A POMC-originated circuit regulates stress-induced hypophagia, depression and anhedonia

Abstract

Chronic stress causes dysregulations of mood and energy homeostasis, but the neurocircuitry underlying these alterations remain to be fully elucidated. Here we demonstrate that chronic restraint stress in mice results in hyperactivity of pro-opiomelanocortin neurons in the arcuate nucleus of the hypothalamus (POMC^ARH neurons) associated with decreased neural activities of dopamine neurons in the ventral tegmental area (DA^VTA neurons). We further revealed that POMC^ARH neurons project to the VTA and provide an inhibitory tone to DA^VTA neurons via both direct and indirect neurotransmissions. Finally, we show that photoinhibition of the POMC^ARH→VTA circuit in mice increases body weight and food intake, and reduces depression-like behaviors and anhedonia in mice exposed to chronic restraint stress. Thus, our results identified a novel neurocircuitry regulating feeding and mood in response to stress.

Introduction

Stress is a physiological and adaptive response to changes in the environment, but can also cause pathological alterations. For example, chronic stress increases vulnerability to anxiety, depression and other mood disorders¹. In addition, stress is also known to change feeding behavior and energy homeostasis^2–4. However, the neurocircuitry underlying stress-induced alterations in mood and energy balance is not fully understood.

The ventral tegmental area (VTA) in the midbrain is one main brain region that responds to various aversive stimuli and coordinates behavioral and metabolic adaptations⁵. Heterogeneous neural types exist within the VTA, including about 65% dopamine (DA) neurons, 30% GABAergic neurons, and 5% glutamatergic neurons^6–8. In particular, depression-like behavior in animals is associated with reduced DA bioavailability⁹, and antidepressant treatment increases DA transmission¹⁰. Optogenetic activation of DA^VTA neurons in mice reduces anhedonia and depression-like phenotypes, as measured in a variety of behavioral assays, including the sucrose preference test (SPT), the forced swim test (FST) and the tail suspension test (TST)^{11, 12}. Further, a step-like stimulation of DA^VTA neurons also increases sucrose preference in mice, although this manipulation does not significantly alter depression-like behaviors measured by FST and TST¹³. Leptin, the anorexigenic hormone, has been reported to act upon DA^VTA neurons to regulate mood^14–16. DA^VTA neurons may also play essential roles in the regulation of food intake, as mice born without DA synthesis develop hypophagia which can be rescued by supplement with a DA analog^{17, 18}. However, the role of DA^VTA neurons in feeding control appears to be complex, since chemogenetic activation of DA^VTA neurons increases meal frequency but reduces meal size, resulting in unaffected total food intake¹⁹.

Pro-opiomelanocortin (POMC) neurons and agouti-related peptide (AgRP) neurons, both located within the arcuate nucleus of hypothalamus (ARH), play essential roles in feeding control^20–24. In particular, POMC^ARH neurons secret α-melanocyte-stimulating hormone (α-MSH) to stimulate melanocortin 3/4 receptor (MC3R/MC4R) and reduce food intake; on the other hand, AgRP inhibits MC3R/MC4R to increase feeding^20–22. Notably, POMC^ARH neurons also release another neuropeptide, β-endorphin, but its role in feeding control is debatable^25–28. It has been reported that chronic restraint stress increases POMC gene expression in the ARH by about 60%³, but functions of POMC^ARH neurons in the context of chronic stress remains unclear.

In the current study, we first examined the neural activities of DA^VTA neurons, POMC^ARH neurons and AgRP^ARH neurons in mice exposed to chronic restraint stress, and observed increased POMC^ARH neuron activity and decreased DA^VTA neuron activity. We further identified a POMC^ARH→VTA circuit and systemically examined neurotransmissions between POMC^ARH neurons and various types of VTA neurons. Finally, we tested the functional relevance of the POMC^ARH→VTA circuit in the regulations of feeding, anhedonia and depression-like behaviors in mice exposed to chronic restraint stress.

Results

Chronic restraint stress induces hypophagia, anhedonia and depression-like behaviors

We first examined effects of chronic restraint stress on energy balance, anhedonia and depression-like behaviors in mice. To this end, wild type mice (males, 3 months of age) were randomly divided into control (CON) and restraint (RST) groups. The RST mice were subjected to 1-hour restraint (9:30am-10:30am) every day for 14 consecutive days (Figure 1A), while the control mice were left intact. During the 14-day period, RST mice displayed significant reductions in their body weight (about 6–7 g) compared to CON mice which plateaued from day 8 (Figure 1B). Food intake was also significantly reduced in RST mice during the entire 14-day period (Figure 1C). These results are consistent with previous report³ and indicate that chronic stress induces body weight loss presumably due to hypophagia.

Figure 1. Chronic restraint stress induces metabolic and neurobehavioral deficits, and alters activity of DA^VTA and POMC^ARH neurons.

(A) A diagram of restraint or control protocol followed by neurobehavioral tests. (B-C) Body weight (B) and food intake (C) measured daily during the 14-day restraint or control protocol. Results are shown as mean ± SEM. **, P＜0.01 and ***, P＜0.001 between RST vs. CON in two-way ANOVA analyses followed by Sidak post hoc test (N=7 mice per group). (D) Sucrose preference measured in the sucrose preference test on day 16. (E) Immobility time measured in the forced swim test on day 17. (F) Immobility time measured in the tail suspension test on day 18. Results are shown as mean ± SEM. *, P＜0.05 and **, P＜0.01 between RST vs. CON in t-tests (N=7 mice per group). (G) Representative microscopic images showing a recorded DA^VTA neuron (TOMATO-labelled in DAT-CreER/Rosa26-LSL-tdTOMATO mice). Scale bar= 10 μm. (H) Representative action potential traces in DA^VTA neurons from mice subjected to CON or RST protocol for 4 days. (I-J) Quantifications of firing frequency (I) and resting membrane potential (J) in DA^VTA neurons from mice subjected to CON or RST protocol for 1, 4 or 14 days. Results are shown as mean ± SEM. *, P＜0.05 and ***, P＜0.001 between RST vs. CON in one-way ANOVA analyses followed by Tukey’s test (N=24–30 neurons from 4 mice per group). (K) Representative microscopic images showing a recorded POMC^ARH neuron (TOMATO-labelled in POMC-CreER/Rosa26-LSL-tdTOMATO mice). Scale bar= 10 μm. (L) Representative action potential traces in POMC^ARH neurons from mice subjected to CON or RST protocol for 4 days. (M-N) Quantifications of firing frequency (M) and resting membrane potential (N) in POMC^ARH neurons from mice subjected to CON or RST protocol for 1, 4 or 14 days. Results are shown as mean ± SEM. *, P<0.05 and ***, P＜0.001 between RST vs. CON in one-way ANOVA analyses followed by Tukey’s test (N=33–40 neurons from 4 mice per group). (O) Representative microscopic images showing a recorded AgRP^ARH neuron (TOMATO-labelled in AgRP-IRES-Cre/Rosa26-LSL-tdTOMATO mice). Scale bar= 10 μm. (P) Representative action potential traces in AgRP^ARH neurons from mice subjected to CON or RST protocol for 4 days. (Q-R) Quantifications of firing frequency (Q) and resting membrane potential (R) in POMC^ARH neurons from mice subjected to CON or RST protocol for 1, 4 or 14 days. Results are shown as mean ± SEM. N=46–49 neurons from 4 mice per group.

We then examined the behavioral consequences in RST mice vs. CON mice after they completed the 14-day paradigm. Similar to observations reported previously²⁹, RST mice displayed anhedonia in a sucrose preference test (SPT) compared to CON mice (Figure 1D). Further, we observed significantly increased despair in RST mice, as demonstrated by increased immobility in both forced swim test (FST) and tail suspension test (TST, Figure 1E–F). Together, these results indicate that chronic stress results in depression-like behaviors.

Chronic stress leads to inhibition of DA^VTA neurons and activation of POMC^ARH neurons

Since DA^VTA neurons play an essential role in regulating depression^{11, 12}, we sought to examine whether RST can regulate activity of DA^VTA neurons. To this end, DAT-CreER/Rosa26-LSL-tdTOMATO mice (after tamoxifen induction) were subjected to CON or RST paradigm for 1 day, 4 days or 14 days, and we recorded firing activity of DA^VTA neurons (TOMATO-labelled) in brain slices prepared from these mice immediately after they were exposed to RST or CON treatment (Figure 1G). We found that compared to CON mice, RST for 1 day significantly reduced the firing frequency of DA^VTA neurons but did not change their resting membrane potential; RST for 4 or 14 days produced similar inhibitory effects to significantly reduce firing frequency and resting membrane potential of DA^VTA neurons (Figure 1H–J and S1A–B). No significant changes were noted in the after-hyperpolarization (Figures S1C–D). Interestingly, RST failed to change firing frequency and resting membrane potential of DA neurons in the substantia nigra (Figure S1E–H).

Given the changes in body weight and feeding induced by RST, we speculated that RST may also influence activity of appetite-regulatory neurons, e.g. POMC^ARH neurons and/or AgRP^ARH neurons. Using POMC-CreER/Rosa26-LSL-tdTOMATO mice (after tamoxifen induction) subjected to CON or RST paradigm, we found that RST for 1 day significantly increased the resting membrane potential of POMC^ARH neurons but did not change their firing frequency; RST for 4 or 14 days produced similar excitatory effects to significantly increase firing frequency and resting membrane potential of POMC^ARH neurons (Figure 1K–N and S1I–J). However, RST failed to significantly alter the firing activity of AgRP^ARH neurons (TOMATO-labelled neurons in AgRP-IRES-Cre/Rosa26-LSL-tdTOMATO mice, Figure 1Q–R and S1K–L).

POMC^ARH neurons inhibit DA^VTA neurons via direct and indirect neurotransmissions

Given the concurrent changes in activities of DA^VTA neurons and POMC^ARH neurons in RST mice, we postulated that POMC^ARH neurons may project to DA^VTA neurons to form a neural circuit. To test this possibility, we stereotaxically injected AAV-FLEX-Synaptophysin-GFP into the ARH of POMC-Cre mice to selectively express Synaptophysin-GFP in POMC^ARH neurons (Figure S2A and S2B-left panel). We observed abundant GFP-labelled neural terminals and boutons in the VTA (Figure S2B-middle panel), indicating that POMC^ARH neurons project to the VTA. Notably, while some of the GFP-labelled boutons were in close proximity to DA neurons (identified as TH-positive neurons), many GFP-labelled boutons appeared to form contacts with non-DA neurons in the VTA (Figure S2B-right panel). Further, we stereotaxically injected AAV-FLEX-GTB and EnvA G-deleted Rabies-mCherry viruses into the VTA of DAT-CreER mice for retrograde tracing from DA^VTA neurons (Figure S2C–S2D). We noted many mCherry-labelled neurons in the ARH, a portion of which were confirmed to be POMC neurons as demonstrated by co-staining of β-endorphin (Figure S2E). Notably, many mCherry-labelled ARH neurons were not POMC neurons (Figure S2E), indicating that DA^VTA neurons also receive synaptic inputs from non-POMC^ARH neurons. Consistently, it has been reported that AgRP^ARH neurons innervate DA^VTA neurons³⁰. Together, these data indicate that a subset of POMC^ARH neurons project to DA^VTA neurons and non-DA^VTA neurons, and that DA^VTA neurons receive synaptic inputs from POMC^ARH neurons and non-POMC^ARH neurons.

Given the complex innervations from POMC^ARH neurons to DA^VTA neurons and non-DA^VTA neurons, we further examined the neurotransmissions between POMC^ARH neurons and several types of VTA neurons. To this end, we used an “upgraded” channelrhdopsin 2 (ChR2)-assisted circuit mapping (CRACM), which we named as WGA-CRACM. Briefly, we stereotaxically injected Cre-dependent AAV expressing ChR2-EYFP into the ARH of POMC-CreER/DAT-CreER/Rosa26-LSL-tdTOMATO mice to express ChR2 specifically in POMC^ARH neurons and their fibers/terminals, which allows photostimulation (as shown Figure S2F). The same mice also received stereotaxic injections of Ad-FLEX-WGA-GFP, an anterograde trans-synaptic viral tracer³¹, into the ARH. Ad-FLEX-WGA-GFP expressed GFP-tagged wheat germ agglutinin (WGA-GFP) in a Cre-dependent manner and therefore only POMC^ARH neurons were filled with WGA-GFP. WGA-GFP is an anterograde trans-synaptic tracer, which labels the soma of downstream neurons that are immediately innervated by POMC^ARH terminals³¹. Thus, GFP-labelled soma in the VTA were identified as the immediate targets of POMC^ARH neurons. Note that these mice also express TOMATO in DA neurons due to the combination of DAT-CreER and Rosa26-LSL-tdTOMATO alleles. Therefore, GFP(+)TOMATO(+) neurons were identified as POMC^ARH-innervated DA^VTA neurons, which account for about 43.66% GFP(+) neurons (POMC^ARH-innervated VTA neurons) and 31.40% TOMATO(+) neurons (DA^VTA neurons) (Figure S2G–I).

We then examined electrophysiological responses in POMC^ARH-innervated DA^VTA [GFP(+)TOMATO(+)] neurons (Figure 4A). Using a voltage clamp protocol, we detected excitatory post-synaptic current (EPSC) only in 9 out of 44 tested GFP(+)TOMATO(+) neurons evoked by blue light pulses (Figure 2B–C and S3A). The EPSC were mediated by glutamatergic neurotransmission as they were blocked by glutamate receptor inhibitors (CNQX and D-AP5); 4-AP and TTX failed to affect the EPSC, confirming a monosynaptic nature (Figure 2B–C). On the other hand, we failed to detect any light-evoked inhibitory post-synaptic current (IPSC) in all 20 tested GFP(+)TOMATO(+) neurons (Figure 2D and S3A). These results indicate that POMC^ARH neurons provide minimal direct glutamatergic and none GABAergic inputs to DA^VTA neurons. Interestingly, under a current clamp mode, we found that blue light photostimulation caused rapid decreases in firing frequency and resting membrane potential in all 19 tested GFP(+)TOMATO(+) neurons (Figure 2E, J and K), indicating that POMC^ARH neurons inhibit DA^VTA neurons. The lack of inhibitory IPSC observed at the POMC^ARH→DA^VTA synapse suggests that neuropeptides from POMC^ARH neurons, e.g. α-MSH and β-endorphin, may mediate the inhibitory effects on DA^VTA neurons. Incubation of SHU9119, an antagonist of MC3R/MC4R, did not affect the inhibitory neurotransmission at the POMC^ARH→DA^VTA synapse (Figure 2F, J and K), indicating that α-MSH is not involved. Another neuropeptide, β-endorphin, could act upon one of 3 opioid receptors, namely δ-opioid receptor (DOR), κ-opioid receptor (KOR) and μ-opioid receptor (MOR). Thus, we tested effects of naltrindole (an antagonist of DOR), norbinaltorphimine (an antagonist of KOR), or cyprodime (an antagonist of MOR), respectively. While naltrindole and norbinaltorphimine did not affect the inhibitory neurotransmission at the POMC^ARH→DA^VTA synapse, incubation of cyprodime completely prevented the decreases in firing frequency and resting membrane potential in DA^VTA neurons evoked by photostimulation of POMC^ARH projections (Figure 2G–K). Further, we collected 4 single neurons double-labelled by GFP and TOMATO (POMC^ARH-innervated DA^VTA neurons), and used RT-qPCR to detect expression of MC3R, MC4R and MOR mRNAs. None of these neurons expressed MC3R, and only one out of 4 neurons expressed MC4R mRNAs (Figure 2L). On the other hand, all 4 neurons expressed abundant MOR mRNAs (Figure 2L). Collectively, these results indicate that POMC^ARH neurons inhibit DA^VTA neurons largely via MOR-mediated neurotransmission.

Figure 4. POMC^ARH neurons indirectly inhibit a subset of DA^VTA neurons.

(A) Recordings in DA^VTA neurons not innervated by POMC^ARH neurons [TOMATO(+) only] in response to photostimulation of ChR2-labelled POMC^ARH-originated fibers within the VTA. (B) The percentage of TOMATO(+) neurons that showed light-evoked EPSCs or no response. (C) Representative traces for light-evoked IPSC, which were blocked by 50 μM bicuculline, by 400 μM 4-AP and 1 μM TTX, or by 50 μM SHU9119. (D) The percentage of TOMATO(+) neurons that showed light-evoked IPSCs or no response in conditions with no blockers or with SHU9119. P=0.0033 in χ2 test. (E-G) Representative action potential traces in TOMATO(+) neurons in response to photostimulation of POMC^ARH-originated fibers within the VTA, in the absence (E) or presence of bicuculline (50 μM, F), or SHU9119 (50 μM, G). (H-I) Light-induced changes in firing frequency (H) and resting membrane potential (I). Results are shown as mean ± SEM. *, P<0.05, **, P<0.01 or ***, P＜0.001 between other group vs. no blocker in one-way ANOVA analyses followed by Sidak post hoc tests (N=11 to 24 neurons from 3 mice per group). (J) A schematic model for POMC^ARH neurons to directly or indirectly inhibit different subsets of DA^VTA neurons.

Figure 2. POMC^ARH neurons directly inhibit a portion of DA^VTA neurons via MOR-mediated mechanisms.

(A) Recordings in POMC^ARH-innervated DA^VTA neurons [GFP(+)TOMATO(+)] in response to photostimulation of ChR2-labelled POMC^ARH-originated fibers within the VTA. (B) Representative traces for light-evoked EPSC, which were blocked by 30 μM D-AP5 and 30 μM CNQX, but not affected by 400 μM 4-AP and 1 μM TTX. (C) The percentage of GFP(+)TOMATO(+) neurons that showed light-evoked EPSCs or no response. (D) The percentage of GFP(+)TOMATO(+) neurons that showed light-evoked IPSCs or no response. (E-I) Representative action potential traces in GFP(+)TOMATO(+) neurons in response to photostimulation of POMC^ARH-originated fibers within the VTA, in the absence (E) or presence of SHU9119 (50 μM, F), naltrindole (100 μM, G), norbinaltorphimine (100 μM, H), or cyprodime (200 μM, I). (J-K) Light-induced changes in firing frequency (J) and resting membrane potential (K). Results are shown as mean ± SEM. **, P＜0.01 and ***, P＜0.001 between cyprodime group vs. no blockers in one-way ANOVA analyses followed by Sidak post hoc test (N=10–19 neurons from 3 mice per group). (L) Relative mRNA levels of MC3R, MC4R and MOR in single VTA neurons labelled by both GFP and TOMATO. Results are shown as mean ± SEM with each data point plotted. N=4 neurons from 2 mice per group.

We then examined blue light-induced responses in POMC^ARH-innervated non-DA^VTA [GFP(+) only] neurons (Figure 3A). Under the voltage clamp mode, we observed light-evoked EPSC in 12 out of 42 tested neurons, which were blocked by CNQX and D-AP5 but were not affected by 4-AP and TTX (Figure 3B–C and S3B). On the other hand, we observed no light-evoked IPSC in all 21 tested neurons (Figure 3D and S3B). These results indicate that POMC^ARH neurons provide minimal direct glutamatergic and no GABAergic inputs to these non-DA^VTA neurons, similar to their inputs to DA^VTA neurons. Under the current clamp mode, we found that blue light photostimulation caused rapid increases in firing frequency and resting membrane potential in 12 out of 14 tested GFP(+) neurons (Figure 3E), indicating that POMC^ARH neurons activate these non-DA^VTA neurons. Interestingly, incubation of SHU9119, the antagonist of MC3R/MC4R, abolished the excitation in these non-DA^VTA neurons caused by POMC^ARH activation (Figure 3F–H), indicating that α-MSH mediates the excitatory neurotransmission at the majority of the POMC^ARH→non-DA^VTA synapse. To explore the chemical identity of these POMC^ARH-innervated non-DA^VTA neurons, we collected single neurons labelled by GFP only, and detected abundant vGAT mRNAs but minimal DAT mRNAs in these neurons (Figure 3I–J), indicating that these non-DA^VTA neurons are likely GABAergic neurons; as a positive control, abundant DAT mRNAs were detected in neurons labelled by both GFP and TOMATO (Figure 3J). In addition, the neurons labelled by GFP only expressed MC3R but no MC4R mRNAs, as well as abundant MOR mRNAs (Figure 3K). Further, we directly recorded from GABA^VTA neurons (TOMATO-labelled VTA neurons in Vgat-IRES-Cre/Rosa26-LSL-tdTOMATO mice, Figure 3L), and found that a MC3R/MC4R agonist, MT-II, depolarized 10 out of 25 tested GABA^VTA neurons in the presence of synaptic inhibitors (TTX, CNQX, D-AP5 and bicuculline); the other 15 GABA^VTA neurons were not affected by MT-II (Figure 3M–N). Collectively, these data indicate that POMC^ARH neurons release α-MSH to directly activate a portion of GABA^VTA neurons via MC3R.

Figure 3. POMC^ARH neurons directly activate GABA^VTA neurons via MC3R-mediated mechanisms.

(A) Recordings in POMC^ARH-innervated non-DA^VTA neurons [GFP(+) only] in response to photostimulation of ChR2-labelled POMC^ARH-originated fibers within the VTA. (B) Representative traces for light-evoked EPSC, which were blocked by 30 μM D-AP5 and 30 μM CNQX, but not affected by 400 μM 4-AP and 1 μM TTX. (C) The percentage of GFP(+) neurons that showed light-evoked EPSCs or no response. (D) The percentage of GFP(+) neurons that showed light-evoked IPSCs or no response. (E-F) Representative action potential traces in GFP(+) neurons in response to photostimulation of POMC^ARH-originated fibers within the VTA, in the absence (E) or presence of SHU9119 (50 μM, F). (G-H) Light-induced changes in firing frequency (G) and resting membrane potential (H). Results are shown as mean ± SEM. N=10 neurons from 3 mice per group. ***, P＜0.001 between SHU9119 group vs. no blocker in t-tests. (I-J) Relative mRNA levels of vGAT (I) and DAT (J) in single VTA neurons labelled by GFP only or by both GFP and TOMATO. Results are shown as mean ± SEM with each data point plotted. ***, P＜0.001 between two groups in t-tests (N=8 or 12 neurons from 2 mice per group). (K) Relative mRNA levels of MC3R, MC4R and MOR in single VTA neurons labelled by GFP only. Results are shown as mean ± SEM with each data point plotted. N=8 neurons from 2 mice per group. (L) Representative microscopic images showing a recorded GABA^VTA neuron (TOMATO-labelled in Vgat-IRES-Cre/Rosa26-LSL-tdTOMATO mice). Scale bar= 10 μm. (M) Representative traces for resting membrane potential recorded in a GABA^VTA neuron depolarized by 300 nM MT-II in the presence of 1 μM TTX, 30 μM CNQX, 30 μM D-AP5 and 50 μM bicuculline. (N) The resting membrane potential at the baseline and after MT-II treatment. Results are shown as individual data point. ***, P＜0.001 between baseline vs. MT-II in paired t-tests (N=10 or 15 neurons from 2 mice per group).

Finally, we examined blue light-induced responses in DA^VTA neurons that are not innervated by POMC^ARH [TOMATO (+) only] neurons (Figure 4A). Under the voltage clamp mode, we did not observe any light-induced EPSC in all 23 tested neurons (Figure 4B and S3C); interestingly, we detected light-induced IPSC in 27 out of 42 tested TOMATO (+) neurons with an averaged latency of 101.30±8.15 ms (Figure 4C–D and S3C). Importantly, these light-evoked EPSC were inhibited by the GABA_A receptor inhibitor, bicuculline, or by 4-AP and TTX (Figure 4C), suggesting an indirect neurotransmission mediated by GABAergic interneurons. Interestingly, in the presence of SHU9119 (the MC3R/MC4R antagonist), the light-evoked IPSC was only observed in 5 out of 21 tested neurons (P=0.0033 in χ2 test vs. the condition with no blockers; Figure 4C–D). Together, these results indicate that POMC^ARH neurons can directly activate a portion of GABA^VTA neurons through a MC3R-mediated mechanism, which in turn provide GABAergic inputs to a subset of DA^VTA neurons. Consistent with this indirect neurotransmission between POMC^ARH neurons and DA^VTA neurons, we found, under the current clamp mode, that blue light photostimulation caused rapid decreases in firing frequency and resting membrane potential in 11 out of 14 tested TOMATO(+) neurons, effects that can be blocked by bicuculline, or by the MC3R/MC4R antagonist, SHU9119 (Figure 4E–I).

In summary, these results identify two parallel mechanisms by which POMC^ARH neurons inhibit DA^VTA neurons. First, POMC^ARH neurons directly innervate a subset of DA^VTA neurons and inhibit these neurons via a MOR-mediated mechanism. In addition, POMC^ARH neurons indirectly inhibit another subset of DA^VTA neurons, largely through activating GABA^VTA interneurons via a MC3R-mediated mechanism (Figure 4J). Notably, POMC^ARH neurons also provide direct glutamatergic inputs to a small portion of DA^VTA and non-DA^VTA neurons. Since subsets of POMC^ARH neurons are reported to be glutamatergic while the other subsets are GABAergic³², we suggest that the VTA-projecting POMC^ARH neurons are glutamatergic but not GABAergic.

Photoinhibition of the POMC^ARH→DA^VTA circuit partially ameliorates effects of chronic restraint stress

Given that POMC^ARH neurons inhibit DA^VTA neurons and that POMC^ARH neurons are activated while DA^VTA neurons are inhibited in RST mice, we sought to examine the functional relevance of the POMC^ARH→VTA circuit in the context of chronic restraint stress. To this end, we used optogenetic approach to selectively inhibit the POMC^ARH→VTA projections in RST mice. Briefly, we stereotaxically injected AAV-FLEX-eNpHR3.0-EYFP into the ARH of POMC-Cre mice to express eNpHR3.0 specifically in POMC^ARH neurons and their fibers/terminals; wild-type (WT) littermates received the same virus injections as controls (Figure 5A). These mice were also implanted with an optic fiber to target the VTA. After recovery, mice were subjected to the 1-hour/day RST paradigm for 14 days coupled to simultaneous 1-hour photoinhibition (Figure 5B). Compared to WT mice which decreased body weight during the 14-day RST period, POMC-Cre mice displayed significantly higher body weight (Figure 5C). Food intake of POMC-Cre mice were also significantly higher than that of WT mice at multiple time points (Figure 5D). These results indicate that inhibition of the POMC^ARH→VTA circuit increases body weight loss and feeding in mice exposed to chronic restraint stress.

Figure 5. Photoinhibition of the POMC^ARH→VTA circuit increases food intake and decreases anhedonia and depression in mice exposed to chronic restraint stress.

(A) Schematic experimental strategy using the AAV-FLEX-eNpHR3.0-EYFP virus to inhibit the POMC^ARH→VTA circuit. (B) The experimental protocol for 14-day restraint with or without photoinhibition of the POMC^ARH→VTA circuit followed by neurobehavioral tests. (C-D) Body weight (C) and food intake (D) measured daily during the 14-day restraint protocol. Results are shown as mean ± SEM. *, P＜0.05 and **, P＜0.01 between WT vs. POMC-Cre mice in two-way ANOVA analyses followed by Sidak post hoc test (N=5 or 6 mice per group). (E) Sucrose preference measured in the sucrose preference test on day 16. (F) Immobility time measured in the forced swim test on day 17. (G) Immobility time measured in the tail suspension test on day 18. Results are shown as mean ± SEM. **, P＜0.01 between WT vs. POMC-Cre mice in t-tests (N=5–6 mice per group).

We then examined the behavioral consequences in these mice after they completed the 14-day RST paradigm. Interestingly, POMC-Cre mice displayed significantly higher sucrose preference in the SPT, and lower immobility in FST and TST than WT mice (Figure 5E–G). These results indicate that inhibition of the POMC^ARH→VTA circuit ameliorates anhedonia and depression-like behaviors induced by chronic restraint stress.

Importantly, the same photoinhibition of the POMC^ARH→VTA projections in mice not subjected to the RST paradigm did not induce any significant changes in body weight, food intake or behaviors (Figure S4).

Discussion

In the current study, we demonstrated that chronic restraint stress leads to decreased DA^VTA neuron activity, associated with increased POMC^ARH neuron activity. We further found that POMC^ARH neurons inhibit DA^VTA neurons via direct or indirect mechanisms. In addition, selective inhibition of the POMC^ARH→VTA circuit partially increases body weight and feeding, and reduces anhedonia and depression-like behaviors in mice exposed to chronic restraint stress. Thus, our results provided evidence that the hyperactivity of the POMC^ARH neurons and their projections to the VTA contributes to the metabolic and mood dysregulations associated with chronic restraint stress.

The existence of a POMC^ARH→VTA circuit has been implicated by the Allen Brain Atlas (Experiment 263369222). Here we used the anterograde and retrograde tracing to further confirm that POMC^ARH neurons send projections to the VTA in the mouse brain. In addition, AgRP^ARH neurons also project to the VTA³⁰. However, the detailed neurotransmissions between the hypothalamic melanocortin neurons and VTA neurons appear to be complex. On one hand, intra-VTA administration of α-MSH has been widely shown to increase DA release and turnover at efferent target sites, such as the nucleus accumbens (NAc) and prefrontal cortex³³. While α-MSH can act upon both melanocortin 3 and 4 receptors (MC3R and MC4R) in the brain³⁴, the stimulatory effects of α-MSH on DA turnover are thought to be mediated primarily by MC4R^{35, 36}. Further, loss of AgRP signals (the endogenous antagonist of melanocortin receptors) results in enhanced DA^VTA neuron activity and elevated DA in basal forebrain³⁰. On the other hand, it has been reported that abundant MC3R, but minimal MC4R, is expressed in the VTA³⁷. Deletion of MC3R increases DA content in the VTA³⁷, suggesting the dominant effect of α-MSH on DA^VTA neurons may be inhibitory. Notably, POMC^ARH neurons can also release β-endorphin which acts upon three different receptors, namely, KOR, DOR and MOR. All these receptors are located on presynaptic terminals of DA neurons to regulate DA release^38–40. However, their effects on DA release are complicated and could be opposite depending on the time course of treatment⁴¹ or on the subdivisions of projection targets⁴⁰. In addition, MOR has been shown to be expressed by GABA^VTA neurons, which receive inhibitory inputs from enkephalinergic neurons in the bed nucleus of stria terminalis (BNST) and inhibit DA^VTA neurons^{42, 43}. Thus, the regulations on DA^VTA neurons are quite complex due to heterogeneous cell types within the VTA, to various neuromodulators originating from distinct upstream neurons, and to different receptor subtypes for these neuromodulators.

Here we focused on the POMC^ARH→VTA circuit, and used the WGA-CRACM approach which allowed us to examine 3 types of neurotransmissions with genetically identified pre-synaptic and post-synaptic nodes: POMC^ARH neurons→DA^VTA neurons innervated by POMC^ARH neurons, POMC^ARH neurons →non-DA^VTA neurons innervated by POMC^ARH neurons, and POMC^ARH neurons→DA^VTA neurons not innervated by POMC^ARH neurons. First, POMC^ARH neurons directly synapse on and inhibit a subset of DA^VTA neurons, effects that can be completely blocked by the MOR antagonist. Consistently, modest levels of MOR mRNAs are present in this subset of DA^VTA neurons, while levels of MC3R and MC4R mRNAs are minimal. These results support a direct mechanism that some POMC^ARH terminals in the VTA release β-endorphin which directly inhibits DA^VTA neurons through MOR-mediated neurotransmission.

On the other hand, POMC^ARH neurons also directly synapse on and activate non-DA^VTA neurons. Importantly, these non-DA^VTA neurons innervated by POMC^ARH neurons are largely GABAergic, and express modest levels of MC3R mRNAs but minimal MC4R mRNAs. We further showed that the α-MSH agonist directly inhibits a portion of identified GABA^VTA neurons, similarly as reported by others⁴⁴. Notably, these non-DA^VTA neurons innervated by POMC^ARH neurons also express abundant MOR mRNAs, although the functional relevance of MOR in this subset of neurons was not directly tested in the current study.

Interestingly, we found that POMC^ARH neurons can also inhibit DA^VTA neurons that are not direct synaptic targets of POMC^ARH neurons. Since these inhibitory effects (both IPSC and decreased firing and resting membrane potential) are blunted by the GABA_A receptor inhibitor or by the α-MSH antagonist, we postulate an indirect mechanism that POMC^ARH neurons activate GABA^VTA neurons, which in turn inhibit a subset of DA^VTA neurons. Consistent with this indirect mechanism, earlier studies indicate that GABA^VTA interneurons provide local GABAergic inhibitory inputs to adjacent DA^VTA neurons^{42, 45, 46}. Together, our data demonstrate that POMC^ARH neurons provide an inhibitory tone to DA^VTA neurons through both direct and indirect neurotransmissions. Notably, a portion of POMC^ARH neurons express molecular machinery for synthesis and secretion of glutamate and GABA^{32, 47}. However, we only observed glutamate-mediated EPSC in a small portion of POMC^ARH-innervated VTA neurons and no GABA-mediated IPSC was observed. We suggest that only a modest subset of glutamatergic POMC^ARH neurons and minimal GABAergic POMC^ARH neurons project to the VTA, while the majority of VTA-projecting POMC^ARH neurons use α-MSH or β-endorphin as the dominant neuro-signal. It is worth mentioning that POMC^ARH neurons have modest projections to the NAc⁴⁸, which may regulate DA release through their actions on the DA^VTA-originated terminals within the NAc. A limitation of our WGA-CRACM approach is that it can only assess the neurotransmission at the cell bodies of VTA neurons, but could not measure potential inputs of POMC^ARH neurons to neural terminals that originate from the VTA.

Restraint stress causes increased activity of POMC^ARH neurons but decreased activity of DA^VTA neurons. Given that POMC^ARH neurons inhibit DA^VTA neurons, we suggest a model that restraint stress results in hyperactivity of POMC^ARH neurons, which in turn inhibit DA^VTA neurons. Further supporting the functional relevance of the POMC^ARH→VTA circuit, the optogenetic study demonstrates that inhibition of the POMC^ARH→VTA projections ameliorates anhedonia and depression-like behaviors induced by chronic restraint stress. Since receptors for β-endorphin and α-MSH both contribute to the POMC^ARH→VTA circuit, we speculate that these receptors within the VTA may be involved in the development of anhedonia and depression-like in animals exposed to chronic restraint stress. While KOR does not seem to regulate anhedonia and depression-like behaviors, genetic deficiency in either DOR or MOR alters animal’s depression level with opposite effects⁴⁹. For example, DOR knockout mice have increased depression⁴⁹, whereas MOR knockout mice show decreased depression-like behaviors⁵⁰. In particular, chronic stress (e.g. social defeat) robustly enhances MOR levels in the VTA^{51, 52}; selective knockdown of MOR in rat VTA prevents social avoidance induced by social stress exposure⁵³. Together with our observations that the inhibitory effects of POMC^ARH neurons on DA^VTA neurons are partially mediated by a direct MOR-dependent neurotransmission, these findings support a possibility that MOR in the VTA contributes to the development of anhedonia and depression-like behavior during chronic restraint stress. Alternatively, α-MSH-mediated activation of GABA^VTA neurons may provide an indirect mechanism to inhibit DA^VTA neurons, which raises the possibility that α-MSH signals within these GABA^VTA neurons may be partly responsible for anhedonia and depression-like behaviors triggered by chronic restraint stress. Consistently, global deletion of MC3R results in increased dopamine bioavailability in the VTA in mice³⁷. Further, administration of MTII, the MC3R/MC4R agonist, into the VTA decreases food intake and sucrose preference in rats, while the MC3R/MC4R antagonist (SHU9119) causes opposite effects^54–56. However, global deletion of MC3R leads to decreased sucrose preference only in female mice but not in male mice³⁷, phenotypes that may have been confounded by loss of MC3R in all types of cells since early development. Thus, the specific roles of MC3R in different subgroups of VTA neurons during adulthood warrant further investigations.

We also found that chronic photoinhibition of the POMC^ARH→VTA projections increases food intake and body weight in RST mice. Consistently, chronic chemogenetic stimulation of POMC^ARH neurons decreases food intake in mice, although acute stimulation of these neurons fails to alter feeding behavior⁵⁷. However, most of anorexigenic effects of POMC^ARH neurons are thought to be mediated by MC4R expressed in the paraventricular nucleus of hypothalamus (PVH)⁵⁸. Our results suggest that the POMC^ARH→VTA circuit is an alternative pathway by which POMC^ARH neurons can reduce food intake at least in the context of chronic stress. Supporting a feeding-suppressing effect of α-MSH in the VTA, it has been reported that injections of MT-II into the VTA decrease food intake, while intra-VTA injections of SHU9119 increase feeding^{55, 56}. Interestingly, knockdown of MOR in the VTA prevents body weight loss induced by social defeat⁵³. Thus, the MOR-mediated neurotransmission we observed in the VTA may also contribute to hypophagia and body weight loss during chronic restraint stress. This possibility remains to be investigated.

Neurocircuits regulating mood and feeding behavior are at least partly overlapping, which provides important defensive mechanisms for animals to coordinate internal states (e.g. energy balance and mood) with complex external environment (e.g. food availability, danger and stress)⁵⁹. In particular, hunger activates AgRP^ARH neurons, which reduce anxiety^{60, 61}, fear⁶² and pain⁶³, and ultimately facilitate food-seeking behavior to ensure survival. Here we showed that alternations of the POMC^ARH→VTA circuit during stress result in abnormal feeding and mood, which is another example that feeding behavior, depression-like behavior and anhedonia are not separate entities, but rather they may influence each other reciprocally. POMC^ARH neurons, via their projections to the reward circuitry, provide one neurobiological basis to coordinate feeding and mood regulations in response to chronic stress. Consistent with our findings in mice, the neural activities of the VTA in human brains are correlated with craving behavior⁶⁴. Further, humans with depression often display reduced neural connectivity from the VTA to its downstream targets, including the striatum⁶⁵ and the dorsal anterior cingulate cortex⁶⁶. Interestingly, mutations in the gene encoding MOR have been associated with human depression⁶⁷, supporting the possibility that POMC-originated signals, e.g. β-endorphin, contribute to development of depression, and that the POMC^ARH→VTA circuit may exist in human brains. Nevertheless, further investigations are needed to confirm such POMC^ARH→VTA circuit in human brains and to explore its role in human depression.

Methods

Mice

Multiple lines of transgenic mice were used in the current study. For electrophysiological studies, we crossed Rosa26-LSL-tdTOMATO allele⁶⁸ (obtained from Jackson Laboratory, #007905) onto DAT-CreER (Jackson Laboratory, #016583), POMC-CreER⁶⁹, AgRP-IRES-Cre mice⁷⁰ or Vgat-IRES-Cre mice⁷¹ (Jackson Laboratory, #028862), respectively. These crosses generated mice with TOMATO selectively expressed in DA, POMC (after tamoxifen inductions), AgRP or GABAergic neurons. We crossed C57Bl6j mice with POMC-Cre mice⁷² to generate POMC-Cre and wild-type littermates. We crossed POMC-CreER, DAT-CreER and Rosa26-LSL-tdTOMATO mice to generate POMC-CreER/DAT-CreER/Rosa26-LSL-tdTOMATO mice. All the breeders have been backcrossed to C57Bl6j background for more than 12 generations. In addition, some C57Bl6j mice were purchased from the mouse facility of Baylor College of Medicine. Mice were housed in a temperature-controlled environment in groups of two to five at 22°C-24°C using a 12 hr light/12 hr dark cycle. The mice were fed standard chow (6.5% fat, #2920, Harlan-Teklad, Madison, WI) ad libitum. Water was provided ad libitum.

RST Protocol

The experiments were performed after the animals had been habituated to the experimental environment for 1 week. In the study described in Figure 1, male C57Bl6j mice (3 months of age) were divided into two weight-matched groups: control group (CON) and restraint stress group (RST). The RST mice were exposed daily for 14 days to 1-hour restraint (9:30–10:30am) in rodent restraint plastic bags (Triangle-shaped Restraints 1800c #89066–356, VWR AIMS_TM) with holes that permit the mice breathing but restrict the movement of the limbs; the CONT mice were left intact. In the study described in Figure 7, male WT and POMC-Cre littermates were anesthetized with isoflurane and received stereotaxic injections of the silencer of AAV-hSyn-eNpHR3.0-EYFP virus (University of North Carolina Gene Therapy Center, Chapel Hill, NC) into both sides of the ARH (250 nL/site, −1.7 mm posterior, ±0.25 mm lateral and −5.8 mm ventral to the Bregma, based on Franklin & Paxinos Mouse Brain Atlas). During the same surgery, an optic fiber was implanted to target the VTA (unilateral, −3 mm posterior, 0.5 mm lateral and −4.3 mm ventral to the Bregma). After a 4-week recovery, these mice (4 months of age) were subjected to the 14-day RST protocol as described above, and during the daily 1-hour restraint, the mice also received yellow light stimulation (589 nm, 10 ms/pulse, 3 pulses per 1 s for 1 hour; MGL-FN-589, CNI LASER). A small hole was made in the restraint plastic bag to allow insertion of the optic fiber. After being restrained, the mice were returned to their home cages and given food and water ad libitum. The food intake and body weight of the mice were monitored daily.

Behavioral tests

All behavioral tests were performed during the afternoon in a dedicated sound-proof behavioral facility by experimenters blind to treatment and genotype information. Mice were brought to the procedure room 1 hour before the start of each behavioral test and remained in the same room throughout the test. At all times, sound was masked with 60–65 Db white noise.

Mice were allowed to rest for one day after the completion of the 14-day RST protocol. Then, sucrose preference tests were performed on day 16. A bottle with tap water and another bottle with 2℅ sucrose solution were given to mice. The mice had been trained (given a free choice between two bottles) for 1 hour every day (12:30pm-1:30pm, on day 8 to day 14) to avoid neophobia. To prevent a possible effect of behavior, the left/right location of the bottles was switched every day. The sucrose preference was calculated as the ratio of consumed sucrose solution to consumed water.

Forced swim tests were performed on day 17. In brief, mice were placed in a glass cylinder (height: 50 cm, diameter: 20 cm) containing water at 23–25 °C and depth of 14 cm, such that mice could neither escape nor touch the bottom. Mice were forced to swim for 6 min. Mice were acclimated for the first 1 min and immobility time was measured for the last 5 min.

Tail suspension tests were performed on day 18. Mice were suspended individually by their tails from a metal rod fixed 30 cm above the surface of a table. The tip of the tail was fixed using adhesive tape (the distance from tip of tail was 2 cm). The subject mouse was suspended by the tail for 6 min. Mice were acclimated for the first 1 min and immobility time was measured for the last 5 min.

All mice that received stereotaxic injections were perfused with 10% formalin after the completion of the behavioral tests. Brain sections were collected, and expression of EYFP was checked in the ARH. Only those with expression of EYFP exclusively in the ARH were included in data analyses. All behavioral analyses were performed by experimenters who were blinded with the treatment groups.

Electrophysiology

To examine the activities of DA^VTA, POMC^ARH and AgRP^ARH neurons in RST mice, DAT-CreER/Rosa26-LSL-tdTOMATO mice (after tamoxifen induction), POMC-CreER/Rosa26-LSL-tdTOMATO mice (after tamoxifen induction), and AgRP-IRES-Cre/Rosa26-LSL-tdTOMATO mice were subjected to RST or CON protocol as described above, for 4 days. Immediately after the last session of restraint (or control), these mice were deeply anesthetized with isoflurane and perfused with a modified ice-cold sucrose-based cutting solution (pH 7.3) containing 10 mM NaCl, 25 mM NaHCO₃, 195 mM Sucrose, 10 mM Glucose, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 2 mM Na-Pyruvate, 0.5 mM CaCl₂, and 7 mM MgCl₂, bubbled continuously with 95% O₂ and 5% CO₂ (Ren et al., 2012). The mice were then decapitated, and the entire brain was removed and immediately submerged in the cutting solution. Slices (250 μm) were cut with a Microm HM 650V vibratome (Thermo Scientific). Coronal brain slices containing the VTA or ARH were obtained for each animal. The slices were recovered for 1 h at 34°C and then maintained at room temperature in artificial cerebrospinal fluid (aCSF, pH 7.3) containing 126 mM NaCl, 2.5 mM KCl, 2.4 mM CaCl₂, 1.2 mM NaH₂PO₄, 1.2 mM MgCl₂, 5.0 mM glucose, and 21.4 mM NaHCO₃) saturated with 95% O₂ and 5% CO₂ before recording.

Slices were transferred to a recording chamber and allowed to equilibrate for at least 10 min before recording. The slices were superfused at 34°C in oxygenated aCSF at a flow rate of 1.8–2 ml/min. TOMATO-labeled neurons were visualized using epifluorescence and IR-DIC imaging on an upright microscope (Eclipse FN-1, Nikon) equipped with a movable stage (MP-285, Sutter Instrument). Patch pipettes with resistances of 3–5 MΩ were filled with intracellular solution (pH 7.3) containing 128 mM K-Gluconate, 10 mM KCl, 10 mM HEPES, 0.1 mM EGTA, 2 mM MgCl₂, 0.05 mM Na-GTP and 0.05 mM Mg-ATP. Recordings were made using a MultiClamp 700B amplifier (Axon Instrument), sampled using Digidata 1440A and analyzed offline with pClamp 10.3 software (Axon Instruments). Series resistance was monitored during the recording, and the values were generally <10 MΩ and were not compensated. The liquid junction potential was +12.5 mV, and was corrected after the experiment. Data were excluded if the series resistance increased dramatically during the experiment or without overshoot for action potential. Currents were amplified, filtered at 1 kHz, and digitized at 20 kHz. Current clamp was engaged to test neural firing frequency and resting membrane potential⁷³. The values were averaged within 2-min bin. After-hyperpolarization potential (AHP) of DA^VTA neurons was calculated by substracting the most negative membrane potential right after the decay of each action potential to the onset membrane potential before the action potential. The AHP value from each neuron was averaged by 3 different single AHP in the same trace. To analyze these parameters (firing frequency, resting membrane potential and AHP), we first treated each neuron as an independent biological sample; in addition, we also calculated the average of all neurons from the same mouse and treated each mouse as an independent biological sample. Both analyses were presented in the figures.

To examine effects of MT-II on GABA^VTA neurons, the VTA-containing brain slices were prepared from 8-week old Vgat-IRES-Cre/Rosa26-LSL-tdTOMATO mice, as described above. The brain slices were incubated with the cocktail inhibitors containing 1μM TTX, 30μM CNQX, 30 μM D-AP5, 50 μM bicuculline in order to block most of the synaptic inputs on GABA^VTA neurons. 300 nM MT-II was perfused to treat the GABA^VTA neurons in the presence of the cocktail blockers. The change of resting membrane potential after MT-II treatment was recorded.

After recording of each neuron, a lucifer yellow dye was injected into the recorded neuron via the pipette. Slices were fixed with 4% formalin in PBS at 4°C overnight and then subjected to post hoc identification of the anatomical location of the recorded neurons within the ARH or the VTA.

Neurotracing

To determine whether POMC^ARH neurons project to the VTA, twelve-week old POMC-Cre mice were anesthetized by isoflurane and received stereotaxic injections of Cre-dependent AAV expressing Synaptophysin-GFP (AAV-FLEX-Synaptophysin-GFP) into the ARH (200 nl). Four weeks after injections, mice were perfused with 10% formalin, and brain sections were cut at 25 μm (5 series). These brain sections were subjected to immunofluorescence staining for TH with rabbit anti-Tyrosine hydroxylase antibody (1:500, AB152, Millipore, Temecula, CA) overnight, followed by the goat anti-rabbit AlexaFluor 405 (1:500, A31556, Invitrogen) for 1.5 h. Slides were cover-slipped and analyzed using a Leica DM5500 fluorescence microscope with OptiGrid structured illumination configuration.

To determine whether DA^VTA neurons receive projections from POMC^ARH neurons, twelve-week old DAT-CreER mice were anesthetized by isoflurane and received stereotaxic injections of AAV8-EF1a-FLEX-GTB (1.16E+12 GC/ml, Salk Institute for Biological Studies) in the VTA (200nl). These mice were recovered for 4 days and received tamoxifen (0.2mg/g, i.p) injection to induce the expression of Cre in DA neurons. One week after tamoxifen injection, the mice received the second stereotaxic injections of EnvA G-deleted Rabies-mCherry (1.08E+9 TU/ml, Salk Institute for Biological Studies) virus in the VTA (200nl). Four weeks after, the mice were perfused as above. Brain sections were then subjected to immunofluorescent staining for β-endorphin. Briefly, sections were incubated with primary rabbit anti-β-endorphin antibody (1:10000, H-022–33, Phoenix Peptide) overnight, followed by the donkey anti-rabbit AlexaFluor 488 (1:500, A21206, Invitrogen) for 1.5 h. The sections were mounted on glass slides and fluorescence images were analyzed using a Leica DM5500 fluorescence microscope with OptiGrid structured illumination configuration. The DA^VTA neurons infected with both AAV8-FLEX-GTB and Env-dG-Rabies and the retrograde traced neurons were labeled by mCherry (red), and POMC neurons were labeled by β-endorphin (green).

WGA-CRACM

We performed the channelrhodopsin-2 (ChR2)-assisted circuit mapping (CRACM), similarly as described by others^{74, 75}, in order to examine the neurotransmissions between POMC^ARH neurons and VTA neurons. Briefly, 12-week old POMC-CreER/DAT-CreER/Rosa26-LSL-tdTOMATO mice were anesthetized by isoflurane and received stereotaxic injections of Ad-FLEX-WGA-GFP (100 nl/site) and AAV-EF1α-DIO hChR2(H134R)-EYFP (100 nl/site) into the ARH. Tamoxifen (0.2 mg/g, i.p.) was injected into these mice to induce POMC-CreER and DAT-CreER. After a 4-week recovery, mice were sacrificed and unfixed brain slices (containing both the ARH and VTA, 120 μm in thickness) were prepared from these mice. These brain slices were subjected to fluorescent microscopy to visualize and quantify GFP(+), TOMATO(+) and GFP(+)/TOMATO(+) neural soma in the VTA. Notably, EYFP expressed by the AAV-EF1α-DIO hChR2(H134R)-EYFP shares overlapping spectra with GFP. However, EYFP will only label the fibers and terminals in the VTA but will not cross the synapse to label the soma of downstream neurons. Thus, the GFP-labelled soma within the VTA must be filled by WGA-GFP fusion proteins. About 150–160 neural soma were counted from each mouse and 3 mice were included. These brain slices were also used to perform electrophysiological recordings, as described below.

Under the voltage clamp mode, light-evoked EPSCs or IPSCs were recorded in GFP(+), TOMATO(+) or GFP(+)/TOMATO(+) VTA neurons in brain slices in response to 473 nM laser blue light (10 Hz, 10 ms pulse, 0.2 mW; MBL-FN-473, CNI LASER). The light sThe light-evoked EPSCs were recorded in whole-cell voltage-clamp mode, by holding the membrane potential at Vh = −60 mV. The pipette solution containing: 125 mM CsCH₃SO₃; 10 mM CsCl; 5 mM NaCl; 2 mM MgCl₂; 1 mM EGTA; 10 mM HEPES; 5 mM (Mg)ATP; 0.3 mM (Na)₂GTP (pH 7.3 with NaOH). The light evoked IPSCs were recorded in whole-cell voltage-clamp mode, by holding the membrane potential at Vh = −70 mV. The CsCl-based pipette solution containing of: 140mM CsCl, 10mM HEPES, 5mM MgCl₂, 1mM BAPTA, 5mM (Mg)ATP, and 0.3mM (Na)₂GTP (pH 7.30 adjusted with NaOH; 295 mOsm kg⁻¹). TTX (1μM) and 4-AP (400 μM) were added to the aCSF in order to confirm the response was monosynaptic responses. EPSCs were recorded when 50 μM bicuculline was applied in the bath solution; after evoked EPSC was detected, 30 μM D-AP5 and 30 μM CNQX were used to confirm the current was mediated by glutamate receptors. IPSCs were recorded when 30 μM D-AP5 and 30 μM CNQX were applied in the bath solution; after evoked IPSC was detected, 50 μM bicuculline was used to confirm the current was mediated by GABA_A receptor. In some experiments, 50 μM SHU9119 was treated after evoked IPSC was detected. Under the current clamp mode, 30 μM D-AP5, 30 μM CNQX and 50 μM bicuculline were applied in the bath solution to block the glutamate and GABA inputs; action potential firing rate and resting membrane potential were recorded in response to 473 nM blue light (10 Hz, 10 ms pulse, 0.2 mW) in the absence or presence of various blockers, such as 50 μM SHU9119⁷⁶, 100 μM naltrindole⁷⁷, 100 μM norbinaltorphimine⁷⁸, 200 μM cyprodime⁷⁹.

Single neuron RT-PCR

To determine the neurochemical identify of POMC^ARH-innervated VTA neurons in the WGA-CRACM study, we manually picked up GFP(+) neurons and GFP(+)/TOMATO(+) neurons, and suspended in PBS buffer. Briefly, the mice brain was removed and immediately submerged in ice-cold sucrose-based cutting solution (adjusted to pH 7.3) containing (in mM) 10 NaCl, 25 NaHCO3, 195 Sucrose, 5 Glucose, 2.5 KCl, 1.25 NaH₂PO₄, 2 Na pyruvate, 0.5 CaCl₂, 7 MgCl₂ bubbled continuously with 95% O₂ and 5% CO₂. The slices (250 μm) were cut with a Microm HM 650V vibratome (Thermo Scientific and recovered for 1 h at 34°C and then maintained at room temperature in artificial cerebrospinal fluid (aCSF, pH 7.3) containing 126 mM NaCl, 2.5 mM KCl, 2.4 mM CaCl₂, 1.2 mM NaH₂PO₄, 1.2 mM MgCl₂, 11.1 mM glucose, and 21.4 mM NaHCO₃ saturated with 95% O₂ and 5% CO₂ before recording. Slices were transferred to a chamber, and GFP(+) neurons and GFP(+)/TOMATO(+) neurons were visualized using epifluorescence and IR-DIC imaging on an upright microscope equipped with a moveable stage (MP-285, Sutter Instrument). Single neurons were manually picked up by the pipette and 2 neurons were combined as a sample for RNA extraction and reverse transcription using the Ambion Single-Cell-to-CT Kit (Ambion, Life Technologies) according to the manufacturer’s instruction. Briefly, 30 μl Single Cell Lysis solutions with DNase I was added to each sample, and the supernatant after centrifuge were used for cDNA synthesis (25°C for 10 min, 42°C for 60 min, and 85°C for 5 min). The cDNA samples were amplified on a CFX384 Real-Time System (Bio-Rad) using SsoADV SYBR Green Supermix (Bio-Rad). Results were normalized against the expression of house-keeping gene (β-actin). Primer sequences for β-actin, DAT and Vgat and DAT: β-actin forward, ATGGAGGGGAATACAGCCC; β-actin reverse, TTCTTTGCAGCTCCTTCGTT; DAT forward, TGATGCACATAGCAGCAACTC; DAT reverse, AGGTCATCAATGCCACGACT; Vgat forward, CGACAAACCCAAGATCACGG; Vgat reverse, AGGATGGCGTAGGGTAGGC; MC3R forward, TCCGATGCTGCCTAACCTCT; MC3R reverse, GGATGTTTTCCATCAGACTGACG; MC4R forward, GGTCGGAAACCATCGTCATT; MC4R reverse, AAAGCAGGCTGCAAATGGAT; MOR forward, CCAGGGAACATCAGCGACTG; MOR reverse, GTTGCCATCAACGTGGGAC.

Statistics

The minimal sample size was pre-determined by the nature of experiments. For most of physiological readouts (body weight, food intake, etc.), 5–7 mice per group were included. For histology studies, 3 mice were included in each group. For electrophysiological studies, at least 10 neurons in each genotype or condition were included. The data are presented as mean ± SEM. Statistical analyses were performed using GraphPad Prism to evaluate normal distribution and variations within and among groups. Methods of statistical analyses were chosen based on the design of each experiment and are indicated in figure legends. P<0.05 was considered to be statistically significant.

Study approval

Care of all animals and procedures were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee.