Noradrenaline activation of hippocampal dopamine D₁ receptors promotes antidepressant effects

Significance

Dopamine receptors in the brain are important targets for the therapeutic treatment of psychiatric disorders. Dopamine receptors are generally thought to be activated by dopamine. In the present study, however, we reveal that noradrenaline can also robustly activate dopamine D₁ receptors in the mouse hippocampus. Noradrenaline-activated D₁ receptor signaling was highly sensitive to the neuronal activity and experience of mice. Chronic stress and voluntary exercise synergistically augmented noradrenaline–D₁ receptor signaling. This augmented noradrenaline–D₁ receptor signaling promoted the induction of hippocampal neuronal plasticity by an antidepressant drug acting on the noradrenergic system. Our results suggest that noradrenaline–D₁ receptor signaling increases the efficacy of antidepressant treatment and therefore can be a unique therapeutic target for augmenting antidepressant medication.

Abstract

Dopamine D₁ receptors (D₁Rs) in the hippocampal dentate gyrus (DG) are essential for antidepressant effects. However, the midbrain dopaminergic neurons, the major source of dopamine in the brain, only sparsely project to DG, suggesting possible activation of DG D₁Rs by endogenous substances other than dopamine. We have examined this possibility using electrophysiological and biochemical techniques and found robust activation of D₁Rs in mouse DG neurons by noradrenaline. Noradrenaline at the micromolar range potentiated synaptic transmission at the DG output and increased the phosphorylation of protein kinase A substrates in DG via activation of D₁Rs and β adrenergic receptors. Neuronal excitation preferentially enhanced noradrenaline-induced synaptic potentiation mediated by D₁Rs with minor effects on β-receptor–dependent potentiation. Increased voluntary exercise by wheel running also enhanced noradrenaline-induced, D₁R-mediated synaptic potentiation, suggesting a distinct functional role of the noradrenaline–D₁R signaling. We then examined the role of this signaling in antidepressant effects using mice exposed to chronic restraint stress. In the stressed mice, an antidepressant acting on the noradrenergic system induced a mature-to-immature change in the DG neuron phenotype, a previously proposed cellular substrate for antidepressant action. This effect was evident only in mice subjected to wheel running and blocked by a D₁R antagonist. These results suggest a critical role of noradrenaline-induced activation of D₁Rs in antidepressant effects in DG. Experience-dependent regulation of noradrenaline–D₁R signaling may determine responsiveness to antidepressant drugs in depressive disorders.

Hippocampal dopamine D₁ receptors (D₁Rs) have been implicated in regulation of learning and memory (1, 2). Accumulating evidence also suggests a critical involvement of hippocampal D₁Rs, especially those expressed in the dentate gyrus (DG) neurons, in antidepressant action (3–7). D₁R activation induces robust synaptic potentiation at the synapse formed by the mossy fiber (MF) axons of the dentate granule cells (GCs) (8). Pharmacological and physical antidepressant treatments augment this D₁R-dependent synaptic potentiation (3, 4) and increase D₁R expression in the DG (4, 5). Furthermore, D₁R activation contributes to antidepressant-induced behavioral changes and neuronal plasticity in the DG (5). Midbrain dopaminergic neurons are the major source of brain dopamine and have been thought to be responsible for activation of dopamine receptors in the DG (9). However, projection of these dopaminergic neurons to the hippocampus including DG is relatively sparse (10–13). In addition, the dopamine content in the hippocampus is much lower than other monoamines (14–16). Therefore, dopamine may mediate only a part of dopamine-receptor–dependent hippocampal functions, and endogenous substances other than dopamine may also be involved.

Noradrenergic fibers from the locus coeruleus abundantly project to the hippocampus (13, 17). The noradrenaline content in the hippocampus is 10- to 100-fold higher than dopamine (14–16). Previous studies have shown that optogenetic manipulation of the noradrenergic projection to the hippocampus modulated performance of learning tasks in a D₁R-dependent manner (13, 18, 19). It is likely that noradrenaline can activate D₁Rs as an endogenous agonist, thereby regulating hippocampal functions. However, none of the previous studies focused on this possibility, because it is generally believed that noradrenaline activates D₁Rs at tens of micromolar concentrations (20), a range that may exceed physiological levels.

In the present study, we examined whether noradrenaline could modulate hippocampal neuronal signaling by activating D₁Rs using electrophysiological and biochemical techniques, with a particular focus on the dentate GCs and their MF synaptic output. We found that noradrenaline at the micromolar range causes robust changes in neuronal transmission in all hippocampal subregions examined and increases protein phosphorylation in the DG via D₁R activation. Notably, noradrenaline–D₁R signaling at the MF synapse was activated by submicromolar noradrenaline and substantially enhanced in an activity- or experience-dependent manner. A possible involvement of this noradrenaline–D₁R signaling in antidepressant effects was further examined. Antidepressant treatments can change the phenotype of mature dentate GCs into an immature-like state, a process characterized as “dematuration” of GCs (5, 21–24). We found that experience-dependent augmentation of noradrenaline–D₁R signaling promotes the induction of dematuration by a noradrenergic antidepressant. Our results suggest that D₁R signaling activated by noradrenaline plays a critical role in regulating the antidepressant efficacy.

Results

Noradrenaline Activates Hippocampal Dopamine D₁ Receptors at Micromolar Concentrations.

Dopamine D₁Rs regulate synaptic transmission and neuronal excitability in the somatodendritic region and at the axonal output of the dentate GCs (8, 25). We first examined the effects of noradrenaline at the MF axon to CA3 pyramidal cell synapse (Fig. 1A). Dopamine potentiates MF synaptic transmission by activating presynaptic D₁Rs (4, 8). Previous studies have shown that noradrenaline had little effect on basal MF synaptic transmission in rat hippocampal slices (26) or inhibited synaptic transmission via α adrenergic receptors in hippocampal slice cultures (27). We found that noradrenaline (10 μM) potentiated excitatory postsynaptic potentials (EPSPs) at the MF synapse in mouse hippocampal slices, with only minor effects on presynaptic fiber volleys (Fig. 1 B and C). This synaptic potentiation was accompanied by a decrease in a presynaptic form of synaptic facilitation induced by repetitive stimulation (Fig. 1D), suggesting that noradrenaline potentiated MF synaptic transmission via presynaptic mechanisms. Noradrenaline-induced synaptic potentiation was partially blocked by an antagonist of β adrenergic receptors, but not of α adrenergic receptors (Fig. 1 E and F). Antagonists of dopamine D₁Rs also partially blocked noradrenaline-induced synaptic potentiation (Fig. 1 E and F). The D₁-antagonist–resistant potentiation was suppressed by the broad β receptor antagonist propranolol or the β₁-receptor–specific antagonist CGP20712 (Fig. 1 G and H), indicating that noradrenaline-induced potentiation is mediated by dopamine D₁ and β₁ adrenergic receptors. Noradrenaline had minimal effects at low micromolar concentrations (Fig. 1C). However, this is due to uptake of noradrenaline into noradrenergic nerve terminals in the hippocampus, because addition of the noradrenaline transporter inhibitor nisoxetine unveiled robust D₁R-dependent potentiation by low micromolar noradrenaline (Fig. 1I). These results revealed that noradrenaline can activate dopamine D₁Rs at the MF output of the GCs at the micromolar concentration range. As compared with noradrenaline, dopamine had larger potentiating effects on both EPSPs and fiber volleys (Fig. 1J). Since dopamine-induced potentiation at the MF synapse is almost exclusively mediated by D₁Rs (4, 8), dopamine is more effective in activating D₁Rs than noradrenaline (see also Fig. 3F).

Fig. 1.

Activation of D₁ receptors by noradrenaline potentiates mossy fiber synaptic transmission in the hippocampus. (A) Schematic diagram of hippocampal excitatory circuit highlighting mossy fiber (MF)–CA3 connection. (B) Effects of noradrenaline (NA) applied at the bar on presynaptic fiber volley (FV) and EPSPs at MF–CA3 synapse. Sample traces show averaged field potentials before and during NA application (scale bars: 10 ms, 0.2 mV). (C) Dependence of synaptic potentiation on NA concentrations. (D) Decrease in triple-pulse facilitation at 200-ms intervals by NA. BL: baseline. Paired t test (t₅ = 11.77, ****P < 0.0001). (E) NA-induced synaptic potentiation in D₁ antagonist SKF83566 (SKF) (200 nM), β antagonist propranolol (Pro) (10 μM) or α antagonist phentolamine (Phento) (20 μM). (F) Summary of antagonist effects on NA-induced synaptic potentiation. SCH: SCH23390 (D₁ antagonist, 50 nM). One-way ANOVA (F_4,33 = 10.2, P < 0.0001) followed by Dunnett's test (**P = 0.0085, ***P = 0.0009, ****P = 0.0001). (G) Block of SKF-resistant potentiation by β₁ antagonist CGP20712 (CGP) (100 nM). (H) Summary of β antagonist effects on SKF-resistant potentiation. One-way ANOVA (F_2,11 = 35.66, P < 0.0001) followed by Dunnett's test (****P = 0.0001). (I) D₁-receptor–dependent synaptic potentiation by low micromolar NA in presence (n = 6) and absence (n = 3) of nisoxetine (1 μM). (J) Dependence of FV and EPSP potentiation on dopamine (DA) concentrations. The number of data is shown in the graph in C and J. Data are presented as means ± SEM in all figures with or without individual values.

Fig. 3.

Activity-dependent enhancement of noradrenaline–D₁ receptor signaling. (A) Effects of NA in control mice (CNT) and mice treated with three times of ECT (ECTx3) in normal saline or in antagonists. (B) Summary of data shown in A. Two-way ANOVA (ECT effect, F_1,35 = 83.22, P < 0.0001; drug effect, F_2,35 = 42.98, P < 0.0001; interaction, F_2,35 = 10.13, ^###P = 0.0003) followed by Tukey's test (NS: not significant, **P = 0.0083, ***P = 0.0002, ****P < 0.0001). (C) Effects of ECTx3 on reduction of triple-pulse facilitation by NA (t₇ = 3.644, **P = 0.082). (D) Effects of β receptor antagonists on SKF-resistant component of NA-induced potentiation in ECT-treated mice. (E) Summary of β antagonist effects on SKF-resistant component of NA-induced potentiation. CGP (100–200 nM), ICI: ICI118551 (β₂ antagonist, 100–200 nM). One-way ANOVA (F_3,21 = 17.37, P < 0.0001) followed by Bonferroni's test (**P < 0.01, ***P = 0.0001). (F) Dependence of D₁-mediated synaptic potentiation on NA and DA concentrations in naive and ECT-treated mice. The DA data in naive mice are the same as those in Fig. 1J. (G) Dependence of D₁-mediated potentiation on NA concentrations in nisoxetine. The number of data is shown in the graph in C, F, and G.

We next examined the effects of noradrenaline at the somatodendritic region of the GCs. Population spikes (PSs) were evoked by stimulating the perforant path input to the DG (SI Appendix, Fig. S1A). Noradrenaline at 10 μM strongly potentiated PSs and had only small potentiating effects on EPSPs (SI Appendix, Fig. S1B). The β and α₁ antagonist labetalol suppressed potentiation of EPSPs and reduced PS potentiation (SI Appendix, Fig. S1 C–F). Propranolol had similar effects (SI Appendix, Fig. S1 D and F), confirming an involvement of β receptors. The D₁ antagonist SKF83566 applied alone had no significant effects (SI Appendix, Fig. S1 C–F). In the presence of labetalol, however, SKF83566 suppressed potentiation of PSs without affecting EPSPs (SI Appendix, Fig. S1 C–F). In the presence of SKF83566 and labetalol, noradrenaline caused inhibition of PSs that was suppressed by the α receptor antagonist phentolamine, and block of α and β receptors isolated D₁R-mediated potentiation of PSs that was abolished by D₁R antagonists (SI Appendix, Fig. S1 C, D, G, and H). The D₁R-mediated potentiation of PSs was observed along the longitudinal axis of the hippocampus with no clear differences between dorsal and ventral regions (SI Appendix, Fig. S1I). These results indicate that activation of D₁Rs by noradrenaline potentiates PSs in the GCs but has no detectable effects on EPSPs. In the CA1 region, noradrenaline potentiated PSs without affecting EPSPs (SI Appendix, Fig. S2 A and B). This PS potentiation was largely suppressed by the β antagonist (SI Appendix, Fig. S2 C and D). As in the DG, SKF83566 significantly suppressed potentiation of PSs only in the presence of the β antagonist, revealing α-receptor–dependent inhibition of PSs (SI Appendix, Fig. S2D). Isolated D₁R-dependent potentiation was detected in the presence of α and β receptor antagonists (SI Appendix, Fig. S2 C and D). Taken together, these results indicate that spike generation in both DG GCs and CA1 pyramidal cells is facilitated by noradrenaline via activation of D₁ and β receptors and inhibited via activation of α receptors. The intact β receptor functioning masked the effect of the D₁R antagonist on noradrenaline-induced potentiation of PSs, suggesting an overlap or redundancy in signaling pathways between D₁ and β receptors (see Discussion). In addition to the electrophysiological studies, we conducted a biochemical assay of D₁R activation by noradrenaline. Noradrenaline strongly increased the phosphorylation of downstream substrates of protein kinase A signaling, GluA1, DARPP32, and ERK2, in slices of the hippocampal DG, but not of the striatum (Fig. 2 A, C, and E). The increased phosphorylation in the DG was blocked by addition of propranolol and SKF83566 (Fig. 2 B, D, and F). However, neither propranolol nor SKF83566 alone had significant effects, again suggesting the overlap in the signaling pathway. Taken together, these results indicate that noradrenaline at the micromolar range activates dopamine D₁ receptors expressed in the hippocampus, but not in the striatum.

Fig. 2.

D₁-receptor–dependent protein phosphorylation by noradrenaline in the dentate gyrus. (A) Effects of NA on GluA1 phosphorylation in dentate gyrus (DG) (n = 7 for each point) and striatum (n = 4). Typical immunoblots for detection of phospho-Ser845 (left) and quantified data (right). One-sample t test (**P = 0.0019, ****P < 0.0001, compared with 1). (B) Effects of SKF83566 (500 nM) and propranolol on NA-induced GluA1 phosphorylation. Typical immunoblots (top) and quantified data (bottom). Two-way ANOVA (SKF effect, F_1,32 = 19.28, P = 0.0001; Pro effect, F_1,32 = 22.07, P < 0.0001) followed by Tukey's test (**P < 0.005, ****P < 0.0001). (C) Effects of noradrenaline on DARPP32 phosphorylation (phospho-Thr34) in slices of DG and striatum. One-sample t test (**P < 0.005). DG: n = 6, striatum: n = 4. (D) Antagonist effects on NA-induced phosphorylation of DARPP32 in DG. Two-way ANOVA (SKF effect, F_1,32 = 8.549, P = 0.0063; Pro effect, F_1,32 = 9.497, P = 0.0042) followed by Tukey's test (***P = 0.0006). (E) Effects of noradrenaline on ERK2 phosphorylation in slices of DG and striatum. One-sample t test (*P < 0.05, **P < 0.01). DG: n = 7, striatum: n = 4. (F) Antagonist effects on NA-induced phosphorylation of ERK2 in DG. Two-way ANOVA (SKF effect, F_1,32 = 11.65, P = 0.0018) followed by Tukey's test (*P = 0.0123, **P = 0.0025). Phospho-proteins were normalized to total proteins and then data were normalized to values obtained with time 0 (A, C, and E) or control without noradrenaline (B, D, and F).

Activity-Dependent Augmentation of Noradrenaline–D₁R Signaling.

A unique feature of D₁R signaling at the MF synapse is marked augmentation by neuronal excitation (4). To further characterize the D₁R signaling activated by noradrenaline, we next examined the effect of electroconvulsive treatment (ECT) that causes massive neuronal excitation and consistently augments dopamine-induced synaptic potentiation at the MF synapse (Fig. 3F) (4). In mice treated with ECT, noradrenaline-induced potentiation at the MF synapse was significantly augmented (Fig. 3 A and B). The reduction of synaptic facilitation by noradrenaline was also augmented by ECT, suggesting presynaptic mechanisms in the effect of ECT (Fig. 3C). Unexpectedly, ECT also changed the sensitivity to the antagonists. While the D₁R antagonist SKF83566 strongly reduced noradrenaline-induced potentiation in both control and ECT-treated mice, the β antagonist propranolol applied alone reduced the synaptic potentiation only in control mice (Fig. 3 A and B). In ECT-treated mice, however, propranolol suppressed the SKF83566-resistant potentiation (Fig. 3E), suggesting that the effect of propranolol in the absence of SKF83566 was masked by intact D₁R functioning. These results suggest that the β receptor signaling overlaps with a part of the D₁R signaling at the MF synapse only after ECT. In addition, the overlapping β receptor signaling was partly mediated by β₂ receptors (Fig. 3 D and E). Therefore, ECT reorganized D₁- and β-receptor–dependent signaling at the MF synapse. Since synthetic β receptor agonists potentiated the MF synaptic transmission via β₂ receptors in naive mice (SI Appendix, Fig. S3), ECT appeared to increase the sensitivity of β₂ receptors to noradrenaline. A comparison of the effects of noradrenaline and dopamine revealed that noradrenaline was more effective in activating D₁Rs after ECT, and there was an apparent increase in potency of noradrenaline (Fig. 3F). In the presence of the noradrenaline uptake inhibitor, even submicromolar noradrenaline was able to activate D₁Rs after ECT (Fig. 3G), excluding a possibility that altered noradrenaline uptake caused the apparent increase in potency. These results indicate that ECT facilitated noradrenaline–D₁R signaling at the MF synapse, possibly via the reorganization of the D₁ and β receptor signaling, rather than simply augmenting D₁R-dependent synaptic modulation. In the DG, ECT did not significantly increase D₁R-dependent potentiation of PSs, although it changed the time course of potentiation (SI Appendix, Fig. S4A). In CA1, ECT tended to reduce D₁R-dependent potentiation of PSs (SI Appendix, Fig. S4B). Therefore, the activity-dependent facilitation of the noradrenaline–D₁R signaling is specific to the MF system.

Interaction between D₁ and β Receptors.

Above results showed the significant overlap between D₁ and β receptor signaling. G-protein–coupled receptors, including dopamine receptors, can form a heteromeric complex (28, 29). Heteromeric D₁ and β receptors may be a molecular basis for such overlap in signaling. To examine formation of D₁–β heteromers, we performed a coimmunoprecipitation assay (Fig. 4A). FLAG-tagged D₁Rs were coexpressed with HA-tagged β₁ or β₂ receptors in HEK293T cells. After immunoprecipitation with a FLAG antibody, blotting with an HA antibody detected β₁ and β₂ receptors coprecipitated with D₁Rs (Fig. 4B). Furthermore, interaction between native D₁ and β receptors was examined using whole hippocampal tissue lysates (Fig. 4D). After immunoprecipitation with a β₁ receptor antibody, a D₁R antibody detected coprecipitated D₁R in control and ECT-treated mice (Fig. 4E and SI Appendix, Fig. S5B). There was no clear difference in the D₁–β₁ coprecipitation between control and ECT-treated mice, while ECT increased D₁R expression levels (SI Appendix, Fig. S5A). These results suggest that D₁ and β receptors can form the heteromeric receptor complex. In HEK293T cells expressing D₁Rs alone, noradrenaline increased cAMP production at the micromolar range (Fig. 4C). Therefore, the complex formation is not required for noradrenaline responsiveness and may regulate efficacy and/or potency of D₁R activation by noradrenaline (see Discussion).

Fig. 4.

Interaction between D₁ and β receptors. (A) Schematic diagram showing experimental design of coimmunoprecipitation assay using tagged receptors. IP: immunoprecipitation, IB: immunoblot. (B) Coimmunoprecipitation of FLAG-tagged D₁ and HA-tagged β₁ or β₂ receptors expressed in HEK293T cells. (C) NA-induced cAMP production in HEK293T cells expressing EGFP or D₁ receptors (n = 2 each). (D) Experimental design of coimmunoprecipitation assay of native receptors. (E) Coimmunoprecipitation of hippocampal D₁ and β₁ receptors in control and mice treated with 11 times of ECT (ECTx11). IgG: immunoglobulin G. Experiments were repeated three times with similar results (SI Appendix, Fig. S5B).

Experience-Dependent Augmentation of Noradrenaline–D₁R Signaling Facilitates Antidepressant Effects.

The noradrenergic system has been implicated in stress-induced alterations of the central nervous system and antidepressant effects (30), and D₁Rs in the dentate GCs contribute to antidepressant effects (5, 7). Finally, we examined an involvement of the noradrenaline–D₁R signaling in antidepressant effects, focusing on possible influence of stress on this signaling. Mice were subjected to 4 wk of restraint stress (Fig. 5A). Noradrenaline-induced MF synaptic potentiation mediated by D₁Rs was recorded in the presence of propranolol. Chronic restraint stress had no significant effect on this noradrenaline–D₁R signaling at the MF synapse (Fig. 5C). On the other hand, increased voluntary exercise by wheel running (Fig. 5A), which often causes neuronal changes opposite to those induced by stress and has antidepressant-like effects (31), significantly increased the noradrenaline–D₁R signaling (Fig. 5 B and C). Low micromolar noradrenaline was able to induce significant D₁R-dependent synaptic potentiation in mice subjected to wheel running (SI Appendix, Fig. S6A). Unexpectedly, chronic stress enhanced the effect of wheel running on the noradrenaline–D₁R signaling (Fig. 5 B and C). Chronic stress combined with wheel running enhanced the D₁R-antagonist–sensitive component of noradrenaline-induced synaptic potentiation (SI Appendix, Fig. S7), confirming the enhancing effect on the noradrenaline–D₁R signaling. The basal synaptic efficacy was not affected by chronic stress or wheel running (Fig. 5D). In addition, chronic stress combined with wheel running had no significant effect on serotonin 5-HT₄-receptor–dependent synaptic modulation that shares the intracellular signaling pathway with the D₁R-dependent modulation (32) (Fig. 5 E and F). These results indicate that chronic stress and voluntary exercise synergistically and specifically enhance the noradrenaline–D₁R signaling.

Fig. 5.

Synergistic augmentation of noradrenaline–D₁ receptor signaling by chronic stress and exercise. (A) Timeline of chronic restraint stress and increased voluntary exercise by wheel running (WR). (B) Effects of chronic restraint stress and WR on NA-induced MF synaptic potentiation mediated by D₁ receptors. Recordings were made in the presence of propranolol. Sample traces show averaged field potentials before and during NA application (scale bars: 10 ms, 0.2 mV). (C) Summary of effects of stress and WR on D₁-mediated potentiation. Two-way ANOVA (stress effect, F_1,23 = 35.29, P < 0.0001; WR effect, F_1,23 = 122.9, P < 0.0001; interaction, F_1,23 = 21.36, P = 0.0001) followed by Tukey's test (**P = 0.0012, ****P < 0.0001). (D) Lack of effects of stress and WR on the basal MF synaptic efficacy assessed by EPSP to fiber volley ratios. The number of data is shown in the graph. (E) Effects of stress and WR on serotonin-induced MF synaptic potentiation. (F) Summary of serotonin-induced potentiation.

We then examined a contribution of this synergistic enhancement of the noradrenaline–D₁R signaling to dematuration of GCs, a proposed cellular substrate for antidepressant action (5, 21–24). Dematuration is characterized by altered expression of maturation stage markers and the appearance of immature-like functional properties in GCs. Prominent frequency facilitation at the MF synapse represents the functional maturity of GCs and is attenuated by serotonergic antidepressants (21, 22). We used the noradrenergic antidepressant desipramine, which acts as a highly potent noradrenaline reuptake inhibitor (33). Acute desipramine significantly augmented D₁R-dependent synaptic potentiation induced by low micromolar noradrenaline (SI Appendix, Fig. S6B). Chronic treatment with desipramine (Fig. 6A) did not change frequency facilitation when applied alone or in combination with wheel running (Fig. 6 B and C). In mice exposed to chronic restraint stress, desipramine alone had no significant effect. However, desipramine combined with wheel running significantly reduced frequency facilitation in stressed mice (Fig. 6 B and C). The reduction of facilitation was blocked by SKF83566 (Fig. 6D), suggesting an involvement of D₁Rs. Furthermore, we found that the expression level of the mature GC marker calbindin was significantly reduced by desipramine combined with wheel running in stressed mice (Fig. 6E). In mice coadministered with SKF83566, the calbindin expression level was higher than that in mice treated with desipramine alone (Fig. 6F). These results suggest that the synergistic enhancement of the noradrenaline–D₁R signaling in stressed mice subjected to wheel running facilitated the induction of GC dematuration by the noradrenergic antidepressant. We also examined behavioral effects of desipramine in stressed mice. While chronic stress did not change anxiety- or depression-related behavior in commonly used behavioral paradigms (SI Appendix, Fig. S8), it significantly reduced voluntary activity in home cages (Fig. 6G). Desipramine combined with wheel running reversed the reduced activity in stressed mice, and SKF83566 blocked this effect (Fig. 6H), suggesting a beneficial effect of the desipramine treatment combined with wheel running on stress-induced behavioral impairment.

Fig. 6.

Experience-dependent augmentation of noradrenaline–D₁ receptor signaling facilitates effects of noradrenergic antidepressant. (A) Timeline of restraint stress, WR, and desipramine (Des) treatment. (B) Effects of chronic Des (30 mg/kg/day for 4 wk) and WR on 1-Hz frequency facilitation at MF synapse in nonstressed (left) and stressed (right) mice. (C) Summary of effects of Des and WR on frequency facilitation in nonstressed and stressed mice. Two-way ANOVA (stress effect, F_1,106 = 1.757, P = 0.1879; treatment effect, F_3,106 = 1.657, P = 0.1809; interaction, F_3,106 = 4.424, P = 0.0057) followed by Dunnett's test (**P = 0.0011). (D) Effects of SKF (1 mg/kg/day) on reduction of frequency facilitation caused by Des and WR in stressed mice. One-way ANOVA (F_2,39 = 7.508, P = 0.0017) followed by Dunnett's test (*P = 0.0247, **P = 0.0015). (E) Effects of Des and WR on calbindin expression in DG. Typical immunoblots (left) and quantified data (right). One-way ANOVA (F_2,8 = 5.809, P = 0.0276) followed by Dunnett's test (*P = 0.0399). (F) Effects of SKF on calbindin expression in DG of Des-treated mice (t₁₃ = 3.164, **P = 0.0075). (G) Effects of chronic stress on nocturnal home cage activity. (H) D₁-receptor–dependent reversal of stress-induced decrease in home cage activity by Des and WR. One-way ANOVA (F_3,28 = 9.789, P = 0.0001) followed by Bonferroni's test (**P < 0.01). The number of data is shown in the graph in C and D.

Discussion

We have revealed that noradrenaline can activate dopamine D₁Rs in the mouse hippocampus, but not in the striatum, at the submicromolar to micromolar concentration range. This noradrenaline–D₁R signaling was strongly augmented by neuronal excitation, a process that may include reorganization of D₁- and β-receptor–dependent signaling. Chronic stress and voluntary exercise synergistically enhanced the noradrenaline–D₁R signaling, thereby facilitating antidepressant-induced dematuration of DG neurons. Our results suggest that experience-dependent plasticity of the noradrenaline–D₁R signaling determines the responsiveness to antidepressant drugs.

Previous in vivo microdialysis studies have estimated that the basal extracellular noradrenaline concentration in the rodent hippocampus is 14–35 nM (34, 35). In general, transmitter concentrations near the release sites are much higher than the basal extracellular levels. While microdialysis studies estimated that the basal glutamate concentration in the hippocampus was 1–3 μM (36, 37), an electrophysiological study suggested that the peak glutamate concentration outside synapses could reach up to about 200 μM (38). Therefore, noradrenaline concentrations may reach micromolar levels around noradrenergic terminals. We showed that micromolar noradrenaline can activate D₁R in all hippocampus subregions examined. In the presence of intact noradrenaline uptake activity, low micromolar noradrenaline activated D₁Rs at the MF synapse in mice subjected to wheel running (SI Appendix, Fig. S6A), but not in naive mice (Fig. 1I), suggesting a critical influence of experience on the sensitivity of D₁Rs to noradrenaline. Our finding is consistent with the previous observations that the noradrenergic projection regulated performance of hippocampus-dependent learning tasks via D₁Rs (13, 18, 19). However, these studies favored an alternative interpretation that dopamine coreleased from the noradrenergic fibers activated D₁Rs. On the basis of the biosynthetic pathway of catecholamines, high-frequency firing of noradrenergic fibers is predicted to increase the vesicular content of dopamine relative to noradrenaline (19). The amount of dopamine released from optogenetically activated noradrenergic fibers projecting to the hippocampus is about 10% of noradrenaline after prolonged high-frequency firing (18). Therefore, assuming micromolar levels of noradrenaline as estimated above, hippocampal D₁Rs near noradrenergic terminals may be primarily activated by noradrenaline, although our results do not exclude the possibility of D₁R activation by coreleased dopamine (also see below).

The overlap or redundancy in D₁ and β receptor signaling was suggested by pharmacological experiments. We observed that the D₁R antagonist failed to cause significant effects in the absence of the β receptor antagonist or vice versa. Especially, at the MF synapse in ECT-treated mice, the effect of the β antagonist propranolol was detectable only after blocking D₁Rs (Fig. 3), suggesting that D₁R activation completely occluded the effect of β receptor activation. Both D₁ and β receptors are coupled with Gs–cAMP signaling pathways. It has been suggested that the cAMP signaling can be compartmentalized (39, 40). If D₁ and β receptors are localized close to each other, activation of either of them may saturate the local cAMP signaling, which could explain our results showing that blocking D₁ or β receptors alone had insignificant effects. Such colocalization can be best exemplified by the formation of heteromeric D₁ and β receptor complex. Indeed, our coimmunoprecipitation studies suggested the formation of the D₁–β receptor complex. As discussed below, we speculate that the formation of the heteromeric receptor complex is essential for regulation of the noradrenaline–D₁R signaling. At the MF synapse, the overlap of D₁ and β receptor signaling was seen in ECT-treated mice, but not in control mice (Fig. 3). ECT causes a marked increase in D₁R expression levels (4), which may be required to induce colocalization of D₁ and β receptors.

The sensitivity of D₁Rs to noradrenaline was different between brain regions. Micromolar noradrenaline caused robust activation of D₁Rs in the hippocampus, but not in the striatum. A bioluminescence resonance energy transfer assay showed that the ligand affinity of recombinant human D₁Rs depends on the subtype of Gα subunit. The D₁R coupled to Gαolf has a lower affinity for noradrenaline than that coupled to Gαs (41). Gαolf is the predominant subtype of Gα coupled with D₁Rs in the striatum (42), which may explain the failure in activation of striatal D₁Rs by micromolar noradrenaline. At the MF synapse, the D₁R sensitive to submicromolar noradrenaline was induced by ECT. ECT also produced the overlap between D₁ and β signaling and increased the sensitivity of β₂ receptors to noradrenaline. Heteromeric β₁–β₂ receptors show a higher affinity to synthetic agonists than homomeric β₁ receptors (29). Since we found that D₁Rs can interact with β₁ and β₂ receptors, a heteromeric complex including D₁ and β₂ or β₁–β₂ receptors is a candidate for the D₁R with the higher affinity for noradrenaline.

ECT augmented the noradrenaline-induced D₁R-dependent synaptic potentiation by about 200% at the maximal concentration of 30 μM, while dopamine-induced synaptic potentiation was increased by about 100% (Fig. 3F). Since D₁Rs solely mediate dopamine-induced potentiation at the MF synapse even after ECT (4), it is conceivable that ECT increased the efficacy of D₁R activation by noradrenaline. While ECT strongly up-regulates D₁R expression (SI Appendix, Fig. S5A) (4), an increase in the number of D₁Rs would equally enhance noradrenaline- and dopamine-induced potentiation. Therefore, we hypothesize that the reorganization of D₁ and β signaling, most probably the formation of the heteromeric D₁–β receptor complexes, underlies the increase in the efficacy as well as potency of D₁R activation by noradrenaline.

We observed coimmunoprecipitation of D₁ and β receptors in the hippocampal tissue lysates from both control and ECT-treated mice. Coimmunoprecipitation of these receptors in control mice supports the idea that the heteromeric receptor complex is the molecular basis for the overlap in the D₁ and β receptor signaling observed by electrophysiological and biochemical studies in naive mice (SI Appendix, Fig. S1 and Fig. 2). Although we hypothesized that ECT caused the reorganization of D₁ and β signaling at the MF synapse via formation of the heteromeric receptor complex, there was no obvious difference in the coimmunoprecipitation between control and ECT-treated mice. Since ECT did not significantly increase noradrenaline-induced potentiation of PSs mediated by D₁Rs in the DG and CA1 (SI Appendix, Fig. S4), the reorganization of D₁ and β signaling may be specifically induced at the MF system. An increase in heteromeric receptor complexes specific to the MF synapse, if any, may not be detected by coimmunoprecipitation in whole hippocampal tissue lysates.

Many lines of evidence suggest a critical role of the hippocampal DG in depression and antidepressant medication (7, 43, 44). We have previously demonstrated that antidepressant drugs and ECT induce immature-like phenotypes in the dentate GCs and at their output MF synapse and proposed this dematuration as the common neuronal mechanism underlying pharmacological and physical antidepressant treatments (5, 21–24). Although serotonergic antidepressants consistently induce GC dematuration in nonstressed mice (5, 21, 23, 45), the noradrenergic antidepressant desipramine failed to induce dematuration even when combined with exercise in nonstressed mice. This is probably related to lower functional levels of endogenous catecholamines at the MF system as compared with serotonin (46). Stress facilitated the induction of dematuration by desipramine, most likely via augmentation of the noradrenaline–D₁R signaling. Since desipramine inhibits uptake of dopamine as well as noradrenaline, D₁R activation by dopamine coreleased from noradrenergic fibers may also contribute to the induction of dematuration by desipramine. As described above, the amount of dopamine that can be released from noradrenergic fibers in the hippocampus is predicted to be at most 10% of noradrenaline. In the presence of desipramine, noradrenaline at 3 μM, a concentration that may be reached around noradrenergic terminals as estimated above, caused strong D₁R-dependent synaptic potentiation in mice subjected to stress and wheel running (SI Appendix, Fig. S6B). In the same condition, a 10-fold lower concentration of dopamine also induced robust potentiation, although it was smaller in magnitude than that by noradrenaline (SI Appendix, Fig. S6B). Therefore, while the noradrenaline–D₁R signaling is likely to play a major role in the antidepressant effect of desipramine, dopamine could also make a significant contribution, depending on its vesicular content. Since the dopamine transporter is expressed, if any, at a very low level in the hippocampus (47, 48), the noradrenaline transporter may be essential for clearance of dopamine released from the dopaminergic fibers in addition to the noradrenergic fibers. Indeed, systemic administration of the noradrenaline transporter inhibitor can raise extracellular dopamine levels in the hippocampus (49). However, the dopamine level is much less sensitive to the noradrenaline transporter inhibitor as compared with the extracellular noradrenaline level (50). Given sparse dopaminergic innervation to the DG and CA3 region (11–13), it is unlikely that dopamine solely mediates D₁R receptor activation involved in the effects of desipramine, although we cannot exclude this possibility. Dematuration by the serotonergic antidepressant fluoxetine requires 5-HT₄ receptors (21, 23), which are abundantly expressed in the GCs and along the MF tract (23, 51). Since both 5-HT₄ and D₁Rs are coupled with Gs, cAMP-dependent signaling is likely to be a common downstream pathway for the induction of dematuration. The noradrenaline–D₁R signaling was observed at the somatodendritic and axonal regions of GCs. At present, it is unknown whether somatodendritic or axonal D₁Rs are required for the induction of dematuration. Experience- or activity-dependent augmentation of noradrenaline–D₁R signaling was prominent at the MF synapse. The extent of fluoxetine-induced dematuration was significantly correlated with the magnitude of 5-HT₄-receptor–mediated potentiation at the MF synapse (21). Therefore, it is possible that axonal cAMP signaling plays a critical role in GC dematuration.

Although chronic restraint stress combined with wheel running enhanced the noradrenaline–D₁R signaling at the MF synapse, it did not significantly change 5-HT₄-receptor–dependent synaptic potentiation (Fig. 5). The exact mechanism underlying this selective enhancement is unknown. The dentate GCs show an acute increase in the c-fos expression, a measure of neuronal activity, after voluntary exercise (52, 53). The D₁R expression in the DG is highly sensitive to neuronal activity. While single ECT strongly up-regulates D₁R expression levels (4), repeated ECT does not significantly change 5-HT₄ receptor expression levels (46). This differential sensitivity to neuronal activity can explain the selective enhancement of the noradrenaline–D₁R signaling. In addition to up-regulation of the D₁R expression, neuronal activity could increase the efficacy of D₁R activation by noradrenaline as demonstrated in ECT-treated mice. Furthermore, chronic restraint stress may boost the activity-dependent up-regulation of the D₁R signaling via corticosterone (22), resulting in the marked enhancement of the noradrenaline–D₁R signaling by restraint stress combined with wheel running.

The central noradrenergic system has long been thought to be an important target of antidepressant drugs, although it is not precisely understood how the noradrenergic system is involved in the pathology of depression and antidepressant effects. Antidepressant drugs have been assumed to reverse dysfunction of monoaminergic systems caused by stress and/or other risk factors for depression (30, 54). However, our finding implies that antidepressant drugs make use of stress-induced adaptive changes in the noradrenaline–D₁R signaling to induce dematuration of GCs. Stress and antidepressant treatment can activate shared cellular responses that mediate resilience to stress (55, 56). The noradrenaline–D₁R signaling may play a role as stress resilience that facilitates the antidepressant efficacy in depressive disorders. Activation of D₁Rs contributes to the effects of the serotonergic antidepressant fluoxetine in the DG (5), including GC dematuration and increased adult neurogenesis, another candidate cellular substrate for antidepressant action (44). Therefore, the noradrenaline–D₁R signaling may be commonly involved in the effects of different classes of antidepressant drugs. Chronic corticosterone treatment that mimics chronic stress exposure augments D₁R signaling at the MF synapse and concomitantly facilitates the induction of dematuration by fluoxetine (22). Chronic corticosterone also facilitates the fluoxetine-induced increase of adult neurogenesis in the DG (57), and antidepressant drugs more consistently increase adult neurogenesis in stressed animals than in naive animals (58). These lines of evidence, including the present study, suggest that antidepressant drugs activate the process of adaptive neuronal plasticity that is initiated by chronic stress rather than simply reversing stress-induced alterations in the brain. Our present finding suggests that the enhancement of the noradrenaline–D₁R signaling plays a pivotal role in linking stress with antidepressant-induced plasticity. The enhancement of the noradrenaline–D₁R signaling by stress was conditional, requiring increased voluntary activity. This synergistic experience-dependent regulation of the noradrenaline–D₁R signaling, possibly influenced by other unknown factors, may determine the antidepressant responsiveness, which suggests a unique target for therapeutic treatments of depressive disorders refractory to antidepressant medication.

Materials and Methods

Adult male C57BL/6J mice were used in all experiments. Animal use and procedures were approved by the Animal Care and Use Committee of Nippon Medical School, Tokyo University of Science, and Kurume University. Electrophysiological recordings were made using acute hippocampal slices (8, 46). In protein phosphorylation analyses, the regions of the dentate gyrus or the striatum were dissected from coronal brain slices (5, 59). Bilateral electroconvulsive treatment was administered to mice anesthetized with isoflurane (24). HEK293T cells transfected with plasmids carrying FLAG-tagged D₁ receptors (60) and/or HA-tagged β₁/β₂ receptors were used for cAMP measurements and receptor coimmunoprecipitation analyses. Mice were treated with oral desipramine, exposed to chronic restraint stress or wheel running, or subjected to combination of these treatments. These mice were used for electrophysiological experiments, home cage activity monitoring (61), and western blotting. For detailed methods, see SI Appendix.

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.