Organization and Plasticity of Cortical Inhibition
Abstract and Keywords
Plasticity of inhibitory synapses keeps inhibition in balance and in register with excitation when changes occur in excitatory synapses. Inhibition has many functions to perform, and there are many kinds of inhibitory neurons to perform various computations and regulate network activity. Different forms of long-term changes in inhibitory synapses have been demonstrated that depend on neural activity. Inhibitory plasticity appears to be partly responsible for the specificity of the inhibitory connections needed to carry out some inhibitory functions. The evolving story of cortical inhibitory plasticity shows that different types of inhibitory interneurons play different roles in a variety of inhibitory functions, that several types of inhibitory plasticity have been attested, and that different forms of plasticity can be expected to have different effects on the organization and specificity of inhibitory connections.
Once it was understood that excitatory connections can be modified by experience, it could be presumed that inhibitory connections ought also to be plastic in order to keep excitation and inhibition in balance and in register when long-term synaptic modifications occur (Froemke, 2015). Plasticity of excitatory connections was predicted (Hebb, 1949; Markram, Gerstner, & Sjöström, 2011) decades before it was confirmed by electrophysiological recordings (Bliss & Lømo, 1973) and by changes in the functional organization of the cortex after sensory deprivation as assessed by single-unit recording (Wiesel & Hubel, 1963), autoradiography (Shatz & Stryker, 1978), and retrograde tracers (Lowel & Singer, 1992). Prior to the advent of molecular genetics approaches to examining different inhibitory subtypes (Vogels et al., 2013), plasticity of inhibitory connections was much more difficult to demonstrate.
Diversity of interneuron form and function. Inhibitory interneurons have diverse origins, locations, shapes, connections, biochemical properties, and behaviors (Fishell & Rudy, 2011; Kubota, Karube, Nomura, & Kawaguchi, 2016; Markram et al., 2004; Rudy, Fishell, Lee, & Hjerling-Leffler, 2011). More than a dozen different types have been described in the neocortex (Fishell & Rudy, 2011; Kubota et al., 2016; Figure 1). Each type has characteristic postsynaptic targets. Some prefer the axon hillock, the soma and proximal dendrites, the apical shaft, or the distal dendritic tuft of pyramidal cells; others prefer inhibitory interneurons (Kubota et al., 2016).
The proper adjustment and calibration of different types of inhibitory interneurons may require different kinds of plasticity (Vogels et al., 2013). It seems likely that their diversity reflects a variety of functional roles (Kepecs & Fishell, 2014), such as these:
• Development. Inhibitory interneurons influence developmental processes in the brain (Gandhi, Yanagawa, & Stryker, 2008; Griffen & Maffei, 2014; Maffei, Lambo, & Turrigiano, 2010; Sur & Leamey, 2001; Trachtenberg, 2015; Van Hooser, Escobar, Maffei, & Miller, 2014; Watt & Desai, 2010).
• Balance. Inhibition must be calibrated so as to balance excitation (Abeles, 1991; Froemke, 2015; Isaacson & Scanziani, 2011; Pozo & Goda, 2010; Vogels, Sprekeler, Zenke, Clopath, & Gerstner, 2011; Xue, Atallah, & Scanziani, 2014); too much inhibition can lead to coma, too little can result in seizures. A fine-scale balance between excitation and inhibition may improve the ability of the cortex to encode information efficiently (Denève & Machens, 2016).
• Scaling. Inhibition helps normalize response levels, maintaining the brain within a useful operating range while the intensity of sensory stimulation ranges over several orders of magnitude (Bortone, Olsen, & Scanziani, 2014; Fino, Packer, & Yuste, 2013; Griffen & Maffei, 2014; Isaacson & Scanziani, 2011; Pozo & Goda, 2010; Schäfer, Vasilaki, & Senn, 2007).
• Adaptation. Sensory neurons adapt to repetitive stimulation. The response to a repeated stimulus is generally weaker than the response to a novel stimulus. This stimulus specific adaptation depends partly on inhibition (Natan et al., 2015).
• Dendritic regulation. Inhibition regulates the integration of excitatory events between dendrites and the axon hillock (Hawkins & Ahmad, 2016; Larkum, 2013; Palmer, Murayama, & Larkum, 2012; Sandler, Shulman, & Schiller, 2016; Schäfer et al., 2007; Sjöström, Rancz, Roth, & Häusser, 2008).
• Timing. In many cortical circuits, inhibitory elements regulate the timing of responses to inputs and the frequency of rhythmic activity (Denève & Machens, 2016; Hawkins & Ahmad, 2016; Isaacson & Scanziani, 2011; Savin, 2014; VanRullen & Thorpe, 2002).
• Learning and plasticity. Inhibition controls learning and gates long-term excitatory plasticity (Froemke, Merzenich, & Schreiner, 2007; Kuchibhotla et al., 2017; Letzkus et al., 2011; Letzkus, Wolff, & Lüthi, 2015), and changes in inhibition are part of what is learned (Froemke et al., 2007; Jasinska et al., 2010; Schäfer et al., 2007; Scheyltjens & Arckens, 2016).
• Response specificity. Many cortical neurons fire action potentials only in response to a small fraction of the excitatory inputs they receive. This response specificity, or sparseness of response, depends in part on inhibition (Barth & Poulet, 2012; Petersen & Crochet, 2013; Wolfe, Houweling, & Brecht, 2010). If inhibition is decreased, cortical cells will generally respond to a wider range of excitatory inputs (Crook & Eysel, 1992; Crook, Kisvárday, & Eysel, 1997, 1998; Griffen & Maffei, 2014; Karnani, Agetsuma, & Yuste, 2014; Lee et al., 2012; Merchant, de Lafuente, Pena-Ortega, & Larriva-Sahd, 2012; Pluta et al., 2015; Sillito, 1977; Wu, Arbuckle, Liu, Tao, & Zhang, 2008). Inhibition plays an important role in determining the response of neurons to input stimuli and the structure of their receptive fields (Crook & Eysel, 1992; Crook, Kisvárday, & Eysel, 1997, 1998; Griffen & Maffei, 2014; Gur, Kagan, & Snodderly, 2005; Kisvárday et al., 2002; Lee et al., 2012; Mao et al., 2012; Martinez, Alonso, Clay Reid, & Hirsch, 2002; Monier, Chavane, Baudot, Graham, & Frégnac, 2003; Palmer et al., 2012; Priebe & Ferster, 2005, 2008; Runyan et al., 2010; Shapley, Hawken, & Ringach, 2003; Wu et al., 2008). Since excitatory plasticity can alter the receptive fields and stimulus selectivity of sensory neurons, corresponding changes in inhibition might be required to maintain certain functions or computations of neural circuits while permitting a degree of flexibility or adaptation.
How do inhibitory connections acquire their specificity? Some of these functions could be carried out by interneurons that connect at random with a large fraction of their neighbors—for example, in the upper layers of mouse frontal cortex, each pyramidal cell is contacted by about three-quarters of the Martinotti cells within a radius of 200 microns of its soma (Fino & Yuste, 2011; Fino, Packer, & Yuste 2013). Other functions appear to require inhibition by interneurons that have a high degree of specificity in contacting their targets—for example, large basket cells in visual cortex extend their distal axons primarily into columns where the preferred orientation is nearly perpendicular to the basket cell’s own preferred orientation or the preferred direction of movement is opposite to the basket cell’s own preference (Kisvárday et al., 2002).
Learning is believed to require specific changes in neuronal connections that depend on patterns of neural activity. Neural networks learn to associate particular circumstances with the firing of particular cells. According to Hebb’s rule (Hebb, 1949) for excitatory connections, “neurons wire together if they fire together” (Lowel & Singer, 1992). As a result, once an important or useful association between a set of circumstances and the firing of a cell has occurred repeatedly, structural and functional changes take place in the excitatory connections that caused the cell to fire so that it becomes easier to associate firing of the cell with the same circumstances again. This captures an important aspect of learning. The Hebbian changes increase the likelihood that a cell will fire under the right circumstances. They do not reduce the likelihood of its firing under the wrong circumstances.
One way to prevent a cell from firing under the wrong circumstances, which the brain almost certainly uses in some situations, is to make excitatory synapses compete (Medini, 2014; Trachtenberg, 2015; von der Malsburg, 1973), so that the stronger ones drive out the weaker ones. Another would be to allow inhibition to weaken inappropriate associations according to an anti-Hebbian rule: If an inhibitory interneuron and its target fire together, its synapse on the target becomes weaker, or is prevented from becoming stronger. This anti-Hebbian rule would weaken the inhibitory synapses on a cell that are active under circumstances when it fires. If inhibitory synapses compete, that would allow inhibitory synapses that are active under other circumstances to become stronger. They might prevent the cell from firing under circumstances in which it should not. Other functions of inhibition may require different rules governing the effects of experience on synaptic change for other sets of synapses (Larsen & Sjöström, 2015).
Some inhibitory synapses have been shown to change strength according to Hebbian rules while others have been shown to change strength according to anti-Hebbian rules (Haas, Nowotny, & Abarbanel, 2006; Lamsa, Heeroma, Somogyi, Rusakov, & Kullmann, 2007; Maffei, Nataraj, Nelson, & Turrigiano, 2006; Woodin, Ganguly, & Poo, 2003). The two kinds of rules have different functional consequences, appropriate for different purposes. The Hebbian rule for inhibitory synapses (cells that fire together wire together) tends to maintain a quantitative balance between excitation and inhibition. According to a Hebbian rule for inhibitory synapses, a cell that receives more excitation also receives more inhibition, tending to keep excitation and inhibition in balance. The anti-Hebbian rule for inhibitory synapses (cells that fire together are less likely to wire together) tends to sharpen the preference of a cell for firing in just those circumstances that have caused it to fire in the past, increasing the likelihood that it will fire in response to circumstances that have been associated with frequent or useful past activity, while decreasing the likelihood that it will fire in other circumstances. In principle, some of the inhibitory synapses on a cell might be subject to a Hebbian rule, balancing inhibition with excitation, while others might be subject to an anti-Hebbian rule, enhancing the specificity of the cell for firing in a particular set of circumstances.
Four key questions:
• Which types of inhibitory interneurons are involved in implementing a given function?
• How specific are the inhibitory connections that implement a given function?
• Which rules govern changes in the strength of inhibitory connections? Hebbian rules? Anti-Hebbian rules?
• If a function requires specific connections, how do they acquire their specificity?
Varieties of Inhibitory Plasticity
Synaptic plasticity can occur at multiple different levels, from structural changes in anatomical connections between axons and dendrites, to functional changes in synaptic weights due to adjustment of receptor expression or transmitter release probability.
Anatomical and Morphological Plasticity
Injury, sensory deprivation, and learning cause turnover in inhibitory synaptic contacts, and injury has also been shown to induce axonal sprouting in both excitatory and inhibitory interneurons.
Growth into and out of a lesioned zone. Retinal lesions in adult macaques induce new growth of the axons of cortical neurons into and out of the lesion projection zone (LPZ) to which the lesioned portion of the retina formerly projected (Marik, Yamahachi, Meyer zum Alten Borgloh, & Gilbert, 2014). Pruning and regrowth begins within hours after the lesion and persists for weeks. Excitatory axons originating in the region surrounding the LPZ grow over distances of hundreds of microns into the LPZ, while inhibitory axons originating in the LPZ grow several hundred microns into the surrounding area. Many of these inhibitory axons probably come from basket cells, which typically project their axons horizontally.
Sensory deprivation changes the rate of synaptic turnover. The dendrites of layer 2/3 (L2/3) pyramidal cells in binocular visual cortex (V1) were observed with 2-photon microscopy in vivo in adult mice both before and after monocular occlusion (Chen et al., 2012). Over an eight-day period of normal vision, approximately 5% of inhibitory synapses on dendritic shafts were either created or eliminated, while approximately 18% of the inhibitory synapses on dendritic spines turned over. During the eight days following the onset of monocular deprivation, the rate of inhibitory dendritic shaft synapse loss doubled, while the rate of creation of new inhibitory dendritic shaft synapses decreased. Changes in the rate of turnover of inhibitory synapses on dendritic spines were in the same direction, but smaller in magnitude. Additions or eliminations of inhibitory synapses tended to cluster within 10 microns of dendritic spines that were either added or eliminated. This clustering may occur around sites of calcium influx.
Pavlovian conditioning can induce synapse formation. Adult mice were subjected to Pavlovian conditioning in which strokes to one of the mystacial vibrissae were paired with a mild tail shock (Jasinska et al., 2010). In L4 of barrel cortex of the conditioned mice, the number of inhibitory synapses on double-synapse dendritic spines (one excitatory synapse and one inhibitory synapse on a single spine) roughly tripled in comparison with control mice. Neither stroking the vibrissa alone nor pseudo-conditioning (in which vibrissa strokes and tail shocks were randomly paired) caused a change in the density of double-synapse spines. However, pseudo-conditioning, unlike conditioning, increased the density of spines with only a single excitatory synapse.
Changes in Inhibitory Synaptic Strength and Efficacy
In addition to forms of structural plasticity such as synaptogenesis or synaptic pruning, the strengths of existing inhibitory synapses can be regulated by activity. Some morphologically identifiable synapses are “silent” (physiologically inactive) until they are awakened by long-term potentiation (LTP; Citri & Malenka, 2008). The strength of a synapse can be changed both by pre- and postsynaptic activity (Kullmann, Moreau, Bakiri, & Nicholson, 2012). The nature of the changes depends on which kind of cell is presynaptic, which kind is postsynaptic (Larsen & Sjöström, 2015; Lu, Li, Zhao, Poo, & Zhang, 2007; Ma, Hu, & Agmon, 2012), on the location of the synapse on the dendritic tree (Froemke, Letzkus, Kampa, Hang, & Stuart, 2010), the detailed timing of presynaptic and postsynaptic activity (D’amour & Froemke, 2015; Froemke & Dan, 2002; Hebb, 1949; Lamsa, Kullmann, & Woodin, 2010; Vogels et al., 2013), and the activity of neuromodulators such as acetylcholine, serotonin, noradrenaline, and oxytocin (Huang, Huganir, & Kirkwood, 2013; Kruglikov & Rudy, 2008; Kuchibhotla et al., 2017; Lisman, Grace, & Duzel, 2011; Marlin, Mitre, D’amour, Chao, & Froemke, 2015).
Spike-timing dependent plasticity (STDP). STDP depends on the relative timing of presynaptic and postsynaptic activity (Citri & Malenka, 2008; Hebb, 1949; Markram, Gerstner, & Sjöström, 2012; Sjöström & Gerstner, 2010). Hebb originally postulated that an excitatory synapse would be strengthened when the presynaptic cell repetitively activated the postsynaptic cell (Hebb, 1949), but it was subsequently realized that some countervailing form of plasticity must prevent an outcome in which all synapses are potentiated to their maximum strength (Markram et al., 2011; Rochester, Holland, Haibt, & Duda, 1956; Stent, 1973; von der Malsburg, 1973). Glutamatergic synapses mediated by NMDA receptors (NMDARs) exhibit a form of STDP (Citri & Malenka, 2008) in which LTP occurs when the presynaptic cell fires shortly before the postsynaptic cell (pre-post pairing), but long-term depression (LTD) occurs when the postsynaptic cell fires shortly before the presynaptic cell (post-pre pairing). NMDA-dependent LTP can be induced when the postsynaptic cell is depolarized, even if its membrane potential does not reach spiking threshold (Sjöström & Gerstner, 2010). Once a pattern of excitatory input induces LTP, it will be easier for the same pattern to depolarize the cell on a subsequent occasion. This is now considered classic Hebbian STDP. There are also anti-Hebbian forms of STDP in which pre-post pairing induces LTD (Woodin et al., 2003), in which concurrent pre- and postsynaptic firing prevents LTP (Maffei et al., 2006), or in which LTP requires hyperpolarization of the postsynaptic cell at the time of a presynaptic spike (Lamsa et al., 2007).
Inhibitory LTP and LTD due to sensory deprivation. Monocular eye closure in mice changed the efficacy of the inhibitory synapses of fast-spiking interneurons (probably basket cells) onto layer 4 star pyramids in binocular V1 (Maffei et al., 2010). The direction of the change depended on the age of the mouse. Prior to the critical period for ocular dominance plasticity, monocular occlusion caused LTD of these inhibitory synapses. During the critical period for ocular dominance plasticity, monocular occlusion seemed to induce inhibitory LTP instead, which occurred when the presynaptic cell fired a rapid train of spikes while the postsynaptic cell was depolarized, but not when the postsynaptic cell was concurrently spiking (Maffei et al., 2006).
This form of inhibitory synaptic plasticity differs in several respects from classic Hebbian NMDAR-dependent STDP. NMDAR-dependent LTP seems to be predominantly postsynaptic (Citri & Malenka, 2008), while LTP of these GABAergic synapses appeared to be predominantly presynaptic, reflecting an increase in quantal content. Both forms of STDP require postsynaptic depolarization concurrent with presynaptic transmitter release, but unlike NMDA-dependent LTP, this kind of LTP at GABAergic synapses was prevented by concurrent postsynaptic spiking and therefore can be considered anti-Hebbian—cells that fire together are prevented from strengthening their connection.
Depending on its timing, post-pre pairing can induce both LTD and LTP. In the inhibitory synapses from fast spiking non-accommodating interneurons (probable basket cells) onto L2/3 pyramidal neurons in 18-day-old rats, both LTP and LTD could be induced by post-pre pairing, depending on the delay between a postsynaptic train of backpropagating action potentials and a subsequent presynaptic spike (Holmgren & Zilberter, 2001). When the delay was less than 300 msec, LTD occurred. When the delay was more than 300 msec but less than 800 msec, LTP occurred. A train of backpropagating action potentials alone without a presynaptic spike induces neither LTP nor LTD. This form of plasticity is probably postsynaptic, with both LTP and LTD simultaneously induced by different mechanisms and LTP predominating with longer delays.
Symmetric LTP. In slices of auditory cortex (A1) from mice two to four weeks old, LTP could be induced in inhibitory synapses on L5 pyramidal cells by pairing an action potential in the recorded L5 pyramidal cell with an extracellular shock in L4 near the apical dendrite (D’amour & Froemke, 2015). Around the time a depolarizing current in the recording electrode triggered an action potential, an extracellular shock excited both inhibitory and excitatory fibers synapsing on the soma and dendrites of the recorded cell. Consistent with theoretical predictions (Vogels et al., 2011), both co-activated excitatory postsynaptic potentials (EPSPs) and inhibitory postsynaptic potentials (IPSPs) expressed STDP when the postsynaptic action potential and the extracellular shock occurred within 10 msec of each other. For the EPSPs, pre-post pairing induced LTP, while post-pre pairing induced LTD—which is classical Hebbian STDP for excitatory synapses. Both pre-post pairing and post-pre pairing induced LTP of IPSPs, which could be considered a form of Hebbian STDP for inhibitory synapses, since cells that fire together tend to form stronger connections. IPSPs recorded at the soma in response to extracellular shocks were probably driven primarily by some type of fast-spiking basket cells (Kruglikov & Rudy, 2008). After pre-post pairing, the combined effect of LTP of EPSPs and LTP of IPSPs was to increase the probability that the recorded cell fired an action potential in response to an external shock. After post-pre pairing, the combined effect of LTD of EPSPs and LTP of IPSPs was to decrease the probability that the recorded cell fired an action potential in response to an external shock. Both the excitatory and the inhibitory STDP required NMDA receptors.
Symmetric LTD. In cultured neurons and brain slices from rat hippocampus, GABAergic IPSPs were recorded from CA1 pyramidal cells in response to stimulation of stratum radiatum interneurons (Woodin et al., 2003). Spikes were induced in the recorded pyramidal cells by current injection through the recording electrode and in the connected inhibitory interneurons either by extracellular shocks in the tissue slices or by current injection in tissue culture. Both pre-post and post-pre pairing with an interspike interval less than 20 msec induced a depolarizing shift in the GABAergic synaptic current reversal potential. GABAergic synaptic currents were measured with the cell voltage-clamped at a hyperpolarized potential. Under these experimental conditions, depolarizing the GABAergic synaptic reversal potential actually increased the GABAergic synaptic currents because it moved the reversal potential farther from the artificially hyperpolarized membrane potential. If the cell had been depolarized by glutamatergic activity in vivo, depolarizing the GABAergic synaptic reversal potential would have brought the reversal potential closer to the membrane potential, reducing the GABAergic synaptic currents and making GABAergic synapses less effective in countering the depolarizing effects of glutamatergic synaptic currents. When the difference in timing between the pre- and postsynaptic spikes was less than 20 msec, the GABAergic synaptic reversal potential was depolarized by a change in intracellular Cl– concentration resulting from a decrease in K-Cl co-transporter (KCC2) activity. The change in intracellular Cl– concentration required GABAergic synaptic transmission but did not require the activation of glutamate receptors. It was blocked by inhibition of L-type Ca2+ channels. When the interval between the presynaptic spike and the postsynaptic spike exceeded 50 msec, pairing induced no change in intracellular Cl– concentration, but did cause a reduction in GABAergic synaptic conductance. Repetitive stimulation of the presynaptic neuron without postsynaptic spiking also induced a reduction in GABAergic synaptic conductance without changing intracellular Cl– concentration. Both depolarizing the GABAergic synaptic reversal potential and decreasing GABAergic synaptic conductance resulted in LTD of the GABAergic synapses, weakening their influence on the cell’s membrane potential in vivo. This form of synaptic change can be considered anti-Hebbian, because cells that fire together tend to form a weaker connection.
Hebbian inhibitory STDP. Inhibitory synapses on stellate cells in entorhinal cortex exhibited a form of inhibitory STDP that depended on intracellular calcium levels mediated by L-type voltage-gated channels (Haas et al., 2006). With a 10 msec interval between an extracellular shock and a postsynaptic spike induced by current injection through the recording electrode, pre-post pairing induced LTP, while post-pre pairing induced LTD. Neither presynaptic stimulation alone nor postsynaptic spiking alone induced long-term changes at these synapses.
Functional Consequences of Changes in Inhibitory Synaptic Efficacy
These diverse kinds of inhibitory plasticity might support different functional consequences (Kullmann et al., 2012).
Symmetric spike-timing dependent inhibitory LTP can balance both feed forward and feedback inhibition with excitation on a cell-by-cell basis (D’amour & Froemke, 2015). Repeated pre-post pairing induces LTP of both EPSPs and IPSPs, regardless whether IPSPs precede or follow postsynaptic spikes. Both excitation and inhibition are strengthened, tending to keep them in balance. The potentiation of inhibition is self-limiting. Sufficiently strong inhibition will prevent the postsynaptic cell from firing and thereby bring an end to the potentiation process. The balance is specific for each postsynaptic cell, depending on which of its inputs tend to be active at approximately the same time.
Hebbian inhibitory STDP can balance feedforward inhibition with excitation on a cell-by-cell basis. Immediate feedback may be depressed. In Hebbian inhibitory STDP, pre-post pairing induces LTP, while post-pre pairing induces LTD (Haas et al., 2006). The excitatory and feedforward inhibitory inputs that are active just before a cell fires will all be potentiated. The potentiation of inhibition is self-limiting. Sufficiently strong inhibition will prevent the postsynaptic cell from firing and thereby bring an end to LTP. Immediate feedback excitation and inhibition that arrive just after a postsynaptic spike could both be depressed. The balance is specific for each postsynaptic cell, depending on which of its inputs tend to be active at the same time and on the relation of their activity to activity in the postsynaptic cell.
The stimulus specificities of some neocortical pyramidal cells could be sharpened by anti-Hebbian plasticity of the distal synapses of large basket cells. Each pyramidal cell has a range of preferred stimuli to which it gives vigorous responses and a range of non-preferred stimuli to which it responds weakly or not at all. In several studies (Crook & Eysel, 1992; Crook, Kisvárday, & Eysel, 1997, 1998; Froemke et al., 2007; Griffen & Maffei, 2014; Kisvárday et al., 2002; Kullmann & Lamsa 2011, 2012; Martinez et al., 2002; Monier et al., 2003; Zhang, Tan, Schreiner, & Merzenich, 2003), the receptive field specificities of some pyramidal cells were sharpened by inhibition of pyramidal cell responses to non-preferred stimuli. In one study of cat V1, selective inhibition of non-preferred stimuli was seen in 40% of the cells studied (Monier et al., 2003). Selective inhibition of non-preferred stimuli is not expressed for every cell. In particular, it does not seem to account for sharpening the orientation selectivity of simple cells in L4 of cat V1 (Liu et al. 2010; Priebe & Ferster, 2008, 2012).
Like pyramidal cells, basket cells have stimulus preferences (Runyan et al., 2010). Since neighboring cortical cells tend to share stimulus preferences, as exemplified by the eye dominance and orientation columns in cat and monkey V1 and the tonotopic organization of A1, neighboring interneurons would inhibit responses to the preferred stimuli of a pyramidal cell rather than its non-preferred stimuli, although at least in rat A1, there may be interesting non-uniform organization of such side-band inhibition (Yoshimura et al. 2005; Zhang et al., 2003). The large basket cells whose distal axons contact a pyramidal cell can have somas far enough away not to share its stimulus preferences. They would be a likely source providing inhibition evoked by non-preferred stimuli. Each pyramidal cell is contacted by many basket cells (Packer & Yuste, 2011). Stronger contacts from basket cells that do not share the pyramidal cell’s stimulus preferences could result from anti-Hebbian plasticity. When a basket cell and a pyramidal cell share the same stimulus preferences, they would tend to fire in near synchrony in response to their preferred stimuli so that anti-Hebbian synaptic plasticity would cause LTD in the basket cell synapses (Holmgren & Zilberter, 2001; Kullmann et al., 2012; Woodin et al., 2003) or block LTP in them (Maffei et al., 2006, 2010). The basket cells that would be able to maintain strong synapses on the pyramidal cell would be those that do not share its stimulus preferences so that they do not tend to fire in synchrony with the pyramidal cell. This argument for anti-Hebbian plasticity applies only to the distal synapses of large basket cells; it is inconclusive for the synapses of small basket cells on their neighbors and for the synapses of the proximal axons of large basket cells on their neighbors.
Selective inhibition of non-preferred stimuli is not the only way that inhibition sharpens stimulus preferences (Griffen & Maffei, 2014; Monier et al., 2003; Priebe & Ferster, 2005, 2008, 2012). Even inhibition selective for a cell’s preferred stimulus can sharpen the cell’s stimulus selectivity if IPSPs are more broadly tuned than EPSPs; and inhibition does not have to be selective at all to sharpen stimulus selectivity by making it more difficult for any but the most preferred stimuli to drive the membrane potential above threshold.
When post-pre pairing induces inhibitory LTD, the result might be to make responses stronger and more prolonged (Holmgren & Zilberter, 2001). This form of plasticity depresses feedback inhibition. Weakened feedback inhibition would allow an excitatory drive to produce a stronger, more long-lasting response (Holmgren & Zilberter, 2001).
Each pyramidal cell may be subject to different forms of inhibitory plasticity. A single pyramidal cell in the neocortex may be contacted by a dozen different kinds of inhibitory interneurons (Fishell & Rudy, 2011; Kepecs & Fishell, 2014; Markram et al., 2004; Rudy et al., 2011; Watts & Thomson, 2005; Figure 1), and diverse kinds of inhibitory interneurons are also present in the hippocampus, the cerebellum, and other brain structures. The synapses of different kinds of inhibitory interneurons may exhibit different kinds of plasticity (Citri & Malenka, 2008; Griffen & Maffei, 2014; Kullmann & Lamsa, 2011; Larsen & Sjöström, 2015; Ma, Hu, & Agmon, 2012; Mao et al., 2012; Naka & Adesnik, 2016; Sandler et al., 2016; Seabrook & Huberman, 2015; Sjöström et al., 2008; Trachtenberg, 2015; Xue et al., 2014). Therefore, a single postsynaptic cell may exhibit several different kinds of plasticity with different functional consequences, affecting different parts of the cell, such as its soma or its apical dendritic tuft.
Plasticity of Synaptic Input onto Inhibitory Interneurons
Synapses onto inhibitory interneuronal somata and dendrites can also be modified, which might alter the extent to which these cells inhibit target cells. Diverse types of plasticity have been observed in synapses on the soma and dendrites of inhibitory interneurons (Kullmann & Lamsa, 2011). Some excitatory synapses onto hippocampal GABAergic interneurons can exhibit NMDA receptor-dependent Hebbian STDP, while others exhibit CP-AMPA receptor-dependent anti-Hebbian LTP requiring concurrent hyperpolarization of the postsynaptic cell (Lamsa et al., 2007). Both Hebbian and anti-Hebbian STDP of excitatory inputs onto inhibitory interneurons have also been noted in the neocortex. Inhibitory synapses onto inhibitory interneurons can also be plastic (Ma, Hu, & Agmon, 2012).
In L2/3 of rat somatosensory cortex, the excitatory synapses of pyramidal cells on fast-spiking interneurons (probably basket cells) exhibited classic Hebbian plasticity; but pyramidal cell synapses on low-threshold spiking cells (Martinotti cells) exhibited only LTD in response to both pre-post and post-pre pairing (Lu et al., 2007).
Functional Consequences of Changes in the Efficacy of Synaptic Inputs to Inhibitory Interneurons
STDP enables changes in a synapse to be adjusted to produce specific functional changes in the behavior of the postsynaptic cell that depend on pre- and postsynaptic patterns of activity. It is less clear how to adjust changes in the synaptic input to an interneuron’s soma and proximal dendrites to produce specific functional changes in the behavior of the cells its axon contacts. STDP depends on the detection of coincident pre- and postsynaptic spikes. It is thought to be governed by diffusible factors originating near the synapse where coincident pre- and postsynaptic spiking occurs, which are typically confined to a dendritic spine or a dendrite (Kullmann et al., 2012; Shouval, Samuel, & Wittenberg, 2010). To reach synapses on an inhibitory interneuron’s soma or proximal dendrites, molecules originating at its axon terminal would have to undergo retrograde transport. They could affect specific synapses on the soma or proximal dendrites if those synapses carried appropriate synaptic tags (Frey & Morris, 1997). There are several other hypothetical mechanisms by which changes in the synaptic inputs to an interneuron might be adjusted to produce specific functional changes in the behavior of the interneuron’s postsynaptic targets:
1. An inhibitory interneuron tends to prevent its postsynaptic targets from responding to the stimuli that the interneuron prefers. Synaptic modifications that affect the stimulus preferences of an inhibitory interneuron thereby change the stimulus preferences of its postsynaptic targets.
2. In feedforward inhibition, it is common for an inhibitory interneuron and its target to receive excitatory input from axon collaterals of the same cells (Cruikshank, Urabe, Nurmikko, & Connors, 2010; Isaacson & Scanziani, 2011; Watts & Thomson, 2005). The same pattern of input may impinge on both an inhibitory interneuron and its target cell and induce LTP in both sets of input synapses. This would tend to preserve the balance of excitation and inhibition.
An inhibitory interneuron and its targets may both be inhibited by the same cells. Disinhibition might create opportunities for both interneurons and pyramidal cells to adjust their responses to the same stimuli (Pi et al., 2013).
3. Similar to pyramidal cells, inhibitory interneurons are subject to neuromodulatory influences that regulate the plasticity of synapses on their somas and dendrites (Froemke et al., 2007; Huang, Huganir, & Kirkwood, 2013; Kuchibhotla et al., 2017; Kruglikov & Rudy, 2008), which might enable coordinated forms of plasticity of pyramidal cells and interneurons.
Some Types of Inhibitory Interneurons
There is growing consensus on the proper classification of inhibitory interneurons (Fishell & Rudy, 2011; Isaacson & Scanziani, 2011; Kepecs & Fishell, 2014; Merchant et al., 2012; Roux & Buzsáki, 2015; Rudy et al., 2011; Figure 1).
In the forebrain, the great majority of inhibitory interneurons are GABAergic, whereas glycine acts as an ionotropic inhibitory neurotransmitter in the spinal cord, the auditory brainstem, and the medulla (Fishell & Rudy, 2011). Almost all GABAergic interneurons in the telencephalon are derived from the embryonic medial ganglionic eminence and caudal ganglionic eminence (Kepecs & Fishell, 2014). As they populate forebrain structures—the neocortex, the hippocampus, the striatum and the amygdala—interneurons of similar provenance share many common properties including their dendritic branching patterns and the presence of molecular markers such as parvalbumin and somatostatin (Fishell & Rudy, 2011; Kepecs & Fishell, 2014).
Almost all neocortical GABAergic interneurons fall into one of three non-overlapping classes: some 40% express the Ca2+ binding protein parvalbumin (PV), about 30% express the neuropeptide somatostatin (SOM), and about 30% express the ionotropic serotonin receptor 5HT3aR (Rudy et al., 2011; Figure 1). Neurons that express these and other molecular markers such as calbindin (CB), vasoactive intestinal protein (VIP), calretinin (CR), and receptors for neuromodulators such as acetylcholine, norepinephrine, and oxytocin can be identified by several means, such as immunofluorescence (Mathes & Kennis, 2016; Rudy et al., 2011; Figure 1). Transgenic animals can be created in which neurons that express a particular molecular marker also express fluorescent markers (e.g., GFP), opsins (e.g., channelrhodopsin-2), or designer receptors exclusively activated by designer drugs (DREADDs; Roth, 2016).
Inhibitory interneurons can further be distinguished by their location in the cortical layers, the sites at which they contact their postsynaptic targets (axon hillock, soma, proximal dendrites, apical dendritic shaft, superficial dendritic tufts), by their shape, and by the branching pattern of their axons (Figure 1). These anatomical features can be identified when an interneuron has been labeled with an injected dye or has been induced to take up a labeling agent that will show its form by autoradiography or horseradish peroxidase staining (Crook, Kisvárday, & Eysel, 1998; Kisvárday et al., 2002; Figure 2). Inhibitory interneurons can also be distinguished by their activity patterns and their sources of input (Fishell & Rudy, 2011; Kepecs & Fishell, 2014; Markram et al., 2004; Merchant et al., 2012).
Many of the types described in the neocortex have homologues in the hippocampus (Fishell & Rudy, 2011; Kepecs & Fishell, 2014). A smaller number have homologues in the paleocortex and the piriform cortex, and a still smaller number in the striatum and the amygdala. Most neurons in the striatum are inhibitory, including the spiny projecting neurons. In the olfactory bulb, GABAergic neurons that seem homologous to neocortical interneurons nonetheless have different layering and connectivity (Rudy et al., 2011).
This multidimensional diversity suggests that different types of inhibitory interneurons may serve different functions, and to do so they may need to exhibit different forms of plasticity (Lamsa et al., 2010; Kepecs & Fishell, 2014). Among the major types of GABAergic cells for which different functions may be imagined are the following.
Basket cells generally express parvalbumin and account for about 40% of all neocortical GABAergic interneurons (Markram et al., 2004). They have homologues throughout the forebrain (Fishell & Rudy, 2011). They target the somata and proximal dendrites of pyramidal neurons and interneurons (Markram et al., 2004; Figure 1). The somata of basket cells are distributed through layers 2 through 6. Many basket cells sustain high-frequency trains of brief action potentials with little spike frequency adaptation (Rudy et al., 2011), have a low input resistance and fast membrane time constants, and mediate fast powerful IPSPs (Fishell & Rudy, 2011). Several subtypes of basket cells can be distinguished by size and axonal extent and the expression of channels and receptors (Rudy et al., 2011; Runyan et al., 2010; Wang, Gupta, Toledo-Rodriguez, Wu, & Markram, 2002). Basket cells are inhibited by activation of muscarine, serotonin, adenosine and GABA receptors (Kruglikov & Rudy, 2008).
In cortical L4, basket cells are the main interneuron target of thalamocortical axons. (Cruikshank et al., 2010; Rudy et al., 2011). They mediate feedforward inhibition and are implicated in the establishment and maintenance of fast gamma frequency cortical rhythms. They are likely to be key contributors to the maintenance of the balance between excitation and inhibition (Xue et al., 2014). In L5, basket cells receive non-selective input from neighboring pyramidal cells and in turn contact a large proportion of their pyramidal cell neighbors (Naka & Adesnik, 2016; Packer & Yuste, 2011). A single basket cell may contact as many as 1000 pyramidal cells.
In mouse V1, some basket cells have broadly tuned receptive fields, while others are sharply tuned for orientation (Runyan et al., 2010). The long horizontal axons of basket cells may selectively make strong contacts on distant pyramidal cells with particular response properties (Crook & Eysel, 1992; Crook, Kisvárday, & Eysel, 1997, 1998; Kisvárday et al., 2002; Runyan et al., 2010) hundreds of microns distant from the basket cell soma. Some large basket cells have long axons that span multiple cortical columns, as long as 2 mm (Markram et al., 2004).
In L2/3 of cat V1, it was noted that the proximal and distal portions of the axons of large basket cells had different preferences for the cells on which they made synaptic contacts. The proximal basket cell axons preferred to synapse on cells that shared the orientation and direction preference of the basket cell itself, while the distal basket cell axons preferred to synapse on cells whose preferred orientation and direction of motion was perpendicular to that of the basket cell (Crook, Kisvárday, & Eysel, 1998; Kisvárday et al., 2002; Figure 2).
The excitatory connection from a pyramidal cell onto a basket cell displays short-term depression. Basket cells are recruited as soon as their inputs from pyramidal cells are fired, but the effect tends to weaken if the pyramidal cell continues to fire. A reciprocally connected pair consisting of a basket cell and a pyramidal cell are likely to share more common excitatory inputs than a pair that are not reciprocally connected (Fino, Packer, & Yuste, 2013). Additionally, for excitatory synapses from pyramidal cells onto basket cells in L2/3 of rat somatosensory cortex, both pre-post pairing and post-pre pairing induced LTD dependent on metabotropic glutamate receptors (Lu et al., 2007).
Martinotti Cells (LTS Cells)
Martinotti cells are also known as low-threshold spiking (LTS) cells (Fishell & Rudy, 2011). They express somatostatin, and never express parvalbumin or VIP. They are strongly excited by muscarinic agonists (Fishell & Rudy, 2011). Inhibition by Martinotti cells has a strong suppressive effect on dendritic calcium spikes. Their somata can be found in L2–6 and their axons project vertically toward L1, where they synapse on the apical dendritic tufts of pyramidal neurons (Figure 1). Their axons can also project horizontally in L1 for millimeters to inhibit the apical tufts of pyramidal cell dendrites in neighboring and distant columns. Martinotti cells in L5 and L6 can also selectively target cells in L4 (Markram et al., 2004).
In L2/3 of mouse frontal cortex, each Martinotti cell contacts a large fraction (as much as 75%) of all the pyramidal cells within a cylinder with a radius of 200 microns with an axis passing through the Martinotti cell and perpendicular to the cortical surface (Fino & Yuste, 2011; Fino, Packer, & Yuste, 2013). Connected pairs of Martinotti cells and pyramidal cells do not share more common inputs than unconnected pairs (Fino & Yuste, 2011; Fino, Packer, & Yuste, 2013).
A step depolarization of a Martinotti cell elicits just a few spikes. The excitatory connection from a pyramidal cell onto a Martinotti cell displays short-term facilitation (Silberberg & Markram, 2007). Martinotti cells are recruited slowly by repeated firing of their excitatory inputs (Ma et al. 2010). An increase in the number of pyramidal cells simultaneously driving a Martinotti cell produces a disproportionately larger increase in the activity of the Martinotti cell (Kapfer, Glickfeld, Atallah, & Scanziani, 2007) so that bursts of activity in just a few pyramidal cells can activate many Martinotti cells and inhibit the dendrites of many nearby pyramidal cells (Naka & Adesnik, 2016). When descending activation of a pyramidal cell apical dendritic tuft coincides with back-propagating action potentials evoked by activation of the soma and proximal dendrites, dendritic Ca2+ spikes in the apical dendrite propagate to the soma to release a burst of spikes from the axon hillock (Larkum, 2013). A few such pyramidal cell bursts in synchrony will presumably activate nearby Martinotti cells, shutting down dendritic activity in all or most nearby pyramidal cell apical dendrites (Naka & Adesnik, 2016). This could create a “winner-take-all” situation in which the first pyramidal cells to be recruited shut down their neighbors (Palmer et al., 2012).
In mouse barrel cortex, Martinotti cells are spontaneously active during periods of quiet wakefulness, but reduce their firing in response to sensory input (Gentet et al., 2012). They may provide tonic inhibition to distal dendrites of pyramidal neurons, which is disinhibited during top-down computations and sensory integration (Karnani et al., 2014). Martinotti cells receive weaker excitatory drive from thalamocortical fibers than basket cells (Cruikshank et al., 2010). Martinotti cells in neocortex bear many resemblances to oriens-lacunosum moleculare (O-LM) interneurons in the stratum oriens of the hippocampus (Fishell & Rudy, 2011). LTP can be induced in excitatory non-NMDA synapses on O-LM interneurons by a combination of high-frequency afferent stimulation and post-synaptic hyperpolarization—an anti-Hebbian protocol (Kullmann & Lamsa, 2011). This form of plasticity may result from changes in the presynaptic glutamate release probability. LTP can also be induced in these cells by a theta-burst stimulation pattern combined with postsynaptic depolarization, a form of plasticity that appears to be due to postsynaptic changes (Kullmann & Lamsa, 2011).
Translaminar cells in layer 6 (Bortone, Olsen, & Scanziani, 2014; Cruikshank et al., 2010; not shown in Figure 1), with somata in layer 6, have axons that branch profusely in L2-6. They are fast-spiking, do not express SOM, and probably do express PV. They are excited by axon collaterals of pyramidal cells in L6 that project to the thalamus. When these cells are activated, they strongly inhibit cells throughout the thickness of the cortex. Thus activation of cortico-thalamic pyramidal cells in L6 results in disynaptic inhibition of cells throughout the cortical column.
Disinhibitory VIP Cells
Among the GABAergic interneurons that express both 5-HT3aR and VIP is a group that primarily synapse on other inhibitory interneurons, mostly interneurons that express somatostatin, but also some that express parvalbumin (Pi et al., 2013). These VIP-containing interneurons were strongly activated by both behavioral rewards and punishments (Pi et al., 2013). In addition to expressing 5-HT3aR receptors for serotonin, they also express receptors for acetylcholine. Many of these interneurons have vertically oriented dendrites and axons (Naka & Adesnik, 2016; see the cell labeled “VIP/CRF” in Figure 1). Consequently, activation of a disinhibitory VIP cell will create a “hole” in the “blanket of inhibition” cast by the background activity of Martinotti cells (Karnani et al., 2014).
Projecting GABAergic Cells
In the striatum, GABAergic medium spiny neurons are the principal neurons that project to the globus pallidus. In the hippocampus, there are several classes of projecting GABAergic neurons, including back projection cells, cells projecting to the septum, and cells of the oriens-retrohippocampal projection. GABAergic cortico-cortical projecting cells have also been described in neocortex (Fishell & Rudy, 2011). Some cortical inhibitory neurons send their axons through the corpus callosum to synapse on cells in contralateral cortex (Roux & Buzsáki, 2015).
Hypotheses about the Specificity of Inhibitory Functionality
With the widespread use of optogenetic techniques, rapid progress is being made in attributing particular functions of inhibition to particular types of inhibitory interneurons. Fast-spiking basket cells are implicated in regulating cortical rhythms (Campanac et al., 2013; Isaacson & Scanziani, 2011). Basket cells with long axons are implicated in increasing the specificity of receptive fields (Crook & Eysel, 1992; Crook, Kisvárday, & Eysel, 1997, 1998; Kisvárday et al., 2002; Lee et al., 2012). Both basket cells and Martinotti cells are implicated in sensory adaptation to repetitive stimuli (Natan et al., 2015) and both basket cells and Martinotti cells are implicated in maintaining a “blanket of inhibition” that contributes to the sparseness of cortical responses (Karnani et al., 2014).
During fetal development, LGN axons driven by the two eyes are extensively overlapped in L4 of the cortex and then gradually segregate themselves into eye-dominance columns. In ferrets, normal eye dominance columns can form even when both eyes are surgically removed before LGN fibers invade the cortex (Chalupa & Huberman, 2004; Sur & Leamey, 2001). The spontaneous activity of the LGN is a potential source of instruction for ocular dominance columns (Sur & Leamey, 2001). Eye dominance columns and orientation columns can be created in the auditory cortex of a ferret that has been rewired neonatally so that the medial geniculate nucleus receives its input from the retina rather than the inferior colliculus (Sur & Leamey, 2001).
The pattern of activity in the retino-cortical pathway can have a marked effect on its organization. If one eye of an animal is closed during a critical postnatal period, there is a marked loss of binocularly driven cells in V1 (Maffei et al., 2006, 2010; Wiesel & Hubel, 1963). GABA-mediated inhibition seems to regulate the beginning and the end of the critical period (Sur & Leamey, 2001; Trachtenberg, 2015). Eye dominance plasticity is likely to be regulated not only by basket cells contacting the soma and proximal dendrites, but also by disinhibitory interneurons and by interneurons, such as Martinotti cells, that regulate competition between excitatory contacts on the distal dendrites (Scheyltjens & Arckens, 2016; Trachtenberg, 2015). Monocular and binocular occlusion can affect basket cell synapses on pyramidal cells and also withdraw excitatory LGN input from basket cells. This reduction in LGN basket cell drive disinhibits the postsynaptic targets of the basket cells, including both pyramidal cells and other inhibitory interneurons (Trachtenberg, 2015).
Optogenetic disinhibition of SOM cells (probable Martinotti cells) in adult mouse V1 permits the restoration of eye dominance plasticity beyond the usual critical period (Fu, Kaneko, Tang, Alvarez-Buylla, & Stryker, 2015; Scheyltjens & Arckens, 2016).
Balance and Scaling
Multiple mechanisms are responsible for maintaining a balance between excitation and inhibition and for keeping neural activity within a suitable operating range (Campanac et al., 2013; Froemke, 2015; Isaacson & Scanziani, 2011; Pozo & Goda, 2010; Roux & Buzsáki, 2015; Vogels et al., 2011; Watt & Desai, 2010; Xue et al., 2014). “Balanced” excitation and inhibition can refer to several different interactions and relations between excitatory and inhibitory processing of a cell or network. The soma, axon hillock, proximal dendrites, and apical dendritic tuft of a pyramidal cell are each targeted by its own set of excitatory and inhibitory inputs (Froemke, Poo, & Dan, 2005; Markram et al., 2004; Merchant et al., 2012; Palmer et al., 2012; Roux & Buzsáki, 2015), which may have to be balanced by separate mechanisms.
Because balancing and scaling are negative feedback mechanisms (Pozo & Goda, 2010), several such mechanisms can be active at the same time without conflict. Different mechanisms may work on different time scales (D’amour & Froemke, 2015; Froemke et al., 2007; Gandhi, Yanagawa, & Stryker, 2008; Watt & Desai, 2010). Several different types of inhibitory interneurons may contribute to balance and scaling according to different rules, affecting different compartments of the postsynaptic cell and balancing different sets of excitatory inputs.
Multiple homeostatic mechanisms maintain a neural network at a suitable level of activity. Each tends to counteract changes in activity level. When the network becomes more active, these scaling mechanisms reduce the efficacy of excitatory synapses and increase the efficacy of inhibitory synapses to lower the activity level; conversely, when the network becomes less active, scaling mechanisms increase the efficacy of excitatory synapses and decrease the effectiveness of inhibition (Pozo & Goda, 2010; Watt & Desai, 2010; Xue et al., 2014). Both presynaptic and postsynaptic changes are involved, with many simultaneously active signaling molecules. Although synaptic scaling tends to affect all the synapses on a neuron (Watt & Desai, 2010), not all of them need be equally affected (Pozo & Goda, 2010; Xue et al., 2014). One consequence of such homeostatic mechanisms would be competition between synapses. If one set of excitatory synapses on a cell becomes stronger, it will increase the cell’s excitatory drive. Homeostatic scaling will then reduce the effectiveness of all excitatory synapses. The net effect will be that the strengthening of one set of synapses weakens all the others.
Balance must also be maintained at the level of individual inputs (Denève & Machens, 2016). In rat A1, EPSC and IPSC responses to tones tend to peak at the same preferred frequency (Froemke et al., 2007; Wehr & Zador, 2003; Zhang et al., 2003). The preferred frequency can be changed from the original preferred frequency to another frequency an octave lower by repeatedly pairing a non-preferred tone with stimulation of the nucleus basalis (Froemke et al., 2007, 2013). The initial effect of this pairing is to disturb the balance between excitation and inhibition by a reduction in IPSC amplitude in response to the paired tone, followed in 10 to 20 seconds by an increase in EPSC amplitude in response to the paired tone. This disinhibition allows Hebbian plasticity to produce LTP of excitatory synapses onto the recorded cell from cells that respond to the paired frequency.
Although nucleus basalis stimulation alone produces a transient non-specific reduction in synaptic transmission from cortical fast-spiking basket cells onto pyramidal cells even in the absence of paired tones (Kruglikov & Rudy, 2008), the enduring reduction in IPSCs after pairing was specific to the paired frequency (Froemke et al., 2007, 2013), This specificity must arise from some interaction between the effects of nucleus basalis stimulation and the activity caused by the paired tone. Later, roughly 10 minutes following pairing, there was also a reduction in both the EPSCs and the IPSCs evoked by certain other frequencies, which depended on the continued presentation of sound (Froemke et al., 2007). Which frequencies were affected depended both on the original preferred tone and on which sounds were presented continuously following pairing (Froemke et al., 2013).
After two hours of continuous auditory stimulation by a broad range of frequencies after pairing had ended, the ratio between excitation and inhibition was eventually restored at the whole range of frequencies. The paired tone became the new preferred frequency both for EPSCs and for IPSCs at the recording locus. This gradual restoration of balance might result from heterosynaptic plasticity in which activity at one synapse leads to long-term modifications at other synapses (Froemke et al., 2013; Froemke, 2015).
Stimulus Specific Adaptation
When a stimulus is frequently repeated, many cells respond more vigorously to its initial presentation than to subsequent presentations (Grill-Spector, Henson, & Martin, 2006; Kovács & Schweinberger, 2016; Natan et al., 2015). This stimulus specific adaptation (SSA) is pervasive and has multiple manifestations and multiple underlying mechanisms.
Adaptation of narrowly tuned inputs. One of the causes of SSA is adaptation of narrowly tuned inputs. When a tone is presented, it will activate only some of the narrowly tuned thalamic inputs to a pyramidal cell in A1. The ones that were activated will be adapted and respond less the next time the tone is presented. A repeated tone would activate only these adapted thalamic inputs, eliciting a weakened synaptic drive and a weakened pyramidal cell response. A deviant tone would activate some thalamic inputs that had not yet been adapted, eliciting a stronger synaptic drive. In A1 of lightly anesthetized mice, adaptation of narrowly tuned thalamic inputs does indeed account for part, but not all, of pyramidal cell SSA (Natan et al., 2015). The further contributions to SSA of cortical inhibitory interneurons were evaluated using optogenetic techniques to selectively and reversibly suppress the activity of SOM interneurons or PV interneurons.
Facilitation of SOM inhibition. SOM interneurons in A1 caused an additional contribution to pyramidal cell SSA because they respond only weakly to a single input, but their response is strongly facilitated by repeated activation. Each repetition of a tone activates SOM interneurons more strongly and more effectively inhibits the pyramidal cells they contact, thereby reducing the pyramidal cell response to repeated tones.
Uniform PV inhibition. PV interneurons were not facilitated by the presentation of repeated tones. PV inhibition decreased pyramidal cells’ response to repeated tones and to deviant tones by the same amount. Because the response to repeated tones was already smaller, PV inhibition had a proportionately larger effect on pyramidal cell responses to repeated tones than on their responses to deviant tones. This also increases the ratio of the response to deviant tones to the response to repeated tones.
A hypothetical explanation for long-lasting SSA. In this preparation reductions in responses to repeated tones took place over a few minutes and depended on mechanisms that are quickly reversible. However, SSA can be detected as long as three days after an initial brief exposure (van Turennout, Ellmore, & Martin, 2000). Hypothetically, long-lasting SSA could result from LTP at the synapses of pyramidal cells on Martinotti cells. Each presentation of a repeated tone produced a few pyramidal cell spikes, but a deviant tone produced a rapid burst of pyramidal cell spikes. When antidromic spikes from the axon hillock of a pyramidal cell invade its apical dendritic tuft at the same time as the tuft receives EPSPs, a long-lasting Ca2+ spike will be elicited in the apical dendrite that can evoke a burst of firing from the axon hillock (Larkum, 2013). This burst of activity can fire somatostatin expressing Martinotti cells which could (1) shut down Ca2+ spikes in the apical dendrites of neighboring pyramidal cells (Scheyltjens & Arckens, 2016) and (2) induce LTP at the synapses of the pyramidal cell onto the Martinotti cells (Lu et al., 2007). Because of this LTP, a future repetition of the stimulus could fire those Martinotti cells more vigorously, creating greater pyramidal cell inhibition and reducing the pyramidal cell response to the repeated stimulus.
Contributions of non-specific inhibition. In sensory cortex, the most effective excitatory sensory stimuli produce EPSPs that tend to reach threshold reliably, while less effective stimuli might produce subthreshold EPSPs on a large fraction of trials. This “iceberg effect” causes the stimulus preferences of the spiking response to be more sharply tuned than the stimulus preferences of EPSPs. Cortical cells are restrained by a “blanket of inhibition” (Karnani et al., 2014), which lowers the entire iceberg so that less of its tip rises above the surface, thereby narrowing the range of inputs that can reach threshold and evoke spikes (Priebe & Ferster, 2008, 2012).
Each Martinotti cell receives excitatory input from a large fraction of the pyramidal cells near it and inhibits a large fraction of the pyramidal cells in its neighborhood (Fino, Packer, & Yuste, 2013; Karnani et al., 2014; Silberberg & Markram, 2007). A Ca2+ spike initiated in the apical dendrite of a pyramidal cell is likely to initiate a burst of spikes from its axon hillock that excite neighboring Martinotti cells. The activated Martinotti cells prevent nearby pyramidal cells from initiating Ca2+ spikes (Karnani et al., 2014; Larkum, 2013; Palmer et al., 2012). This winner-take-all mechanism allows only the earliest, most strongly excited pyramidal cells to fire Ca2+ spikes, helping to sparsify cortical responses.
In many cases, inhibition and excitation in the adult cortex share the same stimulus preferences. Neighboring cells in A1 tend to respond most strongly to tones of the same frequency (Schnupp, Nelken, & King, 2011). In cats and monkeys, neighboring cells in V1 share similar orientation preferences (Hubel & Wiesel, 1968; Ohki, Chung, Ch’ng, Kara, & Reid, 2005). When neighboring cortical cells share similar stimulus preferences and pyramidal cells are inhibited by neighboring interneurons, the EPSPs and IPSPs measured at the soma will share similar stimulus preferences simply because of this anatomical arrangement.
In cat V1, the relation between EPSP orientation preference and IPSP orientation preference varies by cortical layer. In layer 4 and in layer 2/3, IPSPs and EPSPs recorded at the soma generally share the same preferred orientation. In layer 5, however, the preferred orientation for IPSPs is often approximately at right angles to the preferred orientation for EPSPs (Martinez et al., 2002). Although EPSPs and IPSPs originating in the apical dendritic tuft can have important effects on sensory perception (Larkum, 2013; Manita et al., 2015), these PSPs may not be evident in recordings at the soma (Kruglikov & Rudy, 2008).
Specificity of long-range intracortical connections. Cortical cells receive excitatory and inhibitory input not only from their neighbors, but also from cortical cells with long horizontal axons or axon branches. In cat V1, horizontal axons of pyramidal cells can run as far as 6–8 mm parallel to the cortical surface, crossing multiple orientation and eye-dominance columns (Gilbert & Wiesel, 1989). They do not form synapses at random but preferentially form clusters of synapses on cells that share their preferred orientation (Gilbert & Wiesel, 1989; Kisvárday, Toth, Rausch, & Eysel, 1997; Roerig & Chen, 2002; Stettler, Das, Bennett, & Gilbert, 2002).
Both basket cells and Martinotti cells also send their axons horizontally in the cortex, crossing eye dominance columns (Markram et al., 2004). Although they make many contacts on neighboring cells that share their orientation preference, basket cells are more likely than pyramidal cells to synapse on cells that do not share their orientation preference (Kisvárday et al., 1997; Roerig & Chen, 2002). Like all cortical cells, basket cells tend to form synaptic contacts in regularly spaced clusters (Binzegger, Douglas, & Martin, 2007).
Specific inhibition guided by anti-Hebbian plasticity. As noted above, the orientation preferences of some V1 cells seem to be significantly sharpened by inhibition of their responses to non-preferred stimuli. When IPSPs in a cat V1 pyramidal cell prefer an orientation significantly different from the preferred orientation for EPSPs, they are presumably driven by inhibitory interneurons with long horizontal axons whose somas are in distant orientation columns. Both basket cells and Martinotti cells have axons that run horizontally for hundreds or even thousands of microns spanning multiple orientation columns, but IPSPs recorded at the soma are predominantly driven by basket cells (Kruglikov & Rudy, 2008).
The proximal and distal portions of the axon of a large basket cell in layer 2/3 of cat V1 have different patterns of connectivity (Figure 2). Near the soma, the proximal axon contacts pyramidal cells with a broad range of orientation and direction preferences, with the largest number of contacts on pyramidal cells sharing the same orientation and direction preference as the basket cell itself. The distal portions of the basket cell axon may preferentially contact pyramidal cells whose orientation preference is perpendicular to that of the basket cell and whose direction preference is opposite to that of the basket cell (Kisvárday et al., 2002).
The effects of large basket cells on the orientation and direction preferences of their pyramidal cell targets was tested by reversibly inactivating basket cells using a local infusion of GABA (Crook & Eysel, 1992; Crook, Kisvárday, & Eysel, 1997, 1998). An array of electrodes was lowered into V1 (area 17) or the adjacent area 18 in a cat, consisting of a main recording electrode flanked by four iontophoresis electrodes approximately 600 microns distant from the main electrode which were likely to penetrate columns with an orientation preference perpendicular to that of the main electrode. A single cell was recorded by the main electrode and its orientation preference was determined. Multiunit recordings from the iontophoresis electrodes determined the orientation preferences of their columns. Then GABA was released from one or more of the iontophoresis electrodes to silence neurons near their tips. If the orientation preference of the silenced cells was perpendicular to that of the main cell, the main cell was disinhibited and responded to bars at right angles to its preferred orientation (Figure 3). When GABA iontophoresis was stopped, the disinhibitory effect was reversed. Iontophoresis of GABA through electrodes where the orientation preference is the same as that at the main electrode did not change the orientation preference of the recorded cell.
Radioactive retrograde labels introduced near the main electrode were transported to basket cell somas near those iontophoresis electrodes where GABA release changed the orientation preference of the recorded cell. Iontophoresis of an anterograde label through one of those iontophoresis electrodes labeled basket cells that send their axons to the recorded cell, as seen with the horseradish peroxidase reaction.
Together, these observations support the hypothesis that large basket cells made long-distance connections with pyramidal cells with a preferred orientation perpendicular to that of the basket cells. As noted above, anti-Hebbian plasticity of distal basket cell synapses could account for this selectivity. If the distal synapses of these basket cells are subject to anti-Hebbian plasticity, the distal axons of the basket cells would be prevented from forming strong connections onto pyramidal cells that share their stimulus preferences because those pyramidal cells would tend to fire in near synchrony with the basket cells. The basket cells that were able to form strong connections on a pyramidal cell would have to be those that do not share its stimulus preferences.
Although the distal axons of some large basket cells may preferentially form synapses according to an anti-Hebbian rule, their proximal axons tend to synapse on pyramidal cells without regard for orientation preference. Either of two hypotheses could account for this: (1) Their distal and proximal synapses might be subject to different forms of plasticity. (2) Both their distal and proximal synapses might be subject to anti-Hebbian plasticity; their proximal axons form only weak synapses on neighboring pyramidal cells, but their distal axons preferentially enter cortical columns where they succeed in forming strong synapses.
Hebbian synaptic plasticity is presumed to account for the selection of long-distance excitatory connections in V1 by correlated visual activity (Lowel & Singer, 1992). In the case of long-distance connections of large basket cells, anti-Hebbian synaptic plasticity might select the connections.
The evolving story of cortical inhibitory plasticity shows that different types of inhibitory interneurons play different roles in a variety of inhibitory functions, that several types of inhibitory plasticity have been attested, and that different forms of plasticity can be expected to have different effects on the organization and specificity of inhibitory connections. Many interesting questions remain open, such as these:
• Hypotheses. Are some of the speculations presented above correct? Do large basket cells form their distal synapses under the influence of an anti-Hebbian rule? Are the strengths of excitatory synapses on the soma and proximal dendrites of inhibitory interneurons modified in a way that produces specific effects on the postsynaptic targets of the inhibitory interneurons? Does LTP of the synapses of pyramidal cells onto Martinotti cells account for all or part of long-lasting stimulus specific adaptation?
• Basket cells. Several different types of inhibitory plasticity have been demonstrated for basket cells, with different functional consequences. What accounts for the differences? Do different types of basket cells exhibit different types of plasticity? Are different types of plasticity manifested depending on the basket cells’ postsynaptic targets? Are different types of plasticity manifested by the proximal and distal portions of the axon of a single large basket cell?
• Martinotti cells. How specific are the connections of Martinotti cells? Each Martinotti cells appears to contact a large fraction of its pyramidal cell neighbors, but these synapses may not be equally efficacious. If Martinotti cell synapses on pyramidal cell apical dendritic tufts have varying strengths, are they governed by a form of STDP?
• Martinotti cell axons. Do the long horizontal processes of Martinotti cells in layer 1 preferentially contact certain apical dendrites? If so, which ones and what determines which dendrites are contacted?
• Disinhibitory interneurons. What are the specificities of the contacts of VIP expressing disinhibitory interneurons onto other inhibitory interneurons? What, if any, forms of synaptic plasticity are responsible for those specificities?
• Inhibitory projections. What are the specificities of the contacts of projecting inhibitory interneurons? What, if any, forms of synaptic plasticity are responsible for those specificities?
• Inhibition of non-preferred stimuli. It appears that inhibition of non-preferred stimuli sharpens the response specificities of a fraction of cortical pyramidal cells. How widespread is this phenomenon? What determines which cells are affected? Can the results of Martinez et al. (2002) be replicated?
• Calcium spikes. What are the specificities of inhibitory control of the propagation of Ca2+ spikes from the apical dendrite to the axon hillock? What role do these specificities play in determining the function of descending intracortical connections terminating in layer 1?
Inhibition is so pervasive in the brain that inhibitory interneurons are likely to be implicated in almost every aspect of behavior, including development, sensory deprivation, perception, discrimination, movement, learning, forgetting, rhythm, coordination, instinct, and neuropsychiatric disease (Crook & Eysel, 1992; Crook, Kisvárday, & Eysel, 1997, 1998; De Koninck, 2007; Froemke et al., 2007; Griffen & Maffei, 2014; Fu et al., 2015; Isaacson & Scanziani, 2011; Kuchibhotla et al., 2017; Letzkus et al., 2011, 2015; Marlin et al., 2015; Natan et al., 2015; Roux & Buzsáki, 2015; Scheyltjens & Arckens, 2016; Sur & Leamey, 2001; Trachtenberg, 2015; Vogels et al., 2013). Any change in inhibition can be expected to have multiple effects on behavior. Changes in inhibition resulting from inhibitory plasticity can affect the specificity of interneuronal connections and the quantitative balance between excitation and inhibition that is needed to regulate gain and keep the cortex within its operating range.
Inhibition can regulate plasticity (Froemke et al., 2007; Fu et al., 2015; Letzkus et al., 2011, 2015; Vogels et al., 2013), and learned changes to inhibition can gate context-dependent learned behavioral responses (Kuchibhotla et al., 2017). Changes in pyramidal cells can depend on a release from inhibition by the action of disinhibitory interneurons (Fu et al., 2015; Letzkus et al., 2011, 2015; Pi et al., 2013) or by the release of such neuromodulators as acetylcholine, serotonin, noradrenaline, and oxytocin (Froemke et al., 2007, 2015; Kruglikov & Rudy, 2008; Kuchibhotla et al., 2017).
Different types of inhibitory interneurons can play different roles in behavior (Brown et al., 2015; Carlén et al., 2012; Roux & Buzsáki, 2015; Scheyltjens & Arckens, 2016). Parvalbumin expressing basket cells are implicated in regulating brain rhythms and cognitive functions (Carlén et al., 2012), including extinction of fear conditioning (Brown et al., 2015). Somatostatin expressing cells are implicated in regulating learning (Scheyltjens & Arckens, 2016). Both PV cells and SOM cells are implicated in regulating development (Maffei et al., 2006; Scheyltjens & Arckens, 2016), the effects of sensory deprivation (Griffen & Maffei, 2014; Trachtenberg, 2015), and stimulus specific sensory adaptation (Natan et al., 2015). Changes in chloride transport are likely causes of some neurological disorders (De Koninck, 2007) and also account for some forms of inhibitory plasticity (Woodin et al., 2003). Since each cortical pyramidal cell is contacted by several types of inhibitory interneurons, multiple types of interneurons can affect every aspect of behavior.
Current research is starting to reveal the effects of particular types of inhibitory interneurons on particular types of behavior using transgenic animals in which certain types of interneurons are missing and optogenetic techniques to selectively activate or suppress certain types of interneurons (Brown et al., 2015; Carlén et al., 2012; Fu et al., 2015; Lee et al., 2012; Natan et al., 2015; Roth, 2016; Roux & Buzsáki, 2015; Scheyltjens & Arckens, 2016). These approaches do not yet determine how the specificity of interneuronal connections is established and how it affects behavior.
Abeles, M. (1991). Corticonics: Neural circuits of the cerebral cortex. Cambridge, England: Cambridge University Press.Find this resource:
Barth, A. L., & Poulet, J. F. A. (2012). Experimental evidence for sparse firing in the neocortex. Trends in Neurosciences 35(6), 345–355.Find this resource:
Binzegger, T., Douglas, R. J., & Martin, K. A. C. (2007). Stereotypical bouton clustering of individual neurons in cat primary visual cortex. Journal of Neuroscience 27, 12242–12254.Find this resource:
Bliss, T. V. P., & Lømo, T. (1973). Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path. Journal of Physiology 232, 331–356.Find this resource:
Bortone, D. S., Olsen, S. R., & Scanziani, M. (2014). Translaminar inhibitory cells recruited by layer 6 corticothalamic neurons suppress visual cortex. Neuron 82, 474–485.Find this resource:
Brown, J. A., Ramikie, T. S., Schmidt, M. J., Báldi, R., Garbett, K., Everheart, M. G., … Mirnics, K. (2015). Inhibition of parvalbumin-expressing interneurons results in complex behavioral changes. Molecular Psychiatry 20, 1499–1507.Find this resource:
Campanac, E., Gasselin, C., Baude, A., Rama, S., Ankri, N., & Debanne, D. (2013). Enhanced intrinsic excitability in basket cells maintains excitatory-inhibitory balance in hippocampal circuits. Neuron 77, 712–722.Find this resource:
Carlén, M., Meletis, K., Siegle, J. H., Cardin, J. A., Futai, K., Vierling-Claassen, D., … Tsai, L. H. (2012). A critical role for NMDA receptors in parvalbumin interneurons for gamma rhythm induction and behavior. Molecular Psychiatry 17, 537–548.Find this resource:
Chalupa, L. M., & Huberman, A. D. (2004). A new perspective on the role of activity in the development of eye-specific retinogeniculate projections. In M. S. Gazzaniga (Ed.) The cognitive neurosciences III (pp. 85–92). Cambridge, MA: MIT Press.Find this resource:
Chen, J. L., Villa, K. L., Cha, J. W., So, P. T. C., Kubota, Y., & Nedivi, E. (2012). Clustered dynamics of inhibitory synapses and dendritic spines in the adult neocortex. Neuron 74, 361–373.Find this resource:
Citri, A., & Malenka, R. C. (2008). Synaptic plasticity: Multiple forms, functions, and mechanisms. Neuropsychopharmacology 33, 18–41.Find this resource:
Crook, J. M., & Eysel, T. L. (1992). GABA-induced inactivation of functionally characterized sites in cat visual cortex (area 18): effects on orientation tuning. Journal of Neuroscience 12, 1816–1825.Find this resource:
Crook, J. M., Kisvárday, Z. F., & Eysel, U. T. (1997). GABA-induced inactivation of functionally characterized sites in cat striate cortex: effects on orientation tuning and direction selectivity. Visual Neuroscience 14, 141–158.Find this resource:
Crook, J. M., Kisvárday, Z. F., & Eysel, U. T. (1998). Evidence for a contribution of lateral inhibition to orientation tuning and direction selectivity in cat visual cortex: reversible inactivation of functionally characterized sites combined with neuroanatomical tracing techniques. European Journal of Neuroscience 10, 2056–2075.Find this resource:
Cruikshank, S. J., Urabe, H., Nurmikko, A. V., & Connors, B. W. (2010). Pathway-specific feedforward circuits between thalamus and neocortex revealed by selective optical stimulation of axons. Neuron 65, 230–245.Find this resource:
D’amour, J. A., & Froemke, R. C. (2015). Inhibitory and excitatory spike-timing-dependent plasticity in the auditory cortex. Neuron 86, 514–528.Find this resource:
De Koninck, Y. (2007). Altered chloride homeostasis in neurological disorders: a new target. Current Opinion in Pharmacology 7, 93–99.Find this resource:
Denève, S., & Machens, C. K. (2016). Efficient codes and balanced networks. Nature Neuroscience 19, 375–382.Find this resource:
Fino, E., Packer, A. M., & Yuste, R. (2013). The logic of inhibitory connectivity in the neocortex. The Neuroscientist 19, 228–237.Find this resource:
Fino, E., & Yuste, R. (2011). Dense inhibitory connectivity in neocortex. Neuron 69, 1188–1203.Find this resource:
Fishell, G., & Rudy, B. (2011). Mechanisms of inhibition within the telencephalon: “where the wild things are.” Annual Review of Neuroscience 34, 535–567.Find this resource:
Frey, U., & Morris, R. G. M. (1997). Synaptic tagging and long-term potentiation. Nature 385, 533–536.Find this resource:
Froemke, R. C. (2015). Plasticity of cortical excitatory-inhibitory balance. Annual Review of Neuroscience 38, 195–219.Find this resource:
Froemke, R. C., Carcea, I., Barker, A. J., Yuan, K., Seybold, B. A., Martins, A. R. O., … Schreiner, C. E. (2013). Long-term modification of cortical synapses improves sensory perception. Nature Neuroscience 16, 79–88.Find this resource:
Froemke, R. C., & Dan, Y. (2002). Spike-timing-dependent synaptic modification induced by natural spike trains. Nature 416, 433–438.Find this resource:
Froemke, R. C., Poo, M. M., & Dan, Y. (2005). Spike-timing-dependent synaptic plasticity depends on dendritic location. Nature 434, 221–225.Find this resource:
Froemke, R. C., Letzkus, J. J., Kampa, B. M., Hang, G. B., & Stuart, G. J. (2010). Dendritic synapse location and neocortical spike-timing-dependent plasticity. Frontiers in Synaptic Neuroscience 2.Find this resource:
Froemke, R. C., Merzenich, M. M., & Schreiner, C. E. (2007). A synaptic memory trace for cortical receptive field plasticity. Nature 450, 425–429.Find this resource:
Fu, Y., Kaneko, M., Tang, Y., Alvarez-Buylla, A., & Stryker, M. P. (2015). A cortical disinhibitory circuit for enhancing adult plasticity. Elife 4, e05558.Find this resource:
Gandhi, S. P., Yanagawa, Y., & Stryker, M. P. (2008). Delayed plasticity of inhibitory neurons in developing visual cortex. Proceedings of the National Academy of Sciences 105, 16797–16802.Find this resource:
Gentet, L. J., Kremer, Y., Taniguchi, H., Huang, Z. J., Staiger, J. F., & Petersen, C. H. (2012). Unique functional properties of somatostatin-expressing GABAergic neurons in mouse barrel cortex. Nature Neuroscience 15, 607–612.Find this resource:
Gilbert, C. D., & Wiesel, T. N. (1989). Columnar specificity of intrinsic horizontal and corticocortical connections in cat visual cortex. Journal of Neuroscience 9, 2432–2442.Find this resource:
Griffen, T. C., & Maffei, A. (2014). GABAergic synapses: their plasticity and role in sensory cortex. Frontiers in Cellular Neuroscience 8, 91.Find this resource:
Grill-Spector, K., Henson, R., & Martin, A. (2006). Repetition and the brain: neural models of stimulus-specific effects. Trends in Cognitive Sciences 10(1), 14–23.Find this resource:
Gur, M., Kagan, I., & Snodderly, D. M. (2005). Orientation and direction selectivity of neurons in V1 of alert monkeys: functional relationships and laminar distributions. Cerebral Cortex 15, 1207–1221.Find this resource:
Haas, J. S., Nowotny, T., & Abarbanel, H. D. I. (2006). Spike-timing-dependent plasticity of inhibitory synapses in the entorhinal cortex. Journal of Neurophysiology 96, 3305–3313.Find this resource:
Hawkins, J., & Ahmad, S. (2016). Why neurons have thousands of synapses, a theory of sequence memory in neocortex. Frontiers in Neural Circuits 10.Find this resource:
Hebb, D. O. (1949). The organization of behavior: A neuropsychological approach. New York: John Wiley & Sons.Find this resource:
Holmgren, C. D., & Zilberter, Y. (2001). Coincident spiking activity induces long-term changes in inhibition of neocortical pyramidal cells. Journal of Neuroscience 21, 8270–8277.Find this resource:
Huang, S., Huganir, R. L., & Kirkwood, A. (2013). Adrenergic gating of Hebbian spike-timing-dependent plasticity in cortical interneurons. Journal of Neuroscience 33, 13171–13178.Find this resource:
Hubel, D. H., & Wiesel, T. N. (1968). Receptive fields and functional architecture of monkey striate cortex. The Journal of Physiology 195, 215–243.Find this resource:
Isaacson, J. S., & Scanziani, M. (2011). How inhibition shapes cortical activity. Neuron 72, 231–243.Find this resource:
Jasinska, M., Siucinska, E., Cybulska-Klosowicz, A., Pyza, E., Furness, D. N., Kossut, M., & Glazewski, S. (2010). Rapid, learning-induced inhibitory synaptogenesis in murine barrel field. Journal of Neuroscience 30, 1176–1184.Find this resource:
Kapfer, C., Glickfeld, L. L., Atallah, B. V., & Scanziani, M. (2007). Supralinear increase of recurrent inhibition during sparse activity in the somatosensory cortex. Nature Neuroscience 10, 743–753.Find this resource:
Karnani, M. M., Agetsuma, M., & Yuste, R. (2014). A blanket of inhibition: functional inferences from dense inhibitory connectivity. Current Opinion in Neurobiology 26, 96–102.Find this resource:
Kepecs, A., & Fishell, G. (2014). Interneuron cell types are fit to function. Nature 505(7483), 318–326.Find this resource:
Kisvárday, Z. F., Ferecskó, A. S., Kovács, K., Buzás, P., Budd, J. M. L., & Eysel, U. T. (2002). One axon-multiple functions: specificity of lateral inhibitory connections by large basket cells. Journal of Neurocytology 31(3–5), 255–264.Find this resource:
Kisvárday, Z. F., Toth, E., Rausch, M., & Eysel, U. T. (1997). Orientation-specific relationship between populations of excitatory and inhibitory lateral connections in the visual cortex of the cat. Cerebral Cortex 7, 605–618.Find this resource:
Kovács, G., & Schweinberger, S. R. (2016). Repetition suppression: an integrative view. Cortex 80, 1–4.Find this resource:
Kruglikov, I., & Rudy, B. (2008). Perisomatic GABA release and thalamocortical integration onto neocortical excitatory cells are regulated by neuromodulators. Neuron 58, 911–924.Find this resource:
Kubota, Y. (2014). Untangling GABAergic wiring in the cortical microcircuit. Current Opinion in Neurobiology 26, 7–14.Find this resource:
Kubota, Y., Karube, F., Nomura, M., & Kawaguchi, Y. (2016). The diversity of cortical inhibitory synapses. Frontiers in Neural Circuits 10.Find this resource:
Kuchibhotla, K. V., Gill, J. V., Lindsay, G. W., Papadoyannis, E. S., Field, R. E., Hindmarsh Sten, T. A., … Froemke, R. C. (2017). A synaptic and circuit switch for control of context-dependent behavior. Nature Neuroscience, 20(1), 62–71.Find this resource:
Kullmann, D. M., & Lamsa, K. P. (2011). LTP and LTD in cortical GABAergic interneurons: emerging rules and roles. Neuropharmacology 60, 712–719.Find this resource:
Kullmann, D. M., Moreau, A. W., Bakiri, Y., & Nicholson, E. (2012). Plasticity of inhibition. Neuron 75, 951–962.Find this resource:
Lamsa, K. P., Kullmann, D. M., & Woodin, M. A. (2010). Spike-timing dependent plasticity in inhibitory circuits. Frontiers in Synaptic Neuroscience 2, 8.Find this resource:
Lamsa, K. P., Heeroma, J. H., Somogyi, P., Rusakov, D. A., & Kullmann, D. M. (2007). Anti-Hebbian long-term potentiation in the hippocampal feedback inhibitory circuit. Science 315, 1262–1266.Find this resource:
Larkum, M. (2013). A cellular mechanism for cortical associations: an organizing principle for the cerebral cortex. Trends in Neurosciences 36(3), 141–151.Find this resource:
Larsen, R. S., & Sjöström, P. J. (2015). Synapse-type-specific plasticity in local circuits. Current Opinion in Neurobiology 35, 127–135.Find this resource:
Lee, S.-H., Kwan, A. C., Zhang, S., Phoumthipphavong, V., Flannery, J. G., Masmanidis, S. C., … Dan, Y. (2012). Activation of specific interneurons improves V1 feature selectivity and visual perception. Nature 488, 379–383.Find this resource:
Letzkus, J. J., Wolff, S. B. E., & Lüthi, A. (2015). Disinhibition, a circuit mechanism for associative learning and memory. Neuron 88, 264–276.Find this resource:
Letzkus, J. J., Wolff, S. B. E., Meyer, E. M. M., Tovote, P., Courtin, J., Herry, C., & Lüthi, A. (2011). A disinhibitory microcircuit for associative fear learning in the auditory cortex. Nature 480(7377), 331–335.Find this resource:
Lisman, J., Grace, A. A., & Duzel, E. (2011). A neoHebbian framework for episodic memory: role of dopamine-dependent late LTP. Trends in Neurosciences 34(10), 536–547.Find this resource:
Liu, B., Li, P., Sun, Y. J., Li, Y., Zhang, L. I., & Tao, H. W. (2010). Intervening inhibition underlies simple-cell receptive field structure in visual cortex. Nature Neuroscience 13, 89–96.Find this resource:
Lowel, S., & Singer, W. (1992). Selection of intrinsic horizontal connections in the visual cortex by correlated neuronal activity. Science 255(5041), 209–212.Find this resource:
Lu, J., Li, C., Zhao, J.-P., Poo, M., & Zhang, X. (2007). Spike-timing-dependent plasticity of neocortical excitatory synapses on inhibitory interneurons depends on target cell type. Journal of Neuroscience 27, 9711–9720.Find this resource:
Ma, Y., Hu, H., & Agmon, A. (2012). Short-term plasticity of unitary inhibitory-to-inhibitory synapses depends on the presynaptic interneuron subtype. Journal of Neuroscience 32, 983–988.Find this resource:
Ma, W., Liu, B., Li, Y., Huang, Z. J., Zhang, L. I., & Tao, H. W. (2010). Visual representations by cortical somatostatin inhibitory neurons: selective but with weak and delayed responses. Journal of Neuroscience 30, 14371–14379.Find this resource:
Maffei, A., Nataraj, K., Nelson, S. B., & Turrigiano, G. G. (2006). Potentiation of cortical inhibition by visual deprivation. Nature 443(7107), 81–84.Find this resource:
Maffei, A., Lambo, M. E., & Turrigiano, G. G. (2010). Critical period for inhibitory plasticity in rodent binocular V1. Journal of Neuroscience 30, 3304–3309.Find this resource:
Manita, S., Suzuki, T., Homma, C., Matsumoto, T., Odagawa, M., Yamada, K., … Murayama, M. (2015). A top-down cortical circuit for accurate sensory perception. Neuron 86, 1304–1316.Find this resource:
Mao, R., Schummers, J., Knoblich, U., Lacey, C. J., Van Wart, A., Cobos, I., … Sur, M. (2012). Influence of a subtype of inhibitory interneuron on stimulus-specific responses in visual cortex. Cerebral Cortex 22, 493–508.Find this resource:
Marik, S. A., Yamahachi, H., Meyer zum Alten Borgloh, S., & Gilbert, C. D. (2014). Large-scale axonal reorganization of inhibitory neurons following retinal lesions. Journal of Neuroscience 34, 1625–1632.Find this resource:
Marlin, B. J., Mitre, M., D’amour, J. A., Chao, M. V., & Froemke, R. C. (2015). Oxytocin enables maternal behaviour by balancing cortical inhibition. Nature 520(7548), 499–504.Find this resource:
Markram, H., Gerstner, W., & Sjöström, P. J. (2012). Spike-timing-dependent plasticity: a comprehensive overview. Frontiers in Synaptic Neuroscience 4, 8.Find this resource:
Markram, H., Toledo-Rodriguez, M., Wang, Y., Gupta, A., Silberberg, G., & Wu, C. (2004). Interneurons of the neocortical inhibitory system. Nature Reviews Neuroscience 5, 793–807.Find this resource:
Markram, H., Gerstner, W., & Sjöström, P. J. (2011). A history of spike-timing-dependent plasticity. Frontiers in Synaptic Neuroscience 29. doi: 10.3389/fnsyn.2011.00004.Find this resource:
Martinez, L. M., Alonso, J.-M., Clay Reid, R., & Hirsch, J. A. (2002). Laminar processing of stimulus orientation in cat visual cortex. Journal of Physiology 540, 321–333.Find this resource:
Mathes, T., & Kennis, J. T. M. (2016). Editorial: Optogenetic tools in the molecular spotlight. Frontiers in Molecular Biosciences 3, 14.Find this resource:
Medini, Paolo. (2014). “Experience-dependent plasticity of visual cortical microcircuits.” Neuroscience 278: 367–384.Find this resource:
Merchant, H., de Lafuente, V., Pena-Ortega, F., & Larriva-Sahd, J. (2012). Functional impact of interneuronal inhibition in the cerebral cortex of behaving animals. Progress in Neurobiology 99(2), 163–178.Find this resource:
Monier, C., Chavane, F., Baudot, P., Graham, L. G., & Frégnac, Y. (2003). Orientation and direction selectivity of synaptic inputs in visual cortical neurons: a diversity of combinations produces spike tuning. Neuron 37, 663–680.Find this resource:
Naka, A., & Adesnik, H. (2016). Inhibitory circuits in cortical layer 5. Frontiers in Neural Circuits 10, 35.Find this resource:
Natan, R. G., Briguglio, J. J., Mwilambwe-Tshilobo, L., Jones, S. I., Aizenberg, M., Goldberg, E. M., & Geffen, M. N. (2015). Complementary control of sensory adaptation by two types of cortical interneurons. eLife 4, e09868.Find this resource:
Ohki, K., Chung, S., Ch’ng, Y. H., Kara, P., & Reid, R, C. (2005). Functional imaging with cellular resolution reveals precise micro-architecture in visual cortex. Nature 433(7026), 597–603.Find this resource:
Packer, A. M., & Yuste, R. (2011). Dense, unspecific connectivity of neocortical parvalbumin-positive interneurons: a canonical microcircuit for inhibition? Journal of Neuroscience 31, 13260–13271.Find this resource:
Palmer, L., Murayama, M., & Larkum, M. (2012). Inhibitory regulation of dendritic activity in vivo. Neuron 4, L5.Find this resource:
Petersen, C. C. H., & Crochet, S. (2013). Synaptic computation and sensory processing in neocortical layer 2/3. Neuron 78, 28–48.Find this resource:
Pi, H.-J., Hangya, B., Kvitsiani, D., Sanders, J. I., Huang, Z. J., & Kepecs, A. (2013). Cortical interneurons that specialize in disinhibitory control. Nature 503 (7477), 521–524.Find this resource:
Pluta, S., Naka, A., Veit, J., Telian, G., Yao, L., Hakim, R., … Adesnik, H. (2015). A direct translaminar inhibitory circuit tunes cortical output. Nature Neuroscience 18(11), 1631–1640.Find this resource:
Pozo, K., & Goda, Y. (2010). Unraveling mechanisms of homeostatic synaptic plasticity. Neuron 66, 337–351.Find this resource:
Priebe, N. J., & Ferster, D. (2005). Direction selectivity of excitation and inhibition in simple cells of the cat primary visual cortex. Neuron 45, 133–145.Find this resource:
Priebe, N. J., & Ferster, D. (2008). Inhibition, spike threshold, and stimulus selectivity in primary visual cortex. Neuron 57, 482–497.Find this resource:
Priebe, N. J., & Ferster, D. (2012). Mechanisms of neuronal computation in mammalian visual cortex. Neuron 75, 194–208.Find this resource:
Roerig, B., & Chen, B. (2002). Relationships of local inhibitory and excitatory circuits to orientation preference maps in ferret visual cortex. Cerebral Cortex 12, 187–198.Find this resource:
Rochester, N., Holland, J., Haibt, L., & Duda, W. (1956). Tests on a cell assembly theory of the action of the brain, using a large digital computer. IRE Transactions on information Theory 2, 80–93.Find this resource:
Roth, B. L. (2016). DREADDs for neuroscientists. Neuron 89, 683–694.Find this resource:
Roux, L., & Buzsáki, G. (2015). Tasks for inhibitory interneurons in intact brain circuits. Neuropharmacology 88, 10–23.Find this resource:
Rudy, B., Fishell, G., Lee, S. H., & Hjerling-Leffler, J. (2011). Three groups of interneurons account for nearly 100% of neocortical GABAergic neurons.” Developmental Neurobiology 71, 1 45–61.Find this resource:
Runyan, C. A., Schummers, J., Van Wart, A., Kuhlman, S. J., Wilson, N. R., Huang, Z. J., & Sur. M. (2010). Response features of parvalbumin-expressing interneurons suggest precise roles for subtypes of inhibition in visual cortex. Neuron 67, 847–857.Find this resource:
Sandler, M., Shulman, Y., & Schiller, J. (2016). A novel form of local plasticity in tuft dendrites of neocortical somatosensory layer 5 pyramidal neurons.” Neuron 90, 1028–1042.Find this resource:
Schäfer, R., Vasilaki, E., & Senn, W. (2007). Perceptual learning via modification of cortical top-down signals. PLoS Computational Biology 3, e165.Find this resource:
Scheyltjens, I., & Arckens, L. (2016). The current status of somatostatin-interneurons in inhibitory control of brain function and plasticity. Neural Plasticity. doi: 10.1155/2016/8723623.Find this resource:
Schnupp, J., Nelken, I., & King, A. (2011). Auditory neuroscience: Making sense of sound. Cambridge, MA: MIT Press.Find this resource:
Seabrook, T. A., & Huberman, A. D. (2015). Cortical cliques: a few plastic neurons get all the action. Neuron 86, 1113–1116.Find this resource:
Shapley, R., Hawken, M., & Ringach, D. L. (2003). Dynamics of orientation selectivity in the primary visual cortex and the importance of cortical inhibition. Neuron 38, 689–699.Find this resource:
Shatz, C. J., & Stryker, M. P. (1978). Ocular dominance in layer IV of the cat’s visual cortex and the effects of monocular deprivation. Journal of Physiology 281, 267.Find this resource:
Shouval, H. Z., Samuel, S. H., & Wittenberg, G. M. (2010). Spike timing dependent plasticity: a consequence of more fundamental learning rules. Frontiers in Computational Neuroscience 4. doi: 10.3389/fncom.2010.00019.Find this resource:
Silberberg, G., & Markram, H. (2007). Disynaptic inhibition between neocortical pyramidal cells mediated by Martinotti cells. Neuron 53, 735–746.Find this resource:
Sillito, A. M. (1977). Inhibitory processes underlying the directional specificity of simple, complex and hypercomplex cells in the cat’s visual cortex. Journal of Physiology 271, 699.Find this resource:
Sjöström, J., & Gerstner, W. (2010). Spike-timing dependent plasticity. Scholarpedia 5(2), 1362.Find this resource:
Sjöström, P. J., Rancz, E. A., Roth, A., & Häusser, M. (2008). Dendritic excitability and synaptic plasticity. Physiological Reviews 88, 769–840.Find this resource:
Stent, G. S. (1973). A physiological mechanism for Hebb’s postulate of learning. Proceedings of the National Academy of Sciences 70, 997–1001.Find this resource:
Stettler, D. D., Das, A., Bennett, J., & Gilbert, C. D. (2002). Lateral connectivity and contextual interactions in macaque primary visual cortex. Neuron 36, 739–750.Find this resource:
Sur, M., & Leamey, C. A. (2001). Development and plasticity of cortical areas and networks. Nature Reviews Neuroscience 2(4), 251–262.Find this resource:
Trachtenberg, J. T. (2015). Competition, inhibition, and critical periods of cortical plasticity. Current Opinion in Neurobiology 35, 44–48.Find this resource:
Van Hooser, S. D., Escobar, G. M., Maffei, A., & Miller, P. (2014). Emerging feed-forward inhibition allows the robust formation of direction selectivity in the developing ferret visual cortex. Journal of Neurophysiology 111, 2355–2373.Find this resource:
VanRullen, R., & Thorpe, S. J. (2002). Surfing a spike wave down the ventral stream. Vision Research 42, 2593–2615.Find this resource:
van Turennout, M., Ellmore, T., & Martin, A. (2000). Long-lasting cortical plasticity in the object naming system. Nature Neuroscience 3, 1329–1334.Find this resource:
Vogels, T. P., Sprekeler, H., Zenke, F., Clopath, C., & Gerstner, W. (2011). Inhibitory plasticity balances excitation and inhibition in sensory pathways and memory networks. Science 334, 1569–1573.Find this resource:
Vogels, T. P., Froemke, R. C., Doyon, N., Gilson, M., Haas, J. S., Liu, R., … Sprekeler, H. (2013). Inhibitory synaptic plasticity: spike timing-dependence and putative network function. Frontiers in Neural Circuits 7, EPFL-Review-189448.Find this resource:
Von der Malsburg, C. (1973). “Self-organization of orientation sensitive cells in the striate cortex. Kybernetik 14, 85–100.Find this resource:
Wang, Y., Gupta, A., Toledo-Rodriguez, M., Wu, C. Z., & Markram, H. (2002). Anatomical, physiological, molecular and circuit properties of nest basket cells in the developing somatosensory cortex. Cerebral Cortex 12, 395–410.Find this resource:
Watt, A. J., & Desai, N. S. (2010). Homeostatic plasticity and STDP: keeping a neuron’s cool in a fluctuating world. Frontiers in Synaptic Neuroscience 2, 5. doi: 10.3389/fnsyn.2010.00005.Find this resource:
Watts, J., & Thomson, A. M. (2005). Excitatory and inhibitory connections show selectivity in the neocortex. Journal of Physiology 562, 89–97.Find this resource:
Wehr, M., & Zador, A. M. (2003). Balanced inhibition underlies tuning and sharpens spike timing in auditory cortex. Nature 426, 442–446.Find this resource:
Wiesel, T. N., & Hubel, D. H. (1963). Single-cell responses in striate cortex of kittens deprived of vision in one eye. Journal of Neurophysiology 26, 1003–1017.Find this resource:
Wolfe, J., Houweling, A. R., & Brecht, M. (2010). Sparse and powerful cortical spikes. Current Opinion in Neurobiology 20, 306–312.Find this resource:
Woodin, M. A., Ganguly, K., & Poo, M. (2003). Coincident pre-and postsynaptic activity modifies GABAergic synapses by postsynaptic changes in Cl− transporter activity. Neuron 39, 807–820.Find this resource:
Wu, G. K., Arbuckle, R., Liu, B., Tao, H. W., & Zhang, L. I. (2008). Lateral sharpening of cortical frequency tuning by approximately balanced inhibition. Neuron 58, 132–143.Find this resource:
Xue, M., Atallah, B. V., & Scanziani, M. (2014). Equalizing excitation-inhibition ratios across visual cortical neurons. Nature 511, 596–600.Find this resource:
Yoshimura, Y., & Callaway, E. M. (2005). Fine-scale specificity of cortical networks depends on inhibitory cell type and connectivity. Nature Neuroscience 8, 1552–1559.Find this resource:
Zhang, L. I., Tan, A. Y. Y., Schreiner, C. E., & Merzenich, M. M. (2003). Topography and synaptic shaping of direction selectivity in primary auditory cortex. Nature 424(6945), 201–205.Find this resource: