Muscarinic M1 Receptor Coupling to G-protein is Intact in Parkinson’s Disease Dementia

Patients, clinical and neuropathologic assessments

Postmortem tissues from the frontal cortex (Brodmann Area BA9) of 14 PDD patients who formed part of prospective studies in Newcastle, UK were obtained for this study, along with 10 AD + CVD and 14 aged controls from the Newcastle Brain Tissue Resource and London Neurodegenerative Diseases Brain Bank, now part of the UK Brains for Dementia Research network. Longitudinal cognitive assessments, including the Mini-Mental State Examination (MMSE, [10] and medication histories were recorded. None of the AD + CVD patients were taking cholinesterase inhibitors (ChEI) or memantine, while only three PDD patients were on ChEI. Informed consent had been obtained from patients’ next-of-kin prior to the removal of brain, and approval for the study had been granted by research ethics committees of Newcastle upon Tyne Hospital Trust, UK and National University Health System, Singapore. Diagnosis of PDD was based on operationalized neuropathological as well as clinical criteria [11]. Diagnosis of AD + CVD was based on fulfillment of AD diagnosis by the CERAD criteria [12] as well as neuropathological findings of micro- and / or lacunar infarcts and white matter lesions in the subcortical structures without large vessel involvement [13]. Selection of subjects for the study was based on tissue availability and fulfillment of diagnostic criteria as outlined above, and not on Aβ load or any other a priori neurochemical criteria. Control subjects had unremarkable neuropathological findings and no history of neurological or psychiatricdiseases.

Tissue processing

All chemicals and reagents are of analytical grade and purchased from Sigma-Aldrich (St Louis, MO, USA) unless otherwise-stated. Previously fresh frozen brain samples (∼1 cm³) were thawed on ice, dissected free of meninges and white matter and then homogenized using an Ultra-Turnax homogenizer (IKA, Staufen im Breisgau, Germany) on maximum setting, 10 s in 50 mM Tris-HCl buffer, pH 7.5 at 50 mg tissue wet weight / mL. Homogenates were centrifuged for 10 min, 20000 g at 4°C to obtain crude brain membrane pellet, then washed by further resuspension in buffer and centrifugation before storage at –80°C. Aliquots of the homogenate was further treated with 5M Guanidine HCl in 50 mM Tris-HCl, pH 8.0 buffer for enzyme-linked immunosorbent assays (ELISA). Due to limited tissue availability, not all assessments were performed for the entire cohort, and actual N will be stated in the table or figure legends where they differed from cohort N.

Aβ₄₂ ELISA

All assays were performed blind to clinical information. Guanidine / Tris-HCl-treated homogenates were assayed in duplicates using a commercial Aβ₄₂ ELISA kit according to manufacturer’s instructions (Life Technologies, Carlsbad, CA, USA). Measurements were done with a microplate reader (BioTek, Winooski, VT, USA) and expressed in ng / mL.

Muscarinic M1 receptor binding assay

M1 receptors in the brain homogenates were measured as previously described [5, 6] using saturation binding with [³H]pirenzepine (specific activity 84 Ci/mmol, Perkin-Elmer Life Sciences, Waltham, MA, USA). Briefly, eight concentrations of the radioligand (ranging from 0.5–10 nM) were incubated in triplicates with thawed brain homogenate (diluted 1:5) in 50 mM sodium phosphate buffer, pH 7.4 for 60 minutes at 25°C, with 10 μM unlabeled atropine sulphate added to a parallel set of assays to determine non-specific binding. Protein concentrations were measured by a Coomasie Blue assay kit (Thermo Fisher Scientific, Waltham, MA, USA). At the end of incubation, bound radioligand was separated by vacuum filtration through 0.01% polyethylenimine-treated Whatman GF-B glassfibre filters (Whatman, Maidstone, UK) followed by washings with ice-cold phosphate buffer. Radioactivity retained on the filters was measured by liquid scintillation spectrophotometry using a Wallac beta counter (Perkin-Elmer Life Sciences, Waltham, MA, USA). Binding affinity constant (K_D, in nM) and receptor density (B_max, in fmol / mg protein) values were obtained by non-linear regression analyses using the Prism software (version 5.0, GraphPad, San Diego,CA, USA).

M1-G-protein coupling assay

M1-G-protein coupling status was measured using a [³H]pirenzepine-carbachol competition assay based on published methods [5]. Briefly, brain homogenates were incubated in duplicates with 3 nM [³H]pirenzepine and increasing concentrations of carbachol (10^- 9 to 10^- 2 M) in buffer (20 mM Tris–HCl, 1 mM MnCl₂, pH 7.4) for 150 minutes at room temperature. For each binding assay, a parallel series of tubes were set up with the addition of 0.2 mM guanylyl imidodiphosphate (GppNHp), a non-hydrolysable guanosine triphosphate analogue. Non-specific binding was determined in the presence of 10 μM atropine sulphate. Assays were terminated and processed for liquid scintillantion spectrophotometry as described above. IC₅₀ values of specific binding from the [³H]pirenzepine-carbachol competition curves were converted to inhibitory constants for assays with (K_iG) and without (K_i) GppNHp using the Cheng-Prusoff equation, and the ratio of K_iG to K_i (K_iG / K_i) used as a measure of M1-G-protein coupling.

Statistical analyses

Data analyses were performed using SPSS Statistics software (version 21, IBM, Armonk, NY, USA). Variables were first tested for normality to determine the use of parametric or non-parametric tests. Comparisons of demographic and disease variables between controls and diseased groups were performed by one-way analysis-of-variance (ANOVA) with Bonferroni post-hoc tests, or Kruskal-Wallis ANOVA with Dunn’s post-hoc tests. Correlations between variables were likewise tested with either Pearson product moment or Spearman’s rank correlation tests. Results were considered statistically significant if p < 0.05.

Demographic and disease variables

Control and dementia subjects were matched for age and postmortem delay, while the PDD and AD + CVD had similar pre-death MMSE scores, used as an indicator of dementia severity (Table 1). There was a lower proportion of males in the control group, but further analyses showed that none of the neurochemical variables measured were significantly different between males and females, both within each study group and overall (data not shown). For Braak staging of neurofibrillary tangle involvement [14], none of the controls had Braak stage of > II, while only four PDD patients exhibited Braak stage III / IV, with the rest between I – II. In contrast, three AD + CVD patients had Braak stage IV, with the rest between V–VI. Similarly, Aβ₄₂ concentrations were significantly higher in AD + CVD compared to both control and PDD, while Aβ₄₂ in PDD were not significantly different from controls (Fig. 2C). Incidentally, Aβ₄₂ concentrations were below the detectable limit of the assay kit (0.01 ng/ml) in 11 samples (8 controls and 3 PDD), which were assigned the value of 0.01 ng/mL in our analyses. Excluding these samples did not alter the statistical significance of the group differences (Aβ₄₂ concentrations of controls [in ng / mL] = 15.69±8; AD + CVD = 50.3±8; PDD = 6.45±1, Kruskal-Wallis ANOVA with Dunn’s post-hoc p < 0.05 for all pairwise comparisons with AD + CVD, and p > 0.1 for control vs. PDD). Taken together, these results are consistent with a high AD pathological burden in AD + CVD, but not in PDD.

Muscarinic M1 receptor binding densities and coupling to G-proteins

Representative [³H]pirenzipine-carbachol competition curves are shown in Fig. 1. In both the control and PDD subjects, there was a noticeable “right shift” of the curve with the addition of GppNHp, suggestive of a native, coupled state between M1 receptors and G-proteins. In contrast, the absence of a shift with GppNHp in AD + CVD is indicative of an uncoupled state [6]. Consistent with this observation, Fig. 2B shows that mean K_iG / K_i values were significantly reduced only the AD + CVD group. In contrast, receptor densities (B_max) of AD + CVD and PDD were unchanged relative to control (Fig. 2A). Lastly, mean±S.E.M K_D values were also unchanged (control 5.8±0.3 nM, N = 10; AD + CVD 5.9±0.3 nM, N = 6; PDD 5.2±0.2 nM, N = 13; one-way ANOVA, p = 0.164).

Correlations between neurochemical and disease variables

Because subjects with high Aβ₄₂ burden also showed reduced M1 receptor coupling to G-proteins, we performed further correlational analyses for the neurochemical and other variables within our dementia subgroups (AD + CVD and PDD). To prevent data skewing, we excluded the three PDD subjects whose Aβ₄₂ concentrations were below detection limits from the correlational analyses (see above). Figure 3 shows that Aβ₄₂ positively correlated with [³H]pirenzipine B_max values, as well as negatively correlated with K_iG / K_i values. In contrast, demographic and disease factors such as age (rho = –0.04), postmortem interval (rho = –0.16) as well as Braak staging (rho = –0.40) did not correlate with K_iG / K_i (Spearman p > 0.05).

Clinical implications

Gα_q/11-coupled mAChR like M1 receptors have well established roles in learning and memory, and perturbations of these receptors are thought to underlie many of the cognitive and behavioral features of AD [1]. We now show that unlike AD (as previously reported) and AD + CVD, muscarinic M1 receptor coupling to G-proteins is intact in PDD which manifested generally lower amyloid burden. One clinical implication of this finding is that cholinergic replacement therapies which were evaluated, or currently approved for, AD can potentially be repurposed for PDD, and may show improved efficacy due to preserved signal transduction (e.g., protein kinase C [6])downstream of muscarinic receptors. Furthermore, our data suggest that M1 agonists and positive allosteric modulators previously being evaluated for AD therapy [15, 16] may show greater efficacy for PDD as well. This postulate appear to be supported by a few clinical trials on the use of cholinesterase inhibitors (ChEIs) for PDD [17, 18], although another meta-analysis of ChEI trials on PDD have shown only modest improvements in cognitive test scores [19]. Therefore, further evaluations of cholinergic replacement therapies in PDD is required before the reported positive outcomes can be generalized. Conversely, our data suggest that AD + CVD or mixed dementia patients have reduced M1-G-protein coupling similar to AD (see [4–6]), and therefore may show similar, limited efficacy with conventional cholinergic replacement therapies.

Study limitations and future directions

Our first study of M1-G-protein coupling status in PDD and AD + CVD showed that AD + CVD had reduced M1-G-protein coupling and relatively high Aβ₄₂ load, while PDD showed intact M1 coupling to G-protein and lower Aβ₄₂ load. Whilst the finding of low Aβ₄₂ in PDD as a group corroborates previous studies [9, 20], it is important to recognize that Aβ₄₂ burden may be quite heterogeneous within this group [21, 22]. Indeed, if our postulate that Aβ₄₂-associated M1-G-protein uncoupling predicts response to cholinergic therapies is borne out, the heterogeneity in Aβ₄₂ or amyloid plaque burden in PDD may well underlie the variability of clinical response to ChEIs as noted in the previous paragraph. Therefore, follow-up studies in a larger cohort of PDD with varying Aβ₄₂ are needed. Similarly, whilst significant amyloid deposits may be apparent in cognitivelynormal elderly people [23, 24], cholinergic functional status is unclear in these individuals with high amyloid load. Interestingly, Potter et al. [25] has shown that non-demented elderly subjects with high amyloid plaque load had significantly reduced M1-G-protein coupling compared to those without, and follow-up studies will help determine whether early cholinergic therapies may be beneficial for people with high amyloid burden in preclinical stages of dementia.

The mechanisms underlying the preservation of M1-G-protein coupling in PDD but reduced coupling in AD + CVD is at present unclear. It may be related to cortical Aβ₄₂ burden, which we have found to be significantly higher in AD + CVD compared to PDD patients who had levels similar to controls (Fig. 2C). Indeed, we showed in the combined dementia group that cortical Aβ₄₂ correlated negatively with the extent of M1-G-protein coupling (defined by K_iG / K_i, see Fig. 3C), though whether Aβ₄₂ regulates M1 receptor sensitization indirectly via its effects on G-protein coupled receptor kinases (GRK2/5, [26], or directly by the sequestration of M1-associated Gα_q/11 subunits, or by other, as yet uncharacterized mechanisms, remain to be investigated. For example, whilst M1 receptor densities (B_max) were unchanged between the groups, the positive correlation between B_max and Aβ₄₂ may reflect neural plasticity events leading to M1 up-regulation or changes in binding properties in response to Aβ₄₂. However, follow-up studies are required to confirm which, if any, of the proposed mechanisms are relevant in human disease. Additionally, M1 neurochemical parameters need to be assayed in more brain regions to gauge the general applicability of our findings. Furthermore, the state of muscarinic receptor-G-protein coupling, and the effects of Aβ₄₂ as well as aggregated α-synuclein in Lewy body diseases need to be more comprehensively studied. For example, a loss of muscarinic M4 autoreceptor-G-protein coupling has been described in PD striatum which results in the exacerbation of motor symptoms due to a loss of inhibitory signals for acetylcholine release [27], and it would be of interest to see whether such receptor uncoupling may be more widespread in both regional and receptor subtype involvement, and whether these disease processes underlie the loss of neocortical M1 uncoupling currently observed in PDD. Lastly, it would be worthwhile to study M1 receptor coupling status in other dementias with variable amyloid burden. For example, a previous study combining amyloid imaging with ChEI treatment showed that dementia with Lewy body (DLB) subjects with minimal amyloid deposits responded better than subjects with concomitant amyloid pathology [28], and it would be of interest to examine whether such differences could be due to variations in M1-G-protein coupling.

Conclusion

We report here the intact coupling of frontal cortical muscarinic M1 receptors to their G-proteins in PDD with low Aβ₄₂ load, in contrast to decreased M1-G-protein coupling in AD with significant subcortical CVD, which showed high Aβ₄₂ load. Taken together, the existing data suggest a negative correlation between amyloid burden and muscarinic receptor-G-protein coupling in both AD and non-AD neurodegenerative dementias. Our study supports the trial of cholinergic replacement or M1 receptor-targeting pharmacotherapies for PDD and other neurodegenerative dementias which have low Aβ₄₂ burden.