Programmed Death Ligand 1 Is Expressed by Non Hodgkin ... · Programmed Death Ligand 1 Is Expressed by Non–Hodgkin Lymphomas and Inhibits the Activity of Tumor-Associated T Cells

Human Cancer Biology

Programmed Death Ligand 1 Is Expressed by Non–HodgkinLymphomas and Inhibits the Activity of Tumor-Associated T Cells

David J. Andorsky1, Reiko E. Yamada1, Jonathan Said2, Geraldine S. Pinkus3, David J. Betting1, andJohn M. Timmerman1

AbstractPurpose: Programmed death ligand 1 (PD-L1) is expressed on antigen-presenting cells and inhibits

activation of T cells through its receptor PD-1. PD-L1 is aberrantly expressed on some epithelial

malignancies and Hodgkin lymphomas and may prevent effective host antitumor immunity. The role

of PD-L1 in non–Hodgkin lymphomas (NHL) is not well characterized.

Experimental Design: PD-L1 expression was analyzed in cell lines and lymphoma specimens by using

flow cytometry and immunohistochemistry. Functional activity of PD-L1 was studied by incubating

irradiated lymphoma cells with allogeneic T cells with or without anti-PD-L1 blocking antibody; T-cell

proliferation and IFN-g secretion served as measures of T-cell activation. Similar experiments were

conducted using cultures of primary lymphoma specimens containing host T cells.

Results: PD-L1 was expressed uniformly by anaplastic large cell lymphoma (ALCL) cell lines, but rarely

in B-cell NHL, confined to a subset of diffuse large B-cell lymphomas (DLBCL) with activated B-cell features

(3 of 28 cell lines and 24% of primary DLBCL). Anti-PD-L1 blocking antibody boosted proliferation and

IFN-g secretion by allogeneic T cells responding to ALCL and DLBCL cells. In autologous cultures of

primary ALCL and DLBCL, PD-L1 blockade enhanced secretion of inflammatory cytokines IFN-g ,granulocyte macrophage colony-stimulating factor, interleukin (IL)-1, IL-6, IL-8, IL-13, TNF-a, and

macrophage inflammatory protein-1a. In establishing cell lines from an aggressive PD-L1þ mature B-

cell lymphoma, we also noted that PD-L1 expression could be lost under certain in vitro culture conditions.

Conclusions: PD-L1 may thwart effective antitumor immune responses and represents an attractive

target for lymphoma immunotherapy. Clin Cancer Res; 17(13); 4232–44. �2011 AACR.

Introduction

Programmed death 1 (PD-1), a member of the CD28family, is an inhibitory receptor expressed on the surface ofT cells that functions to physiologically limit T-cell activa-tion and proliferation (1). Its ligand, PD-L1 (B7-H1/CD274), is expressed on antigen-presenting cells. Bindingof PD-L1 to its receptor inhibits T-cell activation andcounterbalances T-cell stimulatory signals, such as thebinding of B7 to CD28.

Dysregulation of the PD-1/PD-L1 pathway has beenimplicated in a wide variety of diseases. Impairment ofPD-1/PD-L1 signaling can lead to autoimmune disease inmurine systems (2, 3). In humans, several single nucleotidepolymorphisms in PD-1 have been associated with anincreased risk of developing rheumatologic disease (4).Conversely, upregulation of PD-1 signaling is associatedwith the persistence of chronic infections, includingHIV (5,6),Helicobacter pylori infection (7), and schistosomiasis (8).

PD-L1 is not expressed by normal epithelial tissues, but itis aberrantly expressed on a wide array of human cancers(9). In this context, PD-L1may promote cancer progressionby disabling the host antitumor response. Its expression ontumor cells has been associated with poorer prognosis inrenal cell carcinoma (10–12), breast cancer (13), Wilms’tumor (14), pancreatic cancer (15), ovarian cancer (16),urothelial cancer (17), gastric cancer (18), esophagealcancer (19), and hepatocellular carcinoma (20). In murinesystems, melanoma cells engineered to express PD-L1 areresistant to cytotoxic T lymphocyte (CTL)-mediated lysisand exhibit more aggressive tumor growth than wild-typemelanoma (21). Moreover, melanoma cells expressing PD-L1 can induce apoptosis in tumor-specific CTLs (9).

Compared with solid tumors, the spectrum of expres-sion and biological activity of PD-L1 in lymphomas is

Authors' Affiliations: 1Division of Hematology & Oncology, Department ofMedicine, and 2Department of Pathology & Laboratory Medicine, Univer-sity of California, Los Angeles, Los Angeles, California; and 3Department ofPathology, Harvard Medical School, Boston, Massachusetts

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

D.J. Andorsky and R.E. Yamada are co-first authors.

Corresponding Author: John M. Timmerman, Division of Hematology &Oncology, Center for Health Sciences, Room 42-121,10833 LeConteAvenue, University of California, Los Angeles Medical Center, LosAngeles, CA 90095. Phone: 310-794-4820; Fax: 310-206-5511; E-mail:[email protected]

doi: 10.1158/1078-0432.CCR-10-2660

�2011 American Association for Cancer Research.

ClinicalCancer

Research

Clin Cancer Res; 17(13) July 1, 20114232

Research. on July 21, 2020. © 2011 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

Published OnlineFirst May 3, 2011; DOI: 10.1158/1078-0432.CCR-10-2660

http://clincancerres.aacrjournals.org/

incompletely characterized. Using immunohistochemis-try, Brown and colleagues reported PD-L1 expression on 7of 11 peripheral T-cell lymphomas and 0 of 16 B-cell non-Hodgkin lymphomas (NHL; ref. 22). PD-L1 was detectedby reverse transcriptase–PCR in 5 anaplastic lymphomakinase (ALK)þ anaplastic large cell lymphoma (ALCL) celllines and by immunohistochemistry in 18 primary ALKþ

ALCL specimens (23). Another series reported that PD-L1was expressed in 4 of 14 diffuse large B-cell lymphomas(DLBCL), 0 of 9 T-cell lymphomas, and 8 of 13 classicHodgkin lymphoma (HL) cases (24).In this report, we describe the pattern of expression of

PD-L1 in a large series of primary human lymphomaspecimens (n ¼ 110) and NHL cell lines (n ¼ 34). Usingboth cell lines and primary tumor specimens, we show thatPD-L1 expressed on tumor cells is immunologically activein suppressing the activation of tumor-associated T cells.These results suggest PD-L1 blockade as a potentially usefulstrategy for lymphoma immunotherapy.

Materials and Methods

Cell lines and clinical sample preparationRaji, Ramos, and Daudi human Burkitt lymphoma, and

Jurkat T-cell lymphoblastic leukemia cell lines wereobtained from the American Type Culture Collection(ATCC). SU-DHL-1, SU-DHL-4, SU-DHL-6, SU-DHL-8,SU-DHL-9, SU-DHL-16, BCBL-1, Karpas 299, DEL,Hut78, and SUP-M2 were gifts from Dr. Linda Baum(UCLA, Los Angeles, CA). Granta-519, JeKo-1, and REC-1 were gifts from Dr. William Matsui (Johns HopkinsUniversity, Baltimore, MD). OCI-Ly-2, -3, -7, -10, -19,HBL-1, SU-DHL-2, and U2932 were gifts from Dr. LouisStaudt (National Cancer Institute, Bethesda, MD). SU-

DHL-5, SU-DHL-7, SU-DHL-10, NU-DHL-1, and USC-DHL-1 were gifts from Dr. Alan Epstein (University ofSouthern California, Los Angeles, CA). RC-K8 andMC116 cells were gifts from Dr. Izidore Lossos (Universityof Miami). BJA-B was a gift from Dr. Elliott Kieff (Harvard,Boston, MA). Unless otherwise specified, tumor cells werecultured in RPMI 1640 medium (Invitrogen) plus 10%heat-inactivated fetal calf serum (FCS; Omega Scientific),100 units/mL penicillin/streptomycin, 2 mmol/L L-gluta-mine, and 50 mmol/L b-mercaptoethanol ("RPMI completemedium"; all supplements from Invitrogen), at 37�C in 5%CO2. SU-DHL-6 and SU-DHL-8 cells were cultured in RPMIcomplete medium plus 20% FCS. Hut78 cells were culturedin Isocove’s modified Dulbecco’s medium (IMDM; Invi-trogen) plus 20% FCS. The OCI-Ly series, SU-DHL-2,U2932, and HBL-1 were cultured in IMDM completemedium plus 20% fresh human plasma (heparinized)instead of FCS.

Primary lymphoma specimens were obtained fromlymph node biopsies or involved peripheral blood afterwritten informed consent approved by the UCLA Institu-tional Review Board, enriched by Ficoll–Hypaque sedimen-tation (GE Healthcare), and cryopreserved in liquidnitrogen. For analysis, specimens were thawed quickly ina 37�C water bath and washed twice with warm RPMIcomplete medium before use.

Cytokine stimulation of B-cell linesRamos, Daudi, SU-DHL-4, and SU-DHL-6 cells were

cultured in RPMI complete medium containing 2,000RU/mL IFN-g (R&D Systems), or 10 mg/mL CpG oligo-deoxynucleotide (ODN) 10103 (sequence 50-TCGTC-GTTTTTCGGTCGTTTT-30; Coley Pharmaceuticals Group)plus interleukin (IL)-4 at 2 ng/mL (R&D Systems) for 24 to48 hours. Daudi, Ramos, SU-DHL-4, SU-DHL-6, OCI-Ly-3,and HBL-1 cells were cultured in complete medium con-taining IL-6, IL-10, or both at 50 ng/mL (R&D Systems) for24 to 72 hours. PD-L1 expression was analyzed by flowcytometry.

Flow cytometryMonoclonal antibodies (mAb) used to measure expres-

sion of cell surface markers by flow cytometry includedphycoerythrin (PE)-conjugated anti-human PD-L1/B7-H1(clone MIH1), PD-L2/B7-DC PE (clone MIH18), and PD-1PE (clone MIH4) from eBioscience; and CD3 fluoresceinisothiocyanate (FITC; clone HIT3a), CD3 PE (cloneUCHT1), CD4 PE (clone RPA-T4), CD8 PE (clone RPA-T8), CD20 FITC (clone L27), CD30 PE or CD30 FITC(clone BerH8), EMA/CD227/MUC1 FITC (clone HMPV),and appropriate isotype controls, all from BD Biosciences.Stained tumor cells were analyzed using a BD FACSCaliberflow cytometer (BD Biosciences) with FCS Express software(De Novo Software).

ImmunohistochemistryFrozen sections were cut at 2 to 4 mm and immediately

fixed in cold acetone for 20 minutes at 4�C. After air drying

Translational Relevance

Impaired host immunity is thought to play a role inthe pathogenesis and progression of lymphoma. Expres-sion of the negative T-cell regulator programmed deathligand 1 (PD-L1) seems to facilitate immune toleranceof various carcinomas. Here, we describe the spectrumof expression of PD-L1 among non–Hodgkin lympho-mas (NHL) and evaluate its functional activity in sup-pressing T-cell responses. In vitro experiments usingestablished cell lines and primary lymphoma specimensshow that both T-cell and B-cell lymphomas expressbiologically active PD-L1 and that suppression oftumor-associated T cells can be reversed by PD-L1blockade. Among diffuse large B-cell lymphomas, themost common NHL in adults, we found that PD-L1 isexpressed only in the nongerminal center subtype,which carries a poorer prognosis and frequently recursafter conventional chemoimmunotherapy. Our resultssuggest that targeting PD-L1 may be an effective anti-lymphoma immunotherapy for certain histologic sub-types.

Functional PD-L1 Expression by Non–Hodgkin Lymphomas

www.aacrjournals.org Clin Cancer Res; 17(13) July 1, 2011 4233




for 10 minutes, slides were incubated overnight withmouse anti-PD-L1 Ab (clone MIH1; eBioscience). Slideswere then incubated with DakoCytomation Envisionþ

System horseradish peroxidase (HRP)-labeled polymeranti-mouse for 30 minutes (DAKO), followed by the dia-minobenzidine (DAB) reaction. The sections were counter-stained with hematoxylin.

For formalin-fixed specimens, histologic sections fromparaffin-embedded tissue blocks were subjected to heat-induced epitope retrieval by using a steamer at 95�C for 25minutes in 0.01mol/L citrate buffer, pH 6.0 (for PD-L1), orat 115�C for 3 minutes in 0.1 mmol/L EDTA, pH 8.0, forCD10, BCL6, and MUM1. Sections were incubated withmouse mAbs to CD10 (Vector laboratories), BCL6(DAKO), and MUM1 (DAKO), followed by antibody loca-lization using the DakoCytomation Envisionþ SystemHRP-labeled polymer (DAKO). After 10 minutes of incu-bation with DAB, sections were counterstained with hema-toxylin. Staining for PD-L1 in paraffin sections wasconducted using a mAb (clone 5H1, provided by Dr.Lieping Chen, Johns Hopkins University) at BioPillarLaboratories, using previously described methods (11) orusing a polyclonal rabbit antiserum (Lifespan Biosciences),followed by DAKO DakoCytomation Envisionþ HRP-labeled polymer anti-rabbit detection.

Allogeneic T-cell proliferation assaysT cells were enriched from whole blood obtained from

healthy donors who gave informed consent, using the Roset-teSep T-cell enrichment cocktail (StemCell Technologies),following the manufacturer’s protocol. Enriched T cells (2�105) were cultured in RPMI completemedium at a 10:1 ratiowith irradiated (3000R) Karpas 299 cells in 96-well U-bot-tom plates (Nunc). Cells were fed every 2 days with freshmedium containing 10 IU/mL IL-2 (Chiron). After 1 week, Tcells were harvested, counted, and replated in quadruplicatewith fresh Karpas 299 cells (irradiated 3000R) at effector:target (E:T) ratios of 2:1 and 1:1 with 2� 104 tumor cells perwell in 96-well U-bottom plates with or without 10 mg/mLanti-PD-L1 (clone MIH1) or mouse IgG1 isotype controlmAbs (eBioscience). After 4 days, anti-PD-L1 and controlmAbs were replenished before cells were pulsed with 1 mCi/well 3[H]-thymidine (MP Biomedicals); cells were harvested16 hours later. Incorporated radioactivity (counts per min-ute, cpm) was measured using a b-liquid scintillation analy-zer (PerkinElmer), and results from quadruplicate culturesreported as arithmetic means � SD.

Derivation of PD-L1–expressing lymphoma cell linesLC-96 and RS-27

An 18-year-old women (LC-96) presented with rapidlyprogressive cervical and abdominal lymphadenopathy,ascites, and pleural effusions. Cervical lymph node biopsyconfirmedALKþALCL.Malignant ascites fluidwas collectedat therapeutic paracentesis, and then cells were isolated bycentrifugation and cryopreserved. Cell-free ascites fluid wasobtained by centrifugation and 0.45mmfiltration. A sampleof unmanipulated ascites fluid was placed into immediate

culture supplemented 1:1 with Dulbecco’s modified Eagle’smedium (DMEM) containing 10% FCS, 100 units/mLpenicillin/streptomycin, 2 mmol/L L-glutamine, and 50mmol/L b-mercaptoethanol (Invitrogen) at 37�C in 5%CO2. As tumor cells slowly grew over a period of 1 month,supplementation of the culture with cell-free ascites fluidwas gradually decreased (from 40% to 5%), until theresulting LC-96 cell line could grow in DMEM containing15% FCS. The surface immunophenotypes of the primaryascites cells and the resulting LC-96 cell line were deter-mined by flow cytometry as described above.

Peripheral blood mononuclear cells (PBMC) wereobtained from a patient (RS-27) with peripheral bloodinvolvement with an aggressive DLBCL. Flow cytometryshowed a monomorphic B-cell population expressingCD19, CD20, CD22, and FMC7, with surface k light chainrestriction. The cells did not express BCL1, CD5, CD10, orCD38, and FISH was negative for t(11;14) and c-Myc trans-locations (data not shown). After Ficoll–Hypaque isolation,PBMCs were cryopreserved in liquid nitrogen. Thawed cellswere initially cultured in DMEM complete medium contain-ing 20% FCS plus 10% fresh human serum and 10% 0.45-mm-filtered LC-96 ascites fluid, as described in Results.

Cytokine analysesFor allogeneic experiments, supernatants from cocul-

tures of Karpas 299 cells and healthy donor T cells, asdescribed earlier under allogeneic T-cell proliferationassays, were collected after 4 days of incubation and ana-lyzed for IFN-g by ELISA (R&D Systems). Ninety-six-wellMaxisorp plates (Nunc) were coated with mouse anti-human IFN-g antibody and then washed and blocked with1% BSA in PBS for 1 hour. Supernatants were added andincubated for 2 hours, followed by biotinylated goat anti-human IFN-g antibody (50 ng/mL). Detection was con-ducted using streptavidin-conjugated HRP and hydrogenperoxide–tetramethylbenzidine substrate, and absorbancewas determined at 450 nm/570 nm with a SPECTRAmaxPlus 384 microplate reader (Molecular Devices). Recombi-nant human IFN-g was used to generate a standard curve.

For autologous tumor cell–T-cell cocultures, cryopre-served LC-96 and RS-27 primary tumor specimens werethawed at 37�C, washed twice with warm RPMI completemedium, and 2–2.5 � 105 cells per well were plated in 6replicates in a 96-well U-bottom plate in RPMI completemedium. Phytohemagluttinin (PHA; Sigma) was added at0, 0.5, or 1 mg/mL with or without anti-PD-L1 or mouseIgG1 isotype control mAbs at 10 mg/mL. Cells were incu-bated for 5 days at 37�C in a 5%CO2 humidified incubator.Supernatants were collected and analyzed for IFN-g byELISA.

Cytokine multiplex analysis was conducted on cell-freemalignant LC-96 ascites fluid, spent media from the LC-96cell line, and primary LC-96 cells treated as above with PHAand anti-PD-L1 or isotype control antibody. Supernatantswere analyzed for levels of 16 cytokines [granulocytemacrophage colony-stimulating factor (GM-CSF), IFN-g ,IL-1, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, TNF-a, IL-13, thymus

Andorsky et al.

Clin Cancer Res; 17(13) July 1, 2011 Clinical Cancer Research4234




and activation regulated chemokine (TARC), IFN-a, stro-mal cell-derived factor-1b (SDF-1b), macrophage inflam-matory protein-1a (MIP-1a), and monokine induced byinterferon-g (MIG)], using SearchLight protein array multi-plex sandwich-ELISA (Pierce Biotechnology).

Results

PD-L1 is widely expressed by ALCL but uncommonlyamong B-cell NHL cell linesWe screened a large panel of human lymphoma cell lines

for PD-L1 expression by flow cytometry (Table 1). Repre-

sentative histograms are shown in Figure 1A. Three of the28 B-cell lymphoma cell lines tested expressed PD-L1. Twoof the positive lines (OCI-Ly-10 and HBL-1) have beenclassified as DLBCL of the activated B-cell (ABC) subtypeon the basis of gene expression profiling (25), whereas thethird (RC-K8) is known to have constitutive upregulationof the nuclear factor kB (NF-kB) pathway, a hallmark of theABC phenotype (26). In contrast, 5 of 6 T-cell lymphomacell lines exhibited expression of PD-L1, including all 4ALCL lines tested. PD-L1 expression was strongest amongstthe ALCL lines, with 3 of 4 lines showing 2-log increasesover isotype controls, whereas Jurkat cells (T-cell acute

Table 1. Expression of PD-L1 among 34 human lymphoma cell lines

Cell line Lymphoma subtype PD-L1 PD-L2

B cellDaudi Burkitt � �Raji Burkitt � �Ramos Burkitt � �BJA-B Burkitt � �MC116 Burkitt � �NU-DHL-1 DLBCL (GCB) � �OCI-Ly-2 DLBCL (GCB) � �OCI-Ly-7 DLBCL (GCB) � �OCI-Ly-19 DLBCL (GCB) � �SU-DHL-4 DLBCL (GCB) � �SU-DHL-6 DLBCL (GCB) � �SU-DHL-2 DLBCL (ABC) � �U2932 DLBCL (ABC) � �OCI-Ly-3 DLBCL (ABC) � �OCI-Ly-10 DLBCL (ABC) þþ �HBL-1 DLBCL (ABC) þ �RC-K8 DLBCL (NF-kBþ) þ �SU-DHL-5 DLBCL � �SU-DHL-7 DLBCL � �SU-DHL-8 DLBCL � �SU-DHL-9 DLBCL � �SU-DHL-10 DLBCL � �SU-DHL-16 DLBCL � �USC-DHL-1 DLBCL � �Granta-519 Mantle cell � �JeKo-1 Mantle cell � �REC-1 Mantle cell � �BCBL-1 1� effusion � �

T cellKarpas 299 ALCL þþ �SU-DHL-1 ALCL þþ �SUP-M2 ALCL þþ �DEL ALCL þ �Jurkat T-ALL þ �Hut78 Sezary syndrome � �

NOTE: Expression of PD-L1 and PD-L2 was measured by flow cytometry. "þþ" indicates �2 log MFI above isotype control; "þ"indicates < 2 logs MFI above control.Abbreviation: T-ALL, precursor T acute lymphoblastic leukemia.






lymphoblastic leukemia) had low-level expression of PD-L1, and Hut78 (Sezary syndrome) was negative for PD-L1.PD-L1 mRNA expression correlated with protein expres-sion, with the highest levels found in ALCL and a subset ofABC DLBCL lines (Supplementary Table S1). In addition,all cell lines were screened for PD-1 and PD-L2 expression.Only Jurkat cells expressed PD-1 (data not shown), and nocell line expressed PD-L2.

We then attempted to induce PD-L1 expression in B-celllymphoma lines that did not constitutively express it, asstimulation of some tumor cells with IFN-g can induce PD-L1 expression (9,27). Ramos, Daudi, SU-DHL-4, and SU-DHL-6 were incubated with IFN-g or CpG plus IL-4 for 24to 48 hours, and PD-L1 expression was monitored by flowcytometry. No cell line responded to IFN-g , and only theRamos cell line showed modestly increased PD-L1 expres-sion 48 hours after stimulation with CpG plus IL-4(Fig. 1B). In ALCL, PD-L1 expression is induced by STAT3signaling (23), and IL-6 and IL-10 are both potent inducersof STAT3 (25). Therefore, we asked whether IL-6 and IL-10stimulation of NHL lines could upregulate PD-L1. Daudi,Ramos, SU-DHL-4, SU-DHL-6, OCI-Ly-3, and HBL-1 cells

were cultured with IL-6, IL-10, or both at 50 ng/mL for 24to 72 hours, but none showed significant increase in PD-L1expression (data not shown). Thus, PD-L1 expression is notreadily altered in cultured lymphoma cells by exogenouscytokines.

PD-L1 is expressed by a subset of primary humanDLBCLs

We next tested 68 lymphoma tissue specimens forexpression of PD-L1 [Table 2 (A and B)]. Thirty-threeDLBCLs, 3 primary mediastinal B-cell lymphomas(PMBCL) and 9 HLs were analyzed by immunohistochem-istry in frozen specimens. Single-cell suspensions of 23additional B-cell NHL specimens, including 16 follicularlymphomas (FL), were analyzed by flow cytometry. Expres-sion of PD-L1 among B-cell NHL specimens was hetero-geneous. Twenty-seven percent of DLBCL specimensshowed expression of PD-L1. In contrast, all 3 PMBCLspecimens were PD-L1þ. PD-L1 was not expressed in anycases of FL (n¼ 16), small lymphocytic lymphoma (n¼ 2),marginal zone lymphoma (n¼ 3), or single cases of Burkittor mantle cell lymphoma. Eight of 9 HLs expressed PD-L1

Karpas 299 (ALCL)

B

A

100

75

50

25

0100 101 102 103 104

Jurkat (T-ALL)

HBL-1 (DLBCL) OCI-Ly-10 (DLBCL)

SU-DHL-1 (ALCL) SUP-M2 (ALCL)

Ramos (Burkitt) SU-DHL-6 (DLBCL)

PD-L1

Cou

ntC

ount

Cou

nt

Cou

nt

Cou

nt

Cou

nt

Ramos: No stimulation Ramos: CpG + IL-4 stimulation

100 101 102 103 104

PD-L1100 101 102 103 104

PD-L1100 101 102 103 104

PD-L1

100 101 102 103 104

PD-L1

104

78

52

26

0

104

78

52

26

0

120

90

60

30

0

120

90

60

30

0

100 101 102 103 104

PD-L1100 101 102 103 104

PD-L1100 101 102 103 104

PD-L1

100 101 102 103 104

PD-L1

100 101 102 103 104

PD-L2

100 101 102 103 104

PD-L1

100 101 102 103 104

PD-L2

0

21

41

62

82

Cou

nt

Cou

nt

0 0

24

47

71

94

26

52

77

103

114

86

57

29

0 0

22

44

66

88 112

84

56

28

0 0

30

61

91

121

Figure 1. High-level PD-L1 expression among ALCL but not among most B-cell NHL cell lines. A, flow cytometric analysis of PD-L1 expression is shownfor 8 representative NHL cell lines (among 34 described in Table 1). Consistent high-level PD-L1 expression is a feature of ALCL cell lines but not ofB-cell lines. B, low-level expression of PD-L1 and PD-L2 is induced in Ramos B-cell lymphoma cells after incubation with CpG plus IL-4 for 48 hours.Bold black line represents PD-L1 or PD-L2 staining, blue line isotype control, and shaded histogram unstained.

Andorsky et al.





in Reed–Sternberg cells, in concordance with previousobservations (28).DLBCLs were classified into germinal-center B (GCB) or

non-GCB subtype on the basis of the immunohistochemicalmarkers CD10, BCL6, and MUM1, which correlate with cellof origin subtype as determined by gene expression profiling(29). Of 33 evaluable frozen cases, 19were GCB and 14werenon-GCB. Only 1 of the GCB tumors expressed PD-L1. Incontrast, 8 of 14 (57%) of non-GCB tumors expressed PD-L1(P¼ 0.0004 by Fisher’s exact test). This pattern of expressionparallels that seen in our cell lines, where PD-L1 expressionwas found in3of 6DLBCL lineswithABC features but in 0of7 GCB DLBCL cell lines (Table 1).We next conducted immunohistochemistry for PD-L1

expression in a separate set of 42 formalin-fixed, paraffin-embedded lymphoma specimens [Table 2 (C)], whichrequired a different mAb (5H1), previously used to stainPD-L1 in paraffin sections (10, 30). Cases were consideredpositive when the majority of tumor cells stained for PD-L1.All 7 FLswerenegative for PD-L1,whereas 4 of 5ALCLs and 6of30DLBCLswerepositive for PD-L1.Representative images

are shown in Supplementary Figure S1. Of note, tumor-associated histiocytes stained positive for PD-L1 in 9 of 24(38%)DLBCLs inwhich tumor cellswerenegative. Similarly,PD-L1þ histiocytes were also found in FLs surroundingtumor cell follicles. Of DLBCL specimens, 11 were GCBand 19 were non-GCB. None of the GCB DLBCL stainedfor PD-L1, whereas 6 (32%) of the non-GCB DLBCLs werepositive for PD-L1 (P ¼ 0.061), consistent with results weobtained in frozen sections. Comparative results of thefrozen and paraffin DLBCL series are shown in Table 2 (D).

Of note, we also stained 91 formalin-fixed, paraffin-embedded lymphoma specimens, using a polyclonal rabbitanti-PD-L1 antiserum (Lifespan Biosciences). Using thismethodology (31), 18 of 22 (82%) DLBCLs expressed PD-L1 (data not shown). Because of discordance betweenparaffin and frozen section results, we tested the polyclonalanti-PD-L1 antibody on cell pellets of several PD-L1–nega-tive B-cell lines and a Daudi lymphoma xenograft. Asseveral of these negative controls stained positive, weconcluded that this antibody was unreliable for detectingPD-L1 expression in lymphomas.

Table 2. Expression of PD-L1 in primary lymphoma tissue specimens

PD-L1 detection method Lymphoma subtype Numberof cases

Number ofPD-L1þ

% PD-L1þ

A. Immunohistochemistry:frozen specimens

HLa 9 8 89

DLBCL 33 9 27PMBCL 3 3 100

B. Flow cytometry:cryopreserved specimens

FL 16 0 0

Small lymphocyticlymphoma/chroniclymphocytic leukemia

2 0 0

Marginal zone lymphoma 3 0 0Mantle cell lymphoma 1 0 0Burkitt lymphoma 1 0 0

C. Immunohistochemistry:paraffin specimens

ALCL 5 4 80

FL 7 0 0DLBCL 30 6 20

n GCB Non-GCB P

þ � þ �

D. ImmunohistochemistryFrozen (MIH1 antibody) 33 1 (5%) 18 (95%) 8 (57%) 6 (42%) 0.0004Paraffin (5H1 antibody) 30 0 (0%) 11 (100%) 6 (32%) 13 (68%) 0.0613

NOTE: Immunohistochemistry of frozen sections (A) and flow cytometry (B) of cryopreserved specimens were conducted with theanti-PD-L1 antibody clone MIH1. Paraffin sections (C) were stained with clone 5H1. Frozen and paraffin sections of DLBCL wereclassified as germinal center origin or nongerminal center origin (D). PD-L1 expression in DLBCL occurs almost exclusively in tumorsof nongerminal center origin.aStaining in Hodgkin and Reed–Sternberg cells.






PD-L1 expressed by ALCL inhibits the proliferationand cytokine secretion of allogeneic T cells

We next asked whether PD-L1 expressed by lymphomacells was biologically active in attenuating host immuneresponses. Because PD-L1 was strongly expressed in bothALCL cell lines and tumor specimens, we chose this as ourinitial in vitro model. We hypothesized that antibodyblockade of PD-L1 would result in greater T-cell activity,showing that the presence of PD-L1 on target tumor cellsserves to inhibit T-cell responses. In cultures of donorallogeneic T cells primed for 7 days with irradiated ALCLcells, both T-cell proliferation and IFN-g secretion weremarkedly increased in the presence of a blocking anti-PD-L1 antibody (Fig. 2A). In contrast, anti-PD-L1 did not alterproliferation or IFN-g secretion of T cells incubated in theabsence of tumor targets. As a control, irradiated SU-DHL-4cells, which do not express PD-L1, were used as targets. Inthis case, PD-L1 blockade did not alter the degree of T-cellproliferation or IFN-g secretion (Fig. 2B). Even in 5-daycultures of unprimed normal donor T cells plus irradiatedALCL target cells, IFN-g secretion was uniformly increasedin the presence of anti-PD-L1 (Fig. 2C). The differencesseen in T-cell proliferation were smaller and not consis-tently statistically significant. Nonetheless, these resultsshow functional expression of immunosuppressive PD-L1 by ALCL cells.

PD-L1 expression by primary ALCL attenuates theactivity of tumor-associated T cells

To further study tumor–T-cell interactions, cryopre-served malignant ascites from a patient with newly diag-nosed ALKþ ALCL (LC-96; see Materials and Methods),containing approximately equivalent proportions of PD-L1–expressing tumor cells and tumor-associated T cells,was used as an autologous system (Fig. 3A). The primarytumor cells within the ascites (PD-L1þ, EMAþ, and CD30þ)were associated with a mixture of CD4þ and CD8þ T cells.

Cells from the ascites were incubated for 5 days with anti-PD-L1, isotype control antibody, or media alone, plusdifferent concentrations of PHA, to serve as a polyclonalT-cell activator, and supernatants were assayed for IFN-gsecretion as an indicator of T-cell stimulation (Fig. 3B).Without the addition of PHA (media alone), even with theaddition of IL-2 (10 mg/mL), no IFN-g secretion was seen.Yet in the presence of PHA (0.5 or 1.0 mg/mL), anti-PD-L1provoked a marked increase in IFN-g secretion (P < 0.0001by one-way ANOVA compared with isotype control anti-body or media alone). Thus, PD-L1 expressed by fresh,primary ALCL cells can suppress the function of tumor-associated autologous T cells.

Ascites cells were serially passaged in cell culture to derivea new ALCL cell line, designated LC-96 (see Materials andMethods). The immunophenotype of the LC-96 cell linemirrored that of the primary ascites tumor cells (Fig. 3A,bottom), with strong expression of PD-L1, EMA, and CD30but without expression of PD-1, CD3, CD4, or CD8. FISHanalysis revealed a t(2;5)(p23;q35) nucleophosmin(NPM)–ALK translocation, which was also observed in

the primary clinical specimen (data not shown). WhenLC-96 cells were incubated with PHA, with or without anti-PD-L1 antibody, there was no secretion of IFN-g (data notshown), showing that T cells and not tumor cells are thesource of IFN-g in the primary ascites cultures after PD-L1blockade.

To further characterize the T-cell response induced byPD-L1 blockade, we quantitated the secretion of 16 cyto-kines by using multiplex ELISA. First, we measured thecytokines present in cell-free ascites fluid, which wouldreflect the tumor environment in situ (SupplementaryFig. S2A). Interestingly, the fluid contained high levels ofIL-6, as well as IL-10, and SDF-1b. We next surveyedcytokines in spent culture media from the establishedLC-96 cell line (Supplementary Fig. S2B) to discern whichmight be products of tumor cells themselves. High levels ofIL-8, IL-10, and SDF-1b were observed, indicating these aslikely products of primary tumor cells in vivo.

Next, primary ascites cells were incubated with or with-out PHA and anti-PD-L1 antibody, and the cytokineprofile determined (Supplementary Fig. S2C). WithoutPHA, most cytokines were secreted at low levels. How-ever, the addition of anti-PD-L1 resulted in increasedsecretion of IL-6, IL-8, TNF-a, and MIP-1a comparedwith control antibody or media alone. The addition ofPHA resulted in further enhancement of cytokine secre-tion, including GM-CSF, IFN-g , IL-1, IL-6, IL-8, TNF-a, IL-13, and MIP-1a. Levels of IL-2, IL-4, IL-5, IL-10, IFN-a,TARC, SDF-1b, and MIG were not altered by PD-L1blockade (data not shown).

Functional PD-L1 expression by an aggressive primaryB-cell lymphoma

To show that PD-L1 can also be immunologically activewhen expressed by B-cell lymphomas, analogous experi-ments were conducted using a tumor sample from a patientwith an aggressive DLBCL (RS-27), which lacked expres-sion of CD10 and BCL6, consistent with non-GCB pheno-type. Circulating tumor cells strongly coexpressed PD-L1and CD20, as measured by flow cytometry (Fig. 4A).Unmanipulated RS-27 PBMCs containing approximately75% tumor cells and 20% CD3þ T cells (Fig. 4A) werecultured with PHA in the presence or absence of anti-PD-L1. After 5 days, supernatants were assayed for IFN-gsecretion (Fig. 4B). As described above with the LC-96ALCL tumor cell–T-cell mixture, PD-L1 blockade resultedin increased IFN-g secretion by tumor-associated T cells(P ¼ 0.009 by one-way ANOVA), indicating functionalinhibition of T cells by PD-L1 expressed by the B-celllymphoma.

PD-L1 expression may be lost or attenuated duringserial in vitro passage of lymphoma cells

In deriving the new RS-27 B-cell lymphoma line, wediscovered that expression of PD-L1 could be lost during invitro culture (Fig. 4C). After thawing, primary tumor cellsfrom patient RS-27 were initially cultured in mediumcontaining 20% FCS plus 10% fresh human serum and

Andorsky et al.





2:1 1:1 E alone0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

E:T ratio

Ave

rag

e C

PM

2:1 1:1 E alone0

10,000

20,000

30,000

40,000

E:T ratioA

vera

ge

CP

M

4:1 2:1 0

2,500

5,000

7,500

10,000

12,500

15,000

17,500

E:T ratio

Ave

rag

e C

PM

4:1 2:1 0

5,000

10,000

15,000

20,000

E:T ratio

Ave

rag

e C

PM

4:1 2:1 0

10,000

20,000

30,000

40,000

E:T ratio

Ave

rag

e C

PM

2:1 1:1 E alone 0

25,000

50,000

75,000

100,000

125,000

150,000

175,000

E:T ratio

Ave

rag

e C

PM

T-cell donor 1A

C

BT-cell donor 2 PD-L1–negative targets

T-cell donor 1 T-cell donor 2 T-cell donor 3

P = 0.007P = 0.965

P = 0.010

P = 0.204

P = 0.063

P = 0.019

P = 0.886P = 0.452

2:1 1:1 E alone0

500

1,000

1,500

2,000

E:T ratio

IFN

-γ (

pg

/mL

)

4:1 2:10

100

None

PD-L1 antibody

Control antibody

200

300

400

500

600

700

800

E:T ratio

IFN

-γ (

pg

/mL

)

4:1 2:1 0

500

1,000

1,500

E:T ratio

IFN

-γ (

pg

/mL

)

4:1 2:105

1015202530354045

E:T ratio

IFN

-γ (

pg

/mL

)

2:1 1:1 E alone0

50

100

150

200

250

300

350

E:T ratio

IFN

-γ (

pg

/mL

)

2:1 1:1 E alone0

50100150200250300350400450

E:T ratio

IFN

-γ (

pg

/mL

)

P < 0.001

P < 0.001

P < 0.001

P < 0.001

Figure 2. PD-L1 blockade enhances the activation of T cells cocultured with allogeneic ALCL cells. A, irradiated Karpas 299 ALCL cells (PD-L1þ), werestimulated for 1 week with T cells from 2 healthy donors and then incubated with freshly irradiated Karpas 299 target cells at 2:1 and 1:1 E:T ratiosin the presence of media alone, anti-PD-L1 antibody, or control antibody. After 4 days, supernatants were collected for IFN-g measurement (bottom), orantibody was replenished and cells pulsed with [3H]-thymidine overnight to measure T-cell proliferation (top). Data are represented as mean � SD ofquadruplicate cultures. P values shown are for anti-PD-L1 antibody versus isotype control antibody by one-way ANOVA. B, PD-L1 blockade does not affect Tcells incubated with PD-L1–negative tumor cells. Incubation of healthy donor T cells with a B-cell lymphoma line that does not express PD-L1 (SU-DHL-4)show that anti-PD-L1 antibody does not significantly alter allospecific proliferation (top) or IFN-g secretion (bottom). C, PD-L1 blockade augments activation ofallogeneic T cells directly stimulated with ALCL cells. Irradiated Karpas 299 cells were incubated for 5 days with T cells from 3 healthy donors at 4:1 and 2:1E:T ratios in the presence of anti-PD-L1 antibody, control antibody, or media alone. Supernatants were collected for IFN-g measurement by ELISA (bottom) orcells were pulsed with [3H]-thymidine overnight tomeasure T-cell proliferation (top). Proliferation data are represented asmean� SD of quadruplicate cultures.P values shown are for anti-PD-L1 antibody versus isotype control antibody.






10% 0.45-mm-filtered LC-96 ascites fluid (as a source oflymphoma-derived growth factors). Tumor cells slowlyexpanded under these conditions, and when weanedslowly from LC-96 ascites and fresh human serum, con-tinued to express high levels of PD-L1 and CD20. However,if cells were weaned rapidly (over 2 weeks) into mediacontaining 20% FCS and 5% pooled human AB serum, PD-L1 expression was almost entirely lost. Accordingly, T-cellproliferation and IFN-g production were only increased byPD-L1 blockade when allogeneic T cells were incubatedwith RS-27 PD-L1–positive cells (Supplementary Fig. S3).Culture of PD-L1–negative RS-27 cells for 48 hours inmedia containing 10% fresh human serum, or CpG

plus IL-4, could not restore PD-L1 expression (data notshown). Thus, PD-L1 expression by B-cell lymphomas caneasily be lost upon tumor cell establishment and serialpassage in vitro.

We also observed attenuation of PD-L1 expression underdifferent culture conditions in the established OCI-Ly-10ABC DLBCL cell line (Fig. 4D). When grown in mediasupplemented with 20% human plasma, the cells dis-played bright expression of PD-L1 [mean fluorescenceintensity (MFI) ¼ 465]. However, when the same cellswere transferred to media containing 20% FCS, thecells appeared less healthy, as evidenced by increasednumbers of dead cells with lower forward scatter, and

Primary tumor(ascites)

A

B

CD3 CD4 CD8 PD-1 PD-L1 EMA CD30

Cell line

Cell conditions

IFN

-γγ (p

g/m

L)

PHA 0

0

500

1,000

1,500

2,000

PHA 0.5 PHA 1.0

None

PD-L1 antidbody

Control antibody

P < 0.001

P < 0.001

1,024

768

512

256

00 256 512

FSC-H

SS

C-H

768

1,024

768

512

256

00 256 512

FSC-H

SS

C-H

768 1024

1024

100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104

104

103

102

101

100

100 101

2%

55% 20%

0%24%

0%

95%4%

CD30 FITC

PD

-L1

PE

102 103 104

104

103

102

101

100

100 101

CD30 FITC

PD

-L1

PE

102 103 104

100%

0%0%

0%104

103

102

101

100

100 101

CD30 FITC

PD

-L1

PE

102 103 104

100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104

100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104

Lymphocyte

Tumor

Figure 3. PD-L1 blockade enhances the activation of T cells in the presence of autologous ALCL cells that express PD-L1. A, immunophenotyping ofmalignant ascites from a patient with newly diagnosed ALKþ ALCL and cultured cell line (LC-96) from the same patient. The top 2 rows show results for theprimary tumor (ascites), with far left panel displaying forward and side scatter plots for analysis of tumor (large cell) and lymphocyte (small cell) gates. The redcurve in each histogram represents the surface marker, and unstained cells are shown in gray. Tumor cells (top row) were mostly negative for CD3, CD4, CD8,and PD-1 but expressed PD-L1, EMA, and CD30. The lymphocyte gate (middle row) contains predominantly CD3þ T cells, with a mixture of CD4þ and CD8þ Tcells having low-level expression of PD-1, PD-L1, and CD30. The immunophenotype of the derived LC-96 cell line is shown in the bottom row. Tumor cellsare strongly positive for PD-L1, EMA, and CD30, without expression of CD3, CD4, CD8, or PD-1. B, LC-96 primary ascites cells, containing approximatelyequivalent proportions of PD-L1–expressing tumor cells and tumor-associated T cells, were incubated for 5 days with PHA to activate T cells in the presence ofmedia alone, anti-PD-L1 antibody, or isotype control antibody. In the presence of 0.5 or 1.0 mg/mL PHA, cells incubated with anti-PD-L1 antibodysecreted more IFN-g than controls (P < 0.0001 by one-way ANOVA). Results shown are representative of 3 independent experiments.

Andorsky et al.





PD-L1 expression diminished in the viable cells (MFI ¼72). When the cells were transferred back to 20% humanplasma, PD-L1 expression returned to its previous level(MFI¼ 664; data not shown). Therefore, culture conditionscan alter the expression of PD-L1 even among well-estab-lished lymphoma cell lines.

Discussion

Cancers use multiple mechanisms to thwart endogenoushost antitumor immunity (32). While accumulating dataindicate that expression of the negative T-cell regulatorymolecule PD-L1 by tumor cells or tumor-associated anti-gen-presenting cells represents an important pathwaywhereby cancers evade host immunity (1), only limiteddata have been available regarding the expression of PD-L1

among common NHL subtypes and its ability to suppressautologous T-cell functions.

We studied the spectrum of PD-L1 expression amonghuman lymphomas and showed its capacity to impair thefunction of tumor-associated T cells in both T- and B-celllymphomas. This is the largest reported series of PD-L1expression in human lymphoma cell lines and primarytumors and encompasses the most common B-cell NHLsubtypes. We observed near uniform expression of PD-L1in ALCL cell lines and primary tumors. In contrast, wefound that among B-cell lymphomas, PD-L1 expression isessentially confined to a subset of the clinically importantABC/non-GCB subtype of DLBCL. We further showed thatPD-L1 expressed by lymphoma cells is biologically active,with antagonist antibody blockade resulting in increasedactivation of adjacent T cells. This was true in allogeneic

PHA 0 PHA 0.50

100

200

300

400

500

600

700

A B

C DCell conditions

IFN

-γ (

pg

/mL

)

Primary tumor Weaned rapidlyfrom human serum

Weaned graduallyfrom human serum

OCI-Ly-10, 20% FCS

FSC-HS

SC

-H

Co

un

tC

ou

nt

PD-L1PE

PD-L1PE

RS-27 cell line

Primary RS-27 tumor

MFI = 465

OCI-Ly-10, 20% human plasma

P = 0.009

MFI = 72

104

104

103

75% 11% 9%

103

102

102

101

101

CD20 FITC CD3 FITC CD3 FITC

PD

-L1P

E

CD

4PE

CD

8PE

100

100

104

104

103

103

102

102

101

101

CD20 FITC

PD

-L1P

E

100

100

104

102499

74

50

25

0

768

512

256

256 512 768 1020

0

FSC-H

SS

C-H

102471

53

36

18

0

768

512

256

256 512 768 1020

0

104

103

103

102

102

101

101

CD20 FITC

PD

-L1P

E

100

100

104

104

103

103

102

102

101

101

CD20 FITC

PD

-L1P

E

100

100

104

104

103

103

102

102

101

101

100

100

104

104

103

103

102

102

101

101

100

100

100

101

102

103

104

100

101

102

103

104

None

PD-L1 antibody

Control antibody

Figure 4. PD-L1 expressed in primary B-cell lymphoma is biologically active but readily lost during in vitro culture. A, PBMCs from a patient with aggressivemature B-cell lymphoma were analyzed by flow cytometry. Malignant cells coexpress CD20 and PD-L1 and comprise 75% of the total. The specimenalso contains 11% CD4þ T cells and 9% CD8þ T cells. B, PD-L1 blockade enhances the activation of T cells in the presence of autologous B-cell lymphomaexpressing PD-L1. Patient PBMCs were incubated in triplicate for 5 days with PHA 0.5 mg/mL in the presence of media alone, anti-PD-L1 antibody, or isotypecontrol antibody, and IFN-g wasmeasured in supernatants by ELISA. Data are represented as mean� SD of triplicate cultures. Results are representative of 2independent experiments. C, PD-L1 expression by primary B-cell lymphoma can be lost during serial in vitro passage. Primary tumor cells obtained directlyfrom peripheral blood coexpress PD-L1 and CD20. Cells were initially cultured in vitro with human serum, and weaned either rapidly or gradually to mediumcontaining FCS, while monitoring PD-L1 and CD20 expression by flow cytometry. D, PD-L1 expression may be attenuated on the basis of cell cultureconditions. OCI-Ly-10 cells, which express PD-L1, were cultured in 20% human plasma or 20% FCS. PD-L1 expression was substantially lower whenthe cells were grown in 20% FCS (MFI ¼ 72) than in 20% human plasma (MFI ¼ 465).






models using ALCL or DLBCL tumor cells as targets, as wellas in primary tumor specimens of ALCL and DLBCL con-taining mixtures of lymphoma cells and autologous lym-phocytes. Further studies will be required to confirmwhether PD-L1 blockade results in similar effects inprimary NHL specimens from lymph nodes or other extra-nodal sites of tumor.

Our observations regarding the pattern of expressionof PD-L1 in human lymphomas are consistent withother reports in smaller series. Brown and colleaguesreported PD-L1 expression in 7 of 11 peripheral T-celllymphomas, including ALCL, and in 0 of 16 B-cell NHLs(22). Marzec and colleagues reported PD-L1 staining in100% of 18 ALCLs (23), although the polyclonal anti-PD-L1 antibody used in this study possibly yielded false-positive cases (see later). Wilcox and colleagues reportedPD-L1 expression in 15% of 131 T-cell lymphomas,including 3 of 9 ALCLs (30). Xerri and colleaguesreported PD-L1 expression in 4 of 14 DLBCLs (including2 PMBCLs) but not in follicular (n ¼ 8), mantle cell(n ¼ 4), marginal zone (n ¼ 4), or Burkitt (n ¼ 3)lymphomas (24), similar to our own results. PD-L1 isfrequently expressed in HL within Hodgkin and Reed–Sternberg cells, as reported in a series of 4 cases byYamamoto and colleagues (28), consistent with ourfindings. PD-L1 expression by PMBCL is not surprising,as gene expression profiling has revealed increased PD-L1 and PD-L2 mRNAs in this lymphoma and given theclose biologic relationship of this disease to HL (33).

It seems that PD-L1 expression in B-cell lymphomas isuncommon. Interestingly, we found that among DLBCLcell lines and primary tumor specimens, PD-L1 expressionwas almost entirely confined to the ABC/non-GCB subtype(Tables 1 and 2). The ABC DLBCL subtype identified bygene expression profiling is associated with inferior survivalcompared with the GCB subtype (34), even in cohorts ofpatients treated with rituximab (35, 36). ABC DLBCL ischaracterized biologically by upregulation of NF-kB (37),but numerous other differences from GCB DLBCL exist. Itis possible that PD-L1 expression is one of several "viru-lence factors" that lead to the inferior prognosis amongABC DLBCLs, and we suggest this hypothesis be tested in alarge series of molecularly classified DLBCLs with asso-ciated clinical outcome data. The aggressive B-cell lym-phoma from which we derived the PD-L1þ RS-27 cellline (Fig. 4) was indeed virulent; the patient died of centralnervous system disease despite high-dose chemotherapyand allogeneic stem cell transplantation. Finally, the lowfrequency of PD-L1 expression we observed in DLBCL celllines (50% among ABC-type; Table 1) may underestimatethe true prevalence in primary DLBCL. In establishing theRS-27 cell line, we observed that PD-L1 expression waseasily lost during serial passage of the cells. Thus, loss ofPD-L1 expression may have occurred during the establish-ment of other human lymphoma cell lines, as seems to bethe case in melanoma. Although virtually all primarymelanomas express PD-L1, most melanoma cell lines donot (9).

PD-L1 expressed by leukocytes within the tumor micro-environment may play a role in host immune suppressioneven when not expressed by the tumor cells themselves.We observed that 38% of PD-L1 negative DLBCLs wereinfiltrated by PD-L1þ histiocytes. Increased numbers oflymphoma-associated macrophages have been associatedwith worse prognosis in FL (38) and HL (39), although thesame association is not seen in DLBCL (40, 41). Thus,further studies examining the relationships between lym-phoma-associated macrophages, PD-L1 expression, andclinical outcome are warranted. For such studies, it willbe important to utilize the mAbs MIH1 or 5H1. In ourexperience, the polyclonal anti-PD-L1 antisera used byMarzec and colleagues (23) seems to be less specific forPD-L1 than reported (see Results).

Several investigators have also shown that the PD-1/PD-L1 axis can also influence the function of lymphoma-infiltrating T cells in cases of PD-L1–negative human lym-phomas. Yang and colleagues found that B-cell lym-phoma–associated Treg cells could express PD-L1 andsuppress the function of PD-1þ tumor–associated T cells,an effect partially reversible by PD-L1 blockade (42). Simi-larly, Nattamai andNeelapu found that PD-1wasmarkedlyupregulated on tumor-derived and peripheral blood T cellsin FL in association with impaired Th1 cytokine secretion(43). Antibody blockade of PD-1 improved proliferation oftumor-derived T cells and promoted the activation ofnatural killer cells (44).

As in other series, we found near uniform expression ofPD-L1 in ALCL, a rare but clinically distinct T-cell neo-plasm. Nonetheless, PD-L1 is not a feature of all peripheralT-cell lymphomas. Wilcox and colleagues (30) found thatPD-L1 was expressed by T-cell lymphoma tumor cells inonly a minority of cases, yet often expressed by tumor-associated stromal histiocytes, just as we observed inDLBCL. Our report adds to these findings in showingthe ability of PD-L1 on ALCL and DLBCL cells to suppressthe responses of both allogeneic and autologous tumor-associated T cells (Figs. 2–4 and Supplementary Figs. S2and S3).

In our cultures of primary ALCL ascites cells containingautologous T cells, PD-L1 blockade enhanced the produc-tion of IFN-g , as well as GM-CSF, IL-1, IL-6, IL-8, TNF-a, IL-13, and MIP-1a. We also noted that the ascites fluidrepresenting the tumor microenvironment of this casecontained high levels of IL-6, IL-10, and SDF-1b whereasthe established LC-96 cell line secreted these same cyto-kines, plus IL-8. These cytokines likely play roles in thepathogenesis and clinical manifestations of ALCL, havingbeen detected in ALCL and other lymphomas (45–48).Intriguingly, SDF-1 (CXCL12) is associated with cancermetastasis (49) and angiogenesis (41), and as most ALCLexpress the SDF-1 receptor CXCR4 (50), this suggestsa possible autocrine feedback loop in this lymphomasubtype.

In conclusion, the current work adds to a growing bodyof literature documenting the important role of PD-1/PD-L1 signaling in the pathogenesis of NHL. PD-L1 is highly

Andorsky et al.





expressed in ALCL, HL, and some poor prognosis DLBCLsof the ABC/non-GCB subtype, where it acts to negativelyregulate adjacent T cells. Targeting the PD-1/PD-L1 path-way using antagonistic mAbs may thus be an attractiveapproach to lymphoma immunotherapy.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Authors' Contributions

D.J. Andorsky and R.E. Yamada contributed equally in designing andconducting research, analyzing data, and writing the manuscript. J. Saiddesigned and conducted research and analyzed data. G.S. Pinkus designed

and conducted research. D.J. Betting designed and conducted researchand analyzed data. J.M. Timmerman designed research and wrote themanuscript.

Grant Support

The study is supported by a grant from Gabrielle's Angel Foundation forCancer Research (to J.M. Timmerman) and an American Society of ClinicalOncology Young Investigator Award (to D.J. Andorsky). J.M. Timmerman is aDamon Runyon Clinical Investigator supported in part by the Damon RunyonCancer Research Foundation (CI-26-05).

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received October 1, 2010; revised February 28, 2011; accepted April 13,2011; published OnlineFirst May 3, 2011.

References1. Keir ME, Butte MJ, Freeman GJ, Sharpe AH. PD-1 and its ligands in

tolerance and immunity. Annu Rev Immunol 2008;26:677–704.2. Ansari MJ, Salama AD, Chitnis T, Smith RN, Yagita H, Akiba H, et al.

The programmed death-1 (PD-1) pathway regulates autoimmunediabetes in nonobese diabetic (NOD) mice. J Exp Med 2003;198:63–9.

3. Nishimura H, Nose M, Hiai H, Minato N, Honjo T. Development oflupus-like autoimmune diseases by disruption of the PD-1 geneencoding an ITIM motif-carrying immunoreceptor. Immunity 1999;11:141–51.

4. Okazaki T, Honjo T. PD-1 and PD-1 ligands: from discovery to clinicalapplication. Int Immunol 2007;19:813–24.

5. D’Souza M, Fontenot AP, Mack DG, Lozupone C, Dillon S, Meditz A,et al. Programmed death 1 expression on HIV-specific CD4þ T cells isdriven by viral replication and associated with T cell dysfunction. JImmunol 2007;179:1979–87.

6. Zhang JY, Zhang Z, Wang X, Fu JL, Yao J, Jiao Y, et al. PD-1 up-regulation is correlated with HIV-specific memory CD8þ T-cellexhaustion in typical progressors but not in long-term nonprogres-sors. Blood 2007;109:4671–8.

7. Beswick EJ, Pinchuk IV, Das S, Powell DW, Reyes VE. Expression ofthe programmed death ligand 1, B7-H1, on gastric epithelial cells afterHelicobacter pylori exposure promotes development of CD4þ CD25þ

FoxP3þ regulatory T cells. Infect Immun 2007;75:4334–41.8. Smith P, Walsh CM, Mangan NE, Fallon RE, Sayers JR, McKenzie AN,

et al. Schistosoma mansoni worms induce anergy of T cells viaselective up-regulation of programmed death ligand 1 on macro-phages. J Immunol 2004;173:1240–8.

9. Dong H, Strome SE, Salomao DR, Tamura H, Hirano F, Flies DB, et al.Tumor-associated B7-H1 promotes T-cell apoptosis: a potentialmechanism of immune evasion. Nat Med 2002;8:793–800.

10. Thompson RH, Dong H, Kwon ED. Implications of B7-H1 expressionin clear cell carcinoma of the kidney for prognostication and therapy.Clin Cancer Res 2007;13:709s–15s.

11. Thompson RH, Kuntz SM, Leibovich BC, Dong H, Lohse CM,WebsterWS, et al. Tumor B7-H1 is associated with poor prognosis in renal cellcarcinoma patients with long-term follow-up. Cancer Res 2006;66:3381–5.

12. Thompson RH, Webster WS, Cheville JC, Lohse CM, Dong H, Lei-bovich BC, et al. B7-H1 glycoprotein blockade: a novel strategy toenhance immunotherapy in patients with renal cell carcinoma. Urol-ogy 2005;66:10–4.

13. Ghebeh H,Mohammed S, Al-Omair A, Qattan A, Lehe C, Al-Qudaihi G,et al. The B7-H1 (PD-L1) T lymphocyte-inhibitory molecule isexpressed in breast cancer patients with infiltrating ductal carcinoma:correlation with important high-risk prognostic factors. Neoplasia2006;8:190–8.

14. Routh JC, Ashley RA, Sebo TJ, Lohse CM, Husmann DA, Kramer SA,et al. B7-H1 expression inWilms tumor: correlation with tumor biology

and disease recurrence. J Urol 2008;179:1954–9; discussion1959–60.

15. Nomi T, Sho M, Akahori T, Hamada K, Kubo A, Kanehiro H, et al.Clinical significance and therapeutic potential of the programmeddeath-1 ligand/programmed death-1 pathway in human pancreaticcancer. Clin Cancer Res 2007;13:2151–7.

16. Hamanishi J, Mandai M, Iwasaki M, Okazaki T, Tanaka Y, YamaguchiK, et al. Programmed cell death 1 ligand 1 and tumor-infiltrating CD8þ

T lymphocytes are prognostic factors of human ovarian cancer. ProcNatl Acad Sci U S A 2007;104:3360–5.

17. Nakanishi J, Wada Y, Matsumoto K, Azuma M, Kikuchi K, Ueda S.Overexpression of B7-H1 (PD-L1) significantly associates with tumorgrade and postoperative prognosis in human urothelial cancers.Cancer Immunol Immunother 2007;56:1173–82.

18. Wu C, Zhu Y, Jiang J, Zhao J, Zhang XG, Xu N. Immunohistochemicallocalization of programmed death-1 ligand-1 (PD-L1) in gastric carci-noma and its clinical significance. Acta Histochem 2006;108:19–24.

19. Ohigashi Y, Sho M, Yamada Y, Tsurui Y, Hamada K, Ikeda N, et al.Clinical significance of programmed death-1 ligand-1 and pro-grammed death-1 ligand-2 expression in human esophageal cancer.Clin Cancer Res 2005;11:2947–53.

20. Gao Q, Wang XY, Qiu SJ, Yamato I, Sho M, Nakajima Y, et al.Overexpression of PD-L1 significantly associates with tumor aggres-siveness and postoperative recurrence in human hepatocellular car-cinoma. Clin Cancer Res 2009;15:971–9.

21. Iwai Y, Ishida M, Tanaka Y, Okazaki T, Honjo T, Minato N. Involvementof PD-L1 on tumor cells in the escape from host immune system andtumor immunotherapy by PD-L1 blockade. Proc Natl Acad Sci U S A2002;99:12293–7.

22. Brown JA, Dorfman DM, Ma FR, Sullivan EL, Munoz O, Wood CR,et al. Blockade of programmed death-1 ligands on dendritic cellsenhances T cell activation and cytokine production. J Immunol2003;170:1257–66.

23. Marzec M, Zhang Q, Goradia A, Raghunath PN, Liu X, Paessler M,et al. Oncogenic kinase NPM/ALK induces through STAT3 expressionof immunosuppressive protein CD274 (PD-L1, B7-H1). Proc Natl AcadSci U S A 2008;105:20852–7.

24. Xerri L, Chetaille B, Serriari N, Seriari N, Attias C, Guillaume Y, et al.Programmed death 1 is a marker of angioimmunoblastic T-cell lym-phoma and B-cell small lymphocytic lymphoma/chronic lymphocyticleukemia. Hum Pathol 2008;39:1050–8.

25. Lam LT, Wright G, Davis RE, Lenz G, Farinha P, Dang L, et al.Cooperative signaling through the signal transducer and activatorof transcription 3 and nuclear factor-{kappa}B pathways in subtypesof diffuse large B-cell lymphoma. Blood 2008;111:3701–13.

26. Kalaitzidis D, Davis RE, Rosenwald A, Staudt LM, Gilmore TD. Thehuman B-cell lymphoma cell line RC-K8 has multiple genetic altera-tions that dysregulate the Rel/NF-kappaB signal transduction path-way. Oncogene 2002;21:8759–68.






27. Liu J, Hamrouni A, Wolowiec D, Coiteux V, Kuliczkowski K, Hetuin D,et al. Plasma cells from multiple myeloma patients express B7-H1(PD-L1) and increase expression after stimulation with IFN-{gamma}and TLR ligands via a MyD88-, TRAF6-, and MEK-dependent path-way. Blood 2007;110:296–304.

28. Yamamoto R, Nishikori M, Kitawaki T, Sakai T, Hishizawa M, TashimaM, et al. PD-1-PD-1 ligand interaction contributes to immunosup-pressive microenvironment of Hodgkin lymphoma. Blood 2008;111:3220–4.

29. Hans CP, Weisenburger DD, Greiner TC, Gascoyne RD, Delabie J, OttG, et al. Confirmation of the molecular classification of diffuse largeB-cell lymphoma by immunohistochemistry using a tissue microarray.Blood 2004;103:275–82.

30. Wilcox RA, Feldman AL, Wada DA, Yang ZZ, Comfere NI, Dong H,et al. B7-H1 (PD-L1, CD274) suppresses host immunity in T-celllymphoproliferative disorders. Blood 2009;114:2149–58.

31. Andorsky DJ, Yamada R, Steward KK, De Vos S, Said J, TimmermanJM. Spectrum of expression and biological activity of programmeddeath ligand 1 (PD-L1) in non-Hodgkin's lymphomas. ASH AnnuMeeting Abstr 2008;112:4140.

32. Rabinovich GA, Gabrilovich D, Sotomayor EM. Immunosuppressivestrategies that are mediated by tumor cells. Annu Rev Immunol2007;25:267–96.

33. Rosenwald A, Wright G, Leroy K, Yu X, Gaulard P, Gascoyne RD, et al.Molecular diagnosis of primary mediastinal B cell lymphoma identifiesa clinically favorable subgroup of diffuse large B cell lymphoma relatedto Hodgkin lymphoma. J Exp Med 2003;198:851–62.

34. Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A,et al. Distinct types of diffuse large B-cell lymphoma identified by geneexpression profiling. Nature 2000;403:503–11.

35. Fu K, Weisenburger DD, Choi WW, Perry KD, Smith LM, Shi X, et al.Addition of rituximab to standard chemotherapy improves the survivalof both the germinal center B-cell-like and non-germinal center B-cell-like subtypes of diffuse large B-cell lymphoma. J Clin Oncol2008;26:4587–94.

36. WilsonWH, Dunleavy K, Pittaluga S, Hegde U, Grant N, Steinberg SM,et al. Phase II study of dose-adjusted EPOCH and rituximab inuntreated diffuse large B-cell lymphoma with analysis of germinalcenter and post-germinal center biomarkers. J Clin Oncol 2008;26:2717–24.

37. Davis RE, Brown KD, Siebenlist U, Staudt LM. Constitutive nuclearfactor kappaB activity is required for survival of activated B cell-likediffuse large B cell lymphoma cells. J Exp Med 2001;194:1861–74.

38. Farinha P, Masoudi H, Skinnider BF, Shumansky K, Spinelli JJ, Gill K,et al. Analysis of multiple biomarkers shows that lymphoma-asso-

ciated macrophage (LAM) content is an independent predictor ofsurvival in follicular lymphoma (FL). Blood 2005;106:2169–74.

39. Steidl C, Lee T, Shah SP, Farinha P, Han G, Nayar T, et al. Tumor-associated macrophages and survival in classic Hodgkin's lym-phoma. N Engl J Med;362:875–85.

40. Hasselblom S, Hansson U, Sigurdardottir M, Nilsson-Ehle H, Ridell B,Andersson PO. Expression of CD68þ tumor-associatedmacrophagesin patients with diffuse large B-cell lymphoma and its relation toprognosis. Pathol Int 2008;58:529–32.

41. Lenz G, Wright G, Dave SS, Xiao W, Powell J, Zhao H, et al. Stromalgene signatures in large-B-cell lymphomas. N Engl J Med 2008;359:2313–23.

42. Yang ZZ, Novak AJ, Stenson MJ, Witzig TE, Ansell SM. IntratumoralCD4þCD25þ regulatory T-cell-mediated suppression of infiltratingCD4þ T cells in B-cell non-Hodgkin lymphoma. Blood 2006;107:3639–46.

43. Nattamai D, Neelapu S. PD-1 expression is markedly upregulated onintratumoral CD4þ and CD8þ T cells in follicular lymphoma and isassociated with T-cell exhaustion. Blood (ASH Meeting Abstracts)2007;110:Abstract 2749.

44. Chu F, Foglietta M, Qin H, Sharma R, Yi Q, Rodionov G, et al. In vitroand in vivo effects of CT-011, a humanized anti-PD-1 monoclonalantibody, in combination with rituximab against human B-cell lym-phomas. Blood (ASH Meeting Abstracts) 2009;114:Abstract 724.

45. Merz H, Fliedner A, Orscheschek K, Binder T, Sebald W, M€uller-Hermelink HK, et al. Cytokine expression in T-cell lymphomas andHodgkin's disease. Its possible implication in autocrine or paracrineproduction as a potential basis for neoplastic growth. Am J Pathol1991;139:1173–80.

46. Merz H, Lange K, Gaiser T, M€uller A, Kapp U, Bittner C, et al.Characterization of a novel human anaplastic large cell lymphomacell line tumorigenic in SCID mice. Leuk Lymphoma 2002;43:165–72.

47. Willers J, Dummer R, Kempf W, Kundig T, Burg G, Kadin ME. Pro-liferation of CD30þ T-helper 2 lymphoma cells can be inhibited byCD30 receptor cross-linking with recombinant CD30 ligand. ClinCancer Res 2003;9:2744–54.

48. Boulland ML, Meignin V, Leroy-Viard K, Copie-Bergman C, Bri�ere J,Touitou R, et al. Human interleukin-10 expression in T/natural killer-cell lymphomas: association with anaplastic large cell lymphomas andnasal natural killer-cell lymphomas. Am J Pathol 1998;153:1229–37.

49. Chiang AC, Massague J. Molecular basis of metastasis. N Engl J Med2008;359:2814–23.

50. Weng AP, Shahsafaei A, Dorfman DM. CXCR4/CD184 immunoreac-tivity in T-cell non-Hodgkin lymphomas with an overall Th1� Th2þ

immunophenotype. Am J Clin Pathol 2003;119:424–30.

Andorsky et al.





2011;17:4232-4244. Published OnlineFirst May 3, 2011.Clin Cancer Res David J. Andorsky, Reiko E. Yamada, Jonathan Said, et al. CellsLymphomas and Inhibits the Activity of Tumor-Associated T

Hodgkin−Programmed Death Ligand 1 Is Expressed by Non

Updated version

10.1158/1078-0432.CCR-10-2660doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2011/08/08/1078-0432.CCR-10-2660.DC1Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/17/13/4232.full#ref-list-1

This article cites 48 articles, 26 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/17/13/4232.full#related-urls

This article has been cited by 43 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

SubscriptionsReprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://clincancerres.aacrjournals.org/content/17/13/4232To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-10-2660

http://clincancerres.aacrjournals.org/content/suppl/2011/08/08/1078-0432.CCR-10-2660.DC1

http://clincancerres.aacrjournals.org/content/17/13/4232.full#ref-list-1

http://clincancerres.aacrjournals.org/content/17/13/4232.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/17/13/4232


Programmed Death Ligand 1 Is Expressed by Non Hodgkin ... · Programmed Death Ligand 1 Is Expressed by Non–Hodgkin Lymphomas and Inhibits the Activity of Tumor-Associated T Cells

Documents