Decreased immunoreactivities of neocortical AMPA receptor subunits correlate with motor disability in Lewy body dementias
- First Online:
- Cite this article as:
- Mohamed, N., Howlett, D.R., Ma, L. et al. J Neural Transm (2014) 121: 71. doi:10.1007/s00702-013-1067-0
- 178 Views
Dementia with Lewy bodies and Parkinson’s disease dementia are different clinical phenotypes of Lewy body dementias differentiated by the temporal relationship between parkinsonism and dementia onset. At present, it is unclear whether the glutamatergic system is affected in these disorders. In this study, we measured α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor GluA subunits in the postmortem neocortex of a cohort of prospectively studied Lewy body dementia cases, as well as age-matched controls by immunoblotting. We found losses of GluA2/3/4 immunoreactivities in Lewy body dementias which correlated with higher pre-death Hoehn and Yahr scores and with longer Parkinson’s disease duration before dementia onset, but not with dementia severity, cortical Lewy body burden, or amyloid plaque and neurofibrillary tangle burden. Our study suggests that GluA2/3/4 losses may be a neurochemical marker of motor disability in Lewy body dementias.
KeywordsAMPA receptorsLewy body dementiaNeocortexMotor functionGlutamatergic system
Dementia with Lewy bodies (DLB) and Parkinson’s disease (PD) dementia (PDD) between them account for 10–15 % of the 35 million people worldwide with dementia (Aarsland et al. 1996, 2003; McKeith et al. 2005) and are both associated with underlying Lewy body disease pathology. These two forms of Lewy body dementias (LBD) are characterized clinically by parkinsonism and a dementia syndrome with predominant attentional, visuospatial and executive dysfunction and relatively preserved memory (Ballard et al. 2001a; McKeith et al. 2005). Additional key symptoms are visual hallucinations (McKeith et al. 1996, 2005), cognitive fluctuations (Ballard et al. 2001b), and sleep disturbances such as excessive daytime sleepiness and REM-sleep behavioral disorder (Boeve et al. 2003; McKeith et al. 2005).
The distinction between DLB and PDD as defined within the operationalized clinical criteria for DLB depends on the duration of parkinsonism prior to dementia. An arbitrary cut-off of 1 year was chosen in the original consensus criteria for the operationalized clinical diagnosis of DLB (McKeith et al. 1996), and thus PDD was diagnosed if dementia occurred more than 1 year after onset of parkinsonism, whereas dementia prior to, or within 1 year after the onset of parkinsonism, was classified as DLB. The third report of the DLB Consortium (McKeith et al. 2005) revised the operationalized diagnostic criteria for DLB and highlighted the unresolved issues in the relationship between DLB and PDD by emphasizing the overall clinical (Aarsland et al. 2009; McKeith et al. 2005) and pathological similarities (Ballard et al. 2006) of the conditions, but at the same time maintaining the arbitrary 1-year rule for distinguishing the two syndromes for research studies. The importance of further research to resolve “boundary issues” was also highlighted. Several key conceptual questions still remain unresolved; for example, are these conditions distinct or part of the same spectrum? If they are distinct conditions, is the arbitrary 1-year rule a meaningful distinction between clinical entities with different clinical presentations? Does the 1-year rule reflect measurable differences in brain changes? Addressing these issues is critical to take forward our understanding of this spectrum of conditions; establishing biological markers, determining prognostic indicators and, most importantly, in designing appropriate intervention studies and developing treatment paradigms across the dementias associated with cortical Lewy bodies.
The neurochemistry of LBD is one aspect requiring further study to help address the issues mentioned above. In other neurodegenerative dementias like Alzheimer’s disease (AD), losses of both pre- and postsynaptic markers as well as deficits of several transmitter systems are thought to underlie cognitive and non-cognitive behavioural features (Francis et al. 2010). Within the glutamatergic system, for example, the ligand-gated, ionotropic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors are assembled from heterotetramers of GluA1–A4 subunits and mediate the fast excitatory synaptic transmission known to be critical for learning and cognition (Malinow and Malenka 2002) Furthermore, the GluA2 subunits undergo mRNA editing which regulates Ca2+ permeability across AMPA receptors and may have a role in preventing excitotoxicity (Isaac et al. 2007; Kim et al. 2001). In AD and subcortical ischemic vascular dementia (SIVD), GluA2 is reported to be altered and may underlie certain clinical features of those diseases (Carter et al. 2004; Gong et al. 2009; Mohamed et al. 2011). Previous neuroimaging studies suggest disturbances of glutamate and glutamine in basal ganglia associated with motor symptoms (Modrego et al. 2011), as well as cortical glutamate deficits in non-demented PD and PDD (Griffith et al. 2008a, b). However, it is not clear whether cortical AMPA receptors are involved in cognitive and non-cognitive features of LBD, even with increasing evidence of dysregulated glutamatergic transmission in the basal ganglia of PD with treatment-induced dyskinesias (Sgambato-Faure and Cenci 2012), and proposals for the use of AMPA and other glutamatergic receptor ligands in PD pharmacotherapy (Johnson et al. 2009). In this study, we compared the immunoreactivities of AMPA receptor GluA2, A3 and A4 subunits in the postmortem neocortex of a cohort of prospectively assessed DLB and PDD patients to test the hypotheses that (1) GluA subunits are altered in LBD; and (2) that GluA subunit alterations can neurochemically distinguish clinically defined DLB from PDD. We also investigated potential associations of the neurochemical measures with cognitive and non-cognitive clinical variables.
Materials and methods
Patients, clinical and neuropathological assessments
Demographic and disease variables in a cohort of control and Lewy body dementias
Age at Death, years
80.0 ± 6
80.3 ± 3
80.4 ± 1
3 M/4 F
5 M/6 F
9 M/7 F
Postmortem Interval, hours
42.4 ± 11
37.6 ± 6
36.8 ± 5
6.25 ± 0.1
6.56 ± 0.08
6.36 ± 0.07
MMSE score at pre-death
11.7 ± 3
9.9 ± 2
Neocortical Lewy body (LB) scoreb
11.5 ± 1
11.8 ± 2
PD duration before dementia, years
−1.5 ± 0.7c
9.0 ± 1.1*
Hoehn and Yahr score at pre-death
1.6 ± 0.7
4.7 ± 0.2*
Senile plaque burdend,e
1 (11 %)
2 (15 %)
1 (11 %)
4 (31 %)
4 (44 %)
4 (31 %)
3 (23 %)
3 (33 %)
0 (0 %)
0 (0 %)
4 (44 %)
8 (62 %)
3 (33 %)
4 (31 %)
2 (22 %)
1 (7 %)
Frozen brain tissues from temporal cortex (Brodmann area BA36) were sourced from the Newcastle Brain Tissue Resource and the Brains for Dementia Research (BDR) network. All neuropathological evaluations were performed blind to clinical assignment by two neuropathologists at Newcastle. Lewy bodies were assessed by α-synuclein immunostaining and semi-quantitatively rated on a 0–20 scale which incorporated additional rating points within the “neocortical” range of the DLB consensus criteria (Ballard et al. 2006; McKeith et al. 1996). Coexisting cerebrovascular pathology was assessed based on findings of micro infarcts, lacunae, white matter and small vessel disease in subcortical structures, and none in the present cohort had concurrent SIVD based on operationalized diagnostic criteria (Kalaria et al. 2004). To assess AD pathology, Braak staging was performed to assess neurofibrillary tangle involvement (Braak and Braak 1991), while amyloid senile plaque burden was determined in areas of maximum plaque density and semi-quantitatively rated as “none”, “sparse”, “moderate” and “frequent” based on CERAD guidelines (Mirra et al. 1991). In the present cohort, only three subjects (all DLB) had sufficient pathology to meet CERAD diagnostic criteria for coexisting AD (Mirra et al. 1991).
Tissue processing for immunoblotting
All chemicals used in this study were of analytical grade and from Sigma-Aldrich Co. (USA) unless otherwise stated. Blocks of brain tissues previously stored at −80 °C were thawed on ice, dissected free of white matter and meninges, then homogenized in lysis buffer (Cell Signaling Technology Inc., USA) at a tissue concentration of 50 mg/mL with the addition of 2 mM 4-(2-aminoethyl) benzene sulfonyl fluoride hydrochloride (AEBSF) and Complete Mini™ protease inhibitor tablet (Roche Diagnostics, USA). Measurements of protein were performed using Coomassie Plus reagent (Pierce Biotechnology Inc., USA). Homogenates were then added to Laemmli sample buffer (Bio-Rad Laboratories, USA) at 1:1 vol./vol. and boiled at 95–100 °C for 5 min for immunoblotting. Not all assays were performed for all cases due to limited tissue availability.
Immunohistochemical detection of polyclonal anti-GluA2 and anti-GluA3 (both from Santa Cruz, USA), anti-AMPA receptors which recognizes GluA2/3/4 (Cell Signaling Technology Inc., USA), and anti-cathepsin D (Santa Cruz Biotechnology, USA) was undertaken employing standard procedures. Frontal cortical wax sections (seven microns) of aged controls (one male, 85 years, 46 h postmortem delay; one female, 84 years, 9 h postmortem delay) were de-waxed in xylene followed by washes through descending concentrations of alcohols into water. Thereafter, antigen retrieval was carried out by heating the slides in a microwave oven (1 × 10 min at 900 w followed by 1 × 10 min at 300 w) in 10 mM citrate buffer, pH 6.0. GluA2 and A3 antibodies were applied at a dilution of 1:100, GluA2/3/4 antibody was used at 1:50 dilution, while cathepsin D antibody was used at 1:1,000 dilution. The AMPA subunit antibodies were incubated with sections adjacent to those incubated with cathepsin D antibody. Further development of antibody binding employed appropriate biotinylated secondary antibodies, avidin–biotin complex and DAB (3,3′-diaminobenzidine) kits (Vector Laboratories, UK). Images were captured on a Leica DMRB microscope equipped with DFC420 camera.
Prepared brain homogenates in Laemmli buffer were electrophoretically separated on 10 % polyacrylamide gels, transferred onto nitrocellulose membranes, and blocked in 10 mM phosphate buffered saline, pH 7.4 with 0.1 % Tween 20 and 5 % skim milk (PBSTM) before blotting with primary antibody (GluA2, A3 and A2/3/4, all 1:1,000 dilution) in PBSTM overnight at 4 °C, followed by washings in PBST and incubation with appropriate horseradish peroxidase conjugated secondary antibodies (1:5,000, Jackson Immuno Research Inc., USA). One lane in each membrane consisted of fixed amounts of brain homogenate from a specific case to act as external control for normalization of immunoreactivities across separate membranes. Immunoreactive bands on the membranes were detected by enhanced chemiluminescence and quantified by an image analyzer (UVI tec, UK). Membranes were stripped and re-blotted with anti-β-actin (A1978, mouse monoclonal from Sigma to Aldrich Ltd., 1:5,000 dilution) as loading control. Normalized immunoreactivities were expressed in arbitrary units.
Data were analyzed using SPSS statistics software version 13 (SPSS Inc., USA). Normality of data was first checked using the Kolmogorov–Smirnov tests. Comparisons of group means were performed by one-way analyses of variance (ANOVA) followed by Bonferroni post hoc tests, or by Kruskal–Wallis ANOVA as appropriate. Student’s t tests or Mann–Whitney U tests were used to compare variables applicable to the dementia groups, or to compare controls with combined LBD group. Inter correlations were analyzed by Pearson’s product moment or Spearman’s rank correlation. Results were considered statistically significant if p < 0.05.
Demographic and disease variables of controls and LBD
Demographic and disease variables in the three study groups are listed in Table 1. Age, postmortem interval and cortical pH (as an indicator of tissue quality, Monoranu et al. 2009) were not significantly different among the groups (ANOVA with post hoc Bonferroni, p > 0.05). Table 1 also shows that the severity of dementia (indicated by pre-death MMSE scores), as well as neocortical Lewy body burden was closely matched between DLB and PDD, in agreement with previous findings (Jellinger 2009). PD progression and motor disability (indicated by pre-death Hoehn and Yahr scores) and duration of PD before dementia were expectedly higher in PDD than in DLB. Furthermore, senile plaques and neurofibrillary tangles were common features in the patients, of which DLB seemed to have a higher burden than PDD as indicated by higher percentages of available DLB cases with “moderate” to “abundant” plaques, or with Braak V/VI tangle scores (Table 1).
GluA subunit immunohistochemistry
GluA subunit immunoblotting in controls and LBD: correlations with disease and clinical variables
Summary of correlation coefficients of GluA immunoreactivities and disease variables in a cohort with Lewy body dementias
Variable (available N)
MMSE score at pre-death (24)
Neocortical Lewy body score (24)
PD duration before dementia, years (24)
AMPA receptors are well-known to mediate and regulate synaptic processes involved in learning, memory and behavior, and have been shown to be affected in neurodegenerative dementias. For example, GluA2 and A3 are reported to be reduced in AD, where GluA2 loss has been speculated to contribute to Ca2+ mediated excitotoxicity due to its physiological role in limiting Ca2+ influx (Carter et al. 2004; Gong et al. 2009). On the other hand, GluA2 is found to be upregulated in subcortical ischemic vascular dementia and correlated with milder dementia, which we interpreted as a neuroprotective synaptic plasticity process related to decreased Ca2+ mediated excitotoxicity (Mohamed et al. 2011). In this study, we extend investigations of AMPA receptors to a cohort of LBD (DLB or PDD) patients, and found moderate losses of GluA subunits which correlated with pre-death Hoehn and Yahr scores as well as duration of PD before dementia.
Interpretation and limitations of data
We report here that immunoreactivities of all three antibodies (GluA2-specific, A3-specific, and one which recognized GluA2/3/4 subunits) are reduced in LBD neocortex compared to controls. Because the GluA2/3/4 antibody recognizes GluA2 and A3 in addition to A4, the extent of contribution by GluA4 to changes in GluA2/3/4 immunoreactivities is at present unclear. Although further subgroup analyses indicated that GluA3 was decreased in DLB and PDD, and GluA2/3/4 was decreased only in PDD, the comparable direction of change in immunoreactivities suggests a general loss of GluA2/3/4 in LBD rather than subunit-specific changes (Fig. 2b). Taken as a whole, losses of GluA subunits may be more severe in PDD than in LBD (Fig. 2b), with PDD also manifesting more severe motor disabilities (Table 1). This suggests that GluA alterations may potentially distinguish clinically defined PDD from LBD neurochemically. However, follow-up studies of larger cohorts are needed to confirm this. Furthermore, while we were able to show that a subset of cathepsin D-positive cells also stained for GluA2, A3 and 2/3/4, supporting a neuronal labeling for these antibodies (Fig. 1), it is possible that GluA expression in other cell types (e.g., astrocytes and microglia) may also contribute to the observed neurochemical changes, and further studies are needed.
Because coexisting brain pathologies may contribute to the observed changes, we also assessed the presence of vascular and AD pathology in this cohort. While none of the subjects had sufficient vascular changes to be diagnosed as with SIVD (Kalaria et al. 2004), three DLB subjects met diagnostic criteria for concurrent AD (Mirra et al. 1991). Therefore, we further studied the effects of AD pathology (senile plaque load and Braak staging) on GluA immuno reactivities, and found that plaque and tangle burden did not have further effects on GluA2, A3 and A2/3/4 levels in the LBD cohort. This suggests that whilst both AD and LBD showed decreased AMPA receptor subunits, the effects may not be additive in comorbid diseases. Taken together, our data suggest that GluA2/3/4 loss is a neurochemical marker for motor disability and duration of PD in LBD.
The putative mechanisms by which neocortical GluA2/3/4 alterations may underlie motor disability in LBD are currently unclear. Multiple areas of the cerebral cortex project to the basal ganglia via glutamatergic corticostriatal pathways which interact with the nigrostriatal dopaminergic system to regulate motor and cognitive functions (Strafella et al. 2005), and abnormalities of this glutamate–dopamine interaction are thought be involved in PD pathophysiology (Calabresi et al. 2000; Carlsson and Carlsson 1990). For example, studies on various animal models of PD have shown both increased corticostriatal glutamatergic activation as well as increased striatal neuron excitability (Lindefors and Ungerstedt 1990; Campbell and Bjorklund 1994; Calabresi et al. 2000; Gainetdinov et al. 2001). Interestingly, in an acute mouse model of PD, Gainetdinov et al. (2001) showed that treatment with blockers of glutamate N-methyl-D-aspartate (NMDA) receptor-mediated transmission both exacerbated hyperactive behaviors as well as prevented the inhibitory effects of psychostimulant and serotonergic drugs on hyperactivity. Given that AMPA receptors function to potentiate NMDA receptor activation by increased excitatory postsynaptic potentials and removal of Mg2+ block (Malinow and Malenka 2002), we postulate that motor disability in LBD is likewise related to hampered NMDA receptor function due to GluA2/3/4 losses and associated AMPA receptor hypofunction. Alternatively, the observed GluA subunit changes may simply reflect an adaptive synaptic plasticity arising from long-term corticostriatal dopamine–glutamate imbalance. Another possibility would be the involvement of a common factor e.g., cortical Lewy bodies (LB) which may both affect GluA levels as well as be associated with PD progression, although this was not supported by our data which showed that LB scores did not significantly correlate with either GluA immunoreactivities or pre-death Hoehn and Yahr scores. However, this lack of correlation may also be attributable to (a) a relatively small cohort with a narrow range of cortical LB scores; and (b) the involvement of brain regions and receptors other than those under study. Therefore, to further delineate the potential role of AMPA receptor changes in the clinical features of Lewy body dementias, follow-up studies on larger cohorts and multiple brain regions are needed, including those involved in motor circuits. More comprehensive measurements of motor disability (e.g., motor sub-score of revised Unified Parkinson’s Disease Rating Scale, Goetz et al. 2007) may improve the accuracy and sensitivity of clinical-neurochemical correlations. Potential changes in other subunits (e.g., GluA1) and receptors (e.g., NMDA) at both mRNA and protein levels should also be the areas of active research. Lastly, it may be worthwhile to compare the current findings with further studies using non-demented Lewy body diseases, for e.g., non-demented PD.
Using a clinically well-characterized cohort of patients with Lewy body dementias (DLB and PDD), we report moderate losses of AMPA receptor GluA2/3/4 subunits which correlated with pre-death Hoehn and Yahr scores and PD to dementia interval. In this cohort, GluA2/3/4 immunoreactivities did not correlate with pre-death MMSE scores or cortical LB scores; and also did not seem to be affected by senile plaque or neurofibrillary tangle burden. This suggests that GluA subunit changes in LBD may be a neurochemical marker of severity or chronicity of parkinsonism and related to alterations arising from corticostriatal dopamine–glutamate imbalances. This clinical-neurochemical association appears to be corroborated by the finding that although clinically defined PDD and DLB are two aspects of the same entity (LBD) and share similar brain changes (e.g., GluA deficits), PDD appears to be characterized by higher GluA2/3/4 losses, and also manifests more severe motor disability and longer duration of parkinsonism. Given recent proposals to use AMPA receptor antagonists in PD, especially for levodopa-induced dyskinesia (Johnson et al. 2009), our data suggest both potential and caution in extending such therapeutic rationales to LBD in view of altered levels of receptor drug targets; as well as indicate further studies to elucidate the relationships between parkinsonism, dementia and GluA neurochemical alterations. Therefore, we reiterate the importance of focusing on interactions between the neurochemical, genetic and demographical basis of clinical phenotypes in LBD (Aarsland et al. 2009).
This study is supported by the National Medical Research Council of Singapore with a centre grant to CP Chen and MKP Lai (NMRC/CG/NUHS/2010), as well as a clinician scientist award to CP Chen (NMRC/CSA/032/2011). In the UK, this study is supported by the UK NIHR Biomedical Research Unit on Lewy Body dementias award to the Newcastle upon Tyne Hospitals NHS Foundation Trust. In Norway, this study is supported by the Norwegian Research Council and Western Norway Health Trust. The authors would like to acknowledge Drs Robert Perry and Johannes Attems for the neuropathological assessments. MKP Lai would like to thank AS Yeong and LP Loh for useful discussions.