Scheme showing a nerve terminal from a parvalbumin- (PV-) positive GABA neuron shortly after an action potential triggered Ca²⁺-dependent GABA release, highlighting components currently hypothesized to be altered in schizophrenia. In PV terminals, GABA release is tightly synchronized with Ca²⁺ influx, possibly due to the proximity between voltage-dependent Ca²⁺ channels and release sites. PV is a relatively slow buffer that probably is unable to bind Ca²⁺ before activation of the Ca²⁺ sensor promotes vesicle fusion. Ca²⁺ buffering by PV mainly accelerates the decay of the intraterminal Ca²⁺ transient (see text). GAD65 and GAD67, possibly acting as a dimer, drive GABA synthesis in the cytosol near synaptic vesicles. Vesicles uptake newly synthesized GABA via the vesicular GABA transporter vGAT. Vesicle fusion rapidly and transiently raises GABA concentration in the synaptic cleft, briefly exposing post-synaptic GABA_A receptors (GABA_ARs) to a high concentration of GABA. As GABA escapes from the synaptic cleft after GABA_AR activation, it may be taken up by the plasma membrane GABA transporter GAT1, apparently localized in the extrasynaptic neuronal membrane, as well as in glia. GAT1 therefore regulates the concentration of GABA reaching extrasynaptic GABA_ARs and synaptic GABA_ARs at other synapses (not shown in the scheme). The direction and magnitude of the chloride current produced by postsynaptic GABA_AR activation is regulated by the transporters KCC2 and NKCC1, which uptake and extrude chloride, respectively, setting the equilibrium potential for the GABA_A current, E_GABAA. Since PV accelerates the decay of the intraterminal Ca²⁺ transients, a decrease of PV in schizophrenia may facilitate repetitive GABA release, such as that observed during gamma oscillation episodes. A decrease of GAD67 levels in schizophrenia would reduce the cytosolic GABA concentration near synaptic vesicles. Because vGAT levels appear to be unaffected in schizophrenia, reduced GAD67 may lead to lower intravesicular GABA concentration, therefore decreasing the peak GABA concentration in the synaptic cleft and weakening the postsynaptic response. In schizophrenia, at some synapses postsynaptic GABA_AR density appears to be decreased, further weakening synaptic transmission, whereas at other synapses GABA_AR density is increased, possibly due to compensatory receptor upregulation. In schizophrenia, KCC2 and NKCC1 mRNA levels are normal, but two kinases that strongly regulate KCC2 and NKCC1 may be altered in ways that render an E_GABAA value more depolarizing than normal. Finally, reduced GAT1 in schizophrenia may alter the effects of synaptically released GABA via an exaggerated activation of extrasynaptic and heterosynaptic GABA_ARs. Alternatively, GAT1 activity may be reduced to compensate lower GABA levels due to GAD67 deficiency.

Diversity of perisomatic-targeting GABA neuron-mediated inhibition in cortical circuits may be due to differences in the reversal potential for the GABA_A-mediated current. (a) The scheme illustrates targeting by basket cells (either parvalbumin- or cholechystokinin-positive), which contact the soma and the proximal/perisomatic dendritic compartments, and by chandelier or axoaxonic neurons, which contact the axon initial segment. (b) The schemes illustrate differences in the postsynaptic effect of a GABA_AR conductance according to the value of the reversal potential for the GABA_A current (E_GABAA) relative to the resting membrane potential (V_mr) and the threshold potential to fire spikes (V_th). In (B1) to (B3), the time course of the GABA_AR-mediated conductance, chosen to be identical in all panels, is shown by the black traces and the IPSPs are shown by the blue/red traces. Note that the IPSP time course is always slower than the GABA_A conductance, although in the scheme the difference in time course is somewhat exaggerated for illustration purposes and does not match the actual time scales. (B1) Illustration of cases in which E_GABAA is negative relative to V_mr  (V_th > V_mr > E_GABAA), and the GABA_A conductance generates a hyperpolarizing IPSP (all blue IPSP trace). As the IPSP outlasts the GABA_A conductance, the duration of the inhibitory effect of the synaptic input (shown by the shaded blue rectangle) is extended by the membrane hyperpolarization that remains after the GABA_A conductance decays. (B2) Illustration of cases in which the IPSP is depolarizing because E_GABAA is positive relative to V_mr  (V_th > E_GABAA > V_mr). Just as the hyperpolarizing IPSP, the depolarizing IPSP outlasts the GABA_A conductance; however, in this case E_GABAA is below V_th, and therefore the depolarizing IPSP could have a dual inhibitory/excitatory effect (blue/red IPSP trace), initially producing shunting inhibition which lasts approximately the same time as the GABA_A conductance (shaded blue/red rectangle), followed by an enhanced excitability of the postsynaptic neuron due to the remaining phase of the depolarizing IPSP. (B3) Illustration of cases in which E_GABAA is positive relative to V_th  (E_GABAA > V_th > V_mr). In this case, the depolarizing IPSP has a purely excitatory effect (shaded red rectangle). Basket cells are thought to produce hyperpolarizing GABA_AR-mediated inhibition of pyramidal cells such as that illustrated in (B1). In contrast, the effect of chandelier neuron inputs is currently debatable, some studies suggesting an excitatory as that in (B3), other studies suggesting a purely inhibitory effect. Here, we suggest that depolarizing chandelier cell inputs may have a dual inhibitory/excitatory effect, illustrated in (B2), which could synchronize postsynaptic cells as described in Figure 2, although the depolarizing nature of the IPSP may accelerate the timing of synchronous firing after the postsynaptic cells escape from inhibition.

A model for GABA_AR-mediated synchronization mechanisms. The left panel shows a group of pyramidal cells that are common postsynaptic targets of an inhibitory GABA neuron. Perisomatic-targeting GABA neurons such as that in the scheme produce stronger inhibition than GABA neuron subtypes that target the dendrites. The right panel shows, above (black trace), the membrane potential of the GABA neuron which remains at rest before and after firing a sequence of four action potentials. The red traces below show the membrane potential simultaneously recorded from the postsynaptic pyramidal neurons, which are firing in response to a continuous excitatory input. Note that, before the GABA neuron starts firing, the pyramidal cells fire in an asynchronous manner. Shortly after the first GABA neuron spike, an inhibitory postsynaptic potential (IPSP) is produced (first black arrow) which simultaneously inhibits the firing of all pyramidal neurons. After the IPSP-mediated inhibition decays, the pyramidal neurons fire in nearly synchrony. Note that a similar postinhibition synchronization is observed with each of the IPSPs evoked by the interneuron spikes (each IPSP is denoted by a black arrow). Once the GABA neuron stops firing, pyramidal cell activity rapidly becomes asynchronous. Also note that a single action potential in a GABA neuron would synchronize the pyramidal cells only once, whereas production of a synchronized oscillation requires rhythmic GABA neuron firing. An oscillation episode composed of four cycles is shown in the figure. The production of synchronized oscillations via this mechanism may be impaired by various alterations of GABA_AR-mediated synaptic inhibition in schizophrenia (see Figure 1 and main text).

Relative levels of GAD65 and GAD67 in PVCh, PVBC, and CCKBC terminals. Cryostat sections of monkey PFC tissue (40 μm thick) were quadruple labeled for GAD65, GAD67, PV, and cannabinoid receptor 1 (CB1r). Among inhibitory synaptic boutons, CB1r is exclusively expressed in those CCK-positive neurons and is completely absent in terminals of PV-positive neurons. Presynaptic CB1rs are also present in excitatory synapses; however, the antibody used in the studies illustrated in this figure exclusively reveal CB1r expression in inhibitory boutons (see main text for details). Therefore, CB1r expression is a marker of CCK cell terminals. Single channel (a)-(b) and (d)-(e) and merged (c) and (f) projection images of deconvolved image stacks (2 z-planes 0.25 μm apart). Since four labels cannot be displayed together in a single image, they have been separated into two RGB images that contain GAD67 (red), GAD65 (green), and PV (blue (c)) or CB1r (blue (f)). Arrows: PV cartridge; solid arrowheads: CB1r⁺/GAD65⁺ and GAD67⁻ terminals; open arrowheads: GAD65⁺/GAD67⁺/PV⁺ terminals. Bar = 10 μm.

Intrinsic electrical properties of parvalbumin-positive GABA neurons. (a) Single action potentials in fast-spiking PV-positive neurons (FS) have significantly faster duration than pyramidal cell (PC) spikes or spikes in many other GABA neuron subtypes (not shown). (b) In response to sustained depolarizing current injection (500 ms rectangular current pulses shown below the traces), PCs produce high-frequency firing with significant spike-frequency adaptation as revealed by comparing the first and last interspike intervals (ISI). In contrast, adaptation of FS cell firing is much less significant or absent. (c) In response to gamma wave stimulation (sinusoidal current pulses shown below the traces), PCs show low capacity to respond with firing during each cycle of the gamma wave stimulus. (d) In response to gamma wave stimulation, FS cells show increased firing capacity, initiating spikes in the majority of gamma wave cycles. This property of the FS neuron membrane may contribute to the activation of FS neurons during gamma oscillations in vivo and is likely due to the resonance or frequency-preference properties (see text) that distinguish FS cells from PCs and also from other GABA neuron subtypes.