Journal of the American Chemical Society
Article
Remarkably, the level of selectivity observed with VoLDeMo
for dopaminergic axons in the striatum is comparable to that of
the genetically encoded GFP labeling under the control of the
TH promoter, and to the small molecule, highly hydrophilic
FFN102 probe.32 A similarly high selectivity was observed with
VoLDeCAM targeting AMPAR-expressing terminals in the
cortex and striatum. Although the binding selectivity of VoLDe
probes is governed by a high affinity ligand, the dextran
polymer carrier plays a crucial role in the targeting process.
The high polarity of the dextran polymer compensates for the
lipophilicity of the VSD, while its flexibility (and to some
extent amphiphilic character) likely enables dynamic encapsu-
lation of the dye, leading to relatively low background staining
and high diffusion through the brain tissue. Such dynamic
behavior will be studied further to optimize the targeting
capacity and tissue diffusion of small polymers. Especially
dsRVF5-VoLDeMo (13) will benefit from further optimization
of the polymer carrier to improve the labeling specificity in
highly challenging environments with sparse distribution of
target cells and projections (such as noradrenergic axons in the
cortex; see Figure 5f).
Although state-of-the-art synthetic voltage sensors are
superior to GEVIs in many photophysical parameters such as
voltage sensitivity,69,70 photostability,71 and photon output,72
VoLDe probes cannot compete with overexpressed GEVIs or
hybrid chemo-genetic tools73 in recording single action
potentials. Here, we provide an initial examination of the
limits of optical voltage imaging with synthetic sensors
delivered to protein targets at native expression levels. Both
VoLDeMo and VoLDeCAM probes were successfully used to
record electrically evoked activity in striatal dopaminergic
axons and cortical AMPA-expressing neurons, respectively,
after signal averaging. In addition, VoLDeCAM was used to
record depolarization steps in individual cortical neuronal
bodies after averaging multiple trials. These results are, to our
knowledge, the first examples of optical voltage recordings
from specific cell types in brain tissue targeted via a pure
chemical approach, without any genetic manipulation. This
proof of concept study delineates the boundaries of targeted
voltage imaging without the use of genetic manipulation and
provides a baseline for future improvement of sensors, imaging
software, and hardware to enable high fidelity optical voltage
recordings in wild-type species. Specifically, the VoLDe
platform suffers primarily from the low photon budget defined
by the small number of protein targets to which the probe is
delivered. Consequently, fast voltage imaging at kHz rates in
brain tissue is currently impossible and compromises in spatial
resolution were necessary to provide sufficient signal (see
Figure 6b, where a single pixel comprises 30 μm2, losing the
resolution of individual ∼1 μm2 dopamine release sites).
Significantly brighter VSDs as well as improved instrumenta-
tion enabling more efficient photon collection and noise
attenuation would promote the practical applicability of
VoLDe probes. Nevertheless, nongenetic targeting of probes
is desirable especially in models of brain disorders which often
require multiple genetic modifications,74 making further
manipulation in order to install GEVIs challenging. Overex-
pressed GEVIs can further face a variety of issues including
toxicity or increased membrane capacitance.16
for studying many physiological processes such as behaviorally
relevant changes in circuit activity. For example, midbrain
dopamine cell bodies fire tonically at ∼4 Hz with super-
imposed bursts at ∼20 Hz that are triggered by environmental
stimuli.75 The ability to measure changes due to burst firing
promises a means to image the physiologically appropriate
“language” of these neurons, and to do so from axons, the sites
responsible for neurotransmitter release, is a goal that to date
has eluded the field.
The modular design of dextran-based VoLDe probes offers
many possibilities in terms of the potential applications and
impact. Regarding the payload, we first focused on VSDs, since
these sensors report on membrane potential changesa
fundamental process in the nervous systemand are excep-
tionally challenging to deliver to specific cells and neurites due
to their high lipophilicity. In this sense, VSDs represent a high
bar for testing the targeting efficacy of the designed molecular
delivery constructs. Our findings indicate that the dextran
platform promises a means to accommodate a wide variety of
lipophilic cargo molecules. Imaging modalities such as
photoacoustic (PA) imaging20−22 and short-wave infrared
(SWIR) fluorescence imaging18,19 also utilize lipophilic
synthetic imaging agents, while few genetically encoded PA
probes76,77 and no genetically encoded SWIR probes have
been developed to date. The concept of the VoLDe targeting
platform may therefore offer a promising solution to cell
specific targeting of PA and SWIR probes in the brain. Our
platform may also be applied to include actuators, signaling
modulators, and drugs, thus unlocking chemical and chemo-
genetic approaches to cell specific imaging and pharmacol-
ogy.78
In terms of the cell-targeting component of VoLDe probes,
dichloropane and PFQX can be substituted for a variety of
ligands, unlocking the possibilities of targeting phenotype/
pathology-associated cell surface molecules. Furthermore, the
VoLDe system will be readily adaptable to chemo-genetic
targeting using covalent tagging strategies (e.g., SNAP-tag,79
Halo-tag,80 or TMP-tag81) by substituting the ligand for a
suitable electrophile. In the current version of the VoLDeMo
system, targeting DAT or NET by the dichloropane ligand
perturbs the experimental system by blocking these native
neurotransmitter transporters. This is a general problem with
receptor ligands, as the agonist/antagonist activity of these
compounds perturbs the system under study. We tackled this
issue in the VoLDeCAM system by employing a recently
developed ligand-directed acyl imidazole covalent labeling
approach,43 which we show is compatible with the potentially
nucleophilic dextran polymer carrier. Although full “trace-
lessness” was not achieved in VoLDeCAM, partial retention of
native AMPAR function was achieved under conditions which
provided clear and specific labeling with our sensor. We
hypothesize that improvement of the traceless labeling strategy
will be possible by developing more specific functionalization
of the dextran polymer, which will not lead to stochastic
multivalent species but rather well-defined assemblies contain-
ing one ligand molecule per polymer carrier.
For in vivo applications, a major issue is the delivery of
exogenous diffusible chemical probes to the brain. In small
laboratory animals, local injection into brain parenchyma is
feasible.34 In the long term, less invasive methods are required
for research in animals and to pave the way toward clinical
applications: one of the major rationales for developing
chemically targeted imaging and pharmacological agents.
In the present recordings with the VoLDe probes in brain
tissue, we record voltage changes that occur more slowly than
single action potentials (∼1 ms), specifically in the range of 10
to 1000 ms. This temporal regime could nevertheless be useful
M
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX