Address correspondence and reprint requests to Abbas F. Sadikot, Department of Neurology and Neurosurgery, Montreal Neurological Institute, McGill University, Montreal, Quebec, H3A 2B4, Canada. E-mail: firstname.lastname@example.org
The homeodomain transcription factor Pitx3 is critical for the survival of midbrain dopaminergic (mDA) neurons. Pitx3-deficient mice exhibit severe but selective developmental loss of mDA neurons, with accompanying locomotor deficits resembling those seen in Parkinson's disease (PD) models. Here, we identify specific mDA cell subpopulations that are consistently spared in adult Pitx3-hypomorphic (aphakia) mice, demonstrating that Pitx3 is not indiscriminately required by all mDA neurons for their survival. In aphakia mice, virtually all surviving mDA neurons in the substantia nigra (SN) and the majority of neurons in the adjacent ventral tegmental area (VTA) also express calbindin-D28k, a calcium-binding protein previously associated with resistance to injury in PD and in animal models. Cell-mapping studies in wild-type mice revealed that Pitx3 is primarily expressed in the ventral SN, a region particularly susceptible to MPTP and other dopaminergic neurotoxins. Furthermore, Pitx3-expressing SN cells are preferentially lost following MPTP treatment. Finally, SN mDA neurons in Pitx3 hemizygous mice show increased sensitivity when exposed to MPTP. Thus, SN mDA neurons are represented by at least two distinct subpopulations including MPTP-resistant Pitx3-autonomous, calbindin-positive neurons, and calbindin-negative Pitx-3-dependent cells that display elevated vulnerability to toxic injury, and probably correspond to the subpopulation that degenerates in PD. Impairment of Pitx3-dependent pathways therefore increases vulnerability of mDA neurons to toxic injury. Together, these data suggest a novel link between Pitx3 function and the selective pattern of mDA cell loss observed in PD.
Parkinson's disease (PD) is a progressive multisystem neurodegenerative disorder characterized primarily by locomotor deficits arising from massive loss of midbrain dopaminergic (mDA) neurons (Hornykiewicz 1966; Braak et al. 2003; Jellinger 2012). Multiple nuclei comprise the mammalian midbrain mDA system including the substantia nigra (SN) pars compacta (SNc, nucleus A9), ventral tegmental area (VTA, nucleus A10), and retrorubral field (RRF, nucleus A8). Among these DA neuronal subpopulations, the DA neurons of the ventrolateral SN are particularly susceptible in PD (Sourkes and Poirier 1965; Hornykiewicz 1966; Yamada et al. 1990; German et al. 1992). On the other hand, select DA subpopulations within and outside the midbrain tend to resist the degenerative process (Matzuk and Saper 1985; Fearnley and Lees 1991; German et al. 1992). The factors which determine why some mDA neurons are selectively vulnerable while others resist the degenerative process in PD are poorly understood.
Homeodomain transcription factors play a critical role in the specification and maintenance of mDA neurons (Wallen and Perlmann 2003; Andersson et al. 2006). We and others have previously shown that Pitx3, whose expression in the brain is restricted to mDA neurons (Semina et al. 1997; Smidt et al. 1997), plays a central role in their post-mitotic survival (Hwang et al. 2003; van den Munckhof et al. 2003; Nunes et al. 2003; Smidt et al. 2004). Aphakia (Pitx3ak/ak) mice, which harbor a functional deletion of the Pitx3 gene (Varnum and Stevens 1968; Rieger et al. 2001), exhibit profound mDA cell loss associated with massive nigrostriatal degeneration. Dopamine content in the dorsal striatum of aphakia mice is reduced by over 90% (Hwang et al. 2003; van den Munckhof et al. 2003), leading to locomotor and behavioral deficits, which are partially reversible with L-DOPA administration (van den Munckhof et al. 2003, 2006; Hwang et al. 2005; Jacobs et al. 2009; Ardayfio et al. 2010; Beeler et al. 2010). In contrast, mesolimbic dopaminergic projections are preferentially spared in both Pitx3ak/ak and Pitx3−/− mice (Kas et al. 2008), thus recapitulating a pattern of cell loss resembling that seen in patients with PD. This pattern of mDA cell loss in Pitx3-deficient mice also shows a remarkable similarity to that observed in a number of animal models of PD, suggesting that the observed dependency of these neurons on Pitx3 during developmental apoptosis may be linked to their heightened vulnerability during adulthood. For example, rodents and primates exposed to the dopaminergic neurotoxin MPTP develop a Parkinsonian syndrome, and show loss of mDA neurons in the ventrolateral SNc, which give rise to the nigrostriatal pathway. In contrast, mDA neurons in the dorsal SNc and the adjacent VTA are spared (Yamada et al. 1990; Fearnley and Lees 1991; Lavoie and Parent 1991; German et al. 1992). In addition, an increasing number of studies implicate alterations of the PITX3 gene in both sporadic and early-onset forms of PD (Bergman et al. 2010; Yu et al. 2010; Haubenberger et al. 2011). Given these parallel findings, we hypothesized that subpopulations that are susceptible to loss of Pitx3 function correspond to mDA neurons which are preferentially lost in human PD or following exposure to neurotoxins such as MPTP. Furthermore, we proposed that loss of Pitx3 function in adult mice would render mDA neurons more susceptible to toxic injury.
Here, we have determined the precise subpopulations of mDA neurons, which are lost in Pitx3ak/ak animals, and the relationship between mDA neurons containing Pitx3 and mDA neurons with known resistance to neurodegeneration in PD. Our results show that DA neurons in the ventral SNc are selectively vulnerable to developmental loss in the absence of normal Pitx3 function, with sparing of mDA neurons in the dorsal SNc. Nearly all surviving nigral neurons in Pitx3ak/ak mice are distinguished by their expression of the calcium-binding protein, calbindin-D28k (CB), a marker of resistant mDA neurons in PD and neurotoxin-based models of PD employing either MPTP or 6-hydroxydopamine (6-OHDA) (Yamada et al. 1990; Lavoie and Parent 1991; German et al. 1992). Furthermore, Pitx3- or CB-expressing mDA populations are largely complementary with only a minority subpopulation expressing both markers. Exposure to MPTP in wild-type animals results in preferential loss of Pitx3-expressing neurons in the SNc, while CB-positive/Pitx3-negative mDA neurons were relatively preserved. Reduced Pitx3 gene dosage increases the vulnerability of mDA SN neurons to MPTP, indicating that normal function of a Pitx3-dependent survival pathway is necessary for resistance to toxic injury. Together, these data suggest that multiple populations of mDA neurons, distinguished by their dependence on Pitx3 signaling and resistance to degenerative stresses, coexist in the mammalian midbrain.
Methods and materials
Animals and MPTP treatment
All animal procedures were performed in accordance with the Canadian Council on Animal Care guidelines for the use of animals in research, as administered by the McGill University Animal Care Committee. Aphakia (Pitx3ak/ak) mice were backcrossed and maintained on the C57BL/6J background. To determine the effects of the Pitx3ak/ak genotype on mDA neuron survival in adult and aging animals, male Pitx3ak/ak and Pitx3wt/wt mice were transcardially perfused with 4% buffered paraformaldehyde at post-natal days (P) 1, 21, 35, 100, or > 700. Brains were then removed, post-fixed for 24 h, and immersed in phosphate-buffered sucrose (30%) for 48 h. Sectioning was performed using a freezing microtome (Microm, Walldorf, Germany) at 40 μm. Serial coronal sections were collected in phosphate-buffered saline and immunostained for stereology. To determine whether mDA survival is altered in heterozygous Pitx3ak/wt mice, a separate cohort of Pitx3ak/wt and Pitx3wt/wt littermates were processed in a similar manner. For MPTP experiments, wild-type (Pitx3wt/wt) or heterozygous (Pitx3ak/wt) littermates were used. Genotyping was performed by polymerase chain reaction as described in previous work (van den Munckhof et al. 2003). To determine if Pitx3+ mDA neurons show altered susceptibility to toxic injury, Pitx3wt/wt littermates received four i.p injections (20 mg/kg each) of (MPTP, 5 mg/mL in saline; Sigma, St Louis, MO, USA). Injections were made 4 h apart when animals reached P35. Control animals were injected four times with a similar volume of saline. In a separate experiment, to determine the effect of Pitx3 gene dosage on susceptibility to toxic injury, Pitx3ak/wt or Pitx3wt/wt P35 littermates received a similar MPTP regimen. Animals were killed 10 days after treatment and the brains processed for immunohistochemistry (see below). For each experiment described above, immunohistochemistry was performed simultaneously for comparison groups. Stereological analysis in each comparison group was performed by the same observer.
Statistical analysis was performed using either t-test (unpaired or paired) or two-way anova with Tukey's HSD post hoc test, as appropriate. Main and interaction effects were calculated using SAS software (v. 6.12, Cary, NC, USA) or Datasim (v. 1.1, Drake Bradley, Bates College, ME, USA).
Complete rostrocaudal coronal series representing every sixth section (i.e., 240-μm intervals) were immunostained for tyrosine hydroxylase (TH), CB, or Pitx3 using an avidin–biotin–peroxidase complex method as previously described (van den Munckhof et al. 2003). Briefly, sections were incubated in anti-TH (1 : 1000; Immunostar Inc., Hudson, WI, USA), anti-CB (1 : 2000; Sigma), or anti-Pitx3 antibody (1 : 100; Lebel et al. 2001) overnight after blocking (5% bovine serum albumin in phosphate-buffered saline containing 0.1% Triton-X100). Sections were then washed, incubated for 1 h with biotinylated anti-mouse secondary antibody (1 : 200; Vector), followed by avidin–biotin complex (Vector, Burlingame, CA, USA) as per manufacturer's instructions. The final reaction was revealed by 3,3′-diaminobenzidine. Sections were counter-stained lightly in 0.1% cresyl violet, dehydrated, and coverslipped. For colocalization experiments, double immunofluorescence for TH, CB, or Pitx3 was performed on 5-μm sections from paraffin-embedded mouse brains as previously described (van den Munckhof et al. 2003). Confocal images (40×) were collected using a Zeiss LSM510 microscope (Carl Zeiss Microscopy, Jena, Germany).
Stereology and computer-assisted cell mapping
Neurons positive for TH, CB, or Pitx3 were quantified using the optical dissector method as previously described (van den Munckhof et al. 2003; Landau et al. 2005). The entire rostrocaudal extent of the midbrain was examined in TH-, CB-, or Pitx3-stained coronal sections (1 : 6 series) using an Olympus BX-40 microscope equipped with a motorized XYZ stage and StereoInvestigator software (Microbrightfield, Williston, VT, USA). In non-TH-stained sections, the SNc was defined ventrolaterally by the SN pars reticulata and medially by the VTA or medial lemniscus (Nelson et al. 1996). Dopaminergic neurons of the SNc and VTA were distinguished by their size and mediolateral orientation (Bjorklund and Lindvall 1975). After tracing the SN and VTA at low power, TH-, CB- or Pitx3- cell counts were performed using the Optical Fractionator stereological probe at 100 × magnification (oil, NA 1.4) by employing a 60 × 60-μm counting frame in conjunction with a 12-μm dissector placed 2 μm below the surface of the section. Counting sites were assigned using a randomly placed 125 × 125-μm grid.
Midbrain DA neurons expressing CB are Pitx3 autonomous
While adult Pitx3ak/ak mice show a marked reduction in total mDA neuron number when compared with Pitx3wt/wt littermates, we consistently observed persistent subpopulations of TH-positive cells (Fig. 1) in these animals. In agreement with previous reports, TH+ cell loss was non-uniform and most pronounced within the SNc, with prominent loss of the nigrostriatal projection to the neostriatum (Hwang et al. 2003; van den Munckhof et al. 2003; Nunes et al. 2003). In contrast, TH+ cells in the VTA and mesolimbic mDA projections to the nucleus accumbens and olfactory tubercle were relatively spared (Jacobs et al. 2009, 2011). Remaining mDA neurons in Pitx3ak/ak animals were located predominantly in the dorsal part of the SNc, and also in a scattered distribution within the VTA. This pattern of mDA cell loss was observed in all Pitx3ak/ak mice examined, and is reminiscent of observations in human PD and in animal models using 6-OHDA or MPTP (Lavoie and Parent 1991; German et al. 1992).
In rodents, mDA neurons in the dorsal SNc are typically distinguished by the coexpression of calcium-binding proteins, namely calbindin-D28k (CB) and calretinin (McRitchie et al. 1996; Nemoto et al. 1999; Tsuboi et al. 2000; Fu et al. 2012). Interestingly, CB expression has been well documented in resistant mDA neurons in PD patients (Yamada et al. 1990; German et al. 1992; Damier et al. 1999), suggesting a subpopulation conserved across species capable of withstanding a variety of insults. We therefore examined whether the Pitx3-autonomous subpopulation in the SN and VTA corresponded to neurons expressing this marker, using double immunostaining for CB in persisting TH+ cells in Pitx3ak/ak mice at post-natal day (P) 100 (Fig. 2a–f). Examination of coronal sections at regular intervals across the SN revealed CB immunoreactivity in 92% (SNc) and 87% (VTA) of TH+ cells (Fig. 2g–j). Colocalization of these two markers was also relatively uniform, reaching as high as 97% at some coronal levels examined, indicating that the majority of surviving mDA neurons in Pitx3ak/ak animals are CB+.
In agreement with this observation, stereological quantification of the total number of TH+ neurons in the SNc of 35-day-old Pitx3ak/ak and Pitx3wt/wt animals (Fig. 2k) revealed a significant 80 ± 2.5% (mean ± SEM two-tailed t-test, p < 0.0001) reduction in Pitx3ak/ak compared with Pitx3wt/wt mice. In contrast to TH+ cells, no difference in total CB+ cell number could be detected between the two groups (Fig. 2l). Consistent with this observation, the ratio of total CB+ neurons to TH+ neurons in the SNc of Pitx3ak/ak mice reached 95.5 ± 4.6% compared with only 24.5 ± 1.6% in Pitx3wt/wt animals (mean ± SEM, two-tailed t-test, p < 0.0001, Fig. 2m), strongly supporting the notion that the survival of the CB+ subpopulation of mDA neurons located in the dorsal tier of the SNc is unaffected in the absence of Pitx3.
In agreement with previous studies (van den Munckhof et al. 2003; Smidt et al. 2004), TH+ cell loss in the VTA was less severe than in the SN. Nonetheless, in comparison with Pitx3wt/wt mice, the VTA of Pitx3ak/ak animals also showed a significant 32 ± 1.9% (mean ± SEM, two-tailed t-test, p < 0.01) reduction in the number of TH+ cells, indicating that Pitx3 serves as a survival factor for a subset of mDA VTA neurons. However, in contrast to the SNc, where the CB+ cell number remained unchanged and virtually all remaining TH+ neurons expressed CB, the total number of CB+ cells in the VTA was reduced by 19 ± 4.2% (mean ± SEM, two-tailed t-test, p < 0.01) in Pitx3ak/ak mice. Although the majority of VTA CB+ neurons remained intact (Fig. 2l and m), this cell loss suggests that CB expression alone is not sufficient to protect VTA mDA neurons from degeneration, and contrasts with the observation that CB is a robust marker of Pitx3-independent mDA neurons in the SNc.
Age-dependent loss of Pitx3-dependent mDA neurons
To determine whether mDA subpopulations undergo further degeneration in the absence of Pitx3 function as a function of age, TH+ and CB+ neurons in Pitx3ak/ak and Pitx3wt/wt animals at various ages (P1, 21, 35, 100, and 700) were quantified using stereology (Fig. 3). Within the SNc, the number of TH+ cells stabilized in both Pitx3wt/wt and Pitx3ak/ak animals after the fifth post-natal week with no significant alterations in SNc mDA cell counts at P35, 100, or 700 (Fig. 3a). The number of TH+ mDA in the SNc neurons in Pitx3ak/ak mice was significantly reduced compared with Pitx3wt/wt mice at all ages (two-way between-subject anova with Tukey's post hoc test, p < 0.01). The ratio of CB+ to TH+ neurons in the SNc was significantly higher in Pitx3ak/ak mice at both P35 and P700 (two-way between-subject anova with Tukey's post hoc test, p < 0.01). Indeed, the vast majority of surviving TH neurons in SNc of Pitx3ak/ak mice are CB+. Importantly, the ratios of CB+ neurons to TH+ neurons remained unchanged in SNc between P35 and P700 in both wild-type and Pitx3ak/ak mice (Fig. 3c), suggesting that CB+ neurons remain unaffected in the absence of Pitx3 for prolonged periods, and is concordant with the selective loss of TH+CB− cells in the SNc of Pitx3ak/ak mice.
In contrast, total TH+ cell number in the VTA, which was similar in both genotypes at birth, was reduced in Pitx3ak/ak mice by P21 (two-way between-subject anova with Tukey's post hoc test, p < 0.05), with further reduction at P35–P700 (p < 0.01) (Fig. 3b). Hence, loss of Pitx3-dependent subpopulations in these two regions follow different temporal trajectories with loss in the SNc preceding that of the VTA. Following P35, Pitx3ak/ak mice showed a mild reduction in TH+ cell number in the VTA with aging, although this decrease did not reach statistical significance. Interestingly, preferential sparing of CB+ neurons was not detectable at P35, (Fig. 3d), but with further loss of VTA DA neurons (Fig. 3b), CB−TH+ neurons are preferentially lost compared with CB+ neurons at P700 (Fig. 3d). Thus, while CB expression in SNc mDA neurons strongly correlates with survival in the absence of Pitx3, this relationship is not as robust in the VTA.
Pitx3 expression is heterogeneous in SNc neurons
The presence of a Pitx3-autonomous subpopulation led us to hypothesize that Pitx3 may not be homogenously expressed among mDA neurons as implied by initial gene-expression studies (Smidt et al. 1997; Korotkova et al. 2004; Zhao et al. 2004). Therefore, we directly compared TH and Pitx3 expression by double immunostaining midbrain sections from Pitx3wt/wt mice. Confocal microscopy revealed that the majority of TH+ cells showed strong nuclear Pitx3 immunostaining (Fig. 4). The vast majority of Pitx3 signal was detected only within TH+ cells in the SNc, VTA, RRF, and a small number of TH+ neurons in the periaquaductal gray area (Fig. 4a–i and data not shown). Rostral–caudal mapping of sections revealed that TH+Pitx3- neurons intermingled with TH+Pitx3+ cells within the VTA (Fig. 4c,j). In contrast to this arrangement, the SNc showed a distinctive spatial segregation with TH+Pitx3− cells located predominantly in the dorsal tier (Fallon and Moore 1978; Gerfen et al. 1987) adjacent to TH+Pitx3+ cells that were localized to the ventral tier (Fig. 4f, k, m). Both markers were also detected in the so-called ventrally displaced mDA neurons of the SNc in the pars reticulata (Bjorklund and Lindvall 1975). Rare TH- Pitx3+ profiles were scattered in the SN ventral tiers and VTA, although these profiles did not appear to be nuclear, and most likely represent background staining, or Pitx3+ nuclei, or processes at the top or bottom of sections.
Double immunostaining for CB and Pitx3 (Fig. 5) was also used to confirm the chemical identity of dorsal-tier TH+Pitx3− cells, which also possess conspicuous horizontally oriented processes. CB+ neurons were distributed throughout the VTA and RRF, but restricted in the SNc to the dorsal tier (Fig. 5a, d). The majority (89%) of Pitx3+ neurons in the SNc were not CB+. Furthermore, 96% of CB+ cells lacked Pitx3 staining, although a small proportion of Pitx3+ neurons at the ventral/dorsal interface also expressed CB (Figs. 4 and 5a–d, g). Thus, two largely exclusive mDA subpopulations marked by Pitx3 and CB (Pitx3+CB− and Pitx3−CB+) are present in the SNc. As in the SNc, the majority (81%) of CB+ cells in the VTA were also Pitx3 negative (Figs. 5b-d,h).
Pitx3-expressing mDA neurons are vulnerable to MPTP
The observation that CB+ mDA neurons are spared following MPTP treatment or in the absence of Pitx3 suggests that, in contrast to neighboring Pitx3-autonomous CB+ cells, Pitx3-dependent neurons in the ventral SNc are selectively vulnerable to injury. To test this hypothesis, we quantified the total number of TH+, CB+, and Pitx3+ neurons in wild-type mice following exposure to MPTP using a subchronic dosing regimen (Fig. 6).
Pitx3wt/wt mice treated with MPTP (4 × 20 mg/kg, administered i.p. 4 h apart, n = 5) showed a 25% reduction in total TH+ cell number in the SNc (n = 5) compared with saline-treated (n = 5) animals. Interestingly, total Pitx3+ cell number in SNc was significantly reduced (p < 0.05, two-tailed t-test) in MPTP-treated Pitx3wt/wt mice (1444 ± 293, mean ± SEM) compared with saline-treated Pitx3wt/wt mice (3105 ± 126). This decrease of nearly 50% (Fig. 6a–c, g, h) suggests that Pitx3+ cells accounted for the bulk of SNc cell loss in response to this toxin (mean losses: 1339 TH+ cells and 1661 for Pitx3+ cells). In contrast, the number of CB+ neurons in MPTP- and saline-treated animals was comparable (Fig. 6e, f, i), consistent with their previously reported resistance to insults (Lavoie and Parent 1991; German et al. 1992; Liang et al. 1996). Consistent with the higher proportion of Pitx3-independent cells present relative to the SNc, the VTA did not show a significant change in total TH+ cell number with MPTP treatment (Figs. 2 and 4). Furthermore, Pitx3+ VTA neurons decreased by 23% (Pitx3wt/wt: 2666 ± 387; Pitx3ak/wt: 2079 ± 370; mean ± SEM, p < 0.05, two-tailed t-test) after MPTP administration. In contrast, the number of CB+ cells were actually slightly elevated (Fig. 6h,i). Together, these results indicate that Pitx3-expressing mDA neurons in the ventral SNc and the VTA display increased sensitivity in a commonly used animal model of PD.
Given the observation that SNc neurons vulnerable to MPTP are Pitx3 dependent, as indicated by their lack of CB expression, we determined whether susceptibility to injury is influenced by levels of the Pitx3 gene product. In initial experiments to determine whether reduced gene dosage has an effect on neuronal survival in the heterozygous state, we compared the number of TH+ midbrain neurons in Pitx3ak/wt mice with Pitx3wt/wt littermates. There was no significant difference in total number of TH-IR mDA neurons in Pitx3wt/wt compared to Pitx3ak/wt mice (Pitx3wt/wt: 7690 ± 677 vs. Pitx3ak/wt: 7310 ± 845, mean ± SEM). Next, heterozygous (Pitx3ak/wt, n = 3) and wild-type (Pitx3wtwt, n = 3) mice were subjected to a similar subchronic MPTP protocol as above, after which mDA neuron number was quantified (Fig. 7). Compared with Pitx3wt/wt mice, Pitx3ak/wt animals exposed to MPTP showed a 44% reduction in TH+ neuron number in the SN (3129 ± 324 vs. 1760 ± 74; mean ± SEM, two-tailed t-test, p < 0.01). The number of TH+ neurons in the VTA of MPTP-treated Pitx3ak/wt and Pitx3wt/wt mice were not significantly different. Thus, a reduction in Pitx3 substantially increases the vulnerability of SNc mDA neurons to MPTP, suggesting that Pitx3-dependent pathways may protect mDA neurons from degenerative stress.
The non-uniform, yet consistent, patterns of mDA neuron loss observed in PD (Kish et al. 1988; Damier et al. 1999; Halliday and McCann 2010) and numerous animal models of PD (Yamada et al. 1990; German et al. 1992; Liang et al. 1996) support the existence of multiple cell subpopulations with varying degrees of susceptibility to neurodegeneration. Identifying and characterizing these individual neuronal subpopulations and the factors which contribute to their differential vulnerability is critical toward understanding the mechanisms governing mDA degeneration in PD and related disorders. Among candidate molecules that stand out in this regard are a collection of homeodomain transcription factors whose expression is largely restricted to mDA neurons and play an important role in their specification, development, and/or maintenance. Indeed, disruption of many of these genes, namely Nurr1 (Zetterstrom et al. 1997), Engrailed 1/2 (Alberi et al. 2004; Sgado et al. 2006), Lmx1a (Ono et al. 2007; Friling et al. 2009), Lmx1b (Smidt et al. 2000; Asbreuk et al. 2002), and Pitx3 (Smidt et al. 1997; van den Munckhof et al. 2003; Nunes et al. 2003), have been shown to lead to prominent and selective cell loss in the mesencephalon. However, whereas ablation of master transcriptional molecules such as Nurr1 result in complete agenesis of mDA neurons (Zetterstrom et al. 1997), loss of Pitx3 function results in the loss of a selective, albeit major, subset of this population.
Dependence on Pitx3 segregates mDA neuron populations
Our findings indicate that the pattern of mDA neuronal loss resulting from deficient Pitx3 signaling closely resembles PD, whereby depletion is most severe in ventral SNc neurons, but considerably attenuated in dorsal SNc and VTA neurons. These results are consistent with the presence of multiple/discrete mDA subpopulations distinguished by their dependence on Pitx3 for survival. Importantly, the preservation of specific TH+ neuron subsets in Pitx3ak/ak mice indicates that not all mDA neurons require this transcription factor for developmental survival. We did not detect any further changes to this population in aged (> P700) Pitx3ak/ak mice, suggesting that Pitx3 is also dispensable for the maintenance of the surviving subpopulation throughout adulthood. On the other hand, Pitx3 appears to be critical for survival of ventral SNc neurons as well as for a minority of neurons distributed throughout the VTA. Interestingly, this same Pitx3-dependent subpopulation was also particularly susceptible to degeneration following MPTP treatment in wild-type animals, strongly implicating Pitx3 as a marker of vulnerable mDA subpopulations in this region. Consistent with their dependence on Pitx3, we observed using immunohistochemistry that Pitx3 expression is restricted primarily to ventral tier mDA neurons, and to scattered subpopulations within the VTA of wild-type animals. In addition to confirming that Pitx3-expressing mDA neurons are the main population lost in Pitx3ak/ak mice, this observation indicates that surviving Pitx3-autonomous cells are less vulnerable to stressors such as MPTP.
Our findings confirm previous reports that the vast majority of Pitx3-expressing neurons in the midbrain are dopaminergic (van den Munckhof et al. 2003; Smidt et al. 2004). In contrast, our observation that Pitx3 is differentially expressed among mDA neurons is somewhat at odds with previous reports examining localization of Pitx3 mRNA, suggesting that Pitx3 message is present in all mDA neurons (Smidt et al. 1997; Zhao et al. 2004; Maxwell et al. 2005). A plausible explanation for this perceived discrepancy is that Pitx3 expression depends on mechanisms beyond transcriptional regulation: for example, differential phosphorylation of translation-initiation factors or ribosomal proteins, micro-RNA (miRNA), and small-interfering RNA (siRNA) could result in variations in translation in dorsal-tier SNc neurons and half of VTA DA neurons. Additional studies are required to verify this. Alternatively, the higher resolution of quantitative mapping of Pitx3 and mDA populations using double immunohistochemistry compared with in situ hybridization may account for these differences.
Selective vulnerability in Pitx3+ mDA neurons
Our data point to a heightened susceptibility of Pitx3-dependent mDA neurons to developmental and neurotoxic insults. How this is mediated remains unclear. Interestingly, the majority of CB-expressing mDA neurons in the SNc and the VTA of wild-type mice appear to be Pitx3 autonomous as evidenced by their relative sparing in Pitx3ak/ak mice. Previous reports have shown that this group, which comprises the majority of mDA neurons in the dorsal SNc and a subpopulation of mDA neurons in the VTA in rodents, is relatively resistant to injury in PD (Yamada et al. 1990; Lavoie and Parent 1991; German et al. 1992; Liang et al. 1996). Similarly, CB-expressing mDA neurons are also spared in rodents following exposure to neurotoxins such as MPTP and 6-OHDA. Calcium-binding proteins, such as CB and CR, may provide neuroprotection by buffering intracellular calcium and attenuating the generation of harmful reactive oxygen species (Dauer and Przedborski 2003; Dawson and Dawson 2003; Surmeier et al. 2011). High intracellular levels of calcium may also mediate dopamine toxicity in neurons (Mosharov et al. 2009). The observation that another calcium-binding protein, calretinin, also colocalizes to resistant mDA subpopulations (Liang et al. 1996) further suggests that calcium homeostasis may be a major contributing factor. Along these lines, it has also recently been shown that the Cav1.3 class of voltage-gated L-type calcium channels expressed by SNc neurons renders them particularly susceptible to both MPTP and 6-OHDA exposure, and that blockade by the calcium-channel antagonist isradipine is neuroprotective (Surmeier et al. 2011). CB-expressing neurons in the VTA and dorsal SNc are also overwhelmingly spared in weaver mutant mice containing a mutation of the Girk2 inward-rectifier channel (Gaspar et al. 1994). Furthermore, single-nucleotide polymorphisms in the calbindin-1 gene have been associated with increased risk of PD in a Japanese population (Mizuta et al. 2008), adding to the argument that the buffering capacity of calcium-binding proteins are protective.
Nonetheless, it is noteworthy that previous studies failed to demonstrate detectable changes in mDA neuron survival in CB-null mice challenged with either MPTP or in the presence of the weaver (GIRK2) mutation (Airaksinen et al. 1997), suggesting that additional factors such as differences in target innervations and firing rate (Gerfen et al. 1987; Liss et al. 1999; Brown et al. 2009) may also play a determining role. For example, important regulators of dopamine, including dopamine transporter and vesicular monoamine transporter 2, are also differentially expressed between ventral and dorsal SNc regions (Haber et al. 1995; Sanghera et al. 1997).
The striking resemblance between the pattern of mDA neuron loss seen in Pitx3ak/ak mice, DA-toxin-induced neurodegeneration, and human PD (Yamada et al. 1990; Lavoie and Parent 1991; German et al. 1992) suggests a defect dependent on the Pitx3 pathway may underlie the vulnerability of these mDA subpopulations. During embryonic development, survival of mDA neurons is influenced by Pitx3 via transcriptional regulation of the enzyme aldehyde dehydrogenase 2, an enzyme responsible for retinoid production (Jacobs et al. 2007, 2011). Furthermore, Pitx3 appears to regulate key components of the dopamine metabolic pathway such as vesicular monoamine transporter 2 and Vip by forming a transcriptional complex with Nurr1 (Jacobs et al. 2009), another mDA transcription factor enriched in mDA populations. Pitx3 also participates in the feed-forward regulation of brain-derived neurotrophic factor and glial-derived neurotrophic factor (Peng et al. 2011), both of which are essential to mDA neuron survival. Significant reductions in Pitx3 expression within the CNS (Smidt et al. 1997) and peripheral tissues (Liu et al. 2011) have been reported in human PD, further supporting the view that impaired Pitx3-dependent functions are responsible for the loss of mDA neurons. Interestingly, genetic aberrations at the PITX3 locus or polymorphisms within promoter elements are also associated with increased risk of PD (Kim et al. 2007; Bergman et al. 2010; Haubenberger et al. 2011). Together with our observation that Pitx3 hemizygous mice are more susceptible to toxic challenge, it is likely that reduction in Pitx3 signaling pathway represents an upstream event leading to the impairment of dopamine metabolism and possibly other pathways. While it remains to be determined whether loss of Pitx3 function directly influences mature mDA neurons, our findings suggest that Pitx3 dependency in mDA neurons is closely linked to susceptibility to degeneration, and highlights the need for additional studies on the role of this transcription factor in both physiology and disease.
This work was supported by research grants to AFS and JD from the Canadian Institutes for Health Research and National Science and Engineering Research Council. VVR is a recipient of a Parkinson Society Canada fellowship.
Conflicts of Interest
The authors declare no competing financial interests.