Intra- and inter-species variability in urinary N7-(1-hydroxy-3-buten-2-yl) guanine (EB-GII) adducts following inhalation exposure to 1,3-butadiene

Abstract

1,3-Butadiene is a known carcinogen primarily targeting lymphoid tissues, lung, and liver. Cytochrome P450 activates butadiene to epoxides which form covalent DNA adducts that are thought to be a key mechanistic event in cancer. Previous studies suggested that inter-species, -tissue, and -individual susceptibility to adverse health effects of butadiene exposure may be due to differences in metabolism, as well as other mechanisms. In this study, we aimed to examine the extent of inter-individual and inter-species variability in urinary N7-(1-hydroxy-3-buten-2-yl) guanine (EB-GII) DNA adducts, a well-known biomarker of exposure to butadiene. For a population variability study in mice, we used the Collaborative Cross model. Female and male mice from 5 strains were exposed to filtered air or butadiene (590 ppm, 6 h/day, 5 days/week for 2 weeks) by inhalation. Urine samples were collected, and the metabolic activation of butadiene to DNA-reactive species was quantified as urinary EB-GII adducts. We quantified the degree of EB-GII variation across mouse strains and sexes; then, we compared this variation with the data from rats (exposed to 62.5 or 200 ppm butadiene) and humans (0.004-2.2 ppm butadiene). We show that sex and strain are significant contributors to the variability in urinary EB-GII levels in mice. In addition, we find that the degree of variability in urinary EB-GII in Collaborative Cross mice, when expressed as uncertainty factor for the inter-individual variability (UF_H), is relatively modest (≤ 3-fold), possibly due to metabolic saturation. By contrast, variability in urinary EB-GII (adjusted for exposure) observed in humans, while larger than the default value of 10-fold, is largely consistent with UF_H estimates for other chemicals based on human data for non-cancer endpoints. Overall, these data demonstrate that urinary EB-GII levels, particularly from human studies, may be useful for quantitative characterization of human variability in the cancer risks to butadiene.

Graphical Abstract

Introduction

1,3-Butadiene is a known human carcinogen present in tobacco smoke and also found in automobile emissions, smoke from wood burning, occupational exposures in synthetic polymer manufacturing, and in urban air¹. In laboratory animals, butadiene is a multisite carcinogen and has been shown to induce lung tumors in mice at exposures as low as 6.25 ppm². Human occupational exposure to butadiene is associated with an increased incidence of lymphatic and hemopoietic cancers³. Widespread exposure of human populations to butadiene in tobacco smoke and urban air, as well as the risk for butadiene-associated increased cancer incidence, supports additional studies into the factors accounting for an individual’s risk to butadiene-induced cancer^{4, 5} and for improving scientific justification for uncertainty factors used in deriving cancer risk estimates for butadiene.

Upon exposure, cytochrome P450 monooxygenases metabolize butadiene to the reactive epoxides 3,4-epoxy-1-butene (EB), 1,2,3,4-diepoxybutane (DEB) and 1,2-dihydroxy-3,4-epoxybutane (EBD). Glutathione S-transferases (specifically glutathione S-transferase theta 1 GSTT1) detoxify butadiene epoxides to generate mercapturic acids that are excreted in urine and serve as common biomarkers of butadiene exposure⁶. While butadiene-derived mercapturic acids represent the metabolically inactivated fraction of butadiene⁷, the DNA adducts derived from this pathway serve as important mechanism-based biomarkers of exposure. For example, EB alkylates the N7 position of guanine and forms N7-(1-hydroxy-3-buten-2-yl) guanine (EB-GII) adducts^{4, 8}.

The levels of butadiene-derived DNA adducts represent the formation of electrophilic intermediates that react with nucleophilic DNA bases and other biomolecules. These adducts may be used as the indicator of the biologically active dose available for DNA binding^{9, 10}. While DEB-dependent DNA-DNA crosslinks are considered to be most relevant adducts for butadiene carcinogenicity because they may lead to DNA polymerase errors during replication^{11, 12}, EB-GII adducts are also widely used as biomarkers for studies of butadiene exposure-associated risk¹³. EB-GII adducts are hydrolytically labile, released through spontaneous depurination or actively by base excision repair, and are excreted in urine; this adduct has been previously detected in urine of smokers and occupationally-exposed workers¹⁴, as well as in exposed rats¹⁴. Previous studies demonstrated that urinary EB-GII adducts are stable in human subjects over time^{5, 14, 15}. These studies also revealed that spot urine EB-GII amounts are associated with lung cancer risk⁵.

Human health assessments need to characterize the degree of inter-individual variability in toxic effects of chemicals; however, toxicological data from studies conducted in diverse populations to evaluate inter-individual variability are generally lacking¹⁶. The Collaborative Cross mouse population¹⁷ has been used for both mechanistic toxicology studies¹⁸ and to derive quantitative estimates of variability in effects of a number of chemicals^19-22, including butadiene²³. This genetically diverse mouse population captures greater than 90% of the known mouse genetic variability and was developed to be on par with the extent of genetic heterogeneity observed in the human population²⁴. This population-wide experimental animal model also allows studies of the determinants of interindividual variability in response to toxicants²⁵, as well as quantification of the extent of variability in chemical exposures to inform human health assessments¹⁸.

The goal of this study was to evaluate the extent of inter-individual variability in urinary butadiene-DNA adducts. Mice and humans similarly metabolize butadiene to DNA-reactive intermediates, and the study of genotoxic metabolites can reveal whether the degree of inter-individual variability is concordant between mice and humans. To accomplish this goal, we exposed male and female mice from 5 Collaborative Cross strains to clean air or 590 ppm butadiene by inhalation. The amounts of urinary EB-GII adducts were examined to evaluate strain and sex differences, and were compared to published data from rats (exposed to 62.5 or 200 ppm butadiene) and humans (0.004-2.2 ppm butadiene). Also, the degree of variability in urinary EB-GII levels was quantified as uncertainty factor for the inter-individual variability (UF_H) using data from mice and humans.

Experimental

Chemicals

Chemicals and solvents including LC-MS grade water, methanol, acetonitrile were acquired from Fisher Scientific (Pittsburgh, PA). Synthetic urine, a non-biological material that contains proprietary components that mimic human urine, was purchased from Dyna-Tek Industries (Surine™ 720, Dyna-Tek Industries, Lenexa, KS). The reverse phase Strata X polymeric SPE cartridges (30 mg/1 mL) were acquired from Phenomenex (Torrance, CA). MS vials (300 μL) were acquired from ChromTech (Apple Valley, MN). EB-GII and [¹⁵N₅]EB-GII standards were synthesized in our laboratory as reported previously^{26, 27}. DNA adduct standard stock solutions were prepared in water and stored at −20°C.

Animals and Exposures

Adult male and female mice (~8-12 weeks old on arrival) from five Collaborative Cross (CC) strains (CC026, CC027, CC043, CC049, CC079) were acquired from the University of North Carolina Systems Genetics Core (Chapel Hill, NC). Strains for these experiments were selected based on the availability of mice from the vendor in terms of the number of animals from both sexes that were available in close age range. Female mice were housed with littermates (2-4 per cage, separated by strain) in sterilized cages in a temperature-controlled room (24°C) with 12-hour light/dark cycling within the Texas A&M Institute for Genomic Medicine animal facility. Male mice were individually housed within identical conditions. Mice were provided standard rodent chow (Global 2019 Extruded (irradiated format) rodent diet, Teklad Diets, Madison, WI) and purified water ad libitum. Mice were held in the facility for 30-45 days prior to experiments to allow for post-transportation quarantine and acclimation. All animals were cleared to be pathogen-free before experiments. Prior to initiation of the exposures, animals were weighed and randomized (five mice per group) into three exposure groups (filtered air and two 1,3-butadiene exposure groups) to balance their initial body weight within the same strain and sex (Supplemental Table S1).

The experiment consisted of whole-body exposures to either filtered air or 1,3-butadiene (target concentration 625 ppm; final average concentration 590±150 ppm) for 6 h/day for 5 days/week for two consecutive weeks as detailed elsewhere²⁸. This concentration is within the range of concentrations (6.25–1,250 ppm) that have been used in many previous sub-acute to chronic studies of inhalational exposures in mice, and reported to cause tumors in the lung (among other tissues) of 1,3-butadiene–exposed B6C3F1 mice in chronic inhalation studies¹. Furthermore, 1,3-butadiene is known to have a supralinear exposure–response relationship for DNA damage phenotypes in mice for concentrations up to 625 ppm¹, indicating that the selected exposure concentration was relevant for the study of potential mechanisms of, and variability in responses to, 1,3-butadiene–induced tumorigenicity. This concentration is also relevant to the past occupational human exposures to 1,3-butadiene that were reported to be in the range from tens to hundreds of ppm²⁹. This study’s duration was consistent with other inhalation studies of 1,3-butadiene in mice that investigated DNA damage and epigenetic biomarkers^{28, 30, 31}.

Specifically, animals were placed individually into 2.5 cm diameter and 10 cm long cylindrical stainless-steel mesh holders that were then arranged in the gas exposure chambers in a manner that would not impede the air flow. Chambers were sealed, and butadiene exposures were initiated through air flow controllers, which were adjusted as needed to maintain target butadiene concentrations during exposures. Chamber air samples were taken every 10-15 min throughout 6 h exposure periods and analyzed using gas chromatography (Model 310 Gas Chromatograph-FID, AC, 3’ HAYSEP D, SRI Instruments, Torrance, CA). Exposures were conducted typically between 8:00 and 14:00 each workday. At the end of each exposure period, exposure chambers were flushed with filtered air for 30 min, and the animals were placed in the standard cages individually (to track their body weight and exposure group assignment) with ad libitum access to the standard diet and purified water as detailed above. To alleviate stress of environmental changes, all animals were provided with cotton mesh pads and the stainless-steel exposure tube was placed into their cage overnight. Exposures were conducted separately for female and male mice on consecutive 2-week periods. To accommodate all exposed animals, two 1,3-butadiene exposure chambers were used; therefore, to prevent differences in exposures between chambers, animals were placed in alternate exposure chambers on consecutive days. All mice were weighed twice a week in the morning before exposures. Animal health was monitored multiple times per day before and during exposure. Moribund animals were sacrificed with euthasol upon consultation with staff veterinarian in the animal facility. All experimental procedures involving animals and their husbandry were approved by the Institutional Animal Care and Use Committee (IACUC) of Texas A&M University.

Urine collection was performed from randomly-selected animals using mouse metabolic cages (32×25×36.5 cm; Techniplast USA, West Chester, PA). At the end of the exposure on the third day of each 5-day exposure period, animals were placed into metabolic cages (one mouse per metabolic cage) overnight with free access to food and water. Metabolic cage racks were placed in the same room where all animals were housed to ensure same temperature and light/dark exposure. Next morning, animals were removed from the metabolic cages and placed into exposure chambers and at the end of exposure returned to their standard housing cage. The total urine void from each metabolic cage was collected, placed into microcentrifuge tubes and stored at −80°C until assayed. The same animals were used for urine collection on consecutive exposure weeks.

Creatinine Quantification

Urine samples were allowed to thaw at room temperature before assay. Quantitative determination of creatinine in all urine samples was performed using of the Parameter™ Creatinine Assay Kit (R&D Systems, Catalog: KGE005) as detailed by the manufacturer. Briefly, urine samples were diluted 20-fold in deionized water. An alkaline picrate solution (1:5 NaOH to picric acid) and a seven-fold dilution series of standard were prepared. Within the kit-provided plate, 50 μL of creatinine standards, control (provided within Parameter™ kit), or urine samples were added to wells along with 100 μL of alkaline picrate solution. Plates were incubated at room temperature for 30 minutes before being processed using a plate absorbance reader (Molecular Devices, San Jose, CA; absorbance at 490 nm). Replicate readings for creatinine standards, control samples, and urine samples were averaged before the average zero standard optical density was subtracted (i.e., for sample noise removal). A standard curve was created by plotting the mean absorbance for each Parameter™ standard against the standard concentration. Concentration reads from the generated standard curve were multiplied by the dilution factor. Urine data was linearized by plotting creatinine concentrations (log-form) versus the log of the optical density. A best-fit line was generated by regression analysis.

EB-GII adduct enrichment from urine

Urine samples were first centrifuged at 10,000×g for 15 min to remove particulate matter. Urine samples were diluted 100× with LC-MS grade water. Diluted urine (1%) samples were spiked with [¹⁵N₅]EB-GII (5 fmol, internal standard for mass spectrometry). EB-GII and [¹⁵N₅]EB-GII were purified by solid phase extraction (SPE) and offline HPLC as previously described ¹⁴. HPLC blanks containing only [¹⁵N₅]EB-GII standard were injected in the beginning of each sequence and after every 3 runs to monitor analyte carryover. Offline HPLC fractions were dried under vacuum, resuspended in 0.01% acetic acid in LC-MS grade water (16 μL) processed using nanoLC-ESI⁺-MS/MS as described below¹⁴.

NanoLC/ESI⁺-MS/MS analysis of urinary EB-GII

NanoLC/ESI⁺-MS/MS experiments was performed using a nano-LC HPLC system (Dionex, Sunnyvale, CA) interfaced with an LTQ Quantiva instrument (Thermo Fisher Scientific, Waltham, MA). The HPLC was equipped with a nano-LC column (0.075 x 200 mm) manually packed with Synergi Hydro-RP 80Å (4 μm) chromatographic packing (Phenomenex, Torrance, CA). Solvent A was LC-MS grade water containing 0.01% acetic acid and solvent B was LC-MS grade acetonitrile containing 0.02% acetic acid. Solvent gradient was initially maintained at 0.8 μL/min for 7.5 min at 2% B before reducing the flow rate to 0.3 μL/min for 0.5 min. Solvent B was linearly increased to 25% B for 9 min and then to 50% B in 10 min. Solvent B was then reduced to 2%B in 3 min before the flow rate was increased to 0.8 μL/min in 1 min. The column was equilibrated for an additional 6 min. In this method, EB-GII and [¹⁵N₅]EB-GII elute at 16.8 min.

The TSQ Quantiva triple quadrupole mass spectrometer was operated in the positive ion mode using a spray voltage of 2500V and a capillary temperature of 375°C. These MS parameters were optimized upon infusion of authentic EB-GII solution for maximum sensitivity. Quantitative analysis of EB-GII was performed using selected reaction monitoring mode (SRM). The MS/MS transitions of m/z 222.1 [M + H]⁺ to m/z 151.8 [Gua + H]⁺ and m/z 227.1 [M + H]⁺ to m/z 156.8 [¹⁵N₅-Gua + H]⁺ were monitored for EB-GII and [¹⁵N₅]EB-GII respectively. MS/MS fragmentation was accomplished using collision induced dissociation in Q2, using argon as a collision gas (1.5 mTorr) and a collision energy of 13.8 V. The peak width for Q1 and Q3 were 0.4 amu and 0.7 amu respectively. Quantitation of nanoLC-ESI⁺-MS/MS data was accomplished by comparing the peak areas from the extracted ion chromatograms which corresponded to EB-GII and [¹⁵N₅]EB-GII and fitting to calibration curves prepared with authentic EB-GII and [¹⁵N₆]EB-GII standards as described previously¹⁴. Adduct levels were normalized to urinary creatinine levels.

Variability Analyses and Calculation of Uncertainty Factors

Calculation of the estimates of inter-individual or within-strain variability largely followed the procedures detailed elsewhere²³. As the numerical distribution of adduct values showed strong positive skew, the analyses were conducted on the (natural) logarithmic scale. Accordingly, geometric means and standard deviations were deemed to be most appropriate measures of scale and location on the original scale, with confidence intervals computed following normal assumptions on the log scale ³² and exponentiated. Log-transformed adduct amounts were converted to residual values after correction for covariates. In the human data^{5, 14} the covariates were sex, measured exposure, and pack-years of smoking. Errors in covariates or unobserved sources of variation would tend to increase the estimated total and across-individual variability, leading to conservative (protective) uncertainty factors. From the total variability estimates, intra-class correlation as estimated in the original publication³³, was multiplied by total variance to obtain the within-individual estimated variance. For the rodent datasets (Ref.¹⁴ and this study), sex was used as a covariate, or analyses were conducted separately for males and females. The rat dataset consisted of a single strain, so no population variability measure was computed. For the mouse dataset, linear regression with CC strain as a factor variable was used to decompose the total variance into the within-strain (error) and across-strain (strain effect) portions of the residual variance. Uncertainty factor high (UFH) values were calculated as exp(zσ) values for 95% and 99% standard normal z-quantiles and σ estimated as the square root of the across-individual/strain variance. Confidence intervals were based on standard χ² distributional assumptions for the variance estimate and appropriate degrees of freedom from the linear modeling.

Statistical Analysis

Statistical significance of the differences in means of different groups was compared using either unpaired two-tailed t-test (for two-group comparisons), or two-way ANOVA with Šídák's multiple comparisons test (for comparisons that involved two factors). Analyses of variance for changes in body weight due to treatment, strain, and their interaction were conducted on the differences in post vs pre-treatment weight. P < 0.05 was considered significant for all tests. In each dataset, a test for outliers was conducted using robust regression and outlier removal (ROUT) method based on the False Discovery Rate of 1%³⁴. Statistical calculations and plots were made using GraphPad Prism (v. 9.0; San Diego, CA).

Results

Metabolism of 1,3-butadiene through cytochrome P450-mediated oxidation results in the formation of reactive epoxides which form covalent DNA adducts; of these adducts, EB-GII has been identified as a urinary biomarker of exposure to butadiene¹³. In this study, EB-GII adducts were evaluated in urine of male and female Collaborative Cross mice from 5 strains. To enable detection and quantification of EB-GII concentrations in small-volume urine samples collected from mice exposed to 1,3-butadiene (590 ppm), a sensitive and specific isotope-dilute nanoLC/ESI⁺-MS/MS approach¹⁴ was used (Figure 1).

Figure 1.

Overall study design. Mice from 5 Collaborative Cross (CC) strains were exposed to filtered air or 1,3-butadiene (590 ppm, 6 hr/day, 5 days/week for 2 weeks) for two consecutive weeks and urine samples collected during exposure. Urinary levels of EB-GII were quantified using high-throughput nanoLC/ESI⁺-MS/MS analysis. Variability in EB-GII levels were compared between CC strains and humans exposed to 1,3-butadiene occupationally¹⁴.

First, the quantitative nanoLC/ESI⁺-MS/MS method for detection of EB-GII was validated by processing and analyzing standard solutions containing synthetic urine spiked with EB-GII (0.05-10 fmol) and ¹⁵N₅-EB-GII (5.0 fmol) in triplicate (Figure 2). A strong linear relationship was observed between the actual and calculated amounts of EB-GII in the urine matrix (y = 1.036x, R² = 0.998) (Figure 2B). Analysis of the spiked samples yielded method limit of detection (LOD) and limit of quantitation (LOQ) values to be 0.05 fmol and 0.50 fmol, respectively. A representative nanoLC/ESI⁺-MS/MS ion chromatogram of EB-GII quantitation in mouse urine from a male and female mouse is shown in Figure 2C.

Figure 2.

(A) General metabolic schematic for formation of EB-GII adducts from DNA in butadiene-exposed animals. (B) NanoLC/ESI⁺-MS/MS method validation: correlation between the spiked and the observed amounts of EB-GII spiked into 0.01 μL of synthetic urine. Spiked amounts were 0, 0.1, 0.5, 1, 5, or 10 fmol of EB-GII and 5 fmol of ¹⁵N₅-EB-GII (internal standard), followed by sample processing and MS analysis. (C) Representative nanoLC/ESI⁺-MS/MS ion chromatograms of EB-GII in urine of male and female mice of one mouse strain (CC027) exposed to 1,3-butadiene by inhalation (590 ppm). Top and bottom panels show extracted ion chromatogram MS2 spectra from EB-GII and ¹⁵N₅-EB-GII (internal standard for quantitation), respectively. Mouse urine was spiked with ¹⁵N₅-EB-GII (5 fmol) before analyses.

Next, quantification of urinary EB-GII levels in control and butadiene exposed mice was conducted. Because of the scale of the study (multiple strains exposed to butadiene simultaneously) and to enable direct comparisons, randomly selected butadiene-exposed mice were used for urine collection in metabolic cages. To verify that butadiene exposure at 590 ppm had minimal to no adverse effect on the animals, body weight was recorded (Supplemental Table S1) and it was confirmed that in all groups but one (CC079, males) there were no effects of exposure on terminal body weight (Figure 3). However, both butadiene treatment and strain were significant factors within each sex when using two-way ANOVA analysis (Table 1). These observations match previous reports indicating no significant weight loss following sub-chronic exposure high levels of 1,3-butadiene³⁵. Urine from the exposed animals was processed by the validated nanoLC/ESI⁺-MS/MS method to evaluate EB-GII levels (Figure 2).

Figure 3.

Effects of 1,3-butadiene exposure (590 ppm, 6 hr/day, 5 days/week for 2 weeks) on body weight (g) across CC strains for female (A) and male (B) mice. Body weights shown were recorded at the end of the exposure period. See Supplemental Table S1 for measurements before and during exposures. Box and whiskers plots are shown (box is the inter-quartile range; vertical line is the median and whiskers are min-max range; individual animal’s data are shown as circles). White box and whiskers and open circles denote control (filtered air) animals; gray box and whiskers and black circles denoted treated (1,3-butadiene, 590 ppm, 6 hr/day for 10 days) animals. Asterisk (*) denotes a statistically significant (p<0.05) intra-strain difference between control and treated animals using Šídák's multiple comparisons test from two-way ANOVA analyses (see Table 1 for treatment, strain and interaction effects).

Table 1.

Statistical significance of the comparisons between experimental factors included in the analyses. Data for body weight and urinary EB-GII are included. Shown are the p-values from two-way ANOVA comparisons conducted on the differences in post vs pre-treatment weight (n.s., non-significant).

Factor	Treatment	Strain	Interaction
Body weight, g (F)	n.s.	p<0.01	n.s.
Body weight, g (M)	p<0.05	p<0.0001	n.s.
	Sex	Strain	Interaction
Urinary EB-GII, pg/mL	p<0.0001	p<0.01	n.s.

In agreement with previous studies that evaluated butadiene-associated EB-GII adducts in mouse tissues,^{23, 36} intra-strain and inter-sex variability was observed in urinary levels of EB-GII adducts (Figure 4). Within each strain, mean urinary EB-GII levels were not significantly different between female and male mice when using EB-GII concentrations (Figure 4A); however, creatinine corrected data showed significantly higher levels of EB-GII in exposed female mice from strains CC027, CC043, CC049 and CC079 (Supplemental Figure S1A). Because secretion of creatinine in male mice is twice that in female mice³⁷, and similar sex differences were found in urinary creatinine in the current study (Supplemental Figure S2), unadjusted concentrations of EB-GII adducts (pg/ml) were used for further comparisons. Irrespective of creatinine normalization, urinary levels of EB-GII were significantly higher in female mice when data from all strains were combined (Figure 4B and Supplemental Figure S1B). In addition, two-way ANOVA analysis showed that both sex and strain were significant factors when comparing urinary levels of EB-GII in exposed mice from this study (Table 1).

Figure 4.

Effects of 1,3-butadiene exposure (590 ppm, 6 hr/day, 5 days/week for 2 weeks) on urinary EB-GII levels across CC strains for male (white boxes) and female (gray boxes) mice. Box and whiskers plots are shown (box is the inter-quartile range; vertical line is the median and whiskers are min-max range; individual animal’s data are shown as circles). (A) Intra-strain variability between male (white box and whiskers) and female (grey box and whiskers) mice exposed to 1,3-butadiene. There were no statistically significant (p<0.05) intra-strain difference between sexes based on Šídák’s multiple comparisons test from two-way ANOVA analyses (see Table 1 for sex, strain and interaction effects). (B) Variability between male and female (all strains combined) mice exposed to 1,3-butadiene. Asterisk (****) denotes a statistically significant (p<0.0001) inter-sex difference (unpaired two-tailed t-test). See Supplemental Figure 1 for plots of EB-GII levels corrected for creatinine.

Furthermore, the levels of urinary EB-GII were compared across species and exposure conditions. Previous studies from humans exposed to butadiene in the workplace^{5, 14}, and inhalational exposures of inbred Fisher 344 rats¹⁴ used the same analytical method, and thus these data can be directly compared to those obtained in our study. In humans, estimated air concentrations of butadiene in the workplace varied in low- to sub-ppm range. Rats were exposed to either 62.5 or 200 ppm of butadiene via whole body inhalation in the study design identical to that we used for mice (2 weeks, 6 h/day, 5 days/week)¹⁴. In the mouse study, average exposure concentrations for butadiene were 590 ppm. Figure 5 compares urinary EB-GII levels in these studies. The data for male and female subject/animal were combined to enable species comparisons. It is noteworthy that in rats, urinary EB-GII levels were about 3 orders of magnitude higher than those in humans, even though the exposure concentrations differed by only about 2 orders of magnitude. By contrast, even though the mouse study had about 3-fold greater exposures as compared to rats, urinary EB-GII levels were almost 2 orders of magnitude greater.

Figure 5.

Species comparison of 1,3-butadiene exposure on urinary EB-GII levels in humans, rats and different strains of mice. Box and whiskers plots are shown (box is the inter-quartile range; vertical line is the median and whiskers are min-max range). Data for both sexes were combined within species and exposure groups or strains. Butadiene exposure levels for each group is shown in parentheses. Data are from either previous study in humans^{5, 14} and rats¹⁴, or this study in mice.

Finally, to translate the quantitative data on the population variability of EB-GII levels following 1,3-butadiene exposure in humans and mice into information directly useful in chemical risk assessment, each of the population-based datasets was used to calculate EB-GII-based chemical-specific uncertainty factors for human variability UF_H (Table 2). Interestingly, it was found that variation in urinary EB-GII in the CC mouse population was relatively modest after accounting for within-strain variability. All the resulting UF_H values were less than the default value of 10-fold intended to address combined toxicokinetic and toxicodynamic variability. The largest UF_H value from mouse data was for female mice at a protection level of 99% of the population, resulting in a factor of 3.08 [95% CI: 2.64-3.79]. By contrast, the value of UF_H derived from the human population data was substantially larger, at 13.6 [6.5-74.1] and 39.9 [14.1-440] for protection of 95% and 99% of the population, respectively. These values are comparable both to the typically used default factor of 10-fold, as well as with the limited available human data for other chemicals. For instance, the analysis of human population variability data by World Health Organization’s International Programme on Chemical Safety³⁸ found that at 95% protection, 90% of chemicals would be expected to have UF_H between 1.8 and 14, with the corresponding range at 99% protection being 2.2 to 42. Thus, the values derived here from human exposures to 1,3-butadiene of 14 and 40, respectively, while at the upper end of these ranges, are nonetheless consistent with human data from other chemicals.

Table 2.

Quantitation of variability in urinary EB-GII levels following 1,3-butadiene exposure in humans^{5, 14}, rats¹⁴, and mice (this study). Abbreviations: GM, geometric mean; GSD, geometric standard deviation; σ2 (total), variance of log-transformed EB-GII levels from this study; σ2 (within), variance of log-transformed EB-GII levels within subject or strain; σ2 (across), variance of log-transformed EB-GII levels across subjects or strains; HVF (95%), human variability factor for the 95^th percentile; HVF (99%), human variability factor for the 99^th percentile.

	Human (controlled for exposure, sex, and smoking)	Rat (sex as covariate)	Rat (sex as covariate)	Mouse (sex as covariate)	Mouse (Male only)	Mouse (Female only)
1,3-Butadiene exposure (ppm)	0.004-2.2	62.5	200	600	600	600
GM uncorrected (pg/mL; 95% CI)	1.04 (0.40 −2.72)	927 (607-1,414)	1,982 (1,314-2,987)	89,084 (76,814-103,315)	60,355 (52,208-69,773)	126,054 (100,527-158,063)
GSD uncorrected (pg/mL; 95% CI)	10.3 (6.2-25.7)	1.49 (1.29-2.68)	1.48 (1.28-2.61)	2.56 (2.33-2.88)	1.87 (1.71-2.11)	2.82 (2.46-3.40)
σ2 (total)	3.70	n/a	n/a	0.75	0.32	0.83
σ2 (within)	1.19	0.05	0.09	0.71	0.24	0.60
σ2 (across)	2.51	n/a	n/a	0.04	0.09	0.23
UFH (95%, CI)	13.6 (6.5-74.1)	n/a	n/a	1.36 (1.32-1.42)	1.62 (1.51-1.78)	2.21 (1.99-2.57)
UFH (99%, CI)	39.9 (14.1-440)	n/a	n/a	1.55 (1.48-1.64)	1.97 (1.79-2.26)	3.08 (2.64-3.79)

Discussion

1,3-Butadiene is a known human and rodent carcinogen found in tobacco smoke and exhibits a relatively high cancer risk when compared to other tobacco smoke carcinogens^{39, 40}. Quantitative mass spectrometry methods have been previously applied to evaluate butadiene metabolites^41-43 and DNA adducts in humans^{14, 44-46}. These highly sensitive and quantitative methods have been applied to investigate interindividual differences in butadiene metabolism^{47, 48} and to correlate these biomarkers with smoking³³ and genetic polymorphisms^{4, 7}. Population-based data derived from the CC mouse population has been previously used to investigate 1,3-butadiene toxicity and inter-strain variability in metabolism^{23, 25}. The current study was performed to to examine interindividual variability in EB-GII metabolism.

Using urinary samples from exposed CC mouse strains, the evidence for strain- and sex-specific differences in excretion of EB-GII adducts by butadiene exposed laboratory animals is demonstrated. Butadiene exhibits carcinogenicity in mice at doses ranging from as low as 6.25 ppm and up to 8000 ppm⁴⁹. The selected exposure concentration was within the range of carcinogenic dosage as shown by other studies^{45, 50} and our results demonstrate an increase in 1,3-butadiene-specific DNA damage, as evidenced by urinary EB-GII adducts a known biomarker of 1,3-butadiene-associated genotoxicity⁴⁶, following exposure. The observed interspecies differences in excretion of EB-GII (Figure 5) are consistent with known differences in metabolism of butadiene: mice activate butadiene to genotoxic epoxides more efficiently than rats or humans, which may be responsible for their increased sensitivity toward butadiene induced cancer^51-53. It should be noted, however, that these experiments used different butadiene exposure concentrations, which may lead to differences in metabolism due to metabolic saturation and/or enzyme induction at higher exposure levels^{50, 52, 54}.

The population-based data can be used to quantify variability and permit translation and application in human risk assessment. In particular, one of the key issues raised in the National Academy of Sciences, Engineering, and Medicine report Science and Decisions (2009) was that current practices in cancer risk assessment do not adequately address human variability, and this remains the case today⁵⁵. Because of the long latency and small sample sizes in cancer studies, both in humans and in experimental animals, it can be particularly challenging to estimate differential susceptibility across the population for this endpoint. Moreover, 1,3-butadiene remains a major environmental concern, and was designated a high priority chemical by the U.S. EPA in 2019 for risk evaluation⁵⁶. Furthermore, 1,3-butadiene is among a number of carcinogenic air toxics that are especially of concern among disadvantaged communities, and thus is a major Environmental Justice concern as well⁵⁷.

DNA adducts, both pro-mutagenic and those that are indicative of the metabolic activation of a chemical, provide a potential biomarker for exposures and molecular events that are leading to cancer. They also may be a useful surrogate for quantitative characterization of the inter-individual susceptibility, especially for a mutagenic compound such as 1,3-butadiene. Thus, we derived the EB-GII-specific mouse, rat and human variability factors using urinary adduct data from this study and those previously reported^{14, 33}. These quantitative estimates can be useful in assessing variability in future 1,3-butadiene-derived cancer assessments. Interestingly, the National Research Council report (NASEM 2009) suggested that a factor of 25 may be a useful starting point for human variability in susceptibility to cancer⁵⁵, and the human data-derived values reported here of 14 ~ 40 (depending on the degree of protection being 95% of 99% of the population, Table 2) include this value in its confidence interval. Additionally, as described above, these human-derived values are consistent with the degree of variation previously reported for other chemicals for which human data are available³⁸.

Importantly, the adduct data from five CC strains obtained in this study suggested a much lower degree of variation, at no more than 3-fold, far less than those derived from human data. There are several possible explanations for this difference. First, because the human data were obtained from an observational study, the possibility that there may be additional sources of unaddressed experimental variation cannot be ruled out, and the derived population variation could be an overestimate. Second, it is also possible that the genetic diversity in CC strains used in this study may have been limited with respect to the potential variants that may affect butadiene-induced DNA damage. Even though available CC lines are highly genetically diverse²⁴, in vivo studies of sub-chronic inhalational exposures in dozens of strains are prohibitively complex and costly. Third, this divergence may be due to the much higher exposures in the mouse study as compared to fairly low levels detected in occupationally exposed humans. In particular, because EB-GII is produced by a metabolite of 1,3-butadiene, requiring metabolic bioactivation, high butadiene exposures in mice are known to saturate metabolism across all strains⁵⁰, leading to a considerable shrinkage of population variability estimates on the logarithmic scale. Finally, such high levels of exposure in mice may also result in the saturation of adduct repair as well, given the adduct levels in mouse urine were on average more than 60,000-fold higher than those in humans. Human studies of 1,3-butadiene metabolism suggest a geometric standard deviation for metabolic activation around 1.5 ~ 1.7^{58, 59}, suggesting that at levels below saturation, a UF_H for TK variability alone would be a factor of 2 to 3, equal to or larger than the combined TK and TD UF_H values derived from mice in this study. These results suggest that interpretation of population mouse models as surrogates for human variability needs to be done with care particularly at exposure levels that may saturate TK or TD processes.

In summary, the CC mouse model was useful for investigating population variability in genotoxic effects in response to butadiene. The study results demonstrate how urinary biomarker assessment is useful for calculating butadiene-specific adjustment factors in population variability. These results are useful for evaluating interindividual differences in cancer susceptibility to butadiene. Future studies are required to identify potential genetic and epigenetic mechanisms underlying variability in butadiene-induced genotoxicity and carcinogenicity.

Supplementary Material

Supplemental materials

Figure S1. Creatinine corrected EB-GII quantitation.

Figure S2. Sex differences in creatinine excretion.

Table S1. Body weights at the start and end of exposures.