WO2011015359A1

WO2011015359A1 - Bioanalytical device

Info

Publication number: WO2011015359A1
Application number: PCT/EP2010/004807
Authority: WO
Inventors: Marta Bally; Janos Voeroes; Shoji Takeuchi; Andreas Binkert; Victoria Delange
Original assignee: Eth Zurich
Priority date: 2009-08-05
Filing date: 2010-08-05
Publication date: 2011-02-10

Abstract

The present invention relates to a bioanalytical device comprising a sensor and a thin slice obtainable from a structured three-dimensional construct containing ligands embedded in a permeable matrix in a repetitive manner.

Description

Bioanalytical device

The present invention relates to a bioanalytical device, to a planar microarray and to a process for the production of the planar microarray. Arrays of biomolecules (e.g. proteins or DNA) have emerged as a powerful high-throughput tool to study biomolecular interactions. Applications range from biological research to diagnostics, therapeutics, drug discovery, food technology and environmental monitoring. In particular, arrays of spatially distributed spots of biomolecules deposited onto a solid chip have become more and more popular tools for a variety of bioanalytical applications. Consisting of a multitude of parallelized, miniaturized and position encoded biological binding assays, these so-called microarrays allow for the high-throughput study of biomolecular interactions and can therefore provide systematically a high amount of biological information in short processing times. Thus, biological and biomedical communities seeking knowledge in proteomics and genomics have already, and will in the future, greatly benefit from this emerging technology.

A variety of protein and DNA arrays for the detection of e.g. protein-protein, protein-small molecule, protein- carbohydrate or oligonucleotide-oligonucleotide interactions have been described in a large number of publications, while other biomolecules including small molecules, carbohydrates, aptamers or peptides are now being arrayed as well. Technologies based on the use of microparticles as supports for biological recognition have been presented as an alternative to conventional planar microarrays. In the case of the suspension bead-based assay, multiplexed detection is achieved using populations of encoded microspheres (i.e. the analyte is identified by the bead type e.g. using a dye) and most commonly flow cytometry signal transduction. In this regard, several setups and assays are commercially available. Others have combined planar and suspension arrays for read-out systems requiring the surface immobilization of microparticles. Such approaches include bead immobilization on a planar substrate via a chemical or a biochemical linker and bead entrapment in a polydimethylsiloxane (PDMS) matrix, in microfluidic channels, in flow cells or in microwells.

Planar microarrays are usually manufactured using either photolithography, often combined with in situ synthesis, or robotic printing of the biomolecule of interest (spotting) .

Both processes for preparing planar microarrays commonly require relatively costly devices and appropriate microfabrication facilities (cleanroom facility, spotting robot, etc. ) . Considering the shortcomings of the planar microarrays of the state of the art and the process of their production, the problem to be solved by the present invention is thus to provide a planar microarray, which can be prepared in a large number of copies in a simple and cost-efficient manner without any specific instrumentation. According to another aspect, the problem to be solved by the present invention is to provide a bioanalytical device having the above mentioned advantages.

The problem of the present invention is solved by the subject-matter of the independent claims. Preferred embodiments of the present invention are defined in the dependent claims.

Thus, the present invention relates to a bioanalytical device comprising a sensor and a thin slice obtainable from a structured three-dimensional construct containing ligands embedded in a permeable matrix in a repetitive manner. The thin slice typically has a thickness of 10 nm to 1 mm.

The thin slice according to the present invention thus forms a planar microarray (i.e. a "biochip") of separate areas, each area comprising a ligand. Thereby, the microarray comprises a permeable matrix, in which the ligands are embedded.

Apart from the bioanalytical device, the present invention thus also relates to a planar microarray for bioanalytics .

The device and the microarray are generally used to investigate biomolecular interactions (protein-protein, protein-small molecule, protein-carbohydrate oligonucleotide-oligonucleotide, cell-drug, etc.) in parallel and with high-throughput. Specifically, the device and the microarray thus comprise biomolecules or cells for sensing applications.

According to the present invention, a permeable and generally soft matrix (e.g. a hydrogel) acts as support matrix for the biological ligands. Multiple microarray - A -

copies are obtainable by slicing the corresponding three- dimensional constructs comprising matrix planes or lines/channels containing the ligand. These matrix planes or lines/channels extend in an axial direction. In general, the three-dimensional constructs are sliced in a plane at least approximately perpendicular to the axial direction.

The present invention is thus based on the surprising finding that functional arrays of bioligands and/or bioligand carrying particles can be obtained by cutting a hydrogel construct with the ligands and/or particles embedded therein. Array functionality and compatibility with optical read-out was demonstrated.

Thus, the present invention offers the possibility to produce a large number of microarray copies at low cost and without any specialized instrumentation.

According to a first aspect of the present invention, the three-dimensional construct, from which the planar microarray (e.g. the "thin slice") is obtained, has a layered structure. This layered structure can be obtained by layer-by-layer assembly, the layers being more particularly formed by dipping or spin-coating, in microfluidic channels, or by stacking individual layers.

According to a further aspect, the three-dimensional construct can alternatively or additionally to the layered structure comprise channels filled with the permeable matrix, in which the ligand is embedded. These channels are preferably within a support structure, which can be permeable or not. The ligand is preferably selected from the group consisting of a biomolecule, more particularly a protein or an oligonucleotide, a small molecule, a cell, and a cell fragment. The term "small molecule", as used throughout this application, refers to molecules with a molecular weight below 10 kDa .

In general, the permeable matrix is formed by a compound, a first portion of which is in solid phase and a second portion of which is in liquid phase, the matrix being permeable to biological ligands and compatible with conventional biological assays. In addition, the matrix is preferably also non-toxic.

Preferably, the permeable matrix is a hydrogel, e.g. a sugar based gel, such as dextran, a poly (ethylene glycol) (PEG) based gel, an acrylamide gel, such as those used in protein separation, or a polyelectrolyte gel. Particularly well suited is an agarose gel.

The preferable properties of hydrogels, such as an agarose gel, become particularly apparent if the three-dimensional construct from which the microarrays are obtained comprises channels. Affordable and easy to work with, agarose is ideal for the support structure because it has low fluorescence background and is mechanically stable enough to support the (micro) channels . In addition, proteins can diffuse through an agarose matrix without being non-specifically captured.

The gel used to entrap particles in the channels is in general injectable at room temperature with a gelation procedure that does not denature proteins. To meet these requirements, the channels are typically filled with a low gelation temperature hydrogel. SeaPrep agarose, a hydroxyethylated version of agarose, forms a gel at 18 °C, instead of 37 °C, for 2% (w/v) . Preferably, SeaPrep agarose is only used in the channels because hydroxyethylation also reduces the gel strength, making it less suitable as a support structure.

According to another preferred embodiment, the permeable matrix is embedded in a supporting device. Thus, the preparation of the three-dimensional construct is simplified and a sufficient stability of the planar microarray can be achieved.

The ligands, in particular the bioligands, can be directly attached to the permeable matrix, in particular by covalent coupling. Alternatively, it is also possible that the ligands are "entrapped" in the matrix by non-covalent bonds, for instance by common non-covalent biochemical coupling chemistries, such as biotin-streptavidin or NTA- Ni-histidine tag, or by non-specific interactions, such as electrostatic interactions.

Alternatively or additionally, an additional supporting material, in particular a micro- or nanoparticle, can be used to embed the ligand in the matrix. In this context, polystyrene particles of different sizes can preferably be used as vehicles for biorecognition.

Consequently, the bioligand can be either attached directly to the hydrogel or bound to a support (e.g. a micro/nanoparticle) embedded in the gel matrix. Cells can be directly embedded in the hydrogel matrix.

Depending on the purpose of application, it can be preferred that the individual channels or layers contain different biological samples. Such microarrays are particularly well suited for multiplex binding assays and other microarray assays, and also for reverse microarray applications .

In a preferred embodiment, channels of the three- dimensional construct are part of a microfluidic device. It is thereby particularly preferred that a row of channels is prepared with suitable in- and outlets and stacked together to obtain an array.

As mentioned above, the present invention relates - apart from the bioanalytic device - also to a microarray comprising spatially separated areas, each area comprising a ligand, wherein the areas comprise a permeable matrix, in which the ligands are embedded, and are separated from each other by non-permeable regions. Thus, the microarray comprises several permeable, bioactive areas suitable for performing independent assays, as they are separated from the other permeable areas by non-permeable regions. These microarrays thus allow for the preparation of several completely independent assays in different areas of the microarray and come with a high multiplexing capability.

In an alternative embodiment, the microarray is essentially made of the permeable matrix with the ligands embedded therein. Thus, such a microarray essentially comprises only one bioactive area, which expands over the entire thin slice, but may comprise several different ligands. The preparation of these microarrays is particularly simple.

Depending on the purpose of application, it is further preferred that each area comprises a different ligand. As given above, the microarray is preferably obtainable from a structured three-dimensional construct, which comprises areas extending in an axial direction and containing in each case at least one of the ligands, by slicing the construct in a plane at least approximately perpendicular to the axial direction.

The present invention thus also relates to a process for obtaining a microarray comprising the steps of a) forming a structured three-dimensional construct comprising separate areas extending in an axial direction, said areas containing a ligand embedded in a permeable matrix; and b) slicing the three-dimensional construct in a plane at least approximately perpendicular to the axial direction.

As pointed out above, multiple microarray copies can thus be obtained in a very simple way from a structured three- dimensional construct, in particular from a hydrogel construct. The construct consists either of stacked (hydrogel) layers or of (hydrogel) channels embedded in a solid support.

The present invention thus provides a simple and novel approach for the production of a great number of "biochip" copies without any specialized instrumentation. Specifically, microarrays of functionalized microparticles embedded in a matrix, in particular in a hydrogel matrix, can be obtained from a three-dimensional particle/matrix stack, which is cut into thin slices. As mentioned above, the hydrogel construct can be obtained e.g. by layer-by-layer deposition, using microfluidic devices and laminar flow regimes or by filling a preformed stencil (made e.g. of PDMS). Such a stencil can be obtained, for instance, by molding or by stacking microstructured PDMS. The permeable hydrogel matrix contains the bioligand/cell of interest and is thus the vehicle for biorecognition.

Preferably, manufacturing approaches include layer-by- layer assembly, channel filling approaches or techniques based on laminar flow regimes in microfluidic channels.

According to a first concept, stacks of layers comprising particles embedded into a permeable matrix are obtained using layer-by-layer deposition from consecutive dipping and gelation steps. Agarose, a thermo reversible gel with a gelation temperature below 40 ⁰C, can be used as a support for the particles due to its appropriate gelation temperature, its low fluorescence background, non-fouling properties, and large pore size. The pore size of usually > 100 nm - depending on the gelation conditions, thermal history, and agarose type - permits the diffusion of proteins as large as immunoglobulins (IgGs) (radius: 5-7 nm) , which is highly important for affinity sensing applications. Using dipping as a layer deposition method, a multitude of array replicates can be obtained without any specialized instrumentation, simply using a beaker, a glass slide, a heating plate and a razor blade.

The success of this strategy is mainly based on the following hydrogel characteristics: mechanical stability, interlayer bonding, connectivity, permeability to assay reagents, and non-fouling properties. In order to comply with these characteristics, agarose has been used as a supporting gel matrix. Nevertheless, the approach is not limited to this specific type of hydrogel. Especially, photo- crosslinkable gels, such as high porosity polyacrylamide gels, might constitute a good alternative and could potentially further simplify the experimental procedure by eliminating the need for accurate temperature control as well as fastidious heating and cooling steps.

According to a further preferred embodiment, nano/microparticles are used as vehicles for the biorecognition reaction. In this regard, polystyrene particles of different sizes are preferably used as vehicles for biorecognition. It is also possible to use lipid or polymeric vesicles or large rafts or clay particles. Furthermore, the ligands may be arranged within the nano/microparticles or on the surface thereof.

The use of microparticles as vehicles for biorecognition is motivated by the fact that particles can be easily functionalized and give the system an enormous flexibility in terms of ligand choice. If physical adsorption can be used, particle functionalization can be achieved simply by mixing the particles with the reagents without the need for any coupling chemistry.

Alternatively, it is also possible to use specific coupling reactions, which are well known for such particles from (bio) chemical applications, in oder to couple only certain ligands from a sample to the particles.

The particle-based fabrication is highly flexible, as protein immobilization is not restricted to any particular gel chemistry. Several techniques have been developed for attaching proteins to microparticles, including physical adsorption, covalent coupling, and specific non-covalent attachment with affinity tags. Many varieties of protein- coated particles are also commercially available. In the system according to the present invention, capture probes are immobilized by physical adsorption, making functionalization especially simple. As a result of this flexibility, the technique according to the present invention is not only limited to antibody detection; the agarose channel support can be filled with a broad range of biologically relevant molecules according to the user' s need. With this simple fabrication technique, multiple microarray or biochip copies can be produced quickly and economically. The agarose channel support is ready in 30 min, and preparing a block of microarrays takes less than 8 hrs . Alternatively, the use of particles can be circumvented by coupling the proteins directly to the permeable matrix, and in particular the hydrogel. Besides providing a high surface area, hydrogels have the advantage of providing a quasi-bulk environment with high conformation freedom for the proteins facilitating the interaction between the binding partners.

Alternatively to obtaining the layered constructs by successive dipping and gelation, spin coating of a gel solution can represent an alternative to this procedure with the advantage of allowing a better control of the thickness over a larger thickness range and the preparation of thinner layers, especially when viscous pre-gel solutions are used.

According to a second aspect, the microarrays can be prepared by injecting a biofunctionalized hydrogel into microchannels, as mentioned above. The channels are generally formed using a mold to gel a hydrogel block, in particular an agarose block, around an array of pins.

Alternatively or additionally, it is also possible to obtain the channels by stacking microstructured layers of hydrogel.

In addition, by embedding the hydrogel channels into a stronger non-permeable support (e.g. polydimethylsiloxane (PDMS)), it is possible to integrate a flow-through system for rapid assay analysis using minute sample volumes. For hydrogel-based arrays, accelerating the reaction-diffusion kinetics significantly shortens the analysis times. Flow- through analysis also allows spots to be addressed individually, further increasing the flexibility. PDMS also facilitates denser array fabrication by stacking structures of periodic trenches to form the microchannels obtained with conventional photolithography methods.

Polystyrene microparticles can be functionalized with e.g. immunoglobulins (IgGs) and combined with e.g. low gelation temperature SeaPrep agarose to fill the channels. By using channels, custom two-dimensional arrays can be created; probe positioning is not limited to stripes, as it is in the layer-by-layer appro'ach. Furthermore, the spot material and the surrounding gel can be tuned individually for optimized performance. Agarose is a thermoreversible gel commonly used in electrophoresis, immunology, and as a culture medium for cells and other microorganisms. This hydrogel is well-suited for the system of the present invention because it is protein resistant, affordable, has low fluorescence background, and a large -pore size. The pore size of SeaPrep agarose allows for antibody diffusion, while keeping microparticles immobilized.

Also with regard to this second aspect, the microarray fabrication is very simple and does not require any specialized instrumentation. Also, the system is compatible with standard fluorescence read-out techniques. As a conclusion, a simple and inexpensive method for producing multiple copies of hydrogel microarrays can also be obtained by the channel-based approach. Thereby, arrays can be fabricated without any special instrumentation and limits of detection comparable to standard fluorescence- based immunoassays can be reached.

The channel-based system is more flexible than the layer- by-layer technique, requires less material, and maintains the advantages of rapid and inexpensive array fabrication. A model reverse phase assay shows the multiplexing ability of the microarrays, with low cross-reactivity and low unspecific binding. The limit of detection for a model sandwich assay has been shown to be consistent with standard fluorescent bioanalytical assays. As an additional feature, the microarrays can be dried to concentrate the fluorescent probes on a planar surface, which further increases array sensitivity.

By way of the channel-based system, comparable sensitivity as for the stacked layer system can be achieved using only 4 μl of hydrogel in each channel, which also requires fewer microparticles. By moving from functionalized stripes to spots, the advantages of the stacked layer system can be maintained while increasing the flexibility and degree of multiplexing. In addition, the entire sample is available for spotting and the technology is easily adaptable to the existing automated pipetting systems. It has been found that the microarray of the present invention is compatible with standard fluorescence based read-out techniques such as microscopes, evanescence field based readers or confocal scanners. To this end, model sandwich and reverse phase assays for the detection of various IgGs were performed as a proof of concept. Without any assay optimization, sensitivities in the range of conventional fluorescence based assays can be reached.

The bioassay can thus be performed according to standard procedures typically involving several incubation steps with sample and/or detector molecules. Read-out is performed with optical instrumentation common to microarray technology (e.g. confocal or flatbed scanners).

The present invention is further illustrated by way of the Examples together with the attached Figures, of which Fig. 1 (a) is a schematic representation of a portion of a microarray according to the present invention, in which the ligand is attached to a (micro) particle;

Fig. 1 (b) is a schematic representation of a process according to the present invention for producing a structured three-dimensional construct from which the microarray is obtained;

Fig. 1 (c) is a schematic representation of a structured three-dimensional construct comprising stacked layers; and Fig. 1 (d) is a schematic representation of the microarrays of the present invention obtainable from the structured three-dimensional construct shown in Fig. 1 (C) ; Fig. 2 (a) is a schematic representation of another structured three-dimensional construct comprising microchannels,

Fig. 2 (b) is an enlarged detail of the microchannel showing that the ligands are attached to a (micro) particle;

Fig. 2 (c) is a schematic representation of the microarrays obtainable from the structured three- dimensional construct shown in Fig. 2 (a) ; and

Fig. 3 is a schematic representation of a further three-dimensional construct comprising microchannels and a possible layout of a microfluidic device for the individual filling of the microchannels.

EXAMPLES 1. Preparation of microarrays obtained from a stack of layers comprising a bioanalytic ("stacked layers" approach)

A microarray as schematically shown in Fig. 1 (d) was obtained by preparing a three-dimensional construct according to a process as schematically shown in Fig. 1 (b) and by slicing the resulting structured three- dimensional construct comprising stacked layers as shown in Fig. 1 (c) . 1 . 1 . Experimental

Ma terials

NuSieve GTG low temperature melting agarose (melting temperature (4%): ≥ 65 ⁰C; gelling temperature (4%: ≤ 35 ⁰C) was purchased from Lonza (Japan). Polystyrene particles with diameters of 100 μm, 15 μm and 1 μm were purchased from Microparticles GmBH (Germany) . FluoSpheres 450/480 and 580/605 with a diameter of 15 mm as well as FluoSpheres 350/440 with a diameter of 1 μm were purchased from Invitrogen (Japan) . Bovine serum albumin (BSA), Mouse IgG, Rabbit IgG, anti-mouse IgG (Fc specific, produced in goat) anti-mouse IgG-FITC (fluorescein isothiocyanate) (Fab specific, produced in goat) were purchased from Sigma-Aldrich (Switzerland or Japan) . Anti-rabbit IgG AlexaFluor555 and anti-mouse IgG AlexaFluor488 (produced in goat) were purchased from Invitrogen (Japan) . Borate buffer solution was obtained from 0.1 M Boric Acid solution (Sigma-Aldrich, Switzerland) with a pH adjusted to 8.5. HEPES buffer solution consisted of 10 mM 4- (2-hydroxyethyl ) - piperazine-1-ethane sulfonic acid (MicroSelect , Fluka Chemie GmbH, Switzerland) and 150 mM NaCl, with a pH adjusted to 7.4.

Particle Functionalization

The functionalization of 100 μm particles was carried out in borate buffer. 200 μl beads (10% w/v) were washed in 1.5 ml buffer by centrifugation ( 751 x g, 3 min) followed by supernatant removal. After two washing steps, the appropriate protein solution was added to the beads resuspended in 1 ml buffer. Beads coated with anti-mouse IgG (Fc specific) were obtained by addition of 110 mg of antibody (50 μl) . Beads coated with rabbit or mouse IgG were obtained by addition of a 1:1 (w/w) mixture of BSA and the corresponding antibody (240 μg each, 96 μl) . As a control, beads coated with BSA were obtained by addition of 480 μg of BSA (96 μl) . The suspension was incubated overnight with gentle end-to- end mixing. After washing twice, the beads were blocked by incubation in a 10 mg/ml BSA solution (twice, 30 min incubation) followed by another two washing steps. The beads were stored in 200 μl HEPES buffer until further use. Functionalization of the 1 μm particles was carried out with a similar protocol in HEPES buffer. 20 μl of protein solution containing the appropriate amount of IgG in the presence of 5 mg/ml BSA was added to 20 μl of particles (10% w/v) in 500 μl buffer. After washing and blocking, the beads were stored in 20 μl HEPES buffer. Centrifugation was always performed at 5344 x g (3 min) .

Array preparation

A 10% (w/w) resp 5% (w/w) agarose gel was formed in ultrapure water (Direct Q, Millipore Corporation, Japan) and melted at T > 65 ⁰C. After cooling down to 38 ± 2 ⁰C, the microparticles were added to an equal volume of 10% agarose, except for experiments performed with

1 μm particles where the beads were diluted directly in

5% agarose (dilution factor: 1:30) . If required, the composite was further diluted with 5% (w/w) agarose. Following final particle concentrations were used:

100 μm beads: 4% (w/v) (reverse array), 3.3% (w/v) (sandwich array) ; 15 μm and 1 μm beads: 0.35% (w/v) . As schematically shown in Fig. 1 (b) , hydrogel multilayers consisting of alternating layers of particle containing agarose and plain hydrogel were obtained by dipping successively a glass slide into the appropriate hydrogel solution and letting cool down at room temperature for approximately 45 s. To ensure proper bonding, the array was briefly dipped in a gel at T > 65 ⁰C before being transferred into the next pre-gel solution. After completion of the multilayer build-up, the gels were dipped in HEPES buffer and stored at 4 ⁰C for at least 5 min before being sliced with a razor blade. Adhesion between the gel and the support glass slide was weak: The gel was either released spontaneously during the storage in liquid or by application of a gentle lateral force (with help of tweezers or a razor blade) .

The resulting three-dimensional construct in the form of a stack of layers was then cut manually into thin slices with a cutting tool, in particular a razor blade. Thus, array preparation was fast and simple: The total layer deposition time was roughly one minute per layer so that multiple copies of arrays consisting e.g. of five different bead types were produced in less than 15 min (including the time for manual slicing) . Microarray assay

All assays were performed in HEPES buffer. Arrays prepared from 100 μm beads carrying rabbit resp. mouse IgG were incubated for 3 hours in 5 μg/ml anti-rabbit IgG AlexaFluor 555 or anti-mouse IgG AlexaFluor 488. Arrays prepared from 1 μm beads functionalized with rabbit IgG were incubated overnight in 20 μg/ml anti- rabbit IgG AlexaFluor 555. Sandwich assays were performed by incubating the arrays with mouse IgG in the presence of 10 mg/ml BSA (incubation time: 3 hours), followed by rinsing in buffer (45 min) and incubation (3 hours, 38 μg/ml) with anti-rabbit IgG FITC (Fab specific) . Before imaging all the arrays were rinsed for several hours in HEPES buffer.

Microarray imaging and evaluation

Microarray imaging was performed with a fluorescence microscope Olympus 1X71 (Japan) equipped with a camera QICAM Fast 1394 (Q-Imaging Ltd, United Kingdom) and with the following objectives (Olympus, Japan) : Plan APO 2X N. A. 0.08; U Plan NFL 4X N. A. 0.13; U Plan APO 1OX N. A. 0.4 phi. The following filters from Olympus (Japan) were used: U-MWIG3 (AlexaFluor 555) and U-MWIB2 (FITC and AlexaFluor 488) . Data evaluation and image processing was performed with the software ImageJ (Image processing and analysis in Java, National Institute of Health) . Quantitative data was obtained from images taken with 1OX magnification. Dose-response curves for assays using 100 μm beads were obtained by measuring the average intensity of 6 beads normalized with the average background intensity around the bead (s/n). Dose response curves for the assay with 1 μm beads were obtained by measuring the average layer intensity at three locations on each sample. The limit of detection (LOD) for each experiment was determined from the mean signal intensities of the negative controls (experiments performed with no IgG) incremented with their 2-fold standard deviation. 1 . 2 . Results

Arrays of particles decorated with biomolecules for biorecognition and immobilized within a three- dimensional hydrogel matrix were obtained. In the first step, a stack of several pa'rticle layers was prepared by successive dipping of a support slide into solutions containing the particle of interest and gel formation by cooling on a support slide. The so-obtained hydrogel/particle construct was then cut in thin slices perpendicularly to the deposited layers so that arrays consisting of parallel columns of the different bead populations were obtained. Thus, the strategy presented enables the fast and easy production of a multitude of array without any instrumentation. Using microparticles as platforms for the bioassay has several advantages: Latex particles have long been used in biological and bioanalytical assays so that latex beads with a variety of chemical or biological functionality are commercially available and a variety of protocols for their surface functionalization have been published. With the protocol relying on particle functionalization in solution and particle immobilization within a hydrated hydrogel environment, drying steps common to spotting procedures and potentially harmful to proteins (since they can lead to denaturation and loss of functionality) can be easily avoided. Bead-based systems are also characterized by a great flexibility: The bead populations can be selected from stock solutions and the arrays can be composed freely according to the needs. Because each array element is prepared separately in the bulk, the chemistry and conditions for bioligand surface immobilization can be selected and optimized for each biomolecule individually. Furthermore, particles are three dimensional sensing platforms, which confers the sensor an increased loading capacity compared to the traditional two dimensional configuration. Thus, the array sensitivity can be improved. According to the procedure described above, particle arrays using beads with diameters of 100 μm, 15 μm and 1 μm were prepared. For all particle sizes the bead columns were clearly separated and well-defined. It was found that homogeneous arrays with up to nine distinguishable particle layers can be obtained. In principle, an arbitrary number of layers can be deposited with the approach presented.

Conventional protein microarrays usually consist of spots of several tens of micrometers so that relatively large beads were chosen in the first place. In this case, each sphere can be treated as an individual spot in a planar microarray. Microarrays obtained with 100 μm beads can thus be directly compared to conventional protein arrays and standard low magnification read-out systems, as well as established evaluation protocols are directly applicable. However, for applications aiming at miniaturization and high density arrays, smaller beads might be desirable. With the dipping approach presented, arrays with reduced dimensions were obtained using beads with diameters of 15 μm and 1 μm. The multilayers were obtained by manual dipping and removal of excess solution by pressing the edge of the slide against a paper tissue resulting in limited control over layer thickness. Nevertheless, without any optimization of the dipping protocol, layer thicknesses of less than 100 μm were obtained with the 1 μm beads while the structures obtained with 100 μm beads were usually of several hundreds of micrometers.

Smaller beads also produce arrays with a denser and more homogeneous bead distribution. Finally, it should be noted that, while 15 μm beads can still be individually distinguished and thus evaluated as a single spot, smaller beads produce homogeneous layers, which thus have to be evaluated by average column intensity measurements .

Microarray assays In order to demonstrate the viability and performance of our particle array, several model assays for the detection of proteins were performed. These include reverse phase and sandwich assays using particles with diameters of 100 μm and 1 μm. Assay multiplexing

As a first proof of concept for the protein microarray, arrays consisting of 100 μm beads carrying either BSA-, rabbit IgG, or mouse IgG were produced. Images of three arrays obtained from the same hydrogel stack and incubated either with fluorescent anti-rabbit IgG, fluorescent anti-mouse IgG were produced. The signal was highly specific (s/n for rabbit IgG: 3.8; s/n for mouse IgG: 4.8) with low non-specific binding on the BSA control beads and low antibody cross-reactivity (s/n < 1.15) .

Assay sensitivity

The sensitivity of a microarray of 100 μm beads was first evaluated on a model sandwich assay for the detection of mouse IgG. For this purpose, an array consisting of microparticles carrying either anti-mouse IgG (Fc specific) or BSA (as a negative control) was produced. Mouse IgG was detected after incubation with the IgG containing sample followed by incubation with a fluorescently labeled anti-mouse IgG (Fab specific) . Without any assay optimization, the sensitivity was found to be in the low pM range with a limit of detection of 4±2.6 pM (average and standard deviation of three independent experiments). This value is comparable to sensitivities of standard fluorescent bioanalytical assays .

A reverse phase assay was also performed using 1 μm particles for the immobilization of a model analyte consisting of BSA with IgG spiked in. A typical array image and resulting dose-response curves for two concentration ranges were produced. The signal was linear for approximately three orders of magnitude and the limit of detection for this assay, was 1.6 nM IgG in the presence of 75 μM BSA. This corresponds to a ratio target/total protein content of approximately 1/50000 proteins .

Array platforms with signal enhancement capability, such as the three dimensional matrix presented here, are likely to play an essential role in the implementation of reverse phase arrays. As a matter of fact, in a reverse phase array, the protein of interest is immobilized on the chip in the presence of a complex sample, usually a cell lysate or a biofluid, without any purification step or "fishing out" from solution, as it is typically the case for a capture array format. Thus, a major limiting factor for the assay sensitivity is the number of target proteins immobilized on the spot within the protein mixture. The increase in loading capacity using spherical particles can be estimated from simple geometrical considerations and will depend on the particle diameter, the particle density and the thickness of the gel. Furthermore, the approach presented enables the rapid production of multiple array copies from a small sample volume, another common prerequisite in reverse phase microarray technology.

2. Preparation of m±croarrays obtained from a three- dimensional construct comprising channels containing a bioanalytic ("channel-based" approach)

Alternatively to the "stacked layers" approach, a microarray as schematically shown in Fig. 2 (c) was obtained by preparing a three-dimensional construct comprising channels containing a ligand attached to a (micro) particle, as schematically shown in Fig. 2 (a) and 2 (b) , respectively, and by slicing the resulting structured three-dimensional construct. Thus, in the example shown in Figures 2 (a)-(c), the channels are obtained by moulding.

2.1. Experimental Materials The following antibodies were purchased from Sigma,

Switzerland: anti-human IgG (Fc specific, produced in goat) , human IgG, anti-human IgG-FITC (fluorescein isothiocyanate) (Fab specific, produced in goat), mouse IgG, and rabbit IgG. Alexa Fluor® 488 anti-rabbit IgG (produced in goat) and Alexa Fluor® 633 anti-mouse IgG

(produced in goat) were from Invitrogen, Switzerland.

Fluka agarose for molecular biology (gelling temperature:

34-37 ⁰C) was used for the array support structure (Sigma, Switzerland, #05066) . Ultra low gelling SeaPrep® agarose

(gelling temperature: < 17 ⁰C, melting temperature: 40-

50 ⁰C) was used to fill the channels (Lonza, Switzerland) .

Polybead® carboxylated 0.5 μm polystyrene microspheres were embedded in the channels for biological assays (Polysciences GmbH, Germany) . The surface of the particles was blocked with bovine serum albumin (BSA) (≥98%, Sigma-

Aldrich, Switzerland) . All arrays were prepared in a HEPES buffer solution, consisting of 10 mM 4- (2-hydroxyethyl) piperazine-1-ethane sulphonic acid (Sigma-Aldrich, Switzerland) and 150 mM NaCl, with the pH adjusted to 7.4.

Array Preparation

The channel support structure was prepared by dissolving agarose in HEPES buffer (3% w/v) while applying constant heat. The melted agarose was immediately injected into a mold using a syringe (both preheated to 45 ⁰C) . The mold is a metal chamber containing an array of 25 pins. The pins are 2 cm long, 500 μm in diameter and are arranged in rows of five. The support structure gelled around the pins for 20 min at room temperature. The pins were then gently extracted from the gel block and replaced with an addressing plate. The addressing plate has graduated channels from 700 μm to 500 μm for injecting the hydrogel/particle mixture with a pipette.

To fill the channels, 3.25% (w/v) SeaPrep agarose was added to HEPES buffer and melted at T > 50 ⁰C. The agarose was aliquoted into Eppendorf tubes and cooled to 40 ⁰C in a dry block heating system (Grant Instruments, England) . The 0.5 μm-functionalized polystyrene particles were diluted in agarose (dilution factor 1:2.6), giving final concentrations of 0.1% particles (w/v) and 2% SeaPrep agarose (w/v) , respectively. A pipette was used to inject 4 μl of the particle/gel solution into each channel. The hydrogel array block was immediately submerged in HEPES and left to gel at 4 ⁰C. The channels form a gel after

3 hrs, however a longer gelation time improves the mechanical stability of the arrays. Arrays for reverse phase assays and drying experiments were cooled overnight, while arrays used for the sandwich assays were left at

4 ⁰C for 6 hrs.

The array block was cut with a scalpel into -1-2 mm slices. After incubation and rinsing, the microarray slices were sealed between a coverslip and a microscope slide for imaging.

Particle Functionalization

Polystyrene microspheres were functionalized for the reverse phase assay with either mouse or rabbit IgG and for the sandwich assay with anti-human IgG (Fc specific) .

Beads coated with BSA were the negative control for both assay types. To functionalize the particles, 50 μl of the

0.5 μm beads (2.62% w/v) were washed twice in 1.5 mL of HEPES buffer by exchanging the supernatant with buffer after centrifugation. To remove the supernatant, the beads were spun down with a microcentrifuge (14000 x g, 10 min) .

Following the washing, the beads were shaken overnight

(1000 rpm) in 1 mL of either mouse IgG (250 μg/mL) , rabbit IgG (250 μg/mL), anti-human IgG (Fc specific) (220 μg/mL), or BSA (250 μg/mL) . The beads were then washed two more times in 1 mL of HEPES before blocking with 10 mg/mL BSA. The incubation steps for the blocking were each 30 minutes of shaking at 1000 rpm (Eppendorf Thermomixer, Germany) . BSA was replaced with 50 μl of buffer for storing the beads at 4 ⁰C.

Diffusion

Fluorescence recovery after photobleaching (FRAP) was used to quantify the diffusion of IgG in 2% SeaPrep agarose. To prepare gel samples, 4% (w/v) SeaPrep agarose (prepared as described above) was mixed in equal parts with IgG-FITC (100 μg/mL) and injected into a thin chamber on a glass slide. The gelation chamber is formed by two spacers (thickness: 150 μm) and a coverslip. The gel-filled chamber was sealed and submerged in HEPES buffer for 6 hrs at 4 ⁰C. The gel vertically displaced the spacers, increasing the slice thickness to ~200 μm.

Fluorescence recovery images were taken at irregular intervals after bleaching a circular area (radius: 35 μm) in an IgG-loaded gel. The diffusion coefficient was determined from the fractional fluorescent recovery curves, based on the theories of Axelrod and Soumpasis. The recovery profile was assessed to ensure diffusion is predominantly two-dimensional before fitting the data. FRAP analysis was performed on six measurements from three independent experiments using ImageJ software (Image processing and analysis in Java, National Institutes of Health) .

Microarray Assays Array slices for the reverse phase assay were prepared with mouse IgG-, rabbit IgG- or BSA-coated beads. The slices were incubated overnight on a flat shaker in a HEPES buffer solution containing both AlexaFluor488 anti- rabbit IgG (5 μg/mL) and AlexaFluor633 anti-mouse IgG (5 μg/mL) . The arrays were quickly rinsed three times with HEPES, by injecting and removing the buffer with a pipette, and then gently shaken in 2.5 mL of buffer for 2 hrs . Array slices for the sandwich assay had an alternating pattern of BSA- and anti-human IgG (Fc specific) -coated beads. The slices were incubated overnight under gentle shaking in concentrations of human IgG ranging from 0.1 pM to 1OnM. Human IgG dilutions were prepared in 100 μg/mL of BSA. Arrays were rinsed by gentle shaking in 4 mL of buffer for 2.5 hrs, exchanging the buffer every 30 min. The slices were incubated for 2 hrs in the detection antibody (5 μg/mL of anti-human IgG (Fab specific) -FITC) , followed by the same rinsing procedure. Imaging

Microarrays were imaged using a Zeiss LSM 510 Confocal Laser Scanning microscope (Carl Zeiss, Germany) . Fluorescently tagged antibodies were excited with either a 488 nm Argon laser (FITC, and AlexaFluor488 ) or a 633 nm Helium Neon laser (AlexaFluor633) . The emission filters used were Zeiss LP505 (green) or LP650 (red) . Images for determining array sensitivity were taken with a 10x EC Plan Neofluar objective (N. A. 0.3, optical slice 50.4 μm) , while images of the entire array were composed from a series of images taken with a 5x objective (EC Plan Neofluar N. A. 0.16). Images of fluorescent recovery after photobleaching were taken with a 4Ox LD Plan Neofluar objective (N. A. 0.6, optical slice 17.7 μm)

A planar waveguide-based microarray reader (ZeptoREADER, Zeptosens, Switzerland) was used to image microarrays as the channels dried. An array slice from a reverse assay wa placed on the Ta₂θ₅ waveguide and excited using the red channel of the Zepto READER (635 nm, 3 s illumination, grey filter 1 ) .

Limit of Detection The signal-to-background (s/b) ratio was calculated from the mean intensity of a circular area, 500 μm in diameter and centered over the array spot, divided by the mean intensity of the background. The background was the average signal from a 0.3 mm² border around the image. The dose-response curve is a plot of the mean signal-to- background and standard deviation of three independent experiments. To quantitatively compare images, the detector gain, amplifier offset, laser power, and pinhole were kept constant. For each experiment, twelve microarrays were prepared from the same particle/hydrogel mixture and incubated in different antigen concentrations. The signal-to-background for each array was the average from five spot replicates. The limit of detection was determined from the average signal-to-background of the negative control (arrays incubated in BSA followed by the detection antibody) incremented by 3x the standard deviation.

2.2. Results With the channel-based system, microarrays comprising 25 hydrogel spots, each containing a large number of antibody-coated polystyrene microparticles, were prepared. The channels were formed in agarose by molding the gel around an array of pins. Then SeaPrep agarose was mixed with the particles and injected into the channels (see Fig. 2 (a) ) . These hydrogel blocks were sliced perpendicularly to the channels, producing multiple copies of the microarray, as shown in Fig. 2 (c) . After manual slicing, the 500 μm spots containing biofunctional particles were round and well defined.

Particles with a diameter of 0.5 μm are physically trapped in the hydrogel matrix, as can be demonstrated with a bleaching experiment. It was shown that the particles did not diffuse into the bleached area during a period of 14 hrs, indicating that they are immobilized in the SeaPrep agarose. Even though the beads were physically trapped after only 3 hrs of cooling at 4 ⁰C, it was found that a longer gelation time increases the reliability of producing mechanically stable arrays. When stored in buffer, the hydrogel structure was stable for several months after preparation.

The particle-based fabrication technique uses standard laboratory equipment and fluorescence-based read-out to create custom arrays of biomolecules . This approach is highly flexible, as protein immobilization is not restricted to any particular gel chemistry. Several techniques have been developed for attaching proteins to microparticles, including physical adsorption, covalent coupling, and specific non-covalent attachment with affinity tags. Many varieties of protein-coated particles are also commercially available. In the system according to the present invention, capture probes are immobilized by physical adsorption, making functionalization especially simple. As a result of this flexibility, the technique according to the present invention is not only limited to antibody detection; the agarose channel support can be filled with a broad range of biologically relevant molecules according to the user's need.

With this simple fabrication technique, multiple microarray or biochip copies can be produced quickly and economically. The agarose channel support is ready in 30 min, and preparing a block of microarrays takes less than 8 hrs.

Protein Diffusion

The array spots preferably contain a non-fouling hydrogel with a large pore size (radius of IgG ~7 nm) to ensure that proteins can diffuse quickly to their capture probes. Diffusion of proteins through the hydrogel channels was tested with fluorescence recovery after photobleaching (FRAP) experiments. Full recovery was observed indicating that no proteins were trapped in the gel. The diffusion coefficient for IgG in 2% SeaPrep agarose was found to be 1.34 x 10^"7 ± 0.22 x 10^"7 cm²/s, calculated from six fluorescence recovery curves.

These results illustrate that proteins as large as IgG readily diffuse through low concentrations of SeaPreap agarose, while larger particles remain physically immobilized. According to literature, the pore size of agarose is typically around 100 nm. The exact value depends on the concentration, gelation conditions, gel type, and method used to determine the pore size.

Hydroxyethyl groups decrease the porosity of agarose; the size of the pores in 4% (w/v) SeaPrep agarose is ~42 nm. Despite these findings, the diffusion coefficient reported here is comparable to values found in literature for 2% agarose (2.59 x 10^"7 ± 0.21 x 10^"7 cm²/s). The microarrays were sliced approximately 1-2 mm thick and, as a result, each array spot is a short cylinder of functionalized particles embedded in the agarose support. To demonstrate array functionality and that proteins diffuse through the gel slices, a binding experiment was performed using beads coated with anti-human IgG (Fc specific) . The array slices were imaged at various depths within the gel. A strong fluorescent signal and signal-to- background (s/b) near the middle (2.2 s/b at 0 μm and 1.9 s/b at 500 μm) was detected, indicating that antibodies were able to completely diffuse through the array slices and do not only bind to capture probes on the surface.

Multiplexing

A model reverse phase assay was used as a proof of concept and to demonstrate the multiplexing capability of the system described. In a reverse phase assay, the target biomolecule is directly immobilized on the bead surface along with all other biomolecules that are present in the sample. In the model assay described, IgG was the target and BSA represented the non-specific other molecules. Specifically, a pattern of BSA-, rabbit IgG-, and mouse IgG-coated beads was injected into the channels. The odd rows alternate rabbit IgG with BSA, and the even rows alternate BSA with mouse IgG. The array slices were incubated in a solution of fluorescently labeled anti- mouse and anti-rabbit IgG. Between each fluorescent spot there was a negative control of only BSA-coated beads. The fluorescent signal was highly specific (s/b for rabbit IgG: 3.8; s/b for mouse IgG: 3.1) and had low cross- reactivity (s/b < 1.0) and low unspecific binding on the BSA spots (s/b = 1.0). Detection Limit

The sensitivity of the system was evaluated with a model sandwich assay for detecting human IgG. Beads coated with anti-human IgG (Fc specific) and control BSA beads were arranged in a checkerboard pattern. The arrays were incubated overnight in a BSA solution spiked with concentrations of human IgG ranging from 0.1 pM to 10 nM. The incubation and rinsing times were tested to ensure antibody binding reaches equilibrium and any unbound proteins are washed from the matrix before imaging. The time needed for the sandwich assay is consistent with standard assay protocols.

The average dose-response curve from three independent experiments was determined. The LOD is 12 pM for the human IgG sandwich assay. This value is comparable to standard fluorescent-based immunoassays, even though manual array preparation and bead functionalization decrease experimental reproducibility. The LOD for individual experiments is around 2 pM, representing only the variation between array spots without considering the variations due to sample handling. The fluorescent signal on the BSA beads was s/b = 0.97 ± 0.034, indicating low non-specific binding.

Concentrating the Sample

SeaPrep agarose channels collapse rapidly when exposed to air, but the agarose support maintains structural integrity for longer. As a result, the hydrogel slices can be intentionally dried on a solid surface, after biorecognition in an aqueous environment, to form a planar microarray. Imaging the dried spots could increase the sensitivity of our system because the fluorescent markers embedded in SeaPrep agarose become concentrated on the surface .

An array of fluorescent particles and reference beads was imaged periodically while the hydrogel spots dried. Images were taken with the ZeptoREADER (a microarray reader based on planar waveguide technology) , which uses high illumination intensity to detect fluorescent molecules within the evanescent field (-200 nm) . The ZeptoREADER achieves signal to noise ratios up to 78 times higher than a confocal scanner. However, because of this thin optical slice, it is important to pack a large number of fluorescent probes very near the surface. The particles in our hydrogel microarrays are initially immobilized throughout a gel slice ~10000 times thicker than the optical slice of the ZeptoREADER. This means that the hydrogel slices are initially too thick to take advantage of this sensitivity since, before drying, only very few fluorescent molecules can be detected.

It was found that the intensity drastically increases as the spots dry, indicating that through this process the fluorescent particles are concentrating in the evanescent field. The spot signal determined in this example increases 62 times for the four spots of fluorescent microparticles, the signal to noise (calculating noise on the surrounding Agarose) increases by a factor 1.9.

(Signal from 784 ± 181 to 48800 ± 9070, S/N from 11.9 ±

4.8 to 22.5 ± 2.2) The non-fluorescent reference spots increase their brightness and noise much like the background.

By concentrating all the fluorescent probes from a gel slice into the evanescent field, we improved the detection limit of our system. The potential sensitivity increase is not yet fully exhausted due to the increased background during the drying. As this imaging approach does not affect array fabrication, custom arrays are still produced rapidly and can be stored for extended periods before drying.

As a conclusion, a simple and inexpensive method for producing multiple copies of hydrogel microarrays can also be obtained by the channel-based approach. Thereby, arrays can be fabricated without any special instrumentation and limits of detection comparable to standard fluorescence- based immunoassays can be reached.

As for the stacked layer approach, SeaPrep agarose is compatible with the channel-based approach because it has low non-specific binding, is injectable at room temperature, does not denature proteins during gelation, and has a pore size that permits biomolecular diffusion while physically trapping microparticles . Nevertheless, the technique is not restricted to thermoreversible gels. Protein resistant photo- or chemically crosslinkable gels with a sufficient pore size could replace agarose in the channels to reduce array preparation time (e.g. alginate or polyacrylamide-based gels) . With these modifications, proteins could also be directly coupled to the gel matrix, as an alternative to microparticles as the supports for biorecognition.

As is also the case for the stacked-layer approach, agarose is also advantageous as a support structure because it can be sliced manually. However, thin array slices are fragile and must be handled carefully. By embedding the hydrogel channels into a stronger non- permeable support (e.g. polydimethylsiloxane (PDMS)), it is possible to integrate a flow-through system for rapid assay analysis using minute sample volumes. For hydrogel- based arrays, accelerating the reaction-diffusion kinetics significantly shortens the analysis times. Flow-through analysis also allows spots to be addressed individually, further increasing the flexibility. PDMS also facilitates denser array fabrication by stacking structures of periodic trenches to form the microchannels obtained with conventional photolithography methods. Thus, also by the fabrication method according to the channel-based approach, protein microarrays are more accessible by eliminating the need for costly machinery. The system is highly flexible: A wide range of biological molecules can be immobilized in the channels to create multiple copies of custom arrays. The ability to rapidly produce cheap, sensitive, and flexible arrays is important for any high-throughput application. The process of the present invention thus has great potential as an alternative to traditional robotic spotting and lithographic techniques.

A possible layout of a microfluidic device for the individual filling of the microchannels is shown in Fig. 3. Fig. 3 shows how different microstructured layers can be placed on top of each other to result in a device where arbitrary rows and columns of channels can be individually filled with the permeable matrix (e.g. the hydrogel) . Thus, in the microfluidic device shown in Fig. 3, the channels have been obtained by stacking ("stacked layers" approach) .

Claims

1. A bioanalytical device comprising a sensor and a thin slice obtainable from a structured three-dimensional construct containing ligands embedded in a permeable matrix in a repetitive manner.

2. The device according to claim 1, wherein the ligand is selected from the group consisting of a biomolecule, more particularly a protein and an oligonucleotide, a small molecule, a cell and a cell fragment.

3. The device according to claim 1, wherein the permeable matrix is formed by a compound, a first portion of which is in solid phase and a second portion of which is in liquid phase, the matrix being permeable to biological ligands and compatible with biological assays.

4. The device according to claim 3, wherein the permeable matrix is a hydrogel, in particular an agarose gel.

5. The device according to any of the preceding claims, wherein the permeable matrix is embedded in a supporting device.

6. The device according to any of the preceding claims, wherein the three-dimensional construct has a layered structure, which is in particular obtained by layer- by-layer assembly, the layers being more particularly formed by dipping or spin-coating, in microfluidic channels or by stacking individual layers.

7. The device according to any of the preceding claims, wherein the three-dimensional construct comprises channels filled with the permeable matrix.

8. The device according to any of the preceding claims wherein the ligands, in particular the bioligands, are directly attached to the permeable matrix, in particular by covalent coupling.

9. The device according to any of the preceding claims wherein an additional supporting material, in particular a micro- or nanoparticle, is used to embed the ligand therein.

10. The device according to any of the preceding claims wherein the individual channels or layers contain different biological samples.

11. The device according to any of claims 7 to 10, wherein the channels are part of a microfluidic device .

12. The device according to any of claims 7 to 11, wherein a row of channels is prepared with suitable in- and outlets and stacked together to obtain an array.

13. A planar microarray comprising separate areas, each area comprising a ligand, wherein the microarray comprises a permeable matrix in which the ligands are embedded.

14. The microarray according to claim 13, wherein it is essentially made of the permeable matrix with the ligands embedded therein.

15. The microarray according to claim 13 or 14, wherein it is obtainable from a structured three-dimensional construct, which comprises separate areas extending in an axial direction and containing in each case at least one of the ligands, by slicing the construct in a plane at least approximately perpendicular to the axial direction.

16. The microarray according to any of claims 13 to 15, each area comprising a different ligand.

17. A process for obtaining the planar microarray according to any of claims 13 to 16 comprising the steps of: a) forming a structured three-dimensional construct comprising separate areas extending in an axial direction, said areas containing a ligand embedded in a permeable matrix; and b) slicing the three-dimensional construct in a plane at least approximately perpendicular to the axial direction.