Crystal structures of the fatty acid biosynthesis initiation enzymes in Bacillus subtilis

Abstract

Bacteria use the fatty acid composition of membrane lipids to maintain homeostasis of the bilayer. β-Ketoacyl-ACP synthase III (FabH) initiates fatty acid biosynthesis and is the primary determinant of the fatty acid composition. FabH condenses malonyl-acyl carrier protein with an acyl-Coenzyme A primer to form β-ketoacyl-acyl carrier protein which is used to make substrates for lipid synthesis. The acyl-Coenzyme A primer determines whether an acyl chain in the membrane has iso, anteiso, or no branching (straight chain) and biophysical properties of the membrane. The soil bacterium Bacillus subtilis encodes two copies of FabH (BsFabHA and BsFabHB), and here we solve their crystal structures. The substrate-free 1.85 Å and 2.40 Å structures of BsFabHA and BsFabHB show both enzymes have similar residues that line the active site but differ in the architecture surrounding the catalytic residues and oxyanion hole. Branching in the BsFabHB active site may better accommodate the structure of an iso-branched acyl-Coenzyme A molecule and thus confer superior utilization to BsFabHA for this primer type. The 2.02 Å structure of BsFabHA•Coenzyme A shows how the active site architecture changes after binding the first substrate. The other notable difference is an amino acid insertion in BsFabHB that extends a cap that covers the dimer interface. The cap topology is diverse across FabH structures and appears to be a distinguishing feature. FabH enzymes have variable sensitivity to natural product inhibitors and the availability of crystal structures help clarify how nature designs antimicrobials that differentially target FabH homologs.

1. Introduction

Bacterial fatty acid synthesis (FASII) is an essential process for cell viability that consists of repeating cycles of condensation, reduction, dehydration, and reduction that are carried out by a collection of enzymes (Parsons and Rock, 2013). The FASII substrates are linked to acyl carrier protein (ACP) and are extended by two carbons after each cycle to yield acyl-ACP (Rock and Jackowski, 2002). β-Ketoacyl-ACP synthase III (FabH) uses a ping-pong mechanism to catalyze a decarboxylative Claisen condensation between acyl-Coenzyme A (CoA) and malonyl-ACP to form acetoacyl-ACP and initiate FASII (Fig. 1), and differs from other FASII condensing enzymes because of its selectivity of acyl-CoA as a primer instead of acyl-ACP (Heath and Rock, 1996a; Lu et al., 2004; Tsay et al., 1992). FabH determines the total number of fatty acids synthesized by the cell and is a point of FASII regulation by acyl-ACP (Heath and Rock, 1996a; Heath and Rock, 1996b; Jackowski and Rock, 1987).

Fig. 1. FabH reaction.

FabH catalyzes the formation of acetoacyl-ACP to initiate FASII.

Crystal structures reveal the FabH protein family (EC 2.3.1.180) has a conserved fold belonging to the c.95.1.1 thiolase-related family characterized by two similar subdomains related by a pseudo dyad in the Structural Classification of Proteins extended database (scop.berkeley.edu). Condensing enzymes use a cysteine thiol as a nucleophile for the substrate in the enzyme reaction (Huang et al., 1998; Kauppinen et al., 1988; Tsay et al., 1992), and the FabH cysteine is part of a conserved active site catalytic triad of cysteine, asparagine, and histidine. FabH uses cysteine to attack acetyl-CoA to form the acetyl-FabH intermediate, and asparagine and histidine to interact with the thioester carbonyl of malonyl-ACP and facilitate enolization and the formation of a carbanion on C2 of malonate (Davies et al., 2000) (Boram et al., 2023) (Fig. S1).

Bacteria can encode one or multiple copies of FabH. The soil bacterium Bacillus subtilis encodes two FabH homologs, BsFabH1/BsFabHA and BsFabH2/BsFabHB, that are 43% identical/61% similar to each other. Many Gram-positive bacteria produce odd- and even-number branched chain fatty acids with either iso- or anteiso-methyl branches (Kaneda, 1991). CoA thioesters are used as FASII primers and FabH uses the iso- and anteiso-branched α-keto acids derived from valine, leucine, and isoleucine to initiate biosynthesis of branched chain fatty acids (Wang et al., 1993; Willecke and Pardee, 1971). In biochemical assays, both BsFabHA and BsFabHB use 2-methylbutyryl-CoA, which is a primer for the biosynthesis of anteiso fatty acids, while BsFabHB has 4–10–fold greater activation than BsFabHA of isobutyryl-CoA and isovaleryl-CoA, which are primers for the biosynthesis of iso fatty acids (Choi et al., 2000). A ΔfabHA deletion strain has an increased abundance of straight chain fatty acids and decrease in iso fatty acids relative to the parent strain in B. subtilis lipids, whereas a ΔfabHB deletion strain has a small reduction in straight chain fatty acids and increase in iso fatty acids relative to the parent strain (Kingston et al., 2011). BsFabHB is utilized under the alkaline shock condition that inhibits BsFabHA via the σ^W-dependent stress response (Kingston et al., 2011), and BsFabHB is upregulated 3–4–fold over BsFabHA in the presence of FASII inhibitors (Wenzel et al., 2011).

Natural products have been discovered, like thiolactomycin (Price et al., 2001) and conocandin (Bae et al., 2021) fungal products, and platencin (Wang et al., 2007) and amycomicin (Pishchany et al., 2018) bacterial products, that inhibit FabH and demonstrate FabH is a viable antimicrobial target. A B. subtilis ΔfabHB deletion strain is sensitive to amycomicin whereas a ΔfabHA deletion strain is resistant, and a S. aureus strain overexpressing BsFabHB is resistant to amycomycin whereas the S. aureus parent strain is sensitive (Pishchany et al., 2018). 3-dimensional structures of BsFabHA and BsFabHB would help understand these different phenotypes that are not apparent from the amino acid composition and guide the design of small molecule FabH inhibitors.

In this study we used X-ray crystallography to compare the structures of BsFabHA and BsFabHB with the hypothesis that structural differences may be related to biochemical substrate selectivity and resistance to natural product inhibition. Our findings give us a better understanding of the structural features involved in the FASII initiating condensation reaction and could be applied to basic research for the development of FabH inhibitors.

2. Materials and methods

2.1. FabH expression and purification

DNA fragments encoding BsFabHA or BsFabHB were synthesized and cloned into pET28a with a cleavable N-terminal His tag for purification from Escherichia coli using 5′-NdeI and 3′-EcoRI cloning sites. The plasmids were transformed into E. coli BL21(DE3) cells (Millipore Sigma), and isolates were obtained in Luria broth agar imbedded with 50 μg/μl kanamycin (Gold Biotechnology). Transformants were amplified in Luria broth containing 50 μg/μl kanamycin (Gold Biotechnology) and shaken at 37 °C, 200 rpm. Cells were grown to an OD₆₀₀ of 0.6 and then cooled to 16 °C before overnight induction with 1 mM isopropyl-β-D-thiogalactoside (Gold Biotechnology). Cells were harvested and then lysed in buffer containing 20 mM Tris, pH 8.0, 10 mM imidazole, 200 mM NaCl, and a dissolved tablet of Pierce protease inhibitor (Thermo Fisher Scientific). The BsFabHA and BsFabHB proteins were separated from cell lysates by nickel agarose beads (Gold Biotechnology) and eluted in buffer containing 20 mM Tris, pH 8.0, 250 mM imidazole, and 200 mM NaCl. The eluant was gel filtered into buffer containing 20 mM Tris, pH 7.5, 200 mM NaCl, and 1 mM EDTA using the HiLoad Superdex 200 16/60 column (Cytiva Life Sciences) (Fig. S2). The N-terminal His tag was cleaved from BsFabHA for crystallization using biotinylated thrombin protease, and the biotinylated thrombin was removed using streptavidin agarose (MilliporeSigma). Uncleaved BsFabHA was removed by subtractive nickel column purification and the cleaved BsFabHA was exchanged back into 20 mM Tris, pH 7.5, 200 mM NaCl, and 1 mM EDTA by gel filtration.

2.2. Crystallization

BsFabHA (no His-tag) was concentrated to 15 mg/ml for crystallization, and initial screening was performed at 20 °C against JCSG Core III Suite (Qiagen) by hanging drop vapor diffusion method combining 200 nl protein and 200 nl precipitant (Fig. S3A). Diffraction quality crystals were obtained by combining 1.5 μl protein and 1.5 μl 50% PEG400, 0.2 mM NaCl, 0.1 M N-cyclohexyl-2-aminoethanesulfonic acid (CHES), pH 9.5. BsFabHA crystals were soaked overnight in precipitant containing 10 mM CoA to incorporate CoA into the BsFabHA crystal lattice. Crystals were cryo-protected with the precipitant solution for flash-freezing in liquid nitrogen for X-ray diffraction experiments.

His-tagged BsFabHB was concentrated to 32 mg/ml for crystallization, and initial screening was performed at 20 °C against JCSG Core III Suite (Qiagen) by hanging drop vapor diffusion method combining 200 nl protein and 200 nl precipitant (Fig. S3B). Diffraction quality crystals were obtained by combining 1.5 μl protein and 1.5 μl 20% 2-methyl-2,4-pentanediol (MPD), 0.1 M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 6.5. Crystals were cryo-protected with 25% MPD, 0.1 M HEPES, pH 6.5, and 25% glycerol and then flash-frozen in liquid nitrogen for X-ray diffraction experiments.

2.3. X-ray data collection, structure determination, and refinement

X-ray diffraction data were collected at 1 Å wavelength at the SER-CAT beamlines 22-ID (BsFabHA and BsFabHB) and 22-BM (BsFabHA•CoA binary complex) at the Advanced Photon Source. Diffraction data were processed in DIALS (Winter et al., 2018) and XDS (Kabsch, 2010). Data resolution cutoffs were guided by CC_1/2 and Mean 1/σ_I (Karplus and Diederichs, 2015). The Matthews coefficient probabilities estimated BsFabHA (2.04 ų/Da), BsFabHB (2.30 ų/Da), and BsFabHA•CoA binary complex (1.97 ų/Da) contained 1, 7, and 1 molecule(s) in the asymmetric unit, respectively, using the amino acid sequences. The structures of BsFabHA and BsFabHB were solved by molecular replacement using Staphylococcus aureus FabH (PDB: 1ZOW, 58% identical/ 76% similar to BsFabHA) (Qiu et al., 2005) as a search model for BsFabHA and Thermus thermophilus FabH (PDB: 1UB7, 45% identical/ 62% similar to BsFabHB) (Inagaki, 2003) (Inagaki, 2003) a search model for BsFabHB. The structure of BsFabHA•CoA binary complex was solved by molecular replacement using BsFabHA as a search model. The structures were completed by iterative rounds of refinement using phenix.refine (Afonine et al., 2012) and manual rebuilding using Coot (Emsley and Cowtan, 2004). The refinement was monitored by following the R_free value calculated from a random subset (5%) of omitted reflections. A summary of the data processing and structure refinement statistics is provided in Table 1. The structures of BsFabHA, BsFabHA•CoA, and BsFabHB have been deposited in the Protein Data Bank under accessions 8VD9, 8VDA, and 8VDB, respectively. The figures related to protein structures were generated with PyMOL. Gibbs binding free energy was estimated using HawkDock (Weng et al., 2019).

Table 1.

X-ray data collection and refinement statistics.

Data collection	BsFabHA	BsFabHA•CoA	BsFabHB
PDB code	8VD9	8VDA	8VDB
Diffraction source	APS Beamline 22-ID	APS Beamline 22-BM	APS Beamline 22-ID
Wavelength (Å)	1.0000	1.0000	1.0000
Space group	P41212	P41212	P1
Unit cell	a=61.03, b=61.03, c=156.41	a=61.61, b=61.61, c=153.40	a=91.96, b=92.72, c=94.00
	α=90.00°, β=90.00°, γ=90.00°	α=90.00°, β=90.00°, γ=90.00°	α=111.66°, β=110.13°, γ=105.45°
Resolutiona	61.03 – 1.85 (1.90 – 1.85)	76.70 –2.02 (2.07 – 2.02)	77.14 – 2.40 (2.44 – 2.40)
No. measured reflections	429,510 (19,130)	148,915 (6,354)	333,523 (17,556)
No. unique reflections	25,612 (1,550)	19,511 (1,349)	88,179 (4,618)
Completeness (%)	98.0 (82.6)	99.7 (97.1)	93.2 (94.1)
Redundancy	16.77 (12.34)	7.63 (4.71)	3.8 (3.8)
Rmerge	0.074 (0.584)	0.032 (0.447)	0.156 (0.874)
Rmeas	0.077 (0.609)	0.035 (0.503)	0.182 (1.017)
Mean I/σI	22.99 (3.87)	34.85 (3.22)	4.5 (1.8)
CC1/2	0.999 (0.884)	1.000 (0.836)	0.983 (0.743)
Wilson B-factor (Å2)	31.80	37.84	46.53
Refinement
Resolution	30.51 – 1.85 (1.916 – 1.85)	41.28 – 2.02 (2.092 – 2.02)	45.2 – 2.4 (2.486 – 2.4)
No. reflections used in refinement	25,051 (1,964)	19,026 (1,698)	88,087 (9,017)
No. reflections used for Rfree	1,999 (157)	1,903 (1,70)	1,994 (193)
Rwork	0.2121 (0.2819)	0.1973 (0.2502)	0.2107 (0.2989)
Rfree	0.2495 (0.3137)	0.2370 (0.2676)	0.2299 (0.3405)
Number of non-hydrogen atoms	2487	2481	10316
macromolecules	2348	2348	9916
ligands	13	80	18
waters	126	85	382
Protein residues	311	311	1303
RMS(bonds)	0.004	0.01	0.005
RMS(angles)	0.72	1.04	0.8
Ramachandran favored (%)	97.09	97.41	95.44
Ramachandran allowed (%)	2.59	2.27	4.32
Ramachandran outliers (%)	0.32	0.32	0.23
Rotamer outliers (%)	1.63	2.85	3.8
Clashscore	3.81	5.03	7
Average B-factor	34.62	43.29	51.44
macromolecules	34.26	42.56	51.43
ligands	45.21	79.8	49.94
waters	40.19	42.75	51.9

^aParentheses indicate values for the highest resolution shell

3. Results and discussion

3.1. Structure of BsFabHA

BsFabHA crystallized in the P4₁2₁2 space group, and the 1.85 Å structure was refined to 0.21/0.25 R_work/R_free (Table 1). One protomer was in the asymmetric unit that interacts with a symmetry-related protomer to create the dimer. The presence of dimers in the crystal is consistent with the gel filtration profile of the purified protein (Fig. S2). This result is consistent with other FabH homologs that purify as dimers (Hou et al., 2018; Pereira et al., 2012; Qiu et al., 2001a). The BsFabHA protomer has the prototypical FabH α/β architecture with major and minor repeating domains that are pseudosymmetrically related (Fig. 2A). The first half of the major domain is made up of β1-α3-β4-α4-β5-α5-β6-β7 and the second half with one less β-strand is made up of β8-α6-β11-α7-α8-β12-β13. Each half of the major domain contains a buried mixed β-sheet sandwiched between helices. The major halves meet along the interface created by α5 and α8 that is sandwiched between the β-sheets and establishes the core of the protomer. The first segment of the minor domain is made up of β2-α1-α2-β3-turn which forms the CoA adenosine diphosphate binding site, and the second segment is made up of turn-β9-β10 that provide a surface to the dimer interface. Structural alignment of BsFabHA protomer with Staphylococcus aureus FabH (PDB ID: 1ZOW) showed the structures are highly similar with an RMSD of 0.873 Å across 310 Cα atoms, and enabled identification of the BsFabHA catalytic residues Cys114, His239, and Asn269 at the interface of the protomer domains (Fig. 2B). Cys114 is the amino-terminal residue of helix α5 which positions this key catalytic residue at the partial positive charge of the helix dipole and enhances the nucleophilicity of the cysteine thiol to accelerate catalysis (Hol et al., 1978). Helix dipole α8 is also oriented with its amino terminus directed to Cys114, and helix dipoles α5 and α8 form an 81.29° angle. The organization of the helix dipoles is proposed to contribute a half unit of positive charge and enhance the nucleophilicity of Cys114 to catalyze the thioester exchange reaction (Davies et al., 2000). The BsFabHA protomer has a solvent accessible surface area of ~34,200 Ų and the dimer interface buries ~3,700 Ų (~11%). The overall Gibbs binding free energy for BsFabHA dimerization is favorable and estimated to be −162 kcal/mol. The dimer interface is stabilized by interacting β-strands β4–6 from each protomer that, with β1 and β7, create a mixed ten-stranded β-sheet connecting the major domains from both protomers (Fig. 2C). Additional contacts along the dimer interface between secondary structure elements are interacting β-strands β8 and interacting α-helices α4 and α5. Decomposition of the free energy contribution on a per residue basis (Fig. S4) indicates residues on strand β5 (Met108) and helix α5 (Tyr119, Gln126, and Phe127) are top residues that stabilize dimerization.

Fig. 2. Structure of BsFabHA.

A, The BsFabHA protomer is divided into major and minor domains. The major domain contains key helix dipoles and secondary structural elements that stabilize the overall fold. The minor domain contains the CoA binding helix (α2) and cap subdomain (β9 and β10). B, Catalytic residues are located at the amino termini of helix dipoles that define the BsFabHA active site. Helix dipoles are colored from amino terminus (blue) to carboxy terminus (red) and form an 81.29° angle with respect to each other. C, Secondary structural elements that establish the BsFabHA dimer interface. Interacting α-helices and a β-sheet help stabilize the homodimer.

3.2. Structure of BsFabHB

BsFabHB crystallized in the P1 space group, and the 2.40 Å structure was refined to 0.21/0.23 R_work/R_free (Table 1). Two dimers were observed in the asymmetric unit with an average R.M.S.D. of 0.247 ± 0.026 Å between protomers (across 325 Cα atoms) and 0.222 Å between dimers (across 4,114 total atoms). BsFabHB and BsFabHA are similar with an average protomer R.M.S.D. of 1.250 ± 0.003 Å across 306 Cα atoms, and structural alignment identified the BsFabHB catalytic residues as Cys113, His250, and Asn280 (Fig. 3). BsFabHB has a 12 amino acid extension (Fig. S5) in the second segment of the minor domain that elongates a “cap” subdomain that sits atop the dimer interface. The BsFabHB protomer has a solvent accessible surface area of ~32,300 Ų comparable to BsFabHA but the extension of the minor domain changes the topology and surface electrostatics on top of the dimer interface. BsFabHA has a negatively charged cavity on top of the dimer interface whereas differences in the extension and overall amino acid composition create a positively charged cavity surrounded by a neutral surface in BsFabHB. These features make BsFabHA like Staphylococcus aureus FabH (PDB ID: 1ZOW) with a short cap, and BsFabHB like Escherichia coli FabH (PDB ID: 1EBL) with an elongated cap. The overall Gibbs binding free energy change for BsFabHB dimerization is favorable and estimated to be −271 kcal/mol. Like BsFabHA, decomposition of the free energy contribution on a per residue basis (Fig. S4) indicates residues on strand β5 (Leu107) and helix α5 (Tyr118) are top residues that stabilize dimerization. A distinguishing feature in the BsFabHB Gibbs binding free energy profile is the strong contribution from residues on the surface of the molecule (Asp85 and Tyr86) that are next to a key active site residue (Phe88’), and residues in the cap subdomain (Leu196 and Arg197) that cover the dimer interface.

Fig. 3. BsFabHA vs BsFabHB.

A, Structural alignment of BsFabHA (green) versus BsFabHB (brown). BsFabHB catalytic residues that define the active site are shown. The BsFabHA cap contains β9 and β10, whereas the BsFabHB cap is entirely a flexible loop. B, Surface electrostatic comparison from −5 k_BT/e_c (red) to +5 k_BT/e_c (blue) of the BsFabHA and BsFabHB dimers. The 90° rotated is looking through the cap to the dimer interface.

3.3. Substrate tunnel

The FabH substrate tunnel is a narrow linear passage lined with hydrophobic residues and connects the buried active site catalytic triad with the surrounding bulk solvent. Overall, the substrate tunnels in BsFabHA (Fig. 4A,B) and BsFabHB (Fig. 4C) have similar shapes except BsFabHB has a branch point towards the end of where the pantothenate arm binds. A conserved phenylalanine side chain (Phe299 in BsFabHA and Phe310 in BsFabHB) forms a π-teeing perpendicular T-shaped (edge-to-face) interaction with the catalytic histidine residue and resides on a Gly-Gly-Gly loop (Fig. S5) that contributes a backbone amide to stabilize the oxyanion hole (Davies et al., 2000; Ordentlich et al., 1998) (Fig. S1). Met215 and Phe299 prevent a branching in the BsFabHA substrate tunnel (Fig. 4A,B) that is created by Val226 and Phe310 in the BsFabHB substrate tunnel (Fig. 4C).

Fig. 4. Substrate tunnel.

A, The BsFabHA substrate tunnel (grey wire mesh) is lined by residues from both protomers that shape the architecture. Phe89′ (cyan) is from the opposite protomer. The BsFabHA catalytic residues are Cys114, His239, and Asn269. B, The BsFabHA substrate tunnel (gray wire mesh) undergoes modest architectural change upon binding CoA (yellow) substrate. The Phe299 side chain rotates into a new conformation while there is negligible change in the Phe89′ side chain (green). C, The BsFabHB substrate tunnel (grey wire mesh) is lined by residues from protomer from both protomers that shape the architecture. Phe88′ (magenta) is from the opposite protomer. The BsFabHB catalytic residues are Cys113, His250, and Asn280. The black arrow denotes the branch in the substrate tunnel created by Val226 and Phe310 The catalytic residues create oxyanion holes to accelerate catalysis as shown in Fig. S1.

Phe89’ in BsFabHA and Phe88’ in BsFabHB is the only residue from the opposite protomer that extends in the active site and might convey communication between protomers. This residue is proposed to form the floor of the oxyanion hole with the backbone amide nitrogen atom that may help stabilize the oxyanion transition state (Davies et al., 2000), and the phenyl side chain helps define the primer specificity by interacting with the acyl-FabH intermediate (Qiu et al., 2001b). Resides Ala113 and Thr200 in BsFabHA, and Thr112 and Met211 in BsFabHB, help shape the tunnel architecture.

3.4. CoA binding

The surface of the substrate tunnel is designed to accommodate two acyl substrates (CoA and ACP) so we soaked BsFabHA crystals in CoA to visualize CoA binding in the substrate tunnel (Fig. S6). The 2.02 Å BsFabHA•CoA binary complex structure showed helix α2 is a CoA binding helix that connects antiparallel β-strands β2 and β3 of the minor domain. Helix α2 residue Trp34 and Arg153 from the middle of the minor domain form π-π stacking interactions with the CoA adenine ring (Fig. 5). Helix α6 is the longest helix dipole in the FabH structure and is oriented with its amino terminus alongside but not pointing directly at the CoA pyrophosphate. This orientation may be primarily designed to contribute a partial positive charge to stabilize binding of the second substrate (highly acidic ACP), with the bonus of providing a complementary electrostatic surface for the CoA pyrophosphate. The pyrophosphate is stabilized by hydrogen bonds from the arginine triad of α6 amino terminal residue Arg205, α2 residue Arg38, and α7 residue Arg244 (Fig. 5). The arginine triad also adds to a positively charged surface that makes the substrate tunnel compatible with ACP binding (Zhang et al., 2003a; Zhang et al., 2003b; Zhang et al., 2001). The pantetheine arm of either acyl substrate extends through the tunnel to the catalytic residues (Fig. 4B) as described for other FabH enzymes. Phe299 rotates to a new position in the BsFabHA-CoA structure (Fig. 4B). This same branching and conformation are observed in E. coli FabH-CoA (PDB ID: 1EBL) (Phe304 in E. coli), which the backbone amide is also proposed to bind the malonyl carboxylate of the second substrate malonyl-ACP (Qiu et al., 2001b). The flexibility of the Phe299 side chain and π-teeing perpendicular T-shaped interaction with the catalytic histidine residue has been proposed to exclude water and limit access to the catalytic triad (Qiu et al., 2005). Although E. coli FabH is useful for registering the position of conserved binding and catalytic residues, a difference between E. coli FabH and BsFabHA is E. coli FabH has a substrate preference for short (e.g., C2 acetyl-CoA) primers (Choi et al., 2000). E coli FabH has a small, shallow active site that differs from other FabH enzymes that utilize larger primers.

Fig. 5. BsFabHA-CoA binding interactions.

CoA substrate binding is stabilized by π-π stacking interactions (white dashes an arginine triad, and a helix dipole colored from amino terminus (blue) to carboxy terminus (red). Residue Phe299 is behind the helix dipole.

3.5. FabH cap

Multiple sequence alignment (Fig. S5) shows BsFabHA and BsFabHB differ in the amino acid sequence length of the cap region. The crystal structures of both proteins show this difference impacts accessibility to the dimer interface and surface topology (Fig. 3). The elongated BsFabHB cap contains residues that are predicted to engage in interactions that support dimerization (Fig. S4); however, it is important to note that there is no evidence that variability in the minor domain makes capped FabH enzymes more stable than uncapped FabH enzymes or vice versa.

The Staphylococcus aureus FabH (PDB ID: 1ZOW) contains a truncated minor domain like BsFabHA and a negatively charged cap on top of the dimer interface. E. coli FabH contains a larger minor domain than BsFabHB that covers the dimer interface and creates a channel from bulk solvent to the dimer interface. A substrate-free E. coli FabH structure called “tetragonal FabH” (PDB: 1HNK) shows a disordered cap that coincides with a disordered region of the CoA/ACP substrate binding site, a disordered phenylalanine sidechain that forms the floor of the oxyanion hole (Phe87 in E. coli FabH), a disordered Gly-Gly-Gly loop, and disordered phenylalanine side chain on the Gly-Gly-Gly loop (Phe304 in E. coli FabH) (Qiu et al., 2001b). These changes occur in flexible regions that are crucial for ligand binding and formation of the active site, but “tetragonal FabH” still crystallizes as a dimer with near identical a and b axes and a 3 Å shorter c axis compared with E. coli FabH-CoA complexes that did not have these dramatic changes and were crystallized using the same crystallization condition (Qiu et al., 2001b). These differences suggest a ligand could bind FabH and cause substantial disruption to the protein without compromising the dimeric oligomerization state. These putative ligand-induced conformational changes may reflect a mode of biochemical regulation for capped FabHs.

In conclusion, the crystal structures of BsFabHA and BsFabHB are highly similar. The BsFabHA•CoA binary complex shows BsFabHA shows the prototypical features of a FabH•CoA complex. The FabH cap is a distinguishing feature differentiating BsFabHB from BsFabHA and may influence enzyme sensitivity to natural inhibitors. Complex structures of FabH•inhibitor using natural products that differentially inhibit FabH will clarify the role of unique features (e.g., cap, substrate tunnel branching) in rendering a FabH homolog sensitive or resistant. This insight will help guide the creation of narrow spectrum FabH-selective inhibitors that minimize collateral damage to a microbial community (e.g., microbiome).

Highlights.

• Bacillus subtilis FabHA and FabHB vary in resistance to natural product inhibitors

• B. subtilis FabHA resembles Staphylococcal FabH

• B. subtilis FabHB resembles Escherichia coli FabH

Abbreviations

FabH

β-Ketoacyl-ACP synthase III

FASII

bacterial fatty acid synthesis

CoA

coenzyme A

ACP

acyl carrier protein