Relationship between microglial activation and dopaminergic neuronal loss in the substantia nigra: a time course study in a 6-hydroxydopamine model of Parkinson’s disease

Authors

  • Lilia Marinova-Mutafchieva,

    1. Parkinson’s Disease Research Group, Department of Cellular and Molecular Neuroscience, Division of Neuroscience and Mental Health, Faculty of Medicine, Imperial College London, London, UK
    Search for more papers by this author
  • Mona Sadeghian,

    1. Parkinson’s Disease Research Group, Department of Cellular and Molecular Neuroscience, Division of Neuroscience and Mental Health, Faculty of Medicine, Imperial College London, London, UK
    Search for more papers by this author
  • Lauren Broom,

    1. Parkinson’s Disease Research Group, Department of Cellular and Molecular Neuroscience, Division of Neuroscience and Mental Health, Faculty of Medicine, Imperial College London, London, UK
    Search for more papers by this author
  • John B. Davis,

    1. Neuroscience Centre of Excellence for Drug Discovery, GlaxoSmithKline, Harlow, UK
    Search for more papers by this author
  • Andrew D. Medhurst,

    1. Neuroscience Centre of Excellence for Drug Discovery, GlaxoSmithKline, Harlow, UK
    Search for more papers by this author
  • David T. Dexter

    1. Parkinson’s Disease Research Group, Department of Cellular and Molecular Neuroscience, Division of Neuroscience and Mental Health, Faculty of Medicine, Imperial College London, London, UK
    Search for more papers by this author

Address correspondence and reprint requests to Dr D. T. Dexter, Department Cellular and Molecular Neuroscience, Faculty of Medicine, Imperial College London, Hammersmith Hospital Campus, Du Cane Road, London W12 ONN, UK. E-mail: d.dexter@imperial.ac.uk

Abstract

Cellular interactions between activated microglia and degenerating neurons in in vivo models of Parkinson’s disease are not well defined. This time course study assesses the dynamics of morphological and immunophenotypic properties of activated microglia in a 6-hydroxydopamine (6-OHDA) model of Parkinson’s disease. Neurodegeneration in the substantia nigra pars compacta (SNc) was induced by unilateral injection of 6-OHDA into the medial forebrain bundle. Activated microglia, identified using monoclonal antibodies: clone of antibody that detects major histocompatibility complex (MHC) class II antigens (OX6) for MHC class II, clone of antibody that detects cell surface antigen-cluster of differentiation 11b – anti-complement receptor 3, a marker for complement receptor 3 and CD 68 for phagocytic activity. Activation of microglia in the lesioned SNc was rapid with cells possessing amoeboid or ramified morphology appeared on day 1, whilst antibody clone that detects macrophage-myeloid associated antigen immunoreactivity was observed at day 3 post-lesion when there was no apparent loss of tyrosine hydroxylase (TH)+ve dopaminergic (DA) SNc neurons. Thereafter, OX6 and antibody clone that detects macrophage-myeloid associated antigen activated microglia selectively adhered to degenerating axons, dendrites and apoptotic (caspase 3+ve) DA neurons in the SNc were observed at day 7. This was followed by progressive loss of TH+ve SNc neurons, with the peak of TH+ve cell loss (51%) being observed at day 9. This study suggests that activation of microglia precedes DA neuronal cell loss and neurons undergoing degeneration may be phagocytosed prematurely by phagocytic microglia.

Abbreviations used
6-OHDA

6-hydroxydopamine

CD68

antibody clone that detects macrophage-myeloid associated antigen

DA

dopaminergic/dopamine

LPS

lipopolysaccharide

mfb

medial forebrain bundle

MHC

major histocompatibility complex

NeuN

neuronal nuclei

OX42

clone of antibody that detects cell surface antigen-cluster of differentiation 11b – anti-complement receptor 3

OX6

clone of antibody that detects MHC class II antigens

PBS

phosphate-buffered saline

PD

Parkinson’s disease

SNc

substantia nigra pars compacta

TH

tyrosine hydroxylase

VTA

ventral tegmental area

Parkinson’s disease (PD) is an age-related neurodegenerative disease with a primary pathology characterized by a progressive degeneration of the dopaminergic (DA) neurons of the substantia nigra pars compacta (SNc). Recent evidence has implicated inflammation and microglial activation in the initiation and progression of PD (Qin et al. 2007). Some cases of PD are associated with head trauma or certain viruses or infections, suggesting that microglial activation caused by preceding brain inflammation may trigger the onset of a cascade of events leading to progressive degeneration of DA neurons (Liu et al. 2003; Herrera et al. 2005). Moreover, several studies have demonstrated that neuronal cell loss in the SNc in PD is accompanied by a marked proliferation of reactive microglial cells (Forno 1996; Knott et al. 2000; Mirza et al. 2000). Microglia are reported to be locally activated not only in postmortem idiopathic PD brains (McGeer et al. 1988) but also in MPTP Parkinsonism in humans (Langston et al. 1999). Activation of microglia also has been identified in the SNc and/or striatum in animal models of PD, such as MPTP, lipopolysaccharide (LPS) or medial forebrain bundle (mfb) axotomy models (Kurkowska-Jastrzebska et al. 1999;Cho et al. 2003, 2006; Sugama et al. 2003). In vitro studies demonstrate that the toxicity of LPS to immortalised DA neurons was only observed when microglia were also present in cell culture (Gao et al. 2002) suggesting that LPS is not toxic but activated microglia can be potentially damaging to DA neurons. Additionally, in mfb axotomised brain, it has been demonstrated that activation of microglia preceded neuronal loss (Gao et al. 2002). Similarly, 48 h after striatal 6-hydroxydopamine (6-OHDA) administration, Rodriguez-Pallares et al. (2007) demonstrated significant microglial activation and production of NADPH-derived free radicals prior to DA cell death in the SNc. These findings strongly suggest that activated microglia play an important role in the early stage of disease pathogenesis. Moreover, DA neurons in the SNc may be more vulnerable to microglial-derived neurotoxic factors as the SNc region is particularly rich in microglia (Kim et al. 2000; Polazzi and Contestabile 2002; Liu et al. 2003; McGeer et al. 2005).

Microglial activation is not always deleterious (Glezer et al. 2006; Neumann et al. 2006; Simard and Rivest 2006; Sriram et al. 2006) with microglia playing an important role in phagocytic removal of cell debris and dead neurons. Observations from animal models of PD and cell culture studies have provided rather compelling evidence on the functional role of microglia-derived cytotoxic and neurotrophic factors. In contrast however, limited information is available on the cellular interactions between activated microglia and degenerating neurons in the in vivo models of neurodegenerative disorders.

In this study, we have induced neurodegeneration in the SNc by unilateral injection of 6-OHDA into mfb. For the first time, an immunohistochemical approach has been used to describe the kinetics of immunophenotypic changes of activated microglia and the subsequent fate of activated microglial cells during the initiation and progression of DA neuronal loss in the SNc and loss of striatal DA. Changes in the functional activity of activated microglia were identified using different monoclonal antibodies: OX6 (recognises MHC class II antigens), anti-cluster of differentiation 11b – anti-complement receptor 3 [clone MRC of antibody that detects cell surface antigen-cluster of differentiation 11b – anti-complement receptor 3 (OX42)] (Sugama et al. 2003) and antibody clone that detects macrophage-myeloid associated antigen, a lysosomal enzyme in phagocytic microglia (CD68) (Croisier et al. 2005).

Materials and methods

6-OHDA lesion model

Male Sprague–Dawley rats (Harlan, Bicester, UK) weighing 220–240 g, were housed in groups of three per cage with free access to food and water. They were housed under a controlled temperature (21 + 1°C), relative humidity (60%) and a 12 h light/dark cycle (light on at 8.00 am). All scientific procedures were carried out in accordance Home Office Animals (Scientific procedures) Act, UK, 1986. 6-hydroxydopamine hydrobromide (6-OHDA-12 μg free base in an injection volume of 4 μL) lesions into the left mfb at day 0 in isoflurane (Abbot, Maidenhead, UK) anaesthetised rats were induced as previously described by the authors (Datla et al. 2006) using coordinates of 2.2 mm posterior, 1.5 mm lateral from bregma and 7.9 mm ventral to the dura, according to Paxinos and Watson (1986).

Rats were killed by overdose of CO2 followed by cervical dislocation and decapitation at days 1, 3, 5, 7, 9, 13 and 15 after 6-OHDA injection. The brain was removed and transferred to ice immediately and cut at the level of the infundibular stem forming a hindbrain block containing the SN and a forebrain block containing the striatum. The striata were dissected, snap frozen and stored at −80°C until HPLC analysis for striatal DA concentrations. The hindbrain was transferred to 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS), pH 7.4 and kept for 7 days. Brain specimens were then immersed in 30% sucrose in PBS until sunken (3–5 days) and then frozen at −80°C. Sequential coronal sections (20 μm) of the hindbrain were cut using a cryostat (Bright Instruments, Cambridge, UK). Free-floating sections were collected in 0.1 M PBS (pH 7.4) containing 0.05% sodium azide as a preservative until immunostaining.

Immunohistochemical detection of TH, NeuN, OX42, OX6 and CD68

Dopaminergic neurons were detected with a rabbit polyclonal antibody against tyrosine hydroxylase (TH), the rate-limiting enzyme in DA synthesis, used at dilution of 1 : 1000 (Pel-Freez, Rogers, AZ, USA). The staining was performed on free-floating nigral sections. Briefly, sections were washed in (pH 7.4). Endogenous peroxidase activity was blocked by incubating the sections in 1% hydrogen peroxide/PBS for 30 min. After three washes in PBS, non-specific binding was blocked with 20% normal goat serum in 0.1 M PBS for 1 h. Without further washes sections were then incubated in anti-TH primary antibody in PBS containing 0.05% triton X-100 and 2% normal goat serum overnight at 18–20°C. A secondary biotinylated goat anti-rabbit IgG (1 : 200, Vector Laboratories, Peterborough, UK) was applied for 1 h at 18–20°C in PBS containing 0.05% triton X-100. This was followed by incubation with ABC Elite peroxidase staining kit (Vector Laboratories). The reaction was visualized using 3, 3′-diaminobenzidinne kit (Vector Laboratories). Sections were mounted onto poly-l-lysine coated slides, left to dry and dehydrated.

Microglial cells were immunostained with mouse monoclonal antibodies: OX42 (general marker for microglia not necessarily activated), while activated microglia was detected by staining for OX-6 and CD68 stains phagocytic microglia. The above antibodies were obtained from ABD-Serotec (Kidlington, UK) and applied in dilution 1 : 750. The protocol for staining of the microglial markers was similar to that used for TH staining. Minor changes were: normal horse serum was used to block non-specific binding and the secondary antibody was biotinylated anti-mouse (1 : 200, Vector Laboratories).

Double immunostaining

The co-expression of TH and the general neuronal marker neuronal nuclei (NeuN), was studied using double immunofluorescence with detection of TH (1 : 500) by fluorescent goat anti-rabbit IgG (1 : 300, Alexa 488, Invitrogen, Paisley, UK) and detection of anti-mouse NeuN (1 : 2000, Chemicon, Watford, UK) by incubation with goat anti-mouse IgG conjugated with Cy3 (1 : 300, Invitrogen).

To assess neuronal-microglial interaction, brain sections were stained for TH and OX6 or CD68. Primary antibodies were used at dilutions as followed: TH 1 : 500 and OX6 or CD68 were diluted 1 : 300. TH immunoreactivity was visualised with goat anti-rabbit IgG conjugated with Cy3 (1 : 300, Invitrogen) and OX6 or CD68 expression was monitored by fluorescent goat anti- mouse IgG Alexa 488 (Invitrogen). To assess the proliferative activity of OX6 expressing microglial cells mouse monoclonal OX-6 (ABC-Serotec) at the dilution of 1 : 200 was co-incubated with a rabbit polyclonal anti-Ki67 antibody at a dilution 1 : 300 (clone TEC-3; Dako, Ely, UK). Ki67 is a nuclear mitotic dependent protein that stains the proliferating cells at all steps of the cell cycles except for the resting phase. Ki67 was visualised using Alexa Fluor 488 goat anti-rabbit secondary antibody (1 : 300, Invitrogen) and OX-6 was visualised using Cy3 conjugate goat anti-mouse secondary antibody (1 : 300, Invitrogen). To assess the potential involvement of apoptosis in neuronal cell death in the SN, brain sections were stained for caspase 3 (cysteine-aspartic acid protease) which plays a key role in the execution of cellular apoptosis (1 : 500, Calbiochem, Nottingham, UK) and NeuN (1 : 2000, Chemicon). Caspase 3 was visualised using Alexa Fluor 488 goat anti-rabbit secondary antibody (1 : 300, Invitrogen) and NeuN was visualised using Cy3 conjugate goat anti-mouse secondary antibody (1 : 300, Invitrogen).

To assess the interaction between apoptotic cells with activated microglia, sections were stained for caspase 3 (1 : 500, Calbiochem) and/or CD68 or OX6 (1 : 300, Serotec). Caspase 3 expression was visualised using Alexa Fluor 488 goat anti-rabbit secondary antibody (1 : 300, Invitrogen) and CD68 or OX6 were visualised using Cy3 conjugate goat anti-mouse secondary antibody (1 : 300, Invitrogen).

All sections were also stained with 4′,6-diamidino-2-phenylindole (a fluorescent stains that binds strongly to nuclear DNA) for the visualisation of cellular nuclei (Sigma-Aldrich Ltd., Poole, UK) at the end of the immunostaining procedure. The slides were cover slipped with Vectasheild Hard Set mounting medium (Vector Laboratories) before being examined on a Nikon Eclipse E1000 fluorescence microscope (Nikon UK Ltd., Kingston upon Thames, UK) using appropriate filter sets. Appropriate positive and negative controls were included in all immunohistochemical reactions.

Image capture and quantitative cell counting

To quantify the number of TH+ve (+ve = immunopositive cells) DA neuronal bodies the counting was performed manually using a Nikon Eclipse E800 microscope (Nikon) connected to a JVC camera at 100× magnification. TH+ve neurons were counted in both ipsilateral and contralateral hemispheres of the brain in each of the stereotaxic regions of the SN and adjacent ventral tegmental area (VTA) area, (−4.80, −5.30, −5.60, −5.80 and −6.30 mm from bregma) (Carman et al. 1991) to ensure complete sampling of the entire structures; VTA and SNc. The stereotaxic region in each section was determined by examination of the shape and distribution of TH+ve neurons at low magnification (40×) with reference to the standard rat brain atlas of Paxinos and Watson (1986). The SNc was delineated from the VTA by reference to specific anatomical landmark of the third cranial nerve rootlet (Iravani et al. 2002). To minimise bias all operators were assessed for the consistency of counting in each stereotaxic level of the SNc prior to cell counting. Cells that were intact with a nucleus and clear cytoplasm and axonal processes were included in the counts and cells that appeared deformed with no obvious axonal processes were excluded from the counts (Iravani et al. 2002). For each animal six consecutive sections were counted at each stereotaxic level and these were pooled to give a total mean cell count for ipsilateral and contralateral hemispheres in each group. To ensure an unbiased cell count all operated were blinded to the group under analysis. A similar approach was utilized to count markers of microglial activation.

Analysis of striatal dopamine and metabolites using HPLC-ECD

Dopaminergic/dopamine content of striata was assessed using HPLC with electrochemical detection, as described previously by this group (Vernon et al. 2005). DA concentrations were quantified by an electrochemical analytical cell (model 5011; ESA Analytical Ltd, Aylesbury, UK) attached to a Coulochem II electrochemical detector (ESA Analytical) with electrode one set at −0.020 mV and electrode two set at +0.34 mV with respect to the pallidium reference electrode. All data analysis of the chromatograms was performed using PC-based Chromeleon software (Dionex, Camberley, UK).

Statistical analysis

The data are expressed as the mean ± SEM. Statistical significance was assessed by repeated measures two-way anova followed by post hoc Bonferroni Multiple Comparison tests, comparing the level of significance within and between groups. p < 0.05 was considered statistically significant.

Results

6-OHDA induced loss of dopaminergic neurons and local activation of microglia in the substantia nigra.

To discern the effects of 6-OHDA on neuronal survival in the SNc, we monitored the number of DA neurons, as indicated by the number of TH+ve cells within the SNc at different time points after toxin injection. Comparison of the number of nigral TH+ve neurons between lesioned and unlesioned sides showed no significant loss of TH+ve neurons within the first 7 days after 6-OHDA injection, although the neurons did appear shrunken. However, significant loss of TH+ve neurons started to manifest from day 9 after 6-OHDA injection with the cell loss being 50%, 55% and 58% at days 9, 13 and 15 respectively (Fig. 1). Double staining with TH and NeuN demonstrated that 6-OHDA induced neuronal loss represents neuronal death, rather than a down regulation of TH expression as evidenced by a decline in the total number of neurons in the SNc ipsilateral to the lesion (data not shown). Loss of TH+ve cells appeared specific to the SNc with no significant reduction being observed in the VTA (data not shown). In particular, neurons in the lateral portion and ventral tier of the SNc were most vulnerable. Interestingly, staining for microglia using OX42 antibody, indicated that in the SNc there was a significant increase in the number of ramified OX42+ve cells with time on the lesioned side compared the unlesioned side (F9.53; p < 0.001) and this started to manifest from day 1 after 6-OHDA lesioning (Fig. 2a, confidence intervals Table S1). A similar rapid activation of OX42 expressing cells was also observed in the adjacent VTA, reaching significance at day 5 post-6-OHDA lesioning (Fig. 2b) but this was not associated with DA neuronal cell loss. The greatest increase in activated OX42+ve microglial cells ipsilateral to the lesion in SNc (∼6-fold) and VTA (∼2.5-fold) was observed at day 15 when a significant loss of TH+ve DA neurons was observed in the SNc. Thus, the SNc may be more susceptible to toxic insult when compared to other DA regions.

Figure 1.

 Injection of 6-OHDA resulted in a time-dependent loss of dopaminergic neurons in the SNc but not in the VTA. Male Sprague–Dawley rats were injected with 12 μg 6-OHDA into the mfb (day 0). Animals were killed at the indicated time points and immunostaining was carried out with tyrosine hydroxylase (TH) antibody and TH+ve neurons in SNc were counted. Results are expressed as mean number of cells ± SEM. Level of significance was determined by repeated measure two-way anova test followed by Bonferroni post-test. **p < 0.01, ***p < 0.001 compared to the non-lesioned side. n = 6 animals/per time point.

Figure 2.

 Microglial activation in SNc and VTA after the 6-OHDA lesion. Male Sprague–Dawley rats were injected with 12 μg 6-OHDA into the mfb (day 0). Animals were killed at the indicated time points and immunostaining was carried out with anti-CD11b antibody [cluster of differentiation 11b – anti-complement receptor 3, clone MRC OX42] and OX42+ve microglia in SNc (a) and in VTA (b) were counted. Results are expressed as mean number of cells ± SEM. Level of significance was determined by repeated measure two-way anova test followed by Bonferroni post-test. *p < 0.05, **p < 0.01, ***p < 0.001 compared to the unlesioned side. n = 6 animals/per time point.

Morphological and phenotypic characteristic of activated microglia following 6-OHDA injection

Morphological changes in activated microglia were studied in 3′3-diaminobenzidine-immunolabelled sections. OX6+ve cells (Fig. S1) were densely stained with an activated ramified morphology (enlarged somas and thickened processes) while CD68+ve microglia had a predominant amoeboid macrophage-like appearance. An accumulation of granular staining in the lysosomal membranes was observed in the cytoplasm of CD68+ve cells indicating phagocytic activity.

Extent and localisation of activated microglia

Similarly, there was a significant time dependent increase in the number of activated, OX-6+ve cells, in the 6-OHDA lesioned SNc compared to the unlesioned side (F23.46; p < 0.001). OX6+ve cells were first observed in the medial part and ventral tier of the SNc as soon as day 1 after 6-OHDA lesion induction when no apparent changes were observed in the number of TH+ve cells in the lesion side compared to non-lesioned side of the brain. Their number gradually increased in both the medial and lateral portion of the SNc, reaching significance at day 9 and peeking at day 13 post-6-OHDA lesioning (Fig. 3a, confidence intervals Table S1) when a significant loss of TH+ve neurons in the SNc was also observed (Fig. 1). By day 15, the number of OX6+ve cells ipsilateral to the lesion had decreased but this was still significantly higher than in unlesioned SNc. There was also a time-dependant relocation of activated microglia from the ventral tier to the medial and lateral portions of the SNc suggesting that at least some of the activated microglia accumulated in the SNc might be derived as a result of migration.

Figure 3.

 Injection of 6-OHDA induces accumulation of OX6+ve and CD68+ve cells in the SNc in a time-dependent manner. Graphs (a) and (b) showing time-course quantification of activated microglia: OX6+ve and phagocytic microglia: CD68+ve cells respectively. Results are expressed as mean number of cells ± SEM. Level of significance was determined by repeated measure two way-anova test followed by Bonferroni post-test. *p < 0.05, **p < 0.01, ***p < 0.001 compared to the unlesioned side. n = 6 animals/per time point.

CD68+ve cells were detected in the SNc ipsilateral to the lesion side from day 3 on wards and the increase was time dependent on the 6-OHDA lesioned side of the brain compared to the unlesioned side (F17.55; p < 0.001, Fig. 3b). Their localisation was very similar if not identical to that of OX6+ve cells. CD68+ve cells were mainly confined to the medial aspects of SNc until day 7. Thereafter, the number of CD68 expressing cells progressively increased, reaching significance at day 9 and peeking at day 15 when a large number of CD68+ve cells in the medial and lateral portion of SNc ipsilateral to the lesion were observed. Thus, both OX6 and CD68+ve cells were localised in the SNc on the 6-OHDA lesioned side of the brain and their number increased in a time-dependent manner indicating that the activation of microglia is associated with neuronal degeneration.

The dynamic of OX6+ve microglial cell proliferation and association with degenerating dopaminergic neurons in the SNc

Activated microglial cells showed proliferative activity in the SNc, according to Ki67 immunostaining on days 3, 5, 7 and 9 after 6-OHDA lesion induction. A peak in proliferation was observed on day 7 when a significant proportion of activated microglial cells were positive for OX6 and Ki67 (Fig. 4 lower panel). Thereafter, the amount of Ki67 positive cells gradually declined over the remaining period of the study (cell counts not shown).

Figure 4.

 Top and middle panels show double immunofluorescence images of OX6 (red) and TH (green) at day 9 post-6-OHDA lesion. Note that OX6+ve microglia are specifically adhered to dendrites of degenerating neurons (top panel – see arrows) or are in close proximity to dendritic processes of structurally intact TH+ve neurons (middle panel – see arrows). Lower panel shows double immunofluorescence photograph of OX6 (red) and Ki67 (green) and 4′,6-diamidino-2-phenylindole (DAPI) (blue). Composite image indicates that some of OX6+ve cells are proliferating at day 9 post-lesion.

The double-immunofluorescence staining of TH with OX6 or CD68 at days 3 and 5 post-6-OHDA lesion showed that a considerable number of OX6 and CD68 expressing cells were attached to the morphologically intact TH+ve neurones of the SNc. At day 7 post-6-OHDA lesioning, a large number of OX6 and CD68 expressing cells were observed in close contact to intact DA neurons as well as to the morphologically shrunken neurons with multiple rounded chromatins clumps inside the nuclei indicating neuronal apoptosis (Fig. 4). Moreover, staining with apoptotic marker caspase 3 showed that CD68+ve cells were in close contact to caspase 3 expressing cells (Fig. 5 lower panel) suggesting that phagocytosis of apoptotic cells may take place at very early stage of neuronal loss in SNc. Numerous CD68+ve cells were also observed to be attached to the neuronal fibres. Subsequently, a large number of OX6+ve and CD68+ve microglial cells were found in close proximity or adhered to the degenerating TH+ve cell bodies, axons and cell debris by day 9–15 post-6-OHDA lesioning. (Fig. 4, top two panels). Also, some of CD68+ve microglia were observed to be engulfing apoptotic DA neurons (Fig. 5 top panel). A summary diagram of the increase in microglia activation compared to neuronal loss can be observed in Fig. 6 demonstrating that marked activation occurs prior to significant DA neuronal loss on day 9.

Figure 5.

 Double immunofluorescence image of CD68 (red) and TH (green) or caspase 3 (green) and 4′,6-diamidino-2-phenylindole (DAPI) (blue) at day 7 post-6-OHDA lesion. Top panel: composite image shows engulfment of dopaminergic neurons by CD68+ve microglia. Lower panel: composite image shows caspase 3+ve apoptotic cells are in close contact with CD68+ve phagocytic microglia.

Figure 6.

 Summary of the changes in TH+ve, OX6+ve and CD68+ve cell number ± SEM over time in the 6-OHDA lesion model. ***p < 0.001 numbers of TH+ve and OX6+ve cells in lesioned SNc compared to the unlesioned side.

6-OHDA induced loss of striatal dopamine content

Concomitant with the loss TH+ve DA neurons in the SNc, injection of 6-OHDA into the mfb was followed by significant depletion of striatal DA in the lesioned hemisphere compared to the unlesioned hemisphere (data shown in Fig. S2). DA depletion started to manifest itself from day 9 [lesioned: 4 ± 1.68 (ng/mg), unlesioned: 11.1 ± 0.14 (ng/mg), p < 0.001] similar to the loss of TH+ve neurons in the SNc. Unilateral loss of DA progressed up to day 15 after lesion induction on the side of the lesion compared to the unlesioned hemisphere (lesioned : 4 ± 2.2, unlesioned: 13.02 ± 1.1, p < 0.001) In the sham lesioned animals no significant loss of DA content was observed in the lesioned hemisphere compared to unlesioned hemisphere. A similar pattern of loss of the DA metabolites dihydroxyphenylacetic acid and homovanillic acid was also observed with time after 6-OHDA administration (data not shown).

Discussion

The mechanisms responsible for the progressive loss of DA neurons in PD are poorly understood. In this study, we are the first group to demonstrate in a time course manner that a single injection of 6-OHDA into the mfb initiates early activation of microglia in the SNc, followed by the loss of DA neurons and striatal DA. Specifically, we investigated several aspects of the association of activated microglia in the 6-OHDA model of PD, namely: the time course of microglial activation, morphological changes and immunophenotypic characteristic of activated microglial cells, their interaction with DA neurons in the SN, and the subsequent impact on neuronal loss.

Following CNS injury, microglia activation occurs within a few hours and activated microglia are characterised by an up regulation or de novo expression of several marker proteins, e.g. MHC class I and II (Streit and Kreutzberg 1988; Kaur and Ling 1992; Kreutzberg 1996). A population of activated microglia expresses an antigen, macrophage-myeloid associated antigen, in the lysosomal membranes that is recognized by the monoclonal antibody CD68. The amount of CD68 expression in a single cell can be correlated to phagocytic activity of the respective cell type (Damoiseaux et al. 1994). In fact, it has been demonstrated that CD68+ve microglia are phagocytic in several types of neurodegenerative conditions (Rinaman et al. 1991; Bauer et al. 1994; Frank and Wolburg 1996). We found that OX6+ve microglia, which recognize the rat MHC II, appear in the SNc rapidly after lesioning. These cells appeared earlier than those labelled with CD68. Moreover, it was found throughout all time points of our study that the number of OX6+ve microglia was slightly greater than that of CD68+ve microglia suggesting that a vast proportion of the OX6+ve cells may acquire a phagocytic phenotype but this develops later in the neurodegenerative process (see summary Fig. 6). Similarly, it has been demonstrated that a large number of OX6+ve microglial cells are CD68+ve at the site of neurodegeneration after mfb transection (Cho et al. 2003, 2006). This finding implies that the transformation of microglia to antigen presenting cells expressing MHC II antigens precedes microglial phagocytosis.

The mfb transmits the nigrostriatal fibres from DA neurons of the SNc to the striatum. Unilateral injection of 6-OHDA into the mfb avoids local mechanical trauma to the SNc or striatum and initiate degeneration at the terminals leading to retrograde damage of axons of DA neurons. Activated microglia appeared in the SNc as early as days 1–3 post-6-OHDA lesioning and some activated microglia adhered to the axons or cell bodies of morphologically intact DA neurons, although no significant decrease in the number of nigral TH+ve neurons was detected by 7 days after 6-OHDA, the DA neurons were clearly shrunken. This finding indicates that activated microglia participate in the neurodegenerative process from a very early stage when there has been no apparent loss of DA neurons. Thereafter, a significant number of activated microglia specifically adhered to axons, dendrites, and cell bodies of degenerating DA neurons throughout all time points of the study. This heavy adhesion of activated microglia to the various parts of DA neurons was followed by a significant decrease in the number of DA neuronal cell bodies indicating that activated microglia are directly associated with the cell loss at later time points. Similar observations were made by Rodriguez-Pallares et al. (2007) who at a single 48 h time point after 6-OHDA striatal administration, observed microglial activation coupled with NADPH-derived free radicals administration and shrunken TH+ve DA neurons, but not neuronal loss. Rodriguez-Pallares et al. (2007) proposed that NADPH-derived free radicals from microglia act in concert with the oxidative stress induced by 6-OHDA to bring about DA neuronal cell death in the SNc. Furthermore, in mfb axotomised brain, it has been also demonstrated that activation of microglia preceded neuronal loss and activation of microglia was followed by significant loss of neuronal fibres and nigral DA neurons (Cho et al. 2006). These findings strongly suggest that activated microglia play an important role in the progress of neurodegeneration.

The highly ramified form of resting microglia observed the SNc of normal brains supports their immune surveillance function. Pathological conditions lead to their activation and morphological changes. An increase in their soma size as well as thickened and shortened processes is observed in early stage of activation before their final transformation into amoeboid microglia with phagocytic activity (Streit and Kreutzberg 1988; Mor et al. 1999; Stence et al. 2001). We found numerous activated ramified and amoeboid like microglia on the lesioned side of the SNc. Both forms of activated microglia showed strong CD68 immunoreactivity and attachment to the TH+ve neurons, axons and dendrites which was followed by significant loss of DA neurons. This suggests that activated microglia with ramified morphology can perform active phagocytosis. This is consistent with a previous report where ramified form of activated microglia showed strong expression of CD68 at the lesioned site where the mfb had been cut (Cho et al. 2006). Thus, adhesion of activated microglia to damaged DA neurons represents a process linked with the phagocytic clearance of those neurons. The close proximity of phagocytic microglia very early on in the time course suggested that the microglia may have stripped off the neurites/dendrites of damaged neurons prior to phagocytosis of the cell body. The numbers of phagocytic microglia early in the time course was low but detectable but when significant cell loss was observed the numbers of phagocytic microglia markedly increased, which is probably due to the presence of cell debris from the degenerating neurons.

Activated microglia can also be characterized by their proliferative and migratory properties. Accumulation of activated microglia around regions of the brain undergoing degeneration may be attributed to migration of microglia to the affected sites (Akiyama et al. 1994; Heppner et al. 1998; Schiefer et al. 1999; Marella and Chabry 2004; Rodriguez et al. 2007), as well as their local proliferation (Graeber et al. 1988; Amat et al. 1996; Schwab et al. 2001). We have observed a time-dependant relocation of activated microglia from the ventral tier to the medial and lateral portions of the SNc suggesting that at least some of the activated microglia accumulated in the SNc might be derived as a result of migration. Similar to this observation, movement of amoeboid microglia along dendrites of neurons has been identified in brainstem slices using infrared-gradient contrast microscopy and time-lapse video recording (Schiefer et al. 1999). Using nuclear mitotic protein marker Ki67, we have observed numerous proliferating microglial cells in the SNc at days 3–7 post-6-OHDA lesioning. Thus, accumulation of activated microglia in the SNc ipsilateral to the lesion with 6-OHDA can be explained by both local proliferation and migration.

Injection of 6-OHDA induced apoptosis of DA neurones. 4′,6-diamidino-2-phenylindole staining of the nuclei of DA neurons in the SNc at day 7 post-lesioning showed multiple, rounded chromatin clumps similar to that observed during apoptosis (Tatton and Kish 1997). However, distinguishing the apoptosis/necrosis dichotomy in the CNS is a difficult task, and many variations of apoptotic processes are possible. We observed DNA clumps in the nuclei of shrunken, apparently dead neurons in the lesioned site of SNc. As a continuum of neuronal death occurs post-6-OHDA lesioning, it is reasonable to assume that a few DA neurons degenerated soon after toxin inoculation without having been phagocytosed by microglia. However, at days 3–5 post-lesion, the prevalence of phagocytic microglia in the SNc attached to DA neurons which were apparently morphologically intact with normal nuclear chromatin distribution and were expressing TH led us to conclude that a majority of degenerating neurons were targeted for phagocytosis at very early apoptotic stage. In fact, we observed that the majority of caspase 3+ve cells were in close contact with CD68+ve phagocytic microglia in SNc ipsilateral to the lesion as early as 5–7 days post-lesion. This observation is in accordance with recent reports demonstrating that apoptotic neurons could be phagocytosed by activated microglia in early neuronal apoptosis (Savil and Fadok 2000; Cho et al. 2003; Sugama et al. 2003). However, the presence of CD68+ve phagocytic microglia in close proximity to the DA neurons prior to the detection of caspase 3 would suggest that early microglial phagocytosis of pre-apoptotic neurons might preclude the possibility of their recovery. We observed that CD68+ve activated microglia appeared at day 3 post-lesion induction, and the analysis of TH+ve neuron numbers at this time point indicated no significant difference between the SNc ipsilateral and contralateral to the lesion suggesting that microglial phagocytic activity preceded neuronal death. A major implication of these observations would be that early phagocytosis of apoptotic neurons might prevent neuronal survival by phagocytosing potentially recoverable neurons in neurodegenerative disorders such as PD.

The present study clearly demonstrates that both CD68+ve and OX6+ve microglia specifically adhere to degenerating neurons throughout the whole neurodegenerative process from an early stage phagocytosing damaged neurons. This process is probably in response to expression of apoptotic signals, such as phosphatidylserine and expression of their receptors, including phosphatidylserine receptor, on the phagocyte membrane that initiate phagocytosis (Huynh et al. 2002; Ravichandran 2003), or in response to contents released from dying neurons. Our results indicate that neurons undergoing neuronal cell death may be phagocytosed by activated microglia concomitantly at various stages. The present data imply that blockade of microglial activation or phagocytic activity, might be targets for therapeutic intervention in neurodegenerative disorders such as PD.

Acknowledgement

This work was supported by a grant from GlaxoSmithKline.

Ancillary