Morphology and Connectivity of Parabrachial and Cortical Inputs to Gustatory Thalamus in Rats

Abstract

Ventroposterior medialis parvocellularis (VPMpc) of thalamus, the thalamic relay nucleus for gustatory sensation, receives primary input from parabrachial nucleus, and projects to insular cortex. To reveal unique properties of gustatory thalamus in comparison to archetypical sensory relay nuclei, this study examines the morphology of synaptic circuitry in VPMpc, focusing on parabrachiothalamic driver input and corticothalamic feedback. Anterogradely visualized parabrachiothalamic fibers in VPMpc bear large swellings. At electron microscope resolution, parabrachiothalamic axons are myelinated and make large boutons, forming multiple asymmetric, adherent and perforated synapses onto large caliber dendrites and dendrite initial segments. Labeled boutons contain dense-core vesicles, and they resemble a population of calcitonin gene-related peptide containing terminals within VPMpc. As typical of primary inputs to other thalamic nuclei, parabrachiothalamic terminals are over 5 times larger than other inputs, while constituting only 2% of all synapses. Glomeruli and triadic arrangements, characteristic features of other sensory thalamic nuclei, are not encountered. As revealed by anterograde tracer injections into insular cortex, corticothalamic projections in VPMpc form a dense network of fine fibers bearing small boutons. Corticothalamic terminals within VPMpc were also observed to synapse on cells that were retrogradely filled from the same injections. The results constitute an initial survey in describing unique anatomical properties of rodent gustatory thalamus.

Introduction

The chemical identities of taste stimuli, and psychological aspects of orosensory experience, such as palatability and expectation, collectively influence information processing in the rodent gustatory cortex (Katz et al., 2001; Spector and Travers, 2005; Stapleton et al., 2006; Carleton et al., 2010; Samuelsen et al., 2012). Taste information reaches the gustatory cortex from two pathways: a thalamocortical projection from the venteroposterior medialis parvocellularis nucleus of the thalamus (VPMpc; Kosar et al., 1986a), and limbic projections, including from the lateral hypothalamus and amygdala (Norgren, 1976; Allen et al., 1991). These inputs are distinct in terms of the cortical areas they target, and function. Projections directly from limbic structures tend to be more diffuse within the insular cortex (Allen et al., 1991), and are implicated in affective aspects of gustatory processing (Piette et al., 2012; Samuelsen et al., 2012). In contrast, thalamocortical projections from VPMpc largely terminate in the granular and dysgranular insular cortex, and provide orosensory aspects of taste processing (Sewards, 2004; Samuelsen et al., 2013). Similar to the role of thalamic relay nuclei in other systems (Malpeli, 1983; Poulet et al., 2012), blockade of neuronal activity in VPMpc hampers the ability of cortex to process gustatory information (Samuelsen et al., 2013). Despite its importance in the hierarchy of information flow, our knowledge on the synaptic circuitry of the gustatory thalamus remains rudimentary.

In rat gustatory system, taste information from the oral cavity travels through the nucleus of the solitary tract (NTS), and then to the parabrachial nucleus of the pons (PBN) before reaching VPMpc (Blum et al., 1943; Benjamin and Akert, 1959; Emmers, 1964; Norgren and Leonard, 1971; 1973; Hamilton and Norgren, 1984; Pritchard et al., 1986; Herbert et al., 1990). PBN input to VPMpc arises primarily from the gustatory responsive, waist region of the ipsilateral PBN (Lasiter and Kachele, 1988; Halsell and Travers, 1997; Karimnamazi and Travers, 1998; Krout and Loewy, 2000; Tokita et al., 2010). Presumed relay cells of VPMpc send thalamocortical axons to the insular cortex (Norgren and Wolf, 1975; Kosar et al., 1986a; Allen et al., 1991), which is reciprocally connected to VPMpc (Kosar et al., 1986b). The VPMpc also projects to the thalamic reticular nucleus (TRN; Stehberg et al., 2001), probably via axon collaterals of thalamocortical relay cells. TRN sends inhibitory projections back to the VPMpc (Hayama et al., 1994; Hanamori, 2003) forming an intrathalamic feedback loop.

To describe the morphological properties of putative taste axons originating in PBN and project to VPMpc (henceforth, referred to as parabrachiothalamic terminals) at electron microscope resolution, the present study utilized tract tracing to fill axons anterogradely from PBN waist region that was described to contain neurons responsive to taste stimuli. Corticothalamic axons were studied after anterograde tracer injections into insular cortex. Terminals from both areas were examined on the light microscope for localization within the VPMpc and on the electron microscope for quantitative ultrastructural analysis.

Methods

Animals

A total of 25 adult, female Sprague-Dawley rats (Harlan, RRID:RGD_70508) were used. Fourteen animals received tracer injections in PBN, six received tracer injections in IC, two received injections into VPMpc, and 3 received no injection, but were processed for calcitonin gene-related peptide (CGRP) immunoreactivity. Four PBN-injection, and three IC-injection brains, in which injection site was PBN or IC, and with acceptable structural preservation, were processed for EM. All animals were maintained on a 12:12 light:dark cycle and had ad libitum access to water and standard rat feed. Rats were 3–6 months old, and weighed less than 250 grams at the time of injections. All procedures were approved by University of Virginia Institutional Animal Care and Use Committee.

Tracer injections

Animals were anesthetized with a combination of ketamine (75mg/kg; I.P.) and medetomidine (0.5 mg/kg; I.P.), and placed into a non-traumatic stereotaxic apparatus (Kopf). The skull was exposed to reveal bregma, lambda and sagittal sutures, and a small hole was drilled. For PBN injections, the craniotomy was centered 11.4 mm caudal to bregma and 1.7 mm lateral to the midsagittal suture. PBN injections were angled caudal to rostral at 20° and descended 5.5 mm from the dural surface. VPMpc injections were 4.45 mm caudal to bregma, and 1.1 mm lateral to the midsagittal suture; injections descended perpendicular to the skull 6.6 mm from the dural surface. For IC injections, a lateral craniotomy was performed to expose the intersection of rhinal vein and medial cerebral artery. Three injections were made 2 mm dorsal to this intersection and a along a strip 2 mm anterior (one site) and posterior (two injection sites) to the medial cerebral artery. A glass pipette filled with tracer was descended 1–2 mm perpendicular to the cortex surface. Approximately 200 nl of biotinylated dextran amine (BDA; Invitrogen) in acidic medium (citrate buffer; pH 3.0) or neutral medium was injected using a glass micropipette with fine inner lumen (0.25 mm; A–M systems Carlsborg, WA) and a Picospritzer III (Parker Hannifin, Cleveland, OH) at each site. As acidic medium allows BDA to travel both anterogradely and retrogradely, it was used for both PBN and VPMpc injections, which provided confirmation of input from the NST and PBN, respectively. Non-acidic media that facilitates transport in fine axons was used to study feedback projections in IC injections (Reiner et al., 2000), although these also led to retrograde labeling in thalamus. The volume of tracer injected was calculated based on the inner diameter of the pipette and the advance of tracer fill-line that was plunged after small air puffs applied through the pipette. Following injections, the scalp was sutured and the animal revived with atipamezole (1 mg/kg; SC) and administered the postoperative analgesic buprenex (3 mg/kg; SC). Following a 3 day survival time, the animal was deeply anesthetized with an overdose of Nembutal (excess of 150mg/kg, IP) and transcardially perfused using room-temperature Tyrode's Solution [137 mM NaCl, 5.5 mM Dextrose/Glucose, 1.2 mM MgCl₂, 2 mM KCl, 0.4 mM NaH₂PO₄, 0.9 mM CaCl₂, 11.9 mM NaHCO₃, in 1 L filtered dH₂0] followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB; pH 7.4) for light microscopy processed animals. An additional 0.5% glutaraldehyde was added to the paraformaldehyde for animals used for electron microscopy. Brains were removed and allowed to post-fix overnight in 4% paraformaldehyde at room temperature before blocking and sectioning the brainstem and thalamus in 60 µm coronal sections on a vibratome. Sections that were not processed immediately were treated with 1% sodium borohydrate in PB in order to stop fixation and stored in 0.05% sodium azide in PBS at 4° C.

Tissue Preparation

Sections for light microscopy were processed serially in three interleaved sets to view tracer localization relative to histochemical markers. The first series was permeabilized with 0.1% Triton X-100 in phosphate buffered saline (PBS) for 30 minutes, and incubated in 1:100 dilution of avidin-biotin peroxidase (ABC kit; Vector Laboratories, Cat# PK-7100 RRID:AB_2336827) overnight at 4°C. Tracer-containing neuropil was visualized with diaminobenzidine (DAB). The second series was stained with Cresyl Violet for cell body visualization. The third series was treated with gold hydrochloride in order to view myelination, as described previously (Corson et al., 2012). Briefly, sections mounted onto gelatin subbed slides were rehydrated in 0.01 M PBS for 2 minutes and then incubated in 0.1% HAuCl₄ in 0.02 M PBS (pH 7.0) for 10–12 minutes at 60°C. After fine myelination was differentiated, the slides were incubated in 0.1% KAuCl₄ in saline (pH 7.0) for 3 minutes at 60°C. Following 2 × 2 minute rinses in 0.01 M PBS, the slides were incubated in 1% sodium thiosulfate in dH₂O at 60 °C for 3 minutes, rinsed 5 × 3 minutes in 0.01 M PBS, dehydrated, and coverslipped. These series allowed identification of anterogradely labeled terminal locations in histologically-defined nuclei. All sections were mounted serially on gelatin subbed slides, dehydrated and coverslipped with DPX mounting media (Sigma Aldrich).

Sections for electron microscopy were processed with DAB as above, and embedded in plastic resin using standard procedures. Briefly, the sections were post-fixed with 4% osmium tetroxide for 1 hour, counterstained with uranyl acetate (in 70% ETOH), dehydrated in increasing concentrations of ETOH, followed by treatments in acetone, an acetone and resin mixture, and pure resin. Sections were flat-embedded between two sheets of Aclar Embedding Film (Electron Microscopy Sciences), and polymerized at 60 °C for 24–48 hours, before re-embedding of sections containing VPMpc in capsules for ultrathin sectioning. Capsules were trimmed and 70 nm sections were cut using an ultramicrotome (Leica, Model: EM UC7) at planes quasi-parallel to the surface to obtain uniformly labeled trapezoids. In all brains, labeling was found to have penetrated well into the embedded sections, allowing for several rounds of ultrathin sectioning with sufficient label per block face. Copper mesh grids were used for general circuitry analysis, while copper slot grids (Electron Microscopy Sciences) were used for 3D reconstruction.

Tissue sections used for calcitonin gene-related peptide immunostaining were first treated with 1% sodium borohydrate for 30 minutes, rinsed in 0.01 M PBS until bubbles cleared, and then blocked in 1% bovine serum albumin (BSA, Sigma Aldrich) in 0.01 M PBS for 30 minutes. The sections were then incubated in a monoclonal antibody against CGRP (Sigma-Aldrich Cat# C7113 RRID:AB_259000; 1:1000; Table 1) and 1% BSA in 0.01 M PBS with 0.05% NaN₃ and 0.06% Triton X-100 (ICN Biomedicals Inc.) for 1–3 days. The CGRP antibody (clone #4901) was produced monoclonally against 10 polypeptide sequence (out of 37) of rat CGRP, recognizing both CGRP-I and CGRP-II (Wong et al., 2014). The sensitivity and specificity to rat CGRP were characterized by radioimmunoassay. The #4901 antibody completely abolished the portal release of somatostatin and the inhibition of gastric acid secretion induced by intravenous infusion of rat CGRP in anesthetized rats (Wong et al., 2014). This antibody lacks positive immunoreactivity in histological sections of from alpha-CGRP knockouts (Zhang et al., 2001). The pattern of CGRP immunoreactivity detected in current study was also in agreement with previous studies of CGRP immunoreactivity using other antibodies (Yasui et al., 1989; 1991). The sections were then rinsed in 0.01 M PBS to terminate the primary incubation, and transferred into biotinylated anti-mouse secondary antibody (Vector Laboratories Cat# BA-9200 RRID:AB_2336171; 1:100 dilution) for 2 hours, followed by ABC-DAB visualization.

Table 1.

Name	Host	Antigen Characteristics	Source	Dilution
Anti- Calcitonin Gene- Related Peptide antibody	Mouse (Monoclonal)	C-terminal 10 amino acids of rat CGRP. (Wong et al. 1993)	Sigma Aldrich, Saint Louis, MI (Sigma- Aldrich Cat# C7113 RRID:AB_259000)	1:1,000
Anti-mouse secondary antibody	Goat (Polyclonal)	Mouse IgG, conjugated to biotin	Vector laboratories (Vector Laboratories Cat# BA-9200 RRID:AB_2336171)	1:100

Analysis and Statistics

Mounted sections were analyzed and photographed using a Leica DC100 digital camera mounted on a Leica light microscope (Model LMDC 888011). All images were captured at a resolution to discern single fibers (Myelin or tracer labeled) using 40X or 63X objectives. Large fields of view, such as to capture full coronal sections, were created by tiling adjacent images using Adobe Photoshop CS5 software. For selecting areas to be analyzed on the electron microscope, resin-embedded sections, adjacent to Myelin and Nissl labeled mounted slides, were also imaged on the light microscope. Special care was taken to correctly document anatomical landmarks, such as the mammillothalamic tract, medial lemniscus, fasciculus retroflexus, habenular nucleus and fornix to ensure accurate sampling from within the VPMpc for electron microscopy. Once the landmarks of tissue were identified on the resin-embedded sections, blocks were trimmed to contain VPMpc, and 80 nm ultrathin sections were cut on a Leica Ultracut UTC7 from the surface 2–5 µm of the blocks containing immunolabel. Electron micrographs were taken with a JEOL1010 transmission electron microscope equipped with a 16 megapixel CCD camera. For synapse density analysis, edge-overlapped images were taken at 10,000 magnification to make composite pictures of large regions within VPMpc, along a 30–50 µm strip of tissue at the tissue-resin interface. The composite areas contained 666 synapses in total. This dataset was used for analysis of labeled terminal frequency. For quantification of labeled terminal morphology, a second dataset that was comprised of images of all labeled instances encountered in ultrathin sections were captured. The pixel resolution of EM images was 2.75 nm. Image ProPlus 5.1 (Media Cybernetics) was used to make measurements from micrographs. Data analysis was completed using Microsoft Excel 2007 and SPSS (RRID:rid_000042), while graphs and figures were created with OriginPro 7.5, Adobe Photoshop CS5, and Adobe Illustrator CS5. Three-dimensional reconstructions were completed using the software Synapse Web Reconstruct (Fiala, 2005; RRID:nif-0000-23420).

To estimate dendrite calibers, the ‘feret minimum’ (feret_min) measure of Image ProPlus (Media Cybernetics) was used. This algorithm optimally places a rectangle outside of the dendrite cross-section outline, the shape of which may vary from round to elliptical (at bifurcation points more irregular cross-sections form). While the long side of this rectangle (feret_max) depends on how obliquely the cylindrical dendrite was cut, the short side (feret_min) provides a close estimate of the cylinder caliber. Please note that ‘feret_min’ is different than the ‘shortest diameter’ measurement that is occasionally used in the literature in that, feret_min is not affected by irregularities in dendrite shape, or the relative position of longitudinal cut-plane to the axis of the cylinder. With the exception of unlikely instances that a dendrite may be sectioned at the apex of a bend, feret_min measurements are the most accurate estimate of the caliber of a cylindrical dendrite using two-dimensional cross-sections. For terminal area measurements, labeled and unlabeled terminals were outlined; for synapse density analysis, each synapse was counted, its cross-section length and the small neuropil area (area within each image excluding blood vessels, somata and myelinated axons) was measured. Nonparametric statistical tools were used for comparisons of area measures obtained from terminal or dendrite populations.

Results

Parabrachiothalamic Input to VPMpc

Light Microscopy

Stereotaxic injections of BDA into the gustatory responsive waist region of the parabrachial nucleus led to visualization of parabrachiothalamic axons in the ipsilateral medial thalamus, with a particularly dense projection site situated dorsal and medial to the medial lemniscus (Figure 1 and Figure 2). Using serial coronal sections processed for DAB (tracer), Nissl or myelin in sequence, the region of dense projections was transposed on adjacent Nissl and myelin-stained sections, using medial lemniscus (ml) and blood vessels as landmark structures (Figure 2D–K). Next, the Nissl-DAB-myelin triplets were matched to comparable sections from a digital rat brain atlas (Paxinos and Watson, 2007; Rat Brain Atlas of Paxinos and Watson, RRID:nlx_152120), based on the outlines of ic, cp, habenula, ml, mt and f (Figure 2G and K). As outlined in Rat Brain Atlas, VPMpc extended approximately 1 mm in the medial-lateral axis, and 0.8 mm in the anterior-posterior axis. As consistent with prior studies that described the projection site of PBN neurons located in the waist region (Cechetto and Saper, 1987), the region containing dense anterograde labeling was within the medial portion of VPMpc in all cases. Beyond providing landmarks for cell-sparse regions, Nissl-stained sections did not reveal any demarcations that matched the outlines provided in Paxinos atlas. However, myelin sections were useful in demarcating myelin dense ventral (ml and arrows in Figure 2J) and dorsal (arrowheads in Figure 2J) landmarks. Neither the areas demarcated as VPMpc in the atlas, nor the myelination landmarks, provided a close-fitting envelope for the anterograde projection zone: the densely labeled region in VPMpc was always surrounded by a projection-free zone, the outer borders of which matched dorsal and ventral myelination landmarks (Figure 2J). An additional useful landmark was a large blood vessel that ran at medal-ventral and dorsal-lateral axis, with several lateral bifurcations. The blood vessel was just medial to the dense anterograde labeled region in VPMpc in all cases examined (Figure 2H–K asterisks). In most injection cases, DAB label was also sparsely present in areas outside of VPMpc borders that are demarcated by the myelination landmarks (Figure 2H). Sparse fibers were also found in the contralateral hemisphere, in and around VPMpc, but only the parabrachial fibers within the ipsilateral VPMpc were characterized further using the electron microscope. Using high magnification objectives, we also observed that parabrachiothalamic projections within VPMpc were comprised of a dense bundle of thick fibers with large swellings occurring along these labeled fibers (Figure 3).

Figure 1.

The illustration of tracer injections sites in the parabrachial nucleus (A–C) and corresponding projection sites in the thalamus (D–F). Each color-coded dashed-line outlines the densest region of tracer labeling in the brainstem in coronal sections corresponding to stereotaxic coordinates AP-9.24 (A), AP −9,48 (B) and AP-9.72 (bregma; Paxinos Rat Atlas); these are marked as PBN injection 1–5 (A). For each injection, the projection site in thalamus is also marked in corresponding coronal sections from Paxinos Rat Atlas, at AP-3.72, AP-3.96 and AP-4.20 (D–F). Close-up images of VPMpc (black outline) of each coronal section illustrated in D and E are redrawn in G and H, to illustrate the borders of densest projection from each injection in the thalamus. Blue trapezoid placed in G–H represent the region that was prepared for electron microscopy analysis in cases PBN Injection 2–5.

Figure 2.

Tracer injection site in a representative coronal section of the parabrachial nucleus (this case is illustrated as ‘PBN Injection 5’ in Figure 1). Adjacent sections that represented the core of the injection site were stained for myelin (A), for tracer (B) and for Nissl (C). Section in B corresponds to −9.48 mm AP coordinates (from bregma). The core of the tracer injection site (B) was confined in the waist region (w) of the PBN. Both Nissl and DAB stained sections were useful in identifying the nucleus borders (dashed lines in C). Myelin (D and J), tracer (E and H), and Nissl (F and I) stained adjacent sections through thalamus, containing anterograde tracer projection site in VPMpc. The asterisks mark blood vessel (bv) landmarks that appeared in adjacent Myelin/DAB/Nissl sections. Overlay of the corresponding section from Paxinos Rat Atlas (G and K) marks presumed borders of VPMpc (red outline). Neither Myelin nor Nissl stain outline a discernable border around VPMpc, although medio-laterally coursing myelinated axons (white arrowheads in J) corresponded to the dorsal border of VPMpc. Myelin stain clearly demarcates medial lemniscus (ml), which outlines VPMpc ventrally and tapers down about the midpoint of the mediolateral dimension of VPmpc. In addition, an additional fiber bundle (white arrows in J), represents a local anatomical landmark used for higher resolution imaging. In all cases, tracer-labeled parabrachiothalamic axons were found confined to the VPMpc borders extrapolated from corresponding coronal sections from Paxinos Rat Atlas. Furthermore, parabrachiothalamic projection site was in the ventral-lateral quadrant of these borders at each injection experiment. Panels E and F correspond to −3.7 mm AP coordinates. Scale bars: A = 300 µm (also applies to B–C); F = 600 µm (applies to D–G); J= 300 µm (applies to H–K). bv: blood vessel; other anatomical abbreviations are as indicated in the List of Abbreviations.

Figure 3.

A–B. High-resolution images of anterogradely-labeled parabrachiothalamic fibers displaying en passant swellings (black arrows) and large bouton-like structures (white arrows). Scale bars = 20 µm.

PBN tracer injections that produced dense anterograde labeling in VPMpc also resulted in retrogradely-filled cells in the rostral NTS, indicating accuracy of PBN injections (Figure 4 A–C). Tracer injections placed in VPMpc and surrounding areas led to retrograde filling of cells in the PBN, but not in NTS (Figure 4 D–L), confirming the lack of projections from NTS to thalamus in the rat.

Figure 4.

Retrograde transport of BDA after injections into PBN (A–C) and VPMpc (D–L). Myelin (A) and the adjacent ABC/DAB (B) treated NTS sections after a tracer injection into PBN waist region. NTS borders determined on Myelin section was transferred onto tracer-visualizes section. C. Higher magnification of section in B, illustrating retrogradely labeled neurons in RC subdivision of NTS. Nissl (D), Myelin (E) and ABC/DAB (F) stained near-adjacent sections through thalamus of a rat that received a large retrograde tracer injection into a region of thalamus, including VPMpc. Electrode tract is visible in D and E (black arrow in E), and the injection site in F. This injection resulted retrograde labeling in PBN (G–I), including the waist region (w). However no retrogradely labeled cells were encountered in NTS (J–L). Scale bar in A = 150 µm (applies to B); C =50 µm; D= 300 µm (applies to D–F); G = 300 µm (applies to G–I); J = 300 µm (applies to J–L). AP coordinate of the coronal sections corresponding to Nissl sample of each group is marked on the lower left corner of each Nissl section. Abbreviations are as provided in the List of Abbreviations. NTS subdivisions: rl = rostrolateral, rc = rostrocentral, v = ventral, m = medial. sp5 = spinal trigeminal tract, 4V = fourth ventricle.

Electron Microscopy

All brain sections that were processed for electron microscopy showed the same landmarks as those for light microscopy, including densely labeled fibers in the VPMpc and myelinated axon bundles of medial lemniscus. The ultrastructural morphology of labeled terminals and their postsynaptic targets within parabrachiothalamic projection zone (Figure 5, Figure 6 and Figure 7) were examined systematically in 4 brains, and all labeled profiles encountered (i.e., all parabrachiothalamic axon segments regardless if they displayed a synapse at the cross section) were photographed at high resolution (10,000x magnification, 2.75 nm pixel size). Qualitatively, parabrachiothalamic terminals (labeled profiles that contain vesicles and a synapse) were large and formed multiple, punctate synapses. These terminals often wrapped around large caliber dendrites that bore many dendritic protrusions. Many of these dendritic protrusions became immersed within labeled bouton mass (Figures 5A–B, 6A–B, 7C), hence these can also be named ‘terminal inclusions’. Labeled boutons were invariably filled with vesicles and mitochondria. In lightly-stained terminals, dense-core vesicles were detectable. The synapses formed by parabrachiothalamic terminals displayed postsynaptic densities of varying thickness, while a mesh of electron dense material was often present close to the postsynaptic density, giving an ‘underlined’ appearance to the synaptic zone (Figure 6A–B). While a large area of the lipid bilayer membranes of labeled terminals and postsynaptic dendrites remained apposed to each other, synaptic zones that form between these two elements were punctate. In 3D reconstructions, synaptic zones appeared as patchy, elongated strips that extended along the axis of the dendrite (Figure 7 B). Dendritic protrusions (or unlabeled terminal inclusions) almost always contained a synaptic zone (Figures 5B, 6A, 7C).

Figure 5.

Electron micrographs of anterogradely labeled parabrachiothalamic terminals in VPMpc. A. A large parabrachiothalamic terminal (t_PBN) almost completely surrounds a large caliber dendrite (d), which contains somatic organelles (i.e., rER or ribosomes; small black arrows). Labeled axon segments are often myelinated (ax_PBN). Synapses formed by unlabeled terminals (t) onto small caliber dendrites within the same field are marked with black arrowheads to aid in comparison of terminal and dendrite sizes. Postsynaptic dendrites extend protrusions (*) into the labeled terminal mass. B. Synaptic zones formed by labeled parabrachiothalamic terminals (black arrowhead) small in diameter, patchy and multiple. Postsynaptic dendrite protrusions (*) contain synaptic zones formed by the labeled terminals. Abbreviations: m: mitochondria; ax: axon. Scale bars: A = 1 µm; B = 0.25 µm.

Figure 6.

A–B. Anterogradely labeled parabrachiothalamic terminals (t_PBN) may also display smaller cross sections at the level of synaptic zones (black arrowheads). The presence of somatic organelles (small black arrows) indicates a proximal dendrite. Smaller caliber postsynaptic dendrites were also encountered (B). White arrowheads mark synapses formed by unlabeled terminals. C. Size-distribution histograms of dendrite caliber (measured as feret_min; see Methods for details) of dendrites postsynaptic to labeled parabrachiothalamic (black bars) vs. dendrites that are postsynaptic unlabeled terminals (white bars) within the same regions as the labeled terminals. D. Cross-section area distribution histograms of labeled parabrachiothalamic terminals (black bars) vs. unlabeled terminals (white bars) within the same regions as the labeled terminals. Note that the X axis is truncated at around 8 µm² to be able to illustrate distribution of unlabeled terminal areas. Scale bars: A = 1 µm; B = 0.5 µm.

Figure 7.

A-B. Three-dimensional reconstruction of a labeled parabrachiothalamic terminal (blue), making synapses (orange) onto a dendrite (grey). Small black arrows indicate the position of soma. C. Electron micrograph of sample planes (marked C and an arrow in A) of the reconstructed parabrachiothalamic terminal. Note that the synaptic zones from the labeled terminal are oriented parallel to the axis of the dendrite. Scale bar = 1 µm. D. 3D reconstruction of a VPMpc cell (blue) and anterogradely labeled parabrachiothalamic terminals (orange; PT) that extensively forms synapses onto the primary dendrite segments (d) of the cell. Abbreviations: m=mitochondrion; er = endoplasmic reticulum; r = ribosome; mye = myelin; t = terminal.

To quantify the morphological differences between parabrachiothalamic terminals and other inputs to the same region in VPMpc, terminal cross-section areas of 98 labeled and 637 unlabeled terminals encountered within the same photographs were compared (Figure 6 D). Unlabeled terminal cross-section area was 0.35 ± 0.01 µm² on average (±SEM), with a range of 2.8 µm² (median 0.30 µm²). In contrast, labeled terminals were 1.91 ± 0.31 µm² on average, with a range of 25.2 µm², and a median of 0.84 µm². The size difference between unlabeled and labeled terminal cross-section areas was statistically significant (Mann-Whitney U, p<0.001). The sample population included several large, labeled profiles: a completely vesicle-filled, multiple synapse-forming terminal bouton 25.18 µm² in its cross-section area; another over 10 µm², and two others over 7.5 µm² in cross-section area. Over 70% of all labeled terminals were larger than the average of unlabeled terminals, and over 60% of labeled terminals were larger than unlabeled terminal average area +1 SD. Only 2 out of 637 unlabeled terminals were larger than the average labeled terminals, and only 3% of unlabeled terminals were larger than the median size of labeled terminals (0.84 µm²). Because anterograde tracer injections are not expected to fill all axons projecting to the region and some contralateral parabrachiothalamic projections exist, the large unlabeled terminals may in fact be parabrachiothalamic terminals. The striking size difference of parabrachiothalamic terminals from the rest of the VPMpc terminals suggests that size criterion alone is sufficient to identify parabrachiothalamic terminals.

In quantifying the morphological properties of dendrites postsynaptic to labeled parabrachiothalamic terminals, we found that large caliber dendrites are the main target of labeled terminals (Figure 6 C). The analyzed dataset included 86 dendrites postsynaptic to labeled terminals and 447 dendrites postsynaptic to unlabeled terminals. Dendrite cross-sections that were not contained in their entirety within image frames, or those that did not display unambiguous borders were not included in the analysis. Dendrites postsynaptic to labeled terminals were statistically larger than those postsynaptic to all other terminals (Mann Whitney U test, p<0.001). Labeled terminals were not the exclusive input to large dendrites. Although parabrachiothalamic terminals occupied an overwhelming area surrounding the dendrite, unlabeled terminals also synapsed with the same dendrite as the labeled parabrachiothalamic terminals (e.g., white arrowhead within postsynaptic dendrite in Figure 6A).

The large diameter of dendrites postsynaptic to parabrachiothalamic terminals suggested that these might be proximal regions of the dendrites. In opportune sections, we were able to follow such postsynaptic dendrites into serially adjacent sections to confirm that the primary target of parabrachiothalamic terminals is proximal dendritic segments (Figure 7). Furthermore, many postsynaptic dendrites contained perikaryal organelles (e.g., ribosomes and endoplasmic reticulum) (Figure 6A). 3D reconstruction also revealed that large parabrachiothalamic boutons wrapped around the primary dendrites as the dendrites emerged from the somata, forming synaptic zones of massive total size between a single axon and the postsynaptic somata (Figure 7B).

To quantify the frequency of parabrachiothalamic synapses in VPMpc, we constructed composites of continuous areas from three brains, and examined an area of 6016 µm² within the projection region. Within this restricted area we found 666 synaptic terminals, and only 13 of those contained BDA label. Thus, parabrachiothalamic terminals provided only 1.99% of all synaptic boutons in VPMpc. However, even though the number of terminal boutons was small relative to all other inputs to gustatory thalamus, parabrachiothalamic terminals, which form large and multiple synaptic zones onto their postsynaptic targets, provided 15% of total surface area of synapse active zones in the thalamus circuitry, as measured by total synapse length of labeled and unlabeled terminals within the composite area dataset (28.5 µm synapse length from 13 parabrachiothalamic terminals versus 183.3 µm synapse length from 653 non-parabrachiothalamic terminals).

Corticothalamic Input to VPMpc

Anterograde tracer injections into IC led to visualization of anterogradely-filled fibers in ipsilateral VPMpc (Figure 8). A few retrogradely-filled cells were also present within VPMpc borders (Figure 8 B–C). This confirmed that injections included IC regions that were targeted by VPMpc, and validated the presence of reciprocal connectivity between VPMpc and IC. Serial sections of thalamus processed for DAB and myelin staining confirmed that there was a dense network of anterogradely labeled corticothalamic fibers, located in an ovoid area dorsal and medial to the medial lemniscus (Figure 8B). Anterogradely-filled corticothalamic fibers were fine, numerous and formed en passant swellings as well as short side arms terminating with a small bouton (Figure 8C). Anterogradely-filled boutons were also present in close proximity to retrogradely-filled cell bodies.

Figure 8.

A. Insular cortex injection site from 3 animals. Approximate AP coronal positions of sections are matched to rat brain atlas (Paxinos and Watson, 1989), and marked at the lower corners of each section. Putative borders of AI, DI and GI were also transferred. All three injection site included putative DI and GI. B. Tracer labeling found in VPMpc region after the bottom injection depicted in panel A. Both anterogradely-filled fibers and retrogradely-filled cells (arrows) in VPMpc were found in VPMpc. C. Higher magnification image of the region in VPMpc containing retrogradely-filled thalamocortical cells (arrows). Labeled fibers are fine and bear small swellings. Scale bar in A = 1 mm; B = 500 µm; C = 100 µm; and D = 50 µm. Abbreviations: CPu= caudate putamen; cp= choroid plexus; rf= rhinal fissure; mcer= middle cerebral artery; pir=piriform cortex; lo= lateral olfactory tract.

Thalamic sections from three brains with tracer-visualized corticothalamic projections in the VPMpc were processed for electron microscopy. VPMpc landmarks and tracer label were confirmed by comparisons of myelin stained series and rat brain atlas maps with resin-embedded thick tissue sections under the light microscope. Resin-embedded blocks were trimmed to contain medial VPMpc. In each brain, at least 2700 µm² of tissue (total of 10936 µm²) was imaged and contained a total of 372 labeled terminals. The corticothalamic terminals were small, vesicle-dense, and characteristically lacked mitochondria (Figure 9). Terminals formed single, asymmetric synaptic zones onto postsynaptic dendritic segments. Unlike parabrachiothalamic terminals, labeled corticothalamic terminals contained no dendritic protrusions, (compare Figures 5–6 and 9). Anterogradely-labeled corticothalamic terminals were located in the same region as retrogradely-filled VPMpc cells and dendrites. Synaptic connections between anterogradely- and retrogradely-labeled profiles were observed, suggesting that thalamic neurons that project to IC receive direct synaptic feedback from the same region. Furthermore, retrogradely-filled dendrites received synapses from large unlabeled terminal boutons (that is, from parabrachiothalamic terminals, identified by the size criterion), strongly suggesting that thalamocortical projection neurons receive monosynaptic parabrachiothalamic input. It should be noted that there are currently no VPMpc single cell axon reconstructions in the literature. Therefore, it is not known whether thalamocortical axons give off axon collaterals within VPMpc. If they do, these may also constitute a source of labeled terminal boutons found in VPMpc after IC injections.

Figure 9.

Electron micrographs of anterogradely filled corticothalamic terminals in VPMpc. A–B. Anterogradely labeled corticothalamic terminals (t_CT) making synapses (arrowhead) onto small, unlabeled dendrites (d). C–D. Labeled corticothalamic terminals forming feedback synapses onto retrogradely labeled dendrites of thalamocortical neurons (d_TC). Other synapses (large arrows) made by unlabeled terminals (t) onto unlabeled dendrites are also shown in the graphs. Asterisks indicate an example of labeled corticothalamic terminals with no synaptic contact; such cross-sections are not included in morphometric analyses. Scale bars A = 1 µm (applies to B–C) and D = 1 µm. Abbreviations: m = mitochondria; ax = axon; som = soma.

The cross-sectional areas of corticothalamic terminals (n=372) were significantly smaller than the parabrachiothalamic terminals (0.29 ± 0.22 µm² vs. 1.91 ± 3.1 µm²; mean ± SD; Mann-Whitney U, p< 0.001, Figure 10). Corticothalamic terminals were also statistically smaller than the set of all non-parabrachiothalamic terminals (i.e., unlabeled terminals in parabrachiothalamic experiments; 0.29 ± 0.22 µm² vs. 0.35 ± 0.22 µm²; Mann-Whitney U, p<0.001). Further, we found that unlike parabrachiothalamic terminals, the primary target of corticothalamic terminals was small caliber dendrites (0.74 ± 0.29 µm vs. 1.85 ± 0.90 µm, Mann-Whitney U test, p< 0.001). Dendrites that were retrogradely filled with tracer injected into the insular cortex were also postsynaptic to corticothalamic terminals. No axosomatic synapses were encountered between corticothalamic terminals and retrogradely-filled VPMpc cells.

Figure 10.

A. Composite electron micrographs in VPMpc, showing an anterogradely labeled corticothalamic terminal (t_IC) that makes a synapse (arrowhead) onto a small, unlabeled dendrite (d). A large terminal (t_PBN) in the same field makes a synapse onto a retrogradely-labeled dendrite (i.e., thalamocortical neuron). B–C. Higher magnification views of the synapses shown in A. D. Size distribution histograms of the calibers of dendrites that are post-synaptic to corticothalamic terminals (open bars) and parabrachiothalamic terminals (black bars). (E) Size distribution histograms of cross-sectional areas of corticothalamic terminals (open bars) and parabrachiothalamic terminals (black bars). E: Scale bars: A = 500 nm; B = 500 nm (applies to C).

Calcitonin gene-related peptide (CGRP) terminals in VPMpc

A selective projection into medial VPMpc carries CGRP (Yasui et al., 1989), and these projections are proposed to originate in the PBN (Dobolyi et al., 2005). Furthermore, the EM morphology of CGRP positive fibers in VPMpc (Williamson and Ralston, 1993) are qualitatively similar to the axons anterogradely filled from waist region of PBN in our study. In order to explore the possibility whether CGRP immunolabeling could be used as a marker for parabrachiothalamic terminals in VPMpc, we have compared the morphometry of terminals labeled with CGRP immunolabeling or anterograde tracer labeling from waist region of PBN. The CGRP immunolabeling resulted in dense fiber staining in the regions dorsal and medial to medial lemniscus (Figure 11). CGRP fibers were particularly dense in VPMpc, while CGRP-labeled cell bodies were noted in the dorsal subparafascicular thalamic nuclei and A11 dopaminergic cells, located just medial to VPMpc (Figure 11), confirming previous results characterizing CGRP labeling in thalamus (Orazzo et al., 1993; Coolen et al., 2003). The medial and dorsal portions of VPMpc contained relatively denser labeling. Under high magnification light microscopy, CGRP-labeled fibers bore bouton-like swellings of various sizes. In general, CGRP-labeled fibers were fine with varicosities, while large fiber swellings, which resembled anterogradely-labeled parabrachiothalamic fibers, were also observed.

Figure 11.

A. Immunolabeling with CGRP antibody leads to selectively dark labeling in VPMpc. B. High magnification examination reveals punctate, possibly terminal labeling. C. A few CGRP labeled cells are observed medial to VPMpc, in A11 dopamine cell group (white arrows in A and C), as described before (Orazzo et al., 1993; Kresse et al., 1995). Note that the area of dense CGRP labeling in VPMpc spreads dorsoventrally in medial VPMpc, in contrast to parabrachiothalamic fibers from waist region that we observed to be confined in a ventral position. Scale bars are 600 µm in A and B, and 100 µm in C.

In preparing CGRP immunostained VPMpc sections for electron microscope, the boundaries of VPMpc were confirmed by comparing resin-embedded sections to adjacent Nissl and Myelin stained sections, as described for tracer experiments. A total of 365 CGRP labeled synaptic terminals within 16586 µm² of imaged tissue were analyzed (Figure 12). CGRP labeling filled vesicle-containing profiles in their entirety, although the label was particularly dense over dense-core vesicles (Figure 12A; white arrows). As shown previously (Williamson and Ralston, 1993), there were at least two qualitatively different populations of CGRP terminals in the VPMpc. The first population was comprised of relatively large terminals filled with mitochondria and dense-core vesicles. Postsynaptic dendrites to these larger terminals were more likely to be large in diameter, and they enveloped dendritic protrusions that received synaptic input from those terminals (Figure 12 B–C). With these morphological characteristics, large CGRP terminals resembled anterogradely-labeled parabrachiothalamic terminals (Figure 12A–D, compare with Figure 5). However, not all terminals with parabrachiothalamic-like morphology were CGRP-positive. Large caliber dendrites that received a synapse from CGRP-immunoreactive terminals also received other synapses from unlabeled, large and dense-core vesicle containing terminals (i.e., terminals that also display parabrachiothalamic terminal morphology; Figure 12A). This indicates that CGRP may be present only in a subpopulation of parabrachiothalamic terminals. A second population consisted of smaller terminals (Figure 11D) filled with vesicles and dense-core vesicles. These contained fewer, if any mitochondria, and usually synapsed on smaller caliber dendrites (Figure 12D). No appendages or protrusion formations were observed at those dendrites postsynaptic to smaller CGRP terminals.

Figure 12.

(A–D) Electron micrographs of CGRP-labeled terminals in VPMpc, illustrating large (A–C) and small (D) CGRP-labeled terminals making synapses (arrowheads) onto postsynaptic dendrites (d). A soma is labeled in C showing close position to a labeled large CGRP terminal. The CGRP-DAB was often concentrated on dense-core vesicles (white arrows), yet not all terminals with dense-core vesicles were labeled with CGRP (white arrow in unlabeled terminal, t, in A). (E) The comparison between area distribution histograms of CGRP (dashed line) and parabrachiothalamic (solid line) terminals. Scale bars: A, C, and D = 1 µm; B = 2 µm.

For quantitative comparison of CGRP terminals and anterogradely labeled parabrachiothalamic terminals in VPMpc, all labeled profiles that displayed a synapse at the cross-section were measured. The average terminal area of all CGRP-ir terminals sampled was 1.51 ± 1.44 µm² (mean ± SD) with a range of 16.3 µm² (Figure 12). Although the terminal area frequency histograms of CGRP-ir and parabrachiothalamic terminals were partially overlapping, statistical comparison revealed a significant difference between these populations of terminals (p<0.001, Mann-Whitney U). Furthermore, even though large terminals were present, only about 8% of all CGRP labeled terminals were larger than the average size of non-parabrachiothalamic terminals. As outlined above, over 72% of all parabrachiothalamic terminals were larger than this same criterion. Thus, a great majority of CGRP terminals are more likely to be non-parabrachiothalamic, and CGRP-ir alone cannot be used as a marker for parabrachiothalamic terminals.

Discussion

The current study examined the morphology and connectivity of two main inputs to the gustatory thalamus (VPMpc) in the rat: the primary input from the waist region of the PBN and the feedback projections from the insular cortex. The results demonstrated that parabrachiothalamic and corticothalamic inputs are morphologically distinct from each other, and they target different regions of dendrites (Figure 13). While parabrachiothalamic terminals are large, sparse and highly selective for proximal dendritic segments of VPMpc relay cells, corticothalamic terminals are small, numerous and target distal dendrites. Morphometric properties suggest that size criterion alone is sufficient to reliably identify parabrachiothalamic terminals at the electron microscopic level. The results also provide evidence that while this peptide is not exclusive to parabrachiothalamic input originating from the waist region, CGRP is contained in a subset of parabrachiothalamic terminals.

Figure 13.

Summary illustration of inputs onto thalamocortical projection cells of gustatory thalamic nucleus VPMpc. The parabrachiothalamic axons terminate on initial segments of primary dendrites via large boutons, which also contain CGRP. Feedback corticothalamic axons terminate on distal dendrites, via small terminals. Inhibitory synapses, presumably from local interneurons and thalamic reticular nucleus, also synapse on relay cells. Morphological selectivity of these inputs should be examined with specific markers. Whether the modulatory inputs from other brainstem nuclei, releasing acetylcholine, histamine, dopamine or serotonin, are present in VPMpc is not known.

Considerations on Identification of Taste Thalamus

The extent of the ipsilateral labeling observed in the gustatory thalamus is consistent with previous studies (Norgren and Leonard, 1973; Norgren, 1974; 1976), which documented brainstem projections using lesion and tract tracing approaches. Our results are also consistent with findings in studies using anterograde (Karimnamazi and Travers, 1998; Bester et al., 1999) or retrograde (Krout and Loewy, 2000) tract tracing to indicate that the gustatory region of the VPMpc is restricted to the medial portion. Cytoarchitectonic landmarks such as the fasciculus retroflexus, medial lemniscus, and mammilothalamic tract reliably mark the anterior-posterior zone along VPMpc containing parabrachiothalamic input, although actual borders of the VPMpc are ambiguous with Nissl or myelin stains. The borders drawn in the current study (Figure 1 and 2) are consistent with those drawn by Bester et al (Bester et al., 1999) and Paxinos and Watson (Paxinos and Watson, 2007). Similar to our light microscopy observations, Bester et al (1999) also pointed out that large varicosities of PBN projections clustering in a restricted pattern were found solely inside the VPMpc. These authors have also indicated that the PBN subnuclei that gave projections with those large varicosities were in the main parabrachial gustatory area, including waist region and medial subnuclei. Despite our efforts with cytoarchitectonic or immunohistochemical markers, anterograde labeling from waist region of PBN remains the only unambiguous means to identify the projection site and the terminals of parabrachiothalamic axons.

Parabrachiothalamic Input to VPMpc

The VPMpc contains neurons that relay gustatory activity to the insular cortex. The present analysis of parabrachial terminal areas and postsynaptic dendrite caliber indicates that PBN projections provide a strong excitation on thalamic relay cells in the VPMpc, analogous to the role of a primary, or driver, input axons observed in other primary sensory nuclei of thalamus (e.g., dorsal lateral geniculate nucleus, medial geniculate nucleus and ventral posterior nucleus). Also characteristic for those driver inputs, parabrachiothalamic synapses constitute only a small proportion of all synapses in VPMpc. In contrast, 3D reconstructions of the parabrachial terminals in VPMpc demonstrated extensive interactions with the proximal dendrites of presumed relay cells in a pattern that is unmatched in any other primary sensory thalamic nuclei (Figure 7). Although in visual, auditory and somatosensory thalamic nuclei, principal sensory relay cells also receive their primary input at large caliber, presumably proximal, dendrite segments (Liu et al., 1995; Erisir et al., 1998; Bartlett et al., 2000; Guillery and Sherman, 2002), highly exclusive and invasive targeting of initial dendrite segments of gustatory thalamic cells is unique to the gustatory system.

The functional implications of this innervation selectivity, and what it implies about the elaboration or the modulation of gustatory activity within thalamus are not immediately forthcoming. Thalamic relay cells in other sensory systems transmit sensory input to cortex without much refinement in receptive field properties of their input cells or integration of multiple sensory features, although many converging inputs from modulatory neurotransmitter systems may gate the transmission (Sherman, 2007). The axons from PBN neurons on VPMpc relay cells are situated to exert a strong excitation on individual neurons, possibly providing an excitatory signal capable of overwhelming other modulatory or integrated inputs. It is also worth noting that a major property of sensory thalamic nuclei, namely glomerular arrangements and triads, are missing from the gustatory thalamus. The triads, commonly located within glomeruli provide localized feedback inhibition to relay cells, and may mediate the timing of transmission in temporally sensitive, non-chemical sensory modalities (Sherman and Guillery, 2001; Saul, 2008). The present study demonstrated extensive synapse formation between large parabrachiothalamic terminal boutons and thalamocortical cells of VPMpc. However, it is not known whether all parabrachiothalamic synapses on an individual relay cell belong to a single axon, or convergence of multiple parabrachiothalamic axons onto single relays cells occurs in gustatory thalamus.

Corticothalamic input to VPMpc

As shown previously, tracer injections centered in insular cortex led to anterograde and retrograde labeling in VPMpc (Nakashima et al., 2000; Lundy and Norgren, 2004). Anterogradely labeled axons formed synapses on retrogradely labeled dendrites, providing evidence for a corticothalamic feedback loop in the gustatory system. These corticothalamic terminals formed small boutons with asymmetric synapses, contacting distal dendrites of thalamic relay cells, which is also consistent with the glutamatergic feedback modulation that is encountered in all other sensory thalamic nuclei.

Accumulating evidence suggest that corticothalamic feedback in sensory thalamic nuclei may serve two main functions: enhancing receptive field properties of thalamocortical cells, and altering or strengthening the temporal patterns of activity in thalamic neuron ensembles (reviewed in Sillito and Jones, 2002; Briggs and Usrey, 2008). Relay cell receptive field refinement by corticothalamic activity has been reported for all non-chemical sensory systems (Sillito and Jones, 2002; Li and Ebner, 2007; Zhang and Yan, 2008), where topographic organization of receptive fields can selectively be modified by discretely reciprocal projection patterns of corticothalamic axons. Anatomical projections from insular cortex are as robust as in other sensory nuclei, however, the function of corticothalamic feedback in gustatory thalamus is not forthcoming, especially in the light that the gustatory neurons are broadly tuned and a topographic representation is not evident.

CGRP Terminals in VPMpc

Past studies have shown that VPMpc is heavily innervated by CGRP-ir fibers, and the primary source of CGRP innervation is ventral lateral, external medial and external lateral region of PBN (Yasui et al., 1991; Williamson and Ralston, 1993; de Lacalle and Saper, 2000; Dobolyi et al., 2005). Particularly, using retrograde tracer injections that included VPMpc in thalamus, Yasui et al (Yasui et al., 1989) have demonstrated that while many retrogradely labeled cells in visceral sensory divisions of PBN contained CGRP, a small number of waist region neurons have also contained this peptide. These authors also provided evidence that CGRP-ir neurons in the external medial parabrachial subnucleus sent fibers prominently to contralateral VPMpc region. In the present study, comparing ultrastructural morphological properties of CGRP-ir terminals in VPMpc to anterogradely labeled parabrachiothalamic terminals, we documented the similarities between a subpopulation of CGRP-ir terminals and parabrachiothalamic terminals, namely large boutons containing dense-core vesicles. Furthermore, we have noted that not all CGRP-ir terminals in VPMpc displayed parabrachiothalamic morphology, and not all parabrachiothalamic terminals (as defined by a size criterion) contained CGRP. In addition to eliminating the possibility of using CGRP as a marker for PBN-waist region inputs to VPMpc, these observations have two important implications: First, PBN-waist region inputs to VPMpc are not molecularly uniform; there are at least two separate subpopulations, each with a different chemical phenotype, even though they may be morphologically uniform. The functional significance of CGRP in these chemical sensory axons is not known. Second, given that the origin of small CGRP terminals is not ipsilateral PBN-waist projections (current findings), and brainstem is the most likely source of all CGRP labeling in VPMpc (because transections at superior cerebral peduncle eliminate all CGRP labeling in fibers located in VPMpc; Dobolyi et al., 2005), the external medial parabrachial nucleus, which projects to contralateral VPMpc, may in fact be the origin of small CGRP-ir terminals there. In the current study we have not targeted external medial subdivisions for our injections, nor analyzed contralateral projections. Future experiments should confirm that contralateral parabrachiothalamic axons, which may bring the visceral sensory information to VPMpc are morphologically distinct from ipsilateral, pontine taste axons.

The possibility that gustatory information is modulated by visceral sensory input in rats, as well as in primates, was suggested previously (Yasui et al., 1989; de Lacalle and Saper, 2000). Our findings provide inferential support for this idea, by demonstrating the anatomical evidence for a substantial CGRP input that originates from non-taste subdivisions of parabrachial nucleus. Studying the functional role of CGRP/glutamate co-release in VPMpc will add to the understanding of intrinsic modulation of sensory processing in gustatory thalamus.

Functional implications for VPMpc circuitry

Response properties of VPMpc neurons have been characterized in few studies (Scott and Erickson, 1971; Scott and Yalowitz, 1978; Nomura and Ogawa, 1985; Ogawa and Nomura, 1988; Verhagen et al., 2003) . The presence of three major neuron types, which closely mimic the physiological types encountered in NTS, were proposed: NaCl-, HCl- and sucrose-oriented neurons (Verhagen et al., 2003). A small population of quinine-oriented neurons was also encountered. Response properties of VPMpc cells are consistent with the role of VPMpc as a relay nucleus of taste information: similar to gustatory responses in NTS and PBN cells, VPMpc neurons are broadly tuned, yet tastant-selective (Verhagen et al., 2005). Consistent with the idea of existence of functional neuron types at this relay, neuron groups with selectivity to different tastants respond to selective sodium blocks applied in oral cavity, similar to those in NTS and PBN (Verhagen et al., 2005). Furthermore, the breadth of VPMpc cell tastants-tuning and their overall spontaneous firing rates are intermediate between NTS and cortex cell response patterns. Thalamocortical cells are better able to differentiate tastants than the neurons in the lower gustatory nuclei (Scott and Erickson, 1971), indicating that sharpening of gustatory receptive fields may occur via thalamic gustatory processing. This suggests that some aspects of the tuning properties are refined as taste information is transmitted along the thalamocortical pathway, perhaps utilizing the temporal information to distinguish responses of broadly tuned neurons for stimuli in the same category (Verhagen et al., 2003; Carleton et al., 2010). The synaptic selectivity of parabrachiothalamic axons revealed in the current study also indicates that individual brainstem axons may be sufficient to drive spiking activity in relay neurons. There are no reports available to indicate anatomical phenotypes of neurons in correlation with their response selectivity, nor to identify synaptic connectivity patterns of axons that preserve distinct stimulus properties along the NTS-PBN-VPMpc axis. Our results that imply at least two distinct populations of PBN axons in VPMpc (those that do and do not contain CGRP), may be indicative of a structural basis for differential synaptic circuitry involving functional neuron types in VPMpc.

Chemical vs. non-chemical thalamus

Sensory pathways in mammals follow a well-described hierarchical pattern: Each specific sensory input arrives to corresponding primary cortical area via a thalamic relay nucleus, which is defined as the first-order relay (Sherman, 2007). The primary cortical area, in turn, projects to a higher-order thalamic relay nucleus, which is the source of the primary (or driver) input to secondary sensory cortical areas. The driver inputs to both the first-order and the higher order thalamic relay nuclei are distinguished by a long list of features, including thick axons bearing large terminal boutons and providing surprisingly small percentage of the synapses on relay cells, type of glutamate receptors (only ionotropic), and provision of receptive field properties maintained in the receptor organ or the specific brainstem nuclei. While light microcopy studies have demonstrated that parabrachiothalamic axons project densely to the VPMpc and bear large swellings (Bester et al., 1999), and indicated that the basic connectivity of gustatory thalamic relay may fit the general pattern of all other first order sensory thalamic nuclei, many questions have remained regarding their synaptic properties, which could only be answered at ultrastructural resolution.

The current study examined the ultrastructure of the VPMpc with respect to its main ascending input from PBN, and its main feedback projection from the insular cortex in the adult rat, and revealed the similarities, and unique properties, of gustatory thalamic circuitry with other first-order thalamic relay nuclei. First, the demonstration that parabrachiothalamic terminal boutons are sparse, and that they constitute the largest terminal boutons in taste thalamus, confirms that VPMpc fits the definition of first-order thalamic relay nucleus. Whether there is a higher-order thalamic relay for taste system remains unknown. Second, morphological properties of parabrachiothalamic axons leave no ambiguity in that this input should be sufficient to drive the excitation that will be relayed to IC, in a manner analogous to primary ‘driver’ inputs to visual, auditory and somatosensory thalamic nuclei. Corticothalamic input morphology and targeting properties are also in agreement with known patterns of innervation of sensory thalamic nuclei. However, a hallmark structural component of these non-chemical thalamic nuclei are missing from VPMpc; glomerular structures, that is, glia-encapsulated triadic arrangements among primary and inhibitory axons and relay cell dendrites, are not encountered in VPMpc. In contrast, glomerular arrangements are common in NTS (May et al., 2007; Wang et al., 2012), suggesting that at least some of the specialized processing that are attributed to non-chemical thalamic nuclei circuitry may take place in brainstem nuclei in taste pathways. Lastly, we would like to point out to the possibility that the similarities and unique properties of VPMpc circuitry unveiled in the current study may be peculiar to rodent species. As it has been documented in detail (Beckstead et al., 1980; Pritchard et al., 2000; Lundy and Norgren, 2004), the parabrachiothalamic input to VPMpc replaces the direct input from NTS that is found in primates. This evolutionary divergence is unique not only in considering rodents versus primates, but also in considering chemical versus non-chemical senses in rodents. That is, unlike rodent taste system, brainstem-to-thalamus pathways persist in rodent auditory and somatosensory systems. Therefore, the morphological properties of parabrachiothalamic axons and the VPMpc circuitry we have demonstrated in current study may be unique to rodents as well. Examination of primate thalamus morphology and circuitry may allow direct comparisons of chemical and non-chemical thalamic circuitries and function in primates, including humans.

Conclusions

The current study is a first step in understanding the structural organization of inputs to neurons in gustatory thalamus, and thus the anatomical basis for gustatory processing along the brainstem-thalamus-cortex pathway. It also underlines the unique properties of gustatory thalamus synaptic organization in comparison with other sensory pathways in rodents.

Abbreviations

3V

third ventricle

4V

fourth ventricle

AI

agranular insular cortex

CGPn

central gray of the pons

CGRP

calcitonin gene-related peptide

cp

cerebral peduncle

CPu

caudate / putamen

CT

corticothalamic

DI

dysgranular insular cortex

ec

external capsule

el

external lateral parabrachial nucleus

em

external medial parabrachial nucleus

f

fornix

fr

fasciculus retroflexus

GI

granular insular cortex

IC

insular cortex

ic

internal capsule

m

medial parabrachial nucleus

Me5

mesencephalic trigeminal nucleus

me5

mesencephalic trigeminal tract

ml

medial lemniscus

mt

mammillothalamic tract

NTS

nucleus tractus solitarii

PBN

parabrachial nucleus

PT

parabrachiothalamic

scp

superior cerebral peduncle

sp5

spinal trigeminal tract

spf

subparafascicular thalamic nucleus

TRN

thalamic reticular nucleus

vl

lateral parabrachial nucleus ventral

VPM

ventral posteromedial thalamic nucleus

VPMpc

ventroposterior medialis parvocellularis nucleus

w

waist region of the parabrachial nucleus