of 25
1
Norepinephrine changes behavioral state via astroglial purinergic signaling
Alex B. Chen
1,2,3
*, Marc Duque
2,3
, Vickie M. Wang
2,3
, Mahalakshmi Dhanasekar
4
, Xuelong Mi
5
, Altyn
Rymbek
6
, Loeva Tocquer
4
, Sujatha Narayan
1,7
, David
Prober
6
, Guoqiang Yu
8
, Claire Wyart
4
, Florian
Engert
2
, Misha B. Ahrens
1
*
1
Janelia Research Campus, Howard Hughes Medical Institute; Ashburn, VA 20147, USA
2
Department of Molecular and Cellular Biology, Harvard University; Cambridge, MA 02138, USA
3
Graduate Program in Neuroscience, Harvard Medical School; Boston, MA 02115, USA
4
Sorbonne Université, Paris Brain Institute (Institut du Cerveau, ICM), Institut National de la Santé
et de la Recherche Médicale U1127, Centre National de la Recherche Scientifique Unité Mixte de
Recherche 7225, Assistance Publique
Hôpitaux de Paris, Camp
us Hospitalier Pitié
-
Salpêtrière, Paris,
France
5
Bradley Department of Electrical and Computer Engineering; Virginia Polytechnic Institute and
State University; Arlington, VA 22203, USA
6
Tianqiao
and Chrissy Chen Institute for Neuroscience, Division of Biology and Biological
Engineering, California Institute of Technology, Pasadena, CA, 91125, USA
7
Present address: Allen Institute for Neural Dynamics; Seattle, WA 98109, USA
8
Department of Automation, Tsinghua University; Beijing 100084, P.R. China
*
Corresponding authors:
chena@janelia.hhmi.org
;
ahrensm@janelia.hhmi.org
Abstract
Both neurons and glia communicate via diffusible neuromodulatory substances, but the substrates of
computation in such neuromodulatory networks are unclear. During behavioral transitions in the larval
zebrafish, the neuromodulator norepinephrine drives fas
t excitation and delayed inhibition of behavior
and circuit activity. We find that the inhibitory arm of this feedforward motif is implemented by astroglial
purinergic signaling. Neuromodulator imaging, behavioral pharmacology, and perturbations of neurons
and astroglia reveal that norepinephrine triggers astroglial release of adenosine triphosphate, extracellular
conversion into adenosine, and behavioral suppression through activation of hindbrain neuronal adenosine
receptors.
This work, along with a companion piece by Lefton and colleagues demonstrating an analogous
pathway mediating the effect of norepinephrine on synaptic connectivity in mice,
identifies a
computational and behavioral role for an evolutionarily conserved astroglial purinergic signaling a
xis in
norepinephrine
-
mediated behavioral and brain state transitions.
.
CC-BY-NC-ND 4.0 International license
available under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (which
this version posted May 23, 2024.
;
https://doi.org/10.1101/2024.05.23.595576
doi:
bioRxiv preprint
2
eural circuits perform fast computations through precise
patterns of synaptic connectivity, as well as via direct
electrical coupling through gap
junctions
1,2
, but they can
also be rapidly modulated by diffusible chemical
messengers, including monoamines (e,g. norepinephrine,
dopamine, serotonin) and neuropeptides
3
5
. Such signaling
accounts for a large portion of neural activity patterns
unexplainable by synaptic connectivity alone
6
8
and has
long been known to reconfigure synaptic networks to
orchestrate behavioral states
9
13
. For much of the past
century since the discovery of neuromodulators, their
profound effects on neural circuits have been th
ought to
proceed through activation of cognate receptors on neurons.
However, recent discoveries that astroglia communicate bi
-
directionally with neurons via neuromodulatory signaling
necessitate a retrospection of this dogma and a
consideration that these
non
-
neuronal cells could play more
important roles as neuromodulatory actuators than
previously thought
14
. Astroglial physiology differs
significantly from neuronal physiology. They are
electrically inexcitable, exhibit local and global intracellular
calc
ium transients, and possess complex arbors of processes
that form non
-
overlapping territories and interact with
thousands of individual neuronal synapses
15,16
. However,
the precise role of astrocytes as active modulatory elements
in neural circuits is still unresolved.
Our work here demonstrates that an astroglial bridge
connects two ubiquitous, yet mysterious neuromodulatory
systems
norepinephrine (NE) and ATP/adenosine
to
mediate rapid behavioral state changes. Since its discovery
in the 1940s
17
, NE has been known to strongly affect neural
circuits and behavior through brain wide projections
originating in the locus coeruleus and other clusters
18
. NE
neurons promote rapid arousal
19
, and more generally, they
seem to be engaged under
circumstances that necessitate
transitions of behavioral state, captured in several broad
theories about NE’s function, notably those of global model
failure
20
22
, ‘resetting’ neural circuits
23
, and facilitating
exploration/exploitation
24
. Consistent with its behavioral
effects, NE rapidly reconfigures circuit dynamics by
altering synaptic strength, gating synaptic inputs, and tuning
circuit gain and synchrony
25
28
. While the dominant
assumption over the past eight decades has been that the
profound effects
NE exerts over behavior and
neurophysiology occur through activation of adrenergic
receptors on neurons, recent discoveries that NE strongly
activates non
-
neuronal cells, in particular astroglia,
challenge this assumption. NE is among the strongest
driver
s of astroglial signaling and triggers large intracellular
calcium events caused by
α
1
-
adrenergic receptor (
α
1
-
AR)
activation
29,30
. The discovery that NE leads to astrocyte
calcium activity has led to much work on the function of
astroglia in NE
-
mediated b
ehaviors and modulation of
circuit dynamics
31,32
. However, the specific roles played by
astroglia in noradrenergic modulation remain unclear, as do
the molecular pathways linking NE
-
mediated astroglial
calcium elevation to modulation of neural circuit activity.
As with NE, the purinergic signaling molecules ATP and
adenosine are ubiquitous and critical neuromodulators of
the central nervous system. They play important roles in
sleep
-
wake cycles
33,34
, synaptic plasticity
35
, and motor
pattern generation
36
, among a number of other functions
37
.
Dysfunction in purinergic signaling has been implicated in
panic disorder, depression, and epilepsy
38,39
. As ATP and
adenosine are present as metabolic molecules in all cells,
the source of extracellular ATP/adenosine as a
neur
omodulator
is
contentious.
Much
like
NE,
ATP/adenosine signaling, since its discovery as a
neuromodulatory system in the 1950s, has largely been
conceptualized as primarily neuronal. ATP can be co
-
transmitted with other neurotransmitters through vesicular
release
40
at axons, and can undergo somato
-
dendritic
secretion through unclear mechanisms
34
. While astroglia
have been argued to be a source of extracellular adenosine
through ATP secretion
35,37
and extracellular ATP
-
to
-
adenosine metabolism
41
, the behavior
al relevance of such
release remains, in many cases, controversial
42,43
.
Furthermore, while astroglial calcium elevation appears to
trigger ATP secretion, the behavioral contexts that recruit
astroglial purinergic signaling remain poorly understood,
due to the reliance on exogenous chemogenetic activation
and/or
ex vivo
conditions in existing studies
44
46
. Few
studies have imaged purinergic signaling and astroglial
calcium during behavior. A seminal paper
31
showed that, in
larval
Drosophila
, adenosine recep
tor activation is
necessary for the inhibitory effects of the insect
norepinephrine analog octopamine, but ATP/adenosine
dynamics were not explored. Whether astroglial purinergic
signaling is a major downstream pathway of NE
-
mediated,
astroglial calcium si
gnaling in vertebrates, and its precise
role in NE modulation of behavior and neurophysiology
remain important open questions.
Leveraging the larval zebrafish, in which NE, ATP, and
astroglial calcium can be imaged in conjunction with neural
activity during behavior,
we find that, during rapid
behavioral state transitions, the noradrenergic and
purinergic systems can be conceptualized as,
respectively, fast excitatory and delayed inhibitory arms
of a feedforward motif with astroglia as a coordinating
intermediary.
Therefore, beyond slow modulation of state,
NE also acts through astroglial purinergic signaling to
rapidly reconfigure
circuit dynamics and enact behavioral
state transitions.
NE neurons drive a biphasic futility response
Larval
zebrafish possess an innate tendency to stabilize
their position by swimming in the direction of coherent
visual flow
47,48
. We previously showed that when swims no
longer move the fish forward, futility drives firing in
N
.
CC-BY-NC-ND 4.0 International license
available under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (which
this version posted May 23, 2024.
;
https://doi.org/10.1101/2024.05.23.595576
doi:
bioRxiv preprint
3
hindbrain NE
neurons,
and NE signals
through radial
astroglia to suppress
futile swims
49
. Radial astroglia are a
glial cell type, found in many
vertebrates, with similar
molecular and
functional
characteristics
to
mammalian
astrocytes
50,51
.
Thus,
futility
-
induced passivity in the larval
zebrafish is a rapid, NE
-
mediated
behavioral state transition
in which the
astroglial modulation of neural
circuits
can be
dissected
from the molecular
to behavioral levels
by
leveraging the
larval zebrafish’s unique optical
accessibility.
Here we made use of our previously
published behavioral assay for futility
-
induced
passivity
in
larval
zebrafish
49,52
. Specifically, fish were
immobilized in agarose and their tails
freed. The animals’ tail positions were
then
automatically
tracked,
and
detected swims used to deliver realistic
online
visual
feedback
through
projection of drifting grating stimuli to
the f
loor of the chamber (
Fig. 1A,
Methods
). To encourage robust
swimming behavior, we delivered a
steady,
constant
-
velocity
forward
drifting
gr
ating
47
(
Fig.
1B
).
Simultaneously, we manipulated the
efficacy of the fish’s swims by cycling
between two stimulus conditions:
closed loop, and open loop. During the
closed loop condition, the fish’s swim
Figure 1. Futility triggers a biphasic behavioral and neural response through NE neuron activation.
(
A
) Schematic of
virtual reality behavioral experiments with real
-
time swim detection and visual feedback.
(
B
) Diagram illustrating the difference between closed loop (visual feedback in response to swims) and open loop (no visual
feedback) conditions.
(
C
) Schematic: the known cell types involved in futility
-
induced passivity.
(
D
) Swim trace of an example trial demonstrating closed and open loop swim behavior.
(
E
) Average closed loop and passivity
-
triggered open loop tail angle demonstrating an initial increase in swim amplitude
(excitatory phase) followed by inhibition of swimming (inhibitory phase) in open loop.
(
F
) Neural activity was imaged with a confocal microscope while NE neurons were optogenetically activated using a fiber optic.
(
G
) Optogenetic stimulation
-
triggered average of neural activity in motor areas demonstrating fast excitation and delayed
inhibition, similar to the behavioral futility response.
(
H,I
) Effect of blocking
α
1
-
adrenergic receptors (100
μ
M prazosin) or
β
-
adrenergic receptors (100
μ
M propranolol) on (
H
) open
loop passivity and (
I
) open
-
loop swim vigor.
(
J
) Model of
parallel noradrenergic channels that contribute to the excitatory and inhibitory phases of the futility response and
central problem statement.
.
CC-BY-NC-ND 4.0 International license
available under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (which
this version posted May 23, 2024.
;
https://doi.org/10.1101/2024.05.23.595576
doi:
bioRxiv preprint
4
attempts resulted in visual feedback in the form of backward
drift of the visual stimuli to generate the perception of
successful forward swimming (
Fig. 1B
,
left
). In contrast,
during open loop, swim attempts resulted in no change to
the visual stimulus (
Fig. 1B, right
) and can therefore be
classified as futile. It is known that futility is encoded by a
population of NE neurons in the medulla oblongata known
as NE
-
MO (putatively homologous to mammalian cluster
A2
53
) and that NE
-
MO activation causes passivit
y through
astroglial calcium signaling and activation of GABAergic
neurons in the lateral medulla oblongata (L
-
MO) (
Fig.
1C
)
49
.
Consistent with previous work
49,54
, we found that
behavioral futility signaled by a lack of visual feedback
caused fish, after 10
-
20 seconds of multiple futile
swim
s, to
enter a passive state, in which they do not perform any more
swim attempts (
Fig. 1C,D,
Suppl. Fig. S
1A
-
C
). This
passive state then lasts for tens of seconds. Prior to
passivity, fish exhibited a marked increase in swim vigor as
well as an increase in the probability to perform high
-
amplitude, struggle
-
like swims (
Fig. 1C,D,
Suppl. Fig.
S
1D,
E
). While NE
-
MO activation was previously shown to
cause passivity
49
, the contribution of NE neurons in the
transient upregulation of swim vigor at futility onset has
been less explored. Given NE’s well
-
documented ability to
enhance arousal and effort
19,23
, we tested whether NE
neuron firing immediately promotes the rapid enhancement
of vigor, in addition to driving temporally delayed swim
inhibition (
Fig. 1E
,
Suppl. Fig. S
2
). Indeed, we found that
optogenetic stimulation of NE neurons drove fast excitation
and persistent, but delayed inhibition of hindbrain motor
circuits (
Fig. 1F
,G
,
Suppl. Fig. S
2A
-
C)
.
We further
observed that persistent inhibition of motor circuits
coincides with
sustained activity in L
-
MO, a GABAergic
region previously shown to suppress swimming during
futility
-
induced passivity
49
(
Suppl. Fig. S
2D
). Therefore, in
larval zebrafish, behavioral futility drives a biphasic
response in both behavior and neural
dynamics, which can
be defined as an excitatory phase,
consisting of increased
behavioral vigor, followed by an
inhibitory phase involving
behavioral suppression (
Fig.
1D,F
). Both phases are caused
by NE neuron firing.
NE neurons act on downstream targets via activation of
α
-
and
β
-
ARs by NE, as well as via fast synaptic excitation
through co
-
released glutamate. Astroglial calcium elevation
through
α
1
-
AR activation has been shown to be both
necessary and sufficient for the inhibitory phase of the
futility response in larval zebrafish
49
. However, the
relationship between astroglial calcium signaling and the
excitatory phase of the futility response is less understood.
Two pieces of evidence indicate that astroglial calcium
signaling is not involved in the excitatory phase. First,
optogenetic stimulation of NE neurons elevated astroglial
calcium with a temporal delay too long to account for the
more rapid activation of motor activity (
Suppl. Fig. S
3
).
Second, inhibition of
α
1
-
AR signaling with prazosin,
previously shown to completely abolish NE
-
evoked
astroglial calcium responses
49
, had no effect on futility
-
induced vigor enhancement, but did suppress futility
-
induced passivity (
Fig. 1
H,I
). Thus,
α
1
-
AR signaling, and
therefor
e astroglial calcium elevation, is dispensable for the
excitatory phase. On the other hand, inhibition of
β
-
ARs had
little effect on the inhibitory phase but strongly attenuated
the excitatory phase (
Fig. 1
H,I
), suggesting that different
adrenergic receptor subtypes may contribute to different
aspects of the futility response. Specific astroglial
involvement in the inhibitory phase raises a fundamental
question, central to this work (
Fig. 1
J
): as neurons
ultimately control motor output, how do astroglia sig
nal to
downstream neurons to drive the inhibitory phase by
suppressing swimming?
Futility
-
induced, NE
-
dependent astroglial ATP release
We reasoned that astroglia likely communicate with
downstream neurons by secreting a neuroactive substance.
In particular, we hypothesized that astroglia release
adenosine triphosphate (ATP) in response to calcium
elevation during futility, as glial
-
derive
d ATP can modulate
neural activity in other contexts
45,46,55,56
. To investigate
whether futility
-
induced astroglial calcium elevation leads
to release of ATP, we generated a fish line
(
Tg(gfap:GRAB
ATP
;gfap:jRGECO1a)
) that expresses, in
astroglia, a recentl
y developed extracellular green
fluorescent ATP sensor
57
, as well as an intracellular red
fluorescent calcium sensor (Methods). We then performed
simultaneous brain
-
wide functional imaging of both glial
calcium and secreted ATP while immobilized animals
behaved in virtual reality (Methods) (
Fig. 2A,B
). We found
that, during open
-
loop swimming, both astroglial
intracellular calcium and extracellular ATP around astroglia
exhibited a rapid elevation throughout the hindbrain,
followed by a slower return to base
line over tens of seconds
(
Fig. 2C,D
).
These data suggest that astroglia, and not neurons, release
ATP during futile swimming. However, since the ATP
sensor used is extracellular, it cannot distinguish between
astroglial
-
secreted ATP and ATP released by neurons near
astroglial processes. Howeve
r, five additional lines of
evidence support an astroglial origin for the released ATP.
First, ATP elevation lags behind intracellular astroglial
calcium elevation (
Suppl. Fig. S
4A
-
C
), consistent with
astroglial
calcium elevation causing ATP release. Second,
simultaneous imaging of neuronal calcium activity and
extracellular ATP revealed that neuronal calcium elevation
failed to reliably predict ATP release, whereas glial calcium
events were always accompanied by
ATP elevation (
Suppl.
Fig. S
4D
). Third, inhibition of astroglial calcium with
pharmacological blockade of α1
-
ARs strongly attenuated
futility
-
triggered ATP elevation (
Fig. 2E
). Because this
.
CC-BY-NC-ND 4.0 International license
available under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (which
this version posted May 23, 2024.
;
https://doi.org/10.1101/2024.05.23.595576
doi:
bioRxiv preprint
5
pharmacological manipulation is non
-
specific, we also
generated
a fish
expressing
the
calcium
extruder
hPMCA2
specifically
in
astroglia
(
Tg(gfap:hPMCA2
-
mCherry)
).
Astroglial
-
specific
hPMCA2
expression
inhibited
astroglial
calcium
elevation
during futile
swims
(
Suppl.
Fig. S
4E
)
and accordingly
decreased the
duration
of
passivity
in open
loop (
Suppl. Fig.
S
4F
-
I
).
Consistent
with α1
-
AR
blockade,
inhibiting
glial
calcium
elevation with
hPMCA2 also
suppressed ATP
futile
swim
-
evoked
ATP elevation
(
Fig.
2E
).
Fourth,
pharmacological
activation of α1
-
ARs
was
sufficient to
cause ATP
elevation even when
neural
activity
was
inhibited with
a
sodium channel
blocker
(
Fig. 2F,G
).
Figure 2. Futility drives astroglial release of ATP.
(
A
) Experimental schematic: two
-
color light
-
sheet imaging of extracellular ATP and astroglial calcium (
Tg(gfap:GRAB
ATP
;
gfap:jRGECO1a
) fish) along with fictive behavioral recording.
(
B
-
C
) Fluorescence micrographs of simultaneously collected GRAB
ATP
and jRGECO1a signals in a fish in (
B
) baseline
condition or (
C
) during a futile swim.
(
D
) Futile
-
swim triggered astroglial calcium and extracellular ATP signals averaged across fish.
(
E
) Futile swim
-
triggered GRAB
ATP
signal in fish treated with an
α
1
-
AR blocker (100
μ
M prazosin) or vehicle, and in fish
expressing hPMCA2 in astroglia.
(
F
) Experimental schematic:
ex vivo
confocal imaging during puffing of an
α
1
-
AR agonist (10
μ
M methoxamine) or vehicle in
the presence of a neural activity blocker (170 mg/L MS
-
222, a sodium channel inhibitor).
Tg(gfap:GRAB
ATP
)
fish.
(
G
) GRAB
ATP
signal in fish in experiments described in (F), triggered on puff and onset
-
aligned.
(
H
) Experimental schematic:
in vivo
widefield imaging during chemogenetic activation of
Tg(gfap:rTRPV1
-
eGFP)
fish with
200 nM capsaicin in the presence of a neural activity blocker (170 mg/L MS
-
222).
(
I
) GRAB
ATP
signal in fish treated with capsaicin as described in (F), triggered capsaicin administration and onset
-
aligned.
.
CC-BY-NC-ND 4.0 International license
available under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (which
this version posted May 23, 2024.
;
https://doi.org/10.1101/2024.05.23.595576
doi:
bioRxiv preprint
6
Fifth, direct chemogenetic activation of astroglia was
also sufficient to cause ATP elevation when neural activity
was
inhibited
(
Fig.
2H,I
).
Importantly,
neither
pharmacological nor chemogenetic astroglial activation
affected neuronal activity under conditions of sodium
channel block (
Suppl. Fig. S
4J,K
). These converging lines
of evidence implicate norepinephrine
-
mediated astroglial,
and not neuronal, calcium signaling as critical for
extracellular ATP elevation during behavioral futility.
ATP
promotes passivity via extracellular metabolism into
adenosine
W
e investigated whether ATP elevation promotes
passivity by treating fish with NPE
-
caged ATP (P(3)
-
[1
-
(2
-
nitrophenyl)]ethyl
ester
of
ATP),
which
is
pharmacologically inert until exposed to ultraviolet (UV)
light (Methods) (
Fig. 3A
). Freely swimming fish treated
with caged ATP or vehicle were exposed to UV light, and
the time required for fish to switch from active swimming
to passivity was recorded. Ultraviolet light constitutes an
inescapable aversive stimulus, conceptually similar to the
open loop conditio
ns described in Figure 1. As a result, all
fish exposed to UV light eventually exhibited futility
-
induced passivity following a period of high
-
vigor
swimming (
Fig. 3B,
Suppl. Fig. S
5A
). However, fish
treated with caged ATP became passive significantly more
quickly than vehicle controls (
Fig. 3B
-
C
), while exhibiting
no significant difference in struggle onset or time to peak
swimming (
Suppl. Fig. S
5B
). Therefore, ATP elevation
drives the inhibitory, but not the excitatory, phase of
futility
-
induced passivity
.
Having established that ATP is a central astroglia
-
to
-
neuron signal that can induce passivity, we next sought to
determine the mechanism through which astroglial ATP
release suppresses swimming. ATP directly binds to two
families of purinergic receptors, i
onotropic P2X receptors
.
CC-BY-NC-ND 4.0 International license
available under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (which
this version posted May 23, 2024.
;
https://doi.org/10.1101/2024.05.23.595576
doi:
bioRxiv preprint