In this study, we have adapted a semi-automated method combining multi-step immunolabeling and high-resolution confocal imaging to reconstruct in 3D a total of 195.43 mm of SERT-immunoreactive axons and map the distribution of synaptophysin-immunoreactive boutons forming synaptic triads with excitatory or inhibitory neurochemical synapses throughout the mouse limbic brain. For the first time, we provide a comparative volumetric quantification of the distribution SERT-immunoreactive axons and the density of serotonergic excitatory and inhibitory synaptic triads within several limbic areas of the mouse brain.
Use of the serotonin transporter SERT as a marker of serotonergic axons
Serotonin immunolabeling has been extensively used to label serotonergic axons and quantify their densities in the brain. However, serotonin is a neurotransmitter which can be rapidly metabolized, and thus, the use of antibodies to label serotonin might underestimate the density of serotonin axons. Here, we choose to use an antibody against the serotonin transporter, which has been shown to be a more robust marker of serotonergic axons than 5-HT itself (Nielsen et al.
2006). However, precautions should also be considered when using SERT as a marker of 5-HT neurons due to the subtle differences that may be detected between SERT and 5-HT immunolabeling. During development, some non-5-HT producing neurons in the thalamus, limbic cortex, hypothalamus, retina and superior olivary nucleus transiently express the SERT (Lebrand et al.
1998; Narboux-Nême et al.
2008). In adult rats, although comparable labeling has been observed throughout the cerebral cortex and striatum with only minor differences in the hippocampus, entorhinal cortex and the NAC core, more pronounced differences have been detected in the caudal part of the NAC shell (Brown and Molliver
2000). This previous work has shown that the NAC shell is innervated by two functionally different types of 5-HTergic axons that either contains or lacks the SERT. Our method of labeling of 5-HT fibers with an antibody directed against SERT may have only uncovered a particular subset of serotonergic axons expressing SERT in the NAC shell and is, therefore, likely to have underestimated the density of serotonergic axons and their triadic associations in this region. Further work is needed to determine the distribution and proportion of 5-HT axons lacking the SERT in the mouse NAC shell as well as in other limbic brain regions.
Serotonergic triads, electron microscopy, and physiology
A similar approach to the one used in our study combined array tomography and high-resolution immunolabeling to identify 5-HT synaptic triads in the dorsal raphe nuclei. Both glutamatergic and GABAergic terminals converged onto serotonergic TPH-labeled neurons to modulate of the excitatory activity of serotonergic neurons in rats (Soiza-Reilly et al.
2013; Soiza-Reilly and Commons
2014). We therefore focused on mapping SERT-immunoreactive axons fiber density and distribution of putative serotonergic boutons forming synaptic triads with excitatory or inhibitory neurochemical synapses in mouse limbic brain regions known to receive a high level of 5-HT synaptic input. Our comprehensive assessment and detailed statistical analysis of the distribution of “excitatory vs inhibitory” serotonergic triads and their “pre- vs postsynaptic” location have highlighted some specific differences in the architecture of serotonergic triads within limbic brain areas. These data suggest that 5-HT axons might differentially modulate excitatory or inhibitory transmission in these brain regions, however, further work is needed to confirm the functional significance and ultrastructural distribution of these triads throughout the mouse brain.
In the layer I–III of the prefrontal cortex, the most densely 5-HT-innervated layer in the cortex (Audet et al.
1989), we showed that SYN
SERT+ boutons preferentially formed triads with neurochemical excitatory synapses, suggesting that 5-HT may have a preferential role in the regulation of glutamate transmission in this region. In line with this, electron microscopy studies have observed that junctional and non-junctional appositions of 5-HT-immunoreactive axons to non-5-HTergic axons or dendrites engaged in asymmetrical (excitatory) synapses in a triadic formation were frequently encountered in the upper layers of the frontal cortex in rats (Séguéla et al.
1989), suggesting both a pre- or postsynaptic control of excitatory transmission in this brain region. Electrophysiology studies in rats have also shown 5-HT-mediated pre- and postsynaptic effects in this brain region including increases both glutamate release and the amplitude of glutamatergic postsynaptic currents (EPSCs) (Aghajanian and Marek
1997). These interactions between 5-HT and glutamate within functional triadic contacts could play an important role in the etiology of psychosis and be a promising target for the development of antipsychotics (Marek and Aghajanian
1998) and the treatment of schizophrenia (González-Maeso et al.
2008).
In the NAC core and shell, we show that SYN
SERT+ boutons were equally distributed among excitatory and inhibitory neurochemical synapses with a non-significant preference to the formation of excitatory triads in the NACc. Triadic associations of 5-HT- or SERT-immunolabelled axons to non-serotonergic axons forming both symmetrical (inhibitory) or asymmetrical (excitatory) synapses have been previously observed by electron microscopy in the NAC core and shell of rats (Van Bockstaele and Pickel
1993; Pickel and Chan
1999). The proximity of 5-HT-positive axon terminals to GABA terminals engaged in symmetrical (inhibitory) synapses revealed by electron microscopy in rats (Van Bockstaele et al.
1996) suggests that 5-HT could influence the release of GABA. This is supported by functional studies showing a modulation of GABA release in the NAC by 5-HT2C receptors in rats (Kasper et al.
2015). Interestingly, we found a significantly higher proportion of SYN
SERT+ boutons located closer to the presynaptic component of putative inhibitory synapses within the NAC shell, suggesting that 5-HT could have a modulatory effect on GABA release from GABAergic synapses in this area. Furthermore, functional studies have also revealed 5-HT-mediated control of glutamate release in rat NAC core and shell slices via activation of presynaptic 5-HT1B receptors (Muramatsu et al.
1998), however, we did not observe any preferential location of SYN
SERT+ boutons to the pre- or the postsynaptic component of excitatory triads. Whether 5-HTergic boutons are truly equally distributed toward the pre- and postsynaptic components of glutamate synapses to regulate glutamate signaling in the mouse NAC or whether our methodology using a SERT antibody only focused on a particular subtype of 5-HT neurons in this region remains to be determined.
Among the brain regions analyzed, we found that the highest proportions of SYN
SERT+ boutons engaged in triads were in the BLA and CeA of the amygdala, reaching 61 and 58 %, respectively. We observed a slightly higher number of SYN
SERT+ boutons located in closer apposition toward inhibitory neurochemical synapses than excitatory synapses. In the BLA, serotonergic triads apposed to both symmetrical (inhibitory) and asymmetrical (excitatory) synapses have been identified by electron microscopy in rats (Muller et al.
2007). We have previously shown that in the rat BLA, interneurons contain a significantly higher neurochemical GABAergic synapse density compared with principal neurons (Klenowski et al.
2015). The preferential proximity of SYN
SERT+ boutons to neurochemical inhibitory synapses could therefore suggest that serotonergic axons projecting to the BLA may preferentially target local interneurons and modulate their activity. Although this requires further investigation, previous electrophysiology reports have demonstrated preferential 5-HT mediated effects on interneurons in the BLA (Rainnie
1999). At a presynaptic level, GABA release in rat BLA slices was shown to be inhibited by 5-HT1A (Koyama et al.
1999,
2002; Kishimoto et al.
2000) and activated by 5-HT2A (Jiang et al.
2008) and 5-HT3 (Koyama et al.
2000,
2002) receptors. Combined these results are consistent with 5-HT connectivity structured toward the regulation of inhibitory GABAergic activity in the BLA (Muller et al.
2007). Because the excitability of BLA principal cells, which presumably represents a functional basis of anxiety states, is potently controlled by local GABAergic interneurons (Rainnie et al.
1991; Washburn and Moises
1992; Lang and Paré
1997,
1998), the regulation of GABAergic synapses by 5-HT is in line with a potential role of 5-HT signaling in the BLA in anxiety-related behaviors (Strauss et al.
2013; Vicente and Zangrossi
2014).
The CeA consists primarily of GABAergic projection neurons and interneurons (Sun and Cassell
1993; Veinante and Freund-Mercier
1998) that integrate and modulate glutamatergic inputs from the thalamus, cortex and BLA (LeDoux
2007) to mediate behavioral and physiological responses associated with fear/anxiety (Kalin et al.
2004; Ciocchi et al.
2010) and various negative emotional states including stress/anxiety following alcohol withdrawal (Roberto et al.
2012; Gilpin et al.
2015). Our results showing the preferential locality of SYN
SERT+ boutons to inhibitory synapses in the CeA are consistent with the contribution of 5-HT signaling in the regulation of anxiety-related behaviors (Mo et al.
2008) being facilitated by modulation of the local GABAergic microcircuit in the CeA (Ciocchi et al.
2010; Jiang et al.
2014). Further functional studies are however needed to confirm the role of 5-HT in the regulation of local GABAergic circuitry.
In the VTA, we observed the highest density of SERT
+ fibers and SYN
SERT+ boutons, as well as the largest average fiber diameter. This result was likely given the proximity of the VTA to the raphe nucleus. Importantly, almost half of these total boutons (44 %) were located in close proximity to putative inhibitory synapses and a very low number (<1 %) in the proximity of putative excitatory synapses. Considering the large network of GABAergic neurons in the VTA, this result was somewhat expected and is in line with previous electron microscopy (Hervé et al.
1987) and functional studies. For example, 5-HT1B agonist application was shown to reduce [3H]-GABA release (Johnson et al.
1992; Yan and Yan
2001) and GABA
B-mediated IPSCs in VTA dopamine (DA) neurons (Cameron and Williams
1994). Furthermore, cocaine-induced reductions in GABA
B inhibitory postsynaptic potentials in DA neurons of rat VTA slices were found to be mediated by 5-HT1B receptor activation (Cameron and Williams
1994), which, in turn, facilitates cocaine-induced increases in DA levels in the NACc (Parsons et al.
1999; O’Dell and Parsons
2004). Collectively, these data suggest that 5-HT modulation of VTA signaling occurs primarily via the regulation of local and/or non-local inhibitory synapses. Dopaminergic neurons in the VTA that project to the NAC to form the mesolimbic reward pathway are sensitive to 5-HT/GABA interactions which affect the release of DA in the NAC, notably, in response to cocaine (Cameron and Williams
1994; O’Dell and Parsons
2004), MDMA (Bankson and Yamamoto
2004) and alcohol (Theile et al.
2009). Our methodology could help to identify changes in SERT
+ fiber density and 5-HT connectivity contributing to the dysregulation of DAergic signaling that is associated with the development of addictive behaviors.
Methodological considerations
Electron microscopy studies have revealed that 5-HT axons directly contact the dendrites and the cell bodies of various types of neurons via symmetrical or asymmetrical synapses, and also form triadic contacts with dendrites or axons that are engaged in synapses (for review, see Descarries et al.
2010). Symmetrical and asymmetrical synapses have been proposed to be inhibitory and excitatory, respectively; however, specific markers for the postsynaptic component of 5-HT symmetrical or asymmetrical synapses identified from EM have not yet been determined. Consequently, the quantitative distribution of these synapses cannot be conclusively resolved by fluorescence microscopy. Therefore, our study focused on quantifying the density of serotonergic boutons in close proximity to “conventional” synapses, identified using well-validated neurochemical markers of excitatory and inhibitory pre- and postsynaptic specializations, to estimate the density and distribution of serotonergic excitatory/inhibitory triads throughout the mouse limbic brain.
Interestingly, in some brain regions, including the mPFC, NACc, HIP, and VTA, the proportion of excitatory, inhibitory, and extra-triadic SYN
SERT+ boutons that we observed were in line with the proportion of asymmetrical, symmetrical, and extra-synaptic 5-HT boutons observed in previous electron microscopy studies (Hervé et al.
1987; Séguéla et al.
1989; Oleskevich et al.
1991; Van Bockstaele and Pickel
1993; Smiley and Goldman-Rakic
1996; Miner et al.
2000). If direct asymmetrical and symmetrical synapses made by 5-HT boutons on dendrites are presumed to be excitatory and inhibitory, respectively, our data suggest that the excitatory/inhibitory balance of 5-HT bouton connectivity is a common structural feature related to both 5-HTergic direct synapses and synaptic triads in brain regions that we have investigated. However, additional EM studies that characterize the postsynaptic densities of asymmetrical and symmetrical 5-HT synapses are required to verify this hypothesis.
The methodology described here is highly dependent on the specificity of the antibodies used. Careful consideration should be given to the choice of antibodies and the control of the non-specific labeling of each antibody, to determine the right antibody dilution, labeling sequence and to ensure the best signal-to-noise ratio. Here, we used a combination of specific, well-validated antibodies to achieve high-quality immunolabeling and high-resolution imaging. Future investigations implementing this methodology should also be aware that the generation of consistent quantitative data is highly dependent on slice preparation.
In summary, this method allows for a fast quantitative analysis of 5-HT innervation and connectivity in the mouse brain. The combination of this technique with other methods, for example, with the reconstruction of neurobiotin-filled neurons (Fogarty et al.
2013; Klenowski et al.
2015), offers the possibility for the distribution of 5-HTergic excitatory/inhibitory triads along the dendritic trees, axon or soma of a single intracellularly labeled neuron to be determined in the future studies.