Supplemental Information
EXTENDED EXPERIMENTAL PROCEDURES
B. fragilis Treatment
At 3 weeks of age, saline and poly(I:C) offspring across individual litters were weaned into cages of 4 nonlittermate offspring of the
same treatment group to generate a randomized experimental design (
Lazic, 2013
). Cages within the poly(I:C) versus saline treatment
groups were selected at random for treatment with nonenterotoxigenic
B. fragilis
NCTC 9343 (
Sears, 2009
) or vehicle, every other day
for 6 days. To preclude any confounding effects of early life stress on neurodevelopment and behavior, suspensions were not admin-
istered by oral gavage. For
B. fragilis
treatment, 10
^
10 cfu freshly grown
B. fragilis
was suspended in 1 ml 1.5% sodium bicarbonate,
mixed with 4 ml sugar-free applesauce and spread over four standard food pellets. We find that
42% of
B. fragilis
colony forming
units are recovered from the applesauce inoculum at 48 hr after administration, suggesting that both viable and nonviable
B. fragilis
is
ingested during the treatment. For vehicle treatment, saline and poly(I:C) animals were fed 1.5% sodium bicarbonate in applesauce
over food pellets. Applesauce and pellets were completely consumed by mice of each treatment group by 48 hr after administration.
The same procedure was used for treatment with mutant
B. fragilis
lacking PSA and
B. thetaiotaomicron
.
Intestinal qRT-PCR, Western Blots, and Cytokine Profiles
Gut tissue was flushed with HBSS and either (1) homogenized in Trizol for RNA isolation and reverse transcription according to
Hsiao
and Patterson (2011
) or (2) homogenized in Tissue Extraction Reagent I (Invitrogen) containing protease inhibitors (Roche) for protein
assays. For cytokine profiling, mouse 20-plex cytokine arrays (Invitrogen) were run on the Luminex FLEXMAP 3D platform by the
Clinical Immunobiology Correlative Studies Laboratory at the City of Hope (Duarte, CA). Western blots were conducted according
to standard methods and probed with rabbit anti-claudin 8 or rabbit anti-claudin 15 (Invitrogen) at 1:100 dilution.
Microbial DNA Extraction, 16S rRNA Gene Amplification and Pyrosequencing
For each experimental group, 10 mice (5 males, 5 females) from different litters were randomly selected for single housing at weaning
and treatment with either vehicle or
B. fragilis
, as described above. Bacterial genomic DNA was extracted from mouse fecal pellets
using the MoBio PowerSoil Kit following protocols benchmarked as part of the NIH Human Microbiome Project. The V3-V5 regions of
the 16S rRNA gene were PCR amplified using individually barcoded universal primers containing linker sequences for 454-pyrose-
quencing. Sequencing was performed at the HGSC at BCM using a multiplexed 454-Titanium pyrosequencer.
16S rRNA Gene Sequence Analysis
FASTA and quality files were obtained from the Alkek Center for Metagenomics and Microbiome Research at the Baylor College of
Medicine. 16S data were processed and its diversity was analyzed using QIIME 1.6 software package (
Caporaso et al., 2010b
)as
follows. Sequences < 200 bp and > 1,000 bp, and sequences containing any primer mismatches, barcode mismatches, ambiguous
bases, homopolymer runs exceeding six bases, or an average quality score of below 30 were discarded and the remaining
sequences were checked for chimeras and clustered to operational taxonomic units (OTUs) using the USearch pipeline (
Edgar,
2010; Edgar et al., 2011
) with a sequence similarity index of 97%. OTUs were subsequently assigned taxonomic classification using
the basic local alignment search tool (BLAST) classifier (
Altschul et al., 1990
), based on the small subunit nonredundant reference
database release 111 (
Quast et al., 2013
) with 0.001 maximum e-value. These taxonomies were then used to generate taxonomic
summaries of all OTUs at different taxonomic levels. For tree-based alpha- and beta diversity analyses, representative sequences
for each OTU were aligned using PyNAST (
Caporaso et al., 2010a
) and a phylogenetic tree was constructed based on this alignment
using FastTree (
Price et al., 2009
). Alpha diversity estimates (by Observed Species and Faith’s phylogenetic diversity [PD];
Faith,
1992
) and evenness (by Simpson’s evenness and Gini Coefficient;
Wittebolle et al., 2009
) were calculated and compared between
groups using a nonparametric test based on 100 iterations using a rarefaction of 2,082 sequences from each sample. For beta
diversity, even sampling of 2,160 sequences per sample was used, and calculated using weighted and unweighted UniFrac (
Lozu-
pone and Knight, 2005
). Beta diversity was compared in a pairwise fashion (S versus P, P versus P+BF), from unweighted UniFrac
distance matrixes, using the analysis of similarity (ANOSIM;
Fierer et al., 2010
), permutational multivariate analysis of variance
(PERMANOVA;
Anderson, 2008; McArdle and Anderson, 2001
), permutational analysis of multivariate dispersions (PERMDISP;
Anderson et al., 2006
), and Moran’s I, each with 999 permutations to determine statistical significance.
Identification of Differences in Specific OTUs
Key OTUs, that discriminate between Saline and Poly(I:C) treatment groups, and between Poly(I:C) and Poly(I:C) +
B. fragilis
treat-
ment groups, were identified using an unbiased method from OTU tables, generated by QIIME, using three complimentary analyses:
(1) Metastats comparison (
White et al., 2009
) and (2) the Random Forests algorithm, first under QIIME (
Knights et al., 2011
) and
subsequently coupled with Boruta feature selection, in the Genboree microbiome toolset (
Riehle et al., 2012
), and (3) the Galaxy
platform-based LDA Effect Size analysis (LEfSe;
Segata et al., 2011
). Only OTUs that differ significantly between treatment groups
were candidates for further analyses (p < 0.05 for [1] and [3], and > 0.0001 mean decrease in accuracy for Random Forests and
subsequent identification by the Boruta algorithm). Metastats analyses were done using the online interface (
http://metastats.
cbcb.umd.edu
) with QIIME-generated OTU tables of any two treatment groups. The Random Forests algorithm was used to identify
discriminatory OTUs in the QIIME software package (
Breiman, 2001; Knights et al., 2011
), comparing two treatment groups at a time,
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
S1
based on 1,000 trees and a 10-fold cross-validation, and was further validated and coupled with the Boruta feature selection algo-
rithm, as implemented in the Genboree Microbiome toolset (
Kursa and Rudnicki, 2010; Riehle et al., 2012
). Only those OTUs that were
confirmed by the Boruta algorithm were defined as discriminatory. The ratio between observed and calculated error rates was used
as a measure of confidence for Random Forests Analyses: this ratio was 5.0 for saline versus poly(I:C) (with an estimated error of 0.1
±
0.21) and 2.86 for poly(I:C) versus poly(I:C) +
B. fragilis
(with an estimated error of 0.23
±
0.22). In order to overcome any misidenti-
fication by any one of the three methods only OTUs that were identified by at least two of the three above methods were defined as
discriminatory. For the analyses in
Figure 1
and
Table S1
, OTUs that were significantly altered by MIA were identified by comparing
the saline versus poly(I:C) groups. For the analyses in
Figure 3
, we compared the poly(I:C) versus poly(I:C)+
B. fragilis
groups, and only
report only those OTUs that have also been identified by the analyses in
Figure 1
and
Table S1
. In addition, we compared the results
obtained by Random Forests Analysis with feature selection by Boruta to those obtained by Random Forests Analysis with a cutoff of
0.001 mean decrease in accuracy.
To generate a phylogenetic tree depecting the closest cultured type strains to key OTUs identified, key OTUs were than aligned
using the SINA aligner (
http://www.arb-silva.de/aligner/
;(
Pruesse et al., 2012
), compared to the SILVA reference database release
111 (
Quast et al., 2013
) using Arb (
Ludwig et al., 2004
) and visualized using FigTree (
http://tree.bio.ed.ac.uk/software/figtree/
). Heat
maps of key OTUs were generated by extracting their relative abundance from the OTU table. These data were then normalized
(so that the sum of squares of all values in a row or column equals one), first by OTU and subsequently by sample, and clustered
by correlation using Cluster 3.0 (
de Hoon et al., 2004
). Finally, abundance data were visualized using Java TreeView (
Saldanha, 2004
).
Metabolomics Screening
Sera were collected by cardiac puncture from behaviorally validated adult mice. Samples were extracted and analyzed on GC/MS,
LC/MS and LC/MS/MS platforms by Metabolon, Inc. Protein fractions were removed by serial extractions with organic aqueous
solvents, concentrated using a TurboVap system (Zymark) and vacuum dried. For LC/MS and LC/MS/MS, samples were reconsti-
tuted in acidic or basic LC-compatible solvents containing > 11 injection standards and run on a Waters ACQUITY UPLC and
Thermo-Finnigan LTQ mass spectrometer, with a linear ion-trap front-end and a Fourier transform ion cyclotron resonance mass
spectrometer back-end. For GC/MS, samples were derivatized under dried nitrogen using bistrimethyl-silyl-trifluoroacetamide
and analyzed on a Thermo-Finnigan Trace DSQ fast-scanning single-quadrupole mass spectrometer using electron impact ioniza-
tion. Chemical entities were identified by comparison to metabolomic library entries of purified standards. Following log transforma-
tion and imputation with minimum observed values for each compound, data were analyzed using two-way ANOVA with contrasts.
In Vitro Immune Assays
Methods for Treg and Gr-1 flow cytometry and CD4+ T cell in vitro stimulation are described in
Hsiao et al. (2012
). Briefly, cells were
harvested in complete RPMI from spleens and mesenteric lymph nodes. For subtyping of splenocytes, cells were stained with Gr-1
APC, CD11b-PE, CD4-FITC and Ter119-PerCP-Cy5.5 (Biolegend). For detection of Tregs, splenocytes were stimulated for 4 hr with
PMA/ionomycin in the presence of GolgiPLUG (BD Biosciences), blocked for Fc receptors and labeled with CD4-FITC, CD25-PE,
Foxp3-APC and Ter119-PerCP-Cy5.5. Samples were processed using the FACSCalibur cytometer (BD Biosciences) and analyzed
using FlowJo software (TreeStar). For CD4+ T cell stimulation assays, 10
^
6 CD4+ T cells were cultured in complete RPMI with PMA
(50 ng/ml) and ionomycin (750 ng/ml) for 3 d at 37C with 5% (vol/vol) CO2. Each day, supernatant was collected for ELISA assays to
detect IL-6 and IL-17, according to the manufacturer’s instructions (eBioscience).
B. fragilis
Colonization Assay
Fecal samples were sterilely collected from MIA and control offspring at 1, 2, and 3 weeks after the start of treatment with
B. fragilis
or
vehicle. Germ-free mice were treated with
B. fragilis
as described above to serve as positive controls. DNA was isolated from fecal
samples using the QIAamp DNA Stool Mini Kit (QIAGEN). 50 ng DNA was used for qPCR with
B. fragilis
-specific, 5
0
TGATTCCG
CATGGTTTCATT 3
0
and 5
0
CGACCCATAGAGCCTTCATC 3
0
, and universal 16S primers 5
0
ACTCCTACGGGAGGCAGCAGT 3
0
and 5
0
ATTACCGCGGCTGCTGGC 3
0
according to
Odamaki et al. (2008
).
Behavioral Testing
Mice were tested beginning at 6 weeks of age for PPI, open field exploration, marble burying, social interaction and adult ultrasonic
vocalizations, in that order, with at least 5 days between behavioral tests. Behavioral data for
B. fragilis
treatment and control groups
(
Figure 5
) represent cumulative results collected from multiple litters of 3-5 independent cohorts of mice for PPI and open field tests,
2-4 cohorts for marble burying, 2 cohorts for adult male ultrasonic vocalization and 1 cohort for social interaction. Discrepancies in
sample size across behavioral tests reflect differences in when during our experimental study a particular test was implemented.
Prepulse Inhibition
PPI tests are used as a measure of sensorimotor gating and were conducted and analyzed as in
Geyer and Swerdlow (2001
) and
(
Smith et al., 2007
). Briefly, mice were acclimated to the testing chambers of the SR-LAB startle response system (San Diego Instru-
ments) for 5 min, presented with six 120 db pulses of white noise (startle stimulus) and then subjected to 14 randomized blocks of
either no startle, startle stimulus only, 5 db prepulse with startle or 15 db prepulse with startle. The startle response was recorded by a
S2
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
pliezo-electric sensor, and the percent PPI is defined as: [((startle stimulus only – 5 or 15 db prepulse with startle)/startle stimulus
only)*100].
Open field exploration. The open field test is widely used to measure anxiety-like and locomotor behavior in rodents. Mice were
placed in 50
3
50 cm white Plexiglas boxes for 10 min. An overhead video camera recorded the session, and Ethovision software
(Noldus) was used to analyze the distance traveled, and the number of entries and duration of time spent in the center arena (central
17 cm square).
Marble Burying
Marble burying is an elicited repetitive behavior in rodents analogous to those observed in autistic individuals (
Silverman et al., 2010
).
This test was conducted and analyzed according to methods described in
Thomas et al., (2009
) and (
Malkova et al., 2012
). Mice were
habituated for 10 min to a novel testing cage containing a 4 cm layer of chipped cedar wood bedding and then transferred to a new
housing cage. 18 glass marbles (15 mm diameter) were aligned equidistantly 6
3
3 in the testing cage. Mice were returned to the
testing cage and the number of marbles buried in 10 min was recorded.
Sociability and Social Preference
Social interaction tests were conducted and analyzed according to methods adopted from
Sankoorikal et al. (2006
) and (
Yang et al.,
2011
). Briefly, testing mice were habituated for 10 min to a 50
3
75 cm Plexiglas three-chambered apparatus containing clear inter-
action cylinders in each of the side chambers. Sociability was tested in the following 10 min session, where the testing mouse was
given the opportunity to explore a novel same-sex, age-matched mouse in one interaction cylinder (social object) versus a novel toy
(green sticky ball) in the other interaction cylinder of the opposite chamber. Social preference was tested in the final 10 min session,
where the testing mouse was given the opportunity to explore a now familiar mouse (stimulus mouse from the previous sociability
session) versus a novel unfamiliar same-sex, age-matched mouse. In each session, the trajectory of the testing mouse was tracked
with Ethovision software (Noldus). Sociability data are presented as preference for the mouse over the toy: percent of time in the
social chamber - percent of time in the nonsocial chamber, and social preference data are presented as preference for the unfamiliar
mouse over the familiar mouse: percent of time in the unfamiliar mouse chamber—percent of time in the familiar mouse chamber.
Similar indexes were measured for chamber entries, and entries into and duration spent in the contact zone (7
3
7 cm square
surrounding the interaction cylinder).
Adult Ultrasonic Vocalizations
Male mice produce USVs in response to female mice as an important form of communication (
Portfors, 2007
). We measured adult
male USV production in response to novel female exposure according to methods described in
Grimsley et al. (2011), Scattoni et al.
(2011)
, and
Silverman et al. (2010
). Adult males were single-housed one week before testing and exposed for 20 min to an unfamiliar
adult female mouse each day starting four days prior to testing in order to provide a standardized history of sexual experience and to
adjust for differences in social dominance. On testing day, mice were habituated to a novel cage for 10 min before exposure to a novel
age-matched female. USVs were recorded for 3 min using the UltraSoundGate microphone and audio system (Avisoft Bioacoustics).
Recordings were analyzed using Avisoft’s SASLab Pro software after fast Fourier transformation at 512 FFT-length and detection by
a threshold-based algorithm with 5 ms hold time. Data presented reflect duration and number of calls produced in the 3 min session.
4EPS Synthesis and Detection
Potassium 4-ethylphenylsulfate was prepared using a modification of a procedure reported for the synthesis of aryl sulfates in
Bur-
lingham et al. (2003
) and
Grimes (1959
)(
Figure S7
A). 4-ethylphenol (Sigma-Aldrich, 5.00 g, 40.9 mmol) was treated with sulfur
trioxide-pyridine complex (Sigma-Aldrich, 5.92 g, 37.2 mmol) in refluxing benzene (20 ml, dried by passing through an activated
alumina column). After 3.5 hr, the resulting solution was cooled to room temperature, at which point the product crystallized. Isolation
by filtration afforded 7.93 g of crude pyridinium 4-ethylphenylsulfate as a white crystalline solid. 1.00 g of this material was dissolved
in 10 ml of 3% triethylamine in acetonitrile and filtered through a plug of silica gel (Silicycle, partical size 32-63
m
m), eluting with 3%
triethylamine in acetonitrile. The filtrate was then concentrated, and the resulting residue was dissolved in 20 ml of deionized water
and eluted through a column of Dowex 50WX8 ion exchange resin (K+ form), rinsing with 20 ml of deionized water. The ion exchange
process was repeated once more, and the resulting solution concentrated under vacuum to afford 618 mg (55% overall yield) of
potassium 4-ethylphenylsulfate as a white powder (
Figure S7
A).
1H and 13C NMR spectra of authentic potassium 4-ethylphenylsulfate were recorded on a Varian Inova 500 spectrometer and are
reported relative to internal DMSO-
d
5
(1H,
d
= 2.50; 13C,
d
= 39.52). A high-resolution mass spectrum (HRMS) was acquired using an
Agilent 6200 Series TOF with an Agilent G1978A Multimode source in mixed ionization mode (electrospray ionization [ESI] and atmo-
spheric pressure chemical ionization [APCI]). Spectroscopic data for potassium 4-ethylphenylsulfate are as follows: 1H NMR (DMSO-
d
6
, 500 MHz)
d
7.11 – 7.04 (m, 4H), 2.54 (q,
J
= 7.6 Hz, 2H), 1.15 (t,
J
= 7.6 Hz, 3H); 13C NMR (DMSO-
d
6
, 126 MHz)
d
151.4, 138.3,
127.9, 120.6, 27.5, 16.0; HRMS (Multimode-ESI/APCI) calculated for C8H9O4S [M–K]- 201.0227, found 201.0225.
Authentic 4EPS and serum samples were analyzed by LC/MS using an Agilent 110 Series HPLC system equipped with a photo-
diode array detector and interfaced to a model G1946C single-quadrupole expectospray mass spectrometer. HPLC separations
were obtained at 25
C using an Agilent Zorbax XDB-C18 column (4.6 mm
3
50 mm
3
5 um particle size). The 4EPS ion was detected
using selected ion monitoring for ions of m/z 200.9 and dwell time of 580 ms/ion, with the electrospray capillary set at 3 kV. Authentic
potassium 4EPS was found to possess a retention time of 6.2 min when eluted in 0.05% trifluoroacetic acid and acetonitrile, using a
10 min linear gradient from 0%–50% acetonitrile. For quantification of 4EPS in mouse sera, a dose-response curve was constructed
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
S3
by plotting the total ion count peak area for known concentrations of authentic potassium 4EPS against the analyte concentration
(R
^
2 = 0.9998;
Figure S7
B). Mouse serum samples were diluted 4-fold with acetonitrile and centrifuged at 10,000 g at 4
C for
3 min. 10 ul of supernatant was injected directly into the HPLC system.
SUPPLEMENTAL REFERENCES
Anderson, M.J. (2008). A new method for non-parametric multivariate analysis of variance. Austral Ecol.
26
, 32–46.
Breiman, L. (2001). Random forests. Mach. Learn.
45
, 5–32.
Burlingham, B.T., Pratt, L.M., Davidson, E.R., Shiner, V.J., Jr., Fong, J., and Widlanski, T.S. (2003). 34S isotope effect on sulfate ester hydrolys
is: mechanistic
implications. J. Am. Chem. Soc.
125
, 13036–13037.
Caporaso, J.G., Bittinger, K., Bushman, F.D., DeSantis, T.Z., Andersen, G.L., and Knight, R. (2010a). PyNAST: a flexible tool for aligning sequences
to a template
alignment. Bioinformatics
26
, 266–267.
Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Pen
̃
a, A.G., Goodrich, J.K., Gordon, J.I., et al. (2010b).
QIIME allows analysis of high-throughput community sequencing data. Nat. Methods
7
, 335–336.
de Hoon, M.J.L., Imoto, S., Nolan, J., and Miyano, S. (2004). Open source clustering software. Bioinformatics
20
, 1453–1454.
Edgar, R.C. (2010). Search and clustering orders of magnitude faster than BLAST. Bioinformatics
26
, 2460–2461.
Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., and Knight, R. (2011). UCHIME improves sensitivity and speed of chimera detection. Bioinformat
ics
27
,
2194–2200.
Faith, D.P. (1992). Conservation Evaluation and Phylogenetic Diversity. Biol. Conserv.
61
, 1–10.
Geyer, M.A., and Swerdlow, N.R. (2001). Measurement of startle response, prepulse inhibition, and habituation. Curr Protoc Neurosci
Chapter 8
, Unit 8 7.
Grimes, A.J. (1959). Synthesis of 35S-labelled arylsulphates byintact animals and by tissue preparations, with particular reference to L-tyrosin
e O-sulphate. Bio-
chem. J.
73
, 723–729.
Hsiao, E.Y., McBride, S.W., Chow, J., Mazmanian, S.K., and Patterson, P.H. (2012). Modeling an autism risk factor in mice leads to permanent immune dy
sre-
gulation. Proc. Natl. Acad. Sci. USA
109
, 12776–12781.
Hsiao, E.Y., and Patterson, P.H. (2011). Activation of the maternal immune system induces endocrine changes in the placenta via IL-6. Brain Behav. Im
mun.
25
,
604–615.
Knights, D., Costello, E.K., and Knight, R. (2011). Supervised classification of human microbiota. FEMS Microbiol. Rev.
35
, 343–359.
Kursa, M.B., and Rudnicki, W.R. (2010). Feature Selection with the Boruta Package. J. Stat. Softw.
36
, 1–13.
Lazic, S.E. (2013). Comment on ‘‘Stress in puberty unmasks latent neuropathological consequences of prenatal immune activation in mice’’. Science
340
, 811.
Lozupone, C., and Knight, R. (2005). UniFrac: a new phylogenetic method for comparing microbial communities. Appl. Environ. Microbiol.
71
, 8228–8235.
Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H., Yadhukumar, Buchner, A., Lai, T., Steppi, S., Jobb, G., et al. (2004). ARB: a software envi
ronment for
sequence data. Nucleic Acids Res.
32
, 1363–1371.
McArdle, B.H., and Anderson, M.J. (2001). Fitting multivariate models to community data: a comment on distancebased redundancy analysis. Ecology
82
,
290–297.
Odamaki, T., Xiao, J.Z., Sakamoto, M., Kondo, S., Yaeshima, T., Iwatsuki, K., Togashi, H., Enomoto, T., and Benno, Y. (2008). Distribution of differe
nt species of
the Bacteroides fragilis group in individuals with Japanese cedar pollinosis. Appl. Environ. Microbiol.
74
, 6814–6817.
Portfors, C.V. (2007). Types and functions of ultrasonic vocalizations in laboratory rats and mice. J. Am. Assoc. Lab. Anim. Sci.
46
, 28–34.
Price, M.N., Dehal, P.S., and Arkin, A.P. (2009). FastTree: computing large minimum evolution trees with profiles instead of a distance matrix. Mol. B
iol. Evol.
26
,
1641–1650.
Pruesse, E., Peplies, J., and Glo
̈
ckner, F.O. (2012). SINA: accurate high-throughput multiple sequence alignment of ribosomal RNA genes. Bioinformatics
28
,
1823–1829.
Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., Peplies, J., and Glo
̈
ckner, F.O. (2013). The SILVA ribosomal RNA gene database project:
improved data processing and web-based tools. Nucleic Acids Res.
41
(Database issue), D590–D596.
Riehle, K., Coarfa, C., Jackson, A., Ma, J., Tandon, A., Paithankar, S., Raghuraman, S., Mistretta, T.A., Saulnier, D., Raza, S., et al. (2012). The Ge
nboree Micro-
biome Toolset and the analysis of 16S rRNA microbial sequences. BMC Bioinformatics
13
(
Suppl 13
), S11.
Saldanha, A.J. (2004). Java Treeview—extensible visualization of microarray data. Bioinformatics
20
, 3246–3248.
Sankoorikal, G.M., Kaercher, K.A., Boon, C.J., Lee, J.K., and Brodkin, E.S. (2006). A mouse model system for genetic analysis of sociability: C57BL/
6J versus
BALB/cJ inbred mouse strains. Biol. Psychiatry
59
, 415–423.
Scattoni, M.L., Ricceri, L., and Crawley, J.N. (2011). Unusual repertoire of vocalizations in adult BTBR T+tf/J mice during three types of social enc
ounters. Genes
Brain Behav.
10
, 44–56.
Sears, C.L. (2009). Enterotoxigenic Bacteroides fragilis: a rogue among symbiotes. Clin. Microbiol. Rev.
22
, 349–369.
Segata, N., Izard, J., Waldron, L., Gevers, D., Miropolsky, L., Garrett, W.S., and Huttenhower, C. (2011). Metagenomic biomarker discovery and expl
anation.
Genome Biol.
12
, R60.
White, J.R., Nagarajan, N., and Pop, M. (2009). Statistical methods for detecting differentially abundant features in clinical metagenomic sample
s. PLoS Comput.
Biol.
5
, e1000352.
Wittebolle, L., Marzorati, M., Clement, L., Balloi, A., Daffonchio, D., Heylen, K., De Vos, P., Verstraete, W., and Boon, N. (2009). Initial communit
y evenness favours
functionality under selective stress. Nature
458
, 623–626.
Yang, M., Silverman, J.L., and Crawley, J.N. (2011). Automated three-chambered social approach task for mice. Curr Protoc Neurosci
Chapter 8
, Unit 8 26.
S4
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
TJP1
TJP3
CLDN15
CLDN2
CLDN3
CLDN4
CLDN7
CLDN8
CLDN12
CLDN13
CLDN15
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
mRNA/
ACTB
(fold change)
P
S
*
TJP1
CLDN1
CLDN2
CLDN13
0
1
2
3
4
mRNA/
ACTB
(fold change)
P+BF
P
*
*
n.s.
n.s.
*
n.s.
A
B
C
FGF-basic
GM-CSF
IFN-y
IL-1a
IL-1b
IL-2
IL-4
IL-5
IL-6
IL-12p40/p70
IL-13
IL-17
IP-10
KC
MCP-1
MIG
MIP-1a
TNF-a
VEGF
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
Cytokine/g protein (fold change)
P+BF
P
**
*
*
*
*
**
**
*
*
*
**
*
Figure S1.
B. fragilis
Treatment Has Little Effect on Tight Junction Expression and Cytokine Profiles in the Small Intestine, Related to
Figures
1
and
3
(A) Expression of tight junction components relative to
b
-actin in small intestines of adult saline and poly(I:C) offspring. Data for each gene are normalized to
expression levels in saline offspring. n = 8/group.
(B) Quantification of the effect of
B. fragilis
treatment on expression of notable tight junction components relative to
b
-actin in small intestines of MIA offspring.
Data for saline and poly(I:C) are as in (A). n = 8/group.
(C) Protein levels of cytokines and chemokines relative to total protein content in small intestines of adult saline, poly(I:C) and poly(I:C)+
B. fragilis
offspring.
Data are normalized to expression levels in saline offspring. Asterisks directly above bars indicate significance compared to saline control (norma
lized to 1, as
denoted by the black line), whereas asterisks at the top of the graph denote statistical significance between poly(I:C) and poly(I:C)+
B. fragilis
groups. n = 8-10/
group.
Data are presented as mean
±
SEM. *p < 0.05, **p < 0.01, S = saline+vehicle, p = poly(I:C)+vehicle, P+BF = poly(I:C)+
B. fragilis
.
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
S5
(legend on next page)
S6
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
Figure S2. No Evidence for Persistent Colonization of
B. fragilis
after Treatment of MIA Offspring, Related to
Figures 2
and
4
(A) Evenness of the gut microbiota, as indicated by the Gini coefficient. n = 10/group.
(B) Richness of the gut microbiota, as illustrated by rarefaction curves plotting Faith’s Phylogenetic Diversity (PD) versus the number of sequence
s for each
treatment group. n = 10/group.
(C) Mean relative abundance of OTUs classified by taxonomic assignments at the class level for the most abundant taxa (left) and least abundant taxa (ri
ght) for
poly(I:C) versus poly(I:C)+
B. fragilis
treatment. n = 10/group.
(D) Levels of
B. fragilis
16S sequence (top) and bacterial 16S sequence (bottom) in fecal samples collected at 1, 2, and 3 weeks posttreatment of adult offspring
with vehicle or
B. fragilis
. Germ-free mice colonized with
B. fragilis
were used as a positive control. Data are presented as quantitative RT-PCR cycling thresholds
[C(t)], where C(t) > 34 (hatched line) is considered negligible, and for C(t) < 34, lesser C(t) equates to stronger abundance. n = 1, where each represen
ts pooled
sample from 3-5 independent cages.
(E) Levels of
B. fragilis
16S sequence (top) and bacterial 16S sequence (bottom) in cecal samples collected at 1, 2, and 3 weeks posttreatment of adult offspring
with vehicle or
B. fragilis
. Germ-free mice colonized with
B. fragilis
were used as a positive control. Data are presented as quantitative RT-PCR cycling thresholds
[C(t)], where C(t) > 34 (hatched line) is considered negligible, and for C(t) < 34, lesser C(t) equates to stronger abundance. n = 1, where each represen
ts pooled
sample from 3-5 independent cages.
Data are presented as mean
±
SEM. S = saline+vehicle, p = poly(I:C)+vehicle, P+BF = poly(I:C)+
B. fragilis
, GF+BF = germ-free+
B. fragilis
.
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
S7
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Frequency (%)
Foxp3+ Foxp3+ Foxp3+
CD25- CD25+
n.s.
*
n.s.
**
n.s.
n.s.
0
10
20
30
40
50
60
70
80
Frequency (%)
CD11b+ Gr-1+ CD4+
n.s.
**
n.s.
***
S
P
P+BF
123
0
100
200
300
400
500
600
700
800
900
1000
1100
IL-17 (pg/ml)
S
P
P+BF
n.s.
***
Duration of culture (days)
123
0
50
100
150
200
250
300
350
400
450
500
550
IL-17 (pg/ml)
S
P+BF
P
n.s.
*
Duration of culture (days)
0
5
10
15
20
25
30
35
Center entries
**
*
***
0
10
20
30
40
50
60
70
Center duration (s)
**
**
***
0
10
20
30
40
Distance (m)
0
10
20
30
40
50
Marbles buried (%)
***
S
P+BF
P
P+BF
PSA
ABC
D
E
F
123
0
10
20
30
40
50
60
70
80
90
IL-6 (pg/ml)
S
P
P+BF
n.s.
***
Duration of culture (days)
123
0
5
10
15
20
25
IL-6 (pg/ml)
S
P+BF
P
n.s.
*
Duration of culture (days)
Figure S3.
B. fragilis
Treatment Has No Effect on Systemic Immune Dysfunction in MIA Offspring, Related to
Figure 5
(A) Percent frequency of Foxp3+ CD25+ T regulatory cells from a parent population of CD4+ TCRb+ cells, measured by flow cytometry of splenocytes from ad
ult
saline, poly(I:C) and poly(I:C)+
B. fragilis
offspring. n = 5/group.
(B) Percent frequency of CD4+ T helper cells and CD11b+ and Gr-1+ neutrophilic and monocytic cells from a parent population of TER119- cells, measured
by
flow cytometry of splenocytes from adult saline, poly(I:C) and poly(I:C)+
B. fragilis
offspring. n = 5/group.
(C) Production of IL-17 and IL-6 by splenic CD4+ T cells isolated from adult saline and poly(I:C) offspring, after in vitro stimulation with PMA/ionom
ycin. Treatment
effects were assessed by repeated-measures two-way ANOVA with Bonferroni post hoc test. n = 5/group.
(D) Production of IL-17 and IL-6 by CD4+ T cells isolated from mesenteric lymph nodes of adult saline and poly(I:C) offspring, after in vitro stimulati
on with PMA/
ionomycin. Treatment effects were assessed by repeated-measures two-way ANOVA with Bonferroni post hoc test. n = 5/group.
(E) Anxiety-like and locomotor behavior in the open field exploration assay for adult MIA offspring treated with mutant
B. fragilis
lacking production of poly-
saccharide A (PSA). Data indicate total distance traveled in the 50
3
50 cm open field (right), duration spent in the 17 x17 cm center square (middle) and number of
entries into the center square (left) over a 10 min trial. Data for saline, poly(I:C) and poly(I:C)+
B. fragilis
groups are as in
Figure 5
. poly(I:C)+
B. fragilis
with PSA
deletion: n = 17.
(F) Repetitive burying of marbles in a 6
3
8 array in a 10 min trial. Data for saline, poly(I:C) and poly(I:C)+
B. fragilis
groups are as in
Figure 5
. poly(I:C)+
B. fragilis
with
PSA deletion: n = 17.
Data are presented as mean
±
SEM. *p < 0.05, **p < 0.01, ***p < 0.001. S = saline+vehicle, p = poly(I:C)+vehicle, P+BF = poly(I:C)+
B. fragilis
, P+BF
D
PSA =
poly(I:C)+
B. fragilis
with PSA deletion.
S8
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
0
10
20
30
Center entries
*
**
0
5
10
15
20
25
30
35
40
45
50
55
60
65
Center duration (s)
**
***
**
0
10
20
30
40
Distance (m)
0
5
10
15
20
25
30
35
40
45
50
55
60
65
PPI (%)
S
P
P+BF
P+BT
5 db 15 db
*
*
0
50
100
150
200
250
300
350
400
450
500
550
Number of calls
*
*
0
25
50
75
Duration per call (ms)
*
p
=0.07
***
0
10
20
30
40
Total call duration (s)
*
p
=0.08
***
0
10
20
30
40
50
Marbles buried (%)
**
p
=0.07
*
S
P
P+BF
P+BT
A
B
C
D
Figure S4. Amelioration of Autism-Related Behaviors in MIA Offspring Is Not Specific to
B. fragilis
Treatment, Related to
Figure 5
(A) Anxiety-like and locomotor behavior in the open field exploration assay, as measured by total distance traveled in the 50
3
50 cm open field (right), duration
spent in the 17 x17 cm center square (middle), and number of entries into the center of the field (left) over a 10 min trial. Poly(I:C)+
B. thetaiotaomicron
: n = 32.
(B) Repetitive burying of marbles in a 3
3
6 array during a 10 min trial. Poly(I:C)+
B. thetaiotaomicron
:n=32.
(C) Communicative behavior, as measured by total number (left), average duration (middle) and total duration (right) of ultrasonic vocalizations p
roduced by adult
male mice during a 10 min social encounter. Poly(I:C)+
B. thetaiotaomicron
: n = 10.
(D) Sensorimotor gating in the PPI assay, as measured by percent difference between the startle response to pulse only and startle response to pulse pr
eceded by
a 5 db or 15 db prepulse. Treatment effects were assessed by repeated-measures two-way ANOVA with Bonferroni post hoc test. Poly(I:C)+
B. thetaiotaomicron
:
n = 32.
For all panels, data for saline, poly(I:C) and poly(I:C)+
B. fragilis
are as in
Figure 5
. Graphs represent cumulative results obtained for 3-6 independent cohorts of
mice. Data are presented as mean
±
SEM. *p < 0.05, **p < 0.01, ***p < 0.001. S = saline+vehicle, p = poly(I:C)+vehicle, P+BF = poly(I:C)+
B. fragilis
, P+BT =
Poly(I:C)+B.
thetaiotaomicron
.
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
S9
Figure S5.
B. Fragilis
Treatment Causes Statistically Significant Alterations in Serum Metabolites, with Widespread Changes in Bio-
chemicals Relevant to Fatty Acid Metabolism and Purine Salvage Pathways, Related to
Figure 6
Levels of 103 metabolites that are significantly altered in sera of
B. fragilis
-treated MIA offspring compared to saline controls, as measured by GC/LC-MS. Colors
indicate fold change relative to metabolite concentrations detected in saline offspring, where red hues represent increased levels compared to con
trols and green
hues represent decreased levels compared to controls (see legend on top left). All changes indicated are p < 0.05 by two-way ANOVA with contrasts. p = po
ly(I:C),
P+BF = poly(I:C)+
B. fragilis
. n = 8/group.
S10
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
Figure S6. Synthesis of Autism-Associated Metabolites by Host-Microbe Interactions, Related to
Figure 6
(A) Diagram illustrating the synthesis of 4EPS (found elevated in MIA serum and restored by
B. fragilis
treatment) and
p
-cresol (reported to be elevated in urine of
individuals with ASD) by microbial tyrosine metabolism and host sulfation.
(B) Diagram illustrating the synthesis of indolepyruvate (found elevated in MIA serum and restored by
B. fragilis
treatment) and indole-3-acryloylglycine (reported
to be elevated in urine of individuals with ASD) from microbial tryptophan metabolism and host glycine conjugation.
Solid arrows represent known biological conversions. Dotted arrow represents predicted biological conversions.
Cell
155
, 1–13, December 19, 2013
ª
2013 Elsevier Inc.
S11