Three-dimensional computer-aided analysis of the intraganglionic topography of primary muscle afferent neurons in the rat.код для вставкиСкачать
THE ANATOMICAL RECORD 227:405-417 (1990) Three-Dimensional Computer-Aided Analysis of the lntraganglionic Topography of Primary Muscle Afferent Neurons in the Rat JEAN-MARIE PEYRONNARD, JEAN-PIERRE MESSIER, MARTIN DUBREUIL, LOUISE CHARRON, AND FRANCE LEBEL Centre de Recherche en Sciences Neurologiques and Hbpital Hbtel-Dieu, Universite de Montreal, Montrt?al, Quebec, Canada ABSTRACT A microcomputer system was used to reconstruct, in the L5 dorsal root ganglion (DRG) of the rat, the three-dimensional arrangement of primary neurons which had been labelled by application of horseradish peroxidase (HRP) and fluoro-gold (FG) to various muscle nerves of the leg. Analysis of the data and animation of the reconstructed images with commercially available software were instrumental in identifying the preferential intraganglionic locations of the neurons innervating muscles such as the soleus (SOL), the gastrocnemius lateralis (GL), and medialis (GM), or parts of the GM. These locations appeared to be somewhat related to the position of the muscles in the posterior compartment of the leg. Additionally, the study provided quantitative estimates of muscle afferent neuronal populations, allowed a comparison of the labelling performances of HRP and FG, and finally indicated that few DRG neurons project to two different muscles. The somatotopical organization of many neural sys- Computerized reconstruction and manipulation in 3-D tems is illustrated by the clustering of spinal motoneu- of the image data was undertaken to establish the rerons into cell columns (Burke et al., 1977; Hollyday, spective locations of the neuronal aggregates under 1980; Janjua and Leong, 1984; Landmesser, 1978b; study. McHanwell and Biscoe, 1981; Nicolopoulos-Stournaras MATERIALS AND METHODS and Iles, 1983; Peyronnard et al., 1986; Romanes, 1964) which may be further subdivided into compartment nuSurgical Procedures and Application of Cell Tracers clei innervating specific parts of a muscle (Weeks and Eighteen adult female Sprague-Dawley rats were English, 1985,1987).In the sensory system, cytological anesthetized intraperitoneally with sodium pentobarand electrophysiological evidence points to a topo- bital. The nerve trunks derived from the posterior tibgraphical localization of the central projections of cu- ial nerve and supplying the soleus (SOL), the gastroctaneous and visceral afferents in the dorsal roots nemius lateralis (GL), and the gastrocnemius medialis (Brown and Fuchs, 1975; Heaney e t al., 1984; Koerber (GM) muscles were isolated in the right hindlimb, as and Brown, 1982; Koerber and Mendell, 1988; Light were the two terminal branches of the nerve to the GM and Durkovic, 1984; Wilson et al., 1986), and of their which penetrate the superior (GMS) and inferior (GMI) cell bodies in the spinal and trigeminal ganglia (Arl- parts of the muscle (Fig. 1).The proximal stumps of the hal, 1968; Capra and Wax, 1989; Corner et al., 1978; transected nerves were immersed for 2 hours in a Jacquin et al., 1983; Kausz and Rethelyi, 1985; Lende sealed capillary tube filled either with a 20% solution and Poulos, 1970; Marfut, 1981; Oyagi e t al., 1989; Shi- of HRP (type VI) (Sigma Chemical. Co., St. Louis, MO) genaga e t al., 1989). Primary muscle afferent neurons or a 3% solution of FG (Fluoro-chromes, Engelwood, differ in this respect, a s they lack a n orderly intragan- CO) in 0.9% saline. During this time, diffusion of the glionic distribution according to studies using either markers was prevented by ligating all neighboring the retrograde chromatolytic reaction after peripheral nerves and by wrapping the operation site with a plasnerve injury (Norcio and De Santis, 1976) or cell trac- tic sheet. In order to assess their labelling capacity, ing techniques (Kausz and Rethelyi, 1985; McLachlan HRP and FG were alternatively applied in some aniand Janig, 1983). We ourselves expressed this view in mals to the SOL, GM, and GL; but they were also used a report (Peyronnard et al., 1986) on the neuronal in- simultaneously to map, in the same animal, the muscle nervation of rat hindlimb muscles. However, further afferent neurons supplying either the SOL and GM examination of the material led u s to believe t h a t a definite conclusion required three-dimensional (3-D) analysis of the location of these neurons in dorsal root ganglia (DRG). For this purpose, the present study utiReceived August 16, 1989; accepted October 26, 1989. lized horseradish peroxidase (HRP) and f luoro-gold Address reprint requests to Dr. J.M. Peyronnard, Centre de Re(FG) to mark, in a given animal, the afferent neurons cherche en Sciences Neurologiques, Faculte de Medecine, Universite supplying one or two different muscle nerves, or alter- de Montreal, Case Postale 6208, Succ. A, Montreal, Quebec, Canada natively, the different branches of a muscle nerve. H3C 3T8. 0 1990 WILEY-LISS, INC. 406 J.-M. PEYRON NARD ET AL PL GL SOL Fig. 1. Axillary view of the rat hindlimb showing, in black, the posterior tibia1 nerve supply t o the gastrocnemius lateralis (GL), soleus (SOL), plantaris (PL),and gastrocnemius medialis muscles (GM), including the two terminal nerve branches t o the GM. (SOL-GM), the GM and GL (GM-GL),or the upper and lower parts of the GM (GMS-GMI). Seventy-two hours later, the animals were killed by transcardiac perfusion of 1% paraformaldehyde and 1.25% glutaraldehyde in phosphate buffer (pH 7.4). The right L4, L5, and L6 DRG were dissected, postfixed for 30 min in the same perfusate, and immersed for 3 hours in phosphate buffer containing 10% sucrose (pH 7.4). Tissue Processing and Analysis Before freezing in liquid nitrogen, the DRG were positioned to bring the ventral roots to the bottom, with the dorsal roots surrounded by the bulk of the ganglionic cells lying above (see Figs. 2 , 3 ) .Serial sections, 20 pm thick, were cut longitudinally with a cryostat (-20°C) microtome and processed for HRP histochemistry with tetramethylbenzidine (Mesulam, 1978). In the case of double-labelled ganglia, the fluorescent cells were detected first, with a epifluorescenceequipped Zeiss photomicroscope. These cells were numbered and their diameter was computed (Macintosh Plus) after projection onto a digitizing tablet. The same procedure was then applied to HRP-stained cells, using bright-field illumination. All data were transferred onto photographic prints of the sections enlarged to a final magnification of x 65. After superimposition and marking with three reference points, these prints were used for a 3-D reconstruction of individual cells. In the case of neurons appearing on several adjacent sections, only the level containing the largest diameter was retained. Subsequently, the position of the reference points and the labelled cells, as well a s the contours of the ganglia, ventral and dorsal roots were digitized on a graphic tablet with a 3-D reconstruction software package (MacReco 4.0, E. Otten and G.W. Dreschel, Groningen, The Netherlands). The data were processed on a Macintosh I1 microcomputer equipped with a 5-megabyte random access memory, a 40-megabyte hard disk, and a 13-inch color monitor (Apple). To prevent crowding of the reconstructed figures, DRG contours were digitized a t 4 different levels only (out of approximately 50 sections for a L5 ganglion). The ventral and dorsal root contours were also drawn at every third to fifth section, and no attempt was made to follow the intraganglionic pathway of the dorsal root fibers. To circumvent the limited capacity of the program with respect to figure animation, we employed one of its options to define the X, Y, Z coordinates of all selected points. After calibration measurements, the coordinate documents together with the information on cell diameters were turned over to a spreadsheet (Excel 2.0, Microsoft Corp., Seattle, WA) and transferred to graphic data analysis software (MacSpin 2.0, D2 Software Inc., Austin, TX). The MacSpin program, which we used to reconstruct roots and DRG contours, makes i t possible to draw lines between sequentially connected events. Furthermore, it offers the distinct advantage of allowing a 360"rotation of any 3-D structure in six different directions, while its multiple-window display options facilitates comparison between reconstructed DRG. One of its functions, identified as the centroid connection operation, was found to be most helpful for the visual enhancement of the topographical location and spatial extent of neurons innervating the same muscle target. In this mode of operation, the computer defines the center or centroid point of a neuronal cluster by averaging the X, Y, and Z coordinates of all the cells belonging to this cluster. It then selects the neuron closest to the centroid point and connects this centroid neuron linearly to every other related cell. We also used this function for a quantitative localization of SOL, GM, and GL neurons, as vai-iations in the DRG morphology from one animal to the other made it extremely difficult to use reference anatomical landmarks, such as contours of the ganglia or roots. For this purpose, the centroid neuron of each SOL-GM or GM-GL neuronal pool was determined and served as a reference for positioning the X, Y, Z axes within the ganglia. When viewed along the root or X axis, the Z and Y coordinates delineated four quadrants: dorsal lateral, dorsal medial, ventral lateral, and ventral medial. Within the pools (SOL-GM or GM-GL), the percentages of SOL, GM, or GL neurons projecting in each of these quadrants were determined, and the values obtained for all animals were averaged. The same procedure was followed after rotation in a lateral view along the Y axis. In this position, the Z coordinate NEURONAL ORGANIZATION IN DORSAL ROOT GANGLIA 407 X VENTRAL ROOT DORSAL ROOT LAMINA VERTEBRAE A Fig. 2. A. Oblique orientation of a L5 DRG in vivo. B Position given to all ganglia prior to processing and after rotation along the ventral root (X) axis. the combined use of FG and HRP made i t possible to establish that far more DRG neurons to the GM were supplied by the inferior than the superior nerve branch (Table 3). In one animal (GMS-GMI,), a relatively high RESULTS number of double-labelled neurons were found, possiSegmental Location and Numerical Estimates of DRG bly representing contamination artefacts due to the Muscle Afferent Neurons close proximity of the nerve structures exposed to the After application of the cell tracers to muscle nerves, markers. In all other instances, the percentage of douneurons containing fluorescent FG granules or the ble-labelled DRG cells was rather low, varying beHRP reaction product (Fig. 4A,B) were found in the tween 1 and 8.9 for a mean of 3.9. Finally, for all muslumbar spinal ganglia. As shown in Table 1, with a cles under study, DRG cell diameters ranged from 10 to single exception (animal SOL,), the vast majority (96% 65 pm. on average) of SOL neurons labelled with HRP or FG were located in the L5 DRG, the remainder being found lntraganglionic Location of Muscle Cell Afferent Neurons a t the L4 level. A similar proportion (95%)of afferent Since most SOL, GM, and GL neurons are located in neurons innervating the GM nerve or its two distal branches (Tables 2, 3) was present in the L5 spinal the L5 DRG, only results pertinent to these ganglia ganglia, the rest being scattered in the L6 and occa- will be presented. We have already mentioned that, sionally the L4 DRG. GL neurons were more equally prior to sectioning, the DRG were tilted approximately distributed between the L4 and L5 DRG but prevailed 60" laterally along the roots axis from the oblique poin the latter (Table 4). Quantitative assessment of sition they occupy in vivo (Fig. 2A). In this process, the whole muscle DRG cell populations with one or the bulk of the ganglia normally oriented towards the spiother tracer revealed little interindividual differences. nal canal was moved to a n upward and dorsal position, However, the number of HRP-labelled cells of the SOL while the motor roots were internally displaced in a and GM, averaging 103 and 136, respectively, was completely ventral location (Fig. 2B). By so doing, relower than the mean FG estimates: 144 and 162. Being constructions could proceed from DRG cut at approxiof the same magnitude in single- and double-labelled mately the same angle along the rostrocaudal axis, animals, the difference indicated that FG had not been starting from what could be defined as the top of the affected by treatment needed to reveal the HRP depos- ganglia and moving down towards the ventral roots its. Despite their unequal performance a s cell markers, (Fig. 3A). The fact that contours of the ganglia, ventral, served a s a dividing line between rostra1 and caudal sectors. Selected views of the ganglia were printed on a n Apple Laser Writer I1 NT and edited for illustration. A EXTERNAL VIEW A / Y Rostra1 Caudal Fig. 3. Diagrammatic reconstruction of a L5 DRG depicting A) the relationship between ganglionic cells (dotted areas), the ventral (VR) and dorsal (DR) roots, as well as the plans of histological sections; and B) the pattern of distribution of the dorsal root fibers as they appear at two different cross-sectional levels (a,b)of the DRG. At level a, dorsal root fibers exist as multiple bundles of axons (*) separated by islands of DRG cells. and dorsal roots had been digitized at selected intervals only, for the sake of clarity and without trying to reproduce the complex intraganglionic organization of the dorsal root fibers (Fig. 3B), did not interfere with analysis of the relationships between these anatomical elements and the position of the labelled cells. From the three orthogonal views of a dually labelled L5 DRG shown in Figure 5A-C, one can appreciate without additional treatment of the data that, despite an important spatial overlap, SOL and GM neurons tend to be confined to separate ganglionic areas. The centroid connection operation (Fig. 6A-C) illustrates more vividly that GM cells are mostly located in the dorsal and internal parts of the DRG, being closer to 409 NEURONAL ORGANIZATION IN DORSAL ROOT GANGLIA Fig. 4. Photomicrographs of a section of a dually labelled L5 DRG (animal SOL-GM,). In A, epifluorescent illumination shows GM neurons filled with FG granules, while in B, brightfield illumination reveals a SOL neuron containing HRP deposits. Bar: 20 Fm. TABLE 1. Number of dorsal root ganglion cells labelled after exposure of the soleus nerve to horseradish peroxidase or Fluoro-Gold Animal SOL-GM, 4 SOL, SOL, SOL-GM, 41 1 SOL, SOL-GM, 10 - Y) = HRP L5 L4 97 (2)l - - 56 97 (4) 108 - Total L4 FG L5 Total 101 - - - 97 98 6 - 149 - 155 118 - 3 130 (3) - - 133 No. of double-labelled cells. TABLE 2. Number of dorsal root ganglion cells labelled after exposure of the gastrocnemius medialis nerve to horseradish peroxidase or Fluoro-Gold HRP FG Animal SOL-GM, L4 L5 L6 Total L4 L5 L6 Total - - - - 4 9 183 SOL-GM, - - - - - 170 (2)l 147 17 164 (4) - - 2 156 - - - 139 12 151 GM, GM-GL, - 135 - - 135 - 3 SOL-GM, - - 137 - GM-GL, - 137 (3) - - - - ’( 151 (6) 1 = No. of double-labelled cells. the spinal nerve than most of the SOL neurons which occupy a more rostral, external and ventral position. Drawing envelopes of the roots and ganglion (Fig. 7) allowed a better appreciation of the relationships between these neuronal clusters and the anatomical elements of the DRG. Quantitative analysis performed in three animals revealed a fair constancy in the location of SOL and GM neurons within SOL-GM pools. As evidenced by Figure 8A, a n average of 83.8%of the SOL neurons projected into the ventral and lateral parts of these pools, while 88.2% of the GM neurons were found in the dorsal and medial sectors. This clustering was also apparent in the lateral view, since only 38.7% of the SOL neurons were caudally situated as opposed to 64.7%of the GM cells. It was interesting t h a t the intraganglionic topography of GL neurons in relation to GM cells (Fig. 9) was almost identical to t h a t described for the SOL, these two muscles also sharing a similar 410 J.-M. PEYRONNARD ET AL. Z A . 0 OO 0. 0 . L, Y -X 0. 0 0 ........ ........ ............... 0 0 .....,..,...L................ n,, ..... ................... ..... .............................~,,,,.....,,.. ....:..,..... ..... S .,..,; 0 ,n........ I......nW.....,,,ll...... M...... 22.214.171.124~ Z B 4 Y Ax ------- 0 00 Xo 0 0 . .................................. Oo 0 O0 0 QQ--*--- ,11111: 11111.11.1" 1111111111111111 :: Q C 'rx Y I 200 pm Fig. 5. A-C: Lateral, rostro-caudal, and dorsal views of a L5 DRG (animal SOL-GM,) after exposure of the GM and SOL nerves to FG and HRP, respectively. These drawings prepared from computer print outs indicate the position of the GM- (solid dots) and SOL- (empty dots) labelled cells and the outlines of the ganglion (G) and roots (VR and DR). The few double-labelled cells are not shown. Despite overlapping, it can be appreciated that the GM and SOL neurons are located in different ganglionic areas. 411 NEURONAL ORGANIZATION IN DORSAL ROOT GANGLIA . A B Y C 7 tI Z I X - 200 pin Fig 6 A-C: Reproduction of Figure 5 By drawing the centroid connections, the spatial segregation of the GM and SOL neuronal aggregates IS vlsually enhanced 412 J.-M. PEYRONNARD ET AL TABLE 3. Number of dorsal root ganglion cells labelled after exposure to horseradish peroxidase or Fluoro-Gold of the terminal nerve branches to the gastrocnemius medialis muscle L4 Animal GMS-GMI, - GMS-GMI, GMS-GMI, '( Superior nerve branch: HRP L5 Lfi 67 (26)' - - Total 67 L4 1 - - - - 41 (6) 2 43 - Inferior nerve branch: FG L5 Lfi 164 (26) 121 139 5 (6) Total 165 121 144 1 = No. of double-labelled cells. TABLE 4. Number of dorsal root ganglion cells labelled after exposure of the gastrocnemius lateralis nerve to horseradish peroxidase or Fluoro-Gold HRP Animal GM-GL, L4 37 GM-GL, - GM-GL, 40 '( ) = L5 58 (GI1 FG L6 L* L, Lf3 Total - Total 95 - - - - - - - 33 - 108 - 9 (2) - 146 68 (3) 104 (11) - - No. of double-labelled cells. INTERNAL VIEW Rostra1 200 pm Caudal Fig. 7. Same DRG a s in Figures 5 and 6 after illustration in 3-D through artistic drawing of the root and ganglion contours. To give a sense of depth, the size of the spheres representing GM (black) and SOL (white) neurons varies, depending on their proximity to the viewer. location in the rat leg, lateral to the GM (Fig. 1).The data in Figure 8B averaged from three different GMGL pools of neurons confirmed that 93.2% of the GL neurons were, like SOL cells, laterally and ventrally located, in contrast with the dorsal and medial location of 90.1% of the GM neurons. As we mentioned earlier for SOL neurons, GL and GM neurons were also present in opposite proportions in the rostra1 and caudal sectors of the GM-GL neuronal pool. While use of the centroid point and figure animation helped to de- fine a clear spatial segregation of neuronal innervation of the lateral and medial muscles in the posterior compartment of the leg, interindividual differences were met in the position of cellular aggregates. As a result, there were instances where demarcation between groups of neurons was apparent in certain views (Fig. 10A,C) but not in others (Fig. 10B). Studies dealing with nerve branches to the GM suggest a preferential location of neurons in the dorsal aspect of the L5 ganglion innervating the upper part of the muscle, the NEURONAL ORGANIZATION IN DORSAL ROOT GANGLIA 413 chicken (Hollyday, 1980; Landmesser, 1978131, mouse (McHanwell and Biscoe, 1981), rat (NicolopoulosStournaras and Iles, 1983; Peyronnard et al., 19861, cat (Romanes, 1964) and monkey (Janjua and Leong, 1984). Extant evidence suggests that primary sensory neurons innervating cutaneous and visceral structures or their centripetal processes have an orderly distribution. For instance, in lumbo-sacral DRG of the cat, a link has been found between the position of a neuron in LATERAL MEDIAL the ganglion or that of its peripheral receptive field and the location of its central projection either in the dorsal GM neurons root or in the spinal dorsal horn (Brown and Fuchs, 1975; Burton and McFarlane, 1973; Koerber and Brown, 1982; Koerber and Mendell, 1988; Light and Durkovic, 1984; Wilson et al., 1986). Along the same lines, electrophysiological studies have suggested a somatotopical organization of the saphenous nerve input to the spinal cord in the cat and rabbit (Heaney et al., 1984), and the clustering of cutaneous spinal ganglion cells in the frog (Corner et al., 1978). Finally, microelectrode recordings (Arlhal, 1968) and cell tracing techniques (Kausz and Rethelyi, 1985) have helped to B GL neurons Z demonstrate, in the S2 ganglion of the cat, a topographical arrangement of the neuronal somata supplying the perineum and the urinary bladder. Similar observations have been made in the trigeminal ganglion of rats and cats through the use of HRP or its wheat germ agglutinin conjugate (Capra and Wax, 1989; Jacquin et al., 1983; Marfut, 1981; Oyagi et al., 1989; Shigenaga et al., 1989). In contrast, muscle afferent neurons have been shown to be erratically dispersed throughout the GM neurons spinal ganglia, even in studies employing HRP (Kausz z and Rethelyi, 1985; McLachlan and Janig, 1983; Peyronnard et al., 1986) as a more appropriate method for precise cell identification than the retrograde chromatolytic reaction (Norcio and De Santis, 1976). However, these negative findings were derived solely from the inspection of serial DRG sections, and not from a 3-D reconstruction of the ganglia, as is the case in this report. Fig. 8. Compartmental distribution of SOL, GM, and GL cells within Such a reconstruction faced challenges which could SOL-GM (A) or GM-GL (B) pools, as seen in two orthogonal views. not be met with conventional drawing techniques. InThe X, Y and Z axes, aligned on the centroid neuron (c) of these pools, divide the neuronal populations into defined compartments. Values deed, only a computerized system could be expected 1) are averaged percentages of neurons projecting in a specific compart- to position in 3-D up to several hundred muscle afferment in three different animals. ent cells in relation to structural landmarks, such as the outlines of the roots and the ganglion itself; and 2) to rotate the reconstructed image in order to apprecineuronal supply to the other parts being somewhat in ate, from different perspectives, the relationships bea more ventral position. Finally, it is noteworthy that tween these anatomical elements or to position ganglia no clustering by neuronal size was detected and that from different animals along the same angle for the the few double-labelled cells appeared to be erratically purpose of comparison. Computer-assisted image reconstruction is no longer distributed. the exclusive domain of mainframe systems of the kind DISCUSSION recently used to assemble a rat brain in 3-D (Toga et The image-processing system used in this study was al., 1989). In recent years, numerous systems based on instrumental in revealing that L5 DRG neurons, which microcomputer technology have been commercially desupply specific leg muscles of the rat, have a distinct veloped (Huijsmans et al., 1986; Jarvis, 1988). Howintraganglionic topography linked to some extent to ever, most remain expensive, while some have a limited capacity to manipulate the image or do not offer the anatomical position of muscles in the leg. Until now, muscle afferent nerve cells seemed rather the possibility of working with multiple screen winunique because of their lack of somatotopical organi- dows. We assembled our own system around a mediumzation common to most other primary neuromuscular priced Mac11 microcomputer equipped with a graphic neurons. Indeed, the columnar arrangement of mo- tablet. The input data were photographic prints of DRG toneurons in the ventral horn of the spinal gray matter sections which had to be first superposed to eliminate is a feature observed in many animals, including the redundant neuronal cross-sections, a process which, for A Z w SOL neurons Z A B *-v, Z Y P A X Z 200 Lrn Fig. 9. A-C: Three orthogonal views of the L5 DRG of animal GM-GL,, showing that most HRPlabelled GL neurons (empty dots), like SOL nerve cells (see Fig. 51, are located in the ventral, lateral, and rostra1 part of the ganglion, while a majority of FG-labelled GM neurons (solid dots) are in a more dorsal, internal, and caudal position. A B Y- i X C Y U 200 pm Fig. 10. In this dually-labelled animal (SOL-GM,), the spatial segregation of the SOL (empty dots) and GM (solid dots) neurons is mostly apparent in the lateral and dorsal views of the L5 DRG (A,C) and is difficult to appreciate in the rostrocaudal projection (B). 416 J.-M. PEYRONNARD ET AI, the time being, can hardly be done with scanned or compatibility with peroxidase cytochemistry (Cabrera video-captured images. This procedure allowed the in- e t al., 1988; Pieribone and Aston-Jones, 1988; Schmued clusion of reference marks on the photographs, a step and Fallon, 1986), it also led to neuronal estimates critical to the realignment of the sections during recon- which, on average, were 26% higher than those obstruction. No attempt was otherwise made to correct tained with HRP (Cabrera et al., 1988). There was no the potential distortions resulting from slicing and definite evidence that this difference could be artifactual, resulting, for instance, from the accidental labelmounting the tissue (Huijsmans et al., 1986). The software package (see Materials and Methods) ling of adjacent nerve filaments through diffusion of used to digitize the input images on a graphic tablet led the tracer. In fact, only one animal gave reason to susto a volumetric representation of the DRG which re- pect that such a phenomenon could have taken place, mained nonetheless unsatisfactory due to the limited judging from the high number of dually labelled cells capacity of the program for image animation. This de- after exposure of the superior and inferior branches of ficiency was resolved by exporting the data to another the GM to HRP and FG, respectively. Such double laprogram allowing simultaneous rotation of several 3-D belling was otherwise infrequent and implicated on avDRG images in a n almost unlimited number of direc- erage no more than 3.9% of the neurons. This would tions. However, a special graphic design was needed to confirm that despite the well-documented peripheral unravel the intricate arrangement of muscle afferent dichotomy of DRG cells (Coggeshall, 1985, 1986; Honerve cells. Indeed, the preferential grouping of neu- heisel and Mense, 1987; Langford and Coggeshall, rons innervating specific leg muscles of the rat in cer- 1981; Pierau et al., 1982; Taylor and Pierau, 19821, few tain areas of the L5 DRG became obvious once the cen- afferent neurons send axons to separate nerves in the troid point and its connections had been defined for rat (Devor e t al., 1984). each neuronal aggregate. In so doing, we not only obACKNOWLEDGMENTS served a fair constancy in the location of nerve cell clusters but also a similarity in the intraganglionic The authors are indebted to the Apple Company of topography of GL and SOL neurons in relation to GM Canada for its technical support (of M.D.) and to C. innervation. If a link exists between this observation Gauthier and G. Filosi for their artwork. Gratitude is and the shared location, lateral to the GM, of the GL extended to 0. Da Silva for editing this manuscript. and SOL in the posterior compartment of the leg, it could suggest that, during development, muscle afferLITERATURE CITED ent neurons, like sensory (Honig, 1982; Scott, 1982) Arlhal, A. 1968 Mise en evidence d’une somatotopie a u niveau du and motor cells (Lances-Jones and Landmesser, 1981; ganglion spinal chez le chat. C.R. SOC.Biol. (Paris), 162:1979Landmesser and Morris, 1975; Landmesser, 1978a; 1984. Tosney and Landmesser, 19851, respond to limb- Baron, R., W. Janig, and W. Kollmann 1988 Sympathetic and afferent somata projecting in hindlimb nerves and the anatomical orgaderived guidance cues, anatomically-related muscles nization of the lumbar sympathetic nervous system of the rat. J. preferentially attracting cells located in neighboring Comp. Neurol., 275,460-468. parts of the DRG. An alternative explanation would be Brown, P.B., and J.L. Fuchs 1975 Somatotopic representation of hindlimb skin in cat dorsal horn. J . Neurophysiol., 38:l-9. that the spatial segregation of DRG cells is specified R.E., P.L. Strick, K. Kanda, C.C. Kim, and B. Walmsley 1977 through early central connections established with mo- Burke, Anatomy of medial gastrocnemius and soleus motor nuclei in cat toneuronal columns, in line with the essential role of spinal cord. J. Neurophysiol., 4Ot667-680. motor cells in the development of muscle afferent in- Burton, H., and J.J. McFarlane 1973 The organization of the seventh lumbar spinal ganglion of the cat. J. Comp. Neurol., I49:215nervation, a s documented a t least in the chicken 232. (Honig et al., 1986; Landmesser and Honig, 1986). Not Cabrera, B., F. Portillo, R. Pasaro, and J.M. Delgado-Garcia 1988 unlike the compartmentalization known to exist in the Location of motoneurons and internuclear neurons within the rat lateral (Weeks and English, 1985) and medial (Weeks abducens nucleus by means of horseradish peroxidase and fluorescent double labeling. Neurosci. Lett., 87:l-6. and English, 1987) gastrocnemius motor nuclei of the N.F., and T.D. Wax 1989 Distribution and central projections cat, we observed some degree of partitioning within Capra, of primary afferent neurons that innervate the masseter muscle DRG neurons to the GM, cells innervating the upper and mandibular periodontium: A double-label study. J . Comp. part of the muscle being spatially segregated from Neurol., 279t341-352. Coggeshall, R.E. 1985 An overview of dorsal root axon branching and those supplying the lower part. ventral root afferent fibers. In: Development, Organization, and This study offered the additional opportunity of comProcessing in Somatosensory Pathways. M. Rowe and W.D. Wilparing the performance of two different markers, HRP lis, Jr., eds. Alan R. Liss, Inc., New York, pp. 105-110. and FG, in the quantitative assessment of the neuronal Coggeshall, R.E. 1986 Nonclassical features of dorsal root ganglion cell organization. In: Spinal Afferent Processing. T. Yaksh, eds. innervation of several leg muscles in the rat. As prePlenum Pub., New York, pp. 83-96. viously reported in this animal (Baron et al., 1988; Pey- Corner, M.A., W.A.M. Veltman, R.E. Baker, and J . Van de Nes 1978 ronnard et al., 19861, both tracers concurred to demonTopography of cutaneous spinal ganglion cells in the frog (Rana strate that GM and SOL neurons are mostly located in esculentu) Brain Res., 156:151-156. the L5 DRG, a few cells being present in adjacent gan- Devor, M., P.D. Wall, and S.B. McMahon 1984 Dichotomizing somatic nerve fibers exist in rats but they are rare. Neurosci. Lett., 49: glia, at L4 for the SOL or a t L4 and L6 for the GM. The 187-1 92. segmental location of GL neurons was no different, Heaney, S.K., P.J. Kendell, S.J.W. Lisney, and C.M. Pover 1984 The apart from the presence of a sizeable population of organization of saphenous nerve fibers in the dorsal roots of the rabbit and cat. Somatosens. Res., 2:83-92. nerve cells in the L4 DRG. If compared to values reported earlier (Peyronnard et al., 1986), the present Hoheisel, U., and S. Mense 1987 Observations on the morphology of axons and somata of slowly conducting dorsal root ganglion cells counts of HRP-labelled cells were almost identical for in the cat. Brain Res., 423.269-278. the SOL and somewhat higher for the GM. Using FG, Hollyday, M. 1980 Organization of motor pools in the chick lumbar we found, a s did others investigators, that besides its lateral motor column. J. Comp. Neurol., 194:143-170. NEURONAL ORGANIZATION IN DORSAL ROOT GANGLIA Honig, M.G. 1982 The development of sensory projection patterns in embryonic chick hind limb. J . Physiol. (Lond.),330:175-202. Honig, M.G., C. Lance-Jones, and L. Landmesser 1986 The development of sensory projection patterns in embryonic chick hindlimb under experimental conditions. Dev. Biol., 118:532-548. Huijsmans, D.P., W.H. Lamers, J.A. Los, and J. Strackee 1986 Toward computerized morphometric facilities: A review of 58 software packages for computer-aided three-dimensional reconstruction, quantification, and picture generation from parallel serial sections. Anat. Rec., 216:449-470. Jacquin, M.F., K. Semba, M.D. Egger, and R.W. Rhoades 1983 Organization of HRP-labeled trigeminal mandibular primary afferent neurons in the rat. J . Comp. Neurol., 215t397-420. Janjua, M.Z., and S.K. Leong 1984 Organization of neurons forming the femoral, sciatic, common peroneal and tibia1 nerves in rats and monkeys. Brain Res., 310:311-323. Jarvis, L.R. 1988 Microcomputer video image analysis. J. Microsc., 150333-97. Kausz, M., and M. Rethelyi 1985 Lamellar arrangement of neuronal somata in the dorsal root ganglion of the cat. Somatosens. Res., 2:193-204. Koerber, H.R., and P.B. Brown 1982 Somatotopic organization of hindlimb cutaneous nerve projection to cat dorsal horn. J. Neurophysiol., 48:481-489. Koerber, H.R., and L.M. Mendell 1988 Functional specialization of central projections from identified primary afferent fibers. J. Neurophysiol., 60t1597-1614. Lance-Jones, C., and L. Landmesser 1981 Pathway selection by chick lumbosacral motoneurons during normal development. Proc. R. SOC.Lond. [Biol.], 214:l-18. Landmesser, L. 1978a The development of motor projection patterns in chick hind limb. J. Physiol. (Lond.),284:391-414. Landmesser, L. 1978b The distribution of motoneurons supplying chick hind limb muscles. J. Physiol. (Lond.),284:371-389. Landmesser, L., and M.G. Honig 1986 Altered sensory projections in the chick hind limb following the early removal of motoneurons. Dev. Biol., 118511-531. Landmesser, L., and D.G. Morris 1975 The development of functional innervation in the hind limb of chick embryo. J . Physiol. (Lond.), 249:301-326. Langford, L.A., and R.E. Coggeshalll981 Branching of sensory axons in the peripheral nerve of the rat. J . Comp. Neurol.,203:745-750. Lende, R.A., and D.A. Poulos 1970 Functional localization in the trigeminal ganglion in the monkeys. J . Neurosurg., 32:336-343. Light, A.R., and R.G. Durkovic 1984 Features of laminar and somatotopic organization of lumbar spinal cord units receiving cutaneous inputs from hindlimb receptive fields. J . Neurophysiol., 521449-458. Marfut, C.F. 1981 The somatotopic organization of the cat trigeminal ganglion as determined by horseradish peroxidase technique. Anat. Rec., 201t105-118. McHanwell, S., and T.J. Biscoe 1981 The localization of motoneurons 417 supplying the hindlimb muscles of the mouse. Philos. Trans. R. Soe. Lond. [Biol.], 293r477-508. McLachlan, E.M., and W. Janig 1983 The cell bodies of origin of sympathetic and sensory axons in some skin and muscle nerves of the cat hindlimb. J. Comp. Neurol., 214r115-130. Mesulam, M.M. 1978 Tetramethylbenzidine for horseradish peroxidase neurohistochemistry: A non-carcinogenic blue reactionproduct with superior sensitivity for visualizing neural afferents and efferents. J. Histochem. Cytochem., 28:1255-1259. Nicolopoulos-Stournaras,S., and J.F. Iles 1983 Motor neuron columns in the lumbar spinal cord of the rat. J. Comp. Neurol., 21 7:75-85. Norcio, R., and M. De Santis 1976 The organization of neuronal somata in the first sacral spinal ganglion of the cat. Exp. Neurol., 50t246-258. Oyagi, S., J. Ito, and I. Honjo 1989 Topographic study of the feline trigeminal ganglion via the horseradish peroxidase tracer method. Brain Res., 476:382,383. Peyronnard, J.M., L.F. Charron, J. Lavoie, and J.P. Messier 1986 Motor, sympathetic and sensory innervation of rat skeletal muscles. Brain Res., 373t288-302. Pierau, F.K., D.C.M. Taylor, W. Abel, and B. Friedrich 1982 Dichotomizing peripheral fibres revealed by intracellular recording from rat sensory neurons. Neurosci. Lett., 31:123-128. Pieribone, V.A., and G. Aston-Jones 1988 The iontophoretic application of Fluoro-Gold for the study of afferents to deep brain nuclei. Brain Res., 475:259-271. Romanes, G.J. 1964 The motor pools of the spinal cord. Prog. Brain Res., 11:93-119. Schmued, L.C., and J.H. Fallon 1986 Fluoro-Gold: A new fluorescent retrograde axonal tracer with numerous unique properties. Brain Res., 377t147-154. Scott, S.A. 1982 The development of segmental pattern of skin sensory innervation in embryonic chick hind limb. J. Physiol. (Lond.), 330:203-220. Shigenaga, Y., M. Nishimura, S. Suemune, T. Nishimori, K. Doe, and H. Tsuru 1989 Somatotopic organization of tooth pulp primary afferent neurons in the cat. Brain Res., 477t66-89. Taylor, D.C.M., F.K. Pierau 1982 Double fluorescence labelling supports electrophysiological evidence for dichotomizing peripheral sensory nerve fibres in rats. Neurosci. Lett., 33;l-6. Toga, A.W., M. Samai, and B.A. Payne 1989 Digital rat brain: A computerized atlas. Brain Res. Bull., 22r323-333. Tosney, K.W., and L.T. Landmesser 1985 Specificity of motoneuron growth cone outgrowth in chick hindlimb. J . Neurosci., 5:23362344. Weeks, O.I., and A.W. English 1985 Compartmentalization of the cat lateral gastrocnemius motor nucleus. J . Comp. Neurol., 235:255267. Weeks, O.I., and A.W. English 1987 Cat triceps surae motor nuclei are organized topologically. Exp. Neurol., 96:163-177. Wilson, P., D.E.R. Myers, and P.J. Snow 1986 The detailed somatotopic organization of the dorsal horn in the lumbosacral enlargement of the cat spinal cord. J. Neurophysiol., 55r604-617.