中枢神经系统内神经递质释放研究进展
tion, a process he termed augmentation, and PTP—and elucidated many properties of these processes that affect the Katzian probability, in an history-dependent way, over time scales from a few tens of milliseconds to several minutes; this work was completed when the first issue of Neuron appeared (Magleby, 1987). The dominant mechanism for these processes was believed to be “residual calcium,” that is, a slow return of intraterminal calcium concentration to resting levels following synapse use—a hypothesis based on a number of indirect observations and a single experiment that correlated measured calcium concentrations with synaptic responses (Connor et al., 1986).
It was recognized early that release does not occur at just one specific instant when a nerve impulse arrives, but rather that the vesicle fusion rate (the probability per second that a release will occur) is spread over several hundreds of milliseconds. After a nerve impulse arrival, the probabilistic rate of vesicle fusion increases briefly to a high level and then returns with a double exponential time course (Barrett and Stevens, 1972a, 1972b). The first component is very fast, less than a millisecond at mammalian body temperature, and the second component decays over several hundred milliseconds and was believed to correspond to facilitation and augmentation and to reflect a prolonged increase in intraterminal calcium concentration (residual calcium).
Molecular Biology of Synapses
Only three synaptic vesicle proteins had been cloned by 1988: synapsin I (McCaffery and DeGennaro, 1986), synaptophysin (Buckley et al., 1987; Leube et al., 1987; Sudhof et al., 1987), and VAMP-1 (Trimble et al., 1988) (later also called synaptobrevin). A few other synaptic proteins had been purified—most notable, it would turn out, was a synaptic vesicle protein known then as P65 (Matthew et al., 1981) and later renamed synaptotagmin—but not yet cloned. The understanding of the molecular basis for neurotransmitter release had not advanced very far by that time, because clones for so few of the many players were available and because the actual function for no synaptic protein was known. Nevertheless, more was going on than was apparent from what had been published. Two young molecular biologists, Richard Scheller and Thomas Sudhof (often in collaboration in the early days with Reinhard Jahn, Pietro De Camilli, and Paul Greengard), had both decided that the key to understanding synaptic transmission was to clone all of the proteins associated with synaptic vesicles and all of the proteins associated with the synaptic vesicle proteins. This program, carried out over the next five years, turned out to be central to our current understanding.
Molecular Basis for Membrane Trafficking
A second research stream, the study of membrane trafficking by cell biologists, was also to be of special importance for the advances in the molecular biology of synaptic transmission that were to come. Palade (1975) had long before pointed out that one of the central problems in cell biology was membrane trafficking. He noted that the cell consisted of many membrane bound compartments (the Golgi stack, endoplasmic reticulum, endosomes, vacuoles, and the cell itself defined by its surface membrane) and that material was moved from one compartment to another (and from internal compartments to the cell surface) by the budding off of vesicles from the source compartment and the subsequent fusion of these vesicles with limiting membrane of the destination compartment. Understanding how the vesicle budding and fusion occurred and how the vesicles were directed to the appropriate membrane were thus key for knowing how cells function. Two approaches to this question were being followed in 1988. The first was to isolate the yeast genes involved in membrane trafficking (Rothblatt and Schekman, 1989), discovered in screens for defective trafficking after random mutagenesis, and the second was to reconstitute the basic membrane trafficking in a cell-free system and then use this assay to find out what proteins were required to make it work (Rothman, 1990). What was not fully appreciated, despite a prescient review in the sixth issue of Neuron(Kelly, 1988), was the extent to which the same basic mechanisms responsible for membrane trafficking also underlie synaptic vesicle fusion.
By 1988, many mutants of yeast genes that affected membrane trafficking had been identified, and progress was being made in characterizing these mutants (Rothblatt and Schekman, 1989). Furthermore, the in vitro reconstitution approach had revealed that ATP hydrolysis by a protein complex termed SNAP (soluble NSF attachment protein) was required to maintain vesicle trafficking, and this SNAP was thought to bind to receptors that were directly involved in the fusion reaction (Rothman, 1990).
Synapses in 2003
The Quantal Hypothesis
Katz's basic idea that neurotransmitter is released by the fusion of synaptic vesicles has gained increasing support since 1988, but the formalism Katz used to establish his theory—quantal analysis, which predicts probabilistic variations in the number of vesicles that fuse per nerve impulse—is no longer generally accepted. This fall from favor derives from the continuing questions about what constitutes a release site and from evidence that the Katzian probability p varies across release sites (rather than being constant, as Katz originally assumed).
Several laboratories have shown directly—confirming conclusions reached earlier by less direct experiments (Korn and Faber, 1991; Redman, 1990; Zucker, 1973b)—that individual central synapses, ones thought to have only a single active zone, release only a single vesicle even when the release probability approaches 1 (Dobrunz and Stevens, 1997; Hanse and Gustafsson, 2001; Stevens and Wang, 1995). Of course, a single active zone can release multiple vesicles over time, so the issue of “single release” boils down to a temporal one: over what period is a synapse unable, or very unlikely, to release a second vesicle? After a vesicle release, the probability of an additional release has been found to decrease to essentially zero for about 5 ms and then increase to its normal value over another 5 ms or so (Stevens and Wang, 1995). According to these experiments, then, a Katzian release site is a single active zone and the Katzian probability p corresponds to the release probability, the probability that the active zone will release one of its docked vesicles; most central excitatory synapses have a single active zone, so a Katzian release site is usually a synpase. This probability has been shown to increase approximately linearly (over a specific range of vesicle numbers) with the number of docked vesicles (Dobrunz and Stevens, 1997; Murthy et al., 2001).
These experiments just described all assumed that the postsynaptic membrane was not saturated by a single quantum. Although some early work proposed that a single quantum could saturate the available postsynaptic receptors (Clements, 1996; Edwards et al., 1990), more recent research has established, using more direct methods, that neither AMPA, GABA, nor NMDA receptors are saturated by a single quantum (Frerking et al., 1995; Liu et al., 1999; Mainen et al., 1999; McAllister and Stevens, 2000).
The work summarized above argued that a Katzian release site is a single active zone—that is, that a single active zone can release only one vesicle—but other experiments have, contrary to these conclusions, clearly demonstrated that individual synapses can release multiple vesicles in response to a single arriving nerve impulse (Auger et al., 1998; Oertner et al., 2002; Tong and Jahr, 1994; Wadiche and Jahr, 2001). These results, then, favor the idea that a Katzian release site is a single release-ready vesicle and that Katzian probability p corresponds to the probability that the vesicle will undergo exocytosis with a nerve impulse arrival.
Unfortunately, all of the experiments on both sides of this issue have one flaw or another, so the question “What is a Katzian release site?” is still unanswered, as it was 15 years ago. In the direct experiments (Dobrunz and Stevens, 1997; Hanse and Gustafsson, 2001; Stevens and Wang, 1995), individual synapses were isolated by the method of minimal stimulation (Raastad et al., 1992), and the selection bias of this technique is unknown; the results of these experiments may, then, typify only a very small minority of all synapses in the brain. On the other hand, the experiments demonstrating multiquantal release either could not exclude the possibility of multiple active zones (Auger et al., 1998; Tong and Jahr, 1994; Wadiche and Jahr, 2001) or did not have adequate time resolution to be sure that multiple releases occurred within a 10 ms time window (Oertner et al., 2002) (central synapses, like the neuromuscular junction, exhibit both synchronous and asynchronous release [Goda and Stevens, 1994]).
A further problem for quantal analysis is that, contrary to what the traditional version of the theory assumes, the Katzian probability is not constant across sites (Murthy et al., 1997; Robitaille and Tremblay, 1987; Sakaba and Neher, 2001a, 2001b). This is true whether a release site corresponds to a single active zone (Murthy et al., 1997; Robitaille and Tremblay, 1987) or a single release-ready vesicle (Sakaba and Neher, 2001a, 2001b). Of course, one can extend the Katz formulation of release to assign a distribution of probabilities to the population of release sites, but the form of this distribution is not sufficiently well known to make this approach believable. Certainly, quantal analysis frequently gives the right answer when the Katzian probability is taken as an average value of the single-site probabilities but, because one never knows if the result is right in
- 众说风云 (已有0条评论)
