Anxiolytic-Like Effects of Escitalopram, Citalopram, and R-Citalopram in Maternally Separated Mouse Pups

Selective serotonin reuptake inhibitors (SSRIs) were initially introduced as antidepressants, and their potential as anxiolytics has been observed in the treatment of social phobia, post-traumatic stress disorder, and generalized anxiety disorder (e.g., Nutt et al., 1999). Although each of the SSRIs increases extracellular serotonin (5-hydroxytryptamine, 5-HT) in the brain by blocking the 5-HT transporter SERT, they differ substantially in terms of their selectivity (Nutt et al., 1999; Owens et al., 2001; Sánchez et al., 2003). Actions at other binding sites, such as the muscarinic, histamine, adrenergic, and 5-HT₂ receptors, may contribute to some of the side effects commonly associated with SSRI treatment. Although in general, these side effects are better tolerated than those from tricyclics and benzodiazepines, they are a limitation in the use of SSRIs. Citalopram and its S-enantiomer, escitalopram (Lexapro or Cipralex), are highly selective inhibitors of the SERT with low or no binding to 144 other receptors (Owens et al., 2001; Sánchez et al., 2003). Compared with citalopram and its R-enantiomer, R-citalopram, escitalopram is at least 6-fold less potent in binding to the histamine 1 (H₁) receptor (Owens et al., 2001; Sánchez et al., 2003). Escitalopram and citalopram have similar affinities for the σ (σ₁) receptors, which are about 2-fold less than those of R-citalopram. Escitalopram is believed to confer the pharmacological effects of citalopram (Hyttel et al., 1992; Sánchez et al., 2003), and its effects can be inhibited by coadministration of R-citalopram (Mørk et al., 2003; Sánchez, 2003a). In clinical reports, escitalopram has a faster onset of antidepressant effects compared with its racemate citalopram and has a low incidence of side effects (Montgomery et al., 2001; Burke et al., 2002; Wade et al., 2002; Waugh and Goa, 2003).

Preclinical research on the anxiolytic-like effects of SSRIs is variable; some studies find no effect of SSRIs, whereas others find decreases or increases in anxiety-like behaviors. A viable hypothesis for these SSRI effects depends on whether the experimental procedure measures conditioned or unconditioned behaviors, particularly when the SSRI is given acutely (Griebel, 1995; Miczek et al., 1995). One procedure that has consistently detected an anxiolytic-like effect of acutely administered SSRIs relies on the vocal responses of young rodents when they are briefly separated from their nest and their dam. Separated rodent pups emit ultrasonic vocalizations (USVs) in a frequency range between 30 and 80 kHz that may serve as signals for the dam to initiate retrieval (Noirot, 1972; Brunelli et al., 1994). The species generality and unconditioned nature of vocal reactions distinguish them from many other preclinical measures of anxiety-like behaviors (Panksepp et al., 1980; Miczek et al., 1995; Sánchez, 2003b).

Among the neurotransmitter systems that influence separation calling, GABA and 5-HT are particularly noteworthy. Antianxiety treatments such as benzodiazepines and 5-HT_1A agonists decrease calling, whereas benzodiazepine inverse agonists and pentylenetetrazol can increase calling (Gardner, 1985; Insel et al., 1986; Mos and Olivier, 1989; Nastiti et al., 1991; Winslow and Insel, 1991; Molewijk et al., 1996; Vivian et al., 1997; Fish et al., 2000). Of the SSRIs, citalopram, fluoxetine, fluvoxamine, and paroxetine have been reported to reduce distress USVs (Mos and Olivier, 1989; Winslow and Insel, 1990; Olivier et al., 1998). Similarly, these SSRIs, as well as escitalopram, reduce the shock-induced USVs of adult rats (Sánchez and Meier, 1997; Schreiber et al., 1998; Sánchez, 2003b; Sánchez et al., 2003). Unlike the procedure in adult rats, the separation test can further be used to concurrently assess behavioral specificity, the degree to which drug effects on different behaviors can be dissociated. For example, GABA_A-positive modulators reduce calling, but their effects are often accompanied by sedation, stimulation, and/or increased motor incoordination (Vivian et al., 1997; Fish et al., 2000; Rowlett et al., 2001). 5-HT_1A receptor agonists reduce both motor activity and calling at a similar dose range, whereas 5-HT_1B receptor agonists reduce calling but stimulate motor activity (Fish et al., 2000). Although the doses of citalopram, fluvoxamine, and fluoxetine that reduce separation distress USVs have not been reported to affect motor activity, a high dose of fluoxetine (20 mg/kg) was reported to impair the negative geotaxis response in maternally separated rat pups (Mos and Olivier, 1989).

The objective of the current study was to compare the anxiolytic-like and locomotor effects of escitalopram to those of other SSRIs and the serotonergic/noradrenergic reuptake inhibitor, venlafaxine, in maternally separated 7-day-old mouse pups. Pups of this age were selected because they emit the most separation USVs (Noirot, 1972; Fish et al., 2000; Branchi et al., 2001). To further explore the mechanism through which escitalopram and citalopram reduce separation USVs, the interaction among escitalopram, R-citalopram, and the H₁ receptor antagonist pyrilamine was investigated.

Materials and Methods

Animals. Carworth Farms Webster (CFW) mice (n = 774, number per treatment included in Table 1 and Table 3), from litters of 8 to 12 pups, were 7 days old, weighed 3.5 to 5.0 g, and lived with both parents (Charles River Laboratories Inc., Wilmington, Ma) in clear polycarbonate cages (28 × 17 × 14 cm). Pine bedding covered the cage floor; food (Purina rodent chow; Purina, St. Louis, MO) and water were freely available through a wire lid. The mouse vivarium was 21 ± 1°C with 30 to 40% humidity, and was illuminated for 12 h (lights on at 7:30 AM). The “Guide for the Care and Use of Laboratory Animals” (http://www.nap.edu/readingroom/books/labrats/) was followed while caring for the mice and all procedures were approved by Tufts University's Institutional Care and Use Committee.

Apparatus and Measurements. All behaviors were measured in a sound-attenuated chamber (49.5 × 38 × 34 cm) in a separate procedure room. The chamber had a one-way vision window (19 × 16.5 cm) for observation, was lit by red light (10 W), and held a water bath that maintained the temperature of a square aluminum testing surface (23 × 23 cm) at 19.5 ± 0.5°C. Two-centimeter squares divided the testing surface into grids used to measure motor activity. Equipment similar to that previously described (Vivian et al., 1997; Fish et al., 2000; Rowlett et al., 2001) detected 30- to 80-kHz sounds. A high-frequency condenser microphone (Brüel and Kjær model 4135; Brüel and Kjær, Nærum, Denmark) joined to a preamplifier (Brüel and Kjær model 2633) was suspended 5 cm from the testing surface. Sounds were amplified (Brüel and Kjær model 2610), filtered (Krohn-Hite model 3362; Krohn-Hite Corporation, Brockton, MA) to produce a flat range between 30 and 80 kHz, and then visualized on an oscilloscope (Goldstar 059020A; Goldstar, Cerritos, CA) that was connected to an analog-to-digital converter GWI-AMP (GW Instruments; Somerville, MA). The signal was further amplified before it was connected to a computer (Macintosh II) that ran customized signal detection software. The software counted an ultrasound if it was 30 to 80 kHz, lasted longer than 10 ms, and was separated from the previous sound by at least 20 ms. Body temperature was measured using a lubricated thermo probe (o.d. 0.7 mm; YSI 555 N034; YSI Inc., Yellow Springs, IN) attached to a telethermometer (YSI 2100; YSI Inc.).

Procedure. The test sessions were conducted between 7:30 AM and 7:30 PM, and no differences in baseline USV rates were observed across different times of the day (Fish et al., 2000). An entire litter of pups and some bedding were removed from the home cage, transported to the procedure room, and maintained in an incubator (11.5 × 14 × 5 cm) at 34 ± 1.0°C. Twenty minutes later, the pups were weighed, individually placed in the testing chamber for 30 s to screen for the emission of USVs, and marked for identification. The pups that vocalized more than six times (ca. 80%) were injected with either one dose of the drug or vehicle. After the injection, the lubricated thermo-probe was inserted about 7 mm into the rectum and held in place until the temperature measurement stabilized (ca. 3 s). The pups were returned to the incubator for a specific interval (see “Drugs” section) and a second rectal temperature was taken immediately before a 4-min separation test. The signal detection software automatically counted USVs, while an experimenter, who was blind to the dose of the drug treatment, manually counted the number of grid crossings and rolls. A grid crossing was counted when half of the pup's body crossed into the next grid and a roll was counted when the pup's dorsal surface contacted the testing surface. After the test session, the pups were euthanized by CO₂ inhalation.

Drugs. Citalopram hydrobromide [(±)-(1-[3-(dimethylamino)propyl]-1-(4-fluorophenyl)-1,3-dihydroisobenzofuran-5-carbonitrilmonohydrobromide)], R-citalopram oxalate, escitalopram oxalate (i.e., S-citalopram), paroxetine hydrochloride ((–)-trans-5[4-p-fluorophenyl-3-piperidylmethoxy)-1,3-benzodioxole), fluoxetine hydrochloride ((±)N-methyl-γ-[4-(trifluoromethyl)phenoxy]-benzenepropanamine), and venlafaxine (1-2-[dimethylamino]-1-[4-methoxyphenyl]ethyl cyclohexanol) were provided by Forest Laboratories (Jersey City, NJ). Pyrilamine maleate (mepyramine) (N-[4-methoxy-phenyl]methyl-N′,N′-dimethyl-N-[2-pyridinyl]-1,2-ethanediamine) was purchased from Sigma-Aldrich (St. Louis, MO). All drugs were dissolved in 0.9% saline and injected subcutaneously in a volume of 0.1 ml/10 g body weight. In the interaction studies, drugs and/or saline were administered in two separate injections. To prevent leakage, a small amount of glue was applied to the injection site. The separation test was conducted 45 min after drug administration.

Statistics. USV and grid crossing data (transformed into percentage of vehicle), rolls, and body temperatures (raw values) were analyzed using one-way between-subjects analysis of variance. F-values with p < 0.05 were followed by post-hoc Dunnett's tests to determine which individual doses were significantly different from the vehicle treatment. A two-way between-subjects analysis of variance, followed by post-hoc Tukey's multiple comparison tests, analyzed the interaction between escitalopram and pyrilamine. ED₅₀ values (i.e., doses that reduced the total USVs to 50% of the vehicle mean) were estimated by first-order regression of those doses that were between ca. 20 and 80% of the vehicle mean. r² values for the linear regression ranged from 0.627 for fluoxetine to 0.996 for R-citalopram. Nonoverlapping 95% confidence intervals were considered statistically significant.

Results

Dose-Effect Studies

USVs. All of the treatments, escitalopram (F_(7,92) = 15.0; p < 0.001; Fig. 1; Table 1), citalopram (F_(5,71) = 14.5; p < 0.001; Fig. 1; Table 1), R-citalopram (F_(3,55) = 4.8; p = 0.005; Fig. 1; Table 1), paroxetine (F_(6,72) = 10.1; p < 0.001; Fig. 2; Table 1) fluoxetine (F_(4,86) = 8.9; p < 0.001; Fig. 2; Table 1), and venlafaxine (F_(4,58) = 10.2; p < 0.001; Fig. 2; Table 1) dose dependently reduced the number of separation USVs. Pups treated with the 0.3 to 0.56 mg/kg doses of escitalopram, the 1 to 10 mg/kg doses of citalopram, the 3 to 10 mg/kg doses of R-citalopram, the 0.1 to 3 mg/kg doses of paroxetine, the 3 to 30 mg/kg doses of fluoxetine, or the 3 to 56 mg/kg doses of venlafaxine vocalized significantly less than did the pups treated with vehicle. The histamine H₁ receptor antagonist pyrilamine (F_(5,42) = 4.8; p = 0.001; Table 1) also dose dependently reduced separation USV. Pups treated with the 3 to 30 mg/kg doses of pyrilamine vocalized significantly less than did the pups treated with vehicle (p < 0.05). The ED₅₀ values are shown in Table 4.

Grid Crossing. With the exception of citalopram, the treatments also dose dependently increased grid crossing: escitalopram (F_(7,92) = 3.1; p = 0.006; Table 2), R-citalopram (F_(3,55) = 3.7; p = 0.017; Table 2), paroxetine (F_(6,72) = 4.2; p = 0.001; Table 2), fluoxetine (F_(4,76) = 3.9; p = 0.006; Table 2), and venlafaxine (F_(4,58) = 2.6; p = 0.04; Table 2). Pups treated with the 0.3 and 0.56 mg/kg doses of escitalopram, the 10 mg/kg dose of R-citalopram, the 0.03 and 0.1 mg/kg doses of paroxetine, the 1 and 3 mg/kg doses of fluoxetine, or the 56 mg/kg dose of venlafaxine crossed significantly more grids than did the pups treated with vehicle. Pyrilamine (F_(5,42) = 3.8; p = 0.006; Table 2) also increased grid crossings, significantly at the 10 mg/kg dose.

Rolling. With the exception of escitalopram and venlafaxine, the treatments citalopram (F_(5,72) = 9.3; p < 0.001; Table 2), R-citalopram (F_(3,55) = 6.1; p = 0.001; Table 2), paroxetine (F_(6,72) = 2.6; p = 0.02; Table 2), and fluoxetine (F_(4,76) = 2.8; p = 0.03; Table 2) dose dependently increased rolling. Pups treated with the 1 to 10 mg/kg doses of citalopram, the 10 mg/kg dose of R-citalopram, the 3 mg/kg dose of paroxetine, or the 30 mg/kg dose of fluoxetine rolled significantly more than did the pups administered vehicle. Pyrilamine did not significantly affect rolling.

Body Temperature. None of the drugs, at the doses tested, significantly affected body temperature measured immediately before the separation test (data not shown).

Interaction Studies

USVs. Escitalopram's dose-dependent reduction of separation USVs was confirmed when it was coadministered with either vehicle or pyrilamine (1 or 10 mg/kg) (F_(3,127) = 12.0; p < 0.001; Table 3). Pups treated with the 0.1 and 0.3 mg/kg doses of escitalopram vocalized significantly less than did pups treated with vehicle alone. Pyrilamine did not alter the effects of escitalopram on separation USVs. R-citalopram altered the effects of escitalopram (F_(3,115) = 4.6; p = 0.005; Fig. 3; Table 3), preventing the reduction of USVs by the 0.1 and 0.3 mg/kg doses of escitalopram. In the presence of R-citalopram, higher doses (i.e., 1 and 3 mg/kg) of escitalopram were required to reduce USVs (F_(2,45) = 16.2; p < 0.001). R-Citalopram shifted escitalopram's ED₅₀ to the right from 0.05 (0.02, 0.09) to 0.61 (0.1, 1.7) mg/kg.

Grid Crossing. Escitalopram (F_(3,115) = 7.1; p < 0.001; Table 4) and pyrilamine (F_(2,127) = 3.4; p = 0.04; Table 3) each significantly increased grid crossings at the 0.1 and 0.3 mg/kg and 1 mg/kg doses, respectively. There was no interaction between the effects of escitalopram and those of pyrilamine or R-citalopram. When given with R-citalopram, escitalopram (F_(2,45) = 4.3; p = 0.02; Table 3) significantly increased grid crossing at the 1 mg/kg dose.

Rolling. Neither escitalopram nor pyrilamine significantly affected rolling. However, there was a trend for an interaction between escitalopram and pyrilamine (F_(6,127) = 1.9; p = 0.086; Table 3) but not R-citalopram. This was probably due to a decrease in rolling when the 0.1 mg/kg escitalopram dose was given with the 10 mg/kg pyrilamine dose. When given with R-citalopram, the higher doses (1 and 3 mg/kg) of escitalopram (F_(2,45) = 10.9; p < 0.001; Table 3) increased rolling.

Discussion

Several SSRIs and the mixed noradrenergic-serotonergic reuptake inhibitor venlafaxine reduced the maternal separation-induced USVs of mice and differed in terms of their potency and behavioral specificity. These results confirm earlier studies measuring USVs in neonatal rats after treatment with clinically effective anxiolytic drugs (Gardner, 1985; Mos and Olivier, 1989; Winslow and Insel, 1991; Olivier et al., 1998). The novelty of the present study is the comparison between the enantiomers of citalopram and the potent reduction of USVs by its S-enantiomer, escitalopram. The rightward shift in escitalopram's dose-effect curve by coadministration of R-citalopram suggests that R-citalopram inhibits some of the effects of escitalopram and that escitalopram is the active component of citalopram.

Escitalopram was more potent than citalopram and R-citalopram at reducing separation USVs, a result that is similar to those from behavioral and in vitro binding studies (Hyttel et al., 1992; Owens et al., 2001; Burke et al., 2002; Sánchez et al., 2003). However, the magnitude of the potency difference, about 20- to 125-fold more than citalopram and R-citalopram, is larger than that predicted from the above studies. Based on SERT affinity, escitalopram should be about 2-fold more potent than citalopram, and clinically, escitalopram appears to be at least 2-fold more potent than citalopram (Burke et al., 2002). Development of the 5-HT transporter system could have contributed to these discrepancies because the number and distribution of transporters substantially change with age (e.g., Lebrand et al., 1998).

In addition to reducing USVs, all of the drugs altered motor behavior to varying degrees, as measured by grid crossing and rolling. Escitalopram and venlafaxine were the most behaviorally selective in reducing pup USVs (Table 4), although escitalopram was considerably more potent than venlafaxine (ca. 140-fold). Motor stimulation occurred only at the highest doses of escitalopram (0.3 and 0.56 mg/kg) and venlafaxine (56 mg/kg), which were, respectively, about 30- and 19-fold greater than the minimally effective doses (MEDs) for reducing USVs (Table 4). Unlike the other treatments, escitalopram and venlafaxine did not increase rolling. Only when very high doses (1 and 3 mg/kg) of escitalopram were given in combination with R-citalopram did escitalopram mimic the effects of citalopram and increase rolling. The increased rolling after these treatments is relatively modest, about half of what was observed after benzodiazepine treatment (Rowlett et al., 2001). Although the lack of increased rolling is consistent with the clinical observations that escitalopram is well tolerated (Montgomery et al., 2001; Burke et al., 2002; Wade et al., 2002; Waugh and Goa, 2003), the clinical relevance of motor stimulation observed with high doses of escitalopram and venlafaxine remains to be determined.

Motor stimulation has also been observed after treatment with 5-HT_1B and 5-HT_2A, but not 5-HT_1A, receptor agonists in mouse pups (Fish et al., 2000; E. W. Fish and K. A. Miczek, unpublished data). Interestingly, the citalopram enantiomers and venlafaxine showed dose-effect relationships on motor stimulation different from those of paroxetine and fluoxetine. Lower doses of paroxetine and fluoxetine were stimulating and higher doses were not. In contrast, only the high doses of citalopram enantiomers increased grid crossing. Similar dose-effect relationships between citalopram, paroxetine, and fluoxetine have also been observed in adult mice placed in a novel open-field (Brocco et al., 2002). The different dose-effect curves for the citalopram enantiomers, paroxetine and fluoxetine, may be because of their actions at muscarinic M₁ and 5-HT_2C receptors, respectively (Owens et al., 2001).

One hypothesis for the differences in potency between escitalopram and racemic citalopram is that R-citalopram inhibits the effects of escitalopram. Coadministration of R-citalopram reduced escitalopram's elevation of cortical extracellular 5-HT and reduction of shock-induced USVs (Sánchez, 2003a; Mørk et al., 2003). The results of the interaction between escitalopram and R-citalopram in the present study are consistent with this hypothesis. When given together, R-citalopram caused a 10-fold parallel shift to the right in the USV-reducing effects of escitalopram. There are several mechanisms through which R-citalopram might inhibit the effects of escitalopram. One possibility is that R-citalopram's binding to the H₁ receptor alters the effects of escitalopram. To test this hypothesis, the H₁ receptor antagonist pyrilamine was tested alone and in combination with escitalopram. Pyrilamine (mepyramine) was chosen because it was the ligand for the H₁ receptor in escitalopram binding studies (Owens et al., 2001; Sánchez et al., 2003). Alone, pyrilamine modestly suppressed calling, as was observed in adult rats (Sánchez, 2003b), and its effects were quite variable. When administered with escitalopram, there was no evidence for a shift in escitalopram's dose-effect curve, indicating that H₁ antagonism is an unlikely mechanism for R-citalopram's inhibition of escitalopram's effects. Another hypothesis is that during the interaction between escitalopram and R-citalopram, escitalopram is metabolized faster, leaving higher levels of the less effective R-citalopram to displace escitalopram from the transporter. However, levels of escitalopram are not affected by treatment with R-citalopram in the rat brain (Mørk et al., 2003). A third possibility is that R-citalopram's inhibition of escitalopram's effects is due to noncompetitive binding to a different site on the transporter protein that conformationally inhibits escitalopram's binding to the transporter. Future research into the combined effects of the single enantiomers and determining whether R-citalopram also inhibits the effects of other SSRIs will help elucidate the mechanisms through which R-citalopram inhibits the effects of escitalopram.

The acute anxiolytic-like effects of the SSRIs on neonatal vocalizations (Mos and Olivier, 1989; Winslow and Insel, 1990; Molewijk et al., 1996; Olivier et al., 1998) differ from their acute effects in humans. In humans, the anxiolytic effects of SSRIs emerge only after chronic treatment, and there are some reports that they may initially increase anxiety (for review, see Nutt et al., 1999). In preclinical studies, the acute effects of SSRIs can be detected in several procedures that are used for characterizing anxiolytic drugs, but the nature of these effects varies markedly with the experimental procedure (for review, see Griebel, 1995; Borsini et al., 2002). Consistent anxiolytic-like effects of SSRIs occur on measures of footshock-induced USVs (Sánchez and Meier, 1997; Schreiber et al., 1998; Sánchez et al., 2003), burying behavior (Njung'e and Handley 1991), conditioned freezing (Hashimoto et al., 1996), and increased movement across the electrified grids of the four plate test (Hascoet et al., 2000). Anxiogenic-like effects of SSRIs have been observed on light-dark exploratory behavior (Sánchez and Meier, 1997), novelty-suppressed feeding (Bodnoff et al., 1989), the mouse defensive battery (Griebel, 1995), the social interaction test (File et al., 1999), and the elevated plus maze (File et al., 1999). Interpreting these contradictory results is difficult, and it is clear that novel procedures for assessing anxiety-like behaviors in animals must be developed. Although escitalopram inhibits both neonatal and adult vocalizations and enhances exploratory behavior in the light and dark box, it will be important to extend this assessment to several other procedures.

Procedures measuring vocalizations appear to be particularly sensitive to the acute anxiolytic-like effects of SSRIs, but there are potentially confounding variables to consider (Winslow and Insel, 1991). Young, developing animals may be differentially sensitive to drug treatments than are adults because of differences in the number of receptors as well as pharmacokinetics (e.g., Lebrand et al., 1998; Gow et al., 2001). Manipulations of the 5-HT system in young animals could alter USVs by interfering with its role in ongoing neural development (e.g., Azmitia, 2001). There is also some evidence that certain types of vocalizations in rat pups are influenced by variables such as respiration, thermoregulation, and heart rate (Sokoloff and Blumberg, 1997), particularly after extended separation from the dam. Different neural microcircuits appear to subserve different kinds of vocalizations (Jürgens and Pratt, 1979; Covington and Miczek, 2003). Determining whether these circuits are indeed the targets for the currently studied SSRIs could address whether the present results are significantly influenced by pharmacokinetic, developmental, or physiological variables.

Escitalopram is currently the most selective SSRI for inhibiting the SERT, and its effects in mouse pups are consistent with its clinical efficacy in treating anxiety. Although the maternal separation procedure appears to be particularly sensitive to detect the anxiolytic-like effects of SSRIs, it is important to examine escitalopram in other animal models of anxiety to assess the extent to which these results are species- and age-dependent. As SSRIs emerge as the pharmacotherapy of choice for anxiety disorders, preclinical researchers must continue to develop externally valid procedures for assessing their effects in laboratory animals. Future studies may also consider whether the differences in potency and behavioral selectivity of the SSRIs can be explained by differential activation of particular pre- and/or postsynaptic 5-HT receptors that mediate the USV-reducing effects of escitalopram and other SSRIs. Furthermore, it will be interesting to determine the molecular mechanisms underlying R-citalopram's inhibitory effects on escitalopram's activity.