Life-long epigenetic programming of cortical architecture by maternal ‘Western’ diet during pregnancy

Abstract

The evolution of human diets led to preferences toward polyunsaturated fatty acid (PUFA) content with ‘Western’ diets enriched in ω-6 PUFAs. Mounting evidence points to ω-6 PUFA excess limiting metabolic and cognitive processes that define longevity in humans. When chosen during pregnancy, ω-6 PUFA-enriched ‘Western’ diets can reprogram maternal bodily metabolism with maternal nutrient supply precipitating the body-wide imprinting of molecular and cellular adaptations at the level of long-range intercellular signaling networks in the unborn fetus. Even though unfavorable neurological outcomes are amongst the most common complications of intrauterine ω-6 PUFA excess, cellular underpinnings of life-long modifications to brain architecture remain unknown. Here, we show that nutritional ω-6 PUFA-derived endocannabinoids desensitize CB1 cannabinoid receptors, thus inducing epigenetic repression of transcriptional regulatory networks controlling neuronal differentiation. We found that cortical neurons lose their positional identity and axonal selectivity when mouse fetuses are exposed to excess ω-6 PUFAs in utero. Conversion of ω-6 PUFAs into endocannabinoids disrupted the temporal precision of signaling at neuronal CB1 cannabinoid receptors, chiefly deregulating Stat3-dependent transcriptional cascades otherwise required to execute neuronal differentiation programs. Global proteomics identified the immunoglobulin family of cell adhesion molecules (IgCAMs) as direct substrates, with DNA methylation and chromatin accessibility profiling uncovering epigenetic reprogramming at >1400 sites in neurons after prolonged cannabinoid exposure. We found anxiety and depression-like behavioral traits to manifest in adult offspring, which is consistent with genetic models of reduced IgCAM expression, to suggest causality for cortical wiring defects. Overall, our data uncover a regulatory mechanism whose disruption by maternal food choices could limit an offspring’s brain function for life.

Full text

Molecular Psychiatry
https://doi.org/10.1038/s41380-019-0580-4
IMMEDIATE COMMUNICATION
Life-long epigenetic programming of cortical architecture
by maternal ‘Western’diet during pregnancy
Valentina Cinquina1●Daniela Calvigioni1●Matthias Farlik 2,14 ●Florian Halbritter 2●Victoria Fife-Gernedl2●
Sally L. Shirran 3●Matthew A. Fuszard3,15 ●Catherine H. Botting3●Patrick Poullet4●Fabiana Piscitelli5●
Zoltán Máté6●Gábor Szabó6●Yuchio Yanagawa7●Siegfried Kasper8●Vincenzo Di Marzo5,9 ●Ken Mackie10 ●
Chris J. McBain 11 ●Christoph Bock 2,12 ●Erik Keimpema 1●Tibor Harkany1,13
Received: 7 March 2019 / Revised: 11 October 2019 / Accepted: 24 October 2019
© Springer Nature Limited 2019
Abstract
The evolution of human diets led to preferences toward polyunsaturated fatty acid (PUFA) content with ‘Western’diets
enriched in ω-6 PUFAs. Mounting evidence points to ω-6 PUFA excess limiting metabolic and cognitive processes that define
longevity in humans. When chosen during pregnancy, ω-6 PUFA-enriched ‘Western’diets can reprogram maternal bodily
metabolism with maternal nutrient supply precipitating the body-wide imprinting of molecular and cellular adaptations at the
level of long-range intercellular signaling networks in the unborn fetus. Even though unfavorable neurological outcomes are
amongst the most common complications of intrauterine ω-6 PUFA excess, cellular underpinnings of life-long modifications
to brain architecture remain unknown. Here, we show that nutritional ω-6 PUFA-derived endocannabinoids desensitize CB1
cannabinoid receptors, thus inducing epigenetic repression of transcriptional regulatory networks controlling neuronal
differentiation. We found that cortical neurons lose their positional identity and axonal selectivity when mouse fetuses are
exposed to excess ω-6 PUFAs in utero. Conversion of ω-6 PUFAs into endocannabinoids disrupted the temporal precision of
signaling at neuronal CB1cannabinoid receptors, chiefly deregulating Stat3-dependent transcriptional cascades otherwise
required to execute neuronal differentiation programs. Global proteomics identified the immunoglobulin family of cell
adhesion molecules (IgCAMs) as direct substrates, with DNA methylation and chromatin accessibility profiling uncovering
epigenetic reprogramming at >1400 sites in neurons after prolonged cannabinoid exposure. We found anxiety and depression-
like behavioral traits to manifest in adult offspring, which is consistent with genetic models of reduced IgCAM expression, to
suggest causality for cortical wiring defects. Overall, our data uncover a regulatory mechanism whose disruption by maternal
food choices could limit an offspring’s brain function for life.
Introduction
Establishing the cerebral connectome relies on mechanisms
that produce unique topological specificity for each neuron
to receive up to 30–40,000 synapses [1] for information
exchange. Pyramidal cells and GABA interneurons, modu-
lated by subcortical afferents, construct cortical microcircuits
that represent the cellular backbones of high-order integrative
processes. During the fetal period, telencephalic stem cell
pools contribute neurons to the cerebral cortex [2,3]:
neuroblasts engage in long-distance migration along complex
trajectories for layer-specific homing before their terminal
morphogenesis, including the specification and directional
growth of their axons, can commence [2]. These develop-
mental stages are controlled by intercellular signals
(including activity-dependence, morphogens, and chemotac-
tic cues) that engage transcriptional differentiation programs
intracellularly to define neuronal identity [3].
Fetal development relies on sequential cell divisions that
generate tissue mass and topological specificity for which
nutrients are transferred along maternal-placental-fetal
nutrient interfaces [2]. A critical metabolic demand for
both cell division and morphological differentiation is
the availability of membrane lipid precursors to envelop
*Tibor Harkany
Tibor.Harkany@meduniwien.ac.at
Extended author information available on the last page of the article
Supplementary information The online version of this article (https://
doi.org/10.1038/s41380-019-0580-4) contains supplementary
material, which is available to authorized users.
1234567890();,:
1234567890();,:

developing cells. Accordingly, about half of the dry weight
of the brain is made up by lipids, of which 20–25% are
long-chain polyunsaturated fatty acids (PUFAs) [4], with
20:4 ω-6 arachidonic acid (AA) being a prominent con-
stituent. Incorporation of long-chain PUFAs into biological
membranes allows for well-controlled membrane expansion
as axons and dendrites form [5]. They also play important
roles in trans-membrane transport and receptor-dependent
second messenger signaling [4]. Significantly, AA also
serves as the ultimate precursor to eicosanoids, including
their non-classical endocannabinoid subfamily [6]: (i) sn-2-
AA containing diacylglycerols, which are cleaved by sn-1-
diacylglycerol lipases to generate 2-arachydonoylglycerol
(2-AG; sn designation for stereospecific numbering by
convention; [7]), or (ii) N-arachidonoyl-phosphatidyletha-
nolamine (NAPE), which is cleaved by a NAPE-selective
phospholipase D to generate N-arachidonoylethanolamine
(anandamide, AEA; [6,8]). Significantly, 2-AG in the
developing brain engages CB1cannabinoid receptors
(CB1Rs) on neurons to act as a repulsive guidance cue for
cell migration [9] and to inhibit neurite outgrowth and
morphogenesis [9,10] (Fig. S1a, b). Despite these infer-
ences, causality between maternal PUFA metabolism and
the (de-)regulation of local-acting instructive lipid signals
indispensable for cortical development [4] and influencing
neurodevelopmental outcomes has not been established.
The bulk of fetal PUFAs (including AA) available for
signaling and in lipid depots are of maternal origin, despite
fetal lipogenesis becoming progressively operational from
mid-gestation [11]. Accordingly, maternal dietary choices
commonly impact birth weight, neurological status, postnatal
cognition and life expectancy in both humans and rodent
models [11]. Medical guidelines highlight maternal food
intake during pregnancy as a major factor for child develop-
ment with a near-equivalent ratio of ω-3 (rich in fish and
‘Japanese’diet) and ω-6 PUFAs recommended as dietary
optimum [12] to satisfy basic metabolic demands by sustained
precursor availability and to prevent adverse neurological
outcomes [13]. However, with more than 30% of children
born to overweight/obese mothers [14], a dramatic deviation
from this dietary optimum likely exists for prolonged periods,
as either the relative or the total amount of ω-6 PUFAs being
in excess [15]. Therefore, ‘Western’diets rich in ω-6 PUFA
precursors are likely to be detrimental for intrauterine devel-
opment [14]. The consumption of ω-6 PUFA-rich diets prior
to conception is equally alarming since the composition of
pre-existing maternal fat depots that undergo accelerated
breakdown during the last trimester of pregnancy [14] deter-
mines the molecular diversity of bioactive lipid precursors for
the fetus as it develops. Moreover, in both humans and
rodents [16,17], high-fat diets increase AA in the circulation
that in turn increases the availability of endocannabinoids and,
consequently, modulates the function of the endocannabinoid
system [15]. Yet a mechanistic and causal relationship
between a shift toward maternal ω-6 PUFA preference (even
if retaining a relative ω-3:ω-6 PUFA ratio) and life-long
neurodevelopmental deficits that render the offspring’sner-
vous system prone to postnatal maladaptation remains
unknown.
Materials and methods
Animals, feeding regime and tissue collection
Female C57Bl6/J, cholecystokinin (CCK)BAC/DsRed and
CCKBAC/DsRed::GAD67gfp/+transgenic mice [18] were
housed in groups in clear plastic cages on a 12 h/12 h light/
dark cycle (lights on at 08:00 h) and in a temperature- (22 ±
2 °C) and humidity (50 ± 10%)-controlled environment.
Food and water were available ad libitum. Embryos and
tissues were obtained from timed matings with the day of
vaginal plug considered as embryonic day (E) 0.5. The day
of birth was always registered as postnatal day (P) 0. Group
sizes conformed to those used as the convention in devel-
opmental biology [10,19–21].
Female animals at 6 weeks of age were randomly
assigned to either a hypercaloric diet enriched ~15-fold in
ω-6 PUFAs while maintaining the ω-3:ω-6 PUFA ratio at
1:8 as in standard chow or to a standard diet (Fig. S2a–b1
and Table S1). Diets were from Special Diet Services
(United Kingdom) and quality controlled by mass-
spectrometry. Maternal body weight was recorded 2–3
times per week for the duration of each study (Fig. S2c).
The effect of maternal diets on fetal brain development was
evaluated at E18.5 after both the ‘priming protocol’(con-
suming a high ω-6 PUFA diet starting 2 weeks prior to
conception; Fig. S2b) or ‘programming protocol’(con-
suming a high ω-6 PUFA diet 8 weeks prior to conception,
Fig. S2b1). This experimental approach, which precluded
blinding the investigators, was motivated by (i) the switch
of diets inducing transient changes in appetite and even loss
of body weight, which, if occurring acutely could have
biased fetal development and (ii) changes in copulatory
behaviors affecting both males and females on the first days
of being exposed to the modified diet. Moreover, long-
lasting postnatal effects of maternal ω-6 PUFA-enriched
diets were determined in adult offspring on P75.
Embryos were collected from n≥5 pregnancies to keep the
number of independent observations sufficient for statistical
analyses, sexed if appropriate, and used for histochemistry,
proteomics, lipid analysis, and molecular biology. The use of
specific or mixed genders for particular experiments was
specified. Embryos were collected by Cesarean sections from
mothers anesthetized by isoflurane (5%, 1 L/min flow rate),
weighed and decapitated with their heads either immersed in
V. Cinquina et al.

4% paraformaldehyde (PFA) in 0.1 M Na+phosphate buffer
(PB; pH 7.4) overnight or snap-frozen in liquid N2and stored
at −80 °C until further processing [10,18]. Likewise, adult
offspring were anesthetized in isoflurane (5%, 1 L/min flow
rate), decapitated, with their brain tissues snap-frozen in liquid
N2andstoredat−80 °C.
Ethical approval of animal studies
Experiments on live animals conformed to the 2010/63/EU
European Communities Council Directive and were approved
by the Austrian Ministry of Science and Research (66.009/
0145-WF/II/3b/2014 and 66.009/0277-WF/V/3b/2017). Par-
ticular effort was directed toward minimizing the number of
animals used and their suffering during experiments.
Click-iT EdU labeling
Embryos were exposed to a single maternal intraperitoneal
injection of 5-ethynyl-2′-deoxyuridine (EdU, 33 mg/kg) on
E14.5 (in the ‘programming protocol’only). At E18.5, male
embryos were selected by PCR genotyping (see ref. [19] for
primer pairs and protocol). Embryonic brains were immer-
sion fixed in 4% PFA in PB overnight, rinsed in PB and
cryoprotected in 30% sucrose (in PB) for 48 h. Serial cor-
onal sections (20 μm) were cut on a ThermoFisher NX70
cryostat, thaw-mounted onto fluorescence-free SuperFrost+
glass slides and stored at −20 °C until processing. EdU was
visualized with Alexa 488-azide using the Click-iT labeling
technology (Life Technologies) [22].
Immunohistochemistry and quantitative
morphometry
Immunofluorescence histochemistry was performed
according to published protocols [10,18,19]. Briefly,
sections were extensively rinsed in PB and subsequently
exposed to a blocking solution consisting of 5% normal
donkey serum (NDS; Jackson ImmunoResearch), 1%
bovine serum albumin (BSA; Sigma), and 0.3% Triton
X-100 (Sigma) in PB at 21–24 °C for 2 h. Next, sections
were incubated with either anti-neural cell adhesion mole-
cule L1 (L1CAM; 1:1000, #ABT143, Millipore), or anti-
Ki67 (which marks proliferation/early neuroblasts; 1:100,
#AB9260, Millipore) primary antibody diluted in PB also
containing 0.1% NDS and 0.3% Triton X-100 at 4 °C
for 48 h. Limbic system associated membrane protein
(LSAMP) was localized to cortical territories of male
CCKBAC/DsRed transgenic mice at E18.5 by first exposing
glass-mounted coronal sections to a 0.1M Tris-glycine
solution (pH 7.4; Sigma) at 21–24 °C for 20 min. After
rinsing in 0.05 M phosphate-buffered saline (PBS), sections
were incubated with mouse anti-LSAMP primary antibody
(1:50; #2G9, Developmental Studies Hybridoma Bank) for
48 h (Dr. Aurea F. Pimenta’s suggestions for histochemistry
are greatly acknowledged). DyLight Fluor 488-tagged sec-
ondary antibodies (1:300; Jackson ImmunoResearch)
were used to reveal the localization of primary antibodies.
Sections were coverslipped with Aquamount (Dako).
Images were acquired on Zeiss 700LSM and 880LSM
confocal laser scanning microscopes. Multi-panel images
were assembled in CorelDRAW X7 (Corel Corp.).
Axonal morphometry was performed as described earlier
[10,19]. In brief, the diameter of first-order bundles made
up by corticofugal axons on E18.5 was measured in serial
sections of 20 μm thickness that had been cut and processed
as above. The number of immunoreactive or genetically
tagged neurons in the fetal cerebral cortex was determined
using equivalent and binned surface areas and expressed as
absolute values.
Liquid chromatography-atmospheric pressure
chemical ionization mass spectrometry
Extraction, purification, and quantification of 2-AG and
AEA from cortical tissues of E18.5 embryos and P75 mice
(for embryos also including dorsal hippocampus) fed with
either hypercaloric or standard diet were carried out
according to published protocols [23,24]. Briefly, tissues
were rapidly dissected out, snap-frozen in liquid N2and
stored at −80 °C. After lipid extraction and prepurification
on silica gel columns, 2-AG and AEA levels from n=3
animals/condition were determined by isotope dilution
using liquid chromatography-atmospheric pressure chemi-
cal ionization mass spectrometry (Shimadzu LCMS-2020).
Results were expressed as pmol/mg of tissue.
iTRAQ proteomics
Single cortical hemispheres (n=6/group, mixed and balanced
for sex from independent pregnancies) were obtained from
E18.5 fetuses. Contralateral hemispheres were snap frozen for
target validation by quantitative real-time PCR and/or Wes-
tern blotting (see below). Proteins were extracted by homo-
genization in lysis buffer (pH 10.0) of triethylammonium
bicarbonate (25 mM, Sigma), Na2CO3(20 mM, Sigma) in the
presence of protease inhibitors (cocktail; Sigma). Protein
concentrations were determined by the bicinchoninic acid
assay (BCA; ThermoFisher). After quantification, proteins
were precipitated in 6 volumes of ice-cold acetone. One-
hundred micrograms of proteins per sample were used for
isobaric tagging for relative and absolute quantitation
(iTRAQ) in an 8-plex layout per the manufacturer’s instruc-
tions (ABSciex). In brief, proteins were denatured, reduced,
alkylated, trypsin digested (Promega), individually labeled
with appropriate iTRAQ tags, pooled, concentrated,
Life-long epigenetic programming of cortical architecture by maternal ‘Western’diet during pregnancy

re-suspended in 1.4 mL of loading buffer (10 mM KH2PO4
pH 3.0 in 25% acetonitrile) and sonicated.
Peptides were separated by cation exchange chromato-
graphy on a PolySulfoethyl A column (PolyLC) over 30
min with a KCl gradient increasing up to 0.5 M, and 0.5 mL
fractions collected. Fifteen fractions across the elution
profile of similar peptide concentrations were generated and
concentrated (SpeedVac). Fractions were re-suspended in
0.1% trifluoroacetic acid (TFA) and desalted on C18 spin
columns (PepClean C18, Thermo Scientific). Half of each
fraction was then injected on an Acclaim PepMap 100 C18
trap and an Acclaim PepMap RSLC C18 column (Ther-
moFisher), using a nanoLC Ultra 2D plus loading pump and
nanoLC as-2 autosampler (Eksigent). The peptides were
loaded onto the trap in a mixture of 98% water, 2% acet-
onitrile and 0.05% TFA, and washed for 20 min to waste
before switching in line with the column and were then
eluted with a gradient of increasing acetonitrile, containing
0.1% formic acid (2–20% acetonitrile in 90 min, 20–40% in
a further 30 min, followed by 98% acetonitrile to clean the
column, before re-equilibration to 2% acetonitrile). The
eluate was sprayed into a TripleTOF 5600 electrospray
tandem mass spectrometer (ABSciex) and analyzed in
Information Dependent Acquisition (IDA) mode, perform-
ing 120 ms of MS followed by 80 ms MSMS analyses on
the 20 most intense peaks seen by MS, with the “Adjust
Collision Energy when using iTRAQ reagent”box ticked in
the method.
Data files were processed by ProteinPilot 4.5 (Sciex) using
the Paragon algorithm, searching against the SwissProt data-
base (March 2013 edition). The following settings were
selected: sample type: iTRAQ 8-plex (peptide labeled),
cysteine alkylation: MMTS, digestion: trypsin, instrument:
TripleTOF 5600, species: mouse, ID focus: biological mod-
ifications and amino acid substitutions and search effort:
thorough. Results files were exported to Microsoft Excel
(Table S2), with statistical analysis in SPSS v.21.
RNA isolation and quantitative PCR
RNA was extracted using the RNeasy mini kit (Qiagen)
with a DNase I step performed to eliminate traces of
genomic DNA. Total RNA was then reverse transcribed to a
cDNA library in a reaction mixture using a high-capacity
cDNA reverse transcription kit (Applied Biosystems). The
cDNA library was then used for quantitative real-time PCR
(CFX-connect, Bio-Rad). Pairs of PCR primers specific
for neural cell adhesion molecule L1 (L1cam), neural
cell adhesion molecule 1 (Ncam), limbic system-associated
membrane protein (Lsamp), neurotrimin (Ntm), cannabinoid
receptor 1 (Cnr1), signal transducer and activator of tran-
scription 3 (Stat3), myeloblastosis oncogene (Myb) and
the CCAAT/enhancer-binding protein beta (Cebpb) were
designed with Primer Bank and National Center for Bio-
technology Information (NCBI) Primer Blast software
(Table S3). Quantitative analysis of gene expression was
performed with the SYBR Green Master Mix Kit (Life
Technologies). Expression levels were normalized to the
housekeeping gene encoding glyceraldehyde-3-phosphate
dehydrogenase (Gapdh) for every sample in parallel assays.
Quantitative Western blotting with total protein
normalization
Cerebral samples of E18.5 male embryos and postnatal
mice (mixed for sex) were obtained as described above.
Total protein labeling was initiated by adding carbocyanine
(Cy)5 dye reagent (GE Healthcare) that had been pre-
diluted (1:10) in ultrapure water. Samples were mixed and
incubated for 5 min at 21–24 °C. The labeling reaction was
terminated by adding Amersham WB loading buffer (GE
Healthcare; 20 μL/sample) containing 40 μM DTT. Samples
were then boiled at 95 °C for 3 min with equal amounts (20
μg/40 μL) and subsequently loaded onto an Amersham WB
gel card (13.5%). Electrophoresis (600 V, 42 min) and
protein transfer onto polyvinylidine-difluoride membranes
(100 V, 30 min) were at default settings in an integrated
Amersham WB system (GE Healthcare) for quantitative
SDS-PAGE and Western blotting of proteins with fluores-
cence detection. After blocking, membranes were incubated
with guinea pig anti-CB1R antibody (1:500, kindly pro-
vided by Dr. Masahiko Watanabe) and mouse anti-LSAMP
antibody (1:100; #2G9, Developmental Studies Hybridoma
Bank) overnight. To demonstrate the specificity of the
guinea pig anti-CB1R antibody, we prepared cerebellar
homogenates from wild-type and Cnr1−/−littermates.
A specific band at the calculated molecular weight of this
receptor (53 kDa) was detected in wild-type but not Cnr1−/−
mice (see Results below). Antibody binding was detected
by using species-specific (anti-mouse and anti-guinea pig)
Cy3-labeled secondary antibodies (1:1000; GE Healthcare).
Membranes were dried before scanning at 560 nm (Cy3)
and 630 nm (Cy5) excitation. Automated image analysis
was performed with the Amersham WB evaluation software
with manual optimization if necessary.
Behavioral tests
Anxiety-like behavior in the first generation offspring to
dams fed with ω-6 PUFA-enriched or control chow were
tested in the open-field and elevated plus maze (n=6 male
mice/group from independent pregnancies) as described
[25–27]. In the open-field test, mice were allowed to
explore a 50 × 50 cm arena for 5 min. The total distance
traveled (m), their speed of movement (cm/s) and the time
spent in an inner 15-cm area (s) were measured. For the
V. Cinquina et al.

elevated plus maze, mice were placed in the center of a
standard maze with its perpendicular arms measuring 10 ×
50 cm each. Mice faced an open arm at the start, and were
allowed to explore the maze for 5 min. The distance
traveled (cm) and the time spent (s) in the open and closed
arms were measured and the number of arm entries (n)
were determined. Parameters for the open arms (distance,
time, entries) were expressed as the percentage of total arm
entries.
Cultures of cortical neurons and IncuCyte-assisted
neurite tracking
Cerebral tissues (cortex/hippocampus) of mouse embryos
were isolated at E14.5 (mixed for sex) or E18.5 (males
only). Cells were mechanically dissociated into single cell
suspension by trypsin digestion (0.1%, 3 min) and plated at
a density of 20,000 cells/well onto poly-D-lysine-coated
(PDL; Sigma) 96-well plates. After 24 h, cultures were
exposed to 2-AG (5 μM; endocannabinoid and full CB1R
agonist), AA (10 μM; 2-AG precursor) or to AM251
(200 nM), an inverse CB1R agonist, alone or in combination
for 4 days (pharmacology) or 7 days (neurite outgrowth)
[10,20]. Drug treatment was performed in quadruplicates
with parallel live imaging on an IncuCyte Zoom live-cell
imaging platform (Essen Bioscience). Time-lapse images
were acquired every 2 h. The growth rate of neurites in each
well was determined as the surface area covered by neurites
(soma free mode) and expressed as mm/mm2surface area.
Surface occupancy of neurons after 4 days in culture was
expressed as ‘body cluster area’and computed as percen-
tage surface area covered by live neurons. This parameter
was considered as a precise measure of cell survival.
DNA methylation
Extracted DNA was subjected to the reduced representation
bisulfite sequencing (RRBS) workflow as described pre-
viously [28,29]: 100 ng of DNA were digested at 37 °C for
12 h with 20 units of MspI and TaqI (New England Biolabs)
in 30 μL of 1× NEB buffer. Fill-in and A-tailing were
performed by addition of Klenow Fragment 3′>5′exo-
(New England Biolabs) and dNTP mix (10 mM dATP,
1 mM dCTP, 1 mM dGTP). After ligation to methylated
Illumina TruSeq LT v2 adaptors using Quick Ligase
(New England Biolabs), the libraries were size selected by
performing a 0.75× clean-up with AMPure XP beads
(Beckman Coulter). The libraries were pooled in equal
amounts based on qPCR data and bisulfite converted using
the EZ DNA Methylation Direct Kit (Zymo Research).
Bisulfite-converted libraries were enriched and quality
control was performed using Qubit dsDNA HS (Life
Technologies) and fragment length was assessed using
high-sensitivity DNA chips on a Bioanalyzer 2000 (Agi-
lent). Sequencing was performed on an Illumina HiSeq
3000/4000 instrument in single-end 50 bp mode. RnBeads
[30] were used for quality control and initial analysis of the
DNA methylation data according to established practices
[31]. We summarized CpGs in 1000 kb bins and performed
differential DNA methylation analysis between ω-6 PUFA
(n=3) and control (n=6) datasets using limma [32] (FDR-
adjusted p-value < 0.05, fold change > 1.5, absolute differ-
ence > 25 percentage points). Locus overlap analysis
(LOLA)[33] was used to calculate the relative over-
representation of hypermethylated tiles with respect to three
published ChIP-seq peak lists from its core database, cor-
responding to binding sites of Stat3,c-Myb and Cebpb [34].
Chromatin accessibility
Open chromatin mapping was performed with the assay
for transposase accessible chromatin (ATAC-seq) [35]
with minor adaptations [36]. In each experiment, 1 × 10
cells were incubated in the transposase reaction mix
(12.5 μL of 2× TD buffer, 2 μLofT
N5 transposase
(Illumina), 0.1% NP40 and 10.25 μL of nuclease-free
water) at 37 °C for 30 min. After DNA purification with
the MinElute kit (Qiagen), 1 μL of the eluted DNA was
used in a qPCR reaction to estimate the optimum number
of amplification cycles. Library amplification was fol-
lowed by SPRI size-selection to exclude fragments larger
than 1200 bp. DNA concentration was measured with a
Qubit fluorometer (Life Technologies). Sequencing was
performed on an Illumina HiSeq 3000/4000 instrument in
single-end 50 bp mode. Raw sequencing data were trim-
med using Skewer [37] followed by alignment to the
GRCh38 assembly of the human reference genome with
Bowtie [38] (parameters: --very-sensitive --no-discordant).
Only deduplicated, uniquely mapped reads with mapping
quality ≥30 were retained for further analysis. To identify
accessible genome regions, we used MACS2 [39](para-
meters: -q 0.1 -g hs). Following initial data processing, all
subsequent analyses were performed in R using Bio-
conductor packages. After removing peaks that overlapped
blacklisted regions from the ENCODE consortium [34]
and merging all overlapping 2-AG peaks, we quantified for
each input dataset the number of reads in the retained
peaks. Raw read counts were loaded into DESeq2 [40]for
normalization and differential analysis (FDR-adjusted
p-value < 0.05). In our analysis, we used the sequencing
flowcell as a covariate to account for batch effects. To
further prevent disproportionate normalization between
globally altered chromatin accessibility landscape (as
observed between sample groups), we added random
genome regions to the calculation step for the size factors
of DESeq2 (three times as many as actual peaks). We
Life-long epigenetic programming of cortical architecture by maternal ‘Western’diet during pregnancy

annotated each peak with the closest gene (distance rela-
tive to gene start coordinate based on Ensembl v77) and
used Enrichr for functional enrichment analysis of dif-
ferentially accessible regions [41].
Statistics
The number of independent samples is indicated in the
graphs and the number of animals is indicated in the figure
legends. All values represent the mean ± s.d. of independent
experiments. Samples were tested for equal variance
throughout. Statistically significant differences were deter-
mined by either Student’st-test (two-tailed for independent
groups) or one-way analysis of variance (ANOVA) fol-
lowed by Bonferroni’spost-hoc test (for multiple groups).
Statistical analysis was performed with Prism 6.0 (Graph-
Pad Software Inc.). Statistical significance is indicated
by asterisks (*p< 0.05, **p< 0.01, ***p< 0.001 and
****p< 0.0001).
Results
Dietary ω-6 PUFA enrichment during pregnancy
impacts fetal cortical architecture
To test if fetal corticogenesis in mice undergoes life-long
modifications after a prenatal diet enriched in ω-6 PUFAs,
dams were fed a ‘fast-food-like’hypercaloric diet with its
ω-6 PUFA content increased through supplementation with
linoleic acid and AA (Fig. S2a) by ~15-fold (to ~31% of all
fatty acids) while maintaining the ω-3:ω-6 ratio (Table S1),
for either 2 weeks (‘priming protocol’; Fig. S2b) or 8 weeks
(‘programming protocol’; Fig. S2b1) prior to conception and
during pregnancy. Besides their weight gain (Fig. S2c), ω-6
PUFA-fed dams had difficulties in conceiving (success rate
of pregnancy: ~50% for both protocols vs. ~75% on control
chow), and had fewer viable embryos (8 ± 1.08 (ω-6 PUFA)
vs. 12 ± 1.2 (control)) but retained ~50% sex ratio.
Initially, dams in the ‘programming protocol’were pulsed
with EdU [22] on E14.5, which coincides with the peak of
cortical neurogenesis [42], to address if ω-6 PUFAs modify
the number and positions of new-born neurons destined to
the cerebral cortex. In E18.5 fetuses, EdU+cells accumu-
lated in the subventricular zone/deep cortical plate with a
proportional decrease in neurons reaching the dorsal cortical
plate/marginal zone that forms first in the outside-in hier-
archy of cortical lamination (Fig. 1a) [42]. These data were
verified for both feeding protocols by histochemical detec-
tion of Ki67+neurons (Fig. S3a), and revealed a marked
increase in Ki67+progenies at locations overlapping with
the sites where EdU+cells had accumulated. EdU and Ki67
do not differentiate between radially migrating pyramidal
cells and interneurons commuting tangentially [42]. There-
fore, we used cholecystokinin (CCK)BAC/DsRed mice whose
genetically tagged DsRed+progeny are predominantly
GABA interneurons (co-expressing Gad1, Fig. S3b–b2) and
populate the cerebral marginal zone [18]. Both maternal
feeding regimes led to a significant redistribution of DsRed+
neurons with their loss in the marginal zone (interneurons)
and accumulation in deep cortical layers (likely pyramidal
cells; Figs. 1b and S3d, d1), establishing impaired cortical
lamination upon maternal ω-6 PUFA enrichment. Next,
axon-specific morphometry [18,19] using neural cell
adhesion molecule L1 (L1NCAM; Fig. S3c) revealed that
corticofugal axons coalesced into enlarged fascicles that
spread in the callosal primordium (‘programming protocol’;
Fig. 1c, c1) in E18.5 fetuses whose mothers were on the ω-6
PUFA-enriched diet. This contrasted with control cases,
which had corticofugal axons spaced evenly as fine-caliber
fascicles [19,43]. These data suggest that maternal dietary
choices during pregnancy carries a risk for fetal cortical
abnormalities.
The misplacement of cortical neurons and their axonal
defects observed upon hypercaloric feeding regimes show
striking similarities with failures of cortical wiring upon
genetic [19,43] and pharmacological [43] disruption of
endocannabinoid signaling in the developing nervous sys-
tem. This notion is particularly relevant since CCK+inter-
neurons in the fetal cerebrum preferentially express CB1Rs
[9] and use endocannabinoids as focal cues for chemotaxis
(Fig. S1a, b). Both linoleic acid and AA, available in excess
in the hypercaloric diet (Table S1), can act as endocanna-
binoid precursors (Fig. S2a). This is compatible with altered
AEA and 2-AG levels in the fetal cerebrum (shown from
the ‘programming protocol’; Fig. 1d) with a shift favoring
the availability of 2-AG, a full agonist at CB1Rs, which we
interpreted as a consequence of increased precursor bioa-
vailability and processing. Desensitization of CB1Rs toge-
ther with the expressional deregulation of both the CB1R
and monoacylglycerol lipase (MAGL), the enzyme
degrading the bulk of 2-AG in fetal brain [6,43], is an
effective means to impart a ‘loss-of-function’signature on
cortical architecture [19]. Indeed, both CB1Rs (Figs. 1e and
S4a) and MAGL (Fig. S4b) showed >50% loss at the pro-
tein level with their mRNA transcripts unaffected (Fig. S4c,
d), in fetuses from hypercaloric diet-exposed dams,
regardless of the length of the feeding protocol. Cumula-
tively, these data suggest that excess ω-6 PUFA intake
increases endocannabinoid precursor availability and
signaling to produce CB1R inactivation over time, pheno-
copying mice with genetic Cnr1 loss- of- function [19].
The biological significance of AA-to-endocannabinoid
conversion as a metabolic principle is reinforced by the
AM251 sensitivity of neurite growth retardation upon AA
supplementation (10 μM, Fig. S1a, b).
V. Cinquina et al.

Dietary ω-6 PUFA enrichment reduces IgCAM
expression in fetal cerebrum
We used iTRAQ [19] as a means to search for molecular
determinants underpinning neuronal deficits. In E18.5
cerebri (including cortical plate and dorsal hippocampus)
(Fig. 2a), increased maternal intake of ω-6 PUFAs
(‘programming protocol’) significantly altered the abun-
dance of 208 proteins, which were assigned to 12 major
functional clusters by gene ontology (Fig. 2a1).
Twelve proteins (6%) belonged to the cluster of ‘cell
adhesion/extracellular matrix composition’(Fig. 2a1). A
refined analysis focusing on male embryos returned 215
proteins as significantly different, including 8 ‘cell adhe-
sion molecules’(4%) as topmost male-specifictargets:
members of the immunoglobulin family of cell adhesion
molecules (IgCAMs) [44] were coincidently and sig-
nificantly downregulated along with presynapse-specific
synaptosomal-associated protein 25 (Fig. 2a2), vesicle-
associated membrane protein 2 and syntaxins (Table S2),
indispensable for presynaptic neurotransmitter release at
mature synapses.
Fig. 1 ω-6 PUFA-enriched diets modulate cell migration by disenga-
ging the endocannabinoid system. a,bRepresentative images (left)
and quantification (right) of EdU+(a;‘programming protocol’) and
CCKBAC/DsRed+(b;‘priming protocol’) neurons in cortical layers of
male embryos at E18.5. Cell counts were performed on equal cortical
tissue surfaces in equal binned players (numbered consecutively from
1 to 10), and expressed as absolute numbers. Figure S3 is referred to
for direct comparisons of the two maternal feeding protocols with
regards to their impact on cortical reorganization in the offspring.
Scale bars =20 μm. cp cortical plate, mz marginal zone, svz sub-
ventricular zone. c,c1Immunofluorescence histochemistry and
quantification of the transverse diameter (c1)offirst-order axonal
fascicles in male E18.5 embryos, labeled for L1NCAM, in the cortical
intermediate zone after nutrient-induced reprogramming. dAnanda-
mide (AEA) and 2-AG levels in E18.5 cerebral tissue of mice (both
sexes) after the ‘programming’protocol. eWestern analysis of CB1R
protein levels in E18.5 cerebral tissue of mixed sexes after both
maternal priming (Pri) and programming (Pro) feeding protocols. Total
protein load was visualized by Cy5 dye reagent and used to normalize
CB1R expression at 53 kDa (arrow). Quantitative data are shown to the
right (*p< 0.05, **p< 0.01, n=5 mice/group). Data are means ± s.d.
from three independent experiments
Life-long epigenetic programming of cortical architecture by maternal ‘Western’diet during pregnancy

Next, we focused on limbic system-associated membrane
protein (Lsamp) because of its confinement to cortical
structures in the dorsal telencephalon [45] (Fig. S5a) and
interaction with other IgCAMs to regulate neurite out-
growth [46] (Fig. S5b). By combining qPCR (Fig. 2b) and
Western blot analyses on subcellular fractions (Fig. 2c), we
verified that heightened ω-6 PUFA intake, even for short
periods, disrupted Lsamp mRNA and protein expression
and subcellular localization. Notably, and even if its dis-
tribution is more general in the fetal hippocampus
(Fig. 2d–d4), Lsamp was invariably but not exclusively
found around CCKBAC/DsRed+fetal interneurons
(Figs. 2d1–d4and S5c, d) by E18.5, thus anatomically
linking IgCAM deregulation to the differentiation and
Fig. 2 Dietary ω-6 PUFA enrichment reduces IgCAM expression in
fetal cerebrum. a–a2Origin of cortical tissue (n=4 biological repli-
cates/group) used for quantitative iTRAQ proteomics (a), ontology
classification of significantly altered proteins in mixed sex embryos
through primary function assignment (a1) and topmost modified pro-
tein targets (ω-6 PUFA-enriched diet vs. control) obtained after the
‘programming’protocol in male embryos (a2). bLsamp mRNA
expression after either the ‘priming’(Pri) or the ‘programming’(Pro)
protocol in fetal cortices on E18.5. mRNAs were quantified by qPCR
and normalized to Gapdh as a housekeeping standard. Data were
expressed relative to fetuses from pregnancies on control chow.
cWestern analysis of LSAMP in subcellular fractions (cytosol [cyto];
membrane fraction [mf]) of fetal cerebral tissues after ‘programming’
protocol. d–d4Representative image of LSAMP immunoreactivity
(d1), particularly in the proximity of CCK+interneurons (d2), in E18.5
hippocampi of CCKBAC/DsRed mice. e,fLsamp mRNA in Cnr1−/−
cortices (e) and in E18.5 cortical neurons treated with AM251 for 96 h
(f). g,hIgCAM mRNA levels after both ‘priming’and ‘programming’
protocols (g) and in Cnr1−/−cortical tissue on E18.5 (h). mRNA
expression was normalized to Gapdh; IgCAM expression was nor-
malized to control-fed (c,g) or wild-type (e,h) mice. All data are on
male mice. CA1 cornu Ammonis subfield 1, cng cingulate cortex, Cpu
caudate putamen, DG dentate gyrus, f fimbria, Ht hypothalamus, Ntm
neurotrimin, sctx somatosensory cortex, Th thalamus. Scale bars =
75 μm(d), 8 μm(d, inset); *p< 0.05, **p< 0.01, n=3 mice/group.
Data are means ± s.d. from triplicate experiments
V. Cinquina et al.

placement of CB1R+cell populations [19,43]. We have
genetically linked Lsamp expression to upstream CB1R
activity by demonstrating significantly reduced Lsamp
mRNA expression in both Cnr1−/−fetuses (E18.5; Fig. 2e)
and when exposing primary cortical neurons to AM251
(200 nM), a selective CB1R antagonist (Fig. 2f). We could
similarly implicate L1CAM, NCAM and neuro-
trimin (Ntm), alternative members of the IgCAM family, in
cortical deficits upon maternal ω-6 PUFA overfeeding
because of their mRNA expression being significantly
reduced by both ‘priming’and ‘programming’protocols
(Fig. 2g) and upon CB1R ablation (Cnr1−/−; Fig. 2h).
Overall, these data suggest that IgCAMs are molecular
targets of CB1R-mediated signal transduction cascades.
Moreover, they qualify the transcriptional modification of
IgCAMs as a candidate mechanism (Fig. 3a) through which
maternal ω-6 PUFA preference impairs fetal brain devel-
opment in male embryos, a hypothesis compatible with a
role for endocannabinoids in axonal growth and directional
guidance in the fetal cerebrum [43].
ω-6 PUFA enrichment reduces chromatin
accessibility of transcription factors downstream
from CB1Rs
The design logic of CB1R activity posits the recruitment of
elaborate transcription factor (TF) networks with signal
transducer and activator of transcription 3 (STAT3) being a
prototypic effector to regulate CB1R-dependent neurite
outgrowth [47]. Here, we reasoned that coincident expres-
sional hindrances within the IgCAM family might indicate
consensus TF inactivation. By using publicly available gene
expression microarray data [47] to select TF targets, we
show that both ‘priming’(Fig. 3b) and ‘programming’
protocols (Fig. S6a) significantly reduced the expression of
Stat3, myeloblastosis oncogene (Myb) and CCAAT/enhan-
cer-binding protein β(Cebpb) in fetal cortices (E18.5).
Next, we argued that long-lasting TF repression might be
brought about by epigenetic modifications particularly since
phytocannabinoids that engage CB1Rs disrupt brain devel-
opment by histone modifications [48]. DNA methylation is
a powerful means of gene regulation [49], which leaves the
longest-lasting marks on chromatin. First, we performed
genome-wide DNA methylation profiling in E18.5 fetal
male brains after the ‘priming’protocol and found broad
gene repression along with CpG islands being hyper-
methylated. Second, enrichment analysis based on ChIP-seq
data sets using the LOLA algorithm of genome-wide dif-
ferences in DNA methylation revealed significant enrich-
ments for the TFs STAT3, MYB and CEBPB in
hypermethylated regions (Fig. 3c). Third, we used the assay
of transposase-accessible chromatin (ATAC-seq), which
is suitable to differentiate between effectively open
(potentially active) and closed (potentially repressed)
chromatin regions whilst also identifying the TFs that bind
promoter regions of differential accessibility [50]. To
mechanistically link CB1R activity and changes in the
landscape of chromatin accessibility without confounds that
could influence read-outs in protracted in vivo experiments,
we performed ATAC-seq after treating primary cultures
with 2-AG (5 µM) or AM251 (200 nM) for 7 days (Fig. 3d).
Here, 2-AG reduced both Lsamp and Stat3 mRNA
expression, recapitulating our in vivo results (Fig. 3e; for
AM251 see Fig. S6b). A total of 1423 genomic regions
were identified with differential accessibility upon treat-
ment, with decreased accessibility after both AM251
(Fig. S6c) and 2-AG exposures but with 2-AG showing
higher effect amplitudes (Fig. 3f). This was particularly
notable for the accessibility of the Stat3 promoter (Figs. 3g,
g1and S6d, d1). These observations were interpreted as
2-AG-induced receptor activation followed by permanent
desensitization over 7 days [51]. We then sought to gain
broader insights in the biological relevance of signature
regions with sensitivity to CB1R-mediated signaling by
performing gene set enrichment analysis of all genes asso-
ciated with differentially accessible genomic loci using
Enrichr [52]. Based on ChIP-seq data available from the
ENCODE project, an unexpectedly high fraction of the
genomic loci losing accessibility upon 2-AG treatment
occurred in the proximity of genes that are targets of TFs
including BRCA1, FOS, CREB1 (Fig. 3h and S6e) that
drive CB1R-induced neurite outgrowth [43,47]. As such,
their inactivation provides a transcriptional platform for
endocannabinoid-induced axonal growth retardation upon
CB1R desensitization (Figs. 3i and S6f) [19,43]. Thus, our
results outline an epigenetic regulatory framework to limit
the transcription of gene sets underpinning neuronal mor-
phogenesis upon long-lasting endocannabinoid excess.
Life-long effects of maternal ω-6 PUFA preference in
first generation offspring
Epigenetic reprogramming of TF networks is biologically
advantageous as a means to enforce genomic changes for as
long as the lifetime of an organism [49]. In accordance with
this principle, we have assessed if key parameters of cortical
reorganization endure into the adulthood of ω-6 PUFA-
exposed offspring after programming feeding protocol.
Firstly, AEA and 2-AG contents remained reflective of
those in fetal brain, with significantly increased 2-AG levels
detected on postnatal day 75 (Fig. 4a). At the same time, ω-
6 PUFA-exposed offspring showed significantly reduced
cortical CB1R load (Fig. 4b), which we attribute to the
presence of and adaptation to persistent 2-AG excess.
Secondly, subcellular distribution of LSAMP, particularly
its membrane-bound fraction, remained reduced (Fig. 4c),
Life-long epigenetic programming of cortical architecture by maternal ‘Western’diet during pregnancy

suggesting disrupted cell adhesion. Thirdly, genetic ablation
of Lsamp provokes heightened behavioral responses to
novel stressors in mice [53]andLsamp gene polymorph-
isms correlate with depressive traits in humans [54].
Therefore, we determined anxiety-like behaviors in the
elevated plus-maze and open-field paradigms. Both tests
revealed significant anxiety, reflected by shortened
exploration of open surfaces (center of arena (Fig. 4d) and
open arm entries (Fig. 4e)). Cumulatively, these data sug-
gest that deregulated endocannabinoid signaling could
contribute to the life-long manifestation of depression-like
behaviors in mice born to dams consuming ω-6
V. Cinquina et al.

PUFA-enriched chow during pregnancy. In agreement with
our hypothesis that pre-conception and congenital excess of
dietary ω-6 PUFAs and ensuing elevation of brain 2-AG
levels cause the anxiety phenotype observed here, Mgll−/−
mice [51], which lose the ability to hydrolyze endocanna-
binoids, exhibit chronic CB1R desensitization and present
anxiety-like behaviors. In contrast, life-long n-3-fatty acid
deprivation impairs CB1R-mediated synaptic signaling and
emotional behavior in adulthood [55].
Discussion
In the present report, we show transgenerational con-
sequences of increasing the ω-6 PUFA nutritional content
during pregnancy in mice on the development of the
offsprings’nervous system. We outline that a critical
mechanism involved is the engagement of the endocanna-
binoid system. As such, nutritional ω-6 PUFA-derived
endocannabinoids desensitize CB1Rs thereby altering
neurogenesis, neuroblast commitment to the cerebral cortex
and the formation of axonal connectivity. We suggest a
link between long-lasting changes of cortical architecture
and epigenetic repression of regulatory TF networks
downstream from CB1Rs that control neuronal differentia-
tion [43,47]. The finding that many of the TF networks
associated with axonal growth are epigenetically controlled
by endocannabinoids might be relevant to public health in
view of the ever-increasing world-wide pandemic of excess
‘Western’diet and ensuing metabolic and behavioral dys-
functions [14].
A series of human longitudinal studies (and experimental
reports) [12,15] pinpoint optimal fatty acid composition in
maternal diet during pregnancy as a means to reduce the
risk of childhood diabetes, hypertension, and obesity.
Despite significant efforts in raising public awareness on the
benefits of ω-3 PUFA-enriched diets (‘fish’or ‘Japanese’
diets) for transgenerational disease prevention, the pre-
valence of dietary routines favoring ω-6 PUFAs is raising
given easy access to high-fat fast food with extraordinary
caloric composition. A reason for this is that intake of a
concentrated source of dietary energy is considered
rewarding particularly if it coincides with the intake of
excess sugar or salt to raise flavor and aroma [56].
Accordingly, preschool children born to mothers who pre-
ferred ω-3 PUFA-rich nutrients during pregnancy show
higher cognitive scores [57] and are less prone to neurologic
birth defects [58]. In contrast, extreme pre-pregnancy
weight is linked as a critical risk factor to attention-deficit
hyperactivity disorder (ADHD) [59], schizophrenia [60]
and anxiety [61]. Notably, the endocannabinoid system is a
critical molecular component of the neuropathology of these
neuropsychiatric disorders in both children and adults
[62,63] with an association between lowered CB1R, CCK
and GAD67 expression in cortical interneurons established
as a key variable [64,65]. Our data are compatible with
these considerations in humans and provide a critical and
causal link, at least in rodent models, between ω-6 PUFA-
enriched diets during pregnancy and lactation, permanent
neurochemical modifications that endure into adulthood and
anxiety-like behaviors. Our in vitro finding that AA-to-
endocannabinoid conversion is a cell-intrinsic metabolic
feature is significant to emphasize that not only systemic
increases in circulating endocannabinoid levels but also
excess endocannabinoid production in situ in developing
organs could contribute to and diversify undesired ω-6
PUFA effects. Nevertheless, since AA is a precursor of a
plethora of alternative bioactive products, many being
essential for cell survival and function, we emphasize that
balancing ω-3:ω-6 PUFA intake might be ultimately bene-
ficial for neuronal development.
Even though our study focused on changes in the fetal
nervous system, one ought to consider the constant inter-
play between the brain and peripheral organs through, e.g.,
long-range hormonal mechanisms. This notion is particu-
larly relevant when studying the endocannabinoid system
with its critical roles in the formation of e.g., bone [66],
Fig. 3 ω-6 PUFA-enriched diets reduce chromatin accessibility of
transcription factors regulated by CB1R activation. aSchematic
representation of signaling events and interactions. Gray color indi-
cates the site of action for AM251 used for pharmacological probing.
bStat3,c-Myb and Cebpb transcription factor mRNA levels in E18.5
cortical tissue (‘priming protocol’). mRNAs were quantified by qPCR
and normalized to Gapdh. Expression was normalized to control mice;
*p<0.05, n=3 mice/group. cBar graph showing the relative
enrichment of overlaps of transcription factor binding sites and
genomic regions (1 kb tiles) with increased DNA methylation in E18.5
cortical tissue upon excess ω-6 PUFA exposure (‘priming protocol’).
LOLA was used to test the enrichment of the three selected tran-
scription factors using published ChIP-seq peaks from ENCODE and
CODEX; *p<0.05, **p<0.01, n=3 mice/group. dTimeline of
in vitro CB1R agonist (2-AG, 5 μM) and antagonist (AM251, 200 nM,
see also SI Fig. 6) exposure. DIV =day in vitro, *p<0.05. Data were
expressed as means ± s.d. from three independent experiments.
eLsamp and Stat3 mRNAs in E18.5 primary cortical neurons after
2-AG (5 μM) exposure. mRNA expression (relative to Gapdh) was
normalized to control. fHeatmap of ATAC-seq data showing
decreased chromatin accessibility after 2-AG treatment. Numbers
denote the normalized and scaled read count per ATAC-seq peak. R1-
R6 for both control and 2-AG treatments identify biological replicates
from two independent experiments. g–g1Genome browser plot
showing ATAC-seq signal intensity across cortical neurons in the
vicinity of the Stat3 gene after 2-AG application, and relative quan-
tification (g1). DeSeq2-normalized read counts. hEnrichment analysis
for all differentially accessible regions determined using Enrichr;
****p< 0.0001. iDynamic time-course analysis of neurite outgrowth
after 2-AG exposure for 7 days aided by an IncuCyte automated
imaging system; **p<0.01, ****p< 0.0001. Data represent means ±
s.d. from three independent experiments. All in vivo and in vitro data
are on male mice. An equivalent dataset on tissues from the ‘pro-
gramming protocol’at E18.5 is shown in SI Fig. 6
Life-long epigenetic programming of cortical architecture by maternal ‘Western’diet during pregnancy

muscle [67], fat depots [68], immune system and pancreas
[69] through recruitment of vastly different ligands and
receptor systems. In obesity, CB1R expression increases in
multiple tissues [70] along with elevated circulating 2-AG
levels as shown in both obese individuals [71] and in
insulin-resistant obese postmenopausal women [72].
Despite these inferences, we are acutely aware that the
translational value of experimental data from laboratory
rodents is often limited to human diseases. One confound
we highlight is the C57Bl6/J strain used here, which is a
favored laboratory subject due to its propensity to develop
metabolic syndrome on high-fat diets [16,68,73]. Never-
theless, the fact that both 2-AG and AEA concentrations are
increased in the brain [16] and peripheral tissues [74]in
conjunction with desensitization of brain CB1Rs [16]in
mice upon diet-induced obesity justify our choice of the
animal model. Another concern is the timing and length of
ω-6 PUFA administration because our experimental proto-
cols precluded the pinpointing of a specific developmental
window with peak sensitivity of the nervous system to
imbalanced nutrient supply. A driving force behind our
experimental design was that dietary preferences in humans
are of long-lasting nature, often spanning years and dec-
ades. As such, the reduced probability of conception in
obese females here is reminiscent of population data in
humans [75] and lasts throughout pregnancy. Moreover, the
metabolic multi-enzyme processing of ω-6 fatty acids is
precisely controlled with brief exposures being well-
tolerated and compensated. Therefore, an experimental
design predisposing to heightened ω-6 PUFA levels in the
long term is amenable for discovery research at the cellular
and molecular levels.
In summary, we show that maternal dietary choices
before and during pregnancy define brain development of
the fetus, alike shown earlier for neonates and adolescents
[13], and strongly influence epigenetic check-points that
instruct neuronal differentiation.
Data availability
Proteomic and high-throughput sequencing data
were deposited in PRIDE (accession no. PXD016180)
and Gene Expression Omnibus (GEO; accession
no. GSE140011), respectively.
Fig. 4 Life-long effects of maternal ω-6 PUFA preference in first
generation offspring. aAnandamide (AEA) and 2-AG levels in cor-
tical tissues of mixed sexes on postnatal day (PND) 75 after exposure
to the ‘programming’protocol; **p< 0.01, n=3–4 mice/group.
b,cWestern analysis of CB1R(b) and LSAMP (c) protein load in
cerebral tissue of mixed sexes. CB1R was enriched in membrane
fractions, while LSAMP was detected in both the cytosol (cyto) and
membranes (mf). Quantitative data are from triplicate experiments.
dOpen-field behaviors of male mice born to and weaned from
the ‘programming protocol’, including the total distance traveled (m),
the time spent (s) in the center of the arena and the velocity of
movement (cm/s); *p<0.05, n=4 mice/group. eElevated plus-maze
behaviors of male offspring, such as the distance traveled (cm) on,
the time spent (s) in and the number of entries (n) into open arms;
*p<0.05, **p< 0.01, n=4 mice/group
V. Cinquina et al.

Acknowledgements We thank A. Reinthaler, for her expert laboratory
assistance, A. Alpar and T. Hökfelt for discussions, the Biomedical
Sequencing Facility at the Center for Molecular Medicine of the
Austrian Academy of Sciences for assistance with next-generation
sequencing, M. Watanabe (Hokkaido University) for antibodies, and
A.F. Pimenta for technical advice with LSAMP cytochemistry. GW
Pharmaceuticals (UK) are acknowledged for providing access to an
IncuCyte Zoom (Essen Bioscience) live-cell imaging platform. This
work was supported by the Swedish Research Council (T.H.); Novo
Nordisk Foundation (T.H.); Hjärnfonden (T.H.); European Research
Council (SECRET-CELLS, ERC-2015-AdG-695136; T.H.), intra-
mural funds of the Medical University of Vienna (T.H.) and the
Wellcome Trust (grant number 094476/Z/10/Z, which funded the
purchase of the TripleTOF 5600 mass spectrometer at the BSRC Mass
Spectrometry and Proteomics Facility, University of St. Andrews).
M.F. is supported by a special research program of the Austrian Sci-
ence Fund (FWF-F61).
Author contributions TH conceived the project; VC, MF, CB, VDM,
CJM, EK and TH designed the experiments; CB, CJB and TH pro-
cured funding; VC, DC, MF, VG, MAF, FH, SLS, CHB and FP
performed the experiments; VC, DC, MF, FH, PP, FP analyzed the
data; ZM, GS and KM developed unique reagents and tools for the
project and VC, MF and TH wrote the manuscript with input from all
co-authors.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of
interest.
Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
References
1. Schlingloff D, Kali S, Freund TF, Hajos N, Gulyas AI.
Mechanisms of sharp wave initiation and ripple generation. J
Neurosci. 2014;34:11385–98.
2. Campbell K. Cortical neuron specification: it has its time and
place. Neuron. 2005;46:373–6.
3. Ayoub AE, Oh S, Xie Y, Leng J, Cotney J, Dominguez MH, et al.
Transcriptional programs in transient embryonic zones of the
cerebral cortex defined by high-resolution mRNA sequencing.
Proc Natl Acad Sci USA. 2011;108:14950–5.
4. Bazinet RP, Laye S. Polyunsaturated fatty acids and their
metabolites in brain function and disease. Nat Rev Neurosci.
2014;15:771–85.
5. Pfenninger KH. Plasma membrane expansion: a neuron’s Hercu-
lean task. Nat Rev Neurosci. 2009;10:251–61.
6. Ahn K, McKinney MK, Cravatt BF. Enzymatic pathways that
regulate endocannabinoid signaling in the nervous system. Chem
Rev. 2008;108:1687–707.
7. IUPAC-IUB Commission on Biochemical Nomenclature. The
nomenclature of lipids (recommendations 1976). J Lipid Res.
1978;19:114–28.
8. Bisogno T, Howell F, Williams G, Minassi A, Cascio MG,
Ligresti A, et al. Cloning of the first sn1-DAG lipases points to the
spatial and temporal regulation of endocannabinoid signaling in
the brain. J Cell Biol. 2003;163:463–8.
9. Berghuis P, Rajnicek AM, Morozov YM, Ross RA, Mulder J,
Urban GM, et al. Hardwiring the brain: endocannabinoids shape
neuronal connectivity. Science. 2007;316:1212–6.
10. Keimpema E, Barabas K, Morozov YM, Tortoriello G, Torii M,
Cameron G, et al. Differential subcellular recruitment of mono-
acylglycerol lipase generates spatial specificity of 2-arachidonoyl
glycerol signaling during axonal pathfinding. J Neurosci. 2010;
30:13992–4007.
11. Kirkham TC, Williams CM, Fezza F, Di Marzo V. Endocanna-
binoid levels in rat limbic forebrain and hypothalamus in relation
to fasting, feeding and satiation: stimulation of eating by
2-arachidonoyl glycerol. Br J Pharmacol. 2002;136:550–7.
12. Simopoulos AP. Importance of the ratio of omega-6/omega-3
essential fatty acids: evolutionary aspects. World Rev Nutr Diet.
2003;92:1–22.
13. Manduca A, Bara A, Larrieu T, Lassalle O, Joffre C, Laye S, et al.
Amplification of mGlu5-endocannabinoid signaling rescues
behavioral and synaptic deficits in a mouse model of adolescent
and adult dietary polyunsaturated fatty acid imbalance. J Neurosci.
2017;37:6851–68.
14. Stephenson J, Heslehurst N, Hall J, Schoenaker D, Hutchinson J,
Cade JE, et al. Before the beginning: nutrition and lifestyle in the
preconception period and its importance for future health. Lancet.
2018;391:1830–41.
15. Simopoulos AP. An increase in the omega-6/omega-3 fatty acid
ratio increases the risk for obesity. Nutrients. 2016;8:128.
16. Massa F, Mancini G, Schmidt H, Steindel F, Mackie K, Angioni
C, et al. Alterations in the hippocampal endocannabinoid system
in diet-induced obese mice. J Neurosci. 2010;30:6273–81.
17. Di Marzo V, Matias I. Endocannabinoid control of food intake
and energy balance. Nat Neurosci. 2005;8:585–9.
18. Calvigioni D, Mate Z, Fuzik J, Girach F, Zhang MD, Varro A,
et al. Functional differentiation of cholecystokinin-containing
interneurons destined for the cerebral cortex. Cereb Cortex.
2017;27:2453–68.
19. Tortoriello G, Morris CV, Alpar A, Fuzik J, Shirran SL, Calvigioni
D, et al. Miswiring the brain: Delta9-tetrahydrocannabinol disrupts
cortical development by inducing an SCG10/stathmin-2 degradation
pathway. EMBO J. 2014;33:668–85.
20. Berghuis P, Dobszay MB, Wang X, Spano S, Ledda F, Sousa
KM, et al. Endocannabinoids regulate interneuron migration and
morphogenesis by transactivating the TrkB receptor. Proc Natl
Acad Sci USA. 2005;102:19115–20.
21. Mulder J, Aguado T, Keimpema E, Barabas K, Ballester Rosado
CJ, Nguyen L, et al. Endocannabinoid signaling controls pyr-
amidal cell specification and long-range axon patterning. Proc
Natl Acad Sci USA. 2008;105:8760–5.
22. Zeng C, Pan F, Jones LA, Lim MM, Griffin EA, Sheline YI, et al.
Evaluation of 5-ethynyl-2’-deoxyuridine staining as a sensitive
and reliable method for studying cell proliferation in the adult
nervous system. Brain Res. 2010;1319:21–32.
23. De Marchi N, De Petrocellis L, Orlando P, Daniele F, Fezza F, Di
Marzo V. Endocannabinoid signalling in the blood of patients
with schizophrenia. Lipids Health Dis. 2003;2:5.
24. Matias I, Gonthier MP, Orlando P, Martiadis V, De Petrocellis L,
Cervino C, et al. Regulation, function, and dysregulation of
endocannabinoids in models of adipose and beta-pancreatic cells
and in obesity and hyperglycemia. J Clin Endocrinol Metab.
2006;91:3171–80.
25. Walf AA, Frye CA. The use of the elevated plus maze as an assay
of anxiety-related behavior in rodents. Nat Protoc. 2007;2:322–8.
26. Harkany T, Mulder J, Sasvari M, Abraham I, Konya C, Zarandi M,
et al. N-Methyl-D-aspartate receptor antagonist MK-801 and radical
scavengers protect cholinergic nucleus basalis neurons against beta-
amyloid neurotoxicity. Neurobiol Dis. 1999;6:109–21.
27. Harkany T, O’Mahony S, Kelly JP, Soos K, Toro I, Penke B, et al.
Beta-amyloid(Phe(SO3H)24)25-35 in rat nucleus basalis induces
behavioral dysfunctions, impairs learning and memory and
Life-long epigenetic programming of cortical architecture by maternal ‘Western’diet during pregnancy

disrupts cortical cholinergic innervation. Behav Brain Res. 1998;
90:133–45.
28. Sheffield NC, Pierron G, Klughammer J, Datlinger P,
Schonegger A, Schuster M, et al. DNA methylation hetero-
geneity defines a disease spectrum in Ewing sarcoma. Nat Med.
2017;23:386–95.
29. Klughammer J, Kiesel B, Roetzer T, Fortelny N, Nemc A, Nen-
ning KH, et al. The DNA methylation landscape of glioblastoma
disease progression shows extensive heterogeneity in time and
space. Nat Med. 2018;24:1611–24.
30. Assenov Y, Muller F, Lutsik P, Walter J, Lengauer T, Bock C.
Comprehensive analysis of DNA methylation data with RnBeads.
Nat Methods. 2014;11:1138–40.
31. Bock C. Analysing and interpreting DNA methylation data. Nat
Rev Genet. 2012;13:705–19.
32. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma
powers differential expression analyses for RNA-sequencing and
microarray studies. Nucleic Acids Res. 2015;43:e47.
33. Sheffield NC, Bock C. LOLA: enrichment analysis for genomic
region sets and regulatory elements in R and Bioconductor.
Bioinformatics. 2016;32:587–9.
34. Consortium EP. An integrated encyclopedia of DNA elements in
the human genome. Nature. 2012;489:57–74.
35. Buenrostro JD, Wu B, Chang HY, Greenleaf WJ. ATAC-seq: a
method for assaying chromatin accessibility genome-wide. Curr
Protoc Mol Biol. 2015;109:21.29.1–9.
36. RendeiroAF,SchmidlC,StreffordJC,WalewskaR,DavisZ,
Farlik M, et al. Chromatin accessibility maps of chronic
lymphocytic leukaemia identify subtype-specific epigenome
signatures and transcription regulatory networks. Nat Commun.
2016;7:11938.
37. Jiang H, Lei R, Ding SW, Zhu S. Skewer: a fast and accurate
adapter trimmer for next-generation sequencing paired-end reads.
BMC Bioinformatics. 2014;15:182.
38. Langmead B, Salzberg SL. Fast gapped-read alignment with
Bowtie 2. Nat Methods. 2012;9:357–9.
39. Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein
BE, et al. Model-based analysis of ChIP-Seq (MACS). Genome
Biol. 2008;9:R137.
40. Love MI, Huber W, Anders S. Moderated estimation of fold
change and dispersion for RNA-seq data with DESeq2. Genome
Biol. 2014;15:550.
41. Chen EY, Tan CM, Kou Y, Duan Q, Wang Z, Meirelles GV, et al.
Enrichr: interactive and collaborative HTML5 gene list enrich-
ment analysis tool. BMC Bioinformatics. 2013;14:128.
42. Kepecs A, Fishell G. Interneuron cell types are fit to function.
Nature. 2014;505:318–26.
43. Maccarrone M, Guzman M, Mackie K, Doherty P, Harkany T.
Programming of neural cells by (endo)cannabinoids: from phy-
siological rules to emerging therapies. Nat Rev Neurosci. 2014;
15:786–801.
44. Cavallaro U, Dejana E. Adhesion molecule signalling: not always
a sticky business. Nat Rev Mol Cell Biol. 2011;12:189–97.
45. Reinoso BS, Pimenta AF, Levitt P. Expression of the
mRNAs encoding the limbic system-associated membrane
protein (LAMP): I. Adult rat brain. J Comp Neurol. 1996;375:
274–88.
46. Gil OD, Zhang L, Chen S, Ren YQ, Pimenta A, Zanazzi G, et al.
Complementary expression and heterophilic interactions between
IgLON family members neurotrimin and LAMP. J Neurobiol.
2002;51:190–204.
47. Bromberg KD, Ma’ayan A, Neves SR, Iyengar R. Design logic of
a cannabinoid receptor signaling network that triggers neurite
outgrowth. Science. 2008;320:903–9.
48. Szutorisz H, Hurd YL. Epigenetic effects of cannabis exposure.
Biol Psychiatry. 2016;79:586–94.
49. Lowdon RF, Jang HS, Wang T. Evolution of epigenetic regulation
in vertebrate genomes. Trends Genet. 2016;32:269–83.
50. Buenrostro JD, Giresi PG, Zaba LC, Chang HY, Greenleaf WJ.
Transposition of native chromatin for fast and sensitive epige-
nomic profiling of open chromatin, DNA-binding proteins and
nucleosome position. Nat Methods. 2013;10:1213–8.
51. Imperatore R, Morello G, Luongo L, Taschler U, Romano R, De
Gregorio D. et al. Genetic deletion of monoacylglycerol lipase
leads to impaired cannabinoid receptor CB(1)R signaling and
anxiety-like behavior. J Neurochem. 2015;135:799–813.
52. Farlik M, Halbritter F, Muller F, Choudry FA, Ebert P, Klugh-
ammer J, et al. DNA methylation dynamics of human hemato-
poietic stem cell differentiation. Cell Stem Cell. 2016;19:808–22.
53. Catania EH, Pimenta A, Levitt P. Genetic deletion of Lsamp
causes exaggerated behavioral activation in novel environments.
Behav Brain Res. 2008;188:380–90.
54. Koido K, Traks T, Balotsev R, Eller T, Must A, Koks S, et al.
Associations between LSAMP gene polymorphisms and
major depressive disorder and panic disorder. Transl Psychiatry.
2012;2:e152.
55. Lafourcade M, Larrieu T, Mato S, Duffaud A, Sepers M, Matias I,
et al. Nutritional omega-3 deficiency abolishes endocannabinoid-
mediated neuronal functions. Nat Neurosci. 2011;14:345–50.
56. Drewnowski A. Energy intake and sensory properties of food. Am
J Clin Nutr. 1995;62:1081S–5S.
57. Helland IB, Smith L, Saarem K, Saugstad OD, Drevon CA.
Maternal supplementation with very-long-chain n-3 fatty acids
during pregnancy and lactation augments children’s IQ at 4 years
of age. Pediatrics. 2003;111:e39–44.
58. Crawford MA, Golfetto I, Ghebremeskel K, Min Y, Moodley T,
Poston L, et al. The potential role for arachidonic and doc-
osahexaenoic acids in protection against some central nervous
system injuries in preterm infants. Lipids. 2003;38:303–15.
59. Rodriguez A, Miettunen J, Henriksen TB, Olsen J, Obel C,
Taanila A, et al. Maternal adiposity prior to pregnancy is asso-
ciated with ADHD symptoms in offspring: evidence from three
prospective pregnancy cohorts. Int J Obes. 2008;32:550–7.
60. Schaefer CA, Brown AS, Wyatt RJ, Kline J, Begg MD, Bresnahan
MA, et al. Maternal prepregnant body mass and risk of schizo-
phrenia in adult offspring. Schizophr Bull. 2000;26:275–86.
61. Rodriguez A. Maternal pre-pregnancy obesity and risk for
inattention and negative emotionality in children. J Child Psychol
Psychiatry. 2010;51:134–43.
62. Parolaro D, Realini N, Vigano D, Guidali C, Rubino T. The
endocannabinoid system and psychiatric disorders. Exp Neurol.
2010;224:3–14.
63. Lafenetre P, Chaouloff F, Marsicano G. Bidirectional regulation
of novelty-induced behavioral inhibition by the endocannabinoid
system. Neuropharmacology. 2009;57:715–21.
64. Eggan SM, Stoyak SR, Verrico CD, Lewis DA. Cannabinoid CB1
receptor immunoreactivity in the prefrontal cortex: Comparison of
schizophrenia and major depressive disorder. Neuropsycho-
pharmacology. 2010;35:2060–71.
65. Eggan SM, Hashimoto T, Lewis DA. Reduced cortical cannabi-
noid 1 receptor messenger RNA and protein expression in schi-
zophrenia. Arch Gen Psychiatry. 2008;65:772–84.
66. Ross RA. The enigmatic pharmacology of GPR55. Trends Phar-
macol Sci. 2009;30:156–63.
67. Iannotti FA, Silvestri C, Mazzarella E, Martella A, Calvigioni D,
Piscitelli F, et al. The endocannabinoid 2-AG controls skeletal
muscle cell differentiation via CB1 receptor-dependent inhibition
of Kv7 channels. Proc Natl Acad Sci USA. 2014;111:E2472–81.
68. Ruiz de Azua I, Mancini G, Srivastava RK, Rey AA, Cardinal P,
Tedesco L, et al. Adipocyte cannabinoid receptor CB1 regulates
energy homeostasis and alternatively activated macrophages. J
Clin Investig. 2017;127:4148–62.
V. Cinquina et al.

69. Malenczyk K, Keimpema E, Piscitelli F, Calvigioni D, Bjorklund
P, Mackie K, et al. Fetal endocannabinoids orchestrate the orga-
nization of pancreatic islet microarchitecture. Proc Natl Acad Sci
USA. 2015;112:E6185–94.
70. Pagotto U, Vicennati V, Pasquali R. The endocannabinoid system
and the treatment of obesity. Ann Med. 2005;37:270–5.
71. Bluher M, Engeli S, Kloting N, Berndt J, Fasshauer M, Batkai S,
et al. Dysregulation of the peripheral and adipose tissue endo-
cannabinoid system in human abdominal obesity. Diabetes.
2006;55:3053–60.
72. Abdulnour J, Yasari S, Rabasa-Lhoret R, Faraj M, Petrosino S,
Piscitelli F, et al. Circulating endocannabinoids in insulin
sensitive vs. insulin resistant obese postmenopausal women. A
MONET group study. Obesity. 2014;22:211–6.
73. Gallou-Kabani C, Vige A, Gross MS, Rabes JP, Boileau C,
Larue-Achagiotis C, et al. C57BL/6J and A/J mice fed a high-fat
diet delineate components of metabolic syndrome. Obesity.
2007;15:1996–2005.
74. Alvheim AR, Malde MK, Osei-Hyiaman D, Lin YH, Pawlosky RJ,
Madsen L, et al. Dietary linoleic acid elevates endogenous 2-AG and
anandamide and induces obesity. Obesity. 2012;20:1984–94.
75. Guelinckx I, Devlieger R, Beckers K, Vansant G. Maternal
obesity: pregnancy complications, gestational weight gain and
nutrition. Obes Rev. 2008;9:140–50.
Affiliations
Valentina Cinquina1●Daniela Calvigioni1●Matthias Farlik 2,14 ●Florian Halbritter 2●Victoria Fife-Gernedl2●
Sally L. Shirran 3●Matthew A. Fuszard3,15 ●Catherine H. Botting3●Patrick Poullet4●Fabiana Piscitelli5●
Zoltán Máté6●Gábor Szabó6●Yuchio Yanagawa7●Siegfried Kasper8●Vincenzo Di Marzo5,9 ●Ken Mackie10 ●
Chris J. McBain 11 ●Christoph Bock 2,12 ●Erik Keimpema 1●Tibor Harkany1,13
1Department of Molecular Neurosciences, Center for Brain
Research, Medical University of Vienna, Vienna, Austria
2CeMM Research Center for Molecular Medicine of the Austrian
Academy of Sciences, Vienna, Austria
3School of Chemistry, University of St. Andrews, St. Andrews,
United Kingdom
4Institut Curie & INSERM U900, Paris, France
5Endocannabinoid Research Group, Institute of Biomolecular
Chemistry (ICB), National Research Council (CNR),
Pozzuoli, Italy
6Institute of Experimental Medicine, Hungarian Academy of
Sciences, Budapest, Hungary
7Department of Genetic and Behavioral Neuroscience, Gunma
University School of Medicine, Maebashi, Japan
8Department of Psychiatry and Psychotherapy, Medical University
of Vienna, Vienna, Austria
9Canada Excellence Research Chair, Institut Universitaire de
Cardiologie et de Pneumologie de Québec and Institut sur la
Nutrition et les Aliments Fonctionnels, Université Laval,
Québec, QC, Canada
10 Department of Psychological & Brain Sciences, Indiana
University, Bloomington, Indiana, USA
11 Program in Developmental Neuroscience, Eunice Kennedy-
Shriver National Institute of Child Health and Human
Development, NIH, Bethesda, USA
12 Department of Laboratory Medicine, Medical University of
Vienna, Vienna, Austria
13 Department of Neuroscience, Karolinska Institutet,
Stockholm, Sweden
14 Present address: Department of Dermatology, Medical University
of Vienna, Vienna, Austria
15 Present address: Faculty of Medicine, Martin-Luther University,
Halle-Wittenberg, Halle, Germany
Life-long epigenetic programming of cortical architecture by maternal ‘Western’diet during pregnancy