Involvement of CYP1B1 in interferon γ‐induced alterations of epithelial barrier integrity

Abstract

Background and Purpose

CYP1B1 and CYP1A1 are important extra‐hepatic cytochromes, expressed in the colon and involved in the metabolism of dietary constituents and exogenous compounds. CYP1B1 expression is increased by pro‐inflammatory cytokines, and it has been recently implicated in regulation of blood brain barrier function. We investigated its involvement in the increased permeability of the intestinal epithelial barrier observed in inflammatory conditions.

Experimental Approach

Epithelial monolayers formed by human T84 colon carcinoma cells cultured on transwells, were disrupted by incubation with IFNγ (10 ng·mL⁻¹). Monolayer integrity was measured using transepithelial electrical resistance. CYP1A1 and CYP1B1 inhibitors or inducers were applied apically. Potential mechanisms of action were investigated using RT‐qPCR.

Key Results

IFNγ disrupts the barrier integrity of the T84 monolayers and increases CYP1B1 and HIF1α mRNA expression. CYP1B1 induction is inhibited by the NF‐κB inhibitor ammonium pyrrolidinedithiocarbamate (100 μM) but not by the HIF1α inhibitor 3‐(5′‐hydroxymethyl‐2′‐furyl)‐1‐benzyl indazole (50 μM). Inhibition of CYP1B1 with the selective inhibitor 2,4,3′,5′‐tetramethoxystilbene (100 nM) partly reverses the effects of IFNγ on epithelial permeability.

Conclusions and Implications

These data suggest that increased expression of CYP1B1 is involved in the effects of IFNγ on epithelial permeability. Inhibition of CYP1B1 counteracts the alterations of epithelial barrier integrity induced by IFNγ and could thus have a therapeutic potential in disorders of intestinal permeability associated with inflammation.

Abbreviations

9,10‐DiHOME

9(10)‐dihydroxy‐12Z‐octadecenoic acid

12,13‐diHOME

12(13)‐dihydroxy‐9Z‐octadecenoic acid

11‐HETE

11‐hydroxy‐5Z,8Z,12E,14Z‐eicosatetraenoic acid

12‐HETE

12‐hydroxy‐5Z,8Z,10E,14Z‐eicosatetraenoic acid

15‐HETE

15‐hydroxy‐5Z,8Z,11Z,13E‐eicosatetraenoic acid

5‐HETE

5‐hydroxy‐6E,8Z,11Z,14Z‐eicosatetraenoic acid

9‐HODE

9‐hydroxy‐10E,12Z‐octadecadienoic acid

13‐HODE

13‐hydroxy‐9Z,11E‐octadecadienoic acid

12(S)‐HEPE

12S‐hydroxy‐5Z,8Z,10E,14Z,17Z‐eicosapentaenoic acid

5(S),6(R)‐LXA₄

5S,6R,15S‐trihydroxy‐7E,9E,11Z, 13E‐eicosatetraenoic acid, lipoxin A4

AhR

aryl hydrocarbon receptor

HIF1α

hypoxia‐inducible factor 1α

MDR

multidrug resistance protein 1

PDTC

ammonium pyrrolidinedithiocarbamate

PEST

penicillin + streptomycin

TEER

transepithelial electrical resistance

TMS

2,4,3′,5′‐tetramethoxystilbene

YC‐1

3‐(5′‐hydroxymethyl‐2′‐furyl)‐1‐benzyl indazole

α‐Naphthoflavone

2‐phenyl‐4H‐naphtho[1,2‐b]pyran‐4‐one

Introduction

The gastrointestinal tract is continuously exposed to dietary components and antigens, as well as a diversity of microorganisms. The intestinal epithelium is a selective barrier that plays an essential role in maintaining gut homeostasis. It is permeable to nutrients and essential macromolecules but constitutes an obstacle for luminal antigens, intestinal bacteria and harmful macromolecules (Turner, 2009). The intestinal epithelial barrier also plays a fundamental immunoregulatory function by participating in the coordination of immune responses (Peterson and Artis, 2014). Stimuli such as cytokines, nutrients and bacteria can modulate epithelial permeability (Turner, 2009; Natividad and Verdu, 2013; Bischoff et al., 2014), and such alterations in permeability of the intestinal epithelium have been linked to a number of intestinal diseases such as inflammatory bowel diseases, inflammatory bowel syndrome, colorectal cancer and celiac disease (Bischoff et al., 2014).

Human colon carcinoma cell lines such as Caco‐2 and T84 cells, when grown to confluence in monolayers, have excellent barrier properties, as demonstrated by a low permeability to agents such as mannitol, a high transepithelial electrical resistance (TEER), and the development of tight junctions on the apical side. These cells have long been used for the in vitro study of drug absorption across the intestinal barrier (Artursson et al., 2001) and of the effects of inflammatory stimuli and pathogenic agents upon intestinal barrier function. T84 cells, for example, respond to the presence of pathogens such as the Vibrio cholerae O1 strain C6706 and Escherichia coli MG1655 by producing inflammatory cytokines (Ou et al., 2009b; Monnappa et al., 2016). The cytokine IFNγ plays a pivotal role in inflammation and it has been implicated in the disruption of the intestinal epithelial barrier (Hu and Ivashkiv, 2009; Yang et al., 2014). Treatment of T84 cells with IFNγ increases their permeability in a manner involving the NF‐κB pathway (Youakim and Ahdieh, 1999; Bruewer et al., 2005; Willemsen et al., 2005; Boivin et al., 2009; Yang et al., 2014).

CYP1B1 is one of the main extra‐hepatic cytochromes. It has mostly been studied in cancer development and therapy due to its high expression in tumours compared to surrounding tissue (Murray et al., 2001; Gibson et al., 2003; Kumarakulasingham et al., 2005) and to its ability to metabolize exogenous compounds such as halogenated aromatic hydrocarbons, polycyclic aromatic hydrocarbons and even dietary flavonoids, into their carcinogenic derivatives (Murray et al., 2001; Androutsopoulos et al., 2009). Most of these CYP1B1 substrates are also strong inducers of CYP1B1 and CYP1A1 by activation of the aryl hydrocarbon receptor (AhR) (Murray et al., 2001). CYP1B1 is expressed in extrahepatic human tissues including the colon (Gibson et al., 2003; Kumarakulasingham et al., 2005; Bieche et al., 2007) and in colon cell lines in culture such as Caco‐2, SW480 and HCT116 cells (Niestroy et al., 2011; Patel et al., 2014). Interestingly, while the expression of the other CYP1 family members, CYP1A1 and CYP1A2, is generally down‐regulated by inflammation (Morgan, 2001), CYP1B1 expression is induced by inflammatory cytokines (Piscaglia et al., 1999; Bleau et al., 2003; Umannová et al., 2008). Moreover, CYP1B1 also metabolizes endogenous compounds such as arachidonic acid to give hydroxyeicosatetraenoic acids (HETEs), which are involved in inflammation in general and inflammatory bowel diseases in particular (Masoodi et al., 2013). This oxidation of arachidonic acid by CYP1B1 is of biological significance as CYP1B1‐produced mid‐chain HETEs have been recently implicated in doxorubicin‐induced cardiotoxicity (Maayah et al., 2016).

As CYP1B1 is expressed in the colon, is implicated in the metabolism of dietary constituents and exogenous compounds and is increased by pro‐inflammatory cytokines, it is possible that it is involved in the increased permeability of the intestinal epithelial barrier observed in inflammatory conditions. To investigate this possibility, we used IFNγ to disrupt the epithelial monolayer formed by human T84 colon carcinoma cells and investigate the effects of CYP1B1 on epithelial permeability.

Methods

Cell culture

Human T84 colon carcinoma cells (American Type Culture Collection, Rockville, MD, USA) that had been selected for capacity to form polarized, tight monolayers on semipermeable support (Ou et al., 2009a) and thereafter frozen directly after expansion were cultured at 37°C in a 5% CO₂ environment in DMEM/F12 medium supplemented with 8% FBS, 1% penicillin + streptomycin (PEST) and 1% L‐glutamine. To maintain consistency, cells were kept in culture for 10 passages and used from passage 2 to 10 after defrosting the cryotubes. T84 cells were subcultured after partial digestion with 0.25% trypsin and counted with Trypan Blue before plating. For dose‐dependence testing of the compounds, cells were seeded overnight at a density of 2.5 × 10⁵ cells per well in 24‐well plates. The next day, 2,4,3′,5′‐tetramethoxystilbene (TMS), benzo[a]pyrene, α‐naphthoflavone or vehicle (DMSO, 0.1% v/v) were added (concentration range 0.01–10 μM) for 8 h. For the experiments with the inhibitor of NF‐κB ammonium pyrrolidinedithiocarbamate (PDTC) or with YC‐1, the inhibitor of the hypoxia‐inducible factor 1α (HIF1α), cells were seeded overnight at a density of 2.5 × 10⁵ cells per well in 24‐well plates. The next day, PDTC (100 μM), YC‐1 (50 μM), the combination of both or vehicle (DMSO, 0.1% v/v) were added for 1 h before incubation of the cells with IFNγ (10 ng·mL⁻¹) for 8 h. The same cell number was used to seed cells in the transwells. Medium in transwells (500 and 1500 μL in the apical and basolateral compartments respectively) was changed every day, and the cells were used when TEER was consistently >900 Ω·cm² (usually day 9–10). In transwell experiments, IFNγ (10 ng·mL⁻¹) was always added basolaterally for 24 h and the test compounds apically. For lipid analysis, T84 cells were seeded overnight in 6‐well plates at a density of 1 × 10⁶ cells per well. The next day, the medium was replaced with either fresh medium or medium containing IFNγ (10 ng·mL⁻¹) for 8 or 24 h.

Determination of epithelial monolayer resistance

TEER of the T84 monolayers cultured in transwells was measured using an Evohm epithelial voltohmmeter coupled to the endohm‐12 chamber (World Precision Instruments, Hitchin, UK). On the day of the experiment (t0), TEER was measured and then the medium in the basolateral compartment was replaced with 1.5 mL fresh medium or medium containing 10 ng·mL⁻¹ IFNγ. The medium in the apical compartment was replaced with 500 μL of medium containing either vehicle (DMSO, 0.1% v/v) or the CYP inhibitors/inducers (at the dose determined from the 24‐well experiments). After 24 h of incubation, TEER was measured again to assess the effect of IFNγ and the compounds. The transwells were then either used for mRNA extraction or for paracellular permeability assays (the latter are described in the Supporting Information Figures S1 and S3).

mRNA extraction and RT‐qPCR

T84 cells were seeded overnight into 24‐well plates then incubated with different concentrations of the CYP inhibitors or inducers for 8 h after which the media was removed. The Dynabeads mRNA direct purification kit was used to extract mRNA according to the manufacturer's instructions. Briefly, wells were washed with cold PBS and 300 μL of lysis/binding buffer (from the Dynabeads mRNA direct purification kit) was added before the plates were stored at −80°C until used for assay. For cells cultured in transwells, IFNγ was added to the basolateral compartment and vehicle or compounds were added to the apical compartment for 24 h. TEER was measured before addition of IFNγ and after 24 h, at which time the medium was removed from the inserts and the wells. The inserts were washed with PBS and 600 μL of lysis/binding buffer was added before the plates were stored at −80°C.

The extracted mRNA was quantified in Nanodrop Lite® (Thermo Fisher Scientific), diluted in Tris‐EDTA buffer to a concentration of 5 ng·μL⁻¹ (10 μL total) and mixed with the 2XRT master mix (High capacity cDNA reverse transcription kit). Reverse transcription was run in a Life Touch thermal cycler (BIOER, Hangzhou, China). cDNA was diluted 1/10 and loaded into the assay plates. qPCR was performed with a Eco Real‐Time PCR instrument (Illumina Inc., San Diego, CA, USA). PCR reactions were run using KAPA SYBR FAST qPCR kit Master Mix. Each sample was measured in duplicate. The following conditions were used for amplification: an initial holding stage of 3 min at 95°C, thereafter 45 cycles consisting of denaturation at 95°C for 3 s and annealing/extension at 60°C for 30 s. Products were analysed by performing a melting curve at the end of the PCR reaction. Data were normalized to the 60S ribosomal protein L19 (RPL19). Shown in graphs is the ΔCt [i.e. threshold cycle (Ct) of the gene of interest minus the corresponding Ct of the reference gene (RPL19)]. A ΔCt of −1 corresponds to a doubling of the mRNA. The primers used are shown in Table 1.

Table 1.

Primer sequences used for qRT‐PCR

	Sequence access numbersa	Forward primer (5′ to 3′)	Reverse primer (5′ to 3′)
AHR	NM_001621.4	GGACTTGGGTCCAGTCTAATG	CCTTCCTCATCTGTTAGTGGTC
CLDN2	NM_020384.3	CAAAGACAGAGTGGCGGTAG	TGGTGAGTAGAAGTCCCGTAG
CLDN4	NM_001305.4	CGCACAGACAAGCCTTACTC	GGGAAGAACAAAGCAGAGAGG
CLDN5	NM_001130861.1	ACCTTCTCCTGCCACTAGC	CCGCTCTGCCTATGGAAAC
CYP1A1	NM_000499.3	TGAGAACGCCAATGTCCAGC	GCCGTGACCTGCCAATCA
CYP1B1	NM_000104.3	CTGCGGGTTCCTGTTGACG	CCAAGGGTCGTTCGGGCT
HIF1α	NM_001530.3	GCTGATTTGTGAACCCATTCC	TTCATATCCAGGCTGTGTCG
IL‐8	NM_000584.3	GGCAGCCTTCCTGATTTCTG	GGGTGGAAAGGTTTGGAGTATG
MDR1	NM_001348945.1	CTCCTGACTATGCCAAAGC	TCCTTCCAATGTGTTCGG
OCCLN	NM_002538.3	AGAGCAGGAAGGTCAAAGAG	GGATATTCCCTGATCCAGTCC
RPL19	NM_000981.3	CACATCCACAAGCTGAAGGCA	CTTGCGTGCTTCCTTGGTCT
TJP1	NM_003257.4	TTCAAAGGGAAAGCCTCCTG	CTGAGATGGCTGGGCATAC
TNFα	NM_000594.3	AAGCCTGTAGCCCATGTTGT	GCTGGTTATCTCTCAGCTCCA

^aSequence access numbers used to design the primers. Note that some have several transcript variants and the primers were designed to recognize all transcript variants.

Immunofluorescence

Confluent T84 monolayers were washed with PBS and fixed for 15 min at room temperature with 4% paraformaldehyde in PBS. Inserts were then washed twice with PBS then twice with PBS containing 0.1% BSA before being incubated for 30 min at room temperature in blocking buffer (PBS with 5% BSA and 0.3% Triton X100). After three washes with PBS, the inserts were incubated for 1 h at room temperature with the antibodies diluted in PBS with 1% BSA and 0.3% Triton‐X100 (both at 1/200). Following three washes with PBS, DNA was labelled with 50 ng·mL⁻¹ DAPI for 5 min at room temperature. The inserts were then washed twice with PBS, then the membranes were cut and mounted using Vectashield® HardSet. Images were captured at 40× magnification using a Nikon DS‐Ri1 camera mounted on a Nikon E800 fluorescence microscope.

Extraction and quantification of oxylipins by UPLC‐ESI‐MS/MS

Cell plates were placed on ice immediately after the incubation period and scraped with 1 mL of methanol. Each sample was centrifuged at 2000 g for 15 min (4°C) to sediment cell debris and the methanol extracts were stored at −80°C until LC‐MS analysis was undertaken (never longer than 2 days to avoid chemical decomposition of the metabolites).

Solid phase extraction (SPE) was performed before LC‐MS/MS analysis based on the method reported by Gouveia‐Figueira and Nording (2015). In summary, Waters Oasis HLB cartridges (60 mg of sorbent, 30 μm particle size) were washed with 2 mL of ethyl acetate, followed by 2 × 2 mL of methanol, after which they were conditioned with 2 × 2 mL of wash solution (95:5 v/v water/methanol with 0.1% acetic acid). The methanolic cell extracts, spiked with 10 μL internal standard (50 ng·mL⁻¹ 12,13‐diHOME‐d₄ and 12(13)‐EPOME‐d₄, 25 ng·mL⁻¹ 9‐HODE‐d₄, PGD₂‐d₄, PGE₂‐d₄, 5‐HETE‐d₈, 20‐HETE‐d₆ and TXB₂‐d₄) and 10 μL antioxidant solution [0.2 mg·mL⁻¹ BHT/EDTA in methanol/water (1:1)], were diluted with milliQ water up to a 5% methanol concentration and applied to an SPE cartridge. Another washing step was done with 2 × 4 mL of wash solution and elution of oxylipins was undertaken with 3 mL acetonitrile, followed by 2 mL of methanol and 1 mL of ethyl acetate into polypropylene tubes containing 6 μL of a glycerol solution (30% in methanol). Eluates were concentrated with a MiniVac system (Farmingdale, NY, USA), redissolved in methanol (100 μL), transferred to LC vials and spiked with 10 μL of a recovery standard (CUDA, 50 ng·mL⁻¹). The full chemical names for the oxylipins abbreviated here are given in the abbreviations list.

LC‐MS/MS analysis was undertaken using an Agilent ultra‐performance (UP)LC system (Infinity 1290) coupled with an electrospray ionization source (ESI) to an Agilent 6490 Triple Quadrupole system equipped with the iFunnel Technology (Agilent Technologies, Santa Clara, CA, USA) operating in negative mode. Analyte separation was performed using a Waters BEH C18 column (2.1 mm × 150 mm, 2.5 μm particle size) with mobile phase consisting of (A) 0.1% acetic acid in MilliQ water and (B) acetonitrile : isopropanol (90:10) and 10 μL injection volumes for each run. The gradient used was the following: 0.0–3.5 min 10–35% B, 3.5–5.5 min 40% B, 5.5–7.0 min 42%B, 7.0–9.0 min 50% B, 9.0–15.0 min 65% B, 15.0–17.0 min 75% B, 17.0–18.5 min 85% B, 18.5–19.5 min 95% B, 19.5–21 min 95–10% B and 21.0–25.0 min 10% B.

MS‐ESI parameters employed were capillary and nozzle voltage at 4000 and 1500 V, drying gas temperature 230°C with a gas flow of 15 L·min⁻¹, sheet gas temperature 400°C with a gas flow of 11 L·min⁻¹, the nebulizer gas flow was 35 psi, and iFunnel high and low pressure RF at 90 and 60 V. The dynamic multiple reaction monitoring option was used for all compounds with optimized transitions and collision energies. Manual peak integration was undertaken using MassHunter Workstation software. Quantification was based on the internal standard additions. For further details, see Gouveia‐Figueira and Nording, 2015.

Data and statistical analyses

The data and statistical analysis comply with the recommendations on experimental design and analysis in pharmacology (Curtis et al., 2015). Throughout the text, N refers to the number of separate experiments conducted with 2–6 replicates. For the oxylipin data, n refers to the number of separate samples extracted and analysed from different wells prepared on the same experimental day. Fisher's randomization (permutation) tests using 10⁶ iterations were conducted using the functions aovp and permTS in the lmperm (version 2.1.0) and perm (version 1.0–0.0) packages, respectively, for the R statistical programme versions 3.3.1–3.4.1 (R Core Team, 2017). Levene's test was undertaken using the function leveneTest in the car package (version 2.1‐5) for R. One‐ and two‐way ANOVA, Sidak's multiple comparisons test, paired t‐tests and robust linear regressions were undertaken using the statistical algorithms built in to the GraphPad Prism computer programme (GraphPad Software Inc., San Diego, CA, USA) versions 7a and 7b for the Macintosh. Three‐way ANOVA were undertaken using the function ezANOVA in the package ez (version 4.4‐0) for R. Residual plots were also constructed using R.

Statistical significance was set at P < 0.05. Note, however, that in the lipidomic and qPCR experiments summarized in the tables, the use of a simple P < 0.05 does not allow for identification of potential false positives due to multiple testing. In these cases, we have used a 5% false discovery rate (Benjamini and Hochberg, 1995) to determine the critical values of P and indicated in bold text the P values which are lower than this value and hence significant. In theory, adjusted P values can be calculated (for example with the p.adjust function in R), but since readers may have their own preferences as to the best way to deal with multiple testing in exploratory data, giving the unadjusted P values allows the reader to make her/his own calculations.

With respect to randomization and blinding, cells were randomly seeded in the plates and inserts. When TEER was measured, inserts that were above 900 Ω·cm² were randomly transferred to a new plate. No blinding was done, but all the measurements are quantifiable and not subjective.

Materials

Cell culture reagents [medium (DMEM/F12, GIBCO cat. no. 11330‐057), trypsin, FBS, penicillin + streptomycin (PEST) and L‐glutamine], recombinant human IFNγ and Dynabeads® mRNA direct kit were purchased from Life technologies (Thermo Fisher Scientific, Waltham, MA, USA). Transwell permeable supports cat. no. 3401 were purchased from Corning Inc. (Corning, NY, USA). Anti‐ZO‐1 (ZO1‐1A12) and anti‐occludin (OC‐3F10) antibodies conjugated with Alexa Fluor® 488 were obtained from Thermo Fisher Scientific. The high capacity cDNA reverse transcription kit was purchased from Applied Biosystems, Thermo Fisher Scientific. KAPA SYBR® FAST qPCR master mix was purchased from KAPA Biosystems, (Wilmington, MA, USA). 2,4,3′,5′‐tetramethoxystilbene (TMS), benzo[a]pyrene, 2‐phenyl‐4H‐naphthopyran‐4‐one (α‐naphthoflavone), ammonium pyrrolidinedithiocarbamate (PDTC) and 3‐(5′‐hydroxymethyl‐2′‐furyl)‐1‐benzyl indazole (YC‐1) were purchased from Sigma Aldrich (St. Louis, MO, USA).

Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Southan et al., 2016), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander et al., 2017a,b).

Results

CYP1A1 and CYP1B1 mRNA expression in T84 cells

T84 cells have long been used as a model of intestinal epithelial cells and have been shown to form tight monolayers and acquire the features of mature polarized columnar cells when cultured in transwells for 8 to 14 days (Adams et al., 1993; Ou et al., 2009a). This can be visualized by immunofluorescence of tight junction proteins such as ZO‐1 and occludin (Figure 1A, B) and by measurement of TEER, which was around 1100 Ω·cm² after 9–10 days in culture (Figure 2A). We compared mRNA expression of CYP1A1 and CYP1B1 in T84 cells plated on the same day from the same passage and cultured either in 24‐well plates for 24 h or allowed to form a tight monolayer in transwells. We found that when the cells formed a tight monolayer, mRNA expression of CYP1A1 was increased approximately threefold while that of CYP1B1 was decreased by 65% (Figure 1C, D).

Figure 1.

Immunostaining of T84 cell monolayers for (A) occludin and (B) ZO‐1 showing the formation of a continuous monolayer of cells. (C, D) mRNA levels of CYP1A1 and CYP1B1 cultured in 24‐well plates (‘24w’) and as monolayers on Transwell inserts. The individual ΔCt values (the difference in threshold cycle between the mRNA of the CYP and the housekeeper mRNA [RPL19]) are shown with the means represented by the bars. The right Y‐axes show data where the mean for the 24‐well plates is set to 100%.

Figure 2.

Effect of IFNγ treatment upon the properties of T84 cells cultured in monolayers on Transwells. IFNγ (10 ng·mL⁻¹) was added basolaterally at t0 and the cells were cultured for 24 h. Panel A, TEER values for the controls (N = 31) and IFNγ‐treated (N = 30) cells are pooled from several experimental series. In all the scatterplots presented in this paper, the solid bars show the mean values. A two‐way type 3 ANOVA with time as repeated measure gave F _1,59 values for time, IFNγ and the interaction time × IFNγ of 140, 10.0 and 235, respectively (all statistically significant); *P < 0.05; significantly different as indicated; Sidak's multiple comparisons test. Residual plots (of a corresponding linear mixed model analysis) indicated no obvious heteroscedasticity or non‐normality of the distribution. Panel B, mRNA levels for CYP1A1 (N = 10), CYP1B1 (N = 10), IL8 (N = 8) and TNFα (N = 8) following 24 h of treatment with either vehicle or IFNγ. *P < 0.05; significantly different as indicated; two‐tailed t‐tests.

Effect of IFNγ treatment on TEER and CYP expression in T84 cells

T lymphocytes located basolateral to the tight junctions, both within the epithelium and in the lamina propria, constitute the cellular source of IFNγ in human gut (Forsberg et al., 2002; Melgar et al., 2003). In order to mimic the in vivo situation, IFNγ was added to the basolateral side of polarized tight monolayers of T84 cells. Incubation of T84 cells with 10 ng·mL⁻¹ of IFNγ for 24 h led to a 23% decrease in TEER (Figure 2A). Mean (SD) TEER for the monolayers were 1138 (109) Ω·cm² and 883 (139) Ω·cm² before and after treatment with IFNγ for 24 h respectively (N = 30). This was accompanied by an increased expression of mRNA for the inflammatory cytokines IL‐8 (approximately twofold) and TNF‐α (approximately fourfold) (Figure 2B). A slight, but significant, increase in TEER was seen for the control samples during the 24 h incubation (Figure 2A). In theory, this could be due to incomplete differentiation when the experiments were started, despite the high initial TEER values that appeared to have stabilized in the days preceding the experiments. However, the increase was very small: expressing the 24 h value (t24) as % of the t0 value gave mean values (95% confidence limits and N in brackets) of 103 (101–105, N = 31) %. We regard this very small increase as being statistically, but not biologically, significant. The corresponding values for the IFNγ‐treated cells was 77 (74–80, N = 30).

Involvement of NF‐κB and HIF1α in the effects of IFNγ upon CYP1A1 and CYB1B1 expression by T84 cells

Yang et al. (2014) have reported that inhibitors of either NF‐κB activation (PDTC) or HIF1α (YC‐1) significantly increased the TEER of IFNγ‐treated T84 cells (Yang et al., 2014), raising the possibility that these mediators contribute to the effects of IFNγ upon CYP1A1 and CYP1B1 expression. Here, for IFNγ‐treated cells, the NF‐kB inhibitor PDTC reduced the expression of both CYP1A1 and CYP1B1 mRNA (Figure 3A, B). The HIF1α inhibitor had differential effects on the expression of these two CYPs as it reduced CYP1A1 mRNA expression but increased CYP1B1 mRNA expression (Figure 3A, B).

Figure 3.

Effect of PDTC (100 μM) and YC‐1 (50 μM) upon mRNA expression of (A) CYP1A1 and (B) CYP1B1 for IFNγ‐treated T84 cells (N = 6). Cells were seeded overnight in 24‐well plates and incubated with vehicle or the compounds for 1 h, before incubation with IFNγ (10 ng·mL⁻¹) for 8 h. *P < 0.05; significantly different from control (concomitantly cultured cells in the absence of IFNγ); two‐tailed paired t‐test. For the IFNγ‐treated cells, Levene's test for homogeneity of variance was significant for CYP1B1 and so, we analysed the data using a two‐way permutation test (function aovp in the package lmperm for R). For CYP1A1, a significant P value was found for the main effect of YC‐1. For CYP1B1, significant P values were found for YC‐1 and the interaction PDTC × YC‐1.

Effects of IFNγ upon oxylipin content

The CYP isoforms can not only metabolize arachidonic acid to HETEs but can also metabolize other long chain fatty acids, such as linoleic acid and docosahexaenoic acid (Konkel and Schunck, 2011). Oxylipins such as 5‐, 8‐, 12‐ and 15‐HETE, but not 19‐ or 20‐HETE, are produced by Caco‐2 cells (Le Faouder et al., 2013) and can affect their barrier properties (Rodriguez‐Lagunas et al., 2013; Pochard et al., 2016). CYP1B1 predominantly catalyses the formation of mid‐chain HETEs (11‐ and 12‐HETE) from arachidonic acid, whereas CYP1A1 can catalyse the production of both mid‐ and long‐chain HETEs and particularly 20‐HETE (El‐Sherbeni and El‐Kadi, 2016). This raises the possibility that IFNγ reduces TEER secondary to changes in oxylipin levels in the T84 cells. In consequence, we investigated the levels of oxylipins in T84 cells following treatment for either 8 or 24 h with IFNγ. As HETE synthesis can be stimulated by calcium in some cell types (Hoffman et al., 1988; Edwards and Ritter, 1994), we investigated the effects of IFNγ on oxylipin levels with and without a 20 min stimulation with the calcium ionophore ionomycin. The data are summarized in Table 2. In total, 13 oxylipins could reliably be detected and quantified. Upon three‐way permutation ANOVA with time, cytokine and ionomycin as factors, only one P value was above the critical value of P = 0.0005 upon implementation of a 5% false discovery rate (Benjamini and Hochberg, 1995), and this was for the main effect of incubation time upon 5(S),6(R)‐LXA₄ levels (Table 2). This would argue against robust changes in oxylipin contents being involved in the actions of IFNγ in T84 cells.

Table 2.

Effect of IFNγ treatment of T84 cells upon their oxylipin content (fmol per well)

		Vehicle		Ionomycin
Time	Treatment	median	IQR	median	IQR
Linoleic acid derivatives
9‐HODE
8 h	Control	297	97.0	257	271
8 h	IFNγ	340	238	309	441
24 h	Control	268o	30.7	291	275
24 h	IFNγ	288	465	300o	42.0
13‐HODE
8 h	Control	643	237	640	712
8 h	IFNγ	1081	695	896	464
24 h	Control	705	99.2	690	413
24 h	IFNγ	591	935	628	363
9,10‐diHOME
8 h	Control	42.4	16.8	39.3	10.4
8 h	IFNγ	32.4	7.6	33.6o	12.5
24 h	Control	30.6	9.8	37.6	13.1
24 h	IFNγ	31.0	53.3	33.8	19.1
12,13‐DiHOME
8 h	Control	68.0	18.3	60.7	23.9
8 h	IFNγ	45.1	14.8	46.9o	21.6
24 h	Control	54.6	24.9	48.5	16.3
24 h	IFNγ	46.2o	31.3	63.7	34.6
9,10,13‐TriHOME
8 h	Control	133	105	99.2	15.0
8 h	IFNγ	122	51.8	105o	72.9
24 h	Control	85.3	16.7	73.1	35.6
24 h	IFNγ	82.5	87.8	84.0	39.5
9,12,13‐triHOME
8 h	Control	434	393	375	153
8 h	IFNγ	428	214	384	508
24 h	Control	379	118	267	149
24 h	IFNγ	417	578	373	174
Arachidonic acid derivatives
TXB2
8 h	Control	153	41.0	166	42.2
8 h	IFNγ	188	60.6	154o	81.8
24 h	Control	129	51.6	148	73.0
24 h	IFNγ	147	38.7	156	32.3
5‐HETE
8 h	Control	21.8	13.1	24.3	12.7
8 h	IFNγ	23.4	12.8	22.5o	24.2
24 h	Control	21.5	28.0	24.6	23.2
24 h	IFNγ	31.4	21.0	25.0	20.8
11‐HETE
8 h	Control	6.0	6.4	7.7	4.2
8 h	IFNγ	9.7	5.7	8.8o	8.7
24 h	Control	8.0	3.9	10.8	3.3
24 h	IFNγ	12.1	4.0	12.2	6.4
12‐HETE
8 h	Control	106	18.7	96.3	45.5
8 h	IFNγ	105	22.5	104o	30.8
24 h	Control	84.7	23.4	80.5	21.3
24 h	IFNγ	85.8	41.3	69.0	51.6
15‐HETE
8 h	Control	28.1	18.3	28.6	11.9
8 h	IFNγ	35.1	32.9	32.2o	15.0
24 h	Control	24.3	16.9	33.0	13.5
24 h	IFNγ	30.2	13.0	30.4	32.9
Other fatty acid derivatives
5(S),6(R)‐LXA4
8 h	Control	341	84.3	350	98.1
8 h	IFNγ	277	132	300	346
24 h	Control	421	125	414	110
24 h	IFNγ	479	105	456	84.8
12‐HEPE
8 h	Control	27.9	5.7	24.5o	4.7
8 h	IFNγ	28.6	14.4	21.1o	16.9
24 h	Control	28.1	18.3	28.9	18.6
24 h	IFNγ	23.2	14.6	27.4	17.1

Cells in 6‐well plates were cultured for either 8 or 24 h with vehicle or IFNγ (10 ng·mL⁻¹), after which the oxylipins were extracted. Data are summarized as medians and interquartile ranges (IQR). Thirteen (of a total of 936) lipid quantifications were categorized as outliers (>3 × IQR higher or, in one case lower, than the 75 or 25% quartile, as appropriate, for the group in question) and excluded, this is shown as ^o in the Table. In the case of 11‐HETE, for example, the outlier had a value of 1005 fmol per well. Values using the Tukey definition (>1.5 ×IQR) but which were lower than >3 × IQR were not excluded. The data are reported as medians and IQR, n = 8–9, since 20/104 of the groups did not pass the D'Agostino and Pearson test for normality. P values were determined by three‐way permutation tests (function aovp in the package lmperm for R) with cytokine (cy), ionomycin (io) and time (t) as independent variables. At a 5% false discovery rate (Benjamini and Hochberg, 1995), the critical value of P is for significance is P = 0.00055, and only the main effect of time for 5(S)6(R)‐LXA₄ was significant using this criterion. Given that some readers may regard other ways of correcting for multiple comparisons as being more appropriate, the unadjusted P values are given in Supplementary Table S2 to allow this to be done.

CYP1B1 inhibition by TMS partly reverses IFNγ‐induced alterations of epithelial permeability

TMS is a selective and competitive inhibitor of CYP1B1, showing 50‐fold selectivity over CYP1A1. Its reported IC₅₀ values are 6 and 300 nM for human CYP1B1 and CYP1A1, respectively (Kim et al., 2002; Bruno and Njar, 2007), and some studies have also demonstrated that the compound can reduce CYP1B1 expression (e.g. Chun et al., 2005). Consistent with these potencies, our initial data indicated that at a concentration of 1 μM, effects upon CYP1A1 mRNA expression were seen in addition to those upon CYP1B1 expression, whereas a concentration of 100 nM did not affect CYP1A1 mRNA expression (Supporting Information Figure S1). In consequence, we used 100 nM to achieve selective inhibition of CYP1B1 activity.

The effects of 100 nM TMS upon the TEER for vehicle‐ and IFNγ‐treated T84 cells were investigated in a large series of samples. Analysing the raw data with a three‐way ANOVA with time (0 and 24 h) as the repeated measure, we found a significant interaction time × cytokine × TMS (Supporting Information Table S1). Three‐way ANOVAs are somewhat cumbersome to interpret, and so, we have presented TEER values at 24 h expressed as a percentage of the corresponding values for the same conditions at t = 0. A significant interaction IFNγ‐treatment × TMS was found, due to the ability of the compound to reverse partly the effect of IFNγ (Figure 4A). Calculating the data from each experiment as % reversal of the IFNγ‐induced TEER deficit gave a mean reversal of 47% (95% confidence interval 33–61%) for the 19 experiments where TEER values for control, IFNγ‐treated and IFNγ + TMS treated cells were all measured (Figure 4B). Supporting the results on TEER, TMS partly reversed the increased permeability to [¹⁴C]mannitol produced by IFNγ (Supporting Information Figure S2).

Figure 4.

Panel A, effects of vehicle (V, N = 21 and 20 for control and IFNγ‐treated respectively), and TMS (T, 100 nM; N = 13 and 19 for control and IFNγ‐treated respectively) treatment upon TEER levels in T84 cells cultured in Transwells. Vehicle or IFNγ (10 ng·mL⁻¹) was added basolaterally and vehicle or TMS apically prior to culture for 24 h. In all cases the values shown are the TEER values at 24 h expressed as a percentage of the corresponding values for the same conditions at t = 0. As the sample sizes are different, we calculated ANOVA P values using a permutation test (function aovp in the package lmPerm for R). The P values for IFNγ, TMS and IFNγ × TMS were all <0.05. Pairwise P values (after controlling for α inflation using a Bonferroni correction) were calculated using two‐sided permutation tests using the function permTS in the package perm for R and are shown in the Figure: *P < 0.05; significantly different as indicated; NS, not significantly different. Panel B, % reversal of the effect of IFNγ upon TEER by TMS (N = 19), benzo[a]pyrene (Benzo, 10 μM, N = 6)) and α‐naphthoflavone (α‐Nap, 100 nM, N = 6). For each experiment, % reversal was calculated as [100 – (100 × (ΔTEER_cpd‐con − ΔTEER_con)/(ΔTEER_ifn − ΔTEER_con)], where ‘ΔTEER’ refers to the change (in Ω·cm² seen between t0 and t24 for the different conditions; ‘cpd’ refers to the combination of test compound and IFNγ treatment, ‘con’ to the untreated cells and ‘ifn’ to cells treated with IFNγ and vehicle. A one‐way permutation ANOVA (independence_test in the package coin for R) gave a P value <0.05. One sample t‐tests of each condition versus 0% gave a significant P value for TMS, but not for benzo[a]pyrene or α‐naphthoflavone.

Effects of benzo[a]pyrene and α‐naphthoflavone upon IFNγ‐induced alterations of epithelial permeability

Two other compounds were investigated, α‐naphthoflavone and benzo[a]pyrene. α‐Naphthoflavone is not only an AhR antagonist (Wilhelmsson et al., 1994), leading to a decrease of CYP1A1 and CYP1B1 expression, but also an inhibitor of CYP1A1 and CYP1B1 with reported IC₅₀ values of 60 and 5 nM respectively (Shimada et al., 1998; Bruno and Njar, 2007). In T84 cells, α‐naphthoflavone decreased both CYP1A1 and CYP1B1 mRNA expression at 100 nM but had the opposite effect at 10 μM (Supporting Information Figure S1), consistent with reports that this compound can act as an AhR agonist at higher concentrations (Wilhelmsson et al., 1994). At a concentration of 100 nM, α‐naphthoflavone produced no consistent reversal of the effect of IFNγ on the TEER (Figure 4B). Benzo[a]pyrene, a polycyclic aromatic hydrocarbon, is a substrate as well as an inducer of both CYP1A1 and CYP1B1 via activation of AhR (Nebert et al., 2004; see Supporting Information Figure S1 for our data with T84 cells). No significant reversal by 10 μM benzo[a]pyrene of the reduced TEER produced by IFNγ was seen (Figure 4B).

Mechanistic studies into the mode of action of TMS in T84 cells

In order to shed light into potential mechanisms of action of TMS, we performed two series of experiments. Firstly, TMS is an analogue of resveratrol, and resveratrol can block the activation of NF‐κB, admittedly at higher concentrations than 100 nM (Donnelly et al., 2004; Gonzales and Orlando, 2008; Busch et al., 2012). This raises the possibility that the action of TMS simply reflects an off‐target action of the compound upon NF‐κB, given that this protein is a mediator of IFNγ (Bruewer et al., 2005; Boivin et al., 2009; Yang et al., 2014). If this was the case, then TMS should affect the induction of mRNA for IL8 and TNFα produced by the IFNγ treatment, as these are downstream of NF‐κB (Bonizzi and Karin, 2004). This was not seen as IFNγ‐induced expression of IL8 and TNFα was similar in the presence or absence of 100 nM TMS (Figure 5).

Figure 5.

Effects of vehicle (v) and TMS (T, 100 nM) upon the mRNA for A, IL8 and B, TNFα (N = 6, except for the TMS‐treated control cells, where N = 5) in control and IFNγ‐treated cells. The T84 cells in Transwells were treated as described in Figure 4. For the IFNγ‐treated cells, the mean values for vehicle‐ and TMS‐treated cells were not significantly different.

Effects of IFNγ and TMS upon the mRNA expression of HIF1α, AhR and proteins involved in the barrier properties of T84 cells

A second hypothesis that was tested was that TMS affected the expression either of HIF1α, of AhR, or of other proteins involved in maintaining the barrier properties of T84 cells such as the IFNγ‐sensitive proteins ZO‐1 and claudin‐2 (Youakim and Ahdieh, 1999; Willemsen et al., 2005) and thereby negated the effects of the inflammatory stimulus upon these proteins. We investigated this at the mRNA level using qPCR. The proteins involved in barrier function that were chosen were the multi‐drug resistance gene multidrug resistance protein 1 (MDR1) (involved in drug transport) and the tight junction proteins occludin (OCCLN), ZO‐1 (TJPT1) and claudin (CLDN) proteins 2, 4 and 5. In cultured T84 cells, significant effects of IFNγ upon HIF1α, MDR1, CLDN‐2 and ‐4 mRNA expression were seen (Table 3), whereas no significant main effects of TMS or interaction IFNγ × TMS were seen, arguing against this hypothesis. CLDN‐5 levels were very low in the cells. The pattern of change of CLDN‐2 (decreased expression) and −4 (small increase in expression) produced by IFNγ per se is in line with the immunochemical study of Prasad et al. (2005) who reported that 3 days of treatment of T84 cell monolayers with IFNγ/TNFα reduced the expression of claudin‐2 but redistributed the expression of claudin‐4 from junctional areas to the cytoplasm rather than affecting its expression level. Similarly, the lack of effect upon occludin is consistent with the immunoblot study of Watson et al. (2005), although these authors found a reduced expression of phosphorylated occludin.

Table 3.

Effects of IFNγ and TMS upon the mRNA expression of HIF1, AhR and proteins involved in the barrier function of T84 cells

Gene	Cytokine treatment	Vehicle		TMS		Significant
		mean	SD	mean	SD	P values
HIF1	Control	4.63	0.39	4.74	0.39	Cyt
	IFNγ	4.11	0.50	4.28	0.33
		[144]		[137]
AhR	Control	7.17	0.54	7.31	0.30
	IFNγ	7.16	0.43	7.20	0.44
		[101]		[108]
MDR1	Control	7.16	0.49	7.52	0.50	Cyt
	IFNγ	6.25	0.70	6.46	0.69
		[189]		[208]
Occln	Control	5.12	0.38	5.40	0.36
	IFNγ	4.96	0.31	5.01	0.35
		[112]		[131]
TJP1	Control	5.99	0.66	6.16	0.38
(ZO‐1)	IFNγ	5.22	0.82	5.40	0.78
		[171]		[169]
CLDN2	Control	3.77	0.68	3.91	0.61	Cyt
	IFNγ	5.13	1.02	5.22	1.01
		[39]		[40]
CLDN4	Control	1.63	0.47	1.96	0.52	Cyt
	IFNγ	0.86	0.74	0.92	0.63
		[171]		[206]
CLDN5	Control	14.5	0.61	15.5	1.04
	IFNγ	15.5	0.90	15.0	0.79
		[52]		[140]

The T84 cells in Transwells were treated as described in Figure 4. Shown are means and SD of the ΔCt values, N = 6, except for the vehicle – TMS group where N = 5. P values were determined by two‐way permutation tests (function aovp in the package lmperm for R). At a 5% false discovery rate (Benjamini and Hochberg, 1995), the critical value of P is 0.0083. Thus, P < 0.0083 is considered to be statistically significant and is indicated in bold in the right column for the factor in question (Cyt = cytokine). Given that some readers may regard other ways of correcting for multiple comparisons as being more appropriate, the unadjusted P values are given in Supplementary Table S3 to allow this to be done. Note that a decrease in ΔCt of one unit between control and IFNγ‐treated (i.e. ΔΔCt = 1) corresponds to a doubling of mRNA expression. Numbers in square brackets indicate the mean IFNγ treated value as % of the corresponding mean control value using 2^−ΔΔCt.

Discussion

Apart from the studies in cancer research, not much is known about the role of CYP1B1 on gut homeostasis. More extensive studies have been done in that context on the role of the AhR in inflammatory bowel diseases, where CYP1A1, and to a lesser extent CYP1B1, are only reported as an endpoint to assess AhR activation.

IFNγ has been shown to decrease CYP mRNA expression and activity, including CYP1A1 (Tapner et al., 1996; Calleja et al., 1997; Aitken and Morgan, 2007), but to our knowledge, CYP1B1 has not been investigated in this regard. We found that CYP1B1 mRNA expression is up‐regulated by IFNγ in T84 cells. The NF‐κB pathway has been involved in the effect of IFNγ in T84 cells (Bruewer et al., 2005; Boivin et al., 2009; Yang et al., 2014). This pathway leads to increased levels of downstream inflammatory cytokines such as TNFα (Bonizzi and Karin, 2004) and in other cell systems, inflammatory cytokines such as IL6 and TNFα have been shown to induce CYP1B1 and down‐regulate CYP1A1 (Piscaglia et al., 1999; Bleau et al., 2003; Umannová et al., 2008). Given that the CYP1B1 increase produced by IFNγ was reduced by the NF‐κB inhibitor PDTC in our study, it is reasonable to postulate a causal chain of events IFNγ → NF‐κB activation → production of TNFα and other inflammatory cytokines → induction of CYP1B1 in the T84 cells.

A more complex question concerns whether the chain of events described above causes the reduced permeability of T84 cells produced by IFNγ. In theory, the chain could diverge at some point after NF‐κB activation (Boivin et al., 2009) so that the induction of CYP1B1 and the impairment of barrier function are separate phenomena. However, inhibition of CYP1B1 with TMS partly counteracted the effects of IFNγ on epithelial barrier permeability without affecting IL8 and TNFα mRNA levels, suggesting some form of causality between induction of CYP1B1 and impairment of barrier function. The concentration of TMS used (100 nM) is relatively low and does not produce a complete inhibition of CYP1B1 in experimental systems but selectivity versus CYP1A1 is lost at higher concentrations (Kim et al., 2002). There is thus a trade‐off between selectivity and attainable blockade.

While it is clear that TMS can produce partial reversal of the effects of IFNγ on epithelial barrier permeability, we were not able to elucidate the mechanism(s) involved in this effect, although we could demonstrate that it was not due to a resveratrol‐like off‐target effect on NF‐κB. In T84 cells, IFNγ induces HIF1α, with an approximate doubling in mRNA levels being found after 24 h. Protein expression was also increased, in a manner that was prevented by blockade of NF‐κB activation (Yang et al., 2014). Moreover, the HIF1α inhibitor YC‐1 increases the TEER of IFNγ‐treated T84 cells (Yang et al., 2014). We could confirm the IFNγ‐induced increase in HIF1α in the present study and found additional increases in MDR1 and CLDN4 together with a decrease in CLDN5 levels. HIF1α is also known to interfere with CYP1A1 and CYP1B1 expression. Indeed, both HIF1α and AhR form functional dimers with the aryl hydrocarbon receptor nuclear translocator (ARNT). Therefore, induction of the HIF1α pathway will affect AhR signalling (and thus CYP1A1 and CYP1B1 induction) by competition for ARNT (Schults et al., 2010). Indeed, in human hCMEC/D3 cerebral microvascular endothelial cells, hypoxia results in the down‐regulation of both CYP1A1 and CYP1B1 mRNA in a manner blocked by gene silencing of HIF1α and HIF2α (Jacob et al., 2015). In monkey CV‐1 kidney cells, a feedback loop whereby CYP1A and CYP1B families regulate AhR activity has been reported. Thus, AhR activity was decreased upon transient transfection of the cells with CYP1B1 and increased upon treatment with TMS (100 nM giving a partial response) (Chiaro et al., 2007). We reasoned that the high levels of CYP1B1 produced as a result of the IFNγ → → inflammatory cytokine pathway could reduce the activity of the AhR pathway and thereby favour HIF1α/ARNT‐mediated events, including effects on permeability. Blockade of CYP1B1 by TMS, on the other hand, would block this feedback loop, thereby permitting competition between AhR and HIF1α for ARNT and hence reducing HIF1α‐mediated effects on permeability. Our data, however, do not support this suggestion because (a) the HIF1α inhibitor YC‐1 did not reverse the induction of CYP1B1 by IFNγ; (b) AhR expression was not affected by IFNγ; and (c) neither the induction of HIF1α nor that of MDR1 (which can be regulated by HIF‐1α in T84 cells (Comerford et al., 2002)), by IFNγ was blocked by TMS. Thus, further experiments are required in order to elucidate the mechanism(s) by which TMS can reduce the barrier deficit produced by IFNγ. Such experiments, for example, could focus upon whether the effect of TMS is proximal to, or distal to, the effects of IFNγ upon NF‐κB, and whether the beneficial effect of TMS is due to an ability to prevent the redistribution of barrier proteins such as claudin‐4 that is produced by IFNγ.

In conclusion, we have presented evidence that CYP1B1 inhibition counteracts cytokine‐induced alterations in epithelial barrier integrity. CYP1B1 has been attracting increased attention as a drug target, not least with respect to the treatment of cancer (Bruno and Njar, 2007). Our data, if translated beyond cultured cells, would suggest that new pathways could be explored for the treatment of diseases such as inflammatory bowel diseases, inflammatory bowel syndrome or celiac disease that share, as part of their pathogenesis, a disrupted epithelial barrier. Our findings could also help understand the pathogenesis of some of these diseases where, apart from genetic predisposition, environmental factors play a role in disease progression. The fact that CYP1B1 is expressed in the colon, and its expression can be modulated by dietary or environmental components, could be one component of these environmental factors.

Author contributions

M.A. and C.J.F conceived and designed the experiments and analysed the data. M.A. and S.G.‐F. performed the experiments. M.‐L.H. contributed materials and participated in the design of the experiments. M.A. and C.J.F. wrote the paper.

Conflict of interest

The authors declare no conflicts of interest.

Declaration of transparency and scientific rigour

This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research recommended by funding agencies, publishers and other organisations engaged with supporting research.

Supporting information

Table S1 Three‐way ANOVA output for the effect of TMS upon the TEER for vehicle‐ and IFNγ‐treated T84 cells.

Table S2 Unadjusted P values for the effect of IFNγ treatment of T84 cells upon their oxylipin content (data summarised in Table 2).

Table S3 Unadjusted P values for the effect of of IFNγ and TMS upon the mRNA expression of HIF1, AhR and proteins involved in the barrier function of T84 cells (data summarised in Table 3).

Figure S1 Effects of TMS (T), benzo[a]pyrene (Benzo, B) and α‐naphthoflavone (α‐Nap, αN) upon CYP1A1 and CYP1B1 mRNA levels in A,B: 24 well plates (means and range (when not enclosed by the symbols), N = 3–4); C,D: Transwells.

Figure S2 Effects of vehicle (Veh) and TMS (100 nM) treatment upon the permeability of T84 cells to [¹⁴C]mannitol (0.125 μCi added apically).

Alhouayek, M. , Gouveia‐Figueira, S. , Hammarström, M.‐L. , and Fowler, C. J. (2018) Involvement of CYP1B1 in interferon γ‐induced alterations of epithelial barrier integrity. British Journal of Pharmacology, 175: 877–890. doi: 10.1111/bph.14122.