario is that YbdB evolved within Enterobacteriaceae via divergence of an ancestral gene possibly following its duplication or horizontal transfer.[34,35]

Figure 6

(A) Phylogenetic tree (generated by iTOL) showing bacterial phyla that possess two (red), one (blue), or no (black) YdiI/YbdB homologs. (B) Phylogenetic tree (generated by iTOL) of the Enterobacteriales order depicting species having both YdiI and YbdB orthologs (blue), a YdiI ortholog and a functional analog of YbdB (green), a YdiI ortholog only (red), and last a single YdiI/YbdB homolog having an unknown function (cyan).

The YbdB and YdiI substrate specificity profiles (Table 1) show that both enzymes possess a high level of promiscuity, which promotes evolvability but, at the same time, increases the danger of unwanted hydrolysis of off-target thioester metabolites. As a safeguard, transcription of the YdiI and YbdB genes should be tightly regulated.f In fact, the entCEBA-ybdB operon is under the control of the entCp promoter, which is repressed by the transcriptional regulator protein Fur when the iron concentration in the cell is adequate.[36] Bouveret and co-workers have shown that the production of YbdB in E. coli is induced by iron starvation.[6] Although the mechanism of transcriptional regulation of the menaquinone pathway genes, including ydiI, has not yet been reported, it has been shown that the level of ydiI transcription is significantly elevated during anaerobic growth of E. coli cells on fumarate,[31] conditions which require menaquinone production.

The specialization of biological function in YdiI and YbdB is clearly supported by their divergence in substrate recognition. The aroyl-binding site of YbdB does not accept the napthoyl group, which we have shown for the E. coli enzyme can be attributed, at least in part, to Met68. Conversely, YdiI does not recognize EntB, and this suggests that there are important differences in the structures of the respective regions that accommodate the CoA nucleotide in both enzymes, and the EntB in YbdB only. To gain insight into the structural determinants, which support the divergence in YbdB and YdiI substrate specificity, the X-ray structure determinations reported in the companion paper[2] were carried out.