Parkinson’s disease (PD) is a progressive neurodegenerative disorder characterized by motor symptoms such as bradykinesia, rigidity, tremor at rest and postural instability, which arise mainly from dysfunction of the nigrostriatal dopaminergic pathway (Jankovic, 2008; Braak and Del Tredici, 2009). PD is also accompanied by non-motor symptoms including autonomic dysfunction, cognitive and psychiatric abnormalities (e.g., dementia, depression, apathy, anxiety and hallucination), sleep disorders and sensory abnormalities (Jankovic, 2008). Non-motor symptoms in PD correlate with advancing age and disease severity, although some non-motor symptoms including olfactory problems, constipation and depression, can occur early in the disease (Chaudhuri et al., 2006).

Depression and anxiety appear in approximately 40% (4%–70%) and 50% of all PD patients, respectively (Cummings, 1992; Leentjens et al., 2008). Non-motor symptoms greatly contribute to a reduced quality of life for patients with PD (Yamamoto, 2001; Edwards et al., 2002). The classic treatment L-3,4-dihydroxyphenylalanine (levodopa) has little effect on most non-motor symptoms (Chaudhuri et al., 2006), and chronic administration is associated with a risk of depression and anxiety (Damãsio et al., 1971; Marsh and Markham, 1973; Choi et al., 2000; Nègre-Pagès et al., 2010; Eskow Jeunarajs et al., 2011). There are some differences in neural circuitry dysfunction between major depression and PD-associated depression. For instance, suicidal tendencies and expression of guilt and self-blame are rarely observed in depressed PD patients, in contrast to patients with major depressive disorder (Brooks and Doder, 2001; Lemke, 2008; Starkstein et al., 2008). Postmortem studies of brains from PD patients have shown stage-dependent deposition of aggregated α-synuclein and neuronal loss in multiple brain areas such as serotonergic neurons in the raphe nuclei, noradrenergic neurons in the locus coeruleus, dopaminergic neurons in the substantia nigra and ventral tegmental area, and cortical neurons in regions interconnected with limbic structures (Braak et al., 2004). Conversely, brains from patients with major depression have reductions in gray-matter volume and glial density in the prefrontal cortex and hippocampus (Krishnan and Nestler, 2008), and dysfunction in the prefrontal–subcortical circuits including the amygdala, ventral striatum, hippocampus and dorsal raphe nucleus (Heller, 2016). One of the key pathophysiological mechanisms in major depression is impaired negative feedback control of the hypothalamic–pituitary–adrenal (HPA) axis, resulting in progressively unrestrained glucocorticoid release (Holsboer, 2000). Sustained elevation of glucocorticoid concentration under conditions of prolonged and severe stress, which may damage hippocampal neurons, is not observed in depressed PD patients (McEwen, 2000; Sapolsky, 2000). Although adequate treatment is needed for depression and anxiety in PD, their pathophysiology remains poorly understood.

The classical monoamine hypothesis of depression suggests that the disorder arises from a deterioration of noradrenergic/serotonergic function. In depressed PD patients, there is evidence that levels of norepinephrine (NE; Braak et al., 2004) and serotonin (5-HT; Kish et al., 2008) in the brain and/or cerebrospinal fluid are lower than in healthy people. Serotonergic and noradrenergic neuron dysfunction occurs even in the preclinical period of PD. Therefore, noradrenergic and serotonergic systems may play a significant role in the manifestation of depression and anxiety in PD. Some clinical studies suggested that tricyclic antidepressants such as desipramine and reboxetine (NE reuptake inhibitors), nortriptyline (an NE and 5-HT reuptake inhibitor) and citalopram (a selective 5-HT reuptake inhibitor) improved depressive symptoms in PD patients (Devos et al., 2008; Lemke, 2008; Menza et al., 2009). However, there was insufficient evidence to support the efficacy of antidepressants for the treatment of depression in PD (Liu et al., 2013a). Psychiatric symptoms in PD are still considered difficult to treat, possibly owing to concerns of exacerbation of parkinsonism by antidepressants (Leo, 1996). Therefore, a monotherapeutic agent to treat both motor and non-motor symptoms of PD would be a valuable therapeutic strategy in early PD, eliminating the risk of adverse drug interactions.

Bone marrow stromal cell antigen-1 (BST-1) was first isolated as a cell surface molecule that supported the cell growth of pre-B cells (Kaisho et al., 1994; Ishihara and Hirano, 2000) and clustered in CD157 in Leucocyte Typing VI after genetic cloning (Itoh et al., 1994; Muraoka et al., 1996; Okuyama et al., 1996; Ishihara et al., 1997). CD157/BST-1 is a member of the NADase/ADP-ribosyl cyclase family, to which CD38 also belongs (Hirata et al., 1994; Itoh et al., 1994; Ferrero et al., 1999; Ishihara and Hirano, 2000; Guse, 2005; Malavasi et al., 2006, 2008; Higashida et al., 2012; Lee, 2012). CD157/BST-1 has a variety of roles in the humoral immune response, neutrophil transmigration and hematopoietic stem cell support (Ishihara and Hirano, 2000; Funaro et al., 2004; Podestà et al., 2005; Malavasi et al., 2006; Mouchiroud et al., 2013). CD157/BST-1 is also involved in the pathogenesis of several diseases such as survival of B lymphocytes in rheumatoid arthritis, progression of leukemia, and metastasis of human ovarian carcinoma cells (Kaisho et al., 1994; Shimaoka et al., 1998; Ishihara and Hirano, 2000; Ortolan et al., 2010; Quarona et al., 2013; Lo Buono et al., 2014). Recently, genome-wide association studies and meta-analyses for PD identified intronic single-nucleotide polymorphisms (SNPs) in the CD157/BST1 gene on the human chromosome 4p15 as a new susceptibility locus in Asian and European populations (Satake et al., 2009; Tan et al., 2010; Liu et al., 2011, 2013b; International Parkinson Disease Genomics Consortium et al., 2011; Saad et al., 2011; Simón-Sánchez et al., 2011; UK Parkinson’s Disease Consortium et al., 2011; Zimprich, 2011; Lill et al., 2012; Sharma et al., 2012). In our previous study, CD157/BST1 knockout (CD157 KO) mice showed depression-like behaviors in the forced swimming test (FST) and the tail suspension test, anxiety-like behaviors in the open field test (OFT), the light dark transition test and the elevated plus maze test, and impaired social behaviors in the social preference test, which in part resemble psychiatric symptoms observed in patients with PD (Jankovic, 2008; Kummer et al., 2008). In contrast, CD157 KO mice did not show any degeneration of nigrostriatal dopaminergic neurons, any apparent motor dysfunction, or any alteration in dopaminergic neuron susceptibility to 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP; Lopatina et al., 2014). Genetic and environmental factors may be needed to consider the real pathogenic role of CD157 SNPs or deletion, as suggested by a previous study (Chen et al., 2014). And additionally, we cannot exclude the possibility that psychiatric phenotypes in CD157 KO mice are related to autism spectrum disorder (ASD), because it was reported that CD157/BST1 SNPs showed significant association with ASD (Yokoyama et al., 2015). Although further studies remain to be carried out in order to elucidate how CD157 mutation or deletion contributes to pathogenic process of PD or psychiatric symptoms, our previous study suggests that young adult CD157 KO mice are possibly a genetic rodent model for psychiatry symptoms associated with PD.

Selegiline, a selective and irreversible monoamine oxidase (MAO)-B inhibitor, is widely used for the treatment of PD (Birkmayer et al., 1977), as well as for major and atypical depression at higher doses to inhibit both MAO-A and -B activity (Varga and Tringer, 1967; Mann and Gershon, 1980; Birkmayer et al., 1984). MAO-B inhibitors block the metabolism of dopamine (DA), and increase DA concentration in the synaptic cleft (Youdim, 2013). DA induces motivation, reward and hedonic states, and plays an important role in neuropsychiatric disorders such as depression. Selegiline was reported to have antidepressant effects mediated by the activation of D1 and D2 receptors in normal and depressed mice (Shimazu et al., 2005; Amiri et al., 2016). Moreover, several studies have shown that selegiline enhances the expression of brain-derived neurotrophic factor in cultured murine astrocytes and in the anterior cingulate cortex of mice (Mizuta et al., 2000; Gyárfás et al., 2010), and prevents dopaminergic neurons from degeneration (Zhu et al., 2008; Youdim, 2013; Kong et al., 2015). In de novo PD patients, a double-blind, randomized, placebo-controlled clinical study demonstrated that selegiline monotherapy improved depression scores in the Hamilton Depression Rating Scale and the mental subscale of the Unified PD Rating Scale, and also improved motor scores (Allain et al., 1993). However, to our knowledge, there are no reports of the effects of selegiline in animal models of psychiatric symptoms of PD. In addition, it is not clear whether the antidepressant effect of selegiline reported in the above-mentioned clinical study is independent of its effect on motor symptoms.

Here, we investigated the effect of selegiline on depression- and anxiety-like behaviors in CD157 KO mice, a PD-related genetic model, and on biochemical parameters in their emotion-related brain regions. Furthermore, to clarify whether antiparkinsonian drugs acting on dopaminergic system have commonly the antidepressant action, we compared effects of selegiline on depression-like behavior in CD157 KO mice with those of the antiparkinsonian drugs rasagiline (another MAO-B inhibitor) and pramipexole (a D2/D3 receptor agonist). We demonstrated that selegiline exerted significant antidepressant effects in a dose-dependent manner and showed a tendency to improve anxiety-like behavior in CD157 KO mice. The antidepressant effects of selegiline could be related to an enhancement in dopaminergic signaling, and normalization of dysfunction in the monoaminergic system and HPA axis.

Depression and anxiety are common non-motor symptoms in patients with PD, and have a detrimental effect on quality of life (Yamamoto, 2001; Edwards et al., 2002). Attributed to the monoaminergic dysfunction of PD, depression and anxiety in patients with PD are commonly treated with antidepressants for major depression. However, there is insufficient evidence for the efficacy and safety of this approach (Shabnam et al., 2003; Liu et al., 2013a). In the present study, after exposure to the FST, CD157 KO mice had lower striatal and hippocampal 5-HT content and cortical NE content than WT mice. Administration of selegiline showed a recovery in forced swimming stress-induced decreases in striatal 5-HT and cortical NE concentrations in CD157 KO mice to those in WT mice. Although no differences between genotypes were detected in DA content or turnover in the brain regions examined (cortex, striatum, amygdala and hippocampus) after the FST, the dopaminergic agents pramipexole, selegiline and rasagiline alleviated depression-like behavior in CD157 KO mice. This suggests that depression-like behavior in CD157 KO mice is associated with dopaminergic dysfunction in brain regions we did not examine, such as the mesolimbic pathway. Thus, these data suggest that forced swimming stress leads to serotonergic, and noradrenergic dysfunction in CD157 KO mice, and that pharmacological modification of monoaminergic functions may exert antidepressant effects in CD157 KO mice.

The enzyme MAO exists as two distinct subtypes, MAO-A and B, the former having a higher affinity for NE and 5-HT, and the latter for phenylethylamine. DA is a substrate of both enzymes. In CD157 KO mice, the selective MAO-B inhibitor selegiline (3 and 10 mg/kg) significantly reduced immobility time in the FST, and at 10 mg/kg increased striatal and hippocampal DA content and decreased striatal DA turnover after the FST. Moreover, 10 mg/kg selegiline normalized FST-induced decreases in striatal 5-HT and cortical NE content after the FST in CD157 KO mice. The antidepressant effects of selegiline in the FST may be mediated by the activation of D1 receptors (Shimazu et al., 2005; Amiri et al., 2016). Our data therefore suggest that the antidepressant effect of selegiline is attributable at least partially to the improvement of serotonergic, noradrenergic and dopaminergic dysfunction in CD157 KO mice. From the point of view of effective doses of selegiline, its antidepressant efficacy would be greater in CD157 KO mice than in WT mice or in the normal male ddY mice used by Shimazu et al. (2005). The difference in the effective doses of selegiline between the respective genotypes might arise from changes in monoaminergic dysfunction and/or expression of depression- and anxiety-associated proteins in CD157 KO mice, suggesting that an effective dose of selegiline for the treatment of depression in PD may be lower than that in major depression.

Selegiline (3 and 10 mg/kg) significantly reduced immobility time in CD157 KO mice, and at 10 mg/kg increased climbing time. Single administration of selegiline at 10 mg/kg did not influence the swimming speed of CD157 KO mice in the FST (Supplementary Figure) or the total distance traveled in the OFT in CD157 KO mice (data not shown). These data suggest that selegiline’s antidepressant effects do not necessarily result from increasing climbing behavior and are not elicited by stimulating motor activity, unlike stimulants such as methamphetamine, which reduces immobility time in the FST by causing hyperlocomotion (Kitada et al., 1981; Shimazu et al., 2005). Interestingly, treatment with rasagiline, pramipexole or mirtazapine did not increase climbing time in CD157 KO mice, suggesting that the mechanisms underlying their antidepressant effects are different from those of selegiline. Several groups have demonstrated that NE reuptake inhibitors increase climbing behavior in the FST (Rénéric et al., 2002a,b; Cryan et al., 2005). In the present study, there was no significant correlation between cortical NE content in selegiline-treated CD157 KO mice and their climbing time (data not shown), despite a dose-dependent increase in cortical NE content. Although a 50% inhibitory dose on striatal MAO-A activity is approximately 2.5 mg/kg for both selegiline and rasagiline in rats following i.p. administration (Youdim and Tipton, 2002), at 10 mg/kg, selegiline but not rasagiline significantly increased climbing time in CD157 KO mice. Therefore, the increase in climbing time by selegiline might be related to an enhancement in noradrenergic transmission by other mechanisms such as monoamine reuptake inhibition (Lamensdorf et al., 1999) in brain regions we did not examine. It is widely accepted that climbing behavior in the FST is a specific activity aimed at escaping from the cylinder, and is one of the indexes for antidepressant-like behavior (Cryan et al., 2005; Perona et al., 2008), but we speculate that the increase in climbing time in CD157 KO mice potentially reflects an altered emotional state (Lopatina et al., 2014). Therefore, the emotional state in which saline-treated CD157 KO mice are motivated to try climbing may be qualitatively different from that in selegiline-treated CD157 KO mice. Further investigation is required to clarify the meaning of climbing behavior in CD157 KO mice.

Rasagiline is more potent as an MAO-B inhibitor than selegiline, and has a similar selectivity for MAO-B over MAO-A in vivo (Youdim et al., 2001). Its LED for motor symptoms in patients with PD is one-tenth of selegiline’s LED (Tomlinson et al., 2010). In CD157 KO mice, the antidepressant-like effects of selegiline at 10 mg/kg had a tendency to be more potent than those of rasagiline at 1 mg/kg (t = −1.602, P = 0.120; Figure 1C), suggesting that selegiline is probably more effective for ameliorating depression in PD patients than rasagiline at doses that are therapeutically effective for the motor symptoms of PD. It also suggests that the MAO inhibition cannot entirely account for the antidepressant effects of selegiline. Our data are in line with clinical studies that selegiline improved Hamilton Depression Rating Scale scores in de novo PD patients (Allain et al., 1993), but rasagiline was not effective on Beck Depression Inventory scores (Barone et al., 2015).

There is insufficient evidence to support the efficacy of mirtazapine on depression and anxiety in PD patients, although its antidepressant effect in major depression is comparable to that of tricyclic antidepressants (Watanabe et al., 2011), which might be effective in PD-related depression (Devos et al., 2008; Lemke, 2008; Menza et al., 2009). In CD157 KO mice, repeated administration of mirtazapine ameliorated depression-like behavior in the FST, and anxiety-like behavior and low sociability in the OFT. These results are consistent with the effects of mirtazapine in the tail suspension test and in the elevated plus maze test shown in our previous study (Lopatina et al., 2014). CD157 KO mice showed decreases in 5-HT content in the cortex and amygdala, following the OFT, but mirtazapine failed to influence monoaminergic dysfunction in these brain regions in CD157 KO mice. Therefore, mechanisms underlying anxiety and low sociability of CD157 KO mice and anxiolytic effects of mirtazapine may be attributable principally to other monoaminergic systems, non-monoaminergic systems or neuroplasticity-associated protein expression (Ishima et al., 2014; Kadoguchi et al., 2014; Bittolo et al., 2016).

Like other inescapable stress paradigms, the FST elevates blood corticosterone concentrations in rodents (Steiner et al., 2008; Rogóz et al., 2012). Interestingly, plasma corticosterone concentrations after the FST were lower in CD157 KO mice than in WT mice, although there were no significant differences in baseline levels between genotypes (Figures 2A,B). Surprisingly, following the FST in CD157 KO mice, selegiline (1−10 mg/kg) normalized plasma corticosterone concentrations to those in WT mice (Figure 2B). Selegiline has a tendency to ameliorate dose-dependently the depression-like behavior and post-FST monoamine levels, but not post-FST corticosterone concentrations, in CD157 KO mice, suggesting that such effect of selegiline on post-FST corticosterone concentrations in CD157 KO mice is independent of enhancement of monoaminergic function in brain regions examined. Inputs from the amygdala elicit activation of the HPA axis (Sandi, 2004), and amygdala-lesioned rats show significantly lower plasma corticosterone and adrenocorticotropic hormone levels after certain types of stress than intact animals (Feldman and Conforti, 1981; Beaulieu et al., 1987). In our previous study, the amygdala in CD157 KO mice seemed to be smaller than in WT mice (Lopatina et al., 2014); therefore, atrophy of the amygdala in CD157 KO mice might be related to the reduction in plasma corticosterone concentrations following the FST. Moreover, amygdala dysfunction in CD157 KO mice may also be related to their anxiety and depression-like behavior, because abnormalities in the amygdala might be involved in PD progression and may contribute to elicitation of non-motor symptoms such as anxiety and depression in PD (Huang et al., 2015; van Mierlo et al., 2015; Vriend et al., 2016). Thus, selegiline may ameliorate depressive symptoms by normalizing the hypoactivity of the HPA axis arising from the amygdala.

CD157 KO mice had higher levels of anxiety in the novel environment (session 1) and toward a novel non-social object (session 2) than WT mice, confirming previous results (Lopatina et al., 2014; Mizuno et al., 2015). Moreover, when a novel object was placed in the apparatus (session 2), CD157 KO mice showed a higher level of anxiety than in session 1 (percent of time in center zone in Figure 3B vs. Figure 3F, t = 2.794, P < 0.05). In the experiments with a social target (session 3), CD157 KO mice seemed more nervous than WT mice, and showed weak sociability to the novel mouse, which is characteristic of their phenotype (Lopatina et al., 2014). Furthermore, their phenotype is possibly in line with clinical descriptions of social phobia occurring in 50% of PD patients (Kummer et al., 2008), and depressed PD patients having significantly fewer social ties than non-depressed PD patients (Starkstein et al., 1990). Repeated injection of selegiline at 1 mg/kg for 3 days had a tendency to improve the anxiety-like behavior of CD157 KO mice in the novel environment (session 1), but did not alleviate the higher level of anxiety toward a novel non-social object (session 2) or the weak sociability to the novel social target (session 3). This suggests that selegiline improves mild anxiety but not severe anxiety or low sociability.

In the present study, there was a moderate negative correlation between 3-MT/DA ratio in the cortex and amygdala and immobility time in saline and selegiline-treated CD157 KO mice in the FST (cortex, r = −0.496, P < 0.005; amygdala, r = −0.499, P < 0.01). At 10 mg/kg, selegiline increased the 3-MT/DA ratio in the cortex, amygdala and hippocampus, and reduced the DOPAC/DA ratio in the cortex, striatum and amygdala (data not shown), probably resulting from brain MAO-B inhibition. Intracerebroventricular administration of 3-MT increases behavioral activity (Nakazato and Akiyama, 2002; Sotnikova et al., 2010), and 3-MT and phenylethylamine have affinity for trace amine-associated receptor 1, which is involved in neuropsychiatric disorders including depression (Shi et al., 2016). Therefore, the antidepressant effect of selegiline might be mediated via MAO-B inhibition-induced enhancement of 3-MT and phenylethylamine content.

In conclusion, mice lacking a PD-related gene CD157 show the depression- and anxiety-like behaviors and impairment of their brain monoamine content after exposure to stress. Selegiline exerted some anxiolytic effects in addition to antidepressant effects in CD157 KO mice. These results highlight the potential of selegiline as an antiparkinsonian agent with the efficacy in CD157 mutation-related depressive and anxiety symptoms.

HH, TY, OL and SK conceived and designed the research. SK performed experiments, analyzed data, and prepared the initial draft; HH and KI revised the manuscript. All authors reviewed the final manuscript and approved its publication.

This work was supported by a grant-in-aid from “Integrated research on neuropsychiatric disorders” carried out under the Strategic Research Program for Brain Sciences. It was also supported by the Industry–Academia Collaborative R&D Program from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and was a collaborative research project with FP pharmaceutical corporation.

SK is an employee of FP Pharmaceutical Corporation. The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.