Phrenic motoneuron structural plasticity across models of diaphragm muscle paralysis

Abstract

Structural plasticity in motoneurons may be influenced by activation history and motoneuron-muscle fiber interactions. The goal of this study was to examine morphological adaptations of phrenic motoneurons following imposed motoneuron inactivity while controlling for diaphragm muscle inactivity. Well-characterized rat models were used including unilateral C2 spinal hemisection (SH; ipsilateral phrenic motoneurons and diaphragm muscle are inactive) and tetrodotoxin phrenic nerve blockade (TTX; ipsilateral diaphragm muscle is paralyzed while phrenic motoneuron activity is preserved). We hypothesized that inactivity of phrenic motoneurons would result in a decrease in motoneuron size, consistent with a homeostatic increase in excitability. Phrenic motoneurons were retrogradely labeled by ipsilateral diaphragm muscle injection of fluorescent dextrans or cholera toxin subunit B. Following 2 weeks of diaphragm muscle paralysis, morphological parameters of labeled ipsilateral phrenic motoneurons were assessed quantitatively using fluorescence confocal microscopy. Compared to controls, phrenic motoneuron somal volumes and surface areas decreased with SH, but increased with TTX. Total phrenic motoneuron surface area was unchanged by SH, but increased with TTX. Dendritic surface area was estimated from primary dendrite diameter using a power equation obtained from 3D reconstructed phrenic motoneurons. Estimated dendritic surface area was not significantly different between control and SH, but increased with TTX. Similarly, TTX significantly increased total phrenic motoneuron surface area. These results suggest that ipsilateral phrenic motoneuron morphological adaptations are consistent with a normalization of motoneuron excitability following prolonged alterations in motoneuron activity. Phrenic motoneuron structural plasticity is likely more dependent on motoneuron activity (or descending input) than muscle fiber activity.

Graphic abstract

Phrenic motoneuron morphology in rats was assessed quantitatively using confocal microscopy following unilateral diaphragm muscle paralysis induced by C2 spinal hemisection (SH) or tetrodotoxin phrenic nerve blockade (TTX). Ipsilateral motoneuron somal surface area decreased after SH and increased after TTX, indicating that structural plasticity does not depend on muscle inactivity.

INTRODUCTION

Changes in neuronal activity, including inactivity, induce functional and structural plasticity in various systems (Bi and Poo, 2001; Segal and Andersen, 2000; Streeter and Baker-Herman, 2014). Examples of functional plasticity include both short and long term potentiation and depression in the hippocampus in response to altered activity, which have been attributed to changes in synaptic efficacy (Steward et al., 2001; Tomasulo et al., 1993; Tomasulo and Steward, 1996). Structural plasticity including synaptic remodeling is evident in hippocampal preparations after anoxic/hypoglycemic (Jourdain et al., 2002) or glutamate-induced excitotoxic injury (Goldin et al., 2001). In general, structural plasticity occurs over a longer time period and may involve not only synaptic remodeling but also changes in neuron morphology. Compared to the considerable literature on central and sensory neurons, very little is known about activity-dependent structural plasticity in motoneurons, especially as a result of motoneuron inactivity.

Phrenic motoneurons are highly active in sustaining ventilation. For example, the diaphragm muscle has a duty cycle of ~40% in contrast to the duty cycle of motoneurons innervating hindlimb muscles (~1–2% for soleus and ~15% for extensor digitorum longus muscle) (Sieck, 1995; Sieck and Fournier, 1989; Sieck and Prakash, 1997). Thus, phrenic motoneurons appear well adapted for sustained activity. Accordingly, imposition of inactivity (e.g, cervical spinal cord injury or prolonged mechanical ventilation) may particularly affect phrenic motoneurons. There is currently a paucity of information regarding inactivity-induced phrenic motoneuron plasticity, especially regarding structural adaptations. In models of phrenic motoneuron inactivity, functional adaptations include partial recovery of motor function under increased ventilatory drive following incomplete spinal cord injury (Goshgarian et al., 1989; Hernandez-Torres et al., 2017; Mantilla et al., 2013a; Mantilla et al., 2013b; Martinez-Galvez et al., 2016; Nantwi and Goshgarian, 2001; O’Hara and Goshgarian, 1991) as well as long term potentiation following cervical deafferentation (Kinkead et al., 1998) and pharmacologically-imposed motoneuron inactivity (Streeter and Baker-Herman, 2014). Taken together, these data suggest that phrenic motoneuron inactivation results in adaptations that serve to restore phrenic motoneuron function. Based on Henneman’s size principle (Henneman, 1957; Henneman et al., 1965), intrinsic excitability of neurons is dependent on membrane surface area (capacitance, resistance and rheobase) such that retraction of neuron size (somal and dendritic) leads to increased excitability. In the present study, we hypothesized that inactivity of phrenic motoneurons, induced by a C2 cervical spinal hemisection (SH), would result in a decrease in motoneuron size and since muscle paralysis also results from SH, to control for muscle inactivity per se, we performed additional experiments with tetrodotoxin (TTX) blockade of the phrenic nerve. TTX blockade causes unilateral diaphragm muscle paralysis without affecting the activity of phrenic motoneurons or interrupting axonal transport between the nerve and muscle.

METHODS

Adult male Sprague-Dawley rats (initial body weight ~300 g) were assigned randomly to the following groups: spinal hemisection (SH; n=5), sham spinal hemisection (SHAM-SH; n=3), tetrodotoxin-blockade of the phrenic nerve (TTX; n=5), sham TTX block (SHAM-TTX; n=2) and untreated control (n=4). Female rats were not included given expected differences in body size and possibly in dendritic arborization in this age group (Keil et al., 2017). All experimental procedures were approved by the Institutional Animal Care and Use Committee at Mayo Clinic, and in accordance with the American Physiological Society Animal Care Guidelines. For surgical procedures, animals were anesthetized by intramuscular injection of ketamine (60–90 mg/kg body weight) and xylazine (2.5–10 mg/kg), and administered antibiotics (Penicillin G) and analgesics (acetaminophen) for the first 5 days.

Spinal Hemisection.

Details of the SH surgical procedure have been previously published (Gransee et al., 2017; Gransee et al., 2013; 2015; Mantilla et al., 2007; Miyata et al., 1995; Prakash et al., 1999). Under sterile conditions, a bilateral dorsal C2 laminectomy was performed and the right half of the spinal cord was sectioned lateral to the dorsal fissure, thus transecting only the ventral and lateral funiculi, while preserving the dorsal funiculus on the side of SH. Ipsilateral (right) phrenic motoneuron inactivity following SH was verified by absence of electroneurographic (ENG) activity, and paralysis of the ipsilateral hemidiaphragm was verified by the absence of electromyographic (EMG) activity at the terminal experiment by using pairs of electrode wires that were implanted into the midcostal diaphragm muscle, as previously described (Fig. 1) (Dow et al., 2006; Dow et al., 2009; Khurram et al., 2017; Mantilla et al., 2011; Trelease et al., 1982). In the SHAM-SH group, the spinal cord was exposed by cutting the dura but the cord was not sectioned.

Figure 1.

Two models of diaphragm muscle paralysis were used in this study. Spinal hemisection at C2 (SH) inactivates both phrenic motoneurons and diaphragm muscle, as evidenced by the absence of ipsilateral ENG and diaphragm EMG activity. Tetrodotoxin (TTX) block of the phrenic nerve maintains phrenic motoneuron activity but inactivates the diaphragm muscle, as evidenced by the presence of ipsilateral ENG activity and lack of ipsilateral diaphragm EMG activity.

TTX Blockade.

Details of this model have also been previously published (Mantilla et al., 2007; Mantilla and Sieck, 2009; Prakash et al., 1999; Zhan and Sieck, 1992). Animals were anesthetized and the right phrenic nerve was exposed in the neck. A Silastic cuff (OD 2 mm; ID 1.5 mm, 2 mm long) was placed around the phrenic nerve and sutured to surrounding muscles. A tunneled polyethylene cannula connected the cuff to a miniosmotic pump (OD 7 mm, 30 mm long; weight 1.1 g), which was subcutaneously implanted in the dorsum of the animal. The miniosmotic pump delivered 0.0125% TTX at a constant rate of 0.5 μ1/h over 2 weeks. In the SHAM-TTX group, the pump was filled with saline. The absence of ipsilateral diaphragm EMG activity at the terminal experiment confirmed paralysis of the ipsilateral hemidiaphragm (Fig. 1).

Phrenic Motoneuron Labeling.

Phrenic motoneurons were retrogradely labeled as previously described using intramuscular or intrapleural techniques (Fogarty et al., 2018; Mantilla et al., 2009; Prakash et al., 2000; Prakash et al., 1993; Zhan et al., 2000). For intramuscular injection, two days prior to the terminal experiment, animals were anesthetized, the right midcostal diaphragm muscle was exposed via a subcostal abdominal incision and injected with 5–10 μ1 of either 2% tetramethylrhodamine dextran (MW=3000 kD; ThermoFisher, Waltham, MA) or 1% cholera toxin subunit B (CTB; List Biologicals Labs, Inc., Campbell, CA) at each of 5 separate sites. Following the injections, the incision was closed and the animals recovered for 2 days. For intrapleural injection, Alexa Fluor 488-conjugated CTB (15 μl of a 0.2% solution, ThermoFisher) was administered ipsilaterally (Mantilla et al., 2009).

At the terminal experiment (2 weeks post-SH, TTX or SHAM surgery), animals were anesthetized and transcardially-perfused with heparinized saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). The cervical spinal cord was then rapidly excised and post-fixed in 4% paraformaldehyde at 4°C overnight, followed by 25% sucrose in 0.1 M phosphate buffer for 24 h or more until sectioning. Coronal sections (100 μm thick) were cut on a cryostat (Reichert-Jung) and placed in phosphate buffer.

In a subset of animals injected with CTB, dendritic analyses were conducted using spinal cord sections that were incubated with 10% normal donkey serum in 0.1–1% Triton TBS, followed by anti-CTB (1:8,000; List Biologicals), biotinylated donkey anti-goat IgG (1:1,000; Jackson Immunoresearch, West Grove, PA) and then horseradish peroxidase-conjugated streptavidin (1:1000; Jackson Immunoresearch). Peroxidase was visualized using a glucose oxidase-imidazole-diaminobenzidine solution (Smithson et al., 1984), which yields a brown deposit in labeled neurons easily viewed under brightfield illumination. The sections were washed in TBS, mounted on gel-subbed slides, dehydrated in graded alcohols, and coverslipped with DPX mounting media (Fluka, MilliporeSigma, St. Louis, MO).

In additional studies, the spinal cord was excised 72h after intrapleural tracer injection, fixed in 4% paraformaldehyde, and embedded in 2% agarose. Subsequently, 150 μm coronal sections of the spinal cord were cut using a Vibroslicer (Leica, Buffalo Grove, IL). Sections containing the ventral horn were placed on the bottom of a Sylgard (Dow Corning, Midland, MI)-lined dish and gently affixed with pins. The use of Alexa 488-CTB labeling of phrenic motoneurons aided in guiding the lucifer yellow (LY)-microelectrode through the membrane of superficially-located phrenic motoneurons within the lightly-fixed tissue (Obregon et al., 2009). Retrogradely-labeled phrenic motoneurons were visualized using an Olympus upright microscope equipped with a Photometrics Cascade CCD camera (Photometrics, Tucson, AZ) and injected using custom-fabricated micropipettes filled with 2% aqueous solution of LY (MilliporeSigma) in 0.5 mM lithium chloride (Obregon et al., 2009). Negative current pulses (100 ms, 1.0–1.5 nA, 3–4 Hz) were passed through the microelectrode with a Grass S88 stimulator (Natus Medical Inc, Pleasanton, CA.) (Ermilov et al., 2003; Ermilov et al., 2004). A 10–20 min injection of LY yielded complete filling of phrenic motoneurons, including dendritic trees. Spinal cord sections were then placed on a glass slide while preserving their orientation, dehydrated in graded alcohols, and coverslipped using DPX mounting media.

Phrenic Motoneuron Analyses.

Fluorescently labeled ipsilateral phrenic motoneurons (tetramethylrhodamine dextran or CTB) were visualized using a laser-scanning confocal system with an Olympus DApo x40 oil immersion lens (NA 1.3) and a step size of 1 μm. Images were analyzed using a multidimensional display, processing and measurement package (Analyze; AnalyzeDirect, Overland Park, KS). The procedures for phrenic motoneuron somal volume measurements have been described previously (Fogarty et al., 2018; Issa et al., 2010; Prakash et al., 2000; Prakash et al., 1993; 1994). Surface areas of motoneuron somata were estimated for a prolate spheroid by measuring the major and minor diameters (Fig. 2). The number of primary dendrites of each motoneuron was determined from 3D reconstructions, and dendritic diameter was measured from reformatted images orthogonal to the dendrites (measured at a distance of 15 μm from the somata) (Fogarty et al., 2018; Issa et al., 2010; Prakash et al., 2000; Zhan et al., 2000).

Figure 2.

Representative retrogradely-labeled phrenic motoneurons in a longitudinal section of the rat cervical spinal cord from a control animal. Note clustering of phrenic motoneurons with 3–5 motoneurons easily discerned (a). A single phrenic motoneuron (denoted by *) was sampled per cluster based on the completeness of the dendritic tree (at least 5 primary dendrites were contained in the section; b). Surface areas of motoneuron somata were estimated for a prolate spheroid by measuring the major and minor diameters of each motoneuron from the 3D reconstruction. Primary dendrites were identified and their orthogonal diameter was measured at 15 μm from the cell soma for estimates of dendritic surface area (see Methods).

To render 3D representations of CTB-labeled ipsilateral phrenic motoneurons for dendritic measurements, a computer-controlled neuron tracing and morphometric system (NeuroLucida, MBF Bioscience, Williston, VT) was used. Labeled phrenic motoneurons were visualized on an Olympus BH-2 microscope with an Olympus DPlan x40 lens. The morphometric system was calibrated using a stage micrometer, and the Z-axis was controlled using a stepper motor, with a resolution of 0.1 μm. Dendrite length and branching were measured for intact phrenic motoneurons systematically sampled every 100 μm. Phrenic motoneurons were excluded if they were clearly damaged by the sectioning process. In each animal, at least 15 phrenic motoneurons were sampled.

LY-injected phrenic motoneurons were imaged on an Olympus confocal system with x60 oil immersion lens (NA 1.4) and a step size of 0.8 μm. A sequence of adjacent image stacks were captured for each phrenic motoneuron in order to image entire dendritic trees that extended beyond the field of view, ensuring that images on either side (top and bottom) of visible neuronal structures were included (Fig. 3). The step size was determined based on empirically calculated point spread function, and registration was verified empirically using fluorescently labeled microspheres (10 μm in diameter).

Figure 3.

Representative fuzzy-gradient volume renderings of dendritic tree objects created from optical sections of adult rat phrenic motoneurons. Separate confocal stacks of volume rendered images (a) and merged 3D image (b) from LY filled phrenic motoneurons. (c) XY view of 3D-reconstructed dendritic trees selected for analyses.

Eleven dendritic trees from LY-injected phrenic motoneurons (Fig. 3) were semi automatically segmented in Analyze and assigned to objects based on voxel intensities (Obregon et al., 2009; Prakash et al., 2000). Only phrenic motoneurons showing secondary order branching and greater were used for morphological analyses. The optimal correlation between dendritic tree surface area (A) and primary dendrite diameter (d₀) was obtained using the following power equation:

$$A=1161.5d_0{}^{1.03}$$

This equation was used to estimate the surface area of the dendritic tree from measurements of primary dendrite diameters.

Statistical Analyses.

Morphological parameters were compared across experimental groups using one-way ANOVA. When appropriate, post-hoc analyses were conducted using the Tukey-Kramer honestly significant difference test. Statistical significance was established at the p<0.05 level. Initial analysis revealed no significant difference across SHAM-SH, SHAM-TTX and control groups. Therefore, these groups were pooled in further statistical analyses and will be denoted as control from this point forward. All values presented are mean ± SE, unless otherwise stated.

RESULTS

Animal body weight increased by ~20% across all experimental groups over the 2 week experimental period. By the third day after SH, all animals displayed normal ambulation with only minor deficits in fine motor control. Both SH (n=5) and TTX (n=5) animals displayed altered movements of the rib cage and abdomen resulting from hemidiaphragm paralysis (Miyata et al., 1995; Prakash et al., 1999).

Verification of Surgical Procedures.

Histological examination of the SH spinal cords confirmed lesions restricted to the lateral and ventral funiculi. In TTX rats, there was no histological evidence of phrenic nerve damage at the site of the nerve cuff placement. Phrenic motoneuron inactivity following SH was verified by absence of ENG activity (Fig. 1). In contrast, ENG recordings of the phrenic nerve proximal to the TTX block showed persistent activity of phrenic motoneurons (Fig. 1). In all SH and TTX animals, ipsilateral diaphragm EMG activity was absent immediately after surgery. SH animals were included in this study only if there was absence of ipsilateral diaphragm EMG activity at 14 days after surgery.

Phrenic Motoneuron Labeling.

In all groups, phrenic motoneurons were extensively labeled by both fluorescent dextrans and CTB (Fig. 2 and 4) and only ipsilateral to SH or TTX, forming a column located in the medial aspect of the ventral horn from the C3-C6 segments. There were no apparent differences in the spatial pattern of phrenic motoneuron labeling across the experimental groups, specifically TTX where partial denervation from potential TTX toxicity would have been a confounding factor.

Figure 4.

Representative examples of camera lucida drawings of ipsilateral phrenic motoneuron dendritic architecture from control, C2 spinal hemisection (SH), and tetrodotoxin (TTX) nerve blockade models (shown at 0° and 90° relative to the plane of tissue section). Phrenic motoneurons were retrogradely labeled with cholera toxin subunit B (CTB).

The somata of labeled motoneurons appeared distinctly longer in the rostrocaudal axis, consistent with previous findings (Goshgarian and Rafols, 1981; Prakash et al., 2000; Prakash et al., 1993). Despite the extensive labeling, individual phrenic motoneurons could be easily discerned (Fig. 2), most likely because of discrete injections into the diaphragm muscle. Labeled motoneurons were often clustered in groups of 3–5 motoneurons. In such cases only a single phrenic motoneuron was sampled per cluster based on the completeness of the dendritic tree. Other sampled motoneurons were not included in clusters. This selection facilitated identification and tracking of the discrete dendritic trees of each identified motoneuron.

While extensive dendritic labeling was observed in all experimental groups using both fluorescent dextrans and CTB, more distal dendritic branches (on occasion, up to the fifth order) were often observed mostly with CTB (Fig. 2). There was extensive overlap in the dendritic fields of phrenic motoneurons, which additionally complicated the analyses of distal dendrites. The predominant orientation of dendrites was in the rostrocaudal axis, however dendritic branching in the mediolateral axis was also apparent. Dorsolateral projections were found, but these were not extensive. In most cases, dendritic branching was restricted to the ipsilateral spinal cord; however, some dendrites projected to the contralateral side. There were no apparent qualitative differences in the dendritic patterns between phrenic motoneurons of SH vs. control animals. However, compared to either group, phrenic motoneurons in the TTX group appeared to have greater dendritic arborization along the mediolateral axis. There were no apparent differences in the extent of contralateral projections across experimental groups. Overall, quantitative analyses were performed on 169, 92, and 74 phrenic motoneurons from control, SH and TTX groups, respectively. Only ipsilateral phrenic motoneurons whose cell body and primary dendrites were completely contained within the tissue section (displaying 5 or more primary dendrites) were included in analyses.

Phrenic Motoneuron Somal Dimensions.

The long (F_2,334 = 264; p < 0.001) and short diameter (F_2,334 = 880; p < 0.001) of labeled phrenic motoneurons was different across groups. The long diameter in the SH group (39 ± 1 μm) was significantly less than control (48 ± 1 μm), which was significantly less than the TTX group (79 ± 2 μm). The short diameter in the SH group (21 ± 1 μm) was significantly less than control (22 ± 1 μm), and the TTX group (19 ±1 μm) was significantly less than control and SH. Typically, the long diameter of labeled phrenic motoneurons was approximately 2- to 4-fold greater than the short diameter, supporting the assumption of a prolate spheroidal shape (Prakash et al., 1993). Based on this assumption, and the diameter measurements, somal volume and surface area were estimated.

Phrenic motoneuron somal volume was different across groups (Fig. 5; F₂,₃₃₄ = 116; p < 0.001). In the SH group, phrenic motoneuron somal volume was significantly decreased (range: 5,523–13,729 μm³) compared to control (range: 7,461–19,608 μm³). In contrast, with TTX, somal volume was increased (range: 8,903–23,736 μm³) compared to control as well as SH. Similarly, differences in somal surface area were evident across groups (Fig. 6; F_2,334 = 178; p < 0.001). Somal surface area was also reduced in SH (range: 1,616–4,098 μm²) compared to control (range: 1,981–5,476 μm²), but increased in TTX (range: 2,723–7,740 μm²). The distribution of phrenic motoneuron somal volumes and surface areas was determined using Z-scores based on the mean of each experimental group. In all 3 groups, somal volumes displayed a unimodal distribution, with the control group being similar to that observed in a previous study (Prakash et al., 2000). Compared to control, the distribution of somal volumes in SH animals was shifted to the left with an overall decrease in volume (Fig. 5). In contrast, in TTX animals, somal volume distribution was shifted to the right, with an overall increase in volume (Fig. 5). Similar shifts in the distribution of somal surface area were evident (Fig. 6).

Figure 5.

Phrenic motoneuron somal volume was different across groups. (a) Distribution of phrenic motoneuron somal volumes in control, SH and TTX animals. (b) Distribution of Z-scores for the somal volumes in control, SH and TTX animals. Note the leftward shift in volume distribution with SH, indicating a decrease in the volume of larger phrenic motoneurons. In contrast, with TTX, the distribution is shifted to the right.

Figure 6.

Phrenic motoneuron somal surface area was different across groups. (a) Distribution of phrenic motoneuron surface areas in control, SH and TTX animals. (b) Distribution of Z-scores for somal surface areas in control, SH and TTX animals. Note the leftward shift in surface area distribution with SH, indicating a decrease in the surface area of larger phrenic motoneurons. In contrast, with TTX, the distribution is shifted to the right.

Phrenic Motoneuron Dendritic Dimensions.

The number of primary dendritic branches and the primary dendrite diameter were not significantly different between phrenic motoneurons of control and SH animals (Table 1). Phrenic motoneurons from TTX animals had significantly more primary dendrites (F₂,₃₃₄ = 6.03; p = 0.003) as well as larger primary dendrite diameter (F₂,₃₃₄ = 24.9; p < 0.001) than control and SH animals.

Table 1:

Effect of spinal hemisection (SH) vs. tetrodotoxin (TTX) blockade on phrenic motoneuron dendritic morphology

Parameter	Control	SH	TTX	ANOVA p value
No. of Primary Dendrites	7.6 ± 0.1	7.3 ± 0.2	8.3 ± 0.2*†	p=0.003
Primary Dendrite Diameter (μm)	2.4 ± 0.1	2.3 ± 0.1	2.8 ± 0.1*†	p<0.001
Total Dendritic Length (μm)	3,748 ± 319	4,914 ± 361	4,844 ± 414	p=0.044
Average Dendritic Length (μm)	81.5 ± 4.4	69.1 ± 6.1	93.2 ± 6.7†	p=0.015
Estimated Dendritic Surface Area (μm2)	22,722 ± 848	21,219 ± 990	29,657 ± 1,509*†	p<0.001
Estimated Receptive Field (μm2)	78,764 ± 5,207	99,471 ± 6,744	87,831 ± 6,942	p=0.072
Total Motoneuron Surface Area (μm2)	26,010 ± 899	23,796 ± 1,043	34,371 ± 1,619*†	p<0.001

Data presented are mean ± SE. SH: C2 spinal hemisection; TTX: tetrodotoxin block of the phrenic nerve.

^*indicates statistical significance from control (p<0.05).

^†indicates statistical significance from SH (p<0.05) after post-hoc analyses.

Estimates of dendritic surface area based on primary dendrite diameter were developed using a sample of intracellularly labelled motoneurons from control rats. Eleven full dendritic arbors showing minor damage from sectioning were manually segmented and assigned to objects based on voxel intensities (Fig. 3). Dendritic branching ranged from 3^rd to 7^th order. The space occupied by the neuron was arbitrarily divided into several regions with the landmarks common to all or at least to the adjacent regions in order to capture the entire visible dendritic arbor. For each region a stack of images was separately collected, volume reconstructed and merged, as described previously (Ermilov et al., 2000). Dendritic branches were widely dispersed within the depth of spinal cord, and only the volume rendered images of dendritic trees (n = 11) that were entirely within the field of view and had no visible damages were selected for 3D morphometry, based on a previous study (Obregon et al., 2009). A power model resulted in the best fit correlation between dendritic tree surface area and primary dendrite diameter (r² = 0.880).

Dendritic surface area was estimated from primary dendrite diameter of labeled phrenic motoneurons using this equation. Estimated dendritic surface area was not significantly different between control and SH, but increased with TTX (Fig. 7; F_2,334 = 13.6; p < 0.001). The distribution of estimated dendritic surface area was skewed to the right with TTX. Similarly, total phrenic motoneuron surface area was not significantly different between control and SH, but increased with TTX (Table 1; F_2,334 = 18.3; p < 0.001).

Figure 7.

Estimated phrenic motoneuron dendritic surface area was not significantly different between control and SH, but increased with TTX. (a) Distribution of phrenic motoneuron dendritic surface area in control, SH and TTX animals. (b) Distribution of Z-scores for dendritic surface areas of phrenic motoneurons in control, SH and TTX animals. Note the unchanged distribution for phrenic motoneurons in SH animals, and the significant right shift in the distribution of the TTX group.

In addition to the estimated surface area, the total area across which dendrites from an individual phrenic motoneuron (receptive area) was calculated from a polygon connecting the ends of the most distal dendrites visualized in CTB-labeled phrenic motoneurons. In these measurements, dendrites were found to expand to similar extents as control following SH and TTX (p = 0.072; Table 1). However, the average dendritic length was increased after TTX compared to SH (p = 0.015; Table 1).

DISCUSSION

The present study examined phrenic motoneuron morphology in a model of phrenic motoneuron inactivity induced by SH, and compared these findings to those induced by diaphragm muscle inactivity in the absence of phrenic motoneuron inactivity (utilizing a model of unilateral TTX block of the phrenic nerve). Ipsilateral phrenic motoneuron inactivity resulted in decreased somal dimensions but unchanged dendritic dimension, supporting our hypothesis and consistent with an overall increase excitability in the phrenic motor pool. In contrast, diaphragm muscle paralysis without inactivity of phrenic motoneurons resulted in increased phrenic motoneuron size. Our data suggest that morphological adaptations of phrenic motoneurons to inactivity are likely dependent on intrinsic motoneuron properties and/or altered descending synaptic input (not on target muscle activity). These changes likely occur in order to maintain or restore intrinsic excitability of the phrenic motoneuron pool.

Adaptations of Phrenic Motoneurons to SH and TTX

Although diaphragm muscle paralysis was evident in the SH and TTX models, phrenic motoneurons were inactive in SH animals, while phrenic motoneuron activity was present in the TTX group. In fact, in previous studies (Miyata et al., 1995; Zhan et al., 1997), we found that TTX increased integrated phrenic nerve activity by ~50%; however, whether this increase is maintained for the 2 week survival period is not known. In other studies where diaphragm paralysis is induced by unilateral denervation (Gill et al., 2015; Khurram et al., 2017), EMG activity in the intact hemidiaphragm increases. Regardless, SH is associated with concordant inhibition of phrenic motoneuron and diaphragm muscle activity, while TTX is associated with discordance between motoneuron and muscle activities.

The present study found that SH and TTX differed in the pattern of somal and dendritic alterations. In addition, adaptations in dendritic arborization did not match changes in somal volume and surface areas. Ipsilateral motoneuron somal volumes and surface areas decreased with SH, but increased with TTX. Although dendritic receptive fields were similar with both SH and TTX, the mean branch length was significantly increased following TTX compared to SH. Following SH, there was no change in estimated total surface area, whereas total surface area increased following TTX. However, given that dendritic and total surface area measurements are indeed estimates, they must be interpreted with caution. Regardless, the somal vs. dendritic changes observed with SH are consistent with our hypothesis that phrenic motoneuron adaptations to SH serve to increase excitability, which may contribute to spontaneous restoration of function (Gransee et al., 2017; Gransee et al., 2013; 2015; Hernandez-Torres et al., 2017; Mantilla et al., 2012; Mantilla et al., 2013a; 2017; Mantilla et al., 2014; Mantilla et al., 2013b; Martinez-Galvez et al., 2016). The present study specifically selected animals not displaying recovery of rhythmic diaphragm EMG activity post-SH in order to avoid the confounding effects of restored function, but the underlying mechanisms are expected to be similar in animals displaying recovery vs. those not.

Although muscle inactivity with TTX was persistent for the duration of the study (paralysis was verified at the terminal experiment), it is not possible to ascertain the level of change (if any) in motoneuron activity (at the cell soma). In the absence of direct measurements of motoneuron activity at the terminal experiment (precluded by sampling issues of single motoneuron recordings), we cannot incorporate the morphological changes seen after TTX with changes in motoneuron activity. Based on recent studies from our group examining changes in contralateral diaphragm EMG activity following denervation (Gill et al., 2015; Khurram et al., 2017), there is a suggestion that descending inputs to phrenic motoneurons increase ~50%, but it is not possible to ascertain whether there are motor unit type differences in activation levels in these experiments. It is likely that phrenic motoneuron cell bodies experience greater drive two weeks after TTX, resulting in increased somal volume and surface area, as well as increased dendritic surface area. It is expected that with the TTX phrenic nerve blockade conduction along phrenic afferents is also blocked. However, activity of other afferents relevant to respiratory behaviors (e.g., chest wall or airway afferents) is expected to be unimpaired, although could be affected by changes in the respiratory pattern. Although this is possible, extensive experience in the SH or unilateral denervation models documents minimal changes in respiratory rate, tidal volume or transdiaphragmatic pressure during ventilatory behaviors, and no change in arterial blood gases (Gill et al., 2015; Gransee et al., 2017; Gransee et al., 2013; 2015; Hernandez-Torres et al., 2017; Khurram et al., 2017; Mantilla et al., 2013a; 2017).

There is currently no published information on the effects of inactivation due to SH on gross morphological changes of phrenic motoneurons. However, several studies have demonstrated significant ultrastructural changes in the cervical spinal cord following SH (Goshgarian et al., 1989; Hadley et al., 1999), including increased number of “double synapses” and increased length of dendro-dendritic appositions. These changes, which are more prominent with time, are unlikely to be induced by the SH itself since cold block of descending cervical drive also induces similar changes (Castro-Moure and Goshgarian, 1997). Taken together, these results support ongoing plasticity that aims to maintain or restore excitability of the phrenic motoneuron pool.

Although phrenic motoneuron dendritic arborization is primarily limited to the ipsilateral spinal cord (Lindsay et al., 1991), a previous study suggested there are occasional contralateral projections of phrenic motoneurons (Prakash et al., 2000). These dendritic projections may allow for phrenic motoneuron activation from the contralateral side, and may be especially important following weeks or months of SH in conjunction with any increased (or previously ineffective) contralateral supraspinal axonal input. In the present study, we found only occasional contralateral dendritic projections in both control and SH animals. Since the overall span of dendritic projections did not change significantly with SH, it is likely that axonal decussations from intact contralateral (descending) projections play a more important role in the spontaneous recovery of ipsilateral diaphragm activity following SH (Mantilla and Sieck, 2003).

Phrenic Motoneuron Morphology

As described in previous studies including our own (Fogarty et al., 2018; Furicchia and Goshgarian, 1987; Goshgarian and Rafols, 1981; Issa et al., 2010; Prakash et al., 2000), characteristic features of phrenic motoneurons were observed in the present study, including rostrocaudal phrenic motoneuron clustering with preferred rostrocaudal and mediolateral dendritic orientations. This organization may act to integrate activity of phrenic motoneurons during the respiratory cycle, as well as facilitate interactions between phrenic motoneurons and adjacent motoneurons responsible for controlling accessory respiratory muscles (Scheibel and Scheibel, 1970; 1971). Afferent input from the brainstem has also been shown to follow the somatic and dendritic distribution of phrenic motoneurons (Ellenberger and Feldman, 1988; Ellenberger et al., 1990; Furicchia and Goshgarian, 1987; Wu et al., 2017).

In the present study, we integrated 3D confocal microscopy and dendritic modeling to determine changes in ipsilateral phrenic motoneuron morphology. This integrated approach significantly reduced the analytical burden created by the large volumes of 3D confocal image data. Furthermore, the phrenic motor nucleus spans ~1 cm along the spinal cord making it virtually impossible to create 3D sets of image data of sufficient resolution for detailed analysis of neuronal morphometry. A shape-based stereological technique was used to estimate phrenic motoneuron somal volumes and surface areas. We (Issa et al., 2010; Prakash et al., 1993) and others (Cameron and Fang, 1989; Cullheim et al., 1987a; Ulfhake and Cullheim, 1988) have previously reported that rat phrenic motoneurons can be largely represented by a prolate spheroidal shape. Furthermore, volume and area estimates assuming such geometry are <10% different from actual 3D confocal measurements (Prakash et al., 1993). Thus, the stereological approach significantly reduced the burden of analysis without biasing or altering the results.

Given the complex and extensive dendritic arborization of phrenic motoneurons, it was not possible to perform detailed and complete dendritic analyses from 3D confocal images, especially since only sections of spinal cord were imaged. Commonly employed techniques are based on retrograde labeling of specific motoneuron pools by intramuscular or intraneural injection (Boulenguez et al., 2007; Gordon and Richmond, 1990; Issa et al., 2010; Prakash et al., 2000; Yates et al., 1999; Zhan et al., 2000). The method of intrapleural injection allows specific labeling of phrenic motoneurons and permits near complete labeling of the entire motoneuron pool across developmental ages (Mantilla et al., 2009). However, filling of distal dendrites with any of these retrograde techniques is usually limited, regardless of fluorescent dye or peroxidase selection. In this study, we combined labeling of phrenic motoneurons via intrapleural injection with Alexa Fluor 488-conjugated CTB with intracellular injection of LY into these visually identified neurons in lightly fixed-spinal cord slices (C3-C6) (Obregon et al., 2009). This combined approach allows sampling of a relatively large number of motoneurons per animal and detailed three-dimensional reconstruction of the dendritic tree of phrenic motoneurons from confocal slices. The combination of these labeling and imaging techniques considerably improved image quality and allowed reliable 3D visualization and direct morphometry of the identified phrenic motoneurons with an entire extension of their visible dendritic trees.

Accordingly, analysis of dendritic arborization was performed based on the novel model derived using 3D reconstructions of intracellularly-filled adult rat phrenic motoneurons (Obregon et al., 2009), without morphological assumptions that are used in dendritic models such as those developed by Burke et al. (Burke et al., 1992). For this calculation, dendritic surface area was estimated from primary dendrite diameter (which was reliably measured from the confocal images with <10% error). The reliability of this technique is underlined by the consistency of our estimates of dendritic surface area in control phrenic motoneurons with those previously reported by other investigators for α-motoneurons, including phrenic motoneurons (Cameron et al., 1985; Cameron et al., 1991) and hindlimb motoneurons (Cullheim et al., 1987a; b). In addition, an estimate of dendritic receptive field was obtained based on visible extent of dendritic branching.

Mechanisms Underlying Phrenic Motoneuron Adaptations

One mechanism that may mediate neuronal adaptations to altered activity is the role of trophic factors. Such trophic factors may be muscle-derived or neuron-derived, with both local (autocrine) and distant targets. Furthermore, the specific production of trophic factors or the localization of their receptors may dictate motor unit specificity of the response to trophic factors. Expression of neurotrophic factors such as ciliary neurotrophic factor, brain-derived neurotrophic factor, neurotrophin-3 and neurotrophin-4 change following lumbar spinal cord injury (Bennett et al., 1999; Bregman et al., 1997; Ikeda et al., 2001; Mantilla and Sieck, 2008; Oyesiku et al., 1997; Satake et al., 2000; Widenfalk et al., 2001). However, an important distinction with regard to limb muscles, compared to the diaphragm muscle, is whether spinal transection actually produces hindlimb motoneuron inactivation since the pattern generators are located in the spinal cord (rather than the brainstem as for phrenic motoneurons). In our study, SH clearly inactivated phrenic motoneurons. Recent studies from our laboratory support the role of neurotrophic factors acting on phrenic motoneurons in neuroplasticity underlying recovery of rhythmic diaphragm activity after SH (Gransee et al., 2013; 2015; Hernandez-Torres et al., 2017; Mantilla et al., 2013a; Mantilla et al., 2014; Martinez-Galvez et al., 2016). Future studies could determine whether SH and TTX nerve blockade induce differential expression of neurotrophic factors in phrenic motoneurons to elucidate underlying, intrinsic motoneuron adaptations to inactivity. A previous study comparing SH and TTX induced adaptations at the neuromuscular junction suggested indeed that the predominant changes occurred at the motoneuron axon terminal (Mantilla et al., 2007).

In both SH and TTX models, the potential for muscle-derived neurotrophic influence likely remained intact since axonal transport was unaffected (Miyata et al., 1995; Zhan and Sieck, 1992). However, possible changes in the expression of muscle-derived trophic factors (Dohrmann et al., 1986; Funakoshi et al., 1995; Griesbeck et al., 1995; Henderson et al., 1994) have not been examined following either SH or TTX nerve blockade. Furthermore, the release of trophic factors may be constitutive and may reflect motor unit type differences. The results of the present study are suggestive of such motor unit-specific responses in phrenic motoneuron morphology.