Medial Habenula Output Circuit Mediated by α5 Nicotinic Receptor-Expressing GABAergic Neurons in the Interpeduncular Nucleus

Identity of nAChR α5 subunit-expressing neurons in the habenulo–peduncular–tegmental system

Identification of the specific neurons in which the α5 nAChR subunit is expressed is essential for two principal reasons. First, because it is an accessory subunit, the effect of α5 on nAChR function depends entirely on the specific channel-forming/ligand-binding subunits with which it is assembled (Lindstrom et al., 1996; Ramirez-Latorre et al., 1996; Nelson and Lindstrom, 1999), and the available partners are cell specific. Second, cellular resolution is necessary to identify α5-expressing neurons in brain nuclei containing diverse cell types, such as the IP and MnR, to understand how these neurons may assemble into functional circuits. Three principal methods have been used to describe the expression of nAChRs in the brain: in situ hybridization, receptor autoradiography, and receptor binding in brain homogenates, which have not provided cellular resolution. There are no published reports of α5 subunit localization by immunohistochemistry, presumably because the available antibodies are not suitable for this purpose. Here, we have overcome these limitations by using an α5^GFP BAC transgene (GENSAT), examined by immunofluorescence for the expressed GFP and by cellular-resolution FISH to examine the neurotransmitter phenotype and circuitry of α5-expressing neurons in the brainstem.

To understand the expression of α5 at the cellular level, we first examined the expression of a GFP reporter in α5^GFP mice and compared it with the pattern of α5 mRNA in the ABA. In the telencephalon, α5^GFP-labeled cell bodies were observed in the neocortex (layers 5 and 6), endopiriform nucleus, subiculum, and CA1 region of the hippocampus (Fig. 1A–F and data not shown), entirely in agreement with ABA data. In contrast, no cell body or fiber staining was noted in the medial or lateral habenula (LHb; Fig. 1G,I), also in agreement with ABA data. However, the α5^GFP transgene did not replicate the known expression of α5 mRNA in the VTA and substantia nigra (Fig. 1J–L; Wada et al., 1990). This specific lack of expression of α5^GFP in these midbrain dopaminergic nuclei could occur because the required Chrna5 gene regulatory sequences lie outside the BAC used to generate the transgene. Axonal transport of GFP also allowed visualization of the major projections of the α5^GFP neurons. Fiber labeling, but no cell bodies, were seen prominently in thalamic regions, including the laterodorsal, central medial, ventrolateral, and ventromedial nuclei, representing the expected projections of the cortical layer 5–6 neurons (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Expression of α5^GFP and α5 mRNA in the forebrain and rostral brainstem. A–C, Expression of α5^GFP (A, B) and α5 (C) mRNA in the rostral cerebral cortex. D–F, Expression of α5^GFP (D, E) and α5 (F) mRNA in the hippocampus. The boxed area in F corresponds to the region shown in D. G–I, Expression of α5^GFP (G, H) and mRNA (I) in the habenula (not detected). Immunofluorescence for ChAT reveals cholinergic neurons in the vMHb. J–L, Expression of α5^GFP (J, K) and α5 (L) mRNA in the VTA. Immunofluorescence for tyrosine hydroxylase reveals the location of the dopaminergic neurons in the VTA. In situ hybridization views are derived from the ABA dataset. CA1, Hippocampus region CA1; cc, corpus callosum; CPu, caudate/putamen; dMHb, medial habenula, dorsal part; L5–6, cortical lamina 5–6; SNC, substantia nigra, pars compacta; SNR, substantia nigra, pars reticulata. Scale bars: A, J, K, 400 μm; D, 200 μm; B, E, G, H, 100 μm. ABA views are shown at approximately the same scale as the corresponding immunofluorescence images.

Numerous α5^GFP cell bodies were observed in the IP, consistent with database ISH data (Fig. 2A–C). The α5^GFP-labeled cells were concentrated in the IPR, encompassing part of the region that receives intense cholinergic input from the vMHb (Fig. 2B,D–F). Occasional α5^GFP cells were also observed in the caudal IP (IPC) but not in the lateral IP (IPL), which receives SP-immunoreactive fibers from the dorsal MHb (Fig. 2G,H). The most caudal part of the IP contained few α5^GFP neurons (Fig. 2I), but at this level, scattered cells and GFP fiber staining were observed in the MnR.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Expression of α5^GFP and α5 mRNA in the interpeduncular nucleus. A, Sagittal diagram of the brainstem showing the plane of section in subsequent views. B, Expression of α5^GFP in the IPR. C, ISH signal for α5 mRNA at a plane of section matching B. D–F, Expression of α5^GFP in the caudal part of the IP. The α5-expressing neurons restricted to part of the area innervated by ChAT fibers originating in the vMHb are shown. G, H, Expression of α5^GFP in the caudal part of the IP; the plane of section is similar to D. The α5^GFP does not localize to the region of the IPL containing SP-immunoreactive fibers from the dMHb. I, Expression of α5^GFP at the junction of the IP and MnR. CLi, Caudal linear nucleus raphe; IPA, interpeduncular nucleus, apical; IPF, interpeduncular fossa; Pn, pons; RtTg, reticular tegmental nucleus. Scale bars: B, D–I, 200 μm. ABA views are shown at approximately the same scale as the corresponding immunofluorescence images.

The α5^GFP-expressing neurons were also detected more caudally in the brainstem in the MnR and adjacent PMnR and in the DRI (Fig. 3A–I) but not in the principal part of the dorsal raphe or the central nucleus of the dorsal tegmentum (DTgC). In the raphe areas where they were observed, α5^GFP neurons were distinct from, but adjacent to, the serotonergic cell population marked by the expression of the enzyme Tph2 (Fig. 3C,E,H). In the laterodorsal tegmental nucleus (LDTg), α5^GFP neurons were seen to be intermingled with the cholinergic neurons that characterize this region (Fig. 3J,K). The most caudal expression of α5^GFP was in the posterior dorsal tegmental nucleus (PDTg; Fig. 3L–N). In each of these brainstem areas, there was strong correlation between the distribution of the α5^GFP signal and α5 mRNA expression.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Expression of α5^GFP and α5 mRNA in the hindbrain raphe and dorsal tegmentum. Brains of adult α5^GFP mice were examined in coronal sections for the colocalization of GFP-labeled cells and fibers and markers of selected neurotransmitters. Data are shown from progressively more caudal levels. A, Relationship of α5^GFP cells and fibers to 5-HT neurons of the median raphe. B, Relationship of α5^GFP cells and fibers to 5-HT neurons of the dorsal raphe. C, D, Enlargement of the boxed area in A and the equivalent view from the ABA showing α5 mRNA expression. E, F, Enlargement of the boxed area in B and the equivalent view from the ABA. G, Relationship of α5^GFP cells and fibers to 5-HT neurons of the dorsal raphe, intrafascicular part. H, I, Enlargement of the boxed area in G and the equivalent view from the ABA. J, Relationship of α5^GFP cells and fibers to cholinergic neurons of the dorsal tegmental nucleus. K, Enlargement of the boxed area in J. Fiber terminals near the dorsal raphe are indicated with an asterisk. L, α5^GFP GFP cells and fibers in the posterodorsal tegmental nucleus. M, N, Enlargement of the boxed area in L and the equivalent view from the ABA. Cb, Cerebellum; DR, dorsal raphe; IC, inferior colliculus; LDTgV, lateral dorsal tegmental nucleus, ventral; mlf, medial longitudinal fasciculus; PAG, periaqueductal gray; PnC, pontine reticular nucleus, caudal; SC, superior colliculus. Scale bars: A, B, 400 μm; G, J, L, 200 μm. ABA views are shown at approximately the same scale as the corresponding immunofluorescence images.

Expression of the Chrnα3/β4/α5 gene family in the habenulopeduncular system

The MHb has been proposed as a critical site of action of the α5 receptor subunit in nicotine-related behaviors, yet α5^GFP was not detected in this nucleus, even with immunofluorescent enhancement of the GFP signal. Prior studies have reported mixed findings on the expression of α5 mRNA in the MHb. The initial report of α5 cloning in the rat showed α5 mRNA expression in lamina 5–6 of the neocortex, the subiculum, IP, ventral tegmental area (VTA), and substantia nigra, pars compacta (SNc) of the adult rat but reported that it was not detected in the MHb (Boulter et al., 1990). Subsequent in situ hybridization studies using the same rat full-coding probe in mouse tissue reported low levels of α5 mRNA expression in the MHb (Marks et al., 1992; Winzer-Serhan and Leslie, 2005; Heath et al., 2010). Studies using full-coding probes for α5 in the MHb, especially across species, should be interpreted with caution because β2, β3, β4, and α3 transcripts are also expressed in this nucleus (Allen Institute for Brain Science, 2012), and cross-hybridization must be considered. Thus, to reconcile α5^GFP expression with that of the endogenous gene, we reexamined expression of the α5 mRNA by in situ hybridization and compared expression of the clustered Chrna3/b4/a5 gene transcripts by qPCR in the brain regions of interest.

To design a probe with complete specificity for α5, we amplified α5 cDNA fragments from the CDS and the 3′-UTR from adult mouse interpeduncular nucleus RNA by RT-PCR (Fig. 4A). To verify that the predicted 3′-UTR sequence was transcribed at levels similar to the coding part of the α5 transcript, we compared the expression of the 3′-UTR and coding sequences by qPCR (see Materials and Methods). Expression levels of the coding and noncoding sequences matched within one to two amplification cycles. Nonisotopic in situ hybridization with the α5 CDS and 3′-UTR probes gave a strong signal in the IP (Fig. 4B–D). Neither probe alone, nor a combination of the probes, gave a detectable signal in the MHb (Fig. 4E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Regional brain expression of nAChR mRNAs. A, Map of the Chnra5 gene locus showing the location of qPCR primers, Affymetrix 430v2 array probe sets, and in situ hybridization probes to the Chrna5 coding sequence and 3′-untranslated region. B–E, In situ hybridization for α5 mRNA using individual and combined probes in the interpeduncular nucleus (B–D) and habenula (E). F, Microarray assay of cholinergic receptor subtype mRNA expression in the E16.5 mouse habenula (medial plus lateral). Values shown are means of two independent experiments for each genotype. Null signs indicate “absent” calls, and asterisks indicate significant probability of change values using the Affymetrix MAS5.0 analysis platform, as described previously (Quina et al., 2009). Two probe sets for α5, A:1427401_at (α5C) and B:1442035_at (α5–3′), appear on the array, neither of which has a signal above the detection threshold. The α3, α4, and β4 are significantly decreased in the habenula of Brn3a null mice. G, Real-time qPCR assays of regional brain expression of α3, β4, and α5 mRNA in the hippocampus (CA1), cerebellum (Cb), cerebral cortex layers 2–4 (Ctx2–4), cerebral cortex layers 5 and 6 (Ctx5–6), MHb, and IP. Results were scaled such that expression in the highest-expressing tissue equals 100%. Dashed bars for Chrna5 expression indicate values adjusted for the lower fraction of α5^GFP-expressing neurons in cortical layers 5 and 6 and the hippocampus relative to the IPN, to approximate the level of expression per α5-expressing cells. H, Location of dissected tissues from vibratome sections used to prepare RNA for qPCR assays of nAChR expression. Cerebellum tissue was harvested as a block. Scale bars: B–E, 100 μm.

In a previous study, we used microarrays and bioinformatic analysis of the E16.5 MHb to detect transcripts enriched in the MHb relative to other brain regions and genes that are developmentally regulated in the MHb by the transcription factor Brn3a, encoded by the Pou4f1 gene (Quina et al., 2009). The microarrays used in these studies provide independent expression data for most of the nAChR subunits (Fig. 4F). At E16.5, α3 and β4 transcripts are robustly expressed and are markedly decreased in the Pou4f1 knock-out. Both α5 probe sets on the array, one corresponding to the CDS and one encompassing the 3′-UTR, showed undetectable levels of expression.

To better quantify the relative expression of α5 mRNA in the adult brain regions of interest, we performed qPCR using primer pairs specific for α3, β4, and α5 transcripts (Fig. 4G,H). Expression of α5 mRNA in the MHb was ∼3.5% of that observed in the IP. Although low, this ratio probably still overestimates the cell-specific expression of α5 in the MHb relative to the IP because the cells sampled in the MHb are much more homogenous. The fraction of cells expressing α5 in the MHb sample could not be estimated because expression could not be detected at the cellular level. Conversely, α3 and β4 mRNA levels in the IP were only ∼2% and ∼3% of the MHb levels, respectively. We conclude that the absence of detectable GFP expression in the MHb of α5^GFP mice is consistent with the very low levels of α5 mRNA expressed there and is not attributable to an artifact of low α5^GFP transgene expression. While mRNA abundance does not perfectly predict receptor expression caused by possible post-transcriptional effects on protein levels, it is thus likely that α5 subunits are much more abundant in the IP than in the MHb.

Neurotransmitter phenotypes of α5-expressing neurons

To understand the role of α5-expressing neurons in regulating behavioral pathways, we next examined the principal neurotransmitter phenotype of these cells. None of the α5-expressing cell groups in the CNS expressed ChAT (Fig. 3J–N and data not shown), indicating that α5-expressing neurons are not themselves cholinergic and that the α5 subunit is integrated into postsynaptic receptor complexes, not presynaptic autoreceptors. In the SNc and VTA, the anatomical expression pattern of α5 mRNA is consistent with expression in dopaminergic neurons (Boulter et al., 1990), but this was not assessed directly because the α5^GFP transgene was poorly expressed in this region. In the remainder of the brainstem, α5^GFP did not colocalize with tyrosine hydroxylase (dopaminergic, noradrenergic neurons) or Tph2 (5-HT neurons), indicating that is not expressed in monoaminergic cells outside of the SNc/VTA (Fig. 3A–H and data not shown).

To determine the principal neurotransmitter phenotype of α5-expressing neurons throughout the neural axis, we used DFISH (Thompson et al., 2008) to examine the coexpression of α5 mRNA with the GABAergic marker Gad1(Gad67), the brainstem glutamatergic marker Slc17a6 (Vglut2), and the neuropeptide cotransmitter somatostatin (Sst). In the forebrain, α5⁺ layer 5–6 cortical neurons that project to the thalamus are, as expected, Gad1 negative and therefore glutamatergic (data not shown). The α5⁺ neurons in the ventral subiculum also show no overlap with the Gad1⁺ interneuron population there (Fig. 5A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Neurotransmitter phenotype of α5-expressing neurons. DFISH (A–E) and immunofluorescence (F, G) were used to determine the colocalization of α5 mRNA with other markers in the IP and MnR. A, In the subiculum, α5 does not colocalize with Gad1, indicating that the α5⁺ neurons in this region are glutamatergic. Similar results were observed in layer 6 of the neocortex (not shown). B, In the IPR, >95% of α5⁺ neurons also express Gad1. C, DFISH for α5 and Slc17a6 (Vglut2) mRNA shows no colocalization in IPR, although glutamatergic neurons are present in the nucleus. D, In the MnR/PMnR, >95% of α5⁺ neurons coexpressed Gad1. The dashed line indicates the midline. E, A small number of α5⁺ neurons in the apex of IPR were seen to coexpress Sst. F, The α5⁺ neurons are negative for parvalbumin, which identifies a distinct subset of IPR neurons. G, The α5⁺ neurons are negative for the transcription factor Brn3a, which is expressed in a distinct population of IPR and IPL neurons. A′–E′, Enlargements of boxed areas in accompanying images. Sub_V, subiculum, ventral. Scale bars, 100 μm.

In the IP, essentially all α5⁺ neurons were also Gad1⁺ (Fig. 5B). Conversely, most GABAergic neurons in IPR were α5⁺, but GABAergic neurons in IPC and IPL were mostly α5 negative. Although Slc17a6⁺ neurons were intermingled with α5⁺ neurons in the IPR, colocalization of these markers was not observed (Fig. 5C). Likewise in the MnR/PMnR and DRC/DRI, where α5⁺ cells lie adjacent to 5-HT neurons, essentially all of the α5⁺ neurons also expressed Gad1 (Fig. 5D). We conclude that in the brainstem, α5 expression defines a subset of GABAergic neurons.

In the neocortex, subclasses of GABAergic neurons have been distinguished by the expression of the calcium-binding proteins parvalbumin and calretinin and by neuropeptides, particularly Sst, vasointestinal active peptide, and cholecystokinin (Xu and Callaway, 2009). We examined these markers using DFISH and immunofluorescence in α5^GFP mice to determine whether they also distinguish subsets of GABAergic neurons in the IP or colocalize with α5. A small number of neurons in the most dorsal part of the IPR coexpress Sst and α5 mRNA (Fig. 5E). Parvalbumin immunoreactivity identifies a subset of IP neurons, but these are α5^GFP negative (Fig. 5F), whereas anti-calretinin antisera stained afferent fibers but not cell bodies in the IP (data not shown). Brn3a is also expressed in a subset of neurons in the IPR and IPL, but these are α5^GFP negative (Fig. 5G).

Projections of α5 neurons link the IP and raphe

To begin to understand the function of α5-expressing neurons in the IP, MnR, and DTg, we used transgenic anterograde tract tracing and retrograde tracing with CTB to identify the principal connections of these neurons. Transgenic tracing of α5^GFP-labeled IP neurons shows that their axons project through the MnR to the DTg, both of which also include intrinsic α5 neurons (Fig. 6A). Thus, anterograde tracing of GFP-labeled fibers does not immediately reveal whether α5-expressing IP neurons project all the way to the DTg or terminate at intermediate targets. It is also not clear whether the α5 neurons in the MnR and DTg project rostrally, caudally, or locally.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Caudal projections of α5-expressing neurons in the interpeduncular nucleus. Stereotaxic injections of CTB were used to trace the projections of α5^GFP neurons in the brainstem. A, Sagittal view of the brainstem stained for α5^GFP; rostral is on the left. Lines show the planes of section for the injections shown in B, E, H, and Figure 7A. B, Location of injection site in MnR/dorsal raphe (plane of section at bregma, −4.4). C, D, Retrograde labeling of IPR neurons from MnR. In D, a Z-stack of 20 1 μm optical sections is shown, and arrows indicate neurons double labeled for CTB and GFP (see also H). E, Location of injection site in the midline of the DTg (bregma, −5.3), an area populated with 5-HT neurons of the DRC/DRI, which can be seen in Figure 3H. F, G, Retrograde labeling of IPR neurons from the central DTg. In G, a Z-stack of 20 1 μm optical sections is shown, and arrows indicate neurons double labeled for CTB and GFP (see also I). H, Images of single retrogradely labeled IP neurons from the case shown in B. Single 1 μm optical sections from a confocal Z-series are shown. Each triad of images show, from left to right, a merged image, α5^GFP, and CTB. Images in the left column are taken from the section shown in D; images in the two right columns are taken from another section from the same case. I, Images of single retrogradely labeled IP neurons from the case shown in E. Images in the left column are taken from the section shown in G; images in the two right columns are taken from another section from the same case. All of the cell profiles in both cases are taken from the IPR. Asterisks in H and I show the location of multiple cells in a single field, which may be labeled with α5^GFP, CTB, or both. J–L, Location of injection site in the lateral DTg (bregma, −5.0). K and L show the relationship of the injection to ChAT-positive and Tph2-positive neurons in this area, respectively. M, N, Retrograde labeling of IPDL neurons from the DTg. SP immunoreactivity indicates the input fibers from the dorsal MHb. The IPDL does not contain α5^GFP neurons. 4V, Fourth ventricle; 7n, facial nerve; Aq, aqueduct; Cb, cerebellum; DR, dorsal raphe; IC, inferior colliculus; IPF, interpeduncular fossa; LC, locus ceruleus; LDTgV, lateral dorsal tegmental nucleus, ventral; mcp, middle cerebellar peduncle; mlf, medial longitudinal fasciculus; MM, medial mammillary nucleus; Mo5, motor trigeminal nucleus; PAG, periaqueductal gray; Pn, pons; PnC, pontine reticular nucleus, caudal; PnR, pontine raphe nucleus; Pr5, principal trigeminal nucleus; Pr5DM, principal trigeminal nucleus, dorsomedial; Pr5VL, principal trigeminal nucleus, ventrolateral; py, pyramidal tract; R, red nucleus; RLi, rostral linear nucleus of raphe; s5, sensory root of 5 n; SC, superior colliculus; scp, superior cerebellar peduncle; SNC, substantia nigra, pars compacta; SNR, substantia nigra, pars reticulata; SO, supraoptic nucleus; vtgx, ventral tegmental decussation; xscp, decussation superior cerebellar peduncle ventral part. Scale bars: B, E, J, 400 μm; C, F, 200 μm; D, G, K–N,100 μm; H, I, 25 μm.

We then used stereotaxic injections of CTB to determine the specific projections of the α5-expressing neurons in the brainstem. CTB is a strong retrograde tracer but will also label fibers in an anterograde direction, yet it appears to label fibers of passage weakly or not at all (Luppi et al., 1990). Retrogradely labeled neurons were unambiguously identified by colocalization of CTB with GFP, which fills the cytoplasm of α5^GFP neurons, using confocal microscopy. Injection of CTB into the MnR (Fig. 6B) led to frequent labeling of α5-expressing neurons in the IPR (Fig. 6C,D). Injection of the DTg at the midline (Fig. 6E), an area that is populated with 5-HT neurons of the DRC/DRI, also led to frequent colabeling with α5^GFP in the IPR (Fig. 6F,G) and MnR (data not shown), confirming that α5-expressing IPR and MnR neurons project directly to this region. In single (1 μm) confocal sections, the profiles of IPR α5^GFP neurons retrogradely labeled from the MnR and DTg revealed cell bodies typically 15–20 μm in diameter that were often most intensely labeled with CTB at one pole, probably representing the origin of the axon (Fig. 6H,I). We note that a large fraction of the CTB-labeled neurons in IPR were α5^GFP negative, consistent with the fact that α5^GFP is expressed in only a subset of the IPR neurons. The α5^GFP/CTB double-labeled neurons were smaller than some of the CTB-labeled cells that did not express α5^GFP, consistent with the identity of the α5^GFP neurons as GABAergic interneurons (Fig. 5), which frequently have relatively small somas. We also note that many α5^GFP neurons remained unlabeled in each of these retrograde labeling cases. This is the expected result from a small, focal injection that only labels part of the overall IPR terminal field in MnR, DR, and DTg. In contrast, more lateral injections in the DTg (Fig. 6J), overlapping the region of the LDTg that contains cholinergic neurons (Fig. 6K) but avoiding the DRI (Fig. 6L), seldom labeled α5^GFP neurons in the IPR. Instead, retrograde labeling was seen in the dorsolateral IP (IPDL), mainly contralateral to the injection (Fig. 6M,N). This region did not stain for α5^GFP and could be distinguished from the adjacent IPL by its lack of SP expression. Together, these data demonstrate that α5-expressing neurons in the IP terminate near 5-HT neurons throughout the MnR and DRC/DRI but represent only one class of IP neurons that terminate in these areas, intermingled with more numerous non-α5^GFP cells. In addition to the α5^GFP cell bodies in the IPR, we also noted GFP fiber staining in the IP, especially in IPC (Figs. 2G, 6M), suggesting that α5 neurons also supply afferents to the IP. Injections of CTB into the IP (Fig. 7A) produced intense labeling of both fibers (anterograde labeling) and cell bodies (retrograde labeling) in the MnR (data not shown) and the DTg/DRI (Fig. 7B,D). Confocal imaging of retrogradely transported CTB and α5^GFP revealed that α5 neurons in the PDTg (Fig. 7C) and in the region surrounding the 5-HT neurons of the DRI (Fig. 7E,F) project to the IP. Thus, α5^GFP neurons in the IP and the MnR/DRI form reciprocal connections between these regions (Fig. 7G).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Hindbrain α5^GFP neurons project to the interpeduncular nucleus. Stereotaxic injections of CTB into the IP of adult α5^GFP mice were used to retrogradely trace IP afferents. A, Injection site of CTB in the IP, encompassing IPR, IPC, and IPL on the right side. B, C, Section through the caudal DTg at the level of the facial nerve (7n). Retrograde labeling of α5^GFP neurons in the PDTg is indicated by arrows. Neurons in this area were also retrogradely labeled by injections in the MnR (data not shown). D–F, More rostral section through the DTg. The α5^GFP neurons adjacent to DRI are retrogradely labeled from the IP (E, arrows). The location of these neurons may be equivalent to the PMnR at more rostral levels. F, Location of 5-HT neurons in an adjacent section. G, Schematic of the projections of α5^GFP neurons associated with the IP and raphe, shown in the sagittal view. Abbreviations appear in Figure 6 legend. Scale bars: A, B, 400 μm; D, F, 200 μm; E, 100 μm; C, 50 μm.

Optogenetically assisted circuit mapping of the vMHb connection to α5-expressing IP neurons

Previous work has shown that vMHb neurons use both glutamate and acetylcholine as coneurotransmitters but that fast synaptic connections to IP neurons are mediated mainly by glutamate (Ren et al., 2011). The identities of the postsynaptic IP neurons recorded were not specifically characterized in this study. To assess the synaptic connection between vMHb neurons and α5^GFP neurons in the IP, we used optogenetically assisted circuit mapping. Specific expression of ChR2 in the vMHb was accomplished using a recently described Cre-inducible expression system based on microbial opsin genes targeted to the Gt(ROSA)26Sor locus (Madisen et al., 2012). In this study, we used the transgenic lines Ai27, which allows inducible expression of a Channelrhodopsin2(H134R)-tdTomato fusion gene product (ChR2-tdTomato), and Ai32, which allows expression of a ChR2-EYFP fusion gene product. Ai32 conferred somewhat higher light sensitivity, but the performance of the two ChR2 fusion proteins in electrophysiological experiments was similar. Expression of each ChR2 allele was induced by interbreeding with a Chat^Cre driver line containing an IRES-Cre expression construct targeted to the 3′-untranslated region of the Chat gene (Rossi et al., 2011). In mice with Chat^Cre induction of ChR2, expression of the transgene was confined to the vMHb, with sparing of the dorsal MHb and the LHb (Fig. 8A). In the IP, ChR2-labeled fibers were confined to the rostral and caudal parts of the IP (Fig. 8B), known to be the termination zone of cholinergic vMHb fibers (Eckenrode et al., 1987; Quina et al., 2009). For optogenetic circuit mapping, experimental animals had the genotype Ai27 (or Ai32), Chat^Cre, or α5^GFP. The α5^GFP-expressing IP neurons were readily visualized using GFP fluorescence in vibratome sections of brain tissue prepared for electrophysiology (Fig. 8C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Optogenetically assisted circuit mapping of vMHb inputs to α5-expressing neurons in the IPR. A, Expression of ChR2-EYFP fluorescence and immunofluorescence for ChAT in the MHb of a Chat^Cre, Ai32 mouse. Expression of ChR2-EYFP is restricted to the cholinergic vMHb. B, Marker expression in the IP of a Chat^Cre, Ai32 mouse. Expression is restricted to the termination zone of the cholinergic MHb fibers in the IPR/IPC, sparing the IPL. C, Image of GFP fluorescence superimposed on a differential interference contrast view of a section through the IPR in a α5^GFP, Chat^Cre, Ai27 mouse prepared for “visual patch” electrophysiological recording. The shadow of the patch pipette is marked with a dashed line. D, Loose-seal, cell-attached current-clamp recording of a vMHb neuron from a Chat^Cre × Ai32 mouse (as in A). At baseline, irregular tonic firing at ∼3 Hz is observed. Application of 447 nm blue light pulses (20 ms, 2.0 mW/mm²) at 10 Hz elicits action potentials entrained to the pulse frequency. The maximum stimulation rate for which the cell shown would generate 1:1 action potentials was ∼10 Hz; at higher stimulation frequencies, some light pulses failed to elicit spikes. At the end of a 10 s interval of pulsed light delivery, a period of suppressed firing was observed (∼3 s, not shown), followed by a gradual resumption of firing at the baseline rate. E, Examples of EPSCs from a single α5^GFP IP neuron from a α5^GFP, Chat^Cre, Ai27 mouse evoked by a 10 ms light pulse. EPSCs from 6 of 122 sweeps are shown (56 of 122 sweeps elicited EPSCs in this cell). EPSCs from this cell showed multiple consistent latencies relative to light onset and multiple amplitudes, suggesting input from more than one light-responsive presynaptic unit. F, Histogram of the latencies of all light-evoked EPSCs from the neuron shown in E. Bars depict the probability of observing the onset of an EPSC (event) within an 0.5 ms epoch at the specified times after light onset (latency). Light onset is at time 0; EPSCs were thresholded as described in Materials and Methods. The latencies for this cell are bimodally distributed, with peaks at 6.5–7 ms and 9–9.5 ms, suggesting inputs from two different presynaptic units. Other cells showed a single peak. G, Probability of light-evoked EPSCs in α5^GFP neurons summed over the 3–11 ms period after a light pulse (stimulated, n = 6), after a “sweep” without a light pulse (baseline, n = 6), and after a light pulse in the presence of glutamatergic blockers (n = 4). The probability of evoking a light-stimulated EPSC for the cells shown ranged from 0.16 to 0.94. Inhibition of EPSCs by blockade of glutamatergic signaling also demonstrates that the recorded α5^GFP cells were not intrinsically photostimulated. p = 0.01 (paired t test; n = 6 cells). Results are presented as means (SD). dMHb, medial habenula, dorsal part; fr, fasciculus retroflexus; LHb, lateral habenula. Scale bars: A, B, 100 μm.

To test light activation of ChR2-expressing vMHb neurons, we first used loose-seal, cell-attached current-clamp recordings from vMHb cell bodies (Fig. 8D). At baseline, neurons in the vMHb exhibited slow, irregular tonic firing, consistent with previous reports of spontaneous habenula activity (Kim and Chang, 2005). Trains of blue light pulses elicited 1:1 spiking of action potentials from vMHb neurons at frequencies of up to ∼10 Hz (Fig. 8D). We then used whole-cell voltage-clamp recordings to measure light-evoked EPSCs in α5-expressing IP neurons. In oblique slice preparations that contained the intact habenulopeduncular system, including the vMHb, fasciculus retroflexus, and IP, very frequent spontaneous EPSCs were observed in IP neurons, presumably because of network activity (data not shown). To better isolate the light-evoked EPSCs caused by specific synaptic connections between the vMHb and the IP, experiments were performed in coronal sections through the IP, which included only the fiber terminals and not the cell bodies, of the vMHb glutamatergic/cholinergic neurons.

Light-evoked EPSCs recorded from individual α5^GFP IP neurons frequently showed multiple consistent latencies and amplitudes, suggesting multiple vMHb synaptic inputs to a single cell (Fig. 8E). These light-evoked EPSCs generally occurred with latencies of 5–11 ms from the onset of a 10 ms light pulse. Median latencies of 5, 6.2, 6.5, 6.5, 8.5, and 9 ms were observed for the six cells characterized (the cell with a median latency of 9 ms is shown in Fig. 8F). The median latencies exhibited by these neurons are within the typical latencies for monosynaptic transmission elicited by optogenetic excitation of axon terminals across several studies (Piñol et al., 2012). In the sample of cells recorded, the probability of an EPSC occurring in the interval from 3 to 11 ms after a light pulse ranged from 0.16 to 0.94, compared with 0.01–0.15 for matched time intervals without laser stimulation (p = 0.01, paired t test; n = 6; Fig. 8G). Both spontaneous and light-evoked EPSCs were entirely blocked by inhibitors of glutamatergic signaling (DNQX and d-AP-5), consistent with a previous study showing that glutamate mediates the fast synaptic connection from the vMHb to IP (Ren et al., 2011). We conclude that the vMHb and α5-expressing IP neurons form a direct synaptic circuit in which glutamate is the principal fast neurotransmitter. These results are consistent with a “neuromodulatory” role for acetylcholine in the IP, as described in other CNS regions (Picciotto et al., 2012).

Nicotinic receptor pharmacology of α5^GFP neurons in the IP

The habenulopeduncular system has been hypothesized to regulate behavioral responses to nicotine. Because α5 is an accessory receptor subunit, the other α and β nicotinic receptor subunits with which it interacts will determine the ligand-binding and ion channel properties of α5-containing receptors. In most areas of the brain in which it is expressed, α5 is likely to complex with the widely expressed receptor subunits α4 and β2. However, it is unclear what subunits would interact with α5 in the IP and, specifically, whether sufficient α3 and β4 subunits are expressed to allow for the possible formation of functional α3β4 complexes containing α5 subunits. The principal nAChRs reported from ligand-binding studies of IP homogenates contain α3β4 and α4(or α2)β2 (Grady et al., 2009). Autoradiographic binding studies of nicotinic receptor knock-out mice show β4-containing receptors are required for much but not all of IP nicotinic ligand binding (Baddick and Marks, 2011). However, ligand binding in the IP may be contributed by either the presynaptic vMHb fiber terminals or the postsynaptic IP neurons, and these cannot be resolved by ligand-binding studies in homogenates or autoradiography. Instead, to determine the likely subunits associated with α5^GFP in the IP, we took an electrophysiological approach, using pulsed application of nicotine combined with specific inhibitors of nAChR subtypes. First, however, we considered whether the use of the α5^GFP BAC transgene might lead to the mis-expression of α3 and β4 nAChR subunits, because of the genomic clustering of the Chrna3, Chrnb4, and Chrna5 genes that are all encompassed within the parent BAC. Because α3 and β4 are cotranscribed on the same strand, we chose to examine β4 mRNA expression by in situ hybridization in the MHb and IP. The β4 expression appeared to be increased in the vMHb of α5^GFP mice relative to controls (Fig. 9A,B) but was unchanged in the IP, where only a faint signal could be detected (Fig. 9C,D). This demonstrates that the fidelity of tissue-specific gene expression from this BAC transgene is maintained for the transgenic copies(s) of the Chrnb4 gene and that use of the α5^GFP BAC transgene as a marker to identify specific IP neurons is unlikely to affect the nAChR subunit composition of those neurons.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Nicotinic pharmacology of α5⁺ IPR neurons. A–D, Chrnb4 mRNA expression in the habenulopeduncular system of control and α5^GFP mice. The α5^GFP BAC drives overexpression of Chrnb4 in the habenula but not ectopic expression in the IP. E–L, Pharmacological studies of nicotine-evoked currents in the vMHb and α5^GFP IP neurons. Arrowheads indicate the pulsed application of 0.5 μm nicotine for 90 ms. Black traces, Control or baseline responses to nicotine; red traces, nicotine current responses in the presence of the stated inhibitor in bath; blue traces, nicotine current responses during recovery after washout. E, F, Nicotine-evoked currents in the vMHb are nearly abolished by mecamylamine (Mec; 5 μm) and markedly inhibited by SR16584 (20 μm), a selective inhibitor of α3β4-containing receptors. The arrow in F indicates an action potential evoked immediately after nicotine application but beyond effective space clamp. G, H, Summary of inhibitor effects on nicotine-evoked currents in vMHb neurons. G, Effect of mecamylamine, which was partially reversible. **p < 0.01 compared with control (Friedman test and Dunn's multiple comparisons test; n = 4 cells). H, Effect of SR16584. *p < 0.05 compared with control (Friedman test and Dunn's multiple comparisons test; n = 6 cells). I, J, Nicotine-evoked currents in α5^GFP IP neurons are inhibited by DHβE (2 μm), a partially selective inhibitor of α4β2 receptors, and by SR16584. K, L, Summary of inhibitor effects on nicotine-evoked currents in the IP. K, Effects of DHβE. *p < 0.05 (Wilcoxon test; n = 10 cells). Only 6 of 10 cells appeared to respond to this inhibitor. L, Effects of SR16584. *p < 0.05 (Wilcoxon test; n = 6 cells). All recordings were obtained in the presence of DNQX (10 μm) and d-AP-5 (50 μm) to block glutamatergic signaling. In the “box and whisker” graphs in G, H, K, and L, the box extends from the 25th to 75th percentile, the median is designated by the line in the middle of the box, and the whiskers indicated the smallest and largest values. Abbreviations appear in Figure 4 legend. Scale bars: A–D, 100 μm.

To assess the contribution of specific nAChR complexes to nicotine responses in the vMHb and in α5^GFP neurons in the IP, we examined nicotine-evoked currents in the presence of DHβE, a partially selective inhibitor of α4β2-containing receptors, and SR16584 (Zaveri et al., 2010), a specific inhibitor of α3β4-containing receptors, in the presence of glutamate receptor blockers. To minimize nAChR desensitization known to occur with prolonged nicotine exposure (Covernton and Lester, 2002), responses were evoked by brief pulsed applications of nicotine every 3 min (90 ms pulse, 0.5 mm nicotine; Fig. 9E–L). Pulsed application of nicotine evoked much stronger currents in cholinergic vMHb neurons than in α5^GFP IP neurons. The peak amplitude of the nicotine currents was 1564 ± 1225 pA (range, 633–4076 pA; n = 10 cells) in the vMHb and 297 ± 211 pA (range, 110–853 pA; n = 16 cells) in the IP. The much greater magnitude of evoked currents observed in the vMHb was not surprising given the high levels of nAChR mRNA known to be expressed there and past studies showing intense nAChR ligand binding in the MHb and its efferent fibers (Harris et al., 1996).

In vMHb neurons, the broadly specific nAChR inhibitor mecamylamine abolished nearly all of the inward nicotine currents as expected, and this inhibition was partially reversible (Fig. 9E,G; control, 1398 ± 1097 pA; mecamylamine, 88 ± 29 pA; p < 0.01, Friedman test and Dunn's multiple comparisons test; n = 4 cells). SR16584 inhibited a majority of these currents in the vMHb (Fig. 9F,H; control, 1674 ± 1394 pA; SR16584, 282 ± 127 pA; p < 0.05, Friedman test and Dunn's multiple comparisons test; n = 6 cells), demonstrating that α3β4-containing nAChRs account for most of the vMHb nicotine response; this inhibitor was poorly reversible in washout experiments. In α5^GFP IP neurons, nicotine-induced inward currents were partly inhibited by DHβE in 6 of 10 of the neurons recorded (Fig. 9I,K; control, 259 ± 158 pA; DHβE, 212 ± 155 pA; p < 0.05, Wilcoxon test; n = 10 cells) and by SR16584 in all neurons recorded (Fig. 9J,L; control, 360 ± 286 pA; SR16584, 84 ± 69 pA; p < 0.05, Wilcoxon test; n = 6 cells). These data suggest the presence of both α4β2- and α3β4-containing nAChRs in the IP, either of which could potentially incorporate α5 subunits.