1
Brain Research, 584 (1992) 1-10 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00
BRES 17907
Research Reports
Networks formed by dorsal root ganglion neurites within spinal cord explants: a computer-aided analysis of HRP intracellularly labeled neurons M a r i e - C h r i s t i n e C a l v e t a, J e a n C a l v e t b, J e a n - R e n d T e i l h a c a a n d M a r i e - J e a n n e D r i a n a a
INSERM U 336 and b EPHE Labo lnformatique Appliqu#e h la Neurobiotogie, Montpellier (France) (Accepted 25 February 1992)
Key words: Dorsal root ganglion neuron; Spinal cord; Organotypic culture; Intracellular horseradish peroxidase; Computer microscope; Quantitative morphology
Dorsal root ganglion (DRG) neurons from rat embryos were explanted either alone or with the attached spinal cord (SC). Neuritic processes were mapped out histologically using an intracellular iontophoretic HRP method. Computer reconstructions and morphometric parameters allowed a comparative quantitative analysis of the DRG neurons and of their neuritic processes in both models. The first model, strongly dependent on nerve growth factor (NGF) for its survival, developed large multidirectional processes. The second model showed a bipolar distribution of its neuritic processes, the central one entering predominantly the dorsolateral part of the cord explant and ramifying both homoand heterolaterally. The quantitative data revealed a significant decrease in the overall size of the neuritic networks of the second model (with attached SC). A discriminant analysis permitted the recognition of these two populations of DRG neurons. The role of the spinal cord explant, and more precisely the target cells of the dorsal horn, was considered to be a prominent factor in the development of the DRG neuritic networks.
INTRODUCTION Since the earliest in vivo studies of Caja123, it has been well known that sensory afferents from dorsal root ganglion (DRG) neurons enter and make a majority of their connections within the dorsal horn of the spinal cord (SC). This selectivity can occur under appropriate conditions in vitro and has already been demonstrated by both electrophysiological3'11-14'2t and classic morphological methods 6,22,27,2s. However, these histological procedures have not allowed the complete visualization of the DRG neurites and their terminal arborizations within attached SC explants. A more reliable histological procedure, i.e. horseradish peroxidase (HRP) by iontophoretic extraceUular injections, has been successfully used to trace DRG fibers in mouse cord explants 1'26. In the present study we have applied to the organotypic cultures of DRG neurons the same technique of intracellular iontophoretic injection of HRP as used in our previous studies on cultured Purkinje cells7,9. Only such an individual intraceUular labeling of the DRG neurons is able to allow a complete visualization not
only of the central (within SC) but also of the peripheral neurites and to permit a three-dimensional reconstruction of the whole cell. DRG neurons with attached SC were thus morphologically analysed and compared to isolated DRG neurons cultured alone, and our quantitative morphometric data allowed a good separation of these two populations. MATERIALS AND METHODS
Cultures In one series of cultures, spinal cord (SC) with attached DRGs and meningeal covering from 13-14-day fetal rats was cut into 0.5- to 1-mm-thick transverse cross-sections. In a second series of cultures, isolated DRGs detached from spinal cord were explanted from 16-17-day fetal rats. In a third series the cord explants from 13-day fetal rats were cultured alone for 6-16 days prior to introduction of 15-16-day fetal rat DRGs. In all cases tissues were explanted on collagen-coated coverslips with a drop of nutrient fluid and sealed at 35°C in a Maximow assembly. Except for the isolated DRG and for the DRGs introduced as cocultures which required nerve growth factor (NGF, 7S fraction) (Collaborative Research) at 50 ng/ml of nutrient medium, at least at the explantation time 13'iS'Is, the nutrient (renewed twice a week) consisted of 40% MEM (minimal essential medium), 30% fetal calf serum, 25% Hanks' BSS, 5% chick embryo extract ultrafiltrate (Gibco) supplemented with glucose 600 mg/100 ml. In all cultures the chick embryo extract was suppressed after the
Correspondence: M.C. Caivet, INSERM U 336, USTL Case Postale 106, Place Eugene Bataillon, 34095 Montpellier Cedex 05, France. Fax: (33) (67) 14.33.18.
first week in vitro and replaced by M E M at 45% instead of 40%. In some cultures of D R G s with attached SC (first model), N G F was added to the medium at lower levels than those previously described 13 in order to know if this model was affected by the presence of N G F when cocultured with SC.
HRP lntracellular iontophoretic labeling of DRG neurons For H R P studies of the D R G neurons the culture was transferred to a chamber attached to the stage of a movable inverted phase-contrast microscope s. The selected D R G neurons, viewed under × 250 magnification were first impaled and then stained with H R P according to a previously described technique 7"9'I°. Briefly, single-barrel glass micropipettes were filled with a 10% solution of H R P in Tris-HCl buffer at pH 8.4. After impalement, positive DC current was delivered in 0.6 Hz square wave pulses with an intensity of less than 6nA for 1-3 min. After diffusion, the culture was stained and then incubated in a solution containing diaminobenzidine and hydrogen peroxide as previously described 7.
Computer-assisted acquisition and processing of three-dimensional morphological data Each cultured HRP-filled D R G cell was first drawn through the camera lucida with a x 9 5 oil immersion objective and with a polarizing system (Calvet, unpublished observation) thus allowing visualization of the upper or lower crossings of neuritic processes and their differentiation from true bifurcations. T h e n the cultured neuron was treated by the microprocessor which allowed acquisition and processing of three-dimensional data and coded the locations of neurite origins, bifurcations and terminals 9'1°. All the points of the neuritic arborization, coded in a centrifugal direction, were stored sequentially and all the neuritic processes successively. The resulting file could then be used for all subsequent quantitative studies and for computer reconstruction of the neurons. When the spinal cord explant was present its contours were drawn and reconstructed by the microprocessor in the same way as for the neuritic processes.
Statistical analysis of quantitative morphological parameters T h e method of network analysis outlined by Calvet et al. 9 was used. All the computed values were calculated with their m e a n s and standard deviation (S.D.). Statistical significance was tested by Student's t-test 9. Fourteen variables listed in Table I were thus compared in different D R G neuron populations in organotypic cultures. A m o n g these variables some were a n u m b e r (variables 1, 4, 5, 8, 11) and others a dimension (variables 2, 3, 6, 7, 9, 10, 12, 13, 14). The ratio between the total neuritic length and the neuritic field area (rfla, var. 4) which expressed the branching density and the ratio between the number of terminals and the n u m b e r of primary neurites (rtpn, vat. 11) was also computed. The n u m b e r of primary neuritic processes
(nbne, var. 1), the total number of segments (tnse, var. 5) and the number of rows (nrow, var. 8 expressing the greatest n u m b e r of segments successively arranged in centrifugal order) were thus computed and averaged for all series of cultured D R G neurons. The lengths of the various kinds of segments (mdps, var. 10: mite, var. 13; reals, var. 14), the total segment length (omsl, var. 12) and the total neuritic length (tnle, var. 2) were calculated as were their means. The dimensions of the domain of the neuritic processes could then be evaluated both by measuring the average distances (mple, var. 7), the maximum distance (mdts, var. 9) and the mean lengths (mlts, var. 6) of all the terminal tips from the soma and by calculating the total surface (nfar, vat. 3) located between the maximum and minimum values of the x - y coordinates ,~,1~ Sixteen HRP-stained isolated D R G neurons were thus compared to twenty D R G neurons with attached spinal cord. These neurons, grown from 5 to 38 days in organotypic cultures, were considered as two different populations since the experimental procedures described above (section 1 of Materials and Methods) did not allow separating (1) the isolated D R G neurons from those added to SC explants in which they did not develop any connection and (2) the D R G neurons attached to SC and cultured with N G F from those cultured without NGF. RESULTS
Qualitative morphological data Isolated DRG neurons cultured alone. These neurons cultured alone from 15-17-day-old fetuses or added to and co-cultured with SC explants from 13-14-day-old rat fetuses developed the same morphological patterns. Since our models of D R G neurons added to the SC explant never made any morphological connections with the SC, these neurons were considered as being similar to those explanted alone except that (1) they could survive in culture for several weeks and (2) they did not need to survive the addition of N G F to the nutrient medium. Fig. 1 represents the neuritic processes of about 8 H R P individually labeled D R G neurons from an isolated ganglion explanted from a 16-day-old fetus and cultured for 9 days. The well-developed neuritic processes are numerous and extend in multidirectional
TABLE I
List of the 14 variables used in the quantitative morphometric analysis with their discriminative power (percentage of well-classified individuals in their own group) Note that the values beneath 75% have too low a discriminating power.
Variable
Description
Classification %
1, nbne 2, tnle 3, nfar 4, rlfa 5, tnse 6, mlts 7, mple 8, nrow 9, mdts 10, mdps 11, rtpn 12, omsl 13, mite 14, mals
n u m b e r of primary neurites total neuritic length (microns) neuritic field area (squares of 50 microns) total neuritic l e n g t h / n e u r i t i c field area total n u m b e r of segments m e a n length of terminals to soma (microns) m e a n path length (microns) m e a n distance terminals-soma n u m b e r of segments successively arranged in centrifugal order m a x i m u m distance terminals-soma (microns) m e a n distance first bifurcation-soma (microns) number of t e r m i n a l s / n u m b e r of primary neurites overall m e a n segment length (microns) m e a n length of terminals (microns) m a x i m u m length of all segments (microns)
69 * 84 *84 47 "81 61 58 *75 66 *78 69 58 55 61
3 arrays: neither peripheral nor central (or axonal) branches can be distinguished. D R G neurons cultured with attached SC. A t the time of explantation the spinal ganglia are located in close proximity of the SC explant and lie at the junction of the dorsal and ventral roots 1'2'1~'2a. As both these roots grow the spinal ganglion is driven away from the SC explant but their relative position remains unchanged.
These D R G neurons have mostly bipolarly oriented neuritic processes (Figs. 2 and 3A): peripheral and central branches both emerging from the soma (Fig. 3A) are easily recognizable (Fig. 2). Peripheral processes ramify profusely in several directions (Fig. 2) whereas central axonal branches enter the SC explant via the dorsal roots and then ramify in the dorsal horn area (Fig. 2). In these healthy cultures, cells display
250m
Fig. 1. Darkfield photomicrograph of HRP-stained individual D R G neurons from a 16-day-old fetal rat cultured alone for 9 days. One of the intracellularly HRP-labeled neurons is indicated by an arrowhead. Note that the neuritic processes are numerous and develop in a multipolar way.
4
250jam
Fig. 3. HRP-stained D R G neuronal somata with bipolar processes (A) from a D-day-old fetal rat cultured for 3 weeks with its attached SC. B1 and B2 are HRP-filled growth cones from peripheral processes of similar materials. Bars: A = 20/~m, B = 10/.~m.
several growth cones at their peripheral neuritic tip as shown in Fig. 3 B 1 - B 2 .
Three-dimensional reconstructions of the cultured DRG neurons T h e individual intracellular iontophoretic labeling o f
D R G n e u r o n s with attached SC explant, as used here, allowed a precise visualization of both peripheral and central neuritic processes. This technique eliminated several factors limiting the resolution o f visualizing D R G fibers within cord explants as had b e e n encountered in previous studies 26-28. With our technique it
Brain Research, 584 (1992) 1-10 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00
BRES 17907
Research Reports
Networks formed by dorsal root ganglion neurites within spinal cord explants: a computer-aided analysis of HRP intracellularly labeled neurons M a r i e - C h r i s t i n e C a l v e t a, J e a n C a l v e t b, J e a n - R e n d T e i l h a c a a n d M a r i e - J e a n n e D r i a n a a
INSERM U 336 and b EPHE Labo lnformatique Appliqu#e h la Neurobiotogie, Montpellier (France) (Accepted 25 February 1992)
Key words: Dorsal root ganglion neuron; Spinal cord; Organotypic culture; Intracellular horseradish peroxidase; Computer microscope; Quantitative morphology
Dorsal root ganglion (DRG) neurons from rat embryos were explanted either alone or with the attached spinal cord (SC). Neuritic processes were mapped out histologically using an intracellular iontophoretic HRP method. Computer reconstructions and morphometric parameters allowed a comparative quantitative analysis of the DRG neurons and of their neuritic processes in both models. The first model, strongly dependent on nerve growth factor (NGF) for its survival, developed large multidirectional processes. The second model showed a bipolar distribution of its neuritic processes, the central one entering predominantly the dorsolateral part of the cord explant and ramifying both homoand heterolaterally. The quantitative data revealed a significant decrease in the overall size of the neuritic networks of the second model (with attached SC). A discriminant analysis permitted the recognition of these two populations of DRG neurons. The role of the spinal cord explant, and more precisely the target cells of the dorsal horn, was considered to be a prominent factor in the development of the DRG neuritic networks.
INTRODUCTION Since the earliest in vivo studies of Caja123, it has been well known that sensory afferents from dorsal root ganglion (DRG) neurons enter and make a majority of their connections within the dorsal horn of the spinal cord (SC). This selectivity can occur under appropriate conditions in vitro and has already been demonstrated by both electrophysiological3'11-14'2t and classic morphological methods 6,22,27,2s. However, these histological procedures have not allowed the complete visualization of the DRG neurites and their terminal arborizations within attached SC explants. A more reliable histological procedure, i.e. horseradish peroxidase (HRP) by iontophoretic extraceUular injections, has been successfully used to trace DRG fibers in mouse cord explants 1'26. In the present study we have applied to the organotypic cultures of DRG neurons the same technique of intracellular iontophoretic injection of HRP as used in our previous studies on cultured Purkinje cells7,9. Only such an individual intraceUular labeling of the DRG neurons is able to allow a complete visualization not
only of the central (within SC) but also of the peripheral neurites and to permit a three-dimensional reconstruction of the whole cell. DRG neurons with attached SC were thus morphologically analysed and compared to isolated DRG neurons cultured alone, and our quantitative morphometric data allowed a good separation of these two populations. MATERIALS AND METHODS
Cultures In one series of cultures, spinal cord (SC) with attached DRGs and meningeal covering from 13-14-day fetal rats was cut into 0.5- to 1-mm-thick transverse cross-sections. In a second series of cultures, isolated DRGs detached from spinal cord were explanted from 16-17-day fetal rats. In a third series the cord explants from 13-day fetal rats were cultured alone for 6-16 days prior to introduction of 15-16-day fetal rat DRGs. In all cases tissues were explanted on collagen-coated coverslips with a drop of nutrient fluid and sealed at 35°C in a Maximow assembly. Except for the isolated DRG and for the DRGs introduced as cocultures which required nerve growth factor (NGF, 7S fraction) (Collaborative Research) at 50 ng/ml of nutrient medium, at least at the explantation time 13'iS'Is, the nutrient (renewed twice a week) consisted of 40% MEM (minimal essential medium), 30% fetal calf serum, 25% Hanks' BSS, 5% chick embryo extract ultrafiltrate (Gibco) supplemented with glucose 600 mg/100 ml. In all cultures the chick embryo extract was suppressed after the
Correspondence: M.C. Caivet, INSERM U 336, USTL Case Postale 106, Place Eugene Bataillon, 34095 Montpellier Cedex 05, France. Fax: (33) (67) 14.33.18.
first week in vitro and replaced by M E M at 45% instead of 40%. In some cultures of D R G s with attached SC (first model), N G F was added to the medium at lower levels than those previously described 13 in order to know if this model was affected by the presence of N G F when cocultured with SC.
HRP lntracellular iontophoretic labeling of DRG neurons For H R P studies of the D R G neurons the culture was transferred to a chamber attached to the stage of a movable inverted phase-contrast microscope s. The selected D R G neurons, viewed under × 250 magnification were first impaled and then stained with H R P according to a previously described technique 7"9'I°. Briefly, single-barrel glass micropipettes were filled with a 10% solution of H R P in Tris-HCl buffer at pH 8.4. After impalement, positive DC current was delivered in 0.6 Hz square wave pulses with an intensity of less than 6nA for 1-3 min. After diffusion, the culture was stained and then incubated in a solution containing diaminobenzidine and hydrogen peroxide as previously described 7.
Computer-assisted acquisition and processing of three-dimensional morphological data Each cultured HRP-filled D R G cell was first drawn through the camera lucida with a x 9 5 oil immersion objective and with a polarizing system (Calvet, unpublished observation) thus allowing visualization of the upper or lower crossings of neuritic processes and their differentiation from true bifurcations. T h e n the cultured neuron was treated by the microprocessor which allowed acquisition and processing of three-dimensional data and coded the locations of neurite origins, bifurcations and terminals 9'1°. All the points of the neuritic arborization, coded in a centrifugal direction, were stored sequentially and all the neuritic processes successively. The resulting file could then be used for all subsequent quantitative studies and for computer reconstruction of the neurons. When the spinal cord explant was present its contours were drawn and reconstructed by the microprocessor in the same way as for the neuritic processes.
Statistical analysis of quantitative morphological parameters T h e method of network analysis outlined by Calvet et al. 9 was used. All the computed values were calculated with their m e a n s and standard deviation (S.D.). Statistical significance was tested by Student's t-test 9. Fourteen variables listed in Table I were thus compared in different D R G neuron populations in organotypic cultures. A m o n g these variables some were a n u m b e r (variables 1, 4, 5, 8, 11) and others a dimension (variables 2, 3, 6, 7, 9, 10, 12, 13, 14). The ratio between the total neuritic length and the neuritic field area (rfla, var. 4) which expressed the branching density and the ratio between the number of terminals and the n u m b e r of primary neurites (rtpn, vat. 11) was also computed. The n u m b e r of primary neuritic processes
(nbne, var. 1), the total number of segments (tnse, var. 5) and the number of rows (nrow, var. 8 expressing the greatest n u m b e r of segments successively arranged in centrifugal order) were thus computed and averaged for all series of cultured D R G neurons. The lengths of the various kinds of segments (mdps, var. 10: mite, var. 13; reals, var. 14), the total segment length (omsl, var. 12) and the total neuritic length (tnle, var. 2) were calculated as were their means. The dimensions of the domain of the neuritic processes could then be evaluated both by measuring the average distances (mple, var. 7), the maximum distance (mdts, var. 9) and the mean lengths (mlts, var. 6) of all the terminal tips from the soma and by calculating the total surface (nfar, vat. 3) located between the maximum and minimum values of the x - y coordinates ,~,1~ Sixteen HRP-stained isolated D R G neurons were thus compared to twenty D R G neurons with attached spinal cord. These neurons, grown from 5 to 38 days in organotypic cultures, were considered as two different populations since the experimental procedures described above (section 1 of Materials and Methods) did not allow separating (1) the isolated D R G neurons from those added to SC explants in which they did not develop any connection and (2) the D R G neurons attached to SC and cultured with N G F from those cultured without NGF. RESULTS
Qualitative morphological data Isolated DRG neurons cultured alone. These neurons cultured alone from 15-17-day-old fetuses or added to and co-cultured with SC explants from 13-14-day-old rat fetuses developed the same morphological patterns. Since our models of D R G neurons added to the SC explant never made any morphological connections with the SC, these neurons were considered as being similar to those explanted alone except that (1) they could survive in culture for several weeks and (2) they did not need to survive the addition of N G F to the nutrient medium. Fig. 1 represents the neuritic processes of about 8 H R P individually labeled D R G neurons from an isolated ganglion explanted from a 16-day-old fetus and cultured for 9 days. The well-developed neuritic processes are numerous and extend in multidirectional
TABLE I
List of the 14 variables used in the quantitative morphometric analysis with their discriminative power (percentage of well-classified individuals in their own group) Note that the values beneath 75% have too low a discriminating power.
Variable
Description
Classification %
1, nbne 2, tnle 3, nfar 4, rlfa 5, tnse 6, mlts 7, mple 8, nrow 9, mdts 10, mdps 11, rtpn 12, omsl 13, mite 14, mals
n u m b e r of primary neurites total neuritic length (microns) neuritic field area (squares of 50 microns) total neuritic l e n g t h / n e u r i t i c field area total n u m b e r of segments m e a n length of terminals to soma (microns) m e a n path length (microns) m e a n distance terminals-soma n u m b e r of segments successively arranged in centrifugal order m a x i m u m distance terminals-soma (microns) m e a n distance first bifurcation-soma (microns) number of t e r m i n a l s / n u m b e r of primary neurites overall m e a n segment length (microns) m e a n length of terminals (microns) m a x i m u m length of all segments (microns)
69 * 84 *84 47 "81 61 58 *75 66 *78 69 58 55 61
3 arrays: neither peripheral nor central (or axonal) branches can be distinguished. D R G neurons cultured with attached SC. A t the time of explantation the spinal ganglia are located in close proximity of the SC explant and lie at the junction of the dorsal and ventral roots 1'2'1~'2a. As both these roots grow the spinal ganglion is driven away from the SC explant but their relative position remains unchanged.
These D R G neurons have mostly bipolarly oriented neuritic processes (Figs. 2 and 3A): peripheral and central branches both emerging from the soma (Fig. 3A) are easily recognizable (Fig. 2). Peripheral processes ramify profusely in several directions (Fig. 2) whereas central axonal branches enter the SC explant via the dorsal roots and then ramify in the dorsal horn area (Fig. 2). In these healthy cultures, cells display
250m
Fig. 1. Darkfield photomicrograph of HRP-stained individual D R G neurons from a 16-day-old fetal rat cultured alone for 9 days. One of the intracellularly HRP-labeled neurons is indicated by an arrowhead. Note that the neuritic processes are numerous and develop in a multipolar way.
4
250jam
Fig. 3. HRP-stained D R G neuronal somata with bipolar processes (A) from a D-day-old fetal rat cultured for 3 weeks with its attached SC. B1 and B2 are HRP-filled growth cones from peripheral processes of similar materials. Bars: A = 20/~m, B = 10/.~m.
several growth cones at their peripheral neuritic tip as shown in Fig. 3 B 1 - B 2 .
Three-dimensional reconstructions of the cultured DRG neurons T h e individual intracellular iontophoretic labeling o f
D R G n e u r o n s with attached SC explant, as used here, allowed a precise visualization of both peripheral and central neuritic processes. This technique eliminated several factors limiting the resolution o f visualizing D R G fibers within cord explants as had b e e n encountered in previous studies 26-28. With our technique it
