A highly sensitive novel immunoassay specifically detects low levels of soluble Aβ oligomers in human cerebrospinal fluid
© Yang et al.; licensee BioMed Central. 2015
Received: 3 October 2014
Accepted: 22 January 2015
Published: 22 March 2015
Amyloid β-protein oligomers play a key role in Alzheimer’s disease (AD), but well-validated assays that routinely detect them in cerebrospinal fluid (CSF) are just emerging. We sought to confirm and extend a recent study using the Singulex Erenna platform that reported increased mean CSF oligomer levels in AD.
We tested four antibody pairs and chose one pair that was particularly sensitive, using 1C22, our new oligomer-selective monoclonal antibody, for capture. We applied this new assay to extracts of human brain and CSF.
A combination of 1C22 for capture and 3D6 for detection yielded an Erenna immunoassay with a lower limit of quantification of approximately 0.15 pg/ml that was highly selective for oligomers over monomers and detected a wide size-range of oligomers. Most CSFs we tested had detectable oligomer levels but with a large overlap between AD and controls and a trend for higher mean levels in mild cognitive impairment (MCI) than controls.
Aβ oligomers are detectable in most human CSFs, but AD and controls overlap. MCI CSFs may have a modest elevation in mean value by this assay.
Progress in defining the molecular changes underlying Alzheimer’s disease (AD) has led to a modification of the original amyloid hypothesis to postulate that diffusible assemblies (oligomers) of amyloid-beta (Aβ) protein are the principal neurotoxic species and exist in a complex equilibrium with fibrillar amyloid plaques . A problem in advancing this Aβ oligomer hypothesis has been the difficulty in pinpointing the sizes and levels of the major bioactive oligomers. A bewildering array of oligomeric species has been described, almost entirely based on studies of synthetic Aβ peptides of defined lengths used at supraphysiological concentrations . Going forward, the field should focus on characterizing and quantifying natural Aβ oligomers (oAβ) isolated from human brain tissue and biological fluids.
While natural oAβ can be readily identified in aqueous extracts of AD brain tissue, attempts to detect soluble oAβ in cerebrospinal fluid (CSF) have led to mixed results, with a few laboratories reporting their ready detection but others concluding that their levels are very low or virtually undetectable (discussed in ). In the past 18 months, at least five studies reported measuring oAβ in CSF from AD and control patients [4-8]. All five found considerable overlap between the levels of oligomers detected in AD and control CSF, but four studies reported higher mean levels of soluble oligomers in AD than control subjects. One study examined only postmortem CSF . Two of the studies reported that measurement of soluble oligomers was less good at discriminating AD from controls than were Aβ42 or tau levels [5,6]. Interestingly, Holtta and colleagues found that AD patients with mild and moderate AD had significantly higher mean levels of oAβ than controls, whereas patients with severe AD did not differ significantly from the control group . Most of the groups describing positive detection of natural oligomers did not report the rigorous characterization of the assays; for example, by serial dilutions to confirm the linearity of the signal, by immunodepletion to confirm their Aβ specificity, by appropriately determining the lower limit of quantification, by excluding spurious signals arising from heterophilic antibodies, and by recovery of signal from samples spiked with an oligomer standard.
Two recent examples of detailed studies that were unable to establish the existence of oAβ in human CSF despite the use of well-characterized immunoassays were those by Esparza and colleagues  and by Yang and coworkers . In the latter report, our group postulated very low or virtually undetectable levels of oAβ in the CSF, and we then provided evidence that the relative hydrophobicity of oligomers could lead to very brief dwell times in aqueous compartments such as CSF or interstitial fluid .
To quantify the very low CSF levels of soluble oligomers and show diagnostically meaningful differences between AD and normal subjects will require highly sensitive assays that specifically detect oligomers and are blind to the abundant Aβ monomers present in all CSF samples. In this regard, we were intrigued by a very recent report of successful quantification of soluble oAβ by an immunoassay that utilized the high-sensitivity Singulex Erenna platform (Singulex, Alameda, CA, USA) . The authors even reported observing modestly but significantly higher levels of the oligomers in AD CSF compared with age-matched controls. Accordingly, we decided to try this platform first using a new oAβ-specific monoclonal antibody we generated and then employing the antibodies with the same complementary determining regions as those used by Savage and colleagues . Here, we report achieving a similar degree of sensitivity and specificity with our new sandwich immunoassay, but the signals in two cohorts of AD and age-matched control CSF samples overlapped substantially, while mild cognitive impairment (MCI) subjects showed a trend toward having slightly higher levels. The implications of this new work with regard to the usefulness of oligomers as an AD biomarker and the development of different and even more sensitive Aβ oligomer assays are discussed.
Monoclonal antibody (mAb) 3D6 is highly specific for the extreme N-terminus of Aβ  and was kindly provided by Dr Guriq Basi and Dr Dale Schenk (Elan, plc, South San Francisco, CA, USA). mAb 3B3 was generated by immunizing mice with an Aβ1–42 preparation known as amyloid-beta-derived diffusible ligand (ADDL) and was provided by Dr Bill Goure (Acumen Inc., Beverly Hills, CA, USA) . 3B3 is the murine precursor of the humanized 19.3 mAb used by Savage and colleagues and has the same complementary determining regions, and it demonstrates similar relative preferences to oAβ as assessed by surface plasmon resonance and KinExA (Sapidyne Instruments Inc., Boise, ID, USA). In solution, our novel mAb 1C22 only weakly recognizes Aβ monomer and binds strongly to Aβ aggregates, but in solid phase and tissue sections 1C22 can also bind to fibrils and amyloid plaques . A commercial N-terminal antibody similar to 3D6, 82E1  was obtained from Immuno-Biological Labs, Inc. (Minneapolis, MN, USA).
Preparation of synthetic Aβ standards: ADDLs, S26C dimers, and dityrosine dimers
Synthetic Aβ(1–42), Aβ(1–40) and Aβ(1–40)S26C were synthesized and purified using reverse-phase high performance liquid chromatography by Dr James Elliott at Yale University (New Haven, CT, USA), and the mass and purity of peptides were confirmed by liquid chromatography–mass spectrometry.
ADDLs were prepared essentially as we have described previously . Briefly, Aβ(1–42) peptide was dissolved in hexafluoroisopropanol to a final concentration of 1 mM and the solution was incubated at 37°C for 1 hour and mixed briefly by vortexing every 10 minutes. After 1 hour the hexafluoroisopropanol was evaporated using a speedvac. The dried peptide film was stored over desiccant at −20°C overnight. The film was then dissolved in anhydrous dimethylsulfoxide (Life technologies, Eugene, OR, USA) to produce a solution of 22.5 mg/ml and then diluted 1:50 with Hams F-12 media (Life technologies). The resulting solution was incubated at 4°C for ~14 hours and then centrifuged at 16,000 × g for 10 minutes. The supernatant was recovered and absorbance at 275 nm recorded using a spectrophotometer with a 1 cm path length microcuvette. The concentration of oAβ was determined using the extinction coefficient, ε275 = 1,361/M/cm , and by measuring the percent contribution of oAβ using analytical size exclusion chromatography (SEC) (Additional file 1). Relative oligomer content was estimated by dividing the peak area of the oligomer peak by the peak area of the oligomer peak plus the monomer peak. This value (based on three separate experiments) revealed that Aβ species larger than monomers accounted for 47 ± 6% of the total Aβ present in the ADDL preparations used. Following concentration determination, ADDLs were aliquoted (5 μl), immediately frozen on dry ice and then stored at −80°C. Once thawed, aliquots were used immediately either for characterization (using analytical SEC and electron microscopy ) or as standards in the MSD Multi-Array (Meso Scale Discovery, Gaithersburg, MD, USA) or Erenna immunoassay system (Singulex).
Monomeric Aβ was prepared as described previously . Briefly, Aβ(1–40) was dissolved at 2 mg/ml in 50 mM Tris–HCl, pH 8.5, containing 7 M guanidinium HCl and 5 mM ethylenediamine tetraacetic acid, and incubated at room temperature overnight. The sample was then centrifuged at 16,000 × g for 30 minutes and the upper 90% of supernatant applied to a Superdex 75 10/300 size exclusion column (GE Healthcare Biosciences, Pittsburgh, PA, USA), eluted at 0.5 ml/minute with 50 mM ammonium bicarbonate, pH 8.5, and absorbance monitored at 280 nm. Fractions of 0.5 ml were collected. The UV absorbance at 275 nm was determined for the peak fraction and the concentration of Aβ determined using ε275 = 1,361/M/cm . Dityrosine cross-linked Aβ dimer ((Aβ(1–40))DiY) was prepared as reported previously  and SEC-isolated as described above. The UV absorbance at 283 nm of the dimer peak fraction was measured and the concentration of DiY dimer determined using ε283 = 6,244/M/cm . (Aβ(1–40)S26C)2 was prepared as described previously  and the dimer was SEC-isolated as outlined above. The UV absorbance at 275 nm of the (Aβ(1–40)S26C)2 peak was measured and the concentration of this dimer determined using ε275 = 2,722/M/cm .
Following collection and concentration determination, monomer and dimer peak fractions of the respective oligomer preparations were diluted to 48.5 μM and aliquots (20 μl) were immediately frozen on dry ice and then stored at −80°C. Once thawed, aliquots were used immediately either for characterization (using analytical SEC and electron microscopy ) or as standards in the MSD or Erenna immunoassays.
MSD Aβ oligomer immunoassay
The lower limit of reliable quantification (LLoQ) is defined as the lowest back interpolated standard that provides a signal two-fold the background with a percentage coefficient of variance (CV) ≤20%.
Erenna Aβ oligomer immunoassay
The Erenna Immunoassay System (Singulex) is based on single molecule counting technology  and typically allows a 20-fold to 100-fold increase in sensitivity compared with traditional detection systems. Most experiments employed our mAb 1C22 for capture and 3D6 for detection, but for one set of experiments 3B3 was used for capture and 82E1 for detection. Biotinylated capture mAbs were bound to microparticles (MPs) at a ratio of 12.5 μg biotinylated antibody per milligram of MPs using a kit from Singulex. MPs with bound capture mAb were diluted to 50 μg/ml in assay buffer (Tris buffer; 50 mM Tris, 150 mM NaCl, pH 7.6), with 1% Triton X-100, 0.0005% (w/v) d-desthiobiotin and 0.1% bovine serum albumin, and 100 μl of this suspension was added to an equal volume of sample, standard or blank, and incubated at 25°C for 2 hours. MPs were isolated using a magnet, and unbound material was removed by washing with assay diluent using a HydroFlex plate washer (Tecan Group AG, Männedorf, Switzerland). Fluorescently-labeled detection antibody (20 μl, 100 pg/ml) was added to each well. MPs bearing the antibody–oligomer sandwich were then incubated with agitation using a Jitterbug shaker (Boekel, Feasterville, PA, USA) for 1 hour at 25°C. Unbound detection reagent was removed by washing (four times) with assay buffer, and MPs were transferred to a new plate. The assay buffer was removed by aspiration, and fluorescently-labeled detection antibody released by shaking in elution buffer (10 μl/well) for 5 minutes at 25°C. Eluates were then transferred to the wells of a 384-well plate containing neutralization buffer (10 μl/well). The sample was then drawn into an Erenna instrument and passed through a 100 μm diameter capillary flow cell through which a laser is directed. When fluorescently-labeled antibodies pass through the interrogation space, they emit light that is measured using a confocal microscope lens and a photon detector. The output from the detector is a train of pulses, with each pulse representing one photon that was detected. In this way, three signal outputs can be obtained: detected events (low end signal), event photons (low and medium end signal), and total photons (high end signal). For our experiments, we focused on detected events, generating standard curves from detected event values using a four-parameter curve fit. LOD and LLoQ were calculated as described for the MSD-based assay.
Measurements of tau, phosphorylated tau and Aβ1–42 in CSF
Samples in Cohort 1 were analyzed for Aβ42, t-tau and P-tau181 at the Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, Mölndal, Sweden, by board-certified laboratory technicians using commercially available INNOTEST ELISA kits (Fujirebio Europe, Ghent, Belgium). All analytical procedures were performed according to protocols accredited by the Swedish Board for Accreditation and Conformity Assessment.
CSF samples in Cohort 2 and Cohort 3 were analyzed for Aβ42, t-tau and P-tau181 with an Inno-Bia Alzbio3 kit (Fujirebio Europe) following the manufacturer’s protocol, and were read on a FlexMap3D platform (Luminex Corporation, Austin, TX, USA).
Human brain homogenate preparation
Homogenates of human brains were prepared as described elsewhere . Frozen cortices were provided by Dr M Frosch (MGH and MADRC Neuropathology Core, Harvard, MA, USA) under institutional review board-approved protocols. Frozen samples of temporal or frontal cortex (1 g) were allowed to thaw on ice, chopped into small pieces with a razor blade, and then homogenized with 25 strokes of a Dounce homogenizer (Fisher, Ottawa, ON, Canada) in 4 ml ice-cold 20 mM Tris–HCl, pH 7.4, containing 150 mM NaCl (Tris-buffered saline (TBS)) and protease inhibitors. Water-soluble Aβ was separated from membrane-bound and plaque Aβ by centrifugation at 175,000 × g and 4°C in a TLA 100 rotor (Beckman Coulter, Fullerton, CA, USA) for 30 minutes, and the supernatant (referred to as TBS extract) aliquoted and stored at −80°C. The ethical body approving this study was the Partners Institutional Review Board of the Partners Human Research Committee.
Cerebrospinal fluid sample collection and processing
Samples were collected from the L3/L4 interspace  and transferred into nonabsorbing (polypropylene) tubes. For Cohorts 1 and 2, CSF was mixed by gently inverting three or four times and then transferred to polypropylene storage tubes and stored at −80°C. For Cohort 3, immediately following collection, CSF samples were mixed by gently inverting three or four times and then centrifuged at 400 × g for 10 minutes. The crystal-clear supernatant was removed to a polypropylene tube and centrifuged at 2,000 × g at 4°C for 10 minutes, and aliquots of the supernatant were transferred to polypropylene storage tubes and stored at −80°C.
Developing an Aβ oligomer-selective immunoassay with improved sensitivity
Anti-human amyloid-beta IgG monoclonal antibodies
Aβ oligomer mix
Aβ1–42 ADDL preparation
Aβ oligomer mix
Aβ(1–5) synthetic peptide
Aβ(1–16) synthetic peptide
LLoQ, defined as the lowest back-interpolated concentration of standard that provided a signal two-fold greater than background and had a coefficient of variation <20%, was determined for each mAb pair with each synthetic Aβ preparation.
Comparison of four antibody pairs for sensitivity to detect three different synthetic amyloid-beta oligomer preparations on the MSD platform
Overall the 1C22/3D6B mAb pair most sensitively detected the three forms of synthetic oAβ tested and thus was selected for further study.
Comparison of sensitivities between the MSD and Erenna platforms for different synthetic Aβ oligomers
Comparison of sensitivities (LLoQ) of the 1C22/3D6B amyloid-beta oligomer assay using the MSD versus Erenna platforms
(S26C) 2 (pg/ml)
40.6 ± 9.3
375 ± 84.6
187 ± 54
601 ± 107
Based on these sensitivity results, we chose to employ ADDLs as our Aβ standard and use the Erenna platform in all subsequent experiments.
Selectivity of the oligomer-specific ELISAs versus Aβ monomer
We did not perform selectivity analyses of the 1C22/3D6 ELISA on Aβ42 monomer using the Erenna platform, but did perform such analyses using the MSD platform, and these results (data not shown) indicate much greater selectivity of the ELISA for oligomers over monomers, although as expected Aβ42 (but not Aβ40) can spontaneously form oligomers at very high Aβ42 concentrations, consistent with the fact that this longer peptide is well known to be much more prone to oligomerization. However, our CSF samples do not contain these very high concentrations of Aβ42, which exists at 1/10 the concentration of Aβ40 in normal CSF and is even lower than that in AD and MCI CSF.
Characterization of the oligomeric Aβ species detected by oligomer-specific ELISA
Application to natural human oligomers: dilution linearity of the Aβ oligomer signal in AD-TBS brain extracts
Raw data for Figure 4 : mean amyloid-beta oligomer values in serial dilutions of AD-TBS extracts from nine brains using the 1C22/3D6B oligomer-specific ELISA
35.0 ± 18.7
36.8 ± 6
15.5 ± 0.3
21.5 ± 0.8
8.7 ± 0.4
42.9 ± 5.0
4.7 ± 0.8
31.0 ± 4.8
17.7 ± 1.4
9.2 ± 0.4
7.1 ± 0.6
20.8 ± 0.05
3.2 ± 0.5
15.0 ± 0.8
2.9 ± 0.3
12.9 ± 1.6
8.0 ± 0.9
4.9 ± 0.2
4.9 ± 0.6
13.0 ± 1.16
2.2 ± 0.8
9.9 ± 0.3
1.3 ± 0.2
7.0 ± 0.5
5.0 ± 0.3
4.0 ± 0.5
2.2 ± 0.4
6.1 ± 0.31
0.9 ± 0.1
4.6 ± 0.1
0.8 ± 0.2
3.5 ± 0.3
2.9 ± 0.6
37.5 ± 2.2
2.5 ± 0.5
1.1 ± 0.1
3.6 ± 0.42
0.4 ± 0.1
2.5 ± 0.1
0.3 ± 0
2.0 ± 0.2
1.4 ± 0.1
17.1 ± 1.2
1.6 ± 0.1
0.6 ± 0.1
1.3 ± 0.1
1.4 ± 0.2
0.4 ± 0
1.2 ± 0.3
0.9 ± 0.2
10.0 ± 1.4
1.4 ± 0.1
0.6 ± 0.1
1.2 ± 0.4
0.9 ± 0.1
4.6 ± 0.3
1.0 ± 0.2
AD-TBS (1:2,000) after 1282 + AW7 immunodepletion
1.4 ± 1.5
0.6 ± 0.1
0.4 ± 0.1
Quantifiable levels of Aβ oligomers are present in most human CSF samples
Demographic data and biomarker concentrations for all cerebrospinal fluid samples
Total tau (pg/ml)
Phosphorylated tau (pg/ml)
442.6 ± 27.4
1101 ± 112.5
119.1 ± 8.2
0.9 ± 0.2
(54 to 84)
829.8 ± 57.3
239.5 ± 15.8
35.9 ± 2.1
0.7 ± 0.1
(62 to 81)
184.4 ± 11.2
154.9 ± 22.5
34.4 ± 3.2
0.2 ± 0.0
(50 to 86)
(11 to 20)
250.0 ± 25.6
142.3 ± 15.6
31.3 ± 4.3
0.3 ± 0.1
(76 to 89)
(23 to 28)
251.1 ± 34.1
58.8 ± 6.4
16.7 ± 1.8
0.2 ± 0.0
(75 to 82)
(27 to 29)
386.6 ± 30.0
152.4 ± 29.4
46.1 ± 5.1
0.9 ± 0.1
(60 to 85)
(25 to 30)
475.4 ± 23.1
95.5 ± 20.0
30.7 ± 5.3
0.6 ± 0.1
(55 to 86)
(27 to 30)
Amyloid-beta oligomer levels in cerebrospinal fluid samples before versus after immunodepletion with two amyloid-beta antisera
1.18 ± 0.21
CSF 1 after 1282
0.22 ± 0.25
CSF 1 after 1282 and AW7
3.16 ± 0.31
CSF 2 after AW7
CSF 2 after AW7 and 1282
0.05 ± 0.003
The oAβ values in the AD group of Cohort 1 completely overlapped with those in the control group (Figure 5A). Results from these clinically and biomarker well-characterized samples demonstrate that the current o-ELISA cannot discriminate between AD and controls.
Irrespective of whether or not Tween 20 was added to samples, the 3B3/82E1 assay in our hands yielded values approximately 2.67 times higher than those detected by our 1C22/3D6 assay (mean oAβ: 0.9 pg/ml in AD group and 0.7 pg/ml in controls). Importantly, direct comparison of the values obtained with the 3B3/82E1 assays versus those obtained with 1C22/3D6 revealed little correlation between the two assays, a finding that suggests the assays are not detecting precisely the same species. Nevertheless, none of the three assay formats reveal a significant difference between AD and control subjects.
Intra-assay and inter-assay precision
To examine intra-assay precision, ADDLs standards ranging from 10 to 0.0094 pg/ml were reversed and loaded in triplicate as samples and analyzed using the same ADDL standards. Above the 0.0781 pg/ml concentration of ADDLs, after calculating by averaging the results of the triplicates and calculating the CV percentage, the CVs averaged 11% and ranged between 3 and 18%. The recovery averaged 107% and ranged between 97 and 114%.
Inter-assay precision was controlled by running one CSF sample in triplicate on each of six different plates. Inter-assay precision calculated by averaging results of the six experiments, and the resultant CV percentage was 11% with a range between 3% and 28%.
Our results describe an Erenna-based immunoassay, using mAb 1C22 for capture and fluorescently-labeled mAb 3D6 for detection, which can selectively and sensitively quantify synthetic and natural soluble oAβ of a wide size range. Using this new o-ELISA, we were able to quantify signals from biological samples, including soluble extracts of AD brain and human CSF.
We chose this immunoassay design in an attempt to confirm and extend the recently described Erenna-based o-ELISAs developed by Esparza and colleagues  and Savage and colleagues . First, Esparza and colleagues bound their N-terminal capture antibody HJ3.4 to microtiter plates (rather than beads) and used this same antibody for detection, achieving a LOD of 1.56 pg/ml and a LLoQ of 6.25 pg/ml . With these sensitivities, they could detect oligomers in human brain but not in CSF. Next, Savage and colleagues immobilized their capture antibody (19.3, a humanized antibody with the same complementary determining regions as 3B3) on magnetic beads and detected with mAb 82E1 (an N-terminal specific mAb, like 3D6). The signal-to-noise ratio improved with this bead-based assay, achieving a LOD of 0.09 pg/ml and a LLOQ of 0.18 pg/ml, and oligomers were detected in most CSF samples analyzed .
Here, we compared our oAβ-preferring capture antibody, 1C22, with the murine version of 19.3 (that is, 3B3), combining each with 3D6 or 82E1, initially on the MSD platform. 1C22/3D6 allowed the best sensitivity, achieving: LOD of 4.0 pg/ml and LLoQ of 18.7 pg/ml when using ADDLs as the standard; LOD of 55.6 pg/ml and LLoQ of 187 pg/ml using (S26C)2 as the standard; and LOD of 154 pg/ml and LLoQ of 601 pg/ml using DiY as the standard. In these MSD platform assays, 1C22/3D6B was more sensitive than 3B3/3D6B, 1C22/82E1 and 3B3/82E1. We used the best pair, 1C22/3D6, to create a bead-based Erenna assay.
Based on the LLoQs achieved with each of the three synthetic Aβ standards, we chose ADDLs as the optimal standard for quantifying a broad range of oligomer sizes. We hypothesize that the greater assay sensitivity achieved using ADDLs as a standard occurs because the ADDL mixture contains more high-molecular-weight oligomers than the DiY or (S26C)2 preparations. This provides more potential binding sites and thus greater antibody avidity, resulting in tighter binding and greater sensitivity to detect small quantities of oligomers.
Having selected the best antibody pair and the optimal synthetic standard for a bead-based Erenna o-ELISA, we continued to validate its selectivity and specificity. First, this o-ELISA was shown to have 26,000-fold selectivity for synthetic oligomers over a synthetic Aβ40 monomer preparation that was shown to lack oligomers. Consistent with prior reports [3,22-25], the monomer concentrations in our CSF samples were always less than 10 ng/ml. Thus, even in the worst-case scenario, Aβ monomer could only produce a false-positive oligomer signal of ≤0.47 pg/ml and would by definition produce the highest interference in samples with the highest monomer levels. However, we observed no significant correlation between Aβ42 levels and Aβ oligomer levels in any of the three CSF cohorts studied (Additional file 3). These results indicate that the detected oAβ values are not substantially influenced by cross-reactivity with Aβ monomers. Second, ADDLs were fractionated by SEC using a Superdex 200 column and the fractions used for o-ELISA. The results indicate that the o-ELISA recognizes a broad range of oAβ, but is particularly suited to detect assemblies of masses between ~40 and 670 kDa (that is, within the included volume of a Superdex 200 column). Moreover, the lack of reaction with SEC-isolated Aβ42 monomer further confirms the selectivity of the o-ELISA for oAβ. Third, the o-ELISAs specifically detected natural oAβ in soluble human brain extracts and CSF samples. Quantitative immunodepletion of Aβ (that is, IPs with 1282 and AW7) from the AD-TBS brain extracts or from CSF (each confirmed by western blotting) reduced the o-ELISA signals by 98 to 100% of their original values. The 1C22/3D6 oligomer signal thus arises from oAβ in biological samples.
We then performed initial experiments to ask whether this o-ELISA could be used for quantifying oligomeric Aβ forms in the CSF as an AD biomarker by analyzing three different sets of patient samples obtained from different clinical centers. In two of these cohorts (Cohorts 1 and 2), CSF samples of clinically diagnosed AD patients showed no significant difference in oAβ levels compared with controls. However, there was a small but significant difference between AD and MCI patients in Cohort 2 from the Mayo Clinic CSF collection. In the third cohort, from the Harvard Biomarker Study collection, clinically diagnosed MCI patients also showed a small but significant increase in mean oligomer levels compared with age-matched controls. The three well-established CSF biomarkers for AD – levels of tau, phosphorylated tau and Aβ42 – distinguished control and AD patients much better than did soluble oAβ levels determined by the current o-ELISA.
AD patients might be expected to have increased concentrations of oAβ in CSF since the levels would reflect the high levels of aggregated Aβ accumulating in the brain parenchyma in the form of amyloid deposits. It is possible that patients in the early clinical stages of AD (for example, MCI) may have somewhat higher levels of soluble oAβ in their CSF compared with patients in later stages. We found in Cohorts 2 and 3 that patients with MCI had increased mean levels of CSF oligomers compared with control and AD subjects. Soluble oAβ may thus increase as these species begin to accumulate in brain parenchyma and populate what initially are mostly diffuse plaques. However, the substantial overlap of oligomer levels among control, MCI and AD groups in the three available cohorts we analyzed indicates that, despite similar sensitivities of our Erenna immunoassays, we did not observe the significant differences between AD and control subjects reported in the two cohorts studied by Savage and colleagues . We also did not find a significant correlation between MMSE and oAβ levels in this limited number of subjects. Direct comparison of the 3B3/82E1 antibody pair (Additional file 4), which is very closely similar to that used in Savage and colleagues’ o-ELISA, with our own antibody pair (1C22/3D6) on Cohort 1 CSF samples showed that 3B3/82E1 gave about 2.67 times higher Aβ oligomer values but with the same outcome – that is, no significant difference between AD and control subjects. One explanation for the higher apparent Aβ oligomer levels using 3B3/82E1 may be greater monomer cross-reactivity. The calculated cross-reactivity of 3B3/82E1 was reported by Savage and colleagues to be between 0.03 and 0.05%, so if we assume that the highest level of Aβ monomer in CSF is ~10 ng/ml, the 3B3/82E1 assay could have an oAβ signal emanating from monomers of about 4 pg/ml, compared with just 0.4 pg/ml calculated for our 1C22/3D6 assay.
Savage and colleagues studied CSF samples from two cohorts; that is, a total of 52 AD subjects and 43 controls. oAβ levels ranged from 0.75 to 3.3 pg/ml in AD subjects and from undetectable to 0.66 pg/ml in controls. On average, their Aβ40 values were 1,982 pg/ml for the AD group and 2,208 pg/ml for controls, and their mean Aβ42 levels for AD and control were 98 and 188 pg/ml, respectively. Across our three cohorts (from a total of 80 subjects), we found oAβ values ranged from undetectable to 2.42 pg/ml for AD, from undetectable to 2.74 pg/ml for MCI, and from undetectable to 1.46 pg/ml for controls, with the Aβ42 levels significantly lower in AD and MCI than controls.
Because of the important role that soluble oAβ is believed to have in AD pathogenesis, interest in the specific detection of oligomers in CSF has gained much attention. Five prior studies reported higher mean levels of oAβ in CSF from AD versus controls, but all found considerable overlap of the values. Fukumoto and colleagues  and Savage and colleagues  reported a significant difference of mean oligomer levels in AD versus control groups, and an inverse correlation with the MMSE scores of the patients. Holtta and colleagues  and Herkovits and colleagues  reported that oligomer levels differed between AD and controls but were less effective at discriminating AD from controls than Aβ42 or t-tau and P-tau181, and they reported no statistical correlation between oAβ levels and MMSE scores. Of note, Holtta and colleagues found that AD patients with mild to moderate AD had significantly higher levels of oAβ than controls, but patients with severe AD did not differ significantly from controls, potentially similar to our MCI findings here.
Taken together, current evidence suggests that soluble oligomers can be detected in human CSF, but that oAβ as currently measured are not yet a useful clinical biomarker. However, the study of oAβ is still at an early stage. Only a handful of o-ELISAs have been thoroughly validated and the conditions employed for CSF collection, processing and storage and how these may affect oAβ detection have yet to be investigated. Perhaps more importantly, little effort has been made to develop assays that detect specific populations of oAβ. It is therefore crucial to develop even more specific assays to measure different oligomer subgroups in a manner analogous to how we detect different monomer forms such as Aβ40 and Aβ42. This may require more specific capture and detection antibodies and perhaps use even more sensitive amplification platforms than employed up to now. Given the potential utility of oAβ as theragnostic markers for testing anti-amyloid therapeutics, a continued effort to better understand and accurately measure different oligomeric forms of Aβ is warranted.
oAβ are present in most samples of human CSF, but the levels are very low and their reliable detection requires the use of high-sensitivity platforms. Oligomer levels detected by the current assays tend to be higher in CSF from subjects with MCI than in AD or age-matched control subjects, but there was considerable overlap between the levels found in all three groups, whereas the accepted AD CSF biomarkers Aβ42, tau and p-tau clearly distinguished AD and MCI from controls. Further work, including the development of assays that detect specific populations of oAβ, is required to determine whether measurement of oAβ has diagnostic or theragnostic potential.
- (Aβ(1–40))DiY :
dityrosine cross-linked amyloid-beta dimer
- (Aβ(1–40)S26C)2 :
cystine cross-linked amyloid-beta dimer
amyloid-beta-derived diffusible ligand
coefficient of variance
lower limit of reliable quantification
limit of detection
mild cognitive impairment
Mini-Mental State Examination
oligomer-specific enzyme-linked immunosorbent assay
size exclusion chromatography
The authors thank Anitaben Tailor of Singulex, Inc. for assistance in Erenna assay development; Dr Alexandra Mably and Dr Brian O’Nuallain for preparing 1C22; Dr Ronald Peterson of Mayo Clinic (Rochester, MN, USA) for CSF samples; and Dr Matthew Frosch of the MGH Division of Neuropathology and the Massachusetts Alzheimer’s Disease Research Center for provision of frozen brain tissue.
This work was supported by grants from the National Institutes of Health: R01AG046275 to DMW and DJS, RO1 AG06173 to DJS, and BrightFocus grant A2013059S.
- Selkoe DJ. Alzheimer’s disease is a synaptic failure. Science. 2002;298:789–91.View ArticlePubMedGoogle Scholar
- Shankar GM, Walsh DM. Alzheimer’s disease: synaptic dysfunction and Abeta. Mol Neurodegener. 2009;4:48.View ArticlePubMed CentralPubMedGoogle Scholar
- Yang T, Hong S, O’Malley T, Sperling R, Walsh DM, Selkoe DJ. New ELISAs with high specificity for soluble oligomers of amyloid β-protein detect natural Aβ oligomers in human brain but not CSF. Alzheimers Dement. 2013;9:99–112.View ArticlePubMed CentralPubMedGoogle Scholar
- Bruggink KA, Jongbloed W, Biemans EA, Veerhuis R, Claassen JA, Kuiperij HB, et al. Amyloid-beta oligomer detection by ELISA in cerebrospinal fluid and brain tissue. Anal Biochem. 2013;433:112–20.View ArticlePubMedGoogle Scholar
- Holtta M, Hansson O, Andreasson U, Hertze J, Minthon L, Nagga K, et al. Evaluating amyloid-beta oligomers in cerebrospinal fluid as a biomarker for Alzheimer’s disease. PLoS One. 2013;8:e66381.View ArticlePubMed CentralPubMedGoogle Scholar
- Herskovits AZ, Locascio JJ, Peskind ER, Li G, Hyman BT. A Luminex assay detects amyloid beta oligomers in Alzheimer’s disease cerebrospinal fluid. PLoS One. 2013;8:e67898.View ArticlePubMed CentralPubMedGoogle Scholar
- Tai LM, Bilousova T, Jungbauer L, Roeske SK, Youmans KL, Yu C, et al. Levels of soluble apolipoprotein E/amyloid-beta (Abeta) complex are reduced and oligomeric Abeta increased with APOE4 and Alzheimer disease in a transgenic mouse model and human samples. J Biol Chem. 2013;288:5914–26.View ArticlePubMed CentralPubMedGoogle Scholar
- Savage MJ, Kalinina J, Wolfe A, Tugusheva K, Korn R, Cash-Mason T, et al. A sensitive abeta oligomer assay discriminates Alzheimer’s and aged control cerebrospinal fluid. J Neurosci. 2014;34:2884–97.View ArticlePubMedGoogle Scholar
- Esparza TJ, Zhao H, Cirrito JR, Cairns NJ, Bateman RJ, Holtzman DM, et al. Amyloid-beta oligomerization in Alzheimer dementia versus high-pathology controls. Ann Neurol. 2013;73:104–19.View ArticlePubMed CentralPubMedGoogle Scholar
- Hong S, Ostaszewski BL, Yang T, O’Malley TT, Jin M, Yanagisawa K, et al. Soluble Abeta oligomers are rapidly sequestered from brain ISF in vivo and bind GM1 ganglioside on cellular membranes. Neuron. 2014;82:308–19.View ArticlePubMedGoogle Scholar
- Feinberg H, Saldanha JW, Diep L, Goel A, Widom A, Veldman GM, et al. Crystal structure reveals conservation of amyloid-beta conformation recognized by 3D6 following humanization to bapineuzumab. Alzheimers Res Ther. 2014;6:31.View ArticlePubMed CentralPubMedGoogle Scholar
- O’Nuallain B, Blinder V, Kanmert D, O’Malley T, Mably A, Frost J, et al. Are anti-Abeta aggregate-preferring antibodies the future for AD immunotherapy? In Alzheimer’s Association International Conference (AAIC); 2014 July 12-17; Copenhagen, Denmark. Abstract P4-264.
- Freir DB, Nicoll AJ, Klyubin I, Panico S, Mc Donald JM, Risse E, et al. Interaction between prion protein and toxic amyloid beta assemblies can be therapeutically targeted at multiple sites. Nat Commun. 2011;2:336.View ArticlePubMed CentralPubMedGoogle Scholar
- O’Malley TT, Oktaviani NA, Zhang D, Lomakin A, O’Nuallain B, Linse S, et al. Abeta dimers differ from monomers in structural propensity, aggregation paths and population of synaptotoxic assemblies. Biochem J. 2014;461:413–26.View ArticlePubMedGoogle Scholar
- Walsh DM, Thulin E, Minogue AM, Gustavsson N, Pang E, Teplow DB, et al. A facile method for expression and purification of the Alzheimer’s disease-associated amyloid beta-peptide. FEBS J. 2009;276:1266–81.View ArticlePubMed CentralPubMedGoogle Scholar
- Todd J, Freese B, Lu A, Held D, Morey J, Livingston R, et al. Ultrasensitive flow-based immunoassays using single-molecule counting. Clin Chem. 2007;53:1990–5.View ArticlePubMedGoogle Scholar
- Shankar GM, Li S, Mehta TH, Garcia-Munoz A, Shepardson NE, Smith I, et al. Amyloid-beta protein dimers isolated directly from Alzheimer’s brains impair synaptic plasticity and memory. Nat Med. 2008;14:837–42.View ArticlePubMed CentralPubMedGoogle Scholar
- Teunissen CE, Petzold A, Bennett JL, Berven FS, Brundin L, Comabella M, et al. A consensus protocol for the standardization of cerebrospinal fluid collection and biobanking. Neurology. 2009;73:1914–22.View ArticlePubMed CentralPubMedGoogle Scholar
- Hepler RW, Grimm KM, Nahas DD, Breese R, Dodson EC, Acton P, et al. Solution state characterization of amyloid beta-derived diffusible ligands. Biochemistry. 2006;45:15157–67.View ArticlePubMedGoogle Scholar
- O’Nuallain B, Freir DB, Nicoll AJ, Risse E, Ferguson N, Herron CE, et al. Amyloid beta-protein dimers rapidly form stable synaptotoxic protofibrils. J Neurosci. 2010;30:14411–9.View ArticlePubMed CentralPubMedGoogle Scholar
- Hansson O, Zetterberg H, Buchhave P, Londos E, Blennow K, Minthon L. Association between CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: a follow up study. Lancet Neurol. 2006;5:228–34.View ArticlePubMedGoogle Scholar
- Oe T, Ackermann BL, Inoue K, Berna MJ, Garner CO, Gelfanova V, et al. Quantitative analysis of amyloid beta peptides in cerebrospinal fluid of Alzheimer’s disease patients by immunoaffinity purification and stable isotope dilution liquid chromatography/negative electrospray ionization tandem mass spectrometry. Rapid Commun Mass Spectrom. 2006;20:3723–35.View ArticlePubMedGoogle Scholar
- Bateman RJ, Wen G, Morris JC, Holtzman DM. Fluctuations of CSF amyloid-beta levels: implications for a diagnostic and therapeutic biomarker. Neurology. 2007;68:666–9.View ArticlePubMedGoogle Scholar
- Li J, Llano DA, Ellis T, LeBlond D, Bhathena A, Jhee SS, et al. Effect of human cerebrospinal fluid sampling frequency on amyloid-beta levels. Alzheimers Dement. 2012;8:295–303.View ArticlePubMedGoogle Scholar
- Slemmon JR, Meredith J, Guss V, Andreasson U, Andreasen N, Zetterberg H, et al. Measurement of Abeta1-42 in cerebrospinal fluid is influenced by matrix effects. J Neurochem. 2012;120:325–33.View ArticlePubMedGoogle Scholar
- Fukumoto H, Tokuda T, Kasai T, Ishigami N, Hidaka H, Kondo M, et al. High-molecular-weight beta-amyloid oligomers are elevated in cerebrospinal fluid of Alzheimer patients. FASEB J. 2010;24(8):2716-26.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.