Abstract

Input to apical dendritic tufts is now deemed crucial for associative learning, attention, and similar “feedback” interactions in the cerebral cortex. Excitatory input to apical tufts in neocortical layer 1 has been traditionally assumed to be predominantly cortical, as thalamic pathways directed to this layer were regarded relatively scant and diffuse. However, the sensitive tracing methods used in the present study show that, throughout the rat neocortex, large numbers (mean ∼4500/mm2) of thalamocortical neurons converge in layer 1 and that this convergence gives rise to a very high local density of thalamic terminals. Moreover, we show that the layer 1–projecting neurons are present in large numbers in most, but not all, motor, association, limbic, and sensory nuclei of the rodent thalamus. Some layer 1–projecting axons branch to innervate large swaths of the cerebral hemisphere, whereas others arborize within only a single cortical area. Present data imply that realistic modeling of cortical circuitry should factor in a dense axonal canopy carrying highly convergent thalamocortical input to pyramidal cell apical tufts. In addition, they are consistent with the notion that layer 1-projecting axons may be a robust anatomical substrate for extensive “feedback” interactions between cortical areas via the thalamus.

Introduction

A prominent feature of neocortical organization is that the apical dendritic tufts of pyramidal neurons accumulate in layer 1. Inputs reaching these tufts are increasingly viewed as crucial for the “feedback” interactions in the cerebral cortex that underlie cognitive processes such as associative learning and attention (Sjöström and Häusser 2006; Gilbert and Sigman 2007). Excitatory input to neocortical layer 1 is generally considered to be predominantly cortical (Cauller 1995; Douglas and Martin 2007).

In addition, it has been known for some time that some thalamic axons arborize in layer 1 (Lorente de No 1938; Killackey and Ebner 1973; Herkenham 1979), but these were widely assumed to represent a relatively minor and diffuse (“nonspecific”; Lorente de No 1938) input to cortex. These layer 1–directed pathways have been persistently associated in the literature with the intralaminar and midline thalamic nuclei, despite substantial evidence to the contrary (for reviews, see Herkenham 1986; Avendaño et al. 1990).

Some years ago, fragmentary observations in carnivores and primates led to the hypothesis (Avendaño et al. 1990; Jones 1998) that the thalamocortical layer 1–projecting neurons might actually constitute a substantial cell population originating throughout the thalamus (“matrix” or M-type cells). Importantly, it was noted that the function of these thalamocortical cells would probably be different from the better-known cells that primarily target cortical layer 4 (“core” or C-type): not only did the laminar segregation of the axons imply that each cell type would reach different elements of the cortical circuitry but there was also evidence that each type received different ascending inputs, had a different soma size, and/or expressed different calcium-binding proteins (Jones 2001).

The M-type/C-type dichotomy has recently been singled out as a key organizational principle in thalamocortical pathways (Jones 2007). However, information regarding M-type pathways remains quite scant and fragmentary. There is considerable confusion regarding where exactly M-type neurons are located in the thalamus (Llinás et al. 2002; Kubota et al. 2007) and how prevalent they are. Likewise, evidence regarding the strength of M-type input to the cortex is conflicting. On one hand, thalamocortical pathways targeting layer 1 seem to be relatively few, as indicated by the small numbers of thalamic neurons labeled by retrograde tracer injections in superficial cortical layers in comparison to the number labeled by injections involving the middle layers (reviewed in Avendaño et al. 1990) and by the laminar pattern of the cortical labeling produced by a tracer deposit in most thalamic nuclei (Herkenham 1986; Jones 2007). Moreover, individual thalamic axons in layer 1 reportedly have relatively few branches and sparse boutons (Ferster and LeVay 1978; Arbuthnott et al. 1990; Lu and Lin 1993; Portera-Cailliau et al. 2005). On the other hand, recent immunolabeling observations for the thalamocortical terminal marker vesicular glutamate transporter type 2 (VGluT2; Fujiyama et al. 2001; Kubota et al. 2007) seem to imply that thalamic innervation of layer 1 is substantial.

Here, we examined the origin and convergence/divergence of M-type pathways in the rat cerebral cortex using sensitive pathway-tracing and analysis methods. We found that M-type cells constitute a very sizable population of the thalamus. Furthermore, we demonstrate that M-type cell innervation in cortical layer 1 is both massive and highly convergent, in combinations that are specific for the various areas. We also show that whereas many M-type axons branch over wide swaths of the cerebral hemisphere others arborize within a single area.

Materials and Methods

Animals and Anesthetic Procedures

The brains from a total of 75 adult (4–12 months of age) female Sprague–Dawley rats obtained from the School of Medicine Animal Facility were used for the various experiments reported in this study. From these, the 37 animals with one or more tracer deposit experiments valid for analysis are listed in Supplementary Table SM1. Procedures were carried out in accordance with the Society for Neuroscience Handbook on the Use of Animals in Neuroscience Research and the European Community Council Directive 86 609 and were approved by our university's Bioethics Committee. All surgical procedures were conducted under a combination of ketamine (66 mg/kg, intramuscularly [i.m.]) and xylazine (8 mg/kg, i.m.). Additional partial doses were administered as required during the surgical procedure to keep the animal areflexic, but breathing spontaneously. Throughout the postoperative period, animals were given amoxicillin (3 mg/kg/day). At the time of sacrifice, animals were overdosed with sodium pentobarbital (80 mg/kg, intraperitoneally).

Delivery of Axonal Tracers

To retrogradely label the somata of all thalamocortical neurons innervating a given zone in layer 1, we made epipial applications of Fast Blue (FB, diamidino compound 253/50, Dr Illing GmbH and Co. KG, Groß-Umstadt, Germany) or Diamidino Yellow Dihydrochloride (DY, Dr Illing). Surgical procedures have been described in detail elsewhere (Rubio-Garrido et al. 2007) and will only be briefly explained here. Anesthetized animals were positioned in a stereotaxic frame. A unilateral craniotomy was made, and one or more small tracer-soaked pieces of filter paper were placed on the pia mater of the neocortex. The bone defect was then occluded, the skin and muscle sutured, and the animals allowed to recover.

To anterogradely label thalamocortical axons in layer 1, we made unilateral iontophoretic microinjections of 10 000 Da lysine-fixable biotinylated dextran-amine (BDA, Molecular Probes-Invitrogen, Carlsbad, CA) in the thalamus. Animals were operated as above. A borosilicate micropipette (10- to 25-μm–inner diameter tip; FHC, Bowdoin, ME) loaded with a 10% solution of BDA in 0.01 M phosphate buffer (PB), pH 7.4, was stereotaxically positioned into the ventromedial, the lateral posteromedial, or the dorsal lateral geniculate thalamic nuclei. Using a Midgard Current Source (Stoelting, Wood Dale, IL) square pulses of a 2μA current were delivered for 10 min (7 s on/off cycles). The pipette was left in place for another 15 min and then removed. Bone and skin were closed as above.

Perfusion and Histology

In the experiments with retrograde fluorescent tracers, animals were allowed to survive for 7 days after the FB application; they were then sacrificed and perfused through the left ventricle with saline (5 min) followed by cold (∼4 °C) 4% paraformaldehyde in PB for 45 min. Subsequently, brains were blocked along a coronal plane, removed from the skull, and postfixed by immersion in the same fixative for 24 h at 4 °C. Tissue blocks were cryoprotected by soaking in a sucrose solution (30% in PB) until they sank. Parallel series of 40-μm-thick coronal sections were obtained on a Leica freezing microtome. One series (1 in every 5 sections, spaced 160 μm) was immediately mounted onto gelatin-coated glass slides, air-dried, dehydrated, and coverslipped with DePeX (Serva, Heidelberg, Germany). To allow for a precise delineation of thalamic nuclei and cortical areas, additional parallel series were stained with cresyl violet or else histochemically reacted for either acetylcholinesterase (AChE; Geneser-Jensen and Blackstadt 1971) or cytochrome oxidase (CyO; Wong-Riley et al. 1979) activities. Finally, sections were mounted, dehydrated, and coverslipped as above.

The brain of 3 animals injected in the thalamus with BDA were fixed and coronally sectioned as described for FB experiments (Supplementary Table SM1). In the remaining 9 BDA-injected animals, the cerebral hemispheres were flattened to allow the visualization of labeled axon arbors on a plane that was parallel to the pial surface. To this end, following the initial saline perfusion, the animals were perfused with 1% paraformaldehyde in PB for 3 min followed by 3 min of PB alone. After removal from the skull, the cerebral hemispheres were separated from the diencephalon, and the hippocampus and striatum were removed. The whole cerebral cortex was then placed between 2 glass slides and gently pressed to flatten it, after which it was immersed in cold 4% paraformaldehyde for 48 h. The diencephalon was postfixed, cryoprotected, and coronally sectioned as described above. For reference, the cerebral hemispheres from 2 additional unoperated animals were flattened in the same way.

Thalami containing the BDA injection sites were coronally sectioned and processed with a standard avidin–biotin–peroxidase complex (ABC; Vector Laboratories, Burlingame, CA) + diaminobenzidine (DAB) protocol. Sections were finally mounted on gelatin-coated slides, air-dried, thionin counterstained, and coverslipped. Instead of counterstaining, in 2 cases, thalamic sections were histochemically processed to reveal CyO activity.

To reveal the anterogradely transported BDA in subpial axon arbors, frozen sections of the flattened hemispheres were cut parallel to the pia mater at 60 μm. Labeling was developed on the free-floating sections with ABC + DAB. Sections were then mounted, thionin counterstained, and coverslipped with DePeX. The cerebral hemispheres from the 2 brains that had been flattened and fixed without tracer injections were tangentially sectioned at 60 μm, processed to reveal CyO activity, and mounted and coverslipped as above. This labeling allowed the delineation of cortical domains such as the barrel field in the primary somatic cortex or the primary auditory and visual areas, which could then be correlated with the position of the medial cerebral artery and its branches in the tangential sections.

In the remaining 4 animals that had received BDA injections in the thalamus, labeled axons were directly developed “in toto,” without tissue sectioning. These whole flattened hemispheres were incubated free floating in 2% Triton + ABC in PB for 4 h, washed, and subsequently stained using a standard DAB protocol.

Retrograde Fluorescent Labeling Analysis

Fluorescent labeling was examined with a Nikon 600 Eclipse microscope under 4–40× Wide-Diameter Plan-Apochomat Nikon objectives and a 100-W Hg UV light source. Using a Nikon DMX 1200 digital camera, high-resolution (1490 dpi) RGB TIFF epifluorescence images for each section in every valid case were acquired through the 10× objective and a UV2A Nikon filter set. By means of a motorized stage (Proscan, Prior Scientific Instruments, Cambridge, UK) driven by the Analysis software (Soft Imaging Systems, Münster, Germany) that controls both image acquisition and microscope stage motion, 24–32 epifluorescence high-resolution images were automatically assembled into a single panoramic view of the thalamus (∼100 Mb JPG file). By zooming into these brightly fluorescent images, a thorough high-magnification analysis could be carried out without any photobleaching.

Because the excitation and emission curves of FB and DY overlap extensively, we found it impossible to separate both tracers consistently and unambiguously in our material using a Leica SP2 spectral confocal microscope. Whereas heavily double-labeled cells could be readily identified by the preferential localization of DY or FB to either the nucleus or the cytoplasm, double labeling in weakly labeled cells was often uncertain. As weakly labeled cells represented a significant fraction of the cells labeled in our experiments, we chose to keep the double-labeling analysis qualitative to avoid a substantial underestimation (Schofield et al. 2007). Therefore, present data reflect only the presence or absence of unambiguously double-labeled cells in a given thalamic nucleus but by no means the actual number of double-labeled cells.

A complete series of sections containing the cortex impregnated by the epipial deposit was examined. Deposits produced an intense but superficially restricted tracer impregnation. Any deposit experiments showing signs of tracer spread into cortical layer 2 were discarded (n = 11). Thus, a total of 32 valid different FB or DY epipial deposit experiments were analyzed for thalamic labeling (see list in Supplementary Table SM1). In addition, the depth of tracer uptake was further assessed from the pattern of retrograde labeling in layers 2–7: massive cell somata labeling in layers 2–3, 5, and 7 and a virtual absence of labeling in layers 4 and 6. This pattern reflects the presence in layer 1 of apical dendrites/axon terminals that have originated in the labeled layers (Fig. 1C; Cauller et al. 1998; Rubio-Garrido et al. 2007).

Figure 1.

Epipial fluorochrome applications selectively impregnate a patch of layer 1. (AC) A representative example of a small tracer deposit confined to cerebral cortex layer 1. Adjacent coronal sections taken at the center of a small DY deposit (Case 18b) in the primary visual cortex. Panel (A) shows a Nissl-stained section. Panel (B) shows the epifluorescence image of an adjacent section. Notice the heavily impregnated neuropil in the superficial part of layer 1. At some distance below this zone, numerous cell somata are labeled in layers 2–3, 5, and 7. In contrast, few or no cells are labeled in layers 4 and 6. In panel (C), a line diagram based on panel (B) shows how an epipial deposit works and why the laminar pattern of back labeling observed in deeper layers indicates that tracer uptake occurred only in layer 1. Tracer diffusing from a piece of tracer-soaked filter paper (wavy thick line at the top) applied on the pia mater-impregnated layer 1 (gray shade). Pyramidal neurons whose apical dendrites (layers 2, 3, and 5) or axons (layer 7) arborize selectively in layer 1 are heavily labeled. In contrast, layer 4 or 6 cells, known to not extend axons or dendrites significantly into layer 1, remain unlabeled. (DE) Adjacent coronal sections taken at the center of an FB deposit (Case 20) involving a region of the Cg1 cingulate and M2 motor areas. The laminar distribution of cells labeled underneath the impregnated part of layer 1 is equivalent to that in panels (AC); the narrow unlabeled band at the layer 3–5 border represents layer 4 in these agranular areas. (F) Epifluorescence image of a coronal section from a cerebral hemisphere that received an extensive epipial deposit of FB (Case 41). The pattern of cell somata labeling under the deposit again reveals that tracer uptake was limited to layer 1. (G) Representation on a “flat” cortical surface map of the extent of the 3 deposits shown in panels (AF). For reference, a diagram of a conventional lateral view of the rat cerebral hemisphere is shown. In the flat cortex map, the pial surface of the cerebral hemisphere was “unfolded” by serially aligning the perimeter measurements from all the cortical coronal sections of Paxinos and Watson rat brain atlas (Paxinos and Watson 2005). The perimeters were plotted onto vertical parallel lines aligned along the edge dividing the medial and laterodorsal aspects of the cerebral hemisphere (thick dashed line). Along this axis, perimeter measurements were spaced according to their distance from the interaural coronal plane. The olfactory peduncle is cut (asterisk). Gray shaded areas represent the extent of layer 1 impregnation in each of the 3 experiments illustrated in panels (AF). See Table 1 for abbreviations. Bar = 500 μm.

Figure 1.

Epipial fluorochrome applications selectively impregnate a patch of layer 1. (AC) A representative example of a small tracer deposit confined to cerebral cortex layer 1. Adjacent coronal sections taken at the center of a small DY deposit (Case 18b) in the primary visual cortex. Panel (A) shows a Nissl-stained section. Panel (B) shows the epifluorescence image of an adjacent section. Notice the heavily impregnated neuropil in the superficial part of layer 1. At some distance below this zone, numerous cell somata are labeled in layers 2–3, 5, and 7. In contrast, few or no cells are labeled in layers 4 and 6. In panel (C), a line diagram based on panel (B) shows how an epipial deposit works and why the laminar pattern of back labeling observed in deeper layers indicates that tracer uptake occurred only in layer 1. Tracer diffusing from a piece of tracer-soaked filter paper (wavy thick line at the top) applied on the pia mater-impregnated layer 1 (gray shade). Pyramidal neurons whose apical dendrites (layers 2, 3, and 5) or axons (layer 7) arborize selectively in layer 1 are heavily labeled. In contrast, layer 4 or 6 cells, known to not extend axons or dendrites significantly into layer 1, remain unlabeled. (DE) Adjacent coronal sections taken at the center of an FB deposit (Case 20) involving a region of the Cg1 cingulate and M2 motor areas. The laminar distribution of cells labeled underneath the impregnated part of layer 1 is equivalent to that in panels (AC); the narrow unlabeled band at the layer 3–5 border represents layer 4 in these agranular areas. (F) Epifluorescence image of a coronal section from a cerebral hemisphere that received an extensive epipial deposit of FB (Case 41). The pattern of cell somata labeling under the deposit again reveals that tracer uptake was limited to layer 1. (G) Representation on a “flat” cortical surface map of the extent of the 3 deposits shown in panels (AF). For reference, a diagram of a conventional lateral view of the rat cerebral hemisphere is shown. In the flat cortex map, the pial surface of the cerebral hemisphere was “unfolded” by serially aligning the perimeter measurements from all the cortical coronal sections of Paxinos and Watson rat brain atlas (Paxinos and Watson 2005). The perimeters were plotted onto vertical parallel lines aligned along the edge dividing the medial and laterodorsal aspects of the cerebral hemisphere (thick dashed line). Along this axis, perimeter measurements were spaced according to their distance from the interaural coronal plane. The olfactory peduncle is cut (asterisk). Gray shaded areas represent the extent of layer 1 impregnation in each of the 3 experiments illustrated in panels (AF). See Table 1 for abbreviations. Bar = 500 μm.

Table 1

Anatomical nomenclature abbreviations

AD: anterodorsal nucleus 
AID: agranular insular cortex, dorsal part 
AIP: agranular insular cortex, posterior part 
AIV: agranular insular cortex, ventral part 
AM: anteromedial nucleus 
Au1: primary auditory cortex 
AuD: secondary auditory cortex, dorsal area 
AuV: secondary auditory cortex, ventral area 
AV: anteroventral nucleus 
Cg1: motor cortex, cingulate field 
Cg2: cingulate cortex area 2 
CL: central lateral nucleus 
CM: central medial nucleus 
DI: dysgranular insular cortex 
DLG: dorsal lateral geniculate nucleus 
DLO: dorsolateral orbital cortex 
DP: dorsal peduncular cortex 
Ect: ectorhinal cortex 
FrA: frontal association cortex 
GI: granular insular cortex 
IAM: interanteromedial nucleus 
IL: infralimbic cortex 
IMA: intramedullary thalamic area 
IMD: intermediodorsal thalamic nucleus 
LD: laterodorsal nucleus 
LDDM: laterodorsal nucleus, dorsomedial part 
LDVL: laterodorsal nucleus, ventrolateral part 
LEnt: lateral entorhinal cortex 
LO: lateral orbital cortex 
LPLC: lateral posterior nucleus, caudal part 
LPLC: lateral posterior nucleus, laterocaudal part 
LPLR: lateral posterior nucleus, laterorostral part 
LPMC: lateral posterior nucleus, mediocaudal part 
LPMR: lateral posterior nucleus, mediorostral part 
M1: motor cortex, lateral agranular field 
M2: motor cortex, medial agranular field 
MEnt: medial entorhinal cortex 
MGD: medial geniculate nucleus, dorsal part 
MGM: medial geniculate nucleus, medial part 
MGV: medial geniculate nucleus, ventral part 
MO: medial orbital cortex 
MZMG: marginal zone of the medial geniculate 
PC: paracentral nucleus 
Pf: parafascicular nucleus 
Pir: piriform cortex 
PM: pia mater 
Po: posterior nucleus 
PoT: posterior nucleus, triangular division 
PrL: prelimbic cortex 
PT: paratenial nucleus 
PtA: parietal area 
PVA: paraventricular nucleus, anterior part 
PVP: paraventricular nucleus, posterior part 
Re: reuniens nucleus 
Rh: rhomboid nucleus 
RSA: retrosplenial dysgranular cortex 
RSGa: retrosplenial granular cortex, “a” region 
RSGb: retrosplenial granular cortex “b” region 
Rt: reticular thalamic nucleus 
S1: primary somatosensory cortex 
S1fl: S1, forelimb field 
S1bf: S1, mystacial vibrisae field 
S2: secondary somatosensory cortex 
SG: suprageniculate nucleus 
SPFPC: subparafascicular nucleus, parvicellular part 
Sub: submedius nucleus 
TeA: temporal association cortex 
V1: primary visual cortex 
V2L: secondary visual cortex, lateral area 
V2M: secondary visual cortex, medial area 
VA: ventral anterior nucleus 
VL: ventrolateral nucleus 
VLG: ventral lateral geniculate nucleus 
VO: ventral orbital cortex 
VPL: ventral posterolateral nucleus 
VPM: ventral posteromedial nucleus 
VPPC: ventral posterior nucleus, parvicellular part 
VRe: ventral reuniens nucleus 
WM: subcortical white matter 
ZI: zona incerta 
AD: anterodorsal nucleus 
AID: agranular insular cortex, dorsal part 
AIP: agranular insular cortex, posterior part 
AIV: agranular insular cortex, ventral part 
AM: anteromedial nucleus 
Au1: primary auditory cortex 
AuD: secondary auditory cortex, dorsal area 
AuV: secondary auditory cortex, ventral area 
AV: anteroventral nucleus 
Cg1: motor cortex, cingulate field 
Cg2: cingulate cortex area 2 
CL: central lateral nucleus 
CM: central medial nucleus 
DI: dysgranular insular cortex 
DLG: dorsal lateral geniculate nucleus 
DLO: dorsolateral orbital cortex 
DP: dorsal peduncular cortex 
Ect: ectorhinal cortex 
FrA: frontal association cortex 
GI: granular insular cortex 
IAM: interanteromedial nucleus 
IL: infralimbic cortex 
IMA: intramedullary thalamic area 
IMD: intermediodorsal thalamic nucleus 
LD: laterodorsal nucleus 
LDDM: laterodorsal nucleus, dorsomedial part 
LDVL: laterodorsal nucleus, ventrolateral part 
LEnt: lateral entorhinal cortex 
LO: lateral orbital cortex 
LPLC: lateral posterior nucleus, caudal part 
LPLC: lateral posterior nucleus, laterocaudal part 
LPLR: lateral posterior nucleus, laterorostral part 
LPMC: lateral posterior nucleus, mediocaudal part 
LPMR: lateral posterior nucleus, mediorostral part 
M1: motor cortex, lateral agranular field 
M2: motor cortex, medial agranular field 
MEnt: medial entorhinal cortex 
MGD: medial geniculate nucleus, dorsal part 
MGM: medial geniculate nucleus, medial part 
MGV: medial geniculate nucleus, ventral part 
MO: medial orbital cortex 
MZMG: marginal zone of the medial geniculate 
PC: paracentral nucleus 
Pf: parafascicular nucleus 
Pir: piriform cortex 
PM: pia mater 
Po: posterior nucleus 
PoT: posterior nucleus, triangular division 
PrL: prelimbic cortex 
PT: paratenial nucleus 
PtA: parietal area 
PVA: paraventricular nucleus, anterior part 
PVP: paraventricular nucleus, posterior part 
Re: reuniens nucleus 
Rh: rhomboid nucleus 
RSA: retrosplenial dysgranular cortex 
RSGa: retrosplenial granular cortex, “a” region 
RSGb: retrosplenial granular cortex “b” region 
Rt: reticular thalamic nucleus 
S1: primary somatosensory cortex 
S1fl: S1, forelimb field 
S1bf: S1, mystacial vibrisae field 
S2: secondary somatosensory cortex 
SG: suprageniculate nucleus 
SPFPC: subparafascicular nucleus, parvicellular part 
Sub: submedius nucleus 
TeA: temporal association cortex 
V1: primary visual cortex 
V2L: secondary visual cortex, lateral area 
V2M: secondary visual cortex, medial area 
VA: ventral anterior nucleus 
VL: ventrolateral nucleus 
VLG: ventral lateral geniculate nucleus 
VO: ventral orbital cortex 
VPL: ventral posterolateral nucleus 
VPM: ventral posteromedial nucleus 
VPPC: ventral posterior nucleus, parvicellular part 
VRe: ventral reuniens nucleus 
WM: subcortical white matter 
ZI: zona incerta 

In some experiments, extensive deposits covering a wide area of the dorsal and lateral cortices were made and revealed a large fraction of the M-type cell population in a single experiment (Fig. 1E–F). Other experiments involved small single or double FB/DY deposits in different points of the cerebral hemisphere. The labeling patterns produced by these small injections were used to estimate the convergence of M-type axons as well as to extend the mapping of M-type neurons to cortical regions that were not explored by the single large deposits (Fig. 1A–E,G, see also Supplementary Fig. SM1).

For comparison between experiments, the zone of layer 1 that was impregnated in each of the valid experiments was reconstructed from the serial coronal sections onto a standard “flat” map of the pial surface of the whole cerebral hemisphere (Fig. 1F; Paxinos and Watson 2005). Our valid deposits sampled most of the motor, somatic, auditory, and visual cortices, as well as portions of the medial frontal, retrosplenial, and insular cortices; all told, they covered nearly 60% of the neocortical layer 1 surface. Only the frontal and occipital poles and some zones of the medial cortex or around the rhinal sulcus were not explored, as we found them to be practically inaccessible using this technique.

For each valid case, thalamic labeling was examined and drawn from a complete series of sections. Using vectorial graphics software (Canvas X, ACD Systems, Saanichton, British Columbia, Canada) on Apple computers, nuclear borders were drawn by successively overlaying the digital drawing from the section containing the labeling onto thalamus images from adjacent Nissl-, AChE-, and CyO-stained sections. Because many of the labeled cells were situated at or near nuclear boundaries, this procedure was crucial for consistently ascribing the labeled cells to a given nucleus. We followed the atlas of Paxinos and Watson (2005) for nomenclature and delineation of nuclei. The position of every labeled cell soma was plotted onto the digital drawings from the images. Drawn cells were automatically counted using the software tools.

BDA Labeling Analysis

The location and extent of the BDA injection sites in the thalamus was reconstructed from the serial coronal sections of the diencephalon.

In the flat-sectioned cerebral hemispheres, BDA-labeled thalamocortical axons were examined and photographed using bright- and dark-field optics. Axons in layer 1 were reconstructed from serial digital images taken with the motorized microscope stage as described. We used dark-field optics to better visualize these thin axons. Images were taken with the 10× objective to maximize the depth of focus. To cover the extensive flattened tissue sections, large composite image files were made from 300 to 600 individual images (600 dpi gray scale JPEG) using Analysis and Canvas software. By zooming into the images, we drew labeled axons in layer 1; axons in deeper tissue zones (revealed by the presence of cell somata by the thionin counterstain) were ignored. Because of the impossibility of resolving branching points from intersections in axon arbors at this magnification, labeled axon arbors were drawn as consisting of independent segments. The middle cerebral artery and its branches, as well as tissue landmarks such as the rhinal sulcus or lateral olfactory tract, were used as references for assembling the sections into a panoramic view of the labeled axons over the whole hemisphere. The superior cerebral veins were used as reference for locating the dorsolateral/medial watershed of the hemisphere.

The hemispheres processed “in toto” to reveal BDA were kept floating in 1% sodium azide in PB at 4 °C. For examination, they were transferred to a PB-filled chamber mounted on the microscope stage, flattened under a coverslip, transilluminated, and examined under the 10× objective. Mosaic images were generated and the labeling reconstructed as above. In this unsectioned, non-defattened, and water-mounted tissue, only the labeled axons near the pial surface were visible. Moreover, because of nondehydrated tissue thickness, only a fraction of the labeled axons could be captured in a single focal plane at a microscope magnification sufficient for axon visualization. These factors probably contributed to a considerable underestimation of the labeled axons in layer 1 in these experiments.

VGluT2 Immunohistochemistry

To mass label thalamocortical synaptic sites in the neocortex as an additional reference comparison with our experiments, we immunostained some tissue sections for VGluT2. Coronal cortical sections from 3 animals were immunolabeled with an affinity-purified rabbit anti-VGlut2 serum (1:500 Chemicon, Temecula, CA; Fujiyama et al. 2001). Labeling was visualized via linkage to a biotinylated goat anti-rabbit serum (1:100 Chemicon) following a standard avidin–biotin–peroxidase staining protocol with DAB as chromogen. Parallel sections in which the primary antiserum was substituted by normal rabbit serum yielded no labeling.

Results

Many Thalamic Nuclei Contain Large Numbers of Layer 1–Projecting Thalamocortical Neurons

A large fraction of the M-type cell population was visualized by means of retrograde tracer experiments that created a deposit that was very extensive but remained confined to layer 1. In addition, a large series of small tracer deposits were used to analyze in detail the spatial distribution of M-type inputs within the zones sampled by the large deposits as well as to probe M-type input to areas not reached by the extensive deposits (Fig. 1, see also Supplementary Fig. SM1).

A typical extensive layer 1 deposit is illustrated in Figs 1F–G and 2. In this experiment, the tracer-impregnated layer 1 in most of the lateral frontal and parietal cortices as well as sectors of the temporal and occipital cortices. Massive numbers of cells were labeled in the ventrolateral, ventromedial, anteromedial, posterior, and lateral posterior thalamic nuclei, and many cells were also labeled in the dorsal lateral geniculate, laterodorsal, reuniens, medial geniculate, and anterior intralaminar nuclei. In spite of the large numbers of cells labeled, their distribution was far from uniform. For example, in nuclei such as the ventrolateral or dorsal lateral geniculate nuclei, labeled cells were densely concentrated in one portion of the nucleus, whereas other parts were almost unlabeled. Other nuclei that were virtually unlabeled were the ventral posterior, dorsomedial, anterior, and gelatinous nuclei. Absence of labeling in the ventroposterior nucleus was all the more remarkable because the tracer deposit covered the whole S1 area. Thalamic labeling in the 2 additional experiments with extensive layer 1 deposits in lateral and dorsal neocortical areas (Supplementary Fig. SM1 and Table SM1) confirmed these observations.

Figure 2.

The layer 1–projecting thalamocortical cells are a large population spread throughout the thalamus. (AH) Line drawings of representative coronal thalamic sections. Each black dot represents one labeled cell soma. Fiber bundles such as the mamillothalamic tract, the stria medularis, the retroflex bundle, or the medial lemniscus are shaded in gray. Note that although the tracer deposit covered most of area S1, the ventroposterior nucleus (VP) is largely unlabeled. (A′, B′, F′, G′) Epifluorescence images of the labeling in the rectangles in, respectively, panels (A, B, F, and G). See Table 1 for abbreviations. Further details of the labeled cells can be appreciated by zooming into the images. A high-magnification epifluorescence image of labeled cells is shown in Supplementary Figure SM3. Bars: (AH) = 1 mm; (A′B′), (G′H′) = 500 μm.

Figure 2.

The layer 1–projecting thalamocortical cells are a large population spread throughout the thalamus. (AH) Line drawings of representative coronal thalamic sections. Each black dot represents one labeled cell soma. Fiber bundles such as the mamillothalamic tract, the stria medularis, the retroflex bundle, or the medial lemniscus are shaded in gray. Note that although the tracer deposit covered most of area S1, the ventroposterior nucleus (VP) is largely unlabeled. (A′, B′, F′, G′) Epifluorescence images of the labeling in the rectangles in, respectively, panels (A, B, F, and G). See Table 1 for abbreviations. Further details of the labeled cells can be appreciated by zooming into the images. A high-magnification epifluorescence image of labeled cells is shown in Supplementary Figure SM3. Bars: (AH) = 1 mm; (A′B′), (G′H′) = 500 μm.

Small retrograde tracer deposit in layer 1 in areas not reached by the large deposits yielded labeling in many of the thalamic nuclei already labeled by the large deposits, but also in some additional nuclei. For example, following an insular area deposit (Fig. 3A), labeled cells appeared in the dorsomedial and centralis medialis nuclei; likewise, a deposit in the medial frontal cortex (Fig. 3B) labeled cells in the dorsomedial nucleus; and a deposit in the retrosplenial cortex labeled many cells in the lateral dorsal as well as the anterior nuclei (Fig. 3H).

Figure 3.

In every cortical area, small tracer deposits in layer 1 back label multiple thalamic nuclei. (AI) Comparison of the thalamic labeling in 9 representative experiments, each of which received a small layer 1 deposit in a different cortical area. Thalamic labeling is shown in 6 coronal sections arranged from rostral (left) to caudal (right). Each black dot represents a single cell. Note that some nuclei, like the ventromedial (VM), anteromedial (AM), or lateral posterior medial rostral part (LPMR), contained labeled cells in most or virtually all cortical regions following the layer 1 deposit. In contrast, nuclei such as the dorsal lateral geniculate (DLG, row F) or the medial geniculate (MGD, row I) were labeled only by layer 1 deposits in a particular area. See Table 1 for other abbreviations. Bar = 1 mm.

Figure 3.

In every cortical area, small tracer deposits in layer 1 back label multiple thalamic nuclei. (AI) Comparison of the thalamic labeling in 9 representative experiments, each of which received a small layer 1 deposit in a different cortical area. Thalamic labeling is shown in 6 coronal sections arranged from rostral (left) to caudal (right). Each black dot represents a single cell. Note that some nuclei, like the ventromedial (VM), anteromedial (AM), or lateral posterior medial rostral part (LPMR), contained labeled cells in most or virtually all cortical regions following the layer 1 deposit. In contrast, nuclei such as the dorsal lateral geniculate (DLG, row F) or the medial geniculate (MGD, row I) were labeled only by layer 1 deposits in a particular area. See Table 1 for other abbreviations. Bar = 1 mm.

Single small tracer deposits situated in different locations within the same dorsolateral cortical zone already explored with the extensive tracer deposit experiments showed fewer labeled cells than after the extensive deposits; however, the patterns of cell distribution within the thalamus were strictly consistent (compare Figs 2 and 3C–G).

The combined evidence from all the layer 1 deposits, large and small, shows that the M-type thalamocortical cell population is a large one and is widely distributed in the thalamus. Although our experiments did not address the proportion that M-type cells represent over the total neurons of each thalamic nuclei, the large number of labeled cells in some nuclei strongly suggests that in them M-type cells form the majority. In contrast, some thalamic nuclei such as the ventral posterior nuclear complex, the ventral division of the medial geniculate, the ventral–medial portions of the dorsal lateral geniculate, submedius, anterodorsal, and anteroventral nuclei remained largely free of labeled cells in every experiment.

Thalamic Input to Layer 1 Is a Specific Combination of Afferents from Many Nuclei

Visual comparison of neuron distribution in experiments involving small deposits in different cortical areas (Fig. 3) indicated that the combination of thalamocortical inputs reaching layer 1 varies substantially between areas. For a direct comparison that would be unaffected by differences in absolute neuron number between experiments, we compared the percentage of neurons labeled in each nucleus to the total number of labeled cells in the thalamus (Fig. 4). This analysis clearly shows that thalamic input reaching layer 1 in each cortical area is a combination, in specific proportions, of afferents from a wide array of nuclei.

Figure 4.

Semiquantitative analysis of the distribution of M-type thalamocortical cells projecting to different cortical regions. Upper left corner: flat cortex map showing 12 small epipial deposit experiments, each in a different cortical region. To normalize and compare across these experiments, the percentages that the cells labeled in each nuclei represent over the total for the thalamic cells labeled in each experiment are displayed as levels of gray in the table at the left. In the table, nuclei are listed according to broad functional affinities. The experiment number is indicated atop each column. This analysis confirms the impression gained from the observation of individual thalamic sections that some nuclei disperse their axons over vast swaths of layer 1, whereas other nuclei innervate layer 1 in a few or in even only one area. See Table 1 for abbreviations.

Figure 4.

Semiquantitative analysis of the distribution of M-type thalamocortical cells projecting to different cortical regions. Upper left corner: flat cortex map showing 12 small epipial deposit experiments, each in a different cortical region. To normalize and compare across these experiments, the percentages that the cells labeled in each nuclei represent over the total for the thalamic cells labeled in each experiment are displayed as levels of gray in the table at the left. In the table, nuclei are listed according to broad functional affinities. The experiment number is indicated atop each column. This analysis confirms the impression gained from the observation of individual thalamic sections that some nuclei disperse their axons over vast swaths of layer 1, whereas other nuclei innervate layer 1 in a few or in even only one area. See Table 1 for abbreviations.

In addition, the comparison of the thalamic patterns of labeling in our series of small deposit experiments revealed marked differences in the degree of divergence along the cortical surface of M-type pathways originating in different nuclei (Figs 3 and 4). For example, M-type neurons in nuclei such as the dorsomedial, anterodorsal, or dorsal lateral geniculate were labeled only by layer 1 deposits in a particular cortical field, indicating a limited tangential spread of their axons. In contrast, nuclei such as the ventromedial, the lateral posteromedial, or the posterior nucleus were labeled from deposits located in a variety of cortical areas, suggesting that their axons diverge broadly in the cortex.

Innervation of wide cortical areas from the latter nuclei could result from a widespread dispersal of individual axons, each innervating a single cortical zone, or from the divergent branching of single axons to reach multiple zones. To distinguish between these 2 possibilities at the population level, we placed epipial deposits of FB or DY in distant regions of the cerebral hemisphere in a limited number of experiments (n = 5; experiments numbered with “a” or “b” suffixes in Supplementary Table SM1) and then looked for double-labeled neurons in the thalamus (Supplementary Figs SM2 and SM3DE). Double-labeled cells were found in only a handful of thalamic nuclei, and interestingly, these nuclei (highlighted by asterisks in the table in Fig. 4) were precisely those showing the widest tangential axon dispersal. Because our double deposits were few, relatively small, and placed without a specific pattern, the implication seems to be that the number of cells in these nuclei that branched their axons into widely separated target areas is substantial.

On the other hand, even in the nuclei with axons diverging widely in the cortex, the position of cells labeled in different experiments reveals a crude topographical order. For example, deposits on medial frontal areas labeled dorsal portions of the ventromedial nucleus (Fig. 3A), whereas deposits on parietal areas labeled more caudolateral portions of this nucleus (Fig. 3D). Thus, even if most cells in the ventromedial nucleus branch their axons toward multiple cortical areas, there still seems to be a crude topographic order to these pathways.

Large Numbers of Different Thalamic Axons Converge in Every Point of Layer 1

A striking observation following small layer 1 deposits was that, even when the impregnated layer 1 area was very small, numbers of labeled thalamic cells were consistently high (Fig. 5). Moreover, this observation was repeated in every cortical area explored. The implication is that large numbers of M-type axons converge within every small region of layer 1.

Figure 5.

A small tracer deposit in layer 1 labels a large number of neurons in the thalamus. (A) Extent of a typical small deposit reconstructed on a cortical surface map. In this particular experiment (Case 20, see Fig. 1A–B), the FB deposit impregnated 1.8 mm2 of cortical layer 1. (BF) Panoramic epifluorescence images showing retrogradely labeled neuron somata in the thalamus. The images correspond to the regions indicated by rectangles on the line drawings in panel (G). Further details of the labeled cells can be appreciated by zooming into the images, as well as in Supplementary Figure SM3. (G) Line drawings from an entire series of sections (1 out of 5, 40-μm-thick each, spaced by 200 μm apart). Sections are arranged from rostral (upper left corner) to caudal (lower right corner). Each dot represents a single cell. The total number of neurons labeled in this particular series of thalamic sections was 2211. See Table 1 for abbreviations. Bar: (A) = 2 mm; (BF) = 500 μm; (G) =1 mm.

Figure 5.

A small tracer deposit in layer 1 labels a large number of neurons in the thalamus. (A) Extent of a typical small deposit reconstructed on a cortical surface map. In this particular experiment (Case 20, see Fig. 1A–B), the FB deposit impregnated 1.8 mm2 of cortical layer 1. (BF) Panoramic epifluorescence images showing retrogradely labeled neuron somata in the thalamus. The images correspond to the regions indicated by rectangles on the line drawings in panel (G). Further details of the labeled cells can be appreciated by zooming into the images, as well as in Supplementary Figure SM3. (G) Line drawings from an entire series of sections (1 out of 5, 40-μm-thick each, spaced by 200 μm apart). Sections are arranged from rostral (upper left corner) to caudal (lower right corner). Each dot represents a single cell. The total number of neurons labeled in this particular series of thalamic sections was 2211. See Table 1 for abbreviations. Bar: (A) = 2 mm; (BF) = 500 μm; (G) =1 mm.

The methods employed do not allow a precise quantitative assessment of the number of thalamic neurons converging in any given region of layer 1; however, as an approximation, we counted, for each case, all the labeled neurons in a series of thalamus sections (1:5 sections, each 40 μm thick) and compared the result with the area of layer 1 surface covered by the corresponding cortical deposit reconstructed from the serial cortical sections (Fig. 4). The zone from which tracer uptake and transport effectively occurs following a tracer injection in brain tissue is difficult to determine (Mesulam 1982). However, in our epipial deposits, dendritic back labeling of pyramidal cells in layer 5 provided a reliable internal control for the zone from which tracer transport actually had taken place. Because such back labeling consistently matched the tracer-impregnated surface in layer 1 (Fig. 1B,C,E,F), we are confident that the latter closely corresponds to the area from which tracer transport occurred in each experiment. In addition, to compensate for the shrinkage in surface of the cerebral hemisphere observed between the fresh rat cortex tissue and the fixed, sucrose-embedded, and dry-mounted cortical sections, a correction factor of 1.35 (Avendaño C, personal communication) was applied to the measured surface. This analysis revealed a mean density ratio of 1025 (± 250.5 standard deviation) labeled neurons in the thalamus per square millimeter of fresh layer 1 surface. As cells were counted in only one fifth of the total thalamic tissue, a conservative estimate would then be that the number of different M-type thalamic neurons reaching any given square millimeter of living rat layer 1 must be in the range of 4000–5000. Note that, without correction for surface shrinkage, this density would be even higher.

Finally, neuron numbers in experiments situated in very different points of the cerebral surface, including primary sensory, motor, association, or limbic areas, were compared with the extent of each deposit. This analysis revealed that the massive spatial convergence of thalamic axons in layer 1 is quite similar throughout the cortex (R2 = 0.93; Fig. 6).

Figure 6.

Approximate estimation of the number of cells innervating 1 mm2 of cortical layer 1. Surface area covered by the tracer deposits in each of the 12 experiments shown in Figure 4A after correction for cortical surface shrinkage (Avendaño C, personal communication) was plotted against the total number of retrogradely labeled thalamic neurons counted in a whole rostrocaudal series of thalamus sections (1:5 sections). This density ratio provides a direct estimate of how many thalamocortical cells extend at least some part of their axon into any given 1 mm2 of layer 1. The mean density ratio is 1025 ± 250 cells per mm2. Note, however, that cells were counted in only a fifth of the total thalamic tissue, so the actual number of thalamic cells innervating 1 mm2 of layer 1 must be on the order of 4000–5000. Regression analysis shows a remarkably similar density in every sensory, association, motor, or limbic cortical area examined (R2 = 0.93).

Figure 6.

Approximate estimation of the number of cells innervating 1 mm2 of cortical layer 1. Surface area covered by the tracer deposits in each of the 12 experiments shown in Figure 4A after correction for cortical surface shrinkage (Avendaño C, personal communication) was plotted against the total number of retrogradely labeled thalamic neurons counted in a whole rostrocaudal series of thalamus sections (1:5 sections). This density ratio provides a direct estimate of how many thalamocortical cells extend at least some part of their axon into any given 1 mm2 of layer 1. The mean density ratio is 1025 ± 250 cells per mm2. Note, however, that cells were counted in only a fifth of the total thalamic tissue, so the actual number of thalamic cells innervating 1 mm2 of layer 1 must be on the order of 4000–5000. Regression analysis shows a remarkably similar density in every sensory, association, motor, or limbic cortical area examined (R2 = 0.93).

Distribution in Layer 1 of Tangential M-Type Axon Arborizations

Retrograde tracer deposits in layer 1 seemed to indicate that M-type neurons in some nuclei distribute their axons over wide areas of the cortex, whereas those in other nuclei innervate more limited cortical areas. To directly observe and compare the actual distribution and morphology of these axons within layer 1, we made small BDA injections in some of these thalamic nuclei and examined the labeling in the cortex (Table SM1).

In serial coronal sections (Supplementary Fig. SM4) of brains that received small injections in some of the nuclei that innervate wide regions of layer 1, such as the ventromedial or the medial lateral posterior nucleus, thalamic axons were seen entering the cerebral hemisphere and extending through the subcortical white matter. At multiple points, isolated processes extended radially across the cortical layers, and in occasional favorable sections, they could be followed all the way up to layer 1. Many of these processes arborized also in layers 5 or 3. Within layer 1, they remained confined to a thin stratum (∼70 μm deep) in the uppermost half of layer 1 (sublayer 1a, Vogt 1991) as essentially flat and tangentially oriented axonal arbors. For this reason, coronal sections are clearly inadequate for studying global morphology of these axon arbors. Thus, we examined the axon labeling in layer 1 either directly on whole-mounted unsectioned and flattened hemispheres or in sections parallel to the pial surface taken from flattened hemispheres. From these images, we generated panoramic views of the labeling on the cerebral hemisphere surface (Fig. 7).

Figure 7.

Tangential spread in the cortex varies greatly between M-type axons originated in different thalamic nuclei. (A) Lateral view of the rat's left cerebral hemisphere. For reference, the position of middle cerebral artery (MCA) and its main branches are indicated by dashed lines. Compare with panels (C) and (E). (B) Coronal section showing the center of a small iontophoretic deposit of BDA (arrow) in the ventromedial (VM) thalamic nucleus. CyO counterstain. This deposit backfilled ∼150 ventromedial neuron somata. (C) Reconstruction of thalamocortical axon arbors labeled in layer 1 after the injection in panel (B) as observed on a flattened whole mount of the cerebral hemisphere. MCA is represented by thick dashed gray lines. Thalamocortical axons in layer 1 are thickened ∼100 times to make them visible at this scale. Note that a large swath of the cerebral surface is reached by the axons arising in the small thalamic zone injected with BDA. For comparison, the “barrel” layer 4 domains in the posteromedial barrel subfield (PMBSF) as revealed by CyO histochemistry are also shown. It is to be remembered that C-type axons arising in the ventroposterior medial nucleus (VPM, panel A) have been shown to arborize essentially within the area of 1 or 2 barrel domains (Jensen and Killackey 1987; Arnold et al. 2001). (D) Small iontophoretic deposit of BDA (arrow) in the most medial part of the dorsal lateral geniculate nucleus. Thionin counterstain. (E) Thalamocortical axon arbors labeled in layer 1 in the experiment shown in panel (D) reconstructed from serial tangential sections of a flattened hemisphere. (FG) Following a large iontophoretic BDA deposit about the same location as the previous case (panel D), but involving in this case many thousands of thalamic cells, label layer 1 axons only within the primary visual area. Reconstruction from serial tangential sections. Bar: (A, B, D, F) = 2 mm; (C, E, G) = 200 μm.

Figure 7.

Tangential spread in the cortex varies greatly between M-type axons originated in different thalamic nuclei. (A) Lateral view of the rat's left cerebral hemisphere. For reference, the position of middle cerebral artery (MCA) and its main branches are indicated by dashed lines. Compare with panels (C) and (E). (B) Coronal section showing the center of a small iontophoretic deposit of BDA (arrow) in the ventromedial (VM) thalamic nucleus. CyO counterstain. This deposit backfilled ∼150 ventromedial neuron somata. (C) Reconstruction of thalamocortical axon arbors labeled in layer 1 after the injection in panel (B) as observed on a flattened whole mount of the cerebral hemisphere. MCA is represented by thick dashed gray lines. Thalamocortical axons in layer 1 are thickened ∼100 times to make them visible at this scale. Note that a large swath of the cerebral surface is reached by the axons arising in the small thalamic zone injected with BDA. For comparison, the “barrel” layer 4 domains in the posteromedial barrel subfield (PMBSF) as revealed by CyO histochemistry are also shown. It is to be remembered that C-type axons arising in the ventroposterior medial nucleus (VPM, panel A) have been shown to arborize essentially within the area of 1 or 2 barrel domains (Jensen and Killackey 1987; Arnold et al. 2001). (D) Small iontophoretic deposit of BDA (arrow) in the most medial part of the dorsal lateral geniculate nucleus. Thionin counterstain. (E) Thalamocortical axon arbors labeled in layer 1 in the experiment shown in panel (D) reconstructed from serial tangential sections of a flattened hemisphere. (FG) Following a large iontophoretic BDA deposit about the same location as the previous case (panel D), but involving in this case many thousands of thalamic cells, label layer 1 axons only within the primary visual area. Reconstruction from serial tangential sections. Bar: (A, B, D, F) = 2 mm; (C, E, G) = 200 μm.

In these panoramic views, the extent and number of the axon arbors labeled by some small thalamic injections is evident. For example, following a small (<150 μm diameter) BDA injection in the ventromedial nucleus, one of the nuclei noted in the retrograde experiments to have a high tangential dispersal of its axons (see also Herkenham 1979; Monconduit and Villanueva 2005) also exhibited numerous subpial axon arbors extending over most of the rostral half of the cerebral hemisphere (Fig. 7B–C). These subpial arbors were distributed very unevenly, accumulating in several “patches” and leaving adjacent zones almost unlabeled. Likewise, a slightly larger injection near the ventral and lateral border of the ventromedial nucleus produced even more extensive labeling (Supplementary Fig. SM5). In contrast, comparable BDA injections placed in other nuclei such as the dorsal lateral geniculate (one of the nuclei identified in the retrograde labeling analysis as giving rise to M-type projections with low tangential dispersal) consistently produced a single relatively small zone of layer 1 axon labeling within the primary visual cortex (Fig. 7D–G).

Fine Morphology of Thalamocortical Axon Arbors in Layer 1

Even with our smallest BDA injections in the thalamus, the overlap and high number of labeled layer 1 axon arbors made individual axon reconstructions impossible. Nevertheless, at high magnification, some morphological features could be readily appreciated. Near the point of entry from the underlying layers into layer 1, many axons made a sharp turn, taking a tangential trajectory. Most arbors ramified frequently within sublayer 1a (Fig. 8C–F). In most sites, multiple arbors were overlapping, their branches forming a crisscrossing subpial mesh (Fig. 8B–D). Some few arbors consisted of long, sparingly branched axon segments that followed roughly straight trajectories without an apparent preferred direction. The longest of such segments rarely exceeded 2 mm. This apparent limitation in maximum length was the same in both sectioned and whole-mounted tissue.

Figure 8.

Thalamocortical axons form a dense subpial canopy of crisscrossing terminal branches with numerous varicosities. (A) Labeled thalamocortical axon branches in layer 1 as seen on coronal section of the frontal cortex (area M2) after a small BDA injection in the ventromedial (VM) nucleus. Pial surface is at the top. Cresyl violet counterstain. The few radially oriented, unbranched, and mostly bouton-free axons ascending through layers 2 and 1b contrast sharply with the mesh of tangential, varicose branches present in sublayer 1a. The same pattern was observed in other areas in this case, as well as BDA injection experiments involving other nuclei (Supplementary Table SM1). (B) Low-magnification view of the pial surface of a flattened, non-defattened, unsectioned cerebral hemisphere in aqueous mounting. Numerous layer 1 BDA-labeled thalamocortical axon arbors are visible through the translucent pia mater even at this low magnification. (C) BDA-labeled thalamocortical axon arbors in layer 1 as seen at in a 60-μm-thick tangential section. Parietal cortex after an injection in the lateral posterior medial nucleus. Numerous branching points are visible at this magnification (arrowheads). (D) One of the numerous axon branching points seen in panel (C) (black arrowhead in both panels) is shown at higher magnification. Note that numerous additional branching points in thinner axons (white arrowheads) become visible at this magnification (GF) High-magnification detail of bouton-like swellings in M-type axons in layer 1 as seen in tangential sections. Panel (G) axons were labeled by a BDA injection in the lateral posteromedial nucleus (LPMR), whereas panel (F) axons were labeled from the dorsal lateral geniculate nucleus (DLG). (H) Measurements of interbouton distance axons labeled in layer 1 from the VM nucleus, LPMR or DLG; 400 measurements were taken for each. Note that whereas the mean values are similar, the range of interbouton distances is broader in VM and LPMR axons than in DLG axons. Bars (A) = 50 μm; (B) = 200 μm; (C) = 25 μm; (D) = 10 μm; (EF) = 5 μm.

Figure 8.

Thalamocortical axons form a dense subpial canopy of crisscrossing terminal branches with numerous varicosities. (A) Labeled thalamocortical axon branches in layer 1 as seen on coronal section of the frontal cortex (area M2) after a small BDA injection in the ventromedial (VM) nucleus. Pial surface is at the top. Cresyl violet counterstain. The few radially oriented, unbranched, and mostly bouton-free axons ascending through layers 2 and 1b contrast sharply with the mesh of tangential, varicose branches present in sublayer 1a. The same pattern was observed in other areas in this case, as well as BDA injection experiments involving other nuclei (Supplementary Table SM1). (B) Low-magnification view of the pial surface of a flattened, non-defattened, unsectioned cerebral hemisphere in aqueous mounting. Numerous layer 1 BDA-labeled thalamocortical axon arbors are visible through the translucent pia mater even at this low magnification. (C) BDA-labeled thalamocortical axon arbors in layer 1 as seen at in a 60-μm-thick tangential section. Parietal cortex after an injection in the lateral posterior medial nucleus. Numerous branching points are visible at this magnification (arrowheads). (D) One of the numerous axon branching points seen in panel (C) (black arrowhead in both panels) is shown at higher magnification. Note that numerous additional branching points in thinner axons (white arrowheads) become visible at this magnification (GF) High-magnification detail of bouton-like swellings in M-type axons in layer 1 as seen in tangential sections. Panel (G) axons were labeled by a BDA injection in the lateral posteromedial nucleus (LPMR), whereas panel (F) axons were labeled from the dorsal lateral geniculate nucleus (DLG). (H) Measurements of interbouton distance axons labeled in layer 1 from the VM nucleus, LPMR or DLG; 400 measurements were taken for each. Note that whereas the mean values are similar, the range of interbouton distances is broader in VM and LPMR axons than in DLG axons. Bars (A) = 50 μm; (B) = 200 μm; (C) = 25 μm; (D) = 10 μm; (EF) = 5 μm.

Axon arbors in layer 1 showed numerous irregularly spaced swellings, suggestive of en passant boutons (Fig. 8E–F). Between swellings, axon caliber was variable, but always very small (<0.5 μm). Interbouton distances were measured in large samples of axon segments labeled from the ventromedial nucleus, lateral posterior, and the dorsal lateral geniculate nucleus. No correction was applied for shrinkage in these measurements. Mean interspace bouton distances and their ranges of variation were similar between axons originating in different nuclei (5.23 μm for ventromedial; 5.23 for lateral posterior medial rostral part; and 4.3 for DLG [dorsal lateral geniculate nucleus] axons; Fig. 8H). Comparison with published measurements in typical C-type axons (3.59 μm in ventroposterior nucleus axons innervating layer 4 “barrels” in the rat somatosensory cortex [Lu and Lin 1993]) suggests a comparable abundance of boutons per axon length unit.

Discussion

We mapped the thalamic origin and cortical distribution of thalamocortical pathways innervating cortical layer 1 (M-type neurons). The main findings are that 1) M-type neurons are numerous in many thalamic nuclei; 2) axons from large numbers of M-type thalamocortical neurons converge at any given spot of layer 1; 3) within layer 1, M-type axon arbors extend as a patchy tangential mesh of crisscrossing branches with frequent bouton-like swellings; and 4) M-type pathways are not always “diffuse”: the axons from some nuclei diverge widely, covering large swaths of the cerebral hemisphere, whereas those from other nuclei arborize within only a single cortical area.

The M-Type Cell Population Is Large and Widespread

Because the intensively studied sensory thalamic nuclei (ventroposterior, lateral geniculate, and medial geniculate) are made predominantly of C-type cells, the anatomy, functional properties, and synaptic relationships with the cortex of these cells have been worked out in considerable detail (reviews in Castro-Alamancos and Connors 1997; Sherman and Guillery 2005; Jones 2007). In contrast, M-type thalamocortical cells have received much less attention (Lorente de No 1938; Killackey and Ebner 1973; Frost and Caviness 1980; Herkenham 1986). As a result, data on M-type thalamocortical pathways are scant and fragmentary, and considerable uncertainty remains about their anatomical weight, thalamic origin, and function (Llinás et al. 2002; Kubota et al. 2007; Graziano et al. 2008).

Some years ago, it was hypothesized that these cells might actually be a broad neuron class that was spread widely across the thalamus and one that was fundamentally different from the thalamocortical neurons targeting the middle layers (Avendaño et al. 1990; Jones 1998), as each of the 2 neuron types was observed to receive different inputs and/or have different soma sizes and/or express different calcium-binding proteins (Jones 2001). Moreover, their cortical targets seemed complementary, as thalamic axons arborizing in cortical layer 1 do not arborize in layer 4 (despite often doing so in layers 5a and/or 3b), whereas axons arborizing in layer 4 do not arborize in layer 1 (Boyd and Matsubara 1996; Catalano et al. 1996; Deschênes et al. 1998; Gheorghita et al. 2006; Casagrande et al. 2007). Our finding that M-type neurons are present in many thalamic nuclei seems consistent with the matrix hypothesis, as well as with observations following anterograde tracer injections in individual rat nuclei (Krettek and Price 1977; Herkenham 1986; Berendse and Groenewegen 1991; Van Groen and Wyss 1995; Linke and Schwegler 2000; Kimura et al. 2003; Gauriau and Bernard 2004; Vertes et al. 2006) or superficial retrograde tracer deposits in a single area (Parnavelas et al. 1981; Rieck and Carey 1985; Mitchell and Cauller 2001).

We demonstrate here that M-type neurons are a vast cell population spread widely and unevenly throughout the thalamus. These cells are prevalent in the ventromedial, ventral anterior, and interanteromedial nuclei, as well as in large portions of the ventrolateral, posterior, lateral posterior, dorsal lateral geniculate, parafascicular, and medial geniculate complex (Fig. 9A). The submedius, anterodorsal, anteroventral, and portions of the laterodorsal, dorsomedial, and reuniens nuclei were scarcely labeled; however, they innervate layer 1 profusely in medial and lateral limbic areas not reached by our deposits (Krettek and Price 1977; Herkenham 1986; Shibata 1993; Van Groen and Wyss 1995; Vertes et al. 2006), and so we assume that these nuclei would also contain numerous M-type cells. In contrast, unlabeled portions of the lateral and medial geniculate, ventral lateral nuclei, and the whole ventroposterior nucleus probably lack M-type neurons: we extensively probed their target areas with retrograde tracers and found nothing, whereas anterograde studies by others suggest a C-type neuron predominance (Herkenham 1986; Arbuthnott et al. 1990; Kageyama and Robertson 1993; Kimura et al. 2003). Likewise, we interpret the scant labeling in nuclei like the anterior and posterior intralaminar, paraventricular, intermediodorsal, and peripeduncular nuclei as an indication that they harbor few M-type cells because their main targets seem to be the deep cortical layers, the striatum, and/or amygdala (Herkenham 1986; Berendse and Groenewegen 1991; Deschênes et al. 1996; Linke and Schwegler 2000; Vertes and Hoover 2008).

Figure 9.

Summary of thalamic location and axon branching diversity of rat M-type neurons. Diagrams based on our data as well as from previous studies with anterograde tracer injections in individual nuclei (see text for references). (A) M-type neuron location in the various nuclei is represented by dots on 6 coronal thalamic sections. Orange dots indicate the M-type neurons of “localized axon” subtype (those targeting a single or few adjacent areas), and pink dots represent M-type neurons of the “widespread axon” subtype (targeting multiple, distant cortical fields). Blank regions lack neurons innervating cortical layer 1; these probably contain C-type neurons or neurons projecting diffusely to the infragranular layers (Herkenham 1986). See Table 1 for nuclei abbreviations. (B) Diagram of the wiring pattern of localized axon (orange) or widespread axon (pink) thalamocortical M-type neuron in rodents. These 2 axon forms are intended to represent the extremes of a continuum of variance among M-type neurons. Note that the overlap in sublayer 1a of a great many M-type axon arbors creates a dense axonal canopy. Moreover, as these arbors give off some branches in this layer and have a high spatial frequency of boutons (Fig. 8), the resulting local density of thalamocortical boutons in sublayer 1a may be very high, consistent with the VGluT2 immunolabeling (compare with Supplementary Fig. SM6). M-type axons frequently leave collateral braches to layers 5a or 3 that may contribute to the observed VGluT2 immunolabeling in the middle cortical layers (Fujiyama et al. 2001; see also Supplementary Fig. SM6). In widespread axons, branching in the subcortical white matter allows the simultaneous targeting of multiple and distant cortical zones (Deschênes et al. 1998). Subcortical axon branching is more limited in localized M-type axons. For comparison, a generalized C-type neuron (black line) is included in the diagram, based on published single-axon reconstruction studies in rodents (Jensen and Killackey 1987; Deschênes et al. 1998; Arnold et al. 2001; Gheorghita et al. 2006). Note that, in contrast to M-type axons, C-type axons give rise to a high local bouton density of in layers 4–3 as a result of the profuse branching of individual axons in largely nonoverlapping domains.

Figure 9.

Summary of thalamic location and axon branching diversity of rat M-type neurons. Diagrams based on our data as well as from previous studies with anterograde tracer injections in individual nuclei (see text for references). (A) M-type neuron location in the various nuclei is represented by dots on 6 coronal thalamic sections. Orange dots indicate the M-type neurons of “localized axon” subtype (those targeting a single or few adjacent areas), and pink dots represent M-type neurons of the “widespread axon” subtype (targeting multiple, distant cortical fields). Blank regions lack neurons innervating cortical layer 1; these probably contain C-type neurons or neurons projecting diffusely to the infragranular layers (Herkenham 1986). See Table 1 for nuclei abbreviations. (B) Diagram of the wiring pattern of localized axon (orange) or widespread axon (pink) thalamocortical M-type neuron in rodents. These 2 axon forms are intended to represent the extremes of a continuum of variance among M-type neurons. Note that the overlap in sublayer 1a of a great many M-type axon arbors creates a dense axonal canopy. Moreover, as these arbors give off some branches in this layer and have a high spatial frequency of boutons (Fig. 8), the resulting local density of thalamocortical boutons in sublayer 1a may be very high, consistent with the VGluT2 immunolabeling (compare with Supplementary Fig. SM6). M-type axons frequently leave collateral braches to layers 5a or 3 that may contribute to the observed VGluT2 immunolabeling in the middle cortical layers (Fujiyama et al. 2001; see also Supplementary Fig. SM6). In widespread axons, branching in the subcortical white matter allows the simultaneous targeting of multiple and distant cortical zones (Deschênes et al. 1998). Subcortical axon branching is more limited in localized M-type axons. For comparison, a generalized C-type neuron (black line) is included in the diagram, based on published single-axon reconstruction studies in rodents (Jensen and Killackey 1987; Deschênes et al. 1998; Arnold et al. 2001; Gheorghita et al. 2006). Note that, in contrast to M-type axons, C-type axons give rise to a high local bouton density of in layers 4–3 as a result of the profuse branching of individual axons in largely nonoverlapping domains.

Nuclei such as the ventromedial or posterior were labeled from widely separate cortical areas. The double-labeling experiments (see also Monconduit and Villanueva 2005) show that at least some M-type neurons in these nuclei branch their axons to reach 2 areas. Moreover, the patchy appearance resulting from the multiple small subpial arbors in a wide region of cortical labeling (Fig. 7) is consistent with M-type cells in these nuclei having a “widespread” axon that gives off multiple branches in the subcortical white matter (Fig. 9C; Deschênes et al. 1998; Galazo et al. 2008). In contrast, other thalamic nuclei were labeled from only a single cortical area, indicating that their M-type neurons have a far more “localized” axon distribution. Single-axon studies have in fact shown some M-type axons arborizing within a single area (Deschênes et al. 1998; Casagrande et al. 2007). Overall, the M-type cell class thus seems to show substantial variation regarding 1) axon branching to one or multiple areas (widespread or localized axons); 2) arborization into cortical layers in addition to layer 1 (Deschênes et al. 1998); and 3) calcium-binding protein expression (Jones 2007; Rubio-Garrido et al. 2007).

High Spatial Convergence in Layer 1

Repeated observations that 1) anterograde tracer deposits in the thalamus usually yield the heaviest labeling in the middle layers and 2) fewer neurons are labeled by superficial than deep-layer retrograde tracer injections seem to have cemented the widely held notion that thalamic pathways directed to layer 1 are scant. However, we demonstrate here that virtually every part of cortical layer 1 receives a combined input from a very large number of thalamic neurons. Two factors in our methods may have increased our ability to detect layer 1–projecting neurons. First, previous studies made minute tracer injections or applied tracer-soaked paper on the cortical surface for only few minutes to prevent tracer spread to deep layers. Here, fluorochrome-saturated paper pieces were left on the pia mater for a whole week, probably allowing the labeling of any neuron having an axon segment within the impregnated layer 1 neuropil. Second, the use of high-energy epifluorescence microscopy, wide-diameter objectives, and digital imaging and analysis allowed the confident identification of even weakly labeled cells and eliminated photobleaching.

Results show that every square millimeter of neocortical layer 1 contains at least part of the axon arbor from a very large number (mean ∼4500) of different thalamocortical neurons. No such ratio is currently available for rat C-type neurons, but a crude estimate of ∼3800 axons per mm2 might be inferred for ventroposterior nucleus axons in area S1 (Riddle and Purves 1995; Mooney and Miller 1999; P. Rubio-Garrido, unpublished observations). In carnivores and primates, estimates of C-type neurons innervation of the cortex are even lower (Peters and Payne 1993; Jones 2007). These data seem to suggest that, even in primary sensory areas, the number of M-type axons targeting layer 1 is higher than C-type axons targeting layer 4.

Moreover, single-axon reconstruction studies of rodent C-type axons have shown that they arborize densely and focusedly only within a single or a 2 column–like domains (Jensen and Killackey 1987; Deschênes et al. 1998; Arnold et al. 2001; Gheorghita et al. 2006). Together with published single-axon data, present observations thus suggest that whereas C-type pathways achieve a high local density of terminals in layer 4 by profusely branching individual axons in small, largely nonoverlapping domains, M-type axons may attain similar density in layer 1 by a massive overlapping of tangentially widespread arbors that are less heavily branched (Fig. 9B). Notice that, as a result of such differences in wiring, axon tracing from any given single thalamic nucleus is bound to reveal only a fraction of the total M-type axon population actually present in layer 1 of the target cortical areas.

Our anterograde tracing observations concur with reports of bouton-like varicosities in thalamic axon arbors in layer 1 (Ferster and LeVay 1978; Arbuthnott et al. 1990; Lu and Lin 1993; Hashikawa et al. 1995; Deschênes et al. 1998). Remarkably, the observed bouton frequency in M-type axons within layer 1 is similar to C-type axon bouton frequency reported in layer 4 (Lu and Lin 1993). In mouse thalamocortical axons, these varicosities roughly correlate (1:1.4) with synaptic sites (White et al. 2004).

VGluT2 has been recently shown to be a marker of thalamocortical synapses (Fujiyama et al. 2001; Hur and Zaborszky 2005; Barroso-Chinea et al. 2007). Immunolabeling for VGluT2 thus reveals the local density of thalamocortical synaptic sites in the various cortical layers (Kubota et al. 2007). The heavy VGluT2 immunostaining present in sublayer 1a (Fujiyama et al. 2001; Nakamura et al. 2005; Kubota et al. 2007; see also Supplementary Fig. SM6) is thus consistent with very large number of subpial axonal arbors with numerous bouton-like swellings overlapping within sublayer 1a.

Possible Functional Roles

Although it has been known for many years that thalamocortical pathways targeting layer 1 can modulate cortical firing (see Herkenham 1986 review; Penny et al. 1982; Llinás et al. 2002; Monconduit and Villanueva 2005), functional studies remain scarce and fragmentary. As a result, hypotheses regarding M-type pathway function are still largely inferred from structural data.

Being located in so many thalamic nuclei, it seems clear that M-type neurons could participate in many functional subsystems. At the same time, however, the consistency of features like tangential spread and distal dendrite targeting have suggested some possible common functional roles. For example, tangential spread has been related to the spread of coherent activity across large ensembles of cortical cells (Penny et al. 1982; Jones 2001). Likewise, distal dendrite targeting has been interpreted as evidence that the role of M-type pathways would be “modulatory,” requiring a very high spatial/temporal summation to affect cortical cell firing significantly (Herkenham 1986; Avendaño et al. 1990). In recent years, however, segregation of different inputs between distal and basal dendritic compartments has been shown to greatly expand pyramidal cell computational power (Spratling 2002). Besides, the discovery of complex active properties in the distal tufts of apical dendrites have led to the realization that distal inputs may actually strongly impact both pyramidal cell firing and synaptic plasticity (reviews in Williams et al. 2007; Spruston 2008). For example, in response to synchronous summation of multiple inputs, apical tufts generate calcium spikes that, under narrow time/space constrains, may interact nonlinearly with backpropagating action potentials and effectively drive somatic firing (Larkum et al. 2004). Nonlinear interactions of this kind have been noted between M-type and C-type thalamocortical inputs (Llinás et al. 2002). What is more, even at subthreshold levels of depolarization, asynchronous distal inputs have been shown to provide instructive signals effective for the induction of plasticity at proximal synaptic sites (Dudman et al. 2007).

Electron microscopic studies report that virtually all thalamocortical terminals in layer 1 form asymmetric synapses onto spines, presumably belonging to distal apical dendritic tufts of pyramidal neurons (Arbuthnott et al. 1990; Kubota et al. 2007). Because only some particular rat pyramidal neuron subpopulations extend apical tufts within layer 1 (Hallman et al. 1988; Veinante and Deschênes 2003; Gao and Zheng 2004; Molnár and Cheung 2006; Larsen et al. 2008), M-type input seems bound to primarily reach these pyramidal cells (Fig. 9C).

Selective targeting of layer 1 apical dendritic tufts and convergent organization such as those observed in M-type pathways have been also noted in the other major known excitatory input to layer 1, the “feedback” corticocortical pathways (Felleman and Van Essen 1991; Cauller et al. 1998). In addition to layer 1, these “feedback” corticocortical pathways target layers 5 and 3 and avoid layers 4 and 6, very much like M-type axons do (Fig. 9B). Moreover, it is intriguing that cortical “feedback” pathways originate in higher association, limbic, motor, and nonlemniscal sensory cortical areas, and as shown here, M-type pathways originate mainly in association, limbic, motor, and nonlemniscal sensory thalamic nuclei, which in turn receive their main inputs from the above areas either directly or via subcortical loops through the basal ganglia or cerebellum (Morán and Reinoso-Suárez 1988; Sherman and Guillery 2005). These similarities in global anatomical organization are consistent with the notion that M-type axons may support extensive “feedback” interactions between cortical areas via the thalamus (Crick and Koch 1998).

Funding

Spain's Ministerio de Educación y Ciencia (BFU 2002-04674, BFU 2005-07857) to FC; predoctoral fellowships from Carabobo University, Venezuela (CU448 to FPM); Autónoma de Madrid University (FPU200613 to CP); Spain's Ministerio de Educación y Ciencia (FPU-AP2002-0532 to MJG).

Supplementary Material

Supplementary material can be found at http://www.cercor.oxfordjournals.org/.

Dr Carlos Avendaño generously shared unpublished data on surface shrinkage in freeze-sectioned rat brain tissue and reviewed an earlier version of the manuscript. Dr Fernando Reinoso-Suárez provided insightful comments and encouragement throughout the project. Carol F. Warren provided linguistic advice. Criticism and suggestions for improvement by 2 anonymous reviewers are also gratefully acknowledged. Conflict of Interest: None declared.

References

Arbuthnott
GW
MacLeod
NK
Maxwell
DJ
Wright
AK
Distribution and synaptic contacts of the cortical terminals arising from neurons in the rat ventromedial thalamic nucleus
Neuroscience
 , 
1990
, vol. 
38
 (pg. 
47
-
60
)
Arnold
PB
Li
CX
Waters
RS
Thalamocortical arbors extend beyond single cortical barrels: an in vivo intracellular tracing study in rat
Exp Brain Res
 , 
2001
, vol. 
136
 (pg. 
152
-
168
)
Avendaño
C
Stepniewska
I
Rausell
E
Reinoso-Suárez
F
Segregation and heterogeneity of thalamic cell populations projecting to superficial layers of posterior parietal cortex: a retrograde tracer study in cat and monkey
Neuroscience
 , 
1990
, vol. 
39
 (pg. 
547
-
559
)
Barroso-Chinea
P
Castle
M
Aymerich
MS
Pérez-Manso
M
Erro
E
Tuñon
T
Lanciego
JL
Expression of the mRNAs encoding for the vesicular glutamate transporters 1 and 2 in the rat thalamus
J Comp Neurol
 , 
2007
, vol. 
501
 (pg. 
703
-
715
)
Berendse
HW
Groenewegen
HJ
Restricted cortical termination fields of the midline and intralaminar thalamic nuclei in the rat
Neuroscience
 , 
1991
, vol. 
42
 (pg. 
73
-
102
)
Boyd
JD
Matsubara
JA
Laminar and columnar patterns of geniculocortical projections in the cat: relationship to cytochrome oxidase
J Comp Neurol
 , 
1996
, vol. 
365
 (pg. 
659
-
682
)
Casagrande
VA
Yazar
F
Jones
KD
Ding
Y
The morphology of the koniocellular axon pathway in the macaque monkey
Cereb Cortex
 , 
2007
, vol. 
17
 (pg. 
2334
-
2345
)
Castro-Alamancos
MA
Connors
BW
Thalamocortical synapses
Prog Neurobiol
 , 
1997
, vol. 
51
 (pg. 
581
-
606
)
Catalano
SM
Robertson
RT
Killackey
HP
Individual axon morphology and thalamocortical topography in developing rat somatosensory cortex
J Comp Neurol
 , 
1996
, vol. 
367
 (pg. 
36
-
53
)
Cauller
L
Layer I of primary sensory neocortex: where top-down converges upon bottom-up
Behav Brain Res
 , 
1995
, vol. 
71
 (pg. 
163
-
170
)
Cauller
LJ
Clancy
B
Connors
BW
Backward cortical projections to primary somatosensory cortex in rats extend long horizontal axons in layer I
J Comp Neurol
 , 
1998
, vol. 
390
 (pg. 
297
-
310
)
Crick
F
Koch
C
Constraints on cortical and thalamic projections: the no-strong-loops hypothesis
Nature
 , 
1998
, vol. 
39
 (pg. 
245
-
250
)
Deschênes
M
Bourassa
J
Parent
A
Striatal and cortical projections of single neurons from the central lateral thalamic nucleus in the rat
Neuroscience
 , 
1996
, vol. 
72
 (pg. 
679
-
687
)
Deschênes
M
Veinante
P
Zhang
ZW
The organization of corticothalamic projections: reciprocity versus parity
Brain Res Brain Res Rev
 , 
1998
, vol. 
28
 (pg. 
286
-
308
)
Douglas
RJ
Martin
KA
Mapping the matrix: the ways of neocortex
Neuron
 , 
2007
, vol. 
56
 (pg. 
226
-
238
)
Dudman
JT
Tsay
D
Siegelbaum
SA
A role for synaptic inputs at distal dendrites: instructive signals for hippocampal long-term plasticity
Neuron
 , 
2007
, vol. 
56
 (pg. 
866
-
879
)
Felleman
DJ
Van Essen
DC
Distributed hierarchical processing in the primate cerebral cortex
Cereb Cortex
 , 
1991
, vol. 
1
 
Suppl 1
(pg. 
1
-
47
)
Ferster
D
LeVay
S
The axonal arborizations of lateral geniculate neurons in the striate cortex of the cat
J Comp Neurol
 , 
1978
, vol. 
182
 (pg. 
923
-
944
)
Frost
DO
Caviness
VS
Radial organization of thalamic projections to the neocortex in the mouse
J Comp Neurol
 , 
1980
, vol. 
194
 (pg. 
369
-
393
)
Fujiyama
F
Furuta
T
Kaneko
T
Immunocytochemical localization of candidates for vesicular glutamate transporters in the rat cerebral cortex
J Comp Neurol
 , 
2001
, vol. 
435
 (pg. 
379
-
387
)
Galazo
MJ
Martinez-Cerdeño
V
Porrero
C
Clascá
F
Embryonic and postnatal development of the layer I-directed (“matrix”) thalamocortical system in the rat
Cereb Cortex
 , 
2008
, vol. 
18
 (pg. 
344
-
363
)
Gao
WJ
Zheng
ZH
Target-specific differences in somatodendritic morphology of layer V pyramidal neurons in rat motor cortex
J Comp Neurol
 , 
2004
, vol. 
476
 (pg. 
174
-
185
)
Gauriau
C
Bernard
JF
Posterior triangular thalamic neurons convey nociceptive messages to the secondary somatosensory and insular cortices in the rat
J Neurosci
 , 
2004
, vol. 
24
 (pg. 
752
-
761
)
Geneser-Jensen
FA
Blackstadt
TW
Distribution of acetylcholinesterase in the hippocampal region of the guinea pig. I. Entorhinal area, parasubiculum and presubiculum
Z Zellforsch Mikrosk Anat
 , 
1971
, vol. 
106
 (pg. 
192
-
202
)
Gheorghita
F
Kraftsik
R
Dubois
R
Welker
E
Structural basis for map formation in the thalamocortical pathway of the barrelless mouse
J Neurosci
 , 
2006
, vol. 
26
 (pg. 
10057
-
10067
)
Gilbert
CD
Sigman
M
Brain states: top-down influences in sensory processing
Neuron
 , 
2007
, vol. 
54
 (pg. 
677
-
696
)
Graziano
A
Liu
XB
Murray
KD
Jones
EG
Vesicular glutamate transporters define two sets of glutamatergic afferents to the somatosensory thalamus and two thalamocortical projections in the mouse
J Comp Neurol
 , 
2008
, vol. 
507
 (pg. 
1258
-
1276
)
Hallman
LE
Schofield
BR
Lin
CS
Dendritic morphology and axon collaterals of corticotectal, corticopontine, and callosal neurons in layer V of primary visual cortex of the hooded rat
J Comp Neurol
 , 
1988
, vol. 
272
 (pg. 
149
-
160
)
Hashikawa
T
Molinari
M
Rausell
E
Jones
EG
Patchy and laminar terminations of medial geniculate axons in monkey auditory cortex
J Comp Neurol
 , 
1995
, vol. 
362
 (pg. 
195
-
208
)
Herkenham
M
The afferent and efferent connections of the ventromedial thalamic nucleus in the rat
J Comp Neurol
 , 
1979
, vol. 
183
 (pg. 
487
-
517
)
Herkenham
M
Jones
EG
Peters
A
New perspectives on the organization and evolution of nonspecific thalamocortical projections
Cerebral cortex. Sensory-motor areas and aspects of cortical connectivity
 , 
1986
, vol. 
Vol. 5
 
New York
Plenum
(pg. 
403
-
445
)
Hur
EE
Zaborszky
L
Vglut2 afferents to the medial prefrontal and primary somatosensory cortices: a combined retrograde tracing in situ hybridization study
J Comp Neurol
 , 
2005
, vol. 
483
 (pg. 
351
-
373
)
Jensen
KF
Killackey
HP
Terminal arbors of axons projecting to the somatosensory cortex of the adult rat. I. The normal morphology of specific thalamocortical afferents
J Neurosci
 , 
1987
, vol. 
7
 (pg. 
3529
-
3543
)
Jones
EG
Viewpoint: the core and matrix of thalamic organization
Neuroscience
 , 
1998
, vol. 
85
 (pg. 
331
-
345
)
Jones
EG
The thalamic matrix and thalamocortical synchrony
Trends Neurosci
 , 
2001
, vol. 
24
 (pg. 
595
-
601
)
Jones
EG
The thalamus
 , 
2007
2nd ed
New York
Cambridge University Press
Kageyama
GH
Robertson
RT
Development of geniculocortical projections to visual cortex in rat: evidence early ingrowth and synaptogenesis
J Comp Neurol
 , 
1993
, vol. 
335
 (pg. 
123
-
148
)
Killackey
H
Ebner
F
Convergent projection of three separate thalamic nuclei on to a single cortical area
Science
 , 
1973
, vol. 
179
 (pg. 
283
-
285
)
Kimura
A
Donishi
T
Sakoda
T
Hazama
M
Tamai
Y
Auditory thalamic nuclei projections to the temporal cortex in the rat
Neuroscience
 , 
2003
, vol. 
117
 (pg. 
1003
-
1016
)
Krettek
JE
Price
JL
The cortical projections of the mediodorsal nucleus and adjacent thalamic nuclei in the rat
J Comp Neurol
 , 
1977
, vol. 
171
 (pg. 
157
-
191
)
Kubota
Y
Hatada
S
Kondo
S
Karube
F
Kawaguchi
Y
Neocortical inhibitory terminals innervate dendritic spines targeted by thalamocortical afferents
J Neurosci
 , 
2007
, vol. 
27
 (pg. 
1139
-
1150
)
Larkum
ME
Senn
W
Lüscher
HR
Top-down dendritic input increases the gain of layer 5 pyramidal neurons
Cereb Cortex
 , 
2004
, vol. 
14
 (pg. 
1059
-
1070
)
Larsen
DD
Wickersham
IR
Callaway
EM
Retrograde tracing with recombinant rabies virus reveals correlations between projection targets and dendritic architecture in layer 5 of mouse barrel cortex
Front Neural Circuits
 , 
2008
, vol. 
1
 pg. 
5
  
doi: 10.3389/neuro.04.005.2007
Linke
R
Schwegler
H
Convergent and complementary projections of the caudal paralaminar thalamic nuclei to rat temporal and insular cortex
Cereb Cortex
 , 
2000
, vol. 
10
 (pg. 
753
-
771
)
Llinás
RR
Leznik
E
Urbano
FJ
Temporal binding via cortical coincidence detection of specific and nonspecific thalamocortical inputs: a voltage-dependent dye-imaging study in mouse brain slices
Proc Natl Acad Sci USA
 , 
2002
, vol. 
99
 (pg. 
449
-
454
)
Lorente de No
R
Fulton
J
Cerebral cortex: architecture, intracortical connections, motor projections
Physiology of the nervous system
 , 
1938
London
Oxford University Press
(pg. 
291
-
340
)
Lu
SM
Lin
RC
Thalamic afferents of the rat barrel cortex: a light- and electron-microscopic study using Phaseolus vulgaris leucoagglutinin as an anterograde tracer
Somatosens Mot Res
 , 
1993
, vol. 
10
 (pg. 
1
-
16
)
Mesulam
M-M
Mesulam
M-M
Principles of horseradish peroxidase neurohistochemistry and their applications for tracing neural pathways. Axonal transport, enzyme histochemistry and light microscope analysis
Tracing neural pathways with horseradish peroxidase
 , 
1982
Chichester (UK)
John Wiley & Sons
(pg. 
1
-
151
)
Mitchell
BD
Cauller
LJ
Corticocortical and thalamocortical projections to layer I of the frontal neocortex in rats
Brain Res
 , 
2001
, vol. 
921
 (pg. 
68
-
77
)
Molnár
Z
Cheung
AF
Towards the classification of subpopulations of layer V pyramidal projection neurons
Neurosci Res
 , 
2006
, vol. 
55
 
Suppl 2
(pg. 
105
-
115
)
Monconduit
L
Villanueva
L
The lateral ventromedial thalamic nucleus spreads nociceptive signals from the whole body surface to layer I of the frontal cortex
Eur J Neurosci
 , 
2005
, vol. 
21
 (pg. 
3395
-
3402
)
Mooney
SM
Miller
MW
Effects of prenatal exposure to ethanol on systems matching: the number of neurons in the ventrobasal thalamic nucleus of the mature rat
Brain Res Dev Brain Res
 , 
1999
, vol. 
117
 (pg. 
121
-
125
)
Morán
AM
Reinoso-Suárez
F
Topographical organization of the thalamic afferent connections to the motor cortex in the cat
J Comp Neurol
 , 
1988
, vol. 
270
 (pg. 
64
-
85
)
Nakamura
K
Hioki
H
Fujiyama
F
Kaneko
T
Postnatal changes of vesicular glutamate transporter (VGluT)1 and VGluT2 immunoreactivities and their colocalization in the mouse forebrain
J Comp Neurol
 , 
2005
, vol. 
492
 (pg. 
263
-
288
)
Parnavelas
JG
Chatzissavidou
A
Burne
RA
Subcortical projections to layer I of the visual cortex, area 17, of the rat
Exp Brain Res
 , 
1981
, vol. 
41
 (pg. 
184
-
187
)
Paxinos
G
Watson
C
The rat brain in stereotaxic coordinates
 , 
2005
5th ed
San Diego (CA)
Academic Press
Penny
GR
Itoh
K
Diamond
IT
Cells of different sizes in the ventral nuclei project to different layers of the somatic cortex in the cat
Brain Res
 , 
1982
, vol. 
242
 (pg. 
55
-
65
)
Peters
A
Payne
BR
Numerical relationships between geniculocortical afferents and pyramidal cell modules in cat primary visual cortex
Cereb Cortex
 , 
1993
, vol. 
3
 (pg. 
69
-
78
)
Portera-Cailliau
C
Weimer
RM
De Paola
V
Caroni
P
Svoboda
K
Diverse modes of axon elaboration in the developing neocortex
PLoS Biol
 , 
2005
, vol. 
3
 pg. 
e272
 
Riddle
DR
Purves
D
Individual variation and lateral asymmetry of the rat primary somatosensory cortex
J Neurosci
 , 
1995
, vol. 
15
 (pg. 
4184
-
4195
)
Rieck
RW
Carey
RG
Organization of the rostral thalamus in the rat: evidence for connections to layer I of visual cortex
J Comp Neurol
 , 
1985
, vol. 
234
 (pg. 
137
-
154
)
Rubio-Garrido
P
Pérez-de-Manzo
F
Clascá
F
Calcium-binding proteins as markers of layer-I projecting vs. deep layer-projecting thalamocortical neurons: a double-labeling analysis in the rat
Neuroscience
 , 
2007
, vol. 
149
 (pg. 
242
-
250
)
Schofield
BR
Schofield
RM
Sorensen
KA
Motts
SD
On the use of retrograde tracers for identification of axon collaterals with multiple fluorescent retrograde tracers
Neuroscience
 , 
2007
, vol. 
146
 (pg. 
773
-
783
)
Sherman
SM
Guillery
RW
Exploring the thalamus and its role in cortical function
 , 
2005
Cambridge (MA)
MIT Press
Shibata
H
Efferent projections from the anterior thalamic nuclei to the cingulate cortex in the rat
J Comp Neurol
 , 
1993
, vol. 
330
 (pg. 
533
-
542
)
Sjöström
PJ
Häusser
M
A cooperative switch determines the sign of synaptic plasticity in distal dendrites of neocortical pyramidal neurons
Neuron
 , 
2006
, vol. 
51
 (pg. 
227
-
238
)
Spratling
MW
Cortical region interactions and the functional role of apical dendrites
Behav Cogn Neurosci Rev
 , 
2002
, vol. 
1
 (pg. 
219
-
228
)
Spruston
N
Pyramidal neurons: dendritic structure and synaptic integration
Nat Rev Neurosci
 , 
2008
, vol. 
9
 (pg. 
206
-
221
)
Van Groen
T
Wyss
JM
Projections from the anterodorsal and anteroventral nucleus of the thalamus to the limbic cortex in the rat
J Comp Neurol
 , 
1995
, vol. 
358
 (pg. 
584
-
604
)
Veinante
P
Deschênes
M
Single-cell study of motor cortex projections to the barrel field in rats
J Comp Neurol
 , 
2003
, vol. 
464
 (pg. 
98
-
103
)
Vertes
RP
Hoover
WB
Projections of the paraventricular and paratenial nuclei of the dorsal midline thalamus in the rat
J Comp Neurol
 , 
2008
, vol. 
508
 (pg. 
212
-
237
)
Vertes
RP
Hoover
WB
Do Valle
AC
Sherman
A
Rodriguez
JJ
Efferent projections of reuniens and rhomboid nuclei of the thalamus in the rat
J Comp Neurol
 , 
2006
, vol. 
499
 (pg. 
768
-
796
)
Vogt
BA
Peters
A
Jones
EG
The role of layer 1 in cortical function
Cerebral cortex. Normal and altered states of function
 , 
1991
, vol. 
Vol. 9
 
New York
Plenum
(pg. 
49
-
80
)
White
EL
Weinfeld
E
Lev
DL
Quantitative analysis of synaptic distribution along thalamocortical axons in adult mouse barrels
J Comp Neurol
 , 
2004
, vol. 
479
 (pg. 
56
-
69
)
Williams
SR
Wozny
C
Mitchell
SJ
The back and forth of dendritic plasticity
Neuron
 , 
2007
, vol. 
56
 (pg. 
947
-
953
)
Wong-Riley
MTT
Changes in the visual system of monocularly sutured or enucleated cats demonstrable with cytochrome oxidase histochemistry
Brain Res
 , 
1979
, vol. 
17
 (pg. 
11
-
28
)