1 Introduction

In 1984–1985, Glenner, Masters and colleagues determined that neuritic plaques, one of the defining features of Alzheimer’s disease (AD), consist primarily of the small Aβ peptide (1, 2). This peptide, most commonly consisting of 40 or 42 amino acid residues, varies in length at its C-terminus and is derived from the amyloid precursor protein (APP) by specific proteolytic steps (3). The accumulation of Aβ and its deposition as insoluble β-amyloid plaque in the brain parenchyma is generally thought to be central to the pathogenesis of the disease. Much support for this amyloid-cascade model of AD pathobiology has come from the identification of mutations within the amyloid precursor protein (APP) (4–6) and the presenilin proteins (PS1 and PS2) (7–10) that cause early-onset familial AD (FAD) in humans. These mutations result in either more Aβ generation (11, 12) or increased levels of particularly pathogenic forms of Aβ (e.g., Aβ42 (13–15)). Expression in mice of FAD-mutant human APP transgenes either alone (16–18) or in combination with mutant presenilin transgenes (13, 19, 20) has resulted in animals that deposit substantial β-amyloid. These amyloid-depositing transgenic mice are now perhaps the most important experimental system in which to evaluate potential AD therapies, such as inhibitors of Aβ generation and Aβ immunotherapy (see, e.g., (21)).

Enzyme-linked immunosorbent assay (ELISA) is one of the most commonly employed biochemical techniques for the quantification of the Aβ peptide in the brain of humans and transgenic mice. Following the homogenization of the appropriate tissue (see Chapter 33) and a subsequent extraction to dissociate Aβ from the β-amyloid plaque, ELISA is used to precisely and rapidly measure levels of the peptide. Additionally, C-terminal epitope-specific antibodies are frequently used to differentiate Aβ40 from Aβ42, allowing the investigator to separately quantitate these two Aβ species. In this chapter, we describe the methodology used in our laboratory (21–37) for extracting Aβ from brain tissue and the subsequent quantitation of Aβ40 and Aβ42 by sandwich ELISA.

2 Materials

  1. 1.

    Ultracentrifuge and rotors (e.g., Beckman Coulter Optima TLX or MAX-E ultracentrifuge, TLA-100.1 and TLA 100.3 rotors).

  2. 2.

    Microplate spectrophotometer capable of reading the optical density (OD) at 450 nm of samples in a 96-well plate.

  3. 3.

    Platform rocker.

  4. 4.

    Sonic dismembrator with a probe capable of processing samples of approximately 600 μL.

  5. 5.

    7-mL dounce glass tissue grinder (Kontes brand; available from Fisher Scientific, Pittsburgh, PA, cat. no. K885300-0007).

  6. 6.

    8- or 12-channel pipette, with a minimum range of 50–200 μL.

  7. 7.

    0.5-mL thick-walled polycarbonate ultracentrifuge tubes (8  ×  34 mm, Beckman Instruments, Palo Alto, CA, cat. no. 343776).

  8. 8.

    3.5-mL thick-walled polycarbonate ultracentrifuge tubes (13  ×  51 mm, Beckman Instruments, Palo Alto, CA, cat. no. 349622).

  9. 9.

    96-well high-binding microtiter plates (Nunc-Immuno 96 MicroWell Maxisorp plates recommended, Nalge Nunc International, Rochester, NY, cat. no. 439454; available from Fisher Scientific, Pittsburgh, PA, cat. no. 12-565-135).

  10. 10.

    Adhesive sealing film for microplates (non-sterile, SealPlate brand recommended, cat. no. 100-SEAL-PLT, Excel Scientific, Wrightwood, CA; available from Sigma-Aldrich, St. Louis, MO, cat. no. Z36,965-9).

  11. 11.

    Multichannel pipetter basins (V-shaped bottom, non-sterile; available from Fisher Scientific, Pittsburgh, PA, cat. no. 13-681-100).

  12. 12.

    Centricon Centrifugal Filter, 10-kDa cutoff (Centricon YM-10, Millipore, Bedford, MA, cat. no. 4241).

  13. 13.

    Peroxidase Labeling Kit (Roche Diagnostics Corp., Indianapolis, IN, cat. no. 1 829 696).

  14. 14.

    0.5 mg human Aβ1–40 (lyophilized peptide, American Peptide Co., Sunnyvale, CA, cat. no. 62-0-78A).

  15. 15.

    0.5 mg human Aβ1−42 (lyophilized peptide, American Peptide Co., Sunnyvale, CA, cat. no. 62-0-80A).

  16. 16.

    0.5 mg mouse/rat Aβ1−40 (lyophilized peptide, American Peptide Co., Sunnyvale, CA, cat. no. 62-0-86A).

  17. 17.

    0.5 mg mouse/rat Aβ1−42 (lyophilized peptide, American Peptide Co., Sunnyvale, CA, cat. no. 62-0-84A).

  18. 18.

    Formic acid (minimum 95% purity).

  19. 19.

    FA neutralization solution (1 M Tris base, 0.5 M Na2HPO4, 0.05% NaN3). Store at room temperature; stable for months.

  20. 20.

    0.4% diethylamine, 100 mM NaCl. Store at 4°C; stable for months.

  21. 21.

    0.5 M Tris base, pH 6.8. Store at 4°C; stable for months.

  22. 22.

    10 mM EDTA in PBS.

  23. 23.

    Dimethyl sulfoxide (DMSO).

  24. 24.

    Purified Aβ40- and Aβ42-specific capture antibodies (see Note 1).

  25. 25.

    Purified anti-Aβ antibody with an epitope not overlapping with the epitopes of the capture antibodies (see Note 2).

  26. 26.

    10% NaN3. Store at room temperature. Very toxic, so be careful when preparing stock solution from powder.

  27. 27.

    Coating buffer: 30 mM NaHCO3, 70 mM Na2CO3, 0.05% NaN3, pH 9.6. (2.52 g NaHCO3, 7.42 g Na2CO3, 5 mL 10% NaN3, ddH2O to 1 L; adjust pH to 9.6). Store at 4°C; stable for months.

  28. 28.

    10× PBS: 1,369 mM NaCl, 27 mM KCl, 43 mM Na2HPO4, 15 mM KH2PO4. (80 g NaCl, 2 g KCl, 11.5 g Na2HPO4⋅7H2O, 2 g KH2PO4, ddH2O to 1 L; adjust pH to 7.4). Store at room temperature. Make fresh monthly in a sterile container.

  29. 29.

    Blocking buffer: 1% Block Ace, 0.05% NaN3, in PBS, pH 7.4. (4 g Block Ace powder, 2 mL 10% NaN3, 1× PBS to 400 mL; adjust pH to 7.4). Store at 4°C; stable for months (see Notes 3 and 4).

  30. 30.

    ELISA capture (EC) buffer: 5 mM NaH2PO4, 15 mM Na2HPO4, 2 mM EDTA, 400 mM NaCl, 0.2% bovine albumin, 0.05% CHAPS, 0.4% Block Ace, 0.05% NaN3, pH 7.0. (0.69 g NaH2PO4⋅H2O, 2.13 g Na2HPO4, 0.74 g EDTA disodium salt, 23.3 g NaCl, 2.0 g bovine albumin, 0.5 g CHAPS, 4 g Block Ace powder, 5 mL 10% NaN3, ddH2O to 1 L; adjust pH to 7.0). Store at 4°C; stable for months (see Note 4).

  31. 31.

    PBST: 0.05% Tween 20 in PBS. Store at room temperature. Make fresh monthly in a sterile container.

  32. 32.

    Detection antibody buffer: 3 mM NaH2PO4, 17 mM Na2HPO4, 2 mM EDTA, 400 mM NaCl, 1% BSA, pH 7.0. (0.41 g NaH2PO4⋅H2O, 2.41 g Na2HPO4, 0.74 g EDTA disodium salt, 23.3 g NaCl, 10.0 g bovine albumin, ddH2O to 1 L; adjust pH to 7.0). Sterile filter and make ∼11 mL aliquots under sterile conditions in a tissue culture hood. Store at 4°C; stable for months (see Notes 4 and 5).

  33. 33.

    TMB Microwell Peroxidase Substrate System (Kirkegaard & Perry Laboratories, Gaithersburg, MD, cat. no. 507600).

  34. 34.

    Stop solution: 5.7% o-phosphoric acid. Store at room temperature; stable for months (see Note 6).

3 Methods

3.1 Designing an ELISA and Preparing Solutions and Standards

3.1.1 Designing an ELISA

A sandwich ELISA consists of a “capture” antibody and a “detection” antibody, both of which bind to the peptide one wants to quantitate, but at distinct and nonoverlapping epitopes. Initially, the capture antibody is coated onto the plastic of the microtiter plate. When a complex mixture of proteins (such as those in a brain homogenate) is then incubated in this microtiter plate well, those peptides that are recognized by the capture antibody are bound and “captured” onto the ELISA plate. This reaction is driven by the high concentration and high affinity of the capture antibody, which allows for very small amounts of a peptide, such as Aβ, to be efficiently tethered to the plate via the antibody. Extraneous proteins are then washed away and a second antibody, conjugated to a reporter molecule or enzyme such as horseradish peroxidase, is used to detect the Aβ now bound to the initial capture antibody. Using such a “detection” antibody confers a number of advantages. First, this sandwich method greatly increases the sensitivity of the ELISA. Second, the ELISA gains specificity for a particular protein or peptide by combining the specificities of the two antibodies. For example, we commonly employ antibodies with specificity for the C-terminus of Aβ (e.g., an anti-Aβ40 antibody) to specifically capture Aβ40 from the complex mixture of proteins found in a brain extract (21–37). Importantly, this step eliminates Aβ42, APP, and other APP metabolites that may also be recognized by the detection antibody (see Note 7). The captured Aβ40 is then detected using an antibody that binds to the N-terminal region of Aβ (as well as other APP metabolites, which, if present, would potentially confound the specific quantitation of Aβ40). The following methodology will focus on the use of sandwich ELISAs to quantitate human Aβ40 and Aβ42 from β-amyloid-depositing human APP transgenic mouse brain (see Fig. 1). These protocols are also applicable to human brain from Alzheimer’s disease patients. In these cases, insoluble Aβ that has been deposited into β-amyloid plaques is first solubilized in formic acid (see Subheading 3.2). Additionally, sensitive sandwich ELISAs can also quantitate Aβ from non-β-amyloid-containing tissue, such as APP transgenic mouse brain prior to β-amyloid deposition and nontransgenic mouse brain (see Subheading 3.3.1) or from human or mouse plasma (see Subheading 3.3.2).

Fig. 1.
figure 1_34

Brain Aβ levels in transgenic mice determined by sandwich ELISA. Aβ was quantitated from formic acid-extracted 10% brain homogenates as described in this chapter. Data from three different β-amyloid-depositing transgenic mouse lines analyzed at the indicated ages (mo) are presented: TgCRND8 mice ((42); see also ref. 21) and Tg2576 mice (16) overexpress mutant forms of human APP while the Tg2576  ×  TgPS1 is a cross with a mutant presenilin 1 expressing line (19, 20). Open bars represent Aβ40, solid bars represent Aβ42. The increase in the amount of deposited Aβ40 and Aβ42 that occurs as the TgCRND8 mice age from 3 to 6 months is apparent. Additionally, both the TgCRND8 and Tg2576  ×  TgPS1 lines deposit more Aβ42 relative to Aβ40 than does the Tg2576 line, which is evident from this ELISA analysis. The sandwich ELISAs have high specificity for Aβ40 and Aβ42 due to the use of C-terminal specific anti-Aβ40 and Aβ42 antibodies (see Fig. 2).

Full size image

The specificity and sensitivity of an ELISA is dependent upon the purity, affinity, and epitope specificity of the antibodies used. In general, good monoclonal antibodies make for successful ELISAs; polyclonal antibodies tend to be more variable, and much effort needs to be put into the affinity-purification of the antiserum prior to its use in an ELISA. Thus, for success in any ELISA, the first step must be the identification of appropriate antibodies, preferably in quantities that will permit the investigator to expend the necessary amounts of antibody in the critical task of ELISA development and quality control. To date, we have primarily used monoclonal antibodies to detect Aβ by sandwich ELISA, antibodies which were the kind gift of Dr. Marc Mercken (Johnson and Johnson Pharmaceutical Research and Development/Janssen Pharmaceutica; see, e.g., refs. 21–39). These, and any particular set of antibodies used by others, however, may not be available to a given investigator and much preliminary effort may need to be made to obtain an antibody repertoire that will be suitable. In Notes 1 and 2, we describe commercially available antibodies that have been successfully used in our laboratory and would therefore allow most laboratories to develop a sandwich ELISA to detect human Aβ. The selection of antibodies is best guided by extensive conversations with laboratories and/or commercial suppliers that have had success detecting the protein of interest, prior to initiating any study. The user should be prepared to try a number of antibodies and antibody combinations during the development of an in-house ELISA. The less abundant the protein or peptide of interest, the more will rest on the pairing of the best antibodies in the sandwich ELISA. Commercial ELISA kits are also available from a number of sources for the quantitation of Aβ, and much of the preparation of samples and standards described in this chapter applies to the use of these kits. Covance (Princeton, NJ) currently markets ELISA kits, some of which use antibodies developed or characterized by our group as described in this chapter.

3.1.2 Preparing Solutions

In our laboratory, quantitating a peptide at low fmol/mL levels seems to be remarkably dependent upon the quality of the distilled, reverse osmosis-filtered H2O used to make solutions; contamination of solutions or the H2O source with fungus or other microorganisms can be problematic. We therefore recommend periodically disinfecting water storage containers by autoclaving or other means. For the same reason, PBS, including the 10× stock solution, and PBST should also frequently be made fresh, in sterile containers. Several of the solutions contain 0.05% NaN3 (sodium azide) as a preservative (1/200th volume of a 10% NaN3 stock solution).

3.1.3 Preparing Aβ Standards

Aβ standards for the ELISA are prepared from high quality synthetic peptides. The peptides are dissolved in dimethyl sulfoxide (DMSO), diluted in ELISA capture (EC) buffer, divided into aliquots, and stored at −80°C. To minimize the possibility of aggregation of the Aβ standard, carry out the following steps quickly and keep the EC buffer on ice. Allow vials of lyophilized peptide to equilibrate to room temperature before opening.

  1. 1.

    Preparation of 1 mg/mL master stock solutions: dissolve 0.5 mg of peptide in 500 μL of room temperature DMSO. Make 25 μL aliquots, freeze on dry ice, and store at −80°C. This stock solution is 231 nmol/mL for human Aβ1−40, 222 nmol/mL for human Aβ1−42, 236 nmol/mL for murine Aβ1–40, and 226 nmol/mL for murine Aβ1–42.

  2. 2.

    Preparation of 1 nmol/mL stock solutions: thaw an aliquot of the 1 mg/mL master stock solution from above and dilute 8 μL into the following volumes of EC buffer:

    • 1,840 μL for human Aβ1–40

    • 1,764 μL for human Aβ1–42

    • 1,888 μL for murine Aβ1–40

    • 1,808 μL for murine Aβ1–42

Divide each into 250-μL aliquots, freeze on dry ice, and store at −80°C.

  1. 3.

    Preparation of 5 pmol/mL working stock solutions: thaw an aliquot of 1 nmol/mL solution and dilute 200 μL into 40 mL of EC buffer. Divide into 100-μL aliquots, freeze on dry ice, and store at −80°C. You will need about 400 small tubes to make these aliquots, so label and arrange the tubes before thawing the 1 nmol/mL stock solution.

3.1.4 Preparation of an HRP-Coupled Detection Antibody

Our experience is that the linear range of the sandwich ELISA is best if the detection antibody is directly coupled to horseradish peroxidase (HRP), rather than when coupled to biotin and then detected using streptavidin–HRP. Although suitable biotinylated anti-Aβ detection antibodies can be purchased (see Note 2) and may suffice for some applications, a group that is investing in the development of highly sensitive ELISAs is better off using HRP-coupled antibodies, which they may need to produce themselves. This section describes the preparation of the antibody, the HRP coupling, quality control, and storage of an HRP-coupled detection antibody.

  1. 1.

    Start with the appropriate purified antibody (see Note 2). The antibody cannot contain azide in subsequent steps. Dialyze against PBS if an azide-free antibody source cannot be found.

  2. 2.

    The coupling reaction requires that the antibody concentration be ∼4 mg/mL in a volume of 300 μL. Most antibody solutions are more dilute than this, and if so the antibody first needs to be concentrated using a 10-kDa cutoff Centricon Centrifugal Filter according to the manufacturer’s protocol. Apply a sufficient amount of antibody to the column to obtain the needed amount of concentrated antibody. Ensure that the column never completely dries during centrifugation.

  3. 3.

    Follow the Roche Peroxidase Labeling Kit’s protocol for the coupling of HRP to an antibody. Per the protocol, remember that the PBS-antibody solution must be initially alkalinized before coupling.

  4. 4.

    The HRP-coupled antibody should not be frozen, but stored at 4°C as freezing can inactivate the peroxidase. Do not add azide as this also inactivates the peroxidase. For long-term storage, thimerosal can be added to 0.002% to the coupled antibody. A 2% (1,000×) thimerosal stock solution may be prepared, and should be stored at 4°C. We find that with proper storage HRP-coupled antibodies can remain active for more than 1 year at 4°C.

The efficiency of the coupling reaction must be determined empirically for each batch of HRP-coupled detection antibody. This is done by testing the ability of different dilutions of the newly coupled antibody to detect Aβ standards. When prepared according to the Roche kit protocol, which gives a final antibody volume of approximately 1 mL, an HRP-coupled anti-Aβ antibody can typically be diluted 1:1,000 or more for the ELISA (see Note 8).

3.2 Solubilization of Aβ from β-Amyloid Plaques Prior to ELISA

This procedure dissociates the aggregated Aβ in β-amyloid plaques so that it is possible to quantitate the peptide by ELISA. A small volume of homogenized β-amyloid-containing tissue (prepared as described in Chapter 33) is first sonicated in the presence of formic acid, which solubilizes Aβ from the densely aggregated β-amyloid plaque, and then subjected to high-speed centrifugation. Three layers result: a thin upper lipid layer that is not collected, the predominant intermediate phase that contains the Aβ peptides, and a barely discernable pellet (40). The intermediate phase is recovered and neutralized for the ELISA; the upper lipid layer and the pellet are discarded.

  1. 1.

    Mix 200 μL of a 10% (w/v) brain homogenate (see Chapter 33) into 440 μL of cold formic acid (minimum 95% purity) in a 1.7-mL microcentrifuge tube.

  2. 2.

    Sonicate each sample individually for 1 min on ice: immerse the tip of the probe in the sample and move the tube up and down over the probe while sonicating. Keep the tube on ice during this process by holding it in a small (e.g., 50 mL) beaker filled with ice. Rinse the probe and wipe it dry before processing the next sample.

  3. 3.

    Centrifuge 400 μL of the sonicated mixture at 100,000  ×  g for 1 h at 4°C (48,000 rpm using a Beckman Coulter TLA-100.1 rotor), in a 0.5-mL thick-walled polycarbonate ultracentrifuge tube.

  4. 4.

    Dilute 210 μL of the intermediate phase into 4 mL of room temperature FA neutralization solution. Vortex briefly.

  5. 5.

    Divide the neutralized solution into six 0.5-mL aliquots. Freeze the aliquots on dry ice. Each aliquot is more than sufficient for 4 ELISA well readings (typically Aβ40 and Aβ42 measurements, each in duplicate). These samples can be loaded onto the ELISA plate neat or, more typically, diluted in EC buffer as necessary (see Subheading 3.4).

3.3 Preparation of Non-β-Amyloid-Containing Tissue and Blood Plasma Prior to ELISA

3.3.1 Extraction of Aβ from Tissue Without Plaque Pathology

This section describes the extraction of Aβ from homogenates of tissue without β-amyloid plaque pathology. Although the Aβ present in a tissue that lacks β-amyloid is generally soluble, Aβ is a “sticky” peptide and the uniformity of its recovery and subsequent ELISA quantitation is improved if the Aβ is first extracted in diethylamine (23, 26, 41). Other agents, such as Triton X-100, can be used, but we find that the presence of nonionic detergents in the sample frequently increases the nonspecific background of the ELISA. Additionally, diethylamine extraction has the advantage of eliminating lipids (which are particularly abundant in brain and can nonspecifically increase background) and membrane-associated proteins such as APP, which remain in the pellet fraction following extraction (41). Diethylamine does, however, disrupt membrane vesicles so that intracellular as well as extracellular Aβ is recovered. In contrast to the formic acid extraction technique described in Subheading 3.2, the diethylamine-extracted sample is much less dilute prior to applying to the ELISA (see Subheading 3.5.4). This allows for the quantitation of the typically low levels of Aβ found in tissue without plaques [e.g., young APP transgenic mice prior to β-amyloid deposition and non-transgenic mice (see Note 9)] (22–24, 26–32, 35, 36). Tissue homogenates prepared as described in Chapter 33 are first mixed with diethylamine using a dounce. Following a high-speed centrifugation, Aβ is recovered in the supernatant, which is pH-neutralized and applied to the ELISA plate. Plaque-associated Aβ, however, is not completely extracted with this method and therefore formic acid extraction should be used in tissues containing β-amyloid.

  1. 1.

    Mix 1 mL of a 10% (w/v) brain homogenate (see Chapter 33) with 1 mL of cold 0.4% diethylamine (DEA), 100 mM NaCl with six up and down strokes of a glass pestle in the dounce glass tissue grinder. Keep the thawed homogenates, 0.4% DEA, 100 mM NaCl, and the dounce on ice. Transfer 1.9 mL of the homogenate–DEA mixture to a 3.5-mL thick-walled polycarbonate ultracentrifuge tube.

  2. 2.

    Between samples, rinse the pestle and dounce with H2O and dry.

  3. 3.

    Centrifuge the tube containing the homogenate–DEA mixture at 100,000  ×  g for 1 h at 4°C (43,000 rpm in a TLA 100.3 rotor).

  4. 4.

    Add 1.7 mL of the supernatant to a tube containing 170 μL of 0.5 M Tris base, pH 6.8 and vortex briefly. Divide into four 440-μL aliquots, freeze on dry ice, and store at −80°C. (440 μL is sufficient to run ELISAs for both Aβ40 and Aβ42 when loading 100 μL neat into duplicate wells for each assay.) These samples can be loaded onto the ELISA plate neat or diluted in EC buffer as necessary (see Subheading 3.4).

  5. 5.

    Discard the pellet.

3.3.2 Preparation of Blood Plasma Prior to ELISA

Aβ levels in blood plasma can also be determined by ELISA. The best method to prevent clotting is to mix the blood immediately with EDTA to chelate Ca++ as heparin may bind Aβ and interfere with the ELISA measurement. Plasma prepared this way can be directly applied to an ELISA plate following a freeze–thaw cycle to inactivate endogenous peroxidase. Without this freeze–thaw cycle, the endogenous peroxidase in blood can give a high background when the ELISA is developed using an HRP-coupled detection antibody.

  1. 1.

    Immediately mix the blood sample as it is being collected with an equal volume of 10 mM EDTA in PBS.

  2. 2.

    Gently mix the sample and centrifuge at 10,000  ×  g for 5 min at room temperature.

  3. 3.

    Collect the supernatant and divide the plasma into aliquots prior to freezing at −80°C. These samples can be loaded onto the ELISA plate neat or diluted in EC buffer as necessary (see Subheading 3.4).

3.4 Quantification of Aβ by Sandwich ELISA

Solutions should be kept on ice during the following steps with the exception of the PBS, PBST, and 5.7% o-phosphoric acid, which are stored and used at room temperature. Plates should be tightly sealed with sealing film during all incubations to prevent drying. The sandwich ELISA takes 3 days to complete.

Day 1

  1. 1.

    The capture antibody is coated onto the necessary number of wells of a 96-well high-binding microtiter plate by adding 100 μL/well of antibody diluted in coating buffer. Incubate overnight at 4°C with rocking. Typically, the capture antibodies are diluted to 2–10 μg/mL for coating. The antibody diluted in coating buffer must be free of any other proteins (see Note 1).

Day 2

  1. 2.

    Wash wells twice with PBS. This step can be done by various methods, including a wash bottle containing PBS which is used to spray PBS into the wells. Invert the plate quickly over a sink to discard the solution between washes. Residual wash solution can be removed by inverting the plate and patting on a paper towel.

  2. 3.

    Block nonspecific binding sites on the plastic by adding 200 μL/well of blocking buffer and incubating for 4 h at room temperature with gentle rocking. This blocking step can be extended for significantly longer than 4 h, and plates may even be left at this step for up to 1 week if stored at 4°C.

  3. 4.

    Prepare by thawing and diluting as necessary all samples and standards. Immediately before loading or diluting formic acid extracts, they should be incubated at 37°C for 5 min to solubilize any precipitate and not placed on ice before being diluted or a precipitate will quickly reform. Samples may need to be diluted with EC buffer to generate readings within the linear range of the standards. The approximate dilution needs to be determined empirically, and may require that multiple dilutions be tested in a trial ELISA and/or that assays are repeated with additional dilutions. Typically, formic acid-extracted samples from human tissue or transgenic mouse models with significant β-amyloid deposition will need to be diluted from 1:10 to 1:100 for an ELISA with a linear range of approximately 25–400 fmol/mL. DEA extracts prepared from non-β-amyloid-containing tissue and blood plasma is typically loaded onto the ELISA plate neat or up to a 1:10 dilution. Aβ standards are similarly diluted in EC buffer immediately prior to the ELISA. As previously noted, the concentration of the standards used will depend on the inherent sensitivity of the ELISA and must match the range of the Aβ concentrations found within the samples. As described in Subheading 3.5, care needs to be taken that the values for all samples fall between values obtained for the standards and within the linear range of the standard curve. As an example, standards at 400, 200, 100, 50, 25, 12.5, and 6.25 fmol/mL would be used for samples optimally diluted to give readings of ∼25–200 fmol/mL.

  4. 5.

    Immediately prior to proceeding with the ELISA and after samples are fully prepared for loading, dump the blocking buffer. Quickly add 50 μL/well of EC buffer to prevent the wells from drying while the individual samples are being added.

  5. 6.

    Add 100 μL of standards and samples to wells containing 50 μL of EC buffer (final volume in each well is now 150 μL; see Note 10). For a measurement of background signal, blank wells should contain 150 μL of EC buffer.

  6. 7.

    Incubate overnight at 4°C with rocking.

Day 3

  1. 8.

    Wash wells twice with PBST, then once with PBS.

  2. 9.

    Add 100 μL of HRP-conjugated detection antibody, diluted in detection antibody buffer (see Note 8), to each well. Incubate for 4 h at room temperature with rocking.

  3. 10.

    Wash wells twice with PBST, then once with PBS.

  4. 11.

    Add to the wells 100 μL of a 1:1 mixture of the two solutions (TMB peroxidase substrate and peroxidase substrate solution B) of the TMB microwell peroxidase substrate system (see Note 11). Allow plates to develop until the second to the least concentrated standard has a slight blue color change, and then stop the reaction by adding 100 μL of stop solution to each well (see Note 6). If the samples change color rapidly, the reaction may be stopped more quickly. If the samples have a low signal, however, a longer reaction time may produce more useful results. Regardless, samples and standards need to be stopped as simultaneously as possible. If the plates are allowed to overdevelop, the linear range of the ELISA will be compromised.

  5. 12.

    Read the OD450 with a microplate spectrophotometer. Aβ ­concentrations in the sample are interpolated from the OD450 using a standard curve generated from the known Aβ amounts in the standards (see Fig. 2).

    Fig. 2.
    figure 2_34

    Assessing antibody specificity. Affinity-purified polyclonal antibodies from a commercial source (Covance, Princeton, NJ; see Note 1) were coated on microtiter plates at 5 μg/mL and allowed to capture human Aβ standards. One of the antibodies, made against an Aβ40 epitope, showed sensitivity to sub-fmol/mL levels to Aβ40 (a, open boxes) while exhibiting no cross-reactivity to Aβ42 standards included on the same plate (a, closed circles). The second antibody, made against an Aβ42 epitope, showed similar specificity to Aβ42 but with a higher linear range and lower sensitivity (b). Both Aβ40 and Aβ42 were detected with the human Aβ-specific monoclonal antibody JRF/Aβtot/17, which binds to the N-terminus of Aβ, as previously described (43).

    Full size image

3.5 Assessing Antibody Specificity, Discarding Below Sensitivity and Saturated Points from Standard Curves, Accounting for Dilutions

3.5.1 Assessing Antibody Specificity

Once a combination of capture and detection antibodies has been obtained, pilot assays should be performed to determine the specificity of the antibody combinations for Aβ40 or Aβ42. This is accomplished by coating a microtiter plate with the putative Aβ40- or Aβ42-specific antibody. If an antibody has good specificity for Aβ40 vs. Aβ42, it should not recognize Aβ42 standards (Fig. 2a). Similarly, an Aβ42 antibody should not recognize Aβ40 standards (Fig. 2b). Such specificity assays are very useful in pointing out unexpected cross-reactivity in an ELISA.

3.5.2 Determining Sensitivity and Discarding Measurements Below a Meaningful Sensitivity Limit

The majority of the serially diluted standards in an Aβ ELISA should show a linear relationship, but at some point a “bottoming-out” effect will be observed in which there is no longer a linear change in OD with successively lower concentration standards (Fig. 3). The lowest concentration at which a point is still linear with the other points on the curve, and for which duplicate readings are in close agreement, thus becomes a given assay’s “sensitivity limit.”

Fig. 3.
figure 3_34

Determining sensitivity and discarding standards that are below sensitivity. The OD representing various concentrations of human Aβ40 standard was determined by sandwich ELISA. The bottom two points, representing the lowest concentrations of Aβ40, were below sensitivity and skew a linear curve fit away from the other points (dotted line, r  2  =  0.97). Discarding these two points from the analysis improves the curve fit (solid line, r  2  =  0.99) and would be more appropriate for calculating the amount of Aβ40 in samples containing ∼25–100 fmol/mL Aβ40.

Full size image

Several different concentrations of capture and detection antibodies should be tested in initial pilot assays to determine which antibody concentrations produce the optimal sensitivity for the concentration of Aβ expected in the samples. Since sensitivity will vary somewhat from assay to assay it should be determined for each ELISA, even after pilot assays have demonstrated a general sensitivity limit for a given combination and concentration of antibodies. OD values obtained for standards below this sensitivity limit are meaningless and may skew the standard curve, particularly in the region relevant to the OD measurements obtained from the samples (Fig. 3, dotted line). These “below sensitivity” values should be excluded in a consistent fashion before a curve fit is applied (Fig. 3, solid line).

3.5.3 Determining Linear Range and Discarding Saturated Standards

Frequently, the highest concentration standards develop rapidly, become saturated, and are nonlinear relative to lower concentration standards. Interpolating from a saturated region of the standard curve may cause significant differences between samples to be underestimated. Standard points that show signs of saturation should therefore be discarded and interpolation should only be from within the linear range of the standards (Fig. 4, solid line). This will often require additional dilution of the samples in a subsequent assay to ensure that all sample measurements are within the linear range of the ELISA.

Fig. 4.
figure 4_34

Discarding a saturated point to improve the linear curve fit. Human Aβ42 standards of various concentrations were analyzed by sandwich ELISA to produce a standard curve. The 100 fmol/mL Aβ42 standard was beginning to saturate when this ELISA was developed, and the linear relationship between peptide concentration and OD is not optimal. This produces a relatively poor linear curve fit (dotted line, r   2  =  0.957); removing the 100 fmol/mL Aβ42 standard data point generates a better linear curve (solid line, r  2  =  0.997), which would be more appropriate for calculating Aβ42 concentrations in samples from approximately 5 to 50 fmol/mL.

Full size image

Indeed, an ELISA with a broad linear range is desirable in that this will increase the likelihood that a given sample will be within range on a first ELISA run and that the assay will not have to be repeated. As with optimizing antibody concentrations to produce good sensitivities, pilot assays may be used to determine which antibody concentrations produce the broadest linear range. Interassay variability dictates that the linear range be determined for every assay, however, and that saturated standards be discarded wherever necessary.

3.5.4 Accounting for Dilutions Made During the Assay

Multiple steps throughout this protocol introduce dilution factors that must be taken into account. Aβ amounts are often reported as femto-, pico-, or nano-mole per gram of tissue. If the neutralized formic acid extract of a 10% tissue homogenate is loaded directly into the ELISA well as described in this chapter, the Aβ value determined in fmol/mL by the ELISA needs to be multiplied by 704 to convert to fmol/g of brain tissue (see Note 12). If the neutralized DEA extract of a 10% tissue homogenate is loaded directly onto the ELISA plate, the Aβ value determined in fmol/mL by the ELISA needs to be multiplied by 24.2 to convert to fmol/g of brain tissue (see Note 12).

4 Notes

  1. 1.

    Capture antibodies must be purified IgG and in PBS or a similar protein-free buffer. Any other proteins will compete with the antibody for binding to the microtiter plate. Thus, ascites fluid, antiserum, etc.—which contain many other proteins—are not suitable and only purified IgG or affinity-purified antibody from these sources can be used. Typically, capture antibodies are coated at 2–10 μg/mL. Increasing the coating antibody concentration more than this is often a waste of expensive antibody and will not improve the ELISA’s sensitivity. Using the minimal amount of antibody for coating is cost effective, but this amount must be determined empirically and a useful linear range of standards must be maintained. Covance (Princeton, NJ) sells high-affinity Aβ40 and Aβ42 affinity-purified polyclonal antibodies under their Signet product line; these antibodies were used for the assays shown in Fig. 2. Additional sources of Aβ antibodies are available and others will undoubtedly become available, so the investigator is advised to search for the most cost-effective source of high-affinity antibodies during the development phase of an ELISA.

  2. 2.

    As with the capture antibodies, the detection antibody must be purified IgG and in PBS or a similar protein-free buffer. A commonly used monoclonal antibody that binds to an epitope within the N-terminus of human Aβ is 6E10 (available from Covance, Princeton, NJ). A very serviceable sandwich ELISA for Aβ40 and Aβ42 can be constructed from the C-terminal-specific capture antibodies available from Covance and 6E10. Biotinylated 6E10 can also be purchased, and HRP–streptavidin used to detect the 6E10 bound during the ELISA, but this often results in a narrower linear range and therefore a less useful ELISA than one using directly HRP-coupled antibody. Regardless, if biotinylated antibody followed by HRP–streptavidin is used, much attention needs to be given to ensuring that the ELISA remains linear in the range of all samples (see Subheading 3.5).

An alternative detection strategy is to use a coating antibody of one species (e.g., mouse) and a detection antibody from another (e.g., rabbit). In this case, one could use an HRP-coupled anti-rabbit secondary antibody prior to developing the ELISA. However, we have found that this rarely works in practice, particularly if a high-sensitivity ELISA is required, as very minimal cross-reactivity of the anti-rabbit secondary antibody with murine IgG leads to overwhelming background.

  1. 3.

    Block Ace powder is a product of Dainippon Pharmaceutical Co., Osaka, Japan and is available from Serotec, Raleigh, NC, cat. no. BUF029. Other protein-blocking reagents are often used for ELISAs (BSA, casein). Our laboratory’s experience has been that while these reagents work, Block Ace gives the lowest nonspecific background and the most consistent assays over time, and we therefore recommend it. However, 3% BSA in PBS with azide can be substituted in many applications.

  2. 4.

    Preparing the solutions containing protein (e.g., blocking buffer, EC buffer, and detection antibody buffer) is easiest if all of the dry components are added to half of the final volume of the solution and stirred very vigorously—denaturation of these blocking proteins by foaming is not a problem and may even be advantageous. After the protein is dissolved, the solution is brought up to the appropriate final volume.

  3. 5.

    Do not add sodium azide to detection antibody buffer as the azide will inactivate the horseradish peroxidase conjugated to the detection antibodies in later steps.

  4. 6.

    In addition to 5.7% o-phosphoric acid, other acids such as H2SO4 and HCl are sometimes used to stop the TMB reaction. O-phosphoric acid, however, is more efficient in stopping color development and is recommended by the manufacturer.

  5. 7.

    Aβ sandwich ELISAs are typically done using C-terminal Aβ40- or Aβ42-specific antibodies as the capture antibody. This offers some advantages as N-terminal-directed anti-Aβ antibodies bind all Aβ species and usually recognize APP and other APP metabolites. If sufficiently abundant, these may compete with the Aβ of interest during the initial binding to the capture antibody. In practical application, however, this is not of significant concern when detecting the abundant Aβ extracted from a β-amyloid-depositing mouse, for example. Therefore, in some applications, the capture antibody can be N-terminal directed with other antibodies, recognizing other epitopes within either the Aβ sequence or at the C-terminus, used for detection.

  6. 8.

    The Roche Peroxidase Labeling Kit generally gives quite tractable HRP coupling reactions if the starting antibody concentration is carefully adjusted to 4 mg/mL. Nevertheless, the efficiency of the coupling must always be determined by trying each new batch of detection antibody in a trial ELISA. A straightforward way to do this is to test the newly coupled antibody at various dilutions in detection antibody buffer on a series of Aβ standards. A typical scheme would include the detection antibody diluted 1:1,000, 1:2,000, 1:5,000 and perhaps 1:10,000 against Aβ standards ranging from ∼3 to 200 fmol/mL. Frequently, using the detection antibody at higher concentrations will increase the sensitivity of the ELISA, although eventually higher concentrations will nonspecifically increase the background.

  7. 9.

    Murine and human Aβ differ in their N-terminal region at three residues. Thus, most antibodies that recognize the N-terminus of Aβ are species specific, with the vast majority of the available antibodies recognizing the human and not the murine peptide (e.g., 6E10). However, two well-characterized monoclonal antibodies have been used to quantitate endogenous murine Aβ by ELISA: JRF/rAβ1-15/2 and m3.2 (22–24, 26–32, 35, 36). The m3.2 antibody, unlike JRF/rAβ1-15/2, also recognizes other murine APP metabolites by Western blot (36). Antibody m3.2, which was generated by our group, is available from Covance (Princeton, NJ).

  8. 10.

    As both the samples and standards are diluted equally 2:3 in EC buffer at this point, no arithmetic adjustment for this dilution is necessary when calculating Aβ concentrations from the standard curve.

  9. 11.

    Warm the two bottles of TMB Microwell Peroxidase Substrate System to 37°C, then prepare the developing solution immediately before use by mixing an equal volume from each bottle. Change pipettes between bottles to prevent cross-contamination. For every set of two plates his mixture should be made fresh and used immediately.

  10. 12.

    This correction factor of 704 for the formic acid-extracted homogenate takes into account the 1:11 dilution when the tissue is initially homogenized (see Chapter 33), the further 1:3.2 dilution in formic acid, and the 1:20 dilution when the formic acid extract is neutralized. The correction factor of 24.2 for the DEA-extracted homogenate takes into account the 1:11 dilution when the tissue is initially homogenized (see Chapter 33), the further 1:2 dilution in DEA, and the 1:1.1 dilution when the DEA is neutralized.