Identification, Substrate Specificity, and Inhibition of the Streptococcus pneumoniae (cid:1) -Ketoacyl-Acyl Carrier Protein Synthase III (FabH)*

In the bacterial type II fatty acid synthase system, (cid:1) -ketoacyl-acyl carrier protein (ACP) synthase III (FabH) catalyzes the condensation of acetyl-CoA with malonyl-ACP. We have identified, expressed, and characterized the Streptococcus pneumoniae homologue of Escherichia coli FabH. S. pneumoniae FabH is (cid:1) 41, 39, and 38% identical in amino acid sequence to Bacillus subtilis , E. coli, and Hemophilus influenzae FabH, respectively. The His-Asn-Cys catalytic triad present in other FabH molecules is conserved in S. pneumoniae FabH. The apparent K m values for acetyl-CoA and mal- onyl-ACP were determined to be 40.3 and 18.6 (cid:2) M , respectively. Purified S. pneumoniae FabH preferentially utilized straight short-chain CoA primers. Similar to E. coli FabH, S. pneumoniae FabH was weakly inhibited by thiolactomycin. In contrast, inhibition of S. data analyzed for secondary structure prediction by Softsec for Windows Version 1.2 software provided by the manufacturer (Jasco). Guanidine hydrochloride (GdnHCl) denaturation was carried out while monitoring the secondary structure of FabH by CD at 220 Purified FabH was diluted into denaturant to a 0.2 mg/ml final concentration at 20 °C and incubated for min taking the measurements. Dynamic light scattering analysis was performed on purified protein (1–6 mg/ml) using a DynaPro-MSTC instru- ment (Protein Solutions, Inc.) as described previously (22).

Fatty acid biosynthesis in bacteria, plants and animals is carried out by the ubiquitous fatty acid synthase (FAS) 1 system. In the type I system of animals, including humans, FAS is a homodimer of two large polypeptides, each comprised of sev-eral distinct enzyme domains and an integral acyl carrier protein (ACP) (1,2). In the type II systems of bacteria (3), plants (4), and protozoa (5), the FAS components, including the ACP, exist as discrete proteins. The corresponding enzymes of the type I and II FASs are related in structure and function but generally lack overall sequence homology. The essentiality of type II FAS for bacterial viability together with major differences between it and type I FAS suggest that broad-spectrum anti-bacterial drugs may be obtained by screening for inhibitors of the bacterial components (6 -8).
In the type II FAS system, ␤-ketoacyl-ACP synthase (KAS) enzymes are central to the initiation and elongation steps and play a pivotal role in the regulation of the entire pathway (9). KAS I (FabB), II (FabF), and III (FabH) catalyze the condensation of malonyl-ACP with either acetyl-CoA (in the case of FabH) or the growing ACP-linked acyl chain to form the corresponding ␤-ketoacyl-ACP substrate for the subsequent reduction step in the elongation cycle catalyzed by FabG. FabH acts via a ping-pong mechanism (10,11) and is unique among KAS in that it utilizes acetyl-CoA as an acyl group donor whereas FabB (12) and FabF (13) both utilize acyl-ACPs as primers. Also, unlike FabB and FabF, which are sensitive to both cerulenin and thiolactomycin (TLM), FabH is insensitive to cerulenin and much less sensitive to TLM (3,12,14,15). FabH also possesses acetyl-CoA:ACP transacylase activity in vitro, albeit at a rate ϳ200-fold lower than that of the condensing activity. The physiological role of the transacylase reaction is unknown.
The crystal structure of Escherichia coli FabH was recently solved (16 -18) and shown to display a five-layered core structure that, despite the lack of overall sequence similarity, is similar to that of FabF (19) and other condensing enzymes of known structure (20,21). Consistent with our biochemical data (22), both in the presence and absence of acetyl-CoA, the structure has a quasi-2-fold symmetry (16,17). In addition, cocrystallization of FabH with acetyl-CoA provided direct evidence that Cys-112 is the catalytic nucleophile (16,17).
Interestingly, despite the similar folds, the substrate specificities for FabH from various bacterial species appear to be quite different. E. coli FabH was most selective for acetyl-CoA and inactive with longer chain acyl-CoAs (23). In contrast, Bacillus subtilis FabH displayed low activity with acetyl-CoA but was active with 4-to 8-carbon straight-chain acyl-CoA and all branched-chain acyl-CoA molecules (24), whereas the order of reactivity of Streptomyces glaucescens FabH was butyryl-CoA3acetyl-CoA3isobutyryl-CoA (25). Recently, Choi et al. (26) demonstrated that the FabH homolog of Mycobacterium tuberculosis prefers long-chain (C10-C16) acyl-CoA substrates over either short-or branched-chain primers, thus distinguish-ing it from the other FabH molecules.
The role of FabH in fatty acid biosynthesis in Streptococcus pneumoniae, a clinically relevant Gram-positive bacterium, has been poorly understood. We have recently identified the S. pneumoniae homologue of E. coli FabH, and in this paper we describe its cloning and characterization. Our results show that S. pneumoniae FabH is able to utilize short (C2-C4) straightand branched-chain acyl-CoAs, although activity with the straight-chain acyl-CoA was significantly higher. In addition, our studies have identified SB418011 as an inhibitor of S. pneumoniae FabH that is significantly more potent than TLM and cerulenin. SB418011 also inhibited E. coli and Hemophilus influenzae FabH but not human FAS. The availability of purified and well characterized S. pneumoniae FabH will greatly facilitate structural and kinetic studies and the discovery of potent and broad-spectrum antibacterial inhibitors.

MATERIALS AND METHODS
Reagents-Purified human testes FAS and chicken FAS proteins used for specificity assays were obtained from Dr. Zining Wu (Glaxo SmithKline, King of Prussia, PA) and Dr. Salih Wakil (Baylor College of Medicine), respectively. E. coli FabH was purified as described before (22). Hexahistidine (His 6 )-tagged H. influenzae FabH was expressed and purified from the E. coli cell lysate 2 Reagents otherwise noted were purchased from Sigma.
Cloning of S. pneumoniae fabH in E. coli-The S. pneumoniae fabH gene was cloned via polymerase chain reaction (PCR) taking advantage of the conserved genomic structure between S. pneumoniae and Streptococcus pyogenes. Inspection of the publicly available S. pyogenes sequence revealed the presence of a contig containing the N-terminal coding region of the S. pyogenes fabH gene along with considerable upstream sequence. This contig was used as a query to search the S. pneumoniae sequence database, and a S. pneumoniae contig that was likely to lie upstream of fabH was identified. The primers SPYOH1 and SPYOH2 (Table I) were designed from the upstream contig that shared homology with the S. pyogenes contig whereas the primers EKG8 and EKG10 were designed from the already available partial fabH gene sequence. PCR of S. pneumoniae genomic DNA using all four possible primer combinations yielded products around the expected size of 1.0 to 1.2 kilobase pairs. Sequencing of the PCR products confirmed the presence of the N terminus of fabH. In this way, two of the S. pneumoniae contigs were joined, and the full-length sequence of fabH was obtained.
Expression and Overproduction of S. pneumoniae FabH-A plasmid directing the overproduction of S. pneumoniae FabH with a His 10 tag (decahistidine tag) and thrombin cleavage site (LVPRGS) at its N terminus was constructed. Primers fabH5Ј and fabH3Ј (Table I) were used to amplify fabH from S. pneumoniae genomic DNA. After purification from free primers, the PCR product was digested with NdeI and BamHI and gel-purified. The gel-purified product was then ligated with gel-purified pET16b digested with NdeI and BamHI. Candidate plasmids containing correctly sized inserts were screened for over-production of S. pneumoniae FabH by SDS-PAGE. The inserts of plasmids shown to produce FabH were sequenced to ensure that no mutations were introduced by the PCR process. To obtain soluble protein, E. coli strain BL21 (DE3) was transformed with vector pFabH1 containing the fabH gene. Transformed cells were grown at 37°C to an optical density (A 600 nm ) of 1.0 and then were induced with 0.5 mM isopropyl ␤-Dthiogalactopyranoside for additional 2.5 h (optical density of 2.6). Cells were harvested by centrifugation at 15,000 ϫ g for 30 min, and the resultant cell pellet was stored at Ϫ70°C until further use.
Purification of S. pneumoniae FabH-All lysis and purification steps were carried out at 4°C. 10 g of E. coli cells over-expressing tagged S. pneumoniae FabH were resuspended in 300 ml of lysis buffer (buffer A) containing 20 mM Tris-HCl (pH 8.0), 300 mM sodium chloride, 0.2 mM phenylmethylsulfonyl fluoride, 40 mM imidazole, and 5 mM 2-mercaptoethanol. Cells were lysed twice at 10,000 p.s.i. using a microfluidizer (Microfluidics Corporation). Cell debris were removed by centrifugation (Sorvall RC-5B) at 35,000 ϫ g for 30 min. The supernatant was applied to a 15-ml Ni-NTA metal affinity column (Qiagen) at 2 ml/min equilibrated in buffer A. The column was washed with 10-column-volumes of buffer A and eluted with 20 mM Tris-HCl (pH 8.0), 300 mM sodium chloride, 0.2 mM phenylmethylsulfonyl fluoride, 5 mM 2-mercaptoethanol, and 300 mM imidazole. Fractions containing FabH were pooled, dialyzed against 4 liters of buffer B containing 20 mM Tris-HCl (pH 8.0) and 5 mM 2-mercaptoethanol, and applied to a 15-ml Source Q column (Amersham Pharmacia Biotech) column equilibrated in buffer B. The column was washed with 10-column-volumes of buffer B and then eluted with a 20-column-volume linear gradient of 0 -1.0 M sodium chloride in buffer B. FabH, which eluted at 300 -400 mM NaCl, was next applied to a Superdex 200 size exclusion column (2.6 ϫ 60 cm, Amersham Pharmacia Biotech) equilibrated in 20 mM Tris-HCl (pH 7.5), 50 mM NaCl, and 5 mM DTT. Yield of purified protein was ϳ10 mg/gm wet weight of E. coli. SDS-PAGE, native isoelectric focusing (IEF), N-terminal sequence, and amino acid composition analyses were performed as described (22).
Size pneumoniae FabH. ACP, malonyl-CoA, and FabD were preincubated with buffer and DTT for 10 min at 33°C prior to addition of acetyl-CoA and FabH. FabH was added last to start the reaction, and reactions were then incubated at 33°C for 10 -15 min. 10% trichloroacetic acid was added to stop the reaction, and stopped reactions were incubated on ice for 20 min prior to filtration. Reaction mixtures were filtered and washed twice with 10% trichloroacetic acid on Packard GF/C filter plates using a Packard Filtermate harvester. The filter plates were then dried completely at 60°C, and the radioactivity was quantified using Wallac Supermix scintillation mixture and a Wallac Microbeta 1450 liquid scintillation counter.
Malonyl-ACP formation was determined in the filtration assay format as described above for FabH. To assay different non-radiolabeled acyl-CoAs as primers for FabH, a spectrophotometric assay was employed. ␣-Ketoglutarate dehydrogenase has been used previously to specifically quantify free CoA (27) and was used as coupling enzyme to follow FabH activity.  IC 50 values against human FAS were determined spectrophotometrically by monitoring the oxidation of NADPH to NADP at 340 nm. The reaction mixture contained 50 mM potassium phosphate (pH 6.5), 4 mM DTT, 11 M acetyl-CoA, 190 M NADPH, and 10 nM human FAS. The reaction was started by the addition of 75 M malonyl-CoA and incubated for 20 min at 25°C.
Circular Dichroism (CD) Structure and Stability Analysis-Far-UV CD spectra were recorded on a Jasco J-710 CD instrument using a 0.1-cm path length cuvette cell. Data (seven accumulations) were collected using a time constant of 4 s, at 50 nm/min, and with a 1-nm spectral bandwidth. A buffer-only spectrum was subtracted from the FabH spectrum, and the resulting ellipticity data, (in degrees), were converted to difference molar extinction coefficient data as described (28). Concentration of residues was determined by absorbance at 210 nm using an extinction coefficient of 22 ml/mg and an average molecular mass of 110/residue. The data were analyzed for secondary structure prediction by Softsec for Windows Version 1.2 software provided by the manufacturer (Jasco). Guanidine hydrochloride (GdnHCl) denaturation was carried out while monitoring the secondary structure of FabH by CD at 220 nm. Purified FabH was diluted into denaturant to a 0.2 mg/ml final concentration at 20°C and incubated for 10 min before taking the measurements. Dynamic light scattering analysis was performed on purified protein (1-6 mg/ml) using a DynaPro-MSTC instrument (Protein Solutions, Inc.) as described previously (22).

RESULTS AND DISCUSSION
Sequence Analysis of the fabH Gene and Its Product-The S. pneumoniae fabH gene was cloned via PCR taking advantage of conserved genomic structure between S. pneumoniae and S. pyogenes as described under "Materials and Methods." PCR of S. pneumoniae genomic DNA using all four possible primer combinations yielded products around the expected size of 1.0 -1.2 kilobase pairs. Sequencing of the PCR products confirmed the presence of the N terminus of fabH. In this way two S. pneumoniae contigs were joined and the full-length sequence of fabH was obtained. The open reading frame encodes a protein of 324 amino acids (Fig. 1). The calculated molecular mass of the encoded protein is 34,901 Da, and the isoelectric point is 4.9.
The alignment in Fig. 1 was manually constructed with the aid of the sequence homology search algorithm PSI-BLAST. A FabH sequence profile was generated via five iterations of PSI-BLAST using the S. pneumoniae amino acid sequence as a query against an all-bacteria protein database. The profile was then used to search a Homo sapiens-specific protein database. The alignment was optimized by manual inspection using the x-ray crystal structure of E. coli FabH (16) and the predicted secondary structure of the eukaryotic FAS-␤-ketoacyl synthase domain. The species shown are a subset of the alignment containing 43 members of the FabH family representing 38 different species including green plants and Plasmodium falciparum.
With the exception of Phe-157, the hydrophobic residues that make up the active site pocket in the E. coli structure, including Phe-87Ј that protrudes into the active site of the opposite dimer-pair, are not rigorously conserved in S. pneumoniae (Leu-142, Phe-157, Leu-189, Leu-205, and Phe-87Ј). The E. coli FabH residues, Trp-32 and Arg-151, are invariant within the FabH family, presumably preserving their side-chain interactions with the adenine ring of CoA as suggested in E. coli but different in human and chicken FAS. In addition, the residues that form hydrogen bonds to pantotheinate, Gly-209 and Asn-247, are invariant in the FabH family but again are different in eukaryotic FAS.
Although the high degree of sequence conservation within the bacterial FabH protein family easily distinguishes this member of the ␤-ketoacyl synthase superfamily from FabF and the eukaryotic fatty acid synthases, distant similarities can be found using sensitive search algorithms such as PSI-BLAST. Structural similarities between FabF and FabH are easily recognizable and, from this analysis, one would predict a similar structure for the corresponding domain in eukaryotic FAS.
Purification and Characterization of FabH S. pneumoniae-FabH was expressed in E. coli as an N-terminal decahistidinetagged fusion protein. FabH expressed well and was purified from E. coli cell lysate using a combination of Ni-NTA, anion exchange, and preparative Superdex 200 size exclusion chro-matography as described under "Materials and Methods." The overall yield of the purified FabH was ϳ10 mg/gm (wet weight) of E. coli cells.
Purified FabH was nearly homogeneous as judged by SDS-PAGE ( Fig. 2A). N-terminal sequence analysis confirmed the 34-kDa band as N-terminal decahistidine-tagged S. pneumoniae FabH. The amino acid composition was in complete   (Fig. 2B), which is in close agreement to its theoretical pI of 6.02 calculated from the amino acid sequence of decahistidine-tagged S. pneumoniae FabH.
Our previous studies showed that the molecular mass of E. coli FabH by size exclusion chromatography and dynamic light scattering performed under nondenaturing conditions was ϳ65 kDa, implying the non-covalent dimeric nature of the protein (22). In agreement with the solution studies, the crystal structure of the E. coli FabH in the presence and absence of acetyl-CoA revealed a non-covalent homodimer (16,17). The homodimeric nature of the E. coli FabH was the same as that reported for the E. coli FabB (29) and FabF (19,29) proteins and for the FabH of S. glaucescens (25).
To determine whether S. pneumoniae FabH has similar properties, purified protein was subjected to size exclusion chromatography and dynamic light scattering techniques. On an analytical Superdex S200 column S. pneumoniae FabH eluted as a single species with a molecular mass of ϳ60 kDa (Fig. 2C); similar results were obtained using a Superdex S75 column (not shown). The size exclusion chromatography profile and SDS-PAGE analysis indicate a non-covalent homodimeric structure for S. pneumoniae FabH. S. pneumoniae FabH was monodisperse at lower concentration as judged by dynamic light scattering (not shown). However, unlike the E. coli FabH, at higher concentrations (Ͼ3 mg/ml), the S. pneumoniae FabH appeared to form soluble aggregates as evident from an increased polydispersity index (not shown). Similar results were obtained with purified untagged S. pneumoniae FabH indicating that the decahistidine tag was not the cause of the aggregation. 4 It may be that the hydrophobic residues of S. pneumoniae FabH that are exposed to solvent are responsible for its aggregation. Further work, including modeling and site-directed mutagenesis, may help to determine whether the aggregation is an intrinsic property of S. pneumoniae FabH.
CD Spectroscopy-Our previous circular dichroism spectroscopic studies showed the presence of both ␣-helix and ␤-sheet elements in purified E. coli FabH (22). Consistent with these results, the crystal structure of E. coli FabH displayed a fivelayered core structure, ␣-␤-␣-␤-␣, where each ␣ comprises two ␣-helices and each ␤ consists of a five-stranded, mixed ␤-sheet (16,17). To determine its secondary structure and stability, S. pneumoniae FabH was subjected to CD spectroscopy (Fig. 3). The far-UV circular dichroism spectrum of S. pneumoniae FabH is shown as difference molar extinction coefficient/mole of peptide bond versus wavelength (Fig. 3A). Softsec computer software analysis predicted a ␣-helix content of 26% and a ␤-sheet content of 31%. These values are comparable with those obtained for E. coli FabH based on its crystal structure (39% ␣-helix, 28% ␤-sheet; Ref. 16) and indicate the presence of a similar core structure despite the overall low homology between these two enzymes. To determine its stability, GdnHCl denaturation was carried out, monitoring the secondary structure of S. pneumoniae FabH using the CD signal at 220 nm as indicator. Purified FabH was diluted into denaturant to a 0.2 mg/ml final concentration at 20°C and incubated for 10 min before taking measurements. The denaturation midpoint (1.3 M GdnHCl), m value (1.14 kcal/mol) and ⌬G of unfolding (1.53 kcal/mol) revealed that purified S. pneumoniae FabH is moderately stable (Fig. 3B) and unfolds in a manner consistent with a single cooperative unfolding transition. In contrast, the 4 C. Silverman, S. Khandekar, unpublished data.  Enzyme Kinetics-The kinetic constants for acetoacetyl-ACP synthase activity were determined using the filtration assays described under "Materials and Methods." The malonyl-ACP concentration in the FabH filtration assay was estimated to be 25% of the total ACP added based on the percent conversion calculated from the FabD assay alone. Using this method the apparent K m values for acetyl-CoA and malonyl-ACP were determined to be 40.3 Ϯ 2.3 M and 18.6 Ϯ 1.5 M, respectively. The maximum velocity of the reaction was 7.1 nmoles/min/g. The affinity of acetyl-CoA for S. pneumoniae FabH was nearly identical to that determined for E. coli FabH (23). In contrast, S. glaucescens FabH exhibited higher affinity for acetyl-CoA, with K m of 2.4 M.
Substrate Specificity-FabH species purified from Gramnegative and Gram-positive bacteria, despite their overall similar catalytic mechanism, have displayed significantly different substrate specificities. To determine substrate specificity of Gram-positive S. pneumoniae FabH, purified protein was assayed spectrophotometrically as described under "Materials and Methods," and the results are summarized in Table II. Purified S. pneumoniae FabH was able to utilize short straightand branched-chain acyl-CoAs. Of the CoA substrates examined, the relative order of activity was determined to be acetyl3butyryl3isobutyryl3isovaleryl. Longer-chain substrate such as, lauroyl-CoA and palmitoyl-CoA, showed no detectable activity in this assay.
The substrate specificity of various FabH molecules are summarized in Table III. E. coli FabH preferred short-chain acyl-CoA primers, among which acetyl-CoA was the most preferred substrate (23,24). The enzyme was inactive with longer-chain (longer than C4) primers as well as all branched-chain CoA primers. The smaller binding pocket observed in the crystal structures of E. coli FabH (16,17) and FabH-CoA complex (16) support this finding. Additionally, observing CoA in the crystal indicated that the reaction product has a notable affinity for FabH (16). The B. subtilis FabH homologs, in contrast, were active with 4-to 8-carbon straight-chain acyl-CoAs and all branched-chain acyl-CoAs (24). Similarly, the S. glaucescens FabH efficiently utilized both butyryl-CoA and isobutyryl-CoA (25). Interestingly, FabH of M. tuberculosis (26) preferred only long-chain (C8-C20) acyl-CoA substrates indicating that its substrate-binding pocket is significantly different from those of other FabH molecules. Based on our results and those summarized in Table III, the binding site pocket of S. pneumoniae FabH appears to be more similar to those of S. glaucescens and B. subtilis FabH than those of E. coli and M. tuberculosis FabH molecules. Availability of a FabH crystal structure from a Gram-positive organism will be helpful in comparing the substrate binding pockets to account for the differences in the substrate specificity of various primers.
The fatty acid composition of bacteria appear to vary from species to species. For example, Gram-negative E. coli, in which FabH cannot use branched-chain acyl-CoA primers, does not produce branched-chain fatty acids (30). In contrast, Grampositive bacilli and streptomycetes, whose FabH components are capable of efficiently utilizing branched-chain acyl-CoAs as substrates, produce significant amounts of odd and even carbon-number branched-chain fatty acids (24,31). Interestingly, S. pneumoniae, also a Gram-positive organism, does not produce significant amounts of branched-chain fatty acid structures (30,31). In agreement with the fatty acid composition of S. pneumoniae, the activity of S. pneumoniae FabH with the branched-chain CoAs was significantly weaker than that with the straight-chain CoAs (Table II). Thus, as surmised by Choi et al. (24), substrate specificity of the FabH enzymes appears to be the determining factor in the biosynthesis of branched-or straight-chain fatty acids by type II fatty acid synthase.
To extend these studies to purified S. pneumoniae FabH and its E. coli and H. influenzae FabH homologues, conditions were set at K m concentrations for the substrates as described under "Materials and Methods." The structures of the compounds tested are shown in Fig. 4 and the IC 50 values are summarized in Table IV. In agreement with the earlier reports (14,34), E. coli FabH in our studies was sensitive, albeit weakly, to TLM with an IC 50 of 32.0 Ϯ 12.0 M, and likewise to those studies, cerulenin did not inhibit E. coli FabH in our hands (Table IV). In addition, similar to E. coli FabH, H. influenzae and S.