though LGd neurons received morpho-
logically normal synaptic input from the
single remaining eye. This study further
emphasizes that the binocular competi-
tion critical for the initial formation of
the normal visual system does not re-
quire visual experience since it exercises
its influence before birth. Finally, the
dependence of normal development on
prenatal binocular competition is selec-
tive; segregation of the LGd into magno-
and parvocellular moieties and the lami-
nar distribution of their terminals in the
cerebral cortex developed normally in all
experimental animals. This example of
how an error in development of a single
structure can alter distant but related
structures in the complex primate brain
may offer insight into various abnormali-
ties of lamination or connections that
occur in congenital malformations in hu-
Yale University School ofMedicine,
New Haven, Connecticut 06510
The inability of axons to elongate for
more than a few millimeters within the
damaged central nervous system (CNS)
is believed to be a fundamental feature of
the failure of regeneration in the brain
and spinal cord of adult mammals. On
the other hand, axons from transected
References and Notes
1. T. N. Wiesel and D. H. Hubel, J. Neuorphysiol.
26, 1003 (1963); ibid. 28, 1029 (1965); R. W.
Guillery, J. Comp. Neurol. 149, 423 (1973); R.
D. Lund, T. J. Cunningham, J. S. Lund, Brain
Behav. Evol. 8, 51 (1973); J. A. Robson, C. A.
Mason, R. W. Guillery, Science 201, 635 (1978).
2. F. H. Baker, P. Griggand, G. K. von Norden,
Brain Res. 66, 185 (1974).
3. D. H. Hubel, T. N. Wiesel, S. LeVay, Philos.
Trans. R. Soc. London Ser. B 278, 377 (1977).
4. S. Polyak, The Vertebrate Visual System (Univ.
of Chicago Press, Chicago, 1957); J. H. Kaas, R.
W. Guillery, J. M. Allman, Brain Behav. Evol.
6, 253 (1972); K. J. Sanderson, Aust. J. Optom.
63, 220 (1980).
5. D. H. Hubel and T. N. Wiesel, Proc. R. Soc.
London Ser. B 198, 1 (1977).
7. S. M. Sherman, K. P. Hoffman, J. Stone, J.
Neurophysiol. 35, 532 (1972); P. H. Schiller and
J. G. Malpeli, ibid., p. 788: B. Dreher, Y.
Fukada, R. W. Rodieck, J. Physiol. (London)
258, 432 (1976).
8. R. W. Guillery, Prog. Brain Res. 51, 403 (1979).
9. P. Rakic, Soc. Neurosci. Abstr. 3, 573 (1977).
10. __, Science183, 425 (1974); J. Comp.
Neurol. 176, 23 (1977).
11. __. Nature (London) 261, 467 (1966); Phi-
los. Trans. R. Soc. London Ser. B 278, 245
12. The smaller size of the LGd in the monkeys
enucleated at later gestational ages may be due
to neuronal atrophy, loss of neurons. or both.
This effect may be related to the higher depen-
dence of already committed, more mature LGd
neurons on the proper retinal input.
13. Supported by PHS grant EY 02593.
6 May 1981; revised 28 July 1981
, J. Comp. Neurol. 146, 421 (1972).
peripheral nerves successfully regrow
over long distances when they become
associated with Schwann cells in the
distal nerve stump or in a nerve graft.
Although many mechanisms are proba-
bly involved (1), there is increasing evi-
dence that differences in the capacity of
certain damaged axons to elongate in the
CNS and peripheral nervous system
(PNS) are more dependent on the envi-
ronment in which these axons are locat-
ed than upon intrinsic properties of neu-
rons. This hypothesis, proposed by Cajal
(2), has received additional support from
recent studies. (i) Experiments in adult
mammals have demonstrated that, al-
though PNS axons, with a known capaci-
ty to regenerate, fail to lengthen in a
milieu of CNS glia (3), the axons from
intrinsic CNS neurons grow into periph-
eral nerve segments transplanted into
the transected spinal cord (4). (ii) Tis-
Schwann cells, other nonneuronal cells,
and factors in the culture media exert
various trophic influences on neurons
(5). Using a new experimental model, we
provide evidence that axons from nerve
cells in the injured spinal cord and brain-
stem can elongate for unprecedented dis-
tances when the CNS glial environment
is replaced by that of peripheral nerves.
In adult Sprague-Dawley rats weighing
between 250 and 350 g, segments of
autologous sciatic nerve 35 mm long
were used as "bridges" between the
medulla oblongata and the lower cervical
or upper thoracic spinal cord (Fig. IA).
These "bridges" were placed extraspi-
nally in the subcutaneous tissues along
the back of the animal. One end of the
graft was inserted through a laminecto-
my into the dorsolateral spinal cord, and
the other end was introduced into the
lower medulla through a small opening
made across the craniocervical junction.
We ensured the penetration of the graft
endings into the CNS by using a glass
rod with a 150-p.m tip. Retaining 10-0
sutures were placed at both ends of the
graft. The grafting procedure resulted in
local damage at the site of insertion of
the nerve into the dorsal brainstem and
spinal cord, but the rest of the neuraxis
was left intact. Because axons from spi-
Fig. 1. (A) Diagram of the dorsal surface of the rat CNS, showing a peripheral
nerve "bridge" linking the medulla and the thoracic spinal cord. Cross sections
depict the region where the ends of the nerve graft were inserted. The origin of
axons innervating the graft was determined by retrograde labeling with HRP.
Axonal elongation was measured between the site of HRP application and that
of the labeled cells in the CNS. For this purpose, when neurons were sought in
the medulla, the tracer was applied after sectioning the nerve at a level situated
approximately 30 mm from the brainstem and 5 mm from its caudal insertion
into the cord (group A, Table 1). Conversely, when the growth ofaxons from the
spinal neurons was assessed, the graft was cut close to the brainstem (group B,
Table 1). The short stumps of these nerve grafts were also used for anterograde
labeling. (B) Approximate rostrocaudal position of 1472 labeled CNS neurons
(dots) demonstrated in seven grafted rats. In the brainstem the territory
occupied by 450 of these cells extended along 4 mm, whereas 1022 labeled
neurons were scattered along a 6.5-mm segment of the spinal cord.
0036-8075/81/1 120-0931$01 .00/0
Copyright t 1981 AAAS
Axonal Elongation into Peripheral Nervous System "Bridges"
After Central Nervous System Injury in Adult Rats
Abstract. The origin, termination, and length ofaxonal growth after focal central
nervous system injury was examined in adult rats by means of a newt, experimental
model. When peripheral nerve segments were used as "bridges" between the
medulla and spinal cord, axonsfrom neurons at both these levels greis' approximate-
ly 30 millimeters. The regenerative potential of these central neurons seems to be
expressed when the central nervous system glial environment is changed to that of
the peripheral nervous system.
SCIENCE, VOL. 214, 20 NOVEMBER 1981
on October 24, 2008
nal roots may also innervate these grafts
(4), the two ipsilateral dorsal root ganglia
neighboring the site of the graft insertion
were avulsed. Animals survived without
apparent neurologic deficit and were
killed by systemic perfusion of fixative
between 22 and 30 weeks after grafting
(4). At autopsy, the ends of the grafts
were found to be in gross continuity with
the brainstem and spinal cord. The light
and electron microscope cross-sectional
appearance of the midportion of each
graft was similar to that of a regenerated
peripheral nerve and contained numer-
ous myelinated and unmyelinated fibers
ensheathed by Schwann cells.
In seven rats, the cells of origin and
the termination fields of axons traveling
in the regenerated graft were determined
by means of retrograde and anterograde
(HRP) (6) applied to the tips of the
transected graft (Table 1, groups A and
B). Two additional animals were used
only for the study of the terminal course
offibers. In all rats the HRP was applied
to the graft extraspinally, thereby mini-
mizing the possibility of a spurious label-
ing of neurons due to interstitial spread
of the tracer into the CNS. In addition,
we investigated extracellular diffusion of
the label in three control rats (Table 1,
Fig. 2. Diagrams illustrating
the position of HRP-labeled
neurons in cross sections of
(A) dorsal medulla oblongata
of one rat and (B) spinal cord
of another 22 and 26 weeks
after grafting. In both animals,
the tracer was applied to the
transected graft (G) approxi-
mately 30 mm away from the
level of these two sections.
Calibration bar, 500 ,um.
Fig. 3. Cross sections of (A) the medulla oblongata and (B) the spinal cord, illustrating the
course ofaxons at thejunction ofthe PNS graft (G) and the CNS tissue in two animals 26 and 30
weeks after grafting (dark-field micrographs, 48 hours after labeling with HRP). Bar, 200 ,um.
group C) by crushing the graft with jew-
eler's forceps 5 mm from the rostral and
caudal attachments of the "bridge" and
applying HRP to the nerve in its midpor-
tion. For both retrograde and antero-
grade studies, HRP (20 percent, Sigma
VI) was applied to the tips of the cut
nerve grafts in small. Gelfoam pads for 1
to 2 hours. The exposed portion of the
nerve was laid over a Parafilm sheet and
covered with Vaseline to avoid direct
contamination of the tissues by the trac-
er. Rats were killed 24 to 48 hours after
HRP application. Serial cryostat sec-
tions, 20 to 40 jxm thick, were obtained
from the brainstem and spinal cord and
reacted with tetramethylbenzidine and
H202 (4, 6).
In the seven animals used to demon-
strate both retrograde and anterograde
labeling, 450 neurons were labeled in the
medulla (Figs. lB and 2A) (Table 1) and
1022 in the gray matter of the spinal cord
(Figs. IB and 2B) (Table
neurons of various sizes were scattered
in the neighborhood of the graft inser-
tions at both these levels. Most of these
cells were located ipsilaterally to the
graft, along a territory that extended
rostrocaudally for approximately 4 mm
in the brainstem and 6.5 mm in the spinal
cord (Fig. IB). The following brainstem
nuclei (7) contained labeled neurons: (i)
nucleus intercalatus; (ii) nucleus reticu-
laris lateralis; (iii) nucleus reticularis me-
dullae oblongatae pars dorsalis; (iv) nu-
cleus olivaris inferior; (v) nucleus acces-
sorius olivaris dorsalis; (vi) nucleus ra-
spinalis nervi trigemini; (viii) nucleus re-
ticularis paramedianus; (ix) nucleus retic-
ularis medullae oblongatae pars ventra-
lis; and (x) nuclei gracilis and cuneatus.
In the spinal cord the HRP-labeled neu-
rons tended to be evenly distributed
within the gray matter ipsilateral to the
graft, but the superficial laminae of the
dorsal horn contained few labeled cells.
Labeled neurons were also found in dor-
sal root ganglia above and below the
insertions of the graft.
Because only four labeled cells were
demonstrated in the control animals (Ta-
ble 1, group C), we conclude that the
majority of neurons in the experimental
groups were not labeled spuriously by
diffusion along the graft or hematoge-
In the nine rats in which anterograde
transport of HRP was also investigated,
the tracer was applied to the stump ofthe
graft, approximately 5 mm from the
neuraxis. Labeled axons from the graft
were shown to have penetrated the spi-
nal cord and medulla only for approxi-
mately 2 mm, a distance that represents
SCIENCE, VOL. 214
on October 24, 2008
more than half the normal width of both Download full-text
these structures (Fig. 3, A and B). By
light and electron microscope examina-
tion it was documented that, along their
course within the spinal cord or medulla,
many of these fibers were ensheathed by
Schwann cells that had migrated into the
CNS tissues; other axons were
rounded by glial processes. Many of the
penetrating fibers terminated in close
proximity to CNS neurons, but it is not
known if they formed synapses because
connectivity was not investigated.
The results of these studies indicate
that some of the axons within the PNS
"bridges" originate from neurons in the
spinal cord and brainstem. Under the
conditions of these experiments such ax-
ons have been shown to be capable of a
growth that exceeds 30 mm, a distance
that could be equal to or greater than the
length of axons from some of these neu-
rons in the intact rat.
This new experimental model has sev-
eral advantages for studies of regenera-
tion in the living animal. (i) By selective-
ly positioning the graft, it is possible to
direct the course of axons from and into
specific regions of the CNS.
origin, length, and termination of axons
within the graft can be documented. (iii)
The long extraspinal course of these
grafts should facilitate the electrophysio-
logic investigation of axons within the
"bridges." (iv) Because these animals
are not paralyzed and retain bowel and
bladder control, in contrast to the case
with animals grafted after complete tran-
section (4), their care and survival is
greatly facilitated. (v) If it is eventually
demonstrated that axons from CNS neu-
establish functional connections
with cells in the target regions to which
they have been directed, it may be possi-
ble to devise experimental strategies for
selected populations of axons to bypass
damaged CNS tissue and connect with
specific groups of neurons at a distance.
Whether the central axons
bridging grafts originate by regrowth of
damaged CNS fibers or by sprouting
from uninjured neurons in the proximity
of the graft endings, or both, could not
be decided in this study. Regardless of
the mechanisms involved, the remark-
able elongation of axons in these animals
suggests that PNS tissues exert a striking
facilitation of the growth of axons from
central neurons after CNS injury. Even
though the cells of origin varied in size
and were distributed widely within the
CNS areas neighboring the site of entry
of the graft, it remains to be determined
whether they constitute a special neuro-
nal population or whether their respons-
es are examples of a more general poten-
SCIENCE, VOL. 214, 20 NOVEMBER 1981
numbers of spinal and medullary neurons
labeled in each animal after the application of
HRP to the caudal end of the graft, approxi-
mately 30 mm away from the medulla (group
A) or to the rostral end ofthe graft at the same
distance from the spinal cord (group B). Non-
boldface numbers designate cells labeled by
HRP applied to the shorter, 5-mm long, re-
maining stump of the bridging nerve. Group C
represents findings in control rats in which the
regenerated grafts were crushed approximate-
ly 30 minutes before HRP application.
1. Boldface numbers represent the
tial for regeneration. Our experiments
also demonstrate that regenerating axons
only penetrate the damaged CNS for
short distances. It is possible that elon-
gation fails in the CNS because the cen-
growth-promoting properties of the PNS
or because there are changes in the in-
jured CNS that inhibit fiber growth (8). If
the conclusion is corroborated that inter-
The voluntary motor commands to a
pool of spinal motoneurons can be ana-
lyzed in intact humans by recording the
action potentials of single motor units
through the skin (1). In gtaded voluntary
contractions of a muscle, motor units are
recruited in a stereotyped order at repro-
ducible levels of muscle force (2). This is
usually referred to as Henneman's size
principle (3) because the recruitment se-
quence is correlated with several graded
properties such as the size of the moto-
actions between axons and their immedi-
ate environment play a determinant role
in the success or failure of regeneration,
the study of the molecular basis of these
interdependencies may lead to better ex-
perimental approaches to promote CNS
ALBERT J. AGUAYO*
Neurosciences Unit, Montreal General
Hospital, and Department of
Neurology, McGill University,
Montreal, Quebec, Canada H3G IA4
References and Notes
1. L. Guth, Exp. Neurol. 45, 606 (1974); S. Varon,
ibid, 54, 1 (1977); R. P. Veraa and B. Grafstein,
ibid. 71, 6 (1981); C. W. Cotman and J. V.
Nadler, in Neuronal Plasticity, C. W. Cotman,
Ed. (Raven, New York, 1978), pp. 227-271.
2. S. R. Y. Cajal, in Degeneration and Regenera-
tion of the Nervous System, R. M. May, Ed.
(Oxford Univ. Press, London, 1928).
3. A. J. Aguayo et al., Neurosci. Lett. 9, 97 (1978);
E. L. Weinberg and P. S. Spencer, Brain Res.
162, 273 (1979); L. J. Stensaas, P. R. Burgess,
K. W. Horch, Neurosci. Abstr. 5, 684 (1979); C.
S. Perkins, T. Carlstedt, K. Mizuno, A.
Aguayo, Can. J. Neurol. Sci. 7, 323 (1980).
4. P. M. Richardson, U. M. McGuinness, A. J.
Aguayo, Nature (London) 284, 264 (1980).
5. S. Varon and R. P. Bunge, Annu. Rev. Neur-
osci. 1, 327 (1978); Y.-A. Barde, D. Edgar, H.
Thoenen, Proc. Natil. Acad. Sci. U.S.A. 77.
6. M.-M. Mesulam, J. Histochem. Cytochem. 26,
7. M. Palkovits and D. M. Jacobowitz, J. Comp.
Neurol. 157, 29 (1974).
8. A. J. Aguayo, G. M. Bray, C. S. Perkins, I. D.
Duncan, Soc. Neurosci. Symp. 4, 361 (1979).
9. This work was supported by grants from the
Medical Research Council of Canada, the Mus-
cular Dystrophy Association of Canada, and the
Multiple Sclerosis Society of Canada.
* Address reprint requests to A.J.A., Neurosci-
ences Unit, Montreal General Hospital, 1650
Cedar Avenue, Montreal, Quebec, Canada H3G
13 April 1981; revised 14 July 1981
neuron and the diameter of its motor
axon (4). The orderly recruitment of mo-
tor units that prevails when the muscle is
used as a prime mover undergoes signifi-
cant changes when the same muscle con-
tracts as a synergist in another move-
ment. In contrast to the concept of a
fixed recruitment order, it has been oc-
casionally reported that human subjects,
when provided with visual or auditory
feedback from their active motor units,
can learn to voluntarily activate or sup-
press any arbitrarily chosen motor unit
Copyright© 1981 AAAS
Spinal Motoneuron Recruitment in Man: Rank Deordering
with Direction but Not with Speed of Voluntary Movement
Abstract. Single motor units in human interosseous muscle are recruited in order
from small to large in slow or brisk voluntary abduction ofthe indexfinger. When the
same muscle acts as a synergist as opposed to a prime mover, about 8 percent ofthe
unit pairs consistently reversed their recruitment order. Motor commands appear to
be patterned in terms of movements rather than muscles and to involve different
connectivities to the motoneuron pool ofa muscle executing moverments in different
on October 24, 2008