can be as wide as 0.2–0.91 [25, 26]. Modification of the releasing probability has
been associated to the so-called presynaptic (non NMDA-dependent) long-term
potentiation (LTP) assuming that the efficacy activity dependent of a synapse
depends on the increase of the releasing probability [17, 27–29]. This point is crucial
for the understanding on how the presynaptic neural code is coded synaptically
because it means that not all the presynaptic spikes are coded by an EPSP. Moreover, if this probability changes as a function of the activity, this means that a
different number of EPSP code for a given number of presynaptic spikes depending
on the preceding activity. In terms of information, not all the presynaptic bits are
transferred but only a fraction of it, and the size of the fraction is activity dependent. Moreover, the sequence of EPSPs does not sum linearly at the postsynaptic
side [30, 31]. This means not only that only a part of bits composing the presynaptic
“word” is transferred but also that their postsynaptic representation is extremely
variable and depending on how many bits are transferred (which change as the
probability of release changes with activity) and on the timing between the transferred bits.
Although the vesicular release is considered of quantal type, the release of single
vesicle can produce different responses depending on several presynaptic factors
(see, [2]). An important factor is the position of the vesicle (eccentricity) with
respect to the central axis of the cylinder limited by the AZ and PSD. For a given
configuration of the PSD (see next section), the release of glutamate from a more
peripheral vesicle will produce an EPSC with smaller amplitude than one centered to
AZ-PSD central axis [32–34]. Another important factor is the amount of molecules
into the vesicle. Vesicle concentration, in fact, is extremely variable ranging 60–
210 mM [15, 16, 35] with an average of $140 mM. Assuming an internal radius of the
vesicle with an average of 23 nm, it is clear that the number of molecules of glutamate
released for a single bit of information is extremely variable. A variable number of
molecules produce EPSC with different amplitude [32–34, 36–38]. In our early work
on single glutamatergic response, we have considered the combination of the number
of molecules and the position of release as stochastic factors [8, 32–34]. However, by
considering the large variability of the concentration of glutamate in the vesicles, the
thousand possible combinations of “position-number of molecules,” this could be a
powerful system of presynaptic regulation of the information transfer. In this respect,
an interesting question arise: “what is the mechanism which, for a given presynaptic
spike, ‘decide’ the correct combination ‘position-number of Glutamate molecules’?”
The SNARE complex, because of its different configurations depending on the membrane activity, could be a candidate for this decision role [13, 39, 40].
Although in the larger part of the cases a single vesicle opens with probability
less than 1 for the arrival of a single presynaptic spike, in some cases a
multivesicular release has been observed (see, e.g., [41, 42]). The multivesicular
release found in some experiments opens many other interesting questions. The
most relevant is: what is the relationship on the number of vesicle opened for a spike the
of information transferred? Another interesting question is What is the role of
multivesicle release if usually a single release does not achieve postsynaptic saturation of
the response? [35, 43]. To summarize, the most important presynaptic factor of EPSP
variability are:
• Probability of release of a vesicle following a single presynaptic spike and its
dependence on the past activity
• Probability of multivesicular release
• Number of molecules inside the released vesicle
96
Advances in Neural Signal Processing
been associated to the so-called presynaptic (non NMDA-dependent) long-term
potentiation (LTP) assuming that the efficacy activity dependent of a synapse
depends on the increase of the releasing probability [17, 27–29]. This point is crucial
for the understanding on how the presynaptic neural code is coded synaptically
because it means that not all the presynaptic spikes are coded by an EPSP. Moreover, if this probability changes as a function of the activity, this means that a
different number of EPSP code for a given number of presynaptic spikes depending
on the preceding activity. In terms of information, not all the presynaptic bits are
transferred but only a fraction of it, and the size of the fraction is activity dependent. Moreover, the sequence of EPSPs does not sum linearly at the postsynaptic
side [30, 31]. This means not only that only a part of bits composing the presynaptic
“word” is transferred but also that their postsynaptic representation is extremely
variable and depending on how many bits are transferred (which change as the
probability of release changes with activity) and on the timing between the transferred bits.
Although the vesicular release is considered of quantal type, the release of single
vesicle can produce different responses depending on several presynaptic factors
(see, [2]). An important factor is the position of the vesicle (eccentricity) with
respect to the central axis of the cylinder limited by the AZ and PSD. For a given
configuration of the PSD (see next section), the release of glutamate from a more
peripheral vesicle will produce an EPSC with smaller amplitude than one centered to
AZ-PSD central axis [32–34]. Another important factor is the amount of molecules
into the vesicle. Vesicle concentration, in fact, is extremely variable ranging 60–
210 mM [15, 16, 35] with an average of $140 mM. Assuming an internal radius of the
vesicle with an average of 23 nm, it is clear that the number of molecules of glutamate
released for a single bit of information is extremely variable. A variable number of
molecules produce EPSC with different amplitude [32–34, 36–38]. In our early work
on single glutamatergic response, we have considered the combination of the number
of molecules and the position of release as stochastic factors [8, 32–34]. However, by
considering the large variability of the concentration of glutamate in the vesicles, the
thousand possible combinations of “position-number of molecules,” this could be a
powerful system of presynaptic regulation of the information transfer. In this respect,
an interesting question arise: “what is the mechanism which, for a given presynaptic
spike, ‘decide’ the correct combination ‘position-number of Glutamate molecules’?”
The SNARE complex, because of its different configurations depending on the membrane activity, could be a candidate for this decision role [13, 39, 40].
Although in the larger part of the cases a single vesicle opens with probability
less than 1 for the arrival of a single presynaptic spike, in some cases a
multivesicular release has been observed (see, e.g., [41, 42]). The multivesicular
release found in some experiments opens many other interesting questions. The
most relevant is: what is the relationship on the number of vesicle opened for a spike the
of information transferred? Another interesting question is What is the role of
multivesicle release if usually a single release does not achieve postsynaptic saturation of
the response? [35, 43]. To summarize, the most important presynaptic factor of EPSP
variability are:
• Probability of release of a vesicle following a single presynaptic spike and its
dependence on the past activity
• Probability of multivesicular release
• Number of molecules inside the released vesicle
96
Advances in Neural Signal Processing