Abstract
Zebrafish has long been a model organism for studying the visual system. They have a
similar visual system as higher vertebrates but lack a visual cortex. Instead, higher order
visual processing occurs in the optic tectum. Previous studies of the zebrafish visual
system have used mainly electrophysiological techniques to record wide field responses.
We have developed a new technique for perfusing and calcium imaging of live adult
zebrafish. This technique allows for simultaneously recording of response in large
neuronal populations and activity in neuronal processes as small as 2-5um. Zebrafish was
immersed in and perfused over the gills with physiological 75% Hank's solution in a
custom-built perfusion system. Optic tectum of zebrafish was exposed in a dissection and
microinjected with fluorescent calcium indicator Fluo3. Confocal microscope was used to
image calcium activity in dendrites in the optic tectum receiving input from the retina. A
train of light flashes given in five-second intervals was used as stimulus. Time series of
responses were recorded at two frames per second. Graphical analysis reveals that
dendritic responses to stimuli showed different activity levels and oscillation patterns
compared with control, but responses were not time-locked to stimuli. Responses varied
between dendrites, with both excitation and inhibition evident. Neighboring dendrites
showed a higher correlation in activity, suggesting synchronization of neuronal
populations in response to light stimuli.
Introduction
Previous research in neuroscience has concentrated on studying the nervous
system at the single cell level. However, the question remains, how do these single
neurons and their intricate synaptic connections network together to allow complex
integrative functions of the nervous system? Several advances in this search have come
from researching sensory system integration.
In the visual system, integration of the attributes of an object, such as color, size and
depth, allows us to perceive the object as a unified whole. Amazingly, each of the
different aspects of an object is processed in different regions of the brain. Perceiving a
single object despite localized analysis of its features is referred to as the binding
problem. One proposed solution to the binding problem is synchronization.
Synchronization is defined as the locking or entrainment of the phases of a neural
population with other geographically isolated neural populations (Varela et al., 2001). In
the olfactory system of insects, it has been shown that disruption of synchronization
deteriorates odor discrimination (Stopfer et al., 1997), thus proving synchronization is
necessary for recognition of odor. Synchronization has not yet been proven necessary in
other sensory systems. However, a relatively new technique, in vivo two-photon calcium
imaging, has allowed scientists to witness synchronization in other sensory systems first
hand via real-time monitoring of neural networks (Konnerth, 2003).
The advantage of Ca“ imaging is that it allows for the real-time analyses of both
individual cells and neural networks at the same time. Calcium imaging makes use of the
fact that in living cells, calcium influx is often associated with depolarization due to
action potentials. Hence, one can observe neuron activity through calcium flow. This
technique has been successfully applied by Neill et al. in 2005 to characterize the
direction selectivity of neurons in the optic tectum of Zebrafish. Zebrafish are a model
organism to study visual system integration because they are easy to maintain, and, as
vertebrates, have a similar eye structure to mammals. Unlike mammals, they lack a visual
cortex, where more complex visual integration occurs, simplifying the study of basic
visual integration.
In zebrafish, visual information such as movement, shape, and color are analyzed
and integrated in the optic tectum. Single-cell and multi-cell recordings using
electrophysiology have shown that when humans, as well as other vertebrates, are
presented with a repeated stimuli delivered at a fixed interval, a specific response follows
(Takasaka, 1985). One aim of this study was to observe this response to a repeated light
stimulus at the single cell and network levels simultaneously, using calcium imaging to
record from a living and functional adult zebrafish brain.
Despite zebrafish being recognized as a model organism to study the visual system,
most work has been done on zebrafish larvae. In order to image adult zebrafish neural
networks in vivo using calcium imaging, a new method was required. In order to achieve
our goal of recording the response to a light train in vivo, our first aim was to develop a
method that allowed for imaging in adult zebrafish using confocal microscopy and
calcium sensitive dyes.
Methods
Animal Preparation and dye loading - Experiments were performed on adult zebrafish
raised at 28°C. Adult zebrafish were placed into ice-cold water for one minute to induce
hypothermic shock. They were then paralyzed with 75mL of curare injected near the tail.
The zebrafish were then placed in a petri dish with a specially designed chamber and
secured with pins. Under a dissection microscope, the brain plate was removed with
sharpened titanium tweezers from a suture just below the cerebellum to expose the optic
tectum (Fig. 1B). Only one side of the optic tectum was exposed to minimize damage to
the brain and reduce dissection time. Throughout the dissection, zebrafish were immersed
with 75% Hank's solultion (7.4pH buffered with 2.6g/L HEPES, stored at 4°C). After the
optic tectum was exposed, the gills were perfused with 75% Hank's solution using a
custom-designed setup (Fig. 1A).
The membrane permeable fluorescent calcium dye Fluo3 was used for the
microinjection. The dye was dissolved in DMSO with 20% pluronic to yield a lOmM
stock solution and then further diluted in PBS to a final concentration of ImM (Konnerth,
2003). A micromanipulator was used to insert a glass micropipette into the superficial
layers of the optic tectum. Injections were performed between 10-40ms, using the
smallest duration that could expel dye from tip, and about 7psi. Calcium recordings
started 30-60 minutes after the injection.
Calcium Imaging - Zebrafish were placed into a similar perfusion setup as under the
dissecting microscope, allowing 75% Hanks Solution to pass through their gills. Oxygen
was slowly bubbled into the Hank's solution. The confocal microscope and Fluo View
software were used to visualize the calcium dye. Aperture was set to 1 or 2, and laser was
set to 3 or 6, depending on the amount of dye present. Äfter focusing on an injection site,
a time series was taken in darkness at two frames per second for 60-120 seconds to
observe activity without stimulus. We then recorded another time series from the same
site while providing a light train stimulus consisting of one second flashes followed by
four seconds of darkness. The first flash was given at 20 seconds after recording began.
Brain Slice - Slices of the optic tectum was made from zebrafish injected with the FMI¬
43XM dye, using the same procedure as for Fluo3 injection. FMI-43XM is fluorescent
only when incorporated into the lipid membrane, and was used to outline neurons and
their processes. It was determined from calcium imaging of the slices that signals came
from optic tectum dendrites, which were 2-5um structures (Fig. 1C).
Data Analysis - In FluoView, dendrites of interest were circled as regions of interest and
their mean intensities exported to an excel file (Fig. 1D). Intensities for each circle was
normalized to the average intensity of the first five seconds, and fading was accounted for
using either only the first minute of the series, during which no fading occurred, or by
averaging the last ten seconds (Ffinal) and adding (1-(Ffinal)/240*framef to the normalized
intensity. The normalized values were used for analysis. Igor Pro software was used to
generate a smoothed graph of the data by taking a rolling average of three data points.
Results
While performing the described dissection, perfusion, and injection, one main
concern was for the vitality of the fish. We found that the main indicators that a live fish
were red gills, visible capillaries under the dissecting scope, and visible blood flow
through capillaries under the confocal microscope. Not only were we able to keep the fish
alive during the procedure but we were also able to observe and record optic tectal
activity under the confocal microscope (Fig. 2).
Imaging of brain slices revealed that we could not focus on cell bodies of optic
tectal neurons, which were deeper in the optic tectum, because their fluorescence was
obscured by background activity on the surface. Instead, the long, narrow structures that
were imaged with FMI-43XM were axons and dendrites (Fig. 1C). This test indicated
that, using Fluo3 dye, we successfully imaged calcium activity in the dendrites.
Calcium imaging revealed two particular types of activity. The first was
synchronous activity in local populations, where fluorescence in a large region increased
and decreased as a group (Fig. 2A). The second was individual dendritic activity that
stood out from background activity (Fig. 2B).
Graphical analysis showed that activity patterns of dendrites were different when
light stimulus was delivered, some displaying elevated activity while others were
depressed. Oscillation period also varied between flash and no flash trials in the same
dendrite, and between dendrites, although periods were not time locked to stimulus.
Neighboring dendrites showed more similar activities than dendrites from different
regions. (Fig. 3)
Discussion
By using calcium imaging with the confocal microscope, we have developed a
method to image in vivo neuronal activity in adult zebrafish optic tectum. Other
experiments on zebrafish have used electrophysiology to record overall retinal and tectal
responses to light, and calcium imaging to study development of the visual system in
larvae (Niell, 2005). Taking a step further, our perfusion system allowed us to keep adult
zebrafish injected with calcium dye alive under the confocal microscope for two hours or
more. We were able to resolve calcium flux in cell processes as small as 2-5um in the
integrative region where input axons synapse on dendrites of optic tectal neurons. This
technique can provide information on how low level information from the retina is
integrated and processed in individual neurons in the optic tectum through summation of
inhibition and excitation leading to an action potential.
We selected to use a train of light flashes as our stimulus for several reasons. First
of all, a light flash provided a general stimulus to compare neuronal visual processing
activity to background activity in the dark, as recorded by calcium imaging. Because the
optic tectum is expected to have other integrative functions in addition to visual
processing, neurons in the optic tectum are active even in darkness. This activity is
supported by time series recordings of the optic tectum when the zebrafish was in
completely darkness showing individual flashes and synchronous waves of calcium
activity. In fact, recordings in darkness cannot be distinguished from recordings under
light flashes by simple observation, as both display abundant and non-uniform activity
throughout the field of recording. These results show that information from light stimulus
could not be encoded in the optic tectum in simple on-off fashion as in retinal cells.
Instead, the oscillatory activities suggest that each recording represents integrative
processing in the dendritic regions, where summation of inhibitory and excitatory inputs
determined whether a specific higher order neuron would fire. Furthermore, a large
number of these processes were captured occurring in parallel in each recording.
The calcium imaging technique allowed for more detailed analyses of the data
because it combines the ability to record overall wide field activity, as shown by the time
series, and individual neuronal activity. While overall levels of activity were similar in
the flash and dark conditions, graphical analyses of specific dendrites in each recording
showed that activity did vary between the two conditions. Compared to the dark trial,
each dendrite showed different levels and periods of activity when a light train was
presented (Fig. 3). It is important to note that the peaks and troughs in the graphs do not
represent action potentials as in traditional electrophysiological recordings, as dendrites
do not necessarily fire action potentials in response to depolarization. Rather, they
represent the summation of excitatory and inhibitory inputs from the retina. And it is the
integration of all these potentials as they spread to the neuron cell body that determines if
the neuron can fire an action potential.
Comparing two dendrites in the same time series recording, for example between
figures 3A and 3B, or between figures 3C and 3D, similar patterns of activity can be
observed. For example, in the presence of light flashes, the dendrite in figure 3A showed
elevated activity from 15-35 seconds, which was also apparent in dendrite in figure 3B.
However, comparing figures 3A and 3C, two dendrites that were further apart, the levels
and periods of activity were completely different. These results show that dendrites in
close proximity tended to receive similar input. While we could not determine whether
the dendrites were receiving input from the same axons, or the exact location of the
postsynaptic cell bodies, this correlation between neighboring dendrites suggests that
populations of neurons were stimulated in a similar manner by the light. Each of these
populations may respond to a specific aspect of a visual stimulus. Furthermore, this
correlation can lead to synchronous activity in a population of neurons consistent with the
binding problem model.
While we maintained a 5 second flash interval for all the trials, the effects of the
stimulus on each dendrite were different. Some dendrites displayed a general elevation of
activity (Figs. 3A and B), while some were inhibited compared to control (Figs. 3C and
D). The periodicity of activity in response to stimuli was also different, from -30sec
(Figs. 3A and B) to -15sec (Figs. 3C and 3D). Most importantly, none of the changes
corresponded directly to the timing of each flash. The lack of dependence on the stimuli
is expected because the optic tectum represents a higher order region in the brain. These
results show that timing of a fixed interval stimulus was represented in the optic tectum
in a higher order, more simplified manner. The existence of both excited and inhibited
dendrites further support that neurons representing features matching the stimulus are
excited while neurons that do not match the stimulus are inhibited, allowing synchronous
firing of neurons that represent the stimulus above the background activity.
Another reason for using a light train stimulus was to detect a specific response to
fixed-interval stimuli that represents cognition and expectation of the stimulus. This
response has been recorded in rats (Hurlburt, 1987), cats (Harrison, 1985), and humans
(Takasaka, 1985) at 300 milliseconds following stimulus. Our current setup does not have
enough temporal resolution to resolve activity on this timescale, as the fastest recording is
only two frames per second. As the previously recorded responses came from wide field
electrophysiological techniques, they are unlikely to be reflected in individual dendritic
activity. However, this experiment shows that dendrites also exhibit oscillatory activity,
on a longer timescale, in response to repeated light stimulus. With improved time
resolution, the calcium imaging technique can provide insight into the integrative
processes in individual dendrites and neurons closer to the time of stimulus that can
explain the wide field responses.
As with all new techniques, calcium imaging in adult zebrafish can benefit from
some improvements. Recordings were interrupted by quick, irregular movements of the
zebrafish that made following individual dendrites throughout a time series difficult. We
have yet to resolve the cause of this movement, but a completely stable fish will allow
more accurate recording and analysis of activity. This experiment analyzed dendrites
randomly selected from the area of recording, which was in turn determined by sites of
injection. Being able to inject dye more evenly into the optic tectum and to identify the
same regions between fish will enable recording of reproducible results that can be
mapped to determine integrative circuitry. Finally, with these improvements in place,
different patterns of light stimuli can be tested to elicit specific relationships between
stimulus and neuronal coding in the optic tectum adult zebrafish.
Conclusion
A calcium imaging and perfusion technique was developed to study dendritic
activity in the adult zebrafish optic tectum. Under a fixed-interval light flash stimulus,
dendrites showed a different activity pattern that distinguished them from background
activity and suggest integrative synchronous firing of neurons. However, periodicity of
the response was not locked to the light flashes. This technique can be used together with
electrophysiology to resolve pathways of higher order visual processing in the optic
tectum.
Acknowledgements
The authors would like to thank Professor Stuart Thompson for his guidance and
support; and Christian Reilly and members of the Thompson lab for their comments and
assistance.
Literature Cited
Hurlburt, B.J., Lubar, J.F., Satterfield, M. 1987. Auditory elicitation of the P300 event-
related evoked potential in rat. Physiol. Behav. 39:483-487
Harrison, J. Buchwald J. 1985. Aging changes in cat P300 mimic the human.
Electroencephalogr. Clin. Neurophysiol. 62:277-234.
Stopfer, M., Bhagavan, S., Smith, B. H. & Laurent, G. 1997. Impaired odour
discrimination on desynchronization of odour-encoding neural assemblies. Nature.
390:70-74.
Takasaka, Y. 1985. Expectancy-related cerebral potentials associated with voluntary time
estimation and omitted stimulus. Folia Psychiart. Neuro. Jpn. 39:167-172.
Niell, C.M., Smith, S.J. 2005. Functional Imaging Reveals Rapid Development of Visual
Response Properties in the Zebrafish Tectum. Neuron 45:941-951.
Konnerth, A., Stosieck, C., Garaschuk, O., Hothoff, K. 2003. In vivo two-photon calcium
imaging of neural networks. Proc. Natl. Acad. Sci USA. 100:7319-7324.
Varela, F., Lachaux, J., Rodriguez, E., Martinerie, J. 2001. The Brainweb: Phase
Synchronization and Large-scale Integration. Nature Reviews. 2:229-240.
Figure Legends
Fig. 1. Experimental methods. (A) Life support system: From left- 75% Hank's entered
zebrafish mouth via drip chamber and tubing. The solution passed through the gills of
curare paralyzed fish. Excess solution was pumped out through a second tube. (B) Sites
of dissection and microinjection: dark grey ovals represent optic tectum and light grey
oval represents cerebellum. (C) Brain slice: Calcium imaging of a slice of the optic
tectum injected with FMI-43XM dye. Visible structures are input axons and optic tectum
dendrites in the integrative region. Image was taken with 100x water immersion lens.
(D) Data analysis: Circles indicate dendrites of 2-4um that were selected for graphical
analysis. Image was taken with 40x water immersion lens.
Figure 2. Time sequences reveal synchronization of neural populations. FluoView
software was used to record fluorescence at 2fps. Light stimulus was light flash for one
second followed by darkness for four seconds. (A) Sample sequence of four frames
spanning two seconds. Boxes indicate region of synchronous dendrite activity. (B,C)
Under the same experimental conditions as (A), these time sequences were taken at a
different injection site in the same optic tectum. As in (A), several dendrites are in phase
with one another. In (B), circled bright spots indicates what was defined as a dendrite.
Figure 3. Dendritic activity with and without light stimulus. Each graph represents the
same dendrite recorded with a flash (blue) and without a flash (red). In flash trials, first
flash was given at 20 seconds. Graphed from data recorded at 2fps, and smoothed by
rolling average of 3 consecutive data points. (A) and (B) are two dendrites from the same
region of the optic tectum. (C) and (D) neighboring dendrites from a different region of
the same optic tectum as in (A) and (B).
Figures
Fig. 1.

Site of
Microinjection
Suture from which skull
was removed
Pump
B
Fig. 2.
gle Dendr
Fig. 3.

Fimefsec)
Time(sec)
Flash
No flash



Time(sec)

Timese)