THE ANATOMICAL RECORD 219:315-322 (1987) The Number of Neurons in Dorsal Root Ganglia L4-L6 of the Rat HENNING SCHMALBRUCH Institute of Neurophysiology, University of Copenhagen, The Panum Institute, DK-2200 Copenhagen N, Denmark ABSTRACT The number of neurons in the dorsal root ganglia L4-L6 of the rat was determined because published data are inconsistent and in general incompatible with the number of afferent axons in the sciatic nerve. Nucleoli were counted in serial sections; epoxy-resin sections 3 pm thick, or paraffin sections 5 pm thick, or unstained 12-pm paraffin sections of osmicated tissue were used. Correction factors for split and multiple nucleoli were obtained by counting nucleolar profiles in consecutive sections of identified cells. Dividing the number of nucleolar profiles into the number of cells gave the factor by which the counts of nucleolar profiles had to be multiplied to obtain the number of neurons. The ganglia L4, L5, and L6 contained about 12,000, 15,000 and 14,000 neurons, respectively, when resin sections were used. The standard deviation for the average of 41,000 neurons in the three ganglia was 8% of the mean value. The results compare well with the number of dorsal root fibers, and with the fact that the sciatic nerve at midthigh, to which less than half of the neurons connect, contains 19,000 afferent axons. The data obtained from the paraffin series were 23% smaller, but still considerably higher and less variable than all previously reported data. The main problem with stained paraffin sections was that most small neurons had multiple nucleoli attached to the membrane of the nuclei, which only measured 10 pm in diameter. The nucleoli often projected into the dark cytoplasm and were difficult to identify. The sciatic nerve of the rat is frequently used in experimental work. Its motor and sensory fibers originate from the spinal segments and dorsal root ganglia L4L6. The relative contribution of the individual ventral roots to the motor axon pool of the nerve varies and the right-left difference for individual root pairs is much larger than that for the total number of fibers in the three roots (Schmalbruch, 1984). Similar variations may occur in lumbar dorsal root ganglia as well. The side differences €or thoracic dorsal root ganglia almost vanished when rows of ganglia were compared (Ygge et al., 1981). Because none of the lumbar ganglia exclusively supply the sciatic nerve and because of the likely anatomical variability, experimentally induced changes of the neuron numbers can be quantitated only when all three segmental levels are studied. The normal number of neurons in the ganglia L4 through L6 has been determined in two studies only (Bondok and Sansone, 1984; Arvidsson et al., 1986). These results, and also those of studies in which only one of the ganglia was investigated, are inconsistent (Table 1).The interindividual variation in all papers is much larger than one would expect from the variability of the number of nerve fibers in the sciatic nerve or dorsal roots (Schmalbruch, 1986a,b). All neuron counts appear incompatible with the observation that the sciatic nerve contains 19,000 afferent axons (Schmalbruch, 1986a), although half of the L4 and L5 ganglion cells 0 1987 ALAN R. LISS, INC. have axons in nerves other than the sciatic nerve (Devor et al., 1985). The aim of the present study was to determine the number of neurons in the three ganglia by means of different histological methods, and to elucidate why the neuron counts vary. MATERIALS AND METHODS Twenty Wistar rats of both sexes and 23-96 weeks old were used. The ganglia were taken from rats in which the contralateral sciatic nerve had been cut at birth or at 4 weeks to study induced neuron death (Schmalbruch, 1987).Three rats were used as unoperated controls. The animals were anesthetized with halothane. Twelve rats were perfused transcardially with 2.5%glutaraldehyde in Ringer solution. The lumbar and sacral spinal cord together with roots and ganglia were excised, and the ganglia were identified with the aid of the adherent ventral roots. The ventral root L6 was distinctly thinner than the ventral root L5, which was equal to, or slightly thicker than, the L4 ventral root. There was no reliable criterion to distinguish excised ganglia or dorsal roots directly. The ganglia were postfixed with osmium tetroxide and embedded in epoxy resin (Embed 812). Eight rats were perfused with isotonic buffered 4% paraformaldehyde, and the sacral spinal cord with adherent lumbar ganglia was embedded en bloc in paraffin. Received February 5, 1987; accepted June 1,1987. 316 H. SCHMALBRUCH Complete series of 3-pm sections were cut from the ganglia embedded in resin. The sections were stained individually with p-phenylenediamine and mounted 10 per slide. A typical series for one ganglion comprised 400-600 sections. The paraffin blocks were cut into series of 5-pm cross sections. Sections containing the ganglia L4-L6 were mounted and stained with acidic cresyl violet. A series of 8-pm sections was prepared but the results were discarded (see Sources of Error). A Spencer rotary microtome (American Optic, Buffalo, N.Y.) with glass knives (resin sections) or steel knives (paraffin sections) was used. One set of ganglia was prepared following Hatai’s (1902) procedure. A rat was killed by an overdose of chloroform; the ganglia were excised, fixed for 24 hr with 1%osmium tetroxide in water, and embedded in paraffin, and serial cross sections 12 pm thick were mounted without additional staining. Counts and Measurements use any of the correction formulas for split nucleoli that are based on average nucleolus size and section thickness (see Discussion). Therefore, the empirical method of Coggeshall et al. (1984)was applied. A given area was photographed in 10-20 consecutive sections. In one section, all neurons with nucleolar profiles were identified, and in flanking sections, the number of additional nucleolar profiles contained in the same nuclei was determined. If a nucleus in one section contained more than one nucleolar profile, these were counted as one. Correction factors were computed from the number of cells and nucleolar profiles. The correction factors with which the raw data were multiplied to obtain the number of neurons were 0.84 for 5-pm paraffin sections and 0.61 (L4 and L5) and 0.57 66)for 3-pm resin sections (1,218cells studied in resin sections). The nucleolus counts in 12-pm sections were not corrected. Sources of Error Criteria for counting the nucleoli and their reliability The number of neurons was determined by counting were evaluated on consecutive serial sections. The minucleolar profiles in 10% of the sections. One of the 10 crographs of paraffin sections infrequently showed mulsections on each slide was randomly selected to elimi- tiple nucleoli, but rather often no nucleolus at all could nate the effect of systematic variations in section thick- be seen in adjacent sections. This was frequently so in ness. Raw data for the number of nucleoli (i.e., cells) 8-pm sections and these results were discarded. The were obtained by multiplying the results by 10. Phase identification of nucleoli in resin sections was without contrast (objectiveLeitz Plan 25, N.A. 0.45) was used for problems and at least one nucleolar profile was visible resin sections and brightfield (objective Leitz Apo 24, in each neuron. Two nucleolar profiles in consecutive N.A. 0.65) was used for paraffin sections. The microscope sections of a small neuron were never due to splitting was equipped with an eyepiece graticule to prevent dou- but to the presence of more than one nucleolus (Fig. 1). Displacement of nucleoli by the edge of the knife (Camble counting. Ganglion cells were photographed and the negatives mermeyer, 1967)did not occur in resin-embedded tissue, were projected onto a digitizer tablet (MOP 11, Kontron, but in paraffin sections, large-cell nucleoli were often so FRG) to a final magnification of 800-1,000. The perime- displaced. Tangentially sectioned large-cell nucleoli in ters of all neurons containing nucleoli were measured resin sections sometimes appeared with low contrast, and divided by 3.14 to obtain “idealized” diameters in- and could not be distinguished from low-contrast condependent of the irregular shape of the neuron. The densations of chromatin (Fig. 2). Hence, low-contrast sections to be photographed were at regular intervals structures were not counted as nucleoli. It was essential selected from the series passing through a ganglion to that counting of nucleoli and determining the correction reduce bias by the nonhomogeneous distribution of large factors be done by the same person. and small neurons. R ESULTS The size of nucleoli was determined in the microscope Five sets of ganglia L4-L6 embedded in resin con(objective Zeiss Plan Apo 100 oil, N.A. 1.3)with the aid of a screw micrometer eyepiece. The thickness of the 3- tained an average of 41,000 neurons with a standard pm resin sections was controlled by focusing the 1 0 0 ~deviation of 8% of the mean value. The ganglia L4, L5, oil immersion objective onto the upper and lower sur- and L6 contained 12,000, 15,000, and 14,000 neurons, faces of the section. The accuracy of the microscope drive respectively, with standard deviations for the individual was better than 1 pm (Leitz Ortholux 51.2). Individual ganglia of 7-17% of the mean. 1 pm. The mag- Ten complete sets of ganglia embedded in paraffin section thicknesses varied less than nification of all micrographs was calibrated against mi- were investigated and the mean value for the total numcrographs of a stage micrometer. All counts and ber of neurons was 32,000, with a standard deviation of 8% of the mean value. The ganglia L4-L6 contained measurements were performed by the author. 10,000, 12,000, and 11,000neurons, with standard deviaCorrection Factors tions for the individual ganglia of 11-16% (Table 2). The Study of the resin sections, in particular when consec- deficit for the total number of neurons in the paraffin as utive sections were evaluated, disclosed that almost all compared to the resin series was 23%. The neuron counts in unstained 12-pm sections after small “dark” neurons contained 2-3 nucleoli that were close to the nuclear membrane. Large Yight” neurons, osmium fixation were comparable to those in stained 5however, contained only one centrally placed nucleolus pm sections. The preservation of the cells after osmium (Figs. 1,Z). The diameter of the nucleolar profiles after fixation was better than after formalin fixation, and the resin embedding was 1.1-2.4 pm in small neurons and black myelin sheaths facilitated counting, but it was 4.5-5.5 pm in large neurons. The nucleoli in paraffin necessary to focus through the section for each neuron sections measured 1.6-2.6 pm in small neurons and 3.2- because the section thickness by far exceeded the focal depth of the objective and it was also larger than the 4.0 pm in large neurons. In view of the varying numbers of nucleoli and the diameter of the nuclei of the small ganglion cells. A differences in their size, it did not seem appropriate to series of 12-pmcross sections for one ganglion comprised NUMBER OF NEURONS IN RAT DRG L4-L6 Fig. 1. Two series of consecutive sections of a spinal ganglion L6; resin sections 3 pm thick, phase contrast. A-E: Large neuron. The nucleus contains one large nucleolus, which is centrally placed; it is split and appears in B and C (arrows).The nucleus is almost 20 pm in diameter but only 3 distinct (B-D) and 2 indistinct (A, E) nuclear 317 profiles are seen (see Fig. 4). F-I: Small neurons. The nucleoli are peripheral and may project into the cytoplasm (F, left). The nuclei measure slightly more than 10 pm in diameter but occur in only 2-3 successive sections. One cell contains only one nucleolus (bent arrows). Bar:20 pm. 318 H. SCHMALBRUCH TABLE 1. Comparison of neuron counts in spinal ganglia L4-L6 in the rat L4 L5 L4-L6 8,400 (5,80011,300) Langford and Coggeshall (1979) Bondok and Sansone (1984) L6 5,300 (SD 240) 5,400 (SD 310) 4,100 (SD 330) 14,800 n Technique 6 Paraffin, 8cLm 10 Paraffin/ frozen, 10 pm Yip et al. (1984) 7,000 (SD 1,150) 11 Paraffin, 8cLm Feringa et al. (1985) 2,000 (SD 230) 14 Paraffin, 7pm 16 Paraffin, 10 pm Frozen, lOpm Devor and GovrinLippmann (1985) 7,100 (5,000-9,800) 7,300 (5,200-9,800) Arvidsson et al. (1986) 5,900 (4,600-9,100) 8,900 (5,10011,400) 6,000 (3,30010,900) 20,800 (13,00029,200) 12 Schmalbruch (this study) 12,500 (10,70016,000) 14,800 (13,20017,200) 14,100 (13,40015,100) 41,400 (38,50046,600 5 EPOXY, 3 P 9,600 (7,00011.900) 11,700 (8,20013.800) 10,600 (9,30012.800) 32,000 (28,50035.500) 10 Paraffin, 5pm The figures are mean values, either given by the authors or calculated from the data for individual ganglia. The range is given in brackets; if this was not possible the standard deviation (SD) was calculated from the standard error of mean and the number of observations. n: number of ganglia assessed; technique: embedding medium and thickness of the serial sections. In all papers listed, the number of nucleoli was counted in serial sections and different correction methods for split nucleoli were applied to determine the number of neurons. Fig. 2. Three consecutive sections of a large neuron from a ganglion cap of the nucleolus is seen, but its contrast is low and this structure L5. Resin sections 3 pm thick; phase contrast. The origin of the axon is indistinguishable from other densities within the nucleoplasm. C: is seen in B (arrow), the dark granules in A and B are lipofuscin. A The section passes through the periphery of the nucleus but still The section passes through the middle of the nucleus and contains the contains a complete slice (not a cap) of the nucleus. The delineation is nucleolus. The nucleoplasm is smoothly cut, and the nucleolus is not poor because the nuclear membrane is cut obliquely.The nuclear slice displaced. The nuclear membrane is distinct. B: The section is still has a large and a small face; the contour of the small face, which has close to the center of the nucleus; its size is about the same as in A. A a diameter of 5-6 pm, is visible. Bar: 10 pm. 3 19 NUMBER OF NEURONS IN RAT DRG L4-L6 TABLE 2. The number of neurons in rat spinal ganglia L4-L6 Age o r weieht Resin, 3 pm 23 weeks 38 weeks 39 weeks 43 weeks 96 weeks Mean Standard deviation % L4 L5 L6 Total 11,490 12,810 10,690 16,010 11,330 12,470 13,190 14,800 14,130 17,210 14,760 14,820 15,050 13,690 13,690 13,400 14,570 14,080 39,730 41,300 38,510 46,620 40,660 41,360 2,130 17 1,490 12 700 5 3,120 8 6,980 9,890 11,920 11,070 9,160 10,310 9,840 7,400 10,050 11,110 8,120 9,620 13,760 12,040 12,050 12,750 8,190 30,580 32,250 35,280 35,260 29,680 12,890 12,350 9,350 12,180 11,080 11.660 9,840 10,320 11,310 11,440 12,330 9,480 12,780 10,040 10,620 9,680 9,260 10.650 1,570 16 1,700 15 1,170 11 2,650 8 8,460 10,070 11,240 29,770 Paraffin, 5 pm 300 gm 300 gm 300 gm 300 gm 300 gm 300 gm 72 weeks 72 weeks 77 weeks 77 weeks 89 weeks Mean Standard deviation % Osmium-paraffin, 12 pm 300 gm lost - 35,510 29,790 30,020 32,970 28,460 31,980 The nucleoli were counted in epoxy-resin sections 3 pm thick, in paraffin sections 5 pm thick, and in paraffin sections 12 pm thick. The counts were corrected for multiple and split nucleoli by the empirical method of Coggeshall et al. (1984). In rats of known age, the contralateral sciatic nerve had been cut at birth or at age 4 weeks. 300 gm: unoperated male rats of 250-350 gm weight. PARAFFIN EPOXY RESIN % 8 4 0 b i0 40 Fig. 3.Histograms of the diameter of the perikarya of L5 dorsal root ganglion cells embedded in epoxy resin and paraffin. Note pronounced shrinkage of the paraffin-embedded cells. 200-300 sections of which 20-30 were analyzed. The results differed by less than 2% when the counts from each 20th section (i.e., from only 10-15 sections) were used to calculate the number of neurons. The tissue embedded in paraffin showed more severe shrinkage than after resin embedding, and the large neurons always were detached from their surrounding satellite cells. The size histograms confirmed that shrinkage in paraffin was excessive. No neurons larger than 40 pm were found, whereas 20-25% of the resinembedded neurons were of this size. Very small neurons with diameters of less than 26 pm accounted for 20-25% of the resin-embedded neurons, but in paraffin sections more than 80%of the cells measured less than 26 pm in diameter. The histograms were skewed and probably had two peaks after both embedding methods (Fig. 3). The nuclei of large and of small neurons measured roughly 20 pm and 10 pm in diameter. The large nuclei in the 3-pm series occurred in five consecutive sections; small nuclei were seen in three, sometimes in only two sections (Fig. 1).This was unexpected, considering the diameter of the nuclei and the thickness of the sections. Several series were evaluated to exclude the possibility that the “deficit” of nuclear profiles was due to loss of 320 H. SCHMALBRUCH I H Fig. 4. Schema illustrating the relation between section thickness, orientation of nuclei and nucleoli, and their appearance in sections. I: Nucleus of a large neuron with one nucleolus; 3-pm sections. The slices 2-6 are seen in the microscope, the caps 1 and 7 are lost because of the overlying dark cytoplasm and the poor delineation by the obliquely sectioned nuclear membrane. Note that the delineation is poor for the first and last complete slices (2,6) a s well. The nucleolus may be split and appear in sections 3 and 4 (A), or it may appear in section 4 only because both caps are lost (B). (cf. Figs. 1, 2C.) 11: Nucleus of a small neuron with two nucleoli, 3-pm sections. The nucleus may be cut into 5 (A) or 4 (B, C) slices; the slices 1 and 5 (A), or 1@), or 1and 4 (C), are probably invisible and the nucleus appears in only 3 (A, B), or 2 (C) consecutive sections. Both nucleoli may be in the same (A) or in different sections @, C). The lower nucleolus in B, and both in C, project into the cytoplasm (cf. Fig. 1).111: Nucleus of a small neuron with two nucleoli; 5-pm sections. The nucleus is cut into 3 (A, B, D) or 2 (C) slices of which 1 or 2 are seen in the microscope. Both nucleoli are visible in A and C; the lower one in B may be masked by the dark cytoplasm. The nuclear caps 1 and 3 in D are lost together with both nucleoli. Counting neurons in spinal ganglia is usually desections, or that the sections were mounted in the wrong order. The thickness of these sections was controlled as scribed as routine procedure, and it is stated (or assumed) that each cell contains one nucleolus such that well. one only has to calculate a correction factor for split DISCUSSION nucleoli. Bondok and Sansone (1984) believe that this The ganglia L4-L6 contained 41,000 neurons when problem is eliminated by counting spaced sections-that series of 3-pm resin sections were used for counting. is, their data in reality are raw counts. Only Hulsebosch There were 23% fewer cells counted in series of 5-pm et al. (1986) encountered difficulties. The ganglion cells paraffin sections, but even the counts in paraffin sec- of their rats, as in the present study, had multiple nutions were considerably higher than previously reported cleoli and the counts in T4 and S2 ganglia were unredata. The mean data for the L5 ganglion now range producible when different investigators were involved. from 2,000 (Feringa et al., 1985) to 15,000 cells (this As the identification of nucleolar profiles entailed an article) (Table 1). Most of the present data are from element of arbitrariness, empirical correction factors operated rats, but it is unlikely that contralateral nerve were determined for each investigator (Coggeshall et al., section increases the number of ganglion cells and the 1984). Hulsebosch et al. (19861, however, did not control counts are the same in unoperated rats (Table 2). The whether, in consecutive sections, at least one nucleolus only explanation is that the variation in counts is due was recognized in each cell. Counting nucleolar profiles to different histological methods. The main problem en- in resin sections is unambiguous but the use of 3-pm countered after paraffin embedding in this study was sections may lead to an overestimate because the section the identification of nucleoli in the smallest cells. Their thickness is close to the size of the nucleoli. The correcnucleoli were situated in the periphery of nuclei that tion factors actually represent the main source of error. measured only twice the thickness of a 5-pm sectiort, On the other hand, even the largest possible error in and many projected into the intensely stained cytoplasm. determining a factor does not explain the variation in The thickness of the sections in previous studies of rat neuron counts reported. The prerequisites for applying correction formulas for lumbar ganglia was 7-10 pm. In this study, stained sections thicker than 5 pm were found t o be unsuitable split nucleoli (Abercrombie, 1946; Konigsmark, 1970) for counting nucleoli of small cells. Nevertheless, thick are 1)that each cell have only one nucleolus, 2) that the sections are not the only cause of low neuron counts. nucleoli be cut rather than pushed by the knife, and 3) The number of visible nucleoli may also depend on the that all fragments, or at least those larger than 1 pm preservation of the tissue. Paraffin-embedded specimens (Konigsmark, 19701, be identified. None of these condiin comparison with resin-embedded specimens in this tions was fulfilled in the present study. Nucleoli in parstudy were technically poor, but probably not below the affin sections may be pushed by the knife edge common standard. Feringa et al. (1985) embedded and (Cammermeyer, 1967)and the calculated factors are too cut the entire vertebral column after decalcification with small. The visibility of a structure depends on its density acid. This might have been rather harmful to the struc- in relation to the background and on its delineation. The relative density of a nucleolus projecting into the densely ture of the neurons (Table 1). NUMBER OF NEURONS IN RAT DRG L4-L6 321 stained cytoplasm is low, and the delineation of a nu- presumably myelinated. The number of neurons with cleolar cap is poor because its surface projects obliquely. unmyelinated axons is probably larger in L6 ganglia The effect of poor delineation is obvious for caps of large and the number of ganglion cells at L4-L6 may be nucleoli, but caps of small nucleoli are invisible and the roughly three times that of myelinated dorsal root finucleoli of small neurons never appear split (Figs. 1,4). bers. Nevertheless, the large and nonsymmetric anastoThis indicates that the error due to split nucleoli has moses between adjacent dorsal roots (Jacob and Weddell, previously been overrated, whereas the identification 1975; Schmalbruch, 1987) may distort this relation for problem has practically been neglected. individual ganglia. The fact that even large fragments are lost is illus- The assumed 1:l relation of ganglion cells and nerve trated by the appearance of nuclei in consecutive sec- fibers in dorsal roots or peripheral nerves refers to the tions. Counting nuclei instead of nucleoli to determine established view that each ganglion cell has one central the number of neurons (Smolen et al., 1983) appears and one peripheral process (Dogiel, 1896; Ram6n y Caattractive because nuclei are large and easily seen and jal, 1909). This has been contested because the number because there is only one per cell. The correction factors of central or peripheral axons exceeded the neuronal obtained by using Abercrombie’s (1946) formula for 10- counts in the corresponding ganglia by 30-120% (Langpm and 20-pm nuclei, respectively, for 3-pm sections are ford and Coggeshall, 1979,1981;Hulsebosch et al., 1986). 0.23 and 0.13. This means that a small nucleus should The number of neurons was determined from series of appear in 4-5 sections and a large one in 7-8 sections. paraffin sections and therefore might have been In fact, the number of nuclear profiles that were visible underestimated. in consecutive sections was even smaller than the num- The results presented in this paper indicate that neuber obtained by simply dividing section thickness into ronal counts in rat spinal ganglia embedded in paraffin nucleus diameter (Figs. 1, 4). Smolen et al. (19831, in a tend to be too low because the nucleoli of some small study of the superior cervical ganglion of the rat, found neurons are not detectable, and that this error increases that slices of less than a 6-pm diameter from 9-pm nuclei with the thickness of the sections and with poor tissue were not identifiable in 1.5-pm resin sections. Neuron preservation. Applying calculated correction factors excounts for this ganglion range from 13,000 to 45,000 (for acerbates the problem. Thin paraffin sections may be references see Smolen et al., 1983), and it appears as if suitable for comparing pairs of ganglia after unilateral the discrepant findings in lumbar ganglia are not an experimental procedures, but probably not for comparisolated phenomenon. Neuron counts in other systems ing different populations of rats. Cutting and staining complete series of 3-pm resin series of 6 ganglia for each might need confirmation as well. The counts in unstained 12-pm paraftin sections were rat is tedious and time-consuming and hardly suited for also underestimates but they were larger than previous studies using many animals. Whether other staining counts (Tables 1, 2). The method was used by Hatai methods will provide more satisfying solutions remains (1902), who found 9,000 cells in the L2 ganglion, which to be seen. Thick unstained sections of osmicated tissue has about the same size as the L4 ganglion, and 70% (Hatai, 1902)represent a relatively convenient means t o more cells in the T4 ganglion than Hulsebosch et al. estimate the number of neurons and this method might (1986) found in stained 10-pm sections. Because the in- be improvable by prefixation with aldehyde or the use fluence of split and multiple nucleoli is negligible in 12- of embedding media other than paraffin. pm sections, this difference must be due to the fact that ACKNOWLEDGMENTS the nucleoli are easier to recognize after osmication. Reducing the sample size to half in my counts had very The author thanks Mrs. M. Bjarg for unfailing technilittle effect on the results. This excludes nonsystematic cal assistance and for meticulously cutting the many counting errors. serial sections. The Danish Medical Research Council, The claim that the number of neurons in the lumbar and the Foundations for Experimental Research in Neuganglia supplying the rat sciatic nerve has hitherto been rology and for Progress in Medicine (A.P. Mdller), prounderestimated is supported by circumstantial evi- vided financial support. I am grateful to Dr. Roy J. Levin dence. The nerve at midthigh contains 1,600 motor and for correcting my English. 19,000 sensory axons (Schmalbruch, 1986a).The ventral roots L4-L6 together contain 4,000 myelinated fibers (Schmalbruch, 1984). The sensory-motor fiber ratio in LITERATURE CITED the nerve suggests that the corresponding spinal ganglia rather contain 40,000 than 20,000 or even fewer Abercrombie, M. (1946) Estimation of nuclear population from microtome sections. Anat. Rec., 94:239-247. neurons. The L6 dorsal root contains 10,000-14,000 axons (Langford and Coggeshall, 1979). The dorsal roots Arvidsson, J., J. Ygge, and G. Grant (1986) Cell loss in lumbar dorsal root ganglia and transganglionic degeneration after sciatic nerve L4 and L5 each contain more than 4,000 myelinated resection in the rat. Brain Res., 373:15-21. fibers (Schmalbruch, 1986b);unfortunately data for un- Bondok, A.A., and F.M. Sansone (1984) Retrograde and transganglionic degeneration of sensory neurons after a peripheral nerve myelinated fibers are not available. The myelinated filesion at birth. Exp. Neurol., 86:322-330. bers in thoracic and sacral dorsal roots represent only J. (1967) Artifactual displacement of neuronal nucleoli 23-33% of all fibers (Langford and Coggeshall, 1979, Cammermeyer, in paraffin sections. J. Hirnforsch. 9:209-224. 1981; Hulsebosch and Coggeshall, 1983; Hulsebosch et Coggeshall, R.E., K. Chung, D. Greenwood, and C.E. Hulsebosch (1984) An empirical method for converting nucleolar counts to neuronal al., 1986);hence, the dorsal roots L4 and L5 may contain numbers. J. Neurosci. Methods, 12:125-132. between 13,000 and 19,000 axons. These figures comM., and R. Govrin-Lippmann (1985) Neurogenesis in adult rat pare well with the neuron counts in this study. Harper Devor, dorsal root ganglia. Neurosci. Lett., 61:189-194. and Lawson (19851, in a physiological study of L4 gan- Devor, M., R. Govrin-Lippmann, I. Frank, and P. Raber (1985) Proliferation of primary sensory neurons in adult rat dorsal root ganglia, found 34-45% neurons with fast-conducting axons, 322 H. SCHMALBRUCH glion and the kinetics of retrograde cell loss after sciatic nerve section. Somatosensory Res., 3t139-167. Dogiel, A.S. (1896) Der Bau der Spinalganglien bei den Saugetieren. Anat. Am., 12t140-152. Feringa, E.R., G.W. Lee, H.L. Vahlsing, and W.J. Gilbertie (1985) Cell death in the adult rat dorsal root ganglion after hind limb amputation, spinal cord transection, or both operations. Exp. Neurol., 87:349-357. Harper, A.A., and S.N. Lawson (1985)Conduction velocity is related to morphological cell type in rat dorsal root ganglion neurones. J. Physiol., 359:31-46. Hatai, S. (1902) Number and size of the spinal ganglion cells and dorsal root fibers in the white rat at different ages. J. Comp. Neurol., 12:107-124. Hulsebosch, C.E., and R.E. Coggeshall (1983) A comparison of axonal numbers in dorsal roots following spinal cord hemisection in neonate and adult rats. Brain Res., 265t187-197. Hulsebosch, C.E., E. Coggeshall, and K. Chung (1986) Numbers of rat dorsal root axons and ganglion cells during postnatal development. Dev. Brain Res., 26r105-113. Jacob, M., and G. Weddell (1975) Neural intersegmental connection in the spinal root and ganglion region of the rat. J. Comp. Neurol., 161:115-124. Konigsmark, B.W. (1970) Methods for the counting of neurons. In: Contemporary Research Methods in Neuroanatomy. W.J.H. Nauta and S.O.E. Ebbesson, eds. Springer-Verlag, New York, pp. 315-340. Langford, L.A., and R.E. Coggeshall(1979)Branching of sensory axons in the dorsal root and evidence for the absence of dorsal root efferent fibers. J. Comp. Neurol., 184t193-204. Langford, L.A., and R.E. Coggeshall(1981)Branching of sensory axons in the peripheral nerve of the rat. J. Comp. Neurol., 203t745-750. Ram6n y Cajal, S. (1909) Histologie du Systeme Nerveux de 1’Homme et des Vertebres. Consejo Superior de Investigaciones Cientificas. Madrid Instituto Ramon y Cajal. Reprinted in 1952. Schmalbruch, H. (1984) Motoneuron death after sciatic nerve section in newborn rats. J. Comp. Neurol., 224:252-258. Schmalbruch, H. (1986a) Fiber composition of the rat sciatic nerve. Anat. Rec., 215t71-81. Schmalbruch, H. (198613) The anatomy and fiber composition of the normal rat sciatic nerve. Clin. Neuropathol., 5:115. Schmalbruch, H. (1987) Loss of sensory neurons after sciatic nerve section in the rat. Anat. Rec., 219:323-329. Smolen, A.J., L.L. Wright, and T J . Cunningham (1Y83JNeuron numbers in the superior cervical sympathetic ganglion of the rat: A critical comparison of methods for cell counting. J. Neurocytol., 12t739-750. Ygge, J., H. Aldskogius, and G. Grant (1981) Asymmetries and symmetries in the number of thoracic dorsal root ganglion cells. J. Comp. Neurol., 202t365-372. Yip, H.K., K.M. Rich, P.A. Lampe, and E.M. Johnson (1984)The effects of nerve growth factor and its antiserum on the postnatal development and survival after injury of sensory neurons in rat dorsal root ganglia. J. Neurosci., 4:2986-2992.