Axo-axonic synapses formed by somatostatin

THE JOURNAL OF COMPARATIVE NEUROLOGY 443:1–14 (2002)
Axo-Axonic Synapses Formed by
Somatostatin-Expressing GABAergic
Neurons in Rat and Monkey
Visual Cortex
YURI GONCHAR,* STEPHEN TURNEY, JOSEPH L. PRICE, AND
ANDREAS BURKHALTER
Department of Anatomy and Neurobiology, 8108, Washington University
School of Medicine, St. Louis, Missouri 63110
ABSTRACT
In cerebral cortex of rat and monkey, the neuropeptide somatostatin (SOM) marks a population of nonpyramidal cells (McDonald et al. [1982] J. Neurocytol. 11:809 – 824; Hendry et al. [1984]
J. Neurosci. 4:2497:2517; Laemle and Feldman [1985] J. Comp. Neurol. 233:452– 462; Meineke and
Peters [1986] J. Neurocytol. 15:121–136; DeLima and Morrison [1989] J. Comp. Neurol. 283:212–
227) that represent a distinct type of ␥-aminobutyric acid (GABA) -ergic neuron (Gonchar and
Burkhalter [1997] Cereb. Cortex 7:347–358; Kawaguchi and Kubota [1997] Cereb. Cortex 7:476 –
486) whose synaptic connections are incompletely understood. The organization of inhibitory
inputs to the axon initial segment are of particular interest because of their role in the suppression
of action potentials (Miles et al. [1996] Neuron 16:815:823). Synapses on axon initial segments are
morphologically heterogeneous (Peters and Harriman [1990] J. Neurocytol. 19:154 –174), and some
terminals lack parvalbumin (PV) and contain calbindin (Del Rio and DeFelipe [1997] J. Comp.
Neurol. 342:389 – 408), that is also expressed by many SOM-immunoreactive neurons (Kubota et
al. [1994] Brain Res. 649:159 –173; Gonchar and Burkhalter [1997] Cereb. Cortex 7:347–358). We
studied the innervation of pyramidal neurons by SOM neurons in rat and monkey visual cortex and
examined putative contacts by confocal microscopy and determined synaptic connections in the
electron microscope. Through the confocal microscope, SOM-positive boutons were observed to
form close appositions with somata, dendrites, and spines of intracortically projecting pyramidal
neurons of rat area 17 and pyramidal cells in monkey striate cortex. In addition, in rat and
monkey, SOM boutons were found to be associated with axon initial segments of pyramidal
neurons. SOM axon terminals that were apposed to axon initial segments of pyramidal neurons
lacked PV, which was shown previously to label axo-axonic terminals provided by chandelier cells
(DeFelipe et al. [1989] Proc. Natl. Acad. Sci. USA 86:2093–2097; Gonchar and Burkhalter [1999a]
J. Comp. Neurol. 406:346:360). Electron microscopic examination directly demonstrated that SOM
axon terminals form symmetric synapses with the initial segments of pyramidal cells in supragranular layers of rat and monkey primary visual cortex. These SOM synapses differed ultrastructurally from the more numerous unlabeled symmetric synapses found on initial segments. Postembedding immunostaining revealed that all SOM axon terminals contained GABA. Unlike PVexpressing chandelier cell axons that innervate exclusively initial segments of pyramidal cell
axons, SOM-immunoreactive neurons innervate somata, dendrites, spines, and initial segments,
that are just one of their targets. Thus, SOM neurons may influence synaptic excitation of
pyramidal neurons at the level of synaptic inputs to dendrites as well as at the initiation site of
action potential output. J. Comp. Neurol. 443:1–14, 2002. © 2002 Wiley-Liss, Inc.
Indexing terms: somatostatin; GABAergic synapses; cortex; axon initial segment
Cortical circuits are formed by the connections of pyramidal and nonpyramidal cells (Douglas and Martin, 1991).
Nonpyramidal cells differ from pyramidal cells by their
distinctive dendritic trees and the scarcity of spines
(Fairen et al., 1984). Most nonpyramidal cells synthesize
or contain ␥-aminobutyric acid (GABA) (Ribak, 1978; Somogyi and Hodgson, 1985), which they use as neurotransmitter in axon terminals that contain flattened vesicles
and form symmetric synapses (Somogyi and Soltesz, 1986;
© 2002 WILEY-LISS, INC.
DOI 10.1002/cne.1425
Grant sponsor: NIH; Grant number: EY05935; Grant number: NS30676;
Grant sponsor: the McDonnell Foundation.
*Correspondence to: Dr. Yuri Gonchar, Department of Anatomy and
Neurobiology, 8108, Washington University School of Medicine, 660 South
Euclid Avenue, St. Louis, MO 63110. E-mail: [email protected]
Received 16 February 2001; Revised 18 May 2001; Accepted 25 June 2001
Published online the week of December 17, 2001
2
Thomson et al., 1996). Many GABAergic synapses make
connections with pyramidal cells (Johnson and Burkhalter, 1996), and when activated, they suppress excitation in
postsynaptic neurons (Ferster and Jagadeesh, 1992). The
inhibitory effects may involve hyperpolarization and
blockade of action potential firing (Miles et al., 1996) or
may lead to changes in membrane conductance that reduce the amplitude of excitatory responses (Borg-Graham
et al., 1998). Although, it is known for some time that
interneurons are morphologically heterogeneous (Fairen
et al., 1984), the identification of different types of
GABAergic neurons with distinct intrinsic membrane
properties (Kawaguchi, 1995), discrete molecular characteristics (DeFelipe, 1993; Weiser et al., 1995; Kawaguchi
and Kubota, 1996; Cauli et al., 1997) and diverse synaptic
dynamics (Thomson et al., 1996; Reyes et al., 1998; Gupta
et al., 2000) suggests that different neurons may produce
different postsynaptic effects.
In rat visual cortex, immunostaining for PV, SOM, and
calretinin (CR) revealed three separate families of
GABAergic neurons (Gonchar and Burkhalter, 1997). PV
is expressed in basket cells that preferentially innervate
somata and proximal dendrites and, in addition, is contained in chandelier cells that provide synaptic inputs to
axon initial segments of pyramidal cells (Somogyi, 1977;
Peters et al., 1982; Somogyi et al., 1982; DeFelipe et al.,
1989; Kisvarday et al., 1993; Gonchar and Burkhalter,
1999a: DeFelipe, 1999). CR neurons predominantly innervate other inhibitory nonpyramidal neurons (Gulyas et
al., 1996; Meskenaite, 1997; Gonchar and Burkhalter,
1999b), whereas SOM neurons mainly innervate thin dendritic shafts and spines of pyramidal cells (Hendry et al.,
1984; Meinecke and Peters, 1986; DeLima and Morrison,
1989; Kawaguchi and Kubota, 1997). Although different
inhibitory neurons have dissimilar preferences for innervating pyramidal and nonpyramidal cells, most known
types of nonpyramidal cells form synapses on pyramidal
cell somata and dendrites (Kawaguchi and Kubota, 1997).
One exception is that of PV-containing chandelier cells
that are thought to form exclusively connections with axon
initial segments of pyramidal cells (Somogyi, 1977; DeFelipe et al., 1989; Williams et al., 1992). Based on these
observations, it was proposed that all inputs to axon initial segments originate from chandelier cells (Somogyi,
1977). Several reports, however, suggest that axo-axonic
synapses in rat and monkey cerebral cortex are morphologically heterogeneous and that subpopulations of axon
terminals lack PV and express corticotropin releasing factor or calbindin (DeFelipe et al., 1985; Lewis and Lund,
1990; Peters and Harriman, 1990; Del Rio and DeFelipe,
1997), raising the possibility that in addition to inputs
from PV neurons axon initial segments may be innervated
by other cell types. Synapses at the axon initial segment,
the site where action potentials are generated (Stuart and
Sakmann, 1994), are in a privileged position to control the
output of pyramidal cells. Because different interneurons
have diverse physiological properties (Kawaguchi and
Kubota, 1996), it is important to find out which type(s) of
GABAergic neuron(s) contributes to the innervation of the
axon initial segment.
In human temporal cortex a subset of terminals that
form axo-axonic synapses contain calbindin (Del Rio and
DeFelipe, 1997). Studies in rat cerebral cortex have demonstrated that calbindin is expressed in pyramidal cells
and GABAergic neurons. Some of these GABAergic neu-
Y. GONCHAR ET AL.
rons contain SOM (Kubota et al., 1994; Gonchar and
Burkhalter, 1997). Therefore, SOM neurons might be involved in the innervation of the axon initial segment of
pyramidal neurons. To test this hypothesis, we labeled
SOM terminals and studied the distribution of synaptic
inputs to axon initial segments of pyramidal neurons by
using confocal and electron microscopy. The results show
that SOM neurons innervate axon initial segments, suggesting that SOM axo-axonic synapses participate in a
facilitating form of local inhibition (Thomson and Deuchars, 1997; Reyes et al., 1998; Gibson et al., 1999).
MATERIALS AND METHODS
Experiments were performed on sixteen 6- to 8-week-old
Long-Evans rats (150 –200 g body weight). Tissue from
primate visual cortex was obtained from perfusion fixed
brains of three adult monkeys (Macaca fascicularis) that
were used in experiments that were unrelated to the
present study. All experimental protocols were approved
by the Animals Studies Committee of Washington University and conformed to NIH guidelines and regulations.
Identification of pyramidal neurons
Retrograde labeling of pyramidal neurons. Rats
were anesthetized with ketamine (87 mg/kg body weight)
and xylazine (14 mg/kg) and secured in a headholder. For
retrograde tracing of projection neurons, glass micropipets
(tip diameter, 15 ␮m) were lowered through a small hole
in the skull 1 mm deep into the lateromedial area (LM).
Biotinylated dextran amine (10%, 10,000 molecular
weight, BDA [Molecular Probes, Eugene, OR] in 0.01 mM
phosphate buffer, pH 7.4, 20 nl) was injected by applying
puffs of pressurized air to the back of the pipet, by using a
Picospritzer (General Valve, Fairfax, NJ). To maximize
retrograde filling with BDA, tracer injections were preceded by an injection of 50 nl of N-methyl-D-aspartic acid
(10 mM NMDA) at the same site (Jiang et al., 1993). After
48 hours, survival rats were perfused through the left
ventricle with heparinized (100 units/ml) phosphatebuffered saline (PBS, pH 7.4) followed by a solution containing 4% paraformaldehyde and 0.25% glutaraldehyde
in 0.1 M phosphate buffer (PB, pH 7.4). Brains were removed from the skull and post-fixed for 2 hours at 4°C in
the same fixative. Sections of the visual cortex were cut on
a Vibratome at 25 ␮m and collected in PBS (pH 7.4). The
tissue was then treated with sodium borohydride (1% in
PBS, 1 hour), H2O2 (0.1%, 30 minutes), ethanol (50%, 30
minutes), and transferred for overnight incubation in
avidin-neutralite-Texas Red (1:300; Molecular Probes). A
complete series of sections was mounted on gelatinized
slides, air-dried, dehydrated in ethanols, cleared in citrus
based clearing agent, and cover-slipped in Krystallon
(Harleco, EM Science, Gibbstown, NJ).
Immunofluorescence staining for SMI-32. Pyramidal cells in monkey visual cortex were visualized with an
antibody against nonphosphorylated neurofilaments
(Campbell and Morrison, 1989; Hof et al., 1996). For this
purpose, tissue blocks from occipital cortex of monkeys
perfused with PBS followed by 4% paraformaldehyde and
0.25% glutaraldehyde in PB and post-fixed for 2 hours
were sectioned at 40 ␮m on a Vibratome. Sections were
pretreated with borohydride, H2O2 and ethanol as described above and incubated for 36 hours in SMI-32 monoclonal antibody (1:1,000; Sternberger Monoclonals, Balti-
CONNECTIONS OF SOMATOSTATIN NEURONS IN VISUAL CORTEX
more, MD). The sections were then incubated for 2 hours
in Cy-3–labeled goat anti-mouse secondary antibody (1:
400, Chemicon, Temecula, CA).
Confocal microscopy
Inputs to axon initial segments of pyramidal cells.
Sections from rat and monkey visual cortex that contained
retrogradely BDA-labeled neurons or SMI-32–immunoreactive pyramidal cells, respectively, were used for additional immunofluorescence staining with antibodies
against SOM, PV, or both. For this purpose, rat sections
were treated in 10% fish gelatin and later incubated for 48
hours in a mixture of rabbit anti-SOM antibody (1:500,
gift from Dr. Robert Benoit, McGill University, Montreal/
Canada; Morrison et al., 1983) and mouse anti-PV (1:
1,000, Swant, Bellinzona/Switzerland). Rabbit IgGs were
visualized with a Cy2-labeled goat anti-rabbit secondary
antibody (1:400 in 2% fish gelatin; Amersham, Arlington
Heights, IL), whereas mouse IgGs were stained with a
Cy5-labeled donkey anti-mouse secondary antibody
(1:200, Jackson ImmunoResearch, West Grove, PA).
Monkey sections were treated with 10% fish gelatin and
incubated for 48 hours in a mixture of rat anti-SOM (1:
200, Chemicon) and rabbit anti-PV (1:1,000, Swant). The
rat IgGs were visualized with a Cy2-labeled goat anti-rat
(1:200, Jackson) secondary antibody, whereas the rabbit
IgGs were stained with a Cy5-labeled donkey anti-rabbit
secondary antibody (1:200, Jackson). Sections were
mounted on gelatinized slides and cover-slipped in Krystallon.
Types of interneurons. To determine whether in
monkey cortex SOM neurons are distinct from PV and CR
containing neurons as was shown in rat visual cortex
(Gonchar and Burkhalter, 1997), triple-labeling experiments were performed by using 40-␮m Vibratome sections
of monkey visual cortex. Sections were pretreated successively with sodium borohydride, H2O2 and ethanol as described above, immersed in 10% fish gelatin, and incubated (48 hours, 4°C) in a mixture of rat anti-SOM (1:200,
Chemicon), mouse anti-PV (1:1,000; Swant, Bellinzona/
Switzerland) and rabbit anti-CR (1:1,000, Swant
Bellinzona/Switzerland) antibodies. Staining with secondary antibodies involved goat anti-rat Cy2 (1:200), donkey
anti-mouse Cy5 (1:200) and goat-anti-rabbit Cy3 (1:200;
Chemicon) to reveal SOM, PV, and CR immunofluorescence, respectively.
Analysis. A multiphoton confocal microscope (BioRad MRC-1024 ES) was used to study and compare the
distributions of SOM- and PV-immunoreactive axon terminals on initial axon segments of BDA or SMI-32–
labeled pyramidal cells. The green (Cy2), red (Texas Red,
Cy3), and infrared (Cy5) fluorochromes were excited with
488-, 568-, and 647-nm emission lines of an argon-krypton
laser, respectively. Appropriate barrier filters were used
to split the emitted light into three distinct channels that
were detected by separate photomultipliers. Digitized images (1,024 ⫻ 1,024 pixels, 8 bits/pixel) were displayed on
the screen in three separate windows and superimposed in
the fourth window. Stacks of images (10/stack, optical
thickness 0.5␮m each) of retrogradely filled neurons were
acquired with a low-magnification 10⫻ objective to document the location of a neuron. High-magnification images
(10/stack, optical thickness 0.2␮m each) were acquired
with a 1.4 numerical aperture 60⫻ objective to determine
whether SOM and PV are colocalized in boutons and to
3
study the relationships of SOM and PV terminals with
axon initial segments in three dimensions. Close encounters of SOM- and PV-immunoreactive boutons with axon
initial segments were scored as appositions if (1) SOMand PV-positive terminal boutons were located in the
same focal plane as the target structure, and (2) threedimensional images (Laser Sharp and Confocal Assistant
software, Bio-Rad) viewed at different angles showed no
gap between boutons and axon initial segments.
Electron microscopy
Preembedding staining for somatostatin. For ultrastructural analyses rats were perfused with heparinized
PBS, followed by a phosphate buffered solution containing
4% paraformaldehyde, 0.5% of glutaraldehyde, and 0.1%
of picric acid (pH 7.6, 4°C). The brains were removed and
post-fixed for 2 hours in the same fixative at 4°C. Sections
were cut at 30 ␮m on a Vibratome, pretreated sequentially
in ice-cold 1% sodium borohydride (1 hour), 0.1% H2O2 (30
minutes) and 10% fish gelatin (30 minutes) and then incubated in rabbit anti-somatostatin antibody (gift of Dr. R.
Benoit, McGill University, Montreal/Canada, 1:500, 60
hours, 4°C). Later sections were washed in 2% fish gelatin
(30 minutes) and treated in biotinylated goat anti-rabbit
IgG (Chemicon, 1:200) followed by avidin and biotinylated
horseradish peroxidase (HRP, Vectastain ABC kit, Vector,
Burlingame, CA). The HRP reaction was performed in the
presence of diaminobenzidine tetrahydrochloride (0.05%
DAB) and H2O2 (0.005%). The staining intensity was titrated to reveal immunolabeled structures without masking cellular organelles. Sections were fixed for 40 minutes
with OsO4 (0.5% in PBS), dehydrated for 10 minutes in
ethanols of increasing concentrations, infiltrated with propylene oxide, and flat-embedded in Durcupan (Fluka,
Ronkonkoma, NY).
Resin-embedded sections from rat primary visual cortex
that showed SOM-labeled cell bodies, dendrites, and axon
terminals were further processed for ultrastructural analyses in the electron microscope. Small samples of layer 2/3
of area 17 were sectioned by using a Reichert Ultracut E
ultramicrotome equipped with a diamond knife (Diatome,
Fort Washington, PA). The sections were mounted on
200-mesh nickel grids, additionally stained with 1% uranyl acetate and lead citrate and studied in a JEOL 100
electron microscope. To control for artifactual staining, in
selected sections, contrasting with lead citrate was omitted (Fig. 5). SOM-immunoreactive profiles were identified
by the presence of electron dense, amorphous DAB reaction product. Pyramidal neurons and their initial axon
segments were identified in serial sections, by using structural features described by Peters et al, (1991). Photomicrographs were taken at 5,000⫻ and 20,000⫻. For prints,
the negatives were enlarged 2.5⫻. The distances of SOM
synapses on axon initial segments from the cell body and
the cross-sectional areas of pyramidal cell profiles were
measured on prints by using a digitizing tablet and NIH
image software (http://rsb.info.nih.gov/nih-image/).
Postembedding immunostaining for GABA. Thin
sections of SOM immunoperoxidase stained tissue were
mounted on 300-mesh nickel grids and subsequently
stained on drops of filtered solutions in a humidified
chamber. Sections were rinsed in Tris-buffered saline
(TBS, pH 7.6), incubated for 30 minutes in 1% a bovine
serum albumin (BSA) in TBS containing 0.02% Triton
X-100 (TBST), and treated for 18 hours with mouse anti-
4
Y. GONCHAR ET AL.
Fig. 1. Confocal microscopic images of retrogradely biotinylated
dextran amine (BDA) -filled pyramidal neuron in rat visual cortex.
A: Retrograde BDA-labeling of pyramidal neurons (red) in layer 2/3 of
area 17 indicates that their axons project to the secondary visual area,
the lateromedial area (LM). The cell in the center is labeled in Golgilike manner, revealing large parts of the dendritic arbor and the
descending axon. Thin red fibers represent feedback axons from area
LM and local axon collaterals of retrogradely labeled pyramidal neurons. The green cell bodies are stained with an antibody against
somatostatin (SOM). Green neuropil labeling represents SOMimmunoreactive dendrites, axons, and boutons. The figure is composed of 10 superimposed optical sections, each 0.5 ␮m thick.
B: Higher magnification of the pyramidal neuron (red) illustrated in
A, showing BDA-labeled dendrites, spines, and the axon initial segment. Two green fluorescent SOM-positive boutons (arrows) are directly apposed to the initial segment. Other SOM boutons contact
primary and secondary basal dendrites. The optical thickness of the
image is 2 ␮m. Scale bars ⫽ 20 ␮m in A,B.
GABA monoclonal antibody (Chemicon, 1:8,000 in TBST
containing 0.1% BSA, pH 7.6). The next day, sections were
thoroughly rinsed in TBST containing 1% BSA (pH 7.6),
rinsed in TBST (pH 8.2), and then incubated for 2 hours in
goat anti mouse-IgG conjugated to 15-nm gold particles
(Amersham, 1:25 in TBST, pH 8.2). Sections were rinsed
in TBS (pH 8.2), followed by rinses in TBS (pH 7.6) and in
distilled water. Finally, sections were stained with 1%
uranyl acetate (40 minutes) and Reinold’s lead citrate (10
minutes).
RESULTS
Confocal microscopy
Preparation of figures
Electron photomicrographs were printed from negatives
by using Kodak Polycontrast paper, and/or scanned at
1,200 dpi resolution by using a flatbed scanner (UMAX
Power Look III). Adobe Photoshop 5.0 software was used
to make linear adjustments of brightness and contrast in
the scanned images until the electronic version most
closely matched that of the photomicrograph. Final figures
were printed on a Phaser 440 (Tektronix) digital printer.
Confocal microscopy enabled the simultaneous observation of two to three fluorochromes representing retrogradely BDA-filled (red) or SMI-32 (red) immunolabeled
pyramidal cells, immunolabeled SOM-containing (green)
and PV-containing neurons (infrared, false-colored blue)
(Figs. 1–3).
In the rat, BDA injections into LM retrogradely labeled
projection neurons in layer two to six in ipsilateral and
contralateral area 17 (Figs. 1A,B, 3A). In many cases,
labeling was extensive and filled large parts of the basal
and apical dendritic arbors that resembled those of typical
pyramidal cells (Fig. 1A; Feldman, 1984). Well-labeled
cells showed a single axon that exited at the base of the
soma and descended toward white matter (Fig. 1A,B).
SMI-32 staining of monkey visual cortex revealed labeled neurons in layers 2, 3, 4B and 6 (Fig. 2A). The vast
majority of labeled neurons showed pyramidal cell morphologies. The staining consisted of a mesh of neurofilaments that extended throughout somata and dendrites
Fig. 2. Confocal images of SMI-32 stained pyramidal cells and
GABAergic neurons in supragranular layers of monkey primary visual
cortex. A: Low-power image of SMI-32 immunostained pyramidal neurons in layers 2, 3, and 4. B: Low-power image of the distribution of
parvalbumin (PV; blue, Cy5), calretinin (CR; red, Texas Red), and of
somatostatin (SOM; green, Cy2) -immunoreactive neurons in layers
1–4B of monkey primary visual cortex. Boxed area is shown at higher
magnification in C. Magnification in A and B is the same. C: PV (blue),
CR (red), and SOM (green) are expressed in different nonpyramidal
neurons. D: SMI-32–immunoreactive pyramidal neuron in layer 3.
Staining extends throughout the basal and apical dendritic arbor and
includes the initial segment of the axon. Numerous SOMimmunofluorescent fibers and puncta (green) are apposed to dendrites
and axon of pyramidal cells. Optical thickness, 2 ␮m. E: High-power
image of pyramidal neuron shown in D. Clusters of SOM boutons (arrows) are apposed to the axon initial segment. Optical thickness, 2 ␮m.
Scale bars ⫽ 100 ␮m in B (applies to A,B), 10 ␮m in C, 20 ␮m in D,E.
6
Y. GONCHAR ET AL.
Fig. 3. Confocal images of pyramidal neurons, somatostatin
(SOM), and parvalbumin (PV) -immunoreactive boutons in rat and
monkey primary visual cortex. A: Axon initial segment of retrogradely
biotinylated dextran amine (BDA) -labeled pyramidal neuron (red,
Texas red) in layer 2/3 of rat area 17 is apposed by SOM (green, Cy2)
and by PV (blue, Cy5) boutons. SOM boutons decorate basal dendrites
of pyramidal neuron. B: SMI-32–labeled pyramidal neuron (red) in
layer 3 of monkey primary visual cortex. SOM-immunoreactive axon
(green, Cy-2) approaches axon initial segment of pyramidal cell from
the side and terminates in close proximity of the axon. The same axon
initial segment is apposed by multiple PV-immunoreactive boutons
(blue, Cy5) that are proximal to SOM axon terminals. Scale bars ⫽ 20
␮m in A,B.
(Fig. 2D,E; Campbell and Morrison, 1989). Labeling of
axons was most prominent in the axon hillock and the
axon initial segment (Figs. 2D,E, 3B).
SOM immunostaining in rat primary visual cortex was
expressed in multipolar, bipolar, and bitufted nonpyramidal neurons and was absent from pyramidal cells. The
dendrites of SOM-immunoreactive cells were often beaded
and lacked spines. SOM neurons were present in layers
2– 6 (Fig. 1A). Their distribution was nonuniform and
consisted of small clusters of three to five neurons (Fig.
1A). Layer 1 was devoid of SOM-stained somata but contained labeled axons and dendrites.
SOM-labeled cells in monkey primary visual cortex included multipolar, bipolar, and bitufted nonpyramidal
neurons with spine-free dendrites. Similar to the rat,
SOM neurons in monkey visual cortex were scattered
throughout layers 2– 6 (Fig. 2B,C). Triple-labeling experiments with antibodies against SOM, PV, and CR revealed
that SOM neurons in supragranular layers on monkey
visual cortex lacked PV and CR staining (Fig. 2B,C).
The neuropil in both rat and monkey primary visual
cortex contained a network of SOM-immunoreactive axons that formed terminal boutons and boutons of passage
(Figs. 1B, 2D,E, 3A,B). In layers 2 and 3, axons of variable
lengths criss-crossed the neuropil in all different directions. This pattern of staining differed from the pericellu-
lar baskets and vertically oriented rows of boutons seen in
sections of monkey visual cortex labeled with antibodies
against PV (DeFelipe et al., 1989). The analyses of the
innervation of pyramidal neurons by SOM axon terminals
revealed that in both species SOM-positive puncta formed
close appositions with pyramidal cell somata, and primary
and higher order branches of apical and basal dendrites
(Figs. 1B, 2D,E, 3A,B). Most appositions were formed on
shafts of basal and apical dendrites where they were often
but not exclusively located at branch points (Figs. 1B,
2D,E). Thick dendrites were frequently decorated by short
rows of SOM boutons (Fig. 3A). Occasionally, vertically
oriented rows of SOM boutons were associated with dendrites but never with axon initial segments.
Three-dimensional reconstruction of high resolution
confocal images indicated that in primary visual cortex of
both rat and monkey, SOM-immunoreactive boutons were
apposed to axon initial segments of pyramidal neurons
(Figs. 1B, 2D,E, 3A,B). Typically, SOM-immunoreactive
axons approached axon initial segments at oblique angles
relative to the main vertical axis of the pyramidal cell
axon and formed one to three contacts (Fig. 3B). On fortuitous occasions, we observed individual SOM fibers with
boutons of passage that formed appositions with the axon
initial segment, the somata, and/or a basal dendrite of the
same pyramidal cell.
CONNECTIONS OF SOMATOSTATIN NEURONS IN VISUAL CORTEX
In three cases, SOM fibers were traced from the point of
the axo-axonic apposition back to the nonpyramidal parent cell bodies and were unequivocally identified as axons.
The dendrites of SOM neurons were thick and were easily
distinguished from the much thinner, cylindrical axons,
which makes it unlikely that dendrites adjacent to axon
initial segments were mistaken for axo-axonic appositions.
Electron microscopy
Although the analysis in the confocal microscope
strongly suggested that SOM-expressing axon terminals
provide axo-axonic inputs to pyramidal neurons, ultrastructural studies were performed to determine whether
these inputs represent synaptic connections. In the electron microscope, pyramidal cells were easily identified by
round nuclei surrounded by unfolded membranes, by the
presence of thick apical dendrites, and by exclusively symmetric synapses on the somata (Fig. 4A; Peters and Jones,
1984; Peters et al., 1991). Initial axon segments of pyramidal cells were identified by their origin from the base of
the soma, by the presence of dense axolemmal undercoating, fasciculated microtubules, clusters of ribosomes, and
cisternal organelles (Fig. 4B; Peters et al., 1991).
A total of 11 SOM-immunoreactive axon terminals were
identified that formed synaptic contacts with axon initial
segments of layer 2/3 pyramidal neurons of rat area 17.
Three of these were found by reconstructing eight axon
initial segments from serial sections. In each of these
reconstructions, the axon initial segment was contacted by
a single SOM axon terminal that was located 4 –9 ␮m
distal to the axon hillock (Fig. 4B). Contacts between
SOM-immunoreactive dendrites and axon initial segments of pyramidal neurons were not observed.
To control for lead contrasting artifacts in SOMimmunolabeled tissue, staining with lead citrate was
omitted. As expected, the contrast of the images was suboptimal. Despite this, unambiguous staining with SOM
immunoperoxidase reaction product was observed in axon
terminals that both contained elongated vesicles and
formed symmetric synapses with axon initial segments
(Fig. 5A,B). Interestingly, two of these terminals formed
additional synaptic contacts, one with a dendrite and the
other with a cell body (Fig. 5A,B).
SOM-immunoreactive terminals differed from unlabeled terminals that formed synaptic contacts with initial
segments. Unlike unlabeled, oval-shaped terminals with
smooth outlines, SOM terminals were often elongated and
had highly irregular outlines (Figs. 4B, 6A). In addition,
most unlabeled terminals were flattened against the initial segment with the long axis parallel to the postsynaptic
axon (Fig. 4B), whereas the long axes of SOM-labeled
terminal profiles formed oblique angles with axon initial
segments (Figs. 4B, 6A). Similar to unlabeled terminals in
contact with initial segments most SOM terminals contained one or more mitochondria. SOM-negative synapses
on initial segments always outnumbered synapses formed
by SOM-positive axon terminals. In addition, SOMnegative terminals in axon initial segments formed prominent clusters, whereas SOM-positive boutons tended to
be more dispersed (Fig. 3A,B).
Double-labeling experiments with antibodies against
SOM and GABA were performed to examine whether
SOM terminals contain GABA. The experiments revealed
that all (57 of 57) SOM-stained terminals studied in the
7
double-labeled tissue were GABA-immunogold positive,
including three terminals that formed synapses with axon
initial segments (Fig. 6A–C).
Numbers of SOM synapses
on initial segment
Reassured by the findings in the electron microscope
that SOM boutons formed synaptic connections with axon
initial segments, we used three-dimensional images acquired in the confocal microscope, which we inspected
from different viewing angles, to examine close encounters
of SOM boutons with axon initial segments. The analysis
revealed that, in rat, the majority (21 of 24) of axon initial
segments of forward projecting layer 2/3 pyramidal neurons in area 17 (Figs. 1B, 4A) and of most (10 of 11)
callosally projecting neurons in upper layers were closely
apposed by SOM-labeled boutons. The number of appositions counted as putative synapses varied from one to four
per initial segment and was approximately the same for
neurons with ipsilateral and contralateral projections (Table 1). Interestingly, each of the three forward projecting
neurons encountered in layer 6 lacked SOM boutons associated with the axon initial segment.
Observations in monkey primary visual cortex revealed
close encounters of SOM axon terminals and pyramidal
neurons similar to those found in rat. In all cases, SOMimmunoreactive axons approached initial segments from
the side and formed one to two appositions (Fig. 3B). To
rule out that SOM boutons contain markers that are found
in other types of GABAergic neurons, SMI-32 and SOM
double-labeled sections were in addition stained for PV or
CR. Qualitative observations in triple-labeled tissue from
rat and monkey visual cortex showed that SOM boutons
on initial segments of layer 2/3 pyramidal neurons did not
coexpress PV (Fig. 3A,B), suggesting that SOM boutons
derive from a separate population of GABAergic neurons.
These results extend previous observations of cell body
staining in rat visual cortex that showed that SOM, PV,
and CR neurons represent distinct types of GABAergic
neurons (Gonchar and Burkhalter, 1997). Quantitative
analysis of random samples from supragranular layers of
monkey visual cortex, however, showed coexpression of
PV and SOM in 3–9% of boutons of which most represent
putative inputs to dendrites and somata and only few may
correspond to inputs on axon initial segments (Table 2).
To compare the distribution of SOM inputs with axon
initial segments of layer 2/3 pyramidal neurons in rat area
17 seen in confocal and in electron microscopes, the data
were pooled. This procedure revealed comparable distributions of synaptic contacts and two peaks of inputs at
2– 4 ␮m and 8 –10 ␮m distal to the cell body (Fig. 7A).
Approximately 60% of SOM contacts were located at distances ⬎8 ␮m from the cell body. The distribution of SOM
boutons along the axon initial segment of upper layer
pyramidal neurons in monkey striate cortex resembled
that in rat, except that ⬃76% of contacts occurred distal to
8␮m from the cell body (Fig. 7A; Table 1). In both monkey
and rat, PV terminals were far more numerous than SOM
terminals. In both species, the distribution of PV terminals appeared broader and included more proximal and
more distal parts of the initial segment than those occupied by SOM terminals (Fig. 3A,B).
8
Y. GONCHAR ET AL.
Fig. 4. Electron photomicrographs of somatostatin (SOM) axon
terminals forming synapses with axon initial segment of pyramidal
neuron in layer 2/3 of rat primary visual cortex. A: Low-power picture
of the profile of a pyramidal neuron. The nucleus (Nu) is round,
without infoldings. A thick apical dendrite (AD) is oriented toward the
pial surface. An axon exits at the base of the cell. The boxed area
contains a portion of the axon initial segment (IS). B: Magnified image
of boxed area shown in A. SOM-labeled axon terminal (SOM⫹). Immunoperoxidase reaction product is accumulated in mitochondria and
synaptic vesicles lighter staining extends throughout the cytoplasm.
The intensity of the immunoperoxidase reaction product was titrated
for optimal visualization of organelles. SOM⫹ axon terminal is characterized by irregular outlines, contains elongated synaptic vesicles
and forms symmetric synapse with the axon IS of pyramidal neuron.
Axolemmal undercoating (arrows) and cisternal organelles (CO) are
typical features of axon initial segments. Unlabeled (SOM-) axon
terminal filled with elongated vesicles forms symmetric axo-axonic
synapse distal to the SOM⫹ terminal. Scale bars ⫽ 1 ␮m in A, 0.5 ␮m
in B.
Other targets of SOM axon terminals
that axon initial segments of pyramidal neurons were not
the primary targets of SOM axon terminals. In fact, of a
total of 163 SOM synapses only 1.2% were on axon initial
Although the present study was focused mainly on axoaxonic SOM synapses our observations clearly showed
CONNECTIONS OF SOMATOSTATIN NEURONS IN VISUAL CORTEX
9
Fig. 5. Electron photomicrographs of somatostatin (SOM)
immunoperoxidase-stained synapses on initial segments of axons.
Staining with lead citrate was omitted in the sections shown in A and
B, to control for contrasting artifacts. A: SOM⫹ axon terminal showing accumulations of immunoperoxidase reaction product in synaptic
vesicles and mitochondria. Labeled axon terminal contains elongated
vesicles and forms two symmetric synapses (arrows). The synaptic
contact on the left is with an axon initial segment (IS). The synapse on
the right is with a dendrite (D). Arrowheads point to electron-dense
material that typically lines the plasma membrane of axon initial
segments. B: SOM⫹ axon terminal showing accumulations of immunoperoxidase reaction product in vesicles and mitochondria. Labeled
axon terminal forms synapses (arrows) with an initial segment (IS)
and with the soma of a neuron (N). Scale bar ⫽ 0.5 ␮m in B (applies
to A,B).
segments. The majority of SOM synapses contacted dendritic shafts (68.2%), 25.1% were on spines (necks: 14.7%,
heads: 10.4%), and 5.5% on somata (Fig. 6B,C). These
targets included GABAergic and non-GABAergic structures. Among the GABA-positive targets we encountered
both SOM-positive and SOM-negative somata and dendrites.
channel-rich axon hillock (Wollner and Caterall, 1986) has
attracted considerable attention of anatomists (Palay et
al., 1968; Peters et al., 1968, 1982; Jones and Powell, 1969;
Somogyi, 1977; Sloper and Powell, 1979; Kosaka, 1980;
DeFelipe et al., 1989), who predicted that these synapses
play an important role in the suppression of action potentials. More recently, this hypothesis received considerable
support from recordings in hippocampal pyramidal neurons (Miles et al., 1996).
In rat, cat, and monkey axo-axonic synapses are formed
by a distinct type of nonpyramidal cell that is dedicated to
the innervation of axon initial segment (Somogyi, 1977;
Peters et al., 1982; Somogyi et al., 1982). Hence, the neurons became known as axo-axonic cells (Somogyi, 1977;
Somogyi et al., 1982) or chandelier cells (Peters et al.,
1982; Peters, 1984; DeFelipe, 1999). Historically, however, “chandelier cell” referred to nonpyramidal neurons
innervating apical dendrites of pyramidal cells (Szentagothai and Arbib, 1974). Although axo-axonic or chandelier cells are the main source of synaptic input to the
initial segment of pyramidal neurons, the discovery of
morphologic differences among synapses on this part of
the cell raised the possibility that they originate from
more than a single cell type (Peters and Fairen, 1978;
DISCUSSION
It is known for a long time that the axon initial segment
of cortical pyramidal neurons receives synaptic input (Palay et al., 1968; Peters et al., 1968). The synapses on this
stretch of axon located between the axon hillock and the
beginning of the myelin sheath are symmetric and are
formed by axon terminals that contain elongated vesicles,
glutamic acid decarboxylase, and GABA (Somogyi, 1977;
Peters et al., 1982; Somogyi et al., 1985). Recordings in
hippocampus have shown that GABA release from synapses on axon initial segments leads to fast hyperpolarization of postsynaptic pyramidal neurons and that this
effect depends on the activation of GABAA receptors (Buhl
et al., 1994). GABAergic synapses on the axon initial segment are, therefore, considered to be inhibitory. The
unique location of these inhibitory synapses near the Na⫹
10
Fig. 6. Electron photomicrographs of synapses formed by
␥-aminobutyric acid (GABA)ergic somatostatin (SOM) -immunoreactive axon terminals. A: SOM immunoperoxidase-stained, GABA
immunogold-labeled (black particles) axon terminal (SOM⫹) forming
synapse with axon initial segment (IS). B: SOM-positive immunoperoxidase-stained, GABA-positive (black particles) axon terminal
(SOM⫹) forming synapses with a spine (Sp) and a GABA-negative
Y. GONCHAR ET AL.
dendrite (D). C: Immunoperoxidase-labeled SOM-containing axon terminal (large arrow) that colocalizes GABA immunogold particles
(black particles) and forms a symmetric synapse with a GABAnegative perikaryon (P). Notice, that the outlines of the SOM terminal
are highly irregular. Nearby SOM-negative, GABA-positive terminal
forms symmetric axo-somatic synapse (small arrow) with the
perikaryon (P1). Scale bars ⫽ 0.5 ␮m in A,B, 1 ␮m in C.
CONNECTIONS OF SOMATOSTATIN NEURONS IN VISUAL CORTEX
11
TABLE 1. Number of SOM Boutons Forming Contacts with Axon Initial Segments (IS) of Pyramidal Neurons in Primary Visual Cortex
of Rat and Monkey.
Type of pyramidal neuron
Projecting to area LM (rat), layer
2/3
Projecting to area LM (rat), layer 6
Forming callosal projecting (rat),
layer 2/3
SMI 32 stained in layer 3 (monkey)
Area of the somatic
profile, ␮m2
mean ⫾ SD (range)
Total number of
SOM boutons
apposed to IS
24
137.0 ⫾ 34.9 (64–190)
36
3 : 0; 10 : 1; 8 : 2; 2 : 3; 1 : 4;
1.5 ⫾ 1.0 (0.2)
3
7
112.7 ⫾ 8.0 (97–123)
164.0 ⫾ 26.3 (137–203)
0
11
3:0
1 :0; 2 : 1; 3 : 2; 1 : 3
—
1.6 ⫾ 1.0 (0.4)
11
159.2 ⫾ 34.2 (113–221)
21
3 : 0; 2 : 1; 2 : 2; 2 : 3; 1 : 4; 1 : 5
1.9 ⫾ 1.7 (0.5)
TABLE 2. Colocalization of PV, CR and SOM Immunoreactivity in Boutons
of Layers 2 and 3 of Monkey Primary Visual Cortex. The Numbers Are
Derived from the Analysis of Immunofluorescence in the Confocal
Microscope. The Optical Thickness of the Images Used for Quantification of
Labeled Boutons was 2 ␮m.
PV
PV
CR
SOM
—
344/0 (0%)
164/15 (9.1%)
Average number of
SOM bouton per
IS, mean ⫾ SD
(SEM)
Total number of
neurons
reconstructed
CR
392/0 (0%)
—
191/0 (0%)
SOM
526/15 (2.9%)
372/0 (0%)
—
Somogyi et al., 1982; Peters and Harriman, 1990; DeFelipe and Farinas, 1992).
Recent studies, which increasingly relied on neurochemical attributes to distinguish different types of nonpyramidal neurons and their dendritic and axonal processes, revealed that chandelier cell axon terminals in rat,
monkey, and human cerebral cortex contain PV (Hendry
et al., 1989; DeFelipe, 1993; Condé et al.,1994; Del Rio and
Felipe, 1994; Kawaguchi and Kubota, 1996; Gonchar and
Burkhalter, 1999a). Our results show that SOM neurons
consistently form synapses with axon initial segments of
cortical pyramidal cells. SOM neurons in rat cortex differ
morphologically, physiologically, and chemically from PV
neurons and belong to a distinct family of GABAergic cells
(Kawaguchi and Kubota, 1996; Gonchar and Burkhalter,
1997; Reyes et al., 1998).
Several studies in primate cerebral cortex have shown
that chandelier cells are heterogeneous and comprise axoaxonic terminals that lack PV (Del Rio and DeFelipe,
1994), express corticotropin releasing factor (Lewis and
Lund, 1990), use different types of GABAA receptors
(Nusser et al., 1996), and colocalize PV and calbindin (Del
Rio and DeFelipe, 1997). In rat, calbindin is expressed by
the majority of neocortical SOM neurons (Kubota et al.,
1994; Gonchar and Burkhalter, 1997), which raises the
possibility that some chandelier cells lack PV but contain
SOM and calbindin. However, the evidence presented in
this study argues against this suggestion. Similar to chandelier cell terminals, SOM synapses are found on many
axon initial segments. But unlike chandelier cell inputs
that densely innervate axon initial segments, any given
segment is contacted by only a few SOM synapses (Kosaka, 1980; Peters et al., 1982; Gonchar and Burkhalter,
1999a; this study). SOM synapses on initial segments do
not express PV and do not form vertical arrays of axoaxonic synapses that are a hallmark feature of chandelier
cell terminals (Somogyi, 1977; Peters et al., 1982; Somogyi
et al., 1982; Peters, 1984; DeFelipe et al., 1989). Electron
microscopic analyses further show that SOM terminals
which innervate axon initial segments form additional
Number of IS apposed by:
number of SOM bouton
synapses on dendrites or somata, whereas chandelier cells
are thought to innervate exclusively initial segments (Somogyi et al., 1982; DeFelipe, 1999). Taken together, these
results indicate that SOM neurons which innervate initial
segments differ from chandelier cells and represent a distinct type of neuron. This conclusion agrees with previous
studies that failed to find SOM staining in chandelier cells
(Hendry et al., 1984; DeLima and Morrison, 1989; DeFelipe, 1999).
Studies in rat entorhinal cortex indicate that large numbers of SOM neurons are not GABAergic (Wouterlood and
Pothuizen, 2000). However, in frontal and visual cortex,
GABA is expressed by 90 –100% of SOM neurons (Kubota
et al., 1994; Gonchar and Burkhalter, 1997). Although
these studies suggest that GABA is the main transmitter
used by SOM neurons, it is possible that additional substances are involved in synaptic transmission. The presence of nitric oxide synthase in a subset of SOM neurons
suggests that nitric oxide may play a role (Kubota et al,
1994; Gonchar and Burkhalter, 1997). Somatostatin itself
may activate postsynaptic K⫹ channels through somatostatin receptors and counteract the subthreshold depolarization of pyramidal cells mediated by excitatory synapses (Hicks et al., 1998).
Reconstruction of SOM immunolabeled axon arbors
revealed that at least some axo-axonic synapses found
on area 17 pyramidal neurons originate from neurons
located in primary visual cortex. This finding is consistent with observations from lesioning experiments,
showing that SOM axons arborize locally and make
connections within the area that contains the cell body
(Morrison et al., 1985).
SOM neurons in both rat and monkey were shown to
innervate primarily dendrites, spines, and somata of pyramidal neurons and rarely contact aspiny dendrites and
nonpyramidal somata (Hendry et al., 1984; Meinecke and
Peters, 1986; DeLima and Morrison, 1989). The present
study confirms these observations. Similar to results in
monkey cerebral cortex (DeLima and Morrison, 1989),
SOM terminals in rat have a slight preference for spine
necks over spine head. Unlike other GABAergic terminals,
SOM axon terminals tend to form synapses near dendritic
branch points. Thus, SOM terminals are well-positioned
to influence subthreshold and suprathreshold activity of
pyramidal cells at different levels. The axon initial segment is the last point at which SOM neurons are able to
inhibit postsynaptic pyramidal cells and conceivably control their action potential outputs.
Compared with the massive innervation of dendrites,
soma, and axon initial segment by PV neurons (Kisvarday
et al., 1993; Gonchar and Burkhalter, 1999a; this study),
innervation of pyramidal cells by SOM neurons is much
12
Y. GONCHAR ET AL.
Fig. 7. Cumulative distribution of somatostatin (SOM) boutons
and synapses found at initial segments of pyramidal neurons in rat
and monkey visual cortices(A) and proposed scheme of the innervation by SOM neurons (B). A: White circles indicate SOM boutons in
rat primary visual cortex located at initial segments (IS) of pyramidal
neurons projecting to ipsilateral lateromedial area (LM) or of transcallosally projecting neurons in area 17 (gray circles) that were
observed in the confocal microscope. Black circles indicate synapses
formed by SOM boutons with axon initial segments of pyramidal
neurons in rat area 17. The distribution of SOM boutons apposed to
the axon initial segment resembles that of SOM synapses. Grey
squares indicate SOM boutons apposed to the axon initial segment of
pyramidal neurons in layer 3 of monkey primary visual cortex derived
from three-dimensional reconstructions in the confocal microscope.
B: SOM neurons form synapses on all compartments of pyramidal
neurons, i.e., spines, dendrites, somata, and axon initial segments.
This wide distribution of synapses allows SOM neurons to control
neuronal excitation at multiple input levels, including the site of
action potential output.
weaker. In addition, the pronounced spike frequency adaptation during sustained activation (Kawaguchi and
Kubota, 1996) may render inhibition of pyramidal cell
firing by SOM neurons less effective than by fast-spiking
PV neurons (Kawaguchi and Kubota, 1996; Miles et al.,
1996). On the other hand, trains of pyramidal cell inputs
are known to facilitate excitation of SOM neurons (Thomson et al., 1996; Reyes et al., 1998). In the course of a spike
train, this synaptic enhancement may lead to progressively stronger inhibition of pyramidal cells by SOM neurons. SOM neurons presumably do not receive direct
thalamocortical inputs, and most of their excitatory drive
derives from axon collaterals of pyramidal cells (Beierlein
et al., 2000; Gonchar and Burkhalter, unpublished observations). Therefore, it is attractive to speculate that inhibition from SOM neurons is preferentially directed
against recurrent cortical excitation (Douglas et al., 1995)
and that effective control of this activity relies in part on
axo-axonic synapses from SOM neurons.
ACKNOWLEDGMENTS
We thank Jeff Lichtman for advice in the use of the
confocal microscope and for helpful comments on earlier
versions of the manuscript. This work was supported by
NIH grants EY05935 and NS30676 and the McDonnell
Foundation for Studies of Higher Brain Function.
LITERATURE CITED
Beierlein M, Gibson JR, Connors BW. 2000. A network of electrically
coupled interneurons drives synchronized inhibition in neocortex. Nat
Neurosci 3:904 –910.
Borg-Graham LJ, Monier C, Fregnac Y. 1998. Visual input evokes transient and strong shunting inhibition in visual cortical neurons. Nature
393:369 –373.
Buhl EH, Halsy K, Somogyi P. 1994. Diverse sources of fast and slow
inhibitory postsynaptic potentials in rat neocortex in vitro. Nature
368:823– 828.
Campbell MJ, Morrison JH. 1989. A monoclonal antibody to neurofilament
CONNECTIONS OF SOMATOSTATIN NEURONS IN VISUAL CORTEX
protein (SMI-32) labeled a subpopulation of pyramidal neurons in the
human and monkey neocortex. J Comp Neurol 282:191–205.
Cauli B, Audinat E, Lambolez B, Angulo MC, Ropert N, Tsuzuki K, Nestrin
S, Rossier J. 1997. Molecular and physiological diversity of cortical
nonpyramidal cells. J Neurosci 17:3894 –3906.
Condé F, Lund SJ, Jacobovitz DM, Baimbridge KG, Lewis DA. 1994. Local
circuit neurons immunoreactive for calretinin, calbindin-D28 or parvalbumin in monkey prefrontal cortex: distribution and morphology.
J Comp Neurol 341:95–116.
DeFelipe J. 1993. Neocortical neuronal diversity: chemical heterogeneity
revealed by colocalization studies of classic neurotransmitters, neuropeptides, calcium binding proteins, and cell surface molecules. Cereb
Cortex 3:273–289.
DeFelipe J. 1999. Chandelier cells and epilepsy. Brain 122:1807–1822.
DeFelipe J, Farinas I. 1992. The pyramidal neuron of the cerebral cortex:
morphological and chemical characteristics of the synaptic inputs. Prog
Neurobiol 39:563– 607.
DeFelipe J, Hendry SHC, Jones EG, Schmechel D. 1985. Variability in the
terminations of GABAergic chandelier cell axons on initial segments of
pyramidal axons in the monkey sensory-motor cortex. J Comp Neurol
231:364 –384.
DeFelipe J, Hendry SH, Jones EG. 1989. Visualization of chandelier cell
axons by parvalbumin immunoreactivity in monkey cerebral cortex.
Proc Natl Acad Sci USA 86:2093–2097.
DeLima AD Morrison JH. 1989. Ultrastructural analysis of somatostatinimmunoreactive neurons and synapses in the temporal and occipital
cortex of the Macaque monkey. J Comp Neurol 283:212–227.
Del Rio MR, DeFelipe J. 1994. A study of SMI 32-stained pyramidal cells,
parvalbumin-immunoreactive chandelier cells and presumptive
thalamocortical axons in human temporal neocortex. J Comp Neurol
342:389 – 408.
Del Rio MR, DeFelipe J. 1997. Colocalization of parvalbumin and calbindin
D-28k in neurons, including chandelier cells of human temporal neocortex. J Comp Neurol 342:389 – 408.
Douglas RJ, Martin KAC. 1991. A functional microcircuit for cat visual
cortex. J Physiol Lond 440:735–769.
Douglas RJ, Koch C, Mahowald M, Martin KAC, Suarez HH. 1995. Recurrent excitation in neocortical circuits. Science 269:981–985.
Fairen A, DeFelipe J, Regidor J. 1984. Nonpyramidal neurons: general
account. In: Peters A, Jones EG, editors. Cerebral Cortex, Vol. 1. New
York: Plenum Press. p 201–253.
Feldman ML. 1984. Morphology of the neocortical pyramidal neurons. In:
Peters A, Jones EG editors. Cerebral Cortex, Vol. 1. New York: Plenum
Press, p 123–200.
Ferster D, Jagadeesh B. 1992. EPSP-IPSP interactions in cat visual cortex
studied with in vivo whole-cell patch recording . J Neurosci 12:1262–
1274.
Gibson JR, Beierlein M, Connors BW. 1999. Two electrical networks of
electrically coupled inhibitory neurons in neocortex. Nature 402:75–79.
Gonchar Y, Burkhalter A. 1997. Three distinct families of GABAergic
neurons in rat visual cortex. Cereb Cortex 7:347–358.
Gonchar Y, Burkhalter A. 1999a. Differential subcellular localization of
forward and feedback interareal inputs to parvalbumin-expressing
GABAergic neurons in rat visual cortex. J Comp Neurol 406:346:360.
Gonchar Y, Burkhalter A. 1999b. Connectivity of GABAergic calretininimmunoreactive neurons in rat primary visual cortex. Cereb Cortex
9:683– 696.
Gulyas A, Hajos N, Freund TF. 1996. Interneurons containing calretinin
are specialized to control other interneurons in the rat hippocampus.
J Neurosci 16:3397–3411.
Gupta A, Wang Y, Markram H. 2000. Organizing principles for a diversity
of GABAergic interneurons and synapses in the neocortex. Science
287:273–278.
Hendry SHC, Jones EG, Emson PC. 1984. Morphology, distribution, and
synaptic relations of somatostatin- and neuropeptide Y-immunoreactive neurons in rat and monkey neocortex. J Neurosci 4:2497:2517.
Hendry SHC, Jones EG, Emson PC, Lawson DEM, Heizmann CW, Streit
P. 1989. Two classes of cortical GABA-neurons defined by differential
calcium binding protein immunoreactivities. Exp Brain Res 76:467–
472.
Hicks GA, Feniuk W, Humphrey PP. 1998. Outward current produced by
somatostatin (SRIF) in rat anterior cingulate pyramidal cells in vitro.
Br J Pharmacol 124:252–258.
Hof PR, Ungerleider LG, Webster MJ, Gattass R, Adams MM, Sailstad CA,
13
Morrison JH. 1996. Neurofilament protein in differentially distributed
in subpopulations of corticocortical projection neurons in the macaque
monkey visual pathways. J Comp Neurol 376:112–127.
Jiang X-P, Johnson RR, Burkhalter A. 1993. Visualization of dendritic
morphology of cortical projection neurons by retrograde axonal tracing.
J Neurosci Methods 50:45– 60.
Johnson RR, Burkhalter A. 1996. Microcircuitry of forward and feedback
connections within rat visual cortex. J Comp Neurol 368:383–398.
Jones EG, Powell TPS. 1969. Synapses on the axon hillocks and initial
segments of pyramidal cell axons in the cerebral cortex. J Cell Sci
5:495:507.
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. 1996. Physiological and morphological identification of somatostatin- or vasoactive intestinal polypeptide-containing
cells among GABAergic cell subtypes in rat frontal cortex. J Neurosci
16:2701–2715.
Kawaguchi Y, Kubota Y. 1997. GABAergic cell subtypes and their synaptic
connections in rat frontal cortex. Cereb Cortex 7:476 – 486.
Kisvarday ZF, Beaulieu C, Eysel UT. 1993. Network of GABAergic large
basket cells in cat visual cortex (area 18): implication for lateral inhibition. J Comp Neurol 327:398 – 415.
Kosaka T. 1980. The axon initial segment as a synaptic site: ultrastructure
and synaptology of the initial segment of the pyramidal cell in the rat
hippocampus (CA3 region). J Neurocytol 9:861– 882.
Kubota Y, Hattori R, Yui Y. 1994. Three distinct subpopulations of GABAergic neurons in rat frontal agranular cortex. Brain Res 649:159 –173.
Lewis DA, Lund JS. 1990. Heterogeneity of chandelier neurons in monkey
neocortex: corticotropin-releasing factor- and parvalbumin-immunoreactive populations. J Comp Neurol 392:599 – 615.
Meinecke DL, Peters A. 1986. Somatostatin immunoreactive neurons in
rat visual cortex: a light and electron microscopic study. J Neurocytol
15:121–136.
Meskenaite V(1997. Calretinin-immunoreactive local circuit neurons in
area 17 of the cynomolgus monkey, Macaca fascicularis. J Comp Neurol 379:113–132.
Miles R, Tóth K, Gulyás AI, Haujós N, Freund TF. 1996. Differences
between somatic and dendritic inhibition in the hippocampus. Neuron
16:815– 823.
Morrison JH, Benoit R, Magistretti PJ, Bloom FE. 1983. Immunohistochemical distribution of pro-somatostatin-related peptides in cerebral
cortex. Brain Res 262:344 –351.
Morrison JH, Scherr S, Campbell MJ, Lewis DA. 1985. An experimental
analysis of the intracortical somatostatin-containing neurons. Soc Neurosci Abstr 11:679.
Nusser Z, Sieghart W, Benke D, Fritschy J-M, Somogyi P. 1996. Differential synaptic localization of two ␥-aminobutyric acid type A receptor
subunits in hippocampal pyramidal cells. Proc Natl Acad Sci USA
93:11939 –11944.
Palay SL, Sotelo C, Peters A, Orkand PM. 1968. The axon hillock and the
initial segment. J Cell Biol 38:193–201.
Peters A. 1984. Chandelier cells. In: Peters A, Jones EG, editors. Cerebral
Cortex, Vol.1. New York: Plenum Press. p 361–380.
Peters A, Fairen A. 1978. Smooth and sparsely-spined stellate cells in the
visual cortex of the rat: a study using combined Golgi-electronmicroscope technique. J Comp Neurol 181:129 –172.
Peters A, Jones EG. 1984. Classificaton of cortical neurons. In: Peters A,
Jones EG, editors. Cerebral cortex, Vol. 1. New York: Plenum Press.
p 107–121.
Peters A, Harriman KM. 1990. Different kinds of axon terminals forming
symmetric synapses with the cell bodies and initial axon segments of
layer II/III pyramidal cells. I. Morphometric analysis. J Neurocytol
19:154 –174.
Peters A, Proskauer CC, Kaiserman-Abramof IR. 1968. The small pyramidal neuron of the rat cerebral cortex. The axon hillock and the initial
segment. J Cell Biol 39:604 – 619.
Peters A, Proskauer CC, Ribak C. 1982. Chandelier cells in rat visual
cortex. J Comp Neurol 206:397– 416.
Peters A, Palay AL, Webster HD. 1991. Fine structure of the nervous
system. New-York: Oxford University Press.
Reyes A, Lujian R, Rozov A, Brunashev N, Somogyi P, Sakmann B. 1998.
Target-cell-specific facilitation and depression in neocortical circuits.
Nat Neurosci 1:279 –285.
14
Ribak CE. 1978. Aspinous and sparsely-spinous stellate neurons in the
visual cortex of rats contain glutamic acid decarboxylase. J Neurocytol
7:461– 478.
Sloper JJ, Powell TPS. 1979. A study of the axon initial segment and
proximal axon of neurons in the primate motor and somatic sensory
cortices. Philos Trans R Soc Lond Ser 285:173–179.
Somogyi P. 1977. A specific “axo-axonal” interneuron in the visual cortex of
the rat. Brain Res 136:345–350.
Somogyi P, Hodgson AJ. 1985. Antisera to ␥-aminobutyric acid. III. Demonstration of GABA in Golgi-impregnated neurons and in conventional
electron microscopic sections of cat striate cortex. J Histochem Cytochem 33:249 –257.
Somogyi P, Soltesz I. 1986. Immunogold demonstration of GABA in synaptic terminals of intracellularly recorded horseradish peroxidasefilled basket cells and clutch cells in the cat’s visual cortex. Neuroscience 19:1051–1065.
Somogyi P, Freund TF, Cowey A. 1982. The axo-axonic interneuron in the
cerebral cortex of the rat, cat, and monkey. Neuroscience 7:2577–2607.
Somogyi P, Freund TF, Hodgson AJ, Somogyi J, Beroukas D, Chubb I.
1985. Identified axo-axonic cells are immunoreactive for GABA in the
hippocampus and visual cortex of the rat. Brain Res 332:143–149.
Y. GONCHAR ET AL.
Stuart G, Sakmann B. 1994. Active propagation of somatic action potentials into neocortical pyramidal cell dendrites. Nature 367:69 –72.
Szentagothai J, Arbib MA. 1974. Conceptual models of neural organization. Neurosci Res Prog Bull 12:307–510.
Thomson AM, Deuchars J. 1997. Synaptic interactions in neocortical local
circuits: dual intracellular recording in vitro. Cereb Cortex 7:510 –522.
Thomson AM, West DC, Hahn J, Deuchars J. 1996. Single axon IPSPs
elicited in pyramidal cells by three classes of interneurones in slices of
rat neocortex. J Physiol Lond 496:81–102.
Weiser M, Bueno E, Sekirnjak C, Martone ME, Baker H, Hillman D, Chen
S, Thronhill W, Ellisman M, Rudy B. 1995. The potassium channel
subunit Kv3.1b is localized to somatic and axonal membranes of specific populations of CNS neurons. J Neurosci 15:4298 – 4314.
Williams SM, Goldman-Rakic PS, Leranth C. 1992. The synaptology of
parvalbumin-immunoreactive neurons in the primate prefrontal cortex. J Comp Neurol 320:353–369.
Wollner DA, Catterall WA. 1986. Localization of sodium channels in axon
hillocks and initial segments of retinal ganglion cells. Proc Natl Acad
Sci USA 83:8424 – 8428.
Wouterlood FG, Pothuizen H. 2000. Sparse colocalization of somatostatinand GABA-immunoreactivity in the entorhinal cortex of the rat. Hippocampus 10:77– 86.