Promiscuous and adaptable enzymes fill "holes" in the tetrahydrofolate pathway in Chlamydia species.

Folates are tripartite molecules comprising pterin, para-aminobenzoate (PABA), and glutamate moieties, which are essential cofactors involved in DNA and amino acid synthesis. The obligately intracellular Chlamydia species have lost several biosynthetic pathways for essential nutrients which they can obtain from their host but have retained the capacity to synthesize folate. In most bacteria, synthesis of the pterin moiety of folate requires the FolEQBK enzymes, while synthesis of the PABA moiety is carried out by the PabABC enzymes. Bioinformatic analyses reveal that while members of Chlamydia are missing the genes for FolE (GTP cyclohydrolase) and FolQ, which catalyze the initial steps in de novo synthesis of the pterin moiety, they have genes for the rest of the pterin pathway. We screened a chlamydial genomic library in deletion mutants of Escherichia coli to identify the "missing genes" and identified a novel enzyme, TrpFCtL2, which has broad substrate specificity. TrpFCtL2, in combination with GTP cyclohydrolase II (RibA), the first enzyme of riboflavin synthesis, provides a bypass of the first two canonical steps in folate synthesis catalyzed by FolE and FolQ. Notably, TrpFCtL2 retains the phosphoribosyl anthranilate isomerase activity of the original annotation. Additionally, we independently confirmed the recent discovery of a novel enzyme, CT610, which uses an unknown precursor to synthesize PABA and complements E. coli mutants with deletions of pabA, pabB, or pabC. Thus, Chlamydia species have evolved a variant folate synthesis pathway that employs a patchwork of promiscuous and adaptable enzymes recruited from other biosynthetic pathways. Importance: Collectively, the involvement of TrpFCtL2 and CT610 in the tetrahydrofolate pathway completes our understanding of folate biosynthesis in Chlamydia. Moreover, the novel roles for TrpFCtL2 and CT610 in the tetrahydrofolate pathway are sophisticated examples of how enzyme evolution plays a vital role in the adaptation of obligately intracellular organisms to host-specific niches. Enzymes like TrpFCtL2 which possess an enzyme fold common to many other enzymes are highly versatile and possess the capacity to evolve to catalyze related reactions in two different metabolic pathways. The continued identification of unique enzymes such as these in bacterial pathogens is important for development of antimicrobial compounds, as drugs that inhibit such enzymes would likely not have any targets in the host or the host's normal microbial flora.

INTRODUCTION

Chlamydiae are obligately intracellular bacteria that cause a wide variety of infectious diseases in humans and animals. The bacteria infect epithelial cells and cause diseases ranging from pneumonia to pelvic inflammatory disease to blinding trachoma (1). Chlamydia trachomatis infection is the most commonly reported bacterial sexually transmitted disease in the United States, with a total of 1,422,976 cases of genital chlamydial infection reported during 2012 (2). In addition to being an important sexually transmitted pathogen, C. trachomatis is the causative agent of trachoma, a chronic follicular keratoconjunctivitis, characterized by scarring of the eyelid and corneal opacities. The most recent estimates (from 2008) suggest that there are about 40 million people with active trachoma and that 1.3 million people are blinded as a result of this disease (3). The obligately intracellular organism Chlamydia is notable for its unusual dimorphic life cycle. The infectious form of the organism, the elementary body (EB), is small and displays limited metabolic activity (4). Once inside the host cell, the EB begins a developmental cycle and undergoes conversion into the metabolically active noninfectious reticulate body (RB) within 2 to 4 h. RBs replicate by binary fission and then convert back into the EB form after about 24 to 36 h. Lysis of the infected host cell occurs after 48 to 72 h, depending on the Chlamydia species, and released EBs go on to infect neighboring cells.

Bacterial evolution that leads to obligately intracellular parasitism in a eukaryotic host is thought to be associated with loss of genetic information, especially for genes that become redundant within the host niche (5). These genetic losses are not deleterious for the organism provided that the missing biosynthetic functions can be compensated for by increased transport functions that allow the organism access to essential nutrients found in the intracellular environment. Conversely, metabolic pathways which are either uniquely found in these bacteria or are critical to their intracellular growth are expected to persist despite reductive evolution. Obligately intracellular parasites, like Chlamydia, are dependent on their host cell for nutrients, as they have eliminated many redundant biosynthesis genes.

Folates are tripartite molecules comprising pterin, para-aminobenzoate (PABA), and glutamate moieties to which one-carbon units at various oxidation levels can be attached at the N5 and N10 positions (Fig. 1A). Tetrahydrofolates (THFs) are essential cofactors required by all organisms for DNA and amino acid synthesis and are obtained through either de novo synthesis or transport. Most bacteria, plants, and fungi and some protozoa possess a pathway for de novo synthesis of THF (6, 7), whereas vertebrates do not have a pathway for synthesis of folates and must obtain them through their diet. The pathway for THF synthesis involves 10 enzymes: folEQBK encode enzymes required for formation of the pterin moiety; pabA, pabB, and pabC encode enzymes required for formation of the PABA moiety; and folPCA encode enzymes required for the ligation of the pterin and PABA precursors and the glutamylation and reduction steps (see Fig. 1 for enzyme names and abbreviations and reactions catalyzed) (6).

FIG 1 

The canonical tetrahydrofolate biosynthetic pathway. Abbreviated enzyme names are in bold, with the homologs in C. trachomatis serovar D listed in red. A question mark is present if the Chlamydia homolog is not known. (A) GTP to tetrahydrofolate. FolE, GTP cyclohydrolase I; FolQ, dihydroneopterin triphosphate pyrophosphohydrolase; Ptase, nonspecific phosphohydrolase; FolB, dihydroneopterin aldolase; FolK, 6-hydroxymethyl-7,8-dihydropterin pyrophosphokinase; FolP, dihydropteroate synthase; FolC, dihydrofolate synthetase; FolA, dihydrofolate reductase. Full metabolite names: 7,8-dihydroneopterin triphosphate (H2NTP), 7,8-dihydroneopterin monophosphate (H2NMP), 7,8-dihydroneopterin (H2N), 6-hydroxymethyl-7,8-dihydroneopterin (HMH2N), 6-hydroxymethyl-7,8-dihydroneopterin diphosphate (HMH2NDP), 7,8-dihydropteroate (H2Pt), 7,8-dihydrofolate (DHF), tetrahydrofolate (THF). (B) PABA biosynthesis. PabA, aminodeoxychorismate synthase component II; PabB, aminodeoxychorismate component I; PabC, aminodeoxychorismate lyase.

Unlike other intracellular bacteria, which salvage THF or precursors from their hosts, many Chlamydia species are not folate auxotrophs. C. trachomatis L2, Chlamydia psittaci 6BC, and C. psittaci Cal10 synthesize folates de novo and possess FolA and FolP activity (8). Furthermore, C. trachomatis is sensitive to antibiotics that target enzymes in the folate biosynthetic pathway (8). We previously conducted a comparative genomic analysis of tetrahydrofolate biosynthesis genes in 500 microbial genomes (with only 10 Chlamydiaceae genomes) and reported that the Chlamydia species contained homologs of the folBKPA genes in a physical cluster (9). Proteomic analyses detected FolP in C. trachomatis (10, 11) and Chlamydia pneumoniae (12). However, genes encoding orthologs of FolE and FolQ, the first two enzymes of the pterin branch, as well as PabABC, the three PABA pathway enzymes, and FolC, the glutamylation enzyme, could not be identified in any of the 10 genomes analyzed at the time.

The mystery of the missing folC gene was solved by the identification of a gene now called folC2 (ct611 in C. trachomatis D/UW-3/CX) located in the chlamydial folate biosynthesis cluster. FolC2 is not a member of the FolC/COG0285 family but part of the COG1478 family, which contains enzymes involved in the glutamylation of the archaeal cofactor F420. The activity of this alternate folylglutamate synthase was demonstrated by complementation of an Escherichia coli ΔfolC mutant by a plasmid harboring ct611 (9). This was the first case of nonorthologous displacement in the folate pathway discovered in Chlamydia species.

In E. coli, the folE gene encodes GTP cyclohydrolase I (GCYH-I; EC 3.5.4.16), which catalyzes the conversion of GTP to 7,8-dihydroneopterin triphosphate (H2NTP). In some Bacteria and Archaea, the same reaction is catalyzed by another family of GTP cyclohydrolase I enzymes, the IB family (13), but no members of this second family are found in chlamydiae (9). This suggests that chlamydiae either recruit another enzyme family to perform this reaction or that these organisms scavenge H2NTP from the human host; H2NTP is the only intermediate of the folate pathway found in humans, where it serves as the first intermediate in biopterin biosynthesis (14). Similarly, the source of PABA in chlamydiae remained unidentified until recently, as the absence of the PABA synthesis genes suggests either a de novo PABA synthesis pathway in these organisms or import of PABA from the host. It was found that two members of the COG5424 family, NE1434 from Nitrosomonas europaea and CT610 from C. trachomatis D/UW-3/CX, are involved in a novel PABA synthesis pathway (15). While the precursor used to make PABA remains unknown, NE1434 and CT610 can complement an E. coli ΔpabABC mutant.

In this work, we show an example of Chlamydia’s metabolic plasticity with the discovery of a novel route to initiate folate synthesis that combines the first enzyme of riboflavin biosynthesis with an enzyme of tryptophan synthesis to make the folate pathway intermediate 7,8-dihydroneopterin monophosphate (H2NMP) rather than using the first two canonical enzymes, FolE and FolQ. Additionally, we independently show that CT610 is involved in PABA synthesis, confirming another example of the folate pathway variation found in Chlamydia. Collectively, these findings complete our understanding of folate biosynthesis in these human pathogens, confirm their ability to synthesize folates, and provide an elegant example of enzyme evolution to create new vitamin pathways (16).

RESULTS

Comparative genomic analysis of THF biosynthesis genes in chlamydiae.

The distribution and physical location of folate biosynthesis genes were analyzed with the 43 Chlamydia genomes in the SEED database (17) and is available in the “Folate Tryp Chlamydia” subsystem. Strict conservation of the folate gene cluster folBPKA, folC2, and ct610 was observed in all genomes analyzed and includes no unrelated gene, with the exception of Chlamydia pecorum, which is missing the whole cluster and corresponding genes (Fig. 2). The clusters can be separated into two subgroups: one downstream of rpoN in the C. trachomatis clade and one downstream of recA in the Chlamydophila clade. folK and folP are fused into a single gene in all chlamydiae analyzed. The intergenic region between ct610 and the gene immediately upstream differed slightly between C. trachomatis serovar L2 (3 bp) and C. trachomatis serovar D and Chlamydia muridarum (21 bp), suggesting that ct610 (and its orthologs) might be part of a single transcript of the folate gene cluster. Primers (see Table S1 in the supplemental material) designed to amplify the intergenic region between open reading frames (ORFs) in the folate cluster (folBPKA, folC2, and ctl0874—a homolog of ct610 in C. trachomatis D) of C. trachomatis L2 434/Bu were used to determine transcriptional organization. Reverse transcription-PCR (RT-PCR) analysis revealed that the genes in the folate cluster are transcribed as a single multigenic message which includes ctl0874 (see Fig. S1 in the supplemental material). Here, we refer to ctl0874 with the C. trachomatis D locus name, ct610, for consistency with the literature.

FIG 2 

Clustering of folate pathway genes in Chlamydiae and other intracellular bacteria. The gene organization of Chlamydia spp. and the phylogenetically distant Wolbachia is shown. Numbers in parentheses are numbers of genomes analyzed. ct609 to ct615 are locus tags for C. trachomatis serovar D. They are included for clarity, as several of these loci are cited in previous publications.

Experimental validation of activity of the C. trachomatis FolB, FolK, and FolP orthologs.

Aside from the initial steps of pterin synthesis, the remainder of the folate pathway appears to be intact in most Chlamydia species (Fig. 2) and encoded on an operon (see above). However, the functions of these enzymes have yet to be verified. CT614 of C. trachomatis is currently annotated as a FolX, which suggests that it is a 7,8-dihydroneopterin triphosphate epimerase involved in the biosynthesis of tetrahydromonapterin (18). Since Chlamydia species do not have a pathway for tetrahydromonapterin, CT614 is more likely FolB, the dihydroneopterin aldolase involved in tetrahydrofolate biosynthesis, which would be in agreement with the clustering of ct614 with the other folate synthesis genes (Fig. 2). In order to test this hypothesis, ct614 was cloned into pUC18 (pNEA127), expressed in E. coli ΔfolB, and demonstrated to complement the E. coli ΔfolB mutant on LB in the absence of added thymidine (Fig. 3A). These results are consistent with CT614 functioning as a dihydroneopterin aldolase.

FIG 3 

Activity of the C. trachomatis FolB, FolK, and FolP orthologs. (A) FolB complementation by either folB (pNEA122) or folX (ct614; pNEA127). (B) Complementation of E. coli ΔfolK and ΔfolP. (Top) E. coli ΔfolK is not complemented by any genes in the entire folate cluster of C. trachomatis, as shown by lack of growth on M9 minimal medium without thymidine (dT). (Bottom) E. coli ΔfolP is complemented by the folate cluster of C. trachomatis, presumably by ct613, which encodes FolKP.

ct613, the next gene in the C. trachomatis folate gene cluster, is annotated as a folK-folP fusion. To verify that the C. trachomatis FolKP has both 7,8-dihydro-6-hydroxymethylpterin-pyrophosphokinase (FolK) and dihydropteroate synthase (FolP) activities, pNEA57, a plasmid expressing the whole folate operon from C. trachomatis L2 (i.e., folB-folKP-folA-folC2-ct610), was used to transform E. coli ΔfolK::kan and E. coli ΔfolP::kan, neither of which grows on LB without supplemental thymidine. Interestingly, only the E. coli ΔfolP mutant was complemented by the C. trachomatis folate operon, while E. coli ΔfolK was not (Fig. 3B). Based on these complementation experiments, the FolKPCtL2 fusion protein as expressed in E. coli can carry out FolP activity but not FolK activity, suggesting that CT613 cannot capture the 6-hydroxymethyl-7,8-dihydroneopterin (HMH2N) precursor from E. coli FolB.

Library screen for the Chlamydia homolog of the folE gene of E. coli.

E. coli ΔfolE mutants require thymidine, pantothenic acid, adenine, and Casamino Acids for growth on minimal medium. We screened a library of C. trachomatis L2 DNA in the E. coli ΔfolE::kan mutant P1-7B and selected on minimal medium containing all the required supplements but thymidine. Over 150 transformants that grew on minimal medium without thymidine were recovered in three independent transformations.

Fifteen complementing clones chosen for characterization contained inserts that ranged in size from 3.2 to 9.2 kb. The minimum sequence present in all of the inserts contained a full-length open reading frame annotated as trpF (ctl0581), encoding N′-5′-phosphoribosyl anthranilate (PRA) isomerase (EC 5.3.1.24) with its native promoter as well as the tRNAMet gene just downstream of trpF (Fig. 4A). Two transformants, A4 and C1, were chosen for further characterization. We infer that the trpF in both clones is driven by its native promoter, as trpF in A4 is in the opposite orientation to the plasmid vector promoter, P, and trpF in C1 is more than 1 kb from the vector promoter. Each clone was retransformed into the ΔfolE::kan mutant, and plating efficiency experiments were carried out to determine phenotypic complementation. The results shown in Fig. 4A and in Table S2 in the supplemental material demonstrate that both inserts complemented the folE mutation and that complementation was independent of the vector promoter. The trpF genes from C. trachomatis L2, C. muridarum, and Chlamydia caviae were cloned into pUC18 and tested for complementation of the folE mutation in P1-7B. As shown in Fig. 4B, all trpF alleles tested complemented the folE mutation, albeit with different efficiencies. In E. coli the trpF gene is fused to the trpC gene, encoding indole-3-glycerol phosphate synthase (EC 4.1.1.48). In contrast to the Chlamydia trpF clones, neither the E. coli trpCF gene nor the E. coli trpC or trpF domains expressed alone showed any complementation activity in the ΔfolE mutant (Fig. 4B).

FIG 4 

Phylogenomic context and analysis of trpF genes in Chlamydia. (A) Alignment of library inserts that complemented ΔfolE. Library clones A4 and C1 (highlighted in blue) were used to test efficiency of complementation on M9 minimal medium containing pantothenic acid (1 µg/ml), adenine (40 µg/ml) and Casamino Acids (0.5%). (B) Complementation of E. coli ΔfolE and ΔtrpF mutants with trpF genes from members of the Chlamydiaceae.

Sequence and functional analysis of TrpFCtL2 as a PRA isomerase.

Alignments of TrpF reveal that C. trachomatis L2 TrpF is 98 to 100% identical at the amino acid level to TrpF of the other C. trachomatis ocular and genital serovars and 78% identical to TrpF of C. muridarum, a mouse pathogen, while identity drops to ~48% for the other animal pathogens, C. caviae, C. pecorum, and Chlamydia felis. Moreover, the Chlamydia trpF orthologs differ in genomic contexts, including both gene content and neighborhood. C. pneumoniae, C. psittaci, and C. abortus lack all the trp genes, including trpF, whereas C. trachomatis species have lost the trpC and trpD genes. Thus, the evolutionary history of trpF in Chlamydia seems to have followed a complex path involving genome decay and rearrangements (Fig. 5) coupled with the acquisition of a novel function to serving as a FolE in the folate biosynthesis pathway. Given the ability of Chlamydia trpF orthologs to complement E. coli ΔfolE, we next addressed if evolution of this new enzymatic function was accompanied by a loss of PRA isomerase activity.

FIG 5 

Phylogenetic tree of TrpF homologs and comparative genomics of trpABCDEFGRL and folE genes. The phylogenetic tree was constructed using the WAG model (42). The Verrucomicrobia and Lentisphaera group conserved folE (red clade), while the Chlamydiae group lost folE (green clade and red star). In Chlamydiae, a subclade consisting of C. trachomatis and C. muridarum possess the trpF in a different genomic context (blue star). Genes and their positions are shown in arrows, genes of unknown function are marked in white, multiple genes occurring in both directions are marked with diamonds, and the number inside the diamond indicates how many genes are predicted in this category. Two diagonal black lines indicate that the distance between trp genes and folE spans >15 genes.

To test PRA isomerase (TrpF) activity, a trpCF deletion mutant of E. coli (ATM932) was transformed with a plasmid containing the trpF gene from Chlamydiaceae alone or cotransformed with a plasmid containing the trpC domain of the E. coli trpCF gene (pAM238::trpC, pNEA67), as a trpCF deletion can be complemented when the E. coli trpC and trpF domains are expressed from separate plasmids (pNEA67 and pNEA61, respectively). While the E. coli trp (trpF) domain and the trpF genes of Chlamydiaceae alone fail to complement an E. coli ΔtrpCF mutant, the trpF genes of Chlamydiaceae fully complement the ΔtrpCF mutant when expressed with pNEA67, which contains the E. coli trpC domain (Fig. 4B; also, see Table S3 in the supplemental material). Thus, the Chlamydiaceae trpF encodes a functional PRA isomerase which can act together with the E. coli indole-3-glycerol phosphate synthase (TrpC) to restore tryptophan prototrophy to a ΔtrpCF mutant of E. coli. As expected, the E. coli folE gene (pNEA50) failed to complement the ΔtrpCF mutant even when expressed from pUC19, a high-copy-number plasmid (see Table S3 in the supplemental material).

The PRA isomerase activity was further confirmed using an E. coli ΔtrpF mutant (strain FBG-Wf) which retains the TrpC domain (19, 20). As expected, pASK::trpFCtL2 complemented the ΔtrpF mutation and rescued the tryptophan auxotrophy in vivo (data not shown). In vitro assays were hindered by difficulty in obtaining sufficient amounts of pure TrpFCtL2 protein and a poor signal-to-noise ratio of the enzyme assay, which prevented us from obtaining enzyme kinetics for the conversion of PRA to 1-(o-carboxyphenylamino)-1′-deoxyribulose 5′-phosphate (CdRP). Alternatively, use of cell-free extracts revealed that TrpFCtL2 did indeed catalyze the conversion of PRA to CdRP, but at a level lower than that of the PRA isomerase activity of PriA from Streptomyces coelicolor, which was used as a positive control (see Fig. S2 in the supplemental material).

Biochemical assays for GTP cyclohydrolase I (GCYH-I) activity.

After it was observed that TrpFCtL2 complemented an E. coli ΔfolE mutant, in vitro GCYH-I activity was assessed using a standard fluorescence assay in which the H2NPT produced from GTP is first oxidized to neopterin (13). We observed a small but consistent enhancement of the fluorescent signal (data not shown) in cell-free extracts of E. coli Rosetta BL21(DE3) that expressed TrpFCtL2 as a fusion with the E. coli maltose-binding protein (pMAL-c4x::trpF, pNEA83). However, attempts to measure activity with either the semipurified (affinity) or purified (HiTrap Q; GE Healthcare) fusion protein were unsuccessful, suggesting that the enzyme lost activity or that an essential cofactor during purification or that the enzyme was not functioning as a GCYH-I and may instead catalyze a different but related reaction.

The GCYH-I reaction is especially complex and requires multiple steps (Fig. 6A) (21). The first half of the reaction comprises two sequential hydrolysis reactions that result in purine ring opening and release of formic acid to give 2,5-diamino-6-ribosylamino-4(3H)-pyrimidinone 5′-triphosphate (compound II) (Fig. 6A), which subsequently undergoes ribosyl ring opening, an Amadori rearrangement, and ring closure in the second half of the reaction to give H2NTP. Similar ribosyl ring opening and Amadori rearrangements are also catalyzed by PRA isomerase (TrpF), as well as by the N′-[(5′-phosphoribosyl)formimino]-5-aminoimidazole-4-carboxamide ribonucleotide (ProFAR) isomerase (HisA; EC 5.3.1.16) (22) (Fig. 6B).

FIG 6 

Biochemical reactions catalyzed by FolE and GTP cyclohydrolase I (A) and by TrpF and HisA (B).

Notably, while compound II is an intermediate in the GCYH-I reaction, the monophosphate derivative of compound II, 2,5-diamino-6-ribosylamino-4(3 H)-pyrimidinone 5′-phosphate (compound V) (Fig. 7A), is the product of the GCYH-II enzyme, which in E. coli is designated RibA and is involved in flavin biosynthesis. Since all sequenced chlamydial species possess a RibA ortholog (CT731), we reasoned that the failure to observe GCYH-I activity with TrpFCtL2 might be because it instead functions in the chlamydial folate pathway by converting compound V to H2NMP. The absence of an identifiable folQ gene that clusters with other folate biosynthetic genes in chlamydial genomes is also consistent with this hypothesis, as a pathway that utilized compound V as the precursor to the pterin system would bypass H2NTP by producing H2NMP directly, obviating FolQ.

FIG 7 

Formation of H2NMP from GTP by RibA and TrpFCtL2. (A) Reactions mediated by RibA and TrpFCtL2. (B) Fluorescence assay for H2NMP formation from compound V by TrpFCtL2. Assays were carried out as described in Materials and Methods. ●, TrpFCtL2; ○, BSA; □, no added protein. (C) LC-MS analysis of TrpFCtL2 activity. Total ion chromatograms with mass spectra of peaks at 6.5 min (insets) of authentic 7,8-dihydroneopterin (a), a control assay with no TrpFCtL2 (b), and an assay with TrpFCtL2 (c) are shown. (D) MS/MS analysis of the m/z 254 ion produced in the LC-MS experiments whose results are shown in panel C. (a) Authentic 7,8-dihydroneopterin assay; (b) assay with TrpFCtL2.

To test if TrpFCtL2 catalyzes the conversion of compound V to H2NMP, we carried out coupled assays with purified recombinant E. coli RibA (22, 23) and recombinant TrpFCtL2. In contrast to the assays containing only GTP, a distinct fluorescent signal was observed in the RibA-coupled assays that was both time and TrpFCtL2 dependent (data not shown). To confirm that the signal was in fact due to the conversion of compound V to H2NMP, compound V was produced and purified in large-scale RibA reactions and used in TrpFCtL2 reactions as a putative substrate. Again, only the TrpFCtL2 assays revealed a distinct fluorescent signal, whereas controls that contained no protein or BSA in place of TrpFCtL2 did not (Fig. 7B).

Further analysis of the formation of H2NMP from compound V by TrpFCtL2 was investigated by liquid chromatography-mass spectrometry (LC-MS) (Fig. 7C). In the total ion chromatogram of the TrpFCtL2 assay, a peak with the same retention time as authentic dihydroneopterin was observed, which was absent from the negative control (Fig. 7C), and both peaks possessed a molecular ion (m/z 254.08892 versus m/z 254.08903 for the authentic standard) consistent with dihydroneopterin. Furthermore, tandem MS (MS/MS) analysis of the m/z 254 ion produced an identical fragment ion at m/z 194 in both samples (Fig. 7D).

CT610 functionally replaces PabA, PabB, and PabC in E. coli.

Because of the strict linkage of ct610 and its orthologs with the folate biosynthesis genes in both Chlamydia and the phylogenetically distant Wolbachia (Fig. 2), it was highly probable that this enzyme family was involved in folate biosynthesis; however, its role was a mystery when we began this study.

We looked for a gene which encoded the missing PABA synthesis enzyme by screening a library of C. trachomatis L2 DNA in ATM825, an E. coli ΔpabA::kan mutant, which requires PABA supplementation for growth on minimal medium. Transformants were selected for growth on minimal medium without PABA supplementation. The complementing clone that was isolated contained nucleotides 28 to 696 of the ctl0874 gene, which is a homolog of ct610 in C. trachomatis serovar D. Here, we refer to ctl0874 with the C. trachomatis D locus tag ct610 for consistency with the literature. There is some discrepancy concerning the actual start site of ct610. Mass spectrometry (MS) analysis determined the N-terminal amino acid sequence to be MNFLDQLDLI, indicating a translation start 15 nucleotides downstream of the start site predicted by genomics data at STDgen (http://stdgen.northwestern.edu/) and NCBI (http://www.ncbi.nlm.nih.gov). We cloned the ct610 gene containing the MS-predicted start site into pUC19 (yielding pAJM96) and with the genomics-predicted translation start site into pUC19 (yielding pAJM95). Plasmids containing either of the predicted starts for CT610 complemented the PABA auxotrophy of the E. coli ΔpabA mutant (VDC9500) (Fig. 8A). More importantly, CT610 was also able to complement the PABA auxotrophy of E. coli ΔpabB and ΔpabC mutants (VDC9502 and VDC9504) as well (Fig. 8B), indicating that CT610 is able to make PABA by an alternative route that does not involve the PabA, PabB, or PabC enzymes. Thus, chlamydiae possess a novel pathway for the synthesis of PABA.

FIG 8 

CT610 complements PABA auxotrophy of E. coli PABA mutants. (A) Complementation of VDC9500 (ΔpabA::kan) with either pabA or ct610. (B) Complementation of VDC9502 (ΔpabB::kan) and VDC9504 (ΔpabC::kan) with CT610. All cultures were washed twice in M9 medium without PABA and normalized to an OD600 of 1 in M9 medium without PABA, and 10 µl of 10-fold serial dilutions was spotted onto LB agar, M9 glucose agar, and M9 glucose agar with PABA. Arabinose (0.1% [wt/vol]) was added to plates for complementation of VDC9504 for induction of pBAD24::pabC. All plates contained 100 µg/ml ampicillin.

Search for the potential CT610 substrate.

Chorismate is the precursor for the PABA moiety of folate in the canonical pathway (Fig. 1B). To test if chorismate is also the precursor for PABA via the CT610-mediated reaction, an E. coli ΔaroA mutant was constructed (ATM816). AroA carries out the penultimate step of chorismate biosynthesis (24); thus, aroA mutants are unable to produce chorismate, and growth in minimal medium requires supplementation with the aromatic amino acids, 4-hydroxybenzoate, and PABA. CT610 was still able to complement an E. coli ΔaroA mutant with 72.6% efficiency, allowing growth on M9 medium supplemented with tryptophan, tyrosine, phenylalanine, and 4-hydroxybenzoate but lacking PABA (see Fig. S3 in the supplemental material).

Shikimate and 3-dehydroquinate were investigated next as possible substrates. Both are intermediates in the biosynthesis of chorismate and could serve as precursors for PABA in an alternative pathway that branches earlier than chorismate. In the canonical pathway, 3-dehydroquinate synthase (AroB) catalyzes the cyclization of 3-deoxy-d-arabino-heptulosonate-7-phosphate to 3-dehydroquinate, which is then converted to 3-dehydroshikimate by 3-dehydroquinate dehydratase (AroD). Reduction of 3-dehydroshikimate to shikimate is catalyzed by shikimate dehydrogenase (AroE). CT610 was able to complement both E. coli ΔaroD ΔpabA (VDC9510) and E. coli ΔaroB ΔpabA (VDC9598) mutants (see Fig. S3 in the supplemental material), allowing growth on M9 supplemented with either shikimic acid or just the aromatic amino acids but lacking PABA. CT610 therefore appears to utilize a molecule other than shikimate, 3-dehydroquinate, chorismate, or any intermediate in this pathway as a precursor for PABA.

Since intermediates in the chorismate biosynthetic pathway are not the substrates for CT610, we tested the aromatic compound ubiquinone and its precursor, 4-hydroxybenzoate. Chorismate pyruvate lyase (UbiC) catalyzes the first step of ubiquinone biosynthesis, the conversion of chorismate to 4-hydroxybenzoate (25). An E. coli ΔubiC mutant can grow on glucose as a sole carbon source without 4-hydroxybenzoate or ubiquinone supplementation. To test if CT610 can utilize ubiquinone or 4-hydroxybenzoate as a precursor for PABA, an E. coli ΔubiC ΔpabA mutant was constructed which does not make ubiquinone or PABA (ATM851). Similar to the observations with the chorismate pathway mutants, expression of CT610 in E. coli ΔubiC ΔpabA allowed the strain to grow on M9 with glucose in the absence of PABA, indicating that CT610 does not use ubiquinone or any of the intermediates from chorismate to ubiquinone as a precursor for PABA (see Fig. S3 in the supplemental material).

DISCUSSION

Folate synthesis has long been a target for antimicrobial drug development because the pathway is found in bacteria but not in mammalian cells. C. trachomatis and C. psittaci synthesize folates de novo and are sensitive to sulfonamides, which target enzymes in the folate synthesis pathway (8). However, metabolic reconstruction of the folate pathway in sequenced Chlamydia strains revealed many pathway holes (9). In this study, we performed a comprehensive examination of pathways for synthesis of pterin and PABA, two critical moieties that form folate, and solved the remaining mysteries in this area of metabolism in these pathogenic bacteria.

RT-PCR analysis of the C. trachomatis L2 folate gene cluster showed that the five contiguous genes are expressed as an operon (see Fig. S1 in the supplemental material). folA (8) and folC2 (9) had previously been shown to encode functional enzymes in the THF pathway. Genetic complementation experiments with E. coli mutants allowed us to demonstrate functionality of the chlamydial folB gene, encoding dihydroneopterin aldolase, and the dihydropteroate synthase (FolP) activity of the folKP gene. Our inability to show FolK activity in the folate operon despite an active FolP suggests that the chlamydial enzyme cannot accept the HMH2N precursor from E. coli FolB. More sophisticated biochemical analyses are required to formally validate the inferred chlamydial FolP activity. Thus, four of the five genes in the folate operon have now been assigned a role in THF synthesis. The results obtained by Satoh et al. (15) and our own data demonstrate that the last gene in the folate operon, ct610, is involved not in the pterin biosynthesis pathway but in the synthesis of PABA, the other component of THF.

One outstanding question was how Chlamydia initiates the pathway for THF biosynthesis in the absence of any ortholog of GCYH-I (FolE), the first enzyme in the canonical pathway. We used a combined genetic and biochemical strategy to reveal that TrpFCtL2 initiates a novel route to the formation of a precursor for THF synthesis. While TrpFCtL2 complemented an E. coli ΔfolE mutant, biochemical analysis revealed that purified recombinant TrpFCtL2 catalyzed the conversion of compound V to H2NMP. This observation, plus our inability to demonstrate conversion of GTP to H2NTP, leads us to conclude that TrpFCtL2 functions in Chlamydia not as a classic GCYH-I but instead as an isomerase that converts compound V, the product of GCYH-II, into H2NMP (Fig. 7A). Thus, in Chlamydia, the GCYH-II enzyme catalyzes the first step in both the flavin and folate pathways.

Given the mechanistic complexity of the GCYH-I-catalyzed reaction and the similar chemistry of the latter portion of this reaction to that of the reaction catalyzed by TrpF (Fig. 6), the observation that TrpFCtL2, catalyzes the latter half of the GCYH-I transformation is an elegant solution to the loss of a discrete GCYH-I activity in Chlamydia. Indeed, evolving the ability to convert available compound V to H2NMP in the context of TrpF catalysis is a far more likely process than evolving this ability together with purine ring hydrolysis and deformylation.

PRA isomerase (TrpF) is well known as a promiscuous and evolvable enzyme (26) and is the paradigm of an adaptable enzyme (27). (βα)8-barrel fold enzymes like PRA isomerase are catalytically versatile and excellent candidates for evolutionary selection of diverse activities. For example, a HisA homolog in Actinobacteria possesses dual-substrate specificity of both HisA and TrpF (28, 29). This new isomerization reaction is another example of the plasticity of the TrpF fold that has already been recruited to perform different types of isomerization reactions (Fig. 6) (29). In the case of Chlamydia TrpF, PRA isomerase activity is retained, allowing C. caviae, C. felis, and C. pecorum to carry out the full tryptophan synthesis pathway starting with the anthranilate precursor. Thus, promiscuous enzymes with broad substrate specificity may play a significant role in reductive evolution with far-reaching metabolic implications.

The tryptophan repressor, TrpR, represses trp operon expression when tryptophan is plentiful. In C. trachomatis L2, trpF is not regulated by TrpR (30), but trpF in C. caviae, C. pecorum, and C. felis is in the trp operon and under TrpR control. It is possible that under tryptophan-replete conditions, trpF expression is repressed in these strains. However, it is likely that the level of trpF expression needed for folate synthesis is lower than what is needed for tryptophan synthesis and basal (uninduced) levels of trpF expression are probably sufficient for folate production.

An unresolved question is that of the mechanism of action of CT610. It has been implicated as a type III secreted factor capable of triggering apoptosis in host cells (31). In addition, CT610 is similar to PqqC (32), which catalyzes a ring cyclization and eight-electron oxidation in the final step of pyrroloquinoline quinone (PQQ) biosynthesis. However, the active site residues of PqqC are not conserved in CT610, and we demonstrate here that CT610 can functionally replace PabA, PabB, and PabC, indicating that it acts as a novel PABA synthase. Similarly, Satoh et al. recently reported that both CT610 and its homolog in Nitrosomonas europaea function as PABA synthesis enzymes capable of replacing PabA, PabB, and PabC in E. coli (15). Like the chlamydiae, N. europaea does not contain homologs of pabA, pabB, or pabC, yet it possesses the remainder of the tetrahydrofolate biosynthesis enzyme genes.

Based on the structure of CT610 (32), we predicted that its substrate was likely an aromatic or similar cyclic compound such as chorismate, an intermediate of PABA synthesis, or a downstream product derived from chorismate (i.e., ubiquinone, aromatic amino acids). We tested chorismate, shikimate, 3-dehydroquinate, and ubiquinone as potential precursors to PABA synthesis by genetic complementation of a series of mutations (ΔaroA, ΔaroD, ΔaroB, and ΔubiC) in either wild-type E. coli or an E. coli ΔpabA mutant (a PABA auxotroph), and we found that, in each case, CT610 restored growth in the absence of PABA. Our results are similar to the results with the N. europaea CT610 homolog, confirming that CT610 and its homologs are indeed novel PABA synthases. However, the identity of the substrate used by CT610 to form PABA remains unknown.

CT610 is classified in the superfamily of heme oxygenases (32). Other enzymes in this superfamily catalyze a diverse array of reactions, and include those involved in thiamine salvage (TenA), thiazole biosynthesis (Thi4), and PQQ biosynthesis (PqqC). Interestingly, Thi4p of Saccharomyces cerevisiae has been shown to be a suicide thiamine thiazole synthase which, in contrast to the five enzymes required by bacteria to produce thiamine thiazole (33), forms the thiazole moiety in a single step. Thi4p acts as a cosubstrate for the formation of thiazole through a complex reaction in which a cysteine residue of Thi4p serves as the sulfur source (33). It is intriguing to speculate that CT610 may similarly be a single-turnover enzyme, serving as a cosubstrate to produce PABA. In such a scenario, the conserved lysine residue at position 152 within the predicted binding pocket of CT610 might serve as the amino source for the production of PABA. However, testing of this hypothesis must await the identification of the PABA precursor.

In summary, folate biosynthesis is fully functional in most Chlamydia species, but the pathway is a patchwork of recruited enzymes. FolC2, an enzyme of archaeal origin associated with glutamylation of the F420 cofactor, has been recruited to perform the folate glutamylation reactions (9). The biosynthesis of H2NMP, an early intermediate in the pathway, is performed by the first enzyme of riboflavin biosynthesis (RibA) together with a PRA isomerase (TrpFCtL2) showing broad substrate specificity. Finally, the PABA moiety appears to be synthesized via a unique route using CT610, an enzyme homologous to PqqC, an enzyme involved in the biosynthesis of pyrroloquinoline quinone (PQQ). However, CT610 has no validated role in PQQ synthesis in Chlamydia. Clearly, folate biosynthesis in Chlamydia provides another example of the remarkable metabolic versatility and ingenuity of the species. The unique nature of the Chlamydia folate synthesis enzymes makes them ideal targets for development of highly specific antibacterial agents.

MATERIALS AND METHODS

Comparative genomics.

The BLAST tools (34) and resources at NCBI (http://www.ncbi.nlm.nih.gov/) were routinely used. Sequence alignments were built using ClustalW (35) or Multialin (36). Protein domain analysis was performed using the Pfam database tools (http://pfam.xfam.org/) (37). Analysis of the phylogenetic distribution was performed on the public SEED server (http://pubseed.theseed.org/SubsysEditor.cgi) (38). Physical clustering was analyzed with the SEED subsystem coloring tool or the Seedviewer Compare region tool (38) as well as on the MicrobesOnline tree-based genome browser (http://www.microbesonline.org/) (39).

Phylogenetic reconstructions.

Sequence alignments were done using MUSCLE from within the software SEAVIEW (40). ProtTest (41) was used to define the fittest model for the TrpF phylogenetic reconstruction, and reconstruction was done using a WAG model (42) followed by a Bayesian analysis using MrBayes 3.2.1 (43). Branch supports were estimated using 1,000,000 bootstrap replications.

Bacterial strains and growth conditions.

Strains of E. coli K-12 used in this study are listed in Table 1, and plasmids are listed in Table 2. Construction of strains and plasmids is described in the supplemental material. Strains were grown in Luria-Bertani (LB) medium or M9 minimal medium with aeration or on agar. Medium was supplemented with ampicillin (100 µg/ml), kanamycin (50 µg/ml), chloramphenicol (25 µg/ml), spectinomycin (100 µg/ml), thymidine (75 µg/ml), pantothenate (1 µg/ml), adenine (40 µg/ml), PABA (0.1 µg/ml), 4-hydroxybenzoate (2 µM), phenylalanine (10 or 40 µg/ml), tryptophan (40 µg/ml), tyrosine (40 µg/ml), or Bacto Casamino Acids (0.5%) as necessary, and cultures were incubated at 37°C. Chemicals were purchased from Sigma-Aldrich Company. Phosphate-buffered saline (PBS; Lonza) and buffered saline gelatin (BSG; 150 mM NaCl, 2 mM KH2PO4, 4 mM Na2HPO4, 0.01% gelatin) (44) were used as washes and diluents for bacterial cultures, respectively.

TABLE 1 

Escherichia coli K-12 strains used in this study

Strain	Genotype	Relevant auxotrophic requirements	Reference or source
MC4100	F− araD139 Δ(lacZYA-argF)U169 rpsL150 relA1 deoC1 pt F25 rbsR flbB5301		47
MG1655	F− rph-1 λ− Su°		48
BW25113	F− Δ(araD-araB)567 ΔlacZ4787(::rrnB3) lacIp-4000(lacIq) λ− rph-1 Δ(rhaD-rhaB)568 rrnB3 hsdR514pKD46: araBp-gam-bet-exo oriR101 repA101(Ts) bla		49
Rosetta BL21(DE3)	F− ompT hsdSB (rB− mB−) gal dcm (DE3) pRARE (cat)		EMD Millipore
C600 ΔfolK	C600 ΔfolK::tet	Thymidine, pantothenic acid, adenine, glycine, methionine	50
P1-7B	MG1655 ΔfolE::kan	Thymidine, pantothenic acid, adenine, glycine, methionine	51
ATM816	MG1655 ΔaroA::kan	Phenylalanine, tyrosine, tryptophan, 2,3-dihydroxybenzoate, PABA	This work
ATM825	MG1655 ΔpabA::kan	PABA	This work
ATM842	MG1655 ΔubiC::kan	No auxotrophic requirements on minimal	This work
ATM843	MG1655 ΔubiC	No auxotrophic requirements on minimal	This work
ATM851	MG1655 ΔubiC ΔpabA::kan	PABA, ubiquinone	This work
ATM932	MG1655 ΔtrpCF::cat	Tryptophan	This work
FBG-Wf	ΔtrpF	Tryptophan	19
VDC2250	MG1655 ΔfolP::kan	Thymidine, pantothenic acid, adenine, glycine, methionine	This work
VDC3276	MG1655 ΔfolB::kan	Thymidine, pantothenic acid, adenine, glycine, methionine	52
VDC9500	BW25113 ΔpabA::kan	PABA	This work
VDC9502	BW25113 ΔpabB::kan	PABA	This work
VDC9504	BW25113 ΔpabC::kan	PABA	This work
VDC9508	BW25113 ΔaroD::cat	Phenylalanine, tyrosine, tryptophan, ubiquinone, PABA	This work
VDC9510	BW25113 ΔaroD::cat ΔpabA::kan	Phenylalanine, tyrosine, tryptophan, ubiquinone, PABA	This work
VDC9579	MG1655 ΔaroB	Phenylalanine, tyrosine, tryptophan, ubiquinone, PABA	This work
VDC9598	MG1655 ΔaroB ΔpabA::kan	Phenylalanine, tyrosine, tryptophan, ubiquinone, PABA	This work

TABLE 2 

Plasmids used in this study

Name	Genotype	Description	Source or reference
pGEM-T		Cloning vector for PCR products	Promega
pUC18		Cloning vector with Plac, ColE1 ori, Apr	53
pUC19		Cloning vector with Plac, ColE1 ori, Apr	53
pAM238		Cloning vector with Plac, pSC101 ori, Spcr	54
pMAL-c4x		N-terminal maltose binding protein tag fusion expression vector; Ptac Apr	New England Biolabs
pASK::IBA3plus		Expression vector for C-terminal strep tag-fusion; Ptet Apr	IBA GmbH
pAJM94	pUC18::aroAEc	aroAEc from E. coli MG1655 with Plac; Apr	This work
pAJM95	pUC19::ct610	ct610 (ctl0874) from CtL2 with genomics-predicted start	This work
pAJM96	pUC19::ct610	ct610 (ctl0874) from CtL2 with proteomics-predicted start	This work
pASK		NdeI site of pASK::IBA3plus deleted, His tag from pET15b inserted into the EcoRI and HindIII sites	This work
pASK::ctl0581	pASK::ctl0581	trpFCtL2 (ctl0581) with Ptet, Apr	This work
pASK::priA_Scoe	pASK::priA_Scoe	priA from Scoe with Ptet, Apr	(28)
pJJT22	pUC19::pabAEc	pabA from E. coli MG1655 with Plac; Apr	This work
pJJT23	pUC19::pabBEc	pabB from E. coli MG1655 with Plac, Apr	This work
pJJT70	pBAD24::pabCEc	pabC from E. coli MG1655 with PBAD, Apr	This work
pNEA50	pUC19::folEEc	folEEc with native promoter, Apr	This work
pNEA57	pAM238::folCtL2 genes	Folate gene cluster (folX-folKP-folA-folC2-ct610) from CtL2 with native promoter, Spcr	This work
pNEA59	pUC18::trpCFEc	trpCFEc with Plac, Apr	This work
pNEA61	pUC18::trpFEc	trpFEc domain with Plac, Apr	This work
pNEA64	pAM238::trpFCc	trpFCc with Plac, Spcr	This work
pNEA65	pUC18::trpCCc	trpCCc with Plac, Apr	This work
pNEA67	pAM238::trpCEc	trpCEc domain with Plac, Apr	This work
pNEA69	pUC18::trpFCm	trpFCm with Plac, Apr	This work
pNEA71	pUC18::trpFCtL2	trpFCtL2 (ctl0581) with Plac, Apr	This work
pNEA72	pUC18::trpFCc	trpFCc with Plac, Apr	This work
pNEA79	pAM238::trpCEc	trpFEc from MC4100 with Plac, Spcr	This work
pNEA83	pMAL-c4x::trpFCtL2	N-terminal maltose binding protein-CTL0581 fusion with Ptac, Apr	This work
pNEA122	pUC18::folBEc	folBEc from MC4100 with Plac, Apr	This work
pNEA127	pUC18::folXCtL2	ct614 (ctl0878) from CtL2 with Plac, Apr	This work

Ap, ampicillin; Cc, C. caviae GPIC strain SP6; Cm, C. muridarum Nigg; CtL2, C. trachomatis L2; Ec, E. coli MC4100 or MG1655; Scoe, Streptomyces coelicolor; Spc, spectinomycin.

Library screen.

Freshly prepared P1-7B (E. coli ΔfolE::kan) and ATM825 (E. coli ΔpabA::kan) electrocompetent cells were transformed with ~300 ng of C. trachomatis L2 genomic library (45) and recovered in SOC supplemented with thymidine or PABA overnight at 25°C. After recovery, the culture was washed twice in PBS before plating on M9 medium containing ampicillin, pantothenic acid, adenine, and Casamino Acids for P1-7B and on M9 plates with ampicillin for ATM825. Transformation titers were determined by plating dilutions on M9 medium containing ampicillin, thymidine, pantothenic acid, adenine, and Casamino Acids for P1-7B and on M9 ampicillin PABA for ATM825. Transformants were single-colony purified twice before characterization of phenotype (efficiency of plating) and genotype (DNA sequence of library clone).

Efficiency of plating.

Cultures of P1-7B carrying plasmids of interest were grown overnight in LB containing ampicillin, thymidine, pantothenic acid, adenine, and Casamino Acids. All Δpab strains were grown in M9 medium containing ampicillin and PABA. ATM932 (E. coli ΔtrpCF::kan) was grown overnight in LB containing tryptophan, ampicillin, or spectinomycin as needed. All strains were diluted into fresh medium and grown to mid-log phase (optical density at 600 nm [OD600] = 0.5 to 0.7), washed in PBS, serially diluted in BSG, and then plated at 37°C on appropriate growth media.

Complementation drops.

Strains were grown overnight in LB containing ampicillin or spectinomycin, as needed, and supplemented with thymidine for the E. coli ΔfolB mutants. Strains were diluted into fresh medium, grown to an OD600 of ~1.0, washed in PBS, normalized to an OD600 of 1.0, and serially diluted in PBS. A 10-µl portion of each serial 10-fold dilution was spotted onto appropriate growth media.

RT-PCR.

L2 mouse fibroblasts were infected to a multiplicity of infection of 1 with C. trachomatis L2 434/Bu, and total RNA and DNA were harvested using TRIzol (Invitrogen) at 46 h postinfection. RNA was DNase I treated (Invitrogen) and reverse transcribed with random hexamers using the Thermoscript RT-PCR kit (Invitrogen). Total RNA, total DNA, and cDNA samples were analyzed with Taq polymerase (Fermentas).

Determination of phosphoribosylanthranilate (PRA) isomerase activity.

PRA isomerase activity of TrpFCtL2 was confirmed in an E. coli ΔtrpF mutant, FBG-Wf, as previously described (19, 20) using cell-free extracts of E. coli FBG-Wf carrying pASK::trpF, empty pASK as a negative control, and pASK::priA_Scoe, encoding a phosphoribosyl isomerase A from Streptomyces coelicolor which exhibits both TrpF and HisA activity, as a positive control (28). The cells were grown at 37°C for 72 h in 500 ml of M9 minimal medium without tryptophan, except for the negative control, where tryptophan was added, before induction of the P promoter of pASK with anhydrotetracycline (20 ng/ml). For the PRA isomerase enzyme assay, 62.5 µM of anthranilic acid and 2.6 µg/ml of total protein were used as previously reported (20).

HPLC assay for type I GTP cyclohydrolase activity.

Assays to identify GCYH-I activity of the TrpFCtL2 fusion protein were carried out as described previously (46), except that the reaction volume was 200 µl and solutions were degassed with N2 to minimize postreaction oxidation. Reaction mixtures containing Neisseria gonorrhoeae GCYH-1B (0.5 µM) as a positive control or TrpFCtL2 fusion protein (~5.0 µM) were incubated in the dark at 37°C for 60 min and then for an additional 60 min after the addition of five units of alkaline phosphatase. Protein was removed from the mixture using an Amicon centrifugal device, and the samples were analyzed by reversed-phase HPLC on a Gemini C18 (Phenomenex) column equilibrated in 200 mM ammonium acetate (pH 6.0), at a flow rate of 0.7 ml/min. The following acetonitrile solvent gradient was used to develop the column: 0 min, 0%; 10 min to 30 min, 0 to 30% gradient; 30 min to 40 min, 30 to 0% gradient; and 50 min, 0%.

Fluorescence assay for 7,8-dihydroneopterin formation.

Enzyme assays were run at 37°C in 100-µl reaction mixtures containing 1 mM GTP or 500 µM purified compound V (purification details of compound V are in the supplemental material), 100 mM HEPES (pH 8.0), 100 mM KCl, 1 mM dithiothreitol (DTT), 20 mM MgCl2, and either TrpFCtL2 fusion protein (400 µg/ml), BSA (400 µg/ml), or no protein. The reactions were run for 0, 15, 30, 45, or 60 min, before 12 µl of 1% I2 (wt/vol) and 2% KI (wt/vol) in 1 M HCl were added and incubated at room temperature in the dark for 15 min. Excess iodine was reduced by the addition of 6 µl of 2% ascorbic acid (wt/vol), and the samples were analyzed by fluorescence spectroscopy with a Gemini XPS fluorimeter (excitation at 365 nm, emission at 446 nm). A standard curve using authentic neopterin was generated in tandem with the analysis.

LC-MS analysis of type I GTP cyclohydrolase activity of TrpFCtL2.

GCYH assays were carried out as described above for HPLC analysis of TrpFCtL2 assays along with control assays containing BSA in place of the TrpFCtL2 fusion protein. After protein removal by an Amicon centrifugal device, the filtrate was concentrated by evaporation in a speed-vac. The samples were analyzed by an LCQ Orbi-trap mass spectrometer (Thermo) interfaced with an Accela HPLC system. The interface was operated in the negative mode scanning the range m/z 0 to 800, with the following instrument conditions: capillary voltage, −16 V; capillary temperature, 299°C; tube lens, −90 V; spray voltage, 9 V. The separation was carried out using a reversed-phase column (Phenomenex Gemini [5 μm particle] C18; 250 by 2.00 mm column) using a mobile phase of 20 mM ammonium acetate (pH 6.0) with an acetonitrile gradient at a flow rate of 0.3 ml/min. The injection volume was 10 µl. Authentic dihydroneopterin was analyzed under the same conditions.

RT-PCR analysis of C. trachomatis L2 folate gene cluster. Lines above the operon schematic indicate the intergenic areas amplified by PCR using C. trachomatis L2 cDNA as the template. The amplification of these intergenic regions indicates genetic linkage of the folate genes on a single transcript. Download

Figure S1, PDF file, 0.1 MB

PRA isomerase activity of TrpFCtL2. The TrpF reaction using 2.6 µg/ml of total protein was monitored essentially as described elsewhere (20). Download

Figure S2, PDF file, 0.1 MB

CT610 does not use shikimate, chorismate, or ubiquinone as a substrate. (A) Complementation of ATM816 (E. coli ΔaroA) with ct610. Strains expressing CT610 from pAJM95 are able to grow on M9-glucose with phenylalanine, tyrosine, and tryptophan and without PABA. (B) Complementation drops of VDC9510 (E. coli ΔaroD ΔpabA) with either pabAEc or ct610. Strains expressing CT610 from pAJM95 are able to grow on M9-glucose-Amp with phenylalanine, tyrosine, tryptophan, and without PABA. (C) Complementation drops of VDC9598 (E. coli ΔaroB ΔpabA) with ct610. Strains expressing CT610 from pAJM95 are able to grow on M9-glucose-Amp with shikimic acid, phenylalanine, tyrosine, and tryptophan and without PABA. (D) Complementation drops of ATM851 (E. coli ΔubiC ΔpabA) with ct610. Strains expressing CT610 from pAJM95 are able to grow on M9-glucose-Amp without PABA. Download

Figure S3, PDF file, 0.2 MB

SDS-PAGE of TrpFCtL2 purification. Lane 1, molecular weight markers; lane 2, cell extract from uninduced cells; lane 3, cell extract from induced cells; lane 4, after affinity purification on amylose resin; lane 5, after anion-exchange chromatography on a Hi-Trap Q column. Download

Figure S4, PDF file, 0.1 MB

List of oligonucleotides used in this study.

Table S1, PDF file, 0.1 MB.

Efficiency of plating of E. coli P1-7B (ΔfolE::kan) transformed with two representative clones selected from library screen.

Table S2, PDF file, 0.1 MB.

Complementation of a E. coli ΔtrpCF mutant with trpF genes from members of the Chlamydiaceae

Table S3, PDF file, 0.1 MB.

Supplemental information. Download

Text S1, PDF file, 0.1 MB