Opposing actions of IL-2 and IL-21 on Th9differentiation correlate with their differentialregulation of BCL6 expressionWei Liao, Rosanne Spolski, Peng Li, Ning Du, Erin E. West, Min Ren, Suman Mitra, and Warren J. Leonard1

Laboratory of Molecular Immunology and the Immunology Center, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda,MD 20892-1674

Edited by Anjana Rao, Sanford Consortium for Regenerative Medicine and La Jolla Institute for Allergy and Immunology, La Jolla, CA, and approvedJanuary 27, 2014 (received for review January 18, 2013)

Interleukin 9 (IL-9) is a γc-family cytokine that is highly produced byT-helper 9 (Th9) cells and regulates a range of immune responses,including allergic inflammation. Here we show that IL-2–JAK3–STAT5 signaling is required for Th9 differentiation, with criticalSTAT5 binding sites in the Il9 (the gene encoding IL-9) promoter.IL-2 also inhibited B cell lymphoma 6 (BCL6) expression, and over-expression of BCL6 impaired Th9 differentiation. In contrast, IL-21induced BCL6 and diminished IL-9 expression in wild-type but notBcl6−/− cells, and Th9 differentiation was increased in Il21−/−

and Il21r−/− T cells. Interestingly, BCL6 bound in proximity tomany STAT5 and STAT6 binding sites, including at the Il9 pro-moter. Moreover, there was increased BCL6 and decreased STATbinding at this site in cells treated with blocking antibodies to IL-2and the IL-2 receptor, suggesting a possible BCL6–STAT5 bindingcompetition that influences IL-9 production. BCL6 binding was alsoincreased when cells were Th9-differentiated in the presence ofIL-21. Thus, our data reveal not only direct IL-2 effects via STAT5 atthe Il9 gene, but also opposing actions of IL-2 and IL-21 on BCL6expression, with increased BCL6 expression inhibiting IL-9 production.These data suggest a model in which increasing BCL6 expressiondecreases efficient Th9 differentiation, indicating possible distinctiveapproaches for controlling this process.

Tcells can differentiate into an array of specialized T-helper popu-lations, including Th1 cells, which mediate antiviral responses;

Th2 cells, which mediate host defense to parasites and allergicinflammation; and Th17 cells, which are involved in inflammatoryprocesses and diseases such as psoriasis and inflammatory boweldisease (1–5). Th9 cells are a population of cells differentiated inthe presence of IL-4 and TGF-β to secrete IL-9 and mediate al-lergic inflammation and immunity to intestinal parasites (6–9).The IL-9 receptor consists of IL-9R and the common cytokinereceptor γ chain, γc, which is shared by the receptors for IL-2,IL-4, IL-7, IL-15, and IL-21 (10) and mutated in humans withX-linked severe combined immunodeficiency (11). IL-9R is broadlyexpressed, including on hematopoietic progenitors, mast cells,macrophages, dendritic cells, B cells, airway epithelial cells, im-mature neurons, eosinophils, natural killer T (NKT) cells, naturalkiller (NK) cells, Th9 cells, Th17 cells, and Treg cells (6–9, 12, 13).This distribution helps to explain diverse actions of IL-9. IL-9increases CD4+ T-cell growth, IgE production by B cells, Tregfunction, Th17 differentiation, mast cell growth and survival, ex-pression of FceR1α, production of IL-6 by mast cells, and thematuration of hematopoietic progenitor cells (8, 9, 13). IL-9 alsoinduces the production of IL-8, IL-13, and eotaxin by airwaysmooth muscle cells and goblet cell metaplasia in airway epithelialcells (14). Recently, IL-9–producing cells have also been shown toexhibit robust antitumor immunity for melanoma (15, 16).Like IL-9, IL-2 is a type 1 four α-helical bundle cytokine

produced primarily by CD4+ T cells following antigen activation(10, 17). IL-2 signals via intermediate or high-affinity receptorscontaining IL-2Rβ and the common cytokine receptor γ chain,γc. IL-2 augments Th1 and Th2 differentiation but inhibits Th17

and TFH differentiation (18–23), and interestingly is known to beimportant for IL-9 production (24, 25), but how IL-2 regulatesTh9 differentiation and IL-9 production remains unclear.Here we provide evidence for a direct role for the IL-2–JAK3–

STAT5 signaling pathway in regulating Th9 differentiation. Wealso found that IL-2 and IL-21 have opposing roles in Th9 dif-ferentiation, with IL-2 promoting and IL-21 inhibiting formationof these cells, inversely correlating with their differential regu-lation of BCL6 expression. We also demonstrate that BCL6binds to the STAT5 and STAT6 binding region at the Il9 locus,suggesting possible competitive binding among these factors andconsistent with direct regulation of the Il9 gene by BCL6. Col-lectively, our results support a model in which there is an inverserelationship between BCL6 expression and Th9 differentiation,with cross-regulatory effects of IL-2 and IL-21.

ResultsJAK3 and STAT5 Are Important for IL-2–Induced Il9 Expression. It waspreviously shown that IL-4 + TGFβ could induce IL-9 pro-duction, but this was markedly decreased in Il2−/− T cells unlessexogenous IL-2 was added (24). We confirmed that CD4+ T cellsfrom Il2−/− mice indeed produced very little IL-9, and thatadding IL-2 increased IL-9 production by these cells, albeit notup to WT levels (Fig. 1A). Other γc cytokines, including IL-7,IL-9, and IL-15, had smaller effects on IL-9 production (Fig. 1B),even though they all, like IL-2, also activate STAT5 proteins,indicating specificity for the effect of IL-2. The requirement forIL-2 was confirmed by the ability of blocking antibodies to IL-2 +IL-2R to inhibit IL-9 production in mouse (Fig. 1C) and human

Significance

Interleukin-9 (IL-9) is a γc-family cytokine produced by Th9 cellsthat regulates a range of immune responses, including allergicinflammation. We show that IL-2 via STAT5 is required for Th9differentiation. IL-2 inhibits B cell lymphoma 6 (BCL6), whichinhibits Th9 differentiation, whereas IL-21 induces BCL6. BCL6bound near STAT5 and STAT6 binding sites, including at the Il9(gene encoding IL-9) promoter, and BCL6 binding increased andSTAT binding decreased after treatment with anti–IL-2/IL-2R.Thus, IL-2 and IL-21 have opposing actions on BCL6 expression,which inversely correlates with Th9 differentiation and IL-9 pro-duction, with implications for controlling Th9 differentiation andpotentially allergic inflammation.

Author contributions: W.L., R.S., P.L., and W.J.L. designed research; W.L., R.S., P.L., N.D.,E.E.W., M.R., and S.M. performed research; W.L., R.S., P.L., and W.J.L. analyzed data; andW.L., R.S., P.L., and W.J.L. wrote the paper.

Conflict of interest statement: W.J.L. and R.S. are inventors on patents related to IL-21.

This article is a PNAS Direct Submission.

Data deposition: The data reported in this paper have been deposited in the Gene Ex-pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE41317).1To whom correspondence should be addressed. E-mail: wjl@helix.nih.gov.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1301138111/-/DCSupplemental.

3508–3513 | PNAS | March 4, 2014 | vol. 111 | no. 9 www.pnas.org/cgi/doi/10.1073/pnas.1301138111

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

(Fig. 1D) T cells under Th9 polarizing conditions (IL-4 + TGFβ +anti-IFNγ). We next compared the kinetics of IL-2 versus IL-9production during Th9 differentiation and found that IL-2 wasproduced by day 1, increased by day 3, but then declined by day5, whereas little IL-9 was produced at day 1 but increased by day3 (Fig. 1E). Thus, IL-2 production precedes that of IL-9, con-sistent with a requirement for IL-2 early during Th9 differenti-ation. Because IL-2 is a T-cell growth factor, we investigated ifits effect on Th9 differentiation was due to an effect on pro-liferation. Cells were labeled with carboxyfluorescein succinimidylester (CFSE) and incubated under Th9 conditions for 5 d. BothWT and Il2−/− cells exhibited fairly similar CFSE dilution pro-files (Fig. 1F, Upper), with maximal IL-9 expression by division2–3 (Fig. 1F, Lower). Il9 expression in Il2−/− T cells was less sus-tained, with fewer IL-9−producing cells at each cell division (Fig.1F, Lower). Thus, IL-2 regulates IL-9 production at least partiallyindependent of its role in T-cell proliferation.Because of the importance of transcription factors PU.1 and

IRF4 for Th9 differentiation (26, 27), we investigated if IL-2affected their expression. IL-2 did not induce PU.1 mRNA overthe basal level (Fig. S1A), but it significantly enhanced Irf4 ex-pression in preactivated CD4+ T cells, whereas IL-12 and IL-21had no effect (Fig. S1B), and based on RNA-Seq analysis, Il2−/−Th9 cells had lower Irf4 mRNA expression than did WT Th9cells (Fig. S1C and Dataset S1). Although Il9 mRNA was

somewhat decreased in Irf4−/− T cells, IL-2 could still inducesubstantial Il9 expression in preactivated (Fig. S1D) and Th9-differentiated (Fig. S1E) Irf4−/− T cells, indicating that IRF4 wasnot essential for IL-2–induced Il9 expression. Correspondingly,retroviral transduction of Irf4 did not restore normal IL-9 pro-duction in Il2−/− T cells (Fig. S1F).To clarify how IL-2 regulates Th9 differentiation, we next

examined the signaling pathway(s) required for IL-9 expression.CD4+ T cells were cultured under Th9 conditions for 3 d, restedovernight, inhibitors were added 1 h before IL-2, and Il9 mRNAexpression was measured by real-time PCR 4 h later. Whereasinhibitors of MAPK (PD183161) and NF-κB (Calbiochem no.481406) increased IL-2–induced IL-9 production and a PI3-kinase inhibitor (LY294002) had no significant effect, a JAK3inhibitor (CP690550) potently inhibited IL-2–induced Il9 mRNAexpression (Fig. 2A). We next examined the ability of variouscytokines to induce Il9 mRNA in cells that were Th9-differen-tiated for 3 d and then rested overnight. Although IL-4 inducedIl9 mRNA expression, consistent with its role in driving Th9differentiation, IL-2 was more potent (Fig. 2B). IL-2 is an acti-vator of STAT5, whereas IL-4 activates STAT6. It was thus in-teresting that IL-15, which like IL-2 also activates STAT5, alsoinduced Il9 expression more potently than IL-4, albeit less po-tently than IL-2 (Fig. 2B). The weaker effect of IL-15 was con-sistent with its weaker effect than IL-2 on Th9 differentiation inIl2−/− cells (Fig. 1B). IL-21, which primarily uses STAT3, did not

A

Il2

FSC

IL-9

None IL-2

Il2+/+

F

CFSE

%IL

-9+ c

ells

Il2+/++

Il2

division

% o

f max

0

10

20

E

IL-9

IL-2

anti-IL-2+anti-IL-2RD

IL-2

IL-9

anti-IL-2+anti-IL-2RCtrl AbC

Day 1

Day 3

IL-9

IL-2

Day 5

0

10

20

30

40

50

Il2+/+ noneIL-2 IL-7 IL-9 IL-15

Il2

***

***

*NS

%IL

-9+ c

ells

Ctrl Ab

B

Fig. 1. IL-2 is important for Th9 differentiation. (A) Naive Il2−/− or WT CD4+

T cells were cultured under Th9 conditions as indicated for 3 d. IL-9 wasmeasured by intracellular staining 4 h after restimulation with PMA +ionomycin. (B) Naive Il2−/− or WT CD4+ T cells were Th9-polarized with theindicated cytokines for 3 d. IL-9 was measured as in A. (C and D) Naive CD4+

T cells from WT C57BL/6 mice (C) or sorted CD45RA+ CD25− HLA-DR− humanCD4+ T cells (D) were cultured with isotype control or antibodies to IL-2,IL-2Rα, and IL-2Rβ under Th9 conditions for 3 d. IL-2 and IL-9 were measuredas in A. (E) Naïve WT CD4+ T cells were cultured under Th9 conditions for 1,3, or 5 d. IL-2 and IL-9 were measured by intracellular staining 4 h afterrestimulation with PMA + ionomycin. (F) Naive CD4+ T cells from Il2−/− andWT mice were labeled with CFSE and cultured under Th9 conditions for 5 d.IL-9 was measured as in A. (Upper) Cell division by CFSE dilution. (Lower)Percentage of IL-9+ cells in each division. WT (solid black bars); Il2−/− (openbars). All experiments were performed three times.

A

CD

IL-9

Stat5f/f Ctrl RV Stat5f/f Cre RV

Il9/R

pl7

0.0

0.5

1.0

1.5

2.0

2.5m

ock

PD LY

NF

-B

inJA

K3

in

IL-2

****

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

none

IL-2

IL-4

IL-1

5IL

-21

Il9/R

pl7

***

% o

f IL-

9+ c

ells

0510152025

Ctrl RV Cre RV

**

Il9/R

pl7

0

2

4

6

Ctrl RV Cre RV

*

B

moc

kP

D LY

JAK

3 in

NF

-B

in

E F

Sta

t5a/

Rpl

7

Ctrl RV

CreRV

0

0.2

0.4

0.6

IL-17

Fig. 2. JAK3 and STAT5 are important for IL-2–induced Il9 expression. (Aand B) Naive WT C57BL/6 CD4+ T cells were Th9-polarized for 3 d and restedovernight. (A) Inhibitors were added 1 h before IL-2. Il9mRNA was measuredby RT-PCR 4 h after IL-2 treatment. (B) Il9 mRNA was measured 4 h after theindicated cytokine was added. Shown is the mean ± SEM from at least threeexperiments with similar results. (C–E) Naive Stat5fl/fl CD4+ T cells were Th9-polarized for 1 d, infected with Cre retrovirus, and cultured under Th9conditions for 2 d. (C) GFP+ cells were sorted and Stat5a mRNA measured byRT-PCR to confirm the deletion of Stat5; shown is one of two similarexperiments. (D and E) IL-9 and IL-17 were measured by intracellular staining4 h after restimulation of GFP+ cells with PMA + ionomycin, and shown isa representative experiment (D) as well as the mean ± SEM from twoexperiments comprising a total of seven mice (E). (F) Sorted GFP+ cells fromthe experiment in C–E were rested overnight, and Il9 mRNA measured byRT-PCR 4 h after adding IL-2. *P < 0.05; **P < 0.01.

Liao et al. PNAS | March 4, 2014 | vol. 111 | no. 9 | 3509

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

induce and in fact decreased Il9 mRNA expression (Fig. 2B). Toinvestigate the importance of STAT5, we deleted the Stat5a andStat5b genes using Cre recombinase and Stat5fl/fl T cells (Fig. 2C)and found markedly decreased IL-9–producing cells after Th9polarization (Fig. 2 D and E) as well as a decreased ability ofIL-2 to induce Il9 mRNA expression (Fig. 2F). Thus, STAT5 isnecessary for the normal increase in IL-9–producing cells duringTh9 differentiation.

GAS Motifs in the Il9 Promoter Bind STAT5 and Mediate IL-2–InducedIl9 Expression. STAT5 potentially can act directly at the Il9 lo-cus and/or have indirect effects. To investigate a direct effect,we performed ChIP-Seq experiments during Th9 polarization.Weak STAT5 ChIP-Seq peaks were observed in the Il9 gene at16 h, but strong peaks were observed after two rounds of Th9differentiation (Fig. 3A), with STAT6 binding at similar positions(Fig. 3A). We separately cloned the Il9 promoter and upstreamregions with the strongest STAT5 binding based on ChIP-Seqanalysis (see Fig. S2 for sequences) into pGL4.23, a luciferasereporter vector with a minimal promoter, and constructs weretransfected into CD4+ T cells. The upstream region did not ex-hibit activity, but the Il9 promoter was induced potently by IL-2and weakly by IL-4 but not by TGF-β, and IL-2 + IL-4 gavehigher activity than IL-2 alone (Fig. 3B). The Il9 promoter region

has two gamma interferon activation site (GAS) motif regions(Fig. 3C, Upper), and mutation of either the GAS1 or GAS2motifs diminished promoter activity, whereas mutation of bothessentially abrogated activity (Fig. 3C, Lower). Thus, IL-2 directlyinduces STAT5-dependent Il9 transcription via binding sites inthe Il9 promoter.

IL-21 Inhibits Th9 Differentiation. Above, we found that Il9 mRNAexpression was decreased by IL-21, in contrast to its markedincrease in IL-2–treated cells (Fig. 2B). Consistent with this,IL-21 potently inhibited Th9 differentiation, with fewer IL-9–producing cells (Fig. 4A). Correspondingly, Il21r−/− (Fig. 4B) andIl21−/− (Fig. 4C) T cells exhibited greater Th9 differentiation,and adding IL-21 to Il21−/− T cells markedly decreased Il9mRNA expression (Fig. 4C). Interestingly, however, IL-21 didnot diminish either anti-CD3/CD28 or anti-CD3/CD28 + IL-2–induced Il9 reporter activity (Fig. 4D); thus, the effect of IL-21was indirect rather than a direct effect on the Il9 promoter.

Opposing Actions of IL-2 and IL-21 on Th9 Differentiation Correlatewith Their Differential Regulation of BCL6 Expression. Because ofopposing actions of IL-2 and IL-21 in Th9 differentiation, wesought to identify factors that were differentially regulated bythese cytokines. By performing time-course RNA-Seq experi-ments during Th9 differentiation, we found that 906 transcrip-tion factors were expressed when we used a threshold level ofgene expression of reads per kilobase per million (RPKM) > 1 atone or more time points in naïve cells versus cells Th9-polarizedfor 1, 4, 8, 17, 24, 72, or 168 h (two rounds of polarization). Toexamine which of these factors are bound by STAT5B early inTh9 differentiation, we analyzed STAT5B ChIP-Seq data at 16 hof Th9 differentiation. We identified 283 STAT5B binding sites(peaks) within a total of 173 genes (Fig. 5A, black circle), ofwhich 22 encode transcription factors that are expressed in Th9cells (intersection of black and light gray circles; binding loca-tions of STAT5B in these genes are in Fig. 5B and Dataset S2).To determine if these genes are regulated by IL-2 and/or IL-21 inpreactivated CD4+ T cells, quantitative RT-PCR was performedfor select genes that were differentially expressed in Il2+/+ versusIl2−/− Th9 cells (Dataset S1) with proximal STAT5B sites (within5 kb of transcription start sites) and P value < 10−7 for STAT5Bbinding (Fig. 5 C–H). Of six genes tested (Runx1, Mllt6, Arid5b,Ikzf3, Ikzf4, and Bcl6), Bcl6 had the highest STAT5B binding

C

RLU

pGL4.23

0

2

4

6

8

10

TGF-

RLU

0

2

4

6

pGL4.23 WT GAS1-mut

GAS2-mut

GAS1/2-mut

TGF-

RLU

'!

A

rs1 Il9 Fbxl21

-

_-

_-

_-

_-

_

Il9 promoter

STAT5B-0 h

IgG

Il9 upstream region

Il9

STAT5B-16 h

STAT5B-2 rounds

0

450

450

450

45

STAT6-2 rounds

0

45

B

Fig. 3. GAS motifs located at the Il9 promoter region mediates IL-2–inducedIl9 expression. (A) STAT5B ChIP-Seq analysis in cells subjected to 0 h, 16 h, ortwo rounds of Th9 polarization. Also shown is STAT6 ChIP-Seq data at tworounds of Th9 polarization. Peaks in the Il9 locus are indicated. (B) Reporteractivity (mean ± SEM) of Il9 promoter or upstream regions in CD4+ T cellsTh9-polarized for 1 d. (C) WT or GAS motif mutants of the Il9 promoter (GASmotifs boxed, mutants indicated in lowercase bold) were transfected intoCD4+ T cells as in B, stimulated as indicated, and luciferase activity measured.The experiment was performed three times; shown is the mean ± SEM.

A

IL-9

IL-21

IL-17

B

WT Il21r -/-0

2

4

6

8

Il9/R

pl7

**

C

0

0.04

0.08

0.12 *

**

**

Il9/R

pl7

D

0

10

20

30

40

RLU

pGL4.23 Il9 promoter

+ + + + + + + +Anti-CD3/CD28 IL-2IL-21

Fig. 4. IL-21 inhibits IL-9 production in CD4+ T cells. (A) IL-21 induced re-pression of IL-9 expression in Th9-polarized cells. The experiment was per-formed three times. (B and C) Expression (mean ± SEM) of Il9mRNA in T cellsfrom WT versus Il21r−/− (B) or Il21−/− (C) mice. For the Il21−/− T cells, Il9 ex-pression was also examined after treatment with IL-21. (D) IL-21 did notsignificantly diminish either IL-2– or TCR- (anti-CD3 + antiCD28) induced Il9reporter activity. Shown is the mean ± SEM. *P < 0.05; **P < 0.01.

3510 | www.pnas.org/cgi/doi/10.1073/pnas.1301138111 Liao et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

intensity (tag number) (Fig. S3F). Moreover, when we examinedthe expression levels of these genes (Fig. 5 C–H), we found thatIL-2 significantly augmented expression of Arid5b (Fig. 5E) andIkzf4 (Fig. 5G) but decreased expression of Bcl6 (Fig. 5H), andIL-21 most markedly increased expression of Bcl6, indicatingopposing roles for IL-2 and IL-21 in regulating BCL6 expression,analogous to their opposing effects on Th9 differentiation. Be-cause the ability of IL-21 to induce BCL6 has been reported inother settings (28, 29), we examined the effect of IL-2 and IL-21on Bcl6 expression in Th9-polarized cells and found that IL-21increased Bcl6 (Fig. 5I) but decreased Il9 (Fig. 5J), whereas IL-2repressed Bcl6 (Fig. 5I) but increased Il9 (Fig. 5J), and when IL-2and IL-21 were combined, Bcl6 (Fig. 5I) and Il9 (Fig. 5J) ex-pression levels were similar to those observed with IL-2 alone,suggesting that IL-2 had a dominant effect over IL-21.When we examined the expression level of Bcl6 in Th1, Th2,

Th9, and Th17 cells, we found that Th9 cells had the lowestexpression (Fig. 6A). A time course analysis of Th9-differenti-ated cells inversely correlated Il9 and Bcl6 expression (Fig. 6B).Moreover, RNA-Seq and RT-PCR revealed high Il9 and verylow Bcl6 expression in Th9-polarized WT cells but very low Il9and high Bcl6 expression in Th9 polarized Il2−/− cells (Fig. 6 Cand D and Dataset S1), consistent with a negative regulatory rolefor BCL6 in Th9 differentiation. Importantly, the percentage ofBCL6+ cells dramatically increased when IL-2 signaling wasblocked with antibodies to IL-2 and IL-2R, and virtually no

BCL6+ cells coexpressed IL-9 (Fig. 6E). Moreover, retrovirallymediated overexpression of BCL6 (Fig. 6F) decreased both basaland IL-2–induced Il9 expression (Fig. 6G), indicating an effect ofBCL6 on IL-9 expression. We hypothesized that if BCL6 indeedmediated decreased IL-9 expression by IL-21, then IL-21 shouldnot diminish Th9 differentiation in the absence of BCL6. Indeed,in cells subjected to Th9 differentiation, IL-21 inhibited IL-9production by Bcl6+/+cells but not by Bcl6−/− cells (Fig. 6H).Consistent with these results, when we lowered Bcl6 expressionby introducing a Bcl6 shRNA into WT cells (Fig. 6I) we foundthat IL-21 did not diminish IL-9 production (Fig. 6 J and K).

JI

A B

0.0

0.5

1.0

1.5

Il9/R

PL7

IL-21- IL-2 IL-2+IL-21

***

*

Bcl

6/R

PL7

IL-21- IL-20

1

2

3

4

IL-2+IL-21

***

**

0

2

4

none IL-2 IL-21

0

1

2

3

none IL-2 IL-21

Ikzf

3/R

pl7

0

1

2

none IL-2 IL-21

Run

x1/R

pl7

Mllt

6/R

pl7

C D E

F G H

0

2

4

6

none IL-2 IL-21

Bcl

6/R

pl7

*

*

0

2

4

none IL-2 IL-21

Ikzf

4/R

pl7

*

0

2

4

none IL-2 IL-21

Arid

5b/R

pl7 *

Expressed TFs in Th9

Stat5B-bound genes

884

151

228

4

9

1Bcl6Ikzf4Arid5betc.

Stat5B-Bound TFs

Intron

5 UTR

3 UTR

Promoter

* *

**

Fig. 5. Differential effects of IL-2 and IL-21 on BCL6 expression in Th9-polarizedcells. (A) RNA-Seq analysis from libraries of cells subjected to Th9 polariza-tion for 1, 4, 8, 17, 24, and 72 h or two rounds of Th9 polarization. A totalof 906 transcription factors (TFs) were expressed in Th9 cells, with thresholdsdefined as gene expression levels of RPKM > 1 at one or more time points.Based on ChIP-Seq analysis of cells subjected to Th9-polarizing conditions for16 h, STAT5B bound to 22 TF-encoding genes. (B) STAT5B binding distribu-tion within the 22 TF-encoding genes; location and P value for STAT5Bbinding to these genes are shown in Dataset S2. (C–H) Naïve CD4+ T cellsfrom C57BL/6 mice were preactivated by anti-CD3/anti-CD28 for 2 d, restedovernight, and treated with IL-2 and IL-21 for 4 h. mRNA was measured byRT-PCR. The experiment was done three times; shown is the mean ± SEM.(I and J) Naive CD4+ T cells from C57BL/6 mice were cultured under Th9conditions for 3 d; rested overnight; treated with IL-2, IL-21, or both for24 h; and Bcl6 (I) or Il9 (J) mRNA assessed by RT-PCR. The experiment wasperformed twice; shown is the mean ± SEM. *P < 0.05; **P < 0.01.

A

Bcl

6 lo

g 2

(RP

KM

)

Th1Th2 Th9 Th17

C

E

IL-9

anti-IL-2+anti-IL-2RCtrl Ab

BCL6

1

2

4

8

16

DB

cl6/

Rpl

7

0.0

0.2

0.4

0.6

0.8

1.0** **

Il9/R

pl7

**

0

1

2

3

4

0

5

10

15

20

25

IL-9

-pro

duci

ng c

ell (

%)

noneIL-21

Ctrl shLuc shBcl6

*

Bcl6-/-Bcl6+/+

noneIL-21

IL-9

-pro

duci

ng c

ell (

%)

*2IL-2

0246810

Bcl6

Il2+/+

Il2

Il9

perCP-Cy5

IL-21

none

I

0

5,000

10,000

15,000

20,00020

15

10

5

0

BC

L6/R

pl7

(X10

3 )

F noneIL-2

0

10

20

30

40Il9

/Rpl

7G

noneIL-2

Il2+/+ Il2 Il2+/+ Il2

H

CtrlRV BCL6RV CtrlRV BCL6RV

KCtrl shLuc shBcl6

0 20 40 60 800

5

10

15

0.0

0.5

1.0

1.5Bcl6Il9

Il9

hrs

Bcl

6

B

0.0

0.5

1.0

1.5

2.0

Bcl

6/R

pl7

Ctrl shLuc shBcl6

noneIL-21

IL-9

J

Fig. 6. Down-regulating BCL6 is important for optimal IL-9 production. (A)Expression of Bcl6 as evaluated by RNA-Seq in cells subjected to Th1, Th2,Th9, and Th17 differentiation conditions for 3 d. (B) Bcl6 and Il9 mRNA ex-pression (RNA-Seq) at 0, 1, 4, 8, 17, 24, and 72 h of Th9 differentiation. (C andD) Expression of Bcl6 and Il9 mRNA in Il2+/+ and Il2−/− Th9 cells using RNA-Seq libraries (C) and RT-PCR (shown is the mean ± SEM from one of twosimilar experiments) (D). (E) Naive CD4+ T cells from C57BL/6 mice werecultured under Th9 conditions without or with antibodies to IL-2, IL-2Rα, andIL-2Rβ for 3 d, and IL-9 and BCL6 were measured by intracellular stainingwithout restimulation. The experiment was performed three times. (F and G)Naive CD4+ T cells from C57BL/6 mice were cultured under Th9 conditions for1 d, infected by human BCL6 or control retroviruses, and cultured under Th9conditions for 2 d. GFP+ cells were sorted, rested overnight, and treated withor without IL-2 for 4 h, and BCL6 (F) and Il9 (G) mRNAs were measured byRT-PCR. The experiment was performed three times. (H) Naive CD4+ T cellsfrom Bcl6+/+ and Bcl6−/− mice were Th9-polarized for 3 d, and IL-9 wasmeasured by intracellular staining without restimulation (mean ± SEM fromthree experiments with 2–3 mice per experiment). *P < 0.05. (I and J) NaiveCD4+ T cells from C57BL/6 mice were Th9-polarized for 1 d, infected byshBcl6 or control shLuc lentivirus, and then Th9-polarized for more 2 d.Knockdown efficiency of Bcl6 from GFP+ cells was measured by RT-PCR (I),and IL-9 production was measured in GFP+ cells by intracellular staining with-out restimulation. A representative experiment (J) and summary of threeexperiments (K) are shown. *P < 0.05.

Liao et al. PNAS | March 4, 2014 | vol. 111 | no. 9 | 3511

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

Interestingly, in both Bcl6−/− T cells or WT T cells that receivedBCL6 shRNA, the percentage of IL-9–producing cells was notincreased following Th9 differentiation. Combining this resultwith our observation that Bcl6 expression is very low in Th9 cells(Fig. 6 B–D), it appears that BCL6 can repress Th9 differenti-ation/IL-9 production, but that the lack of BCL6 per se is nota sufficient driving force to enhance Th9 differentiation.

Binding of BCL6 and STAT5/STAT6 at Similar Locations at the Il9Promoter. To investigate the mechanism(s) by which BCL6affects IL-9 production, we generated BCL6 ChIP-Seq librariesfrom Th9 cells differentiated in the presence or absence ofblocking antibodies to IL-2 and IL-2R. We found that BCL6bound to 1,157 binding sites (in 629 genes) that also boundSTAT5B and STAT6 (Fig. 7A and Dataset S3), including at theIl9 promoter (Fig. 7B). BCL6 binding increased, whereas STAT5Band STAT6 binding decreased, when blocking antibodies to IL-2and IL-2R were added (Fig. 7B and Fig. S4). Importantly, BCL6exhibited higher binding at the Il9 promoter when cells weresubjected to Th9 differentiation conditions in the presence ofIL-21 (Fig. 7C). These data suggest that BCL6 may impair IL-9production by competing with STAT5/STAT6 binding, althoughdirect competition has not yet been demonstrated, and loweringBCL6 expression allows efficient Th9 differentiation. The abil-ity of anti–IL-2/anti–IL-2R to diminish STAT5 binding wasexpected; however, its ability to diminish the binding of STAT6,which is activated by IL-4, indicates that anti–IL-2/anti–IL-2Rblockade may be inhibiting the general Th9 differentiation pro-gram, consistent with the role of IL-2 in regulating IL-4Rα ex-pression (23).

DiscussionTh9 cells are involved in allergic inflammation and immunity tointestinal parasites (6–9) and also have been reported to exhibitrobust tumor immunity to melanoma (16). IL-2 was shown toplay an important role in promoting production of IL-9 (24), butthe mechanism has been unclear. We now have shown that IL-2not only directly regulates IL-9 expression via STAT5 binding tothe Il9 promoter, but it decreases expression of BCL6. In con-trast, IL-21 induced expression of BCL6 and diminished IL-9–producing cells. These results suggest that IL-2 and IL-21, two γcfamily cytokines, serve counterbalancing roles in Th9 differenti-ation. While this study was under review, another study demon-strated that thymic stromal lymphopoietin (TSLP)-induced STAT5

also contributes to Il9 expression, providing additional insightsinto the complex control of Th9 differentiation (30).BCL6 has essential roles in B-cell and T-cell biology and is

highly expressed in germinal center B cells as well as in differ-entiating and activated B cells (31) and TFH cells (29, 32, 33). Inthis study, we propose an additional role for BCL6—namely, theability to modulate Th9 differentiation and IL-9 production. Wefound that BCL6 binds to many sites where STAT5/STAT6 canalso bind, including at the Il9 promoter, consistent with thepossibility that BCL6 might compete with STAT5 and STAT6binding and alter regulation of the corresponding genes. BCL6has also been reported to block TGFβ signaling by repressingthe transcriptional activity of SMAD3/4 (34), and SMAD3 hasbeen shown to contribute to IL-9 production (35). However, inBcl6−/− T cells and in WT T cells transfected with BCL6 shRNAand subjected to Th9 differentiation, the percentage of IL-9–producing cells was not increased, suggesting that merely low-ering BCL6 is not sufficient to induce Th9 differentiation, butdifferential regulation by IL-2 and IL-21 of BCL6 and perhapsother factors creates an environment favorable or unfavorablefor Th9 differentiation.IL-2 and IL-21 exert major actions on a range of lymphoid

populations, and in some cases they have opposing effects, forexample related to Th17 and TFH differentiation where IL-21promotes but IL-2 inhibits their differentiation (18, 21, 36, 37). Itwas suggested that competition of IL-2–induced STAT5 forSTAT3 binding sites in the Il17a gene locus mediated IL-2’s in-hibition of Il17a transcription (38), although other effects of IL-2,including its repression of gp130 and induction of T-box tran-scription factor Tbx21 (T-bet), have also been proposed (22).BCL6 has been shown to be critical for TFH differentiation andgerminal center formation, and IL-2 and IL-21 play differentroles in TFH formation by differentially regulating BCL6 (18, 21,36, 37). Our findings related to Th9 cells together with studies onTh17 and TFH cells indicate a fine regulation of T-helper celldifferentiation by IL-2 and IL-21, but in the case of Th9 cells,IL-2 instead of IL-21 promotes differentiation, and BCL6 inhibitsrather than promotes the differentiation of these cells. This re-ciprocal regulation of Th9 versus Th17 and TFH differentiationby IL-2 versus IL-21, in part via effects on BCL6, provides uniqueinsights in T-helper cell differentiation, with potential implicationsfor therapeutic approaches for allergic disease and cancer.

Materials and MethodsMouse Lymphocytes. Il2−/− mice on a Rag2−/− 5C.C7 T-cell receptor (TCR)transgenic background (line 110) and control Il2+/+ Rag2−/− 5C.C7 (line 94)mice were from the National Institute of Allergy and Infectious Diseases/National Institutes of Health (NIH) contract barrier at Taconic Farms. C57BL/6J mice were from the Jackson Lab, Stat5a/Stat5bfl/fl mice were from LotharHennighausen (National Institute of Diabetes and Digestive and KidneyDiseases, NIH, Bethesda) (39), and Bcl6−/− mice from Art Shaffer and LouisStaudt (National Cancer Institute, Bethesda) (40). Animal protocols wereapproved by the National Heart, Lung, and Blood Institute (NHLBI) AnimalCare and Use Committee and followed the NIH Guidelines “Using Animals inIntramural Research.” Naïve CD4+ T cells were purified from spleen andlymph nodes of 5–12-wk-old mice using CD4+ CD62+ T-cell Isolation Kit II(Miltenyi). Cells were cultured in RPMI medium 1640 containing 10 mMHepes, 100 mL/L FBS, 2 mM L-glutamine, and antibiotics, including 50 μM 2-ME,and activated with plate-bound anti-CD3 (2 μg/mL) and soluble anti-CD28(1 μg/mL, PharMingen) or under Th9 conditions (see Th9 Polarization).

Th9 Polarization. Mouse naïve CD4+ T cells were cultured with 2 μg/mL plate-bound anti-CD3 + 1 μg/mL soluble anti-CD28, 40 ng/mL IL-4, 2 ng/mL TGFβ/mL, and 10 μg/mL of anti-IFNγ. Mouse IL-4 and TGFβ were from PeproTech.Neutralizing antibodies to mouse IFNγ (XMG1.2), IL-2 (S4B6), IL-2Rα (PC61),and IL-2Rβ (TM-β1) were from BD Bioscience or BioLegend.

Quantitative RT-PC. RT-PCRwas performed as described (23). Primers and probeswere from Applied Biosystems, Inc. Expression levels were normalized to Rpl7.

RNA-Seq Analysis. Sequenced reads (36 bp, single end) were mapped tothe mouse genome (mm8, February 2006 Assembly) using the Illumina GA

STAT6

STAT5B

BCL6

1157

10692

424592

963

1493 1714

A

C

none IL-21

% o

f inp

ut

0

2

4

6

none IL-21IgG BCL6

STAT5B

BCL6-anti-IL-2/IL-2R

BCL6

STAT5B-anti-IL-2/IL-2R

STAT6

STAT6-anti-IL-2/IL-2R

Il9

B40

40

135

135

139

139

5 kb

Fig. 7. BCL6 and STAT5/STAT6 bind to the Il9 promoter. (A) BCL6, STAT5B,and STAT6 ChIP-Seq analysis in cells subjected to 3 d of Th9 polarization.Overlapping sites are shown in the Venn diagram. (B) BCL6, STAT5B, andSTAT6 binding to the Il9 promoter. ChIP-Seq libraries were from cells sub-jected to 3 d of Th9 polarization with or without antibodies to IL-2, IL-2Rα,and IL-2Rβ. (C) BCL6 bound to the Il9 promoter from cells subjected to 3 d ofTh9 polarization in the presence or absence of IL-21.

3512 | www.pnas.org/cgi/doi/10.1073/pnas.1301138111 Liao et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

pipeline. Mapped reads were normalized by library size (reads per million)and summed in 20 bp sliding windows that were displayed on the Universityof California, Santa Cruz (UCSC) genome browser. Digital gene expressionswere calculated by counting raw reads that fell on exons of each gene, anddifferentially expressed genes were identified using edgeR (41).

ChIP-Seq and ChIP Analysis. ChIP-Seq was performed as previously described(22, 23). Anti-STAT5B (St5B) was from R&D Systems, and anti-STAT6 (M-20,sc-981) and anti-Bcl6 (N-3, sc-858) were purchased from Santa Cruz Bio-technology. Unique sequence reads were centered on corresponding chro-matin fragments, summed in 200 bp windows, and displayed as customtracks on the UCSC genome browser. All ChIP-Seq and RNA-Seq sampleswere sequenced in the NHLBI Sequencing Core. The primers for Il9 promoterin ChIP assay were as follows: forward primer, 5′-TCACTTGACAAAGGCTG-TCTTA-3′, and reverse primer, 5′-AAACACAGACCTGGGCTTTCA.

Intracellular Staining and Flow Cytometric Analyses. Cells were fixed andpermeabilized by Cytofix/Cytoperm Buffer (BD Biosciences) after theywere restimulated with 50 ng/mL of phorbol 12-myristate 13-acetate (PMA),500 ng/mL of ionomycin, and BD GolgiPlug (BD Bioscience) for 4 h or withoutrestimulation. Cells were then stained with isotype control antibodies, orallophycocyanin (APC) anti–IL-9 (RM9A4), or Alexa Fluor 647 anti–IL-17A(eBio17B7, eBioscience), and analyzed on FACSCalibur or FACSCanto II flowcytometers (Becton Dickinson) using FlowJo software (Tree Star, Inc). For BCL6intracellular staining, the cells were fixed and permeabilized by TranscriptionFactor Buffer set (BD Biosciences) and stained with PE anti-BCL6 (K112-91,BD Pharmingen).

Reporter Assays. PCR-generated Il9 promoter and Il9 upstream region(sequences in Fig. S2) fragments were cloned 5′ of the luciferase gene inpGL4.23 and transfected into CD4+ T cells that had been cultured under Th9conditions for 1 d. Cells were rested overnight and not treated or treatedwith 100 U/mL of IL-2, 40 ng/mL of IL-4, 2 ng/mL of TGFβ, or combinations for6 h, and lysates analyzed for luciferase activity.

Retroviral Transduction Experiments. The pLZRS–BMN-1–eGFP–BCL6 humanBCL6 retroviral expression vector (provided by A. Shaffer and L. Staudt)was transfected with the pCl-ECO packaging plasmid into 293T cells. Su-pernatant was mixed with 8 μg/mL polybrene and virus introduced bycentrifugation at 2,465 × g for 45 min at 30 °C into mouse CD4+ CD62L+ Tcells that had been preactivated under Th9 conditions. The supernatantwas replaced with new medium, cells cultured as indicated for 3 d, restedovernight, and IL-2 added for 4 h. BCL6 and Il9 mRNA were measuredby RT-PCR.

Lentiviral Bcl6–shRNA Experiments. shRNA to Bcl6, 5′-gatccCCAGTTTGTGT-CACAGCAACATCTACTCGcttcctgtcagaCGAGTAGATGTTGCTGTGACACAAAC-TGGTTTTTg-3, was inserted into pGreenPuro shRNA cloning and expressionlentivector and transfected with the Lentiviral Packaging Systems (SystemBioscience) into 293T cells. Lentivirus in the supernatant was introducedinto mouse CD4+ CD62L+ T cells that had been preactivated under Th9differentiation conditions, analogous to the method used for retrovirus.Supernatant was replaced with new medium, cells cultured under Th9conditions for 3 d, and then fixed and permeablized for intracellularstaining. The percentage of cytokine-producing cells was measured bygating on live CD4+ T cells and assessed using forward versus side scatterand GFP+ staining.

Statistical Analysis. Two-tailed paired t tests were performed using Prism 4.0(GraphPad software).

ACKNOWLEDGMENTS. We thank Dr. Jian-Xin Lin (National Heart, Lung, andBlood Institute, Bethesda) for critical comments, Drs. Art Shaffer and LouisStaudt (National Cancer Institute, Bethesda) for Bcl6−/− mice and the humanBCL6 retroviral expression vector, Dr. Lothar Hennighausen (National Insti-tute of Diabetes and Digestive and Kidney Diseases, Bethesda) for theStat5a/Stat5bfl/fl mice, and the DNA Sequencing Core, NHLBI for generatingChIP-Seq and RNA-Seq libraries. This work was supported by the Division ofIntramural Research, NHLBI, National Institutes of Health.

1. Korn T, Bettelli E, Oukka M, Kuchroo VK (2009) IL-17 and Th17 cells. Annu Rev Im-munol 27:485–517.

2. Zhu J, Yamane H, Paul WE (2010) Differentiation of effector CD4 T cell populations.Annu Rev Immunol 28:445–489.

3. Ansel KM, Djuretic I, Tanasa B, Rao A (2006) Regulation of Th2 differentiation and Il4locus accessibility. Annu Rev Immunol 24:607–656.

4. Littman DR, Rudensky AY (2010) Th17 and regulatory T cells in mediating and re-straining inflammation. Cell 140(6):845–858.

5. Szabo SJ, Sullivan BM, Peng SL, Glimcher LH (2003) Molecular mechanisms regulatingTh1 immune responses. Annu Rev Immunol 21:713–758.

6. Dardalhon V, et al. (2008) IL-4 inhibits TGF-beta-induced Foxp3+ T cells and, togetherwith TGF-beta, generates IL-9+ IL-10+ Foxp3(-) effector T cells. Nat Immunol 9(12):1347–1355.

7. Veldhoen M, et al. (2008) Transforming growth factor-beta ‘reprograms’ the differ-entiation of T helper 2 cells and promotes an interleukin 9-producing subset. NatImmunol 9(12):1341–1346.

8. Noelle RJ, Nowak EC (2010) Cellular sources and immune functions of interleukin-9.Nat Rev Immunol 10(10):683–687.

9. Jabeen R, Kaplan MH (2012) The symphony of the ninth: The development andfunction of Th9 cells. Curr Opin Immunol 24(3):303–307.

10. Leonard WJ (2001) Cytokines and immunodeficiency diseases. Nat Rev Immunol 1(3):200–208.

11. Noguchi M, et al. (1993) Interleukin-2 receptor gamma chain mutation results inX-linked severe combined immunodeficiency in humans. Cell 73(1):147–157.

12. Elyaman W, et al. (2009) IL-9 induces differentiation of TH17 cells and enhancesfunction of FoxP3+ natural regulatory T cells. Proc Natl Acad Sci USA 106(31):12885–12890.

13. Li H, Rostami A (2010) IL-9: Basic biology, signaling pathways in CD4+ T cells andimplications for autoimmunity. J Neuroimmune Pharmacol 5(2):198–209.

14. Goswami R, Kaplan MH (2011) A brief history of IL-9. J Immunol 186(6):3283–3288.15. Lu Y, et al. (2012) Th9 cells promote antitumor immune responses in vivo. J Clin Invest

122(11):4160–4171.16. Purwar R, et al. (2012) Robust tumor immunity to melanoma mediated by interleukin-

9-producing T cells. Nat Med 18(8):1248–1253.17. Liao W, Lin JX, Leonard WJ (2013) Interleukin-2 at the crossroads of effector re-

sponses, tolerance, and immunotherapy. Immunity 38(1):13–25.18. Ballesteros-Tato A, et al. (2012) Interleukin-2 inhibits germinal center formation by

limiting T follicular helper cell differentiation. Immunity 36(5):847–856.19. Boyman O, Sprent J (2012) The role of interleukin-2 during homeostasis and activa-

tion of the immune system. Nat Rev Immunol 12(3):180–190.20. Cote-Sierra J, et al. (2004) Interleukin 2 plays a central role in Th2 differentiation. Proc

Natl Acad Sci USA 101(11):3880–3885.21. Laurence A, et al. (2007) Interleukin-2 signaling via STAT5 constrains T helper 17 cell

generation. Immunity 26(3):371–381.

22. Liao W, Lin JX, Wang L, Li P, Leonard WJ (2011) Modulation of cytokine receptors byIL-2 broadly regulates differentiation into helper T cell lineages. Nat Immunol 12(6):551–559.

23. Liao W, et al. (2008) Priming for T helper type 2 differentiation by interleukin2-mediated induction of interleukin 4 receptor alpha-chain expression. Nat Immunol9(11):1288–1296.

24. Schmitt E, et al. (1994) IL-9 production of naive CD4+ T cells depends on IL-2, issynergistically enhanced by a combination of TGF-beta and IL-4, and is inhibited byIFN-gamma. J Immunol 153(9):3989–3996.

25. Wilhelm C, et al. (2011) An IL-9 fate reporter demonstrates the induction of an innateIL-9 response in lung inflammation. Nat Immunol 12(11):1071–1077.

26. Chang HC, et al. (2010) The transcription factor PU.1 is required for the developmentof IL-9-producing T cells and allergic inflammation. Nat Immunol 11(6):527–534.

27. Staudt V, et al. (2010) Interferon-regulatory factor 4 is essential for the de-velopmental program of T helper 9 cells. Immunity 33(2):192–202.

28. Ozaki K, et al. (2004) Regulation of B cell differentiation and plasma cell generationby IL-21, a novel inducer of Blimp-1 and Bcl-6. J Immunol 173(9):5361–5371.

29. Nurieva RI, et al. (2009) Bcl6 mediates the development of T follicular helper cells.Science 325(5943):1001–1005.

30. Yao W, et al. (2013) Interleukin-9 is required for allergic airway inflammation me-diated by the cytokine TSLP. Immunity 38(2):360–372.

31. Rui L, Schmitz R, Ceribelli M, Staudt LM (2011) Malignant pirates of the immunesystem. Nat Immunol 12(10):933–940.

32. Crotty S (2012) The 1-1-1 fallacy. Immunol Rev 247(1):133–142.33. Johnston RJ, et al. (2009) Bcl6 and Blimp-1 are reciprocal and antagonistic regulators

of T follicular helper cell differentiation. Science 325(5943):1006–1010.34. Wang D, et al. (2008) BCL6 represses Smad signaling in transforming growth factor-

beta resistance. Cancer Res 68(3):783–789.35. Elyaman W, et al. (2012) Notch receptors and Smad3 signaling cooperate in the in-

duction of interleukin-9-producing T cells. Immunity 36(4):623–634.36. Korn T, et al. (2007) IL-21 initiates an alternative pathway to induce proinflammatory

T(H)17 cells. Nature 448(7152):484–487.37. Vogelzang A, et al. (2008) A fundamental role for interleukin-21 in the generation of

T follicular helper cells. Immunity 29(1):127–137.38. Yang XP, et al. (2011) Opposing regulation of the locus encoding IL-17 through direct,

reciprocal actions of STAT3 and STAT5. Nat Immunol 12(3):247–254.39. Cui Y, et al. (2004) Inactivation of Stat5 in mouse mammary epithelium during

pregnancy reveals distinct functions in cell proliferation, survival, and differentiation.Mol Cell Biol 24(18):8037–8047.

40. Dent AL, Shaffer AL, Yu X, Allman D, Staudt LM (1997) Control of inflammation, cytokineexpression, and germinal center formation by BCL-6. Science 276(5312):589–592.

41. Robinson MD, McCarthy DJ, Smyth GK (2010) edgeR: A Bioconductor package fordifferential expression analysis of digital gene expression data. Bioinformatics 26(1):139–140.

Liao et al. PNAS | March 4, 2014 | vol. 111 | no. 9 | 3513

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021