Science - USA (2022-02-18)

mean of serum concentrations in SELMA
pregnant women. The tested dose range was
adjusted to the experimental models to ac-
commodate their differential responsiveness.

Acute MIX N exposure disrupts human
neurodevelopmental pathways

To define the molecular impact of MIX N,
we started off with primary neural stem cells
sourced from the cortex of human fetuses at 11
and 19 weeks of gestational age [henceforth
human fetal primary neural stem cells (HFPNSCs)
in the figures and fetal progenitors in the text].
Given the potentially nonlinear and nonmo-
notonic dose-response patterns associated with
EDC mixtures ( 17 , 18 ), the experimental design
included five doses of MIX N, ranging in con-
centration from 0.1X to 1000X, and a global
assessment of the early impact on gene expres-
sion to define the proximal targets (fig. S1A).
To this end, RNA sequencing (RNA-seq) was

performed after 48 hours of MIX N exposure,

and patterns of EDC dose-dependent tran-

scriptional responses were determined using

an analysis that considers MIX N dilutions as

distinct categories. By not assuming any par-

ticular response pattern (linearity or monoto-

nicity), this approach allowed us to define a

core list of 16 differentially expressed genes

(DEGs) {false discovery rate (FDR) < 0.05; abso-

lute log–fold change [abs(logFC)] > 0.2 (henceforth

logFC); log–counts per million (logCPM) > 0} that

follow similar dose responses in the two fetal cor-

tical lines (fig. S1, B to E). This core set includes the

genesEPHB2andCLSTN2, an ephrin receptor

and a transmembrane protein of the cadherin

superfamily, respectively. These two genes are

particularly relevant to the phenotype scored

in the SELMA cohort (language delay) because

their down-regulation by MIX N (fig. S1F) is

consistent with their autism spectrum disorder

(ASD)–causative role by haploinsufficiency that

was initially defined in the Simons Simplex

Collection of 2700 families and later validated

in additional cohorts ( 19 – 25 ).

Chronic MIX N exposure uncovers convergence

with genetic causes of ASD

The observation that even acute exposure to

MIX N triggered the dysregulation of a core

set of phenotypically relevant genes in two

genetically independent fetal cortex samples,

sourced at different developmental stages,

motivated us to investigate the molecular

impact of MIX N exposure in a setting that

more closely recapitulates human in utero

exposure.Tothisend,weselectedtwoanchor

doses,1Xand1000XMIXN,andtestedthem

in two experimental models with a minimum

of four replicates per condition: (i) the same

19-weeks-of-gestational-age fetal progenitor

line that was already subjected to acute expo-

sure, now exposed for 2 weeks to mimic the

Caporaleet al.,Science 375 , eabe8244 (2022) 18 February 2022 4 of 15

AB

C

Cortical organoids

CO

MIX N CHRONIC EXPO

DMSO

1X

1000X

Minimum number of replicates

per condition

Cell line 3391B: sex

MIX N CHRONIC EXPO

DMSO

1X

1000X

Minimum number of replicates

per condition

Cell line E3381-1: sex

Human Fetal Primary

Neural Stem Cells

HFPNSC

HFPNSC MIXN DEGs CO MIXN DEGs

Gene Set Enrichment Analysis of HFPNSC DEGs Gene Set Enrichment Analysis of CO DEGs

Fig. 2. MIX N chronic exposure of human cellular models.(A) Schematic
representation of the experimental design for chronic (15 days) MIX N treatment of
HFPNSCs and a heatmap showing the relative gene abundance [z-scores of surrogate
variable analysis (SVA)Ðcorrected values] of the DEGs (FDR < 0.05, logFC > 0.5,
logCPM > 0) identified through differential expression analysis of MIX N dilutions
versus negative control. CNT, unexposed samples; DMSO, samples exposed to 0.1%
DMSO; 1X, samples exposed to 1X MIX N; 1000X, samples exposed to 1000X MIX N.

(B) Schematic representation of the experimental design for chronic (50 days)

MIX N treatment of human COs and a heatmap showing the relative gene abundance

(z-scores of SVA-corrected values) of the DEGs (FDR < 0.05, logFC > 0.5, logCPM > 0)

identified through differential expression analysis of MIX N dilutions versus negative

control. (C) Gene sets significantly (adjustedP< 0.1) associated with MIX N

exposure by GSEA in HFPNSCs and COs, respectively; results are presented as

−log10 adjustedPvalues and normalized enrichment score (NES).

RESEARCH | RESEARCH ARTICLE