Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Mar 2.
Published in final edited form as: Neuron. 2016 Feb 18;89(5):1100–1109. doi: 10.1016/j.neuron.2016.01.025

Developmental inhibition of Gsk3 rescues behavioral and neurophysiological deficits in a mouse model of schizophrenia predisposition

Makoto Tamura 1,4, Jun Mukai 2,3, Joshua A Gordon 1,5, Joseph A Gogos 2,3
PMCID: PMC4783174  NIHMSID: NIHMS761735  PMID: 26898776

SUMMARY

While the genetic basis of schizophrenia is increasingly well-characterized, novel treatments will require establishing mechanistic relationships between specific risk genes and core phenotypes. Rare, highly-penetrant risk genes such as the 22q11.2 microdeletion are promising in this regard. Df(16)A+/− mice, which carry a homologous microdeletion, have deficits in hippocampal-prefrontal connectivity that correlate with deficits in spatial working memory. These mice also have deficits in axonal development that are accompanied by dysregulated Gsk3β signaling and can be rescued by Gsk3 antagonists. Here, we show that developmental inhibition of Gsk3 rescues deficits in hippocampal-prefrontal connectivity, task-related neural activity, and spatial working memory behavior in Df(16)A+/− mice. Taken together, these results provide mechanistic insight into how the microdeletion results in cognitive deficits, and suggest possible targets for novel therapies.

INTRODUCTION

Cognitive impairments in schizophrenia are a core feature of the disease (Barch and Ceaser, 2012) and a key determinant of long-term outcome (Fett et al., 2011; Hofer et al., 2005). Yet these impairments are mostly resistant to available treatments (Buoli and Altamura, 2015; Lett et al., 2014; Millan et al., 2012); greater understanding of their underlying neurobiology could facilitate the development of novel therapies (Arguello and Gogos, 2010). To this end, we have studied the neural basis of cognitive deficits in Df(16)A+/− mice, which model a microdeletion on chromosome 22q11.2 that accounts for up to 1%–2% of sporadic schizophrenia cases (Karayiorgou et al., 1995; Xu et al., 2008). Human carriers of the 22q11.2 microdeletion show a range of cognitive deficits, including working memory impairments (Karayiorgou et al., 2010), and ~30% of them develop schizophrenia (Gothelf et al., 2007; Green et al., 2009; Murphy et al., 1999). Df(16)A+/− mice show impairments in spatial working memory, among other schizophrenia-related phenotypes (Stark et al., 2008). These impairments in working memory correlate with deficits in functional connectivity between the hippocampus (HPC) and medial prefrontal cortex (mPFC) in the mutant animals (Sigurdsson et al., 2010), consistent with the literature suggesting these two structures cooperate during spatial working memory (Jones and Wilson, 2005; Lee and Kesner, 2003). Furthermore, direct inhibition of HPC-mPFC inputs in wild-type (WT) mice causes similar impairments in working memory performance, as well as deficits in the encoding of spatial information within the prefrontal cortex (Spellman et al., 2015). Taken together, these findings suggest a role for the HPC-mPFC circuit in the spatial working memory deficits associated with the 22q11.2 microdeletion.

Attempts to define the molecular and cellular basis of the cognitive and behavioral deficits associated with the 22q11.2 microdeletion have focused on the effects of single gene mutations within the 22q11.2 microdeletion region (Bender et al., 2005; Chun et al., 2014; Fenelon et al., 2013; Harper et al., 2012; Hiramoto et al., 2011; Hsu et al., 2007; Huotari et al., 2004; Mukai et al., 2004; Murphy et al., 1999; Paterlini et al., 2005; Xu et al., 2013). Recently, we demonstrated that mice haploinsufficient for one of these genes, Zdhhc8+/−, recapitulate several key aspects of the full microdeletion, including deficits in spatial working memory and functional connectivity (Mukai et al., 2015). Zdhhc8 haploinsufficiency results in impairments in neurite outgrowth in vitro and axon branching in vivo (Mukai et al., 2015). Axon branching deficits are seen in both cortico-cortical and hippocampo-cortical pathways, are present in mice carrying the full microdeletion, and are accompanied by deficits in synaptic efficacy (Mukai et al., 2015). The Zdhhc8 gene product is a palmitoylation enzyme, and haploinsufficiency results in the mislocalization of several proteins required for normal axonal development. One downstream result of this mislocalization is a reduction in phosphorylated Akt at the axon tip. Phosphorylated Akt normally inhibits Gsk3β, suggesting that increased activity of Gsk3β might contribute to the axonal phenotypes seen in the mutant mice. Indeed, developmental inhibition of Gsk3 restored normal patterns of neurite outgrowth and axonal branching in Zdhhc8+/− mice (Mukai et al., 2015).

These studies raise two important questions. First, are the axonal, physiological and behavioral phenotypes causally related? If so, developmental antagonism of Gsk3 should rescue HPC-mPFC synchrony and working memory performance in Df(16)A+/− mice. Second, what is the impact of the microdeletion, and its rescue by Gsk3 antagonism, on task-related activity in the mPFC? Our previous results (Spellman et al., 2015) would suggest that, because of reduced efficacy of HPC-mPFC input, mPFC neurons in Df(16)A+/− should have reduced spatial representations. In that case, could these altered representations be restored by Gsk3 antagonism? Here we demonstrate that developmental antagonism of Gsk3 indeed rescues impaired neural synchrony, working memory performance, and mPFC spatial representations in Df(16)A+/− mice, consistent with the notion that spatial working memory deficits in these mice are caused by disruptions in the hippocampal-prefrontal circuit as a consequence of disrupted axonal development. Our findings provide crucial evidence that these working memory-related phenotypes are pharmacologically reversible, and suggest a possible pathway towards novel therapies for cognitive deficits in 22q11.2-microdeletion carriers and at least a subset of patients with schizophrenia.

RESULTS

To investigate the effect of developmental inhibition of Gsk3β on behavioral and neurophysiological deficits in Df(16)A+/− mice, mutants and WT littermates were injected with SB-216763 (SB), a potent and selective inhibitor of Gsk3 (Kramer et al., 2012), or vehicle every other day from P7 to P28. At 3–5 months of age, mice were implanted with microelectrodes targeting mPFC, ventral hippocampus (vHPC) and dorsal hippocampus (dHPC) (Figure S1). Local field potential (LFP) and multiple single-unit recordings were obtained from these areas at baseline, as well as during acquisition and performance of a delayed non-match-to-place (DNMTP) T-maze test of spatial working memory (Figure 1A). Similar to our previous findings (Sigurdsson et al., 2010; Stark et al., 2008), Df(16)A+/− mice took significantly longer to reach criterion performance of the task, defined as more than 70% correct for three consecutive days (Figure 1B). This behavioral deficit was reversed by developmental inhibition of Gsk3, such that days to criterion did not significantly differ between WT and SB-treated Df(16)A+/− mice. SB had no effect on acquisition in WT mice.

Figure 1. Developmental inhibition of Gsk3 rescues the spatial working memory deficit in Df(16)A+/− mice.

Figure 1

Open in a new tab

(A) Spatial delayed non-match to sample T-maze task. Each trial of the task comprised a sample and choice phase separated by a 10 second delay. R: reward.

(B) Acquisition (days to criterion) as a function of genotype and treatment. Student’s t-test. n = 11 (WT + vehicle), 12 (Df(16)A+/− + vehicle), 13 (Df(16)A+/− + SB) and 8 (WT + SB). Error bars indicate +/− s.e.m., *p<0.05, **p<0.01, throughout.

We have previously reported that Df(16)A+/− mice have impaired oscillatory synchrony between the dHPC and mPFC across multiple frequency ranges (Sigurdsson et al., 2010). More recently, we have revealed that mPFC-vHPC connectivity is also closely related to the acquisition and encoding of spatial working memory (Mukai et al., 2015; O’Neill et al., 2013; Spellman et al., 2015), and is specifically disrupted in Zdhhc8+/− deficient mice (Mukai et al., 2015). Here, we extend these observations to vehicle-treated Df(16)A+/− mice, which have substantially reduced vHPC-mPFC synchrony at baseline, during habituation to the T-maze (Figure 2). This decreased coherence was evident in both theta- and gamma-frequency ranges, whether measured by coherence of mPFC and vHPC local field potentials (LFPs) (Figures 2A and 2B) or phase-locking of mPFC single units to vHPC LFP oscillations performed using pairwise phase consistency (PPC) (Figures 2C – 2F) and mean resultant length (Figure S2D). Developmental inhibition of Gsk3 significantly rescued theta-frequency coherence and phase-locking in Df(16)A+/− mice, without affecting synchrony in WTs (Figures 2B and 2D). SB-treated Df(16)A+/− mice also showed trends toward higher gamma-range coherence (p=0.10; Figure 2B) and phase-locking of mPFC units to vHPC gamma (p=0.06; Figure 2F) than untreated Df(16)A+/− mice. Similar to our previous finding (Sigurdsson et al., 2010), Df(16)A+/− mice showed impaired synchrony between dHPC and mPFC (Figures S2A–C), although statistical significance was achieved only for phase-locking of mPFC units to dHPC theta oscillations (Figures S2C and S2E). SB treatment did not significantly reversed impaired dHPC-mPFC synchrony regardless of frequency, though trend-level effects were observed (Figures S2B and S2C). LFP power at baseline did not differ by genotype or treatment regardless of frequency or region (Figures S3A–F).

Figure 2. Gsk3 inhibition rescues deficits in theta-frequency synchrony.

Figure 2

Open in a new tab

(A) Coherence spectra between vHPC and mPFC by group. Shaded area indicates +/− s.e.m.

(B) Theta- and gamma-range coherence between vHPC and mPFC by group. Theta: p = 0.0037 (WT vs. Df(16)A+/−), p = 0.014 (Df(16)A+/− vs. Df(16)A+/− + SB). Gamma: p = 0.024 (WT vs. Df(16)A+/−), p = 0.10 (Df(16)A+/− vs. Df(16)A+/− + SB), Student’s t-test. n = 10 (WT + vehicle), 10 (Df(16)A+/− + vehicle), 12 (Df(16)A+/− + SB) and 6 (WT + SB).

(C) Distributions of vHPC theta phases at which action potentials were observed for representative mPFC neurons of each genotype. PPC values are 0.0083 (WT), 0.00007 (Df(16)A+/− ), and 0.0045 (Df(16)A+/− + SB).

(D) Strength of phase-locking of mPFC neurons to theta oscillations for all recorded neurons. p = 0.036 (WT vs. Df(16)A+/−), p = 0.013 (Df(16)A+/− vs. Df(16)A+/− + SB), Student’s t-test. n = 135 (WT), 89 (Df(16)A+/−), and 163 (Df(16)A+/− + SB) units.

(E) Distributions of vHPC gamma phases at which action potentials were observed for representative mPFC neurons of each genotype. PPC values are 0.00041 (WT), −0.00026 (Df(16)A+/− ), and 0.000098 (Df(16)A+/− + SB).

(F) Strength of phase-locking of mPFC neurons to and gamma oscillations for all recorded neurons. p = 0.027 (WT vs. Df(16)A+/−), p = 0.063 (Df(16)A+/− vs. Df(16)A+/− + SB), Student’s t-test. Sample sizes as in (D).

See also Supplementary Figures 1, 2 and 3.

Neurons within the mPFC typically have task-related firing patterns (Baeg et al., 2003; Fujisawa et al., 2008; Jung et al., 1998). In the DNMTP task used here, mPFC neurons represent the spatial location of the goal (Spellman et al., 2015). We speculated that this goal representation might be disrupted in Df(16)A+/− mice, and that this disruption might be reversed by developmental Gsk3 antagonism. Consistent with prior findings, a subset of mPFC single units fired preferentially on one of the two goal arms during the choice (retrieval) phase of the T-maze task (Figures 3A and 3B). A repeated measures analysis of variance (rmANOVA) based on binned spike rates was used to determine whether each unit’s firing rate varied by goal arm, as a function of time within the sample of phase. In WT mice, the percentage of mPFC units showing statistically significant goal-selective firing increased as the animal entered the goal arm, peaking at 21+/−2% near the goal location (Figure 3C). Fewer units in Df(16)A+/− mice showed this goal selectivity, peaking at 12+/−2% (Figure 3C). Developmental inhibition of Gsk3 restored goal selectivity in the mutant animals, without affecting the representation in WT (peaking at 22+/−3% and 26+/−3%, respectively; Figure 3C). Overall firing rates did not differ by group (Figure 3D), indicating that developmental inhibition of Gsk3 rescued spatial representations in mPFC neurons of Df(16)A+/− mice without affecting their overall level of excitation. Similar tendencies were seen in the sample (encoding) phase, though group differences were not significant (Figure S4A). In WT mice, 50 +/− 4.5% of goal-selective units in the sample phase showed significant goal-selective firing in the choice phase (Figure S4C). This percentage in Df(16)A+/− mice reduced to 26 +/− 3.1%. SB treatment significantly restored the similarity of goal-selective firing between task phases without affecting WT mice (Figure S4C). Goal-selective units during the sample phase tended to preferentially fire in the same arm during the choice phase (90% of units; Figure S4D). This preference was decreased to 74% in Df(16)A+/− mice and restored to 89% by SB treatment. The high reliability of arm preference across task phases argues for spatial as opposed to reward-related activity, since the preferred arm could be either rewarded or not rewarded in the choice phase.

Figure 3. Gsk3 inhibition restores goal-arm encoding in Df(16)A+/− mice.

Figure 3

Open in a new tab

(A) Spatial map of firing rate for example mPFC single units.

(B) Normalized firing rate (z-score, top right) and a raster plot (bottom right) of spikes fired by an example left arm-selective single unit across trials aligned to goal arm entrance.

(C) Left: percentage of units that were goal-selective as a function of time from entering goal arm, according to repeated measures ANOVAs performed on binned spike rates. Percentages were calculated within animals, then averaged across animals. Right: mean percentage of units at time zero. p = 0.016 (WT vs. Df(16)A+/−), p = 0.031; (Df(16)A+/− vs. Df(16)A+/− + SB), Student’s t-test. Dashed line represents chance (p = 0.05).

(D) Firing rates by group. Firing rate did not differ by genotype (F1,1341 = 0.01, p =0.94) or treatment (F1,1341 = 3.29, p =0.070, Two-way ANOVA). n = 412 units from 11 animals (WT), 297 units from 12 animals (Df(16)A+/−), 405 units from 13 animals (Df(16)A+/− + SB) and 231 from 8 animals (WT + SB).

See also Supplementary Figure 4.

We also used an undirected approach to examine whether population-level spatial representations in the mPFC were affected by the microdeletion and/or subsequent developmental inhibition of Gsk3.. A maximum margin linear classifier (Rigotti et al., 2013; Spellman et al., 2015) was used to decode choice goal location from binned population firing rate vectors, and to quantify the accuracy of the neural representation of the animal’s contemporaneous location (Figure 4). Binned population firing rate vectors were aligned separately to multiple trial events (Figure 4A). The classifier was trained on firing rate data from half of the trials, and model accuracy was tested on data from the rest of the trials (see Methods). Figure 4B depicts decoding accuracy of the choice goal location using data from 42 mPFC units recorded from a single WT mouse. Choice goal identity was decoded at accuracies above chance from when the animal entered the T-intersection in the maze to when the animal left the goal arm, peaking at the goal location. Model accuracy was computed for each individual animal, and then averaged across animals by group (Figure 4). In WT mice, model accuracy was highest as animals approached the goal. Df(16)A+/− mice showed significantly lower model accuracy, demonstrating that the population spatial representation in mPFC neurons was disrupted (Figure 4D). Population decoding in the mutants was completely restored by developmental inhibition of Gsk3, which had no effect in WT mice (Figure 4D). A similar tendency was obtained from sample goal decoding, although this did not reach statistical significance (Figure S4B).

Figure 4. Disrupted population coding in Df(16)A+/− mice is reversed by developmental inhibition of Gsk3.

Figure 4

Open in a new tab

(A) Schematic of run sequence. Spike data from all isolated mPFC units was structured as peri-event spike histograms centered on key events during the task: Leaving start box (a), entering goal arm (b), reaching goal port (c), leaving goal arm (d), and returning to start box (e).

(B) Accuracy of choice goal decoding during choice run in an individual WT mouse. Solid lines denote mean decoding accuracy, and shaded areas cover 95% confidence intervals for real (red) and shuffled (gray) data.

(C) Decoding accuracy of choice goal as a function of time from entering goal arm, computed for individuals and averaged across animals by group. Solid lines denote mean decoding accuracy across animals, and shaded areas cover s.e.m. n = 10 (WT), 7 (Df(16)A+/−), 9 (Df(16)A+/− + SB) and 8 (WT + SB). Dashed line represents chance level.

(D) Mean decoding accuracy of goal representations at time zero. p = 0.0092 (WT vs. Df(16)A+/−), p = 0.029 (Df(16)A+/− vs. Df(16)A+/− + SB), Student’s t-test. Sample sizes as in (C).

See also Supplementary Figure 4.

We next set out to determine whether the effects of the Df(16)A+/− mutation and its rescue by Gsk3 antagonism on neural synchrony on the one hand, and spatial encoding on the other, were linked. We examined the relationship between spatial information and neural synchrony on an individual neuron-by-neuron basis. We used Mutual Information (MI, see Methods) to quantify spatial information in each neuron as a function of time in the maze, relative to the same landmarks used to create the linear classifier model described above. MI measures how much each of the neuron’s spikes reduces the uncertainty about the goal location (left vs right), and is measured in bits (Onken et al., 2014; Shannon, 1948). Like the accuracy of the population-based model, MI peaked at the choice goal location for most mPFC units (Figure 5A). Averaged across neurons, MI at the goal location was lower in Df(16)A+/− mice than in WTs, and was rescued by developmental Gsk3 antagonism (Figure 5B).

Figure 5. Spatial information during retrieval correlates with vHPC-mPFC synchrony during encoding.

Figure 5

Open in a new tab

(A) Raster plot across trials (bottom) and MI for goal location computed on binned spike rates (top) for an example unit, aligned to goal arm entry during the choice phase.

(B) Mean MI at goal arm entry for all neurons, by genotype and treatment. p = 0.0041 (WT vs. Df(16)A+/−) and p = 0.00023 (Df(16)A+/− vs. Df(16)A+/− + SB), Student’s t-test. n = 412 units from 11 animals (WT), 297 units from 12 animals (Df(16)A+/−) and 405 units from 13 animals (Df(16)A+/− + SB).

(C) Scatter plots depicting MI during the choice phase, plotted against the strength of phase-locking of mPFC units to vHPC gamma during the sample phase, for all neurons recorded from a representative WT (left) and Df(16)A+/− (right) mouse.

(D) Correlation between MI measured during the sample phase and phase-locking measured during the sample (left) and choice (right) phases, averaged across animals. (Differences from r=0 by t-test: WT, n = 10 mice, p = 0.029; Df(16)A+/−: n = 7, p = 0.65, Df(16)A+/− + SB: n = 9, p = 0.15 for sample; WT, p = 0.37; Df(16)A+/−, p = 0.81; Df(16)A+/− + SB, p = 0.27 for choice).

(E) Normalized MI as a function of phase-locking strength to vHPC gamma, binned by quartiles. (Multiple linear regression; p = 0.028 (WT), 0.32 (Df(16)A+/−) and 0.11 (Df(16)A+/− + SB).

(F) MI as a function of phase-locking strength, binned into quartiles and normalized by the mean of the 1st quartile. (paired t-test, p = 0.036 (WT), p = 0.84 (Df(16)A+/−), and p = 0.024 (Df(16)A+/− with SB).

See also Supplementary Figure 5.

Encoding of spatial information about the goal relies on gamma-frequency synchrony between the vHPC and mPFC during the sample phase (Spellman et al., 2015). We therefore asked whether MI about the goal during the choice phase varied as a function of phase-locking of the mPFC unit spiking to gamma-frequency oscillations in the vHPC during the sample phase. In WT mice, the strength of a given neuron’s phase-locking to vHPC gamma during the sample phase predicts the amount of information that neuron carries about goal location in the choice phase, both within individual animals (Figure 5C) and across the sample (by multiple linear regression, Figure 5E, or by comparing MI in the weakest and strongest phase-locked units, Figure 5F). In other words, neurons that phase-lock best to gamma in the vHPC during encoding possess more information about the goal location during retrieval. This relationship was not present in Df(16)A+/− mice – the mean correlation coefficient in the mutants did not differ from zero (Figure 5D); the multiple linear regression was not significant (Figure 5E); and MI in weakly and strongly phase-locked cells did not significantly differ (Figure 5F). SB treatment partially restored the relationship between MI and phase-locking, particularly by increasing the amount of spatial information in strongly phase-locked units (Figures 5C–F). LFP power did not differ by genotype or treatment regardless of frequency, region or task phase (Figures S3G–L). Phase-locking of mPFC spikes to vHPC theta oscillations did not show any correlation with mutual information in any groups (Figure S5).

DISCUSSION

The cognitive deficits associated with schizophrenia remain an unmet treatment need. Here we demonstrate that spatial working memory deficits seen in Df(16)A+/− mice, a mouse model of the schizophrenia-associated 22q11.2 microdeletion, can be rescued through developmental antagonism of Gsk3. We recapitulate, refine and extend our earlier findings demonstrating deficits in hippocampal-prefrontal synchrony in these mice, showing that these deficits are accompanied by impaired spatial representations in the mPFC. Furthermore, we demonstrate that developmental treatment with a Gsk3 antagonist, previously shown to rescue neurite outgrowth deficits in Zdhhc8+/− mice, also rescues the physiological and behavioral impairments. These results have several important implications. First, they underscore the close relationship between hippocampal-prefrontal synchrony and spatial information processing in the mPFC. Second, they elucidate a putative causal chain of events leading from the 22q11 microdeletion to its consequent effects on cognition, crossing multiple levels of analysis in the process. Finally, they provide a potential avenue for novel therapies aimed at enhancing cognition in 22q11.2 deletion carriers and perhaps a subset of patients with schizophrenia.

Hippocampal-prefrontal synchrony and spatial representations in the mPFC

The notion that the HPC and mPFC must cooperate for successful spatial working memory performance arose first from disconnection studies, in which lesioning or silencing the HPC on one side and mPFC on the other disrupts the behavior (Floresco et al., 1997; Izaki et al., 2008; Wang and Cai, 2006). This notion received further support from the subsequent observations that mPFC units synchronize with hippocampal oscillations (Siapas et al., 2005), and that this synchrony is modulated during spatial working memory tasks (Jones and Wilson, 2005; Sigurdsson et al., 2010; Spellman et al., 2015). This synchrony occurs in two different frequency bands, with different task-dependent modulation and different anatomical requirements. In the DNMTP task used here, theta-frequency synchrony is highest during the choice phase (Sigurdsson et al., 2010), while gamma-frequency synchrony is highest during the sample phase (Spellman et al., 2015). The latter finding suggested a role for gamma-frequency synchrony in the encoding of spatial information within the mPFC. Optogenetic inhibition of the direct vHPC-to-mPFC pathway disrupted both gamma-frequency synchrony and the encoding of goal location information within mPFC neurons during the sample phase of the task (Spellman et al., 2015).

Here we provide complementary evidence for a causal relationship between gamma-frequency synchrony and the encoding of spatial information in mPFC neurons, while demonstrating deficits in both in a model of schizophrenia predisposition. In WT mice, mPFC neurons with the strongest phase-locking to vHPC gamma during the sample phase also had the greatest information about goal location during the choice phase, establishing a clear relationship between gamma synchrony during encoding and spatial information during retrieval. This relationship was not present in Df(16)A+/− mice. Coupled with the finding of decreased gamma synchrony in these mice, both here and in Sigurdsson et al. (2010), these results suggest that in the absence of strong gamma synchrony between the vHPC and mPFC, Df(16)A+/− mice cannot successfully encode information about the goal location in the mPFC. Developmental Gsk3 inhibition partially restored the relationship between gamma synchrony and spatial information, despite only a modest (and not statistically significant) rescue of gamma synchrony strength itself. These findings suggest that successful behavioral rescue may be due to successfully reconnecting a subset of mPFC neurons to their vHPC inputs.

From gene to behavior

An important goal of using genetic models of neuropsychiatric disease is to define the chain of events leading from the gene to the behavioral disturbance. The results presented here, coupled with previous studies, allow the construction of preliminary, putative version of such a chain of events leading from the 22q11 microdeletion to spatial working memory deficits. This chain starts at the level of the gene: haploinsufficiency for Zdhhc8, one of the genes within the microdeletion in both humans and the mouse model, recapitulates the working memory phenotype (Mukai et al., 2015). Zdhhc8 codes for a palmitoyl-transferase, and mice with only one copy of the gene have reduced expression of the enzyme and altered palmitoylation of multiple neuronal proteins important for axonal development. This reduced palmitoylation results in mislocalization of several key axonal proteins, including phosphorylated Akt, a negative regulator of Gsk3β (Mukai et al., 2015). Accordingly, Zdhhc8+/− mice, as well as mice with the full microdeletion, have deficits in neurite outgrowth and axonal branching, including decreased branching of vHPC axons within the mPFC, as well as deficits in theta- and gamma-frequency synchrony between the vHPC and mPFC (Mukai et al., 2015). Deficits in gamma-frequency synchrony and spatial working memory behavior are recapitulated by inhibiting these vHPC-to-mPFC axon terminals, a manipulation that also disrupts task-related spatial representations within the mPFC (Spellman et al., 2015). Here we demonstrate deficits in spatial representations in mPFC neurons in Df(16)A+/− mice, tying them together with deficits in gamma-frequency synchrony. Together, these data support a model in which the microdeletion results in inadequate expression of the Zdhhc8 protein, reduced palmitoylation, mislocalization of axonal proteins, dysregulation and hyperactivity of Gsk3β, decreased branching of vHPC axons in the mPFC, decreased vHPC-mPFC synchrony, impaired spatial processing in the mPFC, and disrupted spatial working memory. Supporting this model is the observation, described in Mukai et al (2015) and here, that inhibition of Gsk3 during development restores normal axonal branching, vHPC-mPFC synchrony, mPFC spatial representations, and spatial working memory behavior.

This model is most likely incomplete in its details. For certain, it cannot explain the entirety of the microdeletion picture, including phenotypes such as pre-pulse inhibition that are likely quite relevant to schizophrenia yet not fully recapitulated in Zdhhc8+/− mice. In addition, other genetic pathways disrupted by the microdeletion have been shown to affect spatial working memory (Paterlini et al., 2005; Fenelon et al., 2011; Stark et al., 2008), although whether they impair vHPC-mPFC synchrony, mPFC spatial representations or both remain unknown. Finally, given that SB-216763 inhibits both α- and β-isoforms of Gsk3, we cannot exclude the possibility that Gsk3 inhibition affects aspects of the developing neural circuitry other than axonal branching. For example, it has been shown that signaling through dopamine D2 receptors can also modulate the activity of the Akt/Gsk3 pathway (Emamian et al., 2004; Beaulieu et al., 2004) through a complex including beta-arrestin and PP2A (Beaulieu et al., 2005; Beaulieu et al., 2007). Because components of the dopamine signaling pathway are expressed during mammalian cortical development (Araki et al., 2007), it is conceivable that Gsk3 antagonism may exert its effect also in part via noncanonical modes of dopamine D2/β-arrestin signaling (Allen et al., 2011), although it is currently difficult to predict the direction of such an effect on working memory and vHPC-mPFC synchrony deficits emerging as a result of 22q11.2 deletions. Nonetheless, the model is useful, describing our understanding of this one facet of the 22q11 microdeletion syndrome, crossing boundaries between multiple levels of analysis. It can be used as a framework for future studies, and importantly points to several possible avenues for therapeutic interventions.

Potential for novel treatments

The elucidation of this pathway, and the demonstration that Gsk3 inhibition can reverse these neuroanatomical, neurophysiological and behavioral consequences of the microdeleltion, are important advances. In general, these findings demonstrate the possibility of rescuing the cognitive sequellae of a schizophrenia predisposition genetic lesion. More specifically, they point to manipulation of the Akt/Gsk3β pathway as a potential treatment target (An et al., 2010). Here, pharmacological inhibition was present during an early postnatal time window in the mice, likely corresponding to a late prenatal period in humans (Avishai-Eliner et al., 2002). Given that the cognitive sequellae of the 22q11 microdeletion emerge early in life, it is conceivable that Gsk3β could be pursued as a promising treatment target in this subgroup even if it needed to be administered early to exert its effects. Because prodromal diagnoses are currently unreliable (Debbane et al., 2015) pursuing such early developmental treatments in idiopathic schizophrenia is currently challenging. However, we envision that progress in deciphering the genetic causes of the disease (Xu et al., 2012) and identifying highly penetrant genetic risk lesions (especially CNVs) is likely to facilitate prodromal diagnoses in subsets of patients in the future, enabling implementation of early developmental treatments such as the one proposed here.

Nevertheless, going forward, it will be important to also determine whether Gsk3 antagonism later in life would be similarly efficacious, as the diagnosis of schizophrenia is rarely made before late adolescence. To the extent that successful restoration of cognitive deficits and network activity biomarkers reflects structural restoration of neural circuits, it should be noted that although structural dynamic changes of axonal arbors in the adult cortex are dramatically smaller and slower compared to early development, there is still prominent regional and cell-type specific plasticity of excitatory axonal arbors, which takes place over periods of weeks or months (De Paola et al., 2006). In that respect, although acute Gsk3 antagonism is unlikely to be efficacious, chronic or subchronic manipulation of the Akt/Gsk3β pathway may still be able to promote growth of axonal endings and synaptic communication even in the adult brain. This possibility will be the focus of further investigation.

Of course the generalizability of these findings is unclear. The 22q11.2 microdeletion syndrome represents a specific subset of patients with schizophrenia, accounting for perhaps 1–2% of sporadic cases (Karayiorgou et al., 2010; Xu et al., 2008). Even if these findings led to novel treatments for the cognitive impairments of the 22q11.2 microdeletion syndrome, whether they would apply to schizophrenia cases arising from other causes is unclear given the level of genetic heterogeneity (Rodriguez-Murillo et al., 2012). Supporting the possibility of such generalizability is a body of data suggesting alterations in the Akt/Gsk3β signaling pathway in schizophrenia (Emamian, 2012; Emamian et al., 2004). Moreover, hippocampal-prefrontal dysconnectivity has been described in patients (Lawrie et al., 2002; Meyer-Lindenberg et al., 2005) as well as additional animal models (Esmaeili and Grace, 2013; Jodo, 2013; Phillips et al., 2012; Zhan et al., 2014). This convergence suggests the possibility that other types of treatments aimed at counteracting the effects of diverse disease mutations on the structure and strength of neural connections may be of therapeutic benefit.

Conclusion

In the very least, however, the current findings point out an intriguing, targetable biological pathway leading from a bonafide schizophrenia predisposition locus through multiple levels of analysis to an important behavioral phenotype, making good on the promise of genetic modeling of disease predisposition from a scientific perspective. Coupled with emerging ideas regarding how other genes in the 22q11 region could lead to psychosis (Chun et al., 2014; Fenelon et al., 2011; Fenelon et al., 2013; Paterlini et al., 2005; Xu et al., 2013), they represent a step towards a comprehensive understanding of the neurobiological consequences of the deletion syndrome. Such an understanding may eventually translate into bonafide improvements in the clinical management of patients who suffer from these consequences.

EXPERIMENTAL PROCEDURES

Animal and drug treatment

Df(16)A+/− mice (RRID:MGI_3802827) and their WT littermates were generated on a C57BL/6J background and genotyped as previously described (Stark et al., 2008). SB216763 was obtained from Sigma-Aldrich (St. Louis, MO). The drug was dissolved in dimethyl sulfoxide (20 mg/ml) as the stock solution. The stock solution was freshly diluted to 2 mg/ml with PEG400 at the days of injection. Final concentration for each mouse is 2 mg/kg. Mice were treated with either control dimethyl sulfoxide (WT; n = 11, Df(16)A+/−; n = 12) or SB216763 (2 mg/kg i.p., WT; n = 8, Df(16)A+/−; n = 13) every other day from P7 to P28 at 9 a.m. All procedures were conducted in accordance with NIH regulations and approved by Columbia University and New York State Psychiatric Institute Institutional Animal Care and Use Committees.

Surgery and recording

Three- to 5-month-old mice were anesthetized with isoflurane (Butler Schein, Chicago, IL) and placed in a stereotaxic head holder. A bundle of 14 twisted-wire stereotrodes (12.5 μm) were implanted in the mPFC (1.65 mm anterior to bregma, 0.3 mm lateral to midline, 1.4 mm below brain surface) and single tungsten wire field electrodes (75 μm) were implanted in the dHPC (1.94 mm posterior, 1.5 mm lateral, 1.4 mm ventral) and the vHPC (3.16 mm posterior, 3.0 mm lateral, 4.0 mm ventral). Skull screws were attached above the right olfactory bulb and left cerebellum served as reference and ground, respectively. All wires were connected to a 36-channel interface board anchored to a microdrive. mPFC stereotrodes were regularly advanced to ensure that different cells were recorded in each session. After all behavioral experiment were completed, to verify electrode placements, current (50 mA, 10s) was passed through the electrodes. Recoding locations were verified with Nissl cresyl violet staining.

Behavioral task

Animals were trained and tested in a spatial delayed non-match to sample T-maze task, as described previously (Sigurdsson et al., 2010). The maze consisted of a 55 cm-long center arm and two 32 cm-long goal arms extended from the center arm. Arms were 15 cm high and 10 cm wide. Each trial of the task consisted of a sample and choice phase. In the sample phase, a mouse ran down the center arm of the maze and was directed into one of the goal arms. The mouse returned to the start box where it remained for a delay of 10 seconds. In the choice phase, the mouse was required to enter the arm opposite to that visited during the sample phase to receive a reward. The behavior protocol began with 2 days of habituation to the maze for 10 minutes, followed by 2 days of shaping during which animals were guided by the presence of walls to alternate between goal arms of the maze to receive food rewards. Training sessions of 10 trials per day were then conducted until mice reached criterion performance, defined as performance of at least 70% correct per day for 3 consecutive days, they completed daily sessions composed of 20 trials.

Data Analysis

Data were imported into Matlab (MathWorks, Natick, MA) for analysis using custom written scripts.

Field analysis

Field potentials form vHPC, dHPC and mPFC were obtained while mice were habituated to a T-maze apparatus and they conducted delayed non-match to sample T-maze task. The power and coherence of field potential was computed with the “pwelch” and “mscohere” function in Matlab, respectively. Theta and gamma power and coherence were computed as the mean values in the 4–12-Hz and 30–70-Hz range, respectively. For baseline power and coherence during habituation, only periods of continuous movement (speed ≥ 10 cm/s) were used.

Phase-locking analysis

For single units, neural signals were band-pass-filtered between 600 and 6,000 Hz and waveforms that passed a threshold were digitized at 32,000 Hz. Waveforms were then sorted into single-unit clusters using KlustaKwik, followed by manual adjustment of clusters. LFP signal was digitally band-pass filtered (4–12Hz for theta, 30–70Hz for gamma) using a zero-phase delay filter. The phase of each sample of the filtered field potentials was then calculated by a Hilbert transform, and each spike was assigned to the corresponding phase. Phase-locking of mPFC units to the oscillatory phase of HPC LFPs was performed using PPC, which is known to be unbiased by spike number (Vinck et al., 2010). PPC is based on the average pairwise circular distance, which is defined by

D=2N(N-1)∑j=1N-1∑k=(j+1)Nd(θj,θk),

where d is the absolute angular distance between two samples, θj and θk are the phases of LFP samples assigned to contemporaneous spikes, and N is the number of spikes. PPC is computed by normalizing D as following,

PPC=π-2Dπ.

Notably, uniformly distributed spikes typically yield negative PPC values.

As another measure of phase-locking, MRL of the phase angles was computed as previously described (Sigurdsson et al., 2010). The MRL is the summation of the unit vectors of the phases at which each spike occurred, divided by the number of spikes. MRL is highly dependent on sample size. We therefore calculated MRL of each unit from subsamples of 50 spikes. Subsamples were drawn repeatedly (1000 times) and MRL values were computed for each subsamples and then averaged to estimate MRL values.

Mutual information

To quantify the extent of which the activity of mPFC units represents the animal’s contemporaneous location, the amount of mutual information was estimated between goal arm locations (left vs right). Mutual information (MI) quantifies the reduction in uncertainty of stimuli based on the bias of neuronal response in each stimulus. We used the standard formula which was introduced by Shannon (Shannon, 1948);

MI(S;R)=∑s,rP(r,s)log2P(r,s)P(r)P(s)

where P(r,s) is the joint probability of stimulus s (left vs right goal arm) and neuronal response r and P(r) is the probability of neuronal response across all stimuli and P(s) is the probability of each stimulus. In this study, neuronal response r is estimated as spike counts in 100-msec time bins, i.e. the probability of observing r spikes in the bin. The resultant mutual information was averaged in 5 consecutive bins as a mean MI in 500 msec.

Decoding goal arm location

Goal arm location was decoded using a maximum margin linear classifier as previously described (Rigotti et al., 2013). Population decoding was performed on binned spike vectors (500ms bins, 100ms increments) of mPFC units within animals and averaged across animals by experimental group. Model training was performed using constrained quadratic programming (Anlauf and Biehl, 1989; Barak and Rigotti, 2011) that employed a maximal margin linear classifier (Anlauf and Biehl, 1989; Krauth and Mezard, 1987). The training samples to this algorithm are generated by averaging the recorded spike counts within the relevant conditions that need to be decoded, and across a specified training set of trials. This procedure gives a mean activity vector per condition, in which each component of the vector represents the trial-averaged firing activity of a given neuron at a given condition. The quadratic programing procedure then aims at finding a set of readout weights that maximally separate the mean activity vectors corresponding to the conditions that have to be discriminated. The model was then tested by cross-validating its performance on a test set of recorded trials. At each test phase 100 test vectors per condition are randomly resampled with replacement, and the performance of the model is quantified as the average accuracy in classifying them. For each time bin, model training and testing was performed 100 times (at which point estimates of model accuracy approached asymptote) on non-overlapping subsets of trials (half of trials to train, half to test, random subsampling without replacement), with subsets constrained to include at least one trial corresponding with each feature class under consideration. Since model accuracy is highly dependent on sample size, we only performed decoding for animals from which more than 10 units were recorded. Moreover, to further limit the effects of this bias, we subsampled 10 units so that equal number of units were used for decoding when performing model training and testing.

Statistical analysis

Data are represented as means +/− s.e.m.. Student’s t-test after ANOVA was used for parametric statistics, whereas paired t-test was used for paired comparisons. To test the significance of correlations with multiple data points per animal, multiple linear regression analysis was conducted. To calculate the percentage of goal-arm modulated neurons, repeated measured ANOVA was performed on binned spike rates for each neuron.

Supplementary Material

supplement

Highlights.

  • 22q11.2 microdeletion causes deficits in cognition and HPC-mPFC synchrony.

  • Developmental Gsk3 inhibition rescued deficits in spatial working memory.

  • Gsk3 inhibition reverses deficits in HPC-mPFC synchrony and mPFC representations.

Acknowledgments

We would like to thank Naoko Haremaki for maintaining the mouse colony. We thank Timothy Spellman for sharing his analytic techniques and help in applying them, and Mattia Rigotti and Stefano Fusi for assistance in implementing the linear classifier. This work was supported by NIMH R01 MH096274.

Footnotes

DISCLOSURES

M.T. is an employment of Mitsubishi Tanabe Pharma Corporation.

AUTHOR CONTRIBUTIONS

Conceptualization, and Methodology, M.T., J.M., J.A. Gordon and J.A. Gogos; Investigation, M.T. Formal Analysis, M.T. and J.A. Gordon; Writing – Original Draft, M.T. and J.A. Gordon; Writing – Review & Editing, M.T., J. A. Gordon and J.A. Gogos.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Allen JA, Yost JM, Setola V, Chen X, Sassano MF, Chen M, Peterson S, Yadav PN, Huang XP, Feng B, Jensen NH, Che X, Bai X, Frye SV, Wetsel WC, Caron MG, Javitch JA, Roth BL, Jin J. Discovery of β-arrestin-biased dopamine D2 ligands for probing signal transduction pathways essential for antipsychotic efficacy. Proc Natl Acad Sci U S A. 2011;8:18488–93. doi: 10.1073/pnas.1104807108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. An WF, Germain AR, Bishop JA, Nag PP, Metkar S, Ketterman J, Walk M, Weiwer M, Liu X, Patnaik D, Zhang YL, Gale J, Zhao W, Kaya T, Barker D, Wagner FF, Holson EB, Dandapani S, Perez J, Munoz B, Palmer M, Pan JQ, Haggarty SJ, Schreiber SL. Discovery of Potent and Highly Selective Inhibitors of GSK3β. 2012 Apr 16 [updated 2014 May 13] Probe Reports from the NIH Molecular Libraries Program [Internet] 2014 [Google Scholar]
  3. Anlauf JK, Biehl M. The Adatron - an Adaptive Perceptron Algorithm. Europhys Lett. 1989;10:687–692. [Google Scholar]
  4. Araki KY, Sims JR, Bhide PG. Dopamine receptor mRNA and protein expression in the mouse corpus striatum and cerebral cortex during pre- and postnatal development. Brain Res. 2007;1156:31–45. doi: 10.1016/j.brainres.2007.04.043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Arguello PA, Gogos JA. Cognition in mouse models of schizophrenia susceptibility genes. Schizophrenia bulletin. 2010;36:289–300. doi: 10.1093/schbul/sbp153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Avishai-Eliner S, Brunson KL, Sandman CA, Baram TZ. Stressed-out, or in (utero)? Trends in neurosciences. 2002;25:518–524. doi: 10.1016/s0166-2236(02)02241-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Baeg EH, Kim YB, Huh K, Mook-Jung I, Kim HT, Jung MW. Dynamics of population code for working memory in the prefrontal cortex. Neuron. 2003;40:177–188. doi: 10.1016/s0896-6273(03)00597-x. [DOI] [PubMed] [Google Scholar]
  8. Barak O, Rigotti M. A Simple Derivation of a Bound on the Perceptron Margin Using Singular Value Decomposition. Neural Comput. 2011;23:1935–1943. [Google Scholar]
  9. Barch DM, Ceaser A. Cognition in schizophrenia: core psychological and neural mechanisms. Trends Cogn Sci. 2012;16:27–34. doi: 10.1016/j.tics.2011.11.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Bender HU, Almashanu S, Steel G, Hu CA, Lin WW, Willis A, Pulver A, Valle D. Functional consequences of PRODH missense mutations. American journal of human genetics. 2005;76:409–420. doi: 10.1086/428142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Beaulieu JM, Sotnikova TD, Yao WD, Kockeritz L, Woodgett JR, Gainetdinov RR, Caron MG. Lithium antagonizes dopamine-dependent behaviors mediated by an AKT/glycogen synthase kinase 3 signaling cascade. Proc Natl Acad Sci U S A. 2004;101:5099–5104. doi: 10.1073/pnas.0307921101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Beaulieu JM, Sotnikova TD, Marion S, Lefkowitz RJ, Gainetdinov RR, Caron MG. An Akt/beta-arrestin 2/PP2A signaling complex mediates dopaminergic neurotransmission and behavior. Cell. 2005;122:261–273. doi: 10.1016/j.cell.2005.05.012. [DOI] [PubMed] [Google Scholar]
  13. Beaulieu JM, Gainetdinov RR, Caron MG. The Akt-GSK-3 signaling cascade in the actions of dopamine. Trends Pharmacol Sci. 2007;28:166–172. doi: 10.1016/j.tips.2007.02.006. [DOI] [PubMed] [Google Scholar]
  14. Buoli M, Altamura AC. May non-antipsychotic drugs improve cognition of schizophrenia patients? Pharmacopsychiatry. 2015;48:41–50. doi: 10.1055/s-0034-1396801. [DOI] [PubMed] [Google Scholar]
  15. Chun S, Westmoreland JJ, Bayazitov IT, Eddins D, Pani AK, Smeyne RJ, Yu J, Blundon JA, Zakharenko SS. Specific disruption of thalamic inputs to the auditory cortex in schizophrenia models. Science. 2014;344:1178–1182. doi: 10.1126/science.1253895. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Debbane M, Eliez S, Badoud D, Conus P, Fluckiger R, Schultze-Lutter F. Developing psychosis and its risk states through the lens of schizotypy. Schizophrenia bulletin. 2015;41(Suppl 2):S396–407. doi: 10.1093/schbul/sbu176. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Emamian ES. AKT/GSK3 signaling pathway and schizophrenia. Frontiers in molecular neuroscience. 2012;5:33. doi: 10.3389/fnmol.2012.00033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. De Paola V, Holtmaat A, Knott G, Song S, Wilbrecht L, Caroni P, Svoboda K. Cell type-specific structural plasticity of axonal branches and boutons in the adult neocortex. Neuron. 2006;6:861–75. doi: 10.1016/j.neuron.2006.02.017. [DOI] [PubMed] [Google Scholar]
  19. Emamian ES, Hall D, Birnbaum MJ, Karayiorgou M, Gogos JA. Convergent evidence for impaired AKT1-GSK3beta signaling in schizophrenia. Nature genetics. 2004;36:131–137. doi: 10.1038/ng1296. [DOI] [PubMed] [Google Scholar]
  20. Esmaeili B, Grace AA. Afferent drive of medial prefrontal cortex by hippocampus and amygdala is altered in MAM-treated rats: evidence for interneuron dysfunction. Neuropsychopharmacology: official publication of the American College of Neuropsychopharmacology. 2013;38:1871–1880. doi: 10.1038/npp.2013.64. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Fenelon K, Mukai J, Xu B, Hsu PK, Drew LJ, Karayiorgou M, Fischbach GD, Macdermott AB, Gogos JA. Deficiency of Dgcr8, a gene disrupted by the 22q11.2 microdeletion, results in altered short-term plasticity in the prefrontal cortex. Proceedings of the National Academy of Sciences of the United States of America. 2011;108:4447–4452. doi: 10.1073/pnas.1101219108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Fenelon K, Xu B, Lai CS, Mukai J, Markx S, Stark KL, Hsu PK, Gan WB, Fischbach GD, MacDermott AB, et al. The pattern of cortical dysfunction in a mouse model of a schizophrenia-related microdeletion. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2013;33:14825–14839. doi: 10.1523/JNEUROSCI.1611-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Fett AKJ, Viechtbauer W, Dominguez MD, Penn DL, van Os J, Krabbendam L. The relationship between neurocognition and social cognition with functional outcomes in schizophrenia: A meta-analysis. Neurosci Biobehav R. 2011;35:573–588. doi: 10.1016/j.neubiorev.2010.07.001. [DOI] [PubMed] [Google Scholar]
  24. Floresco SB, Seamans JK, Phillips AG. Selective roles for hippocampal, prefrontal cortical, and ventral striatal circuits in radial-arm maze tasks with or without a delay. The Journal of neuroscience: the official journal of the Society for Neuroscience. 1997;17:1880–1890. doi: 10.1523/JNEUROSCI.17-05-01880.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Fujisawa S, Amarasingham A, Harrison MT, Buzsaki G. Behavior-dependent short-term assembly dynamics in the medial prefrontal cortex. Nat Neurosci. 2008;11:823–833. doi: 10.1038/nn.2134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Gothelf D, Feinstein C, Thompson T, Gu E, Penniman L, Van Stone E, Kwon H, Eliez S, Reiss AL. Risk factors for the emergence of psychotic disorders in adolescents with 22q11.2 deletion syndrome. The American journal of psychiatry. 2007;164:663–669. doi: 10.1176/ajp.2007.164.4.663. [DOI] [PubMed] [Google Scholar]
  27. Green T, Gothelf D, Glaser B, Debbane M, Frisch A, Kotler M, Weizman A, Eliez S. Psychiatric disorders and intellectual functioning throughout development in velocardiofacial (22q11.2 deletion) syndrome. Journal of the American Academy of Child and Adolescent Psychiatry. 2009;48:1060–1068. doi: 10.1097/CHI.0b013e3181b76683. [DOI] [PubMed] [Google Scholar]
  28. Harper KM, Hiramoto T, Tanigaki K, Kang G, Suzuki G, Trimble W, Hiroi N. Alterations of social interaction through genetic and environmental manipulation of the 22q11.2 gene Sept5 in the mouse brain. Human molecular genetics. 2012;21:3489–3499. doi: 10.1093/hmg/dds180. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Hiramoto T, Kang G, Suzuki G, Satoh Y, Kucherlapati R, Watanabe Y, Hiroi N. Tbx1: identification of a 22q11.2 gene as a risk factor for autism spectrum disorder in a mouse model. Human molecular genetics. 2011;20:4775–4785. doi: 10.1093/hmg/ddr404. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Hofer A, Baumgartner S, Bodner T, Edlinger M, Hummer M, Kemmler G, Rettenbacher MA, Fleischhacker WW. Patient outcomes in schizophrenia II: the impact of cognition. Eur Psychiat. 2005;20:395–402. doi: 10.1016/j.eurpsy.2005.02.006. [DOI] [PubMed] [Google Scholar]
  31. Hsu R, Woodroffe A, Lai WS, Cook MN, Mukai J, Dunning JP, Swanson DJ, Roos JL, Abecasis GR, Karayiorgou M, Gogos JA. Nogo Receptor 1 (RTN4R) as a candidate gene for schizophrenia: analysis using human and mouse genetic approaches. PloS one. 2007;2:e1234. doi: 10.1371/journal.pone.0001234. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Huotari M, Garcia-Horsman JA, Karayiorgou M, Gogos JA, Mannisto PT. Damphetamine responses in catechol-O-methyltransferase (COMT) disrupted mice. Psychopharmacology. 2004;172:1–10. doi: 10.1007/s00213-003-1627-3. [DOI] [PubMed] [Google Scholar]
  33. Izaki Y, Takita M, Akema T. Specific role of the posterior dorsal hippocampus-prefrontal cortex in short-term working memory. The European journal of neuroscience. 2008;27:3029–3034. doi: 10.1111/j.1460-9568.2008.06284.x. [DOI] [PubMed] [Google Scholar]
  34. Jodo E. The role of the hippocampo-prefrontal cortex system in phencyclidine-induced psychosis: a model for schizophrenia. Journal of physiology, Paris. 2013;107:434–440. doi: 10.1016/j.jphysparis.2013.06.002. [DOI] [PubMed] [Google Scholar]
  35. Jones MW, Wilson MA. Theta rhythms coordinate hippocampal-prefrontal interactions in a spatial memory task. PLoS biology. 2005;3:e402. doi: 10.1371/journal.pbio.0030402. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Jung MW, Qin Y, McNaughton BL, Barnes CA. Firing characteristics of deep layer neurons in prefrontal cortex in rats performing spatial working memory tasks. Cereb Cortex. 1998;8:437–450. doi: 10.1093/cercor/8.5.437. [DOI] [PubMed] [Google Scholar]
  37. Karayiorgou M, Morris MA, Morrow B, Shprintzen RJ, Goldberg R, Borrow J, Gos A, Nestadt G, Wolyniec PS, Lasseter VK, et al. Schizophrenia susceptibility associated with interstitial deletions of chromosome 22q11. Proceedings of the National Academy of Sciences of the United States of America. 1995;92:7612–7616. doi: 10.1073/pnas.92.17.7612. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Karayiorgou M, Simon TJ, Gogos JA. 22q11.2 microdeletions: linking DNA structural variation to brain dysfunction and schizophrenia. Nature reviews Neuroscience. 2010;11:402–416. doi: 10.1038/nrn2841. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Kramer T, Schmidt B, Lo Monte F. Small-Molecule Inhibitors of GSK-3: Structural Insights and Their Application to Alzheimer’s Disease Models. International journal of Alzheimer’s disease. 2012;2012:381029. doi: 10.1155/2012/381029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Krauth W, Mezard M. Learning Algorithms with Optimal Stability in Neural Networks. J Phys a-Math Gen. 1987;20:L745–L752. [Google Scholar]
  41. Lawrie SM, Buechel C, Whalley HC, Frith CD, Friston KJ, Johnstone EC. Reduced frontotemporal functional connectivity in schizophrenia associated with auditory hallucinations. Biol Psychiatry. 2002;51:1008–1011. doi: 10.1016/s0006-3223(02)01316-1. [DOI] [PubMed] [Google Scholar]
  42. Lee I, Kesner RP. Time-dependent relationship between the dorsal hippocampus and the prefrontal cortex in spatial memory. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2003;23:1517–1523. doi: 10.1523/JNEUROSCI.23-04-01517.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Lett TA, Voineskos AN, Kennedy JL, Levine B, Daskalakis ZJ. Treating Working Memory Deficits in Schizophrenia: A Review of the Neurobiology. Biol Psychiat. 2014;75:361–370. doi: 10.1016/j.biopsych.2013.07.026. [DOI] [PubMed] [Google Scholar]
  44. Meyer-Lindenberg AS, Olsen RK, Kohn PD, Brown T, Egan MF, Weinberger DR, Berman KF. Regionally specific disturbance of dorsolateral prefrontal-hippocampal functional connectivity in schizophrenia. Archives of general psychiatry. 2005;62:379–386. doi: 10.1001/archpsyc.62.4.379. [DOI] [PubMed] [Google Scholar]
  45. Millan MJ, Agid Y, Brune M, Bullmore ET, Carter CS, Clayton NS, Connor R, Davis S, Deakin B, DeRubeis RJ, et al. Cognitive dysfunction in psychiatric disorders: characteristics, causes and the quest for improved therapy. Nature reviews Drug discovery. 2012;11:141–168. doi: 10.1038/nrd3628. [DOI] [PubMed] [Google Scholar]
  46. Mukai J, Liu H, Burt RA, Swor DE, Lai WS, Karayiorgou M, Gogos JA. Evidence that the gene encoding ZDHHC8 contributes to the risk of schizophrenia. Nature genetics. 2004;36:725–731. doi: 10.1038/ng1375. [DOI] [PubMed] [Google Scholar]
  47. Mukai J, Tamura M, Fenelon K, Rosen AM, Spellman TJ, Kang R, MacDermott AB, Karayiorgou M, Gordon JA, Gogos JA. Molecular substrates of altered axonal growth and brain connectivity in a mouse model of schizophrenia. Neuron. 2015;86:680–695. doi: 10.1016/j.neuron.2015.04.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Murphy KC, Jones LA, Owen MJ. High rates of schizophrenia in adults with velo-cardio-facial syndrome. Archives of general psychiatry. 1999;56:940–945. doi: 10.1001/archpsyc.56.10.940. [DOI] [PubMed] [Google Scholar]
  49. O’Neill PK, Gordon JA, Sigurdsson T. Theta oscillations in the medial prefrontal cortex are modulated by spatial working memory and synchronize with the hippocampus through its ventral subregion. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2013;33:14211–14224. doi: 10.1523/JNEUROSCI.2378-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Onken A, Karunasekara PPCR, Kayser C, Panzeri S. Understanding Neural Population Coding: Information Theoretic Insights from the Auditory System. Advances in Neuroscience. 2014;2014:1–14. [Google Scholar]
  51. Paterlini M, Zakharenko SS, Lai WS, Qin J, Zhang H, Mukai J, Westphal KG, Olivier B, Sulzer D, Pavlidis P, et al. Transcriptional and behavioral interaction between 22q11.2 orthologs modulates schizophrenia-related phenotypes in mice. Nat Neurosci. 2005;8:1586–1594. doi: 10.1038/nn1562. [DOI] [PubMed] [Google Scholar]
  52. Phillips KG, Bartsch U, McCarthy AP, Edgar DM, Tricklebank MD, Wafford KA, Jones MW. Decoupling of sleep-dependent cortical and hippocampal interactions in a neurodevelopmental model of schizophrenia. Neuron. 2012;76:526–533. doi: 10.1016/j.neuron.2012.09.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Rigotti M, Barak O, Warden MR, Wang XJ, Daw ND, Miller EK, Fusi S. The importance of mixed selectivity in complex cognitive tasks. Nature. 2013;497:585–590. doi: 10.1038/nature12160. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Rodriguez-Murillo L, Gogos JA, Karayiorgou M. The genetic architecture of schizophrenia: new mutations and emerging paradigms. Annual review of medicine. 2012;63:63–80. doi: 10.1146/annurev-med-072010-091100. [DOI] [PubMed] [Google Scholar]
  55. Shannon CE. A Mathematical Theory of Communication. At&T Tech J. 1948;27:379–423. [Google Scholar]
  56. Siapas AG, Lubenov EV, Wilson MA. Prefrontal phase locking to hippocampal theta oscillations. Neuron. 2005;46:141–151. doi: 10.1016/j.neuron.2005.02.028. [DOI] [PubMed] [Google Scholar]
  57. Sigurdsson T, Stark KL, Karayiorgou M, Gogos JA, Gordon JA. Impaired hippocampal-prefrontal synchrony in a genetic mouse model of schizophrenia. Nature. 2010;464:763–767. doi: 10.1038/nature08855. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Spellman T, Rigotti M, Ahmari SE, Fusi S, Gogos JA, Gordon JA. Hippocampal-prefrontal input supports spatial encoding in working memory. Nature. 2015;522:309–314. doi: 10.1038/nature14445. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Stark KL, Xu B, Bagchi A, Lai WS, Liu H, Hsu R, Wan X, Pavlidis P, Mills AA, Karayiorgou M, Gogos JA. Altered brain microRNA biogenesis contributes to phenotypic deficits in a 22q11-deletion mouse model. Nature genetics. 2008;40:751–760. doi: 10.1038/ng.138. [DOI] [PubMed] [Google Scholar]
  60. Vinck M, van Wingerden M, Womelsdorf T, Fries P, Pennartz CM. The pairwise phase consistency: a bias-free measure of rhythmic neuronal synchronization. Neuro Image. 2010;51:112–122. doi: 10.1016/j.neuroimage.2010.01.073. [DOI] [PubMed] [Google Scholar]
  61. Wang GW, Cai JX. Disconnection of the hippocampal-prefrontal cortical circuits impairs spatial working memory performance in rats. Behavioural brain research. 2006;175:329–336. doi: 10.1016/j.bbr.2006.09.002. [DOI] [PubMed] [Google Scholar]
  62. Xu B, Ionita-Laza I, Roos JL, Boone B, Woodrick S, Sun Y, Levy S, Gogos JA, Karayiorgou M. De novo gene mutations highlight patterns of genetic and neural complexity in schizophrenia. Nat Genet. 2012;44:1365–9. doi: 10.1038/ng.2446. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Xu B, Hsu PK, Stark KL, Karayiorgou M, Gogos JA. Derepression of a neuronal inhibitor due to miRNA dysregulation in a schizophrenia-related microdeletion. Cell. 2013;152:262–275. doi: 10.1016/j.cell.2012.11.052. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Xu B, Roos JL, Levy S, van Rensburg EJ, Gogos JA, Karayiorgou M. Strong association of de novo copy number mutations with sporadic schizophrenia. Nature genetics. 2008;40:880–885. doi: 10.1038/ng.162. [DOI] [PubMed] [Google Scholar]
  65. Zhan Y, Paolicelli RC, Sforazzini F, Weinhard L, Bolasco G, Pagani F, Vyssotski AL, Bifone A, Gozzi A, Ragozzino D, Gross CT. Deficient neuron-microglia signaling results in impaired functional brain connectivity and social behavior. Nat Neurosci. 2014;17:400–406. doi: 10.1038/nn.3641. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplement
NIHMS761735-supplement.pdf (239.2KB, pdf)

RESOURCES