Microdeletions of the 22q11.2 locus are among the most common chromosomal abnormalities, occur predominantly de novo, and account for 1–2% of sporadic schizophrenia cases (Karayiorgou et al., 1995, 2010; McDonald-McGinn et al., 2001; Xu et al., 2008). Individuals with the 22q11.2 microdeletion are characterized by a high incidence of emotional problems and a spectrum of cognitive deficits, and approximately 30% eventually develop schizophrenia or schizoaffective disorder in adolescence or early adulthood (Pulver et al., 1994; Bearden et al., 2001; Woodin et al., 2001; Sobin et al., 2005; Chow et al., 2006).

A mouse model has been created to recapitulate the human deletion syndrome by creating an analogous deletion in the mouse genome that spans a segment syntenic to the 1.5-Mb human 22q11.2 microdeletion that results in the heterozygous deletion of 27 genes (Stark et al., 2008). Behavioral analysis of this mouse indicated both cognitive deficits, as manifested by impaired spatial working memory and contextual and cued fear learning, as well as deficits in sensorimotor gating—paralleling findings in schizophrenia—as demonstrated by reductions in pre-pulse inhibition (PPI) of acoustic startle (Swerdlow and Geyer, 1998; Stark et al., 2008). To further refine the interpretation of behavioral assays, results from novel object recognition and latent inhibition behavioral assays—that require hippocampal function but are largely independent of PFC function—indicated that purely hippocampal-dependent behaviors were intact suggesting that the primary functional deficit in these animals was largely restricted to dysfunction within the PFC (Fénelon et al., 2013).

Electrophysiological assessment of the 22q11.2-DS equivalent mouse showed that while nearly all forms of short-term and long-term plasticity tested in the hippocampus were unaltered (with the exception of modest reductions in 100 Hz induced LTP at CA3-CA1 synapses), synapses onto layer V neurons in the mPFC—although showing normal basic synaptic transmission and paired-pulse ratios—demonstrated enhanced synaptic depression during high-frequency stimulus trains and reductions in the initial phase of the synaptic potentiation (short-term potentiation) induced by the high frequency stimulus train for LTP induction (Drew et al., 2011b; Fénelon et al., 2013). Finally, in vivo recordings from mice performing a spatial working memory task have shown that the functional, synchronous coupling between the microcircuits in the hippocampus and the PFC is significantly reduced during working memory performance in deletion mice suggesting that the working memory deficits observed in these mice might be related to this disconnectivity (Sigurdsson et al., 2010). Overall these findings of cognitive deficits in the face of largely normal hippocampal electrophysiological function suggest that the primary dysfunction in the deletion syndrome mouse is likely due to a primary functional deficit in PFC microcircuit function and the connectivity between the PFC and the hippocampus. In addition to these findings in mice that employ the full 27 gene microdeletion, in order to more precisely dissect the underlying dysfunction, further evaluations have been performed in mice in which only single genes from within this region have been deleted individually.

DGCR8 is a gene within the 22q11.2 microdeletion whose gene product is involved with processing micro-RNAs (mi-RNAs) within the nucleus which are involved with repressing translation of their target mRNAs (Tomari and Zamore, 2005; Karayiorgou et al., 2010). While homozygous deletion of DGCR8 is lethal, heterozygous deletion such as occurs in the 22q11.2 deletion syndrome results in the reduction of numerous mi-RNAs leading to enhanced expression of their targets due to the loss of mi-RNA repression (Stark et al., 2008; Xu et al., 2013). Mice missing one copy of the DGCR8 gene (DGCR8± mice) demonstrate reduced PPI of acoustic startle, impaired performance in spatial working memory, but, in contrast to deletion-syndrome mice, both normal contextual and cued fear learning (Stark et al., 2008). Similar to electrophysiology findings in the deletion syndrome mouse, measures of both short-term and long-term plasticity were found to be normal at CA3-CA1 synapses in the hippocampus of DGCR8(±) mice, while synapses onto layer V neurons in the mPFC showed normal paired-pulse ratios (suggesting normal initial Pr) but enhanced synaptic depression during high-frequency stimulus trains and reductions in the initial phase of the synaptic potentiation induced by high frequency trains of stimuli (Fénelon et al., 2011). Thus, heterozygous deletion of DGCR8 alone seems to be able to account for most of the electrophysiological deficits found in the deletion syndrome mouse while the absence of fear-learning deficits suggests the involvement of other genes within the 27-gene deletion in these behaviors.

PRODH is a gene within the 22q11.2 microdeletion whose gene product is involved with L-proline degradation within mitochondria and its homozygous or heterozygous deletion results in a gene-dose dependent elevation in L-proline in both the periphery and within the CNS (Phang et al., 2001; Karayiorgou et al., 2010). Human subjects with hypomorphic mutations in their PRODH gene have a disease called hyperprolinemia type I that is characterized by seizures and significantly increased incidence of schizophrenia and schizoaffective disorder (Phang et al., 2001; Jacquet et al., 2005; Raux et al., 2007; Clelland et al., 2011). Mice with homozygous functional deletion of the PRODH gene (PRODH knockdown mice) demonstrate reduced PPI of acoustic startle, normal performance in spatial working memory tasks, and reductions in both contextual and cued fear learning (Gogos et al., 1999; Paterlini et al., 2005). Electrophysiological assessment of CA3-CA1 hippocampal synapses with field recordings in PRODH-deficient mice showed increased evoked EPSPs and reduced paired-pulse ratios (depressing compared to facilitating in control animals at a 20 ms interval) both of which suggest increased initial probability in NT release (Pr) in PRODH-deficient mice (Paterlini et al., 2005). These mice also showed decreased CA3-CA1 LTP induction which was notable for a complete absence of its presynaptic component whereas control animals showed increased Pr after LTP induction (Zakharenko et al., 2003; Paterlini et al., 2005). Taken together these electrophysiological results suggest that PRODH deficiency leads to altered presynaptic function with increased initial Pr and an inability of LTP induction to lead to further increases in initial Pr. Interestingly, PRODH-deficient also mice showed evidence of dopaminergic dysregulation within the PFC as evidenced by enhanced hyperlocomotion and dopamine release within the PFC upon amphetamine challenge as well as a possibly compensatory enhanced expression of the dopamine degradative enzyme COMT (a gene located within the 22q11.2 microdeletion that would be expected to be compromised in its ability to undergo an equal compensatory upregulation in patients with the microdeletion). Overall, the findings in PRODH-deficient mice suggest primarily alterations in presynaptic function that—despite normal working memory performance in these mice—would be expected lead to significant alterations in synaptic filtering and network computation within the hippocampus (Tsodyks and Markram, 1997; Abbott and Regehr, 2004). Furthermore, although the underlying mechanism remains to be resolved, PRODH hypofunction also appears to lead to dopaminergic dysregulation and studies in schizophrenics that have revealed a synergy between PRODH and COMT hypofunction support this as a potentially important disease mechanism (Raux et al., 2007).

Disrupted in schizophrenia 1 (DISC1) is a gene that was implicated in schizophrenia through its discovery as the gene disrupted by a chromosomal translocation that cosegregated with mental illness in a family pedigree prominently affected by schizophrenia, bipolar disorder, and major depression (St Clair et al., 1990; Millar et al., 2000; Blackwood et al., 2001). The precise cellular and biological role of the DISC1 gene product remains uncertain and resolution of this question has been hindered by the fact that its protein sequence appears to share no homology with other known proteins (Millar et al., 2000). To determine possible functions of Disc1, studies have identified numerous protein-binding partners of Disc1 the most promising of which suggest that Disc1 may play a significant role in modulating cellular function through alterations in cAMP levels and protein kinase function. Disc1 binds to and regulates the phosphodiesterase, PDE4b, and mice with a DISC1 mutation that models the consequences of the human chromosomal translocation demonstrate enhanced cAMP levels due to decrease PDE4 activity (Millar et al., 2005; Kvajo et al., 2011). Furthermore, Disc1 has also been shown to bind to and inhibit the function of the protein kinase GSK3β (Mao et al., 2009) leading to reduced phosphorylation of some of its protein targets. Although in early development Disc1 shows a broader expression pattern, in the adult brain prominent expression is restricted to the dentate gyrus and, to a lesser extent, the CA1 regions within the hippocampal microcircuitry (Austin et al., 2004). Based upon this expression pattern and given the potential role of adult neurogenesis in schizophrenia and other neuropsychiatric diseases, a number of studies have suggested a potentially important role for DISC1 in regulating microcircuit architecture through its regulation of neurogenesis in the dentate gyrus (Ghashghaei et al., 2006; Reif et al., 2006, 2007; Mao et al., 2009; Ming and Song, 2009; Inta et al., 2011; Kim et al., 2012). Within neurons, Disc1 has been shown to be present in multiple subcellular compartments including dendrites and dendritic spines as well as in growth cones and axonal terminals suggesting the possibility that Disc1 could regulate either postsynaptic or presynaptic function (Millar et al., 2005; Kirkpatrick et al., 2006; Taya et al., 2007; Bradshaw et al., 2008; Brandon and Sawa, 2011; Randall et al., 2014).

Using a genetic mouse model employing a DISC1 mutation that accurately recapitulates the consequences of the human chromosomal translocation, a comprehensive battery of behavioral analyses has revealed a pattern of cognitive deficits in these mice that appears to reflect cognitive deficits found in schizophrenia (Koike et al., 2006; Kvajo et al., 2008). Amongst this battery of cognitive behavioral assays, Disc1 mutant mice displayed an isolated deficit in spatial working memory as indicated by impaired performance on a delayed alternation t-maze task. Significantly, largely hippocampus-dependent behavioral tasks—contextual fear conditioning, novel object recognition, Morris water maze, and the eight-arm radial maze win-shift paradigm—all failed to reveal significant alterations in Disc1 mutant mice (Kvajo et al., 2008). Taken together, the results of these behavioral assays suggest an isolated deficit in the short-term maintenance and utilization of information—as opposed to information storage and retrieval—that likely results from information processing dysfunction in the PFC rather than in the hippocampus (Compte et al., 2000; Baddeley, 2012; Roux et al., 2012).

Electrophysiological assessment of granule cell mossy fiber synapses onto CA3 neurons—motivated by the high expression of Disc1 in the dentate gyrus—showed both normal basic synaptic transmission as well as normal LTP in response to 100 Hz stimulation (Austin et al., 2004; Kvajo et al., 2011). Measures of short-term plasticity at these synapses, however, revealed reduced paired-pulse ratios (suggesting increased Pr) and reductions in the short-term frequency facilitation characteristic of this synapse both pointing to selective deficits in short-term plasticity in Disc1 mutant mice (Nicoll and Schmitz, 2005). Electrophysiological assessment of CA3-CA1 hippocampal synapses, in contrast, showed both normal long-term and short-term plasticity with the exception of an isolated reduction in the initial, short-term synaptic potentiation induced by the 100 Hz stimulus train employed for LTP induction (Kvajo et al., 2008). Interestingly, this isolated deficit in short-term synaptic potentiation at CA3-CA1 synapses may potentially represent a significant pathophysiology for schizophrenia as a similar deficit in this short-term potentiation mechanism was observed at mPFC synapses in both the 22q11.2-DS equivalent mice and DGCR8 (+/−) mice (Fénelon et al., 2011, 2013). While assessment of the function of Disc1 at synapses within the PFC has not been reported in Disc1 mutant mice, a recent study employed in utero electroporation and optogenetic techniques to selectively examine the presynaptic role of Disc1 at synapses onto cortical layer II/III pyramidal neurons (Maher and LoTurco, 2012). This study provided evidence that Disc1 may positively regulate neurotransmitter release as Disc1 knockdown led to increased paired pulse ratios (consistent with reduced Pr) as well as disrupted kinetics of evoked synaptic responses potentially consistent with less synchronized NT release. Taken together, the results of these synaptic function studies suggest a limited role of Disc1 in long-term synaptic plasticity but a potentially significant role in regulating presynaptic short-term plasticity mechanisms that would be expected to significantly impact both synaptic and network computation (Abbott and Regehr, 2004).

Calcineurin dysfunction has been implicated in the pathophysiology of schizophrenia (Gerber et al., 2003; Horiuchi et al., 2007; Liu et al., 2007) with reduced expression or function as the most likely responsible disease mechanism (Eastwood et al., 2005). Calcineurin is a cytosolic Ca²⁺-dependent protein phosphatase that regulates many cellular processes throughout the body including the CNS where is has been shown to regulate various neuronal processes including synaptic plasticity (Groth et al., 2003). Calcineurin-deficient mice display a number of behavioral abnormalities relevant to schizophrenia including severe reductions in working memory performance and deficits in the PPI of acoustic startle (Zeng et al., 2001; Miyakawa et al., 2003). Furthermore, these studies in mice with a forebrain specific calcineurin deficiency have described altered synaptic plasticity and neural network performance suggesting that hypofunction of calcineurin in schizophrenia could likewise alter synaptic plasticity and lead to disruptions of normal information processing (Cottrell et al., 2013; Suh et al., 2013).

It was originally reported that forebrain calcineurin deficiency led most notably to severe working memory deficits as well as to deficits in LTD at Schaffer collateral (CA3)—CA1 hippocampal synapses (Zeng et al., 2001). This study further showed that the frequency dependence of the stimulus required to induce long-term plasticity was shifted in calcineurin-deficient mice such that LTP (vs. LTD) induction was favored at lower than normal stimulus frequencies suggesting that deficiencies in LTD were permissive of LTP induction. Significantly, the LTD deficiency appeared to be due to an alteration of a post-synaptic mechanism since, in the genetic model employed, calcineurin was not deficient in the CA3 presynaptic neurons as these synapses. In further support of a post-synaptic mechanism, presynaptic function and short-term plasticity were unaltered as evidenced by normal PPR and post-tetanic potentiation (PTP). Although this study failed to identify the precise functional defect that led to LTD deficiency, the normal AMPA-R and NMDA-R mediated synaptic responses observed in calcineurin-deficient mice suggested possible dysfunction in downstream postsynaptic LTD induction mechanisms.

A follow-up study by this same group has recently described apparent network dysfunction in the hippocampus of freely behaving calcineurin-deficient mice (Suh et al., 2013). While in vivo recording from dorsal CA1 showed normal place fields and place cell activation during exploratory behavior, during rest periods place cells were overactive and temporally disorganized in their activity during sharp-wave ripple events (SWRs). The authors suggest that, as SWRs during rest are believed to orchestrate sequential “replay” of place cell activity in order to maintain spatial information, the absence of sequential replay of place cells during SWRs observed in calcineurin-deficient mice may underlie the severe spatial WM deficit previously described (Zeng et al., 2001). Finally, in an effort to link synaptic plasticity alterations to network dysfunction, the authors speculate that the LTD deficit present in CA1 might be responsible for the enhanced excitability within this neural circuitry that appeared to underlie the lack of organized “replay” of place cells during SWRs in calcineurin-deficient mice. Taken together the results of these two studies suggest calcineurin deficiency in schizophrenia would most likely have little impact upon short-term plasticity and synaptic filtering but would disrupt the ability to sculpt information processing by neural circuits by limiting the ability of LTD to down-regulate synaptic gain post-synaptically in order to appropriately control network excitability.

In contrast to the possible role of calcineurin in modulating postsynaptic plasticity in the hippocampus, another recent study in calcineurin-deficient mice has reported deficits in presynaptic function within the PFC (Cottrell et al., 2013). Although the cortical layer and synapse under study was not precisely clarified, these synapses in the PFC showed normal PPR and normal synaptic depression during 20 Hz stimulation; however, during 40 Hz stimulus trains calcineurin-deficient mice showed enhanced synaptic depression. Additional experiments in cultured neurons employing synapto-pHluorins revealed impaired synaptic vesicle exocytosis-endocytosis cycling suggesting that the enhanced synaptic depression was due to augmented synaptic fatigue that limited sustained NT release. Parallel experiments at hippocampal mossy fiber–CA3 synapses—where calcineurin remained expressed postsynaptically—showed similar deficits in sustained NT release indicating that presynaptic deficiency of calcineurin was responsible for the observed deficit. Further studies indicated that calcineurin-deficient mice also demonstrated neural circuit dysfunction as evidenced by reduced power of γ-oscillations in both PFC slices and in in vivo PFC recordings from mice exploring a novel environment. As γ-oscillations in the PFC are believed to be critical for working memory and other executive functions and as these high-frequency oscillations rely heavily upon sustained, high-frequency NT release, the observed deficits in network function suggest the possibility that the MW impairments in calcineurin-deficient mice may ultimately result from presynaptic NT release dysfunction. Thus, the combined results of this study provide a possible mechanism sequentially linking presynaptic NT release dysfunction, to neural circuit dysfunction, to cognitive and WM dysfunction that may be highly relevant to understanding how dysfunction in short-term synaptic plasticity could fundamentally underlie schizophrenia pathophysiology.

Taken together, the results from studies in calcineurin-deficient mice indicate prominent deficits in WM and synaptic plasticity while basic synaptic properties appear largely intact. While this series of studies suggests calcineurin may play different roles in regulating synaptic plasticity in different brain regions (postsynaptic responses in the hippocampus vs. presynaptic neurotransmitter release in the PFC), their findings consistently point toward network-dysfunction in the context of alterations in synaptic plasticity. Although none of these studies provides an unequivocal causal link between synaptic plasticity deficits and the observed network dysfunction, these results nonetheless compellingly suggest that calcineurin deficiency or hypofunction in schizophrenia may lead to similar synaptic plasticity alterations that would lead to significant changes in the information processing performed by neural circuits.

Dysfunction in NRG1 signaling pathways has been implicated in the pathophysiology of a variety of neuropsychiatric diseases including schizophrenia (Stefansson et al., 2002; Bertram et al., 2007; Thomson et al., 2007; O'Donovan et al., 2008; Goes et al., 2009; Walker et al., 2010). The manifold NRG1 gene products are EGF-like ligands that signal through ERBB receptor tyrosine kinases—most relevantly through ERBB4—to regulate a large number of cellular processes including neuronal migration, synapse formation, myelination, synaptic transmission, and synaptic plasticity and both their normal signaling and altered signaling in neuropsychiatric diseases have been comprehensively reviewed recently (Mei and Nave, 2014). Although NRG1 has been implicated in an equivocal manner as a schizophrenia risk gene—primarily by human candidate gene studies—only a very small percentage of identified risk variants affect protein coding regions raising significant questions as to whether most identified non-coding variants lead to enhanced or decreased NRG1 signaling (Chen et al., 2006; Walss-Bass et al., 2006; Mei and Nave, 2014). As there is evidence supporting either enhanced or reduced NRG1 signaling in schizophrenia, no clear picture has emerged as to whether human NRG1 disease involves hypofunction or hyperfunction and this has led to the creation of distinct mouse models with opposing types of alterations in NRG1-ERBB signaling (Hashimoto et al., 2004; Petryshen et al., 2005; Hahn et al., 2006; Bertram et al., 2007; Parlapani et al., 2010; Shibuya et al., 2010). Despite the lack of a resolution of this very important issue that critically underlies the generation of valid genetic mouse models of disease, studies in both models of altered NRG1-ERBB signaling nonetheless reveal potential roles and sites of action for these molecules to alter synaptic plasticity and network function in schizophrenia.

Mouse models of both enhanced and reduced NRG1-ERBB signally display a variety of schizophrenia related behaviors. Mice with functional knockdown of NRG1 signaling show locomotor hyperactivity, impaired PPI of startle, impaired working memory, and abnormal social behaviors (Boucher et al., 2007; Chen et al., 2008; Ehrlichman et al., 2009; Shamir et al., 2012). Intriguingly, mice overexpressing components of NRG1-ERBB signaling display a similar pattern of behavioral alterations (Deakin et al., 2009, 2012; Yin et al., 2013; Luo et al., 2014). Importantly, at least some of these behaviors appear to result from semi-acute actions of NRG1-ERBB signaling—rather than complex developmental effects that could significantly impact neural circuit architecture—as perturbing or restoring expression acutely in adulthood can cause or partially reverse behavioral alterations, respectively (Yin et al., 2013; Agarwal et al., 2014; Luo et al., 2014). That many behavioral alterations appear due to acute actions of NRG1 suggests that underlying neural circuit dysfunction may be due to alterations in synaptic transmission and plasticity.

Until recently, the majority of studies exploring the impact of NRG1 signaling have focused upon the function of glutamatergic synapses where ERBB receptors co-localize with PSD95 suggesting a likely role in the regulation AMPA and NMDA receptors (Garcia et al., 2000; Huang et al., 2000; Hahn et al., 2006). Despite this fact, both pharmacological and genetic manipulation of NRG1 signaling does not appear to significantly affect basal glutamatergic synaptic transmission (Huang et al., 2000; Kwon et al., 2005; Bjarnadottir et al., 2007; Iyengar and Mott, 2008; Pitcher et al., 2008). Similarly—and in notable contrast to the findings from most other mouse models based upon schizophrenia risk genes—the preponderance of manipulations of NRG1 signaling have not revealed alterations in short-term glutamatergic synaptic plasticity suggesting NRG1 signaling may preferentially act postsynaptically at these synapses (Huang et al., 2000; Kwon et al., 2005; Iyengar and Mott, 2008; Pitcher et al., 2008; Chen et al., 2010; but see Bjarnadottir et al., 2007; Woo et al., 2007; Deakin et al., 2012). Conversely, NRG1 signaling has a significant effect upon both induction and expression of LTP with manipulations enhancing signaling suppressing LTP while reduced signaling facilitates LTP by removing an apparent tonic suppression (Huang et al., 2000; Kwon et al., 2005; Bjarnadottir et al., 2007; Pitcher et al., 2008; Chen et al., 2010; Agarwal et al., 2014, but see Deakin et al., 2012; Jiang et al., 2013). Although alterations in LTP at these synapses are usually interpreted as due to perturbations of glutamatergic synaptic physiology, complicating this interpretation of NRG1-ERBB signaling is accumulating evidence that NRG1 regulation of hippocampal LTP may be due to more powerful NRG1 facilitating actions at hippocampal GABA-ergic synapse (Chen et al., 2010; Shamir et al., 2012).

NRG1 signaling may particularly impact GABA-ergic transmission as its principal receptor, ErbB4, is selectively expressed in interneurons, particularly within the parvalbumin positive (PV) interneurons known to be fundamental to regulating synchronized network activity (Huang et al., 2000; Bartos et al., 2007; Fisahn et al., 2009; Vullhorst et al., 2009; Fazzari et al., 2010). Enhanced NRG1 signaling increases the amplitude of GABA-ergic synaptic events, most likely through a presynaptic mechanism that increases initial Pr to facilitate GABA release that further leads to altered short-term synaptic plasticity as manifested by enhanced short-term depression (Woo et al., 2007; Chen et al., 2010; Li et al., 2011). A potentially critical importance of the regulation of synaptic function and excitability specifically in PV interneurons by NRG1 is suggested by the alterations in network activity, behavior, and glutamatergic LTP that have been observed both in response to broad alterations in NRG1 signaling as well as when these manipulations are selectively confined to PV interneurons (Woo et al., 2007; Fisahn et al., 2009; Chen et al., 2010; Li et al., 2011; Andersson et al., 2012; Deakin et al., 2012; Shamir et al., 2012; Del Pino et al., 2013).

Overall, the findings from mouse models with a variety of manipulations of NRG1 signaling show prominent deficits in WM, synchronized neural circuit dysfunction likely due to PV interneuron dysfunction, and alterations in both long-term and short-term synaptic plasticity in the context of largely normal basal synaptic function. Furthermore, while on face the prevalence of LT vs. ST synaptic plasticity alterations observed seems at odds with results from mouse models based upon other schizophrenia risk genes, recent findings suggest that these LT plasticity alterations may in fact arise indirectly from altered presynaptic Pr in GABA-ergic neurons that display altered ST plasticity. Yet interpreting the causal impact of NRG1 signaling upon neural circuit function will require much further clarification because, in addition to its more acute effects upon synaptic transmission and synaptic plasticity, its developmental roles in neuronal migration, synapse formation, and myelination further suggests that altered NRG1 signaling in schizophrenia and other disease states may also lead to significant changes in neural circuit architecture.

Reduced expression of the “regulator of g-protein signaling” protein RGS4 has been implicated in schizophrenia. Normally, RGS4 acts a negative regulator of GPCR signaling due to its GAP-activity (GTPase-activating protein) that serves to terminate cycles a G-protein signaling by converting active Gα subunits to their inactive state (De Vries et al., 2000). A number of studies have indicated that RGS4 deficiency leads to enhanced GPCR signaling and that this may ultimately lead to a decrease in the probability of NT release from presynaptic terminals (decreased presynaptic Pr). Such an alteration in presynaptic function would be expected to lead changes in short-term synaptic plasticity that would affect the computation of neural circuits suggesting that such functional changes could contribute to the dysfunction in schizophrenia.

The paper that best describes the possible synaptic mechanism and impact of RGS4 deficiency upon presynaptic function, however, is a report that examines the effects of RGS2 deficiency—rather than RGS4 deficiency—upon synaptic function in cultured hippocampal neurons (Han et al., 2006). This study shows that presynaptic RGS2 deficiency permits enhanced presynaptic mGluR signaling that leads to increased inhibition of presynaptic voltage-gated Ca²⁺ channels (most likely through direct Gβγ inhibition of N-type or P/Q-type channels). The reduced activity of these presynaptic Ca²⁺ channels in RGS2-deficient synapses results in reduced neurotransmitter release in response to APs. When RGS2 is present in presynaptic terminals these hippocampal synapses are either depressing or non-facilitating by paired-pulse paradigms, indicative of a high presynaptic Pr. RGS2-deficient synapses, however, show paired-pulse facilitation suggesting that in the absence of RGS2 the presynaptic Pr is decreased. In additional support of this action at voltage-gated Ca²⁺ channels, RGS2-deficient synapses required more extracellular Ca²⁺ for the same degree of NT release as WT synapses yet both synapses showed indistinguishable NT release in response to hypertonic sucrose challenges. The results of this study indicate that RGS2 deficiency leads to changes in presynaptic function that result in changes in short-term synaptic plasticity and further suggests that a similar scenario may likely result in cases of RGS4 deficiency.

Although whether a similar mechanism of alteration in synaptic function occurs in cases of RGS4 deficiency has not been reported, two studies support this possibility by showing that RGS4 has a functional relationship with voltage-gated Ca²⁺ channel activity analogous to that of RGS2. Rather than exploring effects upon synaptic transmission, however, these two studies that look at RGS4 regulation of voltage-gated Ca²⁺ channels examine the resulting impact upon action potentials (Saugstad et al., 1998; Ding et al., 2006). One study found that conditions that enhanced expression of RGS4 in medium spiny neurons (MSNs) of the striatum led to the removal of muscarinic GPCR inhibition of voltage-gated Ca²⁺ channel activity. The enhanced voltage-gated Ca²⁺ channel activity that resulted from increases in RGS4 expression permitted enhanced Ca²⁺ currents during APs that recruited activation of Ca²⁺-activated K⁺-channels that served to regularize AP rhythms in these striatal neurons. The results of this paper indicate that RGS4 normally serves to reduce voltage-gated Ca²⁺ channel inhibition and suggest that RGS4 deficiency would lead to reduced voltage-gated Ca²⁺ activity.

A second study found that the normal suppression of Ca²⁺-activated K⁺ currents by mGluR activation in CA1 hippocampal pyramidal neurons could be nearly abolished by inclusion of RGS4 in the recording pipette (Saugstad et al., 1998). Although this study did not directly show that the RGS4 enhancement of Ca²⁺-activated K⁺ currents was due to removal of mGluR mediated inhibition of voltage-gated Ca²⁺ channels, this would seem to be the likely mechanism. Taken together, the results of these two studies supporting RGS4-mediated enhancement of voltage-gated Ca²⁺ channel activity, although they did not examine synaptic physiology, support the possibility that RGS4 deficiency could lead to a reduced voltage-gated Ca²⁺ channel activity in presynaptic terminals that would result in reduced NT release and altered short-term synaptic plasticity.

A final study directly explored the impact of RGS4 deficiency upon synaptic plasticity at MSNs in the striatum by employing RGS4-deficient mice (Lerner and Kreitzer, 2012). This study examined an endocannabinoid (eCB) mediated LTD (eCB-LTD) at MSN synapses that is dependent upon mGluR-Gq mediated generation of eCB in the post-synaptic neuron that travels to the presynaptic neuron to inhibit presynaptic NT release by lowering Pr. Normally this LTD pathway can be modulated by the activation of additional postsynaptic GPCRs, but in RGS4-deficient mice this modulation is absent. Through a series of elegant experiments this study showed that this GPCR modulation of LTD occurs through alterations in cAMP/PKA modulation of RGS4 activity that alters the ability of RGS4 to suppress mGluR-dependent production of the eCB retrograde messenger responsible for the LTD. Thus, although by a much more complicated mechanism than described for RGS2, RGS4 deficiency similarly results in alterations in presynaptic NT release probability (Pr). Further, although this study did not directly address changes in ST synaptic plasticity, the inability to modulate presynaptic Pr due to RGS4 deficiency would be predicted to lead to a decreased ability to modulate ST plasticity at this synapse as well.

Taken together, the results of these four studies suggest that the action of RGS4 should normally serve to increase presynaptic NT release probability and that RGS4 deficiency would lead to reductions in Pr causing reduced NT release. While only LT plasticity consequences have been directly explored in RGS4-deficient mice, the results of these studies supporting that presynaptic Pr is a likely target for RGS4 modulation strongly suggest that ST synaptic plasticity—and consequently neural circuit computation—would be strongly impacted by RGS4 deficiency in schizophrenia.

Reductions in dysbindin expression and function have been implicated in schizophrenia and results from a number of studies point toward a role for dysbindin in regulating presynaptic short-term plasticity by modulation of the probability of neurotransmitter release from synaptic terminals. Although a detailed understanding of dysbindin function remains to be determined, its expression in central synaptic terminals in the mammalian CNS and its association with synaptic vesicles suggests a possible role in the regulation of NT release (Benson et al., 2001; Talbot et al., 2006).

Perhaps the most informative study of dysbindin function was performed at the glutamatergic neuromuscular junction (NMJ) in Drosophila (Dickman and Davis, 2009), a system whose amenability to genetic manipulation permits rapid clarification of the detailed molecular mechanisms of gene action. This study demonstrated that presynaptic dysbindin deficiency led to reduced Ca²⁺-sensitivity of NT release, enhanced paired-pulse facilitation of post-synaptic responses, enhanced facilitation of post-synaptic responses during trains of stimuli, and a failure of homeostatic upregulation of the number of NT vesicles released per stimulus (quantal content) upon postsynaptic receptor blockade. Each of these results is consistent with a role of dysbindin in enhancing the Pr of presynaptic NT release (and dysbindin deficiency leading to reduced NT release), most likely through an action of dysbindin that enhances the Ca²⁺ sensitivity of the NT release machinery. This study convincingly illustrated that dysbindin deficiency altered ST plasticity by transforming a modestly facilitating synapse into a strongly facilitating synapse. This finding supports the possibility that dysbindin-deficient synapses in the mammalian CNS would have similar alterations in ST plasticity that would be expected to significantly alter synaptic and neural circuit computation.

A recent study of synaptic transmission onto layer V pyramidal neurons in the mPFC in dysbindin-null mice reported a number of findings consistent with the altered presynaptic function described at the Drosophila NMJ (Saggu et al., 2013). Reduced amplitudes of evoked EPSCs, reduced frequency of mEPSCs, a reduced RRP of synaptic vesicles, reduced SV cycling upon depolarization (as determined by FM1-43 studies), and decreased Ca²⁺ flux into synaptosomes upon depolarization all consistently point toward a decrease in the presynaptic Pr of NT release in dysbindin-deficient synapses mice. Furthermore, although not reaching statistical significance, this study performed assays to determine Pr more directly and showed a ~35% reduction in presynaptic Pr in dysbindin-deficient mice. Thus, although this study did not directly examine either short-term or long-term plasticity by traditional electrophysiological assays, taken together these findings strongly support a reduced presynaptic Pr due to dysbindin deficiency that would be expected to lead to alterations in short-term plasticity in neural circuits in the mammalian mPFC.

Despite the strong support for dysbindin deficiency leading to decreased presynaptic function through reduced Pr (Dickman and Davis, 2009; Saggu et al., 2013), a recent study in the hippocampus of dysbindin-deficient mice did not see alterations in ST plasticity (Tang et al., 2009). This study of Schaffer collateral input to CA1 in the hippocampus reported a nearly doubled magnitude of LTP at these synapses in dysbindin-deficient mice but both normal paired-pulse facilitation and normal LTD. Yet, taken together, these studies of synaptic function in dysbindin-deficient animals suggest that the most likely role of dysbindin is in regulating presynaptic NT release and that dysbindin deficiency leads to decreases in NT release by reducing presynaptic Pr. Furthermore, as presynaptic Pr is the prime determinant controlling short-term plasticity at synapses, dysbindin deficiency in schizophrenia would be predicted to most significantly alter short-term plasticity and lead to alterations in neural circuit computation that could underlie disease symptomology.

Reductions in AKT1 expression and function have been implicated in schizophrenia in a number of recent studies (Emamian et al., 2004; Zhao et al., 2006; Balu et al., 2012). AKT is a protein kinase that has many cytosolic protein targets in a variety signaling pathways both in the CNS as well as in non-neuronal tissues (Yang et al., 2004). AKT1-deficient mice show alterations in the dendritic architecture of mPFC pyramidal neurons as well as working memory deficits that are believed to be representative of analogous deficits found in schizophrenia (Lai et al., 2006). Although the role AKT1 in neuronal and synaptic function remains under-explored, two studies in animal models with reduced AKT1 expression have revealed alterations in synaptic plasticity (Guo and Zhong, 2006; Balu et al., 2012).

While neither of these two studies provides a clear mechanistic explanation of the plasticity alterations observed in AKT1-deficient animals, the study of the glutamatergic synapse at the NMJ of Drosophila provides a more comprehensive analysis (Guo and Zhong, 2006). In this study the authors describe a novel long-term depression (LTD) evoked by high frequency stimulation that quantal analysis revealed to be presynaptic in origin as the quantal size remained unchanged but the quantal content (reflecting the number of released neurotransmitter vesicles per stimulated event) was reduced. Although basic NMJ responses were normal in AKT mutant flies, this novel LTD was significantly reduced in mutant flies with low expression of AKT while normal LTD could be rescued by acute over-expression of WT AKT. Despite the evidence of a presynaptic locus for this LTD, in AKT mutant flies with deficient LTD, measures of baseline presynaptic function such as short-term depression, short-term facilitation, and PTP were found to be normal. Given that these results support that the observed LTD in WT flies was due to reduced presynaptic Pr and that all forms of baseline presynaptic function in AKT-deficient flies were normal—including inferred presynaptic Pr—suggests a role of AKT in the induction of changes in presynaptic Pr during LTD protocols that is reduced or absent in AKT-deficient flies. Interestingly, such a mechanistic deficiency is somewhat similar to the deficient control of presynaptic Pr upon eCB-LTD induction in the striatum of RGS4-deficient mice where RGS deficiency impacts the ability to modulate presynaptic Pr (Lerner and Kreitzer, 2012). Regardless of the details of the mechanism underlying deficient LTD at the NMJ in AKT-deficient flies, an inability to appropriately reduce presynaptic Pr by long-term plasticity mechanisms would consequently lead to an inability to alter presynaptic short-term plasticity normally at AKT-deficient synapses.

In a separate study in AKT1-deficient mice deficits in long-term plasticity were observed at multiple synapses in the hippocampus (Balu et al., 2012). AKT1-deficient mice showed deficits in LTP in CA1 upon Schaffer collateral stimulation (SC-CA1) and deficits in LTP in the dentate gyrus upon medial perforant path stimulation (EC-GC). At both of these sites short-term plasticity of synaptic responses during the tetanus were normal in AKT1-deficient mice suggesting no major alteration in baseline presynaptic short-term plasticity mechanisms. The post-tetanic short-term potentiation, however, although normal at SC-CA1 synapses, was completely absent at EC-GC synapses suggesting a more severe plasticity deficit at the synapses onto granule cells in the dentate gyrus. Despite the profound deficits in synaptic plasticity described in AKT1-deficient mice in this report, given that the mechanisms underlying these deficits were not further experimentally pursued it is difficult to speculate upon the precise role of AKT1 function at these synapses.

Taken together the results of these two studies in AKT-deficient animals indicate that AKT1 may play an significant role in regulating long-term synaptic plasticity and that AKT1 deficiency in schizophrenia may result in similar long-term plasticity alterations that could contribute to disease symptomology.

Genomic studies of neuropsychiatric disease have identified a surprisingly large and diverse array of risk genes. Despite this genetic heterogeneity, the fact that identified risk genes are preferentially involved in synaptic function, growth, and maintenance as well as axonal guidance suggests that genetic risk factors for neuropsychiatric disease may converge upon the common substrate of neural circuit function (Walsh et al., 2008; Glessner et al., 2009; Guilmatre et al., 2009; Akil et al., 2010; Insel et al., 2010; Gilman et al., 2011; O'Dushlaine et al., 2011; Sanders et al., 2011). In addition, despite heterogeneity, a small number of rare but highly penetrant genetic risk variants have been identified that reliably lead to disease that is clinically indistinguishable from the majority of disease cases (Farooqi and O'Rahilly, 2005; Peltonen et al., 2006). Although these highly penetrant genetic variants are usually only responsible for a very small percentage of cases of the broader disease, their high reliability to cause disease renders them invaluable, essential tools in creating mouse models that faithfully recapitulate disease. While a small number of such high-risk genetic variants have been identified in cases of schizophrenia, autism, and AD that have enabled the creation of reliable disease models in mice, for many common neuropsychiatric diseases—e.g., major depression, bipolar disorder, anxiety disorders, attention-deficit hyperactivity disorder—similarly high-risk genetic variants that reliably lead to human disease have yet to be identified thus leaving valid genetic mouse models of these diseases currently out of reach (Pulver et al., 1994; Hardy, 1997; Murphy et al., 1999; Millar et al., 2000; Blackwood et al., 2001; Hatton et al., 2006; Clifford et al., 2007; Ramocki et al., 2009; Vassos et al., 2010; Neul, 2012). While the genetic selective pressures—such as the strong selection pressures against retaining risk genes for socially debilitating and early-onset diseases like autism and schizophrenia that largely preclude successful reproductive transmission—that shape the genetic architecture of disease may be substantially different for diseases such as major depression, it nonetheless seems reasonable to posit (and hope) that high-risk genetic variants for these diseases in fact exist yet have eluded detection and remain to be found. Based upon these considerations, creating reliable genetic mouse models of disease must begin with the identification of highly penetrant genetic variants and in the case of neuropsychiatric diseases in which such genetic variants remain to be identified our understanding of the details of neural circuit dysfunction in these diseases likely must patiently await this first crucial step.

Further complicating the understanding of the causality between genes and disease, genomic studies have revealed a surprising overlapping genetic risk among multiple neuropsychiatric diseases that raises significant questions as to the validity of the boundaries between these diseases (Craddock and Owen, 2005; Craddock et al., 2006; Friedman et al., 2008; Guilmatre et al., 2009; McCarthy et al., 2009; O'Donovan et al., 2009; Williams et al., 2010; Smoller et al., 2013). That is, in many cases the same genes implicated in increased risk for one neuropsychiatric disease have been shown to predispose to increased risk for multiple other distinct neuropsychiatric diseases. Furthermore, although some genetic variants appear to lead to narrower, better-defined clinical phenotypes, genomic studies have also revealed that in some cases the same genetic variant can be found in patients with different clinical diagnoses (St Clair et al., 1990; Hatton et al., 2006; Bijlsma et al., 2009; McCarthy et al., 2009; Ramocki et al., 2009; Shi et al., 2009; Wang et al., 2009; Karayiorgou et al., 2010; Shinawi et al., 2010). The potential significance of these results is further supported by heritability studies that have shown that within family pedigrees the presence of one neuropsychiatric illness is frequently associated with cross-vulnerability to other neuropsychiatric illnesses (Biederman et al., 1991; Maier et al., 1993; Lichtenstein et al., 2009b, 2010; Rommelse et al., 2010; Sullivan et al., 2012). Furthermore, beyond considering these concerns of genetic promiscuity among distinct neuropsychiatric diseases further suggests that beyond choosing highly penetrant genetic variants, the clearest path to understanding distinct diseases through genetic mouse models may further require carefully choosing among genetic variants for those associated with narrow, well-defined clinical presentations.

Beyond these concerns, however, the additional overlapping symptomology and overlapping responsiveness to medication found between some diseases raises the legitimate possibility that the boundaries between “distinct” diseases may be inherently blurred due to a true shared causation (Milberger et al., 1995; Citrome et al., 2005; Lake and Hurwitz, 2007; O'Donovan et al., 2009; Sinzig et al., 2009; Ivleva et al., 2010; Gros et al., 2012; Zbozinek et al., 2012). For example, while NMDA receptor hypofunction has been implicate in the causation of schizophrenia, recent studies in rodent models support NMDA receptor hypofuction in autism suggesting the possibility of a continuum of related dysfunction between and among distinct diseases (Mohn et al., 1999; Gandal et al., 2012; Saunders et al., 2012; Schmeisser et al., 2012; Won et al., 2012). Taken together, these multiple issues of overlapping disease phenomenology suggest the possibility that the traditional diagnostic criteria used to create distinct boundaries between diseases based upon behavioral manifestations—while clinically useful in disease management—may not be ideally suited as a scientific framework from which to determine underlying disease mechanisms which may instead require the development of more refined and quantifiable measures targeted at directly assessing endpoints related to brain function including the function of neural circuits (Geyer, 2006; Gould and Gottesman, 2006; Benes, 2007; Psychiatric GWAS Consortium Coordinating Committee, 2009; Luck et al., 2011; Smoller et al., 2013).

Unlike traditional physiological diseases such as cardiovascular disease, diabetes, or obesity, the understanding of neuropsychiatric disease processes is currently hampered by the underdevelopment and underutilization of analytic measures that provide simple yet precise quantitative readouts of the function and status of the underlying physiological systems that directly contribute to disease pathogenesis (Gould and Gottesman, 2006). For example, psychiatric diagnosis and evaluation as put forth by the DSM-V is based upon a checklist of behavioral signs and symptoms that are considered either present or absent in a binary manner and are far removed from any direct quantitative measure of underlying brain or neural circuit function (American Psychiatric Association, 2013). Furthermore, most current evaluative measures in neuropsychiatric disease are inherently complex—e.g., evaluation of complex phenomena such as social function, judgment, and language—and do not obviously point to precisely circumscribed loci of neural circuit dysfunction. Due to this complexity, these measures—while both interesting and clinically relevant in terms of monitoring the progression and the functional impact of disease—are not terribly useful as scientific bases for a reductionist approach aimed at revealing fundamental disease mechanisms (Insel et al., 2010). Moreover, many of these complex manifestations seem to be uniquely human in nature do not appear to present obvious translatable equivalents that could readily permit correlative evaluation in mouse models. These considerations highlight the intense need in neuropsychiatric diseases for the development of more simplified and precise parameter assays that directly and quantitatively report upon underlying pathophysiological processes such as neural circuit dysfunction (Gould and Gottesman, 2006; Akil et al., 2010; Insel et al., 2010).

Analysis of disease related endophenotypes may represent the most effective strategy to provide the simplified and precisely quantitative functional measures required to efficiently dissect fundamental disease mechanisms. Endophenotypes are relatively simple, quantifiable, and heritable characteristics present in both individuals diagnosed with disease and in unaffected members within their familial pedigrees and are therefore believed to represent disease vulnerability traits (Gould and Gottesman, 2006). Compared to “top level” symptoms in neuropsychiatric disease that often appear to be uniquely human in nature without clear parallels in mouse models, the more rudimentary nature of endophenotypes present the further advantage that their assays are frequently readily translatable into mouse models (Geyer, 2008; Ecker and Murphy, 2014). In consensus recognition of the need for more quantifiable measures in neuropsychiatric disease, an initial, though partial, effort to address these issues strategically has recently begun to be laid out by the NIMH in the form of their proposed Research Domain Criteria (RDoC) project initiative.

While assessment of endophenotypes frequently provides little direct clinical utility, their quantification is scientifically valuable as endophenotypes are believed to represent the most fundamental consequences of the basic pathological processes underlying the more complex manifestations neuropsychiatric diseases. From the standpoint of assessing suspected neural circuit dysfunction in neuropsychiatric disease, particularly appealing among endophenotypes are the neurophysiological measures which quantify rather stereotyped responses to very simplified stimuli (Turetsky et al., 2007; Luck et al., 2011). These assays, through assessing the dysfunction of simplified neural circuits involved with very rudimentary information processing, present an effective and strategic platform from which to reveal the cellular and molecular details of the neuronal and synaptic alterations underlying neural circuit dysfunction in neuropsychiatric disease. To better appreciate the refined analytical power and broad utility such assays present for guiding a detailed mechanistic understanding of neural circuit dysfunction in neuropsychiatric disease, it is useful to specifically examine the strategic approach to understanding neural circuit dysfunction in mouse models of schizophrenia.

Currently the most prevalent approach to understanding neural circuit dysfunction in genetic mouse models of schizophrenia has involved assessing working memory deficits in these animals coupled with electrophysiological recordings from the underlying brain regions believed to contribute most to working memory performance. While not as striking as the positive symptomology of schizophrenia, the pervasive cognitive deficits in schizophrenia have presented a tractable experimental endophenotype for examining disease-relevant alterations in behavior and neural circuit function in genetic mouse models (Glahn et al., 2003; Heinrichs, 2005; Saperstein et al., 2006; Forbes et al., 2009; Arguello and Gogos, 2010, 2012). As such, assessment of working memory in genetic mouse models of schizophrenia has been commonly employed and any working memory deficits detected have been interpreted as reflecting dysfunction of neural circuits in either the hippocampus or the medial prefrontal cortex (Arguello and Gogos, 2010, 2012). Motivated by such findings these brain regions have then been subsequently probed more directly with electrophysiological assays in attempts to detect and localize neuronal dysfunction within specific neurons or synapses and to determine the precise mechanisms underlying this dysfunction and this approach has successfully revealed—consistent with WM deficits—prominent alterations in short-term synaptic plasticity (Zeng et al., 2001; Kvajo et al., 2008, 2011; Mongillo et al., 2008; Jentsch et al., 2009; Sigurdsson et al., 2010; Cottrell et al., 2013; Fénelon et al., 2013).

Despite the strong validity, disease-relevance, and translational potential of working memory tasks as a framework from which to assess and reverse neuronal and synaptic dysfunction in mouse models, the inherent complexity of such tasks limits their potential to identify and precisely localize underlying neuronal dysfunction (Arguello and Gogos, 2010, 2012). Specifically, during a spatial working memory task information processing in disparate brain regions involving sensory, motor, motivational, volitional, appetitive, and fear and anxiety related neural circuits is likely influencing behavioral performance in addition to the neural circuit activation within the mPFC and hippocampus (Gisquet-Verrier and Delatour, 2006; McNab and Klingberg, 2008; Edin et al., 2009; Baddeley, 2012).

As the ultimate goal of investigations in genetic mouse models of neuropsychiatric disease is to reveal the precise mechanisms of neuronal and synaptic dysfunction underlying the neural circuit dysfunction in these diseases, the optimal assays of in vivo neural circuit function should ideally be as simple and uncomplicated as possible such that any dysfunction observed can be confidently and unambiguously attributed to a distinct neural circuit or brain region that can then be directly targeted with invasive, high-resolution electrophysiological assays. Although finding a functional measure that can be purely attributed to a single, well-defined neural circuitry is a non-trivial hurdle, a number of schizophrenia-related neurophysiological endophenotypes involving sensory evoked responses may come the closest to fitting this requirement (Turetsky et al., 2007; Thaker, 2008). While the study of the neural circuits underlying these endophenotypes has yet to be aggressively pursued, due to their exquisite simplicity these endophenotypes may represent the most tractable and efficient analytical path toward understanding the discrete neuronal and synaptic mechanisms underlying neural circuit dysfunction in well-defined and valid genetic animal models of schizophrenia (Goff et al., 1980; Javitt et al., 1994; Freedman et al., 1996; Swerdlow et al., 2001; Grunwald et al., 2003; Kumari et al., 2003; Schall et al., 2003; Luck et al., 2011).

Specifically, among the many neurophysiological assays showing characteristic alterations in schizophrenics, the sensory evoked response endophenotypes involving PPI of acoustic startle, P50 auditory evoked potential suppression (P50), and acoustic mismatch negativity (MMN) appear to reflect information processing abnormalities within relatively simple and isolated neural circuits (Clements et al., 1998; Bramon et al., 2004; Light and Braff, 2005; Swerdlow et al., 2006; Brockhaus-Dumke et al., 2008). From a reductionist point of view, the most compelling characteristic of these three endophenotype assays is that they involve neural circuit processing that is either entirely or largely preattentive—and thus nearly reflexive—in nature (Naatanen, 1992). These measures thus appear to represent read-outs of rudimentary neural circuit computation with minimal contaminating influences from other neural circuitry that might otherwise confound interpretation of results (Naatanen, 2003; Braff and Light, 2004). As such, these assays of neural circuit function that are largely devoid of the complications of attentional, motivational, or cognitive control may be optimal assays of neural circuit function in mouse models of schizophrenia—through use of analogous assays in rodents that have been developed for each of these measures—as they examine hard-wired, relatively inflexible computational elements that present a greatly simplified system in which to explore the discrete neuronal and synaptic mechanisms underlying neural circuit dysfunction (Swerdlow et al., 2001; Gould and Gottesman, 2006; Metzger et al., 2007; Geyer, 2008; Javitt et al., 2008; Ehrlichman et al., 2009; Luck et al., 2011). Furthermore, as these neurophysiological assays employ either pairs or trains of acoustic stimuli delivered at very brief intervals, these assays bear a striking resemblance to electrophysiological assays that employ similarly brief intervals of stimulation to determine short-term plasticity properties of synapses suggesting the possibility that abnormalities observed in neurophysiological assays may reflect underlying alterations in short-term synaptic plasticity (Freedman et al., 1996; Garrido et al., 2009; Blundell et al., 2010). In summary, while no neural circuit exists in isolation from the influence of other circuits within the brain, the neurophysiological assays of preattentive schizophrenia endophenotypes that activate rudimentary neural circuits come close to this ideal and may represent the optimal assays to guide the elucidation of the synaptic and neuronal mechanisms underlying neural circuit dysfunction in valid genetic mouse models of schizophrenia.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.