In this study, we address the potential contribution of deficits in GAD67-mediated GABA synthesis in mice to behavioral and morphological phenotypes reminiscent of schizophrenia pathology. Our data reveal a profound alteration in GAD67+/GFP mice in the socio-emotional domain, including an altered vulnerability to social deprivation and a selective catecholaminergic, presumably dopaminergic, hyperinnervation of area CA1 in the dorsal hippocampus.
Social behavior, locomotor activity, anxiety and sensorimotor gating in GAD67+/GFP mice
Negative symptoms like social withdrawal, avolition and depressive-like behavior are common patterns of human schizophrenia (Samsom and Wong
2015) and also apparent in animal models (Ellenbroek and Cools
2000; Matrisciano et al.
2013). Complex social behaviors and emotional processing such as anxiety are depending on different cerebral cortical areas, the hippocampus and the amygdala (Bicks et al.
2015). On the other hand, motivation and execution of motor plans are processed along parallel cortico-striatal-thalamic-cortical loops (Haber
2003). Alterations in those complex circuits could affect social behavior, locomotor activity, anxiety and other behavioral traits (Guillin et al.
2007). The social interaction test in the open field allows analyzing spontaneous induced behavior largely independent from external influences (File and Seth
2003). Our behavioral analysis showed that
GAD67+/GFP mice spent less time in social contact, marker of a profound disturbance in social interaction. Importantly, this alteration was independent from general motor activity, represented by a comparable distance moved of both genotypes in the open field. Our findings are in accordance with data from our previous study investigating sociability in
GAD67+/GFP mice using the three-chamber social preference test (Sandhu et al.
2014). Additionally,
GAD67+/GFP and
GAD67+/+ mice where tested in the tube test for social dominance. This test was originally developed to analyze social hierarchy in mice through the measurement of aggression (Lindzey et al.
1961), which is known to be modulated by GABAergic function (Miczek et al.
2003). Independent from grouped and isolated housing conditions,
GAD67+/GFP retreated more frequently than
GAD67+/+ mice suggesting a reduced social dominance. This finding is supported by data from our previous study, showing decreased aggressive like behavior of
GAD67+/GFP mice in the resident intruder test (Sandhu et al.
2014). However, we found no genotype differences in the number of aggressive contacts in the social interaction test, which may reflect different forms of social behavior and aggression in these two tasks (Nelson and Chiavegatto
2000).
Social behavior in rodents is highly complex and greatly depends on olfaction (Schultz and Tapp
1973). In our previous study, we found that male
GAD67+/GFP mice show a reduced preference for female mice and a reduced sensitivity to social and non-social odors, maybe reflecting an altered olfactory sense. As demonstrated by expression of the immediate early gene c-Fos we found an unaltered activation of the olfactory bulb in
GAD67+/GFP mice, but decreased neuronal activation in amygdala, bed nucleus of stria terminalis, medial preoptic area and lateral septum. It can be assumed, that
GAD67+/GFP mice show a disturbance in detection or, more importantly, processing of olfactory stimuli in downstream circuits relevant for social behavior (Sandhu et al.
2014).
GAD67+/GFP mice additionally showed altered non-social behavior, an increased rearing activity, in the social interaction test. This could be result of an impaired exploratory motor behavior and habituation of exploration in a novel environment, also reported for the DISC1 animal model of schizophrenia (Walsh et al.
2012). The social interaction test in the open field additionally allowed us to investigate the effects of post-weaning isolation rearing. Isolated male mice of both genotypes showed increased investigation of the unfamiliar test partner shown by higher numbers in sniffing, anogenital sniffing and aggressive contacts. This could be more result of the social isolation rearing (Shoji and Mizoguchi
2011) rather than of GAD67 haplodeficiency. Isolated mice also displayed an increased passive social behavior, which is seen when animals are sitting or lying close to each other, but without direct interaction (Sams-Dodd
1995). Additionally, we found lower rearing activity and increased repetitive self-grooming in isolated
GAD67+/GFP and
GAD67+/+ mice. Our current results reveal that independent from genotype social isolation rearing increases stress and additionally alters specific aspects of explorative and social behavior. The latter is supported by the finding, that GAD67 haplodeficiency itself provokes higher corticosterone levels and enhances maternal and fetal stress vulnerability (Uchida et al.
2011). However, it is important to mention that not only genetic manipulations or social isolation stress could compromise the behavioral phenotype of an animal model. Additionally, the choice of the genetic background strain might influence the outcome of different tests paradigms. This is important, since it was shown, that the genetic background strain influences locomotor activity and anxiety like behavior (Bothe et al.
2005; Voikar et al.
2005).
Negative symptoms of schizophrenia-like poor social drive are commonly associated with depressive-like behavior (Häfner et al.
1999). To analyze depressive-like behavior, we investigated
GAD67+/GFP and
GAD67+/+ mice in the forced swim test (FST) which is based on the assumption that immobility reflects a measure of behavioral despair (Borsini and Meli
1988).
GAD67+/GFP mice showed increased time spent in immobility, indicating an increase of depressive-like behavior. It was shown that the administration of low doses GABA or GABA agonists can ameliorate forced swimming induced depressive-like behavior and are able to potentiate the effect of antidepressants (Borsini et al.
1988; Aley and Kulkarni
1989). On the other hand, GABA antagonists like picrotoxin increase the immobility in FST (Poncelet et al.
1987). Therefore, we suggest GAD67 haplodeficiency provokes depressive-like behavior as found in neuropsychiatric diseases. Social isolation only resulted in higher latency to the first immobility period but did not affect immobility in the FST in general. This is in line with studies (Yates et al.
1991; Hall et al.
1998; Simpson et al.
2012) showing that social isolation in rodents has no effect on immobility in the FST or can promote despair-like immobility only when isolated in a short period of time during early brain development (17–21 day-old animals). Interestingly, group-housed
GAD67+/GFP, compared to
GAD67+/+ mice, showed an increased climbing behavior in the FST, which is reported to be a predictor of an increased motor activity in this test (Lino-de-Oliveira et al.
2005; Vieira et al.
2008). Therefore, we suggest that
GAD67+/GFP show an increased motor activity in FST, possible correlate of the positive symptom domain of schizophrenia (Jones et al.
2011).
Comorbid anxiety disorders are present in more than one third of patients with schizophrenia (Pokos and Castle
2006). The elevated plus maze (EPM) is commonly used as a behavioral assay to study anxiety like behavior in animal models (Walf and Frye
2007). However, our results reveal no differences in time spent in the open arms or locomotor activity between
GAD67+/GFP and
GAD67+/+ mice. This is further supported the study of Smith (
2018) reporting no differences in anxiety related behavior of
GAD67+/GFP on EPM. Modulation of the GABAergic system was shown to affect anxiety like behavior on EPM. GABA
A receptor agonists like diazepam or chlordiazepoxide, increase the proportion of time spent in open arms on EPM, whereas GABA
A-receptor antagonists like picrotoxin reduce this measure (Lister
1987; Rodgers et al.
1992). It is possible that the decrease in GAD67 is not significant enough to provoke alterations of anxiety like behavior in
GAD67+/GFP mice. This assumption is partially supported by the finding, that
GAD67+/GFP mice showed no alterations in exploring the unprotected center area in social interaction task in the open field. Therefore, GAD67 haplodeficiency alone or in combination with social isolation rearing did not affect anxiety related behavior of
GAD67+/GFP mice, as demonstrated in the EPM and open-field test (Sandhu et al.
2014; Smith
2018). Thus,
GAD67+/GFP mice may not be a suitable model to investigate the anxiety-related symptom domain of neuropsychiatric disorders.
Reductions of GAD67 and PARV in mouse cerebral cortex and hippocampus are associated with novelty-induced hyperlocomotion (Belforte et al.
2010) considered to be a sign of schizophrenia-like behavior in human and mutant mice (Laviola et al.
2006). However, spontaneous locomotor activity and motor coordination in social interaction and rotarod test were not affected in
GAD67+/GFP mice, indicating preserved motor functions. Our findings are supported in part by the work of Smith (
2018). This study also showed no genotype differences in EPM test, but showed a mild hyperactivity of
GAD67+/GFP mice in the open field may be result of a longer duration of testing. Social isolation stress induced higher locomotor activity and total arm entries of isolated
GAD67+/GFP mice on EPM, compared to group-housed
GAD67+/GFP. This effect was not significant between isolated and group-housed
GAD67+/+ mice. Since it was shown that postweaning social isolation affects GABAergic function (Hickey et al.
2012; Lim et al.
2012) and increases locomotor behavior on EPM (Abramov et al.
2004; Voikar et al.
2005), it is likely that the decrease of GAD67 alone is not significant enough to alter locomotor activity in
GAD67+/GFP mice. Therefore, we suggest that the additional exposure to social isolation stress as a “second hit” increases the vulnerability of a GAD67 haplodeficiency in
GAD67+/GFP mice, which results in impaired locomotor activity on the EPM. By contrast, spontaneous locomotor activity of isolated
GAD67+/GFP was not increased in the social interaction test in the open field.
It is unlikely that social interaction or aggression is reduced in
GAD67+/GFP mice because of a general deficit in sensorimotor function. Since several neuropsychiatric diseases including schizophrenia are accompanied by deficits in sensory information-processing (Braff et al.
2001), we analyzed
GAD67+/GFP mice and controls for their startle response (ASR) and prepulse inhibition (PPI). However, we found no genotype differences, which suggest that GAD67 haplodeficiency has no influence on ASR and PPI. By contrast, mice lacking GAD67 primarily in PARV-IR neurons show a reduction in PPI indicating the important role of these neurons in sensorimotor gating (Fujihara et al.
2015). It is possible, that the reduction of GAD67 and consequently GABA in
GAD67+/GFP mice is too small to cause sensorimotor gating deficits (Kolata et al.
2018). Since social isolation is used to model deficient sensorimotor gating in schizophrenia (Varty et al.
2006), we assumed that isolation rearing disrupts sensorimotor gating in our mice. Interestingly, neither
GAD67+/GFP nor
GAD67+/+ mice showed ASR or PPI deficits as a result of post-weaning isolation housing. In consequence, we varied the parameters for PPI measurement. First, we increased the interval between prepulse and pulse to 400 ms and, second, varied the prepulse intensities (70, 75 and 80 dB SPL). Additionally, no genotype or housing differences were found (data not shown). We conclude that social isolation does not affect startle activity or sensorimotor gating in
GAD67+/GFP and
GAD67+/+ mice, which is in line with other studies showing that social isolation during the critical developmental period has no effect on sensorimotor gating (Pietropaolo et al.
2008; Kulesskaya et al.
2011). However, PPI deficiency could be compromised by the duration of social isolation housing (Tueting et al.
2008), resulting in a comparatively strong decrease of GABA expression levels in
GAD67+/+ control mice. Thus, long-term social isolation in adulthood could obscure the PPI deficits (Tueting et al.
2008) in
GAD67+/GFP mice compared to
GAD67+/+, which could explain the negative findings in the present study.
Tyrosine hydroxylase-IR neurons and fibers
Dopaminergic dysfunction in association with GAD67 deficiency is implicated in the pathophysiology of schizophrenia and associated with alterations in the hippocampus and amygdala (Laviolette
2007; Lodge and Grace
2011b; Brisch et al.
2014). Therefore, we were interested if GAD67 haplodeficiency is accompanied by alterations of the dopaminergic system. Tyrosine hydroxylase (TH) catalyzes the first and rate-determining step of the catecholamine biosynthesis and is expressed in all catecholaminergic neurons. Therefore, antibodies against TH-positive structures denote dopaminergic as well as noradrenergic neurons and fibers, but with regional differences (Asan
1993). We used tyrosine hydroxylase (TH)-immunoreactivity (IR) as a marker for dopamine since it is known to predominantly represent mesencephalic DAergic input in hippocampal CA1 of mice (Walling et al.
2012; for discussion see Nullmeier et al.
2014). In the amygdala, regionally different DAergic and noradrenergic innervation patterns are described and TH-IR fibers are found to be predominantly dopaminergic afferent fibers (Asan
1997).
GAD67+/GFP mice, compared to
GAD67+/+, showed no difference in density of TH-IR neurons and volume of substantia nigra (SN) and ventral tegmental area (VTA). However, we found that
GAD67+/GFP mice exhibit a significantly higher density of TH-IR fibers in CA1 subfield of dorsal hippocampus, which can be interpreted as a hypercatecholaminergic, presumably hyperdopaminergic, innervation. Additionally, we investigated both genotypes for differences in TH-IR fiber density in the amygdala, but could not find any differences in the investigated subdivisions (Asan
1997).
It has been shown, that the dorsal hippocampus is crucial for spatial and long-term memories (Lodge and Grace
2011b; Brisch et al.
2014; Ragland et al.
2017) and modulates anxiogenic effects in the social interaction test (File et al.
1998). Additionally, dopamine D1-receptors in dorsal hippocampus were reported to mediate social learning and social behaviors in mice (Matta et al.
2017). Thus, it is possible, that the deficits in social interaction of
GAD67+/GFP mice are consequence of a dysfunction of the hippocampal dopaminergic system. However, we only found alterations in dorsal hippocampal CA1, but not in the amygdala and dopaminergic midbrain regions. It is likely, that the higher density of TH-IR fibers in CA1 of
GAD67+/GFP mice is induced locally by a GABAergic deficit in the hippocampus.
GAD67 reduction in the hippocampus is expected to lead to hyperactivity of midbrain dopamine neurons via a polysynaptic pathway (Kalkman and Loetscher
2003; Lodge and Grace
2011b). This appears to be mediated especially by a dysregulation of PARV-containing neurons (Lodge and Grace
2011a). A recent study, investigating GAD67 haplodeficiency showed that maternal stress postnatally decreases especially the density of PARV-positive GABAergic neurons in prefrontal cortex, hippocampus and somatosensory cortex of
GAD67+/GFP mice (Uchida et al.
2014). As a consequence, GABAergic dysfunction in hippocampus could lead to alterations in the dopaminergic pathways and subsequently social behavior, as found in the present study. Additionally, it can be assumed that alterations of local GABAergic circuits in VTA could provoke dopaminergic alterations. Thus, hyperdopaminergic innervation in association with a disturbance in social (Sandhu et al.
2014) and depressive-like behavior may mimic important aspects of neuropsychiatric disorders like schizophrenia (Lisman et al.
2008; Grace
2012).
GAD67+/GFP as a mouse model of schizophrenia and major depressive disorder
Schizophrenia and major depressive disorder (MDD) are considered two distinct neuropsychiatric diseases. However, there is an overlap between negative symptoms of schizophrenia and certain depressive symptoms like anhedonia, avolition and social withdrawal (Siris et al.
1988). Additionally, both disorders were shown to share alterations in biological markers of GABAergic transmission. Similar to schizophrenia, reductions of cortical GAD67 expression and alterations in GABA
A and GABA
B receptor levels were reported in MDD (Fatemi et al.
2005; Abdallah et al.
2015; Fogaca and Duman
2019). In contrast to schizophrenia, cortical GABA levels appear mostly decreased in MDD (Sanacora et al.
1999; Tayoshi et al.
2010; Luscher and Fuchs
2015). It is suggested that this reduction of GABA in MDD could not only result from decreased levels of GAD67, but could also result from a reduction in the density of specific GABA interneuron subclasses (reviewed in Fogaca and Duman
2019). This is supported by studies showing reduced volumes of prefrontal cortex and hippocampus (MacQueen et al.
2008; Savitz and Drevets
2009) and decreased cortical densities of calbindin- and somatostatin-positive GABAergic interneurons in MDD patients (Rajkowska et al.
2007; Sibille et al.
2011; Luscher and Fuchs
2015). PARV-positive interneurons play an important role in cognitive function, emotional response and social interaction (Ferguson and Gao
2018). However, the expression of parvalbumin (PARV) and density of PARV-positive interneurons mostly appear unaltered in MDD, which is in contrast to schizophrenia (reviewed in Fogaca and Duman
2019). On the other hand, rodent studies showed that the exposure to chronic stress or social isolation cause reductions of PARV-positive neurons in prefrontal cortex and hippocampus (Czeh et al.
2015,
2018; Todorovic et al.
2019). That maternal stress, in addition to a heterozygous deletion of GAD67, diminishes neurogenesis of GABAergic neurons was also shown in
GAD67+/GFP mice. Consequently, this postnatally results in a decreased density of PARV-positive interneurons in hippocampus, prefrontal and somatosensory cortex of
GAD67+/GFP mice (Uchida et al.
2014; Wang et al.
2018), similar to that found in human schizophrenia. Therefore, prenatal and social isolation stress could disturb the function of specific interneuron subpopulations and mechanism underlying the control of behaviors related to mood and emotion in neuropsychiatric disorders like schizophrenia and MDD.
GAD67+/GFP mice may provide a useful model for studying the impact of a heterozygous deletion of GAD67 on an enhanced vulnerability to prenatal and social isolation stress.