Stages Of Plasticity

In the preceding section, the patterns of reorganization that were described represented endpoints of a progressive process that occurs over a time course of months and, in some cases, years. The endpoints are achieved by a succession of changes that evolve after deafferentation. For example, some new features of organization usually are apparent immediately, and additional changes emerge during the subsequent weeks and months. The importance of the temporal distinctions for the different stages of plasticity is that they would not be expected if all plasticity was explained by one plasticity mechanism, or by modulation of changes that appear rapidly. Instead, the distinctive temporal signature of each stage indicates that each is subserved by a different cellular process, or mechanism. As a cautionary note, however, at any given point in the plasticity sequence, the observed pattern of reorganization no doubt represents an integration of changes produced by multiple plasticity mechanisms. Thus, the plasticity process may be more akin to a progressive evolution of changes that emerge over an extended time-continuum rather than discernible epochs during which changes occur separated by periods of little plasticity.

13.3.1 Changes that Appear Rapidly

The emergence of new cortical receptive fields after the suppression or removal of the dominant excitatory inputs can be observed instantaneously after a peripheral manipulation. The most elegant demonstrations of this immediate reactivation require that neuronal recordings be maintained at the same site during an inactivation experiment to rule out interpretation errors that could result from small differences in electrode placement before and after the denervation. In the first study of this kind, Calford and Tweedale42,43 implanted electrodes in primary somatosensory cortex of flying foxes. In initial experiments to test the reproducibility of the implanted electrodes in normal animals, little change in receptive field location was observed over the course of 3-4 weeks. However, when a digit on the forelimb was amputated, new receptive fields on nearby, intact parts of the forelimb became apparent immediately after the procedure at some recording sites. Critically, these new fields appeared only at sites where the initial receptive field was restricted to the amputated digit. At sites where the dominant activating inputs remained intact, no changes were observed. The rapid emergence of new receptive fields in primary somatosensory cortex of monkeys was described subsequently by Calford and Tweedale using this same strategy.23 They also showed that if the dominant sensory inputs are deactivated transiently, by subcutaneous lidocaine injections, the original receptive field re-emerged when the anesthetic effects of the lidocaine dwindled and displaced the field that appeared during deactivation.

Another approach to detect rapidly emerging changes in receptive fields after sensory deactivation involves high-resolution mapping. First, in the intact animal, the cortical representation of the sensory surface is mapped. Then, the dominant excitatory inputs to a delimited portion of the mapped zone are deactivated, usually by nerve transection or digit amputation. Promptly thereafter, a second map is made of the sensory representation, usually with electrodes placed in approximately the same locations as before the deactivation. Because of possible errors in electrode placement, it cannot be certain that the same neurons are studied before and after the denervation. Thus, the procedure is not as compelling as the same-site recordings described above. However, the advantage of the mapping procedure is that it can provide insight about the topographic distributions of the acute alterations. The results from the mapping studies indicate that neurons that immediately acquire new receptive fields after sensory deactivations characteristically are positioned at the edge of the deactivated zone, near adjacent sensory representations where activating inputs remain intact (Figure 13.3; see also References 2, 34, and 44). Neurons that are distant from functionally intact inputs typically are not reactivated within a short time frame (Figure 13.3).

Subcortical relays in the lemniscal pathway also have potential for dynamic adjustments in receptive field size and location. The strategy of maintaining recordings at the same site to evaluate effects of deactivation on receptive field organization has been applied in non-primates at multiple levels of the somatosensory pathway, including the ventroposterior nucleus45-48 and the dorsal column nuclei of the brain stem.48-53 At all levels, neurons deprived of their dominant sensory activation can acquire new receptive fields on nearby intact sensory surfaces just as described for cortex, although it is not certain what contribution each of the subcortical relays make to the immediately observed changes found in cortex (see below for more discussion).

There are a few important exceptions to the well-established notion that latent inputs are expressed rapidly after sensory denervation. The most recent comes from a study of the effects of peripheral cold blockade on cuneate neurons in cats by Zhang and Rowe.54 No evidence of new receptive fields was obtained during the period of cold-induced deafferentation. At present there is no explanation for the results, which seem to be in conflict with nearly all other reports. However, limited changes may have been overlooked. For example, if the blockade eliminated both dominant and latent inputs to the majority of the neurons being studied, then only those at the perimetry of the cold blockade, where silent inputs were spared, would acquire new receptive fields immediately. Alternatively, perhaps the time course of the blockade was too short. Tests for new receptive fields began 5 minutes after the cold blockade was established and were continued for no longer than 30 minutes; however, some studies have reported that acute changes only begin to emerge 15-30 minutes after deactivation.46-53 Northgrave and Rasmusson55 also found no new receptive fields in the cuneate nucleus of raccoons when stimulation was applied to non-deprived skin adjacent to the denervated zone, called "off-focus" stimulation. The emphasis in this paper was on inhibitory interactions in the cuneate nucleus, and it appears that only high-threshold stimuli (squeezes) were applied to off-focus sites. Perhaps the activation produced by the high-threshold stimuli suppressed any weak excitatory influences that may have resided in off-focus tactile inputs.

13.3.2 Changes that Develop Over an Intermediate Time Course

In the large majority of studies on the functional effects of sensory denervation in the somatosensory system, the emphasis is on changes that appear a few days after deactivation or within the subsequent weeks and months. The deprivation must be maintained throughout the period of observation, via some manipulation that separates the peripheral inputs from the ascending relays or by making a lesion in a portion of the sensory representation at an early station in the somatosensory pathway. These types of experiments have been performed in a wide range of species using diverse deafferentation strategies. The pattern of reorganization that appears and the time required for the changes to emerge can vary considerably, depending on the manipulation used to invoke the plasticity and the size of the deprived representation centrally. However, two general statements hold true for nearly all the results. First, the extent of reactivation increases with time after sensory dener-vation. Thus, there is more extensive reactivation apparent a few weeks after the deprivation than is observed immediately, and still more reactivation is present months later (Figure 13.5). Second, there are spatial limits to the extent of reactivation that can be mediated by intermediate mechanisms. These conclusions recur in

FIGURE 13.5 Plots showing the changes with time in the size of intact sensory representation that dominates deprived cortex after median nerve transection in a monkey (top) and after sciatic nerve transection in rats (bottom). In the case of the monkey, the size of the dorsal skin representation for the lateral half of the hand was measured from Figure 7 of Reference 2. Evidence for the change in size of the saphenous nerve representation after sciatic nerve cut in rats is from Figure 2 of Reference 67. In both species, the representations double in size from the control values immediately after deafferentation. The representations slowly increase further in size over the course of the following weeks and months, until all or nearly all the denervated neurons are reactivated.

FIGURE 13.5 Plots showing the changes with time in the size of intact sensory representation that dominates deprived cortex after median nerve transection in a monkey (top) and after sciatic nerve transection in rats (bottom). In the case of the monkey, the size of the dorsal skin representation for the lateral half of the hand was measured from Figure 7 of Reference 2. Evidence for the change in size of the saphenous nerve representation after sciatic nerve cut in rats is from Figure 2 of Reference 67. In both species, the representations double in size from the control values immediately after deafferentation. The representations slowly increase further in size over the course of the following weeks and months, until all or nearly all the denervated neurons are reactivated.

virtually all cases where somatosensory deactivation has been followed over extended time frames.

In primates, much of the information on time course of reactivation comes from studies of the effects of median nerve transection, to denervate the lateral half of the palmar hand. The emergence of new receptive fields on the dorsal surface of the hand may require many weeks before the full extent of potential change is expressed (Figure 13.5). Initially, neurons only along the borders of the deactivated zone are responsive to cutaneous stimulation of adjacent intact parts of the hand.2-18'34-44

Gradually over time, the zone of reactivation extends further and further into the core of the denervated representation and ultimately encompasses most or all of the once-deprived representation.1A2a21-56

Another feature of reorganization that evolves over time after the nerve transec-tion is the topographic precision within the reactivated zone. As originally reported by Merzenich's group,2 the new receptive fields that occupy deprived cortex are large initially, but gradually decrease in size. Churchill and colleagues21 have proposed that this receptive field refinement is attributable to a "consolidation" process whereby the most useful synaptic inputs are extracted through a use-dependent process of selection from all available excitatory inputs. In their study, monkeys that were sacrificed 2-4.5 months after median nerve transection were compared with others sacrificed more than 11 months after the same manipulation. The deprived zone was reactivated at the earliest time points, but topography was crude. At the later time point, topographic consolidation had occurred and receptive fields were refined and progressed in a more orderly manner. There were few details about the time course of the consolidation, since many months had ensued between the two time points studied; however, work from the Dykes group in cats that had forelimb denervation suggests that topographic refinements occur gradually over time (see below).

Evidence that there can be spatial limits to the cortical reorganization, even after weeks and months, comes from deafferentation strategies where both the palmar and dorsal surfaces of the hand are denervated so that both the dominant and latent inputs to the central representation are silenced. One such manipulation involves transection of combinations of the nerves to the hand. The median and radial nerves relay sensory information from both the glabrous and hairy surfaces of the lateral half of the hand, and the ulnar and radial nerves relay information from the medial half of the hand. After transection of either combination of nerves, a large region of the affected cortex remained silent for up to 11 months, the longest time point examined (see Reference 57; see also Figure 13.4). There may have been some enlargement of the remaining hand representation along the border of the deprived zone, but not enough to provide new sources of activation to the full extent of the deprived region in the cortical map. Even in early postnatal monkeys, after transec-tion of combined nerves to the hand followed by long recovery, small regions of area 3b remain unresponsive to tactile stimulation.58 The persistence of silent cortex over intermediate time courses suggested that the spatial extent of the deprived cortex was greater than could be reactivated by the available plasticity mechanisms.

Further evidence that the potential for reactivation is spatially limited within a few weeks or months after sensory deprivation comes from studies where one or more digits on the hand are amputated. Digit amputations produce patterns of deprivation centrally much like transection of nerves to the dorsal and palmar surfaces of the hand, in that both the dominant and latent inputs are removed. If only one digit is amputated in monkeys, the extent of the cortical representation that would be affected is only about a millimeter in width. When the electrophysiological recording experiments were performed weeks or months later, the deprived zone was completely reactivated by inputs from the amputated stump or from the adjacent digits.4,22,24 In contrast, if two or more adjacent digits are amputated, the denervated zone would involve a cortical extent of 2 millimeters or more, some of which remain silent for months following the injury.4 With much longer survival times (e.g., years), no evidence of the deactivated cortical neurons remain,24 even if all the digits are amputated.25 These reactivations likely reflect different mechanisms than those that subserve plasticity over an intermediate time frame, and will be discussed in the subsequent section.

The most extensive denervations of the forelimb representation have been produced by cervical dorsal rhizotomy35 and by dorsal column transection,5 which eliminates all or most of the sensory inputs from the forelimb. The only comprehensive evaluation of the time course of reactivation in cortex after large-scale deactivation was done by Jain et al.5 and the outcome reinforces the notion that there are spatial limits to the extent of cortex that can be reactivated over an intermediate time frame. If all sensory inputs from the forearm are removed, neurons in area 3b remain unresponsive to tactile stimulation for many months.5-59 Presumably, there are limited expansions of the intact sensory representations along the border of the deprived zone, yet the large central portion of the affected region is rendered unresponsive. Very slowly developing changes eventually bring about extensive reactivation of deprived cortex, but as for above, these data will be discussed in the subsequent section.

Only a small number of studies have looked carefully at the temporal sequence of reactivation in non-primates, but the data that have emerged from these few are consistent with the scenario proposed for primates. Kelehan and Doetsch60 examined the effects of digit amputation on the somatosensory cortex of raccoons at 1 hour, 1 week, 2 weeks, 4 weeks, and 36 weeks after the injury. In raccoons, the digit representations in somatosensory cortex are greatly magnified, much as in primates, so that the progressive reactivation of the extensive zone of cortex deprived by the amputation was readily apparent. However, because of the vast extent of the deprived zone produced by the amputation, some non-responsive neurons persisted even at the longest survival time. The explanation for this is the same as for monkeys. The latent inputs from adjacent sensory representations do not span the full extent of a digit representation, so that the central core zone has no alternative source of activation after digit amputation. This conclusion is consistent with the findings of Rasmusson and colleagues.61,62 Dykes and colleagues63 studied somatosensory cortex in cats that had transection of multiple nerves in the forelimb to remove all sensory innervation to the forepaw (see also Reference 64). Reactivation appeared initially near the border between the deprived forelimb and the adjacent trunk representation and over time progressed further and further into the denervated cortex. Receptive fields at the leading edge of the reactivated cortex were large and poorly defined, and other abnormal physiological properties were apparent. However, over time, some of the most abnormal features were suppressed, much like the consolidation described by Churchill et al.21 Even up to a year after deafferentation of the forelimb, some neurons in somatosensory cortex of cats remained non-responsive to tactile stimulation.

The time course for reactivation of somatosensory cortex has also been studied in rats after elimination of inputs from the sciatic nerve, which innervates a large portion of the hindlimb.65-67 An important contribution of these studies is that they demonstrated the rates at which the changes occur. After sciatic nerve deafferenta-tion, neurons in the deprived zone of cortex near the border of the intact saphenous nerve territory acquire new receptive fields on saphenous nerve skin, so that the saphaneous nerve representation expands. Quantitative evaluations of the size of the saphenous representation with time after sciatic denervation indicate that the representation expands to more than double its normal size within days of the injury (Figure 13.5; see also Reference 67). Subsequently, the saphenous nerve field again doubles in size; however, the expansion is much slower than that observed within the first few days after injury, spanning up to 8 months (Figure 13.5).

There are a few exceptions to the conclusion that new patterns of sensory representation emerge with time after peripheral denervation. In an early influential study by McMahon and Wall,68 no evidence of receptive field reorganization was found in the dorsal column/trigeminal complex after transection of nerves to the hindfoot in adult rats. Originally, it was presumed that the outcome demonstrated the implastic nature of the dorsal column nuclei under conditions where the sensory neurons remain. However, given the more recent evidence for plasticity under even less disruptive conditions (i.e., temporary anesthetic block49), it is likely that small changes in receptive field organization were overlooked. This same explanation might account for the absence of new receptive fields in the trigeminal brain stem after infraorbital nerve section.69 Another apparent exception comes from rats in which the dorsal columns were cut at thoracic levels.70 Much of the deafferented hindlimb portion of S1 remained unresponsive to tactile stimulation, even after months of recovery.70 Similar results were obtained in adult cats when the dorsal columns and all other ascending afferents from the hindlimb were sectioned by spinal cord transection at lower thoracic levels.71 The emphasis in these studies was the possibility of a functional takeover of the hindlimb representation by forelimb inputs, and although a forelimb expansion was not observed there was no compelling evidence for a complete absence of reorganization. Indeed, as described by McKinley and Smith,71 some reactivation of the deprived hindlimb neurons by trunk inputs was detected. Presumably, this occurred along the border of the deprived zone via mechanisms common to most reports of plasticity.

13.3.3 Changes that Require a Long Time Course

Initially, few studies looked at the long-term (many months to years) consequences of sensory denervation. In cases where the deafferentation spared latent inputs, the process of reorganization was complete in a matter of months.2-17-20 In cases of more extensive denervations, whatever changes had appeared within two or so months after the initial deafferentation were thought to be all that the mature brain allowed.4,57 No systematic inquiries challenged this presumption until Pons et al.35 found evidence of cortical reactivation over distances of more than 11 mm in monkeys that had lived for 12 years or more after dorsal rhizotomy of spinal segments C2-T4. The procedure had eliminated all sensory input from the arm; nonetheless, neurons throughout the arm representation in somatosensory cortex had become responsive to sensory stimulation of other intact sensory surfaces. The vast majority of neurons responded to stimulation of the face.

The specific pattern of change was reminiscent of clinical reports of phantom sensations in humans with amputation of the hand or some portion of the upper limb. Characteristically, these individuals have the sensation that the missing limb is still present,72 and some patients feel touch on the digits of their missing arm, when touched on the side of the face ipsilateral to the amputation.73,74 This suggested that, much as described by Pons et al.35 in monkeys with limb deafferentation, the deprived hand representation in cortex of human amputees might be activated by stimulating the face. This supposition has now been confirmed using non-invasive magnetic source imaging methods. After loss of the forelimb in humans, the region of cortex activated by the face expands to occupy much of the forelimb region.75-81

Reactivation of large extents of forelimb cortex also was detected in monkeys that had hand or forearm amputation.25,26 Additional evidence for large-scale reactivations in somatosensory cortex have been reported after dorsal column section in monkeys.5 In some cases, the deprived cortical representation becomes reactivated by an expanded representation of the face, but in some cases where arm inputs are left intact, such as after hand amputation, most of reorganization involved inputs from the remaining stump of the arm.26 Nonetheless, the major point of the expansions that have been observed after long time courses is not the source of the new activation, but that the spatial extent of the reorganization is much more massive that previously thought possible.

The common explanation for the large-scale reactivation is that it reflects processes that evolve slowly over time. In owl monkeys that had dorsal column section, Jain and colleagues5 found that full reactivation of hand cortex appeared by 8 months after the deactivation. The macaque monkeys that were studied after hand amputation25,26 and those that had dorsal rhizotomy35 had survived for years after the denervation. Thus, there was no information about the time required for the full extent of the reactivations to emerge. However, complete reorganization may take longer in higher-order primates than those reported for owl monkeys by Jain et al.5 The cortical representation of the hand in macaque monkeys is several orders of magnitude larger than in owl monkeys; thus, the progression of reactivation may be considerably more lengthy.

There has been recent argument that the reactivation may not be a protracted process. The emergence of phantom limb sensations can be quite rapid. The regions of skin that evoked phantom limb sensations when stimulated, called "trigger zones," were apparent within 24 hours after amputation in one individual and expanded over the course of the next 8 weeks.82 Such rapid emergence of phantom sensations would not be expected if the neurological mechanism for the sensations was a slowly developing process. The trigger zones that were apparent most rapidly after the injury involved the skin of the stump, the upper arm in the case of the one individual studied.82 Since the upper arm is represented immediately adjacent to neurons that relay information from the forearm in the cuneate nucleus, and separated from hand neurons in the cuneate neurons by less than 200 microns (e.g., Reference 7), the earliest phantoms may result from new patterns of activation that involve only limited reactivations. Of course, the relationship between the perception of phantom limbs and the structural/functional changes observed in primary somatosensory pathway is unknown, and, thus, these arguments are purely speculative, at present.

In non-primates, most of the studies of the effects of limb deafferentation are directed toward understanding developmental mechanisms of plasticity.40-64-71-83-85 Fewer have examined the organization of somatosensory cortex after adult deaffer-entation and the data are not as clear-cut as in primates. In rodents, adult forelimb amputation renders much of S1 non-responsive for up to 16 weeks (the longest time point included).85 Yet, sprouting of hindlimb inputs into denervated cuneate nucleus has been reported within 2 weeks after forelimb denervation through cervical dorsal rhizotomy in adult rats.86 Perhaps the new inputs to cuneate require longer time courses to activate target neurons and initiate new patterns of activation in the ascending pathway, or perhaps GABA mechanisms suppress the new inputs, much as reported by Lane et al.84 in rats that had early neonatal forelimb amputation. Up to a year after forelimb denervation in cats, neurons at only about half the recording sites had acquired new receptive fields,63 but longer time courses were not examined. Finally, in one raccoon that had a forelimb amputation, reactivation of the full extent of the massive forelimb cortex was reported.87 The injury had occurred at some unknown time prior to capture, but appeared to have been a long-standing injury. Thus, the mechanism of reactivation in this animal may have been akin to that described above for primates.

13.4 MECHANISMS OF PLASTICITY 13.4.1 Adjustments in Synaptic Efficacy

As originally suggested by Wall88 the rapid emergence of new receptive fields probably reflects the unmasking of normally suppressed excitatory inputs. Presumably, the latent inputs were suppressed by the activity of the dominant inputs, but become functionally viable when the suppression was released. The premise to this notion is that the full complement of sensory inputs that converge on individual neurons is not expressed under normal conditions (Figure 13.6); there is more divergence and convergence of thalamocortical inputs than is expressed normally in the functional maps of somatosensory cortex of primates.27-28 The dominant inputs presumably are those that were reinforced through correlated patterns of sensory-driven activity.89 However, the process is highly dynamic and the dominance of those inputs reflects on-line adjustments in the relative strengths of diverse excitatory and inhibitory inputs (for review, see Reference 90). The latent inputs that are not expressed are often referred to as a subthreshold "fringe." The rapid expression of the inputs in this fringe region is attributed to a reduction of the excitatory drive of inhibitory cortical neurons, called "afferent-driven inhibition."91 When antagonists of the inhibitory neurotransmitter, GABA, are administered intracortically to eliminate inhibitory influences, somatosensory receptive fields become markedly larger than when inhibition was intact. The rapid acquisition of new receptive fields after peripheral sensory denervation seems to occur simultaneously at multiple levels of the pathway34,48 and it is assumed that the changes at each level are mediated by local activity-dependent modulation of GABAergic inhibitory processes.48


A A,

Axonl Axon 2

Axonl Axon2

0 0

Post a comment