Abstract

Higher-order motor cortices, such as the secondary motor area (M2) in rodents, select future action patterns and transmit them to the primary motor cortex (M1). To better understand motor processing, we characterized “top-down” and “bottom-up” connectivities between M1 and M2 in the rat cortex. Somata of pyramidal cells (PCs) in M2 projecting to M1 were distributed in lower layer 2/3 (L2/3) and upper layer 5 (L5), whereas PCs projecting from M1 to M2 had somata distributed throughout L2/3 and L5. M2 afferents terminated preferentially in upper layer 1 of M1, which also receives indirect basal ganglia output through afferents from the ventral anterior and ventromedial thalamic nuclei. On the other hand, M1 afferents terminated preferentially in L2/3 of M2, a zone receiving indirect cerebellar output through thalamic afferents from the ventrolateral nucleus. While L5 corticopontine (CPn) cells with collaterals to the spinal cord did not participate in corticocortical projections, CPn cells with collaterals to the thalamus contributed preferentially to connections from M2 to M1. L5 callosal projection (commissural) cells participated in connectivity between M1 and M2 bidirectionally. We conclude that the connectivity between M1 and M2 is directionally specialized, involving specific PC subtypes that selectively target lamina receiving distinct thalamocortical inputs.

Introduction

Both primary motor cortex (M1) and higher-order motor areas that are situated more rostrally in the frontal cortex are profoundly involved in the initiation of voluntary movements (Tanji 1994; Dum and Strick 2002). In both primates and rodents, cortical neurons in higher-order motor areas begin to fire earlier than those in M1 during tasks that require the evaluation and selection of multiple possible actions (Nachev et al. 2008; Sul et al. 2011). M1 and some higher-order motor areas are reciprocally connected, and both primary and higher motor areas send projections to the spinal cord (Dum and Strick 1991; Lemon 2008). Therefore, action patterns are likely selected by higher-order areas and then transmitted to the spinal cord both directly and indirectly, through connections to M1. A fundamental question that remains to be addressed is whether corticospinal (CSp) projections that arise from M1 and higher-order motor areas have different functions in generating behavioral output (Miller 1987; Dum and Strick 2002).

Despite the importance of the functional interactions between M1 and higher-order motor areas, little is known about the anatomical basis of corticocortical (CC) connections between these 2 areas. Neighboring neocortical areas are typically connected in the following 2 ways: Direct CC (associational) connections through pyramidal cells (PCs) in various layers and corticothalamocortical loops that originate from layer 5 (L5) PCs (Jones 2007; Shipp 2007; Zikopoulos and Barbas 2007; Theyel et al. 2010).

In sensory pathways, the projections from lower to higher cortical areas are called feedforward, or “bottom-up,” connections, whereas those from higher to lower areas are termed feedback, or “top-down,” connections (Felleman and Van Essen 1991; Salin and Bullier 1995; Rockland 1997; Douglas and Martin 2004). Bottom-up projections originate from PCs that are located in layer 2/3 (L2/3) and terminate primarily in middle layers (layer 4 in primary sensory areas), whereas top-down connections originate from PCs that are located in deep layers and terminate heavily in layer 1 (L1) of target areas. In the somatosensory cortex, subpopulations of L5 PCs project to thalamic nuclei, which subsequently send excitatory afferents to higher-order cortical sensory areas (Theyel et al. 2010; Sherman and Guillery 2011). However, the relationships between direct and transthalamic bottom-up connections are poorly understood, in part because each cortical layer contains multiple subtypes of PCs with complex patterns of axonal collateralization (Douglas and Martin 2004).

In the rat frontal cortex, we have identified 2 nonoverlapping subtypes of L5 PCs that project either to the pontine nuclei (corticopontine [CPn] cells) or to the contralateral cortex (commissural [COM] cells) (Morishima and Kawaguchi 2006; Otsuka and Kawaguchi 2008; Hirai et al. 2012). A similar segregation of these projection neuron subtypes in L5 has been identified in other cortical areas (Hattox and Nelson 2007; Brown and Hestrin 2009). L5 COM cells are heterogeneous in their physiological properties and collateral innervation to the contralateral striatum (Otsuka and Kawaguchi 2008). Both L5 CPn and COM cells have distinct dendritic morphologies and distant axonal targets that are correlated with their vertical position (i.e., depth) within L5 (Morishima et al. 2011; Otsuka and Kawaguchi 2011; Hirai et al. 2012). A portion of CPn cells sends axon collaterals to the thalamus (Hirai et al. 2012). However, it remains to be determined which PC subtypes participate in CC projections between adjacent motor-related areas in the frontal cortex.

The rodent motor cortex is composed of at least 2 areas, such as M1 and the secondary motor area (M2) (Donoghue and Wise 1982; Brecht et al. 2004; Paxinos and Watson 2007), which may correspond to M1 and higher-order motor areas in the primate. To test for functional differentiation between M1 and M2, and to understand their respective roles in behavior, it is necessary to clarify their reciprocal connectivity. In this study, we identified areas M1 and M2 with a combination of microstimulation and immunohistochemical staining, thereby revealing the topography of reciprocal CC connectivity and direct CSp projections in both areas. We further characterized the diversity of CPn cells in these areas. We aimed to clarify which PC subtypes establish CC connections and to determine the relationship between their axonal termination patterns and thalamocortical afferents to M1 and M2. Our results indicate that adjacent motor-related areas are connected by multiple subtypes of L5 PCs innervating specific cortical sublayers that also received distinct thalamic afferent input.

Materials and Methods

Animals

Wistar rats (Charles River Laboratories Japan, Inc., Tsukuba, Japan) of either sex that were 19–23 days, or 3–7 weeks old were used for physiological and histological experiments, respectively. All experiments were conducted in compliance with the guidelines of The Institutional Animal Care and Use Committee of the National Institutes of Natural Sciences.

Intracortical Microstimulation

Animals were anesthetized with a mixture of ketamine (40 mg/kg, intraperitoneal [i.p.]) and xylazine (4 mg/kg, i.p.) and placed in a stereotaxic apparatus. Intracortical microstimulation (ICMS) was conducted while the animals were lightly anesthetized and displayed minimal spontaneous whisker movements. A glass microelectrode with a tip diameter of 20 μm was filled with 2 M NaCl. Extracellular stimulation consisted of 12 monophasic cathodal current pulses (333 Hz; duration, 0.2 ms) that were applied at a depth of 1–1.5 mm from the cortical surface, which corresponded to deep L5 of the frontal motor cortices. Initially, a 60-μA current was used to determine whether any movement was evoked. Stimulation sites were regarded as negative if no obvious movements were evoked with a current strength up to 100 μA. The electrode was inserted in a grid-like fashion at 0.5-mm intervals while using bregma as a reference point and taking care to avoid surface blood vessels.

Retrograde Tracer Injection

Animals were anesthetized with ketamine (40 mg/kg, i.p.) and xylazine (4 mg/kg, i.p.), followed by an injection of glycerol (0.6 g/kg, i.p.) and dexamethasone (1 mg/kg, intramuscular), before being placed in a stereotaxic apparatus. The following tracers were used: Fast Blue (Dr Illing GmbH and Co. KG, Groß-Umstadt, Hesse, Germany; 2% in distilled water), Fluoro-Gold (Fluorochrome LLC, Denver, CO, USA; 5% in distilled water), and Alexa Fluor 555-conjugated cholera toxin subunit B (CTB555, Life Technologies Corporation, Grand Island, NY, USA; 0.2% in distilled water). One or 2 fluorescent tracers with different excitation wavelengths were pressure injected into 1 or 2 target areas using glass pipettes (tip diameter, 50–100 μm; 100 nL in total) and a Pneumatic PicoPump (PV820, World Precision Instruments, Inc., Sarasota, FL, USA). The coordinates for retrograde tracer injections are summarized in Table 1. For a focal injection, CTB555 (0.2% in distilled water) was applied to M1 through iontophoresis (negative current, 0.3 μA; 7-s on/off cycles; 20 min) with glass pipettes (tip diameter, 20–30 μm) that were back filled with 0.9% NaCl.

Table 1

Coordinates of tracer injection sites for histological experiments

Region Coordinates (mm)
 
Insertion angle Retrograde tracer 
Rostrocaudal (from bregma) Mediolateral (from midline) Dorsoventral 
 (From cortical surface) 
M2 +4 to +4.5 1–2 0.2, 0.4, 0.6, 0.8 25°, rostral CTB555, FB 
M1FL 0 to +1.5 1.5–2.5 0.2, 0.4, 0.6 Vertical CTB555, FB 
M1HL −1 to −2 1.5–2 0.2, 0.4, 0.6 Vertical CTB555, FB 
Pn −5.8 to −6 0.8 7.2–7.8 Vertical FB 
vTh −2 to −2.6 1.4–2 5.4–5.8 Vertical CTB555 
Spinal cord (From spinal cord surface) 
 Cervical 1–2  0–1 0.2, 0.4, 0.6, 0.8 Vertical FB, FG 
 Cervical 6–7  1–1.5 1, 1.2, 1.4, 1.6 15°–20°, lateral CTB555, FB 
 Lumbar 2–3  1–1.5 1, 1.2, 1.4, 1.6 15°–20°, lateral FB, FG 
Region Coordinates (mm)
 
Insertion angle Retrograde tracer 
Rostrocaudal (from bregma) Mediolateral (from midline) Dorsoventral 
 (From cortical surface) 
M2 +4 to +4.5 1–2 0.2, 0.4, 0.6, 0.8 25°, rostral CTB555, FB 
M1FL 0 to +1.5 1.5–2.5 0.2, 0.4, 0.6 Vertical CTB555, FB 
M1HL −1 to −2 1.5–2 0.2, 0.4, 0.6 Vertical CTB555, FB 
Pn −5.8 to −6 0.8 7.2–7.8 Vertical FB 
vTh −2 to −2.6 1.4–2 5.4–5.8 Vertical CTB555 
Spinal cord (From spinal cord surface) 
 Cervical 1–2  0–1 0.2, 0.4, 0.6, 0.8 Vertical FB, FG 
 Cervical 6–7  1–1.5 1, 1.2, 1.4, 1.6 15°–20°, lateral CTB555, FB 
 Lumbar 2–3  1–1.5 1, 1.2, 1.4, 1.6 15°–20°, lateral FB, FG 

M1FL: forelimb area of primary motor cortex; M1HL: hindlimb area of primary motor cortex; M2: secondary motor cortex; Pn: pontine nuclei; vTh: ventral thalamic nuclei; CTB555: Alexa Fluor 555-conjugated cholera toxin subunit B; FB: Fast Blue; FG: Fluoro-Gold.

After a survival period of 4–6 days, the animals were deeply anesthetized with sodium pentobarbital (60 mg/kg, i.p.) and perfused transcardially with a prefixative (250 mM sucrose and 5 mM MgCl2 in 0.02 M phosphate-buffered saline [PBS], pH 7.4) solution, which was followed by a fixative (4% paraformaldehyde and 0.2% picric acid) in 0.1 M PB solution. After a postfixation that lasted from 2 h to overnight (or <30 min when combined with chicken ovalbumin upstream promoter transcription factor-interacting protein 2 [Ctip2] immunostaining), the brain was sagittally (Fig. 1A,B) or obliquely (Fig. 1C) cut into 50-μm sections on a vibratome (Leica Microsystems, Inc., Buffalo Grove, IL, USA). In some cases, sections were cut at a thickness of 20 μm in order to observe double labeling. Every 4 sections were collected as a set. The first or third section of each set was used for counting labeled cells. The remaining sections were used for cortical area and layer determination, as described below. In order to count the retrogradely labeled cells, 3–10 discrete oblique or sagittal sections of M2 (1–2 mm lateral to the midline) and sagittal sections of M1 (1–3 mm lateral to midline) from 2 to 4 rats were used. Cell counting was performed in the area that covered the entire cortical layer structures of M2 or M1 with a width of 0.5 mm. To compare the topographical distribution of M2 cells projecting to the forelimb (FL) or hindlimb (HL) areas in M1, all retrogradely labeled cells that were located in the M2 area were counted in each section. CPn and COM cells located in M2 were counted in L5 in an area with a width of 1 mm.

Figure 1.

Histochemical and physiological separation of the rat secondary motor cortex from its surrounding areas. (AC) Immunostaining differences in NF-H (N200 antibody) between the secondary motor cortex (M2) and the surrounding frontal cortical areas. N200 immunoreactivity at L2/3 to upper L5 was weaker in M2 than that in M1 and in the lateral orbital cortex (LO) (A and B, 2 sagittal sections with different lateralities [L 1.5 and L 2 mm, respectively], shown in insets; a pair of arrowheads, area border), and it was much weaker in the anterior cingulate cortex (Cg) than that in M2 (C, oblique horizontal section, shown in inset; a pair of white arrowheads, border between Cg and M2). Cb: cerebellum; Cx: cortex; OB: olfactory bulb; Str: striatum; WM: white matter; D: dorsal; M: medial; R: rostral. (D) Laminar identification by immunofluorescence for NeuN and VGluT2 in M2 (left) and M1 (right). Layer 1 (L1), 2/3 (L2/3), 5 (L5), and 6 (L6) could be identified by the cytoarchitecture. L2/3a has a higher cell density and weaker immunoreactivity for VGluT2 than that for L2/3b. L5 is further divided into 2 sublaminae: L5a, which is weaker, and L5b, which is stronger, for VGluT2 immunoreactivity. L6 was divided into L6a and L6b, which were separated by an intervening neuron-sparse zone. (E) Motor representations evoked by intracortical microstimulation (ICMS) on the cerebral hemisphere of a rat. Black, red, and blue circles correspond to the Vb, FL, and HL regions of M1, respectively. Gray area, a portion of M2 seen from the dorsal surface. A similar map was found in 26 rats (postnatal day 20–43; body weight, 38–178 g). B: bregma; ml: midline; D: dorsal; V: ventral; R: rostral; C: caudal. (F) No movement induction by ICMS (current amplitude: up to 100 μA) was found in the region rostral to the border between M2 and M1. Vb (black circle) or FL (red circle) movements were elicited by ICMS (amplitude: <60 μA) in the region caudal to the M2/M1 border (2 different rats shown). Recording tracks were electrolytically (negative direct current; 7–13 μA, 4–5 min) marked at the most rostral site of motor responses in individual rats (black arrowheads). The areal borders were identified by N200 immunofluorescence (dashed lines).

Figure 1.

Histochemical and physiological separation of the rat secondary motor cortex from its surrounding areas. (AC) Immunostaining differences in NF-H (N200 antibody) between the secondary motor cortex (M2) and the surrounding frontal cortical areas. N200 immunoreactivity at L2/3 to upper L5 was weaker in M2 than that in M1 and in the lateral orbital cortex (LO) (A and B, 2 sagittal sections with different lateralities [L 1.5 and L 2 mm, respectively], shown in insets; a pair of arrowheads, area border), and it was much weaker in the anterior cingulate cortex (Cg) than that in M2 (C, oblique horizontal section, shown in inset; a pair of white arrowheads, border between Cg and M2). Cb: cerebellum; Cx: cortex; OB: olfactory bulb; Str: striatum; WM: white matter; D: dorsal; M: medial; R: rostral. (D) Laminar identification by immunofluorescence for NeuN and VGluT2 in M2 (left) and M1 (right). Layer 1 (L1), 2/3 (L2/3), 5 (L5), and 6 (L6) could be identified by the cytoarchitecture. L2/3a has a higher cell density and weaker immunoreactivity for VGluT2 than that for L2/3b. L5 is further divided into 2 sublaminae: L5a, which is weaker, and L5b, which is stronger, for VGluT2 immunoreactivity. L6 was divided into L6a and L6b, which were separated by an intervening neuron-sparse zone. (E) Motor representations evoked by intracortical microstimulation (ICMS) on the cerebral hemisphere of a rat. Black, red, and blue circles correspond to the Vb, FL, and HL regions of M1, respectively. Gray area, a portion of M2 seen from the dorsal surface. A similar map was found in 26 rats (postnatal day 20–43; body weight, 38–178 g). B: bregma; ml: midline; D: dorsal; V: ventral; R: rostral; C: caudal. (F) No movement induction by ICMS (current amplitude: up to 100 μA) was found in the region rostral to the border between M2 and M1. Vb (black circle) or FL (red circle) movements were elicited by ICMS (amplitude: <60 μA) in the region caudal to the M2/M1 border (2 different rats shown). Recording tracks were electrolytically (negative direct current; 7–13 μA, 4–5 min) marked at the most rostral site of motor responses in individual rats (black arrowheads). The areal borders were identified by N200 immunofluorescence (dashed lines).

Anterograde Tracer Injection

Animals were anesthetized in the same manner as described for retrograde tracer injections. We injected 10% (w/v) biotinylated dextran amine (BDA-10 000 or BDA-10K; Life Technologies Corporation) in 0.5 M potassium acetate into M2 or M1 at 2 sites that were 0.5 mm apart in each area through pressure ejection from glass micropipettes (tip diameter, 50–100 μm).

After a survival period of 7–10 days, the animals were deeply anesthetized with sodium pentobarbital (60 mg/kg, i.p.) and perfused transcardially with a prefixative, which was followed by a fixative. The brain was postfixed for 2 h to overnight at 4°C and cut sagittally into 50-μm sections on a vibratome (Leica Microsystems, Inc.). Every 4 sections were collected as a set. In each set, the first and third sections were used for BDA-10K detection with Alexa Fluor-conjugated streptavidin or 3,3′-diaminobenzidine tetrahydrochloride (DAB), as described below. The remaining sections were used for area and layer identification.

BDA-10K was visualized by incubating sections with the avidin–biotin–peroxidase complex (1%; ABC Elite, Vector Laboratories, Inc., Burlingame, CA, USA) in 0.05 M Tris-buffered saline (TBS) overnight. To enhance the signal, sections were reacted for 30 min with 2.5 μM biotinylated tyramine, 3 μg/mL glucose oxidase, and 2 mg/mL β-d-glucose in 2% bovine serum albumin, which was dissolved in 0.05 M TB solution. Sections were subsequently incubated with Alexa Fluor 488-conjugated streptavidin (Life Technologies Corporation) or ABC Elite (Vector Laboratories, Inc.), which was followed by DAB reaction.

To evaluate the laminar distribution of CC projection targets in M2 and M1, we measured the density of anterogradely labeled signals in an area with a width of 0.1 mm in each motor cortical area. As most CC axons traveled through L6, parallel to the white matter (Vandevelde et al. 1996; our observations), we did not analyze signal density in L6.

In Vitro Electrophysiological Recording of Retrogradely Labeled Cells

Rats (postnatal days 17–21) were anesthetized with a mixture of ketamine (40 mg/kg, i.p.) and xylazine (4 mg/kg, i.p.) and placed in a stereotaxic apparatus. For CSp cell labeling, a fluorescent retrograde tracer (CTB555 [0.2% in distilled water] or red fluorescent “Retrobeads” [red beads; Lumafluor, Inc., Durham, NC, USA]) was injected at 4 different depths (0.2, 0.4, 0.6, and 0.8 mm from the dorsal surface of the spinal cord) of cervical (C1–2) segments. For the double labeling of COM and M1-projecting cells, a red fluorescent retrograde tracer (CTB555 or red beads) was injected into contralateral M2, as described previously (Otsuka and Kawaguchi 2011), whereas a green fluorescent retrograde tracer (Alexa Fluor 488- [Life Technologies Corporation], or fluorescein isothiocyanate-conjugated CTB [Sigma-Aldrich Co. LLC, St. Louis, MO, USA]) was injected into ipsilateral M1. For the double labeling of corticothalamic (CTh) and M1-projecting cells, CTB555 and green beads (Lumafluor, Inc.) were injected into the ipsilateral ventral thalamic nuclei and M1, respectively.

One or 2 days after tracer injection, animals were deeply anesthetized with isoflurane and decapitated. The brain was quickly removed and submerged in ice-cold physiological Ringer's solution. Six slices with a thickness of 300 μm were obtained from M2 or M1 contralateral to the spinal cord injection side. CTh/M1-projecting cells were recorded from M2 ipsilateral to the injection site. To preserve the dendritic structures of the PCs, slices from M2 were obtained in an oblique horizontal direction and those from M1 were obtained with a rostral inclination of 10°. Slices were immersed in a buffered solution containing the following (in mM): 125 NaCl, 2.5 KCl, 2.0 CaCl2, 1.0 MgCl2, 25 NaHCO3, 1.25 NaH2PO4, 10 glucose, and 4 lactic acid. This solution was continuously bubbled with a mixture of 95% O2 and 5% CO2. Lactic acid was omitted during recordings. The membrane potentials of retrogradely labeled cells were recorded in a whole-cell mode at 30–31°C. Labeled cells were identified with epifluorescence microscopy (BX50WI, Olympus Corporation, Tokyo, Japan) with a ×40 water-immersion objective (numerical aperture = 0.8, Olympus Corporation).

The pipette solution for current-clamp recording consisted of the following (in mM): 130 potassium gluconate, 2 KCl, 2 MgCl2, 2 Na2ATP, 0.3 GTP, and 10 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), with 0.75% biocytin. pH of the solution was adjusted to 7.2–7.3 with KOH, and the osmolarity was 290 mOsm. For current-clamp recording from COM cells that also projected to M1, recording pipettes were filled with a solution containing the following (in mM): 130 potassium methylsulfate, 0.5 ethylene glycol tetraacetic acid, 2 MgCl2, 2 Na2ATP, 0.2 GTP, 20 HEPES, and 0.1 leupeptin, with 0.75% biocytin. pH of the solution was adjusted to 7.2 with KOH, and the osmolarity was 290 ± 5 mOsm. The membrane potentials were not corrected for liquid junction potentials. The series resistance of the recording cells was <25 MΩ. The resting membrane potential was recorded soon after whole-cell patch clamping. The input resistance was determined by the linear fitting of voltage responses to hyperpolarizing current injections and recorded within 5 min from whole-cell break-in. The membrane time constant was assessed based on the response to transient hyperpolarizing current pulses (−50 pA). Frequency–current relationships of CSp cells were obtained from all interspike intervals, including initial doublet firings, induced by depolarization pulses of 1 s. Recordings from CSp or CTh/M1-projecting cells were amplified with an Axoclamp 700B amplifier (Molecular Devices, LLC, Sunnyvale, CA, USA), digitized at 10 kHz with a Digidata 1440A apparatus (Molecular Devices, LLC), and collected with pClamp10 (Molecular Devices, LLC). Data were analyzed with IGOR Pro (WaveMetrics, Inc., Lake Oswego, OR, USA) with NeuroMatic functions (http://www.neuromatic.thinkrandom.com, last accessed March 5, 2013). Recordings from COM/M1-projecting cells were performed using an EPC-9 double amplifier (HEKA Electronik, Lambrecht, Germany).

Slices containing biocytin-loaded cells were fixed in 4% paraformaldehyde, 1.25% glutaraldehyde, and 0.2% picric acid overnight at 4°C and resectioned at a thickness of 50 μm. The sections were incubated with 1% ABC Elite (Vector Laboratories, Inc.) in 0.05 M TBS containing 0.04% Triton X-100 overnight at 4°C. After washing in TBS, the sections were reacted with 0.02% DAB and 0.001% H2O2 in 0.05 M TB. Cells were postfixed in 1% OsO4 in 0.1 M PB containing 7% glucose, dehydrated, and embedded on glass slides in Epon. The stained cells were reconstructed 3 dimensionally with the NeuroLucida system (MBF Bioscience, Williston, VT, USA) and analyzed quantitatively with NeuroExplorer (MBF Bioscience).

Immunohistochemistry

For immunofluorescence staining, sections were incubated overnight with a mouse monoclonal antibody against the neurofilament heavy chain (NF-H; N-200 antibody, N0142, Sigma-Aldrich Co. LLC; 1:1000), a mouse monoclonal antibody against neuronal nuclei (NeuN; MAB377, EMD Millipore Corporation, Billerica, MA, USA; 1:5000), a rabbit polyclonal antibody against calbindin D-28K (CB-38a, Swant, Marly, Switzerland; 1:2000), a guinea pig polyclonal antibody against the vesicular glutamate transporter type 2 (VGluT2; AB2251, EMD Millipore Corporation; 1:5000), or a rat monoclonal antibody against Ctip2 (ab18465, Abcam plc, Cambridge, UK; 1:500), in 0.05 M TBS containing 10% normal goat serum, 2% bovine serum albumin, and 0.5% Triton X-100. After washes with TBS, the sections were incubated with an Alexa Fluor-conjugated secondary antibody (1:200; Life Technologies Corporation) for 2–3 h at room temperature. The sections were mounted on glass slides, coverslipped with Krystalon mounting medium (EMD Millipore Corporation), Vectashield mounting medium (Vector Laboratories, Inc.), or Prolong gold antifade reagent (Life Technologies Corporation), and observed with epifluorescence (for Fast Blue, Fluoro-Gold, and Alexa Fluor 350: Excitation, 360–370 nm and emission, 420–460 nm; for Alexa Fluor 488: Excitation, 470–490 nm and emission, 510–550 nm; for Alexa Fluor 555: Excitation, 545–580 nm and emission, >610 nm). The contrast and brightness of the digital images were modified with Adobe Photoshop (Adobe Systems, Inc., San Jose, CA, USA), and images were saved as TIFF files.

Ctip2 Immunostaining in VGAT-Venus Rat

Vesicular gamma-aminobutyric acid (GABA) transporter (VGAT)-Venus transgenic rats, which express the fluorescent protein Venus in GABAergic cells, were used to identify GABAergic cells (Uematsu et al. 2008). VGAT-Venus transgenic rats were generated by Drs Y. Yanagawa, M. Hirabayashi, and Y. Kawaguchi at the National Institute for Physiological Sciences with pCS2-Venus that was provided by Dr A. Miyawaki. VGAT-Venus rats are distributed by The National BioResource Project for the Rat in Japan (http://www.anim.med.kyoto-u.ac.jp:80/nbr/default.aspx, last accessed March 26, 2013). After perfusion with a fixative, as described above, the brains of 4- to 5-week-old VGAT-Venus rats were removed after <30 min of postfixation and washed with 0.1 M PB. We confirmed that the anti-Ctip2 antibody fully penetrated into the 20-μm-thick sections (Fluoview FV1000 confocal system, Olympus Corporation). Subsequently, the Venus expression among Ctip2-positive cells was investigated in 20-μm sagittal sections that were cut on a vibratome (Leica Microsystems, Inc.). Venus-positive cells were counted among Ctip2-positive cells in 3 sections from individual VGAT-Venus rats (width, 1 mm).

To examine the proportions of Ctip2-positive cells among deep-layer excitatory cells, Ctip2 and NeuN were simultaneously immunostained in 8-μm-thick sections of M2 that were cut obliquely on a cryostat (Leica Microsystems, Inc.). Sections were incubated overnight with a mouse monoclonal antibody against NeuN, a chicken polyclonal antibody against GFP/Venus (ab13970, Abcam plc; 1:1000), a guinea pig polyclonal antibody against VGluT2, and a rat monoclonal antibody against Ctip2 in 0.05 M TBS containing 10% normal goat serum, 2% bovine serum albumin, and 0.2% Triton X-100. After washing in TBS, sections were reacted with biotinylated- (for NeuN), Alexa Fluor 488- (for Venus), Alexa Fluor 488- (for VGluT2), and Alexa Fluor 594- (for Ctip2) conjugated secondary antibodies. The NeuN signal was detected by further incubation with Alexa Fluor 350-conjugated streptavidin. The same immunofluorescence signals (Alexa Fluor 488) for Venus (GABAergic cell identification) and VGluT2 (L5 sublayer identification) could be discriminated easily based on labeling pattern differences (somata in the former; fibers and boutons in the latter). Ctip2-positive and Ctip2-negative cells were counted among the Venus-negative/NeuN-positive cells in L5/6 in 7 sections from individual VGAT-Venus rats.

Determination of the Neocortical Area, Layer, and Thalamic Nuclei

We defined the border between M2 and M1 based on differences in NF-H immunoreactivity (Fig. 1A,B). In both M2 and M1, L1, L2/3, and L5 were divided into sublayers by the immunostaining for VGluT2, which is expressed in thalamocortical fibers (Kaneko et al. 2002). VGluT2 expression was stronger in L1a, L2/3b, and L5b than that in L1b, L2/3a, and L5a (Fig. 1D; Morishima et al. 2011). Layer 6 (L6) was divided into L6a and L6b, which were separated by an intervening neuron-sparse zone that was identified with pan-neuronal staining (Fig. 1D). Thalamic nuclei were identified based on calbindin expression patterns and pan-neuronal staining, as reported previously (Ushimaru et al. 2012).

Statistics

Data are presented as the mean ± standard deviation (SD). Two-way analysis of variance (ANOVA) and post hoc Bonferroni tests were used for statistical comparisons of the spatial or laminar distribution patterns of retrogradely labeled cells, or of the laminar distribution patterns of anterogradely labeled fibers in M2 and M1. The sublaminar distribution of retrogradely labeled cells within M2 or M1 was compared with the supposed uniform distribution of the labeled cells among the sublayers with a 1-sample t-test. The Mann–Whitney U-test was used for 2 group comparisons. The morphological properties of M2 CSp cells were subjected to a cluster analysis with Ward's method. Pairwise data of the Ctip2 proportions among CC cells were compared with a chi-squared test. Significance was set at P-values <0.05.

Results

Identification of the Rat Secondary Motor Cortex

The dorsal surface of the rat frontal cortex contains M1 in the caudal region and a smaller section of M2 in the rostral region (Neafsey et al. 1986; Brecht et al. 2004). We found that the area of M2 was histochemically delineated from M1 and other surrounding frontal areas based on immunostaining pattern of NF-H, which has been used to identify various other cortical areas in rodents (Van der Gucht et al. 2007; Van De Werd and Uylings 2008; Hirai et al. 2012). NF-H expression from L2/3 to upper L5 in the rostral frontal cortex corresponding to M2 was weaker than that in caudal and lateral areas corresponding to M1 (Fig. 1AC), or ventral areas corresponding to the lateral orbital area (Fig. 1A,B). However, NF-H expression in M2 was stronger than that in the anterior cingulate area (Fig. 1C). M2 was also overall thicker than M1, mostly due to a higher proportion of L6, although the overall laminar structure was similar between both areas (Fig. 1D).

To confirm that the border between M2 and M1, which was identified histochemically by NF-H expression, corresponded to the functional boundary of these areas, we evoked motor responses by ICMS. In area of M1, motor movements are evoked at lower current intensities than in surrounding areas, including M2 (Brecht et al. 2004). ICMS delivered at an intensity of <60 μA at individual loci within M1 induced the following movements: 1) movements of vibrissa (Vb) were induced close to the midline and anterior to bregma; (2) movements of the FL were induced by stimulations lateral to the Vb region and mainly anterior to bregma; and (3) movements of the HL were induced caudal to the FL region (Fig. 1E). These motor representations correspond well to those reported previously (Miyashita et al. 1994; Huntley 1997; Brecht et al. 2004). Conversely, motor responses were hardly ever evoked in the region rostral to the histochemically delineated M2/M1 border, even when the intensity of the current was increased up to 100 μA (Fig. 1F).

Lack of CSp Projection Topography in M2

Retrograde labeling experiments in which tracers were injected into the cervical and lumbar regions of the spinal cord found that M1 contained both cervical and lumbar spinal cord-projecting CSp cells that had different cortical surface distributions, whereas CSp cells in M2 did not exhibit such differentiated cortical topography. Direct projections to the spinal cord by CSp cells, which are abundantly distributed in M1, are involved in induction of limb movements, as was observed with ICMS. CSp cells are also found in rostral regions (Miller 1987), which probably includes M2. In both M2 and M1, we found CSp cells that projected to the cervical spinal cord in L5b (“Cerv” in Fig. 2A; red dots in Fig. 2B). Cervical spinal cord-projecting cells were distributed all over M2 (retrograde tracer: CTB555, n = 2; Fast Blue, n = 2; and Fluoro-Gold, n = 2). Conversely, CSp cells projecting to the lumbar spinal cord were found in L5b of M1, but were not observed in M2 (“Lumb” in Fig. 2A; blue dots in Fig. 2B). This was verified with Fast Blue (n = 1) and Fluoro-Gold (n = 2). Within M1, CSp cells projecting to the cervical and lumbar spinal cords exhibited different rostrocaudal distributions (Fig. 2B), which largely corresponded, respectively, to the FL and HL regions identified by ICMS (Fig. 1E). A portion of M1 contained fewer CSp cells that projected to either the cervical or lumbar spinal cord (Fig. 2B), and this area largely corresponded to the Vb area identified by ICMS (Fig. 1E). Using retrograde labeling of CSp cells, we confirmed that ICMS in the M2 region containing labeled CSp cells did not induce movements (n = 2; data not shown).

Figure 2.

Distribution of CSp cells and their firing characteristics in M2 and M1. (A) Spinal cord injection of 2 different retrograde fluorescent tracers at the cervical (C6–7; CTB555) and lumbar (L2–3; Fast Blue) levels. In order to avoid labeling from the dorsal CSp tract (DCS), injection pipettes were advanced with a lateral inclination. At the cervical spinal cord, injections were mostly restricted to the gray matter (GM). Cerv: cervical spinal cord; Lumb: lumbar spinal cord; WM: white matter. (B) Distributions of CSp cells projecting to the cervical (red dots) and lumbar (blue dots) spinal cords. Two single sagittal sections, which were different in laterality. M2 CSp cells projected to the cervical, but not to the lumbar, spinal cord. In M1, the cervical and lumbar spinal cord-projecting CSp cells were distributed along the rostrocaudal direction, which is intersected by the CSp cell-sparse zone. Areal borders were identified by N200 immunofluorescence. L5b, between the gray lines (identified by VGluT2 immunofluorescence). (C) Firing responses to a depolarizing current pulse (duration, 1 s). Both M2 and M1 CSp cells projecting to the upper cervical spinal cord typically showed SA firing with the initial doublet (magnified at right). (D) Similar firing frequency–current relationship in M2 (n = 46, gray triangle) and M1 (n = 29, black circle) CSp cells. Error bar, SD.

Figure 2.

Distribution of CSp cells and their firing characteristics in M2 and M1. (A) Spinal cord injection of 2 different retrograde fluorescent tracers at the cervical (C6–7; CTB555) and lumbar (L2–3; Fast Blue) levels. In order to avoid labeling from the dorsal CSp tract (DCS), injection pipettes were advanced with a lateral inclination. At the cervical spinal cord, injections were mostly restricted to the gray matter (GM). Cerv: cervical spinal cord; Lumb: lumbar spinal cord; WM: white matter. (B) Distributions of CSp cells projecting to the cervical (red dots) and lumbar (blue dots) spinal cords. Two single sagittal sections, which were different in laterality. M2 CSp cells projected to the cervical, but not to the lumbar, spinal cord. In M1, the cervical and lumbar spinal cord-projecting CSp cells were distributed along the rostrocaudal direction, which is intersected by the CSp cell-sparse zone. Areal borders were identified by N200 immunofluorescence. L5b, between the gray lines (identified by VGluT2 immunofluorescence). (C) Firing responses to a depolarizing current pulse (duration, 1 s). Both M2 and M1 CSp cells projecting to the upper cervical spinal cord typically showed SA firing with the initial doublet (magnified at right). (D) Similar firing frequency–current relationship in M2 (n = 46, gray triangle) and M1 (n = 29, black circle) CSp cells. Error bar, SD.

Both M2 and M1 contained CSp cells projecting to the cervical spinal cord, but these regions were differentially involved in FL movement induction, as assessed by ICMS. Using in vitro electrophysiological recordings from labeled cervical spinal cord-projecting CSp cells, we found that the difference in ICMS-evoked movement induction was not due to differences in intrinsic physiological properties of CSp cells. CSp cells in both cortical areas showed initial action potential “doublets” at depolarization onset (Fig. 2C, inset, initial doublet firing; Table 2) and slow adaptation of firing during a depolarizing pulse (Table 2; rats, 19–23 days postnatal). These cells also had similar passive membrane properties (Table 2) and frequency–current relationships (Fig. 2D and Table 2) that were similar to, but slightly different from, those observed in the motor or sensory cortex of older mice (Miller et al. 2008; Suter et al. 2013).

Table 2

Electrophysiological properties of M2 and M1 CSp cells

 M2 CSp cells (n = 46) M1 CSp cells (n = 29) 
Passive membrane properties 
 Resting potential (mV) −63.8 ± 3.1 −64.1 ± 3.8 
 Input resistance (MΩ) 53.8 ± 14.4 50.8 ± 15.5 
 Time constant (ms) 13.1 ± 5.6 13 ± 5.9 
Firing properties 
 Mean frequency (Hz) at 0.6 nA 23.3 ± 6 24.6 ± 5.1 
 First ISI (Hz) at 0.6 nA 101.4 ± 49.4 106.7 ± 59.9 
 Third ISI 24.3 ± 9.1 26.3 ± 13 
 Last ISI 21.8 ± 4.8 23.4 ± 6.4 
 M2 CSp cells (n = 46) M1 CSp cells (n = 29) 
Passive membrane properties 
 Resting potential (mV) −63.8 ± 3.1 −64.1 ± 3.8 
 Input resistance (MΩ) 53.8 ± 14.4 50.8 ± 15.5 
 Time constant (ms) 13.1 ± 5.6 13 ± 5.9 
Firing properties 
 Mean frequency (Hz) at 0.6 nA 23.3 ± 6 24.6 ± 5.1 
 First ISI (Hz) at 0.6 nA 101.4 ± 49.4 106.7 ± 59.9 
 Third ISI 24.3 ± 9.1 26.3 ± 13 
 Last ISI 21.8 ± 4.8 23.4 ± 6.4 

Note: Data are presented as mean ± SD.

CSp: corticospinal; ISI: interspike interval.

CPn cells exhibit different morphological and connectional characteristics in L5a and L5b (Morishima et al. 2011). We also found that CSp cells are differentiated according to their somatic depth within L5b, based on differences in dendritic morphology. We reconstructed the apical and basal dendrites of 13 M2 CSp cells. Among them, 12 PCs had a typical pyramidal shape (Fig. 3A), but 1 exhibited an inverted morphology in which the “apical” dendrite extended into L6 (Fig. 3C). Analysis of the reconstructed dendrites revealed that branching patterns of apical oblique and basal dendrites were heterogeneous among CSp cells (Fig. 3A). With cluster analysis, we tentatively divided M2 CSp cells into the following 2 morphologically distinct groups: Cluster I cells had longer apical side branches in L5a and fewer basal dendritic nodes relative to cluster II cells (Fig. 3B). Importantly, cluster I CSp cells were localized more superficially within L5b relative to cluster II CSp cells (Table 3). Apical side branches in L2/3 were also more developed among cluster I CSp cells (Table 3). These observations suggest further, depth-dependent, differentiation of input structures (dendrites) among CSp cells.

Table 3

Morphological properties of superficial and deep types of M2 CSp cells

 M2 CSp cells
 
Cluster I (superficial type) (n = 6) Cluster II (deep type) (n = 6) 
Soma position 
 From pia (μm) 892 ± 99 1022 ± 59* 
 From L5a/L5b border (μm) 140 ± 61 269 ± 47** 
 Relative depth within L5ba 0.39 ± 0.16 0.73 ± 0.13** 
Basal dendrites 
 Primary dendrite number 7.2 ± 1.6 8.2 ± 1.3 
 Branching points (nodes)b 33.8 ± 8.9 51.2 ± 7.5* 
 Internode interval (μm) 23 ± 2.7 20.2 ± 3.7 
Apical dendrites 
 Tuft origin from L1/L2/3 border (μm) 274 ± 147 534 ± 137* 
 Branch length in L2/3 (×103 μm) 979 ± 720 3 ± 7** 
 Branch length in L5a (×103 μm)b 2159 ± 485 657 ± 414** 
 M2 CSp cells
 
Cluster I (superficial type) (n = 6) Cluster II (deep type) (n = 6) 
Soma position 
 From pia (μm) 892 ± 99 1022 ± 59* 
 From L5a/L5b border (μm) 140 ± 61 269 ± 47** 
 Relative depth within L5ba 0.39 ± 0.16 0.73 ± 0.13** 
Basal dendrites 
 Primary dendrite number 7.2 ± 1.6 8.2 ± 1.3 
 Branching points (nodes)b 33.8 ± 8.9 51.2 ± 7.5* 
 Internode interval (μm) 23 ± 2.7 20.2 ± 3.7 
Apical dendrites 
 Tuft origin from L1/L2/3 border (μm) 274 ± 147 534 ± 137* 
 Branch length in L2/3 (×103 μm) 979 ± 720 3 ± 7** 
 Branch length in L5a (×103 μm)b 2159 ± 485 657 ± 414** 

Note: Data are presented as mean ± SD.

aNormalized L5b thickness; L5a/L5b border = 0, L5b/L6 border = 1.

bParameters used in the cluster analysis.

*P < 0.05; **P < 0.01; Mann–Whitney U-test.

Figure 3.

L5b depth-dependent differentiation of CSp cells. (A) Dendrite reconstruction of M2 CSp cells. Cells were tentatively divided into superficial and deep types based on the relative depth of somata within L5b. (B) Two groups of M2 CSp cells identified by cluster analysis with 2 dendritic parameters: Oblique branch length in L5a and node number in basal dendrites. Color-coded scale, relative depth of somata from L5b top (red) to L5b bottom (blue). Roman numeral, cell number common in (A) and (B). (C) Inverted type of CSp cell.

Figure 3.

L5b depth-dependent differentiation of CSp cells. (A) Dendrite reconstruction of M2 CSp cells. Cells were tentatively divided into superficial and deep types based on the relative depth of somata within L5b. (B) Two groups of M2 CSp cells identified by cluster analysis with 2 dendritic parameters: Oblique branch length in L5a and node number in basal dendrites. Color-coded scale, relative depth of somata from L5b top (red) to L5b bottom (blue). Roman numeral, cell number common in (A) and (B). (C) Inverted type of CSp cell.

Topographical Organization of CC Projections from M2 to M1

We have shown that the FL and HL territories in M1 innervate the cervical and lumbar spinal cords, respectively, but that M2 projects only to the cervical spinal cord. To understand the function of M2, it would be informative to characterize the topographical projection from M2 to M1. We revealed that M1-projecting neurons in M2 were topographically represented within M2 according to their projections to the FL and HL regions of M1. We observed that CC cells that projected to the rostral FL region of M1 were distributed in the lateral part of M2 (Fig. 4A,B; CTB555, n = 3), whereas neurons that projected to the caudal HL region of M1 were distributed in the medial part of M2 (Fig. 4A,C; CTB555, n = 3). We injected tracer into ICMS-identified FL or HL cortex (Fig. 4A; FL, CTB555, n = 2; HL, CTB555 and Fast Blue, each n = 1) and confirmed that lateral and medial regions of M2 projected to both the FL and HL sites, respectively. Subsequently, we quantitatively compared the distributions of CC cells that projected to each M1 region according to their position within the mediolateral axis of M2 (Fig. 4B,C; bin, 50 μm). Distances from the midline were shorter in CC cells that projected to the HL territory (caudal M1 region, 1866 cells in total [3 rats]; ICMS-identified HL region, 1855 cells in total [2 rats]) than that in CC cells that projected to the FL territory (rostral M1 region, 5056 cells in total [3 rats]; ICMS-identified FL region, 855 cells in total [2 rats]; P < 0.01, 2-way ANOVA; Fig. 4D; dotted line, median). HL territory-projecting cells were more abundant between 0.9 and 1.5 mm from the midline, whereas FL territory-projecting cells tended to be more lateral than 1.5 mm from the midline (Fig. 4E; *P < 0.05; **P < 0.01; post hoc Bonferroni test).

Figure 4.

Topographical CC projections from M2 to M1. (A) Injection sites of a retrograde tracer (CTB555, 9 sites and Fast Blue, 1 site). Upper, rostrocaudal and mediolateral coordinates of 10 injections. The M1 motor representation was validated in 4 sites by ICMS (FL, 2 sites and HL, 2 sites). B: bregma; ml: midline. Lower, tracer deposits (gray area) in L1 to L5 of the FL (left) and HL (right) regions. Hipp: hippocampus; Str: striatum; WM: white matter. (B) Upper, the distribution of M2 CC cells projecting to the rostral M1 region (FL territory). Each dot corresponds to a single retrogradely labeled cell. Vertical gray dashed lines indicate positions located 0.5, 1, 1.5, 2, 2.5, and 3 mm from the midline. Lower, the distance distribution of rostral M1 region-projecting CC cells from the midline that was obtained from the above section. Cell numbers (175 cells in total) were shown every 50 μm (from the midline to 3 mm lateral; 60 bins in total) and normalized with a maximum bin = 1. Oblique section, 20 μm thick. (C) Upper, the distribution of M2 CC cells projecting to the caudal M1 region (HL territory). Lower, the distance distribution of caudal M1 region-projecting CC cells obtained from the above section (144 cells in total). Oblique section, 20 μm thick. (D) Cumulative distributions of CC cells obtained from 10 different injections shown in (A) (total cell number = 963 ± 796). Arrows, median of distances (dashed line) in individual injections. (E) M2 CC cells projecting to the HL territory were situated more medially than those projecting to the FL territory (mean ± SD; 5 rats; P < 0.01; 2-way ANOVA). HL territory-projecting cells were more abundant between 0.9 and 1.3 mm from the midline, whereas FL territory-projecting ones were more abundant in the region located >1.5 mm lateral to the midline. **P < 0.01; *P < 0.05 (post hoc Bonferroni test).

Figure 4.

Topographical CC projections from M2 to M1. (A) Injection sites of a retrograde tracer (CTB555, 9 sites and Fast Blue, 1 site). Upper, rostrocaudal and mediolateral coordinates of 10 injections. The M1 motor representation was validated in 4 sites by ICMS (FL, 2 sites and HL, 2 sites). B: bregma; ml: midline. Lower, tracer deposits (gray area) in L1 to L5 of the FL (left) and HL (right) regions. Hipp: hippocampus; Str: striatum; WM: white matter. (B) Upper, the distribution of M2 CC cells projecting to the rostral M1 region (FL territory). Each dot corresponds to a single retrogradely labeled cell. Vertical gray dashed lines indicate positions located 0.5, 1, 1.5, 2, 2.5, and 3 mm from the midline. Lower, the distance distribution of rostral M1 region-projecting CC cells from the midline that was obtained from the above section. Cell numbers (175 cells in total) were shown every 50 μm (from the midline to 3 mm lateral; 60 bins in total) and normalized with a maximum bin = 1. Oblique section, 20 μm thick. (C) Upper, the distribution of M2 CC cells projecting to the caudal M1 region (HL territory). Lower, the distance distribution of caudal M1 region-projecting CC cells obtained from the above section (144 cells in total). Oblique section, 20 μm thick. (D) Cumulative distributions of CC cells obtained from 10 different injections shown in (A) (total cell number = 963 ± 796). Arrows, median of distances (dashed line) in individual injections. (E) M2 CC cells projecting to the HL territory were situated more medially than those projecting to the FL territory (mean ± SD; 5 rats; P < 0.01; 2-way ANOVA). HL territory-projecting cells were more abundant between 0.9 and 1.3 mm from the midline, whereas FL territory-projecting ones were more abundant in the region located >1.5 mm lateral to the midline. **P < 0.01; *P < 0.05 (post hoc Bonferroni test).

Different Laminar Distributions of CC Cells Between M2 and M1

By comparing the laminar distributions of retrogradely labeled CC cells in the 2 areas (from M2 to M1, CTB555, n = 3; from M1 to M2, CTB555, n = 3), we found that CC cells in M2 and M1 were differentially distributed within cortical lamina. In both areas, CC cells were found in L2/3–L6 (Fig. 5A), but somata were significantly fewer in L2/3a, L5b, and L6a of M2 and in L6a of M1 than would be expected given the thickness of each sublamina (Fig. 5B,C; *P < 0.05; **P < 0.01; 1-sample t-test). The percentages of CC cells in the superficial (L2/3) and deep (L5 and L6) layers were similar between M2 (37.2% in the superficial and 62.8% in the deep layers, SD = 9%) and M1 (48.2% in the superficial and 51.8% in the deep layers, SD = 4.2%; P = 0.13, chi-squared test). However, the distributions of CC cells in individual sublayers were significantly different in the 2 areas (Fig. 5B; P < 0.01, 2-way ANOVA; 3 rats; 1451 cells in M2 and 1866 cells in M1). CC cells were more abundant in L5a of M2, whereas they were more abundant in L5b of M1 (Fig. 5C; P < 0.05, post hoc Bonferroni test). These sublaminar distribution patterns of M2 CC cells were similar, regardless of whether they were labeled from M1 injections in the FL or HL region identified by ICMS (P = 0.17, 2-way ANOVA; 3 rats in individual areas; 1064 cells projecting to the FL region and 801 cells projecting to the HL region).

Figure 5.

Direction-dependent laminar localizations of CC cells and their axon fibers connecting M2 and M1. (A) Laminar distributions of M2 CC cells projecting to M1 (left) and of M1 CC cells projecting to M2 (right). A fluorescent retrograde tracer (CTB555) was injected into L1 to L5 of individual regions. In order to determine the sublaminar structure, adjacent sections were immunostained for VGluT2 and NeuN. Sections, 0.5 mm wide and 50 μm thick. (B) Laminar proportion of labeled CC cells: M2 cells projecting to M1 (gray bar) and M1 cells projecting to M2 (white bar) (mean ± SD; 3 rats, total cell number = 1451 in M2 and 1866 in M1). CC cells were less dense in L2/3a, L5b, and L6a of M2 and in L6a of M1 compared with the supposed uniform distributions among the sublayers (<*, significantly fewer [P < 0.05]; <**, significantly fewer [P < 0.01]; 1-sample t-test). The sublaminar distribution pattern was different between M2 and M1 (P < 0.01; 2-way ANOVA). L5a CC cells were more abundant in M2, whereas L5b CC cells were more abundant in M1 (*P < 0.05; post hoc Bonferroni test). (C) Schema of the different sublaminar localizations of CC cells connecting M2 and M1. (D) CC fibers labeled with BDA-10K injections into L1 to L5 of individual regions. CC fibers from M1 were distributed predominantly in L2/3 of M2 (left), whereas those from M2 were distributed predominantly in L1 of M1 (right). Sections, 0.5 mm wide and 50 μm thick. (E) Laminar density differences in labeled fibers between M2 and M1 (mean ± SD; 3 rats for M1 and 5 rats for M2; P < 0.01; 2-way ANOVA). Compared with fibers from M1, fibers from M2 had a higher density in L1a, but a lower density in L2/3a and L2/3b of the target area. Fiber density was calculated in a 0.1-mm wide area in each section. **P < 0.01; *P < 0.05; post hoc Bonferroni test. Dotted line, sublayer border.

Figure 5.

Direction-dependent laminar localizations of CC cells and their axon fibers connecting M2 and M1. (A) Laminar distributions of M2 CC cells projecting to M1 (left) and of M1 CC cells projecting to M2 (right). A fluorescent retrograde tracer (CTB555) was injected into L1 to L5 of individual regions. In order to determine the sublaminar structure, adjacent sections were immunostained for VGluT2 and NeuN. Sections, 0.5 mm wide and 50 μm thick. (B) Laminar proportion of labeled CC cells: M2 cells projecting to M1 (gray bar) and M1 cells projecting to M2 (white bar) (mean ± SD; 3 rats, total cell number = 1451 in M2 and 1866 in M1). CC cells were less dense in L2/3a, L5b, and L6a of M2 and in L6a of M1 compared with the supposed uniform distributions among the sublayers (<*, significantly fewer [P < 0.05]; <**, significantly fewer [P < 0.01]; 1-sample t-test). The sublaminar distribution pattern was different between M2 and M1 (P < 0.01; 2-way ANOVA). L5a CC cells were more abundant in M2, whereas L5b CC cells were more abundant in M1 (*P < 0.05; post hoc Bonferroni test). (C) Schema of the different sublaminar localizations of CC cells connecting M2 and M1. (D) CC fibers labeled with BDA-10K injections into L1 to L5 of individual regions. CC fibers from M1 were distributed predominantly in L2/3 of M2 (left), whereas those from M2 were distributed predominantly in L1 of M1 (right). Sections, 0.5 mm wide and 50 μm thick. (E) Laminar density differences in labeled fibers between M2 and M1 (mean ± SD; 3 rats for M1 and 5 rats for M2; P < 0.01; 2-way ANOVA). Compared with fibers from M1, fibers from M2 had a higher density in L1a, but a lower density in L2/3a and L2/3b of the target area. Fiber density was calculated in a 0.1-mm wide area in each section. **P < 0.01; *P < 0.05; post hoc Bonferroni test. Dotted line, sublayer border.

Different Laminar Distributions of CC Fibers Between M2 and M1

In addition to observing differences in the laminar distributions of CC cell somata in source areas (M2 and M1), using anterograde labeling of fibers from M2 or M1, we identified differences in the laminar patterns of CC axonal innervation of target areas. Labeled fibers from M1 were distributed preferentially within L2/3 of M2 (Fig. 5D, left). Conversely, axons arriving in M1 from M2 were densely distributed in L1a of M1 (Fig. 5D, right). To compare quantitatively the sublaminar distribution of CC fibers, we normalized the density of labeled fibers (bin, 1 μm deep × 100 μm wide) to the maximum value (max = 1) in each section and obtained the average value of the normalized density in individual sublayers. Axon densities of sublayers were compared in the 2 cortical areas (Fig. 5E; M2, 15 sections from 5 rats and M1, 9 sections from 3 rats). The laminar distribution pattern of fibers was significantly different between the 2 areas (P < 0.01, 2-way ANOVA). In particular, the density of CC fibers in L1a was lower in M2 than that in M1 (P < 0.05, post hoc Bonferroni test), whereas the density of CC fibers in L2/3a and L2/3b was higher in M2 than that in M1 (L2/3a, P < 0.05 and L2/3b, P < 0.01). These data indicate that CC fibers from M2 and M1 preferentially target, respectively, L1a of M1 and L2/3 of M2. Fiber densities in L5 were similar in M2 and M1, and in M2, we observed a similar laminar distribution of fibers originating from the FL (3 rats) and HL (2 rats) regions.

Collateralization of L5 CPn Cells to the Adjacent Motor Area

Retrograde labeling revealed differential sublaminar localizations of L5 PCs participating in CC connections between M2 and M1. L5 PCs are composed of 2 major subtypes: CPn cells that project to subcerebral regions, including the pontine nuclei, and COM cells that project to the contralateral telencephalon (Molnár and Cheung 2006; Morishima and Kawaguchi 2006; Molyneaux et al. 2007; Otsuka and Kawaguchi 2011). CPn cells can be identified based on their selective expression of the transcription factor Ctip2 (Arlotta et al. 2005). We found that CPn cells are more involved in CC projections from M2 to M1 than that in projections from M1 to M2, using Ctip2 immunoreactivity for a CPn cell marker. Ctip2-positive cells that were positive for the pan-neuronal marker NeuN were much more abundant in deep layers than that in superficial layers. In L1 and L2/3, Ctip2-positive cells were mostly GABAergic cells, whereas, in L5 and L6, they were mostly excitatory cells (but a few were GABAergic cells; Table 4). Ctip2-positive cells accounted for nearly half of the excitatory cells present in L5a, and about 80% of excitatory cells in L5b. In L6, almost all of the excitatory cells expressed Ctip2 (Table 5). Among the L5 excitatory cells, Ctip2 was generally expressed in CPn cells, but only rarely was present in COM cells (Fig. 6A and Table 5). Therefore, we used Ctip2 expression to identify L5 CPn cells participating in CC connections between M2 and M1.

Table 4

Ctip2 expression selectively in GABAergic cells in superficial layers, but mostly in excitatory cells in deep layers

Layer GABAergic cells/Ctip2-positive cells
 
M2 M1 
L1 97.2% (70/72) 98.8% (80/81) 
L2/3 88.7% (211/238) 91.2% (228/250) 
L5 1.9% (29/1518) 2.7% (35/1303) 
L6 5.1% (227/4471) 17.2% (828/4453) 
Layer GABAergic cells/Ctip2-positive cells
 
M2 M1 
L1 97.2% (70/72) 98.8% (80/81) 
L2/3 88.7% (211/238) 91.2% (228/250) 
L5 1.9% (29/1518) 2.7% (35/1303) 
L6 5.1% (227/4471) 17.2% (828/4453) 

Note: Obtained from 2 VGAT-Venus rats.

Ctip2: chicken ovalbumin upstream promoter transcription factor-interacting protein 2; GABA: gamma-aminobutyric acid; L: layer; (n1/n2): n1, number of GABAergic cells positive for Ctip2; n2: total number of Ctip2-positive cells.

Table 5

Ctip2 expression in L5 CPn cells, but not in L5 COM cells

Layer Excitatory cellsa
 
CPn cellsb
 
COM cellsb
 
Ctip2-positive cells
 
L5a 44.6% (2829/6337) 95.5% (234/245) 0.7% (1/153) 
L5b 81.3% (7395/9095) 95.5% (421/441) 4.7% (6/129) 
L6 99.2% (11178/11265) – – 
Layer Excitatory cellsa
 
CPn cellsb
 
COM cellsb
 
Ctip2-positive cells
 
L5a 44.6% (2829/6337) 95.5% (234/245) 0.7% (1/153) 
L5b 81.3% (7395/9095) 95.5% (421/441) 4.7% (6/129) 
L6 99.2% (11178/11265) – – 

Note: Cell numbers are in parenthesis.

CPn: corticopontine; COM: commissural.

aObtained from 2 VGAT-Venus rats.

bObtained from 2 rats.

Figure 6.

Collateralization of Ctip2-positive CPn cells to the adjacent motor area. (A) Ctip2 was expressed in CPn cells, but not in COM cells. CPn and COM cells were retrogradely labeled with a Fast Blue injection into the ipsilateral pons (arrowheads) and a CTB555 injection into contralateral M2 (arrows), respectively. Note that CPn cells were immunopositive for Ctip2 (arrowheads), whereas COM cells were immunonegative for Ctip2 (asterisks). (B) The percentages of Ctip2-positive cells among L5 CC cells, which were labeled retrogradely by CTB555. Upper left, the proportion of L5a and L5b CC cells in M2 (upper panel; 4 rats) and M1 (lower panel; 3 rats). CC cells were much more abundant in L5a of M2 and at similar levels in L5a and L5b of M1. Upper right, the proportion of Ctip2-positive cells among L5 CC cells. Ctip2-positive cells were very scarce in L5b and were mainly in L5a of both M2 and M1. L5a CC cells exhibited more positivity for Ctip2 in M2 than that in M1 (P < 0.01; chi-squared test). Lower panel, M2 CC cells positive (arrowheads) and negative (arrows in left and asterisks in right) for Ctip2.

Figure 6.

Collateralization of Ctip2-positive CPn cells to the adjacent motor area. (A) Ctip2 was expressed in CPn cells, but not in COM cells. CPn and COM cells were retrogradely labeled with a Fast Blue injection into the ipsilateral pons (arrowheads) and a CTB555 injection into contralateral M2 (arrows), respectively. Note that CPn cells were immunopositive for Ctip2 (arrowheads), whereas COM cells were immunonegative for Ctip2 (asterisks). (B) The percentages of Ctip2-positive cells among L5 CC cells, which were labeled retrogradely by CTB555. Upper left, the proportion of L5a and L5b CC cells in M2 (upper panel; 4 rats) and M1 (lower panel; 3 rats). CC cells were much more abundant in L5a of M2 and at similar levels in L5a and L5b of M1. Upper right, the proportion of Ctip2-positive cells among L5 CC cells. Ctip2-positive cells were very scarce in L5b and were mainly in L5a of both M2 and M1. L5a CC cells exhibited more positivity for Ctip2 in M2 than that in M1 (P < 0.01; chi-squared test). Lower panel, M2 CC cells positive (arrowheads) and negative (arrows in left and asterisks in right) for Ctip2.

We found that both Ctip2-positive and Ctip2-negative cells were present among L5 CC cells in M1 and M2; however, the proportion of Ctip2-positive cells was higher among M2 CC cells (Fig. 6B; L5a: 56.3%, 679 cells and L5b: 9.5%, 42 cells, 4 rats) than among M1 CC cells (Fig. 6B; L5a: 33.8%, 225 cells and L5b: 4.8%, 227 cells, 3 rats; P < 0.01, chi-squared test). Among all of the L5 CC cells, Ctip2 expression was found mostly in L5a of both M2 and M1 (Fig. 6B; M2: 53% in L5a and 0.6% in L5b; M1: 16.8% in L5a and 2.4% in L5b). Finally, we found that CPn cells having axonal projections to the spinal cord (L5b CSp cells) did not participate in CC connections in either direction. No double-labeled L5b cells were identified following retrograde double labeling from the upper cervical spinal cord and the other motor area (2 rats; data not shown). This observation is consistent with our result that L5b CC cells did not express Ctip2 (Fig. 6B).

Collateralization of M2 L5 COM Cells to M1

We found that multiple subtypes of M2 L5 COM cells send axons to M1, based on the firing heterogeneity of M2 L5 cells projecting to both M1 and contralateral M2. L5 COM cells in M2 exhibited the following heterogeneous firing properties: fast adapting (FA) and slowly adapting (SA) firing with or without an initial doublet (SA-d) in response to suprathreshold depolarizing current pulses (Fig. 7A; Otsuka and Kawaguchi 2011). Although firing pattern differences are preserved during maturation, adaptation is less pronounced in older animals relative to those that are 3 weeks old (Suter et al. 2013). To examine which types of COM cells are present among M2 CC cells, we recorded from L5 PCs in M2 that were double labeled from ipsilateral M1 and contralateral M2 (Fig. 7B). We calculated the instantaneous firing frequencies of the first, second, and seventh interspike intervals (f1, f2, and f7) during trains of action potentials generated by current pulse injection (Fig. 7C; amplitude, 500 pA; duration, 1 s). From the adaptation index (f7/f2) and the first interspike frequency (f1), we quantitatively identified 3 firing types of COM cells distinguishable by these parameters (Fig. 7C; adaptation index: FA, 0.3 ± 0.11; SA, 0.88 ± 0.09; SA-d, 0.89 ± 0.06; f1: FA, 29.9 ± 7.4; SA, 27.9 ± 4; SA-d, 118.6 ± 10.3 Hz). COM cells projecting to M1 exhibited all 3 firing types (Fig. 7D; FA, 51%; SA, 31%; and SA-d, 18%; 39 cells in total). We have previously found that SA- and FA-type COM cells exhibit different remote projection patterns (Otsuka and Kawaguchi 2011; Hirai et al. 2012): The FA-type COM cells project to the contralateral striatum and to distant cortical areas, such as the posterior parietal and perirhinal cortices. As described above, COM cells were negative for Ctip2. Thus, Ctip2-negative CC cells projecting from M2 to M1 were heterogeneous in their extracortical projections and firing patterns in response to somatic current injection.

Figure 7.

Collateralization of multiple L5 COM cell subtypes in M2 to M1. (A) Firing subtypes of COM cells projecting to M1, which were labeled retrogradely by CTB555 and Alexa Fluor 488-conjugated CTB: FA: fast adapting; SA: slowly adapting; and SA-d: slowly adapting with an initial doublet firing, in response to a somatic depolarizing current pulse. (B) Injection of different fluorescent retrograde tracers into contralateral M2 (CTB555 or red beads) and ipsilateral M1 (Alexa Fluor 488- or fluorescein isothiocyanate-conjugated CTB) to identify L5 COM cells projecting to M1. (C) Relationship between the adaptation index (f7/f2) and first spike frequency (f1) during current pulse injection (500 pA, 1 s) (n = 20, 12, and 7 cells, for FA, SA, and SA-d types). (D) Proportion of firing subtypes among COM cells projecting to M1.

Figure 7.

Collateralization of multiple L5 COM cell subtypes in M2 to M1. (A) Firing subtypes of COM cells projecting to M1, which were labeled retrogradely by CTB555 and Alexa Fluor 488-conjugated CTB: FA: fast adapting; SA: slowly adapting; and SA-d: slowly adapting with an initial doublet firing, in response to a somatic depolarizing current pulse. (B) Injection of different fluorescent retrograde tracers into contralateral M2 (CTB555 or red beads) and ipsilateral M1 (Alexa Fluor 488- or fluorescein isothiocyanate-conjugated CTB) to identify L5 COM cells projecting to M1. (C) Relationship between the adaptation index (f7/f2) and first spike frequency (f1) during current pulse injection (500 pA, 1 s) (n = 20, 12, and 7 cells, for FA, SA, and SA-d types). (D) Proportion of firing subtypes among COM cells projecting to M1.

M1 L1a Innervation by M2 L5a CPn Cells

By localized iontophoretic delivery of a retrograde tracer (CTB555) into M1, we found that L5 cells in M2 preferentially innervated L1 of M1, while L2/3 cells in M2 targeted L2/3 of M1 (Fig. 8A; 2 injection cases shown, #4 and #8). In frontal motor areas, thalamocortical afferents are also segregated based on thalamic nuclei of origin. Fibers from the ventral anterior/ventromedial (VA/VM) nuclei preferentially target L1 (Kuramoto et al. 2009; Rubio-Garrido et al. 2009). Injections delivering a high proportion of retrograde tracer into L1a of M1 produced more retrogradely labeled thalamic cells in the VA/VM nuclei than that in the VL nucleus (Fig. 8A), confirming localization of tracers and their uptake there. For the CC projections, we found that superficial injections that deposited more tracer in L1 than in L2/3 of M1 produced a larger proportion of retrogradely labeled L5 cells in M2 (Fig. 8B; 137 ± 54.6 labeled cells in 8 injections [#1–#8]; proportions of labeled L5 cells in M2 vs. those of L1a or total L1 tracer deposits in M1: Correlation coefficient [c.c.] = 0.78 or 0.85, P < 0.05 or <0.01). Conversely, when tracer deposition was dense within L2/3b of M1, more labeled cells were observed in the VL thalamic nucleus than that in the VA/VM nuclei, while labeled CC cells in M2 were preferentially found in L2/3 (Fig. 8B; proportions of labeled L2/3 cells in M2 vs. those of L2/3b or total L2/3 tracer deposits in M1: c.c. = 0.9 or 0.85, P < 0.01). Finally, we found that, within M2 L5a, Ctip2-positive CPn cells provided more innervation of M1 L1a than did Ctip2-negative COM cells (Figs 6A and 8C). When M1 label was present in L1a, Ctip2-positive CC cells were labeled in L5 of M2 (Fig. 8C, left; 2 rats). In contrast, when L1b, but not L1a, of M1 was labeled with tracer, labeled CC cells in L5 of M2 were mostly negative for Ctip2 (Fig. 8C, right; 2 rats).

Figure 8.

Relationship of the laminar localizations of M2 CC cell bodies and their terminations in M1. (A) The distribution of M2 CC cells labeled by the iontophoretic localized application of CTB555 (2 cases shown). Upper panel, injections in M1. Solid gray circles, injection sites, judging from localized intense fluorescence. Broken gray circles, strong fluorescent region because of tracer incorporation into the neuropil. L1a and L2/3b received denser thalamocortical inputs (stronger immunofluorescence for VGluT2) than did L1b and L2/3a, respectively. Middle panel, retrogradely labeled thalamocortical cells. Tracer deposition in L1a and L2/3 resulted in labeling of the VA/VM nuclei and the ventrolateral (VL) nucleus, respectively. Lower panel, retrogradely labeled M2 cells (black dots). (B) The proportions of L2/3 and L5 labeled cells in M2 were dependent on the laminar localizations of tracer deposits in M1. #1–8, rat number (8 rats). Upper, laminar proportions of M1 injection area. L1a: filled black bar; L1b: open black bar; L2/3a: open gray bar; and L2/3b: filled gray bar. Lower panel, L2/3 (open black bar) and L5 (filled black bar) CC cell distributions in M2 for the individual injections described above. (n), total number of labeled cells from 3 different sections. Note the correlations between the deposit area of M1 L1a and the proportion of labeled M2 L5 cells (c.c. = 0.78) and between the deposit area of M1 L2/3b and the proportion of labeled M2 L2/3 cells (c.c. = 0.9). (C) L5a Ctip2-positive CC cells projecting to M1, which were labeled retrogradely by CTB555. Left panel, M1 injection site and labeled M2 cells of case 4 depicted in (B). Arrowheads, Ctip2-positive cells. Note that the injection site included L1a. Right panel, case 5 depicted in (B). Note the presence of few CC cells expressing Ctip2 and the deposition of the tracer mainly in L1b and L2/3, but not in L1a.

Figure 8.

Relationship of the laminar localizations of M2 CC cell bodies and their terminations in M1. (A) The distribution of M2 CC cells labeled by the iontophoretic localized application of CTB555 (2 cases shown). Upper panel, injections in M1. Solid gray circles, injection sites, judging from localized intense fluorescence. Broken gray circles, strong fluorescent region because of tracer incorporation into the neuropil. L1a and L2/3b received denser thalamocortical inputs (stronger immunofluorescence for VGluT2) than did L1b and L2/3a, respectively. Middle panel, retrogradely labeled thalamocortical cells. Tracer deposition in L1a and L2/3 resulted in labeling of the VA/VM nuclei and the ventrolateral (VL) nucleus, respectively. Lower panel, retrogradely labeled M2 cells (black dots). (B) The proportions of L2/3 and L5 labeled cells in M2 were dependent on the laminar localizations of tracer deposits in M1. #1–8, rat number (8 rats). Upper, laminar proportions of M1 injection area. L1a: filled black bar; L1b: open black bar; L2/3a: open gray bar; and L2/3b: filled gray bar. Lower panel, L2/3 (open black bar) and L5 (filled black bar) CC cell distributions in M2 for the individual injections described above. (n), total number of labeled cells from 3 different sections. Note the correlations between the deposit area of M1 L1a and the proportion of labeled M2 L5 cells (c.c. = 0.78) and between the deposit area of M1 L2/3b and the proportion of labeled M2 L2/3 cells (c.c. = 0.9). (C) L5a Ctip2-positive CC cells projecting to M1, which were labeled retrogradely by CTB555. Left panel, M1 injection site and labeled M2 cells of case 4 depicted in (B). Arrowheads, Ctip2-positive cells. Note that the injection site included L1a. Right panel, case 5 depicted in (B). Note the presence of few CC cells expressing Ctip2 and the deposition of the tracer mainly in L1b and L2/3, but not in L1a.

L5 CPn Cell Diversity in M2

As described above, some CPn cells, especially in L5a, have collaterals to the adjacent motor area; however, CSp cells, which are a subgroup of L5b CPn cells, did not have collaterals to the adjacent motor area. In addition to the spinal cord, some CPn cells send axons to the thalamus (CTh cells; Arlotta et al. 2005; Molyneaux et al. 2007; Hirai et al. 2012). Using retrograde double labeling, we confirmed that the pattern of spinal cord innervation was different among L5a and L5b CTh cells. L5 CTh cells projecting to the ventral thalamic nuclei were more abundant in L5a than that in L5b (Fig. 9A; 3 rats). L5a CTh cells were independent of CSp cells, but some L5b CTh cells also projected to the spinal cord (Fig. 9B; 3 rats). These differences in projections suggest a functional segregation between L5a and L5b CTh cells.

Figure 9.

Comparative laminar localizations of CPn cell subtypes: CTh and CSp cells. (A) Distributions of CTh cells projecting to the ventral thalamic nuclei (labeled by CTB555; red dots) and CSp cells (labeled by Fast Blue; blue dots) in the same animal. CTh cells were more abundant in L5a and L6 and much less abundant in L5b, whereas CSp cells were found exclusively in L5b. Oblique section, 20 μm thick. (B) L5b CTh cell projection to the spinal cord. Some L5b cells were double labeled from the spinal cord by Fast Blue and ventral thalamic nuclei by CTB555 (white arrowheads). Oblique section, 20 μm thick. (C) Sublaminar localization schema of CPn cell subtypes and M1-projecting CC cells in L5 of M2.

Figure 9.

Comparative laminar localizations of CPn cell subtypes: CTh and CSp cells. (A) Distributions of CTh cells projecting to the ventral thalamic nuclei (labeled by CTB555; red dots) and CSp cells (labeled by Fast Blue; blue dots) in the same animal. CTh cells were more abundant in L5a and L6 and much less abundant in L5b, whereas CSp cells were found exclusively in L5b. Oblique section, 20 μm thick. (B) L5b CTh cell projection to the spinal cord. Some L5b cells were double labeled from the spinal cord by Fast Blue and ventral thalamic nuclei by CTB555 (white arrowheads). Oblique section, 20 μm thick. (C) Sublaminar localization schema of CPn cell subtypes and M1-projecting CC cells in L5 of M2.

M1 Innervation by M2 L5a CTh Cells

As described above, M2 L5a contained Ctip2-positive cells projecting to M1 as well as Ctip2-positive CTh cells (Fig. 9C). Using simultaneous injections of 2 tracers into the thalamus and one motor area, we found that the participation of CTh cells in CC connectivity between M1 and M2 is unidirectional. L5a CTh cells in M2 sent axon collaterals to M1, but L5a CTh cells in M1 did not project to M2. M2 L5a CTh cells labeled from the ventral thalamic nuclei were partially double labeled from M1 (Fig. 10A; double labeling in M2: 12.7 ± 3.3% in M1-projecting CC cells and 27.5 ± 17.1% in CTh cells, 3 rats). The double-labeled cells were tufted PCs of the SA type with initial doublet firing (Fig. 10B; n = 12), which is a typical characteristic of L5a CPn cells (Morishima et al. 2011; Hirai et al. 2012).

Figure 10.

Collateralization of M2 L5a CTh cells to M1. (A) Partial double labeling between L5a CTh cells and M1-projecting CC cells in M2. Upper-left panel, cells that were double labeled from M1 by Fast Blue and ventral thalamic nuclei by CTB555 (white arrowheads). Upper-right panel, proportion of double-labeled cells (3 rats). Bottom-left panel, thalamic injection in the ventral thalamic nuclei. Bottom-right panel, proportions of M2 cells projecting to the thalamus or to M1 among all labeled cells in L5a. Sagittal section, 20 μm thick. (B) Dendritic reconstruction and firing pattern of a CTh cell also projecting to M1. Asterisk, initial doublet firing. (C) Few M1 double-labeled cells from the thalamus and M2. Upper-left panel, M2-projecting CC cells, labeled by Fast Blue, were not double labeled from ventral thalamic nuclei by CTB555 (white arrows). Upper-right panel, proportion of double-labeled cells (3 rats). Bottom-left panel, thalamic injection in the ventral thalamic nuclei. Bottom-right panel, proportions of M1 cells projecting to the thalamus or to M2 among all labeled cells in L5a. Sagittal section, 20 μm thick.

Figure 10.

Collateralization of M2 L5a CTh cells to M1. (A) Partial double labeling between L5a CTh cells and M1-projecting CC cells in M2. Upper-left panel, cells that were double labeled from M1 by Fast Blue and ventral thalamic nuclei by CTB555 (white arrowheads). Upper-right panel, proportion of double-labeled cells (3 rats). Bottom-left panel, thalamic injection in the ventral thalamic nuclei. Bottom-right panel, proportions of M2 cells projecting to the thalamus or to M1 among all labeled cells in L5a. Sagittal section, 20 μm thick. (B) Dendritic reconstruction and firing pattern of a CTh cell also projecting to M1. Asterisk, initial doublet firing. (C) Few M1 double-labeled cells from the thalamus and M2. Upper-left panel, M2-projecting CC cells, labeled by Fast Blue, were not double labeled from ventral thalamic nuclei by CTB555 (white arrows). Upper-right panel, proportion of double-labeled cells (3 rats). Bottom-left panel, thalamic injection in the ventral thalamic nuclei. Bottom-right panel, proportions of M1 cells projecting to the thalamus or to M2 among all labeled cells in L5a. Sagittal section, 20 μm thick.

In contrast, a few M1 CTh cells projected to M2 (Fig. 10C; double labeling in M1: 1.5 ± 0.5% in M2-projecting CC cells and 2.9 ± 1.6% in CTh cells; 3 rats), a finding consistent with our observation of lower Ctip2 expression among L5a CC cells in M1 than among those in M2 (Fig. 6B). In these experiments, the percentages of labeled CTh and CC cells in L5a were similar between the 2 areas (Fig. 10A,C, bottom; M2, 67.9% of CC cells and 32.1% of CTh cells among all labeled cells in L5a, n = 1272 for CC, 601 for CTh, SD = 9.1%; M1, 60.8% of CC cells and 39.2% of CTh cells among all labeled cells in L5a, n = 952 for CC, 614 for CTh, SD = 6.8%; 3 rats; P = 0.37, chi-squared test), because both areas densely innervated ventral thalamic nuclei (M2, Ushimaru et al. 2012; M1, data not shown). Since M1 L1a was innervated by M2 L5a CPn cells, these results indicated that L5a CTh cells may connect adjacent frontal motor-related areas through L1a innervations in a direction-sensitive manner.

Discussion

M2, a Higher-Order Motor-Related Area, in the Rat Frontal Cortex

The dorsal frontal cortex in rodents is divided into several cytoarchitectonic areas that include the medial agranular, lateral agranular, and cingulate areas (Donoghue and Wise 1982). These areas contain regions where ICMS can induce movements at low current intensities (<60 μA), and these regions were functionally defined as M1 (Neafsey et al. 1986; Brecht et al. 2004). The rostral region of the medial agranular area, however, was excluded from M1, because movements are not induced there at low current intensities; this region was, therefore, termed the M2 area (Brecht et al. 2004). We found that the rostral border of M1 could be reliably identified based on differences in NF-H immunoreactivity between this area and the rostral frontal area, where ICMS rarely evoked movements, but which contained CSp cells. Thus, we defined this rostral frontal area as M2, which is a motor-related area distinct from M1. M2 was also distinguished from the lateral orbital area, which has few projections to the spinal cord, based on NF-H immunoreactivity patterns.

In M1, the regional representation of FL and HL movements was correlated with direct CSp projections to the cervical and lumbar spinal cords, respectively. In M2, however, CSp cells projected only to the cervical spinal cord; none projected to the lumbar spinal cord. In contrast, the cellular physiology, as indicated by firing patters in response to depolarization, was similar in M1 and M2 CSp cells projecting to the cervical spinal cord. Together with the little movement that was induced by ICMS in M2, these results suggest that M2 innervates target within the cervical spinal cord that are distinct from those innervated by CSp cells in M1.

Cells in M2 projected to the cervical spinal cord, but not to the lumbar spinal cord. In contrast, M2 projections to M1 innervated both the FL region that projects to the cervical spinal cord and the HL region that projects to the lumbar spinal cord. These M2 projections to M1 subareas were segregated, with projections to the FL region originating in lateral M2 and projections to the HL region originating more medially. These findings suggest that M2 projections are spatially differentiated for independent control of multiple movement representations in M1, but that additional M2 projections to the cervical spinal cord contribute in integration of both FL and HL movements.

Axon Collateralization Pattern of CPn Cells Relative to the Sublaminar Structure

CPn cells are distributed exclusively within L5, but their dendritic morphologies and intracortical connections are differentiated further according to their somatic position within 2 subdivisions of L5: L5a and L5b, which have different thalamocortical input densities (Otsuka and Kawaguchi 2008; Morishima et al. 2011; Hirai et al. 2012). In this study, we found that CPn cells in L5a projected to the thalamus, but not to the spinal cord, whereas those in L5b projected to the spinal cord, with only a few projecting to the thalamus. In addition, we found that CSp cells can be differentiated according to their depth within L5b: more elaborated basal dendrites and less developed apical oblique branches were more common in deeper-lying CSp cells. These findings strongly suggest that CPn cells are differentiated according to their distribution between and within L5a and L5b, having segregated projection targets and dendritic arborizations, which may correspond to their distinct input patterns (Hooks et al. 2013).

L5a CPn cells in M2 projected to M1, but those in M1 did not abundantly project to M2. However, L5b CPn cells, including CSp cells, did not participate in CC connections in either direction. It has been shown that CPn cells issue axon collaterals to various regions (O'Leary and Koester 1993; Veinante and Deschênes 2003; Molyneaux et al. 2007). Here, we demonstrate that the axon collateralization of CPn cells is related to their distribution within the L5 sublaminar structure (Figs 9C and 11).

Figure 11.

CC cells connecting M2 and M1: their direction-dependent sublaminar and neuron-type specificity. The lateral and medial subregions of M2 preferentially innervated the FL and HL regions situated in rostral and caudal parts of M1, respectively. Thus, somatotopy was topographically preserved in CC connections between M2 and M1. Both M2 and M1 contained CSp cells in L5b; however, the somatotopy was topographically preserved in CSp projections from M1, but not from M2. CSp cells were independent of the CC connections. In M1, M2-projecting cells were distributed in L2/3a,b, L5a,b, and L6b, whereas in M2, M1-projecting cells were restricted to the L2/3b, L5a, and L6b. The sublaminar localization pattern of CC cells between the 2 areas was most different in L5. L5 CC cells in both areas contained Ctip2-positive (CPn) and Ctip2-negative (COM) cells. In M2, L5 COM cells included FA, SA, and SA-d firing types. FA-type COM cells project to distant cortical areas such as posterior parietal and perirhinal cortices (Otsuka and Kawaguchi 2011; Hirai et al. 2012). Ctip2-positive CC cells were found in L5a, and very few of these cells were found in L5b. In addition, Ctip2-positive CC cells were more abundant in M2 than that in M1. Furthermore, L5a Ctip2-positive CTh cells, projecting to ventral thalamic nuclei, in M2 innervated M1, especially L1a, to a greater extent than those in M1 innervated M2.

Figure 11.

CC cells connecting M2 and M1: their direction-dependent sublaminar and neuron-type specificity. The lateral and medial subregions of M2 preferentially innervated the FL and HL regions situated in rostral and caudal parts of M1, respectively. Thus, somatotopy was topographically preserved in CC connections between M2 and M1. Both M2 and M1 contained CSp cells in L5b; however, the somatotopy was topographically preserved in CSp projections from M1, but not from M2. CSp cells were independent of the CC connections. In M1, M2-projecting cells were distributed in L2/3a,b, L5a,b, and L6b, whereas in M2, M1-projecting cells were restricted to the L2/3b, L5a, and L6b. The sublaminar localization pattern of CC cells between the 2 areas was most different in L5. L5 CC cells in both areas contained Ctip2-positive (CPn) and Ctip2-negative (COM) cells. In M2, L5 COM cells included FA, SA, and SA-d firing types. FA-type COM cells project to distant cortical areas such as posterior parietal and perirhinal cortices (Otsuka and Kawaguchi 2011; Hirai et al. 2012). Ctip2-positive CC cells were found in L5a, and very few of these cells were found in L5b. In addition, Ctip2-positive CC cells were more abundant in M2 than that in M1. Furthermore, L5a Ctip2-positive CTh cells, projecting to ventral thalamic nuclei, in M2 innervated M1, especially L1a, to a greater extent than those in M1 innervated M2.

CC Connection Selectivity in the Sublaminar Innervation and Participation of PC Subtypes

We found that axons from M2 preferentially innervate L1 of M1, whereas those from M1 preferentially innervate L2/3 of M2. Among the sensory cortices of primates and rodents, feedback projections from higher-order areas terminate preferentially in L1 of the primary sensory area, whereas feedforward projections from the primary area do so in the middle layers of higher-order areas (Felleman and Van Essen 1991; Cauller et al. 1998; Dong et al. 2004). Thus, by analogy, M1 may be thought to provide feedforward projections to M2, whereas M2 tends to send feedback projections to M1.

In rat frontal motor-related areas, L1 is innervated by VA/VM thalamic nuclei, which relay basal ganglia (BG) output, whereas L2/3 is innervated by the VL thalamic nucleus, which relays cerebellar output (Kuramoto et al. 2009; Rubio-Garrido et al. 2009). Thus, L1a of M1 receives simultaneous feedback excitation from M2 and afferents from BG-related thalamic nuclei, whereas L2/3b of M2 receives feedforward input from M1 and afferents from the cerebellum (Cb)-related thalamic nucleus. These observations suggested that Cb output may boost the transmission in the feedforward direction (M1 to M2) in frontal motor-related areas, whereas BG output may boost the transmission in the feedback direction (M2 to M1). Therefore, the cortico(Cb)thalamocortical loop, which is important for motor planning and error correction, may be more involved in primary to higher-order information flow. Conversely, the cortico(BG)thalamocortical loop, which is important for action selection and evaluation, may be more involved in top-down feedback from higher-order to primary motor areas. The interaction of these 2 corticothalamocortical loop circuits and CC hierarchical connections between frontal motor-related areas is considered essential for procedural learning and memory formation (Hikosaka et al. 2002; Ito 2011).

Differences in the laminar distributions of CC cells were also observed in M2 and M1, particularly in L5a and L5b (Fig. 11). We found that L5a CC cells were more abundant in M2 than that in M1, whereas L5b CC cells were more abundant in M1 than that in M2. Ctip2-negative CC cells (COM type) were found in L5a of M2 and in L5a and L5b of M1. L5a COM cells that projected from M2 to ipsilateral M1 exhibited both FA and SA firing types. L5a COM cells are known to project to distant cortical areas, such as the posterior parietal and perirhinal cortices, as well as the contralateral striatum (crossed-corticostriatal cells) (Otsuka and Kawaguchi 2011; Hirai et al. 2012). Thus, Ctip2-negative CC cells projecting from M2 to M1 contain 2 subtypes of COM cells that differ in their firing patterns and their distant non-M1 cortical targets.

Among the CC cells connecting M2 and M1, Ctip2-positive CC cells (CPn type) were much more abundant in L5a than that in L5b, and in M2 compared with M1. L5a CTh cells, which were an L5a CPn subtype, participated much more in CC projections from M2 to M1 than that from M1 to M2. In addition, M2 L5a Ctip2-positive cells (i.e., CPn cells, including CTh cells) preferentially innervated L1a of M1. Thus, CC fibers that were distributed in L1a of M1 conceivably originate from L5a CTh cells in M2 (Fig. 11). However, L5a COM type CC cells may prefer to innervate L1b and L2/3a of M1 (Fig. 8).

L5a CTh cells in the frontal cortex send axons to the VA/VM thalamic nuclei, whereas they innervate L5 and L1a locally in the cortex (Hirai et al. 2012). On the other hand, single VA/VM thalamic cells innervate L1 of both M2 and M1 simultaneously (Kuramoto et al. 2009). Our data indicate that M2 cells, including L5a CTh cells, densely innervate L1a of M1. Taken together, these findings suggest that L5a CTh cells in M2 provide direct excitatory input to L1a of both M2 and M1, while also recruiting thalamocortical inputs to L1a of both M2 and M1 through excitation of the cortico(BG)thalamocortical pathway (Fig. 11). Because L5a CTh cells in M1 are unlikely to participate in CC projections to M2, M1 may not provide conjunctive CC and thalamocortical inputs to L1a of M2. These observations suggested differential L1a excitation between adjacent frontal motor-related areas, despite the fact that thalamocortical cells innervate L1a of both primary and higher-order areas simultaneously.

In a previous study involving paired recordings, we found that L5 PCs, which innervate L1b and L2/3a, but not L1a, induce excitatory postsynaptic currents (EPSCs) in L2/3a PCs. However, L5 CTh cells, which innervate L1a, do not induce EPSCs in L2/3a PCs, although the dendritic tufts of L2/3a PCs are well distributed in both L1a and L1b (Hirai et al. 2012). It is conceivable that individual excitatory synapses on pyramidal tuft dendrites in L1a do not effectively generate somatic depolarization on their own because of limitations due to electrotonic distance, but that conjunctive activation of several L1 synapses may effectively induce large local events that are transmitted to the soma of PCs (Larkum et al. 2009). Consequently, the simultaneous firing of L5a CTh cells in M2 and BG-related thalamic cells, both of which innervate L1a of M1, may excite PCs in M1, including CSp cells having well-developed dendritic tufts. Thus, the coactivation of CC and thalamocortical inputs by activity of L5a CTh cells in higher-order areas may be critical for information transfer from higher- to lower-order frontal motor-related areas.

Funding

This work was supported by the Japan Science and Technology Agency (JST), Core Research for Evolutional Science and Technology (CREST), and Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT; KAKENHI No. 21240030 to Y.K.).

Notes

We thank Mika Watanabe and Noboru Yamaguchi for histological assistance and Allan T. Gulledge, Yasuharu Hirai, and Yumiko Hatanaka for comments on the manuscript. Conflict of Interest: None declared.

References

Arlotta
P
Molyneaux
BJ
Chen
J
Inoue
J
Kominami
R
Macklis
JD
Neuronal subtype-specific genes that control corticospinal motor neuron development in vivo
Neuron
 , 
2005
, vol. 
45
 (pg. 
207
-
221
)
Brecht
M
Krauss
A
Muhammad
S
Sinai-Esfahani
L
Bellanca
S
Margrie
TW
Organization of rat vibrissa motor cortex and adjacent areas according to cytoarchitectonics, microstimulation, and intracellular stimulation of identified cells
J Comp Neurol
 , 
2004
, vol. 
479
 (pg. 
360
-
373
)
Brown
SP
Hestrin
S
Intracortical circuits of pyramidal neurons reflect their long-range axonal targets
Nature
 , 
2009
, vol. 
457
 (pg. 
1133
-
1136
)
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
)
Dong
H
Wang
Q
Valkova
K
Gonchar
Y
Burkhalter
A
Experience-dependent development of feedforward and feedback circuits between lower and higher areas of mouse visual cortex
Vision Res
 , 
2004
, vol. 
44
 (pg. 
3389
-
3400
)
Donoghue
JP
Wise
SP
The motor cortex of the rat: cytoarchitecture and microstimulation mapping
J Comp Neurol
 , 
1982
, vol. 
212
 (pg. 
76
-
88
)
Douglas
RJ
Martin
KA
Neuronal circuits of the neocortex
Annu Rev Neurosci
 , 
2004
, vol. 
27
 (pg. 
419
-
451
)
Dum
RP
Strick
PL
Motor areas in the frontal lobe of the primate
Physiol Behav
 , 
2002
, vol. 
77
 (pg. 
677
-
682
)
Dum
RP
Strick
PL
The origin of corticospinal projections from the premotor areas in the frontal lobe
J Neurosci
 , 
1991
, vol. 
11
 (pg. 
667
-
689
)
Felleman
DJ
Van Essen
DC
Distributed hierarchical processing in the primate cerebral cortex
Cereb Cortex
 , 
1991
, vol. 
1
 (pg. 
1
-
47
)
Hattox
AM
Nelson
SB
Layer V neurons in mouse cortex projecting to different targets have distinct physiological properties
J Neurophysiol
 , 
2007
, vol. 
98
 (pg. 
3330
-
3340
)
Hikosaka
O
Nakamura
K
Sakai
K
Nakahara
H
Central mechanisms of motor skill learning
Curr Opin Neurobiol
 , 
2002
, vol. 
12
 (pg. 
217
-
222
)
Hirai
Y
Morishima
M
Karube
F
Kawaguchi
Y
Specialized cortical subnetworks differentially connect frontal cortex to parahippocampal areas
J Neurosci
 , 
2012
, vol. 
32
 (pg. 
1898
-
1913
)
Hooks
BM
Mao
T
Gutnisky
DA
Yamawaki
N
Svoboda
K
Shepherd
GM
Organization of cortical and thalamic input to pyramidal neurons in mouse motor cortex
J Neurosci
 , 
2013
, vol. 
33
 (pg. 
748
-
760
)
Huntley
GW
Correlation between patterns of horizontal connectivity and the extend of short-term representational plasticity in rat motor cortex
Cereb Cortex
 , 
1997
, vol. 
7
 (pg. 
143
-
156
)
Ito
M
The cerebellum: brain for an implicit self
 , 
2011
New Jersey
FT Press
Jones
EG
The thalamus.
 , 
2007
2nd ed
Cambridge (UK)
Cambridge University Press
Kaneko
T
Fujiyama
F
Hioki
H
Immunohistochemical localization of candidates for vesicular glutamate transporters in the rat brain
J Comp Neurol
 , 
2002
, vol. 
444
 (pg. 
39
-
62
)
Kuramoto
E
Furuta
T
Nakamura
KC
Unzai
T
Hioki
H
Kaneko
T
Two types of thalamocortical projections from the motor thalamic nuclei of the rat: a single neuron-tracing study using viral vectors
Cereb Cortex
 , 
2009
, vol. 
19
 (pg. 
2065
-
2077
)
Larkum
ME
Nevian
T
Sandler
M
Polsky
A
Schiller
J
Synaptic integration in tuft dendrites of layer 5 pyramidal neurons: a new unifying principle
Science
 , 
2009
, vol. 
325
 (pg. 
756
-
760
)
Lemon
RN
Descending pathways in motor control
Annu Rev Neurosci
 , 
2008
, vol. 
31
 (pg. 
195
-
218
)
Miller
MN
Okaty
BW
Nelson
SB
Region-specific spike-frequency acceleration in layer 5 pyramidal neurons mediated by Kv1 subunits
J Neurosci
 , 
2008
, vol. 
28
 (pg. 
13716
-
13726
)
Miller
MW
The origin of corticospinal projection neurons in rat
Exp Brain Res
 , 
1987
, vol. 
67
 (pg. 
339
-
351
)
Miyashita
E
Keller
A
Asanuma
H
Input-output organization of the rat vibrissal motor cortex
Exp Brain Res
 , 
1994
, vol. 
99
 (pg. 
223
-
232
)
Molnár
Z
Cheung
AF
Towards the classification of subpopulations of layer V pyramidal projection neurons
Neurosci Res
 , 
2006
, vol. 
55
 (pg. 
105
-
115
)
Molyneaux
BJ
Arlotta
P
Menezes
JR
Macklis
JD
Neuronal subtype specification in the cerebral cortex
Nat Rev Neurosci
 , 
2007
, vol. 
8
 (pg. 
427
-
437
)
Morishima
M
Kawaguchi
Y
Recurrent connection patterns of corticostriatal pyramidal cells in frontal cortex
J Neurosci
 , 
2006
, vol. 
26
 (pg. 
4394
-
4405
)
Morishima
M
Morita
K
Kubota
Y
Kawaguchi
Y
Highly differentiated projection-specific cortical subnetworks
J Neurosci
 , 
2011
, vol. 
31
 (pg. 
10380
-
10391
)
Nachev
P
Kennard
C
Husain
M
Functional role of the supplementary and pre-supplementary motor areas
Nat Rev Neurosci
 , 
2008
, vol. 
9
 (pg. 
856
-
869
)
Neafsey
EJ
Bold
EL
Haas
G
Hurley-Gius
KM
Quirk
G
Sievert
CF
Terreberry
RR
The organization of the rat motor cortex: a microstimulation mapping study
Brain Res
 , 
1986
, vol. 
396
 (pg. 
77
-
96
)
O'Leary
DD
Koester
SE
Development of projection neuron types, axon pathways, and patterned connections of the mammalian cortex
Neuron
 , 
1993
, vol. 
10
 (pg. 
991
-
1006
)
Otsuka
T
Kawaguchi
Y
Cell diversity and connection specificity between callosal projection neurons in the frontal cortex
J Neurosci
 , 
2011
, vol. 
31
 (pg. 
3862
-
3870
)
Otsuka
T
Kawaguchi
Y
Firing-pattern-dependent specificity of cortical excitatory feed-forward subnetworks
J Neurosci
 , 
2008
, vol. 
28
 (pg. 
11186
-
11195
)
Paxinos
G
Watson
C
The rat brain in stereotaxic coordinates
 , 
2007
6th ed
San Diego
Academic Press
Rockland
KS
Rockland
KS
Kaas
JH
Peters
A
Elements of cortical architecture: hierarchy revisited
Cerebral cortex
 , 
1997
, vol. 
Vol. 12: Extrastriate cortex in primates
 
New York
Plenum Press
(pg. 
243
-
293
)
Rubio-Garrido
P
Pérez-de-Manzo
F
Porrero
C
Galazo
MJ
Clascá
F
Thalamic input to distal apical dendrites in neocortical layer 1 is massive and highly convergent
Cereb Cortex
 , 
2009
, vol. 
19
 (pg. 
2380
-
2395
)
Salin
PA
Bullier
J
Corticocortical connections in the visual system: structure and function
Physiol Rev
 , 
1995
, vol. 
75
 (pg. 
107
-
154
)
Sherman
SM
Guillery
RW
Distinct functions for direct and transthalamic corticocortical connections
J Neurophysiol
 , 
2011
, vol. 
106
 (pg. 
1068
-
1077
)
Shipp
S
Structure and function of the cerebral cortex
Curr Biol
 , 
2007
, vol. 
17
 (pg. 
R443
-
R449
)
Sul
JH
Jo
S
Lee
D
Jung
MW
Role of rodent secondary motor cortex in value-based action selection
Nat Neurosci
 , 
2011
, vol. 
14
 (pg. 
1202
-
1208
)
Suter
BA
Migliore
M
Shepherd
GM
Intrinsic electrophysiology of mouse corticospinal neurons: a class-specific triad of spike-related properties
Cereb Cortex
 , 
2013
, vol. 
23
 (pg. 
1965
-
1977
)
Tanji
J
The supplementary motor area in the cerebral cortex
Neurosci Res
 , 
1994
, vol. 
19
 (pg. 
251
-
268
)
Theyel
BB
Llano
DA
Sherman
SM
The corticothalamocortical circuit drives higher-order cortex in the mouse
Nat Neurosci
 , 
2010
, vol. 
13
 (pg. 
84
-
88
)
Uematsu
M
Hirai
Y
Karube
F
Ebihara
S
Kato
M
Abe
K
Obata
K
Yoshida
S
Hirabayashi
M
Yanagawa
Y
, et al.  . 
Quantitative chemical composition of cortical GABAergic neurons revealed in transgenic Venus-expressing rats
Cereb Cortex
 , 
2008
, vol. 
18
 (pg. 
315
-
330
)
Ushimaru
M
Ueta
Y
Kawaguchi
Y
Differentiated participation of thalamocortical subnetworks in slow/spindle waves and desynchronization
J Neurosci
 , 
2012
, vol. 
32
 (pg. 
1730
-
1746
)
Van der Gucht
E
Hof
PR
Van Brussel
L
Burnat
K
Arckens
L
Neurofilament protein and neuronal activity markers define regional architectonic parcellation in the mouse visual cortex
Cereb Cortex
 , 
2007
, vol. 
17
 (pg. 
2805
-
2819
)
Vandevelde
IL
Duckworth
E
Reep
RL
Layer VII and the gray matter trajectories of corticocortical axons in rats
Anat Embryol (Berl)
 , 
1996
, vol. 
194
 (pg. 
581
-
593
)
Van De Werd
HJ
Uylings
HB
The rat orbital and agranular insular prefrontal cortical areas: a cytoarchitectonic and chemoarchitectonic study
Brain Struct Funct
 , 
2008
, vol. 
212
 (pg. 
387
-
401
)
Veinante
P
Deschênes
M
Single-cell study of motor cortex projections to the barrel fields in rats
J Comp Neurol
 , 
2003
, vol. 
464
 (pg. 
98
-
103
)
Zikopoulos
B
Barbas
H
Circuits for multisensory integration and attentional modulation through the prefrontal cortex and the thalamic reticular nucleus in primates
Rev Neurosci
 , 
2007
, vol. 
18
 (pg. 
417
-
438
)