CoChat

In-Vitro and In-Silico Assessment of Per- and Polyfluoroalkyl Substances (PFAS) in Aqueous Film-Forming Foam (AFFF) Binding to Human Serum Albumin

Open access

The DOI matches its Crossref record; 67 of the 68 references it cites were found in a registry or an index; 14 of the 56 sources its claims cite were checked for support.

Abstract

Drinking water contaminated by fluorosurfactant-based aqueous film-forming foams (AFFF) is a source of human exposure to poly- and perfluoroalkyl substances (PFAS). However, assessment of bioaccumulation potentials of diverse PFAS in commercial products such as AFFF have been insufficient and challenging, especially due to a lack of analytical standards. Here we explore the value of suspect screening, equilibrium dialysis, and molecular-docking simulations to identify potentially bioaccumulative PFAS. We exposed human serum albumin (HSA) protein to dilutions of a legacy AFFF produced by 3M in 1999 using equilibrium dialysis and screened in-vitro protein-binding affinities using high-resolution mass spectrometry (HRMS). Through suspect screening, we identified 32 PFAS and 18 hydrocarbon surfactants in the AFFF that bound to HSA. Quantification of noncovalent association constants for 26 PFAS standards confirmed that many PFAS, including the short-chain perfluoropropane sulfonic acid (log Ka= 4.1 ± 0.2 M−1), exhibit strong binding affinities with HSA. At least five PFAS in AFFF (including three PFAS with less than five perfluorocarbons) remained bound to the precipitated HSA pellet after extensive solvent washing—an indication of high PFAS binding potential. Three PFAS (PFBS, PFOS, and PFOA) were confirmed in the protein pellet with analytical standards and quantified after acid digestion—this sample fraction accounted for 5 to 20% of each compound mass in the sample. We calculated pseudo-bioconcentration factors (BCFpseudo) for PFAS that suspect screening flagged as noncovalently bound or potentially covalently bound. Most PFAS exhibiting high BCFpseudo, especially those with seven perfluorocarbons, contained a carboxylic acid or a sulfonic acid. Finally, we used molecular docking to simulate HSA binding affinities for 62 ligands (26 PFAS targets, 18 PFAS qualified in AFFF, and 18 hydrocarbon surfactants qualified in AFFF). We found that molecular docking can effectively separate HSA-binding and -nonbinding compounds in AFFF. In-vitro and in-silico approaches described in this study provide replicable, high-throughput workflows for assessing bioaccumulation potentials of diverse PFAS in commercial products.

Overview

CoChat Summary

TL;DR

This study combined equilibrium dialysis, high-resolution mass spectrometry suspect screening, and molecular docking to assess binding of PFAS from a legacy 3M AFFF to human serum albumin. It found that many AFFF PFAS bind HSA, some remain in precipitated protein pellets after extensive washing, and docking can qualitatively separate HSA binders from nonbinders but not quantitatively predict long-chain PFAS affinities.

Key findings

  • Of 32 PFAS identified in the AFFF, 28 bound noncovalently to HSA; five PFAS were qualified and confirmed in precipitated and washed protein pellets, indicating candidates for ultrastrong noncovalent or potentially covalent binding.
  • For 26 PFAS standards, HSA association constants ranged from 10^4.0 to 10^5.5 M−1; PFHxS and PFHpA had high log KA values of 4.99 ± 0.44 and 5.53 ± 0.39, respectively, and PFBS had log Ka = 4.1 ± 0.2 M−1.
  • PFBS, PFOA, and PFOS were quantified in protein pellets at 7%, 20%, and 5% of total spiked mass (80 ng), respectively; AFFF-spiked pellets contained 1% PFBS, 26% PFOA, and 2% PFOS.
  • Docking showed a positive correlation for short-chain PFAS with experimental KA (slope = 1.02 ± 0.19, r = 0.900) but a negative correlation for long-chain PFAS (slope = −1.05 ± 0.30, r = 0.7680); docking qualitatively separated HSA-binding and nonbinding compounds, though one unbound PFAS and three bound hydrocarbon surfactants were not significantly different (p = 1).

Methods

  • Equilibrium dialysis used a 96-well system with a 10-kDa membrane; HSA at 600 μM was exposed to AFFF dilutions (4000- to 16,000-fold) or an in-house mixture of 26 PFAS at 37 °C, 30 RPM, for 108 h, with six concentration levels and four replicates, plus method blanks and negative controls.
  • Extracts from aqueous, noncovalently bound protein, and residual protein pellet fractions were analyzed by HPLC-QTOF-MS using data-dependent and data-independent acquisition; suspect screening used an in-house PCDL containing 3793 PFAS and 727 hydrocarbon surfactants, and quantification used a 19-ISTD ten-point calibration curve (0.5–250 ng/mL) with whole-method LOQs of 0.025–1 ng/mL.
  • Molecular docking with AutoDock Vina v1.1.2 docked 62 ligands to two HSA crystal structures (1E7G and 1AO6), with 100 repeats for six binding pockets and nine binding modes each, yielding 5400 predictions per PFAS; redocking PFOS gave RMSD < 2 Å.

Limitations

  • The paper states that docking scores cannot be used directly as protein–ligand binding energies, especially for large compounds; AutoDock Vina accuracy is only 60–80%, with modeling assumptions including protein rigidity and removal of water, and docking accuracy declines with increasing rotatable bonds.
  • The paper states that mechanisms explaining PFAS in the protein pellet cannot be disentangled; residual PFAS are considered candidates for ultrastrong noncovalent or potentially covalent binding, but covalent bond formation between carboxylic or sulfonic acids and protein residues has not been observed under physiological conditions, and 14 additional PFAS qualified in dialysis extracts were not further evaluated.
  • Evident from the text: suspect screening depends on an in-house library and confidence-level criteria, some detected PFAS could not be quantified due to lack of available standards, and the study uses one legacy 3M AFFF product, limiting generalizability to other AFFF formulations.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3390/toxics9030063

Cites

67 of 67 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Equilibrium dialysis set-up and observed mass balance for three PFAS with initial dosages of 40–80 ng. PFAS that were free in aqueous solution PFAS, noncovalently bound PFAS, and residual PFAS in the precipitated protein pellet were measured independently. The time required to reach equilibrium (Teq) was previously determined using 26 PFAS standards in Figure S1.2. Detailed mass balance data are available in Table S1.1.
Figure 1 Equilibrium dialysis set-up and observed mass balance for three PFAS with initial dosages of 40–80 ng. PFAS that were free in aqueous solution PFAS, noncovalently bound PFAS, and residual PFAS in the precipitated protein pellet were measured independently. The time required to reach equilibrium (Teq) was previously determined using 26 PFAS standards in Figure S1.2. Detailed mass balance data are available in Table S1.1.CC BY · PubMed Central
Figure 2 Experimentally determined noncovalent association constants (KA) of 26 PFAS targets with human serum albumin (has) in equilibrium dialysis. Three C8 precursor compounds, Perfluoro-1-octanesulfonamide (FOSA), N-methylperfluoro-1-octanesulfonamidoacetic acid (MeFOSAA), and N-ethylperfluoro-1-octanesulfonamidoacetic acid (EtFOSAA) are presented in a separate plot. The error bars represent one standard deviation.
Figure 2 Experimentally determined noncovalent association constants (KA) of 26 PFAS targets with human serum albumin (has) in equilibrium dialysis. Three C8 precursor compounds, Perfluoro-1-octanesulfonamide (FOSA), N-methylperfluoro-1-octanesulfonamidoacetic acid (MeFOSAA), and N-ethylperfluoro-1-octanesulfonamidoacetic acid (EtFOSAA) are presented in a separate plot. The error bars represent one standard deviation.CC BY · PubMed Central
Figure 3 Comparison of experimental log KA with results from molecular simulations with the HSA crystal structure 1E7G. Solid black lines represent the 1:1 line; dotted lines represent one log unit higher or lower. Error bars reflect one geometric standard deviation (GSD).
Figure 3 Comparison of experimental log KA with results from molecular simulations with the HSA crystal structure 1E7G. Solid black lines represent the 1:1 line; dotted lines represent one log unit higher or lower. Error bars reflect one geometric standard deviation (GSD).CC BY · PubMed Central
Figure 4 Residual PFAS quantified in digested HSA pellets, precipitated from equilibrium dialysis experiment.
Figure 4 Residual PFAS quantified in digested HSA pellets, precipitated from equilibrium dialysis experiment.CC BY · PubMed Central
Figure 5 Pseudo-bioconcentration factors (BCFi,pseudo ) of PFAS in aqueous film-forming foams (AFFF) that were bound noncovalently (yellow) or potentially covalently (red) to HSA. Bubble size represents the natural logarithm of the BCFpseudo (legend BCFpseudo = 1). For 12 qualified PFAS with multiple functional groups, a separate bubble of the same size is displayed for each functional group (e.g., noncovalently bound Cl-PFOS contained chloride and sulfonic acid groups and is represented as two bubbles with seven perfluorocarbons).
Figure 5 Pseudo-bioconcentration factors (BCFi,pseudo ) of PFAS in aqueous film-forming foams (AFFF) that were bound noncovalently (yellow) or potentially covalently (red) to HSA. Bubble size represents the natural logarithm of the BCFpseudo (legend BCFpseudo = 1). For 12 qualified PFAS with multiple functional groups, a separate bubble of the same size is displayed for each functional group (e.g., noncovalently bound Cl-PFOS contained chloride and sulfonic acid groups and is represented as two bubbles with seven perfluorocarbons).CC BY · PubMed Central
Figure 6 Violin plots of molecular docking simulated 1E7G docking scores for (a) PFAS with greater than 10 perfluorocarbons and (b) hydrocarbon surfactants identified in AFFF that have greater than 10 carbons. The shape of each violin represents a rotated kernel density plot of 5400 HSA–PFAS binding conformations generated from simulations for six binding pockets. Blue and red colors are used to distinguish experimental results. The compounds with significantly greater peak area (after correction with ISTD peak area) in the noncovalently bound fraction of the protein chamber relative to the chemical chamber are shown in blue. The compounds identified experimentally in the protein chamber that were not significantly greater in peak area relative to the chemical chamber are shown were red. Distributions in red were significantly different than distributions in blue Kruskal–Wallis (p < 0.05) except for those distributions marked with a black X.
Figure 6 Violin plots of molecular docking simulated 1E7G docking scores for (a) PFAS with greater than 10 perfluorocarbons and (b) hydrocarbon surfactants identified in AFFF that have greater than 10 carbons. The shape of each violin represents a rotated kernel density plot of 5400 HSA–PFAS binding conformations generated from simulations for six binding pockets. Blue and red colors are used to distinguish experimental results. The compounds with significantly greater peak area (after correction with ISTD peak area) in the noncovalently bound fraction of the protein chamber relative to the chemical chamber are shown in blue. The compounds identified experimentally in the protein chamber that were not significantly greater in peak area relative to the chemical chamber are shown were red. Distributions in red were significantly different than distributions in blue Kruskal–Wallis (p < 0.05) except for those distributions marked with a black X.CC BY · PubMed Central

Verification

Claims the paper makes

11 Valid · 3 Not valid · 29 Not checked · 77 Own results and interpretations

65 more claims

Create a free account to see all 120 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 34
  • It may state two separate claims as one. 23
  • Borderline: one of the tests for a substantial claim is unclear. 3

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

3 of 3

Citations

Cites

Show all 67 references

Cited by

See all 54

Metadata

Authors
Li, W., Hu, Y., & Bischel, H. N.
Affiliations
University of California, Davis (US)
Corresponding author
Bischel, Heather N.
Venue
Toxics
Publisher
MDPI AG
Published
2021-03-17
Volume
9
Issue
3
Pages
63
ISSN
2305-6304 (ISSN-L)
OpenAlex
W3138131985
Cited by
53
References
68
Citations by year
2026: 10 · 2025: 12 · 2024: 9 · 2023: 10 · 2022: 9 · 2021: 3
FWCI
2.92
Type
journal article
Language
en
Open access
Locations
  • Toxics (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • eScholarship (California Digital Library) (submittedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
UC Davis College of Engineering start-up funds and the National Institute of Environmental Health Sciences of the National Institutes of Health (P30ES023513)
Topics
  • Physical Sciences › Environmental Science › Environmental Chemistry › Per- and polyfluoroalkyl substances research
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Interaction Studies and Fluorescence Analysis
  • Physical Sciences › Environmental Science › Environmental Chemistry › Environmental Chemistry and Analysis
Keywords
per- and polyfluoroalkyl substances, PFAS, aqueous film-forming foam, human serum albumin, protein binding affinity, bioaccumulation potential, suspect screening, equilibrium dialysis, molecular docking, PFBS, PFOS, perfluorooctanoic acid, hydrocarbon surfactants, drinking water contamination, non-covalent binding

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: In-Vitro and In-Silico Assessment of Per- and Polyfluoroalkyl Substances (PFAS) in Aqueous Film-Forming Foam (AFFF) Binding to Human Serum Albumin. https://app.cochat.ai/papers/10.3390/toxics9030063
The paper
Li, W., Hu, Y., & Bischel, H. N. (2021). In-Vitro and In-Silico Assessment of Per- and Polyfluoroalkyl Substances (PFAS) in Aqueous Film-Forming Foam (AFFF) Binding to Human Serum Albumin. Toxics. https://doi.org/10.3390/toxics9030063

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.