THE JOURNAL OF COMPARATIVE NEUROLOGY 403:378–390 (1999)

Dense-Cored Vesicles, Smooth

Endoplasmic Reticulum, and
Mitochondria Are Closely Associated
With Non-Specialized Parts of Plasma
Membrane of Nerve Terminals:
Implications for Exocytosis and Calcium
Buffering by Intraterminal Organelles
ANNA LYSAKOWSKI,1 HUNTER FIGUERAS,2 STEVEN D. PRICE,1 AND YAN-YI PENG2,3*
1Department of Anatomy and Cell Biology, University of Illinois at Chicago,

Chicago, Illinois 60612

2Department of Pharmacological and Physiological Sciences, University of Chicago,

Chicago, Illinois 60637

3Committees on Neurobiology and Cell Physiology, University of Chicago,

Chicago, Illinois 60637

ABSTRACT
To determine whether there are anatomical correlates for intraterminal Ca2⫹ stores to
regulate exocytosis of dense-cored vesicles (DCVs) and whether these stores can modulate
exocytosis of synaptic vesicles, we studied the spatial distributions of DCVs, smooth
endoplasmic reticulum (SER), and mitochondria in 19 serially reconstructed nerve terminals
in bullfrog sympathetic ganglia. On average, each bouton had three active zones, 214 DCVs,
26 SER fragments (SERFs), and eight mitochondria. DCVs, SERFs and mitochondria were
located, on average, 690, 624, and 526 nm, respectively, away from active zones. Virtually no
DCVs were within ‘‘docking’’ (i.e., ⱕ50 nm) distances of the active zones. Thus, it is unlikely
that DCV exocytosis occurs at active zones via mechanisms similar to those for exocytosis of
synaptic vesicles. Because there were virtually no SERFs or mitochondria within 50 nm of any
active zone, Ca2⫹ modulation by these organelles is unlikely to affect ACh release evoked by a
single action potential. In contrast, 30% of DCVs and 40% of SERFs were located within 50 nm
of the nonspecialized regions of the plasma membrane. Because each bouton had at least one
SERF within 50 nm of the plasma membrane and most of these SERFs had DCVs, but not
mitochondria, near them, it is possible for Ca2⫹ release from the SER to provide the Ca2⫹
necessary for DCV exocytosis. The fact that 60% of the mitochondria had some part within 50
nm of the plasma membrane means that it is possible for mitochondrial Ca2⫹ buffering to
affect DCV exocytosis. J. Comp. Neurol. 403:378–390, 1999. r 1999 Wiley-Liss, Inc.

Indexing terms: boutons; presynaptic nerve terminals; bullfrog sympathetic ganglion

Most vertebrate nerve terminals contain both fast-acting Exocytosis of SVs occurs at specialized release sites, i.e.,
neurotransmitters and a neuropeptide. The former is con- active zones. Synaptic vesicles are clustered at active
tained in synaptic vesicles (SVs), whereas the latter is con-
tained in dense-cored vesicles (DCVs). Although neuropep-
tides have many important functions, the presynaptic Grant sponsor: NIH; Grant numbers: DC02521 and NS32429.
regulation of their release from co-transmitting nerve termi- *Correspondence to: Yan-yi Peng, Ph.D., Dept. of Pharmacological and
Physiological Sciences, University of Chicago, 947 E. 58th Street, Chicago,
nals has been studied in only a few systems. One of the IL 60637. E-mail: [email protected]
better-studied co-transmitting systems is the nerve terminals Received 9 March 1998; Revised 2 September 1998; Accepted 4 Septem-
of presynaptic C fibers in the bullfrog sympathetic ganglia. ber 1998

r 1999 WILEY-LISS, INC.

BOUTON MORPHOLOGY RELATED TO DCV RELEASE 379

zones before exocytosis (Heuser et al., 1974). Similarly, total content of these Ca2⫹ stores at nerve terminals. In
exocytosis is the mechanism for release of the contents of order to answer the above questions, we investigated the
dense-cored vesicles (DCVs) in endocrine cells (Chow et al., total content of DCVs, smooth endoplasmic reticulum
1992; Smith and Betz, 1996; Steyer et al., 1997), in fragments (SERFs), and mitochondria in serially recon-
neurohypophysial neurons (Pow and Morris 1989), in structed boutons. We answered the first question by mea-
nerve terminals of the mammalian sympathetic nervous suring the spatial relationships between SERFs and active
system (Fillenz, 1971; Thureson-Klein et al., 1979; Thure- zones and between mitochondria and active zones. To
son-Klein and Stjarne, 1981), and in nerve terminals of answer the second question, we measured the spatial
frog neuromuscular junction (Pecot-Dechavassine and relationships between SERFs closely associated with the
Brouard, 1997). plasma membrane and DCVs as well as between these
However, neither the spatial distribution of DCVs nor SERFs and mitochondria.
the site for DCV exocytosis has been established. DCVs in We found that all these organelles were typically located
presynaptic terminals in the frog sympathetic ganglia away from active zones with average distances of 526 nm
appear to be located at some distance from the active zone (mitochondria), 624 nm (SERFs), and 696 nm (DCVs).
(Taxi, 1967). Incidences of exocytotic figures of DCVs were There were virtually no DCVs, SERFs, or mitochondria
also observed in some nerve terminals to occur at morpho- within 50 nm of any active zone. Therefore, it is extremely
logically non-specialized zones of the plasma membrane unlikely that DCV exocytosis occurs at these zones. Modu-
(Fillenz, 1971; Buma and Nieuwenhuys, 1987, 1988; Mor- lations of intraterminal [Ca2⫹ ] produced by either SER or
ris and Pow, 1988; Pecot-Dechavassine and Brouard, 1997). mitochondria are equally unlikely to affect release of ACh
However, Somogyi et al. (1984) showed that some DCVs from active zones in response to a single action potential.
are anchored and exocytosed at the edges of active zones in On the other hand, about 30%, 40%, and 60% of DCVs,
mammalian central nervous system. Clearly, instances of SERFs, and mitochondria, respectively, were located within
occurrence as observed by the work cited above are not 50 nm of other parts of the plasma membrane. The long
sufficient to answer the question of release sites for DCVs. delay for LHRH release (ⱖ150 ms, Peng and Horn, 1991)
Instead, systematic measurements of the spatial distribu- suggests that the points of entry for the initial Ca2⫹ influx
tion of DCVs in relation to active zones and to other parts must be at some distance from the DCVs that are within
of the plasma membrane are required. In this study, we ‘‘docking’’ distances of the plasma membrane. This sug-
answered this question by systematically measuring the gests that DCV exocytosis is very likely to occur at the
spatial relationships between DCVs and the active zones plasma membrane. Moreover, the SER closest to the
and between DCVs and other parts of the plasma mem- plasma membrane was also intimately associated with
brane in serially reconstructed terminal boutons of presyn- dense-cored vesicles. Thus, this SER is likely to provide
aptic C fibers in the bullfrog sympathetic ganglia. These the Ca2⫹ necessary for the release of neuropeptide trans-
nerve terminals release a fast-acting neurotransmitter, mitter.
ACh, which is contained in synaptic vesicles. They also
release a neuropeptide, luteinizing-hormone–releasing hor-
mone (LHRH), which is assumed to be contained in DCVs. MATERIALS AND METHODS
LHRH has been found to be in DCVs in mammalian
central nervous system (Phillips et al., 1982). Tissue
Release of fast-acting neurotransmitters and release of All procedures involving animals were performed accord-
neuropeptides requires Ca2⫹ influx into terminal boutons. ing to protocols approved by the IACUC of the University
Recently, we have shown that Ca2⫹ released from a ryano- of Chicago. Bullfrog sympathetic ganglia were fixed in situ
dine-sensitive intraterminal Ca2⫹ store, presumably the with a trialdehyde fixative (3% paraformaldehyde, 2%
smooth endoplasmic reticulum (SER), accounted for 46% glutaraldehyde, and 1% acrolein; DeGroot et al., 1987)
of the peak Ca2⫹ elevation evoked by nerve firing in made in frog Ringer’s solution. Besides the aldehydes, the
bullfrog preganglionic C fiber nerve terminals. In fact, final fixative solution contained the following (in mM):
without this intraterminal Ca2⫹ release, [Ca2⫹ ] was never NaCl 115, KCl 2, CaCl2 1.8, and HEPES 2 at pH 7.25. The
elevated beyond the threshold level for LHRH release in salt concentrations and the pH were the same as in normal
half of the terminals (Peng, 1996). Moreover, mitochon- frog Ringer’s solution. Three adult bullfrogs (Rana catesbi-
drial Ca2⫹ buffering limited Ca2⫹ elevation in these termi- ana), measuring 18 cm from nose to rump, were anesthe-
nals in a frequency-dependent manner (Peng, 1997, 1998). tized in a solution of 0.4% MS-222. A slightly biased
Two questions arise from these Ca2⫹ measurements in midline abdominal incision was made, with attention paid
intact nerve terminals. First, it is believed that for fast- to minimizing bleeding and trauma. Internal organs were
acting neurotransmitter release, Ca2⫹ influx occurs in gently pushed aside, and the 9th and 10th lumbar sympa-
response to membrane depolarization elicited by action thetic ganglion were located. The peritoneal lining overly-
potentials through voltage-gated Ca2⫹ channels concen- ing this ganglion was carefully cut away. A syringe contain-
trated at active zones (Robitaille et al., 1990; Roberts et al., ing the fixative solution was placed very close to these
1990; Issa and Hudspeth, 1994). Can SER and mitochon- exposed sympathetic ganglia and the fixative was applied
drial modulation of intraterminal Ca2⫹ affect release of through the syringe. Typically 30 cc of fixative was applied
ACh from bullfrog sympathetic terminals? Second, for over 5 minutes, and then the ganglion was dissected out
neuropeptide release, the source(s) for Ca2⫹ influx are less and placed in fixative. After the postfixation period (typi-
clear. Can SER and mitochondrial modulation of intrater- cally 20 minutes), the connective tissue encasing the
minal Ca2⫹ affect release of LHRH from these terminals? ganglia was removed, and the ganglia were then placed in
Despite the important roles played by the SER and frog Ringer’s solution. After rinsing overnight, ganglia
mitochondria for intraterminal Ca2⫹ dynamics in response were dehydrated through a series of alcohols and then
to nerve firing, there are no reported measurements of the embedded in Araldite.
380 A. LYSAKOWSKI ET AL.

Ganglia 9 and 10 were typically elongated. Thus, longitu-

dinal sections of them were used because each section in
this orientation contained more complete profiles of neu-
rons, i.e., both the nuclear and the axon hillock regions.
Semithin sections in the dorsoventral axis of the ganglia
were cut and examined. This process was stopped when a
single section was found to contain many neurons of
differing sizes. Then, the actual starting point for the
series of ultrathin sections was chosen randomly after this
point. Ribbons of 70-nm-thick silver serial sections (300
sections or more) were cut with a diamond knife (Delaware
Diamond Knives, Wilmington, CT), collected on Formvar-
coated single slot grids, and stained with lead citrate
(Reynolds, 1963) and uranyl acetate.
Grids were examined in a JEOL 100CX electron micro-
scope. First, for a given section, the x- and y-axes of the
largest neuron were determined. These were the B neu-
rons. Then, C neurons were identified by using the crite-
rion that their axes were less than one-third of those for B
neurons in the same section (Dodd and Horn, 1983).
Usually, a given section contained only a few C neuron
profiles that had the nuclear and axon hillock regions, and
we randomly chose one of them to study. Each C neuron
has 30 to 50 boutons synapsing on it (Jan and Jan, 1982;
Peng and Zucker, 1993). We randomly selected a few
boutons from the nuclear region and a similar number of
boutons from the axon hillock region to be serially recon-
structed. Serial photomicrographs of identified terminal
boutons were taken and printed at a final magnification of
⫻82,500. Low-power micrographs (⫻6,250) were taken to
determine the position of each bouton on a cell, either at
the axon hillock end or the opposite (nuclear) end.
Data collection
Several cytoplasmic elements were measured including
the plasma membrane, the active zone, DCVs, mitochon-
dria, and SER. Bouton diameters were measured by
measuring both the minor (x) and major (y) diameters, and
volumes were calculated. All elements were traced from
electron micrographs or from overlays (Fig. 1A,B), by
using a Drawing Slate II tablet connected to a Macintosh
PowerPC 8100/100 computer. The software program Object-
Image 1.58d was used to collect data, and Microsoft Excel
and IGOR were used in data analysis. The plasma mem-
brane was defined as the bouton membrane exclusive of
the active zone(s). Active zones, although a part of the
plasma membrane, were measured separately. SERFs
were identified as clear elongated, tubular organelles.
These structures could only be SERFs because other
single-membrane-bound organelles such as lysosomes, per-
oxisomes, or Golgi bodies do not occur in nerve terminals

Fig. 1. Methods used for data analysis. A,B: An example of the

overlay and the photomicrograph used to prepare it. Comparable
structures are indicated in the photomicrograph (A) and the overlay
(B). SV, synaptic vesicles; DCV, dense-cored vesicle; MT, mitochon-
drion; AZ, active zone; PM, plasma membrane. C: The method of
measuring distances within a bouton in three dimensions. In this
example, the shortest distance between a DCV and a mitochondrion is
冑
measured by using the Pythagorean theorem, C ⫽ A2 ⫹ B2, where C
is the shortest distance between the DCV and the mitochondrion, A
equals the section thickness (70 nm) times number of sections, and B is
the distance between the DCV and the projected location of the
mitochondrion (*) if it were located in the same section as the DCV.
Sections were placed in register by means of fiduciary marks (f). Scale
bar ⫽ 0.25 µm in A.
BOUTON MORPHOLOGY RELATED TO DCV RELEASE 381

(Peters et al., 1991). Most elements measured exceeded the (Hayat, 1981). Because the vehicle used for the fixative
70-nm thickness of the sections and so were present in solution had an osmolarity exactly the same as bullfrog
more than one section. Elements, such as smooth endoplas- Ringer’s solution, shrinkage of our tissue would be mini-
mic reticulum, were followed through all sections in which mal. Empirically, the ranges (0.5–4.4 µm, Table 1) and
they were present in order to prevent double counting. typical sizes (1 or 2 µm in diameter, Table 1) of these
Items were considered to be repeated if they occurred in boutons measured from their electron micrographs were
the same location (according to landmarks and fiduciary similar to those observed in living tissue when these
points) in adjacent sections. boutons were filled with fura-2 for optical recordings (Peng
Diameter measurements were taken in only one section, and Zucker, 1993). For the latter study, these boutons
but all instances of the element were used for mea- ranged from 0.5 to 4.6 µm with typical diameters of 1 or
surements of areas and volumes. Areas were obtained 2 µm.
from the software program by using the digitizing tablet. Twenty-one boutons taken from five cells in one ganglion
If an item, such as an active zone, did not appear in were either completely or nearly completely reconstructed.
two or more adjacent sections, but then appeared in the The latter lacked only one or two sections from either end
next section, the second instance was considered to be a of the terminals. Ten of the boutons were taken from the
second active zone. Closest linear distances were mea- axon hillock region (fully reconstructed), and 11 were
sured as the closest distance in either the x-, y-, or z-axis, taken from the region of the cell opposite the axon hillock
taking into account that boutons are three-dimensional (seven were fully reconstructed; another four lacked only
objects. one or two sections from either end). The number of
For each organelle, i.e., dense-cored vesicle, smooth reconstructed boutons per cell ranged from one to seven.
endoplasmic reticulum, and mitochondria, the distances to Boutons were spindle-shaped. For a given bouton, the
the nearest active zone and plasma membrane were minor (x) and major (y) axes, measured on the largest
measured. The distances between the organelles and the section, averaged 0.9 and 2.3 µm, respectively (Table 1).
active zones were measured in two different ways. For They had an average length (z-axis) of 2.1 µm. Confirming
those organelles that were in the sections where there previous reports of multiple active zones per bouton (Coo-
were active zones, the distances were measured directly. per et al., 1996), most of these boutons also had multiple
For the organelles that were in sections without any active active zones. Nineteen boutons had at least one and as
zones, the measurements were made by using the Pyth- many as seven active zones with an average of 2.9 ⫾ 0.4
agorean theorem (Fig. 1C). The distances between the (mean ⫾ SE, N ⫽ 19) active zones per bouton (Table 2).
organelles and the other parts of the plasma membrane Boutons with portions of two active zones in the same
were directly measured on each section. section are shown in Figure 2A,B. All boutons had dense-
For those SERFs within 50 nm of the plasma membrane, cored vesicles (DCVs), smooth endoplasmic reticulum frag-
the distances between them and the nearest DCVs and ments (SERFs), synaptic vesicles, and mitochondria (Fig.
mitochondria were measured by using the method for 2), although not all elements were present in all sections
measurements between the organelles and the active through a bouton. Two of the 21 boutons, one located at,
zones. and the other away from, the axon hillock region, did not
have any active zones. They are among the smaller
boutons, with total volumes of 0.5 and 0.7 µm3, respec-
RESULTS tively, but they have similar densities of DCVs, SERFs,
Hemorrhage has been shown to produce high-frequency and mitochondria compared to the other terminals. Re-
firing of some sympathetic nerves (Togashi et al., 1990; sults from these two are not reported.
Rea et al., 1991; Koyama et al., 1992). This firing pattern Mean numbers of intraterminal organelles, i.e., DCVs,
has been shown to be the optimal stimulus for release of SERFs, and mitochondria, are given in Table 2. Thirty-
neuropeptide in sympathetic nervous systems of different nine to 672 DCVs were found in a single bouton. Smooth
species including bullfrogs (Lundberg et al., 1989; Peng endoplasmic reticulum fragments ranged from four to 92
and Horn, 1991). Because of this phenomenon, we wanted per bouton, whereas mitochondria ranged from one to 21
to avoid the possibility of an arbitrary and excessive loss of per bouton. Larger boutons had higher numbers of DCVs,
the synaptic vesicles and particularly the DCVs at these SERFs, and mitochondria such that the densities in a
terminals. Precautions were taken to minimize bleeding. single bouton were comparable for all boutons (Fig. 3).
As the normal transcardiac perfusion typically used in Specifically, the number of each of the elements was
tissue fixation for electron microscopy would produce a linearly related to the volume of the bouton. The correla-
great loss of blood, we directly superfused the ganglia in tion between the number of all three elements and the
situ. A midline incision (as small as possible) was made bouton volume was significant at the P ⬍ .01 level (Fig. 3).
slightly to one side so as to avoid damaging any visible (i.e., The weaker correlation between the SERFs and bouton
under 200⫻ magnification) blood vessel. This procedure volume is not surprising because the fragments were not
resulted in very little bleeding. The importance of the uniform in size. In the case of the DCVs, which are uniform
precautions can be appreciated by comparing our boutons in size (i.e., about 100 nm), we found the strongest
(Fig. 2) with those in the literature (Belhumeur and correlation between the number of elements and the
Tremblay 1986; Spacek and Harris, 1997). Our boutons bouton volume. The linear regression lines suggest that
were always very densely packed with hundreds of clear there were 127 DCVs, 3.3 mitochondria, and 8.7 SERFs for
vesicles and other cytoplasmic elements and very little every µm3 of bouton volume. As an example, Figure 2B
empty space. This morphology is not likely to be the result shows a section from a typical bouton (Cell 3L, bouton 3)
of significant tissue shrinkage. The reason for this is that contains a total of 155 DCVs, six mitochondria, and 26
twofold. Theoretically, the effective osmolarity of a fixative SERFs. Data from boutons located at two different regions
solution is that of the vehicle because it is the ions in the of the postsynaptic neurons (axon hillock and its opposite)
vehicle that pass the plasma membrane of the tissue were plotted separately in Figure 3A. There was no
Figure 2
BOUTON MORPHOLOGY RELATED TO DCV RELEASE 383

apparent difference between them. Because there was also

no systematic difference for any of the other measure-
ments reported below, results from all boutons were com-
bined.

Distances between DCVs and active zones

and between DCVs and other parts
of the plasma membrane
In order to find the likely site(s) for LHRH release at
these terminals, we measured the shortest distances be-
tween every DCV and the active zone and between every
DCV and the portion of the plasma membrane that was
nearest to it (Figs. 1, 4).

TABLE 1. Dimensions of 19 Boutons

x-axis y-axis z-axis
(nm)1 (nm)1 (nm)2
Average 903 2,280 2,045
Standard error (SE) 73 162 187
Minimum 563 1,465 1,120
Maximum 1,568 4,357 3,990
1The diameters of the x- and y-axes were the short and long axes, respectively, as

measured from the largest section for each bouton.

2The diameter of the z-axis was calculated by multiplying the number of sections for a

bouton by the thickness (70 nm) of each section.

TABLE 2. Quantities of DCVs, SERFs, and Mitochondria for 19 Boutons1

Number of elements
per bouton
Total
N Mean ⫾ SE Minimum Maximum
Active zones 55 2.9 ⫾ 0.4 1 7
DCVs 4,057 213.5 ⫾ 42.5 39 672
SERFs 502 26.4 ⫾ 5.4 4 92
Mitochondria 147 7.7 ⫾ 1.2 1 21
1DCVs, dense-cored vesicles; SERFs, smooth endoplasmic reticulum fragments.

Fig. 2. Examples of the synaptic bouton elements analyzed in this

study. A: Two separate active zones (AZ1 and AZ2) can be observed in
this photomicrograph and were confirmed in serial sections. When
totally reconstructed, this bouton contained 207 dense-cored vesicles
(DCVs), 26 smooth endoplasmic reticulum fragments (SERFs), and 13
mitochondria (MTs). Although this particular section of the bouton
contains mostly synaptic vesicles, two DCVs, three MTs, and a few
glycogen bodies are also visible. The rest of the bouton is filled with
regular, round synaptic vesicles (SV). The plasma membrane is
delimited by small arrows. B: This bouton section contains several
MTs and DCVs. In total, the bouton contained 155 DCVs, five SERFs,
and six MTs. One active zone (AZ1) and the beginning of another (AZ2)
are seen. Note the proximity of some of the DCVs and the MTs to the
plasma membrane. Glycogen bodies (GLY) are also present. C: This
bouton section contains two SERFs near the plasma membrane. Four
MTs and six DCVs can also be observed. The entire bouton contained
82 DCVs, 92 SERFs, and nine MTs. There are also several clumps of
GLY. D: This bouton section illustrates one MT and several DCVs,
from a total of 672 DCVs, 41 SERFs, and 17 MTs in the bouton. One
DCV, indicated by the thick arrow, is located 100 nm away from the
active zone. Scale bars ⫽ 0.25 µm in A–D.

Fig. 3. Relationships between the numbers of intracellular ele-

ments and the volumes of individual boutons. A: DCVs. B: Mitochon-
dria. C: SERFs. The data were fit with linear functions as indicated by
the heavy lines. The goodness of fit was calculated using a Pearson
coefficient of correlation as indicated in parentheses. The key indicates
the location of the bouton on the cell, either at the axon hillock (A) end
Figure 3
or the opposite nuclear (N) end. The one solid circle in C is an outlier
and was not included in the fit.
384 A. LYSAKOWSKI ET AL.

For distances relative to the active zone, only the

elements (DCVs, SERFs, or mitochondria) that were within
2.3 µm of an active zone were analyzed. This distance, 2.3
µm, was the same as the average diameter of the y-axis
and was more than the average diameter of the z-axis. It
was longer than the length of the x-axis in all of the
boutons (Table 1). Therefore, elements that had distances
greater than 2.3 µm from the nearest active zone were
located closer to the plasma membrane and opposite the
active zone along all axes for most boutons. For the few
exceptionally large boutons (see the maximum values in
Table 1), this distance still places them beyond the center
of these boutons along their longest axis. In short, 2.3 µm
is too large a distance for an element to be specifically
related to an active zone.
All elements were within 0.9 µm of some part of the
plasma membrane. The distances between the elements
and the plasma membrane were grouped into concentric
rings (annular regions) of 100 nm width. The area within
each ring decreases as its distance from the plasma
membrane increases. The area of the innermost region
was assumed to be an oval with its short and long radii,
100 nm and 253 nm, respectively. The ratio between these
radii was calculated from the average x- and y-axis, i.e.,
903/2,280. Assuming the area of the innermost region to be
1, rings with increasing distances from the center have
normalized areas equivalent to 1 ⫹ 2x, where x ⫽ 0, 1, 2,
3. . ., and x ⫽ 0 for the center region. In order to compare
the density of elements within different rings, the recorded
occurrences within any given ring were normalized by the
area of that ring. The total number of elements was not
altered. This procedure effectively normalized the fre-
quency of occurrences in different rings against the vol-
umes of the corresponding cylindrical shells because the
section thickness is the same (70 nm) for all the rings. The
normalized densities of elements are shown in the C
panels of Figures 4–6.
The distances between DCVs and the active zones
nearest them had an apparent unimodal distribution with
a mean of 696 nm as indicated by the heavy vertical bar in
Figure 4A. Moreover, 45% (1,822 of 4,057) of all DCVs were
more than 900 nm away from any active zone. Only 2% of
the DCVs (88 of 4,057) were found within 100 nm of an
active zone. One example of a DCV located 100 nm from an
active zone is seen in Figure 2D. DCVs were rarely (0.27%,
11 of 4,057) located within 20 nm of the nearest active zone
(see inset, Fig. 4A). Moreover, because these 11 DCVs were
found in only eight of the 19 boutons, the majority of the
boutons had no DCVs within 20 nm of any of their active
zones.
The sharp contrasts between the spatial distributions of
DCVs with respect to the active zones, and with respect to

Fig. 4. Spatial distributions of dense-cored vesicles (DCVs) within

boutons. Distributions of the distances between the DCVs and (A) a
nearest active zone (AZ) and (B) other parts of the plasma membrane
(PM). C: Normalized density distribution of the DCVs within 100-nm-
width concentric cylindrical shells of the plasma membrane. See text
for the normalization procedure. D: Proportions of DCVs that were
within 100 nm of plasma membrane in each bouton. In this and the
following two figures, the insets in A and B are number of elements
within 100 nm of a nearest active zone and other parts of the plasma
membrane, respectively. The linear function that fits the data in D is
indicated by the fitting curve, and the goodness of fit was calculated
using a Pearson coefficient of correlation as indicated in parentheses.
BOUTON MORPHOLOGY RELATED TO DCV RELEASE 385

other parts of the plasma membrane, can be appreciated

by comparing A and B in Figure 4. In fact, 59% of the DCVs
(2,386 of 4,057) were located within 100 nm of some point
on the plasma membrane (Figs. 2A–D, 4B). Moreover, this
region also had the highest density of DCVs (Fig. 4C). The
distribution of DCV density is particularly meaningful,
because if the spatial distribution of DCVs was random,
then the density of DCVs for each 100 nm bin would be
similar. Instead, the highly skewed distribution of DCV
density within 100 nm of the non-specialized regions of the
plasma membrane indicates a specific close association of
the DCVs with these parts of the plasma membrane.
Moreover, not only did the grouped data show that the
majority of the DCVs was located within 100 nm of the
plasma membrane, but also a similar proportion (54%) of
the DCVs was located within 100 nm of the plasma
membrane for each bouton (see curve-fitting equation in
Fig. 4D). In fact, the mean distance between DCVs and the
plasma membrane was 113.4 ⫾ 1.8 nm. No DCVs were
found in the intraterminal space more than 900 nm away
from the plasma membrane. Not only were the majority of
the DCVs located within 100 nm of some part of the
plasma membrane, 12.5% (507 of 4,057) of all DCVs were
located within 20 nm of the plasma membrane (see inset,
Fig. 4B).
The long distances between the DCVs and the active
zones, together with the preferred proximity of DCVs to
other parts of the plasma membrane, suggest that in these
terminals exocytosis of DCVs is likely to occur at parts of
the plasma membrane other than the active zones.
Distances between SERFs and active zones
and between SERFs and other parts
of the plasma membrane
Previous work has shown that repetitive nerve firing at
20 Hz induced Ca2⫹ release from a ryanodine-sensitive
store, presumably the SER, in these terminals (Peng,
1996). We investigated the spatial relationships between
SERFs and the active zones as well as between SERFs and
other parts of the plasma membrane. The results are
shown in Figure 5 and summarized in Tables 3 and 4.
The mean distance between SERFs and the nearest
active zones was 624 ⫾ 28.4 nm (heavy bar in Fig. 5A).
Furthermore, 48% (239 of 502) of the SERFs were more
than 700 nm away from any active zone. Only 3% of the
SERFs (15 of 502) were found within 100 nm of an active
zone. Not a single SERF was found within this distance of
an active zone in any bouton (see inset, Fig. 5A). In 19
boutons studied, only a total of four SERFs were found
within 60 nm of an active zone.
Readily releasable synaptic vesicles are considered to be
those located within distances less than 50 nm to the
active zone. This distance, i.e., 50 nm, was calculated from
the ⬍200-µs delay between Ca2⫹ influx and postsynaptic
responses (Llinas et al., 1981). Given this distance, Ca2⫹

Fig. 5. Spatial distributions of smooth endoplasmic reticulum

fragments (SERFs) within boutons. Distributions of the distances
between the SERFs and (A) a nearest active zone and (B) other parts
of the plasma membrane. C: Normalized density distribution of the
SERFs within 100-nm-width concentric cylindrical shells of the plasma
membrane. D: Proportions of SERFs that were within 100 nm of
plasma membrane in each bouton. Figure conventions are described in
Figure 4.
386 A. LYSAKOWSKI ET AL.

TABLE 3. The Number of Elements Within 100 nm of the Nearest Active

Zone in 19 Boutons1
50 to ⬍100 20 to ⬍50 ⬍20
(nm) (nm) (nm)
(mean ⫾ SE) (mean ⫾ SE) (mean ⫾ SE)
DCVs 2.8 ⫾ 0.6 1.1 ⫾ 0.3 0.5 ⫾ 0.2
SERFs 0.4 ⫾ 0.2 0.2 ⫾ 0.1 0.0 ⫾ 0.0
Mitochondria 0.2 ⫾ 0.1 0.4 ⫾ 0.2 0.5 ⫾ 0.3
1DCVs, dense-cored vesicles; SERFs, smooth endoplasmic reticulum fragments.

TABLE 4. The Number of Elements Within 100 nm of the Plasma

Membrane in 19 Boutons1
50 to ⬍100 20 to ⬍50 ⬍20
(nm) (nm) (nm)
(mean ⫾ SE) (mean ⫾ SE) (mean ⫾ SE)
DCVs 49.1 ⫾ 9.8 40.1 ⫾ 8.0 35.4 ⫾ 6.9
SERFs 5.9 ⫾ 1.7 4.9 ⫾ 1.4 4.4 ⫾ 1.2
Mitochondria 1.4 ⫾ 0.2 1.5 ⫾ 0.3 2.3 ⫾ 0.4
1DCVs, dense-cored vesicles; SERFs, smooth endoplasmic reticulum fragments.

release from SER is unlikely to have an effect on the Ca2⫹

transients occuring within this micro-domain for most
terminals. Thus, it is unlikely that SER could potentially
affect the release of synaptic vesicles in response to a
single action potential.
In sharp contrast to their spatial relationship with
the active zones (compare Fig. 5B with Fig. 5A), 63% of
the SERFs (316 of 502) were located within 100 nm of
some point on the plasma membrane, and the mean
distance was 103 ⫾ 4.6 nm (Fig. 5B). Examples of SERFs
that were within 100 nm of the plasma membrane, but
were located at much larger distances away from an active
zone, are shown in Figure 2C. In fact, no SERFs were
found in the intraterminal space that was 700 nm or more
away from the plasma membrane. Among the 316 SERFs
that were located within 100 nm of the plasma membrane,
71 were within 20 nm of the plasma membrane (see inset,
Fig. 5B). This corresponds to 14% of all SERFs. The region
within 100 nm of the plasma membrane also had the
highest density of SERFs (Fig. 5C). As with the DCVs, the
highly skewed distribution of the density of SERFs within
100 nm of the non-specialized regions of the plasma
membrane indicates a specific close association of the
SERFs with these parts of the plasma membrane. Again,
as in the situation for the DCVs, this was not only the case
for the grouped data; a similar proportion (69%) of SERFs
was found within 100 nm of the plasma membrane for each
individual bouton (see curve-fitting function in Fig. 5D).
The close association of many SERFs with the non-
specialized parts of the plasma membrane permits Ca2⫹
release from SERFs to be relevant to exocytosis occurring
at these regions.
Distances between mitochondria and active
zones and between mitochondria and other
parts of the plasma membrane
Besides the smooth endoplasmic reticulum, the mitochon-
drion is another organelle that could be involved in
intraterminal Ca2⫹ dynamics during the neuronal activi-
Fig. 6. Spatial distributions of mitochondria (MT) within boutons. ties of presynaptic terminals. Mitochondria can have two
Distributions of the distances between the MTs and (A) a nearest major effects on intraterminal Ca2⫹ dynamics: 1) net Ca2⫹
active zone and (B) other parts of the plasma membrane. C: Normal-
uptake when Ca2⫹ levels are high and 2) a slow Ca2⫹
ized density distribution of the MTs within 100-nm-width concentric
cylindrical shells of the plasma membrane. D: Proportions of MTs that release when the intraterminal Ca2⫹ concentration falls
were within 100 nm of plasma membrane in each bouton. Figure below the mitochondrial set-point. Indeed, at these termi-
conventions are described in Figure 4. nals, mitochondrial Ca2⫹ buffering has been found to
BOUTON MORPHOLOGY RELATED TO DCV RELEASE 387

strongly inhibit nerve firing–evoked Ca2⫹ elevations (Peng, TABLE 5. Distances Between DCVs and the Mitochondria to the SERFs
That Were Within 50 nm of the Plasma Membrane in 19 Boutons1
1997, 1998) and LHRH release (Cao and Peng, unpub-
lished data). We therefore studied the spatial distributions SERFs within
of the mitochondria in relation to active zones and to other Distances between 50 nm of PM
organelles1 N (mean ⫾ SE)
parts of the plasma membrane. Unlike the small sizes of
the DCVs and the SERFs, some mitochondria were elon- dSERFs⫺PM 181 26 ⫾ 1
dDCV⫺SERF 181 123 ⫾ 8
gated such that for a given section, their long axis could be ------------------------------------------------------------------------------------------------------------------------
a significant fraction of the diameter of the bouton. We dMitochondria⫺SERF⬎210nm 135 ⬎210
dDCV⫺SERF 135 107 ⫾ 9
measured only the nearest distances between any mito- ------------------------------------------------------------------------------------------------------------------------
chondrion to an active zone and to other parts of the dMitochondria⫺SERFⱕ210nm 46 85 ⫾ 9
dDCV⫺SERF 46 140 ⫾ 15
plasma membrane. The reason for making these measure-
ments is because exocytosis of both synaptic vesicles and 1DCVs, dense-cored vesicles; SERFs, smooth endoplasmic reticulum fragments. Dis-

tances between organelles A and B are denoted as dA⫺B .

DCVs occurs at some part of the plasma membrane. Thus,
mitochondrial Ca2⫹ buffering would affect exocytosis only
if it modulates the local [Ca2⫹ ] in the vicinity of the plasma
membrane. were aggregated toward the plasma membrane (with
The results of these measurements are shown in Figures mean distances of about 100 nm). These results suggest
2 and 6, and also in Tables 3 and 4. On average, the that either there are mechanism(s) at or near the active
distance between some part of the mitochondria and the zones that exclude these elements, or there exist mecha-
nearest active zones was 526 ⫾ 35.4 nm (n ⫽ 120) (Fig. 6A). nism(s) that attract them to other parts of the plasma
About half (55%, 81 of 147) of the mitochondria were more membrane, or there is a combination of both factors.
than 500 nm away from the active zones. Also, 7.5% of the
mitochondria had at least a part of them that (11 of 147) Spatial relationships between DCVs, SERFs,
was located within 100 nm of an active zone (inset of Fig. and mitochondria
6A). These 11 mitochondria were distributed in eight of the The similar patterns of intraterminal distributions of
19 boutons. In contrast to the distribution of the SERFs, SERFs and DCVs could conceivably make Ca2⫹ released
2.8% (4 vs. 41) of mitochondria had at least a part of them from the SER available for DCV exocytosis. On the other
that was located within 20 nm to an active zone (see inset, hand, the similar distribution patterns of mitochondria
Fig. 6A). The four mitochondria that were located 5, 12, 14, and DCVs make mitochondrial Ca2⫹ uptake a possible
and 19 nm away from an active zone were distributed in limiting mechanism for DCV exocytosis. Regulation of
four different terminals. Similar to the situation for SERs, intraterminal Ca2⫹ transients that are relevant to DCVs
only a total of nine mitochondria were found within 60 nm exocytosis by the SER and mitochondria depends not only
of active zones (inset of Fig. 6A), therefore, for most on the rates of the release and uptake processes but also on
terminals, mitochondria are unlikely to have an effect on the spatial relationships between these two Ca2⫹ stores
Ca2⫹ transients occurring within this micro-domain such and the DCVs. We therefore measured the distances
that they could potentially affect the release of synaptic between DCVs and mitochondria with respect to the
vesicles in response to a single-action potential. On the SERFs in each bouton. Only the SERFs that were within
other hand, phenomena that are regulated by Ca2⫹ accumu- 50 nm of some part of the plasma membrane were used.
lation during repetitive nerve firing would potentially be The reason for this selection was twofold. First, this is the
affected by the Ca2⫹-buffering mechanisms of the mitochon- shortest distance to the plasma membrane within which
dria (Tang and Zucker, 1997). every bouton had at least one SERF. The numbers ranged
Note, 7.5% of the mitochondria had at least some part from 1 to 36 per bouton. Second, as exocytosis of the DCVs
located within 100 nm of an active zone, whereas only 2% is likely to occur at the plasma membrane, Ca2⫹ release
of DCVs and 3% of SERFs were within this distance of an within a short distance of the plasma membrane is likely
active zone. One implication of this result is that, for some to be relevant.
terminals, Ca ions that enter the terminal through voltage- We analyzed the spatial distribution of both the DCVs
gated Ca2⫹ channels at the active zones, and remain as and the mitochondria in relation to these SERFs (Table 5).
free ions within 100 nm of these zones, are likely to The SERFs had an average distance of 26 nm (n ⫽ 181
encounter mitochondria before binding to Ca2⫹ sensors on SERFs) from the plasma membrane. The DCVs nearest
either the SERs or the DCVs. these SERFs were located at a distance of 123 nm from
In sharp contrast to their spatial relationship with the these SERFs (n ⫽ 181 DCVs). The majority (135 of
active zones (compare Fig. 6B with Fig. 6A) and similar to 181 ⫽ 74.6%) of these SERFs did not have a mitochon-
the relationships that DCVs and SERFs had to the plasma drion within 210 nm of them. The distance between the
membrane, 74% (109 of 147) of the mitochondria had at 135 SERFs and their nearest DCVs was 107 nm. Thus,
least some part located within 100 nm of the plasma 27% of all the SERFs were within 50 nm of the plasma
membrane with a mean distance of 81 nm (Fig. 6B). More membrane and had a DCV within 107 nm, but had no
strikingly, 28% (41 of 147) of all mitochondria had at least mitochondria within 210 nm. It is possible for Ca2⫹ re-
some part within 20 nm of the plasma membrane (Fig. 2, leased from this group of SERFs to promote the exocytosis
see also inset, Fig. 6B). The region within 100 nm of the of DCVs, whereas mitochondrial Ca2⫹ regulation is not
plasma membrane also had the highest density of mitochon- likely to be directly involved in the exocytosis.
dria (Fig. 6C). Moreover, similar proportions (80%) of the On the other hand, the rest of the SERFs that were
mitochondria were within this region (see curve-fitting within 50 nm of the plasma membrane had a mitochon-
function in Fig. 6D) for 17 of the 19 boutons. drion located at an average distance of 85 nm (Table 5).
Despite some quantitative differences, all the organelles Interestingly, for these SERFs, the nearest DCV was
were located far away from the active zones. Moreover, at located further away from the SERFs than the nearest
least DCVs and SERFs, and possibly mitochondria as well, mitochondrion, with an average distance of 140 nm. Exocy-
388 A. LYSAKOWSKI ET AL.

tosis of these DCVs will thus be affected more strongly by

the function of the mitochondria than that of the SER.
Synaptic boutons were rich
in glycogen bodies
Blocking mitochondrial function with a mitochondrial
uncoupler such as carbonyl cyanide m-chlorophenylhydra-
zone (10 mM) for less than 23 minutes did not appear to
affect the ATP/ADP · Pi ratio at most of these terminals
(Peng, 1997). An alternative pathway for ATP production
is cytosolic glycolysis with glycogen as its substrate. In
bullfrog sympathetic terminals, glycogen molecules con-
dense to form glycogen bodies that appear as spherical
electron-dense particles approximately 20 nm in diameter
(Fig. 2A–D). They form different-sized clusters which are
often quite large (not shown). Two ratios, between the
volumes of the glycogen bodies and the entire bouton, were
estimated for four of the 19 boutons. The area of glycogen
bodies on a given section was first measured, and its Fig. 7. The total volume of mitochondria and two different esti-
volume was calculated assuming either a thickness of 70 mates of the total volume of the glycogen bodies are plotted against the
total volume of the same bouton. Data located on the same dotted
or 20 nm. Seventy nanometers was the thickness of a vertical line were from a single bouton. The thicknesses used in
section, which is about the thickness of three glycogen calculating total glycogen body volume are indicated in parentheses in
bodies, whereas 20 nm was the diameter of a single the key.
glycogen body. The former thickness assumed that the
clusters of glycogen bodies found in a section formed a
cylinder through the thickness of a section, whereas the
latter measurement assumed that these clusters had a ally no DCVs within 50 nm of any active zone in 19
thickness of only one glycogen body. The true volume of the completely reconstructed boutons, it is extremely unlikely
glycogen bodies could not be measured because neither the for DCV exocytosis to occur at active zones via mecha-
thickness of any cluster nor its shape in the z-dimension nism(s) similar to that for the exocytosis of synaptic
could be obtained, but the estimate made by assuming a vesicles. Because exocytotic figures of DCVs have been
20-nm thickness of the glycogen body clusters is the lower found at non-specialized regions of the plasma membrane
bound of their true volume. Given these limitations, of nerve terminals (Morris and Pow, 1988; Pecot-Dechavas-
glycogen bodies occupied 6–18% of the bouton volume if sine and Brouard, 1997), it is very likely that exocytosis of
the thickness were assumed to be 70 nm, whereas this DCVs in the nerve terminals of our study also occurs at
range became 1.7–5% if 20 nm were used as the thickness. these regions of the plasma membrane. Exocytosis of
Even the lower estimates constitute a significant fraction DCVs in endocrine cells is thought to be via mechanisms
of the volumes occupied by mitochondria in the same similar to that of synaptic vesicle exocytosis, in which
boutons (Fig. 7). The large content of glycogen in these ‘‘docking’’ of the DCVs at non-specialized regions of the
terminals makes it feasible that glycolysis maintains ATP plasma membrane is believed to be a prerequisite for their
production during periods when its more efficient produc- exocytosis (Henkel and Almers, 1996). The shortest delays
tion by ATP synthase on the mitochondrial membrane is for exocytosis of the DCVs in these cells are 1–40 ms.
reduced. There was no correlation between the volume of a Intriguingly, at the nerve terminals in our study, not only
bouton and either estimate of glycogen body volume (Fig. were 30% of the DCVs within 50 nm of the non-specialized
7). This might be a reflection of the amount of error in regions of the plasma membrane, 12.5% of all the DCVs
these estimations. were indeed within 20 nm. In spite of this, the shortest
delay for the onset of LHRH-induced synaptic current
recorded from the postsynaptic neurons was greater than
DISCUSSION 150 ms (Peng and Horn, 1991). Thus, either exocytosis of
The precautions taken in this study for tissue fixation the DCVs in these nerve terminals does not require
show how densely packed nerve terminals can be. One ‘‘docking’’ as thought to be necessary in all the other
significance of this result is that the ‘‘free’’ intraterminal systems studied, or alternatively, ‘‘docking’’ is required but
space is extremely limited. Thus, for a given amount of the docked DCVs are located far from the site(s) where the
important cytosolic factors such as second messengers, action potential-triggered initial Ca2⫹ influx occurs. There
including Ca ions, their intraterminal concentrations is no evidence to support the first possibility. On the other
should be much higher than would be expected when there hand, the second possibility is consistent with the finding
was a much larger apparent empty space. This fact, i.e., that in half of the terminals Ca2⫹ release through the
that the free intraterminal space was an insignificant ryanodine receptors on the SER accounted for all of the
fraction of the total volume of a bouton, renders the volume [Ca2⫹ ] elevation required for LHRH release (Peng, 1996).
ratios between any of the elements and the entire volume Because release through the ryanodine receptors occurs
of the bouton quite meaningless for attempts to estimate only after repetitive firing of action potentials at 20 Hz
concentrations of cytosolic factors. (e.g., release from ryanodine-sensitive stores started after
Synaptic vesicles that are located within 50 nm of the 500 ms of firing at 20 Hz, see Fig. 5 in Peng, 1996), and 150
active zones are believed to be releasable upon firing of an ms is the time necessary to fire only four action potentials
action potential (see RESULTS). Because there were virtu- at 20 Hz, the long delay of LHRH release can be explained
BOUTON MORPHOLOGY RELATED TO DCV RELEASE 389

despite the high proportion of DCVs located within ‘‘dock- LITERATURE CITED
ing’’ distance of their likely release sites.
Belhumeur C, Tremblay JP. 1986. Fast turnover of clear vesicle membrane
It is obvious that the results on SER fragments for any in frog sympathetic ganglion. Neuroscience 17:1267–1274.
given statistical analysis can not be interpreted as rigor- Buma P, Nieuwenhuys R. 1987. Ultrastructural demonstration of oxytocin
ously as the results for DCVs, because the latter consist of and vasopressin release sites in the neural lobe and median eminence of
a functional unit with a uniform size whereas the former the rat by tannic acid and immunogold methods. Neurosci Lett 74:151–
have units that vary in size. Moreover, when considering 157.
the effect of the SER on Ca2⫹ dynamics, not only the size Buma P, Nieuwenhuys R. 1988. Ultrastructural characterization of exocy-
totic release sites in different layers of the median eminence of the rat.
but also the densities of Ca2⫹ release channels (such as Cell Tissue Res 252:107–114.
ryanodine and IP3 channels) and Ca2⫹ pumps are impor- Chow RH, von Ruden L, Neher E. 1992. Delay in vesticle fusion revealed by
tant factors. The densities of these channels and pumps electrochemical monitoring of single secretory events in adrenal chro-
are likely to vary and are completely unknown for intact maffin cells. Nature 356:60–63.
nerve terminals. This makes quantitative predictions about Cooper RL, Winslow JL, Govind CK, Atwood HL. 1996. Synaptic structural
complexity as a factor enhancing probability of calcium-mediated
SER function impossible.
transmitter release. J Neurophysiol 75:2541–2566.
A previous study (Vaughn and Grieshaber, 1972) showed
DeGroot JCMJ, Veldman JE, Huizing EH. 1987. An improved fixation
that the glycogen content of boutons is much greater in method for guinea pig cochlear tissues. Acta Otolaryngol 104:234–242.
neonatal rats than in adults, whereas mitochondria occupy Dodd J, Horn JP. 1983. A reclassification of B and C neurones in the ninth
a proportionally greater volume of boutons as the animals and tenth paravertebral sympathetic ganglia of the bullfrog. J Physiol
mature. These authors suggest that more glycogen is (Lond) 334:255–269.
present during a period when animals can tolerate longer Fillenz M. 1971. Fine structure of noradrenaline storage vesicles in nerve
terminals of the rat vas deferens. Philos Trans R Soc Lond [Biol]
periods of anoxia and that there is a gradual, progressive 261:319–323.
addition of aerobic pathways to anaerobic pathways. It is Fogelson AL, Zucker RS. 1985. Presynaptic calcium diffusion from various
conceivable that the high glycogen content of the frog arrays of single channels. Implications for transmitter release and
terminals corresponds to a similar earlier stage in develop- synaptic facilitation. Biophys J 48:1003–1017.
ment in the evolutionary sense. Hayat MA. 1981. Fixation for electron microscopy. New York: Academic
The spatial distributions of the SERFs and mitochon- Press. p 23.
dria lead to two conclusions. First, modulation of intrater- Henkel AW, Almers W. 1996. Fast steps in exocytosis and endocytosis
studied by capacitance measurements in endocrine cell. Curr Opin
minal [Ca2⫹ ] by either SER or mitochondria is not likely to Neurobiol 6: 350–357.
affect exocytosis of synaptic vesicles in response to a single Heuser JE, Reese TS, Landis DMD. 1974. Functional changes in frog
action potential. Second, exocytosis of DCVs can be strongly neuromuscular junctions studied with freeze-fracture. J Neurocytol
affected by these two intraterminal Ca2⫹ stores. Although 3:109–131.
SER and mitochondria have been shown to affect intracel- Issa NP, Hudspeth AJ. 1994. Clustering of Ca2⫹ channels and Ca2⫹-
lular [Ca2⫹ ] in the soma of many cell types, we were the activated K⫹ channels at fluorescently labeled presynaptic active zones
of hair cells. Proc Natl Acad Sci USA 91:7578–7582.
first to demonstrate their participation in cytosolic Ca2⫹
Jan LY, Jan YN. 1982. Peptidergic transmitters in sympathetic ganglia of
transients evoked by nerve firing within intact terminals the frog. J Physiol (Lond) 327:219–246.
(Peng, 1996, 1997, 1998). In fact, that finding was the Koyama S, Sawano F, Matsuda Y, Saeka Y, Shibamoto T, Hayashi TJ,
initial motivation to carry out this morphological study. Matsubayashi Y, Kawamoto M. 1992. Spatial and temporal differing
Many attempts have been made to model the dynamics control of sympathetic activities during hemorrhage. Am J Physiol
of the Ca2⫹ micro-domains near the active zones (e.g., 262:R579–R585.
Fogelson and Zucker, 1985; Sala and Hernandez-Cruz Llinas R, Steinberg IZ, Walton K. 1981. Relationship between presynaptic
calcium current and postsynaptic potential in squid giant synapse.
1990; Roberts et al., 1990). Ca2⫹ micro-domains have been Biophys J 33:323–352.
observed at the terminals of the squid giant axon (Llinas et Llinas R, Sugimori M, Silver RB. 1992. Microdomains of high calcium
al., 1992). Similar micro-domains of Ca2⫹ concentration concentration in a presynaptic terminal. Science 256:677–679.
should exist around SERs and mitochondria. The close Lundberg JM, Rudehill A, Sollevi A, Fried G, Wallin G. 1989. Co-release of
proximity between all these three organelles and the neuropeptide Y and noradrenaline from pig spleen in vitro: importance
of subcellular storage, nerve impulse frequency and pattern, feedback
plasma membrane makes it possible for the micro- regulation and resupply by axonal transport. Neuroscience 28:475–486.
domains of [Ca2⫹ ] around these stores, instead of the Morris JF, Pow DV. 1988. Capturing and quantifying the exocytotic event. J
space-averaged [Ca2⫹ ] of the entire bouton, to regulate Exp Biol 139:81–103.
DCV exocytosis. The distances between the SERFs and the Pecot-Dechavassine M, Brouard M.-O. 1997. Large dense-core vesicles at
DCVs will be informative when mathematical models of the frog neuromuscular junction: characteristics and exocytosis. J
these diffusion-limited micro-domains are made. Neurocytol 26:455–465.
The dimensions of these bullfrog peripheral nerve termi- Peng YY. 1996. Ryanodine-sensitive component of calcium transients
evoked by nerve firing at presynaptic nerve terminals. J Neurosci
nals are similar to those typical of boutons in the mamma- 16:6703–6712.
lian central nervous system (Pickel et al., 1993). Thus, Peng YY. 1997. Effects of mitochondrion on calcium transients at intact
some of the features reported here are likely to be relevant presynaptic terminals depend on frequency of nerve firing. Soc Neurosci
to terminals in mammalian CNS. Abstr 23:9.
Peng YY. 1998. Effects of mitochondrion on calcium transients at intact
presynaptic terminals depend on frequency of nerve firing. J Neuro-
physiol 80:186–195.
ACKNOWLEDGMENT Peng YY, Horn JP. 1991. Continuous repetitive stimuli are more effective
than bursts for evoking LHRH release in bullfrog sympathetic ganglia.
We are grateful to the Electron Microscopic Facility of J Neurosci 11:85–95.
the Research Resources Center of the University of Illinois Peng YY, Zucker RS. 1993. Release of LHRH is lineraly related to the time
at Chicago for providing equipment and assistance to integral of presynaptic Ca2⫹ elevation above a threshold level in
conduct this study. bullfrog sympathetic ganglia. Neuron 10:465–473.
390 A. LYSAKOWSKI ET AL.

Peters A, Palay SL, Webster HdeF. 1991. The fine structure of the nervous central nervous system. Immunochemical characterisation, distribu-
system. Neurons and their supporting cells, 3rd ed. New York: Oxford tion, and relationship to catecholamine and enkephalin pathways.
University Press. Brain Res Rev 8:193–230.
Phillips HS, Ho BT, Linner JG. 1982. Ultrastructural localization of Spacek J, Harris KM. 1997. Three-dimensional organization of smooth
LHRH-immunoreactive synapses in the hamster accessory olfactory endoplasmic reticulum in hippocampal CA1 dendrites and dendritic
bulb. Brain Res 246:193–204. spines of the immature and mature rat. J Neurosci 17:190–203.
Pickel VM, Chan J, Sesack SR. 1993. Cellular substrates for interactions Steyer JA, Horstmann H, Almers W. 1997. Transport, docking and exocyto-
between dynorphin terminals and dopamine dendrites in rat ventral sis of single secretory granules in live chromaffin cells. Nature 388:423–
tegmental area and substantia nigra. Brain Res 602:275–289.
424.
Pow DV, Morris JF. 1989. Dendrites of hypothalamic magnocellular neu-
Tang Y, Zucker RS. 1997. Mitochondrial involvement in post-tetanic
rons release neurohypophysial peptides by exocytosis. Neuroscience
32:435–439. potentiation of synaptic transmission. Neuron 18:483–491.
Rea RF, Hamdan M, Clary MP, Joleen Randels M, Dayton PJ, Strauss RG. Taxi J. 1967. Observations on the ultrastructure of the ganglionic neurons
1991. Comparison of muscle sympathetic responses to hemorrhage and and synapses of the frog Rana esculenta L. In: Hydén H, Editor. The
lower body negative pressure in humans. J Appl Physiol 70:1401–1405. neuron. Amsterdam: Elsevier. p 221–254.
Reynolds ES. 1963. The use of lead citrate at high pH as an electron-opaque Thureson-Klein A, Klein RL, Stjarne LL. 1979. Vesicle populations and
stain in electron microscopy. J Cell Biol 17:208–212. exocytosis in noradrenergic terminals of human veins. In: Ustin E,
Roberts WM, Jacobs RA, Hudspeth AJ. 1990. Colocalization of ions Kopin IJ, Barchas J, editors: Catecholamines: basic and clinical fron-
channels involved in frequency selectivity and synaptic transmission in tiers. New York: Pergamon Press. p 262–264.
presynaptic active zones of hair cells. J Neurosci 10:3664–3684. Thureson-Klein A, Stjarne LL. 1981. Dense-cored vesicles in actively
Robitaille R, Adler EM, Charlton MP. 1990. Strategic location of calcium secreting noradrenergic neurons. In: Stjarne L, Lagercrantz H, Wen-
channels at transmitter release sites of frog neuromuscular synapses. nmalm A. Chemical neurotransmission 75 years. London: Academic
Neuron 5:773–779. Press. p 153–164.
Sala F, Hernandez-Cruz A. 1990. Calcium diffusion modeling in a spherical Togashi H, Yoshioka M, Tochihara M, Matsumoto M, Saito H. 1990.
neuron. Relevance of buffering properties. Biophys J 57:313–324. Differential effects of hemorrhage on adrenal and renal nerve activity in
Smith CB, Betz WJ. 1996. Simultaneous independent measurement of anesthetized rats. Am J Physiol 259:H1134–H1141.
endocytosis and exocytosis. Nature 380:531–534. Vaughn JE, Grieshaber JA. 1972. An electron microscopic investigation of
Somogyi P, Hodgson AJ, DePotter RW, Fischer-Colbrie R, Schober M, glycogen and mitochondria in developing and adult rat spinal motor
Winkler H, Chubb W. 1984. Chromogranin immunoreactivity in the neuropil. J Neurocytol 1:397–412.