Tribodies: Fab-scFv fusion proteins as a platform to create multi-functional pharmaceuticals
Nico Mertens
Biotecnol SA, Lagoas Park, Edificio 7, 2741-901 Oeiras (Lisbon), Portugal/ Email: nm@biotecnol.com Abstract Tribodies are multifunctional recombinant antibody derivatives, which utilize the natural in vivo heterodimerization of the heavy chain (Fd fragment) and light chain (L) of a Fab fragment, to form a scaffold, upon which additional functions can be incorporated, such as additional binders - e.g. scFv binding domains. Each chain can be extended preferably at the C-terminus with an additional scFv binder. The chains are co-produced in mammalian cells, where the host-cell BiP chaperone drives the formation of the heavy chain-light chain heterodimer (Fd:L) - this reaction does not appear to be inhibited by the chain extensions, and leads to a very specific heterodimerization, using molecules abundantly present in serum (non-immunogenic) These heterodimers are stable, with each of the binders retaining their specific affinities, with the bivalent tribody having higher affinity, and higher activatation of T-cell proliferation and cytotoxicity in vivo. This design allows easy engineering of multispecificity in a single molecule, e.g. bispecific antibodies bivalent for the target and monovalent for effector activation (e.g. for T-cell activation), or trispecific antibodies pur sang.
1 Introduction ............................................................................................................................................................................................... 2 2 Tribody Structure, Production and Stability ............................................................................................................................................. 4 2.1 What is a Tribody? ........................................................................................................................................................................... 4 2.2 Tribody Expression .......................................................................................................................................................................... 6 2.3 Purification and Stability .................................................................................................................................................................. 8 3 Tribody Activity ......................................................................................................................................................................................... 9 3.1 in vitro binding and T-cell activation ................................................................................................................................................ 9 3.2 In vivo model: mouse BCL1 lymphoma......................................................................................................................................... 10 3.3 PK measurements ......................................................................................................................................................................... 11 4 Tribody Use and Potential...................................................................................................................................................................... 12 5 Conclusion .............................................................................................................................................................................................. 13
N. Mertens, Tribodies as a platform for multi-specific pharmaceuticals In “Bispecific Antibodies” (ed. R. Kontermann), in publication
1
1 Introduction
Our proper immune system is a powerful but highly complex regulated system. Powerful since it is capable of specifically repel bulky cell masses (think of transplantation) and defend us quiet successfully against a pleitropy of invaders , and this with minimal damage to healthy tissue. This is due to a high specificity and self regulation build into a complex system, i.e. a system with a multitude of redundancies and feed-back loops. Antibodies are one way in which the system incorporates specificity. The main function of antibodies is to opsonize bacteria and deliver them to neutrophils and macrophages. They also have a natural function of clearing toxic compounds or circulating viruses out of the blood and target them for clearance. Cellular malignancies such as viral infected cells are more effectively halted by the cellular arm of the system, mainly by the action of T-cells and NK-cells. Only the latter carry a receptor for opsonizing antibodies. Due to the high specificity and high affinity we can find in antibodies and the ease of developing recombinant proteins and adapt them to our needs antibodies became a first means for specific immunotherapy, with a special interest in trying to specifically target cancer cells. Through their Fc part antibodies are able to induce antibody-induced cell death (ADCC) and complement depended cytotoxicity (CDC). Target-cell bound antibodies are recognized through Fcreceptors on effector cells, such as the high affinity Fc-receptor Fc RIII on macrophages and activated neutrophils, but also to inhibiting receptors such as Fc RIIb, and on Fc RIIa complexes on non-cytotoxic cells such as platelets and B-cells. However, experience with antitumor antibodies has shown that effective antibodies also have an influence on the target cell by the nature of the receptor (or even epitope) they target. Ways this can be done is by growth inhibition (e.g. by down-regulation of growth factor receptors), rendering the cancer cells more susceptible to apoptosis, or even induce cell lysis. Receptors without such a function have been targeted with antibodies with much less clinical anti-tumor effectiveness. This created a possibility to develop molecules with a comparable affinity and specificity for these receptors as antibodies can develop, but with a different effector mechanism. Effective routes under investigation are to couple toxic payloads or functions to the targeting molecules (such as the antibodies), that either need to be internalized or stay at the cell surface. The toxins can be of variable nature, such as protein toxins with enzymatic activity, chemical drugs or radionucleotides. Although such fusions are effective one problem is to target enough into the tumor. The percentage of injected dose (%ID) that actually reaches the tumor is often very low in human patients (can be in the order of less than 1%) (Steffens, Boerman et al. 1997). As long as these “armed” antibodies or antibody-derivatives are not bound to the tumor or excreted they remain circulating and can cause toxic effects elsewhere. Since monoclonal antibodies have a special recue mechanism by the FcRn binding in liver cell endosomes, they can have half lives of up to 3 weeks (Borvak, Richardson et al. 1998). This eventually allows good tumor targeting but results in a bad ratio of targeted versus non-targeted (circulating or deposited) antibody. So there is a clear need to develop targeting reagents which have a good tumor binding but are cleared more rapidly from the rest of the body. Smaller fragments still containing the antigen binding properties of the antibody have been used as carrying vehicles of toxic functions.
N. Mertens, Tribodies as a platform for multi-specific pharmaceuticals In “Bispecific Antibodies” (ed. R. Kontermann), in publication
2
However, when these proteins are below 60-80 kDa in size they will be cleared rapidly by the kidneys. Kidneys work as an ultrafiltration device with a 6-10 nm cut-off value. As a comparison: a Fab molecules is about 8 x 5 nm, so it can pass readily through the pores. With a glomerular filtration rate in humans of about 7 liters/hour, a short half life can be predicted for these small fragments. This is indeed experimentally noticed using a variety of small antibody fragments. In most cases, half lives were determined to be less than 1 hour, with disappointing tumor accumulation as a result (Smith, Popplewell et al. 2001). An intermediate sized molecule should avoid kidney clearance and ideally provide a half life sufficient for improved tumor accumulation while also avoiding a body detainment which would hamper targeting toxic function to the tumor. One way of combating tumor cells different from the normal monoclonal antibody functions is by using antibody functions to cross-link cytotoxic T-cells with tumor cells. This is done using bispecific antibodies (BsAb) targeting both the tumor and the T-cell Receptor (TCR). Such a strategy involves T-cells in the induced immunotherapy. T-cells are naturally involved in combating cellular targets but are left inactive when invocing a cure with monoclonal antibodies (moAb). T-cell engaging bispecific antibodies are thus a complementary approach to moAb therapy. In this strategy it is important to have a monovalent binding towards the T-cell since bivalent binding will lead to initial T-cell activation but then induces T-cells anergy. This is why Orthoclone (OKT3 moAb anti-TCR) or its (Fab)2 fragments is used with the aim to tolerize tissue transplants (Todd and Brogden 1989). On the other hand a bivalent binding towards the tumor cell can be more advantageous since the avidity will increase the functional binding and targeting towards the tumor. We have created a manifold that combines two scFv fragments with a Fab fragment. The Fab fragment serves as a specific heterodimerization signal, and the two scFv fragments are each fused to a different Fab chain. In this way we obtain a molecule of intermediate molecular weight (100 kDa) which allows incorporating three different antibody fragments (Schoonjans, Willems et al. 2000). This manifold, which we baptized “tribody”, can be used to create trivalent molecules as well as bispecific molecules with bivalent binding to only one target, as trispecific molecules.
N. Mertens, Tribodies as a platform for multi-specific pharmaceuticals In “Bispecific Antibodies” (ed. R. Kontermann), in publication
3
2 Tribody Structure, Production and Stability
2.1 What is a Tribody?
Tribodies are multifunctional recombinant antibody derivatives, which utilise the natural in vivo heterodimerization of the heavy chain (Fd fragment) and light chain (L) of a Fab fragment, to form a scaffold, upon which additional functions can be incorporated, such as additional binders - e.g. scFv binding domains. Each chain can be extended preferably at the C-terminus with an additional scFv binder. This leads to a very specific heterodimerization, using molecules abundantly present in serum (and hence non-immunogenic and non-antigenic). The only non-natural sequences in the molecule are the peptide linkers connecting the VH and VL domains in the scFv moiety and the linkers connecting the Fab chains and the scFv (Schoonjans, Willems et al. 2001) (Schoonjans, Willems et al. 2000). These sequences can however be chosen to resemble natural human linker sequences or be composed of glycin-polymers which are considered to be neither antigenic (recognized by antibodies) nor immunogenic (recognized by T-cells when presented by antigen presenting cells), probably due to the lack of amino acid side chains. This design allows easy engineering of multispecificity in a single molecule. The Fab chains (Fd and L chain) can be extended at both the N- and C-terminal side. However N-terminal extension at the Fab chains often leads to a hampered binding function of the Fab fragment (unpublished observations), so our preferred model will have the extensions made at the C-terminal side of the Fd and L-chain (Figure 1).
A
VL
Fab L CL L VH CH1 L VH VH
scFv 1 VL
B
VL1 VH1 CL S S CH1 VL VL3 VH3 VH2 VL2
E
VH2 VL2 VL1 CL S S VH1 CH1 CL VL1 S S VH1 VL1 CH1 CL S S VH1 CH1
F
B
A C A A A A A B
Trispecific
VH1
Fab Fd
scFv 2
VL1 CL S S
VH1 CH1
VL1 CL S S
VH1 CH1
VL1
CL S S CH1
C scFv 2 VH3 VL3
VL1
VH1 Fab
D
VL1 VH1 VH2 VL2
VH2 VL2
VL3 VH3
VH2 VL2
VL1 CL S S
VH1 CH1
CL
CH1
IL2
VH2 VL2
Bispecific
VH2 VL2
scFv 1
Trivalent
Figure 1: A) Tribodies are genetically constructed by fusing entities (i.e. to the Fd and L chain of a Fab molecule. As a Fab is a natural disulphide stabilized heterodimer, the position of each function is exactly determined and a single homogeneous product is produced. C) Models of the molecule predict no sterical hindrance between the subunits. D) Even as the scFv are N-terminally linked, their binding site is facing outside. E) Tribodies can lead to 3x1 (trispecific), 2+1 (bispecific) or trivalent molecules, as well as immunotoxins, immunocytokines, enzyme fusions,...F) The model easily allows trispecific, bispecific and trivalent antibody derivatives to be produced.
N. Mertens, Tribodies as a platform for multi-specific pharmaceuticals In “Bispecific Antibodies” (ed. R. Kontermann), in publication
4
anti-TAA
MW
A scFv-TAA VL-TAA VH-TAA CL CH1 scFv-CD3 scFv-CD3
Bs-scFv
B
Fab-(scFv)2
Fab-scFv
C
194 120 87 64 52 39
100 90 80 70 60 50 40 30 20
Tribody in PBSA Bibody in PBSA BssFv in PBSA Tribody in serum Bibody in serum BssFv in serum
Bs-scFv
+/- 50 kDa anti-CD3 anti-TAA
Bibody
+/- 75 kDa
S S
anti-CD3
VL-TAA VH-TAA
CL CH1
scFv-TAA scFv-CD3
anti-TAA
Tribody
+/- 100 kDa
S S
26
10 0 1
D
anti-TAA anti-CD3
E mU A 30.0 25.0 20.0
6
11
16
21
26
15.0
10.0
5.0
0.0
Figure 2: Production and stability. A) Gene and protein structure of a sc(Fv) 2 (BiTE), Fab-scFv (Bibody) and Fab(scFv)2 (Tribody) format. B) Coomassie brilliant blue stained gel of 10 µg of purified fractions. C) Tribody (squares) and Bibody (diamonds) residual activity in T-cell activation assay after 24h incubation at 37°C in PBS (dotted lines) and freshly prepared mouse serum (solid lines) D) Preparative gel filtration of Bibody and tribody structures and column calibration. E) 2 different Tribody structure analyzed on analytical size exclusion before (solid line) and after one (dashed line) or two (dotted line) freeze-thaw cycles (-80°C to 37°C).
The genetic structure of these fusion protein encoding genes thus starts with the coding sequence of either the Fd or the L chain, extended with a linker sequence separating the Fab from the scFv and the scFv molecule. Of course the scFv molecule can be substituted by other binding structures. New developments are producing single domain binders (human, camel, or shark derived) that can substitute for the two-domain scFv molecule. Furthermore, non-antibody domains often selected from more stable proteins allow the insertion of randomized loops and thus can be used for selecting an antibody-like type of binding to any desired target (Table 1). Also other domains or proteins can be included into the scaffold, depending whether the total fusion product can be produced. Examples here are natural protein ligands, cytokines, receptor domains, tags or enzymes. A molecular model of a Tribody molecule is shown in Figure 1. Noteworthy is the orientation of the binding planes as present in the scFv molecules, which are oriented towards the outside (as opposed to orientation towards the Fab fragment), predicting the possibility to cross-link cells and large molecules. Through the choice of flexible linkers of a reasonable size (e.g 15 amino acids), a large span of the molecule can be predicted, comparable to the span range of a
N. Mertens, Tribodies as a platform for multi-specific pharmaceuticals In “Bispecific Antibodies” (ed. R. Kontermann), in publication
5
Table 1: Examples of alternative binding domains in development.
Domain for targeting antibody derived: scFv Human V H or VL Camel/Llama VHH non-antibody derived: Ankyrin repeats Fibronectin Fibronectin Transferrin Lipocalins -Crystallin Cystein knots Synthetic coiled coil Protein A domains Derived from Platform Trade Name SCA dAbs Nanobodies DARPins Tetranectins AdNectins Trans-bodies Anticallins Affilins Microbodies Alphabodies Affibodies Company
Human antibody VH + VL domains Stable human V domains Camel or Llama heavy chain mAbs Synthetic human ankyrin repeats Trimeric human plasminogen binding protein 10th fibronectin type III domain (human) Non-glycosylated form of human transferin Derived from P. brassicae butterly Stable human protein from eye lens E. elaterium trypsin inhibitor II – cyclotide EETI-II De novo design S. aureus protein A domain
monoclonal antibody. We observed however no difference in expression or functionality when using shorter 5 amino acid linkers (Schoonjans, Willems et al. 2000).
2.2 Tribody Expression
The two fusion genes Fd-scFv and L-scFv can be produced in a variety of host cells. Production in Escherichia coli however resulted in a large fraction of precipitated protein as inclusion bodies. Also, a significant fraction of the produced material was degraded. In one attempt, a final purified batch of E. coli produced Tribody yielded only 0.1 mg/L and had a lower specific activity as compared to the same protein produced in mammalian cell culture. Mammalian cell culture is becoming more and more advanced for high level production of antibodies (especially IgG formats). We found that tribodies are easily expressed in a range of mammalian cells, including HEK293, NS0, SP2/0 and CHO cells. Mammial cells all posses the BiP protein, a HSP70 class of chaperone present in the endoplamatic reticulum. This chaperone is involved in guiding the correct assembly of the antibody Fab fragment. It mainly binds to the CH1 domain of the Fd chain and is only displaced after displacement by correct Fd:L chain pairing (Lee, Brewer et al. 1999). As a consequence, no free Fd chains are found in the culture medium. This quality control mechanism does not appear to be inhibited by the C-terminal extensions to the Fab chains. L:L dimers can be secreted by normal and malignent B-cells and also L-scFv:L:scFv dimers are found in the supernatant of transfected cells. We never found them to interfere with overall production since they are easily removed by simple ion exchange chromatography. It is however a good strategy to select a final cell line which produces less fee light chaint and base a purification strategy on the Fd-chain properties. For easy of evaluation, we often incorporate a hexa-histidine tag at the C-terminus of Fd-containing fusion protein (e.g. Fd-scFv-H6). Typical transient expression levels in either adherent or suspension HEK239 cells growing in Tflasks or shake flasks are in the tens of mg/L range. These values are obtained using a vector
N. Mertens, Tribodies as a platform for multi-specific pharmaceuticals In “Bispecific Antibodies” (ed. R. Kontermann), in publication
6
with an actin promoter, a Kozak-optimized mRNA and a standard IgG kappa chain signal sequence. Expression yields using stable transformed cells are of course highly dependent on the cell selection procedure and thereafter from the bioreactor conditions. Without putting effort in selection of high-level producing cells and using laboratory shake flask conditions, we have examples where yields higher than 100 mg/L were obtained. This leads us to assume that highlevel yielding bioreactor runs can be obtained with a > 1 g/L yield as is the case for most IgG productions, but this needs to be proven. Tribodies can also be expressed in yeast cells (Schoonooghe, Kaigorodov et al. 2009). We have demonstrated more or less similar overall yields using the yeast Pichia pastoris with an AOX promoter and the -mating type prepro sequence. We constructed bicistronic expression plamsids to be used for integration since double integration events occur with very low yield. Also in Pichia, no free Fd chains where observed in the medium. This indicates that the yeast chaperone system can substitute for the quality control mechanisms as seen in mammalian cells. Even when using defined medium for growing both host cells, the resulting cleared medium from a Pichia fermentation contained considerably more contaminants than cleared medium from serum-free growing mammalian cells. This resulted in a lower overall recovery yield after purification (20% in Pichia pastoris as compared with 70% from mammalian culture supernatans). The tribodies produced in Pichia showed the same functionality as those derived from mammalian cells. The -mating prepro sequence is highly efficient in guiding recombinant proteins to the medium (as
Figure 3: Binding and T-cell activation. A) A Tribody was constructed having specificity for two different TAA: mouse BCL1 and human placental alkaline phosphatase (hPLAP). The functionality of each of the indicated axes is demonstrated by binding to one antigen and detecting with the second. B) binding of the hPLAPxCD3xBCL1 Tribody to mouse T cells and detecting with hPLAP (which has alkaline phosphatase activity). C) Binding of the tribody to MO4I4 hPLAP transfected cells and detected with fluorescent BBCL1 antigen, and the same detection after binding the tribody to mouse T-cells. D) The hPLAPxCD3xBCL1 trispecific antibody acts as a bispecific antibody in a mouse T-cell proliferation assay primed both with hPLAP-positive cells as well as with BCL1 cells.
+ + + +
T E6 B T
Splenocytes BsAb or TsAb Tumor cells
+ - + + + - + + B B B T T T - - + + - + - + + - +
+ - + + + T T T + - + + - - - +
opposed to a major route to the vacuoles, which is the natural deposit for most yeast secreted proteins). This prepro-sequence is removed by KEX-2 protease.
N. Mertens, Tribodies as a platform for multi-specific pharmaceuticals In “Bispecific Antibodies” (ed. R. Kontermann), in publication
7
This protease prefers a Gln-Ala-Gln-Ala repeat after the cleavage site. These 4 amino acids can be removed by the di-aminopeptidase STE-13. We found the N-terminus of antibody fragments produced with this strategy to be heterogeneous as a result of partial or no removal of the extra amino acids. Also, when producing glycosylated Tribodies in yeast cells, it must be taken into account that yeast glycosylation differs from mammalian type of glycosylation which can lead to severe effects on half live in vivo and antigenicity.
2.3 Purification and Stability
Since most tribodies have a pI value of 7-8, we found a cation exchange fast flow column at pH 5.5 resolved with discrete step elutions an ideal and practical capture step. Further purification was dependent on the antibody. As mentioned, the inclusion of a hexahistidine tag simplifies this by using a common IMAC column. In this way a simple purification step based on immobilized metal chromatography can already lead to a highly enriched protein and produces only Fd-scFv:L-scFv heterodimers. Other purification strategies however could also be designed, e.g. based on protein-L, MEP-hypercel or a combination of ion exchange steps (Willems, Leoen et al. 2003). As a polishing step, gel filtration has the advantage to be able to separate dimers and possible aggregates that might have been co-purifying. Typical, a purified and concentrated sample of tribody shows a small fraction of dimer. The amount of dimerized or even multimerized protein is highly dependent on the nature of the scFv proteins used. This is a behavior known to be associated with scFv molecules (Arndt, Muller et al. 1998). It is possibly a result of the inherent domain swapping property of the scFv molecules, which leads to a diabody type of dimerization. We have noticed this behavior to be even more pronounced when
A anti-BCL1 B anti-BCL1 S S
C
3 H-thymidine incorporation (cpm x 1000)
Bs-scFv
OD (405 nm)
+/- 50 kDa monovalent anti-CD3
Bibody
+/- 75 kDa monovalent
anti-CD3
Tribody
+/- 100 kDa monovalent
anti-BCL1
S S
anti-BCL1 anti-CD3
concentration (nM)
Figure 4. A) In vitro comparison of bispecific antibodies crosslinking BCL1-cells to T-cells in a sc(Fv)2 (BiTE) format, a Fab-scFv (bibody) and Fab-(scFv)2 (tribody) format. B) Apparent affinity of the BsscFv (BCL1xCD3), bibody (BCL1xCD3) and tribody (BCL1xBCL1xCD3). An ELISA plate was coated with the BCL1 IgM tumour antigen and subsequently incubated with respectively the BsscFv (- - ), the bibody (- - ), the tribody (- - ) and an irrelevant bibody (- - ). The bound BsAbs were detected via their His-tag, with an anti-Histag Ab and a secondary alkaline phosphatase conjugated anti-mouse-IgG1. Data are representative for 3 independent experiments. C) In vitro T cell activating potential of the BsscFv, bibody and tribody. Mitomycin C treated BCL1 cells were co-incubated with syngeneic Balb/c spleen cells in the presence of decreasing concentrations of the indicated BsAb. Each condition was tested in triplicate.
N. Mertens, Tribodies as a platform for multi-specific pharmaceuticals In “Bispecific Antibodies” (ed. R. Kontermann), in publication
8
making sc(Fv)2 constructs. Using the same scFv sequences in a tribody structure (i.e. fusing them to the Fab) reduces the dimers and multimers already considerably. The dependency on the nature of the scFv used thus stresses the importance of proper selection of lead candidate scFv molecules, which should not only be characterized for binding and epitope selection but hence also for its lack of tendency to multimerize when incorporated in a more complex manifold (such as even a sc(Fv)2). Remarkably, for many tribodies we could purify the monomeric fraction and this fraction remained monomeric upon concentration and storage. Tribody preparations were stored at 4°C in PBS at 1 mg/ml concentration and as such stable for at least 3 months. However, an even higher stability can be achieved in a storage buffer ith a pH of 5-6, well away of the pI of 7-8 of most molecules. Freeze-thawing stress testing again revealed large differences in behavior depending on the antibody V-domain sequences used and even on the position of these domains (Figure 2), again indicating the need for extensive characterization and monitoring starting with building block selection, but even extending to selecting the optimal configuration within the manifold. Some tribodies could be repeatedly frozen and thawed in PBS buffer lacking any cryoprotectant or stabilizing additive without any sign of aggregation, while other were very prone to aggregate upon such treatment.
3 Tribody Activity
3.1 in vitro binding and T-cell activation
In order to study the use of these Tribodies in a natural immune inveronment we constructed BsAbs to target BCL1 (an IgM/ idiotypic determinant expressed on the murine myeloma cell line BCL1) and murine T-cell receptor determinant CD3 . The BCL1 system has been well characterized and is a model for NHL in a syngeneic immunocompetent Balb/c mouse model (Brissinck, Demanet et al. 1991). As a comparison, we used the “gold standard” for small bispecific antibody formats: the bispecific scFv (BsscFv or BiTE) format. This is the smallest BsAb (50 kDa) used and was composed of two scFv linked together by a peptide linker. This BsscFv(BCL1xCD3) was previously reported to be successful in treating BCL1 lymphoma bearing mice. We engineered BsAbs of intermediate size (75-100 kDa) in the tribody format by fusing single-chain variable fragments (scFv) to the C-terminus of one or both of the Fd and L chains of a Fab fragment. Starting from the anti-BCL1 scFv, a chimeric anti-BCL1 Fab-fragment was constructed by grafting the variable domains onto murine CL and CH1 constant domains of a Fab fragment, We then fused an anti-mouse CD3 2C11 scFv to the C-terminus of the heavy chain of the chimeric Fab-fragment and an anti-BCL1 scFv to the C-terminus of the light chain of the chimeric Fab-fragment. Co-expression of the heavy chain Fab-scFv fusion gene with respectively the light chain Fab gene or the light chain Fab-scFv fusion gene, lead to the production of an (BCL1xCD3) Fab-scFv bibody, or a (BCL1xBCL1xCD3) Fab-(scFv)2 tribody (Figure 4). The tumor binding mode (anti-BCL1) is different in all three BsAbs: In the BsscFv the tumor is recognized via a scFv, in the bibody via a Fab-fragment and in the tribody via both a Fab-fragment and a scFv. An ELISA experiment determined whether these differences resulted in a different functional affinity for the tumor antigen BCL1. To this end, the BCL1 tumor antigen was coated and subsequently incubated with a serial dilution of the respective BsAbs, using equimolar amounts. The bound Bs-Abs were then detected via the C-terminal His-tag, with a
N. Mertens, Tribodies as a platform for multi-specific pharmaceuticals In “Bispecific Antibodies” (ed. R. Kontermann), in publication
9
His-tag specific mAb. The B50 value of the bivalent binding Fab-(scFv)2 tribody was 3 nM, while the monovalent binding Fab-fragment in the bibody had a 10 times higher B50 (30 nM). Binding through the scFv alone in the BsscFv lead to a further decrease in apparent binding affinity. In a T-cell proliferation assay it was determined whether these differences in tumor binding affinity had an effect on the T-cell activating potential of the different BsAbs in vitro. A serial dilution of the different BsAbs, starting from equimolar amounts, was incubated with mitomycine inactivated BCL1 tumor cells and syngeneic spleen cells. All three BsAbs were found to be capable of activating T-cells in the presence of tumor cells. Despite the difference in affinity seen with ELISA, no difference in T-cell activating potential could be observed using both the monovalent binding bibody and the BsscFv. However, the bivalent binding tribody exerted a clearly improved T-cell activating potential. The maximum T-cell proliferation reached with the bivalent tribody was 2 -fold higher than with the monovalent bibody or BsscFv (Figure 4). Moreover, the bivalent tribody remained capable of inducing T-cell proliferation at 4 -fold lower concentration compared to the monovalent bibody and the BsscFv. Also in targeting other tumor markers, the increase in valency lead to better tumor binding (Schoonooghe, Burvenich et al. 2010).
3.2 In vivo model: mouse BCL1 lymphoma
In a next step it was investigated whether the differences in molecular weigh and tumor avidity would also lead to a different therapeutic potential of the BsAbs. Groups of mice (n=13) were inoculated i.p. with 5000 BCL1 cells on day 0 and i.v. (in the tail vein) treated with 4 daily injections on days 2 to 5. Treatment was performed with different equimolar amounts (200, 100 or 50 pmol/inj) of the BsscFv, the bibody or the tribody. In addition, a group of 6 mice were injected with PBS or 200 pmol/inj of anti-Id mAb that targets bivalently to the BCL1 tumor surface. All the animals in the control group receiving PBS or anti-Id mAb treatment developed terminal illness and were euthanized by day 60 or 140, respectively (Figure 5). After dissection the spleen was found to be enormously enlarged by the massive presence of tumor cells. In contrast, the mice treated with 200 pmol/inj of each of the BsAbs demonstrated significant (P
