SIGNAL CONVERTER PROTEINS - Fusion Protein Technologies for Biopharmaceuticals: Applications and Challenges - page 439

One especially attractive candidate was the Fas ligand
(FasL; CD95L; CD178):Fas (CD95; TNFRSF6) signaling
axis, which plays a major role in shaping the repertoires of
both Tand B cells and maintaining peripheral tolerance [17].
Various threads in the literature have pointed to veto-like
function for FasL when naturally expressed on certain cells
[18], and we [19] and others [13,14,18,20-22] have attested
to the veto function of APC with enforced FasL expression,
for example, in the induction of alloantigen-specific T-cell
tolerance and clonal deletion in vivo.
Gene transfer technology was pivotal at the earliest stages
of the AVC story. We established the original link between
CD8 and juxtacrine coinhibition back in 1989 through
antisense RNA-mediated inhibition of CD8 in T cells.
This first application of antisense RNA technology to
T cells enabled the demonstration that veto inhibition by
natural, alloantigen-presenting cloned CD8 þ T cells requires
the presence of CD8 on their surfaces [7]. Having estab-
lished that CD8 on cloned human T veto cells accounts for
their inhibitory activity, we proceeded to demonstrate that
certain allo-APC can be converted into AVC by neoexpress-
ing CD8 on their surfaces via sense gene transfer [8,23].

protein of interest can be produced as a GPI-modified,
protein paint variant. A series of subsequent studies vali-
dated the trans-signaling functionality of artificial GPI
proteins, once appended to cell surfaces [26-31].
Although elegant in conception, GPI protein paints posed
challenges when it came to scaling up protein production
since they necessitated purification from complex mem-
brane lysates. To bypass this inherent limitation, we looked
for alternative protein paints that could be produced more
expeditiously as soluble recombinant proteins. In this vein,
we devised a simple two-component protein transfer
method, which uses protein

Fc g 1 derivatives conjugated to
membrane-incorporating, chemically palmitated protein A.
Significantly, both components of these protein Fc g 1 :palmi-

tated protein A conjugates can be produced efficiently as
soluble recombinant proteins.
The experimental utility of GPI protein and protein

Fc g 1 :
palmitated protein A conjugate paints was established in a
series of immunobiological studies. For instance, painting of
peptide antigen-loaded MHC class I GPI complexes sensi-

tized cellular targets to cytotoxic T-cell-mediated destruc-
tion [27]. Protein transfer of such MHC:peptide antigen
complexes allows for optimization of APC antigenic dis-
plays, and in particular, titration of antigenic density, which
is significant given the importance of optimizing epitope
density on APC surfaces when seeking to stimulate naive
T cells [32].
Moving from first to second APC-to-T-cell signal, cos-
timulator

30.2.2 Protein Painting

While gene transfer offered the simplest experimental
approach for neoexpressing coinhibitors on APC surfaces,
protein transfer soon emerged as a more interesting option.
In principle, protein transfer offers a number of advantages
over gene transfer, in terms of controlling protein levels on
cell surfaces, co-expressing combinations of proteins, limit-
ing exogenous protein half-lives on modified cells, and
accommodating other therapeutic dictates. The first protein
transfer modalities applied to APC engineering were geared
toward coating cell surfaces ex vivo, which we referred to as
protein painting. Over time, the palette of fusion protein
paints was expanded.
The first fusion protein paints were genetically engi-
neered, glycosyl-phosphatidylinositol (GPI)-modified deriv-
atives of transmembrane proteins. Because GPI proteins are
amphiphilic, they remain soluble upon purification, even
after depletion of solubilizing detergents within cell mem-
brane extracts. In turn, this solubility feature makes it
possible to combine purified, detergent-depleted GPI pro-
teins with cells without concern for cell lysis, and when
combined in this way, the fusion proteins spontaneously
incorporate into the cell membranes. Thus, GPI proteins are
inherently protein paints. We [24] and others [25] originally
identified the GPI modification signal sequence within
decay-accelerating factor (DAF; CD55), a natural GPI-
anchored protein, and proceeded to demonstrate that
appending the DAF signal sequence to the carboxyl-termi-
nus of any target protein yields a GPI-modified variant of
that protein. Thus, using this chimerization approach, any

GPI derivatives were shown to retain costimulator
function despite GPI modification and could be used to
enhance tumor cell immunogenicity [28,33,34]. The ability
to finely titrate costimulator Fc g 1 :palmitated protein A con-

jugates onto APC surfaces permitted the first demonstration
of costimulator receptor activation thresholds [35], which
mirrored the T-cell receptor activation threshold phenome-
non that had already been firmly established in the literature.
In turn, the deciphering of quantitative and synergistic
features of costimulation, leveraging protein painting as
an experimental tool, set the stage for the use of costimulator
paints for active cancer immunotherapy. Specifically, our
group demonstrated that effective cellular cancer vaccines
can be generated by painting of tumor cells in situ with a
combination of four costimulator Fc g 1 :palmitated protein A

conjugates [36].
This substantiation of costimulator paints was paralleled
by analogous proof-of-principle validation of coinhibitor
paints. For CD8 qua inhibitor,

GPI [8], as well as a bispecific antibody, with one arm
holding soluble CD8 and the other latching on to MHC class
I resident on the AVCmembrane [12]. Our laboratory further
documented the feasibility of generating AVC by painting
cells with FasL (CD95L) [19]. In the process, we exploited
protein painting to shed light on how responding T cells
quantitatively

this included our CD8

integrate opposing (costimulatory and