Background

Establishing animal models of psychiatric disorders by utilizing genetically engineered mice is essential for investigating the pathogenesis, pathophysiology, and treatment of the disorders [1–5]. Previously, we reported that forebrain-specific calcineurin (also called protein phosphatase 2B) knockout (KO) mice have severe working/episodic-like memory deficits [6] and exhibit a spectrum of abnormal behaviors similar to those of schizophrenic patients [7]. In addition, we identified the PPP3CC gene, which encodes the calcineurin gamma subunit, as a potential schizophrenia susceptibility gene [8]. These studies demonstrated the usefulness of a comprehensive behavioral test battery for genetically engineered mice to efficiently evaluate a mouse model of human psychiatric disorders. Thus, we have applied this approach to test various strains of mice bearing mutations of genes encoding molecules involved in calcineurin signaling pathways or calcineurin-related neural mechanisms [5, 9, 10]. Here we focused on neuronal nitric oxide synthase (nNOS), one of the calcineurin substrates in the nervous system [11, 12].

Nitric oxide (NO) is a highly diffusible gas that acts as an endogenous messenger molecule in various tissues. In the brain, NO has a variety of important roles, including regulation of neurotransmission, synaptic plasticity, gene expression, and neurotoxicity [13–15]. NO is enzymatically synthesized from L-arginine by nitric oxide synthase (NOS). In the mammalian nervous system, NO is primarily produced by nNOS, an isoform predominantly expressed in the brain among three NOS isoforms [14]. nNOS is expressed in a discrete population of neurons in the hippocampus, cortex, striatum, cerebellum, olfactory bulb, and brain stem [16, 17]. nNOS catalytic activity is regulated by the phosphorylation state of the enzyme. The phosphorylation of nNOS by protein kinase C (PKC) and Ca2+/calmodulin-dependent kinases inhibits nNOS activity [18, 19], whereas dephosphorylation by calcineurin activates nNOS [20]. Direct binding of nNOS to PSD-95 protein induces nNOS to localize at a postsynaptic density in the vicinity of NMDA receptors, allowing for an efficient and specific activation of nNOS in response to a glutamate-induced Ca2+ influx [21, 3G; genotype effect, p < 0.0001] and locomotor activity was significantly lower in nNOS KO mice [Fig 3H; genotype effect, p = 0.0041]. These phenotypes were observed both in the light period [Fig 3I, J; genotype effect, mean number of particles, p = 0.0002; activity level, p = 0.0015] and in dark period [Fig 3I, J; genotype effect, mean number of particles, p = 0.0001; activity level, p = 0.0152]. Even at the same activity level, nNOS KO mice tended to remain separated compared to wild-type, indicating that the increased contact in nNOS KO mice was not due to hyperactivity (Fig 3K). ANCOVA applied to the mean number of particles detected, with activity level as a covariate, indicated a significant interaction between activity level and genotype (p = 0.0122) and the effect of genotype remained when activity was used as the covariate in the ANCOVA (p < 0.0001). Analysis of the relationship between activity level and the interaction between two mice indicated that nNOS KO mice showed an increased number of contacts at the same activity level compared to wild-type mice in their home cage. These findings indicate that nNOS deficiency induces an increase in social interaction in the familiar environment.

Crawley's three-chamber social approach test consists of sociability test and a social novelty preference test. These tests assess social interaction that is relatively independent of locomotor activity compared to the other social interaction tests, because the preference of the mice can be quantified based on the time spent around a wire cage containing a stranger mouse vs. an empty cage in the sociability test and stranger mouse vs. a familiar mouse [50]. In the sociability test, both nNOS KO mice and wild-type mice type demonstrated normal sociability [Fig 3L, M; time spent around cage, with stranger vs. empty; wild-type: t(26) = 3.804, p = 0.0008, nNOS KO: t(21) = 2.465, p = 0.0224, paired-t test]. In nNOS KO mice, however, social approach was decreased in the sociability test [time spent around the cage with the stranger, genotype effect, F(1,47)= 5.15, p = 0.0279]. Consistently, nNOS KO mice did not show a preference for the chamber with the stranger [Fig 3N, O; time spent in chambers (stranger 1 side vs. empty cage side); wild-type mice: t(26) = 3.134, p = 0.0042; nNOS KO mice: t(21) = 0.874, p = 0.3920, paired-t test]. The distance traveled in the sociability test was not significantly different between genotypes [Fig 3T; genotype effect, F(1,47) = 1.541, p = 0.2206]. In the social novelty preference test, wild-type mice tended to demonstrate a preference for novelty [Fig 3P; time spent around the cage containing stranger 1 vs. that containing stranger 2: t(26) = 1.969, p = 0.0597, paired-t test], whereas nNOS KO mice did not [Fig 3Q; time spent around the cage containing stranger 1 vs. that containing stranger 2: t(21) = 1.686, p = 0.1067, paired-t test]. A preference between chambers was not detected in either genotype [Fig 3R,S; time spent around cages, wild-type mice: t(26) = 1.234, p = 0.2284, nNOS KO mice: t(21) = 1.592, p = 0.1262, paired-t test]. Although there were no significant difference in the distance traveled in the sociability test (Figure 3T), nNOS KO mice traveled a greater distance in the novelty preference test [Fig 3U; genotype effect, F(1,47) = 11.094, p = 0.0017]. The ratio of distance traveled in the novelty preference test to the distance traveled in the sociability test was higher in nNOS KO mice [Fig 3V; genotype effect, F(1,47) = 5.439, p = 0.0240], suggesting that nNOS KO mice habituated less than wild-type mice.

Performance deficits of nNOS KO mice in the memory tasks

In the eight-arm radial maze test (spatial working memory task), the number of revisiting errors, in which subjects returned to the arms that had been visited previously to retrieve a food pellet, was not significantly different between genotypes during trials without a delay [Fig 4A; genotype effect, F(1,23) = 1.050, p = 0.3161; genotype × trial interaction, F(14,322) = 0.990, p = 0.4636]. The number of different arm choices among the first 8 entries, which is considered a measure of working memory that is relatively independent of locomotor activity levels, and the total number of arm choices were not significantly different between genotypes [Fig 4B; genotype effect, F(1,23) = 2.100, p = 0.1608; genotype × trial interaction, F(14,322) = 1.325, p = 0.1905]. On the other hand, both the number of revisiting errors and the different arm choices among the first 8 entries were significantly greater in nNOS KO mice during trials with delays [Fig 4C; genotype effect, F(1,23) = 5.880, p = 0.0236; genotype × trial interaction, F(2,46) = 1.837, p = 0.1708, Fig 4D; genotype effect, F(1,23) = 5.985, p = 0.0225; genotype × trial interaction, F(2,46) = 1.350, p = 0.2693], suggesting that nNOS KO mice have mildly impaired working memory.

Figure 4
figure 4

Impaired remote spatial memory of nNOS KO mice. (A-D) Eight-arm radial maze test: total number of arms revisited (A, C) and different arm choices among the first 8 entries (B, C) during training were recorded. During trials 32–36, a delay was applied after of the first 4 pellets were consumed (C, D). (E-G) Morris water maze test: latency to escape (E) was recorded during training session and probe tests. Probe tests were conducted 1 day (F) and 7 days after the last training trial (G). The p values indicate genotype effect in two-way repeated measures ANOVA (A-E) and genotype effect in two-way ANOVA (F, G).

Full size image

In the Morris water maze (spatial reference memory task), latency to locate the escape platform during hidden platform training (for 14 successive days) in nNOS KO mice was significantly greater than that in wild-type mice [Fig 4E; genotype effect, F(1,22) = 6.145, p = 0.0213; genotype × trial interaction, F(13,286) = 1.174, p = 0.2981]. Probe trials were performed on day 15 and day 23 (1 week after the last trial). Both nNOS KO mice and wild-type mice spent more time in the previously trained quadrant than in the three untrained quadrants. Time spent in the previously trained quadrant was not significantly different between genotypes on the probe trial of the day 15 [Fig 4F; genotype effect, F(1,22) = 0.346, p = 0.5625]. nNOS KO mice spent significantly less time in the trained quadrant than wild-type mice on day 23 [Fig 4G; genotype effect, F(1,22) = 7.636, p = 0.0133]. These data suggest that nNOS KO mice have impaired spatial remote memory.

In the Porsolt forced swim test, nNOS KO mice showed decreased depression-related behavior [Fig 2A–C; genotype effect, first trial: F(1,28) = 9.660, p = 0.0043, a day after the first trial: F(1,28) = 18.554, p = 0.0002, 7 days after the first trial: F(1,28) = 8.263, p = 0.0076]. nNOS KO mice traveled a greater distance [Fig 2D–F; genotype effect, first trial: F(1,28) = 7.901, p = 0.0089, a day after the first trial: F(1,28) = 6.266, p = 0.0184, 7 days after the first trial: F(1,28) = 7.816, p = 0.0093], reflecting the hyperactivity in the nNOS KO mice. In the second and third trials, immobility during the first minute was similar to that during the last minute in the previous trial in wild-type mice [immobility in last minute of the first trial vs. immobility in first minute of the second trial, t(16) = 1.679, p = 0.1126; immobility in last minute of the second trial vs. immobility in first minute of the third trial, t(16) = 0.500, p = 0.6240, paired-t test], suggesting that wild-type mice remembered the previous event. On the other hand, nNOS KO mice showed less immobility during the first 1 min in the second or third trials compared to that during the last minute in the previous trial [immobility during the last minute of the first trial vs. immobility during the first minute of the second trial, t(12) = 3.442, p = 0.0049; immobility during the last minute of the second trial vs. immobility during the first minute of the third trial, t(12) = 3.875, p = 0.0022, paired-t test]. This finding might indicate that nNOS KO mice have impaired reference memory for stressful events.

Together, these data suggest that nNOS deletion impairs spatial working memory, remote spatial reference memory and reference memory for stressful events.

Increased D1-mediated dopaminergic signaling in brain slices of nNOS KO mice

Glutamate and dopamine regulate DARPP-32 phosphorylation in neostriatal neurons via the activation of multiple signaling cascades [45]. To examine the effect of nNOS deletion on glutamatergic and dopaminergic signaling, we investigated the regulation of protein phosphorylation by glutamate and a D1 receptor agonist in striatal slices from nNOS KO mice (Fig 5).

Figure 5
figure 5

Regulation of protein phosphorylation by glutamate and a D1 receptor agonist in brain slices from nNOS KO mice. Neostriatal slices from wild-type (open circles) and nNOS KO (closed circles) mice were treated with glutamate (5 mM) (A-C) and a dopamine D1 receptor agonist, SKF81297 (1 μM; D-I) for the indicated times. Changes in the phosphorylation of DARPP-32 at Thr34 (B, E) and Thr75 (C, F), GluR1 at Ser845 (G), ERK2 (H), and spinophilin at Ser94 (I) were determined by Western blotting using phosphorylation-state specific antibodies. Typical immunoblots detected with phosphorylation-state specific and total DARPP-32 antibodies are shown in (A, D). Data represent means ± SEM for 3–9 experiments. *p < 0.05, **p < 0.01 compared with untreated slices (time 0) from wild-type mice; †p < 0.05, ††p < 0.01 compared with untreated slices (time 0) from nNOS KO mice; one-way ANOVA followed by Newman-Keuls test. §p < 0.05 compared with values of wild-type mice; two-way ANOVA followed by Bonferroni test.

Full size image

Treatment of neostriatal slices from wild-type mice with glutamate (5 mM) induced a rapid increase in DARPP-32 Thr34 phosphorylation after 30 s incubation, but the effect was transient (Fig 5A, B). We previously reported that the effects of glutamate are mediated via the activation of nNOS/NO/cGMP/PKG signaling [45]. The glutamate-induced increase in Thr34 phosphorylation was absent in nNOS KO mice. Other than the rapid and transient increase in Thr34 phosphorylation, the phosphorylation levels of DARPP-32 at Thr34 and Thr75 under basal conditions and during incubation with glutamate were similar between wild-type and nNOS KO mice [genotype effect on DARPP-32 Thr34, F(1, 31) = 0.4146, p = 0.1927; genotype effect on DARPP-32 Thr75, F(1, 32) = 0.01168, p = 0.6053; Fig 5B,C]. The D1 receptor agonist SKF81297 (Fig 5D–I) increased the phosphorylation of DARPP-32 at Thr34 (Fig 5E), GluR1 at Ser845 (Fig 5G), ERK2 (Fig 5H), and spinophilin at Ser94 (Fig 5I), all of which are phosphorylated by PKA, in both wild-type and nNOS KO mice. The increases in DARPP-32 Thr34 and GluR1 Ser845 phosphorylation were significantly higher in nNOS KO mice than in wild-type mice [genotype effect on DARPP-32 Thr34, F(1, 96) = 12.47, p = 0.0007; genotype effect on GluR1 Ser845, F(1, 94) = 9.199, p = 0.0032], but the increases in ERK2 and spinophilin Ser94 phosphorylation were similar between wild-type and nNOS KO mice [genotype effect on ERK2, F(1, 94) = 2.491, p = 0.1189; genotype effect on spinophilin Ser94, F(1, 96) = 0.6535, p = 0.4212] (Fig 5E,G–I)]. These results suggest that dopamine D1 receptor signaling is upregulated in the striatum of nNOS KO mice in a substrate specific manner.

Treatment of wild-type slices with SKF81297 decreased DARPP-32 Thr75 phosphorylation, presumably via the activation of PP-2A/B56δ by PKA and increased dephosphorylation of Thr75 [57, 58] (Fig 5F). The SKF81297-induced decrease in Thr75 phosphorylation was not observed in striatal slices from nNOS KO mice [genotype effect on DARPP-32 Thr75, F(1, 98) = 12.74, p = 0.0006], suggesting that regulation of PP-2A activity by D1 receptor/PKA signaling is also altered in nNOS KO mice.

Increased D1-mediated dopaminergic signaling in nNOS KO mice in the PPI test

Though low dose of D1 receptor agonist does not alter PPI in mice [59], high dose of it disrupts PPI [60]. D1 receptor antagonist also disrupted PPI [61, 62]. Because dopamine D1 receptor signaling seems to be upregulated in nNOS KO mice, we examined the effect of D1 receptor agonist, SKF81297, on PPI in nNOS KO mice. PPI is a cross-species phenomenon in which the startle response is reduced when the startle stimulus is preceded by a low intensity prepulse, and is disrupted in certain neuropsychiatric disorders that are characterized by abnormal sensorimotor gating, such as schizophrenia [61]. The effect of dopamine agonists on PPI differs between species, and D1 receptor agonists disrupt PPI in mice [62–64]. nNOS KO mice and wild-type mice were tested in the PPI test without drug and with a low dose of the D1 receptor agonist SKF81297 (1 mg/kg, intraperitoneally). In the test without drug, acoustic startle and PPI were not significantly different across genotypes [Fig 6A; genotype effect, F(1,91) = 0.223, p = 0.6383; Fig 6B; genotype effect, F(1,91) = 1.345, p = 0.2492 (110 dB), F(1,91) = 0.424, p = 0.5167 (120 dB)]. Wild-type mice showed a significantly decreased acoustic startle response following injection with SKF81297 [Fig 6C; drug effect, F(1,29) = 5.336, p = 0.0282], but nNOS KO mice did not (Fig 6E). Neither injection with saline nor SKF81297 significantly altered PPI in wild-type mice (Fig 6D). In contrast, PPI was disrupted in nNOS KO mice after injection with SKF81297 [Fig 6F; drug effect, F(1, 25) = 5.115, p = 0.0327 (120 dB)]. These data indicate that PPI is easier to disrupt with a D1 receptor agonist in nNOS KO mice compared to wild-type mice, and suggest that nNOS KO mice have upregulated D1 mediated signaling. Thus, increased D1-mediated dopaminergic signaling was also demonstrated in nNOS KO mice at the behavioral level.

Figure 6
figure 6

D1 receptor agonist-induced disruption of prepulse inhibition in nNOS KO mice. Startle responses at 110 dB and 120 dB were recorded (A). Startle response following 74 dB and 78 dB prepulse inhibition stimuli were recorded (B). Effect of SKF81297 administration on startle responses and prepulse inhibition in wild-type mice (C, D) and in nNOS KO mice (E, F). The p values indicate genotype effect (A, B) and drug effect (C-F) in repeated measures ANOVA.

Full size image

Discussion

In the present study, nNOS KO mice exhibited increased locomotor activity in the open field test. Hyperactivity of nNOS KO mice was also consistently observed in other tests, such as the light/dark transition, elevated plus maze, and social interaction (novel environment) tests. To date, locomotor activity of nNOS KO mice has not been well examined. Although a few studies have reported on the locomotor activity of nNOS KO mice [24, 26, 65], most of the observations were either not quantitative and/or subject animals were not compared with appropriate control animals. Nelson et al. reported that nNOS KO mice displayed normal locomotor activity, but they in fact assessed locomotor balance and coordination [24]. In the study by Bilbo et al., nNOS KO mice showed increased locomotor activity in the open field test only in the second test, performed during the dark phase, but not during the light phase [65]. In addition, they compared nNOS KO mice with non-littermate C57BL/6 controls. Comparing behaviors between mutant mice and non-littermate mice may lead to the detection of an effect of different environments rather than the mutation itself. Behavioral phenoty** of genetically engineered mice should be done with control animals from the same standardized genetic and environmental background as the engineered mice [66, 67]. Otherwise, an observed difference in the phenotype could be caused by differences in the breeding environment or genetic background. Weitzdoerfer et al. used an observational test battery and reported increased locomotor activity in nNOS KO mice [26]. Because their methods depended heavily on human observation, however, the data were not quantitative. They used nNOS KO mice with only three-generations of backcrossing [26]. In mice from an N3 backcross, there is theoretically 12.5% of the donor strain genome, therefore the phenotype could be caused by flanking genes [68]. The donor strain of the nNOS KO mice in their study was 129/Sv, which is hypoactive compared to the C57BL/6 strain. The hyperactivity of nNOS KO mice may have therefore been masked by the low activity of the donor strain in their study. Mice used in the present study were N5 backcrossed and had a predominantly C57BL/6J genetic background (94.5%), assessed by analyzing 100 microsatellite makers. Because further backcrossing into C57BL/6J leads to low fertility in nNOS KO mice, we could not backcross them further. Gryuko et al. also reported that complete elimination of nNOS, including splicing variants, caused infertility [47].

Pharmacologic studies indicate that administration of a NOS inhibitor reduces locomotor activity in rodents [69–71]. NOS inhibitors also cause a lack of motor coordination in rodents, assessed by the rotarod test [71]. There is a high expression of nNOS in the cerebellum and motor coordination is highly cerebellar-dependent. If administration of an nNOS inhibitor induces an acute lack of motor coordination, this may present as reduced locomotor activity. In our study, although muscle strength, as assessed by the grip strength test, was reduced in nNOS KO mice, they performed normally in the rotarod test, suggesting that other molecule(s) or residual splicing variants of nNOS [47] compensate for the loss of nNOS. With normal motor coordination, nNOS KO mice might display increased locomotor activity. Kriegsfeld et al. reported a deficit in the balance and coordination of nNOS KO mice in a balance test on the pole and plank only during the dark phase [25]. In our study, all experiments except the home cage social interaction test were performed during the light phase, therefore the effect of a lack of motor coordination on locomotor activity in nNOS KO mice, if any, would be smaller than that during the dark phase.

Abnormal social behavior of nNOS KO mice

Although there are some studies reporting the social behavior of nNOS KO mice [26, 72] and the social behavior of animals treated with nNOS inhibitors [72–75], the results are inconsistent. Some studies report that treatment with a NOS inhibitor decreases social interaction behavior [72, 75]; others report that treatment with a NOS inhibitor [74] and nNOS knockout [26] does not affect social interaction behavior; and still others report that treatment with a NOS inhibitor increases social interaction behavior [73]. These contradictory findings might be the result of different methods or conditions used to assess social behavior. Additionally, in most of the studies, only one kind of experiment was conducted, and therefore it is difficult to compare the results between them. In the present study, to assess social behavior of nNOS KO mice in various situations, we conducted four kinds of social interaction tests. nNOS KO mice showed 1) an increased number of contacts and an increased total duration of active contacts in a novel environment (one-chamber social interaction test), which might reflect hyperactivity, 2) increased social interaction behavior in their home cage, and 3) decreased social approach behavior in Crawley's three-chamber social approach test.

In the social interaction test in a novel environment (one-chamber social interaction test), the mouse was exposed to a stranger mouse in the chamber and both mice were able to move freely. Weitzdoerfer et al. reported normal social interaction behavior in nNOS KO mice [26]. Their experimental conditions for the social interaction test were similar to those of our one-chamber test. Although the indices they used were the number of social behaviors such as sniffing, grooming, mounting, rubbing, and fighting, all of which are heavily dependent on human observation, their findings were consistent with ours.

The social interaction test in the home cage in the present study revealed increased social interaction behavior in nNOS KO mice. Although there are no reports of increased social interactions in nNOS KO mice, a pharmacologic study showed that the administration of a NOS inhibitor increases social interactions in rats [73]. The finding that NOS inhibition in rats increases social interactions in a novel environment, and nNOS deficiency in mice increases social interactions in familiar environments may reflect an interspecies difference.

In the sociability test (three-chamber), the stranger mouse was in a wire cage and was unable to move freely, and the subject mouse could therefore approach the stranger mouse. Additionally, because this test apparatus (three-chamber) was larger than the one-chamber apparatus, the subject mouse could remain far away from the stranger than in the one-chamber social interaction test. This situation may decrease social investigative behavior in nNOS KO mice. Although nNOS KO mice spent less time around the stranger than the control mice, there was no significant difference between genotypes in the time spent around the empty cage and the distance traveled, indicating that nNOS KO mice showed normal exploration of novel objects and novel environments. Therefore, the decreased social investigative behavior of nNOS KO mice was not due to neophobia.

In the social novelty preference test, a different stranger mouse contained in a wire cage was added to the empty chamber in the sociability test, and then the subject mouse was allowed to explore the cage with the familiar mouse and the cage with the stranger mouse. Distance traveled by wild-type mice in the social novelty preference test was reduced compared to that in the sociability test, probably because those tests were conducted in succession. The ratio between the distance traveled in the social preference test and that in the sociability test was higher in nNOS KO mice than in control mice, indicating that exploration was reduced less in nNOS KO mice than in control mice. Impaired habituation of nNOS KO mice in the three-chamber test may be interpreted as a cognitive impairment. In a study by Bohme et al., rats treated with an NOS inhibitor did not show reduced exploration of a juvenile rat during a second exposure, suggesting that NOS inhibition impaired social recognition [74]. Their finding was consistent with ours in the social novelty preference test. nNOS inhibition also impairs olfactory learning [74], which might be reflected by the impaired habituation of the nNOS KO mice in the present study.

Trainor et al. demonstrated decreased social interaction behavior in nNOS KO mice [72] in which a stranger mouse isolated by a wire barrier was introduced into the home cage and the social behavior of nNOS KO mice was observed. nNOS KO mice spent less time near the barrier compared to control mice, indicating decreased social investigation by the nNOS KO mice. Trainor et al. reported a similar finding in mice administered an nNOS inhibitor. Although their experiments were performed using a home cage, the conditions of the test were similar to those of our three-chamber sociability test regarding the introduction of an animal into the cage. Thus, the results of our three-chamber sociability test are consistent with their findings, i.e., decreased social investigative behavior by nNOS KO mice.

In the present study, nNOS KO mice showed abnormal social behaviors, such as increased social interaction in their home cage, decreased social investigation in a social preference test, and normal social behaviors in the one-chamber social interaction test. Together, the findings from the various social behavior tests indicated that nNOS KO mice demonstrate increased social behavior with a familiar mouse in familiar conditions and they exhibit normal or mildly decreased social behavior with an unfamiliar mouse. Dysregulated social behaviors are often observed in patients with psychiatric disorders such as schizophrenia. Associations between the nNOS gene and schizophrenia have been reported [29, 76–78]. Moreover, a recent study demonstrated significantly more disruption or structural variants in genes involved in NO signaling pathways in schizophrenic patients than in normal controls [31], suggesting the involvement of nNOS and NO signaling pathways in schizophrenia. Thus, abnormal nNOS function might be involved in dysregulated social behavior in a subpopulation of schizophrenic patients.

Impaired reference memory retention and working memory in nNOS KO mice

The eight-arm radial maze task is a hippocampus-dependent task that is generally used to evaluate working memory in rodents [79, 80]. Both nNOS KO and wild-type mice exhibited normal working memory when the delay between each arm choice was 5 s. nNOS KO mice exhibited mild deficits in working memory, however, when the delay was increased to 30 s.

Spatial reference memory is frequently assessed by the hidden platform version of the Morris water maze, another test dependent on hippocampal function [81]. In a previous study, nNOS KO mice showed deficits in reaching the hidden platform 10 to 14 days after training in the Morris water maze; although memory immediately after training was not examined, the findings were interpreted as a memory recall deficit [26]. In the same study, nNOS KO mice performed well in a multiple T-maze test, a less stressful spatial task, leading the authors to conclude that the memory deficit of the nNOS KO mice was observed only under stressful conditions [26]. In the present study, however, both nNOS KO and wild-type mice spent significantly more time in the targeted quadrant than in the other three quadrants 1 day after the last training (day 15), indicating that the nNOS KO mice are able to learn, remember, and recall the platform location normally with short retention delays, even under stressful conditions. Further, these findings indicate that the spatial memory deficit of the nNOS KO mice is not likely due to an abnormal sensitivity to stress. On the other hand, nNOS KO mice failed to search the target quadrant 7 days after the last training (day 23), suggesting that the nNOS KO mice have impaired memory retention rather than impaired memory recall. These results are consistent with an idea that NO acts as an important retrograde message for long-term potentiation (LTP) [13] In hippocampal slices of nNOS KO mice, NO-dependent LTP was only slightly reduced, but otherwise normal, probably because the lack of NO was compensated for by the endothelial isoform of NO (eNOS) [82–84] or by residual nNOS splice variants [85]. A recent study revealed a potential role of nNOS in late-phase LTP [86], a finding that is consistent with the behavior of nNOS KO mice in the water maze test in the present study, because late-phase LTP involvement is implicated in the maintenance/storage of long-term memory [87].

Increased D1 receptor-mediated protein phosphorylation in nNOS KO mice

Activation of dopamine D1 receptors stimulates cAMP/PKA signaling, leading to the phosphorylation of PKA substrates such as DARPP-32 at Thr34 and GluR1 at Ser845 in striatal neurons [37]. Phosphorylation of DARPP-32 and GluR1 induced by the activation of dopamine D1 receptors was enhanced in nNOS KO mice compared to wild-type mice. However, enhanced D1 receptor/PKA signaling was not detected in the analysis of spinophilin Ser94 and ERK2 phosphorylation. The biochemical study clearly demonstrated that D1 receptor/PKA signaling in nNOS KO mice is upregulated in striatal neurons, although the upregulation is substrate-specific. The increase in D1 receptor/PKA signaling detected in the striatum of nNOS KO mice could be applied for brain regions involved in PPI, and supports the findings of D1 receptor-mediated disruption of PPI in nNOS KO mice. In addition, activation of the D1 receptors is known to induce an increase in locomotor activity [88] and a decrease in depression-related behavior in the forced swim test [89]. Hyperactivity and a decrease in depression-related behavior in Porsolt forced swim test, observed in nNOS KO mice, might be explained by the upregulated D1 receptor signaling.

As nNOS deletion in somatostatin-positive interneurons and the subsequent reduction of NO/guanylyl cyclase/cGMP/PKG signaling in medium spiny neurons upregulates D1 receptor/PKA signaling in the striatum, it is possible that the NO/PKG pathway has an inhibitory influence on D1 receptor/adenylyl cyclase/cAMP/PKA signaling in medium spiny neurons, leading to the suppression of DARPP-32 Thr34 and GluR1 Ser845 phosphorylation. Thus, the NO/PKG pathway has bidirectional effects on DARPP-32 Thr34 phosphorylation: PKG and PKA phosphorylate DARPP-32 at Thr34, whereas PKG likely inhibits D1 receptor/adenylyl cyclase/cAMP/PKA signaling upstream of DARPP-32. The molecular mechanisms by which PKG modifies D1 receptor/adenylyl cyclase/cAMP/PKA signaling require further elucidation.

Activation of dopamine D1 receptors decreases the phosphorylation of DARPP-32 at Thr75 (Cdk5-site) via PKA-dependent activation of PP-2A/B56δ [57, 58]. In agreement, DARPP-32 Thr75 phosphorylation was decreased by D1 receptor activation in wild-type mice. In contrast, nNOS KO mice did not show any changes in DARPP-32 Thr75 phosphorylation after D1 receptor activation. The findings are different from the predicted results, because PKA signaling is upregulated and PKG signaling, which increases DARPP-32 Thr75 phosphorylation [90], is downregulated in nNOS KO mice. The reason for the lack of Thr75 dephosphorylation in response to D1 receptor activation in nNOS KO mice is unknown. Phospho-Thr75 DARPP-32 inhibits PKA, and the inhibition is removed when D1 receptor/PKA signaling is activated [57]. The positive feedback loop for PKA activation seems to be impaired in nNOS KO mice possibly due to the high PKA tone. Alternatively, it is possible that activity of PP-2A/B56δ is modulated by PKG, although highly speculative.

PPI of an acoustic startle induces a reduced startle response to the startle stimulus when the stimulus is immediately preceded by a weaker prestimulus [91]. PPI is naturally observed in humans and other animals including rodents, but it is often disrupted in psychiatric disorders such as schizophrenia [92]. PPI is disrupted by pharmacologic manipulations with a psychotomimetic drug, phencyclidine, and dopamine agonists [61]. The role of nNOS in the behavioral effect of phencyclidine has been investigated using the NOS inhibitor and nNOS KO mice. A NOS inhibitor, L-NAME, blocked the phencyclidine-induced decrease in PPI [93, 94]. Interestingly, the effect of phencyclidine in nNOS KO mice was opposite [95]. In that study, treatment with phencyclidine increased PPI in nNOS KO mice but not in wild-type control mice [95]. Although there is a discrepancy between studies using the NOS inhibitor and nNOS KO mice, those studies demonstrate that nNOS/NO signaling plays a critical role in the regulation of PPI.

In the present study, activation of D1 receptors with a low dose of SKF81297 disrupted PPI in nNOS KO mice, but not in wild-type mice. A number of studies demonstrated that manipulations of dopamine signaling alter PPI in rodents [59–62, 64], and pharmacologic studies indicate the relationship between high dopamine signaling and disruption of PPI [61]. SKF81297, a D1 agonist, in a relatively high dose compared to that used in the present study, decreased PPI [60], whereas the D1 antagonist increased PPI in rats [96, 97]. It is likely that dopamine D1 receptor signaling is upregulated in nNOS KO mice as demonstrated by biochemical studies, and therefore PPI is disrupted in nNOS KO mice in response to a low dose of SKF81297.

There are several studies that report abnormal behavior of hyperdopaminergic mice. Dopamine transporter (DAT) knockdown mice, that are known as hyperdopaminergic, displayed hyperactivity [98, 99], perseverative motor behavior [99], and impaired response habituation [98]. In addition, DAT knockout mice also showed hyperactivity, perseverative motor behavior, disrupted prepulse inhibition, and high sensitivity to D1 receptor antagonist [100]. Behavioral phenotypes observed in nNOS KO mice, such as hyperactivity, perseverative motor behavior (increased stereotypic behavior in open filed test), impaired habituation in three-chamber social interaction test, and hypersensitivity to D1 receptor antagonist in PPI test, resemble those in DAT knockdown and knockout mice, suggesting the hyperdopaminergic state of nNOS KO mice.

Conclusion

nNOS KO mice were subjected to a battery of behavioral tests. nNOS KO mice exhibited hyperactivity, impaired memory, decreased depression-related behavior, abnormal social behavior and D1 receptor-mediated disruption of PPI. Biochemical analysis in the striatum revealed the upregulation of dopamine D1 receptor/PKA signaling in nNOS KO mice. Some of behavioral abnormalities in nNOS KO mice such as hyperactivity, decreased depression-related behavior and D1 receptor-mediated disruption of PPI might be explained by high activity of dopamine D1 receptor/PKA signaling.