Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 Mar 1.
Published in final edited form as: J Neuroimmune Pharmacol. 2012 Dec 16;8(1):312–322. doi: 10.1007/s11481-012-9424-6

Amyloid-β(1–42) Protofibrils Formed in Modified Artificial Cerebrospinal Fluid Bind and Activate Microglia

Geeta S Paranjape 1,‡, Shana E Terrill 1,‡, Lisa K Gouwens 1,‡, Benjamin M Ruck 1, Michael R Nichols 1,*
PMCID: PMC3587657  NIHMSID: NIHMS429711  PMID: 23242692

Abstract

Soluble aggregated forms of amyloid-β protein (Aβ) have garnered significant attention recently for their role in Alzheimer’s disease (AD). Protofibrils are a subset of these soluble species and are considered intermediates in the aggregation pathway to mature Aβ fibrils. Biological studies have demonstrated that protofibrils exhibit both toxic and inflammatory activities. It is important in these in vitro studies to prepare protofibrils using solution conditions that are appropriate for cellular studies as well as conducive to biophysical characterization of protofibrils. Here we describe the preparation and characterization of Aβ(1–42) protofibrils in modified artificial cerebrospinal fluid (aCSF) and demonstrate their prominent binding and activation of microglial cells. A simple phosphate/bicarbonate buffer system was prepared that maintained the ionic strength and cell compatibility of F-12 medium but did not contain numerous supplements that interfere with spectroscopic analyses of Aβ protofibrils. Reconstitution of Aβ(1–42) in aCSF and isolation with size exclusion chromatography (SEC) revealed curvilinear β-sheet protofibrils <100 nm in length and hydrodynamic radii of 21 nm. Protofibril concentration determination by BCA assay, which was not possible in F-12 medium, was more accurately measured in aCSF. Protofibrils formed and isolated in aCSF, but not monomers, markedly stimulated TNFα production in BV-2 and primary microglia and bound in significant amounts to microglial membranes. This report demonstrates the suitability of a modified aCSF system for preparing SEC-isolated Aβ(1–42) protofibrils and underscores the unique ability of protofibrils to functionally interact with microglia.

Keywords: Amyloid-beta protein, aggregation, protofibrils, microglia, inflammation

Introduction

Alzheimer’s disease (AD) is a complex neurodegenerative disorder that is fast becoming the number one health concern in the US. The complexity of AD comes from the observation of two distinct protein lesions in the affected brain (Selkoe 2004). These lesions include senile plaques composed of amyloid-β protein (Aβ) and neurofibrillary tangles (NFTs) composed of tau protein. Much of the data supports the hypothesis that Aβ accumulation is the initial pathogenic event followed by NFT formation. Accordingly, the relationship between the two proteins has been described as Aβ being more clearly linked to the cause of AD while tau is more clearly associated with the clinical manifestations of AD (Dickson 2004). Adding to the complexity is the detection of different structural forms of Aβ (soluble and insoluble) in brain tissue. Although insoluble plaques consisting of fibrils are the most prominent type of Aβ deposit in AD, significant attention has now been turned to soluble Aβ forms (Haass and Selkoe 2007; Walsh and Selkoe 2007). These oligomeric species are present in AD brain tissue and cerebrospinal fluid (Kayed et al. 2003; Georganopoulou et al. 2005; Lee et al. 2006; Jin et al. 2011) and may play an important role in cognitive dysfunction (Cleary et al. 2005). The two primary Aβ fragments produced in vivo are Aβ(1–40) and Aβ(1–42) (Selkoe 2004) yet the senile plaques consist overwhelmingly of Aβ(1–42) (Gravina et al. 1995).

Aβ aggregation mechanisms have been studied extensively in vitro and significant evidence confirms a greater aggregation propensity (Jarrett et al. 1993) and a more diverse population of aggregated species (Bitan et al. 2003) by Aβ(1–42). The process by which monomers self-assemble into mature fibrils proceeds through a variety of soluble intermediates including oligomers (Dahlgren et al. 2002; Kayed et al. 2003) and protofibrils (Harper et al. 1997; Walsh et al. 1997). As small soluble precursors to Aβ fibrils, protofibrils are fairly well characterized. They are enriched in β-sheet structure and display toxicity against neurons (Walsh et al. 1999). Aβ protofibrils have also been shown to disrupt ion channels (Ye et al. 2003) and inhibit hippocampal long-term potentiation (O'Nuallain et al. 2010).

A notable component of AD pathology is the presence of inflammation in the affected regions. Inflammation is particularly intense surrounding the Aβ plaques and is evidenced by accumulation of activated microglia (McGeer et al. 1987) and the presence of proinflammatory cytokines (Dickson et al. 1993). A chronic inflammatory state in AD has been suggested as an underlying mechanism of progressive neurodegeneration (McGeer and McGeer 1998). It has been widely demonstrated that Aβ can stimulate microglia in vitro (Meda et al. 1995; Combs et al. 2001; Stewart et al. 2010) which enhances the testing of different structural forms of Aβ as well as different Aβ solution conditions for the ability to trigger an inflammatory response in microglia.

We have previously described robust microglial activation by Aβ(1–42) protofibrils in both BV-2 and primary murine microglia (Paranjape et al. 2012). In that study, the protofibrils were formed, and isolated by SEC, in F-12 medium (without phenol red) supplemented with antibiotics. The use of this medium was compatible with cell treatment, maintained a physiologically relevant pH and ionic strength, and was a passable model for cerebrospinal fluid (CSF). However, there were spectroscopic limitations due to the presence of light-absorbing compounds in the F-12 medium and there were simpler published protocols for artificial CSF (aCSF) that did not contain the complex mixture of components present in F-12 medium. In this report we describe the preparation and characterization of Aβ(1–42) protofibrils in a modified aCSF buffer system and their subsequent evaluation in both microglial binding and activation.

Experimental Procedures

Preparation of Aβ Peptides

Aβ(1–42) was obtained from W. M. Keck Biotechnology Resource Laboratory (Yale School of Medicine, New Haven, CT) in lyophilized form and stored at −20° C. Aβ(1–42) peptides were dissolved in 100% hexafluoroisopropanol (HFIP) (Sigma-Aldrich, St. Louis) at 1 mM, separated into aliquots in sterile microcentrifuge tubes, and evaporated uncovered at room temperature overnight in a fume hood. The following day the aliquots were vacuum-centrifuged to remove any residual HFIP and stored in dessicant at −20° C.

Size Exclusion Chromatography

Lyophilized Aβ (1 mg) was dissolved in 50 mM NaOH to yield a 2.5 mM Aβ solution. The solution was then diluted to 250 µM Aβ in prefiltered (0.22 µm) artificial cerebrospinal fluid, centrifuged at 18,000g for 10 min with a Beckman-Coulter Microfuge 18 and the supernatant was fractionated on a Superdex 75 HR 10/30 column (GE Healthcare) using an AKTA FPLC system (GE Healthcare). Prior to injection of Aβ, the Superdex 75 column was coated with 2 mg of bovine serum albumin taken from a sterile 7.5% fraction V solution (Sigma) to prevent any non-specific binding of Aβ to the column matrix. Following a 1 mL loading of the sample, Aβ was eluted at 0.5 mL min−1 in aCSF and 0.5 mL fractions were collected and immediately placed on ice. Aβ concentrations were determined in line by UV absorbance using an extinction coefficient of 1450 cm−1 M−1 at 280 nm. Aβ(1–42) fibrils were prepared by incubation of SEC-purified monomer at room temperature under gentle agitation at 25° C. Conductivity measurements were made by injecting different buffers or medium through the AKTA FPLC UPC-900 conductivity monitor until a stable conductivity (mS/cm) was obtained.

Thioflavin T Fluorescence

Aβ solutions were assessed by ThT fluorescence as described previously (Nichols et al. 2002). Aβ aliquots were diluted to 5 µM in aCSF pH 7.8 containing 5 µM 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 obtain ThT relative fluorescence values. Buffer controls did not show any significant ThT fluorescence in the absence of Aβ. All ThT fluorescence numbers are reported in relative fluorescence units.

Dynamic Light Scattering

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. Average RH values were obtained with Dynamics software (version 6.7.1). Histograms of percent intensity vs. RH were generated by Dynamics data regularization and intensity-weighted or mass-weighted mean RH values were derived from the regularized histograms.

Electron Microscopy

SEC-purified Aβ(1–42) protofibrils (10 µl) were applied to a 200-mesh formvar-coated copper grid (Ted Pella, Inc.). Samples were allowed to adsorb for 10 minutes at 25°C, followed by removal of excess sample solution. Grids were washed three times by placing sample side down on a droplet of water. Heavy metal staining was done by incubation of the grid on a droplet of 2% uranyl acetate (Electron Microscopy Sciences, Hatfield, PA) for 5 minutes, removal of excess solution, and air drying. Affixed samples were visualized with a JEOL JEM-2000 FX transmission electron microscope operated at 200k eV.

Protein Concentration Determination

In addition to in-line UV absorbance, a bicinchoninic acid (BCA) assay was also utilized to measure Aβ concentrations. A standard curve was constructed using SEC-purified Aβ(1–40) monomer diluted in aCSF typically in the concentration range of 5 µM to 80 µM. Aβ(1–40) standards or Aβ(1–42) samples (10 µl) were added to 100 µl of 50:1 mixture of BCA Reagent A to Reagent B (Thermo Fisher Scientific) and mixed thoroughly in individual wells of a 96 well plate. The plate was then incubated covered for 30 min at 37°C followed by cooling for an additional 15 min. Absorbance was then determined for each well at a wavelength of 562 nm. Aβ(1–40) standard curves were also evaluated using the Bradford method (Bradford 1976). Briefly, 10 µl from Aβ(1–40) standard solutions ranging from 20–60 µM were mixed with 1.5 ml 1× Bradford reagent (BioRad), incubated for 10 min, and the absorbance was determined at 595 nm. Protein concentration was also determined by quantitative amino acid analysis (QAAA) at the W. M. Keck Biotechnology Resource Laboratory (Yale School of Medicine, New Haven, CT) for representative samples. Samples of SEC-isolated Aβ(1–42) protofibrils and monomers were placed in siliconized tubes, sealed, packed within a conical tube and shipped overnight for analysis.

Cell Culture and Primary Microglia Isolation

BV-2 murine microglial cells were cultured as previously described (Paranjape et al. 2012) in Dulbecco’s modified Eagle’s medium (DMEM, 4.5 g/L glucose) (Hyclone) containing 50 U/ml penicillin, 50 µg/ml streptomycin, 50 µM β-mercaptoethanol, and 5% fetal bovine serum (FBS, Hyclone). Primary murine microglia were obtained from 3–4 day old C57BL/6 mouse pups as previously described (Paranjape et al. 2012). Briefly, brains were isolated and meninges were removed under sterile conditions. Minced and trypsinized brain tissue was resuspended in complete DMEM containing 10% fetal bovine serum, 4 mM L-glutamine, 100 U/ml penicillin, 0.1 mg/ml streptomycin and 0.25 µg/mL amphotericin-B (Fisher Scientific), OPI medium supplement (oxalocetate, pyruvate, insulin, Sigma-Aldrich), and 0.5 ng/ml recombinant mouse granulocyte-macrophage colony-stimulating factor (GM-CSF) (Invitrogen). The cell suspension was filtered, centrifuged, resuspended in complete medium and seeded into 150 cm2 flasks (Corning). Cells were cultured at 37°C in 5% CO2 until confluent (1–2 weeks) and microglia were selectively harvested from the adherent astrocyte layer by overnight shaking of the flask at 37°C in 5% CO2 and collection of the medium. The flasks were replenished with fresh medium, and incubated further to obtain additional microglia. Typically, this procedure was repeated 3–4 times for one flask without removal of the astrocyte layer.

Cell Stimulation Assay

For cellular studies, BV-2 murine microglia were removed from culture flasks with 0.25% trypsin and seeded in a sterile 96-well cell culture plate overnight at a density of 5 × 105 cells/ml in growth medium described above. Primary murine microglia were collected as described above by overnight shaking, collection of the cells and seeding in a sterile 96-well cell culture plate for 24 hrs at a density of 5 × 105 cells/ml in growth medium without serum or GM-CSF. Prior to cell treatment of both microglia cell types, medium was replaced with fresh serum-free medium followed by Aβ stimulation at a final concentration of 15 µM unless noted otherwise. The cells were incubated at 37°C for 6 h in 5% CO2 and the medium was collected and stored at −20°C for subsequent analysis by enzyme-linked immunosorbent assay (ELISA). The background cellular response was assessed using the particular buffer vehicle for the Aβ.

ELISA

Measurement of secreted TNFα in the supernatants was determined by ELISA as previously detailed (Udan et al. 2008). Briefly, 96-well plates were coated overnight with monoclonal anti-mouse TNFα capture antibody, washed with phosphate-buffered saline (PBS) containing 0.05% Tween-20 and blocked with PBS containing 1% BSA, 5% Sucrose and 0.05% NaN3 following by a wash step. Successive treatments with washing in between were done with samples or standards, biotinylated polyclonal anti-mouse TNFα detection antibody in 20 mM Tris with 150 mM NaCl and 0.1% BSA, streptavidin-horseradish peroxidase (HRP) conjugate, and equal volumes of HRP substrates 3,3',5,5'-tetramethylbenzidine and hydrogen peroxide. The reaction was stopped by the addition of 1% H2SO4 solution. The optical density of each sample was analyzed at 450 nm with a reference reading at 630 nm using a SpectraMax 340 absorbance plate reader (Molecular Devices, Union City, CA). The concentration of TNFα in the experimental samples was calculated from a mouse TNFα standard curve of 15–2000 pg/ml. When necessary, samples were diluted to fall within the standard curve. TNFα concentrations for absorbance values below the lowest 15 pg/ml standard were determined by extrapolation of the standard curve regression line.

Microglial Binding Assay

BV-2 murine microglia cells (0.1 ml, 5×105 cells/ml were plated in individual MatTek P35GC-1.5-14-C dishes (MatTek Corp., Ashland, MA) overnight in growth medium described above. In some cases cells were plated in MatTek P12G-1.5-14-F 12-well plates and no observable differences were noted. Cells were treated as described above in the cell stimulation assay procedures with either Aβ (15 µM) or aCSF and incubated for 30 min at 4°C (on ice), 25°C, or 37°C. Cells were then rinsed twice with serum-free growth medium and fixed with 3.7% formaldehyde. Cell samples were then washed 3× with PBS containing 0.05% Tween 20 (PBST) prior to a 1 hr incubation with blocking buffer (10% w/v dried milk in PBST) and after each subsequent step. All incubations were conducted with gentle shaking at 25° C. Samples were then treated for 1 hr with the anti-Aβ antibody Ab9 (1:5000 dilution in PBST with 5% w/v dried milk) followed by 1 hr with donkey anti-mouse IgG antibody conjugated to Northern Lights (NL) 493 (R&D Systems). Cell nuclei were visualized after incubation for 5 min with 0.2 mL 0.3 µM 4',6-diamidino-2-phenylindole dihydrochloride (DAPI). After a final wash step, 0.2 mL PBS was applied to the wells to avoid cellular dehydration. Both two-dimensional and Z-stack images were obtained with a Zeiss LSM 700 confocal microscope at 40X magnification using ZEN 2009 imaging and analysis software. Excitation wavelengths used were 358 nm and 493 nm for DAPI and NL493 respectively. For comparison experiments, all optical and electrical settings were fixed at identical levels for each image to ensure that fluorescence intensities were not influenced by different settings.

Results

Comparison between F-12 and aCSF medium

Aβ protofibrils have been prepared in a variety of buffers at physiological pH (Harper et al. 1997; Walsh et al. 1997) or slightly higher (Paranjape et al. 2012). We sought to formulate a simple buffering system that was a good model for aCSF and then characterize the formation of Aβ protofibrils in this system. Somewhat different protocols have been used in previous preparations of aCSF but most typically utilize a phosphate/bicarbonate buffering system with additional ionic strength provided by KCl and NaCl (McNay and Gold 1999; Tzounopoulos et al. 2004). Additional common components found in both F-12 and aCSF are Mg2+, Ca2+ and glucose. We examined and compared these formulations to guide our own preparation of a modified aCSF system (Table 1). There have been no reports of a Mg2+ or Ca2+ requirement for Aβ aggregation so those ions were not included in the current modified aCSF preparation. Glucose was also excluded to reduce the potential for bacterial contamination. The pH was adjusted in order to be consistent with the measured pH (7.8) of F-12 medium utilized in our previous studies (Paranjape et al. 2012). Conductivity measurements were done on F-12 medium, PBS, and the modified aCSF used in this report to determine if overall ionic strength in these systems was within the same range. As expected, purified water (18 MΩ resistance) exhibited no conductivity (0 mS/cm) while F-12 medium, PBS, and modified aCSF had conductivities of 16 mS/cm, 21 mS/cm, and 15 mS/cm respectively. A control solution of 150 mM NaCl gave a value of 16 mS/cm.

Table 1.

Formulations of F-12, PBS, and aCSF. All concentrations are in mmol/L.

Component F-12a PBSb aCSFc aCSFd aCSFe
NaCl 131 154 128 130 130
KCl 3 --- 3 3 3
NaH2PO4 1 6.7 1.3 1.2 1
NaHCO3 14 --- 21 20 15
MgCl2 0.003 --- 1 1.3 ---
CaCl2 0.3 --- 1.3 2.4 ---
Glucose 10 --- 0–3 10 ---
Open in a new tab
a

- Gibco Life / Technologies;

b

- HyClone / Thermo Scientific

e

- current investigation

Characterization of protofibrils formed and SEC-isolated in aCSF

Reconstitution of HFIP-treated and lyophilized Aβ(1–42) in NaOH followed by dilution into aCSF at an Aβ concentration of 250 µM resulted in the rapid formation of protofibrils which were then isolated by SEC on a Superdex 75 column (Fig 1A). The rate of Aβ(1–42) protofibril formation in aCSF was similar to that previously observed in F-12 (Paranjape et al. 2012) and the distribution between protofibrils and monomers was also similar (Fig 1A). From these results it was determined that the absence of Mg2+, Ca2+, and glucose did not negatively affect the extent of protofibril formation. The preparative nature of the SEC isolation allowed further characterization of Aβ(1–42) protofibrils. ThT binding and fluorescence has been used previously as an indirect measure of β-sheet structure in soluble protofibrillar species (Walsh et al. 1999). We have also used this technique to characterize SEC-isolated protofibrils in F-12 medium (Paranjape et al. 2012). Analysis of the SEC fractions following aggregation and elution in aCSF revealed that freshly-purified monomers (5 µM) did not exhibit ThT binding and fluorescence while protofibril fractions (5 µM) displayed significant ThT fluorescence (Fig 1B). These findings were expected and demonstrated that the modified aCSF preparation did not significantly alter the Aβ(1–42) protofibril secondary structure. Transmission electron microscopy (TEM) imaging of the aCSF-eluted SEC void peak showed archetypal short curvilinear protofibrils(Walsh et al. 1997; Kheterpal et al. 2003) <100 nm in length (Fig 1C). Dynamic light scattering (DLS) analysis was conducted on the Aβ(1–42) protofibril peak in Fig 1A and a mean hydrodynamic radius (RH) of 19 nm was determined from a cumulant fit of the autocorrelation curve constructed from 26 5-sec acquisitions. A histogram generated from the regularized autocorrelation data gave a RH value of 23 nm (Fig 1D). Mean cumulant-derived RH measurements of Aβ(1–42) protofibrils from five separate SEC purifications produced an average value of 21.9 ± 4.0 nm standard deviation (data not shown). These size data for Aβ(1–42) protofibrils formed and isolated in aCSF are extremely close to those obtained in F-12 medium (Paranjape et al. 2012) and are quite consistent between replicate preparations.

Figure 1.

Figure 1

Open in a new tab

Characterization of Aβ(1–42) protofibrils formed in, and isolated by SEC, in aCSF. Panel A. Lyophylized Aβ(1–42) (1 mg) was brought into solution with NaOH followed by dilution into aCSF at a final concentration of 250 µM. The supernatant after centrifugation was eluted from a Superdex 75 column and 0.5 ml fractions were collected. UV absorbance at 280 nm was monitored during the elution (solid line). Panel B. Freshly isolated Aβ(1–42) protofibrils and monomers after elution from Superdex 75 in aCSF were diluted to 5 µM in aCSF containing 5 µM ThT and fluorescence emission was measured as described in the Methods. Panel C. Protofibrils (5 µl, 52 µM) applied to a copper formwar grid, and imaged by TEM at a magnification of 59,000. The scale bar represents 100 nm. Panel D. Aβ(1–42) protofibrils were analyzed immediately after SEC-isolation by DLS and a representative regularized histogram of percent intensity versus RH is shown. Intensity-weighted mean RH values were derived from the regularized histograms which for the above data was 23 nm.

Fibril formation in aCSF

The ability of SEC-purified Aβ(1–42) monomer in aCSF to form fibrils was also assessed. Freshly-purified monomer (58 µM) was incubated at 25° C under gentle agitation conditions for 3 days. A significant level of ThT fluorescence was observed at the end point of the incubation and 2/3 of this material was removed from the supernatant following centrifugation at 18,000g for 10 min. Isolation of the Aβ(1–42) pellet and resuspension in aCSF recovered the ThT fluorescence. TEM images of the Aβ(1–42) pellet resuspension revealed long fibers with lengths up to 1 µm and typical widths of 5–10 nm (Fig 2).

Figure 2.

Figure 2

Open in a new tab

SEC-purified Aβ(1–42) monomer in aCSF forms mature fibrils. A solution of freshly purified monomer (58 µM) in aCSF was incubated for 72 h at 25° C with gentle agitation. The solution was centrifuged, supernatant removed and the remaining pellet resuspended in the same volume of aCSF. TEM images of the isolated Aβ(1–42) fibril pellet (58 µM) were obtained at a magnification of 35,000. The scale bar represents 100 nm.

Concentration determination of Aβ(1–42) protofibrils

Aβ concentrations were determined during each SEC-purification using in-line UV absorbance. Previous studies have established that protofibril concentrations can be significantly over-estimated using UV absorbance due to light scattering (Nichols et al. 2002). A BCA assay has been described as an alternative and effective measure of Aβ concentration (Jan et al. 2010). Our use of the BCA assay to determine Aβ(1–42) protofibril concentrations was hampered in previous studies where protofibrils were formed and isolated in F-12 medium. This was due to interference from light-absorbing molecules in F-12 which produced a high background signal and a low signal to noise ratio. The preparation of protofibrils in aCSF allowed BCA assay determination of Aβ concentrations in the low micromolar range. Aβ concentration determination by Bradford assay was also investigated and while this method is less-used than BCA for Aβ studies, standard curves prepared from SEC-purified Aβ(1–40) monomer were nearly as sensitive (data not shown) and affected to a smaller extent by F-12 medium. These findings suggested that the Bradford assay may serve as an alternative for BCA in some situations but the considerable reproducibility and sensitivity of the BCA assay prompted its use. Standard curves were established using freshly-purified Aβ(1–40) monomer solutions (Fig 3A) and a linear regression of the data allowed subsequent measurement of protofibril and monomer concentrations. BCA assay analysis confirmed that concentrations of SEC-purified Aβ(1–42) monomer fractions closely matched those determined by the UV absorbance (Fig 3B). This was not the case with Aβ(1–42) protofibrils as comparison measurements showed that protofibril concentrations determined by BCA assay were typically less than half than those determined by UV absorbance. In a representative experiment, SEC-isolated Aβ(1–42) protofibrils were measured at 47 µM by UV absorbance but only 19 µM by BCA assay (Fig 3B). The preparation and isolation of Aβ(1–42) protofibrils in aCSF also made concentration estimation by quantitative amino acid analysis (QAAA) feasible. This was not possible in F-12 medium due to the presence of numerous amino acids that interfere with QAAA. The SEC-isolated Aβ(1–42) protofibrils shown in Fig 3B were also analyzed by QAAA (Yale University Keck Laboratory) and the Aβ concentration was determined to be 18 µM, a value very close to that obtained from BCA analysis. Monomer concentrations determined by QAAA were much closer to the UV absorbance-determined values.

Figure 3.

Figure 3

Open in a new tab

BCA assay standard curve using Aβ(1–40) monomer. Panel A. Standard solutions of SEC-purified Aβ(1–40) monomer (70, 50, 25, 15, and 7.5 µM) were prepared and subjected to BCA assay analysis as described in the Methods. Each data point represents the mean absorbance for n=3 measurements and error bars signifying standard error for the triplicates are shown. Linear regression analysis of the data points for the standards gave a correlation coefficient (r2) of 0.9987. Panel B. Concentrations (µM) of SEC-isolated Aβ(1–42) monomer and protofibrils were determined by BCA assay using a standard curve constructed as shown in panel A. Duplicates of each sample were assessed and compared to the UV absorbance-determined (280 nm) concentration.

Microglial activation and binding by protofibrils

Aβ(1–42) protofibrils formed, and SEC-isolated, in aCSF were effective stimulators of TNFα production in microglia compared to Aβ(1–42) monomers (Fig 4). Both BV-2 and primary murine microglia were responsive to the protofibrils at a UV-determined concentration of 15 µM. Monomers at the same concentration were much less effective. Subsequent BCA analysis indicated the protofibril concentration was closer to 7 µM while the monomer concentration was confirmed at 15 µM (data not shown).The data demonstrated that Aβ(1–42) protofibrils formed in aCSF exhibited similar inflammatory activity as those formed in F12 medium.

Figure 4.

Figure 4

Open in a new tab

Protofibrils formed and isolated in aCSF are significant stimulators of microglia. SEC-isolated Aβ(1–42) protofibrils and monomers in aCSF were incubated with primary (Panel A) and BV-2 (Panel B) microglia at a final concentration of 15 µM (UV absorbance-determined) for 6 hrs in serum-free medium. Secreted TNFα was measured by ELISA in the conditioned medium. Data bars represent the average ± std error of n=6 trials. Control treatments with an equal volume of aCSF produced TNFα levels of 17 and 1 pg/ml for primary and BV-2 microglia respectively and were subtracted from Aβ-stimulated samples.

In order for Aβ protofibrils to activate microglia it is likely there is a physical interaction between the peptide and the cells that leads to a functional response. The ensuing studies investigated binding of Aβ(1–42) protofibrils to the cell membrane of BV-2 murine microglial cells. Microglia seeded in glass bottom culture dishes were treated with SEC-isolated Aβ(1–42) protofibrils, fibrils, monomers or aCSF buffer control. The dishes were then placed on ice for 30 min to prevent cellular uptake or phagocytosis of Aβ. The cells were then fixed and immunostained for Aβ (NL493-conjugated secondary antibody). Cell nuclei were also chemically stained with DAPI. All four cell samples were imaged with both NL493 and DAPI excitation channels. Aβ(1–42) immunostaining (NL493 fluorescence) was observed in the protofibril-treated, but not monomer- or fibril-treated, cells (Fig 5). DAPI staining confirmed the presence of microglia in all four cell samples. Several areas of intense Aβ staining were noted in the protofibril-treated sample in addition to moderate but widespread Aβ staining around the microglial surface. A Z-stack representation is provided in panel 5C that clearly demonstrates the cell surface-specific binding of the Aβ(1–42) protofibrils to the microglia. Figure 5 is a representative experiment of several that were conducted in which Aβ(1–42) protofibrils preferentially localized to the microglial membrane. The initial protofibril-microglia binding studies were done at 4°C to inhibit Aβ phagocytosis. This procedure allowed us to determine that monomers or fibrils do not bind in significant amounts to microglia compared to protofibrils. Subsequent experiments were conducted at either 25°C or 37°C and very little internalization of the Aβ(1–42) protofibrils was observed. In fact, confocal imaging revealed significant Aβ(1–42) protofibril binding to microglia after a 30 min incubation at 37°C (Fig 6). Z-stack analysis showed significant accumulation of protofibrils specifically at the cell surface. No observable difference in Aβ(1–42) protofibril binding was seen whether the interaction was performed at 37°C or 25°C (data not shown). In order to further correlate Aβ(1–42) protofibril binding with microglial activation, two different Aβ concentrations were examined. BV-2 microglia were treated with either 1 µM or 15 µM Aβ(1–42) protofibrils for 30 min and both TNFα production and microglia binding were assessed. Microglial cells were activated to a greater extent in the presence of the higher Aβ(1–42) protofibril concentration (Fig 7A). 15 µM Aβ(1–42) protofibrils induced a significantly higher level of secreted TNFα compared to 1 µM. The dose-dependent effect on microglial activation was reflected in the extent of Aβ(1–42) protofibril binding to the cell surface. Confocal fluorescence images indicated a relatively low level of binding when BV-2 cells were treated with 1 µM Aβ(1–42) protofibril (Fig 7B, C) compared to the significant binding level observed with 15 µM (Fig 7D). These dose-dependent findings demonstrate a correlation between Aβ(1–42) protofibril binding to microglia and microglial activation and provides additional evidence linking the two processes.

Figure 5.

Figure 5

Open in a new tab

Aβ(1–42) protofibrils, but not monomers or fibrils, bind to microglia. BV-2 microglia were treated with 15 µM Aβ(1–42) protofibrils (PF), fibrils (F), monomers (M) or aCSF buffer control (C) for 30 min at 4°C as described in the Methods. Cell samples were probed with anti-Aβ antibody Ab9 followed by anti-mouse IgG-NL493 conjugate (green). As a final step, cell nuclei were stained with DAPI (blue). Confocal images above were obtained at 40X magnification. The scale bar represents 16 µm. Panel C is a Z-stack representation of PF bound to the microglia cell surface. The thin image on the right side of the panel is a 3-dimensional slice corresponding to the red line while the thin image at the top is a 3-dimensional slice corresponding to the green line.

Figure 6.

Figure 6

Open in a new tab

Aβ(1–42) protofibrils bind to microglia at physiological temperature. BV-2 microglia were treated with 15 µM Aβ(1–42) protofibrils (PF) for 30 min at 37°C. The cell sample was probed as described in Figure 5 legend. A Z-stack confocal image was obtained at 40X magnification (scale bar 16 µm). The thin image on the right side of the panel is a 3-dimensional slice corresponding to the red line while the thin image at the top is a 3-dimensional slice corresponding to the green line.

Figure 7.

Figure 7

Open in a new tab

Aβ(1–42) protofibrils bind and activate microglia in a dose-dependent manner. BV-2 microglia (three separate wells for each condition) were treated with either 1 or 15 µM Aβ(1–42) protofibrils for 30 min at 25°C (trial #1) or 37°C (trial #2). Panel A. Secreted TNFα was measured from each sample (n=6 for each condition). The difference in TNFα production between the two concentrations was statistically significant (*p<0.01). Confocal images were obtained for three separate wells of BV-2 cells treated for 30 min at 37°C at the two concentrations above. Representative images acquired at 40X magnification (scale bar 16 µm) are shown of cells treated with 1 µM (panel B and C) or 15 µM Aβ(1–42) protofibrils (panel D).

Discussion

Aβ protofibrils have been generated under different solution conditions with most preparations using either Tris-HCl or phosphate as the buffering agent at a pH of 7.4 although the concentrations were not always identical (Harper et al. 1997; Walsh et al. 1997; Harper et al. 1999; Walsh et al. 1999; Kheterpal et al. 2003). Some of the preparations included additional ionic strength from NaCl or KCl as well as an antibacterial such as sodium azide. All of these studies isolated protofibrils with SEC and obtained very similar species as determined by TEM. More recent studies have diluted isolated protofibrils into appropriate buffers for neuronal toxicity studies (Jan et al. 2010). We recently took this preparation a step further by forming and isolating protofibrils in phenol red-free F-12 cell culture medium supplemented with antibiotics (Paranjape et al. 2012). The complex mixture of numerous vitamins and amino acids (some light-absorbing) in F-12 medium hampered some spectroscopic analyses so a simpler phosphate/bicarbonate system was tested in the current study for formation and isolation of protofibrils. The SEC-isolated Aβ(1–42) protofibrils readily enhanced ThT fluorescence, exhibited classical protofibrillar morphological features by TEM, and were of similar size (RH) as those prepared in F-12 medium.

Aβ protofibrils are toxic to neurons (Walsh et al. 1999) and have been shown to induce alterations in neuronal and glial excitatory postsynaptic currents (Hartley et al. 1999). Futhermore, Aβ protofibrils were also found to disrupt ion channels in rat cortical neurons (Ye et al. 2003) and inhibit hippocampal long-term potentiation in mouse hippocampal slices (O'Nuallain et al. 2010). These studies demonstrate direct toxicity of protofibrils to neurons yet there are additional mechanisms by which Aβ protofibrils can be deleterious to neurons. We recently found that SEC-isolated Aβ(1–42) protofibrils, but not fibrils or SEC-purified monomers, stimulated a robust inflammatory response in murine microglia (Paranjape et al. 2012). Previous studies have determined that conditioned medium from microglia is toxic to neuronal cells suggesting that one way in which Aβ may create a toxic environment is via activated microglial cells and secretion of proinflammatory products. In the current study we demonstrate the formation and isolation of protofibrils in a modified aCSF solution and strengthen the case for Aβ(1–42) protofibrils as potent microglia activators.

There is a significant amount of data demonstrating microglial activation by Aβ but fewer reports are available on Aβ binding to microglia. Using THP-1 monocytes, a well-established model for microglia, Landreth and coworkers identified a multireceptor complex comprising the SR-B receptor CD36, α6β1-integrin, and the integrin-associated protein CD47 that mediated monocyte binding to fibrillar Aβ and initiation of the proinflammatory response (Bamberger et al. 2003). The interaction was assessed using a quantitative cell adhesion assay that determined monocyte binding to fibrillar Aβ pre-adsorbed to a glass slide. Similar components of the multireceptor complex and the proinflammatory response were also observed in primary murine microglia. Although the binding was established, no images of Aβ binding to microglia were presented. Liu et al. recently demonstrated binding of oligomeric Aβ(1–42) to microglia using confocal microscopy (Liu et al. 2012) however the binding was punctate and not especially widespread yet colocalized with Toll-like receptor (TLR) 2. In comparison, the binding of Aβ(1–42) protofibrils to BV-2 microglia in our current study was substantial and covered a significant amount of cellular surface area. These differences may reflect differences between Aβ binding to BV-2 and primary microglia or differences between protofibrils in our study and oligomers in Liu et al. We have previously shown that TLRs mediate Aβ activation of THP-1 human monocytes (Udan et al. 2008) and others have demonstrated this same role for TLRs in microglia (Fassbender et al. 2004; Reed-Geaghan et al. 2009; Stewart et al. 2010). In our study we noted several areas of high density which may also represent specific Aβ-interacting cell-surface receptors. Furthermore, the lack of microglial binding by SEC-purified monomer in Figure 5 demonstrates the importance of aggregation structure for cellular interaction. It is possible that the Aβ(1–42) monomers bound to microglia but were rapidly internalized and degraded. This scenario is not likely since the binding studies were carried out on ice to slow down internalization pathways.

Acknowledgements

We would like to thank Dr. David C. Osborn in the Microscopy Image and Spectroscopy Technology Laboratory in the Center for Nanoscience at University of Missouri-St. Louis for TEM imaging.

† This work was supported by Award Number R15AG033913 from the National Institute on Aging (MRN).

Abbreviations used

AD

Alzheimer’s disease

Aβ

amyloid-β protein

aCSF

artificial cerebrospinal fluid

BCA

bicinchoninic acid

DAPI

4',6-diamidino-2-phenylindole dihydrochloride

HFIP

hexafluoroisopropanol

SEC

size exclusion chromatography

ThT

thioflavin T

TNFα

tumor necrosis factor α

Footnotes

Conflict of Interest: The authors declare that they have no conflict of interest.

References

  1. Bamberger ME, Harris ME, McDonald DR, Husemann J, Landreth GE. A cell surface receptor complex for fibrillar β-amyloid mediates microglial activation. J Neurosci. 2003;23:2665–2674. doi: 10.1523/JNEUROSCI.23-07-02665.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bitan G, Kirkitadze MD, Lomakin A, Vollers SS, Benedek GB, Teplow DB. Amyloid β-protein (Aβ) assembly: Aβ40 and Aβ42 oligomerize through distinct pathways. Proc. Natl. Acad. Sci. USA. 2003;100:330–335. doi: 10.1073/pnas.222681699. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem. 1976;72:248–254. doi: 10.1016/0003-2697(76)90527-3. [DOI] [PubMed] [Google Scholar]
  4. Cleary JP, Walsh DM, Hofmeister JJ, Shankar GM, Kuskowski MA, Selkoe DJ, Ashe KH. Natural oligomers of the amyloid-β protein specifically disrupt cognitive function. Nat Neurosci. 2005;8:79–84. doi: 10.1038/nn1372. [DOI] [PubMed] [Google Scholar]
  5. Combs CK, Karlo JC, Kao SC, Landreth GE. β-amyloid stimulation of microglia and monocytes results in TNFα-dependent expression of inducible nitric oxide synthase and neuronal apoptosis. J. Neurosci. 2001;21:1179–1188. doi: 10.1523/JNEUROSCI.21-04-01179.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Dahlgren KN, Manelli AM, Stine WB, Jr, Baker LK, Krafft GA, LaDu MJ. Oligomeric and fibrillar species of amyloid-β peptides differentially affect neuronal viability. J Biol Chem. 2002;277:32046–32053. doi: 10.1074/jbc.M201750200. [DOI] [PubMed] [Google Scholar]
  7. Dickson DW. Apoptotic mechanisms in Alzheimer neurofibrillary degeneration: cause or effect? J Clin Invest. 2004;114:23–27. doi: 10.1172/JCI22317. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Dickson DW, Lee SC, Mattiace LA, Yen SHC, Brosnan C. Microglia and cytokines in neurological disease, with special reference to AIDS and Alzheimer disease. Glia. 1993;7:75–83. doi: 10.1002/glia.440070113. [DOI] [PubMed] [Google Scholar]
  9. Fassbender K, Walter S, Kuhl S, Landmann R, Ishii K, Bertsch T, Stalder AK, Muehlhauser F, Liu Y, Ulmer AJ, Rivest S, Lentschat A, Gulbins E, Jucker M, Staufenbiel M, Brechtel K, Walter J, Multhaup G, Penke B, Adachi Y, Hartmann T, Beyreuther K. The LPS receptor (CD14) links innate immunity with Alzheimer's disease. Faseb J. 2004;18:203–205. doi: 10.1096/fj.03-0364fje. [DOI] [PubMed] [Google Scholar]
  10. Georganopoulou DG, Chang L, Nam JM, Thaxton CS, Mufson EJ, Klein WL, Mirkin CA. Nanoparticle-based detection in cerebral spinal fluid of a soluble pathogenic biomarker for Alzheimer's disease. Proc. Natl. Acad. Sci. USA. 2005;102:2273–2276. doi: 10.1073/pnas.0409336102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gravina SA, Ho L, Eckman CB, Long KE, Otvos L, Jr, Younkin LH, Suzuki N, Younkin SG. Amyloid β protein (Aβ) in Alzheimer's disease brain. Biochemical and immunocytochemical analysis with antibodies specific for forms ending at Aβ40 or Aβ42(43) J. Biol. Chem. 1995;270:7013–7016. doi: 10.1074/jbc.270.13.7013. [DOI] [PubMed] [Google Scholar]
  12. Haass C, Selkoe DJ. Soluble protein oligomers in neurodegeneration: lessons from the Alzheimer's amyloid β-peptide. Nat Rev Mol Cell Biol. 2007;8:101–112. doi: 10.1038/nrm2101. [DOI] [PubMed] [Google Scholar]
  13. Harper JD, Wong SS, Lieber CM, Lansbury PT., Jr Assembly of Aβ amyloid peptides: an in vitro model for a possible early event in Alzheimer's disease. Biochemistry. 1999;38:8972–8980. doi: 10.1021/bi9904149. [DOI] [PubMed] [Google Scholar]
  14. Harper JD, Wong SS, Lieber CM, Lansbury PT., Jr Observation of metastable Aβ amyloid protofibrils by atomic force microscopy. Chem. Biol. 1997;4:119–125. doi: 10.1016/s1074-5521(97)90255-6. [DOI] [PubMed] [Google Scholar]
  15. Hartley DM, Walsh DM, Ye CP, Diehl T, Vasquez S, Vassilev PM, Teplow DB, Selkoe DJ. Protofibrillar intermediates of amyloid β-protein induce acute electrophysiological changes and progressive neurotoxicity in cortical neurons. J Neurosci. 1999;19:8876–8884. doi: 10.1523/JNEUROSCI.19-20-08876.1999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Jan A, Hartley DM, Lashuel HA. Preparation and characterization of toxic Aβ aggregates for structural and functional studies in Alzheimer's disease research. Nat Protoc. 2010;5:1186–1209. doi: 10.1038/nprot.2010.72. [DOI] [PubMed] [Google Scholar]
  17. Jarrett JT, Berger EP, Lansbury PT., Jr The carboxy terminus of the β amyloid protein is critical for the seeding of amyloid formation: Implications for the pathogenesis of Alzheimer's disease. Biochemistry. 1993;32:4693–4697. doi: 10.1021/bi00069a001. [DOI] [PubMed] [Google Scholar]
  18. Jin M, Shepardson N, Yang T, Chen G, Walsh D, Selkoe DJ. Soluble amyloid β-protein dimers isolated from Alzheimer cortex directly induce Tau hyperphosphorylation and neuritic degeneration. Proc Natl Acad Sci USA. 2011;108:5819–5824. doi: 10.1073/pnas.1017033108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kayed R, Head E, Thompson JL, McIntire TM, Milton SC, Cotman CW, Glabe CG. Common structure of soluble amyloid oligomers implies common mechanism of pathogenesis. Science. 2003;300:486–489. doi: 10.1126/science.1079469. [DOI] [PubMed] [Google Scholar]
  20. Kheterpal I, Lashuel HA, Hartley DM, Walz T, Lansbury PT, Jr, Wetzel R. Aβ protofibrils possess a stable core structure resistant to hydrogen exchange. Biochemistry. 2003;42:14092–14098. doi: 10.1021/bi0357816. [DOI] [PubMed] [Google Scholar]
  21. Lee EB, Leng LZ, Zhang B, Kwong L, Trojanowski JQ, Abel T, Lee VM. Targeting amyloid-β peptide (Aβ) oligomers by passive immunization with a conformation-selective monoclonal antibody improves learning and memory in Aβ precursor protein (APP) transgenic mice. J Biol Chem. 2006;281:4292–4299. doi: 10.1074/jbc.M511018200. [DOI] [PubMed] [Google Scholar]
  22. Liu S, Liu Y, Hao W, Wolf L, Kiliaan AJ, Penke B, Rube CE, Walter J, Heneka MT, Hartmann T, Menger MD, Fassbender K. TLR2 Is a primary receptor for Alzheimer's amyloid β peptide to trigger neuroinflammatory activation. J Immunol. 2012;188:1098–1107. doi: 10.4049/jimmunol.1101121. [DOI] [PubMed] [Google Scholar]
  23. McGeer EG, McGeer PL. The importance of inflammatory mechanisms in Alzheimer disease. Exp. Gerontol. 1998;33:371–378. doi: 10.1016/s0531-5565(98)00013-8. [DOI] [PubMed] [Google Scholar]
  24. McGeer PL, Itagaki S, Tago H, McGeer EG. Reactive microglia in patients with senile dementia of the Alzheimer type are positive for the histocompatibility glycoprotein HLA-DR. Neurosci. Lett. 1987;79:195–200. doi: 10.1016/0304-3940(87)90696-3. [DOI] [PubMed] [Google Scholar]
  25. McNay EC, Gold PE. Extracellular glucose concentrations in the rat hippocampus measured by zero-net-flux: effects of microdialysis flow rate, strain, and age. J Neurochem. 1999;72:785–790. doi: 10.1046/j.1471-4159.1999.720785.x. [DOI] [PubMed] [Google Scholar]
  26. Meda L, Cassatella MA, Szendrei GI, Otvos L, Jr, Baron P, Villalba M, Ferrari D, Rossi F. Activation of microglial cells by β-amyloid protein and interferon-γ. Nature. 1995;374:647–650. doi: 10.1038/374647a0. [DOI] [PubMed] [Google Scholar]
  27. Nichols MR, Moss MA, Reed DK, Lin WL, Mukhopadhyay R, Hoh JH, Rosenberry TL. Growth of β-amyloid(1–40) protofibrils by monomer elongation and lateral association. Characterization of distinct products by light scattering and atomic force microscopy. Biochemistry. 2002;41:6115–6127. doi: 10.1021/bi015985r. [DOI] [PubMed] [Google Scholar]
  28. O'Nuallain B, Freir DB, Nicoll AJ, Risse E, Ferguson N, Herron CE, Collinge J, Walsh DM. Amyloid β-protein dimers rapidly form stable synaptotoxic protofibrils. J Neurosci. 2010;30:14411–14419. doi: 10.1523/JNEUROSCI.3537-10.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Paranjape GS, Gouwens LK, Osborn DC, Nichols MR. Isolated amyloid-β(1–42) protofibrils, but not isolated fibrils, are robust stimulators of microglia. ACS Chem Neurosci. 2012;3:302–311. doi: 10.1021/cn2001238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Reed-Geaghan EG, Savage JC, Hise AG, Landreth GE. CD14 and Toll-like receptors 2 and 4 are required for fibrillar Aβ-stimulated microglial activation. J Neurosci. 2009;29:11982–11992. doi: 10.1523/JNEUROSCI.3158-09.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Selkoe DJ. Cell biology of protein misfolding: The examples of Alzheimer's and Parkinson's diseases. Nat Cell Biol. 2004;6:1054–1061. doi: 10.1038/ncb1104-1054. [DOI] [PubMed] [Google Scholar]
  32. Stewart CR, Stuart LM, Wilkinson K, van Gils JM, Deng J, Halle A, Rayner KJ, Boyer L, Zhong R, Frazier WA, Lacy-Hulbert A, Khoury JE, Golenbock DT, Moore KJ. CD36 ligands promote sterile inflammation through assembly of a Toll-like receptor 4 and 6 heterodimer. Nat Immunol. 2010;11:155–161. doi: 10.1038/ni.1836. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Tzounopoulos T, Kim Y, Oertel D, Trussell LO. Cell-specific, spike timing-dependent plasticities in the dorsal cochlear nucleus. Nat Neurosci. 2004;7:719–725. doi: 10.1038/nn1272. [DOI] [PubMed] [Google Scholar]
  34. Udan ML, Ajit D, Crouse NR, Nichols MR. Toll-like receptors 2 and 4 mediate Aβ(1–42) activation of the innate immune response in a human monocytic cell line. J Neurochem. 2008;104:524–533. doi: 10.1111/j.1471-4159.2007.05001.x. [DOI] [PubMed] [Google Scholar]
  35. Walsh DM, Hartley DM, Kusumoto Y, Fezoui Y, Condron MM, Lomakin A, Benedek GB, Selkoe DJ, Teplow DB. Amyloid β-protein fibrillogenesis: Structure and biological activity of protofibrillar intermediates. J. Biol. Chem. 1999;274:25945–25952. doi: 10.1074/jbc.274.36.25945. [DOI] [PubMed] [Google Scholar]
  36. Walsh DM, Lomakin A, Benedek GB, Condron MM, Teplow DB. Amyloid β-protein fibrillogenesis: Detection of a protofibrillar intermediate. J. Biol. Chem. 1997;272:22364–22372. doi: 10.1074/jbc.272.35.22364. [DOI] [PubMed] [Google Scholar]
  37. Walsh DM, Selkoe DJ. Aβ oligomers - a decade of discovery. J Neurochem. 2007;101:1172–1184. doi: 10.1111/j.1471-4159.2006.04426.x. [DOI] [PubMed] [Google Scholar]
  38. Ye CP, Selkoe DJ, Hartley DM. Protofibrils of amyloid β-protein inhibit specific K+ currents in neocortical cultures. Neurobiol Dis. 2003;13:177–190. doi: 10.1016/s0969-9961(03)00068-8. [DOI] [PubMed] [Google Scholar]

RESOURCES