Search

The immense complexity of the brain requires that it be built and controlled by intrinsic, self-regulating mechanisms. One such mechanism, the formation of new connections via synaptogenesis, plays a central role in neuronal connectivity and, ultimately, performance. Adaptive synaptogenesis networks combine synaptogenesis, associative synaptic modification, and synaptic shedding to construct sparse networks. Here, inspired by neuroscientific observations, novel aspects of brain development are incorporated into adaptive synaptogenesis. The extensions include: (i) multiple layers, (ii) neuron survival and death based on information transmission, and (iii) bigrade growth factor signaling to control the onset of synaptogenesis in succeeding layers and to control neuron survival and death in preceding layers. Also guiding this research is the assumption that brains must achieve a compromise between good performance and low energy expenditures. Simulations of the network model demonstrate the parametric and functional control of both performance and energy expenditures, where performance is measured in terms of information loss and classification errors, and energy expenditures are assumed to be a monotonically increasing function of the number of neurons. Major insights from this study include (a) the key role a neural layer between two other layers has in controlling synaptogenesis and neuron elimination, (b) the performance and energy-savings benefits of delaying the onset of synaptogenesis in a succeeding layer, and (c) how the elimination of neurons in a preceding layer provides energy savings, code compression, and can be accomplished without significantly degrading information transfer or classification performance.

We investigate whether the inactivation properties of Na^+ channels in squid giant axons show evidence of having been optimized by evolution so as to reduce the energy expenditure associated with propagating action potentials. Using Hodgkin-Huxley-type models of the squid giant axon and varying the voltage dependence and overall scaling of the Na^+ inactivation rate, we find that faster inactivation rates correspond to less costly and more energy-efficient action potentials. However, and in contrast to the case when the ionic conductances and axon diameters are varied, the metabolic energy consumption does not have a minimum when restricted to an isovelocity curve in the phase space associated with these inactivation parameters. This failure-to-optimize encourages other hypotheses about the evolution and functionality of inactivation, such as its effects on information rates.

The hyperpolarization-activated mixed cation and the persistent sodium conductances are compared as linearizing mechanisms for somatodendritic synaptic integration in steady-state systems. In the steady-state model used, the persistent sodium conductance creates a well-defined region of linear synaptic excitation, from -66 to -55mV. This corresponds to a moderate level of synaptic depolarization (total active synaptic conductance ranging from 2 to 5.9nS). In contrast, a model enhanced with a hyperpolarization-activated mixed cation conductance only linearizes over a short voltage range, from -68 to -61mV. This in turns corresponds to very low levels of synaptic activity (0-3.7nS). Given in vivo recordings of firing thresholds for nonbursting neurons ranging from around -60 to -55mV, the persistent sodium current emerges as the better linearizing mechanism for these neurons because it operates across the full physiologically relevant voltage range for perisomatic, proximal dendritic synaptic excitation.

Our hippocampal model depends on randomization. In principle, randomizations, e.g., chaotic activity fluctuations, quantal synaptic failures, or initial state randomization, can be overcome by strong external excitation. However, if external activity is too low, randomization will destroy the information transmitted by the inputs. Here, computer simulations of the transitive inference paradigm reveal an optimal range of external excitation. At lower activity levels, optimal performance occurs when the relative external excitation accounts for 35-40% of the total with activity, while at higher activity, external activity can be as low as 30% of the total.

The relative refractory period of an action potential affects any second closely following impulse. Starting with the constraining conjecture that consecutive action potentials produce minimal interference with each other, we investigate the twin-pulse maximum frequency. Using an updated version of the Hodgkin-Huxley squid giant axon, this constrained maximum firing frequency varies with transmembrane conductance density and diameter. Using the twin-pulse constraint and keeping velocity constant, smaller, higher-conductance density axons are generally preferred to larger, lower-conductance density ones.

A minimal model of the CA3 region of the hippocampus produces gamma oscillations. These oscillations exist across a broad range of conditions. For instance, the oscillations exist in the presence or absence of external input. Additionally, quantal synaptic failures do not eliminate the gamma oscillations. More importantly, there is a sensitivity to the average firing rates of neurons in the simulations. Specifically, increasing the average firing rates leads to a continuous phase transition that is a quenching of the gamma oscillations. Although the minimal model includes an inhibitory neuron, no explicit time constant of this neuron or the inhibitory synaptic conductance corresponds to the observed frequencies of oscillation.

A model of hippocampal function, centered on region CA3, reproduces many of the cognitive and behavioral functions ascribed to the hippocampus. Where there is precise stimulus control and detailed quantitative data, this model reproduces the quantitative behavioral results. Underlying the model is a recoding conjecture of hippocampal computational function. The expanded conjecture includes a special role for randomization and, as recoding progresses with experience, the occurrence of sequence learning and sequence compression. These functions support the putative higher-order hippocampal function, i.e. production of representations readable by a linear decoder and suitable for both neocortical storage and forecasting. Simulations confirm the critical importance of randomly driven recoding and the neurocognitive relevance of sequence learning and compression. Two forms of sequence compression exist, on-line and off-line compression: both are conjectured to support neocortical encoding of context and declarative memory as described by Cohen and Eichenbaum (1993).

We investigate the energy efficiency of interspike interval (ISI) neural codes. Using the hypothesis that nature maximizes the energy efficiency of information processing, it is possible to derive neuronal firing frequencies which maximize the information/energy ratio. With simple assumptions about the encoded ISI and noise distributions, we show that ISI codes can be at least as efficient as discrete binary and frequency codes and that their predicted optimal frequencies are in the same range.

It has been hypothesized that increasing CS longevity affects performance on trace conditioning. Using a hippocampal model, we find that increasing CS and US longevity increases the learnable trace interval. Our simulations show that, over a modest range, the maximal learnable trace interval is approximately a linear function of CS/US longevity.

The function of the CA3 region of the hippocampus can be explained in terms of a sequence predicting recoder. For the CA3 to act as a neural prediction device, each CA3 neuron must also act as a predictor. Thus, such neurons, as prediction devices, compute something that might approximate a conditional probability. In particular, we conjecture that each neuron forecasts its own firing. Here we compare a simple neural network model, based on synaptic encoding of local conditional probabilities to an even simpler model of hippocampal region CA3 that succeeds on a variety of hippocampally dependent learning tasks.

Hodgkin and Adrian's 1975 hypothesis that the squid axon is optimized for maximum conduction velocity is flawed by (i) the inaccurate value of its prediction for channel density, and (ii) the prohibitive energetic expense entailed by their prediction. Here we investigate the metabolic cost of conduction velocity. By manipulating ion channel density or by manipulating the Nernst battery voltages, we demonstrate that action potential velocity has a significant metabolic cost. Thus, in addition to the cost of information transmission (Neural Comput. 8(1996) 531 [9]), there is a cost associated with the timely arrival of such transmitted information.

Using a minimal hippocampal model, previous studies simulating trace conditioning have reproduced the empirically observed learnable trace interval and reproduced the number of training trials required for learning. However, these earlier studies did not address the effects of parameterization on performance. Here, we demonstrate a robust effect of average activity on trace conditioning performance.

If there are many statistically independent excitatory inputs to a neuron, its net excitation is binomially distributed, and this distribution is well-approximated by a Gaussian because there are so many inputs. Because there are so many inputs, quantal failures are essentially harmless. However, the presumption of statistical independence is too simplistic. To reflect statistical dependence among the inputs, we consider mixture distributions. Generally, mixture distributions are a distributional class that can be far from Gaussian even though the individual component distributions might, themselves, be Gaussian. Here we show when quantal synaptic failures can move the kurtosis of a mixture distribution towards the Gaussian value.

Energy efficient use of axons requires achieving a prespecified firing rate. We hypothesize that the failure process of quantal synaptic transmission helps a neuron better approximate this desirable firing rate by moving the neuron's input excitation distribution closer to a Gaussian. If there are many statistically independent inputs per neuron, quantal failures do not help but, essentially, are harmless for achieving the desired firing rate. However, such statistical independence is unrealistic. An input distribution reflecting statistical dependence is a mixture distribution. For certain mixtures, failures can improve the Gaussian approximation and more precisely produce the desired firing rate.

There are three easily distinguishable forms of synaptic plasticity: synaptogenesis, modification of existing synapses, and synapse removal. For any particular synaptic class, only a compatible set of plasticity rules makes sense for these three distinguishable forms. Here we point out two such compatible sets, one for each of two classes of excitatory synapses. In both cases, synaptogenesis is local and anti-correlational, while modification of existing synapses encodes mostly positive correlations as adjusted by a particular negative correlation. Connections that offer little or nothing to the local, postsynaptic information processing is an inefficiency and leads to synapse removal.

The hippocampus is necessary in both humans and rats for learning configural representations in tasks such as transverse patterning. The transverse patterning task, (A+B-, B+C-, C+A-), requires representing individual stimuli in the context of other stimuli. This paper extends a previous application to rat data [INNS World Congress on Neural Networks, 1995; Biol Cybern 6 (1998a) 203] by applying a model of the CA3 region of the hippocampus to human data. A decision function is also added that enables the system to choose among training items. Analysis of the simulations show that configural representations are formed by unique neural codes that depend on temporal and stimuli context. Based on the simulations, we hypothesize that configural representations in biological networks depend on a proper balance of input and context representations. Furthermore, the division of labor between functions in the model is a specific working hypothesis that in learning this task the hippocampus specializes in sequence prediction and the decision function evaluates the predictions.

Quantal synaptic failures are random, independent events in which the arrival of an action potential fails to release transmitter. Although quantal failures destroy information, there exist conditions where such failures enhance learning by a neural network model of the hippocampus. In particular, we show how the appropriate failure rate can allow the model to learn the hippocampally dependent task of transverse patterning. Usefully, since lowered activity levels produce higher memory capacity, synaptic failures lead to robust performance at lowered activity levels. Thus, the synaptic failure mechanism is another example of a random fluctuation that improves neural network computations.

The hippocampus is needed to store memories that are reconfigurable. We have previously shown that a hippocampal-like computational model solves the transverse patterning (TP) problem and transitive inference problems. Here, we show that the same model with the same parameters which solve the TP problem and transitive inference problems reproduces another interesting problem—transverse non-patterning (TNP) problem investigated by Alvarado and Rudy (J. Exp. Psychol.: Anim. Behav. Process 18 (1992) 145). By turning TNP into a problem of sequence learning (stimuli–decision–outcome), a sequence learning, hippocampal-like neural network finds that the TNP problem is unlearnable with the progressive learning paradigm. This unlearnability is what Alvarado and Rudy observed. Thus, both rats and the model learn TP, but fail to learn TNP.

Adding synaptic modification to the inhibitory interneuron in a minimal computational model of hippocampal region CA3 improves average performance of the simulations. After training on two partially overlapping sequences, simulations are tested on a sequence completion problem that can only be solved by using context dependent information. Simulations with dynamic autonomously scaling (DAS) inhibition are more robust than those without. In the DAS model, scaling factors for inhibition are adjusted gradually over time to compensate for the original model's tendency to move away from a pre-set activity level. This variable inhibition modifies more slowly than the local, associative synaptic modification of the excitatory synapses. As a result, activity fluctuations from one time-step to the next continue to occur, but average activity levels show small variability across training. These results suggest that restricting long term activity fluctuations can be beneficial to recurrent networks that must learn context dependent sequences.

The hippocampus is needed to store memories that are reconfigurable. Therefore, a hippocampal-like computational model should be able to solve transitive inference (TI) problems. By turning TI into a problem of sequence learning (stimuli-decisions-outcome), a sequence learning, hippocampal-like neural network solves the TI problem. In the transitive inference problem studied here, a network simulation begins by learning six pairwise relationships: A>B, B>C, C>D, D>E, E>F, and F>G where the underlying relationship is the linear string: A>B>C>D>E>F>G. The simulation is then tested with the novel pairs: B?D, C?E, D?F, B?E, C?F, B?F, and A?G. The symbolic distance effect, found in animal and human experiments, is reproduced by the network simulations. That is, the simulations give stronger decodings for B>F than for B>E or C>F and decodings for B>F and C>F are stronger than for B>D, C>E, or D>F.

We investigate how the strength of entorhinal cortical inputs during training a!ects learned performance using computer simulations of a minimal computational model of hippocampal region CA3. After the model learns two partially overlapping sequences, it is tested on two contradictory prediction problems * disambiguation and goal-"nding. Relative to total activity, the activity level of entorhinal inputs during learning profoundly a!ects performance on each task. The optimal input levels di!er for the two sequence prediction problems, but a small region of overlap exists where both tasks can usually be performed successfully. This sensitivity to relative input activity suggests critical tests of the model. ( 2000 Elsevier Science B.V. All rights reserved.

The organization of training is an important determinant of how well subjects learn a cognitive task. To understand why different training schedules produce different learned performance, we used a hippocampal model to compare three training paradigms for the hippocampally dependent cognitive task called transverse patterning. Simulations reproduce training effects seen in humans and rats. As in behavioral studies, progressive training produces robust learning while random training renders the task essentially unlearnable. The simulations predict that a third training paradigm, called staged learning, will produce more robust learning on average than the progressive paradigm used in published behavioral studies. Possible mechanisms underlying performance differences between paradigms are investigated and discussed.

Using computer simulations of a minimal computational model of hippocampal region CA3, this report investigates how randomization during training alters learned performance. The transitive inference problem is employed for this purpose. Randomizing just the initial network state at the beginning of each training trial profoundly affects learning. That is, no randomization makes the problem unlearnable while a moderate amount of randomized activity optimizes network performance. These results suggest a way to alter learning which may be tested in neuropsychological experiments.

We tested the hypothesis that homosynaptic long-term depression (LTD) can be induced at the CA3–CA3 synapses in the adult, in vivo hippocampus while the CA3–CA1 synapses remain unchanged. Low-frequency conditioning stimulation of the contralateral fimbria significantly depressed the CA3 population response but did not change the simultaneously recorded CA3 response to angular bundle test stimulation. Similarly, in another group of animals, low-frequency conditioning stimulation of the contralateral fimbria depressed the CA3 synaptic response and left the collateral CA1 synaptic response unchanged. Among the possible explanations for this differential induction of homosynaptic LTD at the CA3–CA3 and CA3–CA1 synapses are differential control of intracellular calcium, differing levels of inhibition in these two regions, and the recency of `natural' long-term potentiation in the two regions.

Somatic spines on granule cells have been observed occasionally in the adult rat dentate gyrus in vivo. Here we evaluate the appearance and formation of somatic spines in s. granulosum from adult rat hippocampal slices immediately after slice preparation, 45 min, 3 h, and 5 h later. Initially somatic spines are extremely rare but, after 3 h in vitro, they are readily apparent. Some of these somatic spines form asymmetric synapses that have a spherical vesicle-containing presynaptic bouton and a postsynaptic density. Other somatic spines lack a postsynaptic density and may also lack an apposed presynaptic bouton (free somatic spines) as observed in single thin sections. Both the number of these free somatic spines and the number of somatic spine synapses increase in a time-dependent manner. No somatic spines were observed on CA1 pyramidal cells in the same hippocampal slices. Electrophysiological observations indicate that the formation of somatic spine synapses on granule cells in the hippocampal slice occurs without any apparent granule cell activation. The trigger event for this very rapid synaptogenesis in the adult dentate gyrus remains to be determined.

The present study tested the effect of two non-competitive NMDA receptor antagonists, ketamine and phencyclidine, on the induction of long-term depression (LTD) in the dentate gyrus of urethane-anesthetized rats. Both drugs blocked the induction of LTD as well as long-term potentiation (LTP). NMDA receptor activation thus seems to be required for the induction of both LTD and LTP in the dentate gyrus. High-intensity conditioning stimulation did not overcome the phencyclidine block of LTD. Strong, but brief, postsynaptic depolarization is apparently not the only event needed to trigger LTD.

This study sharpens the temporal definition of an association in the context of long-term synaptic modification of the monosynaptic projections from entorhinal cortex to dentate gyrus in the anesthetized rat. The ipsilateral projection produces a powerful postsynaptic excitation and shows synaptic potentiation following 7.5 ms trains. On the other hand, brief, high-frequency contralateral stimulation just after this powerful postsynaptic excitation is followed by long-term depression (LTD) at the synapses of this pathway. Therefore, based on the differential induction of long-term potentiation (LTP) or LTD, a cell can distinguish associated pre- and postsynaptic activation at a temporal resolution at least as small as 7.5 ms.

The present study quantified ribosomes, as an ultrastructural marker of neuronal protein synthesis, following long-term potentiation (LTP) in the hippocampal CA1 region in vitro. Sixty min after LTP-inducing, high-frequency stimulation of the perforant path, the total number of ribosomes, the number of polysomes, and the number of membrane-bound ribosomes increased significantly. These increases are a postsynaptic morphological correlate consistent with enhanced protein synthesis following the induction of LTP in the perforant path-CA1 system.

This study used urethane-anesthetized rats to investigate the longevity of heterosynaptically evoked depression of the monosynaptic response generated by synapses between entorhinal cortical (EC) afferents and the cells of the dentate gyrus (DG). Brief, high-frequency activation of the converging ipsilateral EC-DG input depressed the synaptic response of the contralateral EC-DG synapses without prior experimentally induced potentiation. This depression lasted for hours. Such observations are consistent with a role for heterosynaptically induced long-term depression in the encoding functions of synapses.

The number of granule cell dendritic spines per micrometer of dendritic length in the dorsal and ventral leaves of the dentate gyrus was quantified using light microscopic-Golgi preparations of normal adult rats. Spines were counted in terms of 3 categories of spine form for the 3 afferent termination zones of the molecular layer and corrected for shading errors. Total spine density averaged 1.6 spines/micron of dendritic length in the dorsal leaf and 1.3 spines/micron of dendritic length in the ventral leaf. Statistically significant differences in spine density existed among the 3 shape categories. Variations in spine density occurred by shape category among the afferent termination zones.

The bilateral, monosynaptic projection from the entorhinal cortex to the dentate gyrus provides a suitable system for studying long-term changes of synaptic strength. In particular, the synaptic response of the sparse contralateral projection can be increased or decreased depencling on the exact conditioning protocol. Since changes in either direction require high-frequency stimulation of the potent, converging ipsilateral system, these synaptic alterations are called associative potentiation/depression. In part, associative potentiation/depression is Anatogous to what others call long-term potentiation. Synaptic phenomena like associative potentiation/depression are central to several theories of associative brain function and neuronal models of cognition. In elaborating such theories, the conditions which alter synaptic physiology as well as the size and direction of the changes are expressed as an algebraic equation. This paper is the initial report attempting to quantify one part of one such ‘synaptic learning equation’. This paper quantifies the term in the equation which expresses the role of presynaptic activity. Through this term the equation can express both synaptic potentiation and depression as a function of the afferent frequency of the particular synapse under consideration. In this term there exists a critical frequency or a balance point. When there is sufficient co-activity among converging excitatory afferents, then the synapse under consideration will either potentiate or depress. If the afferent's frequency is above the balance point, potentiation occurs; if the frequency is below the critical value, depression occurs. For a constant amount of postsynaptic excitation, the amount of change is a nonnegative function of the frequency around this balance point. Prior theoretical proposals postulate that this balance point resides near the average frequency of the afferents. In fact, the value obtained, 100 ± 50 Hz for short duration trains, is orders of magnitude greater than the time-averaged frequency previously observed for these afferents. Other considerations indicate that although a balance point exists it is more likely a variable rather than a constant.

The lateral column-ventral root reflex of the frog spinal cord is a neurophysiologically well-characterized model system that demonstrates eight of the nine parameters of habituation. Because knowledge about the temperature dependence of biological processes can yield clues about the underlying biochemical mechanisms, the current paper investigates the effects of temperature on lateral column-ventral root reflex habituation. Lowering the temperature from 15° to 5°C resulted in profound, quantitative alterations in the appearance of the lateral column-ventral root reflex. These changes included a decrease in response amplitude, an increase in response duration, and an increase in response latency. In spite of these changes, it was concluded that the responses at 15° and 5° were generated by the same, monosynaptic lateral column-motoneuronal system. These changes built up gradually with each degree of cooling and were consistent with the known temperature dependencies of various neurophysiological processes. Lowering the temperature from 15° to 5°C also increased the depression induced by conditioning stimulation for pulse trains of two to ten pulses in length. This increase in habituation with cooling was poorly correlated with changes in response amplitude, and it was therefore concluded that the altered habituation did not result from a stimulus-intensity like effect. Furthermore, understanding the mechanism by which temperature was able to alter habituation did not seem possible without resolving the effects of temperature on the depression induced by each successive stimulus from the effects of temperature on the rate of recovery between stimuli.

Summary Previously accumulated GABA was released from isolated forebrain synaptosomes with repeated calcium stimulation in elevated-potassium medium. Fractional release (calcium-dependent) in response to a second calcium pulse (90–120 sec later) was depressed to approximately 60% of initial release. Neither initial GABA release nor the subsequent depression of release was affected by variations in the labelling duration. Stimulation-dependent depression of labelled GABA and norepinephrine release was demonstrated from both cerebral cortex and cerebellum synaptosomal preparations. In addition, depression resulted from prior stimulation in the presence of veratridine, A23187 or elevated-potassium. Although release of previously accumulated GABA was depressed by calcium stimulation, the release of GABA accumulated between stimulations was not. Release of this recently accumulated GABA was indistinguishable from the initial release of previously accumulated GABA and larger than the subsequently depressed release from the previously accumulated pools. These data imply (1) that the depressed release resulted from a decrease of available transmitter in pools that support secretion processes, (2) that depressed release did not result from a depression of stimulus-secretion coupling processes, and (3) that transmitter accumulated subsequent to release events is released preferentially to transmitter accumulated prior to the intervening stimulation.

A previous study utilizing the powerful ipsilateral and weak crossed projection from the entorhinal cortex to the dentate gyrus in the rat revealed that long-term potentiation is an associative process in these systems. If the weak crossed projection alone receives potentiating stimulation consisting of 8 high-frequency trains 17.5 ms in duration, it does not exhibit long-term potentiation. However, long-term potentiation can be induced in the crossed projection if it is activated concurrently with the converging ipsilateral system. The present study is designed to determine the degree of synchrony required for the associative interactions by varying the timing and order of the potentiating trains delivered to the two converging systems. The associative induction of long-term potentiation does not require perfectly synchronous activation of the converging systems. The order of the trains is crucial, however. Long-term potentiation of the crossed projection can be induced if activity in the ipsilateral system is concurrent with or follows activity in the crossed projection. Indeed, there can be as much as 20ms between the 17.5ms trains in the two systems, and long-term potentiation of the crossed projection is still induced. Activation of the ipsilateral system that precedes activation of the crossed system depresses the responses evoked by the crossed system. If potentiating stimulation of the ipsilateral system follows activation of the crossed projection by too long an interval (200 ms, for example), then the crossed projection is depressed rather than potentiated. These results are discussed with regard to the nature of the associative process permissive for the induction of long-term potentiation and lead us to the conclusion that perfect temporal contiguity is not a requirement of this prototypical elemental memory unit.

This report analyzes long term potentiation (LTP) and associative interactions between synapses of the ipsilateral and crossed entorhinal cortical (EC) pathways to the dentate gyrus (DG). In the anesthetized rat, conditioning stimulation to one EC-DG pathway reliably elicits LTP at the ipsilateral synapses, while the synapses of the collateral, crossed pathway to the contralateral DG do not exhibit LTP. Furthermore, in the DG ipsilateral to the conditioning stimulation the convergent crossed pathway from the contralateral side, which had not been itself conditioned, failed to exhibit heterosynaptic LTP. These results are consistent with a specific 'synaptic' localization of the changes responsible for LTP, and suggest that some critical number of synapses must be activated in order to observe LTP. While the crossed EC-DG projection never exhibited LTP when conditioned alone, the crossed input could be potentiated under certain circumstances. Specifically, paired conditioning of ipsi- and contralateral inputs by nearly simultaneous conditioning stimulation of the EC bilaterally results in LTP in the crossed system. Furthermore, this associatively induced LTP of the crossed system can be reversed by subsequent conditioning of the ipsilateral system alone. Successive potentiating and depotentiating sequences are possible using paired and non-paired stimulation procedures even after lesions which prevent neural loops through the EC. The results are interpreted as evidence for a 'Hebb' type synapse which has the capability for erasure. This synaptic type is not appropriate for classical conditioning without appendant circuitry, but is suited for other forms of associative learning.

The relationship between [Ca 2+ ] 0 and secretion was investigated in crude synaptosomal fractions from the cerebral cortex and the corpus striatum of the rat. The sigmoid nature of the relationship was analyzed with a linearizing modification of the Michaelis-Menten equation. Higher [K + ] 0 increased calcium-dependent release of [ 3 H]dopamine from striatal synaptosomes. Log releaselog [Ca 2+ ] 0 curves were shifted upward and to the left. Higher [K + ] 0 also increased the slope, or sigmoidicity, of the relationship. In the presence of 56 mM [K + ] 0 , release of [ 3 H]dopamine was greater than that of [ 14 C]γ-aminobutyrate from striatal synaptosomes at all the [Ca 2+ ] 0 tested. The differences in release resulted simply from an upward shift of the curves, as no differences were observed in either the apparent K m for calcium or the sigmoidicity of the relationship. Release of [ 14 C]γ-aminobutyrate from cortical and striatal synaptosomes was also investigated. KCl-facilitated, calcium-dependent release of γ-aminobutyrate from cortical synaptosomes was greater than release from striatal synaptosomes. In addition to an upward shift of the log release-log [Ca 2+ ] 0 plot, the curve for cortical synaptosomes exhibited less sigmoidicity. No differences in the K m for calcium were observed. The similarity of K m values for calcium of the secretion systems for different transmitters and for different regions suggests that a similar rate-limiting process(es) may exist in the different secretion systems. However, the differences in sigmoidicity and maximal rates of secretion suggest that the systems do differ in at least some aspects.

Triton X-100 treatment of synaptic plasma membrane fractions selectively dissociates membrane constituents. At an ultrastructural level the synaptic complex resists Triton treatment when Ca2+ is present so that the synaptic complex is dissociated from the bulk of the adjoining plasma membrane. The synaptic complex retains its selective affinity for phosphotungstic acid and can therefore be unambiguously identified in the Triton-insoluble residue. At a chemical level Triton results in differential solubilization of membrane protein, phospholipid and sialic acid. Under conditions favorable for preserving synaptic complex, the insoluble residue retains about 60 % of its initial protein, 70 % of its initial phospholipid phosphorus, and 80 % of its initial sialic acid. Over 75 % of the acetylcholinesterase is solubilized, but essentially all of the 5′-nucleotidase and alkaline phosphatase of the synaptic plasma membrane fraction remains insoluble. The implications of these data for synaptic function and the structuring of synaptic plasma membranes are discussed.