3.4 Enrichment of atezolizumab scFv-expressing yeast using
suspension versus adherent methods
In addition to characterizing yeast/mammalian cell binding interactions
on the biofloating platform, we explored whether a soluble
yeast/mammalian cell interaction system could be exploited for
enrichment of a specific clone spiked into a naïve scFv library.
Enrichment using a soluble yeast/mammalian cell interaction platform has
been conducted previously via FACS to enrich specific yeast clones from
a library of diversity 2.5×107.19 To
increase the throughput of enrichment and allow for selection of higher
diversity libraries, we combined MACS with a suspension cell-based
yeast/mammalian cell interaction platform. Lown et al previously used
magnetic bead-immobilized target-expressing mammalian cells to enrich
yeast displaying a target-specific fibronectin spiked into a pool of
non-displaying yeast.20 Resonating with this work, we
created a mock library consisting of a naïve yeast-displayed scFv
library22 of diversity 109 with
atezolizumab scFv-expressing yeast spiked in at 1:1000. Enrichment of
FLAG-tagged atezolizumab scFv-expressing yeast was tracked over 3 rounds
of selection using the suspension cell-based platform compared to an
adherent cell-based platforms.
For MACS-enabled suspension cell-based selections, mammalian cells were
biotinylated and immobilized on SA-coated magnetic beads. Yeast and
biotinylated mammalian cells were first co-incubated in suspension, and
SA-coated magnetic beads were then added prior to separation over a
magnetic column (Figure 6a). The first round of selections used
1010 yeast cells from the mock library (10-fold
oversampling of scFv diversity) and 107 mammalian
cells, for a 1000:1 yeast:mammalian cell ratio. In each round, a
negative selection was carried out using PD-L1- CHO-K1
cells to deplete the yeast library of scFvs binding to irrelevant
molecules on the mammalian cell surface. A positive selection was then
conducted against PD-L1+ CHO-K1 cells to enrich for
PD-L1-binding yeast-displayed scFvs. Flow cytometry analysis was
performed following each round to track specific scFv enrichment (Figure
6b). Rapid enrichment of atezolizumab scFv-expressing yeast was observed
over the first 2 rounds, with the first round enriching to 10% specific
yeast compared to 0.1% in the unsorted library. The library was almost
fully converged post-round 2 and plateaued post-round 3 (63% and 66%
atezolizumab scFv-expressing yeast, respectively).
For adherent cell-based selections, monolayers of mammalian cells were
co-incubated with the yeast-displayed mock library for directed
evolution following traditional screening methods (Figure 6c). The first
round of selections used 1010 yeast cells from the
mock library (10-fold oversampling of scFv diversity) and
1.8×107 mammalian cells for a 550:1 yeast:mammalian
cell ratio. For each round, negative selections were performed by
clearing binders to PD-L1- CHO-K1 cells prior to
positive selections that enriched for binders to
PD-L1+ CHO-K1 cells. Flow cytometry analysis was
conducted after each round to track scFv enrichment (Figure 6d). Minimal
enrichment of atezolizumab scFv-expressing yeast was detected after
round 1 (1.4% atezolizumab scFv-expressing yeast), whereas dramatic
enrichment was observed post-round 2 (52% atezolizumab scFv-expressing
yeast). Further library enrichment was observed after round 3, reaching
a final atezolizumab scFv-expressing yeast percentage of 71%. Although
comparable library enrichment was observed post-round 3 for the
suspension and adherent cell-based platforms, suspension cell-based
evolution led to faster enrichment compared to adherent cell-based
evolution.