MODULAR ANTIBODY ENGINEERING: ANTIGEN BINDING IMMUNOGLOBULIN Fc CH3 DOMAINS AS BUILDING BLOCKS FOR BISPECIFIC ANTIBODIES (mAb) - Fusion Protein Technologies for Biopharmaceuticals: Applications and Challenges - page 562

which had previously been exposed to the mAb 2 could be
detected (Figure 39.8), demonstrating that the mAb 2 could
bind two different antigens at the same time.
The bispecific nature of mAb 2 HC-TNF461-1 could also
be shown by a similar experimental setup. In this case, the
mAb 2 or its parent trastuzumab were incubated with the
human HER2/neu positive breast cancer cell line SKBR3.
Similar to the experiment described earlier, biotinylated
TNF- a specifically could detect the mAb 2 but not trastuzu-
mab while both antibodies were detectable with similar
sensitivities using antihuman light chain antibodies (not
shown). These data demonstrated that both tested mAb 2
proteins were able to simultaneously bind antigen via their
Fab arms and the binding site in the CH3 domain.
In summary, these data provided proof-of-principle for
the concept that modular replacement of Fc CH3 domains of
conventional monoclonal antibodies with antigen-binding
CH3 domains is a viable strategy to generate bispecific
antibodies (mAb 2 ). The affinities of mAb 2 proteins to their
antigens are unchanged relative to the parental antibody or to
the Fcab, which is the donor of the antigen-binding CH3
domain. In addition, immuno-effector functionalities such as
ADCC or CDC are not compromised in the mAb 2 format.
The functional ADCC data are in-line with Biacore results
demonstrating that binding of the mAb 2 to the relevant Fc
receptor CD16a is unchanged compared to the unmodified
parent. Furthermore, the presence of the antigen-binding site
in the CH3 domain does not interfere with binding to the
FcRn receptor suggesting a long half-life of mAb 2 in vivo.In
conclusion, mAb 2 proteins represent a novel class of bispe-
cific proteins, which feature only minimal structural changes
compared to conventional antibodies while retaining all their
functional attributes.
Treatment of cancer patients with therapeutic monoclo-
nal antibodies has been quite successful. It has become clear,
however, that only a subpopulation of patients will benefit
from such a therapy and that a significant percentage of
treated patients will develop resistance to antibody therapy.
In order to overcome these shortcomings, combination
therapies have been successfully tried in preclinical animal
models using two monoclonal antibodies targeting two
different but interrelated tumor-relevant therapeutic targets.
Such studies have demonstrated additive or even synergistic
effects of the combination therapy versus the individual
antibodies [9-11]. Consequently, clinical trials are now
ongoing to translate these preclinical results into men. A
combination therapy with two therapeutic antibodies could
potentially be achieved with a single molecule such as a
bispecific mAb 2 protein. Other attractive applications for
mAb 2 molecules can be envisioned. For instance, improved
efficacy and lower side effects may be expected by specifi-
cally targeting the mAb 2 to diseased tissue using one of the
two specificities of the antibody while the second specificity
supplies the therapeutic principle. Similarly, one could

ADCC, Calu-3 cells, (HER2 + )

90
80

HC
HC-TNF461-1

70

60

50

HC-TNF461-1 +
100nM TNFa
human lgG1

40

30

20

10

0

mAb ( μ g/mL)

FIGURE 39.6 mAb 2 HC-TNF461-1 is equipotent to trastuzumab
(HC) to kill Calu-3 cells by ADCC. Calu-3 cells were opsonized
with increasing concentrations of antibodies and then incubated
with human NK cells at a Calu-3:NK 1:5 ratio for 4 h. Killed cells

were detected with 7-AAD by flow cytometry.

¼

17 nM, EC 50 RX-TNF353-2: 23 nM) demonstrating the inde-
pendence of the cytokine-binding site in the CH3 domain
from its molecular context (Fcab or mAb 2 ) (Figure 39.7).
Finally, we tested if the mAb 2 was able to interact simul-
taneously with both antigens to prove its true bispecific
functionality. For that purpose, a human CD20 positive B-
cell line (CFB4.2) was incubated with increasing concentra-
tions of rituximab or mAb 2 . Afterwards, cell-bound proteins
were detected either with an antihuman kappa chain specific
monoclonal antibody coupled to a fluorescence dye or with
biotinylated TNF- a followed by streptavidin coupled to a
fluorescent dye. Incubation with the anti-light-chain-specific
antibody revealed that both rituximab and RX-TNF353-2
could interact equally well with CD20 on the cell surface
via their Fab arms indicating that the amino acid changes in
the CH3 domain did not compromise binding to CD20.
Importantly, incubation with TNF- a showed that only cells

5
4
3
2
1
0
-1

TNF353-2
RX-TNF353-2

0 1 2

Antibodies log(ng/mL)

3 4 5

FIGURE 39.7 Comparable TNF- a binding of Fcab TNF353-2
and mAb 2 RX-TNF353-2. TNF- a was coated on ELISA plates and
increasing concentrations of antibodies were added. Bound pro-
teins were detected with antihuman CH2 domain antibodies.