IMMUNORNASE FUSIONS - Fusion Protein Technologies for Biopharmaceuticals: Applications and Challenges - page 336

but the product yield was usually very low. For example, in
the case of dimeric fusion proteins, yield from culture was as
low as

immunoRNases using this expression system in industrial
scale bioreactors should be possible in the near future.
Large-scale purification of secreted fusion proteins should
not cause major problems. Oligo-histidine tags have been
engineered to most immunoRNases to use immobilizedmetal
affinity chromatography (IMAC) for their isolation; this
technique may be used on large scale. Furthermore, it may
be possible to purify larger constructs based on fully func-
tional antibodies [133,135], on scaled-up protein A columns.

20 m g/L [63,86,129]. In an attempt to improve
periplasmic expression yield, Barth et al. [91] introduced a
method entailed the growth of bacteria under osmotic stress
(4% sodium chloride, 0.5M sorbitol) in the presence of
natural osmolytes (glycine betaine and hydroxyectoine)
protecting proteins at high salt concentration. Expression
of angiogenin fused to an anti-CD64 single-chain Fv under
those conditions increased yield to 1 mg/L [89]. However,
this method although improved, was still not robust enough
for large-scale production. In addition to an inadequate
yield, bacterial systems had other limitations (discussed
in Reference 84). For example, it is very difficult to produce
larger proteins of the optimal 60-120 kDa range in bacteria.
Therefore, other approaches had to be sought which led to
the exploration of eukaryotic host systems.
St

22.4 ALTERNATIVES

The concept of immunoRNase fusions encompasses both,
classic immunotoxins and chemical conjugates of RNases to
antibodies, which are specific to cancer antigens. Both
approaches are still under development. Some immunotox-
ins have been effective in the clinical trials. However,
immunotoxins with effector domains that are plant are
usually highly immunogenic. This is a significant problem
because the patient's immune system produces antibodies to
the toxins and the drugs desired effects are neutralized. In
addition, in the clinic immunotoxins have induced serious
hepato- and nefro-toxicities as well as a vascular leakage
syndrome. Immmunogenicity and general toxicity usually
preclude repeated administration of these drugs. Intensive
research toward the reduction of these adverse effects by
modification or replacing the effector moieties is under way
and recent advances were reviewed [176,177].
Rybak [170] recently reviewed Onconase 1 and antibody
chemical conjugates. The enzyme covalently linked with
anti-CD22 antibodies, LL2 [58] and RFB4 [170,178] dem-
onstrated unusually high activity against the human lym-
phoma Daudi cell line. IC 50 values of 20-70 pM were
observed. Both conjugates were also very active in vivo
in a murine model of human lymphoma—significantly
increased the survival of tumor bearing mice. These are
very encouraging results but the drawback of this approach
is the lack a reliable, large-scale production method. The
recently published method of preparation [179] is essentially
the same as the original method [58] and requires chemical
modification of both, the enzyme and the antibody. Intro-
duction of free sulfhydryl groups to the latter by the inter-
action of the antibody (Ab) primary amino groups with 2-
iminothiolate (IT) leads to heterogeneity of the intermediate.
This is because typical IgG contains

oker et al. [84] expressed previously mentioned anti-
CD30 scFv fusion proteins with angiogenin in transiently
transfected human embryonic kidney cells, HEK 293T. The
constructs were not toxic to host cells and were equipped
with leader sequences that mediated secretion of fusion
products into the culture supernatant. The yield, however,
did not exceed 0.5mg/L. CHO cells were also tested as
potential host cells [87]. Unfortunately, in an effort to
produce a humanized anti-CD22 scFv-angiogenin in CHO
cells, a disappointing yield of 0.13 mg/L was obtained.
Further studies with HEK 293 T led to significantly higher
yields of anti-CD30 scFv-RNase 1 (4mg/L) [116]. Recently,
the same group reported that production of anti-CD30 scFv-
Fc-RNase 1 in transiently transfected HEK 293T cells
resulted in the yield of 2-6mg/L. Notably, when a derivative
cell line, HEK293-EBNAwas used; the authors were able to
recover 100mg/L of the immunoRNase [175].
Mousemyelomacellswerealsoemployedtoproduce
immunoRNases. Transiently, transfected NSO or Sp2/0-
Ag14 cells were used as hosts to express the construct that
was composed of two Onconase 1 molecules fused to
humanized hLL1 anti-CD74 antibody [133]. Quite
recently, two groups succeeded in the stable integration
of immunoRNase genes into the genome of myeloma cells,
NSO [88] or Sp2/0 [135]. The expressed fusion proteins
were not toxic to the host cells and they were secreted into
culture supernatants. The former paper reported that the
yield of previously designed [87] anti-CD22 scFv-angiogenin
immunoRNase produced in stably transfected NSO cells was
5-10mg/L. Anti-CD22 specific Onconase 1 fusion proteins
were produced in the same cells with the yield of 20mg/L
[139].
The results from human embryonic and mouse myeloma
cell lines are very promising and it seems that one of the
major obstacles in the clinical development of immuno-
RNases, the lack of efficient production systems, may be
resolved with the use of mammalian hosts. Production of

€

50 Lys residues
bearing e -amino groups reactive with IT. One molecule of
Onconase 1 contains 12 lysine residues (including two
necessary for enzymatic activity). These residues are poten-
tially available for the amino-reactive arm of a hetero-
bifunctional (amino- and sulfhydryl reactive) crosslinker
used for conjugation of both moieties. Consequently, the
conjugates are also heterogeneous. It is possible to minimize
the heterogeneity by careful optimization of the reaction