PFOS Placental Toxicogenomics

What this research found

Which combination of chemical, cell type and molecular pathway is the most defensible starting point for testing how PFAS chemicals act on the developing placenta? Bioactivity data for perfluorooctanesulfonic acid (PFOS) and perfluorooctanoic acid (PFOA) were filtered against compound-specific cytotoxicity thresholds, then crossed against a single-cell atlas of the first-trimester maternal-fetal interface to score all 192 possible chemical-by-cell-state-by-pathway combinations. PFOS acting on the nuclear receptor PPAR-gamma in invasive extravillous trophoblast ranked first with a composite score of 0.899, and the output is a designed laboratory experiment rather than experimental results.

• Once signals that appear only at cell-killing concentrations are discarded, every retained PFOS target is a PPAR receptor: three PPAR-gamma assays at AC50 values of 13.5, 18.2 and 22.5 µM plus PPAR-alpha at 27 µM, all below the 29 µM PFOS cytotoxicity burst. PFOA's retained targets are less potent, starting at 24 µM against a 52 µM burst.
• PPAR-gamma is where the chemistry and the biology meet. The gene is expressed in 89.5% of extravillous trophoblast cells across all seven donors, and its expression rises through gestation, peaking in the week 8 to 10 window when spiral arteries are remodelled.
• The PFOA hit on the xenobiotic receptor PXR has no cellular substrate to act on. PXR is detected in under 1% of cells in every one of the 32 annotated cell states, and its best-scoring combination ranks 153rd out of 192 — a useful negative result.
• The nomination survives a change of scoring rule: PFOS with PPAR-gamma in extravillous trophoblast scores 0.899 under weights that favour target abundance, and remains top-ranked when all five criteria are weighted equally.
• The proposed test pairs trophoblast stem cell-derived extravillous trophoblast with a seven-point PFOS series from 0.1 to 20 µM, keeping a 1.45-fold margin below the cytotoxicity threshold while bracketing the PPAR-gamma AC50 range, and adds a PFOS plus GW9662 co-treatment arm so that any effect can be shown to run through PPAR-gamma rather than around it.

How it was done

Bioactivity records for PFOS and PFOA across nuclear-receptor, xenobiotic and oxidative-stress assays were filtered in two stages: an activity and quality cut, then a specificity cut discarding any assay whose AC50 exceeded that compound's own cytotoxicity burst threshold, estimated from its most potent mitochondrial-toxicity endpoint. Surviving target genes were mapped onto a first-trimester maternal-fetal interface atlas of 64,734 cells from seven donors and 32 cell states, streamed so that only the 132 panel genes were ever retained, and profiled for abundance, donor-to-donor robustness and change across gestation. Crossing six targets with all 32 cell states gave 192 candidate combinations, ranked by a weighted score over five normalised criteria: potency, hit confidence, expression abundance, donor robustness and developmental timing. The winning combination was then written up as a full dosing and readout protocol in a 25-page report that keeps data-grounded evidence and hypotheses in separately labelled sections.

Data sources

• EPA ToxCast and invitroDB v4.2 bioactivity for PFOS and PFOA — 30 curated compound-by-endpoint records across 19 assays
• E-MTAB-6701 — first-trimester maternal-fetal interface single-cell atlas covering gestational weeks 6 to 12, from Vento-Tormo et al., Nature 2018: 64,734 cells, 7 donors, 32 cell states
• Houck et al., Toxicology 457:152789 (2021) — PFAS screening literature underpinning the curated bioactivity table

Figures from this analysis