RESURGENCE OF BISPECIFIC ANTIBODIES - Fusion Protein Technologies for Biopharmaceuticals: Applications and Challenges - page 509

provides for a high antitumor activity and is adding a
vaccination effect to the redirecting lytic activities of the
antibodies [13]. The EpCAM/CD3/Fc g R-trifunctional Tri-
omab catumaxomab (Removab 1 ) has gained market
approval in Europe in 2009 for the treatment of ascites
resulting from EpCAM-expressing cancers [14]. The
HER2/CD3-trifunctional antibody ertumaxomab is in for-
mal Phase II clinical testing after having shown promising
Phase I results [15], and the CD20/CD3- and GD2/CD3
trifunctional antibodies have been in pilot Phase I clinical
studies. A major drawback of the Triomab approach is the
high immunogenicity of rat and mouse antibody
sequences, which trigger a neutralizing antibody response
in nearly every patient, thereby pre-empting a retreatment
[16]. With a serum half-life of 2.1 days and infrequent i.p.
infusions, rather short exposures are achieved that may
limit the time for efficient T-cell engagement. Neverthe-
less, high biological activity has been reported in clinical
trials with distinct Triomabs.
Other groups have introduced structural elements for
heterodimerization into the heavy chains of human, human-
ized, or human/murine chimeric origin with the promise of
generating bispecific antibodies with much reduced immu-
nogenicity. A very subtle, structure-guided approach took
advantage of single amino acid exchanges in the CH3
domains creating bulky hydrophobic amino acid residues
in one heavy chain andmatching hydrophobic pockets in the
other heavy chain (Figure 35.1B). This technology is
referred to as “knobs-into-holes” [17]. Genentech/Roche
published on a bispecific knobs-into-holes IgG that can
cross-link the inhibitory Fc g receptor CD32B with the high
affinity Fc

solution has been developed by researchers fromRoche [22].
Through crossover of heavy and light sequences within a
knobs-into-holes bispecific antibody they crafted so-called
CrossMAbs, which have properly paired variable domains.
A CrossMAb, which is bispecific for VEGF-A and angio-
poetin-2, has been characterized as an antiangiogenic drug
candidate.
Another potential solution to mispairing of light chain has
been worked out by the companies Merus and Genmab.
Merus has developed so-called single V L mAbs, which use
only one light chain species to serve a large repertoire of
human heavy chains [23]. Single V L mAbs come from phage
libraries or, in the future, will be generated by transgenic
mice harboring the human vH gene repertoire and one
human light chain gene. Evidently, the affinity and binding
specificity of such antibodies solely resides within the heavy
chain V H , but seem to have uncomprimized activity. If two
single V L mAbs are simultaneously expressed in a producer
line, a bispecific antibody called “biclonic” is formed (Fig-
ure 35.1E). By engineering the heavy chains, heterodime-
rization of biclonics can be forced to yield a high proportion
of the desired bispecific variant.
On the basis of the natural Fab arm exchange observed for
the IgG4 class [11], a new bispecific format has been
developed by Genmab [24]. The Fab arm exchange by
IgG4 structurally relies on both the IgG4 hinge region
and critical amino acid residues in the IgG4 CH3 domain,
predominantly Lys 409. By conferring these structural ele-
ments to other IgG subtypes, a Fab arm exchange leading to
a bispecific antibody species can be forced in vitro. How this
reaction is made irreversible has not yet been disclosed.

RI onmast cells for the treatment of asthma [18].
Researchers from Amgen have found another solution for
heterodimer formation of heavy chains. They introduced by
recombinant technology charged amino residues in the CH3
domain that interact with matching residues of the opposite
charge [19].
Genentech/Roche presented in January 2010 another
approach for pairing heavy chains called LUZ-Y antibodies
[20]. This technology uses C-terminally fused leucine zipper
domains derived from a heterodimerized transcription fac-
tor, which are later cleaved off the purified bispecific anti-
body with a protease (Figure 35.1C). Researchers from
EMD Serono have likewise focused on the CH3 domain
for heteromeric pairing of heavy chains. They developed a
“strand exchange engineered domain” (SEED) as basis for
their SEEDbodies [21] (Figure 35.1D).
A remaining issue of these three novel approaches is a
potential mispairing of light chains with heterodimerized
heavy chains. In theory, the properly paired bispecific anti-
body species will represent only 25% of possible light
chain/heavy chain combinations. This will decrease yield
and increase the complexity of purification. There are
intense efforts ongoing to solve this issue. One possible

e

35.3 SYMMETRIC IgG-LIKE BISPECIFIC
ANTIBODIES

A number of bispecific antibody formats have recently
emerged that no longer rely on heteromeric pairing of heavy
chains and hence result in a symmetric bispecific antibody.
One such format has been developed by the company
Domantis (now GSK) and is called “dual-targeting” (DT)
Ig. It is combining two so called single-domain antibodies
(dAbs) [25], which have been engineered from isolated
heavy and light chain variable domains. A single-domain
antibody is using only three complimentarity determining
regions (CDRs) for antigen binding and no longer requires
pairing with the respective other variable domain for achiev-
ing stability, solubility, or high binding specificity and
affinity. If two distinct single-domain antibodies are replac-
ing within an IgG molecule the otherwise cooperatively
antigen-binding V H /V L domains, dual specificity is obtained
(Figure 35.2A). A dual cytokine neutralizing antibody bis-
pecific for IL-4 and IL-13, and a bispecific anti-VEGF/anti-
EGFR mAb are in early development [26].