OUP user menu

Degeneration in Different Parkinsonian Syndromes Relates to Astrocyte Type and Astrocyte Protein Expression

Yun Ju C. Song PhD, Glenda M. Halliday PhD, Janice L. Holton PhD, FRCPath, Tammaryn Lashley PhD, Seán S. O'Sullivan MRCPI, Heather McCann BmedSci(Path), Andrew J. Lees MD, FRCP, Tetsutaro Ozawa MD, PhD, David R. Williams PhD, FRACP, Paul J. Lockhart PhD, Tamas R. Revesz MD, FRCPath
DOI: http://dx.doi.org/10.1097/NEN.0b013e3181b66f1b 1073-1083 First published online: 1 October 2009


The reactive changes in different types of astrocytes were analyzed in parkinsonian syndromes in order to identify common reactions and their relationship to disease severity. Immunohistochemistry was used on formalin-fixed, paraffin-embedded sections from the putamen, pons, and substantia nigra from 13 Parkinson disease (PD), 29 multiple-system atrophy (MSA), 34 progressive supranuclear palsy (PSP), 10 corticobasal degeneration(CBD), and 13 control cases. Classic reactive astrocytes were observed in MSA, PSP, and CBD, but not PD cases; the extent of reactivity correlated with indices of neurodegeneration and disease stage. Approximately 40% to 45% of subcortical astrocytes in PD and PSP accumulated α-synuclein and phospho-tau, respectively; subcortical astrocytes in MSA and CBD cases did not accumulate these proteins. Protoplasmic astrocytes were identified from fibrous astrocytes by their expression of parkin coregulated gene and apolipoprotein D, and accumulated abnormal proteins in PD, PSP, and CBD, but not MSA. The increased reactivity of parkin coregulated gene-immunoreactive protoplasmic astrocytes correlated with parkin expression in PSP and CBD. Nonreactive protoplasmic astrocytes were observed in PD and MSA cases; in PD, they accumulated α-synuclein, suggesting that the attenuated response might be due to an increase in the level of α-synuclein. These heterogenous astroglial responses in PD, MSA, PSP, and CBD indicate distinct underlying pathogenic mechanisms in each disorder.

Key Words
  • α-Synuclein
  • Astrocytes
  • Corticobasal degeneration
  • Multiple system atrophy
  • Parkin
  • Parkinson disease
  • Progressive supranuclear palsy
  • Tau


There are differences in astrocytic reactions in different parkinsonian syndromes. In Parkinson disease (PD) cases, reactive astrogliosis is minimal (1-3), but abnormal α-synuclein deposition occurs in astrocytes and relates to the severity of neuronal Lewy body (LB) inclusions (4, 5). In contrast to PD, all other parkinsonian syndrome cases show marked reactive astrogliosis (6, 7). Astrocytes in multiple system atrophy (MSA) do not accumulate α-synuclein (6), whereas in progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD) cases, some reactive astrocytes accumulate phospho-tau (8). The corresponding morphologic changes in tau-accumulating astrocytes in PSP and CBD (i.e. tufted astrocytes and astrocytic plaques, respectively) are diagnostic for these disorders and are thought to directly contribute to the neurodegeneration (7, 8).

Reactive astrogliosis is associated with the upregulation of glial fibrillary acidic protein (GFAP) and cell enlargement (9). These changes occur after many types of brain injury in different astrocyte populations depending on the location of the injury or change. The gray matter is populated by protoplasmic astrocytes that have a mossy appearance with radiating branched processes and express apolipoprotein D (ApoD), a lipid transport protein involved in CNS repair and regeneration (10). Fibrous astrocytes populate CNS white matter and have longer, less branched processes; they do not express ApoD (10). Some PD-associated proteins (e.g. parkin [11], parkin coregulated protein [PACRG] [12], DJ-1 [13], and PINK-1 [14]) are also concentrated in astrocytes, but the type of astrocyte has not been identified. As the expression of Parkin and PACRG are coregulated (15), both proteins were evaluated.

The determinants of the different types of astrocytic changes and whether the changes observed relate directly to the neurodegeneration in the different parkinsonian disorders are not clear. To our knowledge, there have been no studies assessing whether different reactions occur because of the type of astrocyte affected, whether changes in the expression of astrocytic proteins have pathologic correlates, and whether these changes relate to the neurodegeneration observed. We assessed reactive changes in different types of astrocytes in PD, MSA, PSP, and CBD to identify any relationships to the severity of the degenerative changes observed.

Materials and Methods


Formalin-fixed brain tissue samples that fulfilled the pathologic criteria for PD (n = 13) (16), MSA (n = 29) (17, 18), PSP (n = 34) (19), CBD (n = 10) (20), or age-matched controls without significant neuropathologic abnormalities (n = 13) were provided by the Prince of Wales Medical Research Institute Human Tissue Resource Centre after study approval by the Australian Brain Donor Programmes' Scientific Advisory Committee and by the Queen Square Brain Bank, UCL Institute of Neurology. All tissues were collected with appropriate consent from brain donors and/or their next of kin, and the collection programs approved by appropriate institutional or regional human ethics committees. Tissues from some of the cases from the Queen Square Brain Bank have been previously used and published in other studies (6, 21, 22). Unpaired t-tests (SPPS, Inc., Chicago, IL) showed no differences in the demographics (age at onset, age at death, disease duration, or postmortem delay) of the different types of cases from the 2 sources.

Tissue Preparation

In all cases, multiple brain regions were screened using immunohistochemical protocols for abnormal accumulations of pathologic proteins. Tissues blocks from the putamen (dorsolateral level), substantia nigra (SN; transverse level including the red nucleus), and the pontine base (transverse level, including the superior cerebellar peduncle) were analyzed. Sections (5-μm thick) were cut on a microtome, mounted on 3-aminopropyltriethoxysilane-coated slides, and deparaffinized.

The following antibodies were used for identifying pathologic inclusions: mouse monoclonal antibody (mAb) clone 42 to α-synuclein (1:200; BD Biosciences, San Jose, CA), rabbit anti-α-synuclein (1:2000; Abcam, Cambridge, UK), mouse anti-tau mAb (1:600; Autogen Bioclear, Calne, UK), and rabbit anti-tau (Abcam; 1:500). Astrocytic proteins were identified using mouse anti-GFAP mAb (1:1000; DAKO, Ely, UK); rabbit anti-GFAP (1:750; DAKO); mouse anti-ApoD mAb (1:50; Vector Laboratories, Burlingame, CA); mouse anti-proteoglycan NG2 mAb (1:100; Millipore, Billerica, MA); rabbit anti-parkin (1:200; Biosensis, Adelaide, Australia); rabbit anti-PACRG (MC1290, 1:100; kindly provided by Dr. Paul Lockhart).

Immunohistochemical staining and counterstaining with cresyl violet (0.5%) or Mayer hematoxylin were performed as previously described (6, 12, 22). Various antigen retrieval methods (99% formic acid/pressure cooking/citrate buffer pH 6.0) were used before antibody incubation to maximize antigen detection. The specificity of the immunohistochemical reaction was tested by omitting the primary antibody, and no peroxidase reaction was subsequently observed in those sections. The specificity of the parkin antibody was confirmed by negative immunostaining using the same protocol in formalin-fixed, paraffin-embedded brain tissue from quaking mice that have deletions of both parkin and PACRG (Fig. 1) (23). The specificity of the PACRG antibody, that is, Western blot assessment of soluble extracts from human brain cortex to confirm a 29-kDa band and incubating it with purified recombinant PACRG protein, as well as with preimmune serum to show no peroxidase staining observed in these sections, has been previously detailed (12).


Parkin immunoreactivity in mouse striatum. A control mouse (A) shows parkin immunoreactivity, whereas a quaking mouse with parkin and parkin coregulated gene deletions (B) is negative. KO, knockout.

Double labeling immunofluorescence was performed as previously described (12) using the same antigen retrieval methods on additional slides. Rabbit anti-GFAP was mixed with either mouse anti-α-synuclein or anti-tau. Mouse anti-GFAP, anti-ApoD, anti-α-synuclein, or anti-tau mAb was mixed with either rabbit anti-PACRG or anti-parkin. Mouse anti-NG2 was mixed with rabbit anti-tau. The fluorescent probes used for the detection of primary antibodies were secondary antibody anti-mouse conjugated to Alexa 488 (1:500; Molecular Probes, Carlsbad, CA) or tetramethylfluorescein (1:200; Perkin Elmer, Waltham, MA) and anti-rabbit conjugated to Alexa 568 (1:250; Molecular Probes) or tetramethylrhodamine (1:500; Perkin Elmer). To ensure specificity of the immunohistochemical reactions and non-cross-reactivity of secondary fluorescent probes, a section without primary antibodies was included for each staining procedure as a negative control. Additionally, a mixture of the secondary antibodies was applied to sections with only 1 primary antibody incubated on each section. For primary antibodies from the same species (parkin and PACRG), a combined method of immunofluorescence and peroxidase visualization was used as previously described (24). Pretreatment methods and standard peroxidase immunohistochemistry were performed with the incubation of anti-PACRG (1:100) as the first primary antibody. After peroxidase visualization, slides were then incubated with anti-parkin (1:200) as the second primary antibody for 1 hour at 37°C and then incubated with the secondary antibody anti-rabbit conjugated to Alexa 568 (1:250) for 2 hours at room temperature. Similar controls to those previously described revealed no cross-reactivity using this method.


The peroxidase-stained sections from all cases were viewed using a Zeiss Axioskop MC80 DX or an Olympus BX51 microscope to evaluate astrocyte morphology. The extent of reactive GFAP-, PACRG-, and parkin-immunoreactive astrocytes was scored in 3 random 100× magnification fields in each section from each case as follows: none or rare; mild (a sparse distribution); moderate (in close to half the field of view); or severe (densely spread throughout the field of view). Because parkin immunoreactivity was only observed in pathologic and not in control samples, further quantitative assessment with a stereologic approach was performed on these samples using an imaging analysis software program (Image Pro Plus, version 6.2; Media Cybernetics, Bethesda, MD). Using a motorized stage attached to the microscope (Nikon Eclipse 50i), the regions of interest were delineated using a computer mouse and random sites sampled at 400× magnification by the imaging software program to quantify parkin-immunoreactive astrocytes. Intrarater and interrater analysis was less than 5% variance using this method. Nonparametric Kruskal-Wallis and Mann-Whitney U tests (SPPS, Inc.) were performed to identify group differences.

To assess the degree of protein colocalization, double labeled sections were viewed and photographed using Zeiss Axioskop MC80 DX or Olympus BX51 microscopes (Axiocam camera and Axiovision software), and confocal images were taken using a Leica TCS4D confocal microscope with a 3-channel scan head and argon/krypton laser (University College London confocal imaging unit). The proportion of astrocytes with colocalizing proteins was determined in 10 random photographs of each section for each case taken at 200× magnification.

To establish correlates with pathologic severity, a number of indices were evaluated, including neuronal loss in the SN, LB severity (graded by sampling the whole region of interest at 100× magnification as none, 1, 2-6, and 7 or more LB) and Braak LB stage (25) for PD, glial cytoplasmic inclusion (GCI) severity (graded as for astrocytes) and MSA stage (26) for MSA, neurofibrillary tangle severity (graded as for astrocytes) and PSP stage (22) for PSP and overall tau severity (graded as for astrocytes) and disease stage (27) for CBD. Spearman rank tests (SPPS, Inc.) were performed to determine any relationships between the degree of astrocytic reactivity and indices of pathologic severity.


Pathologic Changes in Astrocytes

Glial fibrillary acidic protein-immunoreactive astrocytes in PD cases had normal stellate appearances (Fig. 2A, B), and approximately 45% also showed α-synuclein immunoreactivity using either α-synuclein antibody (Fig. 2C-E). Only the SN showed mildly increased GFAP immunoreactivity (Table 1).

View this table:

Pathologic changes in astrocytes. (A, B) Brightfield photomicrographs of similar typical stellate astrocytes with finely branching processes immunostained for glial fibrillary acidic protein (GFAP) in the putamen of a control (A) and a Parkinson disease (PD) case (B). (C-E) Immunofluorescent images of a typical stellate GFAP-immunoreactive (red Alexa 568) (C, E) astrocyte containing α-synuclein (green Alexa 488) (D, E) in the pons of a PD case. (F) Brightfield photomicrograph of an enlarged and distorted GFAP-immunoreactive astrocyte in a white matter bundle of a multiple system atrophy (MSA) case. (G) Brightfield photomicrograph of an enlarged reactive GFAP-immunoreactive astrocyte in the putamen of a progressive supranuclear palsy (PSP) case. (H) Immunofluorescent merged image of a GFAP-immunopositive (red Alexa 568) reactive astrocyte containing phospho-tau (green Alexa 488) in the putamen of a PSP case. (I) Immunofluorescent merged image of a phospho-tau immunopositive (green Alexa 488), but GFAP-immunonegative (red Alexa 568) tufted astrocyte in the putamen of a PSP case. (J) A brightfield photomicrograph of enlarged reactive GFAP-positive astrocytes in the putamen of a corticobasal degeneration (CBD) case. Brightfield photomicrographs are counterstained with cresyl violet.

In MSA cases, reactive astrocytes had enlarged cell bodies and distorted processes (Fig. 2F). Double labeling immunofluorescence showed no deposition of α-synuclein in GFAP-immunopositive astrocytes, as previously reported (5). The putamen in MSA cases had the most marked increase in GFAP immunoreactivity (Table).

In PSP cases, reactive astrocytes had enlarged stellate morphologies (Fig. 2G), and approximately 40% also had phospho-tau immunoreactivity (Fig. 2H). Tau-immunoreactive tufted astrocytes were not GFAP (Fig. 2I), as previously reported (28). There was obvious reactive astrogliosis in all regions examined in PSP cases, but the most prominent was in the SN (Table).

Reactive astrocytes in CBD cases were similar to those in PSP (Fig. 2J). No tau-immunoreactive astrocytes or astrocytic plaques were observed in the regions examined, although all CBD cases had cortical astrocytic plaques and increased GFAP immunoreactivity in the putamen and SN. The pons was not available for analysis in CBD cases (Table).

In summary, typical astrocytic reactivity was observed in MSA, PSP, and CBD cases with no morphologic changes and only mild increases in numbers of GFAP-immunopositive astrocytes in the SN in PD cases. Approximately 40% to 45% of subcortical astrocytes analyzed accumulated pathologic proteins in PD and PSP but not in MSA or CBD cases.

The slight reactive astrogliosis in PD cases did not correlate with either the severity of SN cell loss or the severity of LB formation (Table). In MSA, the degree of GFAP-immunoreactive astrocytes in the putamen correlated with the degree of α-synuclein-immunoreactive GCI in the pons (Rho = 0.73; p = 0.007) and MSA pathologic stage (Rho = 0.67; p = 0.02). In PSP, the degree of GFAP-immunoreactive astrocytes in both the SN and putamen positively correlated with increasing SN neuronal loss (Rho > 0.53; p < 0.04) and SN neurofibrillary tangle formation (Rho > 0.56; p < 0.03). In CBD, increasing GFAP-immunoreactive astrocytes in the SN correlated with SN neuronal loss (Rho = 0.97; p = 0.007) and CBD stage (Rho = 0.97; p = 0.007). In addition, GFAP-immunoreactive astrocytes in the putamen also correlated with CBD stage (Rho = 0.97; p = 0.007) and overall tau severity in the SN (Rho = 0.95; p = 0.03). Except for PD, astrogliosis directly correlated with indices of neurodegeneration in MSA, PSP, and CBD.

Types of Astrocytes Affected

Apolipoprotein D is selectively expressed in protoplasmic astrocytes (10). All ApoD-immunoreactive astrocytes in controls were also immunoreactive for PACRG (Fig. 3A-C). Despite previous descriptions of parkin-immunoreactive glia (29), only PACRG immunoreactivity was observed in the typical stellate astrocytes in controls in the regions examined (Fig. 3D). Colocalization experiments indicated that approximately 80% of GFAP-immunoreactive astrocytes also contained PACRG immunoreactivity (Fig. 3E). This approximates the expected proportion of protoplasmic versus fibrous astrocytes in these regions and suggests that they all constituently express PACRG and ApoD.


Types of astrocytes affected. (A-C) Double labeled immunofluorescence photomicrographs of parkin coregulated gene (PACRG) immunoreactivity (red Alexa 568) (A, C) colocalizing with apolipoprotein D (ApoD) immunoreactivity (green Alexa 488) (B, C) in a control case astrocyte. (D) Brightfield micrograph of protoplasmic astrocytes with immunolabeled PACRG in a control case. (E) Immunofluorescent merged image of a glial fibrillary acidic protein (GFAP)-immunopositive (green Alexa 488) reactive astrocyte also immunopositive for PACRG (red Alexa 468) in a control case. (F) Merged immunofluorescent image of PACRG-immunopositive (red Alexa 468) astrocytes containing α-synuclein (green Alexa 488) immunoreactivity in the pons of a Parkinson disease (PD) case. (G-I) Double labeled immunofluorescence of enlarged PACRG-immunoreactive stellate protoplasmic astrocytes (red Alexa 568) (G, I) with ApoD immunoreactivity (green Alexa 488) (H, I) in the putamen of a progressive supranuclear palsy (PSP) case. (J-L) Immunofluorescence photomicrographs of parkin-immunoreactive astrocytes (red Alexa 568) (J, L) containing phospho-tau immunoreactivity (green Alexa 488) (K, L) in the putamen of a PSP case. (M) Immunofluorescence merged image of an NG2-immunopositive glial cell (green Alexa 488) immunonegative for phospho-tau immunoreactivity (redAlexa 568) in the putamen of a PSP case.

Different types of astrocytes had distinct reactions in the different parkinsonian syndromes. The distribution of ApoD-/PACRG-immunoreactive astrocytes did not differ from controls in PD or MSA. This suggests that the non-PACRG fibrous astrocytes may be selectively affected in MSA. In PD cases, many ApoD/PACRG-immunoreactive protoplasmic astrocytes also showed α-synuclein immunoreactivity (Fig. 3F) but did not show increased GFAP (Table) and were not associated with significant morphologic changes, whereas in PSP and CBD, most ApoD-/PACRG-immunoreactive protoplasmic astrocytes had enlarged reactive morphologies (Fig. 3G-I). Tau-immunoreactivity was found in approximately 25% of ApoD-/PACRG-immunoreactive and in approximately 20% of PACRG-/parkin-immunoreactive astrocytes only in PSP (Fig. 3J-L). These astrocytes did not have the typical morphologic appearance of tufted astrocytes. Tau-immunoreactive tufted astrocytes were also not GFAP-immunoreactive (Fig. 2I), suggesting they may not be astrocytic. Other glial cells with similar morphology are the NG2 cells, which do not express GFAP but express NG2 chondroitin sulfate proteoglycan and accumulate rapidly in glial scars after injury (30, 31). The tau-immunoreactive tufted astrocytes were also not NG2-immunoreactive (Fig. 3M). Protoplasmic astrocytes were abnormal in PD, PSP, and CBD but not MSA. They were reactive in PSP and CBD but not PD, and accumulated abnormal proteins in PD and PSP but not CBD.

Astrocytic PACRG and Parkin

There were no changes in the densities or morphology of PACRG-immunoreactive astrocytes in the putamen, SN, or pons in PD cases compared with controls (Fig. 4A), and there was no parkin immunoreactivity (Table). In MSA, there was only a mild increase in the density of PACRG-immunoreactive astrocytes in the putamen, no change from controls in the SN or pons, and no parkin immunoreactivity (Table). In PSP and CBD, the SN and putamen showed greater astrocytic PACRG immunoreactivity than in the pons (Table), with morphologic enlargement similar to that observed for GFAP. Enlarged reactive parkin-immunoreactive astrocytes were also observed in PSP and CBD (Fig. 4B). Double labeling experiments showed that approximately 70% of GFAP-immunoreactive astrocytes in PSP and CBD also showed parkin immunoreactivity (Fig. 4C-E), whereas approximately 80% of GFAP-immunoreactive astrocytes also showed PACRG immunoreactivity (Fig. 4F-H). Approximately 90% of PACRG-immunoreactive astrocytes also showed parkin immunoreactivity (Fig. 4I, J). As shown in the Table, the putamen and SN were regions with consistent astrogliosis in PSP and CBD. There was no significant difference observed in the SN, but significantly increased densities of parkin-immunoreactive astrocytes was found in the putamen of PSP compared with CBD (×2; U = 16; p = 0.03; Table). Overall, reactive PACRG-/parkin-immunoreactive astrocytes were observed only in PSP and CBD.


Astrocytic parkin and parkin coregulated gene (PACRG) expression. (A) Brightfield micrograph of PARCG-immunoreactive protoplasmic astrocytes from a Parkinson disease (PD) case. The astrocyte morphology is similar to those in control cases. (B) Brightfield micrograph of enlarged reactive parkin-immunopositive protoplasmic astrocytes from a progressive supranuclear palsy (PSP) case. (C-E) Immunofluorescent photomicrographs of parkin immunoreactivity (red Alexa 568) (C, E) colocalizing with glial fibrillary acidic protein (GFAP) immunoreactivity (green Alexa 488) (D, E) in an astrocyte in a PSP case. (F-H) Immunofluorescent photomicrographs of PACRG immunoreactivity (red Alexa 568) (F, H) colocalizing with GFAP immunoreactivity (green Alexa 488) (G, H) in an astrocyte from a PSP case. (I, J) Double labeled photomicrographs of an enlarged protoplasmic astrocyte from a PSP case that shows colocalization of PACRG (brightfield immunoperoxidase) (I) and parkin (red Alexa 568) (J) immunoreactivity.

In PSP and CBD, the severity of PACRG-/parkin-immunoreactive astrocytes in the putamen positively correlated with the PACRG-/parkin-immunoreactive astrocytes in the pons (Rho = 0.55; p < 0.03) and neuronal loss in the SN (Rho > 0.56; p < 0.04) and in CBD with increasing CBD stage (Rho = 0.97; p = 0.007). The severity of PACRG-/parkin-immunoreactive astrocytes in the SN correlated with increasing neurofibrillary tangles in PSP pons (Rho = 0.72; p = 0.006) and overall tau severity in CBD SN (Rho = 0.92; p = 0.03). This astrocytic change directly correlated with indices of neurodegeneration in PSP and CBD.


Our data show significantly different astrocytic reactions among PD, MSA, PSP, and CBD cases. Different types of astrocytes are selectively involved and their reactions are dependent on the proteins they express and accumulate. In particular, α-synuclein accumulation in PACRG-expressing protoplasmic astrocytes does not seem to be related to neurodegeneration and seems to diminish the reactive capacity of astrocytes in PD. In contrast, astrocytic reactivity is directly related to neurodegeneration in PSP and CBD cases in which the protoplasmic astrocytes express parkin in addition to constituitively expressing PACRG. In MSA, neurodegeneration seems to be related to reactivity of fibrous astrocytes that do not express any of these proteins.

Protoplasmic astrocytes are the dominant glial cell type in CNS gray matter, whereas fibrous astrocytes are mainly found in white matter (9, 32). In addition to PACRG, ApoD (10) and d-serine (a ligand for synaptic NMDA receptors) (33) are expressed in protoplasmic astrocytes. Of these ligands, only PACRG seems to be restricted to protoplasmic astrocytes in human brain tissue, and it was not found in other cell types in the regions sampled. Because PACRG is a tubulin-binding structural component of microtubules (34, 35), it may be an important component of the cytoskeleton of protoplasmic but not fibrous astrocytes; its expression seems to increase along with morphologic reactive changes in these astrocytes.

Reactive changes in protoplasmic astrocytes in the regions examined correlated with nigral cell loss, widespread pathologic phospho-tau deposition, and overall disease stage in CBD and PSP. An association between reactive astrogliosis and pathologic phospho-tau has previously been shown in PSP (7, 36) and to disease stage in CBD (37). In the present study, we found this reactivity of protoplasmic astrocytes to be associated with the expression of parkin. PACRG and parkin are coregulated (12, 38); therefore, parkin expression may be due to the upregulation of the constituent PACRG protein required for the cytoskeletal changes observed in these astrocytes in these disorders. PACRG binds and interacts with tubulin and microtubules (35), and its levels are regulated by the proteasome (12). Parkin is a ubiquitin E3 ligase for proteasomal protein degradation and for stabilizing misfolded proteins on microtubules (39), with particular parkin isoforms enriched in protoplasmic astrocytes (40). We speculate that upregulation of PACRG and parkin in protoplasmic astrocytes would increase microtubule stability as the cell enlarges.

A minority of subcortical protoplasmic astrocytes and no NG2-positive cells (a separate type of glial cell) contained phospho-tau in PSP but not in CBD. The tau isoform that deposits in astrocytes is 4-repeat tau with both PSP and CBD sharing a common tau haplotype (41). Corticobasal degeneration is part of the spectrum of frontotemporal degeneration syndromes that have cortical astrocytic degeneration in association with neuronal loss (42); in PSP, enlarged phospho-tau-depositing tufted astrocytes dominate regions undergoing degeneration. Experimental studies using parkin knockouts and mixed parkin/tau vectors suggest that parkin expression is a normal protective response that reduces tau levels and phosphorylation (43-46). The different end result of such increased astrocytic reactivity in CBD versus PSP points to potential differences in these dynamic intracellular processes. Of interest is the recent finding that there is a genetic association for PSP with the valine-380-leucine polymorphism in the C-terminus region of the parkin gene (47), while the pathology of CBD has many similarities to that observed in parkin-null mice expressing 4-repeat tau (44-46, 48). Certain wild-type parkin isoforms are prone to misfolding under severe oxidative stress, and most types of misfolded parkin are rapidly degraded (49).

Parkin folding is dependent on the C-terminus structure (49), and we speculate that the parkin isoform in PSP protoplasmic astrocytes may be more stable compared with that found in CBD, although there are no functional data to support this at this time. This would lead to the eventual incorporation of misfolded phospho-tau into the cytoskeletal structures of protoplasmic astrocytes in PSP in contrast to their cytoskeletal instability under similar end-stage conditions in CBD. Functional studies on these different parkin isoforms in appropriate cell systems are needed.

In PSP, only a small proportion of protoplasmic astrocytes accumulated phospho-tau, and these exhibited less abnormal morphology than was observed in tufted astrocytes. Although tufted astrocytes have been previously described as protoplasmic based on their morphology and distribution (50), they have also been shown to be GFAP-negative (28). We found that they are also NG2-negative. All protoplasmic astrocytes are GFAP-immunoreactive (9, 32), whereas NG2 glia do not express GFAP and have large, stellate morphologies (31). Based on this morphology, NG2 glia were previously suggested to accumulate phospho-tau and form tufted astrocytes (51). Because no NG2 glia colocalized with tau but a proportion of protoplasmic astrocytes did colocalize phospho-tau, our data support the idea that in PSP protoplasmic astrocytes further metamorphose to incorporate phospho-tau rather than other elements into their enlarged cytoskeleton. The recent identification of significant oxidative damage leading to GFAP fragmentation in PSP (52) may explain the loss of this cytoskeletal element. Our data support the concept that this is a late event (52), and that it is relatively independent of reactive astrogliosis (7), with the loss of normal proteins suggesting significant dysfunction of these protoplasmic astrocytes consistent with their association with increasing neuronal dysfunction (7, 36).

The distribution of α-synuclein-immunoreactive astrocytes parallels the spread of intraneuronal pathology in PD (4, 5, 53). In the present study, however, this astrocytic response was not associated with astrocytic reactivity, it was not directly associated with the severity of neuronal inclusion formation, and it was more widespread than the PD neuronal abnormalities. As in PSP and CBD, the same protoplasmic astrocytes were affected in PD, but their responses differed. Protoplasmic astrocytes in PD do not increase GFAP, PACRG, or parkin; rather, they accumulate α-synuclein. Protoplasmic astrocytes normally express β-synuclein (54), whereas α-synuclein expression occurs in response to certain inflammatory cytokines (55). Increased astrocytic α-synuclein enhances their susceptibility to oxidative stress and induces apoptosis in cell culture (56-58). We speculate that astrocyte apoptosis due to increased α-synuclein might give the impression of a less reactive astrocytic population in PD. In this regard, patients with parkinsonism and mutations in the parkin gene do not accumulate α-synuclein in neurons or glia (59-61) but have a substantial microglial response associated with their neurodegeneration (61), a feature also found in parkin-null mice (48). In parkin-null mice, this is associated with early astrocytic degeneration (48).

The protoplasmic astrocytes in MSA cases did not seem to be affected despite considerable neurodegeneration; there was, however, marked reactivity of fibrous GFAP-positive, PACRG-negative astrocytes. This difference between the astrocyte response in gray versus white matter has recently been highlighted (62). The astrocytic reactivity in MSA correlates with the degree of α-synuclein-immunoreactive GCI, suggesting that oligodendroglial pathology and degeneration drive this process. Such a relationship may have been expected considering the association between fibrous astrocytes and myelinated fibers. What is more remarkable, perhaps, is the lack of reactivity of the protoplasmic astrocytes in regions undergoing neuronal cell death in MSA. Apart from an absence of α-synuclein, the response of protoplasmic astrocytes is similar to that observed in PD and may suggest their targeted degeneration by alternative means.

In summary, this is the first study to identify PACRG as a constituent protein in protoplasmic astrocytes; it is the first to show that different types of astrocytes are affected in different parkinsonian disorders; and it is the first to associate protoplasmic astrocytic parkin expression with reactivity and neurodegeneration in PSP and CBD. Although protoplasmic astrocytes were not spared in PD, a completely different reaction was observed. The protoplasmic astrocytes in PD abnormally accumulated α-synuclein but did not become reactive. A lack of reaction in protoplasmic astrocytes to significant neurodegeneration was also a feature of MSA, although in MSA, the oligodendroglial GCI caused significant reactivity in associated fibrous astrocytes. This suggests that both PD and MSA have an attenuated protoplasmic astrocytic response. Because increased astrocytic levels of α-synuclein increases their susceptibility to oxidative stress and induces apoptosis, degeneration of protoplasmic astrocytes may occur in both these disorders, giving the appearance of an attenuated response. Overall, these studies reveal significantly different astroglial responses in PD, MSA, PSP, and CBD, supporting different underlying pathogenic cellular mechanisms.


The authors thank Heidi Cartwright for assistance with figure work and Karen Murphy for technical support.


  • Song and Halliday contributed equally.

  • Sources of support: Y.J.C.S. was an Australian Postgraduate and Parkinson's NSW Scholar and also funded by a GlaxoSmithKline Australia Postgraduate Support Grant. G.M.H. is a Principal Research Fellow of the National Health and Medical Research Council of Australia. T.R., J.L.H., and A.J.L. are supported by grants from Sara Matheson Trust for Multiple System Atrophy (Margaret Watson Memorial Grant), Alzheimer's Research Trust, and Grant No. LSHM-CT-2004-503039 from the BrainNet Europe II. Some of this work was undertaken at the University College London Hospitals and University College London, which received a proportion of funding from the Department of Health's National Institute for Health Research Biomedical Research Centres funding scheme. Tissues were received from the Prince of Wales Medical Research Institute Human Tissue Resource Centre, which is supported by Enabling Grant No. 282933 from the National Health and Medical Research Council of Australia, and the Prince of Wales Medical Research Institute and from the Queen Square Brain Bank, which are supported by the Reta Lila Weston Institute for Neurological Studies and the Progressive Supranuclear Palsy (Europe) Association.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
View Abstract