Targeting drug or gene delivery to the phrenic motoneuron pool

Abstract

Phrenic motoneurons (PhrMNs) innervate diaphragm myofibers. Located in the ventral gray matter (lamina IX), PhrMNs form a column extending from approximately the third to sixth cervical spinal segment. Phrenic motor output and diaphragm activation are impaired in many neuromuscular diseases, and targeted delivery of drugs and/or genetic material to PhrMNs may have therapeutic application. Studies of phrenic motor control and/or neuroplasticity mechanisms also typically require targeting of PhrMNs with drugs, viral vectors, or tracers. The location of the phrenic motoneuron pool, however, poses a challenge. Selective PhrMN targeting is possible with molecules that move retrogradely upon uptake into phrenic axons subsequent to diaphragm or phrenic nerve delivery. However, nonspecific approaches that use intrathecal or intravenous delivery have considerably advanced the understanding of PhrMN control. New opportunities for targeted PhrMN gene expression may be possible with intersectional genetic methods. This article provides an overview of methods for targeting the phrenic motoneuron pool for studies of PhrMNs in health and disease.

INTRODUCTION

The diaphragm is a thin, dome-shaped muscle forming a barrier between the thoracic and abdominal cavities. Considered the primary muscle of inspiration, the diaphragm also serves a postural function (1) and is active during expulsive behaviors such as coughing and vomiting (2, 3). Voluntary and automatic (e.g., breathing related) diaphragm activation occurs because of phrenic motoneuron (PhrMN) depolarization and subsequent acetylcholine release at the phrenic-diaphragm neuromuscular junction. Thus, PhrMNs are the “final common pathway” (4) used by the central nervous system to control diaphragm contraction.

Broadly, research on the phrenic motor system can be grouped into several categories: 1) the neural regulation of PhrMN activity in health and disease, 2) mechanisms of neuroplasticity in the phrenic motor system, and 3) therapeutic approaches to improve PhrMN function in disease states. Research in these areas often requires targeted delivery of substances to PhrMNs (e.g., drugs, anatomical tracers, and viral vectors). Targeting PhrMNs is not a trivial problem, as these cells are located in the ventral cord (lamina IX) and the surrounding cervical cord contains multiple motoneuron pools and complex interneuron networks (5). In this article, we review methods that can be used to target drug delivery or gene expression to PhrMNs. The goal is to inform researchers studying therapeutic strategies, as well as to inform studies aimed at understanding the neural control of PhrMNs and/or mechanisms of neuroplasticity in health and disease. The text covers techniques for targeting PhrMNs in adult mammals. Many of the methods described here could likely be adapted for use in younger and/or neonatal mammals, but that is not the focus of the present article.

OVERVIEW OF PhrMNs AND THE PHRENIC NERVE

PhrMNs are found bilaterally in longitudinal columns in the midcervical spinal cord (i.e., the phrenic motoneuron pool; Fig. 1). These columns are located in the ventral horn (lamina IX), spanning approximately C3–C6. The C3–C6 distribution has been reported in multiple species including rat (9), mouse (10), rabbit (11), macaque (12), and humans (13). Some variation in the rostral-caudal distribution of PhrMNs has been reported in other species including C4–C6 in cats (14, 15), C4–C7 in guinea pig (16), and C5–C7 in ferret (17). Based on retrograde labeling studies (see targeting phrmns via injection to the intrapleural space, retrograde targeting of phrmns via the phrenic nerve, and delivering substances to the diaphragm to target phrmns), the total number of PhrMNs per side has been estimated at 200 in mice (10) and 220–250 in rats (6).

Figure 1.

The phrenic motoneuron pool. Phrenic motoneurons (PhrMNs) are located in a column extending from approximately the third to the sixth cervical segment. The columns are bilaterally located in the ventral (anterior) gray matter. The drawing in A depicts PhrMN location in the ventral horn. B shows histological images of PhrMNs taken from the plane of section indicated in A. To obtain these images, PhrMNs were retrogradely labeled by intrapleurally delivering cholera toxin β-subunit (CT-β) in an adult Sprague-Dawley rat (see Methods and Limitations of Intrapleural Delivery). Three days following after delivery, tissue was harvested and processed with a CT-β antibody as described previously (6–8). Scale bars, 500 μm.

In the adult mammal, PhrMNs are found in clusters that form a rostro-caudal oriented column (18). The dendrites of PhrMNs display a striking rostro-caudal orientation, with a small proportion of radially projecting dendrites. Some PhrMNs dendrites may project to the contralateral spinal cord, although this is likely to be very rare (7, 19). PhrMNs have a large neuropil, with >95% constituting the dendritic tree and extending over 2–3 mm (20, 21). In the adult rat, PhrMN soma size is unimodally distributed, ranging from 2,000 to 8,000 µm² with an overall left-skewed distribution (22).

PhrMNs of smaller somal surface areas generally innervate slow-twitch, fatigue-resistant (type S) motor units and fast-twitch fatigue-resistant (type FR) motor units. Larger PhrMNs innervate fast-twitch, fatigue-intermediate (type FInt) motor units and fast-twitch, fatigable (type FF) motor units (23). Heterogeneity is present in the distribution of excitatory glutamatergic inputs to PhrMNs (7) as well as the distribution of ionotropic N-methyl-d-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors on PhrMNs (24, 25). Smaller PhrMNs have a higher density of presynaptic glutamatergic inputs as well as a greater density of NMDA and AMPA receptor mRNA compared with the larger motoneurons. These features might contribute to the orderly recruitment of PhrMNs to accomplish behaviors requiring varying degrees of force generation (e.g., breathing, sighing, coughing) (23).

The axons of PhrMNs travel in the phrenic nerve, which provides efferent motor innervation to the entire diaphragm. The phrenic is a mixed motor and sensory nerve containing myelinated efferent axons and myelinated and unmyelinated afferents (26). In rats, the phrenic nerve arises from axons leaving the C3–C6 cervical spinal segments (27, 28). Human cadaveric studies indicate that the phrenic nerve arises exclusively from the C3 and C4 ventral roots in some cases but C3, C4, and C5 in others (29). An accessory phrenic nerve arises from the 5th cervical ventral ramus and then joins the main phrenic nerve inside the thoracic cavity (29, 30). The accessory phrenic nerve contains ∼10% of phrenic motor axons in rats (9).

TARGETING PhrMNs VIA INJECTION TO THE INTRAPLEURAL SPACE

The visceral pleura lines the lungs, and the parietal pleura covers the inner surface of the thoracic cavity, including the diaphragm, mediastinum, pericardium, and thoracic wall. A single layer of mesothelial cells comprises the parietal and visceral pleura, surrounded by connective tissue that has lymphatic and blood vessels giving access to the systemic circulation (31). The pleural cavity is a fluid filled “potential space” between the visceral and parietal pleural membranes.

Delivery of substances such as adenoviral vectors or gene-modified cell lines directly to the intrapleural space was first performed >20 years ago as a strategy to treat pleural tumors (32, 33). Subsequent studies established that intrapleural delivery of viral vectors could effectively drive gene expression in the lung and diaphragm (34, 35). For example, unilateral intrapleural administration of adenovirus (100 µL, 10⁹ plaque-forming units) in mice produces pleural membrane gene expression in both the left and right lung (34). Building upon that foundation, Mantilla and colleagues (6) showed that intrapleural delivery of the retrograde tracer cholera toxin β-subunit (CT-β) produces near-complete PhrMN pool labeling. That is, upon immunohistological evaluation, CT-β can be detected in nearly all PhrMNs. Figure 1 provides an example of PhrMN labeling using the method described by Mantilla et al. The intrapleural method has also been used to target PhrMNs with adeno-associated virus (AAV) (36), wheat germ agglutinin (WGA) (37), and small interfering RNAs (siRNAs) (38). In particular, intrapleural AAV delivery shows promise as a means of driving PhrMN gene expression for therapeutic purposes (36, 39, 40). For example, in a murine model of amyotrophic lateral sclerosis (ALS), intrapleural delivery of an AAV encoding microRNA that suppresses superoxide dismutase 1 mitigated axonal loss in the phrenic nerve, suggesting attenuation of PhrMN death (36).

An intrapleurally delivered substance reaching PhrMNs most likely requires movement across the parietal membrane. The pleural membrane is 10–20 µm thick, comprising mesothelial cells and basal lamina (41). Water and particles < 4 nm can pass directly through the pleural membrane, and particles up to 1,000 nm can cross the basal membrane via transcytoplasmic transport in mesothelial cells (42, 43). Accordingly, CT-β (∼3.6 nm) (44) or AAV particles (∼25 nm) (45, 46) can be transported across the parietal pleura. An intrapleurally delivered substance could enter the PhrMN axon via the neuromuscular junction, or at some point near the distal end of the axon. In the case of intrapleural AAV delivery (36, 39, 40), the process starts with the viral capsid binding to a cell surface receptor (47). Based on the AAV serotype, specific cell surface glycan receptors are required for the initial cellular uptake into the cell (48, 49). AAV is then trafficked into nonmotile endosomes, followed by exocytic vesicles and a retrograde-directed late endosome/lysosome compartment. For CT-β to reach PhrMN soma, it must first bind to ganglioside GM1 present on neuronal membranes (50, 51), and then it can be retrogradely transported.

Methods and Limitations of Intrapleural Delivery

An advantage of the intrapleural technique is the relative simplicity of the surgical procedure (52). The subject is first positioned with a clear view of the rib cage. Under sterile conditions, the retrograde tracer or other substance (e.g., AAV, siRNA) is injected to the intercostal anterior axillary line (typically the 5th intercostal space). Care must be taken to ensure no disruption to the mechanical coupling between the chest wall and lungs to avoid a pneumothorax. This is achieved by restricting the depth to which the needle can penetrate. In the rat, placing tubing around the outer diameter of the needle such that only 5–7 mm can penetrate works well. The size of the particular species as well as the chemical nature of the substance to be injected will impact the overall volume of solution used for intrapleural injection. For example, studies in adult mice using AAV for targeting diaphragm (39) or PhrMNs (36) have used injection volumes of 400 μL. In humans, volumes of ∼50,000 μL have been intrapleurally injected in studies of adenovirus-based targeting of malignant pleural effusion (53, 54). Considerably smaller intrapleural injection volumes (e.g., 15–50 μL) have been used successfully for targeting PhrMNs in the adult rat with retrograde tracers such as CT-β or WGA (6, 37).

The major limitation of intrapleural delivery is that PhrMNs are not uniquely targeted. The original study of intrapleural CT-β delivery reported some labeling in intercostal motoneurons along with robust PhrMN labeling (6). Preganglionic sympathetic neurons may also be targeted, since some sympathetic axons travel in the phrenic nerve (55). Another important consideration is that the lymphatics communicate directly with the pleural space, and intrapleural delivery will therefore unavoidably result in some degree of systemic distribution (31). In fact, gene therapy researchers have used intrapleural delivery to intentionally target widespread viral delivery (31). Vector genome copies can be found in the liver, spleen, kidney, diaphragm, and lungs after intrapleural AAV delivery (56). Widespread distribution can be minimized by using a lower dose of viral vector (or volume of solution) to the pleural space (31). A final consideration of the intrapleural method is the potential danger of misplaced dosing into the lung parenchyma producing pneumothorax. However, the risk of pneumothorax or infection following intrapleural injection is greatly minimized with appropriate surgical technique.

RETROGRADE TARGETING OF PhrMNs VIA THE PHRENIC NERVE

Phrenic nerve application of molecules that can move retrogradely (i.e., from the distal axon toward the cell soma) is an effective way to target PhrMNs. This method has proven successful with a variety of molecules including CT-β (57, 58), Cascade Blue or fast blue (57), FluoroGold (58, 59), rhodamine (60), pseudorabies virus (61), or horseradish peroxidase (HRP) (27). Figure 2 shows an example of PhrMN labeling following application of CT-β to the distal end of a sectioned rat phrenic nerve. Another approach is to “soak” the intact phrenic nerve in a small pool of WGA conjugated to Alexa fluorophore 488 (62). This avoids the complications of phrenicotomy-induced diaphragm paralysis and/or axonal degeneration and produces unilateral labeling of PhrMNs from C3–C6. Direct delivery of a substance to a peripheral nerve via microinjection (i.e., needle penetrating the epineurium) is another viable technique (63), although it has not been used extensively in the phrenic nerve. Texakalidis et al. (64) provided a proof of concept for phrenic nerve injection, although histological analysis of PhrMNs was not reported.

Figure 2.

Example of cervical spinal cord labeling including phrenic motoneurons (PhrMNs) and afferent projections. To obtain these images, the phrenic nerve was treated with cholera toxin β-subunit (CT-β) by the “nerve dip” method (see retrograde targeting of phrmns via the phrenic nerve) as described in our prior publication (57). Tissues were harvested from an adult Sprague-Dawley rat 3 days after the phrenic nerve was cut and exposed to CT-β. Tissues were incubated with a CT-β antibody as described previously (57). The method produces robust labeling of PhrMNs ipsilateral to the CT-β application (A, arrow). Note also the labeling of afferent projections in the dorsal horn of the spinal cord ipsilaterally (C) and contralateral to the delivery (B). Scale bars, 500 µm (A) and 250 µm (B and C).

Methods and Limitations of Retrograde Targeting of PhrMNs via the Phrenic Nerve

Targeting the PhrMN pool via the phrenic nerve requires the substance to reach the axon so that it is transported retrogradely to soma (58, 59). This can be accomplished by needle injection through the epineurium or by “soaking” the distal tip of a sectioned or intact nerve (62). Surgically accessing the phrenic nerve can be accomplished with the subject in a supine or prone position. Either approach requires an invasive surgery to locate the phrenic nerve, usually near the brachial plexus. The phrenic nerve can be sectioned and the retrograde tracing agent applied to the distal tip. This technique works very well, for example, with CT-β, Cascade Blue, or a fluorescent tracer conjugated to a dextran molecule (60, 65, 66) (see Fig. 2). Alternatively, a small vessel holding ∼1.5 µL of fluid can be placed under a surgically exposed, but uncut (i.e., intact), phrenic nerve. This method produced unilateral labeling of PhrMNs when WGA was applied for 1 h (62).

Direct intraneural delivery via microinjection is a relatively straightforward procedure (63). Approximately 1–3 µL of an agent is microinjected into the nerve through a small needle that pierces the epineurium. Typically, the needle is left in place for several minutes after injection to allow the agent to diffuse into the nerve. To our knowledge, this has not been applied to phrenic nerve, but studies of sciatic nerve injections of either recombinant adenovirus or AAV (serotype 2) demonstrate robust expression of transgenes in motoneurons (67).

The mixed composition of the phrenic nerve represents a possible challenge for this approach. Most phrenic nerve axons have soma located in the midcervical dorsal root ganglia (i.e., phrenic afferent neurons) or PhrMNs (57). However, a small proportion of sympathetic nerve fibers that originate from the cervical sympathetic chain also travel in the phrenic nerve (55). Direct application of any substance to the phrenic nerve therefore has the potential to target sympathetic neurons as well as diaphragm sensory afferents (e.g., Fig. 2). However, this is not necessarily a problem per se. This approach has been used specifically for the purpose of labeling diaphragm sensory afferent to enable description of spinal projections of these fibers (57). Additional challenges include the invasive nature of the surgical approach to access the phrenic nerve and also the impact of unilateral diaphragm denervation if phrenicotomy is used. Surgical skill is required for accessing and isolating the phrenic nerve. Section of the phrenic nerve will certainly alter the control of breathing; however, animals exhibit a remarkable compensatory response to unilateral diaphragm denervation (68). Breathing has been reported to be minimally impaired even after a bilateral phrenicotomy in rats (69).

DELIVERING SUBSTANCES TO THE DIAPHRAGM TO TARGET PhrMNs

Diaphragm application also provides a direct route of access to the phrenic motoneuron pool. Note that the substances do not need to be taken up by diaphragm myofibers for PhrMN targeting. Rather, diaphragm application methods are in actuality targeting uptake of the substance to the presynaptic innervation of diaphragm myofibers (e.g., at the neuromuscular junction or at some point along the efferent motor nerve). If this happens, the substance can potentially reach PhrMN soma via retrograde movement. Three general approaches have been used utilized: 1) direct application (i.e., “painting”) of the inferior surface (70); 2) direct intramuscular injection (7, 71), and 3) application of a viscous gel to the inferior surface (72). Table 1 provides a summary of the different agents that have been applied to the diaphragm for targeting PhrMNs.

Table 1.

PhrMN targeting via direct application of substances to the diaphragm

Substance	Species	Method/Volume	Result	Citation
CT-β	Rat	10 × 5-μL injection to each hemidiaphragm *Separate group received injections on 2 consecutive days.	Bilateral PhrMN labeling C3–C6; distal dendrites more prominent when injections made on consecutive days.	(73)
	Rat	35- to 40-μL injection to each side of the diaphragm	Bilateral PhrMN labeling, C3–C5	(74)
	Ferret	3 or 4 × 7- to 10-μL injections were made beneath the peritoneal lining of the diaphragm to include costal and crural regions.	Bilateral PhrMN labeling, C5–C7	(75)
	Mouse	10-μL topical application to each side of the diaphragm	Bilateral PhrMN labeling, C3–C6	(10)
DiAsp	Rat	8–10 × 20-μL injection left hemidiaphragm	Unilateral PhrMN labeling, midcervical	(59)
FluoroGold	Rat	7–10 × 7-μL injection to left hemidiaphragm	PhrMN labeling, C3–C5	(76)
	Rat	8–10 × 20-μL injections to left hemidiaphragm	PhrMN labeling, C3–C6	(59)
Fluoro-Ruby	Rat	8–10 × 20-μL injection left hemidiaphragm	Unilateral PhrMN labeling, C3–C6	(59)
HRP	Cat	30-μL injection	PhrMN labeling, C4–C6	(77)
	Rat	8–10 × 20-μL injection to left hemidiaphragm	Unilateral PhrMN labeling, C3–C6	(59)
PRV	Ferret	5–7 × 7- to 11-μL injections	Bilateral PhrMN labeling, C5–C7 (along with interneurons)	(75)
	Rat	40- to 50-μL topical application to left hemidiaphragm	Unilateral PhrMN labeling C3–C6; (interneurons labeled if PRV incubated >56 h)	(70)
	Mouse	20-μL topical application to right hemidiaphragm	Unilateral PhrMN labeling C3–C6 (along with interneurons)	(10)
Tetramethylrhodamine dextran	Rat	Multiple 5- to 10-μL injections to right hemidiaphragm	PhrMN labeling, C3–C6, robust dendritic labeling	(19)
“Tripartite nanoconjugate” of WGA-HRP, AuNP, and DPCPX	Rat	0.09- to 0.27-μg/kg injections to left hemidiaphragm	Bilateral PhrMN labeling, C3–C6	(78)
WGA-Alexa 488	Rat	5 × 10-μL injection to left hemidiaphragm	Bilateral PhrMN labeling, C3–C6	(62)
WGA-HRP	Rat	5 × 10-μL injection to left hemidiaphragm	Unilateral PhrMN labeling, C3–C6	(79)

Examples are provided that illustrate the range of different retrograde tracers that have produced successful targeting of phrenic motoneurons (PhrMNs). The examples provided used either diaphragm “painting” or intramuscular injection; see targeting phrmns via injection to the intrapleural space for commentary regarding the intrapleural delivery route. Not all publications report details of injection volume and location; such details are included when available. AAV, adeno-associated virus; AuNP, gold nanoparticle; CT-β, cholera toxin (β-subunit); DiAsp, 4-(4-(dihexadecylamino)stryryl)-N-methylpyridinium iodide; DPCPX, 1,3-dipropyl-8-cyclopentylxanthine; HRP, horseradish peroxidase; PRV, pseudorabies virus; WGA, wheat germ agglutinin.

Methods and Limitations of Targeting PhrMNs via the Diaphragm

The diaphragm muscle has a large surface area and has distinct areas including pars sternalis, pars costalis, and pars lumbalis (80). Furthermore, there is a somatotopic organization of the phrenic motoneuron pool (81), and distinct groups of PhrMNs will be labeled if tracers are selectively applied (58). Therefore, to target the entire PhrMN pool, the entire diaphragm should be covered (for topical application) or multiple intramuscular injections performed. In a rat model, delivering three 10-µL intramuscular injections of CT-β across the dorsal-to-ventral region of the diaphragm results in near-complete labeling of the phrenic motoneuron pool (24).

The general surgical approach is as follows. After a surgical plane of anesthesia is attained, the linea alba is incised and skin and muscles are retracted to expose diaphragm muscle. Either substances (Table 1) are topically applied over the abdominal side of the muscle surface or a small volume (∼10 µL) is injected intramuscularly. For topical application, the solution should be spread (“painted”) evenly over the surface of diaphragm (70). For intramuscular injection, the delivery sites should be spread across the dorsal-to-ventral surface of the diaphragm muscle as previously described (7). If done properly, intramuscular diaphragm injection does not cause any structural damage or contractile dysfunction in the diaphragm muscle (71).

Not surprisingly, a challenge of the diaphragm approach is the potential off-target labeling. An excellent demonstration and discussion of this was provided by Boulenguez and colleagues (59). They compared and contrasted the ability of several different traces to uniquely target PhrMNs after diaphragm delivery in rats. When FluoroGold or horseradish peroxidase was topically applied to the diaphragm, in addition to PhrMNs off-target labeling was prominent in the brain stem (area postrema, some motoneurons). This artifact presumably occurred because of these substances entering the bloodstream via diaphragm vasculature. However, other tracers including dextran amine Fluoro-Ruby and carbocyanine DiAsp [4-(4-(dihexadecylamino)styryl)-N-methylpyridinium iodide] produced apparently unique labeling of PhrMNs. The authors suggested that these molecules, because of lipophilicity and high molecular weight, respectively, were less able to enter the bloodstream. Thus, the choice of retrograde tracer is fundamentally important to targeting PhrMNs with minimal off-target labeling.

If using AAV for diaphragm application, the serotype is an important consideration. There are nine main AAV serotypes with >100 variants isolated from different species (82, 83), each with its own unique tissue tropism and immunogenicity (83). Some serotypes of AAV are particularly effective at retrograde movement to motoneurons. For example, AAV1, AAV2, AAV5 (84), AAV6 (85), and AAV9 (86, 87) all effectively drive motoneuron gene expression after delivery to skeletal muscle. Furthermore, a newly designed “retro AAV” variant, rAAV2-retro, has been developed to increase retrograde transport of the virus over existing AAV serotypes (88). Although originally developed for use in the brain, rAAV2-retro has been shown to be highly effective for lower motoneuron transduction after intramuscular injection (89). rAAV2-retro may be more effective at targeting PhrMNs, compared with other AAV variants, but to our knowledge this has not been evaluated.

TARGETING PhrMNs VIA INTRATHECAL DRUG DELIVERY

The size of many experimental or therapeutic molecules makes crossing of the blood-brain barrier (BBB) impossible. Moreover, even if a drug crosses the BBB, it may not exhibit a therapeutic effect because of poor distribution into the brain parenchyma (90). When the target is the spinal cord, intrathecal delivery of drugs can potentially overcome these obstacles. This route of drug administration is particularly useful when localized activation of spinal receptors or ion channels is necessary. In humans, the intrathecal space is commonly targeted to deliver analgesic drugs such as morphine, where they act on spinal neurons to provide analgesia (91). The phrenic motoneuron pool has successfully been targeted with intrathecal drug delivery in animal models, leading to a deeper understanding of the mechanisms associated with spinal respiratory neuroplasticity (92, 93). One strength of intrathecal delivery is that manipulation of spinal receptors can be done while phrenic nerve [or diaphragm electromyogram (EMG)] recordings are made. This allows real-time evaluation of how phrenic motor output is changing. This technique has been essential to unraveling the cellular mechanisms of phrenic long-term facilitation, an enhanced phrenic motor output after exposure to intermittent hypoxia (94). For example, intrathecal delivery of the broad-spectrum serotonin receptor antagonist methysergide can block the induction of long-term facilitation (92). Subsequent studies using intrathecal delivery have demonstrated a role for adenosine (93) and serotonin (95) receptor subtypes in long-term facilitation. The intrathecal delivery is a reliable technique for targeted drug delivery to the PhrMN pool, as evidenced by successful studies from independent laboratories (96–102).

Methods and Limitations of Targeting PhrMNs via Intrathecal Drug Delivery

The general approach for using intrathecal drug application to target the phrenic motoneuron pool is as follows. First, under a surgical plane of anesthesia, a laminectomy and a durotomy are performed over C2–C3. The tip of a polythene or silicone catheter (∼0.6-mm outer diameter) connected to a microliter syringe (e.g., Hamilton) is placed over C4. Chronic intrathecal delivery can be accomplished by subcutaneously implanting a mini-osmotic pump while securing the catheter to the skull (97, 103). For chronic intrathecal delivery, a caudal-to-rostral catheter placement approach with a catheter insertion point at T1 and catheter tip placement above C4 is likely to be more effective. With this approach, the catheter can be secured via sutures in the deep paraspinal muscles, along with securing the osmotic pump subcutaneously.

Drugs delivered to the intrathecal space are likely to reach PhrMNs via diffusion in a dorsal-to-ventral manner. Thus, the major limitation of the intrathecal approach is that it is impossible to restrict diffusion of drugs to PhrMNs or even specifically to the midcervical spinal cord. In studies focused on acute impact of the intrathecally delivered drug, an experiment to test whether drugs reach the brain stem is to simultaneously record hypoglossal and phrenic motor output. Since hypoglossal motoneurons are located in the medulla, their output provides some indication as to whether the drug reached the brain stem (104, 105). In general, off-target spread can be minimized if the volume injected is limited to 10 µL (106). Tilting the head of the animal slightly upward can help to ensure that the drug spread stays restricted to the spinal cord level.

INTRASPINAL MICROINJECTION TO TARGET PhrMNs

Intraspinal microinjection has been successfully used to target the PhrMN pool in many studies (5, 107–111). With this method, a small volume (i.e., nanoliters) of drug, tracer, or a virus is injected directly into the spinal parenchyma. Intraspinal microinjection has been particularly valuable in determining the membrane receptors that contribute to activation or inhibition of phrenic motor output. Chitravanshi and Sapru (108, 112) demonstrated that spinal glutamate and GABA receptors modulate phrenic motor output by microinjecting relevant agonists and antagonists into the C4 ventral horn while recording from the phrenic nerve. For example, when NMDA or AMPA agonists were microinjected into the region of the phrenic motor nucleus, it produced an increase in phrenic nerve output. Conversely, microinjection of antagonists such as 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide (NBQX) and d(−)-2-amino-7-phosphonoheptanoic acid (AP-7) reduced the amplitude of spontaneous phrenic nerve burst. A similar approach has been used for identification of GABAergic circuits contributing to phrenic output (5), and also to treat PhrMN pathology in lysosomal storage disease via microinjection of an AAV encoding a therapeutic protein (109).

The neuronal composition of the midcervical ventral horn merits discussion in the context of targeting the phrenic motoneuron pool with intraspinal microinjections. The cervical spinal “respiratory circuitry,” in addition to PhrMNs, includes interneurons distributed throughout the gray matter. Anatomical (70) and neurophysiological (113, 114) data confirm that propriospinal midcervical neurons can be synaptically coupled to PhrMNs. Interneurons that are monosynaptically linked to PhrMNs (i.e., “prephrenic interneurons”) are found in the cervical dorsal horn, lamina VII, and lamina X (10, 61, 115). Furthermore, a separate and rhythmically discharging group of interneurons is present in the rostral cervical spinal cord (116–118). There are also other motoneuron pools located in close proximity to PhrMNs, including scalene (C4–C8) and pectoralis major (C5–C8) (119). As discussed next, targeting of PhrMNs amidst the complexity of the cervical spinal circuitry requires surgical precision and validation of the injection site.

Methods and Limitations of PhrMN Targeting via Intraspinal Microinjection

Spinal microinjections are typically performed via a dorsal surgical approach, but a ventral approach is also possible (5, 110). A stereotaxic frame can be used to stabilize the vertebral bones and to avoid any damage due to respiratory movements (120). A fine glass capillary/micropipette (tip diameter 30–80 µm) (107) is used to deliver drugs while causing minimal damage to spinal tissue. After anesthesia, a longitudinal incision is made and muscles covering the vertebral column are removed. A dorsal laminectomy (C2 to C5) is performed while avoiding any damage to the underlying spinal cord. Once the spinal cord is exposed, a duratomy is performed over the intended site and the injector tip is gradually lowered to the spinal surface. Once the coordinates are set (1.0–1.2 mm lateral to midline), the tip can be lowered into the cord (1.5–1.8 mm dorsal) (107). This procedure requires penetration of the dura. Rapid lowering of the injector tip can help to reduce local pressure and damage on the cord (120). After infusion, the tip is left in place for 1–2 min before moving on to the next area or removal (121).

The prime advantage of the intraspinal microinjection technique is the relatively focal restriction of the drug to the region of interest (i.e., the phrenic motor nucleus). It is also possible to inject in the vicinity of PhrMNs and to record from the phrenic nerve simultaneously (108, 112). Considerable surgical expertise is necessary to ensure targeting of PhrMN and reproducibility across experiments. Small movements of the spinal cord associated with lung inflation can be problematic when positioning the tip of the micropipette. If the full rostral-caudal distribution of the phrenic nucleus is targeted, multiple microinjections will be necessary (107). Stereotaxic coordinates are of particular importance, and pilot experiments are necessary to ensure optimized coordinates when considering species, age, body weight, sex, etc. In the adult male rat, coordinates that have been successful are 1.0–1.2 mm laterally to the spinal cord midline and at a depth of 1–1.25 mm at the C4 spinal cord level (5). At the conclusion of experiments, histological verification of the pipette tip location is recommended. This can be accomplished via juxtacellular delivery of a tracer (e.g., neurobiotin) using iontophoresis into the recorded neurons (5).

INTRAVENOUS DRUG DELIVERY AND STUDIES OF PHRENIC MOTOR CONTROL

Intravenous drug delivery is widely used in studies of phrenic motor control. This approach obviously cannot directly target the phrenic motoneuron pool, but it has proven useful for establishing fundamental concepts regarding PhrMN control, particularly in regard to neuroplasticity. As one example, to determine whether the sustained increase in phrenic motor output occurring after repeated hypoxia exposure (i.e., phrenic long-term facilitation) is dependent upon serotonin receptor activation, the broad-spectrum serotonin receptor antagonist methysergide was intravenously administered. When the drug was given before exposure to intermittent hypoxia, phrenic long-term facilitation was blocked (122). This experiment confirmed a role of serotonin receptor activation in phrenic motor plasticity and set the stage for future studies that definitively identified the site of action on PhrMNs, first using intrathecal drug application (see targeting phrmns via intrathecal drug delivery) and then more specific retrograde targeting of PhrMNs (see targeting phrmns via injection to the intrapleural space, retrograde targeting of phrmns via the phrenic nerve, and delivering substances to the diaphragm to target phrmns) (52). Another example comes from studies of ampakines, which are positive allosteric modulators of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors. Initial studies used intravenous ampakine delivery to show that modulating AMPA currents increased PhrMN output (123). Subsequent work used spinal (intrathecal) application of ampakine to begin to localize the potential site of action to the midcervical spinal cord (102).

Methods and Limitations of Using Intravenous Delivery

For intravenous delivery to be effective (in the context of PhrMNs), the drug must be able to cross the blood-brain barrier. Veins typically used for drug delivery in rat neurophysiology studies are the femoral and tail veins (124, 125). An advantage of the intravenous delivery route is ease of the procedure and speed with which many drugs will reach the intended target, including easy access to the central nervous system. Intravenous dosing also allows pharmacokinetic (PK)/pharmacodynamic (PD) studies and in theory could help in dose extrapolation from animal to human studies, as it provides for body surface area (BSA) normalization (126). Another advantage is that neurophysiological recordings of PhrMNs, the phrenic nerve, or diaphragm EMG can all be done simultaneously with intravenous drug administration, thereby giving a real-time assessment of the impact of the drug. The limitations are clear: determining the specific site of action is not possible, and off-target effects are also likely. Although peripheral veins are readily visible, they have smaller diameters and care must be taken while injecting solutions to avoid vascular overload (injecting outside the vein or swiftly injecting too large a volume). Hence, some technical skills and training are required to reliably deliver intravenous injections.

TARGETING PhrMNs VIA INTRACELLULAR OR JUXTACELLULAR DELIVERY

Direct, intracellular delivery of substances to PhrMNs is technically challenging, but the “payoff” is exquisite cellular labeling and the ability to do comprehensive histological evaluation of cell morphology. PhrMNs have been intracellularly labeled in cats (127–131) and rats (132). Some of the first studies were done by Baumgarten and colleagues (127), who were able to describe PhrMN discharge patterns and histologically verify the location of the recording. Subseqent studies were able to leverage this approach to study morphology of PhrMNs in more detail. For example, Webber and Pleschka (128) recorded PhrMN discharge and then intracellularly injected procion yellow via the recording micropipette. This produced excellent cellular labeling and enabled a detailed description of the dendritic profiles of PhrMNs including branching patterns and surface area.

A related method that has been used in respiratory neurons (5, 133), although not to our knowledge in PhrMNs, is “juxtacellular recording” (134). This is a powerful technique that enables recording of action potentials by placing a recording electrode/micropipette adjacent to a neuron. After recordings, tracer molecules such as HRP or neurobiotin can be iontophoresed into the membrane of the neuron from which the electrophysiological recording was made. Thus, the method enables study of “structure-function” in individual neurons. The study by Marchenko et al. (5) showed the presence of GABAergic interneuron circuits that contribute to phrenic motor output. These interneurons were labeled juxtacellularly by iontophoresis of neurobiotin to later determine their relative position in the phrenic nucleus.

Methods and Limitations of PhrMN Targeting via Intracellular or Juxtacellular Delivery

Intracellular labeling has proved extremely beneficial in the study of morphological characteristics of the PhrMNs, and it remains the method of choice when detailed location and morphological parameters are under investigation. Recordings typically require that the subject is secured in a stereotaxic frame. The spinal cord is exposed via laminectomy at C2–C5. For intracellular recordings, a glass microelectrode is filled with a suitable dye (procion yellow, HRP, biocytin) usually mixed with an internal solution adjusted to physiological pH and osmolarity. The electrode is advanced with micromanipulators into the midcervical spinal cord. After confirmation of electrical discharge characteristics of PhrMNs (evoked field potentials upon stimulation of phrenic nerve or antidromic identification), a depolarizing current is injected to introduce the dye inside the cell. The tissue is later histologically processed to enable microscopy to evaluate PhrMNs. For the juxtacellular approach, a glass recording electrode is filled with a solution such as biotinamide. After a recording is made with the electrode adjacent to the neuron, anodal current is passed through the solution, which will eject positively charged molecules. This creates a “single-cell electroporation,” and biotinamide molecules can enter the soma. Subsequently, methods such as immunohistochemistry or in situ hybridization can be used to match electrophysiological properties with biochemical evaluation of individual neurons (135).

The primary limitations of these methods are the overall technical difficulty and the relatively low yield of labeled cells. Technical difficulties include lack of precise positioning of the microelectrode at the location of PhrMNs due to the small size of this nucleus. Holding the electrode in the recording position at the cervical cord could also prove difficult because of constant lung inflation. Some of the dye-filled neurons are often discarded from analysis in the instance of cellular injury (128) or inability to appropriately record and fill a cell, making it a low-throughput technique.

INTERSECTIONAL GENETICS: A POSSIBLE APPROACH FOR STUDYING THE PHRENIC MOTOR SYSTEM?

This is a rapidly evolving field that uses genetic and viral approaches to enable targeted gene expression in a highly specific population of cells. Intersectional genetic strategies typically use a “complementation strategy” in which multiple viral injections, or transgenic animals coupled with viral injections, are used to achieve localized gene expression in a specific group of neurons.

Viral vectors such as canine adenovirus (CAV) and AAV are effective in intersectional genetic approaches. For example, Cre recombinase can be expressed in a specific neuronal region via a microinjection of CAV2-Cre or AAV-Cre. A second, Cre-dependent virus can then be delivered to a region that contains known (or hypothesized) neuronal projections to the region expressing Cre recombinase. The second virus can be used to drive expression of fluorophores for tracing neuronal circuits or proteins to alter neuronal excitability (136, 137). Prominent examples of the latter include Designer Receptors Activated by Designer Drugs (DREADDs) (138) or light-activated proteins (e.g., Channelrhodopsin-2 and Halorhodopsin) (139). In recent years, these tools have been widely used to explore the function and composition of neuronal circuits by enabling selective activation or inhibition of target neurons. These approaches are particularly useful for targeting neuronal populations with well-defined and surgically accessible projections, such as the phrenic nerve. As an example, Penzo et al. (140) selectively inhibited neurons in the paraventricular nucleus of the thalamus (PVT) with projections to the lateral division of the central amygdala (CeL). To achieve this, CAV2-Cre was injected bilaterally into the CeL, followed by a PVT injection of AAV that conditionally expresses the hM4Di inhibitory DREADD receptor in the presence of Cre recombinase. This strategy was effective at limiting expression of the inhibitory (Gi) DREADD receptor to PVT neurons projecting to the CeL, allowing them to test the functional output of silencing these PVT neurons. Additional studies have successfully used a similar experimental strategy in the brain (141–146).

Intersectional genetics has enabled significant gains in the understanding of respiratory neural control (147–151). For example, Sherman et al. (150) were able to selectively activate or inhibit glycinergic neurons of the pre-Bötzinger complex (preBötC). To do this, they used transgenic mice that express Cre recombinase from the endogenous glycine transporter 2 locus (Glyt2-Cre) in combination with the Cre-dependent viral vectors AAV2/1-Ef1α-DIO-ChR2-eYFP and AAV2/1-flex-CBA-Arch-GFP, which were microinjected bilaterally into the preBötC. This strategy resulted in expression of ChR2 and Arch exclusively in glycinergic preBötC neurons, which allowed them to test the effects of activating or inhibiting this subpopulation of preBötC neurons on overall respiration (150). In addition, chemogenetic approaches in Cre mouse lines (i.e., using a single viral injection) have been successful in studying spinal respiratory interneuronal circuits. For example, Satkunendrarajah et al. (152) microinjected the Cre-dependent viral vector AAV-DIO-hM3Dq-mCherry into the cervical spinal cord to target midcervical Vglut-positive excitatory interneurons in transgenic Vglut-Cre mice. After high cervical spinal injury, they found that stimulation of midcervical excitatory interneurons via the excitatory DREADD hM3Dq restores respiratory motor function (152).

To our knowledge, intersectional genetic approaches have not yet been employed to directly drive gene expression (such as DREADD) in spinal respiratory motoneurons or interneurons. The anatomical organization of the phrenic motor system should make this approach feasible (Fig. 3). For example, a CAV2-Cre or AAV-Cre virus could be injected into the diaphragm or directly into the phrenic nerve. Retrograde transport of the Cre-expressing virus will drive expression of Cre-recombinase in phrenic afferent and efferent fibers. A second injection of a Cre-dependent AAV that drives expression of an optogenetic channel or DREADD would then be injected into the phrenic motoneuron pool. The overlap of the two viruses will result in selective expression of the optogenetic channel or DREADD in PhrMNs. Therefore, this technique has the potential to selectively modulate the excitability of the phrenic motoneuron pool. Although the general function of PhrMNs is well established, intersectional genetics may be useful for exploring the functions of subpopulations of PhrMNs (e.g., the influence of PhrMNs at C3 vs. C5 on breathing). Alternatively, in neuromuscular diseases such as ALS (153) or Pompe disease (154) where breathing is affected, there may be therapeutic or experimental value in modulating phrenic motor output or expressing therapeutic genes uniquely and specifically to the PhrMN pool.

Figure 3.

An “intersectional genetic” approach to selectively express channelrhodopsin-2 (ChR2) in phrenic motoneurons (PhrMNs). An injection of a canine adenoviral vector (CAV) encoding Cre-recombinase (Cre) directly into the phrenic nerve or intramuscularly in the diaphragm should result in Cre-recombinase expression in PhrMN soma via retrograde movement of CAV. A second injection of a Cre-dependent adeno-associated virus (AAV) encoding ChR2 into the ventral midcervical spinal cord will result in AAV uptake by PhrMNs. Since ChR2 expression is Cre dependent, ChR2 will only be expressed in cells transduced by the CAV-Cre vector. In this instance, the area of overlap will be the PhrMN pool, which will result in ChR2 expression.

SUMMARY AND CONCLUSIONS

PhrMNs are directly impacted in a variety of neuromuscular diseases and injuries. Prominent examples include spinal cord injury (155, 156) and amyotrophic lateral sclerosis (157), both of which can result in PhrMN death and/or dysfunction. Even the normal aging process is associated with PhrMN loss (158). However, only a small portion of the total phrenic motoneuron pool (∼20%) is needed to sustain quiet breathing. Even during intense exercise, <50% of the phrenic motoneuron pool is likely to be needed to sustain the ventilatory response (159). Thus, in disease states (or aging), preservation and/or restoration of function in even a small percentage of PhrMNs will have benefit. The location of the phrenic motoneuron pool, however, poses a challenge for targeting therapies specifically directed at PhrMNs. The same applies for studies designed not for therapeutic purposes but rather to further our understanding of PhrMN control in health and disease. This article has provided an overview of different methods that have been used to target the phrenic motoneuron pool with drugs, viral vectors, or genetic material (e.g., siRNA). These methods are summarized in Fig. 4 and Table 2. Relatively selective targeting of PhrMNs is possible with the use of molecules that are capable of retrograde movement along axons after diaphragm or phrenic nerve delivery. Retrograde methods can be used for robust labeling of PhrMNs [e.g., intrapleural delivery of CT-β (6)] or to deliver therapeutic molecules [e.g., AAV (39)]. Nonspecific approaches (e.g., intrathecal, intravenous, or intraspinal delivery) can also be effective therapeutically or for studying basic mechanisms of phrenic motor control and neuroplasticity. Prominent examples include spinal-directed AAV (109) or intrathecal delivery of trophic molecules (103). As the field of phrenic motor control moves forward, our hope is that these methods, along with new genetic approaches, can be leveraged to continue to advance our understanding of PhrMNs in health and disease.

Figure 4.

Schematic depicting different methods for targeting phrenic motoneurons (PhrMNs). A: intrapleural delivery of small volumes of agents such as cholera toxin β-subunit (CT-β) or adeno-associated virus (AAV) can produce robust phrenic motoneuron labeling/infection. Direct delivery to the diaphragm muscle is similarly effective but requires invasive surgery. B: direct delivery to the phrenic nerve can produce highly specific labeling of PhrMNs or phrenic sensory afferents. This approach may also be used in intersectional methods for targeted gene expression. C: intrathecal delivery enables spinal-directed targeting of drugs and has been particularly useful for studying mechanisms of phrenic motor plasticity. D: intraspinal injection enables relatively focal drug delivery or viral infection in and around the phrenic motoneuron pool. E: intravenous delivery enables screening of how drugs impact PhrMN output. F: intracellular recording and labeling enable a detailed study of PhrMN discharge and morphology. Juxtacellular recording is likely to be technically less challenging and also enables study of structure and function of individual cells.

Table 2.

Techniques for targeting the phrenic motoneuron pool with drugs, tracers, or genetic material

Technique	Examples of Substances Used Successfully	Primary Advantages	Challenges
Intrapleural	• AAV • CT-β • WGA • siRNA	• Simple surgical procedure • Can reach majority of PhrMN pool	• Off-target labeling • Potential for pneumothorax
Phrenic nerve targeting	• CT-β • Cascade Blue or fast blue • FluoroGold • Rhodamine • PRV • HRP	• Robust and unilateral PhrMN labeling is possible.	• Invasive surgical procedure • If phrenic nerve is cut: diaphragm paralysis and axonal degeneration • Labeling of sympathetic neurons and diaphragm sensory afferents
Diaphragm delivery	• CT-β • Fluoro-Ruby • FluoroGold • AAV • PRV • HRP • WGA-HRP	• Near-complete labeling of PhrMN pool is possible.	• Invasive surgical procedure • Off-target labeling is prominent with some tracers.
Intrathecal	• Serotonin receptor agonists/antagonists • Adenosine receptor agonists/antagonists • Ampakine	• Localized spinal receptor activation • Real-time evaluation of phrenic nerve or diaphragm EMG is possible. • Chronic delivery is possible.	• Invasive surgical procedure • Difficult to restrict drug actions to midcervical spinal cord • Impossible to selectively target PhrMNs
Intraspinal	• NBQX • AP-7 • AAV	• Delivery to immediate vicinity of PhrMNs • Real-time evaluation of phrenic nerve or diaphragm EMG is possible.	• Invasive and technically challenging surgical procedure • Multiple injections are needed to target entire PhrMN pool. • Histological verification is necessary.
Intravenous	• Serotonin receptor agonists/antagonists • Adenosine receptor agonists/antagonistsAmpakine	• Simple • Useful for screening the impact of drugs on the phrenic motor system • Real-time evaluation of phrenic nerve or diaphragm EMG is possible.	• Impossible to draw conclusions regarding the location of drug actions • Off-target effects are likely.
Intracellular	• Prussian blue • Procion yellow • HRP • Biocytin	• Excellent cellular labeling	• Technically challenging • A low-throughput technique

See the main text for methodological details with citations. AAV, adeno-associated virus; AP-7, d(−)-2-amino-7-phosphonoheptanoic acid; CT-β, cholera toxin (β-subunit); DiAsp: 4-(4-(dihexadecylamino)stryryl)-N-methylpyridinium iodide; EMG, electromyogram; HRP, horseradish peroxidase, NBQX: 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide; PhrMN, phrenic motoneuron; PRV, pseudorabies virus; siRNA, small interfering RNAs; WGA, wheat germ agglutinin.

GRANTS

This work was supported by Graduate School Funding Award from the University of Florida (E.S.B.), Craig H. Neilsen Foundation (S.R.), R01 HL153140-01 (D.D.F.), R01HL139708-01A1 (D.D.F.), and R01HD052682-11A1 (D.D.F.).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

P.P.T., S.R., E.S.B., and D.D.F. prepared figures; drafted manuscript; edited and revised manuscript; and approved final version of manuscript.