Figure 1.
(A) The model Shaffer collateral axon (blue) from CA3 making an en passant bouton (green) with the dendrite of a CA1 pyramidal neuron showing (right) the physiological spatial distributions and concentrations of ligands and molecules. The simulations were carried out in 0.5 µm×0.5 µm×4 µm volume of the axon including of a cluster of voltage dependent calcium channels (VDCCs), mobile calcium buffer calbindin and plasma membrane calcium ATPase (PMCA) pumps. The active zone was populated by seven docked vesicles each with its own calcium sensor for neurotransmitter release at a prescribed distance, lc from the VDCC cluster. (B) Kinetic model for the calcium sensor with 2 pathways, synchronous and asynchronous. The synchronous release pathway has five calcium binding sites whereas asynchronous release has two calcium binding sites. Note that the neurotransmitter release process has distinct rates, γ, for synchronous release and a slower one, aγ, for asynchronous release. When the refractory period was implemented, the release machinery was disabled after a release event takes place, whether via either synchronous or asynchronous, and was re-enabled with a time constant, ε, of 6.34 ms.
Table 1.
Model parameters.
Figure 2.
(A) The neurotransmitter release profile with no external stimulus illustrating the basal release rate. This steady state release profile is a distinct characteristic of the calcium sensor and is independent of geometry. The transient seen in the data is due to starting the simulation off with the sensor in the completely unbound state. (B) Calcium sensitivity of neurotransmitter release response for a range of distances, lc between the calcium sensor and the VDCCs. The VDCC number is adjusted to give the release probability. A set of non-overlapping curves emerge for various distances. Local peak calcium concentration at the site of the active zone is a measure that is modulated by spatial details.
Figure 3.
(A) Stimulus evoked neurotransmitter release data from dual patch clamp recordings in paired cells using hippocampal pyramidal neurons showing two time scales of release. Figure adapted from Goda and Stevens [1], Fig. 4. (B) Black line shows simulation of neurotransmitter release transient for a synapse with intrinsic pr = 0.2 showing two distinct time scales of release (10 ms bins, compare with 3a). Grey line with shows simulations of kinetic model by Sun et al. [18] in a CA3-CA1 with a single active zone. Dashed grey line describes the average base level (no stimulus) release. (C) Figure adapted from from Scheuss et al. [22], Fig. 6. Measured release transient at the calyx of Held showing a fast timescale of release. (D) A superfast time scale (τsuperfast) emerges for neurotransmitter release rate (pr = 0.2) using finer 1 ms bins (left axis, black line). Compare with the superfast timescale of release described at the calyx in 3C. The calcium pulse measured 10 nm from the calcium sensor in response to 48 VDCCs at lc = 250 nm that triggered neurotransmitter release is superimposed (right axis, red line). The initial superfast part of the release is highly correlated to the calcium pulse (phasic synchronous release) and is followed by a fast timescale of release (delayed synchronous release). (E, F). Release transient in response to an action potential for synapses with pr = 0.6 and pr = 0.95 in 10 ms bins. The insets show the superfast timescale for the same data (1 ms bins). The release transient for pr = 0.6 is generated for synapse with 128 VDCCs placed 400 nm from the sensor and 112 VDCCs placed at 250 nm for pr = 0.95. Even though the maximum amplitudes of the two components of release in a pr-dependent way, the 3 decay time constants τsuperfast, τfast and τslow are insensitive across a wide range of release probabilities. The decay time scales are also independent of ultrasynaptic structure (compare b, d, e, f). For a synapse with pr = 0.2 , 44% of release takes place at τsuperfast , 43% at τfast, and the remainder at τslow. For comparison to Goda and Stevens [1] exponential decay times scales are fit to the equation a0 exp (-t/τfast) +a1 exp (-t/τslow) +a2. For B, τfast = 6.0±0.7 ms, τslow = 160.0±14.1 ms (a0 = 0.025, a1 = 0.00023 and a2 = 0.00012). For E, τfast = 7.0±0.7 ms, τslow = 150.0±14.1 ms (a0 = 0.053, a1 = 0.00070 and a2 = 0.00008). For F, τfast = 8.5±0.7 ms, τslow = 120.0±14.1 ms (a0 = 0.16, a1 = 0.00080 and a2 = 0.00007). The ‘superfast’ timescale with 1 ms binning was fit to the equation b0 exp (-t/τsuperfast) +b1 exp (-t/τfast) +b2 exp (-t/τslow) +b3. For D (inset), τsuperfast = 0.7, τfast = 7±0.7 ms τfast = 160.0±14.1 ms (b0 = 0.01, b1 = 0.0009 and b2 = 0.00005 and b3 = 0.000015).
Figure 4.
Contributions of synchronous and asynchronous release for a range of probabilities.
(A-C): The synchronous pathway is the main contributor of the phasic synchronous and delayed synchronous release. The asynchronous release peaks much later. The overall contribution of the asynchronous release increases with release probability (805 events for pr = 0.2, 1213 events for pr = 0.6 and 1511 events for pr = 0.9). The overall ratio between asynchronous and the first synchronous release however remains small [72]. (D-F): The probability distribution (black line) for the number of released vesicles when the RRP is set to be infinite (no depletion after release). Cumulative probability is shown in grey. Consistent with size of the RRP of CA3-CA1, more than 8 vesicles are rarely released. This validates the binding and unbinding rates of calcium ions for the sensor for vesicle release. Also synapses with higher intrinsic pr are more likely to release more vesicles per stimulus.
Figure 5.
Neurotransmitter release profile for a CA3-CA1 synapse with a single active zone and seven docked vesicles.
(A) Release data histogram in 10 ms bins for a synapse with intrinsic release probability of pr = 0.2 (48 channels at lc = 250 nm). Both transient, refractory period transient (grey) and non-refractory period transient (black) almost exactly overlap. (C) This holds true for a finer 1 ms bin (bottom panel) as well. (B) Release data histogram in 10 ms bins for a high release probability pr = 0.92 (48 channels at lc = 250 nm). The two transients in this case decay with different rates. The synapse without the refractory period decays faster, as depletion of neurotransmitter vesicles cause decreasing release probability. (D) This effect is seen in more detail with 1 ms bins at the same synapse. Only for the synapse with refractory period are the characteristics time scales of decay conserved across the whole range of release probability.
Figure 6.
Differences seen due to refractoriness in components of synchronous and asynchronous release.
(A) For a synapse with refractoriness the synchronous release has a shorter, broader peak than the synapse without refractoriness. (B) The asynchronous release channel encompasses more events for synapse with refractoriness compared to without refractoriness. Neurotransmitter release profile for fast sensor KO and wild type for a synapse with and without refractoriness (1 ms bins). (C) The neurotransmitter release profiles for asynchronous release in wild type and fast sensor KO varieties of the synapse with refractoriness (grey) diverge as they approach shorter time scales of less than 20 ms . Fast release through the synchronous pathway suppresses release from the asynchronous pathway due to the refractory period in the wild type, leading to a dip in asynchronous release. (D) The release profiles of wild type and fast sensor KO run almost parallel through the 400 ms transient in the synapse without (black) a refractory period. The transgenic fast sensor KO in both kinds of synapses (with and without refractoriness) is more elevated than the wild type as there is no depletion of vesicles, through the synchronous pathway, from the limited resource available in the RRP. The release starts 3 ms after initiating the action potential (see Fig. S3, as mentioned in the timescale results on page 10) and we have therefore not included this early period in the graphs having 1 ms binning (C and D).
Figure 7.
Response to a 200 ms at 50 Hz rate stimulus protocol administered to a model CA3-CA1 synapse with seven docked vesicles.
In (A) and (B) a synapse with low intrinsic release probability of pr = 0.2, in (C and D) a synapse with a release probability of pr = 0.6 and in (E and F) a high release probability synapse (pr = 0.9) is shown. In the high pr synapse, depletion quickly overwhelms release. Comparing A to B, C to D and E to F, the base level asynchronous release was higher in the synapse with refractoriness (black) whereas the synapse without refractoriness (grey) had higher peak release rates. This is because the refractoriness inhibits immediate release (less that 6 ms interval) from the synchronous pathway and therefore allows the asynchronous release pathway to contribute more to the release. The rates of facilitation and depression were also characteristically different for these synapses.
Figure 8.
Response to 10 Hz train stimuli.
Release rate for wild type (A) and a simulated asynchronous release sensor (SAKO) (B) plotted in 1 ms bins. The same data is plotted on a log scale to show the elevated long tail of release (black line) due to the presence of asynchronous sensor in the wild type (C). The grey line in (C) is SAKO. In (D) total release rate (100 ms bins) for each stimuli is shown (wild type – black line, SAKO – grey line). The facilitation for the wild-type is 50% as opposed to 35% for the SAKO. In this study vesicle replenishment, which occurs at a timescale of the order of seconds, does not play a role.
Figure 9.
Release profile through the asynchronous pathway with identical vesicle fusion rates for synchronous and asynchronous release, compared with unequal fusion rates (1 ms bins).
There is a sharp peak in the asynchronous release after the stimulus that coincides with the calcium signal at the active zone when the vesicle fusion rate is equal for the synchronous and asynchronous case. This peak seen in the simulations is not consistent with observed data. However, slowing down the fusion rate by a factor of 40 matches the data for spontaneous asynchronous release. The X axis starts at 3 ms, this is the delay in release after initiating the action potential (see Fig. S3, as mentioned in the timescale results on page 10).
Figure 10.
(A) Voltage Gated Calcium Channels (B) PMCA pump and (C) Calbindin. The rates with the respective references appear in Table 1.