Abstract

In monkey prefrontal cortex (PFC), ~50% of the local axon terminals of pyramidal neurons form synapses with the dendritic shafts of GABA neurons. Subclasses of GABA neurons can be distinguished by the presence of different calcium-binding proteins. For example, in monkey PFC, parvalbumin (PV)-containing cells comprise ~25% of GABA neurons and are predominantly located in layers 3b–4, whereas calretinin (CR)-containing cells, which are present in greatest density in layers 2–3a, constitute 50% of GABA neurons. Consequently, in order to determine the cell class and laminar specificity of the dendritic targets of pyramidal neuron local axon collaterals in monkey PFC area 9, we conducted ultrastructural analyses of local axon terminals labeled with the anterograde tracer, biotinylated dextran amine, and dendrites immunoreactive (IR) for PV or CR. In layer 3b, the majority of the local axon terminals targeted PV-IR dendritic shafts, whereas CR-IR dendritic shafts were targeted infrequently. This differential targeting was also present in layers 2–3a, although it was less pronounced. In addition, PV-IR dendrites had a significantly greater density of excitatory inputs than did CR-IR dendrites. These findings indicate that PV-containing interneurons, which have a potent inhibitory effect on pyramidal neurons, are selectively targeted by the excitatory local axon terminals of supragranular pyramidal neurons in monkey PFC. These connections may provide the anatomical substrate for the coordinated activity of pyramidal neurons and fast-spiking GABA neurons during working memory.

Introduction

The neural circuitry of the dorsal prefrontal cortex (PFC) appears to be a critical mediator of certain cognitive processes, particularly those that involve working memory (Goldman-Rakic, 1987; Baddeley, 1992; Fuster, 1997). For example, neurons in the PFC exhibit different patterns of activity during specific phases of the oculomotor delayed-response task, and the sustained firing of neurons during the delay period of this task has been suggested to be the cellular basis of working memory (Goldman-Rakic, 1995). Although much work has focused on the role of pyramidal neurons in this circuitry, recent studies have demonstrated that GABA-mediated inhibition is also critical to these cognitive processes (Sawaguchi et al., 1988, 1989; Wilson et al., 1994; Rao et al., 1999, 2000). For example, during tasks that require spatial working memory, fast-spiking GABA neurons in the monkey PFC display delay-period activity that is selective for memoranda in specific spatial locations (Wilson et al., 1994; Rao et al., 1999, 2000). In addition, adjacent pyramidal and GABA cells display isodirectional tuning (Rao et al., 1999, 2000), suggesting the presence of synaptic connections between neighboring pyramidal cells and a subclass of GABA neurons.

Subclasses of cortical GABA neurons have been identified by electrophysiological and morphological criteria (Lund and Lewis, 1993; Kawaguchi and Kubota, 1997). In addition, different subclasses of GABA neurons may be distinguished by the presence of specific calcium-binding proteins (DeFelipe, 1997). For example, parvalbumin (PV)-containing neurons include chandelier cells and basket-like, wide-arbor neurons, whereas calretinin (CR) is present in double bouquet cells (Lund and Lewis, 1993; Condé et al., 1994; Gabbott and Bacon, 1996a). In addition, in monkey PFC, PV-containing cells comprise ~25% of GABA neurons, whereas CR-containing cells constitute ~50% of GABA neurons (Condé et al., 1994; Gabbott and Bacon, 1996b). Although both PV- and CR-containing cells are present across cortical layers, they are differentially weighted to particular layers. For example, in monkey PFC, CR-containing cells are predominantly located in layer 1 to superficial layer 3, whereas PV-containing neurons are most dense in layers deep 3–4 (Condé et al., 1994; Gabbott and Bacon, 1996a).

The calcium-binding protein subclasses of GABA cells can be further distinguished by their synaptic targets. For example, PV-containing chandelier and wide-arbor neurons (Lund and Lewis, 1993; Condé et al., 1994; Gabbott and Bacon, 1996a) provide the most proximal inhibitory input to pyramidal neurons. Specifically, the axon terminals of chandelier cells are arrayed in distinctive vertical clusters that form Gray’s Type II synapses onto the axon initial segments of pyramidal cells (Somogyi, 1977; Freund et al., 1983; DeFelipe et al., 1985), and wide-arbor neurons (Lund and Lewis, 1993), similar to large basket cells located in sensory cortices, form inhibitory synapses onto the somata and proximal dendritic shafts and spines of pyramidal cells (Freund et al., 1983; Williams et al., 1992). In contrast, CR-containing neurons appear to contact primarily other GABA neurons. For example, in the superficial layers of visual cortex in both rats (Gonchar and Burkhalter, 1999) and monkeys (Meskenaite, 1997), as well as in rat hippocampus (Gulyás et al., 1996), the majority of CR- labeled axon terminals form Gray’s Type II synapses onto GABA-containing dendritic shafts and somata. Some of these postsynaptic targets are also CR-immunoreactive (Gulyás et al., 1996; Gonchar and Burkhalter, 1999).

The activity of these different subclasses of GABA cells depends, in part, upon the source and amount of synaptic inputs that they receive. Interestingly, recent studies suggest that different classes of GABA neurons receive discrete inputs. For example, in rat barrel cortex, PV-containing interneurons are the major GABA cell class to receive excitatory input from the ventroposteromedial thalamic nucleus (Staiger et al., 1996). Similarly, in monkey PFC, dopamine inputs preferentially target PV-labeled dendrites (Sesack et al., 1998) and not CR-labeled dendrites (Sesack et al., 1995a). In addition, in rat hippocampus, an ultrastructural study revealed that PV-containing neurons receive a greater total amount of excitatory synaptic input than do CR-containing neurons (Gulyás et al., 1999). Although this study did not identify the source of the excitatory synapses, it does suggest that GABA neurons differ in the amount of input they receive.

Given these differences across cell classes, the role of GABA neurons in working memory processes may be further informed by determining the sources of excitatory inputs to different types of GABA neurons. We previously demonstrated that both the associational and long-range intrinsic axons of supragranular pyramidal neurons in monkey PFC target almost exclusively the dendritic spines of other pyramidal cells (Melchitzky et al., 1998), whereas ~50% of the local axon collaterals (within 300 μm of the cell body) of these neurons form synapses onto the dendrites of GABA neurons (Melchitzky et al., 2001). Although these GABA neurons include the PV-containing subclass (Melchitzky et al., 2001), it is unknown whether other local circuit neurons receive such input, whether this input differs as a function of laminar location, and whether these inputs reflect the relative amount of excitatory inputs to different GABA cells. Thus, in this study, we used dual-labeling electron microscopy to examine excitatory synapses onto PV- and CR-labeled dendrites in different layers of monkey PFC.

Methods

Surgical Procedures and Tissue Preparation

Tissue sections from the same three male, adult cynomolgus monkeys (Macaca fascicularis) used in our previous study (Melchitzky et al., 2001) were examined in this study. All animals were treated according to the guidelines outlined in the National Institutes of Health Guide for the Care and Use of Laboratory Animals, as approved by the University of Pittsburgh’s Animal Care and Use Committee. As previously described, animals were anesthesized, placed in a stereotaxic apparatus, and a craniectomy was performed over the dorsal PFC. Iontophoretic injections of 10% biotinylated dextran amine (BDA; 10 000 MW; Molecular Probes, Eugene, OR) were made by passing positive current (5 μA, 7 s cycles) through glass pipettes (tip diameter 20–30 μm) for 10 min. Injections were centered at a depth of 1.2 mm below the pial surface in PFC area 9 (Walker, 1940). As shown in Figure 1A, one injection was made in each animal in either the right (two monkeys) or left (one monkey) hemisphere. After a survival time of 8–12 days, monkeys were deeply anesthetized and then perfused transcardially with room temperature (29°C) 1% paraformaldehyde and 0.05% glutaraldehyde in 0.1 M phosphate buffer (PB), pH 7.4, for 5 s followed by 4% paraformaldehyde and 0.2% glutaraldehyde in 0.1 M PB, pH 7.4, for 9 min at a flow rate of 350–400 ml/min. The brains were then removed, and coronal blocks (5–6 mm thick) containing the BDA injection sites were immersed in cold 4% paraformaldehyde for 2 h. For all animals, following post-fixation, tissue blocks were washed in 0.1 M PB, pH 7.4, and sectioned coronally on a Vibratome at 50 μm.

For all three animals, double-labeling experiments (Sesack et al., 1998) were performed to label both BDA- and PV- or CR-containing structures. Briefly, tissue sections were incubated overnight in 0.01 M phosphate-buffered saline (PBS), pH 7.4, containing 0.2% bovine serum albumin, 0.04% Triton X-100, 3% normal goat serum, 3% normal human serum, Vectastain ABC reagents (Vector Laboratories, Burlingame, CA) and either a rabbit anti-CR antibody (diluted 1:2000; Swant, Bellinzona Switzerland) or a rabbit anti-PV antibody (diluted 1:1000, kindly provided by Dr Kenneth Baimbridge, University of British Columbia). The following day, BDA-containing structures were labeled with 3,3′- diaminobenzidine and PV- and CR-containing structures were visualized with silver enhanced immunogold (Sesack et al., 1998; Melchitzky et al., 2001). Sections were then post-fixed in 2% osmium tetroxide for 1 h, dehydrated in ascending alcohols and embedded in Epon 812 (Electron Microscopy Sciences, Fort Washington, PA). The specificity of these antibodies was demonstrated by pre-adsorption experiments using an excess of calcium-binding proteins, including PV, CR and calbindin (Condé et al., 1994; Daviss and Lewis, 1995). Immunoreactivity was blocked by pre-adsorption with the protein against which the antibody was raised and unchanged by incubation with the other two calcium- binding proteins.

Definition of Neuronal and Synaptic Elements

Neuronal elements of relevance to this study were identified using the criteria of Peters et al. (Peters et al., 1991). Axon terminals were gener- ally >0.2 μm in diameter and contained synaptic vesicles and often mitochondria. Dendritic shafts were characterized by the presence of mitochondria, numerous microtubules and neurofilaments, as well as synaptic specializations. Dendritic spines were identified by the absence of both organelles and microtubules and by the presence of a spine apparatus (in optimal planes of section). Gray’s Type I synapses (Gray, 1959) were characterized by widened and parallel spacing of apposed plasmalemmal surfaces, and a thick postsynaptic density. Furthermore, the axon terminals forming these synapses contained round synaptic vesicles. In contrast, Gray’s Type II (Gray, 1959) synapses had thin postsynaptic densities, intercleft filaments, and synaptic vesicles that were usually pleomorphic in shape.

Sampling Regions and Procedures for BDA-labeled Terminals

In order to identify the targets of local axon collaterals of supragranular pyramidal neurons, BDA-labeled axon terminals located in zones ~300 μm from the core of the BDA injection site were sampled (Fig. 1). From each animal, one or two trapezoid blocks were taken from both layers 2 and 3a and layer 3b (Fig. 1D). Coronal blocks were sectioned on a Leica ultramicrotome at 80 nm and three to six ultrathin sections were collected on individual 200 mesh copper grids. The grids were subsequently counterstained with uranyl acetate and lead citrate, and examined on a JEOL 100 CX electron microscope. For each block, two or three grids (separated by at least 10 and at most 20 grids) were analyzed. One section per grid was randomly chosen as the starting point for analysis. Within this section, all BDA-labeled axon terminals within fields of specific CR- or PV-labeling were identified and photographed at ×19 000. BDA-labeled axon terminals were considered to be within a field containing specific CR or PV labeling if both labeled elements were encompassed by an electron micrograph at ×10 000 magnification (66.5 μm2). All profiles of BDA-labeled axon terminals were followed in limited (≤4) serial sections.

For those terminals that had an identifiable synaptic specialization in any of the serial sections, the post-synaptic target was determined. The terminal was then placed into one of the following categories: (i) synapse onto immunolabeled dendrite, (ii) synapse onto unlabeled dendrite, or (iii) synapse onto unlabeled spine. Those terminals without a definite synaptic specialization were excluded from further analyses. In order to avoid false-negative results in the dual-labeled tissue, we sampled only from areas of the trapezoid blocks that contained the tissue–epon interface and we only analyzed fields that contained both specific peroxidase and gold–silver immunolabeling (Sesack et al., 1998).

The proportions of BDA-labeled terminals targeting CR- and PV- labeled dendrites were compared using two-sided, 2 × 2 chi-square analyses. The data presented in Results for inputs to PV-labeled dendrites in layer 3b have been previously reported (Melchitzky et al., 2001).

Determination of the Relative Amount of Excitatory Input to Immunolabeled Dendrites

In order to determine the relative amount of synaptic input to CR- or PV-immunoreactive (IR) dendritic shafts, random fields containing CR- or PV-IR dendritic shafts from the sections used for sampling the synaptic targets of local axon terminals were photographed at ×19 000. Using a computer digitizing system (MCID M5; Research Imaging Inc., St Catherines, Ontario, Canada), the perimeter of each labeled dendritic shaft was outlined to determine the length of the dendritic membrane. In addition, for each apposed, unlabeled axon terminal, the length of the apposition was measured. Measurements were made by two investi- gators, both of whom were blind to dendritic category and layer. The within-rater reliability of length measurements was confirmed by an intraclass correlation coefficient (ICC) of 0.999 [95% confidence interval (CI) = 0.998–0.999]. A between-rater ICC of 0.999 (95% CI = 0.998–0.999) demonstrated the consistency of measurements.

The apposed, unlabeled axon terminals were divided into three categories based upon their synaptic specialization in reference to the labeled dendritic shaft in this single section analysis: (i) apposition without an identifiable synaptic specialization, (ii) Gray’s Type I synapse or (iii) Gray’s Type II synapse. Terminals in category 1 that clearly synapsed onto another structure in the neuropil were considered unlikely to also form a synapse onto the immunolabeled dendrite (Kisvárday et al., 1986; McGuire et al., 1991) and were excluded from further analysis. To determine the proportion of labeled dendrite membrane length that was contacted by unlabeled axon terminals, the ratio of the sum of the lengths of apposed terminals for each dendrite to the total length of the dendritic membrane was computed. In addition, the proportion of dendritic membrane apposed by terminals forming Type I synapses was also calculated. Differences in these proportions were compared using a two-way analysis of variance (ANOVA) with type of labeled dendrite (i.e. PV- or CR-labeled) and layer as the main effects. Because the focus of this portion of the study was on the relative amount of excitatory input to PV- and CR-IR dendrites, the terminals forming Type II synapses were not analyzed. In addition, the use of a single section analysis results in an underestimation of the number of inputs from Type II synapses because their thin postsynaptic density and small synaptic size make Type II synapses more difficult to identify in, and less likely to be sampled by, single sections.

Results

Each of the three BDA injection sites was located in layers 2–3 of Walker’s (Walker, 1940) area 9 (Fig. 1A). These injection sites consisted of a dense core of dark reaction product with a zone containing labeled terminals and neurons immediately surrounding it (Fig. 1B,C). The BDA-labeled neurons within this zone were often intensely labeled and present within layers 2–3. In addition, BDA-labeled axons emanated from the injection site, some of which terminated locally within 300 μm of the injection core, whereas others continued traveling horizontally in both mediolateral and rostrocaudal directions (Fig. 1B).

In the tissue examined, BDA-containing terminals were well-labeled and abundant. However, the goal of this part of the study was to determine if BDA-labeled terminals differentially contacted PV- and CR-IR dendritic shafts. Therefore, in order to avoid confounds introduced by potential differences in the penetration of the antibodies (Sesack et al., 1995b), we quantified only those BDA-labeled axon terminals that were located in the same field (66.5 μm2) as either a PV- or CR-labeled structure. The amount of dual-labeled tissue sampled for CR-containing fields was 314 095 μm2 and 409 976 μm2 for layers 2–3a and layer 3b, respectively. For PV-containing fields, the amount of tissue sampled was 360 381 μm2 for layers 2–3a and 254 581 μm2 for layer 3b. Within these fields, 264 BDA-labeled terminals met the specified criteria and were further analyzed. Of these terminals, 110 were from monkey CM214, 88 from monkey CM222 and 66 from monkey CM223.

Differential Targeting of CR- and PV-labeled Dendritic Shafts

In layers 2–3 of monkey PFC, local axon terminals were found to contact both PV- (Fig. 2A) and CR-labeled (Fig. 2B) dendritic shafts. In fields containing both BDA and PV labeling, 48% (55/115) of the BDA-labeled, local axon terminals were apposed to PV-IR dendrites, a proportion that was significantly greater (χ2 = 12.70; P < 0.001) than the 23% (20/86) of local terminals apposed to CR-IR dendrites in fields containing both BDA and CR labeling. A more marked difference in target specificity was observed when only BDA-labeled local axon terminals with identifiable Type I synapses were examined. Specifically, 34% (10/29) of these terminals formed synapses onto PV-IR dendritic shafts, whereas only 3% (1/34) synapsed onto CR-IR dendritic shafts (χ2 = 10.8; P = 0.002). Of the 63 terminals with identi- fiable Type I synapses, 22 were from monkey CM214, 28 from monkey CM222, and 13 from monkey CM223.

Analysis of these data by laminar location (layers 2–3a versus layer 3b) revealed additional target specificity of the local excitatory axon terminals; that is, the targets of BDA-labeled axon terminals forming appositions or Type I synapses differed as a function of both laminar location and calcium binding protein. In layer 3b, the proportion of BDA-labeled terminals in apposition to PV-IR dendrites (53%, 44/83) was significantly greater (χ2 = 13.70; P < 0.001) than the proportion in appo- sition to CR-labeled dendrites (23%, 14/62). Furthermore, the proportion of BDA-labeled terminals with identifiable Type I synapses onto PV-labeled dendrites (53%; 8/15) was greater than the proportion onto CR-labeled dendrites (5%; 1/19) in layer 3b (Fig. 3; χ2 = 9.95; P = 0.004). In layers 2–3a, although BDA- labeled terminals formed more appositions and synapses onto PV-labeled dendrites (34%, 11/32 and 14%, 2/14, respectively) than onto CR-labeled dendrites (25%, 6/24 and 0%, 0/15, respectively), these differences were not statistically significant (χ2 = 0.57; P = 0.562 and χ2 = 2.30; P = 0.224 for appositions and synapses, respectively).

The proportion of BDA-labeled terminals in apposition to PV-IR dendrites was greater in layer 3b (53%, 44/83) than in layers 2–3a (23%, 14/62), although this difference did not achieve statistical significance (χ2 = 3.22; P = 0.073). However, as shown in Figure 3, the proportion of local axon terminals with identifiable Type I synapses onto PV-labeled dendritic shafts was significantly greater (χ2 = 4.89; P = 0.050) in layer 3b (53%, 8/15) than in layers 2–3a (14%, 2/14). This difference cannot be attributed to local axon terminals forming more synapses in general in layer 3b as opposed to layers 2–3a, because the combined data from the PV- and CR-labeled tissue revealed that 34% (29/85) of the BDA-labeled axon terminals in layers 2–3a formed identifiable synapses, compared to only 19% (34/179) in layer 3b (χ2 = 7.25; P = 0.007).

In addition to contacting labeled dendrites, BDA-labeled axon terminals also formed synapses onto unlabeled dendrites and spines (see Fig. 3), but not cell bodies (labeled or unlabeled). In both the PV- and CR-labeled tissue, approximately one half of the local axon terminals with identifiable Type I synapses targeted dendritic shafts; in the PV condition these shafts were pre- dominantly PV-IR, whereas in the CR condition almost all the targeted dendritic shafts were unlabeled (Fig. 3). As expected from our previous findings (Melchitzky et al., 2001), ~50% of the local axon terminals targeted dendritic spines in both the PV and CR conditions (Fig. 3).

Overall Relative Synaptic Input to PV- and CR-labeled Dendritic Shafts

To determine whether the proportions of BDA-labeled axon terminals that contacted PV- and CR-labeled dendritic shafts, reflected the relative densities of all terminals contacting these two populations, we obtained estimates of the overall synaptic input to each class of dendrite. As shown in Table 1, neither the mean length of membrane per labeled dendrite, nor the mean length of apposed axon terminals differed as a function of calcium binding protein or layer. However, two-way ANOVAs revealed that both the number of apposed axon terminals [F(1,115) = 10.69, P = 0.001] and the proportion of dendritic length apposed by all axon terminals [F(1,115) = 12.32; P = 0.001] were significantly greater for PV- than for CR-labeled dendritic shafts. The same analyses revealed no effect of layer or an interaction between layer and type of labeled dendrite.

As a more stringent test of a possible difference in relative synaptic input between PV- and CR-labeled dendritic shafts, we compared only those axon terminals with identifiable Type I synapses (Table 1). Both the number of Type I synapses [F(1,115) = 9.13; P = 0.003] and the proportion of dendritic length apposed by axon terminals with Type I synapses [F(1,115) = 7.44; P = 0.007] were greater for PV- than for CR-labeled shafts. In contrast, there was no effect for layer or an interaction between layer and type of labeled dendrite in these analyses.

Discussion

We previously reported that the local axon terminals of supragranular pyramidal neurons in monkey PFC synapse onto dendritic spines and dendritic shafts with equal frequency, and that the majority of the dendritic shafts receiving these inputs belonged to GABA neurons (Melchitzky et al., 2001). In the present study, we found that the targeted dendritic shafts principally belong to the PV- and not the CR-IR subclass of GABA neurons, that the extent to which PV-IR dendrites are targeted depends upon their laminar location, and that PV-IR dendrites receive a greater total complement of excitatory synapses than do CR-IR dendrites. Thus, the local axon terminals of supra- granular pyramidal cells preferentially target the dendrites of PV-containing neurons, and not other GABA neurons, in the middle layers of monkey PFC.

The targeting of PV-containing dendrites by these excitatory axon terminals does not appear to be an artifact of technical or sampling issues. For example, even though the immunogold method is less sensitive than immunoperoxidase staining (Chan et al., 1990), the preembedding immunogold technique used in this study provides greater sensitivity than postembedding methods (Chan et al., 1990; Pickel et al., 1993). Thus, although the relative incidence of BDA-labeled, local axon terminals forming synapses onto PV- and CR-IR dendrites may have been underestimated, the degree of this effect would not be expected to differ between the two immunolabels. Furthermore, any differences in the sensitivity of the two antibodies or in the relative densities of PV- and CR-labeled dendritic profiles would not confound our results because our sampling method required the presence of both BDA and PV or CR labeling in the same field (Sesack et al., 1995b). The absence of bias in this sampling approach is supported by the finding that the proportions of BDA-labeled terminals forming synapses onto dendritic spines were similar in the PV- (45%) and CR-labeled (47%) tissue.

Our observation that PV-IR dendrites appear to receive a greater overall density of excitatory inputs than do CR-IR dendrites replicates previous observations in rat hippocampus (Gulyás et al., 1999). Neither this finding, nor the observation that local axon terminals of pyramidal neurons preferentially target PV-IR rather than CR-IR neurons, appear to be a consequence of a greater prevalence of PV- than CR-labeled neurons and dendrites in layers 2–3 of monkey PFC. First, previous studies have documented that PV-containing cells represent ~25% of GABA neurons, whereas CR-containing cells constitute ~50% of GABA neurons in monkey PFC (Condé et al., 1994; Gabbott and Bacon, 1996b). Second, neither the number nor mean membrane length of the labeled dendrites examined differed between PV- and CR-labeled dendrites (Table 1), and the amount of tissue examined for PV- and CR-IR dendrites was of similar magnitude (723 988 μm2 and 668 025 μm2, respect- ively), suggesting that the densities of PV-IR and CR-IR dendrites are similar in layers 2–3 of monkey PFC.

The single section analysis utilized in the present study clearly underestimates the total number of Type I synapses onto labeled dendrites. Indeed, the number of terminals apposed to labeled dendrites is threefold greater than the number of Type I synapses (see Table 1), and it is likely that many of these apposed terminals form synaptic specializations onto immunolabeled dendrites in other sections. However, despite these limitations, single section analyses do provide reliable, relative comparisons of synaptic inputs to PV- and CR-labeled dendrites. The synaptic contacts between BDA-labeled terminals and unlabeled dendritic shafts in layers 2–3a of both the PV- and CR-labeled tissue raises the question of the cell class that receives this input. Unfortunately, the majority of these dendritic shafts were cut in cross-section, and thus many of the hallmark morphological characteristics of interneuron dendrites, such as a varicose shape and a high degree of synaptic input (Smiley and Goldman-Rakic, 1993; Sesack et al., 1995b), could not be assessed. Therefore, the dendritic shafts in layers 2–3a contacted by the local axon terminals could belong to another subclass of GABA neurons and/or to pyramidal cells. Several studies have demonstrated that in combination, the three calcium binding proteins, PV, CR and calbindin (CB), label ~90% of GABA neurons in a cortical region, with CB-containing interneurons comprising ~25% of the GABA cells in monkey PFC (Condé et al., 1994; Gabbott and Bacon, 1996b). However, in many cortical areas, including the PFC, CB is also present in a subpopulation of pyramidal neurons in layers 2–3 (DeFelipe and Jones, 1992; Condé et al., 1994; Gabbott and Bacon, 1996a). Therefore, the identification of synapses from local axon terminals of supragranular pyramidal neurons on CB-IR dendrites in layers 2–3a would not unambiguously define the cell class targeted.

Implications for the Functional Architecture of the Primate PFC

The axons of supragranular pyramidal neurons in monkey PFC furnish three major types of projections: (i) principal axons that pass through the white matter and terminate in other cortical regions, (ii) long-range axon collaterals that travel parallel to the pial surface through the gray matter, giving rise to a series of stripe-like clusters of axon terminals, and (iii) axon collater- als that arborize locally, within 300 μm of the cell body (Goldman-Rakic and Schwartz, 1982; Freund et al., 1990; Kritzer and Goldman-Rakic, 1995; Pucak et al., 1996; Melchitzky et al., 1998). For both the corticocortical connections and long-range projections, ~95% of the axon terminals synapse onto the dendritic spines of other pyramidal cells (Melchitzky et al., 1998). In contrast, only 50% of the local axon terminals synapse onto dendritic spines and the remaining 50% contact the dendritic shafts of GABA cells (Melchitzky et al., 2001). The findings of the present study, that these local axon collaterals of supragranular pyramidal neurons appear to be preferentially directed to the PV- and not the CR-containing class of GABA neurons, provide insight into the possible functional role of these connections. In concert with previous studies (see Introduction), PV-containing neurons appear to be specialized to receive and furnish direct synaptic connections with pyramidal cells. Not only do the local axon terminals of supragranular pyramidal neurons predominantly form synapses onto PV-IR dendrites in layer 3b of the PFC, but the density of Type 1 excitatory synapses is 40–90% greater on PV- than CR-IR dendrites (see Table 1). This greater relative density of excitatory synapses on PV- than CR-IR dendrites may reflect distinct functional properties of PV-IR neurons. For example, electro- physiological recordings in an in vitro slice preparation of monkey PFC have revealed that fast-spiking neurons, which include cells having the morphological features of PV-IR neurons, exhibit significantly shorter duration of EPSPs than do other classes of GABA neurons (González-Burgos et al., 2002). These findings suggest that, compared to other classes of interneurons, PV-IR neurons may require a larger number of inputs to integrate and fire.

These findings may reveal an anatomical basis for the role of GABA-mediated inhibition in the working memory processes subserved by the primate PFC. On the basis of both circuitry analyses (Goldman-Rakic, 1995; Pucak et al., 1996; Lewis and Gonzalez-Burgos, 1999) and modeling (Wang, 2001) studies, it has been suggested that reverberating circuits among recip- rocally connected populations of pyramidal neurons may contribute to the sustained firing of PFC neurons during the delay period of working memory tasks. In addition, in vivo recordings in monkey PFC have revealed that, similar to pyramidal cells, fast-spiking GABA neurons exhibit delay-period activity, and that adjacent pyramidal and GABA cells share similar tuning properties (Wilson et al., 1994; Rao et al., 1999, 2000). In vitro studies in rodent and monkey PFC suggest that these fast-spiking GABA cells include the PV-containing subclass (Kawaguchi, 1993, 1995; Krimer and Goldman-Rakic, 2001; Krimer et al., 2002). Thus, the local, bi-directional connections between supragranular pyramidal neurons and PV-IR GABA cells may be a critical substrate for the regulation of working memory-related neuronal firing in the PFC. In this regard, given the critical role of PFC dopamine in the regulation of working memory processes (Sawaguchi and Goldman-Rakic, 1991; Williams and Goldman-Rakic, 1995; Murphy et al., 1996; Arnsten and Goldman-Rakic, 1998), it is interesting that among GABA neurons, PV-IR neurons preferentially receive synaptic inputs from dopamine axons (Sesack et al., 1995a, 1998) and express the dopamine D1 receptor (Muly et al., 1998).

In conclusion, the results of the present study provide further evidence of the specificity of intrinsic synaptic connectivity in the monkey PFC, and may reveal anatomical substrates for the inhibitory processes that appear to be central for regulating pyramidal cell firing during different phases of working memory tasks (Constantinidis et al., 2002).

Notes

The authors thank Drs Guillermo González-Burgos and Leonid Krimer for critical comments on the manuscript, Mrs Mary Brady and Mr Colin Stebbins for assistance with photography and graphics, and Mr Dustin Kerr for assistance with data collection. These studies were supported by NIH grants MH45156 and MH51234.

Address correspondence to David A. Lewis, Western Psychiatric Institute and Clinic, Biomedical Science Tower, W1651, 3811 O’Hara Street, Pittsburgh, PA 15213, USA. Email: lewisda@msx.upmc.edu.

Table 1

Unlabeled axon terminals forming appositions or synapses with parvalbumin or calretinin dendrites in monkey prefrontal cortex

 Parvalbumin Calretinin Two-way ANOVA results 
 Layer 3b Layers 2–3a Layer 3b Layers 2–3a Labeled dendrite Layer Interaction 
*Excludes those axon terminals that are apposed to immunolabeled dendrites but that form synapses onto other structures. 
Number of dendrites 30 34 25 30 NA NA NA 
Mean (± SD) length of membrane/labeled dendrite (μm) 5.41 (3.06) 5.30 (2.54) 5.56 (2.30) 5.07 (2.99) F(1,115) = 0.008 P = 0.929 F(1,115) = 0.354 P = 0.553 F(1,115) = 0.145 P = 0.704 
Number of apposed axon terminals 94 90 72 61 NA NA NA 
Mean (± SD) length of apposed axon terminals (μm) 0.463 (0.193) 0.459 (0.154) 0.455 (0.333) 0.518 (0.299) F(1,313) = 0.876 P = 0.350 F(1,313) = 1.103 P = 0.294 F(1,313) = 1.423 P = 0.234 
Mean (± SD) number of apposed axon terminals/length of dendrite (μm)* 0.702 (0.252) 0.626 (0.298) 0.546 (0.264) 0.441 (0.306) F(1,115) = 10.69 P = 0.001 F(1,115) = 2.99 P = 0.087 F(1,115) = 0.077 P = 0.782 
Mean (±SD) proportion of dendritic length apposed by axon terminals* 0.332 (0.130) 0.289 (0.145) 0.234 (0.127) 0.214 (0.151) F(1,115) = 12.32 P = 0.001 F(1,115) = 1.54 P = 0.217 F(1,115) = 0.002 P = 0.964 
Mean (± SD) number of axon terminals forming Type I synapses/length of dendrite (μm) 0.279 (0.201) 0.286 (0.194) 0.201 (0.208) 0.147 (0.181) F(1,115) = 9.13 P = 0.003 F(1,115) = 0.413 P = 0.522 F(1,115) = 0.716 P = 0.399 
Mean (± SD) proportion of dendritic length apposed by axon terminals forming Type I synapses 0.130 (0.092) 0.129 (0.101) 0.096 (0.102) 0.068 (0.083) F(1,115) = 7.44 P = 0.007 F(1,115) = 0.735 P = 0.393 F(1,115) = 0.573 P = 0.451 
 Parvalbumin Calretinin Two-way ANOVA results 
 Layer 3b Layers 2–3a Layer 3b Layers 2–3a Labeled dendrite Layer Interaction 
*Excludes those axon terminals that are apposed to immunolabeled dendrites but that form synapses onto other structures. 
Number of dendrites 30 34 25 30 NA NA NA 
Mean (± SD) length of membrane/labeled dendrite (μm) 5.41 (3.06) 5.30 (2.54) 5.56 (2.30) 5.07 (2.99) F(1,115) = 0.008 P = 0.929 F(1,115) = 0.354 P = 0.553 F(1,115) = 0.145 P = 0.704 
Number of apposed axon terminals 94 90 72 61 NA NA NA 
Mean (± SD) length of apposed axon terminals (μm) 0.463 (0.193) 0.459 (0.154) 0.455 (0.333) 0.518 (0.299) F(1,313) = 0.876 P = 0.350 F(1,313) = 1.103 P = 0.294 F(1,313) = 1.423 P = 0.234 
Mean (± SD) number of apposed axon terminals/length of dendrite (μm)* 0.702 (0.252) 0.626 (0.298) 0.546 (0.264) 0.441 (0.306) F(1,115) = 10.69 P = 0.001 F(1,115) = 2.99 P = 0.087 F(1,115) = 0.077 P = 0.782 
Mean (±SD) proportion of dendritic length apposed by axon terminals* 0.332 (0.130) 0.289 (0.145) 0.234 (0.127) 0.214 (0.151) F(1,115) = 12.32 P = 0.001 F(1,115) = 1.54 P = 0.217 F(1,115) = 0.002 P = 0.964 
Mean (± SD) number of axon terminals forming Type I synapses/length of dendrite (μm) 0.279 (0.201) 0.286 (0.194) 0.201 (0.208) 0.147 (0.181) F(1,115) = 9.13 P = 0.003 F(1,115) = 0.413 P = 0.522 F(1,115) = 0.716 P = 0.399 
Mean (± SD) proportion of dendritic length apposed by axon terminals forming Type I synapses 0.130 (0.092) 0.129 (0.101) 0.096 (0.102) 0.068 (0.083) F(1,115) = 7.44 P = 0.007 F(1,115) = 0.735 P = 0.393 F(1,115) = 0.573 P = 0.451 
Figure 1.

(A) Schematic diagram of the dorsal surface of the macaque prefrontal cortex (PFC) illustrating the biotinylated dextran amine (BDA) injection sites in the three animals used in this study. The dashed line delineates Walker’s (1940) area 9 of the PFC. AS, arcuate sulcus; PS, principal sulcus. (B) Schematic diagram illustrating a single BDA injection site. The shaded area represents the stripe-like zone of cortex containing BDA-labeled axons and terminals arising from the core of the injection site (black circle). (C) Photomicrograph of the BDA injection site from CM223. Note the dark core of reaction product and the horizontally oriented, BDA-labeled axons (small arrows) that emanate from it. BDA-labeled neurons (arrowheads) are also visible. (D) Schematic diagram illustrating the sampling of local axon terminals. The BDA-labeled, local axon terminals were sampled in trapezoids cut from layers 2–3a and layer 3b of tissue sections located ~300 μm from core of the injection site. Note that terminals within this area may arise from either pyramidal or interneurons and, thus, may form either Gray’s Type I or II synapses.

Figure 1.

(A) Schematic diagram of the dorsal surface of the macaque prefrontal cortex (PFC) illustrating the biotinylated dextran amine (BDA) injection sites in the three animals used in this study. The dashed line delineates Walker’s (1940) area 9 of the PFC. AS, arcuate sulcus; PS, principal sulcus. (B) Schematic diagram illustrating a single BDA injection site. The shaded area represents the stripe-like zone of cortex containing BDA-labeled axons and terminals arising from the core of the injection site (black circle). (C) Photomicrograph of the BDA injection site from CM223. Note the dark core of reaction product and the horizontally oriented, BDA-labeled axons (small arrows) that emanate from it. BDA-labeled neurons (arrowheads) are also visible. (D) Schematic diagram illustrating the sampling of local axon terminals. The BDA-labeled, local axon terminals were sampled in trapezoids cut from layers 2–3a and layer 3b of tissue sections located ~300 μm from core of the injection site. Note that terminals within this area may arise from either pyramidal or interneurons and, thus, may form either Gray’s Type I or II synapses.

Figure 2.

Electron photomicrographs of BDA-containing, local axon terminals, labeled with peroxidase-diaminobenzidine, and PV- and CR-containing dendritic shafts labeled by immuno-gold, in monkey prefrontal cortex. (A) A BDA-labeled axon terminal forms a Type I synapse (arrow) onto a PV-labeled dendrite (PVd) in layers 2–3a. In addition, an unlabeled terminal forms a Type I synapse (arrowhead) onto the PVd. (B) In layer 3b, a BDA-labeled axon terminal forms a Gray’s Type I synapse (arrow) onto a CR-labeled dendritic shaft (CRd), which also receives a Type I synapse from an unlabeled terminal (arrowhead). (C) In layers 2–3a, a BDA-labeled, local axon terminal forms a Gray’s Type I synapse (arrow) onto an unlabeled dendritic spine in a field containing a CR-labeled dendritic shaft (CRd). Scale bar equals 0.40 μm in A and B and 0.22 μm in C.

Figure 2.

Electron photomicrographs of BDA-containing, local axon terminals, labeled with peroxidase-diaminobenzidine, and PV- and CR-containing dendritic shafts labeled by immuno-gold, in monkey prefrontal cortex. (A) A BDA-labeled axon terminal forms a Type I synapse (arrow) onto a PV-labeled dendrite (PVd) in layers 2–3a. In addition, an unlabeled terminal forms a Type I synapse (arrowhead) onto the PVd. (B) In layer 3b, a BDA-labeled axon terminal forms a Gray’s Type I synapse (arrow) onto a CR-labeled dendritic shaft (CRd), which also receives a Type I synapse from an unlabeled terminal (arrowhead). (C) In layers 2–3a, a BDA-labeled, local axon terminal forms a Gray’s Type I synapse (arrow) onto an unlabeled dendritic spine in a field containing a CR-labeled dendritic shaft (CRd). Scale bar equals 0.40 μm in A and B and 0.22 μm in C.

Figure 3.

Percentages of BDA-labeled local axon terminals that formed identifiable Gray’s Type I synapses onto immunolabeled dendrites (A), unlabeled dendrites (B) and unlabeled spines (C) in layers 2–3a and layer 3b of monkey prefrontal cortex.

Figure 3.

Percentages of BDA-labeled local axon terminals that formed identifiable Gray’s Type I synapses onto immunolabeled dendrites (A), unlabeled dendrites (B) and unlabeled spines (C) in layers 2–3a and layer 3b of monkey prefrontal cortex.

References

Arnsten AFT, Goldman-Rakic PS (
1998
) Noise stress impairs prefrontal cortical cognitive function in monkeys: evidence for a hyper- dopaminergic mechanism.
Arch Gen Psychiatry
 
55
:
362
–369.
Baddeley A (
1992
) Working memory.
Science
 
255
:
556
–559.
Chan J, Aoki C, Pickel VM (
1990
) Optimization of differential immunogold–silver and peroxidase labeling with maintenance of ultrastructure in brain sections before plastic embedding.
J Neurosci Methods
 
33
:
113
–127.
Condé F, Lund JS, Jacobowitz DM, Baimbridge KG, Lewis DA (
1994
) Local circuit neurons immunoreactive for calretinin, calbindin D-28k, or parvalbumin in monkey prefrontal cortex: distribution and morphology.
J Comp Neurol
 
341
:
95
–116.
Constantinidis C, Williams GV, Goldman-Rakic PS (
2002
) A role for inhibition in shaping the temporal flow of information in prefrontal cortex.
Nat Neurosci
 
5
:
175
–180.
Daviss SR, Lewis DA (
1995
) Local circuit neurons of the prefrontal cortex in schizophrenia: selective increase in the density of calbindin- immunoreactive neurons.
Psychiatry Res
 
59
:
81
–96.
DeFelipe J (
1997
) Types of neurons, synaptic connections and chemical characteristics of cells immunoreactive for calbindin-D28K, parv- albumin and calretinin in the neocortex.
J Chem Neuro
 
14
:
1
–19.
DeFelipe J, Jones EG (
1992
) High-resolution light and electron micro- scopic immunocytochemistry of colocalized GABA and calbindin D–28k in somata and double bouquet cell axons of monkey somatosensory cortex.
Eur J Neurosci
 
4
:
46
–60.
DeFelipe J, Hendry SHC, Jones EG, Schmechel D (
1985
) Variability in the terminations of GABAergic chandelier cell axons on initial segments of pyramidal cell axons in the monkey sensory–motor cortex.
J Comp Neurol
 
231
:
364
–384.
Freund TF, Martin KAC, Smith AD, Somogyi P (
1983
) Glutamate decarboxylase-immunoreactive terminals of Golgi-impregnated axoaxonic cells and of presumed basket cells in synaptic contact with pyramidal neurons of the cat’s visual cortex.
J Comp Neurol
 
221
:
263
–278.
Freund TF, Buszáki G, Leon A, Baimbridge KG, Somogyi P (
1990
) Relationship of neuronal vulnerability and calcium binding protein immunoreactivity in ischemia.
Exp Brain Res
 
83
:
55
–66.
Fuster JM (
1997
) The prefrontal cortex: anatomy, physiology, and neuropsychology of the frontal lobe. Philadelphia: Lippincott–Raven.
Gabbott PLA, Bacon SJ (
1996
) Local circuit neurons in the medial prefrontal cortex (areas 24a,b,c, 25 and 32) in the monkey: I. Cell morphology and morphometrics.
J Comp Neurol
 
364
:
567
–608.
Gabbott PLA, Bacon SJ (
1996
) Local circuit neurons in the medial prefrontal cortex (areas 24a,b,c, 25 and 32) in the monkey: II. Quantitative areal and laminar distributions.
J Comp Neurol
 
364
:
609
–636.
Goldman-Rakic PS (
1987
) Circuitry of primate prefrontal cortex and regulation of behavior by representational memory. In: Handbook of physiology (Plum F, Mountcastle V, eds), pp. 373–417. Bethesda, MD: American Physiological Society.
Goldman-Rakic PS (
1995
) Cellular basis of working memory.
Neuron
 
14
:
477
–485.
Goldman-Rakic PS, Schwartz ML (
1982
) Interdigitation of contralateral and ipsilateral columnar projections to frontal association cortex in primates.
Science
 
216
:
755
–757.
Gonchar Y, Burkhalter A (
1999
) Connectivity of GABAergic calretinin-immunoreactive neurons in rat primary visual cortex.
Cereb Cortex
 
9
:
683
–696.
González-Burgos G, Krimer LS, Barrionuevo G, Lewis DA (
2002
) Dynamics of fast spiking interneuron-mediated synaptic inhibition in monkey dorsolateral prefrontal cortex.
Soc Neurosci Abstr
 
28
:
147.4
.
Gray EG (
1959
) Axo-somatic and axo-dendritic synapses of the cerebral cortex: an electron microscopic study.
J Anat
 
93
:
420
–433.
Gulyás AI, Hájos N, Freund TF (
1996
) Interneurons containing calretinin are specialized to control other interneurons in the rat hippocampus.
J Neurosci
 
16
:
3397
–3411.
Gulyás AI, Megías M, Emri Z, Freund TF (
1999
) Total number and ratio of excitatory and inhibitory synapses converging onto single inter- neurons of different types in the CA1 area of the rat hippocampus.
J Neurosci
 
19
:
10082
–10097.
Kawaguchi Y (
1993
) Groupings of nonpyramidal and pyramidal cells with specific physiological and morphological characteristics in rat frontal cortex.
J Neurophysiol
 
69
:
416
–431.
Kawaguchi Y (
1995
) Physiological subgroups of nonpyramidal cells with specific morphological characteristics in layer II/III of rat frontal cortex.
J Neurosci
 
15
:
2638
–2655.
Kawaguchi Y, Kubota Y (
1997
) GABAergic cell subtypes and their synaptic connections in rat frontal cortex.
Cereb Cortex
 
7
:
476
–486.
Kisvárday ZF, Martin KAC, Freund TF, Magloczky Zs, Whitteridge D, Somogyi P (
1986
) Synaptic targets of HRP-filled layer III pyramidal cells in the cat striate cortex.
Exp Brain Res
 
64
:
541
–552.
Krimer LS, Goldman-Rakic PS (
2001
) Prefrontal microcircuits: membrane properties and excitatory input of local, medium, and wide arbor neurons.
J Neurosci
 
21
:
3788
–3796.
Krimer LS, Gonzalez-Burgos G, Lewis DA, Barrionuevo G (
2002
) Inter- neurons of monkey prefrontal cortex from laminae 2–3: physiological and anatomical properties.
Soc Neurosci Abstr
 
27
:
729.14
.
Kritzer MF, Goldman-Rakic PS (
1995
) Intrinsic circuit organization of the major layers and sublayers of the dorsolateral prefrontal cortex in the rhesus monkey.
J Comp Neurol
 
359
:
131
–143.
Lewis DA, Gonzalez-Burgos G (
1999
) Intrinsic excitatory connections in the prefrontal cortex and the pathophysiology of schizophrenia.
Brain Res Bull
 
52
:
309
–317.
Lund JS, Lewis DA (
1993
) Local circuit neurons of developing and mature macaque prefrontal cortex: Golgi and immunocytochemical characteristics.
J Comp Neurol
 
328
:
282
–312.
McGuire BA, Gilbert CD, Rivlin PK, Wiesel TN (
1991
) Targets of horizontal connections in macaque primary visual cortex.
J Comp Neurol
 
305
:
370
–392.
Melchitzky DS, Sesack SR, Pucak ML, Lewis DA (
1998
) Synaptic targets of pyramidal neurons providing intrinsic horizontal connections in monkey prefrontal cortex.
J Comp Neurol
 
390
:
211
–224.
Melchitzky DS, Gonzalez-Burgos G, Barrionuevo G, Lewis DA (
2001
) Synaptic targets of the intrinsic axon collaterals of supragranular pyramidal neurons in monkey prefrontal cortex.
J Comp Neurol
 
430
:
209
–221.
Meskenaite V (
1997
)
Calretinin-immunoreactive local circuit neurons in area
 
17 of the cynomolgus monkey, Macaca fascicularis. J Comp Neurol 379
:
113
–132.
Muly EC, Szigeti K, Goldman-Rakic PS (
1998
) D1 receptor in interneurons of macaque prefrontal cortex: distribution and subcellular location.
J Neurosci
 
18
:
10553
–10565.
Murphy BL, Arnsten AFT, Goldman-Rakic PS, Roth RH (
1996
) Increased dopamine turnover in the prefrontal cortex impairs spatial working memory performance in rats and monkeys.
Proc Natl Acad Sci USA
 
93
:
1325
–1329.
Peters A, Palay SL, Webster DF (
1991
) The fine structure of the nervous system. New York: Oxford University Press.
Pickel VM, Chan J, Aoki C (
1993
) Electron microscopic immuno- cytochemical labeling of endogenous and/or transported antigens in rat brain using silver-intensified one-nanometre colloidal gold. In: Immunohistochemistry II (Cuello AC, ed.), pp. 265–280. Chichester: John Wiley & Sons.
Pucak ML, Levitt JB, Lund JS, Lewis DA (
1996
) Patterns of intrinsic and associational circuitry in monkey prefrontal cortex.
J Comp Neurol
 
376
:
614
–630.
Rao SG, Williams GV, Goldman-Rakic PS (
1999
) Isodirectional tuning of adjacent interneurons and pyramidal cells during working memory: evidence for microcolumnar organization in PFC.
J Neurophysiol
 
81
:
1903
–1916.
Rao SG, Williams GV, Goldman-Rakic PS (
2000
) Destruction and creation of spatial tuning by disinhibition: GABAA blockade of prefrontal cortical neurons engaged by working memory.
J Neurosci
 
20
:
485
–494.
Sawaguchi T, Goldman-Rakic PS (
1991
) D1 dopamine receptors in prefrontal cortex: involvement in working memory.
Science
 
251
:
947
–950.
Sawaguchi T, Matsumura M, Kubota K (
1988
) Delayed response deficit in monkeys by locally disturbed prefrontal neuronal activity by bicuculline.
Behav Brain Res
 
31
:
193
–198.
Sawaguchi T, Matsumura M, Kubota K (
1989
) Delayed response deficits produced by local injection of bicuculline into the dorsolateral prefrontal cortex in Japanese macaque monkeys.
Exp Brain Res
 
75
:
457
–469.
Sesack SR, Bressler CN, Lewis DA (
1995
) Ultrastructural associations between dopamine terminals and local circuit neurons in the monkey prefrontal cortex: a study of calretinin-immunoreactive cells.
Neurosci Lett
 
200
:
9
–12.
Sesack SR, Snyder CL, Lewis DA (
1995
) Axon terminals immunolabeled for dopamine or tyrosine hydroxylase synapse on GABA-immuno- reactive dendrites in rat and monkey cortex.
J Comp Neurol
 
363
:
264
–280.
Sesack SR, Hawrylak VA, Melchitzky DS, Lewis DA (
1998
) Dopamine innervation of a subclass of local circuit neurons in monkey prefrontal cortex: ultrastructural analysis of tyrosine hydroxylase and parv- albumin immunoreactive structures.
Cereb Cortex
 
8
:
614
–622.
Smiley JF, Goldman-Rakic PS (
1993
) Heterogeneous targets of dopamine synapses in monkey prefrontal cortex demonstrated by serial section electron microscopy: a laminar analysis using the silver-enhanced diaminobenzidine sulfide (SEDS) immunolabeling technique.
Cereb Cortex
 
3
:
223
–238.
Somogyi P (
1977
) A specific axo-axonal interneuron in the visual cortex of the rat.
Brain Res
 
136
:
345
–350.
Staiger JF, Zilles K, Freund TF (
1996
) Distribution of GABAergic elements postsynaptic to ventroposteromedial thalamic projections in layer IV of rat barrel cortex.
Eur J Neurosci
 
8
:
2273
–2285.
Walker AE (
1940
) A cytoarchitectural study of the prefrontal area of the macaque monkey.
J Comp Neurol
 
73
:
59
–86.
Wang XJ (
2001
) Synaptic reverberation underlying mnemonic persistent activity.
Trends Neurosci
 
24
:
455
–463.
Williams GV, Goldman-Rakic PS (
1995
) Modulation of memory fields by dopamine D1 receptors in prefrontal cortex.
Nature
 
376
:
572
–575.
Williams SM, Goldman-Rakic PS, Leranth C (
1992
) The synaptology of parvalbumin-immunoreactive neurons in primate prefrontal cortex.
J Comp Neurol
 
320
:
353
–369.
Wilson FA, O Scalaidhe SP, Goldman-Rakic PS (
1994
) Functional syner- gism between putative gamma-aminobutyrate-containing neurons and pyramidal neurons in prefrontal cortex.
Proc Natl Acad Sci USA
 
91
:
4009
–4013.