Article

A Long Noncoding RNA lincRNA-EPS Acts as a

Transcriptional Brake to Restrain Inflammation
Graphical Abstract Authors
Maninjay K. Atianand, Wenqian Hu,
Ansuman T. Satpathy, ...,
Harvey F. Lodish, Daniel R. Caffrey,
Katherine A. Fitzgerald

Correspondence
kate.fitzgerald@umassmed.edu

In Brief
A long noncoding RNA restrains
uncontrolled inflammatory responses in
immune cells and in mice.

Highlights Accession Number

d lincRNA-EPS is downregulated in macrophages exposed to E-MTAB-4088
microbial ligands GSE78873

d lincRNA-EPS represses the expression of immune response

genes (IRGs)

d lincRNA-EPS controls nucleosome positioning and inhibits

transcription of IRGs

d lincRNA-EPS-deficient mice manifest enhanced

inflammation in vivo

Atianand et al., 2016, Cell 165, 1672–1685

June 16, 2016 ª 2016 Elsevier Inc.
http://dx.doi.org/10.1016/j.cell.2016.05.075
 Article

A Long Noncoding RNA lincRNA-EPS

Acts as a Transcriptional
Brake to Restrain Inflammation
Maninjay K. Atianand,1 Wenqian Hu,2,6 Ansuman T. Satpathy,3 Ying Shen,3 Emiliano P. Ricci,4,7
Juan R. Alvarez-Dominguez,2 Ankit Bhatta,1 Stefan A. Schattgen,1 Jason D. McGowan,1 Juliana Blin,1 Joerg E. Braun,4
Pallavi Gandhi,1 Melissa J. Moore,4 Howard Y. Chang,3 Harvey F. Lodish,2 Daniel R. Caffrey,1
and Katherine A. Fitzgerald1,5,*
1Program in Innate Immunity, University of Massachusetts Medical School, Worcester, MA 01605, USA
2Whitehead Institute, Massachusetts Institute of Technology (MIT), Cambridge, MA 02142, USA
3Center for Personal Dynamic Regulomes and Program in Epithelial Biology, Stanford University School of Medicine, Stanford,

CA 94305, USA
4Howard Hughes Medical Institute and Department of Biochemistry and Pharmacology, University of Massachusetts Medical School,

Worcester, MA 01655, USA

5Centre for Molecular Inflammation Research, Department of Cancer Research and Molecular Medicine, NTNU, 7491 Trondheim, Norway
6Present address: Department of Biochemistry and Molecular Biology, Mayo Clinic, Rochester, MN 55905, USA
7Present address: Centre International de Recherche en Infectiologie, Ecole Normale Supérieure de Lyon, Université de Lyon, 69364 Lyon,

France
*Correspondence: kate.fitzgerald@umassmed.edu
http://dx.doi.org/10.1016/j.cell.2016.05.075

SUMMARY for a better understanding of the biological processes that

Long intergenic noncoding RNAs (lincRNAs) are
important regulators of gene expression. Although
lincRNAs are expressed in immune cells, their func-
tions in immunity are largely unexplored. Here, we
identify an immunoregulatory lincRNA, lincRNA-
EPS, that is precisely regulated in macrophages to
control the expression of immune response genes
(IRGs). Transcriptome analysis of macrophages
from lincRNA-EPS-deficient mice, combined with
gain-of-function and rescue experiments, revealed
a specific role for this lincRNA in restraining
IRG expression. Consistently, lincRNA-EPS-defi-
cient mice manifest enhanced inflammation and
lethality following endotoxin challenge in vivo.
lincRNA-EPS localizes at regulatory regions of IRGs
to control nucleosome positioning and repress tran-
scription. Further, lincRNA-EPS mediates these
effects by interacting with heterogeneous nuclear
ribonucleoprotein L via a CANACA motif located
in its 30 end. Together, these findings identify
lincRNA-EPS as a repressor of inflammatory re-
sponses, highlighting the importance of lincRNAs in
the immune system.
INTRODUCTION
Cell-type-specific regulatory circuits control gene expression in
complex, dynamic, and temporally regulated manners (Smale
et al., 2013). Understanding these regulatory networks is critical
contribute to human health and disease. The innate immune sys-
tem represents a first line of defense against microbial infection
and relies on dynamic transcriptional changes, initiated when
germline-encoded pattern-recognition receptors (e.g., Toll-like
receptors [TLRs]) detect microbial products (Janeway and
Medzhitov, 2002; Medzhitov and Horng, 2009). These receptors
trigger signaling cascades that converge on well-defined tran-
scription factors including NF-kB and interferon regulatory
factors (e.g., IRF3), which induce transcription of hundreds of im-
mune response genes (IRGs). As sustained expression of IRGs
can lead to tissue damage and immune pathology, their expres-
sion must be carefully regulated (Chen and Nuñez, 2010). Both
transcriptional and post-transcriptional regulatory checkpoints
ensure that the magnitude and duration of these events are rigor-
ously controlled and that these pathways are turned off in a
timely manner.
A large proportion of the human and mouse genome is tran-
scribed as noncoding RNAs (ncRNAs) (Derrien et al., 2012;
Djebali et al., 2012). Among these ncRNAs, microRNAs are
well-known regulators of gene expression. In addition, long
noncoding RNAs (lncRNAs) also regulate gene expression in
diverse biological contexts (Morris and Mattick, 2014). To distin-
guish them from small noncoding RNAs, lncRNAs are arbitrarily
defined as having R200 nucleotides. These RNAs can be either
intergenic (between protein coding genes; long intergenic non-
coding RNA [lincRNA]), intronic, natural antisense transcripts
(NATs), or transcribed from divergent enhancers and promoters
(Ulitsky and Bartel, 2013). By binding to chromatin-modifying
factors, heterogeneous nuclear ribonucleoproteins (hnRNPs),
or transcription factors, lncRNAs control gene transcription.
Also, lncRNAs act via post-transcriptional mechanisms targeting
the splicing, stability, or translation of host mRNAs (Guttman and
Rinn, 2012). Although lncRNAs have been identified in virtually all

1672 Cell 165, 1672–1685, June 16, 2016 ª 2016 Elsevier Inc.
immune cells, their functions in these cells are only beginning to
emerge (Atianand and Fitzgerald, 2014; Guttman et al., 2009).
For example, lincRNA-Cox2 was identified as a dynamically
regulated gene induced by TLR ligands and that, in turn, acts
to both promote and repress inflammatory gene expression
(Carpenter et al., 2013). Several additional lncRNAs including
THRIL (Li et al., 2014), lnc13 (Castellanos-Rubio et al., 2016),
and an antisense lncRNA, AS-IL-1a (Chan et al., 2015), also
regulate inflammatory gene expression in myeloid cells. In
T cells, NeST regulates IFN-g gene transcription and persistent
infection with Theiler’s virus (Gomez et al., 2013), while the
lncRNA Rmrp regulates effector functions of T-helper 17 cells
(Huang et al., 2015).
In this study, we define lincRNA-EPS (erythroid prosurvival;
also known as Ttc39aos1) as an important transcriptional brake
that curbs inflammatory gene expression in macrophages and in
mice. Macrophages and dendritic cells express lincRNA-EPS,
but it is downregulated in cells stimulated with microbial ligands,
including lipopolysaccharide (LPS), which signals via TLR4/
MD2. Genome-wide transcriptional analysis of macrophages
from wild-type (WT) and lincRNA-EPS/ mice revealed that
lincRNA-EPS specifically repressed the expression of IRGs in
both resting and TLR4-activated cells. Consistently, gain-of-
function and rescue studies further defined lincRNA-EPS as a
potent repressor of IRG expression. We found that lincRNA-
EPS associates with chromatin and interacts with hnRNPL, a
member of a larger family of heterogeneous ribonucleoproteins,
to alter nucleosome positioning and repress IRGs. Finally,
lincRNA-EPS/ mice had increased susceptibility to LPS
challenge in vivo. Collectively, these results demonstrate that
lincRNA-EPS is an inhibitor of IRG expression, acting as a regu-
latory checkpoint that is downregulated prior to the inducible
expression of IRGs.
RESULTS
lincRNA-EPS Is Suppressed in Macrophages Exposed to
Microbial Ligands
Previous work from our lab utilized RNA sequencing to define the
transcriptome of TLR2-activated bone-marrow-derived macro-
phages (BMDMs) and identified lincRNA-Cox2, a highly induc-
ible lincRNA that, in turn, regulated IRG expression (Carpenter
et al., 2013). This approach also identified lncRNAs that
were downregulated in macrophages stimulated through TLR2
(Figure 1A). Among these was an annotated lincRNA, lincRNA-
EPS, originally identified as a regulator of erythrocyte differenti-
ation (Hu et al., 2011). As lincRNA-EPS was highly expressed
in resting macrophages and downregulated following TLR liga-
tion, we hypothesized that lincRNA-EPS might regulate IRG
expression.
We employed qRT-PCR to examine the kinetics of lincRNA-
EPS expression in BMDMs exposed to TLR ligands including
Pam3CSK4 (TLR2/1), LPS (TLR4), and polyinosinic-polycytydilic
acid (polyI:C; TLR3). The expression of lincRNA-EPS was down-
regulated in response to all three TLR ligands in a time-
dependent manner (Figure 1B). TLR4-mediated suppression of
lincRNA-EPS was also observed in bone-marrow-derived den-
dritic cells (BMDCs) (Figure 1C). Notably, lincRNA-EPS suppres-
sion in LPS-stimulated BMDMs was dose- and time-dependent
and inversely correlated with levels of the proinflammatory
gene IL6 (Figures S1A and S1B). TLR-dependent suppression
of lincRNA-EPS was dependent on MyD88 and Trif since
MyD88//Trif / BMDMs had a reduced ability to downregu-
late lincRNA-EPS (Figure 1D). To evaluate the role of NF-kB, a
well characterized regulator of IRGs, in controlling the suppres-
sion of lincRNA-EPS, BMDMs were pretreated with DMSO (con-
trol) or BAY11-7082, an irreversible inhibitor of the IKK kinases,
followed by LPS stimulation for 6 hr. The LPS (TLR4)-mediated
suppression of lincRNA-EPS was impaired in BAY11-7082-
treated BMDMs relative to the control cells (Figure 1E). Induc-
tion of IL-1a mRNA, a known target of the NF-kB pathway,
was impaired in these cells (Figure 1E). Consistent with
these findings, LPS-induced suppression of lincRNA-EPS was
also impaired in IKKb-deficient BMDMs (Figure 1F). We also
observed a similar suppression of lincRNA-EPS expression in
BMDMs infected with Sendai virus or Listeria monocytogenes
(Figure 1G) and in cells exposed to TNFa or type I interferon
(IFN) (Figure 1H). However, stimulation of BMDMs with the
anti-inflammatory cytokine IL-10, did not alter lincRNA-EPS
expression or interfere with TLR4-mediated suppression of
lincRNA-EPS (Figure S1C). We also found that macrophages
express 11 lincRNA-EPS molecules per cell (Figure S1D).
Collectively, these results indicate that lincRNA-EPS levels in
macrophages are dynamically regulated in response to microbial
and inflammatory triggers.
Genetic Deletion of lincRNA-EPS Leads to Enhanced
Basal and TLR4-Induced Expression of IRGs in
Macrophages
To test our hypothesis that lincRNA-EPS regulates IRGs, we
generated mice lacking lincRNA-EPS (Figures S2A–S2C) by
deleting the entire 4 kb genomic locus harboring lincRNA-EPS
and replacing it with a neomycin resistance cassette. The
lincRNA-EPS/ animals were healthy and reproduced at ex-
pected Mendelian frequencies with no gender bias and did
not manifest any gross developmental defects (data not
shown). Although this lincRNA had been implicated in erythroid
differentiation (Hu et al., 2011), features of erythropoiesis
including hematocrit and hemoglobin levels and red blood cell
(RBC) numbers were all normal in lincRNA-EPS/ mice (Fig-
ure S2D). In addition the numbers of macrophages, dendritic
cells (DCs), B cells, T cells, and natural killer (NK) cells were
comparable in spleens of lincRNA-EPS/ and WT animals
(Figure S2E).
To assess the impact of lincRNA-EPS deficiency on IRG
expression in macrophages, we performed unbiased transcrip-
tome profiling using RNA-seq in WT and lincRNA-EPS/
BMDMs stimulated with LPS for 2 and 6 hr. The absence of
lincRNA-EPS altered the expression of 113, 197, and 290 genes
at 0, 2, and 6 hr post-LPS stimulation, respectively (R2-fold
change, Q-value < 0.05) (Figure 2A; Table S1). The majority of
these genes were upregulated in lincRNA-EPS/ BMDMs rela-
tive to WT BMDMs. Furthermore, mRNA levels of these genes
were increased in both resting and LPS-treated conditions.
A gene ontology (GO) enrichment analysis demonstrated that
IRGs were significantly overrepresented in these differentially
Cell 165, 1672–1685, June 16, 2016 1673
 28 11198 8.2
Rik

Gm1259 5.1 ik
A B C

5 R
BMDMs BMDCs

Gm152885C1
linc 1563 M16

Gm23−2 EPS

Gm1182 D09
150

RP RNA 5

AC 00082

14 0
3
100 100 100

45
Gm 000

lincRNA-EPS (%)

lincRNA-EPS (%)
3

1
A2
100
0.0
50
Fold change (log2)

50 50
-0.5
50
-1.0

-1.5 0 0 0 0
time (hr): 0 1 2 6 0 1 2 6 0 1 2 6 time (hr): 0 2 6
-2.0
Pam3CSK4 LPS poly(I:C) LPS
-2.5 (TLR2) (TLR4) (TLR3) (TLR4)

D NT E DMSO F WT
LPS 150
150 BAY 11-7082 100 120 IKKβ-KO

IL-1α mRNA (fold change)

Pam3CSK4
poly(I:C) *
lincRNA-EPS (%)

lincRNA-EPS (%)
100

lincRNA-EPS (%)
80
100 100 80
** 60
60
40 **
50 50 40
20 20
0 0 0 0
WT Trif-KO MyD88/Trif NT LPS NT LPS NT LPS
DKO

G 1 H 1
BMDMs BMDMs
(log10 fold change)

(log10 fold change)

lincRNA-EPS

lincRNA-EPS

0
0
-1
-1
-2

-2 -3
time (hr): 0 1 2 6 1 2 6 time (hr): 0 2 6 2 6

ne
s us Fα N-
I
ge vir TN IF
to dai
y n
oc Se
on
m
L.

Figure 1. Suppression of lincRNA-EPS in Macrophages Exposed to TLR Ligands

(A) Top ten lncRNAs downregulated in TLR2-activated BMDMs. Expression levels (log2-FC) are from the RNA-seq dataset from BMDMs stimulated with
Pam3CSK4 for 5 hr (Carpenter et al., 2013).
(B and C) qRT-PCR analysis of lincRNA-EPS expression in BMDMs (B) and DCs (C) stimulated with TLR ligands.
(D) qRT-PCR analysis of lincRNA-EPS expression in WT, Trif-KO, or MyD88/Trif-DKO BMDMs stimulated with TLR ligands for 6 hr.
(E) WT BMDMs were pretreated with BAY11-7082 (10 mM) or DMSO, followed by LPS stimulation for 6 hr. Expression of lincRNA-EPS (left) and IL-1a (right) was
quantified by qRT-PCR. Data are shown relative to DMSO-treated cells. *p < 0.05; **p < 0.01.
(F) lincRNA-EPS expression in WT and IKKb-deficient BMDMs that were treated or not treated with LPS for 6 hr. **p < 0.01.
(G and H) lincRNA-EPS expression in BMDMs exposed to microbial infections (G) or proinflammatory cytokines (H) for indicated times.
Error bars indicate SD. See also Figure S1.

expressed genes (Figure 2B). The time course indicated that chromosome 5 that included Cxcl10, Cxcl9, Plac8, and several
lincRNA-EPS regulated the expression of many IRGs in a tempo- GBP-family members was upregulated in lincRNA-EPS/
ral manner, since their RNA levels were often altered at specific BMDMs. We further confirmed the upregulation of the chro-
times following LPS treatment (Figure 2C). Among the genes that mosome-5 gene cluster (Figure 2F) and others (Figure S2H) using
were most significantly increased were cytokines and chemo- qRT-PCR in both untreated and LPS-stimulated (2 and 6 hr) WT
kines (Cxcl10, Cxcl9, Tnfsf10, Tnfsf8, and IL-27), as well as inter- and lincRNA-EPS/ BMDMs. Moreover, we found increased
feron-stimulated genes (ISGs) including Ifit2, Rsad2 (Viperin), levels of Cxcl10, Ccl5, and IL6 responses in lincRNA-EPS/
Oasl1, and multiple members of the guanylate-binding protein BMDMs at both the mRNA (Figure 2G) and protein levels (Fig-
(GBP) family (Figures 2D and S2G). Genes that were differentially ure 2H). Notably, lincRNA-EPS/ BMDMs did not display any
expressed between WT and lincRNA-EPS/ BMDMs were changes in the expression of lincRNA-EPS neighboring genes
often IRGs that were clustered within specific chromosomal Eps15 and Ttc39a relative to WT cells (Figure S2I). Furthermore,
locations (Figure 2E). In particular, a large cluster of IRGs on these cells also proliferated normally (Figure S2J). Together,

1674 Cell 165, 1672–1685, June 16, 2016

A B C D RNA-seq
Top 5 GO functional categories Immune Pathway Genes (n) Fold change (KO/WT)
genes (n) dysregulated in KO

(LPS 6 hr, ≥ 2-FC in KO; Q-value < 0.05)

(≥ 2-FC; Q-value < 0.05)

250 dysregulated in KO LPS (hr): 0 2 6

up Ifit2
200 down immune response 6 hr BC147527
Ly6c2
150 immune system process 76
76 Oasl1
Cmpk2
100 defense response Cxcl10
7 21
21
defense response Rsad2
50 to other organism 7 Trim30c
23
23 25
25 Ttc39a
0 innate immune response Gbp4
0 hr 9 2 hr
0 2 6 Cd69
0 10 20 30 Tnfsf10
hours post-LPS −log10 (P−value) Iigp1
Gm14446
Plac8
Pydc4 Ly6c1
E Pyhin1
F Ly6i
Ifit1 Pydc3 Untreated Ly6a
Tnfα Immune genes Tnfsf8
Hspa1l Ms4a6b Ifit3 Ifi204 Mnda
Other genes LPS 2h Cst7
Hspa1a Ms4a4b Ifit2 Mnda1 Ifi203
LPS 6h
Ifi205 128 qRT-PCR Nos2

Fold change (KO/WT)

Hspa1b Ms4a4c X 1 Il27
Tap2 19 2 64
18 6 hr Gbp5 Themis2
Tap1 Mx2 Gbp3 32 Lhx2
17

Mx1
Ly6c2 Gbp2 Aldh1b1
2 hr 16
3

ENSMUSG00000041481
6

Ly6c1
15 1

Tnfsf15 8 Fam26f
Ly6a Tnfsf8 Csf1
4

Ly6i 4
0 hr Tnc P2ry14
Phf11c 2 Slamf7
14

Phf11d 1 Ccrl2
Cxcl10, Cxcl9
13

Phf11b Apol9a
Plac8, Gbp4
l9

10

11
Phf11a Apol9b
ac

bp

bp

bp

bp
6

xc

bp
bp
Gbp6, Gbp8
Pl
12

G
C

Cxcl9

G
G
11 7 Clec4a2 Gbp9, Gbp10
Rgs1
10 8 Clec4n Gbp11 chr5: 92,321,331 - 105,346,472 (mm10) Plekha4
Ccl5 Ccl3 Ccl4 9
Clec4e Slc25a22
Ccl2 Ccl7 Ccl12
Gbp5
Tnfsf4
G WT H WT
Scimp
Ptx3
KO KO 1110059E24Rik
4 15 15 40 6 1.2
(fold change, x1000)

** Clic5
(fold change, x100)

(fold change, x100)

** ** ENSMUSG00000062496
** *** ** 5 1.0
Cxcl10 mRNA

Cxcl10 (ng/ml)
Ccl5 mRNA

3 Phlpp1
Ccl5 (ng/ml)

30
IL6 mRNA

10 10 4 IL6 (ng/ml) 0.8 Trim25

Mdm2
2 20 3 0.6 Kdr
5 5 2 0.4 St3gal5
1 10 Tpx2
*** *** ** 1 0.2
nd log2-fold change
0 0 0 0 0 0
NT LPS NT LPS NT LPS NT LPS NT LPS NT LPS
-3 -2 -1 0 1 2 3

Figure 2. Genetic Deficiency of lincRNA-EPS Leads to Elevated Levels of Basal and TLR4-Induced Expression of IRGs
(A) Numbers of differentially expressed genes (R2-FC over WT; Q-value < 0.05) in lincRNA-EPS/ BMDMs relative to WT cells in resting (0 hr) and LPS-treated
cells (2 and 6 hr). Data are from RNA-seq performed in biological duplicates.
(B) Gene ontology (GO) analysis of differentially expressed genes (R2-FC over WT; Q-value < 0.05 in RNA-seq) between lincRNA-EPS/ BMDMs and WT
BMDMs following LPS treatment for 6 hr. The top five most significantly enriched GO terms (log10 P-values for biological processes) in differentially expressed
genes relative to all other genes in the genome (background model) are shown.
(C) Venn diagram represents the number of immune genes (defined in Supplemental Experimental Procedures) that were differentially expressed (R2-FC over
WT; Q-value < 0.05 in RNA-seq) in lincRNA-EPS/ BMDMs.
(D) Heatmap of top 50 upregulated genes in lincRNA-EPS/ BMDMs relative to WT cells at 6 hr after LPS treatment. log2-FC values were calculated from
RNA-seq (FPKM+1), and are equivalent to lincRNA-EPS KO/WT. A subset of common IRGs is highlighted in red.
(E) Circos plot of differentially expressed genes (R2-FC over WT in RNA-seq) for untreated cells (inner track) and LPS-stimulated cells (middle track, 2 hr; outer
track, 6 hr). Immune genes (defined in Supplemental Experimental Procedures) are highlighted in red, and all other genes are in blue.
(F) qRT-PCR analysis of lincRNA-EPS regulated IRGs in untreated and LPS-treated WT and lincRNA-EPS/ BMDMs.
(G) mRNA levels of cytokine genes in WT and lincRNA-EPS/ BMDMs stimulated with LPS for 6 hr. **p < 0.01; ***p < 0.001.
(H) Protein levels of indicated cytokines in WT and lincRNA-EPS/ BMDMs stimulated with LPS for 12 hr and analyzed by ELISA. **p < 0.01; nd, non-detectable.
Error bars indicate SD. See also Figure S2 and Table S1.

these results indicate that lincRNA-EPS contributes in a specific number of IRGs including cytokines (IL6, IL15, IL1a), chemokines
manner to the temporal regulation of IRGs. (Cxcl10, Cxcl2, Ccl5, Ccl4), and antiviral ISGs (Ifit1, Oas2, Ifi204,
and Rsad2/viperin) were severely impaired in lincRNA-EPS-ex-
Gain-of-Function and Rescue Studies Define lincRNA- pressing BMDMs compared to control cells (Figure 3B). We also
EPS as a Negative Regulator of Inflammatory Responses confirmed these effects for some of these genes using qRT-PCR
in Macrophages (Figure S3A). Also, lincRNA-EPS-mediated suppression of IRGs
We next conducted ‘‘gain-of-function’’ studies by generating was observed in cells stimulated with double-stranded DNA (Fig-
macrophages that ectopically expressed lincRNA-EPS or vector ure S3B). Ectopic expression of lincRNA-EPS also impaired LPS-
control (EV Ctl) (Figure 3A). We stimulated these cells with LPS induced expression of IRGs in J774.1 macrophages (Figures S3C
and measured the mRNA levels of IRGs by NanoString technol- and S3D). These results collectively indicate that ectopic expres-
ogy. These studies revealed that LPS-induced expression of a sion of lincRNA-EPS blocks IRG expression in macrophages.

Cell 165, 1672–1685, June 16, 2016 1675

A C D Cxcl10 Ccl5 Rsad2 IL6
iBMDM 0.003 iBMDM
lincRNA-EPS (fold change)

(lincRNA-EPS/gapdh)
15 300 ** ** 200 * * 400 ** ** 2000
RNA expression
** **

mRNA fold change

0.002 150 300 1500
10 200 WT + EV Ctl
100 200 1000 KO + EV Ctl
0.001 KO + lincRNA-EPS
5 100
50 100 500
0.000 ND
WT + EV Ctl

KO + EV Ctl

KO + lincRNA-EPS
0 0 0 0 0
tl S NT LPS NT LPS NT LPS NT LPS
C
EP
EV A-
N

S
R
nc

EP
li E F
lincRNA-EPS

lincRNA-EPS

EV Ctl

A-
2000

N
l
Ct
B lincRNA-EPS

cR
EV

lin
EV Ctl

EV Ctl

IL-1β (pg/ml)
1500 pro-IL-1β

total cell lysate

ASC
LPS - - + + 1000
aim2 pro-caspase-1
lgp2 Nlrp3
ccl5 500
cd40 β-actin
cxcl10 0
cxcl2
irf1
S

in

li
co
LP

ric
IL6
E.
H EV Ctl
ge

IL1α
ni

ifi204 lincRNA-EPS
+

ifit1 800 lincRNA-EPS + pMSCV neo

S

ccl4
LP

a20 lincRNA-EPS + pMSCV ASC

IL-1β (pg/ml)
cd80 G 600
socs3 lincRNA-EPS
icam1
prdm1
+ ASC

400
EV Ctl

+ neo

socs1
IL15
nlrc5 n.s. 200
rsad2
irf7 ASC
oas2 0
relative expression LPS LPS + nigericin
Row min Row max

Figure 3. Gain-of-Function and Rescue Studies Demonstrate that lincRNA-EPS Acts as a Repressor of IRG Expression
(A) qRT-PCR analysis of lincRNA-EPS levels in iBMDMs expressing ectopic lincRNA-EPS or empty vector control (EV Ctl).
(B) Heatmap of gene expression in iBMDMs expressing ectopic lincRNA-EPS or EV Ctl that were treated or not treated with LPS for 6 hr and analyzed by
NanoString. Data are shown for 23 genes (out of 94 genes) that were suppressed by at least 2-fold in cells expressing ectopic lincRNA-EPS.
(C and D) Rescue of lincRNA-EPS function in lincRNA-EPS/ macrophages through ectopic expression. lincRNA-EPS levels in WT iBMDMs or lincRNA-EPS/
iBMDMs expressing ectopic lincRNA-EPS or EV Ctl were analyzed by qRT-PCR (C). These cells were stimulated with LPS for 6 hr and analyzed by qRT-PCR to
measure lincRNA-EPS-regulated IRGs (D). *p < 0.05; **p < 0.01.
(E and F) iBMDMs expressing ectopic lincRNA-EPS or EV Ctl were stimulated as indicated or infected with E. coli (10 MOI). Levels of secreted IL-1b were
quantified by ELISA (E) and cell lysates analyzed by western blot (F).
(G and H) Addback of ASC expression in iBMDMs expressing ectopic lincRNA-EPS was confirmed by immunoblotting ASC (G) and IL-1b levels measured by
ELISA (H). n.s., non-specific.
Error bars indicate SD. See also Figure S3.

We next used retroviral transduction to restore expression of caspase-1-activating complex that controls the maturation of
the mature RNA transcript in lincRNA-EPS/-derived iBMDMs. interleukin-1b (Lamkanfi and Dixit, 2012). The best-studied
In these cells, lincRNA-EPS expression was restored to levels NLRP3 inflammasome is activated in response to microbial in-
comparable to that seen in WT cells (Figure 3C). qRT-PCR anal- fections, environmental toxins, and endogenous danger signals
ysis of Cxcl10, Ccl5, Rsad2/Viperin, and IL6 expression revealed (Lamkanfi and Dixit, 2012). We therefore tested if activation of
that the LPS-induced levels of these genes, as expected, were the NLRP3 inflammasome was altered in cells expressing
increased in lincRNA-EPS/ iBMDMs (immortalized BMDMs) lincRNA-EPS. Control iBMDMs and those ectopically expressing
as compared to WT cells (Figure 3D). However, ectopic expres- lincRNA-EPS were exposed to LPS alone or LPS together with
sion of lincRNA-EPS in these knockout (KO) cells reduced the nigericin, a pore-forming toxin, or E. coli, which both activate
mRNA levels of these genes back to that observed in WT cells Nlrp3 (Figure 3E). In both cases, production of IL-1b was severely
(Figure 3D). Similar results were obtained for additional IRGs impaired in lincRNA-EPS-overexpressing macrophages com-
(Figure S3E). pared to control cells (Figure 3E). We subsequently compared
In fetal erythroblasts, lincRNA-EPS was shown to regulate the the levels of inflammasome components and substrates by
expression of pro-apoptotic genes (Hu et al., 2011). One of the immunoblotting in WT and lincRNA-EPS-overexpressing cells
proposed target genes, Pycard, encodes ASC (apoptosis-asso- (Figure 3F). While the levels of pro-IL-1b, pro-caspase-1, and
ciated speck-like protein containing CARD), a critical adaptor Nlrp3 were all equivalent in both cell types, the levels of ASC
protein required for the assembly of the inflammasome, a were greatly diminished (Figure 3F). Notably, genetic rescue of

1676 Cell 165, 1672–1685, June 16, 2016

A B C D
Cytosol RNA FISH (Resting BMDMs)
BMDMs BMDMs
Nucleus lincRNA-EPS GFP
120 DNA DNA 100 ** 150

Cells with lincRNA-EPS

**
Relative RNA levels

lincRNA-EPS foci
100 80

(% Total signals)
80 100

foci (%)
60
(% Total)

60
40
40 lincRNA-EPS lincRNA-EPS 50
GFP GFP
DNA DNA 20
20
DIC
0 0 0

ol

us

S
N S
β- T1

β
tin

te
LP
-1

os
EP

le
ac
EA

a
IL

yt

uc

re
A-

nt
N
N

U
cR
lin

E F Chromatin-rich fractions G
Macrophage R2 R1
Fractions: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 5’ 3’
Cross-link Histone H3 -

Retrieved RNA (% Input)

100
lysis
Retrieved RNA (% Input)

22 F4 F13 F15 80
Nuclear Pellet
17 IgG RIP
lysis 12 60 Histone H3 RIP
DNase I fragmentation 7
2 40
Sucrose-gradient 1.0
20
Free RNA 5% 0.5
0
chromatin-rich 0 nd nd

1)

2)

T1
(R

(R

EA
fractions
S

T1

K
β
in

N
U
-1
EP

JP

7S
ct

EP

EP
EA

30%
IL
a
A-

β-

A-

A-
N
N

N
R

cR

cR
nc

lin

lin
Ili

Figure 4. lincRNA-EPS Is Associated with Chromatin in Resting Macrophages

(A) qRT-PCR analysis of RNAs purified from nuclear (red) and cytosolic (black) compartments in BMDMs.
(B) Single molecule RNA FISH detecting endogenous lincRNA-EPS molecules (red) in resting BMDMs. DNA (blue) was stained with DAPI and autofluorescence
detected by staining with probes against GFP (green). A representative image (1003 magnification) is shown. DIC, differential interference contrast.
(C) Quantification of lincRNA-EPS foci detected by RNA FISH in resting BMDMs. Results are shown as % of cells that showed either a nuclear or cytosolic
lincRNA-EPS foci in >500 randomly selected cells. **p < 0.01.
(D) Quantification of cells with lincRNA-EPS foci (nuclear or cytosolic) detected by RNA FISH in BMDMs in response to LPS stimulation for 6 hr. **p < 0.01.
(E) Schematic for the isolation of chromatin-associated RNAs following cross-linking (formaldehyde and glutaryldehyde).
(F) Western blotting of histone H3 (top) in nuclear fractions isolated by sucrose-gradient fractionation. Purified RNA was reverse-transcribed using oligo-dT primer
and analyzed by qPCR (bottom).
(G) Histone H3 RIP followed by qRT-PCR analysis of co-purified RNAs in formaldehyde cross-linked BMDMs.

ASC in these cells fully restored IL-1b responses to levels com- We next determined whether lincRNA-EPS was present in free
parable to that of control cells (Figures 3G and 3H). nucleoplasmic or chromatin-associated fractions. BMDMs were
chemically cross-linked to preserve endogenous chromatin-
lincRNA-EPS Is Chromatin Associated in Resting RNA complexes, and the nuclear lysates subjected to sucrose-
Macrophages gradient fractionation following sonication/DNase I treatment
To understand how lincRNA-EPS regulates IRGs, we examined (Figure 4E). We immunoblotted histone H3 and identified chro-
its localization by performing sub-cellular fractionation and matin-rich fractions 4–15 (Figure 4F, top). We purified RNA
analyzed levels of lincRNA-EPS by qRT-PCR. This analysis re- from three randomly selected chromatin rich-fractions (F4,
vealed that lincRNA-EPS was predominantly nuclear in resting F13, and F15) and performed reverse-transcription using oligo-
BMDMs (Figure 4A). As expected, the mature b-actin and IL-1b dT primers to specifically detect mature full-length transcripts
transcripts were localized to the cytosol, while NEAT1 (a nuclear and avoid detection of nascent (pre-mRNA) transcripts. Analysis
lincRNA) was confined to the nucleus. The nuclear localization of by qRT-PCR revealed that lincRNA-EPS was detected across all
lincRNA-EPS was also confirmed using single molecule RNA three of these chromatin-rich fractions (Figure 4F, bottom).
fluorescent in situ hybridization (FISH) in resting primary BMDMs Importantly, b-actin and IL-1b were undetectable in these
(Figure 4B). In these cells, the majority of detectable lincRNA-EPS samples, while known chromatin associated lncRNAs (NEAT1
foci (>80%) were nuclear (Figure 4C). Furthermore, the lincRNA- and JPX) and snRNAs (U1 and 7SK) were present. Further, his-
EPS FISH signal was reduced in cells exposed to LPS (Figure 4D). tone H3 RNA immunoprecipitation (RIP) followed by qRT-PCR

Cell 165, 1672–1685, June 16, 2016 1677

analysis of co-purified RNA confirmed that the majority of mature
lincRNA-EPS was associated with chromatin (Figure 4G).
lincRNA-EPS Maintains a Repressed Chromatin State
at IRGs
We next explored the possibility that lincRNA-EPS controls the
expression of IRGs by regulating their transcription. We per-
formed chromatin immunoprecipitation (ChIP) followed by
qPCR to monitor the recruitment of RNA polymerase II (Pol II)
to regulatory regions of IRGs that were upregulated in the
absence of lincRNA-EPS. Pol II was recruited near the transcrip-
tion start site (TSS) of Cxcl10, Ccl5, and Rsad2/Viperin in TLR4-
activated cells (Figure 5A). However, these responses were
impaired in BMDMs ectopically expressing lincRNA-EPS. We
also observed similar effects at the TSS of IL6, Irf7, and ASC
genes (Figure S4A). Consistent with these observations, the
inducible recruitment of Pol II at Cxcl10, Ccl5, and Rsad2/Viperin
genes in LPS-treated cells was enhanced in lincRNA-EPS/
BMDMs compared to WT cells (Figure 5B). These results indi-
cate that lincRNA-EPS regulates the expression of these IRGs
at the level of transcription.
Based on these findings, we hypothesized that lincRNA-EPS
may contribute to epigenetic silencing of IRGs to restrain
their basal expression. We performed ChIP-qPCR to quantify
H3K4me3, a mark of active transcription near the TSS of
genes that were transcriptionally suppressed by lincRNA-EPS.
H3K4me3 levels were increased in both resting and LPS-stimu-
lated BMDMs that lacked lincRNA-EPS (Figure 5C). Notably, in
resting cells, the H3K4me3 levels at the Cxcl10, Ccl5, Rsad2/
Viperin, IL6, Irg1, and Ifit1 TSSs were increased in lincRNA-
EPS/ BMDMs compared to WT cells (Figure 5C). Apparently,
lincRNA-EPS/ BMDMs displayed near-saturating levels of
H3K4me3 marks since these levels were comparable to those
detected upon LPS stimulation of WT cells. Similar patterns of
elevated H3K4me3 marks were also observed for Ifit2, Gbp4,
and Cxcl9 (Figure S4B). We also performed ChIP to monitor
Pol II Ser2P, a mark of active/ongoing transcription. In unstimu-
lated conditions, Pol II Ser2P levels were uniformly low but com-
parable between WT and lincRNA-EPS/ BMDMs (Figure 5D).
However, following LPS stimulation, Pol II Ser2P levels at
Cxcl10, Ccl5, Rsad2/Viperin, IL6, Irg1, and Ifit1 loci were greatly
enhanced in lincRNA-EPS/ BMDMs relative to WT cells (Fig-
ure 5D). Similar results showing elevated Pol II Ser2P levels
in lincRNA-EPS/ BMDMs were also observed at additional
IRGs (Figure S4C).
To examine the possibility that lincRNA-EPS binds to regula-
tory region(s) of target genes, we performed RNA antisense pu-
rification (RAP) in nuclear extracts of cross-linked resting macro-
phages using biotinylated, lincRNA-EPS-specific antisense RNA
probes. The RNA probes specifically target lincRNA-EPS and
purify the endogenous lncRNA and its associated chromatin.
RAP followed by qRT-PCR analysis of the retrieved RNA showed
that nearly 50% of the endogenous lincRNA-EPS was specif-
ically pulled down in lincRNA-EPS-specific RAP conditions, rela-
tive to several controls that included no-probes and control RAP
targeting the Firefly luciferase gene (Figure 5E), as well as no-RT
conditions (not shown). Highly abundant b-actin and 18S rRNA
were undetectable in lincRNA-EPS RAP pull-down conditions,
1678 Cell 165, 1672–1685, June 16, 2016
further confirming the specificity of lincRNA-EPS purification.
In the same experiments, we examined Cxcl10, Ccl5, and ASC
genomic regions since these genes were particularly affected
by lincRNA-EPS in both gain-of-function and loss-of-function
studies. Purified DNA was subjected to qPCR analysis, which
showed that the Cxcl10, Ccl5, and ASC genomic regions near
their TSS were highly enriched following lincRNA-EPS RAP, rela-
tive to the control RAP (Figure 5F). These results suggest that
lincRNA-EPS binds (either directly or indirectly) to these gene
loci. We conclude from all of these approaches that the nuclear
localized lincRNA-EPS binds to regulatory regions of IRGs to
prevent their expression in the absence of activating signals.
To investigate if lincRNA-EPS repressed the expression of
IRGs by altering chromatin accessibility, we performed assays
for transposase accessible chromatin (ATAC-seq) (Buenrostro
et al., 2013). ATAC-seq in WT and lincRNA-EPS/ BMDMs
demonstrated dynamic epigenomic changes following LPS
stimulation (Figures S4D and S4E), including increased accessi-
bility at promoters and gene bodies of key immune genes such
as Irg1 and Ccl5 (Figure S4F). While global genome accessibility
was largely unaltered in lincRNA-EPS/ BMDMs relative to WT
cells, several IRGs that are normally suppressed by lincRNA-
EPS showed an increase in chromatin accessibility at their pro-
moters in lincRNA-EPS/ BMDMs at baseline (Figure 5G; Table
S2). These included genes such as Cxcl10, Irg1, and Ifit2, which
show features of increased transcriptional activity in lincRNA-
EPS-deficient macrophages (Figures 5C and 5D). This increase
in chromatin accessibility at IRGs was most pronounced in unsti-
mulated lincRNA-EPS/ BMDMs and was comparable to that
seen in WT cells after stimulation with LPS for 2 or 6 hr (Fig-
ure S4G; Table S2). We next exploited the ability of ATAC-seq
to capture nucleosome positions in order to assess whether
increased chromatin accessibility in lincRNA-EPS/ BMDMs
reflected nucleosome depletion at target gene promoters. To
this end, we mapped the nucleosome positions centered within
a 1 kb window of the TSS of genes using the recently described
NucleoATAC algorithm, which captures nucleosome fingerprints
across regulatory regions of the genome (Schep et al., 2015).
Notably, aggregate nucleosome signals across all lincRNA-
EPS-repressed IRGs (identified by RNA-seq) showed an average
re-positioning of 1 nucleosome further away from their TSSs in
lincRNA-EPS/ BMDMs (Figure 5H; Table S3), while nucleo-
some signals across all genes (background model) were compa-
rable between WT and lincRNA-EPS/ BMDMs (Figure 5H).
These analyses indicate that the nucleosome fingerprints around
the TSS of lincRNA-EPS target genes are selectively altered
in lincRNA-EPS/ cells. Interestingly, changes in nucleosome
positioning were no longer evident in lincRNA-EPS/ BMDMs
compared to WT cells after 2 or 6 hr of LPS stimulation (Fig-
ure S4H), which is analogous to the results obtained for overall
chromatin accessibility changes. These changes in nucleosome
positioning in lincRNA-EPS-deficient cells were better resolved
by examining the genomic loci of individual target IRGs,
including Cxcl10 (Figure 5I), Gpr84 (Figure 5J), Irf7, and Cxcl2
(Figures S4I and S4J). Although the overall chromatin accessi-
bility at promoters of these IRGs was largely comparable be-
tween WT and lincRNA-EPS-deficient cells, there was a clear
depletion of nucleosomes near the TSS of these genes in
A 1.5
B
Cxcl10 1.0 Ccl5 1.0 Rsad2 4 Cxcl10 0.6 Ccl5 1.2 Rsad2
RNA pol II ChIP-qPCR

RNA pol II ChIP-qPCR

0.8 0.8
3 0.9
1.0 0.4
(% Input)

(% Input)
0.6 0.6 EV Ctl WT
lincRNA-EPS 2 0.6 KO
0.4 0.4
0.5 0.2
0.2 1 0.3
0.2

0.0 0.0 0.0 0 0.0 0.0

LPS LPS LPS LPS LPS LPS

C 50 D

RNA polII Ser2P ChIP-qPCR

WT 15 WT
H3K4me3 ChIP-qPCR

40 KO KO

10
(% Input)

(% Input)
30

20
5
10

0 0 ## # #
LPS + + + + + + LPS + + + + + +
Cxcl10 Ccl5 Rsad2 IL6 Irg1 Ifit1 Cxcl10 Ccl5 Rsad2 IL6 Irg1 Ifit1

BMDMs (LPS 0 hr)

E F G

5
Up lincRNA-EPS regulated
no probes lincRNA-EPS RAP
Down genes in RNA-seq
lincRNA-EPS RAP control RAP
80 10
DNA enrichment (log2 FC)
RNA enrichment (% Input)

4
control RAP

Fold Change (KO/WT)

8
RAP RT-qPCR

60 Cxcl10
Ifit2
RAP qPCR

3
Irg1
40 4 Ccl4
Gpr84
2
2
20 Cxcl2
0
# # ## # < 0.02
0 -2
1

Cxcl10 Ccl5 ASC TSS 3’ UTR

S

A
tin
EP

N
ac

rR
A-

lincRNA-EPS Gapdh
β-

S
N

18

0
cR

regulated IRGs control

lin

0 2 4 6 8 10 12
WT promoter ATAC-seq counts (log2)
H
0.05 0.10 0.15 0.20 0.25
nucleosome signals at lincRNA-EPS

WT
regulated genes (LPS 0 hr)

KO
I Open chromatin (ATAC-seq signals)
J Open chromatin (ATAC-seq signals)
20 0.5 kb 35 0.5 kb
WT WT

0 0
20 35
KO KO
0 0
−1000 −500 0 500 1000
Cxcl10 Gpr84
Distance to TSS (bp)
WT Nucleosome position Nucleosome position
0.4

KO 0.6 0.5
nucleosome signals at
all genes (LPS 0 hr)

WT WT
0.3

0 0
0.6 0.5
0.2

KO KO
0 0
0.1

1000 0 -1000 1000 0 -1000

Distance to TSS (bp) Distance to TSS (bp)

−1000 −500 0 500 1000

Distance to TSS (bp)

Figure 5. lincRNA-EPS Controls Nucleosome Positioning and Suppresses the Transcription of IRGs
(A) ChIP-qPCR analysis of Pol II at lincRNA-EPS-regulated IRGs in macrophages expressing ectopic lincRNA-EPS or EV Ctl that were treated or not treated with
LPS for 5 hr. ChIP-purified DNA was analyzed by qPCR targeting the TSSs of indicated genes.
(B) ChIP-qPCR analysis of Pol II at lincRNA-EPS-regulated IRGs in WT and lincRNA-EPS/ (KO) BMDMs that were treated or not treated with LPS for 5 hr.
(C and D) ChIP-qPCR analysis of H3K4me3 (C) and Pol II Ser2P levels (D) in WT and lincRNA-EPS/ BMDMs that were treated or not treated with LPS for 5 hr.
(legend continued on next page)

Cell 165, 1672–1685, June 16, 2016 1679

lincRNA-EPS/ BMDMs compared to WT. These results collec-
tively indicate that lincRNA-EPS promotes nucleosome occu-
pancy at target IRG promoters to repress their transcription at
baseline conditions.
Identification of hnRNPL as a Binding Partner for
lincRNA-EPS
We next attempted to identify the protein partners of lincRNA-
EPS. To this end, we performed RNA-protein binding assays
by incubating in-vitro-transcribed biotinylated lincRNA-EPS or
it’s antisense control RNA with nuclear extracts from BMDMs.
RNA-protein complexes were captured using streptavidin
magnetic beads and resolved on SDS-PAGE, and protein bands
(50–75 kD) that were specifically enriched in lincRNA-EPS pull-
down were subjected to mass spectrometry for identification (Fig-
ure 6A). This approach identified six RNA-binding proteins that
showed R2-fold enrichment in lincRNA-EPS pull-downs relative
to antisense controls (Figure S5A). The most abundant RNA bind-
ing protein identified in these pull-downs was hnRNPL. The ability
of hnRNPL to bind lincRNA-EPS was confirmed by western
blotting (Figure 6B). We confirmed that lincRNA-EPS specifically
interacted with hnRNPL, but not with hnRNP-A2/B1 or the DNA
methyltransferase I (DNMT1) (Figure 6B). We also confirmed the
lincRNA-EPS:hnRNPL interaction in vivo by purifying endogenous
hnRNPL in macrophages and analyzing co-purified RNA by qRT-
PCR (Figures 6C and 6D). Native hnRNPL RIP in non-cross-linked
BMDMs followed by qRT-PCR analysis of co-purified RNAs
showed that lincRNA-EPS was specifically enriched in hnRNPL
immunoprecipitates (Figure 6C). Notably, in formaldehyde
(FA)-cross-linked BMDMs, lincRNA-EPS showed an even higher
enrichment in the hnRNPL RIP condition compared to the control
(Figure 6D). Enrichment of lincRNA-EPS in both native and FA-
cross-linked hnRNPL conditions were notably higher than several
other control RNAs including NEAT1 and U1 snRNA (Figures 6C
and 6D). These results indicate that lincRNA-EPS specifically
interacts with hnRNPL both in vitro and in vivo.
We next generated macrophages with knockdown of hnRNPL
using three independent shRNA lines with distinct non-overlap-
ping sequences (Figure 6E). TLR4-induced expression of
Cxcl10 mRNA and protein levels were enhanced in hnRNPL-
knockdown cells compared to control shRNA cells (Figures 6F
and 6G). Similar results were obtained for Ccl5, IL6, Rsad2,
Cxcl9, and Ifit1 mRNAs, as well as Ccl5 and IL6 protein levels
in these experiments (Figures S5B and S5C). Expression of
IL18, a gene that is not targeted by lincRNA-EPS, was compara-
ble between control and hnRNPL knockdown cells (Figure S5B).
Consistent with our findings in lincRNA-EPS KO BMDMs,
hnRNPL knockdown also led to increased H3K4me3 levels at
the promoters of Cxcl10, IL6, Irg1, and Ifit1 compared to control
cells (Figure S5D). Importantly, knockdown of hnRNPL had no
effect on the levels of either the unspliced or spliced (mature)
forms of lincRNA-EPS (Figure 6H).
We also took advantage of a series of deletion mutants of
lincRNA-EPS to map the hnRNPL-binding region. The results
from in vitro binding assays indicated that hnRNPL interacted
with the 30 -531 nucleotide region of lincRNA-EPS (Figures 6I
and 6J). This region was both necessary and sufficient to bind
hnRNPL (Figure 6J). Consistently, ectopic expression of this re-
gion in lincRNA-EPS KO cells was sufficient to suppress Cxcl10
(Figure 6K). The expression of Tnfa, which is not regulated by
lincRNA-EPS, was not affected in these studies. The relative
expression level of WT and lincRNA-EPS deletion mutants was
comparable (Figure S5E). We next assessed the role of CANACA
motifs, commonly found in hnRNPL-interacting RNAs (Ray et al.,
2013). The lincRNA-EPS sequence contains three CANACA mo-
tifs within the 30 -531 region: CA1, CA2, and CA3 (Figure 6L). To
define the contribution of individual CA motifs, we performed
mutagenesis of CA1, CA2, and CA3 and analyzed hnRNPL bind-
ing and regulation of Cxcl10 in reconstituted cells. This analysis
revealed that CA3 but neither CA1 nor CA2 disrupted lincRNA-
EPS:hnRNPL binding (Figure 6M). Consistently, ectopic expres-
sion of full-length lincRNA-EPS reduced Cxcl10, Ccl5, and Ccl4
expression, while two distinct CA3 mutants (EPS_CA3a and
EPS_CA3b) lost this activity (Figure 6N). The relative expression
of WT and lincRNA-EPS mutants was comparable in these cells
(Figure S5F). The predicted RNA secondary structure of
the lincRNA-EPS 30 -531 region indicates that the CA3 motif
spans a stem-loop region (Figure S5G). A similar prediction of
lincRNA-EPS CA3 mutant reveals that the secondary structure
around this region is vastly altered (Figure S5G). Mutations in a
stem-loop region are more likely to disrupt the secondary struc-
ture than disrupt CA1 and CA2, which reside in the loop region.
These data indicate that lincRNA-EPS interacts with hnRNPL via
a CANACA motif in its 30 -531 region and that this motif is essen-
tial for hnRNPL binding and suppression of IRGs.
lincRNA-EPS Restrains Inflammation In Vivo
As lincRNA-EPS-deficient cells exhibit upregulated expression
of IRGs under both basal and TLR4-stimulated conditions, we
hypothesized that lincRNA-EPS-deficient animals would exhibit
signs of heightened inflammation in vivo. We employed a model
of ‘‘endotoxic shock’’ induced by intraperitoneal (i.p.) injection
of E. coli LPS, characterized by high systemic levels of inflamma-
tory mediators as well as death. WT and lincRNA-EPS/ mice

(E and F) RNA antisense purification (RAP) of endogenous lincRNA-EPS in BMDMs, followed by qRT-PCR analysis of retrieved RNA (E) and qPCR analysis of
co-purified DNA (F). qPCR analysis of Cxcl10, Ccl5, and ASC genomic regions was performed using oligos directed around their TSS. #, not detected.
(G) ATAC-seq results showing fold change of ATAC-seq signals (KO/WT) versus ATAC-seq signals at WT promoters (+/ 1 kb of TSS) for lincRNA-EPS-regulated
genes identified in RNA-seq (red represents upregulated; blue represents downregulated) (Figure 2A; Table S1) and all other genes (gray dots). lincRNA-EPS
target genes showing more open chromatin are highlighted. Dashed lines represent 1.2 fold change.
(H) NucleoATAC analysis showing nucleosome signals in WT (blue) and lincRNA-EPS/ BMDMs (KO; red) at basal state. Aggregate nucleosome signals are
shown within the promoters (+/ 1 kb of TSS) of genes that are regulated by lincRNA-EPS (top) or within the promoters of all genes on a genome-wide level
(bottom). Repositioning of 1 nucleosomes away from TSS in lincRNA-EPS/ BMDMs is highlighted by an arrow.
(I and J) Genome tracks of Cxcl10 (I) and Gpr84 (J) showing chromatin accessibility (normalized ATAC-seq signal) and nucleosome positioning (NucleoATAC
signals) centered around their transcription start sites in WT and lincRNA-EPS/ BMDMs at basal conditions.
Error bars indicate SD. See also Figure S4 and Tables S2 and S3.

1680 Cell 165, 1672–1685, June 16, 2016

A B Streptavidin pulldown C in vivo hnRNPL RIP D in vivo hnRNPL RIP

A
N

S
(native BMDMs) (FA cross-linked BMDMs)

antisense RNA
R

EP

lincRNA-EPS
se

A-
20 IP 50
en
an r

N
ke

no probes
cR
tis

hnRNPL
ar

(hnRNPL RIP/ IgG)

lin

(hnRNPL RIP/ IgG)

m

40

RNA enrichment

Input
Input 15

RNA enrichment
250 -

IgG
150 -
100 - 3x 1x 30
hnRNPL
75 - 10
hnRNPL 20
50 -
5
37 - hnRNP-A2/B1 10

0 0
25 - DNMT1

N A
S S

T1

1
K

G PS
N dh
T1

1
U
N
18 -EP

7S

U
EA
20 -

EA
ap
rR

E
A-
A
N

N
cR

cR
lin

lin
E F Ctl shRNA
G Ctl shRNA
H
240
Cxcl10 mRNA (fold change)

20 125
1 1

RNA level (% Ctl shRNA)

Cxcl10 protein (ng/ml)
200
2 hnRNPL shRNA 2 hnRNPL shRNA
160 15 100
3 3
hnRNPL 120 Ctl shRNA
75
shRNA: Ctl 1 2 3 80 10 hnRNPL sh2
15
hnRNPL 50 * * hnRNPL sh3
10
5
25
β-actin 5
0 0 0
0 hr 2 hr 6 hr 0 5 25 100 hnRNPL unspliced spliced
+ LPS (100 ng/ml) + LPS (ng/ml), overnight lincRNA-EPS

I J K 250 Cxcl10 50 Tnfα

control RNA

biotinylated RNA *
Input (1%)

mRNA fold change

5’ 3’ length (nt) lincRNA-EPS 200 * 40
1 **
beads

2531 deletion mutants EV Ctl

2 2000 150 30
3 1000 1 2 3 4 5 6 7 EPS_WT
4 531 100 20 EPS_2000-2531
5 2000 EPS_1-2000
6 1000 50 10
Streptavidin pulldown
7 500
IB: hnRNPL
8 control RNA 0 0
LPS LPS

L M N Cxcl10 Ccl5 Ccl4

150 ** 25 25
EPS_2000-2531

**
lincRNA-EPS ** **
control RNA

mRNA fold change

CA tract 1 CA tract 2 CA tract 3 full-length 20 20 **

Input (1%)

** EV Ctl
2172 - 2177 2346 - 2351 2386 - 2393 100
beads

15 15 EPS_WT
CA3a
CA3b

WT CAAACA..........CACACA........CATACACA
CA1
CA2
WT

CA1 GTAACA..........CACACA........CATACACA EPS_CA3a

10 10
CA2 CAAACA..........GTGTCA........CATACACA 50 EPS_CA3b
CA3a CAAACA..........CACACA........GTTACACA 5 5
CA3b CAAACA..........CACACA........GTTAGTGA Streptavidin pulldown
hnRNPL motif = CANACA IB: hnRNPL 0 0 0
LPS LPS LPS

Figure 6. Identification of hnRNPL as a Binding Partner of lincRNA-EPS

(A) SDS-PAGE analysis of proteins purified from in vitro binding assay using biotinylated lincRNA-EPS or antisense control RNA and macrophage nuclear
extracts. The highlighted protein bands were subjected to mass spectrometry analysis.
(B) Western blot confirms lincRNA-EPS and hnRNPL interaction in vitro.
(C and D) hnRNPL RIP followed by qRT-PCR analysis of co-purified RNAs in non-cross-linked BMDMs (C) and formaldehyde cross-linked BMDMs (D).
Immunoprecipitation of hnRNPL was assessed by western blot (C).
(E) Western blot assessing the knockdown of hnRNPL in iBMDMs stably expressing shRNAs targeting non-overlapping regions of hnRNPL, and the control
shRNA against GFP.
(F and G) Control iBMDMs or those expressing hnRNPL-specific shRNAs were stimulated with LPS and subjected to qRT-PCR analysis of Cxcl10 mRNA (F) and
ELISA against Cxcl10 protein levels (G).
(H) qRT-PCR analysis of unspliced and spliced forms of lincRNA-EPS in iBMDMs expressing control or hnRNPL-specific shRNAs. *p < 0.05.
(I) Schematic of lincRNA-EPS deletion mutants used in RNA-protein binding assays.
(J) hnRNPL binds 30 -region of lincRNA-EPS. The RNA-protein binding assay was performed using biotinylated full-length or deletion mutants of lincRNA-EPS and
the nuclear extracts isolated from BMDMs, captured using streptavidin beads, and subjected to western blot against hnRNPL.
(K) The 30 -region (2000–2531 nt) of lincRNA-EPS is necessary and sufficient to suppress Cxcl10 expression. lincRNA-EPS/ iBMDMs expressing full-length or
deletion mutants of lincRNA-EPS were stimulated with LPS for 6 hr and mRNA levels analyzed by qRT-PCR. *p < 0.05; **p < 0.01.
(L) Schematic of CANACA motifs in the 30 -region (2000–2531 nt) of lincRNA-EPS. The nucleotide mutations in CANACA tracts are highlighted.
(M) lincRNA-EPS binds hnRNPL through a CANACA motif (2386–2391 nt) embedded in its 30 -region. The RNA-protein binding assay was performed using
biotinylated WT or CANACA tract mutant versions of the full-length lincRNA-EPS, captured using streptavidin beads, and subjected to western blot against
hnRNPL.
(N) Functional role of CANACA motif (2386–2391 nt) of lincRNA-EPS. lincRNA-EPS/ BMDMs expressing WT or the point mutants of the full-length lincRNA-EPS
(defective in hnRNPL binding) were stimulated with LPS for 6 hr and mRNA levels analyzed by qRT-PCR. **p < 0.01.
Error bars indicate SD. See also Figure S5.

Cell 165, 1672–1685, June 16, 2016 1681

were challenged with LPS i.p. for 5 hr and cytokine levels
measured using multiplex assays. Following LPS challenge,
lincRNA-EPS/ animals had significantly elevated levels of in-
flammatory cytokines in serum (Figure 7A), peritoneal fluid (Fig-
ure 7B), liver (Figure 7C), and spleen (Figure 7D). Notably, the
protein levels of key immune genes including Cxcl10, Ccl5,
and IL6 were significantly elevated both locally in peritoneal fluid
as well as systemically in serum, spleen, and liver in lincRNA-
EPS/ animals compared to WT animals. Further, the inducible
expression of IRG mRNAs was also enhanced in the spleen of
lincRNA-EPS-deficient animals relative to WT animals (Fig-
ure 7E). Flow cytometry analysis of peritoneal fluid in LPS-chal-
lenged mice however indicated that the recruitment of immune
cells including neutrophils and inflammatory monocytes was
comparable between WT and lincRNA-EPS/ mice (Figure S6).
We also analyzed the basal expression of a subset of lincRNA-
EPS-regulated IRGs in spleens in vivo by qRT-PCR (Figure 7F).
Interestingly, lincRNA-EPS-deficient animals showed increased
levels of several IRGs relative to WT counterparts (Figure 7F).
These results demonstrate that lincRNA-EPS plays an important
role in controlling both the homeostatic and TLR4-inducible pro-
gram of IRG expression in vivo.
Finally, we assessed the role of lincRNA-EPS in controlling
the survival of these animals. At a sub-lethal dose of LPS chal-
lenge (25 mg/kg mice weight, i.p.), lincRNA-EPS/ mice showed
significantly higher susceptibility to endotoxic-shock (p = 0.0016;
log-rank [Mantel-Cox] test) since 100% of these mice succumbed
to death by 30 hr (Figure 7G, left). In contrast, only 50% of WT an-
imals succumbed to death at this dose and time point. WT ani-
mals that failed to succumb recovered with no visible signs of
illness by day 3 post-LPS challenge. We also performed similar
survival studies at a low (non-lethal) dose of LPS. Notably,
lincRNA-EPS-deficient animals were significantly susceptible to
this non-lethal dose of LPS since only 20% of lincRNA-EPS/
mice survived up to 5 days post-LPS challenge in contrast to
100% survival rate observed for WT mice (Figure 7G, right). Taken
together, these results demonstrate that lincRNA-EPS plays an
important role in restraining lethal inflammatory responses in vivo.
DISCUSSION
Effective immune defense against pathogen challenge is depen-
dent on a broad transcriptional program induced in macro-
phages and other immune cells. Enormous progress has been
made in delineating how this transcriptional program is induced
and regulated. Numerous factors acting at the level of receptor
ligation, signaling, transcription factor activation, and chromatin
state collectively coordinate the intensity and duration of
these responses. Mammalian genomes also encode regulatory
RNAs that control immune gene expression. While the role of
microRNAs in these processes is well established (O’Connell
et al., 2012), the role of lncRNAs in the immune system are still
poorly understood.
In this study, we demonstrate that lincRNA-EPS downre-
gulates IRG expression in macrophages. In resting cells,
lincRNA-EPS suppresses the transcription of immune genes
by controlling nucleosome positioning to maintain a repressed
chromatin state. In response to microbial ligands, however,
1682 Cell 165, 1672–1685, June 16, 2016
this brake is released and IRG expression induced in a temporal
manner (Figure 7H). Notably, the kinetics of lincRNA-EPS sup-
pression in immune cells is consistent with its role as a temporal
regulator since it is expressed in resting macrophages but sup-
pressed soon after the proinflammatory response is initiated.
Using both gain-of-function and loss-of-function approaches,
we have gained critical insights into the biological functions
of lncRNA-EPS at both the cellular and organismal level. Unbi-
ased transcriptome profiling of lincRNA-EPS/ BMDMs demon-
strated that lincRNA-EPS selectively regulated the expression
of a large number of immune genes, all of which play key roles
in host defense against pathogens. Notably, we found that
ectopic expression of lincRNA-EPS in lincRNA-EPS/ macro-
phages reversed the enhanced expression of IRGs. Thus, dele-
tion of the lincRNA-EPS DNA locus did not affect regulatory
element(s) other than those controlling lincRNA-EPS expression.
By controlling levels of ASC, lincRNA-EPS also downregulated
inflammasome-dependent responses. Our work also demon-
strates that lincRNA-EPS restrains inflammation in vivo. In
response to LPS challenge, lincRNA-EPS-deficient mice pro-
duced elevated systemic levels of cytokines. Moreover,
lincRNA-EPS-deficient animals were more susceptible to LPS-
induced lethality compared to their WT counterparts. While we
cannot be certain that all of these effects are due only to the
absence of lincRNA-EPS expression in macrophages, these re-
sults provide compelling genetic evidence for in vivo functions
of lincRNA-EPS in the immune system.
In addition to identifying physiologic functions for this lincRNA
both in isolated cells and in vivo, we provide mechanistic insights
into how lincRNA-EPS controls IRG expression. Consistent with
its nuclear localization, we find that lincRNA-EPS controls the
basal and TLR4-inducible expressions of IRGs at the transcrip-
tional level. The role of lincRNA-EPS in repressing gene tran-
scription is supported by RNA pol II ChIP studies. Moreover,
ATAC-seq and NucleoATAC analysis collectively indicate that
lincRNA-EPS promotes a repressed chromatin state by control-
ling nucleosome positioning at TSSs of target IRGs in the
absence of an activating signal. The evidence from our RNA
FISH experiments, which reveals multiple foci throughout the
nucleus, lend support to the idea that lincRNA-EPS might
act directly at multiple genomic loci, rather than indirectly by
affecting only major regulators in the genome.
Our mechanistic studies also indicate that lincRNA-EPS func-
tions at least in part through its interaction with hnRNPL. Our re-
sults identify a CANACA motif in the 30 -531 region that is critical
for both hnRNPL binding and lincRNA-EPS-dependent suppres-
sion of some IRGs. Despite the presence of three such motifs in
this region, only CA tract 3 (2386–2391 nt), that spans a pre-
dicted stem-loop region is functionally important. These findings
indicate that either the sequence or secondary structure of
this region is involved in mediating lincRNA-EPS functions.
Further studies will be required, however, to determine if all the
lincRNA-EPS-dependent effects we have observed are depen-
dent on its interaction with hnRNPL. hnRNPs are important func-
tional partners for numerous additional lincRNAs. These include
lincRNA-p21, which functions via hnRNPK (Huarte et al., 2010),
Xist and Firre, which both function via hnRNPU (Hacisuleyman
et al., 2014; Hasegawa et al., 2010), as well as lincRNA-Cox2,
A In vivo serum cytokines post LPS-challenge E
Spleen (In vivo)
15 100 100 40 6 10 2.5 20 2.0 2.0
*

PS .
.
, L i.p
*

i.p
ns ns
serum protein (ng/ml)

T, tol
* * * *

KO LPS
80 80

n
* 8 2.0

co
** 30 15 1.5 1.5

T
10 4

W
W
60 60 6 1.5
10 1.0 1.0 WT ifnβ1
20
40 40 4 1.0 KO socs1
5 2 IL4
20 20 10 5 0.5 0.5
2 0.5 IL18
0 0 0 0 0 0 0 0 0 0 IL1a
Cxcl10 Ccl5 IL6 MIP-2 IL-1α IFNγ IL-10 MIG IL-12 p40 IL-12 p70 cox2
mda5
B Peritoneal fluid C Liver rig-i
ifn-γ
** 4000 ns 10 4.0 1.0 0.08
1200 5000 10000 ** 300
**
8 *** ns IL6
ns
** 4000 ** 250 8
** 0.8
IL5
8000

protein (ng/ml)
900 3000 6 3.5 0.06 ccl2
protein (pg/ml)

200 cxcl10
3000 6000 WT 6 0.6 WT
600 2000 150 4 3.0 0.04 cxcl1
KO KO
2000 4000 4 0.4 cxcl9
100
300 1000 2 2.5 0.02 irf7
1000 2000 50 2 0.2
stat1
0 0 0 0 0 0 0 2.0 0.0 0.00 nlrc5
Ccl5 IL6 G-CSF MCP-1 IL-17 Tnfα Ccl5 IL-1α IL6 Tnfα nlrp3
tnfα
D Spleen nos2
20 20 80 60 60 20 8 0.10 0.6 10 15 0.4 IL1ra
* * * ns IL12b
* ** * * ** ns
0.08 IL12a
** *
protein (ng/ml)

15 15 60 15 6 8 0.3
40 40 0.4 10 IL10
0.06
10 WT ifnα4
10 40 10 4 6 0.2
0.04 IL27
KO
20 20 0.2 5 ccl21
5 5 20 5 2 4 0.1
0.02 nlrc4
tgfβ1
0 0 0 0 0 0 0 0.00 0.0 2 0 0.0
Cxcl10 Ccl5 G-CSF KC MIP-2 MCP-1 IL-1α IL-13 IL-17 MIP-1α MIP-1β IL-12 p70 arg1
relative expression

Row min Row max

F In vivo basal expression (Spleen)
normalized mRNA expression

0.015 0.15 0.05 0.004 0.004 0.06 0.004 0.003 ns

*
0.04 * ** * *
** 0.003 0.003 ** 0.003
(/Gapdh)

0.010 0.10 0.04 0.002

0.03 WT
0.002 0.002 0.002
0.02 KO
0.005 0.05 0.02 0.001
0.001 0.001 0.001
0.01

0.000 0.00 0.00 0.000 0.000 0.00 0.000 0.000

Cxcl10 Cxcl9 Ccl5 Ccl4 IL6 Ifit1 IL-1α Gbp4

H
G Steady State Infection/Inflammation
100 100
TLR-MyD88-Trif
(5/5)
80 80 lincRNA-EPS lincRNA-EPS
Percent survival

Percent survival

ON NF-kB
OFF
60 60
WT WT
(6/14) KO hnRNPL
40 40 3’ OFF ON
(2/8) 5’ TF
p = 0.0016
20 20
p = 0.012 IRGs IRGs
(0/28)
0 0
days: 0 1 2 3 days: 0 2 4 6 KO TF
LPS (25 mg/kg) LPS (5 mg/kg)
Permissive chromatin state IRGs
(nucleosome depletion at TSS)
H3K4me3 Transcriptional machinery

Figure 7. lincRNA-EPS Restrains Inflammation In Vivo

(A–D) Cytokine levels in serum (A), peritoneal fluid (B), liver (C), and spleen (D) of WT and lincRNA-EPS/ (KO) mice challenged i.p. with E. coli LPS (5 mg/kg/mice)
for 5 hr. Data are shown as mean ± SEM (n = 4–6 mice per group). *p < 0.05; **p < 0.01, ***p < 0.001; ns, not significant.
(E) Heatmap of gene expression in spleens isolated from WT and lincRNA-EPS/ mice challenged i.p. with LPS for 5 hr and analyzed by NanoString. Image
represents fold change relative to untreated WT mice for all genes that were differentially expressed at least 2-fold in response to LPS (n = 4–6 mice per group).
(F) Basal gene expression profiles of IRGs in vivo in spleens isolated from WT and lincRNA-EPS/ mice and analyzed by qRT-PCR. *p < 0.05; **p < 0.01; ns, not
significant. Error bars indicate SD.
(G) Survival data of WT and lincRNA-EPS/ mice in response to LPS challenge. The numbers of mice that survived in each condition is provided. The statistical
test of differences was calculated using log-rank (Mantel-Cox) test with p < 0.05 considered as significant.
(H) Integrated model depicting lincRNA-EPS as a transcriptional brake that controls basal and inducible expression of IRGs in macrophages. TF, transcription
factor.
See also Figure S6.

Cell 165, 1672–1685, June 16, 2016 1683

which utilizes hnRNP-A2/B1 (and hnRNP-A/B) (Carpenter et al.,
2013). It is well known that hnRNPs are involved in mRNA
biogenesis. In addition, a role for hnRNPL in the transcriptional
regulation of gene expression is also beginning to emerge. For
instance, hnRNPL is a component of the Med23 mediator com-
plex that plays an important role in gene transcription in epithelial
cells (Huang et al., 2012). Moreover, hnRNPL is functionally
linked to the transcriptional program mediated by Aire in devel-
oping T cells (Giraud et al., 2014). To control transcription of
TNFa in THP-1 monocytes, hnRNPL also interacts with an
intronic lncRNA, THRIL (Li et al., 2014). The identification of
hnRNPL as a functional binding partner of lincRNA-EPS adds
further support to the role of hnRNPs in transcriptional regula-
tion, expanding the role of these RNA binding proteins beyond
their well-known functions in mRNA processing.
The mechanism whereby lincRNA-EPS specifically localizes to
the genomic loci of IRGs to mediate gene repression remains to
be fully defined. At least three possibilities exist. First, lincRNA-
EPS could interact directly with nascent pre-mRNAs at their
genomic loci via RNA:RNA interactions to block their ongoing
transcription. Similar interactions involving lncRNA:pre-mRNAs
at specific genomic loci have been recently described for
NEAT1 and MALAT1 (West et al., 2014), as well as U1 snRNA (En-
greitz et al., 2014). Alternatively, lincRNA-EPS might interact
directly with target DNA sequences to form higher-order
DNA:RNA triplex structures, similar to the recently described
model for lncRNA genes PARTICLE (O’Leary et al., 2015) and
MEG3 (Mondal et al., 2015). Finally, lncRNA-EPS may target
genes through a ribonucleoprotein complex that does not rely
on specific base-pairing with either DNA or mRNA. Instead,
such RNP complexes, which in the case of lincRNA-EPS appear
to involve hnRNPL, would target specific genes through a care-
fully coordinated set of multiprotein interactions that resemble
transcription factor recruitment at specific genomic sites.
In summary, we have employed detailed molecular and genetic
approaches to demonstrate that the precise regulation of lincRNA-
EPS expression in macrophages is essential to control inflamma-
tory responses by suppressing the transcription of key immune
genes. Therefore, lincRNA-EPS acts as an important component
of the molecular circuitry to prevent spontaneous activation of im-
mune genes with potential implications in chronic inflammation
and immune pathologies. The insights obtained from this study
further advance our understanding of the physiological roles of
lncRNAs in general and the growing importance of these mole-
cules in the immune system.
EXPERIMENTAL PROCEDURES
Cell Culture and Stimulation
BMDMs were differentiated from bone-marrow cells with 20% L929 superna-
tant for 7 days and stimulated with LPS (100 ng/mL), Pam3CSK4 (200 ng/mL),
poly(I:C) (25 mg/mL), type I IFN (500 U/mL), TNFa (10 ng/mL), and IL-10
(10 ng/mL).
Generation of lincRNA-EPS Knockout Mice
lincRNA-EPS KO mouse was generated by replacing the lincRNA-EPS genomic
locus (4 kb) with a neomycin cassette under control of a Pgk1 promoter.
lincRNA-EPS-targeting vector was electroporated into C57BL/6 mouse embry-
onic stem (ES) cells. Positive ES cells were injected into blastocytes to generate
1684 Cell 165, 1672–1685, June 16, 2016
chimeric mice. lincRNA-EPS heterozygous mice were obtained by gamete line
transmission from mating the chimeric mice with WT C57BL/6 mice.
RNA-Seq and Data Analysis
WT and lincRNA-EPS/ BMDMs (two mice/group) were stimulated with LPS
(100 ng/mL) for 0, 2, and 6 hr. RNA-seq libraries were prepared from total RNA
depleted of rRNA and sequenced on Illumina Hiseq2000 as described (Heyer
et al., 2015).
Cytokine Analysis
Cytokine levels in supernatants were measured by ELISA: Cxcl10 and Ccl5
(R&D Systems), IL6, TNFa, and IL-1b (eBioscience). Multiplex protein analysis
was performed by LUNARIS Mouse 12-Plex Cytokine Kit (AYOXXA Bio-
systems) and Mouse Cytokine/Chemokine Array 31-Plex (EVE Technologies
Corporations).
ATAC-Seq
WT and lincRNA-EPS/ BMDMs were stimulated with LPS (100 ng/mL) for 0,
2, or 6 hr in biological duplicates (two mice/group) and subjected to ATAC-seq
as described (Buenrostro et al., 2013). Briefly, nuclei pellets isolated from
50,000 cells were resuspended in 50 mL transposition buffer containing Tn5
transposase (Illumina) and incubated at 37 C for 30 min. Barcoded ATAC-
seq libraries were sequenced on Illumina Nextseq.
A detailed description of the methods is provided in the Supplemental
Experimental Procedures.
ACCESSION NUMBER
The accession number for the RNA-seq data reported in this paper is
ArrayExpress: E-MTAB-4088. The accession number for the ATAC-seq data
reported in this paper is GEO: GSE78873.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures,
six figures, and four tables and can be found with this article online at http://
dx.doi.org/10.1016/j.cell.2016.05.075.
AUTHOR CONTRIBUTIONS
M.K.A. and K.A.F. designed the research, analyzed results, and wrote the
manuscript. M.K.A. performed the majority of the experiments, with contribu-
tions from W.H., E.P.R., J.R.A.-D., S.A.S., A.B., J.D.M., J.B., J.E.B., and P.G.
D.R.C. performed bioinformatics analyses and contributed to data analysis
and manuscript preparation. A.T.S., Y.S., and H.Y.C. performed ATAC-seq
analysis. H.L. and M.J.M. provided critical reagents and suggestions. All au-
thors read and provided suggestions during manuscript preparation.
ACKNOWLEDGMENTS
We sincerely thank Zhaozhao Jiang and Kelly Army for animal care and tech-
nical help, Mona Motwani and Shruti Sharma for help with flow cytometry, Ali-
cia Schep for NucleoATAC analysis, Rui Li for help during initial stages of
ATAC-seq library preparation, and all members of the Fitzgerald laboratory
as well as Susan Carpenter (UCSC) for their insightful comments. We thank
Michael Karin (UCSD) for providing RNA samples, and Scott Schaffer and
John Leszyck (UMass, Worcester) for mass spectrometry. This study was sup-
ported by American Heart Association grant (14POST18930001) to M.K.A., by
NIH grant (P50-HG007735) to H.Y.C., by NIH grant (DK068348) to H.F.L., and
by grants from the Kenneth Rainin Foundation, Lupus Research Foundation,
and NIH (AI067497) to K.A.F.
Received: November 30, 2015
Revised: April 19, 2016
Accepted: May 25, 2016
Published: June 16, 2016
REFERENCES
Atianand, M.K., and Fitzgerald, K.A. (2014). Long non-coding RNAs and con-
trol of gene expression in the immune system. Trends Mol. Med. 20, 623–631.
Buenrostro, J.D., Giresi, P.G., Zaba, L.C., Chang, H.Y., and Greenleaf, W.J.
(2013). Transposition of native chromatin for fast and sensitive epigenomic
profiling of open chromatin, DNA-binding proteins and nucleosome position.
Nat. Methods 10, 1213–1218.
Carpenter, S., Aiello, D., Atianand, M.K., Ricci, E.P., Gandhi, P., Hall, L.L., By-
ron, M., Monks, B., Henry-Bezy, M., Lawrence, J.B., et al. (2013). A long non-
coding RNA mediates both activation and repression of immune response
genes. Science 341, 789–792.
Castellanos-Rubio, A., Fernandez-Jimenez, N., Kratchmarov, R., Luo, X., Bha-
gat, G., Green, P.H., Schneider, R., Kiledjian, M., Bilbao, J.R., and Ghosh, S.
(2016). A long noncoding RNA associated with susceptibility to celiac disease.
Science 352, 91–95.
Chan, J., Atianand, M., Jiang, Z., Carpenter, S., Aiello, D., Elling, R., Fitzgerald,
K.A., and Caffrey, D.R. (2015). Cutting Edge: A Natural Antisense Transcript,
AS-IL1a, Controls Inducible Transcription of the Proinflammatory Cytokine
IL-1a. J. Immunol. 195, 1359–1363.
Chen, G.Y., and Nuñez, G. (2010). Sterile inflammation: sensing and reacting to
damage. Nat. Rev. Immunol. 10, 826–837.
Derrien, T., Johnson, R., Bussotti, G., Tanzer, A., Djebali, S., Tilgner, H., Guer-
nec, G., Martin, D., Merkel, A., Knowles, D.G., et al. (2012). The GENCODE v7
catalog of human long noncoding RNAs: analysis of their gene structure, evo-
lution, and expression. Genome Res. 22, 1775–1789.
Djebali, S., Davis, C.A., Merkel, A., Dobin, A., Lassmann, T., Mortazavi, A.,
Tanzer, A., Lagarde, J., Lin, W., Schlesinger, F., et al. (2012). Landscape of
transcription in human cells. Nature 489, 101–108.
Engreitz, J.M., Sirokman, K., McDonel, P., Shishkin, A.A., Surka, C., Russell,
P., Grossman, S.R., Chow, A.Y., Guttman, M., and Lander, E.S. (2014).
RNA-RNA interactions enable specific targeting of noncoding RNAs to
nascent Pre-mRNAs and chromatin sites. Cell 159, 188–199.
Giraud, M., Jmari, N., Du, L., Carallis, F., Nieland, T.J., Perez-Campo, F.M.,
Bensaude, O., Root, D.E., Hacohen, N., Mathis, D., and Benoist, C. (2014).
An RNAi screen for Aire cofactors reveals a role for Hnrnpl in polymerase
release and Aire-activated ectopic transcription. Proc. Natl. Acad. Sci. USA
111, 1491–1496.
Gomez, J.A., Wapinski, O.L., Yang, Y.W., Bureau, J.F., Gopinath, S., Monack,
D.M., Chang, H.Y., Brahic, M., and Kirkegaard, K. (2013). The NeST long
ncRNA controls microbial susceptibility and epigenetic activation of the inter-
feron-g locus. Cell 152, 743–754.
Guttman, M., and Rinn, J.L. (2012). Modular regulatory principles of large
non-coding RNAs. Nature 482, 339–346.
Guttman, M., Amit, I., Garber, M., French, C., Lin, M.F., Feldser, D., Huarte, M.,
Zuk, O., Carey, B.W., Cassady, J.P., et al. (2009). Chromatin signature reveals
over a thousand highly conserved large non-coding RNAs in mammals. Nature
458, 223–227.
Hacisuleyman, E., Goff, L.A., Trapnell, C., Williams, A., Henao-Mejia, J., Sun,
L., McClanahan, P., Hendrickson, D.G., Sauvageau, M., Kelley, D.R., et al.
(2014). Topological organization of multichromosomal regions by the long
intergenic noncoding RNA Firre. Nat. Struct. Mol. Biol. 21, 198–206.
Hasegawa, Y., Brockdorff, N., Kawano, S., Tsutui, K., Tsutui, K., and Naka-
gawa, S. (2010). The matrix protein hnRNP U is required for chromosomal
localization of Xist RNA. Dev. Cell 19, 469–476.
Heyer, E.E., Ozadam, H., Ricci, E.P., Cenik, C., and Moore, M.J. (2015).
An optimized kit-free method for making strand-specific deep sequencing
libraries from RNA fragments. Nucleic Acids Res. 43, e2.
Hu, W., Yuan, B., Flygare, J., and Lodish, H.F. (2011). Long noncoding RNA-
mediated anti-apoptotic activity in murine erythroid terminal differentiation.
Genes Dev. 25, 2573–2578.
Huang, Y., Li, W., Yao, X., Lin, Q.J., Yin, J.W., Liang, Y., Heiner, M., Tian, B.,
Hui, J., and Wang, G. (2012). Mediator complex regulates alternative mRNA
processing via the MED23 subunit. Mol. Cell 45, 459–469.
Huang, W., Thomas, B., Flynn, R.A., Gavzy, S.J., Wu, L., Kim, S.V., Hall, J.A.,
Miraldi, E.R., Ng, C.P., Rigo, F., et al. (2015). DDX5 and its associated lncRNA
Rmrp modulate TH17 cell effector functions. Nature 528, 517–522.
Huarte, M., Guttman, M., Feldser, D., Garber, M., Koziol, M.J., Kenzelmann-
Broz, D., Khalil, A.M., Zuk, O., Amit, I., Rabani, M., et al. (2010). A large inter-
genic noncoding RNA induced by p53 mediates global gene repression in the
p53 response. Cell 142, 409–419.
Janeway, C.A., Jr., and Medzhitov, R. (2002). Innate immune recognition.
Annu. Rev. Immunol. 20, 197–216.
Lamkanfi, M., and Dixit, V.M. (2012). Inflammasomes and their roles in health
and disease. Annu. Rev. Cell Dev. Biol. 28, 137–161.
Li, Z., Chao, T.C., Chang, K.Y., Lin, N., Patil, V.S., Shimizu, C., Head, S.R.,
Burns, J.C., and Rana, T.M. (2014). The long noncoding RNA THRIL regulates
TNFa expression through its interaction with hnRNPL. Proc. Natl. Acad. Sci.
USA 111, 1002–1007.
Medzhitov, R., and Horng, T. (2009). Transcriptional control of the inflamma-
tory response. Nat. Rev. Immunol. 9, 692–703.
Mondal, T., Subhash, S., Vaid, R., Enroth, S., Uday, S., Reinius, B., Mitra, S.,
Mohammed, A., James, A.R., Hoberg, E., et al. (2015). MEG3 long noncoding
RNA regulates the TGF-b pathway genes through formation of RNA-DNA
triplex structures. Nat. Commun. 6, 7743.
Morris, K.V., and Mattick, J.S. (2014). The rise of regulatory RNA. Nat. Rev.
Genet. 15, 423–437.
O’Connell, R.M., Rao, D.S., and Baltimore, D. (2012). microRNA regulation of
inflammatory responses. Annu. Rev. Immunol. 30, 295–312.
O’Leary, V.B., Ovsepian, S.V., Carrascosa, L.G., Buske, F.A., Radulovic, V.,
Niyazi, M., Moertl, S., Trau, M., Atkinson, M.J., and Anastasov, N. (2015).
PARTICLE, a Triplex-Forming Long ncRNA, Regulates Locus-Specific Methyl-
ation in Response to Low-Dose Irradiation. Cell Rep. 11, 474–485.
Ray, D., Kazan, H., Cook, K.B., Weirauch, M.T., Najafabadi, H.S., Li, X., Guer-
oussov, S., Albu, M., Zheng, H., Yang, A., et al. (2013). A compendium of RNA-
binding motifs for decoding gene regulation. Nature 499, 172–177.
Schep, A.N., Buenrostro, J.D., Denny, S.K., Schwartz, K., Sherlock, G., and
Greenleaf, W.J. (2015). Structured nucleosome fingerprints enable high-reso-
lution mapping of chromatin architecture within regulatory regions. Genome
Res. 25, 1757–1770.
Smale, S.T., Plevy, S.E., Weinmann, A.S., Zhou, L., Ramirez-Carrozzi, V.R.,
Pope, S.D., Bhatt, D.M., and Tong, A.J. (2013). Toward an understanding of
the gene-specific and global logic of inducible gene transcription. Cold Spring
Harb. Symp. Quant. Biol. 78, 61–68.
Ulitsky, I., and Bartel, D.P. (2013). lincRNAs: genomics, evolution, and mech-
anisms. Cell 154, 26–46.
West, J.A., Davis, C.P., Sunwoo, H., Simon, M.D., Sadreyev, R.I., Wang, P.I.,
Tolstorukov, M.Y., and Kingston, R.E. (2014). The long noncoding RNAs
NEAT1 and MALAT1 bind active chromatin sites. Mol. Cell 55, 791–802.
Cell 165, 1672–1685, June 16, 2016 1685
Supplemental Figures

A C
lincRNA-EPS IL6
1000 BMDMs 1000 BMDMs
120 5
100 4

fold change (log10)

lincRNA-EPS (%)

lincRNA-EPS (%)
lincRNA-EPS (%) 100 0 hr
100

IL6 mRNA
80 3 2 hr
60 2 10 6 hr

40 1 10
1
20 0
0 -1 0.1 1
LPS (ng/ml): time (hr) 0 .5 2 6 24 IL-10 +
0 0.01 0.1 1 10 100 + +
+ IL-10 LPS
6 hr post-stimulation

B lincRNA-EPS IL6 D
120 4 10
mRNA fold change (log10)
100 8

log10 (copies/well)
lincRNA-EPS (%)

3
80 y = -0.2621(x) + 11.362
6
60 2 R2 = 0.997
4
40
1 lincRNA-EPS copy no. in BMDMs
20 2
= 11.12 +/- 1.59
0 0 0
time (hr): 0 1 2 6 12 24 0 5 10 15 20 25
LPS (100 ng/ml) CT

Figure S1. Regulation of lincRNA-EPS Expression in Macrophages, Related to Figure 1

(A) qRT-PCR analysis of lincRNA-EPS and IL6 expression in BMDMs stimulated with indicated doses of LPS for 6 hr.
(B) qRT-PCR analysis of lincRNA-EPS and IL6 expression in BMDMs stimulated for indicated times post LPS (100 ng/ml) treatment.
(C) qRT-PCR analysis of lincRNA-EPS expression in BMDMs stimulated with an anti-inflammatory cytokine IL-10 alone (left panel), and in cells stimulated with
LPS, or LPS and IL-10 for indicated time points.
(D) Copy-number analysis of lincRNA-EPS expression in resting BMDMs by qRT-PCR. Standard curve was generated using plasmid DNA expressing
lincRNA-EPS.
Error bars indicate SD.

Cell 165, 1672–1685, June 16, 2016 S1

A B G
100 kb (mm10) RNA-seq

4
ne

ES ne

ne
n
FPKM+1 (KO/WT)

lo

lo

lo

lo
C

C
109,300,000 109,350,000 109,400,000 109,450,000

ES

ES

ES
LPS (hr): 0 2 6
- WT ENSMUSG00000037849
Eps15 Ttc39a 10 kb -
lincRNA-EPS - Targeted transgene Pydc4
5 kb - I830012O16Rik
Ly6e
Ifit3
C P2 P3 Uchl1
WT KO Phf11b
EcoRI: 10.5 kb P1 P1
Phf11a
WT Rpl5
probe left arm lincRNA-EPS right arm marker +/+ +/- -/- NTC
1000 bp Ifi44
(4.5 kb) (1.2 kb)
Zbp1
P1+P2 Ifi27l2b
6.5 kb 500 bp
KO P1+P3 Xaf1
NeoR Phf11d
Bst2
Oasl2
D Complete Blood Analysis (5 weeks of age) E 40 Spleen BC094916

Cell frequency (%)

Gbp2
10 16 60
Hemoglobin (g/dL)

30 Pyhin1
WT Ms4a6b
Hematocrit (%)
RBC (x106/ul)

9 55 KO
Ifi27l2a
14 20 Slfn1
8 50 Rtp4
12 10 Saa3
7 45 Pdpn
Thbs1
0
10 Irg1
6 40

11 DC

NK DC
CD T c s

lls
M + T ls
s
11 es
l

ro ell
Clec4e

4+ cel
el

ce
CD hag
WT Hets KO WT Hets KO WT Hets KO

CD b+
b-
Ly6c2

CD B

p
Ifit2

8
ac
Oasl1
F Irf7
Top 5 GO functional categories
Ccl5
(≥ 2-FC in KO; Q-value < 0.05) LPS 0 hr LPS 2 hr
Gbp3
response to interferon−beta immune response Isg15
ENSMUSG00000000386
response to virus response to other organism Usp18
Ttc39a
response to other organism response to external biotic stimulus LOC100048759
Ptges
response to external biotic stimulus innate immune response
Arhgap27
defense response to other organism immune system process Ass1
ENSMUSG00000046687
0 5 10 15 0 5 10 15 Ms4a4c
−log10 (P−value) −log10 (P−value) Hes7
Ndufs7
H WT Ly6i
KO ENSMUSG00000075010
3000 * 1000 200 ** 4000 4000
* ** Ly6a
800 ** 3000 Ly6c1
Gbp5 mRNA

IL1α mRNA

150 3000
Ccl2 mRNA
Irg1 mRNA

Ifit2 mRNA

2000 ** 2000 log2-fold change

600
100 2000 1000
* 400
** 60 -2 -1 0 1 2
1000 40
200 ** 50
**
1000
** * 4.05
* 20
20.0 1.8 4.18
0 0 0 0 0
LPS (hr) 0 2 6 0 2 6 0 2 6 0 2 6 0 2 6

I J 4
normalized mRNA levels

0 BMDMs
Cell number (x 105)

Eps15
(/Gapdh; log10)

-1 3
-2 WT WT
KO 2 KO
-3 Ttc39a
1
-4
-5 0
LPS LPS
1

4
y

y
da

da

da

da

Figure S2. Generation of lincRNA-EPS Knockout Mice and Their Immune Phenotype, Related to Figure 2
(A) Strategy used in generating lincRNA-EPS
/
mice. The endogenous lincRNA-EPS locus was replaced with a Neomycin resistance (NeoR) cassette through
the lincRNA-EPS targeting construct containing homology arms (left: 4.5 kb; right: 1.2 kb). EcoRI restriction sites and the DNA probe used in Southern blot are
highlighted.
(B) Confirmation of lincRNA-EPS targeting in mouse embryonic stem (ES) cells. Genomic DNA was isolated from indicated ES clones and analyzed by Southern
blot to detect the WT allele (10.5 kb) and the targeted lincRNA-EPS allele containing NeoR cassette (6.5 kb). ES clones 1, 2, and 4 were selected for the blastocyst
injection.
(C) Genomic PCR strategy to distinguish WT and lincRNA-EPS
/
alleles from intercrosses of lincRNA-EPS+/
littermates. Genomic DNA from mouse tails were
purified, and subjected to PCR analysis using a combination of three indicated primers to simultaneously detect WT (600 bp) and KO (400 bp) alleles. PCR
products were resolved on 2% agarose gels.
(D) lincRNA-EPS
/
mice exhibit normal development of RBCs. Blood was collected from 5-weeks old WT, lincRNA-EPS+/
(Hets) and lincRNA-EPS
/
(KO)
mice and subjected to complete blood analysis (CBC) analysis. Each dot represents one mice.

(legend continued on next page)

S2 Cell 165, 1672–1685, June 16, 2016

(E) Flow cytometry analysis indicating frequencies of various immune cell types in spleens isolated from WT and lincRNA-EPS
/
mice (n = 3 mice per group).
Error bars, SEM.
(F) Immune pathway related genes are most strongly enriched among differentially expressed genes in lincRNA-EPS deficient macrophages. The 5 most
significantly enriched GO terms among differentially expressed genes (R2-FC in KO; Q-value < 0.05) in lincRNA-EPS
/
BMDMs compared to WT cells following
no treatment (left panel) or LPS treatment for 2 hr (right panel) are shown.
(G) lincRNA-EPS deficiency leads to derepression of basal IRG expression in macrophages. Heatmap of top 50 upregulated genes (R2-FC in KO; Q-value < 0.05)
in unstimulated lincRNA-EPS
/
BMDMs relative to WT cells identified via RNA-seq performed in biological duplicates. Relative expression of these 50 genes are
also shown following LPS treatment for 2 and 6 hr. mRNA levels are shown as log2-FC of FPKM+1 (KO/WT) at respective time points. Immune genes are
highlighted red.
(H) lincRNAs-EPS deficiency leads to elevated macrophage responses to TLR4 stimulation. WT and lincRNA-EPS
/
BMDMs were either left untreated or treated
with LPS for indicated times. mRNA levels of IRGs were quantified by qRT-PCR, and shown as fold change over untreated WT cells. *p < 0.05, **p < 0.01.
(I) qRT-PCR analysis of the expression of lincRNA-EPS neighboring genes Eps15 and Ttc39a in WT and lincRNA-EPS
/
BMDMs.
(J) Cell proliferation detected by Trypan-blue staining in WT and lincRNA-EPS
/
BMDMs at indicated days during macrophage differentiation.
Error bars indicate SD.

Cell 165, 1672–1685, June 16, 2016 S3

 A Cxcl10 IL6 Rsad2 Ccl5 Ifit1
800 ** 5000 500 ** 150 ** 400
**
mRNA fold change 4000 400
**
600 300 iBMDM
100 EV Ctl
3000 300
400 200 lincRNA-EPS
2000 200
50
200 100 100
1000
0 0 0 0 0
NT LPS NT LPS NT LPS NT LPS NT LPS

B NanoString mRNA analysis C J774.1

lincRNA-EPS expression
500 15
Cytosolic DNA-induced

400
mRNA fold change

150 iBMDM
EV Ctl 10
100 lincRNA-EPS

50 5

0 0
1α
d2

So 4
1
a

7
0
1
47

N Ctl

S
20
cs
nd

Irf
d4
Irf

at

EP
sa

Ifi

IL
St
Ifi

cR V
R

A-
lin E
D 1200
6070.0
mRNA fold change

400
J774.1
LPS-induced

300
EV Ctl
200 lincRNA-EPS

100

0
0
l9

G 9
10

11

d2

1
cl2
R 6

t1

t2
l1

ac

bp

bp

bp

bp

Irg
IL
xc

Ifi

Ifi
bp

bp

sa

C
xc

Pl

G
C

G
C

chromosome 5
cluster

E 30
**
mRNA fold change

*
lincRNA-EPS-/- iBMDMs
20 + EV Ctl
** + lincRNA-EPS

10 **
** ** ns
**
0
cl9

p4

p6

8
Irg

Ifit

Ifit

p1

-1
Gb

Gb
Cx

IL
Gb

LPS (TLR4)

Figure S3. Gain-of-Function and Rescue Studies Indicate lincRNA-EPS as a Repressor of IRG Expression in Macrophages, Related to
Figure 3
(A) Ectopic expression of lincRNA-EPS suppresses TLR4-induced expression of IRGs in macrophages. iBMDMs expressing lincRNA-EPS or the empty vector
control (EV Ctl) were stimulated with LPS for 6 hr, and mRNA levels were quantified by qRT-PCR. **p < 0.01.
(B) lincRNA-EPS regulates the expression of cytosolic DNA induced IRGs in macrophages. Cells as described above were transfected with poly(dA:dT) for 6 hr,
and mRNA levels analyzed by NanoString technology. mRNA levels of genes showing 2-folds or lower expression in cells expressing ectopic lincRNA-EPS
relative to control cells are shown. Data represent fold change relative to untreated cells.
(C and D) Ectopic expression of lincRNA-EPS suppresses TLR4-induced expression of IRGs in J744.1 macrophages. Stable cell lines expressing ectopic
lincRNA-EPS or EV Ctl were generated by retroviral transduction, and the expression of lincRNA-EPS was confirmed by qRT-PCR (C). Cells were stimulated with
LPS for 6 hr, and mRNA levels of lincRNA-EPS regulated cluster of genes on chromosome-5 (shown in Figures 2E and 2F) and others were analyzed by qRT-
PCR (D).
(E) Restoring lincRNA-EPS expression in lincRNA-EPS
/
macrophages is sufficient to suppress the expression of TLR4-induced IRGs. lincRNA-EPS
/

iBMDMs were transduced with retroviruses expressing lincRNA-EPS or the control pMSCV vector, and mRNA levels of indicated IRGs were analyzed by qRT-
PCR at 6 hr post-LPS treatment. IL-18 mRNA, which is not regulated by lincRNA-EPS, is shown as control. *p < 0.05; **p < 0.01.
Error bars indicate SD.

S4 Cell 165, 1672–1685, June 16, 2016

A B C
IgG IgG IgG
RNA pol II H3K4me3 RNA Pol II Ser2P
0.8 2.0 1.5 100 50 50 8 10 10
IL6 Irf7 ASC Ifit2 Gbp4 Cxcl9 Ifit2 Gbp4 Cxcl9

ChIP-qPCR (% Input)

ChIP-qPCR (% Input)
ChIP-qPCR (% Input)

0.6 1.5 80 40 40 8 8
6
1.0
60 30 30 6 6
0.4 1.0 4
40 20 20 4 4
0.5
0.2 0.5 2
20 10 10 2 2

0.0 0.0 0.0 0 0 0 0 0 0

LPS: + + + + + + WT KO WT KO WT KO WT KO WT KO WT KO
S

S
tl

tl

tl
C

C
Resting BMDMs LPS-stimulated BMDMs
EP

EP

EP
EV

EV

EV
A-

A-

A-
N

N
cR

cR

cR
lin

lin

lin

D ATAC-seq
T WT and KO BMDMs
quality metrics for WT E
TSS enrichment (WT) Insert size distribution (WT)
.000 .005 .010 .015 .020

Chromatin accessibility dynamics in response to LPS (TLR4)

7

0h 0h
2h 2h Color Key
6
signal

6h 6h
Frequency
5
ATAC-seq
4

0 .04 .08 .12

3
T
2

WT
T rep2 0h
WT
T rep1 0h
1

KO rep1 0h
-1000 -500 0 500 1000 0 200 400 600
KO rep2 0h
Distance from TSS (bp) Fragment length WT
T rep2 2h
WT
T rep1 2h
TSS enrichment (KO) Insert size distribution (KO)
KO rep2 2h
.000 .005 .010 .015 .020
7

0h 0h KO rep1 2h
2h 2h
6

WT
T rep2 6h
signal

6h 6h
Frequency

WT
T rep1 6h
5

KO rep2 6h
ATAC-seq
4

KO rep1 6h
3

KO rep1 6h
KO rep2 6h
T rep1 6h
T rep2 6h
KO rep1 2h
KO rep2 2h
T rep1 2h
T rep2 2h
KO rep2 0h
KO rep1 0h
T rep1 0h
T rep2 0h
T
2
1

WT
WT

WT
WT

WT
WT
-1000 -500 0 500 1000
Distance from TSS (bp) Fragment length

F Chromatin accessibility (AT

(ATAC-seq
ATAC-seq signal) G
LPS 2 hr LPS 6 hr
LPS (hr): at LPS induced IRGs in BMDMs 5 kb
6

50 20 Up lincRNA
A-EPS regulated
lincRNA-EPS Up lincRNA
A-EPS regulated
lincRNA-EPS
4

Down genes in RNA-seq

RNA q
RNA-se Down genes in RNA-seq
RNA q
RNA-se
0
5

0
Fold Change (KO/WT)

0
3

WT 2
4

6
3
2

50 20
0
2

0 0
1

KO 2
1
0

6
0

0 2 4 6 8 10 12 0 2 4 6 8 10
WT
T promoter ATAC-seq
T counts (log2) WT
T promoter ATAC-seq
T counts (log2)
Irg1 Ccl5

H LPS 2 hr LPS 6 hr
I J
lincRNA-EPS regulated genes

WT WT
0.35

Open chromatin (ATAC-seq

T signals) Open chromatin (ATAC-seq
T signals)
0.35

KO KO
nucleosome signals at

20 0.5 kb 70 0.5 kb
0.25

WT WT
0.25

0 0
20 70
0.15

0.15

KO KO
0 0
0.05

0.05

Irf7 Cxcl2
0.8

WT WT
0.1 0.2 0.3 0.4 0.5 0.6

Nucleosome position (TSS +/- 1 kb) Nucleosome position (TSS +/- 1 kb)
KO KO
nucleosome signals at

0.6 0.5
0.6

WT WT
all genes

0 0
0.4

0.6 0.5
KO KO
0.2

0 0
TSS TSS
−1000 −500 0 500 1000 −1000 −500 0 500 1000
Distance to TSS (bp) Distance to TSS (bp)

(legend on next page)

Cell 165, 1672–1685, June 16, 2016 S5

Figure S4. lincRNA-EPS Suppresses the Transcription of IRGs by Controlling Their Chromatin States, Related to Figure 5
(A) Ectopic expression of lincRNA-EPS in macrophages leads to reduced RNA pol II occupancy at TLR4-activated IRGs. iBMDMs expressing ectopic lincRNA-
EPS or EV Ctl were treated with LPS for 5 hr, and subjected to RNA pol II ChIP-qPCR analysis targeting the genomic regions around the TSS of indicated IRGs.
(B) lincRNA-EPS deficiency leads to the deposition of active H3K4me3 marks at IRGs in the absence of activating signals. H3K4me3 ChIP followed by qPCR
analysis of genomic regions around the TSS of lincRNA-EPS regulated IRGs in resting (no LPS treatment) BMDMs isolated from WT and lincRNA-EPS
/
mice.
(C) lincRNA-EPS deficiency leads to higher recruitment of RNA pol II Ser2P at TLR4-activated IRGs. WT and lincRNA-EPS
/
BMDMs were stimulated with LPS
for 5 hr, and subjected to ChIP using RNA pol II Ser2P or control IgG antibody, followed by qPCR analysis of genomic regions around the TSS of indicated IRGs.
(D) Quality metrics of ATAC-seq datasets showing ATAC-seq signals around TSS, and the insert size distribution in WT and lincRNA-EPS
/
BMDMs stimulated
with LPS for 0, 2 and 6 hr.
(E) Pearson correlation of ATAC-seq datasets in WT and lincRNA-EPS
/
BMDMs stimulated with LPS for 0, 2 and 6 hr in biological replicates.
(F) Genome tracks showing TLR4-induced dynamic changes in chromatin accessibility (ATAC-seq signals) at the promoters and within gene bodies of Irg1 and
Ccl5 in WT and lincRNA-EPS
/
BMDMs stimulated with LPS for 0, 2 and 6 hr.
(G) Scatter plots showing fold change of ATAC-seq signals (KO/WT) versus ATAC-seq signals at WT promoters (+/
1 kb of TSS) of lincRNA-EPS regulated genes
identified in RNA-seq (red: upregulated; blue: downregulated) (Figure 2A; Table S1), and all other genes (gray dots) in BMDMs stimulated with LPS for 2 hr and
6 hr. Dash lines represent 1.2 fold change.
(H) NucleoATAC analysis showing nucleosome signals in WT (blue) and lincRNA-EPS
/
BMDMs (KO; red) stimulated with LPS for 2 hr and 6 hr. Aggregate
nucleosome signals are shown within the promoters (+/
1 kb of TSS) of genes that are regulated by lincRNA-EPS (above panel), or within promoters of all genes
on a genome-wide level (bottom panel).
(I and J) Genome tracks of Irf7 (I) and Cxcl2 (J) showing chromatin accessibility (normalized ATAC-seq signal), and the nucleosome positioning (NucleoATAC
signals) centered around their transcription start sites in WT and lincRNA-EPS
/
BMDMs at basal conditions.
Error bars indicate SD.

S6 Cell 165, 1672–1685, June 16, 2016

 A B Ctl shRNA hnRNPL shRNA 2
9
hnRNPL shRNA 1 hnRNPL shRNA 3

Total spectral counts

8
7 80 Ccl5 8000 IL6
6

mRNA (fold change)

(log2)
5 60 6000
4
3
40 4000
2
1
0 hnRNPL 20 2000
KHDR1
RBM47
SF01
TOIP1
COR1B
0 0
NT LPS NT LPS
C
20 Ctl shRNA 200 Rsad2 200 Cxcl9

mRNA (fold change)

1
Ccl5 protein (ng/ml)

15 2 hnRNPL shRNA 150 150

3
10 100 100

5 50 50

0 0 0
0 5 25 100 NT LPS NT LPS
+ LPS (ng/ml)
3
250 Ifit1 8 IL18
IL6 protein (ng/ml)

mRNA (fold change)

200
2 6
150
4
1 100

50 2
0
0 0
0 5 25 100 NT LPS NT LPS
+ LPS (ng/ml)

D H3K4me3 ChIP-qPCR
8 3 4 6
Cxcl10 (% Input)

Irg1 (% Input)

Ifit1 (% Input)
IL6 (% Input)

6 3
2 4
Ctl shRNA
4 ** 2
** ** ** hnRNPL shRNA 2
1 2
2 1

0 0 0 0
LPS LPS LPS LPS

E lincRNA-EPS KO F lincRNA-EPS KO G WT CA tract 3 mutant

BMDMs BMDMs A
A A
A
A A
A
A

0.03
U U

1.5
A UU A UU
relative expression (/Gapdh)

AU AU
AU AU
UG UG
relative expression (/ WT)

U U U U
U AC U AC
CG CG
CG

3’ 3’
CG
UA UA
U AU A
A A UA
UA GC
GC AU
AU U
U G
UU G U U
U U
A GCG
GCG C U UA
UA U U
GU

5’
G
U G CA UC

5’
GU G
G C G C
G C G
U C U G A

0.02
A U UAA
UAA G

1.0
G C CA GA
UAA A U A G C A UA
A A CU UA U AA AU A AU
A A AA AC
A AU A UC A A A UG
U U UU U A A CU U G
C
GA A U G GU A A U A G AGAAAAAU G
G A A A U
A A
U G CC U A U U CU C U A
U
U
U GA
A
U U A A G C A A
G AA AG A C A A A
A UGC U C
U C A UA A U U UU U A AU
G AA AG
UA A
CC U U AC C A U C A UA A GU
AA U U A A
A G GU C A AG AU A A UA A
CC U U AC A U
C
U C A
G UA A U A U
A A A GG A C
U A U
A UU
U G GC U A UG A U U A
CG A A GC CG A AC U UA A U UA A
U
CUGG C U U U G GC C U
A
C AU G A U CG A A GC CG
GC
U U
U A U AU A C CUGG C U C
A
A U G UU U G
U G CA UC
G U CA A A AC G G G UA C A U A U AU U C
U C G A U G
A C U G G G GA A G U CA A A AC G UC AG U C UU
U CC A U AC A UA
A A C U G G G GA C AA
G A C G CAA A A
AU U CC A
A A AA A A AUA
AA A

0.01 CA tract 3
U AA A

0.5
C GGU G A UGA A U G A
G UU C CCA A
CUGUUAC
UA C C A U C AAC A A
GC U A U AA AAC A
A C U CAU
CG CACU G UU C
AU UA C A
CG GC UAC
C A CG

CA tract 2
AU
CG
GU

A U AU
CG
C A
AU

G
UA
G CG
A C GU
A

A
C UA

ND
U AC
GC
C
A G A C
G
A C
U AC
A

C
CG

ND
GC
UA C
GC A G
C C CG

0.00 0
UA UA
GC
GU C C

U
CG UA
U
C G GU
U A CG

A
+ lincRNA-EPS_CA3a
+ lincRNA-EPS_CA3b
+ lincRNA-EPS_WT
+ lincRNA-EPS_2000-2531
+ lincRNA-EPS_1-2000

+ lincRNA-EPS_WT

G A

A
U
A A
AU C G
+ EV Ctl

+ EV Ctl

C UU
C U UA U A
C A A U AG GC A
G A A
AU
A G AA A C UU

U A
GC C U UA

C
A C C A A U AG
A A AU A G AA GC
AGC
C G GC A C
AA AA AU

U
GC A A
CG
UG
G C
AA AA
G GC
GC
CG

A
GU

CA tract 1
G CU C C
AU
CG
C
C A UG
GU
G CU C C

C
C

G
AU
A UA C A A C
CG
GC A UA C A A
AU

U
GC
UA AU
G UA
G UG CU G

2386 - 2391
C G
UC

C
CG G UG CU G
C G UA C
A G UC CG
CG C G
G U A G
UA
GU C UA CG
G A C U G U
GU C UA
U G AU G A C U
U G GC U G AU
A A A UG U G GC
AG A
G A A
AG A UG
G G A
G A G
AU G A
CG AU
UA CG
UA UA
GC UA
GC GC
GC
UA UC U
U U U A U CU
U A U U
U A
A A A A
A C C
A
G A A
C G
A A C
C U U
C C GC C
C C GC
AU AU
C C C C
A A A A
CG CG

C
CG

A C
A
lincRNA-EPS_2000-2531 C
CG

A C
A

Figure S5. Knockdown of hnRNPL Expression Leads to Elevated Macrophage Responses to TLR4 Stimulation, Related to Figure 6
(A) lincRNA-EPS interacting proteins identified in mass spectrometry. Biotin-RNA: protein complexes were captured using streptavidin magnetic beads, and
subjected to mass spectrometry analysis. Proteins showing at least 2-fold enrichment in lincRNA-EPS pull-down relative to antisense RNA control are shown in
order of their overall abundance as indicated by total spectral counts: hnRNPL (Heterogeneous nuclear ribonucleoprotein L), KHDR1 (KH domain-containing,

(legend continued on next page)

Cell 165, 1672–1685, June 16, 2016 S7

RNA-binding, signal transduction-associated protein 1), RBM47 (RNA-binding protein 47), SF01 (Splicing factor 1), TOIP1 (Torsin-1A-interacting protein 1), and
COR1B (Coronin-1B).
(B) shRNA mediated knockdown of hnRNPL expression leads to elevated mRNA levels of TLR4-regulated IRGs. iBMDMs expressing shRNAs targeting three
different regions of hnRNPL or a control shRNA (against GFP) were stimulated with LPS for 6 hr. mRNA levels of lincRNA-EPS regulated IRGs were quantified by
qRT-PCR. Expression of IL18 mRNA, which is not regulated by lincRNA-EPS, is shown as control.
(C) Knockdown of hnRNPL expression leads to elevated cytokine secretion in TLR4-activated macrophages. iBMDMs expressing control or hnRNPL specific
shRNAs (described above) were stimulated with indicated amounts of LPS, and the protein levels of Ccl5 and IL6 cytokines in supernatants were quantified using
ELISA.
(D) iBMDMs expressing control or hnRNPL specific shRNA were stimulated with LPS for 5 hr, and subjected to H3K4me3 ChIP-qPCR analysis of genomic regions
around the TSSs of lincRNA-EPS target genes. **p < 0.01.
(E and F) qRT-PCR analysis of the expression of WT and lincRNA-EPS mutants following retroviral transduction of lincRNA-EPS
/
BMDMs.
(G) Predicted secondary structure of lincRNA-EPS 30 -region (2000 – 2531 nt) showing the location of CANACA tracts. The analysis was performed using RNAfold
WebServer with default parameters to generate the minimum free energy based secondary structure (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi).
Error bars indicate SD.

S8 Cell 165, 1672–1685, June 16, 2016

 total cell count CD11b+ CD11b+ Ly6Chi CD11b+ Ly6Ghi
(peritoneal lavage) myeloid cells (inflammatory monocytes) (neutrophils)
10 8 15 6
cell number (x 106)

cell number (x 105)

cell number (x 104)

cell number (x 103)

8
6
10 4
6 WT
4 KO
4
5 2
2
2

0 0 0 0
.

.
ed

ed

ed

ed
i.p

i.p

i.p

i.p
ct

ct

ct

ct
S

S
je

je

je

je
LP

LP

LP

LP
in

in

in

in
un

un

un

un
Figure S6. Recruitment of Inflammatory Cells to Peritoneum in Response to LPS Challenge In Vivo, Related to Figure 7
Flow cytometry analysis of myeloid immune cell populations in peritoneal fluid of WT and lincRNA-EPS
/
mice challenged i.p. with LPS for 5 hr. Error bars
indicate SEM.

Cell 165, 1672–1685, June 16, 2016 S9

Cell, Volume 165

Supplemental Information

A Long Noncoding RNA lincRNA-EPS

Acts as a Transcriptional
Brake to Restrain Inflammation
Maninjay K. Atianand, Wenqian Hu, Ansuman T. Satpathy, Ying Shen, Emiliano P.
Ricci, Juan R. Alvarez-Dominguez, Ankit Bhatta, Stefan A. Schattgen, Jason D.
McGowan, Juliana Blin, Joerg E. Braun, Pallavi Gandhi, Melissa J. Moore, Howard Y.
Chang, Harvey F. Lodish, Daniel R. Caffrey, and Katherine A. Fitzgerald
 Supplemental Materials for Atianand et al.

A long noncoding RNA lincRNA-EPS acts as a transcriptional brake to

restrain inflammation

1
EXTENDED EXPERIMENRAL PROCEDURES

Reagents

Reagents used in the study were obtained from following sources: Pam3CSK4, E. coli LPS, poly(I:C),

nigericin and poly(dA:dT) were purchased from Sigma-Aldrich (St. Louis, MO); recombinant murine GM-

CSF (Peprotech, Inc.), NF-κB inhibitor BAY-7082 (Tocris Bioscience), universal type I IFN (PBL assay

science) and murine recombinant TNFα (Peprotech, Inc.). Sendai virus (Cantrell strain) was purchased

from Charles River Laboratories (Wilmington, MA) and Listeria monocytogenes (clinical isolate 10403s)

was received from V. Boyartchuk (UMass, Worcester, MA). Lipofectamine 2000 was from Invitrogen

(Carlsbad, CA) and GeneJuice was from Novagen (Madison, WI). Customized nCounter gene expression

code-set was obtained from NanoString technologies (Seattle, WA).

Cell culture, stimulation, and transfection

BMDMs were generated by differentiating bone-marrow cells in DMEM supplemented with 10% fetal calf

serum (FCS), 1% Penicillin/Streptomycin (P/S) cocktail and 20% L929 supernatant for 7 days. BMDCs

were generated by differentiating bone-marrow cells in RPMI 1640 medium supplemented with 10% FCS,

1% P/S cocktail and recombinant GM-CSF (20 ng/ml) for 10 days. Immortalized BMDMs (iBMDMs) were

generated from wild-type C57BL/6 and lincRNA-EPS-/- mice using CreJ2 virus. HEK293T, iBMDMs and

J774.1 were cultured in DMEM supplemented with 10% FCS and 1% P/S under standard tissue culture

conditions. For all experiments, cells were plated one day prior to stimulation. Cells were stimulated at

following concentrations (unless mentioned otherwise): LPS (100 ng/ml), Pam3CSK4 (200 ng/ml),

poly(I:C) (25 µg/ml), type I IFN (500 U/ml), TNFα (10 ng/ml) and BAY 11-7082 (10 µM). Transfection of

iBMDMs with poly(dA:dT) (1 µg per million cells) was performed using lipofectamine 2000 (Invitrogen).

HEK293T were transfected using GeneJuice as per manufacturer’s instructions (Novagen). For Nlrp3

inflammasome assays, iBMDMs (2 x 105 cells/well in 96-well plates) were treated with LPS (100 ng/ml) for

6 hr, followed by nigericin (10 µM) treatment for 30 min, or infected with E. coli strain DH5α at a

multiplicity of infection (MOI) of 10 for 12 hr. BMDMs were infected with L. monocytogenes (5 MOI) and

Sendai virus (200 IU/ml) for 1, 2 and 6 hr for mRNA analysis.

2
lincRNA-EPS knockout mice

The lincRNA-EPS knockout mouse was generated by replacing the whole genomic locus (4 kb) encoding

lincRNA-EPS transcript with a neomycin cassette under control of a Pgk1 promoter. Specifically,

fragments of 4.5 kb upstream of lincRNA-EPS transcription start site and 1.2 kb downstream of lincRNA-

EPS transcription termination site were cloned in the PGK-neo-F2L2DTA vector, respectively. The

resulting lincRNA-EPS targeting vector was electroporated into C57BL/6 mouse embryonic stem (ES)

cells, followed by neomycin selection. Positive ES cell clones were analyzed by southern blot and

genotype PCR strategy to confirm the correct integration of the targeting vector into the lincRNA-EPS

genomic locus. Positive ES cells were injected into blastocytes to generate chimeric mice. lincRNA-EPS

heterozygous mice were obtained by gamete line transmission from mating the chimeric mice with wild-

type C57BL/6 mice.

Blood analysis

Complete blood count (CBC) analysis of ~6 weeks old wild-type and lincRNA-EPS-/- mice was performed

at the laboratory of molecular medicine of Boston Children’s Hospital, MA.

Plasmids and stable cell lines

Sequence of lincRNA-EPS (NR_131195.1; Gm12750; also known as Ttc39aos1) has been previously

described (Hu et al., 2011). lincRNA-EPS was cloned in retroviral vectors, pMSCV-PIG (Puromycin-IRES-

GFP) and pMSCV-neo (Clontech Laboratories, Inc.), for stable expressions in mammalian cells. lincRNA-

EPS was sub-cloned in pGEM-T (Promega) for performing in vitro transcription/biotinylation reactions.

lincRNA-EPS deletion mutants were generated by standard molecular biology methods. The CA tract

point mutants of the full-length lincRNA-EPS was generated by site-directed mutagenesis using Phusion

High-Fidelity DNA polymerase (New England Biolabs). pLKO.1 lentiviral vector expressing control shRNA

against GFP (RHS4459) or shRNA targeting different regions of hnRNPL (shRNA 1, TRCN0000112036;

shRNA 2, TRCN0000112038, and shRNA 3, TRCN0000112039) were obtained from GE Healthcare

3
(Dharmacon). Sequences of all expression clones were verified by DNA sequencing (Genewiz). Stable

macrophage lines were generated through transduction as described for pLKO.1 lentiviral vector

(http://www.addgene.org/tools/protocols/plko/). Briefly, HEK293T (2 x 105 cells) in 10-cm dishes were

transfected with pMSCV vector expressing lincRNA-EPS (4 µg) and packaging plasmids, Gag-Pol (1 µg)

and VSVg (1 µg), using GeneJuice as per manufacturer’s instructions (Novagen). For pLKO.1 shRNAs,

the packaging plasmids pSpax (3 µg; Addgene plasmid 12260) and pMD2 (1 µg; Addgene plasmid

12259) were used. Culture supernatants containing viral particles were collected at 48 and 72 hr post-

transfection, pooled together, and filtered through 0.4 µM nitrocellulose filter (Millipore). Cells were

transduced in the presence of Polybrene (8 µg/ml; Thermo Fisher Scientific), and selected for stable

integrants with neomycin (800 µg/ml) or puromycin (2.5 µg/ml) as bulk cultures. All stable cell lines were

used at early passages in experiments.

Chromatin Immunoprecipitation (ChIP)

BMDMs were cross-linked with 1% formaldehyde (Thermo Fisher Scientific) for 10 min at room

temperature, and the cross-linking was quenched twice with 125 mM Glycine/PBS for 5 min each. Cells

were washed twice with ice-cold PBS, and harvested by scraping. Nuclear pellets were isolated by

swelling cross-linked cells in hypotonic lysis buffer (25 mM Hepes pH 7.4, 1.5 mM MgCl2, 10 mM KCl,

0.5% NP-40 and 1 mM DTT) supplemented with 1x Halt protease inhibitor cocktail (Promega) at ice for 15

min, followed by dounce homogenization. Nuclear pellets were suspended in sonication buffer (50 mM

Hepes pH 7.4, 140 mM NaCl, 1 mM EDTA, 1% Triton-X 100, 0.1% Sodium deoxycholate, 0.5% SDS, 1

mM DTT and 1x protease inhibitor cocktail) and incubated at ice for 10 min. Nuclear extracts were

sonicated using Bioruptor UCD-200 (Diagenode Inc., Sparta, NJ) for 10 cycles of “30 sec ON and 30 sec

OFF” at the highest voltage setting to generate 200 - 500 bp chromatin fragments. In each experiment,

chromatin was first processed to confirm DNA shearing to 200 - 500 bp fragments by agarose gel

electrophoresis, and the DNA concentration was measured by NanoDrop 2000. Equal quantities of

sheared chromatin (10 µg per IP) was diluted 1:5 in sonication buffer (no SDS) to the final volume of 1 ml,

and immunoprecipitated overnight with 1 – 2.5 µg of target-specific antibodies or isotype control IgG

4
antibodies at 40 C overnight. The antibodies used in ChIP were : RNA pol II (Active Motif, clone 4H8, cat.

102660), Histone H3 (Abcam, cat. ab1791), H3K4me3 (Abcam, cat. ab8580), RNA pol II Ser2P (Abcam,

cat. ab5095), mouse IgG1 (Imgenex, cat. 20109), and rabbit polyclonal IgG (Abcam, cat. ab37415).

Chromatin complexes were captured using 20 µl Dynabeads Protein G (Invitrogen) at 40 C for 1 hr. Beads

were washed once with sonication buffer (containing 0.1% SDS), two times with high salt buffer (50 mM

Hepes pH 7.4, 500 mM NaCl, 1 mM EDTA, 1% Triton-X 100, 0.1% Sodium deoxycholate, 0.1% SDS),

two times with LiCl buffer (20 mM Tris pH 7.4, 250 mM LiCl, 1 mM EDTA, 0.5% NP-40, 0.1% Sodium

deoxycholate, 0.05% Tween-20), and once with TE buffer (10 mM Tris pH 7.4, 1 mM EDTA). Each wash

was performed at room temperature for 5 min in 1 ml volume. Beads were captured using DynaMag

magnet (Thermo Fisher Scientific). Elution was performed by suspending beads in 100 µl elution buffer

(20 mM Tris pH 7.4, 1% SDS, 50 mM NaHCO3, 1 mM EDTA). ChIP eluates were reverse cross-linked at

650 C for 4 hr, digested with Proteinase K (10 mg/ml) at 550 C for 1 hr and 2 µl RNase cocktail (Ambion)

at 370 C for 30 min. ChIP purified DNA was cleaned using PCR purification columns (Qiagen) and

subjected to qPCR analysis using primers amplifying genomic regions around the transcription start site

(TSS) of indicated genes. Sequences of primers used in ChIP studies are provided Table S4. All ChIP

results are shown as % enrichment relative to input DNA of respective experimental conditions.

Western Blot

BMDMs were lysed in buffer containing 20 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP-40 and 5 mM

EDTA supplemented with 1x Halt Protease inhibitor cocktail (Promega). Clarified lysates were resolved

on 10% SDS-PAGE, and transferred to 0.2 µM nitrocellulose (or PVDF) membrane using Turbo-Blot (Bio-

Rad). Membranes were blocked with PBS supplemented with 5% (w/v) non-fat dry milk (NFDM) for 1 hr,

and probed with primary antibodies overnight in antibody dilution buffer (PBS supplemented with 0.05%

(v/v) Tween-20 and 0.5% (w/v) NFDM). The antibodies used in western blots are: ASC (Santa Cruz

Biotechnology, sc-22514-R), pro-IL1β (R&D Systems, AF-401-NA), caspase-1 (Santa Cruz

Biotechnology, sc-514), β-actin (Sigma), Nlrp3 (Enzo Life Sciences, clone cryo-2), hnRNPL (Abcam,

clone 4D11, ab6106), hnRNP-A2/B1 (Santa Cruz Biotechnology, sc-32316) and Dnmt1 (Imgenex, IMG-

5
261A). Membranes were probed with horseradish peroxidase-conjugated anti-mouse and -rabbit

secondary antibodies (Bio-Rad; cat. 172-1011 and 170-6515). Western blots were developed using ECL

chemiluminescent substrate (Pierce).

Cytokine measurements

Cell culture supernatant, serum and tissue homogenates were assayed for cytokine levels using

commercially available sandwich ELISA kits: Cxcl10/IP-10 (R&D Systems), Ccl5/Rantes (R&D Systems),

IL6 (BD Biosciences), TNFα (eBioscience), IL-1β (BD Biosciences) and IL-1α (BD Biosciences). All

experiments for cytokine analysis by ELISA were performed in biological triplicates. Multiplex cytokine

analysis for in vivo studies was performed by LUNARISTM Mouse 12-Plex Cytokine Kit from AYOXXA

Biosystems (Cat. No. LMC-20121S, Cologne, Germany). Briefly, frozen samples (serum and peritoneal

fluid) were brought to room temperature, and diluted 1:2 using respective assay diluents. Standards

containing known concentrations of all analytes were diluted in a 5-fold dilution series. 5 µl of standard

and samples were loaded in duplicates on the AYOXXA Biochip and incubated for 3 hr at room

temperature. After washing, detection antibody mix was added and incubated for 1 hr. Following washes,

Streptavidin-Phycoerythrin (SA-PE) was added to the wells, and the AYOXXA Biochip was incubated for

additional 30 minutes. After final washing/drying, readout was performed using an Axio Imager

fluorescence microscope (Carl Zeiss, Jena, Germany). Data were analyzed using the

LUNARISTM Analysis Suite Software (Cat. No. LAS-001, AYOXXA Biosystems). In addition, some of the

samples including spleen homogenates from in vivo studies were also analyzed by Mouse

Cytokine/Chemokine Array 31-Plex (EVE Technologies Corporations, Canada). All samples for protein

measurements were harvested at indicated time points, and processed immediately or stored at -800 C.

RNA extraction and RT-qPCR

Total RNA was purified from cells and tissues using RNeasy RNA extraction kit (Qiagen) and TRIzol

reagent (Ambion) respectively as per the manufacturer’s instructions. Genomic DNA in RNA purifications

were eliminated using on-column DNase I digestion (Qiagen), or by treatment with 30U Turbo DNase

6
(Thermo Fisher Scientific) for 30 min at 370 C. RNA quality was assessed using NanoDrop 2000. Equal

amounts of RNA (250 - 1000 ng) was reverse-transcribed using iScript cDNA synthesis kit (Bio-Rad).

Diluted cDNAs (1: 100 final) were subjected to qPCR analysis using iQ SYBR Green Supermix reagent

(Bio-Rad) with the following parameters: 50°C for 2 min, 95°C for 2 min, followed by 40 cycles of 95°C for

15 sec, 60°C for 30 sec and 72°C for 45 sec, followed by melting curve analysis. Oligos used in RT-qPCR

analysis were designed using Primer3web version 4.0 (http://primer3.ut.ee/) to span either an intron (>1

kb) or the exon-exon junction. Primer sequences are provided in Table S4. Primer sequences for U1

snRNA (Li et al., 2013) and JPX (Sun et al., 2013) have been previously published. Gene expression

levels were normalized to Gapdh or β-actin as housekeeping genes. Relative mRNA expressions were

calculated by the change-in-cycling-threshold method as 2−ddC(t). Specificity of PCR amplifications was

assessed by melting curve analysis.

NanoString analysis

Cell stimulation and RNA isolation was performed as described above. The nCounter analysis system

was used for multiplex mRNA measurements using a previously described custom gene expression

code-set against 94 IRGs (Carpenter et al., 2013). Total RNA (100 ng) was hybridized overnight with the

gene expression code-set and analyzed on an nCounter Digital Analyzer (Nanostring Technologies). RNA

hybridization, data acquisition and analysis was performed as per manufacturer’s specifications. RNA

counts were processed to account for hybridization efficiency, and mRNA expressions across

experimental groups were normalized to the geometric mean of six housekeeping genes. Heatmap of

normalized mRNA counts was prepared using Gene-E with default parameters

(http://www.broadinstitute.org/cancer/software/GENE-E/).

RNA-seq Library preparation, Sequencing and Data analysis

RNA-seq was performed in biological replicates (2 mice per group) of wild-type and lincRNA-EPS-/-

BMDMs at 0, 2, and 6 hours after LPS treatment (100 ng/ml). RNA-seq libraries were prepared as

described (Heyer et al., 2015). Briefly, ribosomal RNA (rRNA) was depleted using the Ribozero rRNA

7
removal kit (Epicentre). Purified mRNAs were fragmented with RNA fragmentation reagent (Life

Technologies) for 4 minutes and 30 seconds at 70°C to obtain 100-150 nt long RNA fragments. After

ethanol precipitation and washing, RNAs were re-suspended in 5 µl of water and the 3’ ends

dephosphorylated using PNK (New England BioLabs) for 1 hr at 37°C. RNA fragments with a 3’ OH

were then ligated to a preadenylated DNA adaptor using T4 RNA ligase 2, truncated K227Q (NEB).

Following this, ligated RNAs were reverse-transcribed with Superscript III (Invitrogen) using a bar-coded

reverse-transcription primer that anneals to the preadenylated adaptor. After reverse-transcription, gel

purified cDNAs were circularized using CircLigase I (Epicentre) and PCR amplified using paired-end

primers PE1.0 and PE2.0 (Illumina) for 14 cycles. PCR amplicons were gel purified and submitted for

sequencing on the Illumina Hiseq2000. Tophat version 2.0.12 was used to align single sequence reads to

version 73 of the Ensembl mouse genome (mm10) with options: --library-type fr-firststrand -g 1 -x 1 --

read-mismatches 2. Cufflinks version 2.1.1 was used to estimate RNA abundances with Ensembl version

73 GTF and the --library-type fr-firststrand option. The Cuffdiff program was used to perform differential

expression analysis between wild-type BMDMs and lincRNA-EPS-/- BMDMs at each corresponding time

(0,2, and 6 hours after LPS treatment). To compute fold-change values for genes/transcripts with FPKM

values of zero, a pseudocount of 1 was added to all FPKM values first. All subsequent plotting and

analysis was performed on annotated genes that were specified within the Ensembl GTF file. GO

enrichment analysis was performed using the GO stats package (Falcon and Gentleman, 2007) within R.

Briefly, differentially expressed genes (e.g. ≥ 2-fold change, Q-value < 0.05 at a given time point) from

RNA-seq experiments were compared to all other genes in the genome (background model) for

significantly enriched GO terms. Genes were defined as immune genes (Figure 2C and 2E) when they

were associated with the following GO terms: GO:0045088,GO:0009607, GO:0009615, GO:0006952,

GO:0006954, GO:0032020, GO:0051707, GO:0035455, GO:0035457, GO:0006950, GO:0001866,

GO:0009611, GO:0035458, GO:0071345, GO:0070098, GO:0006955, GO:0032496, GO:0019221,

GO:0035456, GO:0045087, GO:0002376, GO:0098542. The Circos program was used to display RNA-

seq fold-change values on circularly arranged chromosomes (Krzywinski et al., 2009). The RNA-seq

sequence data is available in the ArrayExpress database under accession number (E-MTAB-4088).

8
Single-molecule RNA FISH

A. Probes for single-molecule RNA FISH: A set of 48 DNA oligonucleotides probes (20

nucleotides each) uniquely mapping along the exons of lincRNA-EPS was designed using the online

designer at http://www.singlemoleculefish.com (version 3.0), with a minimum spacing of 2 nucleotides

and a target GC content of 45%. The probe designer applies stringent cutoffs to maximize probe

specificity by masking specific regions of the mouse genome, including repetitive and low complexity

regions. The probe sequences are provided in Table S4. Probes were synthesized with an amine group

at the 3’ end (Biosearch Technologies) and coupled to Alexa Fluor 594 (Invitrogen). Fluorophore-coupled

probes were ethanol precipitated and purified on an HPLC column.

B. Single-molecule RNA FISH: Bone marrow-derived macrophages grown in chambered cover

glasses (Lab-Tek) were fixed with 1-2 ml of 3.7% (v/v) paraformaldehyde, 1x PBS for 10 minutes at room

temperature, and permeabilized by incubating at 4°C in 70% ethanol for at least 16 hrs. Single-molecule

RNA FISH was performed in the chambered cover glasses as described previously (Raj et al., 2008). For

hybridization to DNA probes, cells were rehydrated in wash buffer containing 25% (v/v) formamide and 2x

SSC for 5 mins, and 100µl of hybridization solution, containing labeled DNA probes (2 ng/µl final

concentration) in 25% (v/v) formamide, 2x SSC, 1 mg/ml BSA, 10 mM Vanadyl-ribonucleoside complex,

0.5 mg/ml E. coli tRNA and 0.1 g/ml dextran sulfate, were added to the sample and incubated overnight

at 37°C. Before imaging, cells were washed twice in 25% (v/v) formamide and 2x SSC for 30 min at 37°C,

with 5 ng/ml DAPI added for the second wash for nuclear counterstaining.

C. Image acquisition and analysis: Fluorescence microscopy, image acquisition and analysis were

conducted as previously described (Neuert et al., 2013). For imaging, 200 µl of an oxygen-scavenging

solution, containing 10 mM Tris (pH 7.5), 2x SSC and 0.4% glucose supplemented with 74 µg/ml glucose

oxidase, 74 µg/ml catalase, and 2 mM Trolox, were added to the adherent cells. Images were taken with

a Nikon TI-E inverted fluorescence microscope using a 100x oil-immersion objective, custom filters

designed to distinguish between different fluorophores, and a Photometrics Pixis 1024 CCD camera

(Princeton Instruments) managed by the MetaMorph software (Molecular Devices, Downington, PA).

9
Stacks of images were taken automatically with 0.3 µm between z-slices in the Differential Interference

Contrast (DIC), DAPI, AF594 and GFP channels. For each biological replicate, at least 5 fields of view

covering ~100 cells were imaged. For image processing, the maximum projection of DAPI image z-stacks

was merged with the DIC z-slice of maximum contrast and the composite image was used to identify

individual cells. AF594 images were compared to GFP control images to detect diffraction-limited spots

representing individual RNA molecules using fixed pixel intensity thresholds. For image presentation,

maximum-project AF594 images were merged with maximum-project DAPI images and overlaid with the

DIC z-slice of maximum contrast at 25 - 40% transparency. Enhanced contrast in the DAPI channel was

used to emphasize nuclear counterstaining boundaries.

Purification of Endogenous lincRNA-EPS

Endogenous lincRNA-EPS was purified from unstimulated BMDMs by RNA antisense purification (RAP)

technique (Engreitz et al., 2013). RAP was performed with biotinylated antisense RNA probes against

lincRNA-EPS or control Firefly luciferase RNA probes. Briefly, biotinylated RNAs were transcribed in vitro

using T7 or Sp6 polymerase (Promega) in the presence of 0.5 mM of biotin-16 UTP (Epicentre) and 0.5

mM of rATP, rCTP and rGTP. After transcription, reactions were incubated with RQ1 DNAse I (Promega)

to degrade template DNA. RNAs were ethanol-precipitated overnight, and pellets washed in 70% ethanol

before being resuspended in water. Biotinylated RNAs were then fragmented with RNA fragmentation

reagent (Life Technologies) for exactly 12 mins at 700 C. Fragmented RNAs were resolved on a 10%

acrylamide (19:1), 8 M Urea, 0.5x TBE gel for 1 hr at 35 watts. RNAs of the desired size (30 - 75 nt) were

excised from the gel, crushed, and eluted overnight in buffer containing 300 mM NaCl and 10 mM EDTA

at room temperature under nutation. Eluted RNA probes were clarified using Spin-X column at 10,000 g x

3 min, followed by RNA purification by ethanol precipitation. All steps including cell cross-linking,

chromatin shearing, hybridization and capture of RNA: chromatin complexes were performed as

described (Engreitz et al., 2013). Following RAP, purified RNA was reverse transcribed using

AffinityScript cDNA synthesis kit (Agilent). cDNA and co-purified DNA from RAP purifications were

analyzed by qPCR, and normalized to the input amount.

10
Cell Fractionation

BMDMs were fractionated into cytosolic and nuclear compartments using detergent lysis method (Tsai et

al., 2010). For isolation of chromatin-rich nuclear fractions, BMDMs were cross-linked sequentially with

formaldehyde and glutaryldehyde, followed by isolation of nuclear extracts and fragmentation using

sonication and DNase I treatment as described previously (Engreitz et al., 2013). Sheared nuclear

extracts (420 µl) were loaded on top of 5 - 30% sucrose gradient (15 mM Tris-HCl pH 7, 60 mM KCl, 2

mM EDTA and 1 mM DTT), and spun for 14 hr at 28,000 rpm using the SW-40 rotor. RNA was purified

from individual fractions using TRIzol (Ambion) and reverse transcribed with oligo-dT primers using the

cDNA synthesis kit (Agilent), and subjected to qPCR analysis. Expressions of target genes in individual

fractions were normalized to their expression level in the input RNA, which was set as 100%.

In vitro RNA-Protein binding assay and Mass Spectrometry

Linearized pGEM-T vector (Promega) expressing lincRNA-EPS was used as a template to synthesize

biotinylated lincRNA-EPS or its antisense (AS) control RNA as described above. Biotin-RNA: protein

binding assays were performed as described previously (Tsai et al., 2010). Each binding reaction was

performed with 1 µg of biotinylated RNA and the nuclear lysates prepared from BMDMs (5 million cells

per condition). Biotin-RNA: protein complexes were captured using 25 µl Streptavidin MyOne T1 beads

(Invitrogen). Proteins were eluted by heating samples at 950 C for 5 min in Laemmli buffer containing 1%

SDS. Protein eluates were resolved on 10% SDS-PAGE, and subjected to Silver Staining (Thermo Fisher

Scientific, cat. 24612). A gel slice corresponding to 50 - 75 kDa protein bands from both lincRNA-EPS

and control pulldowns were excised, and subjected to trypsin digestion followed by LC-MS/MS analysis

as previously described (Carpenter et al., 2013). Protein(s) showing at least 2-fold enrichment in lincRNA-

EPS pulldown relative to antisense RNA control were selected, and prioritized in order of their overall

abundance as indicated by total spectral counts.

RNA Immunoprecipitation (RIP)

11
Histone H3- and hnRNPL RIP experiments were performed in nuclear extracts isolated from unstimulated

BMDMs under native or formaldehyde cross-linked conditions. Assays were performed as previously

described (Tsai et al., 2010) except that cells were cross-linked with 1% formaldehyde for 10 min. Nuclear

extracts were treated with 30U DNase I (Thermo Fisher Scientific) at 370C for 20 min to solubilize DNA

prior to immunoprecipitation with 2.5 µg each of histone H3 (Abcam, cat. ab1791), hnRNPL (Abcam,

clone 4D11, ab6106) and isotype-matched control IgG antibodies overnight. RNA-protein-antibody

complexes were captured using Dynabead Protein G. RNA was eluted by adding RLT buffer directly to

magnetic beads and purified using MyOne Silane Dynabeads (Thermo Fisher Scientific) as per

manufacturer’s instructions. cDNA was synthesized using AffinityScript cDNA synthesis kit (Agilent) and

analyzed by qPCR. Results were normalized to input RNA and shown as % enrichment (Figure 4G) or

fold-enrichment over control IgG RIP (Figures 6C and 6D).

Flow Cytometry

Spleen single cell suspensions were prepared and RBCs were lysed prior to staining. Fc receptors were

blocked using supernatant from 2.4G2 hybridoma cells (anti-CD16/32) prior to staining with anti-CD3e-

PE-Cy7, anti-CD19-PerCP-Cy5.5, anti-B220-AlexaFluor 488, anti-CD11b-APC-eFluor 780, anti-CD11c-

eFluor 450, anti-NK1.1-APC (eBioscience), anti-CD4-V500, and anti-CD8a-PE (BD Biosciences). Flow

cytometry analysis of peritoneal fluid isolated from WT and lincRNA-EPS-/- mice challenged with LPS i.p.

for 5 hr was performed as described before (Sharma et al., 2015) . Data acquisition was performed using

a 4-laser LSRII (BD Biosciences). Analysis was performed using FlowJo analysis software (TreeStar).

Assay for Transposase Accessible Chromatin (ATAC-seq)

A. Cell stimulation and Transposase reaction: Wild-type and lincRNA-EPS-/- BMDMs were

stimulated with LPS (100 ng/ml) for 0, 2, or 6 hr and subjected to ATAC-seq as previously described

(Buenrostro et al., 2013). Briefly, 50,000 cells were pelleted, resuspended in 50 µL lysis buffer (10 mM

Tris-HCl, pH 7.4, 3 mM MgCl2, 10 mM NaCl, 0.1% NP-40 (Igepal CA-630)), and immediately centrifuged

at 500 g for 10 min at 40 C. The nuclei pellets were resuspended in 50 µL transposition buffer (25 µl 2X

12
TD buffer, 22.5 µL dH20, 2.5 µL Illumina Tn5 transposase), and incubated at 370 C for 30 min.

Transposed DNA was purified with MinElute PCR Purification Kit (Qiagen), and eluted in 10 µL EB buffer.

B. ATAC-seq Data Preprocessing: Barcoded ATAC-seq libraries were prepared as previously

described and sequenced on an Illumina Nextseq at the Stanford Functional Genomics Facility. Paired-

end reads were trimmed for adapter and transposase sequences using an in-house script and mapped to

mm9 using Bowtie2 v2.1.0 (Langmead et al., 2009) with parameters –very sensitive. Duplicate reads

were removed with Picard v1.79. Peak calling was performed by MACS2 (Zhang et al., 2008) narrow

peak mode with parameters –q 0.01 –nomodel –shift 0.

C. ATAC-seq Data Analysis: Overlapping peaks from all samples were merged together to

generate a consensus peak list. Unique and properly paired reads mapped to each peak for each

individual sample were quantified to calculate the Pearson Correlation. The insert size of fragments was

estimated from the distance between the pair-ended reads and plotted against the frequency in a

histogram. For TSS enrichment analysis, a 2 kb window centered on TSS was divided into 40 equal sized

bins of 50 bp. The number of unique-mapped and properly paired ATAC-seq tags overlapping each bin

was counted. The average fragment count plotted in each bin was normalized to a total of 10 millions of

reads. The per-base level coverage of ATAC-seq reads was normalized to 10 million reads and converted

to the BigWig format for visualization in IGV software (Broad Institute). The number of ATAC-seq reads

within a 2 kb window of TSS of differentially-expressed genes identified by RNA-seq was counted,

normalized by sequencing depth, and then compared between WT and lincRNA-EPS-/- BMDMs using a

two-sample t-test. The results were visualized in two scatter plots: 1) WT promoter ATAC-seq read counts

vs. lincRNA-EPS-/- promoter ATAC-seq read counts (data not shown) and 2) WT promoter ATAC-seq

counts vs. fold-change (lincRNA-EPS-/- /WT).

D. NucleoATAC analysis: Nucleosome positions and occupancy were called using NucleoATAC

software (Schep et al., 2015). The smoothed nucleosome signal in a 2 kb window around TSS was

plotted in WT and lincRNA-EPS-/- BMDMs for comparison. The distance of -1 and +1 nucleosome centers

to the TSS of differentially-expressed genes identified in RNA-seq was calculated and compared between

WT and lincRNA-EPS-/- BMDMs.

13
 E. Data Accession Codes: All ATAC-seq sequencing data have been deposited in Gene

Expression Omnibus (GEO) database under accession number GSE78873.

In vivo Endotoxin Challenge

Age- and sex-matched wild-type and lincRNA-EPS-/- mice were injected intraperitoneally (i.p.) with E. coli

LPS (5 mg/kg/mice) for 5 hr for cytokine and gene expression analysis, or with indicated doses of LPS for

survival studies. All experiments were conducted with mice maintained under specific pathogen-free

conditions in the animal facilities at UMass as per the guidelines set forth by the Institutional Animal Care

and Use Committee.

Statistics

Statistical significance of differences across two experimental groups were calculated using unpaired,

two-tailed Student’s t-test in GraphPad Prism 6.0 software. Survival data in WT and lincRNA-EPS-/- mice

following LPS challenge was analyzed using Log-rank (Mantel-Cox) test. P ≤ 0.05 was considered

statistically significant. Data are shown as mean ± SD (otherwise indicated in the figure legend) of one

representative experiment of at least three independent experiments showing similar results.

14
SUPPLEMENTARY REFERENCES:

Buenrostro, J.D., Giresi, P.G., Zaba, L.C., Chang, H.Y., and Greenleaf, W.J. (2013). Transposition of
native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and
nucleosome position. Nature methods 10, 1213-1218.

Carpenter, S., Aiello, D., Atianand, M.K., Ricci, E.P., Gandhi, P., Hall, L.L., Byron, M., Monks, B., Henry-
Bezy, M., Lawrence, J.B., et al. (2013). A long noncoding RNA mediates both activation and repression of
immune response genes. Science 341, 789-792.

Engreitz, J.M., Pandya-Jones, A., McDonel, P., Shishkin, A., Sirokman, K., Surka, C., Kadri, S., Xing, J.,
Goren, A., Lander, E.S., et al. (2013). The Xist lncRNA exploits three-dimensional genome architecture to
spread across the X chromosome. Science 341, 1237973.

Falcon, S., and Gentleman, R. (2007). Using GOstats to test gene lists for GO term association.
Bioinformatics 23, 257-258.

Heyer, E.E., Ozadam, H., Ricci, E.P., Cenik, C., and Moore, M.J. (2015). An optimized kit-free method for
making strand-specific deep sequencing libraries from RNA fragments. Nucleic acids research 43, e2.

Hu, W., Yuan, B., Flygare, J., and Lodish, H.F. (2011). Long noncoding RNA-mediated anti-apoptotic
activity in murine erythroid terminal differentiation. Genes & development 25, 2573-2578.

Krzywinski, M., Schein, J., Birol, I., Connors, J., Gascoyne, R., Horsman, D., Jones, S.J., and Marra, M.A.
(2009). Circos: an information aesthetic for comparative genomics. Genome research 19, 1639-1645.

Langmead, B., Trapnell, C., Pop, M., and Salzberg, S.L. (2009). Ultrafast and memory-efficient alignment
of short DNA sequences to the human genome. Genome biology 10, R25.

Li, L., Liu, B., Wapinski, O.L., Tsai, M.C., Qu, K., Zhang, J., Carlson, J.C., Lin, M., Fang, F., Gupta, R.A.,
et al. (2013). Targeted disruption of Hotair leads to homeotic transformation and gene derepression. Cell
reports 5, 3-12.

Neuert, G., Munsky, B., Tan, R.Z., Teytelman, L., Khammash, M., and van Oudenaarden, A. (2013).
Systematic identification of signal-activated stochastic gene regulation. Science 339, 584-587.

Raj, A., van den Bogaard, P., Rifkin, S.A., van Oudenaarden, A., and Tyagi, S. (2008). Imaging individual
mRNA molecules using multiple singly labeled probes. Nature methods 5, 877-879.

Schep, A.N., Buenrostro, J.D., Denny, S.K., Schwartz, K., Sherlock, G., and Greenleaf, W.J. (2015).
Structured nucleosome fingerprints enable high-resolution mapping of chromatin architecture within
regulatory regions. Genome research 25, 1757-1770.

Sharma, S., Campbell, A.M., Chan, J., Schattgen, S.A., Orlowski, G.M., Nayar, R., Huyler, A.H., Nundel,
K., Mohan, C., Berg, L.J., et al. (2015). Suppression of systemic autoimmunity by the innate immune
adaptor STING. Proceedings of the National Academy of Sciences of the United States of America 112,
E710-717.

15
Sun, S., Del Rosario, B.C., Szanto, A., Ogawa, Y., Jeon, Y., and Lee, J.T. (2013). Jpx RNA activates Xist
by evicting CTCF. Cell 153, 1537-1551.

Tsai, M.C., Manor, O., Wan, Y., Mosammaparast, N., Wang, J.K., Lan, F., Shi, Y., Segal, E., and Chang,
H.Y. (2010). Long noncoding RNA as modular scaffold of histone modification complexes. Science 329,
689-693.

Zhang, Y., Liu, T., Meyer, C.A., Eeckhoute, J., Johnson, D.S., Bernstein, B.E., Nusbaum, C., Myers, R.M.,
Brown, M., Li, W., et al. (2008). Model-based analysis of ChIP-Seq (MACS). Genome biology 9, R137.

16