Genotoxicity assessment of eight nitrosamines using 2D and 3D HepaRG cell models

Abstract

N-nitrosamine impurities have been increasingly detected in human drugs. This is a safety concern as many nitrosamines are mutagenic in bacteria and carcinogenic in rodent models. Typically, the mutagenic and carcinogenic activity of nitrosamines requires metabolic activation by cytochromes P450 enzymes (CYPs), which in many in vitro models are supplied exogenously using rodent liver homogenates. There are only limited data on the genotoxicity of nitrosamines in human cell systems. In this study, we used metabolically competent human HepaRG cells, whose metabolic capability is comparable to that of primary human hepatocytes, to evaluate the genotoxicity of eight nitrosamines [N-cyclopentyl-4-nitrosopiperazine (CPNP), N-nitrosodibutylamine (NDBA), N-nitrosodiethylamine (NDEA), N-nitrosodimethylamine (NDMA), N-nitrosodiisopropylamine (NDIPA), N-nitrosoethylisopropylamine (NEIPA), N-nitroso-N-methyl-4-aminobutyric acid (NMBA), and N-nitrosomethylphenylamine (NMPA)]. Under the conditions we used to culture HepaRG cells, three-dimensional (3D) spheroids possessed higher levels of CYP activity compared to 2D monolayer cells; thus the genotoxicity of the eight nitrosamines was investigated using 3D HepaRG spheroids in addition to more conventional 2D cultures. Genotoxicity was assessed as DNA damage using the high-throughput CometChip assay and as aneugenicity/clastogenicity in the flow-cytometry-based micronucleus (MN) assay. Following a 24-h treatment, all the nitrosamines induced DNA damage in 3D spheroids, while only three nitrosamines, NDBA, NDEA, and NDMA, produced positive responses in 2D HepaRG cells. In addition, these three nitrosamines also caused significant increases in MN frequency in both 2D and 3D HepaRG models, while NMBA and NMPA were positive only in the 3D HepaRG MN assay. Overall, our results indicate that HepaRG spheroids may provide a sensitive, human-based cell system for evaluating the genotoxicity of nitrosamines.

Introduction

Nitrosamines, or N-nitrosamines, were first described in the nineteenth century (Gushgari and Halden 2018). Nitrosamines, characterized by a nitroso group bound to an amine moiety (Fig. 1), are notorious both for their nearly ubiquitous presence in the environment and because many nitrosamines are carcinogens; some are potent carcinogens (Beard and Swager 2021). Humans are exposed to low levels of nitrosamines via tobacco, water, food, beverages, personal care products, and household products (Horne et al. 2023). Since 2018, nitrosamines have drawn increased attention due to the identification of unacceptable levels of nitrosamines as impurities in several common medicines used for the treatment of hypertension, heartburn, and diabetes. The identification of these impurities has resulted in a new regulatory guidance in addition to drug recalls or withdrawals of over 1800 product lots from the market (Bharate 2021; FDA 2021; Schmidtsdorff et al. 2022; WHO 2019).

Fig. 1.

Structures of eight nitrosamine impurities tested in the study. MW, molecular weight

The U.S. Food and Drug Administration (FDA) has identified seven nitrosamine impurities (Fig. 1) that may be present in drug products: N-nitrosodibutylamine (NDBA), N-nitrosodiethylamine (NDEA), N-nitrosodimethylamine (NDMA), N-nitrosodiisopropylamine (NDIPA), N-nitrosoethylisoporpylamine (NEIPA), N-nitroso-N-methyl-4-aminobutanoic acid (NMBA), and N-nitrosomethylphenylamine (NMPA) (FDA 2021). In addition, another impurity, N-cyclopentyl-4-nitroso piperazine (CPNP, Fig. 1) has been identified recently in rifapentine, an antibacterial drug used to treat tuberculosis (Kao et al. 2022; WHO 2022).

NDMA and NDEA are two well-studied nitrosamines, with both substances classified as Group 2A probable human carcinogens by the International Agency for Research on Cancer (IARC) (IARC 1978). NDMA and NDEA have induced tumors in 16 and 26 different animal species, respectively, including frog, rodent, rabbit, and monkey (Bogovski and Bogovski 1981). NDBA was a more potent mutagen than NDMA or NDEA in the bacterial Ames test (Brambilla et al. 1981) and also induced tumors in multiple tissues in rodents exposed via various routes, thus being classified by IARC as a Group 2B possible human carcinogen (IARC 1978). NMBA was negative in the standard Ames test and mutagenic only when it was oxidized with 5,10,15,20-tetrakis(1- methylpyridinium-4-yl)porphyrinatoiron(III) pentachloride (4-MPy) and tert-butyl hydroperoxide (t-BuOOH) in acetonitrile followed by dichloromethane extraction (Inami et al. 2013). The mutagenicity of NMPA has been challenging to detect in standard Ames tests; however, positive responses have been reported for several tester strains, especially those detecting mutations at A:T base pairs by bulky adducts (Gold et al. 1987; Zielenska and Guttenplan 1988). Genotoxicity and carcinogenicity data on the other three nitrosamines are sparse.

Since nitrosamines belong to a group of highly potent mutagenic carcinogens, they are identified as “cohort of concern” impurities by the International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use (ICH) M7(R2) on Assessment and Control of DNA Reactive (Mutagenic) Impurities in Pharmaceuticals to Limit Potential Carcinogenic Risk (ICH 2023). Approximately 82% of nitrosamines in the expanded Vitic Database (Lhasa Limited), which includes rodent carcinogenicity data on 228 nitrosamines, are considered carcinogenic (Thresher et al. 2020). In addition, analysis of this database confirmed the strong correlation between the Ames and rodent carcinogenicity results. While these observations are consistent with the presumption of a human carcinogenic hazard accorded nitrosamines by their classification in the cohort of concern, the data included in the Vitic Database were generated from bacteria (the Ames test) and rodent animal models which have limitations as predictive models.

Overall, questions remain as to how the Ames test or rodent model observations on carcinogenic potential relate to human risk, especially for a class of chemicals that is ubiquitous in nature. We propose that this question may be addressed, at least partially, by additional data derived from studies conducted in human-cell-based models, especially models that possess human metabolic activation pathways (Guo et al. 2020a; Krewski et al. 2010). This is a particularly important question for human drug impurities where toxicity in humans is the major, if not only concern.

Most nitrosamines require metabolism by cytochromes P450 enzymes (CYPs) to produce the DNA damage that results in mutagenicity and carcinogenicity (Li and Hecht 2022). We previously tested the genotoxicity of six nitrosamines (NDEA, NEIPA, NMBA, NMPA, NDIPA, and NDBA) in human lymphoblastoid TK6 cells transduced with human CYP2A6 (Li et al. 2022). The results showed that all six compounds increased the phosphorylation of histone H2A.X (γH2A.X) and four of them, NDEA, NEIPA, NMBA, and NMPA, induced micronucleus (MN) formation in CYP2A6-expressing TK6 cells. The metabolism of nitrosamines by CYPs has shown substrate specificity and alkyl group selectivity (Lee et al. 1989). CYP2E1 and CYP2A6 have been identified as major catalysts for the metabolic activation of several nitrosamines, while other CYPs, i.e., CYP1A2, 2B6, 2C9, 2C19, and 3A4, also appear to be involved in the activation of some nitrosamines (Schrenk et al. 2023; Yamazaki et al. 1992). Given the number of CYPs potentially involved in nitrosamine activation, their genotoxicity assessment may benefit from conducting tests using cell models that contain multiple CYPs at levels close to those found in humans in vivo.

The human hepatoma HepaRG cell line expresses both Phase I and Phase II enzymes at levels more similar to primary human hepatocytes than other hepatoma cell lines (Duivenvoorde et al. 2021; Seo et al. 2020). Our previous studies characterized the metabolic capacity of two-dimensional (2D) cultures and 3D spheroid cultures of HepaRG cells and demonstrated that both models maintained appreciable levels of CYP activity and gene expression, but that 3D spheroids exhibited relatively higher levels of most CYPs (Seo et al. 2022, 2019). The present study employed both 2D HepaRG cells and 3D HepaRG spheroids to evaluate the genotoxicity of eight nitrosamine drug impurities (Fig. 1) using two high-throughput (HT) genotoxicity assays, the CometChip assay and flow-cytometry-based MN assay. Another HT flow-cytometry-based assay, the MultiFlow DNA damage assay, was conducted in HepaRG spheroids to investigate changes in multiple endpoints associated with DNA damage response pathways. The resulting genotoxicity data were quantified using the benchmark concentration (BMC) approach as described previously (Guo and Mei 2018).

Materials and methods

Chemicals

CPNP (CAS# 61379–66-6) was purchased from Toronto Research Chemicals (Toronto, ON, Canada). Dimethylsulfoxide (DMSO), NDEA (CAS# 55–18-5) and NDMA (CAS# 62–75-9) were obtained from Sigma Aldrich (St. Louis, MO). NEIPA (CAS# 16339–04-1) and NMPA (CAS# 614–00-6) were purchased from Enamine (Monmouth Jct., NJ). NMBA (CAS# 61445–55-4), NDBA (CAS# 924–16-3), and NDIPA (CAS# 601–77-4) were acquired from Chemspace (Monmouth Junction, NJ), TCI America (Portland, OR), and ChemService (West Chester, PA), respectively. All chemicals were stored as recommended by the vendor.

HepaRG cell culture and spheroid formation

The HepaRG cell line was purchased from Biopredic International (Saint Grégoire, France). HepaRG cells were routinely propagated in growth medium as described previously (Seo et al. 2019). The growth medium was prepared by adding growth additive (Lonza; Walkersville, MD) to William’s E Medium (Thermo Fisher; Waltham, MA) supplemented with 2 mM GlutaMax (Thermo Fisher) and 100 μg/ml primocin (InvivoGen; San Diego, CA). After 2 weeks of culture, HepaRG cells were maintained in differentiation medium for 2 more weeks. The differentiation medium was the same as the growth medium except that the growth additive was replaced with differentiation additive (Lonza). Differentiated HepaRG cells were used at passages 16–19 for the treatment with nitrosamines.

Fully differentiated HepaRG cells were dissociated with TrypLE Express (Thermo Fisher) and replated into a 96-well flat bottom plate at a density of 5 × 10⁴ cells/well for 2D cell cultures or into a 384-well ultra-low attachment (ULA) plate (Corning, NY) at a density of 5 × 10³ cells/well to form 3D spheroids (5 × 10³ cells per spheroid) as described previously (Seo et al. 2022, 2019). The plates were incubated at 37 °C in a humidified atmosphere with 5% CO₂ in air. The medium was refreshed every 2–3 days using a VIAFLO 96/384 Electronic Pipette (INTEGRA Biosciences; Hudson, NH).

mRNA expression of Phase I and II metabolizing enzymes

The mRNA expression levels of Phase I and II metabolizing enzymes were determined using the comparative cycle threshold (Ct) method as previously described (Seo et al. 2022). Briefly, total RNA from 2D cultured cells at Day3 (3 days after seeding) and 3D HepaRG spheroids at Day10 (10 days after seeding) was isolated using the RNeasy Mini kit (Qiagen; Valencia, CA). The concentration and quality of the extracted RNA were measured by a NanoDrop 8000 spectrophotometer (Thermo Fisher). cDNA was reverse-transcribed from 1 μg of RNA using a High-Capacity cDNA Reverse Transcription kit (Applied Biosystems; Foster City, CA). Quantitative real-time PCR (qPCR) reactions were carried out in a final volume of 20 μl containing the FastStart Universal Probe Master (Rox) (Roche Applied Science; Indianapolis, IN) and 18 TaqMan probes for 13 Phase I and 5 Phase II enzymes (Supplementary Table 1) on a ViiA 7 Real-Time PCR system (Applied Biosystems). The reverse transcription step involved incubation at 55 °C for 20 min. The amplification protocol included 2 min at 50 °C, 10 min at 95 °C, and 40 cycles at 95 °C for 15 s and 1 min at 60 °C. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a reference gene to normalize levels of gene expression. The gene expression value for each gene evaluated was defined by the equation: E = 2^{−(Ct of test gene – Ct of reference gene)} × 10,000, which was different from our previous study that used fold increases for the comparison of gene expression levels between 2D and 3D cultures (Seo et al. 2022). The expression value represents the relative mRNA expression abundance of a gene, arbitrarily assuming an expression level of GAPDH of 10,000 copies.

Nitrosamine treatment

Stock solutions were prepared by dissolving all nitrosamines in DMSO, except for NDMA, which was dissolved in deionized water. The highest concentrations were prepared by diluting the stock solution at a ratio of 1:100 in differentiation medium and serial dilutions were performed to achieve the final concentrations. The final DMSO concentrations never exceeded 1%. For 3D cultures, approximately 7–8 HepaRG spheroids at Day10 were transferred from 384-well ULA plates into each well of a 96-well round-bottom plate (TPP, Switzerland). Then the spheroids and 2D HepaRG cells at Day3 were exposed to various concentrations of the eight nitrosamines in a total volume of 100 μl treatment medium for 24 h at 37 °C. Treatments were conducted in a humidified atmosphere of 5% CO₂ in air. The experiments were repeated independently at least three times for each chemical.

CometChip assay

Following a 24-h exposure of both 2D and 3D cultures, chemical-induced DNA damage was evaluated by the CometChip assay and cytotoxicity was assessed by the ATP assay. The ATP assay was conducted using the CellTiter-Glo Luminescent Cell Viability Assay kit (Promega) and the CellTiter-Glo 3D Cell Viability Assay kit (Promega) for 2D and 3D HepaRG cultures, respectively, as described previously (Seo et al. 2022).

For conducting the CometChip assay, 2D HepaRG cells and 3D spheroids were dissociated into single cell suspensions using TrypLE Express. The cells then were transferred into the corresponding wells of a 96-well CometChip (R&D Systems; Minneapolis, MN), each well containing ~ 400 micropores. The cells were allowed to settle into the microwells by gravity for 40 min at 37 °C. The CometChip was rinsed gently with Dulbecco’s phosphate buffered saline (DPBS), sealed with 1% low-melting-point agarose, and then submerged in lysis solution (Trevigen; Minneapolis, MN) for 1 h at 4 °C. DNA was unwound in alkaline buffer (0.2 M NaOH, 1 mM EDTA, 0.1% Triton X-100, pH > 13) for 40 min, followed by electrophoresis at 22 V and 300 mA for 50 min at 4 °C. After neutralization and equilibration in Tris–HCl buffer (pH 7.4, Sigma-Aldrich), DNA was stained with 0.2 × SYBR Gold (Invitrogen; Carlsbad, CA) overnight at 4 °C. CometChip images were automatically acquired using a Cytation 5 Cell Imaging Multi-Mode Reader (BioTek; Winooski, VT) and analyzed using Trevigen Comet Analysis Software to score the percentage of DNA in the comet tail (% tail DNA).

HT flow-cytometry-based MN assay

Prior to conducting the MN assay, 2D cultured cells and 3D spheroids were stimulated to divide by adding 100 ng/ml human epidermal growth factor (hEGF) to differentiation media and culturing for additional 3 and 6 days, respectively. The additional incubation allowed the untreated HepaRG cells to go through 1.5–2 cell cycles (Seo et al. 2023), as recommended by the Organization for Economic Co-operation and Development (OECD) Test Guideline 487 for the in vitro mammalian cell MN test (OECD 2016).

Flow-cytometry-based MN analysis was performed using the In Vitro MicroFlow kit (Litron Laboratories; Rochester, NY) as previously described (Seo et al. 2023). The cells were first stained with ethidium monoazide (EMA) and then lysed and stained with lysis solution containing RNase and SYTOX Green. Apoptotic/necrotic cells and chromatin were labeled with EMA and SYTOX Green, respectively. MN events were counted using a FACSCanto II flow cytometer (BD Biosciences; San Jose, CA) equipped with an HT sampler and using a stopping gate of 10,000 intact nuclei. The percentage of MN induction (%MN) was calculated using the count of MN events relative to the nucleated events; relative survival (%RS, as an indicator of cytotoxicity) was calculated using nucleated cell count events for treated cells divided by those for vehicle controls at a specified time point. Cytotoxicity also was evaluated in 3D spheroids (one 5 × 10³ spheroid/well) by an ATP assay following the 24-h treatment plus the additional 6 days of hEGF stimulation (100 ng/ml).

MultiFlow DNA damage assay

DNA damage biomarker responses were investigated using a MultiFlow DNA Damage Kit–P53, γH2A.X, Phospho-Histone H3 kit (Litron Laboratories) according to the manufacturer’s protocol with minor modifications (Guo et al. 2020b). Briefly, HepaRG spheroids were exposed to the eight nitrosamines for 24 h in 200 μl differentiation medium supplemented with 100 ng/ml hEGF. Following the 24-h treatment, the treatment medium was removed and the spheroids were maintained in the same medium without nitrosamines for additional 5 days. Then the spheroids were dissociated with Accutase solution (Thermo Fisher) and the cells were labelled with anti-γH2A.X-Alexa Fluor 647, anti-phospho-histone H3-PE, and anti-P53-FITC antibodies for 1 h at room temperature to determine DNA double strand breaks, mitotic cells, and genotoxic stress, respectively. Flow cytometric analysis was performed using a FACSCanto II Cell Analyzer equipped with a HT sampler (BD Biosciences). Responses of γH2A.X and P53 were expressed as fold-changes relative to the vehicle control based on median fluorescence intensities. Phosphorylated histone H3 (p-H3) events were calculated as %p-H3-positive events relative to total events. These values were then converted to fold-changes relative to the vehicle control.

Quantitative analysis

Both DNA damage and MN data for the eight nitrosamines were quantified using BMC analysis by PROAST web-based software (version 70.1), with ‘compound’ used as a covariate. The concentration-responses were analyzed using both the exponential and Hill models as recommended by the European Food Safety Authority (EFSA) for the analysis of continuous data (EFSA et al. 2017). A critical effect size (CES) of 0.5 (BMC₅₀), corresponding to 50% increases in DNA damage (% tail DNA and γH2A.X) and chromosomal damage (%MN) relative to the vehicle control responses, was chosen to evaluate the relative genotoxicity of the eight nitrosamines. The BMC₅₀s and their upper and lower bounds (BMCU and BMCL, respectively) of the 90% confidence intervals (CIs) were used for quantitative comparisons of relative nitrosamine responses and between the 2D and 3D HepaRG models. The covariate approach was used to improve the BMC precision (Johnson et al. 2016). Responses having non-overlapping BMCU and BMCL intervals were considered significantly different from each other.

Statistical analysis

Data are expressed as the mean ± standard deviation (SD) from at least three independent experiments. Statistical analyses were performed using SigmaPlot 13.0 (Systat Software; San Jose, CA). The statistical significance of mRNA gene expression data was evaluated by the two-tailed Student's t-test. One-way analysis of variance (ANOVA) followed by Dunnett’s post hoc test was used for determining the lowest effective concentration (LEC) in the comet and MN assays and for comparing the CometChip, MN, and Multiflow data between treatment groups and the vehicle control group. A p value of < 0.05 was considered statistically significant.

Results

Increased metabolic gene expression in 3D HepaRG spheroids compared to 2D cultures

Basal mRNA expression of 13 Phase I and 5 Phase II enzymes was investigated in 2D HepaRG cells at Day3 and in 3D spheroids at Day10 using qPCR (Fig. 2). HepaRG spheroids showed higher mRNA expression for 11 Phase I enzymes than did 2D HepaRG cultures. Specifically, compared to 2D cells, the expression of CYP1A1, 1B1, 2C9, 2D6, 2E1, and 3A4 in spheroids increased by 1.4–2.7 fold; the expression of CYP2A6, 2B6, 2C8, and 2C19 increased by 5.3–8.2 fold; and CYP1A2 expression increased 17-fold. There was no significant difference in the expressions of CYP3A5 and two Phase II enzymes (SULT2A1 and UGT1A1) between the two cell models, whereas the expression of CYP3A7 and two Phase II enzymes, NAT1 and UGT1A6, reduced by 17–67% in 3D spheroids compared to 2D HepaRG cells. The Ct values for SULT1A1 gene expression averaged greater than 35 in both 2D and 3D HepaRG cell models (Supplementary Table 1); thus SULT1A1 expression was considered not detectable in the present study. Compared to the basal mRNA expression of Phase I and II enzymes, the expression levels of these enzymes seemed reduced after an additional 5-day stimulation with hEGF, as measured in our previous study (Seo et al. 2023).

Fig. 2.

Gene expression of Phase I and Phase II enzymes in 2D and 3D HepaRG models. Total RNA was extracted from 2D HepaRG cells at Day3 and 5 × 10³ spheroids at Day10 for cDNA synthesis. mRNA expression was measured by quantitative real-time PCR. GAPDH was used as a reference gene. The data are presented as the mean ± SD (n ≥ 3). Part of the raw data are from our previous study (Seo et al. 2022). Significant difference was determined by two-tailed Student’s t-test (*p < 0.05 and the fold change ≤ five-fold; **p < 0.001 and the fold change ≤ 10; and ***p < 0.001 and the fold change > 10). CYP cytochrome P450; NAT N-acetyltransferase; SULT sulfotransferase; UGT UDP-glucuronosyltransferase; GAPDH glyceraldehyde 3-phosphate dehydrogenase; ND non-detectable

Nitrosamine-induced DNA damage and cytotoxicity in 2D and 3D HepaRG models

DNA damage was investigated as DNA strand breaks using the CometChip assay with simultaneous cytotoxicity assessment by the ATP assay. A relative cell viability of at least 70% was used as a cut-off for evaluating DNA damage in both 2D and 3D HepaRG cultures. When limited cytotoxicity was observed, the highest test concentration used for subsequent testing was 10 mM, as recommended by the OECD guidance for in vitro genetic toxicology testing (OECD 2015). Following a 24-h exposure, the cytotoxicity of all the eight nitrosamines reached the limit level at lower concentrations in 3D spheroids compared to 2D cells (Table 1A). NDIPA and NEIPA caused less than 30% cytotoxicity at concentrations up to 10 mM in both 2D and 3D HepaRG models (Supplementary Fig. 1).

Table 1.

Comparison of nitrosamine-induced cytotoxicity and genotoxicity in 2D and 3D HepaRG cultures after a 24-h exposure

A. CometChip assay
Nitrosamine	Max Conc. (μM)a		ATP (%)b		LECc		Fold increased		Outcomee
	2D	3D	2D	3D	2D	3D	2D	3D	2D	3D
CPNP	5000	1250	70.8	86.4	−	312.5	0.9	2.3	−	+ +
NDBA	2000	500	78.5	83.5	2000	125	2.4	2.8	+ +	+ +
NDEA	10,000	1250	73.3	89.6	7500	312.5	2.3	2.6	+ +	+ +
NDIPA	10,000	7500	82.3	84.7	−	5000	1.1	1.8	−	+
NDMA	10,000	625	85.6	98.8	7500	78.1	4.6	6.1	+ +	+ + +
NEIPA	10,000	5000	76.5	82.5	−	2500	1.5	2.4	−	+ +
NMBA	7500	2500	77.0	81.0	−	625	1.0	2.3	−	+ +
NMPA	3000	375	74.8	89.9	−	187.5	0.9	2.3	−	+ +
Positivity									37.5%	100%
B. MN assay
Nitrosamine	Max Conc. (μM)a		RS (%); [%ATP]b		LECc		Fold increased		Outcomee
	2D	3D	2D	3D	2D	3D	2D	3D	2D	3D
CPNP	4000	4000	61.3	73.7 [64.6]		−	1.4	1.4	−	−
NDBA	1500	1500	66.6	58.0 [58.7]	1000	500	1.5	2.3	+	+ +
NDEA	10,000	10,000	80.7	84.1 [65.8]	7500	5000	1.7	2.2	+	+ +
NDIPA	10,000	10,000	77.8	93.5 [73.3]	−	−	1.1	1.2	−	−
NDMA	10,000	5000	78.5	42.6 [48.8]	7500	1250	2.1	3.6	+ +	+ +
NEIPA	10,000	10,000	79.3	89.9 [68.5]	−	−	1.1	1.3	−	−
NMBA	10,000	10,000	63.5	76.7 [66.0]	−	7500	1.4	1.8	−	+
NMPA	4000	4000	87.1	79.9 [56.5]	−	3000	1.2	2.3	−	+ +
Positivity									37.5%	62.5%

^aThe highest concentration tested in the CometChip and MN assays

^b%ATP and RS (relative survival, %) are used as indicators for cytotoxicity, relative to the vehicle control

^cLEC, the lowest effective concentration, determined by one-way ANOVA followed by Dunnett’s test, is the lowest concentration that induced a significant response in the assay

^dThe fold increase of chemical-induced DNA damage or MN formation over the vehicle control at the maximum concentration shown in the table

^eThe ratio ≤ 1.5-fold (p ≥ 0.05 vs. vehicle control, green color); + , 1.5 ≤ ratio < 2; + + , 2 ≤ ratio < 5; and + + + , ratio ≥ 5 (p < 0.05, red color)

CPNP N-cyclopentyl-4-nitrosopiperazine; NDBA N-nitrosodibutylamine; NDEA N-nitrosodiethylamine; NDMA N-nitrosodimethylamine; NDIPA N-nitrosodiisopropylamine; NEIPA N-nitrosoethylisopropylamine; NMBA N-nitroso-N-methyl-4-aminobutyric acid; and NMPA N-nitrosomethylphenylamine

Under the current experimental conditions, three nitrosamines (NDBA, NDEA, and NDMA) induced statistically significant increases in DNA damage in 2D HepaRG cells (2.4-, 2.3-, and 4.6-fold increases over the control, respectively) (Fig. 3A). In contrast, all eight nitrosamines produced positive responses in 3D spheroids (Fig. 3B), with much lower LEC values observed in the spheroids than in 2D cultures for the three nitrosamines (NDBA, NDEA, and NDMA) that were positive in both cell models (Table 1A). Seven nitrosamines, CPNP, NDBA, NDEA, NDMA, NEIPA, NMBA, and NMPA, induced clear concentration-dependent increases in DNA strand breaks (2.3–6.1-fold) in 3D spheroids, while NDIPA produced relatively weak but significant responses (1.8-fold increases compared to the concurrent vehicle controls).

Fig. 3.

Nitrosamine-induced DNA damage and cytotoxicity in 2D A and 3D B HepaRG models. 2D and 3D HepaRG cultures were exposed to eight nitrosamines for 24 h. DNA damage (% tail DNA; left y-axis and black bars) was detected using the CometChip assay. Relative viability (% of control) was measured by the ATP assay (right y-axis and red lines). The data are presented as the mean ± SD (n ≥ 3). Significant difference was determined by one-way ANOVA followed by Dunnett’s test (*p < 0.05, **p < 0.01, and ***p < 0.001 vs. vehicle control). See Fig. 1 for abbreviations of the nitrosamines tested

Nitrosamine-induced MN formation in 2D and 3D HepaRG models

As part of the MN assay, nitrosamine-induced cytotoxicity was evaluated in hEGF-stimulated HepaRG spheroids by %RS and relative cell viability was determined using the ATP assay, providing orthogonal measures for confirmation of responses (Fig. 4 and Supplementary Fig. 2). The concentrations that resulted in < 40% RS and ATP levels (Table 1B) were excluded to minimize misleading false positive results for assessing MN formation. As a result, all nitrosamines had the same highest testable concentration in 2D and 3D HepaRG cultures, except for NDMA, which had a lower highest testable concentration in 3D spheroids compared to 2D cells (Max. conc.; 5 mM vs. 10 mM). Three out of the eight nitrosamines, NDBA, NDEA, and NDMA caused significant increases in %MN in both 2D and 3D HepaRG models, with lower LECs and higher %MN frequencies observed in 3D spheroids (2.3-, 2.2-, and 3.6-fold, respectively, relative to the vehicle control) than in 2D HepaRG cells (1.5-, 1.7-, and 2.1-fold, respectively). Two compounds, NMBA and NMPA, were positive only in 3D spheroids (1.8-, and 2.3-fold increases relative to the vehicle control, respectively). CPNP, NDIPA, and NEIPA, produced negative MN responses in both the 2D and 3D HepaRG models.

Fig. 4.

Nitrosamine-induced MN formation and relative survival in 2D A and 3D B HepaRG models. 2D and 3D HepaRG cultures were exposed to eight nitrosamines for 24 h, followed by an additional 3- or 6-day incubation with hEGF (100 ng/ml). MN formation was measured using a flow-cytometry-based MN analysis. MN frequency is presented as the percentage of micronuclei relative to intact nuclei (%MN; left y-axis and black bars) and cytotoxicity is presented as the percentage of relative survival (%RS; right y-axis and blue line). The data are presented as the mean ± SD (n ≥ 3). Significant difference was determined by one-way ANOVA followed by Dunnett’s test (*p < 0.05, **p < 0.01, and ***p < 0.001 vs. vehicle control). See Fig. 1 for abbreviations of the nitrosamines tested

Nitrosamine-induced DNA damage biomarkers in 3D HepaRG spheroids

DNA damage-associated biomarkers, including γH2A.X and nuclear P53, were measured using the MultiFlow flow-cytometry-based assay in hEGF-stimulated 3D HepaRG models (Fig. 5). All eight nitrosamines induced statistically significant increases in γH2A.X formation in 3D spheroids. Specifically, at the highest test concentrations, NDEA and NDMA induced 2.0–2.3-fold increases in γH2A.X formation, while CPNP, NDBA, NEIPA, and NMBA produced 1.5–1.8-fold increases over the control. NDIPA and NMPA induced weak but statistically significant increases in γH2A.X responses (1.2–1.3-fold). Under the treatment conditions in the present study, only NDMA caused a statistically significant increases in nuclear P53 activation (1.4–1.5-fold) and %p-H3 events (1.6–2.6-fold) at concentrations ≥ 2500 μM (Supplementary Fig. 3).

Fig. 5.

Nitrosamine-induced γH2A.X and nuclear P53 activation in 3D HepaRG models. 3D spheroids were exposed to eight nitrosamines with hEGF (100 ng/ml) for 24 h. After an additional 5-day incubation with hEGF, γH2A.X shift (red circles and lines) and nuclear P53 shift (black squares and lines) were determined by the in vitro MultiFlow DNA damage assay. The data are presented as the mean ± SD (n ≥ 3). Significant difference was determined by one-way ANOVA followed by Dunnett’s test (*p < 0.05, **p < 0.01, and ***p < 0.001 vs. vehicle control). See Fig. 1 for abbreviations of the nitrosamines tested

Relative genotoxicity of eight nitrosamines in HepaRG cells

The relative genotoxicity of the eight nitrosamines was compared using BMC₅₀ values derived by PROAST software from the DNA damage, MN formation, and γH2A.X induction concentration-responses in 2D and/or 3D HepaRG models (Supplementary Fig. 4). The exponential model was used for quantitative comparisons based on its superior fit to the data. NDMA had the lowest BMC₅₀ estimate for inducing DNA strand breakage in the spheroids, followed by NDBA and NMPA, while NDIPA had the highest BMC₅₀ estimate (Fig. 6A and Supplementary Fig. 4A). For the five positive %MN responses in 3D spheroids, NDMA also had the lowest BMC₅₀ estimate, followed by NDBA, NMPA, NDEA, and NMBA (Fig. 6B and Supplementary Fig. 4B). The γH2A.X responses had a similar pattern to the DNA damage responses in 3D HepaRG cultures, with NDBA and NDIPA having the lowest and highest BMC₅₀ value, respectively, among the eight nitrosamines (Fig. 6C and Supplementary Fig. 4C).

Fig. 6.

Genotoxic potency of nitrosamines. Benchmark concentration (BMC)₅₀ value and its upper and lower 90% confidence intervals (CIs, BMCUs and BMCLs) were calculated from 2D and 3D HepaRG CometChip A and MN B data and 3D HepaRG γH2A.X responses C using PROAST software. The exponential model was used for the comparison of BMC values. The bars represent the CIs of BMC₅₀ estimates. Black, 2D HepaRG cells; Red, 3D HepaRG spheroids. See Fig. 1 for abbreviations of the nitrosamines tested

The BMC₅₀ values for the comet and MN assays were compared between 2D and 3D HepaRG models using data generated by the three nitrosamines that were positive in both cell models: NDBA, NDEA, and NDMA. NDMA produced significantly lower BMC₅₀ CIs for both assays in the HepaRG spheroids than in 2D cultures. The BMC₅₀ values derived from the comet data for NDBA and NDEA also were significantly lower in the spheroids than in 2D cultures, while BMC₅₀ CIs derived from the 2D and 3D MN data overlapped each other for these two nitrosamines (Fig. 6).

Discussion

A comprehensive carcinogenicity study that evaluated dose–response data from 4080 rats exposed to NDEA or NDMA indicated that liver was the most sensitive target for nitrosamine tumorigenicity (Peto et al. 1991). Taking advantage of the metabolic capacity of human HepaRG cells, the present study evaluated the genotoxicity of eight nitrosamines using the CometChip and the HT flow-cytometry-based MN and MutiFlow DNA damage assays. Under the conditions used in our study, HepaRG spheroids overall showed a higher positivity rate than 2D cells for detecting the DNA damage and MN formation induced by the eight nitrosamines (100 vs. 37.5% and 62.5 vs. 37.5%, respectively; Table 1). It should be emphasized that these findings may be dependent upon the methods used to prepare the spheroids and 2D cultures used in this study as the expression of enzymatic activity in HepaRG cells is influenced by the methods used to culture and differentiate the cell cultures (Hammour et al. 2022).

In general, NDMA is metabolized mainly by CYP2E1, and CYP2A6 contributes to the bioactivation of small-to-medium-sized nitrosamines, including NDEA, NMPA, and NMBA (Fujita and Kamataki 2001). The other three nitrosamines tested in the present study: CPNP, NDIPA, and NEIPA, also are considered “small-medium” in size. Compared to 2D HepaRG cells, HepaRG spheroids had significantly higher mRNA expression levels for CYP2A6, 2E1, as well as other CYPs, while both cell models had comparable levels of Phase II enzymes (Fig. 2), observations consistent with the significant role Phase I enzymes play in the bioactivation of nitrosamines.

Five out of the eight nitrosamines: NDBA, NDEA, NDMA, NMBA, and NMPA, are considered rodent carcinogens according to the Lhasa Carcinogenicity Database (Table 2). Among them, only three compounds, NDBA, NDEA, and NDMA, were positive in the 2D HepaRG comet and MN assays (Table 1). In contrast, all five rodent carcinogens induced significant DNA damage and MN formation in 3D spheroids. These results suggest that HepaRG spheroids are more sensitive than 2D cultures for detecting N-nitroso rodent carcinogens in standard genotoxicity assays. In addition, all eight nitrosamines produced positive responses in both the comet and γH2A.X assays in the spheroids, indicating both assays were equally sensitive in detecting the DNA damage induced by these compounds. Generally, higher fold-increases over the vehicle control were observed in the comet assay relative to the γH2A.X assay, which might be due to the inherent differences in dynamic range between these two assays, the sampling times used to detect the endpoints, along with the DNA repair capacity of HepaRG cells. The comet assay was conducted immediately after the 24-h nitrosamine exposure whereas the γH2A.X assay was performed after the 24-h exposure followed by an additional 5-day incubation, which provided more time for the removal of DNA damage. In agreement with our previous study using CYP2A6-expressing TK6 cells (Li et al. 2022), the nuclear P53 response was the least sensitive endpoint for detecting the DNA damage induced by nitrosamines in the spheroids.

Table 2.

The reported genotoxicity and carcinogenicity of test nitrosamines and the BMCL₅₀ values in HepaRG models

Nitrosamine	Amesa	ECHAb	IARCc	Carcinogenicity d	Gold TD50 (mg/ kg/day) in ratd	Oral rat LD50 mg/kge	BMCL50 (comet)		BMCL50 (MN)
							2D	3D	2D	3D
CPNP	+	Carc.1B	N/A	N/A	N/A	N/A	−	175	−	−
NDBA	+	Carc.2	2B	+	0.691	1200	206	38.2	398	79.2
NDEA	+	Carc.1A/1B; Muta.2	2A	+	0.0265	280	679	147	3820	1780
NDIPA	+	Carc.1B; Muta.1B	N/A	N/A	N/A	850	−	2530	−	−
NDMA	+	Carc.1B	2A	+	0.0959	40	381	1.01	539	24.7
NEIPA	+	Carc.1B	N/A	N/A	N/A	1100	−	597	−	−
NMBA	−/+f	Carc.2	N/A	+	0.982	6800 g	−	326	−	2540
NMPA	+	Carc.2	N/A	+	0.142	280	−	45.6	−	497

^aUnpublished data from Heflich R. et al

^bData are from the European Chemicals Agency (ECHA) Classification and Labelling Inventory https://echa.europa.eu/information-on-chemicals/cl-inventory-database. Carc.1A, substances known to have carcinogenic potential for humans, classification is largely based on human evidence; Carc.1B, substances which are presumed to have carcinogenic potential for humans, classification is largely based on animal evidence; Carc.2, suspected human carcinogens based on evidence obtained from human and/or animal studies but which is not sufficient for a Carc. 1 classification; Muta.1B, substances known to induce heritable genetic mutations or to be regarded as if they induce heritable mutations in the germ cells of humans. The classification is based on positive mutagenicity test results on cells in humans; Muta.2, mutagens cause concern for human owing to the possibility that they may induce heritable mutations in the germ cells of humans

^cGroups are classified by International Agency for Research on Cancer (IARC) Monographs on the identification of carcinogenic hazards to humans, Volumes 1–133. https://monographs.iarc.who.int/list-of-classifications

^dData are from Lhasa Carcinogenicity Database. https://carcdb.lhasalimited.org/

^eData are from Ambient water quality criteria for nitrosamines by US Environmental Protection Agency. https://www.epa.gov/sites/default/files/2018-12/documents/ambient-wqc-nitrosamines.pdf

^fPositive response was observed only when NMBA was treated with special activation systems (Inami et al. 2013)

^gOral rat LTDLO mg/kg, the lowest dose causing a toxic effect

TD₅₀ Toxic Dose, the median toxic dose of a substance at which toxicity occurs in 50% of the cases; LD₅₀ Lethal Dose, the dose that causes the death of 50% (one half) of a group of test animals; BMCL₅₀ the lower limit of a one-sided 90% confidence interval on the benchmark concentration that produces 50% increase in the responses over the control; N/A not available

Currently, no or limited rodent carcinogenicity data are available in the Lhasa Carcinogenicity Database for CPNP, NDIPA, or NEIPA. However, all three were considered possible human carcinogens (Category 1B) by the European Union legislation regarding Classification Labelling and Packaging (CLP) of substances (Table 2) (ECHA 2023). Limited data are available on the genotoxicity and carcinogenicity of CPNP, while NDIPA has been reported to be a weak carcinogen in Sprague–Dawley rats treated 5 days each week with 1.8 or 12 mg NDIPA for 50 or 40 weeks, respectively (Lijinsky and Taylor 1979). NDIPA and NEIPA have a similar structure, with NDIPA having an additional carbon on the sidechain. It has been shown that the length of the sidechain, ring size, and molecular weight negatively affect nitrosamine carcinogenic potency (Cross and Ponting 2021). NEIPA induced nuclear P53 activation, γH2A.X and MN formation, and G2/M phase cell cycle arrest in CYP2A6-expressing TK6 cells, whereas NDIPA showed positive responses only for γH2A.X formation, with a much weaker response (Li et al. 2022). Likewise, NEIPA, compared to NDIPA, induced positive responses at lower concentrations and had lower BMC₅₀ values in HepaRG spheroids using data from the comet and γH2A.X assays (Figs. 3B and 5), indicating a higher DNA damage potency. CPNP, NDIPA, and NEIPA all are mutagenic in the Ames test (Heflich R.H. et al. manuscript in preparation), and NDIPA has been classified as a Category 1B possible human mutagen based on the EU CLP regulation (ECHA 2023). However, none of these compounds increased MN frequencies in HepaRG spheroids under our experimental conditions (Fig. 4). While not clastogens/aneugens, additional mutagenicity testing using error-corrected next generation sequencing (ecNGS) (Marchetti et al. 2023) may be helpful for determining whether these compounds are mutagenic in HepaRG cells.

NDBA is a relatively strong urinary-bladder carcinogen, but also caused tumors in other organs, including liver tumors in various animal species (IARC 1978). NDBA induced concentration- and time-dependent increases in DNA strand breaks in primary rat hepatocytes (Bradley et al. 1982). NDBA also increased MN formation and sister chromatid exchanges in Namalva cells, a human Burkitt lymphoblastoid cell line, at concentrations as low as 10 μM and 1 μM, respectively (Janzowski et al. 1989). In the present study, NDBA caused DNA damage and MN formation in HepaRG cells cultured in both 2D and 3D formats, resulting in lower LEC and BMC₅₀ values compared to other nitrosamines (Table 1 and Fig. 6). However, NDBA didn’t increase the MN frequency in wild-type TK6 cells or TK6 cells transduced with human CYP1A1, 2A6, 2E1, or 3A4 genes (Li et al. 2022), suggesting that these CYPs played a limited role in metabolizing NDBA. By using liver microsomes prepared from rats, Shu and Hollenberg (Shu and Hollenberg 1996) demonstrated that CYP2B1 specifically catalyzes α-hydroxylation of long chain nitrosamines including NDBA, as well as increasing the metabolism of NDBA via ω-1-hydroxylation. The rat CYP2B1 gene is homologous to the human CYP2B6 gene (Vanoye-Carlo et al. 2015). Given that compared to 2D cultures, the expression of CYP2B6 mRNA was significantly elevated in HepaRG spheroids (Fig. 2) and that higher DNA damage levels and MN frequency were observed in the spheroids, we hypothesize that CYP2B6 may contribute to NDBA-induced genotoxicity.

Two probable human carcinogens, NDMA and NDEA, were positive in all three genotoxicity assays (the comet, γH2A.X, and MN assays) conducted using both 2D HepaRG cells and 3D spheroids (Figs. 3 and 4), demonstrating the capability of the metabolically competent HepaRG cell models for detecting genotoxic/carcinogenic nitrosamine compounds. Compared to the other seven nitrosamines, NDMA showed the most significantly elevated responses for DNA damage and MN formation in 3D spheroids relative to 2D cultures. NDMA is mainly metabolized by CYP2E1, whose gene expression was two-fold higher in the spheroids compared with 2D cells (Fig. 2), a finding consistent with CYP2E1 being responsible for metabolizing NDMA to genotoxic metabolites (Cross and Ponting 2021). Considering the high variability of nitrosamine metabolism among individuals (Camus et al. 1993), our results highlight the possibility that individuals with high CYP2E1 levels may be disproportionately sensitive to NDMA toxicity. Interestingly, NDEA produced stronger responses than NDMA in rodent carcinogenicity and in vivo mutagenicity studies (Johnson et al. 2021). In HepaRG models, however, NDEA consistently had lower genotoxic potential than NDMA across the three assays conducted in 2D and 3D cell models (Fig. 6). This discrepancy requires further investigation.

Both NMBA and NMPA are carcinogenic to rats and are classified as suspect human carcinogens (Category 2) by the EU CLP (ECHA 2023). In rat carcinogenicity studies, NMPA had lower Gold TD₅₀ and oral LD₅₀ values compared to NMBA (Table 2), suggesting that NMPA is a stronger carcinogen. Our previous study found that NMPA was the most potent of six nitrosamines evaluated for genotoxicity in CYP2A6-expressing TK6 cells (Li et al. 2022). In HepaRG cells, neither NMBA nor NMPA were positive in 2D cultures, but both caused DNA damage and MN formation in HepaRG spheroids (Figs. 3 and 4). A plausible hypothesis is that mitochondrial enzymes in spheroids may biotransform long-chain carboxylated nitrosamines (like NMBA) by β-oxidation, and along with the fatty acid degradation process, result in the generation of lipophilic 2-oxopropyl metabolites, which have been found to be genotoxic and carcinogenic (Janzowski et al. 1994). Compared to NMBA, NMPA was more potent inducing DNA damage as measured by BMC₅₀s (Fig. 6), having much lower BMCL₅₀ values in the comet and MN assays (Table 2), and was more cytotoxic in HepaRG cells, all of which is consistent with the previous rat studies. Although displaying weaker carcinogenicity in F344 rats compared to N-nitrosomethyl-N-propylamine derivatives (Lijinsky et al. 1983), NMBA induced DNA strand breakage and chromosomal damage in both HepaRG spheroids and in CYP2A6-expressing TK6 cells (Li et al. 2022).

In conclusion, the present study demonstrated the genotoxic potential of eight nitrosamines found as impurities in human drug products using human-derived metabolically competent HepaRG cells. The data suggest that, under the culture conditions used for our study, 3D HepaRG spheroids are more sensitive to the genetic toxicity produced by nitrosamines than 2D cultures, an observation likely attributed to the increased expression of CYPs relevant to the nitrosamine metabolic activation in 3D spheroids. In addition, NDMA produced the strongest response among the eight nitrosamines, as well as producing the greatest differences between the genotoxicity responses detected in the 2D and 3D HepaRG models. The current study also provides novel genotoxicity data for CPNP. Notably, the fold increases in genotoxicity detected by HepaRG cells for these nitrosamines were not large, compared with the Ames test findings or even the responses detected by our previous study using CYP-transduced TK6 cells. However, unlike other genotoxicity assays used for hazard identification, HepaRG cell responses reflect metabolism by both activating Phase I and potentially inactivating Phase II metabolic pathways. Thus, these data may give a better indication of actual human risk than can assays that employ only Phase I pathways. Our data suggest that 3D HepaRG spheroids may be a useful model that can potentially be used as a New Approach Methodology for detecting the genotoxicity of nitrosamines in a manner that more accurately reflects human risk.

Supplementary Material

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00204-023-03560-x.

Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files.