Alex M Thomson
Department of Physiology, Royal Free and University College Medical School, London

https://doi.org/10.36866/pn.47.17

In a communication to the Society last year we reported a novel presynaptic frequency filter apparent at the synapses made by cortical pyramidal cells with some classes of postsynaptic targets. The frequencies filtered by this mechanism are those that appear in the cortical EEG during arousal and attention (the gamma frequency band), the oscillation in the population activity being due to the synchronous firing of neurones in phase with the rhythm. Pyramidal cells do not typically fire on every cycle, however. It is the fast spiking interneurones and the connections between them that appear to drive the rhythm and provide the temporal framework for the synchronous, if more sporadic, activity in the pyramidal population (Traub et al, 1998). The synchroneity is proposed to result in temporary cell assemblies that form, disassociate and reform in different configurations with each cycle, representing coherent features of the cortical representations of images or events (Singer 1999, 2001). Indiscriminate, reverberating activation of large populations of interconnected pyramidal cells at gamma frequencies would not only obliterate information coded by these assemblies, it could be dangerous. Mechanisms that prevent such indiscriminate recruitment are therefore likely to be important and this communication focussed on a presynaptic mechanism for which we coined the term ‘notch filter’.

The release of transmitter into the synaptic cleft involves an extremely sophisticated series of interactions between proteins and lipids in the vesicle membrane and in the plasma membrane (Thomson 2000, for review).

Before fusion can occur, a vesicle must become attached to an appropriate release site. Proteins whose interactions are essential for the subsequent maturing or priming process are thereby unleashed from the molecular brakes that retard their energy-consuming interactions when they are not attached to an appropriate site. A fusion core complex consisting of proteins in the two membranes forms, binds an ATP-ase and, possibly via cyclical formation and dissociation of the core complex with the resultant expenditure of ATP, primes the vesicle so that it becomes ready for immediate release. The availability of fully mature, primed vesicles forming this immediately releasible pool of transmitter is therefore an important factor in determining whether a release can occur, especially during continued activity. The small synaptic terminals made by pyramidal cell axons in the cortex contain relatively few vesicles, estimates varying between 40 and 60 and only a small proportion of these are in the immediately releasible pool at any one time. Depletion of this pool, factors that retard the entry of vesicles into this pool, or that withdraw primed vesicles from the pool can therefore have a powerful effect on release. The mechanism that probably contributes most significantly to paired pulse depression, release site refractoriness (Betz, 1970), appears to involve the retardation of other primed vesicles by a factor made available during Ca2+-dependent fusion of a vesicle with the plasma membrane. It is therefore a mechanism activated by the release of a vesicle and does not operate in the absence of a release. Terminals that have released transmitter are refractory, while those that did not release remain available and may be facilitated. A mechanism that results in the release-independent depression of subsequent release at very short interspike intervals has also been identified, but appears to be expressed at only a minority of connections, those made with some subclass(es) of interneurones (Thomson and Bannister, 1999).

Another essential component of the release machinery is the population of presynaptic Ca2+ channels. The activity of these channels appears to be regulated by other proteins at the release site so that Ca2+ influx is greatest close to fully mature vesicles. A brief influx of Ca2+ in a spatially restricted region, or micro-domain, triggers fusion of the vesicular and plasma membranes and the release of transmitter (Llinas, Blinks and Nicholson, 1972). The release may be an ‘all or none’ phenomenon, but it does not occur with a probability of 1, nor with equal probability at all release sites or even on every trial at a single site. Four Ca2+ ions must bind to each vesicular membrane bound Ca2+ sensor protein (synaptotagmin), for fusion to be triggered. Small differences in the number of Ca2+ ions that enter during an AP, or in the affinity of the synaptotagmin binding sites for Ca2+ produce dramatic differences in release probability. This requirement for four Ca2+ ions underlies facilitation (and augmentation). Where a single AP does not trigger sufficient Ca2+ entry for a release to occur, some of the Ca2+ nevertheless binds to synaptotagmin. The requirement for a subsequent Ca2+ entry is reduced, since some sites are already occupied and the probability of release for a second AP within a brief period is increased.

In the simplest model proposed to account for the very different patterns of transmitter release at the terminals of pyramidal axons, those with a high release probability exhibit paired pulse (and brief train) depression because they become refractory for a period after each release, while those that have a low release probability exhibit facilitation because they rarely release in response to a single AP and do not experience refractoriness. This model accounted for the differences observed between the typically ‘depressing’ EPSPs elicited by pyramidal cells in other pyramids, and in some classes of inhibitory interneurones and the strongly facilitating EPSPs elicited in other classes of interneurones. It forms the basis for the proposal that connections between pyramidal cells are phasic, responding strongly to novel inputs, but reporting little detail about maintained activity. In striking contrast, facilitating synapses report little or nothing at the beginning of a presynaptic spike train, but respond increasingly powerfully as activity continues; both types of connection retaining a short term memory for preceding activity.

To study the time course of recovery from short interval paired pulse (and brief train) depression, dual intracellular recordings were made in slices of adult rat and cat neocortex. The presynaptic neurone was driven to fire with different patterns and at different frequencies with a range of injected current pulses and postsynaptic responses recorded. Cells were filled with biocytin and processed histologically for identification, and where appropriate, with immuno-fluorescence for parvalbumin. At the connections made by pyramidal axons with some types of interneurones, the early recovery from paired pulse depression appeared adequately described by a single exponential. This could, for example, represent the time course of removal of a/the factor that retards primed vesicles. That this time course is slower when the recovery of later EPSPs in trains is studied might result simply from accumulation of the factor. In these connections, therefore, existing models adequately fit the data. The mechanism responsible for a residual slowly recovering phase of depression, apparent when this curve is extrapolated and equivalent to 10-20% of the 1st EPSP amplitude at intervals >50ms, has not been identified and may be of postsynaptic origin.

At the majority of connections between pyramidal cells and at the pyramidal inputs onto parvalbumin immuno-positive interneurones, however, the picture was rather more complex. At very short interspike intervals (5-10ms) these connections exhibited powerful depression. Recovery from depression was at first rapid and some connections even displayed modest facilitation at intervals around 15ms. This early recovery (or the facilitation) was then interrupted by another phase of depression, peaking at around 20ms and declining again rapidly (half width around 5ms), to be followed at some connections by another phase of modest facilitation. Fluctuation analysis demonstrated that the mechanism underlying this ‘notch filter’ is of presynaptic origin. The proportion of failures of transmission increased during the ‘notch’ and the proportional change in CV-2 (inverse square of the EPSP coefficient of variation = np/[1-p]) was greater than the change in M (mean EPSP amplitude = npq). The ‘notch’ was equally apparent in all EPSPs in brief trains of 3-7 EPSPs, representing therefore a true frequency filter. It was, however, relatively insensitive to the size of the preceding EPSP or to the probability of release at low frequencies, differing therefore from the release-dependent components of depression.

The functional relevance of this ‘notch’ was tested by depolarising the postsynaptic neurone close to firing threshold. The probability of an EPSP initiating a postsynaptic AP correlated with its amplitude at more negative membrane potentials, with EPSPs occurring at very short interspike intervals and those coinciding with the peak of the ‘notch’ being less effective than those at intermediate, or longer intervals. Without significant coincident activity and summation, therefore, or unless some chemical modulator reduces the power of the ‘notch’, the only cells that will respond repeatedly to a presynaptic pyramidal cell firing at gamma frequencies will be certain classes of non-parvalbumin interneurones.

Although much of what we currently understand about synaptic transmission results from studies of the neuromuscular junction in the 40’s 50’s and 60’s (McLachlan, 1978 Katz, 1996, for reviews), we are only just beginning to understand the complexities of the synaptic release machinery, and to provide tentative suggestions for the molecular mechanisms that might underlie each of the frequency-dependent properties observed in electro-physiological experiments. Some of these mechanisms may produce only subtle changes in release probability, or in the availability of the immediately releasible pool, but these subtle differences can result in powerful changes in population activity when many synaptic inputs are involved.

References

Betz, WJ (1970). Depression of transmitter release at the neuromuscular junction of the frog. J. Physiol. 206: 620-644.

Katz B (1996) Neural transmitter release: From quantal secretion to exocytosis and beyond The Fenn Lecture. J. Neurocytology 25: 677-686.

Llinas, RR, Blinks, JB and Nicholson, C (1972) Calcium transients in presynaptic terminal of squid giant synapse: detection with aequorin. Science 176: 1127- 1129.

McLachlan EM (1978) The statistics of transmitter release at chemical synapses. In “International Review of Physiol., Neurophysiology III, Vol 17. Ed R. Porter. Baltimore, MD. University Park. pp 49-117.

Singer W (1999) Neuronal synchrony: a versatile code for the definition of relations? Neuron. 24:49-65, 111-125.

Singer W (2001) Consciousness and the binding problem. Ann N Y Acad Sci. 929:123-146.

Thomson AM and Bannister AP (1999) Release- independent depression at pyramidal inputs onto specific cell targets: dual recordings in slices of rat cortex. J. Physiol. 519: 57-70.

Thomson AM (2000) Molecular frequency filters at cortical synapses. Prog. Neurobiol. 62: 159-196.

Traub RD, Spruston N, Soltesz I, Konnerth A, Whittington MA, Jefferys JGR (1998) Gamma- frequency oscillations: A neuronal population phenomenon, regulated by synaptic and intrinsic cellular processes, and inducing synaptic plasticity. Prog. Neurobiol. 55:563-575