INTRODUCTION
The human gastrointestinal tract represents a complex ecosystem comprised of resident microbiota, dietary elements, and host factors. Within this environment, microbe-microbe and host-microbe interactions are important for maintaining homeostasis (
1,
2). Alterations in the gut ecosystem by factors such as diet, antibiotics, and host susceptibility can lead to dysbiotic conditions wherein a resident microbe adopts pathogenic behaviors and causes disease (
3,
4).
Indole is a microbiota-derived signaling molecule present in the human gut that is known to modulate virulence factors in several enteric bacteria (
5–9) while concurrently functioning to strengthen the host intestinal barrier (
10). Indole and indole derivatives also activate intestinal xenobiotic receptors, such as pregnane X receptor (PXR) (
11), to facilitate anti-inflammatory (
12) and detoxification (
13) responses. The levels of indole in the intestinal lumen are unknown; however, concentrations in human feces can reach 6 mM (
14), with most studies describing ranges between 0.25 and 1.1 mM (
15,
16). Indole is generated as a by-product of the reversible conversion of
l-tryptophan to pyruvate, a key glycolytic end product and Krebs cycle intermediate, by the enzyme tryptophanase (
17). The gene encoding tryptophanase,
tnaA, is ubiquitous among gut bacteria (
18). In
Escherichia coli, the transcription of
tnaA is regulated by carbon catabolite repression; the availability of a preferred carbon source, such as glucose, eliminates the need to convert tryptophan into pyruvate and indole, which is subsequently released (
19). While it is unclear if other gut bacteria regulate indole production in the same way, the abundance and metabolic status of tryptophanase producers could profoundly influence luminal indole levels.
Klebsiella oxytoca is regarded as a human gut commensal, yet in older children and adults treated with β-lactam antibiotics, overgrowth of cytotoxin-producing strains of
K. oxytoca results in antibiotic-associated hemorrhagic colitis (AAHC), a distinct form of non-
Clostridium difficile colitis (
20,
21). Recently, we found that blooms of cytotoxin-producing
K. oxytoca also were associated with the development of necrotizing enterocolitis (NEC) (
22), a devastating intestinal disease of premature infants (
23). Similar strains were found at lower abundance in the majority of non-NEC control infants (
22), which underscores the importance of deciphering how contextual changes in the gut ecosystem impact pathogenicity. In heathy adults, intestinal colonization with
K. oxytoca is reported in 2% to 9% of subjects and ∼1/2 of the strains produce cytotoxin (
24). Recent studies indicate more prevalent carriage of this microbe in the neonatal population (
25) and that many isolates from infants have the capacity to produce toxin (
22,
26).
K. oxytoca (
sensu stricto) is a member of the
K. oxytoca complex, comprising several closely related species within different phylogroups named according to the presence of a
blaOXY variant (β-lactamase) conferring resistance to amino- and carboxypenicillin (
27). Methods used in clinical microbiology laboratories often identify all
K. oxytoca complex isolates as
K. oxytoca (
sensu lato). The identity and contributions of several members to disease remained largely unknown until whole-genome sequencing (WGS) and other advancements led to the identification of newer strains and an expansion of the
K. oxytoca complex (
28–32).
Klebsiella grimontii and other
K. oxytoca complex members subsequently were demonstrated to inhabit the gut of premature infants (
33,
34).
Identification of a biosynthetic gene cluster responsible for generating the pyrrolobenzodiazepine cytotoxins tilimycin and tilivalline has enabled genetic screening for cytotoxin-producing members of the
K. oxytoca complex (
22,
26,
34,
35). Analysis of this gene cluster in the
K. oxytoca AAHC isolate AHC-6 showed that products of
npsA and
npsB are involved in the late-step enzymatic synthesis of tilimycin (also called kleboxymycin), a toxic product of nonribosomal peptide synthesis (NRPS) (
36,
37). In the presence of indole, tilimycin spontaneously incorporates the indole ring to form tilivalline (
36). While both tilimycin and tilivalline are considered toxins, they have distinct molecular targets, and tilimycin is substantially more cytotoxic to mammalian cells than tilivalline (
38). Host factors also are involved in determining sensitivity to exogenous compounds. PXR, a ligand-activated host nuclear receptor, plays an important role in xenobiotic detoxification (
13). PXR also regulates intestinal inflammatory responses through the binding of microbe-specific indoles (
12).
Here, we demonstrate that bacterial indole alleviates cytotoxicity induced by K. grimontii and K. oxytoca via suppression of tilimycin synthesis and enhanced conversion of tilimycin to tilivalline while simultaneously activating a host nuclear receptor, PXR. We establish that tilivalline is a strong agonist of PXR; this interaction upregulates PXR-responsive detoxifying genes and mitigates tubulin acetylation. Our findings strengthen the notion that metabolites produced by gut microbes are critical mediators of microbe-microbe and host-microbe cross talk.
DISCUSSION
Commensal microbes and host factors play a crucial role in regulating the human intestinal ecosystem and maintaining homeostasis. Microbe-microbe and host-microbe interactions are further influenced by environmental changes, such as nutrient availability, which can shape the behaviors and signaling molecules produced by the resident microbiota (
53). The context in which a microbe exhibits different characteristics also can have a dramatic impact on determining if it is friend or foe (
54). Here, we report a novel communication network involving the regulation of
K. grimontii and
K. oxytoca, two closely related species regarded as a human gut commensals but also with the capacity to induce toxin-mediated colitis (
20,
21,
28). Availability of metabolizable carbohydrates, environmental indole concentrations, and host PXR were found to modulate cytotoxic effects in an integrated manner. This study expands the growing body of literature showing that indole is important signaling molecule that reduces pathogenicity (
5–9,
55) while also incorporating unique elements of microbial, metabolic, and host cross talk in the modulation of virulence.
K. oxytoca is commonly isolated from humans and belongs to the early colonizers of the infant gut microbiota (
25). Cytotoxin-producing strains are the causative agent of AAHC (
20,
21) and have been identified in premature infants with and without NEC (
22,
34). Several other studies have implicated
Klebsiella spp. in NEC but failed to discriminate among species (
56,
57) or did not differentiate between toxin-positive and toxin-negative strains (
58). Recent extended genomic analyses of
Klebsiella spp. have subdivided members of the
K. oxytoca complex into various phylogroups, with strains of
K. michiganensis,
K. grimontii, and
K. pasteurii, along with
K. oxytoca, reported to contain the biosynthetic gene cluster (BGC) encoding the enzymatic pathway for tilimycin synthesis (
34). Given this recent reclassification, we performed WGS and ANI/OrthoANI comparisons on UCH-1, isolated from a patient with NEC, and found that it matched closest with
K. grimontii. Similar to
K. oxytoca, isolates of
K. grimontii have been recovered from the fecal stream of subjects with AAHC and asymptomatic carriers (
28).
K. grimontii harboring the tilimycin BGC also has been recovered from preterm (
33,
34) and term (
26) infants.
The link between carbohydrate metabolism and virulence determinants has been demonstrated for several other enteric pathogens, including cytotoxin-producing
Clostridium species (
59). Here, we demonstrate that the presence of a fermentable carbohydrate (i.e., glucose) increases toxin synthesis and cytotoxicity induced by species of the
K. oxytoca complex. These results agree with a prior study which demonstrated variations in cytotoxicity using culture supernatants from
K. oxytoca complex isolate MH43-1 from different culture conditions (
37). In some patients taking β-lactam antibiotics, overgrowth of one or more members of the
K. oxytoca complex can result in AAHC (
20,
21). While expansion of this microbe is generally regarded as the primary driver of AAHC, it remains to be determined if the release of fermentable carbohydrates by commensal bacteria upon antibiotic treatment is contributory, as reported for
C. difficile colitis (
59). A role for
K. oxytoca complex-related infections also has been reported for individuals with diabetes (
60); while it is not clear if the isolates were toxin positive, the presence of this toxin-producing gene cluster could complicate the management of infections caused by multidrug-resistant strains. Presumably, toxin-positive species increase cytotoxin synthesis in response to fermentable carbohydrates to enhance their ability to compete for nutritional resources. As demonstrated previously (
38), and confirmed in the present study, tilimycin exhibits antibacterial activity against inhabitants of the gut microbiota. It is possible, therefore, that intestinal injury invoked by
K. oxytoca complex cytotoxins represent collateral damage.
Malabsorption of carbohydrates has been associated with the development of NEC (
61) and enhances intestinal damage in animal models of NEC (
62); however, these linkages have not been attributed to the pathogenic potential of a specific commensal microbe (
4). Many preterm formulas contain glucose in the form of corn syrup solids. The preterm gut has immature glycosidase activity, which increases the risk of carbohydrate malabsorption (
63). Thus, the nondigested carbohydrates in the preterm gut could serve as a preferred carbon source for tryptophanase producers, thereby decreasing luminal indole levels and inciting cytotoxicity by members of the
K. oxytoca complex. Interestingly, fecal tryptophan levels were reported to be decreased in preterm infants exposed to early antibiotics (
64), and several studies suggest that antibiotic use in preterm infants increases the risk of NEC (
65,
66). Members of the
K. oxytoca complex also utilize a broader array of sugars than other
Klebsiella spp., which may provide the fermentative energy to support toxin production and promote colonization resistance of other enteric bacteria (
67,
68). Perturbations in the gut microbiome leading to excessive Toll-like receptor 4 (TLR4) stimulation have been suggested as an inciting event in NEC pathogenesis (
69,
70), whereas others have challenged the notion that NEC represents a single disease entity (
71). Our findings suggest that cytotoxin production by members of the
K. oxytoca complex represents an additional pathway leading to mucosal disruption in the preterm gut.
Indole and indole metabolites are increasingly recognized as interspecies and interkingdom signaling molecules that reduce pathogenicity (
55). The results obtained from
npsA and -
B transcriptional experiments, MS cytotoxin measurements, and cytotoxicity assays extend these observations to include the regulation of virulence by the
K. oxytoca complex. Furthermore, the data obtained using the murine model confirm that these signaling pathways respond similarly
in vivo. In the gut lumen, microbes sense and respond to indole (
5,
6,
72), while host cells absorb indole to strengthen barrier integrity and facilitate beneficial host-microbe interactions (
10). Among these interactions, the induction of enterocyte PXR facilitates anti-inflammatory and detoxification responses (
12,
13). Indole is generated as a by-product of the conversion of tryptophan to pyruvate by tryptophanase (TnaA), an enzyme that is ubiquitous among gut bacteria (
18). As demonstrated in the present study, the availability of a preferred carbon source, such as glucose, eliminates the need for bacteria to use this tryptophan conversion pathway; consequently, expression of
tnaA is repressed and indole production decreases. Alternatively, in the absence of fermentable carbohydrate,
tnaA is expressed and indole production is high. The general utility of indole as a signaling molecule for bacteria under different environmental conditions is well described (
73). It is possible that
K. oxytoca complex members utilize luminal indole as a barometer for nutrient availability. The production of secondary metabolites such as tilimycin is energetically costly. Repression of
npsA and -
B by indole may minimize energy expenditure and promote mechanisms of persistence, such as biofilm formation (
74,
75), while living in a dense ecosystem with limited nutrients; conversely, when there is greater access to nutrients, such as during antibiotic-induced dysbiosis or carbohydrate malabsorption, derepression of
npsA and -
B by low luminal indole levels may confer a competitive edge.
In the analysis of culture supernatants from
K. grimontii (UCH-1) and
K. oxytoca (AHC-6), grown in LB with added glucose, both tilimycin and tilivalline were detected; however, concentrations of tilimycin far exceeded those of tilivalline, similar to what is reported for fecal contents of humans with AAHC and the murine model of AAHC using AHC-6 (
38). Furthermore, results from the apoptosis experiments performed with UCH-1 grown with and without exogenous indole strongly suggest that tilimycin is the major cytotoxic culprit. Thus, an additional facet of indole regulating
K. oxytoca complex pathogenicity is the conversion of the potent toxin tilimycin to the less toxic product tilivalline.
Indoles and indole analogs/metabolites have been shown to interact with mammalian nuclear receptors that modulate host physiologic and inflammatory responses (
11). To assess the host’s role in regulating responsiveness to tilimycin and tilivalline, we investigated their binding and agonist activity to PXR, a ligand-activated transcription factor and xenobiotic sensor well known for its role in detoxification, drug metabolism, and regulation of intestinal inflammation (
12,
13). In this capacity, PXR functions as a key modulator of inflammatory bowel disease (
76) and experimental NEC (
77). We found that tilivalline, but not tilimycin, which lacks the indole moiety, functioned as an intestine-specific PXR agonist and upregulated PXR-dependent detoxification genes. This interaction also limited the toxic effects of tilivalline on human enterocytes by mitigating tubulin acetylation. Hence, PXR works in a coordinated manner with commensal indole producers following the conversion of tilimycin to tilivalline. Given the wide range of effects that PXR exerts on intestinal homeostasis and host metabolism, these results support the notion of a significant interplay between microbial metabolites and gut physiology (
78). Collectively, these results also indicate that nutrient availability, which profoundly effects the metabolic status of tryptophanase producers, is an important element of this network.
The comparative analysis of
K. grimontii strain UCH-1 and
K. oxytoca strain AHC-6 revealed that expression of
npsA and -
B and tilimycin concentrations were dramatically higher in AHC-6. This strain was originally cultured from an adult subject with AAHC (
20), whereas UCH-1 was from a preterm infant with NEC (
22). Additional studies are needed to determine if differential cytotoxin production is a consistent feature among phylogroups and how the context of cytotoxin synthesis, such as within the preterm gut versus mature gut, impacts pathogenicity. Polymorphisms in upstream regulatory regions could explain the transcriptional differences, which, in turn, impacts cytotoxin biosynthesis, as with
C. difficile “supertoxin” producers (
79). Regardless, both isolates were found to respond to carbohydrate and indole in an analogous fashion, extending these observations across the two phylogroups.
In summary, we have demonstrated a novel indole-based signaling pathway involving intestinal, bacterial, and nutritional components in the regulation of K. oxytoca complex pathogenicity. Deciphering the dysbiotic environmental conditions, genetic diversity of specific strains, and contextual factors are important parameters for understanding how commensal microbes alter their behaviors and cause disease. Such studies are needed to unravel how these factors contribute to K. oxytoca complex-induced intestinal pathology in preterm infants with NEC and older children and adults who develop AAHC.
MATERIALS AND METHODS
Bacterial culture conditions.
For routine culture, all strains were grown in Luria-Bertani broth or agar (LB; BD Biosciences, Franklin Lakes, NJ) and all media were filter sterilized. When required, media were supplemented with glucose (2.5 g/L), indole (Sigma, St. Louis, MO) in 0.1% (vol/vol) N,N-dimethylformamide (DMF), or DMF alone (vehicle). Bacterial cultures were grown overnight in 5 mL of LB broth, harvested by centrifugation at 5,000 rpm, washed once in 5 mL of sterile saline, and suspended in fresh LB broth. This preparation was used to inoculate prewarmed medium to 0.06 unit (absorbance at 600 nm) to prepare the primary experimental culture(s). Growth (optical density at 600 nm [OD600]) was measured every hour for 12 h while cultured in 10:1 (vol/vol) flask-to-medium ratios at 37°C with shaking at 225 rpm.
Bacterial RNA purification and analysis.
Quantities of 5 U (OD600) of culture samples were mixed with twice the volume of bacterial RNAprotect (Qiagen, Valencia, CA) and centrifuged for 10 min at 5,000 rpm. The supernatants from these samples were discarded and cell pellets were mixed with 1 mL of TRIzol (Invitrogen, Carlsbad, CA), transferred to lysing matrix B tubes (MP Biomedicals, Irvine, CA), and subjected to lysis using Disruptor Genie (Scientific Industries, Bohemia, NY) for 4 min at 3,000 rpm. The cell lysate was processed using an RNeasy kit (Qiagen) per the manufacturer’s instructions. The concentration and purity of isolated RNA were determined using a NanoDrop 2000 UV-visible (UV-Vis) spectrophotometer (Thermo Fisher Scientific, Waltham, MA).
cDNA synthesis and RT-qPCR.
A total of 500 ng of RNA was used to prepare cDNA using the iScript master mix (Bio-Rad, Hercules, CA) reverse transcriptase reaction protocol (5 min at 25°C, 20 min at 46°C, and 1 min at 95°C) in a 20-μL volume. Following this reaction, each sample was diluted 10-fold and 5 μL was used for real-time quantitative PCR (RT-qPCR) in a total reaction volume of 25 μL containing 12.5 μL of 2× SsoAdvanced Sybr green supermix (Bio-Rad) and 8.75 pmol of each primer. Primers for the target genes and the internal reference gene (
recA) are listed in
Table S2. The reaction and preliminary data analysis were carried out on a CFX96 real-time PCR detection system and Bio-Rad CFX Manager software version 3.0 (Bio-Rad). Transcripts were plotted as target copies per 100 copies of the reference gene (
80). For human intestinal cell lines and murine intestinal organoids, transcript copy numbers were normalized to the β-actin gene or
Ppib,
Tbp, and
Gusb, respectively.
Detection of tilimycin and tilivalline in bacterial culture supernatants by UPLC-MS/MS.
Samples were analyzed at the University of Connecticut Center for Environmental Sciences and Engineering (Storrs, CT) using a Waters Acquity ultraperformance liquid chromatograph (UPLC) coupled with an Acquity TQD tandem mass spectrometer (Waters Co., Milford, MA) as described previously (
22). A total of 450 μL of test or quality control samples were spiked with 25 μL internal standard solution and 25 μL of methanol. An Acquity UPLC BEH C18 (1.7 μm, 2.1 × 50 mm) column, maintained at 25°C and with a sample injection volume of 10 μL on a 20-μL loop, was utilized for analyte separation. Analyte signal optimization was performed using the Waters IntelliStart, whereas analyte detection, quantification, and statistical analysis were performed using the Waters QuanLynx within MassLynx software v.4.2 in electrospray ionization (ESI) plus tandem MS (MS/MS) mode.
Flow cytometry.
T84 enterocytes were plated at a density of 2 × 10
5 to 3 × 10
5 per well at 37°C with 5% CO
2. Filtered bacterial supernatants were added at a 1:1 dilution for 72 h. Following incubation, the cells were fixed, permeabilized, and stained with propidium iodide as described previously (
22). Flow cytometry was performed using an LSRII (BD Biosciences), and the data were analyzed using FlowJo software (Tree Star Inc., Ashland, OR).
Cytotoxicity of AHC-6 culture supernatants.
HeLa cells (0.8 × 10
5 to 1.0 × 10
5) were seeded in 96-well plates containing Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 μg/mL of penicillin, and 100 μg/mL of streptomycin and allowed to attach for 24 h.
K. oxytoca strains were cultured in tryptic soy (CASO) broth with shaking for 50 h, and samples were taken during cultivation. Conditioned media were cleared by centrifugation (5 min, 10,000 rpm) and filtered through a sterile 0.2-μm syringe filter, and 50 μL of 1:27 dilutions was added per well. After 2 days of incubation at 37°C with 5% CO
2, cell survival was determined via MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay as described previously (
24).
Murine infection model.
Animal experiments were performed as previously described (
35). Adult female C57BL/6NRj mice with specific-opportunistic-pathogen-free status were purchased from Janvier labs and housed under specific-pathogen-free conditions in individually ventilated cages. Mice (age 8 weeks) were administered amoxicillin-clavulanic acid in drinking water to final concentrations of 0.4 mg/mL of amoxicillin and 0.04 mg/mL of clavulanate 24 h before infection with 1 × 10
7 CFU of
K. oxytoca AHC-6
aphA (Kan
r) (
35) or the Δ
tnaA mutant (
36). Fecal pellets were collected 6 h after gavage and daily thereafter. Quantification of bacteria in inocula and in feces of colonized animals (CFU per gram of stool) was performed by plating of serial dilutions on tryptic soy (CASO) agar (40 μg/mL of kanamycin). Fecal metabolites were extracted and quantified as previously described (
43).
Test substances and antibacterial assay.
Tilimycin and tilivalline were synthesized by the Chemical Synthesis and Biology Core Facility, Albert Einstein College of Medicine (Bronx, NY), and Graz University of Technology. Tilivalline also was purchased (Santa Cruz, Dallas, TX). The antibacterial effects of tilimycin and tilivalline were determined using a disk diffusion assay as previously described (
81) at the concentrations indicated in the figure legends.
Microtiter plate biofilm formation assay.
Biofilm formation by AHC-6
ΔtnaA strain was assessed by microtiter plate assay (
82). Briefly, 20 μL of bacterial suspension at an OD
600 of 0.5 was inoculated in each well containing 180 μL of LB broth with increasing concentrations of indole. Following incubation for 48 h, the wells were decanted and washed, and 0.1% crystal violet solution was added for 15 min. The wells then were washed and dried at room temperature for up to 24 h. Finally, the cell-bound crystal violet was dissolved in 30% acetic acid in water. Biofilm growth was quantified using a microplate reader (Bio-Rad) at 570 nm using 30% acetic acid in water as the blank.
Molecular docking analysis of tilivalline.
The PXR crystal structure (PDB code
1M13 [
83]) was prepared for docking studies of tilivalline as described previously (
84). The structure was then energy minimized and refined using molecular dynamics simulations using Amber charges and Amber force field as adopted in the modeling program MOE (Molecular Operating Environment; version 2016.0801). Tilivalline was modeled using the ligand builder module of MOE and optimized for geometry. The ligand binding domain of PXR has been previously validated to bind indole-like molecules (
44). Hence, tilivalline was docked to the ligand binding domain of PXR using GOLD suite version 5.5.0 (CCDC, Cambridge, UK) (
85); 20 independent runs were performed to completely sample the receptor and ligand conformational space.
hPXR competitive ligand binding assay.
hPXR competitive ligand binding assay was carried out using a LanthaScreen time-resolved fluorescence resonance energy transfer (TR-FRET) PXR competitive binding assay kit (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. Test compounds (tilivalline and tilimycin), SR12813 (1 μM; positive control), or dimethyl sulfoxide (DMSO; 1%) to serve as a vehicle control was incubated with the Fluormone PXR (SXR) green (fluorescein-labeled PXR ligand) (40 nM), human PXR-LBD (glutathione S-transferase [GST] labeled) (5 nM), LanthaScreen Tb-anti-GST antibody (10 nM), and dithiothreitol (50 nM) mixed in TR-FRET PXR (SXR) assay buffer. The assay was carried out in a 384- or 96-well plate, and results were quantified using a BMG LABTECH PHERAstar fluorescent plate reader or Tecan Infinite F200 Pro plate reader (Schoeller Instruments, Czech Republic) using an excitation wavelength of 340 nm, emission wavelengths of 490 nm for terbium and 520 nm for fluorescein, a delay time of 100 μs, and an integration time of 200 μs. Half-maximal inhibitory concentration (IC50) for a test compound was determined from the calculated TR-FRET ratio for emission (520:495) and dose-response curve.
Assays with PXR agonists.
PXR-expressing LS180 enterocytes (
45) were plated at a density of 5 × 10
5/well in 12-well plates. Twenty-four hours later, cells were treated with DMSO (0.1% [vol/vol]; vehicle), tilivalline, and rifaximin at the concentrations indicated in the figure legends. Drug-containing medium was renewed at 24 h. RNA was isolated at 48 h and processed for gene expression as described above. Murine small intestinal crypts were isolated from C57BL/6J mice as previously described (
86). Basolateral treatment of intestinal organoids was performed by adding compounds with final solvent concentrations of 0.3%. Plates were incubated at 37°C with 5% CO
2, and media and treatment were renewed every 2 to 4 days. RNA was isolated from 100 to 1,000 organoids/treatment on day 7 as described above. Primary human hepatocytes in monolayer batches Hep22001014 (male, 76 years, unknown ethnicity) and Hep22001015 (male, 72 years, unknown ethnicity) were purchased from Biopredic International (Rennes, France). Primary human hepatocyte culture from multiorgan donor LH79 (male, 60 years, Caucasian) was prepared at the Faculty of Medicine, Palacky University Olomouc. Primary human hepatocyte cultures were cultured in serum-free ISOM medium.
siRNA transfection.
hPXR knockdown in LS180 cells was performed by reverse transfection of small interfering RNA (siRNA) according to the manufacturer’s (Dharmacon, Lafayette, CO) instructions. Briefly, 50 nM ON-TARGET SMARTpool PXR siRNA (L-003415-00-005) or nontargeting siRNA (D-001810-01-05) duplex was complexed with 2.5 μL of Lipofectamine RNAiMAX (Invitrogen, Waltham, MA) in Opti-MEM reduced-serum medium and added to the 12-well plate. A total of 3 × 105 cells then were mixed with siRNA complex in wells and incubated at 37°C with 5% CO2. The PXR knockdown efficiency, determined by qPCR, was greatest at 48 h (80 to 85% of control). At 48 h posttransfection, the cells were treated with DMSO (0.1%), 10 μM tilivalline, or 10 μM rifaximin (Sigma) for another 48 h. At the end of drug treatment, mRNA was extracted and target gene expression was determined using RT-qPCR as described above.
Cell culture and generation of stable clones.
LS180 cells were maintained in modified Eagle’s minimal essential medium (EMEM; Invitrogen) supplemented with 10% heat-inactivated FBS, 2 mM l-glutamine, 1% nonessential amino acids (NEAA), 1 mM sodium pyruvate, and 25 mM HEPES without antibiotics at 37°C with 5% CO2. T84 cells were maintained in DMEM–F-12 (1:1) supplemented with 5% FBS, 2 mM l-glutamine, and no antibiotics at 37°C with 5% CO2. For transfection, T84 cells were plated in 6-well plates at 7 × 105 per well. After 24 h, the cells were transfected with pcDNA3-hPXR or vector plasmid using Lipofectamine-LTXplus (Thermo Fisher Scientific). After 72 h of transfection, the cells were selected with 0.8 mg/mL of G418 for 21 days to establish stable clones. The expression level of PXR was examined with a TaqMan gene expression assay kit (PXR, Hs011265_g1, and ACTB, Hs01060665_g1). Stable clones were maintained with a medium containing 0.4 mg/mL of G418.
Immunoblotting and caspase-3 activity.
T84 enterocytes were rinsed with cold phosphate-buffered saline (PBS) and scraped into radioimmunoprecipitation assay (RIPA) lysis buffer supplemented with sodium orthovanadate and protease inhibitor cocktail (Roche, Basel, Switzerland). Samples were normalized for protein concentration using the bicinchoninic acid (BCA) method according to the manufacturer’s instructions (Pierce, Thermo Fisher Scientific). Whole-cell lysates containing 20 μg of total protein were resolved on 10% SDS-polyacrylamide gels and transferred to nitrocellulose membranes (0.45 μm; Bio-Rad). Membranes were blocked with 5% nonfat dry milk (Bio-Rad) in PBS for 1 h and then incubated for 2 h with specific mouse monoclonal primary antibodies against acetylated tubulin (clone 6-11B-1; Sigma) and total tubulin (clone DM1A; Sigma). Immunocomplexes were detected with horseradish peroxidase (HRP)-conjugated goat anti-mouse secondary antibody (catalog no. 31432; Thermo Fisher) followed by enhanced chemiluminescence (GE, Boston, MA). β-Actin was used as an internal reference and probed with an HRP-conjugated mouse monoclonal antibody (clone AC-15; Sigma). Protein bands were visualized using a ChemiDoc-XRS+ apparatus (Bio-Rad), and band intensity was analyzed with ImageJ 1.46r software (National Institutes of Health, Bethesda, MD). For caspase-3 activity, the cells were plated at a density of 1 × 106 per well in 6-well plates and cultured for 24 h at 37°C with 5% CO2. Test compounds were applied to the cells and 48 h later, the cells were harvested and caspase-3 activity was measured using a caspase-3 assay kit (Abcam, Cambridge, United Kingdom).
Statistics.
The analyses were carried out using GraphPad Prism version 9.0 (GraphPad Software, San Diego, CA). Mann-Whitney U test was used for comparing responses between two groups of bacteria, Kruskal-Wallis test followed by Dunn’s posttest was used for bacterial responsiveness to increasing concentrations of exogenous indole and biofilm formation, and one-way analysis of variance (ANOVA) with multiple comparisons was used for determining induction of PXR target genes. A P value of ≤0.05 was considered significant.
Ethics statement.
Studies with bacterial isolate UCH-1 were approved by the Institutional Review Board of Connecticut Children’s Medical Center (no. 16-001). For studies with primary human hepatocytes, the tissue acquisition protocol complied with the regulation issued by the Ethical Committee of the Faculty Hospital Olomouc, Czech Republic and with Transplantation law 285/2002 Coll. Studies with mice were performed in accordance with the Commission for Animal Experiments of the Austrian Ministry of Science (GZBMWFW-66.007/0002-WF/V/3b/2017) and the local ethics committee.
Data availability.
The sequenced genome of UCH-1 was assembled, annotated, and submitted to the NCBI BioProject database under accession number
PRJNA608440 (
22).
ACKNOWLEDGMENTS
We thank Juan Salazar and James Moore for their encouragement and support, S. Schild for helpful discussions, and M. Neger for technical assistance. We thank G. Raber at the NAWI Graz Central Lab—Environmental Metabolomics.
This work was supported by funds from the Connecticut Children’s Department of Research (to A.P.M., M.C., and J.D.R.), the Connecticut Children’s Stevenson Fund for Microbiome Research (to A.P.M.), the Office of the Assistant Secretary of Defense for Health Affairs through the Peer Reviewed Medical Research Program—Investigator Initiated Research Award under award no. W81XWH-17-1-0479 (to S.M.), Czech Science Foundation grant 20-00449S (to Z.D.), the Austrian Science Fund (FWF) doc.fund Molecular Metabolism (DOC 50 to E.L.Z.) and the DK Molecular Enzymology (W901 to E.L.Z.) and the BioTechMed Flagship “Secretome.”
N.L., M.M., K.R., S.P., A.P., K.K., P.I., Sandhya Kortagere, Sabine Kienesberger, A.C., L.P., and S.G. performed experiments and analyzed data. J.L. and A.K. assisted with experiments and commented on the manuscript. A.P.M., M.C., J.D.R., Y.Z., SM, Z.D., and E.L.Z. designed and directed experiments and interpreted and analyzed data. A.P.M., N.L., M.C., J.D.R., S.M., Z.D., A.P., K.R., and E.L.Z. wrote the manuscript.
We declare that no conflict of interest exists.