Immediate mediators of the inflammatory responseare poised for gene activation through RNApolymerase II stallingKaren Adelmana,1, Megan A. Kennedyb, Sergei Nechaeva, Daniel A. Gilchrista, Ginger W. Musea, Yurii Chinenovb,and Inez Rogatskyb,1

aLaboratory of Molecular Carcinogenesis, National Institute of Environmental Health Sciences, National Institutes of Health, Research Triangle Park,NC 27709; and bHospital for Special Surgery and Department of Microbiology and Immunology, Weill Medical College of Cornell University,535 East 70th Street, New York, NY 10021

Communicated by Keith R. Yamamoto, University of California, San Francisco, CA, September 9, 2009 (received for review June 17, 2009)

The kinetics and magnitude of cytokine gene expression are tightlyregulated to elicit a balanced response to pathogens and result fromintegrated changes in transcription and mRNA stability. Yet, how asingle microbial stimulus induces peak transcription of some genes(TNF�) within minutes whereas others (IP-10) require hours remainsunclear. Here, we dissect activation of several lipopolysaccharide(LPS)-inducible genes in macrophages, an essential cell type mediat-ing inflammatory response in mammals. We show that a key differ-ence between the genes is the step of the transcription cycle at whichthey are regulated. Specifically, at TNF�, RNA Polymerase II initiatestranscription in resting macrophages, but stalls near the promoteruntil LPS triggers rapid and transient release of the negative elonga-tion factor (NELF) complex and productive elongation. In contrast, noNELF or polymerase is detectible near the IP-10 promoter beforeinduction, and LPS-dependent polymerase recruitment is rate limitingfor transcription. We further demonstrate that this strategy is sharedby other immune mediators and is independent of the inducer andsignaling pathway responsible for gene activation. Finally, as astriking example of evolutionary conservation, the Drosophila ho-molog of the TNF� gene, eiger, displayed all of the hallmarks ofNELF-dependent polymerase stalling. We propose that polymerasestalling ensures the coordinated, timely activation the inflammatorygene expression program from Drosophila to mammals.

cytokine gene expression � inflammation � TNF� �transcription initiation and elongation � NELF

Innate immune responses are ancient programs that evolved toprovide protection against invading pathogens. In addition to

mobilizing professional phagocytic cells, e.g., macrophages (M�)and neutrophils, which actively prevent the dissemination of a givenmicrobial agent and create a hostile extracellular environment forits replication, the innate immune system in higher organismstriggers much slower-acting adaptive immunity to mount a responseshould the primary defenses fail to contain the infection.

In recent years our understanding of the key molecular mecha-nisms underlying innate responses in mammals has greatly ad-vanced. Host cells sense the presence of pathogens through a familyof receptors, such as Toll-like receptors (TLRs), which detectconserved microbial components including lipopolysaccharide(LPS), peptidoglycans, single-stranded (ss) and double-stranded(ds) DNA and RNA, and common bacterial surface proteins,collectively known as pathogen-associated molecular patterns(PAMPs) (1, 2). Upon activation, TLRs initiate a signaling cascadethrough adaptors and protein kinases that converge on transcrip-tional regulators nuclear factor (NF)�B, activator protein (AP)1,and interferon (IFN) regulatory factors (IRFs) that, in turn, inducetranscription of diverse proinflammatory cytokines and chemo-kines (3). These will ultimately activate M� and dendritic cells andstimulate their migration to the site of invasion.

Because the innate response is rapid and relatively nonspecific,it imposes an intrinsic threat to the host if not properly contained.

Indeed, uncontrolled production of proinflammatory cytokines,e.g., tumor necrosis factor (TNF)�, IL-1�, and IFNs, is the under-lying cause of the toxic shock syndrome. Persistent inflammation isalso a hallmark of most autoimmune diseases: TNF� has beencausally linked to joint destruction in rheumatoid arthritis, whereaspathogenesis of systemic lupus has been associated with dysregu-lated production of type I IFN (4–7).

Not surprisingly, numerous mechanisms have evolved to limit thelevels of proinflammatory cytokines. In fact, many of them directlytrigger negative feedback loops, including synthesis of anti-inflammatory cytokines such as IL-10 (8). Certain cytokine and che-mokine proteins are synthesized as proforms (IL-1�) or membrane-bound precursors (TNF�), requiring proteolysis for activation (9, 10).Secreted IL-1� can be sequestered by a decoy receptor (IL-1RII) or asoluble receptor antagonist (IL-1Ra), precluding it from signaling (10).mRNAs encoding many proinflammatory mediators are intrinsicallyunstable because of the AU-rich elements (AREs) that bind proteinslike tristetraprolin (TTP), which target the transcripts for mRNA decay(11); TNF�, IL-6, IL-1�, and IL-10 are all thought to be subject toTTP-mediated degradation (12, 13). In fact, recent genomewide anal-ysis described functionally distinct classes of proinflammatory mole-cules on the basis of their mRNA stability (14).

Finally, proinflammatory gene transcription is one of the mostsophisticated and tightly controlled gene expression programs. Indeed,DNA sequences associated with these genes are rich in binding sites fornumerous transcriptional regulators, and the roles of AP1, NF�B/Rel,and IRF families in cytokine gene expression have been studiedextensively long before these factors were identified as the effectors ofTLR signaling. A transcriptional response of a given cytokine gene maybe capped by a sequential recruitment of different regulator familymembers to the element. For example, upon sustained exposure of M�to LPS, cRel/p50 and p65/p50 heterodimers at the TNF� and CXCL2genes are replaced by p50 homodimers, which correlates with cessationof transcription and LPS tolerance (15, 16). Similarly, cofactor switchbecause of a signaling event, leading to modification of either thecofactor itself or the DNA-bound regulator responsible for its recruit-ment, can alter the rate of transcription or even reverse its polarity. Forinstance, cJun homodimers occupying the AP1 site of several proin-flammatory genes in resting M� mediate binding of the corepressorNCoR, which imposes a silent transcriptional state; exposure to LPSleads to cJun phosphorylation, heterodimerization with cFos, NCoRdismissal, and derepression (17). Recently, nucleosome positioning wasproposed to confer a particular transcriptional profile to genes withcommon chromatin structure (18). On the basis of the differential

Author contributions: K.A. and I.R. designed research; K.A., M.A.K., S.N., D.A.G., G.W.M., Y.C., andI.R. performed research; K.A. and I.R. analyzed data; and K.A. and I.R. wrote the paper.

The authors declare no conflict of interest.

1To whom correspondence may be addressed. E-mail: adelmank@niehs.nih.gov orrogatskyi@hss.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0910177106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0910177106 PNAS � October 27, 2009 � vol. 106 � no. 43 � 18207–18212

BIO

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020

requirements for Brg1/BRM and Mi-2b chromatin remodeling com-plexes, the authors described 3 types of genes encoding proinflamma-tory mediators that shared specific induction profiles under conditionstested (18).

Although the studies above explain expression patterns of indi-vidual molecules, coordinate and sharp waves of expression of largesets of proinflammatory genes in response to a common stimulusimply that a broader regulatory mechanism may be in place.Notably, events in the transcription cycle subsequent to sequence-specific factor binding, cofactor recruitment, chromatin remodel-ing, and preinitiation complex (PIC) assembly may representimportant points of regulation. In fact, recent work in Drosophilaand human cells estimates that �20% of genes are regulated afterRNA Polymerase (Pol) II recruitment to the gene promoter, bycontrolling the efficiency of early transcription elongation throughthe promoter-proximal region (19–21). Importantly, this strategyfor gene regulation was found to be enriched at stimulus-responsivegenes in Drosophila (19, 22). How widespread this regulatorycheckpoint is in the mammalian innate immune system is unknown.

To understand the patterns of transcriptional responses to TLRsignaling in M�, we examined transcription complex assembly atTNF�, the prototypic proinflammatory cytokine, and several othermediators of inflammation. Our data indicate that these genes fallinto distinct classes, depending on whether their transcription iscontrolled at the level of PIC recruitment vs. during transcriptionelongation, and that their induction profile correlates with therate-limiting step in a transcription cycle rather than signalingevents leading to their activation.

ResultsRNA Polymerase II Occupies the TNF� Proximal Promoter in M� BeforeGene Activation. TNF� gene expression is strongly induced in responseto microbial products: LPS treatment of the murine RAW264.7 M�-like cell line and of primary bone marrow-derived M� (BMM�) led toa dramatic increase in TNF� transcript (Fig. 1 A and B, Left). Unex-pectedly, chromatin immunoprecipitation (ChIP) analysis revealed asignificant RNA Pol II occupancy near the TNF� promoter under basalconditions in both cell contexts (Fig. 1 A and B, Right). This high ‘‘basal’’Pol II occupancy was focused in the promoter-proximal region; far lessPol II was detected upstream or farther downstream of the transcriptionstart site (TSS). LPS treatment triggered a significant increase in Pol IIsignal particularly within the downstream region of TNF�, suggestingthat LPS-mediated gene activation facilitates the release of Pol II fromthe promoter into the gene.

This Pol II binding pattern resembled that described for geneswhose promoters are regulated at a step subsequent to Pol IIrecruitment and are constitutively occupied by ‘‘stalled’’ or‘‘paused’’ Pol II. Pol II stalling, increasingly appreciated as awidespread regulatory checkpoint in transcription (19–21), occurswhen Pol II initiates and produces a short RNA transcript but stallsduring elongation through the promoter-proximal region.

LPS Target Genes in BMM� Display Different Patterns of Pol IIOccupancy and Phosphorylation. The C-terminal domain (CTD) of thePol II largest subunit contains a series of heptapeptide repeats(YS2PTS5PS) that can be differentially phosphorylated during thetranscription cycle. Interestingly, the stalled Poll II displays a charac-teristic phosphorylation pattern, whereby levels of Serine5 phosphor-ylation (P-S5), a marker for initiated Pol II, are high, whereas phos-phorylation levels at Serine2 (P-S2), a marker for the transition toproductive elongation, are low (23, 24). Hence, we examined CTDphosphorylation across the TNF� gene in resting and LPS-stimulatedM�.

Consistent with polymerase stalling, Pol II adjacent to the TNF� TSSwas phosphorylated on S5 in untreated BMM� [supporting informa-tion (SI) Fig. S1A], whereas P-S2 Pol II was undetectable (Fig. 2A Left).Upon LPS treatment, Pol II was strongly phosphorylated on S2 andassociated with the downstream regions of TNF�. Hence CTD phos-phorylation at the TNF� gene suggested that Pol II initiated transcrip-tion in resting cells but required an additional signal (e.g., LPS-inducedtranscription factor binding) to undergo S2 phosphorylation, be re-leased from the promoter, and enter productive elongation.

Indeed, cyclin T1, the regulatory subunit of the positive tran-scription elongation factor (P-TEFb) kinase complex responsiblefor S2 phosphorylation and release of stalled Pol II into the gene,was recruited to the TNF� proximal promoter specifically inresponse to LPS (Fig. 2A Right). A modest LPS-induced increasein cyclin T1 occupancy was also seen downstream within the TNF�gene. ChIP analysis revealed a similar LPS-dependent increase inoccupancy by Cdk9, the catalytic subunit of P-TEFb, at the TNF�promoter-proximal region (Fig. S1B).

In stark contrast to TNF�, a gene encoding an LPS-induciblechemokine IP-10 (Fig. 2B Right) was devoid of detectible Pol II inresting BMM�; Pol II recruitment to the IP-10 promoter-proximalregion and CTD phosphorylation on S5 and S2 were strictly LPSdependent (Fig. 2B Right and Fig. S1). Consistent with the data atTNF� (Fig. 2A) and previous studies (24), levels of S2 phosphor-ylation were greater on the fully elongation-competent Pol II withinthe IP-10 gene than for the Pol II beginning productive elongationnear the IP-10 promoter. Thus, IP-10 transcription appears to becontrolled at the level of LPS-induced Pol II recruitment andpreinitiation complex assembly.

0

8

16

24

32

40

-- LPS

A

UP PrProx DOWN0

20

40

60

80

-- LPS

B

PrProx DOWN0

10

20

30

40

50

60

concon+LPSPol IIPol II+LPS

RNAtranscriptinduction

RNAtranscriptinduction

0

2

4

6

8

10

Pol IIPol II+LPS

%input

ChIPTNF/45S

Fig. 1. Pol IIoccupies thepromoterof theTNF�genebeforeactivation.RAW264.7cells (A) and day 7 BMM� (B) were treated with LPS (1 �g/mL for 0.5 h and 10 ng/mLfor1h, respectively)andthe inductionofTNF� transcriptwasmeasuredbyreal-timeqPCR (Left) or cells were subjected to ChIP (Right). For RAW264.7 cells (A) ChIPmaterialprecipitatedwithantibodiestoPol IIwasquantifiedbyreal-timeqPCRusingprimerscenteredaroundtheupstream(UP),promoter-proximal(PrProx),anddown-stream(DOWN)regionsoftheTNF�geneandexpressedaspercentageofinputDNAobtainedineachsample.n�3,errorisSEM.ForBMM�(B)ChIPmaterialprecipitatedwith normal rabbit IgG (con) or equivalent amount of antibodies to Pol II was qPCRquantified with primers to PrProx and DOWN regions of the TNF� gene and nor-malized to corresponding signals at the unrelated 45S rRNA gene as an internalcontrol; thevalueof IgGcontrol inuntreatedcellswas set to1forboth locations.n�3, error is SEM.

BA

0

20

40

60

80

100

120

PrProx DOWN

concon+LPSPol IIPol II+LPSP-S2P-S2+LPS

00.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

50

100

150

200

250

300

350

UP PrProx DOWN0

concon+LPSPol IIPol II+LPSP-S2P-S2+LPS

PrProx DOWN0

4

8

12

16

20

concon+LPScycT1cycT1+LPS

ChIPTNF/45S

RNAtranscriptinductionx103

LPS,h 0 1 3

ChIPIP-10/45S

Fig. 2. TNF� but not IP-10 display the hallmarks of Pol II stalling. BMM�derived and treated as in Fig. 1B were subjected to ChIP with IgG control,antibodies to Pol II, P-S2, or cyclin T1, as indicated. Occupancy at the UP, PrProx,and DOWN regions of the TNF� gene (A) or PrProx and DOWN regions of theIP-10 gene (B Right) was assessed as in Fig. 1B. IP-10 RNA induction (B Left) inBMM� treated with LPS for 0, 1, or 3 h was assessed as in Fig. 1B.

18208 � www.pnas.org�cgi�doi�10.1073�pnas.0910177106 Adelman et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020

The TNF� Promoter Is Occupied by the Negative Elongation Factor(NELF) Complex in M�. It has been previously shown that the4-subunit NELF complex plays a key role in Pol II stalling byinhibiting early transcription elongation (25–27). The negativeeffects of NELF are counteracted by phosphorylation of the CTD,and potentially NELF itself, by the P-TEFb kinase, whose recruit-ment is concomitant with release of stalled Pol II from thepromoter region. To evaluate a possible role for NELF in regulatingTNF� transcription in vivo, we first confirmed that NELF wasexpressed in M� (Fig. S2). Furthermore, immunohistochemicalanalysis revealed abundant NELF-E staining (Fig. 3 B and E) in thenuclei of the resident M� occupying the majority of the red pulpof the mouse spleen (Fig. 3 A and D; F4/80 staining); NELF-Eexpression appeared to be particularly high in M� around andimmediately adjacent to the white pulp of the follicles (Fig. 3B;compare to Fig. 3 C and F; CD3 staining for T cells).

We next assessed NELF occupancy at several LPS target genesunder basal and inducing conditions. Consistent with the idea thatNELF inhibits early elongation, both NELF-E and NELF-A werepresent specifically in the promoter-proximal region of TNF� in restingRAW264.7 cells (Fig. 4A Right and Fig. S3A, respectively). Similarly,NELF-A associated with the TNF� promoter in untreated BMM�(Fig. 4B Right). Interestingly, in both cell contexts, NELF occupancyrapidly and transiently decreased upon addition of LPS, which inverselycorrelated with TNF� RNA induction. These data suggested that geneactivation involves the temporary dissociation of NELF from the TNF�promoter, allowing Pol II to efficiently elongate into the gene. Strik-ingly, by 1–2 h of LPS treatment, NELF resumed occupancy at theTNF� promoter (Fig. 4A and Fig. S3).

In contrast, we observed no appreciable NELF-A occupancy abovethe background of control IgG at the IP-10 gene (promoter proximallyor downstream) in either resting or LPS-treated BMM� whereas PolII was, as expected, recruited in response to LPS (Fig. 4C). In the sameexperiment NELF occupied the promoter of Eif4a1, a control genewhose transcription is LPS unresponsive, and importantly this occu-pancy was not altered in LPS-stimulated BMM� (Fig. 4C).

Permanganate Footprinting in BMM� Reveals an Open TranscriptionBubble in the TNF� Promoter-Proximal Region. The above experimentssuggested that Pol II recruited to the TNF� promoter in resting M�initiated transcription but was halted within the promoter-proximalregion by NELF. To confirm the presence of an engaged but stalled PolII at TNF�, we used in vivo permanganate footprinting, a techniquethat capitalizes on the ability of KMnO4 to specifically interact withthymine bases in ssDNA, such as those in the open transcription bubble.When Pol II stalls near the promoter for an extended period, ss regionsassociated with an engaged Pol II can be readily detected by theirKMnO4 hyperreactivity. Fig. 5 (lane ‘‘0’’) illustrates strong permanga-nate reactivity in resting BMM� from near the TNF� TSS through thepromoter-proximal region (dashes). This finding is consistent with thepresence of an open transcription bubble at the uninduced TNF� genegenerated by Pol II, which has produced a short transcript and stalledduring early transcription elongation. LPS treatment both enhanced(particularly at 0.5 h) and extended KMnO4 reactivity further down-stream into the gene (bracket), suggesting that Pol II has been releasedfrom the promoter to enter productive transcription elongation. Thisdownstream expansion of the KMnO4-reactive region is reminiscent ofthe patterns seen at the induced heat-shock genes (28), as is theappearance of detectable reactivity at the TSS (asterisk), which indi-cates a very high level of transcription initiation upon gene activation.

Pol II Stalling at the Genes in the TNF Family Is Highly EvolutionarilyConserved. Interestingly, recent studies identified eiger, the soleDrosophila member of the TNF family, as a gene whose expression

Fig. 3. NELF complex is present in mouse splenic M�. F4/80, NELF-E, and CD3immunohistochemistry in a C57BL/6 mouse spleen is shown. Serial 5-�m sections areshown with 2� (Left) and 40� (Right) magnification. (Scale bar, 100 �m.) F4/80positive cells consistent with M� represent the predominant cell populationthroughout the splenic red pulp (A and D). Note the strong and specific immuno-staining for NELF-E within cell nuclei in these M�-rich areas (B and E). CD3-positive Tcells are largely confined to the periarteriolar lymphoid sheath region of the whitepulp, but small numbers of T cells are also scattered throughout the red pulp (C andF).

0

0.5

1

1.5

2

2.53

UP PrProx DOWN

no LPSLPS 0.5hLPS 1hLPS 2h

%input

NELFA

C

B

UP PrProx DOWN0

10

20

30

40

50

60Pol II no LPSPol II LPS 0.5h

Pol II

UP PrProx DOWN

NELF

0

2

4

6

8

10con no LPScon LPS 0.5hNELF no LPSNELF LPS 0.5h

0369

12151821

UP PrProx DOWN

no LPSLPS 0.5hLPS 1hLPS 2h

Pol II

ChIPTNF/45S

08

162432404856

PrProx DOWN

Pol IIPol II+LPS

0

2

4

6

8

10

PrProx DOWN PrProx

concon+LPSNELFNELF+LPS

IP-10 IP-10 Eif

Chipnormalized

to45SRNA

Fig. 4. The NELF complex dissociates from the TNF� promoter during thepeak of gene activation. LPS-dependent dismissal of NELF from the TNF�

promoter in RAW264.7 cells (A) and primary BMM� (B) is shown. Cells werederived and treated and ChIP assays performed as in Fig. 1. Pol II and NELFoccupancy at the UP, PrProx, and DOWN regions of the TNF� gene is shown.(C) The IP-10 promoter is not occupied by the NELF complex in primary BMM�.Cells were treated with LPS for 1 h, as indicated, and Pol II and NELF-Aoccupancy at the PrProx and DOWN regions of IP-10 and the PrProx region ofthe Eif4a1 gene was assessed as in Fig. 1B.

Adelman et al. PNAS � October 27, 2009 � vol. 106 � no. 43 � 18209

BIO

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020

was attenuated in NELF-depleted cells (19, 22). Eiger, like TNF�,is an essential, stress-inducible regulator of cell death pathways (29,30), critical for protection against extracellular pathogens. Wetherefore examined whether eiger is also regulated by Pol II stalling.Our ChIP analysis in Drosophila S2 cells revealed a striking resem-blance of the Pol II and NELF-E binding profiles at eiger to thoseof Pol II and the NELF complex at the mammalian TNF� gene inM�. Indeed, both Pol II and NELF-E were strongly enriched at theeiger promoter-proximal region, but not in the upstream or down-stream regions (Fig. 6A). Furthermore, RNAi-mediated depletionof NELF significantly decreased Pol II occupancy at eiger (Fig. 6B),consistent with the role of the NELF complex in maintaining stalledPol II at this promoter.

To obtain direct evidence for the transcriptional status of the PolII bound near the eiger promoter, we performed permanganatefootprinting of the gene. As shown in Fig. 6C, Pol II detected by

ChIP is associated with a stalled, open transcription bubble in theeiger promoter-proximal region, as evidenced by KMnO4-sensitivesites at positions between �34 and �41 relative to the TSS. Furthercorroborating the idea that Pol II stalling at the eiger promoter ismediated by the NELF complex (Fig. 6B), NELF-B knockdown byRNAi decreases KMnO4 reactivity (Fig. 6C). Combined, these datademonstrate a remarkable level of evolutionary conservation of thetranscriptional regulatory mechanisms governing TNF gene expres-sion from Drosophila to mammals.

Pol II Stalling and the RNA Induction Profile in LPS-Inducible Genes inM�. In mammals, M� represent an early line of defense againstinfection that respond to invading pathogens by producing a vastnumber of proinflammatory cytokines, chemokines, and factorsthat provide negative feedback loops limiting excessive inflamma-tion. We quantified the expression of several such factors inLPS-treated M� over time. As shown in Fig. 7 A and B, TNF�nascent unprocessed transcript, as measured by the amplification ofthe intronic region, was induced with strikingly rapid kinetics,reaching a maximum in both RAW264.7 cells and BMM� by 0.5–1h and dropping off precipitously thereafter. Interestingly, thisdown-regulation of TNF� transcription coincides with NELF re-loading onto the gene (Fig. 4). Consistent with earlier observations(11), the TTP gene displayed a fast and transient induction profilesimilar to that of TNF� itself (Fig. 7 A and B). In contrast, IP-10induction occurred much more gradually and was sustained forseveral hours; an even slower induction profile was characteristic ofRANTES, another well-established LPS-inducible chemokine. Im-portantly, similarly ‘‘delayed’’ induction of IP-10 or RANTES wasseen whether mRNA or nascent transcript was analyzed.

To begin to address the potential mechanisms underlying suchdramatic differences in the transcriptional response of these 2 pairs

+24

+56+65

+100to+119

A/Gladder

nakedDNA

0 0.5 1 LPS, h

*

Fig. 5. In vivo permanganate probing of the TNF� gene in BMM� reveals apromoter-proximal open transcription bubble consistent with Pol II stalling.Day 7 BMM� were treated with LPS for the indicated times, harvested inice-cold PBS, and subjected to permanganate footprinting (see Methods).Dashes indicate the unpaired thymine bases within the area from the TSS(arrow on the left) to ��100 in untreated cells (‘‘0’’). A bracket marksadditional KMnO4 reactivity appearing from �100 to �119 in response to LPS.Note several reactive bands including that near the promoter (asterisk) thatare specifically enhanced at 0.5 h of LPS treatment and decay thereafter. A �G sequencing ladder and KMnO4 probing of purified ‘‘naked’’ DNA are shown.

%input

A

%input

C

0

0.3

0.6

0.9

1.2

1.5

UP PrProx DOWN bkg

con

Pol II (Rbp3)

NELF-B

0

0.5

1

1.5

2

2.5

3

3.5

PrProx DOWN

mockNELF-KD

0

0.1

0.2

0.3

0.4

0.5

0.6

PrProx

conmockNELF-KD

NELF Pol II

BA+G ladder

naked

DNA

MockNELF-KD

+34-36+41

Fig. 6. Eiger, the Drosophila homolog of TNF�, is a target of NELF-dependentPol II stalling. (A) Pol II and NELF occupy the promoter-proximal region of theuninduced eiger gene. ChIP was performed on Drosophila S2 cells with anti-bodies to the Rpb3 subunit of Pol II or NELF-B or with no antibody (con).Occupancy at the UP, PrProx, and DOWN regions of eiger (or an intergenicregion to assess the background of the assay, bkg) was assessed by qPCR andexpressed as percentage of input; n � 4, error is SEM. (B) Depletion of NELFreduces Pol II occupancy at the eiger promoter. S2 cells were mock treated orNELF-B depleted using RNAi (NELF KD). ChIP with no antibody (con) orantibodies to NELF-B (Left) or Pol II (Right) was performed as in A. (C) Thepermanganate footprint in the eiger promoter-proximal region is dependenton NELF. Lanes depict, left to right: the A � G ladder used to determine theposition of the promoter (arrow), the KMnO4 reactivity of naked DNA as acontrol, and KMnO4 reactivity of eiger in S2 cells (mock-treated or NELF KD).

B

A

0

8

16

24

32

40

0 1 2 3 40

3

6

9

12

15

18

0 1 2 3 40

100

200

300

400

500

600

0 1 2 3 4LPS, h:

TNFTTP

IP-10 RANTES

TTP

048

1216202428

TNF

015

30

45

60

75

90conLPS 0.5hLPS 1hLPS 3hLPS 5h

ChIPnormalizedto45S

C

TTP RANTES

0

10152025303540455055

PrProx DOWN PrProx DOWN

concon 0.5hcon 1hPol IIPol II 0.5hPol II 1h

TTP PrProx RANTES PrProx

concon 0.5hcon 1hNELFNELF 0.5hNELF 1hcycT1cycT1 0.5hcycT1 1h

RANTES

0200400600800

100012001400IP-10

0500

1000

1500

2000

25003000

50

10

1234

56789

RNAtranscript

induction

RNAtranscript

induction

D

Fig. 7. Pol II stalling correlates with rapid and transient RNA induction ofLPS-inducible genes in M�. RAW264.7 cells (A) or BMM� (B) were treated forthe indicated times with LPS, total RNA was isolated, and the expression of theindicated genes was assessed by real-time qPCR, normalized to �-actin anddefined as fold induction over that in untreated cells (set as 1). Shown is arepresentative of 2–3 independent experiments performed in duplicate. (Cand D) TTP but not RANTES promoter is occupied by stalled Pol II in M�. BMM�were treated with LPS for the indicated times and processed for ChIP as in Fig.1B. Occupancy of Pol II at the PrProx and DOWN (C) and of NELF-A and cycT1at the PrProx (D) regions of TTP and RANTES was assessed. Shown is 1 of 2independent experiments done in duplicate.

18210 � www.pnas.org�cgi�doi�10.1073�pnas.0910177106 Adelman et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020

of genes to a common stimulus, we assessed Pol II and NELFoccupancy at TTP and RANTES. Pol II was significantly enrichedat the TTP promoter-proximal region in untreated BMM� (Fig.7C); LPS treatment promoted additional Pol II loading onto thepromoter and, particularly, the downstream region of the TTPgene. In addition, NELF complex occupied the TTP promoter-proximal region in resting M�, was dismissed upon 0.5 h LPSexposure, coinciding with cycT1 recruitment, but resumed pro-moter occupancy by 1 h, as cycT1 occupancy began to decline (Fig.7D). These data were consistent with promoter-proximal Pol IIstalling and LPS-dependent release and elongation at TTP, similarto that occurring at TNF�. In contrast, RANTES was not occupiedby Pol II or NELF in untreated BMM� (Fig. 7 C and D); Pol IIloading onto the RANTES promoter was LPS dependent, indicat-ing that, as with IP-10, Pol II recruitment is a rate-limiting step inRANTES transcription. Thus, in a limited subset of genes tested,Pol II stalling correlated with a rapid and transient expressionprofile.

Pol II Stalling and Not the Nature of an Inducer Determines theInduction Profile of LPS Targets. Most cytokines and chemokines areinduced by multiple PAMPs, which signal through specific TLRs.For example, bacterial lipopeptide Pam3Cys activates TLR2,dsRNA [and its synthetic analog poly(IC)] activates TLR3, andssRNA signals through TLR7 (2). Signal transduction cascadesinitiated by these TLRs are partially overlapping yet distinct,creating a sophisticated regulatory matrix whereby, in principle,each PAMP may trigger a different transcriptional response of agiven gene. We therefore examined the induction profile of TNF�,TTP, IP-10, and RANTES in BMM� treated with poly(IC),Pam3Cys, and ssRNA. We found that TNF� was induced byPam3Cys and ssRNA and that TTP was highly ssRNA responsive,but, in all cases, their expression peaked by 0.5–1 h and subsidedthereafter (Fig. 8A). In contrast, activation of IP-10 by poly(IC) orRANTES by all 3 PAMPs did not peak until 3–5 h irrespective ofthe specific TLR engaged (Fig. 8B). We conclude that for a givengene, the kinetics of induction are specific to the gene itselfregardless of a signaling pathway responsible for activation andreflect a particular rate-limiting step in transcription (Pol II re-cruitment vs. release from promoter-proximal stalling).

DiscussionA key component of the innate immune system, myeloid cellsincluding monocytes, macrophages, and dentritic cells respond topathogens by mounting a surge of proinflammatory mediators.

Among the numerous factors that contribute to the local or systemicelevation of cytokine and chemokines levels, gene expression hasbeen established as a fundamental mechanism that dictates theirbalance. Indeed, cytokine-encoding genes are rich in cis-actingelements for numerous transcriptional regulators. The combinato-rial dynamics of many factors at a given gene are exemplified by theIFN� enhanceosome (31), where transcription initiation requiresthe ordered assembly of the IRF, ATF/AP1, and NF�B familymembers, followed by histone acetylation, nucleosome remodeling,and Pol II recruitment. Consequently, Pol II recruitment to apromoter has been traditionally considered the critical step ingetting transcription underway. Yet, recent studies have begun topaint a very different picture of transcriptional regulation, wherebya sizable fraction of promoters are constitutively occupied by Pol II,and it is the polymerase release into productive elongation that israte limiting.

The notion that the rate-limiting step in transcription does notneed to be Pol II recruitment is not in itself novel: Years ago,elegant studies from J. Lis’s group demonstrated that the Drosoph-ila heat-shock genes were regulated during early elongation (32). Asurprise, however, came from the recent genomewide studies inDrosophila revealing that up to 20% of genes may be controlled ina similar manner (19, 20). Interestingly, many genes that harborstalled Pol II encode inducible factors that respond to environ-mental or developmental triggers (19). If this trend holds true inmammals, genes with promoter-proximally stalled Pol II may beenriched in the innate immune system, which specifically evolved tosense and respond efficiently to pathogens.

Here, we demonstrate that TNF�, a broad spectrum inducer ofinflammatory responses and a key molecule in the pathogenesis ofmany autoimmune and inflammatory diseases, is encoded by astalled gene. Indeed, in resting cells, the TNF� promoter wasco-occupied by NELF and S5- (but not S2-) phosphorylated Pol IIthat was engaged in early transcription elongation. LPS treatmenttriggered P-TEFb loading, a dramatic increase in S2 phosphoryla-tion, a rapid yet transient dismissal of NELF from the promoter,and a release of Pol II into the gene.

Pol II stalling has been described for several mammalian imme-diate early genes including JunB, cFos, cMyc, and a number ofestrogen receptor-regulated genes (33–35). However, the regulateddissociation of NELF from the promoter region during geneinduction, as shown for Drosophila heat-shock genes (36), has neverbefore been reported in a mammalian system. Furthermore, thefact that Drosophila eiger is a NELF target gene, similarly controlledat the level of promoter proximal stalling, illustrates that therate-limiting step in the transcription cycle for a given gene couldbe an ancient evolutionarily conserved mechanism of regulation.

On the basis of these data, we propose the following model forthe role of Pol II stalling in the immune response: Before induction,Pol II and the transcription machinery are preloaded onto thepromoters of many genes, but are held within the promoter-proximal region by the NELF complex. This stalled elongationcomplex establishes a ‘‘poised’’ promoter structure, which mayinclude (but is likely not limited to) the active histone marks recentlyfound at such genes before induction (37). Upon immune challenge,P-TEFb recruitment triggers the rapid transition of stalled Pol II toproductive elongation and NELF release. During the early phase ofinduction, NELF binding and Pol II pausing within the promoter-proximal region would be antagonized by the continued presenceof high levels of P-TEFb. Over time, P-TEFb levels decline,allowing NELF to reassociate with Pol II and gene transcription tosubside. Notably, the decline of the TNF� and TTP transcript levelsis a compound of both a decrease in transcription levels and, aspreviously shown (14), RNA turnover. Future studies, includinggenetic disruption of mammalian NELF in the immune system, willaddress the functional role of this complex in transcriptionalregulation of the stalled genes.

RANTES

IP-10

TTP

0

6

12

18

24

.5 1 3 5 .5 1 3 5 .5 1 3 5pIC Pam3Cys ssRNA

0

6

12

18

24

0 .5 1 3 5 .5 1 3 5 .5 1 3 5pIC Pam3Cys ssRNA

0

80

160

240

320

0 .5 1 3 5 .5 1 3 5 .5 1 3 5

A BTNF

0

7

14

21

28

35

.5 1 3 5 .5 1 3 5 .5 1 3 5

RNAtranscript

induction

RNAtranscript

induction

0

0

Fig. 8. The induction profile of TLR-responsive genes in BMM� reflects therate-limiting step in transcription rather than the nature of an inducer. Day 7BMM� were treated for the indicated times with 5 �g/mL poly(IC), 100 ng/mLPam3Cys, and 10 ng/mL ssRNA, and total RNA was isolated and reversetranscribed. The expression of (A) TNF� and TTP and (B) IP-10 and RANTES wasassessed as in Fig. 7B.

Adelman et al. PNAS � October 27, 2009 � vol. 106 � no. 43 � 18211

BIO

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020

The global role of Pol II stalling in the innate immune responseis currently unknown. Conceivably, genes occupied by the stalledPol II encode acutely inducible factors in the front line of hostdefenses, while those regulated via Pol II recruitment representchemokines and mediators of tissue repair that are required laterin the infectious process. It is tempting to speculate that the intrinsicdanger imposed by potent proinflammatory cytokines such asTNF� also necessitates multiple mechanisms for a rapid shutoff,including the return of NELF to the promoter and arrest ofelongation, prompt degradation of existing mRNA, and inactiva-tion of the signaling pathway via negative feedback. GenomewideChIP-chip or ChIP-seq approaches should reveal whether Pol IIstalling indeed occurs preferentially in a specific functional groupof immune mediators.

TNF� is a prototypic inflammatory cytokine whose critical rolein the pathogenesis of several autoimmune diseases made it a primetherapeutic target in rheumatoid arthritis, psoriasis, inflammatorybowel disease, ankylosing spondylitis, and other disorders (38).Given the extraordinary importance of this cytokine in humandisease, much work has focused on dissecting the pathways respon-sible for its production. This study uncovered a fundamental,evolutionarily conserved mechanism that controls TNF� transcrip-tion in M�, which generate a bulk of TNF� during inflammatoryprocesses. While many questions regarding the molecular determi-nants of Pol II stalling remain, this early elongation checkpoint maybe subverted in disease, as suggested by recent data linking NELFdysregulation to enhanced cellular proliferation in breast cancerand gastrointestinal adenocarcinomas (39, 40). Conceivably, alteredNELF levels in M� may lead to an exaggerated inflammatoryresponse or autoimmunity. A possible correlation of an aberrantPol II stalling profile with quantitatively or qualitatively different

transcriptional response of TNF� would be of great interest topursue in mouse models of systemic inflammation or autoimmuneconditions.

MethodsCell Culture and BMM� Preparation. Mouse RAW264.7 cells were maintained inDMEM (Invitrogen) supplemented with 10% FBS (HyClone), treated with sterilePBS (vehicle) or LPS (1 �g/mL) as described in the figure legends, and harvested.BMM� were prepared from 8-wk-old C57BL/6 mice as described in ref. 41 exceptL-cell conditioned media were used for the 6-day M� expansion; cells were thenscraped, incubated in RPMI-20% FBS overnight, and treated as described in thefigure legends.

Measurement of RNA Levels. Total RNA was isolated from RAW264.7 cells andBMM� and analyzed (41) for the expression of each gene, using �-actin as aninternal control. Primer pairs are listed in Table S1.

Permanganate Footprinting. In vivo permanganate footprinting in Drosophilawas performed as in ref. 19. Day 7 BMM� (�100 � 106) were harvested in ice-coldPBS (400 � g; 5 min), washed in PBS, collected as above, and subjected to KMnO4

probing as in ref. 19 using 1 �g DNA per reaction and 25 cycles of ligation-mediated PCR (see primer sequences in Table S1).

ChIP Experiments and Immunohistochemistry. See SI Methods.

ACKNOWLEDGMENTS. We thank the members of the immunohistochemistrycore at the National Institute of Environmental Health Sciences and Drs. LionelIvashkiv and Perry Blackshear for critical comments on the manuscript. This workwas supported by the Intramural Research Program of the National Institutes ofHealth, National Institute of Environmental Health Sciences Grant Z01 ES101987(to K.A.); by the National Institutes of Health Grant T32 AR07517 (to Y.C.); and bygrants (to I.R.) from the National Institutes of Health National Institute of Allergyand Infectious Diseases (R01 AI068820), the Lupus Research Institute, and theMary Kirkland Center.

1. Kawai T, Akira S (2005) Pathogen recognition with Toll-like receptors. Curr OpinImmunol 17:338–344.

2. Barton GM, Medzhitov R (2003) Toll-like receptor signaling pathways. Science300:1524–1525.

3. Foster SL, Medzhitov R (2009) Gene-specific control of the TLR-induced inflammatoryresponse. Clin Immunol 130:7–15.

4. Bennett L, et al. (2003) Interferon and granulopoiesis signatures in systemic lupuserythematosus blood. J Exp Med 197:711–723.

5. Chu CQ, Field M, Feldmann M, Maini RN (1991) Localization of tumor necrosis factoralpha in synovial tissues and at the cartilage-pannus junction in patients with rheu-matoid arthritis. Arthritis Rheum 34:1125–1132.

6. Rothe J, et al. (1993) Mice lacking the tumour necrosis factor receptor 1 are resistantto TNF-mediated toxicity but highly susceptible to infection by Listeria monocyto-genes. Nature 364:798–802.

7. Pfeffer K, et al. (1993) Mice deficient for the 55 kd tumor necrosis factor receptor are resistantto endotoxic shock, yet succumb to L. monocytogenes infection. Cell 73:457–467.

8. Hu X, Chakravarty SD, Ivashkiv LB (2008) Regulation of interferon and Toll-like receptorsignaling during macrophage activation by opposing feedforward and feedbackinhibition mechanisms. Immunol Rev 226:41–56.

9. Horiuchi K, et al. (2007) Cutting edge: TNF-alpha-converting enzyme (TACE/ADAM17)inactivation in mouse myeloid cells prevents lethality from endotoxin shock. J Immunol179:2686–2689.

10. Dayer JM (2003) The pivotal role of interleukin-1 in the clinical manifestations ofrheumatoid arthritis. Rheumatology 42(Suppl 2):ii3-ii10.

11. Carballo E, Lai WS, Blackshear PJ (1998) Feedback inhibition of macrophage tumornecrosis factor-alpha production by tristetraprolin. Science 281:1001–1005.

12. Sandler H, Stoecklin G (2008) Control of mRNA decay by phosphorylation of tristetra-prolin. Biochem Soc Trans 36:491–496.

13. Baseggio L, Charlot C, Bienvenu J, Felman P, Salles G (2002) Tumor necrosis factor-alphamRNA stability in human peripheral blood cells after lipopolysaccharide stimulation.Eur Cytokine Netw 13:92–98.

14. Hao S, Baltimore D (2009) The stability of mRNA influences the temporal order of theinduction of genes encoding inflammatory molecules. Nat Immunol 10:281–288.

15. Carmody RJ, Ruan Q, Palmer S, Hilliard B, Chen YH (2007) Negative regulation of toll-likereceptor signaling by NF-kappaB p50 ubiquitination blockade. Science 317:675–678.

16. Chan C, Li L, McCall CE, Yoza BK (2005) Endotoxin tolerance disrupts chromatinremodeling and NF-kappaB transactivation at the IL-1beta promoter. J Immunol175:461–468.

17. Ogawa S, et al. (2004) A nuclear receptor corepressor transcriptional checkpointcontrolling activator protein 1-dependent gene networks required for macrophageactivation. Proc Natl Acad Sci USA 101:14461–14466.

18. Ramirez-Carrozzi VR, et al. (2006) Selective and antagonistic functions of SWI/SNF andMi-2beta nucleosome remodeling complexes during an inflammatory response. GenesDev 20:282–296.

19. Muse GW, et al. (2007) RNA polymerase is poised for activation across the genome. NatGenet 39:1507–1511.

20. Zeitlinger J, et al. (2007) RNA polymerase stalling at developmental control genes inthe Drosophila melanogaster embryo. Nat Genet 39:1512–1516.

21. Guenther MG, Levine SS, Boyer LA, Jaenisch R, Young RA (2007) A chromatin landmarkand transcription initiation at most promoters in human cells. Cell 130:77–88.

22. Gilchrist DA, et al. (2008) NELF-mediated stalling of Pol II can enhance gene expressionby blocking promoter-proximal nucleosome assembly. Genes Dev 22:1921–1933.

23. Buratowski S (2003) The CTD code. Nat Struct Biol 10:679–680.24. Peterlin BM, Price DH (2006) Controlling the elongation phase of transcription with

P-TEFb. Mol Cell 23:297–305.25. Yamaguchi Y, et al. (1999) NELF, a multisubunit complex containing RD, cooperates

with DSIF to repress RNA polymerase II elongation. Cell 97:41–51.26. Renner DB, Yamaguchi Y, Wada T, Handa H, Price DH (2001) A highly purified RNA

polymerase II elongation control system. J Biol Chem 276:42601–42609.27. Cheng B, Price DH (2007) Properties of RNA polymerase II elongation complexes before and

after the P-TEFb-mediated transition into productive elongation. J Biol Chem 282:21901–21912.

28. Giardina C, Perez-Riba M, Lis JT (1992) Promoter melting and TFIID complexes onDrosophila genes in vivo. Genes Dev 6:2190–2200.

29. Moreno E, Yan M, Basler K (2002) Evolution of TNF signaling mechanisms: JNK-dependent apoptosis triggered by Eiger, the Drosophila homolog of the TNF super-family. Curr Biol 12:1263–1268.

30. Igaki T, et al. (2002) Eiger, a TNF superfamily ligand that triggers the Drosophila JNKpathway. EMBO J 21:3009–3018.

31. Panne D, Maniatis T, Harrison SC (2007) An atomic model of the interferon-betaenhanceosome. Cell 129:1111–1123.

32. Lis J (1998) Promoter-associated pausing in promoter architecture and postinitiationtranscriptional regulation. Cold Spring Harbor Symp Quant Biol 63:347–356.

33. Krumm A, Hickey LB, Groudine M (1995) Promoter-proximal pausing of RNA polymer-ase II defines a general rate-limiting step after transcription initiation. Genes Dev9:559–572.

34. Aida M, et al. (2006) Transcriptional pausing caused by NELF plays a dual role inregulating immediate-early expression of the junB gene. Mol Cell Biol 26:6094–6104.

35. Kininis M, Isaacs GD, Core LJ, Hah N, Kraus WL (2009) Postrecruitment regulation ofRNA polymerase II directs rapid signaling responses at the promoters of estrogentarget genes. Mol Cell Biol 29:1123–1133.

36. Wu CH, et al. (2005) Molecular characterization of Drosophila NELF. Nucleic Acids Res33:1269–1279.

37. Hargreaves DC, Horng T, Medzhitov R (2009) Control of inducible gene expression bysignal-dependent transcriptional elongation. Cell 138:129–145.

38. Feldmann M, Maini RN (2003) Lasker Clinical Medical Research Award. TNF defined asa therapeutic target for rheumatoid arthritis and other autoimmune diseases. Nat Med9:1245–1250.

39. Aiyar SE, et al. (2004) Attenuation of estrogen receptor alpha-mediated transcriptionthrough estrogen-stimulated recruitment of a negative elongation factor. Genes Dev18:2134–2146.

40. McChesney PA, et al. (2006) Cofactor of BRCA1: A novel transcription factor regulatorin upper gastrointestinal adenocarcinomas. Cancer Res 66:1346–1353.

41. Reily MM, Pantoja C, Hu X, Chinenov Y, Rogatsky I (2006) The GRIP1:IRF3 interaction as atarget for glucocorticoid receptor-mediated immunosuppression. EMBO J 25:108–117.

42. Rogatsky I, Zarember KA, Yamamoto KR (2001) Factor recruitment and TIF2/GRIP1corepressor activity at a collagenase-3 response element that mediates regulation byphorbol esters and hormones. EMBO J 20:6071–6083.

18212 � www.pnas.org�cgi�doi�10.1073�pnas.0910177106 Adelman et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020