|Home | About | Journals | Submit | Contact Us | Français|
Pathological studies have determined that fibrillar forms of amyloid-beta protein (Aβ) comprise the characteristic neuritic plaques in Alzheimer’s disease (AD). These studies have also revealed significant inflammatory markers such as activated microglia and cytokines surrounding the plaques. Although the plaques are a hallmark of AD, they are only part of an array of Aβ aggregate morphologies observed in vivo. Interestingly, not all of these Aβ deposits provoke an inflammatory response. Since structural polymorphism is a prominent feature of Aβ aggregation both in vitro and in vivo, we sought to clarify what Aβ morphology or aggregation species induces the strongest proinflammatory response using human THP-1 monocytes as a model system. An aliquot of freshly-reconstituted Aβ(1-42) in sterile water (100 μM, pH 3.6) did not effectively stimulate the cells at a final Aβ concentration of 15 μM. However, quiescent incubation of the peptide at 4°C for 48-96 h greatly increased its ability to induce tumor necrosis factor-α (TNFα) production, which surprisingly declined upon further aggregation. Imaging of the Aβ(1-42) aggregation solutions with atomic force microscopy indicated that best cellular response coincided with the appearance of fibrillar structures yet conditions that accelerated or increased Aβ(1-42) fibril formation such as peptide concentration, temperature, or reconstitution in NaOH/PBS at pH 7.4, diminished its ability to stimulate the cells. Finally, depletion of the Aβ(1-42) solution with an antibody that recognizes fibrillar oligomers dramatically reduced the ability to induce TNFα production and size-exclusion separation of the Aβ(1-42) solution provided further characterization of an aggregated species with proinflammatory activity. The findings suggested that an intermediate stage Aβ(1-42) fibrillar precursor is optimal for inducing a proinflammatory response in THP-1 monocytes.
The Alzheimer’s disease (AD)1 brain is decorated by a wide array of polymorphic aggregated amyloid-β (Aβ) species. The parenchymal deposits exist as a range of species from dense core neuritic plaques containing fibrillar structures to wispy, loose, and granular diffuse deposits (1) yet the conditions that produce these structures in vivo are not well understood. Furthermore, these diverse Aβ morphologies do not appear to provoke the same in vivo response. For example, cytopathology, such as dystrophic neurites and inflammation, is observed surrounding the plaques (2) in the brains of human patients (3) and those in AD transgenic mouse models (4). Inflammatory markers such as activated microglia (4) stained with proinflammatory cytokines (5) are part of the environment surrounding the plaques while diffuse deposits are devoid of inflammatory cytopathology (1). A chronic inflammatory state induced by accumulated Aβ has been suggested as one of the underlying mechanisms of progressive neurodegeneration in AD (6) and may in fact exacerbate Aβ deposition (7). The plaque-associated activated microglia do not efficiently clear the deposits (4) unless further stimulated by infusion of anti-Aβ antibodies (8).
In vitro aggregation studies of Aβ have been very useful for understanding fibrillogenesis mechanisms and the structural properties of monomers, soluble intermediates, and mature fibrils. These studies have identified a continuum of Aβ species in the assembly process which vary in their size, length, solubility, and morphology (9-13). Monomeric, oligomeric, protofibrillar, fibrillar, and amorphous species possess distinct toxic and biological activities and potencies (14-18).
The in vivo inflammatory response to Aβ has been recapitulated in numerous in vitro cell model systems including both microglial and monocytic cells (19-21). Several receptors have been shown to mediate Aβ-stimulated proinflammatory cytokine production including the receptor for advanced glycation end products (RAGE) (22), a multireceptor complex comprising the scavenger receptor class B (SR-B) receptor CD36, α6β1-integrin, and CD47 (23), and Toll-like receptors (TLR) 2 and 4 (24-26). Furthermore, the SR-A receptor has been shown to mediate Aβ-induced production of reactive oxygen species (27). In all of these studies, induction of an inflammatory response appeared to favor a fibrillar Aβ conformation although different assembly states were not always investigated. Since activated microglial cells are typically observed clustered around the dense core plaques as opposed to the diffuse Aβ deposits (1), their activation appears to be selective for a particular Aβ morphology. We have previously reported that aggregated Aβ(1-42) induces TNFα production from a human monocytic cell line via TLRs (24). Here we further explore the optimal Aβ aggregation state for this process.
THP-1 cells were obtained from ATCC (Manassas, VA) and maintained in RPMI-1640 culture medium (HyClone, Logan, UT) containing 2 mM L-glutamine, 25 mM HEPES, 1.5 g/L sodium bicarbonate, 10% fetal bovine serum (FBS) (HyClone), 50 U/ml penicillin, 50 μg/ml streptomycin (HyClone), and 50 μM β-mercaptoethanol at 37°C in 5% CO2. For cellular assays, THP-1 monocytes were centrifuged, washed, and resuspended in reduced FBS (2%) growth medium. Cells were added to individual wells of a 48-well or 96-well sterile culture plate at a final concentration of 8.5 × 105 cells/ml prior to treatment with 15 μM Aβ and/or sterile water control. In some experiments, polymyxin B sulfate (Sigma, St. Louis, MO) was included to verify the proinflammatory signal was not due to lipopolysaccharide (LPS) contamination. Following incubation of the cells at 37°C for 6-24 h, the content of each well was removed, centrifuged at 2500g for 10 min, and the supernatant was frozen at -20°C for subsequent analysis.
Aβ(1-42) peptides (rPeptide, Bogarth, GA) were dissolved in 100% hexafluoroisopropanol (HFIP) (Sigma) for 1 h, aliquotted into sterile microcentrifuge tubes, dried in a vacuum centrifuge, and stored at -20°C. For experiments, the lyophilized peptides were resuspended in sterile water or treated with 100 mM NaOH at 2 mM Aβ and diluted into phosphate-buffered saline (PBS, HyClone, 6.7 mM phosphate, 150 mM NaCl). Final Aβ concentrations were 100 μM and the solutions were incubated at 4°C unless otherwise stated. THP-1 monocytes were exposed to a final concentration of 15 μM Aβ(1-42). Commercial Aβ lots were endotoxin-tested by several methods as previously described (24). Centrifugation of Aβ solutions was done on either a Beckman-Coulter Microfuge® 18 at 18,000g for 10 min in a 4°C cold room, a Sorvall RC5B refrigerated centrifuge with SS-34 rotor at 50,000g for 1 h at 4°C, or a refrigerated Beckman-Coulter Optima Max ultracentrifuge with TLA120.1 rotor at 50,000g to 150,000g for 1 h at 4°C. Some centrifugation experiments utilized 0.2 μm polytetrafluoroethylene (PTFE) spin filters (Millipore, Billerica, MA) at 12,000g for 3 min to separate small aliquots (~100 μl) of Aβ aggregation solutions. Further separation was done with size exclusion chromatography (SEC). SEC columns were sanitized with 0.5 M NaOH and pretreated with 1 mg bovine serum albumin (BSA) in running buffer to block non specific binding to the resin. 18,000g Aβ(1-42) supernatants were eluted on a Superdex 75 HR 10/30 column (GE Healthcare) in 50 mM Tris-HCl (pH 8.0) at 0.5 ml/min. Collection of fractions (0.5 ml) was initiated after 2 ml of elution volume. Aβ(1-42) concentrations from SEC were determined by absorbance using an extinction coefficient of 1450 cm-1 M-1 as previously described (28). In some cases, Aβ concentrations were determined by the Bradford method (29). Some Aβ aggregation solutions were monitored by thioflavin T (ThT) fluorescence as described previously (30). Briefly, Aβ aliquots were removed and diluted 5-fold (20 μM) into a 5 μM ThT solution prepared in the same solution used for Aβ(1-42) reconstitution (water or PBS). In some cases, Aβ(1-42) reconstituted and aggregated in water was monitored with 5 μM ThT in 50 mM Tris-HCl pH 8.0. ThT fluorescence emission scans (460-520 nm) were acquired on a Cary Eclipse fluorescence spectrophotometer using an excitation wavelength of 450 nm and integrated from 470-500 nm to produce ThT fluorescence values. When necessary, pH was determined in Aβ solutions using a small volume microelectrode (Thermo Scientific Orion).
Measurement of secreted TNFα in the supernatants was determined by ELISA. Briefly, 100 μl of 2-4 μg/ml monoclonal anti-human TNFα/TNFSF1A capture antibody (R&D Systems, Minneapolis, MN) was added to 96-well plates for overnight incubation at room temperature. Wells were washed with PBS (HyClone) containing 0.05% Tween-20 and blocked with 300 μl PBS containing 1% BSA, 5% Sucrose and 0.05% NaN3 for 1 h at room temperature. After washing, successive treatments with washing in between were done with 50 μl samples or standards for 2 h, 100 μl biotinylated polyclonal anti-human TNFα/TNFSF1A detection antibody (R&D Systems) in 20mM Tris with 150 mM NaCl and 0.1% BSA for 2 h, 100 μL streptavidin-horseradish peroxidase (HRP) (R&D Systems) diluted 200 times with PBS containing 1% BSA for 20 min, and 100 μl of equal volumes of 3,3′,5,5′-tetramethylbenzidine and hydrogen peroxide (KPL, Gaithersburg, MD) for 30 min. The reaction was stopped by the addition of 1% H2SO4 solution. The optical density of each sample was analyzed at 450nm with a reference reading at 630nm using a SpectraMax 340 absorbance plate reader (Molecular Devices, Union City, CA). A standard curve was constructed by sequential dilution of a TNFα standard from 2000-15 pg/ml. The concentration of TNFα in the experimental samples was calculated from a TNFα standard curve of 15-2000 pg/ml. When necessary, samples were diluted to fall within the standard curve.
Hydrodynamic radius (RH) measurements were made at room temperature with a DynaPro Titan instrument (Wyatt Technology, Santa Barbara, CA). Samples (30 μl) were placed directly into a quartz cuvette and light scattering intensity was collected at a 90° angle using a 10-second acquisition time. Particle diffusion coefficients were calculated from auto-correlated light intensity data and converted to RH with the Stokes-Einstein equation. Data regularization with Dynamics software (version 6.7.1) generated histograms of percent mass vs. RH. Intensity-weighted mean RH values were derived from the regularized histograms.
Aβ(1-42) aggregation solutions (100 μM) were diluted to 1 μM in water. Grade V1 mica (Ted Pella, Inc, Redding, CA) was cut into 11 mm circles and affixed to 12 mm metal discs. Aliquots (50 μl) were applied to freshly cleaved mica, allowed to adsorb for 15 min, washed twice with water, air dried, and stored in a container with desiccant. For imaging of Aβ(1-42) aggregates that were SEC-separated in 50 mM Tris-HCl, mica surfaces were pre-treated with 1% 3-aminopropyl triethoxysilane (APTES) in 1 mM acetic acid for 10 min, washed with water, and air-dried prior to application of sample. Images were obtained with a Nanoscope III multimode atomic force microscope (Digital Instruments, Santa Barbara, CA) in TappingMode™. Height analysis was performed using Nanoscope III software on flattened height mode images.
Aβ aggregation solutions were diluted to 20 μmol/L in water and 10 μl was applied to a 200-mesh formvar-coated copper grid (Ted Pella, Inc.). Samples were allowed to adsorb for 10 min at 25°C, followed by removal of excess sample solution with a tissue wipe. Grids were washed three times by placing sample side down on a droplet of water. Heavy metal staining of the samples was done in a similar manner by incubation on a droplet of 2% uranyl acetate (Electron Microscopy Sciences, Hatfield, PA) for 5 min, removal of excess solution, and air drying. Affixed samples were visualized with a JEOL JEM-2000 FX transmission electron microscope operated at 200k eV.
Fibrillar oligomers were immunoprecipitated (IP) by addition of OC antisera (2 μl, 1:300 dilution) (gift from Dr. Rakez Kayed, University of Texas Medical Branch) directly to Aβ aggregation solutions (60 μl, 100 μM) and incubation without agitation for 1 h at 4°C. Protein G-sepharose beads (10 μl) (Sigma) were then added to the solution and incubated with slow mixing for an additional 1 h at 4°C. The solution was centrifuged for 15 min at 18,000g and the supernatant (45 μl) was used to treat THP-1 monocytes.
All steps in the dot blot assay were done at 25°C and were modified from (31). Briefly, 5 μl Aβ(1-42) was applied to moist nitrocellulose, allowed to stand for 20 min,and then blocked with 10% milk in PBS-0.2% Tween 20 (PBST). Following a wash step with PBST, the membrane was incubated with OC serum (1:5000) or Ab9 antibody (1:5000) (gift from Dr. Terrone Rosenberry, Mayo Clinic Jacksonville) for 1 h with gentle shaking, washed, and incubated with a 1:1000 dilution of an anti-rabbit IgG (OC) or anti-mouse IgG (Ab9) HRP conjugate (R&D Systems) for 1 h. After washing, the nitrocellulose membrane was then incubated with ECL substrate and exposed to film.
Cell viability was monitored using an XTT (2, 3-bis (2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide) cell assay (32). Cell metabolic activity was probed by mitochondrial-mediated reduction of XTT (Sigma). Briefly, control cells and cells exposed to Aβ for a given period of time were further incubated with final concentrations of 0.33 mg/mL XTT and 8.3 μM phenazine methosulfate (PMS) (Acros, Morris Plains, NJ) for 3 h at 37°C. The extent of XTT reduction was determined by absorbance measurements of the reduced form of XTT at 467 nm.
TNFα, a key product of the proinflammatory response, is measurably increased in post-mortem AD brain sections (5) and microvessels (33), and cerebrospinal fluid (34) of clinically diagnosed AD patients. The THP-1 human monocytic cell line is an excellent model system for inflammatory studies, exhibits responses to stimuli similar to those of microglia, and are a particularly valuable system for studying cell activation and cytokine production by Aβ (19-21). In order to gain further information on the most favorable aggregation state of Aβ for inducing TNFα production, freshly reconstituted solutions of Aβ(1-42) (100 μM) were aged at 4°C and periodically tested for their ability to stimulate TNFα production from THP-1 monocytes. Figure 1 shows that immediately upon reconstitution Aβ(1-42) induced a varying but relatively small amount of TNFα. Aging of the peptide for 48-72 h caused a dramatic increase in TNFα production (Fig 1, lines 1-4). Surprisingly, further aging reduced, and ultimately abolished, the proinflammatory response (Fig 1, lines 3, 4). The results from multiple experiments suggested that an intermediate Aβ(1-42) aggregation structure may be preferred for inducing TNFα production in THP-1 monocytes. Furthermore, the data demonstrated that continued aggregation from that point diminished the response.
The idea that later-stage Aβ(1-42) aggregation species were not effective inducers of TNFα production was tested further by modulating Aβ(1-42) aggregation kinetics. The initial peptide concentration was increased 12-fold thereby accelerating aggregation. Aβ assembly occurs via a nucleation-dependent polymerization process (35). One of the tenets of this type of kinetics is that increased peptide concentration can significantly shorten the lag time for nucleation which is then followed by rapid polymerization and fibril formation. In addition to increased peptide concentration, elevated temperature also accelerates Aβ aggregation (11). The more concentrated solution of Aβ(1-42) (1.2 mM) was incubated at 25°C and aliquots were added to THP-1 monocytes while maintaining the same final concentration (15 μM) as that used for the 100 μM solutions. Although a small level of TNFα was induced at 0 h, the remaining Aβ(1-42) aggregation age time points had no stimulatory activity (Fig 1, circles).
AFM analysis of an Aβ(1-42) aggregation time course was done to visualize morphologies that corresponded to periods of TNFα production shown in Figure 1. Upon reconstitution of Aβ(1-42) in sterile water, images primarily showed a dense field of small punctate species with heights of <2 nm as reported previously (24)(Fig 2A). Dynamic light scattering analysis of the freshly reconstituted Aβ(1-42) showed a predominant peak (95% mass) with an RH of 1.0 nm representing monomeric Aβ and a small population (5% mass) of oligomeric species with an RH of 5.7 nm (Fig 2B). Incubation of the Aβ(1-42) solution at 4°C produced fibrillar structures by 48 h (Fig 2C). The 48 h image in Fig 2C is representative of the intermediate Aβ(1-42) aggregation state that typically stimulated maximal TNFα production (see Fig 1). Height analysis of the 48 h fibrils produced a mean height and standard deviation (SD) of 4.2 +/- 1.4 nm with lengths ranging from 1-3 μm. We have described these features previously in the context of Toll-like receptor activation (24). The fibril height values did not change significantly after 216 h of incubation (4.5 +/- 1.4 nm) (Fig 2D) and lengths were only slightly greater. The most notable change was in the density of fibrils (fibrils/μm2) which increased from approximately 1 fibril/μm2 at 48 h of incubation to 6 fibrils/μm2 by 216 h. ThT fluorescence measurements of the Aβ(1-42) solution indicated no fluorescence at 0 h although longer incubation time gradually increased the fluorescence (Fig 5, circles). A more concentrated Aβ(1-42) solution (1.2 mM) showed accelerated aggregation and fibril production after 24 h at 25°C (Fig 2E). This accelerated aggregation was also reflected in ThT fluorescence levels which were 10-20 times higher in the 1.2 mM solutions than in the 100 μM solutions (data not shown). Treatment of the THP-1 monocytes with the more concentrated sample in Fig 2E at a final Aβ(1-42) concentration of 15 μM did not induce TNFα production (data not shown) consistent with the results shown in Fig 1.
The aggregation rate of the 100 μM Aβ(1-42) solution was accelerated by incubation at higher temperatures. Three solutions of 100 μM Aβ(1-42) were prepared and incubated at 4°C, 25°C, or 37°C. Incubation at higher temperatures significantly decreased the ability of the solution to induce TNFα production. AFM images showed differences in the extent of aggregation at 96 h (Fig 3A-C) yet only the sample incubated at 4°C stimulated TNFα production in THP-1 monocytes (Fig 3D). As in Figure 2, the AFM image of the Aβ(1-42) sample incubated at 4°C contained long flexible fibril structures with a mean height of 5.5 +/- 1.6 nm (SD) along with numerous globular species. The Aβ(1-42) sample at 25°C showed a greater number of fibril structures and also greater dispersity in measured fibril heights (6.9 +/- 2.1 nm). The Aβ(1-42) sample at 37°C was very different by AFM imaging. Although the fibril heights were similar (6.1 +/- 1.6 nm), a lower amount of total Aβ(1-42) species was evident in the image, possibly due to decreased adsorption of the Aβ(1-42) fibrils formed at 37°C to the mica surface. Furthermore, the smaller globular species were no longer observed. Separate experiments measuring ThT fluorescence found much higher levels for Aβ(1-42) incubated at 37°C (470 arbitrary units) compared to incubation at 25°C (80 units) or 4°C (40 units) (data not shown). Aβ(1-42) samples were taken from the three solutions depicted in Figures 3A-C and tested for induction of THP-1 monocyte TNFα production. The dependence of TNFα production on Aβ(1-42) aggregation state for the sample incubated at 4°C (Fig 3D, circles) showed the same profile as in Figure 1, while Aβ(1-42) solutions incubated at higher temperatures were ineffective at inducing an inflammatory response. The cumulative data demonstrated that Aβ(1-42)-induced TNFα production correlated with initial formation of an intermediate fibrillar species yet continued, accelerated, or increased fibril formation abolished the ability of Aβ(1-42) to stimulate the monocyte response.
Two possible explanations for the loss of Aβ(1-42) proinflammatory activity at the later stages of fibril formation was solubility (fibril precipitation) and/or fibril toxicity to the monocytes particularly at later stages when significant numbers of fibrils are present. The first possibility was explored and it was observed that many of the late stage fibrillar species formed in water remained in solution after centrifugation of the sample at 18,000g for 10 min based on AFM images pre- and post-centrifugation (data not shown) This result indicated the fibrils were not easily precipitating out of solution. The toxicity of Aβ(1-42) to the monocytes at both intermediate and late aggregation stages was tested using an XTT cell viability assay. Aβ(1-42) was not toxic to the cells at two stages of aggregation as mitochondrial-mediated reduction of XTT was not affected (Fig 4).
The pH of the Aβ(1-42) aqueous solutions was determined to be 3.6 using a microelectrode. Acidic pH has been shown previously to enhance single fibril formation (11). We hypothesized that Aβ(1-42) at higher ionic strength and buffered at neutral pH may form structures with different proinflammatory-stimulating activity. Aβ(1-42) was either reconstituted in water or treated with 100 mM NaOH followed by dilution into PBS. Both solutions were prepared at a final Aβ concentration of 100 μM and incubated at 4°C. The Aβ(1-42)/PBS solution (pH 7.4) achieved a higher degree of aggregation based on ThT fluorescence measurements compared to the Aβ(1-42)/water solution (Fig 5A) yet Aβ(1-42) incubated in PBS did not stimulate TNFα production to the same extent as Aβ(1-42) incubated in water (Fig 5B). Initially, the Aβ(1-42)/water solutions were monitored by ThT prepared in water which maintained the acidic pH. This method resulted in very low ThT fluorescence (data not shown) and did not accurately reflect the extent of Aβ(1-42) aggregation. Analysis of the (1-42)/water solutions with ThT prepared in Tris-HCl pH 8.0 (Fig 5A, circles) or glycine pH 8.0 (data not shown) produced similar ThT fluorescence values that were both significantly higher than when ThT fluorescence was measured in acidic conditions. Microscopy was utilized to discern morphological differences between Aβ(1-42) incubated in water or PBS. Difficulty was encountered in attempting to adsorb Aβ(1-42) aggregates formed in PBS to mica grids for AFM, therefore TEM was used for morphological evaluation of the two Aβ(1-42) solutions. TEM images obtained of each sample showed that both preparations contained significant fibrillar material. Compared to the Aβ(1-42)/water solution, the Aβ(1-42)/PBS solution possessed more laterally associated fibrils (Fig 5C, D) and a greater number of fibrils when lower magnification was used to observe a larger field (data not shown).
The data to this point indicated that Aβ(1-42) incubated in water at 4°C formed a species that acted as a proinflammatory stimulus. It was of interest to determine if Aβ(1-40) could form the same species under similar conditions. Solutions of Aβ(1-42) and Aβ(1-40) (100 μM in water) were prepared. The Aβ(1-42) solution was incubated at 4°C while the Aβ(1-40) solution was incubated at three temperatures (4°C, 25°C, or 37°C). AFM imaging indicated that even at 4°C Aβ(1-40) did form fibrils (Fig 6A-C) albeit at a much slower rate compared to Aβ(1-42). The Aβ(1-40) fibrils were much longer (> 5 μm) than those formed by Aβ(1-42) and their measured heights were slightly greater. Aβ(1-40) fibrils presented in Fig 6B and 6C had average heights of 5.9 +/- 1.7 nm (SD) with very little change in the fibril morphology from 96 to 216 h. Concurrent treatment of THP-1 monocytes with aliquots from the Aβ solutions revealed that, under these conditions, only Aβ(1-42) effectively stimulated TNFα production (Fig 6D). Incubation of Aβ(1-40) at increased temperature produced more numerous fibrils as observed by AFM (data not shown), yet did not produce a proinflammatory Aβ species.
The data in Figure 1 demonstrated that, upon Aβ(1-42) reconstitution in sterile water, a period of incubation is necessary before an aggregated species conducive for inducing TNFα secretion from THP-1 monocytes is formed. The fact that continued aggregation diminished proinflammatory activity indicated that an intermediate Aβ(1-42) species was optimal. In order to better characterize this species, we subjected an Aβ(1-42) solution incubated for 72 h at 4°C to centrifugation of speeds up to 150,000g for 1 h at 4°C. This treatment failed to significantly suppress the ability of the supernatant to induce TNFα production (Fig 7A) yet was effective at removing many of the Aβ(1-42) fibrils from solution at 150,000g (Fig 7B, C). Centrifugal filter units with a 0.2 μm PTFE membrane were used to separate fibrillar material from Aβ(1-42) aggregation solutions. This was effectively done as the filtrate was devoid of fibrils (Fig 8B). Concentration measurements of the Aβ(1-42) solution pre- and post-filtering determined the filters removed 65% of the Aβ concentration with 35% remaining in the filtrate (data not shown). Separate control filtering experiments with monomeric Aβ indicated only a small loss (~10%) due to non-specific adsorption. Monocyte activation experiments comparing the total solution with the filtrate showed that 0.2 μm filtering of the Aβ(1-42) solution completely abolished the ability of Aβ(1-42) to induce proinflammatory activity (Fig 8C). It has been previously reported that Aβ(1-40) protofibrils will pass through a 0.2 μm filter (36) although in these studies no protofibrillar material was observed in the filtrate (Fig 8B).
The morphology, solubility, and transient appearance of the proinflammatory Aβ(1-42) species suggested similarities to protofibrils or fibrillar oligomers that have been described previously (10, 37). Fibrillar oligomers are conformationally related to fibrils but have been observed across a broad size distribution. Their size appears to overlap with prefibrillar oligomers yet the structural characteristics of the two oligomeric species are distinct. The ability to distinguish between these structural characteristics has been previously demonstrated with OC antisera, which recognizes an epitope common to fibrils and oligomeric fibrillar precursors of varying size (37). We used OC immune serum in this study to investigate whether fibrillar oligomers were involved in Aβ(1-42)-induced proinflammatory response. Aβ(1-42) solutions were immunodepleted of OC-positive species as described in the Methods and examined for remaining OC-positive material by dot blot. Although there was significant Aβ remaining in the supernatant, as detected by a sequence-specific Aβ antibody that is not dependent on conformation (Ab9), very little of it was OC-positive (Fig 9A). AFM analysis of the Aβ(1-42) supernatant after OC IP and centrifugation indicated a loss of both diffuse and fibrillar material (Fig 9B, C). Subsequent experiments looked at the ability of the OC-immunodepleted supernatants to stimulate TNFα production from THP-1 monocytes. Reprobing of the OC-immunodepleted supernatants with OC antisera again showed the amount of OC-positive Aβ(1-42) material was greatly reduced in the Aβ(1-42) solution after OC IP and centrifugation compared to just centrifugation or IP treatment with a rabbit IgG control (Fig 10A). Immunodepletion of OC-positive material in the Aβ(1-42) solution severely diminished the Aβ(1-42) proinflammatory activity compared to untreated 72 h Aβ(1-42) samples, 18,000g supernatants, or supernatants after IP with rabbit IgG and centrifugation (Fig 10B).
Separation techniques such as SEC allow separation of differently sized aggregates but may also suffer a preferential loss of a particular hydrophobic species due to adsorption to the column matrix or dilution-induced dissociation during column purification. OC-positive Aβ species elute across a broad size spectrum on SEC (37). To further support the ultracentrifugation studies that showed sedimentation of a significant percentage of fibrils without a concomitant loss of cellular activity, an Aβ(1-42)/water solution was incubated for 96 h at 4°C, centrifuged at 18,000g, and the supernatant was chromatographed on SEC (Fig 11A). Multiple peaks were observed by UV absorbance including elution peaks at the void volume (Vo), multiple included volumes (peaks 2 and 3), and monomer volume. Selected fractions were assessed for OC-reactivity (Fig 11B) using dot-blot analysis. OC-positive material was found in all tested peaks except for monomer. The OC-positive fractions were then tested for their ability to induce a proinflammatory response which showed that the included peak 2 induced the highest levels of secreted TNFα from THP-1 monocytes (Fig 11C). Subsequent SEC separations of Aβ(1-42) aggregated for 72-96 h were analyzed by DLS and the higher molecular weight fractions 11-14 showed an expected decrease in size (RH) from 100 nm to 10 nm as elution volume increased. The fractions with the highest proinflammatory activity corresponded to RH values between 10-30 nm and exhibited significant ThT fluorescence (data not shown). AFM images of a fraction from included peak 2 showed a significant population of short rod-like structures between 100-200 nm in length with a mean diameter (height) of 5.4 nm ± 1.6 SD for n= 116 measurements.
It has been postulated for some time that a sustained inflammatory response to aggregated Aβ may contribute to progressive neurodegeneration in AD (6). This idea emanated from pathology studies, which revealed inflammatory markers such as dystrophic neurites (2), activated microglia (3), and proinflammatory cytokines (5) surrounding Aβ lesions in the human AD brain (15). Interestingly, even though a vast array of Aβ aggregate morphologies ranging from dense core neuritic plaques to granular diffuse wispy Aβ deposits are observed in the AD brain, only the plaques appear to provoke this particular inflammatory response (1). A recent report by Meyer-Luehmann et al highlighted this phenomenon whereupon rapid plaque formation, and an equally rapid microglial response, was observed in an AD transgenic mouse model (4). Microglia were observed surrounding only the dense core plaques as opposed to the diffuse Aβ deposits.
THP-1 human monocytes have been used extensively to investigate the Aβ-induced proinflammatory response and display a similar pattern of activation to that of microglial cells (19-21). Many of the monocyte/macrophage and microglial studies have utilized preformed fibrillar Aβ and some of those studies included costimulators such as interferon-γ (38) or lipopolysaccharide (39) along with the Aβ treatment. In this study we correlated the time-dependent aggregation of Aβ with the ability to induce TNFα secretion from human THP-1 monocytes. We observed that an intermediate Aβ(1-42) species formed in acidic (pH 3.6-4) aqueous conditions was optimal for stimulating the response. While the peak cellular response coincided with the appearance of Aβ(1-42) fibrils, the cell response did not correlate with fibrillar species based on the observation that increased production of Aβ(1-42) fibrils, Aβ(1-40) fibrils, and Aβ(1-42) fibrils formed in PBS at neutral pH were not effective inducers of TNFα secretion. This observation in combination with the OC-immunodepletion studies and SEC-separation in Figures Figures1010 and and1111 respectively suggested that small fibrillar precursors were transient species that rapidly progressed to fibrils upon their formation. The neutral pH and increased ionic strength conditions in PBS accelerated Aβ(1-42) fibril formation but reduced the peptide’s ability to stimulate a proinflammatory response. There is some physiological support for Aβ(1-42) aggregates formed in acidic conditions. Although much of the attention is focused on the extracellular neuritic Aβ plaques, a significant amount of research now indicates that Aβ(1-42) aggregates may also form intracellularly in an acidic endosomal environment (40). These aggregates, if secreted, may form the structural basis for fibrils that are recognized by phagocytic cells and optimally induce proinflammatory events.
Significant structural polymorphism within Aβ fibrils at the molecular level has been demonstrated in vitro. Solid state NMR measurements revealed that Aβ(1-42), (1-40), and (10-35) fibrils contained in-register, parallel β-sheets (41, 42), while fibrils formed by the shorter peptides Aβ(16-22), (34-42), and (11-25) adopted anti-parallel β-strand alignments (43-45). The scope was expanded further by the observation that pH (45) and physical aggregation conditions (46) could also alter fibril structure and neuronal toxicity (46). Our finding that pH and ionic strength could alter Aβ(1-42) fibril morphology and its proinflammatory properties was consistent with these observations.
Structural differences at the molecular level are not always obvious by imaging techniques such as AFM or EM which are able to survey dimensional and architectural properties of aggregated Aβ. Recently, conformation-specific antibodies have been used to identify structural similarities between amyloid fibrils (47) and soluble oligomers (48) formed from different proteins. These antibodies are able to recognize a particular assembly state and can distinguish structural differences between monomeric, oligomeric, and fibrillar Aβ species (37, 47). OC antisera recognizes fibrils and fibrillar oligomers, which have been described as small soluble aggregates that are conformationally related to mature fibrils (37). Prefibrillar oligomers, which are recognized by the A11 antibody, but not OC, are structurally distinct from fibrillar oligomers (37). The immunodepletion studies in this report demonstrate that selective removal of fibrillar oligomers from an Aβ(1-42) solution by OC antibodies significantly lowered the THP-1 monocyte proinflammatory response to the peptide (Fig 10). This result indicated that an aggregated Aβ(1-42) species with inherent components of fibril structure is necessary to induce TNFα secretion but smaller units of this structure induce the best response. This possibility is supported by the observation that continued, or accelerated, Aβ(1-42) aggregation diminished the monocyte response (Figs (Figs11 and and3).3). Furthermore, the time course suggests that the fibrillar oligomers are transient and appear almost simultaneously with fibrils but rapidly disappear as they likely elongate to form additional fibrils. Fibrillar oligomers may be precursors to protofibrils (10-12) and have similar structural properties although further investigation will be needed in order to make a careful comparison. The inability of Aβ(1-40) to stimulate a proinflammatory response under the same conditions as Aβ(1-42) tested in this study may be due to a lower nucleation propensity which would produce a substantially lower concentration of fibrillar oligomers.
Numerous studies suggest that small Aβ(1-42) oligomers may cause early and significant alterations in synaptic function and then as fibrillar structures are formed, concomitant inflammatory responses appear (reviewed in (49)). This description of Aβ(1-42) aggregation progression towards an inflammatory species is consistent with the data in this report demonstrating that a soluble fibrillar precursor or nuclei is optimal for inducing an inflammatory response in a human monocyte cell line. Plaques consist of fibrillar Aβ at the core (50) although the complexities of plaque composition appear to be quite significant. Hyman and colleagues have characterized Aβ plaques as a reservoir of bioactive molecules and proposed that soluble Aβ species surround the plaques (4). New evidence now indicates a halo of oligomeric Aβ surrounding the plaques based on immunostaining with NAB61 antibody (51). NAB61 is able to recognize both oligomeric and fibrillar pathologic forms of Aβ but does not effectively stain diffuse Aβ (52). These studies make a case that different Aβ aggregation states exist not only throughout the brain parenchyma but even within the plaque area. Our studies in a human monocyte cell line show that soluble fibrillar Aβ(1-42) precursors are optimal for triggering an inflammatory response. These findings provide further information into the complexities of Aβ aggregation, inflammation, and the most favorable Aβ structure for interacting with cell surface receptors.
We greatly appreciate the gift of OC antisera from Dr. Rakez Kayed (University of Texas Medical Branch, George and Cynthia Mitchell Center for Neurodegenerative Diseases, Department of Neurology) and the Ab9 antibody from Dr. Terrone Rosenberry, Mayo Clinic Jacksonville). We would like to thank the Microscopy Image and Spectroscopy Technology Laboratory in the Center for Nanoscience at University of Missouri-St. Louis for technical assistance and equipment.
†This work was supported by grants NIRG-06-27267 (MRN) from the Alzheimer’s Association, R15AG033913 (MRN) from the National Institute on Aging, and by a University of Missouri-St. Louis Dissertation Fellowship (MLDU).
1Abbreviations and Footnotes