16
In practice, these considerations already prove challenging in
the construction of allosterically regulated FP sensors by means of
colony-based screening format: e.g., considering Ca
2+
ions do not
readily diffuse across the Escherichia coli inner plasma membrane,
this means Ca
2+
-specific single-FP sensors need to be exported to
the periplasm to bind Ca
2+
[133]. In comparison, genetically
FRET-based FP sensors that are composed of two FP domains cannot be exported to the E. coli periplasm so that the plasma membrane needs to be selectively permeabilized to allow diffusion of
Ca
2+
[134]. Beyond throughput, it has also been realized that, for
optimal performance, synthetic protein switches are preferably
screened in their operating environment: e.g., Ca
2+
-responsive single- FP sensors for imaging applications in neurobiology are
increasingly screened in human cell lines such as HEK293 [135].
Similarly, HIF1-responsive cytosine deaminases originally screened
and optimized in Escherichia coli have subsequently been shown to
have off-target effects in human cells [136].
For higher throughput assays, fluorescent-activated cell sorting
(FACS) can boost the screening capacity by several orders of magnitude, while the outcome of a screening and selection experiment
can be holistically analyzed by means of next-generation sequencing. The key difference is that FACS-based selection procedures
assay the function associated with a single cell as opposed to an
average output of tens of millions of cells in multiwell plates or a
colony. This imposes a number of technical challenges on FACSbased screening procedures: Firstly, the activity of a synthetic protein switch needs to be assayed either inside or directly on the
surface of a cell. For actuators with catalytic functions, the substrate
and/or the product thus need to be retained inside or attached on
the surface of the cell. Secondly, asynchronies in cell growth and
division as well as bursts in transcription and translation render the
expression of recombinant proteins stochastic. As a result, expression levels and therefore the experimental signal usually vary by an
order of magnitude across individual cells. To some extent, varying
expression levels can be normalized over the size of a cell, e.g., by
normalizing over forward scatter, which is an indicator of cell size,
but does not allow for the same precision as multiwell plate screening assays. This poses challenges if the function of a synthetic protein switch only fractionally improves during every design-build-test
cycle. Furthermore, positive and negative selections need to be performed in a sequential fashion that provides less precise readouts
compared to side-by-side comparison in multiwell plates.
Unsurprisingly, the number of FACS-based screening procedures that have been successfully devised to construct synthetic
protein switches is limited. In one recent example, trehalosespecific single-FP sensors were engineered in Escherichia coli
by inserting allosteric, trehalose-specific PBPs into GFP [137].
4.4 Single-Cell
Screening Based
on FACS
Viktor Stein
In practice, these considerations already prove challenging in
the construction of allosterically regulated FP sensors by means of
colony-based screening format: e.g., considering Ca
2+
ions do not
readily diffuse across the Escherichia coli inner plasma membrane,
this means Ca
2+
-specific single-FP sensors need to be exported to
the periplasm to bind Ca
2+
[133]. In comparison, genetically
FRET-based FP sensors that are composed of two FP domains cannot be exported to the E. coli periplasm so that the plasma membrane needs to be selectively permeabilized to allow diffusion of
Ca
2+
[134]. Beyond throughput, it has also been realized that, for
optimal performance, synthetic protein switches are preferably
screened in their operating environment: e.g., Ca
2+
-responsive single- FP sensors for imaging applications in neurobiology are
increasingly screened in human cell lines such as HEK293 [135].
Similarly, HIF1-responsive cytosine deaminases originally screened
and optimized in Escherichia coli have subsequently been shown to
have off-target effects in human cells [136].
For higher throughput assays, fluorescent-activated cell sorting
(FACS) can boost the screening capacity by several orders of magnitude, while the outcome of a screening and selection experiment
can be holistically analyzed by means of next-generation sequencing. The key difference is that FACS-based selection procedures
assay the function associated with a single cell as opposed to an
average output of tens of millions of cells in multiwell plates or a
colony. This imposes a number of technical challenges on FACSbased screening procedures: Firstly, the activity of a synthetic protein switch needs to be assayed either inside or directly on the
surface of a cell. For actuators with catalytic functions, the substrate
and/or the product thus need to be retained inside or attached on
the surface of the cell. Secondly, asynchronies in cell growth and
division as well as bursts in transcription and translation render the
expression of recombinant proteins stochastic. As a result, expression levels and therefore the experimental signal usually vary by an
order of magnitude across individual cells. To some extent, varying
expression levels can be normalized over the size of a cell, e.g., by
normalizing over forward scatter, which is an indicator of cell size,
but does not allow for the same precision as multiwell plate screening assays. This poses challenges if the function of a synthetic protein switch only fractionally improves during every design-build-test
cycle. Furthermore, positive and negative selections need to be performed in a sequential fashion that provides less precise readouts
compared to side-by-side comparison in multiwell plates.
Unsurprisingly, the number of FACS-based screening procedures that have been successfully devised to construct synthetic
protein switches is limited. In one recent example, trehalosespecific single-FP sensors were engineered in Escherichia coli
by inserting allosteric, trehalose-specific PBPs into GFP [137].
4.4 Single-Cell
Screening Based
on FACS
Viktor Stein