Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Development of a toolbox for global functional brain imaging of wake and sleep states in zebrafish
(USC Thesis Other)
Development of a toolbox for global functional brain imaging of wake and sleep states in zebrafish
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
University of Southern California
Development of a toolbox for
global functional brain imaging
of wake and sleep states in
zebrafish
A dissertation submitted in conformity with the requirements of the
degree of Doctor of Philosophy (BIOMEDICAL ENGINEERING)
By: Andrey Andreev. Advisor: Scott E. Fraser
FACULTY OF THE USC GRADUATE SCHOOL
August 2019
1
Abstract
Sleep is an essential behavior in vertebrates, and it has wide-spread effects on the brain.
Currently, the connection between sleep and neuronal activity is studied either on a large scale,
by collecting information by averaging the activity of thousands of neurons, or on a small scale
with single-cell resolution of a few cells. Given the wide-range nature of sleep effects on the brain
in vertebrate animals, we developed an integrative toolbox that provides whole-brain recording
of neural activity with single-cell resolution. We also manipulated the activity of sleep-regulating
circuits to better describe sleep using zebrafish as a model system. This approach is based on
light-sheet microscopy, a rapidly developing method, which we applied to continuous 24hr-long
imaging of live zebrafish. We assembled an analysis toolbox to quantify sleep and wake states in
zebrafish. This toolbox can describe whole-brain activity changes in several different
experimental conditions. First, we demonstrate the ability to perform whole-brain neural activity
recording during natural sleep and wake states. Second, we analyze changes in neural activity in
animals where the sleep-regulating circuit is stimulated. We provide an approach to describe
whole-brain changes in connectivity and oscillations in sleep and wake states. Using our method,
we found that we can image the zebrafish brain without perturbing sleep behavior, decreased
correlation and activity between hindbrain regions during sleep, lack of single-frequency slow
oscillations across the zebrafish brain during sleep, and that experimental activation of NPVF
neurons induces a sleep-like state. Additionally, we expanded our platform, developing a
protocol to investigate the difference between melatonin-induced sleep and natural sleep. This
work provides a foundation for the detailed description of whole brain activity changes in
sleeping zebrafish with single cell resolution, combining large and small scale studies to elucidate
the molecular and cellular foundation of sleep in vertebrates.
2
Table of contents
Chapter I. Introduction .............................................................................................................................. 4
Section 1.01 Mission .............................................................................................................................. 4
Section 1.02 Sleep is an important behavior ......................................................................................... 4
Section 1.03 Sleep is regulated by redundant circuits ........................................................................... 5
Section 1.04 Sleep behavior adjusted by modern medication .............................................................. 6
Section 1.05 What is missing ................................................................................................................. 6
Chapter II. Zebrafish as a solution for studying brain and sleep ............................................................ 9
Section 2.01 Fish sleep research is established ..................................................................................... 9
Section 2.02 Zebrafish genetics and sleep biology is relevant to humans .......................................... 12
Section 2.03 Zebrafish enables whole-brain imaging .......................................................................... 13
Section 2.04 What is missing in fish sleep field ................................................................................... 20
Section 2.05 What we will do .............................................................................................................. 21
Chapter III. Application of 2P-SPIM to imaging during sleep ................................................................. 22
Section 3.01 Two-photon functional brain imaging in zebrafish ......................................................... 22
Section 3.02 Brain imaging during sleep .............................................................................................. 28
Section 3.03 Achieving volumetric whole-brain imaging .................................................................... 30
Section 3.04 Imaging sleeping zebrafish .............................................................................................. 36
Section 3.05 Conclusions ..................................................................................................................... 39
Chapter IV. Investigation of sleep-regulating NPVF circuit using functional imaging ........................... 40
Section 4.01 Sleep regulation by NPVF-expressing neurons ............................................................... 40
Section 4.02 Methods .......................................................................................................................... 41
Section 4.03 Results ............................................................................................................................. 44
Section 4.04 Conclusions ..................................................................................................................... 47
Chapter V. Brain states during natural and induced sleep ................................................................... 49
Section 5.01 Introduction .................................................................................................................... 49
Section 5.02 Data analysis approach ................................................................................................... 49
Section 5.03 Oscillatory behavior of sleeping brain ............................................................................ 55
Section 5.04 Functional connectivity ................................................................................................... 57
Section 5.05 Statistical analysis ........................................................................................................... 60
Section 5.06 Application of melatonin protocol .................................................................................. 62
Section 5.07 Conclusions ..................................................................................................................... 64
Chapter VI. Discussion ............................................................................................................................ 66
3
Section 6.01 Current state of sleep science would benefit from bridging scales ................................ 66
Section 6.02 Zebrafish and imaging provide the answer..................................................................... 67
Section 6.03 Imaging sleeping vertebrate brain at scale with cellular resolution ............................... 68
Section 6.04 Brain imaging with stimulation of sleep circuits ............................................................. 69
Section 6.05 Tool to better understand induced sleep ....................................................................... 69
Section 6.06 What’s next ..................................................................................................................... 70
References .................................................................................................................................................. 72
Appendix ..................................................................................................................................................... 81
Section 1.01 First attempt at 3D imaging during sleep ....................................................................... 81
Section 1.02 Establishing assay for associative memory in adult zebrafish ........................................ 83
Section 1.03 Brain imaging with 2P-SPIM under different stimuli ...................................................... 85
Section 1.04 Registration code and parameters.................................................................................. 99
Section 1.05 Data storage and management .................................................................................... 100
Section 1.06 Heart rate monitoring during sleep .............................................................................. 101
4
Chapter I. Introduction
Section 1.01 Mission
Sleep and neural activity are interdependent. Sleep regulation is driven through the
neural activity of the whole brain and specific cellular circuits while sleep effects the activity of
the brain and its function. Past studies demonstrated that sleep influences large scale (whole
brain) as well as small scale (single cell) molecular and morphological changes. However due to
technological limitations, these studies were only able to address one of the two scales at a time,
restricting our ability to fully understand the effect of sleep on the brain. The goal of this thesis
is to address this limitation by analyzing neural activity on a new scale, which considers the
activity of both the whole brain and single-neurons of the sleeping vertebrate simultaneously.
To capture the sleeping vertebrate brain at single cell resolution, we expand upon modern
imaging technology, adapt data processing tools and take advantage of the simplicity of the
zebrafish model system, which has not been widely utilized for understanding the connection
between sleep and neural activity. By combining advanced optical imaging, large-scale data
processing tools, and advances in zebrafish sleep experimental methodology, we created a
platform to better understand the nature and genetic regulation of, as well as the difference
between, sleep and sleep states induced by specific circuits or drugs.
Section 1.02 Sleep is an important behavior
Given that sleep is a fundamental biology behavior, there are many connections between
sleep and other biological processes, which can be interrogated with just as many experimental
questions. One can ask whether virtually any biological process is different in animals that are
sleeping, awake, or have been deprived sleep. There is active research as to whether sleep
impacts the immune system (Bryant et al. 2004), the microbiota (Li et al. 2018), brain clearance
(Xie et al. 2013) and a myriad of other processes (Medic et al. 2017).
Sleep allows for important changes in neural connections, manifested by synaptic
rescaling, which is the selective downscaling, or decrease in strength, of synapses during sleep
(de Vivo et al. 2017). This pruning, or removal of synapses that are not contributing significantly
5
to important memories, allows for the “conservation of energy” by only stabilizing strong
synapses. This analysis was performed in fixed brain slices using ultrastructure imaging and
electron microscopy, so it was not able to recover important details, such as what sleep stage is
the most significant for rescaling, or how single individual synapses behave during sleep and
wake.
Sleep behavior effects the whole brain as well as individual cells and cell circuits, which
led to the idea of “local versus global sleep” (Huber et al. 2004a). This idea is supported by the
observation that groups of neurons can display sleep-like oscillatory behavior in mice, while the
animal remains awake. This suggests that sleep is not an inherent behavior of the whole brain,
but can be attributed to individual neuronal circuits or even single cells. Notable example of local
sleep is semi-hemispheric sleep in dolphins and migratory birds, which are able to remain awake
while half of their brain displays sleep state and low arousal (Rattenborg et al. 1999). The ability
to investigate sleep at the single neuron level would provide us with important information about
the function of sleep at the local vs whole scale.
Section 1.03 Sleep is regulated by redundant circuits
Sleep is a vital behavior for animals’ survival, which is why it is regulated by redundant
circuits (Scammell et al. 2017). In mammals, circuits include wake-promoting pathways such as
dopamine, serotonin, acetylcholine, orexin (also known as hypocretin) and others. Sleep-
regulating circuits include GABA and glycine. In lower animals, circuitry is usually simpler (Chiu
and Prober 2013), but in most species sleep regulation projects across the brain and is not
concentrated in one region of the nervous system. The importance of sleep puts pressure on
having multiple redundant systems that regulate wake and sleep states (Adamantidis et al. 2010).
Sleep-regulating circuits can affect different sleep stages in mammals (Mieda et al. 2011),
however, these studies were performed mostly using large-scale recording of neural activity,
without focusing on the behavior of single neurons. It is important to better understand whether
sleep activated by specific circuits have the same effect and serves the same function as natural
sleep, which is regulated by complex interplay between pathways.
6
Section 1.04 Sleep behavior adjusted by modern medication
Given the importance of sleep, a number of sleep medications have been developed
(Pillitteri et al. 1994) and tested. It is interesting to note that these drugs were found by
serendipity, and not design (Robinson and Toledo 2012). One example is melatonin, an over-the-
counter drug widely available in the US and around the world, which is often described as a
natural sleep remedy. The mechanism by which melatonin decreases sleep latency is somewhat
understood, but effects of exogenous melatonin on the nervous system and behavior are poorly
studied. More importantly, we are not currently clear about differences, and potentially negative
effects of administration of melatonin, although mild negative effects have been described
(Andersen et al. 2016).
Modification of sleep behavior with medication has become an important task for the
healthcare industry, and there are efforts to target specific neural circuits that regulate arousal
(Shepard et al. 2005). Such efforts leave open questions of how any given pathway affects brain
activity and whether induced sleep performs the same function as natural sleep. Targeting
specific circuits with stimulation, either with drugs or using optogenetic tools (Adamantidis et al.
2007), may provide an answer to the nature of sleep in animals, by describing the states induced
by individual circuits and determining how these states manifest in behavioral changes.
Section 1.05 What is missing
Sleep is a global process that has effects on multiple sub-circuits and at multiple scales.
To efficiently advance our understanding of sleep, we would like to have an approach that
permits analyses spanning scales and combining different types of data. For example, an ideal
toolset would simultaneously provide information about behavior, neural and glial activity across
the brain, blood flow in the central nervous system, and activity of immune cells in the brain.
Given the current state of technology development, it seems that such a toolbox is possible, at
least in some model systems.
Whole-brain activity imaging with single cell resolution would allow detailed investigation
of the neural basis of complex behaviors. Current tools to observe neural activity in vertebrates
lack either the needed resolution or field of view for such studies. We can visualize this problem
7
by considering that the human brain contains roughly 100 billion neurons. One of the most
widely-used tools to study human brain activity, functional MRI with BOLD contrast (Logothetis
2003), produces three-dimensional images with up to 200,000 voxels, which is roughly 0.0002%
of the number of cells. That means that a single voxel of a fMRI dataset averages activity of
500,000 neurons. Another approach for recording neural activity, electro encephalography (EEG),
captures activity from 0.0000002% of the total number of cells in the brain. EEG provides
information averaged from 30 million neurons per electrode. While these coarse-grain, low
resolution datasets can provide important information about brain activity, ongoing research
demonstrates that the human brain is incredibly heterogeneous, and activity of nearby neurons
can be very different (Churchland and Shenoy 2007). On the other end of the spectrum are
microscopy techniques, with the most advanced techniques allowing imaging of 1000 of
approximately 4 million cortical neurons in the mice brain, merely 0.025% of these cells, in a
single cortical column. By contrast, zebrafish larva is an ideal vertebrate model organism, where
now we can access almost all of its 50,000 neurons in the central nervous system (covering
various brain regions) using advanced optical light sheet microscopy.
There is also a trade-off in temporal resolution. Neural activity is generally defined up to
an action potential below 10 ms in duration. Only a few tools allow recording with comparable
frequency, including single-cell electrophysiological recording, or multi-electrode recording
(Peyrache et al. 2012). In the latter case, signals from several cells can mix together, or cross-talk,
since electrical waves propagate in tissue very well. Functional MRI in humans provides limited
temporal resolution of a few seconds, while optical imaging allows detection of neural activity
with sub-second resolution without cross-talk effects.
Most information about brain function during sleep comes from large scale recordings
with limited resolution. Some details, especially the interaction between brain regions during
sleep, are not able to be captured. On the other end of the spectrum, high resolution methods
provide a precise window into the activity of single cells or even single synapses and their
interactions, but only if the cells are proximal to one another. These limitations stem from both
the model system, as well as the measurement tools (microscope, electrode array, or functional
MRI).
8
These compromises have limited the study of sleep because sleep effects appear across
large temporal and spatial scales. Results from experiments performed on different scales can be
contradicting or inconsistent with each other. For example, fMRI studies show effective
decoupling of nodes in the default mode network of the human brain (Horovitz et al. 2009). In
contrast, results of the analysis of single-neuron activity in cortical areas show that neural activity
correlation increases locally. These contradictions demonstrate there is need of an approach that
can simultaneously shed light on large-scale neural population-level activity and high-resolution
single-neuron data.
To gain access to such meso-scale data about neural activity across the whole brain and
its many regions at single-cell resolution, new approaches need to be developed. These
necessarily must include both novel tools for collecting activity data, such as optical imaging, and
refinement of established, but better suited, model organisms. In the next chapter we discuss
benefits and history of using zebrafish as a model system for imaging-based studies of brain
function in vertebrate sleep.
To permit mechanistic study of the function of sleep, and the roles that each wake- and
sleep-promoting circuit plays, we need to have an approach that allows precise stimulation and
observation of brain-wide effects outside of gross behavior or large-scale oscillations. Such a
combination of tools would allow us to quantify effects of induced sleep compared to natural
sleep, and to highlight the importance of the complex control of sleep and wake for animal
behavior and health.
Neural circuits controlling sleep could be also manipulated by application of drugs and
designed chemicals. We still have a limited understanding of how many sleep-aiding chemicals
work, and more importantly, how sleep induced by them differs from natural sleep. Development
of methods to provide this information will also highlight the nature and function of sleep.
9
Chapter II. Zebrafish as a solution for studying brain and sleep
Section 2.01 Fish sleep research is established
Sleep behavior is well defined in zebrafish. Sleep is usually defined through several
criteria, that can be applied to almost all animals, from humans to zebrafish, from lizards to
jellyfish (Anafi et al. 2019). Sleep behavior is functionally defined as a behavior that:
1. Reversibly renders animal immobile
2. Follows circadian clock
3. Changes arousal threshold
4. Homeostatic, such that it displays property of rebound after deprivation.
Zebrafish sleep behavior follows all these criteria (Prober et al. 2006), both in larvae and
adult animals.
First, diurnal animals display a robust decrease of activity during night, both when guided
by light, and by their circadian clock alone (during so-called “subjective night”). Circadian clock
persists in zebrafish when animals are raised in variable light conditions following natural
light/dark cycle (in experimental conditions it means 11 hours of light and 13 hours of darkness).
Animals show periods of quiescence and activity driven by circadian clock, as early as day five.
Animals that were raised in light/dark cycle conditions will show decrease in activity during
perceived night, even if observed in complete darkness or constant light conditions for up to
several days.
Arousal threshold is robustly increased in zebrafish, as have been demonstrated using
variety of stimuli. Most often, sound or vibration is used to stimulate animal’s response. A similar
approach using tapping on the animals’ plate can be applied as well. Visual response, while easier
to implement technically, is not applicable as stimulation for zebrafish larvae, as these animals
effectively “sleep with eyes closed”, caused by synaptic disassembly of ribbon synapses in retina,
thus rendering animals blind during night (Emran et al. 2010). This limits options for possible
stimuli that can be used to test animal’s arousal threshold. The most common approach is to
deliver mechanical vibration using “tapping” device that can provide well-controlled stimulation.
10
A simpler approach is to transduce sound waves using regular speakers, coupled to container or
directly fish water, and driven by microcontroller or computer (Pantoja et al. 2016).
Historically, neuroscience of sleep has focused on the study of human sleep behavior,
human brain activity patterns, and connection of sleep behavior and health. Studies involving
human subjects are intrinsically limited. Certain invasive experimental procedures are too
dangerous, others are too expensive to do on large scale in human subjects, and the
heterogeneity of human behavior and biology increases variability and noise in the results. To
circumvent these limitations, model organisms are used, such as murine (mice, rats) models,
rapidly developed non-human primate models, and even simpler organisms such as fly (Huber et
al. 2004b) or worm (Raizen et al. 2008).
The history of the model systems for studying sleep was started by the Russian physician
Marie de Manaceine in the late 19
th
century (Bentivoglio and Grassi-Zucconi 1997). Since then,
sleep research has been focused on human disease states and mammalian model systems. More
recently, it has been proposed that field of sleep biology would greatly benefit from adoption of
simpler model organisms (Hendricks et al. 2000). At the time genetic and biochemical techniques
for simpler animal models were developed more rapidly than sleep study methods. So obviously,
it was beneficial to apply rapidly advancing toolsets of molecular biology and neuroscience
toward understanding of sleep function and regulation. One of such models, is zebrafish larvae.
It is beneficial to use model systems because we can genetically modify them, either creating
mutations in genes, or introducing markers of cell type. Smaller animals can be studied on larger
scale at a lower cost. We are also capable of creating experimental groups of animals with
relatively homogenous genetic background, thus minimizing variability of responses.
To advance sleep research by using simpler models, precise criteria of sleep behavior, or
testable definition of sleep has been established. It took another six years to fully realize the
potential that is available in applying zebrafish as a model for sleep research (Zhdanova 2006).
Since then, both adult and larvae zebrafish have need used to study sleep in vertebrate, and this
model has been proven to be very useful for variety of studies, including large-scale drug-screen,
11
investigation of neural circuits regulating sleep, and identification of new genes involved in sleep
regulation. We’ll discuss some of these studies below.
The small size of the zebrafish larvae allows observation of behavior using multi-well
plates (Serbedzija et al. 1999) (Zon and Peterson 2005). In this approach, dozens of animals can
be observed in parallel on the same experimental setup, possibly combining experimental and
control chemical treatments or transgenic and wild-type animals in the exact same experimental
conditions. Unique for zebrafish model is the low-cost possibility of very large-scale screening of
chemicals and drugs. One screen, performed on larval zebrafish using more than 5000 chemicals
(Rihel et al. 2010), demonstrated power of zebrafish model. An individual larva was placed into
each well of standard 96-well plate, and up to 8 chemicals were applied per single plate. Animal
behavior was observed for up to 3 days and automatically quantified using video tracking,
measuring 6 parameters such as length of rest and wake bouts, delay before inactivity onset after
turning lights off, total rest length, waking activity, and number of rest bouts. Hierarchical analysis
allowed grouping of chemicals by their primary role in modulation of activity, either sedative or
arousal function. Statistical analysis of effect variability allowed predicting of biological targets
for some poorly understood compounds. In this manner, MRS-1220 co-clustered with mono-
amine oxidase inhibitors, and its function as MAO-inhibitor was demonstrated in in vitro assay.
This work demonstrated how behavioral profiling can reveal conservation of function of some
psychoactive molecules, as well as uncover new potential targets for pharmaceutical research.
Ability to create transgenic zebrafish rapidly (Irion et al. 2014), allows investigation of
vertebrate sleep-related biochemistry and genetics with detail. Zebrafish sleep-related
biochemistry and biology shares extensive traits with that of other animals, including fly, and,
more importantly, with higher animals, such as mammals. Arousal and sleep states have been
reviewed (Chiu and Prober 2013), and it is interesting that not only biochemical patters, but also
brain regions anatomy are shared with other animals, including mammals. Melatonin circuit,
involved in regulating of circadian clock and sleep, has been described in detail in larval zebrafish
(Gandhi et al. 2015). Melatonin was shown to be essential for maintaining sleep in dark/light
conditions and to promote sleep downstream of circadian clock. This work also showed that
melatonin might connect sleep through adenosine signaling. Adenosine is accumulated during
12
wakefulness (Bjorness and Greene 2009), and this connects circadian mechanism of sleep
regulation with homeostatic mechanisms. The genetic mechanism of melatonin action is now
well understood, effects of melatonin on brain activity during sleep have not been investigated.
Effects of exogenous melatonin are even less clear, but it is important to understand action of
the melatonin as sleep aid.
Zebrafish has advantages over higher vertebrates as a translational model system. Very
important difference between popular mouse model and zebrafish, is that zebrafish is diurnal
animal that sleep at night and so the circuitry that links sleep and circadian clock is conserved
compared to human. Similar sleep schedule and similar sleep-regulating pathways make
zebrafish especially helpful as a model to develop and study human health-related sleep
medication.
Section 2.02 Zebrafish genetics and sleep biology is relevant to humans
In vertebrates, one of the best-characterized wake-promoting systems is a hypocretin
(also known as orexin) neuropeptide system (Yokogawa et al. 2007). Hypocretin is present in
mammals, and zebrafish. Hypocretin and melatonin circuitry are connected in zebrafish, similar
to mammals, as has been shown with imaging in transgenic animal where hypocretin neurons
express fluorescent protein GFP (Appelbaum et al. 2009). Hypocretin neurons, located in
hypothalamus, project to the pineal gland and stimulate production of melatonin, as does
application of hypocretin in cultured cells from pineal gland. Apart from the hypocretin system,
zebrafish behavior also regulated by dopaminergic, histaminergic, cholinergic, and serotoninergic
systems. Similar to the highly distributed innervation patterns seen in mammals, in zebrafish
these circuits project throughout the central nervous system (McLean and Fetcho 2004; Kaslin et
al. 2004).
To understand sleep behavior, we need to better characterize sleep-regulating neural
circuits. Such work is being done on several molecular pathways, such as orexin/hypocretin
system (Naumann et al. 2010). This work showed that neural calcium activity in this particular
circuit decreases during sleep, thus suggesting important role of activity of these neurons in
promoting sleep. To test the hypothesis, another group used approach of activating interesting
13
sleep-related neurons with opto-genetics or chemo-genetics tools. These experiments use
transgenic animals expressing ion channels that activate neurons when appropriate stimulus
(light or chemical) is applied. This approach established the importance of hypocretin neurons
for sleep (Singh et al. 2015) or role of NPVF neurons (Lee et al. 2017). In the latter work, cells are
selectively activated, and researchers observed increase in the amount of sleep.
Another example of how the zebrafish model can benefit sleep research is the ability to
generate mutants, for example animals lacking dopamine beta-hydrolase (dbh-/- mutant). These
animals lack endogenous norepinephrine (NE). Mice mutants carrying this knock-out genotype
could not survive embryonic stage, perhaps due to in utero hypoxia that causes severe heart
defects. The pups, however, can be rescued by feeding parents diet rich in
dihydroxyphenylserine, that gets converted to norepinephrine independently of DBH gene.
However, it has been shown that dbh-/+ pups born to dbh-/- mothers have increased lethality,
probably due to maternal phenotype that displays neglect. Zebrafish, that develop externally and
do not require maternal care, can be generated to be double-negative for the DBH gene (dbh-/-
) and survive till adulthood. This allows way for the function of NE and its connection to sleep
regulation can be studied (Singh et al. 2015). This work showed role of norepinephrine in arousal
of the animals, and provided link between NE and hypocretin neurons.
Section 2.03 Zebrafish enables whole-brain imaging
Different techniques to record neural activity offer different capabilities, mostly
concentrated on opposite sides of the scale-resolution spectrum. We either can have high
resolution-recording of a very limited number of cells, or large-volume imaging with averaging
signals from hundreds of cells. Temporal resolution has similar tradeoff, as we can record fast (at
the speed of action potentials, <10ms) fluctuations from only a limited number of neurons (using
single intra-cranial electrodes), or from signals of many cells mixed together (using EEG).
Analysis of zebrafish brain activity started with recording electrical field fluctuations using
electrodes (Freeman et al. 1980), patch-clamp with a pipette (McMahon 1994), and other
approaches. These tools provide direct measurement of membrane depolarization, true
recording of the activity of the cell. While precise and allowing high temporal resolution, these
14
methods are not suitable for acquiring data from more than few cells at a time. For example,
tetrodes allowed recording of signals from four cells at once (Pezaris et al. 1997). Imaging
approaches have been developed, that transfer membrane depolarization into signal of other
type – light. This allows fast recording from the large number of pixels across the sample, covering
hundreds of neurons in a single image.
Neural activity can be translated into light through several means. There were early
example of voltage imaging (Salzberg et al. 1973), but calcium imaging (Cannell et al. 1987)
became prevalent method due to much higher signal-to-noise ratio (Baker et al. 2005). The
calcium-sensitive dyes are not reporting membrane depolarization, but secondary effect of such
events: release of calcium from cellular storage into the cytoplasm, triggered by action potentials.
Because this increase in calcium is significant, and because it takes longer for calcium to return
to the baseline concentration than membrane potential, calcium indicators offer increased signal
at the price of decreased temporal resolution (Sepehri Rad et al. 2017).
To record changes in calcium concentration, calcium-sensitive fluorescent dye must be
loaded into the cell of interest by injection. Some dyes are able to cross cell membrane, which
allows rapid labeling of many cells (Smetters et al. 1999). The limiting factor of early studies
included low amount of signal, inefficient detection methods (detectors with low sensitivity), as
well as inability to target cells of interest precisely. Genetically-encoded indicators changed that.
First, it is much easier to engineer sequence encoding protein, than complex small molecule such
as a fluorescent dye. Secondly, genetically-encoded proteins can be targeted to be expressed in
the cells of interest, either anatomically (hippocampus or cortex), or by cell type (in motor
neurons but not inter-neurons or glia). Several families of genetically-encoded calcium indicators
(GECIs) have been developed (McCombs and Palmer 2008). In case of one family, GCaMP,
fluorescent protein is “split” in half and joined by calcium-sensitive domain from calmodulin
protein. When this domain binds calcium ions, fluorescent protein recovers its structure and
regain ability to emit light. That design is a general theme of fluorescent indicators, and has been
applied to other important signaling molecules (Lorimier et al. 2002), including detectors of
ubiquitous neurotransmitter glutamate (Hires et al. 2008).
15
Members of GCaMP family belong to intensity-based sensors. These fluorescent proteins
get brighter (in some cases dimmer) when concentration of molecules of interest increases.
Another type of sensors is called ratiometric, as increase in target molecules concentration leads
to shift in spectrum, for example from green fluorescence to red. Sensor of calcium belonging to
GCaMP family ultimately proven to be most simple because single fluorescent protein is used,
most convenient due to single color, and development of these engineered molecules continues
today.
There are several ways to record fluorescence signal, but commonly applied ones are
wide-field, confocal, and light-sheet microscopy techniques. In all cases optical filters are used to
separate excitation light, and pass fluorescence to the detector (a camera or a photodiode).
Widefield microscopy provides large field of view, but almost no axial sectioning, and the sample
is usually illuminated with high intensity of light across the whole sample. Confocal microscopy
(Marvin 1961)(Williams 1990) is based on scanning focused laser beam across the sample, and
collecting fluorescence through the same objective. The sectioning is performed using pinhole
that rejects out-of-focus light in the detection path, which allows removal of light coming out of
the focal plane, thus increasing contrast. The sample is still illuminated with laser light outside of
focal plane, but the registered signal is greatly improved. Light-sheet microscopy (Huisken et al.
2004), discussed below, offers axial sectioning in the illumination side, by using thin sheet of light
as source of excitation. The evolution of microscopy tools allowed us to limit photodamage by
illuminating only sheet of tissue currently in focus, as well as collecting information in that plane
in parallel by using two-dimensional camera.
To deliver calcium-sensitive fluorescent molecules in a tissue-specific manner, transgenic
animal is generated. Designed, artificial DNA sequence is integrated into genome of fertilized
embryo at one-cell stage. This sequence contains tissue-specific promoter, activated only in
tissue of interest, followed by sensor of interest. For example, in zebrafish common promoter
used is promoter of HuC gene, also known as elavl3. It is sometimes useful to have sensor fused
with localization signals, for example, nuclear-localization sequence (NLS) (Kim et al. 2014), to
increase concentration of protein in nuclei, and hence increase signal-to-noise ratio. It is also
helpful for downstream image processing, as images of nuclear-localized fluorescent molecules
16
are essentially bright disks, easily segmented computationally. Concentrating of the sensor
molecules in nuclei comes with a price of lowered temporal resolution, as now changes in calcium
across the cell should propagate to nuclei to affect sensor’s fluorescence. To visualize faster
processes of interest, calcium sensor could be localized to other compartments, such as synapses
by fusing it with synaptophysin (Pech et al. 2015). On one hand, it allows concentrating sensor in
compartment of interest (synapse) because synaptophysin is naturally transported there by cell’s
own machinery, but on the other, can potentially perturb local environment of the synapse, since
it is very crowded and tightly regulated compartment, and molecule fused to the sensor can lose
its original functionality.
Brain imaging using calcium indicators has been successfully applied to study of zebrafish
brain activity. It allows to noninvasively observe neural activity while animal is performing
complex behavior, such as prey capture (Muto et al. 2013), visual and acoustic response
(Thompson et al. 2016), optokinetic response (Vladimirov et al. 2014), and spontaneous activity
reorganization during development (Avitan et al. 2017).
Animal is usually mounted in agarose gel in order to restrict its movement and submerged
in buffered water. Low-melting point agarose (<1% by mass) allows matching its refraction index
very close to that of the water. Together with use of high-quality detection objective, with very
high numerical aperture up to 1.4, allows high resolution imaging deep into the tissue. While such
restriction can be detrimental to animal’s well-being and development (Kaufmann et al. 2012), it
can be sufficiently gentle for up to 52 hour-long imaging of brain activity in 6-7dpf zebrafish larvae
during sleep (Lee et al. 2017).
In the design of light-sheet microscope, or selective plane illumination microscope (SPIM),
thin sheet of excitation light illuminates sample at a right angle to detection optical axis.
Fluorescent signal excited in that thin (2-3um) sheet is imaged using two-dimensional sensor,
usually camera (CCD or sCMOS) with up to 2 million pixels. This technique (Figure 2.1) allows
parallelized collection of signals across the illumination plane, similarly to wide-field microscopy.
However, because excitation light is delivered in selected plane, light-induced damage is
minimized, hence allowing use of higher laser power. Since the phototoxic effects are minimized,
17
we can use higher laser power to excite fluorescence, and achieve higher speed or higher signal-
to-noise ratio.
Fluorescence imaging with SPIM can be performed with single photon (visible, 405-
750nm) or two-photon (near-IR, 800-1200nm) light (Truong et al. 2011). Second case is especially
important for functional brain imaging in transparent animals, such as zebrafish larvae, since
visible light can interfere with animal behavior by over-stimulating its visual processing circuits.
This can lead to both quantitative and qualitative effects. When working with zebrafish larva,
animal’s transparency can be detrimental when using single-photon excitation light. As one study
(Wolf et al. 2015) showed, visual response magnitude can be diminished when animal brain is
imaged with single-photon light, up to a degree of complete lack of response.
To provide full-brain coverage, axial scanning of the sample is required. This can be usually
achieved in two ways: either scan the light sheet and change focus by moving detection objective,
or move the animal through the light sheet, keeping detection objective stationary. Latter case
is less desirable because in might potentially perturb behavior of the animal by applying
acceleration. Design where the light sheet is scanned synchronously with detection objective can
be used instead, by simultaneous application of appropriate waveform from digital-to-analog
cards (such as microcontrollers) or rescaling of the objective position stage though scaling
amplifiers.
18
Figure 2.1. Light-sheet illumination in SPIM setups using single- and two-photon excitation modes. Light-
sheet microscopy allows simulataneous acquisition from a single slice of sample (B), limiting amount of
illuminaiton laser power. Imagign can be performed in single-photon mode (C, middle), as wel las two-
photon mode (C, left). Main advantage compared to scanning (LSM) approach is exclusive illumination of
the focal volume. Adapted from (Truong et al. 2011)
One of the most notable works, showing applicability of SPIM to functional brain imaging
in zebrafish was demonstration of neural activity during motor adaptation (Ahrens et al. 2012).
In this work, paralyzed with alpha-bungarotoxin injection animal was imaging at whole-brain
scale with single-cell resolution during adaptation of motor behavior. It was demonstrated, that
neural substrate for such behavior is located in central nervous system, namely telencephalon
and cerebellum. Using high-resolution imaging in immobilized animal allowed unambiguous
detection of change of activity in single neurons. This work was further developed later
(Vladimirov et al. 2014) to show how ablation of particular neurons changes fictive whole-animal
behavior. It is important to note that study of such “fictive” behavior of animal, when motor
response is not observed but rather recorded by measuring motor neurons’ activation, is
achieved by inducing irreversible paralysis. Not only that could bias results and prevent feedback
signals from muscles to central nervous system, but also limits preparation time, as animals can
survive such treatment for only a few hours.
19
One possible limitation of using SPIM for observation of neural activity in zebrafish is
necessity of animal immobilization, most often using soft polymer such as agarose. It is necessary
to perform long-term observation of neural activity with single-cell resolution. Such restraining
can be deleterious from several points of view. For longer imaging in young animals, agarose
embedding impedes proper development (Kaufmann et al. 2012). Such immobilization can also
possibly change behavior by stimulating efferent neurons, neurons of the lateral line, and other
sensory circuits. New methods are being developed to overcome this limitation and image freely
swimming larval zebrafish (Kim et al. 2017), but only using high power visible light as fluorescence
excitation source. When sufficiently advanced, these would provide much needed tool to observe
neural activity while animal performs unrestricted behavior.
To conclude, the issues facing brain imaging in behaving larval zebrafish, stem from
several points. Imaging of neural activity usually performed with paralyzing of the animal,
restricting imaging session to just a few hours. Imaging modality of choice is mostly single-photon
fluorescent microscopy, that delivers high signal-to-noise ratio, but might perturb visual response
(Wolf et al. 2015). Hence, imaging of neural activity during sensitive and long-lasting behaviors
such as sleep or learning, was difficult because of these factors. Two-photon light-sheet
microscopy can address these points by providing tool for fast brain-wide imaging tool with
single-cell resolution and minimal photo-toxicity.
Microscopy technique that allow access to neural activity during behavior, also should
allow access to some readout of the behavior. There is a tradeoff between how much freedom
we give animal, and how much optical information we are able to acquire. Zebrafish can be
rendered completely paralyzed using, for example, immersion in curare solution (Fetcho and
O’Malley 1995), which would allow very high-resolution imaging with little to no motion artifacts.
This treatment, however, is detrimental to the animal, limiting experiment to few hours. Paralysis
also limits our ability to efficiently readout behavior of the animal, as in the case of zebrafish
most of the behavior is associated with locomotion.
High-resolution brain imaging is very useful in establishing function of neural circuits, but
it is too slow for experiments that require large-scale screening of activity. One notable
20
experiment (Lin et al. 2018) demonstrated trading-off resolution for ability to test activity of 300
chemicals on brain activity. In that work, the imaging was limited to 5 planes of brain, and signals
were averaged over several neurons per voxel, however authors collected enough data to create
characteristic maps of activity corresponding to each drug, and successfully predict activity
patters of some drugs before application.
Zebrafish model system allows also precise imaging of single neuronal circuits. Genetic
targeting of calcium indicator to sleep- or arousal-regulating circuit allows following activity of
these neurons. One example is observation of hypocretin, arousal-controlling neurons in free-
swimming zebrafish, using bioluminescence (Naumann et al. 2010). In these experiment HCRT
neurons were expressing luminescent GFP-apoAequorin protein, that emits lights spontaneously
proportional to the concentration of calcium ions. Animals were observed using sensitive
photomultiplier while freely swimming. This work is perhaps the first demonstration of imaging
single circuit activity in zebrafish during sleep.
Zebrafish brain is transparent, and when combined with ability to perform transgenesis,
allows precise regulation of neural activity using optogenetic tools (Arrenberg et al. 2009). Recent
developments of optogenetics allow activation as well as silencing of circuit of interest. Given
that sleep is governed by wake- and sleep-promoting circuits with complex feedback loops, these
tools are a valuable addition to the experimental approaches used to understand sleep
regulation.
Section 2.04 What is missing in fish sleep field
Zebrafish model system has been established as useful tool for understanding genetics
and regulation of vertebrate sleep and arousal. It offers unique opportunity to dissect sleep
nature with single-cell precision or perform large-scale drug screening.
Recently, experiments with zebrafish brain imaging demonstrated how this system allows
whole-brain investigation of neural activity with single-cell resolution. At the same time, sleep
research in this model system is providing unique insights into vertebrate sleep nature and
function. Bridging together advances in imaging and sleep studies would allow to create system
that can answer questions about sleep on large scale and with high resolution at the same time.
21
Zebrafish is a useful model for studying effects of chemicals on behavior. Few reports
have explored effects of neuroactive drugs on brain activity. There is an opportunity to combine
ability to test drugs in zebrafish on large scale and discover how these chemicals change neural
activity underlying the behavior. It is interesting to ask what effects has sleep medication on brain
activity, and whether natural sleep and induced sleep carry different signatures. This could help
understand such medications better and clarify why sleep is important.
Section 2.05 What we will do
In this work we provide platform for investigation of the neural activity during sleep and
wake, extending existing research in imaging techniques and sleep methods in zebrafish model
system. Our goal is to be able to collect and analyze data that will provide characterization of
behavioral states. The platform should also be able to allow manipulation of specific neural circuit
relevant to sleep regulation. Finally, we would apply this toolbox to characterize difference
between chemically-induced and natural sleep.
22
Chapter III. Application of 2P-SPIM to imaging during sleep
Section 3.01 Two-photon functional brain imaging in zebrafish
In our experiments we used two custom-built light-sheet microscopes. First version,
named “hSPIM”, have been demonstrated and described in (Truong et al. 2011). We will describe
it briefly here, to set up stage for the modification that have been performed. Version hSPIM,
stands for “horizontal Selective Plane Illumination Microscope”, because detection objective is
parallel to the optical table, horizontal plane, as displayed In Figure 3.01. The sample is suspended
vertically and mounted in the agarose cylinder inside glass capillary. The delivery of light is
performed through the agarose and not glass. Illumination is performed by rapidly (1kHz)
scanning of focused beam of laser light, either in 1P or 2P mode. The illumination plane then is
vertical, parallel to the walls of laboratory.
Single snapshot of the data collected by this microscope is shown in Figure 3.01 together
with wide-field overview of the larval zebrafish, collected at the same time. The scanning was
performed by collecting slices in dorsal-ventral direction. In this setup we have used bi-directional
illumination that allows more even delivery of light to the sample, illuminating sample from left
and right sides.
23
Figure 3.1. Imaging brain of zebrafish using light-sheet microscopy. (Left) In hSPIM setup illumination is
provided from two sides of the sample, and detection is performed orthogonally to the illumination sheet.
(Right) Example of acqusiiton of single slice. Gree is colorcoding for fluorescence in Tg(HuC:H2B-GCaMP6f)
animals, image is superimposed with wide-field image of the same animal
In construction of the microscope we used commercially available parts. The detection
was performed though 25X water-immersion objective with 1.4NA (Olympus) on a sCMOS
camera (Andor) with 2048x2048 pixels and 6.5um pixel size. We used relay lens with 100mm
focal length, which provided final image resolution of 0.5um lateral plane. The optical resolution
was diffraction-limited, and smallest feature we can detect is about 0.3um in diameter. Scanning
in axial direction (dorsal-ventral axis) was performed by moving sample via piezo stage
(Newport). Illumination was provided thought 10X water-immersion objectives with 0.8NA
(Nikon). We used ultra-fast Ti:Sapphire laser with to excite fluorescence Chameleon Ultra II
(Coherent). Laser was tuned to 920nm for two-photon excitation of the fluorescence in GCaMP6s
protein. Attenuation of laser power was controlled by using Pockels cell (Conoptics).
When talking about fluorescence microscopy we often consider two main modalities:
single-photon (1P) and two-photon (2P) microscopy. The physics of fluorescence emission in both
cases is the same, however the molecular excitation is different in to modalities. If single-photon
excitation is mostly linear process, with every photon having equal probability to excite
24
fluorescence, then two-photon mode is a non-linear, quadratic phenomenon, that requires two
photons being absorbed by fluorescent molecule within short period, on the order of 1
attosecond (10^-18 s). First theory of two-photon effect was established by Maria Goeppert
Mayer in her thesis (Mayer 1930) in 1930.
Let’s briefly consider how that affects imaging of popular fluorescent protein, GFP. This
molecule is excitable by blue light, often from a 488nm-laser line. At the same time, it can be also
excited by much longer wavelength, 920nm, in two-photon excitation mode (Figure 3.2)
(Stoltzfus et al. 2015).
Figure 3.2 Comparison of physics of 1P and 2P excitation. In case of single-photon excitation, single
photon of visible light excites fluorophor, that then relaxes into ground state while emitting photon of
slightly lower energy than excitation photon. With two-photon excitation, molecule of fluorophore
essentially absorbs two photon of roughly half the energy (double the wavelength) compared to single-
photon excitation. With most common fluorescent proteins, it mean excitation in 850-1050nm range.
Main reason for application of two-photon excitation for biological imaging (Denk et al.
1990) was lower scattering of 2P-photons in the tissues (Jacques 2013), which would allow for
more efficient excitation within deeper structures of the sample. Another benefit of two-photon
excitation is that due to quadratic nature of excitation, the excitation is limited to the regions in
space with very high flux of photons. That drastically limits amount of background, as molecules
far from the focal plane are not getting excited. Compared to confocal microscopy, where
25
background is rejected using pinhole, light-sheet imaging allows optical sectioning with lower
loss of emitted photons.
Single-photon cross-section for GFP, or how efficient it can be excited, is orders of
magnitude larger than two-photon cross-section. That leads to practical limitation: in two-photon
mode we need significantly more excitation photons to excite same level of fluorescence if we
use 2P excitation. At the same time, two-photon excitation provides light that is highly absorbent
by water molecules, as it is basically infrared radiation, or direct heat. That leads to a dilemma:
possibility of photo damage (Ji et al. 2008) (Podgorski and Ranganathan 2016)limits number of
photons we can extract from the sample. That “number of photos” can mean in practical terms
different things. For example, we can use these photons to cover larger field of view with limited
spatiotemporal resolution, or limit our field of view to small area, while collecting very high-
resolution data. This is a common tradeoff in microscopy in general, but in 2P-mode of imaging
of the sensitive behavior such as sleep, it provides additional challenge.
On the other hand, single-photon excitation has obvious limitation in case of transparent
animal such as zebrafish: this light will “leak” into visual system of the animal, possibly changing
behavior or neural response (Wolf et al. 2015). It is a self-evident challenge in experiments where
we seek to understand visual-driven behavior, such as prey capture of ocular-motor response,
but more interesting it might affect other behavioral paradigms. For example, since auditory-
responsive cells are present in optic tectum (Vanwalleghem et al. 2017), primarily responsible for
processing visual information, there might be competition between auditory and visual
processing, modulated by intense illumination due to the microscopy technique.
The core of the stimulation control scheme is a microcontroller (MC) that regulates when
the stimuli is applied and for how long. We selected Arduino as our platform of choice, although
there are multiple alternatives. This device is extremely cheap and easy to program and operate.
We also noticed that these MCs are very durable, some serving in experiments for several years
without fault.
The problem we must solve is synchronization of stimuli delivery with imaging, as well as
recording of stimuli time. This has been resolved by linking imaging and stimulation algorithm
26
together. The imaging camera works in this case as a master clock, sending pulse with each frame
to the MC. MC then works in a “frame counter” mode, counting incoming pulses, and making
decision of whether it is time for stimuli delivery or not yet. Since our framerate is constant, and
we can pre-program MC to deliver stimuli at particular time, we are knowing exactly, ahead of
time, which frames will precede, and which will follow the stimulation period (Figure 3.3).
The stimuli are activated via different means, but generally each device that either
controls light output, or sound, or electrical shock, will have appropriate connectors to trigger
stimuli. For example, in case of the light stimulation, the device (LED controller) has one input
connector that can be set in either HIGH or LOW state by the MC. Microcontroller is programmed
to switch state for precise number of frames or time.
Figure 3.3. Principal scheme of interaction between camera and microcontroller for behavioral
stimulation. Camera sends exposure signals to the controller that counts number of registered signals. At
appropriate frame number, stimulation is applied to the animal (including light, sound, electrical shock, or
vibration). The programming is done using computer before acquisiton
The logic programmed in controller simply counts number of frames taken by camera and
performs necessary actions according to algorithm. Consider for example simple experiment with
using LED light as stimulation for 20 frames in the middle of 120-frame long probe bout, turning
it on at frame 50, and turning it off at frame 70. The logic can be outlines in pseudo-code as:
27
c = 0 // initialize counter
turn LED to LOW // initialize LED light
while True then
if pulse from camera received then
c += 1
if c >= 120 then
c = 0
end
if c == 50 then
turn LED to HIGH
end
if c==70 then
turn LED to LOW
end
end
The microcontroller as well can send additional information to the stimulator, for
example, related to which stimulation episode is that. For example, we might want to send 100%
of LED power on even trials, and only 50% on odd ones. This is done by sending additional signal
for modulation of stimuli intensity. In case of auditory stimulation, the interface was serial, that
is microcontroller would send a message such as “BUZZ” to PC that will in turn produce sound.
Serial connection allows transferring much more diverse information. The connection can also
be reversed, for example, for triggering stimulation based on some image analysis, for example,
after period of animal inactivity. We describe results of light, sound, and vibrational stimulation
in Appendix.
28
Section 3.02 Brain imaging during sleep
To establish imaging protocol for sleep imaging in zebrafish initially, we elected to image
only single plane of the zebrafish brain, and for short periods of time. It can be visualized as
“probing” brain every 15 minutes for 2 minutes. This creates short movies of brain, taken at
different time during night and day. This approach has benefit of limiting amount of laser
exposure of the animal, preventing photo-damage. Because we image not continuously but with
interval (duty cycle of 25-50%) amount of data for analysis is decreased, allowing us process
imaging data quicker.
At each imaging trial, we can apply some type of stimulation to the animal to probe it
behavioral state. This approach allows “plug-in” of different stimuli, controlled through the same
similar protocols. In appendix we describe technical implementation of control schemes for
stimulation using light, sound, and mechanical vibration. To investigate brain activity in sleeping
zebrafish, we started by applying light stimulation using ramping stimuli (triangle wave) explained
in Figure 3.4.
29
Figure 3.4. Probing brain periodically during night. (A) The imaging is performed for short periods (1-5
min) and repeated every 15 minutes (intra-trial interval, ITI), thus providing “probing of brain activity”.
The stimuli is applied eaither every trial or on selected trials, for example in stimuli-no stimuli interleaved
fashion (B) One example of stimuli is LED ramp, triangle wave of gradually increasing red 625nm LED light
power.
One of the criteria of sleep is change in arousal threshold. We hypothesized that zebrafish
larvae, exposed to gradually increased light will show slower increase of response to light
increase. During night we would expect larvae showing less response to same light magnitude,
compared to day-time trials. Indeed, we observed that during night and day, transgenic animals
expressing cytoplasmic GCaMP5G showed decrease of the optic tectum neural activity in
response to light. We also noticed that animals displayed less neural activity before the onset of
stimulation (spontaneous activity).
We observed (Figure 3.5) significant decrease in response to light onset and offset alike
during night, but we cannot attribute it to sleep state, as zebrafish larvae lose vision at night
(Emran et al. 2010). This process is driven by circadian clock and so our experiment with light
stimulation confirmed that during imaging zebrafish retain circadian clock synchronization.
30
Figure 3.5. (Left) Longitudinal imaging of zebrafish brain shows change in response to light during day and
night. (Right) Activity in optic tectum (dashed box) decreases during night and recovers during following
day. Both spontaneous and light-evoked activity changes with circaidan clock time.
Spontaneous activity, thought, is not expected to be regulated by vision, and driven by
internal state of the animal. We observed change in spontaneous activity when comparing day
and night imaging sessions, suggesting that our imaging approach does not perturb intrinsic brain
states of larval zebrafish.
Section 3.03 Achieving volumetric whole-brain imaging
Zebrafish allows whole-brain imaging of neural activity with single-cell resolution, which
is a unique imaging opportunity among all vertebrate model systems. To take advantage of this
approach, we need to develop a tool for three-dimensional imaging. To achieve that, we
developed “upright” light-sheet microscope with capability of fast scanning in axial direction with
two-photon illumination.
The microscope (Figure 3.6), in principle, is similar to other published light-sheet imaging
systems, but we performed several novel modifications. First change to standard light-sheet
microscope is introduction of custom-made masks in the optical path to directly block light
coming into the eyes of an animals. In recent publications more complicated approach of
computational laser light modulation was used, that effectively turned laser off when beam was
about to hit the eyes of an animal.
31
Figure 3.6. Schematics of light sheet Upright-SPIM microscope. The excitaiton light sheets are delivered in
the front and on the side, allowing full illuminaiton of the brain. Light is selectively blocked by physical
masks to avoid exposure to the eyes. Two cameras are continously collecting signal from the brain (top)
and from the tail (bottom). The sample (orange blob) is mounted on a metal z-shaped beam and
completely submerged in liquid. Fluorescence detection objective is immersed as well.
Animal was immobilized (Figure 3.7) in a custom 3D-printed caddy (designed by Matt
Jones, manufactured by Proto Labs) using low-melting point agarose (1%, Seaplaque), and some
of the agarose was cleared around the tail, allowing free movement, while keeping head
immobile. Excitation light was delivered from two sides: right lateral side and from the front,
similar to published setups (Vladimirov et al. 2014). Light sheets were created by rapid (1kHz)
scanning of focused 920nm laser light using galvo-mirrors controlled through programmable
microcontroller (National Instruments) using custom LabView program (Truong et al. 2011).
32
Figure 3.7. Custom 3D-printed caddie for mounting zebrafish sample. Animal is mounted in 1% agarose,
while agarose around tail is carefully removed using fine tweezers. Caddie, made of transparent plastic,
allows observation of tail activity through IR-sensitive camera
To observe brain activity in three dimensions while using only two-dimensional
illumination scheme, we need an appropriate scanning to cover three-dimensional subject.
Detection objective mounted on a piezo stage, that allows precise rapid scanning in axial
direction with up to 10nm precision. Output signal of the piezo is connected (Figure 3.8) to galvo
scanner mirrors (Cambridge Technologies) that synchronize position of detection objective with
offset of light sheet. This is done by employing scanning amplifiers (Stanford Research Systems),
a hardware approach we find better suited than application of programmable microcontrollers.
This is done by using function generator (Stanford Research Systems) to produce scanning
triangle waveform to drive piezo, and triggered function generator to produce uniform exposures
through the scanning range (PicoScope). Parameters of function generator are calculated to
produce desirable span in Z scanning at given frequency. For example, to produce scanning over
250um at 0.3 volume/second, we used parameters displayed in the table below.
33
Figure 3.8. System for synchronization of light sheet with detection objective in Z direction. (Top) All
connections are wired throiuhg a standard BNC patch panel, that allows rapid reconfiguration or
troubleshooting. Piezo controller (middle device) is connected to one of the scaling amplifiers (lower
device) in order to synchronize position of light sheet and detection objective in axial direction. (Bottom)
Principal scheme of controlling Z position of the light sheet. Monitor signal report position of detection
objective piezo collar, that is translated into signal for matching galvo offset, responsible for axial position
of light sheet
34
Second function generator (embedded into PicoScope) is set up to create rectangular
wave of pulses that trigger camera taking fluorescent frames. To set it up properly, we need to
know beforehand how many frames we want to acquire during one Z scan, or conversely, what
exposure time we want to have. These two parameters are linked by reciprocal relationship: the
more frames we want (or re finer axial resolution is required), the faster we will have to acquire
these frames. Roughly, one can equate:
Scan time per volume =
= (Axial Span / Axial Resolution) * Exposure =
= N Frames per Volume * Exposure
The parameters we used to image whole brain of zebrafish are presented below:
Mode of
imaging
Frequency, Hz Offset, V Amplitude, V Exposure, ms Nz deltaZ, um Seconds
per
volume
Whole brain 0.33 1.51 3.02 45 54 4 3
35
Figure 3.9. Example result of 3D acquisiton of brain activity in larval zebrafish Tg(HuC:H2B-GCaMP6s)
using uSPIM. Images show that we can record single neurons (spots) across whole brain volume. Slices of
brain were acquired every 5um over 250um range, here shown only every 4
th
slice. Bidirectional
illuminaiton allows excitaiton fo fluorescence between the eyes of the animal without hiting eyes with
laser. Scale bar 100um.
The imaging described here allowed us to perform whole-brain recording of neural
activity with single-cell resolution (Figure 3.9). It is useful to note that this microscopy technique
can be used to image several fluorescent markers at once, for example integrating neural activity
and immune system activity imaging. The open question is the ability of our experimental setup
to record brain function during sleep and wake. In the next section we describe in detail how we
36
achieve simultaneous behavior and brain observation., and how analysis of tail motion during
imaging allowed us to confirm that animal was sleeping during acquisition.
Section 3.04 Imaging sleeping zebrafish
To simultaneously observe whole brain activity with single-cell resolution, animal head
immobilization is required. At the same time, it is necessary to allow animal freedom to observe
its behavior. We approached this problem by restricting movement of the head while freeing the
animal tail. Observation of tail twitching allows us to gauge behavioral state. Indeed, more
movement is observed during day, and amount of movement, or distance between spontaneous
twitches, gradually increase at the onset of night (Figure 3.10).
Two-photon light-sheet microscopy is not a new has never been applied to tackle problem
of whole-brain imaging in zebrafish with single-cell resolution, especially in attempt to describe
brain activity during sleep and wake. Optical microscopy can be disturbing to the animal, and
even cause tissue damage, so we had to verify that animals under out observation retain sleep
behavior characteristics.
Established definition of sleep in zebrafish larvae is period after 1 minutes of continuous
inactivity. Every following minute of inactivity is defined as “sleep minute” as well. That is, if
animal was stationary for 10 minutes, we regard that as 1 minute of wake followed by 9 minutes
of sleep. Periods less than one minute of inactivity are considered “wake” periods. Schematics of
this analysis are presented in Figure 3.10.
37
Figure 3.10. Simultaneous imaging of t ail motion and brain activity to establish sleep and wake bouts. The
data is aquired simultaneously on two cameras. (Bottom-left) Animal display sleep behavior while imaing,
showing increased immobility during night and recovery in the morning. Black bar labels night-time,
10pm-9am window. Tail motion is used (top-right) to measure animal arousal by detecting twitch. Semi-
automatic processing of tail data provides registration of each twitch, which then transforms with labeling
each brain time point as wkae or sleep (bottom-right)
To quantify sleep during brain imaging, we observed movement of the tail in animals
head-mounted in agarose gel. Animals were imaged while in normal light/dark cycle, and tail
motion was quantified as frequency of tail twitches, without detailed characterization of tail
dynamics, as any motion was regarded as a twitch. We used wide-field camera (iXon) to record
tail movement, and animal was illuminated using invisible infra-red LED light source at 725nm
wavelength (Thorlabs). The tail movement was analyzed by calculating image difference. The
image artifacts were caused by the fact that image of the detection objective was partially in the
frame of tail camera. Hence, movement of detection objective at the end of volume acquit ion
was detected as significant image change. After removing image artifacts through analysis of
background image, we were able to automatically detect >97% of tail twitches as confirmed by
manually reviewing detected twitches and corresponding movies (Figure 3.11).
38
Figure 3.11. Example of raw image difference data of tail movement. Image difference vide (on the right)
highlights frames from image-difference processing resiult that show significant change of the image due
to tail motion.
During imaging we observed that animals slept 70-80% of the night (Figure 3.12), similar
to 60-80% reported in free-swimming animals (Yokogawa et al. 2007), confirming that our
imaging protocol is not perturbing sleep behavior in zebrafish. In our experiments we observed
that animals display similar sleep architecture (number of sleep bouts, and distribution of bout
lengths) as in free-swimming animals (Dvir et al. 2018). Animals in our experiments spent similar
amount of night time asleep, as free-swimming animals. moreover, our approach would allow
more detailed analysis of wake/sleep states, for example, provide information about latency of
sleep onset, as well as splitting sleep into long- and short-sleep bouts.
39
Figure 3.12. Sleep behavior characterization during imaging (Right) Histogram of the distribution of sleep
bouts lengths across 7 experimental fish show 2-fold variability (for example at the 10% mark) but on
average correpond to measurements performed in free-swimming animals. (Left) Animals slept roughly
same fraction of night as in free swimming experiments (citation in the main text). Whiskers are ±SEM
In our experiments we extracted 180-200 sleep and wake bouts per fish. We only analyzed
bouts of wake and sleep that are longer than 1 minute. Outliers that didn’t demonstrate sleep
amount in the 7-10hr range were excluded. To speed up processing we randomly sub-selected
20-40 bouts of each state for each animal, each bout had successfully passed registration to the
atlas.
Section 3.05 Conclusions
To achieve whole-brain imaging in sleeping zebrafish, first, we applied existing two-
photon light sheet microscopy method. Our setup provides light stimulation and other types of
stimulation to the animal that would allow testing arousal threshold. Secondly, we modified the
microscope to allow rapid imaging in three dimensions, preserving single-cell resolution and
minimizing photodamage and thermal damage to the sample. Finally, we performed imaging of
alive, non-paralyzed, animal with its tail free, that allowed us quantification of behavior, and to
separate bouts of sleep and wake. This work establishes platform that provides whole-brain
imaging in sleeping zebrafish with single-cell resolution.
40
Chapter IV. Investigation of sleep-regulating NPVF circuit using functional
imaging
Section 4.01 Sleep regulation by NPVF-expressing neurons
Vertebrate sleep is regulated my many redundant circuits, perhaps due to the importance
of sleep behavior for survival. Serotonin, GABA, and glutamate are among neurotransmitters that
play an important role in regulation of sleep and arousal (Saper and Fuller 2017). In mammals,
most of these circuits are located in hypothalamus, and project throughout the brain. Imaging
studies were very useful in detecting areas activated during sleep and wake (Dang-Vu et al. 2008),
and recently experiments have been developed that allow precise manipulation of circuit of
interest to understand its function and nature for sleep behavior regulation.
Optogenetics is a tool of choice in some of these experiments, as it allows spatio-temporal
precision in activation or suppression of specific circuit. Activation of GABAergic neurons in
mouse ventral medulla shows induction of REM sleep (Weber et al. 2015). Nature of optogenetic
experiments is that it is possible to perform closed-loop experiments, where observation of
current animal state informs application of stimulation. In the same paper, authors demonstrated
that activation of these neurons during REM sleep, prolongates that sleep stage.
Investigation of sleep regulation and function has been done by modification of circuit
activity in different systems (Weber and Dan 2016). In zebrafish, we have strong experimental
evidence of importance of hypocretin (orexin) circuit (Faraco et al. 2006). Suppression of that
wake-promoting circuit induces sleep-0like behavior. Norepinephrine is another example of
sleep control circuit that is present in both mammals and zebrafish and has been studied in detail
in that model system (Singh et al. 2015). Zebrafish model system allows genetic manipulation of
circuit of interest, investigation of connectome, and precise localization of cells of the circuit due
to optical accessibility of the zebrafish.
One important example of sleep-regulating systems is a member of RF-amid (containing
a C-terminal Arg-Phe-NH2 motif) family, neuropeptide VF. It was identified in humans, and
synthetic NPVF peptide injection in rodents affect hormone levels, hunger, and nociception. In
41
rodents and human NPVF is enriched around hypothalamus, suggesting role in regulation of
homeostasis. Zebrafish offers unique opportunity to investigate regulation and function of NPVF
peptide, and specifically answer questions about connection of this family of peptides and sleep.
This work has been published (Lee et al. 2017) and here we will concentrate on details regarding
imaging and demonstrate what this work brings to sleep studies in model systems. Activation of
NPVF neurons promotes sleep in zebrafish, as observed by behavioral experiments. It is inducing
sleep state, according to all criteria, including decreased arousal threshold, and not only decrease
in mobility.
In this work we applied our imaging approach to try to answer the question about nature
of NPVF sleep, and whether NPVF-induced sleep demonstrates significant signature of brain
activity in sleeping zebrafish.
To answer that question, we performed two experiments, addressing two different time
scales. First, we analyzed neural activity in zebrafish after acute activation of NPVF neurons with
optogenetic stimulation. Secondly, we compared brain activity in sleeping zebrafish under long-
term activation of NPVF neurons using chemogenetic activation. In both experiments we
demonstrated decrease in neural activity, and specifically demonstrated that NVPF activation is
not causing brain-wide depression but concentrates effect on sleep portion of day.
Section 4.02 Methods
In this work we used two-photon light sheet microscopy in a horizontal configuration, as
described in previous section. We performed single-plane imaging, with the resolution set at
0.5um in lateral plane (XY). We used 940nm laser light to excite fluorescence. The experiments
divided into two blocks: first, using acute, optogenetic stimulation of NPVF neurons and short,
15-min imaging, and second, with chronic stimulation of these neurons, and 24hr-long imaging.
In our experiments we used 5-7 days old transgenic zebrafish carrying fluorescent calcium
indicator GCaMP6s fused to nuclear-localized histone protein H2B. Indicator was expressed in
pan-neuronal fashion using HuC promoter. These animals are denoted as Tg(HuC:H2B-GCaMP6s).
42
Optogenetic experiments used transgenic animals expressing red-shifted channelrhodopsin ion
channel ReaChR, Tg(HuC:H2B-GCaMP6s;NPVF:ReaChR-YFP).
To perform chronic activation of NPVF neurons, in second set of experiments transgenic
animals were expressing capsaicin-sensitive ion channel TrpV1, in triple-transgenic fashion
Tg(HuC:H2B-GCaMP6s;NPVF:Gal4;UAS:TrpV1-YFP), while control animals were expressing only
two of these transgenes, Tg(HuC:H2B-GCaMP6s;UAS:TrpV1-YFP). Capsaicin was applied (10uM)
to animals in petri dish at day 4 post fertilization, and imaging was performed at day 6.
The fluorescent tags on the transgenic animals is YFP and so indistinguishable from signal
of the GCaMP. To genotype animals after imaging experiments, we perform sodium-hydroxide
embryo homogenization (Meeker et al. 2007) followed by PCR of the extracted genomic DNA.
To amplify ReaChR sequence we used the following primers:
• ReaCH_Fwd: CACGAGAGAATGCTGTTCCA
• ReaCH_Rev: CCATGGTGCGTTTGCTATAA
To genotype for TRPV1 (349bp fragment) we used primers:
• ST32F: CAGCCTCACTTTGAGCTCCT
• ST33R: tcctcataagggcagtccag
To genotype for KalTA4 (270bp fragment) we used primers:
• KalTA4_Forward: ATGCAAAGCTGTGAGTGCAT
• KalTA4_Reverse: TTGTGAGTGGACTTCGCTTG
During optogenetic imaging, each trial consisted of 2 minutes of imaging, with 3 minutes
of intra-trial interval. We imaged brain activity for 5 trails without stimulation, and then applied
stimulating light (625nm LED). The imaging continued for another 10 trials. To record neural
activity during day and night, we performed imaging in 2-minute trials with 10 minutes interval.
The resolution and imaging parameters were same as previously described. We used red LED
light (625nm) to simulate day/night conditions for the animal.
43
Next, we applied PCA/ICA-based segmentation (Mukamel et al. 2009) to recover
information about single neurons present in our dataset and imaging pipeline. This toolbox
combines compression of data into low dimensional space using Principal Component Analysis
(PCA), followed by unmixing of signals from individual neurons in time and space with
Independent Component Analysis (ICA). The unmixing is done both in time and space, allowing
separation of overlapping neuronal images. To segment neurons from imaging data, each
imaging trial was processed separately. The activity of all segmented neurons was averaged as
average activity during trial. Usually, 20 principal components (CPs) contained >90% of variance.
All PCs were used to estimate independent components corresponding to individual neurons. We
used kernel of 5um diameter to extract position of neurons.
Example of segmentation and extracted neural activity traces is shown in Figure 4.1. We
only display here random subset of 20 segmented cells. For visualization it is useful to perform
clustering of neural activity traces, in which case we used k-means clustering algorithm with
correlation distance metric.
44
Figure 4.1. Resut of segmentation neurons from siongle-plane imaging data. (A) raw fluorescence image
with estimated position of neurons in brain segmented using PCA/ICA toolbox (B) Normalized
fluorescence traces for random subset of 20 neurons, clustered using K-means algorithm with k=5
Section 4.03 Results
First, we applied framework to describe brain activity in limited field of view, but with
single cell resolution, in wake, sleep, and in conditions of induced sleep. In these experiments,
we sought effects of experimental activation of NPVF secreting neurons (Lee et al. 2017). We
observed brain activity of larval zebrafish, imaging single plane of the brain for up to 24 hrs.
Analysis of these data showed that zebrafish brain activity robustly decreases by up 5% in
magnitude, during night.
In first set of experiments, we used opto-genetics to activate NPVF neurons (Figure 4.2).
Transgenic animals carrying red-shifted version of channelrhodopsin showed decrease in
mobility and increase in sleep in behavioral experiments after onset of red light. During our
imaging experiments, transgenic animals showed marked decrease in neuronal activation after
illumination with red light, compared to ReaChR-negative siblings.
After onset of the stimulating light, we observed increase in optic tectum activity in all
animals. However, in the transgenic animals carrying optogenetic channel, initial increase in
signal followed by rapid decrease of activity with time constant of approximately 5 minutes. That
change was similar to the decrease in neuronal activity that we detected in animals during 24hr
observation, suggesting that activation of the NPVF neurons induces brain state similar to sleep.
45
Figure 4.2. Schematic and results of optogenetic stimulation of NPVF-ReachR transgenic animals. The
imaging was performed in single plane of the zebrafish brain, and analysis was limited to average activity
in optic tectum (dashed box). In ReaChR-negative anmimals neural activity icnreases but stays relatively
flat after onset of light, while in ReaChR-positive animals GCaMP6s signal exponentially decreases after
intiail increase due to stimulating light. Plot shows timecourse of activity in region of interest averaged
across 3 animasl, with shadow being SEM. Scale bar 50um (100um inset)
Second set of experiments was employing chemo-genetic activation (Figure 4.3). The
transgenic animals were carrying ion channel TrpV1, sensitive to chemical not present in
46
zebrafish, capsaicin (Chen et al. 2016). When applied at low concentration, that chemical shows
activation of ion channel. Transgenic animals carrying TrpV1 channel again showed decrease in
activity and increase in sleep during night, compared to TrpV1-negative siblings. Neural activity
was significantly lower during night for transgenic animals. Interestingly, only night-time neural
activity decrease, and not overall neural activity, which suggest that NPVF regulates sleep, and
not promotes overall neuronal depression.
47
Figure 4.3. Chronic activation of NPVF neurons causes deeper decrease in neural acitivity during sleep. (A)
Example of TRPv1-negative animal average brain activity during day and night. (B) TrpV1-positive animal
activity during day and night after 2 days after application of capsaicin. (C) Profile of average activity for
animals in A and B. (D) Average activity during day in TrpV1-negative and -positive animals is the same. (G)
Normalized night-time activity is decreased in TrpV1-positive animals.
Section 4.04 Conclusions
Experimental modulation of sleep is an important tool to understand sleep regulation as
well as function. Here we provided evidence that zebrafish is suitable model for studying effects
of activation of specific sleep-related circuit, namely NPVF system, in context of whole-brain
neural activity. Sleep is regulated by many different, sometime overlapping systems, working in
a redundant fashion. Having ability to asses effect of each of them on neural activity separately
would provide information necessary for full characterization of sleep-regulating vertebrate
circuitry.
48
The mobility of an animal is a standard behavioral test in zebrafish sleep experiments.
Even though we did not directly track animal’s arousal level in these experiments, we connected
two different areas: brain imaging and behavioral experiments in sleeping zebrafish larvae (Figure
4.4). When in behavioral experiments we saw that animals decreases mobility after onset of
induced or natural sleep, in our experiments with induction of sleep, we observed decrease in
neural activity across large brain regions and many neurons.
Figure 4.4. NPVF experiments bridges established sleep experiments with brain imaging in zebrafish. (Left)
Location of NPVF-positive neurons within larval zebrafish brain. Stimulation of these cells with through
optogenetc causes decrease in light-evoked activity and increase in sleep amount. (Right) Single slice of
brain imaged using hSPIM. Stimulation of animals carrying red light-sensitive ReaChR markedly decreases
GCaMP signal.
This experiment provided first verification that it is possible to perform whole-brain
imaging in sleeping zebrafish, and potentially investigate difference between natural and induced
sleep in that model system. In our work, we performed acute and chronic stimulation of NPVF
neurons, thus outlining protocols for investigation of sleep regulation on two different time
scales.
49
Chapter V. Brain states during natural and induced sleep
Section 5.01 Introduction
Sleep behavior is linked to brain and brain activity on multiple scales. Sleep can be viewed
from a global, and local standpoint, analyzing activity of single neurons, or by looking at whole
brain at once. Methods that would provide comprehensive picture of sleep behavior and its
physiology and function should allow investigation on both ends of scale spectrum. To provide
such tool, we combined modern light-sheet microscopy with zebrafish model system. This
combination allows observation of whole vertebrate brain during sleep and wake while resolving
activity of separate neurons. Our tool allows us to obtain simultaneous activity and behavior data
for sleep study in zebrafish model, something that has not been presented before.
Section 5.02 Data analysis approach
Whole-brain activity imaging poses problem of data analysis, as we must be able to handle
imaging information in 100s of GB range for a single imaging experiment. In our work, we usually
collect about 40GB of tail movement movies, and 350GB of brain activity imaging data. Storing
and managing these datasets represents technical problem, without standard and affordable
solution, so we had to devise strategy ourselves. Some of the discussion of this and description
of our data storage solution is provided in the Appendix. This is a task that only few research
groups have tackled, and we have to develop robust and fast infrastructure to store and process
data. Task of data pre-processing is also an important task that required design and
implementation of specialized software. Both of these tasks currently lacking standard solution,
and the platform for zebrafish brain analysis at large scale has been published only recently (Chen
et al. 2018).
Data processing pipeline is generally described in Figure 5.1 and all code was created as
part of this work. To analyze large dataset resulted from continuous 20-hours recording and put
neural activity into anatomical context, first we constructed our own zebrafish brain atlas, based
on published results, such as the ZBrain atlas (Randlett et al. 2015). Atlas was constructed
primarily by Anna Nadtochi, based on three registered and averaged movies of 3D brain activity
50
in 3 different animals, thus accounting for some biological variability. Atlas was built manually
using Computational Morphometry Toolkit (CMTk), and automatically registered to each
experimental dataset. After that custom MATLAB code was used to collect average activity across
brain regions corresponding to each mask.
Figure 5.1. Overview of the processing pipeline. (Top) Tail and brain are processed independently, each
containing >1.2x10
6
frames. Tail data is used to segment recording into sleep and wake bouts. (A) Bout
has assigned brain movie, because both cameras are synchronized electronically. (B) Brain data further
processed by registration of masks to the bout data (data is not registered to the atlas). Masks for each
bout are then applied to extract average region-wise activity, or to extract power spectra via
supervoxilized approach
Registration of brain images was done using ANTs (Advanced Normalization Tools)
toolbox (Avants et al. 2011). This set of programs was designed for registration of anatomical and
functional MRI datasets together. First, we applied registration using routines in RStudio, ran
inside virtual OpenSUSE machine. Later the process was adapted to be done entirely on Windows
51
platform using command-line utilities of pre-compiled ANTs toolbox. Query was constructed
following ANTs manual [https://github.com/ANTsX/ANTs/wiki/Anatomy-of-an-antsRegistration-
call]. The code that was used to process batch registration is provided in Appendix.
Natively, ANTs do not support registration of 4D dataset (3D+time) to single reference 3D
dataset, so that 4D dataset has to be split into series of 3D datasets corresponding to each time
point. Code for registration if a single 3D volume is wrapped in a function to perform this
processing in parallel through Ruby programming language.
We used segmentation of brain into specific areas based on atlas of 11 regions. First, we
quantified frequency spectrum (Figure 5.2) of the neural activity in sleep and wake bouts. Brain
regions were split into smaller super-voxels 10x10x10um, and signal was hence averaged across
approximately 6 neurons. Fourier transform was used to extract power spectrum. After that we
averaged power spectra across single brain region, yielding average frequency spectrum per
region per bout. In this processing we discarded frames that included tail motion plus 5 seconds
after that, which should decrease bias toward motion-based differences in the data.
52
Figure 5.2. Example of super-voxel segmentation to analyze oscillatory behavior of zebrafish brain. (A)
Brain is split into 10um-large cubes in 3D. (B) Neural activity is averaged across brain regions (left) or
supervoxilized and processed to yield frequency spectra of activity
In control experiment, we presented regular light stimulation to the animal. Red LED light
was pulsed for 10 seconds with 110 seconds interval, and we analyzed data in the same pipeline
with sleep data. This processing successfully extracts activity at stimulation frequency around
100 seconds period (Figure5.3).
53
Figure 5.3. Periodic stimulation with light causes periodic response in Optic Tectum. (Top) Profile of
stimulation using red LED. (Bottom) Frequency analysis of 30 minutes of actrivity recorded during
stimulation with repeated LED light pulses. Frequency spectrum was averaged over 5 trials. Two peaks of
the spectrum (around 100s and 60s) corresponds to the peaks present in the spectrum of rectangular
signal of the stimulating light (higher harmonics of the carrier frequency).
In the same experiment we observed synchronized activity in the hindbrain (Figure 5.4),
at the period of 40s. It matches with the observations of hindbrain oscillator first reported by
(Ahrens et al. 2013) and later confirmed by (Wolf et al. 2017). This confirms that our processing
pipeline is sensitive to pick up oscillations present in large brain regions.
54
Figure 5.4. Power spectra of the spontaneous activity in zebrafish hindbrain, recorded over 30 minutes.
Frequency analysis of activity highlights presence of the hindbrain oscillator with 40s period. Red line is
mean power averaged over ten 3-minute trials, shading is ±SEM range.
To analyze functional connectivity between the regions of the brain, we averaged activity
in each region, and calculated Pearson’s correlation coefficient. Coefficients were transformed
to Z score using Fisher’s transformation. For averaging and statistical testing. Correlations maps
were transformed in “spider-web” to aid reader connecting regions with anatomy (Figure 5.5).
Figure 5.5. Correspondence between regions and vertices of “spiderweb” correlation map
To test correlation coefficients’ statistical significance, we performed random shuffle test
(McIntosh et al. 1996), randomly permuting neural signal timepoints 1000 times, and calculating
resulting distribution of correlation coefficient. The p-value of this test is probability that
randomly permuted signals correlate more than what have been measured. We only included
correlation coefficients that have p-value < 0.05 into further analysis (Figure 5.6).
55
Figure 5.6. Distribution of correlation coefficients in shuffle-test for activity of single pair of brain regions,
for single bout of wake. Higher correlation coefficient usually means higher statistical significance.
Section 5.03 Oscillatory behavior of sleeping brain
Frequency analysis is important tool to quantify neural activity during sleep. In mammals
recording of neural activity using EEG measurements has been used to define stages of sleep.
Oscillatory behavior is also believed to facilitate information processing and transfer of memories
from hippocampus to cortex. In simpler model systems oscillatory behavior also found during
sleep, for example, in fly (Yap et al. 2017).
To perform frequency analysis of activity, we segmented brain into super-voxels, as
described earlier in Methods section. Activity was averaged across all pixels in each super-voxels,
and processed to yield power spectra. For each brain region we averaged power spectra of all
super-voxels to provide information about average region-wise activity pattern.
In our data we observed that most of the changes in activity concentrated in hindbrain.
Our data also allows analysis of states based on time of the day. Day-time activity was higher
across the brain, compared to night time. This effect was observed both for wake and for sleep
states.
We did not observe any organized strong oscillations during sleep or wake, and when
comparing two conditions, at the time resolution that we used (0.33Hz). This is contrast to
mammalian EEG data, that shows significant presence of specific sleep-related frequency band
in neural activity. We observed decrease in overall activity during sleep, and especially in
hindbrain regions of brain. This was detected even though we removed time frames associated
with tail move and for 5 seconds after that (see Section 5.02), which suggest that hindbrain plays
56
important role in defining neural activity during wake and sleep in zebrafish outside of just
promoting motion.
Figure.5.7. Frequency analysis of sleep and wake bouts. (A) and (B) Circadian effects on sleep and wake
activity are widescread and ingeneral neural activity is depressed during night, both in sleep and wake. (C)
and (D) When comparing sleep and wake at any time of the day change in activity is limited to decreased
in hindbrain. N=4 fish, 10-30 bouts/fish, only significant changes (two-sample t-test p<0.05) are shown
(region anatomical labeling as in Fig.5.5).
Change in neural activity power was present in multiple animals. Decrease in power in
hindbrain and cerebellum are illustrated in Figure 5.8 (A). At the same time, zebrafish brain
doesn’t seem to be “quiet” during sleep, and some brain regions display similar power of activity
during sleep and wake, especially during day, as demonstrated in Figure 5.8 (B).
57
Figure 5.8. (A) Sample power spectra for night wake and night sleep states for 3 animals. Decrease in
power is noticeable in sleeping brain in right hindbrain and right cerebellum regions. The difference
between states was statistically significant in some bins (*), tested using two-sample t-test, p<0.05. (B)
Some zebrafish brain regions show similar power during sleep and wake, especially during day. One
example is the right habenula region, where difference was not significant in any bins. In all plots thick
lines are mean of the power averaged across 10-30 bouts in each animal. Shading is ±SEM
Section 5.04 Functional connectivity
Functional connectivity between brain regions is a standard tool to characterize brain
activity during specific behavior. In our experiments we collected correlation maps for 30-50
bouts of sleep and wake per animal. Correlation maps (Fig. 5.9) were averaged across animals
and compared between sleep and wake conditions.
58
Figure 5.9. Example of correlation maps of spontaneous activity in single fish during sleep and wake.
Night-time bouts are shown. Length of each bout and bout type is displayed on top of the web plots.
Colorbar coresponds to Z-score for correlation coefficents. N=4 fish, 30-50 bouts/fish. Significant
correlation coefficients are shown (p<0.05), tested using random shuffle test
We observed change in correlation profiles between wake and sleep periods. Few
observations should be highlighted (Figure 5.10). In sleeping zebrafish brain, we observed mostly
decrease in correlation between regions during sleep. This effect was persistent when comparing
night-time wake and night-time sleep, as well day-time sleep and day-time wake. The change in
correlation is brain-wide and not limited to single region or hemisphere of the brain. The pallium
59
brain region is not affected by sleep state, suggesting that its functional connectivity with the rest
of the brain doesn’t change in that behavior.
Figure 5.10. Averaged difference between correlation maps during wake and sleep. When comparing all
sleep bout and all wake bouts, we observed general decrease in correlation between brain regions during
sleep (N=4 fish, 30-50 bouts/fish, only significant difference shown, two-sample unpaired t-test p<0.05)
The decrease in correlation coefficients between wake and sleep was present in multiple
animals, as illustrated in Figure 5.11A, highlighting connectivity of left cerebellum and right
hindbrain. We observed similar but stronger decrease in connectivity when comparing sleep
during day and sleep during night (Fig.5.11B).
60
Figure 5.11. Correlation decrease during sleep and durign ngiht. (A)Example in correlation change
between left cerebellum and right hindbrain when comparing night-time wake and sleep. (Right)
Averaged correlation coefficient (transformed into Z score) for 3 animals. The difference was statistically
significant (p<0.05, one-tailed paired t-test). (Left) Difference was present in multiple animals, and
illustrated by histograms of correlation coefficients. For each animal the difference was significant
(p<0.05, Kolmogoro-Smirnov test) (B) Same analysis was performed for differences between day-time
sleep and night-time sleep, showing stronger decrease in correlation, for example, between right
cerebellum and right hindbrain regions.
Section 5.05 Statistical analysis
61
To perform statistical analysis of correlations between brain regions, we used two-sample
t-test and compared mean measurements from different samples. We didn’t use false-discovery
rate FDR(FDR) correction (Benjamini and Hochberg 1995) due to small number of samples (low
power of our studies). FDR correction is appropriate to use when comparing correlation
coefficients from brain activity data (Ling et al. 2009), due to large number of measurements that
are not independent, there is a chance of false-negative (accidental rejection of null-hypothesis)
result, or discovery of false difference between correlation coefficients. This is especially
important when analyzing data that is not independent, for example when looking for change in
correlation between brain region A and B as well as B and C and A and C. Correlation between
brain regions can’t be treated as independent variables (Worsley et al. 2002).
To take into account that our statistical tests can yield accidentally significant results, we
performed Kolmogorov-Smirnov 2-sample test on the correlation coefficients (Simpson et al.
2013) extracted from each sample separately, comparing single animal to itself. This allowed us
to test whether correlation coefficients, for example, during wake and sleep come from the same
distribution. We only showed results that were significant when comparing averages as well as
when considering individual animals. demonstrated that difference between sleep and wake
states is not accidental, but present across different animals.
For the frequency analysis, we analyzed data on a bin-by-bin basis, comparing again
samples to itself, and highlighting relative change in power for each bin. We used two-sample t-
test to first find bins where power change between states for each animal separately. To analyze
data together, we calculated mean change in power for each animal, and then pooled these
together and tested the ratio against the null-hypothesis that there is no change (ratio of 1).
When discussing results, it is important to realize that we can observe spurious significance due
to large number of bins, and non-independent nature of the data, so we focused on the changes
that happen across large parts of the spectrum. Changes that are statistically significant but
present only in narrow frequency bin have high chance of being false-positives.
62
Section 5.06 Application of melatonin protocol
Melatonin has become very popular as a safe choice for assisting with mild sleep
disorders. Currently melatonin market is estimated to be around 1.3 billion USD, however part
of its popularity might stem from low magnitude of the effect. How meta-analysis of 19 studies
has demonstrated (Ferracioli-Oda et al. 2013), effects of melatonin on sleep parameters such as
duration, latency, and overall quality, are mild at best. For example, in these studies, averaged
sleep duration, while though significant, was merely 8 minutes longer in experimental group.
Such ambiguous results, as well as shaky foundation for melatonin prescription
recommendations, probably stem from lack of comprehensive biological understanding of
melatonin action as endogenous hormone, as well as medication or supplement. Even though
genetics and effects on sleep of melatonin are studied well, for example in larval zebrafish
(Gandhi et al. 2015), the effects on brain activity during sleep are not addressed. Since sleep is
involving whole brain and involves change in interactions between many brain regions and
neurons, it would be extremely interesting to investigate how melatonin treatment affects those.
However, in our literature review we could not locate single study addressing effects of
melatonin on difference in brain-wide EEG, fMRI-based activity measurements, or other in
natural sleep and melatonin-assisted sleep. Some parameters of sleep, such as time spent in each
sleep stage, were measured in humans (Rajaratnam et al. 2004) and no difference between
placebo and melatonin treatment were found.
The goal of this work is to establish an approach, using zebrafish animal model, that would
allow understanding how melatonin-assisted sleep is changing brain-wide neural activity during
sleep. We seek to analyze how neuronal interactions change when sleep is aided, compared to
natural sleep, as well as how the neural activity the following day is different. To achieve that we
performed two types of experiments in zebrafish larvae: acute application of melatonin in the
middle of the day.
Melatonin experiments followed acute application protocol. We imaged spontaneous
brain activity for 1 hour, after which melatonin (10uM) or vehicle (DMSO) was added directly into
63
imaging chamber. After that the imaging continued for an additional 1 hour. The imaging was
performed using the same parameters as with the natural sleep experiments.
The behavior of zebrafish after application of melatonin has been studied (Zhdanova et
al. 2001) and demonstrated that animals rapidly fall asleep after addition of melatonin in the
water. In our experiments we observed that animals after addition of melatonin reduce number
of tail-motion events and subsequently start display sleep bouts.
Figure 5.11. Protocol for melatonin-induced sleep experiment. (Left) Animal tail motion and brain activity
were continously monitored for 2 hr, and drug was added to imaging chamber at. T=60min. (Right)
Addition of melatonin leads to increase in amount of immobility and subsequently sleep
In these pilot experiments we analyzed acute application of melatonin. Animals were
imaged for one or two hours before drug was added, after which we observed brain activity and
tail motion for another two hours (Figure 5.11). Comparison between data before and after
application was performed using metric described previously. We again compared, in control and
experimental animals, change in correlation maps between activity of 11 brain regions before
and after treatment. Interestingly, we observed striking difference in these changes when natural
sleep was compared with melatonin-induced sleep, even though melatonin application caused
significant increase in sleep behavior (measured through mobility observation).
64
Figure 5.12. Characterization of melatonin-induced sleep. N=2 fish, 5 bouts of wake and sleep/fish. Due to
small sample size we didn’t perform statistical analysis.
When compared to natural sleep, melatonin-induced sleep display change in correlation
relationship between brain regions, and different change in activity spectra (5.12). Melatonin-
induced sleep is characterized by increased intra-regional correlations, where in natural sleep
brain regions are in general became more uncoupled during sleep. Melatonin induced brain-wide
depression of activity, as demonstrated by decrease in power across all power bands in frequency
analysis. These results are providing pilot protocol for more detailed study of difference between
induced and natural sleep, and presented here as an illustration of the abilities our platform.
Section 5.07 Conclusions
Platform, presented here, provides first step in demonstrating feasibility and applicability
of whole-brain imaging during sleep in zebrafish model. It allows continuous recording of activity
from whole brain of behaving zebrafish for more than 24hrs, while preserving single-cell
resolution. The data that have been collected, allows first ever detailed description of neural
correlates of sleep state in zebrafish, and demonstrates potential of this approach to
investigation of sleep regulation by exogenous chemicals.
Our analysis pipeline allows investigation of oscillatory behavior of the sleeping zebrafish
brain. From other model systems it is known that characteristic brain oscillations on different
scales signify function of sleep behavior, and now we can put this information in context of
65
zebrafish sleep studies. In our work, investigating brain activity at frequency <0.3Hz, we did not
find any specific frequency bands that correlate with sleep and wake behavior, but did quantify
decrease in overall neural activity in sleeping zebrafish brain. We also observed that neural
activity in general higher during day, compared to night, both for wake and sleep states. Higher
temporal and spatial resolution description of oscillations would provide more insight into nature
and function of sleep in zebrafish model.
Correlated activity of the brain, on different scales, is a significant characteristic of
sleeping brain in mammals. Local and global correlations are believed to be one of the ways sleep
is regulating brain activity. In our work we looked at low-resolution correlations between
different brain regions and used this approach to describe difference between sleep and wake.
We found that brain regions are less correlated during sleep than during wake. This change is not
limited to just hindbrain and we observed decoupling across brain regions. Future work should
will concentrate on localized correlations within brain regions, as well as large-scale correlations
between individual cells. Functional connectivity approach should be taken to fully describe
effects of sleep on zebrafish brain.
Our framework is designed to help define and characterize natural and induce sleep
states. In our work we observed that sleep induced by melatonin creates patter of activity
different from natural sleep. The results carry low statistical power, and more data need to be
collected to definitely describe these differences. The protocol that we developed is robust and
easy to apply to study melatonin and other chemicals related to sleep regulation. This work
should be expanded to introduce new states of brain, such as anesthesia. We also need to define
more precisely what possible functional changes induction of sleep can create for sleep behavior.
Using functional correlative connectivity and oscillatory, we defined and characterized
behavioral states, based on time of day and type of behavior. We concentrated on splitting sleep
into 4 states, based on time of the day (day/night), and arousal (sleep/wake). These states are
characterized by specific correlation patterns, and this approach is a valuable starting point for
more detail description of zebrafish sleep. We envision that working with the data we collected,
66
we will be able to split sleep states during night, for example, into finer groups, perhaps
separating deep sleep from light sleep, hypothesized to exist in zebrafish (Prober et al. 2006).
Chapter VI. Discussion
Section 6.01 Current state of sleep science would benefit from bridging scales
Sleep is a complex behavior with wide-ranging effects and various regulatory
mechanisms. Its effects are widespread, but current tools used for understanding sleep lack
either resolution or scale. Neural activity, from level of single neurons, to circuits, to whole brain,
plays important role in regulation of sleep, as well as pertains a target for sleep effects. Modern
techniques to understand connection between sleep and neural activity, provide information by
averaging large populations of cells, or by allowing insight into activity of just few separate cells.
There is a gap between scale and resolution, and functional MRI and electrophysiology are
examples of extreme of that spectrum. Here we are laying ground for future work to fill the gap
between scale and resolution, providing tool that would allow investigation of sleeping brain
activity on whole-brain scale with single-cell resolution.
Experiments that record activity on large scale, averaging information from many
neurons, lose information about individual neurons, and locality of sleep. We also are not able to
measure contribution of precise neural circuits or individual cells within circuits when activity of
hundreds of cells is averaged, although many sleep-regulating circuits show redundancy of
activity and cells within it.
Sleep is an inherently large-scale behavior, that affects multiple systems on different
scales. Sleep is required for proper function of single synapses and neurons, re-configuration of
information within whole brain regions, and modulation of whole organism’s behavior. Sleep is
regulated by multiple circuits with brain-wide connections. Altogether, global behavior as sleep
needs to be studied on a large scale while preserving individuality of single cells to come up with
comprehensive model.
Goal of this work was to provide toolbox that can potentially answer that need. We aimed
to bridge the gap between scale and resolution and provide approach to analyze brain activity
67
on whole-brain scale with single-cell resolution in sleeping and awake animal. Modern optical
imaging in zebrafish model system provide an opportunity for us to achieve these goals.
Section 6.02 Zebrafish and imaging provide the answer
Brain imaging in zebrafish is rapidly developing, highlighted by recent publications that
explore whole-brain neural circuits with single-cell resolution, and uncover functional networks
regulating complex behaviors. We take advantage of these developments, providing novel
platform for imaging brain in sleeping zebrafish. The data quality that we achieve is limited by
the analysis pipeline, however provides an opportunity to create first description and
classification of wake and sleep states in zebrafish using brain activity data.
Zebrafish model system is well-suited for understanding function and nature of sleep in
vertebrates. The field is relatively new, which open a lot of opportunity for innovation, but work
has already been done to bridge sleep in zebrafish and known data from other model systems,
including mammals. Zebrafish model system provide unique opportunities for studying sleep
regulation, and especially suitable for imaging experiments. Before this work, we only seen rare
example of analysis of activity of specific circuits, such as hypocretin.
This work brings together field of advanced imaging and sleep based on zebrafish model
system. This is a novel approach, and we aim to lay the groundwork and provide platform for
future experiments. The richness of sleep science and novelty of zebrafish model system for
whole-brain imaging and sleep studies makes this work enabling results. There are many more
questions that will be answerable than description of sleep and wake states in zebrafish.
Several things are still missing in the field of zebrafish brain imaging and sleep studies.
One of the most important are perhaps introduction of robust and standardized sleep deprivation
assay. Currently, there are no such result, and several published results are not giving confidence
that sleep deprivation assay in larval zebrafish is easy to establish and use. Ideally, we would like
to see such assay as a part of imaging platform, presented here, so that we can observe, in real
time, effects of sleep deprivation on sleep behavior and brain activity in living zebrafish. Similar
comment can be made about learning assay, or a method to test memory formation and
consolidation. To date there are several published protocols for induction of memory in larval
68
zebrafish, but additional work on standardization and implementation of learning and imaging
tool is necessary.
This work introduces area of brain imaging to sleep research in zebrafish, and even these
limited results and techniques already provide path for characterization and understanding of
connection between sleep and neural activity in vertebrates. We also show that this platform
allows investigation of sleep control, and detailed answer to the question of nature of sleep
induced or regulated by each specific circuit. Given enough time and resources, it is feasible to
test each known and newly discovered circuit and define sleep states induced by each of them.
Section 6.03 Imaging sleeping vertebrate brain at scale with cellular resolution
Imaging of the whole-brain activity in behaving zebrafish has been an established field for
more than five years. It has rapidly developed and now truly reaching single-cell, whole-brain
level of imaging and analysis. With advancement of these techniques, more and more
experiments become possible. This work provides an example of such experiments, that can
potentially advance not only field of zebrafish neuroscience, but wider understanding of brain
function in vertebrate systems.
Here, we provide analysis pipeline, that focuses on segmenting brain into large regions,
and often averaging activity from large number of neurons. In no way this fully utilizes capabilities
of our approach and even data that has been collected. Better analysis pipeline would include
true single-cell segmentation, as well as dimensionality reduction approaches (such as PCA) that
does not require segmentation but can detect behavior-related cell populations.
We also elected to condense all possible sleep stages into just four, without trying to
statistically separate sleep and wake into finer states. We know from other systems, that sleep is
heterogeneous (for example, there are 4 sleep stages in human), and it is still unknown whether
sleep stages are present in zebrafish. The way we see to solve it, is clustering of sleep bouts based
on brain activity patters, functional connectivity, and oscillatory characteristics.
69
Section 6.04 Brain imaging with stimulation of sleep circuits
Sleep is regulated by redundant circuits and through different, sometimes independent
pathways. Investigation of neural circuits that regulate sleep and arousal usually involve
manipulation, recently mostly focused on optogenetic approaches. Zebrafish is uniquely capable
of providing tools for manipulation of neural circuit activity in precise fashion, combining
transgenesis techniques, as well as optical methods. We introduce manipulation of sleep-
regulating circuits with simultaneous quantification of brain activity.
One of sleep-regulated pathways is RFamid-containing neuropeptides, such as NPVF
peptide. These proteins regulate sleep in invertebrates and effect arousal state in mammals. Here
we developed protocol that allows measurement of brain activity when that circuit is stimulated,
in chronic and acute fashion. Our results show that changes in neural activity level follows change
in behavioral activity and increase in sleep observed in independent experiments.
In our work, we limit analysis to averaged level of neural activity. Our approach, though,
demonstrates that we can answer much more precise questions about effects of certain sleep
and wake-promoting circuits on brain activity. Thus, this work allows advanced description of
effects of specific neural circuits on information processing, sleep regulation, and associated
changes in neural activity. Extension of this work is descriptive work seeking defining of different
types of sleep, for example NPVF-induced sleep, hypocretin-induced wake, or glutamate-induced
sleep. This will allow precisely defining spectrum of states in vertebrate brain and investigation
of interplay and function role of each of these states.
Section 6.05 Tool to better understand induced sleep
Understanding of specific circuits that regulate sleep and wake potentially allows more
precise targeting with pharmaceuticals. TO date, we discover sleep medication by chance, and
without precise understanding of whole-brain effects. One example is widely available melatonin
promoting sleep in vertebrates. Application of current drugs and design of better chemicals
require more detailed picture of effects of the medication on behavior as well as overall brain
activity.
70
This work started as an experiment in defining and contrasting differences between
natural sleep and induced sleep. We demonstrated that melatonin, known to induced sleep in
zebrafish, does not induce same state of sleep, as natural sleep, when considering brain activity.
Difference in these states might shed light not only on function of endogenous melatonin, but
also effects of sleep aid based on that chemical. When combined with more complex behavioral
methods and sleep staging, this work will allow discovery of novel effects of existing sleep
medication or design and introduction of new chemicals.
Section 6.06 What’s next
Necessary for successful sleep study in larvae, sleep deprivation assay was also
demonstrated using controlled water flow as a stimulus (Aho et al. 2017). In these experiments,
authors showed that perturbation of animals’ sleep during night, negatively affects probability of
startle response and long-latency C-bend: key behavioral mechanisms of animal’s escape from
predators. Interestingly, faster response (short-latency C-bend) was not affected by rest
deprivation, probably because of different neuron circuitry of the reflex. Even though sleep
deprivation is reported in zebrafish larvae, it is still work in progress, as more studies need to be
performed using this technique. Sleep deprivation in larval zebrafish is still difficult task, in part
perhaps because of animal’s sensitivity to stimuli: too large perturbation can be deleterious to
animal’s health.
We would like to outline several next steps for extending this work. First, we need to take
full advantage of our datasets, performing single-cell segmentation to quantify brain activity on
the level of every individual neuron. In our work so far, we have been limited by analyzing data
from at least several neurons averaged together. Modern microscopy research employs powerful
tools to segment single neurons, isolating the signal from background, and providing complex
analysis with truly single-cell resolution. These tools are also complex to optimize for any specific
dataset, so this is an effort we would like to make in the future.
Sleep state in vertebrates is heterogeneous and there is an expectation that zebrafish
should display some of that diversity of states. We can attack that from behavioral standpoint,
looking closer at tail motion and length of sleep bouts, or we can ask how many states can be
71
separated from brain activity data, which has been done for humans. We expect that both
approaches will be useful. For example, we would like address separation of sleep into “light”
and “deep” sleep, reported in the zebrafish. Animals that react to light stimuli during sleep will
probably carry different brain activity pattern compared to animals with higher arousal
threshold. This will provide finer description of the sleep function.
We provide a blueprint for experiments to understand action of drugs related to sleep
behavior in zebrafish. This model system is very similar to higher vertebrates and we would like
to continue description of states induced by various chemicals that has been designed or that has
been discovered to have effects on sleep. As has been demonstrated with the epilepsy model in
zebrafish, this has potential of predicting chemicals that would have desired effect on brain
activity. We would look for chemicals that robustly induce sleep-like state with signature of brain
activity close to natural sleep state.
Finally, like chemically-induced sleep, we would like to describe spectrum of states
induced by stimulation or silencing of specific neural circuits. This includes continuation of work
with NPVF neurons, as well as focus on hypocretin cells, as well as dorsal raphe. This work will
provide insight into function and possibly interplay of separate sleep- and wake-promoting
circuits.
72
References
Adamantidis A, Carter MC, de Lecea L (2010) Optogenetic deconstruction of sleep-wake circuitry
in the brain. Front Mol Neurosci 2:31 . doi: 10.3389/neuro.02.031.2009
Adamantidis AR, Zhang F, Aravanis AM, Deisseroth K, de Lecea L (2007) Neural substrates of
awakening probed with optogenetic control of hypocretin neurons. Nature 450:420–424 .
doi: 10.1038/nature06310
Aho V, Vainikka M, Puttonen HAJ, Ikonen HMK, Salminen T, Panula P, Porkka-Heiskanen T,
Wigren H-K (2017) Homeostatic response to sleep/rest deprivation by constant water flow
in larval zebrafish in both dark and light conditions. J Sleep Res 26:394–400 . doi:
10.1111/jsr.12508
Ahrens MB, Li JM, Orger MB, Robson DN, Schier AF, Engert F, Portugues R (2012) Brain-wide
neuronal dynamics during motor adaptation in zebrafish. Nature 485:471–477 . doi:
10.1038/nature11057
Ahrens MB, Orger MB, Robson DN, Li JM, Keller PJ (2013) Whole-brain functional imaging at
cellular resolution using light-sheet microscopy. Nat Methods 10:413–420 . doi:
10.1038/nmeth.2434
Anafi RC, Kayser MS, Raizen DM (2019) Exploring phylogeny to find the function of sleep. Nat
Rev Neurosci 20:109–116 . doi: 10.1038/s41583-018-0098-9
Andersen LPH, Gögenur I, Rosenberg J, Reiter RJ (2016) The Safety of Melatonin in Humans.
Clin Drug Investig 36:169–175 . doi: 10.1007/s40261-015-0368-5
Appelbaum L, Wang GX, Maro GS, Mori R, Tovin A, Marin W, Yokogawa T, Kawakami K,
Smith SJ, Gothilf Y, Mignot E, Mourrain P (2009) Sleep-wake regulation and hypocretin-
melatonin interaction in zebrafish. Proc Natl Acad Sci U S A 106:21942–21947 . doi:
10.1073/pnas.906637106
Arrenberg AB, Del Bene F, Baier H (2009) Optical control of zebrafish behavior with
halorhodopsin. Proc Natl Acad Sci U S A 106:17968–17973 . doi:
10.1073/pnas.0906252106
Arthur D, Levin ED (2001) Spatial and non-spatial visual discrimination learning in zebrafish
(Danio rerio). Anim Cogn 4:125–131 . doi: 10.1007/s100710100111
Avants BB, Tustison NJ, Song G, Cook PA, Klein A, Gee JC (2011) A reproducible evaluation of
ANTs similarity metric performance in brain image registration. NeuroImage 54:2033–
2044 . doi: 10.1016/j.neuroimage.2010.09.025
Avitan L, Pujic Z, Mölter J, Poll MVD, Sun B, Teng H, Amor R, Scott EK, Goodhill GJ (2017)
Spontaneous Activity in the Zebrafish Tectum Reorganizes over Development and Is
73
Influenced by Visual Experience. Curr Biol 27:2407-2419.e4 . doi:
10.1016/j.cub.2017.06.056
Baker BJ, Kosmidis EK, Vucinic D, Falk CX, Cohen LB, Djurisic M, Zecevic D (2005) Imaging
Brain Activity With Voltage- and Calcium-Sensitive Dyes. Cell Mol Neurobiol 25:245–
282 . doi: 10.1007/s10571-005-3059-6
Benjamini Y, Hochberg Y (1995) Controlling the False Discovery Rate: A Practical and Powerful
Approach to Multiple Testing. J R Stat Soc Ser B Methodol 57:289–300
Bentivoglio M, Grassi-Zucconi G (1997) The pioneering experimental studies on sleep
deprivation. Sleep 20:570–576 . doi: 10.1093/sleep/20.7.570
Bhandiwad AA, Zeddies DG, Raible DW, Rubel EW, Sisneros JA (2013) Auditory sensitivity of
larval zebrafish (Danio rerio) measured using a behavioral prepulse inhibition assay. J Exp
Biol 216:3504–3513 . doi: 10.1242/jeb.087635
Bjorness TE, Greene RW (2009) Adenosine and Sleep. Curr Neuropharmacol 7:238–245 . doi:
10.2174/157015909789152182
Bryant PA, Trinder J, Curtis N (2004) Sick and tired: does sleep have a vital role in the immune
system? Nat Rev Immunol 4:457 . doi: 10.1038/nri1369
Cannell MB, Berlin JR, Lederer WJ (1987) Intracellular calcium in cardiac myocytes: calcium
transients measured using fluorescence imaging. Soc Gen Physiol Ser 42:201–214
Chen S, Chiu CN, McArthur KL, Fetcho JR, Prober DA (2016) TRP channel mediated neuronal
activation and ablation in freely behaving zebrafish. Nat Methods 13:147–150 . doi:
10.1038/nmeth.3691
Chen X, Mu Y, Hu Y, Kuan AT, Nikitchenko M, Randlett O, Chen AB, Gavornik JP, Sompolinsky
H, Engert F, Ahrens MB (2018) Brain-wide Organization of Neuronal Activity and
Convergent Sensorimotor Transformations in Larval Zebrafish. Neuron 100:876-890.e5 .
doi: 10.1016/j.neuron.2018.09.042
Chiu CN, Prober DA (2013) Regulation of zebrafish sleep and arousal states: current and
prospective approaches. Front Neural Circuits 7: . doi: 10.3389/fncir.2013.00058
Churchland MM, Shenoy KV (2007) Temporal Complexity and Heterogeneity of Single-Neuron
Activity in Premotor and Motor Cortex. J Neurophysiol 97:4235–4257 . doi:
10.1152/jn.00095.2007
Colwill RM, Creton R (2011) Imaging escape and avoidance behavior in zebrafish larvae. Rev
Neurosci 22:63–73 . doi: 10.1515/RNS.2011.008
Dang-Vu TT, Schabus M, Desseilles M, Albouy G, Boly M, Darsaud A, Gais S, Rauchs G,
Sterpenich V, Vandewalle G, Carrier J, Moonen G, Balteau E, Degueldre C, Luxen A,
74
Phillips C, Maquet P (2008) Spontaneous neural activity during human slow wave sleep.
Proc Natl Acad Sci U S A 105:15160–15165 . doi: 10.1073/pnas.0801819105
de Vivo L, Bellesi M, Marshall W, Bushong EA, Ellisman MH, Tononi G, Cirelli C (2017)
Ultrastructural evidence for synaptic scaling across the wake/sleep cycle. Science 355:507–
510 . doi: 10.1126/science.aah5982
Denk W, Strickler JH, Webb WW (1990) Two-photon laser scanning fluorescence microscopy.
Science 248:73–76
Dvir H, Elbaz I, Havlin S, Appelbaum L, Ivanov PC, Bartsch RP (2018) Neuronal noise as an
origin of sleep arousals and its role in sudden infant death syndrome. Sci Adv 4:eaar6277
. doi: 10.1126/sciadv.aar6277
Eaton RC, Lee RK, Foreman MB (2001) The Mauthner cell and other identified neurons of the
brainstem escape network of fish. Prog Neurobiol 63:467–485
Emran F, Rihel J, Adolph AR, Dowling JE (2010) Zebrafish larvae lose vision at night. Proc Natl
Acad Sci U S A 107:6034–6039 . doi: 10.1073/pnas.0914718107
Faraco JH, Appelbaum L, Marin W, Gaus SE, Mourrain P, Mignot E (2006) Regulation of
hypocretin (orexin) expression in embryonic zebrafish. J Biol Chem 281:29753–29761 .
doi: 10.1074/jbc.M605811200
Ferracioli-Oda E, Qawasmi A, Bloch MH (2013) Meta-analysis: melatonin for the treatment of
primary sleep disorders. PloS One 8:e63773 . doi: 10.1371/journal.pone.0063773
Fetcho JR, O’Malley DM (1995) Visualization of active neural circuitry in the spinal cord of intact
zebrafish. J Neurophysiol 73:399–406 . doi: 10.1152/jn.1995.73.1.399
Freeman JA, Schmidt JT, Oswald RE (1980) Effect of α-bungarotoxin on retinotectal synaptic
transmission in the goldfish and the toad. Neuroscience 5:929–942 . doi: 10.1016/0306-
4522(80)90161-X
Gandhi AV, Mosser E, Oikonomou G, Prober DA (2015) Melatonin is required for the circadian
regulation of sleep. Neuron 85:1193–1199 . doi: 10.1016/j.neuron.2015.02.016
Haesemeyer M, Robson DN, Li JM, Schier AF, Engert F (2015) The structure and timescales of
heat perception in larval zebrafish. Cell Syst 1:338–348 . doi: 10.1016/j.cels.2015.10.010
Hendricks JC, Sehgal A, Pack AI (2000) The need for a simple animal model to understand sleep.
Prog Neurobiol 61:339–351
Hires SA, Zhu Y, Tsien RY (2008) Optical measurement of synaptic glutamate spillover and
reuptake by linker optimized glutamate-sensitive fluorescent reporters. Proc Natl Acad Sci
105:4411–4416 . doi: 10.1073/pnas.0712008105
75
Horovitz SG, Braun AR, Carr WS, Picchioni D, Balkin TJ, Fukunaga M, Duyn JH (2009)
Decoupling of the brain’s default mode network during deep sleep. Proc Natl Acad Sci
106:11376–11381 . doi: 10.1073/pnas.0901435106
Huber R, Ghilardi MF, Massimini M, Tononi G (2004a) Local sleep and learning. Nature 430:78–
81 . doi: 10.1038/nature02663
Huber R, Hill SL, Holladay C, Biesiadecki M, Tononi G, Cirelli C (2004b) Sleep Homeostasis in
Drosophila Melanogaster. Sleep 27:628–639 . doi: 10.1093/sleep/27.4.628
Huisken J, Swoger J, Bene FD, Wittbrodt J, Stelzer EHK (2004) Optical Sectioning Deep Inside
Live Embryos by Selective Plane Illumination Microscopy. Science 305:1007–1009 . doi:
10.1126/science.1100035
Irion U, Krauss J, Nüsslein-Volhard C (2014) Precise and efficient genome editing in zebrafish
using the CRISPR/Cas9 system. Dev Camb Engl 141:4827–4830 . doi:
10.1242/dev.115584
Jacques SL (2013) Optical properties of biological tissues: a review. Phys Med Biol 58:R37-61 .
doi: 10.1088/0031-9155/58/11/R37
Ji N, Magee JC, Betzig E (2008) High-speed, low-photodamage nonlinear imaging using passive
pulse splitters. Nat Methods 5:197–202 . doi: 10.1038/nmeth.1175
Kaslin J, Nystedt JM, Östergård M, Peitsaro N, Panula P (2004) The Orexin/Hypocretin System
in Zebrafish Is Connected to the Aminergic and Cholinergic Systems. J Neurosci 24:2678–
2689 . doi: 10.1523/JNEUROSCI.4908-03.2004
Kaufmann A, Mickoleit M, Weber M, Huisken J (2012) Multilayer mounting enables long-term
imaging of zebrafish development in a light sheet microscope. Dev Camb Engl 139:3242–
3247 . doi: 10.1242/dev.082586
Kim CK, Miri A, Leung LC, Berndt A, Mourrain P, Tank DW, Burdine RD (2014) Prolonged,
brain-wide expression of nuclear-localized GCaMP3 for functional circuit mapping. Front
Neural Circuits 8:138 . doi: 10.3389/fncir.2014.00138
Kim DH, Kim J, Marques JC, Grama A, Hildebrand DGC, Gu W, Li JM, Robson DN (2017) Pan-
neuronal calcium imaging with cellular resolution in freely swimming zebrafish. Nat
Methods 14:1107–1114 . doi: 10.1038/nmeth.4429
Kohashi T, Oda Y (2008) Initiation of Mauthner- or non-Mauthner-mediated fast escape evoked
by different modes of sensory input. J Neurosci Off J Soc Neurosci 28:10641–10653 . doi:
10.1523/JNEUROSCI.1435-08.2008
Lee DA, Andreev A, Truong TV, Chen A, Hill AJ, Oikonomou G, Pham U, Hong YK, Tran S,
Glass L, Sapin V, Engle J, Fraser SE, Prober DA (2017) Genetic and neuronal regulation
of sleep by neuropeptide VF. eLife 6: . doi: 10.7554/eLife.25727
76
Li Y, Hao Y, Fan F, Zhang B (2018) The Role of Microbiome in Insomnia, Circadian Disturbance
and Depression. Front Psychiatry 9: . doi: 10.3389/fpsyt.2018.00669
Lin X, Duan X, Jacobs C, Ullmann J, Chan C-Y, Chen S, Cheng S-H, Zhao W-N, Poduri A, Wang
X, Haggarty SJ, Shi P (2018) High-throughput brain activity mapping and machine
learning as a foundation for systems neuropharmacology. Nat Commun 9:5142 . doi:
10.1038/s41467-018-07289-5
Ling XB, Cohen H, Jin J, Lau I, Schilling J (2009) FDR made easy in differential feature discovery
and correlation analyses. Bioinformatics 25:1461–1462 . doi:
10.1093/bioinformatics/btp176
Logothetis NK (2003) The Underpinnings of the BOLD Functional Magnetic Resonance Imaging
Signal. J Neurosci 23:3963–3971 . doi: 10.1523/JNEUROSCI.23-10-03963.2003
Lorimier RMD, Smith JJ, Dwyer MA, Looger LL, Sali KM, Paavola CD, Rizk SS, Sadigov S,
Conrad DW, Loew L, Hellinga HW (2002) Construction of a fluorescent biosensor family.
Protein Sci 11:2655–2675 . doi: 10.1110/ps.021860
Marvin M (1961) Microscopy apparatus
Mayer MG (1930) Uber Elementarakte mit zwei Quantem sprungen
McCombs JE, Palmer AE (2008) Measuring calcium dynamics in living cells with genetically
encodable calcium indicators. Methods 46:152–159 . doi: 10.1016/j.ymeth.2008.09.015
McIntosh AR, Bookstein FL, Haxby JV, Grady CL (1996) Spatial pattern analysis of functional
brain images using partial least squares. NeuroImage 3:143–157 . doi:
10.1006/nimg.1996.0016
McLean DL, Fetcho JR (2004) Ontogeny and innervation patterns of dopaminergic, noradrenergic,
and serotonergic neurons in larval zebrafish. J Comp Neurol 480:38–56 . doi:
10.1002/cne.20280
McMahon DG (1994) Modulation of electrical synaptic transmission in zebrafish retinal horizontal
cells. J Neurosci 14:1722–1734 . doi: 10.1523/JNEUROSCI.14-03-01722.1994
Medic G, Wille M, Hemels ME (2017) Short- and long-term health consequences of sleep
disruption. Nat Sci Sleep 9:151–161 . doi: 10.2147/NSS.S134864
Meeker ND, Hutchinson SA, Ho L, Trede NS (2007) Method for isolation of PCR-ready genomic
DNA from zebrafish tissues. BioTechniques 43:610, 612, 614 . doi: 10.2144/000112619
Mieda M, Hasegawa E, Kisanuki YY, Sinton CM, Yanagisawa M, Sakurai T (2011) Differential
roles of orexin receptor-1 and -2 in the regulation of non-REM and REM sleep. J Neurosci
Off J Soc Neurosci 31:6518–6526 . doi: 10.1523/JNEUROSCI.6506-10.2011
77
Monroe JD, Manning DP, Uribe PM, Bhandiwad A, Sisneros JA, Smith ME, Coffin AB (2016)
Hearing sensitivity differs between zebrafish lines used in auditory research. Hear Res
341:220–231 . doi: 10.1016/j.heares.2016.09.004
Mukamel EA, Nimmerjahn A, Schnitzer MJ (2009) Automated analysis of cellular signals from
large-scale calcium imaging data. Neuron 63:747–760 . doi: 10.1016/j.neuron.2009.08.009
Muto A, Ohkura M, Abe G, Nakai J, Kawakami K (2013) Real-time visualization of neuronal
activity during perception. Curr Biol CB 23:307–311 . doi: 10.1016/j.cub.2012.12.040
Naumann EA, Kampff AR, Prober DA, Schier AF, Engert F (2010) Monitoring neural activity
with bioluminescence during natural behavior. Nat Neurosci 13:513–520 . doi:
10.1038/nn.2518
Nicolson T, Rüsch A, Friedrich RW, Granato M, Ruppersberg JP, Nüsslein-Volhard C (1998)
Genetic analysis of vertebrate sensory hair cell mechanosensation: the zebrafish circler
mutants. Neuron 20:271–283
Pantoja C, Hoagland A, Carroll E, Schoppik D, Isacoff EY (2017) Measuring Behavioral
Individuality in the Acoustic Startle Behavior in Zebrafish. Bio-Protoc 7: . doi:
10.21769/BioProtoc.2200
Pantoja C, Hoagland A, Carroll EC, Karalis V, Conner A, Isacoff EY (2016) Neuromodulatory
Regulation of Behavioral Individuality in Zebrafish. Neuron 91:587–601 . doi:
10.1016/j.neuron.2016.06.016
Pech U, Revelo NH, Seitz KJ, Rizzoli SO, Fiala A (2015) Optical dissection of experience-
dependent pre- and postsynaptic plasticity in the Drosophila brain. Cell Rep 10:2083–2095
. doi: 10.1016/j.celrep.2015.02.065
Peyrache A, Dehghani N, Eskandar EN, Madsen JR, Anderson WS, Donoghue JA, Hochberg LR,
Halgren E, Cash SS, Destexhe A (2012) Spatiotemporal dynamics of neocortical excitation
and inhibition during human sleep. Proc Natl Acad Sci 109:1731–1736 . doi:
10.1073/pnas.1109895109
Pezaris JS, Sahani M, Andersen RA (1997) Tetrodes for Monkeys. In: Bower JM (ed)
Computational Neuroscience: Trends in Research, 1997. Springer US, Boston, MA, pp
937–942
Pillitteri JL, Kozlowski LT, Person DC, Spear ME (1994) Over-the-counter sleep aids: Widely
used but rarely studied. J Subst Abuse 6:315–323 . doi: 10.1016/S0899-3289(94)90502-9
Podgorski K, Ranganathan G (2016) Brain heating induced by near-infrared lasers during
multiphoton microscopy. J Neurophysiol 116:1012–1023 . doi: 10.1152/jn.00275.2016
Prober DA, Rihel J, Onah AA, Sung R-J, Schier AF (2006) Hypocretin/orexin overexpression
induces an insomnia-like phenotype in zebrafish. J Neurosci Off J Soc Neurosci 26:13400–
13410 . doi: 10.1523/JNEUROSCI.4332-06.2006
78
Raizen DM, Zimmerman JE, Maycock MH, Ta UD, You Y, Sundaram MV, Pack AI (2008)
Lethargus is a Caenorhabditis elegans sleep-like state. Nature 451:569–572 . doi:
10.1038/nature06535
Rajaratnam SMW, Middleton B, Stone BM, Arendt J, Dijk D-J (2004) Melatonin advances the
circadian timing of EEG sleep and directly facilitates sleep without altering its duration in
extended sleep opportunities in humans. J Physiol 561:339–351 . doi:
10.1113/jphysiol.2004.073742
Randlett O, Wee CL, Naumann EA, Nnaemeka O, Schoppik D, Fitzgerald JE, Portugues R,
Lacoste AMB, Riegler C, Engert F, Schier AF (2015) Whole-brain activity mapping onto
a zebrafish brain atlas. Nat Methods 12:1039–1046 . doi: 10.1038/nmeth.3581
Rattenborg NC, Lima SL, Amlaner CJ (1999) Half-awake to the risk of predation. Nature 397:397–
398 . doi: 10.1038/17037
Rihel J, Prober DA, Arvanites A, Lam K, Zimmerman S, Jang S, Haggarty SJ, Kokel D, Rubin
LL, Peterson RT, Schier AF (2010) Zebrafish behavioral profiling links drugs to biological
targets and rest/wake regulation. Science 327:348–351 . doi: 10.1126/science.1183090
Roberts AC, Chornak J, Alzagatiti JB, Ly DT, Bill BR, Trinkeller J, Pearce KC, Choe RC,
Campbell CS, Wong D, Deutsch E, Hernandez S, Glanzman DL (2019) Rapid habituation
of a touch-induced escape response in Zebrafish (Danio rerio) Larvae. PLOS ONE
14:e0214374 . doi: 10.1371/journal.pone.0214374
Robinson DH, Toledo AH (2012) Historical development of modern anesthesia. J Investig Surg
Off J Acad Surg Res 25:141–149 . doi: 10.3109/08941939.2012.690328
Salzberg BM, Davila HV, Cohen LB (1973) Optical Recording of Impulses in Individual Neurones
of an Invertebrate Central Nervous System. Nature 246:508 . doi: 10.1038/246508a0
Saper CB, Fuller PM (2017) Wake-sleep circuitry: an overview. Curr Opin Neurobiol 44:186–192
. doi: 10.1016/j.conb.2017.03.021
Scammell TE, Arrigoni E, Lipton JO (2017) Neural Circuitry of Wakefulness and Sleep. Neuron
93:747–765 . doi: 10.1016/j.neuron.2017.01.014
Sepehri Rad M, Choi Y, Cohen LB, Baker BJ, Zhong S, Storace DA, Braubach OR (2017) Voltage
and Calcium Imaging of Brain Activity. Biophys J 113:2160–2167 . doi:
10.1016/j.bpj.2017.09.040
Serbedzija GN, Flynn E, Willett CE (1999) Zebrafish angiogenesis: A new model for drug
screening. Angiogenesis 3:353–359 . doi: 10.1023/A:1026598300052
Shepard JW, Buysse DJ, Chesson AL, Dement WC, Goldberg R, Guilleminault C, Harris CD, Iber
C, Mignot E, Mitler MM, Moore KE, Phillips BA, Quan SF, Rosenberg RS, Roth T,
Schmidt HS, Silber MH, Walsh JK, White DP (2005) History of the Development of Sleep
79
Medicine in the United States. J Clin Sleep Med JCSM Off Publ Am Acad Sleep Med
1:61–82
Simpson SL, Lyday RG, Hayasaka S, Marsh AP, Laurienti PJ (2013) A permutation testing
framework to compare groups of brain networks. Front Comput Neurosci 7: . doi:
10.3389/fncom.2013.00171
Singh C, Oikonomou G, Prober DA (2015) Norepinephrine is required to promote wakefulness
and for hypocretin-induced arousal in zebrafish. eLife 4:e07000 . doi: 10.7554/eLife.07000
Smetters D, Majewska A, Yuste R (1999) Detecting Action Potentials in Neuronal Populations
with Calcium Imaging. Methods 18:215–221 . doi: 10.1006/meth.1999.0774
Stoltzfus CR, Barnett LM, Drobizhev M, Wicks G, Mikhaylov A, Hughes TE, Rebane A (2015)
Two-photon directed evolution of green fluorescent proteins. Sci Rep 5:11968 . doi:
10.1038/srep11968
Thompson AW, Vanwalleghem GC, Heap LA, Scott EK (2016) Functional Profiles of Visual-,
Auditory-, and Water Flow-Responsive Neurons in the Zebrafish Tectum. Curr Biol CB
26:743–754 . doi: 10.1016/j.cub.2016.01.041
Truong TV, Supatto W, Koos DS, Choi JM, Fraser SE (2011) Deep and fast live imaging with
two-photon scanned light-sheet microscopy. Nat Methods 8:757–760 . doi:
10.1038/nmeth.1652
Vanwalleghem G, Heap LA, Scott EK (2017) A profile of auditory-responsive neurons in the
larval zebrafish brain. J Comp Neurol 525:3031–3043 . doi: 10.1002/cne.24258
Vladimirov N, Mu Y, Kawashima T, Bennett DV, Yang C-T, Looger LL, Keller PJ, Freeman J,
Ahrens MB (2014) Light-sheet functional imaging in fictively behaving zebrafish. Nat
Methods 11:883–884 . doi: 10.1038/nmeth.3040
Weber F, Chung S, Beier KT, Xu M, Luo L, Dan Y (2015) Control of REM sleep by ventral
medulla GABAergic neurons. Nature 526:435–438 . doi: 10.1038/nature14979
Weber F, Dan Y (2016) Circuit-based interrogation of sleep control. Nature 538:51–59 . doi:
10.1038/nature19773
Williams DA (1990) Quantitative intracellular calcium imaging with laser-scanning confocal
microscopy. Cell Calcium 11:589–597 . doi: 10.1016/0143-4160(90)90013-K
Wolf S, Dubreuil AM, Bertoni T, Böhm UL, Bormuth V, Candelier R, Karpenko S, Hildebrand
DGC, Bianco IH, Monasson R, Debrégeas G (2017) Sensorimotor computation underlying
phototaxis in zebrafish. Nat Commun 8:651 . doi: 10.1038/s41467-017-00310-3
Wolf S, Supatto W, Debrégeas G, Mahou P, Kruglik SG, Sintes J-M, Beaurepaire E, Candelier R
(2015) Whole-brain functional imaging with two-photon light-sheet microscopy. Nat
Methods 12:379–380 . doi: 10.1038/nmeth.3371
80
Worsley KJ, Liao CH, Aston J, Petre V, Duncan GH, Morales F, Evans AC (2002) A General
Statistical Analysis for fMRI Data. NeuroImage 15:1–15 . doi: 10.1006/nimg.2001.0933
Xie L, Kang H, Xu Q, Chen MJ, Liao Y, Thiyagarajan M, O’Donnell J, Christensen DJ, Nicholson
C, Iliff JJ, Takano T, Deane R, Nedergaard M (2013) Sleep Drives Metabolite Clearance
from the Adult Brain. Science 342:373–377 . doi: 10.1126/science.1241224
Xu X, Scott-Scheiern T, Kempker L, Simons K (2007) Active avoidance conditioning in zebrafish
(Danio rerio). Neurobiol Learn Mem 87:72–77 . doi: 10.1016/j.nlm.2006.06.002
Yap MHW, Grabowska MJ, Rohrscheib C, Jeans R, Troup M, Paulk AC, van Alphen B, Shaw PJ,
van Swinderen B (2017) Oscillatory brain activity in spontaneous and induced sleep stages
in flies. Nat Commun 8:1815 . doi: 10.1038/s41467-017-02024-y
Yokogawa T, Marin W, Faraco J, Pézeron G, Appelbaum L, Zhang J, Rosa F, Mourrain P, Mignot
E (2007) Characterization of sleep in zebrafish and insomnia in hypocretin receptor
mutants. PLoS Biol 5:e277 . doi: 10.1371/journal.pbio.0050277
Zhdanova IV (2006) Sleep in zebrafish. Zebrafish 3:215–226 . doi: 10.1089/zeb.2006.3.215
Zhdanova IV, Wang SY, Leclair OU, Danilova NP (2001) Melatonin promotes sleep-like state in
zebrafish. Brain Res 903:263–268
Zon LI, Peterson RT (2005) In vivo drug discovery in the zebrafish. Nat Rev Drug Discov 4:35 .
doi: 10.1038/nrd1606
81
Appendix
Section 1.01 First attempt at 3D imaging during sleep
At the time when this work started, there have been no demonstration of whole-brain
imaging in sleeping zebrafish. The whole field of high-resolution, whole-brain imaging in zebrafish
has been merely established with few published reports. It would be suitable to highlight it here,
as we had no experience in this field at all, but only good intuition, skill and experience in
microscopy and biophysics of fluorescence, and confidence that pursuing this problem will be
fruitful.
Our first experiment in 2014 was imaging of whole brain suing hSPIM imaging, by slowly
scanning whole brain through the course of night. It has produced an immediate result,
demonstrating that we can potentially follow neural activity over 15hrs, including 10 hours of
night, in living zebrafish, and even detect changes in neural activity levels, similar to circadian
cycle, and not driven only by the light/dark cycle.
82
Figure 8.0. Whole-brain imaging of Tg(HuC:GCaMP5G) performed with 6min/volume speed allowed us to
detect circadian-associated changes in neural activity across different brain regions.
We used transgenic zebrafish line expressing pan-neuronal cytoplasmic calcium indicator
GCaMP5G. In this experiment, we used very limited laser power, and scanned brain with high
resolution, performing one Z stack imaging in 6 minutes. The analysis was performed by
averaging fluorescent signal over large areas across different brain regions, including optic
tectum (orange), midbrain (green and grey), and hindbrain (yellow and blue).
What we observed is that neural activity changes across day and night and generally
follows circadian rhythm of the animal, and not limited to visual response due to the light we
used to simulate night and day conditions. We also observed that neural activity is differential,
that is parts of the brain show different changes due to night/day. However, these experiments
proven to be very difficult to reproduce, perhaps due to immaturity of our imaging protocol,
animal handling, and that scanning was performed by dragging the animal through the water
(albeit slowly). Nevertheless, these pilot experiments gave us confidence that it is possible, but
perhaps very difficult, to perform whole-brain imaging in zebrafish during night-day transitions,
and perhaps during sleep.
83
Section 1.02 Establishing assay for associative memory in adult zebrafish
Sleep play important role in learning and memory, as has been established in human
experiments and many model systems. Part of the work not included in the thesis is developing
of a learning assay for testing memory formation and consolidation in zebrafish. We started by
setting up assay for associative learning in adult zebrafish.
In these experiments, adult (5-7 months) zebrafish were placed in tanks that had LED light
source and large steel mesh electrodes (Figure 8.1). electrodes were connected to function
generator that provided 5V sinusoidal 60Hz pulses for 0.5 second as unconditional stimuli (US).
Conditional stimuli (LED light) was provided by triggering light source for 10 seconds. We usually
tested 4 fish a time, but the animals were completely isolated. Imaging was provided by
illuminating tanks with uniform invisible IR light, and activity was recorded at 30fps video rate
using custom microscope.
All animals went through habituation phase of 5 CS presentations. In experimental group,
CS stimuli was co-terminated with US during training (conditioning) phase. In control group only,
CS was present during training. Testing of memory was performed at the end of experiment by
presenting 5 CS stimuli and observing speed of animals. Animals were tracked from the movies
using custom MATLAB code developed by Dr. Reza Ardekani (code available at
https://github.com/aandreev0/NeuroScienceFishTracking).
84
Figure 8.1. Schematics of experimental setup and experimental protocol.
In our experiments were analyzed activity of 12 fish in control and experimental groups.
An example of animal has successfully associated LED light with electrical shock is shown in figure
8.2. We also plot maximum speed during CS presentation as metric of animal reactivity.
Experimental animals, that underwent CS+US training protocol, display larger maximal speed.
85
Figure 8.2. (Lefl) Speed of zebrafish undergoing habituation, training, and testing rounds. Red rectangular
wave represent period of LED light (CS), yellow bar represents application of electrical shock. (Right)
Maximum speed in experimental animals (CS+US during training) is higher than maximum speed of
control animals (CS only during training). Each dot represent single animal tested 5 times during testing
round.
Section 1.03 Brain imaging with 2P-SPIM under different stimuli
Change in arousal threshold is part of sleep definition in all model systems. Simply
speaking, it means that during sleep it is hard to wake animal, or it is harder to elicit response in
the animal. Hence, sleep state can be probed by stimulation, as has been done in zebrafish
studies of sleep []. There are number of different stimuli that can be used, roughly divided into
several subcategories, including but not limited to:
1. Visual stimuli, such as simple light flash or complex visual picture such as looming
stimuli (Arthur and Levin 2001)
2. Auditory stimulation, that includes sounds in 100-1000Hz range, as well as
vibrational stimulation (Eaton et al. 2001), (Nicolson et al. 1998), (Pantoja et al.
2017)
86
3. Electrical stimuli (Xu et al. 2007)
4. Heat-based stimulation (Haesemeyer et al. 2015)
5. Touch stimulation (Kohashi and Oda 2008), (Roberts et al. 2019)
We have performed experiments with several types of stimulation, all based on the same
platform and with similar protocols.
The core of the stimulation control is a microcontroller (MC) that regulates when the
stimuli is applied and for how long. We selected Arduino as our platform of choice, although
there are multiple alternatives. This device is extremely cheap and easy to program and operate.
We also noticed that these MCs are very durable, some serving in experiments for several years
without fault.
The problem we must solve is synchronization of stimuli delivery with imaging, as well as
recording of stimuli time. This has been resolved by tying imaging and stimulation algorithm
together. The imaging camera works in this case as a master clock, sending pulse with each frame
to the MC. MC then works in a counter mode, counting incoming pulses, and making decision of
whether it is time for stimuli delivery or not yet. Since our framerate is constant, and we can pre-
program MC to deliver stimuli at time, we are knowing exactly, ahead of time, which frames will
precede, and which will follow the stimulation period.
The stimuli are activated via different means, but generally each device that either
controls light output, or sound, or electrical shock, will have appropriate connectors to trigger
stimuli. For example, in case of broad light stimulation, the device (LED controller) has one input
connector that can be set in either HIGH or LOW state by the MC. Microcontroller is programmed
to switch state for precise number of frames or time.
87
Figure 8.3. Principal scheme of interaction between camera and microcontroller
The logic programmed in controller simply counts number of frames taken by camera and
performs necessary actions according to algorithm. Consider for example simple experiment with
using LED light as stimulation for 20 frames in the middle of 120-frame long probe bout, turning
it on at frame 50, and turning it off at frame 70. The logic can be outlines in pseudo-code as:
88
c = 0 // initialize counter
turn LED to LOW // initialize LED light
while True then
if pulse from camera received then
c += 1
if c >= 120 then
c = 0
end
if c == 50 then
turn LED to HIGH
end
if c==70 then
turn LED to LOW
end
end
The microcontroller as well can send additional information to the stimulator, for
example, related to which stimulation episode is that. For example, we might want to send 100%
of LED power on even trials, and only 50% on odd ones. This is done by sending additional signal
for modulation of stimuli intensity. In case of auditory stimulation, the interface was serial, that
is microcontroller would send a message such as “BUZZ” to PC that will in turn produce sound.
Serial connection allows transferring much more diverse information. The connection can also
be reversed, for example, for triggering stimulation based on some image analysis, for example,
after period of animal inactivity.
89
Visual stimulation was delivered to the animal via red LED, with peak intensity at 625nm
(Thorlabs). We used different modes of stimulation, including triangle and rectangular waves.
Microcontroller connected to LED controller with a coaxial cable with BNC connectors.
Auditory stimulation is a standard assay used to test zebrafish behavioral state, as it can
elicit very strong escape response (Pantoja et al. 2017). It is an important tool because auditory
stimuli can be delivered on scale to many animals studied at the same time in multi-well plates.
In these experiments, researcher would use high-speed camera and image analysis to quantify
escape response to sound, potentially screening drugs of genetic mutants (Colwill and Creton
2011). Auditory response of zebrafish also has been used as a model for studies of hearing
development, defects, and genetics (Bhandiwad et al. 2013).
Auditory stimulation is another modality that can be used to measure and quantify
zebrafish behavior and state, especially during sleep and wake. To understand viability of this
approach, we modified our microscope by using simple sound dynamic attached to a metal plate,
that transduced sound to the imaging chamber. We were unable at the time of experiments to
come up with better solution that would fit our setup, but specialized underwater speaker (such
as used in (Monroe et al. 2016)) would be ideal device to controllably project sound waves onto
animal during imaging. The sound was played from a custom Ruby script, triggered by Arduino
used in a frame counter mode, described previously. The code can be found in the Appendix.
90
Figure 8.4. Setup for delivering sound stimuli to the imaging chamber.
In this experiment we explored which regions of interest in zebrafish brain are activated
by sound. For example, when presented with 0.5 second-long 150Hz sound stimuli in the middle
of 60s fluorescent imaging period, we observed activation of neurons in hindbrain and optic
tectum, as demonstrated by temporal color-coded image. This visualization is useful to show
which neurons are activated specifically at certain point in time, as those are highlighted with
corresponding hue, in this case cyan.
91
Figure 8.5. Recording of auditory response in larval zebrafish.
When we performed this experiment initially, there was no published data on whole-
brain imaging of auditory response in zebrafish. However, our observation is consistent with
detailed study of neural activity due to sound stimuli (Vanwalleghem et al. 2017). Authors there
also observed activity due to sound in optic tectum and mostly in hindbrain, as well as
characterized activity patters with much higher anatomical resolution.
Vibration is another modality that has been used to stimulate zebrafish and evoke
response. Auditory and vibration response is the same modality; however, frequency of vibration
is usually lower compared to auditory stimulation. It was interesting for us to adopt vibration
protocol as modality of stimulation because vibration was used for sleep deprivation in larval
92
zebrafish [Zhdanova 2001]. There have been no reports of experiment measuring brain activity
response to vibration stimulation, so our pilot experiment would provide useful insight into
reliability and viability of using vibration as behavioral stimuli in sleeping zebrafish with
simultaneous imaging.
To apply vibration, we attached small DC vibrational motor, that constitutes off-centered
electrical motor, to the imaging chamber. Application of small current would then cause motor
to vibrate for duration of current. The vibration is passed to the sample indirectly: motion of the
motor is passed to water surrounding chamber, which transfers motion to the sample as sound
waves. As previously described, we used Arduino microcontroller to trigger vibration at specific
time by counting exposure of camera. The duration of vibration was varied in range 100ms-
700ms. The response was measured in hindbrain by averaging over large region of interest. We
observed that response, measured as GCaMP fluorescence change, correlated with duration of
the vibrational stimulation.
93
Figure 8.6. Modification of microscope chamber to allow vibration stimulation
Vibration stimulation is an important and easy-to-adapt protocol for imaging zebrafish. It
was easy for us to adapt and setup, and control with Arduino code is straight-forward. One
limitation is how much flexibility can be provided by off-balanced motor we used for stimulation.
These devices work in very narrow range of voltages, so we effectively cannot control power of
vibration or its spectral composition. Another possible limitation of this scheme is that motor is
attached to the imaging chamber that is connected, via soft membrane, to detection objective.
When active, motor can in turn cause undesirable vibration of the detection objective, inducing
type of “motion blur”. We did not notice significant loss of resolution due to that, but perhaps
higher intensities of vibration would cause such artifact.
94
Figure 8.7. Response to mechanical vibration in zebrafish brain. (Left) Colorcoded projection of single trial,
where color depicts timepoint of activity for each given neuron. A lot of activity in hindbrain (cyan) co-
incides with stimulation in the middle of the trial. (Right) Three trials with sequenctially increaing duration
of vibrational stimuli from 100ms to 700ms. Response progressively increase in all trials. Scale bar 100um
Providing vibrational stimuli to zebrafish larvae during brain imaging is feasible and can
be adopted alter as a tool to measure arousal threshold and behavioral response during sleep
and wake. It is also cheap and simple modification of any existing behavioral protocol, so this
approach can be used in other experiments with larval zebrafish.
95
Response to noxious, or painful stimuli, is a strong behavior that provides evolutionary
advantage. Escape response from unpleasant situation sometimes allows animal to survive. We
explored whether zebrafish shows strong behavioral response to laser-induced heating, with idea
to later use it as a tool to judge animal’s arousal or behavioral state. It is useful to note here as
well, that this experiment provided key insight into setting up associative learning, for different
project.
We started by adding 975nm laser to a custom wide-field microscope (details are
provided in paper associated with project). The laser was triggered by Arduino set up in timer
mode (script in Appendix). Laser spot was aligned to hit animal’s head, as depicted in the figure
below. Animal was immobilized in 1.5% agarose in 5cm Petri dish, with agarose around tail
removed. That allowed precise positioning of laser beam and simultaneous detection of tail
movement. Tail motion detection was done by calculating image difference trace in the manually
selected ROI.
Figure 8.8. Application of heating laser to elicit escape response from larval zebrafish
Our metric of response is tail motion, as zebrafish usual response to noxious stimuli is
escape. Indeed, when turning laser on, we observed that animal would twitch its tail vigorously.
To analyze that data, we calculated image difference of a movie of the tail, subtracting
consecutive frames from each other, then averaging over region of interest selected around the
tail. The movement of the tail ca can be visualized by two-dimensional plot where each line is a
single trial with the same animal. This setup allowed us to quantify average animal response
96
intensity, as well as delay of response compared to stimuli onset. In this example it is interesting
to observe that animal would become more sensitized as later trials take place.
The laser heating approach is a suitable tool to elicit response in zebrafish and it can
potentially be adopted on number of various microscopes. The simple analysis process we
employed already yields useful information about amount of tail activity in zebrafish, as well as
latency and duration. These data can potentially be used to characterize activity and combining
it with brain imaging would provide reach information about how escape response is regulated
and generated.
97
Figure 8.9. Example of heating-laser stimulation. Presentation of laser robustly causes tail deflection in
zebrafish. In this example we also observed sensitization, as latency of first twitch decreases with each
trial
Light-sheet microscopy provides a way for simultaneous acquisition of thin section of the
brain, and for three-dimensional coverage we need to scan this plane across axially (or in Z
direction). First, scanning can be done by moving detection optics and light sheet, or by moving
sample through the light sheet. Secondly, scanning in Z can be done in high-speed fashion, when
single Z volume is collected at each time point, or by repeating experiment, each time imaging at
different Z position. This approach, sometimes called “subframe interpolation”, was used for
famous “speed of light” demonstrations that shows light wave propagation in space around
object such as plastic bottle, claiming “trillion frames per second” [Picosecond Camera for Time-
98
of-Flight Imaging]. At each time point there, authors imaged single line of image with 1.7
picosecond exposure, taking image of scene illuminated by laser with 13ns period and very
narrow pulse duration.
Figure 8.10. Three-dimensional imaging require scanning. Scanning can be performed in two modes.
We used that approach to demonstrate how neural activity, elicited by stimuli can
potentially be recorded with high temporal or spatial resolution without need to quickly move
sample or light sheet. In that experiment, transgenic zebrafish expressing pan-neuronal
cytoplasmic calcium indicator GCaMP5G was used. Single trial was performed by imaging brain
at plane Z=Zi for 60 seconds at 1 frame per second, and LED illumination was turned on for 20
seconds at t=20s. Intra-trial interval was set at 3 minutes, and each trial was performed at
different Z position of light sheet.
After appropriate reorganization of acquired data, we arrived at pseudo-3D image where
each axial plane is collected at 3-minute interval, however due to reproducibility of response to
light, there is clear continuity of response across the brain without significant artifacts of
streaking. We observe strong response to dark-flash (LED turning off), especially in optic tectum.
99
Figure 8.11. Whole-brain activity recording of visual response with repetitive stimulation. Scale bar 100um
This experiment provides tool to record neural response in brain when imaging speed
might be a limitation. The necessary condition is that the response is highly reproducible, and the
animal’s state does not change much between imaging each plane. Similar approach has been
used in [All-optical electrophysiology in mammalian neurons using engineered microbial
rhodopsins] for imaging wave of depolarization in a single neuron, but, to the best of our
knowledge, never yet in living zebrafish brain. For example, combination of voltage indicators
and adoption of subframe interpolation would allow to investigate propagation of signal in living
zebrafish brain with millisecond temporal and sub-micron spatial resolution, while retaining
whole-brain coverage.
Section 1.04 Registration code and parameters
The query is used to registered files saved in fixedImageFile and movingImageFile would
look like:
100
The code is wrapped into a single Ruby function and then applied to each separate
volumetric time point in order to register 4D dataset to 3D template.
Section 1.05 Data storage and management
Hardware for image acquisition has important requirements. Because of the amount of
data collected in light-sheet experiments (up to 0.1-1Gb/s) it is appropriate to use fast solid-state
drives for immediate data storage. Acquisition computers rarely have a lot of storage space, so
fast, optical-fiber network should be built to transfer data to dedicated data storage server.
Current technology allows building relatively cheap ($5000 USD for 60TB) custom storage server,
based on robust FreeBSD operating system and error-resilient and extensible ZFS file system. In
our work we used RAIDZ2 (software version of what is known as RAID6) as a tradeoff between
data resiliency and cost-effectiveness. We used commercially available hardware for the storage
server (Supermicro) with initially 24 bays for hard-drives, but extendable by attaching 45-drive
bay via single SAS2 cable. Potentially this system can be expanded further without compromise
of performance up to 120 total hard drives, potentially bringing total capacity to 500TB.
cmd = 'antsRegistration.exe' +
' --dimensionality 3' +
' --float 0' +
' --output [' + outputFile + ',' + outputFile + '.nii.gz]' +
' --interpolation Linear' +
' --winsorize-image-intensities [0.005,0.995]' +
' --initial-moving-transform [' + fixedImageFile + ',' + movingImageFile + ',1]' +
' --use-histogram-matching 0' +
' --transform Rigid[0.1]' +
' --metric MI[' + fixedImageFile + ',' + movingImageFile +
',1,32,Regular,' + SamplingRate + ']' +
' --convergence [1000x500x250x100,1e-6,10]' +
' --shrink-factors 8x4x2x1' +
' --smoothing-sigmas 3x2x1x0vox '
101
Section 1.06 Heart rate monitoring during sleep
Heart rate is a physiological correlate of sleep, as it regulated by autonomous nervous
system. Sleep stages are characterized by different specific heart rates, and heart rate rapidly
increases after arousal [Sleep stage classification by combination of actigraphic and heart rate
signals]. Role of heart rate variability during sleep is not clear, but it has been noticed that
changes in heart rate due to transition between sleep stages might contribute to risk of adverse
cardiac events [Sleep Processes Exert a Predominant Influence on the 24-h Profile of Heart Rate
Variability].
Zebrafish, with its small size and rapid development, is a useful model for drug screens.
Hear rate monitoring is possible on scale with zebrafish since it is a transparent animal, so optical
observation is sensitive enough to measure heart rate in larvae. Several reports have
demonstrated automatic imaging-based recording of heart rate in zebrafish, using fluorescent
microscopy [ZebraBeat: a flexible platform for the analysis of the cardiac rate in zebrafish
embryos], or transmitted-light microscopy [Automated high-throughput heart rate
measurement in medaka and zebrafish embryos under physiological conditions]. In the latter
report, authors demonstrated 3-day long recording of zebrafish heart rate, noticing that it follows
circadian clock.
In our experiments, we were able to perform recording of zebrafish heart rate during
sleep and wake. First, we developed a protocol for simple imaging-based method to measure
heart rate in groups of larval zebrafish. In this approach, the animals were anesthetized with
tricaine, and fully embedded in a Petri dish in 1% low-melting point agarose, after which the
water was replaced several times to wash out residual tricaine. Animals at 3-5dpf were imaged
under invisible IR light for more than 54hrs. We were able to image up to 16 fish at once. In order
to save processing resources, we imaged animals for 1 minutes every 2 minutes at 10fps, yielding
340GB-datasets. The image was analyzing by manually selecting region of interest over animals’
hearts and calculating image difference for each animal. The averaged image difference then was
plotted as spectrogram to highlight specific high-energy bands corresponding to heart rate of
each fish.
102
Figure 8.12. Heart rate monitoring in immobilized zebrafish. Up to 16 larval zebrafish at 3-4 dpf were
immobilized in agarose and imaged ofr 24 hrs in normal light/dark conditions (D11/L13). (Left)
Spectrogram of heart movememnt for one of th embryos shows decrease in heart rate during night
In one experiment, out of 16 animals imaged we were able to extract useable data for 13
fish (80% success rate). We observed that heart rate decreases at the beginning of recording,
perhaps because of anesthetic, but then recovered to normal level of 4-5Hz, reported previously
[]. During night heart rate on average was 10-15% lower than during day, with spontaneous
increases that perhaps signify night-time wake periods. The next morning heart rate went back
to normal wake-time level, as animal woke up.
There is a potential to easily add heart rate monitoring to our imaging setups, since we
are already using transmitted light to image tail movement. To test this, we used our horizontal
SPIM microscope and image heart rate for short periods of time, suing the same image-
difference-based analysis approach.
103
Figure 8.13. Heart rate detection under hSPIM imaging. Heart movememnt movie is converted to iamge
difference. And then change of image in each of ROIs is processed separately. (Bottom) Power spectra of
ROI image difference for atrium and ventricle.
Abstract (if available)
Abstract
Sleep is an essential behavior in vertebrates, and it has wide-spread effects on the brain. Currently, the connection between sleep and neuronal activity is studied either on a large scale, by collecting information by averaging the activity of thousands of neurons, or on a small scale with single-cell resolution of a few cells. Given the wide-range nature of sleep effects on the brain in vertebrate animals, we developed an integrative toolbox that provides whole-brain recording of neural activity with single-cell resolution. We also manipulated the activity of sleep-regulating circuits to better describe sleep using zebrafish as a model system. This approach is based on light-sheet microscopy, a rapidly developing method, which we applied to continuous 24hr-long imaging of live zebrafish. We assembled an analysis toolbox to quantify sleep and wake states in zebrafish. This toolbox can describe whole-brain activity changes in several different experimental conditions. First, we demonstrate the ability to perform whole-brain neural activity recording during natural sleep and wake states. Second, we analyze changes in neural activity in animals where the sleep-regulating circuit is stimulated. We provide an approach to describe whole-brain changes in connectivity and oscillations in sleep and wake states. Using our method, we found that we can image the zebrafish brain without perturbing sleep behavior, decreased correlation and activity between hindbrain regions during sleep, lack of single-frequency slow oscillations across the zebrafish brain during sleep, and that experimental activation of NPVF neurons induces a sleep-like state. Additionally, we expanded our platform, developing a protocol to investigate the difference between melatonin-induced sleep and natural sleep. This work provides a foundation for the detailed description of whole brain activity changes in sleeping zebrafish with single cell resolution, combining large and small scale studies to elucidate the molecular and cellular foundation of sleep in vertebrates.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Functional consequences of network architecture in rat hippocampus: a computational study
PDF
Negative valence neurons in the larval zebrafish pallium
PDF
Susceptibility-weighted MRI for the evaluation of brain oxygenation and brain iron in sickle cell disease
PDF
Computational investigation of cholinergic modulation of the hippocampus and directional encoding of touch in somatosensory cortex
PDF
On the electrophysiology of multielectrode recordings of the basal ganglia and thalamus to improve DBS therapy for children with secondary dystonia
PDF
Simulating electrical stimulation and recording in a multi-scale model of the hippocampus
PDF
Dynamical representation learning for multiscale brain activity
PDF
Quantitative analysis of mouse medial prefrontal cortex whole brain structural connectivity
PDF
Estimation of cognitive brain activity in sickle cell disease using functional near-infrared spectroscopy and dynamic systems modeling
PDF
The behavioral and neural bases of tactile object localization
PDF
Signal processing methods for interaction analysis of functional brain imaging data
PDF
Applications of graph theory to brain connectivity analysis
PDF
Detection and decoding of cognitive states from neural activity to enable a performance-improving brain-computer interface
PDF
Switching dynamical systems with Poisson and multiscale observations with applications to neural population activity
PDF
Skeletal cell fate plasticity in zebrafish bone development and regeneration
PDF
Learning invariant features in modulatory neural networks through conflict and ambiguity
PDF
Mapping water exchange rate across the blood-brain barrier
PDF
Neural basis of number sense in zebrafish
PDF
Neuronal master regulator SRRM4 in breast cancer cells facilitates CNS-acclimation and colonization leading to brain metastasis
PDF
Synaptic integration in dendrites: theories and applications
Asset Metadata
Creator
Andreev, Andrey
(author)
Core Title
Development of a toolbox for global functional brain imaging of wake and sleep states in zebrafish
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Biomedical Engineering
Publication Date
07/29/2019
Defense Date
08/01/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
brain imaging,microscopy,Neuroscience,OAI-PMH Harvest,Sleep,zebrafish
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Fraser, Scott E. (
committee chair
), Hires, Samuel A. (
committee member
), Mel, Bartlett W. (
committee member
)
Creator Email
aandreev@usc.edu,me@aandreev.net
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-198510
Unique identifier
UC11662781
Identifier
etd-AndreevAnd-7666.pdf (filename),usctheses-c89-198510 (legacy record id)
Legacy Identifier
etd-AndreevAnd-7666.pdf
Dmrecord
198510
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Andreev, Andrey
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
brain imaging
zebrafish