A. L. Gundlach, Howard Florey Institute, Florey Neuroscience Institutes, The University of Melbourne, Victoria 3010, Australia. E-mail: firstname.lastname@example.org
Characterizing the neurocircuits and neurotransmitters that underlie arousal and circadian sleep/wake patterns is an important goal of neuroscience research, with potential implications for understanding human mental illnesses, such as major depression. Recent anatomical and functional studies suggest that relaxin-3 neurons and their ascending projections contribute to these functions via actions on key cortical, limbic and hypothalamic circuits. This study reports the behavioral phenotype of C57BL/6J backcrossed relaxin-3 knockout (KO) mice. Cohorts of adult, male and female relaxin-3 KO and wild-type (WT) littermate mice were subjected to a battery of behavioral tests to assess sensorimotor function and complex behavior. No overt deficits were detected in motor-coordination, spatial memory, sensorimotor gating, anxiety-like behavior or locomotor behavior in novel environments; and no marked genotype differences were observed in response to a chronic stress protocol. Notably however, compared to WT mice, relaxin-3 KO mice displayed robust hypoactivity during the dark/active phase when provided with free home-cage access to voluntary running wheels. This circadian hypoactivity was reflected by reduced time spent and distance traveled on running wheels, coupled with an increase in the time spent immobile, possibly reflecting increased sleeping. Overall, these studies support a role for relaxin-3 signaling in the control of arousal and sleep/wakefulness, and identify the relaxin-3 KO mouse as a useful model to study this role further.
In the mammalian brain, control over sleep and wake states is primarily mediated by an array of neurocircuits and neurotransmitters which are referred to as ‘arousal’ pathways (Jones 2005; Saper et al. 2005, 2010). To date, numerous arousal pathways/networks have been identified including populations of neurons expressing 5-HT (Kocsis et al. 2006), noradrenaline (Carter et al. 2010), orexin (Sakurai 2007), histamine (Takahashi et al. 2006), dopamine (Lu et al. 2006) and acetylcholine (Steriade 2004), located within the raphe, locus coeruleus, lateral hypothalamus, tuberomammillary nucleus, periaqueductal grey, and septum/tegmental system, respectively (Jones 2003, 2005). In addition to basic circadian sleep/wake states, these arousal systems regulate a range of processes required for an alert and active animal, such as exploratory behavior and spatial navigation, motivation and reward, the stress response, feeding and metabolism, and memory.
In line with these possible roles, studies using 129S5:B6 mixed-background relaxin-3 knockout (KO) mice showed two putative phenotypes. In one study, relaxin-3 KO mice displayed a ‘metabolic’ phenotype, whereby they were largely resistant to the obesity observed in wild-type (WT) mice maintained on a moderately high-fat (6%) diet (Sutton et al. 2009). In a concurrent study using independently inbred 129S5:B6 mice, relaxin-3 KO mice displayed hypoactivity in a range of acute tests such as the automated locomotor cell (Smith et al. 2009). Notably though, the metabolic and hypoactive phenotype present in each mixed-background cohort was not reproducible in the other – a problem often encountered using independently inbred mixed-background colonies, as although both colonies contained equal amounts of genetic material from C57BL/6J and 129SV mouse strains, genetic differences remained due to the random fixation of strain-specific alleles at any given locus (Wolfer and Lipp 2000).
Therefore, in this study, we further explored the functional role of relaxin-3 by observing the behavioral consequences of life-long relaxin-3 deficiency, and present the first reports of the behavioral phenotype of backcrossed C57BL/6J relaxin-3 KO mice. Compared to WT controls, relaxin-3 KO mice displayed similar health and well-being, motor-coordination and general performance in a range of acute behavioral paradigms. However, when provided with home-cages fitted with voluntary running wheels, relaxin-3 KO mice ran significantly less distance during the dark/active phase, and spent more time ‘inactive’ (possibly reflecting sleeping). These findings are consistent with the idea that relaxin-3 acts to promote arousal, and suggest relaxin-3 KO mice represent an appropriate model to further explore the roles of relaxin-3 within the brain. Such studies are important, as disrupted arousal pathways have been implicated in a range of human neurological and psychiatric illnesses, including narcolepsy (Lin et al. 1999) and major depression (Werner & Covenas 2010).
Materials and methods
All procedures described were undertaken with approval from the Howard Florey Institute Animal Welfare Committee, in strict compliance with the ethical guidelines issued by the National Health and Medical Research Council of Australia, which are in accordance with the Guidelines laid down by the National Institute of Health (NIH), USA; and all efforts were made to minimize the number of animals used and any discomfort they experienced. Relaxin-3 KO/LacZ knock-in mice (Lexicon Genetics Inc., The Woodlands, TX, USA) were originally obtained on a mixed-background (129S5:B6), before being subjected to successive rounds of 10 generations of backcrossing onto a C57BL/6J background. Four separate cohorts of age-matched male and female WT and relaxin-3 KO littermate mice were produced from heterozygous pairings and genotypes were identified from tail samples. Cohort 1 (n = 18 male WT; 10 male KO; 13 female WT; 11 female KO) and Cohort 2 (n = 13 male WT; 9 male KO; 13 female WT; 16 female KO) were subjected to parallel acute test batteries. The stress regimen was conduced on seven to eight mice per group from each cohort (see below for details). Cohorts 3 and 4 (each consisting of n = 8 male WT; 8 male KO; 7 female WT; 7 female KO) were used in running wheel experiments. Mice were group housed (mixed genotypes) while acute tests were conducted, and single-housed for running wheel and chronic stress studies; and were maintained on a 12-hour light–dark cycle (lights on 0700–1900 h) with ad libitum access to standard chow and water.
Acute behavioral tests
Mice were tested in a battery of acute paradigms during the light phase, with a minimum of 2-day recovery between tests. Mice were routinely habituated to testing rooms 1–3 h before testing. For both cohorts subjected to a parallel battery of acute paradigms, tests were conducted when mice were between 8 and 14 weeks of age in the following order: accelerating rotarod, automated locomotor cell, novel object exploration, Y-maze, elevated plus maze, light/dark box, large open-field, and prepulse inhibition. At the conclusion of this test battery, a smaller group of mice (seven to eight per group) from each cohort was subjected to the 8-week stress regimen.
Mice were subjected to 3 × 2 min learning trials on an accelerating rotarod (Acceler rotarod 7650, Ugo Basile, Comerio, Italy) the day before testing, and a single 2 min training trial 9 min before two test trials 9 min apart, each lasting 5 min, during which time the rotarod speed peaked at 40 r.p.m. after 5 min, and the latency to fall was recorded (Carter et al. 1999).
Automated locomotor cell
Mice were placed in a 27 × 27 cm automated locomotor cell (Tru Scan Photobeam Arena, E63-10; Coulbourn Instruments, Allentown, PA, USA) under dim lighting for 1 h, and the floor plane distance traveled was acquired using Tru Scan 2.0 software (Coulbourn Instruments) (Kondo et al. 2008).
Novel object exploration
Mice were placed in a 38 × 38 cm walled arena for 15 min. After 5 min, an object (white lead cylinder 3 cm in diameter and 10 cm long) was placed in the center of one quadrant of the arena, and at 10 min, a second object (small glass conical flask filled with white weights) was placed in the center of the opposing corner. Mice were filmed from above and their nose-point tracked using EthoVision XT software (Noldus Information Technology, Wageningen, The Netherlands). Mice were defined as exploring an object when their nose point was within 1 cm of the object. [Note: This test was designed to measure time mice spent exploring novel objects when placed in an arena. In this regard, it resembles the frequently used object recognition test (Dere et al. 2007), but it is modified to exclude a ‘memory component' (i.e. there are no separate ‘learning’ and ‘retrieval’ trials) so a continuous time-course of object exploration can be measured without possible confounds of an inter-trial break].
A Y-maze (each arm 10 cm wide and 30 cm long, with 17 cm high walls) illuminated by low light was surrounded by visual cues 40 cm above and outside each arm. During the learning trial, one arm was blocked with a wall insert (designated the ‘novel arm’). Mice were placed in the ‘home arm’ facing the end, and were allowed to explore the apparatus including the other available ‘familiar arm’. Mice were then returned to their home-cage for an inter-trial interval of either 40 or 60 min (see Results), before being returned to the Y-maze for a 5-min retrieval trial with all arms available for exploration. Mice were filmed from above, and the distance traveled and time spent in each arm was recorded using EthoVision XT software (Sarnyai et al. 2000).
Mice were placed in an acoustic startle apparatus (SR-LAB Startle Response System, San Diego Instruments, San Diego, CA, USA) with constant 70 dB background noise. After 5-min acclimatization, 70 individual stimuli were delivered approximately every 15 seconds over the following 20 min, which consisted of P115 stimuli (a 115 dB pulse that lasted for 50 milliseconds with no prepulse), PP4 stimuli (the same as a P115, except preceded by a 74 dB prepulse lasting for 20 milliseconds that finished 100 milliseconds before the P115), PP8 stimuli (the same as a PP4 except prepulse intensity was 78 dB) and PP16 stimuli (the same as a PP4 and PP8, except prepulse intensity was 86 dB). The startle magnitude elicited over 150 milliseconds after the beginning of the 115 dB component of each stimulus was recorded. The first 10 and last 10 stimuli delivered were P115s, while the 50 stimuli in between were made up of 20 P115s, 10 PP4s, 10 PP8s and 10 PP16s, delivered in random order. Percentage prepulse inhibition (%PPI) was calculated by comparing PP4, PP8 and PP16 startle magnitudes to the average of the middle 20 P115 stimuli, which was considered the maximum startle response elicited without a prepulse (Paylor & Crawley 1997).
Elevated plus maze
Mice were placed on the apparatus (each arm 6 cm wide, 30 cm long, elevated 40 cm off the floor, with two opposing ‘closed’ arms with 14.5 cm high walls, and two opposing ‘open’ arms), under dim lighting, and were filmed from above for 10 min. Time spent in and number of entries into the open arms were tracked using EthoVision XT software (McPherson et al. 2010).
A 1 m circular arena was strongly lit (1000 lux in the center and 900 lux at the perimeter), and mice were placed in the ‘center’ region (the circular, 50 cm diameter area), and allowed to explore the arena for 10 min. Mice were filmed from above, and the distance traveled, time spent in, and the number of entries into the center region were tracked using EthoVision XT software.
Automated locomotor cells (described above) were fitted with a box (13.5 × 27 × 37 cm high) in one side of the cell, which was made of a plastic that is impermeable to visible light, but not the locomotor cell photobeams, while the other, ‘light side’ was lit by an array of light emitting diodes (750 lux in the center of the light side, 700 lux in the corners). Time spent in and number of entries into the light side during the 10-min trial were recorded by Tru Scan 2.0 software (McPherson et al. 2010).
Mice were single-housed [considered a mild chronic stressor (Voikar et al. 2005)] for the 8-week duration of the stress study and subjected to a 15-min daily restraint stress for the first seven consecutive days, whereby mice were placed in an 11-cm long, 2.8-cm diameter plastic tube with a hole at one end to allow air flow (Bale et al. 2000). Mice were additionally subjected to 5 min of forced swimming in a container (24 cm high, 17 cm diameter) filled 6 cm from the top with 30°C water, on days 1, 2, 7 and 21. Mice were filmed from above, and scored blindly for the time spent immobile [i.e. in the Porsolt posture (Porsolt et al. 1977), defined as immobility in all four limbs]. At the conclusion of the stress regimen, mice were re-tested in automated locomotor cells (week 7) and the light/dark box (week 8).
Running wheel activity and associated home-cage behavior
Standard home-cages were equipped with 37-cm circumference running wheels connected to reed switches, Activity Wheel Counters (Model 86061, Lafayette Instruments, Lafayette, IN, USA), and USB Computer Interfaces (model 86056A, Lafayette Instruments). Data were collected on a computer using Activity Wheel Monitoring System Software (V.11.04). After an initial 14-day acclimatization period, data were collected continuously in 1-h bins for 21 days and used to construct an ‘average day of wheel running’ for each mouse. These data for each mouse were then used to construct the graphical representations shown. For analysis of home-cage behavior during the dark phase, mice were filmed using night-vision cameras and scored for different behaviors observed at 5-min intervals by a blind observer. Although measurements of time spent ‘sleeping’ vs. ‘awake sitting’ cannot be accurately made from film scoring alone, signs of sleeping, such as curled body posture, slowed breathing and duration of prior immobility were used to determine ‘apparent sleeping’.
Graphical and statistical analysis
Statistical significance was determined using appropriate tests for each data set: Student t tests, using GraphPad Prism V5.0 software; (GraphPad Software Inc, San Diego, CA, USA) two-way analysis of variance (anova), using Sigmastat 3.5 software (Aspire Software International, Ashburn, VA, USA) and Bonferroni post-hoc analysis and three-way repeated measures anova, using Statistica ’98 Edition software (StatSoft Inc, Tulsa, OK, USA) and planned-comparison post-hoc analysis. For the Y-maze, elevated plus maze, large open-field and light/dark box, to determine whether the mean time spent in the novel or aversive area was significantly different than what would be predicted by chance alone (i.e. different from a hypothetical mean calculated from the proportional size that each of these areas constituted), the Wilcoxon signed rank test was employed using GraphPad Prism software. All graphical representations were generated using GraphPad Prism software. All data are presented as means ± SEM and statistical significance is defined as P < 0.05. Asterisk symbols are used to designate statistically significant main effects of genotype.
In multiple independent cohorts of C57BL/6J relaxin-3 KO mice, no marked differences were detected compared to WT littermates in health and well-being, temperament when handled, rates of pup survival, signs of ‘bullying’ (such as whisker barbering or wounds on the back), fecundancy or later-life mortality. In addition, in contrast to an initial report on mixed-background (129S5:B6) mice (Sutton et al. 2009), no genotype differences in body weight were observed, further indicating that relaxin-3 KO mice do not suffer from any gross deficits [three-way repeated measures anova of a representative cohort, weighed weekly from 8–20 weeks of age, main effect of genotype (F(1,12) = 0.055, P = 0.816), main effect of sex (F(1,12) = 102, P < 0.001), interaction (P = 0.83)].
C57BL/6J relaxin-3 KO mice displayed similar performance to WT littermates in a battery of acute behavioral paradigms
Two independent cohorts of adult (>8 weeks of age), male and female C57BL/6J relaxin-3 KO and WT littermate controls were subjected to a battery of acute standard behavioral paradigms during the light phase (0700–1900 h) (Crawley 1999) (Materials and methods for details).
Accelerating rotarod, automated locomotor cell and novel object exploration test
There were no genotype or sex differences in the latency to fall from an accelerating rotarod [two-way anova, main effect of genotype (F(1,100) = 0.16, P = 0.69); sex (F(1,100) < 0.01, P = 0.98); genotype × sex interaction (P = 0.95)], suggesting that relaxin-3 KO mice do not suffer from deficits in motor-coordination (Table 1). In line with this observation, there were no genotype or sex differences in the distance traveled in automated locomotor cells [main effect of genotype (F(1,100) < 0.01, P = 0.97); sex (F(1,100) = 0.67, P = 0.41); genotype × sex interaction (P = 0.81)]. Finally, in the novel object exploration test, there were no genotype or sex differences in the time spent exploring novel objects (main effect of genotype (F(1,95) = 1.86, P = 0.18); sex (F(1,95) = 1.73, P = 0.19); genotype × sex interaction (P = 0.43)], or in the number of interaction events [main effect of genotype (F(1,95) < 0.01, P = 0.93); sex (F(1,95) = 1.62, P = 0.21); genotype × sex interaction (P = 0.14); Table 1].
Table 1. C57BL/6J relaxin-3 KO mice displayed similar performance to WT littermate controls in a range of standard behavioral tests
Tests were conducted for the following durations: accelerating rotarod, 5 min; automated locomotor cell, 30 min; novel object exploration, 10 min; Y-maze, 5 min. Data are mean ± SEM.
(n = 31)
(n = 19)
(n = 27)
(n = 27)
Latency to fall, second trial (seconds)
266 ± 9
262 ± 14
267 ± 12
262 ± 13
Automated locomotor cell
(n = 31)
(n = 19)
(n = 27)
(n = 27)
Distance traveled (m)
62 ± 2
62 ± 2
60 ± 2
60 ± 2
Novel object exploration
(n = 29)
(n = 19)
(n = 25)
(n = 26)
Time exploring novel objects (seconds)
87 ± 3
97 ± 6
97 ± 6
99 ± 4
Number of interaction events
94 ± 4
100 ± 6
95 ± 6
87 ± 4
Y-maze (40-min trial interval)
(n = 18)
(n = 10)
(n = 13)
(n = 11)
% time in novel vs familiar arm
60 ± 2
60 ± 4
65 ± 2
63 ± 2
Y-maze (60-min trial interval)
(n = 13)
(n = 9)
(n = 13)
(n = 16)
% time in novel vs. familiar arm
66 ± 2
61 ± 3
67 ± 2
59 ± 3
The percentage of time spent in the novel vs. familiar arm of a Y-maze was calculated utilizing two different inter-trial intervals (40 and 60 min) for each separate cohort (Table 1). Three-way anova did not detect any significant effect of interval duration (F(1,95) = 0.50, P = 0.48) or sex (F(1,95) = 0.688, P = 0.42). A trend was observed, however, for relaxin-3 KO mice to spend less percentage time in the novel arm, although this effect did not reach statistical significance [main effect of genotype (F(1,95) = 3.60, P = 0.061); genotype × sex interaction (P = 0.48); genotype × interval duration interaction (P = 0.169)]. All groups spent a significantly higher proportion of their time in the novel arm vs. the familiar arm (see Methods), indicating intact spatial memory and neophilia in relaxin-3 KO mice.
Prepulse inhibition of the acoustic startle response
No genotype differences in startle magnitude were detected, but male mice displayed increased startle magnitudes, possibly because male mice are larger and are able to elicit a greater startle vibration in the detection apparatus [three-way repeated measures anova of four consecutive bins of ten 115 dB pulses, main effect of genotype (F(1,3) = 0.07, P = 0.79); sex (F(1,3) = 6.13, P = 0.017)]. Mice displayed habituation, or reduced startle magnitudes with repeated stimulation [main effect of bin (F(3,3) = 15.08, P < 0.001)], and there was no difference between genotypes in the rate of this habituation (genotype × bin interaction (P = 0.13)]. These data suggest relaxin-3 KO mice have similar hearing to WT controls.
No genotype or sex differences were detected in the %PPI achieved with prepulses of 4, 8 and 16 dB above the 70 dB background noise [main effect of genotype (F(1,2) = 0.14, P = 0.71); sex (F(1,2) = 1.19, P = 0.28); genotype × sex interaction (P = 0.618)]. Furthermore, all groups displayed significant increases in %PPI with increased prepulse intensity (main effect of prepulse magnitude (F(2,2) = 230, P < 0.001)], suggesting that relaxin-3 KO mice do not display overt deficits in sensorimotor gating (Fig. 1).
Tests of basal anxiety-like behavior
In three separate paradigms, no genotype differences were observed in basal anxiety-like behavior (i.e. measured in mice with no additional stressors imposed upon them prior to testing), and no consistent genotype differences were observed in motor activity (Table 2). In the elevated plus maze, no significant genotype differences were detected in either the time spent in the open arms [two-way anova, main effect of genotype (F(1,100) = 1.88, P = 0.17)], or in the number of entries into the open arms (F(1,100) < 0.01, P = 0.95). Nevertheless, these parameters were significantly higher in females compared to males [time spent in open arms, main effect of sex (F(1,100) = 4.88, P = 0.029); genotype × sex interaction (P = 0.71); entries into open arms, main effect of sex (F(1,100) = 7.10, P = 0.009); genotype × sex interactions (P = 0.19)]. In the large open-field, there were no genotype or sex differences in the time spent in the center region [main effect of genotype (F(1,100) < 0.001, P = 0.98); sex (F(1,100) = 2.97, P = 0.09); genotype × sex interactions (P = 0.52)], or in the number of entries into the center region [main effect of genotype (F(1,100) = 2.65, P = 0.11); sex (F(1,100) = 0.05, P = 0.82); genotype × sex interactions (P = 0.17)]. Finally, in the light/dark box, there were no genotype or sex differences in the time spent in the light side [main effect of genotype (F(1,100) = 0.66, P = 0.42); sex (F(1,100) = 2.47, P = 0.12); genotype × sex interaction (P = 0.62)], but female mice and WT mice displayed significantly more entries in the light side [main effect of genotype (F(1,100) = 6.63, P = 0.012); sex (F(1,100) = 5.33, P = 0.023); genotype × sex interaction (P = 0.36)].
Table 2. C57BL/6J relaxin-3 KO mice displayed similar basal anxiety-like behavior and locomotor activity to WT littermate controls in three standard behavioral tests
M WT (n = 31)
M KO (n = 19)
F WT (n = 27)
F KO (n = 27)
Tests were conducted for the following durations: elevated plus maze, 10 min; large open-field, 10 min; light/dark box, 10 min. Main effect of genotype *P < 0.05. Data are mean ± SEM.
C57BL/6J relaxin-3 KO mice displayed a similar response to acute and chronic stress as WT littermates
The performance of mice was tested during and after a stress regimen, in light of the demonstrated ability of neurogenic stress to upregulate relaxin-3 expression in rats (Banerjee et al. 2010; Tanaka et al. 2005). The regimen consisted of relatively intensive/frequent stressors during the first week (15 min daily restraint for 7 days, and 5 min forced swimming on days 1, 2 and 7); after which mice were subjected to mild stress for the remaining 8 weeks [single-housing (Voikar et al. 2005), and a forced swim on day 21]. The behavior of relaxin-3 KO and WT mice in various tests measured during this stress regimen are described below.
The mice examined in this study lost between 3 and 5% of their prestress body weight in the week following the onset of stress treatment, before returning to prestress body weight levels 2 weeks after stress treatment, and continuing to gain weight at later time points (main effect of time F(8,8) = 53.9, P < 0.001) (Fig. 2a). Loss in body weight is a characteristic reaction to stress in rodents (Willner et al. 1996). No significant differences were detected between genotypes or sexes [three-way repeated measures anova, main effect of genotype (F(1,8) = 0.22, P = 0.64); sex (F(1,8) = 0.15, P = 0.70); genotype × sex interaction (P = 0.99)]. Notably, these observations are in contrast to a previous report using mixed-background 129S5:B6 relaxin-3 KO mice, whereby at 8 weeks after the commencement of an identical stress treatment relaxin-3 KO mice lost ∼12.5% of their prestress body weight, compared to an ∼5% loss observed in WT littermate controls (Smith et al. 2009). This discrepancy may be due to strain differences or confounding features inherent in using mixed-background mice (see Discussion).
Porsolt swim test
Mice were subjected to a 5-min forced swim test on four separate days throughout the stress study (Fig. 2b). Although mice displayed increased time in the Porsolt posture over subsequent test days [three-way anova, main effect of time (F(3,3) = 38.7, P < 0.001)], no genotype or sex differences were observed [main effect of genotype (F(1,3) = 0.39, P = 0.53); sex (F(1,3) = 1.08, P = 0.31); genotype × sex interaction (P = 0.15)].
Poststress performance in the automated locomotor cell and light/dark box
Mice were tested 7–8 weeks after the commencement of the stress treatment, but no genotype or sex differences in locomotor performance or anxiety-like behavior were detected (Table 3). In automated locomotor cells, this was observed in measurements of distance traveled [main effect of genotype (F(1,59) = 0.56, P = 0.46); sex (F(1,59) = 1.53, P = 0.22); genotype × sex interactions (P = 0.68)]; while in the light/dark box, this was reflected in the time spent in the light side [main effect of genotype (F(1,59) = 0.79, P = 0.38); sex (F(1,59) = 0.42, P = 0.52); genotype × sex interaction (P = 0.97)], and in the number of entries into the light side [main effect of genotype (F(1,59) = 0.36, P = 0.55); sex (F(1,59) = 1.20, P = 0.28); genotype × sex interactions (P = 0.59)].
Table 3. C57BL/6J relaxin-3 KO mice displayed similar performance to WT littermate controls in the automated locomotor cell and the light/dark box, conducted 7–8 weeks after the commencement of a stress regimen
M WT (n = 16)
M KO (n = 16)
F WT (n = 15)
F KO (n = 16)
Tests were conducted for the following durations: automated locomotor cell, 30 min; light/dark box, 10 min. Data are mean ± SEM.
Automated locomotor cell
Distance traveled (m)
50 ± 3
47 ± 2
46 ± 4
45 ± 3
Time spent in light side (seconds)
190 ± 14
202 ± 17
199 ± 14
213 ± 13
Entries into light side
13 ± 1
13 ± 1
15 ± 2
14 ± 1
C57BL/6J Relaxin-3 KO mice displayed chronic hypoactivity compared to WT littermates during the dark phase
In contrast to the general lack of genotypic differences observed in acute behavioral tests, relaxin-3 KO mice displayed a robust and reproducible hypoactive phenotype when examined in home-cages fitted with voluntary running wheels for a total period of 5 weeks (data was collected in 1 h time bins over the last 3 weeks, and averaged to calculate an “average 24 h period).
Home-cage running wheel activity
Relaxin-3 KO mice ran significantly less distance during the total 24 h period compared to WT controls on home-cage voluntary running wheels [three-way repeated measures anova, main effect of genotype (F(1,23) = 7.25, P = 0.009); genotype × sex interactions (P = 0.77)]. In line with similar studies, female mice were observed to run more distance than male mice [main effect of sex (F(1,23) = 5.14, P = 0.027); Fig. 3a]. Planned-comparison post-tests showed that relaxin-3 KOs displayed significant hypoactivity at all hourly time points between 9 pm and 4 am (2100 and 0400 h), while females ran significantly greater distance than males between 12 am and 6 am (0000 and 0600 h). To more clearly illustrate the hypoactivity present in relaxin-3 KO mice, the percentage running wheel activity relative to WT mice was analyzed (where WT activity equals 0%) and showed that between 9 pm and 5 am (2100 and 0500 h), female KO mice displayed an approximate 20–30% reduction in running wheel activity, while male KOs displayed a 20–45% reduction compared to their WT littermate controls (Fig. 3b). An analysis of the total distance traveled during the dark phase also detected significant hypoactivity in relaxin-3 KO mice and increased distance traveled by female mice (two-way anova, main effect of genotype (F(1,56) = 7.65, P < 0.008); sex (F(1,56) = 5.92, P = 0.018); genotype × sex interactions (P = 0.74); Fig. 3c].
General dark phase home-cage behavior
Mice were filmed with night-vision cameras and subsequently scored for the frequency of behavior engaged in during the dark phase. For each behavior, two-way anova did not show a main effect of sex (data not shown), and hence data from males and females were pooled. Relaxin-3 KO mice were observed to spend less time on running wheels than WT littermate controls (Student t test, P = 0.006), and spent more time engaged in immobile inactivity and body posture consistent with sleeping (P = 0.015; see Methods; Fig. 3d).
This study represents the first report on the behavior of C57BL/6J backcrossed relaxin-3 KO mice, whereby cohorts of adult relaxin-3 KO and WT littermates were initially subjected to a comprehensive battery of standard behavioral tests (Crawley 1999). Importantly, relaxin-3 KO mice appeared healthy and did not display any overt ‘fundamental‘, physical or physiological deficits that could negatively impact upon subsequent behavioral testing, and hence we regard this strain of mouse as an appropriate model in which to investigate the effects of relaxin-3 deficiency on complex behaviors. An apparent absence of such profound deficits is in line with the putative role of relaxin-3 as a ‘neuromodulator’ in the brain, rather than a peptide hormone involved in critical stages of embryonic or postnatal development or a transmitter mediating critical primary neuronal functions.
Hypoactive phenotype of relaxin-3 KO mice suggests relaxin-3 promotes arousal
A major finding of these studies is that when single-housed in home-cages equipped with a voluntary running wheel, male and female C57BL/6J relaxin-3 KO mice were hypoactive during the dark/active phase compared to WT controls – running ∼30–40% less distance on running wheels during the majority of the dark phase and spending ∼40% more time in ‘inactive’ body postures that likely reflect sleeping – despite displaying normal motor-coordination. Furthermore, this hypoactive phenotype was unlikely to be due to differences in stress reactivity (as for example, increased stress levels can reduce activity levels), as relaxin-3 KO mice displayed similar performance to WT mice following a chronic stress regimen in paradigms designed to measure locomotor activity and anxiety-like behavior.
The phenotype observed in relaxin-3 KO mice suggests that in WT mice endogenous relaxin-3 acts to promote the motivated active behavior of voluntary wheel running, which it may do by increasing general arousal, motivational drive and/or wakefulness. A number of functional studies conducted in rats are consistent with these findings. For example, intracerebroventricular (ICV) infusion of an RXFP3 agonist has been reported to increase overall locomotor activity (Sutton et al. 2009). Considerable evidence also suggests that relaxin-3 is able to modulate hippocampal theta rhythm, which is a synchronous pattern of brain electrical activity associated with a range of functions such as rapid eye movement sleep, exploratory behavior and memory, and driven by the ‘septohippocampal system’ (Buzsaki 2005; Vertes & Kocsis 1997; Vertes et al. 2004). In this regard, local infusion of an RXFP3 agonist, R3/I5, into medial septum induced an increase in hippocampal theta rhythm in a sensory-poor context (the habituated home-cage), whereas infusion of an RXFP3 antagonist, R3(BΔ23-27)R/I5, into medial septum reduced hippocampal theta rhythm, when rats were introduced into a novel environment (Ma et al. 2009). Similarly, intra-septal RXFP3 antagonist disrupts performance in the spontaneous alternation task, a test of spatial working memory (Ma et al. 2009). Furthermore, stimulation of the nucleus incertus in anesthetized rats promotes theta rhythm, while electrolytic lesions of the nucleus incertus prevent the ability of reticularis pontis oralis stimulation to generate theta rhythm (Nunez et al. 2006; Teruel-Marti et al. 2008).
Further insights into the hypoactive phenotype of relaxin-3 KO mice can also be gained by examining the neuroanatomical distribution of relaxin-3/RXFP3 (Ma et al. 2007; Smith et al. 2010; Tanaka et al. 2005). High levels of relaxin-3 fibers and RXFP3 mRNA/binding sites are present within the intergeniculate leaflet and to a lesser extent within the suprachiasmatic nucleus, which receive photic and non-photic inputs and are involved in circadian rhythm (Harrington 1997; Morin & Allen 2006); and within the cortex, the activation of which is associated with wakefulness. Relaxin-3/RXFP3 is also present within the major nodes of the septohippocampal system, which largely controls the generation of hippocampal theta rhythm (Vertes & Kocsis 1997). Of particular note, the observed reduction in voluntary wheel running in relaxin-3 KO mice may be due to the deficits in reinforced behaviors, as relaxin-3 signaling likely impinges indirectly (directly) on the mesolimbic dopaminergic pathway, which is involved in motivation and reinforcement and undergoes plasticity during voluntary wheel running (Greenwood et al. 2011) and other rewarding activities. In this respect, the comparative behavior of relaxin-3 KO and WT mice in natural and drug reward paradigms such as alcohol or cocaine self-administration and saccharin preference, are of interest. Relaxin-3 projections and RXFP3 are also present within several identified arousal centers, including the raphe nuclei, lateral hypothalamus, tuberomammillary nucleus, periaqueductal grey and septum/tegmental system (see Introduction). Serotonin has been shown to regulate relaxin-3, as relaxin-3 positive cells within the nucleus incertus express the 5-HT1A receptor, and serotonin depletion resulted in an increased relaxin-3 expression in these neurons (Miyamoto et al. 2008). Indeed, the anatomical distribution of relaxin-3 fibers/RXFP3 is remarkably similar to the distribution of 5-HT/raphe neurons and receptors (Vertes et al. 1999) and parallels that of the noradrenaline/locus coeruleus arousal system (Carter et al. 2010), as relaxin-3 is primarily produced in the midline, tegmental nucleus incertus neurons, and their ascending projections innervate key cortical, septohippocampal and hypothalamic RXFP3-rich target regions. Therefore, existing evidence suggests the nucleus incertus/relaxin-3 system represents a newly identified ascending brainstem arousal pathway (cf.,Jones 2003, 2005).
Relaxin-3 KO mice displayed similar performance to WT controls in a range of acute tests
Despite the considerable anatomical and functional evidence that relaxin-3 is able to modulate memory and stress circuits, no differences in performance between relaxin-3 KO and WT mice were detected in the elevated plus maze, light/dark box, large open-field, forced swim test or Y-maze [although a near-significant (P = 0.061) main effect of genotype was observed here]; or in the response of relaxin-3 KO mice to a chronic stress regimen. The absence of such ‘predicted’ phenotypes may be due to compensation by similar systems (i.e. other arousal pathways), as ‘gross redundancy’ exists in arousal control (Jones 2003). Compensation by another ligand which activates RXFP3 seems unlikely however; as such a ligand is yet to be identified in brain. Although relaxin-3 KO mice displayed similar performance to WT littermates in the majority of tests conducted, it must be remembered that this study represents an initial test battery (Crawley 1999), and subsequent testing in more complex paradigms may uncover other phenotypes.
Despite robust dark phase home-cage hypoactivity, relaxin-3 KO mice displayed normal levels of locomotor activity in automated locomotor cells and other paradigms such as the large open-field, during the light phase (data not shown). This discrepancy may partly be due to reduced relaxin-3 tone during the light phase, as reported in rats (Banerjee et al. 2005), although low relaxin-3 tone is unlikely to fully account for this discrepancy, as expression of markers of neuronal activity such as c-Fos increase sharply in the nucleus incertus following handling of rats for behavioral testing during the light phase (Ma et al. 2009) and in a range of other behavioral situations (Goto et al. 2001; Ryan et al. 2011). Furthermore, brains processed from mice killed during the light phase display abundant relaxin-3 mRNA in nucleus incertus neurons, and relaxin-3 immunoreactivity in nucleus incertus soma and nerve fibers/terminations (Ma et al. 2007; Smith et al. 2010; Tanaka et al. 2005). A more likely reason for this discrepancy involves the acute vs. chronic nature of these tests. Handling mice and exposing them to novel environments such as automated locomotor cells increases general arousal and activates arousal pathways (Gompf et al. 2010), which may compensate for the lack of relaxin-3 signaling that is clearly detectable under ‘basal’ home-cage conditions. Although scoring of dark phase home-cage activity of mice detected statistically significant genotype differences in the frequency of locomotion and climbing, as these behaviors were relatively rare, their biological significance is unclear, and it is difficult to speculate about their relevance for relaxin-3 function.
Finally, although central infusion of exogenous relaxin-3 is robustly orexigenic in rats (Hida et al. 2006; McGowan et al. 2005, 2006, 2007), and recent associations have been proposed between relaxin-3/RXFP3 gene polymorphisms and metabolic disorders in human schizophrenic patients (Munro et al. 2011); no genotype differences in body weight were observed in C57BL/6J backcrossed relaxin-3 KO mice, which is perhaps not surprising considering the redundancy of systems controlling this fundamental homeostatic process. These findings are, however, in contrast to a preliminary study of mixed-background 129S5:B6 relaxin-3 KO mice that displayed reduced bodyweight and general metabolic resistance to obesity which was observed in WT mice fed on a moderately high-fat diet (Sutton et al. 2009). Furthermore, in an independently inbred 129SV:B6 mixed-background relaxin-3 strain, female relaxin-3 KO mice displayed robust and reproducible hypoactivity in a range of acute behavioral tests, including automated locomotor cells (Smith et al. 2009). While it is possible that these phenotypes were caused by relaxin-3 deficiency, and that the lack of such phenotypes in backcrossed C57BL/6J relaxin-3 KO mice may be because the penetrance of phenotypes is often strain-dependent (Doetschman 2009; Linder 2006; Wolfer & Lipp 2000); it is also possible that these phenotypes were ‘false positive’ observations due to confounding features inherent in using mixed-background mice, such as increased genetic variability and the effects of gene flanking regions (Gerlai 1996; Wolfer et al. 2002). In any case, the genetic variability present in mixed-background cohorts suggests that these phenotypes may not be reproducible, and for that reason, the use of backcrossed mice is critical.
This study represents the first behavioral profiling of a backcrossed C57BL/6J strain of relaxin-3 KO mice. These mice display normal health and well-being, with no gross deficits compared to WT littermate controls in modalities such as motor-coordination, spatial memory, sensorimotor gating, basal anxiety-like behavior, locomotor activity in novel environments and behavioral response to stress. Notably however, a robust hypoactivity phenotype was observed, whereby relaxin-3 KO mice displayed less running wheel activity and increased time spent in inactive postures suggestive of sleeping in their home-cage during the dark phase. These findings are in agreement with anatomical and functional predictions that relaxin-3 neurons form a newly identified ascending arousal system, which modulates arousal and sleep/wake states via interactions with cortical, limbic and hypothalamic circuits. Therefore, further study of this system using relaxin-3 KO mice and other experimental tools will be valuable, as dysregulation of arousal systems has been implicated in a range of human mental illnesses, and pharmacological manipulation of arousal systems has proven successful for the treatment of depression and other psychiatric diseases.
This research was supported by grants from the National Health and Medical Research Council (NHMRC) of Australia (AJL, ALG), a collaborative research agreement between the Howard Florey Institute and Johnson & Johnson Pharmaceutical Research & Development, LLC (ALG), and the Victorian Government through the Operational Infrastructure Programme. During the course of these studies, CMS was the recipient of an NHMRC (Australia) Dora Lush Postgraduate Scholarship, and is currently a Florey Trust Fellow. AJL and ALG are NHMRC (Australia) Research Fellows. The authors wish to thank Simon Miller for his help in constructing the running wheels. We also thank the Pratt Foundation, the Besen Family Foundation, the Percy Baxter Charitable Trust and the ANZ Trustees Medical Research and Technology in Victoria for their support. Research in the laboratory of ALG was supported by a collaborative research agreement with Johnson & Johnson Pharmaceutical Research & Development, LLC (USA). SWS is a paid employee of Johnson & Johnson Pharmaceutical Research & Development, LLC (USA).