Toxicology 316 (2014) 43–54

Contents lists available at ScienceDirect

Toxicology journal homepage: www.elsevier.com/locate/toxicol

Evaluating the additivity of perfluoroalkyl acids in binary combinations on peroxisome proliferator-activated receptor-␣ activation夽 Cynthia J. Wolf a,∗ , Cynthia V. Rider b , Christopher Lau a , Barbara D. Abbott a a Developmental Toxicology Branch, Toxicity Assessment Division, National Health and Environmental Effects Research Laboratory, ORD, U.S. EPA, Research Triangle Park, NC 27711, United States b National Toxicology Program, NIEHS, Research Triangle Park, NC 27709, United States

a r t i c l e

i n f o

Article history: Received 7 August 2013 Received in revised form 18 November 2013 Accepted 7 December 2013 Available online 26 December 2013 Keywords: Perfluoroalkyl acids PFOA Mixtures Additivity PPAR␣

a b s t r a c t Perfluoroalkyl acids (PFAAs) are found globally in the environment, detected in humans and wildlife, and are typically present as mixtures of PFAA congeners. Mechanistic studies have found that responses to PFAAs are mediated in part by PPAR␣. Our previous studies showed that individual PFAAs activate PPAR␣ transfected into COS-1 cells. The goal of the current study was to determine if binary combinations of perfluorooctanoic acid (PFOA) and another PFAA act in an additive fashion to activate PPAR␣ in the mouse one-hybrid in vitro model. COS-1 cells were transiently transfected with mouse PPAR␣ luciferase reporter construct and exposed to either vehicle control (0.1% DMSO or water), PPAR␣ agonist (WY14643, 10 ␮M), PFOA at 1–128 ␮M, perfluorononanoic acid (PFNA) at 1–128 ␮M, perfluorohexanoic acid (PFHxA) at 8–1024 ␮M, perfluorooctane sulfonate (PFOS) at 4–384 ␮M or perfluorohexane sulfonate (PFHxS) at 8–2048 ␮M to generate sigmoidal concentration–response curves. In addition, cells were exposed to binary combinations of PFOA + either PFNA, PFHxA, PFOS or PFHxS in an 8 × 8 factorial design. The concentration–response data for individual chemicals were fit to sigmoidal curves and analyzed with nonlinear regression to generate EC50 s and Hillslopes, which were used in response-addition and concentration–addition models to calculate predicted responses for mixtures in the same plate. All PFOA + PFAA combinations produced concentration–response curves that were closely aligned with the predicted curves for both response addition and concentration addition at low concentrations. However, at higher concentrations of all chemicals, the observed response curves deviated from the predicted models of additivity. We conclude that binary combinations of PFAAs behave additively at the lower concentration ranges in activating PPAR␣ in this in vitro system. Published by Elsevier Ireland Ltd.

1. Introduction Humans are exposed to a multitude of environmental contaminants. Many environmental contaminants are found not as a single chemical, but as a group of chemicals of the same family. Perfluoroalkyl acids (PFAAs) are one such class of chemicals. Assessing the health risks of PFAAs must take into consideration the joint action of their mixtures.

夽 Disclaimer: This paper has been reviewed by the National Health and Environmental Effects Research Laboratory, US EPA. The use of trade names is for identification only and does not constitute endorsement by the US EPA. The findings and conclusions in this report are those of the authors and do not necessarily reflect the views of the US EPA. ∗ Corresponding author at: US EPA, NHEERL, TAD, MD-67, 109 TW Alexander Drive, Research Triangle Park, NC 27711, United States. Tel.: +1 919 541 5195; fax: +1 919 541 4017. E-mail address: [email protected] (C.J. Wolf). 0300-483X/$ – see front matter. Published by Elsevier Ireland Ltd. http://dx.doi.org/10.1016/j.tox.2013.12.002

PFAAs are man-made surfactants found globally in the environment and in wildlife and human serum as a mixture of individual members of the PFAA class (Lau et al., 2007). These chemicals are ubiquitous as they are found in many consumer and industrial applications, including oil-resistant coatings on food packaging, water-repellants, fire-fighting foams, industrial lubricants, and cosmetics (Kissa, 2001; Renner, 2001) and are generated by degradation of fluoropolymers. PFAAs are present as mixtures in environmental media such as food, water, air and house dust (Eschauzier et al., 2013; Fromme et al., 2007; Vestergren et al., 2012), and in human serum (Calafat et al., 2007; Ode et al., 2013). Manufacturing has ceased for perfluorooctane sulfonate (PFOS) in the U.S. and has been scaled down for some other PFAAs worldwide (Buck et al., 2011; Lau et al., 2007), and this is reflected in the significant decline in levels of PFOS in human serum in the past decade from 30.4 ng/mL to 13.2 ng/mL (Calafat et al., 2007; Kato et al., 2011). However, these levels are still high compared to other PFAAs. Also, this class of chemicals is very stable and levels of PFOS

44

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

and several other PFAAs remain unchanged overall in the arctic environment over the past decade (Butt et al., 2010). In addition, perfluorooctanoic acid (PFOA) levels remain fairly unchanged and levels of perfluorononanoic acid (PFNA), although relatively low, have risen in human serum in recent years (Jain, 2013; Kato et al., 2011). The PFAAs detected together at the highest levels in human serum in the U.S. are PFOS, PFOA, PFNA, and perfluorohexane sulfonate (PFHxS) (Kato et al., 2011). Experimental exposure to PFAAs induces toxic effects including hepatomegaly, respiratory distress, delayed growth and development, and reduced survival in the offspring of rodents exposed during gestation (Grasty et al., 2003; Lau et al., 2003, 2006, 2007; Seacat et al., 2002, 2003). Although the complete mechanism of action for PFAAs is not yet known, one key component mediating their effects is activation of peroxisome proliferator-activated receptor-␣ (PPAR␣). PFAAs activate PPAR␣ in hepatocytes of rat, human and chick (Bjork and Wallace, 2009; Cwinn et al., 2008; Elcombe et al., 2010) and in transfected cell models (Maloney and Waxman, 1999; Shipley et al., 2004; Takacs and Abbott, 2007; Vanden Heuvel et al., 2006). Toxicogenomic analysis showed that PPAR␣ was required for most PFOA-induced changes in gene expression (Rosen et al., 2007) and that PFOS and PFOA upregulated a well recognized group of PPAR␣-dependent genes as the predominant mechanism in fetal liver and lung (Rosen et al., 2009). PPAR␣ was found to be an integral part of PFAA action in developmental effects in vivo, as responses were found after PFOA or PFNA exposure in wild-type but not in PPAR␣-knockout mice (Abbott et al., 2007; Wolf et al., 2010). All PFAAs of chain length C4–C12 that have been evaluated activate mouse PPAR␣ with varying potencies (Buhrke et al., 2013; Wolf et al., 2008, 2012). Some PFAAs are considered partial agonists of the PPAR␣ receptor and were also found to increase expression of other receptors such as CAR and PXR (Abbott et al., 2009; Bijland et al., 2011; Bjork et al., 2011; Cheng and Klaassen, 2008; Elcombe et al., 2012; Rosen et al., 2008, 2010). Considering that the PFAAs appear to share a mode of action that involves activation of PPAR␣, and that multiple PFAAs are present in the environment and human tissues, it is important to understand how these chemicals behave in mixtures. To evaluate the type of interaction (concentration or response additive, greater than additive, less than additive) displayed by mixtures, mathematical models are used to predict expected outcomes under an assumption of additivity among mixture constituents. Observed data are then compared to predicted responses to determine whether the observed mixture responses conform to additivity or are greater than or less than predicted responses. Concentration addition (CA) is commonly applied to chemicals that have a similar mechanism of action. Alternatively, response addition (RA) has been suggested for use with chemicals that have different mechanisms of action (Greco et al., 1995; Rider and LeBlanc, 2005; Wilson et al., 2004). PFAAs have a similar mechanism via activation of PPAR␣, although some may be only a partial agonist of PPAR␣, and could exhibit alternative mechanisms of action. For example, one chemical could act as an agonist of PPAR␣ at high doses, but act as an antagonist in the presence of a strong agonist (Orton et al., 2011). Therefore, both concentration and response additivity models were included to preclude assumptions about similarity of mechanisms and to ensure that any deviations from additivity could be determined in relation to both additivity models. Few studies have been performed on the effects of mixtures or combinations of PFAAs (Ding et al., 2013; Rodea-Palomares et al., 2012; Wei et al., 2009). These studies often used organism lethality or cytotoxicity as the endpoint, included only PFOA and PFOS, or included combinations with other pollutants in aquatic environments. Recently, Carr et al. (2013) studied mixtures of the four

PFAAs most prominently found in human serum, PFOA, PFOS, PFNA and PFHxS, using a fixed-ratio approach at the ratio found in serum with binary and quaternary combinations. They used the same transiently transfected COS-1 cell PPAR␣ reporter assay used in our laboratory to test the four compound mixture and binary mixtures and they reported effects of the PFAAs that were generally less than additive with a potential for additivity in the lower dose–response region. The present study was designed primarily to determine if PFAAs in mixtures exhibit additivity in activating PPAR␣. In this study, we tested five PFAAs commonly found in human serum and measured activation of PPAR␣. PFAAs were tested individually and in binary combinations of PFOA + either PFNA, PFHxA, PFOS, or PFHxS in an 8 × 8 factorial grid design and the responses of mouse PPAR␣ in transiently transfected COS-1 cells were evaluated. Parameters (EC50 and Hillslope) of each individual chemical were input into RA and CA models to predict the responses of the mixtures. Although our previous studies have demonstrated the ability of PFAAs to activate both mouse and human PPAR␣ in the transiently transfected COS-1 model, the mouse PPAR␣ reporter plasmid was selected for the present study of additivity. The basis for this selection was because modeling with both the CA and the RA predictive models requires that a sigmoidal concentration response curve be available to predict EC50 s and estimate Hillslopes for calculation of the predicted responses. The mouse PPAR␣ reporter provides such response curves, while the human PPAR␣ reporter was considerably less responsive and in most cases sigmoidal concentration response curves were not achieved (Wolf et al., 2008, 2012). The mouse model was therefore the appropriate choice for a study designed to determine additivity of PFAA mixtures. Experimental results from mixture exposures were compared to model predictions to determine whether the PFAA combinations conformed to models of additivity. 2. Materials and methods 2.1. Chemicals WY14,643 (4-chloro-6-(2,3-xylidine)-pyrimidinylthioacetic acid) and dimethylsulfoxide (DMSO) were purchased from Sigma–Aldrich (St. Louis, MO). PFOS (potassium salt; CAS# 2795-39-3; purity ≥98%), PFOA (ammonium salt, CAS# 3825-26-1; purity ≥98%) and PFHxA (CAS# 307-24-4; purity ≥97%) were purchased from Fluka Chemical (Steinheim, Switzerland). PFNA (CAS# 375-95-1; purity 97%) was purchased from Aldrich (St. Louis, MO). PFHxS (potassium salt; purity 98.6%) was a gift from 3 M Company (St. Paul, MN). PFOS, PFOA, PFNA and PFHxS were dissolved in deionized, distilled, filtered water (PicoPure Hydro Services and Supplies, Inc., Durham, NC) to make stock solutions. PFHxA was in liquid form and was added to water to prepare a stock solution. PFOS was added to boiling water and the solution was sonicated for 5 min in a 60◦ water bath. 2.2. Plasmid Mouse PPAR␣ plasmid vector was obtained from Dr. Jeffrey Peters and Dr. John Vanden Heuval (Penn State University, PA) and described previously (Takacs and Abbott, 2007). Briefly, the plasmid is an expression vector that expresses a fusion protein containing a construct of the ligand-binding domain of mouse PPAR␣ fused to the DNA-binding domain of Gal4 under the control of an SV40 promoter, and a construct of a UAS-firefly luciferase reporter under the control of a Gal4 DNA response element. This one-hybrid model was used in previous studies in our laboratory (Wolf et al., 2008, 2012). 2.3. Cell culture, trans-activation assay and experimental design Cell culture and assay were performed as previously described (Wolf et al., 2012) with the following modifications. Concentration range finding experiments were performed on individual PFAAs to determine the appropriate concentrations to use in the mixture experiments. Mixture experiments included binary combinations of PFAAs at selected concentrations. Briefly, COS-1 cells (ATCC, Manassas, VA) were maintained in 75 cm2 flasks at 37 ◦ C and 5% CO2 in DMEM with 10% fetal bovine serum (FBS; Hyclone brand, Gibco, Grand Island, NY) and antibiotic (0.2 mg/mL streptomycin and 200 U/mL penicillin; Gibco). For the assay, cells were plated at 104 cells/well in 96 well plates. Cells were transfected with PPAR␣ reporter plasmid using Fugene 6 transfection reagent (Roche Diagnostics, Indianapolis, IN or Promega,

C.J. Wolf et al. / Toxicology 316 (2014) 43–54 Madison, WI) in serum-free DMEM for 3 h followed by incubation with DMEM + 10% FBS for 24 h. Cells in each plate were treated in serum-free DMEM with either WY14643 (positive control; 10 ␮M), DMSO (WY vehicle control; 0.1%), water (PFAA vehicle control), or PFAAs at the following concentrations. For the range finding experiments, the concentrations were PFOA, 1–256 ␮M; PFNA, 1–256 ␮M; PFHxA, 2–1024 ␮M; PFOS, 2–512 ␮M; and PFHxS, 2–2048 ␮M. For the mixture plates, the concentration ranges were PFOA and PFNA, 1–128 ␮M; PFHxA, 8–1024 ␮M, PFOS, 4–384 ␮M; and PFHxS, 8–2048 ␮M. PFAA concentrations for the mixture experiments were selected to give a sigmoidal concentration response curve without inducing cytotoxicity whether exposure was alone or in combination. Concentration ranges consisted of 8 levels that increased in 2-fold increments, except for PFHxS and PFOS. The PFHxS range included 9 concentrations to accommodate a more complete concentration response curve and was distributed across two sets of plates, with a range of 8–1024 ␮M and a range of 16–2048 ␮M. The PFOS concentration range was in 2-fold increments up to 256 ␮M, and the next concentration was only 1.5-fold higher to avoid toxicity when used in combination with PFOA. For the plate design, binary combinations consisted of PFOA and one other PFAA and were applied to cells in an 8 × 8 factorial grid, allowing one well per combination. Each PFAA of the pair was tested alone at the same concentrations on the same plate. Each plate also included 4 wells each of positive control (WY14643) and DMSO, and 8 wells of water vehicle controls. After 24 h of treatment, cells were rinsed, lysed with Luciferase cell culture lysis reagent (Promega), and luciferase activity was measured in relative light units (RLU) on a Lumistar Galaxy luminometer (BMG Lab Technologies, Durham, NC) with d-luciferin (potassium salt; Promega) and reaction buffer. Experiments were performed across multiple assays with 4 plates per assay and 13–16 plates total per PFAA chemical pair. 2.4. Viability and transfection efficiency tests Each PFAA alone in range finding experiments and each binary PFAA combination was tested once for viability on a separate plate using Cell Titer-Blue® cell viability kit and protocol (Promega). In this assay, viability is confirmed by the ability of living cells to convert a redox dye (rezasurin) into a fluorescent product (resorufin). Transfection efficiency was tested in at least one plate per binary combination by adding secreted alkaline phosphatase (SEAP) control vector (Clontech, Mountain View, CA) to the transfection step and normalizing luciferase activity to SEAP activity as previously described (Takacs and Abbott, 2007). 2.5. Data processing, modeling, and statistics Data in RLU were converted to a fold-induction basis over the mean vehicle control response per plate, and then converted to the percent of the maximal response (% maximal response) per plate. The maximal response of each plate was the mean of responses at the highest concentration of the most potent PFAA in the binary mixture across all concentrations of the other PFAA. In most cases, the most potent PFAA in the mixture was PFOA; in the PFOA + PFNA mixture, the most potent PFAA was PFNA. Percent maximal response data for the individual PFAAs tested alone in the mixture plates were log transformed in GraphPad Prism (v4.02) and analyzed with non-linear regression using sigmoidal dose–response (variable slope) with bottom parameter set at 0 and top parameter set to 100 to generate EC50 s and Hillslopes for each chemical. These parameters were used to calculate predicted responses of each mixture combination per plate. Predicted responses of individual PFAAs are defined by the equation: RPFAA = 1/(1 + HillslopePFAA (EC50 /concPFAA) ) where RPFAA is the predicted response of the PFAA based on Hillslope and EC50 for that individual chemical, and is calculated for each concentration of that PFAA. The RA model equation incorporates these predicted responses of each individual chemical. The equation for calculating the predicted RA response for each binary combination in the mixture is R = 1 − ((1 − RPFOA ) × (1 − RPFAA )) × 100. The CA model equation is derived from Bermudez et al. (2010) and Olmstead and Leblanc (2005). The formula for calculating the predicted concentration addition response for each binary combination in the mixture was R = 1/(1 + (1/avgslope(concPFOA/EC50 PFOA+concPFAA/EC50 PFAA) )) × 100, where “avgslope” is the mean of the Hillslopes calculated for the individual chemicals. Mean observed, RA, and CA data for each binary combination were averaged across all plates and plotted as surface plots in Excel. Curves of PFOA response at each concentration of PFAA were plotted in Prism and analyzed with non-linear regression using sigmoidal response (variable slope) followed by analysis using Akaike’s Information Criteria (AIC) to test whether one global curve fits all data sets; observed, RA and CA.

3. Results Preliminary experiments identified appropriate concentration ranges for PFAAs to be used in the mixture study by producing sigmoidal curves for each PFAA (Fig. 1). Most concentration response curves were similar and fairly parallel. PFHxS induced cytotoxicity at concentrations 2× higher than 2048 ␮M, so the concentration response curve was somewhat incomplete at the high end. All

45

PFAAs were difficult to characterize completely at the higher concentrations due to a steep response curve. None of the individual or mixture data analyzed was from PFAA concentrations that affected viability of COS-1 cells. When compared, the individual concentration response curves illustrate a rough estimate of the relative potencies of the PFAAs. The ranking of these PFAAs on PPAR␣ activity appears to be: PFOA ∼ PFNA > PFHxA > PFOS ∼ PFHxS. The rank order of these PFAAs is similar compared to previous ranking of PFAAs (Wolf et al., 2012). The concentration response curves for PFAAs that were run alone on the mixture plates generated EC50 s and Hillslopes for the modeling of the binary combination of these PFAAs loaded on the same plate. The mean predicted responses of PPAR␣ to the binary combinations of PFOA + PFAA that were modeled by RA and CA are illustrated in surface plots in Fig. 2, along with the observed responses, expressed as the mean % maximal response across all plates. Concentrations on the X and Z axes are labeled in ␮M but are plotted on a logarithmic scale. The plots of the observed responses of each binary combination bear a resemblance to the responses predicted by both RA and CA at low concentrations. Variability of the response at higher concentrations of either chemical is also evident in these graphs. In addition, the shapes of the surface plots of observed data are not identical for all PFAA combinations at the higher concentrations. The plots of observed data for the combined carboxylates, PFNA + PFOA and PFHxA + PFOA, appear similar to each other but different from the sulfonates. The plots of observed data for the sulfonate combinations, PFOS + PFOA and PFHxS + PFOA, appear similar, both displaying the same feature of a spike at the highest concentrations. The responses illustrated in the surface plots are further illustrated as separate line graphs of the response to PFOA at each concentration of its paired PFAA in Figs. 3–6. For each PFOA + PFAA pair, the curves for observed data are similar to the curves for RA and CA models at lower concentrations. The probability or percent likelihood that a single global curve fits the RA, CA, and observed data sets is shown for each PFOA + PFAA pair for three different ranges of PFOA concentrations in Table 1. Statistically, all curves fit one global curve in the concentration range of 1–32 ␮M PFOA in combination with every concentration of PFAA for all binary combinations, except for PFOA + PFHxA. For PFOA + PFHxA, curves for modeled data and observed data fit one global curve only at 1–32 ␮M PFOA and at PFHxA concentrations of 8 ␮M and 32–256 ␮M, excluding the two highest concentrations of both PFOA and PFHxA. For each PFOA + PFAA combination, observed responses are often higher than predicted responses at higher concentrations, but the pronounced variability at these concentrations makes their relevance questionable.

4. Discussion Various perfluoroalkyl acids coexist in the environment and are known to exert toxic effects, yet information on mixture effects of perfluoroalkyl acids is often limited. In vitro mixture studies of PFAAs have often focused on toxicity as an endpoint or included only PFOA and PFOS (Ding et al., 2013; Hu and Hu, 2009; RodeaPalomares et al., 2012). The current study provides an analysis of the mixture effects of the PFAAs commonly found in human serum (PFOA, PFOS, PFNA, and PFHxS) and PFHxA in binary combinations with PFOA in each mixture, on PPAR␣ activation, a known component of the mechanistic pathway of PFAA responses. As noted in the introduction, we selected a mouse one-hybrid model to evaluate responses as it has proven to be highly responsive to PFAA activation and is the most suitable model for this analysis. We characterized the individual concentration response relationships of the PFAAs by fitting the response to sigmoidal curves on a logarithmic

46

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

Fig. 1. Individual concentration response curves for the activation of mouse PPAR␣ by (a) PFOA, (b) PFNA, (c) PFHxA, (d) PFOS, and (e) PFHxS. COS-1 cells were transiently transfected with PPAR␣ plasmid and exposed to each PFAA for 24 h at concentrations listed below each plot. Each PFAA was tested in at least 2 plates with 4–8 replicate wells per plate. Curves were generated for determination of appropriate concentration ranges to use for combination exposure and not for statistical analysis.

scale. The differences between the curves reflect a relative potency of the PFAAs that agree with our previous studies (Wolf et al., 2008, 2010). Based on the EC50 s and Hillslopes of the curves, we tested the mixture behavior of binary combinations of these PFAAs on PPAR␣ activation by comparing observed effects with effects predicted by response addition (RA) or concentration addition (CA) models. The predicted responses of the RA and CA models and the observed data were generally the same. The convergence of predictions from the two additivity models is not unique to this study (Cedergreen et al., 2008; Hannas et al., 2012). Continuous endpoints that do not

exhibit clear thresholds of response often result in similar predictions from dose addition and response addition models, especially in the low dose region. We found that binary combinations of PFAAs (PFOA + PFNA, PFOA + PFOS, PFOA + PFHxA or PFOA + PFHxS) appear to behave additively in the lower PFOA concentration range of 1–32 ␮M. At higher PFOA concentrations with PFAAs, the observed responses exceeded the predicted responses using either RA or CA. We found a rank order of potency of the PFAAs that was similar to our previous findings. Whereas our previous studies (Wolf et al., 2008, 2012) assessed relative potencies of PFAAs on a linear

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

47

Fig. 2. Surface plots of RA-predicted (left column), CA-predicted (middle column), and observed responses (right column) of PPAR␣ activation to binary combinations of (a) PFOA + PFNA, (b) PFOA + PFHxA, (c) PFOA + PFOS, (d) PFOA + PFHxS. X- and Z-axes are scaled logarithmically and labeled in units of concentrations used (␮M). Responses shown on Y axis is %maximal response (calculated as detailed in Section 2).

concentration response curve, the current study uses the complete log-scale sigmoidal concentration response curve of PPAR␣ activation to characterize each PFAA and predict the mixture responses. It is evident by subjective observation of the individual curves that PFOA and PFNA were the most potent in activating PPAR␣, followed by PFHxA, PFOS and PFHxS. These results agree with our previous conclusion that activity increases with increasing chain length and is lower in sulfonates compared to carboxylates. These results are also similar to the ranking of PFAAs based on PPAR␣ activity reported by others (Bjork and Wallace, 2009; Buhrke et al., 2013). Our finding that PFAAs behave in an additive fashion in mixtures is in general agreement with other studies on the action of PFAA mixtures. Ding et al. (2013) found that mixtures of PFOA and

PFOS behave additively by both CA and IA (independent action) on zebrafish toxicity at lower concentration ranges, but that observed results deviate from the models at higher concentrations. Their results deviated in measurable ways to indicate synergy, antagonism or additivity between PFOA and PFOS at high concentrations that were dependent on the PFOA:PFOS ratios. Although our study also found that the observed results deviated from the models at high concentrations, it was usually greater than additive and our models did not assess whether these deviations were synergistic. The differences between our studies could be explained by complexity of the biological system and the endpoints measured. When using in vivo models, such as zebrafish, receptors other than PPAR␣ or other biological processes could come into play and contribute to

48

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

Fig. 3. Observed and predicted responses of PPAR␣ activation to binary combinations of PFAAs showing the PFOA concentration response curve at each concentration of PFNA. Observed (), RA (), and CA () data. Data were fitted with non-linear regression, sigmoidal dose–response (variable slope). Responses on Y axis are in % maximal response (calculated as detailed in Section 2).

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

49

Fig. 4. Observed and predicted responses of PPAR␣ activation to binary combinations of PFAAs showing the PFOA concentration response curve at each concentration of PFHxA. Observed (), RA (), and CA () data. Data were fitted with non-linear regression, sigmoidal dose–response (variable slope). Responses on Y axis are in % maximal response (calculated as detailed in Section 2).

50

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

Fig. 5. Observed and predicted responses of PPAR␣ activation to binary combinations of PFAAs showing the PFOA concentration response curve at each concentration of PFOS. Observed (), RA (), and CA () data. Data were fitted with non-linear regression, sigmoidal dose–response (variable slope). Responses on Y axis are in % maximal response (calculated as detailed in Section 2).

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

51

Fig. 6. Observed and predicted responses of PPAR␣ activation to binary combinations of PFAAs showing the PFOA concentration response curve at each concentration of PFHxS. Observed (), RA (), and CA () data. Data were fitted with non-linear regression, sigmoidal dose–response (variable slope). Responses on Y axis are in % maximal response (calculated as detailed in Section 2).

interaction between the PFAAs. In addition, Ding et al. (2013) noted increased toxicity with increasing ratios of PFOS. In other studies using HepG2 cells, Hu and Hu (2009) scored apoptotic effects of mixtures of PFOA and PFOS (50 + 50 ␮mol/L, 150 + 150 ␮mol/L, and 200 + 200 ␮mol/L, PFOA + PFOS) and concluded that the combined effect was additive, and not synergistic or antagonistic. In a study using our transfected COS-1 cell model, Carr et al. (2013) tested PFAAs in a four chemical fixed ratio mixture based on that found in human serum (Kato et al., 2011). These investigators also examined binary combinations of these PFAAs, using the same fixed ratios. They concluded that this fixed ratio mixture of PFOA, PFNA, PFOS and PFHxS had an additive or somewhat less than additive interaction with a tendency to deviate in the direction of antagonism. In contrast, some investigators reported that the response of co-exposure to various PFAAs was other than additive. In cyanobacterium, the binary combination of PFOA + PFOS behaved in an antagonistic fashion in inducing toxicity (Rodea-Palomares et al., 2012). However, in that study, mixtures of PFOA + PFOS with other pollutants were reported to behave synergistically, additively, or

antagonistically depending on the composition of the mixture. Wei et al. (2009) used cultured hepatocytes from minnows to study mixtures of PFAAs, evaluating cell viability and effects on gene expression. In that study, mixtures of PFAAs, including a mixture of PFOA + PFOS, induced gene sets in the hepatocytes that differed from the gene sets induced by each PFAA alone. Wei et al.’s study further illustrates that the modes of action subsequent to co-exposure to the different PFAAs are likely to compete and to involve multiple molecular signaling pathways and cellular functions, and that these responses may differ for mixtures versus individual chemicals. Alternately, some of the genes induced by PFOA or PFOS alone were not induced by the PFOA + PFOS mixture. The difference in gene expression by the mixture was interpreted to indicate interaction between PFOA and PFOS, not additivity. The different results found in mixture behavior of PFAAs between these various studies could be largely attributed to the biological model, the concentrations used, or the particular composition or complexity of the mixture. At the higher concentrations of each PFAA in a mixture in the current study, the observed responses deviated from the RA and CA

52

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

Table 1 Probability (percent likelihood) that one curve fits RA, CA, and observed data for PPAR-alpha activation by PFOA + PFAA combination exposure in COS-1 cells. PFOA concentration range analyzed

PFNA

PFOA 1–32 ␮M 1–64 ␮M 1–128 ␮M

1 ␮M

2 ␮M

4 ␮M

8 ␮M

16 ␮M

32 ␮M

64 ␮M

128 ␮M

98.53* 99.99* >99.99*

>99.99* 31.52 9.33

>99.99* 99.99* 99.98* 99.99* 99.94* 69.16

DNC DNC >99.99*

DNC DNC DNC

PFOS

PFOA 1–32 ␮M 1–64 ␮M 1–128 ␮M

4 ␮M

8 ␮M

16 ␮M

32 ␮M

64 ␮M

128 ␮M

256 ␮M

384 ␮M

>99.99* 99.99* 99.99* 7.19 99.99* 99.99* 99.99* 71.98 0.02

Responses are considered to fit one curve if probability > 95% by Akaike’s Information Criteria. DNC, does not converge.

models of additivity. The RA and CA predictions were based on the individual concentration response curves. Although the PFAAs in the current study have been characterized individually with concentration response curves, the upper ends of the individual curves in some cases may not reach a plateau as expected for a complete sigmoidal fit, due to their toxicity at concentrations higher than that used in the study. This may be the major factor contributing to the difference between the predicted responses and the observed responses at high concentrations. The observed mixture response curves at the higher concentrations were also more variable. Although it is possible that this variability could be reduced with a higher number of trials for higher statistical power, there was little variability at lower concentrations. Variability of the mixture responses at high concentrations could be due to interaction at higher concentrations not found at lower concentrations, as demonstrated by others (Ding et al., 2013; Wei et al., 2009). This variability could also be due to physical properties such as low solubility, transportation into the cell, or receptor kinetics. Mixture studies have been well defined for estrogens. The estrogenic chemicals work well in mixture analysis as they exhibit a complete, well characterized concentration response curve with no toxicity at the plateau phase of the curve (Bermudez et al., 2010; Olsen et al., 2003; Wilson et al., 2004). The accuracy of the curve leads to a more accurate prediction of the mixture response. Regardless of the reason for the variability at the higher end of the concentration range, it should be noted that the concentrations of PFAAs found in human serum (Kato et al., 2011) are lower than all concentrations used in the current study, thus the response data are more relevant at the lower concentrations and become less relevant at the higher end of the concentration range. While this one-hybrid model for PPAR␣ transactivation is useful for screening PPAR␣ ligands and for deciphering the behavior of more than one of these ligands in mixtures, limitations of this model are important to consider. The vector used in this model

contains only the ligand binding domain (LBD) of mouse PPAR␣ and the magnitude of the activation signal is dependent on the SV40 promotor, not the normal PPAR␣ promoter. Activation of PPAR␣ in vivo also usually occurs in heterodimerization with other receptors such as PXR or RXR, which are not in this model. In addition, the concentrations of PFAAs required to activate the receptor vector to a useable degree in this system are orders of magnitude higher than concentrations found in human serum. Endogenous ligands possibly released by PFAAs may also activate this system, although the concentrations of PFAAs used in the current study are much greater than any endogenous ligands that could be produced, so interference via endogenous ligands is unlikely. It is important to note, when comparing this study with other reports of effects of mixtures of PFAAs that differences between these model systems exist and must be considered, as they can impact the outcomes. In summary, our study assessed whether binary combinations of PFOA, PFNA, PFOS, PFHxA, and PFHxS behave additively in activating PPAR␣ in transiently transfected COS-1 cells by modeling their behavior with response addition and concentration addition. We found that these PFAAs in binary combinations with PFOA behave in an additive fashion on PPAR␣ at low concentrations of PFOA (1–32 ␮M) by both predictive models. These findings suggest that other PFAAs may also behave additively at low concentrations in mixtures and are informative for human health cumulative risk assessment. The current study is one of only a few studies investigating mixture effects of PFAAs, and our results are in general agreement with other studies that concluded additivity of PFAA mixtures.

Conflict of interest The authors declare that there are no conflicts of interest.

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

Acknowledgements The authors gratefully acknowledge the technical advice of Dr. Vickie Wilson and Kathy Bobseine and laboratory assistance of Andrew Watkins, and careful review of the manuscript by Dr. Jennifer Seed and Dr. Bethany Hannas. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.tox.2013.12.002. References Abbott, B.D., Wolf, C.J., Das, K.P., Zehr, R.D., Schmid, J.E., Lindstrom, A.B., Strynar, M.J., Lau, C., 2009. Developmental toxicity of perfluorooctane sulfonate (PFOS) is not dependent on expression of peroxisome proliferator activated receptor-alpha (PPAR alpha) in the mouse. Reprod. Toxicol. 27, 258–265. Abbott, B.D., Wolf, C.J., Schmid, J.E., Das, K.P., Zehr, R.D., Helfant, L., Nakayama, S., Lindstrom, A.B., Strynar, M.J., Lau, C., 2007. Perfluorooctanoic acid induced developmental toxicity in the mouse is dependent on expression of peroxisome proliferator activated receptor-alpha. Toxicol. Sci. 98, 571–581. Bermudez, D.S., Gray Jr., L.E., Wilson, V.S., 2010. Modeling the interaction of binary and ternary mixtures of estradiol with bisphenol A and bisphenol AF in an in vitro estrogen-mediated transcriptional activation assay (T47D-KBluc). Toxicol. Sci. 116, 477–487. Bijland, S., Rensen, P.C., Pieterman, E.J., Maas, A.C., van der Hoorn, J.W., van Erk, M.J., Havekes, L.M., Willems van Dijk, K., Chang, S.C., Ehresman, D.J., Butenhoff, J.L., Princen, H.M., 2011. Perfluoroalkyl sulfonates cause alkyl chain lengthdependent hepatic steatosis and hypolipidemia mainly by impairing lipoprotein production in APOE*3-Leiden CETP mice. Toxicol. Sci. 123, 290–303. Bjork, J.A., Butenhoff, J.L., Wallace, K.B., 2011. Multiplicity of nuclear receptor activation by PFOA and PFOS in primary human and rodent hepatocytes. Toxicology 288, 8–17. Bjork, J.A., Wallace, K.B., 2009. Structure-activity relationships and human relevance for perfluoroalkyl acid-induced transcriptional activation of peroxisome proliferation in liver cell cultures. Toxicol. Sci. 111, 89–99. Buck, R.C., Franklin, J., Berger, U., Conder, J.M., Cousins, I.T., de Voogt, P., Jensen, A.A., Kannan, K., Mabury, S.A., van Leeuwen, S.P., 2011. Perfluoroalkyl and polyfluoroalkyl substances in the environment: terminology, classification, and origins. Integr. Environ. Assess. Manag. 7, 513–541. Buhrke, T., Kibellus, A., Lampen, A., 2013. In vitro toxicological characterization of perfluorinated carboxylic acids with different carbon chain lengths. Toxicol. Lett. 218, 97–104. Butt, C.M., Berger, U., Bossi, R., Tomy, G.T., 2010. Levels and trends of poly- and perfluorinated compounds in the arctic environment. Sci. Total Environ. 408, 2936–2965. Calafat, A.M., Wong, L.Y., Kuklenyik, Z., Reidy, J.A., Needham, L.L., 2007. Polyfluoroalkyl chemicals in the U.S. population: data from the National Health and Nutrition Examination Survey (NHANES) 2003–2004 and comparisons with NHANES 1999–2000. Environ. Health Perspect. 115, 1596–1602. Carr, C.K., Watkins, A.M., Wolf, C.J., Abbott, B.D., Lau, C., Gennings, C., 2013. Testing for departures from additivity in mixtures of perfluoroalkyl acids (PFAAs). Toxicology 306, 169–175. Cedergreen, N., Christensen, A.M., Kamper, A., Kudsk, P., Mathiassen, S.K., Streibig, J.C., Sorensen, H., 2008. A review of independent action compared to concentration addition as reference models for mixtures of compounds with different molecular target sites. Environ. Toxicol. Chem. 27, 1621–1632. Cheng, X., Klaassen, C.D., 2008. Perfluorocarboxylic acids induce cytochrome P450 enzymes in mouse liver through activation of PPAR-alpha and CAR transcription factors. Toxicol. Sci. 106, 29–36. Cwinn, M.A., Jones, S.P., Kennedy, S.W., 2008. Exposure to perfluorooctane sulfonate or fenofibrate causes PPAR-alpha dependent transcriptional responses in chicken embryo hepatocytes. Comp. Biochem. Physiol. C Toxicol. Pharmacol. 148, 165–171. Ding, G., Zhang, J., Chen, Y., Wang, L., Wang, M., Xiong, D., Sun, Y., 2013. Combined Effects of PFOS and PFOA on Zebrafish (Danio rerio) embryos. Arch. Environ. Contam. Toxicol. 64, 668–675. Elcombe, C.R., Elcombe, B.M., Foster, J.R., Chang, S.C., Ehresman, D.J., Butenhoff, J.L., 2012. Hepatocellular hypertrophy and cell proliferation in Sprague-Dawley rats from dietary exposure to potassium perfluorooctanesulfonate results from increased expression of xenosensor nuclear receptors PPARalpha and CAR/PXR. Toxicology 293, 16–29. Elcombe, C.R., Elcombe, B.M., Foster, J.R., Farrar, D.G., Jung, R., Chang, S.C., Kennedy, G.L., Butenhoff, J.L., 2010. Hepatocellular hypertrophy and cell proliferation in Sprague-Dawley rats following dietary exposure to ammonium perfluorooctanoate occurs through increased activation of the xenosensor nuclear receptors PPARalpha and CAR/PXR. Arch. Toxicol. 84, 787–798. Eschauzier, C., Hoppe, M., Schlummer, M., de Voogt, P., 2013. Presence and sources of anthropogenic perfluoroalkyl acids in high-consumption tap-water based beverages. Chemosphere 90, 36–41.

53

Fromme, H., Schlummer, M., Moller, A., Gruber, L., Wolz, G., Ungewiss, J., Bohmer, S., Dekant, W., Mayer, R., Liebl, B., Twardella, D., 2007. Exposure of an adult population to perfluorinated substances using duplicate diet portions and biomonitoring data. Environ. Sci. Technol. 41, 7928–7933. Grasty, R.C., Wolf, D.C., Grey, B.E., Lau, C.S., Rogers, J.M., 2003. Prenatal window of susceptibility to perfluorooctane sulfonate-induced neonatal mortality in the Sprague-Dawley rat. Birth Defects Res. B Dev. Reprod. Toxicol. 68, 465–471. Greco, W.R., Bravo, G., Parsons, J.C., 1995. The search for synergy: a critical review from a response surface perspective. Pharmacol. Rev. 47, 331–385. Hannas, B.R., Lambright, C.S., Furr, J., Evans, N., Foster, P.M., Gray, E.L., Wilson, V.S., 2012. Genomic biomarkers of phthalate-induced male reproductive developmental toxicity: a targeted RT-PCR array approach for defining relative potency. Toxicol. Sci. 125, 544–557. Hu, X.Z., Hu, D.C., 2009. Effects of perfluorooctanoate and perfluorooctane sulfonate exposure on hepatoma Hep G2 cells. Arch. Toxicol. 83, 851–861. Jain, R.B., 2013. Effect of pregnancy on the levels of selected perfluoroalkyl compounds for females aged 17–39 years: data from National Health and Nutrition Examination Survey 2003–2008. J. Toxicol. Environ. Health A 76, 409–421. Kato, K., Wong, L.Y., Jia, L.T., Kuklenyik, Z., Calafat, A.M., 2011. Trends in exposure to polyfluoroalkyl chemicals in the U.S population: 1999–2008. Environ. Sci. Technol. 45, 8037–8045. Kissa, E., 2001. Fluorinated Surfactants and Repellents. Marcel Dekker, New York. Lau, C., Anitole, K., Hodes, C., Lai, D., Pfahles-Hutchens, A., Seed, J., 2007. Perfluoroalkyl acids: a review of monitoring and toxicological findings. Toxicol. Sci. 99, 366–394. Lau, C., Thibodeaux, J.R., Hanson, R.G., Narotsky, M.G., Rogers, J.M., Lindstrom, A.B., Strynar, M.J., 2006. Effects of perfluorooctanoic acid exposure during pregnancy in the mouse. Toxicol. Sci. 90, 510–518. Lau, C., Thibodeaux, J.R., Hanson, R.G., Rogers, J.M., Grey, B.E., Stanton, M.E., Butenhoff, J.L., Stevenson, L.A., 2003. Exposure to perfluorooctane sulfonate during pregnancy in rat and mouse. II: postnatal evaluation. Toxicol. Sci. 74, 382–392. Maloney, E.K., Waxman, D.J., 1999. Trans-activation of PPARalpha and PPARgamma by structurally diverse environmental chemicals. Toxicol. Appl. Pharmacol. 161, 209–218. Ode, A., Rylander, L., Lindh, C.H., Kallen, K., Jonsson, B.A., Gustafsson, P., Olofsson, P., Ivarsson, S.A., Rignell-Hydbom, A., 2013. Determinants of maternal and fetal exposure and temporal trends of perfluorinated compounds. Environ. Sci. Pollut. Res. Int. 20, 7970–7978. Olmstead, A.W., Leblanc, G.A., 2005. Joint action of polycyclic aromatic hydrocarbons: predictive modeling of sublethal toxicity. Aquat. Toxicol. 75, 253–262. Olsen, C.M., Meussen-Elholm, E.T., Samuelsen, M., Holme, J.A., Hongslo, J.K., 2003. Effects of the environmental oestrogens bisphenol A, tetrachlorobisphenol A, tetrabromobisphenol A, 4-hydroxybiphenyl and 4,4’-dihydroxybiphenyl on oestrogen receptor binding, cell proliferation and regulation of oestrogen sensitive proteins in the human breast cancer cell line MCF-7. Pharmacol. Toxicol. 92, 180–188. Orton, F., Rosivatz, E., Scholze, M., Kortenkamp, A., 2011. Widely used pesticides with previously unknown endocrine activity revealed as in vitro antiandrogens. Environ. Health Perspect. 119, 794–800. Renner, R., 2001. Growing concern over perfluorinated chemicals. Environ. Sci. Technol. 35, 154A–160A. Rider, C.V., LeBlanc, G.A., 2005. An integrated addition and interaction model for assessing toxicity of chemical mixtures. Toxicol. Sci. 87, 520–528. Rodea-Palomares, I., Leganes, F., Rosal, R., Fernandez-Pinas, F., 2012. Toxicological interactions of perfluorooctane sulfonic acid (PFOS) and perfluorooctanoic acid (PFOA) with selected pollutants. J. Hazard. Mater. 201–202, 209–218. Rosen, M.B., Lee, J.S., Ren, H., Vallanat, B., Liu, J., Waalkes, M.P., Abbott, B.D., Lau, C., Corton, J.C., 2008. Toxicogenomic dissection of the perfluorooctanoic acid transcript profile in mouse liver: evidence for the involvement of nuclear receptors PPAR alpha and CAR. Toxicol. Sci. 103, 46–56. Rosen, M.B., Schmid, J.E., Das, K.P., Wood, C.R., Zehr, R.D., Lau, C., 2009. Gene expression profiling in the liver and lung of perfluorooctane sulfonate-exposed mouse fetuses: comparison to changes induced by exposure to perfluorooctanoic acid. Reprod. Toxicol. 27, 278–288. Rosen, M.B., Schmid, J.R., Corton, J.C., Zehr, R.D., Das, K.P., Abbott, B.D., Lau, C., 2010. Gene expression profiling in wild-type and PPARalpha-null mice exposed to perfluorooctane sulfonate reveals PPARalpha-independent effects. PPAR Res. 2010, http://dx.doi.org/10.1155/2010/794739, Epub 2010 Sep 27. Rosen, M.B., Thibodeaux, J.R., Wood, C.R., Zehr, R.D., Schmid, J.E., Lau, C., 2007. Gene expression profiling in the lung and liver of PFOA-exposed mouse fetuses. Toxicology 239, 15–33. Seacat, A.M., Thomford, P.J., Hansen, K.J., Clemen, L.A., Eldridge, S.R., Elcombe, C.R., Butenhoff, J.L., 2003. Sub-chronic dietary toxicity of potassium perfluorooctanesulfonate in rats. Toxicology 183, 117–131. Seacat, A.M., Thomford, P.J., Hansen, K.J., Olsen, G.W., Case, M.T., Butenhoff, J.L., 2002. Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in cynomolgus monkeys. Toxicol. Sci. 68, 249–264. Shipley, J.M., Hurst, C.H., Tanaka, S.S., DeRoos, F.L., Butenhoff, J.L., Seacat, A.M., Waxman, D.J., 2004. trans-activation of PPARalpha and induction of PPARalpha target genes by perfluorooctane-based chemicals. Toxicol. Sci. 80, 151–160. Takacs, M.L., Abbott, B.D., 2007. Activation of mouse and human peroxisome proliferator-activated receptors (alpha, beta/delta, gamma) by perfluorooctanoic acid and perfluorooctane sulfonate. Toxicol. Sci. 95, 108–117.

54

C.J. Wolf et al. / Toxicology 316 (2014) 43–54

Vanden Heuvel, J.P., Thompson, J.T., Frame, S.R., Gillies, P.J., 2006. Differential activation of nuclear receptors by perfluorinated fatty acid analogs and natural fatty acids: a comparison of human, mouse, and rat peroxisome proliferatoractivated receptor-alpha, -beta, and -gamma, liver X receptor-beta, and retinoid X receptor-alpha. Toxicol. Sci. 92, 476–489. Vestergren, R., Berger, U., Glynn, A., Cousins, I.T., 2012. Dietary exposure to perfluoroalkyl acids for the Swedish population in 1999, 2005 and 2010. Environ. Int. 49, 120–127. Wei, Y., Shi, X., Zhang, H., Wang, J., Zhou, B., Dai, J., 2009. Combined effects of polyfluorinated and perfluorinated compounds on primary cultured hepatocytes from rare minnow (Gobiocypris rarus) using toxicogenomic analysis. Aquat. Toxicol. 95, 27–36. Wilson, V.S., Bobseine, K., Gray Jr., L.E., 2004. Development and characterization of a cell line that stably expresses an estrogen-responsive luciferase reporter

for the detection of estrogen receptor agonist and antagonists. Toxicol. Sci. 81, 69–77. Wolf, C.J., Schmid, J.E., Lau, C., Abbott, B.D., 2012. Activation of mouse and human peroxisome proliferator-activated receptor-alpha (PPARalpha) by perfluoroalkyl acids (PFAAs): further investigation of C4–C12 compounds. Reprod. Toxicol. 33, 546–551. Wolf, C.J., Takacs, M.L., Schmid, J.E., Lau, C., Abbott, B.D., 2008. Activation of mouse and human peroxisome proliferator-activated receptor alpha by perfluoroalkyl acids of different functional groups and chain lengths. Toxicol. Sci. 106, 162–171. Wolf, C.J., Zehr, R.D., Schmid, J.E., Lau, C., Abbott, B.D., 2010. Developmental effects of perfluorononanoic acid in the mouse are dependent on peroxisome proliferator-activated receptor-alpha. PPAR Res. 2010, http://dx.doi.org/10.1155/2010/282896, Epub 2010 Sep 27.