Anat Embryol (1994) 190:1-11

Anatomyand Embryolo 9 Springer-Verlag 1994

Review article

The rostrocaudal organization in the dorsal root ganglia of the rat: a consequence of plexus formation? W.J.To Wessels, H.K.P. Feirabend, E. Marani Neuroregulation Group, Department of Physiology,Universityof Leiden, P.O. Box 9604, NL-2300 RC Leiden,The Netherlands Accepted: 10February 1994

Abstract. The dorsal root ganglia (DRGs) of the rat have a rostrocaudal organization. This organization can most easily be demonstrated in fetal and neonatal rats because the spatial relationships of their DRGs are maintained better in tissue sections than those of mature rats. This review is concerned with the way in which the rostrocaudal organization of the DRGs is generated. Wheat germ agglutinin - horseradish peroxidase/horseradish peroxidase labeling of peripheral nerves of the brachial and lumbar plexuses shows that the position of the somata of the sensory neurons of the labeled nerves can be restricted to rostral or caudal halves of DRGs. Labeling of the thoracic nerve or its branches always results in labeling throughout the entire thoracic DRG. After application of the marker to forelimb nerves, it was observed that whenever a DRG is labeled only partially, its spinal nerve is correspondingly labeled partially as well. These data suggest that the rostrocaudal organization in the DRG is related to the formation of the plexuses. During development nerve fibers can be segmentally labeled, using the subdivision of the DRGs into a rostral and a caudal half to keep together as they find their way through the plexus. Application of label to forelimb skin, hindlimb skin and even thoracic skin can result in labeling of rostral or caudal halves of a DRG. A possible explanation might be that each dermatome can be divided into a skin area innervated by the rostral half of a DRG and a skin area innervated by the caudal half of the same dorsal root ganglion. In the rat, the segmental sensory innervation of muscles during development has not yet been investigated. The question of whether the segmental unit of innervation of a muscle is a whole DRG or half a DRG therefore still remains unanswered. Key words: Dorsal root ganglia - Somatotopy - Development - Plexusformation - Sensory innervation

Correspondence to: W.J.T. Wessels

Introduction The projections of the primary afferent fibers are somatotopically organized in the rat dorsal horn (Arvidsson and Pfaller 1990; Castro-Lopes and Coimbra 1991; Devor and Claman 1980; LaMotte et al. 1991; Molander and Grant 1985, 1986, 1987; Neuheuber and Zenker 1989; Pfaller and Arvidsson 1988; Rivero-Melian and Grant 1990; Smith 1983; Swett and Woolf 1985; Woolf and Fitzgerald 1986; Ygge 1989; Ygge and Grant 1983). In the dorsal root ganglia (DRGs) containing the cell bodies of primary afferent fibers, no somatotopic organization seems to exist. However, recently a rostrocaudal organization in the brachial and lumbar DRGs of fetal and neonatal rats has been demonstrated (Wessels 1991; Wessels et al. 1990 a, b; Wessels and Marani 1993). Injections of wheat-germ agglutinin-horseradish peroxidase (WGA-HRP) into the hindlimb of fetal and neonatal rats resulted in completely and partially labeled DRGs. Partial labeling always concerned the rostral or caudal part of a DRG (Wessels et al. 1990 a, b). Furthermore, labeling of forelimb and hindlimb nerves in neonatal rats showed that the position of the somata of the sensory neurons of the labeled nerves can be restricted to the rostral or caudal half of a DRG (Wessels et al. 1990 a; Wessels and Marani 1993). Labeling of thoracic nerves resulted in labeling of one whole thoracic DRG (Wessels and Marani 1993). The aim of this review is to consider the developmental mechanisms leading to the rostrocaudal organization in the DRGs. How is this rostrocaudal organization in the DRGs generated? This question will be addressed by studying the reliltionship between the rostrocaudal organization in the DRGs and the development of the peripheral innervation pattern. This relationship will be considered in three sections: the nerves, the muscle afferents and the skin afferents. Because most of our data are gathered around the hindlimb, the sensory innervation of the hindlimb is extensively discussed. Thereafter, the sensory innervation of the forelimb is addressed and additional information is provided. The third topic is the sensory

innervation of the thoracic region. This region has its own aspects because no plexuses are involved and the basic segmentation of the body is preserved (Smith 1983, 1986; Ygge 1984; Ygge and Grant 1983). This review will start with a critical examination of the methods used to reveal the rostrocaudal organization in the DRGs during development.

dorsal horn laminae I and II (Nyberg and Blomqvist 1985; Swett and Woolf 1985). WGA-HRP therefore seems to be selectively taken up by fine primary afferents. With free HRP the deeper layers are depicted more strongly than when WGA-HRP is applied (Molander and Grant 1986, 1987), but there is no real difference in the topographic distribution of the labeling in the dorsal horn (Molander and Grant 1986, 1987; Ygge and Grant

19831. Methodological considerations WGA-HRP injections into the limb Evidence for a rostrocaudal organization was acquired after WGA-HRP injections into the hindlimb of rat fetuses (Wessels 1991; Wessels et al. 1990 a, b). Nerve endings and damaged fibers at the injection site take up WGAHRP and transport it (Mesulam 1982). The damaged fibers may innervate the injection site, but may also pass through the injection site to innervate more distal targets. This means that after a proximal injection in the limb labeling of the DRGs may be more extensive than is actually representive for the injected area. Therefore, the precise segmental innervation pattern of proximal injection sites cannot be determined using the WGA-HRP tracing method. Intact myelinated fibers, passing through the injection site, do not take up and transport WGAHRP (Mesulam 1982). However, intact unmyelinated axons can take up WGA-HRP by pinocytosis (La Vail and La Vail 1974). In fetal rats the peripheral nerves are not yet myelinated, as myelination of the rat sciatic nerve starts around birth (Friede and Samorajski 1968; Sima 1974). The presence of unmyelinated fibers may therefore be a second factor obscuring the exact segmental innervation of a more proximal injection site. The more distal the injection site in the limb, the more accurately can the segmental sensory innervation be determined. A WGAHRP injection into the limb labels efferent and afferent nerves of both muscle and skin. Thus, the resultant labeling pattern in the DRGs can be attributed to both muscle and skin afferents from the injection site.

WGA-HRP/HRP labeling of peripheral nerves After application of HRP or WGA-HRP to a dissected peripheral nerve, the exact segmental sensory origin of the nerve can be determined. Depending on the targets of the labeled nerve, the resultant labeling pattern in the DRGs can be attributed to muscle afferents, skin afferents or both. The labeling pattern found after application of free HRP to the cut end of a peripheral nerve depends on the postoperative survival time (Grant 1993). After a short survival time (24 h), mainly small DRG-cells and the superficial laminae I and II of the dorsal horn are labeled, suggesting labeling of fine-caliber primary afferent fibers. With prolonged survival time (72 h), large DRG-cells and deeper laminae of the dorsal horn are also labeled (Grant 1993; Molander and Grant 1986). WGA-HRP application to the cut end of a peripheral nerve results in a preferential labeling of the superficial

The segmental sensory innervation of the hindlimb during development Data relating to the hindlimb WGA-HRP injections into the hindlimb of fetal and neonatal rats result in completely and partially labeled DRGs. Partial labeling of a DRG nearly always involved the caudal half of the rostralmost labeled DRG or the rostral half of the caudalmost labeled DRG (Wessels et al. 1990 a). Reconstructions of the labeled nerves from sagittal sections after injections into the hindlimb of rat fetuses suggested that the somata of the sensory neurons belonging to a particular nerve can be restricted to the rostral or caudal half of a DRG. Labeling of the proximal ends of the cut sciatic and saphenous nerves in neonatal rats showed that during development the sensory part of the sciatic nerve is derived from the caudal half of DRG L4, whole L5 and the rostral half of DRG L6. The saphenous nerve, a branch of the femoral nerve, is derived from whole DRG L3 and the rostral half of L4 (Wessels et al. 1990a). Therefore, the rostrocaudal organization in the DRGs seems to be intimately related to the distribution pattern of the peripheral nerves. The completely and partially labeled DRGs are also related in a special manner to injected areas of the hindlimb. These areas contain both muscle and cutaneous afferents and the resultant labeling pattern in the DRGs must be attributed to both. From the WGA-HRP injection experiments, it was deduced that the ventral side of the lower limb, but not the foot, is innervated by the caudal half of DRG L3 and the rostral half of DRG L4. The dorsal side of the lower limb (but not the foot) is innervatedby the caudal half of DRG L4, whole L5 and the rostral half of L6. The entire lower limb as well as the foot are innervated by the caudal half of DRG L3, whole L4, whole L5 and the rostral half of L6 (Wessels et al. 1990 b), so there is a somatopic representation of the hindlimb over the DRGs. Recent experiments have shown that labeling of the hindlimb skin in neonatal rats also results in completely and partially labeled DRGs. For example, labeling of the skin on the ventrolateral side of the upper hindlimb resulted in dense labeling of the caudal half of DRG L2 and a few labeled cells in the caudal half of DRG L1 and the rostral half of DRG L3 (unpublished observations).

The rostrocaudal organization in the DRGs related to the peripheral nerves It is well established that the formation of the spinal nerves and D R G s is imposed by the somites (Keynes and Stern 1984; Teillet et al. 1987)). The sclerotome part of the somite can be divided into a rostral and a caudal half. Migrating neural crest cells and motor axons emerging from the neural tube traverse only the rostral half of each sclerotome (Bronner-Fraser 1993; Keynes and Stern 1984; Remak 1855; Rickman et al. 1985; Verbout 1985; Teillet et al. 1987). Even if a sclerotome or a piece of the neural tube is reversed before axonal outgrowth, the motor axons still grow into the original rostral half of the sclerotome (Keynes and Stern 1984). Therefore, the segmentation of the peripheral nervous system seems to be due to differences between the rostral and caudal halves of the sclerotome part of the somite. How is the division of the D R G s in rostral and caudal halves, as described above, brought about? The whole hindlimb is innervated exclusively by the ventral rami of the spinal nerves (Bolk 1910; Gray et al. 1980). These ventral rami can be subdivided again into ventral and dorsal divisions (Bolk 1910; Gray et al. 1980). The obturator and tibial nerves are formed by the union of ventral divisions of the ventral rami. The femoral and c o m m o n peroneal nerves are composed of dorsal divisions of the ventral rami. These peripheral limb nerves contain both muscle and cutaneous afferents. It might be that the somata of the sensory neurons of the ventral

~t

DRG1.2L3

II

ventraldivision

[~

dorsaldivision

L4

L5

d nerve

k6

ti~bialnerve~common peronealnerve Fig. 1. Schematic drawing illustrating the hypothetical position of the somata of the sensory neurons of the ventral divisions of the spinal nerves in the rostral halves of the dorsal root ganglia (DRGs) and the somata of the dorsal divisions in the caudal halves of the D R G s (Wessels 1991). However, this hypothesis is considered incorrect in this paper (see text). In this figure the femoral and obturator nerves are derived from the D R G s L2, L3 and L4. The tibial and common peroneal nerves are derived from the D R G s L4, L5 and L6 (Greene 1955)

divisions of the ventral rami are situated in the rostral half of a D R G while the somata of the dorsal divisions are situated in the caudal half of a D R G (Fig. 1). However, this hypothesis seems to be incorrect, since the somata of the neurons of the saphenous nerve, a cutaneous branch of the of dorsal divisions composed of femoral nerve, are situated in whole D R G L3 and the rostral half of D R G L4 instead of the caudal halves of both D R G L3 and D R G L4. Although the rostral or caudal halves of the lumbar D R G s cannot be attributed to either the dorsal or ventral divisions of the ventral rami proper, it can be concluded that the position of the somata of the sensory neurons of a peripheral nerve can be restricted to the rostral or caudal half of a DRG. This does not exclude the possibility that a whole D R G is labeled after labeling of a peripheral nerve.

The rostrocaudal organization in the DRGs related to the muscle afferents In the developing limb, embryonic ventral and dorsal muscle masses are formed first. The flexor muscles are derived from the embryonic ventral muscle mass and the extensor muscles from the embryonic dorsal muscle mass (Romer 1927; Wortham 1948). It would be interesting to know whether the sensory innervation of these pre-muscle masses also shows a division of the D R G s into rostral and caudal halves. Honig (1982) did not report a preferential labeling of either the rostral or caudal half of a D R G after H R P injections into the embryonic ventral muscle mass of the hindlimb of chick embryos. The embryonic ventral and dorsal muscle masses divide further into muscles of the upper hindlimb, muscles of the lower hindlimb, and intrinsic foot musculature (Hollyday 1980). The labeling of the D R G s after a W G A - H R P injection into the hindlimb can be attributed to both muscle and cutaneous afferents. Which part of the labeled cells in the D R G s is due to labeling of muscle afferents and which part is due to labeling of cutaneous afferents can not be determined with the technique used. Assuming that the data apply to the muscle afferents, there seems to be no difference in the segmental sensory innervation of muscles situated in the lower hindlimb and the segmental sensory innervation of the intrinsic foot musculature. Both are innervated by the caudal half of D R G L3, whole L4, whole L5 and the rostral half of L6 (Wessels et al. 1990 b). However, the border of the segmental innervation between muscles situated in the upper hindlimb and muscles situated in the lower hindlimb and foot seems to lie between the rostral and caudal half of D R G L3 (Wessels et al. 1990b). When the injection site extends beyond the kneejoint, the caudal half of D R G L2 , whole D R G L3, L4, L5 and the rostral half of L6 are labeled (Wessels et al. 1990b). Hypothetically, the border between the sensory innervation of the muscles of the trunk and the hindlimb might be situated rostrally between the rostral and caudal half of D R G L2 and caudally between the rostral and caudal half of D R G L6. It could very well be that for certain muscle groups the border between their segmental innervation can be

~ [RG

DRG

1

DRG

\ ~

1~

~

\

2

3 a

b Fig. 2a--e. The segmental sensory innervation of a muscle (M) (Wessels 1991). In a and b the segmental unit of innervation is half a DRG. An example with four DRG-halves is given, a The whole muscle is innervated by the caudal half of DRG 1, whole DRG 2 and the rostral half of DRG 3. b The muscle is separated into four parts. Each part is innervated by its own DRG half. e The segmen-

c tal unit of innervation is the caudal half of one DRG together with the rostral half of the next DRG. The whole muscle is innervated by two segmental units; the caudal half of DRG 1 together with the rostral half of DRG 2, and the caudal half of DRG 2 together with the rostral half of DRG 3

drawn in the middle of the ganglion dividing the ganglion into a rostral and a caudal half. An individual muscle is predominantly innervated by two or three dorsal roots (Dykes and Terzis 1981; Honig 1982). Peyronnard et al. (1986) have even reported that for each muscle in the rat hindlimb most sensory neurons are located in only one of the lumbar DRGs, with the rest in one and occasionally two adjacent DRGs. A D R G contains the sensory neurons of several muscles (Peyronnard et al. 1986). On the basis of our labeling results, it is proposed that the segmental unit of innervation is not a whole D R G but one half of a D R G (hypothesis 1), or the combination of the caudal half of one D R G with the rostral half of the next D R G (hypothesis 2). According to the first hypothesis, depending on the extent of the segmental innervation, a muscle could be innervated by one to six D R G halves (Fig. 2 a; an example with four D R G halves is given) or the muscle could be separated into parts, each part innervated by its own D R G half (Fig. 2 b). According to the second hypothesis, the sensory innervation of a muscle would always come from the caudal half of the rostralmost D R G through the rostral half of the caudalmost D R G (Fig. 2 c). However, not every W G A - H R P injection into the hindlimb of fetal or neonatal rats resulted in partially labeled DRGs. Therefore, the second hypothesis seems unlikely. An individual hindlimb muscle is derived from the myotome part of several adjacent somites. There is a clear correlation between the somitic level of origin of a muscle and its segmental innervation (Lance-Jones 1988). In the mature animal, a myotome is defined as a muscle region served by one spinal nerve. In line with the correlation mentioned above, the myotome in the mature animal could, as proposed by Bolk (1910), enclose a muscle region containing cells derived from the myotome part of a particular somite and served by a spinal nerve of which the segmental level corresponds to that somite. A dermatome is defined as a cutaneous region served by one dorsal root. The boundaries of the dermatome and afferent myotome of the same dorsal root do not necessarily overlap. Moreover, the region of the afferent myotome is generally wider than the region of the matching dermatome (Dykes and Terzis 1981). In the following section, only the cutaneous afferents are discussed.

The rostrocaudal organization in the DRGs related to the cutaneous afferents The dermatomes of adjacent dorsal roots progress in a serially overlapping order along the limb (Bolk 1910; Dykes and Terzis 1981; Sherrington 1893; Werner and Whitsel 1967). Even the cutaneous fields of the rootlets of a dorsal root are arranged within the dermatome in serial overlapping order (Werner and Whitsel 1967). Burton and McFarlane (1973) demonstrated that the receptive fields of the cells within the D R G L7 of the cat showed this same serial overlapping order when the cells were traced from medial (caudal) to lateral (rostral) in the ganglion. Thus, each mediolateral (rostro-caudal) part of a D R G may have its own receptive skin field. A dermatomal map of the rat has not so far been published in the literature. W G A - H R P injections into the hindlimb label both muscle and skin afferents. Assuming that our results apply to skin afferents, a dermatome pattern of the hindlimb as illustrated in Fig. 3 is proposed. Each dermatome is divided into a skin area innervated by the rostral half of a D R G and a skin area innervated by the caudal half of the same DRG. Recent skin-labeling experiments provide evidence that this is indeed a plausible option. In the proposed model, the skin of the whole lower hindlimb is innervated by the caudal half of D R G L3, the whole of L4 and L5 and the rostral half of L6 (see Fig. 3) The skin of the dorsal part of the lower hindlimb is innervated by the caudal half of D R G L4, the whole of L5 and the rostral half of L6. In the proposed dermatome pattern in the rat, the hindlimb dermatomes, when traced rostro-caudally, progress from proximal to distal along the ventral side of the limb, cross the foot from medial to lateral, and ascend along the dorsal side of the limb towards the tail. This sequence of the dermatomes along the hindlimb conforms to that in other species (man, Bolk 1910; cat, Kuhn 1953; squirrel monkey, Werner and Whitsel 1967; chick, Scott 1982). The overlap of sequential dermatomes (Fig. 3) is also consistent with the literature (Dykes and Terzis 1981; Scott 1982; Werner and Whitsel 1967). As shown in the proposed dermatome pattern in the rat (Fig. 3), the distal dermatomes have lost their connection with the dorsal midline. Bolk (1910) put forward the

R

I C

L4

L5

R

I C

R - rostral

C - caudal

M - med/sl

L-lateral

V - venial

D - dorsal

f~

skin area innervated by the rostral half of the DRG skin area innervated by the caudal half of the DRG

Ua

~

1.5

L6

b Fig. 4a, b. Development of the dermatomes according to Bolk (1910), transposed to the rat hindlimb (Wessels 1991). a At first, when the limb bud is just a flat fold, the dermatomes progress in an orderly way from rostral to caudal and all dermatomes have connections with the dorsal midline, b During the growth of the limb, dermatomes connected to the outgrowing limb bud lose their connection with the midline

following developmental theory to explain this phenomenon. The limb bud is at first just a flat fold. In this limb bud the dermatomes progress in an orderly sequence from rostral to caudal, and all dermatomes have connections with the dorsal midline (Fig. 4). During the growth of the limb, the rostral dermatomes shift to the proximal ventral surface of the limb. The middle derm a t o m e s lose their connection with the midline and become located at the distal part of the limb. The caudal d e r m a t o m e s shift over the dorsal surface of the more proximal limb. Although this description of the development of the dermatomes by Bolk (1910) m a y explain the evolution of the d e r m a t o m a l map, it does not describe the underlying fetal development of the dermatomes. Scott (1982) demonstr~/ted in the chick that during devel-

Fig. 3. Theoretical dermatome pattern as proposed for the rat hindlimb (Wessels 1991). The dermatomes are divided into a skin area innervated by the rostral half of a DRG and a skin area innervated by the caudal half of the same DRG

opment the sensory axons which innervate distal skin in the mature chick do bypass nearby skin on the thigh from the outset to innervate more distal skin. Skin does not migrate distally on the limb after being innervated by sensory axons. Skin sensory axons from each D R G grow directly to their target skin and establish their derm a t o m e precisely at its characteristic location (Scott 1982). However, it could very well be that, as proposed by Cole et al. (1968), each D R G innervates skin derived embryologically from the same segmental level, and that consequently skin movements occur before the innervation is established.

The segmental sensory innervation of the forelimb during development

Data relating to the forelimb W G A - H R P or W G A - H R P / H R P injections into the forepaw of neonatal rats gave rise to completely and partially labeled D R G s . Partial labeling always involved the caudal or rostral halves of the rostral-most or caudalmost labeled D R G , respectively. As in the hindlimb, it is possible to relate injected areas of the forepaw to completely and partially labeled D R G s . It was found that the whole forepaw is innervated by the caudal half of D R G C6 through the rostral half of D R G T1. Digit V is mainly innervated by the caudal half of D R G C8 and the rostral half of D R G T1. Digit I and digit II are mainly innervated by whole D R G C7 and D R G C8 (Wessels and Marani 1993). Labeling of the forelimb skin in neonatal rats m a y also result in half-labeled D R G s (unpublished observations). For example, labeling of the skin dorsally on the ulnar side of the forepaw gave rise to labeled cells in the

)[ii ili!i! C7

!I

~I.IIi!:g5511ilII ~%I!!IiliI~]iI IIilii~IIilIiI,,'II~IIIiI~I~,~; I

C7

128

c8

"1"1

musculocutaneotts

nerve TI

median nerve

labeled neurons and fibers of the median nerve

ulnar nerve

Fig. 5 Schematic drawing illustrating the WGA-HRP/HRP labeling pattern in the spinal nerves and DRGs after labeling of the cutaneous branch of the musculocutaneous nerve, the median nerve and the ulnar nerve in neonatal rats caudal half of D R G C8 and the rostral half of D R G T1. Labeling of forelimb nerves in neonatal rats also resulted in whole labeled D R G s and labeled rostral or caudal halves of D R G (Wessels and Marani 1993). It was found that the cutaneous branch fibers of the musculocutaneous nerve arise from cell bodies in whole D R G C6 and the r o s t r a ! h a l f of C7. Labeling of the median nerve at the wrist gave rise to dense labeling of the caudal half of D R G C7 and the rostral half of D R G C8. Labeling of the ulnar nerve halfway down the lower forelimb resulted in dense labeling in the caudal half of D R G C8 and in the rostral half of T1 (Fig. 5). Recent experiments have shown that in neonatal rats, in accordance with the adult (Ygge 1989), the s o m a t a of the sensory neurons of both the muscular and cutaneous branch of the radial nerve are located in whole D R G C7 and D R G C8. A few som a t a of the radial nerve are situated in the rostral half of D R G T1 (unpublished observations).

The rostrocaudal organization in the DRGs related to the peripheral nerves Labeling of the musctdocutaneous, median and ulnar nerves resulted in whole labeled D R G s and labeled rostral or caudal halves of D R G s . The musculocutaneous, median and ulnar nerves are all formed by the union of ventral divisions of the ventral rami of the spinal nerves (Bolk 1910; G r a y et al. 1980; Greene 1955). The alternating position of the sensory cetl bodies of these forelimb nerves in the D R G s is striking (Fig. 5) and suggests a function of the subdivision of the D R G into a rostra1 and

Fig. 6. Reconstruction of the labeled spinal nerves and DRGs in a P4 rat from sagittal sections. Application of WGA-HRP/HRP to the median nerve at the wrist resulted in labeling of the caudal half of spinal nerve C7 and DRG C7 and the rostral half of spinal nerve C8 and DRG C8 caudal half in the formation of the brachial plexus. The rostrocaudal organization in the brachial D R G s continues in the spinal nerve; whenever a D R G is labeled partially, its spinal nerve is labeled partially as well (Fig. 5)(Wessels and Marani 1993). On the basis of these observations, one can speculate that the formation of the brachial plexus is a very orderly process. During development, the fibers m a y be segmentally labeled, using a subdivision into rostra1 and caudal segmental halves to find their way through the plexus. For example our experiments show that the median nerve arises from the caudal half of D R G C7 and the rostral half of C8. The labeled fibers of the spinal nerve C7 and those of C8 remain together within the spinal nerves (Fig. 6), eventually uniting to form the median nerve. Thus, fibers of a spinal nerve which will eventually make up a forelimb nerve m a y stay together while growing in the plexus. In agreement with this, Scott (1982) used H R P injections into lumbosacral D R G s of chick embryos to show that axons from one D R G tend to stay together in a cutaneous nerve trunk. The radial nerve arises from the union of dorsal divisions of the ventral rami (Bolk 1910; G r a y et al. 1980; Greene 1955). After labeling of the ~radial nerve, no rostrocaudal organization in the D R G s was found. In contrast to the nerves formed by the union of the ventral divisions of the ventral rami, the radial nerve does not seem to need a subdivision in the D R G s into rostral and caudal halves to find its way through the plexus. To be segmentally "labeled" seems enough to be guided through the plexus and form the radial nerve.

The rostrocaudal organization in the DRGs related to the muscle afferents From the hindlimb study, assuming that the injection experiments apply to muscle afferents, it was concluded that for certain muscle groups the border between their segmental innervation can be drawn in the middle of the ganglion. Data relating to the sensory innervation of the forelimb are sparse, but assuming that the results of tracer injections into the forepaw can also be applied to muscle afferents, the observation that the whole forepaw is innervated by the caudal half of D R G C6 through the rostral half of T1 lends support to this conclusion. The muscular branch of the radial nerve is the only muscle nerve we investigated. Labeling of this nerve branch did not result in labeling of the rostral or caudal half of a DRG. Therefore, there is no direct evidence that the segmental unit of innervation of a forelimb muscle is half a D R G instead of a whole D R G (Fig. 2). However, this theory is not rejected because labeling of the cutaneous branch of the radial nerve did not result in half labeled DRGs.

The rostrocaudal organization in the DRGs related to the cutaneous afferents Labeling of the cutaneous branch of the radial nerve did not result in half-labeled DRGs, but labeling of the cutaneous branch of the musculocutaneous nerve did. The latter resulted in labeling of whole D R G C6 and the rostral half of D R G C7. Labeling of the forelimb skin results in half labeled DRGs. The theoretical model in which each dermatome can be divided into a skin area innervated by the rostral half of a D R G and a skin area innervated by the caudal half of the same D R G is therefore still a good option (Fig. 3).

observations). The intercostal nerve of T1 was labeled and the branch of T1 that takes part in the brachial plexus was unlabeled, suggesting that the intercostal nerve of T1 arises from the caudal half of D R G T1. In neonatal rats, labeling of the intercostal nerves (Wessels and Marani 1993) (perhaps with the exception of the intercostal nerves T1 and T2) or the lateral cutaneous branch of the dorsal ramus of the thoracic nerve (n -- 3)(unpublished observations) always resulted in labeling of one whole thoracic DRG.

The rostrocaudal organization in the DRGs related to the peripheral nerves In the thoracic region, segmentation is preserved (Smith 1983, 1986; Ygge 1984; Ygge and Grant 1983). In the adult rat no somatotopic organization was found in the thoracic D R G s after labeling various thoracic spinal nerve branches (Ygge 1984). The experiments of Smith (1983) and our own (Wessels and Marani 1993; unpublished observations) in postnatal rats confirm these findings. The intercostal nerve of T1 was never labeled. Maybe labeling of this nerve results in labeling of the caudal half of D R G T1, as suggested by the results of labeling the skin. However, D R G T1 and T2 are considered to be exceptions because they are also partially involved in the brachial plexus in the rat (Greene 1955). So, in contrast to the forelimb and hindlimb nerves, there is no evidence for a restriction of the position of the somata of the sensory neurons of the thoracic nerve or its branches to either the rostral or caudal half of a DRG. These findings support the hypothesis that the rostrocaudal organization in the brachial and lumbar D R G s during development is related to the formation of the plexuses.

The rostrocaudal organization in the DRGs related to the muscle afferents

The segmental sensory innervation of the thoracic region during development Data relating to the thoracic region After W G A - H R P / H R P injections into the thoracic wall of neonatal rats, the labeled cells were found evenly distributed within the thoracic D R G s most of the time. However, in one case a subdivision in the labeling pattern of a midthoracic D R G was noticed. The dorsorostral corner of this thoracic D R G remained unlabeled (Wessels and Marani 1993). Labeling of the thoracic skin in neonatal rats (n = 20) frequently resulted in whole labeled thoracic D R G s without evidence for a somatotopic localization. However in a few cases (n = 4) a clear preference of the labeled cells for either the rostral or caudal half of a thoracic D R G was found (unpublished observations). Labeling of skin on the back just below the forelimb in combination with a piece of skin of the caudal side of the upper forelimb resulted in labeling of the caudal half of D R G C6, the caudal half of D R G T 1 , whole D R G T2 and the rostral half of D R G T4 (unpublished

W G A - H R P injections in the thoracic wall probably label both muscle and cutaneous afferents. In one case, a subdivision in the labeling pattern of a midthoracic D R G was noticed after a W G A - H R P injection into the thoracic wall of a neonatal rat. The dorsorostral corner of the D R G remained unlabeled. Assuming that labeling pattern can be attributed to muscle afferents, a more refined myosomatotopic organization in the thoracic D R G than the subdivision into rostral and caudal halves is indicated. This could even mean that the segmental unit of innervation of a muscle is neither a whole D R G nor half of a D R G (Fig. 2), but rather a quarter of a DRG.

The rostrocaudal organization in the DRGs related to cutaneous afferents Labeling of the thoracic skin in neonatal rats sometimes resulted in labeling of the rostral or caudal half of a DRG. These half labeled D R G s were not, as in case of the forelimb and hindlimb, the rostralmost or caudal-

occur. We therefore propose that a thoracic d e r m a t o m e can be divided into a skin area innervated by the rostral half of a D R G and a skin area innervated by the caudal half of a D R G , as presented in the model in Fig. 7. In this model the thoracic dermatomes do not overlap. In the initial innervation of the rat thoracic skin, d e r m a t o m e boundaries m a y be strictly without overlap, while later on in development the primary afferent fibers sprout into adjacent skin, resulting in overlap of the dermatomes. However, the results of Scott (1982), who showed that the location and a m o u n t of overlap of dermatomes in the chick embryo are similar to those in the mature pattern, are at variance with our proposal. skin area innervated by the rostral half of a DRG skin area innervated by the caudal half of a DRG

Fig. 7. Theoretical dermatome pattern in the rat thoracic region. Only the dermatomes Th 1 to Th 7 are presented. The dermatomes are divided into a skin area innervated by the rostral half of a DRG and a skin area innervated by the caudal half of the same DRG

most labeled D R G (unpublished observations). Labeling of the rostral or caudal half of a thoracic D R G was not correlated with labeling of the skin of the dorsal or ventral side of the neonatal rat. Labeling of the skin of the ventral or dorsal midline should have resulted in labeling of only rostral or caudal D R G halves, which did not

The segmental sensory innervation during adulthood Until now there has been no evidence for a somatotopic organization within the D R G s of the adult rat. Labeling of a peripheral nerve (Molander and G r a n t 1986; Peyronnard et al. 1986; Wessels et al. 1990 a; Ygge 1984, 1989), muscle (Myles et al. 1992; Peyronnard et al. 1986) or skin (Molander and G r a n t 1985) in the adult rat never resulted in a clear somatotopic arrangement of the labeled cells within the D R G s . However, recent forelimb nerve labeling experiments that showed a rostrocaudal organization in the D R G s of the adult rat confirm the rostrocaudal organization in the D R G s of neonatal rats (Fig. 8, Table 1). Application of

Fig. 8. A 40 gm section through the DRG C7 of an adult rat, cut horizontally along the long axis of the spinal cord, showing label in the rostral half of DRG C7 after WGA-HRP/HRP labeling of the median nerve at the wrist, x 92

Table 1. Labeling of the dorsal root ganglia (DRGs) following labeling of the forelimb nerves in the adult rat Series Tracer

Nerve

C6a

C7

C8

T1

RCRCRCRC C5188 C5189 C5194 C5189 C5194

WGA-HRP/HRP WGA-HRP/HRP WGA-HRP/HRP WGA-HRP/HRP WGA-HRP/HRP

Median Median Median Ulnar Ulnar

* + + * * * + + + * * * + + + + * * + +

+ a densely labeled half of a DRG * a few labeled cells within this half of the DRG a The DRGs are divided into a rostral (R) and a caudal (C) half

H R P / W G A - H R P to the proximal stump of the median nerve at the wrist frequently resulted in labeling of the caudal half of D R G C7 and the rostral half of D R G C8. Labeling of the ulnar nerve half-way down the lower forelimb often resulted in labeling of the caudal half of D R G C8 and the rostral half of D R G T1. Similarly, Baron et al. (1988) found that after application of H R P to the central end of rat hindlimb nerves, when two adjacent D R G s were labeled, the labeled somata of the more cranial D R G were mainly located in the mediocaudal part of the D R G , and the cells of the more caudal D R G were mainly located in the laterorostral part. A prerequisite for the demonstration of the rostrocaudal organization in the D R G s is the preservation of the topographic relationships of D R G s during the embedding process. This preservation can be achieved by leaving the D R G s attached to both spinal cord and peripheral nerve plexus. Even then, it cannot be excluded that D R G s are turned and twisted in such a way that the sequence of the dorsal rootlets is disturbed and that the dorsal rootlets and the D R G are not in the same sectioning plane. Consequently, the rostrocaudal organization in the D R G cannot be accurately observed. W o o d b u r y and Scott (1991) also found a rostrocaudal organization in the D R G s of hatchling chicks after labeling of cutaneous hindlimb nerves, although no detailed topographical study of the labeled neurons in the D R G s was conducted. These authors also used tissue pieces in which the D R G s were still attached to the spinal cord. In line with our results in the adult rat, Kausz and R~thelyi (1985) showed in the cat that subcutaneous H R P injections in the forelimb and flank but not H R P injections in the muscles of the forelimb frequently resulted in labeled somata of sensory neurons distributed in elongated proximodistal territories in the D R G s . McLachlan and J/inig (1983) also found that D R G cells of cats labeled from skin but not from muscle nerves had charateristic localizations. Furthermore, Kausz and R6thelyi (1985) found by H R P labeling of the dorsal rootlets that the somata of the sensory neurons located in the lateral (rostral) part of a D R G project in the rostral rootlets of the dorsal root, while the somata of the sensory neurons in the medial (caudal) part of a D R G project in the caudal rootlets of the dorsal root. Electrophysiologically, Burton and McFarlane (1973) had found this

same arrangement in the D R G L7 of the cat. Therefore, it can be concluded that in the adult rat there exists a rostrocaudal organization in the D R G s .

Conclusions In the rat there is a rostrocaudal organization in the D R G s during development. An overview of this organization, as found by investigating body area, peripheral nerves, muscle afferents and cutaneous afferents, is given in Table 2. The organization in the D R G s is especially evident after labeling peripheral nerves of the brachial and lumbar plexuses. Not every peripheral nerve belonging to the plexuses reveals a rostrocaudal organization in the D R G s (for example, the radial nerve which is composed of dorsal divisions of the ventral rami). Labeling of the branches of the intercostal nerve resulted in the labeling of one whole thoracic D R G . After labeling of the forelimb nerves, it was observed that whenever a D R G is labeled partially, its spinal nerve is labeled partially as well. These data provide good evidence for the hypothesis that the rostrocaudal organization in the D R G s is related to the formation of the plexuses. The rostral and/ or caudal half of a D R G with its connecting fibers m a y be looked upon as a separate unit. During development, nerve fibers m a y be segmentally labeled, using the subdivision in the D R G s in a rostral and a caudal half to stay together as they find their pathway through the plexus. Labeling of the branches of the intercostal nerve never revealed a rostrocaudal organization in the thoracic D R G s , but labeling of thoracic skin sometimes resulted in labeling of a rostral or caudal half of a thoracic D R G . A possible explanation might be that each d e r m a t o m e can be divided into a skin area innervated by the rostral half of a D R G and a skin area innervated by the caudal half of the same D R G . The same model could be applied to the forelimb and hindlimb. Labeling of the skin on the limbs of neonatal rats confirmed that an area of the forelimb or hindlimb skin can be innervated by the rostral or caudal half of a D R G . Electrophysiological experiments are required to refine this observation. In the rat, the segmental sensory innervation of muscles during development has not yet been investigated. Whether the segmental unit of innervation of a muscle during development is a whole D R G or half a D R G is therefore still an open question. Arguments in favor of Table 2. The rostrocaudal organization in the rat dorsal root ganglia (DRGs) during development

Body area Peripheral nerve Muscle afferents Cutaneous afferents

hindlimb

forelimb

thoracic region

+ + ? +

+ + ? +

-* ? +

+ Presence of a rostrocaudal organization in the DRGs - absence of a rostrocaudal organization in the DRGs ? no convincing data available * in one case the dorsorostral corner of a DRG remained unlabeled

10 the s e g m e n t a l unit of h a l f a D R G h a v e been p r e s e n t e d in this review (see the h e a d i n g s c o n c e r n i n g the m u s c l e afferents of the f o r e l i m b a n d hindlimb). In the a d u l t r a t the r o s t r o c a u d a l o r g a n i z a t i o n in the D R G s is still p r e s e n t b u t h a r d to d e m o n s t r a t e after labeling e x p e r i m e n t s b e c a u s e of the difficulty of p r e s e r v i n g the s p a t i a l r e l a t i o n s h i p s of the D R G . This p r e s e r v a t i o n is easier in n e o n a t a l rats t h a n in m a t u r e rats b e c a u s e nerve plexus, muscles, D R G s , v e r t e b r a l c o l u m n a n d spinal c o r d can be e m b e d d e d a n d s e c t i o n e d as a whole.

References Arvidsson J, Pfaller K (1990) Central projections of C4-C8 dorsal root ganglia in the rat studied by anterograde transport of WGA-HRP. J Comp Neurol 292:349-362 Baron R, J/inig W, Kollmann W (1988) Sympathetic and afferent somata projecting in hindlimb nerves and the anatomical organization of the lumbar sympathetic nervous system of the rat. J Comp Neurol 275:460-468 Bolk L (1910) De segmentale innervatie van romp en ledematen by den mensch. De Erven F Bohn, Haarlem Bronner-Fraser M (1993) Mechanisms of neural crest cell migration. Bioessays 15:221-230 Burton H, McFarlane JJ (1973) The organization of the seventh lumbar spinal ganglion of the cat. J Comp Neurol 149:215-232 Castro-Lopes JM, Coimbra A (1991) Spinal cord projections of the rat main forelimb nerves, studied by transganglionic transport of WGA-HRP and by the disappearance of acid phosphatase. Brain Res 542:187-192 Cole JP, Lesswing AL, Cole JR (1968) An analysis of the lumbosacral dermatomes in Man. Clin Orthop 61:241-247 Devor M, Claman D (1980) Mapping and plasticity of acid phosphatase afferents in the rat dorsal horn. Brain Res 190:17-28 Dykes RW, Terzis JK (1981) Spinal nerve distributions in the upper limb: the organization of the dermatome and afferent myotome. Philos Trans R Soc Lond [Biol] 293:509-554 Friede RL, Samorajski T (1968) Myelin formation in the sciatic nerve of the rat. J Neur0Pathol Exp Neurol 27:546-570 Grant G (1993) Projection patterns of primary sensory neurons studied by transganglionic methods: somatotopy and target-related organization. Brain Res Bull 30:199-208 Gray H, Williams PL, Warwick R (1980) Gray's anatomy, 36th edn. Churchill Livingstone, Edinburgh Greene EC (1955) Anatomy of the rat. Hafner, New York Hollyday M (1980) Organization of motor pools in the chick lumbar lateral motor column. J Comp Neurol 194:143-170 Honig MG (1982) The development of sensory projection patterns in embryonic chick hind limb. J Physiol (Lond) 330:175-202 Kausz M, R6thelyi M (1985) Lamellar arrangement of neuronal somata in the dorsal root ganglion of the cat. Somatosens Res 3:193-204 Keynes RJ, Stern CD (1984) Segmentation in the vertebrate nervous system. Nature 310:786-789 Kuhn RA (1953) Organization of tactile dermatomes in cat and monkey. J Neurophysiol 16:169-182 La Vail JH, La Vail MM (1974) The retrograde intra-axonal transport of horseradish peroxidase in the chick visual system: a light and electron microscopic study. J Comp Neurol 157:303 358 LaMotte CC, Kapadia SE, Shapiro CM (1991) Central projections of the sciatic, saphenous, median and ulnar nerves of the rat demonstrated by transganglionic transport of choleragenoidHRP (B-HRP) and Wheat Germ Agglutinin-HRP (WGAHRP). J Comp Neurol 311:546-562 Lance-Jones C (1988) The somitic level of origin of embryonic chick hindlimb muscles. Dev Biol 126:394-407 McLachlan EM, J/inig W (1983) The cell bodies of origin of sympathetic and sensory axons in some skin and muscle nerves of the cat hindlimb. J Comp Neurol 214:115-130

Mesulam MM (1982) Tracing neural connections with horseradish peroxidase. IBRD handbook series. Wiley, Chichester Molander C, Grant G (1985) Cutaneous projections from the rat hindlimb foot to the substantia gelatinosa of the spinal cord studied by trans-ganglionic transport of WGA-HRP conjugate. J Comp Neurol 237:476-484 Molander C, Grant G (1986) Laminar distribution and somatotopic organization of primary afferent fibers from hindlimb nerves in the dorsal horn. A study by transganglionic transport of horseradish peroxidase in the rat. Neuroscience 19:297-312 Molander C, Grant G (1987) Spinal cord projections from hindlimb muscle nerves in the rat studied by transganglionic transport of horseradish peroxidase, wheat germ agglutinin conjugated horseradish peroxidase, or horseradish peroxidase with dimethylsulfoxide. J Comp Neurol 260:246-255 Myles LM, Gilmour JA, Glasby MA (1992) Effects of different methods of peripheral nerve repair on the number and distribution of muscle afferent neurons in rat dorsal root ganglion. J Neurosurg 77:457-462 Neuhuber WL, Zenker W (1989) Central distribution of cervical primary afferents in the rat, with emphasis on proprioceptive projections to vestibular, perihypoglossal, and upper thoracic spinal nuclei. J Comp Neurol 280:231-253 Nyberg G, Blomqvist A (1985) The somatotopic organization of forelimb cutaneous nerves in the brachial dorsal horn: an anatomical study in the cat. J Comp Neurol 242:28-39 Peyronnard JM, Charron LF, Lavoie J, Messier JP (1986) Motor, sympathetic and sensory innervation of rat skeletal muscles. Brain Res 373:288-302 Pfaller K, Arvidsson J (1988) Central distribution of trigeminal and upper cervical primary afferents in the rat studied by anterograde transport of horseradish peroxidase conjugated to wheat germ agglutinin. J Comp Neurol 268:91-108 Remak R (1855) Untersuchungen fiber die Entwicklung der Wirbeltiere. Reimer, Berlin Rickman M, Fawcett JW, Keynes RJ (1985) The migration of neural crest cells and the growth of motor axons through the rostral half of the chick somite. J Embryol Exp Morphol 90:437-455 Rivero-Melian C, Grant G (1990) Distribution of lumbar dorsal root fibers in the lower thoracic and lumbosacral spinal cord of the rat studied with choleragenoid horseradish peroxidase conjugate. J Comp Neurol 299:470-481 Romer AS (1927) The development of the thigh musculature of the chick. J Morphol Physiol 43:347-385 Scott SA (1982) The development of the segmental pattern of skin sensory innervation in embryonic chick hind limb. J Physiol (Lond) 330:203-220 Sherrington CS (1893) Experiments in the examination of the peripheral distribution of the fibers of the posterior roots of some spinal nerves. I. Philos Trans R Soc Lond [Biol] 184:641-763 Sima A (1974) Studies on fibre size in developing sciatic nerve and spinal roots in normal, undernourished and rehabilitated rats. Acta Physiol Scand 406:1-55 Smith CL (1983) The development and postnatal organization of primary afferent projections to the rat thoracic spinal cord. J Comp Neurol 220:29-43 Smith CL (1986) Sensory neurons supplying touch domes near the body midlines project bilaterally in the thoracic spinal cord of rats. J Comp Neurol 245:541-552 Swett JE, Woolf CJ (1985) The somatotopic organization of primary afferent terminals in the superficial laminae of the dorsal horn of the rat spinal cord. J Comp Neurol 220:66-77 Teillet M, Kalcheim C, Le Douarin NM (1987) Formation of the dorsal root ganglia in the avian embryo: segmental origin and migratory behavior of neural crest progenitor cells. Dev Biol 120:329 347 Verbout AJ (1985) The development of the vertebral column. Adv Anat Embryol Cell Biol 90:1-122 Werner W, Whitsel BL (1967) Topology of dermatomal projection in the medial lemniscal system. J Physiol (Lond) 192:123-144

11 Wessels WJT (1991) Development of the peripheral sensory innerration of the rat hindlimb and its central connections: a WGAHRP study. Thesis, Leiden Wessels WJT, Marani E (1993) A rostrocaudal somatotopic organization in the brachial dorsal root ganglia of neonatal rats. Clin Neurol Neurosurg 95: $3-Sll Wessels WJT, Feirabend HKP, Marani E (1990a) Evidence for a rostrocaudal organization in dorsal root ganglia during development as demonstrated by intra-uterine WGA-HRP injections into the hindlimb of rat fetuses. Dev Brain Res 54:273-281 Wessels WJT, Feirabend HKP, Marani E (1990b) Somatotopic organization in the sensory innervation of the rat hindlimb during development, using half dorsal root ganglia as subsegmental units. Eur J Morphol 28:394-403 Woodbury CJ, Scott SA (1991) Somatotopic organization of hindlimb skin sensory inputs to the dorsal horn of hatchling chicks (GaIlus g. domesticus). J Comp Neurol 314:237-256

Woolf CJ, Fitzgerald M (1986) Somatotopic organization of cutaneous afferent terminals and dorsal horn neuronal receptive fields in the superficial and deep laminae of the rat lumbar spinal cord. J Comp Neurol 251:517-531 Wortham RA (1948) The development of the muscles and tendons in the lower leg and foot of chick embryos. J Morphol 83 : 105148 Ygge J (1984) On the organization of the thoracic spinal ganglion and nerve in the rat. Exp Brain Res 55:395-401 Ygge J (1989) Central projections of the rat radial nerve investigated with transganglionic degeneration and transganglionic transport of horseradish peroxidase. J Comp Neurol 279:199-211 Ygge J, Grant G (1983) The organization of the thoracic spinal nerve projection in the rat dorsal horn demonstrated with transganglionic transport of horseradish peroxidase. J Comp Neurol 216:1-9