transmembrane tnf-α promotes suppressive activities of ... · the journal of immunology...

13
of April 13, 2017. This information is current as Suppressor Cells via TNFR2 Suppressive Activities of Myeloid-Derived Promotes α Transmembrane TNF- Feng, Zhihai Qin, Bingjiao Yin and Zhuoya Li Guohong Lin, Min Yu, Jing Wang, Xiaodan Jiang, Wei Xin Hu, Baihua Li, Xiaoyan Li, Xianxian Zhao, Lin Wan, http://www.jimmunol.org/content/192/3/1320 doi: 10.4049/jimmunol.1203195 December 2013; 2014; 192:1320-1331; Prepublished online 30 J Immunol References http://www.jimmunol.org/content/192/3/1320.full#ref-list-1 , 31 of which you can access for free at: cites 56 articles This article Subscription http://jimmunol.org/subscription is online at: The Journal of Immunology Information about subscribing to Permissions http://www.aai.org/About/Publications/JI/copyright.html Submit copyright permission requests at: Email Alerts http://jimmunol.org/alerts Receive free email-alerts when new articles cite this article. Sign up at: Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved. Copyright © 2014 by The American Association of 1451 Rockville Pike, Suite 650, Rockville, MD 20852 The American Association of Immunologists, Inc., is published twice each month by The Journal of Immunology by guest on April 13, 2017 http://www.jimmunol.org/ Downloaded from by guest on April 13, 2017 http://www.jimmunol.org/ Downloaded from

Upload: others

Post on 26-Feb-2020

9 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

of April 13, 2017.This information is current as

Suppressor Cells via TNFR2Suppressive Activities of Myeloid-Derived

PromotesαTransmembrane TNF-

Feng, Zhihai Qin, Bingjiao Yin and Zhuoya LiGuohong Lin, Min Yu, Jing Wang, Xiaodan Jiang, Wei Xin Hu, Baihua Li, Xiaoyan Li, Xianxian Zhao, Lin Wan,

http://www.jimmunol.org/content/192/3/1320doi: 10.4049/jimmunol.1203195December 2013;

2014; 192:1320-1331; Prepublished online 30J Immunol 

Referenceshttp://www.jimmunol.org/content/192/3/1320.full#ref-list-1

, 31 of which you can access for free at: cites 56 articlesThis article

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved.Copyright © 2014 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 2: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

The Journal of Immunology

Transmembrane TNF-a Promotes Suppressive Activities ofMyeloid-Derived Suppressor Cells via TNFR2

Xin Hu,*,1 Baihua Li,*,1 Xiaoyan Li,* Xianxian Zhao,* Lin Wan,* Guohong Lin,*

Min Yu,* Jing Wang,* Xiaodan Jiang,* Wei Feng,* Zhihai Qin,† Bingjiao Yin,* and

Zhuoya Li*

It has been reported that TNFR2 is involved in regulatory T cell induction and myeloid-derived suppressor cell (MDSC) accumu-

lation, two kinds of immunosuppressive cells contributing to tumor immune evasion. Because transmembrane TNF-a (tmTNF-a) is

the primary ligand for TNFR2, we hypothesized that tmTNF-a is mainly responsible for the activation of MDSCs. Indeed, we

found that tmTNF-a, rather than secretory TNF-a (sTNF-a), activated MDSCs with enhanced suppressive activities, including

upregulating arginase-1 and inducible NO synthase transcription, promoting secretion of NO, reactive oxygen species, IL-10, and

TGF-b, and enhancing inhibition of lymphocyte proliferation. This effect of tmTNF-a was mediated by TNFR2, as TNFR2

deficiency significantly impaired tmTNF-a–induced release of IL-10 and NO and inhibition of T cell proliferation by MDSC

supernatant. Furthermore, tmTNF-a caused p38 phosphorylation and NF-kB activation, whereas inhibition of NF-kB or p38 with

an inhibitor pyrrolidine dithiocarbamate or SB203580 abrogated tmTNF-a–mediated increased suppression of lymphocyte pro-

liferation by MDSCs. Consistently, our in vivo study showed that ectopic expression of uncleavable tmTNF-a mutant by 4T1 cells

significantly promoted tumor progression and angiogenesis, accompanied with more accumulation of MDSCs and regulatory

T cells in the tumor site, increased production of NO, IL-10, and TGF-b, as well as poor lymphocyte infiltration. In contrast,

enforced expression of sTNF-a mutant by 4T1 cells that only released sTNF-a without expression of surface tmTNF-a markedly

reduced MDSC accumulation and induced more lymphocyte infiltration instead, showing obvious tumor regression. Our data

suggest that tmTNF-a acts as a potent activator of MDSCs via TNFR2 and reveals another novel immunosuppressive effect of this

membrane molecule that promotes tumor immune escape. The Journal of Immunology, 2014, 192: 1320–1331.

Tumor necrosis factor-a exists in two biologically activeforms, transmembrane TNF-a (tmTNF-a) and secretoryTNF-a (sTNF-a). TNF-a is first synthesized as a 26-kDa

transmembrane molecule and can be cleft by the TNF-a–con-verting enzyme, releasing a 17-kDa soluble cytokine. TNF-a ex-ecutes its biological effects through two distinct receptors, TNFR1(p55/p60) and TNFR2 (p75/p80). Although tmTNF-a and sTNF-abind to the same TNFRs, their different biological effects indicatethat both forms of TNF-a trigger distinct signal pathways (1, 2).TNF-a can be produced by tumor and stromal cells in breast,ovarian, colorectal, esophageal, prostate, bladder, and renal cellcarcinomas, melanoma, and hematological malignancies (3). In-

creasing evidence demonstrates that TNF-a acts as a tumor-promoting factor, linking inflammation and cancer. It has beenreported that TNF-a2/2 mice were unexpectedly resistant to skincarcinogenesis (4), and TNF-a in the tumor microenvironmentpromotes tumor survival, proliferation, invasion, metastasis, andangiogenesis (5). Additionally, TNF blockade reduced tumorgrowth as well as decreased colonic infiltration of macrophagesand neutrophils in mice (6), and neutralization of TNF-a withspecific Ab suppressed the hepatic metastases and prolongedsurvival (7, 8).In contrast to the promoting activities of sTNF-a for the tumor

development, the role of tmTNF-a is poorly understood. Ourprevious results showed that tmTNF-a expressed by tumor cellsconfers the resistance to apoptosis induced by sTNF-a through itsreverse signaling (9, 10). Recently it has been reported that in thepresence of IL-2, TNF-a selectively activates regulatory T cells(Tregs), leading to proliferation, upregulation of Foxp3 expres-sion, and enhancement of their suppressive activity in a TNFR2-dependent manner (11). Further study demonstrated that Treginduction involves tmTNF-a but not sTNF-a (12). BecausetmTNF-a is the primary ligand for TNFR2 (13), it is possible thatproinflammatory effects of sTNF-a through TNFR1 (14, 15) syn-ergize with the immunosuppressive effect of tmTNF-a via TNFR2to promote tumor development and immune escape.Myeloid-derived suppressor cells (MDSCs) have been reported

to facilitate tumor immune escape. This population of cells ismarkedly increased in tumor-bearing hosts and suppresses anti-tumor immune response through multiple mechanisms (16–18),including high activity of arginase-1 (ARG1) and/or inducible NOsynthase (iNOS), production of NO and reactive oxygen species(ROS), as well as release of IL-10 and TGF-b. In mice, the cell

*Department of Immunology, Tongji Medical College, Huazhong University of Scienceand Technology, Wuhan 430030, People’s Republic of China; and †National Laboratoryof Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing100730, People’s Republic of China

1X.H. and B.L. contributed equally to this work.

Received for publication November 19, 2012. Accepted for publication November28, 2013.

This work was supported by National Natural Science Foundation of China Grant30771976 and by the Major Research Plan of the National Natural Science Founda-tion of China (Grant 91029709).

Address correspondence and reprint requests to Zhuoya Li and Bingjiao Yin, Depart-ment of Immunology, Tongji Medical College, Huazhong University of Science andTechnology, 13 Hangkong Road, Wuhan 430030, People’s Republic of China. E-mailaddresses: [email protected] (Z.L.) and [email protected] (B.Y.)

Abbreviations used in this article: ARG1, arginase-1; DCFDA, 29,79-dichlorofluor-escein diacetate; iNOS, inducible NO synthase; KO, knockout; MDSC, myeloid-derived suppressor cell; PDTC, pyrrolidine dithiocarbamate; ROS, reactive oxygenspecies; sTNF-a, secretory TNF-a; tmTNF-a, transmembrane TNF-a; Treg, regula-tory T cell; wtTNF-a, wild-type TNF-a.

Copyright� 2014 by TheAmericanAssociation of Immunologists, Inc. 0022-1767/14/$16.00

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1203195

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 3: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

surface phenotype of MDSCs is identified as CD11b+ and Gr-1+.The expansion of MDSCs can be induced by the stimulatory factorsof myelopoiesis such as GM-CSF, stem cell factor, and vascularendothelial growth factor (19–21), released primarily by tumor cells,and the activation of MDSCs is associated with proinflammatorymediators, including IL-1b, IL-6, PGE2, and S100A8/9 proteins (22–25), produced by tumor cells, tumor stroma, and tumor-infiltratingimmune cells. For example, IL-1b has been reported to promotetumor progression by increasing the accumulation of MDSCs (22),whereas IL-1R–deficient mice have a delayed accumulation ofMDSCs and reduced tumor progression.TNF-a is an important proinflammatory cytokine, and a more

recent report (26) showed that TNFR deficiency impairs MDSCaccumulation by promoting apoptosis, suggesting a requirementof TNF-a–mediated signaling for MDSC survival and accumu-lation. However, which form of TNF-a exerts effects and whetherTNF-a affects MDSC functions are still unknown. We hypothe-sized that tmTNF-a contributes to activation of MDSCs. Indeed,we found that tmTNF-a, but not sTNF-a, was able to activateMDSCs with enhanced suppressive activities via TNFR2 andpromoted tumor progression. Our study reveals a novel immu-nosuppressive effect of this membrane molecule.

Materials and MethodsCell lines

The 4T1mammary carcinoma cell line (BALB/c background) was providedby Prof. Zhihai Qin (National Laboratory of Biomacromolecules, Instituteof Biophysics, Chinese Academy of Sciences, Beijing, China) and culturedin DMEM (Life Technologies/Invitrogen) supplemented with 10% FCS(Sijiqing, Hangzhou, China), 100 U/ml penicillin, and 100 mg/ml strep-tomycin.

Plasmids and transfer

pTriEx-4C vector is a shuttle plasmid in which the mouse cDNA (NM-013693.2) for wild-type TNF (wtTNF-a) was inserted at EcoRI andBamHI cloning sites. Δ1–12TNF cDNA encoding a mutant transmembraneTNF-a molecule, which has 12 aa (residues 1–12) deleted to prevent thecleavage of the transmembrane moleclue into the secretory form, was alsocloned into pTriEx-4C. Similarly cloned into pTriEx-4C was sTNF-amutant, which has the region coding for the TNF-a signal peptide (aminoacids 276 to 21) replaced with a sequence coding for the human Ig k-chainsecretion signal peptide (59-ATGGAGACAGACACACTCCTGCTATGG-GTACTGCTGCTCTGGGTTCCAGGTTCCACTGGTGAC-39) to expressa solely secretable form of 17-kDa TNF-a. All of the constructs weresequenced (Invitrogen, Shanghai, China) to confirm the integrity andpreciseness of the described mutations. 4T1 cells were transfected withexpression plasmids pTriEx/Neo (as a control), pTriEx/wtTNF, pTriEx/sTNF-a,or pTriEx/tmTNF-a, respectively, using Lipofectamine 2000 (Invitrogen,Carlsbad, CA) according to the manufacturer’s protocols. Cells were sub-jected to G418 (Life Technologies, Grand Island, NY) selection and clonedby limiting dilution.

Tumor inoculations

BALB/c mice were purchased from the Center of Medical ExperimentalAnimals of Hubei Province (Wuhan, China). TNFR1 or TNFR2 knockout(KO) mice on a BALB/c background were provided by Prof. Zhihai Qin(National Laboratory of Biomacromolecules, Institute of Biophysics, ChineseAcademy of Sciences, Beijing, China). Mice were bred and maintained underspecific pathogen-free conditions and treated according to Tongji MedicialCollege (China) Animal Care Guidelines. The animal study was approved bythe Ethical Committee of Tongji Medical College.

Six- and 10-wk-old female BALB/c, TNFR12/2, or TNFR22/2 micewere inoculated in the mammary fat pad with 53 105 4T1 parental cells orwith 4T1 cells stably transfected with wtTNF-a, sTNF-a, tmTNF-a, orempty vector. For in vitro experiments, MDSCs were isolated from spleenwhen primary tumors were 8–10 mm in diameter at 3–4 wk after inocu-lation of 4T1 cells. For in vivo experiments, tumor (transfected with TNF-aand its mutants) dimensions were measured with calipers every 3 d. Meantumor size was calculated by taking width2 3 length 3 0.52 (27). Allmeasurements were performed in a coded, blinded fashion.

Isolation, purification, and stimulation of MDSCs

Spleens were isolated from normal and tumor-bearing mice and washedtwice with Hank’s solution. After lysis of RBCs, splenocytes were frac-tionated by Percoll density gradient centrifugation as described by Kusmartsevet al. (28). Cells were harvested from the gradient interfaces between 50 and60%, and Gr-1+cells were isolated from this fraction using BD IMag anti-mouse Ly-6G and Ly-6C Particles-DM (BD Biosciences). For the stimu-lation of MDSCs, 300 ng/ml sTNF-a (PeproTech) or surface TNF-a on 1%paraformaldehyde-fixed 4T1/wtTNF-a cells (at an E:T ratio of 10:1) wasadded and incubated for 16 h. The supernatants and the cells were thenharvested for the corresponding assays in vitro.

MTT cell proliferation assay

4T1 cells (1 3 105) stably transfected with wtTNF-a, tmTNF-a mutant,sTNF-a mutant, or empty plasmid were seeded in 96-well microliter platesand incubated at 37˚C, 5% CO2 for the indicated time. Cells were stainedwith glucose-PBS containing 0.45 mg/ml MTT (Sigma-Aldrich) for 4 h,followed by lysis with 0.1 ml 100% DMSO. The OD value at 570 nm wasmeasured on a microplate reader (Tecan, Grodig, Austria).

Flow cytometry

Abs used for flow cytometry were fluorescein-conjugated mAbs, includingPE-anti-CD3 (145-2C11), FITC-anti-CD11b (ICRF44), PE-anti–Gr-1(RB6–8C5), PE-anti–TNF-a (MP6-XT22), PE-anti-TNFR1 (55R-286), PE-anti-TNFR2 (TR75-89), and isotype controls (mouse IgG1, rat IgG2b,Armenian hamster IgG) from BD Biosciences (San Diego, CA) or Bio-Legend (San Diego, CA).

Tumor tissue (100 mg) was minced into small pieces (1–3 mm3) anddigested for 1 h at 37˚C in 5 ml FCS-free RPMI 1640 medium containingtype IV collagenase (160 mg/ml), hyaluronidase (250 mg/ml), and DNase(5 mg/ml). After filtration through a 200-mesh sieve, a single-cell sus-pension was collected and cells were washed three times with PBS. Thesecells from primary tumor tissue or purified MDSCs from spleen were in-cubated with the above Abs in PBS containing 2% BSA for 30 min at 4˚Cand washed twice. Expression of the cell surface molecules was analyzedon an LSR II flow cytometer (Becton Dickinson, San Jose, CA) using BDFACSDiva software.

ELISA for cytokines

Tumor tissue (100 mg) in 1 ml PBS was homogenized at 4˚C with aPolytron homogenizer. After centrifugation at 12,000 3 g for 30 min at4˚C, the supernatants were collected. Commercial ELISA kits were usedto detect IL-10 and TGF-b in supernatants of tumor or of MDSCs fromTNFR1 KO, TNFR2 KO, or wild-type tumor–bearing mice after stimu-lation, as well as sTNF-a in supernatants of transfectants according to themanufacturer’s instructions (BD Biosciences). Plates were read at 450 nmon a microplate reader (Tecan), with 570 nm as the reference wavelength.

NO and ROS analysis

NO in the supernatants of MDSCs was determined by Griess agent witha nitrate/nitrite colorimetric assay kit (Jiancheng Bioengineering, Nanjing,China) as recommended by the manufacturer. Nitrate was converted tonitrite utilizing nitrate reductase, and then Griess reagents were added toconvert nitrite to a deep purple azo compound and the absorbance wasmeasured at 550 nm after 15 min incubation at room temperature.

The production of ROS in MDSCs was determined with the oxidation-sensitive dye 29,79-dichlorofluorescein diacetate (DCFDA; Sigma-Aldrich).Purified MDSCs were incubated at 37˚C in RPMI 1640 in the presence of2.0 mMDCFDA for 20 min after stimulation. Analysis was then performedby flow cytometry as described above.

Cytoplasmic, nuclear, or total protein preparation and Westernblot analysis

MDSCs (1 3 106) were harvested after stimulation and suspended inprecooled buffer A (10 mM HEPES [pH 7.8], 10 mM KCl, 0.1 mM EDTA,1 mM DTT, 0.5 mM PMSF, 5 mg/ml aprotinin, and 5 mg/ml leupeptin) andincubated on ice for 10 min. The cytoplasmic membrane was then brokendown by addition of Nonidet P-40 at a final concentration of 0.5% andvortexed for 10 s. The separation of the cytoplasmic protein from thenucleus was performed by centrifugation at 10,000 3 g for 1 min at 4˚C.The pelleted nuclei were washed three times with ice-cold lysis buffer A,then resuspended in 50 ml ice-cold lysis buffer C (50 mM HEPES [pH 7.8],0.42 M KCl, 0.1 mM EDTA, 1 mM DTT, 0.5 mM PMSF, 5 mg/ml apro-tinin, 5 mg/ml leupeptin, 5 mM MgCl2, and 20% glycerin) and incubated

The Journal of Immunology 1321

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 4: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

on ice for 30 min. After centrifugation at 12,000 3 g for 20 min at 4˚C,the supernatant containing nuclear protein was collected.

For total protein preparation, MDSCs activated with both forms of TNF-a, or 4T1 cells expressing wtTNF-a, sTNF-a, or tmTNF-a were solubi-lized in precooled lysis buffer (30 mM Tris-Cl, 150 mM NaCl, 1% NonidetP-40, 10% glycerol, 0.5 mM PMSF, 5 mg/ml aprotinin, and 5 mg/ml leu-peptin) on ice for 15 min. The total protein in supernatants was obtained bycentrifugation at 13,000 3 g for 20 min at 4˚C.

Fifty micrograms total, cytoplasmic, or nuclear protein was fractionatedby 12% SDS-PAGE and then transferred onto a polyvinylidene difluoridemembrane by electroblotting. The membranes were blocked overnight at4˚C with 5% nonfat dry milk in PBS-Tween 20 (0.05%) and then probedwith different primary Abs to TNF-a, ERK, p-ERK, JNK, p-JNK, p38,p-p38, NF-kB p65, IkB, lamin B1, or b-actin (Santa Cruz Biotechnology,Santa Cruz, CA), followed by incubation with corresponding secondaryAbs conjugated with HRP. Immunoreactive bands were visualized witha chemiluminescence reagent (Pierce) and then exposed to Kodak ImageStation 4000MM (Carestream Heath) for imaging.

RT-PCR and quantitative real-time PCR

Total RNA from MDSCs was isolated with TRIzol reagent (Invitrogen)according to the manufacturer’s instructions. One microgram RNA wasreverse-transcribed to cDNA using the RevertAid first-strand cDNA syn-thesis kit (Fermentas). The sequences of the primers synthesized bySangon were as follows: TNF-a (700 bp), 59-ACCCTCACACTCAGAT-CATCTTCTC-39 and 59-CAGATTGACCTCAGCGCTGAGTTG-39; TN-FR1 (480 bp), 59-GCCTGCTGCTGTCACTGGTGCTCCT-39 and 59-AGTC-CTGGGGGTTTGTGACATTTGC-39; TNFR2 (500 bp), 59-ACGAGTGCC-AGATCTCACAGGAATA-39 and 59-ACGATGTAAGGATGCTTGGAGTTT-G-39; b-actin (300 bp), 59-TCACCCACACTGTGCCCATCTACGA-39and 59-CATCGGAACCGCTCGTTGCCAATAG-39. The reaction was per-formed in 30 cycles (94˚C, 20 s; 58˚C, 20 s; and 72˚C, 30 s).

Real-time PCR was performed in duplicate for each sample on Mx3000P(Stratagene) using the Platinum SYBR Green qPCR SuperMix UDG kit(Invitrogen, Carlsbad, CA). One microliter cDNA was added to a 20 mlreaction mixture containing 10 ml SYBR Green PCR Master Mix and 400nM primers. The primers included ARG1, 59-GGTCCACCCTGACC-TATGTG-39 and 59-GCAAGCCAATGTACACGATG-39; iNOS, 59-GTT-CTCAGCCCAACAATACAAGA-39 and 59-GTGGACGGGTCGATGTC-AC-39; and b-actin, 59-AGTTGCGTTACACCCTTTC-39 and 59-CACC-TTCACCGTTCCAGT-39. The following cycling conditions were used toamplify cDNA: 3 min at 94˚C, followed by 30 s at 94˚C, 40 s at 58˚C, and60 s at 72˚C for 40 cycles. Threshold cycle (Ct) values of the target geneswere normalized to those of b-actin. The 22DDCt method (29) was used tocalculate relative quantitation of gene expression.

T cell proliferation assay

Splenic cells (53 107) from naive BALB/c mice were labeled with 5 mmolCFSE (Molecular Probes, Eugene, OR) (30) at 37˚C for 30 min. Afterwashing with PBS, 5 3 106 CFSE-labeled splenocytes were stimulated for72 h with 10 ng/ml PMA plus 0.5 mg/ml ionomycin (Sigma-Aldrich) in thepresence or absence of the supernatant from MDSCs. For TCR activation,2 3 105 CFSE-labeled splenocytes were cultured for 72 h with 2 mlDynabeads Mouse T-Activator CD3/CD28 (Life Technologies) and 20 IU/mlrecombinant murine IL-2 (Shanghai Haoran Bio Technologies) instead ofPMA and ionomycin. Each sample was in triplicate. Nonlabeled or nontreatedsplenic cells served as controls. Cells were harvested and stained with PE-anti-CD3 Ab (eBioscience). The proliferation of CD3+ T cells was evaluated byflow cytometry.

Immunohistochemistry

Several molecules, including Gr-1, CD4, CD8, Foxp3, and CD31, wereexamined in the tumor tissue by the avidin/biotin complex method. Frozensections were air-dried, fixed in cold acetone for 15 min, and incubated for60 min at room temperature with primary Abs, including goat anti-mouseCD4 or CD8 (Beijing Zhongshan Biotechnology, Beijing, China), rabbitanti-mouse Gr-1 (BD Biosciences), or Foxp3 (United States Biological) andrat anti-mouse CD31 (BD Pharmingen). After a 30-min incubation withbiotin-conjugated secondary Abs against goat, rabbit, or rat IgG (BeijingZhongshan Biotechnology), the sections were then incubated withperoxidase-labeled streptavidin for 20 min. After washing, color was de-veloped by incubation with its substrate diaminobenzidine for 5 min andthe expression of these molecules was observed under a microscope. Intumor tissue, immunoreactive cells were counted in two sections per an-imal and five sights per section under high-power microscopic view in ablinded manner.

Statistical analysis

All data were statistically analyzed by the Student t test or the one-way ANOVAtest. Differences were considered to be statistically significant at p , 0.05.

ResultstmTNF-a, but not sTNF-a, mainly promotes MDSCsuppressive activities

To observe effects of both forms of TNF-a on MDSC bioactivities,Gr-1+CD11b+ cells were freshly isolated from wild-type tumor-bearing mice. The purity of the Gr-1+CD11b+ MDSC populationwas .95% as evaluated by flow cytometry (Fig. 1A, uppercytogram). MDSCs were incubated with 300 ng/ml sTNF-a andtmTNF-a on transfectants (at an E:T ratio of 10:1). We found thattmTNF-a activated MDSCs, manifested as a marked increase inmRNA levels for ARG1 (Fig. 1A) and iNOS (Fig. 1B), and asignificant elevated production of NO (Fig. 1C) and ROS (Fig.1D). In contrast, sTNF-a inhibited production of NO by MDSCs,as compared with control, but it has no effect on ARG1 and iNOSat the mRNA level and on ROS production in MDSCs. Neutrali-zation of tmTNF-a by a specific Ab totally blocked tmTNF-a–induced activation of MDSCs. Furthermore, tmTNF-a has beenshown to result in obviously enhanced release of IL-10 (Fig. 1E)and TGF-b (Fig. 1F); however, sTNF-a only lifted TGF-b se-cretion by ∼2-fold, but it had no significant effect on IL-10 pro-duction. These data indicate that tmTNF-a, rather than sTNF-a,mainly activated MDSCs from tumor-bearing mice to release in-hibitor mediators NO, ROS, IL-10, and TGF-b.It is well known that MDSCs suppress activation and prolifer-

ation of T cells (16). tmTNF-a has been shown above to enhancethe release of IL-10 and TGF-b, two inhibitors for T cell prolif-eration, into the supernatants of MDSCs. To exclude the effect oftmTNF-a on T cells and the competition between MDSCs andT cells for binding to tmTNF-a, we collected supernatants fromMDSC culture to check whether the inhibitory mediators releasedby tmTNF-a–stimulated MDSCs could inhibit T cell proliferationinduced by PMA and ionomycin. As shown in Fig. 1G, PMA andionomycin induced T cell proliferation by ∼40%. Addition of thesupernatant of MDSCs from tumor-bearing mice into the culturepartly inhibited T cell proliferation (p , 0.001), whereas the su-pernatant from tmTNF-a–stimulated MDSCs totally blockedT cell proliferation (p , 0.001). Conversely, T cell proliferationcould be recovered by neutralization of tmTNF-a to abolish theactivation of MDSCs or blockage of IL-10 in the supernatant fromtmTNF-a–stimulated MDSCs using specific Abs.

Activatory effect of tmTNF-a on MDSCs is mediated byTNFR2

Although tmTNF-a is the primary ligand of TNFR2 (13), it isable to effectively trigger both of the TNF receptors (31). To knowwhich receptor, TNFR1 or TNFR2, mediated the activating effectof tmTNF-a on MDSCs, we first tested the expression of TNFR1and TNFR2 by MDSCs at both mRNA and protein levels. RT-PCRresults showed that TNFR2 transcripts were more abundant thanTNFR1 transcripts in MDSCs from both normal and tumor-bearing mice (Fig. 2A). Inconsistent with the results of RT-PCR, flowcytometry results also showed that the expression level of TNFR2 wasmuch higher than that of TNFR1 in MDSCs, whereas the expressionof both TNF receptors on the surface of MDSCs from tumor-bearingmice was slightly increased as compared with those on MDSCs fromnormal mice, but it had no statistical significance (Fig. 2B).Because MDSCs express both TNFRs, we isolated MDSCs from

TNFR1 or TNFR2 gene KO tumor-bearing mice to study whichTNFR mediates the effect of tmTNF-a on MDSCs. As shown in

1322 tmTNF-a ACTIVATES MDSCs VIA TNFR2

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 5: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

Fig. 2C, deficiency of TNFR1 had no influence on the suppressionof T cell proliferation by MDSCs or tmTNF-a–activated MDSCs.However, deficiency of TNFR2 significantly impaired tmTNF-aincreased MDSC inhibition of T cell proliferation by PMA andionomycin in a contact-independent manner. Similar results werealso observed in T cell proliferation induced by anti-CD3/anti-CD28 (Fig. 2D). Because IL-10 in the supernatant released bytmTNF-a–activated MDSCs plays an important role in suppres-sion of T cell proliferation (Fig. 1E, 1G), we further detectedwhether this effect of tmTNF-a on MDSCs is TNFR2-dependent.Indeed, in comparison with tmTNF-a–induced IL-10 release from

MDSCs, either TNFR2 KO (Fig. 2E) or neutralization of TNFR2(Fig. 2F) could totally blocked the effect of tmTNF-a on MDSCs.In contrast, TNFR1 deficiency by KO or neutralization failed toaffect tmTNF-a–induced release of IL-10 by MDSCs. Further-more, sTNF-a and tmTNF-a showed an inhibitory or stimulatoryeffect on NO production of MDSCs, respectively (Fig. 1C), asubstance also responsible for inhibition of T cell proliferation.We found that TNFR1 KO completely removed the suppressiveeffect of sTNF-a, whereas TNFR2 KO totally cleared up theactivatory effect of tmTNF-a on MDSCs to compare with thoseeffects of the both forms of TNF-a on MDSCs from wild-type

FIGURE 1. Activation of MDSCs by tmTNF-a. Purified Gr-1+CD11b+ MDSCs (1 3 106/ml) from wild-type tumor-bearing mice were incubated for 16 h

with 300 ng/ml sTNF-a or with tmTNF-a expressed on the 1% paraformaldehyde-fixed 4T1 cells (13 107) transfected with wtTNF-a (at an E:T ratio of 10:1).

The specificity of TNF-a activity was confirmed by a 30-min preincubation of the transfectants with an mAb against murine TNF-a. The purity of MDSCs was

evaluated by flow cytometry [cytogram at the top of (A)]. (A and B) Total mRNAwas isolated from treated MDSCs and transcripts of ARG1 (A) and iNOS (B)

were analyzed by real-time PCR. The production of NO (C) and ROS (D) by MDSCs was tested by Griess agent and DCFDA, respectively, and the release of

IL-10 (E) and TGF-b (F) from MDSCs was determined by ELISA. (G) The percentage of proliferated CD3+ T cells was tested by CFSE dilution assay. Naive

splenocytes (5 3 106) labeled with CFSE were stimulated for 72 h with 10 ng/ml PMA and 0.5 mg/ml ionomycin in the presence of supernatant (SN) from

MDSCs activated for 24 h by tmTNF-a. Cells were stained with PE-anti-CD3 Ab, and the proliferation of T cells (gated on CD3+ cells) was analyzed by flow

cytometry. Nontreated T cells served as a control. Data are presented as means 6 SD of three independent experiments. *p , 0.05, **p , 0.01,***p , 0.001

versus control (A–F) or PMA plus ionomycin alone (G). ###p , 0.001 versus SN of MDSCs treated with nothing (G).

The Journal of Immunology 1323

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 6: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

FIGURE 2. tmTNF-induced MDSC activation via TNFR2. (A) mRNA levels of TNFR1 and TNFR2 in purified Gr-1+CD11b+ cells from naive or tumor-

bearing mice were tested by RT-PCR and normalized to b-actin gene expression. (B) Expression of TNFR1 and TNFR2 on Gr-1+CD11b+ cells was an-

alyzed by flow cytometry. Upper and middle panels of (A) and (B) show the representative results from one experiment out of three, and lower histograms

show the means 6 SD of three independent experiments. (C and D) MDSCs (1 3 106/ml) from wild-type, TNFR1 KO, or TNFR2 KO tumor-bearing mice

were incubated for 24 h with medium alone or with 1 3 107 fixed 4T1 cells transfected with tmTNF-a (at an E:T ratio of 10:1). The supernatants (SNs)

were added to naive splenocytes labeled with CFSE and incubated for 72 h in the presence of 10 ng/ml PMA and 0.5 mg/ml ionomycin (C) or of 2 ml anti-

CD3/anti-CD28 beads plus 20 IU/ml recombinant murine IL-2 (D). The proliferated T cells (gated on CD3+ cells) were determined by flow cytometry after

staining with a PE-anti-CD3 Ab. Nontreated T cells served as a control. (E and G) MDSCs from wild-type, TNFR1 KO, or TNFR2 KO tumor-bearing mice

were incubated for 24 (E) or 16 h (G) with 4T1 cells transfected with tmTNF-a (at an E:T ratio of 10:1) or 300 ng/ml sTNF-a (G). IL-10 was detected by

ELISA (E), and NO was determined by Griess agent (G). For neutralization of tmTNF-a, a specific Ab (20 mg/ml) was used to pretreat the transfectant for

30 min. (F) MDSCs (1 3 106/ml) from wild-type tumor-bearing mice were pretreated at 4˚C for 30 min with anti-TNFR1, anti-TNFR2, or IgG (eBio-

science) in the concentration of 15 mg/ml and then stimulated with tmTNF-a for 24 h as described above. SNs were collected (Figure legend continues)

1324 tmTNF-a ACTIVATES MDSCs VIA TNFR2

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 7: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

mice (Fig. 2G). These data strongly implied that TNFR2 mainlymediated the activatory effect of tmTNF-a on MDSCs.

tmTNF-a activates MDSCs via the NF-kB pathway

The NF-kB pathway has been reported to be involved in the ac-tivation of MDSCs (32, 33). An inhibitor protein, IkBa, is phos-phorylated in response to stimulation with a variety of agents,including sTNF-a, resulting in its degradation and the transloca-tion of NF-kB from cytoplasm into the nucleus where NF-kBbinds to cognate DNA binding sites and activates the transcriptionof target genes (34). To determine whether the NF-kB pathwaymediated tmTNF-a–induced activation of MDSCs, we observedthe degradation of IkBa and translocation of NF-kB p65 byWestern blot. We found (Fig. 3A) that tmTNF-a significantly

induced IkBa degradation in the cytoplasm and the translocationof NF-kB p65 into the nucleus of MDSCs from tumor-bearingmice, although sTNF-a also had a slightly activatory effect. WhenMDSCs were pretreated for 30 min with pyrrolidine dithiocarbamate(PDTC), an NF-kB inhibitor, tmTNF-a–induced NF-kB activationwas markedly suppressed (Fig. 3B), and the increased inhibition ofT cell proliferation by MDSCs (Fig. 3C) was abolished.

The p38 pathway mediates tmTNF-a–induced MDSCactivation

Because sTNF-a has been well known to be an inducer of thephosphorylation of MAPK, including ERKl/2, JNK, and p38MAPK (35–37), we examined whether the action of tmTNF-a onMDSCs is related to this pathway. As shown in Fig. 4A, the

and the release of IL-10 was detected by ELISA. Results are shown as means6 SD of three independent experiments. *p, 0.05, **p, 0.01, ***p, 0.001

versus TNFR1 (A, B) or stimuli alone for T cell proliferation (C, D) or control (E–G). ##p, 0.01, ###p, 0.001 versus SN of MDSCs from wild-type, TNFR1

KO, or TNFR2 KO mice (C, D).

FIGURE 3. tmTNF-a–induced MDSC activation through the NF-kB pathway. (A and B) Purified Gr-1+CD11b+ MDSCs (1 3 106/ml) from wild-type

tumor-bearing mice were incubated for 30 min with medium alone, with 300 ng/ml sTNF-a, or with 1 3 107/ml fixed 4T1 cells transfected with wtTNF-a

(at an E:T ratio of 10:1). For neutralization of tmTNF-a, fixed 4T1/wtTNF cells were pretreated for 30 min with an mAb against murine TNF-a and washed

three times with PBS before addition to MDSCs. For suppression of NF-kB, MDSCs were pretreated for 60 min with 100 mM PDTC and then washed three

times with PBS before stimulation (B). The cells were harvested and lysed. The translocation of NF-kB p65 from the cytoplasm to the nucleus and

degradation of IkB were determined by Western blotting. Lamin B1 and b-actin were used as a nuclear or a cytoplasm/total protein loading control,

respectively. Results are representative of three independent experiments. (C) The supernatants (SNs) from MDSCs were collected after stimulation with

tmTNF-a for 24 h and added to naive splenocytes (5 3 106) labeled with CFSE in the presence of 10 ng/ml PMA and 0.5 mg/ml ionomycin. After in-

cubation for 72 h, the percentage of proliferated T cells (gated on CD3 stained cells) was measured by flow cytometry. Results represent means 6 SD of

three independent experiments. ***p , 0.001 versus PMA plus ionomycin alone, ###p , 0.001 versus SN of MDSCs treated with nothing.

The Journal of Immunology 1325

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 8: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

phosphorylation level of p38, but not ERK1/2 and JNK, wassignificantly elevated by tmTNF-a treatment, although sTNF-aalso had a weak effect. This action could be completely abrogatedby neutralizing tmTNF-a or using SB203580, a p38 MAPK in-hibitor (Fig. 4B), and T cell proliferation could be recovered frominhibition of tmTNF-a–activated MDSCs (Fig. 4C).

Ectopic expression of tmTNF-a on 4T1 mammary carcinomainduces MDSC accumulation and promotes tumor progression

To confirm the in vivo effect of tmTNF-a on MDSCs, the murinebreast cancer cell line 4T1 was stably transduced with wtTNF-a,which encodes both forms of TNF-a, with a tmTNF-a mutant,which encodes only uncleavable tmTNF-a, and with an sTNF-amutant, which encodes only sTNF-a. Indeed, our results demon-strated in Fig. 5 that 4T1/wtTNF-a expressed the surface moleculedetected by FACS (Fig. 5A) and released the soluble moleculedetermined by ELISA (Fig. 5B). 4T1/tmTNF-a expressed more of

the transmembrane form of TNF-a (94.7%) on the surface thandid 4T1/wtTNF-a (43%), but it released undetectable levels ofthe soluble molecule. However, 4T1/sTNF-a did not express thetransmembrane molecule, but it secreted higher levels of thesoluble molecule than did 4T1/wtTNF-a cells. Additionally, ec-topic expression of wtTNF-a, tmTNF-a, or sTNF-a mutant didnot affect the growth curve of 4T1 tumor cells in vitro (Fig. 5C).Next, we inoculated equal numbers (5 3 105) of different forms

of TNF-transduced 4T1 cells into BALB/c mice. As expected, thetumors that only expressed tmTNF-a grew faster and were largerthan in other tumor groups, displaying a marked tumor progression.Conversely, the tumors that only secreted sTNF-a grew muchslower and were smaller than in other tumor groups, showing asignificant inhibition of tumor development (Fig. 5D). The tumorsthat produced both forms of TNF-a showed a middle growth rate,exhibiting a delayed tumor progression. Additionally, MDSCaccumulation was significantly increased in tmTNF-a–expressing

FIGURE 4. tmTNF-a–induced MDSC activation through the p38 pathway. Purified Gr-1+CD11b+ MDSCs (1 3 106/ml) from wild-type tumor-bearing

mice were incubated for 30 min with medium alone (control), with 300 ng/ml sTNF-a, or with 1 3 107/ml fixed 4T1 cells transfected with wtTNF-a (at an

E:T ratio of 10:1). For the neutralization of tmTNF-a, fixed 4T1/wtTNF-a cells were preincubated for 30 min with an mAb against murine TNF-a. For

suppression of p38, MDSCs were pretreated with 20 mM SB203580 for 60 min and then washed three times with PBS before stimulation (B). The cells

were harvested and lysed, and the phosphorylation of ERK (A), JNK (A), and p38 (A, B) was analyzed by Western blotting. b-actin and the three total

signaling molecules served as loading controls. Results are representative of three independent experiments. (C) MDSCs were activated for 24 h with

tmTNF-a and the supernatants (SNs) were collected and added to 5 3 106/ml naive splenocytes labeled with CFSE in the presence of 10 ng/ml PMA and

0.5 mg/ml ionomycin. After incubation for 72 h, the percentage of proliferated T cells (gated on CD3-stained cells) was measured by flow cytometry. Data

are represented as means 6 SD of three independent experiments. ***p , 0.001 versus PMA and ionomycin alone, ###p , 0.001 versus SN of MDSCs

treated with nothing. SB, SB203580.

1326 tmTNF-a ACTIVATES MDSCs VIA TNFR2

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 9: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

tumor, but was markedly decreased in sTNF-a–secreting tumor(Fig. 5E, 5F), suggesting that tmTNF-a–triggered tumor pro-gression and sTNF-a–induced tumor inhibition are closely asso-ciated with accumulation of MDSCs.

tmTNF-a–expressing tumor produces more inhibitorymediators and promotes Treg accumulation, but it fails toinduce lymphocyte infiltration

Because MDSCs have been shown to accumulate in tmTNF-a–expressing tumors, we further detected whether inhibitory medi-ators that were released at least partially by MDSCs would beelevated in the tumor tissue. Indeed, the release of NO (Fig. 6A),IL-10 (Fig. 6B), and TGF-b (Fig. 6C) was significantly enhancedin tmTNF-a–expressing tumors, but not in wtTNF-a–producingand sTNF-a–secreting tumors. MDSCs are known as an importantregulator of angiogenesis in tumors (38). As expected, CD31+

staining showed an evident increase in the number of small bloodvessels in tmTNF-a–expressing tumors, but a relative decrease inangiogenesis of microvessels in sTNF-a–secreting tumors, indi-cating a pro- or antivascularization effect of tmTNF-a and sTNF-a, respectively (Fig. 6D). This pro- or antivascularization effectseemed to be associated with increased or decreased MDSC ac-cumulation in tmTNF-a or sTNF-a–expressing tumors.

In contrast to the accumulation of MDSCs, immunohisto-chemistry (Fig. 7A, 7B) showed a poor infiltration of lymphocytesin the tmTNF-a–expressing tumor as well as in the control.Conversely, in the sTNF-a–secreting tumor, the infiltration ofeither CD4+ or CD8+ lymphocytes was significantly increased,pointing out a relationship of MDSCs to lymphocyte accumula-tion. Moreover, MDSCs have been shown to induce Treg differ-entiation (39), and therefore we detected the accumulation ofFoxp3+ T cells in the tumor tissue. Indeed, Foxp3+ T cells weresignificantly increased in the tmTNF-a–expressing tumors, butslightly decreased in the sTNF-a–secreting tumors (Fig. 7C), in-dicating a possible association of MDSCs with Tregs. The in vivodata above confirmed again that tmTNF-a, rather than sTNF-a, isresponsible for the accumulation and activation of MDSCs, thuspromoting tumor progression.

DiscussionPrevious studies have demonstrated that various proinflammatorymediators, such as IL-1b (22, 40), IL-6 (23), the bioactive lipidPGE2 (24), S100A8/A9 (25, 41), and complement component C5a(42), are involved in the accumulation and activation of MDSCs intumor-bearing mice and cancer patients. However, whether TNF-a, as a prototype and key proinflammatory cytokine, is also in-

FIGURE 5. Promotion of MDSC ac-

cumulation and tumor progression by

tmTNF-a expressed on 4T1 cells. (A

and B) 4T1 cells stably transfected with

wt-TNF-a, tmTNF-a mutant, sTNF-a

mutant, or empty plasmid were cultured

for 24 h. (A) Transmenbrane TNF-a on

the cell surface of these transfectants

was detected by flow cytometry. Green

indicates 4T1/Neo, and red indicates

4T1/sTNF. The histogram is represen-

tative of one experiment out of three. (B)

TNF-a released in the supernatants was

measured by ELISA. (C) Transfectants

(105) were cultured for indicated time

and the growth of transfectants in vitro

was detected by MTT assay. (D) Mice

were inoculated by s.c. injection in the

mammary fat pad with 5 3 105 4T1/

sTNF-a, 4T1/tmTNF-a, 4T1/wtTNF-a,

or 4T1/Neo cells (six mice per group).

The size of tumors was measured every

3 d and are represented as means 6 SD.

(E) The percentage of Gr1+CD11b+ cells

in tumor tissue at 27 d after tumor in-

oculation was detected by flow cytom-

etry. All quantitative data are represented

as means 6 SD of three independent

experiments. *p , 0.05, **p , 0.01

versus 4T1/Neo (B, E) and 4T1/tmTNF

versus 4T1/Neo (D); #p , 0.05, ##p ,0.01 4T1/sTNF versus 4T1/Neo (D). (F)

Gr-1+ cells (stained brown) were detec-

ted in primary tumor sections from 4T1/

sTNF-a, 4T1/tmTNF-a, 4T1/wtTNF-a,

as well as 4T1/Neo tumor-bearing mice by

immunohistochemistry (original magnifi-

cation 3200) using a specific Ab.

The Journal of Immunology 1327

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 10: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

volved in this pathological process is still obscure. In the presentstudy, we provide firm evidence that tmTNF-a, rather than sTNF-a, is mainly responsible for MDSC activation, manifested asupregulation of iNOS and ARG1, with increased suppressiveactivities including elevated production and release of NO, ROS,IL-10, and TGF-b, as well as enhanced inhibition of T cellproliferation. Although murine splenic MDSCs exert only Ag-specific T cell inhibition in a contact manner, these cells acquirethe ability that tumor-derived counterparts possess to suppress Ag-nonspecific T cell functions when exposed to hypoxia (43). Inter-estingly, our results show that tmTNF-a also conferred on splenicMDSCs the characteristics of tumor-derived MDSCs, includingupregulation of iNOS and ARG1 expression, and thus inhibition ofAg-nonspecific T cell proliferation in a contact-independent manner.Furthermore, it is likely that tmTNF-a–stimulated IL-10 productionby MDSCs was required to suppress Ag-nonspecific T cell prolif-eration, as neutralization of IL-10 completely blocked the inhibitoryeffect of tmTNF-a.Because tmTNF-a is the primary ligand of TNFR2, and our results

showed that MDSCs expressed mainly TNFR2, we hypothesized thattmTNF-a could activate MDSCs through TNFR2. Indeed, our resultssupport this postulate, as TNFR2 deficiency or neutralization com-pletely abolished the activatory effect of tmTNF-a on MDSCs re-leasing IL-10 and inhibiting T cell proliferation. Furthermore,TNFR2 KO totally blocked tmTNF-a–induced elevation of NOproduction by MDSCs, whereas TNFR1 KO entirely preventedsTNF-a–induced inhibition of NO release from MDSCs. It seemsthat both forms of TNF-a affected MDSCs via different types ofTNFRs. Because tmTNF-a has a higher affinity for TNFR2 thansTNF-a (13), this may also account for the difference between two

forms of TNF-a in the activation of MDSCs. Furthermore, in con-trast to TNFR1, TNFR2 is considered to be associated with a nega-tive regulation of inflammatory pathological process (44, 45). It hasbeen reported that TNFR2-expressing CD4+Foxp3+ Tregs comprise∼40% of peripheral Tregs in normal mice and represent the maxi-mally suppressive subset of Tregs (46). Additionally, TNFR2-mediated signaling has also been demonstrated to be required forMDSC survival and accumulation (26). To our knowledge, in thepresent study we describe for the first time the association oftmTNF-a/TNFR2 with the functions of MDSCs. It is likely that theinteraction of tmTNF-a and TNFR2 preferentially leads to the acti-vation of Tregs and MDSCs, two important negative regulatory cellsthat are involved in anti-inflammation and tumor immune evasion.Our study demonstrates that the NF-kB pathway and the p38

pathway mediated tmTNF-induced MDSC activation. This issupported by the following evidence: 1) the level of p38 phos-phorylation in MDSCs stimulated by tmTNF-a was significantlyincreased, whereas the phosphorylation of two other MAPKfamily members, ERK and JNK, was not affected; 2) the NF-kBsignaling pathway was also activated in MDSCs by tmTNF-astimulation, as evidenced by accelerated degradation of IkBa andtranslocation of NF-kB p65 from the cytoplasm into the nucleus;and 3) upon preincubation of MDSCs with the p38 MAPK in-hibitor SB203580 or the NF-kB inhibitor PDTC, suppression ofT cell proliferation by tmTNF-a–activated MDSCs was abro-gated. It has been reported that the MAPK p38 pathway regulatesNF-kB transactivation via direct acetylation of p65 and is requiredfor TNF-mediated NF-kB activation (47, 48). Therefore, it is notsurprising that inhibition of either pathway was sufficient to blocktmTNF-a–induced activation of MDSCs.

FIGURE 6. An increase in the release of suppressive

mediators and promotion of tumor angiogenesis in tmTNF-

a–expressing tumors. (A) NO production in tumors was

detected by Griess agent. IL-10 (B) and TGF-b (C) in

tumors were tested by ELISA. (D) Immunohistochemical

examination of CD31+ cells showed brown-stained abnor-

mal tumor vasculature in the tumors (original magnifica-

tion 3200) (top), and microvessel density was detected in

10 fields per tumor section (two sections per mouse and

four mice per treated group), and results are presented as

the mean number of vessels per field (bottom). All quan-

titative data are represented as means 6 SD. *p , 0.05,

**p , 0.01 versus 4T1/Neo.

1328 tmTNF-a ACTIVATES MDSCs VIA TNFR2

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 11: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

Our in vivo data further confirmed that tmTNF-a, but not sTNF-a, promoted activation and accumulation of MDSCs, and thusfacilitated tumor progression. In comparison with the control,ectopic expression of uncleavable tmTNF-a promoted 4T1 tumorgrowth and progression concomitant with significant MDSC ac-cumulation, whereas expression of sTNF-a led to slower tumorgrowth and significant tumor regression with decreased MDSCrecruitment. However, previous studies including ours (2, 49) haveindicated that tmTNF-a leads to tumor regression. These contra-dictory results are likely due to the different types of tumorstudied and/or the different species of tmTNF-a expressed by thetansductants. In fact, the tmTNF-a transduced in our previousstudy was a human molecule that exerts its effects only viaTNFR1 in mice (50), whereas the tmTNF-a transduced in thisstudy was a murine molecule that exerts its effects via bothTNFR1 and TNFR2. Additionally, TNFR2 is required for TNF-ato activate MDSCs and to induce Tregs (11). In agreement withthe in vitro results, significant release of NO, IL-10, and TGF-bwas observed in the tumor expressing tmTNF-a, although theseinhibitory mediators were only partially produced by MDSCs.These mediators inhibit T cell proliferation, which might accountfor the poor lymphocyte accumulation in tmTNF-a–expressingtumor. Furthermore, it has been reported that MDSCs down-regulate L-selectin on T cells (51) that is needed for T cell homingto lymph nodes and a tumor microenvironment where they canencounter Ags and be activated. Recently, the ability of MDSCs topromote the de novo development of Foxp3+ Tregs in vivo hasbeen described (39). This is in line with our results showing in-creased Foxp3+ Tregs in tmTNF-a–expressing tumor with MDSCaccumulation. The induction of Tregs by MDSCs was found torequire the presence of IL-10 and TGF-b but was independent ofNO (39, 52). tmTNF-a–stimulated secretion of IL-10 and TGF-b

from MDSCs may favor induction of Tregs, in addition to itspossible direct effect on Tregs via TNFR2.MDSCs have nonimmunosuppressive pro-tumor functions, in-

cluding augmenting tumor angiogenesis, invasion, and metastasis(38, 53). Yang et al. (38) reported that MDSCs promote tumorangiogenesis by producing MMP9 and differentiating into endo-thelial cells. Consistent with this observation, our data showed anassociation of angiogenesis with MDSC accumulation in tmTNF-a– or sTNF-a–expressing tumors. Although tmTNF-a expressionhas been shown to be upregulated in the tumor endothelium (54)and play an angiogenic role (55), it is possible that tmTNF-a maybe a factor in tumor microenvironment that would facilitate MDSC-dependent tumor angiogenesis. Briefly, our in vivo data indicatedagain that tmTNF-a plays an important role in the activation ofMDSCs, and thus promotes tumor immune escape, one of themechanisms involved in tumor progression.Interestingly, sTNF-a solely expressed by tumor has opposite

effects. It failed to recruit MDSCs as compared with the control,but it attracted more lymphocytes in the tumor, promoting sig-nificant tumor inhibition. It appears that sTNF-a is an antitumorfactor, instead of being a tumor-promoting factor, in the absenceof tmTNF-a. This is consistent with our previous studies (9, 10,56), showing that when tmTNF-a expression is disturbed, tumorcells become sensitive to the cytotoxic effect of sTNF-a, becausetmTNF-a expressed by tumor constitutively activates NF-kB throughits reverse signaling, thereby protecting tumor from apoptosis andpromoting survival.In the present study, we demonstrated that tmTNF-a expressed

by tumor activates MDSCs through forward signaling via TNFR2to inhibit antitumor immune responses. However, sTNF-a solelyexpressed by tumor is beneficial for antitumor immunity. Al-though the proinflammatory effect of sTNF-a promotes tumor

FIGURE 7. Poor lymphocyte infiltration, but increased Treg accumulation, in tmTNF-a–expressing tumor. Infiltration of CD4+ (upper panel), CD8+

(middle panel), or Foxp3+ (bottom panel) cells was detected in tumor sections from different forms of TNF-a–producing tumors by immunohistochemistry

using specific Abs (original magnification 3200, left). CD4+, CD8+, and Foxp3+ cells were quantitatively analyzed in five fields per section (two sections

per mouse and five mice per treated group) by manual cell counting. The quantitative data of immune reactive cells are represented as means 6 SD and

shown on the right of the corresponding immunohistochemical images. *p , 0.05, **p , 0.01 versus 4T1/Neo.

The Journal of Immunology 1329

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 12: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

development, the inflammation conversely favors the developmentof immune responses. Unfortunately, both forms of TNF-a coexistin the tumor microenvironment. Therefore, selective inhibition oftmTNF-a may prevent tumor-promoting effects of sTNF-a–me-diated inflammation while deriving benefit from it for the tumortherapy through weakening immunosuppression and increasingthe sensitivity of tumor to apoptosis.

AcknowledgmentsWe thank Qinghua Hu for discussion of the data.

DisclosuresThe authors have no financial conflicts of interest.

References1. Chen, H., L. Xiao, H. Zhang, N. Liu, T. Liu, L. Liu, X. Hu, D. Yan, K. Yang,

B. Yin, et al. 2011. The involvement of b-actin in the signaling of transmem-brane TNF-a-mediated cytotoxicity. J. Leukoc. Biol. 89: 917–926.

2. Li, Q., L. Li, W. Shi, X. Jiang, Y. Xu, F. Gong, M. Zhou, C. K. Edwards, III, andZ. Li. 2006. Mechanism of action differences in the antitumor effects of trans-membrane and secretory tumor necrosis factor-alpha in vitro and in vivo. CancerImmunol. Immunother. 55: 1470–1479.

3. Szlosarek, P. W., and F. R. Balkwill. 2003. Tumour necrosis factor a: a potentialtarget for the therapy of solid tumours. Lancet Oncol. 4: 565–573.

4. Moore, R. J., D. M. Owens, G. Stamp, C. Arnott, F. Burke, N. East,H. Holdsworth, L. Turner, B. Rollins, M. Pasparakis, et al. 1999. Mice deficientin tumor necrosis factor-a are resistant to skin carcinogenesis. Nat. Med. 5: 828–831.

5. Balkwill, F. 2006. TNF-a in promotion and progression of cancer. CancerMetastasis Rev. 25: 409–416.

6. Popivanova, B. K., K. Kitamura, Y. Wu, T. Kondo, T. Kagaya, S. Kaneko,M. Oshima, C. Fujii, and N. Mukaida. 2008. Blocking TNF-a in mice reducescolorectal carcinogenesis associated with chronic colitis. J. Clin. Invest. 118:560–570.

7. Egberts, J. H., V. Cloosters, A. Noack, B. Schniewind, L. Thon, S. Klose,B. Kettler, C. von Forstner, C. Kneitz, J. Tepel, et al. 2008. Anti-tumor necrosisfactor therapy inhibits pancreatic tumor growth and metastasis. Cancer Res. 68:1443–1450.

8. Qin, Z., S. Kruger-Krasagakes, U. Kunzendorf, H. Hock, T. Diamantstein, andT. Blankenstein. 1993. Expression of tumor necrosis factor by different tumorcell lines results either in tumor suppression or augmented metastasis. J. Exp.Med. 178: 355–360.

9. Yan, D., N. Qin, H. Zhang, T. Liu, M. Yu, X. Jiang, W. Feng, J. Wang, B. Yin,T. Zhang, et al. 2009. Expression of TNF-a leader sequence renders MCF-7tumor cells resistant to the cytotoxicity of soluble TNF-a. Breast Cancer Res.Treat. 116: 91–102.

10. Zhang, H., D. Yan, X. Shi, H. Liang, Y. Pang, N. Qin, H. Chen, J. Wang, B. Yin,X. Jiang, et al. 2008. Transmembrane TNF-a mediates “forward” and “reverse”signaling, inducing cell death or survival via the NF-kB pathway in Raji Burkittlymphoma cells. J. Leukoc. Biol. 84: 789–797.

11. Chen, X., M. Baumel, D. N. Mannel, O. M. Howard, and J. J. Oppenheim. 2007.Interaction of TNF with TNF receptor type 2 promotes expansion and functionof mouse CD4+CD25+ T regulatory cells. J. Immunol. 179: 154–161.

12. Kleijwegt, F. S., S. Laban, G. Duinkerken, A. M. Joosten, A. Zaldumbide,T. Nikolic, and B. O. Roep. 2010. Critical role for TNF in the induction ofhuman antigen-specific regulatory T cells by tolerogenic dendritic cells. J.Immunol. 185: 1412–1418.

13. Grell, M., E. Douni, H. Wajant, M. Lohden, M. Clauss, B. Maxeiner,S. Georgopoulos, W. Lesslauer, G. Kollias, K. Pfizenmaier, and P. Scheurich.1995. The transmembrane form of tumor necrosis factor is the prime activatingligand of the 80 kDa tumor necrosis factor receptor. Cell 83: 793–802.

14. Kneilling, M., R. Mailhammer, L. Hultner, T. Schonberger, K. Fuchs,M. Schaller, D. Bukala, S. Massberg, C. A. Sander, H. Braumuller, et al. 2009.Direct crosstalk between mast cell-TNF and TNFR1-expressing endotheliamediates local tissue inflammation. Blood 114: 1696–1706.

15. Mazzon, E., E. Esposito, R. Di Paola, C. Muia, C. Crisafulli, T. Genovese,R. Caminiti, R. Meli, P. Bramanti, and S. Cuzzocrea. 2008. Effect of tumournecrosis factor-a receptor 1 genetic deletion on carrageenan-induced acute in-flammation: a comparison with etanercept. Clin. Exp. Immunol. 153: 136–149.

16. Gabrilovich, D. I., and S. Nagaraj. 2009. Myeloid-derived suppressor cells asregulators of the immune system. Nat. Rev. Immunol. 9: 162–174.

17. Ostrand-Rosenberg, S., and P. Sinha. 2009. Myeloid-derived suppressor cells:linking inflammation and cancer. J. Immunol. 182: 4499–4506.

18. Sonda, N., M. Chioda, S. Zilio, F. Simonato, and V. Bronte. 2011. Transcriptionfactors in myeloid-derived suppressor cell recruitment and function. Curr. Opin.Immunol. 23: 279–285.

19. Gabrilovich, D., T. Ishida, T. Oyama, S. Ran, V. Kravtsov, S. Nadaf, andD. P. Carbone. 1998. Vascular endothelial growth factor inhibits the developmentof dendritic cells and dramatically affects the differentiation of multiple hema-topoietic lineages in vivo. Blood 92: 4150–4166.

20. Pan, P. Y., G. X. Wang, B. Yin, J. Ozao, T. Ku, C. M. Divino, and S. H. Chen.2008. Reversion of immune tolerance in advanced malignancy: modulation ofmyeloid-derived suppressor cell development by blockade of stem-cell factorfunction. Blood 111: 219–228.

21. Serafini, P., R. Carbley, K. A. Noonan, G. Tan, V. Bronte, and I. Borrello. 2004.High-dose granulocyte-macrophage colony-stimulating factor-producing vac-cines impair the immune response through the recruitment of myeloid suppressorcells. Cancer Res. 64: 6337–6343.

22. Bunt, S. K., P. Sinha, V. K. Clements, J. Leips, and S. Ostrand-Rosenberg. 2006.Inflammation induces myeloid-derived suppressor cells that facilitate tumorprogression. J. Immunol. 176: 284–290.

23. Bunt, S. K., L. Yang, P. Sinha, V. K. Clements, J. Leips, and S. Ostrand-Rosenberg. 2007. Reduced inflammation in the tumor microenvironment delaysthe accumulation of myeloid-derived suppressor cells and limits tumor pro-gression. Cancer Res. 67: 10019–10026.

24. Sinha, P., V. K. Clements, A. M. Fulton, and S. Ostrand-Rosenberg. 2007.Prostaglandin E2 promotes tumor progression by inducing myeloid-derivedsuppressor cells. Cancer Res. 67: 4507–4513.

25. Sinha, P., C. Okoro, D. Foell, H. H. Freeze, S. Ostrand-Rosenberg, andG. Srikrishna. 2008. Proinflammatory S100 proteins regulate the accumulationof myeloid-derived suppressor cells. J. Immunol. 181: 4666–4675.

26. Zhao, X., L. Rong, X. Zhao, X. Li, X. Liu, J. Deng, H. Wu, X. Xu, U. Erben,P. Wu, et al. 2012. TNF signaling drives myeloid-derived suppressor cell ac-cumulation. J. Clin. Invest. 122: 4094–4104.

27. Hahnfeldt, P., D. Panigrahy, J. Folkman, and L. Hlatky. 1999. Tumor develop-ment under angiogenic signaling: a dynamical theory of tumor growth, treatmentresponse, and postvascular dormancy. Cancer Res. 59: 4770–4775.

28. Kusmartsev, S. A., Y. Li, and S. H. Chen. 2000. Gr-1+ myeloid cells derived fromtumor-bearing mice inhibit primary T cell activation induced through CD3/CD28costimulation. J. Immunol. 165: 779–785.

29. Livak, K. J., and T. D. Schmittgen. 2001. Analysis of relative gene expressiondata using real-time quantitative PCR and the 22DDC(T) method. Methods 25:402–408.

30. Liu, D., J. Yu, H. Chen, R. Reichman, H. Wu, and X. Jin. 2006. Statistical de-termination of threshold for cellular division in the CFSE-labeling assay. J.Immunol. Methods 312: 126–136.

31. Decoster, E., B. Vanhaesebroeck, P. Vandenabeele, J. Grooten, and W. Fiers.1995. Generation and biological characterization of membrane-bound,uncleavable murine tumor necrosis factor. J. Biol. Chem. 270: 18473–18478.

32. Delano, M. J., P. O. Scumpia, J. S. Weinstein, D. Coco, S. Nagaraj, K. M. Kelly-Scumpia, K. A. O’Malley, J. L. Wynn, S. Antonenko, S. Z. Al-Quran, et al. 2007.MyD88-dependent expansion of an immature GR-1+CD11b+ population inducesT cell suppression and Th2 polarization in sepsis. J. Exp. Med. 204: 1463–1474.

33. Tu, S., G. Bhagat, G. Cui, S. Takaishi, E. A. Kurt-Jones, B. Rickman, K. S. Betz,M. Penz-Oesterreicher, O. Bjorkdahl, J. G. Fox, and T. C. Wang. 2008. Over-expression of interleukin-1b induces gastric inflammation and cancer andmobilizes myeloid-derived suppressor cells in mice. Cancer Cell 14: 408–419.

34. Wajant, H., and P. Scheurich. 2011. TNFR1-induced activation of the classicalNF-kB pathway. FEBS J. 278: 862–876.

35. Beyaert, R., A. Cuenda, W. Vanden Berghe, S. Plaisance, J. C. Lee,G. Haegeman, P. Cohen, and W. Fiers. 1996. The p38/RK mitogen-activatedprotein kinase pathway regulates interleukin-6 synthesis response to tumor ne-crosis factor. EMBO J. 15: 1914–1923.

36. Westwick, J. K., C. Weitzel, A. Minden, M. Karin, and D. A. Brenner. 1994.Tumor necrosis factor a stimulates AP-1 activity through prolonged activation ofthe c-Jun kinase. J. Biol. Chem. 269: 26396–26401.

37. Yu, H. N., Y. R. Lee, E. M. Noh, K. S. Lee, E. K. Song, M. K. Han, Y. C. Lee,C. Y. Yim, J. Park, B. S. Kim, et al. 2008. Tumor necrosis factor-a enhancesDMSO-induced differentiation of HL-60 cells through the activation of ERK/MAPK pathway. Int. J. Hematol. 87: 189–194.

38. Yang, L., L. M. DeBusk, K. Fukuda, B. Fingleton, B. Green-Jarvis, Y. Shyr,L. M. Matrisian, D. P. Carbone, and P. C. Lin. 2004. Expansion of myeloidimmune suppressor Gr+CD11b+ cells in tumor-bearing host directly promotestumor angiogenesis. Cancer Cell 6: 409–421.

39. Huang, B., P. Y. Pan, Q. Li, A. I. Sato, D. E. Levy, J. Bromberg, C. M. Divino,and S. H. Chen. 2006. Gr-1+CD115+ immature myeloid suppressor cells mediatethe development of tumor-induced T regulatory cells and T-cell anergy in tumor-bearing host. Cancer Res. 66: 1123–1131.

40. Song, X., Y. Krelin, T. Dvorkin, O. Bjorkdahl, S. Segal, C. A. Dinarello,E. Voronov, and R. N. Apte. 2005. CD11b+/Gr-1+ immature myeloid cells me-diate suppression of T cells in mice bearing tumors of IL-1b-secreting cells. J.Immunol. 175: 8200–8208.

41. Cheng, P., C. A. Corzo, N. Luetteke, B. Yu, S. Nagaraj, M. M. Bui, M. Ortiz,W. Nacken, C. Sorg, T. Vogl, et al. 2008. Inhibition of dendritic cell differen-tiation and accumulation of myeloid-derived suppressor cells in cancer is reg-ulated by S100A9 protein. J. Exp. Med. 205: 2235–2249.

42. Markiewski, M. M., R. A. DeAngelis, F. Benencia, S. K. Ricklin-Lichtsteiner,A. Koutoulaki, C. Gerard, G. Coukos, and J. D. Lambris. 2008. Modulation ofthe antitumor immune response by complement. Nat. Immunol. 9: 1225–1235.

43. Movahedi, K., M. Guilliams, J. Van den Bossche, R. Van den Bergh,C. Gysemans, A. Beschin, P. De Baetselier, and J. A. Van Ginderachter. 2008.Identification of discrete tumor-induced myeloid-derived suppressor cell sub-populations with distinct T cell-suppressive activity. Blood 111: 4233–4244.

44. Hamid, T., Y. Gu, R. V. Ortines, C. Bhattacharya, G. Wang, Y. T. Xuan, andS. D. Prabhu. 2009. Divergent tumor necrosis factor receptor-related remodelingresponses in heart failure: role of nuclear factor-kB and inflammatory activation.Circulation 119: 1386–1397.

1330 tmTNF-a ACTIVATES MDSCs VIA TNFR2

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from

Page 13: Transmembrane TNF-α Promotes Suppressive Activities of ... · The Journal of Immunology Transmembrane TNF-a Promotes Suppressive Activities of Myeloid-Derived Suppressor Cells via

45. Suvannavejh, G. C., H. O. Lee, J. Padilla, M. C. Dal Canto, T. A. Barrett, andS. D. Miller. 2000. Divergent roles for p55 and p75 tumor necrosis factorreceptors in the pathogenesis of MOG35–55-induced experimental autoimmuneencephalomyelitis. Cell. Immunol. 205: 24–33.

46. Chen, X., and J. J. Oppenheim. 2010. TNF-a: an activator of CD4+FoxP3+

TNFR2+ regulatory T cells. Curr. Dir. Autoimmun. 11: 119–134.47. Saha, R. N., M. Jana, and K. Pahan. 2007. MAPK p38 regulates transcriptional

activity of NF-kB in primary human astrocytes via acetylation of p65. J.Immunol. 179: 7101–7109.

48. Vanden Berghe, W., S. Plaisance, E. Boone, K. De Bosscher, M. L. Schmitz,W. Fiers, and G. Haegeman. 1998. p38 And extracellular signal-regulated kinasemitogen-activated protein kinase pathways are required for nuclear factor-kBp65 transactivation mediated by tumor necrosis factor. J. Biol. Chem. 273:3285–3290.

49. Marr, R. A., C. L. Addison, D. Snider, W. J. Muller, J. Gauldie, andF. L. Graham. 1997. Tumour immunotherapy using an adenoviral vectorexpressing a membrane-bound mutant of murine TNF alpha. Gene Ther. 4:1181–1188.

50. Holler, R. A., K. Song, N. Fan, and D. J. Chang. 1992. The p70 tumor necrosisfactor receptor mediates cytotoxicity. Cell 70: 47–56.

51. Hanson, E. M., V. K. Clements, P. Sinha, D. Ilkovitch, and S. Ostrand-Rosen-berg. 2009. Myeloid-derived suppressor cells down-regulate L-selectin expres-sion on CD4+ and CD8+ T cells. J. Immunol. 183: 937–944.

52. Yang, R., Z. Cai, Y. Zhang, W. H. Yutzy, IV, K. F. Roby, and R. B. Roden. 2006.CD80 in immune suppression by mouse ovarian carcinoma-associated Gr-1+

CD11b+ myeloid cells. Cancer Res. 66: 6807–6815.53. Yang, L., J. Huang, X. Ren, A. E. Gorska, A. Chytil, M. Aakre, D. P. Carbone,

L. M. Matrisian, A. Richmond, P. C. Lin, and H. L. Moses. 2008. Abrogation ofTGFb signaling in mammary carcinomas recruits Gr-1+CD11b+ myeloid cellsthat promote metastasis. Cancer Cell 13: 23–35.

54. Clauss, M., C. Sunderkotter, B. Sveinbjornsson, S. Hippenstiel, A. Willuweit,M. Marino, E. Haas, R. Seljelid, P. Scheurich, N. Suttorp, et al. 2001. A per-missive role for tumor necrosis factor in vascular endothelial growth factor-induced vascular permeability. Blood 97: 1321–1329.

55. Rajashekhar, G., A. Willuweit, C. E. Patterson, P. Sun, A. Hilbig, G. Breier,A. Helisch, and M. Clauss. 2006. Continuous endothelial cell activationincreases angiogenesis: evidence for the direct role of endothelium linking an-giogenesis and inflammation. J. Vasc. Res. 43: 193–204.

56. Yu, M., X. Zhou, L. Niu, G. Lin, J. Huang, W. Zhou, H. Gan, J. Wang, X. Jiang,B. Yin, and Z. Li. 2013. Targeting transmembrane TNF-a suppresses breastcancer growth. Cancer Res. 73: 4061–4074.

The Journal of Immunology 1331

by guest on April 13, 2017

http://ww

w.jim

munol.org/

Dow

nloaded from