Structurally Altered, Not Wild-Type, Pentameric C-Reactive Protein Inhibits Formation of Amyloid-β Fibrils Free

C-reactive protein (CRP) is composed of five identical subunits held together by noncovalent bonds and arranged as a cyclic pentamer (1, 2). Each subunit possesses two Ca²⁺ ions that are necessary for the structural integrity of the native pentamer and for binding of CRP to its primary ligand phosphocholine (PCh) (2–4). Because CRP recognizes PCh, a pathogen-associated molecular pattern, CRP is regarded as a pattern recognition molecule of the innate immune system (5, 6). Crystallography of CRP–Ca²⁺–PCh complexes has demonstrated that two Ca²⁺ ions in CRP are coordinated by Asp60, Asn61, and by residues Glu138, Gln139, Asp140, Glu147, and Gln150 present in a loop (7). In the absence of Ca²⁺, the Ca²⁺-coordinating loop moves away from the main body of the CRP molecule, exposing an otherwise hidden site of proteolysis (8–10). Thus, bound Ca²⁺ protects CRP from proteolytic cleavage.

Amyloid-β (Aβ) peptides, 36–43 aa residues long, are normal components of plasma and cerebrospinal fluid. Two main variants of this peptide exist in humans and include Aβ40 and Aβ42, which are 40 and 42 aa residues long, respectively. The most prevalent circulating Aβ variant is Aβ40; however, Aβ42 has been shown to form amyloid fibrils more rapidly (11–14). The aggregation of Aβ peptides into oligomers and fibrils has been implicated in the development and progression of Alzheimer’s disease (AD) (15–18). CRP, remnants of CRP after ligand binding, and CRP-like immunoreactivity have all been seen at sites of inflammation, including in amyloid plaques of AD (19–25). The functions of CRP, however, at the amyloid plaques are not known.

At physiological pH, the binding of CRP to PCh-containing substances is the primary ligand recognition function of CRP, although the binding of CRP to several protein ligands has also been reported (26–29). At acidic pH, the pentameric structure of CRP is altered, and such acidic pH-treated CRP has been shown to bind to a variety of proteins that are denatured and aggregated by immobilization, regardless of the identity of the immobilized protein (10, 30–35). Results similar to that of acidic pH-treated CRP were also observed with H₂O₂-treated CRP (36). CRP has been structurally altered in vitro by employing site-directed mutagenesis to generate CRP mutants, which, like acidic pH-treated and H₂O₂-treated CRP, bound to immobilized proteins, including oxidized low-density lipoprotein (ox-LDL) and complement inhibitor factor H (FH) (37–39). Four such CRP mutants capable of binding to immobilized and aggregated proteins at physiological pH have been reported previously: E42Q, Y40F/E42Q, F66A/T76Y/E81A, and E42Q/F66A/T76Y/E81A (37–40). CRP mutants E42Q and Y40F/E42Q retain their PCh-binding activity, whereas CRP mutants F66A/T76Y/E81A and E42Q/F66A/T76Y/E81A do not bind to PCh, because in these two mutants, the PCh-binding site was mutated (38–42).

The ligand displayed by proteins immobilized on microtiter plates, which is recognized by acidic pH-treated, H₂O₂-treated, and CRP mutants mentioned above, has not been defined yet. We hypothesized that the dense immobilization of proteins onto microtiter plates caused aggregation and denaturation of proteins, resulting in the generation and exposure of a common structure such as an amyloid-like structure that was being recognized by these modified CRP species. We tested this hypothesis by employing anti-Aβ Abs to detect proteins immobilized on microtiter plates. We surmised that if our hypothesis turned out to be correct, then the modified CRP species should also be able to bind to purified Aβ peptides in the fluid phase. Accordingly, we employed CRP mutants E42Q, Y40F/E42Q, F66A/T76Y/E81A, and E42Q/F66A/T76Y/E81A (collectively referred to as “structurally altered pentameric CRP” in this study) and tested their binding to both immobilized and fluid-phase Aβ. The consequences of structurally altered pentameric CRP–Aβ interactions on the formation of Aβ fibrils were also evaluated.

For immobilization, five different randomly selected proteins were used: Aβ peptide 1–42 (Bachem), ox-LDL (Biomedical Technologies), FH (Complement Technology), fibronectin (Fn) (Sigma-Aldrich) and BSA (Sigma-Aldrich). Microtiter wells (Corning, 9018) were coated with 100 µl of 10 μg/ml of each protein in TBS, pH 7.2, in duplicate, and incubated overnight at 4°C. The unreacted sites in the wells were blocked with TBS containing 0.5% gelatin for 45 min at room temperature. Both polyclonal anti-Aβ Abs (Novus Biologicals, NBP2-25093) and monoclonal anti-Aβ Abs (Novus Biologicals, NBP2-13075) were used to detect the amyloid-like structures formed on the proteins following their immobilization. Normal rabbit IgG and normal mouse IgG were use as controls for the Abs. The anti-Aβ Abs (10 μg/ml) diluted in TBS containing 0.1% gelatin and 0.02% Tween 20 were added to the wells and incubated for 1 h at 37°C. After the wells were washed, bound polyclonal anti-Aβ Abs were detected by using HRP-conjugated donkey anti-rabbit IgG (GE Healthcare), and bound anti-Aβ mAbs were detected by using HRP-conjugated goat anti-mouse IgG (Thermo Fisher Scientific). Color was developed, and the OD was read at 405 nm.

Aβ monomers and Aβ fibrils were prepared according to a published method (43). Lyophilized Aβ peptides 1–42 (Bachem, H-1368; m.w. 4514) were dissolved in hexafluoroisopropanol (Sigma-Aldrich) to obtain a 1 mM (4.5 mg/ml) Aβ solution, which was then incubated for 2 h at 37°C for monomerization of the oligomers present in the Aβ peptide solution. Aβ monomers were aliquoted (112.8 μg/25 μl/vial) in glass vials. After hexafluoroisopropanol was removed by evaporation overnight, the vials containing the film of Aβ monomers were stored at −20°C. When needed, 564 µl ice-cold TBS was added to an aliquot to obtain a 200 μg/ml solution of Aβ monomers. To prepare Αβ fibrils, the 200 μg/ml solution of Aβ monomers was vortexed and transferred to a 96-well microtiter plate (Immunochemistry Technologies, Costar 266). The plate was incubated for 3 h at 37°C with shaking at 300 rpm. The resulting Αβ fibrils were stored at −20°C until needed.

The Aβ monomers and Aβ fibrils were visualized by transmission electron microscopy (TEM). For TEM, Aβ monomers or Aβ fibrils (10 μl) were applied to carbon-coated copper grids followed by counterstaining with 2% uranyl acetate and visualized using a Tecnai 120-kV F20 electron microscope at the Molecular Electron Microscopy Core facility, University of Virginia. The results of TEM (Fig. 1) verified the preparations of both Aβ monomers and Aβ fibrils. In addition to TEM, the Aβ monomers and fibrils were also verified by thioflavin T (ThT) assay for Aβ fibrillation, as described below.

The construction of E42Q, Y40F/E42Q, F66A/T76Y/E81A, and E42Q/F66A/T76Y/E81A CRP mutants employed in this study has been reported previously (37–41). CRP mutants were expressed in Chinese hamster ovary (CHO) cells using the ExpiCHO Expression System (Thermo Fisher Scientific) according to the manufacturer’s instructions and as described previously (44).

Wild-type (WT) CRP was purified from discarded human pleural fluid as described previously (45). E42Q and Y40F/E42Q CRP mutants were purified from CHO cell culture supernatants by Ca²⁺-dependent affinity chromatography on a PCh-conjugated Sepharose column (Pierce), followed by ion exchange chromatography on a MonoQ column and gel filtration on a Superose12 column, as described earlier (37, 38). The E42Q/F66A/T76Y/E81A and F66A/T76Y/E81A CRP mutants were purified from CHO cell culture supernatants by Ca²⁺-dependent affinity chromatography on a phosphoethanolamine-conjugated Sepharose column, followed by ion exchange chromatography on a MonoQ column and gel filtration on a Superose12 column, as described earlier (38, 39). Purified CRP in TBS, pH 7.2, containing 2 mm CaCl₂ was stored at 4°C and was used within 1 wk.

The Aβ-binding assays were performed as described previously (32, 39). In brief, lyophilized Aβ peptides were reconstituted in TBS. Microtiter wells (Corning, 9018) were coated with 10 μg/ml of Aβ in duplicate overnight at 4°C. The unreacted sites in the wells were blocked with TBS containing 0.5% gelatin. CRP diluted in TBS containing 0.1% gelatin, 0.02% Tween 20, and 2 mm CaCl₂ (TBS-Ca) was added and then incubated for 2 h at 37°C. After the wells were washed, rabbit polyclonal anti-CRP Ab (Sigma-Aldrich) was used to detect bound CRP. HRP-conjugated donkey anti-rabbit IgG was used as the secondary Ab. Color was developed, and the OD was read at 405 nm. The assays were performed using TBS-Ca, pH 7.2, unless otherwise mentioned.

For inhibition assays, Aβ-binding assays were performed as described above, except that CRP was added along with the inhibitors (Aβ monomers and Aβ fibrils) to Aβ-coated wells. Before CRP was added to the wells, CRP was mixed with Aβ monomers, or with Aβ fibrils, in equal volumes to give a final concentration of 100 μg/ml of the inhibitor.

The fibrillation of Aβ monomers was monitored by employing the ThT assay according to a published method (46). In one assay, the effects of CRP on Aβ fibrillation were determined by adding CRP to Aβ monomers at time 0; that is, CRP and Aβ monomers were mixed together before the beginning of the fibrillation reaction. For this assay, the fibrillation reaction mix was prepared, with and without CRP, in which the final concentrations of Aβ and ThT were 150 μg/ml and 10 μM, respectively. In CRP-containing mixture, the final concentrations of CRP were 1, 10, and 100 μg/ml. After vortexing, 240 μl of each mixture was transferred in triplicate wells in a 96-well microtiter plate (Immunochemistry Technologies, Costar 266). Fluorescence was measured by using the Synergy H1 microplate reader (BioTek) with excitation at 440 nm and emission at 480 nm. After the first measurement at 5 min, the plate was incubated for 3 h at 37°C with shaking at 300 rpm; fluorescence was measured every 15 min for 3 h.

In another assay, the effects of CRP on Aβ fibrillation was determined by adding CRP to Aβ 45 min after the fibrillation began. For this assay, the fibrillation reaction mix was prepared without CRP, in which the final concentrations of Aβ and ThT were 167 μg/ml and 10 μM, respectively. After vortexing, 216 μl of each mixture was transferred in triplicate wells in the microtiter plate. After the first measurement at 5 min, the plate was incubated at 37°C with shaking at 300 rpm. The fluorescence was measured every 15 min. After 45 min, either TBS (24 μl) or CRP (24 μl) was added to the wells to obtain a final concentration of 100 μg/ml CRP. More ThT was added to maintain the ThT concentration at 10 μM. The measurement of fluorescence continued every 15 min for another 2.25 h.

The concentration of Ca²⁺ required for the binding of CRP to PCh was assessed using pneumococcal C-polysaccharide (PnC) (Statens Serum Institut) as the PCh ligand, as described previously (37, 47). In brief, microtiter plates (Corning, 9018) were coated with 10 μg/ml of PnC in TBS, in duplicate wells, overnight at 4°C. CRP diluted in TBS-Ca (with various concentrations of Ca²⁺) was added in duplicate wells. After incubating the plates for 2 h at 37°C, the wells were washed with appropriate TBS-Ca. Purified anti-CRP mAb HD2.4 (0.5 μg/ml), diluted in TBS-Ca, was used to detect bound CRP. HRP-conjugated goat anti-mouse IgG, diluted in TBS-Ca, was used as the secondary Ab. Color was developed, and the OD was read at 405 nm.

The Ca²⁺-binding site-dependent proteolytic cleavage assay of CRP was conducted as described previously (8, 10) with modifications. CRP stored in TBS, pH 7.2, containing 2 mm CaCl₂, was first dialyzed against TBS to remove Ca²⁺. Dialyzed CRP (10 μg; final concentration 2.87 μM) in TBS was incubated with 3 μg protease (Sigma-Aldrich, P8811), with and without 0.5 mM or 5 mM CaCl₂, for 2 h at 37°C and subjected to SDS-PAGE on a 4–20% polyacrylamide gradient gel under reducing conditions. The gels were stained with Coomassie Brilliant Blue. The Bio-Rad Laboratories broad-range marker was used as the m.w. standard.

The results of the experiments in which anti-Aβ Abs were used to identify the expression of amyloid-like structures on immobilized proteins are shown in (Fig. 2. Purified Aβ peptides were immobilized to verify the specificity of the mAbs (Fig. 2A) and polyclonal anti-Aβ Abs (Fig. 2B). Among the other four randomly selected proteins for immobilization, ox-LDL was most reactive with anti-Aβ Abs. Immobilized FH and Fn also reacted with anti-Aβ Abs. These immobilized proteins were recognized by both polyclonal and monoclonal anti-Aβ Abs, although the reactivities of the two Abs with the immobilized proteins were different. Both Abs failed to detect an amyloid-like structure on immobilized BSA. Overall, these data suggest that immobilized proteins express Aβ epitopes or amyloid-like structures and that each protein has different amyloidogenicity; that is, the extent of amyloid-like structures or the number of Aβ epitopes formed on different proteins varies.

We first determined the binding of various structurally altered pentameric CRP species to immobilized Aβ. The results of the solid-phase Aβ-binding assays are shown in (Fig. 3. WT CRP at pH 5.0 and pH 7.2 was included as a positive and negative control, respectively, for the binding of CRP to immobilized Aβ. As shown, and as has been reported previously (31, 32), WT CRP did not bind to Aβ at pH 7.2, but, at pH 5.0, WT CRP bound to Aβ in a CRP concentration-dependent manner. Similar to the binding of WT CRP to Aβ at acidic pH, all four CRP mutants bound to Αβ at physiological pH in a CRP concentration-dependent manner, regardless of whether the PCh-binding site in the CRP mutants was mutated. As shown, the avidity of the binding of E42Q/F66A/T76Y/E81A CRP to Aβ was highest, whereas the avidity of the binding of E42Q CRP to Aβ was lowest, of all other mutants.

Previously, we reported that structurally altered pentameric CRP binds to immobilized proteins, regardless of the identity of the protein (10, 29, 31, 32, 36–39). If the binding of structurally altered pentameric CRP to immobilized proteins was through the ligand Aβ, then CRP mutants employed in this study should be able to bind to purified Aβ in the fluid phase. Accordingly, Aβ-binding inhibition assays were performed in which the inhibition of binding of CRP mutants to immobilized Aβ by fluid-phase Aβ monomers and Aβ fibrils was determined. Two of the four CRP mutants were tested: one that retains the PCh-binding site (Y40F/E42Q) and one with the mutated PCh-binding site (E42Q/F66A/T76Y/E81A). Both Αβ monomers and Αβ fibrils inhibited the binding of Y40F/E42Q CRP to immobilized Αβ, indicating interactions between Y40F/E42Q CRP and Aβ, either monomers or fibrils, in the fluid phase (Fig. 4, left). Greater inhibition by Aβ fibrils than by Aβ monomers suggested that the avidity of binding of Y40F/E42Q CRP to Aβ fibrils was higher than that of Aβ monomers. Similar results were obtained with the CRP mutant E42Q/F66A/T76Y/E81A (Fig. 4, right). These data suggest that Aβ is a ligand of structurally altered pentameric CRP. Inhibition assays for the binding of CRP mutants to immobilized proteins other than Aβ could not be performed, because fluid-phase Aβ interacted directly with immobilized proteins (data not shown).

Under the conditions of the fibrillation assay, the fibrillation of Aβ monomers began within 5 min and continued until ∼2 h, when further fibrillation stopped (Fig. 5, black curves in all panels). WT CRP, even at the highest concentration of 100 μg/ml, did not affect the fibrillation significantly. These results supported our findings shown in Figs. 3 and 4 that WT CRP is incapable of recognizing and binding to Aβ. In contrast to WT CRP, all four CRP mutants inhibited the fibrillation in a CRP concentration-dependent manner. There was no statistically significant difference in fibrillation without or with 1 μg/ml of CRP, except for the Y40F/E42Q CRP mutant. There were statistically significant differences in the fibrillation when CRP mutants were used at 10 μg/ml and 100 μg/ml. As shown in (Fig. 5, CRP mutants significantly inhibited the rate of fibrillation over time and also significantly inhibited the total amount of the fibrils formed by the end of the fibrillation reaction. Fibrillation assays employing acidic pH–treated WT CRP was not performed, because the modified structure of WT CRP at pH 5.0 was reversible at pH 7.0, and the fibrillation assay failed to work at pH 5.0.

Next, we performed fibrillation assays in which CRP was added to Aβ 45 min after the fibrillation began (Fig. 6). WT CRP did not significantly affect the fibrillation of Aβ. Two of the four mutants were tested in this assay: Y40F/E42Q, which binds to PCh, and E42Q/F66A/T76Y/E81A, which does not bind to PCh. Both CRP mutants, as soon as they were added to growing fibrils, completely prevented further fibrillation. The total amount of the fibrils formed at the end of 3 h was significantly less than the total amount of the fibrils formed without CRP. There was no statistically significant difference between the inhibitory effects of the two mutants, suggesting that the effects of CRP mutants on Aβ fibrillation were independent of the PCh-binding site of CRP. Taken together, these data indicate that structurally altered pentameric CRP binds to Aβ monomers and prevents fibrillation. Structurally altered pentameric CRP can also bind to Aβ fibrils to prevent further fibrillation.

We hypothesized previously that the creation of the Aβ-binding site in acidic pH–treated WT CRP was probably due to the loss of one of the two Ca²⁺ from each CRP subunit (32). This hypothesis was based on the findings that acidic pH–treated WT CRP had reduced avidity for Ca²⁺. Because all four CRP mutants bound to Aβ as well as acidic pH–treated WT CRP did, it was possible that the avidity of CRP mutants for Ca²⁺ was also reduced, similar to the reduction in the avidity of acidic pH–treated WT CRP for Ca²⁺. This possibility was explored by employing two different approaches.

In the first approach, we performed Ca²⁺-requirement assays to determine the requirement of Ca²⁺ for the binding of CRP mutants to PCh. Only two of the four CRP mutants, E42Q and Y40F/E42Q, that retained the intact PCh-binding site could be tested employing this assay. The concentration of Ca²⁺ required for 50% binding of each CRP species to PCh was determined; a change in the Ca²⁺ requirement for comparable PCh-binding activities of CRP would reflect a change in the avidity of CRP for Ca²⁺. As shown in (Fig. 7, approximately fivefold more Ca²⁺ was required for equivalent binding of CRP mutants E42Q and Y40F/E42Q to PCh compared with WT CRP. These data suggested that the Ca²⁺-binding site was perturbed in the E42Q and Y40F/E42Q CRP mutants.

In the second approach, we performed protease cleavage assays to assess the Ca²⁺-dependent protection of the protease-sensitive sites in CRP. All four CRP mutants were tested employing this assay. First, we performed a titration assay to determine the concentration of Ca²⁺ needed to protect WT CRP from protease cleavage. As shown (Supplemental Fig. 1), WT CRP was protected from cleavage in a Ca²⁺ concentration–dependent manner. There was no protection of CRP from cleavage at 0.05 mM Ca²⁺; the protection began with a concentration of Ca²⁺ ≥0.5 mM, and 5 mM Ca²⁺ was fully protective. From these data, 5 mM and 0.5 mM Ca²⁺ were chosen to evaluate the Ca²⁺-binding site of structurally altered pentameric CRP. A change in the Ca²⁺ requirement for protecting CRP mutants from cleavage would reflect a change in the avidity of CRP for Ca²⁺.

The results of the cleavage assay performed in the presence of 5 mM Ca²⁺ are shown in (Fig. 8. Protease treatment of WT CRP in the absence of Ca²⁺ generated three fragments (lane 3). The m.w. of the fragments designated A1 and A2 were 17 and 16 kDa, respectively. The 16 kDa A2 fragment is generated because of a secondary cleavage site in the larger fragment A1. The m.w. of fragment B was estimated to be ∼6 kDa, consistent with previously published reports (8, 10). WT CRP was completely protected from this digestion in the presence of 5 mM Ca²⁺ (lane 4). Results of the protease treatment of E42Q and Y40F/E42Q CRP mutants either in the absence of (lanes 6 and 9) or in the presence of (lanes 7 and 10) 5 mM Ca²⁺ were similar to those of WT CRP. Protease treatment of F66A/T76Y/E81A and E42Q/F66A/T76Y/E81A CRP mutants in the absence of Ca²⁺ generated three fragments (lanes 12 and 15); however, the intensity of the cleaved products was stronger than that of WT CRP, indicating that the digestion of CRP mutants was faster than that of WT CRP. In the presence of 5 mM Ca²⁺, F66A/T76Y/E81A and E42Q/F66A/T76Y/E81A CRP mutants were not fully protected from protease cleavage. These data suggested that the accessibility of the proteolytic site in WT CRP to the protease was different from the accessibility of the proteolytic site in CRP mutants F66A/T76Y/E81A and E42Q/F66A/T76Y/E81A, but not in E42Q and Y40F/E42Q, to the protease.

Because the data obtained from the Ca²⁺-requirement assay (Fig. 7) suggested that E42Q and Y40F/E42Q CRP mutants were different from WT CRP, and because the data obtained from the protease cleavage assay (Fig. 8) suggested that E42Q and Y40F/E42Q CRP mutants were not different from WT CRP, we performed another protease cleavage assay with 0.5 mM Ca²⁺ instead of 5 mM Ca²⁺. As shown (Fig. 9), 0.5 mM Ca²⁺ protected WT CRP from cleavage (lane 4) but did not fully protect E42Q and Y40F/E42Q from cleavage (lanes 7 and 10). Because 5 mM Ca²⁺ did not protect the other mutants from cleavage (Fig. 8), it was expected that 0.5 mM Ca²⁺ also would not protect from cleavage (lanes 13 and 16).

Combined data obtained from the Ca²⁺-requirement and protease cleavage assays indicated that the topology of the Ca²⁺-binding site in structurally altered pentameric CRP was perturbed and different from that of WT CRP, despite the fact that the amino acid residues coordinating the two Ca²⁺ were not mutated in any of the CRP mutants.

Our major findings in this study were as follows. (1) Proteins immobilized on microtiter wells expressed amyloid-like structures. (2) Structurally altered pentameric CRP (four different CRP mutants) bound to immobilized Aβ at physiological pH similar to the binding of WT CRP to immobilized Aβ at acidic pH. (3) Structurally altered pentameric CRP also bound to Aβ in the fluid phase. (4) Structurally altered pentameric CRP, not WT CRP, inhibited the fibrillation of Aβ. (5) Structurally altered pentameric CRP also halted the progression of fibril formation when added to growing Aβ fibrils. (6) There was one commonality in all four CRP mutants: The topology of the Ca²⁺-binding site of all four CRP mutants was different from that of WT CRP.

In a previously reported study (48, 49), at physiological pH, WT CRP was also found to bind to Aβ and inhibit fibrillation, which is in contrast to our findings. However, there were methodological differences between the two studies. First, Aβ42 was used in this study, whereas Aβ40 was used in the previous study. These two types of Aβ peptides have different rates of fibrillation; the rate of fibrillation of Aβ42 is faster than that of Aβ40 (50–52). Second, the fibrillation assays were performed while shaking the assay plate in this study, whereas there was no shaking in the previous study. Shaking enhances the rate of fibrillation (50–54). Third, perhaps because of the two differences mentioned above, fibrillation began within 5 min under the conditions of our assay, whereas it began after 30 h in the previous study. On the basis of a previous publication on the stability of purified CRP (55), it was likely that when WT CRP was incubated at 37°C for 30 h (48), a fraction of CRP pentamers might have monomerized. If this happened, then it could be monomeric CRP (mCRP) that bound to Aβ and inhibited fibrillation. Indeed, WT CRP used in the previous study did contain mCRP as determined by gel filtration analysis of the protein (48). Fourth, CRP was used at a concentration of 100 μg/ml and less in this study, whereas it was 500 μg/ml in the previous study. Higher concentration of purified and buffered CRP could have generated enough mCRP and their aggregates, when incubated at 37°C for 30 h, for detectable inhibition of fibrillation. We suspect that in the previous study, if CRP had been added to Aβ after 30 h (i.e., at the time of initiation of fibrillation), and if it had been used at a lower concentration, the results could have been similar to ours. Indeed, in another study on CRP–Aβ interactions (56), biotinylated WT CRP was not found to bind to Aβ.

It has been reported that the binding of WT CRP to certain ligands, including Aβ plaques in vivo, leads to dissociation of pentamers into monomers (23, 24, 35, 57). Such monomerization of pentameric CRP into mCRP involves an intermediate or transitional or nonnative stage when the conformation of the ligand-bound CRP pentamer is different from that of the native pentamer. Several anti-CRP mAbs (3H12, 6B7, 8C10 and 9C9) are available that can differentiate between native and nonnative pentameric CRP; however, these Abs do not differentiate between nonnative pentameric CRP and mCRP (23–25, 35, 57–61). By employing these Abs in immunohistochemistry, it has been concluded that mCRP was deposited at the amyloid plaques of AD (19–25). Because the sites of inflammation, including the site of Aβ deposition, are characterized by acidic and redox conditions (62), we propose that structurally altered pentameric CRP is generated at the site of Aβ deposition first, followed by the binding of structurally altered pentameric CRP to deposited Aβ, which can then lead to monomerization of CRP. It is unclear, though, if CRP, in any structural conformation, before or after monomerization, leaves the surface of the molecule to which CRP was initially bound.

All amyloid deposits contain a serum amyloid P (SAP) component, which is a protein homologous to CRP (63). SAP binds to Aβ fibrils, prevents further fibrillation, and stabilizes the fibrils by protecting against proteolysis (64–68). Because structurally altered pentameric CRP is generated at the site of Aβ deposition, our data raise the possibility of competition between SAP and structurally altered pentameric CRP for binding to Aβ fibrils, and, therefore, the two proteins may be regulating the extent of incorporation of SAP into Aβ fibrils. It remains to be investigated whether the binding of structurally altered pentameric CRP to Aβ fibrils facilitates proteolysis of the fibrils.

Because SAP binds to and prevents proteolysis of Aβ fibrils, both SAP and Aβ fibrils have been therapeutic targets for treating amyloidosis (67). In SAP-deficient mice, in which the gene coding for SAP was inactivated, the deposition of Aβ was found to be delayed (69). Accordingly, treatment with anti-SAP Abs and blocking SAP with small chemical compounds have been shown to remove Aβ deposits in vivo (70). Strategies to prevent the aggregation of Aβ into fibrils and to enable the dissociation of Aβ fibril plaques have also been explored as possible treatments of amyloidosis. Some Aβ-binding synthetic peptides, including one that was derived from CRP, have been found to block the toxic effects of Aβ on neuronal cells in culture (71). Whether exogenously prepared structurally altered pentameric CRP can be delivered to the site of Aβ deposition and whether there would be a protective effect in established animal models and in ex vivo cell culture model systems remain to be explored (69–72).

The amino acid residues that participate in the formation of the Aβ-binding site on structurally altered pentameric CRP have not been identified yet. These amino acid residues must be hidden in WT CRP. Our data suggest that a perturbation of the topology of the Ca²⁺-binding sites is necessary to form the Aβ-binding site. However, there were no mutations of the amino acid residues coordinating Ca²⁺ in any of the four mutants used in this study. Because EDTA-treated, Ca²⁺-free WT CRP does not bind to Aβ (32), it is likely that in all four CRP mutants, there was loss of binding of one Ca²⁺ per CRP subunit, leaving half of the Ca²⁺ site vacant, generating Aβ-binding activity and sensitivity to protease cleavage. Cryo-electron microscopy of all four CRP mutants, as performed for WT CRP at acidic pH (73), would reveal the topology of the Aβ-binding site on structurally altered pentameric CRP.

There are dozens of known amyloidogenic proteins, including ox-LDL employed in this study, that are responsible for clinically significant amyloidoses (74–76). Our hypothesis that dense immobilization of some proteins onto microtiter plates may cause aggregation and denaturation of proteins, resulting in the generation of amyloid-like structures creating Aβ epitopes, was also found to be correct. Considering that immobilized proteins have a common amyloid-like structure on their surface, we conclude that Aβ peptides and amyloid-like structures expressed on immobilized proteins are ligands of structurally altered pentameric CRP. On the basis of the Aβ recognition function of structurally altered pentameric CRP, we propose that CRP should be considered as a dual pattern recognition molecule of the immune system. First, CRP recognizes conserved patterns such as PCh present on the surface of pathogens. Second, CRP recognizes nonconserved, newly created molecular patterns such as amyloid-like structures formed by amyloidogenic and pathogenic proteins. This recognition function is exhibited by structurally altered pentameric CRP.

CRP is a phylogenetically conserved protein (77, 78). Also, a CRP-deficient human has not been reported so far. Therefore, it has been suggested that CRP must have an important host defense function that applies across the animal kingdom. We conclude that CRP is an antiamyloidogenic protein. CRP functions as a scavenger to get rid of misfolded, denatured, deposited, and nonfunctional proteins to prevent the toxicity they generate. This function of CRP would favor the conservation of CRP throughout evolution. The antiamyloidogenic function of CRP also has implications for many human inflammatory and neurodegenerative diseases, such as AD, Parkinson’s disease, the prion-related diseases, and also nonneurodegenerative diseases such as type 2 diabetes that are characterized by abnormal amyloid fibril aggregates (74, 75). The antiamyloidogenic function of CRP also has implications for diseases caused by the deposition of otherwise fluid-phase proteins.

We thank Sanjay Singh, PhD, for his assistance with the preparation of CRP mutants.

The authors have no financial conflicts of interest.