Abstract

We have investigated the dendritic and axonal morphology of connected pairs of L4 spiny neurons and L2/3 pyramidal cells in rat barrel cortex. The ‘projection’ field of the axons of L4 spiny neurons in layers 2/3, 4 and 5 has a width of 400–500 μm thereby defining an anatomical barrel-column. In layer 2/3, the averaged axonal ‘projection’ field of L4 spiny neurons together with the dendritic ‘receptive’ field of the connected L2/3 pyramidal cells form a mostly column-restricted anatomical L4-to-L2/3 ‘innervation domain’ that extends 300–400 μm and includes mostly basal dendrites. In the L4-to-L2/3 innervation domain a single L4 spiny neuron contacts ~300–400 pyramidal cells while in the L4-to-L4 innervation domain it contacts ~200 other L4 spiny neurons. Similarly ~300–400 L4 spiny neurons converge onto a single pyramidal cell and ~200 L4 spiny neurons innervate another L4 spiny neuron. The L2/3 pyramidal cell axon has a vertical projection field spanning all cortical layers, and a long-range horizontal field in layers 2/3 (width 1100–1200 μm) and 5 (700–800 μm) projecting across column borders. The results suggest that the flow of excitation within a barrel-column is determined by the largely columnar confinement of the L4-to-L4 and L4-to-L2/3 innervation domains. A whisker deflection activates ~140 L4 spiny neurons that will generate EPSPs in most barrel-related L2/3 pyramidal cells of a principal whisker column. The translaminar synaptic transmission to layer 2/3 and the axonal projection fields of L2/3 pyramidal cells are the major determinants of the dynamic, multi-columnar map in which a single whisker deflection is represented in the cortex.

Introduction

In the barrel cortex, L4 spiny neurons have a dendritic and axonal domain that in layer 4 is mainly confined to individual, cytoarchitectonically defined barrels (Lorente de Nó, 1922, 1943; Woolsey and van der Loos, 1970; Harris and Woolsey, 1983; Simons and Woolsey, 1984; Lübke et al., 2000). Their axons project throughout all cortical layers and establish synaptic contacts preferentially with other L4 spiny neurons in the same barrel and L2/3 pyramidal cells (Feldmeyer et al., 1999, 2002). These collaterals relay excitation from the thalamus to granular, supragranular and presumably also to infragranular layers.

The electrical representation of a single whisker deflection in layer 2/3 is multi-columnar spreading almost across the entire barrel field (Orbach et al., 1985; Masino and Frostig, 1996; Chen-Bee and Frostig, 1996; Moore and Nelson, 1998; Zhu and Connors, 1999; Brett-Green et al., 2001; Petersen et al., 2003) (M. Brecht, A. Roth and B. Sakmann, unpublished observations). Functional connections between barrels are sparse (Feldmeyer et al., 1999; Petersen and Sakmann, 2000), consistent with the fact that L4 spiny neuron excitation following stimulation of one barrel is restricted to a single barrel-column (Petersen and Sakmann, 2001) and reflecting the predominantly vertical organization of their axons (Lübke et al., 2000). With respect to the anatomical basis of the cortical representation of a whisker deflection the questions arise where the divergence from single-column to multi-columnar excitation occurs and what the exact dimensions of the L4-to-L2/3 projection in supra- and infragranular layers with respect to the barrel borders are.

The functional architecture of connections in the barrel cortex can be altered by modifying sensory input from the whisker pad, often referred to as plasticity of sensory maps (Armstrong-James et al., 1992; Keller and Carlson, 1999; Kossut and Juliano, 1999; Wallace and Fox, 1999) [for a review see Woolsey (Woolsey, 1990)]. Such alterations in the representational areas could be mainly functional, e.g. by changing synaptic efficacy in existing circuits. Alternatively, they could include morphological changes in the density of axon collaterals, synaptic boutons and/or dendritic spines. Sensory deprivation experiments suggest that such plastic changes of representational maps may occur in the vertical connections between excitatory L4 and L2/3 neurons as well as in the horizontal connections within layer 2/3 (Shepherd et al., 2003). To interpret such changes a quantitative anatomy of connections within a barrel-column must be established. For this we developed a method to average the dendritic and axonal domains of neurons.

Here we describe the dimensions of axonal arbors of L4 spiny neurons and their overlap with L2/3 pyramidal cell dendrites as well as the spread of L2/3 axons to delineate the morphological substrate for the vertical and horizontal spread of excitation. For this purpose we analysed the anatomical dimensions of monosynaptically connected pairs of neurons in layers 4 and 2/3, respectively, using reconstructions in relation to barrel borders; the locations of putative synaptic contacts established by the L4 spiny neuron axons on L2/3 pyramidal cells were determined and the average spread of axon collaterals of L2/3 pyramidal cells was measured.

The dimensions of the vertical axonal projections of L4 spiny neurons and of the dendritic arbor of L2/3 pyramidal cells are almost perfectly matched. They form a barrel-column-restricted innervation domain and, together with the projections to infragranular layers, a cytoarchitectonic barrel-column while the long-range horizontal axons of L2/3 pyramidal cells project to adjacent columns. We suggest that the major anatomical substrate for the spread of weak afferent excitation in layer 2/3 is the axonal arbor of L4 spiny cells whereas for strong excitation it is the axonal arbor of ‘barrel-related’ L2/3 pyramidal cells.

Materials and Methods

Materials

The anatomical data are based on 16 pairs of synaptically connected L4 spiny neurons and L2/3 pyramidal cells that were reconstructed after biocytin filling via the recording pipette in acute thalamocortical brain slices. In addition three individual L2/3 pyramidal cells were reconstructed. The morphometric data were derived from 14 reconstructed pairs, the 2D density maps were derived from nine reconstructed pairs. Two pairs of cells were analysed in parallel by camera lucida (2D) drawings and by 3D Neurolucida (MicroBrightField, Colchester, VT) reconstruction for comparison. In two cell pairs the light-microscopically identified synaptic contacts were confirmed by serial EM sectioning. In addition, five pairs of synaptically coupled L4 spiny neurons were analysed morphometrically.

Electrophysiological Recordings

All experiments were carried out in accordance with the animal welfare guidelines of the Max Planck Society and the University of Freiburg. Wistar rats (19–25 days old) were anaesthetized with halothane, decapitated and slices through the somatosensory cortex (barrel cortex) were cut in cold extracellular solution using a vibrating microslicer (DTK-1000, Dosaka Co. Ltd, Kyoto, Japan) and prepared according to methods described elsewhere (Agmon and Connors, 1991; Feldmeyer et al., 1999, 2002). Whole-cell voltage recordings from pre- and postsynaptic neurons were made as in detail described elsewhere (Feldmeyer et al., 1999, 2002). In brief, a postsynaptic cell was recorded from with one pipette and subsequently synaptic connections to this cell were searched with a second pipette in the loose-patch configuration. After establishing a loose seal on a presumed presynaptic cell, action potentials were elicited by applying 10 ms current pulses through the membrane. The extracellularly recorded action potential was visible as a small deflection on the voltage response. When this stimulation resulted in an EPSP in the post-synaptic neuron, the presynaptic cell was patched with a new pipette filled with biocytin-containing intracellular solution and action potentials were elicited in the whole-cell (voltage recording) mode. At the end of each experiment, low power micrographs were taken with the recording pipettes in place to later orient the reconstructions with respect to barrel borders (Feldmeyer et al., 2002). All pairs of neurons used in this study were located in the dorsal end of the posterior medial barrel subfield of the primary somatosensory cortex as indicated by the relatively large barrel size (see Table 3).

Solutions

Slices were continuously superfused with extracellular solution containing (in mM): 125 NaCl, 2.5 KCl, 25 glucose, 25 NaHCO3, 1.25 NaH2PO4, 2 CaCl2 and 1 MgCl2 bubbled with 95% O2 and 5% CO2. The intracellular pipette solution contained (in mM): 105 K-gluconate, 30 KCl, 10 HEPES, 10 phosphocreatine, 4 ATP-Mg, 0.3 GTP (adjusted to pH 7.3 with KOH). The osmolarity of this solution was 300 mOsm. For morphological analysis, 2–3 mg/ml biocytin (Sigma, Munich, Germany) was added to the internal solution and neurons were filled during 0.5–2.5 h of recording.

Histological Procedures

Following recording and intracellular filling with biocytin, brain slices were fixed in 100 mM phosphate-buffered solution (PB, pH 7.4) containing 1% paraformaldehyde and 2.5% glutaraldehyde at 4°C for at least 24 h. Selected pairs were processed for light- and electron microscopy to perform a detailed morphological analysis and to confirm putative light-microscopically identified synaptic contacts as described elsewhere (Lübke et al., 2000). For light-microscopic reconstruction (see below) slices were embedded in a water-based medium (Mowiol; Clariant, Sulzbach i. Taunus, Germany) to reduce tissue shrinkage. For electron microscopy, slices were post-fixed in 0.5% OsO4 (30–45 min), then dehydrated and embedded in Durcupan (Fluka, Deisenhofen, Germany). Ultrathin sections were counterstained and examined with a Philips CM 100 electron microscope (Philips, Eindhoven, The Netherlands).

Morphological Reconstructions

Pairs of synaptically coupled neurons were examined under the light microscope at high magnification (×1200). Only pairs for which a complete physiological analysis was made and that had no obvious truncation of their dendritic and axonal profiles were used for qualitative and quantitative analysis of their morphology. Representative pairs of neurons were photographed at different magnifications to document dendritic morphology, axonal projection and location of synaptic contacts. Neurons were then drawn with the aid of a camera lucida attached to an Olympus BX50 microscope (Olympus, Hamburg, Germany) at a final magnification of ×720 or ×1200. For some pairs of neurons, three-dimensional reconstructions were also made using a ×100, 1.4 NA oil-immersion objective (Zeiss) and Neurolucida software. These reconstructions provided the basis for further quantitative morphological analysis of the following parameters: (i) maximal horizontal field span of the dendrites and axons of the pre- and postsynaptic neurons, (ii) total number and dendritic location of putative synaptic contacts per neuron, and (iii) dendritic and axonal ‘length density’ and total number and density of synaptic boutons counted for the axons of L4 spiny neurons and L2/3 pyramidal cells. Measurements were not corrected for shrinkage. For all data, means ± S.D. were calculated.

Axonal and Dendritic Density Maps

We developed a method to obtain two-dimensional (2D) maps of axonal and dendritic ‘length density’ or axonal ‘bouton density’ from 2D reconstructions. First, the length of all axonal branches was measured or all boutons were counted and the length of all dendritic branches was measured manually in a 25 μm × 25 μm (L4–L4 connections) or 50 μm × 50 μm (L4–L2/3 connections) Cartesian grid superimposed on the drawing, yielding a raw density map. For alignment of these maps with respect to the barrel centre, barrel borders were identified in the low power (×2.5 objective) bright field micrographs made from the acute brain slice (Lübke et al., 2000; Petersen and Sakmann, 2000; Feldmeyer et al., 2002). Spatial low-pass filtering of these maps was performed by 2D convolution with a Gaussian kernel (σ = 25 μm or 50 μm) and continuous 2D density functions were constructed using bicubic interpolation in Mathematica 4.1 (Wolfram Research, Champaign, IL).

Two-dimensional maps constructed from 2D reconstructions involve two projection steps: first, the z component of dendritic and axonal lengths is neglected, and second, an essentially three-dimensional (3D) density map is projected onto a 2D plane. To estimate possible errors in the widths of regions containing 80% of the integrated density (Table 3) due to the use of 2D reconstructions and 2D projections of 3D density maps, two pairs of cells were fully reconstructed in Neurolucida (Micro-brightfield, Colchester, VT). The generation of 2D reconstructions and 2D maps from these 3D reconstructions was then simulated in Mathematica using the same spatial low-pass filtering and interpolation as described above. Horizontal and vertical widths (in a plane containing the soma) of regions containing 80% of the integrated density were measured. Widths of 80% regions (bounded by isosurfaces) in the full 3D maps were 9% to 24% larger than in the corresponding 2D projections, while the additional effect of neglecting the z component of dendritic and axonal lengths in 2D reconstructions was less than 2% change in widths. Furthermore, care was taken to minimize shrinkage in the xy plane by using water-based embedding media (see above).

For the L2/3 pyramidal cell axons the dimensions of the density map may be underestimated when the axonal arbor is not rotationally symmetric. However the axonal spread of L2/3 pyramidal cells described here for 2D projections is comparable to that for 3D reconstructions of identified L2/3 pyramidal cells recorded and filled during in vivo experiments (Petersen et al., 2003) (M. Brecht, A. Roth and B. Sakmann, unpublished observation). Nevertheless the axonal arbor outline of L2/3 pyramidal cells in the 2D map represents only a lower estimate of its width in a projection plane almost parallel to the row of barrels.

Results

The Axonal Arborization of L4 Spiny Neurons and Dendritic Arbor of L2/3 Pyramidal Cells Are Confined to a ‘Home’ Barrel

Infrared video microscopy facilitated the selection of pairs of neurons according to shape, size and location of somata within a single barrel or above the barrel throughout layer 2/3 as described (Feldmeyer et al., 2002). Paired recordings were made exclusively from neurons in the somatosensory cortex. The mean geometric distance of the postsynaptic pyramidal cell from the presynaptic neuron varied between 204 and 403 μm and they were all located above barrels (‘barrel-related’ pyramidal cells). Pre- and postsynaptic neurons were defined as excitatory by their regular action potential firing pattern and the sensitivity of the postsynaptic response to glutamate receptor antagonists [AP5 and NBQX; see Feldmeyer et al. (Feldmeyer et al., 2002)].

Spiny Layer 4 Neurons

From the sample of synaptically coupled pairs of neurons, ~80% of the presynaptic neurons were identified as spiny stellate neurons, the remainder as star pyramidal cells (Feldmeyer et al., 2002). A clear distinction between the two classes of cells was not always possible because of the variability in the morphology of the apical dendrite. Almost all spiny stellate neurons had dendrites with an asymmetric dendritic arrangement oriented towards the centre of a barrel (Figs 1A and 2AC). The somata and dendrites of L4 star pyramidal cells were also exclusively located within layer 4 with the exception of the distal apical dendrite that extends in layer 2/3 without forming a terminal tuft (Fig. 2D). In contrast to spiny stellate cells, star pyramidal cells were located always towards the centre of a barrel and their basal dendrites showed no asymmetry (Feldmeyer et al., 1999; Lübke et al., 2000) (V. Egger and B. Sakmann, unpublished observation). However, the dendritic field of L4 spiny neurons was always restricted to the borders of the ‘home’ column in which the somata of the presynaptic neurons were located.

Axonal Arborization of L4 Spiny Neurons

The axonal collaterals of all synaptically coupled L4 spiny stellate and star pyramidal cells project throughout all cortical laminae from layer 1 to the white matter, but were largely confined to their ‘home’ column (Fig. 2). The majority of axonal collaterals branch off in layer 4 or upper layer 5 and ascend towards layer 2/3 forming a dense axonal projection with a high degree of collateralization. The axonal collaterals of L4 spiny neurons were densely covered with synaptic boutons suggesting a high probability of innervation of target neurons. Morphological characteristics of the presynaptic neurons, the pattern of their dendritic and axonal arborization are summarized quantitatively in Table 1.

Dendritic Arborization of L2/3 Pyramidal Cells

L2/3 pyramidal cells constitute a more heterogeneous population within the class of neocortical pyramidal cells with respect to their dendritic configuration (Gilbert and Wiesel, 1979; Valverde, 1986; Burkhalter, 1989; Gottlieb and Keller, 1997) [for review see DeFelipe and Farinas (DeFelipe and Farinas, 1992)]. The primary basal dendrites emerging from the soma give rise to numerous secondary, tertiary and higher-order basal dendrites of different length (Table 1) that form an almost symmetric ‘receptive’ field (Fig. 2). The basal dendrites of pyramidal cells monosynaptically innervated by L4 spiny neurons were also confined to the same barrel-column (Fig. 2) with one exception (Fig. 2C). The shape of the apical dendrite was, however, highly variable with respect to its length, bifurcation pattern, number of apical oblique dendrites and field size of the terminal tuft in layer 1 (compare pyramidal cells in Fig. 2; see also Table 1). Pyramidal cells in the upper to middle part of layer 2/3 have rather short main apical trunks that often bifurcate in twinned dendrites relatively close to the soma (~40–80 μm) giving rise to an extensive terminal tuft with a field span of ~300 μm (compare postsynaptic neurons in Fig. 2). Pyramidal cells in the lower part of layer 2/3 have a long (150 μm) and more prominent apical dendrite with several oblique dendrites of various length and distance from the soma (Figs 1A and 2D; Table 1).

In one paired recording an ‘unusual’ pattern of dendrites and axonal collaterals of pre- and postsynaptic neurons was observed (Fig. 2C). The postsynaptic L2/3 pyramidal cell was located at the border of the barrel-column close to the pial surface and had an apical dendrite and a terminal tuft that were slightly tangentially oriented. The soma of the L4 spiny stellate neuron was not located directly underneath the postsynaptic neuron, however its axonal projection was largely vertically oriented as shown for other L4 spiny neurons. In this pair of neurons two synaptic contacts were located in the terminal tuft of the target neuron (Fig. 2C).

In Figure 2D the L4 spiny neuron was a star pyramidal cell. Star pyramidal cells had an axonal projection similar to that of spiny stellate neurons (Lübke et al., 2000).

Patch-clamping under visual control involves the selection of neurons to record from and may therefore lead to biased sampling. We tried to minimize this problem by attempting to use a fair sampling strategy, recording from postsynaptic neurons throughout layer 2/3 as well as to the left and right of the barrel-column. Similarly, presynaptic neurons were searched for in the entire barrel. Figures 2 and 3 illustrate the range of positions of both pre- and postsynaptic neurons recorded from.

Axonal ‘Projection’ Fields, Dendritic ‘Receptive’ Fields and Synapse Locations

Overlap of ‘Projection’ and ‘Receptive’ Fields in the L4-to-L2/3 Connection

When all reconstructions of L4-to-L2/3 neuron pairs (n = 9) are superimposed and aligned with respect to the barrel centre (Fig. 3) it is clearly evident that within a barrel-column the ‘projection’ zone of L4 spiny neuron axons overlaps considerably with the dendritic ‘receptive’ field of L2/3 pyramidal cells. To quantify this overlap the axonal length of L4 spiny neurons was measured using a 50 μm × 50 μm grid superimposed on 2D reconstructions of the cell pairs (n = 9). We then constructed a 2D map of the axonal ‘length density’ of L4 axons using bicubic interpolation of the original grid points yielding an ‘average’ axonal projection of L4 spiny neurons. The reference point for alignment of the reconstructions was either the centre of the barrel (Fig. 4A1), the soma of the L4 spiny neuron (not shown) or the soma of the L2/3 pyramidal cell (Fig. 4B1). The map of the L4 axonal ‘length density’ clearly shows that the largest fraction of the L4 axon is restricted to a barrel-column. Comparing the outline of the average barrel-column to the contour line including 80% of the L4 axonal ‘length density’ shows that its majority is restricted to the barrel-column throughout all cortical layers (Table 3). However in layer 2/3 the width of the 2D map is somewhat larger than in L4 and L5 (Fig. 4A1, 4B1, Table 3).

Figure 4A2 illustrates the ‘dendrite density’ maps of L2/3 pyramidal cells. Here, all reconstructed pairs were also aligned with respect to the centre of the ‘home’ barrel. The contour line including 80% of all dendritic length lies within the borders of the barrel-column (Table 3). The dendrite density map, when normalized to the pyramidal cell somata, shows two domains corresponding to the basal and the apical dendrites, respectively (Fig. 4B2).

Assuming that synaptic connections between axons and dendrites in a given region are formed in a random process, we calculated the predicted innervation domain by multiplying the axonal ‘length density’ and the dendritic ‘length density’ (Fig. 4A3); for discussion, see also Stepanyants et al. and Kalisman et al. (Stepanyants et al., 2002; Kalisman et al., 2003). The extent of this predicted innervation domain is limited by the extent of the dendritic arborization. The sharp delineation of this innervation domain of L2/3 pyramidal cells by L4 spiny neurons is particularly clear when L4 axonal and L2/3 dendritic length maps are normalized with respect to the somata of pyramidal cells. The 2D projected dimensions of the innervation domain are ~240 μm (horizontal) and 330 μm (vertical; see Table 3). Only the basal dendritic field is overlapping with axonal arbors of L4 cells (Fig. 4B3). Within this innervation domain, the axonal length of a single L4 spiny neuron was 5519 μm and 3936 μm, with a bouton density of 0.405 ± 0.01 per μm, corresponding to 2235 and 1594 synaptic boutons in the innervation domain. The L2/3 dendritic length was 2724 μm and 2730 μm for the barrel-centred and the soma-centred density maps, respectively, with a dendritic spine density on the basal dendrites of 0.97 ± 0.1 spines/μm (n = 3 cells). From these numbers the fraction of synaptic contacts established by the axon collaterals of L4 spiny neurons can be estimated (see Discussion).

Synaptic Contacts Between L4 Spiny Neurons and L2/3 Pyramidal Cells

In 14 pairs of synaptically coupled L4 spiny neurons and L2/3 pyramidal cells the number and dendritic location of putative synaptic contacts was analysed light- and electron-microscopically (Fig. 1BE, insets in Fig. 2; see also Table 2). The total number of light-microscopically identified synaptic contacts in these cell pairs was 65. Synaptic contacts were exclusively found on the dendrites of L2/3 pyramidal cells and were always located within the ‘home’ column of the presynaptic neuron. For individual connections, their number varied between 4 and 6 [mean: 4.8 ± 0.6; see also Table 2 and Feldmeyer et al. (Feldmeyer et al., 2002)]. Synaptic contacts were preferentially located on the basal dendrites (86.2%, n = 56; Table 2). Of these, 89.3% were en passant synaptic contacts predominantly on second- and third-order basal dendrites. The remaining contacts (10.7%) were found on fourth-order basal dendrites. Synaptic contacts on basal dendrites were located at a distance of 15–130 μm from the soma (see also Table 2). Only a smaller fraction (13.8%) was located on second- and third-order apical oblique dendrites at a distance of 115–200 μm from the soma (Fig. 2C; see also Table 2). Out of 65 light-microscopically identified synaptic contacts only 5 (7.7%) were found on the same dendrite within a distance of 50 μm to each other.

Synaptic contacts were always established by vertical axonal collaterals of higher order of the presynaptic L4 spiny neuron and were located at a distance of 100 μm from the presynaptic soma. In two cell pairs serial EM analysis was used to confirm that putative synaptic contacts, identified light-microscopically, were indeed synaptic contacts according to criteria described elsewhere (Markram et al., 1997). In these two pairs analysed, five and four synaptic contacts were identified by light microscopy, respectively. All five synaptic contacts for the first pair of neurons were confirmed at the EM level. In general, ~85% of putative, light-microscopically identified synaptic contacts were also found at the EM level (Markram et al., 1997; Lübke et al., 2000; Feldmeyer et al., 2002); however, no wrongly identified contacts were revealed by EM analysis. In the other pair an additional contact was found, however, we could not rule out the possibility that this one may also be an autaptic contact. Four contacts were established with dendritic spines, the remaining on the dendritic shaft. Postsynaptic elements were regarded as dendritic shafts if they were larger than spines and contained mitochondria.

Innervation Domain and Location of Synaptic Contacts

To estimate whether the extent of the innervation domain of axonal collaterals of L4 spiny neurons and L2/3 pyramidal cell dendrites corresponds to the actual density of innervation, the location of synaptic contacts was marked in the innervation domain (Figs 5 and 7; blue dots). Most contacts are indeed located within the borders of the innervation domain. This is the case irrespective of whether the reconstructions were centred with respect to barrels (Fig. 5A) or to L2/3 pyramidal cell somata (Fig. 5B). Only three or four of 32 contacts (9–13%; n = 7 connections) were located outside the predicted innervation domain.

L4 spiny cells are also interconnected in layer 4 (Feldmeyer et al., 1999). Postsynaptic L4 dendrites and presynaptic L4 axons form an innervation domain with the vertical and horizontal dimensions of a barrel (Fig. 6). As for the L4-to-L2/3 connection, only 2 of 23 contacts (9%; n = 5 connections one of which was reciprocal) were located outside the predicted innervation domain (Fig. 7, purple dots) in line with an earlier suggestion by Harris and Woolsey (Harris and Woolsey, 1983). The axonal length density of a single neuron in the L4-to-L4 innervation domain was 2254 μm (Figs 6 and 7). Given a bouton density of 0.405 ± 0.010 per mm (see above) this corresponds to 913 boutons. The length of postsynaptic L4 spiny neuron dendrites in the L4-to-L4 innervation domain was 1242 μm with a spine density of 0.454 ± 0.105 per μm (n = 4).

Axons of L2/3 Pyramidal Cells Receiving Monosynaptic Input from L4 Spiny Neurons Project to Adjacent Columns

To identify possible anatomical determinants of the lateral spread of excitation between barrel columns in layer 2/3 the axonal projection of L2/3 pyramidal cells receiving monosynaptic input from L4 spiny neurons was also reconstructed and quantitatively analysed. The extent of the axonal arborization of L2/3 pyramidal cells was variable between different pairs of neurons. However, the axonal arbor always displayed two distinct domains: firstly, a vertically oriented domain of collaterals that emerged from the main axon ascending towards layer 1 where they terminated; and secondly long-range horizontal collaterals (Fig. 8A) in layers 2/3 and 5.

The main axon of L2/3 pyramidal cells originated with a thick axon initial segment either directly from the soma (Fig. 8A) or from one of the primary basal dendrites (not shown). It descends towards the white matter giving rise to numerous vertical and long-range horizontal collaterals in layer 2/3 and layer 5 (Fig. 8A,B; see also Table 1). In the majority of cell pairs analysed these collaterals in layers 2/3 and 5 could be followed to project over distances up to 2.5 mm and 0.8 mm, respectively (Table 1), suggesting that L2/3 pyramidal cells excite neurons in adjacent cortical columns.

Projection Density Map

To obtain a quantitative estimate of the outline of L2/3 axonal projections reconstructions of L2/3 pyramidal cells were aligned with respect to the barrel centres (Fig. 8B). A 2D map of L2/3 axonal length was obtained with the same grid of 50 μm × 50 mm as was used for the quantification of L4 axonal arborizations. The map generated when reconstructions where aligned with respect to the barrel centres illustrates the large horizontal spread of L2/3 pyramidal cell axons across the borders of adjacent barrel-columns (Fig. 8C). In the horizontal plane in layer 2/3 the contour line including 80% of the axonal ‘length density’ extends across the adjacent barrel-columns with a maximal width of ~1.2 mm. A second projection field in layer 5 is narrower (maximal width is ~800 μm), whereas in layer 4 the axonal ‘length density’ is restricted to the width of the barrel (Table 3).

This density map illustrates the ‘average’ projection of axon collaterals of L2/3 pyramidal cells receiving monosynaptic excitatory input from layer 4. Clearly, the main axonal projection of the L2/3 pyramidal cell extends far into the neighbouring barrel-columns both in layers 2/3 and 5. However, it is possible that in slices a fraction of the long-range axon collaterals are truncated and that their maximal horizontal spread is sub-stantially wider.

For a subset of cell pairs (n = 7) bouton counts for both L4 spiny neurons and L2/3 pyramidal cells were determined using the same 50 μm × 50 μm grid as for the axonal ‘length density’ measurements. The L4 and L2/3 bouton density maps thus obtained were very similar to the axonal ‘length density’ maps described above (not shown). Likewise, the resulting innervation domains had comparable dimensions using either bouton density or axonal ‘length density’ maps.

Discussion

To understand the mechanisms of the cortical representation of a sensory stimulus, like a whisker deflection, it is essential to relate the dendritic and axonal morphology and the location of synaptic contacts to the pattern of PSPs and APs evoked by this stimulus. Anatomical connections are static, compared with the time span of a whisker deflection, whereas PSPs and APs change rapidly in time and in space. For example, a few milliseconds after the onset of a whisker deflection its electrical representation in the cortical network is very different from that after tens of milliseconds (Petersen et al., 2003). Here we describe the anatomical constraints for the vertical flow of excitation within and for the subsequent horizontal spread across barrel-columns at the subcellular level. To our knowledge, the dimensions of dendritic and axonal arbors of monosynaptically connected pairs of neurons located in layers 4 and 2/3, their axonal-dendritic overlap and synaptic density distribution have, so far, not been analysed. Three rules seem to govern the L4-to-L2/3 connections: (i) Axons of L4 spiny neurons and dendrites of L2/3 pyramidal cells are restricted roughly to the width of a barrel-column as defined by the cross-sectional area of a barrel. (ii) The axonal ‘projection’ field of L4 spiny neurons is matched to the basal dendritic ‘receptive’ field of L2/3 pyramidal cells thus defining a localized, blob-like L4-to-L2/3 innervation domain in which a single L4 spiny neuron forms synaptic connections with 300–400 pyramidal cells in layer 2/3. (iii) Axons of these L2/3 pyramidal cells extend horizontally from the ‘home’ into adjacent columns in cortical layers 2/3 and 5 forming synapses predominantly with other excitatory neurons.

Columnar Organization of the L4-to-L2/3 Connection

In the somatosensory cortex the restriction of L4 spiny cell dendrites to barrel column and laminar borders is well documented (Lorente de Nó, 1922; Woolsey and van der Loos, 1970; Simons and Woolsey, 1984; Lübke et al., 2000). In addition, the axonal domain of L4 spiny neurons shows also a high degree of columnarity (Harris and Woolsey, 1983; Bernardo et al., 1990; Lübke et al., 2000). Its dimensions are comparable to those of functional barrel columns identified either by 2-deoxyglucose autoradiography (Chmielowska et al., 1986; McCasland and Woolsey, 1988), imaging of cerebral blood flow (Woolsey et al., 1996) or voltage sensitive dye imaging (Laaris et al., 2000; Petersen and Sakmann, 2001; Petersen et al., 2003). However, inter-barrel projections were occasionally observed (Brecht and Sakmann, 2002). The axonal collaterals of barrel-related L2/3 pyramidal cells project across the borders of a barrel column, extending horizontally in layer 2/3 and 5 but largely sparing layer 4. On the other hand, the ‘hourglass’ structure apparent after extracellular dye injection into a single barrel (Bernardo et al., 1990; Miller et al., 2001) may represent axon collaterals of L4 spiny neurons and L2/3 pyramidal cells and possibly also of inhibitory interneurons. The orientation of these axon collaterals is highly asymmetrical in the SI barrel cortex, as barrel-related columns within a row appear more strongly interconnected than those in different rows. This was not found outside the SI cortex (Bernardo et al., 1990). In summary, the morphology of neurons connecting L4 with L2/3 seems to be markedly more stereotyped in the barrel cortex than in other cortices.

Two Innervation Domains in a Barrel-column

The innervation domain of L2/3 pyramidal cells by L4 spiny neurons is located above a barrel, with similar horizontal dimensions. It is largely co-extensive with the distribution of light-microscopically identified synaptic contacts (Figs 5 and 7; blue dots). The same holds true for the L4-to-L4 innervation domain (Fig. 7; purple dots; see below). Whether such co-extension of the predicted innervation domains and the actual distribution of synaptic contacts exists in other cortical connections remains to be determined. An estimate of the number of L2/3 pyramidal cells innervated by the axon collaterals of a single L4 spiny neuron can be obtained from the number of boutons in the L4-to-L2/3 innervation domain which was 2235 and 1594, for the barrel-centred and the soma-centred density maps, respectively. When divided by the mean number of contacts per connection [4.5 (Feldmeyer et al., 2002)] this yields 497 and 354 as the number of L2/3 pyramidal cells innervated by a single L4 spiny neuron. However, since ~15–20% of cortical neurons are GABAergic interneurons, this number has to be reduced to 397 and 283 (assuming for simplicity that connections between L4 spiny neurons and L2/3 interneurons have a roughly similar number of contacts per connection). On the other hand, synaptic connections are also made outside the innervation domain containing 80% of the predicted synaptic density, suggesting that the number of postsynaptic targets is actually higher. In the adult animal, a barrel-column with the dimensions: width, 350 μm; layer 4 height, 300 μm; layer 2/3 height, ~500 μm (Gottlieb and Keller, 1997) and neuron densities of 111 830 per mm3 in layer 4 and 69 290 per mm3 in layer 2/3 (Keller and Carlson, 1999) contains 4110 and 4244 neurons in layers 4 and 2/3 [see also Bruno and Simons (Bruno and Simons, 2002)], respectively, of which 80% are excitatory. As the ratio between layer 4 and layer 2/3 neurons is close to unity, it follows by symmetry that ~300–400 L4 spiny neurons innervate a single L2/3 pyramidal cell. Another way to calculate the L4-to-L2/3 convergence is based on the spine density of L2/3 pyramidal cells [0.97 per μm, comparable to the value of 1.12 per mm reported by Trevelyan and Jack (Trevelyan and Jack, 2002) and the total dendritic length of a neuron in the innervation domain (~2730 μm)]. As the number of contacts per connection is 4.5 (Feldmeyer et al., 2002) and since ~50% of these are made onto spines (Lübke et al., 2000) the number of L4 spiny neurons per L2/3 pyramidal cell would be (0.97 per μm × 2730 μm)/(0.5 × 4.5) = 1177. The difference between the two values for the convergence (300–400 vs 1177) indicates that between one quarter to one-third of all synaptic contacts onto L2/3 pyramidal cells in the L4-to-L2/3 innervation domain are made by vertical inputs from L4. About two-thirds are made by other neurons, most likely by other L2/3 pyramidal cells (Reyes and Sakmann, 1999; Egger et al., 1999; Yoshimura et al., 2000) located in the same and adjacent columns.

L4 spiny neurons also innervate other L4 neurons in the same barrel. They also form a sharply column-restricted L4-to-L4 innervation domain extending just below the L4-to-L2/3 domain. What is the connectivity of these neurons in layer 4? To address this the number of L4 axonal boutons established by a single L4 spiny neuron in the L4-to-L4 innervation domain was calculated (see Results), yielding 913 boutons of which only 0.5 are forming an autaptic contact (Feldmeyer et al., 1999). Dividing this number by the mean number of contacts per connection [3.4 (Feldmeyer et al., 1999)] gave a value of 268 L4 spiny neurons that are innervated by a single L4 spiny neuron. However, since 15–20% of cortical neurons are GABAergic interneurons, this number is reduced to 215 (assuming for simplicity that connections between L4 spiny neurons and interneurons have a similar number of contacts per connection).

By symmetry, on average 215 spiny neurons in layer 4 converge onto this single L4 spiny neuron. Convergence can also be estimated from the spine density of L4 spiny neurons (0.45 ± 0.11 per μm) and the total dendritic length of a neuron in this innervation domain (1242 μm). The number of contacts in a connection between L4 spiny neurons is 3.4 (Feldmeyer et al., 1999) and ~50% of these are made onto spines (Lübke et al., 2000) so that the number of L4 spiny neurons innervating a given L4 spiny neuron would be (0.45 per μm × 1242 μm)/(0.5 × 3.4) = 329. The difference between the two values obtained for the divergence (215 vs 329) indicates that ~65% of all synaptic contacts onto L4 spiny neurons in the L4-to-L4 innervation domain are made by other L4 spiny neurons.

Functional Connectivity

The above values then allow an order of magnitude estimate of the number of L4 spiny neurons and synaptic contacts on L2/3 pyramidal cells activated by a single whisker deflection in anaesthetized rats. In vivo imaging shows that the VSD signal evoked by principal whisker deflection is initially (≤15 ms) restricted to the cross-section of the PW barrel but later (>20 ms) spreads horizontally into surround whisker columns (Petersen et al., 2003). Recent in vivo whole-cell recordings suggest that all L4 neurons in a PW column respond with subthreshold EPSPs. However, only 4% of the recorded cells generate APs following a single whisker deflection during this initial phase [(Brecht and Sakmann, 2002); their Fig. 12D)]. Therefore, whisker stimulation excites ~140 neurons in layer 4 above threshold (4% of ~3400 excitatory neurons in an adult barrel). An individual L4 neuron innervates ~300–400 pyramidal cells in layer 2/3 and an AP in a L4 neuron generates with high reliability near coincident EPSPs in L4 and L2/3 neurons within a column (Feldmeyer et al., 1999, 2002). On the other hand, ~300–400 L4 spiny neurons innervate a single L2/3 pyramidal cell (see above). As 4% of these generate APs during the first 15 ms of a whisker deflection each L2/3 pyramidal cell in the PW column receives synaptic input from 12 to 16 L4 spiny neurons, on average. The probability of a L2/3 pyramidal cell receiving no synaptic input from layer 4 during this time is therefore small and consistent with in vivo whole-cell recordings from barrel-related L2/3 pyramidal cells which indicate that virtually all respond with a compound EPSP to principal whisker deflection (M. Brecht, A. Roth and B. Sakmann, unpublished). The average of these EPSPs is probably generating the initial column-restricted component of the single whisker-evoked VSD response in the neocortex (Petersen et al., 2003). In layer 2/3, the column-restricted VSD signal thus reflects input from ~140 L4 spiny neurons that form 140 × 400 × 4.5 = 252 000 synaptic contacts located mostly on the basal dendrites of L2/3 pyramidal cells.

Relation Between Anatomical and Functional Topographic Maps

How can the overlapping maps of dendritic and axon length density be related to functional maps constructed from receptive fields mapped by EPSPs and/or APs that define the dynamic sub- or suprathreshold electrical representation (‘excitation map’) of a whisker deflection? On the cortical surface subthreshold excitation evoked by deflection of a single whisker should be restricted initially to a relatively small area given by the blob-like L4-to-L2/3 innervation domain. Time resolved VSD imaging in acute slices of barrel cortex shows indeed that electrical stimulation of a single barrel elicits a wave of excitation largely delineated by the borders of a barrel-column (Laaris et al., 2000; Petersen and Sakmann, 2001). However, when inhibition is blocked excitation remains confined to a barrel in layer 4 but spreads horizontally in layers 2/3 and 5 but not in layer 4 (Laaris et al., 2000; Petersen and Sakmann, 2001; Laaris and Keller, 2002). Likewise in vivo the VSD signal evoked by a single whisker deflection on the cortex is initially restricted to the cross-section of the PW column and spreads later horizontally into surround whisker columns. This time dependent spread of excitation in the cortex may reflect the transition from sub- to suprathreshold excitation at L4-to-L2/3 synapses. The L4-to-L2/3 synapses are highly reliable (Feldmeyer et al., 2002) and may thus serve as a ‘gate’ for the horizontal spread of sensory stimulus-evoked excitation in supra- and to a smaller extent also in infragranular layers. The lateral borders of the representational map for a single whisker stimulus when the L4-to-L2/3 gate is ‘open’ are largely determined by the long-range horizontal axons of L2/3 pyramidal cells receiving monosynaptic input from L4 spiny neurons. The representational area of a single whisker movement by subthreshold excitation at later times (>20 ms) after the stimulation is therefore several times larger than a single barrel-column in accordance with results obtained by intrinsic signal imaging (Masino and Frostig, 1996; Chen-Bee and Frostig, 1996; Kleinfeld and Delaney, 1996; Brett-Green et al., 2001). In addition, the horizontal spread of excitation is most likely also controlled by inhibitory connections (Laaris et al., 2000; Petersen and Sakmann, 2001; Laaris and Keller, 2002).

Sensitivity of L2/3 Pyramidal Cell Excitation to Coincident Vertical and Horizontal Inputs

When does the L4-to-L2/3 gate open? Presumably a strong whisker deflection is sufficient to reach threshold in L2/3 pyramidal cells. However the combination of a weaker stimulus with associative excitation from other cortical areas may also be sufficient. In L2/3 pyramidal cells, excitation might become suprathreshold when inputs arriving in layer 1 at the apical dendrite are coincident with inputs from layer 4 and 2/3 to basal dendrites. This was demonstrated for coincident inputs to L5 pyramidal cells (Larkum et al., 1999, 2001). In consequence, the borders of the excitation area in the cortex evoked by a whisker deflection would also be very sensitive to associated synaptic input from afferents originating in other cortical regions and innervating the apical dendrites of L2/3 pyramidal cells.

Notes

The authors would like to thank Drs Imre Vida, Michael Brecht and Fritjof Helmchen for critically reading an earlier version of the manuscript and Moritz Helmstädter for fruitful discussions. We are also grateful to B. Joch, S. Nestel, M. Winter and M. Kaiser for excellent technical assistance.

Address correspondence to Dirk Feldmeyer, Max-Planck-Institut für medizinische Forschung, Abteilung Zellphysiologie, Jahnstrasse 29, D-69120 Heidelberg, Germany. Email: feldmeyr@mpimf-heidelberg.mpg.de.

Table 1

Quantification of different morphological parameters of synaptically coupled layer 4 excitatory spiny cells and layer 2/3 pyramidal cells

Quantity measured Projection (L4 spiny) cell Target (L2/3 pyramidal) cell 
Morphology (means ± SD) of somata and dendrites of the pre- and postsynaptic neurons of 14 synaptic connections between cells in layers 4 and 2/3. The morphology of the pre- and postsynaptic axon was determined from nine connections. Field span refers to the maximal horizontal distance between dendrites or axonal collaterals, respectively. The order of a dendrite is indicated by °. For example, 1° refers to the first branch originating from the soma or the main apical dendrite. For the apical oblique dendrites, starting from the 6° branch through to the 8° branch, the number of observations were, 6, 8 and 3, respectively. Only long-range horizontal collaterals were measured that were ≥500 μm in length which also showed no obvious truncations and growth cones at their tips. Measurements were taken at their origin from the main axon. 
Somata 
    Vertical diameter (μm) 12.9 ± 2.8 18.2 ± 2.5 
    Horizontal diameter (μm) 13.8 ± 2.4 15.6 ± 2.0 
Dendrites 
    Number of primary basals 4.3 ± 1.3 6.6 ± 0.9 
    Number of primary apical obliques  5.9 ± 1.6 
    Maximal field span of basals (μm) 210.5 ± 38.7 270.4 ± 31.7 
    Maximal field span of apical obliques (μm)  192.4 ± 52.9 
    Maximal field span of terminal tuft (μm)  286.8 ± 55.3 
    Main apical dendrite length (μm)  167.5 ± 70.5 
Distance from soma to apical oblique branch (μm) 
    1°  22.9 ± 11.2 
    2°  37.2 ± 14.0 
    3°  45.3 ± 13.9 
    4°  53.7 ± 10.7 
    5°  73.1 ± 14.9 
    6°  69.1 ± 17.6 
    7°  81.0 ± 20.0 
    8°  121.6 ± 18.0 
Axonal collaterals 
    Number of primary horizontal axons 4.1 ± 1.2 7.9 ± 2.5 
    Number of primary vertical axons 7.0 ± 2.2 5.4 ± 1.1 
    Length of single horizontal collaterals in L2/3  1627.5 ± 154.9 
    Length of single horizontal collaterals in L5  923.4 ± 199.6 
    Maximal field span of horizontal axons (μm) 422.6 ± 94.8 1729.7 ± 472.7 
    Maximal field span of vertical axons (μm) 400.9 ± 37.4 673.9 ± 178.3 
Quantity measured Projection (L4 spiny) cell Target (L2/3 pyramidal) cell 
Morphology (means ± SD) of somata and dendrites of the pre- and postsynaptic neurons of 14 synaptic connections between cells in layers 4 and 2/3. The morphology of the pre- and postsynaptic axon was determined from nine connections. Field span refers to the maximal horizontal distance between dendrites or axonal collaterals, respectively. The order of a dendrite is indicated by °. For example, 1° refers to the first branch originating from the soma or the main apical dendrite. For the apical oblique dendrites, starting from the 6° branch through to the 8° branch, the number of observations were, 6, 8 and 3, respectively. Only long-range horizontal collaterals were measured that were ≥500 μm in length which also showed no obvious truncations and growth cones at their tips. Measurements were taken at their origin from the main axon. 
Somata 
    Vertical diameter (μm) 12.9 ± 2.8 18.2 ± 2.5 
    Horizontal diameter (μm) 13.8 ± 2.4 15.6 ± 2.0 
Dendrites 
    Number of primary basals 4.3 ± 1.3 6.6 ± 0.9 
    Number of primary apical obliques  5.9 ± 1.6 
    Maximal field span of basals (μm) 210.5 ± 38.7 270.4 ± 31.7 
    Maximal field span of apical obliques (μm)  192.4 ± 52.9 
    Maximal field span of terminal tuft (μm)  286.8 ± 55.3 
    Main apical dendrite length (μm)  167.5 ± 70.5 
Distance from soma to apical oblique branch (μm) 
    1°  22.9 ± 11.2 
    2°  37.2 ± 14.0 
    3°  45.3 ± 13.9 
    4°  53.7 ± 10.7 
    5°  73.1 ± 14.9 
    6°  69.1 ± 17.6 
    7°  81.0 ± 20.0 
    8°  121.6 ± 18.0 
Axonal collaterals 
    Number of primary horizontal axons 4.1 ± 1.2 7.9 ± 2.5 
    Number of primary vertical axons 7.0 ± 2.2 5.4 ± 1.1 
    Length of single horizontal collaterals in L2/3  1627.5 ± 154.9 
    Length of single horizontal collaterals in L5  923.4 ± 199.6 
    Maximal field span of horizontal axons (μm) 422.6 ± 94.8 1729.7 ± 472.7 
    Maximal field span of vertical axons (μm) 400.9 ± 37.4 673.9 ± 178.3 
Table 2

Number and dendritic distribution of synaptic contacts established by axonal collaterals of L4 spiny neurons on L2/3 pyramidal cells

 Occurrence (% of total) Distance from soma (μm) Number of synaptic contacts 
Fourteen unidirectional synaptic connections were analysed. The number of putative light-microscopically identified synaptic contacts was 65. The percentage of the total number of synapses located on different dendritic sites and their distance from the soma are given as means ± SD. ° refers to the order of a dendritic branch. Basal 1° would be a dendrite arising from the soma, 1° oblique would be a dendrite arising from the main apical trunk. 
Basal 1° – – – 
Basal 2° 35.4 39.4 ± 16.0 23 
Basal 3° 41.6 75.6 ± 15.0 27 
Basal 4° 9.2 96.9 ± 12.2 
Apical oblique 1° – – – 
Apical oblique 2° 9.2 116.2 ± 21.8 
Apical oblique 3° 4.6 119.0 ± 66.9 
 Occurrence (% of total) Distance from soma (μm) Number of synaptic contacts 
Fourteen unidirectional synaptic connections were analysed. The number of putative light-microscopically identified synaptic contacts was 65. The percentage of the total number of synapses located on different dendritic sites and their distance from the soma are given as means ± SD. ° refers to the order of a dendritic branch. Basal 1° would be a dendrite arising from the soma, 1° oblique would be a dendrite arising from the main apical trunk. 
Basal 1° – – – 
Basal 2° 35.4 39.4 ± 16.0 23 
Basal 3° 41.6 75.6 ± 15.0 27 
Basal 4° 9.2 96.9 ± 12.2 
Apical oblique 1° – – – 
Apical oblique 2° 9.2 116.2 ± 21.8 
Apical oblique 3° 4.6 119.0 ± 66.9 
Table 3

Dimensions of 2D maps of dendrite and axon density

Location of measurement Barrel-centred map (μm) L2/3 pyramidal cell soma-centred map (μm) 
Two-dimensional axonal and dendritic ‘length density’ maps were generated from reconstructions of nine unidirectional synaptic connections (see Materials and Methods). Dimensions refer to the maximal vertical or horizontal distance (for layers 2/3 and 5) between the contour line including 80% of all axonal or dendritic length. Horizontal dimensions in layer 4 were determined at the barrel centre. 
Horizontal dimensions 
    L4 spiny neuron axonal arbors 
        in L4 400 381 
        in L2/3 487 447 
 L4 spiny neuron dendrite density 290  
    L2/3 pyramidal cell dendrite density 409 298 
    Predicted L4–L4 innervation density 251  
    Predicted L4–L2/3 innervation density 308 243 
    L2/3 pyramidal cell axonal arbors 
        in L2/3 1152 1113 
        in L4 501 649 
        in L5 795 740 
Vertical dimensions 
    L4 spiny neuron axonal arbors 942 940 
    L4 spiny neuron dendrite density 207  
    L2/3 pyramidal cell dendrite density 557 502 
    Predicted L4–L4 innervation density 186  
    Predicted L4–L2/3 innervation density 466 328 
    L2/3 pyramidal cell axonal arbors 1152 1097 
Average barrel dimensions 
    Horizontal 252  
    Vertical 229  
Location of measurement Barrel-centred map (μm) L2/3 pyramidal cell soma-centred map (μm) 
Two-dimensional axonal and dendritic ‘length density’ maps were generated from reconstructions of nine unidirectional synaptic connections (see Materials and Methods). Dimensions refer to the maximal vertical or horizontal distance (for layers 2/3 and 5) between the contour line including 80% of all axonal or dendritic length. Horizontal dimensions in layer 4 were determined at the barrel centre. 
Horizontal dimensions 
    L4 spiny neuron axonal arbors 
        in L4 400 381 
        in L2/3 487 447 
 L4 spiny neuron dendrite density 290  
    L2/3 pyramidal cell dendrite density 409 298 
    Predicted L4–L4 innervation density 251  
    Predicted L4–L2/3 innervation density 308 243 
    L2/3 pyramidal cell axonal arbors 
        in L2/3 1152 1113 
        in L4 501 649 
        in L5 795 740 
Vertical dimensions 
    L4 spiny neuron axonal arbors 942 940 
    L4 spiny neuron dendrite density 207  
    L2/3 pyramidal cell dendrite density 557 502 
    Predicted L4–L4 innervation density 186  
    Predicted L4–L2/3 innervation density 466 328 
    L2/3 pyramidal cell axonal arbors 1152 1097 
Average barrel dimensions 
    Horizontal 252  
    Vertical 229  
Figure 1.

Unidirectional synaptic connection between a L4 spiny neuron and a L2/3 pyramidal cell. (A) Low power bright field image of a pair of neurons that were filled with biocytin during recording. Note the typical asymmetric dendritic configuration of the presynaptic spiny stellate neuron that is mainly confined to layer 4 whereas the basal dendrites of the L2/3 pyramidal cell show a symmetric arrangement around the soma, and the apical dendrite forms a characteristic terminal tuft in layer 1. The main axons of both neurons could be followed towards the white matter. Scale bar, 100 μm. (BE) High power magnification of putative, light-microscopically identified synaptic contacts (circled areas) that were found exclusively on basal dendrites of different order, established by en passant axonal collaterals of the presynaptic spiny stellate neuron. (F) Autapses were found on the pre- and postsynaptic neurons that had a similar dendritic location as synaptic contacts. Scale bar in BF, 5 μm.

Figure 1.

Unidirectional synaptic connection between a L4 spiny neuron and a L2/3 pyramidal cell. (A) Low power bright field image of a pair of neurons that were filled with biocytin during recording. Note the typical asymmetric dendritic configuration of the presynaptic spiny stellate neuron that is mainly confined to layer 4 whereas the basal dendrites of the L2/3 pyramidal cell show a symmetric arrangement around the soma, and the apical dendrite forms a characteristic terminal tuft in layer 1. The main axons of both neurons could be followed towards the white matter. Scale bar, 100 μm. (BE) High power magnification of putative, light-microscopically identified synaptic contacts (circled areas) that were found exclusively on basal dendrites of different order, established by en passant axonal collaterals of the presynaptic spiny stellate neuron. (F) Autapses were found on the pre- and postsynaptic neurons that had a similar dendritic location as synaptic contacts. Scale bar in BF, 5 μm.

Figure 2.

Dendritic arborization and axonal projection of presynaptic L4 spiny neurons and dendrites of L2/3 pyramidal cells are confined to a barrel-column. (A, B) Camera lucida reconstructions of two synaptically coupled pairs of neurons. The dendritic arborizations of the presynaptic neurons are shown in red, those of the postsynaptic neurons in white and the axons of L4 spiny neurons in blue. Individual barrels are outlined in white. Note the eccentric location of presynaptic neurons within the barrel. The somatodendritic domains of the presynaptic neurons are mainly confined to layer 4. The axonal collaterals span all cortical layers with a preferential dense projection within layer 4 and 2/3. Note that the somatodendritic domain of both neurons and the axons of the presynaptic neurons are confined to a single barrel-column. The insets show the number and dendritic location of putative light-microscopically identified synaptic contacts (blue dots) established by the axonal collaterals of the L4 spiny cell. (C) Reconstruction of a pair of neurons where the postsynaptic pyramidal cell is located close to the pial surface. The postsynaptic pyramidal cell shows a slightly tangential orientation of the apical dendrite and terminal tuft. The presynaptic spiny stellate neuron is located not underneath the postsynaptic neuron, but shifted to the right suggesting that the soma and basal dendrites of the postsynaptic neuron are not located in the same barrel-column as the spiny cell. The inset shows that the axonal projection of the presynaptic neuron remains largely vertical yielding ‘unusual’ locations of putative, light-microscopically identified synaptic contacts (blue dots). (D) L4 star pyramidal cell–L2/3 pyramidal cell connection. In contrast to spiny stellate neurons, star pyramidal cells (red) have a prominent thick apical dendrite that terminates in layer 2/3 without forming a terminal tuft and symmetrically arranged basal dendrites. The axonal projection of star pyramidal cells is similar to that of spiny stellate neurons. The inset shows the number and dendritic location of putative, light-microscopically identified synaptic contacts located exclusively on basal dendrites relatively close to the soma.

Figure 2.

Dendritic arborization and axonal projection of presynaptic L4 spiny neurons and dendrites of L2/3 pyramidal cells are confined to a barrel-column. (A, B) Camera lucida reconstructions of two synaptically coupled pairs of neurons. The dendritic arborizations of the presynaptic neurons are shown in red, those of the postsynaptic neurons in white and the axons of L4 spiny neurons in blue. Individual barrels are outlined in white. Note the eccentric location of presynaptic neurons within the barrel. The somatodendritic domains of the presynaptic neurons are mainly confined to layer 4. The axonal collaterals span all cortical layers with a preferential dense projection within layer 4 and 2/3. Note that the somatodendritic domain of both neurons and the axons of the presynaptic neurons are confined to a single barrel-column. The insets show the number and dendritic location of putative light-microscopically identified synaptic contacts (blue dots) established by the axonal collaterals of the L4 spiny cell. (C) Reconstruction of a pair of neurons where the postsynaptic pyramidal cell is located close to the pial surface. The postsynaptic pyramidal cell shows a slightly tangential orientation of the apical dendrite and terminal tuft. The presynaptic spiny stellate neuron is located not underneath the postsynaptic neuron, but shifted to the right suggesting that the soma and basal dendrites of the postsynaptic neuron are not located in the same barrel-column as the spiny cell. The inset shows that the axonal projection of the presynaptic neuron remains largely vertical yielding ‘unusual’ locations of putative, light-microscopically identified synaptic contacts (blue dots). (D) L4 star pyramidal cell–L2/3 pyramidal cell connection. In contrast to spiny stellate neurons, star pyramidal cells (red) have a prominent thick apical dendrite that terminates in layer 2/3 without forming a terminal tuft and symmetrically arranged basal dendrites. The axonal projection of star pyramidal cells is similar to that of spiny stellate neurons. The inset shows the number and dendritic location of putative, light-microscopically identified synaptic contacts located exclusively on basal dendrites relatively close to the soma.

Figure 3.

Superposition of pairs of monosynaptically connected L4 spiny neurons and L2/3 pyramidal cells. Camera lucida reconstructions of synaptically connected L4 spiny neurons and L2/3 pyramidal cells (n = 9) are superimposed and aligned with respect to the barrel centre. The dendritic domains of the presynaptic neurons are shown in red, those of the postsynaptic neurons in white, the axons of L4 spiny neurons in blue. The average barrel in the centre is outlined in white; two neighbouring barrels are added symbolically.

Figure 3.

Superposition of pairs of monosynaptically connected L4 spiny neurons and L2/3 pyramidal cells. Camera lucida reconstructions of synaptically connected L4 spiny neurons and L2/3 pyramidal cells (n = 9) are superimposed and aligned with respect to the barrel centre. The dendritic domains of the presynaptic neurons are shown in red, those of the postsynaptic neurons in white, the axons of L4 spiny neurons in blue. The average barrel in the centre is outlined in white; two neighbouring barrels are added symbolically.

Figure 4.

Columnar confinement of the axonal arborization of L4 spiny neurons and the dendritic ‘receptive’ field of layer 2/3 pyramidal cells. (A) 2D maps of axonal (A1) and dendritic ‘length density’ (A2) of synaptically coupled L4 spiny neurons and L2/3 pyramidal cells (n = 9), aligned with respect to the centre of the barrel. The predicted innervation domain (A3) of L2/3 dendrites by L4 axons is given by the product of the L4 axonal density and the L2/3 dendritic density. Contours (thin lines) enclosing 80% of the integrated density are shown superimposed. Positions of L4 spiny neuron somata (red dots), L2/3 pyramidal cell somata (white triangles) and outlines of barrels (thick lines) are indicated symbolically. (B) 2D map of axonal and dendritic ‘length density’ of the same nine cell pairs but centred on the location of the L2/3 pyramidal cell somata (white triangles) of each pair of reconstructions. Note the two dendritic ‘receptive’ fields (B2) of which only one is contributing to the innervation domain (B3). Colour bars, 500 μm. For each panel, 100% refers to the peak ‘length density’ measured.

Figure 4.

Columnar confinement of the axonal arborization of L4 spiny neurons and the dendritic ‘receptive’ field of layer 2/3 pyramidal cells. (A) 2D maps of axonal (A1) and dendritic ‘length density’ (A2) of synaptically coupled L4 spiny neurons and L2/3 pyramidal cells (n = 9), aligned with respect to the centre of the barrel. The predicted innervation domain (A3) of L2/3 dendrites by L4 axons is given by the product of the L4 axonal density and the L2/3 dendritic density. Contours (thin lines) enclosing 80% of the integrated density are shown superimposed. Positions of L4 spiny neuron somata (red dots), L2/3 pyramidal cell somata (white triangles) and outlines of barrels (thick lines) are indicated symbolically. (B) 2D map of axonal and dendritic ‘length density’ of the same nine cell pairs but centred on the location of the L2/3 pyramidal cell somata (white triangles) of each pair of reconstructions. Note the two dendritic ‘receptive’ fields (B2) of which only one is contributing to the innervation domain (B3). Colour bars, 500 μm. For each panel, 100% refers to the peak ‘length density’ measured.

Figure 6.

Axonal arborization, dendritic configuration and innervation domain of L4 spiny neuron pairs. 2D maps of axonal (A) and dendritic ‘length density’ (B) of synaptically coupled L4 spiny neuron pairs (n = 5), aligned with respect to the centre of the barrel. The predicted innervation domain (C) of L4 dendrites by L4 axons is given by the product of the L4 axonal and dendritic densities. Contours (thin lines) enclosing 80% of the integrated density are shown superimposed. Positions of L4 spiny neuron somata (red dots in A) and the outline of a single average barrel (thick line) are indicated symbolically. Colour bars, 150 μm.

Figure 6.

Axonal arborization, dendritic configuration and innervation domain of L4 spiny neuron pairs. 2D maps of axonal (A) and dendritic ‘length density’ (B) of synaptically coupled L4 spiny neuron pairs (n = 5), aligned with respect to the centre of the barrel. The predicted innervation domain (C) of L4 dendrites by L4 axons is given by the product of the L4 axonal and dendritic densities. Contours (thin lines) enclosing 80% of the integrated density are shown superimposed. Positions of L4 spiny neuron somata (red dots in A) and the outline of a single average barrel (thick line) are indicated symbolically. Colour bars, 150 μm.

Figure 8.

Transcolumnar projection of long-range horizontal axonal collaterals of L2/3 pyramidal cells. (A) Camera lucida reconstruction of a pair of synaptically coupled neurons. Here the typical horizontal projections of axonal collaterals in layer 2/3 and 5 (green) of L2/3 pyramidal cells are shown receiving monosynaptic input from L4 spiny neurons. Same colour code as in Figure 2. The axonal arborizations of the L4 spiny neurons are omitted for clarity (same pair as illustrated in Fig. 2B). Note that the long-range horizontal axonal collaterals of the pyramidal cell have a maximal field span of more than 2 mm and thus are connecting two or three adjacent columns on each side of the ‘home’ column in which the somata of the pre- and postsynaptic neurons are located. (B) Camera lucida reconstructions of L2/3 pyramidal cells receiving monosynaptic input from layer 4 (n = 9) are superimposed and aligned with respect to the barrel centre (same cells as in Fig. 3). The dendritic domain of the presynaptic L4 spiny neurons is shown in red, that of the pyramidal cells in white and the pyramidal cell axons in green. Three average barrels are outlined in white. (C) 2D map of axonal ‘length density’ of the same L2/3 pyramidal cells aligned with respect to the centre of the barrel. The contour line (white) on the 2D map includes 80% of all L2/3 axonal density. Positions of L4 spiny neuron somata (red dots), L2/3 pyramidal cell somata (white triangles) and outlines of three average barrels (thick white) are indicated symbolically. Colour bar, 500 μm; 100% refers to the peak ‘length density’ measured.

Transcolumnar projection of long-range horizontal axonal collaterals of L2/3 pyramidal cells. (A) Camera lucida reconstruction of a pair of synaptically coupled neurons. Here the typical horizontal projections of axonal collaterals in layer 2/3 and 5 (green) of L2/3 pyramidal cells are shown receiving monosynaptic input from L4 spiny neurons. Same colour code as in Figure 2. The axonal arborizations of the L4 spiny neurons are omitted for clarity (same pair as illustrated in Fig. 2B). Note that the long-range horizontal axonal collaterals of the pyramidal cell have a maximal field span of more than 2 mm and thus are connecting two or three adjacent columns on each side of the ‘home’ column in which the somata of the pre- and postsynaptic neurons are located. (B) Camera lucida reconstructions of L2/3 pyramidal cells receiving monosynaptic input from layer 4 (n = 9) are superimposed and aligned with respect to the barrel centre (same cells as in Fig. 3). The dendritic domain of the presynaptic L4 spiny neurons is shown in red, that of the pyramidal cells in white and the pyramidal cell axons in green. Three average barrels are outlined in white. (C) 2D map of axonal ‘length density’ of the same L2/3 pyramidal cells aligned with respect to the centre of the barrel. The contour line (white) on the 2D map includes 80% of all L2/3 axonal density. Positions of L4 spiny neuron somata (red dots), L2/3 pyramidal cell somata (white triangles) and outlines of three average barrels (thick white) are indicated symbolically. Colour bar, 500 μm; 100% refers to the peak ‘length density’ measured.

References

Agmon A, Connors BW (
1991
) Thalamocortical responses of mouse somatosensory (barrel) cortex in vitro.
Neuroscience
 
41
:
365
–380.
Armstrong-James M, Fox K, Das-Gupta A (
1992
) Flow of excitation within rat barrel cortex on striking a single vibrissa.
J Neurophysiol
 
68
:
1345
–1358.
Bernardo KL, McCasland JS, Woolsey TA, Strominger RN (
1990
) Local intra- and interlaminar connections in mouse barrel cortex.
J Comp Neurol
 
291
:
231
–255.
Brecht M, Sakmann B (
2002
) Dynamic representation of whisker deflection by synaptic potentials in spiny stellate and pyramidal cells in the barrels and septa of layer 4 rat somatosensory cortex.
J Physiol (Lond)
 
543
:
49
–70.
Brett-Green BA, Chen-Bee CH, Frostig RD (
2001
) Comparing the functional representations of central and border whiskers in rat primary somatosensory cortex.
J Neurosci
 
15
:
9944
–9954.
Bruno RM, Simons DJ (
2002
) Feedforward mechanisms of excitatory and inhibitory cortical receptive fields.
J Neurosci
 
22
:
10966
–10975.
Burkhalter A (
1989
) Intrinsic connections of rat primary visual cortex: laminar organization of axonal projections.
J Comp Neurol
 
279
:
171
–186.
Chen-Bee CH, Frostig RD (
1996
) Variability and interhemispheric asymmetry of single-whisker functional representations in rat barrel cortex.
J Neurophysiol
 
76
:
884
–894.
Chmielowska J, Kossut M, Chmielowski M (
1986
) Single vibrissal cortical column in the mouse labeled with 2-deoxyglucose.
Exp Brain Res
 
63
:
607
–619.
DeFelipe J, Farinas I (
1992
) The pyramidal neuron of the cerebral cortex: morphological and chemical characteristics of the synaptic inputs.
Prog Neurobiol
 
39
:
563
–607.
Egger V, Feldmeyer D, Sakmann B (
1999
) Coincidence detection and efficacy changes in synaptic connections between spiny stellate neurons of the rat barrel cortex.
Nature Neurosci
 
2
:
1098
–1105.
Feldmeyer D, Egger V, Lübke J, Sakmann B (
1999
) Synaptic connections between excitatory layer 4 neurones in the ‘barrel field’ of rat somatosensory cortex.
J Physiol (Lond),
 
521
:
169
–190.
Feldmeyer D, Lübke J, Silver RA, Sakmann B (
2002
) Synaptic connections between layer 4 spiny neurone-layer 2/3 pyramidal cell pairs in juvenile rat barrel cortex: physiology and anatomy of interlaminar signalling within a cortical column.
J Physiol (Lond)
 
538
:
803
–822.
Gilbert CD, Wiesel TN (
1979
) Morphology and intracortical projections of functionally characterised neurones in the cat visual cortex.
Nature
 
280
:
120
–125.
Gottlieb JP, Keller A (
1997
) Intrinsic circuitry and physiological properties of pyramidal neurons in rat barrel cortex.
Exp Brain Res
 
115
:
47
–60.
Harris RM, Woolsey TA (
1983
) Computer-assisted analyses of barrel neuron axons and their putative synaptic contacts.
J Comp Neurol
 
220
:
63
–79.
Kalisman N, Silberberg G, Markram H (
2003
) Deriving physical connectivity from neuronal morphology.
Biol Cybern
 
88
:
210
–218.
Keller A, Carlson GC (
1999
) Neonatal whisker clipping alters intracortical, but not thalamocortical projections, in rat barrel cortex.
J Comp Neurol
 
412
:
83
–94.
Kleinfeld D, Delaney KR (
1996
) Distributed representation of vibrissa movement in the upper layers of somatosensory cortex revealed with voltage-sensitive dyes.
J Comp Neurol
 
375
:
89
–108.
Kossut M, Juliano SL (
1999
) Anatomical correlates of representational map reorganization induced by partial vibrissectomy in the barrel cortex of adult mice.
Neuroscience
 
92
:
807
–817.
Laaris N, Keller A (
2002
) Functional independence of layer IV barrels.
J Neurophysiol.
 
87
:
1028
–1034.
Laaris N, Carlson GC, Keller A (
2000
) Thalamic-evoked synaptic interactions in barrel cortex revealed by optical imaging.
J Neurosci
 
20
:
1529
–1537.
Larkum ME, Zhu JJ, Sakmann B (
1999
) A new cellular mechanism for coupling inputs arriving at different cortical layers.
Nature
 
398
:
338
–341.
Larkum ME, Zhu JJ, Sakmann B (
2001
) Dendritic mechanisms underlying the coupling of the dendritic with the axonal action potential initiation zone of adult rat layer 5 pyramidal neurons.
J Physiol (Lond)
 
533
:
447
–466.
Lorente de Nó R (
1922
) La corteza cerebral del ratón.
Trab Lab Invest Biol (Madrid)
 
20
:
41
–78.
Lorente de Nó R (
1943
) Cerebral cortex: architecture, intracortical connections, motor projections. In: Physiology of the nervous system (Fulton JF, ed.), pp. 274–313. London: OUP.
Lübke J, Egger V, Sakmann B, Feldmeyer D (
2000
) Columnar organization of single and synaptically coupled excitatory spiny neurons in layer 4 of the rat barrel cortex.
J Neurosci
 
20
:
5300
–5311.
Markram H, Lübke J, Frotscher M, Roth A, Sakmann B (
1997
) Physiology and anatomy of synaptic connections between thick tufted pyramidal neurones in the developing rat neocortex.
J Physiol (Lond)
 
500
:
409
–440.
Masino SA, Frostig RD (
1996
) Quantitative long-term imaging of the functional representation of a whisker in rat barrel cortex.
Proc Natl Acad Sci USA
 
93
:
4942
–4947.
McCasland JS, Woolsey TA (
1988
) High resolution 2-deoxyglucose mapping of functional cortical columns in mouse barrel cortex.
J Comp Neurol
 
278
:
555
–569.
Miller B, Blake NMJ, Erinjeri JP, Reistad CE, Sexton T, Admire P, Woolsey TA (
2001
) Postnatal growth of intrinsic connections in mouse barrel cortex.
J Comp Neurol
 
436
:
17
–31.
Moore CI, Nelson SB (
1998
) Spatio-temporal subthreshold receptive fields in the vibrissa representation of rat primary somatosensory cortex.
J Neurophysiol
 
80
:
2882
–2892.
Orbach HS, Cohen LB, Grinvald A (
1985
) Optical mapping of electrical activity in rat somatosensory and visual cortex.
J Neurosci
 
5
:
1886
–1895.
Petersen CCH, Sakmann B (
2000
) The excitatory neuronal network of rat layer 4 barrel cortex.
J Neurosci
 
20
:
7579
–7586.
Petersen CCH, Sakmann B (
2001
) Functionally independent columns of rat somatosensory barrel cortex revealed with voltage-sensitive dye imaging.
J Neurosci
 
21
:
8435
–8446.
Petersen CCH, Grinvald, A, Sakmann B (
2003
) Spatiotemporal dynamics of sensory responses in layer 2/3 of rat barrel cortex measured in vivo by voltage-sensitive dye imaging combined with whole-cell voltage recordings and neuron reconstructions.
J Neurosci
 
23
,
1298
–1309.
Reyes A, Sakmann B (
1999
) Developmental switch in the short-term modification of unitary EPSPs evoked in layer 2/3 and layer 5 pyramidal neurons of rat neocortex.
J Neurosci
 
19
:
3827
–3835.
Shepherd GMG, Pologruto TA, Svoboda K (
2003
) Circuit analysis of experience-dependent plasticity in the developing rat barrel cortex.
Neuron
 
38
:
277
–289.
Simons DJ, Woolsey TA (
1984
) Morphology of Golgi–Cox-impregnated barrel neurons in rat SmI cortex.
J Comp Neurol
 
230
:
119
–132.
Stepanyants A, Hof PR, Chklovskii DB (
2002
) Geometry and structural plasticity of synaptic connectivity.
Neuron
 
34
:
275
–288.
Trevelyan AJ, Jack J (
2002
) Detailed passive cable models of layer 2/3 pyramidal cells in rat visual cortex at different temperatures.
J Physiol (Lond)
 
539
:
623
–636.
Valverde, F (
1986
) Intrinsic neocortical organisation: some comparative aspects.
Neuroscience
 
18
:
1
–23.
Wallace H, Fox KD (
1999
) Local cortical interactions determine the form of cortical plasticity.
J Neurobiol
 
41
:
58
–63.
Woolsey TA (
1990
) Peripheral alteration and somatosensory development. In: Development of sensory systems in mammals (Coleman J, ed.), pp. 461–516. New York: Wiley-Liss.
Woolsey TA, van der Loos H (
1970
) The description of a cortical field composed of discrete cytoarchitectonic units.
Brain Res
 
17
:
205
–242.
Woolsey TA, Rovainen CM, Cox SB, Henegar MH, Liang GE, Liu D, Moskalenko YE, Sui J, Wei L (
1996
) Neuronal units linked to microvascular modules in cerebral cortex: response elements for imaging the brain.
Cereb Cortex
 
6
:
647
–660.
Yoshimura Y, Sato H, Imamura K, Watanabe Y (
2000
) Properties of horizontal and vertical inputs to pyramidal cells in the superficial layers of the cat visual cortex.
J Neurosci
 
20
:
1931
–1940.
Zhu JJ, Connors BW (
1999
) Intrinsic firing patterns and whisker-evoked synaptic responses of neurons in the rat barrel cortex.
J Neurophysiol
 
81
:
1171
–1183.