Proc. Natl. Acad. Sci. USA Vol. 94, pp. 11003–11007, September 1997 Neurobiology

Maintenance of a somatotopic cortical map in the face of diminishing thalamocortical inputs (plasticity͞divergence͞cortical magnification)

E. G. JONES*, P. R. MANGER†, AND T. M. WOODS

Department of Anatomy and Neurobiology, University of California, Irvine, CA 92697

Communicated by Vernon B. Mountcastle, Johns Hopkins University School of Medicine, Baltimore, MD, July 25, 1997 (received for review May 29, 1997)

ABSTRACT This study addresses the extent of divergence of monkeys after making destructive lesions of increasing size in the ascending somatosensory pathways of primates. Diver- in parts of the thalamic representation of the hand. The results gence of inputs from a particular body part at each successive show that substantial parts of the thalamic representation of a synaptic step in these pathways results in a potential magni- finger can be removed before the representation of that finger fication of the representation of that body part in the somato- begins to shrink in the cortical somatotopic map. The degree sensory cortex, so that the representation can be expanded of divergence implied by these results is sufficiently great to when peripheral input from other parts is lost, as in nerve explain most cases of representational plasticity in the somato- lesions or amputations. Lesions of increasing size were placed sensory cortex even after substantial peripheral lesions. in the representation of a finger in the ventral posterior thalamic nucleus (VPL) of macaque monkeys. After a survival MATERIALS AND METHODS period of 1–5 weeks, area 3b of the somatosensory cortex ipsilateral to the lesion was mapped physiologically, and the Fifteen macaque monkeys (Macaca mulatta and Macaca nem- extent of the representation of the affected and adjacent estrina) were anesthetized with intramuscular ketamine and fingers was determined. Lesions affecting less than 30% of the maintained on a continuous intravenous infusion of Nembutal. thalamic VPL nucleus were without effect upon the cortical In 10, microelectrodes were introduced into one thalamus representation of the finger whose thalamic representation horizontally from behind, entering the thalamus via the visual was at the center of the lesion. Lesions affecting about 35% of cortex and midbrain. The ventral posterior lateral nucleus the VPL nucleus resulted in a shrinkage of the cortical (VPL) was identified, and receptive field mapping of single representation of the finger whose thalamic representation units and multiunit clusters responding to natural cutaneous was lesioned, with concomitant expansion of the representa- stimulation was used to isolate the extent of the representation tions of adjacent fingers. Beyond 35–40%, the whole cortical of a single finger (which normally occupies a more or less representation of the hand became silent. These results sug- parasagittal lamella, running on a curved trajectory through gest that divergence of brainstem and thalamocortical pro- the anteroposterior extent of VPL; Fig. 1A). Either a small jections, although normally not expressed, are sufficiently electrolytic lesion was made along the length of a large part of great to maintain a representation after a major loss of inputs this trajectory by passing 5–8 mA through the electrode as it from the periphery. This is likely to be one mechanism of was withdrawn or the recording electrode was replaced by a representational plasticity in the cerebral cortex. dual recording͞injection cannula attached to a hydraulic de- livery system and 300 nl of 2 or 10 mM kainic acid (pH 3.5) was Plasticity of the body map in somatosensory cortex permits the injected along the trajectory representing a finger. Larger representation of a body part to expand or contract under the lesions were made by reintroducing the electrode or cannula influence of enhanced or reduced input from sensory recep- and making further electrolytic lesions or injections along tors of skin, muscle, and joints and probably underlies per- trajectories dorsal, ventral, medial, and͞or lateral and parallel ceptual adaptations that occur during learning of new tactile to the first. skills (1) and after nerve injury (2) or amputation (3). The monkeys were allowed to recover; 1–5 weeks after Activity-dependent expansion of part of the body map in the lesioning, they were reanesthetized; and the representation of cerebral cortex at the expense of adjacent parts may depend the upper limb was mapped in area 3b of the somatosensory upon functional expression of divergent but previously sup- cortex ipsilateral to the lesion, by using the same single and pressed inputs from the periphery (4, 5) that reach the cortex multiunit recording of responses to peripheral stimulation as via the ventral posterior nucleus of the thalamus, upon long- for the thalamus. Microelectrodes were introduced at 200- to range intracortical connections (6), and͞or upon sprouting of 500-␮m intervals and were advanced down the posterior bank axons (5) in one or both pathways. of the central sulcus. Five normal monkeys in which the same The extent of thalamocortical divergence is a limiting factor area was mapped served as controls. At the conclusion of the in the potential for expansion of one representation at the cortical mapping, the animals were perfused with 4% para- expense of another (7–10). It can be magnified by similar formaldehyde in 0.1 M phosphate buffer (pH 7.4), the brains divergence of primary afferents to the dorsal column nuclei removed, infiltrated with sucrose, and frozen in dry ice. Serial and of dorsal column nuclear efferents to the ventral posterior frozen sections 30-␮m thick were cut from the thalamus in the thalamic nucleus. The degree of this magnification has been frontal plane and from the cortex in the parasagittal plane. determined in the present study: the representation of the Alternate sections were stained with thionin, histochemically hand was mapped physiologically in the somatosensory cortex Abbreviations: VPL, ventral posterior lateral thalamic nucleus; VPM, The publication costs of this article were defrayed in part by page charge ventral posterior medial thalamic nucleus. *To whom reprint requests should be addressed. e-mail: egjones@ payment. This article must therefore be hereby marked ‘‘advertisement’’ in uci.edu. accordance with 18 U.S.C. §1734 solely to indicate this fact. †Present address: Neurobiology Research 151A3, Sepulveda Veterans’ © 1997 by The National Academy of Sciences 0027-8424͞97͞9411003-5$2.00͞0 Administration Medical Center, 16111 Plummer Street, North Hills, PNAS is available online at http:͞͞www.pnas.org. CA 91343.

11003 Downloaded by guest on September 30, 2021 11004 Neurobiology: Jones et al. Proc. Natl. Acad. Sci. USA 94 (1997)

FIG.1. (A) Representation of the body surface in the normal macaque monkey at a single coronal level of the ventral posterior nucleus (redrawn from ref. 11). The representation of a finger runs in a curved lamella through the full anteroposterior extent of the VPL nucleus. (B–D) Lesions of various sizes in cytochrome oxidase-stained sections from different monkeys. The lesions were made electrolytically along the length of the trajectory representing a finger. All lesions resulted in a reduction of immunostaining for the calcium binding protein parvalbumin (25) in a localized part of the layer IV plexus in area 3b (see Fig. 3C), its size proportional to the size of the thalamic lesion, indicating a loss of thalamocortical fibers.

for cytochrome oxidase, or immunocytochemically for the other fingers occupies somewhat less (11). The curving nature calcium binding protein parvalbumin. of the representations makes it impossible for any small lesions Camera-lucida drawings were made of the Nissl- and cyto- made in the present manner to occupy the representation of chrome oxidase-stained sections through the thalamus, show- one finger throughout its full antero-posterior extent. All ing the extent of the lesion and its relationship to the borders lesions resulted in a reduction of immunostaining for parval- of VPL; the volume of VPL occupied by the lesion was bumin in a localized part of the layer IV plexus in area 3b (Fig. calculated. Nissl-stained sections of the cortex were used to 3C), its size proportional to the size of the thalamic lesion, locate microelectrode tracks in relation to the cytoarchitecture indicating a loss of thalamocortical fibers. of area 3b of the somatosensory cortex, and a map of the hand Small cylindrical lesions 0.5–1 mm in diameter that ran and adjacent parts of the upper limb representation was anteroposteriorly through the thalamus affecting over much of generated on the flattened cortex, by relating electrode tracks their length part of the representation of a single finger (11) to receptive field data collected during the terminal experi- were without effect on the cortical map in area 3b (Fig. 3A). ment. The sections stained immunocytochemically for parv- In these cases, a complete cortical representation of the finger albumin were used to identify regions of area 3b in which the was found, its internal topography was intact, and the size of layer IV plexus of parvalbumin-stained thalamocortical fibers the representation (2.5–10 mm2) was similar to that found in was deficient. the control monkeys (see also ref. 12). Receptive fields of neurons within the finger representation were identical in size RESULTS to those in normal monkeys and their responsiveness to stimuli was indistinguishable from those of normal monkeys within the Electrolytic or kainic acid lesions resulted in the destruction of limitations of the experimental setup. thalamic cells over a region whose size was related to the Lesions centered in the representation of a single finger amount of current passed or volume of excitotoxin injected. were without effect upon the cortical representation of that The lesioned area was distinguishable by loss of Nissl or finger until approximately 30% of the volume of VPL was cytochrome oxidase staining in the histological sections destroyed (Figs. 2 and 3B). When 30–35% of VPL was through the thalamus (Fig. 1). One monkey had a lesion lesioned, the cortical representation of the finger whose tha- affecting 0.25–5% of VPL, five had lesions affecting 5–10%, lamic representation was at the center of the lesion was four had lesions affecting 10–25%, two had lesions affecting reduced in size (Fig. 3B). In the case illustrated, involving a 30–40%, and two had lesions affecting 40% or more of VPL 35% loss of VPL and after a survival of 2 weeks, the cortical (Fig. 2). The representation of the first (D1) and second (D2) representation of the fourth finger was much narrower than in fingers in VPL each occupy 7–10% of its volume; that of the controls, was split by the representation of the middle finger, Downloaded by guest on September 30, 2021 Neurobiology: Jones et al. Proc. Natl. Acad. Sci. USA 94 (1997) 11005

FIG. 2. Camera-lucida drawings of frontal sections in anteroposterior order through the ventral posterior nuclei of one monkey, upon which have been plotted the extents of the lesions in VPL of two different brains. The cortical maps from these cases are illustrated in Fig. 3. VMb, basal ventral medial nucleus; VPI, ventral posterior inferior nucleus.

and was 59% less extensive in area than in normal monkeys (1.5 After lesions that destroyed 35–40% of VPL, and likely mm2 vs. 3.7 Ϯ 0.71 mm2). The receptive fields of many neurons affected the whole thalamic hand representation, the cortical in the diminished area 3b representation were indistinguish- representation of most of the hand failed. In the region of area able in size from those of controls and had equally well-defined 3b normally devoted to the representation of the hand, neu- borders. All parts of the finger appeared to be represented in rons in layer IV were uncharacteristically silent and failed to the shrunken broken finger representation. The reduction of respond even to intense peripheral stimuli, and a topographic the cortical representation of the finger was accompanied by map of the upper limb could not be defined. No neurons in the expansions of the representations of adjacent fingers. In the area of reduced activity responded to stimulation of the trunk case illustrated in Fig. 3, the representations of the third and or face. fifth fingers expanded to 8.3 mm2 and 4.5 mm2 in comparison Behaviorally, monkeys with even the smallest lesions of VPL with means (ϮSEM) of 4.3 Ϯ 0.78 mm2 and 3.2 Ϯ 0.64 mm2, displayed evidence of some reduction in tactile acuity in the respectively, in the five controls. affected digit, failing to withdraw the hand when the digit was

FIG. 3. Representation of the upper limb in area 3b of the ipsilateral somatosensory cortex after lesioning the thalamic representation of a finger. (A) Map for which less than 30% of VPL was destroyed, the lesion being centered on the thalamic representation of the fourth finger (D4); this and lesions of similar size were without effect on the size of the cortical map of the finger whose thalamic representation was at the heart of the lesion. (B) Cases such as this, in which approximately 35% of the volume of VPL was destroyed, revealed a shrunken fragmented cortical representation of the finger (D4) and expansion of the representations of neighboring fingers (D3 and D5). Dots represent sites of microelectrode penetrations. (C) Localized reduction in parvalbumin staining (25) (arrows) in the layer IV plexus of area 3b, indicating destruction of thalamocortical fibers resulting from a lesion in VPL. Downloaded by guest on September 30, 2021 11006 Neurobiology: Jones et al. Proc. Natl. Acad. Sci. USA 94 (1997)

stimulated while their attention was distracted. Larger lesions destruction of thalamic neurons. After the smallest lesions, the gave evidence of a loss of tactile sensation in the palm and in preservation of the cortical finger representation can probably two or more digits, and the largest lesions suggested the be explained by overlap and divergence of projections from presence of complete anesthesia of most of the upper limb. remaining parts of the thalamic finger representation: the curving trajectory of the thalamic representation as it extends DISCUSSION anteroposteriorly through VPL, and its dorso-ventral extent (11), makes it impossible for the orthogonally oriented elec- The present results imply that despite removal of large parts of trodes to have destroyed more than a small part of it. It is the representation of a finger in the thalamus, remaining inconceivable, however, that the largest lesions did not remove thalamocortical connections are capable of maintaining a all or most of the thalamic representation of the target finger. cortical representation of that finger, which after a certain Under these circumstances, the cortical representation of the limit diminishes in size. After a second limit is reached, the same finger held up to some extent but was encroached upon whole representation breaks down. Maintenance of a cortical by that of neighboring fingers. We take this observation to finger representation in the face of reducing thalamic inputs is indicate the diverging projections of relay cells in the still likely to depend upon: (i) preservation of part of the thalamic unremoved parts of the thalamic representations of the neigh- finger representation, reflecting the extended nature of its boring fingers. Beyond a critical point, approaching a 40% thalamic representation and the divergence of its cortical volume loss of VPL, the cortical representation of the whole projection; and (ii) divergence of peripheral inputs from the hand fell silent. At this point, it is assumed, the whole finger to the dorsal column nuclei and of the outputs of these representation of the hand in the thalamus had been removed. nuclei to the thalamus. In this way, inputs from a specific finger The premise upon which the above interpretation is based would reach thalamic cells that represent adjacent fingers. is that, under normal conditions, the divergent projections are Although normally suppressed in the production of the fine weaker or suppressed, probably by GABAergic inhibitory grain somatotopic map in the cerebral cortex, as determined mechanisms (16), only being revealed under activity- physiologically, under the conditions of these experiments they dependent conditions. That is, when their activity is enhanced would be exposed and thus maintain a cortical map. by increased or coherent sensory input (1, 17) from the part The results may be compared with those in the Silver Spring that they represent, the representation of the part that they monkeys (5) in which a 12-year deafferentation of the upper overlap being suppressed, or when inputs to the overlapped limb resulted in transneuronal degeneration of cells in the part of the representation are reduced, as in peripheral nerve cuneate nucleus and in the forelimb representation in VPL section (2, 4, 9, 15), amputation (14, 15, 18, 19), or suppression (13), removing a volume of cells approximately equal to that of activity (20). of the larger lesions in the present study. In these animals, the At the 1- to 5-week survival times used, the silenced cortical cortical upper limb representation, which should have been hand representation was not encroached upon by the adjoining silenced by such a procedure, showed encroachment of an representations of the face. This contrasts with the Silver expanded representation of the lower jaw (5). This could Spring monkeys in which the lower jaw representation, in depend in large part on amplified divergence of the type particular, had invaded cortical territory formerly occupied by demonstrated herein. the representation of the hand (5). In terms of the divergence The extent of divergence in the thalamocortical projections hypothesis, presented above, the failure of expansion of the of VPL and ventral posterior medial thalamic nucleus (VPM) was demonstrated in a previous study (8). It was shown the face representation in the present cases of larger electrolytic adjacent cells within the same part of the thalamic represen- lesions may have stemmed from destruction of fibers of tation of a part of the body, such as a finger, can project to foci passage from VPM as they traversed the lesion in VPL. This in area 3b up to 1.5 mm apart, but no more. Most cells in the explanation would not apply, however, in the case of the kainic thalamic representation send their axons to a common zone in acid lesions. area 3b that constitutes the cortical representation of that It is possible that there is little or no overlap of trigeminal finger. The 1.5-mm divergence, however, ensures that the and dorsal column inputs to the thalamus across the VPM͞ axons of a few cells actually terminate in the representation of VPL border, thus preventing expansion of the face represen- an adjacent finger. The 1.5-mm divergence can account for the tation on the basis of ascending divergence, but this has not ‘‘cortical distance factor’’ (9), the 1- to 1.5-mm expansion of a been explored in detail. There is overlap in the peripheral part of the cortical representation at the expense of adjacent distributions of the mandibular nerve and upper two cervical parts that occurs in the short-term after amputation of a digit nerves in monkeys (21), and this is reflected in overlap of the or division of a single peripheral nerve (2, 14). It cannot representations of the lower jaw and neck in area 3b (22). The account for expansions greater than 1–5 mm, which occur after degree of divergence in the cortical projections of VPM and sectioning of multiple nerves (10) or after even more extensive VPL, coupled with a lack of divergence in the subcortical deafferentations (5, 15). For this, other mechanisms have been inputs to these two nuclei, in these short-term experiments, invoked. Among the proposed mechanisms, the normal diver- could militate against an expansion of the face representation gence of inputs that occurs at each successive relay center in into that of the hand. Long-range intracortical connections do the ascending dorsal column-lemniscal pathway (i.e., in the not extend across the border of the hand and face represen- dorsal column nuclei and thalamus) holds some appeal, since tations to a significant extent (22) and so could not provide a a relatively small degree of divergence of primary afferent basis for short-term expansion of the face representation. In inputs to the dorsal column nuclei or principal trigeminal the long-term, therefore, as in the Silver Spring monkeys, other nucleus could be ‘‘magnified’’ by additional divergence of mechanisms such as sprouting of intracortical connections (23) lemniscal fibers at the ventral posterior nucleus and of its may need to be invoked to explain the expansion of the face projections to area 3b. This would result in the amplification representation. Axon sprouting at subcortical centers (15) may of thalamocortical divergence beyond the 1.5-mm set by the also have been involved. divergence of individual thalamocortical neurons. The present results also imply that the spinothalamic system The present study tested the limits of this amplified diver- is incapable of supporting a hand representation in the absence gence by placing lesions at the point at which subcortical of the thalamic relay for lemniscal fibers. Although some of the divergence would be maximal, i.e., in the thalamus itself. The terminations of the anterolateral system were inevitably de- results are surprising in the extent to which the cortical stroyed by the lesions, substantial numbers end outside VPL in representation of a finger could be maintained after extensive adjacent nuclei such as the posterior and in the small-celled Downloaded by guest on September 30, 2021 Neurobiology: Jones et al. Proc. Natl. Acad. Sci. USA 94 (1997) 11007

zones of calbindin immunoreactive cells around VPM (24), and 11. Jones, E. G. & Friedman, D. P. (1982) J. Neurophysiol. 48, their cortical projections would have been spared. 521–544. The behavioral observations in the lesioned monkeys, al- 12. Kaas, J. H., Pons, T. P., Wall, J. T., Garraghty, P. E. & Cusick, though not quantitatively evaluated, suggest that the presence C. G. (1987) Somatosens. Res. 4, 309–331. of a complete map, as demonstrated with conventional map- 13. Rausell, E., Cusick, C. G., Taub, E. & Jones, E. G. (1992) Proc. ping procedures, can still be accompanied by perturbed sen- Natl. Acad. Sci. USA 89, 2571–2575. sory perception, an effect that is, in fact, commonly associated 14. Garraghty, P. E. & Kaas, J. H. (1991) Proc. Natl. Acad. Sci. USA with amputations and peripheral nerve injuries. The mainte- 88, 6976–6980. nance of a map cannot, therefore, be equated with the 15. Florence, S. L. & Kaas, J. H. (1995) J. Neurosci. 15, 8083–8095. maintenance of normal function. 16. Merzenich, M. M., Nelson, R. J., Stryker, M. P., Cynader, M. S., Schoppmann, A. & Zook, J. M. (1984) J. Comp. Neurol. 224, This work was supported by Grant NS-21377 from the National 591–605. Institutes of Health. 17. Jones, E. G. (1993) Cereb. Cortex 3, 361–372. 18. Clark, S. A., Allard, T., Jenkins, W. M. & Merzenich, M. M. 1. Recanzone, G. H., Merzenich, M. M., Jenkins, W. M., Grajski, (1988) Nature (London) 332, 444–445. K. A. & Dinse, H. R. (1992) J. Neurophysiol. 67, 1031–1056. 19. Manger, P. R., Woods, T. M. & Jones, E. G. (1996) Proc. R. Soc. 2. Merzenich, M. M., Kaas, J. H., Wall, J. T., Sur, M., Nelson, R. J. London Ser. B 263, 933–939. & Felleman, D. J. (1983) Neuroscience 10, 639–666. 20. Calford, M. B. & Tweedale, R. (1988) Nature (London) 332, 3. Ramachandran, V. S., Rogers-Ramachandran, D., Stewart, M. & 446–448. Pons, T. P. (1992) Science 258, 1159–1160. 21. Sherrington, C. S. (1939) in Selected Writings of Sir Charles 4. Kaas, J. H. (1991) Annu. Rev. Neurosci. 14, 137–168. Sherrington, ed. Denny-Brown, D. (Hamish Hamilton Medical 5. Pons, T. P., Garraghty, P. E., Ommaya, A. K., Kaas, J. H., Taub, Books, London), pp. 31–93. E. & Mishkin, M. (1991) Science 252, 1857–1860. 22. Manger, P. R., Woods, T. M., Mun˜oz, A. & Jones, E. G. (1997) 6. Lund, J. P., Sun, G.-D. & Lamarre, Y. (1994) Science 265, J. Neurosci., in press. 546–48. 7. Garraghty, P. E. & Sur, M. (1990) J. Comp. Neurol. 294, 583–593. 23. Darian-Smith, C. & Gilbert, C. D. (1994) Nature (London) 368, 8. Rausell, E. & Jones, E. G. (1995) J. Neurosci. 15, 4270–4288. 737–740. 9. Kaas, J. H., Merzenich, M. M. & Killackey, H. P. (1983) Annu. 24. Rausell, E., Bae, C. S., Vin˜uela, A., Huntley, G. W. & Jones, E. G. Rev. Neurosci. 6, 325–356. (1992) J. Neurosci. 12, 4088–4111. 10. Garraghty, P. E., Hanes, D. P., Florence, S. L. & Kaas, J. H. 25. Jones, E. G. & Hendry, S. H. C. (1989) Eur. J. Neurosci. 1, (1994) Somatosens. Mot. Res. 11, 109–117. 222–246. Downloaded by guest on September 30, 2021