Intermittent apnea elicits inactivity-induced phrenic motor facilitation via a retinoic acid- and protein synthesis-dependent pathway

We identify spinal retinoic acid and protein synthesis as critical components in the cellular cascade whereby repetitive reductions in respiratory neural activity elicit rebound increases in phrenic inspiratory activity.

Abstract

Respiratory motoneuron pools must provide rhythmic inspiratory drive that is robust and reliable, yet dynamic enough to respond to respiratory challenges. One form of plasticity that is hypothesized to contribute to motor output stability by sensing and responding to inadequate respiratory neural activity is inactivity-induced phrenic motor facilitation (iPMF), an increase in inspiratory output triggered by a reduction in phrenic synaptic inputs. Evidence suggests that mechanisms giving rise to iPMF differ depending on the pattern of reduced respiratory neural activity (i.e., neural apnea). A prolonged neural apnea elicits iPMF via a spinal TNF-α-induced increase in atypical PKC activity, but little is known regarding mechanisms that elicit iPMF following intermittent neural apnea. We tested the hypothesis that iPMF triggered by intermittent neural apnea requires retinoic acid and protein synthesis. Phrenic nerve activity was recorded in urethane-anesthetized and -ventilated rats treated intrathecally with an inhibitor of retinoic acid synthesis (4-diethlyaminobenzaldehyde, DEAB), a protein synthesis inhibitor (emetine), or vehicle (artificial cerebrospinal fluid) before intermittent (5 episodes, ~1.25 min each) or prolonged (30 min) neural apnea. Both DEAB and emetine abolished iPMF elicited by intermittent neural apnea but had no effect on iPMF elicited by a prolonged neural apnea. Thus different patterns of reduced respiratory neural activity elicit phenotypically similar iPMF via distinct spinal mechanisms. Understanding mechanisms that allow respiratory motoneurons to dynamically tune their output may have important implications in the context of respiratory control disorders that involve varied patterns of reduced respiratory neural activity, such as central sleep apnea and spinal cord injury.

NEW & NOTEWORTHY We identify spinal retinoic acid and protein synthesis as critical components in the cellular cascade whereby repetitive reductions in respiratory neural activity elicit rebound increases in phrenic inspiratory activity.

breathing is generated by integrated circuits in the brain stem that ultimately provide rhythmic drive to muscles via respiratory motoneuron pools. This drive must be robust and reliable to ensure sufficient gas exchange, dynamic enough to respond to metabolic and respiratory challenges, and adaptable to maintain stability and dynamic responsiveness throughout life. Mechanisms whereby the respiratory control system “tunes” respiratory motor output to meet the demands of an ever-changing organism are not well understood. An emerging principle of neuroscience is that neural activity is sensed and adjusted locally to ensure that neurons continue to operate within an optimal range (Turrigiano 2008). In a series of previous studies, we demonstrated that respiratory motoneuron pools actively monitor inspiratory-related inputs and recruit mechanisms of plasticity that increase inspiratory motor output when synaptic inputs are abnormally reduced, even in the absence of altered blood gases (Baertsch and Baker-Herman 2013, 2015; Baker-Herman and Strey 2011; Broytman et al. 2013; Mahamed et al. 2011; Streeter and Baker-Herman 2014; Strey et al. 2012, 2013). Plasticity induced by reduced respiratory neural activity has been suggested to be a compensatory form of plasticity because the magnitude of enhancement is proportional to the magnitude of activity deprivation (Braegelmann et al. 2017).

Although a variety of inspiratory-related motoneurons exhibit plasticity in response to reduced respiratory neural activity (Baker-Herman and Strey 2011; Mahamed et al. 2011; Strey et al. 2012, 2013), most investigations have focused on mechanisms that sense and respond to reductions in phrenic neural activity, which induces a form of plasticity called inactivity-induced phrenic motor facilitation (iPMF; Baertsch and Baker-Herman 2013, 2015; Baker-Herman and Strey 2011; Broytman et al. 2013; Mahamed et al. 2011; Strey et al. 2012; 2013). Mechanisms local to the phrenic motor pool are necessary for iPMF because spinal application of inhibitors to the iPMF pathway following brain stem-mediated reductions in respiratory neural activity abolish iPMF (Baertsch and Baker-Herman 2015); moreover, reducing activity of a single phrenic motor pool while leaving brain stem and contralateral respiratory neural activity intact elicits iPMF on the ipsilateral, but not contralateral, side (Streeter and Baker-Herman 2014).

Similar to other forms of spinal plasticity (Baker and Mitchell 2000; Baumbauer et al. 2008, 2009), iPMF is pattern sensitive, because it is more efficiently induced by intermittent reductions in respiratory neural activity than a single episode of an equivalent cumulative duration (Baertsch and Baker-Herman 2013, 2015). However, phenotypically similar iPMF can be revealed by a single sustained episode of reduced respiratory neural activity if it is prolonged (~30 min; Baertsch and Baker-Herman 2013; Mahamed et al. 2011). Surprisingly, different induction mechanisms underlie iPMF depending on the pattern of reduced respiratory neural activity, although both converge on a similar downstream pathway. Evidence suggests that prolonged reductions in phrenic synaptic inputs results in the local release of the proinflammatory cytokine TNF-α, which activates atypical PKC (aPKC), in phrenic motoneurons to stabilize a transient increase in phrenic burst amplitude following activity deprivation into long-lasting iPMF (Broytman et al. 2013; Strey et al. 2012; Streeter and Baker-Herman 2014). Whereas iPMF following intermittent reductions in respiratory neural activity ultimately requires spinal atypical PKC activity, it does not require spinal TNF-α (Baertsch and Baker-Herman 2015). Mechanisms that induce iPMF following intermittent episodes of reduced respiratory neural activity are unknown.

The vitamin A metabolite retinoic acid (RA) has many of the requisite characteristics to play a prominent role in iPMF. Although known more typically for its role in neuronal development and differentiation, RA has been shown to underlie rapid, homeostatic increases in synaptic transmission in response to activity deprivation (Aoto et al. 2008; Wang et al. 2011). In cultured hippocampal neurons, RA is not detectable when neurons are active; however, RA synthesis is strongly induced by a loss of synaptic activity (Aoto et al. 2008) or a decrease in dendritic calcium levels (Wang et al. 2011). Evidence suggests that decreases in intracellular calcium during activity deprivation triggers RA synthesis by releasing calcium-constrained RA synthesis enzymes; RA then activates dendritic RA receptors to increase synthesis and insertion of postsynaptic AMPA receptors, thereby increasing synaptic strength. Interestingly, RA is not necessary for all forms of homeostatic plasticity; Chen et al. (2014) analyzed the role for RA in several different forms of homeostatic plasticity and found that RA was only necessary for rapid, protein synthesis-dependent forms, but not for those that take longer to develop.

In the present study, we tested the hypothesis that iPMF following intermittent neural apnea requires spinal retinoic acid and protein synthesis. Using electrophysiological and pharmacological approaches, we demonstrate that iPMF elicited by intermittent reductions in respiratory neural activity requires both spinal retinoic acid and protein synthesis, whereas TNF-α-dependent iPMF elicited by a prolonged reduction in respiratory neural activity does not. Thus our results indicate that different patterns of reduced respiratory neural activity initiate iPMF through distinct spinal mechanisms. Understanding the diverse cellular processes that initiate compensatory plasticity in response to abnormal changes in the activity of motoneuron pools critical for breathing may provide a framework for development of therapeutic strategies for treatment of devastating ventilatory control disorders associated with reduced respiratory neural activity such as sleep apnea and spinal cord injury (Strey et al. 2013).

METHODS

Ethical approval.

All experimental protocols were approved by the Animal Care and Use Committee at the University of Wisconsin, Madison and conformed to the policies of the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Experiments were performed on 2.5- to 3.5-mo-old male Sprague-Dawley rats (n = 61) from Harlan Laboratories (colony 217). Rats were housed 2 per cage with 12:12-h light-dark cycles and food and water ad libitum.

Drugs and materials.

4-Diethlyaminobenzaldehyde (DEAB), an inhibitor of retinoic acid synthesis, and emetine dihydrochloride hydrate, which irreversibly blocks protein synthesis by binding the 40S ribosomal subunit, were purchased from Sigma-Aldrich (St. Louis, MO). Stock solutions of DEAB (100 mM) and emetine (7 mM) were made in 100% DMSO and artificial cerebrospinal fluid (aCSF; containing in mM: 120 NaCl, 3 KCl, 2 CaCl, 2 MgCl, 23 NaHCO₃, and 10 glucose bubbled with 95% O₂-5% CO₂), respectively. Stock solutions were aliquoted and stored at −20°C. Immediately before experimental protocols began, stock solutions of DEAB were diluted to 100 μM in aCSF, a concentration that effectively impairs retinoic acid-dependent increases in excitatory neurotransmission in the hippocampus (Aoto et al. 2008). Stock solutions of emetine were diluted with aCSF to 70 µM, a concentration known to effectively inhibit spinal plasticity when delivered intrathecally (Baker-Herman and Mitchell 2002).

Surgical preparation.

Electrophysiological experiments have been extensively described previously (Baertsch and Baker-Herman 2013, 2015; Broytman et al. 2013; Strey et al. 2012). Rats were induced with isoflurane (2.5–3.5%), tracheostomized, and pump ventilated (~70 breaths/min, 2.5–3 ml; 50% O₂, N₂ balance; model 683 small animal ventilator; Harvard Apparatus, Holliston, MA). A flow-through capnograph (Capnogard; Respironics) was used to measure end-tidal CO₂ (ETco₂) from the expired line of the ventilator circuit as an index of arterial partial pressure of CO₂ ($\text{P}{\text{a}}_{\text{CO}}{}_{{}_2}$). The vagus nerve was isolated using a ventral approach and cut bilaterally at the cervical level to prevent entrainment of respiratory frequency to the ventilator. Tracheal pressure was monitored, and CO₂ was added to the inspired gas mixture as needed to ensure rats continued respiratory efforts throughout the surgery. A tail vein catheter was placed for delivery of fluids (lactated Ringer solution; ~20% sodium bicarbonate, as necessary). The femoral artery was catheterized to monitor arterial blood pressure and draw blood samples for pH and blood gas analysis (ABL800; Radiometer, Copenhagen, Denmark). Rats were then converted to urethane anesthesia (1 ml/100 g of 0.175 g/ml urethane infused at 6 ml/h iv) while isoflurane was gradually withdrawn. Depth of anesthesia was monitored by testing pressor responses to paw pad pinch. A laminectomy was performed at cervical vertebra 2 (C2), a small hole was cut in the dura, and a soft silicon catheter (2-French; Access Technologies) connected to a 50-µl Hamilton syringe was inserted into the intrathecal space and advanced caudally (~5 mm) to rest above spinal segment C4. With the use of a dorsal approach, the left phrenic nerve was isolated, cut distally, desheathed, submerged in mineral oil, and placed on a bipolar silver electrode for recording. Body temperature was monitored with a rectal thermometer (model 700 1H; Physitemp) and maintained near 37.0°C with a custom-heated surgery table. Following surgery, rats were paralyzed with pancuronium bromide (1 mg/kg iv). At least 1 h was allowed following termination of isoflurane anesthesia to ensure adequate washout before experimental protocols were carried out.

Experimental protocols.

Rats were separated into seven groups. Two experimental groups were treated with intrathecal DEAB 20–30 min before (10 µl) and immediately before (5 µl) intermittent neural apnea (n = 8; described below) or prolonged neural apnea (n = 8). Two experimental groups were treated with intrathecal emetine (10 µl) 20–30 min before intermittent neural apnea (n = 8) or prolonged neural apnea (n = 10). Control groups were treated with intrathecal vehicle (aCSF) before intermittent (n = 8) or prolonged (n = 10) neural apnea. Additional control groups did not receive neural apnea but were treated with intrathecal DEAB (n = 4) or emetine (n = 5) at corresponding time points to control for any drug- and/or time-dependent changes in phrenic motor output; there were no statistically significant differences between these groups, so they were combined and referred to collectively as “time controls.”

“Baseline” was established when phrenic nerve burst amplitude and frequency remained stable under isocapnic conditions for at least 15–20 min. Rats were slightly hyperventilated, so a small amount of CO₂ was added to the inspired gas line to maintain ETco₂ at ∼45 mmHg. An arterial blood sample was drawn to obtain baseline $\text{P}{\text{a}}_{\text{CO}}{}_{{}_2}$, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, and pH measurements (temperature corrected). Neural apnea was induced by reducing inspired CO₂ until all rhythmic phrenic activity ceased (i.e., CO₂ apneic threshold). In rats receiving intermittent neural apnea, inspired CO₂ was then immediately increased until respiratory neural activity resumed (i.e., CO₂ recruitment threshold) and baseline ETco₂ was restored. This process was then repeated to induce five ~1.25-min neural apneas, separated by 5-min recovery periods. In rats receiving prolonged neural apnea, ETco₂ was maintained ~2 mmHg below the CO₂ apneic threshold for 30 min, at which point neural apnea was reversed as described above. Importantly, in all cases $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ was maintained >200 mmHg with the ventilator. Arterial blood samples were drawn 5, 15, 30, and 60 min following neural apnea to confirm $\text{P}{\text{a}}_{\text{CO}}{}_{{}_2}$ was within 1.5 mmHg of baseline and to allow maintenance of $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ and pH.

Data analysis.

Phrenic burst activity was amplified (10,000 times), bandpass filtered (0.3–10 kHz; AM Systems), integrated (time constant: 50 ms), and rectified. The signal was digitized and analyzed with PowerLab (AD Instruments; Laboratory Chart 7.0 software). Sixty-breath bins were taken at baseline and 5, 15, 30, and 60 min following either prolonged or intermittent neural apnea or at equivalent time points in time controls and analyzed for average peak phrenic amplitude and frequency. Because the raw voltage recorded depends on the placement of the recording electrode and the quality of the nerve dissection, phrenic nerve burst amplitude is expressed as a percent change from baseline (%baseline). By contrast, burst frequency is not influenced by recording artifact but reflects the physiology of the animal; therefore, burst frequency is expressed as an absolute change from baseline (Δbaseline). Data are means ± SE. Statistical differences between groups and individual time points were determined using a two-way repeated-measures ANOVA design and Tukey’s post hoc tests, respectively (Prism 6; GraphPad Software). A significance level of P < 0.05 was used for all comparisons.

RESULTS

Regulation of physiological variables.

Table 1 lists average $\text{P}{\text{a}}_{\text{CO}}{}_{{}_2}$, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, pH, and mean arterial pressure (MAP) at baseline and 60 min after treatments. In all groups, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ was maintained above 200 mmHg throughout the protocol (including during neural apnea), and $\text{P}{\text{a}}_{\text{CO}}{}_{{}_2}$ post-neural apnea was maintained within ±1.5 mmHg of baseline. Indeed, there were no time-dependent changes in $\text{P}{\text{a}}_{\text{CO}}{}_{{}_2}$ or $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ within any group; thus changes in phrenic nerve activity during our experimental protocols cannot be attributed to changes in chemoreceptor feedback. In some groups there were small but significant time-dependent decreases in pH and/or MAP, which is consistent with this anesthetized rat preparation (Baertsch and Baker-Herman 2015; Dale-Nagle et al. 2011; Streeter and Baker-Herman 2014). These changes were within normal physiological limits and not correlated with changes in phrenic amplitude (P > 0.05); thus they are considered unlikely to have contributed to our results.

Table 1.

Physiological variables during electrophysiological protocols

Treatment	Time	$\text{P}{\text{a}}_{\text{CO}}{}_{{}_2}$, mmHg	$\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, mmHg	pH	MAP, mmHg
Intermittent neural apnea
Vehicle	Baseline	43.4 ± 1.0	280.8 ± 5.6	7.39 ± 0.00	129.7 ± 4.0
	60 min	43.5 ± 1.0	251.9 ± 8.0*	7.36 ± 0.00*	126.4 ± 5.3
DEAB	Baseline	43.9 ± 0.9	261.6 ± 6.9	7.37 ± 0.01	126.4 ± 7.8
	60 min	43.9 ± 1.0	247.8 ± 6.0	7.38 ± 0.01	111.4 ± 8.5*
Emetine	Baseline	43.9 ± 0.9	274.2 ± 6.3	7.38 ± 0.01	127.0 ± 7.6
	60 min	43.9 ± 0.9	259.9 ± 6.4	7.37 ± 0.01	111.4 ± 5.9
Prolonged neural apnea
Vehicle	Baseline	43.9 ± 1.1	284.1 ± 6.0	7.38 ± 0.01	141.0 ± 2.5
	60 min	44.8 ± 1.5	264.7 ± 6.7	7.35 ± 0.01*	134.1 ± 6.8
DEAB	Baseline	46.0 ± 1.3	263.4 ± 10.6	7.35 ± 0.01	139.0 ± 3.7
	60 min	44.3 ± 1.0	260.5 ± 6.0	7.33 ± 0.02	125.0 ± 4.7
Emetine	Baseline	45.3 ± 0.6	282.8 ± 3.4	7.37 ± 0.01	133.2 ± 6.6
	60 min	45.1 ± 0.6	263.2 ± 7.9	7.35 ± 0.01*†	119.0 ± 7.0*
Time control	Baseline	44.6 ± 0.9	272.6 ± 5.7	7.36 ± 0.01	140.7 ± 7.2
	60 min	44.3 ± 1.0	256.1 ± 9.8	7.35 ± 0.02	116.6 ± 9.1*

Values are means ± SE of $\text{P}{\text{a}}_{\text{CO}}{}_{{}_2}$, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$, pH, and MAP at baseline and 60 min following treatment in all rat groups.

^*P < 0.05, significantly different from baseline.

^†P < 0.05, significantly different from prolonged neural apnea + DEAB.

There were no other significant differences between groups.

Intermittent neural apnea elicits retinoic acid and protein synthesis-dependent iPMF.

To test the hypothesis that iPMF following intermittent neural apnea requires retinoic acid synthesis, rats received intrathecal DEAB or vehicle before intermittent neural apnea (Fig. 1, A–C). Representative compressed phrenic neurograms illustrating phrenic burst amplitude at baseline, during, and for the 60 min following intermittent neural apnea, and for a similar duration in a time control rat, are shown in Fig. 1A. Average changes in phrenic burst amplitude and respiratory frequency relative to baseline are shown in Fig. 1, B and C, respectively. In time control rats (no neural apnea), there were no significant changes in phrenic burst amplitude (%baseline; 5 min: 6 ± 2, 15 min: 12 ± 3, 30 min: 10 ± 6, 60 min: 7 ± 7) or frequency (Δbaseline; 5 min: 0 ± 1, 15 min: 1 ± 1, 30 min: 2 ± 1, 60 min: 2 ± 1) from baseline (P > 0.05), indicating that intrathecal DEAB or emetine had no effect on baseline phrenic activity and that our preparations were stable. As expected, in rats receiving intrathecal vehicle, phrenic burst amplitude increased in between successive intermittent neural apnea episodes and remained significantly increased at all time points post-neural apnea (%baseline; 5 min: 50 ± 4, 15 min: 54 ± 3, 30 min: 57 ± 5, 60 min: 56 ± 6) compared with that in time controls (P < 0.0001), indicating iPMF. By contrast, phrenic burst amplitude in rats receiving DEAB before intermittent neural apnea was not significantly different from that in time controls at any time point post-neural apnea (%baseline; 5 min: 11 ± 4, 15 min: 17 ± 7, 30 min: 21 ± 7, 60 min: 22 ± 10; P > 0.05) and was significantly lower than in rats receiving vehicle before intermittent neural apnea (P < 0.0001), indicating that inhibition of retinoic acid synthesis abolishes iPMF.

Fig. 1.

Intermittent neural apnea elicits spinal retinoic acid and protein synthesis-dependent iPMF. A: representative compressed phrenic neurograms depicting integrated phrenic burst amplitude (∫Ph) before (baseline), during, and for 60 min following five ~1.25-min neural apneas separated by ~5 min in rats pretreated with aCSF, 100 µM DEAB, or 70 μM emetine, and an equivalent duration in time control rats pretreated with DEAB or emetine. B and C: mean changes in phrenic burst amplitude and frequency 5, 15, 30, and 60 min following intermittent neural apnea in rats receiving aCSF (black circles), DEAB (blue squares), or emetine (green diamonds) and after no neural apnea (time control; gray triangles), indicating treatment with intrathecal DEAB or emetine blocks the expression of iPMF following intermittent neural apnea. *P < 0.05, significantly different from time control. #P < 0.05, significantly different from emetine + intermittent neural apnea.

Because retinoic acid induces plasticity in response to activity deprivation in the hippocampus via an increase in protein synthesis (Aoto et al. 2008; Chen et al. 2014; Groth and Tsien 2008; Poon and Chen 2008; Sarti et al. 2013; Wang et al. 2011), we next sought to determine if iPMF following intermittent neural apnea requires spinal protein synthesis (Fig. 1, A–C). In rats receiving intrathecal emetine before intermittent neural apnea, phrenic burst amplitude was not significantly different from that in time controls at any time point post-neural apnea (%baseline; 5 min: 6 ± 9, 15 min: 17 ± 5, 30 min: 15 ± 5, 60 min: 10 ± 9; P > 0.05) and was significantly lower than in rats pretreated with vehicle (P < 0.0001), indicating that protein synthesis inhibition abolishes iPMF elicited by intermittent neural apnea.

Intermittent neural apnea elicited a mild transient increase in respiratory frequency (Fig. 1C) that was similar in magnitude between rats treated with vehicle (Δbaseline; 5 min: 6 ± 2, 15 min: 4 ± 1, 30 min: 4 ± 1, 60 min: 3 ± 1), DEAB (Δbaseline; 5 min: 2 ± 1, 15 min: 1 ± 1, 30 min: 2 ± 1, 60 min: 0 ± 2), and emetine (Δbaseline; 5 min: 5 ± 1, 15 min: 3 ± 1, 30 min: 3 ± 1, 60 min: 1 ± 1; P > 0.05); however, frequency was only significantly increased compared with time controls at 5 min post-neural apnea in vehicle-treated rats (P < 0.05). Thus intermittent neural apnea did not elicit long-lasting changes in phrenic burst frequency. Collectively, these data indicate that spinal retinoic acid and new protein synthesis are required for long-lasting increases in phrenic burst amplitude (i.e., iPMF) triggered by intermittent reductions in respiratory neural activity.

iPMF following prolonged neural apnea does not require retinoic acid or protein synthesis.

Because prolonged neural apnea elicits iPMF via spinal TNF-α signaling (Baertsch and Baker-Herman 2015; Broytman et al. 2013), we hypothesized that iPMF elicited by prolonged neural apnea would not require spinal retinoic acid and protein synthesis. To test this hypothesis, rats received intrathecal DEAB, emetine, or vehicle before one 30-min bout of reduced respiratory neural activity (Fig. 2, A–C). Representative compressed phrenic neurograms illustrating phrenic burst amplitude at baseline, during, and for 60 min following prolonged neural apnea, and for a similar duration in a time control, are shown in Fig. 2A. Average changes in phrenic burst amplitude and respiratory frequency relative to baseline are shown in Fig. 2, B and C, respectively. As expected, in rats receiving intrathecal vehicle before prolonged neural apnea, phrenic burst amplitude was significantly increased compared with that in time controls at all time points following resumption of respiratory neural activity (%baseline; 5 min: 60 ± 11, 15 min: 57 ± 11, 30 min: 53 ± 14, 60 min: 54 ± 15; P < 0.05), indicating iPMF. In contrast to what we observed following intermittent neural apnea, rats receiving intrathecal DEAB (%baseline; 5 min: 81 ± 15, 15 min: 79 ± 15, 30 min: 79 ± 17, 60 min: 87 ± 18; P < 0.05) or emetine (%baseline; 5 min: 55 ± 10, 15 min: 59 ± 11, 30 min: 56 ± 11, 60 min: 56 ± 12) continued to express significant increases in phrenic burst amplitude relative to time controls at all time points following prolonged neural apnea (P < 0.01). Furthermore, there were no differences in phrenic burst amplitude between rats treated with DEAB, emetine, or aCSF (P > 0.05), suggesting RA synthesis inhibition or protein synthesis inhibition had no effect on iPMF following prolonged neural apnea. However, we observed a trend for DEAB treated rats to express a larger iPMF following prolonged neural apnea than vehicle controls, suggesting the possibility that RA synthesis may constrain iPMF following prolonged neural apnea. Collectively, these data indicate that iPMF elicited by a prolonged neural apnea does not require spinal retinoic acid or protein synthesis.

Fig. 2.

Prolonged neural apnea elicits spinal retinoic acid and protein synthesis-independent iPMF. A: representative compressed phrenic neurograms depicting integrated phrenic burst amplitude (∫Ph) before (baseline), during, and for 60 min following prolonged neural apnea in rats pretreated with aCSF, 100 µM DEAB or 70 μM emetine, and an equivalent duration in time control rats pretreated with DEAB or emetine. B and C. Mean changes in phrenic burst amplitude and frequency 5, 15, 30, and 60 min following prolonged neural apnea in rats receiving aCSF (black circles), DEAB (blue squares), or emetine (green diamonds), and after no neural apnea (time control; gray triangles), indicating treatment with intrathecal DEAB or emetine has no effect on the expression of iPMF following prolonged neural apnea. Average data for time controls are the same as in Fig. 1. *P < 0.05, significantly different from time control.

Similarly to intermittent neural apnea, prolonged neural apnea elicited a mild, transient increase in respiratory frequency (Fig. 2C) that was similar in magnitude in rats treated with vehicle (Δbaseline; 5 min: 11 ± 2, 15 min: 8 ± 1, 30 min: 6 ± 1, 60 min: 5 ± 1), DEAB (Δbaseline; 5 min: 8 ± 2, 15 min: 5 ± 1, 30 min: 4 ± 1, 60 min: 4 ± 1), or emetine (Δbaseline; 5 min: 10 ± 1, 15 min: 7 ± 1, 30 min: 5 ± 1, 60 min: 6 ± 1). In all groups receiving prolonged neural apnea, phrenic burst frequency was significantly different from that in time controls at 5 and 15 min (P < 0.05), but not by 30 min (P > 0.05). Thus changes in phrenic burst frequency elicited by prolonged neural apnea are transient and do not require spinal retinoic acid or protein synthesis.

DISCUSSION

Our results reveal that spinal mechanisms initiating inactivity-induced phrenic motor facilitation (iPMF) vary depending on the pattern of inactivity experienced by the phrenic motor pool. Specifically, intermittent reductions in respiratory neural activity elicit iPMF via a retinoic acid- and protein synthesis-dependent pathway, whereas prolonged reductions in respiratory neural activity elicit iPMF via a TNFα-dependent pathway (Baertsch and Baker-Herman 2015; Broytman et al. 2013). Multiple pathways to achieve a similar functional outcome (e.g., increased phrenic motor output) may confer system flexibility (Dale-Nagle et al. 2010; Devinney et al. 2013). These results further expand our understanding of inactivity-induced respiratory plasticity and may have important implications for conditions associated with pathological fluctuations in respiratory neural activity such as apnea of prematurity, sleep at altitude, central sleep apnea, spinal cord injury, and chronic heart failure (Strey et al. 2013).

iPMF elicited by intermittent vs. prolonged inactivity: distinct mechanisms of spinal plasticity?

Our data indicate that the pattern of reduced respiratory neural activity is a critical determinant of the mechanism that initiates iPMF. Specifically, inhibition of retinoic acid or protein synthesis in spinal regions containing the phrenic motor pool prevents induction of iPMF during intermittent neural apnea but has no effect on iPMF elicited by prolonged neural apnea. We have previously demonstrated a similar differential, but inverse, requirement for TNF-α, because spinal TNF-α is necessary for iPMF elicited by prolonged (Broytman et al. 2013; Streeter and Baker-Herman 2014) but not intermittent (Baertsch and Baker-Herman 2015) neural apnea. However, once initiated, iPMF seems to require spinal aPKC during the first ~45 min of its expression, regardless of the pattern of inactivity that induced iPMF (Baertsch and Baker-Herman 2015; Strey et al. 2012). Thus we hypothesize that intermittent and prolonged neural apnea initiate distinct signaling cascades, which then converge to elicit phenotypically similar aPKC-dependent iPMF (Baertsch and Baker-Herman 2015) (Fig. 3).

Fig. 3.

Working hypothesis of converging cellular pathways to iPMF, showing distinct cellular pathways activated by different patterns of reduced respiratory neural activity. Our working model suggests that intermittent neural apnea elicits transient reductions in phrenic motor neuron intracellular calcium, increased formation of retinoic acid and rapid protein synthesis of new AMPARs, which are inserted in the membrane through interactions with aPKC/SQSTM1. By contrast, our working model suggests that prolonged neural apnea is sensed by glia, leading to TNF-α release and translocation of existing AMPARs into the synapse by PKCζ/SQSTM1. TNFR, tumor necrosis factor receptor.

It is well established in many neural networks that reduced synaptic activity elicits compensatory changes in synaptic transmission to restore a “target” level of neuronal activity, a phenomenon known as homeostatic synaptic plasticity (HSP) (Fernandes and Carvalho 2016; Lee et al. 2014). Although to our knowledge mechanisms of plasticity elicited by spaced episodes of activity deprivation have not been studied, the requirement for protein synthesis varies among models of HSP induced by sustained inactivity. For example, similarly to iPMF elicited by prolonged neural apnea, blockade of action potential firing in hippocampal neurons using TTX increases synaptic strength via a TNF-α-dependent mechanism (Beattie et al. 2002; Stellwagen and Malenka 2006) that is unlikely to involve protein synthesis because trafficking of existing AMPA receptors via activation of phosphatidyl inositol 3-kinase (PI3K) is required (Leonoudakis et al. 2008; Stellwagen et al. 2005). However, if postsynaptic activity is blocked with 2-amino-5-phosphonopentanoic acid (APV)/6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) in addition to TTX, HSP is induced more rapidly via an alternative mechanism requiring retinoic acid synthesis and local translation of AMPA receptors (Aoto et al. 2008; Chen et al. 2014; Maghsoodi et al. 2008; Sarti et al. 2013). Interestingly, both of these mechanisms of HSP ultimately increase synaptic strength through changes in AMPA receptor trafficking and/or insertion, and aPKC is thought to play a key role in this process (Jiang et al. 2009; Ren et al. 2013; Yao et al. 2008). Therefore, we postulate that intermittent and prolonged respiratory neural activity initiate mechanisms that are distinct in their requirement for TNF-α- and retinoic acid-dependent protein synthesis but ultimately converge on an aPKC-mediated AMPA receptor insertion and/or stabilization giving rise to iPMF (Fig. 3).

The initiation of protein synthesis is often dependent on the pattern and time course of the inducing stimuli (Naqib et al. 2012). In general, synthesis of new protein(s) is preferentially recruited following spaced stimuli, resulting in long-lasting plasticity (Kelleher et al. 2004; Naqib et al. 2011; Scharf et al. 2002). For example, during olfactory conditioning in Drosophila, temporally spaced training and massed training produce two forms of long-lasting memory that differ in their duration and their requirement for protein translation (Tully et al. 1994; Yin et al. 1994). Furthermore, in Aplysia, spaced vs. massed application of the same stimulus leads to different forms of memory that are differentially affected by protein synthesis inhibition (Farah et al. 2009). Protein synthesis-dependent plasticity can occur rapidly (Graber et al. 2013; Li et al. 2012; Skup 2008), including mechanisms of activity-dependent plasticity requiring local dendritic protein translation elicited by episodic stimuli (Dziembowska et al. 2012). Indeed, postsynaptic protein synthesis-dependent plasticity in Aplysia motoneurons can occur within 10 min (Villareal et al. 2007), a time frame comparable to the stimulus period used to initiate iPMF. In contrast to plasticity elicited by intermittent stimuli, many forms of plasticity that are elicited by a single stimulus episode do not require protein synthesis (Bailey et al. 2000; Krahe et al. 2005; Muzzio et al. 2001; Tully et al. 1994); however, plasticity elicited in this way typically lasts for a shorter duration (hours vs. days) (Costa-Mattioli and Sonenberg 2008; Scharf et al. 2002). Although iPMF elicited by intermittent or prolonged neural apnea is phenotypically similar, we can only conclude this for the first ~1 h of its expression because of limitations in the longevity of the in vivo preparations used in these studies. Hence, it is not known how long iPMF persists following different patterns of neural apnea. It is possible that iPMF elicited by intermittent neural apnea persists for a significantly longer duration than iPMF elicited by prolonged neural apnea, perhaps as a result of new protein synthesis. Indeed, in some rat substrains, iPMF is only transient (<30 min) following prolonged neural apnea (Streeter and Baker-Herman 2014) but persists >60 min following intermittent neural apnea (Baertsch and Baker-Herman 2015). Future studies will be needed to determine if iPMF elicited by intermittent and prolonged neural apnea is also distinct in duration.

iPMF and frequency facilitation are independent.

Consistent with previous studies (Baertsch and Baker-Herman 2013, 2015; Broytman et al. 2013), neural apnea elicited an increase in respiratory frequency that was transient in duration. Frequency facilitation likely involves brain stem respiratory rhythm-generating neurons (Baker-Herman and Mitchell 2008; Blitz and Ramirez 2002). Therefore, we expect that spinal iPMF and brain stem-derived frequency facilitation operate independently. Indeed, iPMF and frequency facilitation are distinct in both magnitude and time course, because iPMF following both intermittent and prolonged neural apnea is long lasting, whereas frequency facilitation is transient and modest relative to the magnitude of amplitude facilitation (Baertsch and Baker-Herman 2013, 2015). Moreover, the pattern of neural apnea has a profound effect on amplitude, but not frequency, facilitation (Baertsch and Baker-Herman 2013). Instead, the magnitude of frequency facilitation seems to be related to the duration of neural apnea, because prolonged neural apnea results in a larger increase in respiratory frequency than both intermittent neural apnea and a single brief neural apnea of an equivalent cumulative duration (Baertsch and Baker-Herman 2013, 2015). Cellular mechanisms giving rise to frequency facilitation following neural apnea are not yet known but may involve changes in CSF pH that can increase with time during hypocapnia-induced neural apnea (Edelist and Osorio 1969; Krause et al. 2009), rather than reduced respiratory neural activity per se. This idea is supported by previous studies in which frequency facilitation was not observed when isocapnic respiratory neural activity deprivation was used to elicit iPMF (Mahamed et al. 2011).

Conclusion.

We are beginning to appreciate that the respiratory neural network is capable of exquisite control of the most critical respiratory pump muscle, the diaphragm, via a variety of mechanisms that give rise to phrenic motor plasticity (Dale-Nagle et al. 2010). With this study, we build on this understanding by demonstrating that different patterns of reduced respiratory neural activity elicit phenotypically similar iPMF via underlying spinal mechanisms that are distinct in their requirement for retinoic acid and new protein synthesis. We postulate that multiple pathways to stabilize breathing in response to changes in respiratory neural activity may confer an evolutionary advantage during various respiratory challenges and pathologies that can occur over the lifetime of an organism (Strey et al. 2013).

GRANTS

Research support was provided by National Heart, Lung and Blood Institute Grant R01 HL105511.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

N.A.B. and T.L.B. conceived and designed research; N.A.B. performed experiments; N.A.B. analyzed data; N.A.B. and T.L.B. interpreted results of experiments; N.A.B. prepared figures; N.A.B. drafted manuscript; N.A.B. and T.L.B. edited and revised manuscript; N.A.B. and T.L.B. approved final version of manuscript.