Abstract
In human and nonhuman primates, deep brain stimulation applied at or near the internal medullary lamina of the thalamus [a region referred to as “central thalamus,” (CT)], but not at nearby thalamic sites, elicits major changes in the level of consciousness, even in some minimally conscious brain-damaged patients. The mechanisms behind these effects remain mysterious, as the connections of CT had not been specifically mapped in primates. In marmoset monkeys (Callithrix jacchus) of both sexes, we labeled the axons originating from each of the various CT neuronal populations and analyzed their arborization patterns in the cerebral cortex and striatum. We report that, together, these CT populations innervate an array of high-level frontal, posterior parietal, and cingulate cortical areas. Some populations simultaneously target the frontal, parietal, and cingulate cortices, while others predominantly target the dorsal striatum. Our data indicate that CT stimulation can simultaneously engage a heterogeneous set of projection systems that, together, target the key nodes of the attention, executive control, and working-memory networks of the brain. Increased functional connectivity in these networks has been previously described as a signature of consciousness.
SIGNIFICANCE STATEMENT In human and nonhuman primates, deep brain stimulation at a specific site near the internal medullary lamina of the thalamus [“central thalamus,” (CT)] had been shown to restore arousal and awareness in anesthetized animals, as well as in some brain-damaged patients. The mechanisms behind these effects remain mysterious, as CT connections remain poorly defined in primates. In marmoset monkeys, we mapped with sensitive axon-labeling methods the pathways originated from CT. Our data indicate that stimulation applied in CT can simultaneously engage a heterogeneous set of projection systems that, together, target several key nodes of the attention, executive control, and working-memory networks of the brain. Increased functional connectivity in these networks has been previously described as a signature of consciousness.
Introduction
Deep brain stimulation (DBS) applied at specific sites can enhance arousal and consciousness in some minimally conscious, brain-damaged patients (for review, see Schiff, 2013; Kundu et al., 2018). Clinical studies in these patients pinpointed a site at the center of the thalamus around the internal medullary lamina [“central thalamus” (CT)] as particularly effective for eliciting such arousal effects (Schiff et al., 2007; Schiff, 2010; Yamamoto et al., 2010; Magrassi et al., 2016; Chudy et al., 2018; Lemaire et al., 2018).
Recently, studies in macaques have started to experimentally investigate the neuronal mechanisms behind CT stimulation effects. These studies reported awakening from anesthetic-induced unconsciousness (Redinbaugh et al., 2020; Bastos et al., 2021; Tasserie et al., 2022) or markedly improved performance in attention-dependent tasks in awake animals (Baker et al., 2016; Redinbaugh et al., 2020; Janson et al., 2021) during stimulation with electrodes positioned in CT. Moreover, these effects were correlated with changes in functional connectivity between high-level areas in the frontal, parietal, and cingulate cortices (Redinbaugh et al., 2020; Tasserie et al., 2022). The circuit mechanisms behind the effectiveness of CT DBS, however, remain unclear (Liu et al., 2015; Redinbaugh et al., 2020; Schiff, 2020; Bastos et al., 2021; Janson et al., 2021; Tasserie et al., 2022).
Recruitment of some thalamic output pathways is considered key for the arousal effects of CT DBS (Schiff, 2013; Liu et al., 2015; Janson et al., 2021). Under the parameters used for CT DBS in human and nonhuman primates, the radius of the efficient stimulating area is estimated in a few hundred micrometers (Ranck, 1975; McIntyre et al., 2004; Chiken and Nambu, 2016). Recently, studies in nonhuman primates have narrowed this estimation by modeling from MRI the fibers emanating from specific nuclei (Baker et al., 2016; Janson et al., 2021) or by applying controlled low-intensity microstimulation techniques (Redinbaugh et al., 2020). Currently available evidence thus indicates that a DBS probe positioned at the center of the CT region [in the intralaminar central lateral nucleus (CL) or paracentral nucleus (Pc)] can directly stimulate these nuclei (Redinbaugh et al., 2020; Janson et al., 2021) and, depending on the intensity also adjacent zones of the ventral lateral posterior nucleus (VLp), mediodorsal nucleus (MD; Baker et al., 2016), anterior pulvinar nucleus (APu), and/or the lateral part of the centromedian nucleus (CM; Tasserie et al., 2022). It is even possible that anterograde or retrograde stimulation of fibers traveling in the internal medullary lamina might contribute to the observed effects (Janson et al., 2021). In contrast, stimulation at more lateral or medial thalamic sites reportedly fails to elicit arousal effects (Redinbaugh et al., 2020; Tasserie et al., 2022).
Because of their evolutionary proximity to humans, the study of other anthropoid primate species offers a uniquely relevant window into the organization and function of human brain circuits (Preuss and Wise, 2022; Scott and Bourne, 2022). However, modeling in anthropoid primates the circuit mechanisms involved in CT DBS has been hampered by the limited and fragmentary information available. In these species, the origin and arborization patterns of CT nuclei axons has been directly visualized only for the centromedian nucleus (François et al., 1991; Sadikot et al., 1992; Parent and Parent, 2005) and the lateral MD (Erickson and Lewis, 2004). For the other CT nuclei, axon morphologies remain an inference from the cell-labeling patterns observed in the thalamus after retrograde tracer injections in the cortex (Asanuma et al., 1985; Darian-Smith et al., 1990; Barbas et al., 1991; Cavada et al., 2000; Kultas-Ilinsky et al., 2003; Morel et al., 2005; Stepniewska et al., 2007; Mayer et al., 2019; Gamberini et al., 2020) or striatum (Nakano et al., 1990), or an extrapolation from observations in more distantly related species such as rodents (Berendse and Groenewegen, 1991; Deschênes et al., 1996a,b; Unzai et al., 2017) or carnivores (Royce et al., 1989).
Here, we set out to elucidate the axonal pathways that originate from each of the various neuronal groups in the CT region of the marmoset (Callithrix jacchus), a new-world anthropoid primate that has in recent years become a prime model in experimental neuroscience and brain disease modeling (for review, see Okano, 2021; Scott and Bourne, 2022; Skibbe et al., 2023).
Materials and Methods
Animals.
Six adult marmoset monkeys (C. jacchus; two males and four females; age range, 1–4 years) were used for the experiments in the present report. The animals were bred at the Primatology Center of the Federal University of Rio Grande do Norte (UFRN, Natal, Brazil), and were housed under natural lighting, temperature, and humidity conditions, with food and water available ad libitum. Procedures were approved and supervised by the UFRN Animal Experimentation Ethics Committee (Animal Use Ethics Commission protocol #CEU–020/2012), in accordance with Brazilian law and the guidelines published in the NIH Guide for the Care and Use of Laboratory Animals (available at: http://nap.edu/catalog/12910.html).
Anesthetic and analgesic procedures.
For the tracer injections, animals were first sedated with diazepam 1 mg/kg, i.m. Anesthesia was induced by inhalation of a mixture of 2–4% isoflurane in oxygen, and subsequently adjusted to keep the animal areflexic, but breathing spontaneously, throughout the surgical procedure. The anesthetized animals were placed in a stereotaxic apparatus (Narishige). A water-circulated heating pad (T/Pump, Gaymar) at 37°C was used to maintain body temperature, saline was administered (1.5 ml/h, s.c.) to keep hydration, and the eyes were protected with Vaseline to prevent desiccation. Cardiac and breathing frequency were monitored throughout the surgery. At the end of the surgical session, the isoflurane was discontinued, and animals recovered spontaneously. Buprenorphine hydrochloride (0.01 mg/kg/12 h, i.m.) was administered for postsurgical analgesia. Animals also received a single dose (0.1 mg/kg, s.c.) of the long-acting veterinary antibiotic cefovecin sodium.
After a survival period of 2 weeks after the tracer injection, animals were killed with a lethal dose of sodium pentobarbital (100 mg/kg, i.p.), before perfusion.
Tracer injection experiments.
To label the axons of thalamic neurons, 10 kDa biotinlylated dextran-amine (BDA; Thermo Fisher Scientific) 10% in 0.01 m PB, pH 7.4 was iontophoretically microinjected (one injection per hemisphere) through glass micropipettes into the thalamus. The scalp was cut open at the midline, a small craniotomy was drilled on the parietal bone, and the dura mater was opened. Borosilicate glass micropipettes (FHC; internal diameter, 10 μm) loaded with 10% BDA 10 kDa in 0.01 m PB, pH 7.4 (Thermo Fisher Scientific) were positioned following stereotaxic atlas coordinates (Paxinos et al., 2012). Positive pulses (2–4 µA, 7 s on/off cycles, for 20–40 min) were delivered using a Midgard Precision Current Source (Stoelting). Pipettes were left in place for 5 min after the injection, and the micropipette was then removed. The cortex was covered with hemostatic gelfoam, and the bone defect occluded with acrylic cement. After recovering from the anesthesia, the animals were returned to their housing during the survival period.
Following the procedures for killing described above, animals were perfused through the left ventricle with saline (0.9%, 5 min), followed by 4% paraformaldehyde in PB 0.1 m, pH 7.4, for 30 min and 4% paraformaldehyde plus 15% sucrose in PB 0.1 m, pH 7.4, for 15 min. The brains were dissected, sectioned into blocks adjusted to the stereotaxic coronal plane, photographed, postfixed for 24 h in the last perfusion solution and finally embedded in 30% sucrose in PB 0.1 m, at 4°C.
Histologic procedures.
Six parallel series of 40-µm-thick coronal sections were obtained on a freezing microtome (model SM 2400, Leica) and collected in PB (0.1 m, at 4°C). One series was used to visualize BDA. In these sections, peroxidase activity was blocked via incubation in 2% oxygen peroxide in PB (0.1 m) for 20 min. The free-floating sections were incubated with an avidin-biotin-peroxidase (1:100, 12 h, at 4°C; Vectastain Elite, Vector Laboratories). After several rinses in PB, the bound peroxidase activity was revealed using the glucose oxidase-DAB nickel sulfate-enhanced method (Shu et al., 1988). Sections were serially mounted onto gelatinized glass slides, lightly counterstained with thionine, defatted, and coverslipped with DePeX (SERVA). As an aid for delimitation of thalamic nuclei and cortical areas, parallel sections were stained with cresyl violet (Nissl), histochemistry with acetylcholinesterase (AChE; Geneser-Jensen and Blackstad, 1971; or histochemistry with cytochrome-oxidase (CyO; Wong-Riley, 1979). Remaining series were stored.
Data analysis.
Seven injection experiments were considered valid. BDA labeling was analyzed throughout the whole cerebral hemisphere in 1:6 coronal sections using a microscope (model Eclipse 600, Nikon) under 4× to 40× objectives and bright-field optics. Panoramic images were acquired using the X-Y motorized microscope stage (Proscan II, Prior Scientific Instruments) driven by NIS Elements Image software (version 3.22.15; Nikon). The position of each injection site was delineated by the presence of a core of densely impregnated neuropil and cell somata. Thalamic nuclei were delineated on the thionin-stained sections and by comparison with adjacent Nissl-, AChE-, and CyO-stained thalamus sections. In a complete series (>130 sections), labeled thalamocortical axon arborizations were plotted in highly simplified form (as dots with variable spacing along each section contour line), onto a standard flat representation of the marmoset cortex surface, computationally “unfolded” from the coronal sections (Paxinos et al., 2012). Labeled axons in regions of interest of the cortex or striatum were drawn on 10× digital images.
Results
We injected the sensitive axonal tracer BDA unilaterally or bilaterally in different portions of the “CT region” in six adult marmosets (see Materials and Methods). Each deposit yielded a specific pattern of anterograde labeling in the cerebral cortex and/or striatum (Figs. 1–3).
Selective BDA deposits in different regions of the marmoset CT. A–C, H, Multilabeling delineation of nuclei and fiber tract boundaries in the marmoset thalamus. Adjacent coronal sections containing the center of a BDA deposit (H, Case 5) stained with either cresyl violet (A) or AChE histochemistry (B) or visualized under oblique “dark-field” illumination (C) are shown. The medial and lateral border of the IML is indicated by dashed lines. D–J, Injection sites (black precipitate) of iontophoretic BDA deposits in seven different injection experiments. The areas shaded in pink and blue represent, respectively, the IML or the paralaminar zones of the CT region, as described by the functional literature (see text). Scale bar, 500 µm. CeM, Central medial nucleus; IPu, inferior purlvinar nucleus; LD, laterodorsal nucleus; LGN, lateral geniculate nucleus; LP, lateral posterior nucleus; LPu, lateral pulvinar nucleus; MDM, mediodorsal nucleus, medial part; MGN, medial geniculate nucleus; MPu, medial pulvinar nucleus; Po, posterior nucleus; Pf, parafascicular nucleus; VLp, ventral lateral nucleus, posterior part; VPi, ventral posterior nucleus, inferior part; VPL, ventral posterolateral nucleus; VPM, ventral posteromedial nucleus; VPs, ventral posterior nucleus, superior part.
Cortical area distribution of thalamocortical axons labeled by selective BDA deposits in different regions of the marmoset CT. A, Standard two-dimensional representation of the marmoset cerebral cortex surface. The figurine on top shows a lateral view of the cerebral cortex. Below, the surface of the whole neocortex is shown as a two-dimensional (2D) reconstruction computationally “unfolded” from the coronal sections of the Paxinos et al. (2012) marmoset brain atlas. The lateral cortex surface visible in the figurine at the top is shaded in yellow, while the inferior and medial cortical regions hidden from view in the figurine are shaded in gray. Four coronal section levels are indicated as parallel lines both on the figurine at the top (each marked with its anteroposterior interaural level) and on the 2D map. Note that, because of the nonlinear distortion applied to preserve neighborhood relationships between areas in the 2D map, coronal section levels are projected as wavy lines. B–H, Schematic representation of the distribution of BDA-labeled thalamocortical axons plotted from serial coronal tissue sections onto the 2D cortex map. The position and relative abundance of labeled terminal arborizations is indicated by red dots. The nucleus injected in each case is indicated in the lower left corner of each diagram. Scale bar, 1 cm. A4ab, Area 4 of cortex, parts a and b; A4c, area 4 of cortex part c; A6DC, area 6 of cortex, dorsocaudal part; A6DR, area 6 of cortex, dorsal and rostral part; A6M, area 6 of cortex, medial part; A6Va, area 6 of cortex, ventral part a; A6Vb, area 6 of cortex, ventral part b; A8aV, area 8a of cortex, ventral part; A8b, area 8 of cortex part b; A8C, area 8 of cortex, caudal part; A8aD, Area 8a of cortex dorsal part; A10, area 10 of cortex; A19M, area 19 of cortex, medial part; A23a, area 23a of cortex; A23b, area 23b of cortex; A23V, area 23 of cortex, ventral part; A24c, area 24c of cortex; A24d, area 24d of cortex; A29, area A29 of cortex; A30, area 30 of cortex; A31, area 31 of cortex; A46D, area 46 of cortex, dorsal part; A46V, area 46 of cortex, ventral part; A47D, area 47 of cortex dorsal part; A47L, area 47 of cortex, lateral part; OPt, occipito-parietal transitional area of cortex; PE, parietal area PE of cortex; PEC, parietal area PE, caudal part; PF, parietal area PF of cortex; PFG, parietal area PFG of cortex; PG, parietal area PG of cortex; ProSt, prostriate area of cortex; V1, primary visual cortex; V2, visual area 2; V6, visual area 6; V6A, visual area 6A.
Valid injection experiments and table summary of the striatal and cortical territories containing labeled axons. Age is expressed in months. In the striatum, the morphologic type of terminal axon arborization (Deschênes et al., 1995, Type 1 or Type 2) is also indicated. A8aD, Area 8a of cortex, dorsal part; A8b, area 8b of cortex; A45, area 45 of cortex; A46, area 46 of cortex; A47, area 47 of cortex; AIP, anterior intraparietal area of cortex; LIP, lateral intraparietal area of cortex; MIP, medial intraparietal area of cortex; PGM, parietal area PG, medial part; PreC, precommissural; PostC, postcommissural; ProM, proisocortical motor region; S1, primary somatosensory cortex; S2, secondary somatosensory cortex; TPt, temporoparietal transitional area; V1–V3, visual areas 1–3; V6DM, visual area 6, dorsomedial part; VIP, ventral intraparietal area of cortex. For other abbreviations, see Figure 2.
Because of the histologic intricacy of the CT region, a crucial step for the interpretation of these experiments was to delineate thalamic nuclei borders in an accurate and consistent fashion. To this end, we imaged the extent of the injection site in a series of BDA-stained sections under both bright-field and dark-field optics and matched these images with the images of adjacent serial sections stained to reveal cytoarchitecture (Nissl) or AChE and CyO enzyme activities. From the combined evidence of these multiple histologic criteria, nuclei borders could be ascertained consistently and with high confidence (Fig. 1A–C).
We then scaled to the dimensions of the marmoset thalamus the approximate extent of the CT region as described by functional studies in macaques (Baker, 2016; Janson et al., 2021; Tasserie et al., 2022). Within this region, we distinguished the intralaminar nuclei CL, Pc, and CM that are contained amid the internal medullary lamina (IML) fibers (Fig. 1A–C, dashed white lines, D–I, shaded in pink) from the “paralaminar” laminal portions of the mediodorsal nucleus, lateral part (MDL) and the VLp adjacent to the IML fibers (Fig. 1, shaded in blue).
Five deposits were placed within the consensus boundary of CT. Two of these experiments were confined to the IML. An injection circumscribed to CL and Pc at about anteroposterior (AP) +5.15 mm (Fig. 1D, Case 1) produced robust labeling in the striatum and weaker labeling in the cortex (Figs. 2B, 3, 4A, 5A). In the striatum, the labeled axon arborizations covered an extensive lateral region of the precommissural and postcommissural portions of the putamen, an associative and sensorimotor domain of the striatum (for review, see Haber, 2003). Thalamostriatal branches were mostly simple and elongated, with numerous en passant boutons and occasional varicosity-tipped short appendages (Fig. 5B); this thalamostriatal axon morphology has been dubbed “type 1” and has been shown to correlate with specific structural and functional synaptic features (Deschênes et al., 1995; for review, see Martel and Galvan, 2022). In the cortex, the thalamic projection axons terminate only in dorsomedial motor, premotor, and anterior cingulate areas (Figs. 2B, 3). The labeled intracortical arborizations were spread and relatively sparse, branching in both the supragranular and the infragranular layers of the cortex without a clear preference; however, they mostly spared layer 1 (Fig. 4A).
Laminar distribution of labeled thalamocortical arborizations originated from CT. A, Camera lucida drawings of labeled thalamocortical arborizations in motor (top) or premotor (bottom) cortices following BDA injections in the IML (Cases 1 and 2). B, Thalamocortical axons labeled by injections in the paralaminar zones of CT (Cases 3–5). C, Columnar sample showing BDA-labeled thalamocortical axon arborizations in posterior parietal cortex area PE following an injection in paralaminar VLp (B, blue line frame). Scale bar, 500 µm.
Thalamostriatal projections from CT nuclei and summary diagram. A, Section diagrams of the caudate (Cd) and putamen (Pu) taken at five representative levels of the striatum are shown to illustrate the thalamostriatal axon arborizations (red stippling) labeled by injections in the anterior/dorsal (Case 1) or posterior/ventral (Case 2) portions of the IML. The levels rostral to the interaural coronal plane (AP+) are indicated. B, High-magnification image of thalamostriatal axon arborizations labeled by the CL–Pc injection. Note the diffuse and linear arrangement of the axonal branches and the presence of en passant boutons in them. C, Thalamostriatal arborizations labeled by the BDA injection in CM–Pc. Note the focal, plexus-like morphology of the terminal branches and frequent terminal boutons at the end of a short stalk. D, Thalamostriatal axon arborizations labeled from injections in regions of CT with sparse projections to the striatum. The labeling observed by the injection in Case 6 (APul) is also shown. Conventions are as in A. Scale bars: A–D, 1 mm; B, C, 10 µm.
The injection in Case 2 was located in the dorsolateral portion of CM at about AP +4.5 mm (Fig. 1E). In contrast with the previous case, this experiment labeled several small foci of terminal arborizations in the striatum, and only a few and sparsely branched axons in the cortex (Figs. 2C, 3, 4A, 5A). The axonal arborizations in the striatum were located in the dorsal putamen and lateral caudate. They consisted of clusters of varicose branches, with frequent varicosity-tipped short branchlets (Fig. 5C). This kind of thalamostriatal terminal axon morphology (“type 2”; Deschênes et al., 1995) characterizes the CM–Pf (parafascicular nucleus) axons in primates and rodents and has been shown to correlate with structural and functional synaptic features (Parent and Parent, 2005; Martel and Galvan, 2022) different from all other thalamostriatal connections. The few thalamocortical axon branches labeled in this experiment were located in cortical layers 5 and 6 (Fig. 4A).
Two further deposits (Cases 3 and 5) involved regions medially or laterally adjacent to the IML (paralaminar zone; Fig. 1, shaded in blue) yet spared the IML and the intralaminar nuclei. In contrast with the IML injections, these deposits produced very weak labeling in the striatum yet heavy labeling in the cortex. The deposit in Case 5 (Fig. 1H) was situated in a medial and caudal portion of the VLp nucleus. This injection labeled thalamocortical arborizations in the frontal lobe (premotor, oculomotor, and primary motor areas), as well as in the posterior parietal and cingulate cortices (Figs. 2F, 3). The labeled VLp thalamocortical arborizations were relatively abundant in both infragranular and supragranular layers of the cortex, and their terminal branches showed limited clustering (Fig. 4B,C). The deposit in Case 3 impregnated the MDL (Fig. 1F). In this experiment, the labeled thalamocortical axons targeted mainly the lateral premotor and oculomotor areas (Figs. 2D, 3), and their terminal branching was profuse and clustered into multiple foci. They mostly concentrated in layers 3b–3a and only a few of them reached layer 1 (Fig. 4B). In these two paralaminar injection cases, the thalamostriatal arborizations were relatively sparse, located in the postcommissural putamen and adjacent caudate, and showed type 1 morphology (Fig. 5B,D). A further case involved a more caudal zone of MDL and a portion of the caudal CL (Fig. 1G). Labeling in this experiment was essentially a combination of the patterns observed in Cases 1 and 3, with a more robust labeling in the caudate nucleus, probably reflecting the involvement of caudal CL (Figs. 3, 4B, 5C).
Finally, we examined the labeling produced by two BDA deposits situated further away from the IML, at or immediately beyond the borders of the CT region. These experiments labeled axons almost exclusively in the cerebral cortex. In Case 6, a deposit in the APu with some involvement of the lateral posterior nucleus (Fig. 1I, Case 6) labeled massive numbers of thalamocortical axons in the posterior parietal and cingulate areas (Figs. 2G, 3). Their terminal arborizations showed a clear periodic clustering, suggestive of columnar selectivity. Innervation of layer 1 was substantial. An injection in the central division of MD (MDC; Fig. 1J, Case 7) labeled numerous axons in rostral prefrontal and oculomotor cortices, but not in more caudal cortical areas or in the striatum (Figs. 2H, 3).
Discussion
To visualize the axon pathways arising from the arousal-related CT region of the anthropoid primate thalamus, we transposed to marmosets the CT delineations reported in macaque studies and labeled its output pathways by means of selective BDA microinjections. Results show that the various CT cell groups focus their projections on the associative and sensorimotor striatum and/or dorsolateral prefrontal, premotor, or posterior parietal areas of the cortex (Fig. 6). These data provide an anatomic foundation for modeling the circuit mechanisms that underlie the effects of CT stimulation on attention and arousal.
Our multilabeling protocol allowed us to delineate with confidence the intralaminar nuclei proper (i.e., those cell groups contained amid the fibers of the internal medullary lamina, CL, Pc, and Pf) from the adjacent, paralaminar portions of MD and VLp. This distinction was not always clear in published studies and may have contributed to conflicting results. It is of note that the paralaminar regions of MD and VLp have been specifically distinguished from CL and Pc because of their extensive projections to the parietal and frontal cortices (Avendaño et al., 1990; Erickson and Lewis, 2004; Mayer et al., 2019) and their involvement in volitional saccadic eye movements (Tanibuchi and Goldman-Rakic, 2005; Kunimatsu and Tanaka, 2010).
CT projections to cortex and striatum
Consistent with previous retrograde labeling observations in primates (Parent et al., 1983; Nakano et al., 1990; Jones, 2007) and anterograde labeling studies in rodents (Berendse and Groenewegen, 1991; Van der Werf et al., 2002), our injections confined to the IML show that the intralaminar nuclei send their main projection to the sensorimotor striatum (Fig. 4A, 5A). In contrast, the anterior and posterior intralaminar nuclei innervate the cerebral cortex with different profusion. Cortical projections from CM are sparse and poorly branched (Figs. 2C, 3, 4A), consistent with a report in squirrel monkeys that many CM cells do not even target the cortex, just the striatum and other basal ganglia nuclei (Parent and Parent, 2005), as well as with anterograde labeling studies of the equivalent nuclei in rodents (Deschênes et al., 1996a; Van der Werf et al., 2002; Unzai et al., 2017). In contrast, the projections from the CL and Pc to the premotor cortex are relatively abundant, consistent with previous retrograde tracing observations in primates (Morel et al., 2005) and anterograde studies in rodents (Deschênes et al., 1996b; Van der Werf et al., 2002), although they do not form dense focal plexuses in specific layers or columns, as it is typical of most thalamocortical axons.
Conversely, the paralaminar portions of MDL and VLp send robust projections to the cortex, yet very sparse ones to the striatum. Our observations in marmosets are in general agreement with anterograde (Giguere and Goldman-Rakic, 1988; Erickson and Lewis, 2004) or retrograde (Parent et al., 1983; Nakano et al., 1990; Barbas et al., 1991; Jones, 2007; Mayer et al., 2019) tracer studies in macaques. Neurons in both paralaminar territories target the premotor cortex; the MDL neurons target, in addition, oculomotor and dorsolateral prefrontal areas involved in working memory and cognition, while the paralaminar VLp neurons innervate posterior parietal areas related to visuomotor guidance and extrapersonal space representation (Kawashima et al., 1995; Zaehle et al., 2007; Filippini et al., 2018). At least at a population level, therefore, neurons in the paralaminar region of VLp have simultaneous extensive direct access to both premotor and posterior parietal areas of the cortex.
Recent DBS macaque experiments reported that activation of the medial parietal and anterior and posterior cingulate cortices, in addition to dorsolateral prefrontal areas, is critically correlated with arousal and conscious access, and is dependent on the site of thalamic stimulation (Tasserie et al., 2022). In our experiments, some thalamocortical projections to the medial parietal and cingulate areas were labeled by the injections in CL-Pc (Case 1) and VLp (Case 5). However, labeling in these areas was massive in Case 6, whose injection was situated at the lateral border of our paralaminar CT domain, mostly in APu. These anterograde findings are consistent with connection studies in macaques in which retrograde tracers were injected into the medial parietal and posterior cingulate areas (Buckwalter et al., 2008; Impieri et al., 2018; Gamberini et al., 2020). Interestingly, human behavioral and clinical studies indicate that the equivalent areas play a central role in the forebrain networks of consciousness. These areas are particularly active during self-introspective, awake rest (Gusnard et al., 2001), as well as in a wide spectrum of highly integrated tasks, including visuospatial imagery, episodic memory retrieval, and first-person perspective taking. Moreover, these areas remain deactivated in sleep, hypnosis, general anesthesia, and persistent vegetative state (for review, see Cavanna and Trimble, 2006), and their functional connectivity with the thalamus is altered during the vegetative state yet regains near-normal values if consciousness is recovered (Laureys et al., 2004).
In the cortex, axons from the various CT nuclei distribute across laminae in different patterns. As observed in primates and other mammals, the intralaminar nuclei innervate preferentially the deep cortical layers (Herkenham, 1986; Avendaño et al., 1990; Jones, 2007; Rubio-Garrido et al., 2009). In contrast, MD axons and APu axons arborize densely in cortical layers 3–4 and layer 1. Through these arborizations, CT stimulation might favor synchronous activity in the apical and basal dendritic domains of the same deep-layer pyramidal cells. These cell-level mechanisms have recently been linked to a facilitation of functional connectivity in the thalamocortical network required for consciousness (Aru et al., 2020; Suzuki and Larkum, 2020).
Neural circuit mechanisms of the CT stimulation effects
Our neuroanatomical observations on the marmoset CT may be relevant for the discussion of the neural circuit mechanisms that underlie the effect of CT stimulation on consciousness.
For example, CT stimulation in nonhuman primate models has been shown to induce simultaneous cortical activation in premotor, prefrontal, parietal, and cingulate areas (Tasserie et al., 2022). We show here that all these areas are innervated by the combination of thalamocortical axons originated in CT, particularly from its paralaminar VLp and MDL components. Even if the stimulation probe is positioned in the IML, a critical mass of cells in VLp and/or MDl might be also recruited when the appropriate frequencies or intensities are applied (Liu et al., 2015; Baker et al., 2016; Tasserie et al., 2022).
Likewise, clinical studies in comatose patients and patients in a minimally conscious state have shown that activity of thalamocortical neurons is significantly reduced and dominated by tonic firing mode, instead of the increased inhibition and burst firing that is observed in the physiological unconsciousness of sleep (Magrassi et al., 2018). Conceivably, electrical activation of intralaminar and paralaminar thalamocortical axons could increase the membrane depolarization in frontal and parietal areas to a level sufficient to maintain effective signal transfer between these and other areas (Dehaene and Changeux, 2011; Redinbaugh et al., 2020). The thalamocortical paths activated by CT stimulation would thus differ essentially from the pathways involved in the physiological transition from sleep to the awake state (Saper et al., 2001; Ren et al., 2018).
In addition, our data suggest which thalamic populations might be most directly involved in the cortical activation (Tasserie et al., 2022). For example, the CM axons innervate densely and focally the striatum, and very weakly the cortex. In addition to the cortex, the CL–Pc axons innervate the sensorimotor striatum. Effects from intralaminar nuclei stimulation could thus be both direct and indirect, the latter involving the multisynaptic cortico-thalamo-striatal loops of the basal ganglia system (Alexander et al., 1986; Martel and Galvan, 2022). Interestingly, recent functional connectivity (Crone et al., 2017) or local field potential (Afrasiabi et al., 2021) studies indicate that continued interactions between the cortex and basal ganglia structures are involved in consciousness preservation.
As we investigated the axonal pathways originated from the neurons situated within or around the IML, our data do not rule out that the CT stimulation effects might also involve the anterograde or retrograde activation of some of the fiber systems that travel in the IML from the cerebellum (Asanuma et al., 1983), substantia nigra (Sidibé et al., 2002), pallidum (Sakai et al., 1996; Sidibé et al., 2002), superior colliculus (Graham, 1977), or brainstem reticular formation (Glenn and Steriade, 1982; Steriade and Glenn, 1982; Vogt et al., 2008).
Moreover, our axon-labeling data involve only the ipsilateral monosynaptic pathways originated from neurons whose soma is located within CT. However, it now clear that thalamic cells are high-centrality nodes within multiple large-scale multiregional brain networks (Acsády, 2022; Clascá, 2022). Consequently, behavioral effects of CT DBS can involve not only the direct connections described here, but also a cascade of multisynaptic, and often reciprocal, links across both hemispheres of the brain.
Overall, our data are consistent with the possibility that electrical pulses delivered in CT affect a collection of intralaminar and paralaminar thalamic cell groups that, together, innervate an array of frontal, parietal, and cingulate areas involved in high-level movement planning and guidance (Tanji and Shima, 1994; Assad and Maunsell, 1995), attention (Boshra and Kastner, 2022), and working memory (Bolkan et al., 2017). Interestingly, a coordinated activation of these high-level areas is correlated with conscious perception of sensory stimuli (Dehaene and Changeux, 2011).
Footnotes
This work was supported by Grant PVE 400730/2014-6 from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, Brasil; to J.S.C.) and Grant PHB-H2012-0011 Ministerio de Educación (Spain; to F.C.). We thank Professor John F. Araujo (Federal University of Rio Grande do Norte) for support in the first steps of this project and Begoña Rodriguez (Universidad Autónoma de Madrid) for expert technical help.
The authors declare no competing financial interests.
- Correspondence should be addressed to Francisco Clascá at francisco.clasca{at}uam.es
