Organization of Functional Long-Range Circuits ... - CyberLeninka

Article

Organization of Functional Long-Range Circuits Controlling the Activity of Serotonergic Neurons in the Dorsal Raphe Nucleus Graphical Abstract

Authors Li Zhou, Ming-Zhe Liu, Qing Li, Juan Deng, Di Mu, Yan-Gang Sun

Correspondence [email protected]

In Brief Zhou et al. used slice physiological recording combined with optogenetics to systematically study the long-range functional input from six key upstream brain areas to both serotonergic and GABAergic neurons in the dorsal raphe nucleus. The results reveal the fine circuit mechanisms that functionally balance the activity of serotonergic neurons.

Highlights d

Systematic mapping of the functional organization of longrange inputs to the DRN

d

Upstream brain areas make synapses with serotonergic neurons in different patterns

d

Lateral hypothalamus controls serotonergic neurons via a push-pull mechanism

d

LHb recruits feedforward inhibition to control the activity of serotonergic neurons

Zhou et al., 2017, Cell Reports 18, 3018–3032 March 21, 2017 ª 2017 The Author(s). http://dx.doi.org/10.1016/j.celrep.2017.02.077

Cell Reports

Article Organization of Functional Long-Range Circuits Controlling the Activity of Serotonergic Neurons in the Dorsal Raphe Nucleus Li Zhou,1,2 Ming-Zhe Liu,1,2 Qing Li,1 Juan Deng,1,2 Di Mu,1,2 and Yan-Gang Sun1,3,*

1Institute of Neuroscience, State Key Laboratory of Neuroscience, CAS Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences, 320 Yue-Yang Road, Shanghai 200031, China 2Graduate School, University of Chinese Academy of Sciences, 320 Yue-Yang Road, Shanghai 200031, China 3Lead Contact *Correspondence: [email protected] http://dx.doi.org/10.1016/j.celrep.2017.02.077

SUMMARY

Serotonergic neurons play key roles in various biological processes. However, circuit mechanisms underlying tight control of serotonergic neurons remain largely unknown. Here, we systematically investigated the organization of long-range synaptic inputs to serotonergic neurons and GABAergic neurons in the dorsal raphe nucleus (DRN) of mice with a combination of viral tracing, slice electrophysiological, and optogenetic techniques. We found that DRN serotonergic neurons and GABAergic neurons receive largely comparable synaptic inputs from six major upstream brain areas. Upon further analysis of the fine functional circuit structures, we found both bilateral and ipsilateral patterns of topographic connectivity in the DRN for the axons from different inputs. Moreover, the upstream brain areas were found to bidirectionally control the activity of DRN serotonergic neurons by recruiting feedforward inhibition or via a push-pull mechanism. Our study provides a framework for further deciphering the functional roles of long-range circuits controlling the activity of serotonergic neurons in the DRN. INTRODUCTION Neuromodulators play key roles in dynamically gating the neural network (Bargmann and Marder, 2013). Deficits in the neuromodulatory system lead to numerous neurological disorders (Caspi et al., 2003; Lesch and Waider, 2012; Lotharius and Brundin, 2002; Zill et al., 2004). Serotonin (5-HT) is one of the key neuromodulators and plays critical roles in sensory processing, sleep, reward, emotional regulation, impulsivity, decision making, and neural plasticity (Cools et al., 2008; Homberg, 2012; Lesch and Waider, 2012; Nakamura, 2013; Walker et al., 2006). In the central nervous system, serotonergic neurons are mostly located in the midbrain and can be subdivided into nine groups (Lesch and Waider, 2012). Most of the forebrain-projec-

ting 5-HT neurons are located in the dorsal raphe nucleus (DRN), which contains more than half of the 5-HT neurons in the rodent brain (Descarries et al., 1982). The DRN 5-HT neurons project to most of the forebrain structures, some of which might be the targets of antipsychotic and antidepressant drugs. Although 5-HT neurons are traditionally thought to be homogeneous, they are different in molecular identity, electrophysiological property, and downstream targets (Fernandez et al., 2016; Okaty et al., 2015), suggesting distinct functions of subpopulations of 5-HT neurons. This is supported by pharmacological and histological experiments (Hale and Lowry, 2011). However, it remains unknown whether subpopulations of 5-HT neurons are differentially regulated by the upstream brain areas. Dysfunction of the serotonergic system is implicated in depression and anxiety (Caspi et al., 2003; Sachs et al., 2015). Thus, activity of the 5-HT neurons has to be tightly controlled in order to maintain emotional homeostasis. Previous traditional anatomical tracing studies have shown that the DRN receives inputs from multiple brain areas (Jacobs and Azmitia, 1992), but these studies did not illustrate the synaptic connection between the upstream brain areas and 5-HT neurons in the DRN. With the recently developed rabies-virus-tracing technique, three groups independently studied the presynaptic partners of the DRN 5-HT neurons (Ogawa et al., 2014; Pollak Dorocic et al., 2014; Weissbourd et al., 2014). These studies demonstrated that 5-HT neurons in the DRN are regulated through top-down mechanisms, suggesting that 5-HT neurons serve to integrate information from diverse upstream brain areas. However, how the different brain areas contribute to the functional modulation of DRN 5-HT neurons, especially the functional synaptic connectivity and connectivity pattern of the different inputs beyond structural connection, remains unclear. GABAergic neurons of the DRN also play an important role in the regulation of 5-HT neurons (Challis et al., 2013), possibly via local connection. The same sets of presynaptic partners that innervate 5-HT neurons also make synapses with the GABAergic neurons in the DRN (Weissbourd et al., 2014). It remains to be examined how the local GABAergic neurons are differentially recruited by the long-range input to dynamically regulate the activity of 5-HT neurons. Here, we systematically studied the functional synaptic connection from the upstream brain areas to the 5-HT neurons

3018 Cell Reports 18, 3018–3032, March 21, 2017 ª 2017 The Author(s). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

as well as GABAergic neurons in the DRN with channelrhodopsin-2 (ChR2)-assisted circuit mapping (CRACM) (Petreanu et al., 2007). RESULTS Serotonergic Neurons in DRN Receive Monosynaptic Glutamatergic Input from LHb The lateral habenula (LHb) is implicated in emotional regulation, which is possibly mediated by the LHb-DRN pathway (Amo et al., 2014). We examined the functional synaptic connectivity between LHb and DRN serotonergic neurons with CRACM (Petreanu et al., 2007). The LHb was an example for the experimental procedure, which was also applied to the other brain areas. We utilized the ePet1-Cre:Rosa-tdTomato(Ai9) mice, referred hereafter as Pet1-tdTomato mice, to label the serotonergic neurons in the CNS with red fluorescence. The specificity of the Pet1-tdTomato mice was confirmed by immunofluorescent staining of 5-HT neuron markers. We found that 91.0% of tdTomato-positive neurons in the DRN were positive for serotonin (457/502; eight sections from three mice; Figures S1A and S1B), and 91.7% of them were positive for Tph2, a rate-limiting enzyme for 5-HT synthesis (742/809; nine sections from three mice; Figures S1C and S1D). Next, we infected the LHb of wild-type mice with an adenoassociated virus (AAV) expressing ChR2 fused with mCherry (Figure 1A) and found that mCherry+ fibers were in close contact with the serotonergic neurons in the DRN (Figure 1B). Photostimulation of the ChR2-expressing neurons in the LHb induced inward current and triggered action potentials reliably in ChR2+ neurons (Figures S1E–S1H). To examine the functional synaptic connection between LHb and DRN 5-HT neurons, we injected AAV-ChR2-mCherry into the LHb of Pet1-tdTomato mice and performed patch-clamp recordings in acute brain slices containing the DRN of Pet1-tdTomato mice (Figures 1A and 1C). The putative serotonergic neurons labeled with tdTomato, which were further confirmed to be serotonergic by post hoc immunostaining for the 5-HT neuron marker Tph2 (Figure 1D), exhibited spiking activity in response to depolarizing current injection (Figure 1E) with a relatively wide half-width, consistent with previous reports (Gocho et al., 2013). Moreover, photostimulation (475 nm; 1 ms) in the DRN evoked excitatory postsynaptic currents (EPSCs) in the DRN 5-HT neurons (Figure 1F). The latency of the light-evoked EPSCs was 2.0 ± 0.1 ms (Figure 1G), indicating a monosynaptic connection between LHb and DRN 5-HT neurons. We also analyzed the latency jitter for these neurons; we found that most of the neurons (87/113 neurons; 77.0%) have jitter less than 0.4 ms (Figures S1I and S1J), consistent with monosynaptic connection. For some neurons, the light-evoked EPSCs have multiple peaks (Figure S1J) or have longer jitter (Figure S1K), indicative of possible polysynaptic response. To further confirm the monosynaptic connection, we used tetrodotoxin (TTX)-4-aminopyridine (4-AP) methods (Petreanu et al., 2009). The light-evoked EPSCs were abolished after bath application of TTX (1 mM) but were then reintroduced following bath application of 4-AP (100 mM; in the presence of TTX), a voltage-dependent potassium (K+) channel blocker (Figures 1F and 1H). This result confirmed a monosynaptic connection be-

tween LHb and DRN 5-HT neurons. The light-evoked EPSCs were blocked by an AMPA receptor antagonist NBQX (10 mM), suggesting a glutamatergic nature of this connection (Figures 1F, 1I, and 1J). Out of 210 serotonergic neurons recorded in the DRN (from 18 mice), 147 neurons exhibited light-evoked EPSCs (Figure 1K) with an averaged amplitude of 171.7 ± 18.2 pA (Figure 1L). These results demonstrate that DRN serotonergic neurons receive monosynaptic functional glutamatergic input from the LHb. The Synaptic Properties of Major Functional Inputs to the DRN 5-HT Neurons Numerous brain areas have been shown to potentially make synapses with DRN 5-HT neurons by viral tracing methods (Ogawa et al., 2014; Pollak Dorocic et al., 2014; Weissbourd et al., 2014). We thus determined how the major upstream brain areas control the activity of DRN serotonergic neurons by examining the functional synaptic input to DRN 5-HT neurons from the major upstream brain areas, including prefrontal cortex (PFC), lateral hypothalamus (LH), preoptic area (POA), substantia nigra (SN), and amygdala (AMY), with CRACM. The PFC plays an important role in modulating emotional responses via the PFC-DRN pathway (Warden et al., 2012). We thus defined the property of PFC-DRN projection. We injected AAV-ChR2-mCherry into the PFC of Pet1-tdTomato mice. Photostimulation of the ChR2+ fibers in the DRN evoked short-latency EPSC in 79.3% of DRN serotonergic neurons (65/82; Figures 2A, 2B, 2D, 2V, S2A, and S2B), which was blocked by bath application of NBQX (Figures 2A and 2C), suggesting that the PFC sends glutamatergic projections to DRN 5-HT neurons. This is consistent with a previous study showing that the PFC provides direct synaptic connection to DRN 5-HT neurons (Geddes et al., 2016). In a small percentage of neurons (3/82 neurons), we also observed light-evoked inhibitory postsynaptic current (IPSC), which was not further examined due to the low probability. The percentage of 5-HT neurons receiving EPSC and/or IPSC was shown in Figure 2U, same for the other inputs. The hypothalamus is important for feeding and reward seeking (Stuber and Wise, 2016) and represents another major source of inputs to the DRN, with both LH and POA projecting extensively to the DRN. To investigate the functional synaptic connection between LH and DRN 5-HT neurons, we injected Pet1-tdTomato mice with AAV-ChR2-mCherry in the LH and found that photostimulation of the ChR2+ fibers in the DRN induced EPSC and/or IPSC in 67.0% of DRN serotonergic neurons (71/106; Figures 2E, 2F, 2H, and 2U). The EPSCs and IPSCs were blocked by bath application of the AMPA receptor antagonist NBQX and GABAA receptor antagonist picrotoxin, respectively (Figures 2E and 2G). The short latencies of light-evoked EPSCs and IPSCs (Figures 2V and 2W) indicate that LH is connected with DRN 5-HT neurons via both monosynaptic glutamatergic and GABAergic connections. There could also be a small amount of polysynaptic connections in the DRN, as evidenced by the presence of a small number of postsynaptic currents (PSCs) with longer latencies (Figures S2C and S2D). Similar to the LH, the POA also provided excitatory and/or inhibitory synaptic inputs to 60.7% of serotonergic neurons in the DRN (54/89 neurons; Figures 2I–2L and 2U) with short latencies (Figures 2V,

Cell Reports 18, 3018–3032, March 21, 2017 3019

E

I

C

B

A

F

D

G

J

H

K

L

Figure 1. DRN Serotonergic Neurons Receive Monosynaptic Glutamatergic Input from the LHb (A) Schematic graph of the experiment. (B) Micrograph showing mCherry+ axons (red) and serotonergic neurons (green; 5-HT) in the DRN. The scale bar represents 200 mm. The inset shows the enlargement of the boxed area. The scale bar represents 5 mm. (C) In Pet1-tdTomato mice, whole-cell patch-clamp recordings were made from tdTomato+ neurons. Biocytin-filled cells (green) were reconstructed after recording. The scale bar represents 100 mm. (D) Post hoc staining of a recorded tdTomato+ neuron for Tph2 (green) and biocytin (blue). The scale bar represents 10 mm. (E) The response of a tdTomato+ neuron to current steps. (F) A light-evoked EPSC recorded in a DRN tdTomato+ neuron in ACSF and following sequential application of TTX, 4-AP, and NBQX. Blue bar, LED stimulation (475 nm; 1 ms). (G) Distribution of the latency of EPSCs in the DRN tdTomato+ neurons (n = 146 neurons). The inset shows the averaged EPSC latency. (H) Summary showing the EPSC amplitude before and following application of TTX (1 mM) and 4-AP (100 mM; n = 5). (I) An example experiment showing the NBQX-induced blockade of light-evoked EPSC. (J) Summary data showing the EPSC amplitude before and after NBQX (10 mM; p < 0.001; Wilcoxon signed rank test; n = 12). (K) Summary of the spatial location and connectivity (pie graph; 147/210 neurons received synaptic inputs). (L) Distribution of the EPSC amplitude. The inset shows the averaged EPSC amplitude. Data are represented as mean ± SEM. See also Figure S1.

2W, S2E, and S2F). These data suggest that hypothalamus could bidirectionally regulate the DRN serotonergic neurons. The midbrain SN receives input from DRN serotonergic neurons (Vertes, 1991), and this input may be important to maintain the balance in the basal ganglia circuit. We found that SN neurons were highly connected with 5-HT neurons in the DRN via glutamatergic and/or GABAergic synapses (80.4%; 41/51 neurons; Figures 2M–2P and 2U) with short latencies (Figures 2V, 2W, S2G, and S2H), indicating strong monosynaptic connection between SN and DRN 5-HT neurons. The amygdala has been shown to be connected with DRN 5-HT neurons by viral tracing studies. By recording from the

3020 Cell Reports 18, 3018–3032, March 21, 2017

5-HT neurons labeled with tdTomato in the DRN, we found that the functional connection between amygdala and DRN serotonergic neurons was mediated by glutamatergic and/or GABAergic synapses (Figures 2Q, 2S, 2T, 2V, 2W, S2I, and S2J). The connectivity, percentage of neurons receiving synaptic inputs, was relatively low (24.5%; 23/94; Figures 2R and 2U). Distinct Patterns of Topographic Innervation between the Upstream Brain Areas and the DRN The upstream brain areas of DRN 5-HT neurons selected in this study have dramatically different functional roles. It is possible that these brain areas are connected with different groups of

A

B

C

E

F

G

H

I

J

K

L

M

N

O

P

Q

R

S

T

U

V

D

W

Figure 2. The DRN Serotonergic Neurons Receive Functional Inputs from Multiple Brain Areas (A) (Top) Schematic diagram of the experiment to examine the functional connection from PFC to DRN serotonergic neurons. (Bottom) A representative experiment shows light-evoked EPSCs before (black; in ACSF) and after bath application of NBQX (10 mM; red). Blue bar, LED stimulation (475 nm; 1 ms); same for (E), (I), (M), and (Q). (B) Summary of the spatial location and connectivity of the recorded serotonergic neurons (n = 82 neurons). (C) Summary data showing the amplitude of EPSC evoked by photostimulaton of PFC ChR2+ fibers before and after bath application of NBQX (n = 10 neurons).

(legend continued on next page)


DRN 5-HT neurons, given that subareas of DRN are implicated in different biological processes (Hale and Lowry, 2011). We thus sought to investigate whether the upstream brain areas making synaptic connection with the DRN 5-HT neurons are characterized by different topographic patterns. We first analyzed the patterns of topographic innervation of the different inputs in the DRN after unilaterally injecting AAV-ChR2-mCherry into the PFC, LHb, POA, LH, SN, and amygdala, respectively. Topographic patterns of the axonal projection were different among different inputs (Figures 3A and S3). Notably, PFC and LHb sent axons to the DRN bilaterally, and the distribution of axon terminals was relatively symmetric. We quantified the intensity of mCherry+ fibers across four subregions of the DRN, including the dorsal medial part, the ventral medial part, the ipsilateral lateral wing, and the contralateral lateral wing. The lateral wings are defined as the area that contains serotonergic neurons located in the ventrolateral DRN and ventrolateral periaqueductal gray, which is restricted to the mid-rostrocaudal part of DRN (Hale and Lowry, 2011). We mainly focused on the ipsilateral lateral wing and contralateral lateral wing of DRN to further explore symmetric property of the projection pattern. We found that the axonal projection of PFC and LHb had no obvious bias between the contralateral lateral wing and ipsilateral lateral wing (Figures 3B and 3C). In contrast, POA, LH, SN, and amygdala sent unilateral projections to the DRN, with more axons located within the ipsilateral lateral wing compared with the contralateral lateral wing (Figure 3A), which was further confirmed by quantitative analysis (Figures 3D–3G). We took PFC as an example to examine whether the bilateral axonal projection pattern matches with functional synaptic connectivity. We unilaterally injected AAV-ChR2-mCherry into the PFC of Pet1-tdTomato mice and sampled the light-evoked postsynaptic response of the serotonergic neurons within four subregions of DRN in each acute brain slice (Figure 3H). The functional synaptic connectivity between PFC and DRN 5-HT neurons was comparable across four subregions (p = 0.23; Chi-square test; Figures 3I, 3J, and 3N), suggesting that PFC provides a bilaterally distributed functional innervation to serotonergic neurons. We further compared the synaptic strength of PFC input to DRN 5-HT neurons among different subregions and found that the EPSC amplitudes were comparable for all subfields except

that the neurons in ipsilateral wing receive relatively weaker input (Figure 3O). We next took POA as an example to examine whether the ipsilateral axon projection pattern matches with functional connectivity. The connectivity between POA and the contralateral lateral wing was lower than that in the other three subregions (p = 0.002; Chi-square test; Figures 3K–3M and 3P), which indicates that fewer serotonergic neurons in the contralateral lateral wing received synaptic input from the POA. The amplitude of the EPSCs and IPSCs within each subregion is summarized in Figures 3Q and 3R. Together, these data suggest that DRN serotonergic neurons are differentially regulated by different inputs. Lateral Hypothalamus Provides Convergent Excitatory and Inhibitory Inputs onto Individual DRN Serotonergic Neurons As demonstrated above, the LH, POA, SN, and amygdala provided excitatory and/or inhibitory inputs to DRN 5-HT neurons (Figure 2). The most plausible scenario for the convergent excitatory and inhibitory inputs is that DRN neurons receive inputs from both glutamatergic and GABAergic neurons of the individual brain areas. However, glutamate and GABA can be coreleased from the same neuron group (Shabel et al., 2014). Thus, it remains possible that the excitatory and inhibitory inputs to DRN 5-HT neurons originate from the same group of neurons of individual upstream brain areas. We took the LH as an example and studied this issue further. We used Vglut2-iresCre and Vgat-Cre mice to manipulate glutamatergic and GABAergic neurons, respectively (Figure 4A). A Cre-dependent AAV encoding ChR2 (AAV-DIO-ChR2-eYFP [enhanced yellow fluorescent protein]) was injected into LH of the Vglut2-ires-Cre mice, and serotonergic neurons were identified through post hoc staining for Tph2 (Figure 4B). To block the polysynaptic response, the recording was performed in the presence of TTX/4-AP. We found that photostimulation of Vglut2+ fibers evoked almost exclusively EPSC in DRN serotonergic neurons (94.4%; 17/18 neurons with response; Figures 4C and 4D). We next repeated this experiment in Vgat-Cre mice (Figures 4E and 4F). Photostimulation of Vgat+ fibers in the DRN induced short-latency IPSCs in the DRN serotonergic neurons (100%; 14/14 neurons with response; Figures 4G and 4H). These results

(D) Distribution of the EPSC amplitude of the input from the PFC (n = 65 neurons). (E) Schematic graph of the experiment to examine the functional input from LH to DRN serotonergic neurons. EPSC (black) and IPSC (blue) recorded from an example neuron in ACSF were blocked following sequential application of NBQX (10 mM; red) and picrotoxin (PTX; 50 mM; gray). (F) Summary of the spatial location of recorded neurons and connectivity of input from LH (n = 106 neurons). (G) Summary data showing the EPSC amplitude from LH before and after applying NBQX (left; LH-DRN:5-HT neurons; n = 6), as well as the IPSC amplitude from LH before and after applying picrotoxin (right; LH-DRN:5-HT neurons; n = 5). (H) Distribution of the EPSC amplitude (left; LH-DRN:5-HT neurons; n = 52) and IPSC amplitude (right; LH-DRN:5-HT neurons; n = 58). (I–L) Experiments to examine POA-DRN:5-HT neurons connection. n = 89 neurons (J). The EPSC and IPSC were recorded from different neurons. The IPSC (blue) was recorded in the presence of NBQX. This is the same layout as (E)–(H). (M–P) Experiments to examine SN-DRN:5-HT neurons connection. n = 51 neurons (N). The EPSC and IPSC were recorded from different serotonergic neurons, and the control EPSC and IPSC were recorded in ACSF. This is the same layout as (E)–(H). (Q–T) Experiments to examine AMY-DRN:5-HT neurons connection. n = 94 neurons (R). EPSC and IPSC recorded from an example neuron in ACSF were blocked following sequential application of NBQX and picrotoxin. This is the same layout as (E)–(H). (U) Summary of connectivity of different inputs to DRN serotonergic neurons. Open bars, the proportion of neurons receiving exclusively EPSC; gray bars, the proportion of neurons receiving exclusively IPSC; black bars, the proportion receiving both EPSC and IPSC. (V and W) Summary showing latency of EPSC (V) and IPSC (W) of different inputs to DRN serotonergic neurons. Wilcoxon signed rank test was used for statistical analysis. Note that some p values are the same due to the same sampling number when comparing the data by Wilcoxon signed rank test. Data are represented as mean ± SEM. See also Figure S2.


A

B

C

E

D

G

F

H

I

J

K

L

M

N

O

P

Q

R

Figure 3. Unilateral and Bilateral Patterns of Axonal Topographic Innervation to Serotonergic Neurons (A) The injection sites and axonal projection pattern within the DRN from different inputs. (Upper row) Viral injection sites are shown; (bottom row) mCherry+ axons in the DRN are shown. dm, dorsal medial region; iwi, ipsilateral lateral wing; lwc, contralateral lateral wing; vm, ventral medial region. Rectangular dotted areas indicate the area for measuring axonal density in the lateral wing. The scale bars in the top row represent 500 mm for PFC and AMY; 100 mm for LHb; and 200 mm for POA, LH, and SN. The scale bar in the bottom row represents 100 mm. (B–G) Quantification of the relative density of axons from PFC (B), LHb (C), POA (D), LH (E), SN (F), and AMY (G) within the DRN. Paired t test is shown. n = 3–6 mice. (H–J) Experiments to examine the PFC input to serotonergic neurons in each subregion of DRN. Schematic of the experiment is shown in (H). Example traces recorded from different areas of DRN are shown in (I). The location of recorded cells is shown in (J). This dataset has been included in Figures 2A–2D. Data from three mice are shown. Blue bar, LED stimulation (475 nm; 1 ms; I).



A

B

E

F

I

C

G

J

D

H

K

Figure 4. LH Provides Convergent Excitation and Inhibition to Individual DRN Serotonergic Neurons (A–D) Experiments to examine the projection from LH glutamatergic neurons to DRN serotonergic neurons. (A) Schematic diagram of the experiment is shown. (B) Post hoc identification of a recorded DRN serotonergic neuron by immunostaining for Tph2 is shown. The scale bar represents 10 mm. (C) Example responses of a DRN serotonergic neuron to photostimulation of LH glutamatergic terminals are shown. Blue bar, LED stimulation (475 nm; 1 ms). (D) Summary of the amplitude of EPSCs and IPSCs is shown (p < 0.001; Wilcoxon signed rank test; n = 18). (E–H) Experiments to examine the projection from LH GABAergic neurons to DRN serotonergic neurons. The scale bar in (F) represents 10 mm. Blue bar, LED stimulation (475 nm; 1 ms; G). (I–K) Analysis of the LH-DRN:serotonergic neurons inputs. (I) Examples show that serotonergic neurons received three types of functional input from the LH. Blue bar, LED stimulation (475 nm; 1 ms). (J) The connectivity for the serotonergic neurons receiving different inputs from LH is shown. (K) The spatial distribution for the serotonergic neurons with different inputs or no input is shown (using dots with same color as in J). This is the same dataset (I–K) as shown in Figures 2E–2H. Data are represented as mean ± SEM. See also Figure S4.

indicate that the EPSC and IPSC observed in the DRN 5-HT neurons mostly originate from glutamatergic and GABAergic neurons of LH, respectively. Consistently, the distribution of axons in the DRN from LH Vglut2+ and Vgat+ neurons recapitulated the previously described unilateral pattern (Figure S4). The glutamatergic and GABAergic projections to the DRN from individual brain areas could innervate serotonergic neurons via different scenarios. One extreme scenario is that glutamatergic and GABAergic terminals make synapses with two distinct populations of serotonergic neurons. Another extreme scenario is that glutamatergic and GABAergic terminals converge onto the same population of serotonergic neurons. To determine the organizing principle of the glutamatergic and GABAergic projections in the DRN, we further analyzed the connection between LH and DRN 5-HT neurons using the same experimental setup as Figure 2E. We found that 12.3% of DRN 5-HT neurons (13/106) recorded received exclusively excitatory inputs from the LH,

16.0% of the DRN 5-HT neurons (17/106) received only inhibitory inputs, and most of the DRN 5-HT neurons (38.7%; 41/106) received convergent excitatory and inhibitory inputs. The rest (33.0%; 35/106) received no synaptic input (Figures 4I and 4J). To exclude the possibility that the IPSCs were disynaptic, we recorded 27 cells in the presence of TTX/4-AP and found that 74.1% of the DRN serotonergic neurons received convergent excitatory and inhibitory inputs, suggesting that few, if any, IPSCs were disynaptic. These data were pooled together with the ones recorded without TTX/4-AP. No obvious bias in the location of these three types of serotonergic neurons was observed (Figure 4K). These results indicate that most of the 5-HT neurons in the DRN are innervated by both glutamatergic and GABAergic neurons from the LH and a single serotonergic neuron could receive convergent excitatory and inhibitory inputs. Thus, the LH could control the activity of DRN 5-HT neurons with a push-pull mechanism.

(K–M) Experiments to examine the POA input to serotonergic neurons in each subregion of DRN. Schematic of the experiment is shown in (K). Examples traces recorded from different areas of DRN are shown in (L). The location of recorded cells is shown in (M). This dataset has been included in Figures 2I–2L. Data from four mice are shown. Blue bar, LED stimulation (475 nm; 1 ms; L). (N and O) Summary of the connectivity (N; p = 0.24; Chi-square test) and EPSC amplitude (O) from PFC to DRN serotonergic neurons in each subregion. (P) Summary of the POA-DRN serotonergic neurons connectivity in each subregion (p = 0.002; Chi-square test). (Q and R) Summary of the amplitude of EPSC (Q) and IPSC (R) from POA to DRN serotonergic neurons in each subregion. Data are represented as mean ± SEM. See also Figure S3.


A

D

H

B

C

F

E

I

G

J

K

Figure 5. DRN GABAergic Neurons Receive Functional Input from LHb (A) Schematic showing configuration of the experiment. (B) Micrograph showing mCherry+ axons (red) from LHb and DRN GABAergic neurons (green) in the DRN. The scale bar represents 200 mm. (C) Enlarged graph of the boxed area in (B). The scale bar represents 5 mm. (D) Response of a GABAergic neuron to current steps. (E) A light-evoked EPSC recorded in a DRN GABAergic neuron in ACSF and following sequential application of TTX, 4-AP, and NBQX. Blue bar, LED stimulation (475 nm; 1 ms). (F) Distribution of the latency of light-evoked EPSC. The inset shows the averaged EPSC latency. (G) Summary data showing the EPSC amplitude before and after sequential bath application of TTX (1 mM) and 4-AP (100 mM; n = 5). (H) An example experiment showing the NBQX-induced blockade of light-evoked EPSC. (I) Summary data showing the EPSC amplitude before and after bath application of NBQX (10 mM; p = 0.001; Wilcoxon signed rank test; n = 11). (J) Summary of the spatial location and connectivity (pie graph; 60/86 GABAergic neurons received synaptic input). (K) Distribution of the EPSC amplitude of the recorded GABAergic neurons. The inset shows the averaged EPSC amplitude. Data are represented as mean ± SEM. See also Figure S5.

Functional Glutamatergic Synapses between LHb and DRN GABAergic Neurons Local GABAergic neurons play important roles in controlling the activity of 5-HT neurons in the DRN (Challis et al., 2013). To further decipher the mechanisms underlying long-range circuits that control 5-HT neurons, we systematically investigated the organization of long-range projection to local GABAergic neurons in the DRN. We took the LHb as an example, and these procedures also applied to other brain areas. We injected AAV-EF1a-DIOeYFP into the DRN of Vgat-Cre mice to label the DRN GABAergic neurons and injected AAV-ChR2-mCherry into the LHb of the same mice (Figure 5A). The mCherry+ fibers were in close contact with the GABAergic neurons labeled with eYFP in the DRN (Figures 5B and 5C). Four weeks after viral injection, we performed whole-cell patch-clamp recording in the

DRN slices. The GABAergic neurons exhibited relatively narrow action potentials in response to depolarizing current injection (Figure 5D). Photostimulation (475 nm; 1 ms) of the ChR2+ fibers evoked short-latency EPSCs in the DRN GABAergic neurons (Figures 5E and 5F), indicating a monosynaptic connection between the LHb and DRN GABAergic neurons. This was further confirmed using TTX-4-AP methods (Figures 5E and 5G). The light-evoked EPSCs were blocked by NBQX, suggesting a glutamatergic nature of this synaptic connection (Figures 5E, 5H, and 5I). Out of 86 DRN GABAergic neurons (from 11 mice), 60 neurons exhibited light-evoked EPSCs (Figure 5J). The amplitude of EPSCs was 173.4 ± 25.7 pA (Figure 5K). These results demonstrate that DRN GABAergic neurons receive direct functional glutamatergic input from the LHb. Light-evoked IPSCs were also observed in the DRN GABAergic neurons (11/86 neurons; Figure S5).


The Major Inputs to the GABAergic Neurons in the DRN GABAergic neurons in the DRN might also integrate functional inputs from multiple upstream brain areas as serotonergic neurons do. Thus, it is critical to determine the property of inputs to GABAergic neurons from various brain areas. We examined the functional projection from the PFC, LH, POA, SN, and amygdala to the DRN GABAergic neurons. We first examined the functional input from the PFC to the DRN GABAergic neurons. We injected AAV-DIO-eYFP into the DRN of Vgat-Cre mice to label GABAergic neurons and infected the PFC with AAV-ChR2-mCherry. By recording from the GABAergic neurons (eYFP+) in the DRN brain slices, we found that optogenetic stimulation of the ChR2+ fibers evoked NBQX-sensitive EPSCs in the GABAergic neurons (60.3%; 35/ 58 neurons; Figures 6A–6D). IPSCs were detected in few DRN GABAergic neurons (2/58 neurons) in response to photostimulation of the PFC fibers, which might be due to feedforward inhibition (Geddes et al., 2016). These light-evoked IPSCs were not further examined due to low probability. The distribution of DRN GABAergic neurons with response or no response is shown (Figure 6B). The averaged latency was 2.0 ± 0.1 ms (Figures 6V, S6A, and S6B), suggesting a monosynaptic connection between PFC and GABAergic neurons in the DRN. Hypothalamic nuclei, including the LH and POA, send dense projection to the DRN and are thought to be a major source of inputs to GABAergic neurons in the DRN (Weissbourd et al., 2014). In Vgat-Cre mice injected with AAV-DIO-eYFP into the DRN, we infected LH or POA with AAV-ChR2-mCherry in two separate groups of animals. We found that activation of ChR2+ fiber terminals from LH and POA, respectively, evoked EPSCs and/or IPSCs in the DRN GABAergic neurons (Figures 6E, 6H, 6I, and 6L) with short latencies (Figures 6V, 6W, and S6C–S6F). The EPSCs and IPSCs were blocked by NBQX and picrotoxin, respectively (Figures 6E, 6G, 6I, and 6K). Overall connectivity (including both EPSC and IPSC) was comparable between LH and POA (70.3%, 26/37 neurons versus 60.0%, 27/45 neurons; Figures 6F, 6J, and 6U). Thus, hypothalamus has a strong functional connection with the DRN GABAergic neurons. SN represents another major input to DRN GABAergic neurons. Our results showed that optogenetic activation of SN fibers in the DRN induced short-latency EPSCs and/or IPSCs in the DRN GABAergic neurons (Figures 6M–6P, 6U–6W, S6G, and S6H), suggesting that the SN provides both glutamatergic and GABAergic input to the DRN GABAergic neurons (76.9%; 20/ 26 neurons). We next determined the functional synaptic connection between amygdala and DRN GABAergic neurons. We found that photostimulation of the amygdala fibers in the DRN induced EPSCs and/or IPSCs mediated by AMPA and GABAA receptors, respectively (Figures 6Q–6T), with short latencies (Figures 6V, 6W, S6I, and S6J). However, the percentage of neurons that showed a synaptic response was relatively low (27.6%; 16/58 neurons; Figure 6U), in comparison with other inputs. These data demonstrate that local GABAergic neurons in the DRN could integrate the information from multiple upstream brain areas and play a key role in dynamically modulating the activity of DRN 5-HT neurons.


We next compared the synaptic inputs to DRN serotonergic neurons and to DRN GABAergic neurons. The EPSC connectivity was comparable between DRN serotonergic neurons and GABAergic neurons for each afferent (Figure S6K) and so was the IPSC connectivity (Figure S6L). When comparing the amplitude of synaptic input from individual upstream brain areas, we found that LHb, LH, POA, SN, and amygdala provided comparable EPSC to both serotonergic neurons and GABAergic neurons (Figure S6M). LH, SN, and amygdala provided similar IPSC to both cell types (Mann-Whitney test at a significant level of p = 0.05; Figure S6N). Despite these similarities, PFC provided stronger excitatory input to serotonergic neurons and POA provided stronger inhibitory input to GABAergic neurons (Figures S6M and S6N). Overall, these data suggest that DRN serotonergic neurons receive largely comparable synaptic inputs with those received by GABAergic neurons from six major upstream brain areas. Feedforward Inhibition in Controlling 5-HT Neurons Unlike many subcortical areas, which send both glutamatergic and GABAergic projections to DRN 5-HT neurons, the LHb contains largely glutamatergic neurons and sends glutamatergic projection to DRN 5-HT neurons. We asked whether the LHb could bidirectionally modulate the activity of DRN 5-HT neurons by recruiting local GABAergic neurons in the DRN to generate feedforward inhibition. In Pet1-tdTomato mice injected with AAV-ChR2-mCherry in the LHb, we recorded from the DRN 5-HT neurons at 0 mV and found photostimulation of LHb ChR2+ fibers evoked IPSCs in a fraction of DRN 5-HT neurons (Figure 7A). The IPSC could be blocked by picrotoxin, indicating the IPSC is mediated by GABAA receptor (Figures S7A and S7B). The latency of the IPSCs was 5.7 ± 0.7 ms, significantly longer than that of the EPSCs recorded at 70 mV in the same group of neurons (2.3 ± 0.3 ms; p < 0.001; Wilcoxon signed rank test; Figures 7B and 7C). The onset latency jitter of IPSC was larger than that of EPSC in the same group of neurons (2.6 ± 0.8 ms versus 0.3 ± 0.1 ms; p = 0.008; Wilcoxon signed rank test; n = 9; Figures S7C–S7E), suggesting the disynaptic nature of the feedforward inhibition. The IPSCs induced by photostimulation of the LHb fibers were blocked by application of the AMPA receptor antagonist NBQX (Figures 7D and 7E), further confirming that light-evoked IPSC represents disynaptic GABAergic inhibition. Because the LHb made glutamatergic synapse with GABAergic neurons in the DRN, we speculated that activation of glutamatergic fibers from LHb in the DRN would evoke spiking activity in the DRN GABAergic neurons. In Vgat-Cre mice injected with AAV-DIO-eYFP in the DRN, we injected AAVChR2-mCherry into LHb. We recorded from DRN GABAergic neurons labeled with eYFP in cell attached mode and found that 28.2% of GABAergic cells (20/71 cells) showed spiking activity in response to optogenetic activation of ChR2+ fibers from the LHb (Figure 7F). We further examined the synaptic connectivity between local GABAergic neurons and 5-HT neurons in the DRN. We injected AAV-DIO-ChR2-eYFP into the DRN of Vgat-Cre mice to express ChR2 in the DRN GABAergic neurons (Figure 7G). Recording from the 5-HT neurons, identified by post hoc staining for 5-HT or Tph2, in the DRN, we found that

A

B

C

E

F

G

H

K

L

J

I

D

M

N

O

P

Q

R

S

T

U

W

V

Figure 6. DRN GABAergic Neurons Receive Functional Inputs from Multiple Brain Areas (A) (Top) Schematic diagram of the experiment. (Bottom) A representative experiment shows light-evoked EPSC before (black; in ACSF) and after NBQX (10 mM; red). Blue bar, LED stimulation (475 nm; 1 ms); same for (E), (I), (M), and (Q). (B) Summary of the spatial location and connectivity of the DRN GABAergic neurons (n = 58). (C) Summary data showing the amplitude of EPSC evoked by photostimulaton of PFC fibers before and after NBQX (p = 0.063; n = 5). (D) Distribution of the EPSC amplitude of the input from PFC (n = 35).



A

B

C

Figure 7. Feedforward Inhibition of DRN Serotonergic Neurons by LHb-DRN Inputs

(A) Schematic configuration of the experiment. (B) An example experiment showing that photostimulation of the LHb ChR2+ fibers evoked a shortlatency EPSC (red) and a long-latency IPSC (black) in a DRN 5-HT neuron. Blue bar, LED stimulation (475 nm; 1 ms). (C) Summary of the latencies of light-evoked F D E EPSCs and IPSCs (p < 0.001; n = 13). (D) An experiment showing the blockade of the light-evoked IPSC by NBQX (10 mM). The inset shows the IPSC before (black) and after NBQX (red). Blue bar, LED stimulation (475 nm; 1 ms). (E) Summary of the IPSC amplitude before and after NBQX (p = 0.063; n = 5 neurons). (F) Activation of LHb fibers evoked reliable spikes in an example DRN GABAergic neuron. Example G H I trace (top; scale bars: 400 pA and 100 ms), raster plot of the first ten trials (middle), and summary of all 33 trials (bottom) are shown. Blue bar, LED stimulation (475 nm; 1 ms). (G) Schematic of the experiment to examine the local GABAergic inputs to DRN serotonergic neurons. (H) An example experiment showing the blockade of light-evoked IPSC in a serotonergic neuron by picrotoxin (50 mM). The inset shows the IPSC before (black) and after picrotoxin (red). Blue bar, LED stimulation (475 nm; 1 ms). (I) Summary of the IPSC amplitude evoked by local GABAergic inputs before and after picrotoxin (p = 0.016; n = 7). Data are represented as mean ± SEM. Wilcoxon signed rank test was used for statistical analysis. See also Figure S7.

photostimulation of GABAergic neurons (ChR2+) and their axons induced IPSCs in the DRN 5-HT neurons and the IPSCs were blocked by bath application of picrotoxin (Figures 7H and 7I). The latency of IPSCs was 1.5 ± 0.1 ms, indicative of a monosynaptic connection between GABAergic neurons and 5-HT neurons in the DRN. Together, these results suggest that LHb glutamatergic fibers could bidirectionally modulate the activity of 5-HT neurons by recruiting local GABAergic neurons. DISCUSSION By combining optogenetics and slice electrophysiology, we systematically examined the organization of long-range synaptic input to both DRN serotonergic neurons and GABAergic neurons from six major upstream brain areas, including the

LHb, PFC, LH, POA, SN, and amygdala. We found that these brain areas send functional projections to both 5-HT and GABAergic neurons in the DRN. Although these brain areas do not show much bias in making synapses with either 5-HT or GABAergic neurons in the DRN, they innervate the DRN with different topographic patterns. The PFC and LHb send mostly glutamatergic projections to the DRN with a bilateral pattern. In contrast, the LH, POA, SN, and amygdala send both glutamatergic and GABAergic projections to the DRN, with less connection to the contralateral wing of DRN. We also demonstrated that the upstream brain areas control the activity of 5-HT neurons via different circuit mechanisms: push-pull or feedforward inhibition. Our results provide the basis for further understanding the role of different inputs to the DRN in the context of behavior.

(E) Schematic graph of the experiment. EPSCs (black; in ACSF) were blocked following application of NBQX (red). In another neuron, IPSCs (blue; in ACSF) were blocked following application of picrotoxin (gray). (F) Summary of the spatial location and connectivity of DRN GABAergic neurons receiving input from LH (n = 37 neurons). (G) Summary data of the EPSC amplitude from LH before and after NBQX (left; LH-DRN:GABAergic neurons; p = 0.031; n = 6) and IPSC amplitude from LH before and after applying picrotoxin (right; LH-DRN:GABAergic neurons; p = 0.031; n = 6). (H) Distribution of the amplitude of EPSCs (left; LH-DRN:GABAergic neurons; n = 14) and IPSCs (right; LH-DRN:GABAergic neurons; n = 22). (I–L) Experiments to examine POA-DRN:GABAergic neurons connection. n = 45 neurons (J). EPSC and IPSC recorded from an example neuron in ACSF were blocked following sequential application of NBQX and picrotoxin. This is the same layout as (E)–(H). (M–P) Experiments to examine SN-DRN:GABAergic neurons connection. n = 26 neurons (N). IPSC and EPSC recorded from an example neuron in ACSF were blocked following sequential application of picrotoxin and NBQX. This is the same layout as (E)–(H). (Q–T) Experiments to examine AMY-DRN:GABAergic neurons connection. n = 58 neurons (R). The EPSC and IPSC were recorded from different GABAergic neurons, and the control EPSC and IPSC were recorded in ACSF. This is the same layout as (E)–(H). (U) Summary of connectivity of different inputs to DRN GABAergic neurons. Open bars, the proportion of neurons receiving exclusively EPSC; gray bars, the proportion of neurons receiving exclusively IPSC; black bars, the proportion receiving both EPSC and IPSC. (V and W) Summary showing latency of EPSCs (V) and IPSCs (W) of different inputs to DRN GABAergic neurons. Data are represented as mean ± SEM. Wilcoxon signed rank test was used for statistical analysis. See also Figure S6.


Technical Considerations For the purpose of anterograde tracing, AAV has been shown to be efficient and specific (Wang et al., 2014) and has been used for large-scale anterograde tracing to construct the rodent brain connectome. Its advantage over conventional tracers is the flexibility to introduce various target genes (e.g., ChR2) into specific brain areas or cells. But caveats of using AAV must be taken into consideration. Similar to many anterograde tracers, retrograde infection could occur for AAV, although the degree of AAV retrograde labeling depends on the viral types, brain regions, and cell types (Burger et al., 2004). In experiments examining the axonal projection pattern in the DRN, we found that retrograde labeling is negligible under our experimental conditions. Second, the genetic labeling of 5-HT, glutamatergic, or GABAergic neurons could be non-specific. We confirmed the Pet1-tdTomato mice with more than 90% specificity (Scott et al., 2005). The Vgat-Cre and Vglut2-ires-Cre mice also showed high specificity according to original studies (Chao et al., 2010; Vong et al., 2011). Thus, a small portion of tdTomato+ or eYFP+ neurons in the DRN could be non-serotonergic or non-GABAergic neurons. Finally, in most cases, we identify the monosynaptic input with the standard of short latency and apparently small onset latency variability (jitter). A small fraction of the recorded EPSCs and IPSCs displayed long latency, and some of them displayed large jitter, which may be due to feedforward excitation or feedforward inhibition through DRN local microcircuitry. This is also evidenced by the existence of multi-peak EPSCs and broader time courses of EPSC (Figures 2E, 3K, 5E, and 6Q). Thus, a small amount of polysynaptic input might have been included in the data of long-range monosynaptic input. So the percentage of the monosynaptic input would have been overestimated. However, it is not easy to distinguish feedforward excitation under our experimental condition. Diverse Functional Synapses onto DRN 5-HT and GABAergic Neurons All of the brain areas tested make functional synaptic connection with neurons in the DRN. This is in support of the idea that the DRN functionally integrates the information from multiple forebrain areas (Weissbourd et al., 2014). In addition, DRN receives abundant peptidergic input from the hypothalamus and amygdala (Pollak Dorocic et al., 2014; Weissbourd et al., 2014) and the neuropeptides could play important roles in regulating the neural network via presynaptic or postsynaptic mechanisms (Bargmann and Marder, 2013). Although peptidergic input to the DRN might be important to fully understand the regulation of DRN 5-HT neurons, their effects in the DRN were not examined in our study. Local GABAergic inputs play a key role in the modulation of DRN 5-HT neurons (Challis et al., 2013; Li et al., 2016). This is evidenced from the strong input from local GABAergic neurons to the DRN 5-HT neurons. Actually, it is surprising that the amplitudes of the IPSC from local GABAergic neurons were about 5-fold greater than those produced by different long-range afferents. One reason could be that different number of GABAergic neurons were labeled with ChR2. The local GABAergic neurons are densely distributed in the DRN, which might be easily in-

fected after infusion of AAV. In contrast, for most of the upstream brain areas, the DRN-projecting neurons are dispersed in a relatively larger area (Weissbourd et al., 2014), which could lead to only partial labeling of the GABAergic neurons that project to the DRN. Nevertheless, the DRN GABAergic neurons represent a key circuit element, which could be recruited by the long-range inputs to modulate the activity of DRN serotonergic neurons. Differential Topographic Patterns of Projection and Connection Serotonergic neurons in the DRN are topographically organized, with distinct subpopulations of serotonergic neurons showing different properties (Calizo et al., 2011; Fernandez et al., 2016). Adding to this, our study demonstrated that subareas of the DRN might be differentially modulated by various inputs. Specifically, we found the synaptic connectivity showed bilateral pattern for PFC but ipsilateral for the POA. These results are consistent with the notion that different upstream brain areas play differential roles in the regulation of subareas of the DRN. Further, the DRN might contain functionally distinct subgroups of serotonergic neurons, which is supported by previous pharmacological and morphological experiments (Abrams et al., 2004; Hale et al., 2008). The same brain areas could synapse on both GABAergic and 5-HT neurons in the DRN. However, it is not known whether it is the same population of neurons in a particular upstream brain structure that synapses on both 5-HT and GABAergic neurons. Controlling the Activity of DRN 5-HT Neurons via a PushPull Mechanism Consistent with a previous viral tracing study (Weissbourd et al., 2014), we found that DRN 5-HT neurons receive both excitatory and inhibitory inputs from the same brain areas. It remains possible that the glutamatergic and GABAergic neurons from the same brain areas target at different subsets of 5-HT neurons in the DRN. We took the LH-DRN projection as an example and analyzed the organizing principle of the LH projection to the DRN 5-HT neurons in detail. We showed that most of the DRN serotonergic neurons received both glutamatergic and GABAergic input from the LH, suggesting LH could bidirectionally modulate the activity of DRN 5-HT neurons with a push-pull mechanism. The glutamatergic and GABAergic neurons in the LH are differentially involved in animal behavior, such as feeding (Jennings et al., 2013). Thus, the differential activation of the glutamatergic and GABAergic neurons in the LH will lead to alteration of the activity of DRN 5-HT neurons, which could possibly in turn alter the emotional homeostasis. The other brain areas, including the POA, SN, and amygdala, were not analyzed in detail, but our data suggest that the push-pull mechanism might be a general mode of action for most of the brain areas that send both glutamatergic and GABAergic projection to DRN 5-HT neurons, allowing for a direct control of 5-HT neurons. Modulation of DRN 5-HT Neurons by Feedforward Inhibition Feedforward inhibition is a general mechanism for balancing excitation with inhibition. We determined that the LHb makes


synapses with the local GABAergic neurons in the DRN, which in turn generate feedforward inhibition. Thus, the feedforward inhibition might be important for regulating the homeostasis of LHb-raphe pathway, which is implicated in emotional regulation (Amo et al., 2014). A similar mechanism was demonstrated for the PFC in a recent study (Geddes et al., 2016). Together, these results suggest that the LHb and PFC can bidirectionally control the activity of DRN serotonergic neurons by recruiting local GABAergic neurons. Local GABAergic neurons in the DRN might integrate the inputs from the PFC and LHb, as these inputs might converge onto the same putative GABAergic neurons (Varga et al., 2003). The feedforward inhibition might not be limited to these two inputs, and other inputs could recruit this mechanism as well. Together, DRN serotonergic neurons could integrate different physiologically relevant information encoded by various upstream brain areas via the functional synaptic inputs. Dynamic modulation of the serotonergic neurons could be achieved by convergent excitation and inhibition or feedforward inhibition provided by local GABAergic neurons (Isaacson and Scanziani, 2011). Thus, our study paves the way for further examining the functional role of long-range circuits controlling the activity of 5-HT neurons in the DRN. EXPERIMENTAL PROCEDURES Animals Both adult male and female C57BL/6 (SLAC or Ling Chang), ePet1-Cre (JAX012712), Vgat-Cre (Chao et al., 2010), Vgat-ires-Cre (JAX016962), Vglut2-ires-Cre (JAX016963), and Rosa-tdTomato (Ai9; JAX007905) mice were used. ePet1-Cre mice were crossed with Ai9 to get the double transgenic mice, referred to as Pet1-tdTomato. The results obtained from Vgat-Cre and Vgat-ires-Cre mice were comparable and were therefore combined. The Vgat-Cre and Vgat-ires-Cre mice were hereafter referred to as Vgat-Cre mice. All mice were raised on a 12 hr light/dark cycle (lights on at 7:00 a.m.) with ad libitum food and water. All procedures were approved by the Animal Care and Use Committee of the Institute of Neuroscience, Chinese Academy of Sciences, Shanghai, China. Stereotaxic Surgery Mice were anesthetized with pentobarbital sodium (100 mg/kg body weight), mounted in a stereotaxic apparatus. Injections of AAV viruses were carried out using borosilicate glass pipettes connected to the Picospritzer III (Parker) at a rate of 10–100 nL/min. After the viral injection, the pipettes were left in position for another 10–15 min before withdrawal. Detailed experimental procedures are in Supplemental Experimental Procedures. Histology and Imaging The immunohistochemistry was performed as described previously (Sun et al., 2011). Primary antibodies were anti-5-HT (1:500; Immunostar; 20079), antiDsRed (1:500; Clontech; 632496), and anti-Tph2 (1:500; Millipore; ABN 60). Detailed experimental procedures are in Supplemental Experimental Procedures. Image Analysis To quantify the axon density, the fluorescence of the axons in the DRN was quantified using Fiji software. Images were digitized as 8-bit TIF files. Background was subtracted from the images by using Fiji (value = 50), and the ideal thresholds were calculated as previously described (Beyeler et al., 2016). In Fiji, after thresholding, the area fraction of the region of interest (ROI) was measured for each subregion. Values of the area fraction of each subdivision were normalized within each section. Only sections containing all four subregions were analyzed, including ipsilateral lateral wing (lwi), dorsal medial part (dm), ventral medial part (vm), and


contralateral lateral wing (lwc). Values from one to four sections per mice were averaged and summed together (three mice for PFC, POA, and LH; five mice for LHb and SN; six mice for AMY). Electrophysiology Acute slice electrophysiology was performed as described previously (Sun et al., 2011). Slices were placed on glass coverslips coated with poly-L-lysine (Sigma) and submerged in a recording chamber (Warner Instruments). All experiments were performed at near-physiological temperatures (30 C–34 C) using an in-line heater (Warner Instruments) while perfusing the recording chamber with artificial cerebrospinal fluid (ACSF) at 1–4 mL/min. Whole-cell patch-clamp recordings were made from the target neurons under infrared (IR)-differential interference contrast (DIC) visualization, a charge-coupled device (CCD) camera (IR-1000E; DAGE-MTI) using a fluorescent Olympus BX51WI microscope (Olympus Optical), and a micromanipulator (MPC200; Sutter Instrument). To record the extracellular spiking activity, cell-attached recordings were performed at voltage clamp mode following establishing Giga-seal right before the whole-cell recordings were made. Recording pipettes (2–5 MU; Borosilicate Glass; Sutter Instrument) were prepared by a micropipette puller (model P97; Sutter Instrument) and backfilled with internal solution containing (in mM) CsMeSO3 130, MgCl2 1, CaCl2 1, HEPES 10, QX314 2, EGTA 11, Mg-ATP 2, and Na-GTP 0.3 (pH 7.3; 295 mOsm). The K+-base internal solution was used to monitor the voltage change of the recording neurons at current clamp mode, which contained (in mM) K-gluconate 130, MgCl2 1, CaCl2 1, KCl 1, HEPES 10, EGTA 11, Mg-ATP 2, and NaGTP 0.3 (pH 7.3; 295 mOsm). In most experiments, biocytin (0.2%) was included into the internal solution. Blue light-emitting diode (LED) (475 nm; 11 mW/mm2; UHP-Mic-LED-475; Prizmatix) coupled to a water objective (603; numerical aperture [NA] 1.0) was used to activate ChR2+ cells or fibers. The location of the recorded cells was photographed by an air objective (43; NA 0.13). EPSCs were recorded at a holding voltage of 70 mV and IPSC at 0 mV. Extremely small, low signal-to-noise ratio, and unreliable responses were regarded as no response. NBQX, picrotoxin, and (R)-CPP were purchased from Tocris Bioscience. TTX (0.5–1 mM) was purchased from Hebei Aquatic Product. All other chemicals were obtained from Sigma. A few experiments performed in the presence of TTX/4-AP showed similar results to that in ACSF and thus were pooled together. Data Acquisition and Analysis Voltage clamp and current clamp recordings were carried out using a computer-controlled amplifier (MultiClamp 700B; Molecular Devices). During recordings, traces were low-pass filtered at 4 kHz and digitized at 20 kHz (DigiData 1440; Molecular Devices). Data were acquired by Axon Clampex 10.3. The amplitude and latency of postsynaptic currents (EPSC or IPSC) was measured from the averaged traces manually. EPSC and IPSC amplitudes were calculated as the difference between the peak amplitude in a pre-defined window after the light stimulation onset and the mean amplitude just preceding the EPSC or IPSC. EPSC or IPSC latency was measured as the time from the onset of light stimulation to the first intersection between the baseline and the EPSC or IPSC, which can be easily identified at the place of maximal rising/falling curvature. The EPSCs and IPSCs in our condition sometimes showed multi-peaks, and we measured the latency of the first EPSC or IPSC. The jitter was defined as the SD of the EPSC or IPSC onset latency across individual sweeps per cell (as least five sweeps per cell). Example traces were averaged and filtered using a low-pass-Gaussian algorithm ( 3 dB cutoff frequency = 1,000 Hz) in Clampfit (Molecular Devices). Data were analyzed with Chi-square test, Wilcoxon matched-pairs signed rank test, Mann-Whitney test, or paired t test. All data are expressed as mean ± SEM. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures and seven figures and can be found with this article online at http://dx.doi. org/10.1016/j.celrep.2017.02.077.

AUTHOR CONTRIBUTIONS L.Z. and Y.-G.S. conceived and designed the project. L.Z., M.-Z.L., J.D., and D.M. performed patch-clamp recordings. L.Z. and M.-Z.L. analyzed the data. L.Z., M.-Z.L., and Q.L. performed the immunohistochemical experiments. L.Z., M.-Z.L., and Y.-G.S. wrote the manuscript. ACKNOWLEDGMENTS We thank Judith Homberg for comments on the manuscript. We thank YanJing Zhu for technical support and Dr. Qian Hu from the Optical Imaging Core Facility at ION for the help with image analysis. This work was supported by the National Natural Science Foundation of China (nos. 31371122 and 81322015), Chinese Academy of Sciences Hundreds of Talents Program (to Y.G.S.), Youth Thousand Plan (to Y.G.S.), and the Strategic Priority Research Program of the Chinese Academy of Sciences (grant no. XDB02010000). Received: December 5, 2016 Revised: January 26, 2017 Accepted: February 24, 2017 Published: March 21, 2017

Fernandez, S.P., Cauli, B., Cabezas, C., Muzerelle, A., Poncer, J.C., and Gaspar, P. (2016). Multiscale single-cell analysis reveals unique phenotypes of raphe 5-HT neurons projecting to the forebrain. Brain Struct. Funct. 221, 4007–4025. Geddes, S.D., Assadzada, S., Lemelin, D., Sokolovski, A., Bergeron, R., HajDahmane, S., and Be´ı¨que, J.C. (2016). Target-specific modulation of the descending prefrontal cortex inputs to the dorsal raphe nucleus by cannabinoids. Proc. Natl. Acad. Sci. USA 113, 5429–5434. Gocho, Y., Sakai, A., Yanagawa, Y., Suzuki, H., and Saitow, F. (2013). Electrophysiological and pharmacological properties of GABAergic cells in the dorsal raphe nucleus. J. Physiol. Sci. 63, 147–154. Hale, M.W., and Lowry, C.A. (2011). Functional topography of midbrain and pontine serotonergic systems: implications for synaptic regulation of serotonergic circuits. Psychopharmacology (Berl.) 213, 243–264. Hale, M.W., Hay-Schmidt, A., Mikkelsen, J.D., Poulsen, B., Bouwknecht, J.A., Evans, A.K., Stamper, C.E., Shekhar, A., and Lowry, C.A. (2008). Exposure to an open-field arena increases c-Fos expression in a subpopulation of neurons in the dorsal raphe nucleus, including neurons projecting to the basolateral amygdaloid complex. Neuroscience 157, 733–748. Homberg, J.R. (2012). Serotonin and decision making processes. Neurosci. Biobehav. Rev. 36, 218–236.

REFERENCES

Isaacson, J.S., and Scanziani, M. (2011). How inhibition shapes cortical activity. Neuron 72, 231–243.

Abrams, J.K., Johnson, P.L., Hollis, J.H., and Lowry, C.A. (2004). Anatomic and functional topography of the dorsal raphe nucleus. Ann. N Y Acad. Sci. 1018, 46–57.

Jacobs, B.L., and Azmitia, E.C. (1992). Structure and function of the brain serotonin system. Physiol. Rev. 72, 165–229.

Amo, R., Fredes, F., Kinoshita, M., Aoki, R., Aizawa, H., Agetsuma, M., Aoki, T., Shiraki, T., Kakinuma, H., Matsuda, M., et al. (2014). The habenulo-raphe serotonergic circuit encodes an aversive expectation value essential for adaptive active avoidance of danger. Neuron 84, 1034–1048. Bargmann, C.I., and Marder, E. (2013). From the connectome to brain function. Nat. Methods 10, 483–490. Beyeler, A., Namburi, P., Glober, G.F., Simonnet, C., Calhoon, G.G., Conyers, G.F., Luck, R., Wildes, C.P., and Tye, K.M. (2016). Divergent routing of positive and negative information from the amygdala during memory retrieval. Neuron 90, 348–361.

Jennings, J.H., Rizzi, G., Stamatakis, A.M., Ung, R.L., and Stuber, G.D. (2013). The inhibitory circuit architecture of the lateral hypothalamus orchestrates feeding. Science 341, 1517–1521. Lesch, K.P., and Waider, J. (2012). Serotonin in the modulation of neural plasticity and networks: implications for neurodevelopmental disorders. Neuron 76, 175–191. Li, Y., Zhong, W., Wang, D., Feng, Q., Liu, Z., Zhou, J., Jia, C., Hu, F., Zeng, J., Guo, Q., et al. (2016). Serotonin neurons in the dorsal raphe nucleus encode reward signals. Nat. Commun. 7, 10503. Lotharius, J., and Brundin, P. (2002). Pathogenesis of Parkinson’s disease: dopamine, vesicles and alpha-synuclein. Nat. Rev. Neurosci. 3, 932–942.

Burger, C., Gorbatyuk, O.S., Velardo, M.J., Peden, C.S., Williams, P., Zolotukhin, S., Reier, P.J., Mandel, R.J., and Muzyczka, N. (2004). Recombinant AAV viral vectors pseudotyped with viral capsids from serotypes 1, 2, and 5 display differential efficiency and cell tropism after delivery to different regions of the central nervous system. Mol. Ther. 10, 302–317.

Nakamura, K. (2013). The role of the dorsal raphe´ nucleus in reward-seeking behavior. Front. Integr. Neurosci. 7, 60.

Calizo, L.H., Akanwa, A., Ma, X., Pan, Y.Z., Lemos, J.C., Craige, C., Heemstra, L.A., and Beck, S.G. (2011). Raphe serotonin neurons are not homogenous: electrophysiological, morphological and neurochemical evidence. Neuropharmacology 61, 524–543.

Okaty, B.W., Freret, M.E., Rood, B.D., Brust, R.D., Hennessy, M.L., deBairos, D., Kim, J.C., Cook, M.N., and Dymecki, S.M. (2015). Multi-scale molecular deconstruction of the serotonin neuron system. Neuron 88, 774–791.

Caspi, A., Sugden, K., Moffitt, T.E., Taylor, A., Craig, I.W., Harrington, H., McClay, J., Mill, J., Martin, J., Braithwaite, A., and Poulton, R. (2003). Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene. Science 301, 386–389. Challis, C., Boulden, J., Veerakumar, A., Espallergues, J., Vassoler, F.M., Pierce, R.C., Beck, S.G., and Berton, O. (2013). Raphe GABAergic neurons mediate the acquisition of avoidance after social defeat. Journal Neurosci. 33, 13978–13988, 13988a. Chao, H.T., Chen, H., Samaco, R.C., Xue, M., Chahrour, M., Yoo, J., Neul, J.L., Gong, S., Lu, H.C., Heintz, N., et al. (2010). Dysfunction in GABA signalling mediates autism-like stereotypies and Rett syndrome phenotypes. Nature 468, 263–269.

Ogawa, S.K., Cohen, J.Y., Hwang, D., Uchida, N., and Watabe-Uchida, M. (2014). Organization of monosynaptic inputs to the serotonin and dopamine neuromodulatory systems. Cell Rep. 8, 1105–1118.

Petreanu, L., Huber, D., Sobczyk, A., and Svoboda, K. (2007). Channelrhodopsin-2-assisted circuit mapping of long-range callosal projections. Nat. Neurosci. 10, 663–668. Petreanu, L., Mao, T., Sternson, S.M., and Svoboda, K. (2009). The subcellular organization of neocortical excitatory connections. Nature 457, 1142–1145. €rth, D., Xuan, Y., Johansson, Y., Pozzi, L., Silberberg, G., Pollak Dorocic, I., Fu Carle´n, M., and Meletis, K. (2014). A whole-brain atlas of inputs to serotonergic neurons of the dorsal and median raphe nuclei. Neuron 83, 663–678. Sachs, B.D., Ni, J.R., and Caron, M.G. (2015). Brain 5-HT deficiency increases stress vulnerability and impairs antidepressant responses following psychosocial stress. Proc. Natl. Acad. Sci. USA 112, 2557–2562.

Cools, R., Roberts, A.C., and Robbins, T.W. (2008). Serotoninergic regulation of emotional and behavioural control processes. Trends Cogn. Sci. 12, 31–40.

Scott, M.M., Wylie, C.J., Lerch, J.K., Murphy, R., Lobur, K., Herlitze, S., Jiang, W., Conlon, R.A., Strowbridge, B.W., and Deneris, E.S. (2005). A genetic approach to access serotonin neurons for in vivo and in vitro studies. Proc. Natl. Acad. Sci. USA 102, 16472–16477.

Descarries, L., Watkins, K.C., Garcia, S., and Beaudet, A. (1982). The serotonin neurons in nucleus raphe dorsalis of adult rat: a light and electron microscope radioautographic study. J. Comp. Neurol. 207, 239–254.

Shabel, S.J., Proulx, C.D., Piriz, J., and Malinow, R. (2014). Mood regulation. GABA/glutamate co-release controls habenula output and is modified by antidepressant treatment. Science 345, 1494–1498.


Stuber, G.D., and Wise, R.A. (2016). Lateral hypothalamic circuits for feeding and reward. Nat. Neurosci. 19, 198–205. Sun, Y.G., Wu, C.S., Lu, H.C., and Beierlein, M. (2011). Target-dependent control of synaptic inhibition by endocannabinoids in the thalamus. J. Neurosci. 31, 9222–9230. Varga, V., Kocsis, B., and Sharp, T. (2003). Electrophysiological evidence for convergence of inputs from the medial prefrontal cortex and lateral habenula on single neurons in the dorsal raphe nucleus. Eur. J. Neurosci. 17, 280–286. Vertes, R.P. (1991). A PHA-L analysis of ascending projections of the dorsal raphe nucleus in the rat. J. Comp. Neurol. 313, 643–668. Vong, L., Ye, C., Yang, Z., Choi, B., Chua, S., Jr., and Lowell, B.B. (2011). Leptin action on GABAergic neurons prevents obesity and reduces inhibitory tone to POMC neurons. Neuron 71, 142–154. Walker, S.C., Mikheenko, Y.P., Argyle, L.D., Robbins, T.W., and Roberts, A.C. (2006). Selective prefrontal serotonin depletion impairs acquisition of a detourreaching task. Eur. J. Neurosci. 23, 3119–3123.


Wang, Q., Henry, A.M., Harris, J.A., Oh, S.W., Joines, K.M., Nyhus, J., Hirokawa, K.E., Dee, N., Mortrud, M., Parry, S., et al. (2014). Systematic comparison of adeno-associated virus and biotinylated dextran amine reveals equivalent sensitivity between tracers and novel projection targets in the mouse brain. J. Comp. Neurol. 522, 1989–2012. Warden, M.R., Selimbeyoglu, A., Mirzabekov, J.J., Lo, M., Thompson, K.R., Kim, S.Y., Adhikari, A., Tye, K.M., Frank, L.M., and Deisseroth, K. (2012). A prefrontal cortex-brainstem neuronal projection that controls response to behavioural challenge. Nature 492, 428–432. Weissbourd, B., Ren, J., DeLoach, K.E., Guenthner, C.J., Miyamichi, K., and Luo, L. (2014). Presynaptic partners of dorsal raphe serotonergic and GABAergic neurons. Neuron 83, 645–662. €le, C., Eser, D., Rupprecht, R., Mo¨ller, Zill, P., Baghai, T.C., Zwanzger, P., Schu H.J., Bondy, B., and Ackenheil, M. (2004). SNP and haplotype analysis of a novel tryptophan hydroxylase isoform (TPH2) gene provide evidence for association with major depression. Mol. Psychiatry 9, 1030–1036.