Dendritic Inhibitory Synapses Punch above Their Weight

2 downloads 0 Views 448KB Size Report
Aug 5, 2015 - oped into an epic cinematic version— ..... dendrite (0.5 μm diameter) shows decremental backpropagation that fails at 800 μm from the soma.
Neuron

Previews resuming voltage-clamp Cm recording, they find intensity-dependent increases in exocytosis that do not saturate with increases in the AII voltage integral. Together, these data suggest that AII glycinergic output can respond indefinitely to sustained input, allowing the synapse to signal graded inhibition continuously to OFF CBCs. Balakrishnan et al. describe properties of glycinergic transmission by AII lobular dendrites. AIIs are an important hub for information processing in the retina, contacting R28 different retinal cell types including each type of retinal bipolar cell (Marc et al., 2014), thereby shaping cone-driven responses at high light levels and sharing information between rod and cone pathways at lower light levels. The authors find that in many ways, glycine release from lobular dendrites of AIIs behaves surprisingly like transmission by ribbon-bearing synapses, exhibiting a remarkable capability for sustained release from a large pool of vesicles regulated by Ca2+ influx through L-type channels. The present results are consistent

with other data showing that the ability for rapid sustained release is not limited to ribbon synapses (Hallermann and Silver, 2013). These findings fit with an emerging picture of highly specialized mechanisms operating at different synapses to serve diverse signaling demands (O’Rourke et al., 2012) and open the door to future study of how AII exocytotic properties are specifically suited for carrying retinal signals under different signaling regimes.

Habermann, C.J., O’Brien, B.J., Wa¨ssle, H., and Protti, D.A. (2003). J. Neurosci. 23, 6904–6913. Hallermann, S., and Silver, R.A. (2013). Trends Neurosci. 36, 185–194. Manookin, M.B., Beaudoin, D.L., Ernst, Z.R., Flagel, L.J., and Demb, J.B. (2008). J. Neurosci. 28, 4136–4150. Marc, R.E., Anderson, J.R., Jones, B.W., Sigulinsky, C.L., and Lauritzen, J.S. (2014). Front. Neural Circuits 8, 104. O’Rourke, N.A., Weiler, N.C., Micheva, K.D., and Smith, S.J. (2012). Nat. Rev. Neurosci. 13, 365–379.

REFERENCES Abbott, L.F., and Regehr, W.G. (2004). Nature 431, 796–803. Balakrishnan, V., Puthussery, T., Kim, M.-H., Taylor, W.R., and von Gersdorff, H. (2015). Neuron 87, this issue, 563–575. Bloomfield, S.A., and Dacheux, R.F. (2001). Prog. Retin. Eye Res. 20, 351–384. Demb, J.B., and Singer, J.H. (2012). Vis. Neurosci. 29, 51–60. Fedchyshyn, M.J., and Wang, J. Neurosci. 25, 4131–4140.

Goutman, J.D., and Glowatzki, E. (2007). Proc. Natl. Acad. Sci. USA 104, 16341–16346.

L.Y. (2005).

Schubert, T., Kerschensteiner, D., Eggers, E.D., Misgeld, T., Kerschensteiner, M., Lichtman, J.W., Lukasiewicz, P.D., and Wong, R.O.L. (2008). J. Neurophysiol. 100, 304–316. Strettoi, E., Raviola, E., and Dacheux, R.F. (1992). J. Comp. Neurol. 325, 152–168. Thoreson, W.B., Rabl, K., Townes-Anderson, E., and Heidelberger, R. (2004). Neuron 42, 595–605. Wong, A.B., Rutherford, M.A., Gabrielaitis, M., , T., Go¨ttfert, F., Frank, T., Michanski, S., Pangrsic Hell, S., Wolf, F., Wichmann, C., and Moser, T. (2014). EMBO J. 33, 247–264.

Dendritic Inhibitory Synapses Punch above Their Weight Lea Goetz,1 Arnd Roth,1 and Michael Ha¨usser1,* 1Wolfson Institute for Biomedical Research and Department of Neuroscience, Physiology and Pharmacology, University College London, Gower Street, London WC1E 6BT, UK *Correspondence: [email protected] http://dx.doi.org/10.1016/j.neuron.2015.07.018

Mu¨llner et al. (2015) show that single inhibitory synapses placed in the right location on the dendritic tree can exert a powerful impact on backpropagating action potentials in hippocampal pyramidal neurons by controlling local Ca2+ influx with mm and ms precision. In the book The Hobbit—recently developed into an epic cinematic version— the gloriously evil dragon Smaug appears to be invincible, coated by impenetrable scales that cover his entire body. And yet he has a weakness: a tiny bare patch on his chest, not covered by a scale. Bard the Bowman, tipped off by Bilbo Baggins, fires a single arrow

directly into the spot at exactly the right moment and thus manages to slay the dragon. This idea of a ‘‘weak point’’ that renders an apparently invincible foe vulnerable to even a modest weapon runs throughout mythology and literature, most famously the proverbial ‘‘Achilles’ heel.’’ There are also examples to be found in biology, and in this issue

of Neuron, Mu¨llner and colleagues show that even a single inhibitory synapse can ‘‘slay’’ a backpropagating action potential when placed at the right location in the dendritic tree. Understanding the power of single inhibitory synapses is essential if we are to understand the intricate interplay of excitation and inhibition that is the basis

Neuron 87, August 5, 2015 ª2015 Elsevier Inc. 465

Neuron

Previews of synaptic integration in vivo. Inhibitory interneurons play a crucial role in orchestrating ongoing activity patterns in neural circuits (Klausberger and Somogyi, 2008) and in shaping the responses of individual neurons to sensory input, and defects in inhibitory control of circuit activity can cause disease, such as epilepsy. Importantly, inhibition also regulates synaptic plasticity rules, allowing the brain to learn while ensuring that learning-related changes in neural circuits do not lead to unstable activity patterns. To understand how inhibition achieves its diverse roles, it is therefore essential to understand the impact of single inhibitory synapses on synaptic integration and neuronal output. Early theoretical work provided key insights into the importance of both location and timing for the efficacy of inhibitory synapses. Inhibition was shown to be most effective when it is placed ‘‘on-path’’ between an excitatory synapse and the soma (Jack et al., 1975). This enables logical ANDNOT operations if the inhibitory conductance is large, placed between the excitatory synapse and the soma, and if excitation and inhibition overlap in time (Koch et al., 1983). Supporting this mechanism, it was shown experimentally that for excitatory and inhibitory synaptic inputs in pyramidal cell dendrites to interact, they must be located in close vicinity on the same dendritic branch and must also temporally overlap (Liu, 2004). Thus, to harness these capabilities, excitatory and inhibitory synapses need to spatially target individual dendrites in a specific way. Indeed, across many neural circuits, different types of interneurons have been shown to innervate specific subregions of the dendritic tree of their postsynaptic counterparts (Klausberger and Somogyi, 2008). A notable example are the somatostatin-expressing interneurons in the neocortex, which target some of their axonal boutons to individual spine heads, where highly specific interactions between the inhibitory input and an excitatory synapse made on the same spine can occur (Chiu et al., 2013). These findings have defined the rules for interactions between excitatory and inhibitory inputs in the dendritic tree. However, they have not addressed the

issue of whether single interneurons— and in particular, single inhibitory synapses—can be powerful enough to inhibit the action potential, the dominant electrical signal of the neuron that also forms its output. Paired recordings have demonstrated that single presynaptic interneurons can delay spontaneously generated action potentials in Purkinje cells (Ha¨usser and Clark, 1997) and can also delay spikes in pyramidal cells that are close to threshold (Miles et al., 1996). This can act as a mechanism to synchronize multiple pyramidal cells (Cobb et al., 1995). However, these inhibitory effects were generated by interneurons making multiple contacts with the postsynaptic cells. It appears unlikely that single inhibitory contacts will be sufficiently strong to prevent initiation of an action potential. Nevertheless, one possible ‘‘Achilles’ heel’’ for the action potential is to prevent its spread into the dendritic tree after it has been initiated. Voltage-gated channels in the dendritic tree allow the action potential to backpropagate actively into the dendritic tree of many types of neurons (Stuart et al., 1997). However, since dendrites are only weakly excitable, the propagation of the action potential is decremental in most cell types studied so far. Therefore, unlike somatic action potentials, backpropagating action potentials (bAPs) are not all-or-none events but are bidirectionally modifiable by synaptic inhibition (Tsubokawa and Ross, 1996) and excitation (Stuart and Ha¨usser, 2001). This therefore raises the question of whether single inhibitory synapses may be sufficiently powerful to act as a brake on the bAP. This is the question addressed by Mu¨llner et al. (2015). Instead of measuring dendritic action potential amplitude directly, the authors used dendritic Ca2+ signals as a proxy (building on the pioneering work of Tsubokawa and Ross [1996]), since the backpropagating action potential activates voltage-gated Ca2+ channels (Stuart et al., 1997). This imaging approach allowed them to measure the impact of a single inhibitory synapse on a backpropagating action potential and also to measure its footprint in space and time. The approach used by the authors to study the effect of single identified inhibi-

466 Neuron 87, August 5, 2015 ª2015 Elsevier Inc.

tory synapses is technically virtuosic: they made simultaneous whole-cell recordings from a pyramidal neuron and a dendritetargeting GFP-positive interneuron in organotypic hippocampal slice cultures prepared from GAD65-GFP mice and confirmed monosynaptic connections by eliciting unitary IPSCs in the voltageclamped pyramidal cell, followed by anatomical identification of whether the inhibitory connection consisted of just a single synapse or multiple synapses. Next, they activated action potentials in both neurons and imaged the [Ca2+] transient evoked by the bAP in the postsynaptic neuron at and nearby the inhibitory contact. Mu¨llner and colleagues found that activation of a single inhibitory synapse can reduce the bAP-evoked [Ca2+] transient by up to 70%. The level of inhibition depends both on properties of the inhibitory synapses as well as the [Ca2+] transient itself. The level of inhibition by a single synapse depends on the contact area, and the correlation between impact and contact area also holds for multiple nearby synapses. Interestingly, the amount of Ca2+-transient inhibition by a given inhibitory synapse crucially depends on properties of the inhibited bAP itself—in particular, its amplitude at the location of the inhibitory synapse and their relative timing. The authors measured the spatiotemporal profile of Ca2+-transient inhibition at different distances from the activated inhibitory contact and, under the assumption of an exponential decay, derive space constants for Ca2+-transient inhibition—the distance over which inhibition drops to 1/e of its peak value—of 23–25 mm in the proximal and 23–28 mm in the distal direction from the contact site, respectively. These experimental results are supported by simulations using a biophysical model of a CA1 pyramidal cell, which show that the impact of the inhibitory synapse indeed falls off exponentially along the dendrite. Going further, the authors show that the inhibitory effect is not just local but also branch specific: Ca2+-transient inhibition drops more between branches than predicted by the average electrotonic length of those branches. Finally, the authors analyzed the difference between the inhibition seen in the spine head and the

Neuron

Previews A

B

20 mV

20 mV

2 ms

2 ms

200 µm

20 mV

20 mV 200 µm

200 µm

Figure 1. A Simple Model Demonstrates the Vulnerability of Action Potential Backpropagation to Single Inhibitory Inputs (A) A ball-and-stick model of a neuron with an active soma (20 mm diameter) and a weakly excitable dendrite (0.5 mm diameter) shows decremental backpropagation that fails at 800 mm from the soma (black trace). When a single inhibitory synapse (2 nS) is activated at 500 mm from the soma, the action potential fails irreversibly near that point (red trace). The top traces show the action potential as a function of time, recorded at the indicated locations (pipettes at 400, 500, and 600 mm from the soma), in control (black) and with the inhibitory synapse active (red). The bottom traces show the local bAP amplitude with distance from the soma. (B) A weaker inhibitory synapse (orange, 0.5 nS; red, 1 nS; purple, 1.5 nS) can produce graded and reversible local inhibition of the action potential amplitude.

dendritic shaft. While experimentally they find no difference between inhibition of the bAP-evoked [Ca2+] signal on the shaft or spines, their simulations suggest that in the presence of an excitatory input on the spine, an inhibitory input placed on the dendritic shaft can result in a larger Ca2+-transient inhibition in the spine compared to the shaft. Given its high spatial precision, how temporally precise is inhibition by a single inhibitory synapse? To study the timing dependence of Ca2+-transient inhibition, the authors varied the timing between the presynaptic and postsynaptic AP. They find optimal inhibition with zero delay between bAP and inhibitory activation, and their biophysical model suggests that the spike-timing dependence of Ca2+-transient inhibition is a consequence of the fast synaptic kinetics. Interestingly, the temporal profile of Ca2+-transient inhibition displays a familiar shape: it follows the time course of the synaptic current, as has been observed previously for the boosting of bAPs by excitatory

synaptic inputs (Stuart and Ha¨usser, 2001). Notably, both studies find that the effect on the bAP—boosting or inhibiting—depends on the local amplitude of the bAP. This suggests a mechanism, explored by Mu¨llner et al. (2015) in a detailed model of CA1 pyramidal cells and illustrated using a simple ball-andstick model in Figure 1: the single inhibitory synapse hyperpolarizes the local membrane potential to prevent recruitment of voltage-gated channels that are responsible for the regenerative propagation of the bAP along the dendrite. Consequently, [Ca2+] transients triggered by bAPs propagating just above the regenerative threshold can be reduced non-linearly if insufficient channels can be recruited due to the inhibition. This would ‘‘kill’’ the regenerative bAP, rendering its propagation passive and leading to a steep drop in amplitude (Figure 1A). On the other hand, bAPs sufficiently above the threshold for regenerative propagation, which tend to be associated with larger dendritic [Ca2+] transients, will

experience only a temporary and local ‘‘dent’’ in amplitude as they pass the inhibitory synapse (Figure 1B). Finally, combining the spatial and temporal properties of Ca2+-transient inhibition, the authors find configurations where a single inhibitory synapse can paradoxically boost the [Ca2+] transient, presumably as a result of reduced inactivation of voltage-gated calcium and/or sodium channels. The results of Mu¨llner et al. (2015) therefore elegantly show how single inhibitory synapses can ‘‘punch above their weight’’ and have a significant impact even on the apparently invincible action potential: a well-timed inhibitory input, placed in the dendritic tree just where the backpropagating action potential is most vulnerable—namely near where active backpropagation is about to fail—can actually stop the action potential in its tracks. Inhibitory synapses placed elsewhere in the dendritic tree can have more subtle and spatially precise influences on dendritic [Ca2+] signals associated with backpropagating action potentials. These findings dramatically confirm earlier experimental and theoretical work about the importance of the location of inhibitory synapses for their efficacy and complement recent work showing that the interaction of inhibition with active properties of dendrites can have some surprising and counter-intuitive effects, such as ‘‘action at a distance’’ (Gidon and Segev, 2012; Jadi et al., 2012). These results lead to a number of predictions and suggestions for future experiments. First, since the threshold for regenerative propagation of bAPs also depends on dendritic morphology (Vetter et al., 2001), the impact of single inputs will be stronger at dendritic locations where propagation is particularly vulnerable, such as branch points with large impedance mismatches. Thus, inhibitory synapses could be strategically targeted to such locations to maximize their impact on the bAP. Second, bAPs will encounter different conditions in vivo compared to the dendrites of cultured neurons. In future experiments, modulation of AP backpropagation in vivo could be investigated by activation of individual interneurons or specific interneuron populations. For example, a precise activation of individual synaptic contacts could be

Neuron 87, August 5, 2015 ª2015 Elsevier Inc. 467

Neuron

Previews achieved by targeted optogenetic activation of a single interneuron axon, or via GABA uncaging. Finally, the precise inhibition of backpropagating action potentials in dendrites observed by Mu¨llner et al. (2015) highlights the importance of controlling bAPs as a key mechanism in certain forms of synaptic plasticity. In particular, during STDP the backpropagating AP tells dendritic synapses whether they contributed successfully to generating the postsynaptic AP, implementing Hebb’s rule. The findings by Mu¨llner et al. (2015) therefore make an important prediction: synaptic plasticity can be regulated by a single inhibitory synapse. With precise spatial and temporal modulation of Ca2+-transient inhibition, plasticity could be vetoed very locally in individual dendritic branches, facilitating storage of new patterns of synaptic weights in some branches while preventing the weights of synapses on the inhibited dendritic branch from being

overwritten. To test this prediction, future experiments could explore whether plasticity induction can be blocked specifically in dendritic branches contacted by individual inhibitory contacts, but not in nearby branches. Together, these experiments will help to show how single inhibitory synapses, like Bard the Bowman’s arrow, can have an unexpectedly powerful influence on their targets, on both brief and longer timescales.

Jack, J.J.B., Noble, D., and Tsien, R.W. (1975). Electric current flow in excitable cells (Oxford: Oxford UP). Jadi, M., Polsky, A., Schiller, J., and Mel, B.W. (2012). PLoS Comput. Biol. 8, e1002550. Klausberger, T., and Somogyi, P. (2008). Science 321, 53–57. Koch, C., Poggio, T., and Torre, V. (1983). Proc. Natl. Acad. Sci. USA 80, 2799–2802. Liu, G. (2004). Nat. Neurosci. 7, 373–379. Miles, R., To´th, K., Gulya´s, A.I., Ha´jos, N., and Freund, T.F. (1996). Neuron 16, 815–823.

REFERENCES

Mu¨llner, F.E., Wierenga, C.J., and Bonhoeffer, T. (2015). Neuron 87, this issue, 576–589.

Chiu, C.Q., Lur, G., Morse, T.M., Carnevale, N.T., Ellis-Davies, G.C., and Higley, M.J. (2013). Science 340, 759–762.

Stuart, G.J., and Ha¨usser, M. (2001). Nat. Neurosci. 4, 63–71.

Cobb, S.R., Buhl, E.H., Halasy, K., Paulsen, O., and Somogyi, P. (1995). Nature 378, 75–78.

Stuart, G., Spruston, N., Sakmann, B., and Ha¨usser, M. (1997). Trends Neurosci. 20, 125–131.

Gidon, A., and Segev, I. (2012). Neuron 75, 330–341.

Tsubokawa, H., and Ross, W.N. (1996). J. Neurophysiol. 76, 2896–2906.

Ha¨usser, M., and Clark, B.A. (1997). Neuron 19, 665–678.

Vetter, P., Roth, A., and Ha¨usser, M. (2001). J. Neurophysiol. 85, 926–937.

Norepinephrine and Corticotropin-Releasing Hormone: Partners in the Neural Circuits that Underpin Stress and Anxiety Yajie Sun,1 Sarah Hunt,1 and Pankaj Sah1,* 1Queensland Brain Institute, University of Queensland, St. Lucia, QLD 4072, Australia *Correspondence: [email protected] http://dx.doi.org/10.1016/j.neuron.2015.07.022

Norepinephrine and corticotropin-releasing hormone (CRH) have long been implicated in the response to stress. In this issue of Neuron, McCall et al. (2015) show that CRH projections from the central amygdala drive tonic locus coeruleus activity that evokes acute anxiety responses and place aversion. Fear and anxiety describe particular mental states that are accompanied by characteristic features of vigilance, avoidance, and physiological arousal. These features are seen in all animals and have evolved as a mechanism to adapt to aversive or stressful conditions. While clearly having adaptive value, excessive anxiety is maladaptive and can interfere with normal life. The Diagnostic and Statistical Manual of the American Psychiatric Asso-

ciation (DSM) has classified maladaptive anxiety into a number of mental disorders, such as generalized anxiety, panic disorder, and posttraumatic stress (DSM 5, 2013). It is estimated that these disorders affect up to 20% of the population at some time during their lifetime, and the overall health cost to the community is enormous. A growing body of literature has begun to elucidate the neural circuits that mediate both fear and anxiety, and it

468 Neuron 87, August 5, 2015 ª2015 Elsevier Inc.

is clear that these two interrelated states arise from neural circuits that are distinct but which also share common elements. For example, these studies have shown that both fear and anxiety require the amygdalar complex, and recent studies show that they also engage a wide and complex network including the extended amygdala, hippocampus, and lateral septum (Tovote et al., 2015). Studies to date have concentrated on the circuits