Ammonia generation by tryptophan synthase drives a key genetic difference between genital and ocular Chlamydia trachomatis isolates

Significance

Chlamydia trachomatis is a human bacterial pathogen that causes distinct pathologies upon infecting ocular and urogenital compartments. Previous studies have shown that all urogenital strains can express tryptophan synthase, an enzyme they use to synthesize tryptophan by salvaging indole produced by other bacterial species in the infection microenvironment. In stark contrast, all ocular strains of Chlamydia trachomatis lack tryptophan synthase, typically because of inactivating point mutations. Here, we suggest why ocular strains lose tryptophan synthase activity; activation of this enzyme in an indole-deficient environment, like the eye, results in the deleterious production of ammonia. By identifying the mechanism that underlies this effect, our findings provide strategies to target infections by Chlamydia and other bacteria.

Abstract

A striking difference between genital and ocular clinical isolates of Chlamydia trachomatis is that only the former express a functional tryptophan synthase and therefore can synthesize tryptophan by indole salvage. Ocular isolates uniformly cannot use indole due to inactivating mutations within tryptophan synthase, indicating a selection against maintaining this enzyme in the ocular environment. Here, we demonstrate that this selection occurs in two steps. First, specific indole derivatives, produced by the human gut microbiome and present in serum, rapidly induce expression of C. trachomatis tryptophan synthase, even under conditions of tryptophan sufficiency. We demonstrate that these indole derivatives function by acting as de-repressors of C. trachomatis TrpR. Second, trp operon de-repression is profoundly deleterious when infected cells are in an indole-deficient environment, because in the absence of indole, tryptophan synthase deaminates serine to pyruvate and ammonia. We have used biochemical and genetic approaches to demonstrate that expression of wild-type tryptophan synthase is required for the bactericidal production of ammonia. Pertinently, although these indole derivatives de-repress the trpRBA operon of C. trachomatis strains with trpA or trpB mutations, no ammonia is produced, and no deleterious effects are observed. Our studies demonstrate that tryptophan synthase can catalyze the ammonia-generating β-elimination reaction within any live bacterium. Our results also likely explain previous observations demonstrating that the same indole derivatives inhibit the growth of other pathogenic bacterial species, and why high serum levels of these indole derivatives are favorable for the prognosis of diseased conditions associated with bacterial dysbiosis.

The bacterium Chlamydia trachomatis is an obligate intracellular pathogen that causes urogenital and ocular infections of humans. Distinct serovars of C. trachomatis, classified based on their outer membrane proteins, are associated with infections at these sites. Ocular infections are associated with serovars A–C, while urogenital infections are associated with serovars D–K. Genetic and sequence studies indicate that the genomes of oculotropic and genitotropic strains are >99% identical with two striking differences (1). Urogenital strains encode an intact cytotoxin with a GTPase-inactivating domain (CT166) (2) and, in addition, encode a functional tryptophan synthase that provides them the capacity to synthesize tryptophan via indole salvage (3, 4). The observation that urogenital C. trachomatis isolates express a functional tryptophan synthase extends to genital serovar B isolates, indicating a strong selective pressure to maintain the capacity to salvage indole in the urogenital environment (3, 5). C. trachomatis is a tryptophan auxotroph and cannot synthesize tryptophan de novo. The capacity of urogenital isolates to synthesize tryptophan via indole salvage is proposed to reveal an intricate interplay between C. trachomatis, the IFN-γ–induced host enzyme indoleamine 2,3-dioxygenase (IDO1), and the microbiome at the site of infection (3, 4, 6, 7). When induced, IDO1 degrades tryptophan to kynurenine, thereby starving Chlamydia (8–10). The antibacterial effect of IFN-γ is hampered if the microbiome at the site of infection produces indole, permitting chlamydial tryptophan synthase to generate tryptophan and thus circumvent host-imposed tryptophan starvation (3, 4, 6, 7). The composition of the cervicovaginal microbiome indicates the presence of indole-producing genera during dysbiotic conditions, providing a strong reason for genital isolates to maintain an active tryptophan synthase (6, 7). In contrast, factors that drive the loss of tryptophan synthase in the majority of oculotropic strains are unclear and the focus of this report (5).

Tryptophan synthase has two α-subunits, encoded by trpA, and two β-subunits, encoded by trpB (11, 12). Comprehensive studies by Caldwell et al. (3), evaluating the tryptophan synthase status of a large number of clinical genital and ocular isolates, revealed that no clinical ocular isolates encoded an active tryptophan synthase. With rare exception, this failure arose from point mutations in either trpA or trpB (3). In most bacteria, tryptophan synthase catalyzes two reactions termed α and β. In the α reaction, which is catalyzed solely by TrpA (11, 12), indole-3-glycerol phosphate undergoes a retro-aldol cleavage to release indole and glyceraldehyde-3-phosphate (G3P) (13). Indole is channeled to TrpB by TrpA (11, 14), to undergo a β-replacement reaction with serine to generate tryptophan (11, 13). In addition to β-replacement, in vitro studies using tryptophan synthase from Escherichia coli demonstrated that, in the absence of indole, TrpB deaminates serine via β-elimination to generate ammonia and pyruvate (SI Appendix, Fig. S1) (15–17). The ammonia product of this alternative reaction can be directly antibacterial if it saponifies bacterial lipids (18–20). Alternatively, as first shown for Helicobacter pylori, ammonia kills infected human cells that express the NMDA receptor (21). Given the negative implications of the ammonia-generating β-elimination reaction, it is unsurprising that it is biochemically curtailed in two ways. First, β-elimination proceeds at a slower rate than the tryptophan-generating β-replacement reaction (22, 23). Thus, it is not favored if indole is present. Second, β-elimination is suppressed by allosteric interactions within tryptophan synthase. Biochemical studies using Salmonella typhimurium tryptophan synthase indicate that the G3P product of the α reaction remains bound to TrpA forcing an interaction between loop 6 of TrpA with helix 6 of TrpB (23–27). This allosteric change further decreases the rate of β-elimination. Confirming this, mutations placed in S. typhimurium tryptophan synthase that disrupt the TrpA/TrpB allosteric interaction permit β-elimination to proceed (23–27). Relevant to C. trachomatis tryptophan synthase, studies indicate that the TrpA/TrpB allosteric interaction is not favored for two reasons: (i) TrpA cannot catalyze the α reaction (4); ergo, it is never bound to G3P; and (ii) loop 6 of chlamydial TrpA has mutations (4), which when placed in S. typhimurium tryptophan synthase eliminate allosteric control (25).

These observations led us to posit that ammonia generation by β-elimination may be a negative selection against C. trachomatis tryptophan synthase in an indole-free environment; in this regard, studies reveal the conjunctival microbiome to be dominated by non–indole-producing genera, such as Streptococcus, Corynebacterium, Pseudomonas, and Serratia (28–32). Akin to many other Gram-negative bacteria, expression of the chlamydial trpBA genes are tightly regulated by the tryptophan repressor (TrpR) (33–35). When associated with its corepressor, tryptophan, TrpR prevents transcription by tightly binding the trp operator. Transcription of the trp operon is induced when C. trachomatis-infected cells are subjected to tryptophan depletion, for instance, by exposure to IFN-γ (3, 36). We did not detect increased levels of ammonia in media supernatants under these conditions, possibly because they disfavor translation of TrpB, which has multiple tryptophan residues. Therefore, we sought alternative conditions that permit tryptophan synthase expression without depleting tryptophan.

Molecular studies conducted with Escherichia coli TrpR indicate that some indole derivatives de-repress the trp operon by displacing tryptophan from TrpR, while simultaneously preventing it from binding the trp operator (37–39). Recent studies have demonstrated that some of these indole derivatives are naturally produced by the human gut microbiome and reach concentrations ranging from 1 to 7 μM in blood (40–43). Curiously, these same derivatives act, via an unknown mechanism, as antibacterials against Legionella pneumophila (44–46) and Mycobacterium tuberculosis (47, 48), and possess antiinflammatory properties (40, 49–51). High levels of these indole derivatives in peripheral circulation correlate with protection against inflammatory bowel disease, Crohn’s disease, and type 2 diabetes (49, 52–55); one of them was in clinical trials as an antiinflammatory that reduces Alzheimer’s disease progression (56, 57).

Our results indicate that these indole derivatives effectively kill urogenital C. trachomatis in vitro. They rapidly induce trpBA expression, concomitant with a rise in ammonia levels when indole is not provided. Furthermore, while these derivatives are deleterious to a wild-type genital isolate of C. trachomatis, they have no effect on an ocular isolate or an engineered genital isolate that both lack tryptophan synthase (58). We interpret our results to indicate that trp operon de-repression in the absence of indole underlies the genetic selection against tryptophan synthase in oculotropic C. trachomatis strains. While it remains unclear whether the indole derivatives described by us drive trpBA de-repression in vivo, our results clearly indicate that expression of trpBA in the absence of indole is severely deleterious to C. trachomatis.

Our studies reveal that the ammonia-generating β-elimination reaction can occur when tryptophan synthase is expressed within a live bacterium. Finally, the widely reported antiinflammatory properties associated with these indole derivatives may result from the bacterial burden-reducing consequence of inducing tryptophan synthase activity in an indole-poor environment.

Results

Indole-3-Propionic Acid and Indole-β-Acrylic Acid Are Predicted to De-repress the C. trachomatis trpRBA Operon.

The C. trachomatis trpRBA operon is expressed in conditions where tryptophan is limiting (33, 34), presumably because the TrpR apo-repressor is no longer associated with its corepressor, tryptophan (33, 34). Although C. trachomatis TrpR has diverged from its E. coli homolog (SI Appendix, Fig. S2A), we sought to test whether it could be de-repressed by the same indole derivatives as E. coli TrpR. For E. coli TrpR, indole-3-propionic acid (IPA) and indole-β-acrylic acid (IAA) bind the apo-repressor more tightly than tryptophan but block TrpR from binding trp operator (37, 39). Lawson and Sigler (38) demonstrated that, although IPA and tryptophan bind the same pocket, IPA’s carboxylic acid group repels trpR from the phosphodiester backbone of the trp operator. Marmorstein et al. (37, 39) revealed that IAA de-represses the trp operon similarly. To evaluate whether this mechanism of corepressor and de-repressor association were applicable to chlamydial TrpR, we modeled the structure of chlamydial TrpR with SWISS-MODEL (59), based on the structure of E. coli TrpR (60). Our model (SI Appendix, Fig. S2B) indicates the predicted structure of chlamydial TrpR closely resembles E. coli TrpR. The corepressor binding pocket (SI Appendix, Fig. S3) is highly conserved. For C. trachomatis TrpR, the indole ring of tryptophan is sandwiched between the aliphatic chains of arginines 44 and 74 from one TrpR monomer, and the carboxylic acid group projects toward leucine 31, serine 34, and leucine 33 from the other TrpR monomer. Given this structural conservation, we tested the effect of IAA and IPA on C. trachomatis viability and trpRBA expression.

IPA and IAA Are Deleterious to Genital C. trachomatis and De-repress trpRBA Expression.

HeLa cells, cultured in complete media, were infected with the wild-type C. trachomatis genital strain D/UW-3/CX at a multiplicity of infection (m.o.i.) of 1 and immediately exposed to increasing concentrations of IAA or IPA. Extracts of infected cells were evaluated for inclusion-forming unit (IFU) recovery as described in Materials and Methods (Fig. 1A). IAA and IPA reduced IFU recovery in a concentration-dependent manner. For both compounds, IFU recovery was reduced by ∼2.5 logs at 600 μM. A reduction by ≥1 log was observed at 100 μM IAA and 200 μM IPA. As observed previously (44–46), concentrations of IAA and IPA >3 mM were cytotoxic. We evaluated trpRBA de-repression by quantifying trpB expression relative to 16S rRNA at 24 h postinfection (h.p.i.) 100 μM IAA or 200 μM of IPA increased trpB expression by >1 log relative to control media (Fig. 1B). Therefore, as seen for E. coli (61, 62), IAA and IPA de-repress the C. trachomatis trpRBA operon even during tryptophan sufficiency. The decrease in IFU could result from either reduced replication within infected cells or fewer infected cells. To distinguish these possibilities, primary inclusions were stained at 36 h.p.i. (Fig. 1C). IAA and IPA reduced inclusion size (SI Appendix, Table S1) without changing the fraction of infected cells.

Fig. 1.

IAA and IPA de-repress the C. trachomatis trpRBA operon with deleterious effects. (A) IAA and IPA reduce C. trachomatis D/UW-3/CX IFUs recovered in a dose-dependent manner. HeLa cells were infected with D/UW-3/CX as described (Materials and Methods), following which indicated amounts of IAA or IPA were added. IFU recovery was evaluated at 42 h.p.i. Structures of IAA and IPA are shown as Insets within graphs. (B) IAA and IPA de-repress the C. trachomatis D/UW-3/CX trpRBA operon. RNA extracted at 24 h.p.i. from cells infected with D/UW-3/CX under control conditions or after IAA (100 μM) or IPA (200 μM) exposure. TrpRBA expression was evaluated by RT-qPCR for trpB. (C) IAA and IPA reduce the size of primary inclusions formed by D/UW-3/CX in HeLa cells. Inclusions were stained in cells fixed at 36 h.p.i. as described (Materials and Methods). Chlamydial inclusions are green, while DNA counterstained with Hoechst 33342 is blue. (Scale bar: 20 μm.) *P < 0.05, using the Wilcoxon rank sum test, and **P < 0.01, using the same test.

Indole-3-Lactic Acid Has No Effect on Genital C. trachomatis and Does Not De-repress the trpRBA Operon.

In women, the primary site for genital C. trachomatis infections is the endocervix (63), a microenvironment that often contains Lactobacillus spp. (64), some of which can produce indole-3-lactic acid (ILA) (40). ILA resembles IPA in structure, with the sole difference being that a hydrogen in IPA is substituted by a hydroxyl group in ILA. This difference blocks ILA from de-repressing the trp operon (37, 39) but does not prevent ILA from activating host receptors like the aryl hydrocarbon receptor (AhR) (40, 50, 51, 53), which is also activated by IPA and IAA (40, 50, 51, 53). We tested ILA’s effect on C. trachomatis because it can be in the natural infection microenvironment, and distinguish effects manifested via host cell changes vs. directly on C. trachomatis. ILA did not affect IFU recovery (Fig. 2A) and did not induce the expression of trpBA (Fig. 2B) or alter inclusion formation (Fig. 2C and SI Appendix, Table S1). These data corroborate the interpretation that IPA and IAA negatively impact C. trachomatis through de-repression of tryptophan synthase, rather than by affecting host cell physiology.

Fig. 2.

ILA is not deleterious to C. trachomatis D/UW-3/CX and does not de-repress the trpRBA operon. (A) D/UW-3/CX–infected HeLa cells were exposed to ILA as indicated, and IFU recovery was evaluated at 42 h.p.i. The structure of ILA is shown as an Inset. (B) ILA does not de-repress the D/UW-3/CX trpRBA operon. trpRBA expression was evaluated as described in Fig. 1. (C) Exposure to ILA does not affect the size of primary inclusions formed by D/UW-3/CX. Infected cells exposed to the indicated ILA concentration were fixed and stained at 36 h.p.i. as described in Fig. 1. (Scale bar: 20 μm.)

IPA and IAA Induce Ammonia Production by Genital C. trachomatis-Infected Cells.

Because β-elimination on serine catalyzed by tryptophan synthase creates ammonia (SI Appendix, Fig. S1) (15–17), we tested whether C. trachomatis-infected cells exposed to IPA and IAA produced ammonia. Ammonia is membrane permeant (65, 66) and equilibrates in an aqueous environment to ammonium hydroxide. Therefore, cell supernatants were tested for their ammonia and ammonium levels (Fig. 3). Under control conditions, supernatant levels of ammonia/ammonium were indistinguishable between infected and uninfected cells. Both IPA and IAA dramatically raised supernatant ammonia/ammonium levels only for infected cells, such that they ranged between 40 and 60 µM at 24 h.p.i. In contrast, ILA, which does not de-repress the trpRBA operon, did not induce ammonia production. These results directly prove that serine β-elimination by tryptophan synthase occurs within live bacteria when indole is absent. Furthermore, those indole derivatives that induce ammonia production are deleterious to C. trachomatis. We directly tested whether ammonia was deleterious to C. trachomatis by adding it (as NH₄OH) to cell culture media immediately after infection. As shown (SI Appendix, Fig. S4A), ammonia interfered with IFU recovery in a concentration-dependent manner. Ammonia also affected chlamydial inclusion development (SI Appendix, Fig. S4B).

Fig. 3.

Ammonia is produced by C. trachomatis D/UW-3/CX–infected cells exposed to IPA and IAA, but not ILA. Cells were infected with D/UW-3/CX as described (Materials and Methods). Indicated concentrations of IPA, IAA, and ILA were added immediately postinfection. Ammonia levels in supernatants were evaluated at 24 h.p.i. as described (Materials and Methods). The white bars indicate ammonia levels detected in supernatants from uninfected cells under the specified condition. The gray bars indicate ammonia levels detected in supernatants from infected cells under the specified condition. **P < 0.01, using the Wilcoxon rank sum test.

Indole Blocks the Deleterious Effect of IPA and IAA on C. trachomatis.

In vitro studies indicate that the indole-consuming β-replacement reaction is favored over the ammonia-generating β-elimination reaction on serine (22, 23). Indeed, a central tenet to our hypothesis is that β-elimination only occurs when indole is absent. We tested whether indole blocks the deleterious effect of IPA and IAA (Fig. 4). Provision of 50 μM indole concurrently with IPA and IAA completely restored inclusion size (Fig. 4A) and IFU recovery (Fig. 4B). Indole also reduced supernatant ammonia levels to control levels (Fig. 4C). Importantly, indole did not prevent IPA and IAA from de-repressing trpRBA (Fig. 4D), indicating that indole blocks the deleterious effect of IPA and IAA by shifting catalysis from β-elimination to β-replacement.

Fig. 4.

Indole blocks the deleterious effect of IAA and IPA on C. trachomatis D/UW-3/CX. (A) Fifty micromolar indole along with the indicated concentration of IAA or IPA was added to D/UW-3/CX–infected HeLa cells. Inclusions were stained as described in Fig. 1. (B) IFU recovery was evaluated at 42 h.p.i. from cells infected with D/UW-3/CX and exposed to IAA and IPA alone (gray bars), or in combination with 50 μM indole (white bars). (C) Indole blocks the production of ammonia by HeLa cells infected with D/UW-3/CX and treated with IPA or IAA. Infected cells were treated with IAA (100 μM) or IPA (200 μM) alone, or together with 50 μM indole, following which supernatant ammonia levels were measured at 24 h.p.i. (D) Indole does not prevent de-repression of the D/UW-3/CX trpRBA operon by IAA and IPA. Infected HeLa cells were exposed to 50 μM indole alone (white bar), or in combination with 100 μM IAA or 200 μM IPA (gray bars). TrpRBA expression was evaluated as described. **P < 0.01, using the Wilcoxon rank sum test.

IPA and IAA Do Not Affect a C. trachomatis Serovar A Strain that Lacks Intact TrpA, or an Engineered C. trachomatis Serovar D Point Mutant that Lacks TrpB.

Tryptophan synthase is inactivated in oculotropic C. trachomatis strains predominantly by mutations in trpA or trpB, with trpA mutations being more frequent (3). Studies by the groups of McClarty and Caldwell, demonstrate that while C. trachomatis TrpA lacks catalytic activity, it is essential for TrpB to catalyze β-replacement (4). Consistent with this, although tryptophan depletion induces TrpB production in oculotropic C. trachomatis trpA mutants (3, 4), they are not indole-rescuable (3, 4). We evaluated the effect of IPA and IAA on an oculotropic C. trachomatis trpA mutant (A2497), or an engineered point mutant that inactivates trpB constructed in the same genetic background (D/UW-3/CX) as our wild-type strain (58). IAA and IPA did not affect A2497 or the D trpB mutant at concentrations that reduce IFU recovery for D/UW-3/CX by ≥1 log (Fig. 5A). Consistent with this, IPA and IAA also did not affect primary inclusion formation or size for A2497 or D trpB mut (Fig. 5B and SI Appendix, Table S1), but de-repressed the trpRBA operon in both strains as revealed by RT-qPCR for trpB (Fig. 5C). Finally, consistent with neither strain producing a functional tryptophan synthase, IPA and IAA did not induce them to make ammonia (Fig. 5D). Therefore, for C. trachomatis, ammonia production by β-elimination requires both TrpA and TrpB subunits of tryptophan synthase. The data obtained with A2497 and D trpB mut clearly indicate that tryptophan synthase activity in the absence of indole is highly deleterious to C. trachomatis. Furthermore, trpRBA de-repression in an indole-free environment is likely relevant to the selection and survival of oculotropic C. trachomatis strains.

Fig. 5.

IAA and IPA do not affect C. trachomatis strains that lack tryptophan synthase activity. (A) HeLa cells were infected with a trpA mutant C. trachomatis serovar A strain (A2497), or a derivative of C. trachomatis D/UW-3/CX with an inactivating point mutation in trpB (D trpB mut). Infected cells were exposed to 100 μM IAA or 200 μM IPA for 42 h, following which IFU recovery was evaluated. (B) IAA and IPA do not affect primary inclusion formation by A2497 or D trpB mut. HeLa cells infected with the indicated strain were exposed to control media or IPA and IAA for 36 h following which chlamydial inclusions were stained. (Scale bar: 20 μm.) (C) IAA and IPA de-repress the trpRBA operon in A2497 and D trpB mut. HeLa cells, infected with A2497 or D trpB mut, were treated with 100 μM IAA or 200 μM IPA for 24 h, following which trpRBA expression was evaluated. (D) IPA and IAA do not induce ammonia production from HeLa cells infected with A2497 or D trpB mut. HeLa cells infected with A2497 or D trpB mut were treated with 100 μM IAA or 200 μM IPA for 24 h, following which ammonia levels in cell supernatants were evaluated. Control conditions are indicated by the white bar, while ammonia levels after IPA and IAA exposure are indicated by the gray bars. **P < 0.01, using the Wilcoxon rank sum test.

Physiological Oxygen Tension Permits Physiological Concentrations of IPA and IAA to Be Effective Against Genital C. trachomatis.

Our findings indicate that de-repression of trpBA in the absence of indole is deleterious to genital C. trachomatis, and that this process may be driven by indole derivatives produced by the gut microbiome. However, the concentrations of these derivatives found in serum are between 1 and 2 logs lower than the concentrations we and others have used for experiments in vitro (41, 42, 44–47, 49, 50). Our attempts to dissect this discrepancy were directed by our observation that IAA and IPA had half-lives of ∼6 h in cell culture, independent of the vehicle used to solubilize them. Studies evaluating the neuroprotective properties of IPA in cell culture reveal that while protecting against oxidative radicals, it is oxidized to kynuric acid (57), the structure of which renders it incompatible with functioning as a de-repressor for TrpR. While conducting studies on viral transactivators, we observed that epithelial cells cultured under normoxia (20% O₂) produce oxidative radicals including hydroxyl radicals and superoxides. This production was decreased by culturing cells under hypoxic conditions (4% O₂) (67–69). Our findings correspond to observations made by others with HeLa cells, in that oxidative radicals have been observed at significant levels during normoxic culture (70); in contrast, hypoxia relieves oxidative stress and increases the viability of HeLa cells (71). In this regard, physiological oxygen tensions in the majority of human tissues range from 0 to 5% O₂; for example, vaginal oxygen tension is typically 1–2% O₂, rising briefly to 5% O₂ during sexual stimulation (72); a similar range has been observed for cervical and uterine oxygen tension (73). Conjunctival oxygen tension is also low, ranging 0.5–1% O₂ (74, 75). Therefore, we tested the function of IPA and IAA on genital C. trachomatis under hypoxic conditions (4% O₂). These results are shown in Fig. 6. Under these conditions, 5 μM IPA and IAA reduced IFU recovery by ∼1 log (Fig. 6A) and effectively de-repressed trpBA (Fig. 6B). We also observed the generation of ammonia, albeit at a lower level (Fig. 6C). This decrease either reflects reduced activity of tryptophan synthase under these conditions or that ammonia is blown off during gas exchanges conducted every 4 h to maintain 4% O₂. Finally, the provision of indole rescued the effect of IPA and IAA under hypoxia (Fig. 6D), confirming observations made under normoxic conditions.

Fig. 6.

Physiological concentrations of IPA and IAA are deleterious to genital C. trachomatis under hypoxic (4% O₂) conditions. (A) D/UW-3/CX–infected HeLa cells were treated with 5 and 15 µM of IAA or IPA and incubated in at 4% O₂ as described in Materials and Methods. At 42 h.p.i., cells were harvested to measure IFUs released. (B) Physiological concentrations of IAA and IPA de-repress the D/UW-3/CX trpRBA operon. trpRBA expression was evaluated as described in Materials and Methods. (C) Ammonia is generated by D/UW-3/CX–infected cells incubated with physiological concentrations of IAA and IPA under hypoxic conditions. Ammonia levels were measured from supernatants obtained at 24 h.p.i. as described earlier. (D) Indole alleviates the deleterious effects of IAA and IPA on genital C. trachomatis growth. HeLa cells infected with genital C. trachomatis were exposed to 5 and 15 µM of IAA and IPA concurrently with 50 µM indole and incubated under hypoxic conditions. IFU release was measured at 42 h.p.i. *P < 0.05, using the Wilcoxon rank sum test, and **P < 0.01, using the same test.

Discussion

The selective pressures that cause the loss of tryptophan synthase activity in oculotropic C. trachomatis isolates are enigmatic. We have shown that trpRBA operon de-repression in an indole-poor environment is a powerful selection against expression of tryptophan synthase. Under these conditions, tryptophan synthase generates ammonia via serine deamination, as by demonstrated Crawford and Miles and colleagues (12, 15, 16) in vitro. Our results prove that serine deamination occurs when tryptophan synthase is expressed in indole-free conditions. Using conditions that shift tryptophan synthase’s activity from β-elimination to β-replacement, our results ascribe deleterious effects to the former. Finally, our results indicate that both the TrpA and TrpB subunits of C. trachomatis tryptophan synthase are needed for serine deamination, akin to requirements for β-replacement (4). Mutation in either is protective, explaining why oculotropic strains have mutations in either trpA or trpB (3).

We used IPA and IAA to de-repress the trp operon because these indole derivatives are known gut microbiome products that are found in peripheral circulation in low micromolar concentrations (40–43). Furthermore, the mechanism by which they act on E. coli TrpR for trp de-repression (38) is known, permitting us to evaluate the possibility of their binding chlamydial TrpR before testing them experimentally. Our findings demonstrate these two biologically relevant indole derivatives can de-repress the chlamydial trpRBA operon; our structural predictions suggest that de-repression occurs by their engaging TrpR (SI Appendix, Figs. S2 and S3), although an alternative mechanism of de-repression is not excluded by our studies. The majority of studies evaluating the effect of IPA and IAA on bacteria and host cells have employed them at concentrations that are 1–2 logs higher than their levels in serum (40–43). Pertinently, we have found these compounds to be effective at their serum concentrations provided physiological oxygen tension is maintained during in vitro culture.

IPA and IAA can alter host-cell physiology by binding receptors like the AhR. However, their deleterious effect on C. trachomatis does not result from such interactions because (i) their effect is unabated when C. trachomatis-infected cells are exposed to IPA or IAA in the presence of cycloheximide, which inhibits host-cell protein translation; and (ii) ILA, an indole derivative structurally close to IPA and IAA, binds cellular receptors bound by them, but has no effect on C. trachomatis. Confirming the mechanism ascribed to IPA and IAA, ILA does not de-repress the C. trachomatis trpRBA operon. Our findings with IPA and IAA may explain the previously observed effects of these molecules on Legionella pneumophila (44–46). IPA and IAA reduce L. pneumophila growth under axenic conditions as well as intracellularly within monocytes. Analogous to C. trachomatis, IPA and IAA may be bactericidal by inducing tryptophan synthase, which deaminates serine. A recent screen of gut microbiome metabolites identified IPA as a prominent antitubercular compound that was effective against Mycobacterium tuberculosis (47). While no mechanism was described, it was proposed that IPA be considered for inclusion in antitubercular therapy (47). The approaches we used can be used to determine whether trp de-repression underlies IPA’s bactericidal effects on L. pneumophila and M. tuberculosis. Should that prove true, the simple point mutations in trpA and trpB that permit C. trachomatis ocular strains to evade this deleterious effect dictates that caution be used while considering IPA as an antibacterial.

IPA and IAA are proposed as beneficial indole derivatives produced by desirable genera within the gut microbiome (40, 53). Studies have correlated IPA and IAA levels with better prognosis for individuals with type 2 diabetes, Crohn’s disease, and inflammatory bowel disease (49, 52–55). IPA was also in clinical trials for its effect against Alzheimer’s disease (56, 57). In these conditions, IPA is believed to function as an antiinflammatory. We suggest in conditions associated with bacterial dysbiosis, such as Crohn’s disease and inflammatory bowel disease, IPA may be directly bactericidal by inducing tryptophan synthase and serine deamination. Its observed antiinflammatory effects in conditions like type 2 diabetes may arise from reducing the overall bacterial burden via this mechanism.

IAA is produced by Clostridium sporogenes, and Peptostreptococcus spp. including P. russellii, P. anaerobius, and P. stomatis (40, 42, 49). IPA is produced by the same Peptostreptococcus spp., and a broader spectrum of Clostridium spp. including C. sporogenes, C. botulinum, C. caloritolerans, C. cadaveris, and C. paraputrificum (40, 42, 49, 76–78). We were curious how IPA- and IAA-producing species evaded their bactericidal effects. Pertinently, no bacterial species that makes IAA or IPA has a trp operon or trpR. Thus, IPA and IAA producers are immune to the mechanism by which these compounds are bactericidal. For this reason, we believe that IPA and IAA are made as antibiotics that permit colonization by specific species over other species competing for the same niche.

The invariant capacity of genital C. trachomatis to maintain an active tryptophan synthase is thought to exemplify a pathogen finessing metabolites produced by a niche-specific microbiome to evade an IFN-γ–driven protective host immune response (3, 6, 7). Indole produced by dysbiotic genital microbiomes can be salvaged by C. trachomatis to evade tryptophan starvation imposed by IFN-γ (3, 6, 7). While indole-producing bacterial species can be isolated from vaginal samples (6), IFN-γ levels found at infection sites are likely insufficient by themselves to drive tryptophan starvation (6, 63, 79). Although a role for other cytokines in driving tryptophan starvation is likely, it should be noted that significant levels of cervicovaginal tryptophan continue to be detected in the majority of women who have spontaneously cleared their chlamydial infection (79). Therefore, we suggest that clearance may result from a combination of multiple factors, including (i) a protective immune response that decreases tryptophan availability; and (ii) de-repression of trpRBA by indole derivatives. Lowered tryptophan and de-repression both promote expression of tryptophan synthase. The outcome of tryptophan synthase expression remains dependent on the prevalent genital microbiome. Genital microbiomes that lack indole producers, such as the “normal” vaginal microbiome, favor C. trachomatis clearance by (i) preventing tryptophan synthesis via indole salvage; and (ii) permitting serine deamination to prevail over indole-requiring β-replacement. In contrast, dysbiotic microbiomes with indole producers will hinder clearance by (i) permitting tryptophan synthesis via indole salvage; and (ii) creating conditions that favor indole-requiring β-replacement over serine deamination. It should also be noted that because tryptophan and de-repressors compete for binding TrpR (37–39), dysbiotic genital microbiomes will also decrease the efficacy of de-repressors by increasing intrainclusion tryptophan concentration.

In summary, our studies reveal that expression of tryptophan synthase can be deleterious to C. trachomatis if it occurs in an indole-poor environment. The paucity of indole producers in the ocular microbiome likely provides a strong selective pressure against maintaining this enzyme in that environment. In contrast, for genital infections, indole-producing dysbiotic microbiomes prevent immune clearance by two closely related mechanisms that reflect the activities of tryptophan synthase. The presence of indole blocks serine deamination and simultaneously permits tryptophan biosynthesis to evade starvation. We have demonstrated that gut microbiome-produced indole derivatives can promote tryptophan synthase expression, with deleterious effects in the absence of indole. Our studies suggest a need to evaluate the genital microenvironment for the presence and levels of molecules that can function as trpRBA de-repressors to determine their contribution toward clearance, and the mechanism by which they synergize with IFN-γ.

Materials and Methods

Structural Prediction of TrpR from C. trachomatis D/UW-3/CX.

TrpR from C. trachomatis D/UW-3/CX (accession no. NP_219672) was aligned with E. coli TrpR (accession no. AKD90255) using T-COFFEE (80) with default parameters. SWISS-MODEL (59) was used to create a structural model for C. trachomatis TrpR based on the structure of E. coli TrpR (Protein Data Bank ID code 3WRP) (60). Structures were visualized using EzMol (81). Predicted structures for TrpR and tryptophan synthase from C. trachomatis D/UW-3/CX can be obtained from the Zenodo repository (https://doi.org/10.5281/zenodo.2662318).

Cell Culture, Media Additives, and C. trachomatis Infections.

HeLa cells were cultured as described (82, 83). Postinfection, they were fed DMEM synthesized as described earlier (82, 83), supplemented with 5% triple-dialyzed calf serum. Indicated concentrations of indole derivatives, prepared freshly in DMSO, were added to postinfection media. IPA (Sigma; catalog #57400), IAA (Sigma; catalog # I2273), ILA (Sigma; catalog #I5508), and indole (Sigma; catalog #I3408) were stored under vacuum. The C. trachomatis strains serovar D (D/UW-3/CX), D trpB-null (single point mutation of TrpB subunit of tryptophan synthase), and serovar A (A2497) were used as indicated. Cells were infected as described previously at a m.o.i. of 1 (82, 83). IFU recovery was measured at 42 h.p.i. as described previously (82, 83).

Immunofluorescence, Imaging, and Inclusion Area Measurement.

Control and infected cells were stained using a FITC-conjugated anti-chlamydial LPS antibody (Merifluor; catalog #500111; Meridian Bioscience) as described previously (82, 83). Stained cells were imaged as described previously (82, 83); images were analyzed using Fiji, version 2 (84). At least 75 inclusions were sized for each experimental condition.

Reverse-Transcriptase–qPCR for trpB Gene Expression.

mRNA extracted from infected cells at 24 h.p.i. using the RNeasy kit (Qiagen; catalog #74104) was used to prepare cDNA using the iScript gDNA Clear cDNA Synthesis kit (Bio-Rad; catalog #1725035). qPCRs were performed using the SsoAdvanced Universal SYBR Green Supermix kit (Bio-Rad; catalog #1725271). Reactions were run in a Bio-Rad CFX 384. The following primers were used for amplification: CT-16SrRNA_F, TGAGGAGTACACTCGCAAG; CT-16SrRNA_R, GCGGAAAACGACATTTCTGC; CT-trpB_F, GTGGAACGACAGAAACC; and CT-trpB_R, GGCCGATCCTAAGCAATAG. TrpB mRNA levels were quantified relative to 16S rRNA levels as described (75).

Ammonia/Ammonium Quantification.

Cell supernatants were recovered at 24 h.p.i., 0.2 μm filtered, following which ammonia/ammonium was measured using an Ammonia Assay Kit (Sigma; catalog #AA0100-1KT).

Infections Under Physiological O₂ Conditions.

A premixed gas mixture (91% N₂, 5% CO₂, 4% O₂) was used in these experiments as described previously. Media applied to cells were preexposed to a 4% O₂ atmosphere in sealed modular chambers for at least 24 h for equilibration. HeLa cells were spin infected as described above. Infected cells were fed preequilibrated control media, or media containing IPA or IAA. Cells were placed in sealed chambers, flushed for at least 5 min with the gas mixture, and placed in a 37 °C incubator. Chambers were manually flushed with the gas mixture every 4 h.

Exposure to Ammonia.

HeLa cells were infected as described above and exposed to the indicated concentration of ammonia, which was added to media as ammonium hydroxide. Because ammonia is volatile, control and ammonia-exposed cells were placed in chambers in separate incubators. Media ammonia concentrations were confirmed using the ammonia assay described above.

Statistical Analysis.

Experiments were repeated a minimum of three independent times. Statistical significance was calculated with the Wilcoxon rank sum test (85), using MSTAT, version 6 (N. Drinkwater, McArdle Laboratory for Cancer Research, University of Wisconsin Medical School, Madison, WI).