A fatty acid-binding protein of Streptococcus pneumoniae facilitates the acquisition of host polyunsaturated fatty acids

Streptococcus pneumoniae is responsible for the majority of pneumonia, motivating ongoing searches for insights into its physiology that could enable new treatments. S. pneumoniae responds to exogenous fatty acids by suppressing its de novo biosynthetic pathway and exclusively utilizing extracellular fatty acids for membrane phospholipid synthesis. The first step in exogenous fatty acid assimilation is phosphorylation by fatty acid kinase (FakA), whereas bound by a fatty acid-binding protein (FakB). Staphylococcus aureus has two binding proteins, whereas S. pneumoniae expresses three. The functions of these binding proteins were not clear. We determined the SpFakB1- and SpFakB2-binding proteins were bioinformatically related to the two binding proteins of Staphylococcus aureus, and biochemical and X-ray crystallographic analysis showed that SpFakB1 selectively bound saturates, whereas SpFakB2 allows the activation of monounsaturates akin to their S. aureus counterparts. The distinct SpFakB3 enables the utilization of polyunsaturates. The SpFakB3 crystal structure in complex with linoleic acid reveals an expanded fatty acid-binding pocket within the hydrophobic interior of SpFakB3 that explains its ability to accommodate multiple cis double bonds. SpFakB3 also utilizes a different hydrogen bond network than other FakBs to anchor the fatty acid carbonyl and stabilize the protein. S. pneumoniae strain JMG1 (ΔfakB3) was deficient in incorporation of linoleate from human serum verifying the role of FakB3 in this process. Thus, the multiple FakBs of S. pneumoniae permit the utilization of the entire spectrum of mammalian fatty acid structures to construct its membrane.

The ability to acquire exogenous fatty acids (FA) 2 for membrane phospholipid synthesis is a universal feature of lipid metabolism in Firmicutes, a phylum of Gram-positive bacteria that contains many important human pathogens. The first step in FA incorporation is activation by FA kinase, an enzyme system consisting of a kinase domain protein (FakA) that phosphorylates a FA carried by the FA-binding protein component (FakB) (1). The resulting acyl-phosphate (FAϳP) bound to FakB is either transferred to the glycerol phosphate acyltransferase (PlsY) to initiate phospholipid synthesis or transferred to acyl carrier protein (ACP) by PlsX (2). The resulting acyl-ACP is either utilized by PlsC to acylate the 2-position of 1-acylglycerolphosphate or it may enter the FASII cycle and be elongated. The elongated acyl-ACP is then metabolized via PlsX/PlsY or PlsC like the acyl-ACP derived from de novo biosynthesis. Two major orders of the Firmicutes have distinctly different physiological responses to the presence of extracellular FA. The Bacillales, exemplified by Staphylococcus aureus, do not genetically suppress the genes of type II FA biosynthesis (FASII) in the presence of exogenous FA (3). Rather, FASII continues to produce primarily anteiso15:0 that is placed in the 2-position, and the exogenous long-chain saturated and monounsaturated FA are activated by FA kinase and placed into the 1-position (4,5). The Lactobacillales, exemplified by Streptococcus pneumoniae, have a different response. These organisms strongly suppress the expression of the genes encoding the FASII enzymes in response to exogenous FA. This response is mediated by the FabT transcriptional repressor (6 -9). FabT bound to acyl-ACP tightly binds to promoters within the FASII gene cluster to potently suppress FASII gene expression (10). Thus, these organisms construct their membrane phospholipids almost exclusively using FA obtained from the environment to acylate both positions of the glycerol-phosphate backbone (3,11). Elongation is not a significant fate for exogenous FA in the Lactobacillales because the FabT system suppresses the transcription of the entire gene set responsible for the FASII elongation cycle.
The ability of FA kinase to activate FA encountered in the environment is conferred by the FA selectivities of the FakB component. S. aureus expresses two FakBs (1,5). SaFakB1 is specific for saturated FA (16:0), whereas SaFakB2 selectively binds monounsaturated FA (18:1) (1,5,12). S. aureus does not synthesize unsaturated FA, so SaFakB1 is the housekeeping protein and SaFakB2 is responsible for the uptake of host monounsaturated FA (1,5,12). S. pneumoniae encodes three This work was supported by National Institutes of Health Grant GM034496, Cancer Center Support Grant CA21765, and the American Lebanese Syrian Associated Charities. The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The atomic coordinates and structure factors ( cro ARTICLE FakB-binding proteins suggesting an expanded repertoire of FA-binding capabilities to enable the acquisition of environmental FA to replace de novo biosynthesis. Here, we report that two of these binding proteins, SpFakB1 and SpFakB2, are bioinformatically related to the two FakBs previously characterized in S. aureus (5,12). Heterologous expression, biochemical analyses, and X-ray crystallography confirmed that SpFakB1 specifically bound saturated FA (16:0) and SpFakB2 selectively bound monounsaturated FA (18:1) but also bound saturated FA. Fatty acids are designated by number of carbons:number of double bonds. SpFakB3 is more distantly related to the S. aureus homologs, and heterologous expression and biochemical analyses showed that SpFakB3 expanded the repertoire of substrates to polyunsaturated FA (18:2). The SpFakB3(18:2) structure revealed a different hydrogen bond network anchoring the FA carbonyl and an expanded FA-binding pocket within the hydrophobic interior of the protein that explains its ability to bind FA with multiple cis double bonds. S. pneumoniae strain JMG1 (⌬fakB3) was deficient in 18:2 incorporation verifying the role of SpFakB3 in the incorporation of polyunsaturated FA into S. pneumoniae phospholipids. Thus, the multiple FakBs of S. pneumoniae permit the acquisition and utilization of the entire spectrum of mamma-lian FA structures to facilitate the acquisition of host FA to construct its membrane at the infection site.

Incorporation of FA from human serum
The distinct difference between S. aureus and S. pneumoniae utilization of exogenous FA is illustrated by the effect of human serum supplementation on the molecular species composition of PG, a major membrane phospholipid in both organisms (Fig.  1). S. pneumoniae FASII produced both saturated and monounsaturated FA when grown in CϩY medium without FA supplement (3) (Fig. 1A). The two most abundant PG molecular species had 16:0 paired with either 16:1 or 18:1. S. pneumoniae is like most organisms in that it places saturated FA in the 1-position and unsaturated FA in the 2-position of its membrane phospholipids (13). S. aureus places anteiso15:0 in the 2-position of the glycerol backbone (3) giving rise to S. aureus PG molecular species with only saturated, straight, and branched-chain FA paired with anteiso15:0 when grown in media without exogenous FA (Fig. 1B). These data illustrate the difference in membrane phospholipid structures produced by de novo biosynthesis in these two Gram-positive pathogens. S. pneumoniae strain TIGR4 and S. aureus strain AH1263 were grown in their respective medium or in medium:human serum (1/1, v/v). The PG molecular species were determined under each of the four culture conditions in triplicate, and representative spectra are shown. The molecular species that contain an 18:2 or 18:2-derived FA are highlighted in red. A, S. pneumoniae grown in CϩY media. B, S. aureus grown in Luria broth. C, S. pneumoniae grown with 50% human serum, 50% CϩY medium. The presence of 18:2 in the m/z ϭ 745 peak was verified by fragmentation (inset). D, S. aureus grown with 50% human serum, 50% Luria broth. E, major FA of human serum. Lipids were extracted from triplicate samples of human serum and the total FA composition determined by GC of the derived methyl esters. The weight percent of each FA was calculated. FA less than 1% of the total are not shown. F, triplicate biological replicates were obtained, the areas under each peak in the spectra were summed, and the PG molecular species containing 18:2 as a percent of the total area were calculated to provide an estimate of 18:2 incorporation. Peaks containing elongation products, like 20:2 derived from 18:2, were included in the calculation for the contribution of the parent FA.

Structure and function of S. pneumoniae FakBs
Both S. pneumoniae and S. aureus incorporated FA derived from serum lipids when cultured in the presence of human serum. The S. pneumoniae PG molecular species pattern when grown in human serum was different from the pattern obtained in the absence of an exogenous lipid supplement (Fig. 1C). The most abundant peak, m/z ϭ 747, corresponded to a 16:0/18:1 in both growth conditions, and in the presence of human serum the 16:0/16:1 peak was less prominent. This pattern is consistent with the observation that S. pneumoniae shuts down de novo lipid synthesis when supplied with an exogenous source of FA (3). Peaks in the spectrum arising from the synthesis of molecular species containing 18:2 are highlighted in red (Fig.  1C). A major new peak (m/z ϭ 745) was fragmented to verify that 16:0/18:2 was the most abundant molecular species at this mass position (Fig. 1C, inset). S. pneumoniae cannot make 18:2, thus this FA must be derived from the human serum. The origin of the saturated and monounsaturated FA in PG cannot be unambiguously determined in this experiment because these FA are synthesized by S. pneumoniae and are present in human serum. The analysis of S. aureus grown with human serum showed that most PG molecular species still contained 15:0, but that 18:1 along with its elongation product 20:1 were incorporated from the serum (Fig. 1D). Human serum contains nonesterified FA plus an abundance of glycerolipids, and both S. aureus (geh and lip) (14) and S. pneumoniae (lipA) (15) express genes capable of hydrolyzing these lipids to release FA. The three most abundant FA in the human serum sample used in our study were 16:0, 18:1, and 18:2 (Fig. 1E). Although 18:1 and 18:2 were about equal in abundance, S. aureus incorporated significantly more 18:1 than 18:2 (Fig. 1D). The incorporation of 18:2 into PG was significantly less in S. aureus compared with the more robust incorporation of 18:2 into the S. pneumoniae lipidome (Fig. 1F). These data indicated that S. pneumoniae had a higher capacity to extract polyunsaturated FA from human serum than S. aureus.

Three fakB genes in S. pneumoniae
S. aureus expresses two FakB proteins, one specific for saturated FA and the other specific for monounsaturated FA (5). The fact that S. pneumoniae incorporates 18:2 into its phospholipids suggested that it had a FA-binding protein capable of binding polyunsaturated FA. We identified a single fakA gene (locus tag: Sp0443) and 3 fakB genes (locus tags: Sp1557, Sp1112, Sp0742) in the S. pneumoniae TIGR4 genome. The protein sequences were compared using BLASTP alignment software using the KEGG database sequence entries. The Sp1557 gene was named fakB1 because it encoded a protein that was 35% identical to SaFakB1 and 29% identical to SaFakB2. The Sp1112 gene was named fakB2 because it encoded a protein with 29% identity to SaFakB2 and 28% identity to SaFakB1. The Sp0742 gene encoded a protein that was the least similar to the S. aureus FakBs (18% to SaFakB1 and 26% to SaFakB2), and it was designated fakB3. Both S. pneumoniae and S. aureus are in the phylum Firmicutes and class Bacilli, but S. pneumoniae is in the Order Lactobacillales and S. aureus in the Order Bacillales. Analysis of both Orders using the KEGG database showed that the individual members of both Orders encode between 2 and 4 fakB genes.

FA selectivity of SpFakBs in cells
An experimental system to evaluate the FA selectivities of FA-binding proteins was developed using S. aureus strain JLB31 (⌬fakB1 ⌬fakB2) that was unable to incorporate exogenous FA because it lacks FA-binding proteins but expresses the SaFakA kinase domain protein. A series of plasmids were constructed that expressed the individual FakBs governed by the sarA promoter ( Table 1). The series of JLB31 strains each harboring a plasmid expressing a particular FakB were grown to an A 600 of 0.5 and then a mixture of [d 4 ]16:0, -18:1, and -18:2 (10 M each) was added to the cultures. After a 30-min incubation, the extent of each FA incorporated into PG was determined by MS ( Fig. 2A). The assay was validated using SaFakBs of known FA specificity. The expression of SaFakB1 resulted in robust incorporation of [d 4 ]16:0, whereas the incorporation of the unsaturated FA in the mixture was not detected. In the strain expressing SaFakB2, 18:1 was the most highly incorporated FA, although lower amounts of both [d 4 ]16:0 and -18:2 were detected. Strain JLB31/pCS119 (empty vector) did not express a FakB and did not incorporate any exogenous FA.

FA utilization in a ⌬fakB3 mutant
The physiological role of fakB3 in S. pneumoniae FA uptake was evaluated by constructing a ⌬fakB3 knockout strain. Strain JMG1 (⌬fakB3) was constructed from strain TIGR4 by allelic replacement with a cassette encoding kanamycin resistance (Fig. 5A). PCR analysis showed the presence of the knockout allele and the absence of the fakB3 gene in strain JMG1. The PG molecular species of strain JMG1 (⌬fakB3) was indistinguishable from TIGR4 (WT) when grown in CϩY (Figs. 1A and 5B) but TIGR4 incorporated more 18:2 than strain JMG1 (⌬fakB3) when grown in human serum (Figs. 1C and 5C). The incorpo-ration of 18:2 was restored to WT levels in the complemented strain JMG1/pSpFakB3 (Fig. 5D). These data validate a role for SpFakB3 in the utilization of exogenous 18:2 for membrane lipid synthesis in S. pneumoniae.

Crystal structures of SpFakBs
The basis for the FA preferences of the SpFakB proteins was revealed by the crystal structures of the three S. pneumoniae FA-binding proteins in complex with their respective FA ligands ( Table 2). The SpFakBs all have a two-domain proteinfold very similar to the other FakB protein family members that have been described (5,12,16,17) (Fig. 6). The N-terminal domain consists of an EDD-fold fused to a 3-stranded antiparallel ␤-sheet and one ␣-helix. The carboxyl-terminal domain is a six-stranded ␤-sheet flanked by two ␣-helices on one side and three on the other. The surface arginine residue highlighted in Fig. 6 is conserved and required for high affinity binding to the FakA component of FA kinase (12). Its presence in all FakB structures explains why SpFakBs function interchangeably with SaFakA ( Fig. 2A), and identifies the FakA-FakB binding locale. Furthermore, the overall surface charge distribution is similar in all three SpFakBs (Fig. 6). The conservation of the FakB structure across isoforms and species accounts for their ability to interchangeably interact with their three protein partners: FakA, PlsX, and PlsY (5).

The SpFakB FA-binding tunnels
The unique distinguishing feature of the FakBs is the size and shape of the hydrophobic FA-binding tunnel located in the protein's interior that defines the FA binding selectivity. The  (Fig. 7A), and there is an overall root mean square deviation of 1.45 Å between the two aligned structures. The gently curved shapes of the two FA-binding pockets and the residues used to construct the pocket are similar in the two proteins, and in both cases, the FA carboxyl is fixed by a hydrogen bond network at the mouth of the tunnel. However, a notable difference in SpFakB1 is a kink in the FA chain at carbon-2 that is not present in SaFakB1, which displaces the acyl chains by 1.2 Å at this position. The equivalent residues SpIle28 -SaLeu28 and SpVal264 -SaVal266 are displaced by 1.6 and 1.7 Å, respectively, to accommodate this kink at carbon-2 in the SpFakB1(16:0) structure. The kink generates additional changes in the conformation of the FA in SpFakB1(16:0) compared with SaFakB1(16:0), and these are accommodated by other displacements between aligned equivalent residues in the tunnels: SaGly277-SpGly277, 0.9 Å; SaIle124 -SpPro123, 0.7 Å; SaLeu120 -SpIle119, 1.2 Å; and SaAla158 -SpAla157, 0.6 Å. The SpFakB2(18:1⌬9) structure is also very similar to the SaFakB2(18:1⌬9) structure ( Fig. 7B) with an overall root mean square deviation of 1.27 Å between the two aligned structures. Despite the low level of sequence identity between the two proteins (29%), the amino acids that create their FA tunnels both generate the same distinctive kinked cavities that are required to accommodate the cis double bond in the middle of the acyl chain. Similar to FakB1, the FA carboxyl is fixed by a hydrogen bond network at the entrance of the tunnel that comprises highly conserved and closely aligned residues in SpFakB2 and The bound 18:1 adopts very similar conformations in SpFakB2 and SaFakB2 but begins to diverge at carbon-13 and is separated by 0.5 Å at the terminal methyl. Accordingly, equivalent residues that surround the distal end of the FA are more displaced than those at the proximal end: The SpFakB3(18:2(⌬9⌬12)) complex structure reveals a distinctive gourd-like FA-binding pocket that highlights how the interiors of these proteins have evolved to perfectly accommodate FA of different structures (Fig. 8A). Like FakB1 and FakB2, the proximal end of the FakB3 tunnel is relatively narrow and straight to accommodate the first 7 carbons of the acyl chain. However, at this point, the tunnel expands into a wide cavern min, and the contribution of each unsaturated FA to the PG molecular species was determined by MS. Triplicate biological replicates were obtained, the areas under each peak in the spectra were summed and reported as a percent of the total area, and mean Ϯ S.D. was plotted. Peaks containing elongation products, like 20:1 derived from 18:1, were included in the calculation for the contribution of the parent FA. Data are the mean Ϯ S.D. of 3 individual data sets. Statistical differences between the FA incorporated in each strain was determined using Student's t test.

Structure and function of S. pneumoniae FakBs
that allows FA with multiple cis double bonds to curl up inside the protein. The volume of the tunnel is 110.4 Å 3 , which is larger than the linear FakB1 tunnel (67.5 Å 3 ) and slightly larger than the kinked FakB2 tunnel (102 Å 3 ). Another distinguishing feature of SpFakB3 compared with FakB1 and FakB2 is that it uses a different hydrogen bond network to lock the FA carboxyl in place (Fig. 8B). The typical FakB configuration places Ser on one carbonyl oxygen and a His-Thr dyad on the other side to form a Ser-FA carbonyl-Thr-His network (12). In contrast, in SpFakB3, a tyrosine (Tyr-268) plays the role of histidine and a serine (Ser-63) replaces threonine to form a Ser-FA carbonyl-Ser-Tyr network. There are also two structured water molecules in SpFakB3 hydrogen bonded to the FA carbonyl that mediate hydrogen bond connections between the FA carbonyl and Arg-173, Gly-92, and Ser-63. Arg-173 also forms a hydrogen bond with Tyr-268 and a water-mediated connection with Asn-171.

Further insights into the SpFakB1 and SpFakB2 tunnel specificities
In SpFakB1(16:0), the terminal methyl of the 16:0 chain is packed against the protein at the end of the pocket suggesting that the bend at carbon-2 may be required for the 16:0 chain to fit into the pocket. This idea was tested by determining the structure of SpFakB1 in complex with the shorter 14:0 FA with the prediction that the carbon-2 bend would not be present. An overlay of the SpFakB1(16:0) and SpFakB1(14:0) structures shows that the two FA overlay along the entire length of the binding tunnel and that the bend at carbon-2 is also present in the 14:0 structure (Fig. 9A). SaIle232 acts as a swinging gate to increase or decrease the tunnel length in SaFakB1, but residue SpIle232 did not change orientation or position in the SpFakB1 structures loaded with either 16:0 or 14:0. In the SpFakB1(16:0) structure, the terminal FA methyl carbon is 3.7 Å away from SpIle232. There were small changes in Phe-192 side chain that slightly close the end of the tunnel in the SpFakB1(14:0) structure, but otherwise the two structures are almost identical except for the length of the acyl chain. Thus, the carbon-2 bend is an integral feature of the SpFakB1 FA complex.
Another question is how the substrate selectivity of SpFakB2 is related to its biological function(s). S. aureus does not produce unsaturated FA and thus SaFakB2 does not have a specific role in lipid homeostasis in the organism (5). However, S. pneumoniae does produce monounsaturated FA suggesting that SpFakB2 may have a role in lipid metabolism in this organism. S. pneumoniae de novo biosynthesis produces 18:1, although the primary location of the double bond is at carbon-11 instead of carbon-9 as found in mammals (6). The flexibility of SpFakB2 to bind these two monounsaturated FA was investigated by determining the structure of the SpFakB2(18:1⌬11) complex and comparing it to the structure of SpFakB2(18:1⌬9) (Fig. 9B). The binding pocket has enough space between the 9 and 12 carbons to accommodate either of the sp 2 -hybridized double bonds. The overlay of these two structures shows that they are essentially identical with regard to the size and shape of the FA-binding tunnel and the positioning of the residues that form the binding pocket. The different locations of the sp 2 double bond conformations are clearly seen in the electron density map, and the maximum divergence of the two acyl chains is only 0.7 Å in the immediate area between the locations of the double bonds. These data illustrate that SpFakB2 participates in the activation of both endogenous cis-vaccinate and exogenous oleate derived from the host.

Discussion
The biochemical and structural data lead to a model for FA metabolism in S. pneumoniae outlined in Fig. 10. The biosynthetic pathway in the absence of exogenous FA proceeds via the FASII collection of enzymes to long-chain acyl-ACP (18), which is then distributed to PlsY (via PlsX) or PlsC to produce phosphatidic acid, the precursor to all membrane glycerophospholipids (19). In the presence of exogenous FA, transcription of the FASII genes is potently suppressed by binding of the FabT-acyl-ACP repressor complex to promoters within the FASII gene cluster (6,10). The repression of FASII gene expression by exogenous FA is mediated by acyl-ACP2 (9) (the acpP2 gene that is in an operon with plsX). This regulatory loop shuts off de novo FA synthesis, and only exogenous FA are used for phospholipid synthesis (3). These exogenous FA are not significantly elongated even though they are converted to acyl-ACP because the FASII elongation system is suppressed. The three FakBs of S. pneumoniae characterized in this study participate in this FA sensing pathway by providing a mechanism to activate a spectrum of saturated, monounsaturated, and polyunsaturated FA found at the infection site for FAϳP and acyl-ACP formation. SpFakB1 has the most restricted FA binding preference and effectively excludes unsaturated FA with kinked acyl chains due to the straight shape of its FA-binding tunnel.
SpFakB2 prefers monounsaturated FA. SpFakB2 has a kinked FA-binding tunnel that accommodates both 18:1⌬11, the monounsaturated fatty acid produced by S. pneumoniae, and 18:1⌬9, a prevalent mammalian FA. Although SpFakB2 prefers to bind monounsaturated FA, it does not exclude saturated FA, which are flexible and capable of fitting into the kinked binding pocket. SpFakB2 also has a weak capacity to bind 18:2. SpFakB3 has a wider FA-binding pocket that allows the binding of polyunsaturated FA, like 18:2, with multiple cis double bonds. The gourd-shaped pocket within SpFakB3 means that it can present a variety of FA structures to FakA. The multiple FakBs of the FA kinase activation system coupled with the repression of de novo synthesis means that S. pneumoniae can efficiently and exclusively utilize the entire spectrum of mammalian FA found at the infection site for phosphatidic acid synthesis (Fig. 10).
SpFakB3 is differentiated from most FakBs by having a different hydrogen bond network that interacts with the FA carboxyl group (Fig. 9B). The majority of bacterial FA-binding proteins use a Ser-FA carboxyl-Thr-His hydrogen bond network to fix the FA carboxyl at the entrance of the FakB FA-binding tunnel (12), but in about 5% of the FakB sequences, a tyrosine is substituted at the position of the histidine residue. SpFakB3 is one of the FA-binding proteins with a Ser-FA carboxyl-Ser-Tyr hydrogen bond network. Disruption of this hydrogen bond net-  Table 1. B, representative spectra of S. pneumoniae strain JMG1 (⌬fakB3) grown in CϩY media (C) and grown in 50% human serum, 50% CϩY medium human serum. D, TIGR4, JMG1 (⌬fakB3), and JMG1/pSpFakB3 were grown in 50% human serum, 50% CϩY medium, PG mass spectra from triplicate biological replicates were obtained, the areas under each peak in the spectra were summed, the molecular species containing 18:2 as a percent of the total area were calculated, and the mean Ϯ S.D. was plotted.

Structure and function of S. pneumoniae FakBs
work by mutagenesis shows that it is critical to the stability of the FA-binding protein. For example, the SaFakB2(S93A) mutant severs the connection to one side of the FA carbonyl and results in a 17°C decrease in the denaturation temperature of the protein (12). Whether the hydrogen bond network is associated with polyunsaturated FA binding or has some other mechanistic significance is unclear.
The bacterial FA-binding proteins are very different from their mammalian counterparts (20,21). The mammalian proteins are about half the size of the bacterial proteins and have a ␤-barrel structure consisting of 10 antiparallel ␤-sheets with 2 ␣-helices between the first and second ␤-strands (22)(23)(24). The carboxyl group of the FA is oriented inward and electrostatically bound to Arg-106 in the intestinal FA-binding protein (25), whereas the FA carboxyl in the bacterial binding proteins is exposed as the phosphorylation site in the FakB(FA) complexes. Another key difference is that the mammalian FA-binding proteins lack acyl chain selectivity (20,21,26) and have large cavities (PDB ID 4TJZ ϭ 245 Å 3 ; PDB ID 5CE4, 291 Å 3 ; calculated using CASTp (27)) that in some cases bind two FAs (20,22). The bacterial FA-binding proteins have smaller, narrow tunnels that impart substrate selectivity to these binding proteins. The expanded hydrophobic binding pocket of SpFakB3 is most reminiscent of the large, nonselective hydrophobic pockets that occur in mammalian FA-binding proteins, although the 110 Å 3 volume of SpFakB3 is half the size of the larger mammalian pockets. Mammalian FA-binding proteins are isolated in both their ligand-free and FA-bound forms, and there is no clear relationship between FA binding and protein stability (28). The FA-free FakBs are not stable and only FA-bound FakBs have been purified (5,12). Mutations in the hydrogen bond network that fixes the FA carboxyl significantly destabilize SaFakB2 (12). The mammalian FA-binding proteins pick up and deposit FA into phospholipid bilayers, and the 2-helix motif is thought to act as the entry and exit flap for the FA (22-24, 29, 30). The bacterial FA-binding proteins also exchange FA with FA in phospholipid bilayers (1) and transfer the phosphorylated FA to two enzyme partners, PlsX and PlsY (5). The protein movements in FakB that allow the entry and exit of the FA and acyl-phosphate remain to be defined. a Values in parentheses are for the highest resolution shell. R merge ϭ ⌺(IϪ͗I͘)/⌺(I); where I is the intensity measured for a given reflection, ͗I͘ is the average intensity for multiple measurements of this reflection. R p.i.m. ϭ (⌺[1/(n Ϫ 1)] ∧ 1/2. ⌺͉I Ϫ ͗I͉͘)/⌺(I). R work ϭ ⌺ʈF obs ͉ Ϫ ͉F calc ʈ/⌺͉F obs ͉; where F obs and F calc are the observed and calculated structure factor amplitudes, respectively. R free is R work for the 5% excluded reflections during refinement. b NA, not applicable. c ML is maximum likelihood. Invitrogen, Anza alkaline phosphatase, T4 DNA ligase master mix, and DNA blunt end kit for blunt end ligation. Proteins were expressed, purified, and FA exchange was accomplished as described (1,5,12).

Bacterial strains
DNA sequences were synthesized for each of the S. aureus and S. pneumoniae FakB genes (Invitrogen). The DNA sequences for S. aureus fakB1 (SaUSA300_0733) and fakB2 (SaUSA300_1318) included a ribosomal binding site and a His 6 tag on the N terminus. Each S. aureus fakB sequence was cloned into plasmid pCS119, which has a pCM28 backbone (31), no ribosomal binding site, and uses a sarA P1 promoter to drive expression. DNA sequences for S. pneumoniae strain TIGR4 fakB1 (SP_1557), fakB2 (SP_1112), and fakB3 (SP_0742) genes did not contain a ribosomal binding site or a His 6 tag and were cloned into pET28a (Novagen), to add an N-terminal His 6 tag, and ligated into pPJ480 (32), which is a pCS119 plasmid derivative containing the sarA P1 promoter as well as a ribosomal binding site. Restriction followed by ligation reactions were used to insert the fakB gene of interest into either pCS119 (S. aureus expression) or pPJ480 (S. pneumoniae expression). The ligation mixture was transformed into Top10 Escherichia coli-competent cells (Invitrogen) and selected on Luria broth with carbenicillin (100 g/ml). Purified plasmids were passaged through strain Sa178R1 to acquire the S. aureus DNA methylation pattern, re-isolated, and electroporated into strain JLB31.
Strain JMG1 (⌬fakB3) was constructed by gene splicing by overlap extension PCR, which allowed for the creation of a gene deletion with a kanamycin resistance cassette in its

Structure and function of S. pneumoniae FakBs
place (33). Briefly, primers 1 and 2 were used to amplify the upstream region (ϳ1 kb) and primers 5 and 6 were used to amplify the downstream region (ϳ1 kb) of the fakB3 (locus tag, SP_0742) gene from TIGR4 genomic DNA ( Fig. 5A; Table 1). Primers 3 and 4 were used to amplify the kanamycin cassette (ϳ1.5 kb) from the pABG5 plasmid. These three PCR products underwent another PCR amplification using primers 1 and 6 to generate a linearized hybrid construct with the fakB3 gene replaced by the kanamycin cassette (Fig.  5A). The linear DNA was run on an agarose gel and purified before being transformed. To transform 1 ml of TIGR4 culture in CϩY media (A 600 ϭ 0.07), 3 l of competence pheromone were incubated at 37°C for 13 min before 5 l of the linearized DNA was added and incubated at 37°C for 4 h. Cells were plated on 3% blood agar plates with kanamycin (400 g/ml) and neomycin (20 g/ml) and incubated overnight at 37°C. The colonies that arose were purified and genomic DNA was extracted. PCR was used to confirm the deletion using primers 1 and 6 and 7 and 8 ( Fig. 5A; Table 1).
The SpFakB3 expression plasmid was made by engineering a gene sequence Genewiz) with an EcoRI site on the 5Ј end followed by a ribosomal binding site taken from the pABG5 vector, a His 6 tag, the TIGR4 S. pneumoniae fakB3 nucleotide sequence, and ClaI restriction site at the 3Ј end of the gene. The assembled expression construct was inserted into pUC57. Purified plasmid was passaged through INV110 E. coli-competent cells as the ClaI site was sensitive to Dam methylation. The pUC57 containing the fakB3 insert and pABG5 vector were cut using EcoRI and ClaI, ligated, and transformed into Top10 E. coli-competent cells. The resulting plasmid was purified and amplified using PCR with primers pABGB3BamHI and pABGB3SacI to amplify the fakB3 insert as well as the upstream region of the pABG5 plasmid containing the rofA promoter (34). The pABGB3SacI primer changed the 3Ј restriction site to SacI so that the resulting amplicon could be inserted into pDL278. A restriction digest was performed on the PCR product using BamHI and SacI and the product was run on an agarose gel before the band was extracted and purified. The fakB3 fragment was inserted into linearized pDL278 via blunt end ligation and subsequently transformed into Top10 E. coli-competent cells. The resulting plasmid was purified and sequenced before being transformed into strain JMG1 (⌬fakB3) and plated on 3% blood agar plates with kanamycin (400 g/ml), spectinomycin (150 g/ml), and neomycin (20 g/ml). TIGR4/pDL278 and JMG1/pDL278 were made using the above-mentioned

Structure and function of S. pneumoniae FakBs
transformation protocol and plated on the same plates used to select for strain JMG1.

FA incorporation experiments
S. aureus and S. pneumoniae strains AH1263 and TIGR4, respectively, were inoculated into 5 ml of growth medium (S. aureus, Luria broth; S. pneumoniae, CϩY) at an A 600 of 0.05, and grown at 37°C to an A 600 of 0.5, harvested, and resuspended in 5 ml of 50% human serum/LB (S. aureus) or 50% human serum/CϩY (S. pneumoniae). Cultures were incubated 4 h before cells were harvested, washed twice in growth medium, and once in PBS, and lipids were extracted (35). Plasmids containing S. pneumoniae or S. aureus genes in strain JLB31 were grown in Luria broth at 37°C to an A 600 of 0.5. A FA mixture (10 M each) of [d 4 ]16:0, -18:1, and -18:2, along with Brij-58 (0.1%) was added to the culture, incubated at 37°C for 30 min, and the lipids were extracted. Protocol was the same for polyunsaturated FA incorporation with the exception that a mixture (7.5 M each) of 18:1, 18:2, 18:3, and 20:4 was added to the growth medium.
Lipid extracts were resuspended in chloroform:methanol (1:1). PG was analyzed using a Shimadzu Prominence UFLC attached to a QTrap 4500 equipped with a Turbo V ion source (Sciex). Samples were injected onto an Acquity UPLC BEH HILIC, 1.7 m, 2.1 ϫ 150-mm column (Waters) at 45°C with a flow rate of 0.2 ml/min. Solvent A was acetonitrile, and solvent B was 15 mM ammonium formate, pH 3. The HPLC program was the following: starting solvent mixture of 96% A/4% B, 0 to 2 min isocratic with 4% B; 2 to 20 min linear gradient to 80% B; 20 to 23 min isocratic with 80% B; 23 to 25 min linear gradient to 4% B; 25 to 30 min isocratic with 4% B. The QTrap 4500 was operated in the Q1 negative mode. The ion source parameters for Q1 were: ion spray voltage, Ϫ4500 V; curtain gas, 25 psi; temperature, 350°C; ion source gas 1, 40 psi; ion source gas 2, 60 psi; and declustering potential, Ϫ40 V. The system was controlled by the Analyst software (Sciex). The sum of the areas under each peak in the mass spectra was calculated and the percent of each molecular species present was calculated with LipidView software ( The samples were introduced to the QTrap 4500 by direct injection to perform product scans to verify the fatty acids present in a particular molecular species. The ion source parameters for the negative mode product scan were: ion spray voltage, Ϫ4500 V; curtain gas, 10 psi; collision gas, medium; temperature, 270°C; ion source gas 1, 10 psi; ion source gas 2, 15 psi; declustering potential, Ϫ40 V; and collision energy Ϫ50 V.

Fatty acid composition of human serum
Lipids were extracted from human serum (Millipore-Sigma) by the method of Bligh and Dyer (35), and fatty acid methyl esters were prepared using methanol/hydrochloric acid. The fatty acid methyl esters were analyzed by a Hewlett-Packard model 5890 gas chromatograph equipped with a flame ionization detector and separated on 30 m ϫ 0.536 mm ϫ 0.50-m DB-225 capillary column. The injector was set at 250°C, and the detector was at 300°C. The temperature program was as followed: initial temperature of 70°C for 2 min, rate of 20°C/min for 5 min (final 170°C), rate of 2°C/min for 10 min (final 190°C), hold at 190°C for 5 min, rate of 2°C/min for 15 min (final 220°C), hold at 220°C for 5 min. The identity of fatty acid methyl esters was determined by comparing their retention times with fatty acid methyl ester standards (Matreya). The composition was expressed as weight percentages. Tubes were incubated at 37°C for 5 min before acetic acid (0.6%) was added and 40 l was pipetted onto a DE81 Whatman filter paper disc and discs were washed three times, 20 min each, in ethanol containing acetic acid (1%). Discs were dried and counted by scintillation counting. Apparent K m values were determined performing a nonlinear regression and using the Michaelis-Menten equation. The de novo phospholipid biosynthetic pathway in S. pneumoniae delivers acyl-ACP generated by the FASII system to PlsX to provide FAϳP to the glycerol phosphate (G3P) acyltransferase (PlsY) to form lysophosphatidic acid (LPA) and initiate phospholipid synthesis. Acyl-ACP is also a substrate for PlsC that completes the formation of phosphatidic acid, the universal precursor to membrane glycerolipids (PL). At the infection site the host provides an environment containing a spectrum of saturated (SFA), monounsaturated (MUFA), and polyunsaturated (PUFA) FA. Each of these types of FA are the preferred ligand for binding to unique FakB-binding proteins that carry the FA to FakA, where they are phosphorylated. The resulting FAϳP is a substrate for PlsY or is converted to acyl-ACP by PlsX to act as substrate for PlsC. The long-chain acyl-ACP binds to the FabT transcriptional regulator, and the complex binds tightly to the promoters within the FASII gene cluster to potently suppresses the expression of the entire biosynthetic gene set and shut off acyl-ACP formation from FASII. The acyl-ACP may also enter FASII and be elongated, although this is a minor pathway due to the strong suppression of the elongation cycle enzymes by FabT-acyl-ACP complex.  20:4). The reaction conditions were the same as above with the exception that the total volume of the reaction was 100 l. Eighty microliters was spotted onto DE81 Whatman filter paper discs. Discs were washed three times with an ethanol/acetic acid (1%) mixture for 20 min each wash. Discs were dried and counted by scintillation counting. Experiments were each performed three times in duplicate.