Spinal AMP kinase activity differentially regulates phrenic motor plasticity

Abstract

Acute intermittent hypoxia (AIH) elicits phrenic motor plasticity via multiple distinct cellular mechanisms. With moderate AIH, phrenic motor facilitation (pMF) requires G_q protein-coupled serotonin type 2 receptor activation, ERK MAP kinase activity, and new synthesis of brain-derived neurotrophic factor. In contrast, severe AIH elicits pMF by an adenosine-dependent mechanism that requires exchange protein activated by cAMP, Akt, and mammalian target of rapamycin (mTOR) activity, followed by new tyrosine receptor kinase B protein synthesis; this same pathway is also initiated by G_s protein-coupled serotonin 7 receptors (5-HT₇). Because the metabolic sensor AMP-activated protein kinase (AMPK) inhibits mTOR-dependent protein synthesis, and mTOR signaling is necessary for 5-HT₇ but not 5-HT₂ receptor-induced pMF, we hypothesized that spinal AMPK activity differentially regulates pMF elicited by these distinct receptor subtypes. Serotonin type 2A receptor [5-HT_2A; (±)-2,5-dimethoxy-4-iodoamphetamine hydrochloride] or 5-HT₇ (AS-19) receptor agonists were administered intrathecally at C₄ (3 injections, 5-min intervals) while recording integrated phrenic nerve activity in anesthetized, vagotomized, paralyzed, and ventilated rats. Consistent with our hypothesis, spinal AMPK activation with 2-deoxyglucose or metformin blocked 5-HT₇, but not 5-HT_2A receptor-induced pMF; in both cases, pMF inhibition was reversed by spinal administration of the AMPK inhibitor compound C. Thus, AMPK differentially regulates cellular mechanisms of serotonin-induced phrenic motor plasticity.

NEW & NOTEWORTHY Spinal AMP-activated protein kinase (AMPK) overactivity, induced by local 2-deoxyglucose or metformin administration, constrains serotonin 7 (5-HT₇) receptor-induced (but not serotonin type 2A receptor-induced) respiratory motor facilitation, indicating that metabolic challenges might regulate specific forms of respiratory motor plasticity. Pharmacological blockade of spinal AMPK activity restores 5-HT₇ receptor-induced respiratory motor facilitation in the presence of either 2-deoxyglucose or metformin, showing that AMPK is an important regulator of 5-HT₇ receptor-induced respiratory motor plasticity.

INTRODUCTION

Multiple independent cellular mechanisms give rise to phenotypically similar phrenic motor facilitation (pMF), a persistent poststimulus increase in phrenic nerve activity. Spinal G_q protein-coupled serotonin type 2A receptor (5-HT_2A) activation elicits the “Q pathway” to pMF, which requires ERK MAP kinase, tyrosine receptor kinase B (TrkB), and PKC-θ activity, and new brain-derived neurotrophic factor protein synthesis. Alternatively, spinal G_s protein-coupled serotonin 7 receptor (5-HT₇) activation gives rise to pMF via the “S pathway,” which requires intracellular cAMP signaling, exchange protein activated by cAMP, Akt, and mammalian target of rapamycin complex 1 (mTORC1) activity, and new TrkB protein synthesis (11, 13, 14). Although we have learned a great deal about these distinct mechanisms of pMF, relatively little is known concerning factors that differentially regulate their expression.

Factors such as inflammation and/or energy balance may differentially regulate the Q and S pathways to pMF. For example, neuroinflammation selectively constrains Q pathway-dependent pMF (2) by a p38 MAP kinase (2, 19) and okadaic acid-sensitive protein phosphatase-dependent mechanism (Ref. 44; unpublished observations); inflammation has minimal impact on the S pathway (1). On the other hand, AMP-activated protein kinase (AMPK) regulates energy supply to maintain cellular functions and may uniquely regulate the S pathway. When activated by metabolic stress, such as severe hypoxia, AMPK inhibits mTORC1 signaling (4, 5, 7, 20, 37, 43) and regulates certain forms of hippocampal synaptic plasticity (37). On the other hand, glycolytic flux inhibition with 2-deoxyglucose (2-DG), and presumed AMPK activation, has minimal impact on Q pathway-dependent pMF following moderate acute intermittent hypoxia (27). Thus, AMPK likely regulates some, but not all, forms of phrenic motor plasticity.

Here, we test the hypothesis that AMPK activity regulates the mTORC1-dependent S pathway to pMF, but not the mTORC1-independent Q pathway (11, 13). Specifically, we tested the hypothesis that spinal AMPK activity impairs 5-HT₇, but not 5-HT_2A receptor-induced pMF in anesthetized, paralyzed, and ventilated rats. The Q and S pathways were induced by intrathecal (C₄) injections of selective 5-HT_2A or 5-HT₇ receptor agonists, and AMPK activity was regulated via intrathecal injection of activators (2-DG, metformin) with/without AMPK inhibition [compound C (cC)]. We demonstrate that 2-DG and metformin both abolish 5-HT₇ (not 5-HT_2A) receptor-induced pMF, and this effect is reversed by cC. Differential regulation of phrenic motor plasticity may be particularly advantageous in preventing its expression during metabolic stress to minimize unnecessary energetic costs associated with protein synthesis (10, 32).

METHODS

Experimental animals.

Experiments were performed on adult (300–400 g) male Sprague Dawley rats (208A Colony, Envigo), housed in pairs and maintained on a 12:12-h light-dark cycle with access to food and water ad libitum. All experiments were approved by the University of Florida Institutional Animal Care and Use Committee and were conducted in accordance with the Animal Welfare Act, the Public Health Service Policy on Humane Care and Use of Laboratory Animals, and the National Institutes of Health Guide for the Care and Use of Laboratory Animals (2011).

Neurophysiological experiments.

Surgical procedures and electrophysiological recording techniques were described in detail previously (3, 35). Briefly, rats were induced with 3% isoflurane in 100% O₂ at 3 L/min in a 5-liter Plexiglas chamber. Rats were then transferred to a heated surgical table, and anesthesia was maintained via nose cone. The rats were tracheotomized, ventilated (~2.5 mL, 70 breaths/min; VentElite small animal ventilator; Harvard Apparatus, Holliston, MA), and bilaterally vagotomized in the midcervical region. End-tidal CO₂ was monitored using a flow-through capnograph (Capnogard; Novametrix, Wallingford, CT) and maintained within a normal physiological range. The rats were then slowly converted to urethane anesthesia by reducing isoflurane by 0.5% every 3 min while urethane (2.1 g/kg, diluted in double-distilled water; Sigma-Aldrich, St. Louis, MO) was administered via tail vein catheter (24 gauge; Surflash, Somerset, NJ) with a syringe pump (6 mL/h). Based on isoflurane clearance rates (16), at least 1 h was allowed after anesthetic conversion was complete to minimize isoflurane effects on breathing. Adequate anesthetic depth was assured by the absence of a withdrawal reflex and/or blood pressure response to toe pinch. A polyethylene catheter was inserted in the right femoral artery to monitor blood pressure (Argon Pressure Transducer; DTXPlus, Plano, TX) and to draw arterial blood samples (ABL 90 Flex; Radiometer) in heparinized plastic capillary tubes (electrolyte balanced 70 IU/mL, 70 μL; Radiometer).

Rats were positioned in sternal recumbency, and muscles overlying the dorsal aspect of the cervical spinal cord were exposed through a midline incision. Muscles were separated to allow C₂ laminectomy, and a silicone catheter (2 Fr, 0.6 mm OD; Access Technologies, Skokie, IL) attached to a 50-μL Hamilton syringe was inserted through a small hole in the dura mater. The catheter was advanced caudally to the C₃–C₄ spinal segment and stabilized in position. The left phrenic nerve was isolated and cut distally. The epineurium was gently retracted, and the central end of the phrenic nerve was recorded via saline-filled bipolar glass suction electrodes. Phrenic nerve activity was processed (10,000× amplified and band-pass filtered, 0.3–5 kHz, model 1700; A-M Systems, Sequim, WA), digitized (CED 1401; Cambridge Electronic Design, Cambridge, UK), and analyzed via Spike2 software (version 8.08; Cambridge Electronic Design). Rats were paralyzed with the nicotinic acetylcholine receptor blocker pancuronium bromide (3 mg/kg iv; Sigma-Aldrich; St. Louis, MO).

Drugs.

(±)-2,5-Dimethoxy-4-iodoamphetamine hydrochloride (5-HT_2A agonist) and AS-19 (5-HT₇ agonist) were purchased from Tocris Biosciences (Minneapolis, MN). The AMPK activators 2-DG and metformin were purchased from USP (Rockville, MD) and Tocris Biosciences (Minneapolis, MN), respectively; the AMPK inhibitor cC (dorsomorphin) was purchased from Tocris Biosciences. Upon arrival, drugs were diluted in 100% dimethyl sulfoxide (DMSO), separated into aliquots, and stored at −20°C. On the day of experiments, the stock vials were thawed and further diluted in artificial cerebrospinal fluid (ACSF; in mM: 125 NaCl, 3 KCl, 2.5 CaCl₂2H₂O, 1.25 MgSO₄-7H₂O, 1.25 KH₂PO₄, 25 NaHCO₃, bubbled with 95% O₂-5% CO₂) to appropriate concentrations. Maximum DMSO concentration in the final solution was <10%, varying slightly depending on drug solubility.

Experimental protocols.

In all experiments, the inspired oxygen concentration was 60%, and the inspired CO₂ concentration was adjusted for each rat to maintain end-tidal CO₂ between 2 and 3 mmHg above the CO₂ recruitment threshold. The apneic threshold was determined via stepwise reductions in inspired CO₂ and reported as the end-tidal CO₂ causing a sustained apnea (i.e., 60 s). Recruitment threshold was achieved by progressively increasing the inspired CO₂ concentration and is reported as the end-tidal CO₂ when phrenic activity resumed. Stable baseline phrenic nerve activity was monitored for at least 20 min, and consistency of arterial blood gas values was confirmed with multiple blood samples. The final blood sample during baseline was used as a reference for subsequent blood samples at 30, 60, and 90 min after drug injections. Experiments were analyzed only if they met the following a priori criteria: 1) $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ within 1.5 mmHg and standard base excess (BE) within 3 meq/L of baseline values; 2) $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ above 150 mmHg; and 3) blood pressure changes from beginning to end of experiments <30 mmHg. Only complete experiments were included for statistical analyses.

We initially verified that episodic spinal injections of 5-HT_2A (100 μM, 6 μL × 3) and 5-HT₇ (10 μM, 6 μL × 3) receptor agonists are sufficient to elicit consistent pMF. Rats were then pretreated (intrathecally) with 2-DG (1 mM, 10 µL) 20 min before 5-HT_2A (2-DG + 5-HT_2A agonist) or 5-HT₇ (2-DG + 5-HT₇ agonist) receptor activation. Different rat groups were pretreated (intrathecally) with metformin (2.5 or 5 mM, 10 µL) 20 min before 5-HT_2A (2.5 mM Met + 5-HT_2A agonist) or 5-HT₇ receptor activation (5 mM Met + 5-HT₇ agonist or 5 mM Met + 5-HT₇ agonist). Both drugs used to activate AMPK in this study have known multitarget effects. To confirm the specificity of AMPK involvement in observed effects, cC (2 mM) was added to 2-DG and metformin solutions and administered (10 µL) 20 min before 5-HT₇ receptor activation (2-DG/cC + 5-HT₇ agonist and 2.5 mM Met/cC + 5-HT₇ agonist, respectively). AMPK activator/inhibitor doses were chosen based on previous studies showing consistent physiological effects in response to intrathecal drug injections in the lumbar spinal cord (26).

Data analyses.

Data distribution was considered normal based on visual inspection of histograms and normal probability plots. Raw phrenic nerve activity was smoothed (0.05-s time constant) and rectified for off-line analysis. Integrated phrenic burst amplitude was averaged over 1 min at baseline and 30, 60, and 90 min after drug injection. These 1-min bins at selected times after drug injection were compared with baseline as a delta (time point − baseline) and as a percent change [percent change = (time point – baseline)/baseline × 100]. A mixed two-way ANOVA was used to evaluate overall differences on respiratory frequency and phrenic burst nerve amplitude between groups after drug injections and within groups relative to baseline. A one-way ANOVA was used to compare change in integrated phrenic burst amplitude among groups from baseline to 90 min after intrathecal drug injections. This approach complements pMF analysis as a percent change, since the absolute change in integrated phrenic burst amplitude accounts for respiratory motor output variability within and between groups during baseline, yet is minimally affected by extreme values or potential inconsistencies among group mean values. Tukey’s honestly significant difference tests were used when significant differences were indicated by the ANOVA. Values are expressed as means ± 1SD, and significant differences were considered at α < 0.05. All statistical analyses were performed in R (version 3.4.3).

RESULTS

Episodic spinal 5-HT_2A and 5-HT₇ receptor activation elicits pMF.

pMF following spinal 5-HT_2A and 5-HT₇ receptor activation has been studied extensively using similar experimental protocols (12, 13, 17, 18, 28). Figure 1A shows compressed electrophysiological recordings of integrated phrenic burst activity from representative rats showing that spinal 5-HT_2A (59 ± 54%; n = 4) or 5-HT₇ receptor activation alone (50 ± 31%; n = 6) causes robust pMF (Fig. 1, A and B), i.e., phrenic burst amplitude was significantly higher than baseline levels (P < 0.05), whereas no change was observed in ACSF-treated rats (−3 ± 24%; n = 6; P > 0.05). Changes in phrenic burst amplitude 90 min after intrathecal 5-HT_2A (0.03 ± 0.01 volts) or 5-HT₇ agonist (0.03 ± 0.03 volts) injections were higher than following ACSF (−0.01 ± 0.02 volts; P < 0.05; Fig. 1C).

Fig. 1.

Episodic spinal serotonin type 2A (5-HT_2A) or serotonin 7 (5-HT₇) receptor activation elicits robust phrenic motor facilitation (pMF). A: representative traces of compressed integrated phrenic burst amplitude throughout the experimental protocol. Gray broken line in each trace represents baseline phrenic burst amplitude; arrows on top indicate time of it 5-HT_2A receptor agonist (n = 4), 5-HT₇ receptor agonist (n = 6), or artificial cerebrospinal fluid (ACSF) injections (n = 6). B: %change in phrenic burst amplitude (vs. baseline) at 30, 60, and 90 min after last it injection. C: delta (vs. baseline) in integrated phrenic burst amplitude at 90 min after last it injection. D: respiratory frequency at baseline and 30, 60, and 90 min after last it injection. E: end-tidal CO₂ level at apneic and at recruitment threshold among experimental groups. Data are presented as means ± SD. Mixed 2-way ANOVA or 1-way ANOVA was used for overall group comparison (with Tukey’s adjustment when necessary), and differences were considered significant at P < 0.05. *5-HT_2A receptor agonist different from baseline. #5-HT₇ receptor agonist different from baseline.

No significant changes in respiratory frequency were observed among groups or through the course of experiments (P > 0.05; Fig. 1D), consistent with the interpretation that drugs did not reach the medulla at sufficient concentrations to affect our results. Apneic and recruitment thresholds were also similar among groups (P > 0.05; Fig. 1E). $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, standard BE, pH, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, temperature, and mean arterial pressure were maintained within the a priori specified range (Table 1).

Table 1.

Physiological variables at baseline and 30, 60, and 90 min after last it injection

Time, min	ACSF	5-HT2A	5-HT7
$\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$
Baseline	43.2 ± 3.6	46.7 ± 8.5	43.6 ± 3.1
30	42.9 ± 3.6	45.6 ± 8.5	43.7 ± 3.4
60	43.4 ± 3.1	43.8 ± 5.4	44.0 ± 3.3
90	42.7 ± 2.9	44.9 ± 9.7	44.4 ± 4.1
$\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$
Baseline	274 ± 66	292 ± 26	303 ± 15
30	264 ± 65	290 ± 22	304 ± 21
60	264 ± 66	277 ± 23	293 ± 19
90	255 ± 59	281 ± 26	289 ± 17
MAP, mmHg
Baseline	129 ± 16	121 ± 7	132 ± 19
30	125 ± 20	122 ± 9	126 ± 21
60	129 ± 22	128 ± 12	131 ± 23
90	136 ± 17	130 ± 8	121 ± 28
pH
Baseline	7.39 ± 0.01	7.36 ± 0.04	7.40 ± 0.02
30	7.40 ± 0.04	7.41 ± 0.05	7.40 ± 0.03
60	7.39 ± 0.03	7.41 ± 0.03	7.39 ± 0.03
90	7.39 ± 0.04	7.40 ± 0.05	7.39 ± 0.03
sBE
Baseline	1.3 ± 1.7	1.0 ± 2.3	2.0 ± 1.5
30	1.7 ± 1.7	3.7 ± 2.3	2.0 ± 0.9
60	1.5 ± 1.3	3.0 ± 2.0	1.9 ± 0.9
90	0.7 ± 1.6	2.6 ± 2.1	1.6 ± 0.7
Temperature, °C
Baseline	37.4 ± 0.8	37.2 ± 0.4	37.5 ± 0.5
30	37.6 ± 0.8	37.3 ± 0.2	37.1 ± 0.3
60	37.7 ± 0.7	37.4 ± 0.7	37.3 ± 0.4
90	37.6 ± 0.7	37.2 ± 0.2	37.6 ± 0.3

ACSF, artificial cerebrospinal fluid; 5-HT_2A, serotonin type 2A receptor; 5-HT₇, serotonin 7 receptor; MAP, mean arterial pressure; sBE, standard base excess.

Spinal AMPK activation blocks 5-HT₇-, but not 5-HT_2A-induced pMF.

Figure 2A shows compressed electrophysiological recordings of integrated phrenic nerve burst activity from representative rats, demonstrating that intrathecal 2-DG abolishes 5-HT₇-induced pMF (2-DG + 5-HT₇: −2 ± 14%; n = 7), i.e., 5-HT₇ receptor-induced pMF was minimal and not significantly different from baseline levels (P > 0.05), as observed with 2-DG + ACSF (1 ± 20%; n = 6; P > 0.05). cC with 2-DG restored 5-HT₇-induced pMF; 2-DG/cC + 5-HT₇ agonist (33 ± 21; n = 6) presented higher phrenic burst amplitudes vs. 2-DG + 5-HT₇ agonist without cC or baseline levels (P < 0.05; Fig. 2, A and B). Changes in integrated phrenic burst activity 90 min after intrathecal 2-DG/cC + 5-HT₇ agonist injections (0.03 ± 0.02 volts) were higher than 2-DG + 5-HT₇ agonist (0.00 ± 0.01 volts) or 2-DG + ACSF (0.00 ± 0.01 volts; P < 0.05; Fig. 2C). No significant changes in respiratory frequency were observed among groups or over the course of experiments (P > 0.05; Fig. 2D). Apneic and recruitment thresholds were similar among groups (P > 0.05; Fig. 2E). $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, standard BE, pH, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, temperature, and mean arterial pressure were maintained within the a priori specified range (Table 2).

Fig. 2.

2-Deoxyglucose (2-DG)-dependent AMP-activated protein kinase (AMPK) activation abolishes serotonin 7 receptor (5-HT₇)-induced phrenic motor facilitation (pMF), which is restored by it compound C (cC). A: representative traces of compressed integrated phrenic burst amplitude throughout experimental protocol in 2-DG + artificial cerebrospinal fluid (ACSF, n = 6)-, 2-DG + 5-HT₇ (n = 7)-, and 2-DG/cC + 5-HT₇ (n = 6)-treated rats. Gray broken line in each trace represents baseline amplitude of phrenic burst activity, and arrows on top indicate time of it 5-HT₇ receptor agonist or ACSF injections in rats pretreated with 2-DG or 2-DG plus cC. B: %change in phrenic burst amplitude (vs. baseline) at 30, 60, and 90 min after last it injection. C: delta (vs. baseline) in integrated phrenic burst amplitude at 90 min after last it injection. D: respiratory frequency at baseline and 30, 60, and 90 min after last it injection. E: end-tidal CO₂ level at apneic and at recruitment threshold among experimental groups. Data are presented as means ± SD. Mixed 2-way ANOVA or 1-way ANOVA was used for overall group comparison (with Tukey’s adjustment when necessary), and differences were considered significant at P < 0.05. *2-DG/cC + 5-HT₇ receptor agonist different from baseline. #2-DG/cC + 5-HT₇- different from 2-DG + 5-HT₇-treated rats 90 min after it drug injections.

Table 2.

Physiological variables at baseline and 30, 60, and 90 min after last it injection

Time, min	2-DG + ACSF	2-DG + 5-HT2A	2-DG + 5-HT7	2-DG/cC + 5-HT7
$\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$
Baseline	43.9 ± 3.4	40.8 ± 1.7	43.0 ± 3.8	43.0 ± 2.9
30	43.7 ± 4.2	40.4 ± 2.1	43.8 ± 3.3	43.4 ± 3.4
60	44.6 ± 3.0	42.8 ± 1.3	43.0 ± 3.2	43.8 ± 3.2
90	44.8 ± 3.4	40.7 ± 1.3	43.5 ± 3.0	43.3 ± 3.7
$\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$
Baseline	314 ± 20	304 ± 36	311 ± 27	319 ± 20
30	302 ± 21	280 ± 58	296 ± 23	319 ± 21
60	298 ± 42	282 ± 53	292 ± 31	312 ± 20
90	304 ± 29	267 ± 42	292 ± 35	305 ± 18
MAP, mmHg
Baseline	138 ± 8	127 ± 28	122 ± 18	124 ± 10
30	136 ± 12	129 ± 25	118 ± 20	117 ± 13
60	135 ± 14	130 ± 24	120 ± 19	116 ± 15
90	133 ± 23	127 ± 24	119 ± 19	108 ± 17
pH
Baseline	7.38 ± 0.02	7.41 ± 0.03	7.39 ± 0.03	7.39 ± 0.03
30	7.38 ± 0.04	7.41 ± 0.02	7.40 ± 0.02	7.40 ± 0.03
60	7.38 ± 0.03	7.41 ± 0.02	7.40 ± 0.02	7.39 ± 0.04
90	7.38 ± 0.02	7.39 ± 0.05	7.43 ± 0.13	7.39 ± 0.03
sBE
Baseline	0.7 ± 1.1	1.3 ± 1.5	1.1 ± 1.1	1.2 ± 1.8
30	0.8 ± 1.8	0.9 ± 1.9	2.3 ± 1.0	1.9 ± 0.9
60	1.0 ± 1.3	1.7 ± 1.7	1.6 ± 1.0	1.8 ± 1.5
90	0.9 ± 1.5	0.6 ± 1.6	1.1 ± 1.4	1.2 ± 1.5
Temperature, °C
Baseline	37.9 ± 0.3	37.3 ± 0.5	37.4 ± 0.4	37.5 ± 0.5
30	38.0 ± 0.4	37.4 ± 0.6	37.4 ± 0.4	37.2 ± 0.5
60	37.9 ± 0.6	37.4 ± 0.4	37.5 ± 0.4	37.1 ± 0.5
90	38.0 ± 0.6	37.3 ± 0.4	37.5 ± 0.4	37.2 ± 0.6

2-DG, 2-deoxyglucose; cC, compound C; ACSF, artificial cerebrospinal fluid; 5-HT_2A, serotonin type 2A receptor; 5-HT₇, serotonin 7 receptor; MAP, mean arterial pressure; sBE, standard base excess.

Figure 3A shows compressed electrophysiological recordings of integrated phrenic nerve burst activity from representative rats. Intrathecal 2-DG did not constrain 5-HT_2A-induced pMF (2-DG + 5-HT_2A: 36 ± 19; n = 5); phrenic burst amplitude in this group was higher than baseline (P < 0.05; Fig. 3, A and B), suggesting that AMPK activity selectively constrains the S, but not the Q, pathway to pMF. The change in integrated phrenic burst activity 90 min after intrathecal 2-DG + 5-HT_2A agonist injections (0.06 ± 0.05 volts) was significantly higher than 2-DG + ACSF (P < 0.05; Fig. 3C). No significant changes in respiratory frequency were observed among groups or through experiments (P > 0.05; Fig. 3D). Apneic and recruitment thresholds were similar among groups (P > 0.05; Fig. 3E). $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, standard BE, pH, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, temperature, and mean arterial pressure were maintained within the a priori specified range (Table 2).

Fig. 3.

2-Deoxyglucose (2-DG)-dependent AMP-activated protein kinase (AMPK) activation does not affect normal expression of serotonin type 2A receptor (5-HT_2A)-induced phrenic motor facilitation (pMF). A: representative traces of compressed integrated phrenic burst amplitude throughout the experimental protocol in 2-DG + artificial cerebrospinal fluid (ACSF, n = 6)- and 2-DG + 5-HT_2A (n = 5)-treated rats. Gray broken line in each trace represents baseline amplitude of phrenic burst activity, and arrows on top indicate time of it 5-HT_2A receptor agonist or ACSF injections in rats pretreated with 2-DG. 2-DG + ACSF group is the same reported in Fig. 2. B: %change in phrenic burst amplitude (vs. baseline) at 30, 60, and 90 min after last it injection. C: delta (vs. baseline) in integrated phrenic burst amplitude at 90 min after last it injection. D: respiratory frequency at baseline and 30, 60, and 90 min after last it injection. E: end-tidal CO₂ level at apneic and at recruitment threshold among experimental groups. Data are presented as means ± SD. Mixed 2-way ANOVA or 1-way ANOVA was used for overall group comparison (with Tukey’s adjustment when necessary), and differences were considered significant at P < 0.05. *2-DG + 5-HT_2A different from baseline.

Figure 4A shows compressed recordings of integrated phrenic nerve burst activity from representative rats, demonstrating that intrathecal metformin abolished 5-HT₇-induced pMF (2.5 mM Met + 5-HT₇ agonist: 9 ± 22%; n = 5); phrenic burst amplitude was similar to 2.5 mM Met + ACSF (4 ± 7%; n = 6; P > 0.05) or baseline (P > 0.05). 2.5 mM Met/cC + 5-HT₇ agonist (51 ± 43%; n = 6) exhibited higher phrenic burst amplitudes vs. 2.5 mM Met + ACSF or baseline (P < 0.05; Fig. 4, A and B), consistent with our findings using 2-DG (Fig. 2). Only marginal changes were observed between groups in integrated phrenic burst activity 90 min after intrathecal drug injections: 2.5 mM Met/cC + 5-HT₇ agonist: 0.04 ± 0.04 volts; 2.5 mM Met + 5-HT₇ agonist: 0.01 ± 0.03; 2.5 mM Met + ACSF: 0.01 ± 0.02 volts (P > 0.05; Fig. 4C). Slightly different results were obtained depending on how pMF was analyzed, which was probably due to the fact that normalization (i.e., as percent from baseline) removes baseline variability and, consequently, increases statistical power. No significant changes in respiratory frequency were observed among groups or during experiments (P > 0.05; Fig. 4D). Apneic and recruitment threshold levels were also similar among groups (P > 0.05; Fig. 4E); $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, standard BE, pH, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, temperature, and mean arterial pressure were maintained within a priori specified range (Table 3). Because cC competitively inhibits AMPK-γ regulatory subunits with reasonable specificity (47), the observed effects of the selected AMPK activators are likely specific and account for the findings of this study.

Fig. 4.

Low-dose metformin-dependent AMP-activated protein kinase (AMPK) activation abolishes serotonin 7 receptor (5-HT₇)-induced phrenic motor facilitation (pMF), which is restored by it compound C (cC). A: representative traces of compressed integrated phrenic burst amplitude throughout experimental protocol in 2.5 mM Met + artificial cerebrospinal fluid (ACSF, n = 6)-, 2.5 mM Met + 5-HT₇ (n = 5)-, and 2.5 mM Met/cC + 5-HT₇ (n = 6)-treated rats. Gray broken line in each trace represents baseline amplitude of phrenic burst; arrows on top indicate time of it 5-HT₇ receptor agonist or ACSF injections in rats pretreated with 2.5 mM metformin or 2.5 mM metformin plus cC. B: %change in phrenic burst amplitude (vs. baseline) at 30, 60, and 90 min after last it injection. C: delta (vs. baseline) in integrated phrenic burst amplitude at 90 min after last it injection. D: respiratory frequency at baseline and 30, 60, and 90 min after last it injection. E: end-tidal CO₂ level at apneic and at recruitment threshold among experimental groups. Data are presented as means ± SD. Mixed 2-way ANOVA or 1-way ANOVA was used for overall group comparison (with Tukey’s adjustment when necessary), and differences were considered significant at P < 0.05. *2.5 mM Met/cC + 5-HT₇ receptor agonist different from baseline. #2.5 mM Met/cC + 5-HT₇ different from 2.5 mM Met + ACSF.

Table 3.

Physiological variables at baseline and 30, 60, and 90 min after last it injection

Time, min	Met (2.5 mM) + ACSF	Met (2.5 mM) + 5-HT2A	Met (2.5 mM) + 5-HT7	Met (2.5 mM)/cC + 5-HT7
$\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$
Baseline	45.6 ± 7.0	41.5 ± 3.0	44.0 ± 2.0	45.8 ± 1.3
30	45.4 ± 6.6	42.1 ± 2.2	44.5 ± 2.2	45.9 ± 1.5
60	45.2 ± 6.7	41.4 ± 1.6	44.1 ± 2.5	46.8 ± 1.5
90	45.6 ± 6.4	41.1 ± 3.8	43.7 ± 1.8	46.0 ± 1.4
$\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$
Baseline	300 ± 54	300 ± 23	310 ± 26	310 ± 25
30	298 ± 37	297 ± 19	315 ± 12	301 ± 21
60	292 ± 34	288 ± 21	304 ± 17	293 ± 18
90	295 ± 34	284 ± 24	302 ± 13	282 ± 20
MAP, mmHg
Baseline	129 ± 21	122 ± 10	129 ± 25	131 ± 16
30	128 ± 19	132 ± 12	123 ± 25	126 ± 14
60	128 ± 23	137 ± 14	119 ± 20	123 ± 17
90	131 ± 21	134 ± 13	114 ± 15	119 ± 12
pH
Baseline	7.36 ± 0.05	7.40 ± 0.02	7.38 ± 0.02	7.37 ± 0.03
30	7.37 ± 0.04	7.41 ± 0.05	7.39 ± 0.03	7.37 ± 0.03
60	7.38 ± 0.05	7.42 ± 0.06	7.38 ± 0.02	7.37 ± 0.02
90	7.37 ± 0.04	7.41 ± 0.07	7.39 ± 0.01	7.36 ± 0.02
sBE
Baseline	0.2 ± 1.6	0.6 ± 1.6	0.8 ± 1.5	1.2 ± 2.0
30	1.1 ± 1.8	2.6 ± 5.2	2.1 ± 1.4	0.9 ± 1.9
60	1.0 ± 1.4	2.5 ± 4.0	1.3 ± 1.1	1.8 ± 1.3
90	1.0 ± 0.9	1.0 ± 3.3	1.3 ± 0.9	0.6 ± 1.3
Temperature, °C
Baseline	37.4 ± 0.7	37.7 ± 0.8	37.4 ± 0.5	37.7 ± 0.4
30	37.8 ± 0.4	37.3 ± 0.5	37.4 ± 0.5	37.6 ± 0.4
60	38.1 ± 0.6	37.3 ± 0.4	37.7 ± 0.1	37.6 ± 0.5
90	38.0 ± 0.6	37.5 ± 0.7	37.4 ± 0.3	37.6 ± 0.5

Met, metformin; cC, compound C; ACSF, artificial cerebrospinal fluid; 5-HT_2A, serotonin type 2A receptor; 5-HT₇, serotonin 7 receptor; MAP, mean arterial pressure; sBE, standard base excess.

Figure 5A shows compressed recordings of integrated phrenic nerve burst activity from representative rats, showing that intrathecal metformin did not affect 5-HT_2A-induced pMF (2.5 mM Met + 5-HT_2A: 42 ± 24; n = 4; P < 0.05; Fig. 5, A and B). Changes in integrated phrenic burst activity 90 min after intrathecal injections of 2.5 mM Met + 5-HT_2A agonist (0.03 ± 0.01 volts) were higher than in 2.5 mM Met + ACSF rats (P < 0.05; Fig. 5C). No significant changes in respiratory frequency were observed among groups or over the course of experiments (P > 0.05; Fig. 5D). Apneic and recruitment threshold levels were similar among groups (P > 0.05; Fig. 5E). $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, standard BE, pH, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, temperature, and mean arterial pressure were maintained within the a priori specified range (Table 3). The magnitude of 5-HT₇-induced pMF (without pretreatment) was significantly greater than with 2-DG or 2.5 mM Met pretreatment; however, the magnitude of 5-HT_2A-induced pMF was not affected by either 2-DG or 2.5 mM Met.

Fig. 5.

Low-dose metformin-dependent AMP-activated protein kinase (AMPK) activation does not affect normal expression of serotonin type 2A receptor (5-HT_2A)-induced phrenic motor facilitation (pMF). A: representative traces of compressed integrated phrenic burst amplitude throughout the experimental protocol in 2.5 mM Met + artificial cerebrospinal fluid (ACSF, n = 6)- and 2.5 mM Met + 5-HT_2A (n = 4)-treated rats. Gray broken line in each trace represents baseline amplitude of phrenic burst activity, and arrows on top indicate time of it 5-HT_2A receptor agonist or ACSF injections in rats pretreated with 2.5 mM metformin. Met (2.5 mM) + ACSF group is the same reported in Fig. 2. B: %change in phrenic burst amplitude (vs. baseline) at 30, 60, and 90 min after last it injection. C: delta (vs. baseline) in integrated phrenic burst amplitude at 90 min after last it injection. D: respiratory frequency at baseline and 30, 60, and 90 min after last it injection. E: end-tidal CO₂ level at apneic and at recruitment threshold among experimental groups. Data are presented as means ± SD. Mixed 2-way ANOVA or 1-way ANOVA was used for overall group comparison (with Tukey’s adjustment when necessary), and differences were considered significant at P < 0.05. *Met (2.5 mM) + 5-HT_2A receptor agonist different from baseline. #Met (2.5 mM) + 5-HT_2A different from 2.5 mM Met + ACSF.

No differences in phrenic burst amplitude were observed in rats treated with either 2-DG (2-DG + ACSF) or metformin (2.5 mM Met + ACSF) alone, suggesting that these drugs do not directly affect phrenic nerve activity. A higher metformin dose (5 mM) did not change phrenic burst amplitude in 5 mM Met + ACSF (−9 ± 10; n = 3) vs. baseline (P > 0.05) but prevented 5-HT₇-induced pMF (5 mM Met + 5-HT₇ agonist: 7 ± 25; n = 4; P > 0.05 vs. 5 mM Met + ACSF and vs. baseline). However, at this higher metformin dose, cC was ineffective at restoring 5-HT₇-induced pMF in 5 mM Met/cC + 5-HT₇ agonist (−12 ± 33; n = 3; P > 0.05; Fig. 6, A–C); changes in integrated phrenic nerve burst activity 90 min after drug injections were as follows: 5 mM Met/cC + 5-HT₇ agonist: −0.01 ± 0.04; 5 mM Met + 5-HT₇ agonist: 0.01 ± 0.02; 5 mM Met + ACSF: −0.01 ± 0.01 volts (P > 0.05). No significant changes in respiratory frequency were observed among groups or through the course of experiments (P > 0.05; Fig. 6D). Apneic and recruitment thresholds were also similar among groups (P > 0.05; Fig. 6E). $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, standard BE, pH, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, temperature, and mean arterial pressure were maintained within the a priori specified range (Table 4). There was no significant correlation between baseline $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and pMF magnitude.

Fig. 6.

Schematic illustrating cellular mechanisms giving rise to phenotypically similar pathways to phrenic motor facilitation (pMF). Our findings indicate that spinal AMP-activated protein kinase (AMPK) activation cancels selective pathways to pMF, and possible mechanisms are suggested based on literature, such as 1) directly or indirectly (i.e., via tuberous sclerosis complex activation) AMPK-induced mammalian target of rapamycin complex inhibition (mTORC), and 2) phosphodiesterase (PDE)-induced cAMP downregulation. BDNF, brain-derived neurotrophic factor; TrkB, tyrosine receptor kinase B; EPAC, exchange protein activated by cAMP, mTOR, or mammalian target of rapamycin complex.

Table 4.

Physiological variables at baseline and 30, 60, and 90 min after last it injection

Time, min	Met (5.0 mM) + ACSF	Met (5.0 mM) + 5-HT7	Met (5.0 mM)/cC + 5-HT7
$\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$
Baseline	43.3 ± 3.5	47.2 ± 6.3	45.7 ± 0.3
30	42.1 ± 4.4	47.9 ± 6.8	46.2 ± 1.2
60	43.6 ± 2.8	47.6 ± 5.3	47.1 ± 0.4
90	43.4 ± 3.2	47.7 ± 6.3	46.1 ± 1.3
$\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$
Baseline	305 ± 13	312 ± 49	303 ± 42
30	281 ± 24	294 ± 60	245 ± 76
60	270 ± 22	278 ± 81	227 ± 49
90	266 ± 25	276 ± 72	223 ± 55
MAP, mmHg
Baseline	130 ± 9	123 ± 14	109 ± 33
30	132 ± 11	111 ± 31	114 ± 26
60	131 ± 10	113 ± 35	113 ± 28
90	128 ± 6	109 ± 42	106 ± 22
pH
Baseline	7.41 ± 0.05	7.37 ± 0.05	7.39 ± 0.01
30	7.41 ± 0.06	7.37 ± 0.04	7.36 ± 0.04
60	7.40 ± 0.04	7.36 ± 0.06	7.37 ± 0.02
90	7.39 ± 0.03	7.35 ± 0.05	7.38 ± 0.00
sBE
Baseline	2.4 ± 2.5	2.0 ± 1.1	2.7 ± 0.8
30	1.7 ± 1.7	2.1 ± 1.1	1.0 ± 3.2
60	1.9 ± 3.0	1.0 ± 1.9	1.6 ± 1.6
90	1.1 ± 2.5	0.6 ± 1.5	2.0 ± 0.6
Temperature, °C
Baseline	37.9 ± 0.7	37.3 ± 0.6	37.7 ± 0.8
30	37.9 ± 0.4	37.4 ± 0.5	37.9 ± 0.5
60	38.0 ± 0.5	37.3 ± 0.5	37.3 ± 0.4
90	37.7 ± 0.2	37.2 ± 0.6	37.4 ± 0.3

Met, metformin; cC, compound C; ACSF, artificial cerebrospinal fluid; 5-HT₇, serotonin 7 receptor; MAP, mean arterial pressure; sBE, standard base excess.

DISCUSSION

The major findings of the present study are that pharmacologically induced spinal AMPK activation via 2-DG or metformin abolishes 5-HT₇- but not 5-HT_2A-dependent pMF. The ability of cC to restore 5-HT₇ receptor-induced pMF increases confidence that 2-DG and metformin exerted their effects specifically via AMPK activation vs. unanticipated off-target effects. Thus, spinal AMPK plays an important role in selectively regulating the expression of phrenic motor plasticity. Because AMPK activity varies across the circadian cycle (42), it is possible that AMPK differentially regulates the S pathway to pMF throughout the day. Furthermore, because AMPK is activated by metabolic stress, it likely serves to shut down energetically costly mTOR-dependent protein synthesis required for S pathway expression in conditions such as severe hypoxia that challenge energy supply.

2-DG and metformin are transported into the cells through glucose transporters and organic cation proteins, respectively (33, 38). 2-DG inhibits hexokinase, the first and rate-limiting enzyme in glycolysis (24, 25, 40). In contrast, metformin inhibits complex I of the electron transport chain (34). Both drugs increase the AMP-to-ATP ratio, increasing AMP binding to the AMPK-γ regulatory subunits, leading to persistent AMPK phosphorylation and activation (38, 46). Activated AMPK stimulates ATP production and inhibits nonessential ATP-consuming functions such as protein translation via mTORC1 (39, 41). Synthesis of an immature TrkB isoform is necessary for the S pathway to pMF (14), and this synthesis requires mTORC1 activity (11, 13). Other forms of synaptic plasticity are inhibited by AMPK activity, including hippocampal long-term potentiation (37).

AMPK signaling inhibits mTORC1 activity both directly (7) and indirectly via tuberous sclerosis complex phosphorylation and activation (4, 5, 20, 43). Although mTORC1 signaling is not necessary for moderate AIH-induced Q pathway-dependent pMF, it is required for severe AIH-induced S pathway-dependent pMF (11) and 5-HT₇-induced pMF (13). Thus, AMPK-dependent tuberous sclerosis complex 1/2 activation and mTORC1 inhibition are the most likely mechanism constraining 5-HT₇-induced pMF although this hypothesis has not yet been tested directly. Alternatively, AMPK activates phosphodiesterases, thereby reducing cAMP concentration and downstream signaling in some cell types (21, 45). Although this is another viable explanation for the effects observed here, it is less likely since coactivation of the Q and S pathways to pMF cancels pMF expression due to upstream cross-talk inhibition (35, 36); considering that cAMP activation of protein kinase A is necessary for S to Q pathway inhibition (18, 35, 36), is it unlikely that the effects observed here resulted from phosphodiesterase activation and reduced levels of cAMP. Because neither 2-DG nor metformin affected 5-HT_2A receptor-induced pMF, downstream AMPK effects on tuberous sclerosis complex 1/2 activation are a more likely explanation of effects observed in this study, since it would enable complete differentiation of AMPK effects on the S vs. Q pathways to pMF.

Recent studies from our laboratory demonstrate that moderate AIH-induced 5-HT₂-dependent phrenic long-term facilitation is unaffected by systemic 2-DG (27). Conversely, 2-DG constrains respiratory metaplasticity (enhanced phrenic long-term facilitation) after 4 wk of repetitive AIH preconditioning (27). Although mechanisms underlying respiratory metaplasticity are largely unknown, a loss of cross-talk inhibition between the Q and S pathways to pMF has been postulated, leading to additive contributions from both the Q and S pathways (27). Thus, AMPK activity would eliminate the S but not Q pathway contribution to enhanced pMF.

Although AMPK plays a key role in the neural control of breathing (6, 23, 29), the present study is the first to demonstrate that spinal AMPK differentially regulates distinct mechanisms of phrenic motor plasticity. The hypoxic ventilatory response is reduced by conditional AMPK deletion in catecholaminergic cells (29). On the other hand, carotid chemoafferent neuron activity is unaffected by AMPK deletion (6, 23, 29). Thus, the relevant AMPK modulating the hypoxic ventilatory response must reside within second- and/or higher-order neurons of the carotid chemoreflex (29).

The effects of intrathecal drug injections are not restricted to phrenic motor neurons. Although accumulating evidence indicates that key molecules necessary to both 5-HT_2A- and 5-HT₇-induced pMF are expressed within phrenic motor neurons, the involvement of spinal interneurons and/or glial cells in mediating AMPK-dependent constraint to 5-HT₇-induced pMF cannot be excluded. Both approaches used in this study to activate AMPK consistently blocked 5-HT₇-induced pMF. However, cC was more effective in restoring 5-HT₇-induced pMF following 2-DG vs. metformin. Metformin may lead to prolonged AMPK activation, overcoming cC effects. Alternatively, because 2.5 mM metformin (but not 5 mM) was partially restored by cC, even lower doses might be required.

In conclusion, AMPK activity constrains 5-HT₇ receptor-induced (and presumably adenosine 2A receptor-induced) pMF without detectable effects on the Q pathway elicited by 5-HT_2A receptor activation (Fig. 6). These findings advance our understanding of key molecules regulating the expression of the diverse signaling pathways that lead to pMF (8). It is essential to understand this and other factors regulating motor neuron plasticity as we move along a translational path, attempting to harness intermittent hypoxia to restore respiratory and nonrespiratory movements after chronic spinal injury, or with other clinical disorders that compromise movement (9, 15, 31). For example, diabetes is commonplace in those suffering from spinal cord injury; thus, many people with spinal injury are medicated with metformin (22, 30). In these individuals, the potential benefits of prolonged repetitive acute intermittent hypoxia may be undermined (see Ref. 27). However, future studies are needed to test the hypothesis that AMPK activation undermines phrenic motor plasticity in pathophysiological conditions compromising breathing, such as spinal cord injury or amyotrophic lateral sclerosis.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute (NHLBI) Grant HL-69064 and the McKnight Brain Institute. D. P. Fields was supported by fellowships from the United Negro College Fund and the NHLBI (F30-HL-126351).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.R.P. and D.P.F. performed experiments; R.R.P. analyzed data; R.R.P. and G.S.M. interpreted results of experiments; R.R.P. prepared figures; R.R.P. drafted manuscript; R.R.P., D.P.F., and G.S.M. approved final version of manuscript; D.P.F. and G.S.M. conceived and designed research; D.P.F. and G.S.M. edited and revised manuscript.