Antidepressant Drug-Induced Alterations in Neuron-Localized Tumor Necrosis Factor-α mRNA and α₂-Adrenergic Receptor Sensitivity

Materials and Methods

Animals and Drug Administration Schedule.

Male Sprague-Dawley rats (170–230 g; Harlan Sprague-Dawley, Inc., Indianapolis, IN) are used in all experiments. All procedures are performed according to Institutional Animal Care and Use Committee guidelines. Rats are maintained on a 12-h light/dark cycle in lab animal care facility-accredited pathogen-free quarters with water and food ad libitum. Experimental rats are administered twice daily for 1 day (acute) or 14 days (chronic) either the NE uptake blocker and tricyclic antidepressant drug desmethylimipramine (desipramine) hydrochloride (10 mg/kg i.p.; Sigma, St. Louis, MO) or the serotonin uptake inhibitor and antidepressant drug zimelidine dihydrochloride (3 mg/kg i.p.; RBI/Sigma, Natick, MA) dissolved in sterile saline. Control rats receive equal volumes of saline acutely or chronically. The concentrations of zimelidine as well as desipramine administered are based on work in our laboratory demonstrating alterations in α₂-adrenergic receptor-mediated events after administration of these two drugs (Ignatowski and Spengler, 1994;Nickola et al., 2000). Rats are sacrificed by decapitation 12 h after the last injection and selected brain regions are harvested.

Brain Regions Analyzed.

The locus coeruleus contains the majority of noradrenergic nerve cell bodies in the central nervous system with axons projecting to the hippocampus (Grant and Redmond, 1981). Therefore, the hippocampus and a region of the brain stem containing the locus coeruleus are isolated for investigations. The area referred to as the locus coeruleus consists of a section of the brainstem 1 mm thick, which is medial and inferior to the superior cerebellar peduncle, superior to the nucleus of the Vth cranial nerve, and contains the locus coeruleus and other neurons such as the mesencephalic sensory nucleus of the trigeminal nerve.

In Situ Hybridization.

Modifications of methods published byLewis et al. (1985) and Elner et al. (1991) are used. A region of the brainstem containing the nucleus locus coeruleus, as well as the mesencephalic sensory nucleus of the trigeminal nerve, is placed in OCT Tissue-Tek embedding compound (Miles, Elkhart, IN) and snap frozen in liquid nitrogen. Sections are cut 4 μm thick with a Cryo-cut Microtome cryostat, using morphological landmarks as a guide (bregma, −9.68; lateral, +1.5; vertical, −7.5 mm) (Paxinos and Watson, 1996). Sections are mounted on poly(l-lysine)-coated slides (Sigma) and fixed (15 min) in 4°C 4% paraformaldehyde in PBS, rinsed three times in 4°C 70% ethanol, and stored at 4°C in 70% ethanol until analysis. Before hybridization, sections are postfixed (10 min) in 4% paraformaldehyde in PBS (RT). Sections are then covered with 37°C proteinase K (Promega, Madison, WI), 5 μg/ml in 2× SSC, and incubated at 37°C (15 min); rinsed with 2× SSC at RT; acetylated with freshly prepared 0.5% acetic anhydride in 0.1 M tetraethylammonium, pH 7.2 (5 min); and rinsed in 0.5× SSC. Sections are then prehybridized at 42°C (2 h). The prehybridization/hybridization buffer consists of the following reagents with the final concentration in parentheses: deionized formamide (20%; Sigma), 20× SSC (5×), dextran sulfate (5%) (Sigma), transfer RNA (100 μg/ml) (Sigma), salmon sperm DNA (100 μg/ml) (Sigma), Denhardt's solution (5%), and dithiothreitol (10 mM) (Roche Molecular Biochemicals, Indianapolis, IN). Antisense synthetic cDNA oligonucleotides that are specific for rat TNF mRNA (5′-GTC-CCC-CTT-CTC-CAG-CTG-GAA-GAC-TCC-TCC-3′) are 3′-labeled with³⁵S-dCTP (NEN, Boston, MA) using a terminal deoxynucleotidyl transferase (Roche Molecular Biochemicals). Each tissue section is covered with 100 μl of hybridization buffer containing 1 × 10⁶ bound cpm of labeled oligonucleotide. The slides are coverslipped and placed at 42°C overnight in moist storage boxes. Twenty hours later the slides are rinsed twice in 2× SSC (10 min), once with 1× SSC (1 h at 42°C), and then dehydrated with ascending series of ethanol containing 0.3 M NH₄ acetate (5 min). Dried slides are then dipped in NTB-2 photographic emulsion (Eastman Kodak, Rochester, NY). Dried slides are placed in lightproof boxes and stored at 4°C for 7 to 10 weeks. Slides are then developed and fixed (3 min) using D-19 developer:double distilled H₂O (1:1) and Rapid Fix solution A, respectively (Eastman Kodak), and counterstained with hematoxylin and eosin (1 min). Control samples include hybridizing with unlabeled oligonucleotide, sense-TNF oligonucleotide probe, or RNase-pretreated sections (200 μg/ml for 20 min), all of which lack specific signal detection. Specificity of the 30-mer oligonucleotide probe is verified using a 25-mer oligonucleotide probe specific for a different region of TNF mRNA, which localizes in a similar pattern (data not shown). Image analysis is performed on 10 randomly chosen neurons (verified by immunolocalization of neurofilament 200) from the right and left loci. The contrast of the silver grains is enhanced with Wallis Transformation using Confocal Assistant 4.02. Knowledge of the experimental paradigm was blinded to the investigator performing the digital image analysis. The area of the neuron comprised of silver grains is determined using Image Tool 2.0 (developed at the University of Texas Health Science Center, San Antonio, TX, available from the Internet by anonymous FTP from maxrad6.usthsca.edu).

Immunohistochemistry.

Frozen sections of brain stem containing the neurons of the locus coeruleus are collected in series with those used for in situ hybridization and placed on slides coated with Histostick (Accurate, Westbury, NY). Sections are fixed in acetone (10 min), air dried, and stored in slide boxes at −20°C. On the day of staining, slides are air dried, rehydrated in PBS (pH 7.4), and endogenous peroxidase activity is blocked (50 s) using 0.28% periodic acid (Sigma). Nonspecific binding is blocked (40 min) using normal horse serum (1:10), the species in which the secondary antibody is generated. Tissue sections are incubated overnight at RT with monoclonal anti-neurofilament 200 antibody (phosphorylated and nonphosphorylated; Sigma), at a 1:100 dilution that is determined by checkerboard titration. Mouse IgG₁ (Pharminigen, Minneapolis, MN) at the same concentration as anti-neurofilament 200 antibody is the isotype control. Tissue sections of cerebellum serve as the positive control for neurofilament 200 staining (Gotow and Tanaka, 1994). Amplification of the primary antibody is produced using a horse anti-mouse avidin-biotin-complex secondary antibody system (1:400, Vectastain Elite ABC kit; Vector Laboratories, Burlingame, CA). Visualization of the immunolocalization of neuron-specific neurofilament 200 is performed using cobalt-enhanced 3,3-diaminobenzidine tetrachloride (Sigma) as the substrate for the peroxidase-linked secondary antibody. Slides are counterstained (1 min) with nuclear fast red (Sigma), dehydrated through an ascending series of alcohols and Histoclear xylene substitute (National Diagnostics, Atlanta, GA), and coverslipped using Permount (Fisher, Pittsburgh, PA).

Field Stimulation and Superfusion of Hippocampal Slices: Idazoxan and TNF Concentration-Effect Curves.

The release of preloaded³H-labeled neurotransmitter during electrical field stimulation of brain tissue is an established method for studying neurotransmitter release (Starke et al., 1989). Experiments are performed as previously described (Ignatowski and Spengler, 1994). The hippocampi are isolated and transverse slices (0.4 mm thick) are made. The slices are placed in ice-cold Krebs' physiological buffer solution (118 mM NaCl, 4.8 mM KCl, 1.3 mM CaCl₂, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 25 mM NaHCO₃, 10 mM glucose, 0.06 mM ascorbic acid, and 0.03 mM EDTA 0.03). The slices equilibrate at 37°C for 10 min in the buffer, which is saturated with 95% O₂, 5% CO₂. [³H]NE (levo-[ring-2,5,6-³H]-, specific activity 57.3 Ci/mmol; NEN) is then added to the buffer to a final concentration of 330 nM. The slices are incubated for an additional 15 min, and then transferred to 0.1 ml of superfusion chambers (Brandel Suprafusion 1000; Brandel, Gaithersburg, MD). Each tissue slice is placed between two nylon mesh disks placed between two mesh platinum electrodes that provide depolarizing electrical field stimulation to brain tissue. The slices are superfused with buffer for an initial 30-min period to remove extraneuronal tritium. The tissue slices are superfused at a constant rate of 0.5 ml/min to ensure that changes observed in radioactivity represent overflow of [³H]NE from the tissue. Neuronal release of [³H]NE and sensitivity of α₂-adrenergic autoreceptors are studied by applying nine consecutive field stimulations consisting of trains of square-wave pulses (100 mA, 2-ms duration) at 1- or 4-Hz frequency for 2-min periods. After establishment of baseline-stimulated [³H]NE release, concentrations of the α₂-adrenergic receptor antagonist idazoxan (RBI/Sigma) ranging from 1 nM to 1 μM or murine recombinant TNF (Genzyme, Boston, MA) ranging from 10 pg/ml to 10 ng/ml are added to the buffer perfusing the “experimental” superfusion chambers to generate complete idazoxan or TNF concentration-effect curves, respectively. Control chambers are perfused with Krebs' buffer solution only. Aliquots of the superfusate are collected at 2-min intervals and are analyzed by liquid scintillation counting using Ultima Gold scintillation cocktail (Packard, Meriden, CT). Immediately after collection of the last fraction, tissue slices are removed from the chambers and solubilized in 0.2 ml of 1 N NaOH and tritium remaining in the tissue is determined.

The stimulation-evoked release of [³H]NE in excess of spontaneous efflux is determined by the amount of labeled amine in each aliquot, expressed as a percentage of the total amount of radiolabeled amine in the tissue immediately before the onset of field stimulation (fractional release). The labeled amine released in excess of the spontaneous efflux is greater than 90% unmetabolized [³H]NE. The stimulation-evoked release is calculated as the difference between the total release during the period of stimulation less the basal release (overflow). The percentage of response obtained at various concentrations of idazoxan or TNF is obtained by using nonlinear least-squares regression to fit a curve (GraphPad, Prism, GraphPad Software, Inc., San Diego, CA).

Statistics.

Statistical significance of changes in [³H]NE release, presynaptic sensitivity to TNF, and α₂-adrenergic autoreceptor sensitivity are determined using paired Student's t test or ANOVA. Alterations in accumulated TNF mRNA after antidepressant drug administration are determined using Mann-Whitney U test. All results are expressed as mean values ± S.E.M.

Results

In Situ Hybridization for Neuron-Localized mRNA Specific for TNF.

In situ hybridization performed on brain slices from control rats reveals that neuron cell bodies in a region of the brain containing the locus coeruleus and the mesencephalic sensory nucleus of the trigeminal nerve, as described under Materials and Methods, constitutively expresses mRNA specific for TNF (Fig.1). Immunohistochemical detection of neuron-specific neurofilament 200 in corresponding serial tissue sections verifies that the cell bodies that contain TNF mRNA are neurons. Acute desipramine administration (10 mg/kg) significantly decreases neuron-localized grain density (81.6 ± 1.3%, expressed as percentage of control; p < 0.05) (Fig.2). This decrease is particularly apparent over the nucleus of each neuron. Chronic desipramine administration does not significantly alter the total accumulation of neuron-localized grain density compared with control (90.0 ± 0.94%). Accumulation of neuron-localized mRNA for β-actin does not change after desipramine administration (data not shown).

Presynaptic Sensitivity to TNF after Zimelidine Administration.

Results from field stimulation and superfusion of hippocampal slices obtained from control rats demonstrate that TNF inhibits [³H]NE overflow in a concentration-dependent manner with a maximum response of 30.9 ± 2.7% (Fig. 3). As we have previously demonstrated (Ignatowski and Spengler, 1994), selective in vitro blockade of α₂-adrenergic receptors with idazoxan (10⁻⁷ M) results in an upward shift in the TNF concentration-effect curve (60.1 ± 5.2% maximum inhibition; p < 0.05), demonstrating that TNF-induced inhibition of [³H]NE overflow is tempered by α₂-adrenergic receptor activation. Acute (1-day) administration of zimelidine (3 mg/kg) to rats does not significantly alter the maximum response of TNF-mediated regulation of [³H]NE overflow compared with control, either in the absence (33.3 ± 3.7% inhibition) or presence (56.9 ± 14.7% inhibition) of idazoxan in vitro. Interestingly, chronic (14-day) administration to rats of zimelidine results in a transformation or functional reversal in TNF-mediated regulation of [³H]NE overflow, such that TNF facilitates/potentiates [³H]NE overflow (288.0 ± 18.0% maximum response; p < 0.001). In vitro α₂-adrenergic receptor blockade after chronic zimelidine administration restores TNF-mediated inhibition of [³H]NE overflow to levels comparable to control values (43.0 ± 6.4% maximum response; N.S.) obtained in the absence of idazoxan. The concentration of TNF necessary to reach 50% maximum inhibition or potentiation of [³H]NE overflow (EC₅₀) does not significantly differ between paradigms (data not shown).

The functional reversal in TNF-mediated regulation of [³H]NE overflow after chronic zimelidine administration is similar to that demonstrated after chronic desipramine administration (maximum response = 288.0 ± 18.0% potentiation and 41.3 ± 5.7% potentiation, respectively; Fig. 4). The similarities demonstrated by these two representatives of diverse classes of antidepressant drugs may represent a potentially significant mechanism of antidepressant drug action; i.e., to transform α₂-adrenergic autoreceptors that mediate TNF-induced regulation of NE release.

Fractional Release of [³H]NE from Hippocampal Slices from Control Rats and Rats Administered Zimelidine or Desipramine.

The percentage of [³H]NE released in excess of spontaneous efflux at both 1-Hz (s1) and 4-Hz (s2) frequencies of stimulation is presented in Figs. 5 and7. Field stimulation of hippocampal slices from control rats and rats administered antidepressant drugs demonstrates a frequency-dependent increase in [³H]NE release. In vitro α₂-adrenergic receptor blockade (10⁻⁷ M idazoxan) significantly increases the fractional release of [³H]NE from hippocampal slices isolated from control rats at both 1-Hz (0.55 ± 0.08 versus 1.39 ± 0.47% in the presence of idazoxan;p < 0.05) and 4 Hz (1.64 ± 0.19 versus 4.52 ± 1.61% in the presence of idazoxan; p < 0.001). Chronic zimelidine administration does not significantly alter [³H]NE release at 1-Hz (0.66 ± 0.07%) or 4-Hz (1.84 ± 0.22%) frequencies of stimulation compared with control. However, although in vitro α₂-adrenergic receptor blockade with idazoxan after chronic zimelidine administration does not significantly alter [³H]NE fractional release at 1 Hz (1.02 ± 0.20%; N.S.), the fraction release of [³H]NE is increased at the higher frequency of stimulation (4 Hz, 3.40 ± 0.55%, p < 0.05).

α₂-Adrenergic Autoreceptor Sensitivity after Chronic Zimelidine or Desipramine Administration.

As previously presented (Ignatowski et al., 1996), the α₂-adrenergic antagonist idazoxan (10⁻⁷ M) potentiates [³H]NE overflow in a concentration-dependent manner from control hippocampal slices with a maximum response of 441.3 ± 61.3% (Fig. 6). Chronic zimelidine administration (3 mg/kg) does not significantly change the maximum response (563.3 ± 53.1%) or the EC₅₀ (0.34 ± 0.06 μM for control versus 0.52 ± 0.10 μM; N.S.) of the idazoxan concentration-effect curve. However, chronic administration of desipramine (10 mg/kg) results in a significant downward shift in the idazoxan concentration-effect curve (99.0 ± 23.8% maximum potentiation;p < 0.05), with no change in the EC₅₀ value of idazoxan (0.34 ± 0.06 μM for control versus 0.14 ± 0.04 μM; N.S.).

Discussion

The present study has been undertaken to investigate the reciprocally permissive relationship between the biological effects of TNF and α₂-adrenergic autoreceptor activation after antidepressant drug administration. Alterations in α₂-adrenergic autoreceptors are hypothesized to be a target of action of antidepressant drugs, as demonstrated by decreased α₂-adrenergic receptor density and sensitivity after antidepressant drug administration (Crews and Smith, 1980; Garcia-Sevilla et al., 1981; Garcia-Sevilla and Zubieta, 1986). The results of the present study extend this hypothesis to include α₂-adrenergic autoreceptors because they are associated with the neuromodulatory functions of TNF, which is synthesized de novo in neurons of the brain. Collectively, these data demonstrate a possible common mechanism of action of antidepressant drugs that involves alterations in the dynamic equilibrium between TNF and α₂-adrenergic autoreceptor activation.

Neurons of the hippocampus and locus coeruleus constitutively express TNF protein in a pattern that suggests de novo TNF production (Ignatowski et al., 1997). In situ hybridization reveals that TNF mRNA is constitutively expressed within neurons in a region of the brain stem containing the locus coeruleus (Fig. 1, A and D). Therefore, it follows that neuron-localized TNF protein can be produced and possibly released by neurons (Bartfai and Schultzberg, 1993; Hopkins and Rothwell, 1995). The profound loss of accumulated neuron-localized TNF mRNA (Figs. 1, B and E, and 2) and protein (Ignatowski et al., 1997) after acute desipramine administration may represent transport of existing mRNA out of the soma, as well as increased mRNA translation followed by protein loss, and/or degradation of accumulated TNF mRNA. With continued administration of desipramine (14 days), the loss of neuron-localized TNF protein is maintained (Ignatowski et al., 1997), yet the density of accumulated TNF mRNA localized over the nucleus is restored comparable to control (Figs. 1, C and F, and 2). These data imply that TNF protein is degraded, released, or at levels undetectable by immunohistochemical analysis. Alternatively, these data suggest that TNF mRNA is not translated, which is similar to the regulatory mechanisms that govern the expression of the proinflammatory cytokine interleukin-1 (Schindler et al., 1990; Stevenson et al., 1992).

Neuron-localized TNF protein is decreased after both acute and chronic desipramine administration (Ignatowski et al., 1997). However, a transformation in presynaptic noradrenergic sensitivity to TNF (Ignatowski and Spengler, 1994), and sustained changes in α₂-adrenergic receptor density and sensitivity only occur after chronic antidepressant drug administration (Crews and Smith, 1978, 1980; Garcia-Sevilla and Zubieta, 1986). These data suggest that the continued loss of neuron-localized TNF protein (Ignatowski et al., 1997), occurring before and during the time of the therapeutic efficacy of antidepressant drugs, is paramount in governing alterations in the dynamic equilibrium between TNF and α₂-adrenergic receptor activation, which modulate adrenergic neurotransmission and the expression of mood.

The ability of TNF to inhibit [³H]NE overflow from field-stimulated, superfused hippocampal slices from control rats in a concentration-dependent manner is significantly enhanced during in vitro α₂-adrenergic receptor blockade with idazoxan (Fig. 3) (Ignatowski and Spengler, 1994). These data demonstrate that TNF-mediated inhibition of [³H]NE release occurs in functional association with α₂-adrenergic autoreceptor activation, and suggest that α₂-adrenergic autoreceptor activation tempers or conditions presynaptic sensitivity to TNF. Acute zimelidine administration to rats does not alter the ability of TNF or the α₂-adrenergic autoreceptors that temper the ability of TNF to regulate [³H]NE overflow, whereas chronic administration of zimelidine transforms or functionally reverses presynaptic sensitivity to TNF. Additionally, the restoration in the ability of TNF to inhibit [³H]NE overflow during in vitro α₂-adrenergic receptor blockade demonstrates that the transformation occurs in functional association with α₂-adrenergic autoreceptor activation.

It is interesting to compare these data to our previous results that demonstrate the effects of chronic desipramine administration on presynaptic sensitivity to TNF (Ignatowski and Spengler, 1994). Similar to acute zimelidine administration, acute administration of desipramine does not alter presynaptic sensitivity to TNF. However, chronic desipramine administration, similar to zimelidine administration, dramatically transforms α₂-adrenergic receptor-modulated presynaptic sensitivity to TNF (Fig. 4). Collectively, these data demonstrate that two different classes of drugs efficacious in the treatment of depression similarly alter α₂-adrenergic receptor-mediated presynaptic sensitivity to TNF. This transformation occurs in a time frame corresponding to the therapeutic effects of antidepressant drugs, in a region of the brain associated with the expression of mood (Danysz et al., 1985; Chan-Palay and Asan, 1989).

Alterations in α₂-adrenergic autoreceptors that directly regulate NE release are investigated after chronic desipramine or zimelidine administration (Fig. 6). The concentration-dependent potentiation of [³H]NE release from field-stimulated hippocampal slices in the presence of the α₂-adrenergic receptor antagonist idazoxan represents blockade of α₂-adrenergic autoreceptors that inhibit [³H]NE overflow. Chronic administration of zimelidine does not alter idazoxan-induced potentiation of [³H]NE overflow or fractional release (Figs. 6 and 7), suggesting that chronic zimelidine administration to rats does not alter the density or sensitivity of α₂-adrenergic autoreceptors, or the amount of endogenous agonist (NE) in the vicinity of these receptors. However, decreased maximum response of the idazoxan concentration-effect curve after chronic desipramine administration represents either a decrease in the amount of endogenous agonist in the vicinity of α₂-adrenergic autoreceptors or a decrease in the density or sensitivity of these receptors (Fig. 6). An increase in the fractional release of [³H]NE at 1- and 4-Hz frequencies of stimulation (Ignatowski and Spengler, 1994) demonstrates that the downward shift in the idazoxan concentration-effect curve is due to a decrease in the sensitivity or density of α₂-adrenergic autoreceptors (Fig.7). Decreased α₂-adrenergic autoreceptor sensitivity in the brain after chronic antidepressant drug administration concurs with antidepressant drug-induced decreases in α₂-adrenergic receptor sensitivity in regions of the body other than the brain (Crews and Smith, 1980; Garcia-Sevilla et al., 1981; Garcia-Sevilla and Zubieta, 1986). In vitro α₂-adrenergic receptor blockade increases fractional release of [³H]NE from control field-stimulated hippocampal slices (Fig. 5). Although chronic zimelidine administration transforms α₂-adrenergic autoreceptors that temper TNF-mediated regulation of [³H]NE overflow (Fig. 3), the fractional release of [³H]NE at 1- and 4-Hz frequencies of stimulation is not altered compared with control (Fig. 5). However, the inability of in vitro α₂-adrenergic receptor blockade to alter fractional release of [³H]NE at 1 Hz demonstrates that chronic zimelidine administration does in fact alter α₂-adrenergic autoreceptor function, although in a different manner to that observed after chronic desipramine administration (Crews and Smith, 1980; Garcia-Sevilla and Zubieta, 1986; Ignatowski and Spengler, 1994).

One cannot rule out the possibility that the differential effects of chronic zimelidine administration on α₂-adrenergic autoreceptors that mediate presynaptic sensitivity to TNF (Fig. 3) versus those that directly regulate [³H]NE release (Fig. 7) may represent activation of two functionally different types of α₂-adrenergic receptors. Administration of different concentrations of zimelidine may yield different results because the effects of chronic antidepressant drug administration on α₂-adrenergic autoreceptor sensitivity and density are concentration-dependent (Smith et al., 1981). Additionally, the cellular effects of TNF and α₂-adrenergic receptor activation are reciprocally permissive. Therefore, the loss of neuron-associated TNF after chronic antidepressant drug administration (Ignatowski et al., 1997) (Figs. 1 and 2) and during the process of superfusion analysis (which washes away neurotransmitters and neuromodulators from nerve terminals that no longer receive input from the noradrenergic cell bodies of the locus coeruleus), alters the dynamic equilibrium between TNF and α₂-adrenergic receptor activation. The addition of TNF in vitro during superfusion of isolated hippocampal slices may be necessary to reestablish the dynamic equilibrium, and reveal a common mechanism of action of zimelidine and desipramine administration to potentiate [³H]NE overflow at a time when both drugs exert their therapeutic effects (Fig. 3).

These investigations confirm that TNF is a neuromodulator in the brain and along with alterations in the α₂-adrenergic autoreceptor response is involved in a mechanism elicited by at least some antidepressant drugs (Crews and Smith, 1980; Smith et al., 1981;Garcia-Sevilla and Zubieta, 1986; Ignatowski and Spengler, 1994;DePaermentier et al., 1997). Additionally, the present study reveals a unique common mechanism of action for two different types of antidepressant drugs, such that they both transform α₂-adrenergic autoreceptors that mediate the ability of TNF to regulate NE release. These results offer insight into the mechanisms of the delayed, yet persistent therapeutic effects of antidepressant drugs, and may aid in the development of drugs more efficacious in the treatment of depression.