|Home | About | Journals | Submit | Contact Us | Français|
Numerous studies have demonstrated anatomical and functional neuroplasticity following spinal cord injury. One of the more notable examples is return of ipsilateral phrenic motoneuron and diaphragm activity which can be induced under terminal neurophysiological conditions after high cervical hemisection in the rat. More recently it has been shown that a protracted, spontaneous recovery also occurs in this model. While a candidate neural substrate has been identified for the former, the neuroanatomical basis underlying spontaneous recovery has not been explored. Demonstrations of spinal respiratory interneurons in other species suggest such cells may play a role; however, the presence of interneurons in the adult rat phrenic circuit – the primary animal model of respiratory plasticity – has not been extensively investigated. Emerging neuroanatomical and electrophysiological results raise the possibility of a more complex neural network underlying spontaneous recovery of phrenic function and compensatory respiratory neuroplasticity after C2 hemisection than has been previously considered.
Considerable experimental evidence now exists for post-SCI respiratory neuroplasticity. This potential for intrinsic recovery has been most extensively described in a rat model of high cervical (C2) spinal hemisection (HS) and is often referred to as the crossed-phrenic phenomenon (CPP) (Goshgarian 2003). Initial studies of this injury model showed that recovery of activity in the ipsilateral phrenic nerve and hemidiaphragm can be elicited by contralateral phrenicotomy within minutes to hours after injury. It has since been found that recovery of ipsilateral phrenic nerve and diaphragm activity also can occur spontaneously over the course of weeks (Fuller et al. 2003; Golder et al. 2005; Fuller et al. 2006; Vinit et al. 2007; Fuller et al. 2008) to months (Nantwi et al. 1999; Golder et al. 2001a) following C2HS. For present discussion purposes, this delayed natural recovery is operationally referred to as the “spontaneous crossed-phrenic phenomenon (sCPP)”. While the C2HS lesion model (in which both the CPP and sCPP can be expressed) is not typical of the majority of clinical cases of cervical SCI, it provides a valuable experimental setting for demonstrating proof-of-principle pertaining to respiratory recovery (reviewed in Lane et al. 2008a).
Often lacking in SCI research is detailed identification of neural substrates engaged in functional improvements occurring either naturally or in response to therapeutic interventions. Regarding the CPP and sCPP, the neural substrates, may be quite different as suggested by the striking differences in their expression. To date, the CPP has been attributed to activation of a normally silent crossed intraspinal pathway which can restore inspiratory drive to the ipsilateral phrenic nucleus (Goshgarian 2003). This pathway consists of axonal collaterals from uninjured, contralateral bulbospinal fibers that form monosynaptic projections onto phrenic motoneurons ipsilateral to the C2HS (Goshgarian et al. 1991; Moreno et al. 1992). This basic circuitry, however, does not take into account other neuronal constituents – interneurons in particular – that may affect the nature and extent of spontaneous recovery (i.e., the sCPP) in the chronically injured spinal cord.
Increasing recognition of the neuroplastic reserve of the injured spinal cord has evolved in recent years (Fouad et al. 2008; Harkema 2008; Stein 2008; Blesch et al. 2009; Darian-Smith 2009), and spinal interneurons may play prominent roles in neural circuit remodelling and modulation of motoneuron excitability. For example, one emerging concept is that interneurons with axonal projections traversing multiple spinal levels (i.e., long propriospinal interneurons) can provide alternative substrates capable of mediating locomotion following incomplete lesions of the spinal cord (Bareyre et al. 2004; Courtine et al. 2008). Short (inter- or intrasegmental) propriospinal interneurons, on the other hand, may play critical roles in regulating the expression of spontaneous or therapeutically-driven plasticity by influencing motoneuron activity or adopting novel functional tasks (Jankowska 2001). As once stated, the injured spinal cord represents a “new anatomy” (Dimitrijevic et al. 1997), and interneurons may be an integral part of the altered circuitry. Because spinal interneurons also appear to impact upon plasticity in other functional domains (Krassioukov et al. 2002; de Groat et al. 2006; Llewellyn-Smith et al. 2006; Rabchevsky 2006; Zinck et al. 2006; Hou et al. 2008a; Hou et al. 2008b), the question arises whether such cells may be involved in spinal circuits affecting the expression of protracted, spontaneous respiratory plasticity.
The diaphragm and the circuitry mediating its function (i.e., phrenic motor system) are key components of inspiration (Feldman 1986). Direct monosynaptic innervation of phrenic motoneurons by respiratory bulbospinal axons has been extensively documented in the rat (Ellenberger et al. 1988; Ellenberger et al. 1990). In cats and ferrets, however, both monosynaptic and polysynaptic pathways have been observed (Yates et al. 1999; Lois et al. 2009). Pre-phrenic interneurons (neurons with projections to phrenic motoneurons) have also been identified which appear to coordinate activity within phrenic and other circuits associated with accessory respiratory motoneuron pools (e.g. Aminoff et al. 1971; Sears 1990; Kirkwood et al. 1993). One suggested role for pre-phrenic interneurons is the inhibition of phrenic activity during expiration as shown in the cat (Douse et al. 1993). However, cervical interneurons seem to subserve heterogeneous functions (reviewed in Hilaire et al. 1997). In the cat, for example, interneurons in the lower, but not upper, cervical spinal cord appear to modulate respiratory function during ventilatory-related behaviors such as cough (Grelot et al. 1993; Miller et al. 1993).
Although electrophysiological evidence has suggested some spinal interneurons are associated with respiration in the adult rat (see below and Section 9), very little attention has been given to their neuroanatomical representation within the phrenic motor system. Even less is known about the function of these cells which may differ from that shown in other species that possess respiratory behaviors (e.g., cough, emesis) not present in rats. Furthermore, previous anatomical studies of the CPP in the rat have dismissed an interneuronal contribution to recovery of phrenic activity in that model (Goshgarian et al. 1991; Moreno et al. 1992). However, Lipski et al. (1993) reported evidence for inspiratory neurons in the upper cervical spinal cord of adult rat. While the majority of those cells appeared to have long descending projections extending towards intercostal motoneurons, some axon collaterals were seen directed toward phrenic motoneurons. These findings are consistent with studies in the cat that provided electrophysiological demonstration of “upper cervical interneurons” (e.g. Lipski et al. 1986; Duffin et al. 1987; Nakazono et al. 1994). Collectively, those earlier findings suggest the presence of at least one population of cells in the adult rat spinal cord that may have the capability to influence respiratory, and in particular phrenic, recovery after SCI (see Section 8).
Beyond the study by Lipski et al. (1993), the only other anatomical indication of interneuronal projections onto phrenic motoneurons in rat was shown by Dobbins and Feldman. (1994), who employed a pseudorabies virus (PRV) retrograde transneuronal tracing approach. Their investigation, which primarily focused on identification of medullary neurons projecting onto the phrenic motoneuron pool, revealed interneurons in Rexed laminae VII and X at the second cervical segment and laminae VII and IX at the level of the phrenic motor pool. Given the time-course and pattern of second- and third-order PRV labeling, however, these authors could not definitively determine whether those spinal cells were second-order to phrenic motoneurons (pre-phrenic) or ones with ascending projections to the medulla (i.e., third-order neurons or greater). As the present review shows, both populations exist in the adult rat and refining the use of neuroanatomical methods available enables distinction between them.
A re-evaluation of anatomical relationships between spinal interneurons and the phrenic nucleus of the rat was recently carried out in an effort to gain a greater understanding of the circuitry that may influence spontaneous recovery of phrenic motoneuron and ipsilateral diaphragm functions after C2HS. Our primary focus was on cervical interneurons projecting onto phrenic motoneurons. The aim of initial studies was to establish a baseline for the C2HS model by obtaining a more complete perspective of their numbers, distribution, and connectivity patterns in normal rats. These findings have been described in detail previously (Lane et al. 2008b) and are summarized below.
In the first set of experiments, PRV-Bartha was delivered to the left hemidiaphragm in normal adult female Sprague-Dawley rats. Phrenic motoneurons ipsilateral to the labeled diaphragm were infected with PRV 48 hours later, at which time no labeling was visible in any other spinal or supraspinal cell population. In addition, no PRV-positive cells were observed in the phrenic pool contralateral to the labeled hemidiaphragm. Approximately 64 hours following PRV delivery, and before labeling was visible in the brainstem, PRV-positive interneurons were observed throughout the cervical spinal cord below C2 ipsi- and contralateral to the labeled phrenic motoneurons (Fig. 1). Few PRV-positive interneurons were observed at C1-C2, suggesting limited connectivity within labeled phrenic motoneurons, as suggested by findings previously reported by Lipski et al. (1993). Combined transynaptic (using PRV) and monosynaptic (using cholera toxin beta subunit) retrograde tracing distinguished interneurons from motoneurons (Lane et al. 2008b). Although phrenic motoneurons were labeled with both tracers, cells beyond the defined column of phrenic motoneurons were labeled with PRV only. This finding confirms that cells only labeled with PRV were transynaptically infected and thus operationally considered second-order to phrenic motoneurons (i.e., pre-phrenic or spinal premotor interneurons).
While the majority of second-order cells were observed in laminae VII and X (111 ±24 ipsilateral to labeled phrenic pool; 48 ±14 contralateral), an additional population was seen in the right and left dorsal horns (38 ±15 ipsilateral; 6 ±3 contralateral). These findings suggest that a more substantial population of pre-phrenic interneurons is present in the rat spinal cord than previously appreciated. In addition, we observed another population of PRV-infected cells that usually appeared at ~72 hours post-PRV delivery. These cells were positioned immediately lateral and medial to the labeled phrenic motoneuron pool. Since these interneurons were infected later, they were not included in our original quantitative analyses. However, as discussed below, they may represent a population of cells with potentially important functional implications (referred to in Section 9).
Electrophysiological studies in the cat have revealed medullary inspiratory drive to interneurons in the cervical (Bellingham et al. 1990) and thoracic spinal cord (Kirkwood et al. 1988). There has been some additional evidence in cat and ferret that respiratory interneurons in the cervical spinal cord receive parallel inputs from the vestibular nuclei (Anker et al. 2006) and also from medullary centers that are known to regulate diaphragm function (Yates et al. 1999; Lois et al. 2009). As noted below (Section 8), some electrophysiological evidence has been reported for polysynaptic pathways from the ventral respiratory column (VRC) to phrenic motoneurons in the rat. Correlative neuroanatomical confirmation of these polysynaptic pathways was recently provided by our laboratory (Lane et al. 2008b).
Anterograde tracing studies of projections from VRC neurons on one side of the medulla showed that in addition to direct ipsilateral projections to phrenic motoneurons, some fibers crossed the spinal midline (Goshgarian et al. 1991) and formed terminal arborizations within the contralateral phrenic pool. In addition, decussating fibers branched immediately lateral to the central canal (Goshgarian et al. 1991).
The possibility that VRC fibers innervate spinal interneurons associated with phrenic motoneurons was previously dismissed on the basis of negative transynaptic wheat germ agglutinin-horseradish peroxidase (WGA-HRP) tracing results (Moreno et al. 1992). These results have been replicated and could reflect different WGA-HRP vs. PRV tracing properties (see discussion in Lane et al. 2008b). This also is reflected by the fact that WGA-HRP appears to label only a subset of medullary neurons (e.g. Moreno et al. 1992) compared to results using PRV or rabies virus (Dobbins et al. 1994; Yates et al. 1999; Lois et al. 2009; Rice et al. 2009).
In subsequent experiments, the VRC was electrophysiologically identified by recording spontaneous, rhythmic activity in the medulla in phase with diaphragm electromyogram recordings (Lane et al. 2008b). Discrete iontophoretic injection of Miniruby (fluorescently labeled biotin dextran amine, 10,000kDa) into the region of the VRC and delivery of PRV to the ipsilateral hemi-diaphragm revealed a vast number of BDA-labeled projections to PRV-labeled phrenic motoneurons (Figure 2). To a lesser degree, VRC projections also were observed in the region of the contralateral phrenic pool (Figure 2). Of particular interest, these tracing studies provided evidence for VRC projections to the soma and dendrites of PRV labeled, pre-phrenic interneurons around the central canal ipsi- and contralateral to the labeled phrenic pool (Lane et al. 2008b).
In addition to anterogradely labeled bulbospinal projections, delivery of Miniruby into the VRC resulted in some retrograde labeling. This revealed a separate population of neurons in the cervical spinal cord (Figure 2) with ascending projections to the region of the VRC (Lane et al. 2009). These cells were distributed bilaterally throughout the dorsal horn and medial spinal gray matter, predominantly in the upper cervical segments. Sixty-four hours following delivery of PRV to the left hemidiaphragm, no Miniruby-positive cells were infected with PRV. Therefore, it appears that the Miniruby-labeled neurons observed project to the medulla only and thus are not overlapping with the pre-phrenic interneurons described above. Previous work has supported the notion that some cervical interneurons may represent part of an ascending polysynaptic pathway mediating afferent input to supraspinal respiratory centers (Cleland et al. 1993). To our knowledge, this is the first anatomical indication of such spinal interneurons in the adult rat.
The next question addressed was whether any temporal remodeling of the phrenic circuitry occurs following C2HS as reflected by changes in the number or distribution of PRV labeled cells. PRV labeling via ipsilateral hemidiaphragm delivery was performed two weeks after injury based upon previous documentation of the initial emergence of the sCPP (Golder et al. 2005; Fuller et al. 2006; Fuller et al. 2008). As in spinal-intact rats, analysis of the number of PRV-positive interneurons relative to labeled phrenic motoneurons (ratio of secondary to primary cell labeling) was limited to 64 hours post-PRV delivery to the ipsilateral diaphragm. This method accounts for variation in labeling efficiency between animals (see Lane et al. 2008b for discussion).
Using this approach, the number of second-order neurons in contralateral laminae VII and X were significantly reduced, while the number and distributions of other pre-phrenic interneurons was unaltered (Lane et al. 2008b). More recent findings show that by 12 weeks following C2HS, the ratio of interneurons to motoneurons is significantly reduced in laminae VII and X both ipsi- and contralateral to the injury site (Lane et al. 2009). Meanwhile, the ratio of dorsal horn to phrenic motoneurons remained stable even 12 weeks following injury.
While a variety of studies have shown consistent evidence for spontaneous recovery of phrenic motoneuron excitability and breathing behavior after C2HS, the extent of such recovery remains modest for months post-injury (Fuller et al. 2008). Changes in laminae VII and X second-order cell labeling could indicate that reduced connectivity (Card et al. 1999) within the ipsilateral phrenic circuit may be involved, especially if some interneurons affected exert excitatory influences via a polysynaptic pathway.
Qualitative anterograde tracing observations in animals 2 weeks post-C2HS (Lane et al. 2009) have also revealed an apparent reduction in ipsi- and contralateral VRC projections to the ipsilateral phrenic motoneuron pool compared with what has been described in uninjured animals (compare Figures 2 and and4)4) (Lane et al. 2009). Qualitatively, VRC inputs to persisting pre-phrenic interneurons appeared to be predominantly from the contralateral side of the medulla, suggesting some of these interneurons may be part of the midline pathway typically considered to represent crossed monosynaptic projections (Goshgarian, 2003). These findings are consistent with previous retrograde tracing studies by Boulenguez et al. (2007) which showed that monosynaptic bulbospinal projections from both sides of the medulla to the region of the ipsilateral phrenic motor pool are dramatically reduced following C2HS. Together, these observations suggest that there is insufficient innervation of the affected phrenic circuit by bulbospinal axons that arise predominantly from the contralateral VRC.
Although studies of respiratory plasticity following SCI have focused on recovery of ipsilateral phrenic function following C2HS (i.e., “restorative” neuroplasticity), the functional improvement is limited (as noted earlier), and respiratory deficits persist (Fuller et al. 2008). Phrenic nerve recordings (Fuller et al. 2008) and diaphragm EMG data (Lane, Fuller, Reier – unpublished) show that activity ipsilateral to C2HS remains minimal even 12 weeks following injury. In contrast, compensatory activity in other spinal respiratory circuits can contribute to altered post-injury breathing behavior. It is well-known that following C2HS, animals adopt a rapid, shallow ventilatory pattern in order to maintain appropriate blood-gas levels (Goshgarian et al. 1986; Golder et al. 2001b; Fuller et al. 2008; Lane et al. 2008a). Activity in the contralateral phrenic motoneuron pool is increased following C2HS (Golder et al. 2003; Fuller et al. 2006) and most likely contributes to this altered pattern of breathing. This contralateral activity also may reflect an attempt to partially compensate for ipsilateral respiratory deficit as seen under other experimental conditions (Katagiri et al. 1994; Miyata et al. 1995; Prakash et al. 1999; Sandhu et al. 2009).
At present the direct effect compensation may have on the extent and onset of ipsilateral “restorative plasticity” associated with the sCPP remains poorly defined. While contralateral phrenic output is increased soon after injury, there is some evidence for a reduction in this output as ipsilateral phrenic activity spontaneously recovers (Golder et al. 2001a; Fuller et al. 2006). The altered pattern of activity suggests a close communication between right and left phrenic circuits that we recently demonstrated neuroanatomically in uninjured rats.
Using two variants of the PRV Bartha virus (PRV152 and 614, resulting in neuronal expression of green fluorescent protein and red fluorescent protein respectively) (Lane et al. 2008b), we found bilateral patterns of pre-phrenic interneuron labeling as previously observed with PRV 152 only but which now included a subset of neurons labeled with both PRV variants. This indicated the presence of a population of spinal interneurons that could integrate phrenic circuits on opposite sides of the spinal cord. These cells may partially provide an unrecognized anatomical basis for patterns of synchronized bursting activity between right and left phrenic circuits after C2HS (Sandhu et al. 2009).
We next hypothesized that some spinal premotor interneurons also may integrate with other neuronal networks, given the range of reflexes observed between phrenic, other respiratory (e.g. intercostal (Decima et al. 1967; Decima et al. 1969; Knill et al. 1976; Shannon 1980; Bellingham 1999), abdominal (Shannon 1980)), and non-respiratory circuits (e.g. postural, locomotor (Viala et al. 1979; Eldridge et al. 1981; Viala 1986)). Again, using dual-tracing studies with PRV152 and PRV614, we retrogradely labeled phrenic and intercostal motoneuron pools (Lane et al. 2008b). Tracing of the circuit innervating intercostal muscles revealed a population of second-order spinal cells with a distribution pattern similar to that of pre-phrenic interneurons. These studies also confirmed our hypothesis by revealing a population of interneurons that integrate ipsilateral phrenic and intercostal circuits by virtue of dual PRV labeling.
Given functional changes occurring contralateral to the injury, the question arises whether remodeling of the contralateral phrenic circuit is associated with compensatory phrenic motoneuron responses to C2HS. To address this issue, we recently initiated studies in which PRV was applied to the contralateral hemidiaphragm to transynaptically label the associated phrenic circuit following C2HS (Lane et al. 2009). This resulted in robust second-order infection of interneurons, similar in laminar distribution to that seen in uninjured animals. However, while the number of pre-phrenic interneurons at the C1-2 level is minimal in spinal-intact rats, interneurons associated with the contralateral phrenic motoneuron pool following C2HS were consistently observed at and above the C2 level of injury on both sides of the cord. These initial findings suggest possible recruitment of cells that normally are poorly, if at all, connected with the phrenic circuitry. If these cells represent the same neurons as identified by Lipski et al (1993) in the uninjured spinal cord, future anterograde tracing studies should demonstrate these cells also receive input from the medullary respiratory neurons as seen in rat (Lipski et al. 1994) and cat (Hoskin et al. 1987a; Hoskin et al. 1987b; Mateika et al. 1989). Accordingly, these interneurons would be anatomically positioned to not only mediate increased compensatory drive to the contralateral phrenic motoneuron pool, but also to intercostal or abdominal circuits after C2HS.
A few studies have reported cervical spinal interneurons in the adult rat which show respiratory-related discharge (Lipski et al. 1993; Hayashi et al. 2003), and our recent correlative neuroanatomical data (see above) provide a neural substrate basis for those findings. These and other observations in our laboratory and others, also offer an opportunity to speculate on potential roles of interneurons in post-C2HS and other forms of spinal respiratory plasticity.
Of particular interest is our evidence for putative VRC projections to pre-phrenic interneurons seen below C2. Those data are consistent with previously reported electrophysiological results (Ling et al. 1995; Tian et al. 1996). For example, Ling et al. (1995) showed electrical stimulation of the ventral funiculus on one side of the spinal cord could evoke compound action potentials in the contralateral phrenic nerve with both short (i.e. ~ 1.0 ms) and relatively long onset latencies (i.e. 5-7 ms). The latter were serotonin-dependent and suggested the presence of polysynaptic inputs to phrenic motoneurons (Ling et al. 1995).
Additional evidence was subsequently reported by Hayashi et al. (2003) who studied the contribution of spinal mechanisms to short-term potentiation (STP) of phrenic motoneuron activity. High-frequency stimulation of the lateral funiculus immediately below a complete transection of the C1 spinal cord elicited phrenic STP that peaked several seconds post-stimulation. One explanation for the response latency noted was that interneurons may be interposed between bulbospinal projections and phrenic motoneurons. This was supported by demonstration of extracellular recordings at the level of the phrenic nucleus which showed tonic and superimposed, phasic inspiratory activity that could be amplified by orthodromic spinal stimulation. No effect was obtained with supramaximal antidromic activation of the ipsilateral phrenic nerve thus suggesting these cells were not phrenic motoneurons (Hayashi et al. 2003).
Recent electrophysiological observations in our laboratory corroborate and extend the findings reported by Hayashi et al. (2003). As shown in Figure 5, we have recently obtained virtually identical extracellular recordings (compare Figure 10 in Hayashi et al., 2003) of cells at the level of the phrenic nucleus in non-spinalized rats which were not activated with antidromic stimulation of the ipsilateral phrenic nerve. Moreover, the tonic discharge pattern of this neuron during normoxic “baseline” conditions (Fig. 5i) is not observed in phrenic motoneurons under comparable recording conditions (Lee et al. 2009) providing further evidence that the recording is from a cervical interneuron. Interestingly, the activity of this neuron was enhanced during hypoxia (Figure 5). Therefore, some pre-phrenic interneurons may be stimulated under conditions of heightened inspiratory drive. This supports the suggestion made by Hayashi et al. (2003) that some spinal interneurons may be interposed between medullary centers and phrenic motoneurons serving as an amplifier to “increase the gain of synaptic input to motoneurons”. From an neuroanatomical perspective, however, it should be noted that our evidence for an interneuronal VRC-to-phrenic motoneuron relay is thus far limited to cells in Rexed laminae VII and X, whereas the electrophysiological data presented above have been obtained at the level of the phrenic motoneuron pool itself. As noted earlier (see Section 4) and shown in Figure 1, indications of second-order PRV-infected cells have been observed in that region. For technical reasons (Lane et al. 2008b), we have thus far chosen to be conservative in interpreting them as pre-phrenic interneurons. In view of the neurophysiological evidence cited, further study of these cells is warranted.
It is conceivable that the initial physiological stress associated with unilateral denervation of a phrenic motoneuron pool may result not only in activation of a crossed monosynaptic VRC pathway, but a polysynaptic circuit as well. Given the observed response of some interneurons to hypoxia, however, it may be possible these cells play a greater role in responses to chemical challenge after SCI. In that regard, the Hayashi et al (2003) data represent potential interneuronal contributions to phrenic motoneuron plasticity in an acute, high cervical transection model. This type of interneuronal relay may thus be involved in the CPP/sCPP, as well as other expressions of phrenic motoneuron plasticity including, but not limited to, long-term facilitation in response to intermittent hypoxia (Mitchell et al. 2001; Mitchell et al. 2003). To date, proposed pharmacological mechanisms underlying the effects of acute and chronic intermittent hypoxia have focused primarily on phrenic motoneuron, although potential spinal interneuron contributions have been considered (Golder et al. 2008).
A potentially significant contribution of interneurons to respiratory recovery in the C2HS model has recently been indicated by an intriguing study published by Alilain et al. (2008). Immediately after creating C2HS lesions, a Sindbus viral vector expressing Channelrhodopsin-2 was injected into the spinal cord in proximity to the ipsilateral phrenic motoneuron pool. Four days later, the transduced neurons, most of which were interneurons (Alilain and Silver, personal communication), were exposed to intermittent intervals of photostimulation which resulted in a long-lasting (i.e., beyond the stimulus) synchronous, bilateral diaphragm contraction in the absence of descending respiratory drive. Pharmacological studies revealed that the light-induced activation of phrenic activity in these experiments was NMDA receptor-dependent. Furthermore, some of the transduced neurons had axons that extended across the midline, as has been described for PRV labeled interneurons (Lane et al. 2008b).
In addition to the demonstration of pre-phrenic interneurons with bilateral projections to right and left phrenic circuits, the results of Alilain et al. (2008) raise speculation for the existence of a spinal oscillator that can impart respiratory rhythm. Although dominant respiratory rhythm generation resides at medullary levels (Feldman et al. 2006), some evidence for existence of a spinal respiratory pattern generator also has been reported (e.g. Palisses et al. 1987). However, the basis for this phenomenon has been considered equivocal in terms of the underlying circuitry and mechanisms (reviewed in Feldman (1986)). It is nevertheless tempting to consider that the presence of a spinal pattern generator-like circuitry that is not easily detected in the intact spinal cord may be revealed after injury.
Employing cross-correlation histogram analyses, Duffin and Alpen (1995) previously demonstrated in spinal-intact rats substantial bilateral synchronous excitation of right and left phrenic motoneuron pools via descending projections from VRC neurons. As discussed in detail elsewhere in this Special Issue (Sandhu et al. 2009), recent analyses of cross-correlation data from rats, 12-weeks post-C2HS, showed in some cases a delayed activation of ipsilateral versus contralateral phrenic motoneurons. While other interpretations must be considered (see discussion in Sandhu et al. 2009) the cross-correlation data suggest that the neural substrate associated with recovered motoneuron function may entail polysynaptic, as well as monosynaptic bulbospinal inputs.
It has been previously demonstrated that acute (Goshgarian 1981), as well as chronic (Fuller et al. 2002), dorsal rhizotomies contralateral to C2HS lesions can accelerate functional recovery in the ipsilateral phrenic circuit. This induced restoration was attributed to removal of phrenic afferent projections that innervate phrenic motoneurons on the opposite side of the cord (Goshgarian 1981). More recent pharmacological studies have shown that the disinhibition observed with contralateral dorsal rhizotomy is GABA dependent (Zimmer et al. 2007). This GABAergic inhibition of phrenic activity on the opposite side of the spinal cord was postulated to occur via dorsal horn neurons (Zimmer et al. 2007). Results from neuroanatomical tracing have identified candidate interneurons in the dorsal horn (with either ascending projection toward the VRC (Section 5) or direct projections onto phrenic motoneurons) that support those findings. The neurotransmitter phenotype(s) of those cells and their relationship to primary afferent input, however, remains to be determined. In that regard, further mapping of the phrenic motor circuit will provide greater definition for pharmacological/molecular manipulation of the sCPP.
Increasing attention to neuroplasticity in the injured spinal cord under experimental and clinical conditions has underscored a need for greater understanding of associated neural substrates. Such information can facilitate better definition of therapeutic targets and underlying mechanisms not only in relation to spontaneous recovery processes, but also therapeutically-driven repair. Thus far, the neural circuits mediating most forms of neuroplasticity in the injured spinal cord have proven challenging to identify. In contrast, previous anatomical studies of the CPP have elegantly described a comparatively simple anatomical substrate as a model for respiratory recovery in a partial spinal injury model. Crossed, monosynaptic VRC projections appear to play a key role in mediating recovery of ipsilateral phrenic motoneuron activity. However, the collective neuroanatomical findings described in this review now draw attention to the possibility that the underlying neural substratum associated with spontaneous recovery in the phrenic circuit may be more complex than previously envisioned.
As our transneuronal PRV findings have shown, a substantial contingent of bilaterally distributed interneurons is associated with phrenic motoneurons on each side of the spinal cord. More detailed anatomical-neurophysiological studies are required, but our anatomical findings to date are consistent with past and more recent evidence that suggest some spinal interneurons could be involved in mediating respiratory function in normal and spinal-lesioned rats. Furthermore, the apparent reduction in PRV labeling of some pre-phrenic cells after C2HS suggests a possible impact on the extent of functional recovery in the ipsilateral phrenic circuit.
Formulating hypotheses related to such potentially important neuronal dynamics would be difficult without the benefit of a neuroanatomical framework. One hypothesis suggested by data obtained thus far is that recovery of ipsilateral phrenic motoneuron activity after C2HS involves multiple, and not necessarily mutually exclusive, neural substrates. The induced CPP represents a situation in which the latent pathway can be activated in response to asphyxia following C2HS and contralateral phrenicotomy (Goshgarian, 2003). Under such circumstances, monosynaptic projections would represent the most effective route for enhanced excitatory drive to the ipsilateral phrenic motoneuron pool. On the other hand, in the case of spontaneous recovery, an early indication of restoration is low amplitude rhythmic phrenic motoneuron bursting activity. While this still could be mediated by a crossed, monosynaptic projection system, cross-correlation and Channelrhodopsin data reviewed above imply gradual recruitment of other neuronal populations and unmasking of an endogenous spinal respiratory pattern generator, respectively.
The degree to which a particular animal model permits extrapolations to the human may be critically dependent upon neuroanatomical similarities or differences (Blesch et al. 2008; Darian-Smith 2009). While aspects of respiratory function and neuroanatomy in the rodent are relatively unique (e.g. Duffin et al. 1995), the anatomical organization of spinal circuits in rat now appears more comparable to that described in other species (Yates et al. 1999; Billig et al. 2000; Lois et al. 2009). Our recent studies of the mouse phrenic circuitry (Qiu et al. 2009) demonstrate very similar features which are important given recent reports of apparent CPP-like plasticity in that animal model (Minor et al. 2006).
To conclude, we have provided clear anatomical evidence for spinal interneurons in the adult rat with projections to phrenic and intercostal motoneurons and ascending projections in proximity to the VRC. Such observations assume special significance since electrophysiological studies in a number of species have shown that spinal interneurons with segmental or spino-medullary projections can modulate normal respiratory activity. From a clinical perspective, discussion of respiratory control in the human and the role of polysynaptic drive to inspiratory motoneurons (Butler 2007; Butler et al. 2008) outline the need for a better understanding of the connectivity of spinal interneurons with these circuits. Since the rat is often the model of choice for studies of respiratory neuroplasticity, the cross-species similarities of spinal circuitry strengthens the value of data in the rat that could be subsequently translated to higher animal models of spinal cord plasticity.
Support for this work was provided by grants from the National Institutes of Health (NIH): NIH NS054025 (PJR) and NIH HD052682, NIH 1R01HD052682-01A1 (DDF).. Additional support was also provided by the Anne and Oscar Lackner Chair in Medicine (PJR) and the Craig H. Neilsen Foundation (MAL), and the University of Florida (KZL). PRV152 was were provided to us by L. W. Enquist, Princeton University as a service of the National Center for Experimental Neuroanatomy with Neurotropic Viruses (NCRR P40 RRO118604). PRV614 was generously donated by Dr Bruce Banfield at the University of Colorado. Special thanks go to Barbara O'Steen, Alex Jones and Forest Hunsaker for their excellent technical support.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.