What Are the Mechanisms for Analogue and Digital Signalling in the Brain? Dominique Debanne, Andrzej Bialowas, Sylvain Rama

What are the mechanisms for analogue and digital signalling in the brain? Dominique Debanne, Andrzej Bialowas, Sylvain Rama

To cite this version:

Dominique Debanne, Andrzej Bialowas, Sylvain Rama. What are the mechanisms for analogue and digital signalling in the brain?. Nature Reviews Neuroscience, Nature Publishing Group, 2013, 14, pp.63-69. hal-01766838

HAL Id: hal-01766838 https://hal-amu.archives-ouvertes.fr/hal-01766838 Submitted on 25 Apr 2018

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. examples of graded transmission in the absence of spiking activity have also been What are the mechanisms for described in invertebrate neurons (see REF. 7 for a review). analogue and digital There is now evidence that analogue signalling exists at spiking synapses, where signalling in the brain? neurotransmitter release is evoked by action potentials rather than being tonic. In this article, we discuss recent work suggesting the possibility of mixed analogue–digital signal- Dominique Debanne, Andrzej Bialowas and Sylvain Rama ling at local axonal connections in CNS circuits, with an emphasis on the hippocampus. Abstract | Synaptic transmission in the brain generally depends on action This has important and exciting implications potentials. However, recent studies indicate that subthreshold variation in the for our understanding of information presynaptic membrane potential also determines spike-evoked transmission. processing in this part of the brain. The informational content of each presynaptic action potential is therefore greater than initially expected. The contribution of this synaptic property, Analogue and digital signalling Synapses translating analogue presynap- which is a fast (from 0.01 to 10 s) and state-dependent modulation of functional tic membrane potential fluctuation into coupling, has been largely underestimated and could have important graded tonic release of transmitter display consequences for our understanding of information processing in neural a significantly higher rate of information networks. We discuss here how the membrane voltage of the presynaptic transfer than synapses using presynaptic terminal might modulate neurotransmitter release by mechanisms that do not spike train coding. The dynamic range at involve a change in presynaptic Ca2+ influx. tonic-release, analogue synapses is indeed very large, and a single analogue synapse is able to continuously encode virtually The principal function of the CNS is to is actively propagated along the axon. infinite information levels. For instance, adapt an organism’s behaviour to changes in Neuronal information is therefore transmit- graded synapses in the fly retina trans- the environment — a process that is thought ted to the postsynaptic neuron as discrete mit more than 1,500 bits of information to be mediated by modulation of neuronal amounts of neurotransmitter released by the per second: that is, one or two orders of circuits through synaptic modifications. presynaptic neuron in an all‑or‑none mode. magnitude larger than spiking neurons8,9. Understanding the cellular and molecular This mode of neuronal signalling is thus However, this comes at the price of mechanisms underlying activity-dependent ‘digital’: the neuron either fires or it does high energy consumption. For example, regulation of synaptic strength is therefore not, and neurotransmitter release follows photoreceptors in the retina continuously a major issue in neurobiology. Recent work this binary mode (FIG. 1a). release their neurotransmitter at a high rate indicates that in addition to changes in post- However, neuronal information is not (20–80 vesicles per active zone per second), synaptic receptor expression and synaptic only transmitted digitally, and subthreshold indicating that each active zone may release vesicle release probability, synaptic strength activity that originates in the dendrites and as many as a few million vesicles per day10. in mammalian neurons is regulated by reaches the soma can be conveyed along the Another drawback of analogue signalling subthreshold variations of voltage that are axon to the presynaptic element, in which it is that it is constrained by biophysical laws generated far from the synapse1–4. influences the arriving action potential and such as voltage dissipation along neuronal In most neurons, the proximal region of thus tunes the flow of neuronal information processes over long distances. Therefore, the axon (the axon initial segment (AIS)) via ‘analogue’ coding. Clear examples of pure analogue signalling in neurons is contains a high density of Na+ channels, and analogue transmission of neuronal infor- better suited for local rather than distal is therefore a hotspot for neuronal excita- mation can be found in the inner ear or in transmission of information. tion. Small synaptic potentials that are gen- the retina, in which photoreceptors, bipolar By contrast, digital synapses are able erated in the dendrites summate temporally, and horizontal cells signal photostimula- to signal activity far from the site of spike and if the resulting potential reaches the tion by producing graded potentials with- initiation because neuronal information threshold for action potential generation, out action potentials6. These cells generally encoded in the form of action potentials a spike is initiated at the AIS. According to release transmitter continuously (tonic is carried over very long distances along this view, the AIS is the final site for den- release), and their high rate of spontaneous axons without voltage dissipation, as active dritic integration of synaptic responses5, release is directly modulated by membrane currents regenerate action potentials along which produces an action potential that potential fluctuations (FIG. 1b). Similar the axon11,12. Another major advantage of Digital Analogue AD a b c

Pre

EPSC

Post

Vm soma = –65 mV b1 b2

Vm soma = –45 mV

–45 mV

Vm –65 mV

Axonal distance AD Digital

Figure 1 | Digital, analogue and hybrid (analogue–digital) modes c | Hybrid analogue–digital (AD) transmission. Both subthreshold fluc- Nature Reviews | Neuroscience of synaptic transmission. a | Digital mode of synaptic transmission tuations (red trace) and spiking activity (upper black trace) are transmit- in the CNS. A scheme of two synaptically connected neurons is shown ted. Note that when the presynaptic spike is produced after a prolonged on the left. Transmission (right) is stereotyped and occurs in an period of depolarization (red trace), the spike-evoked synaptic response all‑or‑none (digital) manner (that is, only if a presynaptic action potential (blue trace) is enhanced compared with when there is no prolonged is elicited). Note that subthreshold depolarization (indicated by the depolarization (upper and lower black traces). d | Spatial gradient of arrowhead) produces neither a presynaptic spike nor a postsynaptic hybrid and digital transmission. Because of cable properties, AD trans- response. b | Analogue transmission is a graded mode of transmission of mission is restricted to proximal presynaptic boutons (b1), whereas pure presynaptic voltage fluctuations. The two horizontal dashed lines indi- digital transmission occurs at distal boutons (b2). EPSC, excitatory cate the baselines. The blue trace represents postsynaptic activity. postsynaptic current; TH, spike threshold; Vm, membrane voltage.

digital signalling is its relatively low energy synapses, including cortical2,3,19, cerebel- that a voltage response decays exponentially cost. The kinetics of voltage-gated currents lar20,21 and hippocampal synapses1,22. In along passive cables. The space constant is underlying the action potential are tuned these examples, synaptic transmission the axonal distance for which voltage drops to minimize energetic consumption13,14. evoked by single action potentials is to 37% of its initial value. It depends on both Therefore, in many ways, digital signalling enhanced as the result of analogue-medi- geometrical and electrical factors of the may appear to be an improvement on ana- ated depolarization (10–30 mV) of the pre- axon: the axon membrane resistance, which logue signalling. However, it has also several synaptic element for a few tens to hundreds is determined by the relative contribution limitations. Because of its discrete nature, of milliseconds (FIG. 1c). of conducting (pore-forming proteins) and the coding of information by a single digital non-conducting (lipids) molecules, and synapse is generally poor in comparison Voltage propagation in axons axial resistance, which is controlled by the with analogue synapses. Thus, combining A prerequisite for analogue–digital intra-axonal medium and the axon diam- analogue and digital signalling should in facilitation (ADF) is that analogue voltage eter. Thus, a large space constant is generally principle offer a balance between the dual changes produced in the somatodendritic obtained for low intra-axonal resistance or advantages low energy cost and highly regions are capable of spreading along the high axonal membrane resistance. Excitatory dynamic transfer of neuronal information. axon over long enough distances to reach postsynaptic potentials (EPSPs) that are Hybrid analogue–digital enhance- synapses. This process of analogue volt- generated in the dendritic tree of hippocam- ment of synaptic transmission at spik- age propagation is based on the electrical pal dentate granule cells travel passively ing synapses that was initially described properties of the axon membrane. It was towards the soma and along the axon, and in invertebrates15–18 has more recently first theorized by Rall23 who modelled the the resulting excitatory presynaptic potential been reported in many mammalian CNS axon as a ‘membrane cylinder’ and showed (EPreSP) in mossy fibre terminals can be measured with a patch-pipette1. The axonal a similar spatial dichotomy has recently Inactivation of presynaptic K+ channels. space constant for this transient depolariz- been reported in CA3 pyramidal neurons So far, two independent mechanisms have ing event (~50–100 ms) has been estimated in which ADF is observed at proximal syn- been reported to account for the analogue– as ~450 μm1. The space constant varies as apses established with other CA3 neurons digital enhancement of transmission in a function of the frequency, diminishing but not at synapses with distal CA1 neu- central synapses. First, depolarization of rapidly at high frequency (see supplemen- rons24. In this way, analogue–digital signal- the somatic region of the presynaptic neu- tary information in REF. 2). In neocortical ling increases the computational repertoire ron may enhance synaptic transmission as pyramidal neurons, the axonal space con- of neuronal communication. the consequence of voltage-inactivation stant of steady-state voltage modulation in of a specific type of K+ channel (FIG. 2a). the soma yields values of 420–550 μm2,3. In Mechanisms of ADF Axons contain a high density of voltage- + principle, active conductance such as a per- Principles of ADF. ADF occurs in a highly gated K (KV) channels. They generate a sistent Na+ current could enhance the axon heterogeneous range of synapses in terms current that is strongly inactivated by a space constant by carrying subthreshold of morphology (en passant boutons and small depolarization, suggesting that its depolarization that is initiated in the soma giant terminals), neurotransmitter (GABA contribution to ADF will be important over longer axonal distances. However, the or glutamate) and brain regions (neocortex, even in distal regions of the axon, in large space constant of EPreSPs in hippo- hippo campus and cerebellum). ADF involves which analogue subthreshold depolariza- campal granule cell axons does not result two major steps: the depolarization of the tion is minimal. In neocortical and hippo from the activation of Na+ channels because presynaptic element, which then causes the campal pyramidal neurons, for example, 1 it is unaffected by tetrodotoxin . Therefore, enhancement of neurotransmitter release. the KV1 channels generate a fast-activating the large space constant of granule cell In cerebellar synapses established but slowly-inactivating D‑type current axons may result from specific electrical between GABAergic interneurons of the (ID) that reduces spike duration and thus properties (that is, high membrane resist- molecular layer, subthreshold depolariza- controls neurotransmitter release3,26–28. ance and/or low axial resistance) that need tion facilitates spike-evoked release of Pharmacological inactivation of ID with to be identified. GABA20. In hippocampal dentate granule 4‑aminopyridine or dendrotoxin enhances The notion of an axonal space constant cell axons (mossy fibres), the combina- synaptic strength at hippocampal and is crucial for analogue–digital signalling tion of analogue depolarization spreading neocortical synapses possibly by broaden- because it describes how voltage spreads from the soma in the form of an EPreSP ing the presynaptic spike in the terminal, along the axon and modulates the bio- and digital signalling in the form of an resulting in higher release of neurotrans- physical properties of ion channels located action potential, enhances glutamatergic mitter3,28,29. In a similar manner, blocking in presynaptic terminals. This modulation transmission at the mossy fibre–CA3 cell ID in cortical pyramidal cells also broadens depends on the gating characteristics of the synapse1. For local connections such as the spike in the axon and at presynaptic current. For instance, a modest subthresh- these that occur over relatively short dis- terminals3,28,30 and occludes ADF of synap- old depolarization will be more effective on tances between L5 pyramidal neurons in tic strength3. Furthermore, the kinetics of voltage-gated currents with low activation– the neocortex or GABAergic interneurons the analogue–digital enhancement fit well 2,3 inactivation thresholds (that is, near the in the cerebellum, the analogue-mediated with the inactivation kinetics of ID . Last, resting membrane potential) than on those component of the facilitation of synaptic voltage inactivation of KV1 channels causes with high activation–inactivation thresh- transmission is on average 1–2% per mV a significant enhancement of the Ca2+ olds (that is, far from the resting mem- of somatic depolarization2,3,20. transient in the presynaptic terminal that is brane potential). Consequently, ADF will ADF is mediated by an increase in gluta- evoked by the propagating spike in the pre- spread over long distances if the underly- mate3,25 or GABA20,21 release, as indicated by synaptic terminal31. In fact, Ca2+ chelation ing mechanism involves a low threshold the reduced paired-pulse ratio (PPR): that with EGTA (ethylene glycol tetra-acetic conductance. is, the ratio of synaptic responses for a pair acid) abolishes ADF in L5 pyramidal neu- As many CNS neurons establish local of presynaptic stimuli. Intriguingly, how- rons2. Enhanced neurotransmitter release recurrent connections, analogue–digital ever, in the case of the hippocampal mossy by the inactivation of KV1 channels is not enhancement is therefore likely to have a fibre–CA3 cell synapse, short-term facilita- solely mediated by intrinsic mechanisms major role in a potentially large number tion tested with repeated presynaptic stimuli (that is, depolarization of the presynaptic of cortical, hippocampal and cerebellar is unchanged1, suggesting that (as discussed neuron). In fact, extrinsic signals, such as excitatory and inhibitory circuits2,3,19,20. An later) glutamate release is not changed in a glutamate released from astrocytes, that important consequence of the size of the conventional way. are present abundantly adjacent to hippo axon space constant is that pure digital and Although there are general principles campal axons and their terminals can also hybrid signalling may coexist within the on which ADF mechanisms operate at modulate the width of action potentials in same axon. In cortical pyramidal cell axons, synapses, important differences also exist. axons and enhance synaptic transmission local connections established by boutons For instance, the temporal requirement in an analogue manner32,33. located in the proximal bouquet of axon for ADF is highly heterogeneous. In L5 collaterals are likely to function in a hybrid pyramidal neurons, ADF is observed after Increase of basal Ca2+ concentration. analogue–digital way, whereas long project- several seconds of presynaptic depolariza- Propagated depolarization can also facili- ing axon collaterals may essentially work tion2,3, whereas a few tens of milliseconds tate synaptic transmission by the opening (FIG. 1d) 2+ under a digital mode . For example, of analogue subthreshold depolariza- of voltage-gated Ca (CaV) channels, which at excitatory connections between layer 5 tion is sufficient to induce ADF at mossy results in an increase in basal Ca2+ concen- (L5) pyramidal neurons, the larger ADF is fibre–CA3 cell synapses1 or at cerebellar tration in presynaptic terminals (FIG. 2b). At observed at local connections3. In addition, GABAergic synapses21. rest (that is, –65 mV), intraterminal Ca2+ a K+ concentration is ~100 nM and can increase K+ up to 1 μM upon subthreshold depolariza- 25 Presynaptic Presynaptic tion (~–50 mV ). This increase can easily bouton Vm = –65 mV spike be measured in thin cerebellar axons using fluorescent Ca2+ probes20,21. Alternatively, Axon Ca2+ currents can be directly monitored Soma Postsynaptic EPSC neuron with patch-clamp recording from large terminals such as the calyx of Held25. There is a clear consensus among studies in various brain areas that the Ca channels respon- Vm = –50 mV V sible for this facilitation, whether in the axon terminal20,21,25 or in the axon itself 31, Analogue depolarization are CaV2.1 channels (also known as P/Q- of the soma 2+ type Ca channels). CaV2.2 channels (also known as N‑type Ca2+ channels) that are b located in the axon terminal and the AIS also participate in the increase in baseline Ca2+ signal upon depolarization31. Ca 2.1 Vm = –65 mV V and CaV2.2 channels are activated by high levels of depolarization. Thus, because of voltage attenuation along the axon, their contribution to ADF might be limited to Ca2+ Ca2+ very proximal inputs. The hippocampal mossy fibre terminal, however, appears to

Vm = –50 mV be an exception, as subthreshold depolarization induces a Ca2+ influx that is not only mediated by the classical Ca 2.1 and Ca 2.2 Analogue depolarization V V of the soma channels but also by CaV2.3 channels (also known as R‑type Ca2+ channels)34. c The increase in basal Ca2+ concentra- Axon tion triggered by subthreshold depolarization could have two major consequences. Presynaptic terminal First, it might directly promote neurotransmitter release. In axons of cerebellar Ca 2.2 GPCR V SV SNARE interneurons, for example, subthreshold complex depolarization is sufficient to increase spontaneous, spike-independent release of GABA20. A second possible action of 2+ Postsynaptic neuron an increase in basal Ca concentration is an acceleration of the recruitment of vesicles to the active zone35. The principal Figure 2 | Generic mechanisms for analogue–digital facilitation. a | Facilitation resulting from consequence of this Ca2+-induced prim- voltage-inactivation of voltage-gated K 1 channels. With digital-only signalling (upper neuron), the V ing of vesicles would be an increase in neuronal membrane voltage (V ) is held at –65mV and K 1 channels remainNature open,Reviews resul | Neuroscienceting in a rapid m V synchronous release when the presynaptic termination of the spike response. Analogue subthreshold depolarization (lower neuron; –50mV at spike invades the axon terminal. Both of the soma) propagates from the somatic compartment and along the axon to the terminal (the gradual these effects would result in an increase decay in depolarization is indicated by red-to-blue shading), resulting in closure of KV1 channels and attenuated K+ efflux. This produces a broader action potential and results in enhanced spike-evoked in neurotransmitter release and enhanced Ca2+ entry and incremented neurotransmitter release (single black arrow). b | Facilitation of transmis- synaptic strength. sion resulting from an increase of basal Ca2+ concentration. Under digital signalling conditions (upper 2+ neuron; Vm = –65 mV), the arrival of an action potential is insufficient to open voltage-gated Ca (CaV) Unsolved question: does ADF occur at the channels (shown in red) but is sufficient to evoke release of neurotransmitter (black arrow). During hippocampal mossy fibre synapse? The analogue–digital signalling (lower neuron), partial depolarization (Vm = –50mV) spreading from the cellular mechanisms underlying analogue– soma opens Ca channels, increases Ca2+ influx and basal Ca2+ concentration in the presynaptic bouton V digital enhancement of synaptic strength at and enhances neurotransmitter release (indicated by three black arrows) when a presynaptic action mossy fibre synapses with CA3 cells chal- potential also arrives at the bouton. c | Two putative mechanisms of enhanced exocytosis by neuronal lenge the classical view of synaptic strength Vm. On the left, a subthreshold change in Vm (lightning bolt) is detected by the voltage sensor of CaV2.2 channel (shown in red) that interacts with SNARE (soluble NSF (N-ethylmaleimide-sensitive factor) modulation. First, unlike most examples of attachment protein receptor) proteins (which are part of the exocytosis machinery for synaptic vesi- analogue–digital enhancement, the PPR is cles (SVs)). This results in facilitation of SV fusion and enhanced transmitter release. On the right, a unaffected despite excitatory postsynaptic current (EPSC) facilitation of ~40% in the subthreshold change in Vm is detected by G protein-coupled receptors (GPCRs), such as type 3 metabotropic glutamate receptor, which interact with the release machinery to influence neurotrans- CA3 cell1. One explanation for this is that mitter release. EPSC, excitatory postsynaptic current. the low release probability of glutamate Box 1 | Relationship between Ca2+ and exocytosis expressed at CNS terminals and particularly at mossy fibre boutons34. 2+ 2+ Ca ions are required for exocytosis. However the relationship between the presynaptic Ca The direct contribution of presynaptic 2+ charge (that is, the total amount of Ca ions) and the exocytosis is not linear. At many CNS voltage has been tested at the calyx of Held synapses, the postsynaptic current (PSC) can be described by a power law (PSC = axm, where x is synapse where the release rate increased the presynaptic Ca2+ concentration, a is a constant and m is the power coefficient). The power coefficient is an indicator of the cooperativity of Ca2+ binding to presynaptic Ca2+ sensor only slightly (by 20%) when the presynap- proteins of the vesicle release machinery and varies considerably from synapse to synapse. For tic terminal was depolarized by +80 mV 2+ 47 instance, in hippocampal basket cells, this coefficient is 1.6, indicating relatively low during Ca uncaging . Smaller levels of cooperativity at the output synapses of hippocampal basket cells65, but it may reach 5 at the depolarization (from +10 to +50 mV) were synapse formed by the calyx of Held66. In this latter case, very small variations in presynaptic Ca2+ not tested in this study, so additional experi- concentration resulting from small subthreshold analogue modulation of the spike waveform ments will be necessary to clarify this point. may lead to an extremely large increase in neurotransmitter release. Thus, analogue–digital Other voltage-gated channels such as modulation of synaptic transmission might be extremely heterogeneous among synapses. KV1.1 and KV2.1 interact with presynaptic SNARE proteins and facilitate neurotrans- mission48–50, but it is not yet clear whether (~5%) at this synapse would produce little the subthreshold depolarization, indicating detection of voltage by their voltage sensors vesicle depletion, and thus although more that, at this synapse, facilitation results has a critical role in this facilitation. glutamate is being released (accounting for from the supra-linear summation of intra- the enhanced EPSC), there would be no cellular Ca2+ signals arising from different Do G protein-coupled receptors act as voltage detectable change in the PPR. This hypoth- sources41. sensors at presynaptic terminals? Another esis is plausible but, conversely, relatively In summary, one cannot totally exclude mechanism of voltage-dependent facilita- modest synaptic enhancements, such as that analogue–digital enhancement at tion of exocytosis has been proposed at those produced by LTP (~50%), result in the mossy fibre synapse results from an cholinergic synapses51. Because of charge clear changes in the PPR at the mossy fibre undetectable Ca2+ rise produced by the movement, binding of acetylcholine to the synapse36–38. An alternative explanation, EPreSP that causes supra-linear sum- M2 muscarinic G protein-coupled receptor which is supported by a number of find- mation between conditional (analogue (GPCR) is strongly modulated by membrane ings, is that analogue–digital enhancement depolarization) and spike-evoked Ca2+ potential52. The GPCR is bound to the release at mossy fibre synapses might not entirely signals. However, this solution is still not machinery at rest, producing a tonic inhibi- result from a classical rise in presynaptic entirely satisfying because, in contrast with tion of transmitter release, but is unbound neurotransmitter release. For example, experimental results1, short-term facilita- when there is a depolarized potential53,54 maximal subthreshold depolarization of the tion should be reduced in this case, so other (FIG. 2c). The functional consequence of the mossy fibre terminal in the hippo campus mechanisms must be considered. depolarization produced by the presynaptic produces no detectable change in presyn- spike would be a facilitation of transmitter aptic intracellular Ca2+ signals39. Moreover, Modulation of release by voltage release. A similar mechanism might oper- subthreshold depolarization of the granule Do Ca2+ channels act as voltage sensors in ate at glutamatergic synapses as the kinetics cell soma affects Ca2+ signalling only in presynaptic terminals? Voltage can also of glutamate release at the neuromuscular proximal segments of granule cell axons, induce transmitter release by a direct, Ca2+- junction of the crayfish are also controlled by suggesting that analogue–digital enhance- independent effect on release machinery. GPCRs55, and presynaptic metabotropic glu- ment at more distal mossy fibre boutons Several parallel studies have reported tamate receptors are voltage-sensitive56. The might be largely independent of Ca2+. One facilitation of synaptic transmission that is mossy fibre terminal contains a wide range of may argue that the sensitivity of the fluores- produced by subthreshold depolarization of GPCRs, including type B GABA receptors57, cent Ca2+ probe might not be sufficient to the presynaptic terminal42–44. Furthermore, type 2 and 3 metabotropic glutamate recep- detect very small Ca2+ increases in cellular in a few rare cases, such as in dorsal root tors58 and adenosine A1 receptors59. However, nanodomains39,40. Supporting this view, ganglion neurons, transmitter release is Ca2+- the contribution of GPCRs to analogue– analogue–digital enhancement at the mossy independent but voltage-dependent45. The digital enhancement of transmission at the fibre synapse is partly reduced by buffering underlying mechanisms in this particular mossy fibre synapse remains to be precisely Ca2+ with EGTA1. case remain unknown, but at superior cervi- determined. How could enhanced release be pro- cal ganglion synapses, the synaptic protein It is important to note that even if release duced without any change in the spike- interaction site of CaV2.2 channels that are modulation could result from a direct action evoked Ca2+ transient? The classical view present on the presynaptic membrane and of voltage on Ca2+ channels or GPCRs, the that presynaptic facilitation is necessarily adjacent to synaptic vesicle release sites PPR would still be modulated because vesi- associated with the modulation of the mediates a voltage-dependent facilitation cle depletion would still occur. Therefore, presynaptic Ca2+ signals (BOX 1) has been of transmission46. In this case, presynaptic the lack of PPR modulation observed at convincingly challenged by Felmy and depolarization is likely to be mediated by mossy fibre synapse cannot be satisfactorily 41 co-workers . They showed that a three- the voltage sensor of the CaV2.2 channel and explained by this mechanism. fold increase in synaptic strength can be directly transmitted to the SNARE (soluble induced at the calyx of Held synapse by NSF (N-ethylmaleimide-sensitive fac- Does voltage modulate the ratio of kiss- a prolonged subthreshold depolarization tor) attachment protein receptor) proteins and-run versus full-fusion exocytosis? without any change in Ca2+ during the (FIG. 2c). This mechanism could mediate Whether other molecular mechanisms of presynaptic spike. In this case, Ca2+ concen- analogue–digital enhancement at mossy fibre Ca2+-independent and voltage-dependent tration is raised in the calyx of Held during boutons because CaV2.2 channels are broadly exocytosis contribute to voltage-dependent modulation of the release machinery Dominique Debanne, Andrzej Bialowas and Sylvain 25. Awatramani, G. B., Price, G. D. & Trussell, L. O. Rama are at INSERM, UMR_S 1072, and Aix-Marseille Modulation of transmitter release by presynaptic remains unknown. However, one may resting potential and background calcium levels. Université, UNIS, 13015, Marseille, France. speculate that transmitter release might Neuron 48, 109–121 (2005). Correspondence to D.D. 26. Shu, Y., Yu, Y., Yang, J. & McCormick, D. A. Selective theoretically be enhanced if the relative ratio e-mail: [email protected] control of cortical axonal spikes by a slowly of kiss-and-run versus full-fusion mode of inactivating K+ current. Proc. Natl Acad. Sci. USA 104, 11453–11458 (2007). exocytosis decreases after analogue depo- 27. Shu, Y., Duque, A., Yu, Y., Haider, B. & McCormick, D. A. larization60. The kiss-and-run mode of exo- Properties of action-potential initiation in neocortical 1. Alle, H. & Geiger, J. R. Combined analog and action pyramidal cells: evidence from whole cell axon cytosis corresponds to an incomplete fusion potential coding in hippocampal mossy fibers. Science recordings. J. Neurophysiol. 97, 746–760 (2007). of the presynaptic vesicle by which only a 311, 1290–1293 (2006). 28. Boudkkazi, S., Fronzaroli-Molinieres, L. & Debanne, D. Presynaptic action potential waveform determines 60 2. Shu, Y., Hasenstaub, A., Duque, A., Yu, Y. & small fraction of its content is released . McCormick, D. A. Modulation of intracortical synaptic cortical synaptic latency. J. Physiol. 589, 1117–1131 Thus, compared with the full-fusion mode potentials by presynaptic somatic membrane (2011). potential. Nature 441, 761–765 (2006). 29. Saviane, C., Mohajerani, M. H. & Cherubini, E. An 2+ of release, synaptic transmission is not 3. Kole, M. H., Letzkus, J. J. & Stuart, G. J. Axon initial ID‑like current that is downregulated by Ca modulates maximal because not all postsynaptic recep- segment Kv1 channels control axonal action potential information coding at CA3–CA3 synapses in the rat waveform and synaptic efficacy. Neuron 55, 633–647 hippocampus. J. Physiol. 552, 513–524 (2003). tors are stimulated. Kiss-and-run exocytosis (2007). 30. Foust, A. J., Yu, Y., Popovic, M., Zecevic, D. & has been reported in cultured hippocampal 4. Alle, H. & Geiger, J. R. Analog signalling in mammalian McCormick, D. A. Somatic membrane potential and Kv1 channels control spike repolarization in cortical 61,62 cortical axons. Curr. Opin. Neurobiol. 18, 314–320 neurons , but it has not yet been dem- (2008). axon collaterals and presynaptic boutons. J. Neurosci. onstrated at the mossy fibre synapse. It is 5. Kole, M. H. & Stuart, G. J. Signal processing in the 31, 15490–15498 (2011). axon initial segment. Neuron 73, 235–247 (2012). 31. Yu, Y., Maureira, C., Liu, X. & McCormick, D. P/Q and noteworthy that if neurotransmitter release 6. Werblin, F. S. & Dowling, J. E. Organization of the N channels control baseline and spike-triggered was enhanced in ADF through a decrease retina of the mudpuppy, Necturus maculosus. II. calcium levels in neocortical axons and synaptic Intracellular recording. J. Neurophysiol. 32, 339–355 boutons. J. Neurosci. 30, 11858–11869 (2010). in the kiss-and-run mode of fusion, release (1969). 32. Sasaki, T., Matsuki, N. & Ikegaya, Y. Action-potential probability and the PPR would not change, 7. Marder, E. Neurobiology: extending influence. Nature modulation during axonal conduction. Science 331, 441, 702–703 (2006). 599–601 (2011). as seen at the mossy fibre synapse. 8. de Ruyter van Steveninck, R. R. & Laughlin, S. B. The 33. Debanne, D. & Rama, S. Astrocytes shape axonal The βγ subunits of G proteins are pre- rate of information transfer at graded-potential signaling. Sci. Signal. 4, pe11 (2011). synapses. Nature 379, 642–645 (1996). 34. Li, L., Bischofberger, J. & Jonas, P. Differential gating sent in a wide range of presynaptic termi- 9. Borst, A. & Theunissen, F. E. Information theory and and recruitment of P/Q-, N-, and R‑type Ca2+ channels nals, including mossy fibre synapses, are neural coding. Nature Neurosci. 2, 947–957 (1999). in hippocampal mossy fiber boutons. J. Neurosci. 27, 10. Heidelberger, R. Mechanisms of tonic, graded release: 13420–13429 (2007). thought to influence the opening of the lessons from the vertebrate photoreceptor. J. Physiol. 35. Neher, E. & Sakaba, T. Multiple roles of calcium ions in fusion pore downstream of Ca2+ entry63,64; 585, 663–667 (2007). the regulation of neurotransmitter release. Neuron 11. Debanne, D., Campanac, E., Bialowas, A., Carlier, E. & 59, 861–872 (2008). however, the precise role of voltage in this Alcaraz, G. Axon physiology. Physiol. Rev. 91, 36. Zalutsky, R. A. & Nicoll, R. A. Comparison of two forms process remains unknown. One may specu- 555–602 (2011). of long-term potentiation in single hippocampal 12. Bucher, D. & Goaillard, J. M. Beyond faithful neurons. Science 248, 1619–1624 (1990). late that modifications of the open time of conduction: short-term dynamics, neuromodulation, 37. Salin, P. A., Malenka, R. C. & Nicoll, R. A. Cyclic the fusion pore may not affect the PPR in a and long-term regulation of spike propagation in the AMP mediates a presynaptic form of LTP at axon. Prog. Neurobiol. 94, 307–346 (2011). cerebellar parallel fiber synapses. Neuron 16, classical way. Additional experimental stud- 13. Alle, H., Roth, A. & Geiger, J. R. Energy-efficient action 797–803 (1996). ies will be needed to determine whether potentials in hippocampal mossy fibers. Science 325, 38. Alle, H., Jonas, P. & Geiger, J. R. PTP & LTP at a 1405–1408 (2009). hippocampal mossy fiber–interneuron synapse. Proc. and how subthreshold depolarization may 14. Sengupta, B., Stemmler, M., Laughlin, S. B. & Natl Acad. Sci. USA 98, 14708–14713 (2001). modulate kiss-and-run exocytosis at central Niven, J. E. Action potential energy efficiency varies 39. Scott, R., Ruiz, A., Henneberger, C., Kullmann, D. M. among neuron types in vertebrates and invertebrates. & Rusakov, D. A. Analog modulation of mossy fiber synapses. PLoS Comput. Biol. 6, e1000840 (2010). transmission is uncoupled from changes in presynaptic 15. Shimahara, T. & Tauc, L. Multiple interneuronal Ca2+. J. Neurosci. 28, 7765–7773 (2008). afferents to the giant cells in Aplysia. J. Physiol. 247, 40. Eggermann, E., Bucurenciu, I., Goswami, S. P. & Conclusion 299–319 (1975). Jonas, P. Nanodomain coupling between Ca2+ In summary, synaptic transmission is 16. Shimahara, T. & Peretz, B. Soma potential of an channels and sensors of exocytosis at fast interneurone controls transmitter release in a mammalian synapses. Nature Rev. Neurosci. 13, highly modulated by changes in analogue monosynaptic pathway in Aplysia. Nature 273, 7–21 (2012). voltage resulting from synaptic activity 158–160 (1978). 41. Felmy, F., Neher, E. & Schneggenburger, R. Probing 17. Nicholls, J. & Wallace, B. G. Modulation of the intracellular calcium sensitivity of transmitter that is generated in the dendrites of CNS transmission at an inhibitory synapse in the central release during synaptic facilitation. Neuron 37, neurons. This form of activity-dependent nervous system of the leech. J. Physiol. 281, 801–811 (2003). 157–170 (1978). 42. Cooke, J. D., Okamoto, K. & Quastel, D. M. The role of regulation of synaptic strength is compu- 18. Shapiro, E., Castellucci, V. F. & Kandel, E. R. calcium in depolarization-secretion coupling at the tationally appealing because many corti- Presynaptic membrane potential affects transmitter motor nerve terminal. J. Physiol. 228, 459–497 release in an identified neuron in Aplysia by (1973). cal neurons establish local connections modulating the Ca2+ and K+ currents. Proc. Natl Acad. 43. Llinas, R., Steinberg, I. Z. & Walton, K. Relationship through proximal synaptic inputs. Two Sci. USA 77, 629–633 (1980). between presynaptic calcium current and postsynaptic 19. Zhu, J., Jiang, M., Yang, M., Hou, H. & Shu, Y. potential in squid giant synapse. Biophys. J. 33, principal presynaptic mechanisms have Membrane potential-dependent modulation of 323–351 (1981). been identified in the recent years: a spike- recurrent inhibition in rat neocortex. PLoS Biol. 9, 44. Hochner, B., Parnas, H. & Parnas, I. Membrane e1001032 (2011). depolarization evokes neurotransmitter release in the width-dependent facilitation and a basal 20. Christie, J. M., Chiu, D. N. & Jahr, C. E. Ca2+-dependent absence of calcium entry. Nature 342, 433–435 Ca2+-dependent enhancement. However, enhancement of release by subthreshold somatic (1989). depolarization. Nature Neurosci. 14, 62–68 (2011). 45. Zhang, C. & Zhou, Z. Ca2+-independent but voltage- analogue–digital enhancement of trans- 21. Bouhours, B., Trigo, F. F. & Marty, A. Somatic dependent secretion in mammalian dorsal root mission at mossy fibre synapses may also depolarization enhances GABA release in cerebellar ganglion neurons. Nature Neurosci. 5, 425–430 interneurons via a calcium/protein kinase C pathway. (2002). involve other processes. Alternative mech- J. Neurosci. 31, 5804–5815 (2011). 46. Mochida, S., Yokoyama, C. T., Kim, D. K., Itoh, K. & anisms to the classical Ca2+-induced modu- 22. Debanne, D., Guerineau, N. C., Gahwiler, B. H. & Catterall, W. A. Evidence for a voltage-dependent Thompson, S. M. Action-potential propagation gated enhancement of neurotransmitter release mediated lation of exocytosis have been suggested, + via the synaptic protein interaction site of N‑type Ca2+ by an axonal IA-like K conductance in hippocampus. such as Ca2+-independent but voltage- Nature 389, 286–289 (1997). channels. Proc. Natl Acad. Sci. USA 95, 23. Rall, W. Distributions of potential in cylindrical 14523–14528 (1998). dependent exocytosis. However, the role of coordinates and time constants for a membrane 47. Felmy, F., Neher, E. & Schneggenburger, R. The timing this mechanism remains controversial, and cylinder. Biophys. J. 9, 1509–1541 (1969). of phasic transmitter release is Ca2+-dependent and 24. Sasaki, T., Matsuki, N. & Ikegaya, Y. Effects of axonal lacks a direct influence of presynaptic membrane its precise contribution to analogue–digital topology on the somatic modulation of synaptic potential. Proc. Natl Acad. Sci. USA 100, enhancement has yet to be identified. outputs. J. Neurosci. 32, 2868–2876 (2012). 15200–15205 (2003). 48. Fili, O. et al. Direct interaction of a brain voltage-gated K+ channel with syntaxin 1A: functional impact on channel gating. J. Neurosci. 21, 1964–1974 (2001). 49. Feinshreiber, L., Singer-Lahat, D., Ashery, U. & Lotan, I. Voltage-gated potassium channel as a facilitator of exocytosis. Ann. NY Acad. Sci. 1152, 87–92 (2009). 50. Feinshreiber, L. et al. Non-conducting function of the Kv2.1 channel enables it to recruit vesicles for release in neuroendocrine and nerve cells. J. Cell Sci. 123, 1940–1947 (2010). 51. Parnas, H. & Parnas, I. The chemical synapse goes electric: Ca2+- and voltage-sensitive GPCRs control neurotransmitter release. Trends Neurosci. 30, 54–61 (2007). 52. Kupchik, Y. M. et al. A novel fast mechanism for GPCR-mediated signal transduction — control of neurotransmitter release. J. Cell Biol. 192, 137–151 (2011). 53. Linial, M., Ilouz, N. & Parnas, H. Voltage-dependent interaction between the muscarinic ACh receptor and proteins of the exocytic machinery. J. Physiol. 504, 251–258 (1997). 54. Ilouz, N., Branski, L., Parnis, J., Parnas, H. & Linial, M. Depolarization affects the binding properties of muscarinic acetylcholine receptors and their interaction with proteins of the exocytic apparatus. J. Biol. Chem. 274, 29519–29528 (1999). 55. Kupchik, Y. M. et al. Molecular mechanisms that control initiation and termination of physiological depolarization-evoked transmitter release. Proc. Natl Acad. Sci. USA 105, 4435–4440 (2008). 56. Ohana, L., Barchad, O., Parnas, I. & Parnas, H. The metabotropic glutamate G‑protein-coupled receptors mGluR3 and mGluR1a are voltage-sensitive. J. Biol. Chem. 281, 24204–24215 (2006). 57. Thompson, S. M. & Gahwiler, B. H. Activity-dependent disinhibition. III. Desensitization & GABAB receptor- mediated presynaptic inhibition in the hippocampus in vitro. J. Neurophysiol. 61, 524–533 (1989). 58. Kamiya, H., Shinozaki, H. & Yamamoto, C. Activation of metabotropic glutamate receptor type 2/3 suppresses transmission at rat hippocampal mossy fibre synapses. J. Physiol. 493, 447–455 (1996). 59. Moore, K. A., Nicoll, R. A. & Schmitz, D. Adenosine gates synaptic plasticity at hippocampal mossy fiber synapses. Proc. Natl Acad. Sci. USA 100, 14397–14402 (2003). 60. He, L. & Wu, L. G. The debate on the kiss-and-run fusion at synapses. Trends Neurosci. 30, 447–455 (2007). 61. Richards, D. A. Vesicular release mode shapes the postsynaptic response at hippocampal synapses. J. Physiol. 587, 5073–5080 (2009). 62. Zhang, Q., Li, Y. & Tsien, R. W. The dynamic control of kiss-and-run and vesicular reuse probed with single nanoparticles. Science 323, 1448–1453 (2009). 63. Blackmer, T. et al. G protein βγ subunit-mediated presynaptic inhibition: regulation of exocytotic fusion downstream of Ca2+ entry. Science 292, 293–297 (2001). 64. Chen, X. K. et al. Activation of GPCRs modulates quantal size in chromaffin cells through G and PKC. βγ Nature Neurosci. 8, 1160–1168 (2005). 65. Bucurenciu, I., Bischofberger, J. & Jonas, P. A small number of open Ca2+ channels trigger transmitter release at a central GABAergic synapse. Nature Neurosci. 13, 19–21 (2010). 66. Fedchyshyn, M. J. & Wang, L. Y. Developmental transformation of the release modality at the calyx of Held synapse. J. Neurosci. 25, 4131–4140 (2005).

Acknowledgements The authors are supported by INSERM, Aix-Marseille Université, Centre National de la Recherche Scientifique, Agence Nationale de la Recherche (ANR 11 BSV4 016 01), Fondation pour le Recherche Medicale and the French Ministry of Research. We thank H. Alle, O. El Far, J.-M. Goaillard, V. Marra, M. Seagar and the reviewers for helpful discussion and comments on the manuscript.

Competing interests statement The authors declare no competing financial interests.

FURTHER INFORMATION Dominique Debanne’s homepage: http://www.unis-neuro. com/uk/fiche-membre.php?id=17