CD24 expression does not affect dopamine neuronal survival in a mouse model of Parkinson's disease

Parkinson’s disease (PD) is a progressive neurodegenerative condition that is characterised by the loss of specific populations of neurons in the brain. The mechanisms underlying this selective cell death are unknown but by using laser capture microdissection, the glycoprotein, CD24 has been identified as a potential marker of the populations of cells that are affected in PD. Using in situ hybridization and immunohistochemistry on sections of mouse brain, we confirmed that CD24 is robustly expressed by many of these subsets of cells. To determine if CD24 may have a functional role in PD, we modelled the dopamine cell loss of PD in Cd24 mutant mice using striatal delivery of the neurotoxin 6-OHDA. We found that Cd24 mutant mice have an anatomically normal dopamine system and that this glycoprotein does not modulate the lesion effects of 6-OHDA delivered into the striatum. We then undertook in situ hybridization studies on sections of human brain and found—as in the mouse brain—that CD24 is expressed by many of the subsets of the cells that are vulnerable in PD, but not those of the midbrain dopamine system. Finally, we sought to determine if CD24 is required for the neuroprotective effect of Glial cell-derived neurotrophic factor (GDNF) on the dopaminergic nigrostriatal pathway. Our results indicate that in the absence of CD24, there is a reduction in the protective effects of GDNF on the dopaminergic fibres in the striatum, but no difference in the survival of the cell bodies in the midbrain. While we found no obvious role for CD24 in the normal development and maintenance of the dopaminergic nigrostriatal system in mice, it may have a role in mediating the neuroprotective aspects of GDNF in this system.


Introduction
The primary clinical features of Parkinson's disease (PD)-classically defined as a resting tremor, bradykinesia, postural instability and rigidity [1]-are associated with a >60% loss of dopamine (DA) neurons in the substantia nigra pars compacta (SNpc) of the midbrain [2]. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 The severe cell loss observed in the SNpc is in contrast, however, with that seen in the adjacent midbrain DA populations: the ventral tegmental area (VTA) which has only a 40% reduction in DA cells [2,3] and the retrorubral field (RRF) which exhibits little or no DA cell loss [2,4]. This raises questions as to why the SNpc population of DA neurons is more vulnerable to the PD disease process.
This selective cell loss is though not restricted to the midbrain DA populations. Post-mortem analysis of brains from early stage PD patients has suggested that the loss of the DA neurons in the SNpc is preceded by cell death in specific nuclei further down the brainstem, in particular the dorsal motor nucleus of the vagal nerve (DmnV), the raphe nucleus (RN), the pedunculopontine nuclei (PPN), and the locus coeruleus (LC) [5]. Remarkably, the nuclei neighbouring these structures remain significantly less affected, again raising questions regarding the selective cell loss associated with PD.
Given all this, one can hypothesize that the susceptible cells may share certain common features that makes them vulnerable to the disease, and one such factor that we identified was CD24 (cluster of differentiation 24; also known as Cd24a, M1/69-J11D heat-stable antigen (HSA), Lymphocyte antigen 52 (Ly-52), or Nectadrin). CD24 is a glycosyl phosphatidylinositol (GPI) anchored sialoglycoprotein that is expressed in a variety of tissues [6], and when first cloned, was associated with T-cell co-stimulation [7]. In lymphoid organs, this feature of Cd24 is redundant, only becoming essential in the absence of Cd28 [8,9]. In "co-stimulator ligand rich" organs, such as the central nervous system (CNS), however, the role of Cd24 is more critical. Cd24 deficiency in mouse models of experimental autoimmune encephalomyelitis (EAE) prevents T-cell proliferation and persistence [10], while conditionally over-expressing the gene exaggerates the response [11]. It has also been reported that, through an interaction with SiglecG (in mice) or Siglec10 (in humans), CD24 selectively represses the host response to injury, discriminating danger-associated molecular patterns (DAMPs) from pathogen-associated molecular patterns (PAMPs) [12].
Laser capture microdissection gene expression studies have demonstrated that Cd24 is expressed in the SNpc, but not the VTA [13][14][15] as well as the DmnV [16] and the LC [17]. Given the shared expression of Cd24 amongst the nuclei specifically affected in PD, and the direct interaction of Cd24 with the immune system, we sought to determine whether this protein could have a role in the cell loss seen in PD.
We found that CD24 is expressed in both the mouse and human brain in many of the nuclei affected pathologically in PD, but curiously not the SNpc in the human brain. Furthermore, we found using a classical mouse model of PD that while the absence of Cd24 did not impact on the long-term survival of the DA system, it did affect the neuroprotective effects of Glial cell-derived neurotrophic factor (GDNF) at the level of the striatal dopaminergic fibres. This result may be particularly relevant given the absence of CD24 in the SNpc neurons in the human adult midbrain.

Materials and methods
Mice and post-mortem human tissue Cd24 knockout mice bred on a C57/Black background were used in this study [18]. The wildtype (Cd24+/+) and homozygote (Cd24-/-) littermates were housed with unrestricted access to food and water. All animal procedures were performed in accordance with Cambridge University animal care committee regulations. Experiments were performed under the Home Office licences PPL 80/2366 (expired April 2016) or 70/8411 (valid until 2020).
Samples of human tissue were obtained from the Parkinson's UK Tissue Bank, funded by Parkinson's UK, a charity registered in England and Wales (258197) and in Scotland Behavioural testing Cylinder test. The cylinder test was conducted as previously described with rats [19,20]. Briefly, the mice were placed in a glass cylinder (13cm wide; 18 cm high) with two vertical mirrors positioned behind it, allowing for the paws of each mouse to be visualised at all times. Recordings were made with a Logitech c170 webcam, commencing as soon as the mouse was placed in the cylinder, and were analysed by an experimentally blind investigator (before genotyping). Six recordings lasting 3 minutes each were made over the last nine days before the mice were sacrificed. The first 20 wall contacts were counted in each session, and the frequency of ipsilateral (same side as the lesion) versus contralateral paw contacts was determined. Full extension of the digits of each paw was required for the contact to be counted.
Amphetamine induced rotations. Amphetamine induced rotations were induced on the day before the mice were to be sacrificed. The mice were injected intraperitoneally with 2.5mg/kg of amphetamine and placed in glass cylinders (13cm wide; 18 cm high). The number of rotations over 50 minutes was recorded on a Logitech c170 webcam and counted by an experimentally blind investigator. The number of full rotations was recorded for 1 minute every 5 minutes, and expressed as the net number of ipsilateral turns/min.
Open field locomotion test. The open field locomotion measure was conducted on the day before sacrifice. It involved a 10-minute recording session in which the mice were placed in an arena (70cm x 55cm x 22cm), with a Logitech c170 webcam recording the movement of the mice from above the enclosure. Footage of the test session was analysed using Icy "Mice Profiler Tracking" software (http://icy.bioimageanalysis.org/) [21,22].

Stereotaxic injections
The mice were anesthetized with isoflurane, placed in a stereotaxic frame, and were unilaterally lesioned in the striatum (co-ordinates: AP: +1.5; ML: +0.21; DV: 3.0). A small hole was drilled in the exposed skull and a Hamilton syringe was used for the delivery of the 6-OHDA (5μg/μl; 1.5μl in total, at a rate of 0.5μl per minute). The syringe was left in place for 5 minutes before being slowly removed. The mice were then sutured and allowed to recover until either 12 or 70 days post-surgery. Special husbandry was provided post-surgical procedures, with mice being weighed daily and provided with subcutaneous injections of warm saline (at least 0.3ml/day), mashed pellets, and chocolate balls (Supreme Mini Treats from Datesand) to prevent weight loss.

Tissue processing
Mouse. The mice used in this study were euthanized with a 0.3 ml intraperitoneal injection of Euthatal (pentobarbitone sodium, 200 mg/ml; Merial, UK). They were then perfused transcardially with either 0.9% saline followed by ice-cold 4% paraformaldehyde (0.1 M phosphate buffer, pH 7.4) or formalin solution (10%; Sigma Aldrich). Brains were removed and post-fixed in their respective solutions overnight. The PFA fixed brains were then shifted to 30% sucrose, while the formalin fixed brains were washed in PBS and switched to 70% ethanol for paraffin embedding.
Human. Post-mortem tissues obtained from both the Parkinson's UK brain bank at Imperial College (London) and the Addenbrooke's Brain Bank, were sectioned at source and provided as either fresh frozen sections (12μm thick) or paraffin embedded sections (6μm).
Immunofluorescence was performed using the antibodies mentioned above, in addition to other commercially available antibodies

In situ hybridization
In situ hybridization analysis was conducted on sections of mouse and human brain using the RNAscope Target

Stereological estimations
To quantify the number of TH+ cells in the substantia nigra randomised counts were performed using a standard stereological method (Olympus CAST Grid System) [24]. This analysis was conducted as previously described [23]. Briefly, the substantia nigra was defined by a ventral border passing from the lateral extremity of TH+ cells, down along the dorsal border of the cerebral peduncle to the medial tip of this structure. A vertical line was passed through the medial tip of the cerebral peduncle and represented the medial border of the substantia nigra. The dorsal border of the area of interest extended dorso-laterally along the upper limits of the TH+ cell bodies in the SNpc. On sections where the medial lemniscus was present, TH + cells superior to this structure were not counted-thus excluding the TH+ cells of the VTA.
For all brains, five sections of substantia nigra (at approximately 200μm intervals, between −2.5 and −3.8 relative to Bregma) were counted using a 20x lens. The intact/unlesioned side of the brain was counted with a grid size that was 180x180μm, with the counting frame measuring 80x80μm, while the lesioned side of the brain required a grid size that was 150x150μm, with the counting frame measuring 100x100μm. At least 200 cells were counted per side of the brain and estimates of the total number of neurons were calculated according to the optical fractionator formula, and only a coefficient of error of <0.10 was accepted [25].

Optical density analysis
All of the DAB stained slides in this study were imaged using the scanscope XT slide scanner (Aperio), at a resolution of 0.5 microns per pixel, using a x20 objective. To quantify the fibre density in the striatum and Substantia nigra pars reticulata (SNpr), the mean optical intensity was measured from the TH-positive stained sections [26,27]. For the striatum, the measurements were conducted on 12 coronal levels at intervals of approximately 210μm, corresponding to sections of the structure from between +1.7 and -1.4 relative to bregma. The area of interest did not include the nucleus accumbens and globus pallidus. Fibre density analysis on TH+ fibres of the SNpr was conducted on only 2 coronal sections per brain (210 μm apart; from approximately -2.7 to -3.2 according to Bregma) in the rostral SNpc as only TH+ fibres (and no TH+ cell bodies) are present there. Nonspecific background was determined by taking measurements from the TH-negative corpus callosum and normalizing it to the white light background surrounding the section on the glass slide. This analysis was conducted using Ima-geJ (version 1.42q for Mac OSX, from the National Institutes of Health; http://rsb.info.nih. gov/ij/).

Statistical analysis
All statistical analyses were performed using Prism (version 5.0; Graphpad Software, San Diego, CA, USA). The number of animals and statistical tests used in each experiment is provided in the Results section and in each figure legend (Mann-Whitney (M-W) or ANOVA). For optical density experiments, comparisons were made between the lesioned and intact side of the brain to avoid between-section differences in antibody staining. For all of the stereological estimations of TH+ cells in the midbrain, the lesioned side of the brain was compared with the intact side. Differences were considered significant when P < 0.05 ( Ã P < 0.05; ÃÃ P < 0.005; ÃÃÃ P < 0.0005). Percentages are presented with the standard error (SE) in both the text and figures.

Results
The expression of Cd24 transcripts in the adult mouse brain Using in situ hybridization, we examined Cd24 transcript expression in the adult mouse brain, confirming and extending those that have been previously reported (Allen Institute for Brain Science. Allen Mouse Brain Atlas [Internet]. Available from: http://mouse.brain-map.org/ experiment/show/79591541). Here we shall focus exclusively on those nuclei that are majorly affected in PD, but a full description of Cd24 expression in the adult mouse brain can be found in S1 Fig. When considering the nuclei preferentially affected in PD [28], we find that Cd24 expression in the adult mouse brain overlaps with many of these structures. In the forebrain, robust Cd24 transcript expression is present in the olfactory bulbs (S1A and S1B Fig), though not the anterior olfactory nucleus [29]. We also found strong expression in the amygdala ( Fig 1A)  [30], but not in the nucleus basalis of Meynert although it was present at high levels in the reticular nucleus of the thalamus (S1I Fig).
In the mesencephalon, Cd24 transcripts are present in the DA neurons of the SNpc ( Fig  1B), as confirmed using double immunofluorescent staining for tyrosine hydroxylase (THthe rate limiting enzyme involved in DA production; Fig 1C). Quantifications of TH+ cell bodies with more than two red separate aggregations of signal amplification inside the cytoplasm indicated that 60.8%±3.1 of the cell bodies in the SNpc are TH+/Cd24+, 31.2%±3.7 are TH +/Cd24-and 8.0%±1.2 are TH-/Cd24+ (n = 3 brains). Double staining with astrocyte (GFAP) and microglial (CD11b) markers found no/little co-localisation in the SNpc and VTA regions (data not shown). Cd24 expression in the SNpc is more robust in the rostral portion than more caudal aspects (compare Fig 1D and 1C), and in agreement with others we found very little expression of Cd24 in the VTA when compared to the SNpc [14] (Fig 1E and 1F). This suggests that Cd24 is a gene that is preferentially expressed by the SNpc DA neurons, which are classically recognised as one of the most affected subsets of cells in the PD brain [2,28].
In the hindbrain, we detected strong labelling of Cd24 in the LC and parabrachial nucleus (PBl; Fig 1G) [31], and sparse expression in the dorsal raphe nuclei (RN) and Pedunculopontine nuclei (PPN; S1K Fig) [32] as well as the DmnV (Fig 1H). While numerous nuclei in the adult mouse brain express Cd24, we considered it noteworthy that many of them (such as the amygdala, SNpc, RN, PPN, PBl, LC, and DmnV) are sites of major pathology in PD [28], while others low in expression (e.g. VTA) are relatively well preserved in this condition.

CD24 protein expression in the mouse brain
We followed up the in situ analysis by looking at the protein expression of CD24 in the adult mouse brain. Again, we have focused on the nuclei affected in PD, although a comprehensive representation of CD24 protein localisation is provided in S2 Fig. As previously reported [33], robust CD24 expression was observed in the olfactory bulb (especially the glomerular layer and rostral migratory stream; S2A Fig). We also found a dorsal-ventral gradient of CD24 expression in the rostral striatum (Fig 2A), and strong CD24 expression in the amygdala (S2D  (Fig 2C and 2D), indicative of a membrane bound protein.
In the hindbrain, we observed very little/no CD24 protein in the raphe nucleus (S2H Fig), and moderate CD24 expression in the cell bodies of the LC (Fig 2E). In the medulla, we found CD24 protein in the DmnV (data not shown). Importantly, while again highlighting an expression profile that strongly overlaps with the sites of PD pathology, the presence of CD24 on the neurites of the DA neurons in the SNpc led us to investigate further the possible function of CD24 in these cells.

The dopamine system of the Cd24 mutant mouse is anatomically normal
To determine whether Cd24 has a role in the development or maintenance of the midbrain DA neurons, we analysed the brains of mice that had a targeted deletion of the gene [18]. Immunohistochemistry for CD24 on sections of brain tissue derived from 8week old Cd24-/mice exhibited no labelling, demonstrating that CD24 protein was not present in the brains of these mice (S2I- S2L Fig). Further, TH staining on those same sections indicated no observable difference in the DA system between Cd24-/-mice and Cd24+/+ littermates. In the olfactory bulb, the dopaminergic interneurons in the glomerular layer were in the correct location. The TH+ fibre innervation of the striatum appeared unaltered by the mutation, and stereological  (Fig 3D and 3D') and the morphology of the TH+ cells in the SNpc of Cd24-/-mice (Fig 3E-3F'), when compared to their Cd24+/+ littermates. These mutant mice also displayed the correct expression patterns for SNpc-related genes such as G protein-activated inward rectifier potassium channel 2 (Girk2), and aldehyde dehydrogenase 1 family, member A1 (Aldh1a1; data not shown) [34][35][36]. Together these findings suggested that Cd24 does not play a critical role in the normal anatomical development and maintenance of the murine midbrain DA system. Whether it has a more subtle role in the synaptic release of DA within the striatum was not explored in this study, but may be worth investigating given the putative role of this molecule at the neuromuscular junction [37].
CD24 has no effect on striatal 6-OHDA lesions in mouse models of PD Several reports have highlighted the importance of T-cells in the neurodegeneration associated with neurotoxic and other relevant animal models of PD [38,39]. Given that Cd24 deficient mice exhibit reduced T-cell proliferation and persistence in models of EAE [10] and conditionally over-expressing the gene exaggerates the phenotype [11], we hypothesized that deletion of the gene may lessen the effect of the neurotoxin 6-OHDA on the long term survival of midbrain DA neurons. Blocking CD24 with antibodies in mice has been shown to not be possible due to toxicity issues ( [40] and Prof Peter Altevogt, personal communication), and so we modelled PD in the Cd24-/-mouse using striatal delivery of 6-OHDA.
The delivery of the neurotoxin 6-OHDA to the midbrain nigrostriatal pathway is commonly used to model the DA pathology of PD. Injecting 6-OHDA into the medial forebrain bundle or the SNpc results in a very rapid loss of TH+ cells [41,42], which does not allow for a thorough analysis of the molecular events involved in the loss of cells within the midbrain. To circumvent this, we have previously demonstrated that 6-OHDA delivery to the striatum results in a slower rate of midbrain TH+ cell loss, beginning between 9-12 days post lesion, coinciding with local glial activation [23]. Thus, we chose to adopt this approach given also the possible role of Cd24 in modulating neuroinflammatory processes to injury. This also explains why we examined the brains of the Cd24-/-and Cd24+/+ mice over this period of cell loss following surgery.
At 12 days post-lesion, we found no difference between genotypes in the fibre density of TH+ axonal arborisations in the striatum (33.   Fig  4E).
To determine whether the absence of CD24 had any long-term consequences on the survival of the DA neurons, we next examined the 6-OHDA-striatal lesioned mice 70 days postlesion-a time point when the pathological changes linked to the lesion are known to be complete [43]. Before culling the mice, we performed behavioural testing which included the cylinder test, amphetamine induced rotations, and open field locomotion. We found no difference between the mice in any of these tests (Cylinder test: affected forelimb usage was 64.3±8.

CD24 is involved in the neuroprotective effect of GDNF
Given that the absence of Cd24 had no impact on the overall survival of SNpc DA neurons in the 6-OHDA model of PD, we next sought to determine whether Cd24 may have more of a role in mediating the neuroprotective effects of GDNF. Christophersen et al [44] reported a nine fold increase in Cd24 expression in the substantia nigra following lentiviral delivery of GDNF into the striatum. Thus, we reasoned that Cd24 may be involved in some aspect of the neuroprotective effects of GDNF. To test this hypothesis, we delivered AAV-GDNF into the striatum of both WT and Cd24-/-mice.
One-month post-surgery, robust GDNF expression was observed in the striatum of both groups of mice (optical density analysis indicated a 440.4±56.9% increase in GDNF staining in the injected striatum compared to the uninjected side of the Cd24-/-mice versus with 428.7 ±75.7% in the injected Cd24+/+ mice; M-W U = 12, p = 0.94; S4A Fig). Both sets of mice also exhibited a reduction in body weight in accordance with previous reports [45], suggesting that GDNF functions normally in Cd24-/-mice. Increased levels of CD24 were present in the SNpc of the transfected side of the brain in the Cd24+/+ mice, but absent (as one would expect) in the Cd24-/-mice (S4B and S4B' Fig). Two months post-AAV-GDNF transfection, both groups of AAV-GDNF injected mice (and a control group of WT mice; n = 5 in each group) received a striatal delivery of 6-OHDA on the same side of the brain. By 21 days post lesion, the loss of TH+ fibres in the striatum has plateaued and remains stable. When we compared the 21 day time point data with that of 70 days post lesion (see above), we found TH+ fibre loss of 39.8%±7.2 of intact side and 38.2%  At three weeks post-6-OHDA lesion, we found no difference in the reduction of SNpc DA neurons on the lesioned side of the brain when comparing the two groups of AAV-GDNF injected mice (93.5±5.0% of the unlesioned side in the Cd24-/-mice compared with 96.2 ±3.2% in the Cd24+/+ mice; M-W U = 10, p = 0.66; Fig 6A-6E). In the group of WT mice that received striatal delivery of 6-OHDA, but no AAV-GDNF, only 36.6±2.9% of the SNpc TH + neurons remained compared to the intact side (Fig 6E). These results suggest that CD24 is not involved in the neuroprotective effects of GDNF at the cell body level. However, when we conducted optical density analysis of the TH+ fibres in the striatum, we did observe a significant difference between the two groups of mice that received AAV-GDNF (53.  (Fig 6K and 6L), indicating that the effect was not specific for TH. This result indicates that CD24 plays a role in the GDNF-induced neuroprotection of DA neurons at the level of the striatal neurites, which also fits with our earlier expression data on CD24.

CD24 expression in the human brain
Finally, in order to compare these mouse results with the human situation, we looked at CD24 expression in the human brain. Attempts to label CD24 transcripts on post-mortem brain tissue using conventional in situ hybridization procedures proved very difficult. DIG-labelled in situ probes provided little if any signal above background (including the use of previously published protocols [46]; on both paraffin embedded and fresh frozen tissue from 5 brains; average post-mortem interval = 22hrs; range = 6-47 hours). To resolve this situation, we utilised a DNA-branching protocol (RNAscope, Advanced Cell Diagnostics). Using this approach on fresh frozen, PFA-fixed sections (from 5 normal brains and 5 PD brains supplied by the Parkinson's UK Brain Bank; Table 1), we successfully investigated CD24 expression across 9 regions of the brain: medulla, pons, midbrain, amygdala, thalamus, subventricular zone (SVZ), hippocampus and frontal cortex.
Control and Parkinson's disease (PD) brains used in the present study, with their age of death, sex, post-mortem interval (PMI), and pathological features (based on the pathologist's notes). For the PD brains, the duration of disease is also provided. The brains supplied by the Parkinson's UK Brain bank are indicated in the grey boxes.
Overall, we found less wide spread expression of CD24 when compared to the mouse brain. There was also little difference in transcript expression between normal and PD brains, except in cases where reduced cell numbers in particular nuclei resulted in less labelling. In the brain stem, we observed strong CD24 expression in the large cell bodies of the DmnV (Fig 7A and  7B) and the hypoglossal nucleus of the medulla, with the transcripts being found in the nucleus, cytoplasm and extending into the neurites. CD24 was also expressed by large cell bodies in the nucleus ambiguous (Fig 7A and 7C) but we were unable to localise the LC on any of the samples of the pons. In the midbrain, the lining of the ventricles robustly expresses CD24, but the neuromelanin+ cells of the VTA and SNpc exhibited no CD24 expression. A very small fraction of the medially-located neuromelanin+ cells per section were very weakly positive for CD24 (Fig 7D and 7E), but the more laterally positioned SNpc neuromelanin+ cells were completely void of CD24 transcripts (Fig 7D and 7F). We supplemented the number of midbrain sections with tissue collected from the Cambridge Brain Bank (6 controls and 2 PD; Table 1), but we found only one of the 11 control brains and none of the 7 PD brains exhibited CD24 expression in the SNpc neuromelanin+ cells (control brain CT03 on Table 1). Although this brain exhibited no disease-related pathology in the midbrain, there was unexplained sclerosis in the hippocampus, indicating the possibility of some additional complication in this case that may have affected the staining.
Similar to the mouse brain, we found very strong CD24 transcript signal present in all areas of the SVZ examined (Fig 7G). In the hippocampus, we found robust CD24 transcripts in the dentate gyrus, CA2 and CA3 regions, but not CA1 or CA4 (Fig 7H). The amygdala exhibited weak CD24 expression (Fig 7I), with transcripts also found in layer 3 of the neighbouring periamygdaloid cortex. In the cortex, we found weak CD24 expression in layers 1-3, and little or no expression in the lower levels (data not shown). Collectively, these results indicate that CD24 expression in the human resembles the mouse in many areas of the brain, but importantly it appears to be absent in the neuromelanin+ dopamine neurons of the midbrain.

Discussion
Understanding the mechanisms involved in the selective cell death associated with PD would represent a major step forward in better understanding and possibly treating this condition. We hypothesized that the nuclei preferentially affected in PD may share a common feature that renders them vulnerable to the causative agent/conditions. Given the specific expression and interactions of CD24 in mice, we considered this gene to be worthy of further investigation.  Our initial expression analysis confirmed and extended previous reports of Cd24 expression in the rodent brain ( [33]; Allen Institute for Brain Science. Allen Mouse Brain Atlas [Internet]. Available from: http://mouse.brain-map.org/experiment/show/79591541) and also provides unique data on the expression of CD24 in the human CNS. Similarities in expression localisation across species were of interest, especially in evolutionarily older structures such as the brainstem (e.g. the DmnV and the hypoglossal nucleus). The absence of CD24 expression in the human SNpc, however, was particularly striking, although cross-species differences in gene expression between mice and humans in midbrain DA neurons has been previously reported. Orthodenticle Homeobox 2 (OTX2), for example, is a transcription factor that is expressed by midbrain DA progenitors during development in both mice and humans, and defines the VTA subpopulation in the mouse adult brain, but it is not expressed in any midbrain DA neurons in aged humans [47]. Of the 18 human midbrains analysed in the present study, only one had any CD24 expression in neuromelanin+ cells, and those cells resided in the more medially located VTA population. As indicated in the results section, that particular case had other pathologies which complicates the interpretation of this finding. Despite this absence of CD24 in the human adult midbrain DA neurons, we were still interested in assessing possible functions of CD24 in the mouse SNpc DA neurons.
The absence of a phenotype in the midbrain DA system of the Cd24-/-mouse-out to 2 years of age-indicates that the gene is not required for the normal anatomical development or maintenance of this population of neurons. While this lack of phenotype was useful for subsequent disease modelling experiments, our analysis could not cover all aspects of DA neuronal function (such as dopamine release, etc). It has been reported that Cd24 inhibits the extension and collateral branching of neurites through its interactions with neural recognition molecule L1 [48,49], thus in the absence of the gene it could be hypothesized that neurons expressing Cd24 in the adult brain may exhibit more neurites. Given the density of normal SNpc TH+ axonal arborisations, any increase in branching was difficult to analyse in the Cd24-/-mouse and further microscopic analysis may reveal some subtle specific differences. Cd24 also has a role in synaptic transmission [37]. In the absence of Cd24, neuromuscular junctions exhibited synaptic depression, cyclical periods of complete transmission failure, and a reduction in the pool of recycling vesicles in the terminals [37]. Thus, although it was not investigated in the current study, looking at synaptic transmission in the SNpc DA neurons of Cd24-/-mice would be of interest.
The modelling of PD (using striatally delivered 6-OHDA) in the Cd24-/-mouse showed that the absence of the gene had no impact on the survival of the SNpc DA neurons. However, while CD24 does not exhibit direct neuroprotection, it may be important in mediating the neuroprotective effects of other agents such as GDNF. Thus, we sought to investigate this using AAV-GDNF transfection and 6-OHDA lesions in Cd24-/-and wild-type mice and found that there was a greater TH+ fibre density loss in the striatum of the AAV-GDNF/ 6-OHDA Cd24-/-mice. Previously we have demonstrated that the fibre loss following striatal delivery of 6-OHDA in the mouse is very rapid (>60% of eventual fibre loss occurring within the first 24 hours), and that from 3 days post-lesion the extent of fibre loss remains constant [23]. In the current study, while AAV-GDNF preserved the DA cell bodies in the SNpc of the Cd24-/-mice, it was not able to rescue the density of striatal TH+ fibres at three weeks postlesion. The oxidation and washout of 6-OHDA in the brain within the first 24 hours would render a delayed/slowed loss of TH+ fibres unlikely. One confounding aspect of this result, however, is why the absence of CD24 does not result in a similar loss of TH+ fibres to that observed in the 6-OHDA lesioned control mice. This suggests that additional GDNF-related pathways may be involved, resulting in only a partial reduction in the neuroprotective effect of GDNF at the neurite level. This effect, though, does raise the possibility that CD24 may be playing a role in normal DA neurite outgrowth-an area we are now investigating further in vitro.
Many Tgf-β family members are known to up-regulate the expression of CD24. Lentiviral delivery of GDNF into the striatum resulted in an increase in Cd24 expression in the rodent SNpc [44], and increased Cd24 promoter activity and Cd24 mRNA results from forced expression of Artemin [50]. Recently, a reciprocal relationship between CD24 and TGF-β3 in bone marrow-derived stromal cells has been reported [51]. Given the lack of CD24 expression in the human DA neurons and the reduction in TH+ fibre density in the AAV-GDNF/6-OHDA in the present study, it is tempting to speculate on the role of CD24 and the likely success of clinical trials involving the delivery GDNF to the striatum in people with PD.
As to a specific role CD24 may be playing in PD, it is difficult to speculate. While CD24 has not been associated with PD, polymorphisms in this gene does result in increased risk and progression of several autoimmune diseases, including multiple sclerosis [52,53], Crohn's disease [54], and rheumatoid arthritis [55]. This may be relevant to PD given the recent interest in the possibility that PD may have a primary neuroinflammatory component. In addition while there is a lack of CD24 expression in the DA neurons of the midbrain in humans, the selective expression in other vulnerable regions of the PD brain (particularly in the lower brain stem) suggests that this is a gene worthy of further investigation as it may have some indirect role in the disease process.
In conclusion, we have shown that CD24 has some commonality of expression with the widespread distribution of pathology seen in PD, although we have been unable to show any causal links between this glycoprotein and the loss of dopamine cells in the mouse model of PD. Nevertheless, we have shown some association between the effects of a known neurotrophic factor for dopaminergic nigral neurons, GDNF, and CD24 expression. This merits further exploration not only as this agent is in clinical trials in patients with PD but also because the benefits of GDNF have been shown to critically depend on other pathways within the striatum [56].