Introduction

The neuropathologic hallmarks of Alzheimer’s disease (AD) include β-amyloid (Aβ) containing neuritic plaques, neurofibrillary tangles arising from paired helical filaments of hyperphosphorylated τ proteins, and neuronal degeneration across multiple transmitter systems. In particular, losses of cholinergic as well as serotonergic neurons are thought to underlie much of the neurochemical perturbations found in the AD cortex and contribute to the development of cognitive and behavioral symptoms (Cummings and Back 1998; Francis et al. 1999; Lanari et al. 2006; Lanctot et al. 2001). Dysfunction of acetylcholine and serotonin G-protein coupled receptors, especially those which associate with Gαq/11-subtypes which in turn activate phopholipase C, calcium mobilization and protein kinase C, has been reported in AD. For example, muscarinic M1 receptors are found to be uncoupled to their G-proteins (Flynn et al. 1991; Tsang et al. 2006), while specific reductions of serotonin 5-HT2A receptors occur in the neocortex (Cross et al. 1986; Lai et al. 2005). Given their roles in emotive and behavioral processes, memory and cognition, as well as in intracellular signaling events which regulates amyloidogenic processing of amyloid precursor protein (APP) and τ phosphorylation, neurochemical perturbations of Gαq/11-coupled receptors may be involved in both clinical and neuropathologic features of AD (Hellstrom-Lindahl 2000; Lai et al. 2005; Meneses and Hong 1997; Nitsch et al. 1992, 1996; Tsang et al. 2006, 2007). Besides cholinergic, serotonergic and other well studied neurotransmitter systems, there has been growing awareness of the neuropsychiatric and neurobiological relevance of various “non-classical” neurotransmitters like neuropeptides, nitric oxide and nucleotides, with their status in AD brain only just beginning to be investigated. For example, the large P2Y family of receptors which are activated by purine or pyrimidine nucleotides, are known to be expressed in diverse tissue types in humans and subserve multiple physiological functions (Abbracchio et al. 2006). Of these, the P2Y2 receptors are responsive to both ATP and UTP, while P2Y4 receptors recognize UTP, and P2Y6 is UDP-selective (Nicholas et al. 1996). P2Y2,4,6 receptors have been detected in both neuronal and glial cells in the CNS (reviewed in Hussl and Boehm 2006), where they mediate a variety of processes of potential relevance to AD. For example, P2Y2 receptor activation enhances the proteolytic processing of APP in a manner which precludes the formation of Aβ, the major constituent of neuritic plaques (Camden et al. 2005). P2Y2 receptors also mediate neuronal differentiation via nerve growth factor/TrkA signaling (Arthur et al. 2005), while P2Y2,4,6 receptors have been reported to have neuroprotective, antiapoptotic or proapoptotic functions, as well as the ability to modulate glutamate N-methyl-d-aspartate (NMDA) receptor currents or intracellular calcium (Calvert et al. 2004; Cavaliere et al. 2004, 2005; Kim et al. 2003; Lee et al. 2007; Wirkner et al. 2007). Given that the AD brain is characterized by formation and aggregation of Aβ, loss of trophic support, cell death and apoptosis, derangement of intracellular calcium-mediated signaling pathways, as well as glutamatergic deficits (Behl 2000; Cole et al. 1988; Counts and Mufson 2005; Francis 2003; Selkoe 2001; Tsang et al. 2007), it is reasonable to hypothesize that P2Y2,4,6 receptor dysfunction may be involved in the AD process. However, the status of P2Y2,4,6 receptors in the AD brain is at present unclear. Here, using postmortem materials from a cohort of longitudinally assessed AD patients as well as matched controls, we measured P2Y2,4,6 receptors in two brain regions known to be differentially affected by AD neuropathology (parietal and occipital cortex) and investigated the associations between receptor immunoreactivities and neuropathologic data.

Materials and methods

Patients, clinical and neuropathologic assessments

The study comprised of up to 29 AD patients and 12 elderly controls. The AD patients were derived from a cohort of community-living, clinically diagnosed dementia patients recruited into a longitudinal study of behavior in dementia based in Oxfordshire, UK (Hope et al. 1999). The inclusion and exclusion criteria, point of study entry characteristics, as well as natural history of the study cohort have previously been described in detail (Hope et al. 1997a, b, 1999). From study entry till death, cognitive functioning of the patients was assessed every 4 months with the mini-mental state examination (Folstein et al. 1975). The mean of up to five MMSE scores before death (MMSE5) was used as an indicator of dementia severity to avoid floor effects (Lai et al. 2001). Drug histories were recorded for all patients: eight were on neuroleptics, seven were on minor tranquilizers, and only two were given tricyclic antidepressants in the 8 months before death. No patient was on cholinergic replacement therapies. At death, informed consent was obtained from the patients’ next-of-kin for the removal of brain, and selection of subjects for this study is based on the CERAD neuropathologic diagnosis of AD (Mirra et al. 1991) as well as tissue availability. Only one patient in this study had evidence of mixed AD and vascular dementia. As an additional indicator of disease severity, paraffin-embedded blocks of temporal cortex were stained with methenamine silver and a modified Palmgren stain for a blinded, semi-quantitative rating (from 0, absent to 3, highest) of neuritic plaques (NP) and neurofibrillary tangles (NFT) as previously described (Jacobs et al. 2006). Patients with scores of 0–2 for NP or NFT were categorized under the ‘Low/Moderate’ NP or NFT rating group; while scores of 3 were in the ‘High’ group. The age-matched controls did not have dementia or other neurological diseases.

Tissue processing and brain membrane homogenate preparation

Gray matter from the angular gyrus in the parietal lobe (Brodmann Area, BA 39) and the primary visual cortex in the occipital lobe (BA 17) was homogenized and processed for immunoblotting as previously described with slight modifications (Kirvell et al. 2006). In brief, after dissection to remove meninges and white matter, aliquots of tissue was homogenized in ice-cold 50 mm Tris–HCl buffer (pH 7.4, with 5 mm EGTA, 10 mm EDTA, 2 μg/ml aprotonin, 1 μg/ml E64, 2 μg/ml leupeptin hemisulphate, 2 μg/ml pepstatin A and 20 μg/ml phenylmethylsulphonyl fluoride) using a Teflon–glass homogeniser, followed by storage at −70°C. Protein concentration was measured with the Coomassie Plus reagent (Pierce Biotechnology Inc., Rockford, IL). pH was also measured in the cerebellum as an indicator of agonal status (Hardy et al. 1985). When ready for immunoblotting experiments, brain homogenates were thawed and mixed with Laemmli sample buffer (Bio-Rad Laboratories, Hercules, CA) 1:1 v/v followed by boiling. Samples were not available from both regions for all cases.

Immunoblotting and Immunohistochemistry

Primary antibodies used in this study were anti-P2Y2, anti-P2Y4 and anti-P2Y6 (rabbit polyclonals from Alomone Labs Ltd, Jerusalem, Israel) and anti-β-actin (mouse monoclonal from Sigma-Aldrich Ltd.). Boiled brain homogenate were electrophoretically separated on 10% polyacrylamide gels, transferred onto nitrocellulose membranes, and blocked in 10 mM phosphate buffered saline, pH 7.4, 0.1% Tween 20/5% skim milk (PBSTM) before immunoblotting with primary antibody at the following dilutions in PBSTM: 1:200 (anti-P2Y2 and anti-P2Y6), 1:300 (anti-P2Y4), for 3 h at room temperature (RT) or overnight at 4°C. To assess specificity of antibody binding, certain membranes were pre-incubated with control peptide antigens for 1 h before incubation with primary antibodies. Following washings in PBSTM and incubation with horseradish peroxidase conjugated secondary antibodies (1:10,000, Jackson ImmunoResearch Inc., West Grove, PA), immunoreactive bands on the membranes were detected by enhanced chemiluminescence and quantified by an image analyzer (UVItec, UK). Membranes were then stripped and reblotted with anti-β-actin (1:5,000) to control for sample loading across lanes. One lane of external standard consisting of fixed amounts of protein from one homogenate sample was loaded for each membrane for normalization of data. Normalized immunoblot optical densities are expressed in arbitrary units. Immunoblotting data of anti-synaptophysin (1:20,000, Sigma Chemical Co., Poole, UK) using unboiled aliquots of the same homogenates as the P2Y studies have been previously reported (Kirvell et al. 2006). For immunohistochemical detection of P2Y2 staining, 10 µm paraffin wax sections of rat prefrontal cortex were dewaxed in xylene for 10 min followed by washes through a descending series of alcohols (100–50%) into water. Sections were then immersed in boiling 10 mM citric acid (pH 6) before being microwave pressure cooked for a further 10 min in the same buffer, followed by treatment in 0.3% H2O2 (v/v) in phosphate buffered saline (PBS) for 30 min, RT. Then, the sections were washed in PBS, pre-blocked in 1.5% normal goat serum (v/v)/0.1% bovine serum albumin (BSA) (w/v) in PBS, RT before overnight incubation at 4°C with anti-P2Y2 (1:200) diluted in 0.1% BSA/PBS. On the following day, sections were washed in PBS and further incubated in biotinylated goat anti-rabbit IgG (1:400, Vector labs, USA) for 1 h, RT. Sections were subsequently incubated in avidin–biotin complex (Vector Labs, USA) for 1 h prior to visualization using 0.06% DAB (3,3′-diaminobenzidine) (w/v) prepared in PBS together with 0.01% H2O2 (v/v). Sections were then thoroughly washed in running tap water and counterstained in hematoxylin before being dehydrated through an ascending series of alcohols (50–100%) into xylene and coverslips applied with DPX (dibutyl phthalate xylene) mountant. Slides were then viewed using a Carl Zeiss Axioskop microscope with JVC 3-CCD color camera model KY-F55B and images captured using Carl Zeiss KS300 Image Analysis Software Version 3.

Statistical analyses

Data were first tested for normality for the selection of parametric or non-parametric tests. Potential confounding effects of demographic or disease variables on P2Y receptor immunoblot data, as well as inter-correlations among variables, were studied by Pearson’s product moment or Spearman’s correlation. Demographic variables and P2Y receptor levels between controls and AD were compared with Student’s t tests, while comparisons of immunoblot data among controls, AD with ‘Low/Moderate’ neuropathologic rating, as well as AD with ‘High’ rating, were performed using one-way ANOVA followed by post hoc Scheffé tests.

Results

Specificity of P2Y2,4,6 receptor antibodies

The polyclonal antibodies to P2Y2, P2Y4, and P2Y6 receptors produced major bands at approximately 60, 75 and 60 kDa, respectively, on immunoblotting with control and AD brain homogenates, as well as with neuroblastoma SH-SY5Y cell lysate (Fig. 1), consistent with the manufacturer’s data sheets. Occasionally, there are additional, smaller molecular weight bands (30–50 kDa) detected by the P2Y2 antibody in both AD and control homogenates which may represent degradation products or isoforms of the receptor. Except for a weak correlation of the 50 kDa minor bands in BA17 with cerebellum pH (Pearson r = −0.42, p = 0.04), these minor bands did not correlate with markers of disease and tissue quality (NP, NFT, postmortem interval). Such smaller bands were not routinely seen for P2Y4 and P2Y6 blots. Because we are currently unable to reliably identify the minor bands, analyses were restricted to the major 60 kDa band for P2Y2 (see Fig. 2a). As shown in Fig. 1, the antibodies seemed to specifically detect their target receptors, since preincubation of the antibodies with peptides corresponding to unique sequences of the respective receptors (see Fig. 1 legend) inhibited much of the signal seen in parallel immunoblots without peptide preincubation.

Fig. 1
figure 1

Parallel lanes of brain homogenates from C control, A AD and S SH-SY5Y lysate were immunoblotted with anti-P2Y2, anti-P2Y4 and anti-P2Y6, with (+CA) or without (−CA) preincubation with equimolar control antigen peptides (P2Y2 residues 227–244: KPAYGTTGLPRAKRKSVR; P2Y4 residues 337–350: HEESISRWADTHQD; P2Y6 residues 311–328: QPHDLLQKLTAKWQRQRV). The −CA and +CA blots were run, blotted and visualized using enhanced chemiluminescence in parallel

Full size image
Fig. 2
figure 2

a Representative blots of immunoreactivity to P2Y2, P2Y4 and P2Y6 receptors in control (n = 9–10) and AD (n = 21–22) homogenates from the parietal and occipital cortex. Membranes were stripped and reblotted with anti-β-actin for normalization of protein loading. Lanes are C control, A AD, and E external standard. Indicative positions of molecular weight markers are in kilodaltons (kDa). For P2Y2, only the 60 kDa bands (indicated by arrow head) were analyzed (see text). b Bar graphs of mean immunoblot densities ± SEM (in arbitrary units). *Significantly different from controls. Student’s t test, p = 0.009

Full size image

Demographic, disease and P2Y2,4,6 immunoblot variables in control and AD

Demographic variables were matched between the AD subjects and controls (Student’s p > 0.05) except for lower pH in AD (Table 1), suggesting more severe brain acidosis in AD from prolonged agonal state (Hardy et al. 1985). However, pH, as well as other demographic variables listed in Table 1, did not correlate with P2Y2,4,6 immunoblot densities (Pearson p > 0.05). As expected, both NP and NFT scores were significantly higher in AD compared to controls (Table 1). NP and NFT scores were also highly correlated with each other (Spearman ρ = 0.915, p < 0.001), suggesting an absence of ‘plaque predominant’ or ‘tangle predominant’ neuropathology in this cohort. In the occipital cortex (BA17), P2Y2, P2Y4, and P2Y6 immunoblot densities were not significantly different between controls and AD, while in the parietal cortex (BA39), P2Y2 levels were significantly reduced in AD, with no change in P2Y4, and P2Y6 densities (Fig. 2b). Lastly, immunoblot densities did not differ significantly between patients who were administered neuroleptic or minor tranquilizers in the 8 months before death and those not on such medication (data not shown).

Table 1 Demographic and disease variables in a cohort of prospectively assessed AD patients and controls
Full size table

Association of P2Y2,4,6 immunoblot variables with neuropathologic and clinical features in AD

Of the receptors studied, only P2Y2 density in BA39 inversely correlated with NP (Spearman ρ = −0.69, p < 0.01) and NFT (Spearman ρ = −0.67, p < 0.01) ratings. P2Y4 and P2Y6 in BA39 did not correlate with neuropathologic scores, and all receptors in both regions did not correlate with disease duration or with dementia severity as indicated by MMSE5 (Spearman p > 0.05). Furthermore, among controls and neuropathologic subgroups of AD, P2Y2 densities in BA39 were significantly reduced only in patients in the ‘High’ rating group for NP and NFT (Fig. 3a). Lastly, P2Y2 densities in BA39 significantly correlated with previously reported synaptophysin immunoreactivity in BA39 in the subset of patients for whom both measures were available (Fig. 3b). Synaptophysin levels did not correlate with P2Y4 or P2Y6 densities (Pearson p > 0.05).

Fig. 3
figure 3

a Bar graphs of mean ± SEM normalized P2Y2 receptor immunoreactivity (in arbitrary units) in BA39 across controls (n = 9) and AD with ‘Low/Moderate’ (n = 8) or ‘High’ (n = 9) neuritic plaque (NP) and neurofibrillary tangle (NFT) ratings. b Scatter plot of normalized P2Y2 immunoreactivity with synaptophysin immunoreactivity in BA39 of available subjects (9 controls, 19 AD), both in arbitrary units. *Significantly different from control (one-way ANOVA with post hoc Scheffé p < 0.01). **Significant Pearson correlation

Full size image

Discussion

At present, one potential pharmacotherapeutic utility of P2Y receptor antagonists in AD has been proposed based on their efficacy in limiting ATP-mediated chronic inflammation responses as well as excitotoxicity arising from excessive activation of glutamate receptors (reviewed in Burnstock 2007). However, apart from P2Y1 receptors which have been immunohistochemically localized to neuritic plaques and neurofibrillary tangles (Moore et al. 2000), it is unclear whether other Gαq/11-coupled P2Y receptors are altered in the AD neocortex, and whether these alterations are associated with clinical or neuropathologic features of AD. Here, we report the status of P2Y2, P2Y4 and P2Y6 receptors in the AD neocortex as well as their correlations with dementia severity, neuritic plaque and neurofibrillary tangle burden. Interestingly, P2Y receptor deficits seem to be subtype-specific, as we found reductions of P2Y2 while P2Y4 and P2Y6 receptors remained unchanged. Additionally, P2Y2 deficits seemed to be associated with plaque and tangle burden, since the loss of P2Y2 immunoreactivity was restricted to the parietal cortex, whereas levels were preserved in the occipital cortex, a region considered to be minimally affected by AD neuropathology even in advanced stages of disease (Braak and Braak 1991; Lerch et al. 2005). Furthermore, when AD patients were grouped according to plaque and tangle severity, parietal P2Y2 loss was significant only in subgroups with ‘High’ ratings for NP and NFT, while patients labeled ‘Low/Moderate’ for NP and NFT had intermediate densities of P2Y2 (Fig. 3a). Moreover, decreases in parietal P2Y2 densities also correlated with lower synaptophysin immunoreactivity, a marker of synapse loss (Butler et al. 2007). Additionally, immunohistochemical staining in rat prefrontal cortex suggests that at least some of the P2Y2-expressing cell bodies and processes are derived from pyramidal neurons (Fig. 4), which are known to play important roles in cognitive processes and are vulnerable to neurodegeneration (Francis et al. 1993; Mann 1996; Pearson and Powell 1989). With the putative demonstration of P2Y2 localization in the cortex of normal rodents, it will be of interest to extend the study of the purinergic system to AD animal models in order to determine whether purinergic deficits result from pathophysiological processes which simulate AD, or whether manipulations of purinergic function would affect disease progression in the animal model.

Fig. 4
figure 4

a Immunohistochemistry of anti-P2Y2 in rat prefrontal cortex (4× magnification), the scale bar indicating 50 µm. b 40× Magnification of area indicated by the rectangle from a, arrows indicating some of the pyramidal neurons which were positive for P2Y2 immunoreactivity

Full size image

In contrast to NP and NFT scores, P2Y2,4,6 receptor immunoreactivity did not correlate with dementia severity, suggesting these receptors may play an indirect regulatory, rather than effector, role in processes mediating cognition and memory. For example, P2Y2 receptor activation has been shown to regulate the expression of acetylcholinesterase and acetylcholine receptors in the neuromuscular junction (Tung et al. 2004), although further studies are needed to determine if P2Y2 has similar effects on neurons. Taken together, these findings suggest a specific association between changes in P2Y2 receptor immunoreactivity and neuropathologic features in the AD parietal cortex.

What are the likely mechanisms underlying the effect of P2Y2 receptor loss on AD neuropathology? As P2Y2 activation has been reported to regulate proteolytic processing of APP toward the non-amyloidogenic pathway (Camden et al. 2005), a loss of P2Y2 may expedite the deposition of Aβ and the resultant aggregation of fibrillar Aβ and the formation of plaques, in a manner analogous to that proposed for cholinergic receptor deficits (Auld et al. 2002; Hellstrom-Lindahl 2000). Similarly, although there is as yet no evidence that P2Y2 receptors regulate the phosphorylation of τ or other microtubule-associated proteins, it is conceivable that Gαq/11-mediated signaling arising from P2Y2 activation may lead to decreased τ phosphorylation, as is shown for Gαq/11-coupled muscarinic receptors (Hellstrom-Lindahl 2000; Sadot et al. 1996). Furthermore, the correlation of P2Y2 deficit with synapse loss (as indicated by reduced synaptophysin immunoreactivity) suggest a dysfunction of the trophic support which may normally be mediated partly by P2Y2 receptors (Arthur et al. 2005; Gonzalez et al. 2005). Confirmation or refutation of these hypotheses will await further in vitro studies, but several limitations of the current study as well as further questions which arise need to be considered. Firstly, findings from the present correlational study cannot unequivocally demonstrate the role of P2Y2 changes in plaque and tangle pathogenesis, since it is also possible that P2Y2 receptor losses are a reflection of degenerating P2Y2-expressing cells after exposure to Aβ. For example, perturbations of ATP induced Ca2+ fluxes have been reported in Aβ-treated fetal microglia in vitro (McLarnon et al. 2005). However, the scenario of Aβ-induced P2Y2 deficit does not necessarily negate a potential role of P2Y2 deficit (which resulted from Aβ neurotoxicity) in further exacerbating amyloidogenic processing. Indeed, the hypothesis that Aβ and receptor dysfunction interact in a vicious cycle leading to widespread neurodegeneration deserves further study. Furthermore, unlike cholinergic deficits which can be traced to degeneration of acetylcholine producing neurons in the basal forebrain (Whitehouse et al. 1982), the pathogenesis of reduced P2Y2 immunoreactivity is unclear. It is also not known whether the selective reduction of P2Y2 receptors in AD parietal cortex is due to selective vulnerability of populations of P2Y2-bearing cells, or to specific downregulation or desensitization of P2Y2 receptors. Lastly, although the correlation of P2Y2 with synaptophysin immunoreactivity suggests perturbations in at least a proportion of neuronal P2Y2 receptors, we are unable to conclude in the current study whether, or to what extent, glial P2Y2 receptors are affected. Further studies employing quantitative immunohistochemical methods may be needed to address these questions.

In conclusion, in this first study to investigate the status of uridine nucleotide-activated P2Y2,4,6 receptors in the postmortem neocortex of a cohort of well-characterized AD patients, we found selective reductions of P2Y2 immunoreactivity in the parietal, but not occipital, cortex. We further showed that the loss of parietal P2Y2 immunoreactivity correlated with neuritic plaque and neurofibrillary tangle burden. Although future studies are needed to further characterize the P2Y2 changes, such as the cells types involved, the present results indicate a specific association between P2Y2 receptor perturbations and neuropathologic features in AD. Our study suggests that the possible roles of P2Y2 receptor alterations in regulating molecular processes underlying plaque and tangle formation should be further investigated using various approaches, including cell function assays, animal models and postmortem studies. Furthermore, P2Y ligands may warrant study as potential candidates for AD pharmacotherapy.