Distinct antibody clones detect PD-1 checkpoint expression and block PD-L1 interactions on live murine melanoma cells

Monoclonal antibodies (abs) targeting the programmed cell death 1 (PD-1) immune checkpoint pathway have revolutionized tumor therapy. Because T-cell-directed PD-1 blockade boosts tumor immunity, anti-PD-1 abs have been developed for examining T-cell-PD-1 functions. More recently, PD-1 expression has also been reported directly on cancer cells of various etiology, including in melanoma. Nevertheless, there is a paucity of studies validating anti-PD-1 ab clone utility in specific assay types for characterizing tumor cell-intrinsic PD-1. Here, we demonstrate reactivity of several anti-murine PD-1 ab clones and recombinant PD-L1 with live B16-F10 melanoma cells and YUMM lines using multiple independent methodologies, positive and negative PD-1-specific controls, including PD-1-overexpressing and PD-1 knockout cells. Flow cytometric analyses with two separate anti-PD-1 ab clones, 29F.1A12 and RMP1-30, revealed PD-1 surface protein expression on live murine melanoma cells, which was corroborated by marked enrichment in PD-1 gene (Pdcd1) expression. Immunoblotting, immunoprecipitation, and mass spectrometric sequencing confirmed PD-1 protein expression by B16-F10 cells. Recombinant PD-L1 also recognized melanoma cell-expressed PD-1, the blockade of which by 29F.1A12 fully abrogated PD-1:PD-L1 binding. Together, our data provides multiple lines of evidence establishing PD-1 expression by live murine melanoma cells and validates ab clones and assay systems for tumor cell-directed PD-1 pathway investigations.


B16-F10 melanoma cells express PD-1.
We first validated our previous findings of Pdcd1 expression in murine B16-F10 melanoma cells 5 by real-time qPCR, using two independent Pdcd1 primer sets and positive and negative control cells of varying PD-1 expression level. As already shown previously 5,15,16 , B16-F10 wildtype (WT) cells expressed marked levels of Pdcd1 (qPCR cycle threshold ≤ 25 for both primer sets) that did not substantially differ from those in positive control unactivated syngeneic (C57BL/6) T-cells (Fig. 1a). As expected, Pdcd1 expression was > 100-fold increased in PD-1 OE and > 4-fold in CD3/CD28-activated T-cells compared to WT B16-F10 melanoma cells, but not detected in negative control, PD-1 KO B16-F10 or activated T-cells (Fig. 1a), thus confirming specificity of both primer sets for Pdcd1. Immunoblotting corroborated PD-1 protein expression by B16-F10 WT and PD-1 OE, but not PD-1 KO melanoma cells, and by unactivated and activated WT, but not PD-1 KO T-cells at an expected molecular weight of ~ 37-50 kDa (Fig. 1b), consistent with previous studies 5,15 . Anti-PD-1, but not isotype control ab IP also revealed a predominant band at ~ 50 kDa, and an additional band at ~ 37 kDa, in WT and PD-1 OE B16-F10 melanoma cells, and in activated T-cells (Fig. 1c). PD-1 protein identity was verified in IP eluates by MS sequencing. Together, these results rigorously confirm PD-1 transcript and protein expression by B16-F10 melanoma cells.
Culture conditions can greatly affect PD-1 ab clone reactivity. For example, we found that 29F.1A12 and RMP1-30 binding to B16-F10 cells was increased in 3D versus 2D settings. This result is consistent with our prior demonstration of B16-F10 growth inhibition in 3D tumor spheroid, but not standard 2D, cultures 5 , and with work by others revealing PD-1 induction in B16-F10 cells exposed to hypoxia 15 , a condition prevalent within tumor spheroids 47 . Defining additional factors regulating cell type-specific PD-1 expression level, PTM, and ab recognition will require a future dedicated effort. Nevertheless, our study contributes to the growing body of evidence that tumor cell-PD-1 detection can substantially increase depending on environmental cues 13,15,17,18,29 .
Although multiple groups have identified PD-1 on diverse cancer cell types, including B16-F10 5,7,12,15,16 , one study surprisingly claims that B16-F10 cells are negative for PD-1, and that PD-1 ab clones, such as 29F.1A12, exclusively bind to an off-target nuclear antigen exposed by dead cells 34 . Unfortunately, this work has shortcomings in experimental design and data interpretation, incompletely describes methodological details, and fails to reference already published studies demonstrating PD-1 expression by live B16-F10 cells 7,12 . First, Metzger et al. 34 employed PD-1 abs at an inadequately low concentration of 1 µg/mL (1:200 dilution of the 200 µg/mL 29F.1A12 ab stock) for FACS-based PD-1 detection on B16-F10 cells. This ab amount is > 10-fold lower than previously reported concentrations (10-20 µg/mL) used for staining tumor cell-PD-1 5 and also than clinical PD-1 ab serum titers (> 10-100 µg/mL) achieved in patients at time of administration [48][49][50] . Absent throughout the manuscript were matched isotype control ab FACS plots. We used PD-1 abs at 10-20 µg/mL, proper isotype-controlled gating strategies, and the identical viability dye 34 , and found significant binding of 29F.1A12 and RMP1-30 to overlapping live (FVD − ) B16-F10 and YUMM subpopulations, in contrast to the study in question. We also observed PD-1 ab clone reactivity with dead (FVD + ) melanoma cells, as reported 34 . Nonetheless, such ab binding cannot be exclusively ascribed to off-target recognition of a nuclear antigen in dead cells 34 because both 29F.1A12 and RMP1-30 reacted more avidly with FVD + WT versus PD-1 KO B16-F10 melanoma or T-cells, which differ in PD-1 but not nuclear antigen content. Our findings thus indicate that both ab clones detect PD-1, including in dead cells. On target PD-1 ab binding was further supported by the fact that non-neutralizing 37 RMP1-30 reactivity coincided with, while 29F.1A12 fully blocked, rPD-L1 ligation to B16-F10 cells.
Second, Metzger et al. 34 do not describe Pdcd1 primer sequences and qRT-PCR amplification conditions for detecting PD-1 gene expression in B16-F10 cells. Moreover, the authors omit crucial PD-1 KO negative controls in gene expression analyses that were otherwise prevalently employed elsewhere. The apparent absence of an electrophoresis band for B16-F10 cells 34 does not exclude Pdcd1 amplification by qRT-PCR. Indeed, a short 77 bp fragment 34 is significantly harder to detect than products of greater length, due to decreased dye incorporation and corresponding emission. This is particularly true when product levels are low and agarose gels are used at an atypically high concentration (3%) 34 resulting in increased opacity. Our qRT-PCR analyses rigorously confirmed PD-1 gene expression by WT B16-F10 cells using two independent Pdcd1 primer sets with specified sequences and amplification settings, and multiple control cells of defined PD-1 expression level, including PD-1 KO, OE, WT, and PD-1 + versus PD-1 − FACS (29F.1A12, RMP1-30, rPD-L1)-purified B16-F10 and T-cell cohorts. It should be noted that nucleotide or amino acid sequencing methodologies represent gold standards for validating expression of a gene or protein of interest, respectively. Indeed, we previously amplified and sequenced the full Pdcd1 coding region from B16-F10 cells (GenBank accession KJ865858) 5 . In the current study, we confirmed expression of PD-1 protein in B16-F10 lysates by IP and MS-based sequencing. Together, our work and that of others 7,12,15,16 therefore unequivocally establishes PD-1 expression by B16-F10 cells.
The scientific method encourages spirited debate based on empirical evidence. Replication of results is a pivotal hallmark of knowledge advancement. Reproducibility of research data requires well-defined experimental conditions, careful controls, and an open dialogue between groups to reconcile apparent discrepancies in results. In these respects, the manuscript by Metzger and colleagues 34 rather fell short, as elaborated above. The inability to detect PD-1, or any molecule for that matter, does not definitively exclude its expression, particularly when using conditions distinct from other reports. Moreover, findings restricted to a single cell line, B16-F10, cannot be extrapolated to rule out PD-1 expression by all other tumor cells and even non-T immune cell lineages, as www.nature.com/scientificreports/ overstated by the authors 34 . Such unsubstantiated conclusions can hinder the advancement of science, especially when they dismiss the contributions of countless groups. We appreciate the challenges inherent to studying tumor cell-intrinsic PD-1, because its expression is often restricted to subsets of cancer cells 5,7,12,14,19,20,23 and may vary by ab clone, assay type, and culture condition 13,15,17,18,29 . Nevertheless, there is growing appreciation of tumor cell-intrinsic PD-1 roles in tumorigenesis and response to immune checkpoint therapy across multiple malignancies 5,7,15,17,[19][20][21][23][24][25]27,28,32 . Accordingly, validating reagents, defining assay systems and the experimental conditions that enable mechanistic dissection of cancer cell-PD-1 checkpoint immunobiology are important endeavors. This study represents a crucial step forward in this regard.
Mice. C57BL/6 mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and PD-1 −/− KO C57BL/6 mice 51 maintained and housed at the Brigham and Women's Hospital (BWH) animal facility, as described 5,52 . All mice were female, at least 6 weeks of age, and used in accordance with the National Institutes of Animal Healthcare Guidelines under protocol 2016N000112 approved by the Institutional Animal Care and Use Committee of BWH. The study is reported in accordance with ARRIVE guidelines.

RNA extraction and real-time quantitative RT-PCR analysis.
Total RNA was isolated using the RNeasy Micro Kit or the RNeasy Plus Mini Kit (Qiagen, Germantown, MD), according to the manufacturer's protocol. RNA was subsequently converted to cDNA using the SuperScript VILO cDNA synthesis kit (Thermo Fisher), and samples assayed in triplicate using the Fast SYBR Green Master Mix (ThermoFisher) with primer sets, as below, on a QuantStudio 5 Real-Time PCR system (Applied Biosystems, Waltham, MA). Thermal cycling was carried out at 94 °C for 2 min, followed by 40 cycles at 94 °C for 15 s, 60 °C for 20 s and 68 °C for 1 min, as described 5 , followed by melt curve validation of amplicons. Data was normalized to murine actin, and relative transcript levels calculated using the delta-delta Ct method. Samples with threshold cycle (Ct) numbers above the water negative control were designated as not detected. The primer sequences used for murine Pdcd1 detection were: primer set 1, forward-5′-CGG TTT CAA GGC ATG GTC ATTGG-3′, reverse-5′-TCA GAG TGT CGT CCT TGC T TCC-3′; primer set 2, forward-5′-GGA GCA GAG CTC GTG GTA AC-3′, reverse-5′-AAT GAC CAT GCC TTG AAA CC-3′. For murine actin, primer sequences were: forward-5′-CAT CGT ACT CCT GCT TGC TG-3′ and reverse-5′-AGC GCA AGT ACT CTG TGT GG -3′.
Generation of PD-1 KO B16-F10 melanoma cells. B16-F10 PD-1 KO cells were created using CRISPR/ Cas-9 by inserting the guide (g) RNA, GAG CAG AGC TCG TGG TAA C (ThermoFisher), targeting murine Pdcd1 into the vector pKLV2-U6gRNA5(BbsI)-PGKpuro2ABFP-W (67,974, Addgene, Watertown, MA). We first generated a monogenetic founder B16-F10 tumor cell clone by single cell sorting. The B16-F10 founder cells were co-transfected with the above construct and with Cas9-EGFP (LentiCas9-EGFP, plasmid #63,592, Addgene). Cells were then doubly selected 1-2 days post-transfection in media containing puromycin (5 μg/ml, Life Technologies) and blasticidin (20 μg/ml, ThermoFisher) over 3-5 days, and sorted for BFP and EGFP expression. The B16-F10 transfectant pool was expanded for 3-5 days and then subcultured into 96-well plates by limiting dilution to generate single cell clones. Colonies were determined by visual inspection, expanded and screened for loss of PD-1 gene and protein expression by real-time quantitative RT-PCR, flow cytometry, and immunoblotting, as described herein. To confirm chromosomal knockout of PD-1, genomic DNA was isolated from candidate clones (DNeasy Blood & Tissue Kit, Qiagen), and the Pdcd1 locus was PCR-amplified (Platinum Taq DNA Polymerase, ThermoFisher) with primers bracketing the Pdcd1 gRNA binding site. Thermal cycling was carried out in the presence of 1. 5
Immunoblotting. Cells were lysed in ice-cold RIPA buffer supplemented with protease inhibitor cocktail (Roche, Basel, Switzerland) and vortexed at 4 °C for 30 min, as described 5 . Lysates were centrifuged and protein concentrations measured using the BCA protein assay kit (Thermo Fisher), following the manufacturer's instructions. B16-F10 lysates were boiled in reducing Laemmli sample buffer (Bio-Rad, Hercules, CA) for 7 min, resolved in 7.5% SDS-PAGE gels (Bio-Rad) and proteins transferred to Sequi-Blot PVDF membranes (Bio-Rad). Membranes were blocked in tris-buffered saline (TBS, Boston BioProducts, Milford, MA) containing 5% (w/v) bovine serum albumin (BSA, Sigma) and 0.1% (v/v) Tween-20 (Sigma), for at least 1 h at room temperature (RT), and then incubated overnight at 4 °C with primary ab (10-20 μg/mL), followed by washing and incubation with secondary HRP-conjugated ab (1:1000) for 1 h at RT. Antigens were visualized using the Lumi-Light Western blotting substrate (Roche) on HyBlot CL Autoradiography Films (Thomas Scientific, Swedesboro, NJ) using a Kodak Min-R mammography processor (Kodak, Rochester, NY). For detection of actin, blots were stripped with Restore Western blot Stripping Buffer (Thermo Fisher) according to the manufacturer's protocol, blocked, and incubated for 1-12 h at 4 °C with primary ab (1:1000-2500), and then with secondary HRP-goat anti-mouse ab (1:2500) for 1 h at RT.
Immunoprecipitation. PD-1 IP studies were performed using Pierce Protein A/G Plus agarose beads (Invitrogen), according to the manufacturer's instructions. Briefly, cells were lysed in ice-cold buffer (150 mM NaCl, 50 mM Tris-HCI and protease inhibitor cocktail, Roche), sonicated (three 10 s bursts), vortexed for 2 h at 4 °C in 2% Nonidet P-40 (NP-40, Sigma), and centrifuged. B16-F10 lysates were concentrated using Microcon-10 Ultracel PL-10 filter columns, as above. Cell lysates were precleared for 2 h at 4 °C by incubation with Protein A/G Plus agarose beads previously blocked in ice-cold buffer supplemented with 1% BSA for 1 h at 4 °C, incubated with anti-mouse PD-1 (4-6 µg, RMP1-14) 53 or rat IgG2a isotype control abs for 2 h at 4 °C, and then with Protein A/G Plus agarose beads overnight at 4 °C under continuous rotation. Supernatants were kept for assessment of IP efficiency, Protein A/G Plus agarose beads washed extensively, and IP products eluted in ice-cold buffer, as above, supplemented with 1.5 × Non-reducing Lane Marker Sample Buffer (Thermo Fisher), and boiled at 100 °C for 7min 54 . IP products were then analyzed by immunoblotting, as above, or resolved in 7.5% SDS-PAGE gels, stained with Colloidal Coomassie Blue (Bio-Rad), and subjected to MS sequencing, as described below.
Mass spectrometry. SDS/PAGE gels (7.5%) containing IP products were run at ~ 120 V. Gels were fixed in 40% (v/v) ethanol, 10% (v/v) acetic acid, 50% high-grade water for 15 min, and then stained with Colloidal Coomassie Blue overnight at RT, following the manufacturer's instructions. Gel slices containing counterstained www.nature.com/scientificreports/ protein bands that coincided with the expected PD-1 molecular weight of ~ 32-55 kDa were excised, washed twice in 50% (v/v) acetonitrile, 50% (v/v) high-grade water. Gel slices were then submitted for peptide mass fingerprinting on a Q Exactive™ HF-X Hybrid Quadrupole-Orbitrap™ MS system (Thermo Fisher), as described 55 . Statistics. Experimental groups were compared statistically using the PRISM 9.0 software (GraphPad, San Diego, CA). The Student's t test was used to compare two experimental groups, and one-way ANOVA with Dunnett post-test for comparison of three experimental groups, with p < 0.05 considered statistically significant. Data was tested for normal distribution using the D' Agostino and Pearson omnibus normality test. See also the Supplementary Information.

Data availability
The data generated in this study is available from the corresponding authors upon reasonable request.