Norepinephrine-deficient mice lack responses to antidepressant drugs, including selective serotonin reuptake inhibitors

John F. Cryan*†, Olivia F. O’Leary‡§, Sung-Ha Jin‡, Julie C. Friedland‡, Ming Ouyang‡, Bradford R. Hirsch‡, Michelle E. Page*¶, Ashutosh Dalvi*, Steven A. Thomas‡, and Irwin Lucki*‡ʈ

Departments of *Psychiatry and ‡Pharmacology, University of Pennsylvania, Philadelphia, PA 19104; and §Department of Pharmacology, National University of Ireland, Galway, Ireland

Edited by Bruce S. McEwen, The Rockefeller University, New York, NY, and approved April 13, 2004 (received for review February 17, 2004) Mice unable to synthesize norepinephrine (NE) and epinephrine depletion studies difficult to interpret (10, 11). For these reasons, due to targeted disruption of the dopamine ␤-hydroxylase gene, attempts to resolve the role of NE in the behavioral effects of Dbh, were used to critically test roles for NE in mediating acute antidepressants by using neurotoxin-induced lesions has re- behavioral changes elicited by different classes of antidepressants. mained largely controversial and unresolved (12–16). To this end, we used the tail suspension test, one of the most The generation of mice with a targeted disruption of the gene widely used paradigms for assessing antidepressant activity and for dopamine (DA) ␤-hydroxylase (DBH) provides a genetic depression-related behaviors in normal and genetically modified model for examining the role of NE in complex physiological and -mice. Dbh؊/؊ mice failed to respond to the behavioral effects of behavioral functions (17–21). NE can also be replaced tran various antidepressants, including the NE reuptake inhibitors de- siently by administration of L-threo-3,4-dihydroxyphenylserine sipramine and reboxetine, the monoamine oxidase inhibitor par- (L-DOPS), a synthetic precursor of NE which bypasses DBH by gyline, and the atypical antidepressant bupropion, even though means of metabolism by aromatic L-amino acid decarboxylase ؉/؊ they did not differ in baseline immobility from Dbh mice, which (22). L-DOPS has been used to reverse behavioral, neurochem- have normal levels of NE. Surprisingly, the effects of the selective ical, and physiological phenotypes in the DbhϪ/Ϫ mice (17–22). serotonin reuptake inhibitors (SSRIs) fluoxetine, sertraline, and Thus, DbhϪ/Ϫ mice offer a more complete, specific, and readily paroxetine were also absent or severely attenuated in the Dbh؊/؊ reversible model for examining the role of adrenergic signaling mice. In contrast, citalopram (the most selective SSRI) was equally in the mechanism of antidepressant action. For these reasons, we effective at reducing immobility in mice with and without NE. examined the behavioral effects of antidepressants from several Ϫ Ϫ Restoration of NE by using L-threo-3,4-dihydroxyphenylserine re- classes in Dbh / and littermate control mice. Behavioral effects instated the behavioral effects of both desipramine and paroxetine of antidepressants were studied by using the tail suspension test ؊ ؊ in Dbh / mice, thus demonstrating that the reduced sensitivity to (TST), a well characterized test sensitive to antidepressants from antidepressants is related to NE function, as opposed to develop- different pharmacological classes in the mouse (23). This test is mental abnormalities resulting from chronic NE deficiency. Micro- among the most widely used in studying antidepressant-related dialysis studies demonstrated that the ability of fluoxetine to phenotypes in genetically modified mice (24, 25). The results of ؊ ؊ increase hippocampal serotonin was blocked in Dbh / mice, these studies demonstrate an important role for NE in the acute whereas citalopram’s effect was only partially attenuated. These response to antidepressants from different pharmacological data show that NE plays an important role in mediating acute classes, including the SSRIs. behavioral and neurochemical actions of many antidepressants, including most SSRIs. Materials and Methods Animals. Mice were derived from a hybrid line (129͞SvCPJ and Ϫ Ϫ ϩ Ϫ ntidepressant medications are thought to elicit their ther- C57BL͞6J). Dbh / and heterozygote (Dbh / ) mice were bred Aapeutic effects by increasing synaptic concentrations of the as described (18) and maintained on a 12-h light͞dark cycle monoamines serotonin (5-HT) and͞or norepinephrine (NE) (1). (lights on at 06:00 hours) in a specific pathogen-free facility. Most tricyclic antidepressants enhance NE and 5-HT transmis- Food and water were freely available, and mice were maintained sion because they inhibit presynaptic reuptake, although most of according to National Institutes of Health guidelines. All exper- these drugs have affinity for other neurotransmitter receptors imental procedures were carried out in accordance with proto- that contribute to their side effect profile. More recently, cols approved by the University of Pennsylvania Institutional selective 5-HT reuptake inhibitors (SSRIs) have become widely Animal Care and Use Committee. Genotype was deduced from Ϫ Ϫ prescribed because of their improved safety profile from limited phenotype (Dbh / mice exhibit delayed growth during adoles- interaction with other neurotransmitter receptors (1). However, cence and ptosis), and a subset of mice was confirmed by PCR ϩ Ϫ ϩ ϩ some investigators have recently challenged whether SSRIs (17). Dbh / mice are indistinguishable from wild-type (Dbh / ) produce effects selectively on the serotonergic system over the noradrenergic and dopaminergic systems (2, 3). Indeed, a num- ber of in vivo microdialysis studies have shown that acute (4–6) This paper was submitted directly (Track II) to the PNAS office. and chronic (7–9) treatment with certain SSRIs can increase Abbreviations: NE, norepinephrine; NET, NE transporter; 5-HT, serotonin; SERT, 5-HT ␤ extracellular concentrations of NE in addition to that of 5-HT. transporter; DA, dopamine; DBH, DA -hydroxylase; SSRI, selective 5-HT reuptake inhibitor; L-DOPS, L-threo-3,4-dihydroxyphenylserine; TST, tail suspension test; FST, forced swim test. Attempts to delineate the role of NE in the behavioral effects †Present address: Neuroscience Disease Area, Novartis Institutes for Biomedical Research, of antidepressants have used inhibition of synthetic enzymes to Novartis Pharma AG, CH-4002 Basel, Switzerland. deplete the neurotransmitter or chemical lesion techniques to ¶Present address: Department of Neurobiology and Anatomy, Drexel University College of destroy NE terminals. However, many depletion agents, such as Medicine, 2900 Queen Lane, Philadelphia, PA 19129. ␣ the tyrosine hydroxylase inhibitor -methyl-p-tyrosine or the ʈTo whom correspondence should be addressed at: Department of Psychiatry, University of vesicle disruptor reserpine, affect other neurotransmitters as Pennsylvania, 538A Clinical Research Building, 415 Curie Boulevard, Philadelphia, PA well. Further, incomplete lesions and sprouting and increased 19104-6140. E-mail: [email protected]. turnover at surviving terminals can make negative results with © 2004 by The National Academy of Sciences of the USA

8186–8191 ͉ PNAS ͉ May 25, 2004 ͉ vol. 101 ͉ no. 21 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0401080101 Downloaded by guest on September 24, 2021 mice as to NE and epinephrine levels, as determined in various Statistical Analysis. A two-way ANOVA, with drug treatment and brain regions ex vivo (22). Correspondingly, in all behavioral tests genotype as factors, was carried out on absolute immobility in carried out to date, no difference emerged between Dbhϩ/Ϫ mice studies examining the effects of various antidepressants on and wild-type mice; therefore, Dbhϩ/Ϫ mice were used as con- behavior in the TST. A three-way ANOVA was carried out on trols in all experiments. Adult (3–7 months) male and female the data from the L-DOPS experiment by using pretreatment littermates of each genotype were evenly distributed to experi- with L-DOPS, drug treatment, and genotype as factors. Any mental and control groups for each treatment. All behavioral factors that demonstrated overall significant differences were experimental sessions were conducted between 12:00 and 19:00 analyzed further by using Fisher’s test of least significant differ- hours. ences. The effects of sertraline on behavior were analyzed alone. The behavioral effects of fluoxetine, paroxetine, and citalopram The TST. The TST was carried out essentially as described (23). were investigated in groups of mice treated simultaneously with Thirty min after injection (24 h and 30 min in the case of drug or vehicle, and the data were collated and analyzed pargyline), mice were individually suspended by the tail to a together. The effects of drug treatments on extracellular 5-HT horizontal ring stand bar (distance from floor ϭ 30 cm) using levels were determined by mixed two-factor ANOVA, with adhesive tape (distance from tip of tail ϭ 2 cm). Typically, mice genotype a between-subjects factor and time analyzed by using demonstrated several escape-oriented behaviors interspersed repeated measures. with temporally increasing bouts of immobility. A 6-min test session was used, which was videotaped. Videotapes were sub- Results sequently scored by a trained observer who was unaware of the The Effects of Various Antidepressants in the TST. Baseline behavior treatment and genotype. The parameter recorded was the in the TST was examined after injections of physiological saline number of seconds spent immobile. A subset of the mice in (0.9%) into DbhϪ/Ϫ and littermate control mice 30 min before studies 1 and 2 (i.e., for the catecholaminergic agents, desipra- testing. There were no significant differences of immobility mine, reboxetine, pargyline, and bupropion) had undergone the between Dbhϩ/Ϫ and DbhϪ/Ϫ mice given saline (Fig. 1). Fur- forced swim test (FST) 1 week earlier (20). Parallel studies were thermore, there are no differences in spontaneous activity in the undertaken to demonstrate that there was no difference in home cage between genotypes that would mask a differential baseline or in response to these compounds in the TST 1 week response in the TST (S.A.T., unpublished observations). after FST when compared with mice that have not undergone Desipramine dose-dependently decreased immobility of the FST (data not shown). control mice in the TST (Fig. 1A). In contrast, there was no significant effect in the DbhϪ/Ϫ mice. Similarly, the higher dose Drugs. All drugs were freshly prepared before use and injected of reboxetine (20 mg͞kg) reduced immobility in the control mice ͞ Ϫ/Ϫ i.p. by using a volume of 10 ml kg. Desipramine HCl (Mr 266, but not in the Dbh mice (Fig. 1B). The atypical antidepressant ͞ Sigma); reboxetine HBr (Mr 319, Pharmacia and Upjohn, bupropion (20 mg kg) and the monoamine oxidase inhibitor ͞ ϫ Kalamazoo, MI); pargyline HCl (Mr 159, Sigma) bupropion HCl pargyline (75 mg kg 2) were both effective at reducing ϩ/Ϫ (Mr 240, Research Biochemicals, Natick, MA); fluoxetine HCl immobility in the TST only in Dbh mice (Fig. 1C). Pargyline- Ϫ/Ϫ (Mr 309, Eli Lilly, Indianapolis); citalopram HBr (Mr 324, and bupropion-treated Dbh mice had significantly greater ϩ/Ϫ Lundbeck, Copenhagen); and paroxetine HCl (Mr 329, Smith- immobility than did their Dbh counterparts. Kline Beecham) were dissolved in deionized H2O and were sonicated mildly. Sertraline HCl (Mr 306, Pfizer, Groton, CT) The Effects of the SSRIs in the TST. The consequences of the absence was dissolved in deionized H2O with a drop of Tween-80 and was of NE on the behavioral effects of a series of different SSRIs sonicated mildly. All drug doses were calculated as the base were examined. The effects of sertraline (5–20 mg͞kg) were ͞ weight and expressed as mg kg, although molecular masses (Mr) tested first (Fig. 1D). Sertraline decreased immobility in control are provided for conversion of doses to ␮mol͞kg. Control mice mice in a dose-related fashion, but this effect was completely received saline in a corresponding volume of 10 ml͞kg except for abolished in the DbhϪ/Ϫ mice. The behavioral effects of fluox- sertraline-treated mice, which were given saline with a drop of etine (20–40 mg͞kg), paroxetine (5–20 mg͞kg), and citalopram NEUROSCIENCE Tween-80. L-DOPS (Sumitomo Pharmaceuticals Ltd., Osaka) (5–20 mg͞kg) were studied in different squads of mice run was administered as described (22). Briefly, L-DOPS (1 g͞kg) simultaneously, and data from these groups were collated (Fig. was dissolved at 20 mg͞ml in 0.2 M HCl containing 2 mg͞ml 1E). Fluoxetine reduced immobility at both doses tested (20 and ascorbic acid. The solution was neutralized with 10 M NaOH just 40 mg͞kg) in the Dbhϩ/Ϫ mice, whereas it was ineffective at the before s.c. injection, which was given 4–6 h before behavioral lower dose and increased immobility at the higher dose in the testing. DbhϪ/Ϫ mice. Although paroxetine reduced immobility at both doses tested (5 and 20 mg͞kg) in the Dbhϩ/Ϫ mice, it was Microdialysis. Mice were implanted with custom-constructed mi- ineffective at the lower dose and reduced immobility only crodialysis probes (26) in the ventral hippocampus at Ϫ2.8 mm partially at the higher dose in the DbhϪ/Ϫ mice. In contrast, AP, Ϯ 3.2 mm ML, and Ϫ5.0 DV relative to bregma. After citalopram reduced immobility with equal effect at both doses (5 surgery, the mice were placed in clear polycarbonate cylinders and 20 mg͞kg) in Dbhϩ/Ϫ and DbhϪ/Ϫ mice (Fig. 1E). In a (21.5 cm in height ϫ 17.5 cm in diameter) and allowed to recover separate study, 1 mg͞kg citalopram produced a nonsignificant overnight. The probes were continuously perfused at 1.5 ␮l͞min reduction in immobility of 24 and 27 s in the Dbhϩ/Ϫ and DbhϪ/Ϫ ͞ ͞ with artificial cerebrospinal fluid (147 mM NaCl 1.7 mM CaCl2 mice, respectively, when compared with saline treatment, and ͞ 0.9 mM MgCl2 4 mM KCl). Only data from mice with confirmed the reductions did not differ significantly between genotypes probes in the ventral hippocampus were used. Mice received an (data not shown). injection of fluoxetine or citalopram (at 20 mg͞kg i.p.) after 2 h of baseline sampling. In reverse dialysis experiments, the probes The Effects of Dual Reuptake Inhibition in the TST. The greatly were perfused with artificial cerebrospinal fluid for2hof diminished behavioral effects in the TST of fluoxetine, sertra- baseline sampling and then the perfusion solution was switched line, and paroxetine as compared with citalopram correlate with to artificial cerebrospinal fluid containing 1 ␮M fluoxetine by the selectivity of these compounds for inhibition of 5-HT versus using a liquid switch selector (UniSwitch Bioanlytical Systems, NE reuptake in vitro, with citalopram being the most selective West Lafayette, IN). The amount of 5-HT in the dialysate was (27). One explanation for the differential effects of these SSRIs determined by HPLC coupled to electrochemical detection (26). could be related to a unique aspect of the DbhϪ/Ϫ mice. Evidence

Cryan et al. PNAS ͉ May 25, 2004 ͉ vol. 101 ͉ no. 21 ͉ 8187 Downloaded by guest on September 24, 2021 Fig. 2. The effects of the selective NE reuptake inhibitor reboxetine (10 mg͞kg i.p.) and the SSRI citalopram (5 mg͞kg i.p.) given individually and in combination on immobility in the TST. An ANOVA revealed a significant interaction between drug treatment and genotype [F(3, 72) ϭ 4.22, P Ͻ 0.01]. n ϭ 9–10 mice per group. *, Groups that differed significantly from saline- treated mice of the corresponding genotype (P Ͻ 0.05); #, groups differed significantly from the opposite genotype tested at the same dose (P Ͻ 0.05), according to Fisher’s probable least-squares difference test.

nitude, indicating that the addition of reboxetine to citalopram neither antagonized nor augmented the antiimmobility effect of citalopram in the DbhϪ/Ϫ mice. Similar results were seen with lower concentrations of citalopram and reboxetine (1 mg͞kg each), except that the additive effect of these two compounds together in the Dbhϩ/Ϫ mice was more apparent (data not shown). Taken together, these results argue against the idea that Ϫ/Ϫ Fig. 1. The effects of antidepressants on immobility in the TST. An ANOVA the lack of effect of some SSRIs in the Dbh mice is due to revealed a significant genotype ϫ dose interaction for the tricyclic antide- ectopic increases of extracellular DA rather than deficiency pressant desipramine [F(2, 53) ϭ 7.80, P Ͻ 0.01] (A), for the selective NE of NE. reuptake inhibitor reboxetine [F(2, 48) ϭ 5.83, P Ͻ 0.01] (B), and for the atypical antidepressant bupropion (20 mg͞kg i.p.) and the monoamine oxi- Reinstatement of Behavioral Effects in the TST. To further test the dase inhibitor pargyline (75 mg͞kg i.p. given twice) [F(2, 113) ϭ 4.65, P ϭ 0.01] idea that the behavioral effects of antidepressants may depend Ϫ Ϫ (C). The SSRI sertraline reduced immobility in control but not in Dbh / mice, on adrenergic signaling, NE was restored acutely in the DbhϪ/Ϫ according to a significant genotype ϫ dose interaction [F(3, 68) ϭ 4.64, P Ͻ mice by using L-DOPS. Importantly, acute administration of 0.01] (D). The pattern of effects in DbhϪ/Ϫ mice varied with the different SSRIs, L-DOPS does not reduce the elevated levels of DA found in the fluoxetine, paroxetine, and citalopram [F(6, 211) ϭ 8.97, P Ͻ 0.01] (E). n ϭ 9–38 DbhϪ/Ϫ mice (22). Restoration of NE allowed the DbhϪ/Ϫ mice mice per group. *, Groups that differed significantly from saline-treated mice of the corresponding genotype (P Ͻ 0.05); #, groups differed significantly to respond to both desipramine and paroxetine in the TST (Fig. from the opposite genotype tested at the same dose (P Ͻ 0.05), according to 3). In an independent replication of effects shown in Fig. 1, ϩ Ϫ Fisher’s probable least-squares difference test. There were no differences in desipramine and paroxetine reduced immobility in the Dbh / Ϫ Ϫ immobility of mice treated with saline between genotypes in all experiments. mice only, with no effect in the Dbh / mice pretreated with For Figs. 1–4, data are represented as mean values, with vertical lines indicat- vehicle. However, in mice pretreated with L-DOPS, desipramine ing 1 SEM. and paroxetine significantly reduced immobility in Dbhϩ/Ϫ and DbhϪ/Ϫ mice, and the extent of this reduction was similar between genotypes. Interestingly, DbhϪ/Ϫ mice (but not Dbhϩ/Ϫ suggests that these mice store and release the neurotransmitter mice) pretreated with L-DOPS had significantly lower baseline DA from their adrenergic terminals (22). This result is expected immobility values as compared with their vehicle-pretreated because DA is the endogenous precursor of NE, and DA is a counterpart. This observation suggests that the DbhϪ/Ϫ mice may substrate for both the NE transporter (NET) and the vesicular have enhanced sensitivity to changes in NE levels as compared monoamine transporters (28). In normal animals, the SSRIs with control mice. The ability of L-DOPS to restore sensitivity to fluoxetine, sertraline, and paroxetine augment extracellular lev- desipramine and paroxetine in the DbhϪ/Ϫ mice indicates that els of NE (4–7). As a result, these compounds would be expected the effects of these antidepressants are mediated in part by an to augment extracellular levels of DA near adrenergic terminals acute physiological requirement for adrenergic signaling. It also Ϫ/Ϫ in Dbh mice. It is possible that an ectopic rise in extracellular indicates that the lack of an effect of these antidepressants in DA could counteract the behavioral effects of an increase in untreated DbhϪ/Ϫ mice is not due to abnormal development or extracellular 5-HT. To test this hypothesis, the effects of the two ectopic DA. most selective reuptake inhibitors, citalopram for 5-HT and reboxetine for NE, were examined in the TST when given alone SSRIs and Extracellular Levels of 5-HT. Ϫ Ϫ The role of adrenergic and in combination to control and Dbh / mice (Fig. 2). Post hoc signaling in regulating the effects of SSRIs could be upstream or analysis revealed that citalopram, reboxetine, and both drugs in downstream of the elevation in extracellular 5-HT levels elicited ϩ Ϫ combination reduced immobility in the Dbh / mice. The effects by these compounds. Microdialysis was used to measure the of citalopram given in combination with reboxetine compared ability of the SSRIs to increase extracellular 5-HT levels in the with citalopram given alone were not statistically significant (P ϭ hippocampus of Dbhϩ/Ϫ and DbhϪ/Ϫ mice. Acute systemic Ϫ Ϫ 0.08). In the Dbh / mice, citalopram reduced immobility given administration of fluoxetine (20 mg͞kg) produced a 3-fold either alone or in combination with reboxetine. In contrast to the increase in extracellular 5-HT in control mice but caused no pattern seen in Dbhϩ/Ϫ mice, these effects were equal in mag- increase in the DbhϪ/Ϫ mice (Fig. 4A). On the other hand,

8188 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0401080101 Cryan et al. Downloaded by guest on September 24, 2021 The lack of effect of both desipramine and reboxetine on the TST in DbhϪ/Ϫ mice is consistent with the hypotheses that both compounds elicit their effects by increasing synaptic NE concentrations after selective blockade of NET (32). Reple- tion of NE with the precursor L-DOPS restored the behavioral effects of desipramine. Immobility was also reduced by L- DOPS pretreatment alone in DbhϪ/Ϫ mice, suggesting that mechanisms mediating the effect of L-DOPS in the TST become sensitized with the chronic loss of NE. Furthermore, the lack of any difference in NET messenger RNA, immuno- reactivity, and radioligand binding between DbhϪ/Ϫ and Dbhϩ/Ϫ mice (20, 33), supports the premise that the inability of noradrenergic antidepressants to produce behavioral effects in DbhϪ/Ϫ mice is not due to a reduction in NET, but rather

Ϫ/Ϫ a loss of the endogenous ligand itself. These findings are Fig. 3. Restoring NE in Dbh mice by means of pretreatment with L-DOPS Ϫ/Ϫ reinstates behavioral changes induced by desipramine and paroxetine in the consistent with our previous report (20) that Dbh mice fail TST. There was a significant three-way genotype ϫ drug treatment ϫ L-DOPS to demonstrate behavioral effects to NE antidepressants in the pretreatment interaction [F(1, 160) ϭ 24.22, P Ͻ 0.001], according to ANOVA. FST, and are one of the few demonstrations that the selective n ϭ 9–20 mice per group. *, Groups that differed significantly from saline- depletion of NE prevents the behavioral effects of noradren- treated mice of the corresponding genotype (P Ͻ 0.05); #, groups differed ergic antidepressants (12, 15). This outcome could be because significantly from the opposite genotype tested at the same dose (P Ͻ 0.05); adrenergic lesions are often incomplete (10, 11), or there was $, groups that differed from relevant drug-treated control (P Ͻ 0.05), accord- a failure to target the adrenergic neurons, such as the ventral ing to Fisher’s probable least-squares difference test. NE bundle, that are critically involved (16). The antidepressant bupropion could mediate its effects through potentiation of DA transmission because it has a higher delivering fluoxetine (1 ␮M) directly to the hippocampus by affinity for the DA transporter than either the NET or 5-HT using reverse microdialysis increased extracellular 5-HT levels by ϩ Ϫ Ϫ Ϫ transporter (SERT) (34). However, bupropion and its metabo- Dbh / Dbh / B 3-fold in and mice (Fig. 4 ). In contrast to lite 306U73 inhibit the firing of locus coeruleus noradrenergic fluoxetine, acute systemic administration of citalopram (20 neurons at doses much lower than that needed to inhibit DA cell mg͞kg) caused a 4-fold increase in extracellular 5-HT in control Ϫ/Ϫ firing (35), and bupropion produces effects on both noradren- mice and a 3-fold increase in extracellular 5-HT in Dbh mice ergic and serotonergic systems (36). Likewise, the inability of the (Fig. 4C). Thus, the refractory behavioral effects of fluoxetine in Ϫ/Ϫ monoamine oxidase inhibitor pargyline to produce antidepres- Dbh mice were recapitulated in the effects of systemic sant-like effects in DbhϪ/Ϫ mice suggests that although pargyline fluoxetine on extracellular 5-HT levels and are probably due to would be expected to increase extracellular levels of NE, 5-HT, upstream alterations in adrenergic circuitry that impact 5-HT and DA, its effects on adrenergic transmission are necessary for transmission. Although citalopram’s effects were also somewhat Ϫ/Ϫ the mediation of its behavioral effects in the TST and FST (20). diminished in Dbh mice, 5-HT levels may be increased Furthermore, we have previously shown that the lack of behav- sufficiently by this SSRI to produce behavioral effects. ioral effects of pargyline in DbhϪ/Ϫ mice are not due to any Discussion potential adaptive changes in monoamine oxidase activity sec- ondary to chronic absence of NE (20). The results from the The identification of specific neurochemical substrates that studies employing bupropion and pargyline indicate that en- mediate the behavioral effects of antidepressants is important hancing dopaminergic and͞or serotonergic signaling may not be because it provides insight into neurochemical mechanisms sufficient to observe behavioral effects in the FST and TST in the determining their therapeutic actions and strategies for devel- absence of NE. NEUROSCIENCE oping more effective treatments. The results of the present The most surprising finding of the current study is the studies provide striking evidence that NE is essential for the dramatically blunted behavioral responses to the SSRIs flu- manifestation of the acute behavioral effects of a variety of oxetine, sertraline, and paroxetine in the DbhϪ/Ϫ mice. Intact antidepressants from different classes. 5-HT transmission is required for SSRIs to produce their Interestingly, baseline behavior between mice with and behavioral effects in the TST and other behavioral models without NE was unaltered in the TST. This finding may be because the behavioral effects are prevented after the deple- surprising, given the large body of evidence supporting a role tion of 5-HT with the tryptophan hydroxylase inhibitor p- for this amine in both the pathology of depression and in the chlorophenylalanine (refs. 13 and 31 and O.F.O. and I.L., integration of the stress response (29, 30). However, it agrees unpublished observations). Although two previous studies (37, with our previous data that showed that both genotypes have 38) in rats reported that lesions of the locus coeruleus did not similar baseline behavior in the FST (20), and with studies alter the behavioral effects of fluoxetine or sertraline, the using various noradrenergic neurotoxins, which show no dif- substantial noradrenergic innervation spared by these lesions ference in baseline performance in tests of acquired immobil- could have sustained the effects of SSRIs. Further, Soubrie et ity (13, 15, 16). It is also of interest that depletion of 5-HT by al. (39) have shown that noradrenergic lesions of the hip- p-chlorophenylalanine or 5,7-dihydroxytryptamine fails to al- pocampus, induced by 6-hydroxydopamine, significantly delays ter baseline immobility in both the FST and the TST (refs. 13, the reversal of escape deficits by several antidepressants, 16, and 31 and O.F.O. and I.L., unpublished observations). including the 5-HT reuptake inhibitor clomipramine, in the These observations suggest that NE and 5-HT are not neces- learned-helplessness model of depression. Interestingly, the sary for the tonic regulation of baseline performance in these lesion does not affect baseline acquisition of learned helpless- tasks, although it is possible that compensatory mechanisms ness. The present data challenge the assumption that 5-HT from other neurotransmitters stabilize the behavioral re- regulates the effects of all SSRIs exclusively and demonstrate sponse. In contrast, NE and 5-HT appear to be necessary for that these structurally heterogeneous compounds have various reductions of immobility in these tasks when their levels are requirements for NE to manifest behavioral effects. Previous augmented by antidepressants. investigators have questioned the selectivity of a number of

Cryan et al. PNAS ͉ May 25, 2004 ͉ vol. 101 ͉ no. 21 ͉ 8189 Downloaded by guest on September 24, 2021 measured ex vivo (see supporting information, which is pub- lished on the PNAS web site). Therefore, an impact on NE transmission may participate or even be required for the behavioral effects of these SSRIs to be fully manifested. Citalopram is unique among the clinically available SSRIs in having the greatest selectivity for SERT over NET in vitro (27, 41) and does not, after systemic application, increase NE concentrations in vivo (4) [it has been reported to increase NE after reverse dialysis in the hippocampus (42)]. In accordance with this finding, the present data show that citalopram is the only antidepressant tested that is equally effective in mani- festing its behavioral effects in mice with and without NE. This result suggests that whereas the other antidepressants must recruit adrenergic mechanisms to elicit their behavioral ef- fects, citalopram reduces immobility independent of NE and therefore may be the only truly selective SSRI of those tested. Microdialysis studies examined the impact of the absence of NE on 5-HT efflux directly. Fluoxetine produced no increase of 5-HT levels in DbhϪ/Ϫ mice even though 5-HT levels increased 3-fold in control mice. Although the effects of citalopram were also attenuated in DbhϪ/Ϫ mice when com- pared with the 4-fold rise in control mice, the 3-fold elevation observed in the mutant mice was likely to be sufficient to sustain the behavioral response in the TST. Importantly, local application of fluoxetine increased hippocampal 5-HT in both genotypes, indicating that fluoxetine is active at SERT in the DbhϪ/Ϫ mice, as expected. Serotonergic cells in the dorsal ␣ raphe can be activated by means of 1-adrenergic receptor stimulation by afferent noradrenergic input (43, 44). This point of interaction between NE and 5-HT could account for the absence of increased 5-HT efflux and behavioral effects in DbhϪ/Ϫ mice. It appears that the ability of SSRIs to increase 5-HT efflux depends on increased noradrenergic input to the 5-HT neurons. Further studies are required to determine whether this inability of fluoxetine to increase 5-HT in DbhϪ/Ϫ mice can be reversed by pretreatment with L-DOPS. Citalo- pram seems to have unique properties among SSRIs, perhaps a high selectivity for SERT over NET or other unknown effects, that allow it to bypass the normal requirement for adrenergic input so that it is least affected by the absence of Fig. 4. Effects of citalopram and fluoxetine on extracellular 5-HT concen- NE. Taken together, these data support the idea that the trations in Dbhϩ/Ϫ and DbhϪ/Ϫ mice. Baseline values were determined from effects of most SSRIs, both behavioral and neurochemical, are the average of four samples preceding drug injection or infusion. All infusions diminished by the absence of NE transmission. were begun at time ϭ 0 and all injections were performed at time ϭ 0. (A) A key concept underlying modern antidepressant drug devel- ͞ Ϫ/Ϫ ϭ Effects of systemic administration of 20 mg kg i.p. fluoxetine in Dbh (n opment is whether NE and 5-HT systems are independent ϭ Ϯ ͞ ␮ ϩ/Ϫ ϭ 5; baseline 6.15 0.20 fmol 12- l sample) and Dbh mice (n 5; mechanisms for treating depression or whether they are inter- baseline ϭ 4.34 Ϯ 0.23 fmol͞12-␮l sample). ANOVA indicated a significant genotype ϫ time interaction (F(9, 72) ϭ 4.69, P ϭ 0.001). (B) Local perfusion of dependent (45). Much evidence favoring either independent or 1 ␮M fluoxetine on extracellular 5-HT levels in the ventral hippocampus did interdependent outlooks has been discussed in the literature (46, not produce significantly different effects between DbhϪ/Ϫ (n ϭ 5; baseline ϭ 47). The lack of an increase in extracellular 5-HT after fluoxetine Ϫ Ϫ 2.39 Ϯ 0.08 fmol͞12-␮l sample) and control mice (n ϭ 5; baseline ϭ 1.77 Ϯ 0.04 administration to the Dbh / mice suggests that an interdepen- fmol͞12-␮l sample); P Ͼ 0.05. (C) Effects of systemic administration of 20 dence between 5-HT and NE can exist. However, it is also likely mg͞kg i.p. citalopram in DbhϪ/Ϫ (n ϭ 5; baseline ϭ 3.32 Ϯ 0.18 fmol͞12-␮l ϩ Ϫ that these systems can act independently. The actions of citalo- sample) and Dbh / mice (n ϭ 7; baseline ϭ 4.75 Ϯ 0.08 fmol͞12-␮l sample). Ϫ/Ϫ ϫ ϭ pram are mostly unaffected in the Dbh mice, and depletion ANOVA indicated a significant genotype time interaction (F(9, 90) 4.60, of 5-HT blocks the effects of serotonergic but not noradrenergic P Ͻ 0.001). Data are expressed as mean (fmol͞12-␮l sample) ϩ 1 SEM. antidepressants in mice (ref. 16 and O.F.O. and I.L., unpublished observations). In conclusion, these data confirm that mice lacking endog- SSRIs for the serotonergic over the adrenergic system (2–4, enous NE are a useful tool for dissecting the roles of NE in 40), especially fluoxetine, sertraline, and paroxetine (1). In modulating stress and pharmacologically induced behavioral addition, a number of in vivo microdialysis studies have shown and neurochemical changes of antidepressants. These results ͞ that acute and or chronic treatment with the SSRIs fluox- establish a critical role for NE in the acute behavioral effects etine, sertraline, and paroxetine results in increased NE levels of several classes of antidepressant compounds, including the in various brain regions (4–9). Repletion of NE levels with the SSRIs. However, because DbhϪ/Ϫ mice also lack epinephrine, NE precursor L-DOPS restored the behavioral effects of we cannot exclude a role of this catecholamine in the antide- Ϫ Ϫ paroxetine. The reduced sensitivity to SSRIs in Dbh / mice pressant-induced responses. The value of the TST is supported is not due to an alteration in SERT binding or kinetics or any by the ability to predict drugs used clinically as antidepressants other detectable changes in 5-HT or metabolite concentrations (predictive validity) and because the tests may capture some

8190 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0401080101 Cryan et al. Downloaded by guest on September 24, 2021 aspect of the subjective experience of depression associated must evaluate the role of NE in the manifestation of the effects with evolutionary components of thwarted escape persistence of chronic antidepressant drug treatments. or entrapment (25, 48). The positive findings in DbhϪ/Ϫ mice encourages additional research to evaluate the role of NE in We thank Sumitomo Pharmaceuticals, Ltd. (Osaka) for providing L- other behavioral tests sensitive to antidepressants to ensure DOPS, and Dr. Michael Robinson, Director of the Neurochemistry Core the effects are not idiosyncratic to the TST; although our of the Mental Retardation and Developmental Disabilities Research previous data with catecholaminergic antidepressants in the Center at the Children’s Hospital of Philadelphia (HD26979), for analysis of 5-HT content. This work was supported by U.S. Public Health FST (20) strongly suggest that this is not the case. However, Service Grants MH36262 (to I.L.) and MH48125 (to I.L.); National antidepressant drugs must be given chronically before they Institute of Mental Health Grant MH63352 (to S.A.T.); National produce their full clinical effects, and alterations downstream Institute of Neurological Disorders and Stroke Grant NS37722 (to of elevations in neurotransmitter levels likely contribute to S.A.T.); and a Young Investigator Award from the National Alliance for their efficacy in treating depression. Therefore, future studies Research on Schizophrenia and Depression (to S.A.T.).

1. Frazer, A. (1997) J. Clin. Psychiatry 58, Suppl. 6, 9–25. 25. Cryan, J. F. & Mombereau, C. (2004) Mol. Psychiatry 9, 326–357. 2. Stanford, S. C. (1996) Trends Pharmacol. Sci. 17, 150–154. 26. Knobelman, D. A., Hen, R. & Lucki, I. (2001) J. Pharmacol. Exp. Ther. 298, 3. Owens, M. J., Knight, D. L. & Nemeroff, C. B. (2000) Biol. Psychiatry 47, 1083–1091. 842–845. 27. Popik, P. (1999) J. Clin. Psychopharmacol. 19, 4S–22S. 4. Millan, M. J., Gobert, A., Lejeune, F., Newman-Tancredi, A., Rivet, J-M., 28. Buck, K. J. & Amara, S. G. (1994) Proc. Natl. Acad. Sci. USA 91, 12584–12588. Auclair, A. & Peglion, J-L. (2001) J. Pharmacol. Exp. Ther. 298, 565–580. 29. Stanford, S. C. (1995) Pharmacol. Ther. 68, 297–342. 5. Nakayama, K. (2002) Naunyn-Schmiedeberg’s Arch. Pharmacol. 365, 102–105. 30. Valentino, R. J., Curtis, A. L., Page, M. E., Pavcovich, L. A. & Florin-Lechner, 6. Bymaster, F. P., Zhang, W., Carter, P. A., Shaw, J., Chernet, E., Phebus, L., S. M. (1998) Adv. Pharmacol. 42, 781–784. Wong, D. T. & Perry, K. W. (2002) Psychopharmacology 160, 353–361. 31. Page, M. E., Detke, M. J., Dalvi, A., Kirby, L. G. & Lucki, I. (1999) 7. Thomas, D. N., Nutt, D. J. & Holman, R. B. (1998) J. Psychopharmacol. 12, Psychopharmacology 147, 162–167. 366–370. 32. Wong, E. H., Sonders, M. S., Amara, S. G., Tinholt, P. M., Piercey, M. F., 8. Hajos-Korcsok, E., McTavish, S. F. & Sharp, T. (2000) Eur. J. Pharmacol. 407, Hoffmann, W. P., Hyslop, D. K., Franklin, S., Porsolt, R. D., Bonsignori, A., 101–107. et al. (2000) Biol. Psychiatry 47, 818–829. 9. Beyer, C. E., Boikess, S., Luo, B. & Dawson, L. A. (2002) J. Psychopharmacol. 33. Weinshenker, D., White, S. S., Javors, M. A., Palmiter, R. D. & Szot, P. (2002) 16, 297–304. Brain Res. 946, 239–246. 10. Abercrombie, E. D. & Zigmond, M. J. (1989) J. Neurosci. 9, 4062–4067. 34. Ascher, J. A., Cole, J. O., Colin, J. N., Feighner, J. P., Ferris, R. M., Fibiger, 11. Hughes, Z. A. & Stanford, S. C. (1996) Eur. J. Pharmacol. 317, 83–90. H. C., Golden, R. N., Martin, P., Potter, W. Z., Richelson, E. & Sulser, F. (1995) 12. Kostowski, W. (1985) Pol. J. Pharmacol. Pharm. 37, 727–743. J. Clin. Psychiatry 56, 395–401. 13. Redrobe, J. P., Bourin, M., Colombel, M. C. & Baker G. B. (1998) Psycho- 35. Cooper, B. R., Wang, C. M., Cox, R. F., Norton, R., Shea, V. & Ferris, R. M. pharmacology 138, 1–8. (1994) Neuropsychopharmacology 11, 133–141. 14. Esposito, E., Ossowska, G. & Samanin, R. (1987) Eur. J. Pharmacol. 136, 36. Dong, J. & Blier, P. (2001) Psychopharmacology 155, 52–57. 429–432. 37. Reneric, J. P., Bouvard, M. & Stinus, L. (2002) Eur. Neuropsychopharmacol. 12, 15. Cryan, J. F., Page M. E. & Lucki, I. (2002) Eur. J. Pharmacol. 436, 197–205. 159–171. 16. Lucki, I. & O’Leary, O. F. (2004) J. Clin. Psychiatry 65, Suppl. 4, 11–24. 38. Cervo, L., Grignaschi, G., Rossi, C. & Samanin, R. (1991) Eur. J. Pharmacol. 17. Thomas, S. A., Matsumoto, A. M. & Palmiter, R. D. (1995) Nature 374, 196, 217–222. 643–646. 39. Soubrie, P., Martin, P., el Mestikawy, S. & Hamon M. (1987) Brain Res. 437, 18. Thomas, S. A. & Palmiter, R. D. (1997) Cell 91, 583–592. 323–331. 19. Thomas, S. A. & Palmiter, R. D. (1997) Behav. Neurosci. 111, 579–589. 40. Fuller, R. W., Hemrick-Luecke, S. K., Littlefield, E. S. & Audia J. E. (1995) 20. Cryan, J. F., Dalvi, A., Jin, S. H., Hirsch, B. R., Lucki, I. & Thomas, S. A. (2001) Prog. Neuropsychopharmacol. Biol. Psychiatry 19, 135–149. J. Pharmacol. Exp. Ther. 298, 651–657. 41. Sanchez, C. & Hyttel, J. (1999) Cell. Mol. Neurobiol. 19, 467–489. 21. Weinshenker, D., Szot, P., Miller, N. S., Rust, N. C., Hohmann, J. G., Pyati, U., 42. Hughes, Z. A. & Stanford, S. C. (1998) Br. J. Pharmacol. 124, 1141–1148. White, S. S. & Palmiter, R. D. (2001) J. Neurosci. 21, 7764–7769. 43. Baraban, J. M. & Aghajanian, G. K. (1980) Neuropharmacology 19, 355–363. 22. Thomas, S. A., Marck, B. T., Palmiter, R. D. & Matsumoto, A. M. (1998) 44. Bortolozzi, A. & Artigas, F. (2003) Neuropsychopharmacology 28, 421–434. J. Neurochem. 70, 2468–2476. 45. Delgado, P. L. (2000) J. Clin. Psychiatry 61, Suppl. 6, 7–11.

23. Steru, L., Chermat, R., Thierry, B. & Simon, P. (1985) Psychopharmacology 85, 46. Healy, D. & McMonagle, T. (1997) J. Psychopharmacol. 11, S25–S31. NEUROSCIENCE 367–370. 47. Blier, P. (2001) J. Psychiatry Neurosci. 26, S3–S10. 24. Cryan, J. F., Markou, A. & Lucki, I. (2002) Trends Pharmacol. Sci. 23, 238–245. 48. Gilbert, P. & Allan, S. (1998) Psychol. Med. 28, 585–598.

Cryan et al. PNAS ͉ May 25, 2004 ͉ vol. 101 ͉ no. 21 ͉ 8191 Downloaded by guest on September 24, 2021