Artificial cells as a novel approach to gene therapy - Artificial Cells, Cell Engineering and Therapy - page 271

genesis are more active during cancer, this could have been the rationale of
using the RGD as targeting ligands for tumor cells.
In contrast to the above targeting strategy, angiostatin, which has no known
tumor-targeting ability, appeared to accumulate in tumors! When encapsulated
cells secreting angiostatin were implanted into tumor-bearing mice, no
angiostatin activity could be observed in serum for the first 3 weeks, whereas
as similar implantation into non-tumor-bearing mice resulted in high serum level
of angiostatin even from the first week (Cirone et al., 2003). Therefore, fusion to
angiostatin could be a favorable way to target other molecules to the tumor or
even to the tumor vascular endothelium. Although several receptors for
angiostatin have been identified, the exact molecular mechanism of this tumor-
targeting has yet to be elucidated.
Targeting the brain
Special consideration is required when targeting recombinant molecules to the
CNS. The blood±brain barrier, comprising tight junctions between capillary
endothelial cells in the CNS vasculature, prevents most large molecules,
including proteins, from freely moving into the CNS. While the blood±brain
barrier could be physically circumvented by disruption of the junctions or direct
injection into the CNS, these methods would be quite invasive. To get
peripherally administered therapeutic proteins to cross the blood±brain barrier,
one can take advantage of endogenous receptor-mediated transcytosis systems.
These systems bypass the tight junctions between the endothelial cells by
allowing trans-cellular transport via receptor-mediated endocytosis such as that
for insulin, transferrin and the more recently discovered transport systems of
melanotransferrin (also called P97) (Demeule et al., 2002) and RAP (receptor-
associated protein) (Pan et al., 2004). Fusion techniques such as those described
above should allow functional therapeutic proteins to be combined with ligands
for these receptors to give bi-functional molecular hybrids. Such hybrids could
then be secreted from engineered cells in peripherally implanted microcapsules
but achieving delivery to the CNS without invasive neurosurgery.
The general approach is to use either a natural ligand to the receptor or a
monoclonal antibody to the receptor as the starting point to create the fusion protein
that can cross the blood±brain barrier. Because of species-specificity, the anti-human
insulin receptor mouse monoclonal antibody cannot be used in mice (Pardridge et al.,
1995), but a `humanized' chimeric antibody has been applied to primates and could
be used in humans (Coloma et al., 2000). RAP has been examined as a fusion protein
with the lysosomal enzymes alpha- L -iduronidase (deficient in Hurler/Scheie
syndrome) and acid alpha-glucosidase (deficient in Pompe syndrome) (Prince et
al., 2004), and was shown to retain its receptor binding properties.
For the transferrin receptor (TfR), the natural ligand is not effective for
blood±brain barrier delivery because the receptors are almost saturated at