Archival Report
Biological
Psychiatry
Identification of a Corticohabenular Circuit
Regulating Socially Directed Behavior
Madhurima Benekareddy, Tevye Jason Stachniak, Andreas Bruns, Frederic Knoflach,
Markus von Kienlin, Basil Künnecke, and Anirvan Ghosh
ABSTRACT
BACKGROUND: The prefrontal cortex (PFC) has been implicated in the pathophysiology of social dysfunction, but
the specific circuit partners mediating PFC function in health and disease are unclear.
METHODS: The excitatory designer receptor exclusively activated by designer drugs (DREADD) hM3Dq was used to
induce PFC activation during social behavior measured in the three-chamber sociability assay (rats/mice). Functional
magnetic resonance imaging was combined with hM3Dq-mediated PFC activation to identify novel nodes in the
“social brain” in a hypothesis-free manner. In multiplexed DREADD experiments, hM3Dq and the inhibitory KORDi
were used to bidirectionally modulate PFC activity and measure social behavior and global functional magnetic
resonance imaging signature. To characterize the functional role of specific nodes identified in this functional
magnetic resonance imaging screen, we used anterograde and retrograde tracers, optogenetic and DREADDassisted circuit mapping, and circuit behavioral experiments.
RESULTS: PFC activation suppressed social behavior and modulated activity in a number of regions involved in
emotional behavior. Bidirectional modulation of PFC activity further refined this subset of brain regions and identified
the habenula as a node robustly correlated with PFC activity. Furthermore, we showed that the lateral habenula (LHb)
receives direct synaptic input from the PFC and that activation of LHb neurons or the PFC inputs to the LHb suppresses social preference. Finally, we demonstrated that LHb inhibition can prevent the social deficits induced by
PFC activation.
CONCLUSIONS: The LHb is thought to provide reward-related contextual information to the mesolimbic reward
system known to be involved in social behavior. Thus, PFC projections to the LHb may represent an important
part of descending PFC pathways that control social behavior.
Keywords: Autism, Chemogenetics, DREADDs, fMRI, Lateral habenula, Prefrontal cortex
https://doi.org/10.1016/j.biopsych.2017.10.032
Impairment in social function is one of the core deficits of
autism spectrum disorder (ASD) and is a prominent feature of
schizophrenia. The neural circuit basis of impaired social
function in disease states is only beginning to be understood,
with several lines of evidence pointing to an important role
played by frontal cortical areas. The frontal cortex in patients
with schizophrenia shows a marked gamma-aminobutyric
acidergic deficit in postmortem studies (1). Likewise, ASD is
associated with an aberrant recruitment of frontal regions
during tasks of social cognition (2). Notably, about 50% to
70% of autistic children exhibit electroencephalographic
abnormalities indicative of a hyperactive cortical network (3,4),
and 30% of autistic individuals develop seizures (5). In addition, postmortem cortical samples from patients with ASD
show mutations in genes controlling excitability and synaptic
function (6). Excitation–inhibition coordination in the prefrontal
cortex (PFC) may provide a common circuit substrate for the
various cellular, molecular, and genetic causes of disease
states of social impairment (7,8). Indeed, an acute optogenetic
increase in excitation in the PFC impairs social behavior (9).
However, the descending PFC projections that affect the
prefrontal control of social behavior have not been identified.
Here, we demonstrate that increasing PFC excitation leads
to a robust decrease in sociability in both rats and mice,
extending the initial optogenetic findings from Yizhar et al. (9).
To identify specific brain regions that act in tandem with the
PFC in mediating social dysfunction, we performed a neuroimaging screen for brainwide activity changes caused by a
targeted activation of the PFC. This magnetic resonance
imaging (MRI) screen revealed significant changes in activity in
brain regions related to emotional processing, with a notably
strong activation seen in the habenula. We then performed a
circuit rescue experiment to further refine this subset of brain
regions for coregulations during bidirectional modulation of
PFC activity and identified the habenula as one of the key
components of the PFC network controlling social behavior.
We determined that there are direct projections from the PFC
to the lateral habenula (LHb) and found that directly activating
SEE COMMENTARY ON PAGE 546
ISSN: 0006-3223
ª 2017 Society of Biological Psychiatry. 607
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
LHb neurons or activating PFC projections to the LHb led to a
decrease in sociability without inducing aversion or locomotor
hyperactivity. Finally, habenula inhibition concomitant with
PFC activation prevented the social deficits induced by PFC
activation. Taken together with the emerging role of the LHb in
higher-level cognitive control of reward-related behaviors
(10–14), our results identify the LHb as a novel component of
the descending PFC projections that enable state-dependent
control of social behavior.
Netherlands) and was performed with a minimum interval of 7
days between consecutive tests with a parallel or crossover
design.
METHODS AND MATERIALS
Perfusion Imaging Using Functional MRI
Detailed protocols can be found in the Supplement.
On the day of the experiment, animals were anesthetized with
2.0% to 2.5% isoflurane (Abbott, Cham, Switzerland) in a
mixture of oxygen (20%) and air (80%), and tissue perfusion
was measured in a Bruker Biospec 4.7T/40-cm instrument
(Bruker Biospin, Ettlingen, Germany).
Animals
Rodents (Emx1::Cre [JAX stock # 005628] mice and OFA rats
[Sprague Dawley; Saint-Germain-sur-l’Arbresle, France]) were
maintained on a reversed 12-hour light/dark cycle with ad
libitum access to food and water. Experiments were conducted in strict adherence to the Swiss federal ordinance on
animal protection.
Cell Type–Specific Expression of Designer
Receptors Exclusively Activated by Designer Drugs
and Chronos
Adeno-associated virus (AAV) vectors packaged by the University of North Carolina Vector Core were used to express
designer receptors exclusively activated by designer drugs
(DREADDs) or Chronos. The list of viruses, anesthesia, and
stereotactic coordinates is provided in the Supplement.
Laser Delivery
A 1-mm fiber-optic patch cable (Doric Lenses Inc., Quebec,
Canada) was directly connected to the implanted ferrule with a
rotary joint (FC/FC connector; Doric Lenses Inc.). The light
stimulation consisted of a continuous train of 20-Hz pulses
(473 nm, diode-pumped solid-state laser; OEM Laser Systems,
Midvale, UT) with 15-ms pulse duration. The light stimulus
intensity (5.5 mW) was selected to reduce the probability of offtarget Chronos activation of nearby thalamic fibers (intensity
, 5 mW/mm2; 0.5 mm from fiber tip; http://web.stanford.edu/
group/dlab/cgi-bin/graph/chart.php).
Administration of DREADD Ligands
Behavior was performed 45 and 15 minutes after the hM3Dq
ligand clozapine-N-oxide (CNO) (Enzo, Farmingdale, NY;
0.3–5.0 mg/kg in 0.9% saline; intraperitoneal) and the KORDi
ligand salvinorin B (SALB) (Cayman, Ann Arbor, MI; 17 mg/kg
in 10% dimethyl sulfoxide; subcutaneous) administration,
respectively. One hundred percent dimethyl sulfoxide was
used as the vehicle to improve solubility of SALB (B. Roth,
M.D., Ph.D., personal communication, September 2015).
Three-Chamber Sociability Test
This test involves habituating the subject to the three-chamber
box (10 minutes) and then placing a novel conspecific at one
end of the arena and allowing the subject to explore the arena
for 10 minutes (test). Data are expressed as sociability index =
100 3 (time in social interaction/total interaction time) 2 50.
Behavior was tracked using Ethovision (Noldus, Wageningen,
608
Sucrose Preference
Subject rats were given a free choice of drinking either water or
0.5% sucrose for 4 hours after injection of CNO (1 mg/kg),
consistent with the known maximal effect of CNO on feeding
behaviors (15).
Electrophysiology
The holding potential for voltage clamp recordings was 270
mV (Cl2Erev). LED photostimulation via a fiber-optic tip (w20
mW) generated full field illumination.
RESULTS
Global Functional MRI Signature Associated With
Social Dysfunction
To screen for global activity changes associated with PFC
hyperactivity and social dysfunction, we took advantage of
DREADDs (16–18). DREADDs are engineered G protein–
coupled receptors designed to respond to an otherwise
biologically inert ligand. To increase excitation in the PFC, we
transduced PFC pyramidal neurons (Emx-positive) with the
excitatory DREADD hM3Dq. Using multielectrode arrays, we
confirmed that DREADDs led to a persistent and strong
increase in neuronal activity. Brain slices containing hM3Dq in
PFC pyramidal neurons were placed on multielectrode arrays,
and neuronal activity was recorded in response to the ligand
CNO (Figure 1A–C). Bath application of CNO (0.3–3.0 mM) led
to an increase in activity in the PFC neurons expressing hM3D,
an effect that persisted after CNO withdrawal. Persistent neural
activity was confirmed by reversal with tetrodotoxin. Multiunit
activity from hM3D-negative regions showed no change with
CNO administration (Figure 1C, D). After confirming that
DREADDs led to an increase in neuronal activity, we tested the
consequences of this increase in PFC neural activity on social
behavior measured using the three-chamber sociability test
(19). An acute increase in PFC excitability mediated by CNO/
hM3Dq in Emx-positive neurons led to a significant decrease in
sociability (Supplemental Figure S1A), with no effect on
locomotion during the social behavior (Supplemental
Figure S1B–D) or an independent open field task
(Supplemental Figure S1E–G). We then translated this circuit
model into rats using AAV-CamKII-hM3D, where increasing
PFC activity decreased sociability, thereby demonstrating the
general validity of this circuit model across rats and mice. A
two-way analysis of variance (ANOVA) revealed a significant
time 3 treatment interaction (Figure 1G) (n = 7–9, F1,28 = 4.38,
p = .045). Time spent in each of the three chambers is provided
in Supplemental Table S3. Behaviors in the open field test were
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
also measured in rats and showed no effect of PFC activation
(Supplemental Figure S1H–L).
Our behavioral studies demonstrated a causal link between
PFC excitability and social dysfunction. However, the specific
circuit partners that mediate the effect of the PFC on social
behavior remain unknown. A hyperactive PFC could exert
brainwide effects via its direct or polysynaptic connections to
midline cortical structures, subcortical nuclei, the thalamus, the
hypothalamus, or the brainstem (20,21). To directly address
the circuit-wide response to PFC activation in a hypothesisfree manner, we measured the activity of downstream
circuits with functional MRI (fMRI) upon noninvasive chemogenetic activation of the PFC. We measured neural activity in
distinct anatomically defined brain regions using perfusion
imaging based on continuous arterial spin labeling, the perfusion values of which measure tissue perfusion as a proxy for
neuronal activity (22) and provide a quantitative way of
screening globally for neural activity changes induced by a
targeted chemogenetic manipulation. Remote control of the rat
PFC was achieved by AAV-CamKII-hM3Dq, and a hyperactive
PFC state was induced by the application of CNO (3 mg/kg)
(Figure 1H). CNO induced a robust significant increase in the
fMRI signal in the PFC, the site of hM3Dq expression
(Figure 1H–J) (p , .01).
Our whole-brain neuroimaging screen revealed significant
neural activity changes in a subset of the PFC anatomical
network, including several regions with functional links to the
PFC and emotional behavior. Notably, we observed a robust
increase in activity in the habenula and changes in activity in
the nucleus accumbens (NA), entorhinal piriform cortex, and
medial hypothalamus (Figure 1H–J) (n = 20 rats), brain
regions from the emotional and motivational circuitry that
could contribute to social behavior (23–26). The findings
summarized in Figure 1J highlight the role of the PFC in
sculpting activity in key brain regions involved in emotional
processing. CNO did not elicit a significant change in the
neural activity pattern in animals with no expression of
hM3Dq, with the exception of a change in the septum
(Supplemental Table S2). Collectively, our DREADD–fMRI
signature of PFC activation reveals robust changes in
neural activity in key brain regions that could underpin the
behavioral effects of PFC hyperactivity.
Multiplexed Chemogenetic Rescue of PFC
Hyperactivity and Social Behavior
Our DREADD–fMRI and behavioral data revealed that PFC
hyperactivity causes social dysfunction and modulates neural
activity in key brain regions such as the habenula. We subsequently addressed whether reversing PFC hyperactivity
would rescue social behavior and its associated fMRI signature. For this rescue experiment, we leveraged recent advances in multiplexed DREADDs (18) that enable bidirectional
modulation of PFC activity. To build confidence in this
emerging technology, we designed a behavioral experiment to
independently confirm the link between PFC hyperactivity and
social deficits. We transduced principal neurons in the rat PFC
with the stimulatory hM3Dq (Figure 2A) and the inhibitory
KORDi. We hypothesized that because PFC activation leads to
a decrease in sociability by orchestrating circuit-wide effects,
attenuating PFC activation should restore sociability deficits.
We first verified that KORDi was capable of reversing neuronal
activation by hM3Dq. As a proxy for PFC activity, we recorded
spontaneous excitatory postsynaptic currents (sEPSCs) from
coronal rat brain slices transduced with both hM3Dq and
KORDi (Supplemental Figure S2A, B). Following a 10-minute
period of recorded baseline, application of CNO (3 mM;
3 minutes) increased spontaneous synaptic activity
(Supplemental Figure S2A). Addition of SALB (3 mM) in combination with CNO (3 mM) reduced sEPSC frequency compared
with CNO alone (Supplemental Figure S2B). In contrast,
application of CNO and SALB on brain slices expressing green
fluorescent protein (GFP) alone revealed no significant effect
on sEPSC frequency (Supplemental Figure S2C, D).
In subsequent behavioral experiments, we asked whether
rescuing PFC hyperactivity rescues social behavior deficits.
Consistent with previous data (Figure 1G), hM3Dq-CNO activation led to a significant decrease in sociability index, while
attenuation of hM3Dq-induced PFC hyperactivity with SALB/
KORDi led to the expression of normal social behavior
(Figure 2A–C) (n = 7, two-way ANOVA for time 3 treatment,
F1,24 = 4.8, p = .04). We next investigated the global fMRI
signature associated with bidirectional modulation of social
behavior by adopting a within-session design (as opposed to
the between-session design in Figure 1H) to produce a higher
signal-to-noise ratio. Neural activity was imaged first at baseline and then upon PFC activation and subsequent PFC inhibition (Figure 2D) as part of the same scanning session. As
expected, our DREADD–fMRI data showed that PFC activation
induced by CNO/hM3Dq (24 minutes post-CNO) was attenuated by SALB/KORDi (15 minutes post-SALB). Furthermore,
PFC activation induced perfusion changes in the habenula; the
medial thalamus and septum were significantly reversed on
SALB/KORDi administration (Figure 2E) (n = 10, p , .05, oneway ANOVA followed by post hoc comparison with baseline).
Interestingly, neural activity in the other brain regions showed
different kinetics in response to SALB administration within the
time frame measured and points to the differential sensitivity of
the various functional connections to changes in PFC activity
levels. The complete data can be found in Supplemental
Table S1.
Collectively, our multiplexed DREADD–fMRI and behavioral
data allowed us to refine the subset of brain regions that are
coregulated with PFC activity and altered states of social
behavior (Figure 2F). Of this subset of coregulated brain
regions, the habenula stands out as a brain region that is
robustly regulated across experiments and points to a potential functional connection between activity in corticohabenular
circuits and social behavior.
LHb Receives Synaptic Input From the PFC
One of the striking observations from our imaging study was
that perfusion in the PFC and the habenula was reciprocally
regulated with social behavior. Electron microscopy and lesion
studies have indicated that direct connections exist between
the PFC and the LHb (27–29). To confirm synaptic connectivity
between the PFC and the LHb, we first expressed AAV
synaptophysin-GFP in PFC neurons. Confocal imaging
revealed a network of synaptophysin-GFP-positive fibers in the
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
609
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
610
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
LHb (Figure 3A) but not in the medial habenula (data not
shown). In a subsequent experiment, we injected the retrograde tracer Alexa Fluor 488–conjugated cholera toxin subunit
B into the LHb and allowed it to trace back to the cell bodies
for 3 weeks. Confocal imaging of the PFC cell bodies revealed
Alexa 488 signal in deep-layer cell bodies of the PFC
(Figure 3B). Taken together, these two experiments provide
evidence for a synaptic connection between the PFC and the
LHb.
To physiologically characterize the functional connections
between the PFC and the LHb, we used optogenetic circuit
mapping. Rat PFC neurons were transduced with AAVChronos under the synapsin promoter, and whole-cell
recordings from coronal slices at the level of the LHb were
used to map functional synapses (Figure 3C). Photostimulation
of Chronos-expressing PFC axons (5 3 10 ms, 10 Hz) evoked
time-locked EPSCs in 38% of LHb neurons (Figure 3D), confirming functional connectivity between the PFC and the LHb.
Noting that DREADDs can directly influence synaptic release
(30), we attempted to determine whether the activatory
DREADD hM3Dq used in our behavioral experiments could
likewise be used to map circuit connectivity. We therefore
examined sEPSCs in coronal slices of LHb from rats transduced with either AAV-CamKII-hM3Dq or GFP in the PFC.
Following a 10-minute period of recorded baseline to allow
sEPSC frequency to stabilize, application of CNO (3 mM for 3
minutes) produced increases in synaptic frequency in a subset
of cells in hM3Dq slices but not in control GFP slices
(Supplemental Figure S3). We found that sEPSC frequency
increased by at least 0.3 Hz (cutoff . 2 SD in GFP slices;
Supplemental Figure S3D) in 35% of the recorded LHb neurons (Figure 3E, F) (n = 7/20 cells, p = .02, Wilcoxon test),
demonstrating DREADD-mediated PFC–LHb connections.
PFC-to-LHb Connections Directly Regulate Social
Behavior
The key finding of our anatomical and electrophysiological
data is a direct functional connection between the PFC and the
LHb. With afferent fibers from the PFC and efferent fibers to
midbrain motivational circuitry, the LHb is ideally positioned to
gate contextual information from the PFC into regions controlling socially directed behavior. Furthermore, our DREADD–
fMRI data revealed reciprocal regulation of social behavior and
neural activity in the PFC and LHb. This led us to hypothesize
that a selective LHb activation would be sufficient to disrupt
social behavior. To address this, we transduced LHb neurons
with AAV-CamKIIa-hM3Dq (Figure 4A) and measured social
behavior after an acute application of CNO. To avoid experimental confounds with long-term effects of habenula activation (24), we employed a parallel study design. We controlled
for test–retest effects and nonspecific effects of CNO by
including a group with GFP only and no hM3D expression. Our
behavior results showed that a CNO-mediated increase in LHb
activity led to a strong and selective decrease in sociability in
the rats expressing hM3D in the habenula (Figure 4B).
Administration of CNO to rats expressing GFP alone showed
no significant effect on social behavior, suggesting that these
results are not a result of nonspecific activation by CNO or
test–retest effects. The rats where the stereotactic targeting
missed the habenula and the hM3Dq expression was in the
surrounding regions, such as the medial thalamus and hippocampus, showed no change in social preference on CNO
administration (Figure 4B). Two-way ANOVA for group 3
treatment shows a significant interaction (n = 5–9, F1,16 = 6.2,
p = .02). A schematic representation of the extent of hM3D
expression in all the rats tested in this experiment is included in
Supplemental Figure S4A. Notably, the DREADD-mediated
increase in LHb activity did not alter locomotion and exploration during the habituation and locomotor activity during the
social test (Supplemental Figure S4B–D). In addition, LHb
activation did not induce behavioral avoidance of the social
chamber, as evidenced by the chamber time (Supplemental
Table S3), nor did it affect sucrose preference (Supplemental
Figure S4E). Collectively, these results demonstrate a direct
modulation of social behavior by ongoing activity in the LHb.
We next wanted to address whether this PFC-to-LHb
pathway represents a social behavior pathway that is dissociable from the role of specific LHb outputs in real-time aversion. To directly address the involvement of a corticohabenular
=
Figure 1. Neuroimaging screen for global activity changes associated with social dysfunction. (A) Experimental strategy for expression of the excitatory
designer receptor exclusively activated by designer drugs (DREADD) receptor hM3Dq in pyramidal neurons in the mouse prefrontal cortex (PFC). (B) Bright field
image of a slice expressing hM3Dq in the pyramidal neurons of the PFC placed on a multielectrode array. (C) Representative fluorescent image of hM3Dq-mCherry
expression in the PFC. (D) Neural activity measured using a multielectrode array on application of different doses of clozapine-N-oxide (CNO) (0.3–3.0 mM) in
hM3Dq-positive and -negative regions. Each row represents spikes from one electrode over time. (E) Experimental strategy for expression of hM3Dq in the pyramidal neurons in the rat PFC. (F) Coronal section at the level of the PFC shows hM3Dq-mCherry targeting (scale bar = 100 mm). (G) Sociability index in response to
vehicle and CNO in rats expressing hM3D or green fluorescent protein (GFP) in Ca21/calmodulin-dependent protein kinase-II (CaMKII)-positive neurons in the PFC
(n = 7–9; *p , .05; two-way analysis of variance followed by Fisher’s post hoc comparison). Experimental strategy involves a parallel design. (H) Outline of functional
magnetic resonance imaging (fMRI) experimental design involving vehicle or CNO injections 45 minutes before fMRI. Representative images from planes show
delta normalized perfusion across the rostrocaudal extent of the brain. (I) Percentage increase in normalized perfusion after CNO application (3 mg/kg; intraperitoneal; n = 20 rats; *p , 0.05; one-way analysis of variance followed by post hoc comparisons, controlled for multiple comparisons at a false discovery rate of
10%). (J) Brain regions whose activity is significantly altered by PFC activation (p value cutoff criterion of p , .05; false discovery rate at 10%), expressed as
percentage change in normalized perfusion compared with vehicle. Data represent mean 6 SEM. AAV, adeno-associated virus; ACSF, artifical cerebrospinal fluid;
Amy, amygdala; BmAmy, basomedial amygdala; BNST, bed nucleus of the stria terminalis; cAmy, central amygdala; CoAmy, cortical amygdala; dHIP, dorsal
hippocampus; DP, dorsal peduncular cortex; DR, dorsal raphe; DS, dorsal striatum; Ect, ectorhinal cortex; Ent, entorhinal cortex; EP, entorhinal piriform cortex; Hb,
habenula; IC, inferior colliculus; ICtx, insular cortex; ITh, lateral thalamus; LC, locus coeruleus; lHy, lateral hypothalamus; lTh, lateral thalamus; M1, primary motor
cortex; mAmy, medial amygdala; MEA, multielectrode array; mHy, medial hypothalamus; Min, minutes; MR, medial raphe; mTh, medial thalamus; NA, nucleus
accumbens; NAc, nucleus accumbens core; NAs, nucleus accumbens shell; OFC, orbitofrontal cortex; PAG, periaquaductal gray; pdHip, posterior dorsal
hippocampus; PRC, perirhinal cortex; PVHy, paraventricular hypothalamus; SC, superior colliculus; SMCtx, sensorimotor cortex; SN, substantia nigra; Sub,
subiculum; TTX, tetradotoxin; V1, primary visual cortex; Veh, vehicle; vHip, ventral hippocampus; VP, ventral pallidum; VTA, ventral tegmental area.
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
611
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
612
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
Figure 3. Lateral habenula (LHb) receives direct
functional connections from the prefrontal cortex
(PFC). (A) Experimental strategy and representative
confocal images showing synaptophysin-green
fluorescent protein (GFP) injection in the PFC and
synaptophysin-GFP-positive fibers in the LHb. Scale
bar = 50 mm. (B) Experimental strategy and representative confocal images showing fluorescence of
Alexa Fluor 488–conjugated cholera toxin subunit B
(CTb-Alexa 488) injection in the LHb and Alexa 488–
positive cells in the PFC. Scale bar = 50 mm. (C)
Experimental strategy for electrophysiological recordings. PFC fibers transfected with either hM3D or
Chronos are visible in fiber tracts near the habenula
and penetrate into the LHb. The dashed outline
represents the recorded region in LHbMC (44). (D) In
11 of 29 LHb neurons, photostimulation of Chronosexpressing PFC axons with 5 3 10-ms light pulses
elicited reproducible excitatory postsynaptic currents (EPSCs) at 10 Hz (gray: 10 sample traces;
black: average trace). (E) In response to clozapineN-oxide (CNO) application, spontaneous EPSC
frequency is increased. (F) In 7 of 20 LHb neurons,
CNO increased spontaneous EPSC (sEPSC) frequency (n = 20 cells from 12 rats; *p , 0.05; average
of 24 1-second sweeps collected at 0.2 Hz). PFC–
LHb connectivity estimates with hM3Dq (35%) were
consistent with those obtained with optogenetic
circuit mapping (38%). AAV, adeno-associated
virus; DAPI, 40 ,6-diamidino-2-phenylindole; MHb,
medial habenula; mPFC, medial PFC.
circuit in the generation of normal social behaviors, we performed an optogenetic experiment with stimulation paradigm
designed to activate PFC terminals in the LHb while reducing
the probability of off-target activation (see Methods and
Materials). Optogenetic activation of PFC-to-LHb connections
in rats with Chronos expression led to a significant decrease
in sociability index compared with the GFP controls
(Figure 4C, D). Two-way ANOVA showed a significant
group 3 light epoch interaction (n = 5–8, F1,11 = 13.66,
p = .0035). No significant differences were found in locomotion
and exploration during the habituation and locomotor activity
during the social test (Supplemental Figure S4F–I). To test
whether the decrease in sociability induced by PFC-to-LHb
stimulation is due to an increase in real-time aversion, we
tested these rats in a real-time place preference paradigm. In
the real-time place preference task, the test rat is allowed to
=
Figure 2. Combinatorial bidirectional modulation of prefrontal cortex (PFC) circuit components and social behavior. (A) Experimental strategy for the
expression of hM3Dq and KORD in the pyramidal neurons in the rat PFC. (B) Representative confocal image of PFC neurons expressing both hM3Dq and
KORD in the PFC (arrowhead). Scale bar = 20 mm. (C) Sociability index in response to bidirectional modulation of PFC activity (clozapine-N-oxide [CNO] 1
salvinorin B [SALB]). Sociability index = 100 3 (time in social interaction/total interaction time) 2 50 (n = 7; *p , .05; one-way analysis of variance followed by
Fisher’s post hoc comparison). (D) Outline of magnetic resonance imaging (MRI) experimental design involving CNO and SALB injections (inj.) during MRI.
Representative images from planes show change in normalized perfusion across the rostrocaudal extent of the brain. (E) Percentage change in normalized
perfusion in response to bidirectional modulation of PFC activity (dark bars: CNO; gray bars: CNO 1 SALB; n = 10; *p , .05; one-way analysis of variance
followed by post hoc comparisons; p values for all regions of interest are listed in Supplemental Table S1). (F) Schematic representation of the functional MRI
(fMRI) signature and social behavior in response to bidirectional modulation of PFC activity. Data represent mean 6 SEM. AAV, adeno-associated virus; DAPI,
40 ,6-diamidino-2-phenylindole; Hb, habenula; mTh, medial thalamus; Veh, vehicle.
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
613
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
Figure 4. A corticohabenular circuit regulates socially directed behavior. (A) Experimental strategy for expression of the excitatory designer receptor
exclusively activated by designer drugs receptor hM3Dq in the lateral habenula (LHb) with representative images showing surgical placements. (B) Representative heat maps (10-minute social test) and sociability index measured in response to vehicle (Veh) and clozapine-N-oxide (CNO) in rats expressing hM3Dq
or green fluorescent protein (GFP) in the LHb (n = 5–9; *p = .016; CNO vs. vehicle; two-way analysis of variance followed by Fisher’s post hoc comparison). A
within-subject parallel design with test–retest controls was applied. Bar graphs represent group averages, while gray lines represent individual values. (C)
Experimental strategy for optogenetic activation of prefrontal cortex (PFC)–LHb circuit, with a representative image showing placement of the fiber-optic ferrule.
(D) Representative heat maps (10-minute social test) and sociability index measured in response to optogenetic activation of PFC terminals in the LHb
expressing either Chronos or GFP (n = 5–8; ***p , .0001; ON vs. OFF; two-way analysis of variance followed by Fisher’s post hoc comparison). A within-subject
crossover design was applied. Bar graphs represent group averages, while gray lines represent individual values. (E) Representative heat maps (20-minute realtime place preference task) and duration spent in chambers paired with and without optogenetic stimulation of PFC–LHb projections in rats expressing either
Chronos or GFP (n = 4–9). (F) Experimental strategy for multiplexed activation of PFC and inhibition of LHb circuit with a representative image showing KORDi
expression in the LHb. A within-subject crossover design was applied. Sociability index was measured in response to activation of hM3Dq in the PFC (CNO) or
activation of hM3Dq in the PFC and inhibition with KORDi in the LHb (CNO 1 salvinorin B [SALB]) (n = 7–8; **p = .04; CNO vs. CNO 1 SALB in the designer
receptor exclusively activated by designer drugs–expressing rats; two-way analysis of variance followed by Fisher’s post hoc comparison). Data represent
mean 6 SEM. Sociability index = 100 3 (time in social interaction / total interaction time) 2 50. AAV, adeno-associated virus; MHb, medial habenula.
614
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
explore two different environmental contexts (with different
cues on the walls and flooring), with one of the contexts
coupled to PFC–LHb stimulation. Rats expressing Chronos or
GFP showed no preference for or aversion to the stimulationpaired chamber (Figure 4E). Two-way ANOVA showed no
significant group X treatment interaction (F1,11 = 0.0002827,
p = .9869).
Finally, we addressed whether LHb activation is necessary
for the effects of PFC activation on social behavior. The
excitatory hM3Dq was expressed in the PFC, and the inhibitory
KORDi was expressed in the LHb (Figure 4F). Control animals
had GFP expressed in both the PFC and LHb. Social behavior
was measured in the three-chamber task after administration
of either CNO (1 mg/kg) or both CNO and SALB (17.5 mg/kg).
Data show that administration of SALB after CNO led to
normalized social behaviors in the DREADD-expressing mice.
Two-way ANOVA showed a significant group 3 treatment
interaction (Figure 4F) (n = 6–7, F1,11 = 8.128, p = .0158). Taken
together, these data demonstrate a key role for PFC–LHb
pathways in social behavior and suggest that PFC–LHb
pathways may report social context leading to the expression of normal or pathological social behaviors.
DISCUSSION
Perturbed circuit homeostasis in the PFC has been proposed
as a common neural substrate of social impairments seen in
ASD and schizophrenia. Rodent optogenetic studies have
shown that the PFC governs social behavior via top-down
control over the activity dynamics of the midbrain reward
system (31,32). However, the neural pathways that connect the
PFC to the social reward pathways have not yet been delineated. Here we screened for PFC circuit partners involved in
the control of social behavior and found that increasing PFC
activity led to activity changes in the habenula and other
regions associated with emotional behaviors. Bidirectional
modulation of PFC activity helped us to identify a dynamic
association among PFC activity, habenula activity, and social
behavior. Using cellular tracing, DREADDs, and optogenetic
circuit mapping, we found functional connections from the
PFC to the LHb. The LHb is a key relay center in the midbrain
with strategic projections controlling monoaminergic nuclei
such as the ventral tegmental area (VTA) dopaminergic (DA)
neurons (33,34). Given this, we hypothesized that neural activity in the LHb modulates social behavior. Indeed, a precise
localized activation of the LHb led to a specific decrease in
social preference without any effects on locomotor activity or
sucrose preference. Furthermore, optogenetic activation of
PFC inputs to the LHb led to a specific decrease in sociability
with no effect on real-time aversion. Finally, our multiplexed
DREADD experiment in which LHb activity was decreased
during PFC activation shows that habenula activity is necessary for the social deficits induced by PFC activation. Collectively, our data demonstrate that the corticohabenular pathway
represents a social circuit that can be dissociated from the role
of the LHb in real-time aversion.
The LHb has been shown to be involved in conveying
reward-related cognitive information to dopaminergic circuits
in the VTA (24,35). The ability to engage in socially directed
behaviors is gated by the reward pathways in the VTA (36).
Increased activity in the PFC–LHb pathways might disrupt the
activity dynamics in the VTADA / NA pathway, thereby
interfering with the neural processing of social reward (10,37).
The LHb is thought to exert tonic inhibitory control of VTADA
neurons (38), and reward-predicting cues are thought to
decrease LHb activity and attenuate this tonic inhibition to
signal information about reward to VTADA neurons (11,12).
Based on these observations, we can speculate that PFC–
LHb pathway activation reduces the attribution of positive
salience to the social stimulus such that the rat does not
show a preference to socialize, resulting in the observed
specific social behavioral deficit with no effect on real-time
aversion or locomotor activity. LHb–VTA connectivity comprises a complex set of direct, indirect, and reciprocal connections to distinct subpopulations of VTA neurons (25,39),
and further studies are needed to delineate the complex
circuits that may mediate the influence of the LHb on reward
circuits controlling social behavior. Possible alternative
pathways for a prefrontal control over midbrain circuitry
include parallel circuits from the PFC–bed nucleus of the stria
terminalis, PFC–septum, or PFC–lateral hypothalamus, all of
which could potentially gate contextual information to VTADA
neurons (40,41).
Our DREADD–fMRI methodology combines the quantitative
power of perfusion-based fMRI with behavioral circuit analysis
to identify the activation pattern associated with social
dysfunction. Our approach is a significant addition to existing
approaches combining DREADDs with positron emission
tomography and [18F]fluoro-2-deoxyglucose (17,42) for metabolic mapping. Our fMRI data show that activity in the PFC is
spread across the infralimbic and prelimbic regions, and
therefore chemogenetic activation of the PFC likely reflects
broad activation of the region. Of note, neural dynamics in both
DREADD–fMRI and metabolic mapping methodologies is
measured under anesthesia, which may attenuate the amplitude of the fMRI signal (43), resulting in an underestimation of
circuit activation that occurs in awake-behaving animals.
Nevertheless, the activity changes observed in the PFC and
habenula were reproducible across multiple cohorts.
Our approach of studying circuit dysfunction in neurodevelopmental disorders seeks to identify alterations in a
defined set of neurons in interconnected brain regions that can
be linked to behavioral abnormalities seen in these disorders.
This whole-brain approach led us to uncover the role of the
PFC–LHb pathway in governing social behavior with mechanistic integration into the socially relevant VTA DA circuitry.
Notably, such an approach recognizes that specific cell groups
in multiple brain regions such as the PFC, LHb, VTA, and NA
act in concert to generate social behavioral states. By selectively disrupting activity in the PFC or LHb, our results suggest
that the generation and maintenance of a new circuit activity
set point in a complex disease state may cause behavioral
dysfunction and indicates a potential for correction of PFC
excitation–inhibition coordination as a therapeutic avenue for
ASDs. Although it is difficult to untangle the complex neural
relationships that contribute to disease phenotypes, our results
lend credence to the hypothesis that reciprocal connections
between the PFC and several key subcortical regions comprise
a circuit framework for altered cortical and subcortical function
to modify social behavior.
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
615
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
ACKNOWLEDGMENTS AND DISCLOSURES
This work is partially supported by the Gatsby Charitable Trust (AG) and by
the postdoctoral program at F. Hoffman–La Roche (MB, TJS).
We thank Bryan L. Roth for the DREADD constructs. We thank Claire
Coulon-Bainier for help with surgical procedures and thank Stephanie
Schoeppenthau, Sebastian Debilly, and Ciril Marius Waelti for technical
assistance with the fMRI experiments. We thank Roger Redondo for help
with the in vivo optogenetic experiments. We also thank Laurence Ozmen,
Marie-Therese Miss, Arel Su, and Urs Humbel for their help with experimental setup. We thank Tom Otis, Michael Saxe, Barbara Biemans, Will
Spooren, and members of the Ghosh Lab for valuable discussions.
MB, TJS, FK, AB, MvK, BK, and AG were employed by F. Hoffmann–La
Roche during this study.
15.
16.
17.
18.
19.
ARTICLE INFORMATION
From Neuroscience Discovery, Roche Pharma Research & Early Development, Roche Innovation Center Basel, F. Hoffmann–La Roche Ltd., Basel,
Switzerland.
AG is currently affiliated with Research and Early Development, Biogen,
Cambridge, Massachusetts.
Address correspondence to Anirvan Ghosh, Ph.D., Research and Early
Development, Biogen, 225 Binney Street, Cambridge, MA 02142; E-mail:
anirvan.ghosh@biogen.com.
Received Mar 9, 2017; revised Oct 11, 2017; accepted Oct 31, 2017.
Supplementary material cited in this article is available online at https://
doi.org/10.1016/j.biopsych.2017.10.032.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
616
Curley AA, Arion D, Volk DW, Asafu-Adjei JK, Sampson AR, Fish KN,
et al. (2011): Cortical deficits of glutamic acid decarboxylase 67
expression in schizophrenia: Clinical, protein, and cell type-specific
features. Am J Psychiatry 168:921–929.
Dichter GS, Felder JN, Bodfish JW (2009): Autism is characterized by
dorsal anterior cingulate hyperactivation during social target detection.
Soc Cogn Affect Neurosci 4:215–226.
Lewine JD, Andrews R, Chez M, Patil A-A, Devinsky O, Smith M, et al.
(1999): Magnetoencephalographic patterns of epileptiform activity in
children with regressive autism spectrum disorders. Pediatrics
104:405–418.
Wheless JW, Simos PG, Butler IJ (2002): Language dysfunction in
epileptic conditions. Sem Pediatr Neurol 9:218–228.
Gillberg C, Billstedt E (2000): Autism and Asperger syndrome: Coexistence with other clinical disorders. Acta Psychiatr Scand 102:321–330.
Voineagu I, Wang X, Johnston P, Lowe JK, Tian Y, Horvath S, et al.
(2011): Transcriptomic analysis of autistic brain reveals convergent
molecular pathology. Nature 474:380–384.
Rubenstein JLR, Merzenich MM (2003): Model of autism: Increased
ratio of excitation/inhibition in key neural systems. Genes Brain Behav
2:255–267.
Nelson SB, Valakh V (2015): Excitatory/inhibitory balance and circuit
homeostasis in autism spectrum disorders. Neuron 87:684–698.
Yizhar O, Fenno LE, Prigge M, Schneider F, Davidson TJ, O’Shea DJ,
et al. (2011): Neocortical excitation/inhibition balance in information
processing and social dysfunction. Nature 477:171–178.
Lammel S, Lim BK, Ran C, Huang KW, Betley MJ, Tye KM, et al.
(2012): Input-specific control of reward and aversion in the ventral
tegmental area. Nature 491:212–217.
Matsumoto M, Hikosaka O (2007): Lateral habenula as a source of
negative reward signals in dopamine neurons. Nature 447:1111–1115.
Matsumoto M, Hikosaka O (2009): Representation of negative motivational value in the primate lateral habenula. Nat Neurosci 12:77–84.
Stephenson-Jones M, Yu K, Ahrens S, Tucciarone JM, van
Huijstee AN, Mejia LA, et al. (2016): A basal ganglia circuit for evaluating action outcomes. Nature 539:289–293.
Baker PM, Jhou T, Li B, Matsumoto M, Mizumori SJY, StephensonJones M, et al. (2016): The lateral habenula circuitry: Reward processing and cognitive control. J Neurosci 36:11482–11488.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
Atasoy D, Betley JN, Su HH, Sternson SM (2012): Deconstruction of a
neural circuit for hunger. Nature 488:172–177.
Armbruster BN, Li X, Pausch MH, Herlitze S, Roth BL (2007): Evolving
the lock to fit the key to create a family of G protein-coupled receptors
potently activated by an inert ligand. Proc Natl Acad Sci U S A
104:5163–5168.
Urban DJ, Zhu H, Marcinkiewcz CA, Michaelides M, Oshibuchi H,
Rhea D, et al. (2016): Elucidation of the behavioral program and
neuronal network encoded by dorsal raphe serotonergic neurons.
Neuropsychopharmacology 41:1404–1415.
Vardy E, Robinson JE, Li C, Olsen RH, DiBerto JF, Giguere PM, et al.
(2015): A new DREADD facilitates the multiplexed chemogenetic
interrogation of behavior. Neuron 86:936–946.
Yang M, Silverman JL, Crawley JN (2001): Automated threechambered social approach task for mice. Curr Protoc Neurosci
8:8.26.
Vertes RP (2004): Differential projections of the infralimbic and prelimbic cortex in the rat. Synapse 51:32–58.
Gabbott PL, Warner TA, Jays PR, Salway P, Busby SJ (2005): Prefrontal cortex in the rat: Projections to subcortical autonomic, motor,
and limbic centers. J Comp Neurol 492:145–177.
Williams DS, Detre JA, Leigh JS, Koretsky AP (1992): Magnetic resonance imaging of perfusion using spin inversion of arterial water. Proc
Natl Acad Sci U S A 89:212–216.
Dolen G, Darvishzadeh A, Huang KW, Malenka RC (2013): Social
reward requires coordinated activity of nucleus accumbens oxytocin
and serotonin. Nature 501:179–184.
Stamatakis AM, Stuber GD (2012): Activation of lateral habenula inputs
to the ventral midbrain promotes behavioral avoidance. Nat Neurosci
15:1105–1107.
Stamatakis AM, Jennings JH, Ung RL, Blair GA, Weinberg RJ,
Neve RL, et al. (2013): A unique population of ventral tegmental area
neurons inhibits the lateral habenula to promote reward. Neuron
80:1039–1053.
Nieh EH, Matthews GA, Allsop SA, Presbrey KN, Leppla CA,
Wichmann R, et al. (2015): Decoding neural circuits that control
compulsive sucrose seeking. Cell 160:528–541.
Greatrex R, Phillipson O (1982): Demonstration of synaptic input
from prefrontal cortex to the habenula in the rat. Brain Res 238:
192–197.
Kim U, Lee T (2012): Topography of descending projections from
anterior insular and medial prefrontal regions to the lateral habenula of
the epithalamus in the rat. Eur J Neurosci 35:1253–1269.
Warden MR, Selimbeyoglu A, Mirzabekov JJ, Lo M, Thompson KR,
Kim S-Y, et al. (2012): A prefrontal cortex–brainstem neuronal projection that controls response to behavioural challenge. Nature
492:428–432.
Stachniak TJ, Ghosh A, Sternson SM (2014): Chemogenetic synaptic
silencing of neural circuits localizes a hypothalamus / midbrain
pathway for feeding behavior. Neuron 82:797–808.
Ferenczi EA, Zalocusky KA, Liston C, Grosenick L, Warden MR,
Amatya D, et al. (2016): Prefrontal cortical regulation of brainwide
circuit dynamics and reward-related behavior. Science 351:aac9698.
Yizhar O (2012): Optogenetic insights into social behavior function.
Biol Psychiatry 71:1075–1080.
Herkenham M, Nauta WJ (1977): Afferent connections of the habenular
nuclei in the rat: A horseradish peroxidase study, with a note on the
fiber-of-passage problem. J Comp Neurol 173:123–146.
Herkenham M, Nauta WJ (1979): Efferent connections of the habenular
nuclei in the rat. J Comp Neurol 187:19–47.
Sutherland RJ (1982): The dorsal diencephalic conduction system: A
review of the anatomy and functions of the habenular complex. Neurosci Biobehav Rev 6:1–13.
Gunaydin LA, Grosenick L, Finkelstein JC, Kauvar IV, Fenno LE,
Adhikari A, et al. (2014): Natural neural projection dynamics underlying
social behavior. Cell 157:1535–1551.
Jhou TC, Fields HL, Baxter MG, Saper CB, Holland PC (2009): The
rostromedial tegmental nucleus (RMTg), a major GABAergic afferent to
midbrain dopamine neurons, selectively encodes aversive stimuli and
promotes behavioral inhibition. Neuron 61:786–800.
Biological
Psychiatry
PFC and Lateral Habenula in Social Behavior
38.
39.
40.
41.
Lecourtier L, Defrancesco A, Moghaddam B (2008): Differential
tonic influence of lateral habenula on prefrontal cortex and
nucleus accumbens dopamine release. Eur J Neurosci 27:
1755–1762.
van Zessen R, Phillips JL, Budygin EA, Stuber GD (2012): Activation of
VTA GABA neurons disrupts reward consumption. Neuron 73:
1184–1194.
Jennings JH, Sparta DR, Stamatakis AM, Ung RL, Pleil KE, Kash TL,
et al. (2013): Distinct extended amygdala circuits for divergent motivational states. Nature 496:224–228.
Luo AH, Tahsili-Fahadan P, Wise RA, Lupica CR, Aston-Jones G
(2011): Linking context with reward: A functional circuit from
42.
43.
44.
hippocampal CA3 to ventral tegmental area. Science 333:
353–357.
Michaelides M, Anderson SA, Ananth M, Smirnov D, Thanos PK,
Neumaier JF, et al. (2013): Whole-brain circuit dissection in freemoving animals reveals cell-specific mesocorticolimbic networks.
J Clin Invest 123:5342–5350.
Desai M, Kahn I, Knoblich U, Bernstein J, Atallah H, Yang A, et al.
(2011): Mapping brain networks in awake mice using combined optical
neural control and fMRI. J Neurophysiol 105:1393–1405.
Wagner F, Stroh T, Veh RW (2014): Correlating habenular subnuclei in
rat and mouse by using topographic, morphological, and cytochemical
criteria. J Comp Neurol 522:2650–2662.
Biological Psychiatry April 1, 2018; 83:607–617 www.sobp.org/journal
617