The Organization of Synaptic Vesicles at Tonically Transmitting Connections of Locust Visual Interneurons Gerd Leitinger,* Peter J. Simmons Department of Neuroscience, University of Newcastle upon Tyne, The Medical School, Framlington Place, Newcastle Upon Tyne, NE2 4HH, UK
Received 19 February 2001; accepted 16 July 2001
ABSTRACT: Large, second-order neurons of locust ocelli, or L-neurons, make some output connections that transmit small changes in membrane potential and can sustain transmission tonically. The synaptic connections are made from the axons of L-neurons in the lateral ocellar tracts, and are characterized by barshaped presynaptic densities and densely packed clouds of vesicles near to the cell membrane. A cloud of vesicles can extend much of the length of this synaptic zone, and there is no border between the vesicles that are associated with neighboring presynaptic densities. In some axons, presynaptic densities are associated with discrete small clusters of vesicles. Up to 6% of the volume of a length of axon in a synaptic zone can be occupied with a vesicle cloud, packed with 4.5 to 5.5 thousand vesicles
INTRODUCTION Functional connections between neurons consist of a number of distinct anatomical contacts, each of which includes a densely staining structure, the “presynaptic density,” that is associated with the presynaptic membrane and around which vesicles cluster. Although some connections in the vertebrate central nervous system consist of 10 or fewer contacts (e.g., mammalian spinal cord Ia afferents, Burke, 1998; and inhib-
* Present address: Institut fu¨r Histologie und Embryologie, Karl Franzens Universita¨t Graz, Harrachgasse 21, A-8010 Graz, Austria. Contract grant sponsor: BBSRC (UK) Correspondence to: P.J. Simmons (
[email protected]) © 2002 John Wiley & Sons, Inc.
per m3. Presynaptic densities vary in length, from less than 70 nm to 1.5 m, with shorter presynaptic densities being most frequent. The distribution of vesicles around short presynaptic densities was indistinguishable from that around long presynaptic densities, and vesicles were distributed in a similar way right along the length of a presynaptic density. Within the cytoplasm, vesicles are homogeneously distributed within a cloud. We found no differences in the distribution of vesicles in clouds between locusts that had been dark-adapted and locusts that had been light-adapted before fixation. © 2002 John Wiley & Sons, Inc. J Neurobiol 50: 93–105, 2002; DOI 10.1002/neu.10018
Keywords: synapse; ocellus; ultrastructure; neurotransmitter; insect
itory inputs to fish Mauthner neurons, Korn et al., 1981) many connections, particularly in arthropods, include much higher numbers of contacts. In insects, the best characterized connections include that between a photoreceptor and a large monopolar cell in the housefly compound eye, which consists of about 200 regularly spaced presynaptic densities (Nicol and Meinertzhagen, 1982), and output connections between motor neurons and muscle fibers in fly larvae, where individual boutons include up to 40 presynaptic densities (e.g., Atwood et al., 1993; Meinertzhagen et al., 1998). In the central nervous system, the tangled nature of the neuropile makes it more difficult to trace contacts between pairs of identified neurons. In the locust, where contacts between a few pairs of identified neurons have been traced, functional connections comprise several tens to thousands of discrete ana93
94
Leitinger and Simmons
tomical contacts (Peters et al., 1985; Burrows et al., 1989; Simmons and Littlewood, 1989; Littlewood and Simmons, 1992; Killman et al., 1999). A presynaptic density in arthropods normally has the shape of an elongated bar that, in the central nervous system, usually lies opposite the profiles of two or more postsynaptic neurons. Vesicles are thought to be discharged within an active zone either side of the presynaptic density. Evidence for this comes from freeze-fracture studies (Rheuben and Reese, 1978; Saint Marie and Carlson, 1982); and, in Drosophila neuromuscular junctions, elevating calcium concentration increases the number of vesicles trapped beneath presynaptic densities (Koenig et al., 1993; but see Koenig et al., 1998). The best evidence that membrane-associated vesicles include those that are readily available for release comes from a strong correspondence between physiological measurements of the numbers of quanta of transmitter released and ultrastructural measurements of numbers of membrane-associated vesicles in vertebrate retinal bipolar cells (von Gersdorff and Matthews, 1999) and hippocampal neurons (Schikorski and Stevens, 1997; Murthy et al., 1997). The precise role played by the presynaptic density in regulating vesicle discharge is not known. It may act as an anchor point to which vesicles are tethered and, in support of this, a cluster of vesicles remains attached to a presynaptic density of fly photoreceptors when it moves away from the cell membrane after cold shock (Brandsta¨ tter and Meinertzhagen, 1995). It may also be a center around which calcium channels gather, because freeze-fracture studies show membrane particles arranged alongside presynaptic densities (e.g., Heuser et al., 1979; Fro¨ hlich, 1985). Within a presynaptic terminal, vesicles are stored in distinct functional pools that differ in the extent to which they are available for release. This has been shown in experiments in which the processes by which vesicles are replenished are blocked, for example, by using antibodies to block enzymic pathways (lamprey, Pieribone et al., 1995; squid, Daly et al., 2000), or by using temperature-sensitive mutants of Drosophila (Kuromi and Kidokoro, 1998; Koenig and Ikeda, 1999). At Drosophila neuromuscular junctions, the most readily available vesicles include some that are recycled quite rapidly following release, their membrane probably being retrieved from the site of exocytosis (Koenig and Ikeda, 1999). Other vesicles are recycled more slowly, by formation of clathrin coats near the edge of an active zone (Heuser and Reese, 1973), and these vesicles form a reserve pool that is situated in the central parts of a synaptic bouton (Kuromi and Kidokoro, 1999).
There is considerable diversity in the ways that different synapses operate, and sometimes ultrastructural and physiological specialisations can be correlated. In vertebrates, photoreceptors, retinal bipolar cells and hair cells have a ribbon-shaped presynaptic density, and it has been suggested that this is associated with the way that these cells can sustain a tonic level of transmitter release, which is regulated by small, graded changes in membrane potential rather than by large spikes (von Gersdorff and Matthews, 1999). However, among invertebrates no particular ultrastructural specialization that is associated with an ability to release neurotransmitter tonically has been described. In the locust thorax, for example, output synapses made by nonspiking local interneurons (Watson and Burrows, 1988) have a similar ultrastructure to those made by spiking interneurons (Watson and Burrows, 1985). A second clear example of a difference in physiology is at neuromuscular junctions in crustacea, where tonic motor neurons can sustain transmission throughout a train of impulses, but phasic motor neurons cannot do this, although they release many more quanta of neurotransmitter in response to the first spike (e.g., Atwood and Wotjowicz, 1986; Msghina et al., 1998). Although there are differences in ultrastructure between the boutons made by the two types of motor neuron (King et al., 1996; Msghina et al., 1998, 1999), physiological properties are more likely to be determined primarily by differences at the molecular level (Msghina et al., 1999). In this article, we describe the way that presynaptic densities and synaptic vesicles are arranged at connections that transmit tonically from large secondorder neurons, or L-neurons, in the ocellar system of the locust. Compared with most other central neurons, including the equivalent neurons of the insect optic lobe, output connections made by L-neurons are particularly amenable to physiological and ultrastructural study. The tonically transmitting connections are made by neurons L1–3 of a lateral ocellus with thirdorder neurons (Simmons, 1981, 1993) and with Lneurons 4 –5 (Simmons, 1982). Synapses of these connections are relatively easy to find in electron micrographs because they are located on the wide axons of L1–3 (Simmons and Littlewood, 1989; Littlewood and Simmons, 1992). Sections through some L-neurons often show extensive clouds of vesicles, probably associated with the ability of their output connections to sustain transmission. L-neurons also make phasically transmitting, inhibitory output connections (Simmons, 1982, 1999) that are located on their terminal arbors rather than on their axons (Littlewood and Simmons, 1992), and the neurotrans-
Vesicle Clouds at Tonic Synapses
mitter at both the tonic and the phasic connections is probably acetylcholine (Leitinger and Simmons, 2000). In this article, we describe the three-dimensional structure of these vesicle clouds, their relationship to presynaptic densities, and the way that vesicles are distributed within the clouds. We found that clouds can be very extensive, containing up to a million individual vesicles, and cover many presynaptic densities, which vary in length. Within the cytoplasm, vesicles are evenly distributed within a cloud, indicating that different functional pools cannot be distinguished by their location or by the way in which vesicles are packed.
MATERIALS AND METHODS We studied the ocellar tracts of adult Schistocerca gregaria taken from our colony, and the results presented in this article come from 11 different preparations. Two hours before fixation, the brain was exposed by removing overlying cuticle and muscle, and was bathed with saline to prevent desiccation. Five of the locusts were kept in darkness for the 2 h before fixative was applied, using dim illumination from a red light emitting diode for less than 4 s while fixative was applied to the brain surface, and then in darkness for a further 30 min. The remaining locusts were placed in bright sunlight for 2 h before fixation and then fixed under bright illumination from a tungsten light source. Ocellar neurons continue to function normally for several hours after removal of the top of the brain capsule (e.g., Wilson, 1978). Brains were fixed with an ice-cold mixture of 2% paraformaldehyde, 2% glutaraldehyde, 0.2% picric acid, and 10% sucrose in 0.1 M phosphate buffer at pH 7.2, a mixture that both penetrates tissue rapidly and preserves ultrastructural detail well. After 30 min of fixation the brain was removed from the head capsule, and was then fixed for a further 3.5 h at room temperature. Following three rinses in phosphate buffer, brains were post fixed in 1% osmium tetroxide for 1 h before rinsing again, dehydrating in acetone and embedding in araldite (TAAB). Semithin sections were cut transverse to the neuraxis of the brain, starting where the ocellar nerves join the brain. The sections were stained with toluidine blue and examined under the light microscope, and cutting proceeded until examination of the sections showed that the ocellar tracts had bent to run posteriorly and parallel to each other. From here to where L1–3 start to form their terminal arbors is a length of ocellar tract, about 250 m long, that is known to be rich in tonically transmitting output synapses (Simmons and Littlewood, 1989; Littlewood and Simmons, 1992), and 70-nm sections cut transversely through the tract were taken from this region. In two preparations, individual axons of L-neurons were traced along the whole length of an ocellar tract by taking semithin sections, each accompanied by a few thin sections, at known intervals. In a further two specimens, consecutive series of ultrathin sections were
95
collected in order to create three-dimensional reconstructions of lengths of L-neuron axons. A chosen axon was viewed and photographed with a Philips CM100 or a Zeis EM901 electron microscope at a magnification of 6600 – 28,000. Sections were collected on copper slot grids covered with a pioloform film, and were then stained with uranyl acetate and lead citrate. Negatives of electron micrographs were scanned and saved as computer files for analysis. Most measurements were made on individual micrographs, and when examining serial sections identified synaptic connections were followed through the series. ImagePC Software (Scion Corporation) was used for measurements on micrographs and statistical analyses were performed using SigmaPlot and SigmaStat Software (SPSS). To make three-dimensional reconstructions, montages of adjacent micrographs from single sections were made using Adobe Photoshop. Next, for each section, the cell membrane and the locations of presynaptic densities were traced using a Wacom Ultrapad graphics tablet. The extents of vesicle clouds in the selected axon were traced using the “magic wand” tool in Adobe Photoshop to select an area of specified gray values. Traces of consecutive sections were then aligned relative to each other using Adobe Photoshop, and the contours of the axon surface and of vesicle clouds were reconstructed using IGL Trace software (John C. Fiala, Boston University). Surface areas and the volumes of the reconstructed axons and vesicle clouds were also determined using IGL Trace software. For a more detailed reconstruction of part of an axon membrane, incorporating presynaptic densities, surfaces were rendered using Nuages Software (Bernard Geiger, INRIA, France).
RESULTS Ultrastructural Features of Output Synapses The ocellar tracts are recognizable in sections through the protocerebrum by their locations and the presence of up to seven L-neuron axons, which can be almost 30 m across. A number of smaller processes, including those of third-order neurons (Simmons and Littlewood, 1989), are clustered towards the inside of the tract. One or more L-neuron axons in each section contain clouds of vesicles, located near the axon membrane close to the smaller axons [Fig. 1(A)–(C)]. Small clusters of vesicles are occasionally found in the same regions as vesicle clouds, but vesicles are not found elsewhere in L-neuron axons. Intracellular staining has shown previously that axons that contain clouds of vesicles belong to L1–3 or to ML-neurons (Simmons and Littlewood, 1989). Each cloud of vesicles is associated with a number of presynaptic densities, shown in a tracing of one L-neuron axon profile in Figure 1(D), and in a medium power electron
96
Leitinger and Simmons
Figure 1 Low-power electron micrographs showing accumulations of vesicle clouds in the axons of L-neurons in lateral ocellar tracts. (A–C) Frontal sections through a whole lateral ocellar tract in different animals. (A, B) Anterior to where the median and lateral ocellar tracts diverge and to the arbors of L1–3. (C) This is in the anterior part of the arbor of L1–3, where their axons are beginning to divide into branches. (A, E) Fixed in the light, whereas (B)–(D) were fixed in darkness. Profiles of some L-neurons are labeled (L-). Arrowheads point at some of the larger vesicle clouds, which are near to the sites of synapses with smaller profiles, situated towards the middle of the tract in (A) and (B). Tracheae are labeled “T.” (D) Tracing of the outline of one L-neuron profile [asterisk in (B)], with clouds of vesicles (gray) and locations of presynaptic density. Part of the profile is drawn at a larger scale on the right. (E) Mediumpower electron micrograph of synaptic zones of two Lneurons, each containing clouds of vesicles (vs). Arrows indicate a presynaptic density in each L-neuron axon.
micrograph in Figure 1(E). The border of each vesicle cloud is well defined, but irregular in shape. Most vesicles are electron lucent, but a few dense-core vesicles are scattered among them. We measured the mean diameters of electron lucent vesicles as 28.5 ⫾ 9.9 nm (N ⫽ 266; measurements from membrane center), slightly less than that reported previously using a different fixation technique (Simmons and Littlewood, 1989; Littlewood and Simmons, 1992). Extensive clouds of vesicles were present in locusts that had been either light-adapted [Fig. 1(A) and (E)] or dark-adapted [Fig. 1(B)–(D)] prior to fixation. In different axons within the same tract, most presynaptic densities lie under clouds of vesicles [Fig. 2(A), upper L-neuron profile], but a few are associated with small clusters of vesicles [Fig. 2(A), lower
L-neuron profile; and Fig. 2(B)]. The base of each presynaptic density is located next to two postsynaptic profiles. Most sections across an ocellar tract cut across the bar-shaped presynaptic density but occasionally sections cut along the length of a presynaptic density [Fig. 3(A)]. In some electron micrographs, filaments that apparently attach vesicles to a nearby presynaptic density are visible, as is a regular array of stained material associated with the synaptic cleft [Fig. 3(B)]. Occasionally sections were cut parallel to the long axis of a presynaptic density, and some of these showed vesicles closed associated with the cell membrane [Fig. 3(A)]. A proportion of transverse sections though presynaptic densities also showed membraneassociated vesicles [Fig. 3(B); and Fig. 3(C), which is part of a series of consecutive sections with 3(D) and (E)]. A further feature that occurs in the neighborhood of presynaptic densities is coated pits in the cell membrane [Fig. 3(E)], which are presumed to be sites where some vesicle membrane is retrieved for recycling (Heuser and Reese, 1973; Simmons and Littlewood, 1989).
Figure 2 Details of presynaptic densities associated with different numbers of vesicles. (A) The upper L-neuron axon (L-) contains a cloud of vesicles (vs) associated with three presynaptic densities (arrows) while the lower L-neuron axon (L-) has one presynaptic density (arrow) and few vesicles. Each presynaptic density is apposed to two different postsynaptic profiles, some of which are labeled (po). (B) Detail of a presynaptic density (arrow) associated with few vesicles (vs) and two postsynaptic profiles (po).
Vesicle Clouds at Tonic Synapses
Figure 3 Details of presynaptic densities and vesicles. (A) Presynaptic density seen in transverse section (arrow) and in longitudinal section (double arrowhead) with two postsynaptic profiles (po). Three membrane-associated vesicles are indicated by arrow heads. (B) Presynaptic density (arrow) and two postsynaptic profiles (po). A membrane-associated vesicle is indicated with an arrowhead, and two dense core vesicles (dc) are within the cloud of lucent vesicles. In the synaptic cleft is a regular array of electron dense material (small arrowheads). (C–E) Consecutive sections from a series cut through one synapse within a cloud of vesicles [vs in (C)]. The two postsynaptic profiles are labeled (po) in (C). The presynaptic density is indicated by arrows. Arrowheads indicated a membrane-associated vesicle in (C) and a coated membrane invagination in (E).
97
terminal arbors of L1–3. In the L-neuron from the light-adapted locust [Fig. 4(B)], sections taken over the first 100 m of the series contained relatively small areas of vesicle cloud, but in the section at 200 m the cloud had an area of 9.4 m2, covering 4.5% of the 215 m2 axon profile. Vesicle clouds, therefore, can be extensive, but their size and location varies considerably between one L-neuron and another. The numbers of presynaptic densities in an L-neuron profile also vary, both on different sections of the same L-neuron and between different L-neurons in the same section through an ocellar tract [Fig. 4(C) and (D)]. To learn more about the way that vesicles are arranged in clouds in the axons of L-neurons, we made three-dimensional reconstructions of lengths of axons in which we had found prominent vesicle clouds. Figure 5(A) shows reconstructions of an axon from a locust that had been light adapted prior to fixation, and Figure 5(B) shows reconstructions of an axon from a locust that had been dark-adapted. Each was made from 24 consecutive 70-nm sections, and so represents a 1.7-m length of axon. The distance for which a cloud extends from the cell membrane varies
Vesicle Clouds and Presynaptic Densities in L-Neuron Axons In a single L-neuron axon, clouds of vesicles can be found in sections taken at intervals along the length of an ocellar tract, indicating that a single cloud can extend for several tens or hundreds of microns. This was shown in one dark-adapted [Fig. 4(A)] and one light-adapted [Fig. 4(B)] locust by examining sections taken at known intervals along the length of an ocellar tract, starting just anterior to the region where the left and right ocellar tracts lie parallel to each other. The L-neuron chosen from the dark-adapted locust had a relatively large axon, with a profile of area 550 –900 m2 on a cross section, and the cloud of vesicles in it covered an area of 5.5– 6.9 m2, or up to 1.2% of the axon profile. Because the L-neuron profile in the first section of this series contained quite a large area of vesicle cloud, the cloud probably started in the anterior part of the tract, and it did not extend as far as the
Figure 4 The extent of vesicle clouds and numbers of presynaptic densities in sections through individual axons at different locations long the ocellar tract. In (A) and (C), the preparation was fixed in darkness; and in (B) and (D) it was fixed in the light. (A,B) The total area of clouds of vesicles in profiles of individual L-neurons in sections taken at different intervals along the ocellar tract. (C,D) Numbers of presynaptic densities in three axons, each indicated by a different symbol, from each tract. Filled circles are the same axons as in (A) and (B). The preparations used for this figure were different from those used in other figures in the article.
98
Leitinger and Simmons
Figure 5 Reconstructions of lengths of two axons showing distributions of vesicle clouds (blue) and cell membrane (yellow). Each reconstruction was made from 24 consecutive serial sections, and represents a length of about 1.7 m. (A) from Figure 1(A), fixed in the light; (B) from Figure 1(B) and (D), fixed in darkness. The axon in (A) was 26 m across. Arrows indicate in four different views the location of the same patch of cell membrane. The top view shows the whole reconstruction, and the second view shows a detail of the right side, rotated slightly. The third view is a higher magnification reconstruction of a patch of the inside membrane, with vesicles indicated in gray. The locations of presynaptic densities beneath vesicle clouds are indicated with pink (260-nm scale bars apply to this view only). The bottom view is the whole reconstruction rotated through 180 degrees relative to the top view. The axon in (B) was 28 m across. The top view shows the whole reconstruction, and the lower shows part of the reconstruction at a higher magnification and from a different viewpoint. Arrows indicate the same patch of cell membrane.
quite considerably. Some regions of a cloud can be more than 0.5 m thick, but patches of cell membrane not covered by vesicles occur within clouds. The detailed reconstruction of a patch of membrane approximately 6 m2 in area [Fig. 5(A)] shows that presynaptic densities tend to be oriented parallel
to the long axis of the axon, and that presynaptic densities of a variety of lengths share the same vesicle cloud. Within a cloud, there is no distinct border between the vesicles associated with neighboring presynaptic densities. Lengths of presynaptic densities were measured along the surfaces of three-dimen-
Vesicle Clouds at Tonic Synapses
Figure 6 Distributions of presynaptic density lengths, measured from reconstructions of parts of L-neurons. (A,C) From data used by Simmons and Littlewood (1989), showing synaptic contacts between an L-neuron and a stained third-order neuron. (B,D) New data, including all output synapses made by a single L-neuron. (A,B) Plots of presynaptic density length against number of densities of each length; (C,D) plots of presynaptic density length against the total length of all presynaptic densities of each length in the reconstruction.
sional reconstructions, both from data used by Simmons and Littlewood (1989) for synapses made with an identified postsynaptic neuron [Fig. 6(A)], and from our own reconstructions [Fig. 6(B)]. They vary considerably, and most presynaptic densities were detected in only one or two sections, but a few are more than 1-m long [Fig. 6(A) and (B)]. The number of presynaptic densities of a particular length tends to decrease as length increases so that, within a single
99
reconstruction, the total length of individual presynaptic densities of a particular length is fairly constant [Figs. 6(C) and (D)]. We found no relationship between the length of an individual presynaptic density and how closely spaced it was from its neighbors. Measurements and calculations of various parameters of each length of axon and its vesicle clouds are given in Table 1. Vesicle clouds occupied 4.2– 6.4% of the volume of the reconstructed lengths of axon. The number of vesicles in 1 m3 was estimated by extrapolating from counts of the numbers of vesicles in rectangular areas on micrographs, which assumes that each vesicle in a 70-nm section was both visible and was counted only in a single section. We calculated that vesicles occupy 5–7% of the volume of a cloud and about 0.3% of the volume of an axon (although it should be borne in mind that vesicle size is affected by fixation techniques). In our reconstructions, we estimated that each 1-m length of an axon contained 40 – 60,000 vesicles, and that the total surface area of vesicles was about twice that of the axon cell membrane. Within a cloud, vesicle density was between 4.5 and 5.5 thousand per m3, and measurements from micrographs used in an earlier study (Simmons and Littlewood, 1989) gave similar values. We found no consistent differences in the distribution of vesicle clouds between light-adapted and darkadapted locusts. This suggests that the processes that replenish synaptic vesicles can cope easily with recycling vesicles when they are released at a steady rate in a dark-adapted animal.
Table 1 Measurements and Calculations of Vesicle Clouds in the Two Reconstructions of Lengths of L-Neuron Axons from 24 Serial Sections in Figure 5 Reconstruction number Depth of reconstruction, m Mean profile perimeter, m Mean area of profile, m2 Axon membrane area, m2 Axon volume, m3 Cloud volume, m3 Cloud volume/axon volume Vesicle diameter, nm (mean ⫾ SD) Single vesicle surface area, nm2 Total vesicle surface area, m Vesicle s.a./axon membrane area Vesicle volume, nm3 % of cloud filled with vesicles Vesicle volume/axon volume Vesicles per m3 Vesicles in reconstruction Vesicles per m length of axon
1 1.68 71 219 88 314 13 4.2% 29.4 ⫾ 10.4 N, 81 2709.9 191.5 218% 13,269 7.0 0.29% 5314.3 77680 43,640
2 1.68 70 223 140 336 22 6.4% 27.8 ⫾ 9.7 N, 114 2427.9 243.1 173% 11,249 5.2 0.33% 4628.6 100116 59,593
100
Leitinger and Simmons
structions all of the membrane-associated vesicles occur to one side of the presynaptic density [crosssection views in Fig. 7(A) and (B)]. Vesicle distribution along the lengths of presynaptic densities was also plotted by counting the number of vesicles within 150 nm of a presynaptic density in each of a series of consecutive sections [Fig. 7(E) and (F)]. The number of vesicles around a presynaptic density varied from section to section, but there was no consistent trend for a particular location along the length of the density to have a greater number of vesicles than elsewhere. The number of vesicles around a presynaptic density did not vary with the length of presynaptic density. To investigate this, we examined series of sections through five short presynaptic densities (extending up to three sections), and series of sections though five long presynaptic densities (which we followed through six or more sections). The number of vesicles per presynaptic density per section was 4.0
Figure 7 Distributions of vesicles along the lengths of presynaptic densities. (A–D) Three-dimensional reconstructions made by measuring the position of each vesicle relative to the base of a presynaptic density on serial electron micrographs and using Sigmaplot to plot vesicle positions. (A) One presynaptic density, viewed from end; (B–D) another presynaptic density viewed from: (B) end; (C) top; (D) side. The presynaptic density is represented as a straight bar. (E,F) Numbers of vesicles around presynaptic densities in serial section through two different lengths of L-neuron axons. These figures were made from presynaptic densities that were each associated with their own discrete cloud of vesicles.
Distribution of Vesicles We examined the distribution of vesicles around single presynaptic densities by taking serial consecutive sections. For some presynaptic densities, including those in Figures 7(A)–(D) (which were surrounded by discrete clusters of vesicles), we constructed threedimensional representations of a presynaptic density and the vesicles around it. Vesicles are distributed right along the length of each presynaptic density, overlapping its ends slightly, but the arrangement of vesicles was not regular. For example, in both recon-
Figure 8 The distribution of vesicles at different distances from the base of the nearest presynaptic density. (A) Presynaptic densities that were each associated with a discrete cluster of vesicles. Three presynaptic densities were followed through a series of seven sections. Plotted on the Y axis are the mean ⫹ SD of the number of vesicles per m2 per presynaptic density per section. (B) Presynaptic densities within an extensive cloud of vesicles. Six presynaptic densities were followed through a series of seven sections; Y axis as (A). (C) Measurements of nearest 10 vesicles to individual vesicles at different distances from presynaptic densities within the same vesicle cloud, with mean and ranges of distances for the 10 vesicles. (A,B) Fom different axons in the same preparation, and (C) from a different preparation; light adapted.
Vesicle Clouds at Tonic Synapses
⫾ 0.7 (mean ⫾ SD) for the short presynaptic densities and 4.2 ⫾ 0.7 for the long presynaptic densities. Where vesicles occur in small clusters around single presynaptic densities, their number per unit area tends to decrease with distance from the density [Fig. 8(A)], a pattern that has been described before for neuromuscular junctions in the locust (Usherwood and Rees, 1972). However, within clouds, vesicles are no more concentrated near to presynaptic densities than elsewhere, and are evenly dispersed throughout the cloud [Fig. 8(B) and (C)]. This was determined using two different techniques to make measurements from micrographs. In the first, the distances of all vesicles in a cloud from the base of the nearest presynaptic density were used. For each section examined, the numbers of vesicles in a series of half annuli of the same thickness but with different radii centered on the base of a presynaptic density, and then the mean numbers of vesicles per m2 were calculated and plotted against distance from the base of a presynaptic density [Fig. 8(B)]. Within 100 m of the base of a presynaptic density, the presynaptic density itself occupies much of the available space, but between 100 and 300 m from the base of a presynaptic density, the number of vesicles per m2 remained constant [at about 150 for the vesicle cloud in Fig. 8(B)]. Numbers of vesicles per m2 fell for distances greater than 300 m from the base of the presynaptic density because the areas sampled often included the edge of the vesicle cloud. This technique allows the dispersion of relatively large numbers of vesicles to be analyzed quickly, but is subject to inaccuracies because half annuli with larger radii have a greater area than half annuli with smaller radii. As a check, more direct measurements of vesicle concentration were made by counting vesicles within rectangles of known area placed randomly throughout a vesicle cloud. This gave similar values of numbers of vesicles per m2 of section, ranging from 180 to 500, and gave no indication that the density of vesicles near to presynaptic densities was different from that found within vesicle clouds. The second technique we used was to examine how neighboring vesicles were arranged around particular vesicles at different distances from the bases of presynaptic densities [Fig. 8(C)]. For each central vesicle, the center-to-center distances of the nearest 10 vesicles were recorded. Throughout most of a cloud, there was no consistent relationship between distance from the presynaptic density base and either mean distance to the 10 nearest vesicles or to the nearest vesicle (the only exception was for vesicles within 50 nm of the presynaptic density, where the mean distance to the nearest 10 vesicles was slightly
101
increased because the presynaptic density instead occupies some of the space that could be occupied by vesicles). We found no difference in the distribution of vesicles within clouds between light- and dark-adapted locusts. This suggests that any changes in vesicle distribution caused by long-term exposure to different light levels would be relatively small. We attempted to map the distribution of membrane-associated vesicles with the aim of discovering whether there is a distinct zone near to the presynaptic density in which vesicles might be tethered available for release. However, the proportion of vesicles that had their border within 30 nm of the cell membrane was very small (2.6% of 2626 vesicles in one cloud followed through 12 serial sections; and 1.9% of 573 vesicles in a second cloud flowed through 47 serial sections). Because of this, and because we could not distinguish vesicles that were docked at release sites from those that merely lay alongside the cell membrane, we were unable to draw any firm conclusions about locations of likely release sites.
DISCUSSION Vesicle Clouds It is very likely that the prominent clouds of vesicles in the axons of some L-neurons are associated with the ability of their output synapses to sustain tonic transmission. Some clouds extend for at least 200 m and, from our estimate that a 1-m length of axon can contain about 50000 vesicles, in excess of a million vesicles are located near to the sites of tonically transmitting output synapses. The density of vesicles within clouds, 4000 – 6000 per m3, is of the same order as that reported elsewhere, but relatively high: for example, boutons of synapses in mammalian hippocampus contain about 3000 vesicles per m3 (270 vesicles per 0.09 m3 bouton; Schikorski and Stevens, 1997); and vesicles can be packed at up to 4000 per m3 in photoreceptors of Drosophila compound eye, but their density is usually much less than this (Meinertzhagen and O’Neil, 1991). The proximity of large numbers of vesicles to an active zone means that it can draw on large reserves to sustain transmission, and tonically transmitting connections from L-neurons show no decrement in postsynaptic potential when the presynaptic neuron is held depolarized indefinitely by up to 10 mV from dark resting potential (Simmons, 1993). At phasically transmitting connections made by L1–3, a large postsynaptic potential is produced by of the order of
102
Leitinger and Simmons
50 quanta each less than 0.05 mV in amplitude (Simmons, 1999). The postsynaptic potential has a rapid rise over about 4 ms, so vesicles are released for a short time at a maximum rate of about 12,500/s or, assuming that these connections are composed of about 50 discrete active zones (Littlewood and Simmons, 1992), about 250 vesicles/s per active zone. This figure is comparable with an estimate of 55 vesicles/s/active zone, which has been made for sustained transmission from fly compound eye photoreceptors (Laughlin et al., 1987, 1998), so a reasonable guess would be that tonically transmitting outputs from an L-neuron release 50 –250 vesicles/s/active zone. In vertebrates, release rates vary quite considerably, from 30 –300 vesicles/s/active zone for terminals of spiking neurons (e.g., Goda and Stevens, 1994; Borges et al., 1995; Schneggenburger and Neher, 2000). Retinal bipolar cells, which transmit small graded potentials, can achieve a peak rate of about 3000 vesicles/s/active zone for the first 10 ms of stimulation, declining after a second to about 40 vesicles /s/active zone (Neves and Lagnado, 1999; von Gersdorff and Matthews, 1999). Although L-neuron axons clearly accumulate large numbers of vesicles near to presynaptic densities, their distribution gives no clues about the existence of different functional pools or of routes by which they are mobilized during release. We found that vesicle clouds are homogeneous structures, with no tendency for vesicles to cluster more closely together near to presynaptic densities than in the region that borders the interior of the axon. A reason for this might be that movement of vesicles between pools is relatively rapid compared with the time taken for the paraformaldehyde and glutaraldehyde fixatives to immobilize vesicles.
Presynaptic Densities and Active Zones Although it is reasonable to conclude that vesicles are likely to be discharged near to presynaptic densities, the precise extent and location of the active zones in which vesicles are discharged are unknown. In retinal bipolar cells of vertebrates, there is evidence that vesicles are discharged from the base of the ribbon (von Gersdorff and Matthews, 1999), and some of our sections that were cut parallel to the long axis of presynaptic densities did show vesicles closely opposed to the cell membrane. However, within clouds, only about 2% of the vesicles were opposed to the cell membrane, and sections cut across presynaptic densities only rarely showed vesicles located at the junction of a presynaptic density with the cell membrane, but this could be the time taken for a vesicle to discharge
is very much less than the time taken for fixative to penetrate into tissues and immobilize cell organelles. In fly eyes, freeze fracture electron micrographs of output synapses made by photoreceptors show membrane particles aligned on either side of a presynaptic density (Fro¨ hlich, 1985), although the nature of these particles and any function they have in the process of vesicle release is unknown. Output connections made by L-neurons have a much less regular structure than neuromuscular junctions and photoreceptor terminals in arthropods, although the way their synapses are arranged may be representative of most neurons in the central nervous system where the more tangled nature of the neuropile makes it difficult to reconstruct the distribution and detail of individual synapses. Outputs from photoreceptors of compound eyes of the flies Musca and Drosophila are arranged in an exceptionally regular manner in which the number (Nicol and Meinertzhagen, 1982; Meinertzhagen, 1993), length (Fro¨ hlich, 1985), and spacing (Meinertzhagen and Hu, 1996) are all regulated within quite narrow limits. In the compound eye, geometry is severely constrained by the requirement to keep the visual image in register as it is distributed from a photoreceptor to a number of different postsynaptic elements in the lamina. Ocelli are not image forming eyes (Wilson, 1978), so Lneurons do not a need such precise geometry. A fairly regular arrangement is also found at neuromuscular junctions (e.g., crab, Florey and Cahill, 1982; crayfish, Wojtowicz et al., 1994; King et al., 1996; fly larvae, Atwood et al., 1993; Jia et al., 1993; Meinertzhagen et al., 1998), but far fewer elements are involved in making connections than in the central nervous system. Clouds of vesicles that extend over regions containing several different presynaptic densities have been reported at synaptic zones of interneurons in locusts (Watson and Burrows, 1983) and crayfish (Sato et al., 1993), but are much less extensive than those in locust L-neurons. Presynaptic densities in L-neuron vesicle clouds vary considerably in length, with most extending through only one or two sections but others having lengths up to 1.5 m. At arthropod neuromuscular junctions most presynaptic densities extend into only one or two serial ultrathin (70 nm) sections (Atwood et al., 1993; Jia et al., 1993; Wojtowicz et al., 1994; King et al., 1996), although Atwood and Tse (1988) reported a pattern similar to the one we find for L-neurons at crayfish neuromuscular junctions. Similarly, presynaptic densities at output synapses made in a locust thoracic ganglion by motor neurons (Watson and Burrows, 1982) and an intersegmental interneuron (Watson and Burrows, 1983) can be
Vesicle Clouds at Tonic Synapses
longer than 1 m, although most are considerably shorter. In the fly lamina, photoreceptor output synapses show remarkably little variation in presynaptic density length (Fro¨ hlich, 1985), but the feedback synapse from one of the large monopolar cells onto photoreceptors is much more variable (Kral and Meinertzhagen, 1989) possibly because this synapse turns over daily (Pyza and Meinertzhagen, 1993). There are a number of possible reasons why presynaptic densities vary in length. First, they might grow and divide continually. The lengths of presynaptic densities in fly photoreceptors increase during development (Fro¨ hlich and Meinertzhagen, 1982, 1983), and presynaptic densities are assembled at one every three minutes during recovery from a cold shock in an adult (Brandsta¨ tter and Meinertzhagen, 1995). Some presynaptic densities in the fly compound eye also respond to changes in physiological activity or to neurotransmitter (Rybak and Meinertzhagen, 1997; Pyza and Meinertzhagen, 1993). In crayfish, intense and prolonged stimulation of neuromuscular junctions leads to the appearance of pairs of joined presynaptic terminals or of presynaptic densities closer together than usual within sections (Wojtowicz et al., 1994), suggesting that they are dividing. If presynaptic densities respond to the degree of synaptic activation by growing and dividing, tonically transmitting connections, such as those made by Lneurons, would have a relatively high rate of presynaptic density growth and division. Alternatively, variation in presynaptic density length may simply be because a presynaptic density develops in locations where a presynaptic profile lies next to the correct postsynaptic profiles. This explanation is consistent with the presence of two different postsynaptic elements at output synapses from L-neurons, which is the usual pattern for synapses in the central nervous systems of arthropods. For output synapses from photoreceptors of the compound eyes of Musca and Drosophila (which have four output elements) it has been proposed that the synaptic sites that survive during development to assemble the organelles required of a functional synapse in the adult are those at which the correct numbers and identities of neuronal elements are represented (Fro¨ hlich and Meinertzhagen, 1983; Meinterzhagen, 1989). Does the variability in presynaptic length have any implications for the ways that the connections function? Our measurements of vesicle distribution do not reveal any evidence for functional specialisation along the length of a presynaptic density, so that long presynaptic densities have more vesicles available for release than shorter ones, and it would be reasonable to expect a long active zone to have a greater probability to release a
103
vesicle than a shorter one. However, size of an active zone is not necessarily a good prediction of its physiological properties. A good example is provided by crayfish neuromuscular junctions, where phasically transmitting connections have presynaptic densities that are slightly longer than tonically transmitting connections, but release a thousand times more quanta of neurotransmitter in response to a single spike (Msghina et al., 1999). In L-neurons, the rate of release of neurotransmitter is controlled by small, graded changes in membrane potential, and the way that the release of individual quanta of neurotransmitter is controlled by such signals is unclear. An important feature of phasically transmitting output connections of L-neurons is that large postsynaptic potentials have no greater scatter in their amplitude than smaller postsynaptic potentials (Simmons, 1999). If an increase in the amount of neurotransmitter released were due to an increase in the probability of release of individual vesicles throughout the whole of a connection, the scatter in postsynaptic potential amplitudes would also, inevitably, increase. It is, therefore, likely that individual active zones differ in the probability with which they release neurotransmitter at particular membrane potentials, and a reasonable working hypothesis would be that the probability of neurotransmitter release at a single active zone is related to its length. We thank Mr M.W. Bendall and Mr H. Smith for technical help; Dr F.C. Rind and the referees for comments on the manuscript; and Prof. M.A. Pabst (University of Graz, Austria) for use of facilities.
REFERENCES Atwood HL, Tse FW. 1988. Changes in binomial parameters of quantal release at crustacean motor axon terminals during presynaptic inhibition. J Physiol 402:177–193. Atwood HL, Govind CK, Wu CF. 1993. Differential ultrastructure of synaptic terminals on ventral longitudinal abdominal muscles in Drosophila larvae. J Neurobiol 24:1008 –1024. Atwood HL, Wojtowicz JM. 1986. Short-term and longterm plasticity and physiological differentiation of crustacean motor synapses. Int Rev Neurobiol 28:275–362. Borges S, Gleason E, Turelli M, Wilson M. 1995. The kinetics of quantal transmitter release from retinal amacrine cells. Proc Natl Acad Sci USA 92:6896 – 6900. Bransta¨ tter JH, Meinertzhagen IA. 1995. The rapid assembly of synaptic sites in photoreceptor terminals of the fly’s optic lobe recovering from cold shock. Proc Natl Acad Sci USA 92:2677–2681. Burke RE. 1998. Spinal cord: ventral horn. In: Shepherd GM, editor. The synaptic organization of the brain. New York: Oxford University Press. p 77–120.
104
Leitinger and Simmons
Burrows M, Watson AHD, Brunn, DE. 1989. Physiological and ultrastructural characterization of a central synaptic connection between identified motor neurones in the locust. Eur J Neurosci 1:111–126. Daly C, Sugimori M, Moreira JE, Ziff EB, Llina´ s R. 2000. Synaptophysin regulates clathrin-independent endocytosis of synaptic vesicles. Proc Natl Acad Sci USA 97: 6120 – 6125. Florey E, Cahill MA. 1982. The innervation pattern of crustacean skeletal muscle—morphometry and ultrastructure of terminals and synapses. Cell Tissue Res 224:527– 541. Fro¨ hlich A. 1985. Freeze-fracture study of an invertebrate multiple-contact synapse—the fly photoreceptor tetrad. J Comp Neurol 241:311–326. Fro¨ hlich A, Meinertzhagen IA. 1982. Synaptogenesis in the 1st optic neuropil of the fly’s visual system. J Neurocytol 11:159 –180. Fro¨ hlich A, Meinertzhagen IA. 1983. Quantitative features of synapse formation in the fly’s visual system. 1. The pre-synaptic photoreceptor terminal. J Neurosci 3:2336 – 2349. Goda Y, Stevens CF. 1994. Two components of transmitter release at a central synapse. Proc Natl Acad Sci USA 91:12942–12946. Heuser JE, Reese TS. 1973. Evidence for recycling of synaptic vesicle membrane during transmitter release at the frog neuromuscular junction. J Cell Biol 57:315–344. Heuser JS, Reese TS, Dennis MJ, Jan Y, Jan L, Evans L. 1979. Synaptic vesicle exocytosis captured by quick freezing and correlated with quantal transmission release. J Cell Biol 81:275–300. Jia XX, Gorczyca M, Budnik V. 1993. Ultrastructure of neuromuscular-junctions in Drosophila— comparison of wild-type and mutants with increased excitability. J Neurobiol 24:1025–1044. Killmann F, Gras H, Schu¨ rmann FW. 1999. Types, numbers and distribution of synapses on the dendritic tree of an identified visual interneuron in the brain of the locust. Cell Tissue Res 296:645– 665. King MJR, Atwood HL, Govind CK. 1996. Structural features of crayfish phasic and tonic neuromuscular terminals. J Comp Neurol 372:618 – 626. Koenig JH, Ikeda K. 1999. Contribution of active zone subpopulation of vesicles to evoked and spontaneous release. J Neurophysiol 81:1495–1505. Koenig JH, Yamaoka K, Ikeda K. 1993. Calcium-induced translocation of synaptic vesicles to the active site. J Neurosci 13:2313–2322. Koenig JH, Yamaoka K, Ikeda K. 1998. Omega images at the active zone may be endocytotic rather than exocytotic: implications for the vesicle hypothesis of transmitter release. Proc Natl Acad Sci USA 95:12677–12682. Korn H, Triller A, Mallet A, Faber DS. 1981. Fluctuating responses at a central synapse—n of binomial fit predicts number of stained presynaptic boutons. Science 213: 898 –901. Kral K, Meinertzhagen IA. 1989. Anatomical plasticity of
synapses in the lamina of the optic lobe of the fly. Philos Trans R Soc Lond B 323:155–183. Kuromi H, Kidokoro Y. 1998. Two distinct pools of synaptic vesicles in single presynaptic boutons in a temperature-sensitive Drosophila mutant, shibire. Neuron 20: 917–925. Kuromi H, Kidokoro Y. 1999. The optically determined size of exo/endo cycling vesicle pool correlates with the quantal content at the neuromuscular junction of Drosophila larvae. J Neurosci 19:1557–1565. Laughlin SB, de Reuter van Steveninck R, Anderson JC. 1998. The metabolic cost of neural information. Nat Neurosci 1:36 – 41. Laughlin SB, Howard J, Blakeslee, B. 1987. Synaptic limitations to contrast coding in the retina of the blowfly Calliphora. Proc R Soc Lond B 231:437– 467. Leitinger G, Simmons PJ. 2000. Cytochemical evidence that acetylcholine is a neurotransmitter of neurons that make excitatory and inhibitory outputs in the locust ocellar visual system. J Comp Neurol 416:345–355. Littlewood PMH, Simmons PJ. 1992. Distribution and structure of identified tonic and phasic synapses between l-neurons in the locust ocellar tract. J Comp Neurol 325:493–513. Meinertzhagen IA. 1989. Fly photoreceptor synapses: their development, evolution and plasticity. J Neurobiol 20: 276 –294. Meinertzhagen IA. 1993. The synaptic populations of the fly’s optic neuropil and their dynamic regulation—parallels with the vertebrate retina. Prog Retin Res 12:13–39. Meinertzhagen IA, Hu X. 1996. Evidence for site selection during synaptogenesis: the surface distribution of synaptic sites in photoreceptor terminals of the flies Musca and Drosophila. Cell Mol Neurobiol 16:677– 698. Meinertzhagen IA, O’Neil SD. 1991. Synaptic organization of columnar elements in the lamina of the wild-type in Drosophila melanogaster. J Comp Neurol 305:232–263. Meinertzhagen IA, Govind CK, Stewart BA, Carter JM, Atwood HL. 1998. Regulated spacing of synapses and presynaptic active zones at larval neuromuscular junctions in different genotypes of the flies Drosophila and Sarcophaga. J Comp Neurol 393:482– 492. Msghina M, Govind CK, Atwood HL. 1998. Synaptic structure and transmitter release in crustacean phasic and tonic motor neurons. J Neurosci 18:1374 –1382. Msghina M, Millar AG, Charlton MP, Govind CK, Atwood HL. 1999. Calcium entry related to active zones and differences in transmitter release at phasic and tonic synapses. J Neurosci 19:8419 – 8434. Murthy VN, Sejnowski TJ, Stevens CF. 1997. Heterogeneous release properties of visualized individual hippocampal synapses. Neuron 18:599 – 612. Neves G, Lagnado L. 1999. The kinetics of exocytosis and endocytosis in the synaptic terminal of goldfish retinal bipolar cells. J Physiol 515:181–202. Nicol D, Meinertzhagen IA. 1982. An analysis of the number and composition of synaptic populations formed by photoreceptors of the fly. J Comp Neurol 207:29 – 44.
Vesicle Clouds at Tonic Synapses Peters BH, Altman, JS Tyrer NM. 1985. Synaptic connections between the hindwing stretch-receptor and flight motor neurons in the locust revealed by double cobalt labeling for electron-microscopy. J Comp Neurol 233: 269 –284. Pieribone VA, Shupliakov O, Brodin L, Hilfikerrothenfluh S, Czernik AJ, Greengard P. 1995. Distinct pools of synaptic vesicles in neurotransmitter release. Nature 375: 493– 497. Pyza E, Meinertzhagen IA. 1993. Daily and circadianrhythms of synaptic frequency in the 1st visual neuropil of the housefly’s (Musca domestica) optic lobe. Proc R Soc Lond B 254:97–105. Rheuben MB, Reese TS. 1978. Three-dimensional structure and membrane specializations of moth excitatory neuromuscular junctions. J Ultrastruct Res 65:95–111. Rybak J, Meinertzhagen IA. 1997. The effects of light reversals on photoreceptor synaptogenesis in the fly Musca domestica. Eur J Neurosci 9:319 –333. Saint Marie RL, Carlson SD. 1982. Synaptic vesicle activity in stimulated and unstimulated photoreceptor axons in the housefly. A freeze fracture study. J Neurocytol 11:747– 761. Sato M, Nagayama T, Hisada M. 1993. Distribution of synapses on 2 types of ascending interneurons in the crayfish, Procambarus clarkii. Cell Tissue Res 271:9 –21. Schikorski T, Stevens CF. 1997. Quantitative ultrastructural analysis of hippocampal excitatory synapses. J Neurosci 17:5858 –5867. Schneggenburger R, Neher E. 2000. Intracellular calcium dependence of transmitter release rates at a fast central synapse. Nature 406:889 – 893. Simmons PJ. 1981. Synaptic transmission between 2ndorder and 3rd-order neurons of a locust ocellus. J Comp Physiol 145:265–276. Simmons PJ. 1982. Transmission mediated with and with-
105
out spikes at connections between large 2nd-order neurons of locust ocelli. J Comp Physiol 147:401– 414. Simmons PJ. 1993. Adaptation and responses to changes in illumination by second- and third-order neurones of locust ocelli. J Comp Physiol A 173:635– 648. Simmons, PJ. 1999. The performance of synapses that convey discrete graded potentials in an insect visual pathway. J Neurosci 19:10584 –10594. Simmons PJ, Littlewood PMH. 1989. Structure of a tonically transmitting synapse between identified interneurone in the locust brain. J Comp Neurol 283:129 –142. Usherwood PNR, Reese D. 1972. Quantitative studies of the spatial distribution of synaptic vesicles within normal and degenerating motor axons of the locust, Schistocerca gregaria. Comp Biochem Physiol 43A:103–118. von Gersdorff, H Matthews G. 1999. Electrophysiology of synaptic vesicle cycling. Annu Rev Physiol 61:725–752. Watson AHD, Burrows M. 1982. The ultrastructure of identified locust motor neurons and their synaptic relationships. J Comp Neurol 205:383–397. Watson AHD, Burrows M. 1983. The morphology, ultrastructure, and distribution of synapses on an intersegmental interneurone of the locust. J Comp Neurol 214:154 – 169. Watson AHD, Burrows M. 1985. The distribution of synapses on the two fields of neurites of spiking local interneurones in the locust. J Comp Neurol 240:219 –232. Watson AHD, Burrows M. 1988. The distribution and morphology of synapses on nonspiking local interneurones in the thoracic nervous system of the locust. J Comp Neurol 272:605– 616. Wilson M. 1978. The functional organisation of locust ocelli. J Comp Physiol 124:297–316. Wojtowicz JM, Marin L, Atwood HL. 1994. Activity-induced changes in synaptic release sites at the crayfish neuromuscular junction. J Neurosci 14:3688 –3703.
