Full Terms & Conditions of access and use can be found athttps://www.tandfonline.com/action/journalInformation?journalCode=idrt20

Journal of Drug Targeting

ISSN: 1061-186X (Print) 1029-2330 (Online) Journal homepage: https://www.tandfonline.com/loi/idrt20

A small-sized protein binder specific for humanPD-1 effectively suppresses the tumour growth intumour mouse model

Sumin Son, Jinho Park, Hyodeok Seo, Hyun Tae Lee, Yong-Seok Heo & Hak-Sung Kim

To cite this article: Sumin Son, Jinho Park, Hyodeok Seo, Hyun Tae Lee, Yong-Seok Heo& Hak-Sung Kim (2019): A small-sized protein binder specific for human PD-1 effectivelysuppresses the tumour growth in tumour mouse model, Journal of Drug Targeting, DOI:10.1080/1061186X.2019.1669042

To link to this article: https://doi.org/10.1080/1061186X.2019.1669042

View supplementary material

Accepted author version posted online: 16Sep 2019.Published online: 26 Sep 2019.

Submit your article to this journal

Article views: 103

View related articles

View Crossmark data

ORIGINAL ARTICLE

A small-sized protein binder specific for human PD-1 effectively suppresses thetumour growth in tumour mouse model

Sumin Sona, Jinho Parka, Hyodeok Seoa�, Hyun Tae Leeb, Yong-Seok Heob and Hak-Sung Kima

aDepartment of Biological Sciences, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea; bDepartment of Chemistry,Konkuk University, Seoul, Korea

ABSTRACTImmune checkpoint inhibitors have drawn a consider attention as an effective cancer immunotherapy,and several monoclonal antibodies targeting the immune checkpoint receptors, such as human pro-grammed cell death-1 (hPD-1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), are clinicallyused for treatment of various cancers. Here we present the development of a small-sized protein binderwhich specifically binds to hPD-1. The protein binder, which is composed of leucine-rich repeat (LRR)modules, was selected against hPD-1 through phage display, and its binding affinity was maturated up to17nM by modular evolution approach. The protein binder was shown to be highly specific for hPD-1,effectively inhibiting the interaction between hPD-1 and its ligand, hPD-L1. The protein binder restoredT-cell function in vitro, and exhibited a strong anti-tumour activity in tumour mouse model, indicatingthat it acts as an effective checkpoint blockade. Based on the results, the developed protein binder spe-cific for hPD-1 is likely to find a potential use in cancer immunotherapy.

ARTICLE HISTORYReceived 18 July 2019Revised 6 September 2019Accepted 14 September 2019

KEYWORDSCancer immunotherapy;PD-1; immune checkpointinhibitor; non-antibodyscaffold; repebody

Introduction

Immunotherapy has drawn much attention as an effective way oftreating various cancers by boosting patients’ own immune sys-tems. Several types of immunotherapy have been used to treatcancers, and such treatments can help the body’s immune systemfight cancer cells directly or stimulate the immune system in amore general way. Immune checkpoints, such as programmed celldeath-1 (PD-1) and cytotoxic T-lymphocyte-associated protein 4(CTLA-4), have been considered promising targets for immuno-therapy. PD-1 is an inhibitory receptor which is normallyexpressed on activated T-cells [1], and it has two ligands, namelyprogrammed death 1-ligand 1 (PD-L1) (also known as B7-H1) andPD-L2 (also known as B7-DC). It has been known that binding ofPD-1 to PD-L1 hampers T-cell proliferation and cytokine produc-tion through downregulation of ZAP70-mediated T-cell receptor(TCR) signalling pathways [2,3]. Importantly, cancer cells alsoexpress PD-L1 in tumour microenvironment to make T-cells“exhausted” as one of the mechanisms to avoid immune-surveil-lance in the body [4,5].

As an approach to stimulating the body’s immune system forcancer treatment by inhibition of the PD-1 and PD-L1 interaction,monoclonal antibodies targeting PD-1 have been developed, andnivolumab and pembrolizumab are clinically used [6–8]. Anti-PD-1antibodies are known to have significant effects on patients withvarious cancers, such as metastatic melanoma, renal cell carcin-oma (RCC) and non-small cell lung cancer (NSCLC) [9–11].Recently, more attempts have been made to broaden their use inother cancer types such as glioblastoma and pancreatic cancer

through combination therapy [12,13]. Despite widespread useof monoclonal antibodies as therapeutics, they have several draw-backs. Their large size (�150 kDa) and disulphide bonds areknown to cause aggregation and low penetration into tumoursite, especially in solid tumours, resulting in low efficacy [14–17].Their long half-life mediated by FcRn recycling might trigger unex-pected immune-related toxicities [18–21]. In an effort to overcomesuch shortcomings, peptides and peptidomimetics targetingimmune checkpoints were developed [22–26]. Peptide-basedinhibitors were shown to have an antitumor activity, but theyhave some limitations, including low binding affinity and suscepti-bility to hydrolysis. Variants of PD-1 with high binding affinity forits ligand were also attempted [27,28], but they were expressed asinclusion body in E.coli, requiring a refolding process for purifica-tion mainly due to their conserved immunoglobulin (Ig)-likedomains. Thus, development of new immune checkpoint inhibi-tors is of great significance.

As alternatives to immunoglobulin antibodies, small-sized non-antibody protein scaffolds have attracted considerable attention[29,30]. We previously developed a small-sized (�30 kDa) non-antibody scaffold, termed ‘repebody’, which is composed ofleucine-rich repeat (LRR) modules [31]. The potential of the repe-body scaffold was demonstrated by showing its use as antitumoragents, targeted drug delivery vehicles, and diagnosis tools of dis-ease biomarkers [32–35]. Here, we present the development of arepebody which is highly specific for human PD-1 (hPD-1) throughphage display and modular evolution approach. The utility of thehPD-1-specific repebody as an immune checkpoint blockade wasinvestigated in vitro and in vivo. Details are reported herein.

CONTACT Hak-Sung Kim e_hskim@kaist.ac.kr Department of Biological Sciences, Korea Advanced Institute of Science and Technology, Daejeon, 34141 Korea�Present address: Research Group of Natural Materials and Metabolism, Korea Food Research Institute (KFRI), Jeollabuk-Do, KoreaSupplemental data for this article can be accessed here.

� 2019 Informa UK Limited, trading as Taylor & Francis Group

JOURNAL OF DRUG TARGETINGhttps://doi.org/10.1080/1061186X.2019.1669042

Materials and methods

Development of a repebody specific for hPD-1

A repebody library was constructed by mutagenic PCR using NNKdegenerate codon as described elsewhere [31]. For selection ofinitial binders, three amino acids on LRRV2 (I91, T93, and G94)and three amino acids on LRRV3 (V115, V117, and E118) wererandomised. Another four amino acids on LRRV4 (Y137, N139,A141, and H142) were randomly mutated to construct a secondlibrary for affinity maturation. The constructed library was insertedinto a phagemid pBEL118M vector, and transformed into E. coliTG-1 strain by electroporation for a phage display. Repebodiesspecific for human PD-1 were selected by phage display and sol-uble biopanning against extracellular domain (ECD) of hPD-1using DynabeadsVR M-280 Streptavidin magnetic beads (Invitrogen,Carlsbad, CA). ECD of biotinylated hPD-1 protein (Acro biosystems,Newark, DE) was immobilised onto Dynabeads and blocked with ablocking buffer (1X PBS, 0.05% Tween-20, and 1% BSA). Toexclude phages which non-specifically bind to the surface of mag-netic beads, 1ml of phage solution containing 1.0� 1012 cfu/mLwas added to beads without hPD-1. After 1 h incubation at roomtemperature, phages were moved to hPD-1-immobilised beads forpositive selection. Following incubation for 1 h at room tempera-ture, unbound phages were washed out, and bound phages wereeluted using 0.2M Glycine-HCl (pH 2.2) and neutralised with 1MTris-HCl (pH 9.0). Eluted phages were infected to TG-1 (OD600 ¼0.5) for generation of phage solution for the next round of bio-panning. This process was repeated five times for selection of ini-tial binders and three times for affinity maturation, respectively.After enrichment of phages, repebodies specifically binding tohPD-1 were identified through phage ELISA.

Protein expression, purification and removal of LPS

Six histidine residues were fused to C-terminus of a repebody,cloned into pET21a vector, and transformed to E. coli strainOrigami B. The cells were cultured in LB medium followed by add-ition of 0.5mM IPTG when OD600 reached 0.5. A repebody waspurified by affinity chromatography using Ni-NTA resin (Qiagen,Germany). Lipopolysaccharide (LPS) was removed using TritonX-114 (Sigma-Aldrich, St. Louis, MO), as described elsewhere[36,37]. Briefly, protein solution was treated with 1–2% TritonX-114 followed by incubation for 30min at 4 �C with gentle shak-ing, and the mixture was further incubated at 37 �C for 10min.The aqueous phase was separated by centrifugation at13,000 rpm. After repeating this step four times, residual TritonX-114 in protein solution was removed by Bio-Beads SM-2 (Bio-Rad, Hercules, CA).

Inhibition assay of the PD-1 and PD-L1 interaction

Inhibition of the interaction between PD-1 and PD-L1 by a repe-body was assayed using competitive ELISA. Each well of a 96-wellplate was coated with 1 mg/mL of extracellular domain (ECD) ofhPD-1 (Sino Biological, China) by overnight incubation at 4 �C.After blocking with 250 mL of a blocking buffer (1X PBS, 0.05%Tween 20, 1% BSA), 100mL of a working solution (0.6mg/mL ofECD of hPD-L1 (Mouse Fc-tagged, Sino Biological) with differentrepebody concentrations) was added. Following incubation for 1 hat room temperature, each well was washed five times with PBSTbuffer (1X PBS, 0.05% Tween 20). HRP-conjugated goat-anti-mIgGantibodies (Bio-Rad) were added and incubated for 1 h at roomtemperature to detect remaining hPD-L1, followed by addition of

100 mL of TMB to each well. The reaction was stopped with 100 mLof 1 N H2SO4, and absorbance at 450 nm was measured.Competitive ELISA between PD-L1 and a repebody or nivolumabwas carried out using the same procedure as described above.1 mg/mL of hPD-1 was coated on 96-well plate. 100mL of workingsolution containing 0.6mg/mL of ECD of hPD-L1 (Biotinylated(Acrobiosystems)) and 200 nM of repebody or nivolumab (Anti-hPD1-Ni-hIgG4 (S228P), InvivoGen, San Diego, CA) was treated.Signals from hPD-L1 was detected by HRP-conjugated streptavidin(Biolegend, San Diego, CA). For competitive ELISA between a repe-body and nivolumab, a 96-well plate coated with 2 mg/mL of hPD-1 was blocked by a blocking buffer followed by addition of 60 nMof a myc-tagged repebody with nivolumab at different molarratios. Remaining repebody was detected by anti-c-myc-HRP anti-body (Santa Cruz Biotechnology, Dallas, TX).

Flow cytometry

CHO-K1 cells expressing hPD-1 were constructed by transienttransfection. Briefly, 3 mg of hPD-1/pCMV3 (Sino Biological) wasintroduced into 1� 105 CHO-K1 cells using lipofectamine transfec-tion reagent (Invitrogen). After incubation for 48 h post transfec-tion, CHO-K1 cells were harvested and resuspended in 1ml 1%BSA, DPBS. 200 mL of cell suspension (�1� 105 cells) were treatedwith 10mg/mL of a myc-tagged repebody for 30min on ice,washed three times and stained with goat-anti-myc-FITC antibody(Abcam, UK). After incubation for 30min on ice, cells were washedand analysed by LSR Fortessa (BD Biosciences, San Diego, CA) andFACS-Diva software (BD Biosciences).

In vitro bioassay for blockade of PD-1 and PD-L1 interaction

Blockade of the interaction between PD-1 and PL-L1 in vitro wasassayed using the commercial kit (PD-1/PD-L1 Blockade Bioassaykit, Promega, Fitchburg, WI). PD-1 effector cells (hPD-1-expressingjurkat T-cells) were manipulated to express luciferase by an NFATresponse element (NFAT-RE) through TCR activation. PD-1 effectorcells and PD-L1-expressing aAPC/CHO-K1 cells were co-cultured inthe presence of a repebody or nivolumab. After incubation, lumi-nescence was measured using a luminescence plate reader. EC50values were determined using GraphPad Prism software.

Analysis of cytokine production

Human PBMCs were isolated by density gradient centrifugationusing Ficoll-PaqueTM PLUS (GE Healthcare, Chicago, IL). CD4þ

T-cells were isolated from PBMCs using Dynabeads CD4 positiveisolation kit (Invitrogen). Monocytes were isolated from PBMCsusing Monocyte isolation kit II (Miltenyi Biotec, Germany), andused for generating dendritic cells (DCs). Briefly, isolated mono-cytes were cultured in RPMI-1640 medium containing 10% FBS,5 ng/mL GM-CSF (R&D Systems, Minneapolis, MN), and 20 ng/mLIL-4 (R&D Systems), and the medium was refreshed every 2 days.At day 6, the medium was refreshed with RPMI-1640 containing10% FBS, 5 ng/mL GM-CSF, 20 ng/mL IL-4, and 10 ng/mL TNF-a(R&D Systems), followed by incubation for 24 h and harvest.For assay of a mixed lymphocyte reaction (MLR), CD4þ T-cells(1� 105) were co-cultured with allogeneic DCs (1� 104) in a96-well cell culture plate in the presence of off-target repebody,anti-PD-1 repebody, or nivolumab. After 5 days of incubation,the supernatant was collected to determine the levels of IL-2 and

2 S. SON ET AL.

IFN-c by ELISA kit (BD OptEIATM Set for human IL-2 and humanIFN-c, BD Biosciences) according to the manufacturer’s protocols.

In vivo experiments

All animal studies were approved by the Institutional Animal Careand Use Committee (IACUC, Korea) and carried out according tothe guidelines described by the committee.

The pharmacokinetic profile of the repebody was analysedusing male balb/c mice with 4–5weeks of age (Joong Ah Bio,Korea). Mice were intravenously administered with the repebody(r_G9) (10mg/kg, 100 mL), and blood samples were obtained attime intervals after single dosing. Serum concentrations of therepebody were determined by sandwich ELISA. Briefly, 10mg/mLof polyclonal anti-repebody antibody were coated on a 96-wellELISA plate by overnight incubation at 4 �C. Each well was blockedby SuperBlock T20 (PBS) Blocking Buffer (Thermo Fisher Scientific,Waltham, MA), followed by addition of 100 mL of serum samples.After incubation for 2 h at room temperature, 1 mg/mL of biotiny-lated polyclonal anti-repebody antibody and HRP-conjugatedstreptavidin diluted to 1/1000 (Biolegend) were sequentiallytreated to detect the levels of the repebody in serum. Signalsfrom HRP were developed by TMB, and the reaction was stoppedwith 100 mL of 1 N H2SO4. Absorbance measured at 450 nm wereconverted to percent for relative comparison of serum concentra-tion of the repebody. The initial and terminal half-lives of therepebody were determined using GraphPad Prism software.

For generation of the xenograft mouse model, immune defi-cient NOD/ShiLtJ-Rag2em1AMC Il2rgem1AMC (NRGA) mice (6–8weeksof age, female) were purchased from Joong Ah Bio. NCI-H292tumour cells (2� 106 in 25 mL) and freshly isolated human PBMC(5� 105 in 25 mL) were mixed with 50 mL of BD Matrigel Matrix (BDBiosciences), and subcutaneously implanted into the flanks ofNRGA mice (8 mice per each group). After inoculation of tumourcells, 100 mL of anti-PD-1 repebody or off-target repebody wasintraperitoneally (IP) administered at day 1, 3, 7 and 10. DPBS wasalso used as a negative control. Dose of a repebody was 10mg/kgper each injection. The body weight and tumour volumes weremonitored every 3 or 4 days from a week post implantation.Tumour volumes were measured using calliper and calculatedaccording to the formula: V ¼ (W2 � L)/2 (mm3) (V is tumour vol-ume, W is tumour width, and L is tumour length). Mice weremonitored for a month and sacrificed at day 31 for extirpation oftumours. Statistical analysis was performed by one-way ANOVAusing GraphPad Prism software.

Results

Development of a repebody specific for hPD-1

To develop a repebody that specifically binds to hPD-1, we con-structed a repebody library by randomising six variable sites ontwo nearby modules (LRRV2 and LRRV3 modules) using NNKdegenerate codon followed by phage display selection againstECD of hPD-1 (Figure 1(A)). After five rounds of soluble biopan-ning, a repebody clone (r_A1) with the highest binding affinitywas selected. The selected repebody was shown to have threemutations on LRRV2 (I91N, T93W, and G94E) and additional threemutations on LRRV3 (V115T, V117F, and E118L) (Figure 1(B)).The binding affinity (KD) of r_A1 for hPD-1 was estimated to be617 nM through isothermal titration calorimetry (ITC)(Supplementary Figure S1A). Although hPD-1 is known to have alow binding affinity for hPD-L1 (KD ¼ 8.2mM) [38], a repebody

with higher affinity for hPD-1 is expected to be more effective forinhibiting the interaction between PD-1 and its ligand, PD-L1. Toincrease the binding affinity of r_A1, random mutations were fur-ther introduced into additional four variable sites on LRRV4 mod-ule of r_A1 (Figure 1(A)), followed by phage display selection andthree rounds of soluble biopanning. As a result, r_G9 with twoadditional mutations (A141K and H142F) was finally selected(Figure 1(B)). The binding affinity of r_G9 for hPD-1 was deter-mined to be about 17.6 nM, showing a 35-fold increase comparedto the initially selected repebody (Supplementary Figure S1B). It isknown that hPD-1 has four N-glycosylation sites on ECD (49N,58N, 74N, and 116N). We thus checked whether glycosylation ofhPD-1 affects its binding to r_G9. To this end, we measured abinding affinity of r_G9 for ECD of hPD-1 expressed by E. coli. TheKD value of r_G9 for non-glycosylated ECD of hPD-1 was esti-mated to be 28.9 nM (Supplementary Figure S1C), and this impliesa negligible effect of glycosylation on its binding to r_G9.

We next investigated a binding specificity of r_G9, andobserved that it had high specificity for hPD-1 (Figure 1(C)). Eventhough hPD-1 has the sequence identity as high as 64% withmouse PD-1 (mPD-1), r_G9 showed a negligible cross-reactivityagainst mPD-1 as well as human and mouse PD-L1. We checked ifthe repebody binds to a serum albumin, because its binding to aserum albumin could affect its half-life and distribution [39]. Therepebody was shown to have a negligible binding activity forserum albumins from mouse, rat, rabbit, and human(Supplementary Figure S2), verifying high specificity for hPD-1.

Inhibitory effect on the PD-1 and PD-L1 interaction

To check a potential use of r_G9 as an immune checkpoint inhibi-tor, we tested its inhibitory effect on the interaction betweenhPD1 and hPD-L1 through competitive ELISA. As a result, r_G9was shown to effectively block the interaction between hPD-1 andhPD-L1 (Figure 2(A)). In contrast, inhibition by initially selectedrepebody (r_A1) was much lower than r_G9, and this seems to bedue its lower binding affinity for hPD-1. Off-target repebody(r_Off) had a negligible inhibitory effect on the interactionbetween hPD-1 and hPD-L1. We tested the effect of the r_G9 con-centration, and observed a dose-dependent inhibition on thebinding of hPD-1 to hPD-1, whereas off-target repebody had amarginal effect (Figure 2(B)). Interestingly, r_G9 was shown toexhibit a comparable inhibitory effect to anti-PD-1 antibody, nivo-lumab (Figure 2(C)). These results indicate that r_G9 specific forhPD-1 has a potential as an immune checkpoint inhibitor. We fur-ther attempted to get insight into a binding region of r_G9 bycomparing with nivolumab of which crystal structure in complexwith hPD-1 is available [40]. When nivolumab was added to a mix-ture of hPD-1 and r_G9 at the increasing molar ratio, the bindingof r_G9 to hPD-1 decreased, but became saturated even thoughthe molar ratio was above 1 (Figure 2(D)). This seems to be dueto the fact that binding affinity of nivolumab (�3 nM) for hPD-1 iscomparable to r_G9 [41]. Based on the result, it is likely that r_G9shares the epitope with nivolumab.

Restoration of T-cell function

We tested if the hPD-1-specific repebody can restore T-cell func-tion. For this, we first examined if r_G9 binds to native hPD-1on the cell surface using hPD-1-expressing CHO-K1 cells(Supplementary Figure S3). FACS analysis revealed a shifted peakonly when r_G9 was treated with hPD-1-expressing CHO-K1 cells(Figure 3(A)). It is thus likely that r_G9 specifically binds to native

JOURNAL OF DRUG TARGETING 3

hPD-1 expressed on the cell surface as well as ECD domain ofhPD-1 in solution. Inhibition assay for the interaction betweenPD-1 and PD-L1 was carried out to assess the biological activity ofr_G9 in vitro. Human PD-1-expressing jurkat T-cells were

co-cultured with aAPC/CHO-K1 cells expressing PD-L1 followed byaddition of r_G9, and expression of luciferase which is driven byactivation of TCR signalling pathway was measured. As a result,TCR-mediated luminescence signal was shown to increase with

Figure 1. Construction of a library and selection of anti-PD-1 repebodies through phage display. (A) Randomised sites for a library construction are shown. Six variablesites on LRRV2 and LRRV3 modules were mutated for screening of initial binders (yellow color). Additional four variable sites on LRRV4 module (colored in purple)were mutated for affinity maturation. (B) Amino acid sequences of selected repebodies from the first round of biopanning (r_A1) and the second round of affinity mat-uration (r_G9). (C) Binding specificity of r_G9 was tested by ELISA. Each ECD of hPD-1, hPD-L1, mPD-1, and mPD-L1 was coated on a 96-well plate. Binding of myc-tagged r_G9 was detected using anti-c-myc-HRP antibody. BSA and Off-target repebody (r_Off) were used as negative control. Data represent the mean 6 standarddeviation (n¼ 3).

4 S. SON ET AL.

the increasing concentration of r_G9, and EC50 of r_G9 was esti-mated to be 0.84mg/mL, which is comparable to nivolumab (EC50¼ 0.32 mg/mL) (Figure 3(B)). This result indicates that r_G9 effect-ively inhibited the interaction between PD-1 and PD-L1 in vitro,leading to activation of TCR signalling pathway.

We next performed a mixed lymphocyte reaction (MLR) tocheck if r_G9 could restore T-cell function. PD-1 blockades havebeen known to result in an increase in cytokine secretion throughT-cell activation [42]. T-cells isolated from human PBMCs were co-cultured with allogeneic DCs, and a repebody was treated for5 days, and the levels of IL-2 and IFN-c secreted from T-cells weremeasured. As expected, r_G9 was observed to effectively enhancethe secretion of IL-2 and IFN-c in a dose-dependent manner,whereas off-target repebody had a negligible effect (Figure4(A,B)). Interestingly, r_G9 exhibited a comparable effect on thesecretion of cytokines by T-cells to nivolumab (Figure 4(C,D)). Asfor the secretion levels of IL-2 and IFN-c, no significant differencewas observed between r_G9 and nivolumab (p values of 0.88 and0.48, respectively). The results indicate that r_G9 binds to hPD-1and inhibits PD-1 signalling pathway, resulting in activation ofTCR signalling.

Pharmacokinetics and in vivo anti-tumour activity

We analysed the pharmacokinetic profile of the repebody in miceto assess its half-life. To this end, r_G9 (10mg/kg) was intraven-ously administered into 4–5weeks old male balb/c mice. Bloodsamples were collected over time, and the levels of the blood-

circulating repebody were determined (Supplementary Figure S4).The initial and terminal serum half-lives were estimated to be 0.2and 4.5 h, respectively. The repebody with a small size (�30 kDa)seems to have a short half-life due to fast renal clearance com-pared to immunoglobulin antibodies [43].

We next evaluated anti-tumour activity of the repebody intumour mouse model. NCI-H292 tumour cells were co-implantedwith human PBMCs into the right flank of immune deficient NRGAmice, and tumour growth was traced. Compared to the controlsusing DPBS and off-target repebody, r_G9 was shown to signifi-cantly suppress the tumour growth (p< 0.001) (Figure 5(A)). Whenr_G9 was administered four times over two weeks after tumourimplantation, tumour progression was significantly blocked until31 days. Average tumour volumes of mice treated with r_G9 weremaintained within the range of 97mm3 to 106mm3, whereastumours were grown to 501mm3 and 582mm3 for DPBS- and off-target repebody-treated mice, respectively. Representative imagesof tumours from sacrificed mice on day 31 also confirmed astrong anti-tumour activity of r_G9 (Figure 5(B)). This result indi-cates that r_G9 blocked effectively the interaction between PD-1and PD-L1, thus significantly suppressing the tumour growth.Based on the results, it is likely that r_G9 can act as an efficaciousimmune checkpoint inhibitor.

Discussion

We demonstrated an effective anti-tumour activity of a small-sizedrepebody specific for hPD-1 as an immune checkpoint blockade.

Figure 2. Inhibitory effect of a repebody on the interaction between PD-1 and PD-L1. (A) Inhibitory effects of a repebody on the interaction between PD-1 and PD-L1were tested by competitive ELISA. Off-target repebody (r_Off) was used as a negative control. (B) Effect of the repebody concentration on the inhibition of PD-1 andPD-L1 interaction. Competitive ELISA was carried out in the presence of different repebody concentrations. (C) Inhibitory effect of r_G9 was compared with nivolumabby competitive ELISA. Concentrations of r_G9 and nivolumab were 200 nM. Signals from biotinylated hPD-L1 were generated using HRP-conjugated streptavidin. (D)Binding competition between r_G9 and nivolumab was assayed by competitive ELISA. Nivolumab was added with r_G9 to hPD-1 at different molar ratios (nivolumab/repebody ¼ 0, 0.5, 1, and 2), and relative binding portion of r_G9 was determined. Data represent the mean 6 standard deviation (n¼ 3).

JOURNAL OF DRUG TARGETING 5

The repebody, r_G9, which has a binding affinity of 17.6 nM forhPD-1, showed high specificity for hPD-1. The repebody wasshown to bind to hPD-1 about 470-times stronger than hPD-L1(KD of hPD-L1¼ 8.2mM) [38], effectively inhibiting the interactionbetween hPD-1 and its ligand PD-L1. The repebody was predictedto share the epitope with clinically used monoclonal antibody,nivolumab. As an effective antagonist of PD-L1, the repebodyrestored the TCR signalling pathway, stimulating cytokine secre-tion. In vivo study using tumour mouse model showed that therepebody has a strong anti-tumour activity, significantly suppress-ing the tumour growth. These results strongly support that therepebody specific for hPD-1 can have a potential use for cancerimmunotherapy.

Cancer immunotherapy with immune checkpoint inhibitors hasshown notable successes, and several monoclonal antibodies,including ipilimumab (anti-CTLA-4 antibody), nivolumab and pem-brolizumab (anti-PD-1 antibody), have been approved by FDA,and are clinically used. However, monoclonal antibodies have sev-eral disadvantages, including aggregation and poor tumour pene-tration due to a large size, difficulty with engineering, highproduction cost and immune-related adverse effect [18,44,45]. Asan alternative to immunoglobulin antibodies, the repebody scaf-fold was shown to offer some favourable properties: overproduc-tion by bacterial expression system, easy and simple affinity

maturation, easy engineering and design, high tissue penetration[31,46,47]. In this study, the hPD-1-specific repebody (r_G9) withhigh binding affinity was successfully developed through phagedisplay and modular evolution approach. Its high binding affinitycan be a distinct advantage over other small molecules or pepti-des showing weak binding affinity for hPD-1. High specificity ofthe repebody seems to stem from intrinsic property of the LRRproteins with b-strands at the concave region. Considering themode of action of immune checkpoint inhibitors, high tissuepenetration of the repebody due to small size (�30 kDa) would beanother distinct property in their applications to solid tumourscompared to immunoglobulin antibodies[48,49]. Although engi-neered high affinity PD-1 or PD-L1 variants with a similar molecu-lar size to the repebody were attempted, they were expressed asinclusion body in E. coli or obtained from insect cell expressionsystem [27,28].

In recent, combination therapy has attracting much attentionbecause of an increased therapeutic efficacy. Anti-PD-1 repebodymight be used with conventional immune checkpoint antibodies(e.g. anti-CTLA-4 antibody, ipilimumab) since dual blockade ofCTLA-4 and PD-1 is known to be most effective combination interms of therapeutic efficacy [50]. It is also worth a combinationwith other repebodies targeting various pathways related to can-cers. As an example, VEGF-targeting repebody [51], which was

Figure 3. Binding of a repebody to hPD-1 expressed on the cell surface and activation of TCR signalling. (A) Binding of r_G9 to hPD-1 expressed on the cell surfacewas analysed by FACS. Native CHO-K1 cells (PD-1 negative) or PD-1-expressing CHO-K1 cells (PD-1 positive) were treated with myc-tagged repebody and stained withgoat-anti-myc-FITC antibody followed by FACS analysis. (B) Inhibitory effect on the interaction between PD-1 and PD-L1 was assayed in vitro by measuring the lumines-cence intensity resulting from TCR activation, and EC50 value was determined using GraphPad Prism software. PD-1-expressing T-cells and PD-L1-expressing aAPC/CHO-K1 cells were co-cultured in the presence of r_G9, r_Off, or nivolumab, and luminescence intensity was measured. Data represent the mean 6 standard devi-ation (n¼ 3).

6 S. SON ET AL.

previously developed, might be a candidate for combination ther-apy because combinatorial inhibition of the VEGF and PD-1 path-ways exhibited a synergistic effect in advanced RCC [52,53]. At thesame time, anti-PD-1 repebody can be employed for in vivo imag-ing as companion diagnostics to check the expression of PD-1 onthe cells [35,54]. In conclusion, anti-PD-1 repebody is likely to finda potential use as an immune checkpoint inhibitor in cancerimmunotherapy.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

The research was supported by the Global Research Laboratory[Grant No. NRF-2015K1A1A2033346], the Mid-Career ResearcherProgramme [Grant No. NRF-2017R1A2A1A05001091], the NewDrug Performance Platform Technology Development Programme[Grant No. NRF-2017M3A9F5031419] funded by Ministry of Scienceand ICT and Brain Korea21 funded by Ministry of Education. Son,S. was supported by The Hyundai Motor Chung Mong-Koo Foundation.

References

[1] Greenwald RJ, Freeman GJ, Sharpe AH. The B7 family revis-ited. Annu Rev Immunol. 2005;23:515–548.

[2] Parry RV, Chemnitz JM, Frauwirth KA, et al. CTLA-4 and PD-1 receptors inhibit T-cell activation by distinct mechanisms.Mol Cell Biol. 2005;25(21):9543–9553.

Figure 5. In vivo anti-tumour activity of a repebody. (A) Xenograft mice with NCI-H292 and human PBMC (E:T ratio ¼ 1:4) were administered intraperitonially withthe PD-1-specific (r_G9) or off-target repebody (r_Off) at 10mg/kg or DPBS(vehicle). A repebody or DPBS was injected at the time indicated by arrows aftertumour inoculation. Tumour volumes were measured by calliper, and data indicategeometric means of tumour volumes 6 standard error of the mean (n¼ 8).Statistically significant difference is indicated, ���p< 0.001 (one-way ANOVA). (B)Representative images of tumours from sacrificed mice after treatment with r_G9,r_Off, and DPBS. Tumours were excised from sacrificed mice on day 31 aftertumour inoculation. Scale bar: 5mm.

Figure 4. Effect of a repebody on the cytokine production by T-cells. (A, B) Allogeneic MLR (mixed lymphocyte reaction) in the presence of different repebody concentra-tions. After 5 days of treatment, the levels of IL-2 and INF-c in the supernatant were determined by ELISA. (C, D) Comparison of cytokine secretion between r_G9 andnivolumab by allogeneic MLR. The levels of IL-2 and INF-c in the supernatant were determined by ELISA. Data indicate the mean± standard error of the mean (n¼ 4).

JOURNAL OF DRUG TARGETING 7

[3] Bennett F, Luxenberg D, Ling V, et al. Program death-1engagement upon TCR activation has distinct effects oncostimulation and cytokine-driven proliferation: attenuationof ICOS, IL-4, and IL-21, but not CD28, IL-7, and IL-15Responses. J Immunol. 2003;170(2):711–718.

[4] Drake CG, Jaffee E, Pardoll DM. Mechanisms of immuneevasion by tumors. Adv Immunol.. 2006;90:51–81.

[5] Iwai Y, Ishida M, Tanaka Y, et al. Involvement of PD-L1 ontumor cells in the escape from host immune system andtumor immunotherapy by PD-L1 blockade. Proc Natl AcadSci USA. 2002;99(19):12293–12297. 200

[6] Topalian SL, Hodi FS, Brahmer JR, et al. Safety, activity, andimmune correlates of anti-PD-1 antibody in cancer. N EnglJ Med. 2012;366(26):2443–2454.

[7] Sunshine J, Taube JM. PD-1/PD-L1 inhibitors. Curr OpinPharmacol. 2015;23:32–38.

[8] Patnaik A, Kang SP, Rasco D, et al. Phase I study of pem-brolizumab (MK-3475; anti-PD-1 monoclonal antibody) inpatients with advanced solid tumors. Clin Cancer Res. 2015;21(19):4286–4293.

[9] Berger R, Rotem-Yehudar R, Slama G, et al. Phase I safetyand pharmacokinetic study of CT-011, a humanized anti-body interacting with PD-1, in patients with advancedhematologic malignancies. Clin Cancer Res. 2008;14(10):3044–3051.

[10] Brahmer JR, Drake CG, Wollner I, et al. Phase I study ofsingle-agent anti-programmed death-1 (MDX-1106) inrefractory solid tumors: safety, clinical activity, pharmaco-dynamics, and immunologic correlates. JCO. 2010;28(19):3167–3175.

[11] Ohaegbulam KC, Assal A, Lazar-Molnar E, et al. Human can-cer immunotherapy with antibodies to the PD-1 and PD-L1pathway. Trends Mol Med. 2015;21(1):24–33.

[12] Wang X, Guo G, Guan H, et al. Challenges and potential ofPD-1/PD-L1 checkpoint blockade immunotherapy for glio-blastoma. J Exp Clin Cancer Res. 2019;38(1):87

[13] Looi C-K, Chung FF-L, Leong C-O, et al. Therapeuticchallenges and current immunomodulatory strategies intargeting the immunosuppressive pancreatic tumor micro-environment. J Exp Clin Cancer Res. 2019;38(1):162.

[14] Lee CM, Tannock IF. The distribution of the therapeuticmonoclonal antibodies cetuximab and trastuzumab withinsolid tumors. BMC Cancer. 2010;10(1):255.

[15] Francis DM, Thomas SN. Progress and opportunities forenhancing the delivery and efficacy of checkpoint inhibitorsfor cancer immunotherapy. Adv Drug Deliv Rev. 2017;114:33–42.

[16] Deng R, Bumbaca D, Pastuskovas CV, et al. Preclinicalpharmacokinetics, pharmacodynamics, tissue distribution,and tumor penetration of anti-PD-L1 monoclonal antibody,an immune checkpoint inhibitor. MAbs. 2016;8(3):593–603.

[17] Fujimori K, Covell DG, Fletcher JE, et al. A modeling analysisof monoclonal antibody percolation through tumors: abinding-site barrier. J Nucl Med. 1990;31(7):1191–1198.

[18] Naidoo J, Page DB, Li BT, et al. Toxicities of the anti-PD-1and anti-PD-L1 immune checkpoint antibodies. Ann OncolOff J Eur Soc Med Oncol. 2015;26:2375–2391.

[19] Fecher LA, Agarwala SS, Hodi FS, et al. Ipilimumab and itstoxicities: a multidisciplinary approach. Oncologist. 2013;18(6):733–743.

[20] Ribas A, Hodi FS, Kefford R, et al. Efficacy and safety of theanti-PD-1 monoclonal antibody MK-3475 in 411 patients

(pts) with melanoma (MEL). JCO. 2014;32(15_suppl):LBA9000.

[21] Guan M, Zhou Y-P, Sun J-L, et al. Adverse events of mono-clonal antibodies used for cancer therapy. Biomed Res Int.2015;2015:1.

[22] Zhan M-M, Hu X-Q, Liu X-X, et al. From monoclonal anti-bodies to small molecules: the development of inhibitorstargeting the PD-1/PD-L1 pathway. Drug Discov Today.2016;21(6):1027–1036.

[23] Chang H-N, Liu B-Y, Qi Y-K, et al. Blocking of the PD-1/PD-L1 interaction by a D-peptide antagonist for cancerimmunotherapy. Angew Chem Int Ed. 2015;54(40):11760–11764.

[24] Li Q, Quan L, Lyu J, et al. Discovery of peptide inhibitorstargeting human programmed death 1 (PD-1) receptor.Oncotarget. 2016;7:64967–64976.

[25] Li K, Tian H. Development of small-molecule immunecheckpoint inhibitors of PD-1/PD-L1 as a new therapeuticstrategy for tumour immunotherapy. J Drug Target. 2019;27(3):244–256.

[26] Boohaker RJ, Sambandam V, Segura I, et al. Rational designand development of a peptide inhibitor for the PD-1/PD-L1interaction. Cancer Lett. 2018;434:11–21.

[27] Maute RL, Gordon SR, Mayer AT, et al. Engineering high-affinity PD-1 variants for optimized immunotherapy andimmuno-PET imaging. Proc Natl Acad Sci USA. 2015;112(47):E6506–E6514.

[28] Liang Z, Tian Y, Cai W, et al. High-affinity human PD-L1 var-iants attenuate the suppression of T cell activation.Oncotarget. 2017;8(51):88360–88375.

[29] Richards DA. Exploring alternative antibody scaffolds: anti-body fragments and antibody mimics for targeted drugdelivery. Drug Discov Today Technol. 2018;30:35–46.

[30] Vazquez-Lombardi R, Phan TG, Zimmermann C, et al.Challenges and opportunities for non-antibody scaffolddrugs. Drug Discov Today. 2015;20(10):1271–1283.

[31] Lee S-C, Park K, Han J, et al. Design of a binding scaffoldbased on variable lymphocyte receptors of jawless verte-brates by module engineering. Proc Natl Acad Sci. 2012;109(9):3299–3304.

[32] Lee J, Kim HJ, Yang C-S, et al. A high-affinity protein binderthat blocks the IL-6/STAT3 signaling pathway effectivelysuppresses non–small cell lung cancer. Mol Ther. 2014;22(7):1254–1265.

[33] Lee J, Kang JA, Ryu Y, et al. Genetically engineered andself-assembled oncolytic protein nanoparticles for targetedcancer therapy. Biomaterials. 2017;120:22–31.

[34] Seo H-D, Lee J-J, Kim YJ, et al. Alkaline phosphatase-fusedrepebody as a new format of immuno-reagent for animmunoassay. Anal Chim Acta. 2017;950:184–191.

[35] Yun M, Kim D-Y, Lee J-J, et al. A High-affinity repebody formolecular imaging of EGFR-expressing malignant tumors.Theranostics. 2017;7(10):2620–2633.

[36] Liu S, Tobias R, McClure S, et al. Removal of endotoxinfrom recombinant protein preparations. Clin Biochem.1997;30(6):455–463.

[37] Teodorowicz M, Perdijk O, Verhoek I, et al. Optimized TritonX-114 assisted lipopolysaccharide (LPS) removal methodreveals the immunomodulatory effect of food proteins.PLoS One. 2017;12(3):e0173778.

[38] Cheng X, Veverka V, Radhakrishnan A, et al. Structure andinteractions of the human programmed cell death 1 recep-tor. J Biol Chem. 2013;288(17):11771–11785.

8 S. SON ET AL.

[39] Smith DA, Di L, Kerns EH. The effect of plasma proteinbinding on in vivo efficacy: misconceptions in drug discov-ery. Nat Rev Drug Discov. 2010;9(12):929–939.

[40] Lee JY, Lee HT, Shin W, et al. Structural basis of checkpointblockade by monoclonal antibodies in cancer immunother-apy. Nat Commun. 2016;7:13354.

[41] Tan S, Zhang H, Chai Y, et al. An unexpected N-terminalloop in PD-1 dominates binding by nivolumab. NatCommun. 2017;8:14369

[42] Wang C, Thudium KB, Han M, et al. In vitro characterizationof the anti-PD-1 antibody nivolumab, BMS-936558, and invivo toxicology in non-human primates. Cancer ImmunolRes. 2014;2:20–22.

[43] Centanni M, Moes D, Troc�oniz IF, et al. Clinical pharmaco-kinetics and pharmacodynamics of immune checkpointinhibitors. Clin Pharmacokinet. 2019;58(7):835–857.

[44] Coulson A, Levy A, Gossell-Williams M. Monoclonal antibod-ies in cancer therapy: mechanisms, successes and limita-tions. W Indian Med J. 2014;63(6):650–654.

[45] Chames P, Van Regenmortel M, Weiss E, et al. Therapeuticantibodies: successes, limitations and hopes for the future.Br J Pharmacol. 2009;157(2):220–233.

[46] Hwang D-E, Shin Y-K, Munashingha PR, et al. A repeat pro-tein-based DNA polymerase inhibitor for an efficient andaccurate gene amplification by PCR. Biotechnol Bioeng.2016;113(12):2544–2552.

[47] Hwang D-E, Choi J-M, Yang C-S, et al. Effective suppressionof C5a-induced proinflammatory response using

anti-human C5a repebody. Biochem Biophys Res Commun.2016;477(4):1072–1077.

[48] Sun Q, Ojha T, Kiessling F, et al. Enhancing tumor penetra-tion of nanomedicines. Biomacromolecules. 2017;18(5):1449–1459.

[49] Cabral H, Matsumoto Y, Mizuno K, et al. Accumulation ofsub-100 nm polymeric micelles in poorly permeabletumours depends on size. Nat Nanotechnol. 2011;6(12):815–823.

[50] Rotte A. Combination of CTLA-4 and PD-1 blockers fortreatment of cancer. J Exp Clin Cancer Res. 2019;38(1):255

[51] Hwang D-E, Ryou J-H, Oh JR, et al. Anti-human VEGF repe-body effectively suppresses choroidal neovascularizationand vascular leakage. PLoS One. 2016;11(3):e0152522.

[52] Wang M, Liu Y, Cheng Y, et al. Immune checkpoint block-ade and its combination therapy with small-molecule inhib-itors for cancer treatment. Biochim Biophys Acta RevCancer. 2019;1871(2):199–224.

[53] Yasuda S, Sho M, Yamato I, et al. Simultaneous blockadeof programmed death 1 and vascular endothelial growthfactor receptor 2 (VEGFR2) induces synergistic anti-tumour effect in vivo. Clin Exp Immunol. 2013;172(3):500–506.

[54] Broos K, Lecocq Q, Raes G, et al. Noninvasive imaging ofthe PD-1:PD-L1 immune checkpoint: Embracing nuclearmedicine for the benefit of personalized immunotherapy.Theranostics. 2018;8(13):3559–3570.

JOURNAL OF DRUG TARGETING 9