The World Alzheimer Report 2016 [1
] estimates that dementia syndromes affect over 46 million people worldwide, and approximately 50% of them are diagnosed with Alzheimer’s disease (AD). AD-related pathophysiological changes can be detected as early as 10–20 years before the appearance of cognitive impairment. The detection of these changes by biochemical assays measuring levels of biomarkers in the cerebrospinal fluid (CSF), up to now the reference matrix for analysis of AD biomarkers, has led to a conceptual shift in the field of early Alzheimer’s disease diagnostics [2
]. Human tau protein is one of the most advanced and accepted biomarkers for AD and tauopathies diagnosis in general [4
], attracting increasing interest for developing anticipated and accurate molecular diagnostics [8
]. Bioanalytical approaches to tau protein detection have recently been reviewed, and we refer readers to these works for a detailed description of the currently available analytical platforms [4
]. Tau protein (50–65 kDa) belongs to the microtubule-associated protein family (MAP) and is expressed in human neurons and glia cells as oligodendrocytes and astrocytes. Tau protein interacts with tubulin, with a role in microtubules polymerization and stabilization. Its structure can be divided into four regions (Figure 1
), the N
-terminal region; the proline-rich domain; the microtubule-binding domain; and the C-terminal region, preferentially adopting a ‘paperclip’ conformation in solution [12
At present, the available detection methods for tau protein are based in large part on immuno-based recognition. Declared ranges for calibrators of enzyme-linked immunosorbent assay (ELISA)-based kits are 50–2000 ng L−1
for total tau, with a limit of detection (LOD) = 34 ng L−1
for all six protein isoforms. Very recently, the diagnostic role of non-phosphorylated tau in CSF has been addressed and an LOD of 25 ng L−1
has been reported [2
]. A nanoplatform used for Surface-Enhanced Raman Spectroscopy (SERS) fingerprint identification of β-amyloid and tau protein in blood with detection limit at 0.1 ng L−1
has also been described [13
]. Moreover, two ultrasensitive assays have been reported [10
]. The first is related to the detection of 3-repeat (3R) tau isoforms in Pick’s disease brain extracts and undiluted cerebrospinal fluid based on real-time fluorescence detection of thioflavin T (ThT) incorporated by protein seeds [10
]. The second allows the direct and rapid fluorescence detection of β-amyloid, tau, and phosphorylated tau in CSF, saliva, serum, and urine using magnetic nanoparticles [11
In the new diagnostic framework for preclinical Alzheimer’s disease, comprising biomarker-based research criteria developed by the workgroups of the National Institute on Aging–Alzheimer’s Association (NIA-AA) and those proposed by the International Working Group (IWG-2), the availability of label-free, real-time and low-cost methods based on biosensing tau biomarker in CSF represents an important possibility. Very recently, we reported Surface Plasmon Resonance (SPR)-based sensing of tau protein at the nanomolar level in artificial CSF (aCSF) by employing Multi-Walled Carbon Nanotubes (MWCNTs) in a sandwich-like detection strategy to enhance the analytical performances of the biosensor [14
]. Although affinity sensing based on QCM transduction has been successfully applied to aggregation kinetic studies [15
] and AD clinical diagnostics [16
] based on amyloid-β detection, to our best knowledge its application to tau protein detection has not been reported so far. Contrarily, some encouraging findings with electrochemical [18
] and conventional and/or localized SPR-based [14
] have been reported in the recent past. QCM relies on a quartz crystal alternating current-induced oscillations with an inverse relationship between the oscillation frequency and interface phenomena such as mass loading on the surface or changes of the viscosity or density of the media surrounding the sensor. For this reason, biochemical interactions can be directly displayed in real time and without the use of any label, allowing for simple detection of key biomarkers using consolidated chemistries, e.g., thiol chemistry coupled to amino coupling for receptor attachment, and several assay formats. In this context, we explored for the first time QCM capabilities in tau detection through the development of an immunosensor for tau protein with application in a simulated matrix (aCSF). Direct and sandwich-based detection strategies are compared here by using two monoclonal antibodies able to recognize different epitopes, as represented in Figure 1
. Furthermore, we exploited the well-known and specific recognition between tau and tubulin as an alternative to classical immuno-based sandwich assays. To the best of our knowledge, this is the first attempt at using tubulin as a recognition element by piezoelectric biosensing.
2. Materials and Methods
2.1. Reagents and Buffers
Human tau protein (isoform 2N4R, MW 46 kDa) was purchased from Enzo Life Sciences (Lyon, France). Primary antibody (mAb1, monoclonal, clone 39E10 produced in mouse) was from Biolegend (San Diego, CA, USA). This antibody recognizes the middle domain (amino acids 189–195) shared by all isoforms of tau protein. Secondary monoclonal antibody (mAb2, monoclonal, clone Tau12, produced in mouse) was purchased from Merck (Daarmstaad, Germany) and recognizes the N-terminus of tau. Human serum albumin (HSA) was from Sigma-Aldrich, Milan, Italy. Tubulin (bovine brain, >99% pure, lyophilized) was from Cytoskeleton Inc. and purchased from Società Italiana Chimici, Rome, Italy. Tubulin consists of a heterodimer of one alpha and one beta isotype, each tubulin isotype is 55 kDa in size. Typically, the molar equivalent of tubulin is defined as the heterodimer that has a molecular weight of 110 kDa. 11-mercapto-1-undecanoic acid (MUA), 1-ethyl-3-(3-dimethyl-aminopropyl) carbodiimide (EDAC) and ethanolamine HCl (EA) used for the modification of the gold surface of the sensor and for the immobilization of the antibodies were purchased from Sigma (Milan, Italy) as well as N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES), 1,4-Piperazinediethanesulfonic acid (PIPES), 2-(N-morpholino)ethanesulfonic acid (MES) hydrate, Ethylene glycol-bis(2-aminoethylether)-N,N,N′,N′-tetraacetic acid (EGTA), Guanosine 5′-triphosphate sodium salt hydrate (GTP) and glycerol. N-hydroxysuccinimide (NHS) was from Fluka (Milan, Italy). Ammonia (28%) and hydrogen peroxide (30%) were obtained from Merck (Milan, Italy). Ethanol, sodium acetate and all the reagents for the buffers were purchased from Merck (Italy). All solutions were prepared using double-distilled MilliQ water, unless otherwise stated. Antibody immobilization buffer was 10 mM CH3COONa (pH 4.5). Artificial cerebrospinal fluid (aCSF) consisted in 150 mM NaCl, 3 mM KCl, 1.4 mM CaCl2, 0.8 mM MgCl2, 0.8 mM Na2HPO4, 0.2 mM NaH2PO4 with 100 mg L−1 HSA, pH 7.3. The HBS-EP buffer (pH 7.4) contained 10 mM HEPES-Na, 150 mM NaCl, 3 mM EDTA and 0.005% Tween 20. Binding tests with tubulin were carried out in different buffer solutions: Phosphate buffer solution (PB, 20 mM Na2HPO4, 20 mM NaH2PO4, pH 7.2); MES solution (200 mM MES, pH 6.5); PEM buffer (80 mM PIPES, 2 mM MgCl2 0.5 mM EGTA, and 1 mM GTP, pH 6.9).
2.2. Piezoelectric Apparatus
The quartz crystal analyzer used for the measurements was the Model QCA922 (Seiko EG&G, Chiba, Japan). Sensorgrams were recorded in real time by a computer connected to the instrument interface using home-made software.
2.3. Sensor Chip Functionalization and Detection Format
Quartz crystals (9.5 MHz AT-Cut, 14 mm) with gold evaporated (42.6 mm2 area) onto both sides were purchased from Elbatech (Marciana (LI), Italy). The gold surface of the sensor chip was modified through thiol self-assembled monolayer (SAM) to which the primary monoclonal antibody (mAb1) was attached. Prior to use, the quartz crystal was cleaned for 10 min in a boiling solution of H2O2 (30%), NH3 (28%) and MilliQ water in a 1:1:5 ratio; then thoroughly washed with distilled water, dried, and used immediately afterward. The crystal was then immersed in 1 mM MUA ethanol solution for 48 h at room temperature to allow film formation via the strong Au–thiol bond with the tail carboxylic group exposed at the monolayer–liquid interface. Then the crystal was rinsed thoroughly with ethanol and sonicated in ethanol for 10 min. After rinsing with ethanol and drying with nitrogen, the MUA modified crystal was fixed into a methacrylate cell, where only one side of the crystal was in contact with the solution. The immobilization chemistry involved the covalent peptide bond forming process. The activation of the carboxylic groups was achieved by reacting the MUA-modified surface with an aqueous solution of 50 mM NHS and 200 mM EDAC for 15 min. An aliquot (100 µL) of mAb1 (100 mg L−1) was introduced to the activated MUA surface and lasted for 40 min. The surface was then rinsed with buffer to remove nonspecifically bound mAb1 before adding 1 M EA (pH 8.5) for 15 min to block any remaining active esters on the surface. After baseline equilibration, the crystal was ready for measurements.
2.4. Detection of Tau Protein
All measurements were carried out at room temperature (~20 °C) and in static conditions. Frequency shifts (in Hertz) due to the mass change at the sensing surface were measured by subtracting the baseline frequency recorded in the proper blank solution (HBS-EP buffer or simulated matrix, i.e., aCSF) from the frequency recorded after the binding of the relative ligands, i.e., tau protein, the secondary monoclonal Ab (mAb2), or tubulin. In the direct assay, tau protein (100 µL in HBS-EP buffer or aCSF) at different concentrations was added to the sensing surface and left in contact with the immobilized primary antibody for 15 min. The surface was washed to remove the excess of unbound protein and the change in frequency signals before and after protein incubation was recorded. For the sandwich-based assay, 100 µL of secondary antibody (mAb2, 10 mg L−1 in HBS-EP buffer) were added to the sensing surface after the binding of tau with the mAb1 and incubated for 15 min. Subsequently, the immunocomplex was washed to remove unbound mAb2. The frequency change was measured before and after each addition to evaluate the relative binding shifts. When the sandwich strategy was carried out, mAb2 was first used as a secondary binder. Alternatively, tubulin was explored as a secondary receptor (100 µL of 5 µM tubulin solution in PB, MES, or PEM buffers) and added to the sensing surface after the formation of the tau–mAb1 complex. After incubation (15 min), the surface was rinsed with buffer. In all the measurements, the surface of the biosensor was regenerated with 20 mM NaOH and 20 mM HCl in sequence for 30 s and equilibrated in buffer before starting a new measuring cycle.
QCM-based sensing for human tau protein in artificial cerebrospinal fluid (aCSF) has been achieved with a detection limit for the direct assay of 50 nM and 42 nM in HBES-EP and aCSF, respectively. The analytical performances of direct, classic sandwich, and tubulin-based secondary recognition strategies are tested and compared here. tau–tubulin binding has shown an extended working range coupled with signal improvement in comparison with the conventional secondary antibody-based approach. Tau–tubulin interaction conditions (PB, PIPES, MES, and PEM buffers; the presence of Mg2+ and GTP) were tested, with the best results in terms of non-specific adsorption and sensitivity recorded for the PEM buffer. Under the optimized conditions, tau within a 0–500 nM concentration range was assayed in aCSF, clearly showing a dose–response trend at a lower tau concentration than usually investigated (micromolar range) and closer to physiological levels in CSF, the reference matrix for protein tau biomarker, with important reflections on the real sensor applicability. In fact, the early identification of AD patients and the proper monitoring of the disease has a positive impact on defining effective therapeutic treatments and supplying patients with the most suitable treatment at the right time and dose. Our preliminary findings represent an interesting and innovative attempt to detecting Tau protein through a label-free, real-time, and low-cost biosensing strategy in a real matrix.