No evidence for new dopaminergic neurons in the adult mammalian substantia nigra

Abstract

A recent report by Zhao et al. [Zhao, M., Momma, S., Delfani, K., Carlen, M., Cassidy, R. M., Johansson, C. B., Brismar, H., Shupliakov, O., Frisen, J. & Janson, A. M. (2003) Proc. Natl. Acad. Sci. USA 100, 7925–7930] suggests that dopaminergic neurons, the cell type lost in Parkinson's disease, are continuously generated in the adult substantia nigra pars compacta. Using similar methodological procedures to label dividing cells, we found no evidence of new dopaminergic neurons in the substantia nigra, either in normal or 6-hydroxydopamine-lesioned hemi-Parkinsonian rodents, or even after growth factor treatment. Furthermore, we found no evidence of neural stem cells emanating from the cerebroventricular system and migrating to the substantia nigra. We conclude that it is unlikely that dopaminergic neurons are generated in the adult mammalian substantia nigra.

Neurogenesis in the adult mammalian brain is generally accepted to occur in two discrete regions: the hippocampus and olfactory bulb (1, 2). In recent years, there have been reports that neurogenesis may also occur in other regions of the adult brain under normal conditions, such as the neocortex (3) and amygdala (4). Moreover, various brain insults have been shown to induce the production of new neurons in the striatum (5–8) and cortex (9, 10). However, others have been unable to replicate some of these reports (11, 12), and some have been challenged on methodological grounds. In some cases, it has been argued that cells considered as newborn because of labeling by nucleotide analogues are not newborn but actually undergoing DNA repair (13, 14). In other cases, critique has been raised claiming that the criteria for identifying cells double-labeled with markers of cell division and neuronal proteins have not been applied strictly enough (15). Recently, Zhao et al. (16) suggested that new dopaminergic neurons are continuously generated in the adult mouse substantia nigra pars compacta (SN). They proposed that there are sufficient dopaminergic neurons generated to completely replace the SN during the adult lifetime of a mouse. They also reported that lesions of the mouse SN, made by using the toxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), result in a dramatic increase in the generation of new dopaminergic neurons. If these findings can be extended to humans, they are potentially extremely important clinically, because induced enhancement of neurogenesis in the SN could be an approach to replace dopaminergic neurons in Parkinson's disease (15). However, using different techniques or experimental paradigms, other studies have failed to detect neurogenesis in the SN (17, 18). Various factors could explain the discrepancies between reports, such as differences regarding the rodent species used, the manner in which the proliferative marker was administered, or the time delay after mitotic labeling. To resolve this issue, we set out to replicate parts of the Zhao et al. study (16), following as accurately as possible their methodological procedures, but using a higher number of animals per group and also rats rather than only mice.

First, we tried to repeat the part of the experiment in which the highest number of cells colabeled with BrdUrd, the marker for proliferating cells, and the dopamine-synthesizing enzyme tyrosine hydroxylase (TH) had been found in nonlesioned mice. It was also reported (16) that mice prelesioned with MPTP, which selectively destroys dopaminergic neurons in the SN, exhibit nearly double the number of BrdUrd/TH-colabeled cells per animal. Therefore, we investigated whether another model of Parkinson's disease, in which the rat SN was destroyed by 6-hydroxydopamine (6-OHDA) (19), would display neurogenesis in the SN. To potentially further improve conditions for neurogenesis, some lesioned rats were infused with brain-derived neurotrophic factor (BDNF), which has been shown to enhance neurogenesis in other regions of the adult rat brain (20).

Zhao et al. (16) suggest that the newly born cells in SN divide in the subependymal layer of the lateral ventricles and migrate along the ventral midline to the SN, where they differentiate into dopaminergic cells, and that this phenomenon can be traced by using an intracerebroventricular (i.c.v.) injection of the fluorescent tracer 1,1′-dioctadecyl-6,6′-di-(4-sulfophenyl)-3,3,3′,3′-tetramethylindocarbocyanine (DiI). DiI is highly lipid soluble and is passively transported within the plasma membrane of cells it has come in contact with. Furthermore, it has been used to retrogradely label dopamine neurons in the SN by intrastriatal injections of the fluorescent tracer (21). To investigate whether already existing nigral neurons might be retrogradely labeled via the nigrostriatal pathway also after i.c.v. injections, we compared results when injecting DiI into intact animals and into rats in which the nigrostriatal pathway had been destroyed by prior unilateral injection of 6-OHDA. We found no newly born dopamine neurons, as detected by BrdUrd incorporation, in either normal or lesioned animals. Furthermore, we argue that results obtained by using DiI to label newly born neurons in the SN have to be interpreted with extreme caution because DiI may retrogradely label already existing nigral neurons.

Materials and Methods

BrdUrd Labeling. Animal handling and surgical procedures were carried out according to the ethical regulations and guidelines set by the Malmö–Lund Ethical Committee. Singly housed adult male C57BL/6 mice (9–10 weeks of age) and Sprague–Dawley rats (9–10 weeks of age and housed two to three rats per cage) were implanted with Alzet 2004 osmotic minipumps delivering 0.25 μl/h BrdUrd (Sigma) for 21 days into the right lateral ventricle (stereotaxic coordinates for mice: lateral 0.8 mm to midline at bregma and -2.0 mm below dura; and for rats: 0.3 mm posterior to bregma, 1.5 mm lateral to midline, and 4.5 mm below the dura) (16). Zhao et al. (16) used a BrdUrd concentration of 150 mg/ml in their minipumps. However, it was not possible to dissolve BrdUrd at such a high concentration while maintaining a physiological pH (manufacturers recommend a maximal concentration of 10–20 mg/ml in solution without addition of a base). By heating (60°C) and shaking for >1 h, we were able to dissolve BrdUrd in solution up to a maximal concentration of 75 mg/ml at physiological pH (in PBS, pH 7.8) and therefore used this concentration. Mice and rats were perfused (see below) 21 days after pump implantation.

6-OHDA Lesions. Adult female Sprague–Dawley rats (B & K Universal, Sollentuna, Sweden), weighing 220–250 g and housed in pairs, had their SN lesioned by injecting 6-OHDA (Sigma) into the right ascending mesostriatal forebrain bundle (see ref. 22 for procedural details). Three weeks after 6-OHDA injections, the completeness of the lesion was assessed with the amphetamine-induced rotation test by injecting d-methylamphetamine (2.5 mg i.p.) and placing the rat in an automated rotometer bowl for 90 min (22). Rats exhibiting an average of >4.7 net ipsilateral full turns per min were selected. Three weeks after lesioning, rats had Alzet model 2002 osmotic minipumps (200 μl; flow rate: 0.5 μl/h, 12 μl/day) implanted into the right lateral ventricle (see above) to deliver either BrdUrd (1 mg/ml in PBS, 0.5 μl/h) alone or BrdUrd with BDNF (0.5 ng/h; R & D Systems) (20) over 10 days. Rats were then perfused (see below) 5 weeks after pump implantation.

DiI Labeling. By using the same protocol as above, four female and four male Sprague–Dawley rats were 6-OHDA-lesioned in the right median forebrain bundle and tested for amphetamine rotation (22). At 3 weeks after lesion, rats either received a unilateral injection in the left lateral ventricle (coordinates: anterioposterior, -0.3 mm; lateral, +1.5 mm; and ventral, -4.5 mm relative to bregma) of 10 μl (n = 2), bilateral striatal injections (coordinates: anterioposterior, +1 mm; lateral, ±3 mm; and ventral, -4.5 mm) of 5 μl per side (n = 4) or a unilateral injection in the left striatum of 10 μl (n = 2) of 0.2% wt/vol DiI (Molecular Probes) in dimethyl sulfoxide. As controls for the first group, four intact rats received unilateral i.c.v. injections of the same preparation and dose of DiI as above. All rats weighed 300–400 g at the time of DiI injections. They were perfused (see below) at 3 weeks after the injections, the time point at which Zhao et al. (16) first reliably observed DiI and TH colabeling in the SN.

Immunohistochemistry. The animals were killed and transcardially perfused with ice-cold saline followed by 4% paraformaldehyde (PFA). The brains were postfixed in 4% PFA overnight and transferred to 20% sucrose in phosphate buffer. After sinking, they were sectioned at 30 μm by using a sliding microtome. Every brain section containing the SN from the animals that had received BrdUrd at 75 mg/ml, and every 10th section from the 6-OHDA/BrdUrd-treated rats, was immunohistochemically processed for BrdUrd/TH immunohistochemistry. By using free-floating sections, DNA was denatured by incubation in 1 M HCl at 67°C for 30 min and then preincubated in the relevant sera before being incubated over 2 nights at 4°C in rat anti-BrdUrd (1:100 monoclonal; ImmunologicalsDirect, Oxfordshire, U.K.) and mouse anti-TH (1:2,000 monoclonal; Chemicon). The sections were incubated in secondary antibodies, Cy3-conjugated donkey anti-rat antibody (1:200; Jackson ImmunoResearch) and biotinylated horse anti-mouse antibody (1:200; Vector Laboratories) for 2 h, followed by incubation of Alexa 488-conjugated streptavidin (1:200; Molecular Probes) for 2 h. For triple-labeling stainings, we used BrdUrd (same protocol as above), mouse anti-NeuN (1:100; Chemicon), and rabbit anti-TH (1:500; Pel-Freez), with the secondary antibodies listed above, plus Cy5-conjugated goat anti-rabbit (1:200; Jackson ImmunoResearch). Doublecortin antibody (1:400 monoclonal; Santa Cruz Biotechnology catalog no. 8066) was applied with a similar staining procedure to that listed above, by using the biotinylated secondary antibodies. DiI-treated animals were transcardially perfused at 21 days after DiI injection, the brains were postfixed and sectioned, and every fifth section was then processed for TH immunohistochemistry as described above.

Cell Quantification. Total numbers of BrdUrd-immunopositive cells and TH/DiI-colabeled cells were estimated by using unbiased computer-assisted stereological counting methods (cast-grid, Olympus) (23). All cells that appeared to be BrdUrd/TH-colabeled when examined in a conventional fluorescence microscope were subsequently subjected to three-dimensional analysis in a confocal laser-scanning microscope (Leica), by using orthogonal reconstruction of sections scanned at 1 μm thickness. DiI colocalization with TH was assessed by using confocal laser-scanning on representative cells from all of the DiI-injected animals.

Results

No Evidence for Ongoing Turnover of Newly Generated Dopamine Neurons in the SN. Consistent with other reports (16–18), we observed BrdUrd-immunopositive cells in the SN (Fig. 1a). The mean number of BrdUrd-labeled cells in the mouse SN matched those reported by Zhao et al. (16) (Table 1), suggesting that we labeled the same quantity of dividing cells despite using only half the BrdUrd concentration. This finding is consistent with another report showing that BrdUrd incorporation plateaus with high concentrations, probably because of the saturation of the nucleoside transporter systems (24). Many proliferating cells were found in the vicinity of TH-immunopositive neurons in the SN (Fig. 1a), and up to 16 BrdUrd-labeled cells per rodent SN were located immediately adjacent to TH neurons (18). When studied in a standard fluorescence microscope, up to three cells per rodent SN appeared to be colabeled with BrdUrd and TH and therefore potentially could be newly born nigral dopaminergic neurons. However, when all of these cells in each animal were individually further scrutinized with three-dimensional laser confocal analysis, none of them convincingly exhibited both markers colocalized (Table 1). The closest example to colocalization in the SN from all of the tissue examined is illustrated in Fig. 1b.

Fig. 1.

Progenitor cells in the adult rodent SN do not display neuronal or dopaminergic phenotypes. (a) Distribution of BrdUrd (red) and TH (green) immunoreactive cells in the mouse SN after 21 days of BrdUrd (75 mg/ml) minipump infusion into the right lateral ventricle. (b) Laser confocal (Leica) image of a BrdUrd-labeled cell adjacent to a TH-labeled cell, viewed in three dimensions. Note that this is not a double-labeled cell because the orthogonal view reveals that the BrdUrd-labeled nucleus is not completely encapsulated by the TH-labeled cytoplasm. (c) Distribution of BrdUrd (red) and TH (green) immunoreactive cells in the 6-OHDA-lesioned rat SN after BDNF infusion (20). Note the dramatic reduction of TH-labeled cells in the SN as compared with the nonlesioned condition in a. (d) Confocal image of a BrdUrd-labeled cell adjacent to a TH-labeled cell from c. No BrdUrd+ cells coexpressed TH. (e) Confocal image of a red BrdUrd-labeled cell adjacent to a blue TH-labeled cell with its own green NeuN-labeled nucleus, demonstrating that the BrdUrd nucleus did not originate from the TH neuron. (Scale bar = 100 μmin a and c and 5 μm in b, d, and e.)

Table 1. BrdUrd labeling and BrdUrd/TH colabeling in the SN.

Animals	Treatment	Total BrdUrd+ cells	BrdUrd+ / TH+ cells
Mice* (n = 2)	150 mg/ml BrdUrd	2,110 ± 320	22 ± 2
Mice (n = 6)	75 mg/ml BrdUrd	1,903 ± 267	0
Rats (n = 4)	75 mg/liter BrdUrd	3,025 ± 1,051	0

We implanted adult mice and rats (9–10 weeks old) with osmotic minipumps filled with BrdUrd (75 mg/ml delivering 0.25 μl/h for 21 days) into the right lateral ventricle. Values represent mean ± SEM.

^*Data are from Zhao et al. (16)

No Newly Generated Dopamine Neurons in the SN of Hemi-Parkinsonian Rats or After Growth Factor Treatment. Similar to an earlier report (20), we also observed abundant BrdUrd-immunopositive cells in the subventricular zone (SVZ) and in structures surrounding the lateral ventricles, e.g., the striatum, after BDNF administration. Immunopositive BrdUrd cells were also detected in the 6-OHDA-lesioned SN of both vehicle and BDNF (217 ± 48 and 493 ± 223 cells per right SN, respectively) (Fig. 1c). However, confocal analysis of 381 BrdUrd-labeled cells closely associated with any remaining TH cells revealed that none of them coexpressed TH (closest example in Fig. 1d), which is consistent with previously published results (18). Moreover, we further determined that the BrdUrd cells closely associated with TH neurons were indeed two different cells, because all TH neurons examined had a NeuN-labeled nucleus that was distinctly separate from the BrdUrd nucleus (Fig. 1e). Under no conditions did we observe any cells immunoreactive for the neuroblast marker doublecortin (25) in the SN.

No Evidence of Migrating Neural Stem Cells to the SN from the Cerebroventricular System. Intracerebroventricular injections of DiI resulted in fluorescent labeling of the lining of the entire ventricular system and of various brain structures, including the SN, ventral tegmental area, thalamus, hypothalamus, and neocortex (Fig. 2 a–c). However, unlike the labeling pattern observed after i.c.v. BrdUrd administration of the previous experiment, no DiI-labeled cells were observed in structures directly adjacent to the lateral ventricles. However, weakly labeled DiI fragments (i.e., extracellular, string-like formations) could be detected in all brain regions directly bordering the ventricular system (Fig. 2a). Contrary to previous reports (16), we did not observe a distinct stream of DiI-labeled cells migrating along the ventral midline to the SN (Fig. 2a). Direct injection of DiI into the striatum resulted in a zone of strong DiI labeling in and around the injection tract in the absence of labeling of the ependymal cells lining the ventricular walls (Fig. 2d). Striatal DiI injections also resulted in several strongly labeled cells with a neuron-like morphology in the SN, thalamus, and restricted layers of the neocortex (consistent with the pyramidal cell layers that project to the striatum). Because there was no labeling of ependymal cells, it is highly unlikely that these DiI-positive cells located outside the striatum were derived from newly divided ependymal cells. When comparing the total number of DiI-labeled cells coexpressing TH in the SN, no significant differences were detected between rats receiving i.c.v. (886 ± 102 ipsilateral and 789 ± 44 contralateral to the DiI injection site; not significant) versus striatal injections (927 ± 147 per side) (Fig. 2 e and g). Moreover, i.c.v. injections resulted in DiI labeling restricted to the medial aspect of the SN in all rats, whereas striatal injections consistently resulted in more lateral labeling of the SN (Fig. 2 e and g). The pattern of DiI labeling within the TH neurons, with numerous intensely fluorescent granule-like bodies throughout the perikarya, appeared to be identical regardless of whether the rat had received an intrastriatal or i.c.v injection (Fig. 2 f and h). To examine whether DiI-labeling in the SN was contingent on retrograde transport in an intact nigrostriatal pathway, some rats first underwent a unilateral 6-OHDA lesion of the SN. In the striatum, DiI was injected contralateral or ipsilateral plus contralateral to the lesion site; in the i.c.v. injection, DiI was injected contralateral to the lesion. In all of these rats, there were no detectable DiI-labeled cells in the SN on the lesioned side, regardless of whether DiI was injected into the ventricle or the striatum (Fig. 2 i and j). As with the previous experiment, no doublecortin immunopositive cells were observed in either the intact or lesioned SN of DiI-treated rats, again suggesting that no new neurons reside in the adult SN.

Fig. 2.

DiI acts as a retrograde tracer when injected i.c.v. (a) At 21 days after DiI injection i.c.v., labeling of the midbrain aqueduct is still evident, with very few cells observed in the adjacent parenchymal structures. Similar to striatal injections, i.c.v. DiI injections resulted in retrograde labeling of the major striatal afferent pathways: selective layers of the neocortex (b) and various thalamic nuclei (c) are shown. (d) Striatal DiI injections did not label the ependymal lining of the lateral ventricular wall. (e and g) A similar quantity of DiI (red) and TH (green) immunoreactive cells after either i.c.v. (e) or striatal (g) DiI injections. A three-dimensional laser confocal image of DiI incorporation into a TH cell (green) reveals no differences in the labeling pattern between i.c.v. (f) and striatal (h) DiI injections. Rats having their SN lesioned with 6-OHDA exhibit no DiI labeling, regardless of the injection location. (Scale bar = 100 μm in a–d, 20 μm in e, g, i, and j, and 5 μm in f and h.)

Discussion

Our findings are in disagreement with those reported by Zhao et al. (16), who suggest that new neurons are continuously added to the SN and that this process is up-regulated during injury. Using similar procedures, we were unable to find definitive proof of even one newly generated dopamine neuron in the SN. In a study published by Lie et al. (18), there was also no evidence for newly born dopaminergic cells in the adult SN. However, the report by Zhao et al. (16) suggests that the rate of neurogenesis in the SN is low. Therefore, it could be argued that the study by Lie et al. (18) used brain sections that were too thick (40 μm) and were sampled too infrequently (every sixth section was examined) to guarantee that any rare newly born cells would be detected. They administered BrdUrd by daily i.p. injections (18) instead of continuous infusion into the ventricular system (16). In the present study, we optimized chances of detecting BrdUrd/TH double-labeled cells by administering the BrdUrd into the right lateral ventricle over 21 days, examining every single section at 30-μm thickness. Based on the study by Zhao et al. (16), we should expect to observe as many as 22 BrdUrd/TH colocalized cells per mouse SN, or ≈1 BrdUrd/TH coexpressing cell per 96 BrdUrd-immunopositive cells in the SN. Instead, we observed no BrdUrd/TH-colabeled cells whatsoever, in either mice or rats. In our study, we used the same delay between BrdUrd administration and analysis of the brains as in ref. 16. We believe it is unlikely that the discrepancy between the results can be explained by lower BrdUrd incorporation in our experiment because we observed similar numbers of newly born cells in the mouse SN (Table 1) as reported previously (16). Moreover, according to results obtained in a different group of mice presented in the same study (16), daily intermittent i.p. administration of BrdUrd, which results in a lower level of BrdUrd incorporation in the brain (as revealed by a lower number of labeled glia and neurons) than continuous i.c.v. infusion, is still enough to detect at least one BrdUrd/TH coexpressing cell per mouse SN. We also demonstrate that this event is not dependent on the species, because we found no new nigral dopaminergic neurons in either mice or rats by using the same procedures (Table 1). To conclude, there are a few minor discrepancies between our methodologies and those used in the earlier study (16), such as the concentration and pH of BrdUrd, the tissue processing, antibodies used, and the housing conditions. However, we believe it is highly unlikely that any of these variables could truly account for our inability to detect even one newly generated dopamine neuron in the SN. Possibly, we adopted stricter criteria for the definition of truly double-labeled BrdUrd/TH-positive neurons than the earlier study that reported neurogenesis in the adult SN (16). We were unable to determine the phenotype of the BrdUrd-labeled cells in the animals receiving BrdUrd i.c.v., because all sections containing the SN were used for BrdUrd/TH double-immunohistochemistry. However, the study by Lie et al. (18) showed that almost half of BrdUrd cells in the SN express early glial progenitor markers, but few cells express mature glial markers, when analyzed 4 weeks after BrdUrd administration (18). Others have shown that SN BrdUrd cells do not coexpress most established markers of mature brain cells, suggesting that these newborn cells may remain as uncommitted progenitors (17). Alternatively, BrdUrd incorporation in the SN may not specifically occur during mitosis. It has been speculated that BrdUrd can be incorporated with DNA nuclear repair (14) or, more recently, in neurons undergoing apoptotic cell death (see below).

According to Zhao et al. (16), the rate of SN neurogenesis increases with partial dopaminergic lesions to the SN. We opted for injections of 6-OHDA into the median forebrain bundle, which results in complete destruction of the SN with minimal likelihood of dying dopaminergic neurons 3 weeks after the injection (19). With this model, we detected increases in cell proliferation in the SN, similar to those reported with the MPTP models (16, 17). However, none of these newly born cells expressed a dopaminergic phenotype, as has been reported elsewhere (18). It has been argued that a nearly complete loss of SN neurons achieved with 6-OHDA lesioning may not be permissive to neurogenesis (16). However, in an earlier study, Kay and Blum (17) were unable to detect new neurons in the mouse SN after partial lesion caused by MPTP administration.

As an alternative explanation, it has been demonstrated that mature SN dopaminergic neurons undergoing 6-OHDA-induced apoptotic cell death can express various cell-cycle markers, as well as incorporate BrdUrd (26), and dying cells can express these markers for extended periods of weeks after brain insults (P. Rakic, personal communication). Although speculative, it is plausible that the higher concentration and/or pH of the BrdUrd solution used by Zhao et al. (16) may have induced an inflammatory response that resulted in extended cell death in mature TH-expressing neurons that subsequently incorporated BrdUrd. Moreover, it has been demonstrated that MPTP treatment may cause SN dopaminergic neurons to become atrophic and dysfunctional rather than to die (27), possibly implicating that these compromised cells may also incorporate BrdUrd. Because of the potential complications in interpreting neurogenesis with BrdUrd incorporation, our study also relied on another independent marker of neurogenesis, namely doublecortin (25). As expected, there were numerous doublecortin-labeled cells in the dentate gyrus of the hippocampus. However, we found no doublecortin immunopositive cells in the SN, suggesting that there is no neurogenesis in the adult SN.

The study by Zhao et al. (16) used DiI to label newly generated neurons originating from the subependymal layer of the lateral ventricles (i.e., SVZ) to further substantiate their findings of neurogenesis in the adult SN. After i.c.v. injections of DiI, they showed that the numbers of cells coexpressing DiI and TH in the SN increased the longer the delay after DiI administration, with the first observations of positive cells reported to occur at ≈10 days after injection. Moreover, some DiI/TH double-labeled cells were even found to be labeled by fluorogold, when this retrograde tracer was injected into the striatum (16). The findings were interpreted as support for newly born cells migrating from the SVZ, maturing into a dopaminergic phenotype, and extending axons to the striatum. It is well documented that DiI, via retrograde transport, can label nigral neurons when injected to the striatum (21), and it has been suggested that 7–14 days is the minimum time required for reliable retrograde DiI labeling of TH neurons to appear in the SN (21) (T. Hagg, personal communication) after intrastriatal DiI injections. Therefore, we examined whether DiI injected into either the ventricle or the striatum could label similar numbers of SN dopaminergic neurons as those reported to be newly born in the study of Zhao et al. (16). When we injected DiI into the ventricle of rats, we replicated the initial findings (16) in that we observed massive fluorescent labeling of the ependymal lining of the ventricles and several hundred DiI-labeled TH-immunopositive neurons in the SN. However, when we purposely injected DiI directly into the striatum, without any of the neighboring ependymal cells taking up the DiI, we also observed a comparable number of DiI-labeled TH-positive nigral neurons. The known mediolateral topography of the rodent nigrostriatal system would suggest that DiI injections placed in the medial striatum would primarily be transported retrogradely by dopamine neurons located in the medial SN (28). For injections placed in the central striatum, one would expect that neurons in a more lateral aspect of the SN would be labeled. Indeed, these predictions fit well with the results concerning the location of DiI-labeled nigral neurons in our experiment. There were also numerous DiI-labeled cells showing mature neuronal morphologies and excellent anatomical integration, e.g., in the thalamus and neocortex (29), all of which are regions that are known to project to the striatum. Thus, the pattern of DiI labeling was entirely consistent with the dye acting as a retrograde tracer. Indeed, several earlier studies have already used DiI as a retrograde tracer to map afferents to the striatum (21, 30). Importantly, in our study, the rats receiving DiI injections into the lateral ventricles also exhibited numerous DiI-labeled cells in regions known to project to the striatum. This observation was not presented in the original study by Zhao et al. (16), with the exception that they described cells retrogradely labeled by DiI in the raphe nucleus. Therefore, we asked whether the DiI-labeled cells in the SN could be the result of retrograde transport of the lipophilic dye in the nigrostriatal pathway. As could be expected, in animals with complete and prior 6-OHDA lesions of the SN, the injection of DiI into the striatum resulted in no labeled cells in the SN on the lesioned side. Perhaps less expected in light of the study by Zhao et al. (16), there were no DiI-labeled neurons in the SN on the lesioned side when the dye was injected into the lateral ventricle of rats with prior 6-OHDA lesions. Thus, although we could not exclude that the 6-OHDA lesion rendered the nigral microenvironment unfavorable to migration and maturation of newly born neurons, our observations suggest that no measurable neurogenesis takes place in the adult SN. Instead, the previously reported DiI labeling in the SN is likely to have been the result of retrograde transport after uptake of the tracer from striatal areas close to the injected ventricular system, where the fluorescent label could be seen to diffuse. In support, dopaminergic fibers are known to densely innervate the SVZ close to the ventricle (31).

Zhao et al. (16) also applied the antimitotic agent cytosine-d-arabinofuranoside (Ara-Cyt), which completely abolished DiI-labeling in the SN (16). Ara-Cyt does not exclusively prevent mitosis of stem cells in the brain, and cytotoxicity to mature neurons is a well known side effect in patients treated with Ara-Cyt for hematologic malignancies (32). Likewise, several reports show that fully differentiated neurons undergo apoptotic cell death as a result of Ara-Cyt treatment in vitro (33, 34). Therefore, it cannot be concluded that a reduced number of DiI-labeled cells after Ara-Cyt treatment is due exclusively to a selective inhibition of mitosis. They further demonstrate that MPTP lesions increase the total number of DiI/TH-labeled neurons in the SN (16). MPTP treatment can cause a compensatory increase in dopaminergic fiber sprouting in the striatum over time (35). Therefore, one can speculate that neurons with a sprouting terminal network can take up more DiI from the ventricles and thereby, through retrograde labeling, increase the proportion of TH/DiI cells in the SN.

In conclusion, we suggest that if neurogenesis occurs in the adult rodent SN, it is an exceptionally rare event that is not readily detectable with standard methods. Here, we have not addressed certain findings (e.g., [³H]thymidine labeling of adult nigral dopamine neurons) reported in by Zhao et al. (16); therefore, we cannot absolutely rule out that new dopaminergic cells can be generated in the SN under extremely defined conditions. Because the motor disturbances associated with Parkinson's disease are symptoms of loss of dopaminergic neurons, the previously reported findings of SN neurogenesis (16) were of potentially great importance with respect to using endogenous neurons to replace lost dopaminergic cells in Parkinson's disease (15). However, given that we, and others, have used several approaches and still fail to find any conclusive evidence of actual occurrence of neurogenesis in the adult SN, we are skeptical of this therapeutic approach for Parkinson's disease.