Nanoscience enables ultrasensitive detection of Alzheimer's biomarker

Alzheimer's disease (AD), which afflicts an estimated 16 million people worldwide, is a neurodegenerative dementia characterized by memory loss and cognitive impairment (1). Symptoms begin with mild cognitive impairment that cannot be distinguished from other more benign forms of age-related dementia. Even as the disease progresses, clinical diagnosis can be made with only 65–90% accuracy. AD cannot be definitively diagnosed until after death, when brain tissue can be examined for the senile plaques and neurofibrillary tangles characteristic of the disease (2). The plaques result from aggregation of amyloid-β peptides and were long thought to be responsible for AD pathogenesis; however, their presence does not always correlate with neurological symptoms. Smaller, soluble oligomers of these peptides, referred to as amyloid-β-derived diffusible ligands (ADDLs), have recently been hypothesized as the causative agent in AD-related memory loss (3). Support for the role of ADDLs comes from their neurotoxicity (4), presence at elevated levels in the brains of AD patients as compared with age-matched controls (5), and mouse studies that indicate a reversal of memory loss upon injection of amyloid-β antibodies (6, 7). The work of Klein, Mirkin, and coworkers in this issue of PNAS (8) takes a significant step forward by demonstrating elevated ADDL concentrations in the cerebrospinal fluid (CSF) of patients who had been diagnosed (postmortem) with AD, as compared with healthy subjects. The correlation of CSF ADDL levels with disease state offers promise for improved AD diagnosis and early treatment. This finding was made possible by combining ADDL-specific monoclonal antibodies (9, 10) with an ultrasensitive, nanoparticle-based protein detection strategy termed bio-barcode amplification (BCA) (11).

Although ultrasensitive detection has become routine for nucleic acids, it remains challenging for proteins. A primary reason for this is the lack of a direct amplification method, analogous to the PCR for nucleic acids. Although proteins cannot be directly copied, they can be indirectly amplified by a process termed immuno-PCR (IPCR). In IPCR, a sandwich immunoassay is performed on a solid support, similar to traditional ELISA, except that the secondary antibody is covalently coupled to a DNA oligonucleotide rather than an enzyme (12). This DNA is then PCR-amplified and can be detected afterward by, e.g., gel electrophoresis, or in real time by specially designed fluorescent probes, e.g., TaqMan (13). IPCR provides substantial increases in sensitivity as compared with ELISA but requires conjugation of the secondary antibodies to DNA strands as well as the thermocycling and enzymatic amplification common to PCR. Variations have been reported in which improved reagents or other forms of enzymatic amplification provide improvements in sensitivity, quantification, or ease of use (e.g., by avoiding thermocycling) (14–21). For example, IPCR using oligovalent streptavidin-DNA assemblies has been shown to provide 1,000-fold increases in sensitivity as compared with traditional ELISA (18, 19).

The bio-barcode amplification strategy makes clever use of nanoparticles as DNA carriers to improve sensitivity.

The BCA strategy used by Klein, Mirkin, and coworkers (8, 11) makes clever use of nanoparticles as DNA carriers to enable millionfold improvements over ELISA sensitivity. Fig. 1 illustrates the approach. CSF is first exposed to monoclonal anti-ADDL antibodies bound to magnetic microparticles. After ADDL binding, the microparticles are separated with a magnetic field and washed before addition of secondary antibodies bound to DNA:Au nanoparticle conjugates. These conjugates contain covalently bound DNA as well as complementary “barcode” DNA that is attached via hybridization. Unreacted antibody:DNA:Au nanoparticle conjugates are removed during a second magnetic separation, after which elevated temperature and low-salt conditions release the barcode DNA for analysis. Importantly, each Au nanosphere carries hundreds of identical barcode DNA strands, providing substantial amplification. A second advantage is the ability to perform the BCA assay in homogeneous suspension, where faster kinetics are possible and where excess binding sites can be used to drive protein adsorption simply by addition of more particles. In addition, because the resulting DNA is separated from the particles and sample matrix before quantification, it can be detected by any method ranging from gel electrophoresis to electrochemistry (11, 22).

Fig. 1.

Bio-barcode amplification (BCA).

Klein, Mirkin, and coworkers (8, 23) took advantage of the “scanometric” DNA detection method to quantify the DNA produced by ADDL BCA. This approach, introduced by Mirkin and coworkers in 2000 (23), provides ultrasensitive DNA analysis in a surface-based sandwich hybridization assay for which the probe strands are arrayed on a solid support and the detection strand is bound to an Au nanosphere (Fig. 2). The selectively assembled Au nanospheres then act as nucleation sites for Ag deposition upon the chemical reduction of Ag⁺ from solution. The resulting Ag deposits, which can be quantified by a simple desktop scanner such as is used to scan documents for computer manipulation (hence scanometric), indicate the presence and amount of target DNA (23). BCA with scanometric detection has previously provided sufficient amplification to enable 30 attomolar (30 × 10^-18 M) sensitivity for prostate-specific antigen even without a PCR step and 3 attomolar sensitivity when PCR was used (11). This high sensitivity enabled measurement of ADDL levels of ≈200 aM in the CSF of normal patients as well as the higher levels in AD patients, where the mean concentration was 1.7 fM (8). Although a single protein target was detected in this study, the BCA approach is capable of simultaneously monitoring several targets. This is accomplished by conjugating antibodies against these proteins to Au nanospheres carrying different sequences of barcode DNA, which is then used to identify the proteins (24).

Fig. 2.

Scanometric detection.

The use of BCA to correlate ADDLs in CSF with Alzheimer's disease state marks one of the first real applications of nanotechnology. Although countless publications on nanoparticle-based diagnostics and other devices have appeared, the vast majority of these are in the proof-of-principle stage. Klein, Mirkin, and coworkers (8) have gone beyond this stage, applying the nanoparticle-based BCA and scanometric assays to current problems in AD research. Their findings show that a correlation exists between ADDL concentrations in CSF and AD in humans. Certainly much remains to be done; this study was restricted to 30 patients, and a larger sample size will provide greater confidence in the connection between ADDLs and AD. Importantly, the approach used here has the potential to test simultaneously for several proteins in CSF. It could provide a powerful tool for distinguishing among the several forms of dementia that begin with mild cognitive impairment (25). Because the pathology of AD is thought to begin decades before the first symptoms, it would be very interesting to learn at what stage of disease progression ADDL levels in the CSF rise above those in healthy individuals. The prospect of early diagnosis is particularly exciting in light of the success of antibodies against ADDLs in reversal of memory loss in mice (6, 7); presumably, any therapeutics will be most effective when used at the earliest possible stage. The work by Klein, Mirkin, and coworkers in this issue takes an important step toward answering these and other questions.