Ultrastructural localization of the norepinephrine transporter in superficial and deep layers of the rat prelimbic prefrontal cortex and its spatial relationship to probable dopamine terminals.
ABSTRACT The prefrontal cortex (PFC) is a likely site of action for the therapeutic efficacy of antidepressants that inhibit norepinephrine (NE) reuptake. Moreover, drugs that block the NE transporter (NET) increase extracellular levels of both NE and dopamine (DA), an interaction that may contribute to their therapeutic properties. To examine the subcellular localization of NET and to investigate the spatial relationships between presumed NE and DA axons within the rat prelimbic PFC, we combined immunogold-silver localization of NET with immunoperoxidase staining for the catecholamine synthetic enzyme tyrosine hydroxylase (TH). An additional aim was to quantify the proportion of profiles dually labeled for NET and TH to test the common observation that TH immunolabeling is relatively selective for DA axons. NET-immunoreactive (NET-ir) axonal profiles were typically unmyelinated and occasionally were observed to form symmetric axodendritic synapses. The majority of immunogold NET labeling was unexpectedly observed in the cytoplasm rather than on the plasma membrane. Furthermore, in tissue dually labeled for both NET and TH, only 8-10% of profiles contained both markers. Unlike observations for singly labeled profiles, gold-silver particles for NET in dually labeled axons were localized primarily to the plasmalemma. A systematic survey of terminals labeled only for TH revealed that they were typically separated by at least 1.2 mum from NET-ir varicosities, and the two profile types were not seen to contact common targets. These results suggest that, in the rat PFC, NE axons (1) contain predominantly cytoplasmic NET, (2) infrequently contain TH immunolabeling, and (3) may interact with probable DA afferents by means of extrasynaptic mechanisms.
- Annals of the New York Academy of Sciences 10/1994; 733:193-202. · 4.38 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: Extracellular levels of serotonin [5-hydroxytryptamine (5-HT)] in the nucleus accumbens (NAc) can influence both cognitive and motor functions involving extensive connections with the frontal cortex. The 5-HT levels reflect vesicular release and plasmalemmal reuptake through the serotonin transporter (SERT). We used electron microscopic immunocytochemistry to determine the sites for SERT activation in the limbic shell and motor-associated core of the rat NAc. Of the SERT-immunoreactive profiles in each region, >90% were serotonergic axons and axon terminals; the remainder were nonserotonergic dendrites and glia. Axonal SERT immunogold labeling was seen mainly at nonsynaptic sites on plasma membranes and often near 5-HT-containing large dense core vesicles (DCVs). SERT-labeled axonal profiles were larger and had a higher numerical density in the shell versus the core but showed no regional differences in their content of SERT immunogold particles. In contrast, immunoreactive dendrites had a lower numerical density in the shell than in the core. SERT labeling in dendrites was localized to segments of plasma membrane near synaptic contacts from unlabeled terminals and/or dendritic appositions. Our results suggest that in the NAc (1) reuptake into serotonergic axons is most efficient after exocytotic release from DCVs, and (2) increased 5-HT release without concomitant increase in SERT expression in individual axons may contribute to higher extracellular levels of serotonin in the shell versus the core. These findings also indicate that SERT may play a minor substrate-dependent role in serotonin uptake or channel activity in selective nonserotonergic neurons and glia in the NAc.Journal of Neuroscience 09/1999; 19(17):7356-66. · 6.91 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: The dopamine transporter (DAT) critically regulates the duration of the cellular actions of dopamine and the extent to which dopamine diffuses in the extracellular space. We sought to determine whether the reportedly greater diffusion of dopamine in the rat prefrontal cortex (PFC) as compared with the striatum is associated with a more restricted axonal distribution of the cortical DAT protein. By light microscopy, avidin-biotin-peroxidase immunostaining for DAT was visualized in fibers that were densely distributed within the dorsolateral striatum and the superficial layers of the dorsal anterior cingulate cortex. In contrast, DAT-labeled axons were distributed only sparsely to the deep layers of the prelimbic cortex. By electron microscopy, DAT-immunoreactive profiles in the striatum and cingulate cortex included both varicose and intervaricose segments of axons. However, DAT-labeled processes in the prelimbic cortex were almost exclusively intervaricose axon segments. Immunolabeling for tyrosine hydroxylase in adjacent sections of the prelimbic cortex was localized to both varicosities and intervaricose segments of axons. These qualitative observations were supported by a quantitative assessment in which the diameter of immunoreactive profiles was used as a relative measure of whether varicose or intervaricose axon segments were labeled. These results suggest that considerable extracellular diffusion of dopamine in the prelimbic PFC may result, at least in part, from a paucity of DAT content in mesocortical dopamine axons, as well as a distribution of the DAT protein at a distance from synaptic release sites. The results further suggest that different populations of dopamine neurons selectively target the DAT to different subcellular locations.Journal of Neuroscience 04/1998; 18(7):2697-708. · 6.91 Impact Factor
Ultrastructural Localization of the
Norepinephrine Transporter in
Superficial and Deep Layers of the Rat
Prelimbic Prefrontal Cortex and Its
Spatial Relationship to Probable
LEEANN H. MINER,1SALLY SCHROETER,2RANDY D. BLAKELY,3
AND SUSAN R. SESACK1*
1Departments of Neuroscience and Psychiatry, University of Pittsburgh,
Pittsburgh, Pennsylvania 15260
2Pharmacia Corporation, Chesterfield, Missouri 63017
3Department of Pharmacology, School of Medicine, and Center for Molecular Neuroscience,
Vanderbilt University, Nashville, Tennessee 37232
The prefrontal cortex (PFC) is a likely site of action for the therapeutic efficacy of antide-
pressants that inhibit norepinephrine (NE) reuptake. Moreover, drugs that block the NE trans-
porter (NET) increase extracellular levels of both NE and dopamine (DA), an interaction that
may contribute to their therapeutic properties. To examine the subcellular localization of NET
and to investigate the spatial relationships between presumed NE and DA axons within the rat
prelimbic PFC, we combined immunogold–silver localization of NET with immunoperoxidase
staining for the catecholamine synthetic enzyme tyrosine hydroxylase (TH). An additional aim
was to quantify the proportion of profiles dually labeled for NET and TH to test the common
observation that TH immunolabeling is relatively selective for DA axons. NET-immunoreactive
(NET-ir) axonal profiles were typically unmyelinated and occasionally were observed to form
symmetric axodendritic synapses. The majority of immunogold NET labeling was unexpectedly
observed in the cytoplasm rather than on the plasma membrane. Furthermore, in tissue dually
labeled for both NET and TH, only 8–10% of profiles contained both markers. Unlike observa-
tions for singly labeled profiles, gold–silver particles for NET in dually labeled axons were
localized primarily to the plasmalemma. A systematic survey of terminals labeled only for TH
revealed that they were typically separated by at least 1.2 ?m from NET-ir varicosities, and the
two profile types were not seen to contact common targets. These results suggest that, in the rat
PFC, NE axons (1) contain predominantly cytoplasmic NET, (2) infrequently contain TH immu-
nolabeling, and (3) may interact with probable DA afferents by means of extrasynaptic mecha-
nisms. J. Comp. Neurol. 466:478–494, 2003.
© 2003 Wiley-Liss, Inc.
Indexing terms: antidepressants; catecholamine; depression
The prefrontal cortex (PFC) is involved in complex cog-
nitive processing and the organization of goal-directed
behavior (Fuster, 1997), and pathology of the PFC may
underlie some of the cognitive deficits observed in depres-
sion, attention deficit hyperactivity disorder (ADHD), and
schizophrenia (Berman and Weinberger, 1990; Drevets et
al., 1992; Soares and Mann, 1997; Mostofsky et al., 2002).
Moreover, the norepinephrine (NE) innervation of the
PFC is also implicated in these disorders, as well as the
therapeutic properties of antidepressants, antipsychotics,
and drugs used to treat ADHD (Nutt et al., 1997; Li et al.,
1998; Westerink et al., 1998; Nelson, 1999; Frazer, 2000;
Bymaster et al., 2002). Indeed, intact NE afferents to the
PFC are involved in normal cognitive processes like vigi-
lance (Aston-Jones et al., 1991; Florin-Lechner et al.,
1996; Arnsten, 1997). Other monoamines such as dopa-
Grant sponsor: National Institute of Mental Health; Grant number:
MH50314 (S.R.S.); Grant number: MH58921 (R.D.B.); Grant number: F32
*Correspondence to: Susan R. Sesack, Department of Neuroscience, 446
Crawford Hall, University of Pittsburgh, Pittsburgh, PA 15260.
Received 31 October 2002; Revised 26 March 2003; Accepted 30 June 2003
Published online the week of October 6, 2003 in Wiley InterScience
THE JOURNAL OF COMPARATIVE NEUROLOGY 466:478–494 (2003)
© 2003 WILEY-LISS, INC.
mine (DA) also modulate these functions, and NE and DA
are likely to interact both in normal cortical regulation
and in the therapeutic response to psychoactive com-
The NE transporter (NET) is critical for the removal of
NE from the extracellular space (Axelrod and Kopin, 1969;
Iversen, 1971; Trendelenburg, 1989; Bonisch and Bru ¨ss,
1994), and NET is also a target for antidepressant drug
actions (Nelson, 1999; Frazer, 2000; Kent, 2000; Mo ¨ller,
2000). Moreover, there is some evidence that depression
may be associated with a loss of NET protein (Klimek et
al., 1997). Neurochemical studies have suggested that
some antidepressants exert their therapeutic action by
blocking NET specifically within the PFC (Tanda et al.,
1994), consistent with the known role of NE in modulating
An antibody directed against the carboxy terminus of
(Schroeter et al., 2000) and shown to be specific for NET on
the basis of Western blot analysis, preadsorption methods,
light and electron microscopic distribution, and loss of
immunoreactivity after specific neurotoxic lesions. The
presence of NET protein in the PFC has been documented
initially (Schroeter et al., 2000). However, the ultrastruc-
tural localization of NET within this region has not been
well characterized. Therefore, the first aim of the present
study was to describe the cellular and subcellular distri-
bution of NET within the medial PFC of the rat by using
electron microscopic immunocytochemistry. As part of this
analysis, we also performed dual immunolabeling for NET
and dopamine ?-hydroxylase (D?H), the enzyme that cat-
alyzes the conversion of DA to NE.
The availability of NET antibodies as reliable markers
of NE neurons and their processes (Schroeter et al., 2000)
allows us to address a separate issue, namely the repeated
observations by light microscopy that the catecholamine
synthetic enzyme tyrosine hydroxylase (TH) is relatively
absent from cortical NE axons. Previous immunocyto-
chemical studies in primates have shown that antisera for
TH and D?H label distinct populations of terminals in
areas that have both NE and DA afferents (e.g., PFC).
Moreover, the distribution patterns of TH-immunoreactive
terminals are more similar to those of DA afferents than
to NE inputs (Lewis et al., 1987, 1988; Noack and Lewis,
1989). Furthermore, lesions of the locus coeruleus, the
source of the NE cell bodies that project to the cortex
(Swanson and Hartman, 1975; Loughlin et al., 1982), sub-
stantially reduce cortical D?H immunostaining but not
TH immunoreactivity (Lewis et al., 1987, 1988). Finally, a
previous ultrastructural comparison of terminals in the
rat PFC labeled for either TH or DA revealed no differ-
ences in their synaptic contacts (Sesack et al., 1995).
These findings suggest that immunocytochemical label-
ing of TH-containing terminals appears to be a sensitive
and specific method for identifying DA axons in the PFC.
However, the extent to which some NE terminals are also
labeled by using TH immunocytochemistry has never been
directly investigated in the rat PFC by using ultrastruc-
tural methods. Therefore, the second aim of the present
study was to describe the extent to which profiles immu-
noreactive for NET also expressed TH by using a dual
labeling procedure. We used NET rather than D?H im-
munoreactivity to identify NE axons because of the
greater sensitivity of NET labeling. Specifically, the vesic-
ular localization of the D?H enzyme requires antibodies to
penetrate both the plasmalemmal and vesicular mem-
branes to bind to D?H. Hence, this marker generally pro-
duces weak immunolabeling for ultrastructural studies
that use low detergent levels.
To the extent that TH antibodies are largely selective
for DA axons in the PFC, an assertion for which we now
present quantitative evidence, we can also examine the
spatial relationship between afferents expressing NET
and inputs that are labeled only for TH and, therefore,
presumably DA. This was of interest because local appli-
cation into the PFC of drugs that block NET increases
extracellular levels of both NE and DA (Jordan et al.,
1994; Tanda et al., 1994, 1997; Gresch et al., 1995;
Yamamoto and Novotney, 1998; Bymaster et al., 2002).
Moreover, it has been hypothesized that the subsequent
increase in DA levels in the PFC may be involved in the
ability of these drugs to treat depression (Tanda et al.,
1994) or ADHD (Bymaster et al., 2002). These studies
indicate that local alterations in NET functioning in the
PFC induce changes in DA levels, in part because the NET
may serve as an important clearance mechanism for DA
(Horn, 1973; Raiteri et al., 1977; Gu et al., 1996). There-
fore, the third aim of the present study was to investigate
the spatial relationship of profiles expressing NET and
presumed DA terminals singly labeled for TH in the rat
PFC. TH was considered the most optimal marker for this
purpose, because the fixative required for DA immunola-
beling is incompatible with NET staining, and because DA
transporter immunoreactivity is low in this region (Sesack
et al., 1998). Given the level of functional interaction be-
tween NE and DA systems, it was hypothesized that the
two types of profiles would occur in close juxtaposition.
More specifically, we speculated that the profiles express-
ing NET or TH would frequently be apposed to each other
and would possibly contact common postsynaptic ele-
MATERIALS AND METHODS
The rabbit polyclonal NET antiserum was raised
against the peptide WERVAYGITPENEHHLVAQRDVR
(amino acids 585-607, peptide N585) located in the car-
boxy terminus of mouse NET (Fritz et al., 1998). The
generation and purification of this antibody has been de-
scribed previously in extensive detail and specificity has
been demonstrated by several methods, including pread-
sorption with peptide (Schroeter et al., 2000). The mouse
monoclonal antibody against rat TH (Chemicon Interna-
tional, Temecula, CA) recognizes an epitope that does not
include the regulatory N-terminus of the enzyme. This
antibody and the remaining antisera used in the present
study were commercially available, and the specificity of
each has been determined or described in previous publi-
cations by using techniques such as Western blot analysis
and radioactive immunoassay (Steinbusch and Tilders,
1987; Wolf et al., 1991; Sesack et al., 1995).
Twelve naive adult male Sprague-Dawley rats (Hilltop
Lab Animals, Scottsdale, PA) weighing 300–350 g were
used in the present investigation. Prior to the perfusion,
all animals were housed in the departmental animal fa-
cility for at least 3 days and received ad libitum food and
479NOREPINEPHRINE TRANSPORTER IN PFC
water. The processing of tissue from each of these 12 rats
was as follows: (1) 5 rats, dual immunolabeling for NET
immunogold–silver labeling for NET, electron microscopy;
(3) 2 rats, dual immunolabeling for D?H and NET, elec-
tron microscopy; (4) 1 rat, single immunoperoxidase label-
ing for NET or TH, light microscopy; (5) 1 rat, single
immunoperoxidase labeling for D?H, light microscopy; (6)
1 rat, single immunoperoxidase labeling for DA, light mi-
croscopy. The experiments were conducted in accordance
with the National Research Council’s Guide for the Care
and Use of Laboratory Animals (1996), and animal use
protocols were approved by the University of Pittsburgh’s
Institutional Animal Care and Use Committee.
The rats were anesthetized with sodium pentobarbital
(50 mg/kg i.p.) and given sodium diethyl dithiocarbamate
(1 g/kg i.p.) to eliminate artifactual labeling of zinc by the
silver used for intensification of immunogold labeling
(Veznedaroglu and Milner, 1992). Fifteen minutes after
administration of the zinc chelator, rats were transcardi-
ally perfused with 5 ml of heparinized saline (1,000 U/ml)
followed by 50 ml of 3.75% acrolein in 2% paraformalde-
hyde and then 250 ml of 2% paraformaldehyde in 0.1 M
phosphate buffer, pH 7.3 (PB). To examine whether treat-
ment with sodium diethyl dithiocarbamate affects NET
localization, one rat was perfused by using the same fix-
ative but without the zinc chelator. The brains were re-
moved, trimmed into 5-mm blocks, and post-fixed at room
temperature in 2% paraformaldehyde for 30 minutes. Tis-
sue from the rat processed for DA immunoperoxidase la-
beling was taken from a previous study (Sesack et al.,
1995) and was perfused by using 500 ml of 5% glutaral-
dehyde in PB containing 0.4% sodium metabisulfite. Coro-
nal sections (50 ?m) through the PFC were then cut by
using a vibrating microtome and placed in PB.
For ultrastructural investigation, immunoperoxidase
was used to label TH or D?H, and immunogold–silver was
used to label NET. There were two principle reasons for
choosing this combination. First, we wished to examine
the subcellular distribution of NET, and immunogold–
silver provides better subcellular detail than immunoper-
oxidase. Second, we anticipated that NET-ir profiles
would have low levels of TH and less accessible D?H, and
we wished to use the more sensitive immunoperoxidase
technique (Chan et al., 1990) for detecting these enzymes.
Finally, immunoperoxidase labeling was used for compar-
ative light microscopic examination of all the markers:
NET, TH, D?H, and DA.
All sections except those dually
immunolabeled for D?H and NET (see below) were incu-
bated in 1% sodium borohydride in PB for 30 minutes and
then rinsed in PB. They were then rinsed further in 0.1 M
Tris-buffered saline, pH 7.6 (TBS), and incubated for 30
minutes in blocking solution containing 1% bovine serum
albumin (BSA) and 3% normal goat serum in TBS. The
blocking solution for electron microscopy contained 0.04%
Triton X-100, whereas the solution for light microscopy
contained 0.2% Triton X-100. The sections were then
transferred into blocking solution containing the primary
antibodies and left at room temperature overnight (mini-
mum of 12 hours). The primary antibodies were rabbit
anti-NET (1:2,000) and mouse anti-TH (1:4,000; Chemicon
(1:2,000; Chemicon International) or rabbit anti-DA (1:
4,000; Dr. Harry Steinbusch, University of Limburg,
Maastricht). The next day, the sections to be processed for
immunoperoxidase were rinsed in TBS and then incu-
bated for 30 minutes in biotinylated horse anti-mouse or
goat anti-rabbit IgG (Vector Laboratories, Burlingame,
CA) diluted 1:400 in blocking solution. Finally, the sec-
tions were incubated in avidin–biotin–peroxidase complex
(Vectastain Elite, Vector Laboratories) diluted 1:100 in
TBS for 30 minutes and then rinsed. The peroxidase was
visualized by incubating the sections in 0.022% 3,3?-
diaminobenzidine (DAB) and 0.003% hydrogen peroxide
in TBS for 3 to 4 minutes.
Sections dually labeled for D?H and NET were first
incubated in 1% sodium borohydride in 0.01 M phosphate-
buffered saline, pH 7.4 (PBS), for 20 minutes and then
rinsed in PBS. They were then incubated in a cryopro-
tectant solution containing 8% glycerol and 20% sucrose in
0.04 M PB for 20 minutes at room temperature and then
at ?80°C for 20 minutes. Next, the tissue was incubated
at room temperature in decreasing concentrations of the
cryoprotectant (100%, 70%, 50%, and 30%) for 10 minutes
each and then rinsed for an additional 10 minutes in PBS.
Sections were then incubated in blocking solution contain-
ing 1% normal goat serum and 0.02% Triton X-100 in PBS
for 30 minutes. The sections were subsequently incubated
in blocking solution containing the primary antibodies at
room temperature for 1 hour and then at 4°C for 40 hours.
The primary antibodies were rabbit anti-NET (1:2,000)
and mouse anti-D?H (1:2,000; Chemicon). They were then
rinsed for 1 hour in several changes of PBS and incubated
for 1 hour in biotinylated horse anti-mouse IgG (Vector)
diluted 1:400 in blocking solution. The sections were in-
cubated for 2 hours in avidin–biotin–peroxidase complex
(Vectastain Elite, Vector) diluted 1:100 in PBS and then
rinsed. The peroxidase was visualized by incubating the
sections in DAB and hydrogen peroxide in PBS for 3 to 4
Sections to be dually labeled were
first processed for TH or D?H by immunoperoxidase stain-
ing and then incubated for 30 minutes in a different block-
ing solution containing 0.8% BSA and 0.1% fish gelatin in
PBS. They were then incubated for 3 hours in blocking
solution containing a 1:50 dilution of goat anti-rabbit con-
jugated to 1-nm gold particles (Amersham Life Sciences,
Buckinghamshire, England). The tissue was rinsed in
PBS, placed in 2% glutaraldehyde in PBS for 10 minutes,
and thoroughly rinsed in PBS again. Sections were then
rinsed in 0.2 M sodium citrate buffer (pH 7.4), and the gold
labeling was intensified by incubation for 5 to 10 minutes
in a silver solution (IntenSE M Silver Enhancement Kit,
Amersham Life Sciences). The silver reaction was termi-
nated by rinsing in sodium citrate buffer and then PB.
Some sections from each condition were mounted onto
glass slides, dehydrated through an ascending ethanol
series, cleared with xylene, and cover-slipped by using
DPX (Electron Microscopy Sciences, Fort Washington,
PA). Representative sections were photographed by using
a Zeiss Axiophot light microscope and a digital camera.
The images were then adjusted for contrast and exposure
in Adobe Photoshop.
480L.H. MINER ET AL.
The sections to be processed for electron microscopy
were incubated for 1 hour in 2% osmium tetroxide in PB
and rinsed in PB. The sections were then dehydrated
through a series of ascending concentration ethanol solu-
tions and finally propylene oxide. They were incubated
overnight in a 1:1 solution of propylene oxide and Epon
(EM-bed 812, Electron Microscopy Sciences). The next
day, the sections were transferred to straight Epon for 2
hours and then flat-embedded between plastic sheets and
heated for 18 hours at 60°C.
Pieces of the prelimbic PFC (Krettek and Price, 1977)
were glued onto solid Epon capsules and trimmed into
trapezoids. The trapezoids were then cut at 70 nm by
using an ultramicrotome (Leica, Buffalo, NY). Ribbons of
the ultrathin sections were collected onto copper mesh
grids (400 mesh, Electron Microscopy Sciences) and coun-
terstained with 5% uranyl acetate and Reynold’s lead
citrate. Tissue was examined by using a Zeiss 902 trans-
mission electron microscope (Oberkochen, Germany).
To characterize the ultrastructural properties of NET-
immunoreactive (NET-ir) profiles in the PFC, a minimum
of one Vibratome section each from superficial (I–III) and
deep layers (V/VI) from tissue dually labeled for NET and
TH was examined for each animal. Random fields at the
tissue–plastic interface containing neuropil were selected
at 7,000? and recorded by using the computer coordinate
system on the electron microscope. Then, photographs
were taken at each of the recorded sites at 12,000?. Areas
containing soma or bundles of myelinated fibers were ex-
cluded, and care was taken not to analyze the same fields
in serial ultrathin sections. Sampling was always re-
stricted to the surface of the tissue where immunoreagent
penetration was maximal. Equivalent numbers of photo-
graphs, and hence amounts of tissue, were analyzed for
superficial and deep layers (Table 1).
Specific immunogold labeling for NET was defined as a
minimum of two gold particles on the membrane or three
particles within the cytoplasm. These criteria were similar
to those used to characterize the serotonin transporter in
this region (Miner et al., 2000) but were more conser-
vative than those used by other investigators for other
transporters (Garzon et al., 1999; Pickel and Chan,
1999). Because the perimeter-to-area ratio is higher for
small versus large profiles, a single random gold parti-
cle has a greater probability of being associated with the
plasmalemma than with the cytoplasm of smaller pro-
files. However, the presence of two particles associated
with the membrane of the same profile is unlikely to
occur by random chance.
For each photograph, the number of NET-ir structures
was recorded, and the area and diameter of each was
determined by using an image analysis system (see be-
low). In addition, the following characteristics were noted:
(1) the number of gold particles per unit area; (2) the
location of gold particles (membrane versus cytoplasm);
(3) the number, type, and target of synaptic contacts being
formed; and (4) the presence or absence of TH immunore-
To determine the spatial relationships between TH-
immunolabeled and NET-ir profiles, at least one Vi-
bratome section each from superficial and deep layers was
examined from each rat. It should be noted that the sec-
tions used for this portion of the study were different from
the ones used for the general characterization of NET-ir
profiles described above. Tissue was systematically exam-
ined at the surface at 20,000? magnification to identify
immunoperoxidase labeling for TH. The number of TH-
immunoreactive (TH-ir) terminals was recorded, and if a
NET-ir profile was within the same field (defined as the
13.8 ?m2area within the photographic brackets on the
viewing screen), a photograph was taken for later image
analysis. Closely adjacent ultrathin sections were ex-
cluded from analysis to avoid resampling of profiles in
serial. The amount of tissue sampled (Table 3) was esti-
mated based on the percentage of tissue vs. Epon plastic in
each of the grid squares analyzed and the known area of a
grid square, 3,025 ?m2.
Quantitative image analysis was performed to calculate
the relative diameter and area of the NET-ir profiles and
the spatial distances between TH-ir profiles and NET-ir
structures. Electron microscopic negatives were scanned
by using a digital camera into a Presage image analysis
system (Advanced Imaging Concepts, Princeton, NJ).
NET-ir profiles were traced according to the following
guidelines: (1) when possible, the actual membrane was
traced; (2) when the membrane was obscured by immuno-
gold particles, the perimeter was approximated between
areas of visible membrane; and (3) when two NET-ir pro-
files were apposed to each other, the inner portions of the
apposed membranes were outlined, so that the binary
images would not merge. An unbiased, computerized al-
gorithm was then used to determine the area and maxi-
mum diameter of each profile. For a detailed description,
see Sesack et al. (1998). To calculate the distances be-
tween the TH- and NET-ir profiles, locations were chosen
on the outer edges of the membranes of each profile at
points of closest vicinity. The analysis system then drew a
line between the locations and measured the distance.
Significant differences were determined by using t tests or
the nonparametric Fisher’s exact test (?? 0.05).
By light microscopic examination (Fig. 1), NET-ir fibers
were distributed across all layers of the rat prelimbic PFC
in a relatively uniform manner (Fig. 1A). The fibers la-
TABLE 1. Profiles Singly Immunoreactive for NET in a Random Sample
From the Superficial and Deep Layers of the Rat Prefrontal Cortex1
Total area examined (?m2)
Number of NET-ir profiles observed
Area (?m2) of NET-ir profiles (mean ? SD)
Diameter (?m) of NET-ir profiles (mean ? SD)
Density of immunogold–silver labeling
(particles per unit area; mean ? SD)
Percentage of immunogold–silver labeling on
the plasma membrane
Number (%) of NET-ir profiles forming
0.43 ? 0.27
0.42 ? 0.14
19.79 ? 23.5
0.46 ? 0.28
0.41 ? 0.14
24.9 ? 60.2
20 ? 27% 24 ? 32%
25 (33%) 12 (20%)
1All comparisons between superficial and deep layers not significant, P ? 0.12. NET,
norepinephrine transporter; -ir, immunoreactive.
481 NOREPINEPHRINE TRANSPORTER IN PFC
beled for NET were thick, and relatively smooth, with
occasional beading. They coursed parallel to the pial sur-
face in layer VI and prominently at the layer I/II border.
In middle layers, NET-ir fibers had a primarily tangential
orientation. The laminar distribution and orientation of
NET-ir fibers markedly resembled axons labeled for D?H
(Fig. 1B), although the latter fibers had a more beaded
appearance. In contrast, TH-ir fibers were most dense in
layers V and VI of the prelimbic PFC and decreased pro-
gressively toward the pial surface (Fig. 1C). In general,
TH-ir fibers were considerably more varicose than those
labeled for NET. The morphologic appearance and orien-
tation of TH-ir fibers was comparable to axons labeled by
immunocytochemistry for DA (Fig. 1D), with the exception
that a thin band of TH-ir fibers ran parallel to the pial
surface at the border between layers I and II. An analo-
gous band of fibers was observed after labeling for NET
and for D?H.
ing of norepinephrine transporter (NET), dopamine ?-hydroxylase
(D?H), tyrosine hydroxylase (TH), or dopamine (DA) axons within the
prelimbic portion of the rat prefrontal cortex. A: The relatively
smooth, moderately branched NET-immunoreactive (-ir) profiles ex-
hibit a uniform density across the cortical layers, with an additional
dense plexus of axons coursing tangentially along the border between
layers I and II. NET-ir fibers in layer VI also run tangentially,
whereas those in layers II–V have an irregular, radial orientation. B:
The overall laminar distribution and orientation of NET-ir axons is
Light photomicrographs showing immunoperoxidase label-
similar to those labeled for D?H. C: TH-ir axons are more heavily
branched and beaded than NET-ir fibers. They are distributed pri-
marily in deep layers V–VI and become more sparse in superficial
layers, with the exception of a band of TH-ir axons coursing tangen-
tially within the deep portion of layer I, which is also seen for NET
and D?H. D: The latter band of fibers is not seen in tissue immuno-
labeled for DA, although other aspects of morphology and laminar
distribution are comparable between DA-ir and TH-ir axons. Scale
bar ? 200 ?m in D (applies to A–D).
482L.H. MINER ET AL.
Characterization of profiles singly labeled for NET.
A random sample of 85 fields from the neuropil of super-
ficial and deep layers (170 fields total) resulted in 157
immunogold–silver labeled NET profiles, of which 141
were only singly labeled for NET (Table 1). The majority of
NET-ir profiles were unmyelinated axons and varicosities
that were not observed to form traditional synaptic con-
tacts in single sections (Fig. 2). They contained small,
clear vesicles and occasional dense-cored vesicles. In these
profiles, immunogold–silver labeling for NET was ob-
served predominantly within the cytoplasm and only oc-
casionally along the plasma membrane (Fig. 2). Intracel-
lular NET labeling was associated with small vesicles
likely to represent synaptic vesicles (Fig. 2A) as well as
with larger tubulovesicles. Immunogold–silver particles
for NET were also sometimes observed in the cytoplasm
unassociated with organelles in that plane of section.
Some NET-ir profiles did form synapses in single sec-
tions (Table 1; 37 of 157, 24%); the observed synaptic
incidence was slightly but not significantly higher in su-
perficial layers (33%) than in deep layers (20%; Fisher’s
exact test, P ? 0.18). Synapses formed by NET-ir profiles
were typically of the symmetric type (26 of 37, 70%; Fig.
3A–C) and less frequently asymmetric (11 of 37; Fig. 3D).
The synaptic targets of NET-ir profiles were predomi-
rine transporter–immunoreactive (NET-ir) axons (a) and varicosities
(large arrows) within the superficial and deep layers of the rat pre-
frontal cortex. NET-ir profiles are characteristically unmyelinated,
contain primarily small clear vesicles, and do not form obvious syn-
aptic junctions in single sections. NET-ir profiles are sometimes ob-
Electron photomicrographs illustrating typical norepineph-
served in close proximity to blood vessels (bv in D). The majority of
immunogold–silver particles for NET occur within the cytoplasm,
often in association with small vesicles (small arrows in A). Only a few
NET immunogold–silver particles are associated with the plasma
membrane. Scale bar in D ? 0.30 ?m in A,B, 0.48 ?m in C,D.
483NOREPINEPHRINE TRANSPORTER IN PFC
nantly dendritic shafts (24 of 37, 65%; Fig. 3A,B) and
occasionally dendritic spines (13 of 37; Fig. 3C,D). Within
profiles forming synapses, NET immunoreactivity was
also predominantly cytoplasmic (Fig. 3A,C,D).
Of the profiles that were singly labeled for NET, the
diameters ranged from 0.09 ?m in axons to 0.73 ?m in
varicosities, with an average of 0.40 ?m. The mean area of
the profiles was 0.44 ?m2, with a range of 0.03 ?m2to 1.64
?m2. The mean density of immunogold–silver labeling,
calculated as the number of gold particles per unit area,
was 19.8 in superficial layers and 24.9 in deep layers. This
laminar difference was not significant [t(139) ? 0.69, P ?
0.49]. The percentage of immunogold–silver labeling that
occurred on the plasmalemma was 20% and 24% in super-
ficial and deep layers, respectively.
A few immunogold–silver particles were routinely ob-
served in some dendrites within the PFC (not shown),
similar to the sparse labeling of serotonin transporter that
was also observed in dendrites of this region (Miner et al.,
2000). This labeling may have resulted from a nonspecific
immunogold–silver method, based on several observa-
tions: NET mRNA has not been reported in neurons of the
adult cerebral cortex (Lorang et al., 1994; Hoffman et al.,
1998); immunogold–silver labeling in dendrites is absent
in tissue incubated only in normal serum; NET labeling in
dendrites is rarely observed when immunoperoxidase is
used to localize this protein (unpublished observations).
NET immunogold–silver labeling was also occasionally
observed in a few glial processes (not shown). The pres-
transporter–immunoreactive (NET-ir) varicosities (large arrows) that
form synaptic contacts. A,B: Most frequently, NET-ir profiles form
symmetric synapses (small, straight arrows) onto dendritic shafts (d).
C,D: Less commonly, NET-ir profiles form symmetric or asymmetric
(curved arrow) synapses onto dendritic spines (s). The immunogold–
Electron photomicrographs illustrating norepinephrine
silver labeling for NET is typically not associated with the synapse
but is more often cytoplasmic. One exception is the axon (a) in B,
which exhibits gold–silver particles along the membrane (smallest
arrows). In some cases, the dendrites postsynaptic to NET-ir profiles
receive additional synaptic input from unlabeled terminals (Ut in
A,B). Scale bar in D ? 0.30 ?m in A,C, 0.48 ?m in B,D.
484L.H. MINER ET AL.
ence of NET in glia has not been previously described;
therefore, we cannot rule out the possibility that it is
artifactual. Because we cannot distinguish between arti-
factual labeling and a low level of actual protein expres-
sion, we used a more conservative criterion for specific
immunogold labeling in axons to avoid potential false-
positive results (see Materials and Methods section).
Characterization of profiles dually labeled for NET
By using modifications of our standard immu-
noperoxidase staining method, we were able to reliably
obtain immunoreactivity for D?H within some axons and
varicosities in the rat PFC, although the number of la-
beled profiles seemed to underrepresent the expected den-
sity of NE axons in this region. Light D?H labeling was
typically observed throughout immunoreactive profiles
but occasionally showed clustering within some portions.
The morphologic features of D?H-immunolabeled profiles,
including size, extent of myelination, and vesicle content,
were similar to those labeled for NET. Moreover, in many
instances, D?H and NET were colocalized within axonal
profiles, and in these cases, NET gold–silver particles
were primarily localized in the cytoplasm and only occa-
sionally on the plasma membrane (Fig. 4). However, over-
all immunolabeling for NET was substantially reduced in
this tissue, presumably because the staining conditions
that favored D?H were not optimal for NET labeling.
Therefore, it was not possible to obtain reliable quantita-
tive estimates of the extent of D?H and NET colocaliza-
Characterization of profiles dually labeled for NET
Profiles immunoreactive for both NET and TH
(NET?TH-ir) were infrequently observed (16 of 157, 10%;
Fig. 5A–C; Table 2). Moreover, it was evident that some of
the structural characteristics of these profiles were differ-
ent from the majority of profiles singly labeled for NET.
For instance, in the random sample, NET?TH-ir varicos-
ities were never observed to form classic synaptic contacts
in single sections. Furthermore, they contained NET-ir
gold–silver particles located predominantly along the
plasma membrane (76%) in both superficial and deep lay-
ers. Gold–silver particles were distributed to both the
inner and outer surfaces of the plasma membrane in du-
ally labeled profiles. A similar subcellular distribution of
NET was also noted in a minority of axons observed in
tissue that was processed only for NET (Fig. 5D). The
predominantly plasmalemmal localization of NET in du-
ally labeled profiles was significantly different compared
with that observed in singly labeled profiles [t(155) ? 7.24,
P ? 0.001]. Occasionally, the gold–silver labeling for NET
was not uniformly distributed along the immunoreactive
profile, suggesting that the proportion of TH-ir profiles
that also contained NET may have been underestimated,
depending on the plane of section.
Profiles immunoreactive for both NET?TH were more
frequently encountered in deep than in superficial layers
(Table 2), although this was not a significant difference
(Fisher’s exact test, P ? 0.19). This observation was some-
what surprising, in that we expected a higher proportion
of dually labeled fibers in superficial layers, given the
similar light microscopic distribution of labeled axons at
the border of layers I and II in TH, NET, and D?H immu-
nostained tissue. That we did not observe this distribution
may reflect that we did not systematically sample from
this border region. The diameter of NET?TH-ir profiles in
tive for both norepinephrine transporter (NET) and dopamine
?-hydroxylase (D?H, large arrows) in the superficial layers of the rat
prefrontal cortex. Light immunoperoxidase labeling for D?H is local-
ized throughout these profiles, with heavier reaction product clus-
Electron photomicrographs showing profiles immunoreac-
tered in some portions. Immunogold–silver particles for NET appear
predominantly in the cytoplasm and less frequently along the plas-
malemmal surface. Different portions of the same dually labeled axon
are shown in A and B, whereas a dually labeled profile from another
animal is shown in C. Scale bar ? 0.30 ?m in C (applies to A–C).
485NOREPINEPHRINE TRANSPORTER IN PFC
our sample ranged from 0.06 ?m to 0.66 ?m, with a mean
of 0.30 ?m. The area of labeled profiles ranged from 0.01
?m2to 1.0 ?m2, having an average of 0.38 ?m2. Although
the area measurements varied between the layers, there
were no significant differences [t(14) ? 1.15, P ? 0.27].
The density of gold–silver labeling was greater in deep
than in superficial layers (13.2 vs. 83.7 particles per unit
area); however, this difference was also not significant
[t(14) ? 0.98, P ? 0.34], most likely due to the small
sample and large variance.
Because the dorsal striatum receives dense DA inputs
but does not contain many NE terminals (Swanson and
Hartman, 1975), dual-labeling immunocytochemistry for
NET and TH was performed on sections of the striatum
nephrine transporter–immunoreactive (NET-ir) profiles (large ar-
rows) in the rat prefrontal cortex. A–C: In tissue processed for both
NET by immunogold–silver and tyrosine hydroxylase by immunoper-
oxidase, profiles containing both markers exhibit gold–silver particles
that are predominantly on the plasma membrane. D: In tissue pro-
cessed only for NET by immunogold–silver, occasional profiles are
Electron photomicrographs illustrating atypical norepi-
observed that have a plasmalemmal distribution of NET similar to
dually labeled profiles and markedly different from the majority of
axons singly labeled for NET in this study. For NET-ir profiles show-
ing primarily plasmalemmal labeling, gold–silver particles are dis-
tributed along either the inner or outer surface of the membrane.
Scale bar in D ? 0.48 ?m in A, 0.30 ?m in B–D.
486L.H. MINER ET AL.
and examined by electron microscopy to verify the speci-
ficity of single and dual labeling for NET. Many TH-ir
profiles, presumably DA terminals, were observed in this
region. However, NET-ir profiles were sparse, and no pro-
files were found to be dually labeled for NET?TH (data
not shown). Finally, it should be mentioned that the sub-
cellular distribution of NET and the degree of NET?TH
dual labeling in the PFC was not attributable to the ad-
ministration of a zinc chelator before the perfusion, as the
same pattern of labeling was observed in tissue from the
rat in which the zinc chelator was omitted.
Spatial relationships between profiles labeled for
NET or TH.
In addition to the random sample, we con-
ducted a systematic sample of the PFC to determine the
spatial relationships of NET-ir and TH-ir profiles with
each other. In this systematic sample, 234 TH-ir profiles
were observed in superficial layers and 366 in deep layers.
Of the total number of TH-ir profiles, 25 were dually
labeled for NET in superficial layers (10%) and 27 in deep
layers (7%). Therefore, only 8% of TH-ir profiles expressed
immunoreactivity for NET, with the remaining 92% of
profiles being immunopositive only for TH in single sec-
tions. Data from these singly labeled axons are presented
in Table 3. The NET?TH-ir profiles in this systematic
sample possessed similar characteristics to dually labeled
profiles observedwith random
immunogold–silver particles for NET were located mainly
on the plasmalemma, and the frequency of observed syn-
aptic contacts was rare.
The spatial relationships between profiles in adjacent
regions of the neuropil labeled for TH or NET (Fig. 6) are
outlined in Table 3. In superficial layers, 15% of TH-ir
profiles were within the same 13.8 ?m2field as NET-ir
profiles; in deep layers, 18% of TH-ir axons had this prox-
imity to NET-ir structures. Occasionally, one TH-ir profile
was within the same field as multiple NET-ir profiles (Fig.
6A). Typically, the TH-ir and NET-ir profiles were sepa-
rated by an average distance of 1.20 ?m. However, the
range of distances varied from 0.02 ?m to 3.63 ?m. No
laminar difference in the mean distance between profiles
was observed [t(101) ? 0.04, P ? 0.97].
TH-ir and NET-ir profiles were rarely in apposition to
each other (Fig. 7), although it is interesting that, in one
case, the immunogold–silver labeling for NET was ob-
served near the site of apposition with a TH-ir axon (Fig.
7B). More frequently, these two profile types were sepa-
rated by unlabeled structures. TH-ir and NET-ir profiles
were never seen to contact the same dendritic structures
in the present study. Occasionally, the TH-ir or NET-ir
profiles were seen to contact or synapse onto separate
dendritic shafts or spines within closely adjacent regions
of the neuropil (Fig. 6C).
This study provides the first detailed ultrastructural
investigation of NET immunoreactivity in the rat PFC
and its relationship to TH immunolabeling, both within
NE axons and in presumed DA afferents. Using dual-
labeling immunocytochemistry, the present study re-
vealed that NET-ir profiles in the rat prelimbic PFC ex-
hibited the morphologic features of NE fibers, and the
majority of these profiles contained mostly cytoplasmic
NET and no detectable TH immunolabeling. The NET-ir
fibers that were dually labeled for TH (approximately
10%) rarely were found to form synapses and contained
predominantly plasmalemmal NET. NET-ir profiles were
typically separated from TH-ir, presumably DA, profiles
by an average distance of at least 1.20 ?m that did not
vary across the cortical layers. Finally, the NET-ir profiles
were infrequently apposed to TH-ir profiles, and these
immunolabeled axons never were observed to contact the
same dendritic structures.
The technical limitations of the present study are sim-
ilar to those previously described (Miner et al., 2000).
Most importantly, the findings represent the relative, not
the absolute, quantity of NET-ir and TH-ir profiles in the
rat PFC for several reasons. First, the low detergent levels
required for ultrastructural preservation of tissue for elec-
tron microscopy decreased the total amount of labeling
that could be observed. Moreover, the degree of immuno-
staining is dependent on antibody penetration, which is
maximal only at the tissue surface. To minimize this con-
tributor to false negatives, sampling was performed,
therefore, only at the edge of the tissue–plastic interface.
Second, the number of NET profiles in the present study
was likely to be further underestimated, because the
immunogold–silver procedure is less sensitive than immu-
noperoxidase, likely due to limited penetration of the 1-nm
gold particle on the secondary antibody (Chan et al.,
1990). In addition, we adopted a more conservative crite-
rion for determining specific immunogold–silver labeling
in the current study than used to identify transporters in
other brain regions (Garzon et al., 1999; Pickel and Chan,
TABLE 2. Profiles Dually Immunoreactive for NET ? TH in a Random
Sample From the Rat Prefrontal Cortex1
Total number of NET-ir profiles observed
Number of NET ? TH-ir profiles observed (% of
Area (?m2) of NET ? TH-ir profiles (mean ? SD)
Diameter (?m) of NET ? TH-ir profiles
(mean ? SD)
Density of immunogold–silver labeling (particles
per unit area; mean ? SD)
Percentage of immunogold–silver labeling on the
Number (%) of NET ? TH-ir profiles forming
6 (7%)10 (14%)
0.47 ? 0.27
0.32 ? 0.06
0.30 ? 0.31
0.28 ? 0.19
13.2 ? 7.183.7 ? 173.6
68 ? 34% 84 ? 28%
1All comparisons between superficial and deep layers not significant, P ? 0.19. NET,
norepinephrine transporter; TH, tyrosine hydroxylase; -ir, immunoreactive.
TABLE 3. Associations Between Profiles Immunoreactive for TH or NET in
a Systematic Sample From the Rat Prefrontal Cortex1
Total area examined (?m2)
Number of singly-labeled TH-ir profiles
Number (%) of TH-ir profiles in the same
field* as NET-ir profiles
Number of NET-ir profiles observed
Number of TH-ir and NET-ir profiles in
Number of TH-ir and NET-ir profiles
contacting common dendritic targets
Distance (?m) between TH-ir and NET-ir
profiles in the same field* (mean ? SD)
32 (15%)60 (18%)
1.20 ? 0.85
1.21 ? 0.85
*area encompassed by an electron micrograph at 20,000? magnification (13.8 ?m2)
1all comparisons between superficial and deep layers not significant, P ? 0.31. TH,
tyrosine hydroxylase; NET, norepinephrine transporter; -ir, immunoreactive.
487NOREPINEPHRINE TRANSPORTER IN PFC
distance between profiles immunoreactive for tyrosine hydroxylase
(TH, open arrows) or norepinephrine transporter (NET, closed ar-
rows) within adjacent regions of the neuropil. Most typically, the two
profile types are separated by unlabeled elements in the microenvi-
ronment. In A, two NET-immunoreactive (-ir) axons are within the
same field as one TH-ir profile, whereas in B, one NET-ir varicosity is
Electron photomicrographs illustrating the typical spatial
located close to two TH-ir profiles. In C, the immunoreactive profiles
contact different dendritic spines (small, straight arrows) by forming
a close apposition (TH-ir axon) or a synapse (NET-ir axon). Distances
between TH-ir and NET-ir profiles: A, 1.72 ?m and 1.66 ?m; B, 1.36
?m and 0.29 ?m; C, 1.24 ?m. Scale bar ? 0.30 ?m in C (applies to
488L.H. MINER ET AL.
1999). In part, this decision was based on our observation
of a low level of immunogold–silver particles in some den-
drites and glia in the PFC. Because these structures are
not known to express mRNA for NET (Hoffman et al.,
1998), it was difficult to rule out the possibility of artifac-
tual labeling. Our more conservative criterion for specific
gold labeling, therefore, reduced the incidence of false-
positive identification of axonal profiles. However, this
approach may have increased the likelihood that a portion
of lightly labeled NET profiles, possibly including some
dually labeled for TH, was underestimated in the present
study. In addition, the inclusion of small profiles with
predominantly plasmalemmal gold particles may have
contributed to an underestimation of the extent of cyto-
plasmic NET labeling.
Despite these limitations, the use of immunogold–silver
labeling for NET was desirable for several reasons. First,
gold–silver particles do not obscure membranes and or-
ganelles, making the description of morphologic charac-
teristics more accurate. Second, the presence of immuno-
gold particles is a more reliable indicator of NET’s
subcellular location (e.g., cytoplasmic vs. plasmalemmal),
which was a specific aim of this study. Finally, using
immunogold–silver for NET allowed the use of the more
sensitive avidin–biotin–peroxidase method (Hsu et al.,
1981) for localization of TH, so that the proportion of
NET-ir profiles that also contained TH (approximately
10%) was a more accurate estimate.
It should be noted that the distance measurements re-
ported in the present investigation are given for relative
quantitative purposes only, as there is likely to be tissue
shrinkage during the processing for immunocytochemis-
try and electron microscopy (Hillman and Deutsch, 1978).
Furthermore, the relationship between NET-ir and TH-ir
profiles was examined only in coronal sections, and this
relationship might vary in different planes of section. Fi-
nally, we measured the distances only between NET-ir
and TH-ir profiles that were observed within the same
13.8 ?m2field, so that naturally many more labeled pro-
files fall outside this arbitrary boundary. For these rea-
sons, we have noted that the average distance between
these profile types is at least 1.2 ?m.
Profiles expressing NET
Several observations suggest that immunoreactivity for
NET represents a specific marker for NE axons and ter-
minals. First, the light microscopic labeling for NET ob-
served in the present and previous studies (Schroeter et
al., 2000) is more similar in orientation and laminar dis-
tribution to D?H immunoreactivity than to TH or DA
immunostaining. The less-beaded morphology of NET- as
opposed to D?H-labeled fibers probably reflects the distri-
bution of antigens within different intracellular compart-
ments. Moreover, the present study provides the first ul-
trastructural evidence that NET and D?H are colocalized
in the same axonal profiles. Second, the ultrastructural
characteristics of NET-labeled profiles match those of NE
varicosities in other parts of the rat cerebral cortex: lack of
myelination, occasional content of dense-cored vesicles,
and moderate rate of observed synaptic incidence (Beau-
det and Descarries, 1978, 1984; Se ´gue ´la et al., 1990). The
frequency of synapses observed in the present analysis of
the PFC was actually somewhat higher than previously
immunoreactive for tyrosine hydroxylase (TH, open arrows) or nor-
epinephrine transporter (NET, large, closed arrows) that lie in close
association to each other. Infrequently, direct appositions (arrow-
heads)are observed between
A,B: Electron photomicrographs illustrating rare profiles
these profiles.B:When TH-
immunoreactive (-ir) and NET-ir profiles are directly apposed, immu-
nogold labeling for NET is sometimes seen near the site of apposition.
Distances between TH-ir and NET-ir profiles: A, 0.02 ?m; B, 0.02 ?m.
Scale bar ? 0.30 ?m in A (applies to A,B).
489 NOREPINEPHRINE TRANSPORTER IN PFC
reported, although this finding may reflect regional differ-
The nearly undetectable level of NET labeling in control
striatal sections further argues for the specificity of the
NET antibody for labeling NE fibers. Moreover, NET
mRNA is localized exclusively in neurons immunoreactive
for D?H (Lorang et al., 1994; Hoffman et al., 1998), and
NET immunoreactivity is absent in sections from NET
knockout animals (Savchencko and Blakely, unpublished
observations). Finally, NET immunolabeling disappears
(Schroeter et al., 2000).
In addition to specificity, it is likely that NET immuno-
reactivity also is a sensitive method for visualizing the
majority of NE terminals, in that subpopulations of NE
cells that do not have the capacity to synthesize NET have
never been described (Lorang et al., 1994; Hoffman et al.,
1998). This is important, given that other monoamine
neurons have been reported to include populations with
low or nondetectable transporter expression (Kosofsky
and Molliver, 1987; Mamounas et al., 1991; Shimada et
al., 1992; Axt et al., 1995; Sesack et al., 1998).
In the present study, NET was frequently localized
within the cytoplasm, sometimes with no obvious associ-
ation to organelles in single sections. In other cases, NET
immunolabeling was found along the edges of apparent
synaptic vesicles and larger tubulovesicles (Nirenberg et
al., 1997). It has been hypothesized that membrane traf-
ficking is a major mechanism through which transporters
are functionally regulated (Beckman and Quick, 1998;
Blakely et al., 1998). Therefore, it should be considered
that cytoplasmic NET within PFC varicosities represents
a pool of transporters that could be readily inserted into
the membrane in response to changes in cellular activity.
It is likely that plasmalemmal gold particles for NET
represent functional transporters, similar to what has
been demonstrated for membrane-associated receptors
(Dumartin et al., 1998), although whether this is specifi-
cally the case for NET requires experimental demonstra-
tion. The predominantly cytoplasmic localization of NET
stands in distinct contrast to the mainly plasmalemmal
localization of the serotonin transporter in the PFC (Miner
et al., 2000) or the DA transporter in the striatum (Niren-
berg et al., 1996; Hersch et al., 1997). Cytoplasmic NET
may reflect the possibility that NE was not being released
from most PFC varicosities at the time of fixation, and
thus, NET did not need to be inserted in the membrane.
Alternatively, NE may have been undergoing active re-
lease from varicosities, but because NET was internalized
at the same time, NE was permitted to diffuse farther. It
will be important in future studies to determine the exact
conditions under which cytoplasmic NET is mobilized to
the plasma membrane of PFC NE axons.
Localization of TH within NET-ir profiles
The demonstrated specificity and sensitivity of NET
immunoreactivity made this an optimal marker to test the
coexpression of TH in identified NE axons in the rat PFC.
Moreover, theuseof the
peroxidase method to label TH ensured that the propor-
tion of NET-ir profiles that contained TH was not substan-
remarkable that, in the present study, only a small pro-
portion of the profiles expressing NET exhibited immuno-
reactivity for TH. It should be noted that profiles dually
labeled for NET and TH always contained abundant TH-
ir. The all-or-none nature of these results suggests that
NET-labeled axons with low levels of TH immunoreactiv-
ity were not overlooked. However, it is always possible
that NET-ir axons in the rat PFC contain levels of TH too
low to detect even with the most sensitive methods.
Our findings are consistent with previous investigations
reporting that several TH antisera, both monoclonal and
polyclonal, label predominantly DA axons in cortical areas
receiving both NE and DA innervation (Pickel et al.,
1975b; Lewis et al., 1987, 1988; Noack and Lewis, 1989).
Many of these were light microscopic studies that used
higher detergent levels for maximal antibody penetration
of the tissue. It should be noted that TH antibodies that
fail to label cortical NE terminals do stain NE cell bodies
and proximal dendrites within the locus coeruleus, albeit
to a lesser degree than TH in DA neurons (Pickel et al.,
1975a, c). This finding suggests that the antibodies have
the potential to recognize TH where it is present. More-
over, it is unlikely that the antibody used in the present
study does not recognize the isoform of TH expressed in
NE neurons, as only one isoform of TH has been reported
in the rodent brain (Brown et al., 1987; Ichikawa et al.,
1991). The functional basis for the lack of detectable im-
munoreactivity for TH within most NE terminals in the
PFC is not known. A complete review of this literature is
beyond the scope of the study, although some of the more
salient possibilities are discussed here.
As previously mentioned, the concentration of the TH
enzyme may be too low in cortical NE profiles to be de-
tected by immunocytochemical methods (Pickel et al.,
1975b; Noack and Lewis, 1989). Indeed, biochemical stud-
ies suggest that cortical NE terminals have less TH than
do cortical DA varicosities (Emson and Koob, 1978;
Schmidt and Bhatnagar, 1979). The mechanism whereby
TH fails to reach detectable levels in the axons of neurons
that otherwise synthesize this enzyme is not known.
It is possible that TH is not detectable in most cortical
NE terminals due to variations in the TH molecule itself.
For instance, reported differences in the molecular weight
(Joh and Reis, 1975; Reis et al., 1975) and the pH optimum
(Acheson et al., 1981) of TH in NE vs. DA neurons might
influence immunostaining. However, it is unlikely that
the phosphorylation state of the TH enzyme altered anti-
body recognition for at least two reasons. First, the phos-
phorylation of TH occurs in the N-terminal regulatory
domain (Campbell et al., 1986; Haycock, 1990), a portion
of the enzyme not targeted by the monoclonal antibody.
Second, the dephosphorylation of TH is reportedly rather
rapid (D. Lewis and J. Haycock, personal communication)
and so is unlikely to account for the absence of detectable
immunolabeling in histologically prepared tissue.
The relative absence of TH immunoreactivity in most
cortical NE axons raises the obvious question of how NE
terminals can synthesize and release NE. One potential
mechanism may rely on the ability of NET to transport DA
(Horn, 1973; Raiteri et al., 1977; Gu et al., 1996). In areas
where NE and DA afferents overlap, NE terminals might
be able to synthesize NE from extracellular DA. This
suggestion is especially consistent with existing data in
the rat PFC, because of the low expression of the DA
transporter (Sesack et al., 1998), the greater extracellular
diffusion of DA (Garris et al., 1993; Garris and Wightman,
1994; Cass and Gerhardt, 1995), and the demonstrated
clearance of DA by NE terminals in this region (Gresch et
490L.H. MINER ET AL.
al., 1995; Tanda et al., 1997; Yamamoto and Novotney,
1998). However, one potential problem with a mechanism
that relies on NET clearance of extracellular DA is the
predominantly cytoplasmic localization of NET. Possibly,
the amount of plasmalemmal NET that is present is suf-
ficient for this purpose. Based on our findings, it is not
possible to speculate on what happens to DA that is trans-
ported by NET into NE profiles. Most probably, it is con-
verted by D?H into NE. Alternatively, it is possible that
DA is released by NE profiles (Devoto et al., 2001). In this
case, DA may be operating as a false transmitter within
NE terminals (Vanhatalo and Soinila, 1995).
It is possible that sparse TH immunoreactivity in the
majority of PFC NE axons reflects a low activity state
associated with minimal NE release. In the absence of
neurochemical data from our rats, this idea must remain
speculative. However, if true, then increasing the activity
of NE neurons that project to the PFC would be predicted
to increase TH expression. In this regard, it should be
noted that certain manipulations such as electroconvul-
sive shock, reserpine treatment, or acute stress alter TH
mRNA and/or activity levels within the locus coeruleus
(Reis et al., 1974; Zigmond et al., 1974, 1995; Kapur et al.,
1993; Serova et al., 1999; Sands et al., 2000). However, it
is difficult to predict the exact manipulations that might
increase TH in the PFC, particularly given that mRNA
levels do not always predict the amount of protein synthe-
sized (Kumer and Vrana, 1996).
If the concept of different activity states is viable, it
further suggests that the small population of NET-labeled
profiles that also contain TH may operate at higher activ-
ity levels that require greater TH expression. This idea is
also consistent with the primarily plasmalemmal localiza-
tion of NET in those dually labeled profiles, suggesting
that higher activity would be associated with greater NE
reuptake after release. Alternatively, rather than reflect-
ing a differential activity state, profiles dually labeled for
NET and TH may represent a subpopulation of NE termi-
nals in the PFC that arise from an area other than the
locus coeruleus. The possibility that these axons derive
from the nucleus of the solitary tract (Zagon et al., 1994) or
another brainstem site needs to be examined. Finally, it
should be considered that NET-ir axons lacking TH might
synthesize and release trace amines such as tyramine,
which is a substrate for NET (Burnette et al., 1996).
Relationship between profiles expressing
NET and presumed DA terminals
The low degree of colocalization of TH and NET sup-
ports prior observations that TH immunoreactivity is pri-
marily representative of DA rather than NE axons in
cortex (Pickel et al., 1975b; Lewis et al., 1987, 1988; Noack
and Lewis, 1989). The use of single labeling for TH as a
means for identifying DA axons is strengthened by the
ability in ultrastructural dual labeling studies to exclude
profiles labeled for both TH and NET as representing
cortical NE fibers. Based on these considerations, we re-
garded axons singly labeled for TH in this study as likely
to be DA, although we acknowledge that a few may have
represented NE fibers in which NET was undetected.
Because local application of compounds that block the
NET increases not only extracellular levels of NE but also
DA (Jordan et al., 1994; Tanda et al., 1994, 1997; Gresch
et al., 1995; Yamamoto and Novotney, 1998), it was pre-
dicted that profiles expressing NET would be in close
spatial association with DA terminals in the PFC. How-
ever, such a relationship was not observed in our study.
Rather, it was unusual to find the two profile types ap-
posed to each other. Nevertheless, the average spatial
separation of only 1.2 ?m observed between these two
systems suggests that NE and DA can readily interact by
extrasynaptic means, as has been proposed by others
(Descarries and Umbriaco, 1995; Mundorf et al., 2001). In
addition, NE and DA afferents also were never observed to
synapse onto common postsynaptic elements, although it
is possible that they target the same cortical neurons at
spatially distant portions of the dendritic tree.
The degree of functional interaction between NE and
DA of course depends on the localization of the receptors
for these transmitters. Of interest, ?2A-adrenoceptor im-
munoreactivity has been observed on axons in the PFC
(Aoki et al., 1994, 1998). Moreover, administration of ?2-
adrenoceptor antagonists alters extracellular DA levels in
this region (Matsumoto et al., 1998; Hertel et al., 1999),
although the exact mechanism for this effect is unclear.
The possibility that some adrenergic receptors act as het-
eroreceptors on DA axons within the PFC is supported by
observations that DA neurons within the ventral tegmen-
tal area, the source of the DA innervation of the PFC
(Berger et al., 1976), express ?2C-adrenoceptor immuno-
reactivity (Lee et al., 1998). However, it remains to be
determined whether these receptors are trafficked into
cortical DA axons. Much less is known about the localiza-
tion of DA receptors on NE axons in the PFC. Presynaptic
DA receptors have been occasionally observed in that area
(Smiley et al., 1994; Bergson et al., 1995). However, to our
knowledge, there are presently no reports of mRNA for DA
receptors in locus coeruleus neurons.
Drugs that block either the NET or the serotonin trans-
porter have high efficacy in the treatment of depression
(Nelson, 1999; Frazer, 2000; Kent, 2000; Mo ¨ller, 2000).
However, it takes several weeks for maximal therapeutic
benefit, suggesting that compensatory effects of chronic
drug administration may be more relevant for antidepres-
sant action. These compensatory processes might involve
biochemically distinct mechanisms, depending on whether
noradrenergic or serotonergic transmission is altered. On
the other hand, it may be that activation of either neuro-
transmitter system results in similar effects on an inter-
mediary process (Frazer, 2000). One such intermediary
may be the DA system. In line with this hypothesis, the
ability of both antidepressant drug treatment and acute
stress to regulate PFC DA levels is altered after chronic
administration of these drugs (Carlson et al., 1996; Tanda
et al., 1996; Dazzi et al., 2001; Page and Lucki, 2002).
Further research is necessary to determine the mecha-
nisms underlying these functional interactions. However,
our findings suggest that primarily extrasynaptic pro-
cesses are involved, as NE or serotonin (Miner et al., 2000)
must diffuse approximately 1 ?m or more to encounter DA
Acheson AL, Kapatos G, Zigmond MJ. 1981. The effects of phosphorylating
conditions on tyrosine hydroxylase activity are influenced by assay
conditions and brain region. Life Sci 28:1407–1420.
Aoki C, Go C-G, Venkatesan C, Kurose H. 1994. Perikaryal and synaptic
491 NOREPINEPHRINE TRANSPORTER IN PFC
localization of alpha2A-adrenergic receptor-like immunoreactivity.
Brain Res 650:181–204.
Aoki C, Venkatesan C, Go C-G, Forman R, Kurose H. 1998. Cellular and
subcellular sites for noradrenergic action in the monkey dorsolateral
prefrontal cortex as revealed by the immunocytochemical localization
of noradrenergic receptors and axons. Cereb Cortex 8:269–277.
Arnsten AFT. 1997. Catecholamine regulation of the prefrontal cortex.
J Psychopharmacol 11:151–162.
Aston-Jones G, Chaing C, Alexinsky T. 1991. Discharge of noradrenergic
locus coeruleus neurons in behaving rats and monkeys suggests a role
in vigilance. Prog Brain Res 88:501–520.
Axelrod J, Kopin IJ. 1969. The uptake, storage, release, and metabolism of
noradrenaline in sympathetic nerves. Prog Brain Res 31:21–32.
Axt KJ, Molliver ME, Qian Y, Blakely RD. 1995. Subtypes of 5-HT axons
differ in their expression of serotonin transporter. Soc Neurosci Abstr
Beaudet A, Descarries L. 1978. The monoamine innervation of rat cerebral
cortex: synaptic and nonsynaptic axon terminals. Neuroscience 3:851–
Beaudet A, Descarries L. 1984. Fine structure of monoamine axon termi-
nals in cerebral cortex. In: Descarries L, Reader TR, Jasper HH, edi-
tors. Monoamine innervation of the cerebral cortex. New York: Liss. p
Beckman ML, Quick MW. 1998. Neurotransmitter transporters: regulators
of function and functional regulation. J Membr Biol 164:1–10.
Berger B, Thierry A-M, Tassin JP, Moyne MA. 1976. Dopaminergic inner-
vation of the rat prefrontal cortex: a fluorescence histochemical study.
Brain Res 106:133–145.
Bergson C, Mrzljak L, Smiley JF, Pappy M, Levenson R, Goldman-Rakic
PS. 1995. Regional, cellular, and subcellular variations in the distri-
bution of D1and D5dopamine receptors in primate brain. J Neurosci
Berman KF, Weinberger DR. 1990. The prefrontal cortex in schizophrenia
and other neuropsychiatric diseases: in vivo physiological correlates of
cognitive deficits. Prog Brain Res 85:521–537.
Blakely RD, Ramamoorthy S, Schroeter S, Qian Y, Apparsundaram S,
Galli A, DeFelice J. 1998. Regulated phosphorylation and trafficking of
antidepressant-sensitive serotonin transport proteins. Biol Psychiatry
Bonisch H, Bru ¨ss M. 1994. The noradrenaline transporter of the neuronal
plasma membrane. Ann N Y Acad Sci 733:193–202.
Brown ER, Coker GT, O’Malley KL. 1987. Organization and evolution of
the rat tyrosine hydroxylase gene. Biochemistry 26:5208–5212.
Burnette WB, Bailey MD, Kukoyi S, Blakely RD, Trowbridge CG, Justice
JB Jr. 1996. Human norepinephrine transporter kinetics using rotat-
ing disk electrode voltammetry. Anal Chem 68:2932–2938.
Bymaster FP, Katner JS, Nelson DL, Hemrick-Luecke SK, Threlkeld
PG, Heiligenstein JH, Morin SM, Gehlert DR, Perry KW. 2002.
Atomoxetine increases extracellular levels of norepinephrine and
dopamine in prefrontal cortex of rat: a potential mechanism for
efficacy in attention deficit/hyperactivity disorder. Neuropsychop-
Campbell DG, Hardie DG, Vulliet PR. 1986. Identification of the four
phosphorylation sites in the N-terminal region of tyrosine hydroxylase.
J Biol Chem 261:10489–10492.
Carlson JN, Visker KE, Nielsen DM, Keller RW Jr, Glick SD. 1996.
Chronic antidepressant drug treatment reduces turning behavior and
increases dopamine levels in the medial prefrontal cortex. Brain Res
Cass WA, Gerhardt GA. 1995. In vivo assessment of dopamine uptake in
rat medial prefrontal cortex: comparison with dorsal striatum and
nucleus accumbens. J Neurochem 65:201–207.
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–
Dazzi L, Spiga F, Pira L, Ladu S, Vacca G, Rivano A, Jentsch JD, Biggio G.
2001. Inhibition of stress- or anxiogenic-drug-induced increases in do-
pamine release in the rat prefrontal cortex by long-term treatment
with antidepressant drugs. J Neurochem 76:1212–1220.
Descarries L, Umbriaco D. 1995. Ultrastructural basis of monoamine and
acetylcholine function in CNS. Semin Neurosci 7:309–318.
Devoto P, Flore G, Pani L, Gessa GL. 2001. Evidence for co-release of
noradrenaline and dopamine from noradrenergic neurons in the cere-
bral cortex. Mol Psychiatry 6:657–664.
Drevets WC, Videen TO, Price JL, Preskorn SH, Carmichael ST, Raichle
ME. 1992. A functional anatomical study of unipolar depression. J Neu-
Dumartin B, Caille ´ I, Gonon F, Bloch B. 1998. Internalization of D1
dopamine receptor in striatal neurons in vivo as evidence of activation
by dopamine agonists. J Neurosci 18:1650–1661.
Emson PC, Koob GF. 1978. The origin and distribution of dopamine-
containing afferents to the rat prefrontal cortex. Brain Res 142:249–
Florin-Lechner SM, Druhan JP, Aston-Jones G, Valentino RJ. 1996. En-
hanced norepinephrine release in prefrontal cortex with burst stimu-
lation of the locus coeruleus. Brain Res 742:89–97.
Frazer A. 2000. Norepinephrine involvement in antidepressant action.
J Clin Psychiatry 61(Suppl 10):25–30.
Fritz JD, Jayanthi LD, Thoreson MA, Blakely RD. 1998. Cloning and
chromosomal mapping of the murine norepinephrine transporter.
J Neurochem 70:2241–2251.
Fuster JM. 1997. The prefrontal cortex: anatomy, physiology and neuro-
psychology of the frontal lobe. New York: Lippincott-Raven.
Garris PA, Wightman RM. 1994. Different kinetics govern dopaminergic
transmission in the amygdala, prefrontal cortex, and striatum: an in
vivo voltametric study. J Neurosci 14:442–450.
Garris PA, Collins LB, Jones SR, Wightman RM. 1993. Evoked extracel-
lular dopamine in vivo in the medial prefrontal cortex. J Neurochem
Garzon M, Vaughan RA, Uhl GR, Kuhar MJ, Pickel VM. 1999. Cholinergic
axon terminals in the ventral tegmental area target a subpopulation of
neurons expressing low levels of the dopamine transporter. J Comp
Gresch PJ, Sved AF, Zigmond MJ, Finlay JM. 1995. Local influences of
endogenous norepinephrine on extracellular dopamine in rat medial
prefrontal cortex. J Neurochem 65:111–116.
Gu HH, Wall S, Rudnick G. 1996. Ion coupling stoichiometry for the
norepinephrine transporter in membrane vesicles from stably trans-
fected cells. J Biol Chem 271:6911–6919.
Haycock JW. 1990. Involvement of serine-31 in phosphorylation of tyrosine
hydroxylase in PC12 cells. J Biol Chem 265:11682–11691.
Hersch SM, Yi H, Heilman CJ, Edwards RH, Levey AI. 1997. Subcellular
localization and molecular topology of the dopamine transporter in the
striatum and substantia nigra. J Comp Neurol 388:211–227.
Hertel P, Nomikos GG, Svensson TH. 1999. Idazoxan preferentially in-
creases dopamine output in the rat medial prefrontal cortex at the
nerve terminal level. Eur J Pharmacol 371:153–158.
Hillman H, Deutsch K. 1978. Area changes in slices of rat brain during
preparation for histology or electron microscopy. J Microsc 114:77–
Hoffman BJ, Hansson SR, Mezey E, Palkovits M. 1998. Localization and
dynamic regulation of biogenic amine transporters in the mammalian
central nervous system. Front Neuroendocrinol 19:187–231.
Horn AS. 1973. Structure-activity relations for the inhibition of cate-
cholamine uptake into synaptosomes from noradrenaline and dopa-
minergic neurons in rat brain homogenates. Br J Pharmacol 47:332–
Hsu S-M, Raine L, Fanger H. 1981. Use of avidin-biotin-peroxidase com-
plex (ABC) in immunoperoxidase techniques. J Histochem Cytochem
Ichikawa S, Sasaoka T, Nagatsu T. 1991. Primary structure of mouse
tyrosine hydroxylase deduced from its cDNA. Biochem Biophys Res
Iversen LL. 1971. Role of transmitter uptake mechanisms in synaptic
neurotransmission. Br J Pharmacol 41:571–591.
Joh TH, Reis DJ. 1975. Different forms of tyrosine hydroxylase in central
dopaminergic and noradrenergic neurons and sympathetic ganglia.
Brain Res 85:146–151.
Jordan S, Kramer GL, Zukar PK, Moeller M, Petty F. 1994. In vivo biogenic
amine efflux in medial prefrontal cortex with imipramine, fluoxetine,
and fluvoxamine. Synapse 18:294–297.
Kapur S, Austin MC, Underwood MD, Arango V, Mann JJ. 1993. Electro-
convulsive shock increases tyrosine hydroxylase and neuropeptide Y
gene expression in the locus coeruleus. Brain Res 18:121–126.
Kent JM. 2000. SNaRIs, NaSSAs, and NaRIs: new agents for the treat-
ment of depression. Lancet 355:911–918.
Klimek V, Stockmeier C, Overholser J, Meltzer HY, Kalka S, Dilley G,
492 L.H. MINER ET AL.