Synaptic targets of pyramidal neurons providing intrinsic horizontal connections in monkey prefrontal cortex
Post on 06-Jun-2016
Synaptic Targets of Pyramidal NeuronsProviding Intrinsic Horizontal
Connections in MonkeyPrefrontal Cortex
DARLENE S. MELCHITZKY,1 SUSAN R. SESACK,1,2 MICHELE L. PUCAK,1
AND DAVID A. LEWIS1,2*1Department of Psychiatry, University of Pittsburgh, Pittsburgh, Pennsylvania 15213
2Department of Neuroscience, University of Pittsburgh, Pittsburgh, Pennsylvania 15260
ABSTRACTIn monkey prefrontal cortex, the intrinsic axon collaterals of supragranular pyramidal
neurons extend horizontally for considerable distances through the gray matter and give riseto stripe-like clusters of axon terminals (Levitt et al.  J. Comp. Neurol. 338:360376).Because understanding the functional role of these connections requires knowledge of theirsynaptic targets, we made injections of biotinylated dextran amine (BDA) into layer 3 ofmacaque prefrontal area 9 and examined the labeled intrinsic axon collaterals by electronmicroscopy. Labeled axon terminals formed exclusively asymmetric synapses, and 95.6% ofthe postsynaptic structures were dendritic spines, presumably belonging to other pyramidalneurons. The remaining postsynaptic structures were dendritic shafts, many of which had themorphological characteristics of local circuit neurons. The prefrontal injections also labeledassociational projections that traveled through the white matter to terminate in other areas ofprefrontal cortex. All of the synapses formed by these associational axons were asymmetric,and 91.9% were onto dendritic spines. The similarities in synaptic targets of the prefrontalintrinsic and associational axon terminals suggested that these projections might arise fromthe same neurons, an interpretation confirmed in dual label, retrograde tracing studies.
To determine the specificity of the synaptic targets of these prefrontal connections, twoadditional comparisons were made. In the posterior parietal cortex (area 7a), 94.2% of thesynapses furnished by BDA-labeled intrinsic collaterals of supragranular pyramidal neuronswere also with dendritic spines. In contrast, only 75.6% of unlabeled asymmetric synapses inthe prefrontal cortex were onto dendritic spines. These comparisons suggest that the axons ofsupragranular pyramidal neurons in primate association cortices are preferentially directedto specific targets.
Finally, after injections of BDA, a small number of retrogradely labeled pyramidalneurons were observed within the anterogradely labeled clusters of intrinsic axon terminals.At the ultrastructural level, synapses between anterogradely labeled axon terminals andretrogradely labeled dendritic spines were identified. These findings suggest that reciprocal,monosynaptic connections may exist between pyramidal neurons located in different stripe-like clusters, providing a potential anatomical substrate for reverberating excitatory circuitswithin the primate association cortices. J. Comp. Neurol. 390:211224, 1998. r 1998 Wiley-Liss, Inc.
Indexing terms: axon collateral; asymmetric synapse; posterior parietal cortex; local circuit neuron
In the primate brain, the prefrontal cortex (PFC) sub-serves a number of higher-order cognitive functions, includ-ing those involving working memory (Goldman-Rakic,1987; Baddeley, 1992) and the temporal integration ofinformation (Fuster, 1985). Disturbances of these abilitiesin disease states, such as schizophrenia and Alzheimers
Grant sponsor: USPHS; Grant numbers:AG05133, MH51234, and MH45156;Grant sponsor: NIMH Independent Scientist Award; Grant number: MH00519.
*Correspondence to: David A. Lewis, Western Psychiatric Institute andClinic, University of Pittsburgh, Biomedical Science Tower, W1651, 3811OHara Street, Pittsburgh, PA 15213.E-mail: firstname.lastname@example.org
Received 13 May 1997; Revised 28 July 1997; Accepted 8 August 1997
THE JOURNAL OF COMPARATIVE NEUROLOGY 390:211224 (1998)
r 1998 WILEY-LISS, INC.
disease, have been associated with abnormalities in pyra-midal neurons, especially those located in supragranularlayers 2 and 3 (Hof et al., 1990; Glantz and Lewis, 1995).The involvement of these excitatory neurons in the patho-physiology of cognitive disorders may be related, at least inpart, to the fact that the majority of their extrinsic axonalprojections are directed to other cortical association re-gions (Schwartz and Goldman-Rakic, 1984; Barbas andPandya, 1989). In addition, the intrinsic axon collaterals ofsupragranular pyramidal neurons in monkey PFC spreadhorizontally for considerable distances and give rise todiscrete, stripe-like clusters of axon terminals that spanlayers 13 (Levitt et al., 1993). Retrograde labeling studies(Kritzer and Goldman-Rakic, 1995; Pucak et al., 1996)have demonstrated that the neurons that contribute axoncollaterals to this intrinsic circuitry are also arranged in astripe-like fashion. These clusters of retrogradely labeledneurons are coregistered with anterogradely labeled axonterminals arising from the same injection site (Pucak etal., 1996), suggesting that reciprocal intrinsic connectionslink spatially segregated clusters of supragranular pyrami-dal cells into neuronal networks (Goldman-Rakic, 1995;Lewis and Anderson, 1995).
These networks of supragranular pyramidal neurons inmonkey PFC have been suggested to provide a means forcoordinating the activity of neuronal populations thatshare the same response properties (Levitt et al., 1993;Goldman-Rakic, 1995), analogous to the apparent role ofthe intrinsic horizontal connections in sensory cortices(Lund et al., 1993). For example, in the primary visualcortex, horizontally oriented axon collaterals predomi-nantly interconnect orientation domains of similar prefer-ences, as well as zones of similar ocular dominance (Tso etal., 1986; Gilbert and Wiesel, 1989; Malach et al., 1993;Yoshioka et al., 1996). In addition, the sustained activity ofspecific populations of prefrontal pyramidal neurons dur-ing the delay period of delayed response tasks (Fuster etal., 1982; Funahashi et al., 1989) may depend, at least inpart, on reciprocal intrinsic connections (Lewis and Ander-son, 1995; Pucak et al., 1996). That is, these connectionshave been suggested to provide the anatomical substratefor a reverberating excitatory circuit that maintains thefiring of prefrontal neurons in the absence of externalstimulation, a critical feature of working memory (Funaha-shi et al., 1989; Goldman-Rakic, 1995). However, furthertesting of this hypothesis requires knowledge of the synap-tic targets of these intrinsic axon collaterals.
In other species, the intrinsic axon collaterals of differ-ent subpopulations of pyramidal neurons preferentiallyform synaptic contacts with certain types of structures(Czeiger and White, 1993). For example, in mouse somato-sensory cortex, over 80% of the intrinsic axon collateralsarising from callosally projecting neurons target dendriticspines (Elhanany and White, 1990), whereas dendriticshafts are the postsynaptic targets of over 90% of theintrinsic axon collaterals of corticothalamic neurons (Whiteand Keller, 1987). However, the postsynaptic targets of theintrinsic axon collaterals of supragranular pyramidal neu-rons in primate PFC are not known. Consequently, in thisstudy, we placed injections of the tracer biotinylateddextran amine (BDA) into the superficial layers of monkeyPFC to identify the structures that receive synaptic inputfrom the intrinsic axon collaterals of supragranular pyra-midal neurons. In addition, the relative specificity of thesetargets was assessed through comparisons with the neural
elements that receive synaptic input from other types ofcortical excitatory terminals, both within the PFC and inanother higher-order association region, the posterior pari-etal cortex.
MATERIALS AND METHODSInjections of biotinylated dextran amine
Surgical procedures. Four male, adult cynomolgusmonkeys (Macaca fascicularis) were used in this study. Allanimals were treated according to the guidelines outlinedin the National Institutes of Health Guide for the Care andUse of Laboratory Animals. After injections of ketaminehydrochloride (25 mg/kg), dexamethasone phosphate (0.5mg/kg), and atropine sulfate (0.05 mg/kg), an endotrachealtube was inserted, and the animal was placed in a stereo-taxic apparatus. Anesthesia was maintained with 1%halothane in 28% O2/air. Guided by stereotaxic coordinates(Szabo and Cowan, 1984), a craniectomy was performedover either the dorsal prefrontal or the posterior parietalcortex. By using a 5.0-l Hamilton syringe (26-gaugeneedle), one injection (0.3 l) of 10% BDA (10,000 molecu-lar weight; Molecular Probes, Inc., Eugene, OR) in 0.01 Mphosphate buffer, pH 7.3, was placed into either dorsalarea 9 or area 7a (Fig. 1A). For each animal, one injectionwas made at a depth of approximately 1.0-mm below thepial surface. After tracer injections, the scalp was closed,and the animals were treated with an antibiotic (chloram-phenicol, 15 mg/kg) and an analgesic (hydromorphone,0.02 mg/kg).
Tissue preparation. After a survival time of 8 to12days, monkeys were deeply anesthetized with ketaminehydrochloride (25 mg/kg) and pentobarbital sodium (30mg/kg) and then perfused transcardially with cold 4%paraformaldehyde in phosphate buffer (Pucak et al., 1996).The brain was removed, and coronal blocks (4-mm-thick)were immersed in 0.12 M phosphate buffer, pH 7.3,containing 4% paraformaldehyde and 0.2% glutaralde-hyde, for 2 hours. Tissue blocks were then washed in 0.1 Mphosphate buffer, pH 7.3, and sectioned on a Vibratome at50 m.
Histochemistry. The histochemical procedures usedto visualize BDA were adapted from those previouslydescribed (Pucak et al., 1996; Woo et al., 1997). Briefly,tissue sections were incubated in 0.05 M phosphate-buffered saline (PBS) containing 4.5% normal humanserum (NHuS), 0.04% Triton X-100, and 0.05 mg/ml bovineserum albumin (BSA) and then placed in 0.05 M PBS, pH7.3, containing avidin-biotin complex (ABC) reagents (Vec-tastain Elite, Vector Laboratories, Burlingame, CA) and4.5% NHuS for approximately 16 hours at 4C. Sectionswere then rinsed and incubated in 0.1 M phosphate buffer,pH 7.3, containing 0.5 mg/ml diaminobenzidine and 0.006%H2O2 for 5 minutes. After a rinse in 0.1 M phosphate buffer,pH 7.3, sections for electron microscopy were post-fixed in2% osmium tetroxide in 0.1 M phosphate buffer pH 7.3 for1 hour, dehydrated, and embedded in Epon 812. Ultrathinsections were collected, counterstained with uranyl ac-etate and lead citrate, and examined on a Zeiss or JEOLelectron microscope. Sections for light microscopy weremounted onto gel-coated slides and intensified with silveras previously described (Pucak et al., 1996).
212 D.S. MELCHITZKY ET AL.
Definition of neuronal and synaptic elementsNeuronal elements of relevance to this study were
identified by using the criteria of Peters et al. (1991). Axonterminals were generally greater than 0.2 m in diameterand contained synaptic vesicles and, occasionally, mitochon-dria. Dendritic shafts were characterized by the presenceof mitochondria, numerous microtubules, and neurofila-ments, as well as postsynaptic specializations. Dendriticspines were identified by the absence of mitochondria andmicrotubules and by the presence of a spine apparatus (inoptimal planes of section). In addition, dendritic spinesinvariably received asymmetric synapses, which werecharacterized by widened and parallel spacing of apposedplasmalemmal surfaces, and a thick postsynaptic density.The axon terminals forming these synapses always con-tained round synaptic vesicles.
Sampling regions and proceduresAs previously described (Pucak et al., 1996), BDA-
labeled axon terminals within the PFC were organizedinto distinct clusters that could be identified as originatingfrom intrinsic or associational axon projections. Intrinsicclusters were composed of arborizing axons and terminalsconfined to layers 13, were associated with labeled preter-minal horizontally oriented axons in the adjacent graymatter, and lacked labeled axons in the underlying whitematter (Figs. 1B, 2A). In contrast, associational clustershad arborizing axons and terminals in both the superficialand deep cortical layers (although the density of labeledaxons in different layers varied across regions), lacked
labeled preterminal axons in the gray matter and ap-peared to arise from labeled axons in the underlying whitematter (Figs. 1B, 2B). Finally, over 95% of the axonclusters with the characteristics of intrinsic projectionswere located within 6 mm of the injection site, whereas90% of the associational clusters were located more than 8mm from the injection site (Pucak et al., 1996).
Three types of labeled axon terminals were examined inthis study: intrinsic prefrontal, associational prefrontaland intrinsic parietal. For each type of terminal, blocks forelectron microscopic analysis were taken from two to fivedifferent clusters of labeled axon terminals per animal. Allintrinsic clusters selected for study were located 1 to 5 mmfrom the injection site. Within each cluster, blocks werecentered in layers 2 and 3. For each block, three ultrathinsections, separated by 10 to 20 sections, were examined.The analysis consisted of identifying all profiles of antero-gradely labeled axon terminals within the entire ultrathinsection, and determining which terminals had synapticspecializations. For those terminals forming synaptic spe-cializations, the postsynaptic structures were recorded.
As an additional check on the relative specificity of thesynaptic targets of the prefrontal axons, unlabeled axonterminals from the general prefrontal neuropil were alsoanalyzed. Random fields from the sections used for sam-pling the intrinsic and the associational prefrontal connec-tions were photographed at 319,000. The total number ofunlabeled asymmetric synaptic profiles in each field werecounted, and the structures postsynaptic to them wererecorded.
Fig. 1. A: Schematic drawing of the lateral view of the cynomolgusmonkey brain showing the approximate location of each biotinylateddextran amine (BDA) injection site in the four animals used in thisstudy. In each animal, one injection was placed in either prefrontalarea 9 or parietal area 7a. AS, arcuate sulcus; CS, central sulcus; IPS,intraparietal sulcus; PS, principal sulcus; STS, superior temporalsulcus. B: Schematic drawing of a coronal section through the
prefrontal cortex (PFC; at the level indicated by the arrow in panel A)illustrating the cytoarchitectonic areas and types of axon clusterssampled. After an injection of BDA in area 9 (closed circle), clusters ofintrinsic axon terminals (i; see Fig. 3A) were sampled from area 9,whereas clusters of associational axon terminals (a; see Fig. 3B) weresampled from area 46. Note that injection site size and neuronalelements are not shown to scale. Scale bars 5 ,1 cm in A, 5 mm in B.
EXCITATORY CONNECTIONS IN MONKEY PREFRONTAL CORTEX 213
Individual 2 3 2 x2 analyses were performed to comparethe synaptic targets of the intrinsic prefrontal axon te...