Bioorthogonal
Labeling
Enables
In Situ Fluorescence
Imaging
of
Expressed
Gas Vesicle
Nanostructures
Erik Schrunk,
Przemysław
Dutka,
Robert
C. Hurt, Di Wu,
*
and Mikhail
G. Shapiro
*
Cite This:
Bioconjugate
Chem.
2024,
35, 333−339
Read
Online
ACCESS
Metrics
& More
Article
Recommendations
*
sı
Supporting
Information
ABSTRACT:
Gas
vesicles
(GVs)
are proteinaceous
nanostructures
that,
along
with
virus-like
particles,
encapsulins,
nanocages,
and
other
macromolecular
assemblies,
are being
developed
for potential
biomedical
applications.
To facilitate
such
development,
it would
be
valuable
to characterize
these
nanostructures’
subcellular
assembly
and localization.
However,
traditional
fluorescent
protein
fusions
are
not tolerated
by GVs’
primary
constituent
protein,
making
optical
microscopy
a challenge.
Here,
we
introduce
a method
for
fluorescently
visualizing
intracellular
GVs
using
the bioorthogonal
label
FlAsH,
which
becomes
fluorescent
upon
reaction
with
the six-
amino
acid
tetracysteine
(TC)
tag. We engineered
the GV subunit
protein,
GvpA,
to display
the TC tag and showed
that
GVs
bearing
TC-tagged
GvpA
can
be successfully
assembled
and
fluorescently
visualized
in HEK
293T
cells.
Importantly,
this was achieved
by replacing
only
a fraction
of GvpA
with
the tagged
version.
We used
fluorescence
images
of the tagged
GVs
to study
the GV size and
distance
distributions
within
these
cells.
This
bioorthogonal
and
fractional
labeling
approach
will
enable
research
to provide
a greater
understanding
of GVs
and
could
be adapted
to similar
proteinaceous
nanostructures.
■
INTRODUCTION
Gas
vesicles
(GVs)
are air-filled
protein
nanostructures
(
∼
85
nm
diameter,
∼
500
nm
length)
1
that
are entering
use
in
biomedical
applications
alongside
other
proteinaceous
macro-
molecular
assemblies
such
as encapsulins,
virus-like
particles,
and nanocages.
2
−
4
In particular,
GVs
have
recently
emerged
as
promising
agents
for biomolecular
ultrasound:
they
have
been
expressed
recombinantly
in both
bacterial
and mammalian
cells
and
have
been
used
as cavitation
nuclei,
5
ultrasonic
reporters
of cancer,
6
acoustic
actuators
for selective
cellular
manipu-
lation,
7
and
more.
8
−
11
These
GV-based
technologies
�
and
those
involving
other
macromolecular
complexes
�
could
benefit
from
the knowledge
of these
structures’
subcellular
localization,
as this knowledge
could
enable
the engineering
of
systems
targeted
to specific
organelles
or cellular
compart-
ments.
However,
there
are currently
no reported
methods
to
fluorescently
label
GVs
within
cells.
In part,
this is because
the
composition
of GVs
as assemblies
of small,
highly
conserved
subunit
proteins
makes
it difficult
for them
to accommodate
substantial
fused
functionalities,
such
as fluorescent
proteins.
Here,
we describe
a method
to optically
visualize
GVs
inside
cells
by genetically
modifying
the GV shell
protein
GvpA
with
the
tetracysteine
(TC)
motif,
allowing
the
GVs
to be
fluorescently
labeled
with
the bioorthogonal
FlAsH
reagent
for visualization
of their
subcellular
localization.
FlAsH
is a
membrane-permeant
fluorogenic
molecule
that
reacts
specifi-
cally
with
the TC tag (Cys-Cys-Pro-Gly-Cys-Cys)
and
which
turns
on fluorescence
upon
reaction.
12
−
15
We
sought
to
introduce
this tag into
the major
GV structural
protein,
GvpA,
such
that
expressed
intracellular
GVs
would
be able
to react
with
FlAsH
and
turn
on fluorescence.
We screened
for TC-
containing
GvpA
mutants
in bacteria
and,
once
we identified
a
suitable
variant,
expressed
TC-tagged
GVs
(“tcGVs”)
in HEK
293T
cells.
Using
these
tcGVs,
we were
able
to directly
visualize
the 3D distribution
of GVs
in the cell,
observing
that
they
tend
to form
clusters
in the cytosol.
Notably,
these
tcGVs
are produced
using
a mixture
of wild
type
(WT)
and modified
GvpA
genes,
where
only
a fraction
of the GvpA
subunits
are
tagged
with
the TC tag. In addition
to enabling
the study
of
GVs,
this
fractional
labeling
approach
could
inform
similar
studies
of other
genetically
encoded
protein
nanostructures.
■
RESULTS
AND
DISCUSSION
C-Terminus
of GvpA
Is Amenable
to Single
Sub-
stitutions
to Cysteine.
To engineer
tcGVs,
we sought
to
incorporate
the TC motif
into
the GV shell
protein
GvpA.
We
Received:
November
29, 2023
Revised:
January
21, 2024
Accepted:
January
23, 2024
Published:
February
12,
2024
Article
pubs.acs.org/bc
© 2024
The Authors.
Published
by
American
Chemical
Society
333
https://doi.org/10.1021/acs.bioconjchem.3c00518
Bioconjugate
Chem.
2024,
35, 333
−
339
This article is licensed under CC-BY 4.0
looked
to introduce
the motif
into
a region
of GvpA
that
faces
the GV
exterior
�
thereby
making
it accessible
to cytosolic
FlAsH
�
and
which
is tolerant
of mutations
to cysteine,
such
that
the introduction
of the TC tag does
not abrogate
GvpA
expression
and GV assembly.
To predict
which
region
of GvpA
would
best
accommodate
the TC tag, we looked
to structural
models
of GvpA.
1,16
In a random
mutagenesis
experiment,
the
C-terminus
of GvpA
was found
to be tolerant
of many
different
point
mutations
�
more
so than
any
other
region
of the
protein
�
suggesting
that
this
region
could
be the
most
amenable
to the substitution
of the six-amino
acid
TC tag.
1
Furthermore,
the C-terminus
of GvpA
is on the exterior-
facing
region
of the protein
1
(Figure
1a
−
c).
We
therefore
selected
the C-terminus
of GvpA
as our target
location.
Before
attempting
the substitution
of four
cysteines
into
GvpA,
we first
tested
the ability
of individual
positions
within
its C-terminus
to accommodate
single-Cys
mutations.
We
screened
mutants
in
Escherichia
coli
using
the
bARG
Ser
construct,
which
uses
GV
genes
derived
from
Serratia
sp.
39006,
including
the
GvpA
homologue
GvpA1
6
(
∼
92%
similarity
to GvpA,
sequence
alignment
in Figure
S1).
We
mutated
each
of the final
eight
amino
acids
in GvpA1
to Cys
Figure
1.
C-terminus
of the GV structural
protein
GvpA
is tolerant
of point
mutations
to cysteine.
(a) Illustration
of a single
GV.
(b) Atomic
model
of two
adjacent
“ribs”
composed
of GvpA
(PDB
8GBS).
1
Interior-
and exterior-facing
sides
of the GV shell
are indicated.
(c) Linear
and atomic
models
of a single
GvpA
molecule
with
α
helix
and
β
-strand
regions
are indicated.
1
The
C-terminal
region
of GvpA
is shown
as a dashed
box in the
linear
representation
of GvpA.
(d) Schematic
of the C-terminal
variants
of GvpA1
screened.
Each
red box represents
a point
mutation
to cysteine,
and each
blue
box represents
the amino
acid
in the WT
GvpA1.
(e) Graph
of opacity
for induced
and uninduced
bacterial
patches
transformed
with
plasmids
encoding
mutant
GV expression.
Colony
opacity
is indicative
of GV expression.
Representative
images
of induced
and uninduced
patches
are displayed
above
their
corresponding
columns
in the graph.
N
= 8 patches
per condition.
Patches
with
a plasmid
encoding
green
fluorescent
protein
(GFP)
expression
were
included
as a GV-negative
control.
Asterisks
represent
statistical
significance
by unpaired
t
-tests
(
****
:
p
< 0.0001,
ns: not significant).
Error
bars
represent
mean
±
SEM.
Bioconjugate
Chemistry
pubs.acs.org/bc
Article
https://doi.org/10.1021/acs.bioconjchem.3c00518
Bioconjugate
Chem.
2024,
35, 333
−
339
334
(Figure
1d,e),
and then
expressed
the mutant
GVs
in bacterial
patches
on Petri
dishes
containing
the inducer
arabinose.
We
then
measured
the opacity
of the patches
as a proxy
for GV
expression,
as GVs
scatter
visible
light.
17
−
19
We observed
GV
expression
in all mutants,
with
only
modest
reductions
at
positions
69
−
71
(Figure
1e),
and
concluded
that
the
C-
terminus
of GvpA1
could
tolerate
point
mutations
to Cys.
C-Terminus
of GvpA
Is Amenable
to Substitutions
to
the TC Tag.
With
the knowledge
that
each
amino
acid
in the
C-terminus
of GvpA1
could
be individually
substituted
to Cys,
we next
tested
multiposition
substitutions
to introduce
the TC
tag.
We
cloned
three
variants
of the GvpA1
gene
with
the
minimal
TC
tag
(Cys-Cys-Xxx-Yyy-Cys-Cys)
in all three
possible
C-terminal
positions
(Figure
2a),
leaving
the middle
two non-Cys
amino
acids
of the tag unchanged
relative
to WT
GvpA1
(denoted
by Xxx
and
Yyy)
to minimize
sequence
disruption.
We found
GV expression
in all cases
(Figure
2b),
although
at reduced
levels
compared
to that
of the WT.
The
variant
with
the TC tag at the most
C-terminal
position,
called
TC3,
had
the highest
opacity
(Figure
2b)
and
the healthiest
patch
morphology
(Figure
S2),
suggesting
that
this variant
was
the best
tolerated
by cells
expressing
the resulting
GVs.
As an
additional
test,
we converted
the two non-Cys
residues
in TC3
to Pro-Gly
to create
a full TC tag (Cys-Cys-Pro-Gly-Cys-Cys)
(Figure
2a) and noted
that
this mutant,
called
TC4,
expressed
GVs
as well
(Figure
2b).
We
concluded
that
the
optimal
positioning
of the TC tag in the C-terminus
of GvpA1
was at
the most
C-terminal
position.
TC-Tagged
GvpA
Can
Be Incorporated
into
GVs
Expressed
in Mammalian
Cells
and Imaged
Fluores-
cently
by FlAsH.
After
establishing
tcGV
expression
in
bacteria,
we translated
our approach
to mammalian
cells.
We
inserted
the TC tag into
the GvpA
of the mARG
construct
6
at
the same
location
as the best-performing
TC-tagged
GvpA1
from
our bacterial
screen
(TC3)
and called
the resulting
gene
“tcGvpA.”
We
transfected
human
HEK
293T
cells
with
mARG,
replacing
10, 20, 25, and
100%
of the WT
GvpA
(“wtGvpA”)
plasmid
with
the
tcGvpA
plasmid
in the
transfection
mixture
(molar
ratio),
and observed
GV formation
in all but the 100%
tcGvpA
condition
(Figures
3b and
S3).
This
shows
that
while
some
wtGvpA
is necessary
for GV
formation,
tcGvpA
expression
is well-tolerated
by mammalian
cells.
To determine
whether
tcGvpA
is incorporated
into
the
GVs,
we next
treated
the transfected
cells
with
FlAsH
and
found
that
FlAsH
readily
labeled
the GVs
in those
cells
(Figure
3b).
Control
cells
expressing
WT
GVs
(“wtGVs”)
without
tcGvpA
did not show
labeling
(Figure
3b).
This
demonstrated
that
tcGvpA
is incorporated
into
mammalian
GVs
when
coexpressed
with
wtGvpA
and
that
the
resulting
chimeric
tcGVs
can be labeled
intracellularly
by FlAsH.
tcGVs
expressed
in cells
transfected
with
10 and 25%
tcGvpA
could
be labeled
with
FlAsH
(Figure
S3)
similarly
to those
with
20%
tcGvpA
(Figure
3b),
suggesting
that
FlAsH
labeling
does
not require
a
very
precise
ratio
of tcGvpA
and
wtGvpA.
The
codelivery
of tcGvpA
and wtGvpA
for tcGV
expression
is an important
aspect
of these
findings,
as it demonstrates
that
not every
protein
subunit
of the GVs
needs
to be TC-tagged
for the GV
itself
to be sufficiently
reactive
toward
FlAsH.
Therefore,
the use of a mixture
of wtGvpA
and
tcGvpA
�
notably
with
a significant
majority
of the wtGvpA
gene
�
to
express
tcGVs
highlights
the utility
of this
approach
when
attempting
to fluorescently
label
proteinaceous
nanostructures
Figure
2.
C-terminus
of GvpA
tolerates
substitution
to the TC motifs
CC
−
CC
and CCPGCC.
(a) Schematic
of the C-terminal
TC mutants
of
GvpA1
screened.
Each
red “C”
box
represents
a point
mutation
to cysteine,
and
each
blue
box
represents
the amino
acid
in WT
GvpA1.
TC1
through
TC3
are the minimal
TC tag Cys-Cys-Xxx-Yyy-Cys-Cys,
while
TC4
is the full TC tag Cys-Cys-Pro-Gly-Cys-Cys.
(b) Graph
of opacity
for
induced
and
uninduced
bacterial
patches
transformed
with
plasmids
coding
for mutant
GV
expression.
Colony
opacity
is indicative
of GV
expression.
Representative
images
of induced
and uninduced
patches
displayed
above
their
corresponding
columns
in the graph.
N
= 8 patches
per
condition.
Patches
with
a plasmid
encoding
GFP
expression
were
included
as a GV-negative
control.
Asterisks
represent
statistical
significance
by
unpaired
t
-tests
(
****
:
p
< 0.0001,
***
:
p
< 0.001,
ns: not significant).
Error
bars
represent
mean
±
SEM.
Bioconjugate
Chemistry
pubs.acs.org/bc
Article
https://doi.org/10.1021/acs.bioconjchem.3c00518
Bioconjugate
Chem.
2024,
35, 333
−
339
335
within
cells;
even
if TC-tagging
a protein
subunit
is not ideal,
spiking
in a small
fraction
of TC-tagged
subunits
can
be
sufficient
for informative
labeling.
In addition,
the multimeric
nature
of GVs
contributes
to
their
high
contrast
labeling
with
FlAsH,
which
may
require
the
concentration
of the tagged
protein
to be higher
than
several
μ
M to overcome
background
fluorescence.
13
We estimate
that
the
concentration
of GvpA
within
a typical
imaging
voxel
containing
one
GV
is on the order
of 300
μ
M
(Supporting
Informtion
Methods
S1),
such
that
a tcGV
containing
only
a
few
percent
of FlAsH-tagged
GvpA
would
be sufficient
for
selective
imaging.
GVs Expressed
in HEK 293T
Cells
form
Clusters
in the
Cytosol.
After
demonstrating
that
intracellular
tcGVs
could
be
fluorescently
labeled
with
FlAsH,
we sought
to determine
their
subcellular
location
in mammalian
cells.
Knowledge
of the
localization
of GVs
within
cells
could
improve
our
under-
standing
of the biosynthesis
and
degradation
of these
protein
structures
and
inform
efforts
to target
GVs
to specific
organelles
or cellular
structures.
Although
phase
contrast
microscopy
can be used
to observe
the presence
of GVs
within
cells
due
to their
differential
refractive
index,
19
it does
not
provide
reliable
information
about
their
subcellular
localization
due to poor
depth
resolution.
On the other
hand,
imaging
the
GVs
using
confocal
microscopy,
now
enabled
by FlAsH
labeling,
would
allow
the
determination
of their
precise
subcellular
location
in 3D.
To demonstrate
this
capability,
we acquired
multiple
horizontal
planes
of the cells
expressing
tcGVs
labeled
with
FlAsH
and
simultaneously
stained
the
nucleus
with
DAPI
and
the
plasma
membrane
with
a
membrane-trafficked
fluorescent
protein
20
(Lck-mScarlet-I)
(Figure
4a; 3D renderings
of additional
cells
are in Figure
S4).
After
rendering
the cells
in 3D,
we found
that
GVs
form
distinct
clusters
within
the cell that
vary
considerably
in size,
ranging
from
the size of single
GVs
(around
0.003
μ
m
3
) to 20
μ
m
3
, with
an average
of 1.4
μ
m
3
and
a standard
deviation
of
2.9
μ
m
3
, and
together
occupy
between
0.21
and
1.1%
of the
total
cell volume.
We then
computed
the distances
between
the GV
clusters
to the nuclear
and
plasma
membranes
and
found
that
virtually
all GV clusters
were
not in direct
contact
with
the nucleus
or the plasma
membrane
and remain
localized
to the cytosol,
with
the average
GV cluster’s
center
being
2.9
±
2.2
μ
m away
from
the nucleus
and
2.6
±
0.92
μ
m away
from
the plasma
membrane
(Figure
4b-d).
■
CONCLUSIONS
In summary,
our
results
show
that
intracellularly
expressed
GVs
within
HEK
293T
cells
can be fluorescently
labeled
and
imaged
by confocal
microscopy
for the first
time.
The
C-
terminus
of GvpA
proved
to be quite
tolerant
of mutations
to
cysteine,
allowing
for the substitution
of the TC tag into
GvpA
without
a major
disruption
of the GV expression.
Also,
while
Figure
3.
tcGVs
can be successfully
expressed
and labeled
with
FlAsH
in HEK
293T
cells.
(a) Schematic
of an expressed
tcGV
cluster
becoming
fluorescent
with
FlAsH.
tcGVs
are composed
of wtGvpA
and tcGvpA.
After
addition
of FlAsH,
the tcGVs
become
fluorescent
as FlAsH
reacts
with
tcGvpA.
(b) Images
of wtGV
and tcGV
(20%
tcGvpA)
clusters
in fixed
HEK
293T
cells
(indicated
by arrows).
All GVs
are visible
under
bright-
field
imaging
(first
column),
but only
tcGVs
have
any FlAsH
signal
above
the background
(second
column).
The
bright-field/FlAsH
overlay
(third
column)
demonstrates
that
the strongest
FlAsH
signal
overlaps
with
the tcGV
clusters.
All scale
bars
are 5
μ
m.
GV-expressing
cells
are outlined
in
white.
Bioconjugate
Chemistry
pubs.acs.org/bc
Article
https://doi.org/10.1021/acs.bioconjchem.3c00518
Bioconjugate
Chem.
2024,
35, 333
−
339
336
HEK
293T
cells
could
not synthesize
GVs
made
entirely
of
tcGvpA,
we found
that
delivering
a mixture
of wtGvpA
and
tcGvpA
led to the expression
of FlAsH-labelable
tcGVs.
We
demonstrated
the utility
of FlAsH
labeling
of in situ-expressed
GVs
by studying
their
intracellular
distribution
with
higher
spatial
precision
than
ever
before
and found
that
they
generally
localize
to the
cytoplasm.
While
in this
study
we used
monocistronic
cotransfection
to deliver
the
tcGvpA
and
wtGvpA
genes,
we expect
that
this
labeling
approach
could
be improved
with
bicistronic
expression
where
the two
genes
are linked
using
the IRES
21
sequence
or the SEMPER
system
22
to provide
a finer
control
of their
relative
stoichiometry
within
the same
cell.
In addition,
future
studies
of the relative
protein
composition
of tcGvpA
and
wtGvpA
within
a tcGV
as a
function
of gene
ratio
could
inform
the
design
of these
multicistronic
systems.
We anticipate
that
our
approach
will
become
a tool
that
not only
furthers
the development
of GV-
based
technologies
but also
one
that
can
be applied
to the
study
of other
genetically
encoded
polymeric
proteinaceous
structures.
Figure
4.
Fluorescence
imaging
of tcGVs
elucidates
the size and spatial
distributions
of the GV clusters
in HEK
293T
cells.
(a) 3D rendering
(left)
of a fixed
tcGV-expressing
HEK
293T
cell reacted
with
FlAsH.
The
membrane
is shown
in red (Lck-mScarlet-I),
the nucleus
in blue
(DAPI),
and
tcGVs
in green
(FlAsH).
All scale
bars
are 5
μ
m.
Three
z
-slices
of the 3D rendering
are depicted
at right.
The
border
colors
of the
z
-slices
indicate
the heights
of the slices
within
the original
3D image:
z
= 0.4
μ
m (orange,
bottom),
z
= 2.2
μ
m (red,
middle),
and
z
= 5.6
μ
m (purple,
top)
above
the base
of the cell.
Corresponding
notches
in the 3D rendering
mark
the approximate
heights
of the
z
-slices.
(b)
−
(d)
Box-and-whisker
plots
of the
distributions
of the GV cluster
volumes
(b),
distances
to the nucleus
(c), and distances
to the membrane
(d).
N
= 6 cells
and 122 GV clusters
were
analyzed.
Bioconjugate
Chemistry
pubs.acs.org/bc
Article
https://doi.org/10.1021/acs.bioconjchem.3c00518
Bioconjugate
Chem.
2024,
35, 333
−
339
337