Amyloid Plaques Dissociate Pentameric to Monomeric C‐Reactive Protein: A Novel Pathomechanism Driving Cortical Inflammation in Alzheimer's Disease?

Abstract

Beta‐amyloid (Aβ) plaques and local inflammation are central to the pathogenesis of Alzheimer's disease. Although an association between circulating pentameric C‐reactive protein (pCRP) and Alzheimer's disease has been reported no pathomechanistic link has been established. We hypothesized that Aβ plaques induce the dissociation of pCRP to individual monomers (mCRP), which possess strong pro‐inflammatory properties not shared with pCRP and localizing inflammation to Alzheimer's plaques. pCRP was incubated with Aβ plaques generated in vitro and with non‐aggregated Aβ₄₂ peptide. pCRP dissociation to mCRP was found only when co‐incubated with Aβ plaques. Furthermore, sections of frontal cortex from brains of patients with and without Alzheimer's disease were stained with antibodies specific for mCRP and pCRP. There was significantly more mCRP in the cortex of Alzheimer's disease patients (P ≤ 0.01). In contrast, there was no significant difference in pCRP staining. These findings establish that Aβ plaques possess a previously unrecognized potential to dissociate pentameric CRP to monomeric CRP. The existence of mCRP but not pCRP in the brains of Alzheimer's disease patients strongly indicates that this newly described biological effect of Aβ plaques is relevant in Alzheimer pathobiology; potentially localizing and amplifying inflammation via the strong pro‐inflammatory effects of locally generated mCRP.

INTRODUCTION

Alzheimer's disease (AD) is the most common form of irreversible dementia (14). Despite its enormous clinical burden our understanding of the pathological processes central to the progression of AD is incomplete. The current therapeutic agents available to clinicians serve predominantly to modify patient symptoms and are unable to halt the progressive decline in cognitive function that is the hallmark of AD (5). It is hoped that new generations of agents will be able to slow or even halt the progression of dementia. However, it is likely that any significant therapeutic advance will rely upon an improvement in our current understanding of the pathophysiology of AD, enabling novel treatment approaches to be developed.

There is extensive data that strongly supports the central role of beta‐amyloid (Aβ) in the pathogenesis of AD (35). The Aβ peptide is cleaved from the transmembrane amyloid precursor protein (APP) by β‐ and γ‐secretases. The most important Aβ peptide fragment in AD is an insoluble 42 amino acid peptide (Aβ₄₂). Aβ₄₂ peptides aggregate and are the dominant component of the diffuse senile plaques (29) found in the brain tissue of AD patients.

Senile plaque formation initiates the inflammatory processes, which are responsible for neuronal cell death leading to the progression of AD. Consistent with this understanding, increased levels of the pro‐inflammatory cytokine interleukin‐6 have been found in the brains of AD patients (36). Further supporting data has documented complement and microglial activation (28), along with increased numbers of reactive astrocytes (2). The level of the acute phase reactant C‐reactive Protein (CRP) is also increased (36) and co‐localization of CRP with senile plaques has been shown in brain sections from AD patients (15). Although elevated levels of CRP are associated with AD, there is currently no direct evidence linking CRP to AD pathophysiology.

CRP circulates as a pentamer (pCRP) consisting of five identical non‐covalently bound 23 kDa subunits (25). It has been shown that following binding to a variety of lipid surfaces including activated cell membranes (9) and liposomes (17) pCRP undergoes dissociation to individual monomers (mCRP). mCRP has distinct pro‐inflammatory properties not shared with pCRP including C1q fixation (17), platelet activation (21) and monocyte chemotaxis (9). This raises the possibility that (m)CRP may contribute directly to the pathogenesis of AD. This hypothesis is supported by the previous identification of mCRP in atherosclerotic plaques (9).

In this study, we investigated the specific CRP isoform associated with amyloid plaques in AD patients 15, 38 and identify mCRP but not pCRP in association with senile plaques in AD. Furthermore, we demonstrate that aggregated Aβ plaques but not Aβ peptide fragments are able to dissociate pCRP to mCRP, providing a mechanistic basis for the generation of mCRP and thus localization of inflammation in these regions.

MATERIALS AND METHODS

Antibodies

Antibodies to mCRP (clone 9C9) and to pCRP (clone 8D8 and clone 1D6) were a gift from Dr Potempa (College of Pharmacy, Roosevelt University, Chicago, IL, USA). Antibody NAB228 to Aβ was obtained from Cell Signaling Technology (Danvers, MA, USA) and rabbit polyclonal antibody H‐55 against C1q was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). The secondary antibody Alexa Fluor 488 labeled goat anti‐rabbit was purchased from Invitrogen (Carlsbad, CA, USA) and Alexa Fluor 555 labeled anti‐mouse was purchased from Cell Signaling Technology.

Amyloid beta plaque growth

Aβ₄₂ (GenScript, Piscataway, NJ, USA) was dissolved in 100% methanol and aliquots were frozen at −20°C until use. Amyloid beta plaque formation was performed using a modification of the method previously described (24). Briefly, 55 µM of Aβ₄₂ peptides were added to phosphate buffered saline (PBS) with 0.9 mM calcium (to avoid self‐dissociation of CRP described previously (18)) and 0.49 mM magnesium (Ca²⁺/Mg²⁺) in a polypropylene tube. The tube was placed on a horizontal rotor at ∼1 Hz with visible plaque formation after 24–48 h.

For immunohistologic analysis 275 µM of Aβ₄₂ was added to PBS with Ca²⁺/Mg²⁺ in a polypropylene tube and placed on a horizontal rotor at ∼1 Hz for 24–48 h. The supernatant was removed and the plaque was fixed in low‐melting agarose (Sigma‐Aldrich, St. Louis, MO, USA). After setting the agarose containing plaque was removed from the tube, placed in Tissue Tek optimum cutting temperature (O.C.T.) Compound (Sakura, Torrance, CA, USA) and frozen at −80°C.

Western blot

Artificial plaque (55 µM), non‐aggregated peptide (55 µM) or PBS with Ca²⁺/Mg²⁺ was incubated with pCRP (Merck Chemicals, Darmstadt, Germany) at a final concentration of 25 µg/mL for 60 minutes at 37°C. Nonreducing loading buffer was added to samples before they were run under native polyacrylamide gel electrophoresis (PAGE) conditions and blotted onto a nitrocellulose membrane as previously described (9). Membranes were stained with antibody clones 1D6 (pCRP specific) or 9C9 (mCRP specific).

Immunohistology of artificial plaques

The artificial plaque fixed in agarose was cut on a CM 3000 cryosection system (Leica Microsystems GmbH, Wetzlar, Germany) into 6‐µm sections and stained with the ABC Vectastain kit (Vector Laboratories, Inc., Burlingame, CA, USA) (30). Sections were thawed for 30 minutes before being fixed in acetone at −20°C for 20 minutes, incubated in 3% v/v H₂O₂ in methanol for 10 minutes, blocked with 10% normal rabbit serum (Vector), followed by avidin and biotin blocking for 20 minutes. The slides were incubated with primary antibody clones 8D8, 9C9 or NAB228 at 4°C overnight. Sections were then incubated with a biotinylated anti‐mouse secondary antibody (diluted 1:1000) for 30 minutes prior to incubation with the tertiary antibody for 30 minutes. After the slides were incubated with 3,3′‐diaminobenzidine‐tetrachloride (DAB) substrate chromogen solution for 5 minutes. Sections were finally dehydrated with 100% ethanol, cleared in xylene and mounted with DePeX^® (Sigma).

Patient samples

Tissues were received from the Victorian Brain Bank Network (VBBN), supported by The University of Melbourne, The Mental Health Research Institute of Victoria, Alfred Hospital and the Victorian Forensic Institute of Medicine and funded by Australia's National Health and Medical Research Council, The Helen Macpherson Smith Trust, Parkinson's Victoria and Perpetual Philanthropic Services. In all cases identified as AD, dementia had been clinically diagnosed for several years ante mortem. The diagnosis was confirmed post‐mortem by an experienced neuropathologist according to the diagnostic recommendations defined in 1997 (1). All the patients were classified as Braak tangle stage 5–6 by Dr Catriona McLean, Head of the VBBN. Control patients had no clinical or pathological features consistent with dementia or AD. The two groups show no significant differences in age, sex and the prevalence of cardiovascular diseases, such as myocardial infarction, coronary heart diseases or other thromboembolic diseases (Table 1).

Table 1.

The Alzheimer patients and the patients selected as control group show no significant differences in age, sex and the presence of cardiovascular diseases.

	AD cases	Non‐AD controls	P‐value	Summary
Age	80.69 (74.6–88.4)	78.25 (72.6–82.7)	P > 0.05	NS
PMI (h)	42.57	42.00	P > 0.05	NS
Cases (female/male)	7 (4/3)	6 (2/4)	P > 0.05	NS
CVD	0.29	0.67	P > 0.05	NS

AD, Alzheimer's disease; PMI = post‐mortem interval, CVD = cardiovascular diseases; NS = not significant.

Human brain tissue from seven patients diagnosed with AD and seven non‐AD cases as controls were analyzed. 6‐µm thick formalin‐fixed and paraffin‐embedded sections were obtained from the frontal cortex.

Staining was performed as described (34) with some modification. Briefly, sections were de‐paraffinized, rehydrated and boiled for 10 minutes in concentrated citric acid pH 6.0 as described previously (30), treated with 0.5% H₂0₂ in methanol for 30 minutes and then blocked with 10% normal rabbit serum for 30 minutes. Primary antibodies were diluted in 2% normal rabbit serum with 0.3% triton X‐100, added to the slides and incubated overnight at 4°C. Biotinylated anti‐mouse secondary antibody (diluted 1:1000 in 2% normal rabbit serum) was added for 1 h before addition of the tertiary antibody (ABC Vectastain Elite kit, Vector) for 1 h. The slides were incubated with DAB substrate chromogen solution for 10 minutes and then counter stained with haematoxylin and eosin for 30 s. Finally, sections were dehydrated with 100% ethanol, cleared in xylene and covered with DePeX^® and cover slipped.

Quantification was performed by identification and measurement of the DAB positive tissue area of seven randomly selected cortical regions per slide with Image‐Pro^® (Media Cybernetics Inc., Bethesda, MD, USA); two slides per patient were analyzed. All 14 pictures were used for statistical tests.

For co‐localization assays with fluorochrome labeled antibodies sections were de‐paraffinized, rehydrated and incubated in 0.1% CaCl₂ at 50°C overnight and then blocked with 10% bovine serum albumin (BSA) for 30 minutes. Primary antibodies were diluted in 1% BSA added to the slides and incubated overnight at 4°C. Anti‐mouse and anti‐rabbit secondary antibodies (diluted in 1% BSA) were added for 2 h in the dark at room temperature (RT). Thioflavin T staining was performed with 3 mM freshly filtered aqueous thioflavin T solution for 10 minutes at RT, which were followed by two washing steps in 80% ethanol and one washing step in 95% ethanol with subsequent washing in distilled water. Finally, sections were mounted with Aquatex^® mounting media and covered with a cover slip. Pictures were taken on an Olympus IX81 microscope at 60× magnification. Microscope settings were obtained with single fluorescence stained sections and controls in which primary antibodies had been omitted. One non‐AD case was excluded because of a typical senile plaque pattern seen in the Aβ (NAB228) staining.

Studies were approved by local research ethics committees, and all individuals gave informed consent.

Statistical analysis

All statistical analyses were performed by using GraphPad Prism v5.0 (GraphPad Software Inc., La Jolla, CA, USA). The data are shown as mean ± standard error of the mean (SEM). A P‐value with P ≤ 0.05 was considered as statistically significant.

RESULTS

Artificial senile plaques dissociate pCRP to mCRP

Aβ plaque formed in vitro and non‐aggregated peptide that had not undergone plaque formation were incubated with pCRP for 1 h at 37°C. The contents were then analyzed for the presence of mCRP using native Western blot. This blot demonstrated mCRP formation induced by Aβ plaques. In contrast, the peptide alone was not able to dissociate the pentameric isoform of CRP to its monomeric isoform (Figure 1).

Figure 1.

Aβ plaque‐induced dissociation of pCRP to mCRP. After incubating pCRP (25 µg/mL) with an artificial Aβ plaque for 60 minutes at 37°C a native PAGE and subsequent Western blot was performed. Membrane was stained for mCRP with antibody clone 9C9. A clear mCRP band was visible, whereas no band was seen after incubation of the non‐aggregated Aβ42 peptide alone. Neither the artificial plaque alone nor the non‐aggregated Aβ42 peptide show any staining for mCRP (last two lanes). As a positive control for the dissociation of pCRP into mCRP we included pCRP in PBS incubated at 95°C for 10 minutes (first two lanes). PAGE = polyacrylamide gel electrophoresis; mCRP = monomeric C‐reactive protein; pCRP = pentameric C‐reactive protein.

In a separate experiment, Aβ plaque was also incubated with pCRP, fixed in agarose and sectioned for immunohistological staining. Clone 9C9 (mCRP) and clone 8D8 (pCRP) were used to identify CRP isoforms. Strongly positive staining (brown color) is visible for slides treated with 9C9, whereas almost no staining is detectable in those with 8D8 staining (Figure 2). Control experiments performed to exclude the possibility of nonspecific binding to Aβ plaque confirmed the specificity of the primary and secondary antibodies. Finally, plaques were stained with the antibody clone NAB228 to confirm the homology between the plaques generated in vitro and those found in vivo (Figure 2).

Figure 2.

Histological proof of mCRP generation by an artificial Aβ plaque. An artificial Aβ plaque incubated with pCRP (left lane) and fixed in agarose showing positive (brown) staining with antibody clones 8D8 (pCRP), 9C9 (mCRP) and NAB228 (beta‐amyloid). Another artificial plaque incubated without pCRP (right lane), fixed and stained showed no staining for 8D8 or 9C9. Positive staining (brown) is seen with NAB228. mCRP = monomeric C‐reactive protein; pCRP = pentameric C‐reactive protein.

Characterization of artificial Aβ42 plaques

The artificial plaques formed via aggregation of Aβ42 peptide form beta sheet structures similar to Alzheimer plaques and can therefore serve as a model for the plaques found in the brain of Alzheimer patients. Thioflavin T has the property of emitting a green fluorescence after excitation at 485 nm, when bound to beta sheet structures, and thus indicates the presence of beta sheet structures. Artificial plaques but not the non‐aggregated Aβ42 peptide demonstrated this characteristic (Figure S1).

mCRP but not pCRP can be detected in brain tissue of patients with AD

Sections of frontal cortex from patients diagnosed with AD and non‐AD cases were stained with antibody clones 9C9 (mCRP), 8D8 (pCRP) and NAB228 (Figure 3A). Sections from AD patients were strongly positive (brown color) for mCRP staining but not for pCRP, whereas neither mCRP nor pCRP could be found in the healthy controls.

Figure 3.

Identification of mCRP but not pCRP in human brain tissue from AD patients with Aβ plaques. A. Human brain tissue from the frontal cortex from a patient with AD (left lane) demonstrates positive staining (brown) with the antibody clones 9C9 (mCRP‐specific) and NAB228 (beta‐amyloid‐specific), whereas almost no staining is detectable with clone 8D8 (pCRP‐specific). In contrast, almost no staining is visible in the human brain tissue from a healthy control (right lane) for the antibody clones 8D8, 9C9 or NAB228. These images are representative of observations for 7 AD cases. B. For data analysis the positive stained area of 7 AD cases and 6 non‐AD cases has been evaluated as described in methods. mCRP = monomeric C‐reactive protein; pCRP = pentameric C‐reactive protein; AD = Alzheimer's disease.

Clone NAB228 was used to identify the distribution pattern of the Aβ plaques and confirm their presence in the samples obtained from AD patients. All patients assigned to the AD group had significant plaque burden on histological staining.

The staining in all sections was analyzed by quantifying tissue staining using Image‐Pro^® Plus (Figure 3B). Of the cortex of AD patients, 2.53% ± 0.72% were positive for mCRP, whereas only 0.46% ± 0.12% of the cortex of the controls were positive (P < 0.01). There was only a small amount of pCRP detected in sections from AD patients when compared with controls (0.35% ± 0.05% vs. 0.28% ± 0.09%, P not significant). Comparison of total mCRP and pCRP staining demonstrated significantly more mCRP than pCRP (2.53% ± 0.72% compared with 0.35% ± 0.05%, P < 0.01). As expected, and similar to results observed for mCRP formation, Aβ plaques were far more prevalent in sections from AD patients when compared with controls (2.04% ± 0.49% vs. 0.22% ± 0.04%, P ≤ 0.001).

To investigate a possible effect of the staining technique on pCRP dissociation, an additional set of sections was stained with an alternate method of antigen retrieval, which did not expose tissue sections to conditions that potentially facilitate CRP dissociation (Figure S2A). The antigen retrieval solution contained sufficient free calcium to prevent dissociation and heat treatment was not severe enough to degrade pCRP into monomers as shown previously (32). Additionally, a native PAGE and subsequent Western blot with staining for mCRP (9C9) and pCRP (8D8) were performed to verify that the newly used antigen retrieval method does not cause pCRP dissociation (Figure S2B). The sections that were processed according to this alternate method show a similar staining for mCRP and Aβ42 plaques. Most importantly, there was no significant staining of pCRP in any samples indicating that it undergoes complete dissociation after coming into contact with amyloid plaques. This supports the notion that it is mCRP and not pCRP, which is present in brain tissue of patients with AD.

mCRP is co‐localized with β‐amyloid plaques

mCRP detected by staining with antibody clone 9C9 and Alexa fluor 555 labeled anti‐mouse secondary antibody is co‐localized with Alzheimer plaques in sections of the frontal cortex from patients with AD (Figure 4). Alzheimer plaques have been detected by thioflavin T binding and subsequent imaging at excitation and emission wavelengths of 480 and 550 nm, respectively. There was no staining for mCRP and no fluorescence following thioflavin T staining in non‐AD controls.

Figure 4.

mCRP (red staining) co‐localizes with β‐amyloid plaques (green staining) in the brain of patients with AD. Plaques were identified by staining with thioflavin T and fluorescence imaging at an excitation and emission wavelength of 485 and 550 nm, respectively. mCRP was detected with antibody clone 9C9 and Alexa fluor 555 labelled secondary antibody. No β‐amyloid plaques and almost no staining for mCRP are visible in tissue of healthy controls. The left column shows overlays of bright field and fluorescence pictures taken at 60× magnification. The right column shows fluorescence pictures only. The bottom row shows the co‐localization of mCRP and β‐amyloid plaques at higher magnification (note the different scale bars). mCRP = monomeric C‐reactive protein; pCRP = pentameric C‐reactive protein.

C1q co‐localizes with mCRP in human brain sections of Alzheimer patients

Sections of frontal cortex from patients diagnosed with AD and non‐AD cases were stained with the antibody clone 9C9 (mCRP) and antibody H‐55 (C1q). Samples were then stained with anti‐mouse or anti‐rabbit fluorochrome labeled secondary antibodies (Figure 5). Sections from AD patients showed positive staining for mCRP (red fluorescence) and for C1q (green fluorescence) and co‐localization can be seen in the overlays, whereas neither mCRP nor C1q could be found in the healthy controls.

Figure 5.

mCRP (red staining) and C1q (green staining) are co‐localized in human brain tissue of patients with Alzheimer disease. In contrast, there is almost no staining for mCRP or C1q in brain tissue of healthy controls. mCRP was detected with antibody clone 9C9 and Alexa fluor 555 labeled secondary antibody. C1q was detected with rabbit mAb against C1q and Alexa fluor 488 labeled secondary antibody. The left column shows overlays of bright field and fluorescence pictures taken at 60× magnification. The right column shows fluorescence pictures only. The bottom row shows the co‐localization of mCRP and C1q at higher magnification (note the different scale bars). mCRP = monomeric C‐reactive protein; pCRP = pentameric C‐reactive protein; AD, Alzheimer's disease.

DISCUSSION

In this study, we have demonstrated that the monomeric isoform of CRP but not the pentameric isoform is associated with Aβ plaques found in the cortical tissues from patients with AD. Furthermore, we were able to demonstrate a mechanistic basis for this discovery; Aβ plaques but not Aβ peptides are able to dissociate pCRP into individual monomers. Previous studies have shown the presence of CRP in the brain tissue of AD patients using nonspecific antibodies to detect CRP 8, 15, 38. In this study, using conformation‐specific antibodies, we demonstrate that it is mCRP, not pCRP, which is associated with beta amyloid plaques and that mCRP is co‐localized with the complement factor C1q, which drives cortical inflammation in patients with Alzheimer disease.

The potential source of CRP identified in brain is a matter of controversy. The majority of circulating CRP is synthesized in the liver (4); controlled by cytokines including interleukin‐6, tumor necrosis factor‐α and interleukin‐1. Other potential sources such as atherosclerotic plaques (39) have been identified but the significance of their contribution remains to be elucidated. In brain tissue, the pathophysiology of CRP accumulation is complex because of the requirement of systemically produced CRP to cross the blood–brain barrier (BBB) to reach the cerebrospinal fluid (CSF) and brain tissue. However, it has been established previously that during inflammatory conditions, the BBB becomes dysfunctional, enabling proteins normally only found in serum to enter the CSF (31). Elevated serum CRP levels in AD have been described in several studies 7, 13, 40 and are associated with more rapid progression of disease (37). Although transit of circulating CRP across the BBB is the most likely potential source of cerebral CRP (23), there are also studies that suggest that it may be produced in situ. Yasojima et al found that CRP mRNA is up‐regulated in cells within the affected areas of AD brains (38); potentially obviating the need to transit the BBB. Additionally, Ciubotaru et al showed mCRP production by U937‐derived macrophages (6). It is therefore possible that microglia, which function as resident macrophages of the brain, produce mCRP locally. A local production of mCRP would also be in accordance with the findings of Slevin et al who showed that stroke‐affected neurons expressed mCRP in cytoplasm and nuclei whereas normal areas of the cortical neurons only showed weak cytoplasmic expression of mCRP (30). However, which of these mechanisms operate in vivo and their relative importance remains to be established.

In this study, we show that Aβ protein, which has aggregated into plaques in vitro, has the capacity to dissociate pCRP to mCRP. In contrast, individual peptides that have not undergone plaque formation cannot dissociate pCRP. The physiological mechanism resulting in pCRP dissociation on Aβ plaques is not known yet. One potential cause is the displacement of calcium bound to pCRP by amyloid plaques. Calcium chelation is a well‐recognized CRP dissociative mechanism (27). Calcium has been shown to co‐accumulates with Aβ plaques (20) and the donation of the pCRP bound calcium to the plaque may lead to pCRP dissociation (4). An alternate etiology would be conversion on the membranes of activated cells associated with the Aβ plaque. Activated cell membranes are currently considered the primary mediator of pCRP dissociation (10) and it is therefore likely that this mechanism also is involved in mCRP formation in AD.

The pathophysiological role played by CRP in AD remains an area of significant research interest. In animal models, the intra‐cerebroventricular injection of CRP has been shown to be a potent initiator of intracerebral inflammation leading to progressive long‐term memory loss (19). Previous work has established the importance of complement in mediating the pro‐inflammatory effects of CRP (12). Rogers et al showed immunohistochemical co‐localization of C1q, membrane attack complex (MAC) and thioflavin S‐positive senile plaques in vivo (28). In a myocardial infarction model, complement depletion prevented the pathogenic effects of CRP (12). Consistent with these findings we were able to demonstrate mCRP co‐localization with C1q (Figure 4). The pathological role of the complement system in AD is still unresolved (33). Interestingly, in 2006 Ji et al showed C1q fixation by surface‐bound mCRP, which regulates the classical complement pathway (CCP) by activating the complement system (16). A study by Biro et al showed that mCRP but not pCRP binds C1q and can activate the CCP (3). It is this distinct functional change initiated upon CRP dissociation that is the crucial intermediate step linking CRP to complement activation. The dissociation process also acts to localize complement activation to areas of plaque formation, accounting for the focal changes observed in pathological studies in AD.

Additional evidence supporting the central importance of inflammation in degenerative brain disease has been obtained from the study of traumatic brain injury. In these individuals, the initial external impact itself can cause less damage to the brain tissue than the secondary inflammatory response this insult triggers (22). This discovery suggests that anti‐inflammatory strategies may be of benefit in the prevention and treatment of degenerative brain diseases, a novel approach with significant potential. Akiyama et al presented a diagram which depicts the role of microglia in the pathogenesis of AD, inflammation and neuronal degeneration; we have adapted this scheme by adding our new findings postulating a direct role of mCRP in AD (Figure 6).

Figure 6.

A postulated role of CRP in the development of Alzheimer's disease. This adapted scheme outlines the pathomechanism of Alzheimer's disease including a possible role of CRP in the development and neuronal degeneration of Alzheimer's disease. As a first step the Aβ peptide starts to aggregate, which leads to the dissociation of pCRP to mCRP. mCRP itself as a strong pro‐inflammatory agent recruits/activates macrophages/microglia and has the ability to fix C1q initiating the complement cascade. As a final step the C5b‐9 membrane attack complex contributes directly to neuronal degeneration as seen in patients with Alzheimer's disease. mCRP = monomeric C‐reactive protein; pCRP = pentameric C‐reactive protein; APP = amyloid precursor protein.

It remains a matter of great controversy as to whether CRP is a direct mediator of inflammatory processes or merely a marker of systemic inflammation. Animal studies investigating myocardial infarct size after coronary ligation (12) or provoked cerebral ischemia (11) have shown a complement‐dependent increase in ischaemic lesion size following infusion with CRP. These findings prompted the development of the first described direct CRP inhibitor, 1,6 bis‐phosphocholine (bisPC). This agent was used in a rat model of myocardial infarction and was able to completely inhibit CRP mediated necrosis (26). As pCRP appears to have only limited pro‐inflammatory properties, it is more likely that this effect is mediated by blocking its dissociation to the more pro‐inflammatory monomeric isoform. These findings underline the fundamental significance of the finding that mCRP and not pCRP is found in the brain of patients with AD.

There are now a large number of observational and experimental studies that have clearly implicated CRP in the development and progression of AD. However, these studies have not identified the mechanism by which CRP mediates these effects. Introducing the intermediate step of pCRP dissociation to mCRP potentially solves this dichotomy and integrates previously contradictory data.

In conclusion, we show that Aβ plaques can induce the dissociation of pentameric to monomeric CRP and thereby localizes inflammation to Alzheimer plaques. Consistent with this pathophysiological concept, we could identify monomeric CRP together with Aβ plaques in the cortical tissues from patients with AD. This finding potentially links CRP to the inflammatory processes underlying the progression of AD and potentially other degenerative brain diseases. This finding contributes to the understanding of the pathophysiology of AD, and it may enable the development of novel therapeutic approaches for this most insidious disease.

Supporting information

Figure S1. Artificial plaques formed through aggregation of Aβ42 peptide form beta sheet structures and can therefore serve as a model for the plaques found in the brain of patients with Alzheimer disease. The presence of beta sheet structures is shown by thioflavin T binding and subsequent emission of green fluorescence. Left column: Artificial plaques and non‐aggregated Aβ peptide in agarose were stained with 3 mM thioflavin T solution and pictures were taken at excitation and emission wavelengths of 485 and 550 nm, respectively. Green fluorescence can be seen in artificial plaques, but not in Aβ42 peptide. Right column: Artificial plaque or Aβ peptide in agarose stained with monoclonal antibody NAB228 against beta‐amyloid. Staining for amyloid can be seen in both peptide and artificial plaque. In the sections of the non aggregated peptide staining is evenly distributed in the agarose, whereas aggregates can be seen in the sections with the artificial plaque.

Figure S2. A. A modified antigen retrieval method, which does not cause pCRP dissociation does not affect staining for monomeric and pentameric CRP. Human brain tissue from the frontal cortex of patients with Alzheimer disease demonstrates positive staining (brown) with the antibody clones 9C9 (mCRP‐specific) and NAB228 (beta‐amyloid‐specific), whereas almost no staining is detectable with clone 8D8 (pCRP‐specific). B. Incubating pCRP in 0.1% CaCl₂ at 50°C overnight does not cause dissociation into its monomeric isoform. pCRP was diluted in 0.1% CaCl₂ and then incubated at 50°C for 12 h and then run on a 10% native gel followed by native Western blotting and staining with conformation‐specific antibodies (clone ID6: pCRP; clone 9C9: mCRP). As a positive control for dissociation, pCRP in PBS was incubated at 95°C for 10 minutes.

Supporting info item