Novel method for screening functional antibody with comprehensive analysis of its immunoliposome

Development of monoclonal antibody is critical for targeted drug delivery because its characteristics determine improved therapeutic efficacy and reduced side-effect. Antibody therapeutics target surface molecules; hence, internalization is desired for drug delivery. As an antibody–drug conjugate, a critical parameter is drug-to-antibody ratio wherein the quantity of drugs attached to the antibody influences the antibody structure, stability, and efficacy. Here, we established a cell-based immunotoxin screening system to facilitate the isolation of functional antibodies with internalization capacities, and discovered an anti-human CD71 monoclonal antibody. To overcome the limitation of drug-to-antibody ratio, we employed the encapsulation capacity of liposome, and developed anti-CD71 antibody-conjugated liposome that demonstrated antigen–antibody dependent cellular uptake when its synthesis was optimized. Furthermore, anti-CD71 antibody-conjugated liposome encapsulating doxorubicin demonstrated antigen–antibody dependent cytotoxicity. In summary, this study demonstrates the powerful pipeline to discover novel functional antibodies, and the optimal method to synthesize immunoliposomes. This versatile technology offers a rapid and direct approach to generate antibodies suitable for drug delivery modalities.

In recent years, antibody-drug conjugates (ADCs) in which monoclonal antibodies are attached to cytotoxic payloads by non-cleavable or cleavable linkers have emerged as highly potent pharmaceutical modalities because of target specificity and target-binding affinity that collectively contribute to targeted delivery of the drugs as well as reduced side effects [1][2][3] . Among these components, monoclonal antibodies are critical because therapeutic properties of ADCs are partially dependent on the characteristics of their antigens. The efficacies of ADCs are primarily dependent on the expression patterns of the targeted antigens; however, recent reports revealed that successful ADCs possess internalization properties that will facilitate them to be transported into the cells and enhance their pharmacological effects 4,5 . Therefore, a proper screening process to identify suitable monoclonal antibodies is essential for development of ADCs.
Traditionally, common assays to screen for monoclonal antibodies include enzyme-linked immunosorbent assay, flow cytometry, and immunoblotting analyses. While these procedures may assist in characterizing target specificity and target-binding affinity, they are not suitable to determine internalization properties of the antibody. To this end, we previously reported a cell-based immunotoxin screening system to facilitate the isolation of functional antibodies with internalization capacities 6 . Functional antibodies were selected from the hybridoma library through formation of immunotoxins via Fc-mediated coupling of antibodies with engineered diphtheria toxin DT3C, and screening for conditional lethality of the target cells if the immunotoxins were internalized. This rapid and straightforward screening strategy possesses the following advantages: 1) selection of potent antibodies that effectively bind to the cell-surface epitopes, and 2) selection of the antibodies that are efficiently internalized into the cells. Collectively, our approach allows direct discovery of potent ADC-compatible antibodies.

Results
Immunotoxin screening of anti-U87 hybridoma library identified 214D8 clones. A hybridoma library was generated through immunization of BALB/c mice with U87 cells, fusion of immunized splenocytes and P3U1 myeloma cells, and selective growth of hybridomas in the 96-well plates containing culture media with hypoxanthine-aminopterin-thymidine (HAT) (Fig. 1a). In total, 312 out of 384 wells (81%) contained at least one colony of hybridoma cells (data not shown). Supernatants from the hybridoma library were pre-incubated with engineered toxin DT3C to form immunotoxins (Fig. 1b). DT3C is a recombinant protein that consists of diphtheria toxin (DT) without the receptor-binding domain but containing the Fc-binding domains of Streptococcus protein G (3C) 10 . As summarized in Fig. 1c, if the antibody:DT3C immunocomplex recognizes an antigen expressed on the cell surface, then the immunotoxin is internalized wherein DT3C is cleaved by the cytosolic furin protease, and catalytic domain of DT3C is released into the cytoplasm. The catalytic domain causes ADPribosylation of elongation factor (EF)-2, which subsequently leads to cytotoxicity via inhibition of the protein translation machinery.
Exploiting the unique principle of DT3C immunotoxin assay, we performed primary screening to identify antibody-secreting hybridomas that were capable of inducing DT3C-dependent cytotoxicity. Visual observation of U87 morphology, as well as immunotoxin assay revealed 214A2 and 214D8 as putative hybridomas that secreted functional antibodies with DT3C-dependent cytotoxicity ( Supplementary Fig. 1). These positive hybridomas were subjected to limiting dilution for cloning, and supernatants from the clones were used to perform secondary screening for confirmation. Collectively, we established 214D8 clone that produced functional antibody with capacity for DT3C-dependent cytotoxicity. The immunotoxin is then internalized wherein translocated terminus of DT3C is cleaved by the cellular furin protease, and catalytic domain of DT3C (Cat) is released into the cytoplasm. Subsequently, the catalytic domain ADP-ribosylates elongation factor (EF)-2, which leads to cytotoxicity via inhibition of the protein translation machinery. www.nature.com/scientificreports/ 214D8 antibody recognized CD71/TFRC. To identify a putative antigen of 214D8 antibody (subclass: IgG 1 kappa), cell surface proteins were biotinylated with sulfo-NHS-biotin prior to immunoprecipitation (Fig. 2a). Glioblastoma cell line A172 was treated with sulfo-NHS-biotin, lysed with 1% NP40, and the putative antigen was subsequently immunoprecipitated by using 214D8. A 98-kDa band was detected by probing with streptavidin-HRP (Fig. 2b). We noticed similarities in the detected band patterns between 214D8 and 6E1 antibodies, a previously published antibody against human CD71 (hCD71)/TFRC 10 .
To determine if CD71 was an antigen of 214D8 antibody, purified recombinant hCD71 (rhCD71; 77.4 kDa) was immunoprecipitated by using 214D8 and 6E1 antibodies, and a 77 kDa band was detected by probing with anti-hCD71 polyclonal antibody (Fig. 2c). Additionally, sandwich ELISA also revealed immunoreactive binding between the rhCD71 and these antibodies (Fig. 2d). Human CD71-OFP expression vector was transiently transfected into the CHO cells, and these transfected cells were assessed by flow cytometry. As expected, while 214D8 and 6E1 antibodies did not react with CHO mock cells, both antibodies demonstrated reactivities against CHO hCD71-OFP cells wherein the number of OFP + FITC + double positive subpopulation reached 9.96% and 10.3%, respectively [ Fig. 2e; quadrant 2 (Q2)]. Additionally, we performed immunoprecipitation/immunoblotting by using the cell lysates prepared from the transiently transfected CHO cells, and determined that, while 214D8 and 6E1 did not immunoprecipitate any detectable proteins in the CHO mock lysate, both antibodies immunoprecipitated a 125 kDa protein from the CHO hCD71-OFP lysate, indicative of hCD71-OFP (Fig. 2b). Based on the experimental results above, we concluded that CD71 was a genuine antigen of 214D8 and 6E1 antibody.
Next, we used flow cytometry to assess expression patterns of the antigen recognized by 214D8 antibody and found varying immunoreactive patterns against 4 cell lines, A172, U87, SH-SY5Y, and H4 (Fig. 2f) To determine the subcellular localization of CD71, A172 cells were fixed with 4% PFA, incubated with primary antibodies and subsequently with Alexa555-conjugated anti-mouse IgG secondary antibody, and counterstained with DAPI. Both 214D8 and 6E1 antibodies demonstrated similar membranous and cytoplasmic staining patterns (Fig. 2g). Taken together, these results highly suggested that CD71 was a genuine antigen of 214D8 antibody.
Immunotoxin activity of anti-CD71 antibodies. To assess cellular cytotoxicity of newly established 214D8 antibody, we treated cell lines with mIgG, 6E1, and 214D8 antibodies alone or the corresponding antibody:DT3C immunotoxins. Cytotoxicity was evaluated after incubation of the analytes for 3 days by WST-1 assay for measuring cellular viability. As shown in Fig. 3a and Supplementary Fig. 2a Anti-CD71 immunoliposome demonstrated immunoreactivity against A172 cells. Given the limited quantity of cytotoxic payloads that can be attached to monoclonal antibody, we then extended our studies by exploring application of functional antibodies through conjugation with liposome. Immunoliposome was generated by the ethanol injection method whereby lipids and DiOC 18 (3) were dissolved in ethanol and rapidly injected into aqueous buffer, extruded through 0.2 µm, 0.1 µm, and 0.05 µm polycarbonate membrane filters to produce approximately 100 nm liposome, and immediately conjugated with antibody through crosslinking reaction with N-hydroxysuccinimide (NHS) ester (Fig. 4a). Optimized procedure resulted in utilizing NHS ester and full immunoglobulin as key reaction components, leading to multiple advantages including prompt reaction as well as preserved immunoreactivity. Physical properties of the immunoliposomes are summarized in Table 1. When compared with liposome without antibody conjugation, our mIgG-, 6E1-, and 214D8-conjugated liposomes displayed slightly larger size and less negative zeta potential, but when compared amongst 3 antibody-conjugated immunoliposomes, physiological properties did not differ significantly. Importantly, antibody conjugation percentages were highly controlled: 76% for mIgG-conjugated liposome, 69% for 6E1-conjugated liposome, and 60% for 214D8-conjugated liposome.
To further confirm immunoreactivity of our immunoliposomes, A172 cells were subjected to flow cytometric analysis by using liposome without antibody conjugation at the phospholipid (PL) concentration equivalent of containing 0.25 µg, 0.83 µg, 2.5 µg antibody (Fig. 4b). These results indicated that liposome without antibody conjugation did not demonstrate reactivity. In cases of 214D8-and 6E1-conjugated liposomes, both immunoliposomes demonstrated comparable immunoreactivities at the PL concentration equivalent of containing 2.5 µg antibody (Fig. 4c). With reduction of liposomal concentration, only 6E1-conjugated liposome maintained strong signal. Median fluorescent intensity levels of these immunoliposomes are summarized elsewhere (Supplementary Fig. 3a-c). These results indicated that both 214D8-and 6E1-conjugated liposomes were immunoreactive against A172 cells.
To determine equilibrium dissociation constant (K D ) of 214D8-and 6E1-conjugated liposomes, we utilized LigandTracer, an instrument for real-time measurement of binding of fluorescently labeled molecules to immobilized cells (Fig. 4d). We incubated A172 cells at 37 °C overnight, and subsequently measured baseline, association We attempted measuring K D values of Alexa488-conjugated antibodies, but we could not quantify their binding capacities due to low signal (data now shown).
Anti-CD71 antibody-conjugated liposome demonstrated enhanced cellular uptake. To evaluate cellular uptake of immunoliposomes by A172 cells, we encapsulated the immunoliposomes with DiOC 18 (3), which allowed fluorescent visualization and quantitative assessment of cellular uptake (Fig. 5a). Remarkably, at 10 µM liposomal (phospholipid) concentration, 214D8-conjugated liposome was gradually taken up by the cells, reaching its apex at 69 h (Fig. 5b). When compared with the control mIgG immunoliposome, 214D8and 6E1-conjugated liposomes exhibited 2.21-(214D8: 66 h) and 4.25-fold (6E1: 51 h) increases of the cellular uptake (Fig. 5c). Phase contrast and fluorescent images at 72 h after treatment are shown in Fig. 5d, indicating enhanced cellular uptake of fluorescence-labeled immunoliposome. With increase of liposomal concentration to 30 µM, the control mIgG-conjugated liposome was also taken up by the cells, minimizing the ratio between 214D8-or 6E1-conjugated liposomes and the control. Since the immunoliposomes were permanently administrated without washing steps, these results indicated that lowest concentration tested was sufficient to validate functionality of antigen-antibody dependent cellular uptake of the immunoliposome.

Discussion
Through generation of anti-U87 hybridoma library and screening of hybridoma supernatants treated with DT3C, we identified functional 214D8 antibody that targeted the surface molecule of U87 cells and were internalized to induce cytotoxicity. These findings provided evidence, such as targeting of cell surface molecules and cellular internalization, supporting 214D8 as a potential drug delivery agent. Subsequent biochemical and cellular analyses strongly indicated that the antigen of 214D8 antibody was CD71/TFRC. CD71 is a type 2 membrane protein expressed as a homodimer in the cell membrane that mediates internalization of diferric holo-transferrin by clathrin-mediated endocytosis 11,12 . While generally ubiquitous, CD71 is expressed on the surface of immature erythroid cells and placental tissues of healthy individuals 13,14 . Additionally, CD71 is expressed on the surface of brain capillary endothelial cells that form the blood-brain barrier (BBB), as well as choroid plexus epithelial cells that form the blood-cerebrospinal fluid (CSF) barrier; both of which play an important role in the developing brain 15 .
A majority of research efforts on CD71 has been conducted in the cancer research because increased expression levels of CD71 have been reported in the multiple forms of cancer including glioblastoma 16,17 . Since expression of CD71 has been demonstrated in the human glioblastoma cell line U87 18 , and we used U87 for immunization of BALB/c mice, it is conceivable that we obtained the monoclonal antibody against CD71 through our screening strategy. Given the enhanced expression of CD71 in the various types of cancer, CD71 has generated a great interest in producing monoclonal antibodies that can either induce cytotoxicity of cancer cells through direct inhibition of the receptor function or deliver therapeutic agents 19 . In case of 214D8 antibody, while we observed strong immunoreactivities against A172, U87, and SH-SY5Y cells by flow cytometry, we did not detect cytotoxicity of these cells by administration of the antibody alone. These findings suggested that 214D8 is more suited for delivery of therapeutic agents. In fact, when 214D8 was formed with DT3C, the immunotoxin was To maximize drug load while minimizing structural changes and instability of the antibody, we generated a drug delivery modality that consisted of phospholipid bilayer liposome conjugated with monoclonal antibody that was collectively designed for targeted delivery to the cells expressing its corresponding antigen. There are multiple advantages to liposomes including capacity to compartmentalize both hydrophilic and hydrophobic drugs [20][21][22] . In principle, various types of clinically approved or presently developed therapeutic agents including levodopa are potential drug candidates for liposomal encapsulation 23,24 , as well as contrast agents or radionuclides [25][26][27] for diagnostic and theranostic application. Through the ethanol injection method, followed by extrusion and antibody conjugation, we were able to rapidly and efficiently generate immunoliposomes with multiple types of antibodies. When compared amongst mIgG-, 6E1-, and 214D8-conjugated liposomes, physiological properties did not differ significantly. Previously, antibody conjugation ranged from 6 antibodies per a liposome 28 to approximately 50 antibodies per a liposome 29 . Given the antibody conjugation percentages and the quantity of liposome particles, our optimized method allowed conjugation of approximately 100 antibodies per a liposome.
To test our concept of maintaining functionality of antigen-antibody dependent binding, and maximizing drug load while minimizing structural changes and instability of antibody, we treated A172 cells with DiOC 18 (3)-encapsulated immunoliposomes. Interestingly, we observed antigen-antibody dependent uptake wherein 214D8-and 6E1-conjugated liposomes exhibited 2.21-and 4.25-fold increases of the cellular uptake when compared with the control immunoliposome. We next determined antigen-antibody dependent delivery www.nature.com/scientificreports/ of a therapeutic drug by encapsulating doxorubicin into our immunoliposomes. Doxorubicin was selected since liposomal formulation of this anti-cancer agent (Doxil) has already been well-characterized and FDAapproved [30][31][32] . Remarkably, when A172 cells were treated with 100 µM 214D8-or 6E1-conjugated liposomes, we observed 31% and 29% reduction of relative cell viability. Quantitative assessment of the doxorubicin-encapsulated immunoliposomes demonstrated that approximately 2000 doxorubicin molecules were encapsulated per antibody. Taken together, these findings confirm the potential of our immunoliposome as an antigen-antibody dependent drug delivery modality that conceivably can overcome the limitation imposed by DAR. As one of the potential targets of anti-CD71 antibody, the BBB is a selective barrier between systemic blood circulation and brain parenchyma, and because of the complexity of the central nervous system and the impermeable BBB, delivery of therapeutic drugs across the BBB is a well-documented hurdle [33][34][35] . Despite the notion that CD71 is expressed on the surface of the BBB, only recently, technological tools have been reconceptualized to exploit the receptor-mediated transcytosis with promising potentials 36 including favorable features of anti-CD71 antibody variants with low affinity that allowed the increased release from the BBB into the brain 37,38 . Recently, Johnsen et al. 29 reported that intravenous injection of OX26 (a mouse monoclonal antibody against rat CD71)conjugated and oxaliplatin-loaded liposome resulted in detection of higher concentration of platinum in the rat brain parenchyma compared to the control rat IgG-conjugated liposome. Additionally, a BBB-penetrating fusion protein, JR-141, which is composed of anti-hCD71 antibody and intact enzyme has recently been reported for treatment of lysosomal storage disease 39 . As a new modality, anti-CD71 targeted immunoliposome in this study would become advantageous format with high capacity for encapsulation.
In summary, through our screening strategy, we obtained functional 214D8 antibody that targeted the cell surface molecule CD71 with capacity for cellular internalization. To overcome the limitation of DAR, we generated anti-CD71 antibody-conjugated liposome, evaluated its cellular internalization property, and observed enhanced cellular uptake and cytotoxicity. Taken together, our pipeline with optimized method is a rapid and efficient procedure to isolate potent antibodies with therapeutic potentials as ADCs and to synthesize functional immunoliposomes.
Production of hybridoma library. BALB/c mice (Japan SLC, Inc.; Hamamatsu, Japan) aged 7-10 weeks were immunized weekly or bi-weekly for at least 19 weeks by intra-peritoneal injection of 5 × 10 6 U87 cells. These animals received boost immunization at 3 days before being sacrificed by dislocation or injection of pentobarbital (Kyoritsu Seiyaku Corp.; Tokyo, Japan). Immunized 1 × 10 8 splenocytes and 3 × 10 7 P3U1 myeloma cells were fused with Hybri-Max PEG/DMSO solution (Sigma-Aldrich; St. Louis, MO, USA). Hybridoma library was cultured in RPMI-1640 media (Nacalai Tesque, Inc.; Kyoto, Japan) supplemented with 10% Hyclone Super Low IgG Defined FBS (GE Healthcare Life Sciences; Utah, USA), PSA, sodium pyruvate, MEM non-essential amino acids, 2-mercaptoethanol (Life Technologies Corp.; Grand Island, NY, USA), and hypoxanthine-aminopterinthymidine (HAT) (Life Technologies Corp.; Grand Island, NY, USA), and incubated at 37 °C in the 5% CO 2 incubator for approximately a week for proper HAT selection. The library was screened by using toxin DT3C as described below. For cloning, positive hybridomas from the library were subjected to limiting dilution. Immunoglobulin (Ig) subclass was determined by using IsoStrip mouse monoclonal antibody isotyping kit (Sigma-Aldrich; St. Louis, MO, USA). The animal experimental protocol used in this study was approved and carried out in accordance with the guidelines and regulations from the institutional review board for animal experiment at Juntendo University School of Medicine (#1294), as well as the ARRIVE guidelines.
Immunoprecipitation and immunoblotting.   Active loading of doxorubicin into immunoliposome. Following the conjugation of antibodies to liposomes as described above, the immunoliposomes were ultracentrifuged at 100,000×g for 20 min, and the pelleted immunoliposomes were resuspended with 10 mM histidine (pH 6.5) + 10% sucrose solution. The immunoliposomes were mixed with doxorubicin hydrochloride (Toronto Research Chemicals, Inc.; Toronto, Canada) at the concentration of 0.2 mg doxorubicin/1 mg phospholipid 40 , and incubated at 42 °C overnight. The resulting doxorubicin-encapsulated immunoliposomes were ultracentrifuged at 100,000×g for 20 min three times, and stored in 10 mM histidine (pH 6.5) + 10% sucrose solution at 4 °C until further use. To quantify doxorubicin, the immunoliposomes were treated with 2% Tween-20 solution, and incubated at 60 °C for 30 min. After disruption of liposomal lipid bilayers, absorbance at 495 nm corresponding to doxorubicin was measured.

Analysis of equilibrium dissociation constant.
To determine equilibrium dissociation constant of immunoliposomes, 2 × 10 6 A172 cells were transferred to a tilted 100-mm cell culture dish, and incubated at 37 °C overnight. The seeded dish was placed into LigandTracer (Ridgeview Instruments AB; Vange, Sweden), and baseline, association phase, and dissociation phase were determined for at least 1 h. For association phase, 30 µM, 100 µM, and 300 µM immunoliposomes were applied. Association rate constant (K a ), dissociation rate constant (K d ), and equilibrium dissociation constant (K D ) were analyzed by using TraceDrawer software ver. 1.6 (Ridgeview Instruments AB; Vange, Sweden; https ://www.ligan dtrac er.com/).

Cytotoxicity of doxorubicin-encapsulated immunoliposomes.
For cytotoxicity analysis, 1.0 × 10 4 A172 cells were seeded in the 96-well plates and incubated at 37 °C overnight, followed by the treatment with doxorubicin-encapsulated immunoliposomes at the concentration of 3-100 µM phospholipids at 4 °C for 1 h. The plates were then washed once with culture media, and incubated at 37 °C for 3 days. After 3 days of cultivation, relative cell viability was measured by using WST-1 reagent, and data analysis was performed by using Prism 7.
Statistical analysis. For cytotoxicity analysis of doxorubicin-encapsulated immunoliposomes, two-way ANOVA with Sidak test was performed for multiple group comparisons. P < 0.01 was determined as statistically significant.