Ultrastructural contributions to desensitization at cerebellar mossy fiber to granule cell synapses

Abstract

Postsynaptic AMPA receptor desensitization leads to depression at some synapses. Here we examine whether desensitization occurs at mossy fiber to granule cell synapses and how synaptic architecture could contribute. We made whole-cell voltage-clamp recordings from granule cells in rat cerebellar slices at 34 degrees C, and stimulated mossy fibers with paired pulses. The amplitude of the second EPSC was depressed by 60% at 10 msec and recovered with tau approximately 30 msec. This fast component of recovery from depression was reduced by cyclothiazide and enhanced when release probability was increased, suggesting that it reflects postsynaptic receptor desensitization. We evaluated the importance of synaptic ultrastructure to spillover and desensitization by using serial electron microscopy to reconstruct mossy fiber glomeruli. We found that mossy fiber boutons had hundreds of release sites, that the average center-to-center distance between nearest release sites was 0.46 microm, and that these sites had an average of 7.1 neighbors within 1 microm. In addition, glia did not isolate release sites from each other. By contrast, desensitization plays no role in paired-pulse depression at the cerebellar climbing fiber, where glial ensheathment of synapses is nearly complete. This suggests that the architecture of the mossy fiber glomerulus can lead to desensitization and short-term depression. Modeling indicates that, as a consequence of the close spacing of release sites, glutamate released from a single site can desensitize AMPA receptors at neighboring sites, even when the probability of release (p(r)) is low. When p(r) is high, desensitization would be accentuated by such factors as glutamate pooling.

Fig. 1.

Depression is reduced by cyclothiazide. All experiments were performed at 34°C in the presence of 5 μm CPP. Stimulus artifact has been removed for clarity.A, Two separate experiments showing the effect of cyclothiazide. Left, Average EPSCs recorded in 2 mm Ca_e after stimulation with interpulse interval Δt = 10 msec. Right, Average EPSCs after washing in cyclothiazide. Traces are averages of 5–13 trials. B, PPR at Δt = 10 msec in 2 and 3 mm Ca_e, with and without 50 μm cyclothiazide. Ratios are average ± SE of 12–51 experiments. C, Increase in PPR for Δt = 10 msec after addition of cyclothiazide. Ratios are average ± SE of 12–21 experiments.

Fig. 2.

Cyclothiazide specifically affects AMPA receptor desensitization. A, No effect of cyclothiazide on NMDA EPSCs. Sample experiment in the presence of 10 μm NBQX and no CPP shows the NMDA EPSC in control (left), after addition of cyclothiazide (middle), and these two traces overlaid (right). The same lack of effect was found in four other experiments. B, Relationship between PPR at Δt = 10 msec in control conditions and cyclothiazide, in 2 mm Ca_e and 3 mmCa_e. Each point is from a single experiment. Dotted line is unity. C, Relationship between control PPR at Δt = 10 msec and the increase in PPR after addition of cyclothiazide, in 2 mm Ca_e and 3 mm Ca_e. D, Relationship between the increase in EPSC amplitude and the increase in PPR after addition of cyclothiazide for a number of experiments. Lines are fits to the data (EPSC₁: intercept = 1.48 ± 0.13, slope = −0.08 ± 0.06; EPSC₂: intercept = 0.43 ± 0.22, slope = 1.08 ± 0.10).

Fig. 3.

Fast component of depression is reduced by cyclothiazide. A, Single experiment showing EPSC paired-pulse plasticity in 2 mm Ca_e. The mossy fiber was stimulated twice with varying intervals. Eachtrace is the average of five to six trials.B, Separate experiment from A showing paired-pulse plasticity in the presence of 50 μmcyclothiazide. Each trace is the average of 10–13 trials. C, Paired-pulse plasticity in 2 mmCa_e, in the presence and absence of 50 μm cyclothiazide. Each point is the average of 9–11 experiments. D, Paired-pulse plasticity in 3 mm Ca_e, in the presence and absence of 50 μm cyclothiazide. Each point is the average of 8–10 experiments.

Fig. 4.

Mossy fiber ultrastructure. A, Single electron microscopic section of a mossy fiber glomerulus. The mossy fiber axon is shaded blue, granule cell dendritic processes that receive synaptic contacts from the mossy fiber in this or adjacent sections are shaded pink, and glial processes are shaded yellow. Other structures, including granule cell somata or dendritic processes that make no synaptic contact, are not shaded. Release sites are marked byasterisks. The boxed area is enlarged inB. B, Serial sections through one release site. The release site consists of a cluster of presynaptic vesicles, active zone material, widening of the synaptic cleft, and postsynaptic density.

Fig. 5.

Stereo pairs of reconstructed mossy fibers. Mossy fibers were reconstructed from 96 to 177 serial sections. The mossy fiber axon is depicted in blue. Mossy fiber membrane that contacts glia is colored yellow. Postsynaptic densities on opposing dendritic processes are indicated inred.

Fig. 6.

Distances between release sites. A, Distribution of distances to nearest neighbor. The distribution was determined separately for each mossy fiber and then normalized and averaged. Inset, Cumulative frequency histogram of nearest neighbor distances for four mossy fibers. B, Average number of release sites encountered at varying distances from a starting release site. This was determined separately for each mossy fiber and then averaged together. Inset, Close-up of the same relationship over the first 1 μm away from the average release site.

Fig. 7.

Reconstructed climbing fiber segments. The climbing fiber axon is depicted in blue. Climbing fiber membrane that contacts glia is colored yellow. Postsynaptic densities on opposing dendritic spines are indicated inred. Scale bar, 2 μm. These climbing fibers are from the data set of Xu-Friedman et al. (2001).

Fig. 8.

Simulation of glutamate diffusion and AMPA receptor desensitization. A, Glutamate concentration as a function of distance from the release site at a series of time points. The model was of glutamate diffusion from a single vesicle at a single release site as described by Equation 2, with vesicle radius ρ = 25 nm, initial concentrationC₀ = 100 mm, cleft width λ = 20 nm, and glutamate diffusion constantD = 0.4 μm²/msec.B, Glutamate concentration as a function of time since release at various distances from the release site. C,D, AMPA receptor desensitization as a function of time since release at various distances from the release site. Two AMPA receptor models were used, based on Purkinje cell outside-out patches (C) (Model 1) (Wadiche and Jahr, 2001) and on chick magnocellularis outside-out patches (D) (Model 2) (Raman and Trussell, 1995).