Automated Analysis of Rare Biomarkers Using a Magnetic Bead Surface Coverage Assay Coupled with Chaotic Mixing on a Microfluidic Chip

Extremely sensitive protein detection is crucial for early diagnosis and monitoring of diseases. Present day’s limit of detection for an antibody-antigen recognition-based immunoassay in serum without having to use special target antigen amplification techniques, like immuno-linked polymerase chain reaction (PCR), is 10 fg/mL, a sensitivity achieved thanks to the use of relatively large sample volumes (40-100 μL). In this thesis, a new method is developed to analyze rare protein biomarkers in a serum sample using a magnetic bead surface coverage assay. The assay protocols are realized on a microfluidic chip having channels, valves, an integrated mixer, and a protein micro-array. The microfluidic chip provides automated analysis, minute sample consumption (5 μL), resulting in an unprecedented 2 fg/mL detection performance for Tumor Necrosis Factor-α (TNF-α) in serum. A first part of the thesis consists in fabricating protein micro-arrays using the surface of immobilized ‘small’ (1.0 μm size) superparamagnetic beads that are placed in microfluidic flow channels. The small beads are immobilized on a glass chip with pre-patterned aminopropyl- trietoxysilane (APTES) structures using an electrostatic self-assembly method. Different target antigens are captured from a flow passing over the micro-array inside the microfluidic channel, and are specifically bound on the small bead surfaces. The presence of target antigens is detected as a fluorescent signal by realizing a sandwich immunocomplex on the capture antibody-functionalized surfaces of the small beads. However, the fluorescent detection method is abandoned due to high fluorescence background levels, resulting in only a 0.25 ng/mL detection limit for mouse immunoglobulin in phosphate-buffered saline. In a second part of the thesis, a polydimethylsiloxane (PDMS) chip having microfluidic channels, valves and an integrated mixer is designed. A new active mixing strategy is performed in the chip using multiple source-sink microfluidic flows. To do so, four different pressure- controlled actuation chambers are arranged on top of the 5 μL volume of the mixing chamber. After microfluidic valves, realized using ‘microplumbing technology’, seal the mixing volume, a virtual source-sink pair is created by pressurizing one of the membranes and, at the same time, releasing the pressure of a neighboring one. This creates microfluidic flows from the squeezed region (source) to the released region (sink) where the PDMS membrane is turned into the initial state. Perfect mixing is 7000-fold faster than achievable with pure diffusion-based mixing. This excellent performance is possible thanks to the implementation of chaotic advection by inducing vortices inside the mixing chamber. A DNA isolation and purification protocol using magnetic beads in the mixing chamber is also performed on-chip, showing the enhanced DNA binding performance due to active mixing. Finally, a new protein detection method is realized in a microfluidic chip. ‘Large’ (2.8 μm size) functionalized superparamagnetic beads specifically capture target antigens from a serum matrix under active microfluidic mixing. Subsequently, the large beads loaded with the antigens gently expose antibody-functionalized small bead patterns on the glass detection chip. During the exposure, attractive magnetic bead dipole-dipole interactions improve the contact between the two bead types and help the antigen-antibody immunocomplex formation, while non- specific large bead adsorption is limited by exploiting viscous drag forces in the microfluidic channel on the detection chip. This efficient transport mimics a biological process, where leukocytes roll and slow down on blood vessel walls by selectin-mediated adhesion for rapid recognition of tissue molecules. Afterwards, the antigen concentration is detected by simply counting the surface pattern-bound large magnetic beads. The new technique allows detection of only a few hundreds of TNF-α molecules (600 yoctomole) spiked in the 5 μL serum sample. This performance, in terms of detectable number of molecules, is at least 20 times better than today’s most-sensitive PCR-free antibody-based methods. Furthermore, the method has the supplementary advantage of being rapid, resulting from the automated assay protocols and the extremely fast (1 minute) antigen capture.

Microchannel-guided capture antibody patterning on beads for multiple protein detection arrays

Martinus Gijs, Venkataragavalu Sivagnanam, Hüseyin Cumhur Tekin, Sébastien Vaneberg

We present the rapid fabrication of custom protein micro-arrays, in less than 35 minutes, on the surface of immobilized beads using microchannel-guided flows driven by a single syringe pump. The commercially available beads provide robust micro-array surfaces to graft capture antibodies. The different target antigens are subsequently specifically bound and analysed in parallel in different microfluidic channels on a single micro-array chip. The presence of target molecules is detected as a fluorescent signal by realizing a sandwich immunocomplex on the specific capture antibody spot of the micro-array. With this technique, multiple proteins can be easily detected in a single channel.

2011

Microelectronics Meets Biology

Emile Dupont

Microfluidics plays a key role in the design of automated platforms for realizing biological assays in a miniaturized format. Several advantages are offered by a microfluidic system, namely, low sample and reagent consumption (typically a few microliters), the possibility to mass-produce and to integrate several analysis modules enabling portable automated applications, fast analysis and low cost. Furthermore, the combination of advanced Complementary Metal-Oxide-Semiconductor (CMOS) technology and microfluidics can result in flexible and easy-to-use hybrid systems that hold large potential for application in an analytical laboratory or at the point-of-care. In this Thesis, two different monolithic silicon CMOS chips were designed, fabricated and tested, demonstrating state-of-the-art performance utilising a novel fluorescence detection principle. They allowed the detection of monoclonal antibodies in a solution or the precise counting of specifically labelled fluorescent cells in a mixed-cell sample. The proposed detection method does not require the utilization of a fluorescence camera and filter set. The first part of this Thesis presents different methods, fabrication processes and materials for microfluidic cartridges that can be integrated in a hybrid way with silicon CMOS chips. Glass and SU-8 capillaries are tested, a novel material (NOA 63) is proposed for fabricating a natively hydrophilic microfluidic cartridge, and the fabrication of custom-designed polydimethyl(-siloxane) (PDMS) cartridges for easy integration on top of a CMOS chip is introduced. The second part of this Thesis presents the combination of microfluidics with a monolithic and fully-integrated CMOS chip for the manipulation and detection of fluorescent magnetic beads, inside a PDMS microchannel that is loosely positioned on top of the chip. Magnetic manipulation is done by current actuation of microcoils on the chip; detection is achieved using single photon avalanche diodes (SPADs), located in the centre of each microcoil, that count the fluorescent photons originating from a single labeled magnetic bead. This approach permits in principle microscope-less fluorescence detection with high sensitivity. Using sandwich immunoassays on the beads' surfaces, the selective detection of the cancer biomarker 5D10 monoclonal antibody (mAb) from a non-purified hybridoma cell-culture medium is demonstrated. In the third part, a CMOS chip comprising a matrix of SPADs is designed, fabricated and tested. The recognition of single fluorescently-labelled cells from the breast cancer cell line MCF-7 in a test solution of mixed MCF-7 and non-fluorescent Jurkat cells is demonstrated, without the need of a bulky and expensive fluorescent microscope setup. Moreover, due to the use of a SPAD matrix spanning the complete width of the microchannel, the system does not require any cell focusing mechanism, needed when using a single detector. This allowed the use of a very simple, easily replicable and disposable plastic microfluidic cartridge.

EPFL2010

Dose-response curve of a microfluidic magnetic bead-based surface coverage sandwich assay

Matteo Cornaglia, Martinus Gijs, Thomas Lehnert, Hüseyin Cumhur Tekin, Raphaël Etienne Jean Trouillon

Magnetic micro-and nanoparticles ('magnetic beads') have been used to advantage in many microfluidic devices for sensitive antigen (Ag) detection. Today, assays that use as read-out of the signal the number count of immobilized beads on a surface for quantification of a sample's analyte concentration have been among the most sensitive and have allowed protein detection lower than the fg mL(-1) concentration range. Recently, we have proposed in this category a magnetic bead surface coverage assay (Tekin et al., 2013 [1]), in which 'large' (2.8 mm) antibody (Ab)-functionalized magnetic beads captured their Ag from a serum and these Ag-carrying beads were subsequently exposed to a surface pattern of fixed 'small' (1.0 mm) Ab-coated magnetic beads. When the system was exposed to a magnetic induction field, the magnet dipole attractive interactions between the two bead types were used as a handle to approach both bead surfaces and assist with Ag-Ab immunocomplex formation, while unspecific binding (in absence of an Ag) of a large bead was reduced by exploiting viscous drag flow. The dose-response curve of this type of assay had two remarkable features: (i) its ability to detect an output signal (i.e. bead number count) for very low Ag concentrations, and (ii) an output signal of the assay that was non-linear with respect to Ag concentration. We explain here the observed dose-response curves and show that the type of interactions and the concept of our assay are in favour of detecting the lowest analyte concentrations (where typically either zero or one Ag is carried per large bead), while higher concentrations are less efficiently detected. We propose a random walk process for the Ag-carrying bead over the magnetic landscape of small beads and this model description explains the enhanced overall capture probability of this assay and its particular non-linear dose response curves. Research Paper