BIOTRANSFORMATION OF POLYCYCLIC AROMATIC HYDROCARBONS BY TROUT LIVER S9 FRACTIONS: EVALUATION OF COMPETITIVE INHIBITION USING A SUBSTRATE DEPLETION APPROACH

Abstract

Environmental contaminants frequently occur as part of a chemical mixture, potentially resulting in competitive inhibition among multiple substrates metabolized by the same enzyme. Trout liver S9 fractions were used to evaluate the biotransformation of three polycyclic aromatic hydrocarbons (PAHs): phenanthrene (PHEN), pyrene (PYR) and benzo[a]pyrene (BAP), tested as binary mixtures. Initial rates of biotransformation were determined using a substrate depletion approach. The resulting data were then fitted by simultaneous non-linear regression to a competitive inhibition model. In each case, the PAH possessing the lower Michaelis-Menten affinity constant (K_M) competitively inhibited biotransformation of the other compound. Inhibition constants (K_i) determined for the lower-K_M compound were generally close to previously determined K_M values, consistent with the suggestion that phase I biotransformation of PAHs is largely catalyzed by one, or a small number of cytochrome P450 enzymes. The use of a substrate depletion approach to perform enzyme inhibition studies imposes practical limitations on experimental design and complicates the interpretation of derived kinetic constants. Nevertheless, the resulting information may have utility for chemical hazard assessments as well as the design and interpretation of controlled laboratory studies. Depletion experiments informed by measured chemical concentrations in tissues may also provide a means of determining whether enzyme inhibition occurs under relevant environmental conditions.

INTRODUCTION

Many environmental contaminants exist as part of a complex mixture. Nevertheless, chemical hazard assessments are generally performed on a chemical-by-chemical basis. If one chemical impacts the accumulation of a second substance with which it co-exists, failure to recognize this interaction could lead to over- or under-prediction of the second compound’s true potential for bioaccumulation and toxicity. One mechanism by which this type of mixture interaction could occur involves chemical biotransformation. Biotransformation may substantially reduce the extent to which a chemical accumulates in fish; however, the enzymes responsible for this activity are subject to a variety of mixture effects including induction and inhibition. Induction typically involves binding of an inducing substance to a transcription factor, such as the aryl hydrocarbon hydroxylase receptor (AhR), resulting in coordinated expression of membrane transporters and biotransformation enzymes (Xu et al., 2005; Aleksunes and Klaassen, 2012). This process may take several days to achieve full expression of the response and is generally studied in controlled in vivo experiments. Alternatively, a chemical may interact directly with an enzyme, inhibiting its ability to catalyze the biotransformation of a second compound. These interactions occur on very short time scales and are amenable to study using in vitro test systems. Well-known mechanisms of enzyme inhibition include competitive, uncompetitive, and mixed-type inhibition. For reactions that follow classical Michaelis-Menten kinetics, these different types of inhibition result in characteristic patterns of change in maximum reaction velocity (V_max; pmol/min/mg protein) and the Michaelis-Menten affinity constant (K_M; μM) (Cornish-Bowden, 2012).

The potential for chemical mixture effects to impact chemical bioaccumulation in fish has been known for decades. Research is needed, however, to determine the likelihood of these interactions in relevant experimental and environmental settings. The purpose of this study was to examine the biotransformation of three polycyclic aromatic hydrocarbons (PAHs), phenanthrene (PHEN), pyrene (PYR), and benzo[a]pyrene (BAP), tested as binary mixtures using trout liver S9 fractions. Initial reaction rates were measured using a substrate depletion approach instead of the traditional product formation method. In doing so, we sought to evaluate the strengths and limitations of the substrate depletion approach for characterizing enzyme inhibition. The results were then interpreted in the context of in vivo laboratory testing with fish, current use of vitro-in vivo extrapolation (IVIVE) methods to predict in vivo rates of biotransformation in fish, and possible inhibition of enzyme activity in fish that inhabit sites contaminated with PAHs.

MATERIALS AND METHODS

Chemicals

Phenanthrene, PYR, BAP, ethoxyresorufin (ER), 1--chloro-2,4-dinitrobenzene (CDNB), p-nitrophenol (p-NP), uridine 5′-diphosphoglucuronic acid (UDPGA), reduced glutathione (GSH), and alamethicin were purchased from Sigma-Aldrich (all > 95% pure), β-nicotinamide adenine dinucleotide phosphate (β-NADPH; > 95% pure) was obtained from the Oriental Yeast Co. 3’-Phospho-adenosine 5’-diphosphosulfate (PAPS; 80% pure) was purchased from Millipore Sigma. All other chemicals and solvents were purchased from Sigma-Aldrich and were reagent grade or higher in quality.

Animals

Rainbow trout (Oncorhynchus mykiss) weighing ~ 50 g were obtained from the USGS Upper Midwest Environmental Sciences Center in La Crosse, WI, and grown up to the size used in these studies. The fish were fed commercial trout chow (Classic Trout; Skretting USA) and maintained at 11 ± 1 °C on a 16:8-h light:dark cycle. Water used for fish holding was obtained directly from Lake Superior (single pass, sand-filtered and UV treated) and had the following characteristics: pH 7.6 to 7.8; total ammonia, < 1 mg L⁻¹; alkalinity, 41 to 44 mg L⁻¹ as CaCO₃; dissolved oxygen, 85 to 100% of saturation. The fish used in this study were approximately 1.5 yr old at the time of sacrifice.

Preparation and characterization of pooled trout liver S9 fraction

Five trout (1 ♂, 4♀), averaging 448 ± 48 g, were used to generate a single pool of liver S9 fractions. The sexual maturity of each animal was evaluated by determining its gonadosomatic index (GSI; mass of gonads/total body mass × 100). Measured GSI values indicated that all fish were in very early stages of sexual maturation (Gomez et al. 1999; Le Gac et al. 2001). An earlier study performed using isolated hepatocytes from trout of the same age, strain, and stage of sexual maturity showed that there were no gender-related differences in hepatic biotransformation (Fay et al. 2014).

The methods used to generate the S9 fractions are described elsewhere (Johanning et al. 2012). Briefly, individual livers were cleared of blood and homogenized in 2 volumes of homogenization buffer using 4 to 5 strokes of a Potter-Elvehjem mortar and pestle. The homogenization buffer (pH 7.80 ± 0.05) consisted of 150 mM KCl, 50 mM Tris, 1 mM dithiothreitol, 2 mM EDTA and 250 mM sucrose. The pooled homogenate was centrifuged at 13,000 × g for 20 min at 4 °C. Individual aliquots (0.5 mL) were then flash-frozen in liquid nitrogen and stored at −80 °C until use.

Metabolic activity of the pooled S9 sample was evaluated using model substrates for cytochrome P450 1A (CYP1A), glutathione-S-transferase (GST), and uridine 5′-diphospho-glucuronosyltransferase (UGT). Cytochrome P450 1A activity was characterized by measuring the rate of 7-ethoxyresorufin O-dealkylation (EROD assay; Burke and Mayer [1974]). Glutathione-S-transferase activity was assessed by measuring glutathione conjugation of CDNB (Habig et al. 1974). Uridine 5′-diphospho-glucuronosyltransferase activity was characterized by measuring the glucuronidation of p-NP (Ladd et al. 2016). All assays were performed in quadruplicate at a temperature (11 ± 1 °C) and pH (7.80 ± 0.05) appropriate for trout. Sample protein content was measured using Peterson’s modification of the Lowry method (Sigma technical bulletin TP0300). Total CYP content was determined using a dithionite difference spectroscopy method (Matsubura et al. 1976), modified for use with trout (Nichols et al. 2013b).

Theory

The primary pathway for PAH biotransformation in fish involves CYP-mediated hydroxylation followed by glucuronidation of hydroxylated products (Varanasi et al. 1989). CYP1A plays a major role in catalyzing these hydroxylation reactions, although other CYPs (e.g., CYP3A) may contribute (Schlenk et al. 2008). If phase I biotransformation of PAHs is largely catalyzed by one or a small number of CYPs, competitive inhibition among paired chemicals can be expected. The theory presented below addresses this possibility. Experimental results were then evaluated to determine concordance with this theory.

Competitive inhibition of an enzymatic reaction can occur when an inhibitor (I) binds reversibly to the enzyme active site but is not catalyzed to form a product. This interaction may be depicted as

$$\mathit{EI}\rightleftharpoons I+E+S\rightleftharpoons \mathit{ES}\to E+P$$ (1)

where E is the enzyme, S is an enzyme substrate, and P is the product of the reaction. If the reaction follows classical Michaelis-Menten kinetics, the equation that describes this interaction is given by

$$v=\frac{V_{\mathit{max}}[S]}{K_M\left(1+\frac{[I]}{K_i}\right)+[S]}$$ (2)

where v is the reaction velocity (pmol min⁻¹ mg protein⁻¹), V_max is the maximum reaction velocity (pmol min⁻¹ mg protein⁻¹), [S] is substrate concentration, K_M is the Michaelis-Menten affinity constant (μM), and K_i is the enzyme inhibition constant (μM). Equation 2 predicts that the K_M value for S will appear to increase in the presence of I; however, v may approach V_max at sufficiently high concentrations of S, even when I is present.

If two substrates are transformed by the same enzyme, their binding at the catalytic site is mutually exclusive, and they will competitively inhibit each other’s reactions. Under these conditions, the reaction scheme depicted in Equation 1 must be expanded to

$$E+P_2\leftarrow {\mathit{ES}}_2\rightleftharpoons S_2+E+S_1\rightleftharpoons {\mathit{ES}}_1\to E+P_1$$ (3)

where S₁ and S₂ are substrates, and P₁ and P₂ are reaction products. If the rates of formation of P₁ and P₂ are slow relative to attainment of the two equilibria, the rate of formation of P₁ is described by

$$v_1=\frac{V_{\mathit{max}}^1[S_1]}{K_M^1\left(1+\frac{[S_2]}{K_M^2}\right)+[S_1]}$$ (4)

while the rate of formation of P₂ is given by

$$v_2=\frac{V_{\mathit{max}}^2[S_2]}{K_M^2\left(1+\frac{[S_1]}{K_M^1}\right)+[S_2]}$$ (5)

where v₁ and v₂ are reaction velocities (pmol min⁻¹ mg protein⁻¹), $V_{\mathit{max}}^1$ and $V_{\mathit{max}}^2$ are the maximum velocities of each reaction (pmol min⁻¹ mg protein⁻¹), [S₁] and [S₂] are substrate concentrations (μM), and $K_M^1$ and $K_M^2$ are Michaelis-Menten affinity constants (μM). An examination of Equations 4 and 5 indicates that the inhibition constant (K_i) in Equation 2 has been replaced by the K_M value for the inhibitory substrate. In this special case, therefore, the K_i value for a competitive inhibitor can be estimated from a previously determined K_M value (i.e., by assuming K_i = K_M).

Equations 4 and 5 predict that a substrate possessing higher affinity for an enzyme (lower K_M value) will tend to out-compete a lower-affinity substrate (higher K_M value) if both compounds are present at similar concentrations (i.e., S₁ ≈ S₂). This will result in an apparent increase in K_M for the higher-K_M chemical. If the concentration of the higher-K_M chemical is increased sufficiently, however, it will out-compete the lower-K_M chemical resulting in biotransformation rates that approach its own V_max.

Assessment of in vitro mixture interactions

In vitro assays were performed using a substrate depletion approach, wherein declining concentrations of parent chemical are monitored over time (Johanning et al. 2012; OECD 2018). The experiments were conducted using a single vial method with 1 mg mL⁻¹ S9 protein. The 1-mL reaction mixture consisted of 100 mM potassium phosphate buffer (pH 7.80 ± 0.05), 2 mM β-NADPH, 2mM UDPGA, 0.1 mM PAPS, 5 mM GSH, and 10 μg/mL alamethicin (introduced in methanol; 0.1% v/v final concentration). Test chemicals dissolved in acetone (0.5% v/v final concentration) were pipetted into the system to start the reaction. Previous work has shown that 0.5% acetone has no discernable effect on PAH biotransformation by trout liver S9 fractions (Nichols et al. 2018). Subsamples (100 μL) withdrawn from the vial at predetermined time points were pipetted into 300 μL of acetonitrile to stop the reaction and extract the analyte. Triplicate vials, loaded with the same buffers and starting biological material, were run for each set of tested conditions. A fourth enzymatically inactive control vial (heat-treated) was sampled in parallel with each set of active samples to account for potential non-metabolic losses including adsorption and volatilization.

In the first set of experiments, PHEN and PYR were tested individually across a range of starting substrate concentrations. These tested concentrations of PHEN (nominal) ranged from 0.007 μM to 30.0 μM, while tested concentrations of PYR ranged from 0.03 μM to 4.0 μM. Preliminary studies were performed with each chemical to optimize the sampling schedule for the highest tested concentration. Starting concentrations were then decreased, requiring progressive adjustments to the sampling schedule to account for observed increases in the reaction rate. The goal of each experiment was to measure substrate concentrations across 5 or more sampling times, while limiting the overall loss of parent chemical to < 25%. Most of the assays were performed in 30 min or less, although several were run out to 90 min. The resulting control curves were analyzed by non-linear regression (NLR) to estimate the Michaelis-Menten constants V_max and K_M for each compound.

The three PAHs were then tested as binary mixtures: PHEN/PYR, PHEN/BAP, and PYR/BAP. Guidance for performing enzyme inhibition studies generally recommends that chemical concentrations be adjusted to ‘surround’ the substrate K_M and the inhibitor K_i (Mandan et al. 2002; Obach 2008). Obach (2008) also recommended testing substrate and inhibitor concentrations that differ by a factor of at least three (higher and lower) from the respective K_M and K_i values.

Each binary mixture was evaluated by testing multiple concentrations of ‘substrate’ (the higher-K_M compound) in the presence of a fixed concentration of ‘inhibitor’ (the lower-K_M compound). Preliminary studies were performed at 2 concentrations of substrate (0.5 K_M and 4 K_M) and 4 concentrations of inhibitor (1.3K_i, 4K_i, 10K_i, and 40K_i). For these experiments, we assumed that the inhibitor’s K_i value was equal to its previously determined K_M, consistent with the special case described above (two substrates metabolized by the same enzyme). In earlier work performed using liver S9 fractions from trout, the K_M for BAP was shown to be lower than that for PHEN or PYR (Nichols et al. 2018). In the present study, therefore, BAP was used exclusively as an inhibitor, eliminating the need to generate a substrate control curve. The K_M values for PHEN and PYR were obtained from control curves developed in the present study, while the K_M for BAP was assumed to be equal to that given previously by Nichols et al. (2018).

Data obtained from these preliminary studies were combined with control data (no inhibitor) and analyzed by simultaneous non-linear regression (SNLR; Kakkar et al. 1999) using a competitive inhibition model to obtain an initial estimate of K_i for each inhibitor; that is, a K_i based on measured inhibitory activity and not the assumption that K_i = inhibitor K_M. Since their introduction in the late 1990s, SNLR methods have largely replaced older graphing methods for analysis of enzyme inhibition data due to their ease of use, reduced data requirements, and greater parameter estimation accuracy (Kakkar et al. 1999, 2000).

Additional mixture studies were then performed by evaluating two fixed concentrations of inhibitor (C_i,low and C_i,high) at 7 or 8 substrate concentrations. These low and high inhibitor concentrations were designed to approximate 1/3 K_i and 3 K_i, while substrate concentrations were based on previously-determine substrate K_M values, starting at approximately K_M and increasing with regular log unit spacing (0.3 or 0.5).

Analytical methods

The test chemicals PHEN, PYR, and BAP were analyzed by high-performance liquid chromatography. Samples were injected onto an Agilent 1260 HPLC system equipped with a fluorescence detector. Chromatography was performed using a reverse phase Hypersil Green PAH column (3 μm, 100 × 2.1 mm; Thermo Fisher Scientific). Solvent A consisted of 90% water (Milli-Q; Millipore) and 10% ACN. Solvent B consisted of 5% water and 95% ACN. The solvents were run isocratically at a flow rate of 0.7 mL/min, but the proportions of solvents A and B were changed for each analyte to optimize run time and peak symmetry. The excitation/emission wavelengths (nm) were 260/350 (PHEN), 237/390 (PYR), and 255/420 (BAP). The limits of detection for PHEN, PYR, and BAP were approximately 0.5, 0.2, and 0.1 nM, respectively. The PAH concentration in the S9 extracts was quantified by comparing peak areas to those of known standards.

Data analysis

The standard deviation of reaction rates measured in control and mixture studies tended to increase with starting substrate concentration. Curve fitting (NLR and SNLR analysis) was therefore performed with 1/Y² weighting using reaction rates measured in each depletion experiment (n = 3 observations at each substrate concentration). Mixture data sets were analyzed by setting the inhibitor concentration equal to the mean of measured starting concentrations. The rational for this decision is provided under Results, PAH Mixture Data. All analyses were performed in GraphPad Prism 5. All statistical tests were performed using a significance level of 0.05.

RESULTS

Characterization of pooled trout liver S9 fraction

Measured EROD, UGT, and GST activities of the pooled S9 fraction averaged 7.1 ± 0.3 pmol min⁻¹ mg protein⁻¹, 623 ± 53 pmol min⁻¹ mg protein⁻¹, and 504 ± 14 nmol min⁻¹ mg protein⁻¹, respectively (mean ± SD, n = 4). The initial protein content of the pooled sample was 25.6 ± 0.6 mg mL⁻¹, while the CYP content was 68.9 ± 1.6 pmol P450 mg protein⁻¹. All values are within the ranges of values determined previously for S9 fractions isolated from trout of the same age and strain (Nichols et al. 2013a, 2018).

Substrate concentration-dependence of PAH biotransformation (control curves)

When tested individually, PHEN and PYR exhibited significant depletion (negative slope ≠ 0) at each tested concentration (Figure S1). There were no apparent chemical losses from negative controls. Had these reactions been carried out for longer periods of time, log-linear depletion would have been expected. By restricting incubation times, however, we sought to measure initial reaction rates from the early (nearly linear) part of the depletion curve. In most cases, the r² value associated with a linear regression of untransformed substrate concentrations against time was > 0.95. Based on these observations, reaction rates were calculated directly from the slope of the line. The relative standard deviation (RSD) of reaction rates determined at each tested substrate concentration was generally less than 10%, indicating a very high level of assay reproducibility.

Control curves for PHEN and PYR were well-described by the Michaelis-Menten model (r² > 0.96; Figure S2). The fitted K_M for PHEN (0.84 μM) is 1.6 times greater than that (0.52 μM) determined previously using trout liver S9 fractions (Nichols et al. 2018), and 11 times larger than the K_M determined here for PYR (0.075 μM; Table 1). In contrast, the fitted K_M for PYR is nearly identical to that (0.07 μM) determined in an earlier study (Nichols et al. 2018). The fitted V_max for PHEN (26.3 pmol min⁻¹ mg S9⁻¹) is smaller than that for PYR (34.9 pmol min⁻¹ mg S9⁻¹); however, these values differed by less than 35% (Table 1). Both V_max estimates are within 17% of values determined previously using trout liver S9 fractions (Nichols et al. 2018).

Table 1.

Michaelis-Menten kinetics constants (K_M and V_max) for biotransformation of phenanthrene, pyrene, and benzo[a]pyrene in trout liver S9 fractions^a

Present study				Nichols et al. (2018)
	KM (μM)	Vmax (pmol min−1 mg S9−1)	r2	KM (μM)	Vmax (pmol min−1 mg S9−1)
Phenanthrene	0.84 (0.07)	26.3 (1.36)	0.98	0.52	22.5
Pyrene	0.075 (0.004)	34.9 (0.63)	0.97	0.07	40.9
Benzo[a]pyrene				0.03	28.7

^aK_M and V_max values estimated in the present study were obtained by fitting initial rates of substrate depletion to the Michaelis-Menten model. Values given by Nichols et al. (2018) were estimated by fitting first-order depletion rate constants to an empirical relationship given by Obach et al., (2002). The numbers in parenthesis represent the SE of derived terms.

Evaluation of PAH mixtures using a substrate depletion approach

In each of the mixture studies, the concentration of the lower-K_M inhibitor declined more rapidly than that of the higher-K_M substrate, relative to starting concentration values (Figure 1). This difference in observed depletion kinetics was most apparent for the PHEN/BAP mixture. In several cases, data points were trimmed from the end of one or both depletion curves. These adjustments were made to minimize depletion of the inhibitor and achieve a similar extent of inhibitor depletion across an entire set of mixture experiments (one pair of PAHs). In no case were more than 3 data points eliminated from an individual data set, leaving at least 5 to define each curve.

Figure 1.

Representative substrate depletion curves from competitive inhibition studies. The selected plots represent test conditions close to the substrate K_M value, determined in the presence of a high inhibitor concentration (~ 3K_i). Each panel shows data for the higher-K_M substrate (open circles, squares and triangles) and the lower-K_M inhibitor (solid dots, squares and triangles). All data represented by a given symbol were generated using one reaction vial. At many time points the three symbols are nearly overlapping, giving the appearance of a single point. Stated concentrations associated with each plot refer to measured starting values. A. 11.04 μM phenanthrene and 1.29 μM pyrene. B. 4.44 μM phenanthrene and 0.43 μM benzo[a]pyrene. C. 0.28 μM pyrene and 0.062 μM benzo[a]pyrene.

In some cases, a plot of inhibitor concentration against time yielded a linear pattern of declining concentration, while in others a log-linear pattern was observed. In contrast, substrate concentrations declined in a consistently linear manner even when the inhibitor concentration had been substantially reduced. These differences in curve shape were most apparent in mixture studies with PHEN/BAP (Figure 1B). Across all tested concentrations of PHEN, depletion of BAP exceeded 90%; nevertheless, r² values associated with the fit of PHEN depletion data to a linear model were consistently greater than 0.95. Thus, large changes in inhibitor concentration had little effect on the linearity of substrate depletion curves. From the perspective of calculating an initial reaction rate, this result is advantageous. In addition, it suggests that the inhibitor concentration may be set equal to the initial value when modeling these data sets since the slope of the substrate depletion curve at very early time points (i.e., when the inhibitor concentration is still close to the starting value) was maintained throughout.

At high substrate concentrations, initial rates of substrate depletion were low even in the absence of inhibitor, so any inhibition made it difficult to obtain a measurable rate of activity during the working lifetime of the assay (here limited to 90 min). Each depletion data set was evaluated to determine if the fitted slope was significantly different from 0. If one or more of the 3 curves for a set of tested conditions did not evidence significant depletion, all three curves were excluded from the analysis. This exclusion step resulted in the truncation of several data sets used to estimate enzyme inhibition constants.

Selection of a preferred inhibition model

Initial reaction rates for PHEN and PYR in the presence of inhibitor were fitted by NLR to the Michaelis-Menten model to obtain estimates of K_M and V_max under inhibited conditions (Table 2). A comparison of these values to fitted K_M values from control curves showed a clear pattern of increase in K_M with increasing inhibitor concentration. This finding ruled out the possibility of uncompetitive inhibition, which predicts that K_M will decrease in the presence of inhibitor. Changes in V_max with the addition of inhibitor were less clear, leaving open the possibility of mixed-type inhibition, which predicts that V_max will decrease with increasing inhibitor concentration.

Table 2.

Enzyme kinetic constants for biotransformation of PAHs, tested as binary mixtures using trout liver S9 fractions^a

Data set(s) evaluated	Inhib. conc.	Michaelis-Menten model		Competitive inhibition model	Model fit
	Ci (μM)	KMb(μM)	Vmaxb (pmol min−1 mg S9−1)	Kic (μM)	r2
Phenanthrene/Pyrene
Control	0.0	0.84 (0.71–0.98)	26.32 (23.49–29.14)	-----	0.98
Low inhibitor (Ci,low)	0.14	1.59 (1.42–1.75)	24.86 (23.04–26.69)	-----	0.99
High inhibitor (Ci,high)	1.29	11.37 (9.20–13.54)	32.14 (26.97–37.30)	-----	0.99
Control + Ci,low	0.14	-----	-----	0.14 (0.11–0.16)	0.98
Control + Ci,high	1.29	-----	-----	0.13 (0.12–0.15)	0.98
Control + Ci,low and Ci,high	0.14, 1.29	-----	-----	0.13 (0.12–0.15)	0.98
Phenanthrene/Benzo[a]pyrene
Control	0.0	0.84 (0.71–0.98)	26.32 (23.49–29.14)	-----	0.98
Low inhibitor (Ci,low)	0.10	1.57 (1.35–1.78)	26.88 (25.04–28.72)	-----	0.99
High inhibitor (Ci,high)	0.43	4.31 (3.50–5.13)	27.71 (24.25–31.17)	-----	0.99
Control + Ci,low	0.10	-----	-----	0.13 (0.08–0.17)	0.97
Control + Ci,high	0.43	-----	-----	0.11 (0.10–0.13)	0.98
Control + Ci,low and Ci,high	0.10, 0.43	-----	-----	0.11 (0.10–0.13)	0.98
Pyrene/Benzo[a]pyrene
Control	0.0	0.075 (0.067– 0.083)	34.87 (33.57–36.17)	-----	0.97
Low inhibitor (Ci,low)	0.010	0.110 (0.090–0.130)	36.38 (34.24–38.53)	-----	0.93
High inhibitor (Ci,high)	0.062	0.305 (0.274–0.335)	34.38 (32.73–36.03)	-----	0.98
Control + Ci,low	0.010	-----	-----	0.039 (0.014–0.045)	0.95
Control + Ci,high	0.062	-----	-----	0.019 (0.017–0.022)	0.98
Control + Ci,low and Ci,high	0.010, 0.062	-----	-----	0.019 (0.016–0.022)	0.96

^aThe tested pairs of PAHs are identified by listing the higher-K_M substrate first, followed by the lower-K_M inhibitor. All kinetic constants are reported as the mean (95% confidence intervals).

^bDetermined by fitting initial rates of substrate depletion to the Michaelis-Menten model using non-linear regression.

^cDetermined by fitting initial rates of substrate depletion from control and enzyme inhibition studies to a competitive inhibition model using simultaneous non-linear regression.

Selection of a preferred inhibition model was accomplished by using SNLR to fit full data sets from the three mixture studies (control plus both inhibitor curves) to both competitive and mixed-type inhibition models. For two of the three PAH mixtures (PHEN/PYR and PHEN/BAP), this analysis yielded an ambiguous result (essentially no difference in model fit). For the third mixture (PYR/BAP), the competitive inhibition model was clearly preferred (lower AIC criteria value; Akaike, 1974). All subsequent analyses were performed using the competitive inhibition model.

Characterization of competitive inhibition

Preliminary studies with each pair of PAHs indicated a strong inhibitory effect of the lower-K_M inhibitor on biotransformation of the higher-K_M substrate (Figures S3). SNLR analysis of these data sets yielded initial estimates of K_i for each substrate/inhibitor combination. The K_i value for PYR, determined in mixture studies with PHEN, was 0.25 μM. The K_i for BAP, when tested with PHEN, was 0.086 μM, while that determined in mixture studies with PYR was 0.019 μM.

A final set of inhibition experiments was then performed to obtain definitive estimates of K_i for PYR and BAP (Figures 2–4 and Table 2). To evaluate the contribution of low and high inhibitor curves to the overall analysis, SNLR was performed using the control + C_i,low, control + C_i,high, and control + C_i,low and C_i,high data sets. In general, K_i values determined for each pair of PAHs using the three different data sets exhibited good agreement (factor of 2 difference or less). A closer examination of these values indicates, however, that K_i estimates obtained using all data (control + C_i,low and C_i,high) or just the control + C_i,high data set were identical, while those determined using the control + C_i,low data set were slightly higher.

Figure 2.

Competitive inhibition of phenanthrene biotransformation by pyrene. Phenanthrene biotransformation rates were measured in the presence of 0.0 μM, 0.14 μM, or 1.29 μM pyrene (starting values). Solid dots – control data. Open circles – low inhibitor concentration. Solid squares – high inhibitor concentration. In this and the following figures, each point represents the mean (± SD) of three rate determinations. Solid lines were obtained by fitting a competitive inhibition model to the entire data set using simultaneous non-linear regression.

Figure 4.

Competitive inhibition of pyrene biotransformation by benzo[a]pyrene. Pyrene biotransformation rates were measured in the presence of 0.0 μM, 0.010 μM, or 0.062 μM benzo[a]pyrene (starting values). Solid dots – control data. Open circles – low inhibitor concentration. Solid squares – high inhibitor concentration.

The K_i for PYR, determined in the PHEN/PYR mixture study (all data), was 0.13 μM. For comparison, the K_M determined from the control curve was 0.075, or a factor of 2 lower. The K_i for BAP, determined in the PHEN/BAP mixture study (all data), was 0.11 μM. This value is approximately 3.7 times greater than the K_M for BAP (0.03 μM) determined earlier using trout liver S9 fractions (0.03 μM; Nichols et al. 2018). The K_i for BAP, estimated using all data from the PYR/BAP mixture study, was 0.019 μM, which is approximately 0.6 times the K_M for BAP given previously (0.03 μM; Nichols et al. 2018).

DISCUSSION

In vitro enzyme inhibition studies are generally performed by evaluating the effect of an inhibitor on the rate of formation of a reaction product. This approach requires prior knowledge of the product and a quantitative method for its analysis. Detailed methods for performing these studies, including guidance on experimental design, analysis of resulting data, and the interpretation of derived kinetic constants are provided in several recent texts (Purich 2010; Cornish-Bowden 2012; Bisswanger 2017). Additional sources describe the application of these procedures to drug discovery and development (Madan et al. 2002; Obach 2008).

For some applications, however, it may be of interest to characterize inhibitory effects on biotransformation of a parent chemical without knowledge of or regard for specific reaction products. One example involves chemical hazard assessments performed by chemical regulatory authorities. Most of these assessments focus, at least initially, on properties (e.g., persistence, bioaccumulation, toxicity) associated with a parent chemical. A second example involves in vivo experimentation wherein multiple chemicals are evaluated simultaneously. This type of testing is performed to achieve efficiencies in terms of time, cost, and use of animals. However, inhibitory effects on chemical biotransformation could lead to findings that do not reflect the behavior of these chemicals when tested individually.

In vitro substrate depletion methods have been used for over two decades for preclinical screening of drug candidates (Iwatsubo et al. 1997; Obach et al. 1997, 1999; Carlile et al. 1998). The goal of such studies is to measure in vitro intrinsic clearance (CL_int; mL min⁻¹ mg protein or 10⁶ cells⁻¹), defined as the rate of clearance under non-saturating conditions ([S] << K_M). This CL_int value is then extrapolated to an estimate of hepatic clearance (CL_H) using appropriate scaling factors and a physiological model of the liver. More recently, these methods have been employed by environmental toxicologists to measure CL_int in fish, as a means of refining modeled predictions of chemical bioaccumulation (Cowan-Ellsberry et al. 2008; Dyer et al. 2008; Gomez et al. 2010; Han et al. 2007, 2009; Laue et al. 2014). Substrate depletion methods have also been employed by mammalian toxicologists to obtain information needed to perform human health hazard assessments for large chemical inventories (Rotroff et al. 2010; Wetmore et al. 2012, 2013). In each of these applications, the primary need is to estimate CL_H of a parent chemical occurring by all possible biotransformation pathways.

Depletion curves obtained in control studies with PHEN and PYR were highly linear. This finding was not unexpected, given that the loss of parent chemical was limited to 25% or less. Under these conditions, a log-linear decrease in concentration, which should in theory occur in a closed reaction system, is difficult to distinguish from simple linear behavior. More surprising was the behavior of substrate depletion curves obtained in competitive inhibition studies. A distinguishing feature of these data sets is that measured concentrations of the inhibitory (lower-K_M) PAH declined more rapidly than those of the higher-K_M substrate (Figure 1). These changes might have been expected to result in substrate depletion curves that bend down at later time points, indicating a progressive increase in biotransformation rate as the inhibitor disappears from the system. Instead, these depletion curves were, like those obtained in control studies, remarkably linear. For the PHEN/PYR and PHEN/BAP mixtures, large differences in K_M values for the substrate and inhibitor (11-fold and 28-fold, respectively) may have obscured the effect of declining inhibitor concentrations, since even small amounts of the inhibitory PAH would have substantially inhibited the reaction. Alternatively, the tendency for substrate curves to bend down may have been offset by a tendency for the system to exhibit first-order kinetics, which causes depletion curves to flatten out over time.

Previously, Obach and Reed-Hagen (2002) provided a method for using substrate depletion data to estimate K_M and V_max values. Briefly, depletion experiments are performed across a range of substrate concentrations. First-order depletion rate constants (k_dep; min⁻¹) are obtained by linear regression of log-transformed data against time. These derived k_dep values are then fitted to an empirical relationship, providing an estimate of K_M (from the inflection point) and CL_int. Recognizing that CL_int = V_max/K_M, V_max may then be estimated as the product CL_int K_M. Subsequent work has shown that this method is theoretically sound, provided that k_dep values are obtained during the initial phase of each reaction when the depletion kinetics are mono-exponential (Nath and Atkins 2006).

The use of SNLR methods to estimate K_i requires, however, that enzyme activity be expressed as an initial reaction rate (product formation or substrate loss) normalized to the concentration of biological material (e.g., pmol min⁻¹ mg protein⁻¹). Multiplication of a fitted k_dep value by a starting substrate concentration provides the desired units of activity. In principal, therefore, an initial reaction rate can be calculated from a log-linear depletion curve. As noted above, however, depletion curves obtained in this effort were highly linear. Initial rates were therefore obtained directly from the slopes of these curves. This approach is comparable to the traditional method of characterizing enzyme kinetics by measuring initial rates of product formation but differs in the respect that derived kinetic constants represent the net result of all enzymatic pathways acting on the substrate simultaneously. As such, these derived constants are properly viewed as apparent values, and not as values that can be ascribed to one enzymatic reaction.

A disadvantage of the substrate depletion approach is that enzyme activity is difficult to measure at high starting substrate concentrations (i.e., near-saturating conditions) or when reaction rates are low. In either case, the slope of the line described by these data sets is small, and the ability to detect measurable activity is complicated by the presence of a high parent chemical ‘background.’ In contrast, all the product formed in a traditional product formation assay can be attributed to the reaction. This makes it possible to measure very low rates of activity, provided that the analytical method exhibits high sensitivity and precision. Moreover, product concentrations increase with increasing substrate concentration up to the point of enzyme saturation. Thus, in comparison to the substrate depletion approach, the product formation method is better suited to measuring low rates of activity and/or estimating V_max.

In the final set of experiments, the target concentrations for the low and high inhibitor curves were 1/3K_i and 3K_i. A post-hoc comparison of measured inhibitor concentrations and derived K_i values (all data for each tested pair) shows that the low inhibitor concentration ranged from 0.52K_i (PYR/BAP) to 1.0K_i (PHEN/PYR) while the high inhibitor concentration ranged from 3.3K_i (PHEN/BAP) to 9.9K_i (PHEN/PYR). In each case, the high inhibitor concentration resulted in good differentiation of control and inhibitor curves. When using the substrate depletion approach, inhibitor concentrations greater than 10K_i may not be useful as this high level of inhibition further complicates the task of measuring biotransformation at high substrate concentrations. Inhibitor concentrations substantially less than K_i are also problematic if they do not result in differentiation of the control and inhibitor curves.

As a generalization, the use of a substrate depletion approach for competitive inhibition studies becomes more challenging as the difference in K_M values for the inhibitor and substrate increases. If the K_M for an inhibitor is low enough relative to that of a competing substrate, it would be possible to have situation wherein the inhibitor is eliminated from the system before there is detectable depletion of the substrate. This would make it impossible to reliably estimate K_i.

Interpretation of fitted Ki values

The K_i for BAP, determined in the PYR/BAP mixture study (all data), was very close (factor of 0.6 times lower) to its previously reported K_M value. This finding is consistent with the possibility that PYR and BAP are metabolized by one enzyme. Alternatively, multiple enzymes may contribute in a proportionately similar manner to biotransformation of both chemicals. In contrast, the K_i for PYR, determined in the PHEN/PYR mixture study, was about 2 times higher than its measured K_M. Similarly, the K_i for BAP, determined in the PHEN/BAP study (all data), was approximately 3.7 times higher than its previously reported K_M. These differences between estimated K_i and K_M values for PYR and BAP, observed in competitive inhibition studies with PHEN, are relatively small. Nevertheless, it is interesting to note that in each case the direction and magnitude of the difference is similar. A possible explanation for these findings is that multiple enzymes which operate similarly against PYR and BAP contribute in a proportionately different way to biotransformation of PHEN.

For each of the PAH mixtures, K_i values derived using only one inhibitor curve (control + C_i,low or control + C_i,high) exhibited good agreement with the K_i obtained using all data (control + Ci,low and Ci,high). This finding suggests that a good estimate of K_i can be obtained from a data set that consists of a control and one inhibitor curve, provided that the mechanism of inhibition is known. A similar conclusion was reached by Kakkar et al. (2000) in a study that employed SNLR to analyze simulated data. The same authors concluded that data for three or more inhibitor concentrations are required to accurately discriminate different mechanisms of inhibition.

Potential applications

The present study shows that substrate depletion methods may be used to study enzyme inhibition. Here we examined the special case where two chemicals competitively inhibit one another’s biotransformation, as this provides a basis for prior estimation of K_i from a measured K_M value. However, the same methods can be used to study other mechanisms of inhibition, provided that preliminary experiments are performed to obtain an initial estimate of K_i.

Depending on the question being addressed, these procedures can be used to determine whether inhibition is likely to occur under a prescribed set of conditions, or alternatively, to predict circumstances under which it could occur. For example, if an investigator is interested in performing an in vivo study with two or more chemicals, one of which is known to undergo biotransformation, it may be of interest to show that inhibition is unlikely to occur. For this evaluation, it would not be necessary to characterize a mechanism of inhibition or determine a K_i value. Instead, all that would be required is to perform an in vitro depletion assay with the chemicals of interest under conditions that can be related to the in vivo exposure. The results could then be compared with those obtained by testing the same chemicals individually (control curves). Alternatively, if the K_i value for a given pair of chemicals is determined in vitro, this information could be used to adjust in vivo exposure conditions to avoid enzyme inhibition.

Because enzymes operate against chemicals that are freely dissolved in aqueous solution, comparisons between an in vitro assay result and in vivo chemical exposure must be made on an unbound chemical concentration basis. If a compound is relatively hydrophilic, there may be little difference between total and unbound chemical concentrations in vitro. For more hydrophobic compounds, the free concentration in vitro may be a small fraction of the total. Several algorithms have been provided to estimate chemical binding in trout S9 fractions (Han et al. 2009; Lee et al. 2017; Nichols et al. 2018). Additional data are needed, however, to evaluate and refine these relationships (Saunders et al. 2019).

In a waterborne chemical exposure, the unbound chemical concentration in a fish will approach that in water, provided the exposure proceeds to steady-state and that biotransformation is negligible. Biotransformation reduces the extent to which chemicals accumulate in fish. Further, this activity may reduce the chemical concentration in the liver tissue itself to a level well below that of other tissues. In the absence of a relatively sophisticated model, it is difficult to predict an unbound chemical concentration in the liver from an assumed concentration in water and measured rate of biotransformation. One can with confidence, however, assume that it is lower than the free concentration in the exposure water. Thus, the free concentration in water provides an upper limit that can be used to make in vitro-in vivo comparisons.

As noted previously, in vitro substrate depletion methods have been used to support modeled bioaccumulation assessments for fish (Cowan-Ellsberry et al. 2008; Dyer et al. 2008; Gomez et al. 2010; Han et al. 2007, 2009; Laue et al. 2014). The present work shows that if two substrates are transformed by the same enzyme(s), depletion of both chemicals can be studied simultaneously with no discernable competitive inhibition, provided the concentrations of both substrates are well below their respective K_M values, and if available analytical methods possess the required degree of sensitivity. This finding is consistent with theory and hinges on a close relationship of K_i to the inhibitor’s K_M value. For a binary mixture of this type, chemical concentrations (substrate and inhibitor) 1/10^th K_M or lower are probably adequate to achieve this result. Extrapolation procedures used to predict CL_H from substrate depletion data generally assume that the assay is conducted under first-order conditions, as this provides an estimate of true CL_int. This requires, in turn, that the assay be performed at a starting substrate concentration << K_M. Thus, test conditions recommended for estimating CL_int would, if applied to two chemicals in a binary mixture, ensure little or no competitive inhibition, provided these chemicals are transformed by the same metabolic pathway(s). Simultaneous testing of three or more substrates for the same enzymes(s) would be more challenging, since biotransformation of a higher K_M substrate would be inhibited, albeit to different degrees, by all lower-K_M chemicals. Moreover, the lowest-K_M chemical would tend to inhibit in vitro clearance of all intermediate-K_M chemicals, causing them to be retained within the system. Under these conditions, it would be difficult to estimate chemical concentrations that would result in little appreciable inhibition.

When a compound inhibits an enzyme but does not itself undergo biotransformation, the relationship of K_i to K_M becomes moot. Alternatively, the K_i for an inhibitor may differ quite substantially from its K_M value. This would be true, for example, if a compound inhibits one enzyme but is largely metabolized by a second. Given these complications, it seems reasonable to advise against performing substrate depletion assays with more than one chemical if the goal is to use the data to support modeled bioaccumulation assessments. Although inhibition could, in many cases, be avoided by performing adequate preliminary work, this would eliminate the cost savings originally sought by testing multiple chemicals simultaneously.

Environmental significance

Efficient biotransformation of PAHs in fish may be inferred from the presence of metabolic products as well as the fact that measured concentrations of parent chemicals tend to be much lower than those predicted from simple partitioning to tissue lipids (Varanasi et al. 1989). Indeed, it may be difficult to measure some well-metabolized PAHs in fish tissue, even though they are present at relatively high concentrations in water and sediments. These observations do not, however, indicate whether biotransformation is inhibited, and if so for which compounds. Complicating matters further, many PAHs have been shown to impact CYP1A expression levels in fish and fish liver lines. Thus, PYR and BAP operate via an aryl hydrocarbon receptor (AhR)-based mechanism to induce CYP1A (Gearhart and Carlson 1978; van der Weiden et al. 1994; Bols et al. 1999; Fent and Bätscher 2000; Zapata-Perez et al. 2002; Barron et al. 2004), while fluoroanthrene may cause down-regulation of the same enzyme (Willett et al. 2001). It is unlikely, therefore, that K_i values determined from in vitro experiments can be used to determine whether competitive inhibition impacts PAH biotransformation in a real-world setting.

These complications do not, however, preclude the design of in vitro studies based on the measured PAH concentrations in fish tissues. For example, PAH concentrations in fish liver have been reported by several authors (Pointet and Milliet 2000; Xu et al. 2011; Zhao et al. 2014). In principal, this information could be used to design experiments that would show whether accumulated levels of PAHs can inhibit biotransformation of other PAH substrates. Finally, it should be noted that the CYPs responsible for PAH biotransformation in fish have been shown to metabolize a wide range of hydrophobic environmental contaminants (Schlenk et al. 2007). Thus, the methods outlined here may have utility for investigating competitive inhibition among chemicals representing several different chemical classes.

Figure 3.

Competitive inhibition of phenanthrene biotransformation by benzo[a]pyrene. Phenanthrene biotransformation rates were measured in the presence of 0.0 μM, 0.10 μM, 0.43 μM benzo[a]pyrene (starting values). Solid dots – control data. Open circles – low inhibitor concentration. Solid squares – high inhibitor concentration.