Endogenous Tryptophan-Derived Ah Receptor Ligands are Dissociated from CYP1A1/1B1-Dependent Negative-Feedback

Chemicals and reagents

All chemicals unless otherwise stated were from Sigma. Indole-3-propionic acid and indole-3-lactic acid were from Alfa Aesar (Heysham, UK). FICZ was purchased from TOCRIS Biosciences (Bristol, UK). TCDD was kindly provided by Steve Safe (Texas A&M University).

Cell culture

Hepa 1 and Caco2 cells were obtained from American Type Culture Collection and maintained in α-minimal essential medium (Sigma-Aldrich, St. Louis, MO) supplemented with 10% or 15% fetal bovine serum (Hyclone Laboratories, Logan, UT), respectively, and 100 U/mL penicillin and 100 µg/mL streptomycin. Cells were maintained at 37°C in a humidified incubator in 95% air and 5% CO₂.

CYP1A1 microsomal activity assay

Caco2 cells and Hepa 1 cells were treated with 5 nM 2,3,7,8-tetrachlorodibenzo-p-dioxin for 24 hours to induce CYP1A1 protein expression. Microsomes were isolated from trypsinized cells washed with phosphate-buffered saline. The cell pellet was resuspended in 0.25 M sucrose, 10 mM Tris-HCl (pH 7.5), with 1 X protease inhibitors (cOmplete Mini, Roche). The cells were homogenized in a Dura-Grind Dounce-Type stainless steel homogenizer (Wheaton). The resulting cell lysate was centrifuged at 10 000×g for 10 minutes at 4°C and the supernatant transferred and centrifuged at 100 000×g for 90 minutes at 4°C. The microsomal pellet was resuspended in the homogenization buffer, aliquoted and stored at −80°C for CYP1A1 activity assay. Microsomal protein content was determined using the BCA protein assay system (Pierce, Rockford, IL). The P450-Glo CYP1A1/1B1 assay kit, NADPH regeneration system and luciferase assay kit were purchased from Promega (Promega, Madison, WI). P450-Glo assays were performed following manufacturer protocol. Briefly, 25 µL 4X reaction mixtures were assembled, comprising 10 µg microsomal protein, test compounds (4X final concentration), 120 µΜ luciferin substrate (CEE-luciferin, CYP1A1/B1), and 400 mM potassium phosphate buffer, pH7.4. 4X reaction mixtures were preincubated at 37°C for 10 minutes in white-walled opaque 96-well plates. Reaction mixtures were simultaneously initiated and diluted to 1X by adding of 25 µL NADPH regenerating system. Reactions were incubated at 37°C for 25 minutes. Reactions were terminated by addition of 50 µL luciferin detection reagent. Following incubation at 37°C for 10 minutes, luciferin-based luminescence was quantified using a Synergy HTX multi-mode plate reader (BioTeK, Winooski, VT). Assays were performed in triplicate and data expressed as mean ± SD percent P450 activity relative to vehicle treated.

Mouse experiments

C57BL6/J mice were obtained from Jackson Laboratory (Bar Harbor, ME), bred in-house and maintained on a chow diet. Ahr^+/−, and Ahr^−/− mice were kindly provided by Dr. Christopher Bradfield (University of Wisconsin-Madison) and bred using female Ahr^+/− and male Ahr^−/− mice. Mice were housed on corncob bedding in a temperature- and light-controlled facility and given access to food and water ad libitum. Mice were maintained in a pathogen-free facility and treated humanely with approval from the Animal Care and Use Committee of the Pennsylvania State University and methods were carried out in accordance with approved guidelines. Mice in the 8 to 10-week-old range were utilized in all experiments and were euthanized by over exposure to CO₂, blood was collected after cutting the portal vein. Serum samples were obtained from mice exposed to PCB126 for 6 days as previously described. ¹⁵ Briefly, each mouse was fed a dough pill containing 24 µg/kg PCB126 every 24 hours for 5 days and mice were sacrificed on day 6.

Quantification of tryptophan metabolites in mouse serum

Serum stored at −80°C was thawed on ice, 25 µL was mixed with 100 µL extraction solvent of ice-cold methanol containing isotope labeling standards (indole-3-acetic acid-d4 and kynurenic acid-d5). Mixture was vortexed and stored at −20°C for 30 minutes to precipitate protein. Following centrifugation at 12 000×g for 15 minutes at 4°C, the supernatants were collected and subsequently evaporated to dryness (Thermo Scientific, Waltham, MA) and dissolved in 45 µL of 10% acetonitrile. After centrifugation at 12 000×g for 15 minutes at 4°C, supernatants were transferred to autosampler vials for LC-MS analysis. Quantitative analysis of tryptophan metabolites was performed by reverse phase UHPLC using a Prominence 20 UFLCXR system (Shimadzu, Columbia, MD) with a water (Milford, MA) BEH C18 column (2.1 × 100 mm × 1.7 µm particle size) maintained at 55°C with a 20 minutes aqueous acetonitrile gradient, at a flow rate of 250 µL/minute. Solvent A was LC/MS grade water with 0.1% formic acid and Solvent B was LC/MS grade acetonitrile with 0.1% formic acid. The initial conditions were 97% A and 3% B, increasing to 45% B at 10 minutes, 75% B at 12 minutes then held at 75% B until 17.5 minutes before returning to the initial conditions. The eluate was delivered into a 5600 (QTOF) TripleTOF using a Duospray™ ion source (all Sciex, Framingham, MA). The results were quantified by comparing the integrated peaks against a standard curve. The recovery and matrix effects for the tryptophan metabolites in mouse serum were used to normalize the quantitative data as previously described. ¹⁴

Computational CYP modeling and docking analysis

The ligand binding affinity and target oxidation sites for a panel of tryptophan metabolites, including FICZ, kynurenine, kynurenic acid, indole-3-acetate, indole-3-lactate, indole-3-propionate, and indole-3-carboxylate were investigated using Autodock Tools 4 and Autodock Vina, and 4 pre-validated, structure-based, computational models of mouse and human CYPs 1A1 and 1B1.^{16 -18} Crystal structures of human CYP1A1 (4I8V) and human CYP1B1 (3PM0) were utilized to prepare docking models for the human enzymes, and as templates for the mouse homology models, using the SWISS-MODEL server.^{19 -21} Homology models were validated for overall quality, clashes, and outliers using the SWISS-MODEL server, SAVESv6.0 server (https://saves.mbi.ucla.edu/), and ChimeraX using the ISOLDE plugin.^{22 -26} Computational docking models for each substrate tested were obtained from the PubChem database and optimized for docking using Autodock Tools. Autodock Vina was run using standard settings (exhaustiveness = 10), using Grid box parameters generated in Autogrid for a 30 to 60 Å ³ docking grid centered on the heme center, as described previously.^27,28 Docking results were analyzed using Autodock Tools 4 and The PyMOL Molecular Graphics System, Version 2.52 Schrödinger, LLC. ²⁹ Active site cavity size was calculated using Caver 3.0. ³⁰ Model quality was validated using an Autodock Vina re-docking analysis with α-naphthoflavone (ANF) in structure-based models. RMSD error was calculated using the program LigRMSD, and models with RMSD error <2.0 Å for ANF (as compared to the crystallographic coordinates) were considered sufficiently accurate for comparative docking studies. ³¹

Statistical analysis

Values were expressed as mean ± SEM. Data were compared using one-way analysis of variance with Tukey multiple comparison post-test in GraphPad Prism 9.1.1 (GraphPad Software, Inc, La Jolla, CA, USA) to establish statistical significance between different groups. The value of P < .05 was considered statistically significant (****P < .0001). All experiments were repeated at least twice.

Tryptophan metabolites are not significant CYP1A1/1B1 substrates/inhibitors

We have developed a microsomal assay system that examines CYP1A1/1B1 enzymatic activity. The microsomes utilized are isolated from either mouse hepatoma or human colonic tumor cell lines treated with TCDD and express significant levels of CYP1A1 as previously described. ³² TCDD prompts a significant increase in CYP1A1 expression but importantly does not exhibit CYP1A1 substrate/inhibitor activity. Secondly, a luciferase substrate (luciferin 6′-chloroethylether, CEE) is used that is highly specific for CYP1A1/1B1. This assay was utilized to test whether tryptophan metabolites that can activate the AHR are also CYP1A1/1B1 substrates. Each metabolite was tested at between 0.1 and 3 µM concentrations, which encompasses the concentrations of the 6 significant tryptophan metabolites that are also AHR ligands found in human serum. ¹⁴ None of these metabolites examined exhibited a dose-dependent competition with the CYP1A1/1B1 substrate in either mouse or human microsomal CYP1A1/1B1 assay (Figure 1A and B). In these assays ketoconazole was utilized as a positive control. In addition, FICZ is a potent inhibitor/substrate in this assay, consistent with previous reports (Figure 2). ¹³

OPEN IN VIEWER

Figure 1.

Tryptophan metabolites are not CYP1A1/1B1 competitive substrates/inhibitors. (A) Six tryptophan metabolites; kynurenine, kynurenic acid, indole-3-acetic acid, indole-3-propionic acid, indole-3-lactic acid, or indole-3-carboxaldehyde were co-incubated with Luc-CEE substrate in a Hepa 1 microsomal assay or (B) Caco2 microsomal assay and compared with ketoconazole. The data are mean ± SEM; Tukey’s test.

OPEN IN VIEWER

Figure 2.

FICZ is a CYP1A1/1B1 inhibitor/competitive substrate. Indicated concentrations of FICZ were co-incubated with Luc-CEE substrate in conjunction with Hepa 1-derived microsomes.

CYP1A1 modeling of tryptophan metabolites

While the results in Figure 1 would indicate that the tryptophan metabolites are not CYP1A1 substrates or inhibitors, it is important to recognize that this is in the context of the conditions utilized in the assay. To gain further insight into whether these metabolites have the potential to be inhibitors/substrates we utilized a molecular modeling docking approach. The substrate binding properties of several tryptophan metabolites, including indole-3-acetate, indole-3-lactate, indole-3-propionate, indole-3-carboxaldehyde, kynurenine, kynurenic acid, and 6-formylindolo[3,2-b]carbazole (FICZ) were studied via computational docking analysis with Autodock Vina. Structure-based docking models of human cytochromes P450 1A1 (hCYP1A1) and 1B1 (h1B1), and homology models of mouse cytochromes P450 1A1 (mCYP1A1) and 1B1 (mCYP1B1) were developed and validated to explore CYP enzyme- and species-specific differences in tryptophan metabolite substrate recognition and metabolism (Figure 3; Table 1).

OPEN IN VIEWER

Figure 3.

Comparative Tryptophan Metabolite Docking Study in Structure-based Model of Human CYP1A1. Substrate binding properties of tryptophan related metabolites, FICZ, kynurenine, kynurenic acid, and 4 gut microbiome-derived indoles (3-carboxylate, 3-acetate, 3-lactate, and 3-propionate) were explored with Autodock Vina and a structure-based human CYP1A1 (hCYP1A1) homology model. (A) Here, α-naphthoflavone (ANF) is shown docked in our computational model of hCYP1A1. The heme-centered active site and substrate binding pocket is highlighted, with key active site residues as shown. Important hydrophobic and electrostatic ligand contact distances to key amino acids are shown (in angstroms; Å; yellow dash). (B) FICZ also binds the hCYP1A1 active site with low picomolar affinity, positioning the C4 carbon 4.4 Å from the heme center, identical to the C4 target carbon distance for ANF. FICZ interacts closely with several highly conserved, active site residues, including F123, F224, F258, D313, T321, V382, I385, and L496. (C) Kynurenine docks in hCYP1A1 in an allosteric pocket above the active site, positioning no target carbons within 10 Å of the heme center. This allosteric site is defined by several well-conserved residues (S116, S122, F123, F224, N255, F258, D313) that converge near the top of the canonical active site, which extends deeper into the enzyme, over the heme center. Lower affinity binding modes for kynurenine that positioned the C4 carbon within 5.1 Å of the heme were also observed. (D) The heterocyclic derivative kynurenic acid did not position the substrate significantly closer to the heme center. A secondary docking prediction for kynurenic acid, with ~3-fold lower affinity was also observed that positioned the C3 carbon 4.5 Å from the heme center. (E) Indole-3-carboxylate binds hCYP1A1 in the same allosteric site as kynurenine. (F) Indole-3-acetate also docks in the same allosteric site, positioning no target carbons within 11 Å of the catalytic center. (G) Indole-3-lactate displayed slightly higher affinity to hCYP1A1 than either the carboxylate or acetate derivatives, but still docked far from the heme. (H) Indole-3-propionate displayed higher affinity for hCYP1A1 than the other indole-derivatives, despite binding in the same allosteric site above the catalytic center.

OPEN IN VIEWER

Table 1.

Computational docking analysis of tryptophan metabolites in human and mouse CYP1A1 and CYP1B1.

Substrate		Autodock Vina
		Structure-based homology models
		Human Cyp1A1	Mouse Cyp1a1	Human Cyp1B1	Mouse Cyp1b1
		Binding energy a (kcal/mol) [KD] nM b	Binding energy (kcal/mol) [KD] nM	Binding energy (kcal/mol) [KD] nM	Binding energy (kcal/mol) [KD] nM
FICZ	Max c	−13.8 [0.082]	−13.5 [0.135]	−14.6 [0.021]	−10.1 [3.98]
	Avg d	−13.44 ± 0.37 [0.17 ± 0.10]	−12.98 ± 0.50 [0.43 ± 0.34]	−13.38 ± 1.08 [0.19 ± 0.22]	−9.45 ± 0.83 [197 ± 188]
Kynurenine	Max	−7.9 [1575]	−8.0 [1332]	−8.0 [1332]	−7.5 [3074]
	Avg	−7.49 ± 0.31 [3512 ± 1726]	−7.24 ± 0.45 [4481 ± 2701]	−7.33 ± 0.57 [4874 ± 5940]	−6.88 ± 0.38 [6665 ± 3686]
Kynurenic acid	Max	−9.1 [212]	−9.1 [212]	−9.4 [128]	−8.8 [350]
	Avg	−8.64 ± 0.38 [537 ± 305]	−8.29 ± 0.33 [678 ± 277]	−8.61 ± 0.83 [377 ± 240]	−8.02 ± 0.61 [1879 ± 1567]
Indole-3-carboxylate	Max	−7.8 [1862]	−7.6 [2601]	−8.3 [807]	−7.6 [954]
	Avg	−7.28 ± 0.27 [4353 ± 1835]	−7.34 ± 0.24 [4273 ± 1652]	−7.70 ± 0.27 [2378 ± 889]	−7.22 ± 0.35 [4895 ± 4249]
Indole-3-acetate	Max	−7.9 [1575]	−8.2 [954]	−8.3 [807]	−7.9 [2600]
	Avg	−7.43 ± 0.35 [4049 ± 2259]	−7.47 ± 0.22 [3594 ± 1377]	−7.90 ± 0.36 [1880 ± 1332]	−7.32 ± 0.45 [4006 ± 2874]
Indole-3-lactate	Max	−8.2 [954]	−8.6 [489]	−8.6 [489]	−7.9 [1575]
	Avg	−7.89 ± 0.28 [1769 ± 867]	−7.73 ± 0.59 [3004 ± 2384]	−8.21 ± 0.31 [1053 ± 538]	−7.44 ± 0.38 [3988 ± 2426]
Indole-3-propionate	Max	−8.4 [683]	−8.5 [577]	−9.0 [250]	−8.1 [1127]
	Avg	−7.68 ± 0.48 [2865 ± 1635]	−8.04 ± 0.31 [1378 ± 628]	−8.56 ± 0.30 [587 ± 282]	−6.58 ± 0.93 [27 378 ± 28 083]

a

Substrate Binding Energies were derived computationally using Autodock Vina.

b

Dissociation binding constants (K_D) (in the nanomolar [nM] scale), were derived computationally from Autodock Vina data using the Autodock 4.2 inhibitory constant conversion scale, based on the equation: y = 0.5982ln(x) − 12.304.

c

Max refers to the low-energy Vina docking solution (in kcal/mol), with the highest predicted binding affinity [K_D], for a given CYP model. If multiple results had the same binding energy, the pose with the closest contact distance to the heme center was selected.

d

Avg refers to the average or cumulative dissociation constant and binding energy (± the standard deviation) for the Vina docking poses obtained for each individual substrate:model docking combination (N < 10).

In Figure 3A, the α-naphthoflavone (ANF) docking, an established CYP1A1 substrate, is shown in the heme-centered, substrate binding pocket of human CYP1A1 (hCYP1A1), with key active site residues depicted. ANF docks in the hCYP1A1 model with low picomolar affinity (−13.7 kcal/mol; K_D = 97 pM; see Table S1), with low root-mean-square error (0.44 Å) compared to the authentic structural coordinates (PDB:4I8V; see Table S1). Comparative tryptophan metabolite docking revealed that the endogenous ligand FICZ also binds hCYP1A1 with low picomolar affinity (−13.8 kcal/mol; K_D = 82 pM; Table 1), in a conformation similar to ANF that positions the C4 carbon 4.4 Å from the heme center (Figure 3B). Similar docking results were obtained for mouse CYP1A1 (mCYP1A1) and human (hCYP1A1) and mouse (mCYP1B1) forms of CYP1B1 (Table 1).

While di-indolyl-compounds like FICZ were found to bind tightly to hCYP1A1 similar to ANF, smaller, aromatic ketone derivatives of tryptophan, like kynurenine, did not (Figure 3A-C). Kynurenine docks hCYP1A1 with low micromolar affinity (−7.9 kcal/mol; K_D = 1.58 µM), in an allosteric pocket formed above the active site, positioning no target carbons within 10 Å of the heme center. Lower affinity docking results for kynurenine that positioned the C4 carbon within 5.1 Å of the heme were also observed (−7.7 kcal/mol; 2.20 µM; not shown), but were less common in general, among all 4 models tested. It is notable that the conversion of kynurenine to the larger, heterocyclic derivative kynurenic acid increased substrate binding affinity to hCYP1A1 over 7-fold (−9.1 kcal/mol; K_D = 212 nM), without forcing the substrate much deeper into the heme pocket (Figure 3D). A secondary docking prediction for kynurenic acid, with ~3-fold reduced affinity (−8.5 kcal/mol; K_D = 577 nM) was also observed, which positioned the C3 carbon only 4.5 Å from the heme center. The low affinity docking results for kynurenine and kynurenic acid suggest they may feasibly be substrates for CYP1A1 and CYP1B1, similar to FICZ, but only when present at higher (micromolar) levels in the cell. This hypothesis is reinforced by convergent docking results for FICZ, kynurenine, and kynurenic acid in all 4 docking models (Table 1).

To further expand our understanding of tryptophan metabolite recognition by the CYP1 family, we next explored the substrate binding properties of 4 additional, microbiome-derived indole compounds, also known to modulate the AHR activity. Tryptophan metabolites, like indole-3-carboxylate, bind hCYP1A1 with, at most, low micromolar affinity (−7.8 kcal/mol, K_D = 1.86 µM) in the same conserved, allosteric site as kynurenine and kynurenic acid (Figure 3E). Indole-3-acetate binds hCYP1A1 with similar, low micromolar affinity (−7.9 kcal/mol, K_D = 1.58 µM), in the same allosteric site, positioning no target carbons within 11 Å of the catalytic center (Figure 3F). Indole-3-lactate, which has a longer side chain than either the carboxylate or acetate derivatives, displayed higher affinity to hCYP1A1 (−8.2 kcal/mol; K_D = 954 nM), despite sharing a similar indole ring binding configuration in the outer, allosteric pocket (Figure 3G). Enhanced, nanomolar range affinity was also observed for indole-3-propionate (−8.4 kcal/mol; K_D = 683 nM), despite sharing a similar docking configuration in the distal, allosteric pocket of the CYP1 family active site (Figure 3H).

Collectively, our docking results suggest that tryptophan metabolites containing only one indole ring, are poor CYP1 family substrates, because of their tendency to bind allosteric sites outside (near the mouth) of the enzyme’s narrow, hydrophobic active site. Furthermore, results with kynurenine and kynurenic acid suggest that bacterial oxidations that generate a thiazole ring structure in the tryptophan derivative can increase substrate affinity significantly, improving accessibility to the catalytic center. Indole-derivatives with larger functional groups (eg, lactate or propionate) also revealed enhanced affinity to the enzyme, despite binding far from the heme center. Both CYP1A1 and CYP1B1 have a stronger binding preference for elongated and planar, polycyclic aromatic hydrocarbons, and heterocyclic compounds with multiple rings, than simple aromatics, like kynurenine, and the small microbiome-derived indoles.

The lack of AHR expression or AHR activation in mice fails to alter tryptophan metabolite levels

CYP1A1 expression is essentially dependent on AHR expression in mice. ⁴ Therefore, we tested whether in mice the level of circulating tryptophan metabolites differs in Ahr^−/−, Ahr^+/−, and Ahr^+/+ mice (Figure 4A). We included Ahr^+/− mice in this analysis because they are from the same litters as Ahr^−/− mice, this would help rule out genetic differences between different mouse lines. The six tryptophan metabolites analyzed in the in vitro studies were examined in vivo utilizing serum samples from mice on a chow diet. All of the metabolites examined failed to exhibit a significant difference in Ahr^−/− versus Ahr^+/− mice. In contrast, comparison of Ahr^−/− to Ahr^+/+ mice revealed that indole-3-acetic acid was significantly lower in Ahr^+/+ mice, while all other metabolites examined exhibited similar serum levels. The reason is the difference in indole-3-acetate between the Ahr^−/− to Ahr^+/+ mice is not known. Serum samples from a previous study, where mice were exposed to the potent AHR ligand PCB126 for 5 days, were examined for levels of the 6 tryptophan metabolites. ¹⁵ The results indicate that there was no significant difference in the level of these metabolites (Figure 4B), even though Cyp1a1 was induced ~225-fold in liver. ¹⁵

OPEN IN VIEWER

Figure 4.

Serum levels of the tryptophan metabolites examined are not altered by AHR phenotype or PCB126-mediated AHR activation in vivo. LC-MS analysis was utilized to quantitate tryptophan metabolites in mouse serum. (A) Concentration of serum tryptophan metabolites were determined in Ahr^+/+, Ahr^+/−, and Ahr^−/− mice. (B) Concentration of serum tryptophan metabolites in mice treated with PCB126 for 6 days. There were 6 mice in each group in both studies, except the Ahr^+/+ mice in panel A utilized 11 mice. The data are mean ± SEM; Tukey’s test.