Blocking slow exocytosis with slow Ca²⁺ buffers slows recovery from depression

Synaptic transmission is a hungry process. During periods of elevated neuronal activity, sustained exocytosis can consume the pool of available synaptic vesicles and lead to profound synaptic depression (Betz, 1970). At some synapses depression may act as a useful computational feature, employed by neural circuits for the purposes of gain control and novelty detection (Abbott & Regehr, 2004). But many synapses must maintain strength in order to reliably transmit information during high‐frequency firing. This requires efficient resupply of vesicles to the readily releasable pool (RRP). For years it has been assumed that synapses employ an active mechanism to increase the rate of vesicle replenishment during periods of intense activity, a phenomenon known as “Ca²⁺‐dependent recovery from depression”. In this issue of The Journal of Physiology Ritzau‐Jost et al. (2018) provide an elegant analysis and a new perspective on Ca²⁺‐dependent recovery from depression, and suggest that – for at least one synapse – the phenomenon that has been observed may have little to do with vesicle replenishment.

Twenty years ago a seeming paradox was discovered: recovery from depression can be accelerated by strong stimulation, despite the fact that strong activation also drives greater depletion of the RRP. Following 100 Hz stimulation, recovery at the calyx of Held proceeds with a single exponential time constant of 4–5 s (at room temperature), a rate typical for many central synapses (von Gersdorff et al. 1997). Stimulating the calyx of Held at higher frequencies (300 Hz) reveals a component of recovery with a much faster time constant (∼80 ms) (Wang & Kaczmarek, 1998). This accelerated vesicle replenishment appears to be driven by the elevated cytoplasmic Ca²⁺ that persists after action potentials, known as “residual calcium”. Both high‐frequency stimulation and manipulations that increase Ca²⁺ influx during action potentials accelerate recovery from depression. Recovery from depression is slowed when residual calcium is clamped at lower levels by introducing the slow Ca²⁺ buffer EGTA into nerve terminals (Dittman & Regehr, 1998; Stevens & Wesseling, 1998). This led to the hypothesis that an active, Ca²⁺‐sensitive mechanism speeds up replenishment and helps synapses avoid premature vesicle depletion.

Subsequent studies suggested a mechanism for Ca²⁺‐dependent recovery from depression. Using the calyx of Held, Sakaba & Neher (2001) showed a role for calmodulin in accelerating recovery. Calmodulin binds to Ca²⁺ with relatively high affinity, making it a promising candidate for sensing the sub‐micromolar residual Ca²⁺ signals. Calmodulin inhibitors slow the rate of recovery from depression (Sakaba & Neher, 2001). Munc13, which plays an important role in regulating the RRP, binds to Ca²⁺‐activated calmodulin to form a complex that recruits new vesicles to the RRP (Junge et al. 2004). Knock‐in mice with a mutated form of Munc13 that does not bind to calmodulin exhibit slowed recovery from depression at the calyx of Held (Lipstein et al. 2013). However, this mutation did not fully block the acceleration in replenishment following 300 Hz stimulation, suggesting that other mechanisms may play a role in Ca²⁺‐dependent recovery from depression. One such mechanism might be an interaction between calmodulin and myosin light chain kinase (MLCK), which has also been shown to affect vesicle recruitment at the calyx of Held (Srinivasan et al. 2008; Lee et al. 2012).

Ritzau‐Jost (2018) and colleagues set out to evaluate the role of these replenishment mechanisms at cerebellar mossy fibre boutons (cMFBs). The cMFB synapse exhibits fast replenishment and can reliably transmit high‐frequency firing rates (Saviane & Silver, 2006), but the Ca²⁺ dependence of recovery has yet to be directly studied at this synapse. The authors show that at the cMFB, like the calyx of Held, recovery following 300 Hz stimulation exhibits both a fast and a slow component. Application of the slow Ca²⁺ buffer EGTA‐AM abolishes most of the fast component of recovery, suggesting accelerated Ca²⁺‐dependent recovery from depression.

However, as the authors tested each of the putative mechanisms for Ca²⁺‐dependent vesicle replenishment they discovered that none could explain the apparent Ca²⁺ dependence of recovery from depression. Recovery was unaffected by calmidazolium, a membrane‐permeable calmodulin inhibitor. Knock‐in mice with the mutated form of Munc13 that does not bind calmodulin also showed no deficits in fast recovery from depression. Finally, the authors patched presynaptic boutons in order to introduce a membrane‐impermeable peptide inhibitor that blocks the calmodulin–MLCK interaction. Again, disrupting this pathway had no effect on recovery. However, by directly patching presynaptic boutons they were able to depolarize terminals for longer than a normal action potential, which revealed that vesicle release proceeds in multiple phases driven by two pools of synaptic vesicles. A fast pool of vesicles fuses quickly after depolarization, while a slow pool of reluctant vesicles takes longer to release. Intriguingly, the fast‐releasing pool of vesicles recovers with a slow time constant, while the slow‐releasing pool recovers very quickly.

If the Ca²⁺ dependence of recovery from depression at cMFBs is not driven by any of the previously elucidated mechanisms, how can it be explained? To account for their findings the authors constructed models in which synaptic transmission is driven by two vesicle pools with different Ca²⁺ sensitivities, kinetics of release and rates of replenishment. The pool with a high initial probability of release (p _r) is responsible for the initial phase of exocytosis, but this pool depletes quickly and recovers slowly. After the fast pool is depleted, the slow‐releasing low‐p _r pool takes over, and is able to maintain synaptic transmission during prolonged, high‐frequency stimulation due to its faster rate of replenishment. The apparent Ca²⁺ dependence of recovery from depression actually reflects the differences in Ca²⁺ dependence of release for the two pools. The effect of EGTA can be explained by its ability to preferentially block release of the slow‐releasing pool, either by lowering p _r or by blocking facilitation, a Ca²⁺‐dependent process that increases the probability of release (Jackman & Regehr, 2017).

The present study provides solid evidence against active Ca²⁺‐dependent mechanisms for recovery from depression at cMFBs, and a compelling explanation for the observed phenomenon. It also suggests that synapses where Ca²⁺‐dependent recovery from depression has been observed should be re‐examined in rigorous, quantitative detail. However, the idea that vesicles in the RRP possess heterogeneous properties is not new. Soon after the discovery of Ca²⁺‐dependent recovery, Wu & Borst (1999) showed that prolonged stimulation at the calyx of Held results in the release of two distinct pools of vesicles with differing kinetics of both release and replenishment They proposed a model with striking similarities to that of Ritzau‐Jost et al. (2018). Importantly, Ritzau‐Jost et al. (2018) suggest a novel way to test for Ca²⁺‐dependent recruitment using EGTA: if the first few EPSCs triggered by a stimulus train are not reduced by EGTA and only the rate of recovery from depression is affected then the process is genuinely an accelerated Ca²⁺‐dependent recruitment of vesicles.

Perhaps the most pressing question raised by Ritzau‐Jost and colleagues is what defines the properties of the two vesicle pools. Could it simply be a matter of location, with fast‐releasing high p _r vesicles residing closer to Ca²⁺ channels? In support of this hypothesis, the authors could recapitulate many aspects of their data with a positional, 3‐dimensional model of release sites (Fig. 1). Alternatively, are there molecular distinctions between the pools, such as the presence of different Ca²⁺ sensors for exocytosis? At the cerebellar Purkinje cell synapses, another giant CNS synapse, the high‐affinity Ca²⁺ sensor that drives facilitation, synaptotagmin‐7 (Jackman et al. 2016), appears to be critical for the release of a reluctant pool of vesicles during prolonged stimulation (Turecek et al. 2017). Intriguingly, knockout of synaptotagmin‐7 at this synapse significantly accelerates recovery from depression. It also remains unclear why the pool of slow‐releasing vesicles is able to replenish so much faster than the fast‐releasing pool. Further studies are clearly needed to understand the mechanisms that contribute to vesicle heterogeneity.

Figure 1. Two vesicle pools: fast and slow.

A fast‐releasing pool of vesicles fuses quickly after depolarization and Ca²⁺ influx, while a larger pool of reluctant vesicles takes longer to release. Both pools can contribute to neurotransmitter release during a stimulus train. The fast‐releasing pool of vesicles recovers with a slow time constant, while the slow‐releasing pool recovers very quickly. Introducing the slow Ca²⁺ buffer EGTA into nerve terminals blocks the release of the fast‐replenishment pool, which appears to slow the recovery from synaptic depression. Caution is required when interpreting the Ca²⁺ dependence of recovery from depression.

Additional information

Competing interests

No competing interests.

Author contributions

All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

H.v.G. was funded by National Institutes of Deafness and Communication Disorders grants DC012938 and DC004274.