Locally Collateralizing Glutamate Neurons in the Dorsal Raphe Nucleus Responsive to Substance P Contain Vesicular Glutamate Transporter 3 (VGLUT3)

Introduction

Three vesicular glutamate transporter isoforms (VGLUT1-3) have the capacity to fill synaptic vesicles with the neurotransmitter glutamate. VGLUT1 is highly expressed in the cerebral and cerebellar cortices, whereas subcortical areas predominantly utilize VGLUT2 (Fremeau et al., 2001; Herzog et al., 2001 Varoqui et al 2001). VGLUT3 is distinguished from VGLUT-1 and -2 in its' localization to brain areas and neurons that are not thought to use glutamate as a neurotransmitter. For example, VGLUT3 is found in interneurons of the hippocampus, cholinergic striatal interneurons and within the serotonergic dorsal and median raphe nuclei (Fremeau et al., 2002; Gras et al., 2002; Hioki et al., 2004; Schafer et al., 2002; Shutoh et al., 2008). Thus there is a trend for VGLUT3 to localize in axons that may use multiple fast neurotransmitters. However, in some cases VGLUT3 is required for glutamate neurotransmission, most notably at the synapse of the ear's inner hair cell (Seal et al., 2008).

Within the dorsal raphe nucleus (DR), one of the major sources of forebrain serotonin (5-HT), the relationship between VGLUT3 and 5-HT neurons has been examined in rat (Gras et al., 2002; Shutoh et al., 2008) and Syrian hamster (Mintz and Scott, 2006). These studies have consistently reported co-existence of VGLUT3 with markers for 5-HT within cell bodies and, to a varying extent, the existence of neurons that contain VGLUT3 in the absence of 5-HT and vice versa. Examination of VGLUT3 with respect to 5-HT axons in the forebrain reveal that some, but not all, forebrain 5-HT axons also contain VGLUT3 (Boulland et al., 2004; Hioki et al., 2004; Shutoh et al., 2008).

5-HT neurons within the DR are not homogeneously distributed, rather they can be divided into distinct subgroups based upon their cytoarchitecture and neurochemical features (reviewed by (Jacobs and Azmitia, 1992; Michelsen et al., 2007; Steinbusch, 1984)). However, it has never been examined if VGLUT3 axons, with or without 5-HT, have any sub-region specific characteristics within the DR. In fact, little is known about the role of VGLUT3 in DR circuits. Therefore, we examined the topographic distribution VGLUT3 containing axons with respect to potential co-transmitters. In addition, non-5HT VGLUT3-containing cell bodies were characterized for both their distribution and relationship to the receptor for substance P (neurokinin 1, NK1), which has previously been associated with glutamate neurotransmission in the DR (reviewed by Valentino and Commons, 2005).

Methods

The Animal Care and Use Committee at The Children's Hospital, Boston approved the animal procedures used. Fifteen male Sprague-Dawley rats (250-500 g) were anesthetized with sodium pentobarbital (75 mg/kg i.p.) to minimize pain and transcardially perfused. The perfusate consisted of 10 ml of saline followed by 250 ml of 4% paraformaldehyde in 0.1 M phosphate buffer (PB), pH 7.4. After perfusion, brains were removed and left in the same fixative overnight. Brains were then saturated in 30% sucrose, frozen in crushed-dry ice and sectioned 40 μm thick using a rotary microtome with a chuck cooled with a Peltier device. All sections were collected in 0.1 M PB with 0.9% saline (PBS) and processed free-floating.

Table 1 summarizes the antisera used. Overall, the strategy used to confirm specificity of all of the antisera was to characterize the immunolabeling attained in comparison to the known distribution of each of the antigens in the DR or other regions where they have been described. Since in each case, individual descriptions of gene expression and protein localization are well established, appropriate distribution in other regions provides a strong line of evidence to support specificity particularly when supported by additional evidence from the literature. Using the VGLUT3 antisera the distribution of VGLUT3 immunolabeled cell bodies was identical to prior descriptions in the ventral forebrain and striatum and the hippocampus. In addition, the distribution of VGLUT3 immunolabeled axons was consistent with previous descriptions, including for example localization to stratum pyramidale in the CA1 region of the hippocampus. This is consistent with the observations of others who compared immunolabeling with this antisera to that of others (Gabellec et al., 2007) and with the pattern of source neurons expressing the mRNA for this transporter (Gras et al., 2002; Schafer et al., 2002). In addition for VGLUT3, previous studies have reported that VGLUT3 immunolabeling is lost when the VGLUT3 antiserum is preadsorbed to the antigenic peptide (Gabellec et al., 2007; Shutoh et al., 2008)). Immunolabeling for 5-HT and SERT was definitive for the known distribution of 5-HT cells and axons, that is, cell bodies immunolabeled for these antigens were located in the raphe nuclei and absent from other areas of the brain including other monoamine nuclei. For VMAT2 the antibody is specific for 88 kDa protein by Western blot (Sandoval et al., 2002) and identifies neurons in the raphe nuclei as well as neurons in all other monoamine nuclei. Immunolabeling for GABAergic axons was determined using a well-characterized monoclonal antibody, GAD6 that recognizes GAD-65 (Chang and Gottlieb, 1988), obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA, as well as with a GAD-67 antiserum obtained from Millipore/Chemicon (Varea et al., 2005). These antisera were tested on sections containing the hippocampal formation, and there distribution was characteristic for the known localization of these GAD isoforms. The antiserum used to detect VAChT has previously been shown to identify cholinergic cell bodies and axons with high sensitivity (Arvidsson et al., 1997), and in our tests immunolabeling with this antisera defined both ventral forebrain and pontine cholinergic cell groups. A rabbit antiserum to detect NK1 receptor was raised against a C-terminal portion of the rat NK1 receptor (Vigna et al., 1989), which yields a pattern of immunolabeling identical to known descriptions in the hippocampus and striatum, as well as the DR (Commons and Valentino, 2002). An affinity purified anti-tryptophan hydroxylase (TPOH) polyclonal antisera raised in sheep was used to label the somatodendritic extent of 5-HT neurons. This TPOH antiserum identifies all 5-HT labeled cells in the raphe nuclei and exhibits variable cross-reactivity to neurons in catecholamine containing nuclei probably due to some similarity between TPOH and tyrosine hydroxylase.

Immunohistochemical processing took place with continuous light agitation. Between incubations the tissue was rinsed several times with PBS. For immunolabeling, antisera were diluted in PBS and 0.5% bovine serum albumin and tissue was incubated either overnight at room temperature, or 2-3 days in the coldroom with constant mild agitation. For visualization of axons 0.1% TRITON X-100 was added to the buffer but was excluded to maximize cell body labeling for VGLUT3. Fluorophore-conjugated secondary antisera raised in donkey were purchased from Jackson ImmunoResearch (CY3 conjugated) or Molecular Probes (Alexa-488 and Alexa-647 conjugated). Tissue was incubated with secondary antisera diluted 1:100 for 90 minutes at room temperature. After rinsing, immunolabeled tissue was mounted on glass slides from 0.5 M PB, dried and cover-slipped using mounting medium containing 90% glycerol and 10% PBS pH 8.5-9.

Results

VGLUT3 immunolabeling overall appeared enriched within the DR in comparison to the neighboring periaqueductal gray (Fig. 1). Examination at higher magnification revealed that immunolabeling for VGLUT3 identified a population of intensely labeled axonal boutons as well as less intense and more diffuse labeling within cell bodies (Fig. 2). Since previous studies have reported that 5-HT and VGLUT3 colocalize, mapping the colocalization of VGLUT3-immunolabeled axons with 5-HT was the first focus of our study.

In most subregions of the DR only a few axons appeared to be double labeled for 5-HT and VGLUT3 (Figs. 3, 4A). This lack of co-localization was confirmed by dual labeling of VGLUT3 with SERT (not shown). However, at all levels 5-HT axons within the ependyma were obviously dually labeled for VGLUT3 (Fig. 5). Innervation of the ependyma arises from clusters of cells at the base of the aqueduct in the caudal DR and additional cells located ventral to these (Mikkelsen et al., 1997; Simpson et al., 1998). The caudal dorsal clusters of 5-HT cells were densely surrounded by axons containing both VGLUT3 and 5-HT (Fig. 5). In addition, the caudal ventral DR had a higher level of colocalization between VGLUT3 and 5-HT than other regions of the DR.

VGLUT3 immunolabeling was next examined with respect to that for VMAT2 in axons to evaluate if VGLUT3 could be colocalized with another monoamine in the remainder of the DR. VMAT2 immunolabeling yielded clearer definition of axon-like boutons with less neuropil labeling when compared to 5-HT (Fig. 4). However, estimates of co-localization between VGLUT3 and VMAT2 were just slightly higher than those for 5-HT and VGLUT3 in all DR subregions (Figs. 3-4).

To determine if VGLUT3 colocalized with GABA within axon terminals, tissue was dually labeled for VGLUT3 and GAD65 or separately for VGLUT3 and GAD67. Immunolabeling for either GAD isoform resulted in abundant labeling of axonal boutons throughout the DR (Fig. 4). Co-localization was studied throughout the rosto-caudal extent of the nucleus. Only in the caudal ventral DR was there a trend for increased co-localization for VGLUT3 and GAD65, however these accounted for a small fraction of VGLUT3 labeled axons.

To determine if VGLUT3 colocalized with acetylcholine within axon terminals, tissue was dually labeled for VAChT. VAChT positive cells and axons were found in the lateral dorsal tegmental nucleus and elsewhere within the same tissue sections. However, VACHT-immunolabeling yielded sparse axonal labeling through the DR, and these were not detectably dually labeled for VGLUT3 in any subregion (Fig. 3-4).

Therefore with the exception of the caudal dorsal DR, VGLUT3 did not apparently colocalize with another neurotransmitter within axons. VGLUT3-containing cell bodies were investigated to further understand their identity. Since the colocalization of VGLUT3 in 5-HT cells has previously been evaluated using both immunohistochemistry and in situ hybridization techniques (Gras et al., 2002; Mintz and Scott, 2006; Shutoh et al., 2008), we focused on VGLUT3 cell bodies that lacked 5-HT immunoreactivity. Cell bodies immunolabeled for VGLUT3-only were mapped with respect to 5-HT at different levels, revealing there were present along the entire rostral-cuadal axis of the DR (Fig. 5). They appeared most prevalent at a mid-dorsoventral and mid-rostrocaudal location (Fig. 5B) although they were spread along the entire rostocaudal extent of the DR. Typically cells single labeled for VGLUT3 favored the midline (Fig. 5).

Previously glutamate-immunoreactive cells containing the substance P receptor, neurokinin 1 (NK1), were described in the DR (Commons, and Valentino, 2002), therefore it was determined if non-5-HT cells with VGLUT3 had immunolabeling for the NK1 receptor. For this, 5-HT cells were identified by the presence of the synthetic enzyme tryptophan hydroxylase (TPOH). NK1 distribution was identical to that previously published in the DR (Barbaresi, 1998; Commons and Valentino, 2002). Dually labeled cells were commonly found, particularly at mid-rostrocaudal levels of the DR, and appeared to account for the majority of non-5-HT VGLUT3 containing cells (Fig. 6). In an arbitrary selection of fields with triple labeling, 88 of 122 VGLUT3 labeled cells were also labeled for NK1 (72%). VGLUT3- and NK1-dually labeled cells were smaller than TPOH labeled cells, and were approximately 14 um in diameter vs. average of 21 um in diameter for TPOH cells. NK1 immunolabeling was almost entirely associated with the plasma membrane of cells and therefore was easily distinguished from overlying cytoplasmic VGLUT3 labeling within cell soma.

Discussion

The DR is enriched with VGLUT3-containing cell bodies and axons, yet the relationship of these neurons to 5-HT cells has remained incompletely understood. Immunolabeling and fluorescence microscopy revealed in this study that VGLUT3-containing axons in the DR are comprised of at least two populations. VGLUT3 colocalizes with 5-HT in some axons, and these preferentially innervate the caudal dorsal portion of the DR. Many VGLUT3 axons lacked 5-HT or another obvious co-transmitter, and at the same time many VGLUT3-labeled cell bodies had the receptor for substance P. This evidence suggests that non-5-HT VGLUT3 cells comprise a population of locally collateralizing glutamate neurons. Both of these populations likely play a role regulating excitation of 5-HT neurons within the DR.

All of the antigens examined in this study have well known independent patterns of distribution, providing evidence for the specificity of their immunolabeling and colocalization. However, the differing propensity of antigens to be enriched in distinct subcellular compartments could influence the accuracy of quantifying colocalization. Although VGLUT3 immunolabeling in cell bodies can be diffuse and much less intense then axon terminal labeling, the different subcellular distribution of labeling for VGLUT3 and NK1 receptors facilitated estimating colocalization between these antigens. With respect to colocalization of VGLUT3 and 5-HT in cell bodies, the technical approach used in the current study did not allow further insight beyond previously published careful descriptions using both immunohistochemistry or dual in situ hybridization (Gras et al., 2002; Mintz and Scott, 2006; Shutoh et al., 2008). With respect to axons, individual 5-HT boutons were partially obscured by subcellular punctate labeling of 5-HT in soma and dendrites combined with a high degree of neuropil immunolabeling. However, the lack of co-localization of 5-HT and VGLUT3 was confirmed by both SERT and VMAT2. That is, VMAT2 colocalized with VGLUT3 in only a slightly greater population of axons than 5-HT either because of the greater efficiency of detecting VMAT2 within axons and/or the additional fractional colocalization of VGLUT3 with another monoamine labeled by VMAT2.

The results of this study show that within different DR subregions there is a differing extent of colocalization of VGLUT3 and 5-HT within axons. This result parallels previous descriptions of the relationship between VGLUT3 and 5-HT in the forebrain. VGLUT3 and 5-HT colocalize in axons in the ependyma (Shutoh et al., 2008), and this innervation is sometimes referred to as the supraependymal plexus. These axons are well characterized to arise from the caudal DR, both from clusters of cells at the base of the aqueduct and additional ventral cells (Mikkelsen et al., 1997; Simpson et al., 1998). The density of 5-HT innervation of the supraependymal plexus correlates with ciliated regions (Mikkelsen et al., 1997) and may function in regulating the beating rate of cilia and subsequently the flow of cerebral spinal fluid (Nguyen et al., 2001). In addition, prominent colocalization between VGLUT3 and 5-HT in forebrain exists in the dorsal part of the lateral septum (Shutoh et al., 2008) and projections to this area could also arise from the caudal DR (Vertes, 1991; Waselus et al., 2006). Thus the high level of colocalization in these projection areas correlates with the high level of colocalization within the caudal DR, particularly dorsally. The selective colocalization of VGLUT3 and 5-HT in this area gives insight into the organization of re-current axon collateral innervation of the DR. That is, this caudal area that provides dually labeled innervation to the ependyma and/or the septum likely provides recurrent feedback innervation that is also dually labeled. At the same time, these dual 5-HT-VGLUT3 labeled axons provide only limited innervation of other subregions of the DR. These observations provide circumstantial evidence for region-autonomous feedback regulation, at least for this group of cells.

In contrast, colocalization of VGLUT3 and 5-HT is sparse or negligible in subregions of the amygdala, which are innervated by cells located at mid-levels of the DR (Imai et al., 1986; Vertes, 1991), and the striatum, which is preferentially innervated by the rostral DR (Steinbusch et al., 1981; Vertes, 1991; Brown and Molliver, 2000; Waselus et al., 2006). Consistent with these observations, this study showed that 5-HT axons which innervate the mid and rostral DR, and in part could be composed of recurrent axon collaterals, have little colocalization with VGLUT3. However, at the level of cell soma many 5-HT cells have consistently been reported to contain VGLUT3 (Gras et al., 2002; Hioki et al., 2004; Mintz and Scott, 2006; Shutoh et al., 2008). Evidently low expression of VGLUT3, or expression that is post-translationally regulated, does not fill the axonal arbor of these cells. Consistent with this conclusion, a previous study describing the colocalization of VGLUT3 and 5-HT in the forebrain noted that even within a single 5-HT fiber, some, but not all, varicosities may have detectable VGLUT3 (Shutoh et al., 2008).

Within the DR, axons that colocalize 5-HT and VGLUT3 appear to represent a minority population, with single-labeled axons predominating in most areas. This partial colocalization distinguishes the relationship between VGLUT3 and 5-HT in the DR from that between VGLUT3 and acetylcholine in the striatum, which appears to be a more complete co-existence (Gras et al., 2002). The perception that VGLUT3 is widespread in 5-HT axons is created by the high expression of VGLUT3 in the DR, however non-5-HT cells may be a major contributor to this expression.

In this study, it was demonstrated that many non-5-HT cells that contained VGLUT3 were dually labeled for the NK1 receptor. Previous studies have shown that application of substance P in a DR slice preparation increases glutamatergic postsynaptic potentials in 5-HT neurons (Liu et al., 2002). Parallel anatomic studies confirmed that the NK1 receptor is present on glutamate-immunolabeled DR neurons (Commons and Valentino, 2002). Finally, using in vivo single-unit recordings the effects of substance P in the DR are blocked by a non-selective glutamate receptor antagonists, kynurenate (Valentino et al., 2003; Valentino and Commons, 2005). Thus the colocalization between NK1 and VGLUT3 provides strong evidence that VGLUT3 containing neurons within the DR have local axon collaterals that release glutamate onto resident neurons. Consistent with this conclusion, VGLUT3-containing axons do not prominently colocalize with other monoamines, GABA or acetylcholine.

In this study it was observed that 5-HT and VGLUT3 immunolabeling tend to colocalize within axons only in some regions of the DR, and these areas correspond to origins of dually-immunolabeled projections. VGLUT3 and 5-HT more commonly represent distinct populations of axons in the DR. Non-5-HT VGLUT3 containing cells were often immunolabeled for the NK1 receptor, and the circuitry underlying effects of activating NK1 receptors has been well studied (reviewed by (Valentino and Commons, 2005)). Specifically, activation of NK1 receptors within the DR leads to activation of local glutamatergic axons. Thus the results of this study place VGLUT3 within a predefined functional context in the DR nucleus.