|Peter Jonas and Gyorgy Buzsaki (2007), Scholarpedia, 2(9):3286.||doi:10.4249/scholarpedia.3286||revision #91565 [link to/cite this article]|
The concept of inhibition entails several meanings, including interruption or blockade of activity and restriction of activity patterns in both space and time. The importance of inhibition in the brain is aptly illustrated by the fact that in addition to excitatory principal cells, the brain contains diverse classes of specialized inhibitory interneurons that selectively innervate specific parts of the somatodendritic surfaces of principal cells and other interneurons. In the cortex, axon terminals of interneurons release gamma amino butyric acid (GABA) onto their synaptic targets, where the inhibitory action can compete with the excitatory forces brought about by the principal cells. However, inhibitory interneurons do much more than just provide stop signals for excitation. Proper dynamics in neuronal networks can only be maintained if the excitatory forces are counteracted by effective inhibitory forces. With only excitatory cells, it would be difficult to create form or order or secure some autonomy for transiently active groups, the hypothetical "cell assemblies", because in interconnected networks, excitation begets more excitation. Interneurons, by way of their inhibitory actions, provide the necessary autonomy and independence to neighboring principal cells. The functional diversity of principal cells can also be enhanced by the membrane domain-specific actions of GABAergic interneurons. Additionally, the opposing actions of excitation and inhibition often give rise to membrane and network oscillations which, in turn, provide temporal coordination of the messages conveyed by principal cells.
Inhibition is mediated by multiple receptors
Inhibition in the mammalian brain is mediated by two fast transmitters, glycine and gamma-aminobutyric acid (GABA). Glycine is the major inhibitory transmitter in the spinal cord, whereas GABA is more abundant in higher brain regions (e.g. the hippocampus and the neocortex). GABA released from presynaptic terminals activates three different types of receptors, termed GABA receptors (GABARs) type A, B, and C. GABAARs, like GlyRs, are ligand-gated ion channels permeable to Cl-. The probably most abundant GABAAR in the brain has a subunit composition alpha1, beta2, and gamma2 (with a putative stoichiometry of 2 : 2 : 1). GABAARs are not only permeable to Cl-, but also show a small permeability to bicarbonate ions (HCO3-; Kaila et al., 1994). This explains why the synaptic reversal potential is slightly more positive than the Cl- equilibrium potential that can be calculated from the Nernst equation. In contrast, postsynaptic GABABRs are heptahelical receptors coupled to inwardly rectifying K+ channels (Kir channels, e.g. Kir 3.2) via guanosine triphosphate (GTP)-binding proteins (G proteins). Finally, GABACRs are ligand-gated Cl- channels comprised of rho subunits. These receptors are primarily expressed in the retina. The mechanisms of inhibition in the brain are highly diverse. First, "presynaptic" and "postsynaptic" inhibition have to be distinguished. During presynaptic inhibition (classically described in the spinal cord) inhibitory transmitters reduce the amount of glutamate release from adjacent excitatory synapses converging on a given neuron. This can occur via both ionotropic and metabotropic receptors on presynaptic terminals. Excitatory presynaptic terminals not only contain metabotropic receptors for inhibitory transmitters (e.g. GABAB receptors), but also for a variety of other transmitters (such as muscarinic acetylcholine receptors and adenosine receptors). It is thought that these metabotropic receptors inhibit presynaptic Ca2+ channels. As the presynaptic Ca2+ inflow is essential for the vesicular release of neurotransmitters, this reduces the efficacy of the excitatory synapses converging on a given neuron. Thus, the neuron is inhibited indirectly. In contrast, during postsynaptic inhibition, the inhibitory transmitter receptors are located on the postsynaptic neuron. Activation of the receptors leads to an increase in postsynaptic conductance, a change in the membrane potential of the postsynaptic neuron, or a combination of both. Ultimately these effects can lead to the inhibition of action potential generation in the postsynaptic cell. Second, inhibition can be "phasic" or "tonic". Phasic inhibition is a short-lasting inhibition typically generated by the activation of GABAA receptors following action potentials in a presynaptic interneuron. However, there are also more long-lasting forms of inhibition. One form is the activation of GABAB receptors by spillover of GABA caused by GABA release from specialized interneurons. A second form is the "tonic" GABAA conductance activated by ambient GABA in the extracellular space (Farrant and Nusser, 2005). This form of inhibition is mediated by molecularly and functionally specialized GABAA receptors. These receptors contain alpha6 or delta subunits, which display a high affinity for GABA binding. Furthermore, a long-lasting form of inhibition can be mediated by asynchronous or spontaneous vesicular release of GABA.
Finally, "hyperpolarizing" and "shunting" inhibition can be distinguished. As GABAA receptor-mediated inhibition is mediated by Cl- channels, the concentration gradient for Cl- across the cell membrane determines the nature of the inhibitory effect. However, the synaptic reversal potential of GABAA receptor-mediated currents varies widely during development, among cell types, and probably even between different compartments of the same cell. If the synaptic reversal potential is below the resting potential, inhibition will be hyperpolarizing. In contrast, if the synaptic reversal potential is between the resting potential and the threshold for the generation of action potentials, GABAergic synapses will have "shunting" effects (Alger and Nicoll, 1979; Andersen et al., 1980; Gulledge and Stuart, 2003; Fig. 1). If the synaptic reversal potential is above the action potential threshold, GABAergic synapses can be even excitatory. This scenario is found early in development (Ben-Ari, 2002). The concentration gradient for Cl- is, in turn, determined by the balance between the activity of the Na+ - K+ - 2 Cl- cotransporter (NKCC), which pumps Cl- into the neuron, and the K+ - Cl- cotransporter (KCC), which pumps Cl- out of the cell. The driving force for the two transporters is provided by the concentration gradients for the co-transported ions, Na+ and K+. Shunting inhibition is termed "shunting" because the synaptic conductance short-circuits currents that are generated at adjacent excitatory synapses. If a shunting inhibitory synapse is activated, the input resistance is reduced locally and, following Ohm’s law, the amplitude of subsequent excitatory postsynaptic potentials (EPSPs) is reduced. This simple scenario arises if the synaptic reversal potential is identical to the resting potential (Fig. 1C). More generally, however, the synaptic reversal potential lies between the resting potential and the action potential threshold. Under these conditions, the effects of shunting inhibition are more complex, consisting of two temporal phases (Fig. 1D). In the first phase, the conductance dominates, leading to inhibition of the postsynaptic cell. In the second phase, the conductance has decayed, but the membrane potential remains depolarized. Thus, the input resistance is high and the membrane potential is shifted towards the threshold for action potential initiation, implying that excitability is increased. Shunting inhibition also has unique spatial properties. The conductance-dominated phase is spatially focal. Consequently, efficient control of action potential generation requires that the shunting synapse is close to the site of action potential initiation, which is presumably the axon initial segment. In contrast, the depolarizing component spreads over a much longer distance, which is determined by the length constant of the dendritic cable of the neuron. Therefore, if an inhibitory synapse from a dendrite-targeting interneuron is located on the distal dendrites of a target cell, only the second depolarizing component propagates to the soma, and only this component is relevant for the control of spiking. Shunting synapses are more effective if the inhibitory synapse is on-path between an excitatory synapse and the action potential initiation site. In addition, the effects of shunting inhibition may be activity-dependent. If GABAergic synapses are activated repetitively, Cl- can accumulate in the postsynaptic cell, and the depolarizing phase of shunting inhibition will become more accentuated (Kaila, 1994).
Feed-forward, feedback and other network forms of inhibition
Networks built from both excitatory and inhibitory elements can self-organize and generate complex properties, the understanding of which is a subject of intense research. However, even in the simplest pairing involving a principal cell and an interneuron, the pattern of firing depends on the exact wiring scheme (Fig. 2). In a feedback or recurrent inhibitory circuit, increased firing of the principal cell elevates the interneuron’s discharge frequency which, in turn, may decrease the principal cell’s output, providing a regulatory mechanism somewhat analogous to the action of a thermostat. In a feed-forward inhibitory configuration, increased discharge of the interneuron, as the primary event, results in the decreased activity of the principal cell. Such simple pairing of excitation and inhibition can substantially increase the temporal precision of firing (Buzsáki, 1984); depolarization of the principal cell, initiated by the excitatory input, is reduced quickly by the repolarizing or shunting effect of feed-forward inhibition, thereby narrowing the temporal window of non-zero discharge probability (Pouille and Scanziani, 2001). Any departure from the simple feedback or feed-forward partnership inevitably increases the complexity of the firing patterns in the participating cells. For example, when two interneurons are activated simultaneously, their combined effect on the target principal cell depends primarily on the interaction between the interneurons, which can be non-linear and hard-to-predict. An extension of feedback inhibition is lateral inhibition. This occurs when the activation of a principal cell recruits an interneuron, which, in turn, suppresses the activity of surrounding principal cells. Suppose that two principal cells are excited by the same input but that the input to principal cell A is stronger than the input to principal cell B. If neuron A and B share a common inhibitory interneuron, the sustained activation of the interneuron by a spike train of neuron A may prevent the spiking of neuron B. A similar outcome is expected if the inputs strengths to neurons A and B are equal but the interneuron-principal cell B synapse is slightly stronger than the interneuron-principal cell A connection. The initial minor difference in the inputs results in a large difference in the outputs of the two neurons. The same asymmetry can be produced if the input to neuron A arrives slightly earlier than the input to B. These static illustrations become more meaningful if one takes into account that synapses are dynamic (i.e., their strength can increase or decrease with repetition; Markram et al., 2004; Watts and Thomson, 2005). Such increased autonomy by competition is also known as "winner-take-all", a non-linear selection or segregation mechanism. In general terms, the inhibitory interneuron system plays a critical role in determining the non-linearity and functional complexity of cortical networks.
Inhibitory interneurons provide spatio-temporal coordination in cortical networks
Some basic functions accomplished by neuronal networks are pattern completion and pattern separation, functions that are related to integration and differentiation. Separation of inputs is difficult in a network with only excitatory connections. However, with inhibitory connections, the competing cell assemblies and even neighboring excitatory neurons can be functionally isolated and excitatory paths can be re-routed by the traffic-controlling ability of coordinated interneuron groups (Fig. 3). The specific firing patterns of principal cells in a network will depend largely on the temporal and spatial distribution of inhibition. As a result, in response to the same input, the same network can potentially produce several different output patterns at different times, depending on the state of inhibition. Coordinated inhibition can ensure that excitatory activity recruits the right numbers of neurons in the right temporal window and that excitation spreads in the right direction. These important features of cortical processing could not be achieved efficiently by principal cells acting alone. Furthermore, the rivalry between excitatory and inhibitory neurons ensures the stability of global neuronal firing rates over extended territories of the cortex, and yet also allows for dramatic increases of local excitability in short time windows, something which is necessary for sending messages and modifying network connections. Competition between opposing forces and feedback control by inhibition are also essential principles for oscillations; interneuron networks are the backbone of many brain oscillators (Buzsáki and Chrobak, 1995; Salinas and Sejnowski, 2001; Engel et al., 2001).
Interneuron diversity multiplies the computational ability of principal cells
Brain systems utilizing mostly locally organized circuits and parallel computational, such as the cerebellum or basal ganglia, evolved only a few neurons types. In contrast, cortical structures specialized for distributed or global computation have evolved several types of principal cells and numerous classes of GABAergic inhibitory interneurons (Freund and Buzsáki, 1996; Somogyi and Klausberger, 2005; Markram et al., 2005). The addition of qualitatively different interneuron types to the same network, even in small numbers, offers a dramatic expansion of computational possibilities. Virtually every segment of the somatodendritic surface of cortical principal cells is under the specific control of a unique interneuron class, and often, multiple classes of interneurons target the same domain (such as the soma). This is a clever way of multiplying the functional repertoire of principal cells, using mostly local interneuron wiring. Neurons with different levels of complexity in their dendritic arbors or neurons with similar geometry but different distribution of ion channels generate uniquely different outputs in response to the same input. So one way evolution can lead to functional diversity through the generation of numerous principal cell types. However, dividing the full computational power of principal cells into numerous subroutines that can be flexibly used according to momentary needs would present an enormous advantage. This important service is provided with ease by the interneuron system. For example, interneurons can functionally "eliminate" a dendritic segment or a whole dendrite, selectively deactivate Ca2+ channels, or segregate dendrites from the soma or the soma from the axon (Miles et al., 1996; Buzsáki et al., 1996; Tsubokawa and Ross, 1996). Such actions of interneurons are functionally equivalent to replacing a principal cell by a morphologically different type. Although researchers agree on the rich diversity of inhibitory interneurons, to date, no widely acceptable taxonomy exists (http://www.columbia.edu/cu/biology/faculty/yuste/petilla/petilla-webpages/petilla.html). Novel interneuron types are being discovered with accelerated speed. Because the main function of interneurons is to control the activity of principal cells, one classification scheme divides them according their axonal targets. Accordingly, three or four first major divisions can be made. (1) The perisomatic group. The first and largest family of interneurons, basket cells and axo-axonic (or chandelier) cells, controls the spiking output of principal cells by providing perisomatic inhibition. (2) The dendrite-targeting group. Interneurons in this family target specific dendritic domains of principal cells. Every known excitatory pathway in the cortex has a matching family of interneurons, which innervates the same dendritic domain . Several additional subclasses in this group seek out two or more (overlapping or non-overlapping) dendritic regions, and yet other subclasses innervate the somata and nearby dendrites with similar probability. The different domains of principal cells have different functional dynamics, and interneurons innervating these specific domains appear to effectively and specifically control the kinetic properties of their target domains (e.g. by suppressing the activity of Ca2+ channels (Miles et al., 1996)). Not surprisingly, members of the dendrite-targeting interneuron family display large variability. (3) The Interneuron-specific group. These interneurons have the distinguishing characteristic that their axons preferentially contact other interneurons but avoid principal cells. (4) The long-range group. Members of this morphologically diverse group have axon trees that span two or more anatomical brain regions. Some axon collaterals of long-range interneurons even cross the hemispheric midline and/or innervate subcortical structures. Their large-caliber axons provide fast communication between the innervated areas. Since this group of inhibitory cells projects over large distances, the term 'interneuron' is not strictly accurate. Nevertheless, for historical reasons all GABAergic cells in the cerebral cortex are referred to as “inhibitory interneurons”. With perhaps 20 or more distinguished interneuron types in the cerebral cortex, the complexity of their wiring must be enormous, although neither the critical details nor the proportions of the different classes of interneurons in various cortical regions are currently known (Freund and Buzsáki, 1996; Markram et al., 2004; Somogyi and Klausberger, 2005). Despite being maximally sensitized to external perturbations, neuronal networks with multiple levels of excitatory and inhibitory constituents are resilient systems, capable of absorbing large external effects without undergoing functional breakdown. Alterations of these interactions, however, may result in epilepsy and various forms of psychiatric diseases.
The interneuron system as a distributed clock
Despite its multifarious wiring, the principal cell system alone cannot carry out many useful computations. The inhibitory neuronal network, when coupled to the principal cells, provides the flexibility needed for the complex operations of the brain. Competition between opposing forces, such as excitation and inhibition, often gives rise to rhythmic behavior. Providing rhythm-based timing to principal cells is one of the most important roles of interneurons. Once a collective oscillatory pattern arises, it constrains the timing freedom of its members and decreases the windows of opportunity for the principal cells to discharge; as a result, principal cells become synchronized. Synchronization by oscillation occurs at multiple time scales, covering time epochs from tens of seconds to milliseconds. The duration of the oscillation, in turn, regulates the length of messages that can be transmitted, as well as the spatial extent of the involved neuronal pools. By way of oscillations, inhibition can create multiple temporal and spatial organizations of principal cells in the cerebral cortex. Among the interneuron classes, basket cells that express the Ca2+ binding protein parvalbumin play a particularly important role in the generation of network oscillations, especially for oscillations in the gamma frequency range (30 – 80 Hz). Basket cells are highly active during gamma activity and fire action potentials that are precisely phase-locked to the oscillations. Computational analysis has revealed that network models of mutually connected basket interneurons can generate synchronized oscillations if exposed to a tonic excitatory drive. The high abundance, locally extensive axonal arborization, and fast signaling properties of parvalbumin-expressing basket cells are very suitable for the generation of fast network oscillations. In a simple analogy, the interneuron network acts as a clock, providing a timing signal to the principal cell ensemble. However, basket cells have spatially confined axon arbors and thus can generate only local synchronization. But how can neuronal populations be synchronized over larger distances, perhaps even across hemispheres? Spatially widespread synchrony can be brought about by region-spanning axon collaterals of principal cells or by the family of long-range interneurons. The latter interneuron type can perhaps provide the necessary fast conduit for synchronizing distantly operating oscillators and allow for coherent timing of a large number of principal neurons that are not connected directly with each other (Buzsáki and Chrobak, 1995; Bartos et al., 2007). Understanding the synaptic, molecular mechanisms of inhibition and deciphering how these elementary processes contribute to the complex dynamics of neuronal networks remain an important agenda for future research.
Alger B.E. and Nicoll R.A. (1979) GABA-mediated biphasic inhibitory responses in hippocampus. Nature 281:315-317.
Andersen P., Dingledine R., Gjerstad L., Langmoen I.A., Laursen A.M. (1980) Two different responses of hippocampal pyramidal cells to application of gamma-amino butyric acid. J. Physiol. 305:279-296.
Bartos M., Vida I., Jonas P. (2007) Synaptic mechanisms of synchronized gamma oscillations in inhibitory interneuron networks. Nat. Rev. Neurosci. 8:45-56.
Ben-Ari Y. (2002) Excitatory actions of GABA during development: the nature of the nurture. Nat. Rev. Neurosci. 3:728-739.
Buzsáki G. (1984) Feed-forward inhibition in the hippocampal formation. Prog. Neurobiol. 22:131-153.
Buzsáki G. (2005) Similar is different in hippocampal networks. Science 309:568-569.
Buzsáki G. and Chrobak J.J. (1995) Temporal structure in spatially organized neuronal ensembles: a role for interneuronal networks. Curr. Opin. Neurobiol. 5:504-510.
Buzsáki G., Penttonen M., Nadasdy Z., Bragin A. (1996) Pattern and inhibition-dependent invasion of pyramidal cell dendrites by fast spikes in the hippocampus in vivo. Proc Natl Acad Sci U S A 93:9921-9925.
Engel A.K., Fries P., Singer W. (2001) Dynamic predictions: oscillations and synchrony in top-down processing. Nat. Rev. Neurosci. 2:704-716.
Farrant M, Nusser Z. (2005) Variations on an inhibitory theme: phasic and tonic activation of GABAA receptors. Nat. Rev. Neurosci. 6:215-229.
Freund T.F. and Buzsáki G. (1996) Interneurons of the hippocampus. Hippocampus 6:347-470.
Gulledge A.T. and Stuart G.J. (2003) Excitatory actions of GABA in the cortex. Neuron 37:299-309.
Kaila K. (1994) Ionic basis of GABAA receptor channel function in the nervous system. Prog. Neurobiol. 42:489-537.
Markram H., Toledo-Rodriguez M., Wang Y., Gupta A., Silberberg G., Wu C. (2004) Interneurons of the neocortical inhibitory system. Nat. Rev. Neurosci. 5:793-807.
Miles R., Toth K., Gulyas A.I., Hajos N., Freund T.F. (1996) Differences between somatic and dendritic inhibition in the hippocampus. Neuron 16:815-823.
Pouille F. and Scanziani M. (2001) Enforcement of temporal fidelity in pyramidal cells by somatic feed-forward inhibition. Science 293:1159-1163.
Salinas E. and Sejnowski T.J. (2001) Correlated neuronal activity and the flow of neural information. Nat. Rev. Neurosci. 22:539-550.
Somogyi P. and Klausberger T. (2005) Defined types of cortical interneurone structure space and spike timing in the hippocampus. J. Physiol. 562:9-26.
Tsubokawa H. and Ross W.N. (1996) IPSPs modulate spike backpropagation and associated [Ca2+]i changes in the dendrites of hippocampal CA1 pyramidal neurons. J Neurophysiol 76:2896-2906.
Vida I., Bartos M., Jonas P. (2006) Shunting inhibition improves robustness of gamma oscillations in hippocampal interneuron networks by homogenizing firing rates. Neuron 49:107-117.
Watts J, Thomson AM (2005) Excitatory and inhibitory connections show selectivity in the neocortex. J Physiol 562:89-97.
- Peter Redgrave (2007) Basal ganglia. Scholarpedia, 2(6):1825.
- Valentino Braitenberg (2007) Brain. Scholarpedia, 2(11):2918.
- Olaf Sporns (2007) Complexity. Scholarpedia, 2(10):1623.
- James Meiss (2007) Dynamical systems. Scholarpedia, 2(2):1629.
- Eugene M. Izhikevich (2007) Equilibrium. Scholarpedia, 2(10):2014.
- Eugene Roberts (2007) Gamma-aminobutyric acid. Scholarpedia, 2(10):3356.
- Jeff Moehlis, Kresimir Josic, Eric T. Shea-Brown (2006) Periodic orbit. Scholarpedia, 1(7):1358.
- John Dowling (2007) Retina. Scholarpedia, 2(12):3487.
- Philip Holmes and Eric T. Shea-Brown (2006) Stability. Scholarpedia, 1(10):1838.
- Arkady Pikovsky and Michael Rosenblum (2007) Synchronization. Scholarpedia, 2(12):1459.