TandAbs: potent and well-manufacturable bi-specific antibodies for immunooncology

The initial scientific attraction of biotherapeutics lay in their exquisite selectivity and specificity. The anticipation was that their action should be more predictable and on-target than that afforded by small molecules. While the history of biotherapeutics began with the replacement of endogenous proteins that were either absent or at insufficient concentration to afford normal physiological function, resulting in disease, the most successful biotherapeutics sought to harness the agent of the humoral immune system, the antibody. The current generation of therapeutic antibodies is powerful, but clinical experience indicates that their full potential is yet to be realised. In particular, it was recognised that a bi-specific antibody, functioning either to simultaneously engage two cell-surface signaling molecules to therapeutically modulate aberrant signaling pathways, or to recruit and activate mediators of cellular immunity at the site of a tumour, could have more significant clinical impact. Here we describe one such bi-specific antibody format, the TandAb (tandem diabody), which has reached human clinical trials. The TandAb format provides highly potent and stable antibodies that allow scalable manufacture for clinical trials, and is suited for the commercial development of immune effector cell-engaging antibodies.

TandAb technology

The tetravalent bi-specific TandAb is used to recruit T cells or NK cells via their activating receptors to tumour cells. While regular antibodies cannot engage T cells for tumour cell killing, TandAbs are capable of recruiting either T cells, the most potent immune effector cells, or NK cells. One TandAb domain always remains invariant, anti-CD3 or anti-CD16A respectively, whereas the other domain is selected to target a tumour-specific antigen. A variety of methods were employed to develop the anti-CD3 and anti-CD16A domains with optimal specificity and affinity profiles, reduced immunogenicity, and enhanced stability to maximise their potency and drug-like properties.

A screening scheme, from tumor antigen-specific scFv selection to GMP manufacturing of a clinical lead, is presented in Figure 1. The process consists of several discrete work packages that we describe below.

Selection of target-specific scFv and TandAb formatting – high affinity antibodies

Development of the TandAb tumour-targeting domain begins with the selection of variable fragments (Fvs) that are specific for tumour antigens of interest. This is accomplished by screening single-chain Fv (scFv) phage display libraries, although in some cases the Fvs of antibodies obtained through hybridoma technology are utilised. For this purpose, we generated several large IgM-based phage display libraries of human scFv sequences, each containing more than 10¹⁰ sequentially and structurally diverse scFvs. These libraries are subjected to several rounds of in vitro selection to enrich for binders specific to the desired antigen. The in vitro selection is steered to favour species-cross-reactive binders or affinity for a desired homolog to avoid off-target activity. Enzyme-linked immunosorbent assay (ELISA) screens of soluble scFv identify hits matching the predefined specificity profile. For instance, when developing a domain targeting a specific deletion variant of EGFR with no residual binding to native EGFR, the library was depleted of binders to native EGFR by pre-incubation with EGFR-Fc prior to each panning round. Further selection from the resulting scFv depends on the target biology; it includes the verification of binding to the target protein on the mammalian cell surface and affinity measurements employing surface plasmon resonance (SPR). scFv with affinities of <10 nM can be isolated from the libraries. Should the affinities of the selected antibody candidates be inadequate for the particular application, affinity maturation may be applied; this increases affinities of scFv by 10- to 100-fold depending on the method chosen.

Expression of TandAbs – early assessment of manufacturability

To assess producibility of TandAb leads, and to obtain protein material for biochemical and functional characterisation, stably transfected CHO suspension cells containing a single genomic copy of the TandAb gene are generated using Flp-recombinase-mediated targeted integration. For each new TandAb construct, one standard transfection, combined with stringent antibiotic selection, is sufficient to yield a stable isogenic cell pool expressing the lead. Stable cell pools recover approximately 1-2 weeks post-transfection, and after reaching a viability of >95 % are used for cryopreservation and the inoculation of fed-batch production cultures in shake flasks (Figure 3). We have optimised this process to maximise lead throughput and, to minimise effort and time, in order to obtain high quality protein material for subsequent characterisation. Thus, we have produced more than 500 candidates. TandAb-expressing cell pools exhibit robust performance, with high cell densities (above 10⁷ cells/mL) and viabilities of >80% maintained over a period of 10 days in fed-batch shake flask cultures (Figure 3A). During fed-batch production the secreted TandAbs accumulate at titers between 50 mg/L and 250 mg/L (Figure 3B). The expression titers and the integrity of the TandAbs, which are the main protein products in the cell culture supernatant, are assessed by SDS-PAGE (Figure 3C). This production system satisfies all material requirements from early research to late preclinical stages of development. Note that the suspension CHO expression system is also used for the production of clinical TandAb material at external contract research organisations.

Lab-scale purification of TandAbs – homogeneous preparation

TandAbs are purified from the cleared cell culture supernatant using a two-step purification procedure. The first is affinity chromatography and the second is size exclusion chromatography (SEC). The homodimer content of the TandAb after affinity chromatography is an indicator of good drug-like property and cost-of-goods during product development. It ranges from 10% to over 90%, which makes a second purification step mandatory. Preparative SEC is performed to obtain a TandAb content ranging from ≥90% to ≥95%.

Biophysical characterisation of TandAbs – stable product

During development, several stability and degradation parameters must be identified, characterised, and controlled. Charge- and size-based HPLC, reducing and non-reducing SDS-PAGE, UV-Vis spectroscopy, and differential scanning techniques are the initial characterisation methods. The purification and characterisation scheme depicted (Figure 4) applies to a repertoire of approximately one hundred TandAbs during the selection of a therapeutic lead. Biophysical data generated during this phase are used to rank order among candidates, to identify candidates matching the Target Product Profile (TPP), and to understand the contributions of Fv order, V_H and V_L topography, linker lengths, etc. to the overall stability of a given TandAb. For purification and biophysical characterisation, the TandAb TPP includes: i) a TandAb content of ≥60% after the first capture step, ii) stability at pH 3.5 with soluble product recovery of >90%, of which >95% is the TandAb , iii) stability after three freeze-thaw cycles (from -70°C to RT) with minor changes (≤2%) in TandAb content, iv) endurance of accelerated stability testing at 40°C for up to 7 days, and v) thermal melting above 55°C. At this stage more than 80% of candidates meet the required TPP. Since the biophysical properties of TandAbs are difficult to predict based on the properties of the parental scFvs, such characterisation is required for each lead selection campaign.

Functional characterisation of TandAbs – potent immune effector engagers

After biophysical characterisation, TandAb leads are assayed for affinity and specificity for their target antigens. If recombinant soluble antigens are available, candidates are assayed in ELISA and binding kinetics is characterised by SPR. A more relevant binding characterisation is performed on cell lines and primary cells using flow cytometry. In cell-binding assays, His-tagged TandAbs are incubated in serial dilutions on cells and cell-bound TandAbs are detected by flow cytometry using an anti-His antibody. For comparison and selection of TandAb leads, K_D values are calculated. Control antibodies, such as bi-valent mono-specific IgG or bi-valent bi-specific diabodies, having the same antigen specificity, are used to determine whether TandAbs bind mono- or bi-valently to the target antigens. These flow cytometric cell-binding assays are used for screening panels of 50-100 TandAbs.

TandAb leads meeting the TPP binding criteria are then selected based on potency and efficacy of mediating target cell lysis in in vitro cytotoxicity assays. For assays with short incubation times (up to 6 hours) a calcein-release assay is employed. This assay measures the fluorescent calcein in the cell culture supernatant that is released from labelled target cells upon antibody-mediated lysis by effector cells. If TandAb-mediated target cell lysis requires incubation periods longer than 6 hours, a flow cytometric cytotoxicity assay is used; it employs target cells labeled with a fluorescent membrane dye that allows their discrimination from unlabeled effectors. The viability of target cells is detected using an apoptosis-specific fluorescent dye. Both cytotoxicity assays can be performed in 96 well microplates, and are suitable for selecting the most potent lead from a large panel of candidates.

Since recruitment of immune effectors for the lysis of target cells results in their activation, in vitro safety assessments, for example, induction of cytokine release, proliferation, and activation assays, are performed to evaluate potential differences among the leads in on-target and off-target secondary pharmacodynamic studies. These studies are performed in cultures of peripheral blood cells, in the presence and absence of antigen-positive target cells, for the selection of leads; only those leads that display strict target-dependent activation of effectors are selected.

In vivo pharmacodynamics of (T cell-recruiting) TandAbs – effective tumour elimination

The assessment of the in vivo pharmacodynamic properties of TandAbs requires the transient establishment of a human effector cell population in immunodeficient mice. Here we present the example of a CD19-targeting T cell-recruiting TandAb. Peripheral blood mononuclear cells (PBMC) are isolated from fresh blood donations or buffy coats and purified by Ficoll density gradient centrifugation. NOD/scid mice are either reconstituted with complete PBMC preparations or with negatively enriched T cells. The establishment of tumours from certain cell lines requires purified T cells, since the NK cell activity in some PBMC donor preparations exhibits a strong graft-versus-tumour effect resulting in poor or no tumour take. The pharmacodynamic profile of the CD19/CD3 TandAb was assessed in a Burkitt Lymphoma xenograft model with subcutaneously growing solid tumours in a prophylactic setting. NOD/scid mice were xenotransplanted on day 0 by subcutaneous injection with a suspension of Raji cells pre-mixed with purified human PBMC from healthy donors. Animals were treated starting several hours post tumour-implantation for 5 consecutive days with either vehicle (control) or the TandAb at four different dose levels. Animal groups that received the TandAb exhibited dose-responsive tumour volume reduction relative to the control group. Treatment with the highest TandAb dose (5 mg/kg) completely prevented tumour development in all animals, whereas the lowest dose (0.005 mg/kg) exhibited 60% tumour-growth inhibition.

cGMP manufacturing of TandAbs – scalable process

Sufficient amounts of TandAb material can be produced in cGMP-regulated environments for clinical trials. A manufacturing campaign consists of cell culture and harvest operations, followed by a multi-step purification, formulation, and fill-and-finish process.

Cell line development, leading to a master cell bank (MCB), starts with TandAb DNA transfection of an untagged DNA construct and cell cloning. For sub-cloning and expansion of single-cell clones, selected chemically-defined media can be used. The final cell clone is selected based on growth performance and TandAb production in shake flasks and small-scale bioreactor runs; it is then cryopreserved in a safety cell bank. A cGMP compliant MCB is prepared from the safety cell bank.

The TandAb clinical drug substance manufacturing process is initiated when cells are thawed from a MCB for inoculum expansion. Initial cell expansion is performed in either shake flasks or wave bags. Cell expansion is continued in a production bioreactor, which is run under fed-batch culture conditions.

Similar to lab-scale expression, process monitoring includes parameters, such as cell density, cell viability and product concentration. Pre-harvest samples are tested for microbial contamination, mycoplasma, and adventitious virus in vitro. TandAb-containing cell culture is harvested by dead-end filtration, where depth filters and membrane filters are employed to efficiently remove cells and cell debris, providing cell-free supernatant for further purification of the product. The TandAb is purified via a chromatographic process, as described above, capable of efficiently removing process- and product-related impurities. In addition, virus removal/inactivation steps are included to ensure clearance of potential viral contaminants.

Based on this robust process, TandAb material of high quality is produced. The quality is assayed employing the analytical methods that were developed during the setup of the process-flow leading to the specification definition and the identification of release acceptance criteria. The analytical procedures are implemented for process development, in-process controls and quality control release testing, stability testing, and product characterisation. They encompass generic, compendial, and product-specific methods, and are established to assess the appearance and description, general tests, identity, heterogeneity, purity, impurities, binding activity/potency, quantity and microbial tests. Reference standard material is produced and characterised. The analytical methods are qualified and validated according to the requirements of the clinical development phase.

Formulation development is performed by screening for excipients and conditions which are favourable for the stability of the TandAb protein. The drug substance and drug product are formulated in the identified formulation buffer. Since lyophilisation is an option for long-term storage during early clinical development, conditions are also identified to support this formulation. During process development, product quality is evaluated prior- and post-lyophilisation to evaluate process impact. The test methods used during development are stability-indicating assays such as measurement of opalescence, measurement of sub-visible particle content, high pressure size exclusion, and the ion exchange chromatography profile.

The aseptic filling and lyophilisation operations are carried out in the pharmaceutical filling area where the TandAb drug product is manufactured. Analytical procedures identical to those used for drug substance lead to release, with the following additional tests performed for the drug product: appearance of the lyophilisate, reconstitution time, uniformity of content and residual moisture of the lyophilised powder, appearance – visible particles, particulate contamination – sub-visible particles, and sterility of the reconstituted solution.

Accelerated and long-term storage stability assessments are performed for both the drug substance and the drug product. The data sets show excellent stability for the TandAb material spanning three years (Figure 5).

Conclusion

In conclusion, the TandAb platform enables the production of bi-specific effector-recruiting antibodies with excellent drug-like properties including long-term storage, high potency and efficacy, and an advantageous PK profile permitting bolus infusion administration. The process for selecting TandAbs against specific target antigens having the desired mechanism-of-action (T or NK cell recruitment) is robust and has been successfully applied to several programs currently at various clinical investigation stages.

References

1. Kipriyanov SM, Moldenhauer G, Schuhmacher J et al. Bispecific tandem diabody for tumor therapy with improved antigen binding and pharmacokinetics. J. Mol. Biol. 293(1), 41–56 (1999).
2. Kipriyanov SM. Generation of bispecific and tandem diabodies. Methods Mol. Biol. 562, 177–193 (2009).
3. Kipriyanov SM, Moldenhauer G, Braunagel M et al. Effect of domain order on the activity of bacterially produced bispecific single-chain Fv antibodies. J. Mol. Biol. 330(1), 99–111 (2003).
4. Le Gall F, Reusch U, Little M, Kipriyanov SM. Effect of linker sequences between the antibody variable domains on the formation, stability and biological activity of a bispecific tandem diabody. Protein Eng. Des. Sel. 17(4), 357–366 (2004).

Biographies

Dr. Michael Weichel joined Affimed in 2014 as Head of Protein Chemistry at Affimed Therapeutics AG and has broad experience in the purification and physicochemical characterisation of monoclonal antibodies as well as alternative binding protein scaffolds. Dr. Weichel studied Chemistry at the Technical University in Darmstadt, Germany, and holds a PhD degree in Biochemistry from the University of Zurich, Switzerland. Dr. Weichel started his industrial career at Boehringer Ingelheim Pharma GmbH & Co. KG. From 2008 to 2013 he headed the Antibody Purification and Characterization Group at Ganymed Pharmaceuticals AG where he contributed to DSP and formulation development of two monoclonal antibody therapeutics currently in Phase I and II clinical trials.

Dr. Kristina Ellwanger joined Affimed in 2011 as Head of the Protein Expression and Cell Engineering Group. She received her Diploma in Biology at the University of Stuttgart in 2004 and completed her PhD degree at the German Cancer Research Center (DKFZ) and the University of Heidelberg in 2008. She started her industrial career at Boehringer Ingelheim Pharma, as a Senior Scientist where she was responsible for the development of innovative approaches in molecular expression and cell line engineering to improve the quality and productivity of recombinant proteins in mammalian cells.

Ivica Fucek (Dipl.-Ing.) is Head of the Lead Generation Group at Affimed Therapeutics. He joined Affimed in January 2007 as a Research Scientist. In 2013, he was promoted to Head of Lead Generation. He has broad experience in molecular biology and the SPR technique. He received his Dipl.-Ing. degree in biotechnology at the University of Applied Science in Darmstadt. His diploma thesis on the identification and characterisation of tuberculosis-associated antigens was performed at Ganymed Pharmaceuticals.

Dr. Stefan Knackmuss is Head of Preclinical Development at Affimed Therapeutics. He received his PhD from the Ruprecht-Karls-University where he graduated in Biochemistry and Molecular Biology. Dr. Knackmuss joined Affimed in 2000 as a Research Scientist responsible for establishing and screening recombinant antibody libraries. He was involved in the isolation and in vitro/in vivo characterisation of a wide range of antibodies directed against various targets in oncology and inflammation. In 2008, he was promoted to Head of Preclinical Development with responsibility for the IND/IMPD enabling preclinical programs.

Dr. Erich Rajkovic is Head of Business Development and Alliance Management at Affimed Therapeutics. He joined Affimed in 2007 as Scientist in antibody discovery and engineering. In 2010, he joined the Business Development team and was promoted to Director of Business Development in 2011. Since 2013 he has been responsible for Business Development & Alliance Management. Prior to Affimed Dr. Rajkovic worked for Kwizda Pharma (Austria) and did a postdoctoral fellowship at the University of Graz. He received the Ph.D. in Protein Chemistry and Biophysics from the University of Graz in 2006. In 2014 he completed an MBA at the SRH Fernhochschule Riedlingen.

Dr. Uwe Reusch is Head of Biassay Group at Affimed Therapeutics. He graduated in Biology at the Eberhard-Karls University Tübingen and received his PhD from the Ludwig-Maximilians University Munich. In 2001, he joined the start-up company Affimed and established several in vitro assays for testing recombinant antibodies.

Dr. Claudia Wall is Head of Project Management, Regulatory Affairs and Quality Management. She joined Affimed in 2002 as Research Scientist. In 2008 she was promoted to Head of Project Management, Regulatory Affairs and Quality Management. In this role, she was responsible for the successful establishment of the cGMP- compliant production processes of both lead projects AFM13 (CD30/CD16A TandAb) and AFM11 (CD19/CD3 TandAb). Prior to joining Affimed, from 1997 to 2001, she was a Scientific Associate at Hoffmann-LaRoche AG. She received her PhD in Pathobiochemistry and General Neurochemistry from the University of Heidelberg.

Dr. Eugene Zhukovsky joined Affimed in 2011 as Chief Scientific Officer. He has over 20 years experience in the field of biotherapeutics research and development. Prior to Affimed, he was a Senior Research Fellow at Boehringer Ingelheim Pharmaceuticals Inc. where he led antibody discovery efforts. Prior to that he was an Associate Director at Xencor Inc. where he led translational research efforts. Prior to that he developed genomics technologies at Lynx Therapeutics and utilized phage display technology for development of catalytic antibodies at Neurex Corporation. Dr. Zhukovsky performed a postdoctoral fellowship at Genentech; he received the PhD in Biochemistry from Brandeis University.