SINGLE-AXON TRACING OF THE
CORTICOSUBTHALAMIC HYPERDIRECT
PATHWAY IN PRIMATES
Dymka Coudé, André Parent and Martin Parent*
Laboratoire de Neurobiologie
Centre de recherche Université Laval CERVO 2601, Chemin de la Canardière, Local F-6500
Beauport, Québec, Canada, G1J 2G3
Abbreviated title: The hyperdirect pathway in monkeys
Key words: basal ganglia, subthalamic nucleus, primary motor cortex, single-axon reconstruction, corticosubthalamic projections, monkeys.
*Correspondence to: Martin Parent, Ph.D.
CERVO Brain Research Centre 2601, Canardière, F-6500 Quebec City (Quebec) Canada, G1J 2G3
Tel: (418) 663-5747 (ext 6736) Fax: (418) 663-8756
E-mail: martin.parent@fmed.ulaval.ca
Grant sponsor: The study was supported by research grants from the Canadian Institutes of Health Research (CIHR MOP-153068) and the Natural Sciences and Engineering Research Council of Canada (NSERC 2018-06264 and 2018-522690) to M.P. who also benefited from of a Junior II career award from the Fonds de Recherche du Québec - Santé (FRQ-S). D.C. was recipient of MSc fellowship from FRQ-S. The authors have no conflict of interest to declare.
ABSTRACT
Individual axons that form the hyperdirect pathway in Macaca fascicularis were visualized following microiontophoretic injections of biotinylated dextran amine in layer V of the primary motor cortex (M1). Twenty-eight singly-labeled axons were reconstructed in 3D from serial sections. The M1 innervation of the subthalamic nucleus (STN) arises essentially from collaterals of long-ranged corticofugal axons en route to lower brainstem regions. Typically, after leaving M1, these large caliber axons (2-3 µm) enter the internal capsule and travel between caudate nucleus and putamen without providing any collateral to the striatum. More ventrally, they emit a thin collateral (0.5-1.5 µm) that runs lateromedially within the dorsal region of the STN, providing boutons en passant in the sensorimotor territory of the nucleus. In some cases, the medial tip of the collateral enters the lenticular fasciculus dorsally and yields a few beaded axonal branches in the zona incerta. In other cases, the collateral runs caudally and innervates the ventrolateral region of the red nucleus where large axon varicosities (up to 1.7 µm in diameter) are observed, many displaying perisomatic arrangements. Our ultrastructural analysis reveals a high synaptic incidence (141%) of cortical VGluT1-immunoreactive axon varicosities on distal dendrites of STN neurons, and on various afferent axons. Our single-axon reconstructions demonstrate that the so-called hyperdirect pathway derives essentially from collaterals of long-ranged corticofugal axons that are rarely exclusively devoted to the STN, as they also innervate the red nucleus and/or the zona incerta.
INTRODUCTION
The subthalamic nucleus (STN) occupies a pivotal position in the functional organization of the basal ganglia (BG). Being the only glutamatergic component of the BG, the STN is often viewed as a major driving force of this set of subcortical structures (Parent & Hazrati, 1995b; Mink, 1996; DeLong & Wichmann, 2007; Gerfen & Bolam, 2017). Our previous single-axon tracing studies in primates have revealed that STN neurons are endowed with a highly collateralized axon, which allows them to exert a direct excitatory influence on the two major output structures of the BG, namely, the internal pallidum and the substantia nigra pars reticulata, as well as on the external pallidum and the striatum (Sato et al., 2000). Besides these multiple efferent projections, the STN receives afferents from different brain regions (Parent & Hazrati, 1995b) and is often considered an important input station of the BG, as it acts in parallel with the striatum, the main entry structure, to directly collect and integrate cortical information (Nambu et al., 1996; Nambu et al., 2002). Although significant glutamatergic projections arising from the caudal intralaminar thalamic nuclei exist (Sadikot et al., 1992), the STN receives most of its glutamate-mediated excitatory input from neurons located in layer V of the primary motor cortex (M1) and the supplementary motor area (Hartmann-von Monakow et al., 1978; Rouzaire-Dubois & Scarnati, 1985; Canteras et al., 1990; Nambu et al., 1996), as well as from other prefrontal cortical regions (Haynes & Haber, 2013). This projection system, which has been called the hyperdirect pathway (Nambu et al., 1996), allows cortical information to be directly relayed to the STN, without being processed by striatal neurons (Steiner & Tseng, 2016). Because it bypasses the striatum, the hyperdirect pathway is seen as a route whereby cortical information can influence the BG output structures (the internal pallidum and the substantia nigra pars reticulata) with shorter latencies than through the so-called direct and indirect striatofugal pathways.
The abnormal activity of STN neurons in Parkinson's disease (PD) is believed to be associated with motor symptoms that characterize this neurodegenerative disorder, including rigidity, bradykinesia, resting tremors and postural instability (Galvan & Wichmann, 2008; Kita & Kita, 2011). For this reason, the STN has become a target of choice for chronic deep
(Wichmann et al., 2017). Moreover, antidromic activation of the motor cortex through the hyperdirect pathway following DBS of the STN is thought to play a key role in the mediation of DBS therapeutic effects in PD patients (Wichmann & DeLong, 2016; Anderson et al., 2018). Therefore, in the hope to reach a better understanding of the anatomical and functional organization of the primate BG, while obtaining new insights into the cellular mechanisms of DBS, we undertook, for the very first time, a detailed single-axon tracing study of the corticosubthalamic hyperdirect pathway in cynomolgus monkeys. Our main findings indicate that the hyperdirect pathway derives essentially from collaterals of long-ranged corticofugal axons, en route to lower brainstem regions.
MATERIALS AND METHODS
Injection procedures
A total of five adult cynomolgus monkeys (Macaca fascicularis) of both sexes, with a body weight that ranged from 3-4 kg, were used in the present study. All experimental procedures were approved by the Comité de Protection des Animaux de l’Université Laval, in accordance with the Canadian Council on Animal Care’s Guide to the Care and Use of Experimental Animals (Ed2). Maximum efforts were made to minimize the number of animals used. The animals were first anesthetized with ketamine (75 mg/kg) plus xylazine (5 mg/kg) and their head placed in a specifically designed stereotaxic apparatus. They were then maintained under propofol (10 mg/ml, i.v.) anesthesia, while microiontophoretic injections of biotin dextran amine (BDA 10 000 MW, Thermo Fisher Scientific, Waltham, MA, Catalog no. D1956) were being made bilaterally in the forelimb area of the primary motor cortex (M1), as identified on physiological maps previously established (Woolsey, 1958; Kwan et al., 1978; Stepniewska et al., 1993). Because corticosubthalamic projections were found to originate mainly from layer V in primates (Nambu et al., 1996), BDA injections were centered upon this cortical layer. In some cases, neurons located in supragranular layers were also labeled. However, these neurons were very few in number, their staining was weak, and their axon could not be traced outside the cerebral cortex. They most likely represent neurons
whose neuronal processes (axons or dendrites) have transported a small amount of tracer from the periphery of the main injection loci.
Four injections were made in each animal, two on each side of the brain. We used the stereotaxic coordinates of the atlas of Szabo and Cowan (Szabo & Cowan, 1984) and microiontophoretic labeling was carried out with glass micropipettes (tip diameter 2-3 µm) filled with a solution of potassium acetate (0.5M) plus 2% BDA. These electrodes had impedance ranging between 10-15 MΩ and were used to monitor the extracellular activity of the neuronal populations encountered during the penetration of the micropipette. Layer V of M1 was easily recognizable by the characteristic bursting firing pattern of its neurons under propofol anesthesia (Fig. 1C). Once in the chosen target, the micropipette was connected to a high compliance iontophoresis device (NeuroData) and the tracer was injected by passing positive current pulses of 350 nA (1s on/ 1s off) for 25 min.
Tracer visualization and cytochrome oxidase staining
After a survival period of 8-10 days, the animals were deeply anesthetized with sodium pentobarbital and perfused transcardially with 1 liter of saline solution (0.9%) followed by 2 liters of a fixative solution containing 4% paraformaldehyde (PFA) in phosphate buffer (PB, 0.1M, pH 7.4) and 1 liter of 10% sucrose solution in PB. The brains were dissected out and placed in a cryoprotective solution composed of 1/3 PFA (4% solution in PB) and 2/3 sucrose (30% solution in PB) for 24h at 4°C. Sagittal (1 monkey) or frontal (3 monkeys) frozen sections of 70 µm were obtained from a sliding microtome. The sections were collected serially in phosphate buffer saline (PBS, 0.1M, pH 7.4) and processed for the visualization of BDA according to the avidine-biotin-peroxydase method (ABC Elite kit, Vector Labs, Burlingame, CA) with 3,3’diaminobenzidine (DAB; Sigma, St. Louis, MO) as the chromogen. In brief, the sections were incubated overnight at 4°C in a solution containing ABC diluted 1:100 in 0.1M PBS, pH 7.4, plus 1% normal rabbit serum and 1% triton X-100. They were then rinsed twice in PBS and once in Tris buffer. The bound peroxidase was revealed by incubating the sections in a solution containing 0.05% DAB, 0.3% nickel- ammonium sulfate, and 0.005% H2O2 in 0.05M Tris buffer (pH 7.6) for 8-10 minutes at room
temperature. The reaction was terminated by a rinse in Tris buffer followed by two rinses in PBS.
To help identify cortical layers, nuclei and structures that harbored labeled neurons and axons, sections were counterstained for cytochrome oxidase, according to the histochemical protocol of Wong-Riley (Wong-Riley, 1979). The counterstaining was performed before BDA revelation, and nickel-cobalt-intensified DAB (dark blue reaction) and unintensified DAB (diffuse brown precipitate) were used to reveal BDA and cytochrome oxidase, respectively.
Material analysis and neuronal reconstructions
All sections were mounted on gelatin-coated slides, dehydrated in graded alcohols, cleared in toluene, and coverslipped with Permount. Labeled axons were reconstructed in three dimensions by using a light microscope equipped with a motorized stage and an image analysis software (Neurolucida, MicroBrightField, Colchester, VT). Entire and individual axonal reconstructions were obtained from serial sagittal or transverse sections, each containing at least one axonal segment. By going from one section to another, we were able to follow and reconstruct individually injected axons. The terminal fields of labeled neurons were mapped at lower magnifications to determine their topographic localization according to the atlas of Szabo and Cowan (Szabo & Cowan, 1984). The photomicrographs were digitally captured (camera model DC 300, Leica, Wetzlar, Germany) and processed with the Adobe Photoshop software (version CS6, Adobe, San Jose, CA). The terminal arborization of each axon at the STN level was carefully examined. A neuron was considered projecting toward lower brainstem regions when its labeled axon could be traced into the cerebral peduncle, ventral and caudal to the substantia nigra. Each axonal varicosity encountered along the various branches of individual axons was precisely charted and the total number of such varicosities was determined to estimate the strength of the synaptic input provided by a single corticofugal axon.
Preparation of samples for electron microscopy
The brain of one cynomolgus monkey was prepared for electron microscopy. This monkey was transcardially perfused with 400 mL of ice-cold PBS (50 mM; pH 7.4) followed by 600 mL of 3.0% acrolein in PBS and by 700 mL of cold 4% PFA with 0.2% glutaraldehyde and finally by 600 mL of cold 4% PFA. The brain was rapidly dissected out, post-fixed by immersion in PFA for 1h at 4°C and cut with a vibratome (model VT1200 S, Leica) into 50 µm-thick transverse sections collected in PBS. Three 50 µm-thick transverse sections were taken through the STN, at 12.1 mm relative to the interaural plane (Szabo & Cowan, 1984), and immunostained for the vesicular glutamate transporter 1 (VGluT1), a faithful marker of corticofugal axons (Fujiyama et al., 2006; Raju et al., 2006). Briefly, these free-floating sections were sequentially incubated at room temperature in: (1) a blocking solution of PBS, containing 2% normal goat serum and 0.5% gelatine (1h); (2) the same solution containing a 1:1000 dilution of guinea pig polyclonal antibody against VGluT1 (overnight, Catalog no. AB5905; EMD Millipore Corporation) and (3) a 1:1000 dilution of biotinylated goat anti- guinea pig antibody (1h, Vector Labs) diluted in the same solution. After rinses in PBS, sections were incubated for 1h at 4°C in ABC diluted 1:100 in blocking solution. They were then rinsed in PBS and Tris-saline buffer (TBS; 50 mM; pH 7.4), and the bound peroxidase was revealed by incubating the sections for 3 min, at room temperature, in a 0.05% solution of DAB diluted in Tris, to which 0.005% H2O2 was added. The reaction was stopped by
several washes in TBS followed by PB. Sections were then incubated for 30 min in a 1% solution of OsO4 diluted in PB, followed by several rinses in PB. They were then dehydrated
in graded ethanol and in propylene oxide and flat-embedded in Durcupan (Fluka, Buchs, Switzerland). Quadrangular pieces of the dorsolateral STN were cut from the flat-embedded VGluT1-immunostained sections and glued on the tip of a resin block and cut ultrathin (~80 nm) with an ultramicrotome (model EM UC7, Leica). After being collected on bare 150- mesh copper grids and stained with lead citrate, the ultrathin sections were examined by using a transmission electron microscope (Tecnai 12; Philips Electronic, Amsterdam, Netherlands), at 100 kV, and an integrated digital camera (MegaView II; Olympus, Müster, Germany). The VGluT1-positive (+) axon varicosities were randomly sampled at a working magnification of 11,000x by taking a picture every time one was encountered. VGluT1+ axon varicosities were analyzed, using the public domain IMAGE J processing software from NIH (v.1.45),
for the long and short axis and cross-sectional area. They were then categorized as containing or not a mitochondrion, and as showing or not a synaptic junctional complex, i.e. a localized straightening of apposed plasma membranes associated with a slight widening of the intercellular space and a thickening of the pre- and/or postsynaptic membrane. All synaptic junctions were also characterized as symmetrical or asymmetrical, the synaptic target identified, and the length of junctional complexes measured. The synaptic incidence observed in single section was then extrapolated to the whole volume of varicosities by means of the formula of Beaudet and Sotelo (Hardman et al., 2002), using the long axis as diameter, according to Umbriaco et al. (Umbriaco et al., 1994).
RESULTS
General labeling features
As defined anatomically using physiological maps of monkey cerebral cortex, the M1 area targeted in this study corresponds mainly to the forelimb area (Woolsey, 1958; Kwan et al., 1978; Stepniewska et al., 1993). The injection sites were centered upon the cortical layer V since previous electrophysiological as well as anterograde and retrograde cell labeling studies indicate that corticosubthalamic projections in rats and primates arise mainly from cortical neurons located in this layer (Rouzaire-Dubois & Scarnati, 1985; Canteras et al., 1990; Nambu et al., 1996). Under propofol anesthesia, cortical neurons in layer V of M1 display characteristic bursting firing patterns (Fig. 1C) that facilitate their identification. Almost all injection loci have a dense core composed of BDA precipitate (Fig. 1A) surrounded by several neurons labeled in a Golgi-like manner (Fig. 1B). Neurons labeled in M1 layer V have a triangular cell body ranging between 15 and 30 µm in diameter. Usually, 2 to 5 horizontal basal dendrites arborizing as far as 400 to 700 µm away from the cell body were seen, as well as a single apical dendrite (Fig. 1B, C). The remarkable length of these dendrites explains why labeled neurons were occasionally found at some distance from the core of the injection site. Intensely labeled axons could be seen to emerge either from the core of the injection sites or from individually labeled neurons located peripherally. In the latter case,
the axons emerge from the basis of the cell body and then invade the subcortical white matter where they initiate their downward course.
Only axons of M1 cortical neurons projecting to the STN were traced in the present study since reconstructions were initiated in the STN itself. Twenty-eight corticosubthalamic axons were traced. Unfortunately, because of the dense core of BDA precipitate surrounding the injection sites, none of these reconstructed axons could be directly connected to their parent cell body. However, a detailed analysis of the material was made to ensure that they were emerging from the core of the injection loci. All brain regions that are known to project to M1, including the thalamus, were carefully examined for the presence of potentially confounding retrogradely-labeled neurons, but no such retrogradely-labeled neurons were observed. The single-axon reconstruction approach used here is a very powerful technique that yields a detailed view of single-neuron axonal arborization (Parent et al., 2001; Parent & Parent, 2005; Parent & Parent, 2006; Gagnon & Parent, 2014; Parent & Parent, 2016). The fact that each charted axonal segment and axonal collateral belongs to the same axonal unit was ensured by carefully examining all serial brain sections in which these axonal segments were encountered. However, because of the considerable length of these corticofugal axons, the most distal portions of certain axonal branches, particularly those that reached the lower brainstem regions, became too faintly labeled to be accurately traced, although they were clearly in continuity with the rest of the axonal branching. In such cases, the distal ends of axonal branches were labeled with an arrow in figures. This applies principally to the long- ranged axonal branches, en route to the lower brainstem regions and/or spinal cord that were lost in the cerebral peduncle, ventral and posterior to the substantia nigra.
Initial axonal trajectory
Based on their target sites and axonal branching patterns, different types of corticosubthalamic projection axons arising from M1 forelimb area were identified (Table 1). Apart from the STN in which the tracing was initiated, brain regions that are targeted by reconstructed axons of the hyperdirect pathway are the lower brainstem (25/28 neurons), the zona incerta (ZI, 15/28 neurons), the red nucleus (6/28 neurons), the reticular thalamic
widespread axonal branching patterns, all M1 reconstructed axons forming the hyperdirect pathway follow a similar initial trajectory. Primary axons endowed with a large diameter (2 – 3 µm) emerge from the core of the injection site to invade the subcortical white matter and initiate a sinuous downward course. The vast majority of reconstructed corticosubthalamic axons (25/28) have their major axonal branch projecting toward lower brainstem regions. This principal axonal segment travels in the internal capsule, between the caudate nucleus and the putamen (Fig. 2B). Interestingly, one axon was seen passing through the caudate nucleus, but without leaving any boutons en passant in this striatal region (Fig. 2A). The main axonal branch then reaches the cerebral peduncle, ventral and posterior to the substantia nigra. Of all reconstructed corticosubthalamic axons, none were seen to emit collateral innervating the striatum, indicating that the hyperdirect pathway comes from a distinct cortical neuronal population than the one at the origin of the corticostriatal projections. Most of the major axonal branches traced (23/25) exhibit large and beaded axon varicosities as they course in the internal capsule, with an average of 37 axon varicosities per axon. No axon collaterals were seen to cross the midline.
Subthalamic nucleus
Our results indicate that the hyperdirect pathway is largely composed of axons that are not solely dedicated to the STN. Indeed, only one axon was seen to innervate exclusively the STN. In the internal capsule, axon of this neuron was thinner (~ 1 µm) than that of the other long-ranged axons and its small diameter remains constant throughout its course. It then enters the dorsolateral tip of the STN, where it branches into 3 varicose axon collaterals that arborize in the dorsolateral region of the nucleus. The STN innervation provided by all other reconstructed axons derives essentially from 1 or 2 thin axon collaterals that depart at right angle from the main and larger axonal branch that runs downward in the cerebral peduncle. These thinner axon collaterals have a diameter ranging from 0.5 to 1.5 µm and depart from the main axon at the level the dorsolateral tip of the STN to invade this BG component. These branches usually run lateromedially within the dorsal region of the STN, providing boutons en passant (see photomicrographs in Fig. 2B and Fig 5B). Occasionally, the lateromedially coursing collateral(s) breaks out into even thinner and varicose branches that plunge ventrally within dorsolateral region of the STN. An example of such neuron is provided in figure 2A.
Its main axon travels in the internal capsule, where it yields 16 large and beaded axon varicosities in this myelinated fiber bundle. Once it reaches the height of the dorsolateral tip of the STN, the main axon emits a thinner and perpendicular oriented branch that enters the STN. In the STN, this smaller collateral breaks out into numerous thinner and varicose axonal