Crosstalk between the aryl hydrocarbon receptor and hypoxia-inducible factor 1α pathways in human islet models

ABSTRACT

Background

We previously showed that 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD – a persistent organic pollutant) activates the aryl hydrocarbon receptor (AHR) in pancreatic islets. The AHR is known to crosstalk with hypoxia-inducible factor 1α (HIF1α) in other cell types but AHR-HIF1α crosstalk has not been previously examined in islet cells. Islet cell function is sensitive to hypoxia; we hypothesize that AHR activation by environmental pollutant(s) will interfere with the HIF1α pathway response in islets, which may be detrimental to islet cell function and survival during periods of hypoxia.

Methods

We assessed AHR-HIF1α crosstalk by treating human donor islets and stem cell-derived islets (SC-islets) with 10 nM TCDD ± 1% O₂ and measuring gene expression of downstream targets of AHR (e.g. CYP1A1) and HIF1α (e.g. HMOX1).

Results

In SC-islets, co-treatment with TCDD + hypoxia consistently suppressed CYP1A1 induction compared with TCDD treatment alone. In human islets, TCDD + hypoxia co-treatment suppressed CYP1A1 induction, but only in 2 of 6 donors. Both SC-islets and human donor islets displayed hypoxia-mediated suppression of glucose-6-phosphate catalytic subunit 2 (G6PC2) expression. Glucose-stimulated insulin secretion (GSIS) in human donor islets was impaired by hypoxia exposure, but unaffected by TCDD exposure.

Conclusion

Our study shows consistent AHR-HIF1α crosstalk in SC-islets and variable crosstalk in primary human islets, depending on the donor. In both cell models, hypoxia exposure interfered with activation of the AHR pathway by TCDD but there was no evidence that AHR activation interfered with the HIF1α pathway. In summary, our data show that co-exposure to an environmental pollutant and hypoxia results in molecular crosstalk in islets.

GRAPHICAL ABSTRACT

1. Introduction

Diabetes prevalence is increasing rapidly on a global scale. In 2021, 537 million adults worldwide were estimated to be living with diabetes and the incidence rate is projected to increase by 46% by 2045¹. Type 2 diabetes (T2D) typically develops through myriad factors causing β-cell death or dysfunction, alongside peripheral insulin resistance^2–4. Traditional risk factors for T2D include genetics, diet, and lifestyle. However, emerging research implicates exposure to man-made persistent organic pollutants (POPs) as an additional risk factor for T2D incidence⁵.

POPs are globally distributed, resulting in ubiquitous human exposure⁶. Dioxins and dioxin-like compounds are a class of POPs characterized by a molecular structure of chlorinated benzene rings and a high affinity for the aryl hydrocarbon receptor (AHR)^7,8. Although serum concentrations of dioxins are decreasing globally⁹, epidemiology studies have reported positive associations between dioxin exposure and T2D incidence in diverse populations with lipid-adjusted serum dioxin concentrations ranging from 0.3 to 3388 ppt^5,10–14. The most toxic and widely studied dioxin is 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). Although the mechanism linking TCDD exposure and T2D incidence remains unclear, our laboratory has shown that TCDD exposure to mice in vivo and to mouse or human islets ex vivo upregulates cytochrome P450 1A1 (Cyp1a1/CYP1A1) expression in islet cells^15–17, an indication of AHR activation. The effects of TCDD on glucose tolerance, plasma insulin levels, and insulin sensitivity were abolished in β-cell specific Ahr knockout mice, suggesting that the AHR pathway in β-cells is mediating the adverse metabolic effects of TCDD¹⁸.

The AHR is a class I bHLH/PAS protein associated with cellular and xenobiotic metabolism^19–24. Cytosolic AHR is inactive, but when bound by a ligand AHR translocates to the nucleus where the ligand-AHR complex dimerizes to the class II bHLH/PAS protein, aryl hydrocarbon nucleus translocator (ARNT). The AHR-ARNT complex recognizes and binds to xenobiotic response element (XRE) promoter sequences^25–27, leading to upregulation of CYP1A1 enzymes, which are required for the oxidation and subsequent excretion of xenobiotics^23,28. ARNT is a promiscuous Class II protein that shows diverse protein–protein interactions^29–31, including dimerization with the Class I bHLH/PAS protein hypoxia-inducible factor 1α (HIF1α)^30–32. During periods of insufficient oxygen availability (i.e., hypoxia, < 5% O₂) stable HIF1α protein translocates into the nucleus and heterodimerizes with ARNT. The HIF1α-ARNT transcription complex is recognized by hypoxia response element (HRE) promotor sequences, leading to induction of genes associated with glucose metabolism, cell survival, vascular tone, and vascular growth^27,33–35.

Competitive binding of AHR and HIF1α to ARNT has been reported in epithelial lung cells, lymphocytes, breast cancer cells, adipocytes, and hepatocytes^{27,32,36–38}. In hepatocytes and hepatoma cells exposure to hypoxia or hypoxia-mimetics consistently interferes with expression of AHR gene targets, but exposure to AHR ligands does not consistently interfere with induction of HIF1α gene targets^{27,36,39–41}. For example, mouse hepatoma cells simultaneously exposed to hypoxia and an AHR ligand showed a 30% reduction in the AHR target Cyp1a1, but no change in the HIF1α target Vegf compared with cells exposed to normoxia and the AHR ligand⁴⁰. Crosstalk between AHR and HIF1α pathways has not been reported in pancreatic islets. We hypothesized that AHR-HIF1α crosstalk likely occurs in human islets, similar to hepatocytes, where HIF1α pathway activation interferes with the AHR pathway; we also speculate that the reverse may occur in islets—i.e., an AHR ligand interfering with HIF1α activation.

There are numerous scenarios where pancreatic islets are exposed to periods of hypoxic stress and need to mount a HIF1α response. Pancreatic islets normally experience oxygen partial pressure (pO₂) of ~ 31–37 mmHg (4–5% O₂) in vivo⁴² and ~25 mmHg (3% O₂) at the islet surface when cultured in vitro⁴³. Tightly regulated low physiological pO₂ is required for pancreatic development and has been reported in subpopulations of mature pancreatic islets^34,43. However, sustained glucotoxic and/or lipotoxic stress results in pathological hypoxia ( < 2% O₂) that shifts cellular metabolism from aerobic to anaerobic and severely impairs glucose stimulated insulin secretion (GSIS)⁴⁴. In islet transplantation, a robust HIF1α-driven response is critical for graft revascularization and the success of transplantation^45,46. It is important to understand how environmental factors, such as ubiquitous exposure to environmental AHR ligands⁴⁷, may influence the ability of islets to mount an appropriate hypoxia response. Additionally, given that AhRactivation in β-cells appears to mediate the adverse metabolic effects of TCDD in mice¹⁸, it is also important to understand if hypoxic stress could interfere with adaptation of islets to AHR ligands.

Human embryonic stem cell (hESC)-derived islet-like endocrine cells (SC-islets) are a developmental research model with potential for clinical diabetes treatment⁴⁸. In the present study, we use a genetically modified hESC line that expresses EGFP under control of the INSULIN promotor (INS-2A-EGFP cell line⁴⁹), to generate human SC-islets in vitro. Although differentiation protocols are continually improving^50–52, SC-islets are not fully mature and do not show robust GSIS comparable with primary islets^53–55. An advantage of SC-islets compared to primary islets from human organ donors is that SC-islets are environmentally naïve and thus have limited biological variability between differentiation batches. In contrast, human donor islets are sourced from a diverse population with a broad range of ages, genetic backgrounds, health, lifestyles, and previous environmental exposures. While donor variability is a challenge of studying human islets, this also serves as an important opportunity to examine the range of responses encountered in human islets.

In the current study, we examine AHR-HIF1α pathway crosstalk in human islet cells by exposing primary donor islets and SC-islets to hypoxia alone, TCDD alone, or hypoxia + TCDD combined for 48 hours. We measured gene targets of AHR (e.g., CYP1A1) as an indicator of AHR activation and gene targets of HIF1α (e.g., HMOX1) as an indication of HIF1α activation⁵⁶. GSIS was also measured in primary islets to determine the effect of AHR-HIF1α crosstalk on human islet function.

2. Methods

2.1. hESC differentiation and maintenance

SC-islets were derived from INS-2A-EGFP WA01 hESCs by Dr. Francis Lynn (University of British Columbia, BC, Canada) using a 27+ day differentiation protocol, as previously described^52,57. INS-2A-EGFP hESCs were authenticated and tested for contamination in the Lynn lab. SC-islets were then shipped from Vancouver to Ottawa and maintained in Stage 6 media [CMRL-1066 (VWR, #CA45001–114; Radnor, PA) containing 20 g/L bovine serum albumin (Sigma-Aldrich #10775835001), 1X Glutamax (Gibco #35050079), 1% penicillin-streptomycin (Gibco #15140122), 0.5 mM pyruvate (Sigma-Aldrich #S8636–100 mL), 0.5X Insulin Transferrin-Selenium-X (Thermo Fisher #51500056), 35 nM zinc sulfate (Sigma-Aldrich #Z0251-100 G), 1 mM N-acetyl-L-cysteine (Sigma-Aldrich #A9165-5 G), 10 ug/L Heparin (Sigma-Aldrich #H3149-10KU), 1:2000 Trace elements A (Corning 25–021-CI), 1:2000 Trace Elements B (Corning 25–022-CI #MT99175CI), 10 nM T3 (Sigma, #T6397-100 MG), 0.5 μM ZM 447,439 (Cedarlane Labs, #S1103-10 MM/1 ML; Burlington, ON), and 1:2000 lipid concentrate (Fisher Scientific #11905031)]. SC-islets were maintained at 37°C, 21% O₂, and 5% CO₂ on an orbital shaker (106 RPM) until analyses, with media changes every two days.

SC-islets are composed of a mixture of pancreatic endocrine cells, including ~ 30–70% INS-EGFP⁺ cells (Supplemental Figure S1). SC-islets aged to day 50–60 are capable of showing a modest GSIS response⁵². Therefore, we used SC-islets aged day 50+ in our experiments. SC-islets were visually assessed for cluster integrity and GFP⁺ fluorescence upon arrival and were not included in this study if quality appeared poor. Each SC-islet differentiation was considered one “biological replicate” (n = 3 technical replicates per condition per differentiation; n = 3 biological replicates).

2.2. Human organ donor islet cell culture

Human islets were provided by the Alberta Diabetes Institute IsletCore at the University of Alberta (www.bcell.org/adi-isletcore) with the assistance of the Human Organ Procurement and Exchange (HOPE) program, Trillium Gift of Life Network (TGLN), and other Canadian organ procurement organizations. Islet isolation was approved by the Human Research Ethics Board at the University of Alberta (Pro00013094). All donors’ families gave informed consent for the use of pancreatic tissue in research. All research using human islets was approved by the Research Ethics Board at Carleton University.

Islet purity ranged between 75% and 90% for this study (Figure 4H). Donor islets were shipped overnight to Ottawa in CMRL media (Fisher Scientific, #11–530–037). Upon arrival, islets were visually assessed for islet integrity and transferred to low glucose DMEM media (LG-DMEM; Fisher Scientific, #11–885–084) supplemented with 10% FBS (Sigma-Aldrich #F1051-500 ML) and 1% penicillin-streptomycin (Gibco, #15140122; Billings, MT) and cultured at 37°C, 21% O₂, and 5% CO₂ overnight to allow for recovery. See Figure 4H for donor characteristics. Human islets from separate donors were considered as biological replicates (n = 3–5 technical replicates per condition per donor; n = 3 biological replicates per sex).

Figure 4.

Co-treatment of human islets to TCDD + hypoxia impairs induction of CYP1A1 by TCDD in 2 of 6 donors. SC-islets and human islets were exposed to TCDD (10 nM) ± hypoxia (1% O₂) for 48 hours. Gene expression was measured in SC-islets and human islets for (A,C) CYP1A1 and (B,D) HMOX1, and in human islets for (E) AHRR, (F) ARNT, and (G) G6PC2. Gene expression is relative to DMSO-treated islets. (A,B) Separate SC-islet biological replicates (i.e., different differentiations) and (C-G) separate human islet donors are shown for each gene. Data are mean ± SEM and individual data points represent technical replicates. Significance within each biological replicate was determined by a two-way ANOVA and Dunnett post-doc test. Overall treatment and interaction effects are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001). Multiple comparison analysis showing treatment groups that are significantly different from each other are indicated by different letters (p < 0.05). (H) Human islet characteristics are summarized for each individual donor.

2.3. Experimental conditions

Two experimental paradigms were used in this study (Figure 1). First, we treated SC-islets with TCDD ± hypoxia continuously for 48 hours or pre-treated SC-islets with either TCDD alone or hypoxia alone for 24 hours prior to co-treatment with TCDD + hypoxia for 24 hours (Figure 1A). Lastly, we compared the effects of TCDD ± hypoxia co-treatment for 48 hours in SC-islets versus human donor islets (Figure 1B). A concentration of 10 nM TCDD (3,219 ppt) was selected following previous findings that 10 nM TCDD was sufficient to upregulate CYP1A1 mRNA and enzyme activity in human donor islets¹⁶. Hypoxia exposure was set at 1% O₂ (7.6 mmHg O₂) to mimic physiological hypoxia shown to activate the HIF1α pathway⁵⁸.

Figure 1.

Summary of treatment conditions applied to SC-islets and human donor islets. (A) SC-islets were exposed to TCDD (10 nM) ± hypoxia (1% O₂) for 48 hours continuously or to TCDD or hypoxia pre-treatment for 24 hours followed by TCDD + hypoxia co-treatment for 24 hours. (B) SC-islets or human donor islets were exposed to TCDD ± hypoxia co-treatment continuously for 48 hours. Made with BioRender.com.

SC-islets or human donor islets were hand-picked and transferred to six-well non-TC treated plates (VWR, #10861–554) to undergo the treatment condition outlined above: (i) DMSO (#276855-100 ML, Sigma-Aldrich, Oakville, ON, Canada) vehicle control; (ii) normoxia (21% O₂, 5% CO₂); (iii) 10 nM TCDD (#AD404SDMSO10X1ML, Accustandard, Chromatographic Specialties, Brockville, ON, Canada); (iv) hypoxia (1% O₂) in a triple-gas (N₂-O₂-CO₂) incubator (ThermoForma, Series II, 3130, Marietta, Ohio).

2.4. Quantitative real time PCR

Immediately following treatment, SC-islets or human islets were transferred to buffer RLT (#79216, Qiagen, Hilden, Germany) and stored at −80°C. mRNA was isolated using RNeasy Micro kit (#74004, Qiagen, Hilden, Germany) as per the manufacturer’s instructions. DNase treatment was performed prior to cDNA synthesis with the iScript^TM gDNA Clear cDNA Synthesis Kit (#1725035, Bio-Rad, Mississauga, ON, Canada). Real time (RT) qPCR was performed using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, #1725271) and run on a CFX394 (Bio-Rad). PPIA was used as the reference gene. Primer efficiency ranged between 87% and 110%, and sequences are listed in Table S1. Data were analyzed using the 2ΔΔCT relative quantitation method and normalized to the DMSO vehicle control.

2.5. Glucose-stimulated insulin secretion

SC-islets displayed limited glucose responsiveness in the Bruin lab, despite being modestly glucose responsive in other labs⁵². Therefore, only human islets were used to measure the impact of TCDD ± hypoxia on GSIS (n = 3–5 technical replicates per donor; n = 6 biological replicates, i.e., donors). Immediately following treatment, 20 islets per condition were moved to pre-warmed (37°C) Krebs – Ringer bicarbonate buffer (KRBB) solution (115 mmol/l NaCl, 5 mmol/l KCl, 24 mmol/l NaHCO₃, 2.5 mmol/L CaCl₂, 1 mmol/l MgCl₂, 10 mmol/l HEPES, 0.1% (wt/vol.) BSA, pH 7.4) with 2.8 mmol/L glucose (low glucose, LG) for a 60 min pre-incubation in 1.5 mL Eppendorf tubes. Supernatant was removed and replaced with 500 μL of LG KRBB for 60 min, followed by transfer to 500 μL of KRBB with 16.7 mmol/L glucose (high glucose, HG) for 60 min at 37°C. The LG KRBB and HG KRBB samples were centrifuged, and the supernatant stored at − 20°C. The GSIS assay was conducted under normoxic conditions and in the absence of TCDD, regardless of previous treatments. To measure insulin content, supernatant was removed and replaced with an acid-ethanol solution of 1.5% vol/vol HCl in 75% vol/vol ethanol at 4°C overnight and neutralized with 1 mol/L Tris base (pH 7.5) before long-term storage at − 20°C. Low and high glucose GSIS samples were diluted 1:10 in KRBB and acid-ethanol lysed samples were diluted 1:100 in KRBB prior to measurement of human insulin concentration by ELISA (#80-INSHU-CH10, ALPCO, Salem, NH, USA).

2.6. Statistics

All statistics were performed using GraphPad Prism 10.0.0 (GraphPad Software Inc., La Jolla, CA). For all analyses, p < 0.05 was considered statistically significant and any potential outliers were tested for significance using a Grubbs’ test with α = 0.05. We tested for variance between biological replicates within model using a nested one-way ANOVA. We tested for main and interaction effects using an ordinary two-way ANOVA with a Tukey’s multiple comparisons test. Comparisons between different treatment groups are visualized on graphs by showing different letters above each bar to indicate significant differences (i.e., “a” is different than “b” and “ab” is not different from either “a” or “b”). Non-parametric statistics were used in cases where the data failed normality or equal variance tests. A Spearman (rho) correlation test was used to assess correlation between donor islet quality and clinical parameters (age, BMI, HbA1c, cold ischemia time, and islet particle index) against degree of CYP1A1 induction by TCDD exposure relative to CYP1A1 induction by TCDD + hypoxia exposure. Data are presented as means ± SEM. Figure legends specify whether individual datapoints are technical or biological replicates.

3. Results

3.1. Co-exposure of SC-islets to TCDD + hypoxia impairs CYP1A1 induction by TCDD, regardless of pre-treatment conditions

To determine if hypoxia interferes with TCDD-mediated AHR activation in islets, we measured CYP1A1 and HMOX1 induction in SC-islets following exposure to TCDD ± hypoxia (Figure 1A). TCDD significantly upregulated CYP1A1 expression in SC-islets ~41-fold compared to the vehicle (Figures 2A and 3A). However, when SC-islets were co-treated with TCDD + hypoxia, the magnitude of CYP1A1 induction was suppressed to ~ 21-fold (Figures 2A and 3A). Hypoxia exposure induced HMOX1 ~24-fold in SC-islets compared to normoxia with no effect of co-treatment with TCDD (Figures 2B and 3B).

Figure 2.

Hypoxia interfered with CYP1A1 upregulation by TCDD in SC-islets. (A) CYP1A1 and (B) HMOX1 gene expression in SC-islets co-treated with TCDD (10 nM) ± hypoxia (1% O₂) for 48 hours or following either TCDD or hypoxia pre-treatment for 24 hours and TCDD + hypoxia co-treatment for 24 hours. Gene expression is relative to DMSO-treated SC-islets. Data are mean ± SEM and individual data points represent biological replicates (i.e. different differentiations). Significance was determined by a one-way ANOVA and Tukey post-hoc test. Multiple comparison analysis showing treatment groups that are significantly different from each other are indicated by different letters (p < 0.05).

Figure 3.

TCDD + hypoxia co-treatment causes interactive effect for CYP1A1 and G6PC2 expression in SC-islets. SC-islets were exposed to TCDD (10 nM) ± hypoxia (1% O₂) for 48 hours and gene expression was measured for (A) CYP1A1, (B) AHR, (C) AHRR, (D) HMOX1, (E) VEGFA, (F) ARNT, (G) MNSOD, (H) SLC2A1, (I) MAFA, and (J) G6PC2. Gene expression is relative to DMSO-treated SC-islets. Data are mean ± SEM and individual data points represent biological replicates (i.e. different differentiations). Significance was determined by a two-way ANOVA and Tukey post-hoc test. Overall treatment and interaction effects are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001). Multiple comparison analysis showing treatment groups that are significantly different from each other are indicated by different letters (p < 0.05).

We also investigated whether the degree of crosstalk between AHR and HIF1α in SC-islets is influenced by which pathway is activated first—i.e., a first-come, first-serve effect. We hypothesized that activation of either the AHR or HIF1α pathway before competitive co-exposure would dictate the dimerization partner of ARNT. SC-islets were pre-treated with either TCDD or hypoxia for 24 hours prior to co-treatment with TCDD + hypoxia for 24 hours (Figure 1A). We found that all TCDD + hypoxia co-treatment conditions significantly impaired the degree of CYP1A1 induction compared to treatment with TCDD alone (Figure 2A). Pre-treatment of SC-islets with either TCDD or hypoxia did not alter this effect (Figure 2A). Similarly, all hypoxia treatment conditions upregulated HMOX1 in SC-islets, irrespective of co-treatment or pre-treatment with TCDD (Figure 2B).

3.2. Co-exposure of SC-islets to TCDD + hypoxia also causes an interaction effect on G6PC2

To further explore the impact of molecular crosstalk in SC-islets, we focused on the first 4 conditions: vehicle, TCDD alone, hypoxia alone, or TCDD + hypoxia for 48-hours (Figure 1B). We measured CYP1A1 expression alongside other AHR targets (AHR, AHRR), HIF1α targets (HMOX1, VEGFA), ARNT, biomarkers of β-cell maturity (MAFA – a transcription factor required for β-cell maturity and function⁵⁹, SLC2A1 – a glucose transporter required for glucose-stimulated insulin secretion⁶⁰, and G6PC2 – glucose 6 phosphatase catalytic subunit 2, an islet-specific enzyme involved in the futile glucose flux with glucokinase and positively associated with fasting blood glucose levels in rodents and humans^61–65, and MNSOD – an antioxidant that protects against reactive oxygen species produced by mitochondrial oxidative stress^66,67(Figure 3). As shown in Figure 2A and re-plotted here, there was a significant interaction effect of TCDD + hypoxia co-treatment on CYP1A1 expression in SC-islets (Figure 3A). The AHR repressor gene, AHRR, was significantly upregulated by TCDD exposure, as expected, but unlike CYP1A1 there was no interaction effect with hypoxia (Figure 3C). Exposure of SC-islets to hypoxia significantly upregulated AHR, HMOX1, VEGFA, MNSOD, and SLC2A1 (Figure 3B,D,E,G,H) and downregulated MAFA expression (Figure 3I); none of these changes were affected by co-treatment with TCDD. Both TCDD and hypoxia exposure alone modestly upregulated ARNT expression, but there was no interaction effect on ARNT (Figure 3F). Interestingly, an unexpected interaction effect was seen for G6PC2 expression (Figure 3J). Both TCDD alone and hypoxia alone significantly downregulated G6PC2 expression compared with vehicle/normoxia control conditions (Figure 3J). Hypoxia-mediated downregulation dominated the interaction effect seen following TCDD + hypoxia co-treatment (Figure 3J).

3.3. TCDD consistently induces CYP1A1 in human donor islets but AHR-HIF1α crosstalk varies between donors

We next examined whether crosstalk between AHR and HIF1α pathways also occurs in primary human islets from deceased organ donors. Human islets from 3 male and 3 female donors were included in our study. Average donor age was 49.3 (±8.0) years, BMI was 27.2 (±5.3), and HbA1c was 5.2 (±0.6) (Figure 4H). Human donor islets were co-exposed to TCDD ± hypoxia continuously for 48 hours (Figure 1B). Unlike SC-islets, which showed consistent treatment effects across all biological replicates (i.e., different differentiations; Figure 4A,B), the effects of TCDD ± hypoxia on human islets varied substantially between donors. These variable effects are summarized in Table 1, detailed statistics are provided in Supplementary Table S2, and additional donor information can be found at humanislets.com⁶⁸.

Table 1.

SC-islets and human donor islets treated with TCDD ± hypoxia show variability for gene expression changes.

	CYP1A1			HMOX1			AHRR			ARNT			G6PC2
Cell source	TCDD	Hypoxia	Crosstalk	TCDD	Hypoxia	Crosstalk	TCDD	Hypoxia	Crosstalk	TCDD	Hypoxia	Crosstalk	TCDD	Hypoxia	Crosstalk
SC-Islets	↑ (41-fold)	↑ (1.1-fold)	✓	-	↑ (24-fold)	x	↑ (6-fold)	-	x	↑ (1.2-fold)	↑ (1.2-fold)	x	↓	↓	✓
R425 - ♂	↑ (10-fold)	↓	✓	-	↑ (2-fold)	✓	↑ (1.3-fold)	-	✓	-	-	x	-	↓	x
R434 -♂	↑ (3-fold)	-	x	↑ (1.4-fold)	↑ (5-fold)	x	↑ (2-fold)	-	x	↑ (1.7-fold)	↑ (1.7-fold)	x	↑ (2-fold)	↓	x
R444 -♂	↑ (11-fold)	-	x	↑ (2.5-fold)	↑ (2.6-fold)	x	↑ (3-fold)	-	x	↑ (2-fold)	-	x	↑ (2-fold)	-	x
R427 -♀	↑ (3-fold)	↓	x	-	↑ (3-fold)	x	-	-	x	-	-	x	-	↓	x
R430 -♀	↑ (3-fold)	-	✓	-	↑ (3-fold)	x	↑ (1.1-fold)	-	✓	-	-	✓	-	↓	x
R461 -♀	↑ (15-fold)	↑ (2-fold)	✓	-	↑ (3-fold)	x	↑ (2-fold)	-	x	-	-	x	-	↓	x

A summary table of the main effects of TCDD (10 nM) ± hypoxia (1% O₂) on SC-islets and human donor islets for expression of CYP1A1, HMOX1, AHRR, ARNT, and G6PC2. Bolded fold-increase indicates significance for TCDD or hypoxia treatments alone, compared with vehicle control.

TCDD exposure significantly increased CYP1A1 expression in all donors, albeit to varying degrees (2.5-fold to 15.8-fold, Figure 4C, Table 1). We saw a significant interaction effect for TCDD + hypoxia co-treatment suppressing the magnitude of CYP1A1 induction compared with TCDD alone in 2 of the 6 donors (Donors R425 and R461, Figure 4C, Table 1). There was also a trending interaction effect in Donor R430 (p = 0.0552, Figure 4C, Table 1). No interaction effect for CYP1A1 expression was found in the other donor islets co-exposed to TCDD + hypoxia (Figure 4C, Table 1).

We next correlated various islet and donor parameters (age, BMI, HbA1c, cold ischemia time, and islet particle index) against CYP1A1 fold induction change by co-exposure (i.e., CYP1A1 fold induction following TCDD relative to CYP1A1 fold induction following TCDD + hypoxia). Interestingly, cold ischemia time positively correlated with the CYP1A1 fold induction change by co-exposure (rho = 0.89, p = 0.03; Supplemental Table S3). This preliminary analysis suggests that islets with longer cold ischemia time had a greater degree of crosstalk between AHR and HIF1α pathways.

3.4. TCDD consistently upregulates AHRR expression in human islets; hypoxia consistently induces HMOX1 and downregulates G6PC2 expression in human islets

All donors showed a main effect or trend for modest HMOX1 upregulation by hypoxia (2.2-fold to 5.0-fold, Figure 4D, Table 1). A modest interaction effect by TCDD + hypoxia co-exposure augmenting HMOX1 upregulation was seen for one donor (R425, Figure 4D, Table 1).

All donors except one (Donor R427) showed a main effect for TCDD exposure upregulating AHRR expression (Figure 4E, Table 1); this is consistent with the effect seen in SC-islets (Figure 3C, Table 1). In addition, two donors (R425 and R430) showed a modest TCDD + hypoxia interaction effect for AHRR (Figure 4E, Table 1).

Changes in ARNT expression were highly variable between donors. Three donors showed no change in ARNT expression across treatments (Donors R425, R427, R461; Figure 4F, Table 1). Donor R430 showed a modest interaction effect, Donor R434 showed a main effect for both hypoxia and TCDD, and Donor R444 showed a main effect for TCDD exposure only (Figure 4F, Table 1).

All human donors except R444 showed a significant main effect for hypoxia to downregulate G6PC2 expression (Figure 4G, Table 1). Donors R434 and R444 additionally showed a main effect for G6PC2 upregulation by TCDD (Figure 4G, Table 1). TCDD-driven upregulation of G6PC2 in donor islets contrasts the modest but consistent TCDD-driven G6PC2 downregulation seen in the SC-islets (Figure 3J, Table 1). No interaction effects were found for G6PC2 expression among the donor islets, which also differs from the significant interaction effect seen in SC-islets.

3.5. Hypoxia, but not TCDD, negatively impacts donor islet GSIS

To determine whether TCDD ± hypoxia exposure impacted islet function we assessed GSIS immediately following 48-hour exposure conditions. Across all donors, TCDD did not impact basal or glucose-stimulated insulin secretion compared with vehicle control (Figure 5A,B). Exposure to hypoxia dramatically reduced insulin secretion in response to high glucose (Figure 5A,B). This effect was maintained but not amplified when cells were co-treated with hypoxia + TCDD (Figure 5A,B). Hypoxia treatment resulted in mild reduction of total insulin content in human islets, regardless of co-treatment with TCDD (Figure 5C). When insulin secretion was normalized to insulin content, there was no significant effect of treatment on GSIS (Figure 5D).

Figure 5.

Hypoxia exposure negatively impacts glucose stimulated insulin secretion. Human islets were exposed to TCDD (10 nM) ± hypoxia (1% O₂) for 48 hours and (A) insulin secretion was measured following 1 hour in low (2.8 mM) and high (16.7 mM) glucose concentrations. (B) Stimulation index is expressed as secreted insulin concentration following high glucose relative to low glucose stimulation. (C) Total insulin content in islets following acid ethanol lysis. (D) Insulin secretion following low and high glucose stimulation normalized by total insulin content. Data are mean ± SEM and individual data points represent biological replicates (i.e. different islet donors). Significance was determined with (A, D) a two-way ANOVA and Šidák post-hoc test, (B) a brown-forsythe ANOVA and Dunnett post-hoc test, and (C) a one-way ANOVA and Dunnett post-hoc test. Multiple comparison analysis showing treatment groups that are significantly different from each other are indicated by different letters (*p < 0.05).

4. Discussion

This study investigated whether there is crosstalk between AHR and HIF1α pathways in SC-islets and human donor islets. TCDD exposure consistently activated the AHR pathway, as indicated by CYP1A1 upregulation, in SC-islets and all human islet donors. Interestingly, the degree of CYP1A1 induction was more pronounced in SC-islets than human islets and varied substantially between human islet donors. Hypoxia exposure consistently interfered with TCDD-mediated CYP1A1 induction in the SC-islet model, which suggests possible crosstalk between AHR and HIF1α pathways. We also saw evidence of AHR-HIF1α crosstalk in human islets but only in 2 of 6 donors.

There are many variables that may contribute to the consistency of AHR-HIF1α crosstalk in SC-islets compared to human donor islets. First, we speculate that previous environmental exposures likely influence the sensitivity of donor islets to TCDD and/or hypoxia exposure ex vivo. SC-islets are environmentally naive whereas human islets have been exposed to a broad variety of environmental contaminants throughout their lifetime^69,70, which impact baseline AHR pathway activation between donors. Furthermore, dietary habits, smoking or drug use, and other environmental factors can lead to cellular hypoxia^71–73, which likely impacts baseline HIF1α pathway activation. Donor age is another important source of biological variation; the human islets used in our study were isolated from donors between 36 to 61 years of age. AHR activation and downstream effects are known to vary with age^74,75 and baseline AHR and ARNT levels vary by age and tissue state⁷⁶. Our sample size of human donors was insufficient to do a well-powered correlation analysis to determine which clinical parameters are driving variation in AHR-HIF1α crosstalk, but we did note a positive correlation between cold ischemia time and suppression of CYP1A1 induction by co-exposure to TCDD + hypoxia. Duration of cold ischemia time does not impact donor islet retrieval or function and does not impair transplantation in appropriately selected recipients^77–79. However, these data suggest an extended period of cold ischemia may sensitize donor islets to AHR-HIF1α crosstalk ex vivo. In summary, the differences in robustness and variability of AHR-HIF1α crosstalk between SC-islets and human donor islets likely arises from a mixture of these considerations and reflect the need for caution when interpreting cell-line responses as a proxy for primary human cell biology. The variability in responses between different human donors emphasizes the need for large sample sizes in future work examining AHR-HIF1α crosstalk in human donor islets. A larger sample size is critical to ensure reproducibility and allow for appropriate stratification based on donor characteristics, such as biological sex.

One surprising finding from this study was the lack of impaired GSIS in human islets after 48-hour exposure to TCDD ex vivo. In our previous publication, we reported modestly impaired insulin secretion under high glucose conditions in TCDD-exposed islets from 5 different human donors, although this did not reach statistical significance¹⁶. Here, using 6 additional donors, the effects of TCDD on static GSIS are highly variable between donors and the overall effect does not approach statistical significance under either low or high glucose conditions. More detailed perifusion analysis would provide additional insight into the dynamics of insulin secretion (e.g., first-phase versus second-phase responses) from human islets following ex vivo TCDD exposure. Furthermore, markers of mitochondrial function, endoplasmic reticulum stress, and/or cell viability can be examined in future studies for a clearer understanding of the impact of TCDD on human islets. We also speculate that TCDD may have more pronounced effects on insulin secretion following longer-term exposure and/or exposure alongside a secondary metabolic stressor, based on findings from the epidemiology literature⁵ and animal studies^15,17,18,,80. In summary, despite the consistency in activation of the AHR (i.e., CYP1A1 and AHRR induction) across all human islet donors tested to date, there remains a wide spectrum of functional responses to TCDD; more human islet donors are needed to fully appreciate the nuances of how TCDD impacts human islet function.

In our study, the HIF1α pathway dominated over the AHR pathway in islets, even when cells were pre-treated with TCDD for 24 hours. HIF1α pathway dominance was indicated by the lack of TCDD-mediated interference with hypoxia-mediated HMOX1 induction in both SC-islets and human islets. However, our study is limited by the absence of protein measurements. Future studies should examine protein expression, particularly stability of the AHR and HIF1α, in human islet models to strengthen our understanding of AHR-HIF1α crosstalk. The dominance of hypoxia-mediated effects was also clear in the GSIS response of human islets. Hypoxia exposure profoundly blunted the GSIS response, as expected based on previous findings^81–86, and this effect was seen irrespective of co-treatment with TCDD. We speculate that ARNT binding affinity is likely driving AHR-HIF1α crosstalk in human islets. Notably, there is little to no evidence of ARNT-independent gene induction by HIF1α^87–91. Gradin et al. used coimmunoprecipitation to show stabilized HIF1α has a high binding affinity with ARNT in a human hepatoma cell line. Furthermore, this group found the HIF1α pathway dominates over AHR in binding with ARNT in a rabbit reticulocyte lysate⁴¹. The exact mechanism(s) driving the HIF1α-ARNT binding dominance over AHR-ARNT in SC-islets and human donor islets warrants further investigation.

The presence of AHR-HIF1α crosstalk in SC-islets and human donor islets has interesting implications for islet transplantation. Human islets are transiently hypoxic following clinical transplantation^44,46 and need to mount an efficient and robust HIF1α response to induce vascularization. Given that the AHR and HIF1α pathways both rely on access to ARNT, it is possible that sustained exposure to an AHR ligand(s) could interfere with the HIF1α-mediated hypoxia response post-transplant of either SC-islets or human donor islets. While our experiment models showed the HIF1α pathway dominating AHR-HIF1α crosstalk in islets, this was a short-term (48-hr) window of crosstalk in vitro. It remains possible that chronic exposure of patients to AHR ligands could compromise engraftment post-transplant via interfering with the HIF1α pathway. Additionally, we do not fully understand the consequences of AHR pathway activation within islets, so it is difficult to appreciate the potential impact that post-transplant hypoxia may have on islet health if HIF1α activation interferes with the AHR-mediated detoxification response in engrafted islets.

The consistent hypoxia-mediated downregulation of G6PC2 in both SC-islets and human donor islets was an unexpected and interesting finding. The consistent changes in HMOX1, CYP1A1, and AHRR expression were expected based on well-established effects of hypoxia^34,44 and TCDD³¹, in islets and other cell types. Hypoxia exposure upregulates G6pc expression in mouse liver tissue and in a human liver cell line (HepG2)^92,93, but to our knowledge the connection between hypoxia exposure and G6PC2 downregulation in human islets has not been previously described. In cancer patients and cancer cell lines G6PC was described as part of the battery of gene changes associated with the Warburg effect (aerobic glycolysis)^94,95, and hypoxia-exposed rats showed G6pc downregulation in liver tissue⁹⁶. In humans the G6PC2 isoform is islet-specific⁶¹, associated with autoimmunity⁹⁷, and responsible for tight regulation of fasting blood glucose levels^62,98–100. G6PC2 expression is downregulated in human islets from donors diagnosed with T2D, a condition associated with islet hypoxia^85,101–103. We speculate that beyond regulating blood glucose levels in diabetes, downregulation of G6PC2 is protective by increasing intracellular glucose levels available for anaerobic glycolysis and ATP production necessary for insulin secretion. In support, Rahim et al. (2022) showed G6pc2 knockout improved GSIS response by modulating glycolysis in a mouse pancreatic β-cell line¹⁰⁴. Hypoxia-mediated G6PC2 downregulation may work in concert with other hypoxia-mediated gene expression changes in the islet to promote glycolysis and suppress gluconeogenesis^44,105.

5. Conclusions

In conclusion, our study is the first to show evidence for AHR-HIF1α crosstalk in islets. This crosstalk was dominated by the HIF1α pathway and was consistently observed in SC-islets. However, in human islets there was substantial biological variability in both the extent that TCDD exposure upregulated CYP1A1 expression and the presence of AHR-HIF1α crosstalk. We believe further study is warranted to examine why HIF1α dominates AHR-HIF1α crosstalk in islets, and what factors contribute to the variability seen in human donor islets. Furthermore, we believe the previously unreported finding of hypoxia-mediated G6PC2 downregulation in SC-islets and human donor islets requires further investigation.

Funding Statement

This research was supported by a Canadian Institutes of Health Research-Breakthrough T1D (CIHR-Breakthrough T1D) Team Grant (#ASD-173663/5-SRA-2020-1059-S-B) and Natural Sciences and Engineering Research Council (NSERC) Discovery Grants to JEB (#RGPIN-2017-06265) and WGW (#RGPIN-2017-06414). NG was supported by an NSERC Canadian Graduate Student-Doctoral (CGS-D) award and KAV was supported by an NSERC-Collaborative Research and Training Experience (CREATE) award provided by the Canadian Islet Research and Training Network (CIRTN). JEB is supported by an Early Researcher Award from the Ontario Government and a Dorothy Killam Fellowship.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Abbreviations

AHR

Aryl hydrocarbon receptor

AHRR

Aryl hydrocarbon receptor repressor

ARNT

Aryl hydrocarbon nuclear translocator

bHLH

Basic helix loop helix

CYP1A1

Cytochrome P450 1A1

GSIS

Glucose stimulated insulin secretion

G6PC2

Glucose-6-phosphate catalytic subunit 2

HIF1α

Hypoxia inducible factor 1α

HRE

Hypoxia response element

MAFA

V-maf musculoaponeurotic fibrosarcoma oncogene homolog A

MNSOD

Manganese superoxide dismutase

PAS

Period/ARNT/Sim

POP

Persistent organic pollutant

SC-islets

Stem cell-derived islets

T2D

Type 2 diabetes

TCDD

2,3,7,8 tetrachlorodibenzo-p-dioxin

XRE

Xenobiotic response element

Author contributions

NG: Conceptualization, Methodology, Formal analysis, Investigation, Writing – Original draft, Writing – Review & editing, Visualization. JEB: Conceptualization, Methodology, Formal analysis, Resources, Writing – Review & editing, Visualization, Supervision, Funding acquisition. WGW: Conceptualization, Methodology, Resources, Writing – Review & editing. KVA: Investigation, Writing – Review & editing. FCL: Investigation, Resources, Writing – Review & editing.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19382014.2025.2526871