5-HT_1B autoreceptors differentially modulate the expression of conditioned fear in a circuit-specific manner

Abstract

Located in the nerve terminals of serotonergic neurons, 5-HT_1B autoreceptors are poised to modulate synaptic 5-HT levels with precise temporal and spatial control, and play an important role in various emotional behaviors. This study characterized two novel, complementary viral vector strategies to investigate the contribution of 5-HT_1B autoreceptors to fear expression, displayed as freezing, during contextual fear conditioning. Increased expression of 5-HT_1B autoreceptors throughout the brain significantly decreased fear expression in both wild type (WT) and 5-HT_1B knockout (1BKO) mice when receptor levels were increased with cell type-specific herpes simplex virus (HSV) vector injected into the dorsal raphe nucleus (DRN). Additional studies used an intersectional viral vector strategy, in which an adeno-associated virus containing a double-floxed inverted sequence for the 5-HT_1B receptor (AAV-DIO-1B) was combined with the retrogradely transported virus canine adenovirus-2 expressing Cre (CAV-Cre) in order to increase 5-HT_1B autoreceptor expression only in neurons projecting from the DRN to the amygdala. Surprisingly, selective expression of 5-HT_1B autoreceptors in just this circuit led to an increase in fear expression in WT, but not 1BKO, mice. These results suggest that activation of 5-HT_1B autoreceptors throughout the brain may have an overall effect of attenuating fear expression, but activation of subsets of 5-HT_1B autoreceptors in particular brain regions, reflecting distinct projections of serotonergic neurons from the DRN, may have disparate contributions to the ultimate response.

Serotonin (5-hydroxytryptamine, 5-HT) is a monoamine neurotransmitter that is involved in a wide variety of emotional and cognitive behaviors; dysfunction of the serotonergic system is implicated in psychiatric illnesses such as depression, post-traumatic stress disorder, and anxiety. Serotonergic neurons in the brain have cell bodies that reside in the raphe nuclei of the midbrain and brainstem but project extensively to nearly every brain region (Jacobs and Azmitia, 1992). The dorsal raphe nucleus (DRN) provides most of the serotonin fibers to the forebrain and is somatotopically organized; subpopulations of serotonergic neurons that have unique anatomical connections may also impact different elements of complex emotional behavior (Lowry et al., 2008, Gaspar and Lillesaar, 2012).

5-HT_1B receptors are metabotropic Gα_i/o receptors (Bouhelal et al., 1988, Schoeffter and Hoyer, 1989) that are primarily localized to nerve terminals (Boschert et al., 1994, Ghavami et al., 1999, Riad et al., 2000) and exist both as autoreceptors on serotonergic neurons and as heteroreceptors on nonserotonergic cells (Hen, 1992). Since 5-HT_1B autoreceptors are more strongly activated during bouts of intense serotonergic activity at the sites of release, they provide localized autoregulation of serotonin neurotransmission (Sari, 2004). Several lines of evidence suggest that the level of 5-HT_1B autoreceptor expression is a key determinant of stress reactivity (Neumaier et al., 2002a, Kaiyala et al., 2003), and several studies suggest that drugs targeting the 5-HT_1B receptor have promise as adjunctive therapy with SSRIs (Roberts et al., 1999, Gardier, 2009, Ruf and Bhagwagar, 2009).

Differentiating between 5-HT_1B autoreceptors and 5-HT_1B heteroreceptors is crucial in revealing their individual roles in brain circuits and behavior but poses a technical challenge. 5-HT_1B autoreceptors and heteroreceptors are identical in sequence and structure but are expressed in different neuron types. 5-HT_1B autoreceptors are diffusely distributed at nerve terminals throughout the entire brain, but are intermixed with 5-HT_1B heteroreceptors in most brain regions. Therefore, it is difficult to target 5-HT_1B autoreceptors exclusively using conventional methods like drug administration, as 5-HT_1B heteroreceptors will also be affected. The 5-HT_1B knockout mice currently available for research have a constitutive deletion of the gene, and while they have a remarkable phenotype with decreased anxiety and increased aggression (Gingrich and Hen, 2001, Groenink et al., 2003, Guilloux et al., 2011), it is unclear whether these behavioral effects relate to the loss of 5-HT_1B autoreceptors or heteroreceptors during early development or adulthood. One approach for manipulating 5-HT_1B autoreceptor function is via systemic injections of the brain-penetrant agonist CP-94,253 at low doses, which preferentially activates autoreceptors over heteroreceptors (Sarhan et al., 2000, McDevitt et al., 2011); however, it is possible that heteroreceptors may also be affected even at low doses, tending to confound any conclusions based solely on pharmacological experiments. An alternative technique utilizes viral-mediated gene transfer (Clark et al., 2002, Clark et al., 2004, McDevitt et al., 2011, Hagan et al., 2012) to target neurons residing in a particular brain region.

Novel recombinant viral vectors are continuously developed to achieve greater control over transgene expression levels and cell type-specificity (Lowenstein and Castro, 2002, Luo et al., 2008, Bouard et al., 2009, Papale et al., 2009, Zhang et al., 2010). Herpes simplex virus (HSV) has been commonly used for local manipulation of expression in specific brain targets and is advantageous in offering neuron-specific gene expression with a low immune and inflammatory response (Papale et al., 2009). In recent years, AAV viruses with a double-floxed inverted open reading frame (DIO) have gained popularity. In these constructs, the transgene is inserted in the antisense orientation and flanked by two sets of lox sites; when Cre recombinase is present, an inversion occurs at the lox sites to flip and lock the transgene into the sense orientation (Atasoy et al., 2008). AAV-DIO viruses may be used in combination with transgenic animals expressing Cre in a cell type-specific manner or with retrogradely transported, Cre-expressing canine adenovirus (CAV-Cre) to achieve greater specificity of transgene expression (Schnutgen et al., 2003, Saunders et al., 2012, Nair et al., 2013).

Previous work in our lab has used the neuron-specific HSV to increase 5-HT_1B receptor expression in the rat DRN and demonstrated that viral-mediated 5-HT_1B receptor expression localized correctly to axon terminals throughout the brain and increased autoreceptor activity (Clark et al., 2002, Clark et al., 2004, Hagan et al., 2012). This technique has permitted in vivo studies to elucidate the specific contribution of 5-HT_1B autoreceptors to emotional behaviors including fear, anxiety, depression, and stress. In rats, increasing expression of 5-HT_1B receptors in neurons in mid-rostrocaudal DRN decreased anxiety in the open field test, as well as fear potentiation of the startle response (Clark et al., 2002, Clark et al., 2004), but when the caudal DRN was targeted instead, expression of transgenic 5-HT_1B receptors had no effect on anxiety (McDevitt et al., 2011). Additionally, overexpression of 5-HT_1B receptors decreased fear expression in contextual fear conditioning but had no effect on acquisition; similarly, context-paired administration of a low dose of CP-94,253 to preferentially activate 5-HT_1B autoreceptors showed the same decrease in fear. Taken together, these data suggest a protective role of 5-HT_1B autoreceptors in these emotional behaviors, with differences along the rostrocaudal axis. Interestingly, these anxiolytic and fear-attenuating properties of 5-HT_1B autoreceptors are abolished when animals were exposed to stress or received the 5-HT_1B antagonist SB224289 prior to behavioral testing (Clark et al., 2002, Clark et al., 2004, McDevitt and Neumaier, 2011).

While it has been suggested that conditioned aversive stimuli activate the serotonergic system, which, in turn, projects to a variety of brain regions to elicit different behavioral responses (Deakin and Graeff, 1991), the specific circuits involved are still unknown. The amygdala is a brain area that receives strong serotonergic projections and is heavily implicated in fear and anxiety behaviors (Davis, 1992, Ebner et al., 2004, Lowry et al., 2005, Ciocchi et al., 2010, Haubensak et al., 2010, Herry et al., 2010, Johansen et al., 2010). Although approximately 10% of serotonergic neurons in the DRN project to the amygdala, nearly all of the neurons that project from the DRN to the amygdala are serotonergic (Ma et al., 1991); within the amygdala, serotonergic innervation is densest in the basolateral amygdala and more diffuse in the central amygdala (Vertes, 1991). However, it is still unclear how the serotonergic system acts in the amygdala in fear-related responses. One hypothesis proposed that the DRN-to-amygdala circuit is important for eliciting fear and anticipatory anxiety, while the DRN-to-periaqueductal gray circuit is important for preventing inappropriate freezing or fight-or-flight responses (Deakin and Graeff, 1991, Graeff et al., 1996). Because 5-HT_1B autoreceptors modulate serotonergic neurotransmission locally at the site of release, it is important to understand their function in specific circuits, as this may lead to greater insight about the role of serotonin in various brain regions.

To examine the selective contribution of 5-HT_1B autoreceptors to emotional behaviors, we designed and utilized two novel viral vectors. First, to overexpress 5-HT_1B autoreceptors in nerve terminals throughout the brain, we employed an HSV viral vector using the SERT promoter to induce transgene expression only in serotonergic cells. Second, to overexpress 5-HT_1B autoreceptors in nerve terminals in one particular circuit, we used a conditional AAV-DIO viral vector in combination with CAV-Cre viruses or cell type-specific Cre driver mouse lines, such as Pet1-Cre mice that use the serotonergic marker Pet-1 to selectively drive Cre expression. We harnessed these complementary techniques to further elucidate the role of 5-HT_1B autoreceptors in contextual fear conditioning. We hypothesized that 5-HT_1B autoreceptors attenuate fear expression but that their contribution to contextual fear conditioning is dependent on the specific circuit in which the 5-HT_1B autoreceptors are expressed.

EXPERIMENTAL PROCEDURES

Vector construction

To create a serotonin-specific viral gene expression system, we utilized a plasmid containing the 1.7 kb fragment of the human SERT promoter, which was received as a generous gift from Ovie Wiborg. In rat raphe precursor cells, the 1250 bp proximal to the SERT gene shows the greatest promoter/enhancer activity (Mortensen et al., 1999), so this section, along with GFP and the SV40 polyadenylation sequence, was cloned using PCR. Two SV40 polyA fragments were ligated together, and the ligation reaction was subjected to PCR to form a double polyA site. All PCR fragments were subcloned into the TOPO-Blunt cloning vector. All three components were inserted into the a pHSV-prPUC plasmid, developed by Dr. Rachael Neve for use in a replication-defective HSV gene transfer system (Neve and Geller, 1995), to form an expression cassette. This cassette was cloned in the opposite orientation relative to the HSV promoter to avoid nonspecific induction of transgene transcription in nonserotonergic cells, and was tested in cell culture using the serotonergic-like CA-77 cell line (Greene and Tischler, 1976) and the noradrenergic-like PC-12 cell line (Clark et al., 1995). Virus produced from this construct (pSERT-GFP) was injected into rat striatum, ventral tegmental area, and dorsal raphe nucleus and was observed to express selectively in DRN (Sexton et al., 2006).

To create the AAV-DIO construct that conditionally expresses the 5-HT_1B receptor (AAV-DIO-1B), we constructed a DNA block using annealed synthetic oligonucleotides that was comprised of GSG, 2A skip, and a hemagglutinin (HA) tag with inclusive restriction sites. We then inserted mCitrine before the GSG sequence, and the coding sequence for the “humanized” rat 5-HT_1B (N357T, which confers low affinity for ß-antagonists (Adham et al., 1994)) without the Kozak sequence immediately after the HA tag; thus, the final construct was mCitrine-GSG-2A-HA-5-HT_1B. The 2A skip was modeled on the porcine teschovirus-1 (Szymczak et al., 2004). We utilized the backbone of pAAV-FLEX-hM3D-mCherry (Krashes et al., 2011) to gain the AAV origins of replication (LITR/RITR), FLEX double lox sites (Lee and Saito, 1998), woodchuck hepatitis virus posttranscriptional response element (WPRE) (Loeb et al., 1999, Loeb et al., 2002), and antibiotic selectivity. Briefly, the pAAV-FLEX-hM3D-mCherry was cut with AscI and NheI to remove the hM3D-mCherry portion and insert mCitrine-GSG-2A-HA-5-HT_1B in the reverse orientation. The final construct was packaged into AAV serotype 8 by the Vector Core at the University of North Carolina at Chapel Hill. The original CAV viral vector was a generous gift from from Dr. Eric Kremer (Kremer et al., 2000) and was propagated and packaged by our lab.

The novel AAV-DIO-1B viral vector incorporated two important elements. First, because the double-floxed and inverted ORF design requires the presence of Cre recombinase to induce expression, the selection of the source of Cre gives an additional level of control; DIO viral vectors may be used in combination with transgenic mice that express Cre in a cell type-specific manner (Krashes et al., 2011), or with retrogradely transported CAV-Cre viral vectors, which may be used to study a single circuit in the brain (Nair et al., 2013, Boender et al., 2014). Second, the 2A ribosomal skip peptide allows the expression of two separate proteins from a single cassette, which offers the advantage of a small cassette size but avoids any issues associated with fusion proteins altering protein function.

Animal care

All animal procedures were approved by the University of Washington’s Institutional Animal Care and Use Committee and carried out in accordance with National Institutes of Health guidelines. Care was taken to minimize animal discomfort. Adult male and female ePet1-Cre (generously provided by Dr. Evan Deneris (Scott et al., 2005)), 129S6/SvEvTac, and C57BL/6J mice were housed two to five per cage with access to food and water available ad libitum. 129S6/SvEvTac mice were either wild type (WT) or had constitutive deletions of the 5-HT_1B receptor (1BKO) (generously provided by Drs. Rene Hen and Suzanne Ahmari (Malleret et al., 1999)). Mice were housed on a 12/12 hour light cycle with lights on at 0600. All experiments were performed during the light cycle. Mice were 12-24 weeks of age and weighed 15-26 g at the time of surgery. Mice used in behavioral experiments were no more than 16 weeks of age at the time of testing.

Stereotaxic surgery

Mice were anesthetized with isoflurane (1-2% in oxygen) or a ketamine/xylazine cocktail (100 mg/kg ketamine and 10 mg/kg xylazine, i.p.), the scalp was shaved and cleaned with betadine and ethanol, and the mice were placed in a Stoelting stereotaxic device. The scalp was incised to reveal the skull, and small holes were drilled at the sites of injection. A 27 gauge needle was slowly inserted, and virus or fluorescent microspheres were injected using a microprocessor-controlled pump. The needle was left in position for five minutes following the completion of the injection, and then slowly withdrawn. The scalp was closed with sutures and Vetbond (3M), and mice were placed in a recovery chamber warmed by a heating pad until they recovered and moved about freely before returning to their home cage. Analgesia was provided using meloxicam (0.4 mg/kg) or ketoprofen (5 mg/kg) administered subcutaneously during surgery.

Injection of fluorescent microspheres

Fluorescent microspheres (Molecular Probes) were diluted 1:1 in sterile water and injected bilaterally, 250 nL per side at a rate of 100 nL/sec, to target the amygdala using four test coordinates: 1) AP −1.1, ML ±2.8, DV −4.7; 2) AP −1.25, ML ±2.8, DV −4.6; 3) AP −1.4, ML ±2.8, DV −4.8; or 4) AP −1.55, ML ±2.8, DV −4.6. Animals were sacrificed 5 days after the injections and perfused with 1× PBS followed by 4% paraformaldehyde. Brains were post-fixed in 4% paraformaldehyde for four hours before switching to 1× PBS, sliced, and mounted onto slides with Prolong Gold with DAPI (Molecular Probes).

Injection of viral vectors

The DRN was targeted at an angle of 28° from the vertical with the coordinates AP −4.8, ML +1.5, DV −3.4 from bregma. Bilateral injections of CAV-Cre or CAV-GFP into the amygdala were performed at the coordinates AP −1.25, ML ±2.8, DV −4.6 from bregma. Each site received 1 μL of virus injected at a rate of 200 nL/min.

Immunohistochemistry

Animals were sedated with Beuthanasia-D (585 mg/kg pentobarbital sodium and 75 mg/kg phenytoin sodium), then intracardially perfused with 1× PBS followed by 4% paraformaldehyde before removing the brains, and the brains were post-fixed in 4% paraformaldehyde for four hours before switching to 1× PBS, then sliced with a vibratome into 40 μm sections. Sections were washed once for ten minutes in 1× PBS/0.5% Triton X-100, then blocked for two hours with 5% normal goat serum (NGS)/1× PBS/0.25% Triton at room temperature with gentle shaking. Primary antibodies were diluted in 2.5% NGS/1× PBS/0.25% Triton X-100 and incubated with brain slices overnight at 4°C with gentle shaking. The primary antibody dilutions used were: rabbit anti-GFP 1:400 (Cell Signaling), rabbit anti-HA 1:400 for single-labeling experiments and 1:1000 for double-labeling experiments (Cell Signaling), and sheep anti-tryptophan hydroxylase (Tph) 1:400 (Millipore). Sections were washed five times with 1× PBS before incubation with secondary antibodies diluted in 2.5% NGS/1× PBS for two hours at room temperature with gentle shaking. The secondary antibody dilutions used were goat anti-rabbit and goat anti-sheep at 1:400 dilutions (Molecular Probes). Brain sections were washed two more times before mounting onto slides with Prolong Gold with DAPI (Molecular Probes).

Contextual fear conditioning

Behavior was performed at an interval after viral vector infusion that allowed stabilized transgene expression (discussed later); this was after seven days for SERT and after 20 days for the Cre-dependent DIO vector system. Training and testing occurred in a mouse test cage measuring 7 × 7 × 12 inches with a shock floor (Coulbourn Instruments) with a 1% acetic acid odor. Scrambled shocks were delivered manually with the Precision Animal Shocker (Coulbourn Instruments). For fear training, mice were placed in the middle of the test cage and allowed to acclimate for two minutes, then received three 2-second, 0.7 mA footshocks at two minute intervals. Following the third and final footshock, mice remained in the test cage for an additional minute before being returned to their home cage. After 24 hours, mice were placed back in the test cage for a five minute test session. Training and testing sessions were digitally recorded, and freezing was analyzed using ANY-maze software (Stoelting Co.).

Data analysis

Statistical analyses were performed with GraphPad Prism. Fear expression was measured as percent of time freezing during the five minute test session. Freezing over the course of the training session was quantified in 60 second bins. Pain sensitivity was measured as the latency from the onset of the first shock to the first episode of freezing. Two-way ANOVA was used to test for statistical significance for behavior, with virus and genotype as the between subjects variables, except for the analysis of freezing during the training session, which used two-way repeated measures ANOVA, with experimental group as the between subjects variable and time as the within subjects variable.

RESULTS

1.1 Viral mediated expression of 5-HT_1B autoreceptors

The first set of novel viral vectors was generated to incorporate the known sequence for the human SERT promoter to drive transgene expression (Figure 1). Using a similar strategy that we previously used to target the direct and indirect pathway neurons of striatum (Ferguson et al., 2011), we cloned the human SERT promoter and either eGFP (SERT-GFP) or hemagglutinin (HA)-tagged 5-HT_1B receptor (SERT-1B) into the HSV amplicon. Since the HSV IE4/5 promoter/enhancer that is intrinsic to the standard HSV amplicon is very strong, we cloned the SERT-GFP or SERT-HA1B cassette in the reverse orientation relative to the HSV promoter.

Figure 1. HSV viral vectors utilize the SERT promoter to drive transgene expression.

Plasmid maps for the SERT-GFP (left) and SERT-1B (right) vectors are shown. The human SERT promoter sequence is located directly upstream of the transgene to be expressed. HSV viral elements are the origin of DNA replication (oriS), the constitutive IE4/5 promoter, and the HSV packaging signal.

1.2 SERT promoter selectively drives transgene expression in serotonergic cells

To test the expression time course and cell type-specificity of the SERT promoter-driven viral vectors, SERT-GFP was injected into the DRN of WT mice. In contrast with conventional HSV vectors that peak around day 3 and dissipate by about day 7 (Carlezon et al., 2000, Pliakas et al., 2001, Clark et al., 2002, Ferguson et al., 2009, McDevitt et al., 2011), immunohistochemistry revealed low GFP expression at day 4, and peak expression at day 7 that persisted until at least day 14 (Figure 2a). This time course is delayed and longer lasting than that associated with HSV IE4/5 (Neve et al., 2005, Ferguson et al., 2009) but is similar to what we observed when using preprodynophin or preproenkephalin promoters in a similar constellation (Ferguson et al., 2011). Immunoreactivity for the serotonergic marker Tph colocalized with GFP, demonstrating that transgene expression was restricted to serotonergic neurons (Figure 2b). A similar pattern of expression was observed in rat brain (data not shown).

Figure 2. HSV viral vectors utilize the SERT promoter to express transgenes selectively in serotonergic neurons, with expression peaking at day 7.

SERT-GFP was injected into the DRN of WT mice. (a) Viral expression was assessed at days 4, 7, and 14, with peak expression at day 7. (b) Viral GFP expression colocalized with Tph immunoreactivity, demonstrating the selectivity of using the SERT promoter.

2.1 Cre-dependent 5-HT_1B receptor expression using viral mediated gene transfer

As shown in Figure 3, the novel AAV-DIO-1B viral vector drives the expression of two distinct proteins: mCitrine, a fluorescent protein allowing for easy detection of viral infection, and HA-tagged 5-HT_1B receptor. This bicistronic construct takes advantage of the 2A ribosomal skip peptide, which allows the expression of two proteins encoded by a single messenger RNA (Donnelly et al., 2001, Luke et al., 2008, Ibrahimi et al., 2009, Sharma et al., 2012). This open reading frame (ORF) is inserted in an inverted orientation with respect to the rest of the plasmid, and is flanked by loxP and lox2272 sites as shown. In the presence of Cre recombinase, the ORF inverts at the lox sites to allow transcription of the transgene sequences. The transgene is under the control of the human synapsin promoter, which restricts expression to neurons (Thiel et al., 1991, Kugler et al., 2003). Several AAV serotypes are readily available; we packaged these constructs into AAV8 as they infect DRN neurons well and support strong transgene expression (Klein et al., 2006).

Figure 3. AAV viral vector utilizes a DIO design and the ribosomal 2A skip to express mCitrine and HA-5-HT_1B in the presence of Cre recombinase.

Plasmid map for the AAV-DIO-1B vector is shown. The ORF, encoding mCitrine and an HA-tagged 5-HT_1B receptor separated by a 2A ribosomal skip peptide, is in an inverted orientation flanked by two sets of lox sites. Cre recombinase flips the ORF into the correct orientation, with transcription driven by the human synapsin (hSyn) promoter, restricting expression of the transgenes to neurons.

2.2 AAV-DIO-1B viral vector expresses mCitrine and HA-tagged 5-HT_1B receptors in the presence of Cre recombinase

Hemizygous Pet1-Cre mice received injections of the AAV-DIO-1B viral vector into the DRN. Robust viral expression was observed in the DRN 24 days following surgery (Figure 4). As shown by costaining for HA and Tph, expression of HA-5-HT_1B was limited to serotonergic neurons expressing Cre recombinase under the control of the Pet-1 promoter, demonstrating the integrity of the DIO viral construct (Figure 4a). Injection of 1 μL of AAV-DIO-1B was sufficient for robust expression (Figure 4b). Viral-mediated HA-5-HT_1B receptor expression correctly localized to axon terminals, as evidenced by HA immunoreactivity in beaded fibers that are characteristic of collateral axons in DRN (Figure 4c). In the DRN, mCitrine was visible readily without immunohistochemistry, and costaining with HA showed colocalization of the two viral protein products in the same cells (Figure 4d).

Figure 4. Injections of AAV-DIO-1B into Pet1-Cre mice produce robust mCitrine and HA-5-HT_1B expression in the DRN.

Viral expression was observed in the DRN 24 days following surgery. (a) Dual labeling of HA-5-HT_1B (green) and Tph (red) showed colocalization only in serotonergic neurons, demonstrating that the DIO viral vector was only expressed in the presence of Cre recombinase. Staining for HA-5-HT_1B was prominent throughout the DRN (b), in both cell bodies as well as in axon collaterals (c). (d) Both mCitrine (unstained) and HA (red) were readily detectable, and expression of the two separate proteins were perfectly colocalized.

2.3 DRN-to-amygdala projections are targeted with intersectional viral vector approaches using AAV-DIO-1B in combination with CAV-Cre

We confirmed effective stereotaxic coordinates for retrograde transport to the caudal DRN by injecting fluorescent microspheres into the amygdala; the most effective coordinates for retrograde transport back to the caudal DRN were AP −1.25, ML ±2.8, DV −4.6 from bregma (Figure 5a). The same coordinates were used for CAV-Cre injections into the amygdala combined with AAV-DIO-1B injections into the DRN. CAV-Cre was retrogradely transported from the amygdala to the DRN, where Cre was expressed and inverted the DIO transgene into the correct reading frame, allowing expression of HA-5-HT_1B receptors in DRN neurons that projected to the amygdala. Immunohistochemistry for HA-tagged 5-HT_1B receptors showed immunoreactivity at nerve terminals in the amygdala (Figure 5b), as well as in the DRN in both cell bodies and neuronal projections (Figure 5c, top), demonstrating not only expression but also correct localization of the receptor to nerve terminals as is seen with endogenous 5-HT_1B receptors. Colabeling with Tph (Figure 5c, middle) showed most, but not full, colocalization of HA-5-HT_1B with Tph (Figure 5c, bottom). The number of infected neurons detected was modest but consistent with what we would expect with viral expression, especially since only about 10% of neurons from the DRN project to the amygdala (Ma et al., 1991).

Figure 5. Intersectional use of the AAV-DIO-1B viral vector in combination with CAV-Cre can selectively target specific brain circuits.

DRN-to-amygdala projections are validated with immunohistochemistry. (a) When red fluorescent microspheres were injected into several targets within the amygdala, the coordinates AP −1.25, ML ±2.8, DV −4.6 from bregma showed the greatest level of retrograde transport to the caudal DRN. Injection of AAV-DIO-1B into the DRN combined with injection of CAV-Cre into the amygdala resulted in visible transgene expression at nerve terminals in the amygdala (b; anti-HA staining in green) and in the DRN (c; anti-HA staining in green, anti-Tph staining in red), with HA-5-HT_1B receptors in both cell bodies and axon collaterals; HA-5-HT_1B colocalizes with Tph immunoreactivity in most, but not all, cells.

3.1 Increased levels of 5-HT_1B autoreceptors in nerve terminals throughout the brain reduce the expression of conditioned fear

By injecting the SERT-1B viral vector into the DRN, we tested the effect of viral-mediated expression of 5-HT_1B autoreceptors in WT and 1BKO mice on contextual fear conditioning. Analysis of freezing during the five minute test session revealed a significant reduction in freezing due to the 5-HT_1B viral vector, but no effect of genotype and no interaction between virus and genotype (Figure 6; two-way ANOVA; F_1,28 = 5.553, p = 0.025 for virus; F_1,28 = 0.825, p = 0.371 for genotype; F_1,28 = 1.121, p = 0.299 for interaction). WT and 1BKO mice that received the control SERT-GFP viral vector showed similar levels of freezing, which is consistent with previous reports suggesting that typical levels of contextual fear are observed in constitutive 1BKO mice as a result of compensatory adaptations arising from developing without 5-HT_1B receptors (Malleret et al., 1999). 1BKO mice receiving the SERT-1B viral vector, inducing transgenic 5-HT_1B autoreceptor expression while lacking 5-HT_1B heteroreceptors, showed a dramatic reduction in freezing, while WT mice receiving the SERT-1B viral vector, which were essentially overexpressing 5-HT_1B autoreceptors in caudal DRN neurons, showed an intermediate phenotype in between that of the controls and of the 1BKO mice receiving the SERT-1B viral vector.

Figure 6. Increased levels of 5-HT_1B autoreceptors reduce the expression of conditioned fear.

WT and 1BKO animals receiving the SERT-1B viral vector showed significantly less freezing in the test session following contextual fear conditioning compared to animals receiving the SERT-GFP control vector, with no effect of genotype on freezing or interaction between virus and genotype (two-way ANOVA, p = 0.025 for virus, p = 0.371 for genotype, p = 0.299 for interaction).

3.2 5-HT_1B receptor overexpression in the DRN-to-amygdala circuit increases fear expression in WT, but not 1BKO, mice

WT and 1BKO mice were injected with AAV-DIO-1B into the DRN in combination either with CAV-Cre into the amygdala to increase 5-HT_1B autoreceptors in the DRN-to-amygdala circuit, or with a CAV-GFP control vector. HA staining of cell bodies in the DRN were only observed when CAV-Cre was injected into the amygdala, indicating that recombination and expression of HA-5-HT_1B was Cre-dependent; no HA staining was observed in animals that received CAV-GFP injections into the amygdala. 5-HT_1B autoreceptor overexpression in this circuit significantly increased freezing in the test session in WT mice but not in 1BKO mice (Figure 7a; two-way ANOVA; F_1,39 = 4.497, p = 0.040 for interaction; F_1,39 = 1.035, p = 0.315 for virus; F_1,39 = 2.431, p = 0.127 for genotype). This was not due to differences between the two groups in anxiety as measured by the open field test (Figure 7b; two-way ANOVA; F_1,19 = 0.190, p = 0. 668 for virus; F_1,19 = 0.063, p = 0.804 for genotype; F_1,19 = 0.063, p = 0.804 for interaction) or differences in pain sensitivity as measured by the latency to the first freezing episode following the first shock (Figure 7c; two-way ANOVA; F_1,39 = 0.029, p = 0.865 for virus; F_1,39 = 1.386, p = 0.246 for genotype; F_1,39 = 0.068, p = 0.796 for interaction). Interestingly, while all groups showed increased freezing over the course of the seven minute training session, significant differences were found when comparing 1BKO mice receiving CAV-Cre both with WT mice receiving CAV-Cre (at 300, 360, and 420 sec) and with WT mice receiving CAV-GFP (at 420 sec) (Figure 7d; two-way repeated measures ANOVA; F_6,234 = 29.15, p < 0.0001 for time; F_{18, 234} = 2.063, p = 0.0079 for interaction; F_3,39 = 1.383, p = 0.262 for experimental group).

Figure 7. Increased levels of 5-HT_1B receptors in the DRN-to-amygdala circuit increases fear expression in WT mice.

WT and 1BKO mice received injections of AAV-DIO-1B into the DRN, and either CAV-Cre for 5-HT_1B receptor overexpression or CAV-GFP as a control. (a) 5-HT_1B receptor overexpression in the DRN-to-amygdala circuit significantly increased freezing following contextual fear conditioning in WT but not 1BKO mice (two-way ANOVA, p = 0.04 for interaction, p = 0.315 for virus, p = 0.127 for genotype). Viral-mediated expression of 5-HT_1B receptors in the DRN-to-amygdala circuit had no effect on (b) anxiety in the open field test (two-way ANOVA, p = 0. 668 for virus, p = 0.804 for genotype, p = 0.804 for interaction) or on (c) pain sensitivity, as measured by latency to the first freezing episode following the first shock (two-way ANOVA, p = 0.865 for virus, p = 0.246 for genotype, p = 0.796 for interaction). (d) Freezing during the training session increased over time, and freezing in 1BKO mice receiving CAV-Cre was significantly different from WT mice receiving CAV-Cre at 300, 360, and 420 sec and significantly different from WT mice receiving CAV-GFP at 420 sec (two-way repeated measures ANOVA, p < 0.0001 for time, p = 0.0079 for interaction, p = 0.262 for experimental group).

DISCUSSION

We describe two new viral vector strategies for manipulating 5-HT_1B expression using either phenotype-specific gene expression (with SERT promoter control of transgene expression) or intersectional transgene expression (with DIO vector and retrograde Cre expression). Both strategies transduce 5-HT_1B receptors predominantly in serotonergic neurons, achieving a greater degree of specificity than previous viral vector strategies allowed. Development of the serotonin-specific SERT-1B vector allows for greater control over viral expression, since a heterogeneous population of cells may be infected by an HSV injection, but only serotonergic neurons will have the necessary elements to induce transgene expression. Additionally, viral expression is easily visualized, suggesting that the SERT promoter is sufficiently strong to drive transgene expression. These qualities are highly advantageous for manipulating 5-HT_1B autoreceptor levels, as the viral vector can be injected into serotonergic cell bodies located in the DRN but the receptor itself will be trafficked to the nerve terminals just as are endogenous receptors. We did observe some staining of cell bodies as well, but no staining of dendrites.

The novel AAV-DIO-1B viral vector showed strong expression in the Pet1-Cre transgenic mouse line (Figure 4) and no expression in the absence of Cre (data not shown). Both mCitrine and the HA-tagged 5-HT_1B receptor are clearly detectable; while we did not quantify the expression of each protein, the strong expression of both products suggests that the 2A ribosomal skip peptide may have lower failure rates compared with other polycistronic viral vector approaches such as the use of the internal ribosome entry site, which tends to cause lower expression levels of the second protein encoded in the sequence (de Felipe, 2002, Chan et al., 2011, Ho et al., 2013). This intersectional strategy allows for exquisite control of specificity and permits the targeting of subpopulations of neurons as well as of discrete neural circuits.

One caveat of viral vector use is the potential for undesired anterograde or retrograde infection. This potential has been demonstrated for both HSV (Ugolini et al., 1987, McGovern et al., 2012) and for certain serotypes of AAV (Boulis et al., 2003, Burger et al., 2004, Rothermel et al., 2013). We previously observed that injection of HSV-GFP into nucleus accumbens induced many hundreds of neurons to express GFP at the injection site, whereas only a few VTA neurons were retrogradely infected (Neumaier et al., 2002b). Therefore, we conclude that the infected neurons at the HSV injection site account for the observed results in this and previous studies.

Interestingly, we observed very different behavioral results when 5-HT_1B autoreceptors were manipulated in the DRN using these two different strategies. Using the SERT-1B viral vector, which increases transgene expression in serotonergic neurons projecting all over the forebrain, 5-HT_1B autoreceptor overexpression decreases the expression of conditioned fear, which is consistent with a previous study in our lab that used a conventional HSV injected in the DRN of rats to drive 5-HT_1B receptor expression (McDevitt et al., 2011). The use of the selective SERT-1B viral vector verifies that it is indeed 5-HT_1B autoreceptors in serotonergic neurons that are mediating the reduction in fear. Notably, injection of SERT-1B into 1BKO mice, which lack both 5-HT_1B auto- and heteroreceptors, also results in fear attenuation; after the viral infection these mice express 5-HT_1B autoreceptors only, showing that 5-HT_1B autoreceptors alone are sufficient for this effect.

In contrast, increasing 5-HT_1B autoreceptors in the DRN projections to the amygdala did not reduce freezing; rather, overexpression of 5-HT_1B receptors in this circuit increases the expression of conditioned fear. This observation is not a result of differences in baseline freezing, fear acquisition, or pain sensitivity between control and experimental mice. This seemingly contradictory finding in the DRN-to-amygdala circuit suggests that 5-HT_1B autoreceptors in the amygdala may play a role in promoting fear expression, while 5-HT_1B receptors in other projections may be responsible for the attenuation of fear. Ethologically, the appropriate expression of fear responses in the face of danger offers animals a selective advantage (Blanchard et al., 1993), so it is possible that 5-HT_1B receptors in the DRN-to-amygdala circuit facilitate this process. This may occur through 5-HT_1B-mediated reductions in synaptic 5-HT levels, via increased reuptake through SERT (Daws et al., 2000, Hagan et al., 2012, Montanez et al., 2013), suppression of 5-HT synthesis (Hjorth et al., 1995), or inhibition of 5-HT release (Middlemiss and Hutson, 1990). Indeed, recent reports support this hypothesis, as depletion of 5-HT levels in the amygdala potentiates fear behaviors (Tran et al., 2013, Wellman et al., 2013), while increased 5-HT levels in the amygdala decreases fear expression (Burghardt and Bauer, 2013, Kitaichi et al., 2014). Other serotonin receptors also play a role in fear and anxiety behaviors, such as 5-HT₃ receptors, which modulate fear extinction (Kondo et al., 2014). Additionally, 5-HT_2A and 5-HT_2C receptors produce anxiety-like behaviors and enhance fear conditioning (Campbell and Merchant, 2003, Burghardt et al., 2007) while 5-HT_1A heteroreceptors in the amygdala decrease the expression of conditioned fear when activated (Li et al., 2006, Ferreira and Nobre, 2014). However, in contrast to these postsynaptic receptors, 5-HT_1B autoreceptors work through regulating synaptic 5-HT levels, which, in turn, affect signaling through all downstream 5-HT receptors.

Remarkably, manipulation of 5-HT_1B receptors in the DRN-to-amygdala circuit did not have the same effects on 1BKO mice as it did on WT mice. This may be due to developmental differences in 1BKO mice as a result of the constitutive absence of 5-HT_1B receptors. Because the amygdala receives strong serotonergic projections from the DRN, the absence of 5-HT_1B receptors throughout life may have led to fundamental changes in this circuit, such that reintroduction of 5-HT_1B autoreceptors in later life are unable to mediate the same functions as seen in WT mice. Although the virally expressed 5-HT_1B receptors are correctly trafficked to the axons and nerve terminals throughout the brain, it is difficult to quantify the amount of receptors that are being expressed and impossible to detect the percent change in 5-HT_1B autoreceptor expression as a result of viral infection.

One striking observation resulting from the intersectional use of the AAV-DIO-1B and CAV-Cre viral vectors is the degree of collateralization observed in the DRN. Injections of AAV-DIO-1B into the caudal DRN combined with CAV-Cre into the amygdala were expected to increase 5-HT_1B autoreceptor expression in the serotonergic nerve terminals in the amygdala; however, immunohistochemical staining for HA shows that the HA-tagged 5-HT_1B receptor is abundant in axon collaterals in the DRN as well (Figure 5c). It is unknown whether collaterals from these amygdala-projecting neurons synapse onto other amygdala-projecting neurons to form an integrated DRN-to-amygdala circuit, or onto neurons that project to other brain regions. In the latter case, it is conceivable that serotonergic neurons with projections to the amygdala may also project to and alter the signaling of other serotonergic neurons that send efferents to different brain regions, thereby affecting multiple circuits. Another open question is whether these collaterals might have complementary or opposing roles in regulating responses to fear. In any event, the contribution of 5-HT_1B autoreceptors in axon collaterals within the raphe nuclei has received insufficient attention but may be an important regulatory mechanism that exists alongside somatodendritic 5-HT_1A autoreceptors.

One limitation of the intersectional viral vector strategy using AAV-DIO-1B in combination with CAV-Cre to target the DRN-to-amygdala circuit is the inability to target specific cell types within the DRN. The DRN is a heterogenous structure, containing both serotonergic and nonserotonergic cells (Jacobs and Azmitia, 1992). Interestingly, nonserotonergic cells in the DRN also send efferents to the amygdala (Halberstadt and Balaban, 2008), so it is possible that 5-HT_1B receptor expression in nonserotonergic cells may contribute to the behavioral observations. As shown in Figure 5c, transgene expression is detected in both Tph-positive and Tph-negative cells. Recent evidence suggests that subdivisions in the DRN amongst serotonergic neurons themselves may have distinct functional roles in behavior, not only in their somatotopic organization (Paul and Lowry, 2013), but also in their expression of various genes that had previously been thought to define whether a cell was serotonergic, such as 5-HT_1A autoreceptors (Gaspar and Lillesaar, 2012, Kiyasova et al., 2013). A greater understanding of the organization of the DRN is necessary, as well as an investigation into the anatomical connections between subdivisions of the DRN and the different amygdaloid nuclei.

These new findings shed light on the role of both 5-HT_1B autoreceptors and, more broadly, the serotonergic system on fear behaviors. Further work is warranted given the complexity of the system, particularly the functionally distinct subdivisions of both the amygdala and the DRN (Pare et al., 2004, Sierra-Mercado et al., 2011, Asan et al., 2013, Paul and Lowry, 2013, Paul et al., 2014) and the collateralization of serotonergic neurons in the DRN (Steinbusch, 1981, Waselus et al., 2011).

Highlights.

• Two novel viral vectors express 5-HT_1B receptors in a cell type-specific manner.

• The effect of 5-HT_1B autoreceptors on fear depends on the circuitry involved.

• Increased levels of 5-HT_1B autoreceptors in the brain decrease fear expression.

• Increased 5-HT_1B receptors in raphe neurons projecting to amygdala increase fear.

Abbreviations

1BKO

5-HT_1B receptor constitutive knockout

AAV

adeno-associated virus

CAV

canine adenovirus

DIO

double-floxed inverted open reading frame

DRN

dorsal raphe nucleus

HA

hemagglutinin

HSV

Herpes simplex virus

NGS

normal goat serum

ORF

open reading frame

SERT

serotonin transporter

SSRI

selective serotonin reuptake inhibitor

Tph

tryptophan hydroxylase

WT

wild type