Uptake of Biotin by Chlamydia Spp. through the Use of a Bacterial Transporter (BioY) and a Host-Cell Transporter (SMVT)

Chlamydia spp. are obligate intracellular Gram-negative bacterial pathogens that cause disease in humans and animals. Minor variations in metabolic capacity between species have been causally linked to host and tissue tropisms. Analysis of the highly conserved genomes of Chlamydia spp. reveals divergence in the metabolism of the essential vitamin biotin with genes for either synthesis (bioF_2ADB) and/or transport (bioY). Streptavidin blotting confirmed the presence of a single biotinylated protein in Chlamydia. As a first step in unraveling the need for divergent biotin acquisition strategies, we examined BioY (CTL0613) from C. trachomatis 434/Bu which is annotated as an S component of the type II energy coupling-factor transporters (ECF). Type II ECFs are typically composed of a transport specific component (S) and a chromosomally unlinked energy module (AT). Intriguingly, Chlamydia lack recognizable AT modules. Using 3H-biotin and recombinant E. coli expressing CTL0613, we demonstrated that biotin was transported with high affinity (a property of Type II ECFs previously shown to require an AT module) and capacity (apparent K(m) of 3.35 nM and V(max) of 55.1 pmol×min−1×mg−1). Since Chlamydia reside in a host derived membrane vacuole, termed an inclusion, we also sought a mechanism for transport of biotin from the cell cytoplasm into the inclusion vacuole. Immunofluorescence microscopy revealed that the mammalian sodium multivitamin transporter (SMVT), which transports lipoic acid, biotin, and pantothenic acid into cells, localizes to the inclusion. Since Chlamydia also are auxotrophic for lipoic and pantothenic acids, SMVT may be subverted by Chlamydia to move multiple essential compounds into the inclusion where BioY and another transporter(s) would be present to facilitate transport into the bacterium. Collectively, our data validates the first BioY from a pathogenic organism and describes a two-step mechanism by which Chlamydia transport biotin from the host cell into the bacterial cytoplasm.


Introduction
The Chlamydia are obligate intracellular Gram-negative bacterial pathogens infecting economically-important animals as well as humans. C. pneumoniae causes pneumonia in humans and may be a risk factor for the development of cardiovascular disease [1,2]. C. trachomatis serovars A-C are responsible for trachoma, one of the leading causes of infectious blindness, while serovars D-K and LGV are the leading cause of bacterial sexually transmitted infections (STI) worldwide [3,4]. Chlamydial STIs can result in decreased fertility in men and women as well as sterility, pelvic inflammatory disease, and life threatening ectopic pregnancies. C. psittaci, an avian pathogen, causes psittacosis in humans while other Chlamydia spp. cause disease in a variety of wild and economically important animals with the most notable being C. abortus, which causes infectious abortus in sheep, cattle, and goats [5].
Chlamydia spp. undergo a unique two-stage developmental cycle initiating with attachment of the extra-cellular form, known as an elementary body (EB), to the eukaryotic cell surface [6]. After attachment, the EB is internalized inside a host derived membrane vacuole termed an inclusion. Within the inclusion the EB differentiates into the replicative form, the reticulate body (RB), which undergoes binary fission. Approximately 18 hours post infection, the RBs differentiate asynchronously back into EBs, eventually lysing the cell after 40-72 hours (depending upon the species). During growth and development, Chlamydia interact extensively with the host cell using a type 3 secretion system to transport effectors into the cytoplasm leading to multiple alterations of normal host cell processes including the re-routing of lipid vesicles to the inclusion and modification of the lipid and protein content of the inclusion membrane [7,8].
The genome sequences of eight different species and more than 30 different strains reveals that Chlamydia are closely related (C. pneumoniae and C. trachomatis share .80% of their proteome while C. trachomatis serovars D, A, and L2 share .95% of their proteome [9,10,11]). Despite strong genome conservation, Chlamydia spp. show different host and tissue tropisms and the mechanisms responsible for these differences remain unclear. Chlamydia spp. show multiple signs of reductive evolution typical of obligate intracellular bacteria and are auxotrophic for many essential compounds including certain nucleotides, amino acids, vitamins, and cofactors [12]. However, auxotrophic requirements, and thus metabolic capacity, vary both between species and in some cases within species [11]. In general, species previously named Chlamydophila (now Chlamydia) show slightly greater synthetic capabilities than Chlamydia trachomatis and Chlamydia muridarum with minor variations present amongst strains of a given species. While overall metabolic variation is subtle, minor variations may greatly affect infectious outcomes. The best studied example in C. trachomatis, trpBA (the tryptophan synthase genes), which are active in C. trachomatis genital serovars, but are present as pseudogenes in ocular serovars, allow the genital serovars to produce tryptophan from indole [13]. Functionally, this allows the genital serovars to overcome tryptophan starvation conditions brought on by IFN-c induction of the tryptophan degrading enzyme indoleamine 2,3dioxygenase [IDO] through the use of indole (an IDO insensitive, TrpBA substrate) supplied by the vaginal microflora. This difference in metabolism is believed, in part, to explain the ocular versus genital tissue tropism found amongst C. trachomatis serovars. Variations in metabolism leading to tissue tropisms also have been reported in other pathogenic bacteria and fungi, highlighting the importance of continued studies of metabolic specialization to provide insight into why highly related species infect and replicate better in some niches than others [14,15,16,17,18]. Therefore, we sought to define biotin metabolism, which differs amongst Chlamydia spp. to further assess the potential effects of metabolism on tropism.
Biotin, an essential vitamin for all forms of life, functions as a cofactor in carboxylation, decarboxylation, and transcarboxylation reactions by acting as a carrier of CO 2 [19]. For biotin to be functional, it must be ligated to the e-amino group of lysine within an apoenzyme. This process is performed by biotin ligases that in some organisms also function as regulators of genes involved in the metabolism of biotin [20]. Mammalian cells are auxotrophic for biotin and utilize two transporters for biotin uptake; the monocarboxylate transporter 1 (MCT1), which is used for biotin uptake by lymphoid cells and mitochondria, and the more widely disseminated sodium multivitamin transporter (SMVT) [21]. In contrast, many bacteria synthesize biotin from pimeloyl-CoA (which is made from different precursors) using a conserved four step process ( Figure 1A) while others are auxotrophic and must acquire biotin from an exogenous source [22,23]. Bacteria like Escherichia coli transport biotin using an uncharacterized mechanism while others use an energy coupling -factor transporter (ECF) [24,25]. This recently described transporter family consists of an S component that provides substrate specificity, a membrane protein T component, and an ATPase A component [26]. For biotin transport, these genes have been designated BioY(S), BioM(A), and BioN(T) ( Figure 1B). These modular units can be encoded as an operon (Type I ECFs) or as separate S components that utilize a shared AT module (type II ECFs). Many pathogenic bacteria utilize Type II ECF transporters making them attractive antibiotic targets due to the presence of a shared AT module that facilitates uptake of multiple essential compounds. Genome annotations of the pathogens Chlamydia and Rickettsiales (and other environmental bacteria) show the presence of a bioY in the absence of AT modules. While the BioY from Rhodobacter capsulatus functions independently from BioM and BioN in a recombinant Escherichia coli transport model, BioMN (the AT module) increases biotin transport affinity [24]. Thus it is unclear if the orphaned BioY transporters function as high or low affinity transporters or perhaps interact with highly divergent AT modules not recognized through similarity searches.
Analysis of chlamydial genomes identified differences across species (and in one case within a species) in how biotin is acquired. Some species possess the biotin synthesis genes (bioF_2ADB), the biotin transporter bioY, or both biotin synthesis and transport genes. We used recombinant E. coli expressing BioY (CTL0613) from C. trachomatis 434/Bu to determine that BioY functions as a high affinity and capacity transporter in the absence of the BioM and BioN AT module. This is the first study characterizing an orphaned type II ECF BioY and our results indicate that some S components might not require an AT module for high affinity substrate transport. Mammalian acquisition of biotin via protein transporters indicates that cell membranes are poorly permeable or impermeable to biotin. Hence, the host derived Chlamydia inclusion membrane also may require a biotin transporter. Immunofluorescence microscopy studies on infected cells demonstrated that SMVT colocalizes with the chlamydial inclusion membrane. These findings indicate that Chlamydia redirects a portion of the SMVT pool away from the cell membrane to the inclusion membrane to facilitate transport of biotin into the inclusion space where it can then be transported by BioY into the bacterium. SMVT also transports the essential compounds lipoic acid and pantothenic acid (for which Chlamydia are auxotrophic) into the host cell [12,27,28]. While previous work has shown that Chlamydia redirect cellular proteins to the inclusion membrane for its benefit, this is the first example to our knowledge of Chlamydia recruiting a cellular transporter and may serve as a paradigm for how small molecules gain access to the intra-inclusion space.

Utilization of biotin by Chlamydia
Chlamydia encode at least two well characterized biotin utilizing proteins, the biotin carboxyl carrier protein AccB and the biotin protein ligase BirA. AccB acts in concert with AccACD to form acetyl CoA carboxylase, which converts acetyl-CoA to malonyl-CoA in the first committed step in fatty acid synthesis [29]. The biotin ligase, BirA, ligates biotin to the e-amino group in a conserved lysine residue within AccB [30]. The chlamydial AccB proteins share 25-32% amino acid sequence identity with the E. coli homolog and possess an altered biotin-ligation motif sequence (consensus: AMKM, chlamydial: AMKV, Figure S1) [20]. The chlamydial BirA is a Class I biotin ligase (Class II biotin ligases contain an N-terminal helix-turn-helix motif involved in regulating the expression of biotin synthesis genes), which lacks the conserved C-terminal region found in many Class I and II BirA proteins ( Figure S2) [20]. In addition to AccB and BirA, Chlamydia spp. encode accACD along with genes comprising downstream fatty acid biosynthesis pathways indicating that Chlamydia utilize biotin for fatty acid synthesis.
Western blot analysis (using an HRP-streptavidin conjugate) of EB samples ( Figure 2A) and cells infected for 20 hours (when RBs are the dominant developmental form) ( Figure 2B) identified at least one Chlamydia specific biotinylated protein. E. coli (which contains a single covalently biotinylated protein, AccB) was used as a positive control. Samples were run under denaturing conditions resulting in the detection of only covalently modified biotinylated proteins. These blots identified a unique biotinylated protein at ,20 kDa present in chlamydial samples that was absent in control cell lysates. The predicted mass of the chlamydial AccB is 18.2 kDa. Under our conditions, the E. coli AccB ran larger (,20 kDa) than its predicted mass of 17.7 kDa. Banding patterns at the top of the blots represent biotinylated mammalian enzymes: propinoyl-CoA (60-80 kDa), methanoyl-CoA (80 kDa), pyruvate-CoA (130 kDa), and acetyl-CoA 1 and 2 (265 and 280 kDa, respectively) [31]. Control Hsp60 Western blots were run to confirm the presence of Chlamydia in the appropriate samples.

Genome analysis of biotin acquisition strategies in Chlamydia
Analysis of the available chlamydial genomes predicts two separate pathways for biotin acquisition: synthesis starting with pimeloyl-CoA using BioF_2(BioF_1)/BioA/BioD/BioB or biotin transport using BioY ( Figure 1 and Table S1). No homologs to proteins known to comprise the pimeloyl-CoA synthesis pathways in Gram-negative or Gram-positive bacteria were identified through BLAST analysis of chlamydial genomes. The chlamydial ancestor Parachlamydia acanthamoebae encodes proteins with weak homology to BioC (PUV_13520, 29% identity) and BioH (PUV_07140, 24% identity) while Waddlia chondrophila encodes a BioH homolog (wcw_1298, 26% identity). Similar to other bacteria, the biotin synthesis genes are arranged in a putative operon (bioBF_2DA) [32]. Chlamydia do not encode homologs to the known biotin synthesis gene regulators BioQ, BioR (the C. suis TetR protein is homologous to BioR, but likely is present due to its role in regulating tet(C) [33]), or the Class II BirA.
With the exception of C. abortus strains S26/3 and LLG [34], variation in biotin metabolism was not observed within species (Table S1). Alignment of reference genomes anchored using the C. trachomatis 434/Bu dapB/asd/lysC/dapA gene cluster shows the localized area of genome variation accounting for divergence between biotin synthesis and/or biotin transport species ( Figure 3A). A genome insertion between aroA and dapB in the group formally termed Chlamydophila contains the predicted biotin synthesis genes bioBF_2DA arranged in a putative operon (the lone exception is C. caviae which still contains an insertion relative to C. trachomatis). The genes are not arranged in the order of biotin synthesis and, consistent with the absence of known biotin gene regulators in Chlamydia, biotin regulator binding sites are not present upstream of the bioB initiation codon.
Chlamydia possess two bioF homologs, bioF_1 and bioF_2, which are predicted to encode a 7-keto-8-aminoperlargonate (KAPA) synthase. BioF carries out the first committed step in biotin synthesis converting pimeloyl-CoA to KAPA ( Figure 1A). The bioF_2 gene is encoded as part of the biotin synthesis gene cluster and BioF_2 is more closely related to the characterized BioF from E. coli than BioF_1 ( Figure S3). The bioF_1 gene is encoded separately from the biotin synthesis/biotin transport gene region where it converges upon and overlaps with the 39 end of priA ( Figure 3B). While both BioF_1 and BioF_2 contain a pyridoxal phosphate-dependent aspartate aminotransferase fold, BioF_1 lacks residues important for activity of the E. coli BioF including a stretch of 21 N-terminal amino acids ( Figure S3) [35]. Based on genome annotations, the bioF_1 gene in C. pneumoniae, C. pecorum, C. trachomatis, and C. muridarum also encode an alternative initiation codon, TTG, which was verified using Sanger sequencing of the bioF_1 region in C. pneumoniae TW-183 and C. trachomatis 434/Bu. Consistent with these sequence differences, neighbor joining analysis separates BioF_1 and BioF_2 (clustering BioF_2 closer to BioF from E. coli) and distinguishes between BioF_1 with and without an ATG start codon ( Figure S3B).
The bioY gene is located approximately one to six kb downstream from dapA in all strains (except C. pneumoniae which is bioY-negative) and displays ,52% amino acid identity (42%-85%) with respect to other Chlamydia and ,19% amino acid identity (14%-24%) with the previously characterized Class I ECF BioY transporters ( Figure S4) [24,36,37,38]. The absence of the biotin energy module genes BioM and BioN adjacent to the chlamydial bioY leads to a Class II ECF designation and would predict that an AT module should be encoded elsewhere to be shared by solute-specific S transporter proteins [26]. However, unlike standard type II ECFs, Chlamydia encode no identifiable AT gene homologs. Biotin transport across the bacterial membrane To characterize the transport properties of BioY, we constructed a recombinant N-terminal His-tagged version of CTL0613 from C. trachomatis 434/Bu for expression in E. coli (E. coli ATM1172). The obligate intracellular life-style of Chlamydia and the absence of characterized gene expression systems often necessitate the use of an alternative bacterial host for the study of chlamydial proteins. C43(DE3), the E. coli host strain used for BioY expression, is derived from E. coli BL21(DE3) [39]. Consistent with a previous study that rigorously analyzed bacteria for the presence of ECFs, our analysis of the BL21(DE3) genome revealed no ECF AT modules that would potentially interfere with our analysis of the orphaned chlamydial BioY [26,40]. Anti-Histag Western blotting demonstrated that BioY was expressed by E. coli ATM1172 and migrated primarily at a calculated mass of 37 kDa with a minor band at 21 kDa ( Figure S5). The expected mass of a His-tagged BioY monomer and dimer are 24 and 48 kDa, respectively.
We next analyzed the transport properties of BioY using a 3 Hbiotin uptake assay. Preliminary experiments demonstrated that BioY transport properties were not adversely affected by the presence of the N-terminal His-tag ( Figures S6 and S7) and that uptake of biotin by E. coli ATM1172 was significantly higher than E. coli ATM1135, the host strain carrying the empty vector ( Figure  S6). E. coli ATM1172 transported biotin with an apparent K(m) of 3.35 nM (95% confidence interval 3.27 to 3.42) and an apparent V(max) of 55.1 pmol6min 21 6mg 21 (95% confidence interval of 55.0 to 55.3) ( Figure 4A). Consistent with carrier mediated transport, 3 H-biotin transport was linear over the first minute of incubation and reached saturation within approximately 10 minutes ( Figure 4B). The transport specificity of BioY was analyzed by measuring the inhibition of 3 H-biotin transport in the presence of biotin, the sulfur containing compounds lipoic acid and Lmethionine, or the biotin precursor desthiobiotin. Transport of 3 H-biotin was significantly inhibited only in the presence of biotin ( Figure 5).

Transport of biotin across the inclusion membrane
Since the membrane of the Chlamydia inclusion is derived from the host cell which is poorly permeable of impermeable to biotin [19] and previous studies were unable to measure inclusion permeability to molecules between 45 and 520 Da (biotin has a mass of 244 Da), we hypothesized that the inclusion also requires a transporter to import biotin [41,42]. SMVT localization was initially studied using antibodies that recognize the C-terminus of SMVT (H56) in both infected and mock-infected HeLa cells. Antibodies against the C. trachomatis inclusion protein IncA were used to define the inclusion. In uninfected cells, SMVT had a diffuse, punctate staining pattern across the cell surface ( Figure 6A). In contrast, infected cells had an intense SMVT signal that colocalized with the IncA antibody staining pattern ( Figure 6B). To confirm these findings, experiments also were performed using an antibody raised against the N-terminus of SMVT (N14). While detection of SMVT with N14 in uninfected cells was less obvious than with H56, colocalization with IncA staining was easily visualized ( Figure S8A and B). Similar to experiments with C. trachomatis 434/Bu, SMVT (detected using the H56 antibody) also localized to the chlamydial inclusion in HeLa cells infected with C. trachomatis E, C. psittaci 6BC, and C. caviae GPIC ( Figure S8C). Both C. trachomatis E and C. caviae GPIC are biotin auxotrophs while C. psittaci 6BC possesses genes for biotin synthesis and transport (Table S1). To demonstrate that these results were not a HeLa specific phenomenon, we also tested for colocalization of SMVT with C. trachomatis 434/Bu inclusions (using anti-SMVT H56 and anti-IncA antibodies) in infected McCoy mouse cells and obtained identical results ( Figure S9).

Discussion
Obligate intracellular bacteria exist under a unique ''living arrangement'' enabling them to entirely or partially shed energetically costly metabolic pathways if they possess mechanisms to parasitize the host for the essential intermediate or end products. However, the ensuing evolutionary reduction in metabolic capacity and subsequent reliance on the host cell may result in host or tissue tropisms due to growth restrictions arising from differences in cellular nutrient content and bioavailability. Thus, bacteria risk becoming overly specialized as a consequence of their diminished metabolic versatility. Elucidation of the underlying genotypic and phenotypic reasons for tropisms has the potential to illuminate an Achilles heel to be targeted by chemotherapeutics or vaccines. Of course, a pathogen's metabolic repertoire will not be the sole factor(s) driving tropism, but will function in concert with more traditional virulence factors.
Despite their close genomic relationship Chlamydia spp. have different host and tissue tropisms. The previously classified Chlamydophila family members, with the exception of C. caviae, generally have a greater host range and can propagate in more tissue types than C. trachomatis and C. muridarum [43]. Genome sequences for all species except C. suis have now been published and genome comparisons reveal that divergence primarily lies within the plasticity zone (which includes two toxin-like genes along with tryptophan synthesis and purine/pyrimidine salvage pathways), the type V transporter repertoire (Pmps), putative T3SS effectors (which includes the Inc family proteins), and metabolism [11]. Analysis of metabolic capacity of Chlamydia spp. based upon genome annotations shows differing abilities to salvage nucleotides, synthesize tryptophan, transport pantothenic acid, degrade arginine, and acquire biotin. However, functional analysis of AaxB and AaxC from C. trachomatis serovars determined that missense mutations led to inactive proteins [44]. Loss of activity was not predicted based upon sequence analysis alone, hinting that additional pseudogenes may be present and that metabolic divergence may be greater than appreciated.
Biotin has long been known to be an essential vitamin for mammals with biotin deficiencies and mutations in biotin utilizing enzymes resulting in multiple defects apparent at both the cellular and organismal level [45]. Mammals primarily acquire biotin from dietary sources via carrier mediated uptake. The primary biotin uptake system is the SMVT, a 12 transmembrane protein that is widely expressed throughout the body [46]. SMVT also facilities the acquisition of two other essential compounds, pantothenic acid and lipoic acid. At least one other transporter, MCT1, is known to be utilized for uptake of biotin by lymphoid cells and potentially mitochondria [19]. Once inside the cell, biotin is covalently attached to the e-amino group of a conserved lysine within apoenzymes by a biotin protein ligase. Mammalian cells utilize five biotin requiring enzymes for various trans-, de-, and carboxylation reactions in which biotin serves as a CO 2 carrier.
Bacteria generally encode fewer biotin requiring enzymes than mammalian cells, with E. coli known to contain only one covalently modified enzyme (AccB) along with the biotin protein ligase BirA to which biotin is bound non-covalently as AMP-biotin [30]. Similar to E. coli, Chlamydia only contain one annotated biotinylated protein, AccB, and a single biotin protein ligase, BirA. HRP-streptavidin blotting analysis of both the EB and RB developmental forms revealed a single band that was absent in protein samples from uninfected cells. This band size is larger than the expected size of the chlamydial AccB (.20 kDa versus 18.2 kDa). However, similar results were seen for the E. coli AccB (.20 kDa versus 17.7 kDa) and are consistent with other studies [47,48]. Thus, it is highly likely that the single reactive band from Chlamydia is AccB. Higher weight reactive bands seen in all EB, RB, and uninfected cell samples correspond to the predicted sizes of the mammalian biotin-utilizing enzymes. Contamination of EB preparations with host proteins is common and RB samples were prepared from lysates of infected HeLa cells [49]. The bands migrating between AccB and the higher weight bands (.50 kDa) could be AccB multimers or biotinylated histones [50]. Since biotin is not covalently attached to BirA we did not expect to see a BirA band under denaturing gel conditions. Proteomics studies have identified all of the acetyl-CoA carboxylase complex components (ACC, comprised of AccABCD) in C. trachomatis EBs and RBs, AccBA in C. pneumoniae EBs, and lipid analysis has demonstrated that Chlamydia synthesize fatty acids de novo [49,51,52,53]. Our data further support the role of the ACC and suggest that the holoenzyme may be present in EBs (and, not surprisingly, RBs) indicating that fatty acid synthesis may be required during immediate early infection. If this model is correct, then ACC would make a prime target for antimicrobials.
BirA from Chlamydia is a Class I biotin protein ligase (BPL). Class I BPLs lack the N-terminal helix-turn-helix DNA binding motif (utilized for regulation of biotin synthesis genes) that defines the Class II BPLs [20]. The chlamydial BirA also lack a conserved C-terminal region of unknown function found in most Class I and II BPLs. The likely presence of a biotinylated AccB-like protein in Chlamydia demonstrates that BirA is functional and transcriptome data indicate that BirA is present throughout development [54,55]. The Class I BPL in Chlamydia and the absence of homologs to other known biotin gene regulators [32] suggests that either the biotin synthesis/transport genes do not need to be regulated or that regulation occurs through a novel mechanism. Species with a single biotin acquisition strategy (BioY or synthesis) may not require regulation. However, for species possessing both bioY and synthesis genes a lack of regulation would result in the use of redundant pathways. As biotin synthesis is a costly metabolic pathway, biotin transport via BioY would seem like the preferred pathway except when biotin is limiting [22]. Analysis of biotin tissue content in rats and chickens found that biotin levels are low in lung and heart tissue compared to other organs [56]. This could explain why C. pneumoniae, which primarily infects the lungs, has dispensed with bioY. Furthermore, species with greater host and tissue range (C. psittaci, C. abortus, and C. felis) maintain both biotin synthesis and transport genes which may allow them to switch between energy intensive synthesis versus low-energy transport depending on biotin availability at the site of infection. The recently reported sequence of C. abortus LLG was found to have a putative degraded biotin synthesis pathway (Table S1) making it the only example of biotin metabolism variation within a species [34]. C. abortus can grow in a number of hosts and tissues, but is best known for causing abortions. Pregnant animals are known to incur transient biotin deficiencies and mouse models have shown that reduced biotin levels in dams correlates with reduced levels of biotin in the fetus [57,58]. The loss of biotin synthesis by the LLG strain might explain its reduced pathology in a pregnant mouse abortion model and small plaque phenotype and makes it an ideal candidate to test for biotin dependent phenotypes compared to ''dual'' biotin metabolism C. abortus strains [59].
The biotin synthesis bioBF_2DA genes in Chlamydia would facilitate biotin synthesis starting from pimeloyl-CoA. Global expression analysis indicates that these genes are expressed in C. pneumoniae showing maximum levels during the mid-cycle when bacterial division and fatty-acid synthesis would be maximal [55]. In Gram-negative bacteria, pimeloyl-CoA is synthesized by BioC and BioH (Figure 1) [22]. Chlamydia lack homologs to BioC/BioH, but do possess a FabH which shares 33% identity with BioZ (a FabH-like protein) from Mezorhizobium sp. strain R7A. In M. RZA  BioZ is predicted to function in both fatty acid synthesis (as FabH) and biotin synthesis as BioH due to relaxed substrate specificity compared to the E. coli FabH [60]. A similar mechanism would not be surprising for Chlamydia spp. which are known to encode other rare or novel bifunctional enzymes [61,62]. It is unclear if Chlamydia encode a BioC homolog but the best candidate based on similarity is UbiE, which shares a SAM dependent methyltransferase domain with BioC.
Annotations of the biotin synthesis pathway in Chlamydia list two bioF genes, bioF_1 (located next to priA) and bioF_2 (encoded within the biotin synthesis operon), which share an aspartate aminotransferase fold type I. BioF_2 shows greater similarity to the characterized synthetic BioF proteins, shares more active site residues with BioF, and is present with the other biotin genes making it likely to be the bona fide BioF. We predict that BioF_1 participates in an amino-transferase reaction separate from biotin synthesis, consistent with its maintenance and expression in biotin synthesis deficient species, and should be reclassified as a conserved hypothetical protein until functional analysis can assign a proper role.
Biotin transport in bacteria has been recognized for over 35 years [25,63], however the actual transport proteins have only recently been determined and still remain elusive for E. coli. The ECF transport family described by Rodoniov et al. comprises a collection of small molecule transporters with diverse substrate preference, but conserved membrane energy coupling components [26]. Studies with the Rhodobacter capsulatus BioYMN have demonstrated that the S component can function independently as a low affinity, high capacity transporter, but is converted to a high affinity, low capacity transporter in the presence of the BioMN module [24]. The Class I BioYMN from other bacteria also function similar to the BIOYMN from R. capsulatus (Table S2). Whether the non-R. capsulatus Class I BioY transporters show altered transport in the absence of BioMN is unknown. Genome analysis by Rodionov et al. for ECF transporters identified BioY as the only S component to be encoded by bacteria (including the Rickettsiales and Chlamydia) in the absence identifiable AT modules suggesting that BioY may function at a physiologically relevant level in its orphan form which may be a unique property of BioY [26,64].
Surprisingly, in our heterologous expression system BioY showed high affinity and moderate capacity transport in the absence of BioMN (bioinformatic analysis did not identify AT modules in E. coli BL21(DE3)). Previous heterologous studies of BioYMN used an E. coli K12 derivative S1039 due to its absence of the E. coli high affinity biotin transport [24,40]. However, the high affinity biotin transport system in E. coli has not been defined and the mechanism underlying its absence in S1039 is not understood. Our preliminary experiments indicated that expression of BioY in the BL21(DE3) background significantly increased transport above background levels ( Figure S6) and the apparent K(m) and V(max) were both superior to those determined for E. coli ATM1135 (Table S2). The ability of BioY to function as a high affinity transporter in the absence of BioMN may have made BioMN dispensable, leading to loss of the AT module in a subset of bacteria including Chlamydia. Berntsson et al. argue in a recent report that BioY from Lactococcus lactis (a Class II transporter) does not function as a transporter in the absence of a shared MN module, but merely binds biotin [65]. However, the absence of identifiable MN-encoding genes in Chlamydia and the differences in BioY amino acid sequences between these species (,25%) suggests that the chlamydial BioY may have altered functionality. Amino acid sequence analysis does not reveal obvious reasons for these transport differences and the regions of BioY that interact with BioN or BioM are not known. Work focusing on the stoichiometry of ECF components predict an SA 2 T for non-BioY ECFs while the BioYMN system is predicted to exist in an S 2 A 2 T ratio or possibly as an S 2 A 2 T 2 [40,64,66]. Consistent with this observation our recombinant BioY ran primarily as a dimer ( Figure S5). Interestingly, BioY is present in many bacteria in two copies (including the Chlamydia relative Simkania) [26] suggesting that perhaps a low affinity complex (requiring BioMN) and high affinity complex could be assembled with the loss of the low affinity complex in some species accompanied with the loss of the AT module. Further biotin transport studies using Class I and Class II BioY transporters will be required to address this point including comparisons of the transport affinities of BioY from species encoding two copies of bioY with and without a heterologous or homologous BioMN as well as studies to differentiate binding versus transport in these systems.
The dependence on exogenous biotin by some Chlamydia species presents a unique problem, namely, how does biotin get across the inclusion membrane? Studies have determined that the inclusion membrane is permeable to small ions, but is impermeable to polymers larger than 520 Da leaving a window of 45-520 Da in which permeability is unclear [41,42]. During infection Chlamydia maintain intimate interactions with the host cell altering host cell immune function, organelle architecture, and vesicle trafficking patterns [67,68,69]. Using the microtubule network the inclusion positions itself within the exocytic pathway, avoiding lysosomal fusion, while at the same time situating itself to intercept various vesicles and proteins [70]. These exocytic vesicles either fuse with the growing inclusion membrane or are taken up into the inclusion space where they interact directly with RBs [71,72]. Chlamydia can also retarget host proteins normally found in the cytosol or ER to the inclusion membrane [73,74]. While vesicles could potentially delivery some molecules to the inclusion space upon fusion with the inclusion membrane, not all compounds for which Chlamydia are auxotrophic would be contained in vesicles.
The redirection of vesicles and proteins to the inclusion membrane led us to seek a host cell protein to solve Chlamydia's biotin ''dilemma''. SMVT is an ideal candidate for a number of reasons including: 1) its ability to transport of biotin, lipoic acid, and pantothenic acid for which Chlamydia are auxotrophic, 2) its presence in tissues infected by Chlamydia, and 3) its reliance for transport on microtubules and dynein, cellular components that Chlamydia are known to modulate [45,70,75]. Immunofluorescence microscopy demonstrated co-localization of SMVT with the inclusion membrane in two different host cells using biotin auxotroph species and a facultative biotin prototroph. The K(m) of SMVT for biotin was determined to be 15.1 mM [27], which is significantly higher than for BioY (3.35 nM) and should allow Chlamydia to establish influx of biotin from the host cell into the bacterium. While not tested, we predict that SMVT also would colocalize to the C. pneumoniae inclusion despite its biotin prototrophy due to its need for exogenous lipoic acid and pantothenic acid. It should be noted that the requirement for exogenous pantothenic acid is less certain than for lipoic acid [28] due to the absence of annotated genes to carry out the conversion of pantothenic acid to its functional form, CoEnzyme A. How lipoic and pantothenic acid would then be transported into the bacterium is unclear. The formerly classified Chlamydophila species possess a PanF homolog, which functions as a Na/pantothenic acid symporter in E. coli while all Chlamydia lack homologs to known lipoic acid transporters.
The co-localization of SMVT with the inclusion membrane is consistent with, but does not prove, that it is inserted into and utilized by the inclusion to transport biotin and other vitamins across the inclusion membrane. Use of SMVT would further support the sensitivity of Chlamydia to microtubule disrupting agents as these compounds would not only affect lipid acquisition, but also would result in the loss of nutrients due to the failure of SMVT to route to the inclusion. The use of SMVT would set a precedent for the requirement of transporters to move compounds between 45 and 520 Da, such as nucleotides, amino acids, and SAM (for which bacterial transporters have been characterized) into the inclusion. Recent work from Saka et al. identified Npt1, the chlamydial ATP/ADP translocase, in inclusion membrane fractions as well as in RB fractions providing another mechanism to circumvent the inclusion permeability problem [51]. Coupled with our findings, it seems that Chlamydia may utilize both bacterial and host transporters to acquire compounds from the host cytosol. It will be interesting to determine if other chlamydial transporters also reside in the inclusion and why some transporters can function as bacterial and inclusion transporters while others only function in one locale. Together these data expand upon our understanding of basic chlamydial biology, further defining its metabolic requirements and ability to modulate host cell factors to its advantage.

Cell culture conditions
Cell lines (listed in Table S3) were routinely cultured in T-175 culture flasks (BD Falcon) using Dulbecco's Modified Eagle Medium+GlutaMAX TM [GibcoH] (DMEM) supplemented with 10% heat-inactivated Fetal Bovine Serum (FBS, Hyclone) at 37uC with 5% CO 2 . During infection experiments, cell medium was supplemented with 16MEM Non-Essential Amino Acids Solution (Sigma) and 0.2 mg/ml cyclohexamide (Sigma). Cells were routinely monitored for Mycoplasma infection by PCR using DNA lysates (QIAgen DNeasy Kit) with degenerate oligos that prime to the 16S rRNA of Mycoplasma spp.

Cell infection and bacterial growth conditions
The bacterial strains used in this study are listed in Table S3. EBs were harvested from either L2 cells or Hep2 cells (for C. pneumoniae) and stored at 280uC in sucrose phosphate glutamic acid buffer [7.5% W/V sucrose, 17 mM Na 2 HPO 4 , 3 mM NaH 2 PO 4 , 5 mM L-glutamic acid, pH 7.4] until use. Stocks were titered using an infection forming unit assay (IFU) and tested for Mycoplasma contamination using PCR. E. coli were routinely cultured in Luria Bertani medium at 37uC with 100 mg/ml ampicillin (Sigma) as appropriate.
Construction of E. coli ATM1172 and E. coli were used to clone CTL0613 in frame with the N-terminal His-tag sequence in pET19b (EMD chemicals). Primers Y-His5 [59-ATACCATGGTTGCTAAGAGCATTAC-39] and Y19b3 were used to create a tagless CTL0613 construct still expressed from the T7 promoter in pET19b. Restriction sites are underlined. Both sequences were PCR amplified using Platinum Taq DNA polymerase (Life Technologies) as directed. PCR products were subcloned into pGEM-T (Promega), digested using either NdeI and XhoI (for cloning with the His-tag) or NcoI and XhoI (for cloning without the His-tag), and ligated into similarly digested pET19b. Final vectors were sequence verified using T7 promoter (59-TAATACGACTCACTATAGGG-39) and T7 terminator (59-GCTAGTTATTGCTCAGCGG-39) primers at the Uniformed Services University Biomedical Instrumentation Center (Bethesda, MD) and transformed into E. coli C43(DE3) (Lucigen) to create E. coli ATM1172 and E. coli ATM1173. C43(DE3) carrying pET19b (E. coli ATM1135) also was constructed for control experiments.

Western blot detection of proteins
Detection of biotinylated proteins. To detect biotinylated proteins from mid-developmental cycle Chlamydia, confluent L2 cell monolayers grown in 60 mm tissue culture dishes were infected with EBs at an MOI of 1 (or mock infected) as previously described [44]. Detection of Hsp60 and His-tagged BioY. Samples prepared for biotinylated protein analysis also were used for Hsp60 Western blot analysis. After transferring proteins to nitrocellulose, blots were blocked with 5% milk/TBS and probed with a mouse anti-Hsp60 antibody (A57-B9, provided by Dr. Dan Rockey, Oregon State University) diluted 1:1000 in 5% milk/ 0.05%Tween-20/TBS. Blots were then washed with 0.05%Tween-20/TBS, probed with a HRP-conjugated goat anti-mouse antibody (GE Healthcare), and finally washed again in 0.05%Tween-20/TBS. To detect His-tagged BioY, E. coli ATM1172 was grown in 2xTY medium (8 g tryptone, 5 g yeast extract, and 5 g NaCl per liter adjusted to pH 7.0) with 100 mg/ ml ampicillin (Sigma) to an OD 600 nm of ,0.6 and then induced with 1 mM IPTG (Sigma) at 37uC for 2.5 hours. Bacteria were then pelleted and lysed in Laemmli buffer containing 16 Halt Protease Inhibitor Cocktail and b-mercaptoethanol, boiled, and run on 12% SDS-PAGE gels. Protein was transferred to nitrocellulose and blotted as described for Hsp60 with the exception that a mouse anti-His tag antibody (GE Healthcare) diluted 1:1000 was substituted for the anti-Hsp60 antibody. Blots were developed using Supersignal West Pico Chemiluminescent Substrate and visualized using an ImageQuant LAS 4000.

H-biotin transport assays
To measure biotin transport, E. coli ATM1172, E. coli ATM1173, or E. coli ATM1135 were grown overnight in 2xTY medium with 100 mg/ml ampicillin at 37uC. Bacteria were then subcultured into 2xTY containing 100 mg/ml ampicillin and 0.1 mM IPTG and grown at 37uC to an OD 600 nm of ,1. Cultures were placed on ice and then pelleted to concentrate samples. The pellets were washed twice with ice cold M9 minimal salts and finally suspended in M9 minimal salts to ,4610 9 bacteria per ml. Experiments were performed with 50 ml samples containing ,2610 8 bacteria and ,0.1 to 0.5 mg/ml protein.
Protein levels were quantified using the Bio-Rad Protein Assay. For 3 H-biotin uptake experiments, bacteria were mixed with 50 ml of D- [8,  Competitor studies were performed using samples prepared identically to uptake experiments. Samples were then incubated simultaneously with 3 H-biotin and inhibitor (lipoic acid, Lmethionine, biotin, or desthiobiotin [all from Sigma]) or no inhibitor for 1 minute at 30uC. Reactions were stopped and quantified as described for transport assays.

SMVT immunofluorescence localization studies
HeLa or McCoy cells were grown until subconfluent on 35 mm glass bottom cell culture dishes (World Precision Instruments, Inc). Cells were then infected with Chlamydia at an MOI ,0.1 and incubated for 20 to 24 hours prior to fixing in 3% paraformaldehyde/3% sucrose/PBS and permeablized with 0.2% Triton X-100/PBS. Prior to antibody treatment, samples were blocked in 10% FBS/DMEM. SMVT localization was assessed using either SMVT H56 (Santa Cruz, rabbit polyclonal against C-terminal peptide, used at 1:100) or SMVT N14 (Santa Cruz, goat polyclonal against N-terminal peptide, used at 1:100). For C. trachomatis 434/Bu infections, IncA was detected using a mouse anti-IncA antibody (from Ted Hackstadt, Rockey Mountain Laboratories, used at 1:100). AlexaFlour 594 chicken anti-goat, AlexaFlour 594 goat anti-rabbit, and AlexaFlour 488 goat antimouse (all from Life Technologies) were used at a concentration of 1:1000. Images were obtained using an Olympus BX51 microscope fitted with an Olympus DP-70 digital camera. Merged images were created using DP controller/manager software version 1.1.1.71.

Bioinformatics
Analysis of biotin synthesis and fatty acid synthesis pathways was performed using KEGG PATHWAY maps (http://www. kegg.com/kegg/pathway.html). Genome synteny analysis was performed using the Genome Region Comparison tool from JCVI CMR (http://cmr.jcvi.org/cgi-bin/CMR/shared/ MakeFrontPages.cgi?page = GenomeRegionComparison) and the NCBI genome database (http://www.ncbi.nlm.nih.gov/sites/ genome). Protein homology searches were performed using NCBI blastp (default settings). E-values #1e 210 were considered significant for homology analysis. Proteins were aligned using ClustalW2 through the EMBL-EBI server (http://www.ebi.ac.uk/ clustalw/) with default settings and the results were used to report percent amino acid sequence identity and guide trees using neighbor joining analysis. ClustalW2 alignment files were modified using BioEdit for visual sequence comparison [76].

Sequencing of bioF_1
To confirm the genomic sequence data for the start sites of bioF_1, the 59 regions of bioF_1 from C. pneumoniae TWAR (CpB0956, NC_000922.1) and C. trachomatis 434/Bu (CTL0146, NC010287.1) were amplified using primers bioF1cpF (59-GTTCCCTAGT GGTGTGTGGTATTCC-39) and bioF1cpR (59-CGATAACTGAAGAAGGCCCTACC-39) or bioF1ctF (59-AAATGGTTAGCCTTGAGGCCATGGAG-39) and bioF1ctR (59-ACCTGCTCGT CCCATAGCACATAATCC-39), respectively. Genomic template was obtained using the QIAgen DNeasy kit and EBs from C. pneumoniae TW-183 or C. trachomatis 434/Bu. DNA was PCR amplified using Fermentas PCR Master Mix (Thermo Scientific) as directed. PCR products were purified using the MinElute PCR purification kit from QIAgen and directly sequenced in both directions with the same primers used for the PCR reaction. Samples were sequenced at the Uniformed Services University Biomedical Instrumentation Center. Figure S1 AccB sequence comparison. AccB from E. coli and select Chlamydia (listed at the end of the alignment plot) were aligned using ClustalW. Identical (black shading) and similar (grey shading) amino acids are shown using a 100% sequence identity cut-off. The biotinylated lysine residue is shown in green. A neighbor joining tree with distances is shown in panel (B). (TIFF) Figure S2 BirA sequences comparison. BirA sequences from select Chlamydia (listed at the end of the alignment plot) were aligned with BirA from E. coli using ClustalW. Identical (black shading) and similar (grey shading) amino acids are shown using a 100% sequence identity cut-off. The helix-turn-helix DNAbinding region of the E. coli Class II BirA is highlighted in yellow. The BPL-ligase domain is shown in blue and the conserved biotin protein ligase C-terminal region is shown in green. A neighbor joining tree with distances is shown in panel (B). (TIFF) Figure S3 BioF_1 and BioF_2 sequence comparisons. BioF_1 and BioF_2 sequences were aligned using ClustalW. Identical (black shading) and similar (grey shading) amino acids are shown using a 100% sequence identity cut-off. The organisms used for analysis are shown at the end of the sequence plot. Residues marked in yellow indicate amino acids important for activity (for the E. coli BioF) that show variance across the chlamydial BioF_1 and BioF_2 sequences while those in green indicate conserved active site residues. Blue shading indicates active site amino acids that cluster separately between BioF_1 and BioF_2. Green shading indicates the conserved, catalytic lysine residue. Organisms shown in red font use a TTG initiation codon for bioF_1. A neighbor joining tree with distances is shown in panel (B). (TIFF) Figure S4 BioY sequence comparison. An alignment of BioY transporters was created using ClustalW. The bacteria species used are listed to the right of the plot (green font indicates organisms with functionally validated type I ECF biotin transporters). Percent amino acid identity versus the Rhodobacter capsulatus BioY is shown to the right of the second strand in the plot. Identical (black shading) and similar (grey shading) amino acids are shown using a 60% sequence identity cut-off. A neighbor joining tree with distances is shown in panel (B). McCoy cells were infected for 20-24 hours, fixed and permeablized, and probed with anti-SMVT (H56) antibodies (shown in red) and anti-IncA antibodies (shown in green). The inset in panel C was created by merging the red and green channel micrographs. The SMVT H56 antibody was raised against a C-terminal peptide from the human SMVT (amino acids 528-580). Species cross reactivity between the human and mouse SMVT was expected with H56 because the mouse peptide region shares 49/52 amino acids with the human antigenic peptide region. (TIFF)