Here we present a protocol to study the in vivo localization of antibodies in mice tumor xenograft models.
Monoclonal antibodies are high affinity multifunctional drugs that work by variable independent mechanisms to eliminate cancer cells. Over the last few decades, the field of antibody-drug conjugates, bispecific antibodies, chimeric antigen receptors (CAR) and cancer immunotherapy has emerged as the most promising area of basic and therapeutic investigations. With numerous successful human trials targeting immune checkpoint receptors and CAR-T cells in leukemia and melanoma at a breakthrough pace, it is highly exciting times for oncologic therapeutics derived from variations of antibody engineering. Regrettably, a significantly large numbers of antibody and CAR based therapeutics have also proven disappointing in human trials of solid cancers because of the limited availability of immune effector cells in the tumor bed. Importantly, nonspecific distribution of therapeutic antibodies in tissues other than tumors also contribute to the lack of clinical efficacy, associated toxicity and clinical failure. As faithful translation of preclinical studies into human clinical trails are highly relied on mice tumor xenograft efficacy and safety studies, here we highlight a method to test the tumor and general tissue distribution of therapeutic antibodies. This is achieved by labeling the protein-A purified antibody with near Infrared fluorescent dye followed by live imaging of tumor bearing mice.
FDA approved the first monoclonal antibody targeting CD3 (OKT3, Muromonab) in 19861,2. Since then for the next twenty years, there has been a rapid explosion in the field of antibody engineering due to the overwhelming success of antibodies against immune checkpoint inhibitors3. Beside indirect activation of immune system, antibodies are being aimed to directly flag cancer cells to precisely engage immune effector cells, trigger cytotoxicity via death receptor agonist, block tumor cell survival signaling, obstruct angiogenesis (growth of blood vessels), constrain immune checkpoint regulators, deliver radioisotopes, chemotherapy drugs and siRNA as a conjugated agents2. In addition, studying the single chain variable fragments (scFv) of various antibodies on the surface of patient derived T-cells and NK cells (CAR-T and CAR-NK) is a fast growing area of clinical investigations for cell-based therapies4.
The ultra-high affinity of antibody-based drugs that provides selectivity to antigen expressing tumor cells makes it an attractive agent. Likewise, the targeted delivery and tumor retention of a therapeutic antibody (or a chemical drug) is the key to balance efficacy over toxicity. Therefore, a large number of protein engineering based strategies that include but are not limited to bispecific5 and tri-specific antibodies6 are being exploited to significantly enhance avidity optimized tumor retention of intravenously (IV) injected therapeutics5,7. Here, we describe a simple fluorescence-based method to address the tumor and tissue distribution of potentially effective anti-cancer antibodies.
Because animal tissues possess auto-fluorescence when excited in visible spectrum, the antibodies were initially labeled with near Infrared dye (e.g., IRDye 800CW). For proofs of concept investigations, we have made use of folate receptor alpha-1 (FOLR1) targeting antibody called farletuzumab and its derivative called Bispecific anchor Cytotoxicity activator (BaCa)7 antibody that co-targets FOLR1 and death receptor-5 (DR5)8 in one recombinant antibody. FOLR1 is a well-defined overexpressed target receptor in ovarian and TNBC cancer cells, tumor xenografts and patient tumors9. Notably, there are multiple efforts to clinically exploit FOLR1 using antibody-based approaches to engage immune effector cells and antibody drug conjugates (ADC) for ovarian and breast cancers10,11.
In this methods paper, we cloned, expressed and purified clinical anti-FOLR1 (farletuzumab) along with other control antibodies using CHO expression system. IgG1 isotype and a clinical anti-idiotype mucin-16 antibody called abagovomab12 were used as negative controls. Following protein-A purification, indicated antibodies were labeled with IRDye 800CW and were administered into the tail vein of nude mice either bearing ovarian tumor xenografts or stably transfected human FOLR1 expressing murine colon cancer xenografts. The antibody localization was tracked by live imaging using in vivo imaging spectrum at multiple different time points7. This method does not require any genetic modification or injection of the substrate to enable light emission and is significantly quicker, cost effective and efficient. The general cloning, expression, purification and labeling protocol described below can be applied to any clinical and nonclinical antibody if heavy and light chain sequences are available.
All the procedures involving animals handling and tumor xenografts studies were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) here at the University of Virginia and conform to the relevant regulatory standards
1. Expression and purification of antibodies
2. Fluorescent labeling
NOTE: Antibodies are labeled with the infrared dye that contains an NHS ester reactive group, which couples to proteins and form a stable conjugate. This reaction is pH sensitive and works best at pH 8.5. Fluorescent conjugates labeled with the dye display an absorption maximum of 774 nm, and an emission maximum of 789 nm. pH 8.5 is key for effective conjugation.
3. Mouse xenograft studies
4. Antibody localization using an in vivo imaging system
NOTE: In vivo imaging equipment (see Table of Materials) used in this experiment uses a set of high efficiency filters and spectral un-mixing algorithms for noninvasive visualization and tracking of cellular and genetic activity within a living organism in real time. System provides both fluorescence and bioluminescence monitoring capability.
In the described methodology, first we cloned antibodies targeting folate receptor alpha-1 (FOLR1) named farletuzumab, and a bispecific antibody called BaCa consisting of farletuzumab and lexatumumab along with control antibodies such as abagovomab (sequences provided in Supplementary File 1). Details of representative variable heavy (VH) and variable light (VL) domains in DNA clones (pVH, pVL) are shown in Figure 1A. To confirm the positive clones, we carried out colony PCR using signal peptide forward (SP For) and CK Rev/CH3 Rev primers (sequences provided in Supplementary File 1). Representative results of colony PCR confirm the expected sizes of light and heavy chains (Figure 1C). Positive antibody clones were, also, confirmed using Sanger sequencing. Following confirmed pVH and pVL DNA cloning, transfections were carried out using CHO suspension cultures, followed by protein-A column affinity purifications of antibodies at 4 °C (see Figure 1B and protocol for detail steps).
Representative results of purified farletuzumab along with other control IgG1 (except BaCa antibody run on non-reducing and reducing SDS PAGE are shown in Figure 2A. As evident, heavy and light chain produced 50 and 25 KDa bands after reduction. This was followed by binding confirmation of antibodies to native proteins on the cell surface. Representative farletuzumab binding to human FOLR1 on OVCAR3 cells surface is shown using flow cytometry (Figure 2B).
Next fluorescently labeled antibodies were tail vein injected into animals grafted with FOLR1 expressing tumors (Figure 3). Animals were live imaged at multiple time points using IVIS. The data confirms selective enrichment of FOLR1 and BaCa antibodies into the FOLR1+ tumors (Figure 4 and Figure 5). Importantly control antibodies (negative for tumor antigen) did not localize into tumors.
Figure 1: Schematic of antibody cloning, expression and purification.
(A) Shows the detail schematic of heavy (pVH) and light (pVL) chain vectors. (B) pVH vector carrying variable domain sequence of an IgG1 heavy chain and pVL vector carrying variable domain sequence of a light chain were mixed together (1:2 ratio) along with transfection reagents (such as mirus or PEI) before adding to the suspension culture of CHO or HEK cells. Cells were fed with supplemental feed next day and cultures were monitored for 10 additional days with intermittent feeding. At day 11, cells were harvested through 0.2 µm PES filters followed by affinity chromatography using protein-A. Purified antibodies were next analyzed for % monomer (FPLC), binding activity (ELISA or SPR), and binding to the target antigen (FACS) on live cells. Antibodies can also be checked for in vivo assays (e.g., cell growth inhibition, cell viability assays, or inhibition of signaling intermediate phosphorylation etc). Antibodies can also be conjugated with far-red dyes e.g., IRDye 800CW for in vivo imaging. ± IRDye 800CW must be tested for activities prior to tumor and tissue distribution studies. (C) Agarose gels of colony PCR confirming the sizes of positive heavy (1.4 Kb) and light chain (0.8 Kb) clones. Please click here to view a larger version of this figure.
Figure 2: A gel-based reduction assay to confirm antibody integrity.
(A) Four different IgG1 antibodies (schematic shown on top) were added ± reducing agent (such as BME or DTT) at 95°C for 10 min. Antibodies were next loaded on a 10% SDS-PAGE gel followed by protein staining and imaging. Gel image in left and right clearly show the intact antibody and two separate polypeptides (~50 KDa and ~25 KDa) respectively. Please also see pVH and pVL vector maps (Figure 1) carrying cDNA corresponding to VH/VL, CH1/CK, CH2 and CH3 domains. (B) Flow cytometry confirmation of unlabeled and IRDye 800CW labeled farletuzumab binding to native FOLR1 on ovarian cancer cells. Non-reducing = Antibody run on gel with non-reducing dye, Reducing = Antibody run on gel with reducing dye, HC = Heavy chain, LC = Light chain, VL = Variable domain of light chain, VH = Variable domain of heavy chain, CK = Kappa chain Please click here to view a larger version of this figure.
Figure 3: Experimental schematic of tumor generation and antibody treatments.
6-8 weeks old mice strains such as: Immunodeficient athymic nude/NSG/ Immunocompetent C57BL/6 or Balb/C mice could be easily grafted with tumor cells via subcutaneous (SQ) tumors. Similar studies could be carried out using breast fat-pad and intraperitoneal (IP) tumors. 3-4 weeks later (tumor ~200 mm3), mice were IV injected with an IRDye labeled indicated antibodies that were ± selective against tumor-overexpressed receptor. This was followed by live in vivo imaging. Please click here to view a larger version of this figure.
Figure 4: Live in-vivo imaging of human OVCAR3 tumor bearing mice.
Randomly selected 6 to 8 weeks old (Age) and 20-25 gram (Weight) hairless athymic Nude Foxn1nu/Foxn1+ (Envigo) were grafted with FOLR1+ ovarian tumors (OVCAR-3 cells). After 3 weeks with evident tumors, mice were tail vein injected with IRDye 800CW labeled IgG1 control, abagovomab (CA-125 anti-idiotypic antibody), farletuzumab (anti-FOLR1 antibody) and BaCa (anti-FOLR1-DR5 antibody) followed by live imaging at indicated times. Please click here to view a larger version of this figure.
Figure 5: Live in-vivo imaging of human FOLR1 expressing murine MC38 cell derived tumor bearing mice.
(A) Randomly selected 6 to 8 weeks old (Age) and 20-25 g NOD.Cg Prkdcscid Il2rgtm1Wjl/SzJ or nude mice were SQ injected with murine MC38 cells stably expressing human FOLR1. Upon tumor appearance, mice were tail vein injected with IRDye 800CW labeled IgG1 control, abagovomab (CA-125 anti-idiotypic antibody), farletuzumab (anti-FOLR1 antibody) and BaCa (anti-FOLR1-DR5 antibody) followed by live imaging at indicated times. (B) After 7 days animals were euthanized and isolated key organs (as indicated) were imaged together along with grafted tumors for relative antibody signal (IRDye 800CW) distribution. As expected, tumors remained negative with IRDye 800CW signal in IgG1 control and CA-125 anti-idiotypic antibody, abagovomab injected animals. Please click here to view a larger version of this figure.
Supplementary File 1: Sequences of all antibodies and primers. Please click here to download this file.
Selective and tumor tissue specific delivery of anti-cancer therapeutic agent is the key to measure efficacy and safety of a given targeted therapy13. Here we have described a quick and efficient approach to investigate the detailed tissue and tumor distribution of clinical, farletuzumab and a nonclinical BaCa antibody. The described approach is applicable to any newly generated antibody and can be used alongside of a clinically effective antibody (with desired qualities) for its tumor/organ distribution properties. Considering most antibody target receptors (such as HER2 in breast cancer) are highly overexpressed in tumor cells (tissues), in most cases their suboptimal expression and function is also critical in cell types other than tumor cells14. For example, a significant proportion of EGFR targeting clinical antibodies in colon cancer patients accumulate and cause toxicity to skin tissue15, a noncancerous tissue whose growth and differentiation requires EGFR signaling and function. Therefore, we strongly believe these sorts of preliminary tissue distribution investigations in combinations with hepatotoxicity and tissue histochemistry assays in a larger cohort of animals are key to comprehensively assess safety and therapeutic viability of the newly generated antibody. Moreover, described tissue distribution studies would also be highly applicable in immune competent mice xenograft studies, if the newly generated antibody maintains cross-reactivity to murine counterpart antigen/receptor. In syngeneic animal studies, along with tumor distribution and tissue histochemistry studies, detailed blood cytokine analysis can used to strengthen the efficacy and safety data. An attractive feature of the described approach is that it allows near accurate quantitation of antibody distribution if data is additionally supported with ELISA (against target antigen) from the tumor and other significant tissue lysates (such as liver, heart, lung, spleen, kidney etc)7. Another important feature of described method over single photon emission computed tomography (called SPECT) and position emission tomography (PET) is the cost-effectiveness16. Both SPECT and PET are very expensive and makes use of radioactive tracers for imaging, making the whole process cumbersome if testing a large cohort of animals17. In addition SPECT and PET imaging facilities are not very standard in laboratories and vivariums to study small animal models of diseases such as mice18.
One limitation with described method to achieve a near accurate quantitation of antibody tumor distribution is the dependence on high affinity target antigen binding. It is because high affinity antigen-antibody interactions may result in “target-mediated drug disposition (TMDD)” by enhanced endocytosis and shuttling to lysosomes19. Therefore, the results of the described approach will vary depending on the particular target receptor in a particular tumor type. Thus, we strongly recommend testing labeled antibody/antibodies in a large cohort of animals with tumor xenografts generated with more than one tumor cell line(s) having variable (heterogeneous) expression of target antigen receptor. It is also greatly recommended to make use of more than one fluorescent conjugate dye(s) and mice strain(s) for the proposed studies.
Considering that small size antibodies such as Fabs, scFvs, BiTes, DARTs, etc. (lacking salvage recycling by neonatal Fc receptor (FcRn) clear more rapidly from tumors (with serum half times being minutes to hours), care should be taken to compare data between different tumor types having highly variable FcRn expression. Furthermore, larger molecules (such as dual and trispecificity antibodies) that are engineered with Fc domain for salvage recycling have tissue/tumor penetration issues. In those scenarios, the described approach will not be suitable to compare tissue/tumor distribution of antibodies that differ significantly in sizes6. In terms of significance however, the described tumor and detailed tissue distribution studies along with their counterpart monospecific antibodies would serve as a key factor in an effective dual and trispecificity antibody platform design. Finally, since higher affinity and avidity-optimized antibodies generally have a significantly homogenous distribution in tumors, the target tumor epitope selection (lacking TMDD), overall antibody affinity and biological activity in a particular cancer model should always be considered before making conclusion of tumor penetration, safety and efficacy.
In summary, we have described a quick and simple method for monitoring the tumor and tissue distribution of intravenously injected antibodies. The described approach has added potential to analyze antibody-siRNA conjugates (where siRNA is labeled), antibody-drug conjugates (where drug is labeled) and antibody-nanoparticles (where nanoparticle lipids are labeled with fluorescent dye). Likewise, a uniquely engineered cysteine residue in a tumor targeting scFv (if fluorescently labeled with melamide chemistry) of chimeric antigen receptor T-cells (CAR-T) and CAR-NK will be a cost effective approach to analyze tumor/tissue distribution of these cell based therapies independent of viral transfection based GFP/RFP signals strategies.
The authors have nothing to disclose.
We are thankful to University of Virginia Cancer Center Core Imaging Facility, Biomolecular Analysis Facility, Advanced Microscopy Facility and the Core Vivarium Facility for Assistance. J. T-S is an early career investigator of Ovarian Cancer Academy (OCA-DoD). This work was supported by NCI/NIH grant (R01CA233752) to J. T-S, U.S. DoD Breast Cancer Research Program (BCRP) breakthrough level-1 award to J. T-S (BC17097) and U.S. DoD Ovarian Cancer Research Program (OCRP) funding award (OC180412) to J. T-S
FreeStyle CHO media | Gibco Life Technologies | Cat # 12651-014 | |
Anti-Anti (100X) | Gibco Life Technologies | Cat # 15240-062 | |
Anti-Clumping Agent | Gibco Life Technologies | Cat # 01-0057DG | |
BD Insulin Syringe | BD BioSciences | Cat #329420 | |
Caliper IVIS Spectrum | PerkinElmer | Cat #124262 | |
CHO CD EfficientFeed B | Gibco Life Technologies | Cat #A10240-01 | |
Corning 500 mL DMEM (Dulbecco's Modified Eagle's Medium) | Corning | Cat # 10-13-CV | |
Corning 500 mL RPMI 1640 | Corning | Cat # 10-040-CV | |
Cy5 conjugated Anti-Human IgG (H+L) | Jackson ImmunoResearch | Cat # 709-175-149 | |
GlutaMax-I (100X) | Gibco Life Technologies | Cat # 35050-061 | |
HiPure Plasmid Maxiprep kit | Invitrogen | Cat # K21007 | |
HiTrap MabSelect SuRe Column | GE Healthcare | Cat # 11-0034-93 | |
Infusion | Takara BioScience | STO344 | |
IRDye 800CW NHS Ester | LI-COR | Cat # 929-70020 | |
Isoflurane, USP | Covetrus | Cat # 11695-6777-2 | |
Lubricant Eye Ointment | Refresh Lacri-Lube | Cat #4089 | |
Matrigel | Corning | Cat # 354234 | |
PEI transfection reagent | Thermo Fisher | Cat # BMS1003A | |
Slide-A-Lyzer Dialysis Cassettes | Thermo Scientific | Cat # 66333 | |
Steritop Vacuum Filters | Millipore Express | Cat #S2GPT02RE | |
Trypsin-EDTA | Gibco Life Technologies | Cat # 15400-054 | |
Experimental Models: Cell lines | |||
Human: OVCAR-3 | American Type Culture Collection | ATCC HTB-161 | |
Human: CHO-K cells | Stable transformed in our lab | ATCC CCL-61 | |
Mouse: 4T1 | Kind gift from Dr. Chip Landen, UVA | ||
Mouse: MC38 | Kind gift from Dr. Suzanne Ostrand-Rosenberg, UMBC | Authenticated by STR profiling | |
Mouse: MC38 hFOLR1 | Generated in our laboratory (This paper) | ||
Experimental Models: Animal | |||
Mice: athymic Nude Foxn1nu/Foxn1+ | Envigo | Multiple Orders | |
Mice: NOD.Cg Prkdcscid Il2rgtm1Wjl/SzJ | Jackson Laboratory | Multiple Orders |