- Email: [email protected]
Comparing the efficacy of two fluorescent retrograde tracers in labeling the motor and sensory neuron populations of the rat sciatic nerve
Journal of Neuroscience Methods 114 (2002) 159– 164 www.elsevier.com/locate/jneumeth
Comparing the efficacy of two fluorescent retrograde tracers in labeling the motor and sensory neuron populations of the rat sciatic nerve C.T. Byers a, R. Fan b, A. Messina b, W.A. Morrison c, M.P. Galea a,* a b
School of Physiotherapy, The Uni6ersity of Melbourne, 200 Berkeley Street, Park6ille, Victoria 3052, Australia Bernard O’Brien Institute of Microsurgery, St Vincent’s Hospital Melbourne, Fitzroy, Victoria 3065, Australia c Department of Surgery, The Uni6ersity of Melbourne, Victoria 3010, Australia Received 12 April 2001; received in revised form 30 November 2001; accepted 30 November 2001
Abstract We compared the efficacy with which the fluorescent tracers Fast Blue (FB) and Diamidino Yellow (DY) retrogradely label neutrons. Trace crystals were applied to the sciatic nerve exclusively (single label) or serially (double label). Unbiased cell counts showed that FB and DY label similar numbers of motoneurons (P = 1.00, df 5) or DRG neurons (P= 0.95, df 5) when applied exclusively. Plotting of motoneurons revealed a similar pattern of distribution of FB and DY labeled neurons. When the tracers were applied serially, 79% of labeled motoneurons and 77% of labeled DRG neurons were double-labeled irrespective of which tracer was applied first. Equal proportions of the remaining labeled neurons were single-labeled with FB or DY. These data show that FB and DY label equal numbers of motor and sensory neurons of the sciatic nerve following exclusive or serial application of tracers. These findings support the use of FB and DY together in serial fluorescent labeling experiments. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Fast Blue (FB); Fluorescent tracers; Diamidino Yellow (DY); Motoneurons; Dorsal root ganglion (DRG)
1. Introduction The application of a series of different fluorescent retrograde tracers to peripheral nerves allows for the comparison of pre- and post-injury neuron populations that project to the lesion site (Richmond et al., 1994; Puigdellı´vol-Sa´nchez et al., 2000). However, when two different tracers are applied serially for this purpose, the value of qualitative and quantitative comparisons depends on the similarity of uptake characteristics of the respective tracers. If valid quantitative comparisons are to be made, two principal issues should be considered. Firstly, it is important to establish that the two tracers being used label similar numbers of neurons and with a similar distribution in the tissue being studied. Although the * Corresponding author. Tel.: + 61-3-8344-4118; fax: +61-3-83444188. E-mail address: [email protected] (M.P. Galea).
efficacy of different fluorescent retrograde tracers labeling neurons projecting to the peripheral nervous systems has been previously investigated (Horikawa and Powell, 1986; Haase and Payne, 1990; Richmond et al., 1994; Puigdellı´vol-Sa´nchez et al., 2000), few studies have utilized modern, unbiased stereological techniques, such as the optical disector (Gundersen et al., 1988), to estimate labeled neuron numbers. Thus, variables such as cell size and shape, the thickness of histological sections and easily-violated assumptions bring the results of studies using profile and assumption-based counting methods into question (Coggeshall and Lekan, 1996). Secondly, if tracers are to be applied serially, it is also important to establish whether the presence of the initial tracer affects labeling by the subsequently applied tracer. A recent study reported a ‘blocking’ effect of the tracer Fast Blue (FB) over Diamidino Yellow (DY) when they were applied serially as aqueous solutions to the proximal stump of a transected rat sciatic
0165-0270/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 5 - 0 2 7 0 ( 0 1 ) 0 0 5 2 0 - 9
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164
160
nerve (Puigdellı´vol-Sa´ nchez et al., 2000). However, the possibility of a ‘blocking’ effect of FB over DY following serial applications warrants further investigation, to determine if such an interaction occurs when using our previously published method of applying tracer crystals to the transected rat sciatic nerve (Sangster et al., 1999). This study was undertaken with the purpose of establishing whether: 1. FB and DY applied separately labeled a similar number of motoneurons and DRG neurons and a similar distribution of motoneurons, and; 2. Pre-labeling with one tracer affected the subsequent labeling by a second tracer applied at a later time.
within one episode of surgery, utmost care was taken to prevent tracer cross-contamination. In particular, only one sciatic nerve was left exposed at a time and for each operative side a separate set of surgical and tracer application instruments were used. Animals in group B underwent a second episode of surgery 4 days after the first. For each initial surgery site the incision was reopened and the sciatic nerve was trimmed to remove the distal 5 mm of the proximal stump. This enabled application of the complementary crystal tracer (i.e. DY where FB was applied first and vice-versa) to a freshly cut nerve end. Table 1 compares tracer application and experimental time frames for groups A and B.
2. Materials and methods
2.2. Tissues preparation and sectioning
2.1. Surgery and tracer application
All solutions used for perfusion, tissue fixation and cryoprotection were buffered with 0.1 M sodium phosphate at pH 7.4. At the experimental endpoint, the rats were deeply anaesthetized with Nembutal (75 ml/kg IP), injected intracardially with approximately 500 units of heparin, then perfused with 250 ml of chilled 0.9% saline followed by 250 ml of 4% paraformaldehyde. During dissection, the sciatic nerves were traced back via the DRGs to the spinal cord. The DRGs were confirmed as being L4, L5 and L6 by counting caudally from the first dorsal roots (C2) of the entire intact spinal cord. From this tissue, the spinal cord between L1 and S2 and the L4-6 DRGs were harvested and post-fixed in 4% paraformaldehyde for 12 h, then cryoprotected in 30% sucrose for 24 h (both at 4 °C). Longitudinal 50 mm sections of lumbar spinal cord tissue were cut serially using a freezing microtome, mounted on 0.1% gelatinized slides, air-dried and then coverslipped with DPX. DRGs placed in TissueTek O.C.T. Compound (Sakura Finetek, USA) were snap frozen in liquid nitrogen–cooled isopentane and 20 mm sections were cut using a cryostat and mounted serially on 0.1% gelatinized slides. O.C.T. was removed from the air-dried sections by rinsing in distilled water before the slides were once again dried and then coverslipped with DPX.
All experiments were performed on adult male Sprague–Dawley rats, approximately 200 g in weight. Animal care and surgical procedures were approved by The University of Melbourne Animal Experimentation Ethics Committee and carried out in accordance with its guidelines. Two groups of animals (A, n =6 and B, n= 6) underwent surgery using pentobarbitone sodium anaesthesia (Nembutal, Boehringer, Artarmon, Australia, 50 ml/kg IP). The procedure used for group A and the first surgery for group B was bilateral sciatic nerve transection at mid-thigh level. Surrounding tissue was shrouded by fine-weave gauze and then tracer crystals (FB or DY, EMS-Chemie GmbH, Groß-Umstadt, Germany) were applied directly to the proximal stump of each nerve using a small spatula (Fritzsch and Northcutt, 1992) and allowed to ‘set’ for approximately 10 min. Any loose crystal not bonded to the proximal stump was removed by lightly wiping it as the gauze was being discarded. Surrounding tissue was then rinsed with sterile 0.9% saline, the distal nerve end reflected (but not sutured) into muscle and the wound was closed in layers with 5/0 silk. As both FB and DY were being applied to separate transected sciatic nerves
Table 1 Tracer application and experimental time frames for groups A and B Group
Procedure
A
FB to right sciatic nerve, DY to left
B
Apply tracer 1 (FB one nerve, DY other)
4-day interval
Perfusion and tissue collection Apply tracer 2 (complement)
4-day interval Perfusion and tissue collection
Group A, n =6; group B, n=6. Both groups underwent bilateral sciatic nerve transection at first surgery. At second surgery, group B animals had the sciatic nerve re-exposed, trimmed and the complementary tracer applied.
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164
161
For group B tissue, the proportions of the fluorescent-labeled motoneuron and DRG neuron subpopulations (i.e. single-labeled FB, single-labeled DY and double-labeled FB–DY) were calculated directly from the sampling data. For group A tissue, the Cavalieri method was used to estimate FB- and DY-labeled neuron numbers in both DRGs and spinal cords.
2.4. Statistical analysis Fig. 1. Motoneurons displaying ‘a’ FB–DY fluorescence (DY nucleus and FB cytoplasm) and ‘b’ DY fluorescence only. The neighbouring neuron above ‘b’ is FB single-labeled, but its characteristically dim, unlabeled nucleus is out of the plane of focus. It is an example of a neuron that may not be counted within a specified focal range using the optical disector method.
2.3. Microscopic examination and stereological analysis Fluorescent-labeled neurons in the prepared spinal cord and DRG sections were identified using an Olympus BX60 fluorescence microscope with a standard UV filter and wavelength of 360 nm. FB-labeled neurons in spinal cord and DRG sections had a characteristic appearance of blue fluorescent cytoplasm usually with an unlabeled nucleus, whereas DY-labeled neurons had a distinctly yellow fluorescent nucleus surrounded by a less intense granular yellow fluorescent cytoplasm. Double-labeled neurons had a yellow fluorescent nucleus and paler blue or blue-green fluorescent cytoplasm (Fig. 2). DY labeling of some glial cells was noted, but these were distinguished from the nuclei of DY- and double-labeled neurons on the basis that they were often oval in shape, and were not surrounded by a fluorescent-labeled cytoplasm. Furthermore, in spinal cord tissue the DY-labeled motoneuron nuclei were far larger than labeled glial cells. Labeled motoneurons in every fourth spinal cord section from animals in group A were plotted under fluorescence microscopy using a digitizer (Minnesota Datametrics Corp., MN, USA) and plotting software (MD Plot 4.0) to enable comparison of FB- and DY-labeled motoneuron distributions in the transverse plane and rostro-caudally. Labeled neuron populations in every fourth spinal cord section and every eighth DRG section for each animal from groups A and B were sampled using the optical disector method (Gundersen et al., 1988), which has recently been validated for analysis of frozen tissue (Messina et al., 2000). To compensate for the problem of uneven distribution of labeled neurons within a spinal cord and DRG tissue, a set sampling path was followed from a random starting point of each section sampled. A neuron was counted only if its nucleus could be clearly identified (Fig. 1).
As the counts of both group A and B were found to be normally distributed, parametric tests were applied to the data. The number of FB- and DY-labeled neurons in group A animals were compared using a Student’s t-test for paired samples. The group B data was consolidated to reflect the labeling efficacy of each tracer (i.e. the total proportion of neurons within a population labeled by either tracer, including both double- and single-labeled neurons, see Section 3, Table 4) and then analysed using a two-way analysis of variance (ANOVA). This test was used to determine firstly whether there was a difference in the labeling efficacy between the two tracers and secondly, whether any difference was associated with the order of tracer application.
3. Results
3.1. Group A— separate FB and DY labeling of motoneurons and DRG neurons Plotted motoneurons from group A animals showed very similar distributions of FB and DY labeling both rostro-caudally and in the transverse plane (Fig. 2). The mean estimated number of motoneurons labeled by FB (27899339) and DY (27989308) were not significantly different (P= 1.00, df 5; Table 2 for individual
Fig. 2. A stack of plotted longitudinal sections from a group A animal superimposed on each other, illustrating the similar distribution of motoneurons labeled with FB (*, on the right) and DY ( + , on the left).
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164
162
Table 2 Group A— the estimated number of FB- and DY-labeled motoneurons and DRG neurons following separate application of the two tracers Animal Number of neurons Moto-neurons DRG-neurons
1
2
3
4
5
6
Mean
SD
FB DY
3201 2575
2475 2939
2341 2761
2986 3317
3041 2432
2741 2761
2798 2798
339 308
FB DY
15 034 14 439
23 702 20 698
24 995 19 794
22 660 22 324
17 027 18 009
22 490 18 668
20 985 18 989
3990 2700
and group mean results). Similarly, there was no significant difference between the mean estimated number of DRG neurons labeled by the two tracers (FB =20 98593990, DY=18 989 9 2700, P = 0.95, df 5; Table 2).
3.2. Group B—serial labeling of motoneuron populations with FB and DY Table 3 shows individual and group mean results for the proportions of the three labeled motoneuron subpopulations (double-labeled FB– DY, single-labeled FB and single-labeled DY). Within this table, the data is further divided into groups depending on the first tracer applied. The group mean labeled motoneuron subpopulation proportions were 79% FB– DY, 10% FB and 11% DY when FB was the first tracer applied, and 79% FB –DY, 11% FB and 10% DY when DY was the initial tracer. Omitting the division based on initial tracer shows labeling rates of 79% FB– DY, 10% FB and 11% DY in labeled motoneuron populations. As the proportion of labeled motoneurons that were not double-labeled was evenly divided between FB- and DY-single labeling, there was little difference between the labeling efficacy of the two tracers (i.e. the overall proportion of neurons labeled by each tracer, %FB+ %FB –DY and %DY +%FB – DY; Table 4). Table 4 shows that when FB was applied first, the respective labeling efficacies were FB=89% and DY= 90% compared with FB=89% and DY=89% when DY was the initial tracer. A two-way ANOVA showed that there was no significant effect of tracer order on labeling efficacy of the two tracers (F=0.012, P =0.914, df 1) and no significant difference between the mean labeling efficacy of FB (8996%) and DY (89 96%) (P =0.935, df 11) in serial labeling of motoneurons.
FB applied first the labeling proportions were 75% FB –DY, 12% FB and 13% DY, compared with 78% FB –DY, 11% FB and 11% DY when DY was the initial tracer. Omitting the division based on initially applied tracer, the group means were 77% FB–DY, 11% FB and 12% DY. Analysis of the consolidated data (Table 4) showed that the comparative efficacy of serially-applied FB and DY in labeling DRG was similar to that in labeling motoneurons. The proportions labeled were FB=87% and DY= 87% when FB was applied first, and FB= 89% and DY=90% when DY was the initially applied tracer. Once again, this is due to the proportion of single-labeled DRG neurons being equally divided between FB and DY (Table 3). A two-way ANOVA showed that there was no significant effect of tracer order on labeling efficacy of the two tracers (F=0.301, P=0.595, df 1) and no significant difference between the mean labeling efficacy of FB (889 8%) and DY (899 6%) in serial labeling of DRG neurons (P= 0.838, df 11).
Table 3 Group B— the proportion of FB–DY-, FB- and DY-labeled neurons in fluorescent-labeled motoneurons and DRG neurons
Motoneurons
DRG neurons
3.3. Group B—serial labeling populations with FB and DY
of
DRG
Proportion of neurons
Mean
SD
FB 1st tracer
%FB–DY %FB %DY
79 10 11
9 6 6
DY 1st tracer
%FB–DY %FB %DY
79 11 10
8 6 5
FB 1st tracer
%FB–DY %FB %DY
75 12 13
10 7 9
DY 1st tracer
%FB–DY %FB %DY
78 11 11
10 6 8
neuron
The group mean labeled DRG neuron subpopulations showed similar results to the motoneuron labeling after serial application of FB and DY (Table 3). With
The group means show no obvious effect of order of tracer application on labeling pattern.
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164 Table 4 Group B— efficacy of FB and DY in labeling motoneurons and DRG neurons Proportion of neurons Motoneurons
DRG neurons
Mean
SD
FB 1st tracer
%FB+FB–DY 89 %DY+FB–DY 90
6 6
DY 1st tracer
%FB+FB–DY 89 %DY+FB–DY 89
5 6
FB 1st tracer
%FB+FB–DY 87 %DY+FB–DY 87
9 7
DY 1st tracer
%FB+FB–DY 89 %DY+FB–DY 90
8 6
4. Discussion When using fluorescent retrograde tracers to illustrate neuron population changes following peripheral nerve injury, several issues should be considered. It is desirable that the tracers being used are able to label a large number of neurons, have comparable uptake characteristics and a relatively rapid uptake time, are stable compounds giving fluorescence longevity, and can be clearly distinguished from one another, enabling the identification of cells labeled with either or both tracers. With such a broad range of tracers currently available, it is not difficult to select two tracers that fulfil many of these criteria. The combination of FB and DY has previously been suggested as a convenient and effective combination because of their distinct fluorescent appearance under the same ultraviolet wavelength (Horikawa and Powell, 1986). Notwithstanding, the same investigators questioned the efficacy of the two tracers. They reported that in the peripheral nervous system FB and DY label far fewer neurons than wheatgerm agglutinin conjugated HRP (WGA– HRP), which is often seen as a ‘gold standard’ in retrograde labeling. Unfortunately, previous investigators using WGA– HRP to retrogradely label neurons projecting to the rat sciatic nerve (Swett et al., 1986) did not utilize unbiased counting methods, which makes comparison with our results difficult. Our previous findings indicate that FB shows similar labeling rates for other peripheral nerves to FluoroGold (FG, in preparation) but outperforms the fluorescent tracer tetramethyl rhodamine dextran (TMRD), in retrogradely labeling motoneurons and DRG neurons of the rat sciatic nerve (Messina et al., 2000). Other work in progress suggests that FB has a labeling efficacy of up to 100% in labeling DRG neurons projecting from the L5 DRG to the sciatic nerve (approximately 70% of sensory axons in the sciatic at mid-thigh level project from the L5 DRG, and FB labels at least that proportion of L5 DRG neurons when applied to the cut nerve).
163
With regard to comparative efficacy of FB and DY, we have found no previously published studies that directly compared the performance of these two tracers when applied separately to the rat sciatic nerve. A previous study did find that DY had a poorer efficacy than other fluorescent retrograde tracers such as True Blue (Haase and Payne, 1990). Our results using unbiased estimates, however, have shown that applying DY tracer crystals to a transected sciatic nerve achieves the same labeling efficiency as FB for both motoneurons and DRG neurons, which is in turn far better than other fluorescent retrograde tracers, such as TMRD. Furthermore, DY presents practical advantages over other tracers with comparable labeling rates to FB, such as FG, DY and FB are complementary tracers where DY labels the nucleus and FB labels the cytoplasm. Both tracers are visible using the UV filter on a fluorescence microscope and double labeling is therefore easily observed and documented. Another significant advantage is that DY displays good fluorescence longevity in cut, refrigerated tissue. Using FB and FG together would present practical difficulties in identifying double-labeling, as both tracers label the cytoplasm, and would therefore be difficult to distinguish. Aside from the comparability of labeling efficiency of FB and DY applied to separate nerves, Fig. 2 also shows that these two tracers label a very similar distribution of rat sciatic motoneurons, which is further evidence of their performance similarities, and the advantage of using this combination in serial fluorescent labeling experiments. On the tissue of tracer interaction following serial application of FB and DY, Puigdellı´vol-Sa´ nchez et al. (2000) reported a ‘blocking’ effect when FB application to the rat sciatic nerve preceded DY either immediately or by up to 2 months. This resulted in a lower rate of double labeling and a higher proportion of single-labeled FB neurons in both motoneuron and DRG neutrons populations. However, utilising a crystal tracer application method and an uptake period between the two tracers of only 4 days we have achieved a doublelabeling rate of 79% in motoneurons and 77% in DRG neurons. Furthermore, there were no consistent differences attributable to the order of tracer application. The results of these serial labeling experiments do not reveal evidence of one tracer predominating over the other, with the labeling efficacy of FB and DY in both motoneurons and DRG neurons approximately equal. A remaining issue from these results is the explanation for equal labeling efficacies of FB and DY in serial labeling experiments. It is possible that the two tracers label slightly different distributions of neuron profile sizes. The analysis of profile size for FB labeled neurons would not be difficult, as it clearly and consistently labels the cytoplasm of both motoneurons and DRG
164
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164
neurons. The ability to analyse accurately the size of DY labeled neurons is questionable, as it is predominantly a nuclear fluorescent label and so judgment of cell borders may therefore be inconsistent. Another possible contributing factor to different labeling patterns between FB and DY following serial applications is whether extraneous pathways have contributed significantly to the labeling of neurons. When applying tracers, there is the potential for tracer uptake and labeling via pathways other than directly from the injured nerve (for example via muscle or skin damaged during surgery). This may affect the distribution of neuron populations labeled by tracers applied on different occasions and thus the consistency of results. For this reason many investigators take time-consuming precautions in order to minimize tracer contamination of surrounding tissues. However, our preliminary work involving blockage of neuronally-mediated labeling following the crystal tracer application technique indicates that labeling via extraneous pathways is minimal. Regardless of the reason for the differences in labeling patterns following serial tracer application, our results have shown that the average proportion of single labeled neurons (approximately 20– 23%) remains evenly divided between FB and DY in spinal cord and DRG tissue. This causes no discrepancy between the labeling efficacy of the two tracers in control animals. It therefore stands to reason that in experimental animals with an interposed injury time between FB and DY applications, a difference between the group mean efficacy of these two fluorescent tracers would indicate an actual difference in pre- and post-injury neuron numbers, and not be simply a result of differing tracer performance. Our findings support the combined use of FB and DY to label pre- and post-injury neuron populations in the rat sciatic nerve transection model. Our methods provide a useful tool for demonstrating and analysing the changes that take place following a peripheral nerve injury. Using a reliable, accurate counting method and a crystal tracer application technique we have established that a large number of motor and sensory neurons projecting to the rat sciatic nerve can be labeled by FB and DY.
There was no statistically significant difference between the number and distribution of neurons labeled by the two tracers when applied separately. More importantly, the finding that FB and DY label a very similar proportion of motoneurons and DRG neurons following serial application also supports their use in this way, even though 20–23% of labeled motoneuron or DRG neuron populations will not be double labeled in uninjured animals.
References Coggeshall RE, Lekan HA. Methods for determining numbers of cells and synapses: a case for more uniform standards of review. J Comp Neurol 1996;364:6 – 15. Gundersen HJG, Bagger P, Bendtsen TF, Evans SM, Korbo L, Marcussen N, Møller A, Nielsen K, Nyengaard JR, Pakkenberg B, Sørensen FB, Vesterby A, West MJ. The new stereological tools: disector, fractionator, nulceator and point sampled intercepts and their use in pathological research and diagnosis. APMIS 1988;96:857 – 81. Fritzsch B, Northcutt RG. A plastic embedding technique for analyzing fluorescent dextran-amine labeled neuronal profiles. Biotech Histochem 1992;67:153 – 7. Haase P, Payne JN. Comparison of the efficiencies of true blue and diamidino yellow as retrograde tracers in the peripheral motor system. J Neurosci Methods 1990;35:175 – 83. Horikawa K, Powell EW. Comparison of techniques for retrograde labeling using the rat’s facial nucleus. J Neurosci Methods 1986;17:287 – 96. Messina A, Sangster CLC, Morrison WA, Galea MP. Requirements for obtaining unbiased estimates of neuronal number in frozen sections. J Neurosci Methods 2000;97:133 – 7. Puigdellı´vol-Sa´ nchez AP, Prats-Galino A, Ruano-Gil D, Molander C. Fast Blue and Diamidino Yellow as retrograde tracers in peripheral nerves: efficacy of combined nerve injection and capsule application to transected nerves in the adult rat. J Neurosci Methods 2000;95:103 – 10. Richmond FJR, Gladdy R, Creasy JL, Kitamura S, Smits E, Thomson DB. Efficacy of seven retrograde tracers, compared in multiple-labelling studies of feline motoneurones. J Neurosci Methods 1994;53:35 – 46. Sangster CL, Galea MP, Fan R, Morrison WA, Messina A. A method for processing fluorescent labeled tissue into methacrylate: a qualitative comparison of four tracers. J Neurosci Methods 1999;89:159 – 65. Swett JE, Wilkholm RP, Blanks RHI, Swett AL, Conley LC. Motoneurons of the rat sciatic nerve. Exp Neurol 1986;93:227 –52.
Comparing the efficacy of two fluorescent retrograde tracers in labeling the motor and sensory neuron populations of the rat sciatic nerve C.T. Byers a, R. Fan b, A. Messina b, W.A. Morrison c, M.P. Galea a,* a b
School of Physiotherapy, The Uni6ersity of Melbourne, 200 Berkeley Street, Park6ille, Victoria 3052, Australia Bernard O’Brien Institute of Microsurgery, St Vincent’s Hospital Melbourne, Fitzroy, Victoria 3065, Australia c Department of Surgery, The Uni6ersity of Melbourne, Victoria 3010, Australia Received 12 April 2001; received in revised form 30 November 2001; accepted 30 November 2001
Abstract We compared the efficacy with which the fluorescent tracers Fast Blue (FB) and Diamidino Yellow (DY) retrogradely label neutrons. Trace crystals were applied to the sciatic nerve exclusively (single label) or serially (double label). Unbiased cell counts showed that FB and DY label similar numbers of motoneurons (P = 1.00, df 5) or DRG neurons (P= 0.95, df 5) when applied exclusively. Plotting of motoneurons revealed a similar pattern of distribution of FB and DY labeled neurons. When the tracers were applied serially, 79% of labeled motoneurons and 77% of labeled DRG neurons were double-labeled irrespective of which tracer was applied first. Equal proportions of the remaining labeled neurons were single-labeled with FB or DY. These data show that FB and DY label equal numbers of motor and sensory neurons of the sciatic nerve following exclusive or serial application of tracers. These findings support the use of FB and DY together in serial fluorescent labeling experiments. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Fast Blue (FB); Fluorescent tracers; Diamidino Yellow (DY); Motoneurons; Dorsal root ganglion (DRG)
1. Introduction The application of a series of different fluorescent retrograde tracers to peripheral nerves allows for the comparison of pre- and post-injury neuron populations that project to the lesion site (Richmond et al., 1994; Puigdellı´vol-Sa´nchez et al., 2000). However, when two different tracers are applied serially for this purpose, the value of qualitative and quantitative comparisons depends on the similarity of uptake characteristics of the respective tracers. If valid quantitative comparisons are to be made, two principal issues should be considered. Firstly, it is important to establish that the two tracers being used label similar numbers of neurons and with a similar distribution in the tissue being studied. Although the * Corresponding author. Tel.: + 61-3-8344-4118; fax: +61-3-83444188. E-mail address: [email protected] (M.P. Galea).
efficacy of different fluorescent retrograde tracers labeling neurons projecting to the peripheral nervous systems has been previously investigated (Horikawa and Powell, 1986; Haase and Payne, 1990; Richmond et al., 1994; Puigdellı´vol-Sa´nchez et al., 2000), few studies have utilized modern, unbiased stereological techniques, such as the optical disector (Gundersen et al., 1988), to estimate labeled neuron numbers. Thus, variables such as cell size and shape, the thickness of histological sections and easily-violated assumptions bring the results of studies using profile and assumption-based counting methods into question (Coggeshall and Lekan, 1996). Secondly, if tracers are to be applied serially, it is also important to establish whether the presence of the initial tracer affects labeling by the subsequently applied tracer. A recent study reported a ‘blocking’ effect of the tracer Fast Blue (FB) over Diamidino Yellow (DY) when they were applied serially as aqueous solutions to the proximal stump of a transected rat sciatic
0165-0270/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 5 - 0 2 7 0 ( 0 1 ) 0 0 5 2 0 - 9
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164
160
nerve (Puigdellı´vol-Sa´ nchez et al., 2000). However, the possibility of a ‘blocking’ effect of FB over DY following serial applications warrants further investigation, to determine if such an interaction occurs when using our previously published method of applying tracer crystals to the transected rat sciatic nerve (Sangster et al., 1999). This study was undertaken with the purpose of establishing whether: 1. FB and DY applied separately labeled a similar number of motoneurons and DRG neurons and a similar distribution of motoneurons, and; 2. Pre-labeling with one tracer affected the subsequent labeling by a second tracer applied at a later time.
within one episode of surgery, utmost care was taken to prevent tracer cross-contamination. In particular, only one sciatic nerve was left exposed at a time and for each operative side a separate set of surgical and tracer application instruments were used. Animals in group B underwent a second episode of surgery 4 days after the first. For each initial surgery site the incision was reopened and the sciatic nerve was trimmed to remove the distal 5 mm of the proximal stump. This enabled application of the complementary crystal tracer (i.e. DY where FB was applied first and vice-versa) to a freshly cut nerve end. Table 1 compares tracer application and experimental time frames for groups A and B.
2. Materials and methods
2.2. Tissues preparation and sectioning
2.1. Surgery and tracer application
All solutions used for perfusion, tissue fixation and cryoprotection were buffered with 0.1 M sodium phosphate at pH 7.4. At the experimental endpoint, the rats were deeply anaesthetized with Nembutal (75 ml/kg IP), injected intracardially with approximately 500 units of heparin, then perfused with 250 ml of chilled 0.9% saline followed by 250 ml of 4% paraformaldehyde. During dissection, the sciatic nerves were traced back via the DRGs to the spinal cord. The DRGs were confirmed as being L4, L5 and L6 by counting caudally from the first dorsal roots (C2) of the entire intact spinal cord. From this tissue, the spinal cord between L1 and S2 and the L4-6 DRGs were harvested and post-fixed in 4% paraformaldehyde for 12 h, then cryoprotected in 30% sucrose for 24 h (both at 4 °C). Longitudinal 50 mm sections of lumbar spinal cord tissue were cut serially using a freezing microtome, mounted on 0.1% gelatinized slides, air-dried and then coverslipped with DPX. DRGs placed in TissueTek O.C.T. Compound (Sakura Finetek, USA) were snap frozen in liquid nitrogen–cooled isopentane and 20 mm sections were cut using a cryostat and mounted serially on 0.1% gelatinized slides. O.C.T. was removed from the air-dried sections by rinsing in distilled water before the slides were once again dried and then coverslipped with DPX.
All experiments were performed on adult male Sprague–Dawley rats, approximately 200 g in weight. Animal care and surgical procedures were approved by The University of Melbourne Animal Experimentation Ethics Committee and carried out in accordance with its guidelines. Two groups of animals (A, n =6 and B, n= 6) underwent surgery using pentobarbitone sodium anaesthesia (Nembutal, Boehringer, Artarmon, Australia, 50 ml/kg IP). The procedure used for group A and the first surgery for group B was bilateral sciatic nerve transection at mid-thigh level. Surrounding tissue was shrouded by fine-weave gauze and then tracer crystals (FB or DY, EMS-Chemie GmbH, Groß-Umstadt, Germany) were applied directly to the proximal stump of each nerve using a small spatula (Fritzsch and Northcutt, 1992) and allowed to ‘set’ for approximately 10 min. Any loose crystal not bonded to the proximal stump was removed by lightly wiping it as the gauze was being discarded. Surrounding tissue was then rinsed with sterile 0.9% saline, the distal nerve end reflected (but not sutured) into muscle and the wound was closed in layers with 5/0 silk. As both FB and DY were being applied to separate transected sciatic nerves
Table 1 Tracer application and experimental time frames for groups A and B Group
Procedure
A
FB to right sciatic nerve, DY to left
B
Apply tracer 1 (FB one nerve, DY other)
4-day interval
Perfusion and tissue collection Apply tracer 2 (complement)
4-day interval Perfusion and tissue collection
Group A, n =6; group B, n=6. Both groups underwent bilateral sciatic nerve transection at first surgery. At second surgery, group B animals had the sciatic nerve re-exposed, trimmed and the complementary tracer applied.
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164
161
For group B tissue, the proportions of the fluorescent-labeled motoneuron and DRG neuron subpopulations (i.e. single-labeled FB, single-labeled DY and double-labeled FB–DY) were calculated directly from the sampling data. For group A tissue, the Cavalieri method was used to estimate FB- and DY-labeled neuron numbers in both DRGs and spinal cords.
2.4. Statistical analysis Fig. 1. Motoneurons displaying ‘a’ FB–DY fluorescence (DY nucleus and FB cytoplasm) and ‘b’ DY fluorescence only. The neighbouring neuron above ‘b’ is FB single-labeled, but its characteristically dim, unlabeled nucleus is out of the plane of focus. It is an example of a neuron that may not be counted within a specified focal range using the optical disector method.
2.3. Microscopic examination and stereological analysis Fluorescent-labeled neurons in the prepared spinal cord and DRG sections were identified using an Olympus BX60 fluorescence microscope with a standard UV filter and wavelength of 360 nm. FB-labeled neurons in spinal cord and DRG sections had a characteristic appearance of blue fluorescent cytoplasm usually with an unlabeled nucleus, whereas DY-labeled neurons had a distinctly yellow fluorescent nucleus surrounded by a less intense granular yellow fluorescent cytoplasm. Double-labeled neurons had a yellow fluorescent nucleus and paler blue or blue-green fluorescent cytoplasm (Fig. 2). DY labeling of some glial cells was noted, but these were distinguished from the nuclei of DY- and double-labeled neurons on the basis that they were often oval in shape, and were not surrounded by a fluorescent-labeled cytoplasm. Furthermore, in spinal cord tissue the DY-labeled motoneuron nuclei were far larger than labeled glial cells. Labeled motoneurons in every fourth spinal cord section from animals in group A were plotted under fluorescence microscopy using a digitizer (Minnesota Datametrics Corp., MN, USA) and plotting software (MD Plot 4.0) to enable comparison of FB- and DY-labeled motoneuron distributions in the transverse plane and rostro-caudally. Labeled neuron populations in every fourth spinal cord section and every eighth DRG section for each animal from groups A and B were sampled using the optical disector method (Gundersen et al., 1988), which has recently been validated for analysis of frozen tissue (Messina et al., 2000). To compensate for the problem of uneven distribution of labeled neurons within a spinal cord and DRG tissue, a set sampling path was followed from a random starting point of each section sampled. A neuron was counted only if its nucleus could be clearly identified (Fig. 1).
As the counts of both group A and B were found to be normally distributed, parametric tests were applied to the data. The number of FB- and DY-labeled neurons in group A animals were compared using a Student’s t-test for paired samples. The group B data was consolidated to reflect the labeling efficacy of each tracer (i.e. the total proportion of neurons within a population labeled by either tracer, including both double- and single-labeled neurons, see Section 3, Table 4) and then analysed using a two-way analysis of variance (ANOVA). This test was used to determine firstly whether there was a difference in the labeling efficacy between the two tracers and secondly, whether any difference was associated with the order of tracer application.
3. Results
3.1. Group A— separate FB and DY labeling of motoneurons and DRG neurons Plotted motoneurons from group A animals showed very similar distributions of FB and DY labeling both rostro-caudally and in the transverse plane (Fig. 2). The mean estimated number of motoneurons labeled by FB (27899339) and DY (27989308) were not significantly different (P= 1.00, df 5; Table 2 for individual
Fig. 2. A stack of plotted longitudinal sections from a group A animal superimposed on each other, illustrating the similar distribution of motoneurons labeled with FB (*, on the right) and DY ( + , on the left).
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164
162
Table 2 Group A— the estimated number of FB- and DY-labeled motoneurons and DRG neurons following separate application of the two tracers Animal Number of neurons Moto-neurons DRG-neurons
1
2
3
4
5
6
Mean
SD
FB DY
3201 2575
2475 2939
2341 2761
2986 3317
3041 2432
2741 2761
2798 2798
339 308
FB DY
15 034 14 439
23 702 20 698
24 995 19 794
22 660 22 324
17 027 18 009
22 490 18 668
20 985 18 989
3990 2700
and group mean results). Similarly, there was no significant difference between the mean estimated number of DRG neurons labeled by the two tracers (FB =20 98593990, DY=18 989 9 2700, P = 0.95, df 5; Table 2).
3.2. Group B—serial labeling of motoneuron populations with FB and DY Table 3 shows individual and group mean results for the proportions of the three labeled motoneuron subpopulations (double-labeled FB– DY, single-labeled FB and single-labeled DY). Within this table, the data is further divided into groups depending on the first tracer applied. The group mean labeled motoneuron subpopulation proportions were 79% FB– DY, 10% FB and 11% DY when FB was the first tracer applied, and 79% FB –DY, 11% FB and 10% DY when DY was the initial tracer. Omitting the division based on initial tracer shows labeling rates of 79% FB– DY, 10% FB and 11% DY in labeled motoneuron populations. As the proportion of labeled motoneurons that were not double-labeled was evenly divided between FB- and DY-single labeling, there was little difference between the labeling efficacy of the two tracers (i.e. the overall proportion of neurons labeled by each tracer, %FB+ %FB –DY and %DY +%FB – DY; Table 4). Table 4 shows that when FB was applied first, the respective labeling efficacies were FB=89% and DY= 90% compared with FB=89% and DY=89% when DY was the initial tracer. A two-way ANOVA showed that there was no significant effect of tracer order on labeling efficacy of the two tracers (F=0.012, P =0.914, df 1) and no significant difference between the mean labeling efficacy of FB (8996%) and DY (89 96%) (P =0.935, df 11) in serial labeling of motoneurons.
FB applied first the labeling proportions were 75% FB –DY, 12% FB and 13% DY, compared with 78% FB –DY, 11% FB and 11% DY when DY was the initial tracer. Omitting the division based on initially applied tracer, the group means were 77% FB–DY, 11% FB and 12% DY. Analysis of the consolidated data (Table 4) showed that the comparative efficacy of serially-applied FB and DY in labeling DRG was similar to that in labeling motoneurons. The proportions labeled were FB=87% and DY= 87% when FB was applied first, and FB= 89% and DY=90% when DY was the initially applied tracer. Once again, this is due to the proportion of single-labeled DRG neurons being equally divided between FB and DY (Table 3). A two-way ANOVA showed that there was no significant effect of tracer order on labeling efficacy of the two tracers (F=0.301, P=0.595, df 1) and no significant difference between the mean labeling efficacy of FB (889 8%) and DY (899 6%) in serial labeling of DRG neurons (P= 0.838, df 11).
Table 3 Group B— the proportion of FB–DY-, FB- and DY-labeled neurons in fluorescent-labeled motoneurons and DRG neurons
Motoneurons
DRG neurons
3.3. Group B—serial labeling populations with FB and DY
of
DRG
Proportion of neurons
Mean
SD
FB 1st tracer
%FB–DY %FB %DY
79 10 11
9 6 6
DY 1st tracer
%FB–DY %FB %DY
79 11 10
8 6 5
FB 1st tracer
%FB–DY %FB %DY
75 12 13
10 7 9
DY 1st tracer
%FB–DY %FB %DY
78 11 11
10 6 8
neuron
The group mean labeled DRG neuron subpopulations showed similar results to the motoneuron labeling after serial application of FB and DY (Table 3). With
The group means show no obvious effect of order of tracer application on labeling pattern.
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164 Table 4 Group B— efficacy of FB and DY in labeling motoneurons and DRG neurons Proportion of neurons Motoneurons
DRG neurons
Mean
SD
FB 1st tracer
%FB+FB–DY 89 %DY+FB–DY 90
6 6
DY 1st tracer
%FB+FB–DY 89 %DY+FB–DY 89
5 6
FB 1st tracer
%FB+FB–DY 87 %DY+FB–DY 87
9 7
DY 1st tracer
%FB+FB–DY 89 %DY+FB–DY 90
8 6
4. Discussion When using fluorescent retrograde tracers to illustrate neuron population changes following peripheral nerve injury, several issues should be considered. It is desirable that the tracers being used are able to label a large number of neurons, have comparable uptake characteristics and a relatively rapid uptake time, are stable compounds giving fluorescence longevity, and can be clearly distinguished from one another, enabling the identification of cells labeled with either or both tracers. With such a broad range of tracers currently available, it is not difficult to select two tracers that fulfil many of these criteria. The combination of FB and DY has previously been suggested as a convenient and effective combination because of their distinct fluorescent appearance under the same ultraviolet wavelength (Horikawa and Powell, 1986). Notwithstanding, the same investigators questioned the efficacy of the two tracers. They reported that in the peripheral nervous system FB and DY label far fewer neurons than wheatgerm agglutinin conjugated HRP (WGA– HRP), which is often seen as a ‘gold standard’ in retrograde labeling. Unfortunately, previous investigators using WGA– HRP to retrogradely label neurons projecting to the rat sciatic nerve (Swett et al., 1986) did not utilize unbiased counting methods, which makes comparison with our results difficult. Our previous findings indicate that FB shows similar labeling rates for other peripheral nerves to FluoroGold (FG, in preparation) but outperforms the fluorescent tracer tetramethyl rhodamine dextran (TMRD), in retrogradely labeling motoneurons and DRG neurons of the rat sciatic nerve (Messina et al., 2000). Other work in progress suggests that FB has a labeling efficacy of up to 100% in labeling DRG neurons projecting from the L5 DRG to the sciatic nerve (approximately 70% of sensory axons in the sciatic at mid-thigh level project from the L5 DRG, and FB labels at least that proportion of L5 DRG neurons when applied to the cut nerve).
163
With regard to comparative efficacy of FB and DY, we have found no previously published studies that directly compared the performance of these two tracers when applied separately to the rat sciatic nerve. A previous study did find that DY had a poorer efficacy than other fluorescent retrograde tracers such as True Blue (Haase and Payne, 1990). Our results using unbiased estimates, however, have shown that applying DY tracer crystals to a transected sciatic nerve achieves the same labeling efficiency as FB for both motoneurons and DRG neurons, which is in turn far better than other fluorescent retrograde tracers, such as TMRD. Furthermore, DY presents practical advantages over other tracers with comparable labeling rates to FB, such as FG, DY and FB are complementary tracers where DY labels the nucleus and FB labels the cytoplasm. Both tracers are visible using the UV filter on a fluorescence microscope and double labeling is therefore easily observed and documented. Another significant advantage is that DY displays good fluorescence longevity in cut, refrigerated tissue. Using FB and FG together would present practical difficulties in identifying double-labeling, as both tracers label the cytoplasm, and would therefore be difficult to distinguish. Aside from the comparability of labeling efficiency of FB and DY applied to separate nerves, Fig. 2 also shows that these two tracers label a very similar distribution of rat sciatic motoneurons, which is further evidence of their performance similarities, and the advantage of using this combination in serial fluorescent labeling experiments. On the tissue of tracer interaction following serial application of FB and DY, Puigdellı´vol-Sa´ nchez et al. (2000) reported a ‘blocking’ effect when FB application to the rat sciatic nerve preceded DY either immediately or by up to 2 months. This resulted in a lower rate of double labeling and a higher proportion of single-labeled FB neurons in both motoneuron and DRG neutrons populations. However, utilising a crystal tracer application method and an uptake period between the two tracers of only 4 days we have achieved a doublelabeling rate of 79% in motoneurons and 77% in DRG neurons. Furthermore, there were no consistent differences attributable to the order of tracer application. The results of these serial labeling experiments do not reveal evidence of one tracer predominating over the other, with the labeling efficacy of FB and DY in both motoneurons and DRG neurons approximately equal. A remaining issue from these results is the explanation for equal labeling efficacies of FB and DY in serial labeling experiments. It is possible that the two tracers label slightly different distributions of neuron profile sizes. The analysis of profile size for FB labeled neurons would not be difficult, as it clearly and consistently labels the cytoplasm of both motoneurons and DRG
164
C.T. Byers et al. / Journal of Neuroscience Methods 114 (2002) 159–164
neurons. The ability to analyse accurately the size of DY labeled neurons is questionable, as it is predominantly a nuclear fluorescent label and so judgment of cell borders may therefore be inconsistent. Another possible contributing factor to different labeling patterns between FB and DY following serial applications is whether extraneous pathways have contributed significantly to the labeling of neurons. When applying tracers, there is the potential for tracer uptake and labeling via pathways other than directly from the injured nerve (for example via muscle or skin damaged during surgery). This may affect the distribution of neuron populations labeled by tracers applied on different occasions and thus the consistency of results. For this reason many investigators take time-consuming precautions in order to minimize tracer contamination of surrounding tissues. However, our preliminary work involving blockage of neuronally-mediated labeling following the crystal tracer application technique indicates that labeling via extraneous pathways is minimal. Regardless of the reason for the differences in labeling patterns following serial tracer application, our results have shown that the average proportion of single labeled neurons (approximately 20– 23%) remains evenly divided between FB and DY in spinal cord and DRG tissue. This causes no discrepancy between the labeling efficacy of the two tracers in control animals. It therefore stands to reason that in experimental animals with an interposed injury time between FB and DY applications, a difference between the group mean efficacy of these two fluorescent tracers would indicate an actual difference in pre- and post-injury neuron numbers, and not be simply a result of differing tracer performance. Our findings support the combined use of FB and DY to label pre- and post-injury neuron populations in the rat sciatic nerve transection model. Our methods provide a useful tool for demonstrating and analysing the changes that take place following a peripheral nerve injury. Using a reliable, accurate counting method and a crystal tracer application technique we have established that a large number of motor and sensory neurons projecting to the rat sciatic nerve can be labeled by FB and DY.
There was no statistically significant difference between the number and distribution of neurons labeled by the two tracers when applied separately. More importantly, the finding that FB and DY label a very similar proportion of motoneurons and DRG neurons following serial application also supports their use in this way, even though 20–23% of labeled motoneuron or DRG neuron populations will not be double labeled in uninjured animals.
References Coggeshall RE, Lekan HA. Methods for determining numbers of cells and synapses: a case for more uniform standards of review. J Comp Neurol 1996;364:6 – 15. Gundersen HJG, Bagger P, Bendtsen TF, Evans SM, Korbo L, Marcussen N, Møller A, Nielsen K, Nyengaard JR, Pakkenberg B, Sørensen FB, Vesterby A, West MJ. The new stereological tools: disector, fractionator, nulceator and point sampled intercepts and their use in pathological research and diagnosis. APMIS 1988;96:857 – 81. Fritzsch B, Northcutt RG. A plastic embedding technique for analyzing fluorescent dextran-amine labeled neuronal profiles. Biotech Histochem 1992;67:153 – 7. Haase P, Payne JN. Comparison of the efficiencies of true blue and diamidino yellow as retrograde tracers in the peripheral motor system. J Neurosci Methods 1990;35:175 – 83. Horikawa K, Powell EW. Comparison of techniques for retrograde labeling using the rat’s facial nucleus. J Neurosci Methods 1986;17:287 – 96. Messina A, Sangster CLC, Morrison WA, Galea MP. Requirements for obtaining unbiased estimates of neuronal number in frozen sections. J Neurosci Methods 2000;97:133 – 7. Puigdellı´vol-Sa´ nchez AP, Prats-Galino A, Ruano-Gil D, Molander C. Fast Blue and Diamidino Yellow as retrograde tracers in peripheral nerves: efficacy of combined nerve injection and capsule application to transected nerves in the adult rat. J Neurosci Methods 2000;95:103 – 10. Richmond FJR, Gladdy R, Creasy JL, Kitamura S, Smits E, Thomson DB. Efficacy of seven retrograde tracers, compared in multiple-labelling studies of feline motoneurones. J Neurosci Methods 1994;53:35 – 46. Sangster CL, Galea MP, Fan R, Morrison WA, Messina A. A method for processing fluorescent labeled tissue into methacrylate: a qualitative comparison of four tracers. J Neurosci Methods 1999;89:159 – 65. Swett JE, Wilkholm RP, Blanks RHI, Swett AL, Conley LC. Motoneurons of the rat sciatic nerve. Exp Neurol 1986;93:227 –52.