INTRODUCTION

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY ASSOCIATED WITH THE... MORPHOLOGICAL PLASTICITY... MORPHOLOGICAL CHANGES IN THE... ASSOCIATION OF PHRENIC NUCLEUS... IS THE PLASTICITY ASSOCIATED... EFFECTS OF SEROTONIN ON... OTHER EXAMPLES OF PLASTICITY... BEHAVIORAL EFFECTS OF SPINAL... CLOSING REMARKS REFERENCES

TRAUMATIC INJURY TO THE CERVICAL spinal cord rostral to the level of the phrenic nucleus may interrupt the descending bulbospinal respiratory pathways and cause respiratory muscle paresis or paralysis. The resultant respiratory insufficiency is often treated in human cases of cervical spinal cord injury by placing the patient on long-term mechanical ventilator support. Such treatment is not always optimal and could lead to serious complications, such as infection, pneumonia, atelectasis, or even death (20, 45). Moreover, quadriplegic patients who are tethered to mechanical ventilators do not have the freedom to fully participate in physical and occupational rehabilitative strategies. It is important, therefore, to understand the underlying mechanisms related to plasticity and recovery of the respiratory pathways after spinal cord injury, since this information may lead to therapies directed toward resolving the problem of spinal cord injury-induced respiratory insufficiency in humans.

In 1895, Porter (107) provided one of the first demonstrations of functional restitution in the respiratory system after spinal cord injury. He paralyzed the ipsilateral hemidiaphragm by spinal cord hemisection rostral to the level of the phrenic nucleus in dogs and rabbits and showed that subsequent transection of the contralateral phrenic nerve immediately restored function to the previously paralyzed hemidiaphragm. This example of functional restitution was later termed the "crossed phrenic phenomenon" (CPP; Ref. 113). Subsequently, several laboratories confirmed the presence of the CPP in dogs (23, 79, 113), cats (23, 79, 111, 116), rabbits (16, 111-113), rats (50, 102), guinea pigs (44, 59), and even woodchucks (113). There was disagreement, however, in the earlier studies (before 1951) regarding the conditions and factors required to induce the CPP. The disagreement may have been due in part to the mechanical diaphragmatic recording techniques used in the early studies, which were not sufficiently sensitive to detect small movement and were inevitably affected by movements transmitted by the contralateral hemidiaphragm (79). The physiological basis for the CPP was not elucidated until 56 years after Porter's (107) investigation. In 1951, Lewis and Brookhart (79) used more sensitive diaphragm electromyographic recordings to show that the CPP was attenuated by artificial respiration and enhanced by conditions that resulted in an increase in central respiratory discharge (e.g., a cold block of the phrenic nerve contralateral to hemisection). The authors concluded "that the amount of crossed respiratory activity is directly proportional to the intensity of the central respiratory discharge and is entirely independent of the state of conduction in the direct phrenic nerve. The CPP occurs only when experimental conditions permit an augmentation of central respiratory discharge to take place as a secondary result of interruption of the direct phrenic nerve" (79).

	ANATOMIC SUBSTRATE FOR THE CPP

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY ASSOCIATED WITH THE... MORPHOLOGICAL PLASTICITY... MORPHOLOGICAL CHANGES IN THE... ASSOCIATION OF PHRENIC NUCLEUS... IS THE PLASTICITY ASSOCIATED... EFFECTS OF SEROTONIN ON... OTHER EXAMPLES OF PLASTICITY... BEHAVIORAL EFFECTS OF SPINAL... CLOSING REMARKS REFERENCES

Porter (107) demonstrated that bulbospinal respiratory impulses cross the spinal cord at the level of the phrenic nuclei during the induction of the CPP. He hypothesized that some phrenic motoneuron dendrites cross the spinal cord midline and receive uncrossed bulbospinal synaptic connections from the opposite side, thus suggesting an anatomic basis for the CPP. Cameron et al. (11), after injecting horseradish peroxidase (HRP) intracellularly into phrenic motoneurons of the adult cat, demonstrated phrenic dendrites that cross the spinal cord midline, but Furicchia and Goshgarian (46) did not identify any crossing phrenic dendrites in the adult rat after injecting choleratoxin-HRP into the diaphragm. However, Lindsay et al. (80) did find that a significant proportion of phrenic dendrites do cross the midline of the spinal cord in neonatal rats, and, more recently, Prakash et al. (109) determined that the percentage of phrenic dendrites crossing over to the contralateral spinal cord was markedly higher in 21-day-old rats than in adult rats (26.3 ± 2.0 vs. 3.1 ± 1.5%, respectively). These data suggest that contralaterally projecting rat phrenic dendrites may either retract back to their cell bodies of origin during postnatal maturation (109) or there may be a higher growth rate of motoneurons compared with the spinal cord (80). Thus there may be differences regarding the anatomic substrate over which the CPP is mediated in different species.

In an attempt to delineate the pathways over which the CPP is mediated in adult rats, Moreno et al. (92) injected wheat germ agglutinin conjugated to horseradish peroxidase (WGA-HRP) into functionally recovered hemidiaphragm muscle during the CPP. When WGA-HRP is injected into muscle or a peripheral nerve it is retrogradely transported back to the motoneuron cell bodies and also across synapses to other neurons within the synaptic chain (63, 64, 70, 108). Furthermore, the transynaptic transport is selectively mediated across physiologically active connections rather than silent or less active connections (63, 64, 73). The results indicated that the neurons that drive phrenic motoneurons in spinal-hemisected rats during the CPP are located bilaterally in the rostral ventral respiratory group (rVRG) of the medulla. No transneuronal labeling of propriospinal neurons was noted despite careful analysis of the spinal cord from C₁ to C₈ (92). The only spinal motoneurons labeled were in the phrenic nucleus ipsilateral to the hemisection and hemidiaphragm injection. Thus the results suggested that both crossed and uncrossed bulbospinal axon pathways from the rVRG collateralize to both left and right phrenic nuclei and that propriospinal neurons do not relay respiratory drive to phrenic motoneurons (92). A diagram of the bulbospinal pathways involved in the CPP is shown in Fig. 1. The arrows in the figure illustrate the pathway taken by respiratory impulses during the CPP.

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Surgical procedures and pathways involved in the crossed phrenic phenomenon (CPP). Inspiratory drive to phrenic motoneurons is mediated by medullary neurons in the rostral division of the medullary respiratory group (rVRG). These neurons project bilaterally to the phrenic nuclei. Moreno et al. (92) showed that both crossed and uncrossed descending respiratory pathways have spinal decussating collaterals that project to both phrenic nuclei (i.e., the crossed phrenic pathways). Hemisection rostral to the phrenic nucleus interrupts (dotted lines) the major bulbospinal drive to the ipsilateral phrenic nucleus, which results in paralysis of the left hemidiaphragm. Transection of the right phrenic nerve immediately after hemisection paralyzes the right hemidiaphragm and induces the CPP in most mammalian species. Arrows indicate the pathways followed by respiratory impulses during the CPP to restore function to the hemidiaphragm paralyzed by spinal cord injury. [Reprinted from Ref. 14 with permission by Academic Press.]

Before the Moreno et al. (92) study, it was known that respiratory rVRG axons descend via both crossed and uncrossed pathways to bilaterally innervate the phrenic nuclei (39, 40, 42) as depicted in Fig. 1. What was novel about the Moreno et al. study was the suggestion of spinal decussating rVRG axon collaterals, which also bilaterally innervate the phrenic nuclei (i.e., the "crossed phrenic pathways"). Subsequently, Goshgarian et al. (51) provided direct anatomic evidence for the decussation of bulbospinal respiratory axons at the level of the phrenic nuclei in adult rats that were injected with the anterograde neuronal tracer, Phaseolus vulgaris leucoagglutinin, into the rVRG. Thus, at least in the adult rat, respiratory impulses transmitted across the midline of the spinal cord during the CPP are most likely mediated by bulbospinal axons rather than phrenic dendrites as Porter (107) initially suggested. However, Prakash et al. (109) showed that a small proportion of spinal decussating phrenic dendrites also exist in adult rats; therefore, it is possible that respiratory impulses cross the midline over both anatomic substrates.

	PLASTICITY ASSOCIATED WITH THE CPP

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY ASSOCIATED WITH THE... MORPHOLOGICAL PLASTICITY... MORPHOLOGICAL CHANGES IN THE... ASSOCIATION OF PHRENIC NUCLEUS... IS THE PLASTICITY ASSOCIATED... EFFECTS OF SEROTONIN ON... OTHER EXAMPLES OF PLASTICITY... BEHAVIORAL EFFECTS OF SPINAL... CLOSING REMARKS REFERENCES

In most mammalian species, the CPP can be expressed within minutes after spinal hemisection by contralateral phrenicotomy and is the result of an increase in central respiratory discharge (79). Thus, according to the definitions offered by Mitchell and Johnson (90), the expression of the CPP actually does not represent "plasticity" but rather "modulation" of respiratory activity. Specifically, plasticity is defined as "a change in future system performance based on experience" and modulation is defined as "a neurochemically induced alteration in synaptic strength or cellular properties, adjusting or even transforming neural network function" (90). The term plasticity, as used by these authors, implies a long-lasting change, whereas modulation suggests a temporary change that is reversed once the modulating stimulus is removed. The fact that the expression of the CPP is an example of modulation is exemplified by the work of Aserinsky (1), who showed in cats that, when a local anesthetic was applied to the contralateral phrenic nerve after spinal hemisection, the CPP was induced, but, when the effects of the anesthetic block wore off, activity returned to the blocked hemidiaphragm and crossed phrenic activity in the hemidiaphragm ipsilateral to hemisection was attenuated. The same results were obtained when the experiment involving anesthetic block of the contralateral phrenic nerve was repeated in rats (50). Goshgarian (50) extended Aserinsky's finding by carrying out an additional study that showed that, even when the inducing stimulus is prolonged for several weeks by crushing the contralateral phrenic nerve, crossed phrenic activity disappears after the crushed phrenic axons regenerate and reestablish function in the denervated hemidiaphragm (50). Subsequent transection of the axons in the regenerated phrenic nerve once again induced the CPP (50).

There are aspects of the CPP, however, that do involve plasticity-related changes in the respiratory pathways. Although the CPP was demonstrated in several mammalian species (16, 23, 79, 113), it was not readily elicited in guinea pigs (59, 113). When spinal hemisection was followed within minutes by contralateral phrenicotomy, both sides of the diaphragm remained paralyzed and these animals always died of asphyxia. Guth (59) found that the CPP could be induced in guinea pigs if an interoperative interval of several months elapsed between the spinal hemisection and contralateral phrenicotomy; subsequently, Goshgarian and Guth (53) showed that the response could be induced if the interoperative interval was as short as 3.5 h. It was concluded that plasticity-related changes occurred within hours after spinal cord injury that converted "functionally ineffective" synapses in the crossed phrenic pathway to "functionally latent" connections, named so because they do not restore hemidiaphragm activity under normal conditions. The "functionally latent" synapses, however, become "functionally effective" after contralateral phrenicotomy (53). Thus specific plasticity-related alterations occurred in the respiratory neural circuitry within hours after spinal cord injury that changed the system response to the modulatory stimulus (i.e., asphyxia).

Functionally ineffective synapses were also demonstrated in the respiratory pathway of young (5-9 wk old) Osborn Mendel female rats (49). The synapses could be converted to functionally effective ones in those animals provided that 1-7 days elapsed between the spinal hemisection and contralateral phrenicotomy (49). This early study employed diaphragm electromyographic recording techniques in spontaneously breathing animals to document a yes/no presence of the CPP. Using a more sensitive quantitative electrophysiological recording technique from the phrenic nerve, O'Hara and Goshgarian (102) showed in Sprague-Dawley female rats that the time course for the conversion of functionally ineffective synapses was similar to that of guinea pigs. Specifically, the study showed a statistically significant increase in crossed phrenic activity at 2 h posthemisection compared with a weak expression of activity recorded at 30 min posthemisection. There were further smaller increases at 4 and 6 h posthemisection, but thereafter the same activity level persisted at 12 and 24 h posthemisection (102). Interestingly, it was discovered that, in older adult Osborn Mendel rats (6 mo old), the CPP could be induced without any delay between the spinal hemisection and the contralateral phrenicotomy (49). This suggested in this case that the functional conversion of ineffective synapses also occurred normally (i.e., without injury to the spinal cord) as the animals aged from young maturity to older maturity. Once again, with the use of more sensitive quantitative recording techniques from the phrenic nerve, the experiment was repeated on young (9-10 wk old) and older (9-10 mo old) female Spraque-Dawley rats (132). In both groups, the CPP was induced 30 min after spinal hemisection. The results indicated that there was almost a fourfold enhancement of crossed phrenic nerve activity generated in older rats compared with young rats. On the basis of this observation, it was concluded that as rats age from young maturity to older maturity, there is a gradual reorganization of the respiratory neural circuitry that enables older animals to more strongly express the CPP (132).

	MORPHOLOGICAL PLASTICITY ASSOCIATED WITH THE CONVERSION OF FUNCTIONALLY INEFFECTIVE SYNAPSES

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY ASSOCIATED WITH THE... MORPHOLOGICAL PLASTICITY... MORPHOLOGICAL CHANGES IN THE... ASSOCIATION OF PHRENIC NUCLEUS... IS THE PLASTICITY ASSOCIATED... EFFECTS OF SEROTONIN ON... OTHER EXAMPLES OF PLASTICITY... BEHAVIORAL EFFECTS OF SPINAL... CLOSING REMARKS REFERENCES

Although the above studies were the first to associate functionally ineffective synapses with the unmasking of a latent respiratory motor pathway that restores function to muscle paralyzed by spinal cord injury, functionally ineffective synapses had been demonstrated physiologically by other investigators on sensory neurons throughout the central nervous system (CNS). This type of sensory connection had been demonstrated in spinal cord (6, 24-28, 117, 125), visual system (9), somatosensory cortex (44, 126), thalamus (34, 110), trigeminal complex (33), and dorsal column nuclei (32, 88, 89). Although several physiological mechanisms underlying the rapid conversion of functionally ineffective synapses to a functionally effective state have been hypothesized (22, 29), the precise mechanism is still unknown.

An attempt was made to determine whether a morphological basis for the conversion of functionally ineffective synapses in the crossed phrenic pathway could be demonstrated at the light or electron microscopic level. Because there was little information available in the literature pertaining to the morphological organization of the phrenic nucleus in the rat when this work was initiated (78), a series of studies were published that detailed the anatomy of the phrenic nucleus at the light microscopic level (54), including the neurons that give rise to the accessory phrenic nerve (30) and the dendritic organization of phrenic motoneurons (46). In addition, a study of the origin and distribution of phrenic primary afferent nerve fibers (56) and a detailed electron microscopic analysis of the ultrastructure and synaptic architecture of phrenic motoneurons in the spinal cord of the adult rat (55) were completed.

After completion of the studies of the normal anatomy of the rat phrenic nucleus, a detailed quantitative morphometric analysis of the ultrastructure of the rat phrenic nucleus was conducted in both normal animals and in animals subjected to a C₂ spinal cord hemisection ipsilateral to the analyzed nucleus (58). Phrenic motoneurons were identified at the electron microscopic level by retrograde HRP labeling. On the basis of the early guinea pig studies, which indicated that functionally ineffective synapses were converted by 3.5 h posthemisection (53), the phrenic nucleus was analyzed in one group of rats at 4 h posthemisection. In addition to 4 h, analyses were also made in separate groups of rats at 1, 2, and 4 days after injury. Only the morphological characteristics of the phrenic nucleus pertinent to the conversion of functionally ineffective synapses will be discussed here.

Briefly, HRP labeling of phrenic motoneurons in noninjured rats revealed that the phrenic nucleus is a column of cells at the C₃-C₆ spinal levels (30, 58, 78). Although there are dendrites that radiate out into the spinal cord at almost right angles from the cell column (46), the majority of dendrites run rostrocaudally in the phrenic nucleus and surround phrenic motoneuron cell bodies (58). These longitudinally oriented dendrites are arranged in bundles that appear to be tightly fasciculated at both light microscopic and low-power electron microscopic levels. High-power electron microscopy of transverse sections through the phrenic nucleus, however, revealed that the longitudinally oriented dendrites lying in the phrenic neuropil adjacent to HRP-labeled cell bodies are not as tightly fasciculated as that suggested by the low-power observations. In fact, the membranes of adjacent dendrites are seldom in direct apposition. Intervening between adjacent dendrites are thin astroglial processes that separate the dendrites and isolate them from one another in the neuropil (Fig. 2A). Occasionally, direct dendrodendritic membrane apposition in the phrenic nucleus occurs, but the frequency of these appositions is low before spinal cord hemisection (58).

View larger version (40K): [in this window] [in a new window]   Fig. 2.   A series of schematic drawings illustrating the ultrastructural features of the normal rat phrenic nucleus (A) and the main changes of the phrenic neuropil that occur 4 h (B) and 4 days (C) after ipsilateral spinal cord hemisection at C2. A: normally, the longitudinally oriented phrenic dendrites (D1-D6, shown here in cross section) are not tightly fasciculated and for the most part are isolated from each other by astroglial processes (dotted areas). Occasionally, short dendrodendritic direct membrane appositions (e.g., between D1 and D2 and between D5 and D6) with punctum adhaerens (open arrow, top left) are seen. The majority of the synaptic terminals in the normal phrenic nucleus (T1-T4) form single synapses with postsynaptic dendritic profiles, but an occasional double synapse (between T5 and D5, D6) is also observed. N, here and in the other drawings, represents part of the nucleus of a phrenic motoneuron. B: by 4 h after hemisection, the mean length of previously existing dendrodendritic appositions increases (i.e., between D1 and D2 and D5 and D6). Such an increase in mean length of the appositions may result from active retraction of glial processes (exemplified by the direction indicated by the solid arrows). In addition, the number of double synapses increases significantly over normal levels by this time (e.g., between T1 and D1, D2 and between T4 and D5, D6). C: at 4 days posthemisection, the mean length of a dendrodendritic apposition reverts back to normal levels, but the percentage of appositions increases significantly over normal values. In addition, a slight decrease in the number of double synapses is accompanied by a corresponding increase in the number of triple and quadruple synapses (e.g., between T5 and D2, D3, D5, and D6). At this time, some degenerated terminals (DT representing T2 in the previous drawings) were incorporated into glial processes. [Reprinted from Ref. 58 with permission by Alan R. Liss.]

The other morphological feature pertinent to the conversion of functionally ineffective synapses relates to the synaptic architecture in the phrenic nucleus before spinal cord hemisection. Most of the axon terminals in the normal phrenic nucleus form single synapses with their postsynaptic targets. That is, each terminal establishes one or more synaptic active zones (the actual site of vesicle accumulation, membrane densities, and transmitter release), with only one postsynaptic profile in the same plane of section (Fig. 2A). Occasionally, a double synapse, i.e., a single terminal forming synaptic active zones with two postsynaptic profiles in the same plane of section, was found in the normal phrenic nucleus (Fig. 2A).

	MORPHOLOGICAL CHANGES IN THE PHRENIC NUCLEUS 4 H POSTHEMISECTION

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY ASSOCIATED WITH THE... MORPHOLOGICAL PLASTICITY... MORPHOLOGICAL CHANGES IN THE... ASSOCIATION OF PHRENIC NUCLEUS... IS THE PLASTICITY ASSOCIATED... EFFECTS OF SEROTONIN ON... OTHER EXAMPLES OF PLASTICITY... BEHAVIORAL EFFECTS OF SPINAL... CLOSING REMARKS REFERENCES

One of the first observations made within the phrenic nucleus was a statistically significant increase in the number of double synapses occurring by 4 h posthemisection (Fig. 2B). The double synapses consisted of clusters of spherical vesicles associated with asymmetrical synaptic membrane densities. Synapses with this characteristic morphology have been described as excitatory rather than inhibitory (105, 124). The morphological features of the double synapse did not change at later posthemisection survival periods.

Another morphological change noted in the phrenic nucleus at 4 h posthemisection was a statistically significant increase in the length of phrenic dendrodendritic membrane appositions (Fig. 2B). Although qualitative observations suggested that there may be a higher number of appositions in the phrenic neuropil at 4 h posthemisection, quantitative analysis revealed only a trend toward increasing numbers at 4 h. There was a statistically significant increase in the number of dendrodendritic appositions, however, by 2 days posthemisection (58). It was suggested that the increases in length and numbers of dendrodendritic appositions were due to the active retraction of astroglial processes normally intervening between adjacent dendrites in the phrenic nucleus (58).

The morphological changes seen in the phrenic nucleus at 4 h posthemisection were also observed at later posthemisection intervals. Thus the rapid changes seen in the phrenic nucleus within hours after spinal hemisection are not transient changes but can be observed at 1, 2, and 4 days posthemisection. Certain aspects of these changes, however, were modified at some of the later intervals (see Ref. 58 for details). Other changes, not observed at 4 h posthemisection, were detected at the later intervals. Specifically, in addition to double synapses, triple and quadruple synapses, defined as the synaptic active zones formed by a single presynaptic terminal with three and four postsynaptic profiles in the same plane of section, respectively, were also observed in the phrenic nucleus at the later posthemisection intervals (Fig. 2C). For quantitative purposes, double, triple, and quadruple synapses were combined into a single "multiple" synapse category (58).

Finally, in addition to the above-described rapid morphological changes, the well-known anterograde degenerative changes of the injured spinal cord were also observed. Beginning at 2 days posthemisection, electron-dense degenerating myelinated axons and terminals were observed with frequency in the phrenic nucleus (Fig. 2C).

Subsequent to the above quantitative morphometric study of the phrenic nucleus (58) and the early physiological studies (49, 50, 53), O'Hara and Goshgarian (102) used more sensitive quantitative recording methods from the phrenic nerve in rats and showed that there was an initial weak expression of the CPP that was significantly enhanced as early as 2 h posthemisection. Thus, in rats, the shortest time required for the conversion of functionally ineffective synapses to a more effective state was 2 h posthemisection. In light of this finding, a second morphometric study was conducted to determine whether ultrastructural changes could be detected in the rat phrenic nucleus as early as 2 h posthemisection (119).

The results of the second ultrastructural study indicated that, by 2 h posthemisection, there was a statistically significant increase in the mean percentage of phrenic dendrodendritic membrane appositions compared with normal controls. Thus astroglial processes retract from their normal intervening position between adjacent phrenic dendrites as early as 2 h after hemisection (119). Furthermore, although a significant increase in the number of double synapses could not be demonstrated, there was a documented significant increase in the length of the synaptic active zones in the terminals of the phrenic nucleus at 2 h posthemisection compared with controls (119). These observations led to the development of a hypothesis that related the morphological alterations observed in the phrenic nucleus within hours after spinal hemisection to the conversion of functionally ineffective synapses in the crossed phrenic pathway.

	ASSOCIATION OF PHRENIC NUCLEUS ULTRASTRUCTURAL CHANGES TO THE CONVERSION OF FUNCTIONALLY INEFFECTIVE SYNAPSES

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY ASSOCIATED WITH THE... MORPHOLOGICAL PLASTICITY... MORPHOLOGICAL CHANGES IN THE... ASSOCIATION OF PHRENIC NUCLEUS... IS THE PLASTICITY ASSOCIATED... EFFECTS OF SEROTONIN ON... OTHER EXAMPLES OF PLASTICITY... BEHAVIORAL EFFECTS OF SPINAL... CLOSING REMARKS REFERENCES

The increase in synaptic active zone length may have functional consequences, especially since other studies show that neural injury induces plasticity that is manifest as a change in synaptic morphology. (17, 69). Hillman and Chen (69) demonstrated change in the size of postsynaptic densities following lesion of parallel fiber afferents to Purkinje cells in the cerebellum. They found that the average length of postsynaptic densities greatly increased when reduction of afferents was between 50 and 70%. In their study, decreases in the number of afferent connections onto Purkinje cells caused the remaining sites to enlarge, thus suggesting constancy in total contact area. The authors suggested that the presynaptic area may enlarge to liberate sufficient transmitter to fully depolarize the neurons without having to alter the threshold level. In a more recent study, Chen and Hillman (17) demonstrated that a reduction of cortical input to the neostriatum generated a rapid and lasting increase in the average contact area of synaptic sites on principal cells of the striatum. This study further supported conservation of total synaptic contact area on target neurons, as demonstrated by the reciprocal relationship between synaptic size and number (17).

On the basis of the above, it is possible that lengthening of active zones in the phrenic nucleus after spinal cord hemisection could enhance the effectiveness of preexisting synapses, resulting in more stable and efficient connections. It is possible that lengthening of the synaptic active zones after spinal cord hemisection may play an important role in mediating the CPP and thereby restoring function to the previously paralyzed hemidiaphragm.

A significant increase in double and multiple synapses in the phrenic nucleus was not detected at 2 h posthemisection (119), but an increase in these specialized synapses was detected by 4 h posthemisection (58). Thus, between 2 and 4 h after hemisection are required for full differentiation of double and multiple synapses. It was hypothesized that the ongoing process of synaptic active zone elongation in terminals that remain in the phrenic nucleus after spinal cord hemisection is causally related to the documented increase in the number of double and multiple synapses. Figure 3 illustrates this relationship. Briefly, there may be several synaptic active zones contacting different postsynaptic targets in rVRG axon terminals in the normal phrenic nucleus. However, because the synaptic active zones are relatively short, a random section (Fig. 3, dotted lines) through the terminal may reveal only one of the active zones, thus identifying the terminal as a single synapse. After hemisection interrupts the main ipsilateral bulbospinal pathways, synaptic active zones in terminals of the crossed phrenic pathway elongate. Thereafter, random sections are more likely to reveal terminals with synaptic active zones contacting more than one postsynaptic target (i.e., a "multiple" synapse) because the zones are longer. The above hypothesis was strengthened by two additional ultrastructural studies (52, 121). In the first (52), phrenic motoneurons ipsilateral to a C₂ spinal hemisection were retrogradely labeled with HRP, whereas respiratory bulbospinal terminals in the phrenic nucleus were anterogradely labeled after the injection of a mixture of WGA-HRP and HRP into the rVRG. This study provided direct anatomic evidence that bulbospinal respiratory neurons from the rVRG provide a source of double synapse formation in the phrenic nucleus after C₂ hemisection. In the second, ultrastructural study (121), terminals in the phrenic nucleus were immunochemically stained for glutamate and GABA, the main excitatory and inhibitory neurotransmitters projecting to the phrenic nucleus. It has been shown that glutamate is the excitatory amino acid that depolarizes phrenic motoneurons during inspiration (84, 86), whereas the termination of the inspiratory burst and the subsequent quiescent expiratory phase of respiration are mediated by a disfacilitation of phrenic motoneurons and their active inhibition by GABA (8, 41). The study indicated that there were fewer glutamate and GABA-labeled terminals in the phrenic nucleus 30 days after C₂ hemisection compared with controls (121). However, among the terminals that remained, the average number of active zones per terminal was greater in the hemisection group than in the control group, with the active zones in the glutamate terminals mainly accounting for the difference. Moreover, the length of the active zones in the glutamate terminals was significantly longer in the hemisection group compared with that in controls. Interestingly, the length of the active zones in the GABA terminals was also found to be longer in the hemisection group compared with that in controls (121). These data taken together support the notion that respiratory bulbospinal synaptic input to the phrenic nucleus lost by spinal cord injury is compensated for in part by an increase in contact area of the remaining synaptic input.

View larger version (94K): [in this window] [in a new window]   Fig. 3.   Diagram suggesting a role of synaptic active zone enlargement leading to a greater number of multiple synapses in the phrenic nucleus neuropil. A and B represent a bundle of dendrites in the phrenic neuropil and their contact with synaptic terminals from rVRG neurons in the medulla. A: characteristics of the synaptic morphology of a control (nonhemisected) animal. B: modifications occurring in synaptic active zones in animals subjected to hemisection. The blue ovals correspond to the synaptic active zones in rVRG terminals. The terminals are depicted at the end of axons descending either ipsilaterally (the 2 axons projecting down from the top of each panel) to the phrenic dendrites or contralaterally from axons of the crossed phrenic pathway (the 2 axons projecting to the bundle from the left of each panel). Random transverse sections through the dendrite bundle (dotted lines at S1 and S2) are depicted diagrammatically to the right of each panel. Before hemisection (A) the synaptic active zones (blue ovals) in each terminal are relatively short. Thus random sections through the dendrite reveal a predominance of single synapses (S1 and S2 in A). Within hours after hemisection (B), there is an elongation of synaptic active zones in the crossed phrenic pathway. Because of the synaptic active zone enlargement, random transverse sections through the dendrite bundle now reveal a predominance of multiple synapses (S1 and S2 in B).

There may also be a functional significance to the astroglial process retraction observed in the phrenic nucleus after hemisection. It has been suggested that glial process retraction from in between adjacent neurons in other CNS centers, such as the supraoptic and paraventricular nuclei of the hypothalamus, would result in a rise in the extracellular potassium concentration because the glial processes would no longer be available to absorb the potassium that is released from the physiologically active cells (66-68). Elevated extracellular potassium would increase neuronal membrane excitability by partial depolarization (66, 67). In addition, the increase in neural membrane apposition resulting from glial retraction may create the development of electrical field effects around groups of neurons. The result of these field effects would be to synchronize the firing of cells, especially if electrical excitability were already elevated by depolarizing extracellular ion concentrations (66).

Enhanced phrenic motoneuron excitability, augmentation of crossed phrenic presynaptic input, and synchronization of phrenic motoneuron output are all manifestations that would predispose to the conversion of functionally ineffective synapses in the crossed phrenic pathway after spinal cord injury. It is possible that functionally ineffective synaptic terminations from the crossed phrenic pathway may be physiologically normal but quantitatively insufficient to fully depolarize phrenic motoneurons in spinal-hemisected young adult rats. Thus, when the animals are subjected to contralateral phrenicotomy within minutes after spinal cord hemisection, the summated action potentials of the crossed phrenic input may not be able to fire the cells. Within hours after spinal cord injury, however, astroglial process retraction and increases in phrenic membrane apposition could result in enhanced excitability of phrenic motoneurons. The postsynaptic changes resulting in enhanced phrenic motoneuron excitability coupled with the rapid presynaptic augmentation of input mediated by synaptic active zone elongation and the associated double and multiple synapse formation could be the mechanism enabling previously ineffective synapses of the crossed phrenic pathway to activate the phrenic motoneurons (58, 119).

	IS THE PLASTICITY ASSOCIATED WITH THE CONVERSION OF FUNCTIONALLY INEFFECTIVE SYNAPSES INJURY INDUCED OR ACTIVITY DEPENDENT?

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY ASSOCIATED WITH THE... MORPHOLOGICAL PLASTICITY... MORPHOLOGICAL CHANGES IN THE... ASSOCIATION OF PHRENIC NUCLEUS... IS THE PLASTICITY ASSOCIATED... EFFECTS OF SEROTONIN ON... OTHER EXAMPLES OF PLASTICITY... BEHAVIORAL EFFECTS OF SPINAL... CLOSING REMARKS REFERENCES

Although very specific, rapidly occurring morphological alterations were documented in the phrenic nucleus after spinal cord hemisection, it was not clear whether the observed plasticity was the result of the complex multifactoral events associated with the physical trauma to the spinal cord (i.e., edema, ischemia, local tissue hypoxia, and so forth) or whether it was the result of the more specific effect of interruption of the descending respiratory drive onto phrenic motoneurons. To address this issue, the descending bulbospinal respiratory drive to the phrenic nucleus was blocked for 4 h by cold application at the C₂ level of the spinal cord while avoiding major damage to the cord (13, 14). Complete block of axon transmission of the respiratory pathways running unilaterally in the ventral as well as in the lateral funiculus was achieved by approximation of a cold probe to the ventral surface of the spinal cord. The spinal cord surface temperature was lowered to 7°C and was maintained by a cold recirculated alcohol system. The efficacy of the reversible block was assessed by continuous bilateral electromyogram (EMG) activity recorded from the hemidiaphragms ipsilateral and contralateral to the cold application. Sham-operated control animals were employed to confirm that the surgical exposure of the cord and/or the chronic placement of the probe and the administration of intravenous dopamine given to maintain stable blood pressure did not affect respiration. An example of the results of the reversible hemispinalization of the rat spinal cord by a cooling device is shown in Fig. 4. No significant change occurred in EMG hemidiaphragmatic activity in any of the sham-operated control animals. The descending pathway from the rVRG to the phrenic nucleus was completely and continuously blocked for 4 h in all the experimental animals as demonstrated by abolition of the EMG hemidiaphragmatic signal ipsilateral to cold block (Fig. 4). In all experimental animals, hemidiaphragmatic activity returned when the cold block was removed, demonstrating the reversibility of the block and the fact that no significant damage to the spinal cord occurred at the site of the cold-block application (13). The recovered EMG activity was significantly greater than the preblock values (Fig. 4). This suggested that persistent plasticity-related changes occurred in the ipsilateral respiratory neural circuitry within 4 h of cold block. Interestingly, EMG activity contralateral to the block did not change significantly from preblock values after the block was removed but was significantly enhanced during cold block (Fig. 4). The temporary enhancement of activity contralateral to cold block most likely reflects an increase in central respiratory discharge during cold block as expected when one-half of the diaphragm becomes paralyzed. In addition, there is an increase in both burst frequency and burst duration in the right hemidiaphragm during cold block, further suggesting that cold block had altered respiratory drive (Fig. 4).

View larger version (25K): [in this window] [in a new window]   Fig. 4.   Electromyogram (EMG) recordings from the right (R) and left (L) hemidiaphragm of a control and experimental animal subjected to 4 h of cold block of the bulbospinal respiratory drive at C2 on the left. A: before cold block. B: 1 h after cold block was initiated. C: 4 h after the initiation of cold block and 1 min after the cold probe was removed from the surface of the spinal cord. No major change in EMG activity is observed in the control animal. In B, the right EMG signal contralateral to cold block is enhanced, whereas the left EMG response is absent. After the cold probe was removed (C), activity in the previously quiescent left hemidiaphragm is enhanced, but the enhanced activity on the right seen in B returns to baseline levels (compare with A). (Reprinted from Ref. 13 with permission by Academic Press.)

Immediately after the 4 h application of the probe at the C₂ level of the spinal cord in the experimental and sham-operated animals, the phrenic nucleus ipsilateral to probe placement was prepared for quantitative morphometric analysis using the same techniques employed in the C₂ hemisection studies (58, 119). The following morphological alterations were documented in cold-block animals compared with controls: 1) a significant increase in the number of multiple synapses, 2) a significant increase in the number of dendrodendritic appositions, and 3) a significant increase in the length of symmetric and asymmetric synaptic active zones. The above changes are similar to the changes induced in the phrenic nucleus following C₂ hemisection. It was concluded, therefore, that injury to the spinal cord is not a requirement for the morphological plasticity thought to underlie the conversion of functionally ineffective synapses in the crossed phrenic pathway. Rather, the induced changes are activity dependent and are likely caused by the interruption of the descending bulbospinal respiratory drive to the phrenic nucleus (14).

One additional experiment was conducted to verify the above conclusion. Both hemisection and hemispinalization by cold block at the C₂ spinal level cause not only a loss of the ipsilateral descending respiratory drive, (i.e., a "functional deafferentation" of the ipsilateral phrenic nucleus; Ref. 14) but also an increase in the remaining contralateral descending respiratory drive. The contralateral respiratory pathways connect with phrenic motoneurons ipsilateral to cold block or hemisection by decussating collateral axons that cross the spinal cord midline below the hemisection/cold-block site. Thus the observed phrenic nucleus plasticity after hemisection/cold block could be caused by a loss of the ipsilateral descending drive, an enhancement of the contralateral respiratory drive, or both. To differentiate between these two possible inducers of the plasticity, both the glial and synaptic cytoarchitectures of the phrenic nucleus were assessed morphometrically in nonoperated rats exposed to 48 h of hypoxia in an atmosphere chamber. The hypoxia exposure produces an increased descending respiratory drive without functional deafferentiation. The 48-h time interval was based on the work of Hatton (66). He noted morphological alterations in the supraoptic and paraventricular hypothalamic nuclei of noninjured rats 48 h after dehydration similar to those observed after C₂ hemisection (58, 119). Phrenic nucleus morphometric analysis revealed that there were alterations in the astroglial cytoarchitecture similar to those induced by hemisection and cold block, but there were no alterations of the synaptic cytoarchitecture (14, 57). From these results, it was concluded that the enhancement of descending respiratory drive by hypoxia will induce astroglial process retraction, but the retraction does not necessarily lead to synaptic remodeling in the phrenic nucleus (57). Additionally, it was concluded that no synaptic plasticity occurs in the phrenic nucleus without functional deafferentation, despite an increase in descending respiratory drive. Therefore, functional deafferentation or, in other words, a loss of ipsilateral descending respiratory drive may be the primary inducer of phrenic nucleus synaptic plasticity occurring after hemisection or cold block (14).

	EFFECTS OF SEROTONIN ON THE PLASTICITY ASSOCIATED WITH THE CONVERSION OF FUNCTIONALLY INEFFECTIVE SYNAPSES

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY ASSOCIATED WITH THE... MORPHOLOGICAL PLASTICITY... MORPHOLOGICAL CHANGES IN THE... ASSOCIATION OF PHRENIC NUCLEUS... IS THE PLASTICITY ASSOCIATED... EFFECTS OF SEROTONIN ON... OTHER EXAMPLES OF PLASTICITY... BEHAVIORAL EFFECTS OF SPINAL... CLOSING REMARKS REFERENCES

It is generally agreed that 5-hydroxytryptamine (5-HT, serotonin) is a neuromodulator that has an excitatory effect on spinal motoneurons, including those in the phrenic nucleus (5, 81, 91, 127, 128). However, there have been studies that have shown that 5-HT may have inhibitory effects on phrenic motoneurons (31). Moreover, there is a disagreement regarding the mechanism of 5-HT-enhanced excitability of spinal motoneurons. Some studies suggest that 5-HT directly stimulates the spinal motoneurons by depolarizing membrane potential beyond threshold via a decrease in potassium conductance (81, 123). Other studies have shown that the application of 5-HT alone does not evoke spinal motoneuron action potentials (128). Nevertheless, excitatory effects on spinal motoneurons mediated by glutamate or ventral/dorsal root stimulation are augmented by iontophoretic application of 5-HT, whereas systemic injection of the 5-HT antagonist, metergoline, reduces the ability of 5-HT to enhance glutamate-evoked motoneuron activity (127, 128). These results suggest that 5-HT may act as a neuromodulator that decreases the threshold for glutamate-evoked action potentials and thus enhances the effects of excitatory inputs to motoneurons. Indeed, our laboratory as well as others have demonstrated the neuromodulatory effects of 5-HT on either revealing subthreshold ineffective respiratory pathways or enhancing the expression of the CPP (82, 134-136). Because these studies focus more on the excitatory neuromodulatory role of 5-HT and do not directly implicate 5-HT in the plasticity of the respiratory pathways after spinal cord injury, they will not be discussed here. There are three other studies, however, that do suggest two different mechanisms involving 5-HT directly with the conversion of functionally ineffective synapses in the crossed phrenic pathway (61, 62, 122). The first (122) suggests that plasticity of 5-HT synaptic terminals after spinal cord injury may be part of the morphological substrate leading to the functional conversion of these specialized synapses in the respiratory pathways, whereas the other two (61, 62) suggest that normal levels of serotonin must be present in the spinal cord to initiate the plasticity underlying the conversion of ineffective synapses.

It was hypothesized that other neurotransmitters in the phrenic nucleus such as 5-HT may help to compensate for the loss of glutamatergic and GABAergic terminals after spinal cord injury and thus contribute to the conversion of functionally ineffective synapses in the crossed phrenic pathway (122). The conversion of these synapses then leads to the recovery of the hemidiaphragm paralyzed by an ipsilateral C₂ hemisection. The above hypothesis is particularly salient because 5-HT has been implicated in playing an important role in the process of functional recovery of other motor and sensory systems after spinal cord injury (65, 114, 133).

In light of the above, an electron microscopic morphometric immunochemical analysis of 5-HT-labeled terminals in the phrenic nucleus was conducted before and 30 days after an ipsilateral C₂ spinal cord hemisection to document injury-induced synaptic architectural changes involving this neurotransmitter (122). The results indicated that, although the total number of all types of terminals in the phrenic nucleus was significantly less 30 days after hemisection compared with nonhemisected controls, the number of 5-HT-immunoreactive terminals in the hemisected group well exceeded that of the control group. Although 5-HT-immunoreactive terminals contained both symmetrical (presumably inhibitory) and asymmetrical (presumably excitatory) synaptic membrane densities, the 5-HT terminals with asymmetrical synapses mainly accounted for the difference between the two groups. Similarly, the mean number of asymmetrical synaptic active zones in all 5-HT terminals increased significantly 30 days after hemisection over the control level. Furthermore, the mean length of both asymmetrical and symmetrical synaptic active zones in 5-HT terminals was significantly longer in the hemisection group compared with that in controls. Finally, the total number of 5-HT terminals with multiple synapses was significantly greater at the 30-day postlesion survival time than the value observed in control animals (122).

The above data indicate that serotonergic axons and terminals possess a remarkable capacity for plasticity after spinal cord injury that results within 30 days after C₂ hemisection in a greater than 100% replacement of the serotonergic projections lost by the cord injury (122).

Spinal cord injury, however, is not the only stimulus that can induce plasticity of serotonergic projections to the phrenic nucleus. Cervical dorsal rhizotomy (CDR) also enhances serotonergic innervation of phrenic motoneurons and serotonin-dependent long-term facilitation of respiratory motor output in rats (76). Kinkead et al. (76) showed that the number of serotonin-immunoreactive terminals within 5 µm of labeled phrenic somata and primary dendrites increased more than twofold ~1 mo after CDR vs. that shown after sham operation. Moreover, the authors showed that CDR enhances serotonin-dependent long-term facilitation of phrenic motor output following periods of intermittent isocapnic hypoxia (76).

It has been shown that administration of 5-HT in cultured sensory neurons or in excised pleural ganglion cell clusters of Aplysia triggers the adenylate cyclase cascade by acting on G-protein-coupled 5-HT receptors. The activation of the receptors causes a significant elevation of the second-messenger cAMP concentration in the presynaptic terminal of the neuron (4). The elevated level of cAMP in the terminals subsequently activates protein kinase A, which phosphorylates and closes potassium channels. The closing of potassium channels blocks potassium influx into the presynaptic terminals and thus elevates the extracellular potassium concentration, which increases the excitability of the terminal (21, 77, 104, 120). The stimulated terminals would thus increase the amount of neurotransmitter released in the synaptic cleft and thereby enhance the postsynaptic receptor response. It is possible that the enhanced 5-HT innervation documented in the phrenic nucleus after spinal hemisection (122) may be related to an enhanced release of glutamate from surviving presynaptic terminals originating from neurons in the medullary respiratory centers. If this were true, this would explain in part the mechanism for the conversion of functionally ineffective synapses to effective ones in the crossed phrenic pathway. This proposed mechanism is even more likely if 5-HT were colocalized with glutamate in the same terminal.

Because 5-HT has been shown to colocalize in the CNS with several neurotransmitters, including GABA (74), substance P (131), and thyrotropin-releasing hormone (131), it is also possible that 5-HT may coexist with glutamate in synaptic terminals terminating on phrenic motoneurons. By employing a combination of retrograde tracing and tricolor immunohistofluorescence techniques, Nicholas et al. (101) showed that glutamate immunoreactivity is present in virtually all substance P and 5-HT immunoreactive neurons in the medulla that have projections to the spinal cord. In addition, the triple immunoreactive terminals were demonstrated in an area immediately adjacent to motoneurons in the spinal cord (101). Furthermore, Johnson (74) showed in intracellular recording studies from 5-HT neurons in microculture that 40% of the neurons evoked excitatory glutamatergic potentials in themselves or in target neurons. Finally, in the study by Tai et al. (122), a comparison of the percentage of 5-HT-labeled terminals (47% of control) with the percentage of both glutamatergic and GABAergic terminals in the phrenic nucleus (81% combined in the control group, Ref. 121) suggests a major overlap between these two populations. Thus it is feasible that 5-HT and glutamate may colocalize in the same terminal in the phrenic nucleus.

Because previous physiological studies have shown that the conversion of functionally ineffective synapses in the crossed phrenic pathway is likely to occur within 2 h after C₂ hemisection in rats (102), injury-induced changes in the 5-HT innervation of the phrenic nucleus would also have to occur within hours after injury before the changes could be related directly to the conversion of the synapses. In the quantitative morphometric study by Tai et al. (122), it was necessary to employ a 30-day posthemisection interval to clearly differentiate between the terminals of injured 5-HT axons and the terminals of uninjured axons, which presumably sprout across the midline of the spinal cord (114) to innervate the deafferented phrenic motoneurons. A 30-day posthemisection interval was selected to allow for the complete degeneration of the injured terminals. However, as already suggested, the enhanced 5-HT innervation of phrenic motoneurons documented by the study of Tai et al. (122) at 30 days posthemisection could be related to an enhanced release of glutamate from the surviving terminals of the medullary respiratory centers (i.e., in the crossed phrenic pathway). What needs to be stressed here is that, based on the morphology of the terminals undergoing the most prominent alterations in the phrenic nucleus within hours after C₂ hemisection (58, 119), it is likely that many of the synaptic terminals contain 5-HT. Specifically, Sperry and Goshgarian (119) showed that the mean length of both asymmetrical and symmetrical synaptic active zones in unlabeled terminals elongated significantly in the phrenic nucleus 2 h after hemisection compared with control. The same finding was made for labeled 5-HT terminals at 30 days posthemisection (122). In the study by Sperry and Goshgarian (119), however, because of the short posthemisection interval, it was unclear whether bulbospinal connections interrupted by hemisection or the remaining crossed bulbospinal connections were undergoing synaptic modification, an issue resolved by the study of Tai et al. (122). Finally, it should be stressed that in both studies the majority of the elongating synapses were asymmetrical and presumably excitatory (119, 122).

In summary, elongation of synaptic active zones in 5-HT terminals after spinal cord hemisection may be a form of functional compensation for spinal cord injury-induced synaptic loss. As 5-HT is capable of reducing the threshold for glutamate-evoked action potentials on spinal motoneurons (127, 128), more neurotransmitter released by elongated 5-HT synaptic active zones in the phrenic nucleus could significantly augment phrenic motoneuron excitability mediated by glutamate after injury.

The issue of the role of serotonin in the conversion of functionally ineffective synapses of the crossed phrenic pathway was approached in another way. Although it was clear that 5-HT levels were significantly altered in time after spinal cord injury, it was unclear whether normal levels of 5-HT must be present before spinal cord injury to help initiate the morphological plasticity thought to be related to the conversion of functionally ineffective synapses. To address this issue, the effects of para-chlorophenylalanine (p-CPA), a serotonin synthesis inhibitor, on plasticity in the rat phrenic nucleus 4 h after C₂ hemisection were investigated (61, 62). Hemisected control rats (vehicle-injected) demonstrated typical morphological changes in the ipsilateral phrenic nucleus, including: 1) an increased number and length of synaptic active zones and 2) an increased length and number of dendrodendritic membrane appositions. p-CPA treatment 3 days before hemisection reduced 5-HT levels and resulted in an attenuation of the above changes in the ipsilateral phrenic nucleus 4 h after hemisection compared with hemisected controls (61). In addition, p-CPA treatment attenuated injury-induced alterations in immunohistochemical staining of glial fibrillary acid protein (GFAP) in astrocytes, although Western blot analysis demonstrated that overall levels of GFAP did not differ significantly between groups (61).

The change in GFAP immunohistochemical staining noted in hemisected control rats was related to GFAP filaments undergoing depolymerization and redistribution toward astroglial soma during hemisection-induced astroglial process retraction (61). The extent to which GFAP is polymerized is related to the phosphorylation state of GFAP filaments (71). Several neurotransmitters are known to alter the phosphorylation of GFAP (85, 130). Because p-CPA treatment attenuates alterations in GFAP immunohistochemical staining after spinal cord injury, 5-HT may influence GFAP phosphorylation through a receptor-mediated mechanism that alters the phosphorylation of GFAP (61). Thus astroglial process retraction in p-CPA-treated animals is attenuated. It is not known whether the attenuation of astroglial process retraction in p-CPA-treated animals after hemisection is directly related to the attenuation of the synaptic plasticity also noted in these animals. It may be that the reduction of 5-HT in terminals of the phrenic nucleus before hemisection is responsible for the overall reduction in synaptic plasticity. Furthermore, it needs to be stressed that p-CPA also decreases norepinephrine and dopamine levels. Thus the effects described above may not be due to the depletion of 5-HT alone (61).

Correlative morphological/physiological studies indicated that, when the morphological substrate believed to underlie the conversion of functionally ineffective synapses was present in the phrenic nucleus (i.e., within hours after C₂ hemisection or cold block), there was an augmentation of the expression of the CPP (13, 14, 58, 102, 119). Administration of p-CPA before hemisection attenuated the morphological alterations in phrenic nucleus cytoarchitecture that typically occur within hours after hemisection as described above (61). A correlative physiological study tested the effects of p-CPA administration on the expression of the CPP 4 h after C₂ hemisection in rats (62). p-CPA was administered 3 days before hemisection as in the morphological study. Eight of eight saline-injected control rats expressed crossed phrenic nerve activity at 4 h posthemisection, whereas only four of eight rats in the p-CPA-treated group expressed recovery in the phrenic nerve ipsilateral to hemisection (62). Quantification of integrated phrenic nerve waveforms indicated that the mean amplitude of respiratory-related activity in the ipsilateral phrenic nerve was significantly lower in p-CPA-treated rats than that in controls. Moreover, saline controls demonstrated significant increases in mean respiratory frequency and mean amplitude of contralateral phrenic nerve activity during asphyxia, compared with normocapnia. However, p-CPA-treated rats did not express significant differences in either frequency or amplitude during asphyxia compared with normocapnia. These results suggest that the depletion of normal levels of 5-HT by p-CPA before hemisection attenuates the enhanced expression of phrenic nerve activity seen within hours after hemisection. Coupled with the morphological data (61), the results further suggest that the attenuation of crossed phrenic activity in p-CPA-treated rats may be due to the failure of the hemisection to induce the plasticity in the phrenic nucleus thought to underlie the enhanced expression of activity. However, the fact that p-CPA-treated animals do not show the characteristic increase in frequency during asphyxia suggests that 5-HT depletion may also affect brain stem respiratory centers (62). Moreover, the absence of an increase in phrenic nerve amplitude in p-CPA-treated rats during asphyxia may reflect either spinal or supraspinal mechanisms. Another interpretation is that p-CPA blunted chemoreception, which could, in this case, have a peripheral (e.g., carotid body) component.

	OTHER EXAMPLES OF PLASTICITY IN THE RESPIRATORY PATHWAYS FOLLOWING SPINAL CORD INJURY

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY ASSOCIATED WITH THE... MORPHOLOGICAL PLASTICITY... MORPHOLOGICAL CHANGES IN THE... ASSOCIATION OF PHRENIC NUCLEUS... IS THE PLASTICITY ASSOCIATED... EFFECTS OF SEROTONIN ON... OTHER EXAMPLES OF PLASTICITY... BEHAVIORAL EFFECTS OF SPINAL... CLOSING REMARKS REFERENCES

The majority of the work involving the underlying mechanisms associated with the conversion of functionally ineffective synapses in the crossed phrenic pathway focused on events that occur rapidly within hours after spinal cord injury. Control studies from this early work indicated that there was no spontaneous recovery of the hemidiaphragm paralyzed by C₂ hemisection out to 30 days posthemisection; therefore, it was concluded that the hemidiaphragmatic paralysis was permanent (50). More recently, it was discovered that this is not the case (95). Female Sprague-Dawley rats subjected to C₂ spinal cord hemisection and allowed to survive for prolonged posthemisection periods (4-16 wk) show spontaneous recovery of the paralyzed hemidiaphragm, which occurs in a time-dependent manner. Specifically, spontaneous recovery did not occur in any of the animals tested 4 wk after injury. However, spontaneous recovery in the phrenic nerve and hemidiaphragm ipsilateral to hemisection was evident at 6 wk postinjury. The number of animals showing recovery increased progressively with time. By 12 and 16 wk, all animals demonstrated spontaneous recovery (95). The amount of recovery was expressed quantitatively in two ways: 1) as a percentage of activity in the contralateral phrenic nerve in the same hemisected animal and 2) as a percentage of activity in the homolateral phrenic nerve of noninjured animals. The recovery amounted to 56.76 ± 5.10% of the activity in the contralateral phrenic nerve of hemisected rats or 28.75 ± 3.19% of the activity in the homolateral nerve of noninjured rats (95). Ongoing studies in the Goshgarian laboratory are designed to determine what the underlying mechanisms of this newly discovered aspect of plasticity in the respiratory pathways may be.

Another very recent study has suggested a drug-induced aspect of plasticity in the respiratory pathways that has heretofore not been described (93). Eldridge et al. (37, 38) first demonstrated in intact cats that the systemic administration of the methylxanthine, theophylline, enhances respiratory drive by blocking adenosine receptors. Subsequently, several studies have been published that show that the systemic administration of theophylline into C₂ spinal- hemisected rats also increases central respiratory discharge, activates the crossed phrenic pathway, and thus restores function to the paralyzed hemidiaphragm (94, 96-99). The shortened phrenic burst duration and increased burst amplitudes in the neurograms reported by Nantwi et al. (94) suggest that respiratory drive was increased. The above studies, however, involve straightforward modulatory influences of this drug on the respiratory system and do not involve plasticity. Recently, however, an aspect of theophylline-induced plasticity in the respiratory pathways was discovered (93) and will be briefly described here. Rats subjected to a left C₂ hemisection received chronic oral theophylline administration (3 times/day) for 3, 7, 12, or 30 days. Some rats were assessed for recovery of left phrenic nerve activity at the end of the drug administration period, whereas others were weaned from drug administration for an additional 7, 12, or 30 days before assessment of respiratory recovery. After an electrophysiological recording, theophylline serum analyses were conducted and the spinal cord of each animal was removed for in situ hybridization and immunochemistry to assess adenosine A₁ receptor mRNA levels in the phrenic nucleus.

The results indicated that chronic theophylline administration induced recovery in the phrenic nerve ipsilateral to hemisection over a theophylline serum range of 1.2-1.9 µg/ml (93). These data are in accord with the previous studies involving theophylline-induced activation of the crossed phrenic pathway (94, 96, 97-99). Interestingly, theophylline-induced recovered activity persisted for as long as 30 days after the animals were weaned from theophylline and serum levels of the drug were virtually undetected. Previous work indicated that theophylline mediates its effects in the C₂ spinal hemisection model by blocking adenosine A₁ receptors (99). There were no significant changes, however, in adenosine A₁ receptor mRNA expression after chronic theophylline administration (93). Thus it was concluded that the persistent recovery of respiratory activity that follows chronic theophylline administration and weaning from the drug is not related to changes in adenosine A₁ receptor mRNA expression (93). This interesting aspect of drug-induced plasticity in the respiratory pathways is in marked contrast to the modulatory actions of asphyxia induced by placing a topical anesthetic on the phrenic nerve contralateral to hemisection (1, 50) or crushing the phrenic nerve (50). Regardless of how long the modulatory effects were induced (i.e., several hours or even weeks), when the stimulus was removed, the CPP was attenuated (1, 50). However, weaning a rat from theophylline after as little as only a 3-day exposure to the drug will not attenuate activity transmitted over the crossed phrenic pathway (93). The mechanism underlying this drug-induced plasticity is unknown, but ongoing studies are directed toward resolving this issue.

	BEHAVIORAL EFFECTS OF SPINAL CORD INJURY AND CLINICAL RELEVANCE

TOP ABSTRACT INTRODUCTION ANATOMIC SUBSTRATE FOR THE... PLASTICITY