RECOMBINANT ALBUMIN FUSION PROTEINS - Fusion Protein Technologies for Biopharmaceuticals: Applications and Challenges - page 168

mechanical and thermal stress and the formation of these
aggregates was associated with increased immunogenicity
in mice [87]. In the case of the albumin-hGH fusion protein,
the use of nonionic surfactants such as polysorbate 20 and
polysorbate 80 was found to minimize aggregation during
fermentation, purification, freeze-drying, shipping, and stor-
age [88], and so this approach was applied to IFN- a -2b-rHA.
The addition of polysorbate 80 improved the stability of
IFN- a -2b-rHA in response to agitation, but not to thermal
stress [87]. In fact, the surfactant increased aggregate for-
mation in aqueous solution during inert storage. In an
alternative approach to improve stability, the Cys34 residue
of albumin was replaced with serine. The stability of the
resulting fusion protein (IFN- a -2b-rHA(C34S)) was better
than that of IFN- a -2b-rHA. After either agitation for 72 h or
incubation at 60 C for 2 h, more than 90% of the protein
content remained monomeric. In addition, immunogenicity
was reduced after being subjected to these conditions.
Pharmacokinetic analysis in rats indicated similar pharma-
cokinetic properties to that of IFN- a -2b-rHA; both fusion
proteins had a half-life of

the C-terminus of rFVIIa. The length of the linker (31 amino
acids) was chosen to minimize potential interactions
between the two proteins while retaining a high specific
activity of rFVIIa [11,76].
While the use of a flexible glycine/serine linker was
successfully used to produce rVIIa-FP, the same approach
could not be applied to FIX, as it produced a fusion protein
with low FIX-related activity [12,80,90]. A possible explan-
ation for this is that the albumin portion of the protein
interferes with the interactions between FIX and its cofactor
and substrate, FVIIIa and FX [90]. This led to the develop-
ment of a FIX-recombinant albumin fusion protein using a
linker sequence that was derived from the natural cleavage
site of FIX responsible for proteolytic activation [12]. The
linker sequence is based on amino acids 137-153 derived
from the N-terminus of the activation peptide of FIX.
Activation of FIX-recombinant albumin fusion protein by
either FXIa or FVIIa/TF cleaves the linker in parallel,
separating the FIXa and rHA moieties of the fusion protein.
The liberation of FIXa from recombinant albumin at the time
of activation allowed for 50 h.

Another approach to improve the homogeneity, stability,
and activity of IFN- a -2b-rHA was to modify the linker
[86,87,89]. Variations in linker size and conformation can
alter the interaction between the two fused proteins and
between the therapeutic protein and its target. Various linker
sizes (1, 2, 5, or 10 amino acids) and conformations (flexible
glycine/serine linker (GGGGS) or rigid linker (PAPAP) or
helix-forming linker (AEAAAKEAAAKA)) were inserted
between rHA and IFN- a -2b, and the properties of the
resulting fusion proteins were characterized [88]. It was
found that a linker length of five residues allowed the fusion
proteins to migrate as a single band on SDS-PAGE, facili-
tating correct disulfide bond formation. Aggregate forma-
tion after incubation at 37 C for 10 days was also reduced
when these linkers were used. The different linkers varied in
terms of susceptibility to hydrolysis; the rigid linker was the
least susceptible at pH 6 or 7. Activity assays showed that
structured linkers separated rHA and IFN- a -2b more effec-
tively compared to a flexible linker, as they increased the
antiviral activity more significantly. A rigid, flexible, or
helix-forming linker increased antiviral activity in vitro
by 39, 68, and 115%, respectively [89].

largely enhanced biological

activity [12].

10.5 CHALLENGES

One of the inherent challenges of recombinant albumin
fusion technology is maintaining specific activity of the
therapeutic protein, which in most cases has been affected
by the presence of the albumin moiety. While the use of
linkers and the flexibility in orientation of the fused genes
can help optimize biological activity, the presence of recom-
binant albumin can interfere with interactions between the
therapeutic protein and its target(s). The development of
rVIIa-FP and rIX-FP, as discussed earlier, illustrate some of
the typical challenges and solutions encountered in recom-
binant albumin fusion technology, and highlight the fact
that, while recombinant albumin fusion technology is rela-
tively simple and easily applied to various settings, each
novel protein poses unique conditions and challenges that
must be considered. To retain as much specific activity as
possible, it may therefore be necessary to develop a specific
concept for each fusion protein.
As with all therapeutic proteins, albumin fusion proteins
carry a risk of toxicity and immunogenicity [2,13]. However,
in a Phase I/II study of alb-IFN in patients with CHC (mostly
genotype 1) for whom prior IFN- a therapy had failed, the
risk of immunogenicity appeared to be low [50]. In most
patients, titers of anti-albumin-IFN- a antibodies remained
low throughout the study. Just over 10% of patients had anti-
albumin-IFN- a antibodies at baseline, which is not surpris-
ing since prior IFN- a -based therapy had not been successful.
The immunogenicity of alb-IFN did not correlate with
patient demographics, disease status, prior treatment history,

10.4.2 FVII and FIX

Early attempts to prolong the half-life of FIX using recom-
binant albumin fusion technology did not or did only
moderately improve the pharmacokinetics of FIX [80].
Similar results were expected for rFVIIa, as the two proteins
are structurally similar members of the same family of
vitamin-K-dependent coagulation factors [11]. A novel
rFVIIa fusion protein (rVIIa-FP) was developed using a
flexible glycine/serine linker inserted between albumin and