Binding of Per- and Polyfluoroalkyl Substances (PFAS) to Serum Proteins: Implications for Toxicokinetics in Humans

Abstract

Per- and polyfluoroalkyl substances (PFAS) are a diverse class of highly persistent anthropogenic chemicals that are detectable in the serum of most humans. PFAS exposure has been associated with many adverse effects on human health including immunotoxicity, increased risk of certain cancers, and metabolic disruption. PFAS binding to the most abundant blood serum proteins (human serum albumin [HSA] and globulins) is thought to affect transport to active sites, toxicity, and elimination half-lives. However, few studies have investigated the competitive binding of PFAS to these proteins in human serum. Here, we use C18 solid-phase microextraction fibers to measure HSA-water and globulin-water distribution coefficients (D_HSA/w, D_glob/w) for PFAS with carbon chains containing 4 to 13 perfluorinated carbons (η_pfc = 4-13) and several functional head-groups. PFAS with η_pfc < 7 were highly bound to HSA relative to globulins, whereas PFAS with η_pfc ≥ 7 showed a greater propensity for binding to globulins. Experimentally measured D_HSA/w and D_glob/w and concentrations of serum proteins successfully predicted the variability in PFAS binding in human serum. We estimated that the unbound fraction of serum PFAS varied by up to a factor of 2.5 among individuals participating in the 2017-2018 U.S. National Health and Nutrition Examination Survey. These results suggest that serum HSA and globulins are important covariates for epidemiological studies aimed at understanding the effects of PFAS exposure.

Keywords: exposure; globulins; human serum albumin; interindividual variability; solid-phase microextraction; unbound fraction.

Figure 1.

Results of solid-phase microextraction (SPME) C18 fiber-water distribution experiments for PFAS. Panel (A) shows concentration ratios in phosphate buffered saline (PBS) and fibers for five PFAS measured over 48 hours in batch experiments (n=3 per time point). The solid lines represent a first-order model fit. Panel (B) shows sorption isotherms for five PFAS and corresponding slopes of the linear isotherm fits (n=4 per concentration), where m denotes the fitted slope of linear isotherms.

Figure 2.

Variability in PFAS distribution coefficients to C18 fibers (${\mathit{D}}_{\mathbf{C}18/\mathbf{w}}$), human serum albumin (${\mathit{D}}_{\mathbf{H}\mathbf{S}\mathbf{A}/\mathbf{w}}$), γ-globulins (${\mathit{D}}_{\mathbf{g}\mathbf{l}\mathbf{o}\mathbf{b}/\mathbf{w}}$), and human serum (${\mathit{D}}_{\mathbf{s}\mathbf{e}\mathbf{r}\mathbf{u}\mathbf{m}/\mathbf{w}}$) as a function of their molecular weight (${\mathbf{M}}_{\mathbf{w}}$). Panels show distribution coefficients of PFAS to (A) the C18 fiber; (B) human serum albumin (HSA); C) γ-globulins, and D) human serum. PFAS classes are indicated by the color and shape of symbols with orange circles for carboxylates, blue triangles for sulfonates, yellow squares for sulfonamides, and grey hexagons for fluorotelomers. Turquoise diamond = ADONA.

Figure 3.

Panel (A) shows average fractions unbound of the study PFAS in the surrogate human serum (see SI section S1) measured with C18 fibers. Panel (B) shows the comparison of experimentally measured distribution coefficients of PFAS to serum (log ${\mathit{D}}_{\mathbf{s}\mathbf{e}\mathbf{r}\mathbf{u}\mathbf{m}/\mathbf{w}}$) to distribution coefficients predicted by serum protein concentrations and distribution coefficients for globulins and human serum albumin (HSA). Serum protein concentrations were based on the average concentrations measured in the NHANES 2017–18 survey cycle (n=1919). Turquoise diamond = ADONA.

Figure 4.

Variability in the unbound fraction of PFAS in serum ($f_{\mathrm{u}\mathrm{n}\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}$). Center plot shows the range in unbound fractions ($f_{\mathrm{u}\mathrm{n}\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}$) of PFAS for individuals participating in the NHANES 2017–18 survey cycle (n=1919) predicted from measured albumin and globulin concentrations in blood and corresponding PFAS distribution coefficients. Left and right panels show the estimated unbound/bioactive fractions ($f_{\mathrm{u}\mathrm{n}\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}$) of perfluoroheptanoate (PFHpA) and perfluorotridecanoate (PFTriDA) in serum for the same individuals. Black horizontal lines indicate the median and 75th and 25th percentiles of $f_{\mathrm{u}\mathrm{n}\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}$. Coefficients of variation (CV) and maximum/minimum ratios of $f_{\mathrm{u}\mathrm{n}\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}$ are reported. Turquoise diamond = ADONA.