Biomedical Engineering Reference
In-Depth Information
functioned as highly specific and effective blockers of IL-6
and CNTF. Moreover, neither complex cross-inhibited, that
is, the IL-6:gp130 heterodimer did not block CNTF, and the
CNTF:gp130 heterodimer did not block IL-6, nor did they
inhibit signaling initiated by leukemia inhibitory factor
(LIF) (AN Economides and N Stahl, unpublished results),
another IL-6 cytokine family member whose receptor uti-
lizes gp130 in conjunction with LIFR
b
[130]. This first
version of cytokine Traps provided proof-of-concept that
heterodimeric proteins composed of R1:R2 ectodomains
could act as highly specific blockers of their cognate cyto-
kine. Furthermore, this work illustrated that the ectodomain
of gp130 is by itself a poor blocker for the different members
of the IL-6 family of cytokines (an observation indepen-
dently verified [131]), but that when it is placed in a
preformed complex with a specificity-determining receptor
ectodomain, then the resulting hybrid protein acts as an
extremely specific blocker for its cognate cytokine. (A
notable exception to the low effectiveness of soluble
gp130 as a blocker is its ability to selectively inhibit IL-6
trans-signaling [132]; this is discussed in more detail below.)
Conversely, this work demonstrated that the properties of the
soluble versions of IL-6R, IL-11R, and CNTFR could be
altered from potentiators to blockers when engineered as
obligate heterodimers with the ectodomain of gp130.
(IL-6R-Fc)
2
and (gp130-Fc-His6)
2
—by nickel affinity chro-
matography [133]. The technical choice of His6-tagging
gp130-Fc was driven by the need to completely remove any
traces of IL-6R-Fc from the IL-6R-Fc:gp130-Fc preparation,
as IL-6R-Fc would act as a potentiator of signaling. The
first-generation Traps utilizing Fc differentially tagged with
a hexahistidine tag were generated for IL-6, IL-11, CNTF,
and LIF, all using combinations of gp130-Fc-His6 as the
common R2 component [53] (Economides and Stahl,
unpublished results). All of these Traps were shown to be
highly specific and effective blockers for their cognate
cytokine. However, their production and purification
remained problematic for two reasons: (a) the desired
species, R1-Fc
R2-Fc-His6 heterodimer, was only about
half of the total Fc-tagged protein, and (b) both the R1-
Fc
R2-Fc-His6 heterodimers and the R2-Fc-His6 homo-
dimers have hexahistidine tags. Although, isolation of the
desired species (R1
R2 heterodimers) away from the
“contaminating” homodimers could be accomplished to
generate enough material for laboratory tests, it was deemed
difficult and costly for large-scale production (K Bailey,
unpublished results).
Therefore, a second generation of cytokine Traps was
developed by engineering “in-line” fusion proteins com-
posed of the ectodomain of a specificity-determining subunit
(R1) fused to a shared subunit (R2) and tagged with Fc,
either in an R1-R2-Fc or an R2-R1-Fc configuration. These
were generated for several cytokines and then tested to
determine which version worked best. These were the
R1-R2-Fc configuration of the IL-6 Trap (IL-6R-gp130-
Fc), and the R2-R1-Fc of the IL-4 Trap (IL-2R
g
-IL-4RA-
Fc) and the IL-1 Trap (IL-1RAP-IL-1R1-Fc; Figure 9.2).
Unlike their first-generation counterparts, these in-line
fusion traps were expressed as homogeneous homodimeric
proteins that could be easily purified. In addition, just like
their first-generation counterparts, these in-line fusions were
highly potent and specific blockers and bound their cognate
cytokines with high affinity, displaying dissociation con-
stants in the low or subpicomolar range. Furthermore, they
were much more effective as blockers than the correspond-
ing homodimeric receptor-Fc fusions [53].
The successful engineering of in-line fusion Cytokine
Traps enabled preclinical testing of their pharmacokinetic
properties and their ability to block cytokine activity in
animal models. This work focused on two Cytokine Traps of
therapeutic interest, the IL-1 Trap and the IL-4 Trap. Both of
these Traps were shown to persist in circulation for many
days after a single subcutaneous or intravenous injection in
cynomolgus monkeys. In vivo, the IL-1 Trap was shown to
block responses to exogenously administered interleukin
1-
b
(IL-1
b
), and then in a mouse model of collagen-induced
arthritis. Similarly, the IL-4 Trap was shown to be effective
in blocking IL-4 responses in vivo, both in mice and in
cynomolgus monkeys [53].
9.3.2 Turning Heterodimeric Soluble Receptor-Based
Ligand Traps into Therapeutics
Although this early work provided proof-of-concept, the
production of these molecules was inefficient and their
purification was difficult. A mixture of homodimers and
heterodimers was generated, and the homodimers had to be
removed because IL-6R and CNTFR homodimers can
potentiate—rather than block—IL-6 or CNTF activity,
respectively [123,126] (Figure 9.3). In addition, there was
the concern that these artificial heterodimeric hybrid pro-
teins would be immunogenic and display poor pharmaco-
kinetic profiles. The concern for immunogenicity arose from
the fact that these were fusions with leucine zipper domains
of two transcription factors, that is, proteins that are not
naturally found in circulation.
Therefore, taking advantage of the experience gained
with immunoadhesins [47,48], and mirroring the experience
with TNFR-Fc [46], we engineered the first generation of
Fc-based ligand traps. This presented a new engineering
problem: how does one bring together the ectodomains of
two different receptors and specifically heterodimerize them
via the Fc domain of IgG, which will drive dimerization to
form homodimers of each ectodomain as well as the desired
heterodimers? One solution to this problem was to tag the C-
terminus of gp130-Fc with a hexahistidine tag, as this would
allow the purification of IL-6R-Fc:gp130-Fc-His6 hetero-
dimers from the corresponding homodimeric proteins—
Search WWH ::
Custom Search