CovX-BODIES - Fusion Protein Technologies for Biopharmaceuticals: Applications and Challenges - page 548

was used as a reactive hapten, designed to trap the lysine
e -amino group and facilitate the formation of an enamine
intermediate. The diketone hapten, after coupling to keyhole
limpet hemocyanin was used in the immunization of 129 G IX þ
mice [3]. Twenty hybridomas producing antibodies that bound
the 1,3-diketone were isolated from the mice. Two of these
antibodies, m38C2 and m33F12, were shown to catalyze the
synthetic and retrosynthetic aldol reactions, whereas the
remaining 18 antibodies were found to be inactive as catalysts.
Mechanistic studies of m38C2 and m33F12 revealed that the
reaction with the hapten proceeded via the proposed enamine-
vinylogous amide intermediate and these antibodies facilitated
the aldol and retro-aldol reactions via amechanismthatmimics
the natural class I aldolases. The catalytic activity of m38C2
and m33F12 can be attributed to an active site Lysine with a
perturbed pKa. The structure of Fab' of m33F12 was deter-
mined at 2.15 A resolution and in general was found to be
identical toother knownFabstructures [4]. The antigenbinding
pocket was found to be 11A deep and comprised of mostly
nonpolar amino acid residues with a lysine residue, LysH93 at
the base of the pocket. In most antibodies, H93 is an alanine
next to a polar arginine H94; which in the case of m33F12 was
found to be replaced by a nonpolar isoleucine residue. The
structure revealed that no charged residues were close enough
to form a salt-bridge or an H-bond with LysH93. It is hypothe-
sized that this unusual hydrophobic microenvironment along
with absence of an H-bond or a salt-bridge perturbs the pKa of
e

of antibody m38C2 can be formed in vivo and this complex
increases the circulating half-life of the integrin targeting
small molecule and effectively inhibits tumor growth in
animal models of Kaposi's sarcoma and colon cancer.
The in vivo half-life of this prototype murine CovX-Body
was projected to be around 2 h (based on the k off rates for the
complex of m38C2 and a 1,3-diketone linker, as determined
by surface plasmon resonance). Remarkably, the authors
found that the circulatory half life of the RGD targeting
CovX-Body to be around 3 days. Additional successful
examples of chemical programming of m38C2 with small
molecule pharmacophores resulted in generation of highly
potent and selective CovX-Bodies targeting the endothelin
and CCR5 receptors (Figure 38.3) [6,7].
To develop CovX-Bodies for human therapeutic appli-
cations, the murine aldolase antibody m38C2 was human-
ized through grafting of the murine complementarity
determining regions (CDRs) into a human antibody frame-
work sequence. Human V k gene DPK9 and human J k gene
JK4 were used as frameworks for the humanization of the k
light chain variable domain. Human V H gene DP47 and
human J H gene JH4 were used as frameworks for the
humanization of the heavy chain variable domain of
m38C2. Related frameworks from the same light and heavy
chain subgroups, V k I and V H III, respectively, have been
used in the humanization of widely used therapeutic anti-
bodies, such as Herceptin 1 [8,9]. Combinations of light
chains derived from germ line DPK9 with heavy chains
derived from germ line DP47 are also frequently found in
native human antibodies and have been used as templates for
synthetic human antibody libraries [10,11]. All CDR resi-
dues, as well as defined framework residues in both light and
heavy chain variable domains, were grafted from murine
38C2 onto the human framework. The selection of the
grafted framework residues was based on the crystal struc-
ture of the murine antibody m33F12 described earlier. The
grafted framework residues consist of five residues in the
light chain and seven in the heavy chain and encompass
residues that are likely to participate directly or indirectly in
the catalytic activity of the aldolase antibody. These include
the reactive LysH93, which is positioned in the framework
region 3 (FR3) of the heavy chain. In addition to the CDR
residues, a number of framework residues line the active site
pocket. Among these, LeuL37, GlnL42, SerL43, ValL85,
PheL87, ValH5, SerH40, GluH42, GlyH88, IleH89, and
ThrH94 were grafted on the human framework. Six residues,
SerH35, ValH37, TrpH47, TyrH95, TrpH103, and PheL98
which are conserved between murine monoclonal antibodies
33F12 and 38C2, are within a 5-A radius of LysH93. These
residues were conserved in the humanization process. The
resulting humanized variable domains were grafted on
human constant domains. The G1m(f) heavy chain and
Km3 light chain allotypes were selected due to their high
prevalence

-amino group of LysH93 to around 6. This enables LysH93 to
function as a strong nucleophile at neutral pH when all other
lysine side chains are protonated and non-nucleophilic. The
structure of m38C2 was expected to be identical to m33F12 as
all the residues that are in van der Waals contact with LysH93
are conserved in both the antibodies, except for m33F12
IleH94, which is a threonine in m38C2.
In the course of mechanistic studies of the aldolase
antibodies m38C2 and m33F12 it became apparent that
these antibodies readily react with 1,3-diketones to form
a so-called reversible covalent bond via an enamine docking
mechanism (Figure 38.3). Thus, any molecule that is func-
tionalized with a 1,3-diketone linker will react with LysH93
of the aldolase antibody and be fused to the antibody via a
reversible covalent bond. This effectively results in the
catalytic, untargeted antibody being chemically pro-
grammed with the binding specificity of the molecule
carrying the 1,3-diketone linker. This concept was demon-
strated in 2003 by Rader et al. who synthesized a 1,3-
diketone derivative of a potent RGD peptidomimetic small
molecule that bound integrins a v b 3 and a v b 5 . They showed
that upon fusion with m38C2 the RGD peptidomimetic
effectively mediated the cell surface binding of m38C2 to
integrin expressing cells [5]. The antibody m38C2 was
incapable of binding the integrin expressing cells in absence
of a fused peptidomimetic. Through mouse experiments
they were able to show that the integrin targeting complex

in pooled blood donor populations and