A neuronal circuit for activating descending modulation of ......ARTICLES NATURE NEUROSCIENCE g a c...

Articleshttps://doi.org/10.1038/s41593-019-0481-5

1Department of Physiology and Pharmacology, University of Calgary, Calgary, Canada. 2Alberta Children’s Hospital Research Institute, University of Calgary, Calgary, Canada. 3Hotchkiss Brain Institute, University of Calgary, Calgary, Canada. 4Cumming School of Medicine, University of Calgary, Calgary, Canada. 5Department of Psychiatry & Behavioral Sciences and Department of Bioengineering, Howard Hughes Medical Institute, Stanford University, Palo Alto, CA, USA. 6These authors contributed equally: Junting Huang, Vinicius M. Gadotti. *e-mail: [email protected]; [email protected]

The human dorsal lateral PFC (dlPFC) plays an important role in the cognitive and affective modulation of pain1–3. Noninvasive stimulation of the dlPFC in humans leads to acute pain modulation and effectiveness in the treatment of chronic pain states4,5. The medial PFC (mPFC) is the rodent homolog of the primate dlPFC6. In mice, parvalbumin-expressing, γ-aminobutyric acid (GABA)ergic interneurons (PVINs) mediate increased feed-forward inhibition of mPFC output after peripheral nerve injury, leading to reduced mPFC output7,8. Optogenetic inhibition of these PVINs consequently mediates analgesia, as well as revers-ing increased place avoidance and escape behaviors associated with neuropathic pain7,8. How this occurs at the cellular, molecular and network level is not understood. Specifically, it has remained unclear to what extent specific inputs into the mPFC are modified after peripheral nerve injury, and which projections originating from the mPFC might be the downstream effectors of this dysregu-lated mPFC activity. The mPFC exhibits reciprocal connections with the BLA, a brain region that receives ascending sensory inputs from the spinal cord via the parabrachial nucleus9–11. This then raises the possibility that the observed alterations in mPFC function may originate from BLA inputs. In the present study, optogenetic approaches were used to examine the role of specific glutamatergic BLA inputs into the mPFC, reveal the molecular basis of dysregula-tion of mPFC signal processing and elucidate the downstream brain circuitry involved in neuropathic chronic pain states.

ResultsNerve injury-induced dysregulation of BLA inputs into the mPFC. To establish a functional link between the mPFC and BLA, first PVINs were genetically labeled by crossing parvalbumin-cre (PV-cre) mice with cre-dependent Ai9 tdTomato reporter mice (PV-cre::Ai9), allowing a record to be made of postsynaptic activity

from labeled layer 5 PVINs and nonlabeled pyramidal neurons in prelimbic mPFC slices. To achieve selective activation of inputs from the BLA, ChR2(H134R)-eYFP was virally expressed in pro-jection neurons originating from the BLA under the control of the glutamatergic, cell-specific, CaMKIIα promoter (Fig. 1a). Enhanced yellow fluorescent protein (eYFP)-expressing fibers were observed in the BLA, as well as in layers 2/3 and 5 of the mPFC (Fig. 1b). The experiments focused on prelimbic layer 5 because it is the major output region of the mPFC11. Current-clamp recordings showed that blue laser pulses induced time-locked action potential firing in AAV-ChR2(H134R)-eYFP-infected neu-rons in the BLA and in mPFC layer 5 neurons that receive BLA inputs (Fig. 1c). Voltage clamp recordings in prelimbic layer 5 PVINs revealed that blue laser light-induced short-latency evoked excitatory postsynaptic currents (eEPSCs) when the membrane voltage was held at −70 mV (Fig. 1d). These eEPSCs were blocked by bath application of tetrodotoxin (TTX) and could be rescued by bath application of the potassium channel blocker 4-amino-pyridine (4-AP), which is believed to cause calcium influx and thus facilitate vesicular release on optical activation of axon ter-minals (Fig. 1d)12. This is consistent with a monosynaptic nature of the BLA inputs to PVINs. An apparent small evoked inhibi-tory postsynaptic current (eIPSC) disappeared after TTX, and was not rescued with 4-AP, revealing a bisynaptic nature to this current. Similarly, voltage clamp recordings from prelimbic, layer 5 pyramidal neurons (glutamatergic projections or PNs) revealed blue light-induced eEPSCs as well as eIPSCs, both of which were eliminated by TTX (Fig. 1e). Only the excitatory events could be rescued by 4-AP, but not the disynaptic feedforward inhibition mediated by GABAergic interneurons. Therefore, glutamatergic BLA input into the mPFC forms monosynaptic connections with both PVINs and PNs.

A neuronal circuit for activating descending modulation of neuropathic painJunting Huang1,2,3,4,6, Vinicius M. Gadotti1,2,3,4,6, Lina Chen1,2,3,4, Ivana A. Souza1,2,3,4, Shuo Huang1,2,3,4, Decheng Wang1,2,3,4, Charu Ramakrishnan 5, Karl Deisseroth5, Zizhen Zhang 1,2,3,4* and Gerald W. Zamponi 1,2,3,4*

Neuropathic pain can be a debilitating condition with both sensory and affective components, the underlying brain circuitry of which remains poorly understood. In the present study, a basolateral amygdala (BLA)–prefrontal cortex (PFC)–periaqueductal gray (PAG)–spinal cord pathway was identified that is critical for the development of mechanical and thermal hypersensitivity after peripheral nerve injury. It was shown that nerve injury strengthens synaptic input from the BLA onto inhibitory inter-neurons located in the prelimbic medial PFC, by virtue of reduced endocannabinoid modulation. These augmented synaptic connections mediate a feedforward inhibition of projections from the PFC to the ventrolateral PAG region and its downstream targets. Optogenetic approaches combined with in vivo pharmacology reveal that these BLA–PFC–PAG connections alter pain behaviors by reducing descending noradrenergic and serotoninergic modulation of spinal pain signals. Thus, a long-range brain circuit was identified that is crucial for pain processing and that can potentially be exploited toward targeting neuropathic pain.

NAtuRe NeuRoSCIeNCe | www.nature.com/natureneuroscience

mailto:[email protected]:[email protected]://orcid.org/0000-0002-3474-6332http://orcid.org/0000-0001-8069-3756http://orcid.org/0000-0002-0644-9066http://www.nature.com/natureneuroscience

Articles Nature NeuroscieNce

g

ca

h

EP

SC

PV

IN/E

PS

CP

N

*

Sham SNI

0

2

4

6

8

BLA inputPN

PVIN or

0

2

4

6

8

eIP

SC

/eE

PS

C

*

BLA input PN

PVIN

20 m

V

400 ms

20 Hz

BLA PrL

20 m

V

400 ms

20 Hz

BLA

PrL

d

L1 L2/3 L5

b

AAV-CaMKIIα-ChR2-eYFP

PFC

BLA

f

PN

PVIN

BLA input

i

Control TTX

TTX + 4-AP

PN

–70 mV

0 mV

–70 mV

0 mV

–70 mV

0 mV

–70 mV

0 mV

PVIN–70 mV

0 mV

–70 mV

0 mV

Control TTX

TTX + 4-APPN

BLA input

PVIN PN

BLA input

PVIN

2.5

0

Control + AM281

** *

eEP

SC

(nA

)

PN

PVIN

BLA inputP

PR

*

**

**

**

0

0.4

0.8

1.2

0 1 2 3 4 5 6 7 8

eEP

SC

(nA

)

Stimulation (V)

SNI (14)Sham (18)

1.6

j

e

2

0

0.0

0.5

1.0

1.5

2.0

2.5

CB

1 pr

otei

n/α-

Tub

ulin

Sham SNI

*

50 kDa

50 kDa

CB1

α-Tubulin

0

1

2

3

4

5

CB

1 m

RN

A/β

-act

in

Sham SNI

*

k l

1.5

0.5

1.0

2.0

1

(8) (6)

(11) (17) (14) (13)

(18)

(14) (14) (13)

(5) (7)

(9)

(10)

(4) (4) (4) (4)

Sham

ipsi

Sham

cont

ra

SNI ip

si

SNI c

ontra

Sham SNI Sham SNI Sham SNI Sham SNI

2 weeks 4 weeks Ipsi Contra

Fig. 1 | effects of nerve injury on BLA inputs into the prelimbic mPFC. a, Experimental configuration showing AAV-CaMKIIα-ChR2-eYFP injection into the BLA. b, Sample image showing eYFP expression in BLA (left) and the prelimbic mPFC (right). Scale bar, 200 μm (n = 3 mice) c, Sample current-clamp traces showing light-evoked action potential firing in a ChR2-eYFP-expressing BLA neuron (left, representative of five cells) and in a layer 5 prelimbic mPFC neuron receiving BLA inputs (right, representative of eight cells). d, Sample voltage-clamp recording of blue laser-induced synaptic currents in a PVIN at Vh −70 mV or 0 mV, without (control) and with TTX (1 μM) or TTX and 4-AP (100 μM). e, Same as in d but in a layer 5 PN. The scale bars in d,e reflect 200 pA and 20 ms. f, Amplitude ratio of optically evoked eEPSCs from pairs of PVINs and PNs within the same brain slices (P = 0.0452, t = 2.234). g, Optically evoked eIPSC and eEPSC ratio from the same layer 5 PNs of prelimbic mPFC (P = 0.0122, t = 2.685). h, Amplitude of eEPSCs recorded in PVINs after electrical stimulation of the BLA inputs (4 V: P = 0.0337, t = 2.225; 5 V: P = 0.0212, t = 2.431; 6 V: P = 0.0123, t = 2.664; 7 V: P = 0.0147, t = 2.596; 8 V: P = 0.0031, t = 3.195). Paired pulse ratio (right) was obtained from the same experiments (P = 0.0492, t = 2.067). i, Amplitude of eEPSCs recorded in layer 5 PVINs after electrical stimulation of BLA inputs (stimulation voltage = 8 V) and CB1 antagonist (AM281, 1 μM) application (P = 0.0050, F = 4.778). j, Quantitative PCR analysis of CB1 receptor mRNA levels in the contralateral prelimbic mPFC from sham and SNI mice at 2 and 4 weeks after injury (P = 0.0486, t = 2.245 at 2 weeks; P = 0.3308, t = 1.001 at 4 weeks). k, Western blot examining CB1 receptor protein levels in the prelimbic mPFC at 4 weeks after SNI surgery. This blot represents four independent repetitions; in one additional experiment no change was observed, and this was not included in the quantification shown in l. l, Densitometry analysis of band intensities for CB1 receptor western blots (P = 0.0324, F = 3.773). The insets in various panels illustrate optogenetic electrophysiological configuration, with the laser stimulation indicated in blue. Data are presented as mean ± s.e.m. Two-tailed, unpaired Student’s t-tests were used for f,g,h (left and right) and j, one-way ANOVA with Newman–Keuls multiple comparison test for i and Tukey’s multiple comparison test for l. The asterisks denote significant differences. In f–i, numbers in parentheses denote numbers of cells. In j and l, numbers in parentheses reflect numbers of mice and numbers of independent assays, respectively. Contra, contralateral; ipsi, ipsilateral; L, layer; PrL, prelimbic area of mPFC.


http://www.nature.com/natureneuroscience

ArticlesNature NeuroscieNceGiven that the BLA projects to PNs both directly and indirectly

via PVINs, it is important to determine the net contribution of these synaptic inputs to overall PN output. To address this issue, blue laser (473 nm, 3–5 mW)-induced maximal eEPSCs were recorded in pairs of prelimbic mPFC layer 5 PNs and PVINs within the same brain slices in either sham-operated mice or spared nerve injury (SNI) mice, 7–10 weeks after AAV(ChR2(H134R)) injection into the BLA. Importantly, this approach eliminates differential viral expression of ChR2 between different slices or mice as a potential confound-ing factor. The eEPSC-PVIN:eEPSC-PN ratio was increased in SNI mice compared with sham controls (Fig. 1f), indicating a nerve injury-induced strengthening of excitatory inputs into GABAergic interneurons, which in turn are expected to mediate enhanced feedforward inhibition of PNs. As layer 5 PNs integrate both excit-atory and inhibitory input before sending output signals to other cortical and subcortical areas, the relative excitatory and inhibitory balance of these neurons is important in determining their overall activity. Notably, layer 5 PNs receive inputs from numerous sources, including GABAergic inputs from interneurons in layers 2/3 and 5 within the prelimbic cortex11,13, the connections of which are pre-served in slice preparations. Thus, first the synaptic strength of the direct glutamatergic BLA input into PNs was compared with that of their BLA-derived disynaptic feedforward inhibition (originat-ing from layer 2/3 and/or layer 5 PVINs) via recording of the blue laser-evoked maximum eEPSCs and eIPSCs in the same layer 5 PNs. These experiments revealed that nerve injury causes an augmenta-tion of inhibitory over excitatory events (Fig. 1g), thus leading to a potential net inhibition of mPFC output.

To further pinpoint the origin of the plasticity changes, fiber bundles originating from the BLA were electrically stimulated and recordings were conducted specifically from tdTomato-labeled layer 5 PVINs in mPFC slices. Compared with sham-operated mice, the amplitude of eEPSCs in PVINs from SNI animals was increased across the entire range of test intensities (Fig. 1h, left). Concomitantly, there were alterations in paired pulse facilitation in neurons from SNI animals (Fig. 1h, right). These data indicate that there is a nerve injury-induced increase in synaptic strength between BLA inputs and mPFC PVINs. Previous reports showed that cannabinoids play a role in controlling GABA release in the infralimbic cortex14 and that cannabinoid receptor type 1 (CB1) receptors are present on glutamatergic afferent inputs in the pre-limbic area15,16. To determine whether altered cannabinoid signal-ing may underlie nerve injury-induced alterations in synaptic input, eEPSCs in tdTomato-labeled PVINs were recorded in the absence and the presence of the CB1 antagonist AM281. Application of AM281 selectively augmented synaptic inputs in sham mice to lev-els that were similar to those seen in SNI mice (Fig. 1i). By contrast, AM281 had no effect in slices from SNI animals (Fig. 1i), indicating that cannabinoid signaling is already weakened after nerve injury. Quantitative PCR and Western blot analyses revealed a down-regulation of CB1 receptor messenger RNA (Fig. 1j) and protein (Fig. 1k,l) levels in the prelimbic mPFC SNI mice. The mRNA lev-els of enzymes involved in the regulation of endocannabinoids were not notably altered after nerve injury, with the exception of a reduc-tion in N-acetylphosphatidylethanolamine-hydrolyzing (NAPE)-phospholipase D (an enzyme involved in anandamide production) 4 weeks after injury (see Supplementary Fig. 1). Altogether, these data suggest that nerve injury leads to weakened endocannabinoid signaling in layer 5 of the mPFC, in turn causing a disinhibition of glutamatergic inputs into PVINs. It is possible that a similar dys-regulation may occur in layers 2 and 3, but this was not examined in the present study.

Functional role of BLA–mPFC inputs in neuropathic pain. To ascertain whether BLA inputs into the mPFC affect pain responses, in vivo optogenetic experiments were performed. Viruses were

injected into the BLA, leading to expression of opsins in the pre-limbic mPFC (see Fig. 1a,b and Supplementary Fig 2a,b) where they could be activated by an implanted optic cannula. Activation of BLA inputs with blue laser light (10 Hz, 5–10 mW) in mice that express AAV5-ChR2(H134R)-eYFP in the prelimbic mPFC (Fig. 2a) notably enhanced both mechanical and thermal hyperalgesia in SNI mice (Fig. 2b,c), but did not change the response of sham con-trol mice (see Supplementary Fig. 3a,b). Analogous experiments (Fig. 2d and see Supplementary Fig. 2c,d,e) with viral delivery of inhibitory opsins (Acrh3.0, NpHR3.0) in the BLA revealed that yellow light inhibited both mechanical and thermal hyperalgesia in SNI mice, but had no effect in sham-operated animals (Fig. 2e,f and see Supplementary Fig. 3c–h). Arch3.0-mediated inhibition of BLA–mPFC projections also led to analgesic effects in models of cold allodynia (Fig. 2g) and dynamic tactile plantar allodynia (Fig. 2h), and reversed place escape/avoidance (PEA) responses to mechanical stimulation of the hindpaw in SNI mice (Fig. 2i), without affecting locomotor activity (see Supplementary Fig. 4). The PEA response is consistent with a recent report showing that inhibition of BLA inputs into the mPFC alters avoidance responses during conditioned, context–induced retrieval of morphine with-drawal memory17. In vivo effects of optogenetic manipulations of the BLA-to-mPFC inputs were further supported by c-fos staining (see Supplementary Fig. 5). As expected, ChR2-mediated activation of BLA–mPFC inputs increased c-fos expression in PVINs in layer 5 of prelimbic mPFC, whereas Arch3.0-mediated inhibition of BLA–mPFC inputs increased c-fos expression in PNs in the same area.

Altogether the data reveal that peripheral nerve injury alters syn-aptic inputs from the BLA to the prelimbic mPFC, and optogenetic manipulation of this circuit modulates both sensory and affective components of pain. By contrast, although nerve injury was associated with increased anxiety, this was insensitive to optogenetic manipula-tion of the BLA–mPFC projections (see Supplementary Fig. 6), indi-cating that the observed changes in pain responses are not associated with a possible anxiolytic effect of BLA input manipulation.

Mapping of mPFC to PAG projections. How does altered mPFC output contribute to the pain neuraxis? The PAG area is an impor-tant hub that processes pain signals in the descending pathway18,19, with glutamatergic and GABAergic neurons mediating opposing effects20. The PAG is known to receive projections from the PFC13,21–25. However, whether this connection regulates pain responses and how they process pain signals have remained unclear. Thus the anatomical and functional connections between the mPFC and the PAG were investigated together with their roles in neuropathic pain. First, AAV5-CaMKIIα-ChR2(H134R)-eYFP was injected into layer 5 of the prelimbic mPFC region and then the PAG region was examined for YFP expression. These experiments revealed that mPFC layer 5 projects predominantly to the posterior ventrolateral (vl) PAG region (~−4.6 to −4.8 mm from the bregma) (Fig. 3a), with very few terminals projecting to other parts of the PAG (see Supplementary Fig. 7a). Then this connection was retrogradely traced with injection of the cholera toxin subunit B555 (CTB555, Fig. 3b) and green retrobeads (see Supplementary Fig. 7b) into the vlPAG, revealing selective layer 5 labeling within the prelimbic mPFC, with these retrotracers, but scant labeling of the anterior cin-gulate cortex, infralimbic area and orbital cortex (data not shown). In addition, the existence of direct mPFC–vlPAG projections was further confirmed by concomitant injections of AAV9-hEF1α-DIO-synaptophysin-mCherry virus into layer 5 of the prelim-bic mPFC and of Cav2-cre retrovirus into the vlPAG (Fig. 3c). As cre-dependent synaptophysin-mCherry expression can occur only in neurons infected by both viruses, expression of synaptophysin-mCherry confirms the synaptic connectivity of the projections from layer 5 of the prelimbic mPFC to the vlPAG. To further determine the identity of projections from the prelimbic mPFC to the vlPAG,




Vglut2-cre and GAD2-cre transgenic mice were crossed with Ai9 tdTomato reporter mice, respectively. CTB488 was injected into the vlPAG and it was found that the projection neurons from the prelimbic mPFC to the vlPAG were of glutamatergic rather than GABAergic origin (Fig. 3d). Furthermore, the glutamatergic pro-jections from the mPFC were found to innervate both CaMKIIα-positive projection neurons and GABAergic neurons in the vlPAG (see Supplementary Fig. 7c).

Role of mPFC–PAG connections in neuropathic pain. To deter-mine how these glutamatergic mPFC–vlPAG projections regulate nociceptive responses, AAV5-CamKIIα-ChR2(H134R)-eYFP was injected into layer 5 of prelimbic mPFC and an optic fiber was implanted in the vlPAG (Fig. 4a). It was found that activation of glutamatergic terminals in the vlPAG with blue light (473 nm, 20 Hz and 5–10 mW) inhibited both mechanical allodynia and thermal hyperalgesia on the ipsilateral paws of SNI mice, whereas there

was no effect on the contralateral paws or in sham-operated mice (Fig. 4b–e). In contrast, when virus was injected into the infralim-bic area of the PFC, blue light stimulation in the vlPAG did not affect mechanical hypersensitivity (see Supplementary Fig. 8a,b). These data are consistent with in vitro observations that neuro-pathic pain states are associated with decreased neuronal activity in the prelimbic, but not the infralimbic, cortex of the mPFC which projects to the PAG7,8,13. It was also investigated whether expression of the inhibitory opsin NpHR3.0 in the mPFC–vlPAG projections could enhance pain responses. As expected, delivery of yellow light (589 nm, solid light, 5 mW) in the vlPAG reduced the mechanical threshold on the ipsilateral side of SNI mice (Fig. 4f,g), thus exac-erbating neuropathic pain. Taken together, these data support a mechanism in which enhanced feedforward inhibition of PNs in the mPFC reduces the activity of mPFC–vlPAG projections and exacer-bates pain responses, whereas boosting the firing of these projection neurons via optogenetics has analgesic properties.

a

f

0

10

20

30

***

Contra Ipsi

Light off Light on

*

0

5

10

15

***

Contra Ipsi

Light off Light on

***

e

g

AAV-CaMKIIα-Arch3.0-eYFPAAV-CaMKIIα-NpHR3.0-eYFP

PFC

BLA

0

10

20

30

Contra Ipsi

Light off Light on

* *

b

0

PW

thre

shol

d (g

)

5

10

15

Contra Ipsi

Light off Light on

*** *

c

AAV-CaMKIIα-ChR2-eYFP

PFC

BLA

d

h

0

1

2

3

4

Sham SNI

*

Noc

ifens

ive

resp

onse

(s)

0

20

40

60

80

100

Contra Ipsi

***

Per

cent

age

with

draw

al

i

****

Light off Light on Light off Light on

Cold allodynia Dynamic plantar

20

40

80

Sham SNI

** ***

Light off Light on

Tim

e (p

erce

ntag

e in

brig

ht s

ide)

60

PEA

(8) (8) (8) (8) (6)

(8) (8) (8) (8) (8) (8) (8) (8)

(6) (6)

(6)

(4)

(4)

(6)

(7)(4) (5)

(4) (5)

(4) (4)

(4)

(5)

PW

late

ncy

(s)

PW

thre

shol

d (g

)

PW

late

ncy

(s)

Fig. 2 | effects of optogenetic manipulation of BLA inputs into the prelimbic mPFC. a, Experimental configuration showing AAV-CaMKIIα-ChR2-eYFP injection into the BLA and optic cannula implantation in the prelimbic mPFC. b,c, Mechanical paw withdrawal (PW) threshold (b; P

ArticlesNature NeuroscieNce

Signal processing of mPFC inputs into the vlPAG region. To determine how glutamatergic inputs from the mPFC are processed in the vlPAG, electrophysiological recordings were performed using vlPAG brain slices in PrL AAV-ChR2 injected mice (Fig. 4h,i,j). Optical activation of these glutamatergic inputs from the mPFC triggered excitatory synaptic events in both glutamatergic (Fig. 4i)

and GABAergic (Fig. 4j) cells within the vlPAG. These were found to be exclusively monosynaptic because light-evoked synaptic events could be rescued with 4-AP after TTX treatment. Such light-evoked glutamatergic events were observed in approximately 40% and 36% of glutamatergic and GABAergic cells examined in the vlPAG, respectively (Fig. 4k), indicating that not all cells in

a

c

d

b

PFCPAG

AAV5-ChR2

PFCPAG

CTB555

PFCPAG

CTB488

Vglut2-cre-tdTomato

GAD2-cre-tdTomato Merge

MergeCTB488

PFC

L5

L2/3

vl

dl/l

dm

vlPAG

L1

PFC

L2/3 L5

PFC PAG

AAV9-synapto-physin Cav2-cre

vlPAG

PFC

L5L1 L2/3

L1 L2/3 L5

CTB488

PFC

L1 L2/3 L5

Fig. 3 | Circuit projection from the prelimbic mPFC to the vlPAG. a, AAV5-CaMKIIα-ChR2 (H134R)-eYFP was injected into the prelimbic mPFC. Images show eYFP virus expression in the mPFC (left) and the vlPAG (middle and right, n = 4 mice). Scale bars, 200 μm (left), 50 μm (middle) and 20 μm (right). b, CTB555 was injected into the vlPAG. Images show CTB555 expression in the vlPAG (left) and retrograde transport to layer 5 of prelimbic mPFC (middle and right, n = 4 mice). Scale bars, 200 μm (left), 50 μm (middle) and 20 μm (right). dm, dorsomedial; dl, dorsolateral; l, lateral; vl, ventrolateral. c, AAV9-hEF1α-DIO-synaptophysin-mCherry virus was injected into the prelimbic mPFC, and Cav2-cre virus was injected into the vlPAG. Expression of mCherry is observed in PFC cell bodies (left), as well as at vlPAG terminals (arrows, middle and right, n = 4 mice). Scale bars, 20 μm (left), 50 μm (middle) and 20 μm (right). d, CTB488 was injected into the vlPAG of Vglut2-cre::Ai9 and GAD2-cre::Ai9 mice to determine the identity of projection neurons from the mPFC to the vlPAG. Images show that glutamatergic (arrows in upper panel), but not GABAergic (lower panel), neurons project from the prelimbic mPFC to the vlPAG (n = 3 mice). Scale bar, 20μm. L, layer.




the vlPAG receive direct inputs from the mPFC. Consistent with these slice electrophysiology data, in vivo optogenetic activation of mPFC-to-vlPAG projections resulted in an increase in c-fos expres-sion in glutamatergic neurons within the vlPAG (see Supplementary Fig. 9a,c). A similar pattern was also observed in the vlPAG on opto-genetic inhibition of BLA inputs into the mPFC (see Supplementary

Fig. 10a,b,c), supporting the concept that further light-induced reduction of feedforward inhibition of mPFC outputs is transmit-ted to the vlPAG. On the other hand, activation of glutamatergic projections from the mPFC to the vlPAG yielded an unexpected decrease in c-fos expression in GABAergic cells (see Supplementary Fig. 9b,d). The most salient explanation of this finding may be the

a b c

PFC PAG

Light

AAV-CaMKIIα-ChR2-eYFPAAV-CaMKIIα-NpHR-eYFP

SNI Sham

0

2

4

6

8

10

Contra Ipsi

**** ****Light onLight off

Ipsi

10

8

6

4

2

0Contra

SNI Sham

Ipsi

10

8

6

4

2

0

Light off Light on10

Ipsi

8

6

4

2

0Contra Contra

d

** **

Ipsi

20

15

10

5

0

Ipsi

20

15

10

5

0Contra

Contra

SNI

Sham

Light off Light on

g

50 ms

40 p

A

PN

Control

TTX

TTX + 4-AP

PAGPFC input

50 ms

40 p

A

GABA

Control

TTX

TTX + 4-AP

PAGPFC input

GABAPN

Recording sites in vlPAG

PN (2

5)

GABA

(47)

GABA

(27)

Per

cent

age

resp

ondi

ng c

ells

100

0

EPSC responderEPSC nonresponderIPSC nonresponder

h

i

NS

NS

NS

0

0.2

0.4

0.6

0.8

1.0Light offLight on

**A

rea

unde

r th

e cu

rve

–1

0

1

2

3

4

5 Light off Light on

Z-s

core

1 2 3 4 5 6

Time (s)

PFC PAG

BLA

Arch3.0-eYFPYellow

light

AAV-GCaMP6f-tdTomato

Fiber photometryprobe

0 200 400 600 800 1,000 1,200–0.02

–0.01

0

0.01

0.02

0.03

0.04

ΔF

/F

NL, no stim YL, no stimNL, VF stim YL, VF stim

Time (s)

j k l

m

e f

n o

(8) (8) (8)(8) (8)

(8)

(8) (8)(12) (12) (12) (12)

(11) (11) (11) (11) (4) (4) (4) (4) (5) (5) (5) (5)

(3)

PW

thre

shol

d (g

)

PW

thre

shol

d (g

)

PW

late

ncy

(s)

PW

late

ncy

(s) *** ****

PW

thre

shol

d (g

)

PW

thre

shol

d (g

)

Fig. 4 | effects of optogenetic manipulation of the prelimbic mPFC–vlPAG circuit. a, Schematic diagram for optogenetic manipulation of the prelimbic mPFC–vlPAG circuit with virus injection into the mPFC and optical cannula implantation in the vlPAG. b–e, Blue light activation of the prelimbic mPFC–vlPAG inputs increases paw withdrawal (PW) threshold (b; P

ArticlesNature NeuroscieNceexistence of two populations of GABAergic cells—one that receives direct glutamatergic inputs from the mPFC with activity that would be expected to increase on ChR2 stimulation, and a second group of cells that might receive feedforward GABAergic inhibition, thus accounting for an overall decrease in c-fos expression. In slice recordings, among 27 GABAergic cells that were tested for light-evoked synaptic events, none showed an eIPSC (Fig. 4k), suggesting an absence of functional GABAA-receptors in these GABA neurons. However, GABAergic cells did respond with an outward current, and a decrease in input resistance to bath application of the GABAB-receptor agonist baclofen (see Supplementary Fig. 11a,b,c), suggest-ing that such a putative feedforward inhibition of GABA cells could potentially occur via volume transmission (see Supplementary Fig. 11d). However, considerable additional validation of such a possible mechanism is required before firm mechanistic conclu-sions can be drawn. Suffice it to say, mPFC inputs into the vlPAG exert differential effects on GABAergic and glutamatergic cells.

To further examine changes in neuronal activity within the vlPAG in vivo in real time, the activity of GABAergic cells was investigated in freely moving nerve-injured mice using fiber photometry (Fig. 4l). GCaMP6f and tdTomato were bicistronically and selectively expressed in GABAergic vlPAG neurons using GAD2-cre mice, in which Arch3.0 was separately expressed in the BLA-to-mPFC pro-jections (see Supplementary Fig. 12a,b). Bulk fluorescence, a mea-sure of population activity, was monitored in response to hindpaw stimulation with a von Frey filament. As shown in Fig. 4m–o, stimu-lation of the hindpaw triggered time-locked transient increases in calcium fluorescence in vlPAG GABAergic cells. Concomitant opto-genetic inhibition of BLA–mPFC inputs (which boosts mPFC out-put to the vlPAG; see Supplementary Fig. 10) led to a reduction in evoked calcium transients in the vlPAG, which is consistent with the reduction in c-fos staining of GABAergic cells (see Supplementary Fig. 9b,d). These data are important in that they provide direct real-time confirmation of activity changes in the vlPAG that are induced by alteration of upstream BLA inputs into the mPFC.

Ablation of mPFC inputs into the vlPAG. To rule out the pos-sibility that the analgesic effects of inhibiting BLA inputs into the mPFC might involve other (parallel) pathways, projections within this pathway were selectively ablated using dual injection of AAV-mCherry-flex-dtA and Cav2-cre virus. This leads to expression of diphtheria toxin subunit A (dtA) only in neurons connecting the two injected brain regions, and thus their ablation. SNI mice expressing Arch3.0 in the BLA-to-mPFC projections were injected with AAV-mCherry-flex-dtA in the prelimbic mPFC and Cav2-cre virus in the vlPAG, and then a fiberoptic cannula was implanted in the prelimbic mPFC (Fig. 5a). This procedure greatly decreased the numbers of neurons in layer 5 of the prelimbic mPFC (Fig. 5b,c) and prevented the analgesic effects of inhibiting BLA–mPFC inputs (Fig. 5d), indicating that the mPFC–vlPAG projection is a neces-sary downstream pathway for the control of pain responses by the BLA–mPFC.

Link between BLA–mPFC–PAG connections and descending modulation of pain. The PAG is known to be an important relay station for descending modulation of pain through projections to the spinal cord via the rostroventral medulla (RVM) and the locus coeruleus (LC)19. Injection of retrotracers into these two brain structures, followed by an analysis of vlPAG slices, revealed that the vlPAG sends GABAergic and glutamatergic projections to both the LC (see Supplementary Fig. 12c,d) and the RVM (see Supplementary Fig. 12e,f). The dtA-mediated ablation of either the projections from the vlPAG to the LC or those from the vlPAG to the RVM prevented the analgesic effects associated with opto-genetic inhibition of BLA–mPFC inputs (Fig. 5d), indicating that both pathways are required for top-down control of neuropathic

pain responses by the mPFC. It was noted that a bifurcation of pain signals from the PAG to the LC and RVM has also been implicated in opioid analgesia26.

The RVM and LC are known to send descending serotoninergic and nordadrenergic projections to the spinal cord, respectively27. To ascertain how these pathways are involved in the BLA-mediated modulation of pain signals, optogenetics was combined with in vivo pharmacology (Fig. 5e). AAV-Arch3.0 was injected into the BLA and a fiberoptic cannula was implanted into the prelimbic mPFC of SNI mice, as described above. The mechanical withdrawal thresh-old was measured in the absence and presence of Arch3.0-mediated inhibition of BLA-to-mPFC inputs. These paradigms were then repeated after intrathecal delivery of vehicle, the serotonin (5-HT3) receptor antagonist ondansetron, the 5HT1/2-receptor antagonist metergoline, or the α2A-adrenergic-receptor antagonist BRL-44408 at effective doses that were determined in a separate series of in vivo dose–response experiments (not shown). As clearly evident from the data presented in Fig. 5f, neither intrathecal delivery of vehicle nor blocking of 5-HT3 receptors could prevent the analgesic effects of yellow light-mediated inhibition of BLA–mPFC projections. In contrast, blocking of either 5HT1/2- or α2A-adrenergic receptors abolished the effects of inhibiting BLA-to-mPFC inputs, consistent with the ablation experiments described above. None of the com-pounds affected hyperalgesia of SNI mice in conditions in which the BLA–mPFC pathway was not optogenetically inhibited to boost the mPFC-to-vlPAG inputs (see Supplementary Fig. 13). This altogether suggests a loss of serotoninergic and adrenergic control during neu-ropathic states, which precludes further pro-nociceptive effects by 5HT1/2- or α2A-adrenergic-receptor antagonists. Finally, as an addi-tional line of evidence, and consistent with the pharmacological studies, ablation of noradrenergic descending projections with intra-thecally delivered N-(2-chloroethyl)-N-ethyl-2-bromobenzylamine (DSP4) also extinguished the analgesia produced by yellow light in the prelimbic mPFC (Fig. 5f, and see Supplementary Fig. 14). It was acknowledged that DSP4 may ablate noradrenergic projec-tions originating from other brainstem regions that may contribute to descending modulation28. However, the observation that deletion of vlPAG-to-LC projections fully prevented the analgesic effects of inhibiting BLA-to-mPFC inputs suggests that yellow light modula-tion in the mPFC does not engage DSP4-sensitive pathways other than the LC in these optogenetic experiments. Collectively, the data of the present study thus indicate that BLA inputs regulate neuro-pathic pain by altering the top-down control of the mPFC over par-allel descending pain pathways via vlPAG.

DiscussionThe processing of pain signals in the brain is immensely complex, and involves both sensory and affective components. Many brain regions, including the thalamus29, anterior cingulate cortex30,31 and PFC1,32 have been implicated in processing pain-related informa-tion. These regions interact with brain structures that are involved in pathologies such as anxiety, reward seeking and depression, thus further complicating the analysis of brain circuits that process pain-related information33,34. In the present study the functional connec-tomics of a linear circuit have been described that is involved in the development of neuropathic pain. The data reveal that peripheral nerve injury leads to a selective augmentation of BLA inputs into GABAergic interneurons in the prelimbic area of the PFC as a result of weakened endocannabinoid signaling. Indeed, the strengthening of these synaptic inputs can be observed in PFC slices that contain the axonal tracts, indicating that this dysregulation resides within the PFC itself and is supported by the overall reduction in CB1-receptor expression in this region. Although quantitative PCR and western blot analyses do not allow determination of whether CB1-receptor density is decreased in specific BLA-originating presyn-aptic terminals, the electrophysiological analysis is consistent with




such a mechanism. A role for the mPFC cannabinoid system in neu-ropathic pain is supported by a recent study showing that sublingual administration of the CB1 agonist δ9-tetrahydrocannabinol medi-ated analgesia by reducing the connectivity of the dlPFC (the equiv-alent of which is the mPFC in rodents) and chronic pain network components in neuropathic pain patients35. The selective strength-ening of glutamatergic inputs into PVINs leads to an overall inhi-bition of layer 5 pyramidal cell output toward the vlPAG. Within the vlPAG, this leads to reduced activity of glutamatergic neurons and increased activity of GABAergic neurons. Increased inhibition together with less excitation is expected to reduce serotoninergic and noradrenergic output from the RVM and LC to the spinal dorsal horn, respectively, thus leading to compromised descending modu-lation of ascending nociceptive inputs (see Supplementary Fig. 15). It is possible that the loss of descending adrenergic modulation

could give rise to a progressive cycle in which increased inputs into the BLA sustain a tonic inhibition of mPFC function. Specific opto-genetic inhibition of BLA inputs into the mPFC weakens the feed-forward inhibition, thus boosting mPFC inputs into the vlPAG and its downstream projections to the RVM, LC and spinal cord, culmi-nating in analgesia. Ablation experiments of the mPFC-to-vlPAG projections, and the downstream ablation between vlPAG and LC or RVM fully prevented the analgesic effects of optogenetic modula-tion of the BLA-to-mPFC input, suggesting that BLA inputs into the mPFC do not modulate pain behavior through mPFC-originating pathways that bypass the vlPAG. It is interesting to reiterate that these analgesic effects required the functionality of both branches of the descending pathway, as if the neuronal circuitry within the spinal cord acted as a coincidence detector for both serotoniner-gic and noradrenergic inputs. Additional work will be required to

dtAa b

d

e

PFC PAG

Cav2-cre

BLA

Arch3.0-eYFP

AAV-mCherry-flex-dtA

Yellowlight

dtA + Cav2

c

BLA

Arch3.0-eYFP

Intrathecal injectionof drugs

L5PAG

PFC

Yellowlight

0

2

4

6

8

10 ** * **NS

NS

Light offLight on

Light off – no drugLight on – drug

PBS Ond Meterg BRL PBS DSP4

NS

0

2

4

6

8

10 Light off Light on

**

Sham

SNI

NS NSNS

0

20

40

60

80**

dtA + Cav2dtA

No.

of p

roje

ctio

n ne

uron

sin

pre

limbi

c P

FC

f

(3)

(4)

(7) (10) (7) (6) (3)

(7) (5) (5) (11) (4) (4)

PW

thre

shol

d (g

)

dtA

only

PFC-

PAG

PAG-

RVM

PAG-

LC

PW

thre

shol

d (g

)dtA + Cav2dtA

Fig. 5 | optogenetic and pharmacological manipulation of the BLA–mPFC–vlPAG–spinal cord circuit and its effect on mechanical allodynia. a, Schematic diagram for dtA-mediated ablation of mPFC–vlPAG projections. b, Images for mCherry-labeled neurons in layer 5 of the prelimbic mPFC with dtA (left) versus dtA + Cav2 (right) injection (n = 3 mice for the dtA group and n = 4 mice for the dtA + Cav2 group). Scale bar, 20 μm. c, Quantification of neuronal loss in layer 5 of the prelimbic mPFC (P = 0.0069, t = 4.415). d, Effect of dtA-mediated ablation of mPFC-to-vlPAG projections, vlPAG-to-RVM projections and vlPAG–LC projections on the analgesic effects of Arch3-mediated inhibition of the BLA–mPFC circuit (P

ArticlesNature NeuroscieNcedetermine whether blockers of GABAergic, opioidergic or glyciner-gic signaling in the spinal cord are also effective in interfering with yellow light-mediated modulation of BLA–mPFC inputs.

Altogether, the data presented in the present study constitute an in-depth analysis of the functional connectomics of the BLA–mPFC–PAG–spinal cord pathway, and establish a causal link between the dysregulation of this circuit and sensory and affec-tive components of pain. How this pathway interacts with other brain structures involved in the regulation of pain signals will be an important area of future research, and it will be important to deter-mine how the pathway examined in the present study interacts with other circuits involved in the cortical modulation of pain signals. Furthermore, it is fully acknowledged that other pathways such as anterior cingulate cortex–spinal cord and S1/S2–spinal cord con-nections contribute to the processing of neuropathic pain signals36,37 and may play an important role in descending pain modulation; it is possible that different pathways are engaged in different types of acute and chronic pain states. Nevertheless, it must be noted that the present findings are consistent with observations that enhanced BLA inputs into the mPFC have been associated with pain-related cognitive deficits in a rat model of arthritis38. Moreover, two recent human functional magnetic resonance imaging studies showed a communication of pain signals between the PFC and the PAG39,40, thus supporting the relevance of these findings beyond the rodent brain. In this context, the present data hint at possible strategies for combatting mechanical allodynia and thermal hyperalgesia in neuropathic pain sufferers via noninvasive modulation of the BLA–mPFC–vlPAG neuraxis.

online contentAny methods, additional references, Nature Research reporting summaries, source data, statements of code and data availability and associated accession codes are available at https://doi.org/10.1038/s41593-019-0481-5.

Received: 18 April 2019; Accepted: 25 July 2019; Published: xx xx xxxx

References 1. Lorenz, J., Minoshima, S. & Casey, K. L. Keeping pain out of mind: the

role of the dorsolateral prefrontal cortex in pain modulation. Brain 126, 1079–1091 (2003).

2. Apkarian, A. V. et al. Chronic back pain is associated with decreased prefrontal and thalamic gray matter density. J. Neurosci. 24, 10410–10415 (2004).

3. Devoize, L., Alvarez, P., Monconduit, L. & Dallel, R. Representation of dynamic mechanical allodynia in the ventral medial prefrontal cortex of trigeminal neuropathic rats. Eur. J. Pain. 15, 676–682 (2011).

4. Seminowicz, D. A. & Moayedi, M. The dorsolateral prefrontal cortex in acute and chronic pain. J. Pain 18, 1027–1035 (2017).

5. Taylor, J. J., Borckardt, J. J. & George, M. S. Endogenous opioids mediate left dorsolateral prefrontal cortex rTMS-induced analgesia. Pain 153, 1219–1225 (2012).

6. Seamans, J. K., Lapish, C. C. & Durstewitz, D. Comparing the prefrontal cortex of rats and primates: insights from electrophysiology. Neurotox. Res. 14, 249–262 (2008).

7. Zhang, Z. et al. Role of prelimbic GABAergic circuits in sensory and emotional aspects of neuropathic. Pain Cell Rep. 12, 752–759 (2015).

8. Lee, M. et al. Activation of corticostriatal circuitry relieves chronic neuropathic pain. J. Neurosci. 35, 5247–5259 (2015).

9. Little, J. P. & Carter, A. G. Synaptic mechanisms underlying strong reciprocal connectivity between the medial prefrontal cortex and basolateral amygdala. J. Neurosci. 33, 15333–15342 (2013).

10. McGarry, L. M. & Carter, A. G. Prefrontal cortex drives distinct projection neurons in the basolateral amygdala. Cell Rep. 21, 1426–1433 (2017).

11. Cheriyan, J., Kaushik, M. K., Ferreira, A. N. & Sheets, P. L. Specific targeting of the basolateral amygdala to projectionally defined pyramidal neurons in prelimbic and infralimbic cortex. eNeuro 3, https://doi.org/10.1523/ENEURO.0002-16.2016 (2016).

12. Collins, D. P., Anastasiades, P. G., Marlin, J. J. & Carter, A. G. Reciprocal circuits linking the prefrontal cortex with dorsal and ventral thalamic nuclei. Neuron 98, 366–379 (2018).

13. Cheriyan, J. & Sheets, P. L. Altered excitability and local connectivity of mPFC-PAG neurons in a mouse model of neuropathic pain. J. Neurosci. 38, 4829–4839 (2018).

14. Woodhams, S. G., Chapman, V., Finn, D. P., Hohmann, A. G. & Neugebauer, V. The cannabinoid system and pain. Neuropharmacology 124, 105–120 (2017).

15. Fortin, D. A. & Levine, E. S. Differential effects of endocannabinoids on glutamatergic and GABAergic inputs to layer 5 pyramidal neurons. Cereb. Cortex 17, 163–174 (2007).

16. Caballero, A. & Tseng, K. Y. Association of cannabis use during adolescence, prefrontal CB1 receptor signaling, and schizophrenia. Front. Pharmacol. 3, 101 (2012).

17. Song, J. et al. Crucial role of feedback signals from prelimbic cortex to basolateral amygdala. Sci. Adv. 5, eaat3210 (2019).

18. Reynolds, D. V. Surgery in the rat during electrical analgesia induced by focal brain stimulation. Science 164, 444–445 (1969).

19. Basbaum, A. I. & Fields, H. L. Endogenous pain control systems: brainstem spinal pathways and endorphin circuitry. Annu. Rev. Neurosci. 7, 309–338 (1984).

20. Samineni, V. K. et al. Divergent modulation of nociception by glutamatergic and GABAergic neuronal subpopulations in the periaqueductal gray. eNeuro 4, https://doi.org/10.1523/ENEURO.0129-16.2017 (2017).

21. Hardy, S. G. & Leichnetz, G. R. Frontal cortical projections to the periaqueductal gray in the rat a retrograde and orthograde. Neurosci. Lett. 23, 13–17 (1981).

22. An, X., Bandler, R., Ongür, D. & Price, J. L. Prefrontal cortical projections to longitudinal columns in the midbrain periaqueductal gray in macaque monkeys. J. Comp. Neurol. 401, 455–479 (1998).

23. Ferreira, A. N., Yousuf, H., Dalton, S. & Sheets, P. L. Highly differentiated cellular and circuit properties of infralimbic pyramidal neurons projecting to the periaqueductal gray and amygdala. Front. Cell Neurosci. 9, 161 (2015).

24. Franklin, T. B. et al. Prefrontal cortical control of a brainstem social behavior circuit. Nat. Neurosci. 20, 260–270 (2017).

25. Rozeske, R. R. et al. Prefrontal-periaqueductal gray-projecting neurons mediate context fear discrimination. Neuron 97, 898–910 (2018). e896.

26. Kim, J. H. et al. Yin-and-yang bifurcation of opioidergic circuits for descending analgesia at the midbrain of the mouse. Proc. Natl Acad. Sci. USA 115, 11078–11083 (2018).

27. Ossipov, M. H., Dussor, G. O. & Porreca, F. Central modulation of pain. J. Clin. Invest. 120, 3779–3787 (2010).

28. Nam, H. & Kerman, I. A. Distribution of catecholaminergic presympathetic-premotor neurons in the rat lower brainstem. Neuroscience 324, 430–445 (2016).

29. Craig, A. D., Bushnell, M. C., Zhang, E. T. & Blomqvist, A. A thalamic nucleus specific for pain and temperature sensation.pdf. Nature 372, 770–773 (1994).

30. Johansen, J. P., Fields, H. L. & Manning, B. H. The affective component of pain in rodents: direct evidence for a contribution of the anterior cingulate cortex. Proc. Natl Acad. Sci. USA 98, 8077–8082 (2001).

31. Bliss, T. V., Collingridge, G. L., Kaang, B. K. & Zhuo, M. Synaptic plasticity in the anterior cingulate cortex in acute and chronic pain. Nat. Rev. Neurosci. 17, 485–496 (2016).

32. Ohara, P. T., Vit, J. P. & Jasmin, L. Cortical modulation of pain. Cell. Mol. Life Sci. 62, 44–52 (2005).

33. Barthas, F. et al. The anterior cingulate cortex is a critical hub for pain-induced depression. Biol. Psychiatry 77, 236–245 (2015).

34. Jahn, A., Nee, D. E., Alexander, W. H. & Brown, J. W. Distinct regions within medial prefrontal cortex process pain and cognition. J. Neurosci. 36, 12385–12392 (2016).

35. Weizman, L. et al. Cannabis analgesia in chronic neuropathic pain is associated with altered brain connectivity. Neurology 91, 1285–1294 (2018).

36. Chen, T. et al. Top-down descending facilitation of spinal sensory excitatory transmission from the anterior cingulate cortex. Nat. Commun. 9, 1886 (2018).

37. Liu, Y. et al. Touch and tactile neuropathic pain sensitivity are set by corticospinal projections. Nature 561, 547–550 (2018).

38. Ji, G. et al. Cognitive impairment in pain through amygdala-driven prefrontal cortical deactivation. J. Neurosci. 30, 5451–5464 (2010).

39. Roy, M. et al. Representation of aversive prediction errors in the human periaqueductal gray. Nat. Neurosci. 17, 1607–1612 (2014).

40. Tinnermann, A., Geuter, S., Sprenger, C., Finsterbusch, J. & Büchel, C. Interactions between PFC-PAG-spinal cord mediate value effects in nocebo hyperalgesia. Science 358, 105–108 (2017).

AcknowledgementsThis work was supported by a Foundation Grant to G.W.Z. from the Canadian Institutes of Health Research, and by the Canada–Israel Health Research Initiative, jointly funded by the Canadian Institutes of Health Research, the Israel Science Foundation, the


https://doi.org/10.1038/s41593-019-0481-5https://doi.org/10.1038/s41593-019-0481-5https://doi.org/10.1523/ENEURO.0002-16.2016https://doi.org/10.1523/ENEURO.0002-16.2016https://doi.org/10.1523/ENEURO.0129-16.2017http://www.nature.com/natureneuroscience

Articles Nature NeuroscieNceInternational Development Research Centre and the Azrieli Foundation. G.W.Z. is a Canada Research Chair in Molecular Neuroscience. V.M.G. is supported through the Vi Riddell Program in Pediatric Pain of the Alberta Children’s Hospital Research Institute. S.H. is supported by a studentship from Alberta Innovates and a University of Calgary Eyes-High studentship. We thank T. Fuzesi and the Cumming School of Medicine Optogenetics facility for technical assistance with fiber photometry.

Author contributionsZ.Z., J.H. and G.W.Z. conceived the project. J.H., Z.Z. and V.M.G. designed and performed the experiments, analyzed the data, prepared the figures and contributed to the writing. L.C. performed the tissue extractions and immunostaining experiments. I.A.S. performed the western blotting. Z.Z., D.W. and S.H. performed the electrophysiology experiments. G.W.Z. directed and supervised the study and co-wrote the manuscript. C.R. and K.D. provided the reagents and discussion, and edited the manuscript.

Competing interestsThe authors declare no competing interests.

Additional informationSupplementary information is available for this paper at https://doi.org/10.1038/s41593-019-0481-5.

Reprints and permissions information is available at www.nature.com/reprints.

Correspondence and requests for materials should be addressed to Z.Z. or G.W.Z.

Peer review information: Nature Neuroscience thanks C. Woolf and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© The Author(s), under exclusive licence to Springer Nature America, Inc. 2019


https://doi.org/10.1038/s41593-019-0481-5https://doi.org/10.1038/s41593-019-0481-5http://www.nature.com/reprintshttp://www.nature.com/natureneuroscience

ArticlesNature NeuroscieNceMethodsAnimals. Animal experiments were approved by the Animal Care Committee of the University of Calgary. Male C57BL/6J wild-type and transgenic mice were purchased from Jackson laboratories. Cre transgenic mice (PV-cre: B6;129P2-Pvalbtm1(cre)Arbr/J, Vglut2-cre: Slc17a6tm2(cre)Lowl/J and GAD2-cre: Gad2tm2(cre)Zjh/J) were crossed with tdTomato reporter mice (Ai9: B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J) for labeling of parvalbumin, glutamatergic and GABAergic neurons, respectively, in imaging and brain-slice (mPFC and vlPAG) electrophysiological recordings. Mice were kept on a 12-h light/dark cycle (lights on at 7am) in a holding room kept at a temperature of 23 ± 1 °C; food and water were freely available.

Electrophysiology. Acute coronal forebrain slices (300 μm) containing prelimbic mPFC or coronal brainstem slices (260 μm) containing the vlPAG were obtained from C57BL/6J or transgenic mice. Briefly, coronal sections were cut using a Vibratome (Leica VT1200S, Leica Biosystems) in an ice-cold sucrose- or N-methyl-d-glucamine-based solution. Brain slices were first incubated at 33.5 °C for 50 min, then transferred to room temperature in normal external solution for at least 1 h before recording. Whole-cell patch clamp recordings were performed using MultiClamp 700B and Digidata 1440 A (Molecular Devices). Normal external solution contained (mM): NaCl 120, NaH2PO4 1.25, NaHCO3 26, glucose 25, CaCl2 2.5, KCl 2.5, MgSO4·7H2O 1.3. The intracellular solution for current clamp was potassium gluconate based, and the solution for voltage clamp was cesium methanesulfonate based (supplemented with 5 mM QX314). For eEPSC and eIPSC recordings, Vh was held at –70 mV (calculated from Cl– reversal potential) and 0 mV (reversal potential for EPSC), respectively. TTX (1 μM) was used to block action potential-based synaptic transmission. Both TTX (1 μM) and 4-AP (100 μM) were used to re-establish monosynaptic transmission under blue light. The eEPSC recordings in Fig. 1h,i were performed without GABA blockers in the extracellular solution. The recorded eEPSCs in these experiments had short constant latency with low jitters, and followed 20-Hz stimulation (not shown), indicating the monosynaptic nature of these eEPSCs. To record both eEPSCs and disynaptic eIPSCs in the prelimbic area of the mPFC in Fig. 1g, GABA blockers were not typically used, but bicuculline methiodide (10 μM) was added in the extracellular solution in experiments shown in Fig. 1f. For experiments in Figs 1d,e and 4h,i,j,k, both TTX and 4-AP were used to restore glutamatergic, monosynaptic, evoked current9. In some experiments eEPSCs were verified using 6,7-dinitroquinoxaline-2,3-dione (DNQX, 10 μM) and AP5 (50 μM), added in extracellular solution at the end of recording. For experiments in Supplementary Fig. 11a, b, DNQX (20 μM), AP5 (50 μM) and bicuculline (10 μM) were added to the extracellular solution. Mice injected with AAV5-CamKIIα-ChR2-eYFP virus in the BLA (mPFC slice recording) or in the prelimbic area of mPFC (vlPAG slice recording) were allowed to express ChR2 in the axon terminal area (PrL or vlPAG, respectively) for at least 8–10 weeks before brain slice recording. Both diode-pumped solid-state (DPSS) lasers (Laserglow Technologies) and light-emitting diodes (LEDs; Thorlabs) were used to illuminate brain slices for recording. The recording temperature was 33 ± 1 °C. All chemicals were purchased from Tocris or Sigma.

Stereotaxic surgeries and injections. Mice (~8 weeks old) were anesthetized with isoflurane (4–5% for induction, 1.5–2.0% for maintenance) in a stereotaxic frame (Stoelting). Viral titers of adeno-associated virus (AAV) vectors were between 2 × 1012 and 9.9 × 1012 viral genomes per milliliter. To express opsins in brain areas of interest, AAV5-CaMKIIα-ChR2 (H134R)-eYFP (500–650 nl), AAV5-CaMKIIα-Arch3.0-eYFP (500–600 nl) or AAV5-CaMKIIα-NpHR3.0-eYFP (500–600 nl) was injected using an automatic nanoliter injector (Nanoject II, Drummond Scientific) positioned into the prelimbic area of the PFC (anteroposterior (AP) +1.8–1.9 mm; right (R) −0.5 mm; dorsoventral (DV) −2.1 mm), BLA (AP −1.3 mm; R −3.4 mm; DV −4.9 mm) or infralimbic PFC (AP +1.9 mm; R −0.5 mm; DV −2.5-2.6 mm). In some cases the AAV8 serotype was used in place of AAV5, with identical results. To retrogradely trace the PFC–PAG long projection, cholera toxin subunit B (0.1%, w/v, 500-600 nl, CTB488 and CTB555, Thermo Fisher Scientific) or retrobeads (100–150 nl, Lumafluor Inc.) were injected into the vlPAG (AP −4.6-4.7 mm; R −0.5 mm; DV −3.3 mm) and allowed 5–7 d for sufficient retrograde transport41,42. To map the vlPAG–LC and vlPAG–RVM projections, CTB488 (0.1%, w/v, 600 nl) was injected into the LC (AP −5.4 mm; R −0.9 mm; DV −4.0 mm) and RVM (AP −5.8 mm; R 0 mm; DV −6.2 mm), respectively. To further confirm the selective projection from the prelimbic mPFC to the vlPAG, AAV9-hEF1α-DIO-synaptophysin-mCherry virus (200 nl) was injected into the prelimbic mPFC (AP +1.9 mm; R −0.5 mm; DV −2.2 mm) and Cav2-cre virus (100–150 nl)43 was injected into the vlPAG (AP −4.6–4.7 mm; R −0.5 mm; DV −3.3 mm).

AAV-ChR2, AAV-Arch3.0 and AAV-NpHR3.0 viruses were obtained from either the Deisseroth lab or the University of North Carolina Vector Core. Cav2-cre virus was obtained from Dr E. J. Kremer (Montpellier, France) and AAV9-hEF1α-synaptophysin-mCherry virus was obtained from Dr R. L. Neve (Massachusetts Institute of Technology, Cambridge, MA, USA). Injections were performed using a digital stereotaxic frame with a glass capillary. The viral suspensions, CTB or retrobeads were delivered at 50 nl min–1. After completing the injection, the injection capillary remained in position for 5 min and then was raised 100 μm with an additional 10 min wait to allow for the virus/CTB/retrobeads to diffuse

at the injection site, and then the glass needle was slowly withdrawn. Viruses were allowed to express themselves for at least 6–8 weeks before optogenetic manipulation. For projection tracing, the numbers of mice given in the figure captions reflect successful repetitions. There were cases in which projections could not be detected in the target area (for example, due to off-target injections of viruses or retrograde tracers) and data from such animals are not included in these totals. For Fig. 1b, additional successful BLA-to-mPFC projections were visually observed during mPFC slice recordings, but not included in the total numbers of mice because no images were acquired.

SNI surgeries. Surgeries were performed on 7-week-old, wild-type (C57BL/6J) or transgenic mice as previously described44. Briefly, under isoflurane anesthesia a 0.5-cm skin incision was made on the left thigh to expose the three branches of the sciatic nerve (that is, common peroneal, tibial and sural nerves). Common peroneal and tibial nerves were tightly ligated with a 6/0 silk suture (Ethicon) and transected together. A 1-mm piece of the nerves was removed. Precaution was taken to avoid damage to the sural nerve. After surgery, the overlaying muscles and skin were separately closed with 6/0 silk and 4/0 Vicryl sutures, respectively. After surgery animals were placed in a new cage for recovery for at least 10–12 d before further brain cannula implantation or brain slices recordings. In sham-operated animals, exactly the same procedure was performed but without ligation and transection of the nerves. Most of the SNI mice displayed neuropathic pain response (reduced thermal withdrawal latency and mechanical threshold) 10–14 d after SNI surgery. Animals that did not exhibit neuropathic phenotype after 14 d were discarded without further experiments. Baseline measurements were taken before the surgeries.

Optogenetic manipulation and in vivo testing. For optogenetic manipulation, a fiberoptic cannula (2.5 mm ceramic ferrule, 2.0 mm length for the prelimbic mPFC and 4.0 mm length for the vlPAG, Thorlabs), with core diameter 200 µm, 0.39 numerical aperture (NA) fiber, was implanted into the right PFC (A: +1.9 mm; R −0.5 mm; DV −2.1 mm) or right vlPAG (AP −4.6-4.7 mm; R −0.5 mm; DV −3.3 mm) 10 d after virus injection. For sensory pain analysis, testing was carried out between 7 and 8 weeks after virus injections. Thermal withdrawal latency and mechanical withdrawal threshold were measured with a Hargreave’s Apparatus and a Dynamic Plantar Aesthesiometer (Ugo Basile), respectively. Mice were placed individually in Plexiglass chambers on top of a glass platform or a grid floor, and allowed to habituate for at least 90 min before testing. Mice were then tested at the ipsilateral hindpaws (nerve-injured side) and contralateral paws before light stimulation. After 30 min, mice were stimulated with blue or yellow light, beginning 3 min before pain assessments and with constant stimulation until the end of the testing session. Mechanical thresholds or thermal latencies were measured three times for each mouse. DPSS lasers (blue 473 nm and yellow 589 nm; Laserglow Technologies) with 5–15 mW at the fiber tip (S130C power sensor, Thorlabs) were used for all the behavioral testing.

Acetone evaporation test for cold allodynia. The acetone evaporation test was performed as previously described, but with minor modifications45. Acetone 20 µl was applied to the plantar surface of the left hindpaw using a 0.5-ml syringe. Mice were divided into four different groups: sham (light OFF), sham (light ON), SNI (light OFF) and SNI (light ON) and observed for 1 min after the acetone application. Nocifensive behavior (withdrawal, licking, flinching) was considered to be a positive response and quantified with a chronometer. Each paw was tested only once and the total duration of nocifensive responses recorded.

Dynamic tactile allodynia. Dynamic tactile allodynia testing was performed between 7 and 8 weeks after virus injection and assessed by lightly stroking the lateral external side of the surface of the injured hindpaw, in the direction from heel to toe with a von Frey monofilament (0.4 g) without bending the filament. After acclimatizing the mice for about 90 min in the observation chambers on top of a grid mesh, both contralateral and ipsilateral paws were tested 10 times, with intervals of 10–15 s, and the percentage of withdrawal for each paw was considered as the pain response. Mice were tested before opto-light stimulation (light OFF) and after 45 min they were stimulated with yellow light on (light ON) for 3 min and both hindpaws were tested.

Open field test for anxiety and locomotor activity. An open field test was used to access anxiety-related behavior and locomotor activity. The apparatus consists of a wooden box measuring 40 × 60 × 50 cm3 with a frontal glass wall46. The floor of the arena is divided into 12 equal squares and placed in a sound-free room. Animals were placed in the rear left square and allowed to explore freely for 6 min. Crossings reflect the number of grid lines that were crossed with all paws (crossing). Normalized data (periphery/center ratio) reflect anxiety-related behavior. The apparatus was cleaned with a 70% alcohol solution and dried after each individual mouse session.

PEA paradigm. A PEA paradigm was used to assess affective pain behavior. Mice were assessed for the time spent in the dark and bright sides of a rectangular enclosure (length × width × height: 65 × 21 × 30 cm3, with one side of the walls



Articles Nature NeuroscieNcepainted dark and the other side painted white), placed on top of a grid floor platform. Under normal conditions, the mice spend most of their time in the dark side. On stimulation of the ipsilateral paw with a von Frey filament (0.4 g) while at the dark side, the mice displayed PEA behavior (that is, moving to the bright side) which was exacerbated by peripheral nerve injury.

Fiber photometry in vivo recording. GAD2-cre: Gad2tm2(cre)Zjh/J) mice (~8 weeks old) were used. For in vivo recording, AAV5-CaMKIIα-Arch3.0-eYFP (500–600 nl) and AAV-EF1a-DIO-GCaMP6f-P2A-nls-tdTomato (600 nl) were injected into the BLA (AP −1.3 mm; R −3.4 mm; DV −4.9 mm) and vlPAG (AP −4.6 mm, R −0.5 mm, DV −3.3 mm), respectively. SNI surgery was performed ~2 weeks after virus injection. Fiberoptic calcium-recording cannulas were implanted in the prelimbic area of the PFC (AP +1.9 mm; R −0.5 mm; DV −2.1 mm) for yellow laser stimulation and into the vlPAG (AP −4.6 mm; R −0.5 mm; DV −3.3 mm) ~6 weeks after virus injection, with 2.0-mm (200 μm core, 0.39 NA, ceramic ferrule) and 5.0-mm length fibers (400 μm core, 0.39 NA, SS ferrule) (Thorlabs), respectively. In vivo recording was performed ~8 weeks after virus injection. GCaMP6f virus was purchased from Addgene.

Mice were habituated for 3 d before fiber photometry. The photometry recording consisted of four sessions totaling 20 min: 5 min recording with light off and no stimulation, 5 min recording with light off but with 0.4 g von Frey stimulation, 5 min recording with yellow light but no von Frey stimulation, and 5 min recording with both yellow light and von Frey stimulation. The injured hindpaw was stimulated by a von Frey filament (0.4 g) every 30 s ten times. Calcium signals for the bulk fluorescence signals were acquired and analyzed with custom-written MATLAB software. Corrections for bleaching and motion were conducted before analysis. The calcium signals were recorded with two LED lights at 405 nm and 465 nm (30 μW, Doric)47. The 405 channel was used as the control channel and change in fluorescence (ΔF) was calculated as (465-nm signal − fitted 405-nm signal), and ΔF/F was calculated by dividing each point in ΔF by the 405-nm curve at that time point. A 6-s window around the von Frey stimulation point was analyzed, with the period −3 s before stimulus onset taken as baseline. The Z-score of a population of neurons was calculated using the formula: Z = (x − �x

I)/s.d. (where x = ΔF/F and �x

I = mean of ΔF/F for baseline)

Ablation of projections. Male C57 mice (~8 weeks old) were used in these studies. For PFC–vlPAG ablation, AAV5-CaMKIIα-Arch3.0-eYFP (600 nl), AAV-mCherry-flex-dtA (600 nl) and Cav2-cre virus (200 nl) were injected into the BLA (AP −1.3 mm; R −3.4 mm; DV −4.9 mm), the prelimbic area of the PFC (AP +1.9 mm; R −0.5 mm; DV: −2.1 mm) and the vlPAG (AP −4.6 mm; R −0.5 mm; DV −3.3 mm), respectively. SNI surgery was performed at ~2 weeks and cannula fibers were implanted in the prelimbic area of the PFC (AP +1.9 mm; R −0.5 mm; DV −2.1 mm) ~6 weeks after virus injection. Mechanical paw withdrawal threshold was tested ~8 weeks after virus injection with and without optogenetic manipulation of the PFC–vlPAG projection. Similarly, for vlPAG–LC and vlPAG–RVM ablation, AAV5-CaMKIIα-Arch3.0-eYFP (600 nl) and AAV-mCherry-flex-dtA (600 nl) were injected into the BLA (AP −1.3 mm; R −3.4 mm; DV −4.9 mm) and vlPAG (AP −4.6 mm; R −0.5 mm; DV −3.3 mm), respectively. Cav2-cre virus (200 nl) was injected into the LC (AP −5.4 mm; R −0.9 mm; DV −3.9-4.0 mm) or the RVM (AP −5.8 mm; R 0 or −0.1 mm; DV 5.9–6.2 mm). Optic fibers were implanted in the vlPAG (AP −4.6 mm; R −0.5 mm; DV −3.3 mm). Mechanical allodynia was measured ~8 weeks after virus injection.

For ablation of descending noradrenergic projections to the spinal cord, conscious mice were treated with 10 µl DSP4 (50 μg, intrathecally) or vehicle (phosphate-buffered saline (PBS), 10 µl, intrathecally), delivered over a period of 10–15 s per session, 2 weeks before measurements of mechanical withdrawal thresholds with and without optogenetic inhibition of BLA-to-mPFC inputs (AAV-Arch3.0 virus was injected into the BLA 7–8 weeks before these experiments). Briefly, mouse furs were shaved and they were manually restrained, then a 27-g half-needle attached to a PE20 poly(ethylene) tube connected to a 50-µl Hamilton microsyringe was inserted into the subarachnoid space between the L4 and L5 vertebrae. Mice were allowed to acclimatize for about 90 min in the observation chambers, and subsequently had the ipsilateral hindpaws tested before opto-light stimulation (light OFF). Some 45 min after the pre-opto-stimulation, mice were then stimulated with yellow light (light ON) and mechanical withdrawal thresholds were taken again.

Combined in vivo pharmacology and optogenetic stimulation. For assessments of mechanical withdrawal thresholds of mice injected intrathecally with antagonists and opto-stimulation, testing was also carried out between 7 and 8 weeks after BLA AAV-Arch3.0 virus injection. At the testing day, mice were placed individually in observation chambers on top of a grid platform to acclimatize for about 90 min, where they would subsequently have the ipsilateral hindpaws tested before opto-light stimulation and drug treatment. Some 30 min after the pre-drug delivery and opto-stimulation test, they received an intrathecal injection of PBS, ondansetron, metergoline or BRL-44408. Intrathecal injections were performed in fully conscious mice over a period of 10 s as routinely performed in the lab44. After 15 min of intrathecal injections, mice were then stimulated with yellow light and mechanical withdrawal thresholds were taken. The 15-min time point and the

effective doses of antagonists were selected based on the ability of such compounds to reverse analgesia produced by selective agonists observed in a separate series of experiments (not shown).

SYBR Green quantitative real-time PCR. Fresh prelimbic PFC tissues were taken from SNI/sham mice 2 and 4 weeks after surgery. Tissues were homogenized in ice-cold TRI Reagents to extract total RNA from the samples as previously described48. The primer sequences were as follows: ACCCGCGAGCACAG CTTCT (forward) and GCCTCGTCACCCACATAGGAGTCC (reverse) for β-actin, TCCGGATGCTGGAGTATTA (forward) and CAGGTTCATGTAGGACGAAG (reverse) for CB1, GAA GGAGCCACCGCGTTTTTA (forward) and TCGGCGAGAAGTTGTATTCC T (reverse) for NAPE-phospholipase D, CAGTGATGGTGGCTGCTCTT (forward) and GCCAGCCGA GGAAACAGAG (reverse) for fatty acid amide hydrolase, GCAGTGTCAGGAG CAAGTCT (forward) and ACAGCAGCAGAA GCTCTACG (reverse) for diacylglycerol lipase, and ACAAGTCGGAGGGTTCTGCT (forward) and GAGGAC GTGATAGGCACCTT(reverse) for monoacylglycerol lipase. The oligos were synthesized by the University of Calgary Core DNA service. RNA 250 ng was reverse transcribed using SuperScript IV Reverse Transcriptase (catalog no. 18090050, Life Technologies) following the manufacturer’s instructions. Complementary DNA synthesis was conducted at 23 °C for 10 min to extend primers, 53 °C for 10 min to synthesize cDNA and 80 °C for 10 min to inactivate the reaction. The cDNA samples were amplified using Fast SYBR Green Master Mix (catalog no. 4385612, Thermo Fisher Scientific) using Applied Biosystems QuantStudio 3 (Thermo Fisher Scientific). The PCR reaction started at 95 °C for 20 s to activate AmpliTaq Fast DNA polymerase, followed by 40 cycles at 95 °C for 3 s to denature and 60 °C for 30 s to extend the cDNA templates. The mRNA expression levels were normalized to β-actin.

Western blot analysis. Prelimbic PFC tissue from sham and SNI mice was homogenized in modified radioimmunoprecipitation buffer (in mM: Tris 50, NaCl 150, ethylenediaminetetraacetic acid 5, 1% Triton-X, 1% NP-40, 0.2% sodium dodecylsulfate, pH 7.4) containing protease inhibitors. Homogenates were centrifuged at 16,000g for 15 min at 4 °C and supernatant protein quantified using the Bio-Rad protein assay dye. Homogenates, 100 µg, were resolved by sodium dodecylsulfate/polyacrylamine gel electrophoresis and transferred to a polyvinylidene fluoride membrane. The membrane was blocked using 3% bovine serum albumin in Tris-buffered saline (containing 0.1% Tween 20) for 1 h at room temperature and analyzed by western blotting using anti-CB1 receptor (rabbit, 1:250, Abcam, ab23703) with ECL anti-rabbit immunoglobulin (Ig)G, horse radish peroxidase-linked whole antibody (donkey, 1:5,000, GE Healthcare, catalog no. NA934) and anti-tubulin (mouse, 1:5,000, Abcam, ab7291) with peroxidase-conjugated, affinity, sheep anti-mouse Ig(heavy + light) (Ig(H+L) (1:5,000, Jackson ImmunoResearch, catalog no. 515-035-003). Blots were quantified using ImageJ.

Immunohistochemistry. Staining with c-fos was used to assess the effect of in vivo optogenetic manipulations on neuronal activity in the PFC and PAG. Mice were stimulated with blue or yellow light for 10 min. They were then transcardially perfused with 0.1 M PBS and 4% paraformaldehyde in PBS about 90 min after light stimulation. Brains were removed and fixed in 4% paraformaldehyde for 2 h and then transferred to 30% sucrose at 4 °C overnight. Coronal sections (30 μm) of the PFC and PAG were collected using a cryostat (Leica CM3050 S) and immunohistochemistry was performed. Briefly, sections were washed five times and were permeabilized with 0.3% Triton-X for 1 h. Then sections were blocked with 5% normal goat serum and 2% bovine serum albumin in PBS for 2 h. The brain sections were then incubated with anti-c-fos antibody (rabbit, 1:1,000, abcam, ab190289; rabbit, 1:400, SantaCruz, sc-52), anti-CaMKIIα (mouse, 1:100, Thermo Fisher Scientific, MA1-048), anti-GAD67 (mouse, 1:100, abcam, ab26116), anti-tyrosine hydroxylase (mouse, 1:1,000, Sigma-Aldrich, T1299) and anti-parvalbumin (mouse, 1:2,000, Millipore, MAB1572) at 4 °C overnight. Alexa Fluor 488 donkey anti-mouse (1:400, Life Technologies, A21202), Alexa Fluor 546 goat anti-rabbit (1:400, Life Technologies, A11035), Alexa Fluor 546 goat anti-mouse (1:1,000, Invitrogen, A11030) and Alexa Fluor 633 goat anti-mouse (1:400, Invitrogen, A21050, shown in green color in the figures) were incubated at room temperature for 2 h. Images were visualized using a 40 × 0.4 NA or 20 × 0.75 NA objective lens on a Zeiss LSM 510 META and ConfoCor systems, running Velocity 6 and 20 × 0.75 NA or 10 × 0.4 NA objective lens on a Leica TCS SP8. For some of the tracing experiments, whole brain sections were stained with DAPI (1:10,000) and were scanned using a slide scanner (Olympus, System VS110, Model BX61VSF). For each mouse, the data from the various sections were averaged and the mean ± s.e.m. calculated for groups of mice.

Randomization, blinding and data exclusion. For electrophysiology, animals were randomly assigned to each day’s experiments by alternating sham and SNI each day. Mice were randomly assigned to light or no light stimulation, drug delivery, and SNI or sham for in vivo experiments. Blinding between SNI and sham conditions was conducted for a subset of behavioral experiments involving



ArticlesNature NeuroscieNceoptogenetic studies of the BLA–PFC projection and ablation experiments. There are two limitations to the effectiveness of blinding: first, during optogenetics, the experimenter can easily see whether the laser is ON or OFF. Second, after SNI, mice often display alterations in gait due to neuropathic pain. The inability to blind these studies fully is mitigated by the fact that mechanical and thermal withdrawal thresholds and latencies are determined in an automated fashion (that is, with a digital aesthesiometer) and results were replicated across several cohorts conducted by different investigators. Criteria for data exclusion are described in the Reporting Summary accompanying this manuscript.

Statistical analysis. Data were analyzed with GraphPad Instat v.3.0, Graphpad Prism v.6.0, Synaptosoft and Clampfit (Axon pClamp software v.10.3), ImageJ and Matlab software, and data are presented as the mean ± s.e.m. Data for miniature ESPCs (mEPSCs) or eEPSCs were analyzed using the Mini Analysis Program (Synaptosoft) and Clampfit (Molecular Devices). Paired or unpaired Student’s t-tests and one-way analysis of variance (ANOVA), followed by Newman–Keuls multiple comparison tests, were used for comparison of electrophysiological data. One-way ANOVA, followed by Tukey’s multiple comparison test, was used for analysis of western blot data. Animal behavioral data were analyzed by one- or two-way ANOVA followed by Bonferroni’s post-hoc corrections, Newman–Keuls multiple comparison test or two-tailed, paired or unpaired, Student‘s t-tests. For immunofluorescence analysis, data were analyzed using one-way ANOVA with Bonferroni’s post-hoc correction or unpaired Student’s t-tests. Statistical significance was accepted at the level of P < 0.05 with asterisks in figures denoting P values as follows: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. No statistical methods were used to predetermine sample sizes, but our sample sizes are similar to those reported in previous publications7,29,37,41. Data distribution was assumed to be normal but this was not formally tested in all cases.

Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availabilityData for analysis of in vivo calcium measurements will be made available upon reasonable request. The data that support the findings of this study are available from the corresponding author upon request.

Code availabilityThe code for analysis of in vivo calcium measurements will be made available upon reasonable request.

References 41. Penzo, M. A. et al. The paraventricular thalamus controls a central amygdala

fear circuit. Nature 519, 455–459 (2015). 42. Allen, W. E. et al. Thirst-associated preoptic neurons encode an aversive

motivational drive. Science 357, 1149–1155 (2017). 43. Beyeler, A. et al. Divergent routing of positive and negative information from

the amygdala during memory retrieval. Neuron 90, 348–361 (2016). 44. Zhang, F. X., Gadotti, V. M., Souza, I. A., Chen, L. & Zamponi, G. W. BK

potassium channels suppress Cavalpha2delta subunit function to reduce inflammatory and neuropathic pain. Cell Rep. 22, 1956–1964 (2018).

45. Brenner, D. S., Golden, J. P. & Gereau, R. W. A novel behavioral assay for measuring cold sensation in mice. PLoS One 7, https://doi.org/10.1371/journal.pone.0039765.g001 (2012).

46. Sestakova, N., Puzserova, A., Kluknavsky, M. & Bernatova, I. Determination of motor activity and anxiety-related behaviour in rodents: methodological aspects and role of nitric oxide. Interdiscip. Toxicol. 6, 126–135 (2013).

47. Simone, K., Fuzesi, T., Rosenegger, D., Bains, J. & Murari, K. Open-source, cost-effective system for low-light in vivo fiber photometry. Neurophotonics 5, 025006 (2018).

48. Huang, J. et al. Targeting the D series resolvin receptor system for the treatment of osteoarthritis pain. Arthr. Rheumatol. 69, 996–1108 (2017).


https://doi.org/10.1371/journal.pone.0039765.g001https://doi.org/10.1371/journal.pone.0039765.g001http://www.nature.com/natureneuroscience

1

nature research | reporting summ

aryO

ctober 2018

Corresponding author(s): Gerald Zamponi, Zizhen Zhang

Last updated by author(s): Jun 25, 2019

Reporting SummaryNature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.

StatisticsFor all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed

The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement

A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly

The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.

A description of all covariates tested

A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons

A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings

For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes

Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated

Our web collection on statistics for biologists contains articles on many of the points above.

Software and codePolicy information about availability of computer code

Data collection Axon pClamp software 10.3 (Clampex and Clampfit), Zeiss LSM 510, Leica TCS SP8, slide scanner (Olympus, System VS110, Model BX61VSF), 1-site 2-color Fiber Photometry System (Doric), Applied Biosystems QuantStudio 3

Data analysis Axon pClamp software 10.3 (Clampex and Clampfit) ; Synaptosoft inc Mini Analysis Program; Image J 2.0.0; Leica Application Suite X; GraphPad Software Prism 6; GraphPad Instat 3.0, MATLAB R2018b; Adobe Illustrator CC; Microsoft office Excel 2013/2016, Word2013/2016; Adobe Acrobat Pro 2017. Custom code for fiber photometry is available upon request.

For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

DataPolicy information about availability of data

All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability

A statement that data will be made available upon reasonable request has been included in the methods.

2

nature research | reporting summ

aryO

ctober 2018

Field-specific reportingPlease select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences

For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Life sciences study designAll studies must disclose on these points even when the disclosure is negative.

Sample size We didn't use statistical methods to predetermine the sample size, but our study used similar sample sizes to previously studies in behavior neuroscience field (PMID: 26212331,PMID: 7695716,PMID: 30209395,PMID: 25600269 ).

Data exclusions 1) In electrophysiology recordings, any cells with the following properties were excluded, with established a priori based on standard practice in this field: a leak current more than -150 pA, or a membrane potential more positive than -50 mV, or a series resistance more than 20 M ohms after series resistance compensation, or with a 20% increase at the end of the experiments were excluded. 2) As stated in the methods section, in tracing experiments, mouse brains with negative labeling were not considered when this was caused by off target injections of tracers such as viruses/CTB/GCaMP6f. 3) Animals without pain responses after SNI surgery were not used for experimentation. 4) Calcium imaging traces without significant transients were excluded from analyses.

Replication Numbers of replicates for all experiments are clearly denoted in all figures. No problems in replicating results in different mice throughout the duration of the study were encountered and results were replicated across several cohorts conducted by different investigators.

Randomization For electrophysiology, animals were randomly assigned to each day’s experiments and by alternating Sham and SNI each day. Mice were randomly assigned to light or no light stimulation, drug delivery, and SNI or Sham for in vivo experiments.

Blinding Blinding between SNI and sham conditions was conducted for a subset of behavioral experiments involving optogenetic studies of the BLA-PFC projection and ablation experiments. There are two limitations to the effectiveness of blinding: First, during optogenetics, the experimenter can easily see whether the laser is ON or OFF. Second, after SNI, mice often display alterations in gait due to neuropathic pain. The inability to blind these studies fully is mitigated by the fact that mechanical and thermal withdrawal thresholds and latencies are determined in an automatized fashion (i.e., with a digital aesthesiometer) and results were replicated across several cohorts conducted by different investigators.

Reporting for specific materials, systems and methodsWe require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.

Materials & experimental systemsn/a Involved in the study

Antibodies

Eukaryotic cell lines

Palaeontology

Animals and other organisms

Human research participants

Clinical data

Methodsn/a Involved in the study

ChIP-seq

Flow cytometry

MRI-based neuroimaging

AntibodiesAntibodies used anti-c-fos antibody (rabbit, 1:1000, abcam, Cat#: ab190289, Lot#: GR3188743-1; rabbit, 1:400,SantaCruz, Cat#:sc-52),

anti-CaMKIIα (mouse, 1:100, Thermo Fisher Scientific, Cat#:MA1-048, Lot#: SE248714，Clone#: 6G9), anti-GAD67(mouse, 1:100, abcam, Cat#:ab26116, Lot#: GR3188743-1, Clone #: K-87 ) anti-Tyrosine Hydroxylase (mouse,1:1000, Sigma-Aldrich,Cat#:T1299, Lot#:036M4860V, Clone#: TH-2) anti-parvalbumin (mouse,1:2000, Millipore, Cat#:MAB1572,Lot#:2552333, Clone#: PARV-19) anti-CB1 receptor (rabbit, 1:250, Abcam, Cat#:ab23703, Lot#:GR3239384-2) anti-tubulin (mouse, 1:5000, Abcam, Cat#:ab7291, Lot#: GR66309-4, Clone #: DM1A) Alexa Fluor®488 donkey anti-mouse (1:400, Life Technologies, Cat#: A21202, Lot#:1423052), Alexa Fluor® 546 goat anti-rabbit (1:400, Invitrogen, Cat#: A11035, Lot#:1904467)

A neuronal circuit for activating descending modulation of ......ARTICLES NATURE NEUROSCIENCE g a c...

Documents

Transcript of A neuronal circuit for activating descending modulation of ......ARTICLES NATURE NEUROSCIENCE g a c...