Mice lacking the Parkinson's related GPR37/PAEL receptor show non-motor behavioral phenotypes: age and gender effect


  • S. Mandillo,

    Corresponding author
    • CNR-National Research Council, IBCN-Institute of Cell Biology and Neurobiology, EMMA-Infrafrontier-IMPC, Monterotondo Scalo, Rome, Italy
    Search for more papers by this author
  • E. Golini,

    1. CNR-National Research Council, IBCN-Institute of Cell Biology and Neurobiology, EMMA-Infrafrontier-IMPC, Monterotondo Scalo, Rome, Italy
    Search for more papers by this author
  • D. Marazziti,

    1. CNR-National Research Council, IBCN-Institute of Cell Biology and Neurobiology, EMMA-Infrafrontier-IMPC, Monterotondo Scalo, Rome, Italy
    Search for more papers by this author
  • C. Di Pietro,

    1. CNR-National Research Council, IBCN-Institute of Cell Biology and Neurobiology, EMMA-Infrafrontier-IMPC, Monterotondo Scalo, Rome, Italy
    Search for more papers by this author
  • R. Matteoni,

    1. CNR-National Research Council, IBCN-Institute of Cell Biology and Neurobiology, EMMA-Infrafrontier-IMPC, Monterotondo Scalo, Rome, Italy
    Search for more papers by this author
  • G. P. Tocchini-Valentini

    1. CNR-National Research Council, IBCN-Institute of Cell Biology and Neurobiology, EMMA-Infrafrontier-IMPC, Monterotondo Scalo, Rome, Italy
    Search for more papers by this author

Corresponding author: S. Mandillo, CNR-IBCN, Via E. Ramarini, 32—Monterotondo Scalo (RM), 00015 Italy. E-mail: silvia.mandillo@cnr.it


Non-motor symptoms in Parkinson's disease (PD) have been often described at different stages of the disease but they are poorly understood. We observed specific phenotypes related to these symptoms in mice lacking the PD-associated GPR37/PAEL receptor. GPR37 is an orphan G-protein-coupled receptor highly expressed in the mammalian central nervous system. It is a substrate of parkin and it is involved in the pathogenesis of PD. GPR37 interacts with the dopamine transporter (DAT), modulating nigro-striatal dopaminergic signaling and behavioral responses to amphetamine and cocaine. GPR37 knockout (KO) mice are resistant to MPTP and exhibit several motor behavioral abnormalities related to altered dopaminergic system function. To evaluate non-motor behavioral domains, adult and aged, male and female GPR37 KO mice and their wild-type (WT) littermates were analyzed in a series of cross-sectional studies. Aged GPR37 KO female mice showed mild improvements in olfactory function, while anxiety and depression-like behaviors appeared to be significantly increased. A reduction of the startle response to acoustic stimuli was observed only in adult GPR37 KO mice of both genders. Furthermore, HPLC analysis of major neurotransmitter levels revealed gender differences in the striatum, hippocampus and olfactory bulb of mutant mice. The absence of GPR37 receptor could have a neuroprotective effect in an age and gender-dependent manner, and the study of this receptor could be valuable in the search for novel therapeutic targets.

Parkinson's disease (PD) is no longer considered a uniquely ‘movement’ disorder. The traditional triad of symptoms, tremors, rigidity and bradykinesia, are often accompanied by various ‘non-motor’ disturbances that affect the majority of patients either in the early premorbid phases or during later stages of the disease. Often, non-motor or behavioral symptoms in PD are under recognized (Shulman et al. 2002) while they produce substantial impairment in the quality of life (Witjas et al. 2002). Most common ‘non-motor’ disturbances are in the domains of olfaction, gastrointestinal function, mood, sleep, learning and sensorimotor processing; these deficits have been linked to dopaminergic and other neurotransmitter system alterations (Chaudhuri & Odin 2010; Chaudhuri et al. 2005; Langston 2006). Several studies have attempted to evaluate the treatment of non-motor symptoms in PD (Obeso et al. 2010; Zesiewicz et al. 2010).

There is a male prevalence in PD, and it is known that gender influences its clinical manifestation (Miller & Cronin-Golomb 2010; Shulman & Bhat 2006) with non-motor psychiatric symptoms more distressing in female patients (Negre-Pages et al. 2010; Scott et al. 2000). Estrogens have an important moderating role (Dluzen & Horstink 2003; Leranth et al. 2000) and more specific therapies are evaluated for women with PD (Price & Shulman 2008). Studies that have reported non-motor behavioral impairments in mouse models related to PD (Mcdowell & Chesselet 2012; Taylor et al. 2010) rarely addressed gender as a factor. Gender differences and effects of estrogens were observed in parkin null mouse mutants (Rodriguez-Navarro et al. 2008). Parkin is mutated in juvenile autosomal recessive PD and ubiquitinates various substrates including the PD associated GPR37/PAEL receptor (Imai et al. 2001).

In this study, we extended the behavioral characterization of GPR37/PAEL receptor KO mice to non-motor functions. GPR37 is an orphan G-protein-coupled receptor highly expressed in the mammalian brain (Marazziti et al. 1997, 1998). It is found in Lewy bodies and its insoluble aggregates accumulate in brain samples of PD patients (Murakami et al. 2004). The absence of the receptor is beneficial for the nigrostriatal dopaminergic neurons as GPR37 KO mice are resistant to MPTP, although their striatal dopamine content is reduced with consequent motor behavioral abnormalities (Marazziti et al. 2004). GPR37 colocalizes with the dopamine transporter (DAT) modulating its presynaptic functional expression (Marazziti et al. 2007). It also affects nigrostriatal dopaminergic signaling and behavioral response to amphetamine and cocaine (Marazziti et al. 2011). Moreover, the expression of the GPR37 gene is upregulated in PD and downregulated in major depression patients (Cantuti-Castelvetri et al. 2007; Sibille et al. 2009).

We performed behavioral analysis of male and female GPR37 KO mice of different ages to evaluate olfactory and intestinal functions, anxiety and depression-like behaviors, emotional memory and sensorimotor abilities, and to investigate if a particular phenotype could be influenced by age and gender. Increased anxiety and depression-like behaviors and reduced acoustic startle response was observed in the mutant mice, such abnormalities were gender and/or age dependent.

Materials and methods

Subjects and housing

Homozygous Gpr37−/− mutant mice (KO) and their wild-type littermates (WT) were generated from heterozygous matings on a mixed genetic background (75% C57BL/6J, 25% 129P2/OlaHsd) (Marazziti et al. 2004) or on N7 backcrossing to C57BL/6J (used in acoustic startle response experiment only).

Mice were born and bred in a mouse facility under standard housing conditions in ventilated racks. All cages were provided with a transparent red polycarbonate igloo house (Datesand, Manchester, UK) and at each weekly cage-change with wood shavings contained in single cellulose bags (Scobis Uno bags, Mucedola, Settimo Milanese, Italy) as forms of environmental enrichment. At weaning mice were housed in groups of four or five same sex littermates.

At least 1 week before the experiments mice were transferred to the behavioral phenotyping unit and housed in a room with a 12-h light/dark cycle (lights on at 07:00 a.m.), relative humidity 50–60% and temperature 21 ± 1°C with ad libitum access to food (Harlan, Teklad 2018, Mucedola, Settimo Milanese, Italy) and water.

Animals were subjected to experimental protocols approved by the Veterinary Department of the Italian Ministry of Health, and experiments were conducted according to the ethical and safety rules and guidelines for the use of animals in biomedical research provided by the relevant Italian laws and European Union's directives (no. 86/609/EEC and subsequent). All adequate measures were taken to minimize animal pain or discomfort.

General testing procedure

To investigate the effect of age on non-motor behavioral domains, a series of cross-sectional studies were carried out on GPR37 WT and KO female mice on a mixed genetic background. Most of tests were conducted on two age groups, at 4–6 (Adult) and 16–18-months-old (Aged). Olfactory function was probed in three types of tests (buried food, block test and habituation to nonsocial odors); intestinal function was assessed by colonic motility test. To evaluate anxiety and depression-like behaviors mice were observed in the elevated plus-maze, open field, light–dark exploration and forced swim tests. Cognitive domains as emotional memory and sensorimotor abilities were assessed measuring fear conditioned freezing and the startle response and prepulse inhibition (PPI) to acoustic stimuli.

Furthermore, gender comparison studies were carried out in most of the tests using different cohorts of male and female littermate mice. Additionally, one cohort of aged female mice (age 19 months) was used to assess the effect of acute estrogen administration on elevated plus-maze and forced swim tests. After the tests, levels of serum estradiol were measured in this group, and in an additional cohort of ovariectomized female mice (age 5 months).

In some cases tests were run in batteries in order to reduce the number of animals and from the least to the most stressful to minimize previous test influence. Appropriate resting intervals were given to the mice in between tests.

Aged males and females from one last cohort were sacrificed and their brain was removed to measure levels of monoamines and amino acids with HPLC analysis of microdissected olfactory bulb, striatum and hippocampus.

All behavioral studies were carried out by experimenters blind to the genotype of the animals. Testing took place between 1000 and 1700 h. Number of subjects used in each experiment is specified in the appropriate figure legend.

Behavioral tests

Olfactory tests

Female mice were individually housed in standard plastic cages (cm 30 × 18 × 15 h) and maintained as such for the duration of the olfactory tests (Fleming et al. 2008; Nathan et al. 2004; Taylor et al. 2009; Tillerson et al. 2006). A detailed description of methods for the following tests is reported in the Supporting Information: Buried food test to measure general olfactory function; Block test to evaluate the ability of mice to discriminate between social odors, specifically between self and non-self; Habituation/dishabituation test to measure the ability of mice to discriminate between nonsocial odors.

Colonic motility

To analyze colon transit, a stool frequency measure is used. Mice were placed individually in clean empty plastic cages and observed for 1 h. Fecal pellets were collected immediately after expulsion and placed in sealed 1.5 ml tubes. Tubes were weighed to calculate the wet stool weight, and then left open overnight at 65°C to allow the stools to dry. The stool water content was obtained as the difference between the wet and dry stool weight (Anderson et al. 2007; Li et al. 2006).

Open field

This test is used to assess exploratory locomotion and anxiety-related behaviors based on the conflict to venture into an exposed space. Mice were tested in a square arena (cm 44 × 44 × 33.5 h) made of gray PVC. Locomotor activity was recorded for 15 min by an automated video tracking system (ViewPoint, Lyon, France) and expressed as total distance traveled (cm). Time spent in the center of the arena was also measured and expressed as percentage of time spent in the center over the total time (index of anxiety-like behavior) (Carola et al. 2002).

Light–dark test

This test is used to analyze anxiety-related phenotypes in mice (Crawley & Goodwin 1980). The apparatus consisted of a standard rat cage (cm 36.5 × 20.7 × 14 h) equally divided in a light (580 lux) and a dark compartments (cm 18 × 20.7 each) connected by a small open door. Each mouse was placed in the illuminated side facing the wall opposite to the dark box. The latency to enter the dark side was recorded when the mouse entered the box with all four paws. After the first entry, the number of transitions between dark and light side and the time spent in each compartment was recorded for a total of 5 min. Other behaviors such as rearing, stretched attend posture and risk assessment were scored later from digital video recordings using a manual keyboard assisted custom-made software (MouseWatch 1.1, R. H. Butler, personal communication).

Elevated plus-maze

This test is used to assess anxiety-related behaviors in mice (Lister 1987). The apparatus was made of dark grey PVC and it consisted of two open arms (cm 30 × 5 with 0.3 cm ledges), two enclosed arms (cm 30 × 5 with 13 cm high side- and end-walls) and a connecting central platform (cm 5 × 5). The maze was raised to 50 cm above the floor. Illumination level in the maze was approximately 30 lux in the enclosed arms and about 100 lux in the open arms. After 30 min of acclimation to the testing room animals were placed in the central platform facing one of the open arms. During the 5-min test session the following parameters were scored manually from digital video-recordings using a custom-made computer software for event timing (MouseWatch 1.1, R. H. Butler, personal communication): number of open- and closed-arm entries (all four paws crossing), time spent in open and closed arms, frequency of protected and unprotected head dippings, frequency of reaching the far edge of open arms, frequency of stretched-attend postures, duration of grooming. After the elevated plus-maze test, from a group of adult females vaginal smears were collected with a smoothed glass pipette and placed on glass slides to assess cycling and estrous phase at the time of testing. Unstained material was checked under a light microscope (10×, 40×) and estrous phases (proestrus, estrus, metestrus and diestrus) were scored and assigned to each animal over several days (Marcondes et al. 2002).

Forced swim test

This test is used to evaluate depressive-like behavior. Each mouse was placed in a glass beaker (diameter 12.5 × 21 cm high) filled with 14 cm of water maintained at 26°C. The test session lasted 6 min and each mouse was videotaped from the side of the beaker. Time spent immobile was measured in the last 4 min; immobility was defined as passive floating on the surface (Porsolt 1979; Porsolt et al. 1977; Taylor et al. 2009).

Fear conditioning

In the fear conditioning procedure, a neutral conditioned stimulus (CS) such as light or tone is paired with an aversive unconditioned stimulus (US) such as mild footshock. Concomitantly, animals associate the background context cues with the CS. After conditioning, the CS or the spatial context elicits a central state of fear in the absence of the US, expressed as reduced locomotor activity or total lack of movement (freezing). Freezing behavior is the tendency for rodents to remain in a motionless, defensive posture after being shocked and percentage of freezing is used as a measure of learning/memory performances (Crawley 1999). Fear conditioning was assessed using an automated system (Freeze Monitor, San Diego Instruments, CA, USA). On day 1, the mouse was placed in the chamber and allowed to explore it for 9 min (conditioning session). During the 9 min the animal was exposed to a light + tone (140 lux, 92 dB) CS presented for 20 seconds and paired at the end with a mild (1 second, 0.4 mA) footshock (US) for two times, after 4 and 6 min from the beginning of the session. After 24 h the mouse was returned to the same testing chamber and its activity was recorded for 6 min where no cues or shocks were presented (context session). Four hours after the context session the chamber was modified so to change the context of the testing environment: room lights were turned off, the position of the chamber was rotated of 90°, the floor and walls were covered with black PVC plates, a vanilla scent was sprayed inside the chamber. The mouse was placed in this modified chamber and allowed to explore for 8 min. After 2 and 6 min from the beginning of the session, the light + tone cues were presented for 2 min (cue session). In all sessions, freezing behavior was automatically and manually scored (every 10 seconds) and expressed as percentage of freezing (100 × number of freezing episodes/total number of possible freezing episodes per interval).

Acoustic startle reflex and PPI

The acoustic startle reflex is characterized by an exaggerated flinching response to an unexpected auditory stimulus. This response can generally be attenuated when it is preceded by a weaker stimulus, prepulse inhibition. PPI provides an operational measure of sensorimotor gating, which reflects the ability of an animal to integrate sensory information (Swerdlow & Geyer 1998).

Acoustic startle and PPI were measured using the SR-Lab apparatus (San Diego Instruments, San Diego, CA, USA). After placing each mouse individually in the apparatus, a 5-min acclimation period was followed by 10 different trial types: acoustic startle pulse alone (white noise, 110 dB/40 ms); four different prepulse trials (76, 80, 88, or 100 dB) in which 10 ms long stimuli were presented alone or preceded the pulse (110 dB/40 ms) by 50-ms interval; one trial in which only the background noise (BN, 72 dB) was presented to measure the baseline movement of the animals. Each trial was presented 10 times in random order. The intertrial interval (ITI) was 25 seconds in average (20–30 seconds). In an acoustic startle only experiment pulse alone stimuli of 80, 85, 90, 100 or 110 dB were presented, 10 times in random order, each for 10, 20 and 40 ms. In this last test were used adult male and female mice from seven backcrossings to C57BL/6J inbred strain. The average response was used as the dependent variable to measure the startle reflex. PPI was expressed as %PPI = 100 × (pulse-prepulse)/pulse (Mandillo et al. 2008).

Determination of monoamine and amino acid levels in brain tissues

One week after the completion of behavioral tests (light/dark, colon motility and forced swim) a group of aged (15 months) females and their male littermates were sacrificed, brains removed for the dissection of olfactory bulb, hippocampus and striatum and samples frozen for subsequent HPLC analysis (Pronexus Analytical AB, Karolinska Institutet Science Park, Stockholm, Sweden). Levels of dopamine (DA), noradrenaline (NA), serotonin (5-HT) and their acidic metabolites DOPAC, HVA and 5-HIAA as well as amino acids including glutamate, glutamine and GABA in tissue homogenates were determined as described elsewhere (Galter et al. 2009; Kehr et al. 2010).

Estradiol administration and serum level measurements

β-Estradiol 3-benzoate (E2) and sesame oil (Vehicle) were purchased from Sigma-Aldrich (Saint Louis, MO, USA). E2 was dissolved in sesame oil and injected s.c. to GPR37 KO and WT intact aged female mice (19 months) at the dose of 0.25 mg/kg in a volume of 5 ml/kg. Each mouse was administered either E2 or Vehicle 1 h before the elevated plus-maze and 1 week later, 1 h before the forced swim test (Fig. S5). Dose and regimen of administration were chosen according to previous reports (Kastenberger et al. 2012; Walf & Frye 2010). Levels of circulating serum estradiol were measured in these mice right after the forced swim test (1 h from last E2 or Vehicle injection). Levels of estradiol in the Vehicle group was also compared to those of a group of 5 months old GPR37 KO and WT female mice, 6 weeks after bilateral ovariectomy (OVX). Blood was collected by submandibular vein bleed, serum was separated from whole blood, and E2 was measured by ELISA (Mouse/Rat Estradiol ELISA kit, Calbiotech Inc., Spring Valley, CA, USA) following manufacturer instructions. This kit was chosen for its high sensitivity to low levels of estrogens (<3 pg/ml) and for its good reliability (Haisenleder et al. 2011).

Statistical analysis

All data were analyzed by t-test, simple factorial analysis of variance (ANOVA) and repeated measures (RM) ANOVA with genotype, age and gender as between-subjects factors and day, trial, time, conditioning session, pulse intensity and/or duration as within-subjects factors using the StatView 5.0 PowerPC (SAS Institute Inc., Cary, NC, USA) and Prism 5.0a (GraphPad Software Inc., La Jolla, CA, USA) software packages. For each parameter, individual subject performances that scored ± 2 SD from the group mean were considered as outliers and discarded from the analysis. Post hoc analysis was performed where allowed. Level of significance was set at P < 0.05. Where heterogeneity of variance was encountered, nonparametric Wilcoxon Signed Rank test or Mann–Whitney U-test were used. Data are presented as mean ± SEM. Factor analysis of behavioral variables in elevated plus-maze, open field and fear conditioning tests using a principal component solution with unrotated factors was also performed using StatView 5.0. Factor loadings > |0.4| were reported.


Increased olfactory discrimination in aged GPR37 KO female mice

Impairments in olfactory function are found in 70–95% of PD patients, especially as a premorbid manifestation of the disease (Haehner et al. 2011). In the buried food test, both WT and KO adult female mice were able to detect and locate the hidden food although KO mice appeared slower to learn (Fig. 1a; RM ANOVA, effect of day F(2,34) = 8.72, P = 0.0009; split by genotype: WT, F(2,18) = 11.02, P = 0.0008; KO, F(2,16) = 1.58, n.s.). To monitor that mice showed similar interest and attention to the food stimulus, the latency to reach and eat a visible food pellet placed on top of the cage bedding was measured. KO mice showed a longer latency to reach the visible food than WT littermates but this difference was not statistically significant (Fig. 1a).

Figure 1.

Olfactory tests. (a) Latency to grasp a buried or visible food pellet in adult female mice (*P < 0.05, **P < 0.005 vs. day 3 in WT group, paired t-test). (b) Time spent sniffing familiar blocks in trials 1–3 (average of all four blocks) or familiar and novel blocks in trial 4 in adult female mice, 6-days procedure (**P < 0.01 vs. blocks A–C, paired t-test). (c) Time spent sniffing familiar or novel blocks in trial 4 in aged female mice, 24-h procedure (*P < 0.05 familiar vs. novel block, Wilcoxon Signed Rank test). (d) Time spent sniffing a cartridge filled with habituated (trials 1–4) vs. novel (trial 5) scents in adult female mice (**P < 0.005 trial 4 vs. trial 1 and trial 5 in WT group, Fisher's PLSD test). (e) Time spent sniffing a cartridge filled with habituated (trials 1–4) vs. novel (trial 5) scents in aged female mice (*P < 0.05 trial 4 vs. trial 5 in KO group; #P < 0.05 WT vs. KO, unpaired t-test). Adult females WT, n = 9–10; KO, n = 9; aged females WT, n = 5; KO, n = 6.

In the social odor discrimination task, no differences between genotypes were found in the time spent sniffing the familiar blocks during all trials. On trial 4, both genotypes spent significantly more time sniffing the novel block compared to all other blocks. Interestingly, KO mice seemed to spend more time sniffing the novel block compared to WT littermates although this difference was not statistically significant (Fig. 1b,c). Additionally, aged female KO mice, tested in the more challenging 24-h procedure, showed a shorter latency to contact the novel block also indicating a better odor detection (WT = 8.20 seconds, KO = 4.16 seconds; Mann-Withney U-test, P < 0.05). In the habituation/dishabituation test, the time spent sniffing the scented cartridge decreased significantly from trial 1 to trial 4 in adult WT female mice (habituation) and then significantly increased in trial 5 (novel scent). Conversely, adult KO littermates did not seem to habituate and to discriminate the novel scent (Fig. 1d; RM ANOVA split by genotype, effect of trial in WT, F(2,18) = 7.83, P = 0.0036; KO, F(2,16) = 2.33, n.s.). When a different group of female mice aged 12 months was exposed to the cartridge filled with similar odors, both genotypes failed to habituate to the peach odor (Fig. 1e). However, the time spent sniffing the cartridge was significantly higher in the KO mice (ANOVA, genotype F(1,9) = 7.09, P < 0.05, trial F(1,9) = 4.14, P < 0.05, genotype × trial F(2,18) = 1.40, n.s.). Although the interaction genotype × trial was not significant, only KO mice showed a significant increase in the time spent sniffing the novel scent (apricot) in trial 5 (RM ANOVA split by genotype, effect of trial in WT F(2,8) = 1.96, n.s; KO F(2,10) = 4.19, P < 0.05) possibly indicating a tendency to better discriminate between similar odors. No differences were observed in locomotor activity or motivation to explore because both genotypes showed a similar decrease in the number of transitions along the length of the cage across the five test trials (data not shown).

Mild alteration of gastrointestinal function in adult GPR37 KO female mice

Autonomic impairments such as gastrointestinal dysfunction have been frequently reported in PD (Pfeiffer 2003). Gastrointestinal function was evaluated in male and female GPR37 KO and WT mice of different ages. Aged females showed increased stool frequency (ANOVA, F(1,34) = 4.38, P < 0.05) compared to adult females. Aged male mice showed a tendency to increased stool frequency (Fig. 2a; ANOVA, F(1,37) = 3.11, P = 0.08) and statistically significant decrease in percentage of solid matter (Fig. 2b; ANOVA, F(1,35) = 8.38, P = 0.0065). A reduction in the percentage of stool solid matter was observed in GPR37 KO adult female mice compared to their WT littermates (unpaired t-test, t(12) = 3.02, P < 0.05) and to aged females (unpaired t-test, t(17) = 2.25, P < 0.05) (Fig. 2b; ANOVA, genotype × age F(1,28) = 3.71, P = 0.06). This reduction corresponds both in males and adult GPR37 KO females to increased water content (Fig. S2) and may thus indicate poorer intestinal water absorption.

Figure 2.

Colonic motility. (a) Stool frequency over 1-h collection (*P < 0.05 adult vs. aged, unpaired t-test) and (b) Percentage of stool solid matter (*P < 0.05 vs. KO aged females; #P < 0.01 males vs. females; °P < 0.05 WT vs. KO, unpaired t-tests). Adult females: WT, n = 10; KO, n = 9. Aged females: WT, n = 7; KO, n = 12. Males: WT, n = 10; KO, n = 12.

Anxiety and depression-like phenotypes in aged GPR37 KO female mice

Neuropsychiatric disturbances like anxiety and depression have been observed in about 40% of PD patients (Brown et al. 2011) and gender and age may influence the occurrence of these symptoms (Scott et al. 2000). Male and female mice of different ages were tested in the elevated plus-maze, open field and light/dark tests to evaluate anxiety-like behavior and in the forced swim test to assess depression-like phenotype. Mice tend to avoid high, open and brightly lit places (anxiety) and adopt an immobile posture that is interpreted as behavioral despair and inability to cope with an inescapable and stressful situation (depression) (Cryan & Holmes 2005; Porsolt et al. 1977).

Aged GPR37 KO female mice showed a lower percentage of entries and percentage of time spent in the open arms of an elevated plus-maze compared to WT littermates and adult KO females (Fig. 3a, ANOVA: % entries, age F(1,89) = 31.31, P < 0.0001, genotype F(1,89) = 4.05, P < 0.05, genotype × age F(1,89) = 6.31, P < 0.05; % time, age F(1,89) = 39.34, P < 0.0001, genotype F(1,89) = 0.56, n.s., genotype × age F(1,89) = 5.49, P < 0.05). Table S1 shows ethological parameters measured on the elevated plus-maze confirming that aged GPR37 KO female mice displayed more anxious-like behaviors. Namely, aged GPR37 KO female mice showed fewer head dippings and a higher number of stretched-attend postures than WT littermates thereby indicating a decrease in exploration and increase in risk assessment behaviors. However, no statistically significant difference between genotypes was observed in the open field and light/dark test (Fig. S3), possibly indicating an anxiety phenotype specific to the elevated plus-maze as already described (Bailey et al. 2007; Holmes et al. 2003) and confirmed by Factor analysis (see Supporting Information and Table S3).

Figure 3.

Elevated plus-maze and forced swim test. Females, age comparison: (a) Percentage of entries in the open arms, percentage of time spent in open arms and number of entries in the closed arms (***P < 0.0001 vs. adult KO group or adult vs. aged; **P < 0.005 WT vs. KO in aged group; *P < 0.05 vs. adult WT group; #P < 0.05 adult vs. aged, unpaired t-tests). Adult females: WT, n = 20; KO, n = 19. Aged females: WT, n = 25; KO, n = 29. Gender comparison: (b) Percentage of entries in the open arms, percentage of time spent in open arms and number of entries in the closed arms in aged male and female mice (***P < 0.0005 males vs. females and male vs. female WT group; **P < 0.005, WT vs. KO in female group, males vs. females; ##P < 0.01 vs. female WT group, WT vs. KO in female group, unpaired t-tests) Males: WT, n = 16; KO, n = 23. Females: WT, n = 32; KO, n = 26. (c) Percentage of time of immobility over 4 min of observation in adult and aged female and aged male mice (*P < 0.05 WT vs. KO in aged females; #P < 0.01 vs. adult female and aged male KO groups, unpaired t-tests). Adult females: WT, n = 10; KO, n = 9. Aged females: WT, n = 7; KO, n = 11. Males: WT, n = 8; KO, n = 12.

Figure 3b and Table S2 show the gender comparison of behaviors observed in aged male and female mice. Male mice showed overall more anxiety-like behaviors than females (ANOVA: % entries, gender F(1,93) = 13.45, P < 0.0005; % time, gender F(1,93) = 10.78, P < 0.005). Among GPR37 KO mice, only females showed a significantly lower percentage of entries and percentage of time spent in the open arms compared to WT littermates (ANOVA: % entries, genotype F(1,93) = 7.21, P < 0.01, gender × genotype F(1,93) = 4.095, P < 0.05; % time, genotype F(1,93) = 4.79, P < 0.05, gender × genotype F(1,93) = 4.43, P < 0.05). Males and females of the two genotypes did not differ in the frequency of closed arm entries to confirm that general levels of activity were not affecting the anxiety-related parameters. Ethologically relevant parameters scored in the elevated plus-maze are shown in Table S2. The higher number and percentage of protected stretched-attend postures and protected head dippings confirm an increased level of anxiety in aged males compared to females and, within females, in the KO compared to their WT littermates. We observed no effect or correlation of estrus on anxiety-related parameters (data not shown).

Figure 3c shows the percentage of time spent immobile in the forced swim test. Aged female GPR37 KO mice spent significantly more time immobile than their WT littermates and than adult GPR37 KO female mice. Aged GPR37 KO females also showed significantly higher percentage of immobility time than their KO male littermates. No difference between genotypes was observed in adult females or in aged males (ANOVA, genotype effect split by age and gender: adult females F(1,17) = 1.96, n.s.; aged females F(1,16) = 6.65, P < 0.05; aged males F(1,18) = 0.04, n.s.).

To evaluate the effects of acute E2 administration on the anxiety and depression phenotypes observed in aged female mice, an additional cohort of mice was used and their levels of circulating estradiol was measured and compared to those of adult OVX females. Results are described in Supporting Information.

Slightly impaired context-dependent learning in adult GPR37 KO female mice

Mild cognitive impairment such as visuospatial, attentional, executive, memory and emotion recognition deficits may be frequently observed in PD (Aarsland et al. 2011). We measured emotional memory in female mice in the fear conditioning test (Fig. 4). Overall, aged female mice exhibited more freezing than adults in both context- and cue-dependent learning. Both context- and cue-dependent learning were not altered depending on genotype but a tendency to reduced freezing to context was observed only in adult GPR37 KO mice. (RM ANOVA: Fig. 4a, context session: genotype F(1,80) = 4.21, P < 0.05; age F(1,80) = 4.21, P < 0.05; conditioning F(1,80) = 149.7, P < 0.0001; conditioning × age F(1,80) = 5.10, P < 0.05. Fig. 4b, cue session: age F(1,80) = 4.95, P < 0.05; conditioning F(1,80) = 247.05, P < 0.0001). When exposed to a new environment, aged GPR37 KO female mice showed less baseline freezing before the shock and in the new context, pre-CS. This could be due to both increased motor response and/or anxiety in a new context, as also observed in the open field test and as emerged in the factor analysis (Fig. S3 and Table S3).

Figure 4.

Fear conditioning in adult and aged female mice. (a) Percentage of freezing during baseline (first 2 min of conditioning session) and 24 h after administration of mild foot shock (US) in the same context (first 2 min of context session). (b) Percentage of freezing was measured for 2 min in a new context (cue session) 4 h after the context session, before and during the presentation of the CS (*P < 0.05, **P < 0.005 adult vs. aged and #P < 0.05, ##P < 0.005 WT vs. KO, unpaired t-test). Adult: WT, n = 22; KO, n = 23. Aged: WT, n = 19; KO, n = 20.

Reduced acoustic startle response but no deficits in PPI were observed in adult GPR37 KO female mice

Adult female GPR37 KO mice showed reduced startle response to acoustic stimuli (76–100 dB/10 ms) while no differences between genotypes were found in aged female mice (Fig. 5a,b; RM ANOVA split by age, adults: genotype F(1,50) = 5.75, P < 0.05; stimulus intensity F(4,200) = 61.67, P < 0.0001; stimulus intensity × genotype F(4,200) = 2.39, P = 0.05; aged: stimulus intensity F(4,136) = 26.24, P < 0.0001). No age or genotype effect was found in the response to the startle pulse of 110 dB/40 ms (mean startle response ± SEM; adults: WT = 88.36 ± 9.06; KO = 75.58 ± 8.55; aged: WT = 115.02 ± 18.73; KO = 96.44 ± 15.73).

Figure 5.

Acoustic startle response and PPI. (a,b) Startle response to acoustic stimuli and (c,d) percentage of inhibition of the startle response (%PPI) in adult and aged female mice (*P < 0.05, WT vs. KO, unpaired t-test). Adult: WT, n = 29; KO, n = 23. Aged: WT, n = 14–15; KO, n = 19-21. (e,f) Startle response to acoustic stimuli in adult mice (C57BL/6J genetic background; *P < 0.05, **P < 0.005, WT vs. KO, unpaired t-test). Males: WT, n = 17; KO, n = 10. Females: WT, n = 11; KO, n = 18.

When the 110 dB/40 ms startling pulse was preceded by shorter prepulses (10 ms) of lower intensity (Fig. 5c,d), no differences were observed between genotypes in the inhibition of the acoustic startle response (%PPI) but aged females showed a lower %PPI compared to adults (RM ANOVA, age F(1,81) = 42.29, P < 0.0001; age × stimulus intensity F(2,162) = 3.499, P < 0.05; stimulus intensity × genotype F(2,162) = 3.38, P < 0.05 and stimulus intensity × age × genotype F(2,162) = 4.22, P < 0.05). When the data were analyzed separately by age groups, a slight increase in %PPI was found in aged GPR37 KO females (Fig. 5d, ANOVA, stimulus intensity × genotype F(2,62) = 3.14, P = 0.05).

To further evaluate the startle response using a more refined protocol and to verify the presence of gender differences, stimuli were presented to adult male and female mice (C57BL/6J genetic background) at different intensities (80, 85, 90, 100, 110 dB) and durations (10, 20, 40 ms) (Fig. 5e,f). GPR37 KO mice of both genders showed a reduced startle response compared to WT littermates (RM ANOVA, genotype F(1,52) = 17.93, P < 0.0001; stimulus duration × intensity × genotype F(8,416) = 4.28, P < 0.0001). Male mice showed higher acoustic startle response than females especially at higher stimulus intensity irrespective of genotype (RM ANOVA, stimulus intensity × gender F(4,208) = 2.99, P < 0.05).

Altered dopamine, serotonin and GABA levels in striatum, hippocampus and olfactory bulb of aged female KO mice

Brain tissue samples from 15-months-old male and female mice were collected 1 week after the completion of behavioral tests (light/dark, colon motility and forced swim). Right striatum, hippocampus and olfactory bulb were analyzed for content of monoamines NA, DA, 5-HT and their metabolites (DOPAC, HVA, 5-HIAA) as well as several amino acids (among others, glutamate, glutamine and GABA). Genotype and gender differences were found in the levels of some of these substances in each brain region analyzed (Fig. 6).

Figure 6.

Brain monoamine, metabolite and amino acid levels. Amounts of monoamines (ng/mg of tissue), metabolites over monoamines ratio and amounts of amino acids (nmol/mg of tissue) in brains of aged male and female mice: (a–c) striatum (*P < 0.05 vs. males KO, **P < 0.005, ***P < 0.0005 males vs. females, unpaired t-tests); (d–f) hippocampus (**P < 0.005, males vs. females, ##P < 0.005 vs. KO males, unpaired t-tests); and (g–i) olfactory bulb (g, *P < 0.05 vs. KO females, **P < 0.005 vs. WT females, ***P < 0.0005 vs. WT males, #P < 0.05 vs. WT females; h,i, *P < 0.05 males vs. females, **P < 0.005 vs. KO males, #P < 0.05 vs. WT females, unpaired t-tests). Males: WT, n = 10; KO, n = 12. Females: WT, n = 7; KO, n = 12.

In the striatum, although no significant differences were detected between genotypes, levels of dopamine were found higher (16.9%) in females compared to male KO mice. Lower levels of DA metabolites were found in female mice compared to males, regardless of genotype; female mice also showed significantly higher levels of serotonin and consequently lower metabolite levels indicating a higher turnover (Fig. 6a,b). As for amino acid levels, GABA was also found increased in the striatum of female mice (Fig. 6c).

In the hippocampus, no statistically significant differences between genotypes or genders have been detected in monoamines or their metabolites (Fig. 6d,e). Reduced levels of GABA have been observed in female compared to male mice (ANOVA, gender F(1,37) = 10.84, P < 0.005) and in particular in female KO mice compared to males (−33.2%; Fig. 6f).

In the olfactory bulb, significantly reduced levels of DA were found in female KO compared to female WT mice (−16%) and compared to males of the same genotype (−12.7%; ANOVA, genotype × gender F(1,37) = 6.40, P < 0.05). Levels of 5-HT also resulted significantly reduced in female KO mice (−7.3%) compared to female WT and among WT, males showed lower levels than females (−15%; Fig. 6g; ANOVA, gender F(1,37) = 13.05, P = 0.0009; genotype × gender F(1,37) = 6.80, P < 0.05). Reduction of 5-HT levels was also accompanied by an increase of 5-HT metabolite 5-HIAA ratio in female KO mice compared to males (Fig. 6h; ANOVA, gender F(1,37) = 4.19, P < 0.05; genotype × gender F(1,37) = 5.49, P < 0.05). Additionally, it was observed a reduction of glutamate levels in female compared to male mice (ANOVA, gender F(1,37) = 4.59, P < 0.05) and a 15.4% reduction of GABA levels in KO females compared to WT female mice (Fig. 6i).


This study revealed that aged female mice lacking the Parkinson's related GPR37 receptor showed increased anxiety and depression-like behaviors. Other domains as olfactory function, colonic motility, startle and PPI responses were however ‘improved’ or unaffected in comparison with the expected age-related decline observed in the WT and in some cases also in KO males.

Olfactory function

Odor detection and social odor discrimination was comparable in adult GPR37 WT and KO female mice while in non-social odor discrimination adult GPR37 KO female mice showed a mild deficit compared to WT. Interestingly, aged female KO mice seemed to detect and discriminate better than WT novel social and nonsocial olfactory stimuli (Fig. 1). Studies on non-motor symptoms in PD mouse models have found deficits in olfaction that were explained with a specific dopaminergic dysfunction (Fleming et al. 2008; Taylor et al. 2009; Tillerson et al. 2006). In this study, only aged female GPR37 KO mice showed a mild but statistically significant reduction of DA, 5-HT and GABA levels in the olfactory bulb accompanied by increased 5-HT but not DA turnover (Fig. 6). GPR37 is expressed in olfactory sensory neurons (Sammeta et al. 2007) and both 5-HT and DA are key players in olfactory dysfunction of PD and related disorders but their specific regulation is not completely understood (Doty 2012). If the mild neurochemical changes in the olfactory bulb here reported could explain the slightly ‘improved’ olfactory function in aged GPR37 KO female mice is a hypothesis that cannot be fully supported and thus it requires further investigation.

Colonic motility

Adult GPR37 KO females showed a reduced percentage of fecal solid matter compared to age-matched WT littermates and to aged GPR37 KO females (Fig. 2b). This may indicate poorer intestinal water absorption and possibly shorter transit time thereby suggesting potential gastrointestinal dysfunction due to impaired contractility in adult but not in aged KO females.

PD patients with gastrointestinal dysfunction mainly reported constipation (Pfeiffer 2003). Remarkably, this symptom was already described by James Parkinson in 1817 (Parkinson 2002) and Lewy bodies have also been found in the colonic myenteric plexus (Kupsky et al. 1987). Increased stool frequency was instead reported in several mouse models (Anderson et al. 2007; Taylor et al. 2009; Wang et al. 2008b). Also, DAT and D2-R KO mice display altered colonic motility explained by an inhibitory role of endogenous enteric dopamine (Li et al. 2006; Walker et al. 2000). Along with dopamine, it was reported that Gpr37 mRNA is highly expressed in the muscle-myenteric nerve layer (Ito et al. 2009) and interestingly, mice lacking endothelin-B receptor, the closest GPR37 homolog, display a megacolon phenotype reminding of Hirschsprung's disease (Hosoda et al. 1994). Even though we have not measured dopamine or DAT levels in the intestine of GPR37 KO mice, we hypothesize that the lack of this receptor affects DAT functionality in the colon as already observed in the brain (Marazziti et al. 2007).

Anxiety and depression

Aged GPR37 KO female mice showed more anxiety and depression-related behaviors than their WT littermates (Fig. 3). Anxiety-like behaviors and cognitive deficits have been reported in adult and aged parkin null mice, although gender of animals was not disclosed (Zhu et al. 2007). Few mouse models of PD (Mcdowell & Chesselet 2012; Taylor et al. 2010) have focused on anxiety and depression related tests. A VMAT2-deficiency model showed age-related anxiety-like and depressive phenotypes explained by monoamines storage reduction (Taylor et al. 2009), and young adult alpha-synuclein transgenic mice exhibited a decrease in anxiety-like behaviors (George et al. 2008).

Anxiety and mood disorders in women are strongly correlated with age and estrogen fluctuations (Douma et al. 2005), they are common non-motor symptoms of PD (Brown et al. 2011; Richard 2005) and are prevalent in PD women patients (Leentjens et al. 2011). The GPR37 gene is upregulated in substantia nigra of PD female patients (Cantuti-Castelvetri et al. 2007) and downregulated in amygdala and hippocampus of major depressive disorder patients, and in one rodent model of depression (Duric et al. 2010; Sibille et al. 2009; Surget et al. 2009).

The phenotypes observed in this study seemed specific to females, occurred only in older animals and could be explained by the expected decline of estrogen levels. However, preliminary results on the effect of acute administration of estradiol in aged female mice before performing the elevated plus-maze and forced swim test (Supporting Information) did not show differences in the responses of the mutants compared to WT. Also, circulating levels of serum estradiol were not different between 19-months-old females of the two genotypes (Table S4). Nonetheless, a role of GPR37 in modulating estrogen receptors activation in anxiety and depression contexts cannot be excluded and would require further research.

Emotional memory and sensorimotor processing

GPR37 KO female mice of both ages displayed a tendency to impaired context-dependent but no cue-dependent fear learning (Fig. 4). Adult GPR37 KO female and male mice showed significantly reduced acoustic startle but no significant genotype difference was found in PPI (Fig. 5). Differently from anxiety and depression, the acoustic startle phenotype did not demonstrate a gender effect. The reduced startle response of adult GPR37 KO male mice could be associated to the previously reported reduced levels of striatal DA and open field locomotion observed in these mutants (Marazziti et al. 2004). No such reduction in activity was however observed in adult female KO mice in which DA levels were not measured. However, striatal dopamine levels measured in aged females, are higher in KO than WT mice and interestingly, the opposite was observed for aged male mice. Notably, a tendency to higher locomotor activity and PPI has been observed only in aged female KO mice and not in adults (Figs S3 and 5c,d). Deficits in auditory startle as well as PPI have been observed in PD patients and linked to DA dysfunction (Kofler et al. 2001; Miller et al. 2009; Perriol et al. 2005).


We evaluated several non-motor functions focusing on female mice, often underrepresented in animal models (Zucker & Beery 2010). We compared male and female mice to see if specific non-motor phenotypes have a gender bias since this could unveil the mechanism of pathological processes. We demonstrated that the inactivation of Gpr37 produces motor and non-motor phenotypes relevant for PD. Namely, the absence of GPR37 appeared to produce deficits in olfactory function, colonic motility and acoustic startle only in adults and not in aged KO mice. On the other hand, the lack of GPR37 seemed to induce anxiety and depression phenotypes only in aged females.

Interestingly, GPR37 adult male KO mice were neuroprotected from MPTP and 6-OHDA toxins (Imai et al. 2007; Marazziti et al. 2004), GPR37 overexpression resulted in increased sensitivity to 6-OHDA (Imai et al. 2007) and, if combined with parkin deletion, lead to progressive catecholaminergic neuronal loss via unfolded protein stress (Kitao et al. 2007; Wang et al. 2008a). In humans, unfolded GPR37/PAEL-R polypeptide accumulates in brain samples of PD patients (Imai et al. 2001; Murakami et al. 2004) and recently GPR37 has been recognized as a candidate gene in the pathophysiology of PD (Chen et al. 2011).

Thus, far from being considered as a model of PD, the GPR37 null mutant strain could however be instrumental to elucidate the role of this receptor in the behavioral responses associated to DA and other neurotransmitter alterations that often lead to PD symptoms.

From the analysis of neurotransmitter levels it appeared that aged female KO mice showed a reduction of DA, 5-HT and GABA in the olfactory bulb, a decrease of GABA also in the hippocampus, but a significant increase of DA in the striatum (Fig. 6). This analysis revealed differences in neurotransmitter levels that paralleled some of the behavioral phenotypes observed in aged animals. In humans, DA-based treatments of PD non-motor symptoms are frequently ineffective, considering that some of those may have a non-dopaminergic correlate (Fox et al. 2008). In this respect, it will be necessary to further analyze the influence of age and gender when investigating the interactions of GPR37 with other players in the mechanisms of neurodegeneration/neuroprotection in the search for novel therapeutic targets.


We are very grateful to C. Gross (EMBL) for critical reading of the manuscript. We greatly thank R. H. Butler for providing the MouseWatch software; A. Giuliani for precious assistance in statistical analysis; S. Gobessi for invaluable help in estrogen level analysis and A. Gambadoro for help in ovariectomy; G. Di Franco, G. D'Erasmo, and A. Ventrera for excellent technical assistance; A. Ferrara and T. Cuccurullo for secretarial work. This study was supported by Italian Ministry of Research grants (FIRB-Idee Progettuali, 2005; CNR-Progetto di interesse strategico Invecchiamento, 2012–2014), and European Union Framework Programme 6 and 7 contracts (EUMODIC, Phenoscale). The authors declare no conflicts of interest.