Abstract

The contribution and precise role of intracortical circuits in generating orientation tuned responses in visual cortical neurons is still controversial. To address this question, the relationship between excitatory and inhibitory synaptic connections and orientation maps in ferret striate cortex was investigated by combining in vivo optical imaging and in vitro scanning laser photostimulation. Excitatory and inhibitory inputs to pyramidal cells originated preferentially from regions with similar orientation preference. Prominent cross-orientation inhibition was not observed, arguing against cross-orientation models of orientation selectivity. The tuning of inhibitory inputs was significantly broader in both layer 2/3 and layer 5/6 pyramidal neurons compared to the tuning of excitatory inputs. Local excitatory inputs were more prominent in the 0–20° tuning difference range between pre- and postsynaptic cells than inhibitory inputs, whereas inhibition dominated in the 20–40° tuning difference range. These differences in tuning of excitatory and inhibitory inputs onto individual cells are consistent with the predictions of recurrent models of orientation selectivity.

Introduction

In the mammalian visual system, complex receptive field properties such as orientation and direction selectivity emerge for the first time in the primary visual cortex. The mechanisms generating orientation selectivity have been the subject of numerous experimental and theoretical studies. It was originally proposed (Hubel and Wiesel, 1962) that the spatial alignment of thalamic afferents in layer 4 could produce orientation tuning in simple cells, whose outputs were subsequently pooled to create complex receptive field properties. More recent models of orientation selectivity invoke contributions from intracortical connections in shaping orientation tuning in striate cortex. One group of models proposes that cross-orientation inhibitory inputs suppress non-preferred responses (Sillito, 1979; Sillito et al., 1980; Heggelund, 1981). Recurrent models, in contrast, assert that an initial orientation bias provided by thalamic afferents is amplified by local excitatory inputs of similar orientation, balanced by more broadly tuned inhibition (Woergoetter and Koch, 1991; Berman et al., 1991; Somers et al., 1995).

Each class of models makes different predictions regarding the tuning of local excitatory and inhibitory connections converging onto individual neurons. In vivo experiments investigating this issue have yielded conflicting results. Several investigations have demonstrated that both excitatory and inhibitory inputs are strongest at the preferred orientation of the postsynaptic neuron (Blakemore and Tobin, 1972; Nelson and Frost, 1978; Ferster, 1986, 1988; Nelson, 1991). Other studies, however, indicate the presence of inhibitory inputs tuned to cross-orientations (Sillito, 1975; Sillito et al., 1980; Morrone et al., 1982; Crook and Eysel, 1992; Pei et al., 1994). Thus, the orientation tuning of intracortical inhibition as compared to excitation remains controversial.

To help distinguish between current hypotheses, we combined in vivo optical imaging of orientation preference maps with in vitro scanning photostimulation of intracortical synaptic inputs (Roerig and Katz, 1998; Roerig and Kao, 1999) to determine the relative orientations of monosynaptic inhibitory and excitatory inputs to individual cortical neurons. We find that intracortical excitatory and inhibitory inputs to both layer 2/3 and deep layer pyramidal neurons preferentially originate from iso-orientation domains. However, the input tuning histograms for inhibitory inputs are significantly broader than the tuning histograms for excitatory inputs. The relationship between excitatory and inhibitory synaptic connectivity patterns is thus consistent with recurrent models of orientation selectivity, but does not support cross-orientation models.

Materials and Methods

Successful recordings were obtained in 27 slices from layers 2/3 and in 24 slices from layers 5/6. We recorded from 29 upper layer pyramidal cells and 21 deep layer pyramidal cells. Photostimulation maps of sufficient size to be included in summary histograms were obtained from 19 layer 2/3 cells and 17 deep layer neurons. Three layer 2/3 cells were excluded due to their location in the vicinity of an orientation center.

Cells were dialyzed with an internal solution containing 12 mM Cl, which sets the chloride reversal potential around –37 mV. Inhibitory and excitatory synaptic inputs were distinguished by shifting the holding potential from –60 to –20 mV (Katz and Dalva, 1994). Postsynaptic cells were filled with biocytin and cell types were identified based on the morphology of the dendritic tree and axonal projection patterns. All recordings included in this study were from pyramidal cells.

Optical Imaging of Orientation Preference Maps

Juvenile ferrets (P37-P48, Marshall Farms, New Rose, NY) were anesthetized with a mixture of Nembutal (40 mg/kg, i.m.) and xylazine hydrochloride (2 mg/kg, i.m.). To prevent occlusion of the trachea, 0.05 mg/kg atropine were injected s.c. Animals were intubated and ventilated with a mixture of 2:1 N2O/O2 and 1.5–2 % halothane during surgery. A craniotomy (8 × 12 mm) was made covering area 17 and adjacent parts of area 18. The dura was removed and a stainless steel chamber was mounted on the skull, filled with saline and sealed with a coverglass. During recording sessions animals were paralysed with pancuronium bromide (2 mg/kg, i.p.) to prevent eye movements and anesthesia was kept at 1:1 N2O/O2 and 0.5 % halothane. Expired CO2 was kept at 4–5 %. Pupils were dilated with 2% atropine and corneas protected with zero power contact lenses. Visual stimulation was provided monocularly through the contralateral eye. Images were acquired and amplified using an enhanced video acquisition system (Optical Imaging Inc., New York). The cortex was illuminated with red light (620 nm) and a 50 × 50 tandem lens combination; a CCD camera (Pentax) was used for imaging. The total size of the imaged region was 6.5 × 8 mm. Visual stimuli were presented at a distance of 30 cm. For imaging orientation domains, a high-contrast bar grating pattern was used (1.2° bar width, 6° bar spacing, 18°/s drift). Four pairs of orientations (0/90, 45/135, 22/112 and 67/157°) were imaged per animal. Forty trials were averaged per angle pair and the acquisition time per trial was 15 s. To enhance the signal-to-noise ratio, differential images were computed by subtraction of the responses to orthogonal orientations. Images were divided by a baseline image of the unstimulated cortex (blank) to correct for uneven illumination. The eight differential images were vector-summed to produce an angle map of orientation preference (Bonhoeffer and Grinvald, 1993).

Extracellular Single Unit Recording

Since most of the optically recorded signal is generated by neurons in layers 2/3 and 4 we controlled for consistency of orientation tuning in a cortical column using extracellular single unit recordings at different depths during penetrations made perpendicular to the cortical surface. To this end, the coverglass sealing the recording chamber was removed, the saline drained and the cortical surface was protected by a sheet of 50 μm thick silicone. Extracellular recording electrodes (16–18 MO, FHC, Brunswick, ME) were advanced through a hole in the silicone. Cortical pulsation was reduced by agar (2–4 %) applied over the silicone sheet. The visual stimulus was a single high-contrast bar (1.2° bar width, 40° length) drifting at 18 deg/s. Extracellular spike activity was recorded for nine different stimulus orientations and two opposite directions per orientation. Eight trials were averaged per direction of motion. Recording and wave-form discrimination were done using SPIKE2 software (Cambden Elektronic Design, Cambridge, UK). To determine orientation tuning curves, the peak discharge rate at each direction of motion was determined during a 1.0–1.5 s period as the stimulus crossed the central region of the receptive field. Background spontaneous activity was subtracted.

Following optical recording four to six small injections of either florescein or rhodamine conjugated latex microspheres (200–300 nl) were made in the imaged area using glass pipettes and a Picospritzer (General Valve, Fairfield, NJ). To facilitate alignment, the reference injections were made close to blood vessel branch points or into orientation domains. Following the injections the animal was deeply anesthetized with Nembutal (100 mg/kg) and the lateral gyrus containing areas 17 and 18 was completely removed.

Slice Preparation

The block of tissue removed from the animal was placed in chilled artificial cerebrospinal fluid (sucrose-ACSF, composition: 248 mM sucrose, 5 mM KCl, 5.3 mM KH2PO4, 1.3 mM MgSO4, 3.2 mM CaCl2, 10 mM dextrose, 25 mM NaHCO3, 1 mM kynurenic acid), oxygenated with a mixture of 95% O2 and 5% CO2, pH 7.4. Tangential slices (400 μm thickness) of primary visual cortex were prepared using a vibratome (Ted Pella Inc., Redding, CA). Dissections were made in artificial cerebrospinal fluid, chilled to 4°C. Three to four slices of different cortical layers were obtained per imaged hemisphere. Slices were maintained in an interface chamber at a temperature of 33°C and in an atmosphere of 95% CO2/5% O2 as previously described (Durack and Katz, 1996). Sucrose–ACSF was replaced with standard ACSF (composition: 125 mM NaCl, 5 mM KCl, 5.3 mM KH2PO4, 1.3 mM MgSO4, 3.2 mM CaCl2, 10 mM dextrose, 25 mM NaHCO3) after 1 h.

Photostimulation

Individual slices were transferred to a recording chamber mounted on the stage of an upright microscope (Zeiss Axioskop FS) and continuously superfused with ACSF at room temperature. Fluorescent bead marks were viewed using epifluorescence and either a rhodamine or fluorescein filter set (exciter G 546 nm, beam splitter FT 580, barrier LP 590 for rhodamine; 450–490 excitation, FT 510 dichroic mirror, LP 520 barrier filter for fluorescein; Zeiss). Bead injections were visible in the living slices and could directly be used to guide positioning of patch pipettes. Slice overview images were taken using a Sony XC-75 CCD camera and a SNAPPY (Play) video frame acquisition module in conjunction with SNAPPY software. Electrophysiological recordings from single neurons were performed using standard whole-cell, patch-clamp methods. The intracellular solution consisted of 110 mM d-gluconic acid, 110 mM CsOH, 11 mM EGTA, 10 mM CsCl, 1 mM MgCl2, 1 mM CaCl2, 10 mM HEPES, 1.8 mM GTP, 3 mM ATP, pH 7.2, and contained 0.5% N-(2-amino-ethyl)biotinamide (Neurobiotin, Molecular Probes, Eugene, OR). Voltage clamp recordings were conducted using an Axopatch 1D amplifier (Axon Instruments, USA). The holding potential was either –60 or –20 mV. Recordings were filtered at 1 kHz and digitized at 8 kHz. Series resistances ranged from 11 to 17 MΩ; a 30–50% compensation was usually achieved using the amplifier adjustments. Presynaptic inputs were stimulated using the scanning laser photostimulation approach detailed previously (Katz and Dalva, 1994). Slices were bathed in a 250 μM solution of γ-CNB caged glutamate (Molecular Probes). An argon ion laser (Coherent Enterprise 261) was used as a UV light source for local uncaging. The laser beam was focused into the slice preparation (spot size ~15 μm) through a 40× 1.3 NA oil immersion objective (Plan-Neofluar, Zeiss). The objective was attached to a motorized XY stage and moved within an oil droplet below the quartz glass bottom of the recording chamber. Opening of the external shutter, scanning of the laser beam and data acquisition were controlled by a National Instruments AD board (AT-MIO/AI E-10) and custom-written software. (Labview, National Instruments). The flash duration was 5 ms and the interstimulus interval 3–5 s. The acquisition period was 1 s, with the shutter opening occurring after 500 ms. The prestimulus acquisition period served to monitor spontaneous synaptic activity. Cells with high frequencies of spontaneous activity (>20 Hz) were discarded to avoid ‘contamination’ of the evoked signal. Photo-stimulation-evoked responses were analysed within the first 100 ms following uncaging. The spacing of stimulation sites was 50 or 100 μm. Typical maps consisted of 500–2000 stimulation sites; sampling was done in ‘four-pass’ mode, i.e. only every fourth spot was stimulated during one round and the map was filled in during four acquisition cycles. Maps were recorded at –20 mV to distinguish between inhibitory and excitatory inputs, at –60 mV to allow for a larger driving force for small excitatory events and in the presence of tetrodotoxin (TTX, 2 μM) to distinguish between dendritic direct activation and local synaptic inputs. Only one or two cells were recorded per slice to facilitate alignment and unambiguous assigning of input maps to postsynaptic cells.

Histology

Following recording, slices were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS, pH 7.4) for subsequent histological processing of neurobiotin-filled cells. Slices were resectioned at 70 μm on a freezing microtome and labeled cells were visualized by standard immunoperoxidase staining techniques (Durack and Katz, 1996). No heavy metal intensification was performed in order to preserve the fluorescent alignment markers in the histological sections.

Alignment of Orientation Preference Maps and Synaptic Input Maps

Alignment of orientation maps obtained in vivo and photostimulation maps were guided by the fluorescent bead injections. The in vivo images, the video image of the living slice and the histological section were overlaid with the photostimulation map using the layer menu of Adobe Photoshop. The position of the stained postsynaptic cell in the histological section and the direct activation area in the photostimulation maps served as additional markers. No other transformations but linear scaling and rotation were used. To determine orientation and direction tuning differences between pre- and postsynaptic sites, functional maps recorded in vivo were aligned with the synaptic input map determined by photostimulation in vitro (Fig. 5). To this end, linear scaling and rotation was applied to the two images until the bead marks were at least 50% overlapping. The bead marks were 100–200 μm in diameter, which resulted in a maximum alignment error of 75 μm. This corresponds to an orientation preference error of <10° (Weliky et al., 1995), which is not critical for cells located in orientation domains, where orientation preference changes smoothly.

The curved surface of ferret V1 can impair a layer-specific analysis of the synaptic input patterns. We have taken several precautions to reduce this problem: (i) the effect of curvature is most prominent in the lateral regions of the cortex, we have therefore restricted both our bead injections and the recording sites in slices to the more central regions of V1; (ii) laminar boundaries can be seen in cresyl violet stained sections prepared from the recorded slices. We have discarded recordings from slices that showed clear laminar boundaries.

Analysis

The analysis of photostimulation data was partially automated. Synaptic events were either completely manually analysed or manually selected and entered into a second Labview program. Sign, amplitude, latency and number of postsynaptic events within 100 ms following the laser flash were determined for each stimulation site. Plots of synaptic input patterns were generated using Transform (Fortner). Synaptic input maps were then superimposed on the orientation angle map using the layer menu of Adobe Photoshop. For each site giving rise to a synaptic input, as well as for the location of the postsynaptic cell, the orientation value was calculated as the mean of four pixels. In all figures the number of sites generating an excitatory or inhibitory input has been used, not the number of individual excitatory postsynaptic currents (EPSCs) or inhibitory postsynaptic currents (IPSCs) generated by one stimulation site.

The orientation difference between location of the postsynaptic cell and the site of origin of synaptic inputs was calculated for each event and the difference values were used to generate orientation tuning histograms. In addition, the distance between pre- and postsynaptic site was measured for each individual event. The assignment of location of origin of synaptic input to the corresponding location on the orientation map as well as the calculation of the tuning differences were not automated. Tuning deviation (0 ± 90°) can be in either direction; distributions shown were rectified unless otherwise stated. Events were binned into 10° categories. Data from all recorded neurons were pooled to generate the histograms shown in Figures 6 and 7.

To determine the spatial distribution of inhibitory and excitatory inputs, the ‘input density’ was calculated as number of sites generating a postsynaptic response per mm2. Calculations were made for concentric zones spaced 50 μm apart. Plotting of histograms as well as statistical comparison of distributions were done using the programs Sigmastat and Sigmaplot (Jandel) and Origin (Microcal).

Results

Orientation Tuning is Constant Throughout Vertical Columns in Ferret Visual Cortex

In this study we have aligned optical imaging maps obtained in vivo with synaptic input maps obtained in vitro, in order to determine the orientation specificity of intracortical synaptic connections. The optical signal mainly derives from layers 2/3 and 4. To be able to align slices obtained from cortical layers 5/6 with the optical imaging map with confidence, it was necessary to control for constancy of orientation tuning during vertical penetrations. To this end, we have performed single unit recordings (n = 123) in four different animals. Single units were recorded in a total of 27 penetrations made perpendicular to the cortical surface (Fig. 1DF). Although we used rather young animals in this study, visual response properties as well as functional architecture as determined with optical imaging techniques are mature at this age range in the ferret (Weliky et al., 1997). Our single unit recordings also show sharp orientation tuning in layers 2/3 and 5/6 between P37 and P58. Although receptive field properties may still mature further past this period, the majority of intracortical circuits underlying these properties appear to be already established. The majority of neurons in all layers of ferret visual cortex, including layer 4, were orientation tuned and either direction tuned or direction biased. To assess the constancy of orientation tuning within a vertical column we have plotted the orientation tuning difference between adjacent recording sites (Fig. 1E), as well as between the most superficial and the deepest recording site in each penetration (Fig. 1F). Within a vertical penetration, orientation preference did not change considerably (Fig. 1E,F). We therefore felt confident to align slices derived from the deep layers with the optical imaging map.

To examine the relationships between orientation preference maps and synaptic inputs onto individual cells, we used intrinsic signal imaging in the visual cortex of juvenile ferrets (n = 16, age range postnatal days 37–48) to generate orientation preference maps. Following the imaging session, red or green fluorescent latex microspheres (Katz and Iarovici, 1990) were injected into the imaged area to guide subsequent alignment of orientation maps with photostimulation maps and histological sections (Fig. 1). Following removal of the visual cortex and preparation of tangential slices, whole-cell, patch-clamp recordings were established onto individual neurons. Scanning laser photostimulation with caged glutamate was then used to assess the spatial patterns of monosynaptic excitatory and inhibitory connections converging onto the recorded cell (Katz and Dalva, 1994). To avoid confusion, no more than one or two cells were recorded per slice.

Following photostimulation mapping, slices were fixed and processed for biocytin to reveal the position and morphology of recorded neurons. Using the microsphere injections as fiducial marks, the in vivo maps were aligned with the brain slice (see Materials and Methods for details). The putative orientation selectivity of recorded neurons and their excitatory and inhibitory inputs were determined by reference to the in vivo orientation map. For most data analysis, we calculated the absolute value of the difference between the recorded neuron's orientation preference and that of its individual synaptic inputs, a value that varied between 0° (iso-orientation inputs) and 90° (orthogonal inputs). The number and sign (inhibitory or excitatory) of these synaptic inputs were used to construct orientation histograms for excitatory and inhibitory inputs (see below).

Spatial Organization of Excitatory and Inhibitory Inputs

We recorded maps of excitatory and inhibitory inputs from a total of 36 cells in both layer 2/3 (n = 19) and layer 5/6 (n = 17). On average, the maps covered 1.5 mm2 and consisted of 800–2400 stimulated sites. By recording at a holding potential of –20 mV, we could assess the relative contributions of excitatory and inhibitory inputs originating from the photostimulated sites. At a holding potential of –20 mV, excitatory inputs produced inward deflections of the current trace, while inhibitory inputs produced outward deflections (Fig. 2C). Close to the cell body, direct activation of glutamate receptors on neuronal dendrites led to large inward currents that persisted in the presence of TTX, which blocks action potential mediated activity. Because these large currents obscured the smaller synaptic inputs, we excluded the zone within 50 μm of the cell body from further analysis.

In both layers 2/3 and 5/6, the vast majority of excitatory and inhibitory inputs originated from a zone within a 500 μm radius of the cell body. To estimate the density of synaptic inputs as a function of distance from the recorded cells, we counted the number of stimulation sites that generated an inhibitory or excitatory synaptic response in 50 μm concentric rings centered a the cell body and calculated the density of inputs per mm2 (Fig. 3). In layer 2/3, ~90% of the excitatory and inhibitory inputs were evoked from the region <500 μm from the recorded cell (Fig. 3A,B); in layer 5/6, 87% of excitatory and 82% of inhibitory inputs originated from this region (Fig. 3C,D). Thus, the overwhelming majority of intracortical synaptic connections are highly local. However, in some neurons, excitatory synaptic inputs could be evoked from up to 2 mm away from the recorded neuron and were occasionally found in clusters (Fig. 4), consistent with previous reports (Katz and Dalva, 1994). Inhibitory inputs rarely originated from regions >1 mm from the recorded cell and were not clustered. The only obvious differences in the input patterns between the supragranular and infragranular layers were a lower overall density of both excitatory and inhibitory inputs in the infragranular layers (Fig. 3). The lower synaptic input density is probably due simply to the lower overall cell density in layer 5, while the somewhat longer-range inhibition observed in some deep layer neurons may result from a population of inhibitory interneurons, such as basket cells, with longer axon collaterals.

Orientation Specificity of Intracortical Synaptic Connections

Our study was mainly conducted to clarify which models of orientation selectivity are in line with the orientation specificity of intracortical synaptic connections. To determine the orientation specificity of intracortical synaptic inputs, we have aligned a map of locations of origin of synaptic inputs to individual cells with an orientation preference map of the scanned area obtained in vivo. Since the average size of orientation domains in ferret visual cortex is ~500 μm, a fraction of local inputs will inevitably originate from iso-orientation domains. All neurons included in this study were located within orientation domains., i.e. the iso-orientation tuning of their very local connections is also determined by their location on the map. We have recorded from three neurons located close to an orientation singularity. To reduce the heterogeneity of our sample we have excluded these cells from summary diagrams.

Both local and long-range (horizontal) excitatory connections were iso-orientation tuned in layers 2/3 and layers 5/6 (Fig. 6). Inhibitory inputs showed a broader tuning in both upper and lower layers (Figs 6C,F and 8). The difference in tuning width of local excitatory and inhibitory inputs to cortical neurons is an important parameter in models designed to explain the contribution of intracortical synaptic inputs to the emergence of orientation tuning in the primary visual cortex (Somers, 1995). To address this, the percentages of EPSCs and IPSCs falling into each orientation difference category (0–10 to 80–90° orientation tuning difference) have been calculated for each individual neuron and for the entire sample of layer 2/3 and layer 5/6 neurons (Fig. 6).

We have calculated the half width at half height for each EPSC and IPSC's distribution from Gaussian fits to the distributions (Fig. 8). The half width at half height of the input tuning curves yielded an average value of 18.8° (SEM 1.9, n = 19) for EPSCs and 36.8° (SEM 1.6, n = 17) for IPSCs in layer 2/3 pyramidal cells, in layers 5/6 the values were 23.7° (SEM 1.4, n = 17) for excitatory inputs and 39.6° (SEM 1.5, n =15) for inhibitory inputs. The half width at half height of EPSC and IPSC distributions were statistically compared (paired t-test). IPSC distributions were significantly broader (P < 0.001) both in layers 2/3 and in layers 5/6 (Fig. 8).

The efficacy of a synaptic input related to its impact on tuning properties is not only determined by its specificity, but also by its amplitude. To determine whether iso-orientation tuned inputs are also the strongest ones, we have plotted the average amplitude of EPSCs and IPSCs recorded from each neuron against the orientation tuning difference between pre- and post- synaptic cells. Summary histograms are shown in Figure 7. In most neurons the iso-orientation tuned inputs were also the strongest ones; however, the average amplitudes of long-range EPSCs and IPSCs in the upper layers tended to be more evenly distributed among orientation difference ranges than local EPSCs (Fig. 7). We have also multiplied the event counts by the mean event amplitude in each tuning difference category for each neuron and applied a paired t-test to each category. Only events recorded within a 500 μm radius around the postsynaptic cell were included. This analysis revealed a statistically significant dominance of local excitation over inhibition in the tuning difference ranges of 0–10° and 10–20° for both layer 2/3 (P < 0.002, n = 17, paired t-test) and layer 5/6 neurons (P < 0.001, n = 15). In the tuning difference range of 20–30 and 30–40°, inhibition was stronger than excitation (P < 0.001) in all layers. These results strongly support recurrent models of orientation selectivity (Somers, 1995; Sengpiel et al., 1997).

Comparison of Tuning Histogram Shapes with ‘Random Inputs’

Our data indicate that excitatory synaptic connections are, to a large extent, specifically formed between neurons of similar orientation selectivity. However, theoretically an orientation bias of input tuning curves can be either due to an actual specificity of synaptic connections between iso-orientation tuned neurons or due to an organizational feature of the orientation map per se. To distinguish between these possibilities we have constructed artificial tuning histograms of ‘simulated inputs spaced 50–100 μm apart (Fig. 9). The orientation tuning difference of these sites and the positions of actual postsynaptic cells (n = 11, six layer 2/3 cells, five layer 5/6 neurons) was calculated and the orientation tuning histograms constructed. Although these histograms are centered on iso-orientation, they are considerably broader than the distributions for excitatory inputs to layer 2/3 neurons (average half width at half height: 49.1°, SEM 18.4°, n = 11). The tuning of actual excitatory inputs to layer 2/3 and layer 5/6 pyramidal neurons was statistically significantly sharper than the tuning of ‘artificial’ inputs in both the 500 μm range (P < 0.001, Student's t-test) and the 500–1500 μm range (P < 0.03, Student's t-test).

Discussion

Methodological Considerations

The optical signal recorded during the in vivo part of our experiments derives mainly from cortical layers 2/3 and 4; the contribution of the deep layers is very low. In order to be able to align slices from the deep cortical layers with the optical imaging maps, we had to ascertain that there were no significant changes in orientation tuning between upper and lower cortical layers. To this end, we have performed single unit recordings through all layers during vertical penetrations through area 17. Our recordings showed no evidence for significant changes in orientation tuning between upper and lower cortical layers. We thus feel confident to align synaptic input maps derived from layer 5/6 neurons with the orientation preference maps.

In the in vitro photostimulation part of our experiments we have used the holding potential to discriminate between glutamatergic EPSCs and GABAA receptor mediated IPSCs. Our method does not allow for detection of GABAB receptor mediated inhibitory synaptic inputs, i.e. we can only draw conclusions regarding the orientation specificity of GABAA receptor mediated inhibitory synaptic inputs. However, our preliminary pharmacological experiments (B. Chen and B. Roerig, unpublished observation) conducted in tangential slices obtained from ferrets at P32–P46 do not show any significant contribution of GABAB receptor mediated synaptic currents to photo-stimulation-evoked responses.

Comparison with Other Species

The majority of studies addressing receptive field properties in the primary visual cortex, in particular orientation selectivity, have been conducted in cats. However, the pigmented ferret is becoming an increasingly popular species since it is hardier, recordings are more stable, especially in young animals and, due to its premature birth compared to the cat, the ferret is more suitable for developmental studies. The development of orientation tuning and the percentages of tuned neurons in the adult cortex are similar for cat and ferret (Chapman and Stryker, 1993). We have used the ferret as a model system in this study since (a) the smaller size of the primary visual cortex as compared to the cat facilitates large-scale mapping studies and makes the ferret an ideal model system for combined in vivo–in vitro studies and (b) due to its premature birth the ferret is more suitable for future developmental studies.

Orientation Tuning of Intracortical Synaptic Connections

A number of current models proposed to explain the initial emergence of orientation tuning in the primary visual cortex assume a prominent role for intracortical synaptic connections, in particular inhibitory connections. The goal of our study was to analyse the orientation tuning of excitatory and inhibitory intracortical connections to clarify how the specificity and spatial organization of intracortical circuits fits into current models of orientation selectivity. In general, the orientation tuning of both excitatory and inhibitory synaptic inputs to pyramidal neurons showed a strong bias towards iso-orientation. This was observed in both upper and lower cortical layers.

It was originally proposed (Hubel and Wiesel, 1962) that the spatial alignment of LGN afferents with on/off center surround receptive field structure accounts for the emergence of a preferred axis of orientation in first-order cortical neurons, i.e. layer 4 simple cells. This simple model does not require intracortical synaptic inputs. Two recent studies corroborate this model: inactivation of the intracortical circuitry by cooling (Ferster et al., 1996) or electrically evoked inhibition (Chung et al., 1998) does not significantly affect orientation selectivity in simple cells. However, cooling of the cortex also reduces the efficacy of the thalamocortical input, thus resulting in a pre-selection for the strongest input. Using intracortical inhibition to inactivate cortical circuits recruits the very synapses that under control conditions may contribute to the emergence or refinement of orientation tuning, leaving a potential contribution of intracortical circuits unresolved.

Other models put forward to explain the emergence of orientation selectivity in V1 propose a role for intracortical inputs. Some groups have reported tremendous alterations of tuning properties following removal of intracortical inhibition (Sillito, 1975, 1980; Morrone et al., 1982; Crook et al., 1992, 1996), indicating a prominent role for intrinsic inhibitory synaptic connections. How intracortical inhibition exactly operates to create or sharpen orientation tuning, however, remains to be resolved. Current models of orientation selectivity mainly differ in their assumptions about the properties of intracortical inhibition. Intracortical inhibition could itself be orientation selective and thus specifically suppress orthogonally tuned excitation, as proposed in the cross-orientation model of orientation selectivity (Sillito, 1979, 1980; Heggelund, 1981). Cross-orientation models (Sillito, 1979, 1980; Heggelund, 1981) postulate that inhibitory connections arising from regions preferring opposite stimulus orientations suppress excitatory inputs originating from cross-orientation domains. Alternatively, the excitatory input could have some orientation selectivity and inhibition could provide a threshold nonlinearity restricting the range of orientations over which the cell will fire (recurrent models of orientation selectivity). Recurrent models assume isoorientation preference for excitatory inputs and iso-orientation centered, but broader tuning curves for intracortical inhibition (Blakemore and Tobin, 1972; Hata et al., 1991; Berman et al., 1991; Somers et al., 1995). The broadness of inhibitory tuning required by recurrent models is in apparent conflict with some experimental results. Inhibitory synaptic potentials recorded in vivo in cat visual cortex simple and complex cells were well tuned for the same orientation as excitatory inputs to the same neuron (Ferster, 1986; Douglas et al., 1991). On the other hand, there is also evidence for non-iso-orientation tuned inhibitory inputs to first-order cells in cat area 17 in vivo, which have been hypothesized to contribute to time-dependent sharpening of orientation tuning (Morrone et al., 1982; Crook et al., 1992, 1997; Pei et al., 1994). In addition, studies reporting the presence of cross-orientation inhibition usually show broad tuning width for inhibition rather than sharp tuning for non-iso orientations (Morrone et al., 1982; Crook et al., 1992, 1997).

In summary, the intracortical connectivity patterns underlying the generation of orientation tuning are still not clear. Our combined in vivo-in vitro approach allows the analysis of the circuit patterns that may create these properties. In experiments using electrical stimulation (Weliky et al., 1995) there is often no clear distinction between mono- and polysynaptic connections. Polysynaptic activation of both long-range and locally projecting inhibitory interneurons is likely to occur in these experiments. Hence, the location of GABAergic interneurons relative to the orientation map is not known. Photostimulation, on the other hand, mainly activates monosynaptic inputs. A fraction of the orientation specific inhibitory inputs observed both in vivo and in vitro may result from polysynaptic activation of local inter-neurons. In this scenario interneurons would not have to be located in exactly the same orientation domain as the postsynaptic cell and still may mediate disynaptic, orientation-specific inhibition. Excitatory horizontal connections arising from iso-orientation domains may recruit GABAergic interneurons from a relatively broad annulus around a given cell. This mechanism would explain the functional iso-orientation tuning of both excitatory and inhibitory inputs. However, both theoretical and experimental studies indicate that even broadly tuned circular inhibition can result in sharpening of orientation tuning and in the emergence of direction selectivity (Blakemore and Tobin, 1972; Woergoetter et al., 1991). Our study corroborates this: there was no evidence of a predominance of cross-orientation tuned inhibitory inputs in any of the recorded pyramidal cells, although a fraction (up to 15–20 % in some cells) of inhibitory inputs were frequently evoked from domains differing in orientation preference by 80–90° from the post-synaptic neuron. The sum tuning histograms for inhibitory inputs, however, were biased towards iso-orientation. The tuning width of inhibitory input distributions was significantly broader than the tuning width of excitatory inputs. This was the case for both upper and lower layer neurons. Thus, the predominant role of intracortical inhibition may be the thresholding of excitation rather than providing an orientation specific input, in line with recurrent models of orientation selectivity (Somers, 1995; Sengpiel et al., 1997).

Notes

We would like to thank Larry Katz for his continuous support and for making significant contributions to the preparation of this manuscript. We further thank Darin Nelson for designing the acquisition and analysis software for photostimulation experiments. We also wish to thank Megan Gray for excellent histology. This work was supported by a Human Frontier Science Program postdoctoral fellowship (B.R.).

Figure 1.

(AC) Schematic of the strategy used to align in vivo functional maps with the patterns of synaptic inputs determined by scanning photostimulation in vitro. (A) Blood vessel reference image showing the location of the fluorescent bead injection sites which served as reference marks to guide the alignment. (B) Orientation map recorded in vivo from the cortical area shown in (A). Following optical recording, the imaged hemisphere was removed and tangential slices of the imaged portion of visual cortex were prepared. The fluorescent bead injection sites are visible in the living slice preparation and aided the choice of recording and stimulation sites. Standard whole-cell recordings were established and presynaptic inputs mapped using scanning photostimulation (see Materials and Methods). (C) Histological section (70 μm thickness). The rhodamine conjugated latex beads are visible even in standard brightfield transillumination. The black arrow marks the position of the biocytin-filled postsynaptic cell. Alignment of the photostimulation map with the orientation map was achieved by positioning the bead marks in the histological sections over the marks in the stored video images of the slice preparation and in the orientation map. (D) Orientation and direction selectivity of cortical neurons determined by extracellular single unit recordings. In this study we have aligned photostimulation maps obtained from neurons located in different cortical layers with optically recorded orientation maps. Since most of the optical signal is derived from cortical layers 2/3 and 4, we made extracellular single unit recordings at different depths along electrode penetrations perpendicular to the pial surface to control for invariant orientation tuning in upper and lower cortical layers. The visual stimulus was a single high-contrast bar (1.2° bar width, 40° length, speed: 18 deg/s) presented at nine different orientations. The bar moved in two opposite directions orthogonal to the stimulus orientation. Diagrams show orientation tuning curves for five units recorded at increasing cortical depth. The y-axis shows the average firing rate. The x-axis shows the 18 different directions of motion (0–340°). Bar orientation was always perpendicular to the direction of motion. The two peaks in the histograms correspond to the preferred orientation presented at two opposite directions of movement. Orientation preference (here, horizontal) did not change across cortical layers. The incidence of direction selective units increased with cortical depth. Sample recordings shown here were obtained from a P42 ferret. (E) Quantitative plot showing the direction tuning difference between adjacent unit pairs recorded in 68 different vertical penetrations. (F) Plot of the orientation tuning difference between the most superficial and the deepest recording site in each of the 68 penetrations. The quantitative analysis clearly shows that orientation tuning varies very little between units throughout all cortical layers.

Figure 1.

(AC) Schematic of the strategy used to align in vivo functional maps with the patterns of synaptic inputs determined by scanning photostimulation in vitro. (A) Blood vessel reference image showing the location of the fluorescent bead injection sites which served as reference marks to guide the alignment. (B) Orientation map recorded in vivo from the cortical area shown in (A). Following optical recording, the imaged hemisphere was removed and tangential slices of the imaged portion of visual cortex were prepared. The fluorescent bead injection sites are visible in the living slice preparation and aided the choice of recording and stimulation sites. Standard whole-cell recordings were established and presynaptic inputs mapped using scanning photostimulation (see Materials and Methods). (C) Histological section (70 μm thickness). The rhodamine conjugated latex beads are visible even in standard brightfield transillumination. The black arrow marks the position of the biocytin-filled postsynaptic cell. Alignment of the photostimulation map with the orientation map was achieved by positioning the bead marks in the histological sections over the marks in the stored video images of the slice preparation and in the orientation map. (D) Orientation and direction selectivity of cortical neurons determined by extracellular single unit recordings. In this study we have aligned photostimulation maps obtained from neurons located in different cortical layers with optically recorded orientation maps. Since most of the optical signal is derived from cortical layers 2/3 and 4, we made extracellular single unit recordings at different depths along electrode penetrations perpendicular to the pial surface to control for invariant orientation tuning in upper and lower cortical layers. The visual stimulus was a single high-contrast bar (1.2° bar width, 40° length, speed: 18 deg/s) presented at nine different orientations. The bar moved in two opposite directions orthogonal to the stimulus orientation. Diagrams show orientation tuning curves for five units recorded at increasing cortical depth. The y-axis shows the average firing rate. The x-axis shows the 18 different directions of motion (0–340°). Bar orientation was always perpendicular to the direction of motion. The two peaks in the histograms correspond to the preferred orientation presented at two opposite directions of movement. Orientation preference (here, horizontal) did not change across cortical layers. The incidence of direction selective units increased with cortical depth. Sample recordings shown here were obtained from a P42 ferret. (E) Quantitative plot showing the direction tuning difference between adjacent unit pairs recorded in 68 different vertical penetrations. (F) Plot of the orientation tuning difference between the most superficial and the deepest recording site in each of the 68 penetrations. The quantitative analysis clearly shows that orientation tuning varies very little between units throughout all cortical layers.

Figure 2.

Sample recordings from a layer 5 pyramidal neuron. Age of animal: P38. (A) Overview of the photostimulation map. The spacing between stimulation sites was 50 μm. In this example, 608 sites were stimulated. The postsynaptic neuron is located in the upper left (position 1). The numbers indicate the positions that gave rise to the responses shown in (C). (B) Video image of the biocytin-filled postsynaptic neuron taken from the histological section. (C) Sample traces of photostimulation-evoked postsynaptic responses evoked from the different locations indicated on the map in (A). Photostimulation maps were recorded first at a holding potential of –20 mV to distinguish between excitatory and inhibitory synaptic inputs, second, at –60 mV to allow for a greater driving force for small excitatory inputs that might not be detected at –20 mV and, third, in the presence of tetrodotoxin (TTX) to distinguish very local excitatory inputs from direct activation of the postsynaptic cell's dendritic tree. Position 1 represents the location of the postsynaptic cell body. The first panel shows traces recorded at –20 mV. The early, fast inward current represents the direct activation of the cell, which is followed by large outward currents representing strong, local inhibitory connections. The following panel shows recordings following stimulation at the same positions at –60 mV. Both direct activation and GABAergic inputs are inward-going. The last panel shows traces recorded in the presence of 2 μM TTX, which removes the synaptic input and shows the direct glutamate response of the postsynaptic neuron in isolation. Panels 2–5 show synaptic responses evoked from the indicated positions in the map. In panel 3, recordings at both 20 and –60 mV are shown to illustrate the effect of the increasing driving force at –60 mV on these small excitatory synaptic inputs. (D,E) Reliability of photostimulation approach to evoke synaptic inputs. In (D), 10 excitatory responses evoked successively from the same stimulation site are shown (interstimulus interval, 2 s). In (E), 10 inhibitory responses evoked successively from the same stimulation site are shown (interstimulus interval, 2 s). Both response amplitudes and latencies are fairly constant, indicating reliable stimulation of monosynaptic inputs (vertical line, time of shutter opening).

Figure 2.

Sample recordings from a layer 5 pyramidal neuron. Age of animal: P38. (A) Overview of the photostimulation map. The spacing between stimulation sites was 50 μm. In this example, 608 sites were stimulated. The postsynaptic neuron is located in the upper left (position 1). The numbers indicate the positions that gave rise to the responses shown in (C). (B) Video image of the biocytin-filled postsynaptic neuron taken from the histological section. (C) Sample traces of photostimulation-evoked postsynaptic responses evoked from the different locations indicated on the map in (A). Photostimulation maps were recorded first at a holding potential of –20 mV to distinguish between excitatory and inhibitory synaptic inputs, second, at –60 mV to allow for a greater driving force for small excitatory inputs that might not be detected at –20 mV and, third, in the presence of tetrodotoxin (TTX) to distinguish very local excitatory inputs from direct activation of the postsynaptic cell's dendritic tree. Position 1 represents the location of the postsynaptic cell body. The first panel shows traces recorded at –20 mV. The early, fast inward current represents the direct activation of the cell, which is followed by large outward currents representing strong, local inhibitory connections. The following panel shows recordings following stimulation at the same positions at –60 mV. Both direct activation and GABAergic inputs are inward-going. The last panel shows traces recorded in the presence of 2 μM TTX, which removes the synaptic input and shows the direct glutamate response of the postsynaptic neuron in isolation. Panels 2–5 show synaptic responses evoked from the indicated positions in the map. In panel 3, recordings at both 20 and –60 mV are shown to illustrate the effect of the increasing driving force at –60 mV on these small excitatory synaptic inputs. (D,E) Reliability of photostimulation approach to evoke synaptic inputs. In (D), 10 excitatory responses evoked successively from the same stimulation site are shown (interstimulus interval, 2 s). In (E), 10 inhibitory responses evoked successively from the same stimulation site are shown (interstimulus interval, 2 s). Both response amplitudes and latencies are fairly constant, indicating reliable stimulation of monosynaptic inputs (vertical line, time of shutter opening).

Figure 3.

Spatial distribution of excitatory and inhibitory synaptic inputs to upper and lower layer pyramidal neurons. To estimate the spatial extent of local and long-range synaptic connections in our preparation, we determined the density of synaptic inputs evoked by photostimulation for layer 2/3 and deep layer neurons. Histograms show mean values and SEM. Data from all recorded neurons were pooled. (A–D) Density of synaptic inputs measured as the number of stimulation sites generating an excitatory or inhibitory input/mm2 in concentric rings spaced 50 μm apart. Data were pooled for all 19 layer 2/3 neurons (A,B) and all 17 deep layer neurons (C,D) included in this study. Local excitatory and inhibitory inputs showed a similar spatial organization: most inputs originated within a distance of <500 μm around the postsynaptic cell body; however, a small fraction of inputs originated >1 millimeter away. As a general trend, input densities were lower in the deep layers as compared to layers 2/3.

Figure 3.

Spatial distribution of excitatory and inhibitory synaptic inputs to upper and lower layer pyramidal neurons. To estimate the spatial extent of local and long-range synaptic connections in our preparation, we determined the density of synaptic inputs evoked by photostimulation for layer 2/3 and deep layer neurons. Histograms show mean values and SEM. Data from all recorded neurons were pooled. (A–D) Density of synaptic inputs measured as the number of stimulation sites generating an excitatory or inhibitory input/mm2 in concentric rings spaced 50 μm apart. Data were pooled for all 19 layer 2/3 neurons (A,B) and all 17 deep layer neurons (C,D) included in this study. Local excitatory and inhibitory inputs showed a similar spatial organization: most inputs originated within a distance of <500 μm around the postsynaptic cell body; however, a small fraction of inputs originated >1 millimeter away. As a general trend, input densities were lower in the deep layers as compared to layers 2/3.

Figure 4.

Spatial organization and orientation tuning of synaptic inputs to a deep layer neuron. (A) Orientation preference map without superimposed synaptic input map. Age of animal, P42. (B) Photostimulation-evoked excitatory synaptic inputs superimposed on the orientation preference map (white dots). The black star indicates the position of the postsynaptic neuron. (C) Inhibitory synaptic inputs (black dots) recorded from the same neuron superimposed on the orientation map. (D) Combined excitatory and inhibitory input pattern. The map was dominated by inhibitory inputs arising from a wide range of locations. Recordings were made at –20 mV. (E) Sample traces recorded following stimulation of the sites indicated by white numbers on the orientation map. Panel 1 shows the direct glutamate response of the postsynaptic cell. Panels 2–6 show GABAergic synaptic inputs evoked from the positions indicated by the white numbers in (D). Panel 3 shows a small, excitatory input evoked from a remote location. (F) Orientation tuning histograms for excitatory inputs (white bars) and inhibitory inputs (black bars).

Figure 4.

Spatial organization and orientation tuning of synaptic inputs to a deep layer neuron. (A) Orientation preference map without superimposed synaptic input map. Age of animal, P42. (B) Photostimulation-evoked excitatory synaptic inputs superimposed on the orientation preference map (white dots). The black star indicates the position of the postsynaptic neuron. (C) Inhibitory synaptic inputs (black dots) recorded from the same neuron superimposed on the orientation map. (D) Combined excitatory and inhibitory input pattern. The map was dominated by inhibitory inputs arising from a wide range of locations. Recordings were made at –20 mV. (E) Sample traces recorded following stimulation of the sites indicated by white numbers on the orientation map. Panel 1 shows the direct glutamate response of the postsynaptic cell. Panels 2–6 show GABAergic synaptic inputs evoked from the positions indicated by the white numbers in (D). Panel 3 shows a small, excitatory input evoked from a remote location. (F) Orientation tuning histograms for excitatory inputs (white bars) and inhibitory inputs (black bars).

Figure 5.

Spatial organization and orientation tuning of synaptic inputs to a layer 2/3 neuron. (A) Orientation preference map with superimposed synaptic input map. Age of animal, P45. White dots, excitatory inputs; black dots, inhibitory inputs; black star, position of postsynaptic neuron; pink circles, bead injection sites. (B) Orientation tuning histogram of excitatory synaptic inputs for the example neuron shown in (A). (C) Orientation tuning histogram of inhibitory synaptic inputs (A). Although the majority of both excitatory and inhibitory inputs are generated in iso-orientation tuned regions of the map, the distribution of IPSCs is broader than the distribution of EPSCs.

Figure 5.

Spatial organization and orientation tuning of synaptic inputs to a layer 2/3 neuron. (A) Orientation preference map with superimposed synaptic input map. Age of animal, P45. White dots, excitatory inputs; black dots, inhibitory inputs; black star, position of postsynaptic neuron; pink circles, bead injection sites. (B) Orientation tuning histogram of excitatory synaptic inputs for the example neuron shown in (A). (C) Orientation tuning histogram of inhibitory synaptic inputs (A). Although the majority of both excitatory and inhibitory inputs are generated in iso-orientation tuned regions of the map, the distribution of IPSCs is broader than the distribution of EPSCs.

Figure 6.

Summary orientation tuning histograms of excitatory and inhibitory inputs to layer 2/3 (A–C) and layer 5/6 (D–F) pyramidal neurons and relationship between tuning of excitatory and inhibitory inputs to upper layer (G) and layer 5/6 neurons (H). Bars represent the percentage of total EPSCs/IPSCs falling into each orientation difference category compared to the postsynaptic neurons. Each site generating an EPSC or IPSC was counted only once, even if multiple events were generated from one site. Histograms show mean values and SEM. (A,B) Excitatory inputs to layer 2/3 cells (n = 19). The tuning histogram for local inputs (evoked from a distance <500 μm from the postsynaptic neuron) are shown in (A), the inputs evoked from larger distances (>500 μm) are shown in (B). (C) Inhibitory inputs to layer 2/3 cells (n = 15). (D,E) Excitatory inputs to deep layer cells (n = 17). Excitatory inputs were again subdivided into events evoked from within 500 μm distance from the postsynaptic cell (D) and events originating from distances >500 μm (E). (F) Inhibitory inputs to deep layer cells (n = 12). Since inhibitory inputs were predominantly of local origin, all IPSCs recorded from upper (C) and lower layer (F) pyramidal neurons were pooled.

Figure 6.

Summary orientation tuning histograms of excitatory and inhibitory inputs to layer 2/3 (A–C) and layer 5/6 (D–F) pyramidal neurons and relationship between tuning of excitatory and inhibitory inputs to upper layer (G) and layer 5/6 neurons (H). Bars represent the percentage of total EPSCs/IPSCs falling into each orientation difference category compared to the postsynaptic neurons. Each site generating an EPSC or IPSC was counted only once, even if multiple events were generated from one site. Histograms show mean values and SEM. (A,B) Excitatory inputs to layer 2/3 cells (n = 19). The tuning histogram for local inputs (evoked from a distance <500 μm from the postsynaptic neuron) are shown in (A), the inputs evoked from larger distances (>500 μm) are shown in (B). (C) Inhibitory inputs to layer 2/3 cells (n = 15). (D,E) Excitatory inputs to deep layer cells (n = 17). Excitatory inputs were again subdivided into events evoked from within 500 μm distance from the postsynaptic cell (D) and events originating from distances >500 μm (E). (F) Inhibitory inputs to deep layer cells (n = 12). Since inhibitory inputs were predominantly of local origin, all IPSCs recorded from upper (C) and lower layer (F) pyramidal neurons were pooled.

Figure 7.

Amplitudes of photostimulation-evoked EPSCs and IPSCs as a function of orientation tuning difference between pre- and postsynaptic sites. Bars represent the average amplitude of all synaptic currents falling into a particular orientation tuning difference category recorded from all layer 2/3 (A–C) or layer 5/6 (D–F) pyramidal neurons. Histograms show mean values and SEM. (A,B) Excitatory inputs to layer 2/3 cells (n = 19). (C) Inhibitory inputs to layer 2/3 cells (n = 15). (D,E) Excitatory inputs to deep layer cells (n = 17). (F) Inhibitory inputs to deep layer cells (n = 12). Excitatory inputs were subdivided into events evoked from within 500 μm distance from the postsynaptic cell (A,D) and events originating from distances >500 μm (B,E). Since inhibitory inputs were predominantly of local origin, all IPSCs recorded from upper (C) and lower layer (F) pyramidal neurons were pooled. Local iso-orientation tuned excitatory inputs were larger in amplitude than inputs originating in sites showing a larger tuning difference as compared to the location of the postsynaptic neuron (A,D). Long-range excitatory inputs showed the same tendency in layers 5/6 (E), but it was less obvious in the upper layers (B).

Figure 7.

Amplitudes of photostimulation-evoked EPSCs and IPSCs as a function of orientation tuning difference between pre- and postsynaptic sites. Bars represent the average amplitude of all synaptic currents falling into a particular orientation tuning difference category recorded from all layer 2/3 (A–C) or layer 5/6 (D–F) pyramidal neurons. Histograms show mean values and SEM. (A,B) Excitatory inputs to layer 2/3 cells (n = 19). (C) Inhibitory inputs to layer 2/3 cells (n = 15). (D,E) Excitatory inputs to deep layer cells (n = 17). (F) Inhibitory inputs to deep layer cells (n = 12). Excitatory inputs were subdivided into events evoked from within 500 μm distance from the postsynaptic cell (A,D) and events originating from distances >500 μm (B,E). Since inhibitory inputs were predominantly of local origin, all IPSCs recorded from upper (C) and lower layer (F) pyramidal neurons were pooled. Local iso-orientation tuned excitatory inputs were larger in amplitude than inputs originating in sites showing a larger tuning difference as compared to the location of the postsynaptic neuron (A,D). Long-range excitatory inputs showed the same tendency in layers 5/6 (E), but it was less obvious in the upper layers (B).

Figure 8.

Examples of Gaussian fits to the EPSC (A) and IPSC (B) tuning histograms for a layer 2/3 neuron. Gaussian fits were used to compare the tuning width of EPSCs and IPSCs. The IPSC distributions were significantly wider in the majority of both upper and lower layer neurons tested. Fitting parameters: (A) χ2 = 3.4, R2 = 0.97, width = 12.7; (B) χ2 = 9.8, R2 = 0.8, width = 16.6.

Figure 8.

Examples of Gaussian fits to the EPSC (A) and IPSC (B) tuning histograms for a layer 2/3 neuron. Gaussian fits were used to compare the tuning width of EPSCs and IPSCs. The IPSC distributions were significantly wider in the majority of both upper and lower layer neurons tested. Fitting parameters: (A) χ2 = 3.4, R2 = 0.97, width = 12.7; (B) χ2 = 9.8, R2 = 0.8, width = 16.6.

Figure 9.

Comparison of real and ‘random’ synaptic inputs for an example layer 2/3 neuron. (A) Orientation preference map with superimposed pattern of photostimulation-evoked excitatory synaptic inputs (white dots). Black star: position of postsynaptic layer 2/3 neuron. (B) Orientation preference map with superimposed pattern of photostimulation-evoked inhibitory synaptic inputs (black dots). Black star: position of postsynaptic layer 2/3 neuron. (C) Orientation preference map with superimposed pattern of ‘random’ inputs (white dots). Black star: position of postsynaptic layer 2/3 neuron. (D) Tuning histogram for ‘real’ excitatory synaptic inputs. As in most neurons in our sample, the majority of inputs originate in iso-orientation regions. (E) Tuning histogram of inhibitory inputs. The distribution is significantly broader than the distribution of excitatory inputs. (F) Tuning histogram of ‘random’ inputs. The orientation tuning of artificial inputs is significantly broader than the distribution of actual excitatory inputs. This indicates that the iso-orientation tuning of actual synaptic inputs is due to the specificity of synaptic connections between cortical neurons and not simply a product of the intrinsic structure of the orientation preference map.

Figure 9.

Comparison of real and ‘random’ synaptic inputs for an example layer 2/3 neuron. (A) Orientation preference map with superimposed pattern of photostimulation-evoked excitatory synaptic inputs (white dots). Black star: position of postsynaptic layer 2/3 neuron. (B) Orientation preference map with superimposed pattern of photostimulation-evoked inhibitory synaptic inputs (black dots). Black star: position of postsynaptic layer 2/3 neuron. (C) Orientation preference map with superimposed pattern of ‘random’ inputs (white dots). Black star: position of postsynaptic layer 2/3 neuron. (D) Tuning histogram for ‘real’ excitatory synaptic inputs. As in most neurons in our sample, the majority of inputs originate in iso-orientation regions. (E) Tuning histogram of inhibitory inputs. The distribution is significantly broader than the distribution of excitatory inputs. (F) Tuning histogram of ‘random’ inputs. The orientation tuning of artificial inputs is significantly broader than the distribution of actual excitatory inputs. This indicates that the iso-orientation tuning of actual synaptic inputs is due to the specificity of synaptic connections between cortical neurons and not simply a product of the intrinsic structure of the orientation preference map.

References

Berman NJ, Douglas RJ, Martin KA, Whitteridge D (
1991
) Mechanisms of inhibition in cat visual cortex.
J Physiol
 
440
:
667
–722.
Blakemore C, Tobin EA (
1972
) Lateral inhibition between orientation detectors in the cat's visual cortex.
Exp Brain Res
 
15
:
439
–440.
Bonhoeffer T, Grinvald A (
1993
) The layout of iso-orientation domains in area 18 of cat visual cortex: optical imaging reveals a pinwheel-like organization.
J Neurosci
 
13
:
4157
–4180.
Chapman B, Stryker MP (
1993
) Development of orientation selectivity in ferret visual cortex and effects of deprivation.
J Neurosci
 
13
:
5251
–5262.
Chung S, Ferster D (
1998
) Strength and orientation tuning of the thalamic input to simple cells revealed by electrically evoked cortical suppression.
Neuron
 
20
:
1117
–1189.
Crook JM, Eysel UT (
1992
) GABAergic-induced inactivation of functionally characterized sites in cat visual cortex (area 18): effects on orientation tuning.
J Neurosci
 
11
:
1347
–1358.
Crook JM, Kisvarday ZF, Eysel UT (
1997
) GABA-induced inactivation of functionally characterized sites in cat striate cortex: effects on orientation tuning and direction selectivity.
Vis Neurosci
 
14
:
141
–158.
Durack JC, Katz LC (
1996
) Development of horizontal projections in layer 2/3 of ferret visual cortex.
Cereb Cortex
 
6
:
178
–183.
Douglas RJ, Martin KAC, Whitteridge D (
1991
) An intracellular analysis of the visual responses of neurones in cat visual cortex.
J Physiol (Lond)
 
44
:
659
–696.
Ferster D (
1986
) Orientation selectivity of synaptic potentials in neurons of cat primary visual cortex.
J Neurosci
 
6
:
1284
–1301.
Ferster D (
1988
) Spatially opponent excitation and inhibition in simple cells of cat visual cortex.
J Neurosci
 
6
:
1172
–1180.
Ferster D, Chung S, Wheat HS (
1996
) Orientation selectivity of thalamic input to simple cells of cat visual cortex.
Nature
 
380
:
249
–252.
Hata Y, Tsumoto T, Sato H, Tamura H (
1991
) Horizontal interactions between visual cortical neurons studied with cross-correlation analysis in the cat.
J Physiol
 
441
:
593
–614.
Heggelund P (
1981
) Receptive field organization of simple cells in cat striate cortex.
Exp Brain Res
 
42
:
89
–98.
Hubel DH, Wiesel TN (
1962
) Receptive fields, binocular interactions and functional architecture in the cat's visual cortex.
J Physiol (Lond)
 
160
:
106
–154.
Katz LC, Dalva MB (
1994
) Scanning laser photostimulation: a new approach for analyzing brain circuits.
J Neurosci Meth
 
54
:
205
–218.
Katz LC, Iarovici DM (
1990
) Green fluorescent latex microspheres: a new retrograde tracer.
Neuroscience
 
34
:
511
–520.
Morrone MC, Burr DC, Maffei L (
1982
) Functional implications of cross-orientation inhibition of cortical visual cells.
Proc R Soc Lond
 
216
:
335
–354.
Nelson SB (
1991
) Temporal interactions in the cat visual system. I. Orientation selective suppression in the visual cortex.
J Neurosci
 
11
:
344
–356.
Nelson JI, Frost B (
1978
) Orientation selective inhibition from beyond the classical receptive field.
Brain Res
 
139
:
359
–365.
Pei X, Vidyasagar TR, Volgushev M, Creutsfeldt OD (
1994
) Receptive field analysis and orientation selectivity of postsynaptic potentials of simple cells in cat visual cortex.
J Neurosci
 
14
:
7130
–1760.
Roerig B, Kao JPY (
1999
) Organization of intracortical circuits in relation to direction preference maps in ferret visual cortex.
J Neurosci
 
19
:
1
–5.
Roerig B, Katz LC (
1998
) Relationship between synaptic input patterns and orientation preference maps in ferret visual cortex.
Soc Neurosci Abstr
 
24
:
765
.
Sengpiel F, Sen A, Blakemore C (
1997
) Characteristics of surround inhibition in cat area 17.
Exp Brain Res
 
116
:
216
–228.
Sillito AM (
1975
) The contribution of inhibitory mechanisms to the receptive field properties of neurones in the striate cortex of the cat.
J Physiol
 
250
:
305
–329.
Sillito AM (
1979
) Inhibitory mechanisms influencing complex cell orientation selectivity and their modification at high resting discharge levels.
J Physiol.
 
289
,
33
–53.
Sillito AM, Kemp JA, Milson, JA and Berardi N (
1980
) A reevaluation of the mechanisms underlying simple cell organization selectivity.
Brain Res
 
194
:
517
–520.
Somers DC, Nelson SB, Sur M (
1995
) An emergent model of orientation selectivity in cat visual cortical simple cells.
J Neurosci
 
15
:
5448
–5465.
Weliky M, Kandler K, Fitzpatrick D, Katz LC (
1995
) Patterns of excitation and inhibition evoked by horizontal connections in visual cortex share a common relationship to orientation columns.
Neuron
 
16
:
541
–552.
Woergoetter F, Koch C (
1991
) A detailed model of the primary visual pathway in the cat: comparison of afferent excitatory and intracortical inhibitory connection schemes for orientation selectivity.
J Neurosci
 
11
:
1959
–1979.