AMP-activated protein kinase slows D2 dopamine autoreceptor desensitization in substantia nigra neurons

Abstract

Dopamine neurons in the substantia nigra zona compacta (SNC) are well known to express D2 receptors. When dopamine is released from somatodendritic sites, activation of D2 autoreceptors suppresses dopamine neuronal activity through activation of G protein-coupled K⁺ channels. AMP-activated protein kinase (AMPK) is a master enzyme that acts in somatic tissues to suppress energy expenditure and encourage energy production. We hypothesize that AMPK may also conserve energy in central neurons by reducing desensitization of D2 autoreceptors. We used whole-cell patch-clamp recordings to study the effects of AMPK activators and inhibitors on D2 autoreceptor-mediated current in SNC neurons in midbrain slices from rat pups (11-23 days post-natal). Slices were superfused with 100 μM dopamine or 30 μM quinpirole for 25 min, which evoked outward currents that decayed slowly over time. Although the AMPK activators A769662 and ZLN024 significantly slowed rundown of dopamine-evoked current, slowing of quinpirole-evoked current required the presence of a D1-like agonist (SKF38393). Moreover, the D1-like agonist also slowed the rundown of quinpirole-induced current even in the absence of an AMPK activator. Pharmacological antagonist experiments showed that the D1-like agonist effect required activation of either protein kinase A (PKA) or exchange protein directly activated by cAMP 2 (Epac2) pathways. In contrast, the effect of AMPK on rundown of current evoked by quinpirole plus SKF38393 required PKA but not Epac2. We conclude that AMPK slows D2 autoreceptor desensitization by augmenting the effect of D1-like receptors.

Graphical Abstract

1. Introduction

AMP-activated protein kinase (AMPK) is a heteromeric enzyme complex composed of an alpha catalytic subunit, a beta scaffold subunit, and a regulatory gamma subunit (Rutter and Leclerc, 2009). When energy supplies are low, the ATP-to-AMP ratio falls, and AMP binds to the gamma subunit, which then permits activation by phosphorylation of the alpha subunit by upstream kinases CaMKKβ or liver kinase B1 (Carling et al., 2008). In somatic tissues, AMPK acts to conserve energy while also promoting energy production. Thus, AMPK inhibits ATP-utilizing processes such as gluconeogenesis and synthesis of lipids and proteins, while promoting glucose uptake and optimizing mitochondrial function (Hardie et al., 2012). AMPK also increases the surface expression of ATP-sensitive K⁺ (K-ATP) channels in pancreatic beta-cells and thereby fine tunes the regulation of insulin release (Beall et al., 2013; Rutter and Leclerc, 2009; Chen et al., 2013). Although AMPK is widely expressed, its action in neurons is only beginning to be defined (Amato and Man, 2011). We recently reported that activators of AMPK increase calcium-dependent K-ATP currents in subthalamic nucleus neurons (Shen et al., 2014). More recently, we reported that AMPK activation augments current evoked by K-ATP channel openers in dopamine neurons of the substantia nigra zona compacta (SNC) and ventral tegmental area (VTA) (Wu et al., 2017; Shen et al., 2016). Because K-ATP channels open a K⁺ conductance, they exert a hyperpolarizing, inhibitory influence on neuronal excitability. By suppressing action potentials, this results in reduced Na⁺ and Ca²⁺ influx that would otherwise need to be removed by ATP-utilizing exchange mechanisms. This led us to hypothesize that AMPK may act to conserve energy in neurons by suppressing neuronal excitability, in addition to its other duties that maintain energy homeostasis.

The D2 dopamine autoreceptor exerts an important inhibitory influence on the excitability of SNC dopamine neurons. Expressed on the surface of dopamine neurons, these G protein-coupled receptors open inwardly rectifying K⁺ channels that hyperpolarize the cell membrane, which thereby limits further dopamine release (Uchida et al., 2000; Cheramy et al., 1981). As such, the D2 autoreceptor exerts an important negative feedback mechanism on dopamine neuronal activity (Ford, 2014; Beckstead and Williams, 2007). However, overstimulation of D2 autoreceptors causes desensitization and a reduction in this inhibitory influence. Many studies suggest that autoreceptor desensitization may be caused by β-arrestin-dependent internalization of the receptor (Beaulieu and Gainetdinov, 2011), although uncoupling from G protein is an alternative explanation (Robinson et al., 2018). Loss of the inhibitory influence of the autoreceptor facilitates further dopamine release and has been linked to behavioral changes. For example, desensitization of D2 autoreceptors in VTA dopamine neurons has been shown to facilitate behavioral reinforcement to cocaine and other drugs of abuse (Bello et al., 2011). Although an effect of AMPK on D2 autoreceptor desensitization has not been examined previously, studies of other transmitter systems suggest it might slow desensitization or otherwise enhance receptor function. In the hippocampus, AMPK has been shown to phosphorylate the GABA-B receptor and slow the rundown of current evoked by the GABA-B agonist baclofen (Kuramoto et al., 2007). AMPK activation has also been reported to increase the surface expression of GABA-A receptors in hippocampus and increase GABA-A receptor-mediated synaptic activity (Fan et al., 2019). In fact, AMPK has been reported to increase the surface expression of a variety of ion channels and receptors including the GLUT3 glucose transporter (Weisová et al., 2009), ATP-sensitive K⁺ channels (Wu et al., 2015), and delayed rectifier K⁺ channels (Wu et al., 2015). Thus, it is possible that AMPK could slow apparent desensitization of D2 autoreceptors by either slowing internalization or promoting surface expression.

We used patch pipettes to record whole-cell currents from presumed dopamine-containing neurons in slices of rat SNC. Slices were superfused with high concentrations of dopamine or the D2 agonist quinpirole while measuring outward currents that diminished over time despite the continued presence of the agonist. Activators of AMPK slowed the rundown of current evoked by dopamine, but slowing of current rundown by quinpirole required concurrent stimulation of D1-like receptors. Current clamp recordings showed that AMPK prolonged the inhibition of spontaneous firing of action potentials by dopamine. These results suggest that AMPK slows D2 autoreceptor desensitization and are consistent with the view that AMPK acts to conserve energy by reducing neuronal excitability.

2. Materials and Methods

2.1. Animals and slice preparation

Sprague-Dawley rat pups (11-23 days post-natal) were obtained from Envigo, USA. A total of 217 animals were used in our experiments. Protocols were approved and reviewed by the animal care committee at the Veterans Affairs Portland Health Care System (Oregon, USA). Every precaution was taken to minimize animal stress and the number of animals used.

Rat pups were anesthetized with isoflurane (Baxter, USA) and euthanized by rupture of major thoracic vessels. The brain was quickly removed and submerged in ice-cold sucrose cutting solution made of (in mM) sucrose (196), KCl (2.5), MgCl₂ (3.5), CaCl₂ (0.5), NaH₂PO₄ (1.2), NaHCO₃ (26), and glucose (20) that had been bubbled with 95%/5% O₂/CO₂ blended (carbogen) gas. A brain block containing the ventral midbrain was mounted on a platform with cyanoacrylate glue and horizontal slices (250 μm thick) containing the substantia nigra were cut with a vibrating microtome (Leica Biosystems, USA) while submerged in chilled cutting solution bubbled with carbogen. Slices were placed in a holding chamber containing artificial cerebral spinal fluid (aCSF) saturated with carbogen gas at room temperature. The ACSF was composed of (in mM): NaCl (126), KCl (2.5), CaCl₂ (2.4), MgCl₂ (1.2), NaH₂PO₄ (1.2), NaHCO₃ (19), and glucose (11).

2.2. Electrophysiology

After at least one hour in the holding chamber, a hemi-slice containing the SNC was transferred into a submersion recording chamber mounted on an upright microscope (Zeiss Microscopy, USA) and held down with platinum wire pieces and continuously perfused with warmed (35°C) ACSF flowing at 2 mL/min. Using bright-field optics, slices were visualized under 5x objective (50x final magnification) to initially center the microscope over the SNC. We used tissue from rat pups because neurons are more easily visualized and recorded with patch pipettes compared to adult rat tissue. Presumed dopamine neurons in the SNC were recorded in the region immediately rostral to the medial terminal nucleus of the accessory optic tract (Paxinos and Watson, 1986). Microscope output was captured with a digital camera (Orca 5G, Hamamatsu, Japan) and projected onto a computer monitor.

Patch pipettes were pulled from thick-walled 1.5 mm borosilicate glass capillary tubing (World Precision Instruments, USA) on a vertical microelectrode puller (PC-10, Narishige Instrument Co, Japan). Immediately before use, the pipettes were backfilled with chilled intracellular solution (pH 7.3) containing (in mM): potassium gluconate (138), MgCl₂ (2), CaCl₂ (1), EGTA (11), HEPES (10), ATP (1.5), and GTP (0.3). The ATP and GTP salts were added to a thawed aliquot of internal solution stock at the start of each week of experiments.

Pipette positioning onto neurons was visualized with a submersible 40x objective (400x final magnification) using differential interference contrast optics under infrared illumination. Currents were recorded from singe SNC neurons under voltage clamp (−60 mV) using patch pipettes (3-6 MΩ) in whole-cell configuration. Current was filtered (2 kHz low-pass) and amplified (Axopatch-1D, Molecular Devices, USA), digitized (Digidata 1550, Molecular Devices, USA) and saved to computer storage with digitizer-associated software (Clampex v10, Molecular Devices, USA). Presumed dopamine-containing neurons in the SNC were identified by having relatively broad (1-2 ms) spontaneous action currents and large hyperpolarization-induced inward currents (Lacey et al., 1989). Recordings were corrected for a liquid junction potential offset of 10 mV.

2.3. Drugs and chemicals

All drugs were dissolved in either purified water or dimethyl sulfoxide (DMSO). Stock solutions containing DMSO were made to require at least 1000-fold dilution to working strength to minimize final DMSO concentration in the superfusate. A769662, ZLN024, STO609 acetate, PKA Inhibitor fragment(6-22) amide (PKI), forskolin and (−)-quinpirole hydrochloride were purchased from Cayman Chemical (USA). Dopamine hydrochloride, dorsomorphin dihydrochloride (Compound C), tetrodotoxin (TTX) and (−)-sulpiride were purchased from Sigma-Aldrich (USA). SCH39166, U73122, and ESI05 were obtained from Tocris/Bio-techne (UK). A769662, ZLN024, and PKI were dissolved to working strength in internal pipette solutions; these agents diffused passively from pipette solutions into the cytosol during whole-cell recordings. All other drugs were dissolved to final concentration in aCSF and added to the brain slice superfusate. Dorsomorphin, STO609, SCH39166, ESI05, and U73122 were added to the superfusate 10 min before adding dopamine or quinpirole. For agents that were added to the pipette solution (A769662, ZLN024, and PKI), we waited at least 10 min before applying dopamine or quinpirole to the superfusate. Typically 60 sec were required for solution changes to traverse the tubing system to show initial effect at the neuron. Dopamine solutions were made daily and kept on ice. A summary of agents used is shown in Table 1.

Table 1. Summary of compounds used.

Concentrations (Conc) refer to final concentrations in slice superfusate or pipette solutions, whereas Solvent refers to stock solutions.

Agent	Action	Conc	Solvent	Mode of application	Protocol
Dopamine	D1 & D2 agonist	100 μM	aCSF	Superfusate	Superfused for 25 min
Quinpirole	D2 agonist	30 μM	aCSF	Superfusate	Superfused for 25 min
Sulpiride	D2 antagonist	1-10 μM	aCSF	Superfusate	Superfused for 5 min at end of recording
A769662	AMPK activator	10 μM	DMSO	Pipette solution	Diffusion into cell at least 10 before DA or Quin
BAPTA	Ca2+ chelator	10 mM	aCSF	Pipette solution	Passive diffusion into cell at least 10 before Quin
Dorsomorphin	AMPK inhibitor	30 μM	aCSF	Superfusate	Superfusion started 10 min before DA or Quin
ESI05	EPAC2 inhibitor	10 μM	DMSO	Superfusate	Superfusion started 10 min before Quin
Forskolin	AC activator	10 μM	aCSF	Superfusate	Superfused simultaneously with Quin
PKI	PKA inhibitor	10 μM	aCSF	Pipette solution	Passive diffusion into cell at least 10 before Quin
STO609	AMPK inhibitor	10 μM	aCSF	Superfusate	Superfusion started 10 min before DA or Quiin
SKF38393	D1 agonist	10 μM	aCSF	Superfusate	Superfused simultaneously with Quin
SCH39166	D1 antagonist	1 μM	aCSF	Superfusate	Superfusion started 10 min before Quin
Tetrodotoxin	Channel blocker	0.3 μM	aCSF	Superfusate	Superfusion started 10 min before Quin
U73122	PLC inhibitor	5 μM	DMSO	Superfusate	Superfusion started 10 min before Quin
ZLN024	AMPK activator	10 μM	DMSO	Pipette solution	Diffusion into cell at least 10 before DA or Quin

2.4. Data measurement and analysis

Slices were superfused continuously with either dopamine or quinpirole for 25 min, and currents were measured at peak value and at 5, 10, 20 and 25 min after starting superfusion. Sulpiride was added to the superfusate at the 20 min time point to measure the amplitude of current blocked by sulpiride (sulpiride-sensitive current). Decay plots of dopamine- or quinpirole-induced current were analyzed at the 5, 10 and 20 min time points using a linear mixed model with the following parameters: treatment and time were fixed effects (sum-of-squares method III), and repeated measures, from each neuron, were modeled as repeated within-subject, using first-order autoregressive covariance structure (AR1). The restricted maximum likelihood estimator was used. Post-hoc pairwise comparisons of estimated means were adjusted using Sidak’s method. Normality of linear mixed model datasets were inspected on Studentized residual plots. Data points that exceeded three standard deviations were excluded. Between-group analysis of currents blocked by sulpiride were done using either Student’s or Welch’s unpaired t tests, depending upon equal versus unequal variance as determined by Levene’s test. All statistics were computed with SPSS v25 (IBM North America, USA). Numerical data in the text and error bars in figures are expressed as mean average ± standardized error of the mean. Significance was accepted with P < 0.05.

3. Results

3.1. AMPK activators slow rundown of dopamine-induced current

In order to study autoreceptor desensitization, our protocol utilized a relatively high concentration of dopamine (100 μM) that is known to cause rundown of current over time (Lacey et al., 1987). As illustrated in Fig 1 (top trace), slices were superfused continuously for 25 min with 100 μM dopamine while recording current under voltage-clamp. Sulpiride (1 - 10 μM) was added to the superfusate during the last 5 min of dopamine superfusion in order to measure the magnitude of residual D2 autoreceptor-mediated current. As shown in the current trace, dopamine-induced outward current declined to near zero after 5 min, and there was little or no sulpiride-sensitive current at the end of the recording. However, the middle and lower current traces in Fig 1 show that AMPK activating agents caused significant slowing of the rundown of dopamine-induced current. The AMPK activators A769662 (middle trace) and ZLN024 (lower trace) slowed rundown of dopamine-induced current, and caused noticeable sulpiride-sensitive currents at the end of recordings. We showed previously that A769662 increases phosphorylation of the AMPK alpha subunit at thr-172 in rat midbrain slices (Shen et al., 2014), which is considered a requirement for AMPK activation (Hawley et al., 1996; Göransson et al., 2007). In contrast, ZLN024 promotes AMPK activation by protecting AMPK Thr-172 against dephosphorylation (Zhang et al., 2013). Both A769662 (10 μM) and ZLN024 (10 μM) were contained in internal pipette solutions in separate experiments. Dopamine superfusion was begun 10-15 min after rupture of the membrane for whole-cell recordings.

Fig. 1. Current traces showing that activators of AMPK slow desensitization of dopamine-evoked current.

Currents were recorded with pipettes that contained normal internal solution (control), A769662 (10 μM), or ZLN024 (10 μM). Slices were superfused with dopamine (100 μM) for 25 min, and sulpiride (1 μM) was added to the superfusate during the last 5 min. Each current trace was recorded from a different neuron.

Experiments with A769662 and ZLN024 are summarized in Fig. 2A. Under control conditions, dopamine-induced outward current reached an average peak of 56.9 ±5.1 pA at 2.1 ±0.1 min after beginning superfusion (n = 21). This outward current then decayed in amplitude, typically diminishing to baseline levels within 5 to 10 minutes. In contrast, rundown of dopamine-induced current was significantly slowed by A769662. Mixed model analysis showed there was a significant effect of A769662 on dopamine-induced current (F_(1,42.56) = 91.41, P = 3.6812e-12). In the presence of A769662, the peak outward current produced by dopamine (61.0 ± 3.7 pA, n = 25) did not differ significantly from the control value (t(44) = 0.660, P = 0.513, t test), nor did the time to reach peak amplitude (2.2 ± 0.1 min; t(44) = 1.49670, P = 0.142, t test). ZLN024 also greatly slowed the rundown of dopamine-induced current. Mixed model analysis showed there was a significant effect of ZLN024 on current evoked by dopamine (F_(1,34.19) = 35.24, P = 0.000001). With ZLN024 in pipettes, neither the peak dopamine-evoked current (56.1 ± 4.9 pA) nor the time of peak response (2.2 ± 0.1 min, n = 15) differed significantly from control values (t(34) = 0.118, P = 0.907 and t(34) = 0.898, P = 0.376, respectively, t tests). Control experiments were done with pipette solutions that contained both A769662 and ZLN024; as shown in Fig. 2A, the combination of A769662 and ZLN024 evoked no current in the absence of dopamine (n = 5).

Fig. 2. Summary graphs showing that AMPK activators slow rundown of current evoked by dopamine (100 μM).

A) Both A769662 and ZLN024 significantly slowed the rundown of dopamine-induced current compared to control. The combination of A769662 plus ZLN024 evoked no outward current in the absence of dopamine. B) A769662 and ZLN024 increased sulpiride-sensitive currents recorded in the last 5 min of dopamine superfusion. C) Bath application of AMPK inhibitors dorsomorphin (30 μM) and STO609 (10 μM) reversed the slowing of current rundown that is produced when pipettes contained A769662 (10 μM). The combination of dorsomorphin and STO609 evoked no outward current in the absence of dopamine D) Dorsomorphin and STO609 blocked the ability of A769662 to increase sulpiride-sensitive currents. Current-decay plots were analyzed with a mixed model followed by Sidak pairwise comparison tests, whereas sulpiride-sensitive currents were analyzed with Welch’s t tests: **, P < 0.01; ***, P < 0.001.

Dopamine evoked significant sulpiride-sensitive currents in the presence of A769662 and ZLN024, as shown in Fig. 2B. In the continued presence of dopamine, sulpiride-sensitive current in A769662 (16.3 ± 3.1 pA, n = 14) was significantly greater than that evoked by dopamine in the control condition (0.6 ± 0.9 pA, n = 14; t(15.31) = 4.928, P < 0.000171, Welch’s t test). Sulpiride-sensitive current in ZLN024 (19.2 ± 3.4 pA, n = 7) was also significantly greater than that recorded with dopamine under the control condition (t(6.861) = 5.199, P < 0.001336, Welch’s t test). As a control experiment, dopamine evoked no significant outward current (0.3 ± 0.9 pA, n = 4) after a 5 min superfusion with sulpiride (t(3) = 0.280, P = 0.798, paired t test); this supports the conclusion that dopamine-induced currents are mediated by D2 autoreceptors. Overall, these results are consistent with the hypothesis that activation of AMPK slows D2 autoreceptor desensitization.

3.2. AMPK blocking agents antagonize the effect of A769662 on dopamine current rundown

Figure 2C shows the ability of AMPK blocking agents to prevent the slowing of dopamine-induced current rundown in the presence of A769662. AMPK blocking agents were added 10 min before dopamine and superfused continuously throughout experiments. Dorsomorphin (compound C), which inhibits AMPK activity by blocking phosphorylation of Thr-172 by upstream kinases (Kunanusornchqai et al., 2016; Zhou et al., 2001), significantly accelerated the rundown of dopamine-induced current when recorded with pipettes that contained A769662. Mixed model analysis showed that there was a significant effect of dorsomorphin on the A769662-enhanced current evoked by dopamine (F_(1,32.28) = 45.57, P = 1,206e-7). We also examined the effect of STO609, which blocks the ability of the upstream kinase CaMKKβ to phosphorylate Thr-172 and thereby prevents AMPK activation (Hawley et al., 2005; Tokumitsu et al., 2002). Similar to dorsomorphin, STO609 greatly accelerated the rundown of dopamine-induced current in the presence of A769662. Mixed model analysis showed that there was a significant effect of STO609 on the A769662-enhanced current evoked by dopamine (F_(1,33.12) = 61.67, P = 4.6501e-9). Both dorsomorphin (30 μM) and STO609 (10 μM) were added to the superfusate 10-15 min before beginning superfusion with dopamine (100 μM). As shown in Fig. 2C, the combination of dorsomorphin and STO609 evoked no current in the absence of dopamine (n = 8).

Effects of dorsomorphin and STO609 on sulpiride-sensitive currents, recorded in the presence of A769662, are shown in Fig. 2D. In the continued presence of dopamine and A769662, sulpiride-sensitive current in dorsomorphin (−0.3 ± 0.6 pA, n = 11) was significantly less than that evoked under the A769662 control condition (16.3 ± 3.1 pA, n = 14; t(14.035) = 5.520, P = 0.000107, Welch’s t test). Sulpiride-sensitive current evoked in STO609 plus A769662 (−1.4 ± 1.8 pA, n = 8) was also significantly less than that evoked by dopamine in the A769662 control condition (t(19.333) = 4.978, P = 0.000080, Welch’s t test). It is important to note that A769662 and STO609 block AMPK activation by different mechanisms, which reinforces the conclusion that the action of A769662 to slow dopamine-evoked current rundown is mediated by AMPK.

3.3. Characterization of currents evoked by the selective D2-like agonist quinpirole

Although slowing of dopamine-induced current rundown could be due to a direct action of AMPK on dopamine D2 autoreceptors, results thus far do not rule out an effect of AMPK on D1-like receptors or the dopamine transporter. Although dopamine neurons may not express the D1 receptor subtype, immunohistological and neurophysiological studies suggest that these neurons express the D5 receptor subtype (Khan et al., 2000; Schilström et al., 2006). Therefore, we proceeded to test the effect of A769662 on current evoked by quinpirole, which is a D2-like receptor agonist that has no D1-like activity and is not a substrate for the dopamine transporter (Bolan et al., 2007; Ledonne et al., 2010). In a method analogous to our dopamine protocol, slices were superfused with a relatively high concentration of quinpirole (30 μM) for 25 min, and sulpiride (10 μM) was added during the last 5 min in order to test for residual D2 autoreceptor-mediated current. As shown in the top current trace in Fig. 3A, quinpirole evoked an outward current that decayed slowly over time. Sulpiride, which superfused during the last 5 min of quinpirole superfusion, had little or no effect on current. We then repeated the experiment with pipettes that contained A769662 (10 μM), and to our surprise found that A769662 had no effect on rundown of quinpirole-induced current (see summary data in Fig. 3B). Because the AMPK activator affected current evoked by dopamine but not quinpirole, we considered the possibility that AMPK required activation of D1-like receptors to have an effect. Therefore, we next performed experiments in which slices were superfused simultaneously with quinpirole (30 μM) plus the D1-like agonist SKF38393 (10 μM). As shown in the middle trace in Fig. 3A, we found that SKF38393 slowed the rundown of quinpirole-induced current, even without the addition of an AMPK activator. The bottom trace shows that A769662 caused further slowing of quinpirole-induced current rundown when recorded with SKF38393. A summary of these experiments is shown in Fig. 3B. Under control conditions, quinpirole evoked an outward current of 55.8 ± 5.0 pA which peaked 2.1 ± 0.1 min after starting superfusion (n = 9). Rundown of quinpirole-induced current was somewhat slower than that of dopamine, but rundown was complete 20 min after starting superfusion. However, the rundown of quinpirole-evoked current was significantly slowed by SKF38393. Mixed model analysis showed there was a significant treatment effect of SKF38393 on quinpirole-evoked current (F_(1,18.93) = 9.93, P = 0.005273). Rundown of current evoked by quinpirole plus SKF38393 was slowed further when recording with pipettes that contained A769662. Mixed model analysis showed there was a significant effect of A769662 on current evoked by quinpirole plus SKF38393 (F_(1,17.39) = 9.42, P = 0.006811). Figure 3B also shows that recording with pipettes that contained A769662 had no effect on currents evoked by quinpirole alone (F_(1,14.18) = 0.587, P = 0.455983). Furthermore, Fig. 3B shows that slices superfused with SKF38393-alone evoked no current (n = 4). To control for possible remote effects of SKF38393, we found that superfusing slices with TTX (0.5 μM) had no effect on rundown of currents evoked by quinpirole plus SKF38393 (F_(1,1484) = 0.043, P = 0.838985; data not shown). These control experiments suggest that slowing of quinpirole-induced current rundown by AMPK activation requires D1-like receptor stimulation, and this effect is not mediated indirectly by release of endogenous transmitters.

Fig. 3. AMPK activation slows rundown of quinpirole-induced current in the presence of a D1-like receptor agonist.

A) Current traces show that rundown of current evoked by quinpirole (30 μM) is slowed by the D1 agonist SKF38393 (10 μM), and the rundown is further slowed by A769662. Quinpirole and SKF38393 were superfused simultaneously, and A769662 (10 μM) was contained in the pipette solution. Each current trace is from a different neuron. B) Summary graph showing that the addition of SKF38393 slowed the rundown of quinpirole-induced current (*), and recording with pipettes that contained A769662 further slowed the rundown of currents evoked by quinpirole plus SKF38393 (#). Note that A769662 did not affect rundown of quinpirole-induced current when recorded without SKF38393. SKF38393, when superfused alone, evoked no current. C) Sulpiride-sensitive currents evoked by quinpirole were increased by SK38393, and these currents evoked by quinpirole plus SKF38393 were increased further by A769662. Current-decay plots were analyzed with a mixed model followed by Sidak pairwise comparison tests, whereas sulpiride-sensitive currents were analyzed with Welch’s t tests: * or #, P < 0.05; ** or ##, P < 0.01; ***, P < 0.001.

Effects of SKF38393 and A769662 on sulpiride-sensitive currents evoked by quinpirole are summarized in Fig. 3C. During the last 5 min of recordings, sulpiride blocked significantly more outward current when evoked by quinpirole plus SKF38393 (13.5 ± 1.9 pA, n = 11) compared to that blocked when quinpirole-alone was superfused (2.4 ± 0.7 pA, n = 6; t(15) = 4.156, P = 0.000844, t test). Furthermore, sulpiride-sensitive current recorded in quinpirole plus SKF38393 was significantly larger when pipettes contained A769662 (31.0 ± 5.7 pA, n = 7) compared to recordings without A769662 (t(7.408) = 2.918, P = 0.021036, Welch’s t test). In contrast, sulpiride-sensitive current evoked by quinpirole with A769662 (2.8 ± 2.1, n = 8) was not significantly different from quinpirole alone (t(12) = 0.180, P = 0.860, t test). These results support the conclusion that slowing of D2 autoreceptor desensitization by AMPK requires D1-like receptor stimulation.

3.4. Antagonist pharmacology of quinpirole-induced currents plus SKF38393 and A769662

We used selective blocking agents to characterize the ability of SKF38393 and A769662 to slow the rundown of quinpirole-induced current. Each blocking agent was added to the superfusate 10 min before beginning superfusion with quinpirole plus SKF38393. Figure 4A shows that the D1-like antagonist SCH39166 (1 μM) effectively blocked the ability of SKF38393 to slow the rundown of quinpirole-induced current. Mixed model analysis showed there was a significant effect of SCH39166 on current evoked by quinpirole plus SKF38393 (F_(1,16.80) = 22.14, P = 0.000210). Notice that the D1-like antagonist produced a moderate but significant acceleration of the rundown of current evoked by quinpirole alone (F_(1,12.66) = 6.09, P = 0.028642), perhaps due to the block of D1-like receptors stimulated by endogenous dopamine. As expected, the AMPK blocking agent dorsomorphin prevented the ability of A769662 to slow the rundown of current evoked by quinpirole plus SKF38393, as shown in Fig. 4B. Mixed model analysis showed that there was a significant effect of dorsomorphin on current evoked by quinpirole plus SKF38393 when pipettes contained A769662 (F_(1,12.22) = 10.05, P = 0.007902). In contrast, dorsomorphin had no significant effect on current evoked by quinpirole plus SKF38393 when recording with pipettes without A769662 (F_(1,16.75) = 0.15, P = 0.704020; data not shown), and dorsomorphin had no effect on sulpiride-sensitive currents in the absence of A769662 (t(4.626) = 0.626, P = 0.561, Welch’s t-test). Finally, Fig. 4C shows that the D1-like antagonist SCH39166 also blocked the ability of A769662 to slow the rundown of current evoked by quinpirole plus SKF38393. Mixed model analysis showed there was a significant effect of SCH39166 on current evoked by quinpirole plus SKF38393 when recorded with pipettes that contained A769662 (F_(1,11.31) = 15.63, P = 0.002142).

Fig. 4. Blocking agents prevent the effects of SKF38393 and A769662 on quinpirole-induced currents.

A) The D1-like receptor antagonist SCH39166 (1 μM) blocks the ability of SKF38393 to slow the rundown of quinpirole-induced current. B) The AMPK blocking agent dorsomorphin prevents the ability of A769662 to slow the rundown of current evoked by quinpirole plus SKF38393. C) The D1 antagonist SCH39166 prevents the ability of A769662 to slow the rundown of current induced by quinpirole plus SKF38393. D) SCH39166 blocks sulpiride-sensitive currents evoked by quinpirole plus SKF38393. Sulpiride-sensitive currents evoked by quinpirole plus SKF38393 and A769662 are reduced by dorsomorphin and are completely blocked by SCH39166. Current-decay plots were analyzed with a mixed model followed by Sidak pairwise comparison tests, whereas sulpiride-sensitive currents were analyzed with Welch’s t tests: *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Effects of the above blocking agents on sulpiride-sensitive currents are shown in Fig. 4D. Sulpiride-sensitive current evoked by quinpirole plus SKF38393 was significantly smaller in the presence of SCH39166 (0.7 ± 0.4 pA, n = 7) compared to control current evoked by quinpirole plus SKF38393 (13.5 ± 1.9 pA, n = 11; t(10.763) = 6.525, P = 0.000047, Welch’s t test). Sulpiride-sensitive current recorded with dorsomorphin (12.1 ± 2.8 pA, n = 7) was significantly less than that evoked by quinpirole plus SKF38393 and A769662 (31.0 ± 5.7 pA, n = 7; t(12) = 2.990, P = 0.011272, t test). Furthermore, pretreatment with SCH39166 caused a significantly greater reduction in sulpiride-sensitive current (1.0 ± 0.5 pA, n = 5) compared to dorsomorphin, in the presence of quinpirole, SKF38393, and A769662 (t(6.406) = 3.867, P = 0.007310, Welch’s t test). These data show that the D1-like antagonist is able to block the effects of both SKF38393 and A769662. Taken together, results suggest that the ability of AMPK activation to slow the rundown of D2-mediated current requires D1-like receptor activity.

3.5. cAMP and calcium mediate the effect of SKF38393 on D2 current rundown

We next investigated mechanisms that might underlie the ability of SKF38393 to slow the rundown of quinpirole-induced current. Because D1-like receptors are well known to cause G protein-coupling to adenylyl cyclase, we investigated whether or not the adenylyl cyclase activator forskolin could substitute for SKF38393. Forskolin (10 μM) and quinpirole were superfused simultaneously, and SCH39166 (10 μM) was added to the superfusate 10 min before quinpirole and forskolin to block any D1 receptor-mediated actions of endogenous dopamine. As shown in Fig. 5A, forskolin caused slowing of rundown of quinpirole-induced current. Mixed model analysis showed there was a significant effect of forskolin on current evoked by quinpirole (F_(1,14.32) = 17.34, P = 0.000912). Forskolin also caused a significant increase in sulpiride-sensitive current (18.5 ± 2.5 pA, n = 7) compared to quinpirole alone (2.4 ± 0.7 pA, n = 6; t(6.877) = 6.267, P = 0.000448, Welch’s t test). Forskolin (plus SCH39166) evoked no current in the absence of quinpirole (n = 8).

Fig. 5. Cyclic AMP and calcium mediate the ability of D1-like receptor stimulation to slow the rundown of quinpirole-induced current.

A) Summary graph showing that forskolin mimics the ability of SKF38393 to slow the rundown of quinpirole-induced current (*). Quinpirole and forskolin were superfused simultaneously. The effect of forskolin on quinpirole-induced current was completely blocked by the PKA inhibitor PKI (#) but not by the Epac2 inhibitor ESI05. All recordings were done in the presence of SCH39166. Forskolin (plus SCH39166) evoked no current in the absence of quinpirole. B) The SKF38393-dependent slowing of quinpirole-induced current rundown was completely blocked by the combination of PKI and ESI05, but PKI by itself had no effect on current rundown. C) The PLC inhibitor U73122 plus PKI completely blocked the slowing of current rundown induced by quinpirole plus SKF38393, whereas U73122 applied alone did not affect current rundown. D) The calcium chelator BAPTA accelerated the rundown of current evoked by quinpirole plus SKF38393. Forskolin (10 μM), U73122 (5 μM) and ESI05 (10 μM) were added to the superfusate, whereas PKI (10 μM) and BAPTA (10 mM) were contained in pipette solutions. Data were analyzed with a mixed model followed by Sidak pairwise comparison tests: * or #, P < 0.05; **, P < 0.01; *** or ###, P < 0.001.

Having implicated cAMP in the slowing of quinpirole-induced current rundown, we proceeded to test if protein kinase A (PKA) or Epac (exchange protein directly activated by cAMP) was also involved. Because Epac2 has been implicated in second messenger signaling in midbrain dopamine neurons (Tone et al., 2017), we examined the effect of ESI05, which selectively inhibits Epac2 (Tsalkova et al., 2012). ESI05 (10 μM) was added to the superfusate 10 min before starting superfusion with quinpirole and forskolin. But as shown in Fig. 5A, ESI05 had no effect on the ability of forskolin to slow rundown of quinpirole-induced current (F_(1,11.35) = 0.07, P = 0.796901). Subsequent experiments examined the effect of PKI, which is a specific inhibitor of PKA (Hilfiker et al., 2001). Figure 5A shows that slowing of quinpirole-induced current rundown by forskolin was completely prevented when PKI was included in the pipette internal solution (10 μM). Shifts in baseline holding current over 25 min were negligible for PKI (0.6 ± 1.7 pA, n = 8) and ESI05 (0.4 ± 0.3 pA, n = 6). Mixed model analysis showed there was a significant effect of PKI on current evoked by quinpirole plus forskolin (F_(1,11.35) = 16.12, P = 0.002383). These data suggest that cAMP generation slows the rundown of quinpirole-induced current, and the effect of forskolin depends upon PKA activation.

We next performed experiments with SKF38393 to determine if its actions are also cAMP dependent. In contrast to the action of forskolin, Fig. 5B shows that the PKA inhibitor PKI had no effect on rundown of current evoked by quinpirole plus SKF38393 (F_(1,15.98) = 0.12, P = 0.735804). Moreover, PKI had no significant effect on sulpiride-sensitive current recorded in quinpirole plus SKF38393 (t(6.942) = 0.605, P = 0.564487, Welch’s t test). However, the combination of PKI and ESI05 produced a dramatic acceleration of current rundown evoked by quinpirole plus SKF38393. Mixed model analysis showed there was a significant effect of PKI plus ESI05 on current evoked by quinpirole and SKF38393 (F_(1,15.93) = 29.81, P = 0.000053). Moreover, sulpiride-sensitive current evoked by quinpirole plus SKF38393 (13.5 ± 1.9 pA, n = 11) was completely blocked by PKI plus ESI05 (0.1 ± 0.1 pA, n = 5; t(10.067) = 7.082, P = 0.000033, Welch’s t test). These results confirm the involvement of cAMP in the actions of SKF38393.

Because Epac has been shown to activate phospholipase C (PLC) (Baljinnyam et al., 2010), we next tested whether or not U73122, a PLC blocker (Bleasdale et al., 1990), could block the action of SKF38393. Slices were superfused continuously with U73122 (5 μM) 10 min before starting superfusion with quinpirole plus SKF38393. As shown in Fig. 5C, U73122 had no significant effect on rundown of current evoked by quinpirole plus SKF38393 (F_(1,16.69) = 0.90, P = 0.355820; mixed model analysis). However, the ability of SKF38393 to slow rundown of quinpirole-induced current was significantly decreased by U73122 when pipettes contained PKI (F_(1,16.16) = 19.92, P = 0.000384; mixed model analysis). Moreover, sulpiride-sensitive current in quinpirole plus SKF38393 (13.5 ± 1.9 pA, n = 11) was completely blocked by PKI plus U73122 (0.2 ± 0.6 pA, n = 5; t(14) = 4.686, P = 0.000350, t test). U73122 produced a negligible change in holding current over 25 min (1.4 ± 0.6 pA, n = 6). These results suggest that the action of SKF38393 is mediated in part by PLC, which can be activated by Epac2. However, our results also suggest that slowing of rundown of quinpirole current by SKF38393 requires activation of either PKA or Epac2, and both pathways must be blocked to prevent SKF38393 from slowing rundown of quinpirole-induced current. It should be noted that the dependence of the SKF38393 effect on either PKA or Epac2 differs from the effect of forskolin, which was dependent upon PKA but not Epac2.

Both PLC and PKA are well known to increase intracellular levels of calcium (Taylor, 2017; Rebecchi and Pentyala, 2000; Kim et al., 2006). Moreover, calcium has been shown to slow the D2 receptor desensitization via activation of neuron calcium sensor (NCS1) (Dragicevic et al., 2014; Kabbani et al., 2002). We therefore examined whether or not BAPTA can prevent the slowing of rundown of current evoked by quinpirole plus SKF38393. In these experiments, EGTA in internal pipette solutions was replaced by BAPTA (10 mM), which is a much more efficient chelator of calcium (Adler et al., 1991). Figure 5D shows that BAPTA caused a significant acceleration of rundown of current evoked by quinpirole plus SKF38393 (F_(1,17.69) = 26.84, P = 0.000066; mixed model). Moreover, sulpiride-sensitive current in quinpirole plus SKF38393 (13.5 ± 1.9 pA, n = 11) was significantly reduced by BAPTA (0.0 ± 0.1 pA, n = 7; t(10.020) = 7.046, P = 0.000035, Welch’s t test). These results suggest that the slowing of quinpirole-induced current by SKF38393 is dependent upon intracellular calcium.

3.6. PKA underlies the slowing of D2 current rundown by AMPK

Having characterized the pharmacology of SKF38393, we next investigated second messenger systems that might underlie the ability of an AMPK activator to slow rundown of quinpirole-induced current. We proceeded to evaluate a role for cAMP by testing whether or not forskolin could substitute for SKF38393 in the slowing of quinpirole-induced current by A769662. Quinpirole and forskolin were superfused simultaneously; SCH39166 (10 μM) was added to the superfusate 10 min before in order to block possible D1-like receptor stimulation by endogenous dopamine.

Although A769662 did not produce a statistically significant change in rundown of quinpirole-induced current when recorded in the presence of forskolin (F_(1,15.00) = 2.86, P = 0.111459; mixed model), Fig. 6A shows that the effect of forskolin plus A769662 on current rundown was reduced substantially when recording with pipettes that contained PKI. Mixed model analysis showed there was a significant effect of PKI on rundown of current evoked by quinpirole plus forskolin in the presence of A769662 (F_(1,11.74) = 23.18, P = 0.000450). In fact, rundown of quinpirole-induced current with forskolin, A769662 and PKI was not significantly different from rundown of current evoked by quinpirole alone (F_(1,9.80) = 3.04, P = 0.112545; mixed model), which suggests that PKI completely removed the influence of forskolin and A769662 on current mediated by D2 autoreceptors. Effects of these agents on sulpiride-sensitive currents are shown in Fig. 6B. Sulpiride-sensitive current evoked by quinpirole, forskolin and A769662 (34.0 ± 4.9 pA, n = 7) was significantly larger than that evoked by quinpirole plus forskolin (18.5 ± 2.5 pA, n = 7; t(8.875) = 2.808, P = 0.020721, Welch’s t test). Furthermore, the addition of PKI caused a significant reduction in sulpiride-sensitive current (0.2 ± 0.1 pA, n = 4) compared to that recorded in quinpirole, forskolin and A769662 (t(6.008) = 6.866, P = 0.000467, Welch’s t test).

Fig. 6. cAMP and PKA mediate the ability of AMPK stimulation to slow the rundown of current evoked by quinpirole plus SKF38393.

A) Slowing of current rundown by A769662 is completely blocked by the PKA inhibitor PKI. B) A769662 increases sulpiride-sensitive current evoked by quinpirole plus forskolin, and the effect is blocked by PKI. C) PKI prevents the ability of A769662 to slow the rundown of current evoked by quinpirole plus SKF38393 (*). The addition of U73122 to PKI further accelerates the rundown of current evoked by quinpirole plus SKF38393 compared to quinpirole, SKF38393 and PKI (#). D) PKI decreases sulpiride-sensitive current evoked by quinpirole plus SKF38393 and A769662. The addition of U73122 to PKI completely blocks sulpiride-sensitive currents evoked by quinpirole plus SKF38393 and A769662. Current-decay plots were analyzed with a mixed model followed by Sidak pairwise comparison tests, whereas sulpiride-sensitive currents were analyzed with Welch’s t tests: * or #, P < 0.05; ** or ##, P < 0.01; *** or ###, P < 0.001.

Because results with forskolin suggest that cAMP generation enables the slowing of quinpirole-induced current rundown by AMPK, we returned to experiments with SKF38393 to test if PKA and/or the Epac2/PLC pathway might also be necessary for the action of AMPK. Results in Fig. 6C show that the rundown of quinpirole-induced current with SKF38393 and A769662 was not affected by the Epac2 inhibitor ESI05 (F_(1,13.25) = 0.113, P = 0.742). In contrast, PKI accelerated the rundown of quinpirole-evoked current when recorded in SKF38393 and A769662. Mixed model analysis showed there was a significant effect of PKI on rundown of current evoked by quinpirole plus SKF38393 and A769662 (F_(1,13.71) = 10.03, P = 0.007007). Rundown of quinpirole-induced current in SKF38393, A769662 and PKI was not significantly different to the rundown of current evoked by quinpirole plus SKF38393 (F_(1,18.27) = 0.05, P = 0.832974), which suggests that PKI blocked the AMPK component while leaving the effect of D1-like receptor stimulation unchanged. Finally, the combination of PKI and U73122 caused a further acceleration of rundown of current evoked by quinpirole plus SKF38393 in the presence of A769662, beyond that of PKI alone (F_(1,11.66) = 17.73, P = 0.001285); indeed, the resulting current decay profile was not different from that of quinpirole alone (F_(1,11.32) = 2.98, P = 0.111316; mixed model analysis).

Effects of these agents on sulpiride-sensitive currents are shown in Fig. 6D. Sulpiride-sensitive current evoked by quinpirole plus SKF38393 and A769662 was not affected by ESI05 (36.2 ± 7.7 pA, n = 6; t(11) = 0.553, P = 0.591, t test). However, recording with PKI in pipettes caused a significant reduction in sulpiride-sensitive current (12.7 ± 3.3 pA, n = 8) compared to that evoked by quinpirole plus SKF38393 and A769662 in the absence of PKI (t(13) = 2.880, P = 0.012893, t test). The addition of U73122 to PKI caused a further significant reduction in sulpiride-sensitive current (0.1 ± 0.2 pA, n = 5) compared to that recorded in quinpirole, SKF38393, A769662 and PKI (t(7.038) = 3.817, P = 0.006501,Welch’s t test). These results suggest that PKA is required for the AMPK-dependent slowing of rundown of current evoked by quinpirole plus SKF38393. However, inhibition of both PKA and Epac2/PLC pathways are required to fully block the effect of D1-like receptor stimulation on rundown of quinpirole-induced current.

3.7. AMPK augments the inhibitory effect of dopamine on firing rate

We recorded SNC neurons under current clamp to study the effect of the AMPK activator A769662 on dopamine-induced inhibition of action potential discharge. As in voltage-clamp recordings, dopamine was superfused for 25 min, and sulpiride was added during the last 5 min. As shown in Fig. 7A, superfusion with dopamine (100 μM) caused a brief hyperpolarization and inhibition of firing. But as shown in Fig. 7B, the duration of dopamine-induced hyperpolarization was prolonged significantly when the pipette contained A769662 (10 μM). On average, neurons began firing spontaneously 9.5 ± 2.3 min after starting dopamine superfusion under the control condition (n = 6), whereas neurons recorded with A769662 did not fire spontaneously until sulpiride was added to the superfusate (n = 5; t(5.000) = 5.068, P = 0.003875, Welch’s t test). Treatment with sulpiride (1 μM) completely reversed the dopamine-induced hyperpolarization in cells recorded with A769662 (n = 5). In contrast, the maximum dopamine-induced hyperpolarization under the control condition (6.5 ±1.7 mV, n = 7) was not significantly different from that evoked by dopamine with A769662 (8.8 ±1.6 mV, n = 5; t(10) = 0.989, P = 0.346, t test). A control experiment, illustrated in Fig. 7C, shows that A769662 does not affect firing rate in the absence of dopamine. Spontaneous firing rates were measured in 5 min epochs at the beginning of recording, after 20 min of recording, and after sulpiride superfusion. The initial firing rate (0.57 ± 0.08 Hz, n = 5) did not differ from that after 20 min of recording (0.65 ± 0.08 Hz; t(4) = 1.633, P = 0.178, paired t test). In none of the recordings with A769662 did the neuron stop firing in the absence of dopamine. Furthermore, the firing rate during sulpiride treatment (0.69 ± 0.06 Hz, n = 5) did not differ from that measured before sulpiride (t(4) = 1.170, P = 0.307). Taken together, these results show that an AMPK activator significantly prolongs the inhibitory effect of dopamine on neuronal excitability while having no significant effect on basal neuronal excitability in the absence of dopamine.

Fig. 7. AMPK activation prolongs inhibition of firing by dopamine (100 μM).

A) Control recording showing that continuous superfusion of dopamine evokes a brief hyperpolarization and inhibition of firing rate. B) Dopamine-induced hyperpolarization is prolonged when the pipette contained A769662 (10 μM). Vertical deflections during the hyperpolarization are artifacts created by passing depolarizing current pulses to test for evoked firing. C) Recording with A769662 in the pipette had no effect on spontaneous firing rate in the absence of dopamine.

4. Discussion

Although our results show that activators of AMPK slow the rundown of current evoked by a desensitizing concentration of dopamine, slowing of quinpirole-evoked current required the presence of a D1-like agonist (SKF38393). Moreover, the D1-like agonist also slowed the rundown of quinpirole-induced current even in the absence of an AMPK activator. Pharmacological antagonist experiments showed that this effect of a D1-like agonist required activation of either PKA or Epac2. BAPTA accelerated the rundown of current evoked by quinpirole plus SKF38393, which suggests involvement of calcium. Finally, the effect of AMPK on rundown of current evoked by quinpirole plus SKF38393 required PKA but not Epac2. The proposed pathways involved in the slowing of D2 autoreceptor desensitization by D1-like receptors and AMPK are summarized in Fig. 8.

Fig. 8. Schematic showing proposed interactions between D1-like receptors, AMPK and D2 autoreceptor desensitization.

Stimulation of D1-like receptors can slow D2 autoreceptor desensitization by activation of either the PKA or Epac2 pathway, whereas AMPK only requires PKA activity to cause further slowing of D2 autoreceptor desensitization. Both PKA and PLC can increase intracellular calcium, which activates neuronal calcium sensor (NCS1) to slow D2 autoreceptor desensitization.

4.1. cAMP generation by D1-like receptors mediates D2 autoreceptor desensitization

Our results show that AMPK slows D2 autoreceptor desensitization by augmenting the effect of D1-like receptors. Therefore, in order to understand the action of AMPK, it was important to first characterize how D1-like receptors slow D2 autoreceptor-dependent current rundown.

Our finding that forskolin could mimic a D1-like agonist clearly suggests that slowing of D2 autoreceptor desensitization is mediated by a cAMP-dependent mechanism. Results with antagonists of PKA and Epac2 suggest that activation of either these pathways could slow rundown of D2-evoked current, and block of both pathways was needed to prevent the effect of a D1-like agonist. It is interesting to note that slowing of current rundown by forskolin was sensitive to block of PKA but not to block of the Epac2/PLC pathway. Because forskolin has been shown to activate both PKA and Epac pathways in a variety of tissues (García-Morales et al., 2017; Mei et al., 2002), it is surprising that we found the effect of forskolin was dependent only upon PKA. However, it is possible that the subcellular distribution of Epac2 in SNC dopamine neurons does not permit activation by cAMP that is generated by forskolin. Also, it has been reported that forskolin activates the transmembrane but not the soluble subtype of adenylyl cyclase (Kamenetsky et al., 2006). Thus, it is possible that forskolin might not activate the subtype of adenylyl cyclase that is linked by cAMP to Epac2 (Kamenetsky et al., 2006). These factors might explain why the effect of forskolin was sensitive to block by PKA and not Epac2.

4.2. Role of calcium in slowing D2 autoreceptor desensitization

Our finding that BAPTA abolished the effect of a D1-like agonist to slow quinpirole-induced current rundown suggests that intracellular calcium is required to slow D2 autoreceptor desensitization. Epac2 likely causes release of intracellular stores of calcium through its ability to activate PLC and subsequent generation of inositol triphosphate (IP3) (Baljinnyam et al., 2010; Ferrero et al., 2013). Parallel activation of PKA can also cause release of internal calcium stores by phosphorylating the IP3 receptor and thus sensitizing the receptor to IP3 (Taylor, 2017). Although PKA can also increase calcium influx through voltage-gated calcium channels (Wang and Sieburth, 2013; Kim et al., 2006), this is unlikely in our study because experiments were performed under voltage clamp. Calcium-dependent slowing of D2 receptor desensitization has been shown to be mediated by activation of neuronal calcium sensor-1 (NSC-1) in SNC dopamine neurons (Dragicevic et al., 2014). Moreover, Kabbani et al (2002) showed that the NSC-1-dependent slowing of D2 receptor internalization could me mimicked by forskolin, suggesting involvement of a cAMP-dependent mechanism. Although these studies linked receptor internalization to desensitization, other studies suggest that D2 receptor desensitization could be mediated by a reduction in G protein coupling (Robinson et al., 2018).

In contrast to our results, other studies have reported that intracellular calcium chelation or block of calcium release from internal stores prevented rundown of currents evoked by dopamine in midbrain dopamine neurons (Beckstead and Williams, 2007; Perra et al., 2011; Gantz et al., 2015). These studies generally looked at relatively brief applications of dopamine, and it is possible that longer durations of D2 and D1 receptor stimulation are required to observe D2 receptor internalization mediated by β-arrestin, as opposed to short-term desensitization caused by uncoupling of G protein (Beaulieu and Gainetdinov, 2011). These studies also utilized the NMDA channel blocker MK-801 in preparing brain slices, which may be important because D1-like receptors have been shown to have actions on NMDA receptors that affect sensitization to dopaminergic drugs (Schilström et al., 2006; Heshmati, 2009; Mao et al., 2011). Moreover, MK801 has been shown to affect behavioral responses to D1 and D2 agonists (Kim et al., 1996; Asin et al., 1996). Finally, a difference in rodent species (rats versus mice) might be a significant factor in these conflicting results on the calcium dependence of D2 autoreceptor desensitization.

4.3. D1-like receptors in the SNC

Our results showed that the antagonist SCH39166 blocked the slowing of quinpirole-induced current rundown in the presence of the D1-like agonist SKF38393 with or without an AMPK activator. These results clearly implicate involvement of D1-like receptors in slowing D2 autoreceptor desensitization. Because selective D1 versus D5 receptor antagonists are not available, our findings could be mediated by either of these receptors. D1 receptors are plentiful in the substantia nigra, but the vast majority is located on nerve terminals of the striatonigral pathway (Caille et al., 1996; Levey et al., 1993). Although several studies failed to find evidence of D1 receptor expression in SNC dopamine neurons (Hurd et al., 2001; le Moine et al., 1991), immunochemical studies have shown that midbrain dopamine neurons express D5 receptors (Ciliax et al., 2000; Khan et al., 2000). One immunohistological study reported weak staining for both D1 and D5 receptors in human midbrain dopamine neurons (Reyes et al., 2013). In our studies, AMPK activators, as well as the PKA inhibitor, were contained in internal pipette solutions, which increase the likelihood that these agents were acting within the recorded neuron. Also, our studies showed that TTX did not block the ability of a D1-like agonist to slow the rundown of quinpirole-induced current. Although these considerations reduce the possibility that a D1-like agonist or AMPK activator acts remotely on adjacent cells to slow D2 autoreceptor desensitization, this possibility cannot be ruled out.

4.4. Controversies on the effect of D1 on D2 receptor desensitization

In apparent agreement with our results, Momiyama et al (1993) showed that microiontophoresis of SKF38393 onto VTA dopamine neurons augmented the effect of D2 receptor agonists to reduce neuronal firing evoked by antidromic stimulation of the nucleus accumbens. However, ligands were applied for a short duration (60 sec), which might be too brief to induce receptor desensitization. In contrast to our results, Nimitvilai and Brodie (2010) reported that a D1-like agonist facilitated recovery from quinpirole-induced inhibition of spontaneous firing of VTA neurons, a phenomenon they termed “dopamine-inhibition reversal”. When applied alone, inhibition of firing by quinpirole showed no desensitization. However, they reported that desensitization to the effect of quinpirole could be induced by a phorbol ester but not by forskolin, which suggested desensitization was mediated by protein kinase C (PKC) (Nimitvilai et al., 2012a; Nimitvilai et al., 2012b). We agree that there is much evidence to suggest that PKC-dependent mechanisms can mediate D2 receptor internalization (Beaulieu and Gainetdinov, 2011; Celver et al., 2013). However, our data suggests that D1-like receptor stimulation predominantly acts to slow D2 autoreceptor desensitization--via cAMP-dependent mechanisms--rather than to promote desensitization. Possible reasons why our results differ from those of others include the age of animals (young versus adult rats), site of recordings (VTA versus SNC), and method of recordings (extracellular versus whole-cell patch clamp). Perhaps the most important difference is the concentration of ligands, in that the study by Nimitvilai and Brodie (2010) used low concentrations of dopamine agonists intended to inhibit spike firing, whereas we utilized high concentrations with the intention to induce autoreceptor desensitization. Because we used high concentrations of dopamine and quinpirole, our results might be more applicable to situations relevant to strong dopamine receptor stimulation, such as in stimulant abuse or in the treatment of Parkinson’s disease, as opposed to physiologic levels of dopaminergic neurotransmission.

It should be noted that Aversa et al (2018) reported that the dopamine transporter can have a significant effect on the rundown of dopamine- and baclofen-induced current in whole-cell recordings of SNC dopamine neurons. These authors showed that the apparent rundown of current evoked by 100 μM dopamine or 2 μM baclofen was accelerated by inward current generated when dopamine was taken up by the transporter. In our studies of current rundown using quinpirole and SKF38393, a role of the dopamine transporter can be ruled out because neither ligand is a substrate for the transporter. However, it is possible that the transporter could play a role in our studies on the effects of AMPK activators on rundown of dopamine-induced current. Further studies are warranted to investigate a possible role of the transporter on the effects of AMPK on rundown of current evoked by dopamine.

4.5. AMPK slows D2 desensitization by augmenting the effect of D1-like receptors

Results of our study clearly show that stimulation of D1-like receptors is required for an AMPK activator to cause further slowing of D2 autoreceptor desensitization. However, we found that the effect of AMPK activation was completely blocked by inhibition of PKA, despite evidence that D1-like receptors can slow cAMP-dependent D2 autoreceptor desensitization by activating either the PKA or Epac pathways. Although block of PKA prevented the effect of AMPK, it is not clear if AMPK acts directly on PKA or downstream to PKA to augment the slowing of D2 autoreceptor desensitization. There are, however, several examples in the literature showing that PKA activity can lead to phosphorylation and activation of AMPK (Park et al., 2012; Hurtado de Llera et al., 2014). There are also reports that AMPK can facilitate PKA activity (Chen et al., 2013), which raises the possibility that the PKA and AMPK pathways represent a positive feedback loop (Stone et al., 2012). Further research is needed to describe the mechanism by which the ability of PKA to reduce D2 autoreceptor desensitization is amplified by AMPK activity.

AMPK might slow the rundown of D2 autoreceptor-mediated current by reducing the rate of D2 receptor internalization. In support of this hypothesis, Kabbani et al (2002) showed that the calcium-dependent activation of NCS-1 reduced D2 receptor internalization in HEK cells. However, it is also possible that AMPK increases the surface expression of D2 receptors, such as has been shown for K-ATP channels in pancreatic beta-cells (Chen et al., 2013; Wu et al., 2015), GABA-A receptors in hippocampus (Fan et al., 2019), and the glucose transporter in cerebellar neurons (Weisová et al., 2009). Finally, it is possible that AMPK acts to increase the coupling of D2 receptors to G protein-coupled second messengers (Robinson et al., 2018). Further work is needed to evaluate these possible mechanisms for slowing D2 receptor desensitization by AMPK and D1-like receptors.

4.6. Functional implications of slowed D2 autoreceptor desensitization

Our data on spontaneous firing of action potentials (Fig. 7) illustrates that an AMPK activator can augment the inhibitory influence of D2 autoreceptors on neuronal excitability. Our results that used high concentrations of dopamine may be especially relevant under conditions of intense and prolonged D2 receptor stimulation that encourage receptor desensitization. Because D2 autoreceptors exert negative feedback on the excitability of dopamine neurons, a reduction in autoreceptor function could significantly increase dopamine release from nerve terminals and thereby potentiate the psychomotor effects of drugs of abuse (Ford, 2014; Bello et al., 2011). In VTA dopamine neurons, D2 autoreceptor desensitization has been shown to facilitate the development of long-term potentiation (LTP) (Madhavan et al., 2013). Furthermore, LTP in VTA neurons has been shown to facilitate dopamine release from nerve terminals (Stuber et al., 2008) and has been shown to underlie some aspects of behavioral sensitization to drugs of abuse (Zweifel et al., 2008; Heshmati, 2009). Previous studies have shown that development of LTP in VTA neurons depends upon activation of D1-like receptors (Schilström et al., 2006; Thomas, Jr. et al., 1996). However, our studies suggest that activation of D1-like receptors should slow desensitization of D2 autoreceptors and would thus interfere with development of LTP. It is not clear if regulation of autoreceptors in the VTA is the same as in SNC dopamine neurons, and to our knowledge LTP has not yet been described in the SNC. It is also unknown how AMPK activation might influence LTP in VTA neurons, let alone in the SNC. Because the dopaminergic input from SNC to striatum has been implicated in many behaviors including habit formation, learning, and goal-oriented behavior (Aosaki et al., 1994; Schultz, 1998; Shan et al., 2015), it will be important to investigate how alterations in autoreceptor function by AMPK and D1-lke receptors might influence dopaminergic transmission in the nigrostriatal pathway.

4.7. AMPK and conservation of energy

In somatic tissues, AMPK acts to conserve energy and promote ATP synthesis. Thus, AMPK inhibits gluconeogenesis and synthesis of protein and fatty acids while also promoting glucose uptake and increasing mitochondrial function (Hardie et al., 2003; Lang and Föller, 2014). Because action potentials cause Na⁺ and Ca²⁺ influx that must be removed by ATP-dependent exchange mechanisms, we speculate that AMPK might conserve energy in neurons by reducing excitability. In support of this hypothesis, AMPK has been reported to reduce desensitization of GABA-B receptors and increase the surface expression of GABA-A receptors in hippocampus (Kuramoto et al., 2007; Fan et al., 2019). AMPK has also been reported to shift the voltage dependence of Kv2.1 K⁺ channels to more negative potentials (Hardie et al., 2012), reduce L-type Ca²⁺ currents (Huang et al., 2015), and reduce Ca²⁺ oscillations in cerebellar granule cells (Weisová et al., 2013). Our own studies have shown that AMPK activation augments K-ATP currents in subthalamic nucleus neurons and in VTA and SNC dopamine neurons (Shen et al., 2014; Shen et al., 2016; Wu et al., 2017). In an apparent opposition to the energy conservation hypothesis, AMPK has been reported to inhibit the large conductance calcium-activated K⁺ channel in carotid body cells, making them more excitable (Ross et al., 2011). However, the end result of this action is to signal cardiorespiratory centers in the medulla to increase respiration, which might be interpreted as a mechanism to promote net energy production. The present study, which showed that AMPK slows the desensitization of D2 autoreceptors, is consistent with the hypothesis that AMPK acts to inhibit neuronal excitability. Although the hypothesis that AMPK conserves energy by inhibiting neuronal excitability may be overly simplistic, it may provide a useful framework for future studies.

5. Conclusions

Our studies suggest that stimulation of D1-like receptors slows the rundown of D2 autoreceptors in SNC dopamine neurons. The cAMP-dependent action of D1-like receptors can be activated by either the PKA or Epac2 pathway, and both pathways must be blocked to prevent the slowing of D2 autoreceptor desensitization. Results also suggest that AMPK causes further slowing of D2 autoreceptor desensitization by augmenting the effect of D1-like receptors. But in contrast to D1-like receptors, the action of AMPK is completely dependent upon the PKA pathway. Because slowed D2 autoreceptor desensitization exerts an inhibitory influence on dopamine neuronal excitability, this action of AMPK is consistent with its role to conserve neuronal energy.

Highlights:

• Prolonged dopamine exposure desensitizes D2 autoreceptors in SNC dopamine neurons

• AMPK activators slowed dopamine-induced D2 autoreceptor desensitization

• Slowing of quinpirole-induced desensitization by AMPK required D1-like stimulation

• The effect of AMPK was completely dependent upon PKA activity

• Reduced D2 desensitization augments negative feedback control of dopamine neurons

Abbreviations:

aCSF

artificial cerebrospinal fluid

AMPK

5’-AMP-activated protein kinase

BAPTA

1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid

EGTA

ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid

EPAC

exchange protein directly activated by cAMP

IP3

inositol triphosphate

K-ATP channel

ATP-sensitive K⁺ channel

LTP

long-term potentiation

NCS

neuron calcium sensor

PKA

protein kinase A

PKC

protein kinase C

PLC

phospholipase C

SNC

Substantia nigra zona compacta

VTA

Ventral tegmental area