Does chronic systemic injection of the DREADD agonists clozapine-N-oxide or compound 21 change behavior relevant to locomotion, exploration, anxiety, and depression in male non-DREADD-expressing mice?

Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) are chemogenetic tools commonly-used to manipulate brain activity. The most widely-used synthetic DREADD ligand, clozapine-N-oxide (CNO), is back-metabolized to clozapine which can itself activate endogenous receptors. Studies in non-DREADD-expressing rodents suggest CNO or a DREADD agonist that lacks active metabolites, such as Compound 21 (C21), change rodent behavior (e.g. decrease locomotion), but chronic injection of CNO does not change locomotion. However, it is unknown if chronic CNO changes behaviors relevant to locomotion, exploration, anxiety, and depression, or if chronic C21 changes any aspect of mouse behavior. Here non-DREADD-expressing mice received i.p. Vehicle (Veh), CNO, or C21 (1mg/kg) 5 days/week for 16 weeks and behaviors were assessed over time. Veh, CNO, and C21 mice had similar weight gain over the 16-week-experiment. During the 3rd injection week, CNO and C21 mice explored more than Veh mice in a novel context and had more open field center entries; however, groups were similar in other measures of locomotion and anxiety. During the 14th-16th injection weeks, Veh, CNO, and C21 mice had similar locomotion and anxiety-like behaviors. We interpret these data as showing chronic Veh, CNO, and C21 injections given to male non-DREADD-expressing mice largely lack behavioral effects. These data may be helpful for behavioral neuroscientists when study design requires repeated injection of these DREADD agonists. Highlights Acute injection of CNO changes behavior of non-DREADD-expressing mice It’s not known if chronic CNO or alternative agonist C21 also changes mouse behavior DREADD agonists or Veh were given chronically to non-DREADD-expressing mice CNO and C21 don’t change locomotion and have a mixed effect on anxiety-like behavior 1 mg/kg CNO and C21 can be injected repeatedly without non-specific behavior effects


Introduction
The preclinical use of chemogenetics, such as designer receptors exclusively activated by designer drugs (DREADDs), has enhanced manipulation of the brain activity in awake and behaving rodents [1][2][3]. The prototypical DREADD activator, Clozapine-N-oxide (CNO), was initially thought to be biologically-inactive [3]. However, CNO is now presumed to cause off-target endogenous receptor activation due to its back-metabolism to clozapine [4][5][6][7]. Clozapine is a clinical antipsychotic [8] and its acute administration to non-DREADD-expressing rodents decreases locomotion in a dose-and time-dependent manner [9][10][11][12] and can be either anxiolytic- [11][12][13] or anxiogenic-like [10,14] depending on dose. Thus, it is important to understand if the behavioral changes documented in studies that have used CNO are due to specific activation of DREADDs or to off-target effects via the non-DREADD-specific actions of clozapine. To avoid potential off-target effects, other designer drugs have been developed to interact with DREADDs [15,16], such as Compound 21 (C21) which has no back-metabolism to clozapine [16]. Notably, it is unknown if chronic administration of CNO or C21 to non-DREADD-expressing rodents changes behavior relevant to locomotion, exploration, anxiety, or depression. This is an important knowledge gap, as many studies administer DREADD agonists repeatedly [17,18]. Identifying the behavioral effect -or lack thereof -of chronic CNO or C21 in non-DREADD-expressing rodents would enable researchers to best adhere to the principles of the 3R's (replacement, reduction, refinement) [19].
Some data support an off-target behavioral effect of acute CNO [5,7]; for example, non-DREADDexpressing mice given acute CNO (1mg/kg) locomote less than control mice when examined 2-3h post-injection, the predicted time point when back-metabolized clozapine concentration is highest [7].
However, other data do not support a behavioral or physiological effect of back-metabolized clozapine [6,21,22]; for example, in non-DREADD expressing animals, CNO (<5mg/kg) does not substitute for clozapine. To understand this apparent discrepancy -1mg/kg acute CNO decreases 4 locomotion, but <5mg/kg CNO is not discriminated -it is important to directly assess the influence of CNO on fundamental behaviors (e.g. locomotion) in non-DREAD-expressing rodents, and to examine these behaviors at the interval post-CNO when back-metabolized clozapine is thought to be highest.
Also, since clozapine's effect in both humans and mice is greater after chronic vs. acute administration [8,23], it is important to assess the behavioral effect of chronic CNO injections in non-DREADD-expressing rodents.
Here we assessed the behavioral effect of giving male mice chronic injections of Veh or the DREADD agonists CNO and C21. In line with best practices [5,6], non-DREADD-expressing mice were given Veh or a DREADD agonist at an experimentally-relevant dose (1mg/kg) [4,16] to discern off-target effects unrelated to DREADD activation. We hypothesized that relative to chronic Veh, chronic (like acute [6,7]) CNO would result in back-metabolized clozapine and thus decrease locomotion and exploration, while chronic C21 injections would result in similar performance in tasks relevant to locomotion, exploration, anxiety, and depression.

Ethics
Experiments were approved by the Institutional Animal Care and Use Committee at the Children's Hospital of Philadelphia (CHOP) and performed in compliance with NIH's Guide for the Care and Use of Laboratory Animals.

Statistics
Main effects and interactions were considered significant at P<0.05, and Bonferroni post-hoc tests were then performed. Results and Supplementary Table1 provide effect sizes (omega-squared, ω 2 ; partial omega-squared, ωp 2 ) and P-values to four significant digits. All data are presented in the Results, but conclusions are not stated when subject numbers fell below the predetermined threshold (Supplementary materials).

Body weight, Weeks 0-16.
6 Veh, CNO, and C21 groups gained a similar amount of weight at a similar rate ( Fig. 1B

Tests relevant to anxiety and depression, Weeks 15-16.
In marble burying (Supplementary Fig. 2A), Veh, CNO, and C21 mice buried a similar percentage of marbles (F2, 20=0.09887, P=0.906, ωp 2 =-0.09), indicating similar behavior relevant to anxiety and/or repetitive action. In the sucrose splash test (Supplementary Fig. 2B-C), data suggest Veh, CNO, 8 and C21 mice had similar measures on latency to grooming and grooming events; however, the Veh subject number was too low to state a conclusion.

Discussion
Here we examined the behavioral effects of chronic CNO or C21 injections in non-DREADDexpressing mice. We administered 1mg/kg based on the ability of this dose to change behavior in DREADD-expressing rodents when injected acutely [16,27] or chronically [18,28]. We assayed most behaviors 3h post-injection, when CNO-to-clozapine back-metabolism peaks [4,7], and tested locomotor activity 0-3h post-injection. For the non-DREADD-expressing mouse line, we selected CamKIIa-icre mice given the use of this forebrain glutamatergic neuron cre-expressing line in DREADD studies by us and others [1,18,24,29]. Data from the 3rd injection week show these non-DREADD-expressing mice given chronic CNO or C21 explored a novel context more (but had similar locomotion) relative to mice given Veh, and showed mixed responses in anxiety tests. In all other tests during the 16-week experiment, Veh, CNO, and C21 mice had similar behavior; the effect of chronic CNO or C21 on depressive-like behavior (splash test) and acute CNO on locomotion was inconclusive due to loss of subjects. Overall, our data suggest that when appropriate control groups are used, non-DREADD-expressing mice can be injected repeatedly with either DREADD agonist without causing gross behavioral effects.
Our largely-negative data are consistent with reports that chronic 1mg/kg CNO given to non-DREADD-expressing mice does not change behavior/physiology [28,30], but our data importantly show no effect on locomotion and a mixed effect on anxiety-like behavior. While the literature is unanimous that chronic 1mg/kg CNO does not grossly change mouse behavior in non-DREADDexpressing mice, the literature is mixed on the effects of acute CNO given to non-DREADDexpressing rodents. Several studies report acute CNO or C21 does not change behavior in non-DREADD-expressing rodents (1mg/kg [our data, 22] or 3 mg/kg [4,21] CNO, or 3.5mg/kg C21 [4]). Of 9 studies that do report a change in behavior in non-DREADD-expressing rodents after acute CNO, two saw behavioral effects in rats with acute 1mg/kg CNO [7,26] and two in mice and rats with higher CNO doses than the present work [4,6]. Further work is needed to understand what factors (mouse strain, behavior parameters, etc.) underlie these divergent results with acute CNO.
CNO and C21 are both rapidly-metabolized in mice [4], but brain C21 levels last slightly longer than brain CNO levels [4]. Thus, C21 may be a better designer drug for studies requiring longer duration of DREADD activation. While C21 binds to a range of G-protein coupled receptors (including dopamine D1 and D2 and histamine H4 receptors) at doses >3mg/kg, below this dose it appears to be a reliable activator since unlike CNO it has no detectable conversion to either clozapine or CNO [4,16].

Conclusion
While Non-DREADD-expressing mice given CNO or C21 (1mg/kg i.p) for 5 days/week for 3-16 weeks perform indistinguishably from Veh mice in tests relevant to locomotion, mice given CNO or C21 have increased exploration during the 3rd injection week; results on anxiety measures were mixed. These data suggest with appropriate dose and control groups, CNO and C21 can be used as DREADD agonists for studies that require long-term, repeated injection of these compounds without concern for gross non-specific behavioral or physiological effects.

Declaration of Competing Interests:
Authors have no conflicts of interest to declare.          week for the study remainder. After this move, one C21 mouse had unexplained weight loss and was therefore removed from the study.

S1.4 Overview of behavior testing for mice in the chronic injection study
Testing in mice given chronic injection of Veh, CNO, or C21 proceeded as noted in Fig. 1A. In the 3rd injection week, mice underwent activity monitoring in novel context 1 (2 consecutive days), open field [1], and elevated plus maze [1]. In the 14th injection week, mice underwent activity monitoring in novel context 2. In the 15th and 16th injection weeks, mice underwent marble burying [2] and sucrose splash test [1], respectively. Initial study design included the forced-swim test, but as three Veh mice died during initial forced-swim testing, no other mice were run on this test. Prior to each test, mice were habituated to the testing room for ~1h under red light (30-50 lux).

S1.4.1 Activity in Novel Context 1 (3rd injection week)
Activity in a novel context (termed novel context 1) was tested 30min/day for two consecutive days Locomotion was defined as total or partial movement detected by chamber activity beams (movement between wall and reward hopper).

S1.4.2 Open Field (3rd injection week)
Three hours after injection, mice underwent open field testing [1] during the 3rd injection week. The test was performed at 35-50 lux. Mice were placed in a square (L42xW42xH42cm) opaque-white Plexiglas chamber (Nationwide Plastics). A single mouse was allowed 5-min of free movement in the chamber. Total movement distance, center zone (L14xW14cm) entries and duration, and peripheral zone (L5xW5cm) entries and duration were scored via EthoVisionXT software (Noldus Information Technology) using nose-center-tail tracking.

S1.4.3 Elevated Plus Maze (3rd injection week)
Three hours after injection, mice were placed in the elevated plus maze apparatus

S1.4.4 Activity in Novel Context 2 (14th injection week)
Square chambers (L27xW27xH20cm, Med Associates Inc., #ENV-510), each placed inside an opaque sound-attenuating chamber, were used to measure activity via infrared beam breaks over 240min.The test was performed at lux 0. After 60min, mice were removed from the square chambers, injected (Veh, CNO, or C21), and immediately placed back into the chamber for the final 180min. 5 Measures of total ambulatory distance, ambulatory duration, number of movement events, and total number of jumps were scored with Activity Monitoring 5 software (Med Associates Inc., #SOF-811).
Each measure was analyzed separately.

S1.4.5 Marble Burying (15th injection week)
Three hours after injection, marble burying [2]  were placed on the bedding (4x5rows). Mice were placed in the cage for 20min. The number of marbles buried (covered ⅔ or more in bedding) by the end of the test was scored by two independent observers and averaged.

S1.4.6 Sucrose Splash Test (16th injection week)
The day before the sucrose splash test, mice were singly-housed in fresh microisolator cages. The next day mice were given their respective injection, and the nestlet, food hopper, and red plastic hut were removed from the home cage right injection. For the test itself (performed at 35-50 lux), freshlymade 10% sucrose in water was sprayed onto the mouse's back [1]. Video recordings (5min, Panasonic, HC-V270) were manually scored by an observer unaware of treatment group. Measures reported are total grooming duration, latency to groom, and total grooming events.

S1.5 Activity in Home Cage after Acute Veh or CNO
Activity monitoring in a home cage (Supplementary Fig. 1) was measured as previously described [3]. In brief, mice received a single i.p. injection of Veh, 0.3mg/kg CNO, or 1 mg/kg CNO and were immediately and individually placed in a fresh home cage (same dimensions as housing cage, but 6 fresh bedding, food, and water) between photocells. The test was run under dim light (30-50 lux; 5pm-6:15pm, 6:15pm-9:00am) and red light (30 lux; 6:15pm-6:15am). A computer-controlled photobeam activity system (San Diego Instruments) recorded total movement of mice in the XY plane, with photocell beam breaks recorded with 15min-bins overnight (16h, from 17:00 to 09:00).

S1.6 Rigor, statistical analyses, and effect size determination
Experimenters were blinded to treatment until behavior was complete. Data are reported as mean±SEM and were analyzed as noted in Supplementary two-way ANOVA was performed for activity beam breaks over time. Due to varied group size, body weight and activity were analyzed via mixed measure (within-subject repeated measure, betweensubject non-repeated). Bonferroni post-hoc tests were performed when P-values<0.05 for the main effect of Treatment or when there was an interaction between two independent variables. In this work, the word "threshold" is applied to a main effect when 0.05<P<0.1, and confidence intervals (CI, 95%) for one-way ANOVAs are provided in Supplementary Table 1. For estimation statistics, omegasquared (ω 2 ) was used for one-way ANOVA and partial omega-squared (ωp 2 ) was used for RM twoway ANOVA. In line with the literature [4][5][6][7], ω 2 and ωp 2 effect sizes were considered as following: ≤0, similar to 0; 0.1, small; 0.06, medium; and 0.13, large. All effect sizes (ω 2 and ωp 2 ) are provided in the Results and Supplementary Table 1 regardless of NHST P value, and the classification of ω 2 and ωp 2 effect sizes are only provided in the Results for "large" effect sizes. All behavioral data collected 7 were analyzed and are presented in the Results. However, conclusions are not stated for the splash test or acute injection activity as these subject numbers fell below the predetermined threshold.