Subcortical contributions to massive cortical reorganizations.

Abstract

The dynamic nature of sensory representations in the adult brain gives us the lifelong potential to adapt to changes in our environment and to compensate for injury. Reorganizations in sensory systems typically involve relatively limited shifts in the topography of the functional representations, and in most reorganizations of the somatosensory cortex only neurons within 1–2 mm of the borders of the affected zone in cortex acquire new or altered receptive fields (11Merzenich M.M Nelson R.J Stryker M.P Cynader M.S Schoppman A Zook J.M J. Comp. Neurol. 1984; 224: 591-605Crossref PubMed Scopus (973) Google Scholar). In fact, for many years, it was assumed that this was the maximal distance for plasticity in the adult central nervous system. Yet, the adult brain is capable of much more extensive changes, involving much larger extents of the nervous system. Large-scale peripheral deafferentations, such as limb amputation or spinal cord damage, lead to extensive reactivations of the large regions of somatosensory cortex deprived by the injury. These unusually large changes in cortical organization are not easily explained by cellular mechanisms involving the potentiation of previously existing connections. However, there is now evidence that new connections may grow into regions deprived of primary afferent inputs as a result of injury. Subcortical sensory representations are smaller than their cortical counterparts; therefore, even limited new growth subcortically can lead to massive cortical reactivations. The first suggestion that the somatosensory system is capable of extensive reorganization came from a report of a single raccoon that had lost a forearm at some unknown time prior to its capture (15Rasmusson D.D Turnbull B.G Leech C.K Neurosci. Lett. 1985; 55: 167-172Crossref PubMed Scopus (38) Google Scholar). The large representation of the forepaw in primary somatosensory cortex of this animal had been completely reactivated by an expanded representation of the stump. However, the finding received little attention, perhaps because the age of the animal when the injury occurred was not known, so it was not conclusive that the extensive change reflected mechanisms of adult plasticity. Several years later, a subsequent report of extensive cortical change convinced the research community that the adult brain is capable of enormous change. 12Pons T.P Garraghty P.E Ommaya A.K Kaas J.H Taub E Mishkin M Science. 1991; 252: 1857-1860Crossref PubMed Scopus (840) Google Scholar studied the impact of long-standing sensory deafferentation of the arm on somatosensory cortex in macaque monkeys. The sensory inputs from the periphery project to a series of cortical fields, called areas 3a, 3b, 1, and 2, which each contain orderly somatotopic representations of body surface and muscle receptors. The sensory deafferentations eliminated the normal sources of activation from the large hand, wrist, and arm representations of these and other somatotopic fields. Quite surprisingly, microelectrode recordings from the monkeys revealed that the deprived cortex was largely or completely reactivated by inputs from the face (12Pons T.P Garraghty P.E Ommaya A.K Kaas J.H Taub E Mishkin M Science. 1991; 252: 1857-1860Crossref PubMed Scopus (840) Google Scholar). Neurons throughout the explored zone of reactivated cortex responded vigorously to light touch and brushing of hairs on the face, much as in the normal cortical representations of the face. The reactivated cortex extended some 10–14 mm mediolaterally and 9–11 mm rostrocaudally, a much greater extent than found previously after more limited nerve damage (11Merzenich M.M Nelson R.J Stryker M.P Cynader M.S Schoppman A Zook J.M J. Comp. Neurol. 1984; 224: 591-605Crossref PubMed Scopus (973) Google Scholar). Moreover, similar reactivations undoubtedly took place in higher-level somatosensory representations such as the second somatosensory area, S2, and the parietal ventral area, PV, since these areas depend on inputs from anterior parietal cortex. The report of extensive reactivation of deprived cortex in these monkeys was soon followed by a description of a patient with an arm amputation who felt light touch to the face as being both on the face and on the fingers of the missing arm (14Ramachandran V.S Rogers-Ramachandran D Stewart M Science. 1992; 258: 1159-1160Crossref PubMed Scopus (321) Google Scholar). This observation suggested that the brains of such individuals reorganize after amputation, as in the monkeys with arm deafferentation, and that touching the face leads to activation of both face and reactivated hand zones of cortex, hence the misplaced and double sensations. This supposition was soon given further credibility when it became possible to record from the cortex of three monkeys with long-standing therapeutic amputations of the hand or forelimb. Recordings from area 3b in these monkeys revealed a reactivation of the deprived hand and forelimb cortex by inputs from the stump of the arm and the face (Figure 1; see also 6Florence S.L Kaas J.H J. Neurosci. 1995; 15: 8083-8095PubMed Google Scholar). In addition, there is evidence from noninvasive imaging of evoked activity in the brains of humans with arm amputations that cortex formerly devoted to the missing hand comes to be activated when the face is stimulated (e.g.5Flor H Elbert T Knecht S Wienbruch C Pantev C Birbaumer N Larbig W Taub E Nature. 1995; 375: 482-484Crossref PubMed Scopus (1372) Google Scholar). Finally, after deactivation of large extents of somatosensory cortex by transection of the dorsal columns of the spinal cord at a high cervical level (9Jain N Catania K.C Kaas J.H Nature. 1997; 386: 495-498Crossref PubMed Scopus (157) Google Scholar), the large unresponsive zones of cortex in areas 3a, 3b, and 1 became responsive to inputs from the face and the few dorsal column afferents from the anterior arm that entered the spinal cord above the cut (Figure 1). These related findings demonstrate beyond any doubt that large extents of somatosensory cortex, when deprived of normal sources of activation, can become responsive to remaining peripheral sensory inputs. Of course these observations, by themselves, do not tell much about how or where the reorganization occurs. It seems unlikely that the changes occur only in cortex, since alterations at all subcortical levels of the somatosensory pathway have been documented after peripheral nerve injury and other losses of activation. Yet, in the few cases where the question of a causal relationship between the subcortical and cortical changes have been explored, outcomes have differed. In a study of use-dependent changes on the somatosensory system, monkeys had very specific patterns of reorganization in primary somatosensory cortex after undergoing training to detect specific stimulus sequences; however, there were no detectable changes in the ventroposterior nucleus of the thalamus (19Wang X Merzenich M.M Sameshima K Jenkins W.M Nature. 1995; 378: 71-75Crossref PubMed Scopus (363) Google Scholar). Yet, in another study that used sensory deafferentation of the glabrous surface of the hand to induce plasticity, the deprived representations in both the ventroposterior nucleus and in somatosensory cortex were similarly reactivated by the dorsal hand afferents (8Garraghty P.E Kaas J.H Neuroreport. 1991; 2: 747-750Crossref PubMed Scopus (122) Google Scholar). Thus, the potential for a contribution to cortical plasticity from the subcortical somatosensory relays appears to vary and may depend on the sensory manipulation. To directly address the potential for subcortical changes after sensory denervations such as limb amputations, recordings were made from the thalamus of human amputees as part of a clinical procedure. Neurons in the ventroposterior nucleus that would normally be responsive to stimuli on the missing limb were responsive instead to touch on the stump of the missing limb (3Davis K.D Kiss Z.H.T Luo L Taskar R.R Lozano A.M Dostrovsky J.O Nature. 1998; 391: 385-387Crossref PubMed Scopus (184) Google Scholar). The pattern of reactivation in the ventroposterior nucleus seemed to mirror that which had been found in cortex of limb amputees (e.g.5Flor H Elbert T Knecht S Wienbruch C Pantev C Birbaumer N Larbig W Taub E Nature. 1995; 375: 482-484Crossref PubMed Scopus (1372) Google Scholar). Additionally, in our recent recordings from monkeys with accidental forearm injuries and therapeutic amputations, or with unilateral cervical dorsal column section, neurons throughout the completely deprived portions of the ventroposterior nucleus were responsive to light touch on the face or stump of the amputated arm. Thus, subcortical centers can contribute to the changes seen in cortex after injury. This is not to say that cortical plasticity, such as that mediated through use-dependent processes, is not a component of the overall outcome. Indeed, cortical mechanisms can suppress the effects of subcortical alterations. In a recent study of the developmental effects of neonatal forelimb amputation in rats, hindlimb inputs expanded into the deprived representation of the forelimb in the cuneate nucleus, but this hindlimb activation was suppressed in somatosensory cortex by GABA-mediated inhibitory mechanisms (10Lane R.D Killackey H.P Rhoades R.W J. Neurophysiol. 1997; 77: 2723-2735PubMed Google Scholar). Nonetheless, the substrate exists for subcortical contributions to cortical reorganization. The important question that still remains is: “How do these types of reactivations occur?” A common explanation for most types of adult plasticity, both in cortex as well as subcortically, is that previously existing but hidden inputs in the normal anatomical framework of the somatosensory system become expressed (18Wall P.D Philos. Trans. R. Soc. Lond. B Biol. Sci. 1977; 278: 361-372Crossref PubMed Scopus (252) Google Scholar). Connections with only subthreshold effects become potentiated through local modifications of synaptic weights and come to activate neurons that were previously unresponsive to the inputs. Such changes in synaptic strengths could involve the participation of NMDA receptors in activity-dependent plasticity, the regulation of inhibitory amino acid (e.g., GABA) levels, and/or the release of neuromodulators and other regulatory molecules (for a more extensive review, see 4Ebner F.F Rema V Sachdev R Symons F.V Semin. Neurosci. 1997; 9: 47-58Crossref Scopus (25) Google Scholar). Yet these mechanisms cannot be the only basis for the extensive reactivations sometimes seen after amputation and spinal cord lesion when the forearm representation comes to be largely reactivated by inputs from the face. The existing connectional framework allows for some overlap of the hand and face representations in cortex, but none that could account for the full extent of the injury-induced changes. The alternative explanation for the extensive reactivation is that new connections grow. This would seem improbable, because there has been a long-held view that new growth, often called sprouting, is limited in the mature brain. However, evidence has been accumulating that sprouting does occur in the adult central nervous system. Indeed, injections of anatomical tracers into the stump of monkeys with long-standing arm amputations labeled sensory afferents from the arm that had sprouted beyond the original termination zones in the cuneate nucleus of the brain stem, into the hand and digit portions of the nucleus (6Florence S.L Kaas J.H J. Neurosci. 1995; 15: 8083-8095PubMed Google Scholar). More recently, similar injections were made into the skin of the face of two monkeys with an arm amputation and two monkeys with dorsal column section at a high cervical level. In these monkeys, sparse but obvious connections were observed from the trigeminal (face) nucleus to the cuneate (hand) nucleus. Thus, it appears that afferents from the face that normally terminate in the trigeminal nucleus had sprouted into the cuneate nucleus. These anatomical demonstrations suggest that the growth of even a limited number of new connections at the brain stem level of the somatosensory system reactivates some of the deprived neurons that they contact. These reactivations could be relayed to thalamus and cortex and provide the framework for functional reactivation over very large extents of cortex. Finally, there is evidence from monkeys with long-standing forelimb injuries that horizontal connections in cortex are more extensive than usual, so that neurons in the deprived cortical zones extend outside their normal territory and project to nearby nondeprived cortex; conversely, neurons in nondeprived cortical zones sprout into the deprived region of cortex (7Florence S.L Taub H.B Kaas J.H Science. 1998; 282: 1117-1121Crossref PubMed Scopus (353) Google Scholar). The new cortical growth no doubt contributes to the final pattern of reorganization that is expressed in cortex and perhaps participates in additional dynamic processes that reflect the experiential history of the individual. Similar new growth also may accompany the plastic changes that have been reported in other systems. Sprouting of horizontal connections has been reported in the deprived portion of visual cortex after retinal lesions in cats (Darian-Smith and Gilbert, 1995); the new connections are proposed to provide new sources of activation for neurons in deprived cortex. In the auditory system, the possibility of new growth has not been addressed; however, much as in other sensory systems, primary auditory cortex that is deprived by peripheral lesions comes to be reactivated by other intact inputs (13Rajan R Irvine D.R.F Wise L.Z Heil P J. Comp. Neurol. 1993; 333: 17-49Crossref Scopus (286) Google Scholar). In the motor system, the basis for much of the cortical reorganization that has been described in human amputees is proposed to be a reduction of GABAergic inhibition (1Chen R Corwell B Yaseen Z Hallett M Cohen L.G J. Neurosci. 1998; 18: 3443-3450PubMed Google Scholar); the issue of new growth has not been addressed. Other examples of sprouting have been reported in the adult central nervous system; however, there is little information about how the growth is facilitated. Growth factors and neurotrophins may be involved. Growth-associated protein 43 (GAP-43) is upregulated in the central terminations of sensory nerves after injury (20Woolf C.J Reynolds M.L Molander C O’Brien C Lindsay R.M Benowitz L.I Neuroscience. 1990; 34: 465-478Crossref PubMed Scopus (302) Google Scholar), and the synthesis of neurotrophins is modulated by changes in levels of excitatory activity, at least in the developing brain (reviewed by 17Thoenen H Science. 1995; 270: 593-598Crossref PubMed Scopus (1666) Google Scholar). Thus, changes induced either by the injury itself or by the subsequent alterations in activity levels may alter the synthesis of growth promoters and provide an environment that is supportive of new growth. Massive injuries such as amputation and spinal cord injury produce an initial surge of excitatory activity followed by sharply reduced levels of activity throughout the somatosensory pathway, and the sheer scale of this shift could have a substantial impact on these types of mechanisms. Perhaps another impact of the altered synaptic activity after injury is to reduce the expression of growth inhibitors that normally limit growth in the mature nervous system (see 16Schwab M.E Kapfhammer J.P Bandtlow C.E Annu. Rev. Neurosci. 1993; 16: 565-595Crossref PubMed Scopus (505) Google Scholar). A related issue is whether the new growth is sufficient to activate neurons rendered nonresponsive by injury. Presumably the new connections, even though sparse, could be greatly potentiated by local alterations in synaptic strengths and numbers. In both somatosensory and visual cortex, sensory deprivation leads to a robust downregulation of inhibitory neurotransmitters, and this may allow previously ineffective excitatory inputs to be expressed. Additionally, levels of excitatory neurotransmitters may increase, and the local growth of axon and dendrite arbors may occur as well. Thus, we conclude that the growth of new connections at both subcortical and cortical levels of the system, in combination with changes in connectional strengths at these same levels, mediate the major reorganizations in cortex. Functionally, such major reorganizations are likely to be detrimental, resulting in perceptual errors and unwanted sensations. Compelling evidence for these types of undesirable outcomes was provided by microstimulation studies in the thalamus of humans suffering from postamputation pain (3Davis K.D Kiss Z.H.T Luo L Taskar R.R Lozano A.M Dostrovsky J.O Nature. 1998; 391: 385-387Crossref PubMed Scopus (184) Google Scholar). When neurons in the region of thalamus that previously innervated the amputated limb were activated by tactile stimulation to the stump, sensations continued to be felt as if they were on the limb. Thus, the stimulated neurons maintained their original role in perception, rather than assuming the new roles that would be consistent with their new source of activation. Under other circumstances, neurons that come to be activated by new sources of inputs might be reintegrated into the computational networks and perhaps produce much more useful clinical outcomes. Empirically derived treatments for sensory loss and brain damage may improve with a sound theoretical approach based on maximizing beneficial plasticity and reducing detrimental changes in the brain.*To whom correspondence should be addressed (e-mail: [email protected]).