Optical control of zebrafish behavior
Aristides B. Arrenberg, Filippo Del Bene, and Herwig Baier1
Department of Physiology, Program in Neuroscience, University of California, San Francisco, 1550 4th Street, San Francisco, CA 94158-2324
Edited by Lily Y. Jan, University of California, San Francisco, CA, and approved August 19, 2009 (received for review June 8, 2009)
Expression of halorhodopsin (NpHR), a light-driven microbial chlo-
ride pump, enables optical control of membrane potential and
reversible silencing of targeted neurons. We generated transgenic
zebrafish expressing enhanced NpHR under control of the Gal4/
UAS system. Electrophysiological recordings showed that eNpHR
stimulation effectively suppressed spiking of single neurons in
vivo. Applying light through thin optic fibers positioned above the
head of a semi-restrained zebrafish larva enabled us to target
groups of neurons and to simultaneously test the effect of their
silencing on behavior. The photostimulated volume of the ze-
brafish brain could be marked by subsequent photoconversion of
co-expressed Kaede or Dendra. These techniques were used to
localize swim command circuitry to a small hindbrain region, just
rostral to the commissura infima Halleri. The kinetics of the
hindbrain-generated swim command was investigated by com-
bined and separate photo-activation of NpHR and Channelrhodop-
sin-2 (ChR2), a light-gated cation channel, in the same neurons.
Together this ‘‘optogenetic toolkit’’ allows loss-of-function and
gain-of-function analyses of neural circuitry at high spatial and
temporal resolution in a behaving vertebrate.
central pattern generator ? channelrhodopsin ? Danio rerio ? reticulospinal
neural basis of behavior (1–8). The light-gated chloride pump
halorhodopsin (NpHR) from the archaebacterium Natronomonas
pharaonis has recently been introduced into neuroscience along
with enhanced derivatives (9–14) and enables superior temporal
and spatial control. Other light-controlled silencing methods are
being developed (15–17), but require covalent attachment of a
photo-switchable affinity label. NpHR silencing has been demon-
strated electrophysiologically (10, 12) and has been used to revers-
ibly paralyze Caenorhabditis elegans expressing NpHR in motor
peripheries (12). Despite its promise, however, NpHR has so far
found only limited applications for circuit analysis in vivo. In this
study, we have devised a versatile and cost-effective optical stim-
ulation strategy for manipulation of animal behavior with this tool.
Zebrafish are ideal models for testing and applying light-
controlled channels and pumps in vertebrates, since they are
translucent and display a number of quantifiable behaviors during
et al. (22) used a re-engineered, light-gated glutamate receptor
(LiGluR), to induce swimming by photostimulation of a rare type
responses by activating ChR2 in single zebrafish mechanosensory
cells. The adaptation of the Gal4/UAS method from Drosophila
melanogaster (24) to zebrafish enables targeting transgene expres-
sion to specific brain areas and cell types (25–29) and will further
Here we report on the generation of UAS:NpHR transgenic
zebrafish lines. Using a Gal4 line that drives NpHR broadly in
neurons, we show that enhanced NpHR (eNpHR) is targeted
efficiently to the surface of neurons in vivo and mediates light-
induced suppression of spikes. We then use a non-invasive fiber
echnology for the inactivation of specific neurons within an
otherwise intact circuit promises to accelerate research into the
optics approach to stimulate small (ca. 30 ?m) CNS areas, while
simultaneously monitoring the fish’s behavioral responses. We
combine NpHR silencing with ChR2-mediated excitation, to iden-
tify a critical role for a small cell group in the caudal hindbrain in
the control of forward swimming. The ability to selectively silence
neurons in vivo with precise temporal and spatial control is likely
and neuronal plasticity.
Enhanced Halorhodopsin (eNpHR) Is Targeted to the Cell Surface of
Zebrafish Neurons In Vivo. Different versions of NpHR have been
reported to vary in their intracellular distribution and surface
four transgenic lines, UAS:NpHR-eYFP, UAS:NpHR-mCherry,
UAS:eNpHR-eYFP, and UAS:eNpHR-mCherry. Transgenic fish
were crossed to carriers from the enhancer trap line Gal4s1101t,
which expresses the transcriptional activator Gal4-VP16 broadly in
most neurons (29) (Fig. 1A). eNpHR was engineered to have
expected, eNpHR-eYFP proteins (Fig. 1B) trafficked to the cell
surface and did not form the intracellular blebs that were observed
for NpHR-eYFP (Fig. S1) and NpHR-mCherry (Fig. 1B).
Two lines (NpHR-mCherry and eNpHR-eYFP) were further
investigated for surface localization by co-expressing membrane-
targeted fluorophores, Dendra-kras or mCherry-kras, respectively
(Fig. 1C). While most of the NpHR-mCherry protein remained
intracellular, a fraction co-localized with Dendra-kras at the cell
surface. In contrast, virtually all eNpHR-eYFP signal was co-
localized with mCherry-kras at the plasma membrane. Membrane
cell morphologies including neurites (Fig. S2 shows radial glial cells
in the optic tectum). For unknown reasons, fusion proteins con-
taining the identical opsin but different fluorescent tags were
sometimes distributed in apparently different cellular compart-
ments (Fig. S1 and Fig. 1B). Together, eNpHR-eYFP appeared to
be superior, as it combined excellent surface localization with
effectiveness in suppressing spikes (see below).
Single-Unit Electrophysiology Confirms Silencing in Vivo. To deter-
loose-patch recordings in hindbrain neurons. To activate NpHR, a
bandpass filter centered on its activation spectrum maximum (HQ
585/70, Fig. 2A) was used. The neuron in Fig. 2B (top two traces)
was silenced during illumination periods, and no spikes were
generated. After stimulation, the cell resumed firing at a rate
comparable to the average firing rate before stimulation. This
experiment suggested that NpHR was an effective and reversible
contributed new reagents; A.B.A. analyzed data; and A.B.A. and H.B. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
October 20, 2009 ?
vol. 106 ?
silencer of neuronal activity in larval zebrafish. Conversely, the
activation of the light-gated cation channel ChR2 (ChR2-H134R)
in Gal4s1101t; UAS:ChR2-eYFP animals induced firing rates up to
130 Hz for many seconds (Fig. 2B bottom trace).
We next assessed the magnitude of the silencing effect across the
population of recorded hindbrain neurons. NpHR expressing cells
had much lower firing rates during illumination (F2) than without
illumination (F1; Fig. 2C). For quantification, the distributions of
firing rate ratios (F2/F1) of NpHR expressing cells and NpHR
non-expressing cells (wild-type, Fig. 2D and Fig. S3) were com-
pared. The silencing effect in NpHR-expressing cells was highly
significant (P ? 0.0001 for both eNpHR-eYFP and NpHR-
mCherry, Ranksum and KS test). Furthermore, the median firing
rate ratio (F2/F1) was 0.2 for both eNpHR-eYFP and NpHR-
mCherry (see Fig. S4 for the light intensity dependence of the
effect). This means that NpHR photostimulation suppressed, on
average, 80% of all spontaneous spikes. A fraction of cells (?15%)
were not significantly inactivated; very few even increased their
spike rate (permutation test with alpha ? 0.01, Fig. S5).
In control experiments with wildtype cells, we noted that illu-
mination had a small effect on firing rate in 26% (8/31) of the cells
(permutation test, P ? 0.01). These light responses could be due to
synaptic input from the visual system. We therefore recorded from
form (30), and found that the fraction of cells reproducibly respon-
sive to light decreased to 14% (3/23; Fig. 2D). Since the hindbrain
of lakritz/atoh7 mutants receives no input from the retina, the
significant firing rate change in the few light-responsive cells could
stem from inputs from the pineal organ, from intrinsic photosen-
sitivity of the recorded cells, or from a thermal effect of the
Rebound from Inhibition. The Gal4s1101ttransgene drives NpHR-
mCherry broadly in CNS neurons. We therefore expected an
extensive and almost instantaneous inhibition of movements upon
(627 nm), larvae frequently stopped moving and lost coordination
(Movie S1 for NpHR-expressing animals and Movie S2 for their
non-expressing siblings). Strikingly, the offset of illumination trig-
across the population (Movie S1). A similar response could be
elicited in agarose-embedded larvae whose tails were left free to
of the induced tail movement were in the range of those previously
reported for routine swim scoots (31).
Since the forward swims were induced by switching the illumi-
nation off, it seemed likely that cells rebounding from inhibition
triggered the behavior. Indeed, a heightened firing rate (up to
10-fold) in the first second after light offset was observed in the
majority (60%) of the recorded hindbrain cells (Fig. 3B; the
remaining 40% of cells did not show detectable changes). While we
of the behavioral response in the same animal, recordings similar
to those in Fig. 2 showed that rebound spiking could start within
tens of milliseconds after light offset, which is in agreement with
the fast NpHR kinetics (12) (?OFF ? 40 ms). The forward
swimming could last up to 5 s, unless the illumination was turned
back on, which again blocked all swimming movements (see
below; Movie S3).
Fiber Optics Enables Targeted, High-Resolution Silencing of Neuronal
Activity. To investigate which neural structure was triggering the
rebound swim behavior, we decided to map its origin in the
scanning.’’ First, we needed to determine the spatial resolution
achievable with this method. For local activation of NpHR we used
multimode optical fibers that were coupled to lasers. To verify that
the light was confined to a small cone exiting the fiber we moved
the fiber (10 or 50 ?m in diameter) across a recorded cell and
measured the firing rate ratio F2/F1 for each position (Fig. 3 D and
E). In three of eight cells recorded in this manner, silencing only
cells, firing rates were influenced by illuminating neighboring
positions away from the cell bodies, possibly due to local intercon-
the tip of the recording pipette had no influence on the firing rate
of the cell in Fig. 3E. This observation suggests that non-invasive
fiber optics used for silencing has a spatial resolution of 30 ?m or
better in vivo.
To determine the approximate penetration depth and the extent
of light scatter within the tissue as the stimulus light passes through
the tissue, we used a photoconvertible fluorescent protein. Fish
carrying both UAS:Dendra-kras and Gal4s1101texpressed the mem-
brane-targeted Dendra protein in most neurons. These larvae were
embedded in agarose and illuminated with an optic fiber (50 ?m)
coupled to a blue laser for several minutes. Selective-plane illumi-
nation microscopy (32) (SPIM) was used to image the distribution
of green-to-red photoconverted Dendra in the hindbrain from two
orthogonal directions (Fig. S6). As expected, red cells formed a
narrow column whose width was 50 ?m at the surface and widened
slightly with increasing depth. The divergence angle of the cone of
converted cells matched the theoretical value for low numerical-
aperture optic fibers (12°, NA ? 0.22). This experiment shows that
blue light emitted from thin optic fibers penetrates the entire depth
of the zebrafish brain with little scattering.
5 dpf animals. (B) Expression of NpHR in Gal4s1101t; UAS:(e)NpHR-XFP animals.
(i and ii) NpHR-mCherry and eNpHR-eYFP, respectively. Intracellular blebs
(arrows) are labeled in (i). (C) Surface targeting of NpHR. Co-expression of
kras; UAS:NpHR-mCherry animals (i) reveals suboptimal surface targeting.
Co-expression of eNpHR-eYFP and membrane-bound mCherry-kras in
Gal4s1013t; UAS:eNpHR-eYFP; UAS:mCherry-kras animals, shows complete sur-
face targeting of eNpHR-eYFP. [Scale bars, 100 ?m in (A); 5 ?m in (B and C).]
Expression of NpHR in zebrafish. (A) Expression pattern of Gal4s1101t;
Arrenberg et al.PNAS ?
October 20, 2009 ?
vol. 106 ?
no. 42 ?
NpHR-Assisted Optogenetic Scanning Identifies Swim Command Neu-
rons in the Hindbrain. Having shown that photostimulation with
optic fibers is spatially precise, we set out to map the rebound
swimming behavior to particular locations in the zebrafish CNS.
spinal cord and activate an array of segmentally repeated neural
circuits, called central pattern generators (CPGs), which drive
muscles via motoneurons (33). We placed a thick fiber (200 ?m in
diameter) on the dorsal surface of the animal and photostimulated
various positions in the brain (Fig. 3C). The probability of evoking
swimming was maximal after offset of hindbrain illumination [Fig.
4A, P ? 1.0, n ? 17 trials, 95% confidence interval (0.82 1.00)] and
small (P ? 0.3) after forebrain, midbrain, or spinal cord were
illuminated. Non-expressing siblings did not show swimming be-
havior correlated to the light pulse (P ? 0.0, n ? 23 trials, 95%
confidence interval [0 0.14]). The difference between expressors
and controls was significant (P ? 0.01, z-test for proportions).
neurons in the hindbrain with greater precision (Fig. 4B). The
maximum response probability (P ? 0.8) was seen following
the midline of the animal. We used the UAS:Kaede transgene and
a UV laser to label the illumination position via photoconversion
the behavior with a probability of P ? 0.8 were located in the
caudal-most part of the reticular formation (34, 35), just rostral to
the commissura infima Halleri (36, 37). This commissural tract
defines the division between spinal cord and hindbrain. At the
caudally adjacent positions, it was also possible to sometimes elicit
swimming behavior (P ? 0.3–0.7), but regions further down the
spinal cord or more rostral in the brain did not affect swimming
(P ? 0.0). The caudal hindbrain contains previously identified
groups of neurons that extend axons into the spinal cord (38, 39)
‘‘Reversible Spinalization’’ Can Be Used as a Method to Study De-
scending Swim Commands. Spinal cord CPGs are able to generate
normal locomotor muscle activations following surgical lesion of
the descending connections (spinalization) (40–42). In our exper-
iment, these connections were left physically intact, but reversibly
10 s.) (C) Scatter plot of the firing rate during stimulation (F2) vs. the firing rate without stimulation (F1). NpHR-mCherry cells had a reduced firing rate F2, while cells
(with/without stimulation) in cells from different lines. Cells from NpHR lines were silenced (curve shift to the left) and cells from the ChR2 line were facilitated (shift
to the right). The differences between wild-type and eNpHR-eYFP or NpHR-mCherry were highly significant (P ? 0.0001, KS test and Ranksum test).
tion is plotted over time. (B) The recorded cell showed an above-average firing
rate directly after the stimulation (yellow shaded box) was stopped (arrow). (C)
Fiber-optic setup for the mapping of the locomotion phenotype. An optic fiber
was coupled to a laser and placed over the head-restrained larva with a micro-
fibers. (D) The optic fiber (10-?m inner core diameter) was moved across the
recorded cell in five steps. False-colored stripes in the image represent the
scattered light that was detected by the camera when light was sent through
theoretical divergence angle of the fiber (8.5°, white solid lines). Dashed lines
of a single cell is plotted for different fiber positions. The cell was only silenced
when it was directly illuminated, which demonstrated the precise light applica-
tion with optic fibers.
Use of fiber optics to control locomotion with NpHR. (A) Release from
www.pnas.org?cgi?doi?10.1073?pnas.0906252106 Arrenberg et al.
inactivated. This allowed us to ask to what extent swimming, once
that rebound-evoked swimming behavior could be blocked by
turning the NpHR stimulation back on during or after buildup of
activity in the caudal hindbrain (n ? 5 animals; Fig. 5A). This
indicates that there is a defined time during which the swimming
command is sensitive to hindbrain perturbations.
To begin to investigate the kinetics of this control, we used an
animal (5 dpf) with long-lasting induced forward swims (3.6 ? 0.3 s
SEM). The latency of the first tail undulation following light-offset
was highly reproducible (267 ? 14.0 ms, standard deviation, n ?
C). With longer time intervals, the animal initially started swim-
ming. However, the amplitude (Fig. 5B and Movie S4) and the
number of tail undulations were dependent on the re-illumination
time point. The later the animal was illuminated, the stronger and
not result in more vigorous swimming movements. Resetting CPG
activity by hindbrain inputs, once set in motion, appeared to be
almost immediate for low locomotor activity and took a few
hundreds of milliseconds (?300 ms) for higher levels of locomotor
activity (Fig. 5 B and C).
ChR2, Co-Expressed with NpHR and Activated Separately, Can Be Used
to Control Locomotor Behavior. We tested whether ChR2 activation
We found that the latency and amplitude of locomotion were
correlated with the magnitude of the induced depolarization. For
high light intensities, the latency was 10 ms; for very low intensities,
the latency was ?1,000 ms. These results suggest that the amount
of activity in the caudal hindbrain specifies the intensity of the
same cells and activated separately. The activation spectra of ChR2
and NpHR (Fig. 6B, inset) partially overlap, and we tested whether
independent activation was possible in triple-transgenic Gal4s1101t;
UAS:NpHR-mCherry; UAS:ChR2-eYFP animals (Fig. 6). Since
ChR2 is activatable with much lower light intensities than NpHR,
we used a red laser (633 nm) instead of a green one for NpHR
activation (Fig. 6B, inset). Using the locomotor behavior described
above, we found that ChR2 induced locomotion and NpHR-
rebound induced locomotion could be triggered independently by
using medium intensity blue light (488 nm, 30 mW/mm2) and high
intensity red light (633 nm, 710 mW/mm2), respectively. Further-
more, the ChR2-evoked locomotion could be interrupted by addi-
very reliable for all stimulation protocols, as the probability of
experiments confirmed that ChR2 and NpHR can be activated
independently in the same neurons.
We have generated zebrafish lines for four NpHR variants, making
system, although NpHR-mCherry was equally effective in silencing
of neurons. Our electrophysiological experiments demonstrated
effective and rapid suppression of spikes in NpHR-expressing cells,
variable (see Fig. 2). This could be due to differences in expression
level or in intensity of illumination, which varies somewhat with
position in the tissue. Another source of variability is the intrinsic
connectivity of the network. The relative weights of excitatory and
inhibitory synaptic connections to a recorded cell should impact its
from NpHR silencing (A) The probability of observing a forward swim is plotted
versus the position of stimulation (200-?m optic fiber) in 3 dpf zebrafish. Mean
probabilities of NpHR-mCherry expressors (black) and siblings (gray, only third
position) are shown with 95% confidence intervals (Wilson score interval for
binomial distributions). (B) The hindbrain region in (A) (asterisk) was mapped in
detail with a 50-?m fiber. The probability was maximal in a small area along the
(B) (n ? 7 animals). (C) The phenotype was mapped in an animal transgenic for
Gal4s1101t; UAS:NpHR-mCherry; UAS:Kaede. The region where the animal re-
sponded reliably was illuminated with UV light to photo-convert Kaede. (D)
a few cells in every optical section. The maximal response probability was found
in the region of the commissura infima Halleri (CI, dashed lines), which demar-
cates the border between spinal cord (SC) and hindbrain (HB). The upper right
in (C) and 10 ?m in (D).]
Mapping of the locomotion phenotype that was induced by rebound
induced forward swimming was blocked by reactivating NpHR. Top: Without
reactivation of NpHR, the animal (5 dpf) started to move 267 ? 3.5 ms (SEM)
Bottom: When the animal was re-illuminated after 248 ms, the locomotion
duration of the tail contractions depended on the re-illumination time point.
Early re-illuminations permitted only smaller amplitudes (see also Movie S4).
Images are an average-z-projection of four consecutive minimum-intensity-
z-projections. (C) The time difference between the cessation of locomotion
and the re-illumination is plotted versus the re-illumination time point. Inter-
vals shorter than 190 ms never permitted tail contractions. Between 190 ms
value of 263 ? 14 ms (SEM) at 300 ms. Trials in (B) are labeled with an asterisk
in (C). (D) In ChR2 expressing animals, the response latency depended on the
illumination intensity. Stronger illuminations elicited shorter latencies. Trials
from a single 3-dpf animal are plotted.
Kinetics of the rebound-evoked swim command. (A) The NpHR
Arrenberg et al.PNAS ?
October 20, 2009 ?
vol. 106 ?
no. 42 ?
if overall reduction of the cell’s inputs leads to a net increase in
a random network of neurons receiving external inputs of varying
magnitude (Fig. S9). Global hyperpolarization of all network
neurons produced variable changes in firing rates, suggesting that
intrinsic network properties alone should produce a broad range of
A requirement for the dissection of behavioral circuitry with
NpHR is the spatial restriction of activated NpHR. This can in
principle be achieved by genetic targeting (e.g., using cell-type
specific Gal4 lines) (25, 27–29) in combination with global illumi-
defined cell population. Moreover, global illumination inadver-
tently activates the visual and other photosensitive systems and
influences behavior. We therefore relied on optical targeting to
restrict the light-activated volume. This was done initially in a
broadly expressing line, but could be combined in the future with
specific Gal4 drivers to accomplish additional levels of spatial
control. Our approach is particularly useful for behavioral studies,
because, (i) the position of the light beam can be varied indepen-
provide great flexibility in behavioral setups; (iii) light application
the size of the brain region that is stimulated can be large or small
depending on the fiber diameter (10 to 1,500 ?m). Notably,
alternative light application methods with superior spatial resolu-
tion (43) exist, for example, digital micromirror devices (44) and
laser-scanning units (45) and will be used to expand this approach
We used Kaede and Dendra to mark the photostimulated cells
and estimate light scatter and penetration depth. Light delivered by
the optic fiber photoconverted a narrow column, whose diameter
increased with depth and extended several 100 ?m into the brain
the theoretical value for low numerical-aperture optic fibers (12°,
is that the photoconverted volume is not identical to the photo-
stimulated volume, because Kaede and NpHR require different
wavelengths and intensities. The radial spread of light application,
however, will be slightly larger for Kaede than for NpHR, because
scattering is stronger for shorter wavelengths, for example, by a
factor of 5 for 405 compared to 633 nm (Mie theory of scattering).
We calculated that the divergence angle difference is small (?1%
for 405 compared to 633 nm). The photoconverted volume is
therefore expected to provide a very close, upper-bound estimate
for the radial extent of photostimulation.
To demonstrate the utility of this toolkit for functional neuro-
thin optic fiber was positioned above the head of a semirestrained
while simultaneously monitoring the animal’s behavior. Using this
method, we identified a small region in the caudal hindbrain, just
rostral to the commissura infima Halleri, that initiates a locomotor
command when released from inhibition. The swim-inducing re-
gion identified by this approach contains bilaterally symmetric
groups of neurons called IC, CC, and T that project axons into the
spinal cord (38, 39). Because photostimulation of the region
effect, we consider it extremely unlikely that the behavior was
triggered by activity of en passant axons originating from more
rostrally positioned cells. Identified reticulospinal cells have been
assigned functions in specific behaviors, including escape, turning,
and pursuit of prey (46–51). Here we add to this emerging map of
behavioral functions in the zebrafish reticular formation a role for
caudal cells in the control of forward swimming. The observation
that activity in the brain region containing these neurons is neces-
locomotor command neurons.
We made use of the temporal resolution of NpHR and ChR2 to
study the kinetics of rebound-induced swimming. NpHR-assisted
in the spinal cord are under tight control of descending projections.
After light offset, swimming starts with a delay of a little less than
300 ms (267 ms in the animal shown in Results). This latency is the
composite of several processes that occur in the hindbrain, spinal
cord, and muscle. In the hindbrain, NpHR needs to cease its
chloride pump activity, the neurons need to rebound from hyper-
polarization and generate spikes, and enough neurons need to be
recruited to form a ‘‘swim command.’’ To determine the sum of all
back on after various intervals, while monitoring the animal’s
behavior. We found that up to a certain interval length (190 ms in
the tested animal), the swim command appeared to be completely
dependent on hindbrain activity; i.e., the fish never started swim-
ming when the light was turned back on within 190 ms. After this
time, for intervals of 190–300 ms, enough activity appeared to be
building up to drive the CPGs, that is, the fish swam. Longer
intervals (?300 ms) did not lead to more vigorous swimming. This
300 ms and for the first 200 ms is completely hindbrain-dependent.
Animals transgenic for Gal4s1101t; UAS:NpHR-mCherry; UAS:ChR2-eYFP were
illuminated with red or blue light, or both. Illumination with blue light
When red light was followed by blue and red light, no locomotion was
induced (iii). In (iv), blue light evoked locomotion was blocked with red light
three consecutive times. (B) Experiments in (A) were highly reproducible. The
probability of locomotion is plotted for four different genotypes (divided by
vertical lines) and four different stimulation protocols. Error bars are 95%
confidence intervals for binomial distributions (Wilson score, n ? 238 trials in
total, n ? 2 animals for each genotype). For triple transgenic animals, the
with (A iii) and without (A i) red illumination (P ? 0.0004, z-test for propor-
tions). The inset shows the activation spectra of ChR2 and NpHR (after ref. 12)
and the laser lines for ChR2 (488 nm) and NpHR (633 nm).
NpHR and ChR2 can be combined and activated separately. (A)
www.pnas.org?cgi?doi?10.1073?pnas.0906252106Arrenberg et al.
The fact that the fish start swimming less than 300 ms after light
offset (of which a large part, roughly 200 ms, is attributable to
caudal hindbrain with ChR2—here swimming could be elicited
continue for some time after hindbrain silencing. Indeed, the time
lag between re-illumination time and the cessation of the behavior
not be explained by the NpHR rise time alone (12) (?ON? 36 ms).
We interpret this finding to indicate that, while the CPGs do not
require continuous excitation, their autonomous activity is not
sustained for longer than a fraction of a second in the absence of
Our preliminary analysis of the topography and kinetics of
information transfer from hindbrain to CPGs was confirmed by
optic fiber stimulation of ChR2, alone or in combination with
NpHR. First, as expected, activation of the caudal-most hindbrain
with ChR2 elicited vigorous swimming, which could be blocked by
wavelength-separated activation of NpHR in the same neurons.
light intensity and was as short as 10 ms for the brightest light. In
summary, we have shown here, using electrophysiology and behav-
ior, that the microbial opsins NpHR and ChR2 are potent and
versatile tools for the dissection of circuit function in an intact
vertebrate nervous system.
Materials and Methods
The stimulation protocol consisted of alternated (no stimulus, with stimulus)
repeated trials lasting 5–20 min in total.
Optic Fiber Setup. For NpHR activation, a laser system of a confocal microscope
AOTF. The single mode fiber of the AOTF was coupled into multimode fibers
using an FC-to-SMA adapter. Fibers were prepared according to ref (52) and the
maximal output intensities were 58 mW/mm2(200 ?m fiber, 633 nm) and 712
mW/mm2(50 ?m fiber).
Additional Methods. Descriptions of plasmids, transgenic fish lines, electrophys-
iology, modeling, optic fiber setup, swimming behavior, and statistics are avail-
able in the SI Text.
of California, Berkeley, CA) for providing us with the electrophysiological setup
used in this study; Karl Deisseroth (Stanford University, Palo Alto, CA) for the
halorhodopsin plasmid; Jan Huisken for helping with the laser setup and per-
the linear firing rate model; and German Sumbre, Tod R. Thiele, and Estuardo
Robles for comments on the manuscript. A.B.A. was supported by a Boehringer-
Ingelheim Foundation fellowship (B.I.F.). This work was supported by National
medicine Development Center for the Optical Control of Biological Function
Grant PN2 EY018241, the David and Lucile Packard Foundation, a Sandler Op-
portunity Award, and the Byers Award for Basic Science (H.B.).
1. Lerchner W, et al. (2007) Reversible silencing of neuronal excitability in behaving mice
by a genetically targeted, ivermectin-gated Cl- channel. Neuron 54:35–49.
2. Slimko EM, McKinney S, Anderson DJ, Davidson N, Lester HA (2002) Selective electrical
silencing of mammalian neurons in vitro by the use of invertebrate ligand-gated
chloride channels. J Neurosci 22:7373–7379.
by selective suppression of activity in individual neurons. Nature 420:414–418.
4. Hua JY, Smear MC, Baier H, Smith SJ (2005) Regulation of axon growth in vivo by
activity-based competition. Nature 434:1022–1026.
5. Hong ST, et al. (2008) cAMP signalling in mushroom bodies modulates temperature
preference behavior in Drosophila. Nature 454:771–775.
6. Pitman JL, McGill JJ, Keegan KP, Allada R (2006) A dynamic role for the mushroom
bodies in promoting sleep in Drosophila. Nature 441:753–756.
7. van der Bliek AM, Meyerowitz EM (1991) Dynamin-like protein encoded by the
Drosophila shibire gene associated with vesicular traffic. Nature 351:411–414.
8. Sorensen JB, et al. (2002) The SNARE protein SNAP-25 is linked to fast calcium trigger-
ing of exocytosis. Proc Natl Acad Sci USA 99:1627–1632.
9. Gradinaru V, Thompson KR, Deisseroth K (2008) eNpHR: A Natronomonas halorho-
dopsin enhanced for optogenetic applications. Brain Cell Biol 36:129–139.
10. Zhao S, et al. (2008) Improved expression of halorhodopsin for light-induced silencing
of neuronal activity. Brain Cell Biol 36:141–154.
11. Han X, Boyden ES (2007) Multiple-color optical activation, silencing, and desynchro-
nization of neural activity, with single-spike temporal resolution. PLoS ONE 2:e299.
12. Zhang F, et al. (2007) Multimodal fast optical interrogation of neural circuitry. Nature
13. Schobert B, Lanyi JK (1982) Halorhodopsin is a light-driven chloride pump. J Biol Chem
14. Kalaidzidis IV, Kalaidzidis YL, Kaulen AD (1998) Flash-induced voltage changes in
halorhodopsin from Natronobacterium pharaonis. FEBS Lett 427:59–63.
excitability. Nat Methods 5:331–338.
16. Banghart M, Borges K, Isacoff E, Trauner D, Kramer RH (2004) Light-activated ion
channels for remote control of neuronal firing. Nat Neurosci 7:1381–1386.
17. Szobota S, et al. (2007) Remote control of neuronal activity with a light-gated gluta-
mate receptor. Neuron 54:535–545.
18. Borla MA, Palecek B, Budick S, O’Malley DM (2002) Prey capture by larval zebrafish:
Evidence for fine axial motor control. Brain Behav Evol 60:207–229.
19. Orger MB, et al. (2004) Behavioral screening assays in zebrafish. Methods Cell Biol
larval zebrafish. J Neurophysiol 93:3177–3188.
21. Burgess HA, Granato M (2007) Sensorimotor gating in larval zebrafish. J Neurosci
spinal cord. Nature 461:407–410.
and generating dominant phenotypes. Development 118:401–415.
25. Davison JM, et al. (2007) Transactivation from Gal4-VP16 transgenic insertions for
tissue-specific cell labeling and ablation in zebrafish. Dev Biol 304:811–824.
26. Scheer N, Campos-Ortega JA (1999) Use of the Gal4-UAS technique for targeted gene
expression in the zebrafish. Mech Dev 80:153–158.
27. Asakawa K, et al. (2008) Genetic dissection of neural circuits by Tol2 transposon-
mediated Gal4 gene and enhancer trapping in zebrafish. Proc Natl Acad Sci USA
28. Asakawa K, Kawakami K (2008) Targeted gene expression by the Gal4-UAS system in
zebrafish. Dev Growth Differ 50:391–399.
29. Scott EK, et al. (2007) Targeting neural circuitry in zebrafish using GAL4 enhancer
trapping. Nat Methods 4:323–326.
requires lakritz, a Zebrafish atonal Homolog. Neuron 30:725–736.
31. Budick SA, O’Malley DM (2000) Locomotor repertoire of the larval zebrafish: Swim-
ming, turning and prey capture. J Exp Biol 203:2565–2579.
32. Huisken J, Swoger J, Del Bene F, Wittbrodt J, Stelzer EH (2004) Optical sectioning deep
33. Grillner S, Wallen P, Saitoh K, Kozlov A, Robertson B (2008) Neural bases of goal-
directed locomotion in vertebrates–an overview. Brain Res Rev 57:2–12.
34. Kimmel CB, Powell SL, Metcalfe WK (1982) Brain neurons which project to the spinal
cord in young larvae of the zebrafish. J Comp Neurol 205:112–127.
35. Metcalfe WK, Mendelson B, Kimmel CB (1986) Segmental homologies among reticu-
lospinal neurons in the hindbrain of the zebrafish larva. J Comp Neurol 251:147–159.
36. Myers PZ (1985) Spinal motoneurons of the larval zebrafish. J Comp Neurol 236:555–
Neurol Psychol 18:409–431.
38. Mendelson B (1986) Development of reticulospinal neurons of the zebrafish. II. Early
axonal outgrowth and cell body position. J Comp Neurol 251:172–184.
repeating cells in the zebrafish hindbrain. J Comp Neurol 233:365–376.
40. McDearmid JR, Drapeau P (2006) Rhythmic motor activity evoked by NMDA in the
spinal zebrafish larva. J Neurophysiol 95:401–417.
41. Wallen P, Williams TL (1984) Fictive locomotion in the lamprey spinal cord in vitro
compared with swimming in the intact and spinal animal. J Physiol 347:225–239.
42. Graham-Brown T (1911) The intrisic factors in the act of progression in the mammal.
Proc R Soc Lond B Biol Sci 84:308–319.
43. Lutz C, et al. (2008) Holographic photolysis of caged neurotransmitters. Nat Methods
44. Wang S, et al. (2007) All optical interface for parallel, remote, and spatiotemporal
control of neuronal activity. Nano Lett 7:3859–3863.
45. Ayling OG, Harrison TC, Boyd JD, Goroshkov A, Murphy TH (2009) Automated light-
based mapping of motor cortex by photoactivation of channelrhodopsin-2 transgenic
mice. Nat Methods 6:219–224.
46. Gahtan E, Baier H (2004) Of lasers, mutants, and see-through brains: Functional
neuroanatomy in zebrafish. J Neurobiol 59:147–161.
47. Liu KS, Fetcho JR (1999) Laser ablations reveal functional relationships of segmental
hindbrain neurons in zebrafish. Neuron 23:325–335.
stem escape network in larval zebrafish. J Neurophysiol 87:608–614.
49. O’Malley DM, Kao YH, Fetcho JR (1996) Imaging the functional organization of
zebrafish hindbrain segments during escape behaviors. Neuron 17:1145–1155.
50. Gahtan E, Tanger P, Baier H (2005) Visual prey capture in larval zebrafish is controlled
by identified reticulospinal neurons downstream of the tectum. J Neurosci 25:9294–
51. Orger MB, Kampff AR, Severi KE, Bollmann JH, Engert F (2008) Control of visually
guided behavior by distinct populations of spinal projection neurons. Nat Neurosci
52. Campagnola L, Wang H, Zylka MJ (2008) Fiber-coupled light-emitting diode for local-
Arrenberg et al. PNAS ?
October 20, 2009 ?
vol. 106 ?
no. 42 ?