Genotoxicity evaluation using primary hepatocytes isolated from rhesus macaque (Macaca mulatta)

Abstract

Non-human primates (NHPs) have played a vital role in fundamental, pre-clinical, and translational studies because of their high physiological and genetic similarity to humans. Here, we report a method to isolate primary hepatocytes from the livers of rhesus macaques (Macaca mulatta) after in situ whole liver perfusion. Isolated primary macaque hepatocytes (PMHs) were treated with various compounds known to have different pathways of genotoxicity/carcinogenicity and the resulting DNA damage was evaluated using the high-throughput CometChip assay. The comet data were quantified using benchmark dose (BMD) modeling and the BMD₅₀ values for treatments of PMHs were compared with those generated from primary human hepatocytes (PHHs) in our previous study (Seo et al. Arch Toxicol 2020, 2207–2224). The results showed that despite varying CYP450 enzyme activities, PMHs had the same sensitivity and specificity as PHHs in detecting four indirect-acting (i.e., requiring metabolic activation) and seven direct-acting genotoxicants/carcinogens, as well as five non-carcinogens that are negative or equivocal for genotoxicity in vivo. The BMD₅₀ estimates and their confidence intervals revealed species differences for DNA damage potency, especially for direct-acting compounds. The present study provides a practical method for maximizing the use of animal tissues by isolating primary hepatocytes from NHPs. Our data support the use of PMHs as a reliable surrogate of PHHs for evaluating the genotoxic hazards of chemical substances for humans.

1. Introduction

Human-derived metabolically competent cells are generally considered to be superior to cells from laboratory animals as in vitro models for evaluating chemical substances for their potential toxicity in humans (NAS, 2017). Human liver-derived cell lines contain various levels of phase I and II enzymes that catalyze the activation and detoxification of toxicants, and thus have been used as alternative models for genetic toxicity testing (Guo et al., 2020a). Primary human hepatocytes (PHHs) are recognized as the “gold standard” since they closely reflect human in vivo liver metabolism and functionality. In addition, several human hepatoma cell models, i.e., HepG2, HCC1.2, HuH6, and Hep3B, have phase I and II enzyme activities ranging between 5–10 % and <50 % of those found in PHHs, respectively (Misik et al., 2019). Our previous studies compared the performance of three human hepatic cell models, HepG2, HepaRG, and PHHs, using the high-throughput high-content (HTHC) CometChip assay; DNA damage dose–responses from these studies were analyzed using benchmark dose (BMD) potency ranking analysis (Seo et al., 2019, 2020). The results showed that PHHs had a higher sensitivity (90 %) for detecting the DNA damage induced by ten known genotoxicants/carcinogens than either HepG2 (60 %) or HepaRG (70 %) cells. Additionally, most of the compounds had comparable BMD values for DNA damage induction in PHHs and HepaRG cells. Despite their superior predictive ability, it is not practical to use PHHs in routine toxicity studies due to their limited availability and high costs (Zeilinger et al., 2016).

Non-human primates (NHPs) are physiologically and genetically similar to humans, and are susceptible to many human diseases (Mudry et al., 2011). NHPs are uniquely capable of reducing uncertainty/ambiguity in toxicological assessments and are often used for evaluating the preclinical toxicity of drugs intended for human use (Grow et al., 2016).

Both in vivo and in vitro NHP models have been used in a limited number of studies for genotoxicity assessment. A single dose of 50 mg/kg of N-ethyl-N-nitrosourea (ENU) significantly increased phosphatidylinositol glycan class A (Pig-a) and hypoxanthine-guanine phosphoribosyl transferase (Hprt) mutant frequencies in peripheral blood lymphocytes of male rhesus macaques, and the spectrum of Hprt mutation was similar to ENU-induced mutation determined for other species (Dobrovolsky et al., 2009). Further, both single and repeated administration of cyclophosphamide (CPA) significantly increased micronucleated reticulocytes in peripheral blood and bone marrow of rhesus macaques (Hotchkiss et al., 2008). Administration of chemotherapeutic agents (thiotepa, etoposide, or paclitaxel) also induced chromosomal aberrations in bone marrow of rhesus macaques (Rao et al., 2005). Quantitation of γ-H2AX foci was used as a biodosimeter for analyzing total-body exposure to ionizing radiation in a rhesus macaque model (Redon et al., 2010). Additionally, high levels of DNA adducts have been detected in several organs, particularly liver, kidney, and heart, in cynomolgus monkeys dosed with three heterocyclic aromatic amines, 2-amino-3-methylimidazo-[4,5-f]-quinoline (IQ), 2-amino-3, 8-dimethylimidazo[4,5-f]-quinoxaline (8-MeIQx), and 2-amino-1-me-thyl-6-phenylimidazo-[4,5-b]-pyridine (PhIP) (Thorgeirsson et al., 1995).

With regards to in vitro NHP studies, to the best of our knowledge, genotoxicity studies have only been conducted with an African green monkey kidney cell line (Vero). Treatment of Vero cells with 1.3 mM of the antifungal drug fluconazole induced DNA damage and micronucleus formation, likely due to the drug’s oxidative properties (Correa et al., 2018). An analgesic drug, dipyrone, also caused significant increases in DNA damage in Vero cells as evaluated by the comet assay (Gomes et al., 2021). These studies demonstrated the potential for NHP-based models in genotoxicity assessment.

Here, we evaluated whether primary macaque hepatocytes (PMHs) can be used as a surrogate for PHHs in genotoxicity assessments by testing 16 compounds known to have different genotoxic/carcinogenic modes of action (MoAs) using the HTHC CometChip assay (Table 1). The PMHs were isolated from liver after in situ whole liver perfusion following scheduled euthanasia of adult male rhesus macaques. The resulting DNA damage responses were quantified using BMD modeling and the BMD₅₀s and their confidence intervals (CIs) compared with those generated using PHHs in our previous study (Seo et al., 2020).

Table 1.

Reported genotoxicity and carcinogenicity of test chemicals.

Group	Chemical	CAS#	Ames	Genotoxicity assay findings		Carcinogenicity*	References
				in vitro	in vivo
1. Indirect-acting genotoxicants/carcinogens	2,4-DAT	95-80-7	(+)	(+) for MLA, CA, MN, comet, and UDS; (−) for comet in HepaRG and Hprt	(+) for UDS and comet; (+/−) for MN in BM	(+) for liver tumor; Group 2B	(Guo et al., 2020b; Kirkland et al., 2016; Le Hegarat et al., 2014; Seo et al., 2019, 2020; Severin et al., 2005)
	B[a]P	50-32-8	(+)	(+) for MLA, CA, MN, Pig-a, comet, Hprt, and UDS	(+) for comet, MN, Pig-a and gene mutations	(+) for liver, lung tumor, etc.; Group 1	(Dertinger et al., 2019; Graupner et al., 2014; Guo et al., 2020b; Kirkland et al., 2016; Le Hegarat et al., 2010; Shah et al., 2018, 2016; Wang et al., 2018)
	CPA	6055-19-2	(+)	(+) for MLA, CA, and MN	(+) for CA, MN, Pig-a, TG, and comet	(+) for liver, lung, skin tumor, etc.; Group 1	(Bhalli et al., 2013; Dertinger et al., 2019; Dertinger et al., 2012; Kirkland et al., 2016; Le Hegarat et al., 2010, 2014; Seo et al., 2019, 2020; Yusuf et al., 2000)
	DMBA	57-97-6	(+)	(+) for MLA, CA, MN, comet, and UDS	(+) for MN and Pig-a in blood, comet in liver, CA in BM, gene mutation	(+) for mammary liver, skin tumor, etc.	(Kirkland et al., 2016; Le Hegarat et al., 2014; Shi et al., 2011)
2. Direct-acting genotoxicants/carcinogens	4-NQO	56-57-5	(+)	(+) for CA, MLA, Pig-a, comet	(+) for MN, Pig-a, TG	(+) for oral tumor	(David et al., 2018; Dertinger et al., 2012; Kanojia and Vaidya, 2006; Le Hegarat et al., 2014)
	CdCl2	10108-64-2	(+/−)	(+) for MN, CA, MLA, Hprt, comet	(+) for CA and MN, comet in liver	(+) for lung, prostate, etc.; Group 1	(Guo et al., 2016; Kirkland et al. 2016; Seo et al., 2019)
	Cisplatin	15663-27-1	(+)	(+) for MN, Pig-a, comet	(+) for MN, Pig-a, TG	(+) for lung tumor, leukemia; Group 2A	(Bemis et al., 2018; Bhalli et al., 2013; Dertinger et al., 2019)
	Colchicine	64-86-8	(−)	(+) for MN, CA, MLA; (−) for comet	(+) for MN in BM and liver	No adequate studies	(FDA and CDER, 2013; Kirkland et al., 2016; Seo et al., 2020)
	ENU	759-73-9	(+)	(+) for MLA, CA, MN, Pig-a, Hprt, comet, and UDS	(+) for CA, MN, Pig-a, TG, comet	(+) for kidney, mammary tumor, etc.; Group 2A	(Bemis et al., 2018; David et al., 2018; Dertinger et al., 2012; Habas et al., 2017; Kirkland et al., 2016; Le Hegarat et al., 2014; Seo et al., 2019, 2020)
	HQ	123-31-9	(−)	(+) for MLA, MN, CA, comet	(+) for CA, MN; (−) for comet and TG in liver and stomach	(+) for liver and kidney adenoma; Group 3	(IARC, 1999; Kirkland et al., 2016; Peng et al., 2013)
	MMS	66-27-3	(+)	(+) for MLA, CA, MN, Pig-a, Hprt, comet, and UDS	(+) for CA, MN, UDS, Pig-a, TG, comet	(+) for nasal, nervous system tumor, etc.; Group 2A	(Bemis et al., 2018; Dertinger et al., 2012; Habas et al., 2017; Kirkland et al., 2016; Liu et al., 2019; Seo et al., 2019, 2020)
3. Non-carcinogens that are negative or equivocal for genotoxicity in vivo	2-Ethyl-1,3-hexanediol	94-96-2	(−)	(+) for CA + S9; (−) for MLA, Hprt, MN, and SCE	(−) for MN, CA	(−) in dermal studies	(Ballantyne, 2005; Kirkland et al., 2016; Slesinski et al., 1988)
	o-Anthranilic acid	118-92-3	(−)	(+) for CA, MLA, SCE; (+/−) for MN	(−) for Pig-a, MN, CA, comet	(−) in rats and mice; Group 3	(Dertinger et al., 2012; Kirkland et al., 2016; (NTP, 1978a)
	Curcumin	458-37-7	(−)	(+) for oxidative DNA damage, CA; (+/−) for MN, SCE	(−) for MN	(−) in male rats; E in mice and female rats	(Damarla et al., 2018; Fowler et al., 2012; Li et al., 2008; NTP 1993; Sebastia et al., 2012)
	Ethionamide	536-33-4	(−)	(+) for MLA; (+/−) for CA; (−) for SCE, MN	(−) for comet	(−) in rats and mice; Group 3	(Kirkland et al., 2016; (NTP, 1978b)
	Resorcinol	108-46-3	(−)	(+) for MLA, SCE, CA; (+/−) for MN	(+/−) for MN	(−) in rats and mice; Group 3	(Kirkland et al., 2016; NTP, 1992; Soeteman-Hernandez et al., 2016)

2,4-DAT: 2,4-diaminotoluene; B[a]P: benzo[a]pyrene; CPA: cyclophosphamide; DMBA: 7,12-dimethylbenzanthracene; 4-NQO: 4-nitroquinoline1-oxide; CdCl₂: cadmium chloride; ENU: N-ethyl-N-nitrosourea; HQ: hydroquinone; MMS: methyl methanesulfonate.

E, equivocal; BM, bone marrow; CA, chromosome aberration; Hprt, hypoxanthine-guanine phosphoribosyltransferase; MLA, mouse lymphoma assay; MN, micronucleus; Pig-a, phosphatidylinositol glycan class A; SCE, sister chromatid exchange; TG, transgenic mutation; UDS, unscheduled DNA synthesis.

^*Groups were classified by IARC Monographs on the identification of carcinogenic hazards to humans, Volumes 1–125 (https://monographs.iarc.who.int/list-of-classifications). +/–, both positive and negative results were reported.

2. Materials and methods

2.1. Chemicals and materials

Chemicals purchased from Sigma-Aldrich (St. Louis, MO) include: 16 test compounds (Table 1), chemicals and reagents used for hepatocyte isolation [bovine serum albumin (BSA), collagenase type IV, ethylene glycol-bis(2-aminoethylether)-N,N,N’,N’-tetraacetic acid (EGTA), D-(−)-fructose, d-(+)-glucose, heparin, 4-(2-hydroxyethyl)-1-piper-azineethanesulfonic acid (HEPES), hydrocortisone 21-hemisuccinate, insulin, L-15 medium (Leibovitz), percoll, trypsin inhibitor], six chemicals used for CYP450 enzyme substrates (phenacetin, bupropion, chlorzoxazone, diclofenac, dextromethorphan, and midazolam) and their metabolites (4-acetamidophenol, hydroxybupropion, 6-hydroxychlorzoxazone, 4-hydroxydiclofenac, dextrorphan, and 1-hydroxymidazolam, respectively). A CellTiter-Glo^® 2.0 Cell Viability Assay kit (ATP assay) was obtained from Promega (Madison, WI). The CometChip and other comet supplies were purchased from R&D Systems (Minneapolis, MN).

2.2. In situ whole liver perfusion

All animal procedures were approved by the National Center for Toxicological Research (NCTR) Institutional Animal Care and Use Committee. Eight 15-year-old male rhesus macaques (body weight ranging from 11.4 to 14.7 kg) were humanely euthanized for brain tissue collection as part of a long-term neurotoxicological assessment (Rodriguez et al., 2010; Wilkinson et al., 2019; Zhang et al., 2016, 2021). Other tissues, including livers, were collected from these animals at the time of sacrifice in support of the 3Rs of animal testing. All animals had been treatment free for at least one year before their euthanasia and 7 of the 8 animals had livers with no signs of gross pathology at the time of sacrifice. No animals were sacrificed for the sole purpose of harvesting tissue to support this study. Under the guidance of a veterinarian, animals were sedated with ketamine (10 mg/kg, I.M.). Once appropriately sedated, the animals were administered Euthasol^® (60–90 mg/kg, I.V., Virbac Corporation, Westlake, TX). After sedation, the brain and upper body of the animals were perfused with approximately 2 L of heparinized phosphate buffered saline (PBS, Thermo Fisher Scientific, Waltham, MA) for other tissue collection requirements. Then, the liver was flushed clean of residual blood by perfusion with washout solution (9.2 mM HEPES-NaCl buffer containing 0.5 mM EGTA, 16.65 mM fructose, and 10 U/mL heparin) at approximately 40 °C. After 20 min, perfusion was switched to the digestion solution (9.2 mM HEPES-NaCl buffer containing 6.7 mM KCl, 5 mM CaCl₂, 16.65 mM fructose, 0.3 % BSA, 0.06 % collagenase type IV, 100 mg/L hyaluronidase, and 100 mg/L trypsin inhibitor) at approximately 40 °C by use of a 3-way stopcock, without stopping the peristaltic pump to avoid the incorporation of air bubbles into the system. The perfusion was stopped after 20–30 min of perfusion with digestion solution. The endpoint of digestion was determined when the liver possessed an overall soft consistency (Ulrich et al., 1990).

2.3. Primary macaque hepatocytes isolation

After perfusion, a portion of liver was removed and placed in a sterile beaker containing 200 mL suspension medium (L-15 medium supplemented with 0.2 % BSA, 1 M HEPES, 5 μg/mL insulin, and 5 % FBS) on ice and transported within 5 min into a biosafety cabinet. The Glisson’s capsule of the digested liver was cut with sterile scissors and the parenchyma was mechanically dissociated and then gently pushed through two sieves of decreasing sizes (200 and 80 μm). The first cell suspension was centrifuged at 60 ×g for 2 min. After two washes with suspension medium, the cells were purified with 90 % isotonic percoll and resuspended in plating medium (L-15 medium supplemented with 1× nonessential amino acids, 100 U/mL penicillin, 100 μg/mL streptomycin, 1× GlutaMax, 1.4 μM hydrocortisone, 100 nM insulin, and 17 % FBS). Cell number and viability were measured using the trypan blue exclusion assay, and 5 × 10⁴ cells in 100 μL plating medium were plated in each well of collagen I coated 96-well plates and cultured at 37 °C in a humidified incubator containing 5 % CO₂. PMHs were allowed to attach for 4 h, and the medium was replaced with fresh plating medium. PMHs were incubated overnight at 37 °C in a humidified incubator containing 5 % CO₂.

2.4. Determination of CYP450 activities

Activities of six major cytochrome P450 (CYP450) enzymes were measured using established methods (Seo et al., 2019). Briefly, after being cultured overnight, the PMH plating medium was changed to 100 μL/well treatment medium (L-15 medium supplemented with 1× nonessential amino acids, 100 U/mL penicillin, 100 μg/mL streptomycin, 1× GlutaMax, 17 μM hydrocortisone, 100 nM insulin, and 10 % FBS) containing 250 μM chlorzoxazone (CYP2E1) or a cocktail of five substrates, i.e., 100 μM phenacetin (CYP1A2), 100 μM bupropion (CYP2B6), 20 μM diclofenac (CYP2C9), 20 μM dextromethorphan (CYP2D6), and 50 μM midazolam (CYP3A4), in quadruplicate. Following a 2-h incubation, 100 μL medium were collected and the metabolites were quantified by high performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) (Dierks et al., 2001; Seo et al., 2019). The remaining cells were immediately washed with cold PBS and lysed in 30 μl RIPA buffer (Thermo Fisher Scientific, Waltham, MA) on ice for 30 min. The protein concentrations were measured using a Pierce BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA) as described previously (Guo et al., 2015; Seo et al., 2021). Final CYP450 activities were averaged and expressed as pmol metabolite/min/mg protein.

2.5. Cell treatments

Following a 21-h incubation after plating, PMHs were treated for 24 h with 16 compounds that have been tested in PHHs in our previous study (Seo et al., 2020). The highest concentrations were selected based on the cytotoxicity of compounds in PHHs. Specifically, the 16 test compounds included four indirect-acting (requiring metabolic activation) genotoxicants/carcinogens [2,4-diaminotoluene (2,4-DAT), benzo [a]pyrene (B[a]P), cyclophosphamide (CPA), 7,12-dimethylbenz[a] anthracene (DMBA)], seven direct-acting genotoxicants/carcinogens [4-nitroquinoline 1-oxide (4-NQO), cadmium chloride (CdCl₂), cisplatin, colchicine, ENU, hydroquinone (HQ), and methyl methanesulfonate (MMS)], and 5 non-carcinogens that are negative or equivocal for genotoxicity in vivo (2-ethyl-1,3-hexanediol, curcumin, ethionamide, o-anthranilic acid, and resorcinol). All chemical stock solutions were prepared as described previously (Seo et al., 2020). Two wells received 1× and 0.75× of the desired top concentrations of test article in treatment medium. A serial dilution was performed from these top two concentrations to generate 11 closely spaced concentrations with 100 μl per well on the 96-well plates. Dimethyl sulfoxide (DMSO) was used as the vehicle control with a final concentration of ≤ 0.1 %.

2.6. Cytotoxicity and CometChip assays

The cytotoxicity and CometChip assays were performed following a 24-h treatment. Cytotoxicity was measured using the ATP assay as recommended by the manufacturer (Promega). Briefly, the treatment medium was removed following treatment and the ATP reagent was added into each well at a ratio of 1:10, followed by an incubation of 10 min at room temperature. Luminescence was measured using a Cytation 5 Cell Imaging Multi-Mode Reader (BioTek, Winooski, VT). The relative viability (%) was calculated by the ratio of intensity levels of treated cells to those of vehicle controls.

The CometChip assay was performed as described previously with minor modifications (Seo et al., 2020). Briefly, the treated cells were detached by adding 30 μL Accutase^® solution into each well of the 96-well plates, followed by the addition of 120 μL treatment medium to stop the reaction. Then all treated cells were transferred into each well of a 96-well CometChip for conducting the comet assay according to the manufacture’s protocol. Comet images were acquired automatically using the Cytation 5 (BioTek, Winooski, VT) and analyzed using Trevigen Comet Analysis Software. DNA damage effects were expressed as the percentage of tail DNA.

2.7. Quantification of DNA damage concentration-responses

The PMH comet data were quantified using PROASTweb (version 70.1), which was developed by the Dutch National Institute for Public Health and the Environment (RIVM). The BMD value, along with its lower and upper bounds of the 90 % CIs, BMDU and BMDL, were calculated simultaneously at the defined critical effect size (CES) or benchmark response (BMR). The present study calculated BMD₅₀ values, with the rationale that 0.6 (60 % increase in response over vehicle control, BMD₆₀) has been proposed as a suitable CES for evaluating in vivo rat liver comet data (Zeller et al., 2016) and that the BMD₅₀ is commonly used in BMD quantification studies. The BMD₅₀ values were calculated using exponential and Hill models and values from the exponential model were used for potency ranking in the present study. The PHH comet data from our previous study (Seo et al., 2020) were reanalyzed using PROAST and resulting BMD₅₀ values were compared with those from the PMH comet data.

2.8. Statistical analysis

Comet data were averaged from at least three independent experiments using hepatocytes isolated from different animals and presented as the mean ± standard deviation (SD). The lowest effective concentration (LEC) and pairwise comparisons between treatment groups and the vehicle control were determined using one-way analysis of variance (ANOVA) followed by Dunnett’s test. The LEC was defined as the lowest concentration giving a positive response in the comet assay. Statistical significance was determined when p < 0.05 using SigmaPlot 13.0 (Systat Software, San Jose, CA). Correlation of BMD₅₀ values between PMHs and PHHs was evaluated by Spearman’s rank correlation coefficient using GraphPad Prism version 6.04 (GraphPad Software, La Jolla, CA, USA).

3. Results

3.1. Cell viability and morphology

High quality PMHs were isolated from seven out of eight animals in the present study. After percoll purification, the viability of freshly isolated hepatocytes from the seven animals ranged from 75 to 95 %, with an average viability of 87.4 ± 6.8 %. Hepatocytes isolated from one primate (#4), which had an enlarged liver and scars in the left lateral lobe of the liver, displayed low cell viability (38 %) and were excluded from toxicity assays.

Cells were attached to the collagen-I-coated 96-well plates approximately 3 h after plating (Fig. 1). Live and attached primary hepatocytes exhibited a polygonal or typical hexagonal shape after a 4-h incubation, with nonviable cells observed on the top of the monolayer. After a 24-h incubation, cells appeared to have flattened morphology, with clearly outlined membranous boundaries. Most cells were single or binucleated and contained dense cytoplasmic granules, but multi-nucleated cells were also occasionally seen in the plate.

Fig. 1. Cell morphology of primary macaque hepatocytes (PMHs) after plating.

Light micrograph illustrates PMHs after 0 h (A), 4 h (B), and 24 h (C) in culture (200 × magnification).

3.2. CYP activities in PMHs derived from macaques

The basal activities of six major CYPs (CYP1A2, 2B6, 2C9, 2D6, 2E1, and 3A4) were examined (Fig. 2). Although variations were seen between animals, all NHPs consistently showed high levels of activities of CYP1A2, 2B6, 2D6, and 3A4, with averages between 64–93 pmol/min/mg protein. The activities of CYP2C9 and 2E1 in the 8 animals were less than 1 pmol/min/mg protein, which were significantly lower than those of the other four CYPs tested. No CYP3A4 activity could be detected in the PMHs of primate #2 likely due to technical issues.

Fig. 2. CYP450 activity in PMHs isolated from eight rhesus macaques.

Six major CYP450 activities (CYP1A2, 2B6, 2C9, 2D6, 2E1, and 3A4) were measured 24 h after plating. CYP450 activities are expressed as pmol/min/mg protein and the data are presented as the mean ± SD (n ≥ 3). PHHs and HepaRG cells data are from our previous study (Seo et al., 2020). < LOD, below the limit of detection; NT, not tested.

The CYP activities in PMHs also were compared with those in human hepatic cells, i.e., PHHs and differentiated HepaRG cells (Fig. 2). CYP3A4 activity was comparable among the three types of liver cells. Interestingly, PMHs had much higher activities of CYP1A2, 2B6, and 2D6 than those in PHHs and HepaRG cells (with an average of 64–85 vs. 3–14 pmol/min/mg protein). Conversely, the CYP2C9 activity of PMHs was markedly lower than those in PHHs and HepaRG cells (0.2 vs. 7–31 pmol/min/mg protein). PMHs had low CYP2E1 activities, with values of 0.2–1.4 pmol/min/mg protein, while the CYP2E1 activity in HepaRG cells was below the detection limits. PHHs from three donors showed 11–126 pmol/min/million cells CYP2E1 activities as reported by the supplier (Seo et al., 2020). Regrettably, CYP2E1 activity was not tested in PHHs in our previous study, and we were unable to make a direct comparison between PMHs and PHHs in the present study.

3.3. Cytotoxicity profiles of the 16 compounds

Following a 24-h exposure, all 16 compounds induced cytotoxic effects in PMHs as measured by the ATP assay (Table 2). The maximum concentration of each chemical used in the CometChip assay was defined as the highest concentration producing cell viability of ≥ 70 % or 10 mM when no cytotoxicity was observed (OECD, 2015). Nine out of the 16 compounds induced the same cytotoxicity in PMHs and PHHs, as evaluated by the maximum concentration and cell viability. CPA displayed the most distinct cytotoxic effects between the two types of primary hepatocytes, with cell viability of 50 % at 2500 μM in PMHs and 88 % at 10,000 μM in PHHs. Three compounds, DMBA, ENU, and o-anthranilic acid, had cytotoxic effects over the same range of concentrations, but only in PMHs. Cisplatin (50 μM) and HQ (200 μM) were slightly more cytotoxic in PHHs, with a cell viability of 78–81 % in PMHs vs. 55–63 % in PHHs. CdCl₂ at 4 μM induced slightly higher cytotoxicity (43 %) in PMHs than in PHHs (25 %).

Table 2.

Comparison of cytotoxicity and DNA damage responses between primary macaque (PMH) and human (PHH) hepatocytes.

Group	Chemical	Max. con. (μM)		LECa (μM)		Cytotoxicityb		DNA damagec
		PMH	PHH	PMH	PHH	PMH	PHH	PMH	PHH
Indirect-acting genotoxicants/carcinogens	2,4-DAT	8000	8000	8000	1414	+	+	++	++
	B[a]P	100	100	37.5	6.3	+	+	+++	++
	CPA	5000	10,000	937.5	1768	+++	+	+++	++
	DMBA	500	250	31.25	62.5	+	−	+++	++
Direct-acting genotoxicants/carcinogens	4-NQO	7.5	5	7.5	2.5	+	+	++	++
	CdCl2	4	4	2	1	++	+	+++	+
	Cisplatin	50	50	9.4	6.3	+	++	++	++
	Colchicine	4	4	−	−	+	+	−	−
	ENU	3200	3200	2400	800	+	−	+++	+++
	HQ	200	200	−	−	+	++	−	−
	MMS	500	500	125	62.5	+	+	+++	+++
Non-carcinogens that are negative or equivocal for genotoxicity in vivo	2-Ethyl-1,3-hexanediol	10,000	10,000	−	−	+	+	−	−
	o-Anthranilic acid	5000	5000	−	−	++	−	−	−
	Curcumin	40	40	−	−	+	+	−	−
	Ethionamide	2000	2000	−	−	+	+	−	−
	Resorcinol	1000	2000	−	−	++	++	−	−

PHH data are from our previous study (Seo et al., 2020).

^aLEC, the lowest effective concentration, determined by one-way ANOVA followed by Dunnett’s test, is the lowest concentration that gives a positive response in the comet assay.

^bCytotoxicity was determined by the ATP assay following a 24-h treatment. −, cytotoxicity of 0–10 %; +, 11–30 %; ++, 31–50 %; +++, > 50 %.

^cDNA damage was expressed as the relative ratio calculated by comparing the %DNA in tail at the highest concentration to those of the vehicle control. −, The ratio ≤ 1.5-fold and p ≥ 0.05; +, 1.5 < ratio ≤ 2, and p < 0.05; ++, 2 < ratio ≤ 5; +++, the ratio > 5.

3.4. DNA damage profiles of the 16 compounds

Although different levels of CYP activities and cytotoxicity were seen in PMHs and PHHs, both cell types demonstrated the same positive/negative calls for all 16 compounds tested, whether or not the compounds required metabolic activation (Table 2, Fig. 3). Overall, PMHs induced higher percentages of tail DNA in the comet assay than PHHs for four out of the nine positive compounds. At the highest acceptable concentration, B[a]P, CPA, DMBA, and CdCl₂ induced 6.1–8.1-fold and 1.8–4.4-fold increases in % tail DNA (DNA damage) in PMHs and PHHs, respectively, compared to the vehicle control. Interestingly, out of the nine positive compounds, the LECs of seven compounds were 1.5–6-fold higher in PMHs than those in PHHs, while the LEC values for two compounds (CPA and DMBA) were 2-fold higher in PHHs than in PMHs. Both PHHs and PMHs also showed a 100 % specificity for detecting the five non-carcinogens that are negative or equivocal for genotoxicity in vivo (Fig. 3C).

Fig. 3. DNA damage and cytotoxicity of 16 compounds in PMHs.

PMHs were exposed to four indirect-acting (A), seven direct-acting genotoxicants or carcinogens (B), and five non-carcinogens that are negative or equivocal for genotoxicity in vivo (C) for 24 h. The relative cell viability (% of control, indicating cytotoxicity) was measured by the ATP assay (right y-axis and red line) and DNA damage (% tail DNA; left y-axis and black bar) was detected using the CometChip assay. The data are expressed as the mean ± SD (n ≥ 3). *p < 0.05, **p < 0.01, and ***p < 0.001 vs. control. See Table 1 for abbreviations of the compounds tested.

3.5. BMD modeling on DNA damage responses from nine positive compounds in PMHs

By using a PROAST covariate approach, the nine concentration-responses were roughly divided into six groups having non-overlapping BMDLs and BMDUs (Fig. 4A). CdCl₂, showed the lowest BMD₅₀ value of 0.37 μM (Table 3), indicating it was the most potent DNA damage inducer in PMH among the nine compounds. Three compounds (4-NQO, B[a]P, and DMBA) had BMD₅₀ values of 3.5–5.5 μM; MMS and cisplatin had similar BMD₅₀ values of 21–22 μM; while CPA, ENU, and 2,4-DAT had the highest BMD₅₀ values of 253, 683, and 2831 μM in PMHs, respectively. The width of the CI, i.e., the BMDU/BMDL ratio, represents the uncertainty in the BMD estimate and reflects the quality of the concentration-response data (Slob, 2014; Wills et al., 2017). The ratios of the BMDU to the BMDL in the present study were between 2.3–3.2 (Table 3), indicating relatively high precision for the BMD estimates.

Fig. 4.

Comparison of BMD₅₀ and its upper and lower 90 % confidence intervals (BMDU and BMDL) of chemical-induced DNA damages between PMHs and PHHs. The DNA damage concentration-response data in PHHs were obtained from our previous study (Seo et al., 2020). BMD and its BMDU and BMDL were calculated from PMH (A) and PHH (B) comet data using exponential (upper line) and Hill (lower line) models of PROAST. (C) The bars represent the range between BMDUs and BMDLs and are used to differentiate between responses based on non-overlapping confidence intervals (CIs). Red, PHHs; Blue, PMHs. (D) Correlation of BMD₅₀ values between PMHs and PHHs is illustrated by BMD and its CIs for PMH comet data against those of PHHs. See Table 1 for abbreviations of the compounds tested.

Table 3.

Comparison of the benchmark dose (BMD₅₀) and potency ranking between primary macaque and human hepatocytes.^a

Chemical	Primary macque hepatocytes			Primary human hepatocytes
	BMD50 (μM)	BMDL50–BMDU50 (μM)	BMDU/BMDL	BMD50 (μM)	BMDL50–BMDU50 (μM)	BMDU/BMDL
2,4-DAT	2831 (9)b	1790–4660	2.6	880.6 (9)b	489–1580	3.2
4-NQO	3.5 (2)	2.2–6.3	2.9	0.3 (1)	0.1–0.6	4.4
B[a]P	5.5 (4)	3.4–8.5	2.5	9.2 (5)	5.2–16.3	3.2
CdCl2	0.4 (1)	0.2–0.5	2.3	0.9 (2)	0.5–1.6	3.2
Cisplatin	21.3 (5)	12.4–39.2	3.2	2.5 (4)	1.3–5.3	4.2
CPA	253.8 (7)	165–372	2.3	248.2 (8)	135–437	3.2
DMBA	4.0 (3)	2.6–6.1	2.4	15.1 (6)	8.5–25.8	3.0
ENU	683 (8)	439–1030	2.3	47.3 (7)	25.3–83.5	3.3
MMS	22.3 (6)	14.2–33.7	2.4	1.0 (3)	0.5–1.9	3.5

Primary human hepatocyte data are from our previous study (Seo et al., 2020).

^aBMDs for the CometChip data were calculated using PROAST.

^bPotency ranking.

3.6. Comparison of BMD values between PMHs and PHHs

The BMD values and upper and lower 90 % CIs for each test compound also were used for the comparative evaluation of DNA damage potency in PMHs and PHHs. For both PMHs and PHHs, CdCl₂ and 4-NQO were the most potent, and CPA, ENU, and 2,4-DAT were the least potent DNA damage inducers (Table 3 and Fig. 4A and B). DMBA was more potent (ranked #3) in PMHs than in PHHs (ranked #6), while MMS was more potent in PHHs, with a BMD₅₀ value 22-fold lower than that in PMHs.

Among the four indirect-acting compounds, the BMD₅₀ CIs for both B [a]P and CPA tested in PMHs and PHHs overlapped with each other (Fig. 4C). The BMD₅₀ for 2,4-DAT was 3.2-fold higher in PMHs than in PHHs, while DMBA had a 3.8-fold higher BMD₅₀ value in PHHs than in PMHs (Table 3). Four out of the five positive direct-acting compounds, 4-NQO, cisplatin, ENU, and MMS, had significantly (8–22-fold) lower BMD₅₀ values in PHHs than in PMHs. CdCl₂ had a lower BMD₅₀ value in PMHs than in PHHs (0.37 vs. 0.89 μM), but the 90 % CIs overlapped for the DNA damage responses in the two types of primary hepatocytes.

A correlation evaluation using the Spearman rank correlation test on comet data in PHHs and PMHs showed that BMD₅₀ values had a correlation coefficient of 0.8 (p < 0.05), indicating a strong correlation between the PMH and PHH data.

4. Discussion

The phylogenetic similarity of NHP species to Homo sapiens makes them unique in their ability to mimic toxicity testing in humans. Due to their homology with humans in terms of physiology and drug metabolism, NHPs are widely used in preclinical studies for drug and antibody development (Uno et al., 2016). The present study used in situ whole liver perfusion, as was initially used in rats (Oldham et al., 1979), for isolating primary hepatocytes from rhesus macaques. In situ perfusion of rodent livers maintains liver function for up to 2 h and has better cell yield and viability compared to isolated organ perfusion (Aiken et al., 1990; Choi et al., 2019). Isolated PMHs in the present study had a relatively high average cell viability of 87.4 %. An optimal liver perfusion/digestion and normal liver morphology are likely critical factors for the successful isolation of high-quality primary hepatocytes. The hepatocytes with low viability (38 %) were obtained in the present study from one animal with an enlarged/scarred liver. However, the poor cell viability could also be due to procedural or technical issues.

CYP450 enzymes are essential for the bioactivation and/or detoxification of many xenobiotics and drugs. The present study compared the activities of six major CYPs in PMHs, PHHs, and HepaRG cells. CYP3A4/5 has been reported to be the most abundant (40 % of total) enzyme in the liver of both humans and NHPs (Gundert-Remy et al., 2014; Uehara et al., 2011), and CYP3A4 showed similar activities in the three types of cells (Fig. 2). However, significant differences in the activities of several CYPs were seen in human and rhesus macaque hepatocytes, despite CYP450 enzymes displaying 95 % homology between humans and NHPs (Martignoni et al., 2006). Interindividual variability in CYP450 activities also was observed in the present study even though the 8 animals were from the same species (rhesus macaque) and were all males. Regardless of the varying CYP450 activities and cytotoxic effects, PMHs in the present study displayed identical sensitivity and specificity for detecting the eleven genotoxicants/carcinogens and five non-carcinogens that are negative or equivocal for genotoxicity in vivo (Fig. 2 and Table 2). These results demonstrate the usefulness of PMHs as a surrogate to PHHs for predicting genotoxic risks for the xenobiotics humans are likely to be exposed to.

As genetic toxicology is moving from a qualitative, “screen and bin” binary approach to quantitative dose-response analysis and risk assessment (White and Johnson, 2016), we conducted a quantitative comparison of DNA damage potency for the-comet-assay-positive compounds between PMHs and PHHs. BMD₅₀ values were estimated using PROAST software, with covariate analysis used to improve the precision for DNA damage potency ranking of the nine concentration-response relationships (Wills et al., 2016). Our data showed that seven out of nine compounds had a similar rank order of potency in the two types of primary hepatocytes, regardless of requiring metabolic activation or not. Thus, CdCl₂ and 4-NQO ranked #1–2, B[a]P and cisplatin ranked #4–5, and CPA, ENU, and 2,4-DAT ranked #7–9 (Table 3). In contrast, DMBA, a polycyclic aromatic hydrocarbon mainly metabolized by CYP1B1/1A1 (Lin et al., 2012), showed a higher DNA damage potency ranking in PMHs (#3), with a BMD₅₀ value 3.75-fold lower (i.e., more potent) compared to PHHs (ranked #6). CYP1A1, 1A2, and 1B1 are three highly conserved members of the CYP1 family. We didn’t measure the activities of CYP1A1 and CYP1B1 in the present study because both are expressed predominantly in extrahepatic tissues, with very low levels of expression in the liver of human and other species including NHPs. In contract, CYP1A2 is mainly expressed in the liver (Wang et al., 2016). Our results indicate that PMHs isolated from the eight NHPs all had significantly higher CYP1A2 activities compared to PHHs from three human donors (Fig. 2). Given that CYP1A2 appeared as a gene duplication event from CYP1A1 and ancestors of the CYP1A and CYP1B subfamily diverged from one another >400 million years ago (Jorge-Nebert et al., 2010), we speculate that PMHs may possess higher levels of CYP1B1 and 1A1 activities than PHHs. Consequently, more mutagenic metabolites of DMBA might be produced in PMHs than in PHHs.

To further evaluate the appropriateness of PMHs for predicting human genotoxicity risk, we compared BMD₅₀ values and their CIs derived from the CometChip data for PMHs with those for PHHs. These comparisons led to several interesting findings. First, the DNA damage responses in PMHs and PHHs for two out of four (50 %) indirect-acting compounds had overlapping BMD₅₀ upper and lower CIs, while only for one out of the five (20 %) direct-acting compounds had overlapping BMD₅₀ CIs for DNA damage (Fig. 4C). Second, for the six compounds that had non-overlapping BMD₅₀ CIs, the four direct-acting compounds (4-NQO, cisplatin, ENU, and MMS) had distinctly different BMD₅₀ values in PMHs and PHHs compared to the two indirect-acting compounds (2,4-DAT and DMBA). These observations demonstrate that primary hepatocytes isolated from rhesus macaques can detect indirect-acting human genotoxic compounds both qualitatively and quantitatively, although the CYP activity levels varied between PMHs and PHHs. Third, all but one compound (DMBA) that had non-overlapping BMD₅₀ CIs had more conservative BMD₅₀ values in PHHs than in PMHs. In our previous study, 4-NQO showed remarkably higher DNA damage potency in PHHs from one donor compared to the other two donors (Seo et al., 2020). Even though we assume that the significant difference in BMD₅₀ values between the two types of hepatocytes is due to inter-species differences rather than intraspecies individual variability, based on the observation that the BMD₅₀ of 4-NQO was 12-fold higher in PMHs than in PHHs (Table 3). It is worth noting that the PMHs used in the present study were freshly isolated from the liver of euthanized animals, while PHHs were cryopreserved for several years before they were used for these studies. This freezing and thawing process may have resulted in slightly compromised cell viability and higher background DNA damage values (Gajski et al., 2020). Cryopreservation of primary hepatocytes provides constantly available resources for studying in vitro drug metabolism and evaluating the toxicity of various drugs intended for human use (Hengstler et al., 2000). The present study has optimized the conditions for cryopreserving freshly isolated PMHs and stored them in liquid nitrogen for future use. However, further studies are required to validate cryopreservation techniques by comparing metabolic functions and the sensitivity for detecting genotoxic compounds between fresh and frozen PMHs.

In conclusion, by comparing DNA damage responses between macaque and human primary hepatocytes qualitatively and quantitatively, our results demonstrated that PMHs had the same sensitivity and specificity as PHHs in evaluating the genotoxicity of 16 compounds, indicating that PMHs may be an appropriate surrogate for PHHs in hazard identification studies. However, despite the fact that NHPs are the closest relatives to human, quantitative differences were seen between PMHs and PHHs regarding CYP450 enzyme activities as well as DNA damage responses. Differences in DNA damage responses were especially notable for direct-acting genotoxicants/carcinogens. Currently, animal genotoxicity and carcinogenicity data are used by regulatory bodies to derive point-of-departure metrics; and cross-species extrapolation is used to determine health-based guidance values that specify a maximum human exposure, e.g., acceptable daily intake or permitted daily exposure (White et al., 2020). Our results demonstrate the existence of differences in toxicity, even for closely related primate species. However, given the close phylogenetic relationship between NHPs and humans and the strong correlation of the BMD₅₀ values between the PMH and PHH comet data, our data support the use of PMHs as a surrogate for PHHs for evaluating the potential genotoxic hazard of compounds in humans.