Article

Role of anterior cingulate cortex inputs to

periaqueductal gray for pain avoidance
Graphical abstract Authors
Jeong-Yun Lee, Taeyi You,
Choong-Hee Lee, Geun Ho Im,
Heewon Seo, Choong-Wan Woo,
Seong-Gi Kim

Correspondence
[email protected] (J.-Y.L.),
[email protected] (S.-G.K.)

In brief
Using optogenetic fMRI in mice, Lee et al.
reveal abnormal circuit changes of the
ACC in chronic pain and identified its
function to be involved in sensorimotor
integration rather than sensory
transmission in pain processing. In
particular, the ACC-PAG circuit is
vulnerable to chronic pain and induces
active defensive behavior to noxious
stimuli.

Highlights
d The ACC is involved in the maintenance of chronic pain
hypersensitivity

d ofMRI shows increased network strength of the ACC in the

chronic pain model

d The ACC contributes to sensorimotor integration in pain

processing

d The ACC-dl/lPAG circuit induces active defensive behaviors

against threat signals

Lee et al., 2022, Current Biology 32, 2834–2847

July 11, 2022 ª 2022 Elsevier Inc.
https://doi.org/10.1016/j.cub.2022.04.090 ll
 ll

Article
Role of anterior cingulate cortex inputs
to periaqueductal gray for pain avoidance
Jeong-Yun Lee,1,5,* Taeyi You,1,2,3,5 Choong-Hee Lee,1 Geun Ho Im,1 Heewon Seo,1,4 Choong-Wan Woo,1,2,3
and Seong-Gi Kim1,2,3,6,*
1Center for Neuroscience Imaging Research (CNIR), Institute for Basic Science (IBS), Suwon 16419, Republic of Korea
2Department of Biomedical Engineering, Sungkyunkwan University, Suwon 16419, Republic of Korea
3Department of Intelligent Precision Healthcare Convergence, Sungkyunkwan University, Suwon 16419, Republic of Korea
4Department of Chemistry and Biochemistry, Oberlin College, Oberlin, OH 44704, USA
5These authors contributed equally
6Lead contact

*Correspondence: [email protected] (J.-Y.L.), [email protected] (S.-G.K.)

https://doi.org/10.1016/j.cub.2022.04.090

SUMMARY

Although pain-related excessive fear is known to be a key factor in chronic pain disability, which involves the
anterior cingulate cortex (ACC), little is known about the downstream circuits of the ACC for fear avoidance in
pain processing. Using behavioral experiments and functional magnetic resonance imaging with optoge-
netics at 15.2 T, we demonstrate that the ACC is a part of the abnormal circuit changes in chronic pain
and its downstream circuits are closely related to modulating sensorimotor integration and generating active
movement rather than carrying sensory information. The projection from the ACC to the dorsolateral and
lateral parts of the periaqueductal gray (dl/lPAG) especially enhances both reflexive and active avoidance
behavior toward pain. Collectively, our results indicate that increased signals from the ACC to the dl/lPAG
might be critical for excessive fear avoidance in chronic pain disability.

INTRODUCTION functions to react to a threat5 and serves as the hub for descend-
ing pain control.6 The dorsolateral/lateral part of the PAG (dl/
Pain is a sensory and affective experience in response to actual lPAG) prompts active defensive responses, such as jumping
or potential tissue damage for survival in threatening situations. and running, whereas the ventrolateral part (vlPAG) produces
Traditionally, pain processing in the brain is known to have a sen- analgesia and passive defensive responses, such as freezing
sory-discriminative dimension, which is based on the sensory and immobility.7 However, research on the defensive behaviors
information of location, quality, and an affective-motivational toward pain has been mostly focused on the freezing response
dimension, which establishes the unpleasantness and aversive to footshock8; therefore, little is known about the active defen-
experiences of pain.1 More recently, a fear-avoidance model sive responses to pain.
focused on contextual fear conditioning has been proposed to Functional magnetic resonance imaging (fMRI) based on
explain individual differences in chronic pain disability.1,2 In blood-oxygenation-level-dependent (BOLD) signals is a nonin-
these models, the context of pain determines the level of fear vasive neuroimaging technique for measuring whole-brain
and the patient’s attitudes to ‘‘avoidance’’ or ‘‘confrontation.’’ activity. Since pain is a multidimensional experience with sen-
The persistence of pain-related fear can result in the generaliza- sory-discriminative, affective-motivational, and cognitive-evalu-
tion of avoidance responses even in non-painful conditions, ative aspects involving multiple brain regions,1 systems-level
leading to chronic pain disability. However, the neural mecha- exploration is essential.9 In human fMRI studies, correlative
nisms underlying the fear avoidance of pain are poorly methods, such as functional connectivity, are powerful tools to
understood. understand pain processing at the systems level,10 but they
Defensive behaviors against fear require the selection of show a limited ability to provide causal interpretation.11 In
appropriate actions for an active or passive response. In the de- rodents, fMRI combined with optogenetics (ofMRI), a neuro-
fense cascade model, fight or flight is an active defensive modulation technique that uses light-sensitive ion channels or
response to an imminent threat, enabling an organism to boost pumps controlled by specific wavelengths, provide a unique op-
energy supplies for movement.3 By contrast, freezing is a pas- portunity for investigating downstream circuits in the whole
sive defensive response to an uncertain threat in which the ‘‘fight brain.12 Therefore, we aimed to understand neuroplasticity in
or flight’’ status is put on hold and motor function is halted to chronic pain by applying ofMRI to an animal pain model.
detect a possible threat.3 In other words, if the choice between Here, we explored the function of the anterior cingulate cortex
immediate avoidance and attack is difficult to determine, an (ACC) and its downstream circuits involved in the active defen-
organism withholds action selection.4 In particular, the periaque- sive responses in the context of pain chronification. We focused
ductal gray (PAG) coordinates the integration of sensorimotor on the ACC because it is a hub for the pain-related

2834 Current Biology 32, 2834–2847, July 11, 2022 ª 2022 Elsevier Inc.
 ll
Article

Figure 1. Effects of the optogenetic silencing of the ACC on pain-related behavior

(A and B) Schematic design for the experiment and experimental schedule for VGAT-ChR2 mice.
(C–E) The effect of ACC silencing on nociception (C), capsaicin-induced acute pain (D), and CFA-induced chronic pain (E); VGAT-ChR2 n = 6–7 and
littermate n = 4.
(F) Experimental schedule and schematic design for AAV-CaMKII-eArch.
(G and H) The effect of ACC silencing on nociception (G) and CFA-induced chronic pain (H); CaMKII-eArch-eYFP n = 5–7 and CaMKII-eYFP n = 4. Cross denotes
significance levels compared with littermate or CaMKII-eYFP. (C paired t test, Da unpaired t test, Dc, E, G, and H two-way ANOVA.)

multidimensional experience.13 In rodent studies, chronic pain pyramidal neurons (Figures 1A and 1F). We used transgenic
changes the synaptic transmission in the ACC,14–16 and the mice expressing channelrhodopsin2 (ChR2) under the vesicular
increased neural activity of the ACC engages in descending GABA transporter promoter (VGAT) in GABAergic interneurons
pain facilitation and fear memory formation.17–20 The ACC is (VGAT-ChR2 mice) to depolarize interneurons with light
known to generate a teaching signal for the sources of danger (473 nm) (Figures 1A–1E) or an adeno-associated virus (AAV) ex-
and predict future dangers.21–24 Our results revealed that the pressing archaerhodopsin (Arch) under a calcium/calmodulin-
ACC and its downstream neural circuit are vulnerable to chronic dependent protein kinase II (CaMKII) promoter to hyperpolarize
pain. Using ofMRI at an ultrahigh field of 15.2 T, we determined excitatory pyramidal neurons with light (532 nm) (Figures 1F–
that the function of the ACC is closely related to sensorimotor 1H). The thermal pain threshold was estimated by measuring
integration and movement generation. In particular, the ACC the reflexive paw withdrawal latency under infrared stimulation.
has numerous inputs into the dl/lPAG, and the ACC-dl/lPAG To investigate the role of the ACC in the transition from nocicep-
circuit modulates both reflexive and active avoidance responses tion to chronic persistent pain, we injected capsaicin (CAP) for
to noxious stimuli. acute pain and complete Freund’s adjuvant (CFA) for chronic
pain into the hind paw. In the animal pain model, it is known
RESULTS that persistent pain for 3 weeks induces changes in brain
circuitry, leading to anxiety and decreased motivation.25,26
The ACC is a part of the neural circuits associated with Thus, we injected CFA twice to prolong the CFA-induced persis-
pain-induced neural plasticity tent pain for more than 3 weeks.
To examine the role of the ACC in pain processing, we observed The optogenetic activation of GABAergic interneurons (VGAT-
pain-related behaviors during ACC silencing (Figure 1). The ChR2, 3–5 mW, 20 Hz, and 10 ms) in the cortex can depolarize
silencing of the ACC pyramidal neurons was induced by acti- interneurons and consequently hyperpolarize excitatory pyrami-
vating inhibitory interneurons or suppressing excitatory dal neurons.27,28 The optogenetic activation of ACCVGAT neurons

Current Biology 32, 2834–2847, July 11, 2022 2835

 ll
Article

Figure 2. Brain-wide BOLD fMRI during optogenetic silencing of ACC in naive and CFA-induced chronic pain model
(A) The effect of ketamine (KET) and dexmedetomidine (DEX) on the thermal pain threshold (a), capsaicin (CAP)-induced spontaneous pain (b), and pain
hypersensitivity (c); vehicle n = 10, KET n = 5–6, and DEX n = 7.
(B) Experimental design for ofMRI; naive n = 7 and CFA n = 7.
(C) Group activation maps of naive and CFA mice during ACC inhibition; Z > 2.8, cluster corrected at p = 0.05. To visualize PAG activity, maps were generated with
a threshold of Z > 1.96, cluster corrected at p = 0.05, and presented as insets.
(D) Time courses of BOLD fMRI in representative ROIs.
(E) Connectivity matrix of all responsive ipsilateral ROIs. 3 denotes significance of p = 0.05 family-wise error rate (FWER) corrected for multiple comparisons.
(legend continued on next page)

2836 Current Biology 32, 2834–2847, July 11, 2022

 ll
Article

did not affect the thermal pain threshold in naive mice (Figure 1C). and interconnected network strength of downstream regions
Similarly, the optogenetic inhibition of ACCCaMKII neurons such as the lateral pathway for a sensory-discriminative pain (pri-
caused no change in the thermal pain threshold (Figure 1G). To mary somatosensory hindlimb cortex [S1HL]), the medial
examine whether ACC suppression influences spontaneous pathway for an affective-motivational pain (the ACC and medial
pain and pain hypersensitivity, ACCVGAT neurons were immedi- dorsal thalamus [MD]), the descending pathway (PAG), and the
ately activated for 10 min after CAP injection. The silencing motor-related system (primary motor hindlimb cortex [M1hl],
of the ACC failed to block the CAP-induced spontaneous secondary motor cortex [M2], and ventral anterolateral thalamus
pain behavior and the induction of pain hypersensitivity [VAL]) (Figures 2D, 2F, and 2G).
(Figures 1Da and 1Db). However, the activation of ACCVGAT neu-
rons reversed the maintenance of CAP- or CFA-induced pain The ACC is involved in sensorimotor integration in pain
hypersensitivity (Figures 1Dc and 1E). Similarly, the optogenetic processing
inhibition of ACCCaMKII neurons blocked pain hypersensitivity in To further dissect the downstream circuits of the ACC for pain
the CFA-induced chronic pain model (Figure 1H). Thus, the processing, we combined the optogenetic silencing of the ACC
ACC is engaged in the maintenance of pain hypersensitivity with electrical stimuli (Figure 3). Electrical stimulation (4 Hz) is
rather than nociception. transmitted as a noxious signal (Figure S3). Noxious electrical
stimulation of the whisker pad (WP) showed a stable positive
The ACC and its downstream circuit are involved in BOLD response in the pain pathway and sensorimotor-related
chronic pain brain regions (Figures 3B–3F). The modularity-based community
To explore the spontaneous neural activity of the ACC and its detection of the functional connectivity patterns identified two
downstream circuits, we performed optogenetic silencing fMRI modules: WP-related and polymodal ROIs (Figure 3D). WP-
under light anesthesia (Figure 2).29 Since anesthesia can sup- related ROIs are areas directly related to the whisker somatosen-
press pain circuits, we investigated the effect of anesthesia on sory network, as described previously.30,33 When noxious
pain behaviors at a subanesthetic dose of ketamine (KET) and stimulation was combined with optogenetic ACC silencing
dexmedetomidine (DEX) (Figure 2A), which are suitable for (WPOG, WP and optogenetic stimulation), BOLD amplitudes
mouse fMRI.30 Unlike KET, DEX suppressed nociception (Fig- were reduced by the ACC-modulated downstream circuits
ure 2Aa) but did not block pain hypersensitivity (Figure 2Ac). (Figures 3B–3F). The inter-region connectivity matrix was group-
Therefore, we chose DEX to investigate the role of the ACC in ed into three functional modules (Figure 3D): (1) WP-dominant
pain hypersensitivity. ROIs (WP-d), which did not have a significant modulation from
Both activations of ACCVGAT neurons and inhibition of ACC silencing, primarily make up somatosensory regions such
ACCCaMKII neurons caused negative BOLD responses (Figures 2 as primary somatosensory barrel field (S1BF) and ventral poste-
and S1). Since silencing the excitatory pyramidal neurons by the rior thalamus (VP); (2) ACC-dominant ROIs (ACC-d) with no WP
stimulation of VGAT interneurons is more effective than that by response have a significant negative BOLD response, creating
the optogenetic inhibition of CaMKII neurons,27 we chose to a negative correlation; and, (3) importantly, optogenetic ACC
focus on the ofMRI of VGAT-ChR2 mice. To investigate the silencing significantly reduced noxious stimulation-induced
circuit changes caused by chronic pain, we compared the ofMRI BOLD responses in ACC-modulated areas (ACC-m), including
responses in naive mice and the CFA-induced chronic pain the descending pain pathway (dl/lPAG and vlPAG) and motor-
model (R3 weeks after CFA injection). The activation of related pathways (M2, M1, and superior colliculus motor region
ACCVGAT neurons caused larger negative BOLD responses in [SCm]) (Figures 3C–3F). These data indicate that the down-
the ACC and several downstream regions of the CFA-induced stream targets of the ACC are closely related to sensorimotor
chronic pain model compared with the naive mice (Figures 2C– integration and movement generation rather than sensory
2G and S2). Moreover, the functional connectivity strength be- discrimination in pain processing.
tween all the responsive regions of interest (ROIs) was enhanced Next, to examine the efferent projections of the ACC, we in-
in the CFA-induced chronic pain model (Figure 2E). Despite the jected an anterograde viral tracer (AAV5) expressing eYFP under
controversy regarding the origin of negative BOLD responses,31 the CaMKII promoter into the ACC (Figure 4A). Anterograde label-
the appearance of these responses in optogenetic silencing ing showed dense projection from the ACC to the neural circuits
fMRI reflects the deactivation of excitatory pyramidal neurons for defensive behaviors such as PAG and SCm,8,34,35 whereas
at the target region.29 The amplitude of the fMRI signal reduction somatosensory circuits such as S1, S2, and VP received less
is closely dependent on a reduction in the spontaneous neural input from the ACC (Figures 4A and S4A). Compared with anter-
activity from baseline.32 Since ACC silencing suppresses the ograde monosynaptic tracing, which represents only direct pre-
excitatory output to downstream pathways, the magnitude of synaptic input, ofMRI can reveal brain activity by multisynaptic
the downregulated neuronal activity reflects the degree of inter- pathways. The optogenetic activation of ACCCaMKII neurons
regional communication under basal conditions. Thus, persis- induced a positive BOLD response throughout the brain regions,
tent pain increases the spontaneous neural activity of the ACC but the magnitude of the BOLD signal was dependent on the

(F and G) Absolute area under the curve (AUC) of the BOLD response in each ROI; mean (F) and individual data (G).
Ordering of ROIs in (F) denotes ordering in (E). ROI abbreviations: see STAR Methods.
Cross denotes significance levels in comparison with baseline or before CAP. (Aa and Ac: two-way ANOVA, Ab: one-way ANOVA, E: unpaired t test, F and G:
Mann-Whitney U test.)
See also Figures S1 and S2.

Current Biology 32, 2834–2847, July 11, 2022 2837

 ll
Article

Figure 3. Effects of ACC silencing on the BOLD responses induced by the noxious electrical stimulation of whisker pad
(A) Experimental design for silencing ofMRI combined with electrical stimulation.
(B) Group activity maps from the noxious stimulation of whisker pad (WP) without and with optogenetic ACC silencing (WPOG); Z > 2.8, cluster corrected at
p = 0.05.
(C) Time courses of BOLD fMRI in representative ROIs.
(D) Connectivity matrix of all responsive ROIs. 3 denotes p = 0.05 FWER; two modules in WP (WP-related, polymodal,) and three modules in WPOG were
classified (WP-d, WP dominant, ACC-m, ACC modulated, and ACC-d, ACC dominant).
(E and F) AUC of the WP-induced BOLD response without and with optogenetic silencing; ROI order in (E) denotes the order in (D). Mean (E) and individual data (F);
n = 7–8. (D: paired t test; E and F: Wilcoxon test.)
See also Figure S3.

strength of the synaptic activity.36 The BOLD signal decreased ACCVGAT neurons showed that a stronger inhibition was found
with increasing distance from the stimulated ACC region (rostral in the ACC and dl/lPAG compared with S1 or VP (Figure 2).
to caudal), and a significant difference was observed between the Collectively, our results suggest that the function of the ACC is
ipsilateral and contralateral thalamus regions (Figure 4Ca). At a closely related to the dl/lPAG regions.
high threshold, the optogenetic activation of ACCCaMKII neurons In addition to the ACC, the prelimbic cortex (PrL) also belongs
showed BOLD activation patterns in the brain regions receiving to the medial frontal cortex (MFC) in rodents. In the chronic pain
direct input from the ACC (Figure 4B). Similar to the viral tracing model, decreased input from the PrL to the vlPAG induces pain
of ACCCaMKII neurons, the optogenetic activation of ACCCaMKII hypersensitivity.37–39 Thus, we determined the innervation site of
neurons induced a higher BOLD response in the dl/lPAG than the PAG from the ACC and the PrL (Figures 4D, S4B, and S4C).
in S1 (Figure 4Cb). Similarly, the BOLD response of the dl/lPAG, The vlPAG received dense input from the PrL rather than the
although not significant, was also higher than those of the vlPAG ACC, whereas the dl/lPAG received dominant inputs from the
and dmPAG (Figure 4Cb). In accordance, the activation of ACC.

2838 Current Biology 32, 2834–2847, July 11, 2022

 ll
Article

Figure 4. Efferent projections of the ACC and functional connections under the optogenetic activation of the ACC
(A) Whole-brain tracing maps of the axonal projections from the ACC neurons with distances relative to bregma; subcortical sections indicated by dashed boxes
were expanded for better visualization.
(B) Mean BOLD activation maps induced by the optogenetic activation of the ACCCaMKII neurons with distances relative to bregma; Z > 3.29, cluster corrected at
p = 0.05. The dashed boxes were used for expansion in (b).
(C) AUC of the BOLD response in ROIs during the activation of ACCCaMKII neurons (a), and the comparison of fMRI responses in ipsilateral ROIs (b); n = 6.
(D) Anterograde tracing from the PrL or ACC into subregions of the PAG. (Ca Mann-Whitney U test and Cb Kruskal-Wallis test.)
See also Figure S4.

The ACC projection to the PAG contributes to pain ACC-PAG circuit decreased the thermal pain threshold (Fig-
behavior ure 5B), whereas the optogenetic inhibition increased the
To assess the function of the ACC-PAG circuit, we injected an thermal pain threshold in naive mice and suppressed CFA-
anterograde virus containing the CaMKII promoter with ChR2 induced pain hypersensitivity (Figure 5D). Additionally, KET
or Arch into the ACC and modulated the axon terminal in the (NMDA receptor antagonist) partially blocked the pain hyper-
dl/lPAG with light (Figure 5). The optogenetic activation of the sensitivity in the hind paw contralateral to the optogenetically

Current Biology 32, 2834–2847, July 11, 2022 2839

 ll
Article
Figure 5. Effects of the optogenetic modu-
lation of the ACC-PAG inputs on pain-
related behavior
(A) Circuit-specific modulation scheme and histo-
logical confirmation.
(B) The effect of the optogenetic activation of ACC-
PAG on the thermal pain threshold at 3–4 weeks
(3W) or 4 to 5 weeks (4W) after virus injection;
CaMKII-ChR2-eYFP n = 7 and CaMKII-eYFP n = 6.
(C) The effect of ketamine on pain hypersensitivity
by the activation of ACC-PAG; contralateral/ipsi-
lateral paw in relation to the transfected pathway.
Individual data comparing before (baseline) and af-
ter ketamine injection (KET) (b); n = 7.
(D) The effect of the optogenetic inhibition of ACC-
PAG on the thermal pain threshold (a) and the CFA-
induced chronic pain (b); CaMKII-eArch-eYFP n =
5–7 and CaMKII-eYFP n = 4.
(E) The effect of bidirectional modulation of ACC-
PAG on the thermal pain threshold. Thermal la-
tency was measured 4 or 8 times in 7 mice at
each condition and plotted together.
Cross denotes significance levels in comparison
with CaMKII-eYFP or ipsilateral paw. (B–D: two-
way ANOVA and E: one-way ANOVA.)
See also Figure S5.

The ACC projection to the PAG

contributes to fear response
We next investigated the role of the ACC-
PAG in innate fear response by using fox
urine (Figures 6C, 6D, and S5B). The op-
togenetic activation of the ACC-PAG cir-
cuit increased the speed of movement,
similar to the results of the open-field
test (Figure 6Da), whereas the optoge-
netic inhibition of the ACC-PAG circuit
slightly decreased the speed of move-
ment (Figure S5Bc). After exposure to
predator odors, mice avoided the places
with odors (zone 1). After 2 min, the
continuous exposure to predator odor
alleviated the avoidance behavior toward
zone 1 (Figures 6Db and S5B). Zone 2
activated ACC-PAG circuit (Figure 5C), indicating glutamater- (away from odors) preference was still high in the control and
gic synaptic neurotransmission. ACC-PAG inhibition groups, but the ACC-PAG activated
Pain behavior can be enhanced by anxiety, which is (CaMKII-ChR2-eYFP) mice preferred to be in zone 1 (Fig-
closely related to the ACC.13 Thus, we examined whether ure S5Bd). Thus, in an uncertain threat, the increased input
the ACC-PAG circuit is associated with the anxiety level from the ACC to PAG induces active movement related to
by using an open-field test (Figures 6A, 6B, and S5A). As anx- exploratory behavior toward the threat environment.
iety levels increased, mice showed reduced exploratory Finally, we examined the role of the ACC-PAG circuit in
behavior and locomotion and avoided the center zone. During learned fear response to noxious stimuli with an active place
the optogenetic activation of the ACC-PAG circuit, the speed avoidance task (APAT) (Figures 6E–6H). In the APAT, mice
of movement increased without abnormal motor behavior receive electrical shocks whenever they enter a certain section
(Video S1), but the time spent in the center zone did not (60 shock zone) on a counterclockwise rotating platform.
change (Figure 6B). On the other hand, the optogenetic inhibi- After learning the shock zone location via spatial cues, mice
tion of the ACC-PAG circuit decreased the duration in the freely and actively avoid the shock zone. Before the optoge-
center zone (Figure S5A). Therefore, the ACC-PAG circuit is netic experiment (day 1), mice were trained to avoid a 60
unlikely to cause pain hypersensitivity by increasing the anxi- shock zone, and the number of shocks was counted to ensure
ety levels. that the mice learned the shock zone (Figure 6G). On day 2, the

2840 Current Biology 32, 2834–2847, July 11, 2022

 ll
Article

Figure 6. Effects of the optogenetic activation of ACC-PAG inputs on anxiety, innate fear, and active avoidance response
(A) Experimental design for the open-field test.
(B) The effect of the optogenetic activation of ACC-PAG on the speed of movement (a) and the duration in center/border zone (b); n = 9.
(C) Experimental design for the innate fear response induced by fox urine.
(D) The effect of the optogenetic activation of ACC-PAG on the speed of movement (a) and the durations spent in zone 1 and zone 2 (b); n = 9.
(E) Experimental design (a) and schedule (b) for active place avoidance task.
(F) Representative heat map of time spent.
(G) Number of entries into the shock zone on day 1.
(H) The effect of the optogenetic activation of ACC-PAG on the number of entries into the shock zone (a) and the angle from the shock zone (b and c) on day 2; n = 9–10.
Ba, Da, and Hb: repeated measures one-way ANOVA and B, D, and Hc: paired t test.)
See also Figures S5 and S6 and Videos S1 and S2.

number of shocks was close to zero and independent of the of their position was measured from the shock zone
optogenetic activation of the ACC-PAG circuit (Figure 6Ha). (Figures 6Hb and 6Hc). The optogenetic activation of the
Interestingly, in one mouse that received multiple shocks on ACC-PAG circuit increased the angle from the shock zone
day 1, the optogenetic activation of the ACC-PAG circuit (Figures 6F, 6Hb, and 6Hc). These data indicate that increased
decreased the number of shocks (Figure S6). To avoid shock, inputs from the ACC to PAG contribute to active avoidance
mice showed a clockwise running behavior, so the angle behavior to noxious stimuli.

Current Biology 32, 2834–2847, July 11, 2022 2841

 ll
Article

DISCUSSION signals at the spinal cord.6 Since the activation of the ACC mod-
ulates the nociceptive input at the spinal cord,17,20,55 the ACC-dl/
In this study, we demonstrate the function of the ACC for active lPAG circuit may engage in descending pain facilitation. Consis-
defensive response in pain processing. We found that the ACC is tent with previous studies,52,56 our anterograde tracing showed
a part of neural circuits associated with pain-induced neural that the ACC densely innervates the dl/lPAG (Figures 4 and
plasticity. The downstream circuit of the ACC is closely related S4). The silencing of the ACC in the CFA-induced chronic
to the dl/lPAG, which is a key region for active defensive pain model shows an increase in the network strength of its
response. The activation of the ACC-dl/lPAG circuit enhances downstream regions, such as the dl/lPAG (Figure 2). Previous
both reflexive and active avoidance behavior. Our results sug- studies have suggested that the ACC-PAG synapse is
gest that increased input from the ACC to the PAG might be a glutamatergic.57,58 Injection of a retrograde tracer into the PAG
critical factor for abnormal circuit changes in chronic pain. followed by immunohistochemical staining indicates that the
retrogradely labeled neurons in the ACC contain glutamate-like
The role of the ACC in pain chronification and fear immunoreactivity.57 Moreover, pharmacological approaches
formation have demonstrated that the ACC is one of the glutamatergic in-
The ACC receives direct nociceptive inputs and encodes the un- puts to the PAG.58 In this study, the activation of the ACC-dl/
pleasantness of pain.13 In both animal and human studies, the lPAG induced pain hypersensitivity, whereas KET reduced this
activity in the ACC correlates with the intensity of acute pain pain hypersensitivity (Figure 5), indicating glutamatergic synaptic
stimuli,40,41 and chronic persistent pain causes hyperactivity in neurotransmission.
the excitatory pyramidal neurons of the ACC.40,42,43 Since the In fear response, the neural activity in the ACC is closely
optogenetic activation of GABAergic interneurons in the cortex related to terminate freezing.23,59 Exploratory behavior is neces-
suppresses the spontaneous activity of excitatory pyramidal sary for recognizing the safety and dangers of the environment,
neurons, the resulting negative BOLD response reflects baseline which affects avoidance decisions. In the presence of a persis-
activity.28,29 In our data, the negative BOLD response after ACC tent threat signal, the activation of the ACC-dl/lPAG circuit
silencing was greater in the chronic inflammatory pain model enhanced exploratory behavior toward predator odor
(Figure 2), suggesting that the baseline neural activity of the (Figures 6D and S5B). Therefore, when an uncertain threat per-
excitatory pyramidal neurons in the ACC increased from chronic sists, the ACC-PAG circuit induces an active response rather
persistent pain. than a passive response withholding the action selection. In
In rodent studies, both electrical stimulation and chemical learned fear response to noxious stimuli with APAT, the activa-
activation of the glutamate receptors in the ACC produced a pro- tion of ACC-dl/lPAG also enhanced the avoidance behavior to-
nociceptive and aversive response.17–20,44,45 Compared with ward the approaching footshock zone (Figure 6H). Therefore,
electrical and chemical activation, optogenetic stimulation has the ACC-dl/lPAG circuit is involved in both pain and active defen-
the advantage of modulating specific cell types at precise time sive responses.
windows. The optogenetic activation of the ACC pyramidal neu- The intrinsic excitability of PrL endogenously suppresses pain
rons (CaMKII or Thy1) facilitates pain-related behaviors.46,47 and anxiety.39,60,61 In the chronic pain model, in contrast to the
Consistent with previous studies,47,48 our study showed that increased activity of the ACC, the excitability of PrL is
direct (via CaMKII-eArch) and indirect (via VGAT-ChR2) inhibition decreased,39,60 and decreased inputs from the PrL to vlPAG
of the ACC pyramidal neurons selectively suppresses pain hy- are critical for the development of chronic pain.37–39 Thus, the
persensitivity without affecting nociception (Figure 1). Further- PrL and ACC project into different subregions of the PAG with
more, the modulation of the glutamatergic synaptic plasticity in opposing actions in pain and defensive behavior (Figures 4D
the ACC was sufficient to relieve chronic pain without affecting and S4), suggesting that the frontal cortical projections to the
nociception.49–51 Thus, the ACC pyramidal neurons are impor- PAG are important for pain and defensive response.
tant for synaptic alteration of chronic pain.13
The artificial activation of the ACC also induces fear experi- Relevance to pain fMRI studies in humans
ence even without aversive stimuli.18,19 The inhibition of the The human MFC, which consists of the ACC and the medial pre-
ACC during the fear-acquisition phase does not impair nocicep- frontal cortex (mPFC), is a key structure for pain, negative
tion or innate fear response but reduces the fear response during emotion, and cognitive control.62 The rodent MFC has been a
the fear-expression phase.18,52 Furthermore, the ACC is respon- controversial issue since granular parts, which are critical for
sible for paying attention to pain-related signals.23,24,53 In trace high-level cognitive function in human mPFC, do not exist.63
fear conditioning with a time gap between footshock and a The Brodmann classification divided the cerebral cortex based
conditioned stimulus, distraction interferes with the formation on a cytoarchitectural organization of cells, such as the granular
of the fear experience, so the attentional role of the ACC is layer. By the Brodmann nomenclature, the rodent ACC (Area 24)
essential in trace fear.24,54 This evidence suggests that the func- appears to be homologous to the human dACC or anterior mid-
tional plasticity of the ACC generates a teaching signal for aver- cingulate cortex (aMCC) regions and the rodent PrL (Area 32) to
sive experiences and is engaged in fear formation. the human pgACC.64 In human fMRI studies, imminent danger
requiring active avoidance increases the activity in the dACC
The role of the frontal cortical projections to the PAG in and dACC-PAG connectivity, whereas high shock probability in-
pain and fear response creases the activity in the pgACC.65 Furthermore, pain-related
The PAG is a critical hub for the descending pain modulatory prediction errors lead to the selection of actions expected to
system, originating in the brainstem and regulating the pain have a low probability of pain in which the failure to avoid pain

2842 Current Biology 32, 2834–2847, July 11, 2022

 ll
Article

due to incorrect prediction causes higher activity in aMCC and model (Figure 2F). BNST activity regulates pain aversion, with
PAG.66 Cognitive tasks such as visual attention produce atten- recent studies hypothesizing its involvement in goal-directed
tional analgesia, which involves the ACC and PAG.67–69 Given defensive response.89,90 Ventrolateral OFC’s (OFCvl) role in anti-
that the ACC is activated by both reward and pain to generate nociception is well reported, with little known about the OFCl’s
attention for learning and prediction,70–72 attention reallocation involvement aside from its roles in decision making and valence
in the context of pain is likely to be involved in defensive re- integration.91–93 Taken together, these areas may provide moti-
sponses. In this study, we proposed a hypothesis for different vational and cognitive control of pain with the ACC.
connections of ACC subregions to PAG (dACC-dl/lPAG versus Finally, to our surprise, the anteromedial thalamus (AM)
pgACC-vlPAG) and their different roles in defensive responses showed the second-highest response in the chronic pain model
in pain processing (Figure S4), which is worthy of future study (Figure 2F). The AM reacts to predator and social threats94,95 and
by human fMRI. has reciprocal connections with the ACC, which contribute to
contextual fear memory.52,96 However, since the AM did not
The role of the ACC in motor output respond to noxious stimulation (Figure 3E), further studies are
The ACC is known to engage in shaping the motor output based needed to determine the role of AM in pain chronification.
on the sensory input related to expected reward or punish- In conclusion, our findings offer insights into the neural
ment.23,73–75 In this study, the modulatory effect of the ACC mechanisms underlying the fear-avoidance response to
was pronounced in motor-related systems (Figures 2 and 3). pain, a key factor for chronic pain disability. Furthermore,
In particular, the SCm is essential for the integration of we sought to apply ofMRI to explore pain circuits at the sys-
sensorimotor function and the transmission of visual threat tems level, which will provide a much-needed avenue for pain
signals,34,76,77 and the ACC-SC circuit is known to coordinate research.
specific action selection.78 The ACC has dense projections to
the SCm (Figure 4), and noxious stimuli induce a positive STAR+METHODS
BOLD response in SCm, which is modulated by the optogenetic
silencing of ACC (Figure 3). Therefore, our data suggest that the Detailed methods are provided in the online version of this paper
ACC-SC circuit may be involved in pain-related defensive and include the following:
behavior.
The nigro-striatal-thalamocortical network, which includes the d KEY RESOURCES TABLE
substantia nigra (SN), dorsal lateral striatum (DLS), VAL, and M1/ d RESOURCE AVAILABILITY
M2, constitutes a key pathway in motor integration and execu- B Lead contact
tion.79–81 The ACC has a dense input to the nigro-striatal-thala- B Materials availability
mocortical network (Figure 4), and chronic pain appears to B Data and code availability
enhance the network strength of these downstream regions (Fig- d EXPERIMENTAL MODEL AND SUBJECT DETAILS
ure 2E). Furthermore, the optogenetic silencing of the ACC B Animals
significantly reduced the noxious stimulation-induced BOLD re- d METHOD DETAILS
sponses in the nigro-striatal-thalamocortical network (Figure 3E). B Stereotaxic surgery
Thus, the ACC seems to have a critical role in the motor output B Optogenetics
related to pain processing. B Behavior tests
The function of PAG also is closely related to the motor B MRI
cortex.82,83 However, during the unilateral modulation of the B Viral tracing
ACC-PAG circuit, we did not observe any abnormal motor d QUANTIFICATION AND STATISTICAL ANALYSIS
behavior such as contralateral rotations (full 360 turns) related B Behavior analysis
to the motor cortex84 (Videos S1 and S2). Furthermore, the mod- B GLM statistical mapping
ulation of the ACC-PAG circuit did not affect the speed of move- B BOLD quantification
ment (i.e., the movement distance) compared with the control B Functional connectivity analysis
group in the open-field test, whereas the optogenetic inhibition
significantly decreased movement distance in the innate fear SUPPLEMENTAL INFORMATION
test (Figure S5). Thus, the effect of the ACC-PAG circuit on loco-
Supplemental information can be found online at https://doi.org/10.1016/j.
motion seems to be associated with the affective aspect rather
cub.2022.04.090.
than the direct activation of the motor system.
ACKNOWLEDGMENTS
Potential targets for pain processing
The ACC is also involved in the motivation for goal-directed or This project was funded by the Institute for Basic Science in Korea (IBS-
effortful behavior.85,86 In particular, the nucleus accumbens R015-D1 to S.-G.K).
(NAc) is a key region for decreased motivation in chronic
AUTHOR CONTRIBUTIONS
pain,25 and the ACC-NAc circuit modulates pain and fear re-
sponses.87,88 In our study, ofMRI allowed the identification of
S.-G.K. obtained funding for the study and guided the project. J.-Y.L., T.Y.,
other potential target brain regions. Enhanced response in the and G.H.I. conducted the ofMRI experiments, collected data, and analyzed
bed nucleus of the stria terminalis (BNST) and the lateral the results. J.-Y.L., C.-H.L., and H.S. conducted the behavior experiments,
orbitofrontal cortex (OFCl) was detected in the chronic pain collected data, and analyzed the results. J.-Y.L. conducted histological

Current Biology 32, 2834–2847, July 11, 2022 2843

 ll
Article
experiments. J.-Y.L., T.Y., C.-W.W., and S.-G.K. wrote the manuscript. 15. Xu, H., Wu, L.J., Wang, H., Zhang, X., Vadakkan, K.I., Kim, S.S.,
C.-W.W. and S.-G.K. supervised the study. All the authors reviewed the manu- Steenland, H.W., and Zhuo, M. (2008). Presynaptic and postsynaptic am-
script and gave final approval. plifications of neuropathic pain in the anterior cingulate cortex.
J. Neurosci. 28, 7445–7453. https://doi.org/10.1523/JNEUROSCI.
DECLARATION OF INTERESTS 1812-08.2008.
16. Koga, K., Descalzi, G., Chen, T., Ko, H.G., Lu, J., Li, S., Son, J., Kim, T.,
The authors declare no competing interests. Kwak, C., Huganir, R.L., et al. (2015). Coexistence of two forms of LTP in
ACC provides a synaptic mechanism for the interactions between anxiety
Received: December 7, 2021 and chronic pain. Neuron 85, 377–389. https://doi.org/10.1016/j.neuron.
Revised: March 7, 2022 2014.12.021.
Accepted: April 28, 2022
17. Calejesan, A.A., Kim, S.J., and Zhuo, M. (2000). Descending facilitatory
Published: May 23, 2022
modulation of a behavioral nociceptive response by stimulation in the
adult rat anterior cingulate cortex. Eur. J. Pain 4, 83–96. https://doi.org/
REFERENCES
10.1053/eujp.1999.0158.

1. Melzack, R., and Casey, K.L. (1968). Sensory, motivational, and central 18. Johansen, J.P., and Fields, H.L. (2004). Glutamatergic activation of ante-
control determinants of pain: a new conceptual model. In The Skin rior cingulate cortex produces an aversive teaching signal. Nat. Neurosci.
Senses, D. Kenshalo Springfield, ed. (C.C. Thomas), pp. 423–439. 7, 398–403. https://doi.org/10.1038/nn1207.

2. Vlaeyen, J.W.S., Crombez, G., and Linton, S.J. (2016). The fear-avoid- 19. Tang, J., Ko, S., Ding, H.K., Qiu, C.S., Calejesan, A.A., and Zhuo, M.
ance model of pain. Pain 157, 1588–1589. https://doi.org/10.1097/j. (2005). Pavlovian fear memory induced by activation in the anterior
pain.0000000000000574. cingulate cortex. Mol. Pain 1, 6. https://doi.org/10.1186/1744-8069-1-6.

3. Kozlowska, K., Walker, P., McLean, L., and Carrive, P. (2015). Fear and 20. Chen, T., Taniguchi, W., Chen, Q.Y., Tozaki-Saitoh, H., Song, Q., Liu,
the defense cascade: clinical implications and management. R.H., Koga, K., Matsuda, T., Kaito-Sugimura, Y., Wang, J., et al. (2018).
Harv. Rev. Psychiatry 23, 263–287. https://doi.org/10.1097/HRP. Top-down descending facilitation of spinal sensory excitatory transmis-
0000000000000065. sion from the anterior cingulate cortex. Nat. Commun. 9, 1886. https://
doi.org/10.1038/s41467-018-04309-2.
4. Yilmaz, M., and Meister, M. (2013). Rapid innate defensive responses of
mice to looming visual stimuli. Curr. Biol. 23, 2011–2015. https://doi.org/ 21. Ortiz, S., Latsko, M.S., Fouty, J.L., Dutta, S., Adkins, J.M., and Jasnow,
10.1016/j.cub.2013.08.015. A.M. (2019). Anterior cingulate cortex and ventral hippocampal inputs to
5. Koutsikou, S., Apps, R., and Lumb, B.M. (2017). Top down control of spi- the basolateral amygdala selectively control generalized fear.
nal sensorimotor circuits essential for survival. J. Physiol. 595, 4151– J. Neurosci. 39, 6526–6539. https://doi.org/10.1523/JNEUROSCI.
4158. https://doi.org/10.1113/JP273360. 0810-19.2019.

6. Benarroch, E.E. (2008). Descending monoaminergic pain modulation: 22. Bian, X.L., Qin, C., Cai, C.Y., Zhou, Y., Tao, Y., Lin, Y.H., Wu, H.Y., Chang,
bidirectional control and clinical relevance. Neurology 71, 217–221. L., Luo, C.X., and Zhu, D.Y. (2019). Anterior cingulate cortex to
https://doi.org/10.1212/01.wnl.0000318225.51122.63. ventral hippocampus circuit mediates contextual fear generalization.
J. Neurosci. 39, 5728–5739. https://doi.org/10.1523/JNEUROSCI.
7. Lanius, R.A., Boyd, J.E., McKinnon, M.C., Nicholson, A.A., Frewen, P.,
2739-18.2019.
Vermetten, E., Jetly, R., and Spiegel, D. (2018). A review of the neurobi-
ological basis of trauma-related dissociation and its relation to cannabi- 23. Steenland, H.W., Li, X.Y., and Zhuo, M. (2012). Predicting aversive events
noid- and opioid-mediated stress response: a transdiagnostic, transla- and terminating fear in the mouse anterior cingulate cortex during trace
tional approach. Curr. Psychiatry Rep. 20, 118. https://doi.org/10.1007/ fear conditioning. J. Neurosci. 32, 1082–1095. https://doi.org/10.1523/
s11920-018-0983-y. JNEUROSCI.5566-11.2012.
8. Gross, C.T., and Canteras, N.S. (2012). The many paths to fear. Nat. Rev. 24. Han, C., O’Tuathaigh, C.M., van Trigt, L., Quinn, J.J., Fanselow, M.S.,
Neurosci. 13, 651–658. https://doi.org/10.1038/nrn3301. Mongeau, R., Koch, C., and Anderson. (2003). Trace but not delay fear
9. Coghill, R.C. (2020). The distributed nociceptive system: a framework for conditioning requires attention and the anterior cingulate cortex. Proc.
understanding pain. Trends Neurosci. 43, 780–794. https://doi.org/10. Natl. Acad. Sci. USA 100, 13087–13092.
1016/j.tins.2020.07.004. 25. Schwartz, N., Temkin, P., Jurado, S., Lim, B.K., Heifets, B.D., Polepalli,
10. Davis, K.D., Flor, H., Greely, H.T., Iannetti, G.D., Mackey, S., Ploner, M., J.S., and Malenka, R.C. (2014). Chronic pain. Decreased motivation dur-
Pustilnik, A., Tracey, I., Treede, R.D., and Wager, T.D. (2017). Brain imag- ing chronic pain requires long-term depression in the nucleus accum-
ing tests for chronic pain: medical, legal and ethical issues and recom- bens. Science 345, 535–542. https://doi.org/10.1126/science.1253994.
mendations. Nat. Rev. Neurol. 13, 624–638. https://doi.org/10.1038/ 26. Jin, Y., Meng, Q., Mei, L., Zhou, W., Zhu, X., Mao, Y., Xie, W., Zhang,
nrneurol.2017.122. X., Luo, M.H., Tao, W., et al. (2020). A somatosensory cortex input to
11. Buckner, R.L., Krienen, F.M., and Yeo, B.T. (2013). Opportunities and lim- the caudal dorsolateral striatum controls comorbid anxiety in
itations of intrinsic functional connectivity MRI. Nat. Neurosci. 16, persistent pain. Pain 161, 416–428. https://doi.org/10.1097/j.pain.
832–837. https://doi.org/10.1038/nn.3423. 0000000000001724.
12. Lee, J.H., Durand, R., Gradinaru, V., Zhang, F., Goshen, I., Kim, D.S., 27. Li, N., Chen, S., Guo, Z.V., Chen, H., Huo, Y., Inagaki, H.K., Chen, G.,
Fenno, L.E., Ramakrishnan, C., and Deisseroth, K. (2010). Global and Davis, C., Hansel, D., Guo, C., et al. (2019). Spatiotemporal constraints
local fMRI signals driven by neurons defined optogenetically by type on optogenetic inactivation in cortical circuits. eLife 8, e48622. https://
and wiring. Nature 465, 788–792. https://doi.org/10.1038/nature09108. doi.org/10.7554/eLife.48622.
13. Bliss, T.V., Collingridge, G.L., Kaang, B.K., and Zhuo, M. (2016). Synaptic 28. Moon, H.S., Jiang, H., Vo, T.T., Jung, W.B., Vazquez, A.L., and Kim, S.G.
plasticity in the anterior cingulate cortex in acute and chronic pain. Nat. (2021). Contribution of excitatory and inhibitory neuronal activity to BOLD
Rev. Neurosci. 17, 485–496. https://doi.org/10.1038/nrn.2016.68. fMRI. Cereb. Cortex 31, 4053–4067. https://doi.org/10.1093/cercor/
14. Zhao, M.G., Ko, S.W., Wu, L.J., Toyoda, H., Xu, H., Quan, J., Li, J., Jia, Y., bhab068.
Ren, M., Xu, Z.C., and Zhuo, M. (2006). Enhanced presynaptic neuro- 29. Jung, W.B., Jiang, H., Lee, S., and Kim, S.G. (2022). Dissection of brain-
transmitter release in the anterior cingulate cortex of mice with chronic wide resting-state and functional somatosensory circuits by fMRI with
pain. J. Neurosci. 26, 8923–8930. https://doi.org/10.1523/JNEUROSCI. optogenetic silencing. Proc. Natl. Acad. Sci. USA 119, https://doi.org/
2103-06.2006. 10.1073/pnas.2113313119.

2844 Current Biology 32, 2834–2847, July 11, 2022

 ll
Article
30. You, T., Im, G.H., and Kim, S.G. (2021). Characterization of brain-wide 46. Barthas, F., Sellmeijer, J., Hugel, S., Waltisperger, E., Barrot, M., and
somatosensory BOLD fMRI in mice under dexmedetomidine/isoflurane Yalcin, I. (2015). The anterior cingulate cortex is a critical hub for pain-
and ketamine/xylazine. Sci. Rep. 11, 13110. https://doi.org/10.1038/ induced depression. Biol. Psychiatry 77, 236–245. https://doi.org/10.
s41598-021-92582-5. 1016/j.biopsych.2014.08.004.
31. Kim, S.G., and Ogawa, S. (2012). Biophysical and physiological origins of 47. Kang, S.J., Kwak, C., Lee, J., Sim, S.E., Shim, J., Choi, T., Collingridge,
blood oxygenation level-dependent fMRI signals. J. Cereb. Blood Flow G.L., Zhuo, M., and Kaang, B.K. (2015). Bidirectional modulation of hy-
Metab. 32, 1188–1206. https://doi.org/10.1038/jcbfm.2012.23. peralgesia via the specific control of excitatory and inhibitory neuronal
activity in the. ACC. Mol. Brain 8, 81. https://doi.org/10.1186/s13041-
32. Logothetis, N.K. (2008). What we can do and what we cannot do with
015-0170-6.
fMRI. Nature 453, 869–878. https://doi.org/10.1038/nature06976.
48. Elina, K.C., Moon, H.C., Islam, J., Kim, H.K., and Park, Y.S. (2021). The
33. Petersen, C.C.H. (2019). Sensorimotor processing in the rodent barrel
effect of optogenetic inhibition of the anterior cingulate cortex in neuro-
cortex. Nat. Rev. Neurosci. 20, 533–546. https://doi.org/10.1038/
pathic pain following sciatic nerve injury. J. Mol. Neurosci. 71,
s41583-019-0200-y.
638–650. https://doi.org/10.1007/s12031-020-01685-7.
34. Evans, D.A., Stempel, A.V., Vale, R., Ruehle, S., Lefler, Y., and Branco, T. 49. Wu, L.J., Toyoda, H., Zhao, M.G., Lee, Y.S., Tang, J., Ko, S.W., Jia, Y.H.,
(2018). A synaptic threshold mechanism for computing escape Shum, F.W., Zerbinatti, C.V., Bu, G., et al. (2005). Upregulation of fore-
decisions. Nature 558, 590–594. https://doi.org/10.1038/s41586-018- brain NMDA NR2B receptors contributes to behavioral sensitization after
0244-6. inflammation. J. Neurosci. 25, 11107–11116. https://doi.org/10.1523/
35. Lischinsky, J.E., and Lin, D. (2019). Looming danger: unraveling the cir- JNEUROSCI.1678-05.2005.
cuitry for predator threats. Trends Neurosci. 42, 841–842. https://doi. 50. Wang, H., Xu, H., Wu, L.J., Kim, S.S., Chen, T., Koga, K., Descalzi, G.,
org/10.1016/j.tins.2019.10.004. Gong, B., Vadakkan, K.I., Zhang, X., et al. (2011). Identification of an ad-
36. Logothetis, N.K., Pauls, J., Augath, M., Trinath, T., and Oeltermann, A. enylyl cyclase inhibitor for treating neuropathic and inflammatory pain.
(2001). Neurophysiological investigation of the basis of the fMRI signal. Sci. Transl. Med. 3, 65ra3. https://doi.org/10.1126/scitranslmed.
Nature 412, 150–157. https://doi.org/10.1038/35084005. 3001269.
51. Li, X.Y., Ko, H.G., Chen, T., Descalzi, G., Koga, K., Wang, H., Kim, S.S.,
37. Huang, J., Gadotti, V.M., Chen, L., Souza, I.A., Huang, S., Wang, D.,
Shang, Y., Kwak, C., Park, S.W., et al. (2010). Alleviating neuropathic pain
Ramakrishnan, C., Deisseroth, K., Zhang, Z., and Zamponi, G.W.
hypersensitivity by inhibiting PKMzeta in the anterior cingulate cortex.
(2019). A neuronal circuit for activating descending modulation of neuro-
Science 330, 1400–1404. https://doi.org/10.1126/science.1191792.
pathic pain. Nat. Neurosci. 22, 1659–1668. https://doi.org/10.1038/
s41593-019-0481-5. 52. de Lima, M.A.X., Baldo, M.V.C., Oliveira, F.A., and Canteras, N.S. (2022).
The anterior cingulate cortex and its role in controlling contextual fear
38. Yin, J.B., Liang, S.H., Li, F., Zhao, W.J., Bai, Y., Sun, Y., Wu, Z.Y., Ding,
memory to predatory threats. eLife 11, e67007. https://doi.org/10.
T., Sun, Y., Liu, H.X., et al. (2020). dmPFC-vlPAG projection neurons
7554/eLife.67007.
contribute to pain threshold maintenance and antianxiety behaviors.
J. Clin. Invest. 130, 6555–6570. https://doi.org/10.1172/JCI127607. 53. Descalzi, G., Li, X.Y., Chen, T., Mercaldo, V., Koga, K., and Zhuo, M.
(2012). Rapid synaptic potentiation within the anterior cingulate cortex
39. Drake, R.A., Steel, K.A., Apps, R., Lumb, B.M., and Pickering, A.E. (2021). mediates trace fear learning. Mol. Brain 5, 6. https://doi.org/10.1186/
Loss of cortical control over the descending pain modulatory system de- 1756-6606-5-6.
termines the development of the neuropathic pain state in rats. eLife 10,
54. Gilmartin, M.R., Balderston, N.L., and Helmstetter, F.J. (2014). Prefrontal
e65156. https://doi.org/10.7554/eLife.65156.
cortical regulation of fear learning. Trends Neurosci. 37, 455–464. https://
40. Zhao, R., Zhou, H., Huang, L., Xie, Z., Wang, J., Gan, W.B., and Yang, G. doi.org/10.1016/j.tins.2014.05.004.
(2018). Neuropathic pain causes pyramidal neuronal hyperactivity in the
55. Senapati, A.K., Lagraize, S.C., Huntington, P.J., Wilson, H.D., Fuchs,
anterior cingulate cortex. Front. Cell. Neurosci. 12, 107. https://doi.org/
P.N., and Peng, Y.B. (2005). Electrical stimulation of the anterior cingulate
10.3389/fncel.2018.00107.
cortex reduces responses of rat dorsal horn neurons to mechanical stim-
41. Büchel, C., Bornhovd, K., Quante, M., Glauche, V., Bromm, B., and uli. J. Neurophysiol. 94, 845–851. https://doi.org/10.1152/jn.00040.2005.
Weiller, C. (2002). Dissociable neural responses related to pain intensity, 56. Fillinger, C., Yalcin, I., Barrot, M., and Veinante, P. (2018). Efferents of
stimulus intensity, and stimulus awareness within the anterior cingulate anterior cingulate areas 24a and 24b and midcingulate areas 24a’ and
cortex: a parametric single-trial laser functional magnetic resonance im- 24b. Brain Struct. Funct. 223, 1747–1778. https://doi.org/10.1007/
aging study. J. Neurosci. 22, 970–976. s00429-017-1585-x.
42. Zhou, H., Zhang, Q., Martinez, E., Dale, J., Hu, S., Zhang, E., Liu, K., 57. Beitz, A.J. (1989). Possible origin of glutamatergic projections to the
Huang, D., Yang, G., Chen, Z., and Wang, J. (2018). Ketamine reduces midbrain periaqueductal gray and deep layer of the superior colliculus
aversion in rodent pain models by suppressing hyperactivity of the ante- of the rat. Brain Res. Bull. 23, 25–35. https://doi.org/10.1016/0361-
rior cingulate cortex. Nat. Commun. 9, 3751. https://doi.org/10.1038/ 9230(89)90159-7.
s41467-018-06295-x.
58. Coutinho, M.R., and Menescal-de-Oliveira, L. (2010). Role of homocys-
43. Baliki, M.N., Chialvo, D.R., Geha, P.Y., Levy, R.M., Harden, R.N., Parrish, teic acid in the guinea pig (Cavia porcellus) anterior cingulate cortex in
T.B., and Apkarian, A.V. (2006). Chronic pain and the emotional brain: tonic immobility and the influence of NMDA receptors on the dorsal
specific brain activity associated with spontaneous fluctuations of inten- PAG. Behav. Brain Res. 208, 237–242. https://doi.org/10.1016/j.bbr.
sity of chronic back pain. J. Neurosci. 26, 12165–12173. https://doi.org/ 2009.11.047.
10.1523/JNEUROSCI.3576-06.2006. 59. Jhang, J., Lee, H., Kang, M.S., Lee, H.S., Park, H., and Han, J.H. (2018).
44. Zhang, L., Zhang, Y., and Zhao, Z.Q. (2005). Anterior cingulate cortex Anterior cingulate cortex and its input to the basolateral amygdala control
contributes to the descending facilitatory modulation of pain via dorsal innate fear response. Nat. Commun. 9, 2744. https://doi.org/10.1038/
reticular nucleus. Eur. J. Neurosci. 22, 1141–1148. https://doi.org/10. s41467-018-05090-y.
1111/j.1460-9568.2005.04302.x. 60. Wang, G.Q., Cen, C., Li, C., Cao, S., Wang, N., Zhou, Z., Liu, X.M., Xu, Y.,
45. Zugaib, J., Coutinho, M.R., Ferreira, M.D., and Menescal-de-Oliveira, L. Tian, N.X., Zhang, Y., et al. (2015). Deactivation of excitatory neurons in
(2014). Glutamate/GABA balance in ACC modulates the nociceptive re- the prelimbic cortex via Cdk5 promotes pain sensation and anxiety.
sponses of vocalization: an expression of affective-motivational compo- Nat. Commun. 6, 7660. https://doi.org/10.1038/ncomms8660.
nent of pain in guinea pigs. Physiol. Behav. 126, 8–14. https://doi.org/10. 61. Zhou, H., Martinez, E., Lin, H.H., Yang, R., Dale, J.A., Liu, K., Huang, D.,
1016/j.physbeh.2013.12.004. and Wang, J. (2018). Inhibition of the prefrontal projection to the nucleus

Current Biology 32, 2834–2847, July 11, 2022 2845

 ll
Article
accumbens enhances pain sensitivity and affect. Front. Cell. Neurosci. 78. Huda, R., Sipe, G.O., Breton-Provencher, V., Cruz, K.G., Pho, G.N.,
12, 240. https://doi.org/10.3389/fncel.2018.00240. Adam, E., Gunter, L.M., Sullins, A., Wickersham, I.R., and Sur, M.
62. Kragel, P.A., Kano, M., Van Oudenhove, L., Ly, H.G., Dupont, P., Rubio, (2020). Distinct prefrontal top-down circuits differentially modulate
A., Delon-Martin, C., Bonaz, B.L., Manuck, S.B., Gianaros, P.J., et al. sensorimotor behavior. Nat. Commun. 11, 6007. https://doi.org/10.
(2018). Generalizable representations of pain, cognitive control, and 1038/s41467-020-19772-z.
negative emotion in medial frontal cortex. Nat. Neurosci. 21, 283–289. 79. Pan, W.X., Mao, T., and Dudman, J.T. (2010). Inputs to the dorsal striatum
https://doi.org/10.1038/s41593-017-0051-7. of the mouse reflect the parallel circuit architecture of the forebrain.
Front. Neuroanat. 4, 147. https://doi.org/10.3389/fnana.2010.00147.
63. Laubach, M., Amarante, L.M., Swanson, K., and White, S.R. (2018).
What, if anything, is rodent prefrontal cortex? eNeuro 5, 0315-18.2018, 80. Bosch-Bouju, C., Hyland, B.I., and Parr-Brownlie, L.C. (2013). Motor thal-
https://doi.org/10.1523/ENEURO.0315-18.2018. amus integration of cortical, cerebellar and basal ganglia information: im-
n, M. (2017). What constitutes the prefrontal cortex? Science 358, plications for normal and parkinsonian conditions. Front. Comput.
64. Carle
Neurosci. 7, 163. https://doi.org/10.3389/fncom.2013.00163.
478–482. https://doi.org/10.1126/science.aan8868.
81. Mallet, N., Delgado, L., Chazalon, M., Miguelez, C., and Baufreton, J.
65. Mobbs, D., Marchant, J.L., Hassabis, D., Seymour, B., Tan, G., Gray, M.,
(2019). Cellular and synaptic dysfunctions in Parkinson’s disease: step-
Petrovic, P., Dolan, R.J., and Frith, C.D. (2009). From threat to fear: the
ping out of the striatum. Cells 8, 1005. https://doi.org/10.3390/
neural organization of defensive fear systems in humans. J. Neurosci.
cells8091005.
29, 12236–12243. https://doi.org/10.1523/JNEUROSCI.2378-09.2009.
82. Alloway, K.D., Smith, J.B., and Beauchemin, K.J. (2010). Quantitative
66. Roy, M., Shohamy, D., Daw, N., Jepma, M., Wimmer, G.E., and Wager,
analysis of the bilateral brainstem projections from the whisker and fore-
T.D. (2014). Representation of aversive prediction errors in the human
paw regions in rat primary motor cortex. J. Comp. Neurol. 518, 4546–
periaqueductal gray. Nat. Neurosci. 17, 1607–1612. https://doi.org/10.
4566. https://doi.org/10.1002/cne.22477.
1038/nn.3832.
83. Pagano, R.L., Fonoff, E.T., Dale, C.S., Ballester, G., Teixeira, M.J., and
67. Petrovic, P., Petersson, K.M., Ghatan, P.H., Stone-Elander, S., and
Britto, L.R.G. (2012). Motor cortex stimulation inhibits thalamic sensory
Ingvar, M. (2000). Pain-related cerebral activation is altered by a distract-
neurons and enhances activity of PAG neurons: possible pathways for
ing cognitive task. Pain 85, 19–30. https://doi.org/10.1016/s0304-
antinociception. Pain 153, 2359–2369. https://doi.org/10.1016/j.pain.
3959(99)00232-8.
2012.08.002.
68. Valet, M., Sprenger, T., Boecker, H., Willoch, F., Rummeny, E., Conrad,
84. Magno, L.A.V., Tenza-Ferrer, H., Collodetti, M., Aguiar, M.F.G.,
B., Erhard, P., and Tolle, T.R. (2004). Distraction modulates connectivity
Rodrigues, A.P.C., da Silva, R.S., Silva, J.D.P., Nicolau, N.F., Rosa,
of the cingulo-frontal cortex and the midbrain during pain–an fMRI anal-
D.V.F., Birbrair, A., et al. (2019). Optogenetic stimulation of the M2 cortex
ysis. Pain 109, 399–408. https://doi.org/10.1016/j.pain.2004.02.033.
reverts motor dysfunction in a mouse model of Parkinson’s disease.
69. Oliva, V., Gregory, R., Davies, W.E., Harrison, L., Moran, R., Pickering, J. Neurosci. 39, 3234–3248. https://doi.org/10.1523/JNEUROSCI.
A.E., and Brooks, J.C.W. (2021). Parallel cortical-brainstem pathways 2277-18.2019.
to attentional analgesia. Neuroimage 226, 117548. https://doi.org/10. 85. Holroyd, C.B., and Yeung, N. (2012). Motivation of extended behaviors
1016/j.neuroimage.2020.117548. by anterior cingulate cortex. Trends Cogn. Sci. 16, 122–128. https://
70. Bryden, D.W., Johnson, E.E., Tobia, S.C., Kashtelyan, V., and Roesch, doi.org/10.1016/j.tics.2011.12.008.
M.R. (2011). Attention for learning signals in anterior cingulate cortex. 86. Nass, S.R., Hahn, Y.K., McLane, V.D., Varshneya, N.B., Damaj, M.I.,
J. Neurosci. 31, 18266–18274. https://doi.org/10.1523/JNEUROSCI. Knapp, P.E., and Hauser, K.F. (2020). Chronic HIV-1 Tat exposure alters
4715-11.2011. anterior cingulate cortico-basal ganglia-thalamocortical synaptic cir-
71. Hayden, B.Y., Heilbronner, S.R., Pearson, J.M., and Platt, M.L. (2011). cuitry, associated behavioral control, and immune regulation in male
Surprise signals in anterior cingulate cortex: neuronal encoding of un- mice. Brain Behav Immun Health 5, 100077. https://doi.org/10.1016/j.
signed reward prediction errors driving adjustment in behavior. bbih.2020.100077.
J. Neurosci. 31, 4178–4187. https://doi.org/10.1523/JNEUROSCI. 87. Woo, C.W., Roy, M., Buhle, J.T., and Wager, T.D. (2015). Distinct brain
4652-10.2011. systems mediate the effects of nociceptive input and self-regulation on
72. Schneider, K.N., Sciarillo, X.A., Nudelman, J.L., Cheer, J.F., and Roesch, pain. PLoS Biol 13, e1002036. https://doi.org/10.1371/journal.pbio.
M.R. (2020). Anterior cingulate cortex signals attention in a social para- 1002036.
digm that manipulates reward and shock. Curr. Biol. 30, 3724–3735.e2. 88. Smith, M.L., Asada, N., and Malenka, R.C. (2021). Anterior cingulate in-
https://doi.org/10.1016/j.cub.2020.07.039. puts to nucleus accumbens control the social transfer of pain and anal-
73. Shima, K., and Tanji, J. (1998). Role for cingulate motor area cells in gesia. Science 371, 153–159. https://doi.org/10.1126/science.abe3040.
voluntary movement selection based on reward. Science 282, 1335– 89. Deyama, S., Nakagawa, T., Kaneko, S., Uehara, T., and Minami, M.
1338. https://doi.org/10.1126/science.282.5392.1335. (2007). Involvement of the bed nucleus of the stria terminalis in the nega-
74. Paus, T. (2001). Primate anterior cingulate cortex: where motor control, tive affective component of visceral and somatic pain in rats. Behav.
drive and cognition interface. Nat. Rev. Neurosci. 2, 417–424. https:// Brain Res. 176, 367–371. https://doi.org/10.1016/j.bbr.2006.10.021.
doi.org/10.1038/35077500. 90. Hulsman, A.M., Terburg, D., Roelofs, K., and Klumpers, F. (2021). Roles
75. Williams, Z.M., Bush, G., Rauch, S.L., Cosgrove, G.R., and Eskandar, of the bed nucleus of the stria terminalis and amygdala in fear reactions.
E.N. (2004). Human anterior cingulate neurons and the integration of Handb. Clin. Neurol. 179, 419–432. https://doi.org/10.1016/B978-0-12-
monetary reward with motor responses. Nat. Neurosci. 7, 1370–1375. 819975-6.00027-3.
https://doi.org/10.1038/nn1354. 91. Nogueira, R., Abolafia, J.M., Drugowitsch, J., Balaguer-Ballester, E.,
76. Wei, P., Liu, N., Zhang, Z., Liu, X., Tang, Y., He, X., Wu, B., Zhou, Z., Liu, Sanchez-Vives, M.V., and Moreno-Bote, R. (2017). Lateral orbitofrontal
Y., Li, J., et al. (2015). Processing of visually evoked innate fear by a non- cortex anticipates choices and integrates prior with current information.
canonical thalamic pathway. Nat. Commun. 6, 6756. https://doi.org/10. Nat. Commun. 8, 14823. https://doi.org/10.1038/ncomms14823.
1038/ncomms7756. 92. Ong, W.Y., Stohler, C.S., and Herr, D.R. (2019). Role of the prefrontal cor-
77. Zhou, Z., Liu, X., Chen, S., Zhang, Z., Liu, Y., Montardy, Q., Tang, Y., Wei, tex in pain processing. Mol. Neurobiol. 56, 1137–1166. https://doi.org/
P., Liu, N., Li, L., et al. (2019). A VTA GABAergic neural circuit mediates 10.1007/s12035-018-1130-9.
visually evoked innate defensive responses. Neuron 103, 473–488.e6. 93. Huang, J., Zhang, Z., Gambeta, E., Chen, L., and Zamponi, G.W. (2021).
https://doi.org/10.1016/j.neuron.2019.05.027. An orbitofrontal cortex to midbrain projection modulates hypersensitivity

2846 Current Biology 32, 2834–2847, July 11, 2022

 ll
Article
after peripheral nerve injury. Cell Rep. 35, 109033. https://doi.org/10. information flow. Proc. Natl. Acad. Sci. USA 118, https://doi.org/10.
1016/j.celrep.2021.109033. 1073/pnas.2023265118.
94. de Lima, M.A., Baldo, M.V., and Canteras, N.S. (2017). A role for the an-
teromedial thalamic nucleus in the acquisition of contextual fear memory 98. Cimadevilla, J.M., Wesierska, M., Fenton, A.A., and Bures, J. (2001).
to predatory threats. Brain Struct. Funct. 222, 113–129. https://doi.org/ Inactivating one hippocampus impairs avoidance of a stable room-
10.1007/s00429-016-1204-2. defined place during dissociation of arena cues from room cues by rota-
tion of the arena. Proc. Natl. Acad. Sci. USA 98, 3531–3536. https://doi.
95. Rangel, M.J., Jr., Baldo, M.V.C., and Canteras, N.S. (2018). Influence of
org/10.1073/pnas.051628398.
the anteromedial thalamus on social defeat-associated contextual fear
memory. Behav. Brain Res. 339, 269–277. https://doi.org/10.1016/j.
99. Bullmore, E., and Sporns, O. (2009). Complex brain networks: graph
bbr.2017.10.038.
theoretical analysis of structural and functional systems. Nat. Rev.
96. Nelson, A.J.D. (2021). The anterior thalamic nuclei and cognition: a role Neurosci. 10, 186–198. https://doi.org/10.1038/nrn2575.
beyond space? Neurosci. Biobehav. Rev. 126, 1–11. https://doi.org/10.
1016/j.neubiorev.2021.02.047. 100. Zalesky, A., Fornito, A., and Bullmore, E.T. (2010). Network-based statis-
97. Jung, W.B., Im, G.H., Jiang, H., and Kim, S.G. (2021). Early fMRI re- tic: identifying differences in brain networks. Neuroimage 53, 1197–1207.
sponses to somatosensory and optogenetic stimulation reflect neural https://doi.org/10.1016/j.neuroimage.2010.06.041.

Current Biology 32, 2834–2847, July 11, 2022 2847

 ll
Article

STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER

Bacterial and virus strains
AAV5-CaMKIIa-hChR2(H134R)-eTFP Addgene Cat#26969-AAV5;RRID:Addgene_26969
AAV5-CaMKIIa-eArch3.0-eYFP UNC Vector N/A
AAV5-CaMKIIa-eYFP UNC Vector N/A
AAV5-CaMKIIa-mCherry Addgene Cat#114469-AAV5
Chemicals, peptides, and recombinant proteins
Capsaicin Sigma-Aldrich Cat#M2028
Freund’s Adjuvant, Complete (CFA) Sigma-Aldrich Cat#F5881
Experimental models: Organisms/strains
Mouse:C57BL/6J OrientBio (South Korea) N/A
Mouse: VGAT-ChR2-eYFP (B6.Cg-Tg The Jackson Laboratory RRID:IMSR_JAX:014548
(Slc32a1-COP4*H134R/eYFP 8Gfng/J)
Software and algorithms
MATLAB R2020b MathWorks https://www.mathworks.com; RRID: SCR_001622
Brain Connectivity Toolbox MATLAB Brain Connectivity Toolbox https://sites.google.com/site/bctnet/; PRID:SCR_004841
FSL Analysis Group, FMRIB, Oxford, UK https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/; RRID: SCR_002823
AFNI SSCC at NIMH https://afni.nimh.nih.gov/; RRID: SCR_005927
SPM12 Wellcome Department of Cognitive https://www.fil.ion.ucl.ac.uk/spm/; RRID: SCR_002823
Neurology, London, UK
GraphPad 9.0 Prism www.graphpad.com; RRID:SCR_002798
EthoVision XT Noldus https://www.noldus.com/ethovision-xt; PRID:SCR_000441

RESOURCE AVAILABILITY

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Seong-Gi
Kim ([email protected]).

Materials availability
This study did not generate new unique reagents.

Data and code availability

d All data reported in this paper will be shared by the lead contact upon request.
d This paper does not report original code.
d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animals
All experimental procedures were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at
Sungkyunkwan University and followed relevant guidelines and regulations confirmed by IACUC. Adult male C57BL/6 mice
(n = 42) aged 7 to 10 weeks were purchased from Orient Bio (Seongnam, Korea), and VGAT-ChR2-eYFP (B6.Cg-Tg
(Slc32a1-COP4*H134R/eYFP 8Gfng/J) mice were bred in-house from breeding pairs obtained from Jackson Laboratory
(Bar Harbor, ME, USA). Both male (n=5) and female (n = 15) VGAT-ChR2-eYFP mice and VGAT-ChR2-eYFP negative female
littermates (n = 4) were used. The exact number of mice for each experiment can be found in the figure legends. Mice were
housed independently in ventilated cages post-surgery in a temperature/humidity-controlled facility and maintained with
standard lab chow and water ad libitum under a 12-hour dark-light cycle. Behavior and fMRI experiments were conducted

e1 Current Biology 32, 2834–2847.e1–e5, July 11, 2022

 ll
Article

three-four weeks post-surgery. Behavior differences between genders were not observed and thus both genders were grouped
for analysis.

METHOD DETAILS

Stereotaxic surgery
The adeno-associated viruses (AAVs) were purchased from Addgene or UNC Vector Core; AAV5-CaMKIIa-hChR2(H134R)-eYFP
(Addgene), AAV5-CaMKIIa-eArch3.0-eYFP (UNC Vector), AAV5-CaMKIIa-eYFP (UNC Vector), AAV5-CaMKIIa-mCherry (Addgene).
Mice were initially inducted with inhaled isoflurane (Hana, Korea) and then anesthetized by intraperitoneal (IP) injection of ketamine
(Yuhan, Korea) and xylazine (RompunⓇ, Bayer, Korea) mixture (100 mg/kg and 10 mg/kg) before fixing onto a stereotaxic frame. To
maintain the anesthetic conditions throughout surgery duration, isoflurane (0.5-1.0 %) in oxygen and air gases (1:4 ratio) was used,
based on physiological signal changes. To alleviate pain and inflammation, lidocaine (Daihan, Korea) was subcutaneously injected
into the scalp, and Metacam (5 mg/kg, meloxicam, Boehringer Ingelheim, Korea) was injected subcutaneously into the loose skin
over the neck. Male C57BL6 mice were used for viral infection into left side of ACC (anteroposterior [AP], 1.0 mm relative to bregma;
mediolateral [ML], 0.3 mm; dorsoventral [DV], 1 mm relative to surface) or left side of PrL ([AP], 2.5 mm relative to bregma; [ML],
0.3 mm; [DV], 1 mm relative to surface). A glass capillary was connected to the injection pump (Nanoliter 2010, World Precision
Instruments) and a total 1 ml of viral suspension was filled. After virus injection, optic fiber (105 mm inner core diameter, NA = 0.22,
2 mm, Thorlabs, USA) for optogenetic modulation was immediately implanted at the same [AP] and [ML] coordinates, except for
0.8 mm [DV] relative to surface. For the optogenetic modulation of ACC input to PAG, the optic fiber (100 mm inner core diameter,
NA = 0.22, 3.5mm, Doric Lenses Inc, Canada) was inserted into left side of PAG ([AP], -4.0 mm relative to bregma; [ML], 0.5 mm;
[DV], 1.9 mm relative to the surface), following virus injection into the ACC. For VGAT-ChR2 group, only the optic fiber was implanted
without virus injection. We used bio-compatible silicon elastomer (Kwik-Sil, World Precision Instruments, Sarasota, FL, USA) to seal
the fiber implantation site and dental cement (SB, Sun-Medical Co., Shiga, Japan) to fix the optic fiber onto the skull. Behavior or
ofMRI experiments were conducted at least 3-4 weeks after virus injection/fiber implantation.

Optogenetics
The location of optic fiber was confirmed by an anatomical image using MRI. For the behavior test, the implanted optic fiber was
connected to a fiber-optic cable (Doric Lenses Inc, Canada) coupled to 473 nm, 532 nm, or 566 nm diode-pumped solid state laser
(Changchun New Industries Optoelectronics Tech. Co., Ltd, Changchun, China). The constant output power was calibrated to be 3-5
mW at the tip of the optic fiber, as measured by a power meter (PM100D, Thorlabs, USA). The experimental procedure for ofMRI was
previously described in detail.97 The implanted optic fiber was connected to Low Profile Patch Cord Adapter (LPPA) to modify the
direction path of a patch cord without bending the fiber cable. 437 nm, 532 nm, or 566 nm diode-pumped solid state laser (Chang-
chun New Industries Optoelectronics Tech. Co., Ltd, Changchun, China) was used, and the constant output power was calibrated to
be 3 mW at the tip of the optic fiber. The intensity of output power was adjusted according to the BOLD response. Given our previous
research,28,29 the light stimulus (437 nm) was delivered with 20 Hz/10ms parameters to silence excitatory pyramidal neuron with
VGAT-ChR2 mice. 532 nm (for behavior test) and 566 nm (for ofMRI) light was given continuously to silence excitatory pyramidal
neuron using AAV5-CaMKIIa-eArch3.0-eYFP.

Behavior tests
Thermal pain test (Hargreaves test)
To assess thermal-evoked pain, paw withdrawal latency was measured using the Hargreaves method (IITC Life Science Plantar Test,
Victory Blvd Woodland Hills, CA, USA). Before the test, mice were habituated in an acrylic observation chamber (size ranges 10 x 10 x
15 cm3) on a glass plate. The infrared stimulation (48-50% or 80% intensity) was applied to the hind paw. Paw withdrawal latency was
measured 4 to 5 times by an observer blinded to the treatment and then averaged, which was represented by thermal pain threshold.
Pain model
Before the induction of the pain model, the baseline of thermal pain threshold was measured. The induction of capsaicin/CFA-evoked
pain hypersensitivity was confirmed by the decrease in thermal pain threshold compared to baseline or contralateral paw. For the
capsaicin-induced acute pain model, capsaicin (Sigma, USA) was dissolved in DMSO, and 20ml of capsaicin (1mg, diluted in 0.9%
saline) was injected into the plantar surface of the right hind paw. The time mice spent licking was measured during 10 minutes
by an observer who was blinded to the genotype of mice. The thermal pain test was performed 20 minutes after capsaicin injection.
For the CFA-induced chronic pain model, 20ml of undiluted CFA (Sigma, USA) was injected twice into the plantar surface of the right
hind paw. The thermal pain test was performed 3 or 4 days after the first injection of CFA (before the second injection of CFA). Sub-
sequently, the mice received a second injection of CFA 7 days after the first and the thermal pain test was performed once more.
Open field test
To assay locomotor activity and anxiety, the duration in the center or border zone and the speed of movement were measured by
automated analysis software (EthoVision). The sound insulated behavioral room was illuminated by indirect dim lighting. Single
mice were placed in the center of an acrylic apparatus (size ranges 50 x 50 x 38 cm3) consisting of the center zone (24 x 24 cm2)
and the border zone, and then were allowed to explore freely. Optogenetic stimulations (473nm, 35mW, 20Hz, 10ms) were given
over one minute with one-minute intervals between each. The ON-OFF period was repeated 3 times each. To quantify the anxiety

Current Biology 32, 2834–2847.e1–e5, July 11, 2022 e2

 ll
Article

behavior, we calculated the speed of movement and the duration in the center zone or border zone. The effect of optogenetic stim-
ulation was defined as the differences between ON and OFF.
Fox urine-induced innate fear test
Before the test, mice were habituated for 2 minutes in an acrylic apparatus (size ranges 35 x 25 x 23 cm3). Mice were exposed to fox
urine in an inescapable environment consisting of zone1 (12 x 25 cm2) close to fox urine and zone2 (12 x 25 cm2) away from fox urine.
Three sheets of KimWipes laboratory tissue were stacked, folded in half, and fixed in zone1. 2 ml of fox urine were applied. Optogenetic
stimulation (473 nm, 35 mW, 20 Hz, 10 ms) was given over one minute with an interval of one minute between consecutive trials. The
ON-OFF period was repeated 3 times each. Using automated analysis software (EthoVision), the duration in zone1 or zone2 and the
speed of movement were measured. The effect of optogenetic stimulation was defined as the differences between ON and OFF.
Active place avoidance task
To evaluate the active avoidance behavior, we tested mice using APAT.98 A custom-built rotating circular maze (radius: 17 cm) with
transparent acrylic walls (height: 25 cm) and floor consisting of a metal grill (3.4 mm diameter bars with 1 cm spacing) was used for
APAT training. The apparatus was placed in a well-lit isolation booth, which had distinct black-and-white visual patterns on four walls
(vertical stripes, square grids, diagonal stripes, and polka dots). The rotation speed of the maze was given in rotations per minute
(RPM). The position of the animal during the task was analyzed instantly with video tracking software (EthoVision). One to three
days before the test, mice were habituated for 30 minutes inside the apparatus without any shocks or rotation. APAT training was
conducted over two days following the habituation session. On day 1, mice underwent two sessions with different maze rotation
speeds (3 RPM then 2 RPM) with an inter-session interval of two hours. In these sessions, whenever mice entered a pre-defined
shock zone spanning one-sixth of the maze, the mice received a footshock (0.2 mA, 60 Hz) for 500ms. If the mice failed to exit
the shock zone, additional footshocks were given every 1.5 seconds. On day 2, the same set of sessions were conducted with an
addition of optogenetic stimulation. In each session, optogenetic stimulations (473 nm, 35 mW, 20 Hz, 10 ms) was given in an alter-
nating pattern over five minutes with an interval of five minutes. To quantify the active avoidance behavior, we calculated the number
of entries into the shock zone and mean angular distance from the shock zone for each of 5 min time bins. The effect of optogenetic
stimulation was defined as the differences in angular distance between time bins when the first optogenetic stimulation was turned on
(ON1) and the first 5 min bin when the laser was turned off (OFF1). Additionally, we used the differences in angular distance between
the first two time bins (OFF1 and OFF2) with no optogenetic stimulation as a baseline.

MRI
Animal preparation
All MRI experiments were performed on a 15.2T MRI scanner with an actively shielded 6-cm diameter gradient (Bruker BioSpec,
Billerica, MA, USA) and a 15 mm ID surface coil for radiofrequency (RF) transmission/reception. BOLD fMRI was conducted on
CFA (n = 8), naı̈ve (n = 9), and ACCCaMKII (n = 7) group of mice under self-breathing IV continuous infusion dexmedetomidine/isoflur-
ane (0.05 mg/kg/hr/0.3%) anesthesia described in detail previously.30 Mice were inducted with 2-4 % isoflurane during which the tail
vein was canulated for IV infusion of dexmedetomidine (Precedex, Hospira, NC, USA). Next, mice were secured to a custom-made
MR cradle via ear and bite bar. The cradle was transferred to the magnet and connected to a small animal ventilator (Model 1030,
Small Animal Instrument Inc., Stony Brook, USA). A bolus of 0.05 mg/kg dexmedetomidine was given after MR preparation followed
by discontinuation of isoflurane. Infusion of dexmedetomidine at (0.05 mg/kg/hr) was started 10 min post bolus. Isoflurane was
continued at 0.3% 30 min post dexmedetomidine bolus. During the experiments, heart rate and motion-sensitive respiratory sig-
nals were continuously monitored (PhysioSuite, Kent Scientific Corp,) along with body temperature of the animals being maintained
at 37 ± 0.5  C with a warm-water heating system and a rectal thermometer.
Data acquisition
Data were acquired with Paravision 6.0.1 software. Image acquisition started after placing the mouse brain at the isocenter of the
magnet and correcting for field inhomogeneity via the MAPSHIM protocol. Anatomical images were acquired using fast low angle
shot (FLASH) sequence with the following parameters: TR/TE = 3000/45 ms, matrix size = 120 x 58, field of view (FOV) = 15.80 x
7.65 mm2, and 20 0.5-mm-thick slices. Functional images were acquired using gradient-echo single-shot echo planar imaging
(EPI) with the following parameters: TR/TE = 1000/11.5 ms, matrix size = 120 x 58, FOV = 15.84 x 7.65 mm2 (0.13 x 0.13 mm2 in-plane
resolution), 20 0.5-mm-thick slices, and 50 flip angle. To ensure best group alignment, the 6th slice was matched to the individual
mouse’s anterior commissure.
Functional scans were conducted in a block design of 60s baseline – 20s stimulation – 60s interstimulus interval – 20s stimulation –
60s recovery. The same optogenetic parameters used for behavior were used. In the naı̈ve group of mice, a further whisker-pad (WP)
stimulation was conducted in which the right whisker-pad was electrically stimulated with electrodes placed on the pad at 0.4mA,
4Hz.30 Simultaneous WP and optogenetic stimulation (WPOG) were also conducted on the same group of mice.
fMRI data pre-processing
BOLD fMRI data were processed as previously described using AFNI, FSL, and SPM12.30 Images were scaled 10x, despiked (AFNI,
3dDespike), motion-corrected (AFNI, 3dvolreg), and slice-time corrected (AFNI, 3dTshift). EPI images were then temporally averaged
(fsl, fslmaths -Tmean) and skull stripped (fsl, bet) to create an EPI template for registration. Anatomical images were skull stripped (fsl,
bet) and co-registered to the EPI template (SPM12). Linear affine parameters and nonlinear regularizations were calculated between
the skull-stripped co-registered anatomical image and Allen Mouse Brain Common Coordinate Framework v3 (CCFv3) template, af-
ter which the deformations were applied to the functional images with SPM12’s Normalize function.

e3 Current Biology 32, 2834–2847.e1–e5, July 11, 2022

 ll
Article

Viral tracing
Mice were sacrificed after behavior and ofMRI experiments were completed. Mice received a lethal dose of pentobarbital (> 60 mg/
kg) and were perfused with 0.9% saline followed by 4% paraformaldehyde (PFA). The whole brain was post-fixed by 4% PFA, cry-
opreserved in 30% sucrose, and embedded in OCT compound. The frozen specimen was transversely sectioned into 40 mm with a
cryotome. Sections were mounted on a slide glass and nuclear staining was performed with DAPI solution (1:10000, sigma). Fluo-
rescence images were obtained using a fluorescence microscope (Leica, Germany).
Region of interest abbreviations
Abbreviations for all ROIs are the following: ACCd/v anterior cingulate cortex dorsal/ventral, AId dorsal agranular insula, AM antero-
medial thalamus, APN anterior pretectal nuclei, AV anteroventral thalamus, BNST bed nucleus of striata terminalis, CA1 hippocam-
pus CA1 subfield, dl/lPAG dorsal-lateral lateral periaqueductal gray, DLS dorsolateral striatum, dmPAG dorsomedial PAG, GPe
globus pallidus externa, IL infralimbic, LD lateral dorsal thalamus, LP lateral posterior thalamus, M1hl primary motor hindlimb region,
M1wf primary motor whisker/forepaw region, M2 secondary motor cortex, MD mediodorsal thalamus, MRN midbrain reticular
nucleus, OFCl lateral orbitofrontal cortex, PF parafascicular nucleus, PrL prelimbic cortex, PO posterior thalamus, PPCa posterior
parietal cortex, RE nucleus reuniens, RSCd/v retrosplenial cortex dorsal/ventral, S1BF/FL/HL primary somatosensory barrel/fore-
limb/hindlimb, S2 secondary somatosensory, SCm motor superior colliculus, SN substantia nigra, SUB subiculum, V2al secondary
anterolateral visual cortex, VAL ventral anterolateral thalamus, vlPAG ventral lateral PAG, VM ventromedial thalamus, VP ventral pos-
terior thalamus, ZI zona incerta.
Abbreviations in anatomical tracing data follow: CM central medial thalamus, CL central lateral thalamus, cp cerebral peduncle,
CPu caudate putamen, Ect ectorhinal cortex, ic internal capsule, PRh perirhinal cortex, PRN pontine reticular nucleus, Rt reticular
thalamus, VA ventral anterior thalamus, VL ventral lateral thalamus, VPL ventral posterolateral thalamus, VPM ventral posteromedial
thalamus

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical details for every experiment are provided in the figure legends with ‘‘n’’ representing the number of animals per group.
Group differences with p < 0.05 were considered significant (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001).

Behavior analysis
Statistical analysis was performed using GraphPad Prism (GraphPad Software, USA). For comparison between two groups, statis-
tical analyses were carried out using paired t-test or unpaired t-test (two-tailed) as appropriate. For multiple comparisons, data were
analyzed using one-way ANOVA, repeated measures one-way ANOVA, or two-way ANOVA followed by Bonferroni test as appro-
priate. Data are presented as mean ± SEM.

GLM statistical mapping

Group BOLD response maps were generated by general linear model (GLM) by deconvolving the registered images with a design
matrix containing a linear detrend and our previously designed dexmedetomidine hemodynamic response function convolved
with the stimulation blocks as regressors (AFNI, 3dDeconvolve). Unpaired (CFA vs naı̈ve), paired (WP vs WPOG), or one sample (AC-
CCamKII) t-test was used to produce group response maps (AFNI, 3dttest++). The resultant maps were smoothed 2x the voxel size and
cluster corrected by finding first-nearest neighbor clusters that pass the threshold of P<0.005 uncorrected and correcting them with a
family-wise error correction to P<0.05 unless otherwise stated (AFNI, 3dClustsim).

BOLD quantification
The registered functional images were normalized to the baseline to calculate percent signal change. Anatomical region of interests
(ROI) based on the CCFv3 was used to extract the signal from each region. Areas such as dmPAG, dl/lPAG, and vlPAG did not have a
segmented atlas and thus a square 9 voxel ROI was drawn in each respective area within the PAG. For signal quantification, the ex-
tracted time series were temporally smoothed with a 6-second window to remove sharp outliers and area-under-curve (AUC) was
calculated by integrating the area between stimulation start and five seconds after cessation of stimulation of both stimulation blocks,
after which the mean AUC was calculated (cumtrapz, MATLAB). Exclusion criteria were applied by excluding any mouse that was
outside the outer fence using the interquartile range method. The AUCs of the ACC response (CFA, naı̈ve, and ACCCaMKII) or
S1BF (WP and WPOG) were used to calculate the outer fences. A final sample size was the following: CFA (n = 7/8), naı̈ve (n =
7/9), WP/WPOG (n = 7/9), and ACCCaMKII (n = 6/7). Comparison between two groups were conducted with GraphPad Prism. Statis-
tical analyses of the AUC values were carried out using the Wilcoxon test or Mann-Whitney U test (one-tailed) as appropriate. For
multiple comparisons, data were analyzed using the Kruskal-Wallis test followed by Dunn’s multi test as appropriate.

Functional connectivity analysis

Functional connectivity matrixes were made from calculating the Pearson correlation coefficient (MATLAB, corr) between all respon-
sive ROIs (n = 40), ipsilateral to ACC stimulation site, using the whole unsmoothed time series resulting in a 40 x 40 connectivity matrix
for each animal. Matrixes were Fisher-z transformed for group-averaging and statistical tests. Results were converted back to r for
presentation. The group connectivity matrix was further used to calculate modules using graph theory-based Newman’s spectral

Current Biology 32, 2834–2847.e1–e5, July 11, 2022 e4

 ll
Article

community detection via the Brain Connectivity Toolbox (BCT).99 Modules refer to groups of nodes with high intra-connectivity and
low inter-connectivity. As we only chose responsive ROIs, modularity was calculated with weighted, undirected matrixes. Resolution
parameters were adjusted between 0.9 (larger modules) – 2 (smaller modules) until modularity values were maximized. CFA and naı̈ve
dataset did not result in logical modules with low modularity and thus was sorted based on anatomical groups. WP dataset resulted in
two maximized modules using 0.9 resolution while WPOG dataset resulted in three modules with similar ordering to WP modules
using 1.2 resolution. WPOG ROI ordering was shifted to match WP module ordering. Each subject matrix was reordered based
on the calculated maximized module and unpaired (CFA vs naı̈ve) or paired (WP vs WPOG) statistical testing using network-based
statistic (NBS)100 with a threshold of 2.3, 1,000 permutations, and significance of p=0.05 after family-wise error rate multiple
correction was conducted to find significant networks.

e5 Current Biology 32, 2834–2847.e1–e5, July 11, 2022