Effects of PFAS on human liver transporters: implications for health outcomes

Abstract

Per- and polyfluoroalkyl substances (PFAS) have become internationally recognized over the past three decades as persistent organic pollutants used in the production of various consumer and industrial goods. Research efforts continue to gauge the risk that historically used, and newly produced, PFAS may cause to human health. Numerous studies report toxic effects of PFAS on the human liver as well as increased serum cholesterol levels in adults. A major concern with PFAS, also dubbed “forever chemicals,” is that they accumulate in the liver and kidney and persist in serum. The mechanisms responsible for their disposition and excretion in humans are poorly understood. A better understanding of the interaction of PFAS with liver transporters, as it pertains to the disposition of PFAS and other xenobiotics, could provide mechanistic insight into human health effects and guide efforts toward risk assessment of compounds in development. This review summarizes the current state of the literature on the emerging relationships (eg, substrates, inhibitors, modulators of gene expression) between PFAS and specific hepatic transporters. The adaptive and toxicological responses of hepatocytes to PFAS that reveal linkages to pathologies and epidemiological findings are highlighted. The evidence suggests that our understanding of the molecular landscape of PFAS must improve to determine their impact on the expression and function of hepatocyte transporters that play a key role in PFAS or other xenobiotic disposition. From here, we can assess what role these changes may have in documented human health outcomes.

perfluorinated chemicals, xenobiotic transporters, biliary excretion, toxicokinetics, hepatotoxicity

Per- and polyfluoroalkyl substances (PFAS) are a class of persistent organic pollutants that have been used since the 1940s in a variety of manufacturing processes for consumer products and industrial or military applications. The inclusion of certain PFAS in these processes has been phased out since the 2009 Stockholm Convention. However, due to their stable chemical structures, these compounds tend to persist in the environment and throughout ecological networks. Since this ban, new PFAS with differing chemical structures have been manufactured and introduced into the environment with limited or no publicly available data on their potential risks to human health (Glüge et al., 2020). Humans commonly are exposed to PFAS via food, drinking water, or dust, and exposure levels can vary depending on region and occupation (Sinclair et al., 2020; Sunderland et al., 2019). PFAS are frequently categorized by their chemical structures and historical usage (ie, legacy vs emerging) (Table 1).

Table 1.

Structures of legacy and emerging PFAS.

Legacy perfluorinated alkyl substances
Perfluoroalkyl carboxylic acids
Structure	n	Abbreviation		Full name
	2 4 5 6 7 8 9 11	PFBA PFHxA PFHpA PFOA PFNA PFDA PFUnDA PFTrDA		Perfluorobutanoic acid Perfluorohexanoic acid Perfluoroheptanoic acid Perfluorooctanoic acid Perfluorononanoic acid Perfluorodecanoic acid Perfluoroundecanoic acid Perfluorotridecanoic acid
Perfluoroalkane sulfonic acids
Structure	n	Abbreviation		Full name
	3 5 7	PFBS PFHxS PFOS		Perfluorobutane sulfonic acid Perfluorohexane sulfonic acid Perfluorooctane sulfonic acid
Precursors to perfluorooctane sulfonic acid
Structure		R	Abbreviation	Full name
		—	6:2 FTS	6:2 fluorotelomer sulfonic acid
		H CH3 CH2CH3	FOSA MetFOSA EtFOSA	Perfluorooctane sulfonamide Methyl perfluorooctane sulfonamide Ethyl perfluorooctane sulfonamide
Emerging per- and polyfluoroalkyl ether-based substances
Structure		n	Abbreviation	Full name∗
		1 2 3	HFPO-DA HFPO-TA HFPO-TeA	Hexafluoropropylene oxide dimer acid Hexafluoropropylene oxide trimer acid Hexafluoropropylene oxide tetramer acid
		—	H-PFMO2OSA	7-hydro-perfluoro-4-methyl-3,6-dioxooctane sulfonic acid
		—	PFMO2HpA	Perfluoro-2-methyl-3,6-dioxo-heptanoic acid
		—	PFMO3NA	Perfluoro-2-methyl-3,6,8-trioxo-nonanoic acid

Classification and full chemical names of PFAS discussed in this review. PFAS may be classified as perfluoroalkyl acids or per- and polyfluoroalkyl ether-based substances. The legacy compounds are classified as perfluoroalkyl acids, which are further subdivided into perfluoroalkyl carboxylic acids (PFCAs) and perfluoroalkane sulfonic acids (PFSAs). Several compounds are environmental precursors to the PFSA, perfluorooctane sulfonic acid. Manufacturers creating emerging compounds tend to avoid alkylated backbones and instead often create ether-based fluorinated chemicals. Created with ChemDraw Professional software.

*

There may be discrepancies in the nomenclature used for emerging PFAS, but for the purpose of this review, all chemicals are named based on previously published studies.

PFAS have been associated with a variety of human health effects including developmental delay, reduction in vaccine responsiveness, thyroid disease, kidney and testicular cancer, liver toxicity, and increased serum cholesterol, as reviewed in Fenton et al. (2021). In particular, elevated serum cholesterol is a finding that could predispose populations with high PFAS exposure to an increased risk of cardiovascular disease, although no such association has been found (Andersen et al., 2021). Epidemiologic and mechanistic studies have demonstrated that PFAS can associate with or induce dyslipidemia, leading to perturbations in circulating levels of fatty acids and cholesterol, and hepatic steatosis (Dunder et al., 2022; Qi et al., 2023). After exposure, PFAS tend to accumulate in the blood, liver, kidney, and lung in humans (Pérez et al., 2013). Long-chain legacy PFAS, in particular PFOA and PFOS, tend to accumulate in the liver (Pérez et al., 2013; Pizzurro et al., 2019). This warrants a closer examination into the propensity for certain PFAS to accumulate in the liver over other organs, and how they affect specific molecular pathways related to liver-specific outcomes.

A number of studies have evaluated liver-specific outcomes in both epidemiological analyses and experimental models. In humans, PFAS exposure may have positive correlations with liver function biomarkers, metabolic dysfunction-associated steatotic liver disease (MASLD; previously, nonalcoholic fatty liver disease, NAFLD), and liver cancer, though these variables are inconsistently associated with PFAS levels (Bassler et al., 2019; Borghese et al., 2022; Cao et al., 2022b; Costello et al., 2022; Qi et al., 2023; Sen et al., 2022; Sinclair et al., 2020; Zhang et al., 2023). Animal models frequently show increased liver weight, steatosis, fibrosis, and cholestasis after PFAS dosing; however, the high-dose exposures make it difficult to extrapolate what these responses may mean for human risk assessment (Guo et al., 2022; Zhang et al., 2018). Interspecies toxicokinetic differences also play a key role in the difficulty underlying extrapolation from animal models to humans (Pizzurro et al., 2019). Although human epidemiological studies report increases in serum cholesterol, animal models react differently to PFAS exposure with decreased serum cholesterol, as reviewed in Fragki et al. (2021). Some emerging PFAS undergo biotransformation in vivo, which necessitates species comparison in metabolic pathways, though these pathways tend to be conserved across species (Kolanczyk et al., 2023). However, certain species such as rats differ in endogenous metabolic pathways that may confound human relevancy of PFAS exposure outcomes based on rodent data (Blais et al., 2017).

Besides these complications, the dose and duration of PFAS exposure in humans may influence PFAS toxicokinetics. A phase I dose-escalation trial of PFOA administered to cancer patients reported a much shorter half-life and a statistically significant decrease in serum cholesterol, which differed from correlative association studies and animal studies (Convertino et al., 2018). Additionally, biomonitoring studies on occupational workers exposed to high levels of PFAS have also occasionally reported no change or a reduction in serum cholesterol (Toxicological Profile for Perfluoroalkyls, 2018). These findings indicate that different toxicokinetic phenomena occur at varying levels and duration of PFAS exposure that could be attributed to saturable mechanisms (ie, transporter interactions).

Hepatocyte transporters are essential mediators of bile acid homeostasis as well as drug and xenobiotic disposition and excretion. The localization of each transporter discussed in this review is depicted in Figure 1. Due to their structure, many legacy PFAS are ionized at physiological pH, which may limit passive diffusion (Cao et al., 2022a). The interaction between PFAS and hepatocyte transporters becomes important when considering why some of these chemicals tend to accumulate in the liver and plasma, and why rates of excretion become protracted.

Figure 1.

Localization of proteins in hepatocytes and a summary of the general trends in transporter function or regulation of mRNA expression reported for the interaction between PFAS and hepatocyte transporters in human (uppercase) and in rodent (lowercase) models. Red boxes indicate that one or more PFAS (brown circles) act as substrates for the highlighted transporter, as discussed in this review. Red “X” symbols indicate that PFAS have been shown to inhibit transport of substrates of indicated transporters. Blue: uptake transporters, purple: basolateral efflux transporters, green: apical efflux transporters. BCRP/Bcrp, breast cancer resistance protein; BSEP/Bsep, bile salt export pump; CD36, fatty acid translocase; FAs, fatty acids; FABP, fatty acid binding protein; FATP, fatty acid transport protein; MATE1/Mate1, multidrug and toxin extrusion protein; MCT1, monocarboxylate transporter; MRP/Mrp, multidrug resistance-associated protein; NTCP/Ntcp, sodium-taurocholate co-transporting polypeptide; OATP/Oatp, organic anion transporting polypeptide; OAT/Oat, organic anion transporter; OCT/Oct, organic cation transporter; OSTα/β/Ostα/β, organic solute transporter; Octn2, novel organic cation transporter 2; MDR1 P-gp/Mdr1a/b, multidrug resistance protein, also known as P-glycoprotein. Created with BioRender.com.

Our understanding of bile acid/xenobiotic transport protein interactions with PFAS as substrates or inhibitors, and the dynamics of PFAS control over transporter expression and membrane localization, is at an early stage. Recent publications have demonstrated that incorporating transporter kinetic parameters into physiologically based toxicokinetic (PBTK) models can enhance our understanding of how PFAS binding to transport proteins may lead to prolonged exposure in vivo with long half-life estimates (Cheng and Ng, 2017; 2021; Fragki et al., 2023). The most rigorously studied example is in the kidney, where modeling and simulation revealed that reabsorption by transporters in proximal tubule cells contributed greatly to the persistence of PFAS in the circulation. In this study by Lin et al., changes to the parameters for a hypothetical hepatocyte basolateral efflux transporter did not affect PFOA clearance. However, removal of this suppositional clearance component reduced circulating concentrations ascribed to stores of PFOA retained in the liver. This model successfully capitulated the reasoning behind the differences in half-life estimates between a high, short-term dose and biomonitoring studies for PFOA in humans: high doses resulted in saturation of a transporter responsible for reabsorption of PFOA in proximal tubule cells, which reduced its serum half-life (Lin et al., 2023). This molecular mechanism is essential to understanding idiosyncratic toxicity risk and predictions for other PFAS whose clearance may be governed by hepatocyte transport. Incorporation of PFAS protein binding into PBTK models, along with kinetic parameters for bile acid and xenobiotic transporters, would give insight into the distribution of PFAS, especially with respect to species and dosing differences. From here, it would be possible to evaluate the relative concern for population variability or disease state on PFAS kinetics from transporter polymorphisms and transport protein abundance. These could aid in enhancing our understanding of the impact of genetic variability on worldwide susceptibility to PFAS exposures. As such, this review will assess the available literature and discuss the remaining knowledge gaps on PFAS interactions with hepatocellular bile acid and xenobiotic transporters.

Basolateral uptake transporters

Transporter-mediated uptake of PFAS

The ability of human (denoted by upper case letters) and rodent (denoted by lower case letters) transporters to move various PFAS into cells has only been evaluated in a handful of studies. Uptake of PFBS, PFHxS, PFOS, PFOA, and PFNA by the sodium/taurocholate co-transporting polypeptide (NTCP; SLC10A1) and organic anion transporting polypeptides (OATP; SLCO) 1B1, 1B3, 2B1, 1a1, and 1b2 has been demonstrated in transfected cell lines (Lin et al., 2023; Ruggiero et al., 2021; Zhao et al., 2015, 2017). To date, only NTCP has been shown to transport PFDA to a significant degree, though uptake by OATP1B1, 1B3, and 2B1 is also possible, even though the data did not reach statistical significance (Ruggiero et al., 2021; Zhao et al., 2017).

Modulation of transporters by PFAS

Only two studies have evaluated how PFAS functionally impact basolateral uptake transporters. PFBS, PFHxS, and PFOS inhibit taurocholate uptake via human NTCP in a chain-length dependent manner (ie, PFOS was the most potent inhibitor). Interestingly, rat NTCP function was most potently inhibited by PFHxS, illustrating an important species distinction for future in vivo studies (Zhao et al., 2015). Similarly, another study reported that the perfluoroalkyl carboxylic acids (PFCAs) exhibit chain-length dependent inhibition of NTCP transport with the most potent inhibition via PFDA (Ruggiero et al., 2021). Additionally, PFBS, PFHxS, PFOS, PFOA, PFNA, and PFDA tend to downregulate Slco10a1 mRNA expression in mice, with linkage to peroxisome proliferator-activated receptor-alpha (PPARα) activation (Bijland et al., 2011; Cheng and Klaassen, 2008; Marques et al., 2022; Zhang et al., 2018). With PFOS in particular, a high-fat diet potentiated this effect (Marques et al., 2020). Mice treated with the emerging compound, PFMO3NA, displayed reduced levels of Ntcp protein (Guo et al., 2022). Downregulation of SLCO10A1 mRNA expression has also been shown in PFOS-, PFOA-, PFNA- and PFDA-treated human cell lines, with more robust downregulation at lower concentrations and longer durations of PFOS exposure (Behr et al., 2020; Lim et al., 2022; Louisse et al., 2020). With regard to the mechanism(s) facilitating these differential gene expression patterns, some research indicates that the effect of PFAS-mediated PPARα regulation on transporter gene expression in mice may not be relevant in humans. Using human in vitro models, PFAS treatment did not induce PPARα to the extent reported in rodents in transactivation assays (Bjork et al., 2011). Indeed, chain-length dependent downregulation of SLCO10A1 in HepaRG cells by PFOA, PFNA, and PFDA was not linked to PPARα or constitutive androgen receptor (CAR) regulation. Instead, the potential linkage was to nuclear factor erythroid 2–related factor 2 (Nrf2), a mechanism that remains to be further investigated (Lim et al., 2022). On the other hand, PFBS, PFOS, PFDS, and PFOA have all shown robust perturbation of PPARα in human spheroids at environmentally relevant doses (Rowan-Carroll et al., 2021). Thus, data still support PPARα-induced regulatory effects in human in vitro models, though the effect of PFAS on SLCO10A1/Slco10a1 expression has not been attributed to a specific upstream regulator.

Similarly, most of the studies examining the effect of PFAS on human and rodent OATPs/Oatps have focused on the regulatory nature of these interactions. Generally, long-chain perfluoroalkyl acids tend to downregulate OATP/Oatp mRNA expression. In mice, this downregulation was occasionally linked to PPARα activation and the effects were more pronounced in mice fed high-fat diets, while in human cell lines the mechanism was potentially linked to Nrf2 activation (Behr et al., 2020; Cheng and Klaassen, 2008; Lim et al., 2022; Luo et al., 2017; Marques et al., 2020; Pfohl et al., 2021; Wang et al., 2023; Zhang et al., 2018). This effect is conserved in more sophisticated human in vitro models, where environmentally-relevant exposures of PFBS and HFPO-DA to HepaRG cells cultured on a liver-on-a-chip model reduced SLCO1B3 mRNA levels. However, this study did not report significant changes in SLCO1B1 expression (Solan et al., 2023). Mice fed a diet supplemented with fibers such as pectin and inulin alongside chronic PFOS exposure typically responded with less extensive or abrogated downregulation of Oatp mRNA expression (Deng et al., 2022). These trends in Oatp downregulation are also conserved in zebrafish models (Jantzen et al., 2016, 2017). In one study examining Oatp mRNA expression in relation to relative liver bioavailability of PFOA and HFPO-TA in mice, the authors reported a moderate positive correlation with Slco1a1, Slco1b2, and an organic anion transporter (Oat2/Slc22a7) (Cui et al., 2022). Mice treated with PFOA, HFPO-DA, PFMO2HpA, and PFMO3NA have also displayed reduced levels of Oatp2b1 protein (Guo et al., 2022). Similar to the trend reported for OATP/Oatps, PFNA, PFOS, HFPO-DA, and H-PFMO2OSA have been shown to downregulate Slc22a7 mRNA expression, although an increase in protein levels also was detected and the role of PPARα activation is debatable (Deng et al., 2022; Heintz et al., 2022; Pfohl et al., 2021; Wang et al., 2023; Zhang et al., 2018). Studies have also reported downregulation of organic cation transporter (OCT/Oct1; SLC22A1/Slc22a1) mRNA levels (Deng et al., 2022; Louisse et al., 2020).

Interestingly, the only uptake transporter for which consistent upregulation was reported by various types of PFAS in animal models is the organic cation transporter novel family member 2 (Octn2/Slc22a5) (Conley et al., 2019; Deng et al., 2022; Gray et al., 2021; Guruge et al., 2006; Heintz et al., 2022; Ho et al., 2006; Li et al., 2019; Liu et al., 2017; Rosen et al., 2017). This protein is key in transporting carnitine into hepatocytes for delivery into peroxisomes or mitochondria to aid in fatty acid oxidation (Juraszek and Nałęcz, 2019). Upregulation of Slc22a5 suggests that hepatocytes may require additional resources for fatty acid metabolism upon exposure to PFAS, indicating a response to a cellular deficiency.

Implications for health outcomes and knowledge gaps

PBTK models require Michaelis-Menten kinetic parameters such as binding affinity (K_m), potency of inhibition (K_i), and maximum velocity of transport (V_max) to assess the contribution of a specific transporter to the disposition of a substrate (Vivian and Polli, 2014). To date, kinetic parameters for hepatic transport of PFAS have only been generated for NTCP and OATPs in transfected cell lines, and K_m values range from 23 µM to as high as 5.3 mM; see Supplementary Table 4 (Lin et al., 2023; Ruggiero et al., 2021; Zhao et al., 2015; 2017). Of note, these studies used exposure concentrations ranging from 10 µM to 3 mM. For context, the geometric means of serum PFOA and PFOS concentrations in the total U.S. population in 2015–2016 were 3.77 and 9.44 nM, respectively (Fourth National Report on Human Exposure to Environmental Chemicals, 2021). This does not render these parameters unreliable; nevertheless, further studies are suggested to determine transporter-mediated PFAS clearance pathways from blood at lower exposure concentrations to support risk assessment efforts. However, use of lower exposures may limit the quantifiable detection of cellular PFAS accumulation, especially in uptake studies where the duration of exposure is on the order of seconds to minutes.

Moreover, the tendency for rodent and human cell lines to downregulate most uptake transporters that interact with PFAS either as substrates or inhibitors suggests that this may be protective for cells to limit intracellular accumulation of PFAS. Additionally, PFAS are known to associate with markers of oxidative stress, which can provide another mechanistic pathway by which the expression of these uptake transporters is reduced (Omoike et al., 2021; Roma and Sanchez Pozzi, 2008). Thus, this downregulation could simply be a marker of cellular stress or death due to direct toxicity. Of note, these responses have largely been studied in acute high-dose exposures and both regulatory effects and toxicokinetic parameters may differ widely between rodents and humans. Additionally, the referenced studies have not always tested overt toxicity of the concentrations used in cell lines or animal models prior to endpoint measurements. For example, 100 µM doses of PFOA induce oxidative stress in primary rat hepatocytes, which can explain some of the results described here and in further sections (Liu et al., 2017). Similarly, several animal studies referenced here have used extraordinarily high doses that have caused lethality in similar experimental designs of other studies (Eldasher et al., 2013; Luo et al., 2017; Maher et al., 2008). Thus, it is important from a human health perspective to determine the impact of PFAS on transporter function and/or expression at chronic, low-dose exposures that are not lethal. Additionally, investigators should assess whether the adaptive response reflected in mRNA expression described in the literature corresponds to functional changes at the protein level in human tissue or cell lines.

Basolateral efflux transporters

Transporter-mediated efflux of PFAS

So far, only Zhao et al (2015) have specifically evaluated whether basolateral hepatocyte efflux transporters are involved in PFAS disposition. Transport of PFBS, PFHxS, and PFOS was quantified in HEK293 cells overexpressing organic solute transporter-alpha/beta (OSTα/β; SLC51A/B) (Zhao et al., 2015). It is important to note that OSTα/β, which is not highly expressed at the protein level in healthy human liver, can function as either an uptake or efflux transporter (Malinen et al., 2018; Beaudoin et al., 2020). OSTα/β is included in this section because of documented upregulation of this transporter in diseases involving cholestasis. In this context, OSTα/β is hypothesized to serve as a “safety valve” to excrete bile acids across the basolateral membrane to help protect the hepatocyte from excess bile acid accumulation (Ballatori et al., 2009).

Modulation of transporters by PFAS

Only one study evaluated the impact of PFOS and PFOA on multidrug resistance-associated protein 4 (MRP4) function using membrane vesicles derived from MRP4-transfected HEK293 cells and found a dose-dependent inhibition of uptake (Dankers et al., 2013). The remaining data discussed in this section are related to changes in transporter gene expression. Studies in HepaRG cells have reported PFOA- and PFOS-mediated upregulation of SLC51A/B mRNA levels (Behr et al., 2020). In other transcriptomic studies using HepaRG cells or mice, PFBA, PFOA, PFOS, PFNA, PFDA, HFPO-DA, HFPO-TeA, and H-PFMO2OSA upregulated mRNA levels of MRP3/Mrp3 and MRP4/Mrp4 (ABCC3/Abcc3 and ABCC4/Abcc4, respectively), with some evidence of contribution via Nrf2 activation (Behr et al., 2020; Deng et al., 2022; Heintz et al., 2022; Lim et al., 2022; Louisse et al., 2020; Luo et al., 2017; Maher et al., 2008; Wang et al., 2023, 2017; Zhang et al., 2018). A few studies in HepaRG cells and primary mouse and human hepatocytes found that PFBS, PFHxS, PFOS, and PFCAs of various lengths downregulated ABCC3/Abcc3 and Abcc4 (Behr et al., 2020; Rosen et al., 2008, 2013). Similarly, there was a moderate negative correlation between Abcc4 mRNA expression and bioavailability of PFOA- and HFPO-TA-spiked chow relative to control chow fed to mice (Cui et al., 2022). These studies did not report overt toxicity in response to the selected concentrations for dosing cells and animals; however, it is possible that the decrease in MRP mRNA could be a response to hepatocellular stress and/or death.

Implications for health outcomes and knowledge gaps

The ability of long-chain legacy PFAS to upregulate ABCC3 and ABCC4 and inhibit MRP4 could demonstrate protection against, or facilitation of, cholestatic potential, respectively (Chai et al., 2012; Gradhand et al., 2008). Enhanced basolateral efflux combined with downregulation of hepatocyte uptake transporters may serve as protective mechanisms to reduce hepatocyte exposure to PFAS. These mechanistic pathways would also be corroborated by further studies that assess PFAS transport and quantify changes in protein levels of these transporters after PFAS exposure. Additionally, because OSTα/β can function as either an uptake or an efflux transporter, it would be helpful to determine whether OSTα/β can efflux PFAS.

Apical efflux transporters

Transporter-mediated efflux of PFAS

From a functional standpoint, the number of studies reporting PFAS inhibition of apical efflux transporters far surpasses the number of studies evaluating PFAS as substrates for these transporters. Dankers et al. (2013) reported that PFOA can be transported by the breast cancer resistance protein (BCRP; ABCG2), but not by multidrug resistance protein 1, also known as P-glycoprotein (MDR1 P-gp; ABCB1).

Modulation of transporters by PFAS

In vesicular studies, PFBS, PFHxS, and PFOS inhibited transport of MRP2 (ABCC2), BCRP, and bile salt export pump (BSEP; ABCB11) (Dankers et al., 2013; Eldasher et al., 2013; Zhao et al., 2015, 2017) using standard probe substrates. Similarly, PFOS and PFOA inhibited MDR1 P-gp transport in vesicle studies (Dankers et al., 2013). In contrast, one study reported increased MRP2 transport measured as increased fluorescence intensity of the MRP2 substrate 5-carboxy-2′,7′-dichlorofluorescein in response to PFOS and PFOA in HepaRG cell cultures. However, this was more likely due to the dilation of bile canaliculi from cytoskeletal disorganization, and may not reflect enhanced MRP2 functional activity (Behr et al., 2020).

The regulatory response of the apical efflux transporters to PFAS exposure seems to be transporter-dependent rather than the clear trends evident for the uptake and basolateral efflux transporters. For example, ABCC2/Abcc2 generally has been reported to be upregulated by long-chain PFAS (Behr et al., 2020; Bjork et al., 2008; Deng et al., 2022; Ho et al., 2006; Luo et al., 2017; Ren et al., 2010; Rosen et al., 2007; Weatherly et al., 2021; Yu et al., 2011; Zhang et al., 2018), though two studies reported downregulation of Abcc2 mRNA and protein levels (Johnson and Klaassen, 2002; Yoo et al., 2023). Similarly, PFBS, PFHxS, PFOA, PFOS, and PFNA downregulated ABCB11/Abcb11 mRNA (Beggs et al., 2016; Behr et al., 2020; Bijland et al., 2011; Bjork et al., 2011; Louisse et al., 2020; Pouwer et al., 2019; Zhang et al., 2018), though a lower-dose study in primary human hepatocytes and analyses of fetal rat livers reported upregulated levels (Bjork et al., 2008, 2011). Meanwhile, long-chain PFCAs, PFBS, PFHxS, PFOS, and HFPO-DA consistently upregulate mRNA levels of ABCG2/Abcg2 with some linkage to PPARα or Nrf2 activation (Deng et al., 2022; Eldasher et al., 2013; Heintz et al., 2022; Lim et al., 2022; Louisse et al., 2020; Solan et al., 2023; Zhang et al., 2018). Few studies have examined changes in mRNA expression in ABCB1, the multidrug resistance protein 3 (MDR3/Mdr3; ABCB4/Abcb4), and the multidrug and toxin extrusion protein (MATE1/Mate1; SLC47A1/Slc47a1); thus, it is difficult to generalize the effects of PFAS on these proteins (Dankers et al., 2013; Deng et al., 2022; Heintz et al., 2022; Louisse et al., 2020; Rosen et al., 2013; Rusiecka and Składanowski, 2008; Wang et al., 2023; Weatherly et al., 2021; Zhang et al., 2018). Some of these transcriptomic changes were supported by changes in protein levels, such as those reported for Mrp2, Bcrp, and MDR1/Mdr1b P-gp (Eldasher et al., 2013; Johnson and Klaassen, 2002; Rusiecka and Składanowski, 2008; Wang et al., 2023). Despite the trends, it is unclear why there are differing responses of the various apical efflux transporters to PFAS. With human cell lines, standard culture conditions include higher-than-physiologic levels of serum (ie, 10%) in culture media, which bind and reduce the amount of unbound PFAS that can enter cells (Bangma et al., 2020). It is possible that high exposure levels of various PFAS cause saturation of serum protein binding, increasing the amount of free PFAS available to reach and perturb apical transporters. To date, there are no studies determining whether PFAS are substrates of other apical efflux transporters, how PFAS are moved from hepatocytes to bile in vivo, or whether these mechanisms contribute to hepatocyte accumulation and/or toxicity.

Implications for health outcomes and knowledge gaps

Impaired function or downregulation of expression of these apical efflux transporters could contribute to altered bile acid homeostasis and/or cholestatic mechanisms that would, in part, explain the associations of PFAS with increased serum cholesterol. Hepatic cytochrome P450 (CYP) 7A1 is the first and rate-limiting step in the conversion of cholesterol to bile acids and serves to promote cholesterol metabolism. Increased uptake of bile acids signals a negative feedback mechanism that inhibits CYP7A1 via the nuclear receptor, farnesoid X receptor (FXR) (Chiang and Ferrell, 2018). Bile acid accumulation in hepatocytes also may inhibit CYP7A1 as an additional strategy to limit further production of bile acids from cholesterol (Chiang, 2017). Bile acids are transported from hepatocytes to bile predominately by BSEP, although MRP2 plays a role in the biliary excretion of glucuronide and sulfate conjugates of bile acids; MDR1 P-gp also has been implicated in this process (Dawson et al., 2009). Disruption of bile flow, or cholestasis, leads to the accumulation of bile acids in the liver and serum, thereby disrupting cholesterol metabolism and promoting hypercholesterolemia (Li and Apte, 2015).

Animal models have consistently shown that PFAS can increase serum and liver bile acids and produce cholestasis upon histological examination (Cheng and Klaassen, 2008; Maher et al., 2008; Wang et al., 2023; Zhang et al., 2018). Studies with human serum also have reported positive correlations with PFAS levels and increased bile acids (Salihović et al., 2020; Sen et al., 2022). The relationship between PFAS exposure and bile acid disposition is summarized in Table 2. In both human and animal models, many PFAS inhibit CYP7A1 expression and function, although this may be a function of higher exposure doses (Behr et al., 2020; Yoo et al., 2023). Human epidemiological studies also have shown a linkage between PFAS levels and markers of cholestasis such as elevations in gamma-glutamyltransferase and bilirubin. However, data are inconsistent and subject to high variability based on location and exposure level (Toxicological Profile for Perfluoroalkyls, 2018). The inhibition of BSEP, MRP2, and BCRP function by PFAS provides mechanistic evidence for the theory that PFAS have the potential to induce hypercholesterolemia via cholestasis (Zhao et al., 2015). The environmental relevance of the exposure level in vitro and the use of membrane vesicles prepared from cells transfected with the transporter(s) of interest limits the extrapolation of these data to in vivo. Nevertheless, the potential for PFAS to modulate the apical efflux transporters and contribute to cholestasis could provide a reason for the consistent epidemiological reports of the PFAS-hypercholesterolemia correlation.

Table 2.

PFAS impact on disposition of total bile acids or targeted bile acid species.

Total or targeted	PFAS	Dose	Duration	Model	Method	Results	Reference
Targeted	PFOA PFOS	1–250 µM 1–50 µM	24 and 48 h	Serum-starved HepaRG cells	Ultra-performance liquid chromatography coupled to tandem mass spectrometry of cell lysates and supernatant	Biphasic response: dose-dependent decrease in concentrations of bile acids, particularly GCDCA and TCDCA, in both cell lysates and supernatant in cells treated with mid-range concentrations of PFOA/PFOS (50–100/10–25 µM); increased bile acid concentrations at higher PFOA/PFOS concentrations (250/50 µM).	Behr et al., 2020
Total	PFOA PFDA	40 mg/kg of PFOA, 80 mg/kg PFDA IP, once	48 h	WT, FXR-null, PXR-null, CAR-null, Nrf2-null, and PPARα-null male C57BL/6 mice	Commercial colorimetric assay kit using serum	PFDA but not PFOA significantly increased serum bile acid concentrations.	Cheng and Klaassen, 2008
Targeted	PFNA	0.1 mmol/kg in a volume of 5 ml/kg IP, once (46.4 mg/kg)	1 week, 2 weeks, and 3 months	WT, PPARα-null, CAR-null male C57BL/6 mice	Ultra-performance liquid chromatography coupled to tandem mass spectrometry of serum and liver; histological examination of liver	Microvesicular steatosis in PPARα-null mice. Evidence of cholestasis in WT mice at 2 weeks and 3 months after exposure. Increased serum bile acids at the highest level in WT mice, followed by CAR-null mice, but not in PPARα-null mice. Reduced liver unconjugated, glycine-conjugated, taurine-conjugated, and total bile acid concentrations in WT-PFNA mice. Only unconjugated BAs decreased in the livers of PPARα- and CAR-null-PFNA mice.	Zhang et al., 2018
Total	PFBS PFHxS PFOS	30 mg/kg/day, 6 mg/kg/day, 3 mg/kg/day respectively, oral	4–6 weeks	APOE*3-Leiden CETP male mice	Gas chromatography of feces	PFHxS and PFOS decreased fecal bile acid excretion.	Bijland et al., 2011
Total	PFOA	10, 300, and 30 000 ng/g/day, oral	4-6 weeks	APOE*3-Leiden CETP male mice	Gas chromatography of feces	No change in fecal bile acid excretion.	Pouwer et al., 2019
Targeted	PFDA	80 mg/kg, IP, once	5 days	WT and PPARα-null 129/Sv mice	Ultra-performance liquid chromatography coupled to tandem mass spectrometry of serum	Serum bile acids (CA, TUDCA, MCA, and TCA) were significantly increased only in WT mice.	Luo et al., 2017
Total	PFOS H-PFMO2OSA	1 and 5 mg/kg/day, oral	28 days	WT and PPARα-null B6 mice	Commercial bile acid assay kit using serum and liver	Only H-PFMO2OSA increased total bile acids in serum at 5 mg/kg/day. Only PFOS increased total bile acids in liver at both concentrations.	Wang et al., 2023
Total	PFOA PFDA	40–80 mg/kg IP, once	48 h	WT, Nrf2-null, and PPARα-null male C57BL/6 mice	Commercial bile acid assay kit using serum	Increased total bile acid levels in serum.	Maher et al., 2008
Targeted	Various PFCAs and PFSAs from C4–C14	N/A	N/A	Serum collected from blood samples of human donors	Ultra-performance liquid chromatography coupled to tandem mass spectrometry of serum	A majority of individual bile acids as well as total bile acids were negatively correlated with serum PFAS levels. Significant positive correlation between LCA, GLCA, and TLCA levels and PFOA, PFNA, PFOS, PFDA, PFUnDA, and PFTrDA.	Salihović et al., 2020
Targeted	PFOS PFOA PFNA	N/A	N/A	Serum collected from blood samples of human donors	Ultra-performance liquid chromatography coupled to tandem mass spectrometry of serum and liver	Positive correlation between serum levels of all 3 PFAS with TCA, GCDCA, and TCDCA in female donors only. Positive correlation between PFOA and liver levels of DCA, GHCA, and GUDCA in female donors only.	Sen et al., 2022
Total and targeted	PFOA HFPO-DA PFMO2HpA PFMO3NA	0.4, 2, and 10 mg/kg/day, oral	28 days	BALB/c male mice	Commercial bile acid assay kit using serum and liquid chromatography coupled to tandem mass spectrometry using liver	Total bile acids in serum increased in response to all doses except 0.4 mg/kg/day of PFOA, PFMO2HpA, HFPO-DA, and 2 mg/kg/day PFMO2HpA. Primary conjugated bile acids in the liver were primarily increased.	Guo et al., 2022

Effects of PFAS on bile acid disposition. CA, cholic acid; CAR, constitutive androgen receptor; DCA, deoxycholic acid; FXR, farnesoid X receptor; GCDCA, glycochenodeoxycholic acid; GHCA, glycohyocholic acid; GLCA, glycolithocholic acid; GUDCA, glycoursodeoxycholic acid; IP, intraperitoneal; LCA, lithocholic acid; MCA, muricholic acid; Nrf2, nuclear factor erythroid 2-related factor 2; PPARα, peroxisome proliferator-activated receptor-alpha; PXR, pregnane X receptor; TCA, taurocholic acid; TCDCA, taurochenodeoxycholic acid; TLCA, taurolithocholic acid; TUDCA, tauroursodeoxycholic acid; WT, wild-type. Refer to Table 1 for full-length chemical names and classifications for abbreviated PFAS.

Additional considerations regarding the contribution of PFAS-transporter interactions to PFAS disposition

The enterohepatic recycling theory of PFAS persistence

One study used matched human serum and bile samples to calculate the biliary excretion and reabsorption rates of PFOA and PFOS. This study demonstrated that PFOA and PFOS were excreted into bile and readily reabsorbed into the circulation, which provided a basis for the enterohepatic recycling theory to explain PFAS persistence in humans (Harada et al., 2007). The ability of NTCP, OATPs, and the intestinal apical sodium-dependent bile acid transporter (ASBT) to transport PFAS has also provided some evidence for the commonly stated hypothesis that PFAS mimic bile acid reabsorption. When PFAS become “trapped” in enterohepatic recirculation, their tendency to accumulate in the liver may be enhanced (Lin et al., 2023; Ruggiero et al., 2021; Zhao et al., 2015, 2017). These findings are frequently cited to bolster the hypothesis that PFAS persistence is partly due to enterohepatic recirculation, although a more accurate assessment of this potential is only possible by considering the capability of apical efflux transporters to excrete PFAS into bile. With this theory in consideration, a computational study was conducted using the binding affinities of PFAS to NTCP and ASBT, calculated from previous studies in transfected cell lines. The authors used these values to determine the fraction of reabsorption (f_reabs) relative to the bile acid, taurocholic acid (TCA), which is highly (95%) reabsorbed from the intestine due to enterohepatic circulation. The authors reported that long-chain PFCAs tend to have high f_reabs compared to other PFAS. Similarly, perfluoroalkane sulfonic acid reabsorption increases significantly by chain length, even by addition of just one carbon. Some PFAS structurally resemble TCA, which may explain their affinity for NTCP. The authors concluded that the affinity of PFAS to ASBT is a strong predictor of their long serum half-life estimates, while affinity to NTCP does not correlate with serum half-life, as determined by molecular docking analyses (Cao et al., 2022a). Although this would provide a clear explanation for why PFAS persist in the blood, it is unknown whether all PFAS undergo enterohepatic recirculation in vivo in humans, and which transporters contribute to this mechanism. As discussed previously, Dankers et al. (2013) reported that PFOA can be transported by BCRP, but this is the only study available to support transporter-mediated biliary excretion of PFAS. Bile cannulation studies in rodents have yet to be performed, though these results would have to be cautiously interpreted due to rodent-specific differences in transporter expression and affinity to PFAS.

Additional transport mechanisms

Multiple mechanisms may be responsible for PFAS tissue distribution and the long half-life estimates reported for these compounds. For legacy PFAS, extensive bioaccumulation initially was attributed to integration within the phospholipids of cell membrane bilayers. More recent work has focused on protein binding interactions, which explain more of the characteristics of bioaccumulation than the phospholipid theory alone (Ng and Hungerbühler, 2014). In particular, PFAS are highly bound to liver fatty acid binding protein (L-FABP) and human serum albumin (HSA), which may explain, in part, the tendency for these compounds to aggregate in the liver (Zhao et al., 2023). It is increasingly apparent that the binding of PFAS to L-FABP and/or HSA cannot fully explain the disposition of PFAS, which necessitates considering other protein interactions (ie, transporters) (Cheng and Ng, 2017). Although L-FABP is used as a screening tool to assess potential hepatotoxicity of PFAS based on binding affinity, this approach often relies on in silico methods that do not include physiologic processes at the hepatocellular level (Yang et al., 2020; Zhao et al., 2023). Additionally, studies in zebrafish led to the hypothesis that PFAS-bound L-FABP is delivered to nuclear receptors, such as PPARα, to exert effects potentially related to perturbations in cholesterol metabolism (Zhang et al., 2012). However, some researchers argue that environmentally-relevant levels of PFAS are not comparable to the doses necessary to induce transcription of PPARα-regulated genes in human cell lines and animal models (Andersen et al., 2021), and this is further discussed in the “Human relevance of regulatory modes of action and experimental models” section. Thus, the corresponding cumulative effects that transport proteins play in addition to L-FABP and HSA interactions with PFAS remain to be evaluated in humans.

In addition to bile acid and xenobiotic transporters, hepatocytes also express fatty acid transport proteins (FATP) 2, 3, 4, and 5 as well as fatty acid translocase (CD36). FATP5 is a liver-specific member of this family expressed on the basolateral membrane of hepatocytes (Alves-Bezerra and Cohen, 2017; Gimeno, 2007; Kazantzis and Stahl, 2012). These transport proteins increase fatty acid uptake when overexpressed and have been implicated in the pathogenesis of diseases such as MASLD (Kazantzis and Stahl, 2012; Rada et al., 2020). PFAS can modulate FATP expression and may rely on CD36 expression to enter retinal cells (Bijland et al., 2011; Camdzic et al., 2022; Conley et al., 2021; Roth et al., 2021). Thus, it is possible that long-chain PFCAs, due to their structural resemblance to fatty acids, may use these transporters as a means to enter hepatocytes and/or disrupt bile acid or lipid metabolism. The nature of PFAS binding to these transporters has yet to be elucidated. The potential interaction between PFAS and these fatty acid transporters could play a role in PFAS disposition and perturbation in cholesterol metabolism, and warrants further investigation.

Additionally, Lin et al. (2023) observed that mock-transfected HEK293 cells have basal levels of PFOA uptake compared with liver and renal transporter-transfected HEK293 cells. This was attributed to the presence of the monocarboxylate transporter 1 (MCT1), a ubiquitous transporter expressed in virtually all organs in humans. The role of MCT1 in normal liver physiology is not well described, but part of MCT1 function is to transport lactate into hepatocytes for gluconeogenesis, which may be altered during pathophysiological states (Droździk et al., 2020). MCT1 may provide a route of entry for PFAS, specifically for PFCAs, which contain one carboxylate group and as such may resemble prototypical substrates for MCT1. An in vitro study to characterize the affinity of various PFCAs to MCT1 could provide additional information regarding PFAS accumulation and persistence in the liver and other tissues.

Human relevance of regulatory modes of action and experimental models

To understand the variability between the studies discussed in this review and how data can be extrapolated to humans, it is necessary to consider the human relevance of proposed modes of action as well as the choice of experimental models. PFAS are known PPARα agonists, and as discussed in previous sections, this could explain their impact on hepatocellular transporter expression. However, the organ-specific expression of PPARα (eg, higher hepatic expression of PPARα in rodents than humans) and the relationship of PPARα activation to downstream liver-specific health effects differ between humans and rodents (Abbott, 2009). Species differences in liver pathology and transcriptional pathways have been noted after rodent or human-model exposure to PPARα agonists, including PFAS (Foreman et al., 2021; McMullen et al., 2020; Su et al., 2022). Thus, it is possible that rodents are more sensitive to PPARα activation. However, humanized and human models still show that PPARα activation is necessary for some of the outcomes related to adverse liver outcomes, although perhaps not for changes in transporter expression (Evans et al., 2022; Schlezinger et al., 2020; Wang et al., 2023).

The choice of experimental models also impacts the interpretation of the data discussed in this review, especially in the context of transporter function and expression. Some important differences in hepatic transporters exist between humans and rodents, and rodents also may have sex-specific differences in transporter expression that impact PFAS toxicokinetics. Transfected cell lines (eg, HEK293, Chinese hamster ovary cells) expressing high levels of the transport protein(s) of interest are invaluable to determine whether PFAS are substrates and/or inhibitors of specific transporters. However, these models have some disadvantages: transfected transporter expression levels may be much higher than observed in vivo; they do not account for potential compensation by other transporters in hepatocytes; and non-liver cell lines may lack hepatocellular regulatory machinery that can impact transporter expression, trafficking, and/or function (Balakrishnan et al., 2007; Brouwer et al., 2022; Zamek-Gliszczynski et al., 2013). Studies in human hepatocytes are ideal, although it is more challenging to attribute outcomes to a specific transporter, and donors may have variability in transporter abundance. Additionally, human hepatocyte models de-differentiate in culture without overlay, and even sandwich-cultured human hepatocytes may only be suitable for short-term chemical exposures (eg, up to approximately 5 days in culture) (Kaur et al., 2023). To address this limitation, 2D micropatterned co-culture models such as HepatoPac more closely resemble the liver microarchitecture, which improves hepatocellular viability and function for a longer culture period (Wang et al., 2021). Human spheroid models using primary hepatocytes co-cultured with non-parenchymal cells may be another approach, but these models have yet to be characterized for transporter localization and function over long-term cultures (Bell et al., 2018). Thus, model selection to evaluate PFAS effects on transporter-specific outcomes should consider human relevance; maximum allowable time in culture, which influences PFAS exposure duration; and baseline transporter abundance, regulation, and function.

Conclusion and future directions

In summary, the potential importance of transport protein binding in the disposition of PFAS is a mechanism that has only recently been recognized. Interaction with liver bile acid and xenobiotic transport proteins may explain unanswered questions regarding the tendency for PFAS to accumulate in the liver, persist in serum, and impact cholesterol levels. At this point, transcriptomic data and signal transduction mechanisms in preclinical species and human cell lines outweigh the available data assessing hepatic transporter function. It would be impactful to survey a broader range of PFAS to determine which are substrates for liver transporters, and which physicochemical characteristics of PFAS can predict transporter affinity. Information is sparse with respect to changes in transporters at the protein level, and further studies are needed to estimate kinetic parameters to effectively generate accurate PFAS PBTK models. Additionally, most studies have only focused on single short-term exposures to long-chain legacy PFAS, with a limited number of studies using longer-term exposures (eg, weeks to months) or emerging PFAS with extensive data gaps. Nevertheless, there are clear trends in the transcriptomic changes in basolateral transporter expression levels induced by PFAS across various strains of rodent models and different human cell lines. These data reveal new questions about how PFAS exposure may affect the disposition of other drugs and xenobiotics. The impact of diet and lifestyle could also exacerbate or ameliorate some of the observed effects, as noted in animal models. Further studies are needed to understand the contribution of enterohepatic recirculation to accumulation and persistence as well as subsequent health outcomes. Thus, a large knowledge gap remains in the interaction between these liver transporters and PFAS exposure.

Supplementary data

Supplementary data are available at Toxicological Sciences online.

Acknowledgments

The authors would like to thank Nicole Chang for her assistance in reviewing tables for accuracy.

Funding

E.V. was supported by the National Research Service Award T32 ES007126 from the National Institute of Environmental Health Sciences of the National Institutes of Health. K.L.R.B. was supported, in part, by the National Institute of General Medical Sciences of the National Institutes of Health (NIH) under Award Number R35 GM122576.

Declaration of conflicting interests

Dr Kim L. R. Brouwer is a coinventor of the sandwich-cultured hepatocyte technology for quantification of biliary excretion (B-CLEAR^®) and related technologies, which have been licensed exclusively to BioIVT. All other authors declare no conflicts of interests.