Chrna5-Expressing Neurons in the Interpeduncular Nucleus Mediate Aversion Primed by Prior Stimulation or Nicotine Exposure

Electrophysiology of the IP response to ACh and nicotine

To address directly the significance of the expression of the Chrna5 subunit in the rostral interpeduncular nucleus, we performed acute slice recordings from sibling WT (α5^+/+) and α5^KO mice. The intrinsic properties of IPR neurons were recorded and compared by genotype (Table 1). There were no significant differences on measures of input resistance, resting membrane potential, capacitance, spike amplitude, or spike threshold.

We first investigated how the absence of Chrna5 subunit expression affected current response to bath application of ACh (1 mm). Data for both the WT and α5^KO groups failed the Shapiro–Wilk test for normality (WT, n = 27, W = 0.75, p < 0.0001; α5^KO, n = 27, W = 0.72, p < 0.0001, Fig. 1-1 relating to Fig. 1), necessitating the use of nonparametric analyses. The amplitude of current responses was markedly reduced in a5^KO IP neurons (WT, 227 ± 50 pA, n = 27; α5^KO, 37 ± 7 pA, n = 27, Mann–Whitney U = 70.0, p < 0.0001; Fig. 1A,B) and this genotype difference was observed across a range of concentrations (100 μm to 1 mm; WT, n = 8; α5^KO n = 6; F_{1, 34} = 16.6, p = 0.0003; Fig. 1C,D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Acetylcholine and nicotine currents in IPR neurons of WT and α5KO mice. A, Example traces of ACh (1 mm) current responses in WT and α5^KO IPR neurons performed in voltage clamp at a holding potential of −75 mV. B, Mean ACh responses are significantly larger in WT compared with α5^KO neurons (Mann–Whitney U = 70.0, p < 0.0001). C, Example responses of IPR neurons from WT and α5^KO mice to 100 μm, 300 μm and 1 mm ACh. D, Quantitative comparison of current responses of WT and α5^KO neurons show significant differences are maintained across this range of ACh concentrations (F_1,34 = 16.6, p = 0.0003). E, Example traces of current responses to ACh stimulation in IPR neurons from WT and α5^KO mice in the presence of the synaptic blockers APV, CNQX, and bicuculline. F, Quantification of ACh currents in the presence of synaptic blockers demonstrates that marked differences persist in the direct ACh response in WT compared with α5^KO neurons (Mann–Whitney U = 4.0, p = 0.005). G, Example traces of ACh current responses in WT and α5^KO neurons before and after bath application of the α4β2* nAChR antagonist DHβE (10 μm). H, DHβE significantly reduces the mean current amplitude of responses to ACh in both WT and α5^KO neurons (Wilcoxon W = −45.0, p = 0.004). I, Example traces of nicotine (1 μm) current responses in WT and α5^KO IPR neurons. J, Mean nicotine responses are significantly larger in WT compared with α5^KO neurons (Mann–Whitney U = 31.0, p = 0.005). See Figure 1-1 (in table form). *p < 0.05, **p < 0.01, ****p < 0.0001.

IP neurons are predominantly GABAergic and receive synaptic inputs from the MHbV that are both glutamatergic and cholinergic (Ren et al., 2011; Hsu et al., 2013). The MHbV terminals are rich in nAChRs that could act presynaptically to facilitate glutamate release. In the presence of synaptic blockers (APV, CNQX, and bicuculline), there was a significant reduction in the current amplitude of responses to ACh in within-cell experiments (WT baseline, 252 ± 75, synaptic blockers, 147 ± 52 pA, n = 9; α5^KO baseline, 41 ± 11, synaptic blockers, 19 ± 5 pA, n = 6; Wilcoxon W = −110.0, p = 0.006, data not shown). However, in the presence of synaptic blockers, there remains a strong and significant genotype difference in cholinergic currents (WT: 147 ± 52 pA, n = 9; α5^KO: 18 ± 5 pA, n = 6; Mann–Whitney U = 4.0, p = 0.005; Fig. 1E,F). This demonstrates that glutamatergic synaptic transmission contributes to the ACh response of IP neurons, but their intrinsic response to ACh is strongly affected by the presence of the α5 subunit.

To determine the principal subtype of channel-forming nAChRs mediating the cholinergic response in the IPR and thus potentially associated with α5 subunits in this nucleus, we blocked ACh-induced currents with the β2* nAChR antagonist DHβE (10 μm, 10 min). Cholinergic currents were substantially reduced by DHβE in both genotypes (WT: baseline, 360 ± 143; DHβE, 75 ± 25 pA; n = 5; α5^KO: baseline, 44 ± 14; DHβE, 8 ± 3 pA; n = 4; Wilcoxon W = −45.0 p = 0.004; Fig. 1G,H).

Because the current responses to ACh application were largely mediated by nAChRs, we next applied nicotine (1 μm) in the bath to assess differences in response between genotypes. Inward currents elicited by nicotine were significantly greater in WT neurons (WT, 122 ± 35 pA, n = 14; α5^KO, 16 ± 4 pA, n = 12; Mann–Whitney U = 31.0, p = 0.005; Fig. 1I,J). These results indicate that the Chrna5 gene is critical for the normal response of IPR neurons to nicotinic stimulation.

Contrasting nicotine-induced excitability of WT and α5^KO neurons in the IP and MHbV

To investigate the question of where α5 exerts its maximal effects within the habenula–IP pathway, we examined the ability of nicotine to excite IP neurons and MHbV cholinergic neurons in adult WT and α5^KO mice (Fig. 2). Nicotine (1 μm) was bath applied to WT and α5^KO neurons in the IPR in current-clamp conditions (Fig. 2A). The peak nicotine-elicited spike frequency was greater in neurons from WT (13.8 ± 5 Hz, n = 15) than α5^KO mice (1.6 ± 1 Hz, n = 12, Mann–Whitney U = 46.0, p = 0.02; Fig. 2B–D; data failed the Shapiro-Wilk test for normality, necessitating the use of non-parametric analyses, see Fig. 2-1). Voltage-clamp recordings confirmed the significant difference in nicotinic currents observed above (WT, 144.9 ± 50 pA, n = 7, α5^KO, 14.5 ± 9 pA, n = 12, Mann–Whitney U = 3.0, p < 0.001). Membrane and action potential properties did not differ significantly between genotypes except for greater input resistance in the α5^KO neurons in this sample (WT, 348 ± 90 MΩ, n = 8, α5^KO, 812 ± 164 MΩ, Mann–Whitney U = 18, p = 0.03), which would suggest that they were IP cells of a type that would otherwise have the most prominent responses to nicotine (Shih et al., 2014).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Assessing the effects of nicotine on neuronal excitability in the MHbV and IP in WT and α5^KO mice. A, Schematic of anterior recording site for IP neurons. B, Example traces of IPR neuron responses to nicotine in WT and α5^KO mice. C, Average IPR neuronal firing frequency across baseline, nicotine (1 μm, 1 min), and washout recording. D, Peak nicotine-elicited firing frequency is significantly larger for WT than α5^KO IPR neurons (Mann–Whitney U = 46.0, p = 0.02). E, Schematic illustrating recording sites for MHbV neurons. F, Example traces of MHbV neuron responses to nicotine at resting membrane potential in WT and α5^KO mice. G, Average MHbV neuronal firing frequency across baseline, nicotine (1 μm, 1 min) and washout recording. H, Peak nicotine-elicited firing frequency does not differ between WT and α5^KO mice in MHbV neurons at resting membrane potential (Mann–Whitney U = 49.0, p = 0.2). I, Example traces of MHbV neuron responses to nicotine in WT and α5^KO mice recorded from −70 mV. J, Average MHbV neuronal firing frequency across baseline, nicotine (1 μm, 1 min) and washout recording from −70 mV. K, Peak MHbV neuronal firing frequency does not differ between WT and α5^KO mice recorded from −70 mV (Mann–Whitney U = 58.0, p = 0.3). See Figure 2-1 (in table form). *p < 0.05.

Because cholinergic neurons in the MHbV are the vital part of the habenula–IP pathway, we also tested the response of these neurons to nicotine. We recorded from the tonically firing, YFP-positive ChAT-ChR2 neurons in the MHbV of either hemisphere (Fig. 2E,F) and confirmed the presence of ChR2 (see Materials and Methods) in both WT and a5^KO transgenic mice. Intrinsic properties of MHbV neurons were measured and compared by genotype and no significant differences between genotypes were found on any measures (Table 2). In the first set of experiments, nicotine (1 μm) was applied while neurons were firing tonically at rest in current clamp, increasing their firing frequency (Fig. 2F). In contrast to the IP recordings, however, peak nicotine-elicited firing frequency did not differ significantly between MHbV neurons from WT (9.3 ± 1.4 Hz, n = 12) and α5^KO (7.9 ± 1.4 Hz, n = 13) mice (Mann–Whitney U = 58.0, p = 0.29; Fig. 2F–H). To avoid a potential ceiling effect, we next assessed nicotine-induced firing in current clamp from an initial membrane potential of −70 mV, similar to the starting membrane potential in the IP experiments. Nicotine depolarized MHbV cholinergic neurons, as illustrated in Figure 2I, yet no difference was found in nicotine-elicited peak firing frequency between genotypes (WT, 4.3 ± 0.8 Hz, n = 12, α5^KO, 3.4 ± 1.1 Hz, n = 12, Mann–Whitney U = 49.0, p = 0.19; Fig. 2I–K). Additional voltage-clamp experiments with neurons held at −75 mV did not detect any significant difference in the response to nicotine between genotypes (WT, 20.6 ± 5 pA, n = 12, α5^KO, 15.0 ± 3.8 pA, n = 11, Mann–Whitney U = 52.5, p = 0.4). In a final series of experiments, we recorded from nonlabeled, tonically firing neurons in the MHbV of WT and a5^KO mice from the original a5^+/− cross. In these MHbV neurons recorded at rest, nicotine-elicited peak firing frequencies also did not differ significantly between genotype (WT, 9.2 ± 1.6 Hz, n = 12, α5^KO, 11.9 ± 1.2 Hz, n = 5, Mann–Whitney U = 20.5, p = 0.3), nor were there significant differences in MHbV membrane properties between genotypes (data not shown).

Generation and characterization of Chrna5^Cre transgenic mice

The cell-specific expression of the α5/α3/β4 nAChRs determines the possible repertoire of pentameric receptors in a given neuron and thus the function of these receptors in specific brain regions. The Chrna3 and Chrnb4 genes are transcribed on the same DNA strand and, in rodents, they are highly expressed in the MHbV. Chrna5, which is transcribed on the opposite strand, is highly expressed in the IP and moderately expressed in the VTA and is also found in other CNS areas not necessarily correlated with transcription of Chrna3/b4, including layer 6 of the cerebral cortex, CA1 of the hippocampus, and the median/paramedian raphe (Wada et al., 1990; Marks et al., 1992; Winzer-Serhan and Leslie, 1997; Salas et al., 2003; Hsu et al., 2013). Mouse genomic BACs modified to express GFP or Cre-recombinase have been used to mark and manipulate gene expression in neurons expressing one or more of the genes in the Chrna5/a3/b4 cluster (Frahm et al., 2011; Hsu et al., 2013). Because the three genes at this locus are tightly linked, BAC transgenic models that target one of the three genes at this locus have the disadvantage that they may misexpress the other genes (Ables et al., 2017). To address these limitations and to provide a better model for studies of α5-expressing IP neurons, we constructed a new BAC-derived Chrna5^Cre transgene that completely eliminates the Chrna3 and Chrnb4 transcribed sequences (Fig. 3) and thus cannot overexpress these subunits. The parent BAC clone RP24–318N16 was modified as part of the GENSAT project by the insertion of a Cre recombinase expression cassette immediately after the Chrna5 translation start site, blocking the native Chrna5 open reading frame (Gerfen et al., 2013). This BAC clone contains most of the intergenic region between Chrna5 and the upstream locus Psma4, the entire Chrna5 expressed region, and the entire Chrna3 and Chrnb4 loci, including several regions of high interspecies conservation (Fig. 3C,D). In humans, sequence variants in this genomic region appear to affect CHRNA5 expression levels (Smith et al., 2011). We also examined database information for CTCF binding sites within the locus determined by chromatin immunoprecipitation as part of the ENCODE project (Fig. 3E). Such sites are associated with chromosomal insulators that isolate enhancer function (Ali et al., 2016). We note that a cluster of CTCF-binding sites are found in the short intergenic region between Chrna5 and Chrna3, suggesting that these genes can be independently regulated and the 3′ end of the transgenic construct was chosen to coincide with these sites. We then derived a ∼50 kb subclone of the parent BAC that included all of the available Chrna5 upstream flanking sequence plus the inserted Cre-transcribed region and the Chrna5 exonic, intronic, and 3′ untranslated regions, ending between the closely opposed Chrna5 and Chrna3 loci (Fig. 3F,G, and Materials and Methods).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Structure of the mouse Chrna5/a3/b4 gene cluster and Chrna5-derived transgenic models. The upper section (A–E) shows the structure of the Chrna5/a3/b4 genomic locus and BAC clones within the locus that have been used for transgenic studies. The lower section (F,G) shows the new transgenic construct reported here. The locus structure and design of transgenic constructs is based on NCBI build 37 of the mouse genome. A, Genomic reference coordinates for the Chrna5/a3/b4 locus. B, Location of BACs used, in truncated form, in the present work (RP24-318N16) and in previously reported studies: RP24-318N16 (Hsu et al., 2013), RP23-366O6 (Frahm et al., 2011). C, Location of transcribed genes in and flanking the Chrna5/a3/b4 locus. Chrna5 is transcribed on the forward strand and Chrna3 and Chrnb4 are transcribed on the reverse strand as shown. D, Mammalian species conservation within the locus. E, ChIP-seq data for CTCF transcription factor binding sites (TFBS) in mouse CH12 cells (ENCODE project data from UCSC genome browser). These sites tend to occur at “insulator” sites between independently regulated regions of chromatin. Arrow indicates the CTCF-binding peak used to define the distal end of the Chrna5^Cre transgenic construct. F, Structure of the parent BAC-GFP and BAC-Cre constructs from the GENSAT project. G, Structure of the modified BAC-Cre construct used in the present study. Cre recombinase is inserted at the translation start site (ATG) of the Chrna5 gene, disrupting expression of the native gene. Dashed red lines indicate the extent of the transgenic construct in the context of the genomic locus. The transcription start site is indicated by an arrow. The Chrna3 and Chrnb4 transcription units are entirely deleted from the transgenic construct.

The BAC-derived transgenic construct Chrna5^Cre was injected into mouse oocytes and three founder lines were derived. Founders were crossed with a Cre-inducible reporter strain, Ai6, which gives ZsGreen expression in the presence of Cre-recombinase (Fig. 4). Overall, the expression pattern of the Cre-induced reporter expression in Chrna5^Cre/Ai6 mice resembled the previously described expression of a Chrna5^GFP transgene derived from the same parent BAC (Hsu et al., 2013). In the IP, expression driven by all three founder lines closely resembled expression of the endogenous mRNA (Fig. 4A–F), including expression in the IPR, the dorsolateral IP subnucleus (IPDL), and the intermediate and central IP subnuclei (IPI/IPC) and lack of expression in lateral IP subnucleus (IPL). Chrna5-expressing neurons in IPR and IPDL consist predominantly of GABAergic projection neurons (Hsu et al., 2013), whereas IPI contains a preponderance of Chrna5-expressing GABAergic neurons that are local interneurons (Ables et al., 2017). In the VTA, expression of the transgene was very sparse (lines 1 and 2) or absent (line 3; Fig. 4G–I) and clearly did not reproduce the native pattern of Chrna5 mRNA expression in dopaminergic VTA neurons (Fig. 4J). A Chrna5^GFP BAC transgene derived from the same genomic region is also not expressed in the VTA (Hsu et al., 2013), so it appears that enhancers controlling VTA expression reside outside the region included in the parent BAC.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Chrna5^Cre-mediated reporter expression in subcortical areas. Chrna5^Cre mice from three founder lines were interbred with the Ai6 reporter strain, conditionally expressing the marker ZsGreen to characterize cell-specific cre-mediated expression. A–C, ZsGreen fluorescence in the interpeduncular nucleus corresponding to axial level bregma −3.5 in a standard atlas (Paxinos and Franklin, 2001). Expression is nearly identical when driven by each of three Cre-driver lines. D, E, Chrna5 ISH at levels slightly rostral (D) and slightly caudal (E) to that shown in A–C. F, Chrna5/TH DFISH at a level close to that shown in A–C. G–I, ZsGreen fluorescence is sparse (lines 1 and 2) to absent (line 3) in the VTA (bregma −3.0). J, FISH data for Chrna5 and TH mRNA in the VTA at a level similar to G–I demonstrating that the expression of the endogenous Chrna5 mRNA is more extensive than that of the Chrna5^Cre-driven reporter. K–M, ZsGreen fluorescence in the habenula. N–P, Comparison of expression for Chrna5 (undetectable), VAChT, and Chrna3 mRNA in the habenula aligned approximately with the level shown in K–M. Data in D, E, N, O, P are from the Allen Brain Atlas; data in F and J are from the Allen Transgenic Characterization Project (courtesy of Dr. Hongkui Zeng). All dark-field sections shown are counterstained with DAPI (blue). LHb, Lateral habenula; MHb, medial habenula (MHbD, dorsal, and MHbV, ventral subnuclei); SNc, substantia nigra, pars compacta; SNr, substantia nigra, pars reticulata. Scale bar, 200 μm.

Two Chrna5^Cre transgenic lines showed some expression in the MHb and LHb, but in different sets of neurons (Fig. 4K,L). The third transgenic line showed negligible habenula expression (Fig. 4M). In the adult habenula, Chrna5 mRNA is weakly expressed and is not detectable by nonisotopic ISH or FISH (Fig. 4N and data not shown), but has been detected by ³⁵S-ISH (Franceschini et al., 2002; Salas et al., 2003) and qPCR (Hsu et al., 2013). Because of the weak ISH signal and the lack of available antibodies for immunohistochemistry, the expression of Chrna5 has not been assigned to any specific habenula subnucleus. Marker expression in the two lines that showed habenula signal did not correlate with mRNA expression for the MHbV cholinergic markers vesicular acetylcholine transporter (VAChT) and Chrna3 (Fig. 4O,P). Because of this inconsistency, the habenula expression in these lines can be considered ectopic. Chrna5^Cre line 3 was used for subsequent functional studies of the IP because of its lack of potentially confounding expression in the habenula and VTA.

Chrna5 mRNA is also expressed in layer 6 of the cerebral cortex and cortical neurons in α5^KO mice show altered responses to acetylcholine (Bailey et al., 2010). Crosses of the three Chrna5^Cre founder mice with the Ai6 reporter line revealed transgene expression in cortical layer 6 in one line (Fig. 5A–C), but with this was associated with significant ectopic expression in a radial pattern consistent with clonal expression (Fig. 5B). To determine the neurotransmitter phenotype of the Chrna5^Cre-expressing neurons, Chrna5^Cre-mice were generated expressing a Cre-dependent tdTomato reporter, Ai14 (Madisen et al., 2010), and a Cre-independent marker of GABAergic neurons, Gad67-GFP (Tamamaki et al., 2003). This is of interest because α5* receptors have been hypothesized to mediate nicotinic responses in VIP-expressing GABAergic interneurons in the mouse prelimbic cortex (Koukouli et al., 2017). Chrna5^Cre-expression was not detected in GABAergic interneurons (Fig. 5D), implying that it is restricted to glutamatergic pyramidal neurons. This could be because the regulatory sequences contained in the Chrna5^Cre transgene do not drive correct expression in GABAergic neurons or because the Chrna5 mRNA is not significantly expressed in these cells. For this reason, we examined the expression of mRNA for Gad1, a GABAergic marker, and Chrna5 in the prelimbic region of WT mice by DFISH (Fig. 5E). Coexpression of these markers was very rarely observed either within pyramidal cell layers 5 and 6, where most of the Chrna5-expressing neurons reside, or in more superficial layers, where Chrna5 neurons are scattered. However, occasional Gad1-expressing neurons appeared to express low levels of Chrna5 mRNA. This suggests that Chrna5^Cre-driven reporter expression correctly reports that the α5 subunit is expressed in layer 6 cortical pyramidal neurons, but only rarely and/or at very low levels in the interneurons of any layer.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Chrna5^Cre-mediated reporter expression in the cerebral cortex. A–C, Expression of ZsGreen in the cerebral cortex of Chrna5^Cre mice from three founder lines interbred with the Ai6 reporter strain. Lines 1 and 3 exhibit low expression; line 2 exhibits appropriate expression in deep cortical layers, but expression is also seen sporadically in more superficial cells that appear to be clonal in origin and are likely to result from occasional developmental activation of the reporter (asterisks). Coronal sections at bregma 1.9. D, Colabeling of GABAergic interneurons in green (Gad67-GFP) and Chrna5^Cre neurons in red (Chrna5^Cre and Ai14/tdTomato). Inset detail shows that Chrna5^Cre neurons are exclusively excitatory pyramidal cells. Also shown is a coronal section at bregma 2.3. E, DFISH for Chrna5 and Gad1/Gad67 mRNA. Little overlap in expression is observed, but occasional Gad1-expressing cells appear to express low levels of Chrna5 (inset, arrows). Fml, Forceps minor, corpus callosum; DP, dorsal peduncular cortex; IL, infralimbic cortex; lv, lateral ventricle; Mo, motor cortex; PrL, prelimbic cortex. Scale bar, 200 μm. See Figure 5-1.

Remarkably, although reporter crosses of Chrna5^Cre line 3 founder mice showed little cortical expression, these mice bred one generation into a C57BL/6 background (N1) showed extensive, appropriate expression in the cortex (Fig. 5-1). Analysis of the transmission frequency of the Chrna5^Cre allele in line 3 was consistent with Mendelian transmission of a single insertion site (∼50% transmission), ruling out the presence of multiple insertion sites driving different expression patterns. We infer that the variable expression in the cortex is due to epigenetic effects on cortical expression of the Cre transgene. No such variability of Cre-dependent transgene expression was observed in the IP, where the induction of ROSA26-targeted ZsGreen (Ai6) and ChR2-YFP (Ai32) transgenes was 100% penetrant in all generations. The Chrna5^Cre transgenic lines derived here contain only the Chrna5 gene and upstream regulatory sequences and do not contain coding or regulatory information from the Chrna3/b4 region of the combined Chrna5/a3/b4 locus. Therefore, these results demonstrate the autonomous transcriptional regulation of the Chrna5 gene in specific brain regions (IP, cortex) independent of enhancers that might reside in the Chrna3/b4 region of the gene locus. In other cell types, such as the VTA, information from outside the transgenic construct used here appears to be required for correct expression. These findings may help to provide a framework for the interpretation of the noncoding polymorphisms found in smoking-related haplotypes (Barrie et al., 2017).

Specific projections of Chrna5-expressing neurons in the IP and the mesopontine raphe

The IP subnuclei have distinct afferents and efferents. Chrna5^Cre is expressed in IPR and IPC/IPI, which receive input from MHbV, and excluded from IPL, which receives input from the dorsal MHb (Quina et al., 2017). To map the specific efferents of the Chrna5^Cre neurons, we used a Cre-dependent AAV tract-tracing strategy. Two viruses were coinjected, one expressing synaptophysin-EGFP (sypGFP), which prominently labels the presynaptic area of efferent fibers, and one expressing tdTomato, which labels cell bodies and fibers but is not specifically transported to synaptic areas. In each injected area, the majority of labeled neurons coexpressed both markers. In case IP-a1, AAV was injected into the IPR of a Chrna5^Cre mouse (Fig. 6A,B), with a small number of labeled neurons also noted in the adjacent median/paramedian raphe (MnR/PMnR; Fig. 6C,D). Prior work has suggested that SST-expressing GABAergic neurons in IPR make presynaptic connections onto afferents from the MHbV, which may modulate habenula input (Zhao-Shea et al., 2013). These SST neurons in IPR have been shown to be a subpopulation of the Chrna5-expressing neurons (Hsu et al., 2013). However, a synaptic connection by these neurons within the IP has not been directly demonstrated. In case IP-a1, labeled cell bodies in the IP were restricted to IPR, but labeled synaptic puncta were observed in IPC and in symmetrical bands running between IPI and IPL (Fig. 6A). The synaptic labeling was sometimes associated with ChAT labeling in MHbV afferents (Fig. 6E–G), although it is not possible to determine with certainty whether these synapses are associated with these afferents or with local IP neurons. Paradoxically, no tdTomato-labeled fibers of passage were observed between IPR and the ventral synaptic puncta, suggesting that these synapses may be derived from the small population of labeled PMnR neurons. To test this hypothesis, in case MnR-a1, we injected sypGFP+tdTomato expressing viruses into the PMnR of a Chrna5^Cre mouse caudal to the IP (Fig. 6H). Although fewer neurons were labeled compared with case IP-a1, dense synaptic labeling was observed in the ventral IP (Fig. 6I–K). To further characterize the raphe–IP projection, in case MnR-a2, we injected a GFP tract-tracing virus into the PMnR of an SST^Cre mouse (Fig. 6L). This also produced strongly labeled afferents to IPC and the border between IPI and IPL (Fig. 6M). DFISH for Chrna5 and SST mRNA showed that the SST-expressing neurons in IPR are a subset of the Chrna5-expressing neurons found there (Fig. 6N). However, in the MnR/PMnR, expression of these transcripts does not overlap (Fig. 6O). We conclude that IPR neurons do not project within the IP as previously hypothesized, but rather, the IP receives input from distinct populations of Chrna5- and SST-expressing neurons in the raphe.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Anterograde tracing of local projections of Chrna5^Cre neurons in the IP and Chrna5^Cre and SST^Cre neurons in the median raphe. A–K, Mixture of Cre-dependent (flex) AAVs encoding a red marker in cell bodies and axons (tdTomato) and a green marker targeted to presynaptic areas (sypGFP) was injected into the IPR (Case IP-a1; A–G) and MnR/PMnR (Case MnR-a1; H–K) of Chrna5^Cre mice. A, Injected area in the IPR (bregma 3.5). The sypGFP signal is overexposed to show faint synaptic labeling in the IPC and at the border between IPI and IPL. B, Confocal image of the injected area showing cellular colocalization of tdTomato and sypGFP. No synaptic labeling or projecting fibers are observed in the IPR beyond the injected area. C, D, A more caudal section (bregma −4.2) shows that virally labeled cell bodies are also found among Chrna5^Cre-expressing neurons in the PMnR caudal to the IP. 5HT neurons in the MnR are identified by immunostaining for Tph2. Overlay in C is from a standard atlas at bregma −4.24 (Paxinos and Franklin, 2001). E, Section adjacent to that shown in A immunostained for ChAT to show afferent fibers from the MHbV. F, G, Thick and thin confocal Z-stacks of sypGFP-labeled synapses in IPC. Arrows indicate synaptic labeling adjacent to cholinergic MHbV fibers. H, Case MnR-a1. Flex-sypGFP and flex-tdTomato viruses were injected in the PMnR of a Chrna5^Cre mouse. Overlay is at bregma −4.36. I, Labeled Chrna5^Cre PMnR neurons project to IPC, and the border between IPI and IPL. No labeled cell bodies are observed in IPR, confirming that the injection is localized to PMnR. J, Thin Z-stack confocal image of sypGFP-labeled synapses in IPC. Arrows indicate synaptic labeling adjacent to cholinergic MHbV fibers. K, Confocal image of sypGFP-labeled synapses at the border between IPI and IPL, which do not appear to overlap MHbV fibers. L, M, Case MnR-a2, viral tract-tracing of SST^Cre-expressing neurons in the PMnR. L, Site of injection of a Cre-dependent AAV encoding GFP in cell bodies and axons in the PMnR. Overlay is at bregma −4.24. M, Labeled fibers in the IPC and the border between IPI and IPL. No labeled cell bodies are observed in the IP, confirming that the injection is localized to PMnR. N, O, DFISH for Chrna5 and SST mRNA. Chrna5 and SST are coexpressed in IPR (bregma −3.7; N), but identify distinct, intermingled neurons in the PMnR (bregma −4.3; O). ATg, Anterior tegmental nucleus; MnR, median raphe nucleus; PMnR, paramedian raphe nucleus; Rbd, rhabdoid nucleus; mlf, medial longitudinal fasciculus; xscp, decussation of the superior cerebellar peduncle. Scale bars: A, E, H, M–O, 200 μm; L, 400 μm.

Prior work has shown that IP efferents project predominantly to the raphe and mesopontine tegmentum, with distinct projections for the IP subnuclei (Quina et al., 2017). Viral tract tracing of the efferents of Chrna5^Cre-expressing neurons in IPR shows synaptic labeling in a subset of these areas (Fig. 7A–I). In the median raphe, synaptic puncta from IPR projections generally appear in PMnR, surrounding but usually not overlapping 5HT neurons in MnR proper (Fig. 7A,B). Labeling in the rostral part of dorsal raphe was sparse (DR; Fig. 7C,D). Intense synaptic labeling was observed in the rhabdoid nucleus (Rbd; Fig. 7C,E), which can be identified by immunofluorescence for the orphan receptor Gpr151. The expression of this orphan receptor in Rbd can probably be attributed to inputs from Gpr151-expressing excitatory neurons in the LHb (Broms et al., 2015), showing an interesting confluence of inhibitory IP and excitatory LHb inputs to this nucleus. Projections from Chrna5^Cre IP neurons are sparse in the laterodorsal tegmental nucleus (LDTg) and are not closely associated with the cholinergic neurons there (Fig. 7F,G). Projections are absent from the central dorsal tegmental nucleus (DTg; Fig. 7H,I), although there is strong synaptic labeling in the surrounding the central pontine gray (CGPn) and near the midline, in the nucleus incertus (NI; Fig. 7I). Viral tract tracing of Chrna5^Cre neurons in MnR (case MnR-a1; Fig. 7J–L) shows that these neurons also form synaptic connections to the mesopontine raphe and dorsal tegmentum in a pattern that overlaps that of the IPR efferents but spares Rbd. The caudal projections of Chrna5^Cre IP neurons in IPR appear similar to IP projections described in the rat (Lima et al., 2017).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Efferents of Chrna5-expressing neurons to the pontine raphe and tegmentum. A–I, Caudal projections of IPR neurons labeled with Cre-dependent AAVs encoding a red marker in cell bodies and axons (tdTomato) and a green marker targeted to presynaptic areas (sypGFP), as shown in Figure 5A. A, B, Synaptic labeling in the MnR/PMnR (bregma 4.2). Serotonergic neurons are marked by immunofluorescent staining for Tph2. Confocal image in B shows that labeling surrounds but usually does not overlap Tph2-expressing cell bodies. C, D, Synaptic labeling in the rhabdoid nucleus and MnR/PMnR (bregma 4.5). Confocal view shows that fibers are very sparse in the DR. 5HT neurons are identified by immunostaining for Tph2. E, Coimmunostaining for Gpr151 and sypGFP in the rhabdoid nucleus. F, G, Synaptic labeling in the caudal MnR/PMnR, DR, and dorsal tegmentum (bregma 5.0). Cholinergic neurons are marked by immunofluorescent staining for ChAT. Sparse sypGFP labeling is noted in the LDTg, which does not overlap the ChAT-expressing neurons. H, I, Synaptic labeling in the pontine tegmentum (bregma 5.3). Synaptic labeling is prominent in DRI and NI, but absent from DTg and PnR. J–L, Caudal projections of Chrna5^Cre-expressing MnR neurons in case MnR-a1, labeled as shown in Figure 5H. J, Prominent synaptic labeling in DRI and in PMnR ipsilateral to the injection (bregma 4.8). K, Labeling in CGPn and DRI sparing DTg/LDTg (bregma 5.0, comparable to F). L, Projections to the caudal part of the pontine tegmentum, including NI (bregma 5.3, comparable to H). MnR, Median raphe nucleus; PMnR, paramedian raphe nucleus; Rbd, rhabdoid nucleus; mlf, medial longitudinal fasciculus; xscp, decussation of the superior cerebellar peduncle; DTg, dorsal tegmental nucleus; LDTg, laterodorsal tegmental nucleus; PnR, pontine raphe nucleus; NI, nucleus incertus; CGPn, central gray of the pons; DR, dorsal raphe nucleus; DRI, dorsal raphe nucleus, interfascicular part; DRD, dorsal raphe nucleus, dorsal part; DRVL, dorsal raphe nucleus, ventrolateral part; DRC, dorsal raphe nucleus, caudal part. Scale bar, 200 μm.

Activation of IP Chrna5 neurons is aversive

To test the behavioral function of the IP Chrna5 neurons, we generated an optogenetic model using Cre-inducible expression of ChR2. Chrna5^Cre mice were interbred with mice homozygous for a floxed-stop ChR2-EYFP cassette targeted to the Gt(ROSA)26Sor gene locus (Madisen et al., 2012). ChR2 expression was observed in the IP, but not in adjacent structures such as the VTA, which might confound behavioral experiments (Fig. 8A). IP^ChR2 neurons and fibers were found in IPR, IPC, and IPI, areas that receive glutamatergic/cholinergic afferents from the MHbV (ChAT label; Fig. 8B), but not in IPL, which receives glutamatergic/peptidergic inputs from the MHbD (SP label; Fig. 8C). Therefore, IP^ChR2 mice allow specific functional testing of the MHbV-IP pathway most often implicated in the effects of nicotine.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Optogenetic activation of Chrna5^Cre neurons using transgenic expression of ChR2. A, Expression of ChR2-EYFP in a Chrna5^Cre/Ai32 mouse (A1) and the same section also showing immunofluorescence for ChAT expressed in afferents from the MHbV (A2). B, An adjacent section stained for SP expressed in MHbD afferents to the IPL, which does not contain Chrna5^Cre-expressing neurons. C, Current-clamp recording of the optogenetic stimulation of a typical Chrna5^Cre-expressing neuron in IPR. The cell may be entrained to fire action potentials in response to light pulses (473 nm, 10 ms) delivered at 5 Hz, 10 Hz, and 20 Hz (C1–C3), but when stimulated at 50 Hz (C4) fires a single action potential, followed by excitatory block characterized by continuous depolarization and superimposed oscillations from ChR2 currents. D, Cell-attached recording of a less-frequently observed response by a different Chrna5^Cre-expressing IPR neuron, which could be initially entrained to light pulses delivered at 50 Hz, but after ∼1 s responded sporadically.

To assess the light responsiveness of IP^ChR2 neurons, we recorded cellular responses in IPR neurons in acute brain slice preparations (see Materials and Methods) using whole-cell current-clamp (n = 3) and cell-attached (n = 5) recordings. Of eight recorded cells, only two were spontaneously firing at <1 Hz; the others were silent in the absence of light stimulus. Light responsiveness was tested using 2 s trains of 10 ms light pulses at stimulus frequencies of 5, 10, 20, and 50 Hz. Eight of 8 neurons tested were able to fire action potentials at 1:1 correspondence with the 5 Hz and 10 Hz stimuli and 6/8 neurons were able to follow a 20 Hz stimulus (Fig. 8C1–C3). However, in the presence of a 50 Hz stimulus, some neurons produced an action potential in response to the first light pulse and then entered a depolarization block (5/8 cells; Fig. 8C4), whereas others briefly fired action potentials at a 1:1 correspondence, but were unable to do so for 2 s (2/8 cells; Fig. 8D). Therefore, for all behavioral experiments, a 20 Hz stimulus frequency was used except when assessing somatic nicotine withdrawal signs, for which a specific published stimulus protocol was followed (see Materials and Methods).

To assess the effect of IP stimulation on behavior, IP^ChR2 mice and controls were implanted with fiber-optic cannulas just dorsal to or slightly within IPR (see Materials and Methods; Fig. 9A, Fig. 9-1). Light stimulation from cannulas in this dorsal position resulted in the induction of cFos expression in neurons throughout the IP (Fig. 9-2). Because visible light and drug administration can in themselves affect behavior, all behavioral experiments were structured as a comparison between genotypes, in which IP^ChR2 mice and controls received identical light stimulation and/or pharmacological treatment. To assess for any effect of IP activity on acute locomotion, IP^ChR2 mice were tested in an open-field arena (Fig. 9B–E). During the stimulus phase of the trial, a small-magnitude increase in distance traveled during the light on intervals was noted, but there was no effect of genotype [repeated-measures (RM)-ANOVA, genotype × light, effect of genotype: F_(1,18) = 0.24, p = 0.63, effect of light: F_(1,18) = 13.89, p = 0.0015; Fig. 9B,C], nor were significant differences noted between genotypes in the three phases of the experiment (multiple-hypothesis-corrected t test; baseline t₍₅₄₎ = 1.10, p = 0.83, stimulation t₍₅₄₎ = 0.49, p > 0.99, poststimulation t₍₅₄₎ = 0.24, p > 0.99; Fig. 9D). In the open-field test, avoidance of the center of the enclosure has been used as a model of anxiety (Bailey and Crawley, 2009). A small-magnitude decrease in center occupancy was observed during the lights-on intervals, but there was no effect of genotype (RM-ANOVA: genotype × light, effect of genotype: F_(1,18) = 1.26, p = 0.28, effect of light: F_(1,18) = 9.468, p = 0.0065; Fig. 9E). Similarly, no difference between genotypes was detected in another assay of anxiety, the light/dark box (unpaired t test between genotypes: t₍₂₈₎ = 0.97, p = 0.34; Fig. 9F).

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Activation of Chrna5^Cre neurons does not affect locomotion, produce acute anxiety, or elicit somatic signs of nicotine withdrawal. A, Placement of fiber-optic cannulas in IP^ChR2 and control mice. Data for three behavioral cohorts of mice are shown. Blue circles represent fiber placement in control mice; red circles represent placement in IP ChR2 mice. Fiber termini are shown on images from a standard atlas closest to their rostrocaudal position (Paxinos and Franklin, 2001). In all cases, the optical fibers were intact and transmitted light efficiently when examined postmortem after the experimental protocol. A small number of cannulas were located outside the predicted area for effective illumination of IPR, such as the most dorsal cannula at bregma −3.8 and the most dorsal two cannulas at bregma −4.16, but the intensity of target illumination cannot be precisely determined. To avoid selection bias, all subjects were included in the statistical analysis except for two subjects in which the cannula placement obstructed the aqueduct and induced hydrocephalus (data not shown). The blue dashed line indicates the limit of 5mw/mm² irradiance, as measured from the point indicated by the yellow asterisk. B–E, Open-field behavior was assessed in IP^ChR2 and control mice over three 10 min intervals: baseline, stimulation, and poststimulation. The stimulation period consisted of 30 s periods of 20 Hz light pulses, alternating with 30 s periods of lights off. B, Distance traveled during 30 s intervals of pulsed light (+) and no light (−) during the stimulation period. Cohort of 12 IP^ChR2 and 8 control mice. C, Summary of distance traveled during light on time and light off time during the stimulation period; no effect of light or genotype was observed. D, Summary of distance traveled during the baseline, stimulation, and poststimulation periods. No effect of period or genotype was observed. E, Time spent in the center of the open-field enclosure during the stimulation period. No effect of light or genotype was observed. F, Time spent in the dark chamber of the light/dark box; no effect of light or genotype was observed. Cohort of 14 IP^ChR2 and 16 control mice. G, Somatic signs of nicotine withdrawal. Intermittent pulsed stimulation was delivered to the IP of nicotine-naive IP^ChR2 and control mice (see Materials and Methods) and mice were assessed for six somatic signs of nicotine withdrawal. No effect of genotype was observed. Cohort of 8 IP^ChR2 and 9 control mice. Cli, Caudal linear nucleus raphe; MnR, median raphe; PMnR, paramedian raphe; Pn, pons; PnO, pontine reticular nucleus, oral; PN, paranigral nucleus; RLi, rostral linear nucleus raphe; RRF, retrorubral field; SNc, substantia nigra, pars compacta; SNr, substantia nigra, pars reticulata. Scale bar, 0.5 mm. For extended data regarding cannula placement, see Figure 9-1. Pulsed light stimulation induced the expression of cFos in the IP in a pattern similar to that previously obtained with systemic nicotine activation (Ren and Sagar, 1992). For extended data on cFos induction by optogenetic stimulation, see Figure 9-2.

Prior work has reported that optogenetic stimulation of GABAergic neurons in the IP can elicit somatic signs associated with nicotine withdrawal, which include scratching, shaking, nodding, backing, rearing, and chewing (Zhao-Shea et al., 2013). To exactly match the stimulation protocol used in the prior work, we used intermittent optogenetic stimulation at 50 Hz in a standard housing cage (see Materials and Methods). IP^ChR2 mice showed no difference from controls in any of the somatic signs tested (t test between genotypes for each behavior, p > 0.99 for all seven behaviors; Fig. 9G). In the prior study of somatic signs, the IP of Gad2^Cre mice was injected with a conditional AAV vector expressing ChR2 in all GABAergic neurons in the injected area. The amount of injected virus (1 μl) was approximately twice the entire volume of the IP and is likely to have diffused into the surrounding tegmental nuclei, all of which contain GABAergic neurons, although the extent of viral spread was not discussed. Therefore, the prior result may represent off-target effects.

To determine whether the stimulation of α5-expressing IP neurons is reinforcing, aversive, or neutral, we used RTPP in a two-compartment shuttle box. To control for effects of preexisting preference or habituation, RTPP was performed with a 2 d reversal structure. On day 1, mice were assigned randomly to receive the light stimulation on the right or left side of the box, referred to here as the “A” side. On the subsequent day of testing, light stimulation was delivered on the opposite “B” side. No significant preference or avoidance was observed on day 1 of a 2 d RTPP experiment; however, on day 2, a significant avoidance of the stimulated side was observed in IP^ChR2 mice (RM-ANOVA: genotype × day, effect of genotype: F_(1,18) = 14.71, p = 0.0012, effect of day: F_(1,18) = 0.83, p = 0.38; Fig. 10A). Division of the test period into 5 min intervals for IP^ChR2 mice on the two test days revealed an incipient trend toward avoidance of the stimulated side on day 1, which became a significant difference between the start and end of the trial on day 2 (RM-ANOVA, day × epoch, effect of day: F_(2,33) = 5.71, p = 0.0074, effect of day: F_(1,33) = 8.67, p = 0.0059; Fig. 10B). In principle, the enhancement of avoidance seen on day 2 could either be a learning effect, in which a specific side preference is retained from day 1, or a “priming” effect, in which recent prior stimulation sensitizes the real-time avoidant response. However, given the reversal of the stimulated side between days, a choice to avoid the stimulated side on day 2 would result from a learned preference for the chamber stimulated on day 1. This seems unlikely based on the trend to develop avoidance of the stimulated side over the course of the test on both days. To test for the retention of a learned preference, we performed a 2 day experiment in which one side was stimulated on day 1 and the animals were returned to the shuttle box on day 2 without a light stimulus. No retained side preference was exhibited in the absence of the stimulus (RM-ANOVA, genotype × day, effect of genotype: F_(1,24) = 2.13, p = 0.16, effect of day: F_(1,24) = 2.00, p = 0.17; Fig. 10C). In a second cohort of mice, we tested the reversibility of the avoidance effect. A trend toward avoidance in IP^ChR2 mice on day 2 disappeared by day 10, demonstrating the transient nature of the priming effect (RM-ANOVA, day × genotype, effect of genotype: F_(1,15) = 3.774, p = 0.07, effect of day: F_(2,30) = 0.62, p = 0.55; Fig. 10D). Although the trend toward an effect of genotype did not reach statistical significance in this experiment, analysis of day 1 and day 2 data for the combined cohorts showed a clear aversive effect in IP^ChR2 mice on day 2 (RM-ANOVA, day × genotype, effect of genotype: F_(1,35) = 24.86, p < 0.0001, effect of day: F_(1,35) = 0.46, p = 0.50; Fig. 10E,F). Finally, to eliminate any positive or negative learned effect of side in the primed aversion effect, we provided the initial stimulation in a standard housing cage. Prior stimulation in a separate cage was sufficient to prime the aversive effect (t test between genotypes: t₍₄₁₎ = 2.47, p = 0.0176; Fig. 10G).

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

IP stimulation in IP^ChR2 mice produces aversion when primed by a recent prior stimulus. Optogenetic stimulation of the IP was performed in a two-compartment shuttle box to elicit RTPP. Video tracking was used to detect the location of the mouse during each 15 min trial and the presence of the animal in one side of the box (“stimulated side”) triggered pulses from a 473 nm laser at 20 Hz. The stimulated side was reversed in each trial and is indicated by blue shading. A, A 2 d reversal trial shows no significant effect of IP stimulation on day 1, but significant aversion in IP^ChR2 mice on day 2 relative to controls. Note that the control animals have a slight, persisting preference for side B, shown as a negative preference for the stimulated side on day 1 and a positive preference for the stimulated side on day 2. In contrast, IP^ChR2 animals show a shift to specific avoidance of the stimulated side on day 2. ****Post hoc test: multiple-hypothesis-corrected t test of day 2: t₍₃₆₎ = 5.21, p < 0.0001. Cohort of 12 IP^ChR2 and 8 control mice. B, Time course of the development of aversion in IP^ChR2 mice. On day 2 a significant difference is observed between the initial and final 5 min epoch. *Post hoc test: multiple-hypothesis-corrected t test t₆₆ = 2.807, p = 0.0197. Cohort of 12 IP^ChR2 mice from (A). C, Two-day RTPP trial conducted without stimulation on the second day. No side preference is retained from the first day of the trial. Cohort of 11 IP^ChR2 and 15 control mice. D, Reversal of primed aversion. In an extended experiment with three RTPP trials, a trend toward aversion appeared on day 2, but disappeared on day 10 after 8 d of no stimulation. Cohort of 7 IP^ChR2 and 10 control mice. E, Combined data for 2 d side reversal RTPP experiments. RTPP across the 2 d trial shows development of side aversion in IP^ChR2 mice on day 2. The cohort of mice used in A plus the cohort used in C were used for the combined analysis, n = 19 IP^ChR2 mice and 18 controls. **Post hoc test: t test between genotypes for day 2: t₇₀ = 3.43, p = 0.002. F, Time course of the acquisition of place aversion for IP^ChR2 mice in the combined cohort in 5 min epochs. G, Primed aversion after home cage stimulation. On day 1, intermittent light pulses were delivered for 15 min in a standard housing cage in a pseudorandom pattern (see Materials and Methods). RTPP was assessed on day 2 and significant aversion was observed in IP^ChR2 mice. *Main effect of genotype: t₍₄₁₎ = 2.473, p = 0.0176. Combined cohort of 18 IP^ChR2 and 25 control mice. H, IP^ChR2 mice do not respond to ICSS training. Light pulses were delivered in response to turning the active response wheel in a 2-wheel choice paradigm in an 8 d experiment with reversal of the active response wheel on day 5. No increase in response rate was observed over time and mice did not develop a preference for the active wheel on day 4 or day 8 of training, when a maximum preference would be expected (RM-ANOVA, day × wheel for day 4 and 8: F_(1,8) = 1.45, p = 0.26, effect of wheel: F_(1,8) = 0.02, p = 0.90). Wheel turning resulted in a low rate of stimulation that did not change over the course of the experiment. One stimulation (train of light pulses) was delivered per two wheel turn events observed within a specific time interval. The number of stimulations delivered is less than one-half the number of active wheel turns because isolated single wheel turn events were not used to trigger the stimulus (see Materials and Methods).

In prior studies, we used an optogenetic ICSS protocol to demonstrate that activation of the dorsal subnucleus of the medial habenula is reinforcing (Hsu et al., 2014, 2016). In this paradigm, mice are given access to paired response wheels, one of which triggers a laser stimulus to the area of interest. After 4 d of training, the active wheel is switched for four reversal trials. Increased wheel turns, preference for the active wheel after 1–2 d of training, and reversal of the preference with change of the active wheel are indicative of an ICSS effect. As expected given the aversive nature of IP stimulation in the shuttle box test, IP^ChR2 mice did not show a preference for the stimulus-associated wheel in this paradigm on day 4 or day 8, when a maximum response would be seen (Fig. 10H). The observed infrequent, sporadic wheel-turning behavior resulted in a very low frequency of light stimulation, which did not demonstrate reinforcement and could not be shown to be aversive because of the low rate of stimulation at baseline.

α5-expressing IP neurons receive input from cholinergic/glutamatergic neurons of the MHbV. The role of cholinergic transmission in this system has not been fully defined, but acetylcholine released at the MHbV-IP interface has the potential to both directly activate IP nicotinic receptors and to act presynaptically to increase glutamate release (Ren et al., 2011; Hsu et al., 2013; Frahm et al., 2015). This suggests that, like direct stimulation of the IP, exposure to nicotine could prime the aversive effect of IP activation in RTPP. Nicotine-primed RTPP was performed with a 2 day reversal design. The initial trial and reversal trial were separated by 10 day to avoid the effect of light stimulation priming, which is lost after this amount of time (Fig. 10D). On day 1, IP^ChR2 and control mice were given a subcutaneous injection of nicotine (0.3 mg/kg, see Materials and Methods) and placed a standard housing cage. The chosen dose of nicotine is by itself reinforcing or neutral, not aversive, in place preference assays (Neugebauer et al., 2011; Grieder et al., 2017). Acute locomotion, measured from 1 to 16 min after nicotine administration on day 1, was reduced compared with that observed in a prior trial in which nicotine was not delivered (RM-ANOVA, drug × genotype, effect of genotype: F_(1,24) = 4.71, p = 0.04, effect of drug: F_(1,24) = 46.01, p < 0.0001; Fig. 11A). RTPP was measured 2 h after nicotine administration. The rapid metabolism of nicotine in mice, with a half-life of <10 min (Matta et al., 2007), predicts that the residual blood level from this single dose of nicotine was low at the time of RTPP and is consistent with prior results (Bailey et al., 2010). The acute inhibition of locomotion after nicotine administration had resolved by the time RTPP was conducted. IP^ChR2 mice exposed subacutely to nicotine in this way showed a marked aversion to IP stimulation (RM-ANOVA, day × genotype, effect of day: F_(1,21) = 0.70, p = 0.42; effect of genotype: F_(1,21) = 7.856, p = 0.0107; Fig. 11B–D) with a strong reversal of side preference between day 1 and day 10 (Fig. 11C). In experiments with activity-primed RTPP, the aversive response increased over the 15 min trial of IP stimulation (RM-ANOVA, day × 5 min epoch, effect of day: F_(2,54) = 5.184, p = 0.0087; effect of epoch: F_(1,54) = 5.963, p = 0.0179; Fig. 10B), whereas in nicotine-primed RTPP, strong aversion was observed in IP^ChR2 mice in the first 5 min period of the trial on day 1 (Fig. 11E). We also noted that, after nicotine administration, control mice showed slight aversion to IP stimulation, spending ∼40% fractional time in the laser-paired compartment. This is consistent with the small effect of light seen in control animals in open-field locomotion and was not sufficient to mask the very strong genotype-dependent aversion observed in IP^ChR2 mice.

• Download figure
• Open in new tab
• Download powerpoint

Figure 11.

Subacute exposure to a single dose of nicotine primes aversion to IP stimulation in IP^ChR2 mice. Nicotine primed aversion was tested in RTPP on day 1 of the experiment and the light-simulated side of the test compartment was reversed on day 10. A, Home cage locomotion is suppressed by acute administration of nicotine independent of genotype. IP^ChR2 and control mice were given subcutaneous nicotine injections (0.3 mg/kg) and returned to the home cage for activity monitoring from 1 to 16 min after administration on day 1. Baseline activity was measured in the same cohort of mice in a prior trial of home cage locomotion without nicotine. ****Main effect of drug, F_(1,24) = 46.01, p < 0.0001. Cohort of 11 IP^ChR2 and 15 control mice. B–E, Results of a two-session nicotine-primed RTPP experiment with side reversal on day 10 in a cohort of 9 IP^ChR2 and 14 control mice. B, IP^ChR2 mice treated with nicotine show aversion to IP stimulation on day 1. *Post hoc test: multiple-hypothesis-corrected t test of day 1 effect of genotype: t₍₄₂₎ = 2.325, p = 0.05. C, Comparison of day 1 and day 10 of the reversal experiment shows that IP^ChR2 mice change side preference consistent with nicotine-primed aversion for the light stimulus. ****Post hoc test: multiple-hypothesis-corrected t test of Cre effect of day: t₍₂₁₎ = 6.185, p < 0.0001. D, Examples of the pattern of shuttle box activity in control and ChR2 mice selected from the day 1 results in B. The side paired with light stimulation is shaded in blue. E, Rapid development of nicotine-primed aversion in the day 1 trial shown in B. A significant difference is observed between genotypes in the initial 5 min epoch. *Post hoc test: multiple-hypothesis-corrected t test between genotypes for first 5 min epoch: t₍₆₃₎ = 2.54, p = 0.04.