Immune regulation of canine tumour and macrophage PD-L1 … · cancer,checkpoint...

Original Article DOI: 10.1111/vco.12197

Immune regulation of canine tumourand macrophage PD-L1 expression

G. Hartley, E. Faulhaber, A. Caldwell, J. Coy, J. Kurihara, A. Guth,D. Regan and S. Dow

Department of Clinical Sciences, Flint Animal Cancer Center, Colorado State University, Ft. Collins, CO, USA

AbstractExpression of programmed cell death receptor ligand 1 (PD-L1) on tumor cells has been associated

with immune escape in human and murine cancers, but little is known regarding the immune

regulation of PD-L1 expression by tumor cells and tumor-infiltrating macrophages in dogs. Therefore,

14 canine tumor cell lines, as well as primary cultures of canine monocytes and macrophages, were

evaluated for constitutive PD-L1 expression and for responsiveness to immune stimuli. We found that

PD-L1 was expressed constitutively on all canine tumor cell lines evaluated, although the levels of

basal expression were very variable. Significant upregulation of PD-L1 expression by all tumor cell

lines was observed following IFN-𝛾 exposure and by exposure to a TLR3 ligand. Canine monocytes

and monocyte-derived macrophages did not express PD-L1 constitutively, but did significantly

upregulate expression following treatment with IFN-𝛾 . These findings suggest that most canine

tumors express PD-L1 constitutively and that both innate and adaptive immune stimuli can further

upregulate PD-L1 expression. Therefore the upregulation of PD-L1 expression by tumor cells and by

tumor-infiltrating macrophages in response to cytokines such as IFN-𝛾 may represent an important

mechanism of tumor-mediated T-cell suppression in dogs as well as in humans.

Keywordscancer, checkpointmolecule, dog, immune

Introduction

The adaptive immune system depends on a net-work of numerous signals, both stimulatory andinhibitory. Balance in these signals is crucial tomaintain self-tolerance while still protecting thehost from pathogens and diseases like cancer.Under normal conditions, immune checkpointmolecules prevent autoimmunity by downregulat-ing the tempo of T-cell responses. Dysregulationof immune checkpoint molecule expression is,however, a mechanism by which many cancersevade elimination by the host. The PD-1/PD-L1axis is an important immune checkpoint pathwaythat plays a critical role in maintaining peripheraltolerance by downregulating T-cell activation andproliferation.1

Programmed death protein 1 (PD-1) is a cellsurface protein molecule commonly upregulatedon tumour-infiltrating lymphocytes,2 and its

primary role is to limit the activity of T-cells duringinflammatory responses.3 It is induced on T-cellsupon activation as a mode of self-regulation andis highly expressed on regulatory T-cells, whichis thought to enhance their proliferation to pro-vide physiological homeostasis.4 Expression ofthe transmembrane-protein programmed death-1ligand 1 (PD-L1) is increased on many tumours,5

and PD-L1 is also expressed on tumour-infiltratinglymphocytes.6 Tumour-associated macrophagesexpress PD-L1 as well and this is thought tobe one of the mechanisms by which thesemacrophages immunosuppressive effects in thetumour environment.7–9 Chronic exposure toantigen and to PD-L1 leads to T-cell exhaustion,which is characterized by poor effector function,low proliferative activity and the inability to persistas memory T-cells.10

PD-L1 expression on human tumour cellshas been found to be negatively correlated with

Correspondence address:Dr Steven DowDepartment of ClinicalSciencesFlint Animal Cancer CenterColorado State UniversityFt. CollinsCO 80523USAe-mail:[email protected]

© 2016 John Wiley & Sons Ltd 1

2 G. Hartley et al.

prognosis and patient survival in colon, cervical,pancreatic, breast, ovarian, renal cell, hepato-cellular, non-small cell lung, melanoma andoesophageal cancers11–20. In mice, PD-L1 hasbeen found to be expressed by ovarian, myeloma,lung, melanoma and mammary cancers,2,21–23

where it is involved with the escape of the tumourcells from the immune system. A recent study utiliz-ing bovine lymphocytes reported that crosslinkingof PD-L1 by PD-1-Ig increased cell death anddecreased cytokine production in PD-L1 express-ing cells, suggesting a role for PD-L1 in inducingimmunosuppression.24 The upregulation of PD-L1in inflamed tumours has been found to be mediatedprimarily by IFN-𝛾 produced by tumour infiltratingT-cells in humans and mice.25,26

The innate immune response has also been foundto play a role in the regulation of PD-L1 expressionin tumours. Toll-like receptors (TLRs) can recog-nize a wide range of pathogen-associated molecularpatterns (PAMPs), which may also include certaintumour antigens.27 TLR expression is upregulatedon many human tumours including colon, gastric,prostate, breast, ovary and brain tumours, wheresignalling by the TLRs has been implicated inaugmenting the secretion of immune-modulatorycytokines.28 For example, TLR2 and TLR4 ligandshave been shown to protect leukaemia blast cellsfound in acute myeloid leukaemia from T-cellkilling through the induced expression of PD-L1.29

Canine PD-1 and PD-L1 genes are conservedamong canine breeds, and cross-reactive antibod-ies that also recognize canine PD-L1 have beenidentified.30 However, the expression of PD-L1by a broad panel of different canine tumour typeshas not been previously investigated, nor hasthe immune regulation of canine PD-L1 expres-sion on tumour cells been studied. Therefore, thepurpose of the current study was to investigatePD-L1 expression by a series of canine tumourcell lines and to determine how PD-L1 expressionon these cells was regulated by IFN-𝛾 and also byTLR ligands. Because PD-L1 is also expressed bytumour-infiltrating macrophages in humans, weinvestigated the expression of PD-L1 by caninemacrophages generated from blood monocytes,and the effects of cytokines on macrophage PD-L1expression. A recently developed canine PD-L1

antibody was used in these studies, and expressionof PD-L1 by canine tumour cell lines and caninemonocytes and macrophages was assessed by flowcytometry and by immunocytology. The tumourPD-L1 response to treatment with recombinantcanine IFN-𝛾 and TLR ligands was also assessed,as was the PD-L1 response to cytokine-enrichedconditioned medium derived from activated canineT-cells.

We found that all canine tumour cell linesscreened expressed low basal levels of PD-L1,though there was a high degree of variabilitybetween tumour cell lines, especially amongsthaematopoietic-origin tumours. In all tumour linesevaluated, PD-L1 expression could be stronglyupregulated by IFN-𝛾 and certain TLR ligands. Incontrast, canine monocytes and monocyte-derivedmacrophages did not constitutively express PD-L1though canine monocyte-derived macropahgeswere responsive to treatment with IFN-𝛾 andexhibited significant PD-L1 upregulation. Thesefindings indicate that canine tumours may broadlyexpress the checkpoint molecule ligand PD-L1, andthat upregulation of PD-L1 expression on tumourcells and tumour infiltrating macrophages by T-cellcytokines such as IFN-𝛾 may be an importantdeterminant of overall tumour immunity in dogs.

Materials and methodsCell lines

Fourteen different canine tumour cell lines wereevaluated in this study. All cell lines were val-idated and screed to be genetically unique.31

Melanoma (MEL) cells included Talsky, Shadowand Jones (Colorado State University – CSU).Canine osteosarcoma (OS) cells included Abrams(CSU), D17 (ATCC) and McKinley (CSU). Caninehemangiosarcoma (HSA) cells included DEN-HSA(University of Wisconsin) and SB (University ofMinnesota). The Bliley cell line was a transitionalcell carcinoma (TCC) from CSU while the Oswaldcell line was a T-cell lymphoma (LSA) from OhioState University – OSU. The C2 cell line was amast cell tumour (MCT) from the University ofCalifornia, San Francisco. The CTAC cell line was athyroid cancer (TA) from Auburn University andcanine histiocytic sarcomas (HS) included Nike(CSU) and DH82 (ATCC).

© 2016 John Wiley & Sons Ltd, Veterinary and Comparative Oncology, doi: 10.1111/vco.12197

Regulation of canine PD-L1 expression 3

Tumour cell culture

Tumour cells were grown in MEM medium (Gibco,Grand Island, NY, USA) supplemented with 10%FBS (Atlas Biologicals, Fort Collins, CO, USA)and 5% CTM [10 000 μg/mL Pen/Strep, 200 mML-glutamine, 10 mM essential amino acids withoutL-glutamine, 10 mM non-essential amino acids and7.5% bicarbonate solution (all from Gibco)]. Thecells were cultured in standard plastic tissue cultureflasks (Cell Treat, Shirley, MA, USA) and stronglyadherent cells were harvested by treatment with0.25% Trypsin/1 mM ethylenediaminetetraaceticacid (EDTA) (Gibco) followed by Trypsin/EDTAinactivation with media. Viability of cells wasdetermined using 0.4% trypan blue stain (Gibco)for live and dead cell discrimination.

After tumour cells were harvested from culture,1.0× 105 viable cells were plated in triplicate wellsof 24-well polystyrene cell culture plates (Falcon,Durham, NC, USA). The cells were then treatedwith cytokines, supernatants from Concavalin A(ConA) activated peripheral blood mononuclearcells (PBMCs), or TLR ligands for 24 h at 37 ∘C. Thesupernatants were generated by incubating primarycanine PBMCs in 10 μg/mL conA (Sigma-Aldrich,St. Louis, MO, USA) overnight, collecting thesupernatants and centrifuging to remove theremaining cells out of the supernatants. For anal-ysis, the cells were trypsinized and washed withmedia before transfer to a round bottom 96-wellplate (Falcon) for immunostaining. The wells werewashed with FACS buffer and centrifugation tocollect the cell pellets, and the cells were stained forflow cytometry.

Macrophage culture

Canine macrophages were derived from peripheralblood monocytes obtained from healthy dogs.Briefly, PBMCs were obtained using lympho-cyte separation medium (LSM; MP Biomedicals,Solon, OH, USA) and plated on fibronectin-coatedwells in 24-well polystyrene cell culture plates(fibronectin was obtained from Sigma-Aldrich).Non-adherent cells were removed after 18 h inculture and adherent monocytes were cultured forseven days in 10 ng/mL human M-CSF (Peprotech,Rocky Hill, NJ, USA) in DMEM medium (Gibco)

supplemented with 10% FBS. The growth mediawas changed and fresh huM-CSF (10 ng/mL) wasadded every 2 days.

Cytokines and TLR reagents

Recombinant canine IFN-𝛾 was obtained fromR&D Systems, Minneapolis, MN and titratedat 0.1, 1, 10 and 100 ng/mL in tumour culturemedium. Tumour cells were cultured in the pres-ence of IFN-𝛾 in complete medium for 24 hprior to analysis of PD-L1 expression. For themajority of studies, IFN-𝛾 was used at 10 ng/mLbased on the results of the titration studies. Polyi-nosinic:polycytidylic acid, (pI:C) was purchasedfrom InVivogen (San Diego, CA, USA) and wasused at 10 μg/mL. Plasmid DNA (pDNA) was pro-duced by Juvaris Biotherapeutics, Inc (Pleasanton,CA, USA) and used at 10 μg/mL. Lipopolysaccha-ride (LPS, E. coli serotype 0111:B4) was obtainedfrom Sigma-Aldrich and used at 1 μg/mL. R848from InVivogen was used at 10 μg/mL. Theseconcentrations were chosen based on the work-ing concentration ranges established by in vitrotitration studies and by ranges suggested bymanufacturers. The activity of TLR agonists instimulating canine TLRs was assessed by stimula-tion of canine monocyte-derived macrophages andmeasurement of IL-6 secretion (data not shown).

Antibodies

A murine anti-canine PD-L1 monoclonal antibody(clone 4 F9) was used in these studies to detectPD-L1 expression. This antibody was developedby Merck Research Laboratories (Kenilworth, NJ,USA) and was used for flow cytometry, immunoflu-orescence staining and Western blotting studies.An isotype matched, irrelevant antibody was usedat the same concentration (eBioscience, San Diego,CA, USA). Binding of mouse mAbs was detectedusing a labelled donkey anti-mouse secondary anti-body (Jackson ImmunoResearch, West Grove, PA,USA). Mouse anti-human CD11b (clone Bear1)was obtained from Immunotech-Beckman Coulter(Marseilles, France) and used for flow cytomet-ric evaluation of primary canine macrophages.A polyclonal goat anti-canine IFN-𝛾 antibodyfrom Novus Biologicals (Littleton, CO, USA) was


4 G. Hartley et al.

Figure 1. PD-L1 expression by canine tumour cell lines. Canine tumour cells were cultured in 24-well plates overnight incomplete medium and stained the following day for detection of PD-L1 expression by flow cytometry. The cells wereincubated with the appropriately matched concentration of irrelevant isotype (grey filled line) or PD-L1 antibody (bold whiteline) and the data were analysed using FlowJo Software. PD-L1 expression was determined with respect to the geometricmean fluorescence intensity of the isotype. This experiment was repeated four times with similar results each time.

used at 5 μg/mL to neutralize IFN-𝛾 in activatedPBMC-conditioned media for 30 min prior totreating the tumour cells, and an irrelevant controlwas a polyclonal goat IgG antibody from JacksonImmunoResearch, used at the same concentration.

Western blot

A standard Western blotting protocol from(Bio-Rad, Hercules, CA, USA) was followed.Briefly, the tumour cells were harvested by scrap-ing, then lysed in the presence of protease inhibitors

(Thermo Fisher Scientific, Waltham, MA, USA)and the protein concentration was determinedusing a Pierce BCA protein assay kit (ThermoFisher Scientific). Samples were prepared undernon-reducing, denaturing (boiled) conditions,and 20 μg total protein was loaded into a 4–15%gradient mini PROTEAN TGX BioRad gel. Non-fatdry milk (5%) was used for blocking, and 4 F9 mAbwas used to probe for PD-L1 protein, followedby a peroxidase-conjugated donkey anti-mousesecondary. The blots were visualized using BioRadClarity Western ECL substrate.



Figure 2. PD-L1 expression by canine tumour cell lines as assessed by immunofluorescence imaging. Canine melanoma,osteosarcoma, transitional cell carcinoma and histiocytic sarcoma cells were grown as monolayers on glass coverslips andstained for immunofluorescence with either PD-L1 (red) or irrelevant isotype antibody and with DAPI nuclear stain (blue).The coverslips were mounted on glass slides and the images were obtained using confocal microscopy, with instrumentsetting held constant between slides in order to compare the relative intensity of PD-L1 staining between tumour cell lines.

Flow cytometry

Tumour cells were harvested from culture using0.25% trypsin/1 mM EDTA (Gibco), washed withPBS and incubated in 10% normal donkey serum(Jackson ImmunoResearch) to reduce non-specificbinding of antibodies. To detect expression ofcanine PD-L1 on the surface of the cells, appro-priately diluted PD-L1 4 F9 mAb or concentrationmatched isotype control MOPC-21 mAb (BioXcell,West Lebanon, NH, USA) were added to tumourcells for 20 min at room temperature. The cells werethen washed in FACS buffer and incubated withbiotin-conjugated donkey-anti-mouse (JacksonImmunoResearch) antibody followed by washingand then incubation with streptavidin-conjugatedPE (eBioscience, San Diego, CA, USA). After afinal wash, cells were resuspended in FACS stain-ing buffer and 7-AAD viability dye (eBioscience)was added for dead cell exclusion. The cells wereevaluated for PD-L1 expression using a BeckmanCoulter Gallios flow cytometer (Brea, CA), anddata was analysed using FlowJo Software (Ashland,OR, USA).

Monocyte-derived macrophages were detachedfrom tissue culture plastic with 2 mM EDTA inice-cold PBS for 30 minutes on ice followed bygentle pipetting to wash off cells from the plate.The cells were blocked with 5% normal donkeyserum and dog serum before incubating with 4 F9(or isotype control) along with CD11b, and wereafterwards stained following the same protocol asfor tumour cells.

Immunofluorescence imaging

Tumour cells were cultured on glass coverslipsovernight, with or without IFN-𝛾 (10 ng/mL).The next day, the coverslips were washed, fixedin ice-cold acetone for 10 min, and rehydrated in1X PBS prior to staining. After blocking with 5%donkey serum and Streptavidin block (Vector Lab-oratories, Burlingame, CA, USA), 4 F9 or isotypeantibody with Avidin block (Vector Laboratories)was added in appropriate concentrations. This wasfollowed by biotin-conjugated donkey-anti-mouseIgG and streptavidin-conjugated Cy-3 (Invitrogen).Lastly, the cells were stained with DAPI (Molecular


6 G. Hartley et al.

A

Figure 3. Effect of IFN-𝛾 stimulation on PD-L1 expression by canine tumour cells. (A) Effects of tumour incubation withtitrated concentrations of recombinant canine IFN-𝛾 . Canine melanoma, osteosarcoma, transitional cell carcinoma andhistiocytic sarcoma cells were cultured with 0 (grey line), 1 (dotted line), 10 (dashed line) or 100 ng/mL rcIFN-𝛾 (bold line)in complete medium overnight. The cells were then immunostained for detection of PD-L1 expression, and flow cytometrywas used to compare the mean fluorescence intensity of each treatment group. This titration experiment was conducted fourtimes, with similar results each time. (B) Treatment of all canine tumour cell lines with 10 ng/mL IFN-𝛾 . Tumour cell lineswere treated with 10 ng/mL rcIFN-𝛾 overnight and stained with an irrelevant isotype antibody (grey filled line) or PD-L1antibody (unstimulated is the dotted line and IFN-𝛾 treated is the bold line) and analysed by flow cytometry for PD-L1expression. These data are representative of six different independent experiments.

Probes, Eugene, OR, USA) and mounted ontoSuperfrost slides (VWR, Radnor, PA, USA)with Fluoromount Aqueous Mounting Medium(Sigma-Aldrich).

Primary tumour tissues were frozen in OCT(Optimal Cutting Temperature compound) and cutto a thickness of 5 μm. After fixing in ice-cold ace-tone for 5 min, they were rehydrated with 1X PBSand stained the same way as the tumour cell linesgrown on coverslips.

RT-PCR

Total RNA was isolated from canine tumour celllines with and without IFN-𝛾 treatment using anRNeasy Mini Kit (Qiagen, Frederick, MD, USA),and this was transcribed into cDNA using a Quan-tiTect Reverse Transcription Kit (Qiagen). SYBRGreen-based PCR (Bio-Rad, Berkely, CA) was con-ducted with the following primer sequences toamplify PD-L1 mRNA (Integrated DNA Technolo-gies, Coralcille, IA): 5′-CCG CCA GCA GGT CAC



B

Figure 3. Continued.

TT-3′ (forward) and 5′ TCC ATT GTC ACA TTGCCA CC-3′ (reverse). This primer pair was vali-dated with an amplification efficiency of 102% andan R2 value of 0.991. Data analysis was based on thefold change= 2-ΔCt method, with normalization ofthe data to the GAPDH housekeeping gene. Rela-tive quantification was calculated using the averagevalue of duplicate samples.

Results

PD-L1 expression and responsiveness to IFN-𝜸

The expression of PD-L1 was assessed on 14 dis-tinct canine tumour cell lines, using flow cytometricanalysis. We found all 14 tumour cell lines expressed

PD-L1 under basal conditions, with the two his-tiocytic sarcoma cell lines expressing the highestlevels (Fig. 1). The cell lines with the lowest levelsof surface PD-L1 expression were the lymphomacell line and the two hemangiosarcoma cell lines.We also assessed the intracellular expression ofPD-L1 by fixing and permeabilizing tumour cells oncoverslips and assessing PD-L1 expression micro-scopically (Fig. 2). We found that all four tumourcell lines screened expressed PD-L1, when stainingintensity was compared with that of the isotype con-trol antibody.

IFN-𝛾 has been reported to upregulate PD-L1expression in human and mouse systems.25,26

Therefore, the effects of canine rIFN-𝛾 on PD-L1


8 G. Hartley et al.

Figure 4. Effect of IFN-𝛾 stimulation on PD-L1 expression by canine tumour cell lines, as assessed by immunofluorescenceimaging. Canine melanoma, osteosarcoma, transitional cell carcinoma and histiocytic sarcoma cells were grown asmonolayers on glass coverslips in medium alone or with 10 ng/mL rcIFN-𝛾 overnight. The next day they were stained forimmunofluorescence with PD-L1 (red) or irrelevant isotype antibodies and DAPI nuclear stain (blue). The coverslips weremounted on glass slides and the images were obtained with the same microscope settings to show the relative intensity ofPD-L1 staining between unstimulated and rcIFN-𝛾 treated cells. This experiment repeated three times to demonstrateconsistency of the assay.

expression on canine tumour cell lines was assessed.We found that there was a titratable PD-L1 upreg-ulation effect from exposure of tumour cells toincreasing concentrations of rIFN-𝛾 (Fig. 3A). Amid-range concentration of IFN-𝛾 (10 ng/mL) wasthereafter selected for the remaining studies. Thisconcentration of IFN-𝛾 was found to upregulatePD-L1 expression by all the tumour cell lines eval-uated except the lymphoma cell lines, though todiffering degrees depending on the specific cell line(Fig. 3B). For example, histiocytic sarcoma cellswere the most responsive to IFN-𝛾 stimulation,upregulating PD-L1 significantly more than othercell lines when comparing MFI between unstim-ulated and IFN-𝛾 treated cells (P-value< 0.001).qRT-PCR for five different cell lines with andwithout IFN-𝛾 treatments showed that, in general,Pdl1 transcript levels increased after exposure ofthe cells to IFN-𝛾 but do not directly correspondto the amount of PD-L1 protein expressed by the

cells when compared between tumour cell lines(Fig. S1, Supporting Information). These resultswere also demonstrated using immunocytology,where PD-L1 expression was increased after treat-ment with IFN-𝛾 (Fig. 4). Finally, the 4 F9 caninePD-L1-specific antibody was shown to bind a pro-tein of the correct predicted molecular weight ofglycosylated PD-L1, 50 kDA. In addition, expres-sion of PD-L1 was increased after treatment ofhistiocytic sarcoma cells with IFN-𝛾 (Fig. 5).

Cytokine regulation of PD-L1

The effect of other immune-regulatory cytokines onPD-L1 expression was also investigated. BecauseT-cell cytokines other than IFN-𝛾 (e.g., IL-4,GM-CSF, TGF-𝛽) have been shown to upregu-late PD-L1 expression on different murine celltypes,32,33 we were interested to determine the neteffect of all cytokines produced by activated T-cellson canine tumour PD-L1 expression. A diverse



Figure 5. Western blot of PD-L1 expression by HS1 cells with and without IFN-𝛾 treatment. Equal numbers of caninehistiocytic sarcoma cells were cultured overnight, with and without IFN-𝛾 treatment. Cell lysates were prepared 24 h laterand immunoblotted, as described in Methods. Lane 1=MW standard, Lane 2=HS-1 lysates, unstimulated, Lane 3=HS-1lysates, IFN-𝛾-stimulated.

mixture of different T-cell cytokines was generatedfrom in vitro activated canine PBMC, using ConAstimulation. The canine IFN-𝛾 concentration in thesupernatants of conA-stimulated PBMC was foundto be 8 ng/mL by ELISA. The T-cell cytokine mix-ture was then evaluated for upregulation of tumourPD-L1 expression. We found that the T-cellcytokine mixture strongly upregulated PD-L1expression (Fig. 6A), above PD-L1 expression lev-els observed following incubation with equivalentamounts of 8 ng/mL IFN-𝛾 alone (Fig. 6B). Thesefindings suggested that IFN-𝛾 in combinationwith other T-cell cytokines can stimulate evengreater upregulation of tumour PD-L1 expression.When IFN-𝛾 was neutralized in the supernatantsof conA-stimulated PBMCs, PD-L1 upregulationby histiocytic sarcoma cells was markedly inhibited(Fig. 7). These findings suggest that IFN-𝛾 is theprimary cytokine produced by activated canineT-cells responsible for upregulating PD-L1 expres-sion on canine tumour cells and macrophages,though other T-cell derived cytokines appeared toplay a role as well in amplifying the overall PD-L1stimulatory effect of IFN-𝛾 .

Regulation of PD-L1 by innate immune stimuli

We also tested the effects of TLR activation ontumour expression of PD-L1. Canine tumourtumour cell lines were treated with ligands forTLR3 (pIC), TLR9 (pDNA), TLR4 (LPS) andTLR7/8 (R848) for 24 h and their PD-L1 expres-sion level was assessed by flow cytometry. Asshown in Fig. 8, treatment with pIC consistentlyupregulated PD-L1 on all four cell lines screened(P-value= 0.048). However, only the histiocyticsarcoma cell line was responsive to the other 3TLR ligands as well with respect to upregulation ofPD-L1 expression (P-value< 0.001).

PD-L1 expression by monocyte-derivedmacrophages

Studies in rodents and humans have found thatPD-L1 is also expressed by myeloid cells, especiallymacrophages, in addition to tumour cells.8,34 There-fore, adherent canine monocytes were evaluated forPD-L1 expression after overnight culture (Fig. 9A)and again after 7 days in culture in human M-CSF(Fig. 9B). Freshly isolated monocytes expressed lowto negative levels of PD-L1 following brief overnight


10 G. Hartley et al.A

B

Fig

ure

6.Eff

ecto

fcyt

okin

espr

oduc

edby

conc

anav

alin

Aac

tivat

edca

nine

T-ce

llson

tum

ourP

D-L

1ex

pres

sion

and

com

pari

son

with

IFN

-𝛾.(

A)P

D-L

1up

regu

latio

nfo

llow

ing

expo

sure

tose

rial

dilu

tions

ofsu

pern

atan

tsfr

omco

nA-s

timul

ated

PBM

Cs.

Can

ine

mel

anom

a,os

teos

arco

ma,

tran

sitio

nalc

ellc

arci

nom

aan

dhi

stio

cytic

sarc

oma

cells

wer

eex

pose

dto

four

dilu

tions

ofsu

pern

atan

tsfr

omC

onA

-stim

ulat

edPB

MC

over

nigh

t.G

roup

sinc

lude

med

ium

only

(ligh

tgre

yfil

led

line)

,1:1

00di

lutio

nof

Con

Asu

pern

atan

ts(d

otte

dlin

e),1

:10

dilu

tion

ofC

onA

supe

rnat

ant(

dash

edlin

e)an

d1:

1di

lutio

nof

Con

Asu

pern

atan

twith

med

ium

(bol

dlin

e).Th

ece

llsw

ere

then

harv

este

dan

dst

aine

dw

ithPD

-L1

forfl

owcy

tom

etri

can

alys

isas

prev

ious

lyde

scri

bed.

This

expe

rim

entw

asre

peat

edfo

urtim

es.(

B)C

ompa

riso

nof

PD-L

1up

regu

latio

nby

supe

rnat

ants

from

conA

-stim

ulat

edPB

MC

sand

by4

ng/m

Lrc

IFN

-𝛾.Th

esa

me

tum

ourc

elll

ines

test

edin

(A)w

ere

incu

bate

dov

erni

ghtw

ithm

ediu

mon

ly(g

rey

fille

dlin

e),4

ng/m

Lrc

IFN

-𝛾(d

otte

dlin

e),o

ra1:

1di

lutio

nof

PBM

CC

onA

supe

rnat

ants

(bol

dlin

e)an

dth

ece

llsw

ere

imm

unos

tain

edfo

rPD

-L1

expr

essio

n24

hla

ter.

Thes

eda

taar

ere

pres

enta

tive

ofth

ree

inde

pend

ente

xper

imen

ts.



Figure 7. IFN-𝛾 neutralization in conditioned mediumfrom conA-stimulated PBMCs. Canine histiocytic sarcomacells were cultured overnight in regular media, conditionedmedium from conA-stimulated PBMCs and conditionedmedium with an IFN-𝛾 neutralizing antibody, orconditioned medium with control antibody. The next day,the cells were then harvested and stained for PD-L1upregulation by flow cytometric analysis, as described inMethods.

culture. Moreover, monocyte-derived macrophagesalso did not express PD-L1 even after 1 week inculture with human M-CSF. However, caninemonocyte-derived macrophages were responsiveto IFN-𝛾 stimulation, as their PD-L1 expressionwas significantly increased (P-value= 0.001) afterovernight treatment with canine rIFN-𝛾 .

PD-L1 expression on primary tumours

Expression of PD-L1 by a canine histiocytic sar-coma tumour biopsy was assessed to compare thestaining intensity to in vitro histiocytic sarcoma celllines. Fresh frozen tumour samples were immunos-tained for PD-L1 expression. PD-L1 was found to behighly expressed in the histiocytic sarcoma tumourtissues, with expression located on both the cellmembrane and within the cytoplasm (Fig. 10). Thelevels of in vivo expression by histiocytic sarcomacells were similar to those expressed by the HS-1cell line in vitro, suggesting that canine histiocyticsarcomas may be somewhat unique with respectto high levels of PD-L1 expression. A larger studyis currently in progress to assess PD-L1 expres-sion by canine tumours, using a panel of differ-ent canine tumour tissue biopsies (Faulhaber, et al.,manuscript in preparation).

Discussion

In this study we observed that all the canine tumourcell lines evaluated expressed detectable levels ofPD-L1 under basal (i.e., unstimulated) conditions(Figs. 1 and 2). Moreover, all the canine tumourcell lines were highly responsive to treatment withIFN-𝛾 (Figs. 3 and 4). The histiocytic sarcomalines were the most responsive to IFN-𝛾 . In addi-tion, the histiocytic sarcoma cell lines were also themost responsive to stimulation with several differ-ent TLR agonists (Fig. 8). The increased expressionof PD-L1 and increased overall sensitivity to acti-vation with immune stimuli may be related to theorigin of the histiocytic sarcoma cell lines from cellsof the macrophage/dendritic cell lineage, whichare normally very responsive to activation by TLRligands.35,36

IFN-𝛾 is recognized as the principle cytokine thatupregulates PD-L1 expression on many tumourtypes in human and mouse models.25,26 Recently,it was reported that IFN-𝛾 also upregulates PD-L1expression on certain canine tumour lines whenused at a high concentration.30 We observed thatall of the canine tumour cell lines evaluated inthe present study upregulated PD-L1 expressionin response to IFN-𝛾 in a dose-dependent man-ner. Moreover, we also found that cytokine-richsupernatants from activated canine T-cells alsoinduced a significant increase in PD-L1 expres-sion over that induced by treatment with IFN-𝛾alone. This finding suggested that the presence ofother immune-modulating cytokines may furtherupregulate PD-L1 expression by canine tumourcells, or that other cytokines may increase theactivity of IFN-𝛾 in inducing PD-L1 upregulation.For example, it has been reported that IL-4 andTNF-𝛼 together can synergistically induce PD-L1expression by human renal cell carcinoma cells.32

Thus, cytokines such as IL-4 and TNF-𝛼 fromT-cells may potentially be active in upregulatingPD-L1 expression on canine tumour cells.

Human tumour cells have been found to expressvarious TLRs and to upregulate PD-L1 followingactivation by ligands for TLR2, TLR3 and TLR4.29,37

We found that the TLR3 agonist pI:C inducedupregulation of PD-L1 on melanoma, osteosar-coma, transitional cell carcinoma, and histiocytic


12 G. Hartley et al.

Figure 8. PD-L1 expression response to tumour cell activation with TLR ligands. Canine melanoma, osteosarcoma,transitional cell carcinoma and histiocytic sarcoma cells were treated with media only (grey filled line), pDNA (long dashedline), LPS (dotted line), pI:C (bold line) or R848 (dashed line) overnight before analysis for PD-L1 expression by flowcytometry. This experiment was repeated six times.

sarcoma cells, while all four TLR agonists evaluated(i.e., agonists for TLR3, TLR4, TLR7/8 and TLR9)triggered upregulated PD-L1 expression on caninehistiocytic sarcoma cells. These data suggest thatinnate immune responses may also play a role inregulating canine tumour PD-L1 expression.

For the canine tumour cell lines evaluated inthis study, PD-L1 expression was observed to belocated on both the cell membrane and withinthe cytoplasm of the same cell. This is differentthan some studies of PD-L1 expression on humantumour cells, where PD-L1 has been reported tobe expressed only on the plasma membrane or inthe cytoplasmic compartment of tumour cells, evenwithin the same tumour sample.38 However, other

studies of human tumour cells have demonstratedboth cytoplasmic and surface staining of PD-L1in the same cells.39,40 Our findings also add addi-tional tumour cell lines and mechanistic insights toa previous study that evaluated PD-L1 expressionby canine tumour cell lines.30

Tumour-associated macrophages (TAM) com-prise a large population of tumour stromal cells.34

We did not observe constitutive PD-L1 expressionfollowing short-term cultures of canine mono-cytes, which is in contrast to findings from aprior study in mice.33 We also have not observedexpression of PD-L1 on circulating canine mono-cytes (Coy et al.; manuscript in preparation).However, we did find that treatment with IFN-𝛾



A B C

Figure 9. PD-L1 expression and regulation by IFN-𝛾 in canine monocyte-derived macrophages. Primary canine monocyteswere cultured as described in Methods, then detached and immunostained with PD-L1 antibody for analysis by flowcytometry for PD-L1 expression. (A) Primary canine monocytes plated overnight. (B) Monocytes incubated with huM-CSF.Primary canine monocytes were cultured in huM-CSF for 1 week to drive macrophage differentiation (isotype is light greyfilled and PD-L1 is bold). (C) Monocyte-derived macrophages stimulated with IFN-𝛾 . Monocyte-derived macrophagescultured in M-CSF were treated overnight with rcIFN-𝛾 (isotype is light grey filled, baseline PD-L1 is dotted and IFN-𝛾treated is bold). These data are representative of results from three separate studies.

Figure 10. PD-L1 expression by canine histiocytic sarcoma tissue biopsy. Tumour biopsy was obtained from a dog withmalignant histiosarcoma and imbedded in OCT and cryosectioned. Tumour tissues were immunostained as described inMethods and imaged by confocal microscopy. Intense expression of PD-L1 (red) was observed throughout the tumoursection. Cell nuclei were stained with DAPI (blue). Immunostaining with an isotype control antibody is displayed in the inset.

significantly upregulated PD-L1 expression bycultured canine macrophages, which is in agree-ment with previous rodent studies. It should alsobe noted that the source of macrophages was dif-ferent in our study (blood) as compared to therodent studies (peritoneal cavity), which may alsoaccount for the differences in constitutive PD-L1expression.

IFN-𝛾 is widely accepted to be apro-inflammatory cytokine. However, IFN-𝛾has also been shown to play a role in inducing theproduction of anti-inflammatory cytokines suchas IL-1Ra and IL-18BP.41These counter-regulatory,immune-suppressive activities of IFN-𝛾 allowIFN-𝛾 to also protect the host from tissue damagecaused by uncontrolled inflammation. Therefore,



the induction of PD-L1 expression may be anothermeans by which IFN-𝛾 acts to dampen the durationand intensity of host inflammatory responses.

Upregulated expression of PD-L1 by tumourcells is an example of a phenomenon known asadaptive immune resistance.42 PD-L1 expressionby tumour cells may therefore be useful as a markerof the level of adaptive immune resistance exhib-ited by tumours,21,43 as well as a predictor of theeffectiveness of PD-1/PD-L1 blockade for cancerimmunotherapy.9,44 It has recently been shown thatthe use of PD-L1 blocking antibodies enhancescytokine secretion by peripheral blood mononu-clear cells and tumour-infiltrating cells in vitro indogs,30 suggesting that PD-L1 blockade may be aneffectiveness means of immunotherapy for certaincanine tumours. Studies are currently underwayin our laboratory assessing the patterns of PD-L1expression by canine tumour cells in vivo (usingtumour biopsies) to determine the degree of PD-L1expression heterogeneity and tumour-type asso-ciation (Faulhaber; manuscript in preparation).The tremendous potential for checkpoint moleculeblockade, based on the results of recent humanclinical trials,45–47 suggests that such an approachalso has considerable merit for treatment of caninecancer.

Acknowledgements

These studies were supported by Merck ResearchLaboratories and by the Shipley Foundation.

Supporting Information

Additional Supporting Information may be foundin the online version of this article:

Figure S1. qRT-PCR of canine tumour cell lineswith and without IFN-𝛾 treatment. Five caninetumour cell lines were cultured overnight in regu-lar media or media with IFN-𝛾 . RNA was extractedfor qRT-PCR, and Ct values were normalized to thehousekeeping gene GAPDH. Fold increase was cal-culated as 2−ΔCt.

References1. Freeman GJ, Long AJ, Iwai T, Bourque K, Chernova

T, Nishimura H, et al. Engagement of the PD-1

immunoinhibitory receptor by a novel B7 familymember leads to negative regulation of lymphocyteactivation. The Journal of Experimental Medicine2000; 192: 1027–1034.

2. Duraiswamy J, Freeman GJ and Coukos G.Therapeutic PD-1 pathway blockade augments withthe other modalities of immunotherapy T-cellfunction to prevent immune decline in ovariancancer. Cancer Research 2013; 73: 6900–6912.

3. Ishida Y, Agata Y, Shibahara K and Honjo T.Induced expression of PD-1, a novel member of theimmunoglobulin gene superfamily, uponprogrammed cell death. The EMBO Journal 1992;11: 3887–3895.

4. Francisco LM, Salinas VH, Brown KE, Vanguri VK,Freeman GJ, Kuchroo VK, et al. PD-L1 regulates thedevelopment, maintenance, and function of inducedregulatory T cells. The Journal of ExperimentalMedicine 2009; 206: 3015–3029.

5. D’Inecco A, Andreozzi M, Ludovini V, Rossi E,Capodanno A, Landi L, et al. PD-L1 expression inmolecularly selected non-small-cell lung cancerpatients. British Journal of Cancer 2014; 112(1):95–102. doi:10.1038/bjc.2014.555.

6. D’Angelo SP, Shoushtari AN, Agaram NP, Kuk D,Qin LX, Carvajal RD, et al. Prevalence oftumor-infiltrating lymphocytes and PD-L1expression in the soft tissue sarcomamicroenvironment. Human Pathology 2014; 46(3):357–365. doi:10.1016/j.humpath.2014.11.001.

7. Mantovani A, Sica A, Sozzani S, Allavena P, VecchiA and Locati M. The chemokine system in diverseforms of macrophage activation and polarization.Trends in Immunology 2004; 25: 677–686.

8. Chanmee T, Ontong P, Konno K and Itano N.Tumor-associated macrophages as major players inthe tumor microenvironment. Cancers 2014; 6:1670–1690.

9. Kang FB, Wang L, Li D, Zhang YG and Sun DX.Hepatocellular carcinomas promotetumor-associated macrophage M2-polarization viaincreased B7-H2 expression. Oncology Reports 2015;33: 274–282.

10. Wherry EJ. T cell exhaustion. Nature Immunology2011; 12: 492–499.

11. Shi SJ, Wang LJ, Guo ZY, Wei M, Meng YL, YangAG, et al. B7-H1 expression is associated with poorprognosis in colorectal carcinoma and regulates theproliferation and invasion of HCT116 colorectalcancer cells. PLoS ONE 2013; 8.doi:10.1371/journal.pone.0076012.

12. Karim R, Jordanova ES, Piersma SJ, Kenter GG,Chen L, Boer JM, et al. Tumor-expressed B7-H1and B7-DC in relation to PD-1+ T-cell infiltration



and survival of patients with cervical carcinoma.Clinical Cancer Research 2009; 15: 6341–6347.doi:10.1158/1078-0432.CCR-09-1652.

13. Nomi T, Sho M, Akahori T, Hamada K, Kubo A,Kanehiro H, et al. Clinical significance andtherapeutic potential of the programmed death-1ligand/programmed death-1 pathway in humanpancreatic cancer. Clinical Cancer Research 2007;13: 2151–2157.

14. Muenst S, Schaerli AR, Gao F, Daster S, Trella E,Droeser RA, et al. Expression of programmed deathligand 1 (PD-L1) is associated with poor prognosisin human breast cancer. Breast Cancer Research andTreatment 2014; 146: 15–24.

15. Hamanishi J, Mandai M, Iwasaki M, Okazaki T,Tanaka Y, Yamaguchi K, et al. Programmed celldeath 1 ligand 1 and tumor-infiltrating CD8+ Tlymphocytes are prognostic factors of humanovarian cancer. Proceedings of the National Academyof Sciences of the United States of America 2007; 104:3360–3365.

16. Thompson RH, Kuntz SM, Leibovich BC, Dong H,Lohse CM, Webster WS, et al. Tumor B7-H1 isassociated with poor prognosis in renal cellcarcinoma patients with long-term follow-up.Cancer Research 2006; 66: 3381–3385.

17. Gao Q, Wang W, Qiu S, Yamato I, Sho M, NakajimaY, et al. Overexpression of PD-L1 significantlyassociates with tumor aggressiveness andpostoperative recurrence in human hepatocellularcarcinoma. Clinical Cancer Research 2009; 15:971–979.

18. Azuma K, Ota K, Kawahara A, Hattori S, Iwama E,Harada T, et al. Association of PD-L1overexpression with activation EGFR mutations insurgically resected non-small cell lung cancer.Annals of Oncology 2014; 25: 1935–1940.

19. Hino R, Kabashima K, Kato Y, Yagi H, NakamuraM, Honjo T, et al. Tumor cell expression ofprogrammed cell death-1 ligand 1 is a prognosticfactor for malignant melanoma. Cancer 2010; 116:1757–1766.

20. Ohigashi Y, Sho M, Yamada Y, Tsurui Y, Hamada K,Ikeda N, et al. Clinical significance of programmeddeath-1 ligand-1 and programmed death-1 ligand-2expression in human esophageal cancer. ClinicalCancer Research 2005; 11: 2947–2953.

21. Iwa Y, Ishida M, Tanaka Y, Okazaki T, Honjo T andMinato N. Involvement of PD-L1 on tumor cells inthe escape from host immune system and tumorimmunotherapy by PD-L1 blockade. Proceedings ofthe National Academy of Sciences in the United Statesof America 2002; 99: 12293–12297.

22. Noman MZ, Desantis G, Janji B, Hasmim M, KarrayS, Dessen P, et al. PD-L1 is a novel direct target ofHIF-1𝛼, and its blockade under hypoxia enhancedMDSC-mediated T cell activation. The Journal ofExperimental Medicine 2014; 211: 781–790.

23. Barsoum IB, Smallwood CA, Siemens DR andGraham CH. A mechanism of hypoxia-mediatedescape form adaptive immunity in cancer cells.Cancer Research 2014; 74: 665–674.

24. Ikebuchi R, Konnai S, Okagawa T, Yokoyama K,Nakajima C, Suzuki Y, et al. Influence of PD-L1cross-linking on cell death in PD-L1-expressing celllines and bovine lymphocytes. Immunology 2014;142: 551–561.

25. Spranger S, Spaapen R, Zha Y, Williams J, Meng Y,Ha TT, et al. Up-regulation of PD-L1, IDO, andTregs in the melanoma tumor microenvironment isdriven by CD8+ T cells. Science TranslationalMedicine 2013; 5: 200ra116.

26. Abiko K, Matsumura N, Hamanishi J, Horikawa N,Murakami R, Yamaguchi K, et al. IFN-𝛾 fromlymphocytes induces PD-L1 expression andpromotes progression of ovarian cancer. BritishJournal of Cancer 2015; 112: 1501–1509.

27. Milner RJ, Salute M, Crawford C, Abbot JR andFarese J. The immune response to disialogangliosideGD3 vaccination in normal dogs: a melanomasurface antigen vaccine. Veterinary Immunology andImmunopathology 2006; 114: 273–284.

28. Olbert P, Kesch C, Henrici M, Subtil F, Honacker A,Hegele A, et al. TLR4- and TLR9- dependent effectson cytokines, cell viability, and invasion in humanbladder cancer cells. Urologic Oncology 2014; 33(3):110.319–327. doi:10.1016/j.urolonc.2014.09.016.

29. Berthon C, Driss V, Liu J, Kuranda K, Leleu X, JouyN, et al. In acute myeloid leukemia, B7-H1 (PD-L1)protection of blasts from cytotoxic T cells is inducedby TLR ligands and interferon-gamma and can bereversed using MEK inhibitors. Cancer Immunology,Immunotherapy 2010; 59: 1839–1849.

30. Maekawa N, Konnai S, Ikebushi R, Okagawa T,Adachi M, Takagi S, et al. Expression of PD-L1 ontumor cells and enhancement of IFN-𝛾 productionfrom tumor-infiltrating cells by PD-L1 blockade.PLoS ONE 2014; 9: e98415.

31. O’Donoghue LE, Rivest JP and Duval DL.Polymerase chain reaction-based speciesverification and microsatellite analysis for caninecell line validation. Journal of Veterinary DiagnosticInvestigation 2011; 23: 780–785.

32. Quandt D, Jasinski-Bergner S, Muuler U, Schulze Band Seliger B. Synergistic effects of IL-4 and TNF𝛼on the induction of B7-H1 in renal cell carcinoma



cells inhibiting allogeneic T cell proliferation.Journal of Translational Medicine 2014; 12: 151–162.

33. Yamazaki T, Akiba H, Iwai H, Matsuda H, Aoki M,Tanno Y, et al. Expression of programmed death 1ligands by murine T cells and APC. The Journal ofImmunology 2002; 169: 5538–5545.

34. Labiano S, Palazon A and Melero I. Immuneresponse regulation in the tumormicroenvironment by hypoxia. Seminars inOncology 2015; 42: 378–386.

35. Sakuma C, Sato M, Takenouchi T, Chiba J andKitani H. Critical roles of the WASP N-terminaldomain and btk in LPS-induced inflammatoryresponse in macrophages. PLoS ONE 2012; 7:e30351.

36. Sakuma C, Sato M, Oshima T, Takenouchi T, ChibaJ and Kitani H. Anti-WASP intrabodies inhibitinflammatory responses induced by toll-likereceptors 3, 7, and 9, in macrophages. Biochemicaland Biophysical Research Communications 2015;458: 28–33.

37. Boes M and Meyer-Wentrup F. TLR3 triggeringregulates PD-L1 (CD274) expression in humanneuroblastoma cells. Cancer Letters 2015; 361(1):49–56.

38. Bigelow E, Bever KM, Xu H, Yager A, Wu A, TaubeJ, et al. Immunohistochemical staining of B7-H1(PD-L1) on paraffin-embedded slides of pancreaticadenocarcinoma tissue. Journal of VisualizedExperiments 2013; 71: e4059.

39. Chen L, Deng H, Lu M, Xu B, Wang Q, et al. B7-H1expression associates with tumor invasion andpredicts patient’s survival in human esophagealcancer. International Journal of Clinical andExperimental Pathology 2014; 7: 6015–6023.

40. Cedres S, Ponce-Aix S, Zugazagoitia J, Sansano I,Enguita A, et al. Analysis of expression of

programmed cell death 1 ligand 1 (PD-L1) inmalignant pleural mesothelioma (MPM). PLoSONE 2015; 10: e0121071.

41. Muhl H and Pfeilschifter J. Anti-inflammatoryproperties of pro-inflammatory interferon-𝛾 .International Immunopharmacology 2003; 3:1247–1255.

42. Pardoll D. The blockade of immune checkpoints incancer immunotherapy. Nature Reviews Cancer2012; 12: 252–264.

43. Madore J, Vilain R, Menzies AM, Kakavand H,Wilmott JS, Hyman J, et al. PD-L1 expression inmelanoma shows marked heterogeneity within andbetween patients: implications for anti-PD-1/PD-L1clinical trials. Pigment Cell & Melanoma Research2014; 28(3): 245–253. doi:10.1111/pcmr.12340.

44. Topalian SL, Hodi S, Brahmer JR, Gettinger SN,Smith DC, McDermott DF, et al. Safety, activity, andimmune correlates of anti-PD-1 antibody in cancer.The New England Journal of Medicine 2012; 366:2443–2454.

45. Brahmer JR, Tykodi SS, Chow LQ, Hwu W,Topalian SL, Hwu P, et al. Safety and activity ofanti-PD-L1 antibody in patients with advancedcancer. The New England Journal of Medicine 2014;366: 2455–2465.

46. Ohaegbulam KC, Assal A, Lazar-Molnar E, Yao Yand Zang X. Human cancer immunotherapy withantibodies to the PD-1 and PD-L1 pathway. Trendsin Molecular Medicine 2014; 21(1): 24–33.doi:10.1010/j.molmed2014.10.009.

47. Taube JM, Klein A, Brahmer JR, Xu H, Pan X, KimJH, et al. Association of PD-1, PD-1 ligands, andother features of the tumor immunemicroenvironment with response to anti-PD-1therapy. Clinical Cancer Research 2014; 20:5064–5074.


Immune regulation of canine tumour and macrophage PD-L1 … · cancer,checkpoint...

Documents

Transcript of Immune regulation of canine tumour and macrophage PD-L1 … · cancer,checkpoint...