PERK is a key regulator of translation control pathways known to be involved in learning and memory formation. Previous studies have shown that global inactivation of PERK causes severe developmental defects, precluding a comprehensive analysis of its role in cellular and molecular processes underlying memory and cognitive function. In our studies, we used the Cre-lox expression system to achieve a temporally and spatially restricted inactivation of Perk in the postnatal forebrain. Our findings reveal a previously unrecognized role of PERK-dependent translational regulation in cognitive function and provide molecular insights into the pathophysiology of cognitive impairment. We found that disruption of the PERK-eIF2α-ATF4 signaling pathway in the mouse prefrontal cortex recapitulates multiple behavioral phenotypes consistent with impaired cognition and information processing. Furthermore, our studies identify the modulation of eIF2α phosphorylation as a potential molecular target for therapeutic agents designed to prevent cognitive symptoms associated with a wide variety of neurological and neuropsychiatric disorders.
Earlier studies provide strong evidence to suggest that gene-specific translation of ATF4 is critical for the modulation of hippocampus-dependent long-term synaptic potentiation and memory formation. In particular, transgenic mice expressing a dominant negative inhibitor of C/EBP proteins were reported to have reduced ATF4 expression, which was associated with a facilitation of hippocampus-dependent longterm synaptic plasticity and memory formation (Chen et al., 2003
). Moreover, a reduction of ATF4 expression in mice lacking the eIF2α kinase GCN2 and heterozygous knockin mice with a mutation on serine 51 of eIF2α results in a lowered threshold for eliciting long-lasting LTP and memory (Costa-Mattioli et al., 2005
; Costa-Mattioli et al., 2007
). Taken together, these studies suggest that regulation of ATF4 expression, mediated by GCN2-dependent phosphorylation of eIF2α, is required for activity-dependent, enduring changes in neuronal function. Expanding on these findings, we show that in the prefrontal cortex of PERK-deficient mice, reduced eIF2α phosphorylation and ATF4 expression is associated with severe behavioral inflexibility. Thus, our data suggests that the regulation of ATF4 mRNA translation, mediated by PERK-directed phosphorylation of eIF2α, is critical for normal cognitive function. Because of its conserved role as a memory repressor in diverse phyla (Bartsch et al., 1995
; Chen et al., 2003
; Costa-Mattioli et al., 2005
; Yin et al., 1994
), we speculate that ATF4 normally acts to destabilize the initial memory trace and consequently, when ATF4 expression is reduced in the prefrontal cortex, the initial memory trace prevails despite changes in sensory and contextual information.
Our data suggests that reduction of PERK expression and ATF4 translation in the frontal cortex may specifically contribute to the pathophysiology of human schizophrenia (). Interestingly, ATF4 has previously been shown to interact with Disrupted-In-Schizophrenia 1 (DISC1) (Chubb et al., 2008
; Muir et al., 2008
; Sawamura et al., 2008
), a genetic risk factor for mental illnesses, including mood disorders and schizophrenia. In addition, mutations in selective regions of the DISC1 gene result in a loss of interaction with ATF4 (Morris et al., 2003
). Moreover, the ATF4 gene is positioned at chromosome 22q13, a hotspot for several schizophrenia-related susceptibility genes (Lewis et al., 2003
; Mowry et al., 2004
; Williams et al., 2003
). Finally, polymorphisms in the ATF4 locus have been associated with schizophrenia in male patients (Qu et al., 2008
). Thus, multiple studies support the notion that ATF4, whose translation is tightly regulated by eIF2α phosphorylation, is correlated with schizophrenia. It should be noted that in addition to the central regulator ATF4, other genes recently have been identified that are preferentially translated by eIF2α phosphorylation, which suggests that additional factors participate downstream of PERK to regulate cognitive function (Dey et al., 2010
; Jackson et al., 2010
Behavioral studies with the PERK mutant mice revealed multiple phenotypes consistent with cognitive and information processing deficits, which have been implicated as core features of numerous mental illnesses, including schizophrenia, bipolar disorder, attention deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) (Bora et al., 2009
; Goos et al., 2009
; Lesh et al., 2011
; Solomon et al., 2009
). In particular, PERK-deficient mice exhibited reduced prepulse inhibition (), a sensorimotor gating mechanism that restricts processing of sensory information (Bitsios et al., 2006
; Braff et al., 2001
). In addition, PERK cKO mice showed enhanced preference for the familiar object compared with the novel object in a hippocampus- and entorhinal cortex-dependent novel object recognition task (). One explanation for these results is that PERK is important for frontal and temporal cortex-dependent sensory information processing. Thus, in the absence of PERK, mice are unable to inhibit responses to sensory or cognitive information and have enhanced perseveration. Consistent with this notion, our results from the MWM and Y-maze tests indicate that PERK cKO mice have reduced inhibitory control of a previously-reinforced response, which causes enhanced perseveration, impaired reversal learning, and behavioral inflexibility ( and and Movies S2
). Furthermore, the results from our fear extinction studies highlight an equally important role for PERK in PFC-directed updating of behavior (). Collectively, these results suggest that PERK regulates sensory information processing, thereby inducing deficits in various cognitive paradigms when eliminated.
Previous studies have shown that a reduction of eIF2α phosphorylation in mice lacking the eIF2α kinase GCN2 and in heterozygous knockin mice with a mutation on serine 51 of eIF2α results in a lowered threshold for the consolidation of long-term memory (Costa-Mattioli et al., 2005
; Costa-Mattioli et al., 2007
). Similar to the PERK cKO mice, GCN2 and eIF2α-S51A mutant mice showed reduced eIF2α phosphorylation and ATF4 expression, although these reductions were global and constitutive rather than forebrain-specific and post-developmental. However, the behavioral phenotypes of the GCN2 and eIF2α-S51A mutant mice were quite different from the PERK cKO mice. Unlike the PERK cKO mice, we found that GCN2 KO mice showed normal reversal learning in the Y-water maze reversal task (Figure S5
). These complementary studies indicate that even in the face of similar biochemical profiles, such as reduced eIF2α phosphorylation and ATF4 expression, additional mechanisms and levels of regulation exist to further modulate pools of eIF2α that are eventually reflected by distinct behavioral phenotypes. It also is important to emphasize that the PERK cKO mice are post-developmental knockout mice where the disruption of PERK occurs approximately two to three weeks after birth. In contrast, GCN2 KO and eIF2α-S51A mutant mice, as well as PKR (another eIF2α kinase) KO mice, are all global, constitutive knockout mice. Thus, the behavioral phenotypes displayed by the GCN2, eIF2α-S51A, and PKR mutant mouse lines could be due to developmental complications, whereas the PERK cKO mice are not.
One current model for the pathophysiology of PCP-induced psychosis and schizophrenia involves the hypofunction of NMDA-R in GABAergic fast-spiking interneurons (Belforte et al., 2010
; Lisman et al., 2008
; Nakazawa et al., 2011
). Loss of NMDA-R function in interneurons is thought to result in disinhibition of pyramidal neurons in the cortex and hippocampus, asynchronous pyramidal neuron activation, hyperexcitability of cortical networks, and cognitive impairment. Intriguingly, our findings suggest that while selective ablation of PERK in pyramidal neurons does not alter NMDA-R function (), chronic NMDA-R hypofunction elicits a decrease of eIF2α phosphorylation in the prefrontal cortex (). Based on these results, we speculate that NMDA-R hypofunction in interneurons causes cortical excitation, dysregulation of PERK, and decreased eIF2α phosphorylation in pyramidal neurons, resulting in impaired cognition. Consistent with this notion, a recent study showed that deletion of the eIF2α kinase PKR in mice results in reduced GABAergic transmission, increased network excitability, and altered cognition (Zhu et al., 2011
). However, whether selective ablation of PERK in pyramidal neurons associates with altered cortical network excitability remains to be determined. Furthermore, it is possible that reduced NMDA-R function causes a disruption of PERK-directed translation specifically in GABAergic interneurons to dampen inhibitory control of pyramidal neurons and impair cognitive function. Future studies examining the role of PERK in various neuronal subtypes, in particular GABAergic interneurons, will provide molecular insight into the role of eIF2α in the pathophysiology of cognitive impairment associated with multiple neurological disorders.
Interestingly, it has been shown that enhancement of NMDA-R function by treatment with the GlyT1 inhibitor SSR504734 improves behavioral flexibility, reversal learning, and overall cognitive function (Black et al., 2009
; Depoortere et al., 2005
; Singer et al., 2009
). Consistent with these studies, our findings indicate that SSR504734 uniquely enhanced behavioral flexibility ( and Movies S5
) without altering enhanced vertical activity and sensorimotor gating impairments displayed by the PERK cKO mice. Furthermore, chronic SSR504734 treatment was found to restore aberrant eIF2α phosphorylation and ATF4 expression, but not disrupted PERK levels in the PFC of PERK cKO mice (). Thus, our results indicate that chronic inhibition of GlyT1 can normalize disrupted PERK-regulated translation, presumably by either activating other eIF2α kinases such as GCN2 and/or PKR, or by inhibiting the protein phosphatase 1/GADD34 complex that dephosphorylates eIF2α (Ma and Hendershot, 2003
). Future studies are required to address whether chronic SSR504734 treatment enhances GCN2 activity, enhances PKR activity, or inhibits protein phosphatase 1/GADD34 complex in the prefrontal cortex of PERK mutant mice.
Under normal physiological conditions, it has been reported that systemic administration of SSR504734 improves behavioral flexibility and cognitive function in wild-type mice (Singer et al., 2009
). In contrast, we found that although chronic SSR504734 treatment had no effect on the performance of the wild-type mice (), it restored the behavioral flexibility of the PERK cKO mice (). Moreover, we found that SSR504734 could not only rescue the behavioral deficits, but also could restore the molecular anomalies exhibited by the PERK cKO mice (). Thus, our findings provide direct evidence that SSR504734 can modulate eIF2α phosphorylation and ATF4 expression that is tightly correlated with reversal of the behavioral inflexibility displayed by the PERK cKO mice. Future studies are required to determine whether chronic SSR504734 treatment can restore the dysregulated eIF2α-ATF4 axis in wild-type mice treated chronically with agents such as MK-801 and PCP that induce NMDA-R hypofunction and are used to model schizophrenia.
A rapidly expanding list of neurological disorders and neurodegenerative diseases including Alzheimer’s disease, Parkinson’s disease, fragile X syndrome, tuberous sclerosis complex, and ASD have been linked to dysregulated protein synthesis (Auluck et al., 2010
; Hoeffer and Klann, 2010
; Kelleher and Bear, 2008
; Ozcan et al., 2008
; Palop and Mucke, 2010
; Santini and Klann, 2011
). Consistent with these findings, our results suggest that post-developmental disruption of PERK-regulated translational control is sufficient to trigger cognitive control impairments consistent with several disorders, including schizophrenia. In conclusion, these findings emphasize the critical importance of PERK in normal cognitive processes. Further studies elucidating the specific role of PERK-regulated translation in the brain may provide new avenues to tackle such widespread and often debilitating neurological disorders.