Summary. Programmed death-ligand 1 (PD-L1) is aninhibitory transmembrane protein that can preventautoimmune response. Upregulated PD-L1 serves as apredictive biomarker for patients who may respond wellto immune checkpoint therapies. However, variableassociations of PD-L1 level with prognoses have beenreported. In this study, a short peptide sequencecorresponding to PD-L1 amino acids 172-187 (from theextracellular Ig-like C-type domain, and with highpredicted antigenicity and hydrophilicity) was used togenerate a monoclonal antibody (mAb). The resultantPD-L1 mAb, clone HC16, was examined for bindingspecificity and reactivity in cancer cell-lines, as assessedby immunocytochemical, immunoblotting, and co-immunoprecipitation. The potential diagnostic andclinical applicability of clone HC16 was further testedusing malignant tissue arrays derived from variouscancer types analyzed with an automatedimmunohistochemical (IHC) staining platform.Additionally, tumor samples from patients diagnosedwith non-small cell lung cancer (NSCLC) were analyzedby western blotting. Clone HC16 showed obviousstaining activity in lung and breast cancer tissues.Interestingly, we observed that PD-L1 level wasnegatively associated with clinical stage in NSCLC.Strong PD-L1 expression tended to be found in patientsdiagnosed with bronchioloalveolar carcinoma (BAC).These results demonstrate that clone HC16 harbors goodtarget specificity and is suitable for further developmentin diagnostic tools to assess PD-L1 expression in humantissues. In addition, our findings also suggest a role forPD-L1 in a non-invasive subtype of lung cancer.

Key words: Programmed death-ligand 1, Monoclonalantibody, Non-small cell lung cancer (NSCLC)

Introduction

Programmed death-ligand 1 (PD-L1, also called B7-H1) is encoded by the cluster of differentiation 274(CD274) gene and functions as an immune inhibitoryreceptor ligand expressed by lymphocytes (Dong et al.,1999) and various cancer cells (Dong et al., 2002).Interaction of PD-L1 with its receptor, PD-1, regulatesthe adaptive immune response to danger signals bypreventing excessive activation of CD8+ T cells and/orCD4+ helper cells as well as cytokine production. PD-1knockout mice exhibit enhanced autoimmunity,indicating a negative regulatory role for the receptor andits ligand. (Nishimura et al., 1999, 2001). PD-L1/PD-1-mediated actions require the phosphatases, Srchomology region 2 domain-containing phosphatase-1(SHP-1) and SHP-2, which bind to the immunoreceptortyrosine-based switch motif (ITSM) and subsequentlytrigger immunosuppression, prevent auto-immunity, andmaintain immune system homeostasis (Freeman et al.,2000; Chemnitz et al., 2004; Keir et al., 2007).

Cancer cells express tumor antigens that arerecognized and stimulate cell lysis by specific CD8+cytotoxic T lymphocytes (CTLs) (Urban and Schreiber,1992; van der Bruggen and Van den Eynde, 1992).However, tumor rejection may be inactivated bynegative regulatory signals in the tumor and itsmicroenvironment (Gajewski et al., 2006). The silencingof tumor-infiltrating lymphocytes (TILs) may be at leastpartially induced by PD-L1-expressing cells, such astumor-associated macrophages (TAMs) (Hartley et al.,2018), stromal cells (O'Malley et al., 2018), or tumorcells (Freeman et al., 2000; Keir et al., 2007; Pardoll,

Novel PD-L1 mAb HC16 reveals upregulation of PD-L1 in BAC subtype Bor-Chyuan Su1*, Chen-Hung Ting2*, Kang-Yun Lee3,4,5, Sheng-Ming Wu4,5, Po-Hao Feng4,5, Yao-Fei Chan4 and Jyh-Yih Chen2,61Department of Anatomy and Cell Biology, School of Medicine, College of Medicine, Taipei Medical University, Taipei City, 2MarineResearch Station, Institute of Cellular and Organismic Biology, Academia Sinica, Jiaushi, Ilan, 3Graduate Institute of ClinicalMedicine, College of Medicine, 4Division of Pulmonary Medicine, Department of Internal Medicine, Shuang Ho Hospital, 5Division ofPulmonary Medicine, Department of Internal Medicine, School of Medicine, College of Medicine, Taipei Medical University, Taipei Cityand 6The iEGG and Animal Biotechnology Center, National Chung Hsing University, Taichung City, Taiwan*equal contribution

Histol Histopathol (2021) 36: 77-89

http://www.hh.um.es

Corresponding Author: Jyh-Yih Chen, Marine Research Station, Instituteof Cellular and Organismic Biology, Academia Sinica, 23-10 DahuenRoad, Jiaushi, Ilan 262, Taiwan. e-mail: zoocjy@gate.sinica.edu.twDOI: 10.14670/HH-18-272

istology andistopathologyH

ORIGINAL ARTICLE Open Access

©The Author(s) 2021. Open Access. This article is licensed under a Creative Commons CC-BY International License.

From Cell Biology to Tissue Engineering

2012). The immune checkpoint inhibitors (ICIs), such asPD-1 and PD-L1 antibodies, were developed tocounteract these immune escape mechanisms bydisrupting immune checkpoint ligand-receptorinteractions and have shown efficacy in many clinicaltrials (Pardoll, 2012; Gonzalez-Cao et al., 2015). Manystudies have indicated that expression of PD-1 or PD-L1may be valuable prognostic markers (Ghebeh et al.,2006; Thompson et al., 2006; Pyo et al., 2017; Chen etal., 2018; Shen and Zhao, 2018; Kahlmeyer et al., 2019;Saito et al., 2019; Shigemori et al., 2019). Indeed, somePD-L1 mAbs, including clones 28-8, 22C3, SP142 andSP263, are commonly used in clinical diagnostic tests(Shukuya and Carbone, 2016). However, heterogenicand variable intra-tumor expression of PD-L1 have beenobserved (Ilie et al., 2016; Parra et al., 2018). As aresult, the prognostic value of PD-L1 is still uncertain. Ameta-analysis of six studies revealed that PD-L1expression was significantly higher in poorlydifferentiated tumors and in patients with poor overallsurvival (OS) in non-small cell lung cancer (NSCLC)(Wang et al., 2015). On the other hand, PD-L1expression was analyzed with the 22C3 antibody in atissue microarray representing a large cohort of NSCLCpatients; the study revealed that higher PD-L1expression is associated with longer OS, indicating thatPD-L1 may be a favorable prognostic factor in early-stage NSCLC (Cooper et al., 2015). Clone 22C3 wasepitope mapped and recognizes the extracellular domainof PD-L1 at amino acid residues 156-178 and 196-206(Schats et al., 2017). Another report indicated that PD-L1 expression was associated with poorer OS andProgression Free Survival (PFS) in stage III NSCLCpatients receiving chemo-radiotherapy (Adam et al.,2015). This study was conducted using the clone E1L3Nantibody, which targets the intracellular domain of PD-L1 (Schats et al., 2017). Yet another study wasconducted to compare the immunohistochemical (IHC)staining of different PD-L1 clones (E1L2N, E1J2J,SP142, SP263, 28-8, and 22C3) in NSCLC (Parra et al.,2018). The results showed similar membrane stainingpatterns for all antibodies, but clone SP263 identifiedmore PD-L1-positive cases (Parra et al., 2018). CloneSP263 was generated using a short peptide based onamino acid residues 272-290 as an immunogen; it wasepitope mapped to both intracellular and extracellulardomains (Schats et al., 2017). These findings indicatethat the expression patterns for PD-L1 in IHC studies areinfluenced by features of the antibody used, such as thebinding affinity and target site. Notably, about 10% ofpatients with low PD-L1 expression still respond to ICIs(Shukuya and Carbone, 2016), indicating that theprognostic and/or predictive value of different PD-L1antibodies remains sub-optimal (Ilie et al., 2016;Shukuya and Carbone, 2016).

Membrane-localized PD-L1 is considered to be theactive form and most useful for clinical evaluation ofmalignant tumors (Escors et al., 2018). Here, wegenerated a novel mAb targeting the extracellular

domain of PD-L1. The antibody was validated usingIHC and immunoblotting of cancer cells and tumortissues from NSCLC patients. Furthermore, PD-L1expression patterns were determined by IHC in variousmalignant tissues.Materials and methods

PD-L1 epitope identification and monoclonal antibodygeneration

To identify possible antigenic regions of human PD-L1 (accession number: EAW58763), in-built algorithmsof the MacVector™ software (v.16.0) (Waterbeach,Cambridge, UK) were used to predict epitopeantigenicity (Parkers and Protrusion antigenicity;Kyte/Doolittle and Hopp/Woods hydrophilicity).Selected peptide sequences were analyzed by an onlineprogram for predicting B-cell epitopes (BepiPred-2.0,Lyngby, DK) (Jespersen et al., 2017) and by the BasicLocal Alignment Search Tool (BLAST) (Rockville Pike,MD, USA) to identify any sequence overlap with otherproteins. The mAb was generated by Cyrusbioscience(New Taipei City, Taiwan). The HC16 peptide(HQVLSGKTTTTNSKRE, PD-L1 residues 172-187)and N-terminus biotinylated full-length and truncatedversions (HQVLSGKT, LSGKTTTT, KTTTTNSK,TTTNSKRE) were synthesized and purified (GLBiochem Ltd., Shanghai, China). For antibodygeneration, mice were immunized with HC16 peptide bystandard protocols. Briefly, mice were subcutaneouslyinjected with synthesized HC16 plus complete adjuvant(Sigma-Aldrich, St. Louis, MO, USA) at day 1 followedby three immunizations done at day 21, 35, 51 withsynthesized HC16 plus incomplete adjuvant (Sigma-Aldrich). Blood samples were obtained from mice formeasurement of serum antibodies by enzyme-linkedimmunosorbent assay (ELISA). The fifth immunizationwas done at day 65 by intravenous injection ofsynthesized HC16 without adjuvant. At day 68, micewere euthanized and spleens were removed for in vitrohybridoma cell production through the 38%polyethylene glycol (PEG)-mediated cell fusionapproach (Greenfield, 2018). The murine myelomaSp2/0-Ag14 cell was used as a fusion partner to producehybridoma. After selection, hybridoma clones weremaintained and processed for the production ofmonoclonal antibody by the mouse ascites method(Jackson et al., 1999). For the epitope mappingexperiment, 100 ng biotinylated peptide, full lengthHC16 (FL) or truncated peptides, were incubated withstreptavidin coated plate (Pierce™, ThermoFisherScientific) for 2 h at room temperature (RT). After threewashes by PBS with 0.05% Tween-20 (PBST), each wellwas incubated with the HC16 antibody (1:1000) for 1 hat RT followed by donkey anti-mouse secondaryantibody (1:5000) (GE Healthcare Life Science,Buckinghamshire, UK) for 1 h at RT. After three washes,developing solution and stop solution were added, and

78Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

the optical density (OD) at 450 nm was measured usinga spectrophotometer (SpectraMax® i3, MolecularDevices, Lagerhausstrasse, Wals, Austria) within 5 min,with a reference wavelength of 655 nm.Cell culture

Cells were purchased from the BioresourceCollection and Research Center (BCRC, Hsinchu,Taiwan) - A549 (BCRC60074), MDA-MB231 (BCRC60425), MCF7 (BCRC 60429) - or the American TypeCulture Collection (ATCC, VA, USA) - HCC827 (ATCCCRL-2868). Cells were cultured as recommended.Hybridoma cells were grown in Iscove’s ModifiedDulbecco’s Medium (Thermo Fisher Scientific Inc.,Grand Island, NY, USA) supplemented with 15% fetal-bovine serum (FBS), 2 mM L-glutamine, andpenicillin/streptomycin at 37°C with 5% CO2. Halfmedium replacements were made until the cell densityreached 1-6×106 cells/mL.Western blotting and co-immunoprecipitation

Total protein extracts were prepared and Westernblots were performed by standard protocols. Primaryantibodies were as follows: PD-L1 (1:1,000), α-tubulin(1:2,000, clone DM1A, Cell Signaling, MA, USA), andGlyceraldehyde-3-phosphate dehydrogenase (GAPDH,1:10,000, clone 6C5, EMD Millipore, Burlington, MA).Membranes were visualized with enhancedchemiluminescence (Merck Millipore, Billerica, MA,USA) and detected by an imaging system (UVP,BioSpectrum™ 500). For the competition assay, themembrane was incubated with the HC16 blockingpeptide prior to the addition of the primary antibody. Forthe co-immunoprecipitation (co-IP), 500 μg equalprotein lysate form the HCC827 cells were harvested forIP using Dynabeads™ magnetic beads (Thermo FisherScientific Inc., Dynal, Oslo, Norway), in accordancewith the recommended procedures.Immunocytochemical and immunohistochemical studies

For immunocytochemical studies, cells stained withantibodies (PD-L1/1:500; Integrinα5/1:500, CellSignaling Inc., MA, USA) were mounted in ProLongGold Antifade Reagent (ThermoFisher Scientific).Samples were observed with a FV3000 laser-scanningconfocal microscope (Olympus), and images wereacquired with FV31S software (Olympus). Tissue arrayswere purchased from SUPER BIO CHIPS (CC5, CCA4,CBA4, and MB4). Paraffin sections were processed bystandard protocols with antigen retrieval for 32 min.Slides were stained using the Ventana BenchMark XTautomated stainer (Ventana, Tucson, AZ). The slideswere incubated with the PD-L1 antibody (1:200) for 16min at 37°C and then overnight at 4°C. After three rinsesin PBS, the slides were incubated with the secondaryantibodies (EnVisionTM System, HRP, anti-mouse,

DakoCytomation). Slides were visualized by 3,3'-diaminobenzidine tetrahydrochloride (DAB) substratechromogen solution (DakoCytomation) and counter-stained with hematoxylin, dehydrated, and mounted.Images were obtained with an inverted microscope(Olympus, IX71) coupled to a digital camera (Olympus),using a 4× (UPlanFI 10× /0.13 PhL) and 20×(LUCPlanFI 20× /0.40 Ph1) objective lens. CellSensstandard software (Olympus) was used for imageacquisition. Quantitation of the PD-L1 signal intensitywas performed with ImageJ software (National Institutesof Health) using the IHC tool box plugins.PD-L1 blockade assay

PD-L1 blockade assay was performed using PD-1[Biotinylated]:PD-L1 Inhibitor Screening Assay Kit(BPS Bioscience) following the recommended protocol.Briefly, 50 μL diluted PD-L1 (100 ng) was coated on a96-well plate. Uncoated wells served as the Ligandcontrol. After three washes with 1× Immuno buffer andblocking for 1 h at RT, 25 μL reaction mixtures (10 μL3× Immuno buffer + 15 μL H2O) were added into eachwell. Five microliters of the same solution was added tothe Ligand control and Positive control groups, while 5μL PD-L1 antibody solution (125, 250 or 500 ng) wasadded into the Sample groups. Diluted biotinylated PD-1 (10 ng) was then added into each well and incubatedfor 2 h at RT. After three washes with 1× Immunobuffer and blocking for 10 min at RT, dilutedstreptavidin-HRP was added and incubated for 10 minat RT. After washing, freshly prepared chemi-luminescent reagent was added to each well.Chemiluminescence was measured with a luminescencereader (Molecular Devices).Statistical analysis

Data from repeated assays were analyzed by Prism 7software (GraphPad Inc). Values represent the mean ±SD, and statistical significance was determined byapplying a two-tailed unpaired Mann-Whitney t-test orone-way ANOVA test. Differences were consideredstatistically significant at P<0.05.Results

Generation of a mAb targeting human PD-L1

A short peptide sequence covering 16 amino acids(residues 172-187) within the Ig-like C2 type domain ofhuman PD-L1 (Fig. 1A) was used for mAb production.Selection of the target peptide was based on antigenicityprediction algorithms (Fig. 1B, boxed by a red line) andvalidated by a web server that predicts B-cell epitopes(Jespersen et al., 2017). The selected antigen partlyoverlaps with that of clone 22C3 but not with the othercommercially available clones (Fig. 1A). Synthesizedpeptide was then used to immunize mice. Antibody-

79Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

producing cells were collected for fusion with myelomacells, and one hybridoma clone with the strong anti-PD-L1 activity was selected for cell expansion and mAbproduction in mice.

Expression pattern of PD-L1 in cancer cell lines

Immunocyctochmical staining of PD-L1 withpurified clone HC16 was performed in epidermal

80Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

Fig. 1. Antigenic analysis of human PD-L1. A. Schematic shows the domain characteristics of PD-L1. The epitope selected for antibody production isshown in yellow. Several epitopes recognized by commercially available PD-L1 antibodies are indicated, including clone 28-8 (brown), clone 22C3(shown in blue; boxed by blue dashed line), clone SP263 (green), clone SP142 (grey line), and clone E1L3N (red line). B. Parker, Protrusion antigenicindex and Kyte/Doolittle, Hopp/Woods hydrophilicity index algorithms were used to predict the potential antigenic sequences. Red boxed regionindicates amino acids 172-187.

81Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

Fig. 2. Expression patterns of PD-L1 in cancer cell-lines. A.Localization of PD-L1 in cancer cells. Cells were stained for PD-L1(green), Integrin5α (red), and nuclei (blue; Hoechst33258). Boxedregions are magnified to the right of the merged image. B. Totallysates from cancer cells were analyzed by Western blot usingantibodies against PD-L1 and GAPDH. For the competition assay,blocking peptide was incubated with the primary antibody beforeblotting. C. HCC827 cell lysates were harvested for co-IP using thePD-L1 antibody. IP eluates were analyzed by Western blot for PD-L1. D. Biotinylated HC16 peptide (FL) and its truncated variants(named a to d) were incubated with streptavidin-coated wells andrecognized by PD-L1 antibody. BSA served as a negative control.E. Total cell lysates were analyzed by Western blot using

antibodies against PD-L1 with or without the incubation of HC16 peptide or its truncated variants. F. Biotinylated PD-1 was incubated with PD-L1-coated wells without or with HC16 antibody. PD-L1 and PD-1 binding was determined with a fluorescence reader and relative luminescence wascalculated by dividing the signal by its ligand control (Lig.). The ELISA assays shown in (D) and (F) represent at least duplicate wells. Data werecollected from three independent assays. Results represent the mean ± SD. *P<0.05; **P<0.01; ***P<0.001. Scale bars: 20 μm.

82Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

Fig. 3. Immunohistochemical analysis of PD-L1 profile inlung cancer. A. Matched NAT (n=9) and lung cancersamples (n=93) were stained for PD-L1 with DAB signalenhancement. Red boxed regions are magnified to theright. B. Quantification of PD-L1 staining intensity in lungcancer and matched NAT samples. C. PD-L1 staining

intensity was normalized to matched NAT and stratified by disease stage. D-F. Correlation analysis of relative PD-L1 level and survival status in stage I(D), II (E), and III NSCLC patients (F). On the right, survival data are shown according to relative PD-L1 expression level. S: Strong; H: high; L: low PD-L1 expression. Blue dashed line divides low and high/strong PD-L1 expression. Orange and black dots respectively denote dead and surviving patients.G-I. PD-L1 expression in stage I (G), stage II (H), and stage III (I) NSCLC patients with various cancer types. Red and green dashed line denote thevalue 1.5 and 3.0; a value less than 1.5 was considered weak PD-L1 expression, between 1.5 and 3.0 was defined as high, and over 3.0 was definedas strong. The red and black dots respectively denote dead and alive patients. ADC: Adenocarcinoma; BAC: Bronchioloalveolar carcinoma; SCC:Squamous carcinoma; LCC: Large cell carcinoma; MEC: Mucoepidermoid carcinoma; CS: Carcinosarcoma; LCNC: Large cell neuroendocrinecarcinoma; MM: Maligant mesothelioma; Ad-SCC: Adenosquamous carcinoma; ACC: Adenoid cystic carcinoma. Quantitative results in (B) and (C)show mean ± SD. Mann-Whitney t-test: ns, not significant; *P<0.05; **P<0.01. Scale bars: left column, 200 μm; right column, 50 μm.

growth factor receptor (EGFR)-mutated HCC827 cells,EGFR-wild-type A549 lung cancer cells, triple-negative MDA-MB-231 breast cancer cells, andestrogen receptor/progesterone receptor (ER/PR)-positive MCF7 cells. PD-L1 was expressed on themembrane and intracellular regions of all tested cells(Fig. 2A, pink arrows indicate membrane-localized PD-L1). Interestingly, different levels of PD-L1glycosylation were observed in different cancers undernormal culture conditions (Fig. 2B), suggesting thatdistinctive mechanisms regulate PD-L1 in the cell lines(Fig. 2B,C). The antibody binding specificity wasconfirmed by competition assays (Fig. 2B, middlepanel). Clone HC16 only recognized full length peptide(FL) and not truncated versions in ELISA (Fig. 2D) andcompetition assays (Fig. 2E). PD-L1 residues Ala-121to Arg-125 are conserved across species and are

required for the interaction with PD-1 (Lin et al.,2008). This site is far from the binding site for cloneHC16, so we tested whether the antibody mightinfluence the PD-L1-PD-1 interaction by an ELISA-based approach. The result showed that clone HC16only has minor effects on the interaction between PD-1and PD-L1 (Fig. 2F).PD-L1 expression may be associated with clinical stageand subtype in lung cancer

We subsequently determined the profiles of PD-L1expression in cancer tissue arrays by IHC with cloneHC16. The intensity of PD-L1 staining was quantifiedin each tissue sample and normalized to the average ofnine matched normal adjacent tissue (NAT) samples(Fig. 3Aa). A normalized value less than 1.5 was

83Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

Fig. 4. Immunoblotting of PD-L1 withlung cancer tissues. Tissue extractsfrom eight lung cancer patients staged Ito IV were analyzed by Western blotusing antibodies against PD-L1 andαTubulin. For the competition assay,blocking peptide was incubated with theprimary antibody before blott ing.General information for each patient isshown below the blot.

considered weak expression (Fig. 3Ab); 1.5 to 3.0 wasdefined as high (Fig. 3Ac), and over 3.0 (Fig. 3Ad)indicated strong expression. Staining was observedwithin tumor cells both intracellularly andextracellularly (Fig. 3Ad). A significant elevation ofPD-L1 intensity was observed in lung cancer tissuesover NAT (Fig. 3B, n=93). Interestingly, the PD-L1level was significantly higher in stage I samplescompared to those from stage II/III patients (Fig. 3C).Correlation analysis further showed that PD-L1expression might be weakly correlated (not statisticallysignificant) with follow-up survival data in stage I/IINSCLC patients (Fig. 3D,E, Spearman r=0.1472 instage I and 0.1984 in stage II patients); while a negativecorrelation was observed in stage III patients (Fig. 3F,Spearman r =-0.06717). Among the 31 stage I patients,21 showed high or strong expression of PD-L1 (Fig.3D). More surviving patients were observed in thisgroup (61.1% [11/18] surviving stage I patients) thanthe stage II/III groups (43.8% [7/16] stage II patientsand 33.3% [3/9] stage III patients were surviving) (Fig.3E,F). Among different cancer types, bronchiolo-alveolar carcinoma (BAC), squamous carcinoma (SCC),and adenocarcinoma (ADC) groups had more patientswith strong PD-L1 expression at stage I, but thenumbers were reduced in stage II/III groups (Fig. 3G-I).Strikingly, all stage I patients (n=6) with BAC showedstrong PD-L1 immunoactivity (Fig. 3G), while PD-L1staining in stage II/III patients was relatively low (Fig.3H,I). Lung cancer tissues were obtained from 12female and 47 male individuals. The median follow-uptime was 70.78 months (range 2-149 months). Tovalidate the results of our IHC studies, we alsomeasured PD-L1 levels in NSCLC clinical samples byimmunoblotting. The results show that three out of fivestage I patients (diagnosed as BAC, SCC and ADC) hadstrong PD-L1 immunoactivity; meanwhile, the level ofPD-L1 was lower in three other patients staged II-IV(Fig. 4).

A similar analysis was then used to evaluate PD-L1expression in breast cancer tissue arrays (Fig. 5A). PD-L1 staining was observed within the tumor cellsintracellularly and extracellularly (Fig. 5Ad).Significantly higher PD-L1 immunoactivity wasobserved in breast cancer tissues compared to nine NATs(Fig. 5B, n=48). No significant differences in PD-L1levels were observed between the stage II and IIIpatients (Fig. 5C). Including all breast cancer types,34.6% (9/26) of stage II patients and 23.8% (5/21) ofstage III patients had strong PD-L1 expression (Fig. 5D-F). All breast cancer tissues were obtained from femalepatients. The median follow-up time was 70.75 months(range 9-84 months). Finally, we analyzed theassociation of PD-L1 expression with three knownpredictive markers in breast cancers (38/48 patients; 22stage II cases and 16 stage III cases). The results showedthat PD-L1 expression was detectable in various breastcancer subtypes (Fig. 5F).

Discussion

In this study, we introduce a novel mAb, cloneHC16, with unique target specificity to human PD-L1.Its utility was validated using cancer cell lines and arraysof malignant tissues. Significant elevation of PD-L1level was observed in both lung cancer and breast cancertissues, and strong PD-L1 expression was particularlynotable in stage I NSCLC patients.

Clone HC16 targets the Ig-like C2 type domain ofPD-L1 (Fig. 1A), which is also targeted by clone 22C3as one of two different binding sites (Schats et al., 2017).The partially overlapping binding site includes residues172-178 (Fig. 1A). An earlier report utilized clone 22C3to show that PD-L1 expression is a favorable prognosticfactor in early stage NSCLC (Cooper et al., 2015). Wealso observed that PD-L1 level is significantly higher instage I patients than stage II/III patients (Fig. 3C) in alimited number of cases. A larger sample size is neededto confirm our findings. The study with clone 22C3further showed that PD-L1 tended to be more highlyexpressed in SCC and LCC than ADC, but thedifferences were not statistically significant; also, highPD-L1 expression in NSCLC was significantlycorrelated with longer OS (Cooper et al., 2015). Thesurvival analysis of stage I NSCLC patients in our studyrevealed that individuals with high or strong PD-L1staining activity might have improved survival status(4/6 surviving BAC patients; 2/3 surviving ADCpatients; 4/7 surviving SCC patients; 2/2 surviving LCCpatients; 1/1 surviving MEC patients) (Fig. 3E-G). Inaddition, we found that all patients diagnosed with BACshowed strong PD-L1 staining (Fig. 3E). BAC is asubtype of ADC characterized by the growth ofmalignant cells over the alveoli and bronchioles butwithout invasion (Brambilla et al., 2001). Patients withsmall non-invasive BACs (tumor size <2 cm) haveexcellent survival after surgical resection (Noguchi et al.,1995). Extensive BAC with an aggressive malignantcomponent may occur with invasion and destruction ofthe pulmonary architecture, and is categorized as amixed subtype of ADC (Brambilla et al., 2001).Therefore, the presence or absence of invasion seems tobe a critical prognostic factor. Follow-up survival datarevealed that only 2/3 ADC patients with strong PD-L1expression (among 8 total stage I patients) were alive(Fig. 3G). Moreover, PD-L1-positive cases tended todecrease with stage: 21/33 stage I cases (63.6%), 16/38stage II cases (42.1%), 9/22 stage III cases (40.9%) (Fig.3D). These results echo the previous finding that PD-L1staining with clone 22C3 may be of prognostic value forearly-stage NSCLC (Cooper et al., 2015). Furthermore,the finding that PD-L1 expression is more relevant inBAC than ADC subtypes suggests that PD-L1 regulatoryactivities may differ among lung cancer types. miR-197is a reported prognostic indicator for NSCLC thatregulates PD-L1 expression through the cyclin-dependent kinase CDC28 protein kinase regulatory

84Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

85Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

Fig. 5. Immunohistochemistry for PD-L1 in breast cancer. A.Matched NAT (n=9) and cancer samples (n=48) were stained for PD-L1 with DAB signal enhancement. Red boxed regions are magnifiedto the right. B. Quantification of PD-L1 staining intensity in cancerand matched NAT samples. C. PD-L1 staining intensity wasnormalized to matched NAT and stratified by disease stage. D-F. PD-L1 expression in stage II (D) and stage III (F) breast cancer patients

with various cancer types. Red and green dashed lines denote the values of 1.5 and 3.0; less than 1.5 was considered weak PD-L1 expression,between 1.5 and 3.0 was defined as high, and over 3.0 was defined as strong. IDC: Infiltrating duct carcinoma; AMC: Atypical medullary carcinoma;MC: Metaplastic carcinoma; IPC: Intraductal papillary carcinoma; SC: Sarcomatoid carcinoma; LC: Lobular carcinoma. Quantitative results shown in (B)and (C) represent the mean ± SD (two-tailed unpaired Mann-Whitney t-test: ns, not significant; *P<0.05). F. Expression of PD-L1 is shown for varioussubtypes of breast cancer at different stages. ER: estrogen receptor; PR: progesterone receptor; HER2: human epidermal growth factor receptor 2;TNBC: ER-/PR-/HER2-. Scale bars: left column, 200 μm; right column, 50 μm.

subunit 1B (CKS1B) and signal transducer and activatorof transcription 3 (STAT3) signaling cascade in chemo-resistant NSCLC cells (Fujita et al., 2015; Mavridis etal., 2015). Expression of miR-197 is associated withlarger tumor size and SCC histotype in early or advancedNSCLC (Mavridis et al., 2015). In addition, a variantform of the cancer stem cell (CSC) marker, CD44v, mayserve as a negative prognostic marker in SCC (Nguyenet al., 2000; Leung et al., 2010). However, in ADC,improved OS was observed for patients with CD44-expressing tumors (Leung et al., 2010). Intriguingly,ADC tumors may contain a population of CD44v-expressing cells that are highly proliferative and resistantto chemotherapy (Nishino et al., 2017). Tumor cells witha stemness phenotype (i.e. CD44v positive) exhibitWNT-mediated constitutive activation of PD-L1 in triplenegative breast cancer (TNBC), which facilitatesimmune evasion (Castagnoli et al., 2019). In addition,PD-L1 promotes OCT4 and Nanog expression in breastCSCs through activation of PI3K/AKT (Almozyan et al.,2017). Constitutive and inducible expression of PD-L1in CD44-positive head and neck SCC suppresses T-cell-mediated immunity (Lee et al., 2016). Furthermore, PD-L1 binds to H-Ras, subsequently triggering Erk-mediated epithelial mesenchymal transition (EMT)signaling to promote Glioblastoma multiforme formationand invasion in a rodent model (Qiu et al., 2018). Atranscriptomic study conducted in human glioblastomacells revealed that PD-L1 significantly alters expressionof genes involved in cell growth/migration/invasion (Qiuet al., 2018), suggesting that PD-L1 is associated withEMT in malignant cells. Alternatively, anti-tumoractivity of PD-L1 has also been reported. Cells with lowPD-L1 expression isolated from two humancholangiocarcinoma cell lines harbor pluripotent stemcell-like characteristics and are highly tumorigenic inimmunodeficient NOD⁄SCID⁄γcnull mice, butknockdown of PD-L1 in the cells suppressestumorigenicity (Tamai et al., 2014). Survival analysisrevealed that low PD-L1 expression correlates with poorprognosis in cholangiocarcinoma patients (Tamai et al.,2014). Taking our finding that PD-L1 is stronglyexpressed in the BAC subtype in NSCLC together withother studies showing high PD-L1 expression at earlystages of lung cancer may have improved survival status,one may hypothesize that a distinct population ofNSCLC cells with very different oncogenic propertiescould exist, and their correlation with the overall PD-L1profile should be carefully evaluated.

Strong PD-L1 expression was detected in differentbreast cancer subtypes by clone HC16 (Fig. 5D-F). Nosignificant difference in PD-L1 staining was observedbetween stage II and III breast cancer patients, but morepatients with strong PD-L1 activity were found in thestage II group (Fig. 5F). Among patients with knownpredictive factors, 4/10 TNBC cases showed strong PD-L1 activity (Fig. 5Ad). Significant upregulation of PD-L1 in TNBC has been reported and validated with IHC(Gatalica et al., 2014; Mittendorf et al., 2014; Ali et al.,

2015; Polonia et al., 2017). However, the expression ofPD-L1 in TNBC tissues can vary when using differentmAbs (Sun et al., 2016). Clones 28-8 and E1L3Nproduced more positive tumor cell staining than cloneSP142; moreover, clone SP142 was almost devoid ofimmune cell staining compared to the two other clones(Sun et al., 2016). The staining profile with clone HC16is much more similar to clones 28-8 and E1L3N, withpositive staining of both immune and tumor cells (Fig.5Ad). Clones 28-8 and E1L3N both recognizeintracellular domains of PD-L1, while clone SP142targets both intracellular and extracellular epitopes(Schats et al., 2017). In this work, we did notexperimentally verify the HC16 binding sites in PD-L1.Thus it is still unclear whether HC16 can also recognizeintracellular portions of PD-L1.

Studies of the correlations between PD-L1expression and breast cancer subtypes or prognosisremain controversial (Stovgaard et al., 2019). Forexample, PD-L1 expression is frequently detected inbasal-like breast cancers and is correlated with improveddisease-free survival (DFS) in ER- breast cancers (Ali etal., 2015). A high level of PD-L1 in TNBC patients wascorrelated with the prevalence of TILs and hadsignificantly longer DFS and OS (AiErken et al., 2017).Increased PD-L1 expression in tumor cells was stronglyassociated with better DFS but not OS in TNBC (Botti etal., 2017). By contrast, other studies showed PD-L1 isassociated with the TNBC subtype and decreased OS(Polonia et al., 2017; Adams et al., 2018; Choi et al.,2018). Indeed, different PD-L1 antibodies show differentbinding characteristics that may lead to distinctive IHCpatterns. Splicing variants of PD-L1 have been identifiedand their cellular functions have been characterized (Heet al., 2005; Nielsen et al., 2005; Hassounah et al., 2019;Mahoney et al., 2019). A secreted form of PD-L1, whichlacks the C-terminal transmembrane domain andcytoplasmic region, has been found to function as aninhibitory immune regulator within the tumor withoutthe need for cell-cell interactions (Mahoney et al., 2019).Conversely, the PD-L1 splicing isoform 2, which lacksthe Ig-like V-type domain (residues 19-132), shows anintracellular localization pattern (He et al., 2005).Therefore, failure of an antibody to bind specificdomains of PD-L1 may not fully reveal the specificexpression patterns of all isoforms.

Overall, a single anti-PD-L1 antibody clone may notfully reveal the status of every patient, since post-transcriptional modulation of PD-L1 may result indifferential immune regulation. Considering therequirement for high diagnostic accuracy to identifypatients who should receive immune therapy, two ormore positive tests may be required to determine themost appropriate treatment options. The novel PD-L1mAb, clone HC16, has excellent PD-L1 targetspecificity. However, further investigations to fullycharacterize the targeted epitope and PD-L1conformation are still required before furtherdevelopment of diagnostic tools.

86Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

Acknowledgements. We thank Dr. Marcus J. Calkins for languageediting. We also thank Cyrusbioscience for the generation of PD-L1mAb and Taipei Medical University, Shuang Ho Hospital in New TaipeiCity for providing the human samples from NSCLC patients. Funding. This work was financially supported by the iEGG and AnimalBiotechnology Center from the Feature Areas Research CenterProgram within the framework of the Higher Education Sprout Project bythe Ministry of Education (MOE-109-S-0023-A) in Taiwan. This workwas partly financially supported by the 109-2313-B-001-007-MY3, 109-2622-B-001-002-CC1, or 108-2313-B-001-006—from the Ministry ofScience and Technology (Taiwan) and a PI quota to Dr. Jyh-Yih Chenfrom the Marine Research Station, Institute of Cellular and OrganismicBiology. This research was also funded by Taipei Medical University,grant number: TMU108-AE1-B22. The funders played no part in studydesign, data collection and analysis, decision to publish, or preparationof the manuscript. Availability of data and materials. The materials and data generatedand/or analyzed in this study are available from the correspondingauthor upon reasonable request.Authors' contributions. BC Su and CH Ting performed the experiments.KY Lee, SM Wu, PH Feng, and YF Chan designed the study. BC-Su,CH-Ting, and JY Chen wrote the manuscript. Ethics approval and consent to participate. The use of human tissuespecimens was approved by the Institutional Review Board at ShuangHo Hospital, Taipei Medical University (IRB No: N201702026). Thisstudy was conducted in accordance with the Declaration of Helsinki.Patient consent for publication. Not applicable.Competing interests. The authors have no competing financial intereststo declare.

References

Adam J., Boros A., Lacas B., Lacroix L., Pignon J.P., Caramella C.,Planchard D., Levy A., Besse B. and Le Pechoux C. (2015).Prognostic value of PDL1 expression in stage III non-small cell lungcancer (NSCLC) treated by chemo-radiotherapy (CRT). Eur. J.Cancer. 51, S604-S604.

Adams T.A., Vail P.J., Ruiz A., Mollaee M., McCue P.A., Knudsen E.S.and Witkiewicz A.K. (2018). Composite analysis of immunologicaland metabolic markers defines novel subtypes of triple negativebreast cancer. Mod. Pathol. 31, 288-298.

AiErken N., Shi H.J., Zhou Y., Shao N., Zhang J., Shi Y.W., Yuan Z.Y.and Lin Y. (2017). High PD-L1 expression is closely associated withtumor-infiltrating lymphocytes and leads to good clinical outcomes inchinese triple negative breast cancer patients. Int. J. Biol. Sci. 13,1172-1179.

Ali H.R., Glont S.E., Blows F.M., Provenzano E., Dawson S.J., Liu B.,Hiller L., Dunn J., Poole C.J., Bowden S., Earl H.M., Pharoah P.D.P.and Caldas C. (2015). PD-L1 protein expression in breast cancer israre, enriched in basal-like tumours and associated with infiltratinglymphocytes. Ann. Oncol. 26, 1488-1493.

Almozyan S., Colak D., Mansour F., Alaiya A., Al-Harazi O., Qattan A.,Al-Mohanna F., Al-Alwan M. and Ghebeh H. (2017). PD-L1promotes Oct4 and Nanog expression in breast cancer stem cells bysustaining PI3K/AKT pathway activation. Int. J. Cancer 141, 1402-1412.

Botti G., Collina F., Scognamiglio G., Rao F., Peluso V., De Cecio R.,

Piezzo M., Landi G., De Laurentiis M., Cantile M. and Di Bonito M.(2017). Programmed death ligand 1 (PD-L1) tumor expression isassociated with a better prognosis and diabetic disease in triplenegative breast cancer patients. Int. J. Mol. Sci. 18.

Brambilla E., Travis W.D., Colby T.V., Corrin B. and Shimosato Y.(2001). The new World Health Organization classification of lungtumours. Eur. Respir. J. 18, 1059-1068.

Castagnoli L., Cancila V., Cordoba-Romero S.L., Faraci S., Talarico G.,Belmonte B., Iorio M.V., Milani M., Volpari T., Chiodoni C., Hidalgo-Miranda A., Tagliabue E., Tripodo C., Sangaletti S., Di Nicola M. andPupa S.M. (2019). Wnt signaling modulates PD-L1 expression in thestem cell compartment of triple-negative breast cancer. Oncogene.38, 4047-4060.

Chemnitz J.M., Parry R.V., Nichols K.E., June C.H. and Riley J.L.(2004). SHP-1 and SHP-2 associate with immunoreceptor tyrosine-based switch motif of programmed death 1 upon primary human Tcell stimulation, but only receptor ligation prevents T cell activation.J. Immunol. 173, 945-954.

Chen R.Q., Liu F., Qiu X.Y. and Chen X.Q. (2018). The prognostic andtherapeutic value of PD-L1 in glioma. Front. Pharmacol. 9, 1503.

Choi S.H., Chang J.S., Koo J.S., Park J.W., Sohn J.H., Keum K.C., SuhC.O. and Kim Y.B. (2018). Differential prognostic impact of strongPD-L1 expression and 18F-FDG uptake in triple-negative breastcancer. Am. J. Clin. Oncol. 41, 1049-1057

Cooper W.A., Tran T., Vilain R.E., Madore J., Selinger C.I., Kohonen-Corish M., Yip P., Yu B., O'Toole S.A., McCaughan B.C., YearleyJ.H., Horvath L.G., Kao S., Boyer M. and Scolyer R.A. (2015). PD-L1 expression is a favorable prognostic factor in early stage non-small cell carcinoma. Lung Cancer 89, 181-188.

Dong H., Zhu G., Tamada K. and Chen L. (1999). B7-H1, a thirdmember of the B7 family, co-stimulates T-cell proliferation andinterleukin-10 secretion. Nat. Med. 5, 1365-1369.

Dong H., Strome S.E., Salomao D.R., Tamura H., Hirano F., Flies D.B.,Roche P.C., Lu J., Zhu G., Tamada K., Lennon V.A., Celis E. andChen L. (2002). Tumor-associated B7-H1 promotes T-cell apoptosis:a potential mechanism of immune evasion. Nat. Med. 8, 793-800.

Escors D., Gato-Canas M., Zuazo M., Arasanz H., Garcia-Granda M.J.,Vera R. and Kochan G. (2018). The intracellular signalosome of PD-L1 in cancer cells. Signal. Transduct. Target Ther. 3, 26.

Freeman G.J., Long A.J., Iwai Y., Bourque K., Chernova T., NishimuraH., Fitz L.J., Malenkovich N., Okazaki T., Byrne M.C., Horton H.F.,Fouser L., Carter L., Ling V., Bowman M.R., Carreno B.M., CollinsM., Wood C.R. and Honjo T. (2000). Engagement of the PD-1immunoinhibitory receptor by a novel B7 family member leads tonegative regulation of lymphocyte activation. J. Exp. Med. 192,1027-1034.

Fujita Y., Yagishita S., Hagiwara K., Yoshioka Y., Kosaka N., TakeshitaF., Fujiwara T., Tsuta K., Nokihara H., Tamura T., Asamura H.,Kawaishi M., Kuwano K. and Ochiya T. (2015). The clinicalrelevance of the miR-197/CKS1B/STAT3-mediated PD-L1 networkin chemoresistant non-small-cell lung cancer. Mol. Ther. 23, 717-727.

Gajewski T.F., Meng Y., Blank C., Brown I., Kacha A., Kline J. andHarlin H. (2006). Immune resistance orchestrated by the tumormicroenvironment. Immunol. Rev. 213, 131-145.

Gatalica Z., Snyder C., Maney T., Ghazalpour A., Holterman D.A., XiaoN.Q., Overberg P., Rose I., Basu G.D., Vranic S., Lynch H.T., VonHoff D.D. and Hamid O. (2014). Programmed cell death 1 (PD-1)and its ligand (PD-L1) in common cancers and their correlation with

87Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

molecular cancer type. Cancer. Epidemiol. Biomarkers Prev. 23,2965-2970.

Ghebeh H., Mohammed S., Al-Omair A., Qattan A., Lehe C., Al-QudaihiG., Elkum N., Alshabanah M., Bin Amer S., Tulbah A., Ajarim D., Al-Tweigeri T. and Dermime S. (2006). The B7-H1 (PD-L1) Tlymphocyte-inhibitory molecule is expressed in breast cancerpatients with infiltrating ductal carcinoma: correlation with importanthigh-risk prognostic factors. Neoplasia 8, 190-198.

Gonzalez-Cao M., Karachaliou N., Viteri S., Morales-Espinosa D.,Teixido C., Sanchez Ruiz J., Molina-Vila M.A., Santarpia M. andRosell R. (2015). Targeting PD-1/PD-L1 in lung cancer: currentperspectives. Lung Cancer (Auckl). 6, 55-70.

Greenfield E.A. (2018). Polyethylene glycol fusion for hybridomaproduction. Cold. Spring. Harb. Protoc. 2018, pdb prot103176.

Hartley G.P., Chow L., Ammons D.T., Wheat W.H. and Dow S.W.(2018). Programmed cell death ligand 1 (PD-L1) signaling regulatesmacrophage proliferation and activation. Cancer Immunol. Res. 6,1260-1273.

Hassounah N.B., Malladi V.S., Huang Y., Freeman S.S., BeauchampE.M., Koyama S., Souders N., Martin S., Dranoff G., Wong K.K.,Pedamallu C.S., Hammerman P.S. and Akbay E.A. (2019).Identification and characterization of an alternative cancer-derivedPD-L1 splice variant. Cancer Immunol. Immunother. 68, 407-420.

He X.H., Xu L.H. and Liu Y. (2005). Identification of a novel splicevariant of human PD-L1 mRNA encoding an isoform-lacking Igv-likedomain. Acta. Pharmacol. Sin. 26, 462-468.

Ilie M., Hofman V., Dietel M., Soria J.C. and Hofman P. (2016).Assessment of the PD-L1 status by immunohistochemistry:challenges and perspectives for therapeutic strategies in lungcancer patients. Virchows Arch. 468, 511-525.

Jackson L.R., Trudel L.J., Fox J.G. and Lipman N.S. (1999). Monoclonalantibody production in murine ascites. II. Production characteristics.Lab. Anim. Sci. 49, 81-86.

Jespersen M.C., Peters B., Nielsen M. and Marcatili P. (2017).Bepipred-2.0: Improving sequence-based B-cell epitope predictionusing conformational epitopes. Nucleic Acids Res. 45, W24-W29.

Kahlmeyer A., Stohr C.G., Hartmann A., Goebell P.J., Wullich B., WachS., Taubert H. and Erlmeier F. (2019). Expression of PD-1 andCTLA-4 are negative prognostic markers in renal cell carcinoma. J.Clin. Med. 8, 743.

Keir M.E., Freeman G.J. and Sharpe A.H. (2007). PD-1 regulates self-reactive CD8+ T cell responses to antigen in lymph nodes andtissues. J. Immunol. 179, 5064-5070.

Lee Y., Shin J.H., Longmire M., Wang H., Kohrt H.E., Chang H.Y. andSunwoo J.B. (2016). CD44+ cells in head and neck squamous cellcarcinoma suppress T-cell-mediated immunity by selectiveconstitutive and inducible expression of PD-L1. Clin. Cancer. Res.22, 3571-3581.

Leung E.L., Fiscus R.R., Tung J.W., Tin V.P., Cheng L.C., Sihoe A.D.,Fink L.M., Ma Y. and Wong M.P. (2010). Non-small cell lung cancercells expressing CD44 are enriched for stem cell-like properties.PLoS One 5, e14062.

Lin D.Y., Tanaka Y., Iwasaki M., Gittis A.G., Su H.P., Mikami B.,Okazaki T., Honjo T., Minato N. and Garboczi D.N. (2008). The PD-1/PD-L1 complex resembles the antigen-binding Fv domains ofantibodies and T cell receptors. Proc. Natl. Acad. Sci. USA 105,3011-3016.

Mahoney K.M., Shukla S.A., Patsoukis N., Chaudhri A., Browne E.P.,Arazi A., Eisenhaure T.M., Pendergraft W.F., Hua P., Pham H.C., Bu

X., Zhu B.G., Hacohen N., Fritsch E.F., Boussiotis V.A., Wu C.J. andFreeman G.J. (2019). A secreted PD-L1 splice variant thatcovalently dimerizes and mediates immunosuppression. Cancer.Immunol. Immunother. 68, 421-432.

Mavridis K., Gueugnon F., Petit-Courty A., Courty Y., Barascu A.,Guyetant S. and Scorilas A. (2015). The oncomir miR-197 is a novelprognostic indicator for non-small cell lung cancer patients. Br. J.Cancer 112, 1527-1535.

Mittendorf E.A., Philips A.V., Meric-Bernstam F., Qiao N., Wu Y.,Harrington S., Su X., Wang Y., Gonzalez-Angulo A.M., AkcakanatA., Chawla A., Curran M., Hwu P., Sharma P., Litton J.K., MolldremJ.J. and Alatrash G. (2014). PD-L1 expression in triple-negativebreast cancer. Cancer Immunol. Res. 2, 361-370.

Nguyen V.N., Mirejovsky T., Melinova L. and Mandys V. (2000). CD44and its v6 spliced variant in lung carcinomas: relation to NCAM,CEA, EMA and UP1 and prognostic significance. Neoplasma 47,400-408.

Nielsen C., Ohm-Laursen L., Barington T., Husby S. and Lillevang S.T.(2005). Alternative splice variants of the human PD-1 gene. Cell.Immunol. 235, 109-116.

Nishimura H., Nose M., Hiai H., Minato N. and Honjo T. (1999).Development of lupus-like autoimmune diseases by disruption of thePD-1 gene encoding an ITIM motif-carrying immunoreceptor.Immunity 11, 141-151.

Nishimura H., Okazaki T., Tanaka Y., Nakatani K., Hara M., MatsumoriA., Sasayama S., Mizoguchi A., Hiai H., Minato N. and Honjo T.(2001). Autoimmune dilated cardiomyopathy in PD-1 receptor-deficient mice. Science 291, 319-322.

Nishino M., Ozaki M., Hegab A.E., Hamamoto J., Kagawa S., Arai D.,Yasuda H., Naoki K., Soejima K., Saya H. and Betsuyaku T. (2017).Variant CD44 expression is enriching for a cell population withcancer stem cell-like characteristics in human lung adenocarcinoma.J. Cancer 8, 1774-1785.

Noguchi M., Morikawa A., Kawasaki M., Matsuno Y., Yamada T.,Hirohashi S., Kondo H. and Shimosato Y. (1995). Smalladenocarcinoma of the lung. Histologic characteristics andprognosis. Cancer 75, 2844-2852.

O'Malley G., Treacy O., Lynch K., Naicker S.D., Leonard N.A., Lohan P.,Dunne P.D., Ritter T., Egan L.J. and Ryan A.E. (2018). Stromal cellPD-L1 inhibits CD8+ T-cell antitumor immune responses andpromotes colon cancer. Cancer Immunol. Res. 6, 1426-1441.

Pardoll D.M. (2012). The blockade of immune checkpoints in cancerimmunotherapy. Nat. Rev. Cancer 12, 252-264.

Parra E.R., Villalobos P., Mino B. and Rodriguez-Canales J. (2018).Comparison of different antibody clones for immunohistochemistrydetection of programmed cell death ligand 1 (PD-L1) on non-smallcell lung carcinoma. Appl. Immunohistochem. Mol. Morphol. 26, 83-93.

Polonia A., Pinto R., Cameselle-Teijeiro J.F., Schmitt F.C. and ParedesJ. (2017). Prognostic value of stromal tumour infiltrating lymphocytesand programmed cell death-ligand 1 expression in breast cancer. J.Clin. Pathol. 70, 860-867.

Pyo J.S., Kang G. and Kim J.Y. (2017). Prognostic role of PD-L1 inmalignant solid tumors: A meta-analysis. Int. J. Biol. Markers 32,e68-e74.

Qiu X.Y., Hu D.X., Chen W.Q., Chen R.Q., Qian S.R., Li C.Y., Li Y.J.,Xiong X.X., Liu D., Pan F., Yu S.B. and Chen X.Q. (2018). PD-L1confers glioblastoma multiforme malignancy via ras binding andRas/Erk/EMT activation. Biochim. Biophy. Acta Mol. Basis. Dis.

88Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype

1864, 1754-1769.Saito H., Shimizu S., Kono Y., Murakami Y., Shishido Y., Miyatani K.,

Matsunaga T., Fukumoto Y., Ashida K. and Fujiwara Y. (2019). PD-1 expression on circulating CD8+ T-cells as a prognostic marker forpatients with gastric cancer. Anticancer Res. 39, 443-448.

Schats K., Van Vre E.A., Schrijvers D., De Meester I. and Kockx M.(2017). Epitope mapping of PD-L1 primary antibodies (28-8, SP142,SP263 E1l3N). J. Clin. Oncol. (Suppl. 15) 35, 3028.

Shen X. and Zhao B. (2018). Efficacy of PD-1 or PD-L1 inhibitors andPD-L1 expression status in cancer: Meta-analysis. Br. Med. J. 362,k3529.

Shigemori T., Toiyama Y., Okugawa Y., Yamamoto A., Yin C., NarumiA., Ichikawa T., Ide S., Shimura T., Fujikawa H., Yasuda H., Hiro J.,Yoshiyama S., Ohi M., Araki T. and Kusunoki M. (2019). SolublePD-L1 expression in circulation as a predictive marker for recurrenceand prognosis in gastric cancer: direct comparison of the clinicalburden between tissue and serum PD-L1 expression. Ann. Surg.Oncol. 26, 876-883.

Shukuya T. and Carbone D.P. (2016). Predictive markers for theefficacy of anti-PD-1/PD-L1 antibodies in lung cancer. J. Thorac.Oncol. 11, 976-988.

Stovgaard E.S., Dyhl-Polk A., Roslind A., Balslev E. and Nielsen D.(2019). PD-L1 expression in breast cancer: expression in subtypesand prognostic significance: a systematic review. Breast. CancerRes. Treat. 174, 571-584.

Sun W.Y., Lee Y.K. and Koo J.S. (2016). Expression of PD-L1 in triple-negative breast cancer based on different immunohistochemicalantibodies. J. Transl Med. 14, 173.

Tamai K., Nakamura M., Mizuma M., Mochizuki M., Yokoyama M., EndoH., Yamaguchi K., Nakagawa T., Shiina M., Unno M., Muramoto K.,Sato I., Satoh K., Sugamura K. and Tanaka N. (2014). Suppressiveexpression of CD274 increases tumorigenesis and cancer stem cellphenotypes in cholangiocarcinoma. Cancer Sci. 105, 667-674.

Thompson R.H., Kuntz S.M., Leibovich B.C., Dong H., Lohse C.M.,Webster W.S., Sengupta S., Frank I., Parker A.S., Zincke H., BluteM.L., Sebo T.J., Cheville J.C. and Kwon E.D. (2006). Tumor B7-H1is associated with poor prognosis in renal cell carcinoma patientswith long-term follow-up. Cancer Res. 66, 3381-3385.

Urban J.L. and Schreiber H. (1992). Tumor antigens. Annu. Rev.Immunol. 10, 617-644.

van der Bruggen P. and Van den Eynde B. (1992). Molecular definitionof tumor antigens recognized by T lymphocytes. Curr. Opin.Immunol. 4, 608-612.

Wang A., Wang H.Y., Liu Y., Zhao M.C., Zhang H.J., Lu Z.Y., FangY.C., Chen X.F. and Liu G.T. (2015). The prognostic value of PD-L1expression for non-small cell lung cancer patients: a meta-analysis.Eur. J. Surg. 41, 450-456.

Accepted October 28, 2020

89Novel PD-L1 mAb reveals regulation of PD-L1 in BAC subtype