of 26
Peer
Review
File
Manuscript
Title:
TRANSFORMATIONS
AND
FUNCTIONS
OF
NEURAL
REPRESENTATIONS
IN
A
SOCIAL
BEHAVIOR
NETWORK
Editorial
Notes:
Redactions
unpublished
data
Parts
of
this
Peer
Review
File
have
been
redacted
as
indicated
to
maintain
the
confidentiality
of unpublished
data.
Reviewer
Comments
&
Author
Rebuttals
Reviewer
Reports
on
the
Initial
Version:
Referees'
comments:
Referee
#1
(Remarks
to
the
Author):
In
“BNST
promotes
regional
male
bias
within
a
female
-
biased
circuit
controlling
social
behavior
in
male mice” Yang and Anderson start with knowledge from a previous Shah lab study of bulk activity
of the
AB+
BNSTpr
neurons
which
concluded
that
they
encode
representations
of gender in
the
male
brain in an intensity
-dependent manner. The present study seeks to dissect the role of BNST-
ESR
neurons (a subset of AB+ neurons) that project to either the MPOA or the VMHvl using single cell
imaging. The approach is well used and the depth of the analysis performed is complete and
convincing, leading to technically rigorous
experiments. However, the difference between what was
previously shown and what is now understood is subtle and less convincing. Moreover, neither the
previous or the current study convince me that these neurons are actually encoding sex
representations, largely because in the previous study they were not found to encode gender in the
female brain (and it was not studied here). Is it likely that the female brain has evolved an entirely
different mechanism to encode such a key percept? The analysis in this study nicely supports that
they are responding to an analog comparison but it could be another function, such as motivation,
that drives their activity. Though it may be intuitive to think that they are encoding sex and is
supported by the PC and in silico analysis, it is not empirically shown by experimental means. It is
correlative.
The
author’s statement
that it
is “paradoxical that silencing
of BNSTpreEsr1
neurons
had
no effect on our ability to decode intruder sex from population activity in either the MPOA or
VMHvl” indicates that the data, model, and hypothesis are not yet fully aligned. The other major
finding from this study is the flexibility of the neural code downstream of the BNST. This part of the
conclusion echos back to earlier work from the Anderson lab finding scalable control of mounting
and attack in the VMH, and work from the Lin lab has shown heterogeneity in the VMH
-Esr
po
pulation
that
correlates
with
promoting
different
social
behaviors
(Hashikawa,
2017).
This
extends
some of that previous work by focusing on the function of the BNST to gate differential activity, but
there is no mechanistic understanding of how these neur
ons are dynamically recruited or how their
addition alters natural social behavior. Though I understand the meaning of the title, I do not agree
that the statistical difference of sampling of several 100 neurons in one area of the brain should be
interpret
ed
as functional
male
or female
bias.
Though
the
experiments
in this
study are
elegant,
they
do not serve to clarify the significance of the BNST in the social behavior circuit and the
representation of sex in the brain.
Additional
concerns:
1) From fig 3 onwards, where the chemoinhibition
of BNST Esr1 neurons is being used to draw
conclusions about the information content in the VMHvl Esr1 and the MPOA Esr1 neurons, the
stated hypothesis is "that inhibiting BNSTprEsr1 neurons should alter sex representations in
MPOAand VMHvl."
To establish
this, it would be clearer to inhibit BNST Esr1 neurons, and then
image
neurons
in the
VMHvl
or MPOA
that
express
Esr1
AND
are
downstream
of the
BNST
Esr1
(or
BNST) neurons. In the absence of this kind of experiment, it is difficult to formally state that the
BNST
- VMHvl
or BNST
-MPOA
projection
has
anything
to do
with
the
distortions
in representations
at the
VMHvl
or
MPOA.
2)
The
above
concern
is
also
applied
to
behavior.
When
the
BNST
-
Esr
to
-
VMH
-
Esr
or
to
MPOA
-
Esr
is
specifically
silenced,
what
is
the
effect
on
behavior?
3)
The
rationale
to
focus
on
the
Esr
population
is
not
clear.
Are
the
ESR
negative
subset
of
AB
neurons
not
responding
to male
and
female
cues
or projecting
to the
MPOA
and
VMH?
If not,
what
is their role in the model?
4)
Many
of
the
BNST
-
Esr+
neurons
are
not
active
in
the
presence
of
either
a
male
or
female
(Figure
2). Are these neurons also projecting to the MPOA or VMH? The use of mating and aggression as a
behavioral proxy for sex identification is not very granular. Maybe the BNST contributes to other
aspects of social motivation that result in the phenotypes in figure 1. Do the authors have another
measure
for
sex
recognition
that
could
support
the
functional
conclusions
of the
Esr
neurons
driving
sex recognition?
5)
Are
all
of
the
BNST
-
Esr
neurons
that
project
to
the
two
studied
targets
inhibitory?
If
not,
this
could
confound
the
model.
6)
Is
it
possible
to
manipulate
the
sensory
signals
so
that
BNST
and
VMH
activity
can
be
monitored
as
they
mis
-identify
the
sex
of the
partner?
A gender
illusion?
This
would
control
for
all
other
aspects
of
behavior and neural activity.
7)
The
model
(EDF8)
suggests
that
absolute
incoming
sensory
activity
is
weighed,
more
female
activity
= mating,
more
male
activity
= aggression.
Can
you
manipulate
this
by
allowing
your
subjects
to interact with multiple females and a single male simultaneously? It looks like the BNST activity
persists as long as a female is present (fig 2j). Is this correct? If the subject first interacts with
females
and
then
a male
is also
added
subsequently,
does
the
simultaneous
presence
of a male
alter
the balance of activity in the VMH? (In both cases, I expect the presence of a male would promote
aggression, but the representation of male and female shoul
d not change.)
8)
Figure
4a,
it
appears
that
with
CNO
(BNST
-
)
the
female
responding
cells
are
largely
spatially
segregated
from
the
male
responding
cells
and
about
half
of the
male
responding
cells
remain
male
responding. Are these two populations, those that switch vs retain sex tuning without BNST input
different molecular subsets of the VMHvl-
ESR population? On repeated trials, is the same neuron
able to switch sometimes and remain stable other times, or
are they set to be either flexible or
fixed?
Minor
Concerns:
1)
I cannot
find
CNO
only
controls
on
neural
activity
and
behavior,
and
some
quantification/analysis
for the silencing of BNST
-Esr expressing hm4di neurons.
2)
The
term
chemoscope
doesn't
seem
to add
much,
it is fundamentally
chemogenetics
and
regular
miniscope imaging combined.
3)
Fig
2:
Representative
images
of GCaMP
infections
in BNSTEsr1
neurons
will
be
useful.
4)
Fig
3:
Representative
images
of DREADDs
infections
in BNSTEsr1
neurons
+ GCaMP
in
MPOA/VMHvl Esr1 neurons will be useful.
5)
Ext
Fig
4
:
Why
are
error
bars
missing
in
k?
Referee
#2
(Remarks
to
the
Author):
Neural
circuits
that
mediate
social
behaviors
like
mating
and
aggression
are
not
well understood.
This
paper is a technical tour
-de
-force, using in vivo calcium imaging in several limbic system nuclei of
awake
mice
to provide
single
neuron
representations
during
male
and
female
social
encounters.
This
study significantly advances our understanding of sensory coding in
the BNST (a limbic system
structure that receives chemosensory inputs), revealing that different neurons are tuned to male
and female sensory cues,
with smaller groups
of neurons activated during behavioral displays rather
than by sensory cues per se. Pri
or studies involving fiber photometry in this region were important,
claiming a female
-bias in the overall response, but as they lacked single neuron resolution, they
missed the presence of male-
selective neurons, which are sparser, as well as the distribu
tion of
neurons tuned to sensory vs motor actions. Moreover, Yang and Anderson used chemogenetics to
show that BNST inputs re
-shape sex representations in a downstream hypothalamic nucleus (the
VMH). I am enthusiastic about this study, but also note that s
ome additional controls are needed to
validate
claims and
that
some questions remain
related
to the
mechanism underlying the interesting
BNST
-VMH transformation.
1.
It
is
interesting
that
the
VMH
and
MPOA
still
display
sex
-
biased
responses
after
chemogenetic
inhibition
of the
BNST.
This
could
be
due
to factors
discussed,
such
as a role
for
another
brain
region,
but also
could
be
due to
incomplete
BNST inhibition
in chemogenetic
experiments, related to
either
(1)
the
extent
of coverage
of the
large
BNST
area
by
AAV
injections,
and
2) the
Cre
lines
used.
For
(1),
what
% of Cre
-expressing
BNST
cells
are
labeled
and
silenced
by
AAV
injections,
and
for
(2)
are
there
sex
-selective BNST neurons that do not express aromatase and/or ESR and could
be
contributing to
downstream representations?
2.
The
authors
should
ensure
that
AAV
injection
in
the
BNST
does
not
label
neurons
in
the
VMH
or
MPOA
(for
example
by
retroactive
labeling
of neurons
providing
feedback
control),
which
could
confound
interpretations.
3.
The
authors
use
a
nicely
comprehensive
set
of
stimuli
in
Figure
2j,
and
showed
that
BNST
response variance due to intruder sex was larger than the associated behavior. It would be worth
discussing
these
observations
in the
context
of the
lab's
previous
work
on
VMH/MPOA.
Is there
an
increased response variance due to behavior in the VMH and MPOA, suggesting further input
transformation as information moves to the hypothalamus?
4.
The
authors
focus
here
on
neuronal
r
epresentations
in
male
mice;
do
similar
sex
biases
in
BNST
and
VMH
exist
in
female
mice?
5.
Both
the
title
and
last
sentence
of
the
abstract
should
be
edited
for
clarity
to
make
the
manuscript
more
accessible
to a general
audience.
(I would
go
with
something
like
'Transformations
of sex
representations in the ascending limbic system' but of course this is just a suggestion!)
6.
For
ED
4b,
it
seems
like
activity
is
synchronized
to
particular
events
in
the
time
series
-
if
true,
it
would
be
helpful
to provide
annotation
of whether
such
synchronized
events
correspond
to
sniffing
or other social episodes.
Referee
#3
(Remarks
to
the
Author):
Yang
&
Anderson
demonstrate
that
chemogenetic
silencing
of
BNSTprEsr1
alters
sexual
behavior
and
aggression. Optogenetic silencing of this same neuronal population showed that activity of
BNSTprEsr1
neurons
is required
for
the
transition
from
appetitive
to consummatory
social
behaviors
towards both sexes (i.e. initiation and duration of aggression and
sexual behavior).
These findings (reported already in previous studies, as noted by the authors) set the foundation to
determine
how
sex
is presented
in this
neuronal
population.
The
authors
performed
calcium
imaging
of the BNSTprEsr1 neuronal population in sexually experienced male mice during social interaction
with an intruder male or female, using a miniature head
-mounted microscope, and specifically
monitored the decoding of the intruder's sex, at a single cell level. They identified subpopulations
that were female preferring or male preferring, with or without physical interaction with the
intruders, suggesting that sex is represented by population coding.
Next,
they
combined
chemogenetic
silencing
of BNSTprEsr1
neurons,
with
microendoscopic
imaging
of VMHvlEsr1 or MPOAEsr1 neurons expressing. They revealed that such manipulation led to a
decrease in the neuronal response to male intruders in both MPOA and VMHvl neurons, and an
increase in the response to female intruders in VMHvl. Moreover, they revealed that silencing
BNSTprEsr1 neurons inverted the 2:1 ratio of male
- to female
-preferring units in VMHvl, to the
female
-dominant ratio seen in MPOA.
The
main
novel
findings
and
conclusion
of the
authors
is that
"the
activity
of BNSTprEsr1
neurons
is
not
required for the coding of intruder sex identity by MPOAEsr1
and
VMHvlEsr1
neurons.
Rather,
it is required
to invert,
in VMHvl,
the
female
bias
in population
representations of intruder sex seen in BNSTpr, MPOA and MeApd, to a male bias".
The
set
of
methodologies
used
is
impressive,
contain
appropriate
controls
and
the
data
is
analyzed
well.
No
doubt
it
will
provide
a
great
dataset
for
better
understanding
of
how
sex
-
specific
stimuli
are
encoded
in neuronal
networks,
at single
cell
resolutions.
However, the neuroimaging-
chemogenetic data set (the main part which provide novel findings),
although interesting, is purely descriptive and missing any mechanistic explanation (beyond the
proposed hypothesis). Namely, how does sex
-biased neu
ronal coding in the MPOA/VMH/BNST (or
the altered sex-
bias in the VMH following chemogenetic silencing) encode sex
-typical stimuli and
control different reproductive behaviors (mating and aggression) towards males and females? As
stated by the authors them
selves in the discussion, the manuscript does not provide any
experimental data to answer whether and how the changes in the ratio of female-
male responsive
neurons induced by silencing of BNSTprEsr1 are required for sex discrimination, sex
-typical
sexual/
aggressive
behavior,
or any
other
phenotype.
Moreover,
it is essential
to confirm
the
findings
with additional complementary manipulations such as chemogenetic or optogenetic activation of
the same neuronal population.
Additional
concerns:
1. The use
of restrained
individuals as social
stimui (first set of
experiment) is
very problematic, as
it
may
trigger
a massive
stress
response
in both
mice.
If the
authors
wish
to examine
the
response
to
a
conspecific mouse without enabling attack or mount responses,
they can present the conspecific
separated by a perforated barrier, or use a stimuli such as urine or soiled bedding.
2.
The
link
between
the
behavioral
and
neural
effects
of
the
chemogenetic
silencing
is
poorly
explained,
thus
it
is
not
clear
what
is
the
biological/functional
significance
of
the
neural
findings.
3.
In
continuous
to
the
prior
comment,
did
the
authors
notice
any
sex
-
reversed
behaviors
during
the
silencing
period
(i.e.
attack
of subject
females
or sexual
behaviors
towards
males)?
These
behavior
should be quantified for pre
-
CNO and CNO segments.
4.
Also,
is
the
silencing
effect
reversible?
What
would
happen
in
optogenetic
/
chemogenetic
activation
of
these
neurons?
5.
In
order
to
support
their
sex
discrimination
claim,
the
authors
need
to
conduct
a
separate
discrimination
assay,
during
BNSTprEsr1
silencing.
6.
Do
the
response
characteristics
of
specific
neurons
remain
stable
for
days/weeks?
Author
Rebuttals
to
Initial
Comments:
Nature
2021
-
03
-
04882
Yang
et
al.
Point
-
by
-
point
response
Reviewer
#1
We thank this reviewer for their incisive and helpful questions and comments. While it was
not
possible
to perform
every
experiment
requested,
both
for
technical
reasons
and
because
of the contraction of our mouse
colonies necessitated by the COVID pandemic, we have
done
our
best
to
provide
new
data
to
address
them.
Comment
1.
“neither
the
previous
or
the
current
study
convince
me
that
these
neurons
are
actually
encoding sex
representations,
largely
because
in the
previous
study
they
were
not
found to encode gender in the female brain (and it was not studied here).”
Response
: We appreciate the reviewer’s comment, but find it puzzling. We do not
understand why evaluating the function of a neuronal population i
n males is
dependent on knowing what a similar population does in females (or vice-
versa),
especially when the circuit we are studying is well
-known for sexual dimorphisms in
structure and function. Perhaps the reviewer thinks that by
“sex representations”
we
mean a representation of the animal’s OWN sex?
If so, that is a misunderstanding:
we meant the representation of the sex of a conspecific intruder, either male or
female (see Remedios et al. (2017)
Nature
550
:388
-392). In any case, as detailed
below our new data indicate that while BNSTpr may contain a neural representation
of intruder
sex
(or
a sex
-specific
internal
state),
it is not
required
functionally
in
males
to identify
and
distinguish
male
from
female
conspecifics.
This
in turn
argues
that
the
requirement for BNSTpr in mounting and attack behavior is not an indirect
consequence of deficient sex identification. These new findings have substantially
changed our view of BNSTpr function, as explained in our responses below and in
the
revised
man
uscript.
Comment 2.
“The analysis in this study nicely supports that they are responding to an
analog comparison, but it could be another function, such as motivation, that drives their
activity. Though
it may
be intuitive
to think
that
they
are
encodin
g sex
and
is supported
by
the PC and in silico analysis, it is not empirically shown by experimental means. It is
correlative.”
Response
: The reviewer is correct that we cannot distinguish a function in encoding
sex from encoding a
motivational state that is closely associated with the intruder’s
sex.
Our
experiments
using
functional
silencing
demonstrate
that
BNSTpr
Esr1
neurons
are necessary for the transition from sniffing to consummatory behavior (mating or
fighting).
We
have
now
revised
our
manuscript
to indicate
that
BNSTpr
could
encode
either
intruder
sex,
or a motivational
state
that
is strongly
correlated
with
intruder
sex,
on
pg. 10
and
12.
Comment 3.
“From fig 3 onwards, where the chemo
-
inhibition of BNST Esr1 neurons is
being used to draw conclusions about the information content in the VMHvl Esr1 and the
MPOA Esr1 neurons, the stated hypothesis is "that
inhibiting BNSTprEsr1 neurons should
alter sex
representations in MPOA and VMHvl
." To establish this, it would be clearer to
inhibit
BNST Esr1
neurons,
and
then
image
neurons
in the
VMHvl
or MPOA
that
express
Esr1
AND are downstream of
the BNST
Esr1 (or BNST) neurons
. In
the
absence of
this
kind of experiment, it is difficult to formally state that the BNST
-VMHvl or BNST
-MPOA
projection
has
anything
to
do
with
the
distortions
in
representations
at
the
VMHvl
or
MPOA”.
Response
: The experiment the reviewer suggests (
bolded
above) is not technically
feasible
at present.
To
our
knowledge,
there
is no method
that
will
efficiently
transfer
GCaMP anterogradely from a pre
-synaptic neuron to its post
-synaptic target while
maintaining
the
viability
of the
latter
cells
for
imaging.
However, we
now
present
new
data
in
ED
Fig.
2
showing
that
silencing
of BNSTpr
Esr1
terminals
in VMHvl
and
MPOA
blocks attack and mating, respectively (but not vice-
versa). We also present new
data showing that the female-
biased representations of intruder sex are present in
BNSTpr
neurons
that
project
to MPOA
and
VMHvl
(
Extended
Data
Fig.
9
).
Together
these data make it highly likely that the BNSTpr
VMHvl and BNSTpr
MPOA
projections control sex representations in these two hypothalamic nuclei.
Nevertheless,
we
have
included
a caveat
to state
that
we
cannot
formally
exclude
indirect
influences
of BNSTpr
silencing
on the
altered
representations
in VMHvl
or
MPOA (
pg. 11)
.
Comment
4.
“The
above
concern
is
also
applied
to
behavior.
When
the
BNST
-
Esr
to
-
VMH
-
Esr
or
to
MPOA
-
Esr
is
specifically
silenced,
what
is
the
effect
on
behavior?”
Response
:
We
have
now
performed
these
experiments,
as
suggested
by
the
reviewer.
We
have
silenced
BNSTpr
Esr1
terminals
in
VMHvl
and
MPOA
using
Halorhodopsin,
during
resident
intruder
assays.
Silencing
of
the
BNSTpr
Esr1
VMHvl
projections strongly inhibited ongoing aggression towards males, while causing a
modest reduction in the transition from approach/sniff to mounting towards females;
Silencing
of the
BNSTpr
Esr1
MPOA
projections
strongly
inhibited
the
transition
from
approach/sniff to mounting towards females, while having little effect on male-
male
aggression. These results phenocopy the effects of silencing BNSTpr cell bodies
(
Fig. 1
) and therefore cannot be ascribed to
paradoxical effects of eNpHR3.0 at
nerve
terminals.
We
have
now
included
these
results in
the
revised
manuscript
pg. 3
and
in
ED
Figure
2.
Comment
5.
“The
rationale
to
focus
on
the
Esr
population
is
not
clear.
Are
the
ESR
negative
subset
of AB
neurons
not
responding
to male
and
female
cues
or projecting
to
the
MPOA and VMH? If not, what is their role in the model?”
Response
:
We
focused
on
the
Esr1
+
populations
in
all
3
structures
because
1)
these
neurons have been functionally implicated in sniffing,
mounting and attack by
perturbation experiments in VMHvl and MPOA; 2) it allowed us to silence Esr1
+
neurons in BNSTpr while simultaneously imaging Esr1
+
neurons in VMHvl and
MPOA, using the same Esr1
-Cre driver, simplifying the genetics; Aromatase-
Cre is
not
useful
for
that
purpose.
In response to the reviewer’s question, AB neurons constitute a relatively minor
population in BNSTpr. Aromatase
marks
a subset of Esr1
+
neurons, and
the Esr1
negative
subset
of AB
neurons
is even
smaller.
[REDACTED]
The
curiosity
-
driven
question
of determining
the
role
of the
Esr1
-negative
subset of AB
neurons
would
require
complex,
expensive
and
time
-consuming
genetic
intersectional
strategies,
and
is
tangential
to
the
central
point
of
this
paper.
Comment 6.
“Many of
the BNST
-
Esr+ neurons are not active in the presence of either a
male
or female
(Figure
2).
Are
these
neurons
also
projecting
to the
MPOA
or VMH?
The
use
of mating and aggression as a behavioral proxy for sex identification is not very granular.
Maybe
the
BNST
contributes
to
other
aspects
of
social
motivation
that
result
in
the
phenotypes
in
figure
1.
Do
the
authors
have
another
measure
for
sex
recognition
that
could
support
the
functional
conclusions
of
the
Esr
neurons
driving
sex
recognition
?”
Response
: In response to the reviewer’s initial question, we believe that most of the
BNSTpr
Esr1
neurons
that
are
not
active
in the
presence
of either
a male
or female
are
either interneurons, or cells that do not project to VMHvl
or MPOA. To confirm this,
we have performed a new
imaging analysis of BNSTpr
Esr1
VMHvl and BNSTpr
Esr1
MPOA
projection
neurons
(labeled
by retrograde
delivery
of GCaMP),
and
show
that
most
of the
back
-labeled
BNSTpr
Esr1
neurons
respond
to either
male
or female
cues (
ED Figure 9g
-
l
).
To address the reviewer’s second point (italics), we have performed optogenetic
silencing of
BNSTpr
Esr1
neurons
during
male
vs.
female
urine
preference
assays
and
(pencil cup-
enclosed) male vs. female preference tests.
Our results indicate that the
preferences
for
female
vs.
male
urine,
as well
as the
preference
for
interacting
with
a
female vs. a male restrained in a pencil cup, are lost upon BNSTpr
Es1
silencing,
consistent with results reported in Bayless et al. (2019)
. Surprisingly, however, we
found that despite the loss of
preference
for female cues when BNSTpr is silenced,
ultrasonic vocalizations (USVs) towards females or female urine but not males or
male
urine;
Karigo
et
al.
2021
Nature
)
remained
intact
(
ED
Figure
1o
-
r
)
.
This unexpected new finding has caused us to revise our original conclusion. It
suggests that in sexually experienced males, activity in BNSTpr is not required to
identify or recognize intruder sex, but rather to exhibit a
preference
for
female cues
over male cues. (By analogy, a person who once preferred apples to oranges may
lose
that
preference,
while
still
being
able
to identify
apples
vs.
oranges
apart
by
their
smell.) The observation that sex recognition is still intact following BNS
Tpr silencing
fits with our original observation that intruder sex
can still be efficiently decoded from
neuronal activity in either MPOA
Esr1
or VMHvl
Esr1
neurons following such silencing
(Fig. 3l, 3m). These data suggest that the deficits in mounting and attack caused by
silencing BNSTpr neurons cannot be explained by a failure of sex identification,
[REDACTED]. Rather, when taken together with our new results from silencing
BNSTpr
MPOa/VMHvl terminals during social interactions (see
ED Figure 2
and
response
to Comment
4, above),
the
data
suggest
that
BNSTpr
Esr1
silencing
prevents
mounting and attack as a consequence of its effects on neural activity and sex
representations
in
VMHvl
and
MPOA.
We
have
now
included
these
results
in
the
revised
manuscript
on
pg.
10
,
and
illustrated
them
with
a
new
diagram
in
Fig.
5n
.
In
summary,
our
new
data
argue
for a
view
of BNSTpr
function
that
is
substantially
different
from
that
suggested
[REDACTED].
We
thank
the
reviewer
for
suggesting
the
key
experiments
that
brought
this
important
revision
to
light.
Comment
7.
“Are
all
of
the
BNST
-
Esr
neurons
that
project
to
the
two
studied
targets
inhibitory?
If
not,
this
could
confound
the
model.”
Response
: We thank the reviewer for raising this important question. Recently
published BNSTpr single
-cell RNA sequencing data (Welch et al., 2019, Cell, 177,
1873–
1887) indicate that 95% of the BNSTpr
Esr1
neurons are inhibitory. Thus, it’s
reasonable to assume th
at the majority of the BNSTpr
Esr1
projection neurons are
inhibitory as well. To confirm this, we have performed new experiments using cre-
dependent
retrograde
AAVs
expressing
mNeongreen
or mScarlet
in MPOA
or
VMHvl
of VGAT
-Cre mice. The results (
ED Fig. 9
a-f
) indicate that most of the back labeled
VGAT+ neurons are Esr1
+
. We have now included both the citations and our
retrograde
tracing
results
in
the
revised
manuscript
on
pg.
2
.
Comment
8.
“Is
it
possible
to
manipulate
the
sensory
signals
so
that
BNST
and
VMH
activity
can
be
monitored
as they
mis
-identify
the
sex
of the
partner?
A gender
illusion?
This
would
control for all other aspects of behavior and neural activity.”
Response
:
We
appreciate
the
utility
of
the
experiment
suggested
by
the
reviewer,
in
principle.
Unfortunately, there is no validated procedure for creating “gender
illusions” in mice. However, we present dual
-site (VMHvl and MPOA) fiber
photometry data as a reviewer figure (Reviewer Figure 1) from triadic interactions
with bot
h a male and a female intruder, in which paradoxical male-
directed sexual
behavior (which could reflect a “gender illusion” of the sort referred to by the
reviewer) was observed. These results are described in detail in our response to
comment
9
below.
Comment 9.
“The model (EDF8) suggests that absolute incoming sensory activity is
weighed, more female activity = mating, more male activity = aggression. Can you
manipulate
this
by allowing
your
subjects
to interact
with
multiple
females
and
a single
male
simultaneously?
It
looks
like
the
BNST
activity
persists
as
long
as
a
female
is
present
(fig
2j).
Is this correct?
If the subject first interacts with females and then a male is also added
subsequently,
does
the
simultaneous
presence
of a male
alter
the
balance
of activity
in
the
VMH
? (In both cases, I expect the presence of a male would promote aggression, but the
representation
of
male
and
female
should
not
change.)
Response
: The exact experiment the reviewer requested is extremely
difficult to
perform, for reasons that will become apparent in the following description. As an
approximation
to this
experiment,
we
have
now
included
bulk
calcium
measurements
acquired by dual fiber
-photometry performed simultaneously in VMHvl and MPOA
Es
r1+
neurons
in the
same animal
(described and validated in in Karigo
et al., 2021,
Nature
), recorded during
triadic interactions
between a resident male, an intruder
female and an intruder male present in the same cage
(Reviewer Fig. 1
), as the
reviewer
suggested.
We
provide
examples
from
two
such
interactions.
In both
cases,
the male resident first interacts with the female intruder and then interacts with the
male
intruder.
In the first, characteristic example (
Reviewer Fig. 1
upper recording)
,
when
the
recorded
male
is sniffing
or mounting
the
female
(purple
outlined
box;
red
and
green
rasters, respectively), activity in MPOA (purple trace) is higher than in VMHvl (gray
trace). The converse (VMHvl>MPOA activity) is observed when the resident is
sniff
ing the male intruder (gray box; blue raster). Towards the end of the recording
session, as the animal switches from mounting a female to sniffing a male, MPOA
activity
rapidly
declines
and
VMHvl
activity
increases
(red
box).
These results from a triadic interaction confirm observations made on dyadic pairs
published in Karigo et al. (2021), which showed that in the presence of a female
intruder, MPOA>>VMHvl activity, while in the presence of a male intruder,
VMHvl>MPOA
activit
y. These
opposite
ratios
of MPOA:VMHvl
activity
in bulk
calcium
measurements reflect the fact that in MPOA (as in BNSTpr), female-
tuned neurons
outnumber male tuned neurons by ~2:1, whereas the converse is true in VMHvl
(Remedios
et
al.,
2017;
Karigo
et
al
.
2021).
The
lower recording in
Reviewer Fig. 1
shows a
rare case
of what appears to
be a
“gender illusion,” as mentioned by the reviewer in comment 8.
In this [different]
triadic assay, following a sniffing/mounting interaction with the f
emale (purple box,
red and green rasters), the male exhibits paradoxical
sexual
mounting (marked by
ultrasonic vocalizations, USVs; see Karigo et al. 2021) towards the intruder male
(yellow raster). During this USV
+
male mounting behavior, MPOA activity initially
increases
relative
to VMHvl
activity
(red
box
1, dashed
vertical
line),
the
opposite
to
what is typically observed (see above description of upper recording).
The
fact that
the
resident
male
sings
when
mounting
the
male
intruder
confirms
that
he
perceives
the intruder as a female (Karigo et al., 2021). Following this male-
directed USV
+
mounting
bout,
the
animal
resumes
sniffing
the
male.
During
that
period,
MPOA
[REDACTED]
activity
slowly
falls
while
VMHvl
activity
increases
(lower
recording,
red
box
1,
traces
to right of dashed vertical line). This suggests the resident male now perceives this
intruder as a male. Eventually, following another interaction with the female, during
sniffing of the male (gray box), the relative level of
MPOA vs. VMHvl activity flips,
reflecting the typical pattern (upper panel): MPOA shows a characteristic decline in
activity, while VMHvl activity rises.
The
data
in the
second
(lower)
recording
suggest
that
the
initial
attempted
sexual
mounting
of
the
mal
e
(red
box
1)
may
reflect
a
spontaneous
“gender
illusion,”
in
which
the
resident’s
brain
mistakenly
identified
the
male
intruder as
a
female
(MPOA
activity>VMHvl
activity).
They
reinforce
our
previous
conclusion
that
activity
in
MPOA
and VMHvl does not simply reflect sensory responses to male-
or female
-specific
cues, but rather a percept of intruder sex (Karigo et al., 2021). These data further
support the
reviewer’s inference
that more
“female activity” (MPOA>VMHvl) leads
to
mat
ing
and
more
“male
activity”
(VMHvl>MPOA)
leads
to aggression,
consistent
with
the female-
vs. male
- tuning bias in MPOA vs. VMHvl, respectively. Unfortunately,
the low frequency of these spontaneous “gender illusion” events (1 in ~20 mice
tested) makes it
impractical to perform single-
cell calcium imaging during such
events,
due
to the
very
large
number
of animals
that
would
have
to be
implanted
with
miniscopes
to
image
activity
during
even
a
single
such
behavioral
event.
Comment
10.
“Figure
4a,
it
appears
that
with
CNO
(BNST
-
)
the
female
responding
cells
are
largely spatially segregated from the male responding cells and about half of the male
responding cells remain male responding.
Are these two populations, those that switch vs
retain sex tuning without BNST input
different molecular subsets
of the VMHvl
-ESR
population
? On repeated trials, is the same neuron able to switch sometimes and remain
stable
other
times,
or
are
they
set
to
be
either
flexible
or
fixed?
Response
: The apparent spatial
segregation of female
-
vs. male
-
responding cells to
which the reviewer refers is not a consistent observation across animals. To our
knowledge, there is no method that would allow us to track the neurons that switch
vs. retain sex tuning following BNST sil
encing via microendoscopic calcium imaging
in freely
moving animals, and
then determine
the
molecular profile of these neurons,
as the reviewer suggested. (Such an experiment might be possible in head-
fixed
animals,
but
they
will
not
perform
the
social
behaviors
studied
here.)
We
look
forward
to the time when such technology has been developed. In answer to the last
question, we did not perform repeated CNO trials in the same animals, in order to
prevent experience-
dependent changes in the animals that coul
d confound the
results.
Comment
11.
“I
cannot
find
CNO
only
controls
on
neural
activity
and
behavior,
and
some
quantification/analysis
for
the
silencing
of
BNST
-
Esr
expressing
hm4di
neurons.”
Response
:
We
have
now
reported
the
percentage
of
BNSTpr
-
Esr1
neurons
that
express
hM4Di,
and
included
CNO
only
controls,
in
ED
Figure
1c,
l,
m.
Comment
12.
“Fig
2:
Representative
images
of
GCaMP
infections
in
BNST
Esr1
neurons
will
be useful.
Response:
We
have
now
provided
a representative
image
of GCaMP
infections
in
BNST
Esr1
neurons
in
ED
Figure
3i.
Comment
13.
“Fig
3:
Representative
images
of
DREADDs
infections
in
BNSTEsr1
neurons
+
GCaMP
in
MPOA/VMHvl
Esr1
neurons
will
be
useful.”
Response
:
We
have
added
the
requested
images
in
ED
Figure
5b
-
o
,
as
well
as
schematic
diagrams
to
facilitate
their
interpretation
by
the
reader
(
ED
Fig.
5a,
f,
k
).
Comment
14.
“Ext
[Data]
Fig
4:
“Why
are
error
bars
missing
in
k?”
Response
:
We
only
imaged
a
single
animal
(the
only
one
of 2
implanted
animals
that
exhibited adequate GCaMP expression), to confirm that separate populations of
female-
and male-
tuned neurons exist in the BNST aromatase+ (AB) population.
Therefore
no
error
bars
were
shown.
This
experiment
is simply
an
existence
proof
to
demonstrate
that
the
difference
between
our
conclusions
and
those
of Bayless
et al.
(2019) is not due to the use of different Cre drivers.
We
again
thank
this
reviewer
for
their
insightful
and
thoughtful
questions,
and
hope
that
our
efforts
to
address
them
have
improved
the
paper.
Reviewer
#
2
We
thank
the
reviewer
for
their
helpful
questions
and
suggestions,
to
which
we
respond
below.
Comment 1.
“It is interesting that the VMH and MPOA still display sex
-
biased responses
after
chemogenetic
inhibition
of the
BNST.
This
could
be
due
to factors
discussed,
such
as
a
role for another brain region, but also could be due to incomplete BNST inhibition in
chemogenetic
experiments, related to either (1) the extent of coverage of the large BNST
area by AAV injections, and 2) the Cre lines used. For (1), what % of Cre
-expressing BNST
cells are labeled and silenced by AAV injections, and for (2) are there sex
-selective BNS
T
neurons that do not express aromatase and/or ESR and could be contributing to
downstream
representations?”
Response
: These are reasonable questions, for which we thank the reviewer. To
address them, we have now included a panel showing
that over 90% of the Esr1
+
BNSTpr neurons are labeled by AAV expressing hM4D
-mCherry, as the reviewer
requested (
ED figure 1c
). Single cell RNA
-seq data (Welch et al., 2019, Cell, 177,
1873–
1887) indicate that 25% of the inhibitory population in BNSTpr is
Esr1
+
. We
cannot exclude that there would be a stronger behavioral or circuit
-level perturbation
if additional, non
-Esr1
+
BNSTpr neurons were silenced. That said, if silencing the
Esr1
+
subset
of BNSTpr
neurons
was
inadequate
to strongly
perturb
the
syste
m,
then
we would have expected to see NO behavioral phenotype.
To the contrary, we
observed very robust and highly penetrant behavioral phenotypes, including the loss
of preference for female over male cues, and the inhibition of both mounting and
attack
(Fig.
1f,
k
and
ED
Fig.
1b).
Comment
2
.
“The
authors
should
ensure
that
AAV
injection
in
the
BNST
does
not
label
neurons
in the
VMH
or MPOA
(for
example
by retroactive
labeling
of neurons
providing
feedback control), which could confound
interpretations.”
Response
:
The
reviewer
is
correct
to
bring
up
this
important
concern.
We
have
added
a figure
demonstrating
the
lack
of cell
body
labeling
in VMHvl
and
MPOA
following AAV injections into BNST in
ED Figure 5a
-
o
.
Comment 3
. “The
authors use a nicely comprehensive set of stimuli in Figure 2j, and
showed that BNST response variance due to intruder sex was larger than the associated
behavior.
It would
be
worth
discussing
these
observations
in the
context
of the
lab's
previous
work on
VMH/MPOA. Is there an increased response variance due to behavior in the VMH
and MPOA, suggesting further input transformation as information moves to the
hypothalamus?”
Response
: The reviewer brings up a great point.
There is indeed an increase
d
fraction of variance explained by behavior in MPOA, relative to BNST, and we have
now mentioned those data in the
revised manuscript pg. 9 and Figure 5g
. In
VMHvl,
the
fraction
of variance
due
to intruder
sex
is larger
than
that
due
to
behavior,
as descr
ibed in Karigo et al. (2021). These results are now summarized in a new
diagram
in
Fig.
5m.
Comment
4
.
The
authors
focus
here
on
neuronal
representations
in
male
mice;
do
similar
sex
biases
in
BNST
and
VMH
exist
in
female
mice?
Response
: This is an important question that we appreciate, but it would require
repeating all of the imaging experiments in female mice, with animals analyzed at
different
stages
during
the
estrus
cycle.
This
represents
a large
amount
of
additional
work
that
likely
would
raise
more
questions
than
it would
answer.
We
therefore
feel
it
is
more
appropriate
for
a
separate,
follow
-
up
study.
Comment 5
. “Both the title and last sentence of the abstract should be edited for clarity to
make
the
manuscript
more
accessible
to a general
audience.
(I would
go with
something
like
'Transformations
of sex
representations in
the ascending limbic
system' but of course
this is
just
a
suggestion!)”
Response
:
We
appreciate
the
suggestion
and
have
modified
the
title
along
the
lines
of
the
reviewer’s
suggestion.
Comment
6
.
“For
ED
4b,
it
seems
like
activity
is
synchronized
to
particular
events
in
the
time
series
- if true,
it would
be
helpful
to provide
annotation
of whether
such
synchronized
events correspond to sniffing
or other social episodes.”
Response
:
We
thank
the
reviewer
for
this
suggestion,
and
have
now
added
such
behavioral
annotation
to
ED
Figure
4b.
Reviewer
#3
We
thank
this
reviewer
for
taking
the
time
to
read
the
manuscript
carefully,
and
for
their
thoughtful
comments.
Comment 1
. “The neuroimaging
-
chemogenetic data set (the main part which provide novel
findings),
although
interesting,
is purely
descriptive
and
missing
any
mechanistic
explanation
(beyond the proposed hypothesis). Namely, how does sex
-biased neuronal coding in the
MPOA/VMH/BNST (or the altered sex
-bias in the VMH following chemogenetic silencing)
encode sex
-typical stimuli and control different reproductive behaviors (mating and
aggression)
towards
males
and
females?”
Response
:
We
respectfully
disagree
with
the
reviewer’s
characterization
of
our
experiments as “purely descriptive.” We have performed a genetically based
perturbation
experiment
(silencing
of
BNSTpr)
and
have
characterized
the
phenotypes of that perturbation at both the behavioral and circuit levels, in two
different downstream targets of BNSTpr, a first
-in-class study.
Our observations
reveal
an
unexpected
transformation
performed
by BNS
Tpr
Esr1
neurons
(inverting
the
ratio
of
female:male
-
specific
neurons
in
VMHvl,
relative
to
MPOA).
We originally provided two complementary mechanistic hypotheses to explain how
the
observed
perturbations
of neural
activity
could
explain
the
observed
perturbations
in social behavior, following BNSTpr silencing. [REDACTED] In the interests of
distinguishing these hypotheses, and thereby gaining further mechanistic insight as
the reviewer requested, we have performed an extensive additional series of
Che
moScope experiments to monitor activity in MPOA and VMHvl before vs. after
silencing BNSTpr, during unrestrained male-
male and male-
female social
interactions.
(As
the
reviewer
may
recall,
our
original
ChemoScope
experiments
were
performed using dangled intruders, preventing free social interactions). In this
experiment, we performed the DREADD
-mediated silencing of BNSTpr
Esr1
neurons
unilaterally
(rather than bilaterally as in the case of behavioral assays, e.g., Fig. 1).
Because there are very few inter
-hemispheric connections in this circuit, the
unperturbed contralateral side of the brain is able to compensate for silencing on the
ipsilateral (imaged/silenced) side, and therefore social behavior is intact. This design
eliminates
the
confound
that
any
obs
erved
changes
in neural
activity
in
MPOA/VMHvl
are simply a reflection of changes in behavior.
It also allows us to temporally
correlate alterations in MPOA or VMHvl activity caused by silencing BNSTpr, with
specific behavioral states or transitions made by the animal. The results of these
experiments
are
now
presented
in
a
new
Figure
5
,
as
well
as
in
ED
Figures
8
and
9.
These
new
data
include
unexpected
results
that
strongly
favor
one
hypothesis
over
the
other,
and
provide
mechanistic
insights.
There
are
3
basic
findings
we
report:
1)[REDACTED] a requirement for BNSTpr in sex recognition was inferred
from the
observation
that
a
preference
for
female
over
male
cues
was
lost
when
Aromatase
+
neurons
in
BNSTpr
were
ablated
or
chemogenetically
silenced.
We