Resonant nano-cluster devices . J. Haglmuller, H. Rauter, G. Bauer, F. Pittner and T. Schalkhammer Abstract: The resonance-enhanced absorption (REA) by metal clusters on a surface is an effective technique on which to base bio-optical devices. A four-layer device consisting of a metal mirror, a polymer or glass-type distance layer, a biomolecule interaction layer and a sub-monolayer of biorecognitively bound metal nano-clusters is reported. Experiments indicate a strong influence of the resonator homogeneity on the absorption maximum. Layer stability plays an important role in the overall performance of the device. Techniques and optimised lab protocols to set up biochips that use the REA process in the detection are presented. The sensors show one to three narrow reflection minima in the visible and or infra-red (IR) part of the spectrum and therefore they do not suffer from the spectral limitations associated with spherical gold colloids. Metal clusters (synthesised by thermal step reduction) as well as metal- dielectric shell clusters (synthesised by various shell deposition processes) are used to precisely shift the readout of the device to any frequency in the visible and near IR range. Disposable single-step protein chips, DNA assays as well as complex biochip arrays are established that use various DNA/RNA, antigen–antibody and protein–protein interaction systems.
1
Introduction
Metal nano-particles, -colloids and -clusters are chosen for use as signal transducers in molecular interaction and biorecognition studies because of their superb extinction coefficients (Fig. 1). No conjugated organic chromophore is able to compete with the high energy uptake of metal nanoclusters. Metal clusters are more intensely coloured than any conjugated organic molecule. This fundamental property has resulted in gold and silver staining methods being used since the mid-20th century. Colloid capture assays enable us to visualise the binding of biomolecules at a given surface by a bound layer of metal clusters. A wide variety of
photon
plasmon
− e - hole - pair non-radiative decay
radiative intra-band Ef
decay h
intra-band d-band
Fig. 1
Optical cluster spectra
r IEE, 2005 IEE Proceedings online no. 20045013 doi:10.1049/ip-nbt:20045013 Paperreceived 22nd December 2004 J. Haglm.uller and F. Pittner are with the Institute of Biochemistry and Molecular Cell Biology, University of Vienna, Austria H. Rauter is with the Attophotonics Bioscience Schalkhammer KG, Alland, Austria G. Bauer is with the November AG, Erlangen, Germany T. Schalkhammer is with the Institute of Analytical Biotechnology, Technical University Delft, The Netherlands E-mail:
[email protected] IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
simple bio-assays based on deposition and aggregation of colloids and clusters have been developed. In comparison direct approach to detection using organic or inorganic chromophore particles turn out to be either very insensitive (without using additional amplification by an enzyme) or require large particles resulting in severe limitations due to reduced diffusion and precipitation. Same modern detection schemes make use of the light scattering of gold nano-clusters to detect DNA hybridisation on glass slide biochips. However, it is often difficult to discriminate between dust particles and metal clusters. The success of colloidal techniques is based on the wide variety of physical and chemical techniques available to synthesise clusters. The advantageous properties of a simple synthetic route, a narrow size distribution and a straightforward surface modification with bio-ligands enables the application of colloidal clusters as transducers in biorecognitive binding and molecular structure studies. The fundamental properties are: (i) monodisperse particles; (ii) simple chemical synthesis; (iii) efficient coating by biomolecules; (iv) coupling with various ligands; (v) electron-dense materials; (vi) strong particle plasmons; (vii) direct visualisation; (viii) cluster light-scattering; (ix) catalytic size enhancement; (x) coating with other metals; and (xi) no photobleaching under illumination. Based on these key advantages nano-clusters have become the basis of devices that employ cluster resonance, cluster-cluster interactions and cluster field effects. 2 Resonance-enhanced nano-cluster colour effects The reason for noble metal colloids to be intensely coloured can be deduced from theory. The first observations of ruby coloured gold sols date back to Faraday. The fine tuning of cluster composition, size, solubility, plasmon effects, scattering, and stability has received considerable attention in the last decade. Resonance-enhanced absorption (REA) effects are observed if a light-absorbing cluster or cluster layer is 53
positioned at a nano-distance from a reflective mirror and illuminated from the particle side (Fig. 2). At a certain wavelength the reflected electromagnetic wave has the same phase as the incident wave at the position of the absorbing nano-cluster particle. This assembly is a nano-interference system. Its feedback mechanism intensifies the absorption of
light at a well defined wavelength, resulting in ‘anomalous absorption’. At a specified mirror – cluster distance and a well defined angle of light reflection (and observation) only a subset of waves, in a narrow range of wavelengths, is in phase. The almost complete absorption of those light waves and the high reflectivity at any other wavelength results in a visible colour response (Fig. 3). The REA setup when combined with a biological recognition system (such as DNA-DNA or protein-protein) is able to generate an optical signal in response to: (i) changes in the mirror – cluster distance; or (ii) changes in the packing density; (iii) the exchange in the spatial arrangement of a cluster layer; and (iv) changes in the number of clusters [1–10].
metal cluster ligand + receptor resonance
distance layer mirror CHIP
Fig. 2
Such changes modify the colour due to a spectral shift of the absorption maximum or the colour intensity due to a
The REA-based assay
0−2 absorbance
increasing layer thickness
400 500 600 700 800 900 1000 1100
400 500 600 700 800 900 1000 1100
400 500 600 700 800 900 1000 1100
wavelength, nm
wavelength, nm
wavelength, nm
0−2 absorbance
increasing layer thickness
400 500 600 700 800 900 1000 1100
400 500 600 700 800 900 1000 1100
wavelength, nm
wavelength, nm
AFM determind layer thickness, nm
a 80 70 60 red 50 yellwo 40 light blue
30 20
dark blue
10 dark browm 0 0
1
2
3
4
5
6
7
8
9
sputter time b
Fig. 3 Spectral response of a REA - chip with increasing thickness of the distance layer (1-5) and sputter time (distance layer) versus resonance color. A silver mirror, an organic polymer (hexylacrylate) spacer layer and a gold cluster top layer were used to obtain the spectra given below. Spectra were recorded using a reflectance module inserted in an Hitachi photometer averaging over an active optical area of 6 mm2. a Changes in the spectral response of a REA chip as the thickness of the distance layer is increased. A silver mirror, an organic polymer (hexylacrylate) space layer and a gold cluster top layer were used to obtain the spectra. A reflectance module inserted into Hitachi photometer averaging over an active optical area of 6 mm2 was used to record the spectra b The layer thickness as a function of sputter time also shown is the resulting resonance colour 54
IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
change in light absorption (at a defined wavelength, often in the visible or IR spectrum). The main application area of REA devices is in biochips such as DNA or protein arrays. These are used in various fields including DNA diagnostics and drug- and antigen screening. A background colour of the reflecting mirror surface as well as thin film colours created by the multilayer system are additional effects modifying the REA colour spectra. Given that a maximum dynamic range is desirable the setup needs to be built of layers with well adjusted refractive indices. However, biological samples can introduce additional optical absorptions. In such a system the colour response is tuned to compete with the optical background either by adjusting the thickness of the distance layer or by changing the light path (actually the angle of incidence and observation, angle-dependent colour, see Fig. 4). Thin film colours are created by interference and are computed from the refractive indices of the materials used in REA chip design. Using a high-quality mirror and distance layer materials with low refractive indices minimises those thin film colours. Another possibility is the introduction of sandwich layers (with precisely-adjusted refractive indices).
type of reducing agent and added stabilisers define the final product. Quite a number of protocols for the fabrication of gold colloids have been developed in recent years. The most important method remains the one employed by Frens [11] because of its convenience, reproducibility and since it provides mono-disperse, aqueous particle dispersions. Like for any other application in nanotechnology, working with ultra-pure water and clean dishes is crucial for good results!
3.1.1 Clusters and cluster coating: In general, cluster sizes between 5 and 100 nm are used, extended storage times may further influence the choice to be well below 100 nm (due to precipitation problems). Large clusters may exhibit a low diffusion, small clusters a low signal. Thus, the choice depends on the average diffusion distance between the cluster-bio-component and its target and on the available concentration of the bio-component. A low bio-component concentration limits the number of clusters at the target position, in such cases larger clusters are favoured. A gold cluster diameter of around 20 nm is a viable compromise.
3.1.2 Standard technique: gold clusters via the sodium citrate reduction method: This technique was first reported by Frens [11]. The size of the clusters can be adjusted to be in between 6 and 60 nm by adding varying amounts of reducing agent. Protocol for 17 nm colloids:
Fig. 4 Angle-dependent colour of cluster resonance on a goldcoated chip
3
Methods and techniques
The techniques necessary to set up a cluster chip are deduced from colloidal techniques (to design and coat the nanoparticles), from thin film technology (to design and coat the chip with the resonator layer) and finally the (bio)chemical interaction layers.
3.1 Techniques to prepare noble metal colloids Approaches to prepare metal colloids, can be divided into two subgroups: (i) the chemical reduction of metal salts to solid metal; and (ii) physical deposition of the metal using vapour coating techniques. Here, we will focus on the reduction of metal salts in aqueous solutions since bionanotechnological applications usually depend on water dispersed nano-particles. Various clusters can be used in the REA-based assay, including colloids of gold, silver, platinum or gold-coated silver clusters. Gold clusters are easy to produce, stable and easy to use, therefore our main focus will be on them. Gold clusters (being the most important material for colloidal assays, SPR, chip-based readout systems and REA applications) (are usually produced from tetrachloroauric(III) acid (HAuCl4). This salt is dissolved, the charged ions are reduced to atomic gold and (by a nucleation and catalytic growth process) clusters are formed. The choice of solvent, gold ion concentration, temperature, amount and IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
1. 10 mg HAuCl4 xH2O in 100 ml of water is heated to boiling. 2. 4 ml of filtered 1% (w/v) trisodium citrate dihydrate is rapidly added under constant stirring. 3. In about 25 s the slightly yellow solution will turn a faint blue colour (nucleation). 4. After approximately 2 min the blue colour changes into a dark red, indicating the formation of mono-dispersed spherical particles. 5. The solution is boiled for another 10 min to ensure complete reduction of the gold chloride. 6. The suspension can be stored at +41C for several months. The final product shows an absorption maximum (around abs ¼ 1.2) at 520 nm. Reproducibility over the course of more than 20 reactions was observed. If application in the near-IR is necessary Pd-doped Au-clusters allow signal boosting. Reduction of the metal salt with additional tannic acid [12] or aspartic acid [13] is feasible. In our experience citrate is the only reducing agent that has a good reproducibility and it also has the advantage of a clear protocol for clusters of various sizes.
3.1.3 Preparation of coated gold particles with a small diameter: These gold clusters do not show as extensive a plasmon absorption as the ones produced by other methods (e.g. the citrate reduction route of Frens [11]). Due to the fast reduction by the sodium boro-hydride a high number of nucleation cores is created which leads to the formation of numerous particles with a small diameter (o 6 nm). Surfactants (usually thiols with the correct solubility) have to be used for stabilisation. 1. Dissolve 2 mg HAuCl4 in 20 ml of water (0.1 mg/ml). 55
2. Surfactants: add 0.8 mg mercaptosuccinic acid (80 ml of a 10 mg/ml stock) add 1.45 mg mercaptopropionic acid (145 ml of a 10 mg/ml stock). 3. Reduction: drip in sodium borohydride (15 mg/ml, fresh!, cold!) under vigorous stirring until an orange solution develops. Synthesis and stabilisation (arenethiolate, alkanethiolate,y) of gold clusters in a size range of 1,5 to 6 nm is possible using modified versions of the Brust protocol. [14–16]
3.2
Stabilisation of gold clusters
3.2.1 Coating with mercaptohexadecanoic acid: Due to the quite fast exchange and loss of the thiols bound to the metal cluster surface, additional fixation of the surface coating is necessary. By using additional-SH moieties or polymers the exchange rate can be lowered and thus the stability of the colloid is increased. Another simple approach is to employ long-chain mercaptoalcanic acids with an ultra-low solubility in water (e.g. mercaptohexadecanoic acid, MHDA). Three features of these surfactants are important: (i) the -SH group should have a high affinity for the gold cluster; (ii) the long -CH2 chain accounts for hydrophobicity of the core; and (iii) the carboxylic acid makes the outer surface hydrophilic and available for chemical coupling. MHDA protected gold clusters 1. Prepare gold clusters by citrate reduction. 2. To 10 ml of cluster solution add 40 ml of ethanol (for solubility of MHDA). 3. Add 1 mg of the acid (stock: 1 mg/ml in ethanol). 4. Incubate for 2 h at room temperature. 5. Evaporate ethanol. 6. Purify/modify coated clusters. The gold clusters are now trapped in a chemically-bound ‘MHDA-micelle’. Even if the single bond of the mercapto group to the gold core is broken, the molecule cannot leave the shell since mercaptohexadecanoic acid is not soluble in pure water. Shielding of the charge (by the addition of electrolytes) leads to a colour shift from red to blue. In this case aggregation is reversible: simple purification (removal of the salts) and pH adjustment restores the initial colour indicating sole interaction of the shells instead of coagulation of the gold cores.
3.2.2 Silica-coated gold nano-particles: Another approach to stabilise gold colloids is to coat them with a solid shell of e.g. silica, [17, 18]. Most optical properties are preserved: although the plasmon peak (B520 nm for gold nano-particles) broadens and shifts to longer wavelengths. Stabilisation is achieved by a seeded polymerisation technique that derives from the St.ober method for creation of silica particles. Silane (usually tetraethyl orthosilicate, TEOS) is added to an ethanolic/aqueous solution of (gold) particles. The polymerisation is catalysed by addition of ammonia. Nucleation as the initial step in particle formation is not needed here (since the gold particles act as nucleation cores). Growth of the silica shell can be controlled by stopping the reaction after some time or by adjusting the amount of TEOS and letting the reaction run to completion. For reasons of convenience and reproducibility the second possibility is usually chosen. 17 nm gold particles with a 20 nm silica shell: 1. Prepare 17 nm gold clusters by the citrate reduction route. 56
2. To 10 ml of cluster solution (fresh) add 40 ml of ethanol. 3. Add (at least) 0.9 ml of TEOS (for thicker silica shells use up to 20 ml). 4. Add 1480 ml of ammonia (B25%) under continuous stirring. 5. React overnight at 301C. The absorption maximum shifts from 520 nm (gold clusters) towards 529 nm for silica-coated clusters. Multicore (one particle contains more than one gold cluster) and silica beads (no gold cluster included) formation can be significantly limited with this method. Aggregation, coagulation and coalescence of the gold cores can be overcome by silanisation. One of the drawbacks is the rather thick shell that is needed: for stabilisation of a B20 nm gold colloid the silica shell has to exceed B15 nm. This means that the final particle exhibits a diameter of roughly 50 nm!
3.3
Clusters and biomolecules
Experiments have shown that gold nano-clusters are composed of a gold-zero core and a gold-one shell. This is a direct result of the incomplete reduction at the outer surface. Similar behaviour is found in most other cluster systems such as silver, copper and platinum. The reducer used to synthesise the clusters coats the surface via physisorption or chemisorption (e.g. citrate ions in gold colloid synthesis). Added salts turn the ruby red colour to blue. This indicates the beginning of precipitation. However, if polymers or proteins are added under the correct conditions these molecules adsorb on to the metal cluster surface and stabilise it. The Binding of proteins to metal clusters is more or less irreversible due to multiple attachment sites including-SH,-COOH and -NH2. Most of the proteins retain their biological activities, at least in part. The stabilisation by protein (or DNA) shell layers can be proven by the salt precipitation test. Whereas unstabilized clusters precipitate at around 20 mM of sodium chloride stabilised clusters will tolerate 200 mM or more without any aggregation. Ease of preparation allied to chemical and optical robustness make protein and DNAcoated clusters useful tools in various assays and biochips.
3.3.1 Cluster coating with protein: The Biological activity of protein shells depends on their structural integrity and will vary from protein to protein. Binding of the protein to the cluster and its activity are pH dependent. Although there is no general rule, a pH of around one-step above the isoelectric point proved to be best for most proteins. Note: glass electrodes are prone to irreversible coating by colloids! Low salt buffers should be used in order to prevent cluster aggregation. On the average, one cluster is covered with tens to hundreds of proteins (10 nm cluster Bten proteins, 20 nm B50, 40 nm B150). Thus, even poor immobilisation or biological activity results in quite useful cluster reagents for enzymatic applications. Due to the superior signal intensity most assays employ clusters of around 40 nm in size loaded with 100-200 protein molecules. For proteomics very tiny amounts of the proteins are available. In this case it is vital to determine the minimum amount of stabilising protein or to supplement it with cheap proteins such as BSA. Appropriate amounts of protein can be found by adding increasing amounts of protein and assessing stabilisation by addition of NaCl. Protein coating can often be done within 20 min if proper buffer conditions are chosen. Incubation in a 2-3-fold excess for several hours IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
may help if a low binding affinity is observed. Excess protein has to be removed by washing in several centrifugational runs. Stabilisation can be improved by addition of skimmed milk powder, glycerol or PVP. Biotinylated poly-L-lysine monolayers can be adsorbed onto gold clusters and subsequently be used for avidin conjugation. Antibodies yield excellent recognition clusters. Preparation and purification of protein-coated gold clusters: 1. Prepare gold colloid by the citrate reduction route. 2. Add protein, incubate, block with additional BSA (if required). 3. Glycerol gradient centrifugation: 50% (v/v) glycerol on 80% (v/v) glycerol. 4. Speed and running time are adjusted to the point at which the clusters of desired size are found in the 50% glycerol layer. 5. After centrifugation, unbound protein is in the upper water layer. Colloid aggregates are in the 80% glycerol layer and on the bottom of the centrifuge tube. 6. Contamination of the stabilised clusters can be avoided by removing them sideways out of the 50% glycerol layer with a syringe. 7. The obtained colloids in glycerol are diluted and the glycerol is removed in five centriprep-30 runs. 8. If required, the suspension is concentrated. 9. The suspension is stored at +41C.
3.3.2 Cluster coating with DNA: Covalent attachment of amino-terminated DNA oligos to MHDAprotected gold clusters by carbodiimide (e.g. EDC) chemistry is feasible. The drawbacks of this method are possible side-reactions with the phosphate groups on the DNA backbone and formation of urea- derivatives during reaction. Because these compounds destabilise the gold clusters, fine tuning of the stabilisation procedures and reaction conditions is required. The stabilisation of gold clusters using SH-DNA was shown in [19–21] and is useful for REA chip applications. A terminally-modified DNA oligonucleotide (thio-linker) is used to coat the colloids. Site-specific interaction of DNA is used to bind clusters to complementary sequences. Preferentially, clusters of around 40 nm in diameter are used. The gold surface is coated with a dense layer of DNA/ oligonucleotides of 50-200 molecules (depending on size, length and DNA conformation). It should still be kept in mind that thiol-coupled molecules are bound to any gold surface via labile bonds and a thermo stability of 4 401C requires further stabilisation by multiple thiol anchors or silane chemistry. Coating of gold clusters with oligonucleotides: 1. Prepare the gold colloid using the citrate reduction route. 2. Adjust to B20 nM in particle concentration. 3. Add excess of a thio-functionalised oligonucletoide (3.5 mM) in water. 4. Incubate for 24 h at room temperature. 5. Centrifuge to remove excess DNA. An approach to bind DNA to the clusters that has the advantage of an easy protocol and commercially available formats is as follows: IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
1. Follow the protocol ‘Cluster coating with protein’ with avidin as the protein. 2. Purify the avidin-coated colloid according to the protocol ‘Preparation and purification of protein-coated gold clusters’. 3. Dilute biotin-labelled DNA in dH2O to approximately 20 mM. 4. Incubate the purified avidin-coated colloid with biotinlabelled DNA for 20 min at ambient temperature on a shaking platform. 5. Purify the DNA-biotin-avidin coated colloid according to the protocol ‘Preparation and purification of protein-coated gold clusters’. A clear correlation between the maximum signal and the speed of signal development as a function of the cluster size is given in Fig. 5.
100 80 60 45
40
percent colour intensity dots visible , s cluster size, nm
20 12
0 1
2
3
4
5
Fig. 5 Kinetic analysis of the REA-based cluster-capture protein assay with quantification via a reflectivity chip scanner. An antigenantibody assay employing IgG was used to characterise the maximum signal and assay time as a function of the cluster size. An aluminum mirror, a nitride spacer layer and gold clusters of variable sizes (coated with the antibody via adsorption) set up the chip
3.4
Chip design
REA effects are observed if nano-clusters are positioned ten to a few hundred nanometres away from a mirror. To setup a REA biochip a number of elements need to be designed:
3.4.1 Support: Any flat surface can be used as a support, but the mirror coating process as well as surface activation processes define minimum requirements for thermal and chemical stability. Polymer sheets such as polyethylene terephtalate foils are commercially available in metallised forms. Printing with mechanical microarrayers on plastic sheets sometimes damages the surface. However, these materials are well suited for non-contact printing. Highly polished metal surfaces such as aluminium disks have the intrinsic advantage that the mirror is already part of the chip per se. However, microspotting with pin–and–ring systems cannot easily be performed on this type of substrate due to the irreversible bending of the material. Microscopic glass slides, being the standard format of most commercial instruments, are an excellent choice for REA biochips.
3.4.2 Mirror: One essential feature for REA-based assays is a highly reflective mirror, significantly limiting the choice of materials. Aluminium and silver fulfill this requirement, although these materials are chemically unstable and have to be coated for corrosion protection. Adhesion of silver on glass is rather poor thus an adhesion layer of titanium or tungsten and for some applications an interlayer of palladium is required. 57
Many other materials can be used as mirror materials including tungsten, platinum, gold or chromium but they exhibit a considerable thin film colour. Such a background colour decreases the signal-to-noise ratio considerably, especially if it competes with the REA colour. Silver and aluminium mirrors can be applied by evaporation, sputter coating techniques or via chemical metallisation. However, the application of these materials must be carefully optimised for a high reflectivity as well as adhesion of the metal film to the support.
3.4.3 Distance layer (resonance layer): As cited above, the small diameters (tens of micrometres) of the spotted proteins or DNA requires a high quality for the material. The REA effect will transduce all in-homogeneities of the chip surface, which can result in a colour shift on a slide surface or within a batch of slides. Although, polymer films have the advantages of a low refraction index (which reduces thin film colours) and ease of chemical surface activation, these materials are rather difficult to apply homogeneously to large surfaces. Furthermore, some polymer materials react to changes in the environment, such as humidity, temperature or light. Glass, metal oxides and metal nitrides on the other hand have higher refraction indices. This results in a significant background colour, therefore the mirror quality is of greater importance for this approach. Metal nitrides are chemically rather stable, can with stand changes in the environment and are effectively coated by reactive sputtering. Glass as an interlayer material has the advantage of a well developed chemistry for chemical activation (and biomolecule attachment), but cannot be applied by standard sputter coating techniques.
3.4.4 Protocol to manufacture REA slides 1. Thoroughly clean a microscopic glass slide for use as a support. 2. Coat with tungsten or titanium to act as an ‘adhesion’ layer. Thickness: 5 – 10 nm. Application by sputter coating. 3. Coat with silver or aluminium (as the mirror layer) by sputter coating or evaporation techniques (B100 nm thick). The thickness of this layer can be checked using light studies transmission: the eye should qualify the slide as ‘not transparent at all’. 4. Apply tin nitride using sputter coating (tin as the target, nitrogen as the sputter gas), alternatively zinc oxide, titanium oxide, aluminium oxide or tin oxide or combinations of these materials to act as the distance layer (e.g. by sputter coating, zinc or tin as the target, oxygen as the sputter gas). The height of this layer should be chosen according to a strong colour readout (e.g. 70 nm optical thickness). 5. Store slides, dust free, in a plastic box until use. For a typical chip and a colour calibration bar see Fig. 6. Remark 1: When silver is used as a mirror layer the interlayer needs to be applied immediately to protect the silver from corrosion (sulphide!). Alternative protocol: 1. Aluminium (highly reflective) sheets or discs as the support. 2. Punch out or cut (do not bend!) 3. Sputter coat of aluminium oxide (aluminium as the target, oxygen as the sputter gas), or by an eloxation 58
Fig. 6 Optical appearance of the REA chips with a cluster-colour control strip tuned to pink and dark-blue resonance colours
process. Alternatively zinc oxide or tin nitride as the distance layer. Remark 2: The aluminium substrates are rarely perfectly flat and this might cause problems with contact printers. The application of aluminium oxide with a sputter coater requires a professional device. Standard DC coaters will not work. A quick and easy protocol to produce protein binding REA chips is to coat the mirror with a nanometre-scale layer of polystyrene (PS). A solution of PS (20 mg/ml) in toluene is directly spincoated (acceleration: 4500, final speed 3100 rpm) onto the mirror surface. The binding capacity of the PS layer makes this approach very useful for REAbased protein chips. Clean surfaces are crucial for homogeneous REA readouts. Figure 7 shows a spincoating defect due to a dust particle.
Fig. 7 Optical appearance of the REA chips: the spin-coating defect is due to a glass particle
3.4.5 Surface activation: From the variety of surface activation protocols available, one has to be chosen that is compatible with the surface as well as with binding the desired bio-components. In general, for proteins it can be assumed that an amino group will be accessible. DNA probes can be synthesised (and ordered) with amino-linkers. Polymerase chain reaction (PCR) products might need harsh conditions to link to backbone moieties; nitrocellulose interlayers are used to attach PCR products via UV photocrosslinking. IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
Most proteins readily adsorb an to surfaces. However, a covalent link enables the use of more effective washing protocols and therefore improves the signal-to-background ratio. In the case of polymer layers an easy and efficient way of activation is achieved by oxygen plasma etching which results in hydrophilicity of the distance layers surface. Metallic glasses are functionalised by a variety of protocols, often using silane-based chemistry. Amino groups introduced in this way can be modified and activated, using for example carbodiimide or divinylsulphone chemistry, which enables the surface to bind to the carboxyl or amino groups of proteins. Cystamine/divinylsulphone activation for metal oxides such as ZnO or SnO2 surfaces: 1. Immerse the slide surface in 0.1% cystamine dihydrogen chloride in dH2O for 15 min at 401C or at room temperature overnight. 2. Wash the slide with dH2O. 3. Dry the slide in a vacuum centrifuge or with an air gun. 4. Cover the slide surface completely with 0.5% divinylsulphone in 0.1 M Na2CO3/NaHCO3 buffer pH 9 for 15 min at room temperature. 5. Dry the slides in a vacuum centrifuge or with an air gun. Since divinylsulphone is not stable, activated slides should be immediately used for immobilization. 3.4.5.1 Silane-based activation: vapour coating, liquid silanisation: The substrates are put into a vacuum chamber together with the silane (or dilutions thereof), a vacuum is applied and silanisation takes place overnight. Closslinking of the silane is necessary, usually at temperatures between 60 and 1101C, depending on substrate- and silane stability. For the introduction of epoxy groups glycidoxypropyltrimethoxysilane (GOPS) is the silane of choice. Immersing the clean substrates in a fresh 1% (v/v) solution of GOPS in toluene for 1 h, washing the slide three times in toluene and crosslinking at 801C for 10 min yields good results. Alternatively, silanisation in 5% (v/v) GOPS in 95% (v/v) ethanol for 15 min is possible. This protocol is used for toluene-sensitive substrates. Vapour coating techniques are only employed for silanes with a considerable gas pressure. The surface of the substrate needs to hold a considerable amount of available and stable hydroxy groups. This is not the case for most ZnO surfaces.
with 1% BSA phosphate buffer containing 1% Triton, finally wash twice with dH2O. 7. Dry slides, store dust free. Slightly modified versions of this protocol can be used to immobilise PCR products, proteins and sensitive enzymes. Increasing the incubation times and temperature (e.g. 601C for 1 h) enhances PCR product binding. The addition of small amounts of DTT and glycerol can be beneficial for sensitive proteins.
3.6
Assays
DNA hybridisation assays and conventional enzyme linked immuno sorbent assay (ELISAs) can be adapted to the cluster detection protocol, reporting DNA-DNA, antigenantibody or general protein-protein interactions. We present a number of protocols that are adaptable to other biorecognitive interactions. It should be mentioned that the following protocols are described for a contact (pin-andring) micro-spotter but can of course be used with other machines.
3.6.1 DNA: For DNA-hybridisation-based assays the application of amino- labelled probes or PCR products is done as previously described. A typical experiment using 30 bp oligonucleotides is given in Fig. 8. After hybridisation of the labelled target strand the gold cluster is positioned at the correct distance from the mirror and a color signal is created. DNA cluster hybridisation assay: 1. Mix cluster-labelled DNA (probe) 1:1 with 2 hybridisation buffer (according to protocol, usually 3 SSC+0.1% SDS). 2. Transfer target slide to hybridisation chamber. 3. Apply 30-120 ml of probe solution (amount depends on cover slips). 4. Hybridisation time and temperature according to choice of probes and target. 5. Wash slide with 2 SSC+0.1% SDS for 15 min (stirring or shaking). 900 800 700
Spotting/dispensing/immobilisation
Probes can be applied to the chip surface via dispensing, spotting or jetting using commercial devices. Additionally, the choice will depend on the surface activation and stability of the probe. Immobilisation of amino- linked DNA on divinylsulphone activated surfaces: 1. Dilute amino-labelled DNA probes in 50 mM NaHCO3/ Na2CO3 pH 9.0 buffer to 20 mM. 2. Transfer probes into a low binding micro-well plate. 3. Spot the probes onto the surface. 4. Transfer slide into a humid chamber and incubate for 1 h at room temperature. 5. Wash slide in dH2O, 0.1% Triton X-100 and again in dH2O (immerse and shake for 2 min). 6. Block slides with 0.1% milk powder (alternatively, protein solution or ethanolamine) for 30 min, then wash IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
scanner signal
600
3.5
500 400 300 200 100 0 260
520
1050
2100
4200
8300 16 700 33 000
DNA concentration, ng/ml
Fig. 8 REA-based capture hybridisation assay: perfect match (bar) compared to a two-base-pair mismatch (line) for a oligonucleotides conduction of 30 bp. The quantification was performed using a reflectivity chip scanner 59
6. Wash slide with 2 SSC for 15 min (stirring or shaking). 7. Wash slide with 0.2 SSC for 15 min (stirring or shaking). 8. Dry slides. 9. Read results. To withstand high salt conditions the clusters need to be stabilised, otherwise precipitation will occur. Instead of binding the DNA clusters directly, avidin clusters can be bound to biotinylated hybridisation partners. This system has the advantage of a cluster independenthybridisation. Tests with single, double and quadraple mismatches have been perfect and the results prove that the cluster-based REA-based assay does not decrease the performance and precision of DNA-DNA hybridisation. Using large metal nano-clusters of 410 nm in diameter a correct geometrical design of the molecular binding sites is required to avoid steric hindrance effects. Thus, a linker of 1–3 nm in length (e.g. poly-T) is required to allow correct probe binding. Unspecific adsorption or precipitation of metal clusters is observed similar to the unspecific background of any biological assay. Whereas precipitation and adsorption of metal clusters is mostly reversible and loosely bound clusters can be removed by washing, cluster coagulation (as an irreversible phenomenon caused by reagent instability and degradation) needs to be avoided to obtain the sensitivity given in Fig. 9. Typical numbers of unspecifically bound clusters are a few nanoclusters per square micrometre. The background can be reduced significantly by proper assay design (no cluster deposition) and surface-charge tuning of the chip and the nano particles avoiding deposition via strong electrostatic attraction.
‘Reverse assays’ work by binding an antibody-labelled cluster solution to a slide with immobilised antigens. These assays are used for antibody screening and medical diagnostics. In an antibody screening assay the antibody of interest is screened with B5000 samples in a one-step protocol. Furthermore, this setup avoids drying of the sensitive antibodies during the dotting procedure. The third type of assays, ‘sandwich assays’, is used if a high selectivity and sensitivity are required. Capture antibodies are immobilised on the surface, during incubation with the unlabelled sample with which the antigen binds, thereby concentrating on the proper capture antibody spots. A labelled detector antibody couples with the cluster to the antigen, creating the colour micro-dot. A sensitivity suitable for most bio-medical applications is obtained (see Fig. 9). REA ‘direct’ or ‘reverse’ assays: 1. Follow ‘Cluster coating with proteins’ protocol to label the sample solution. 2. Follow the protein immobilisation protocol to produce the REA chip using DTT and glycerol to stabilise the proteins (if necessary). 3. Transfer slide to humid chamber. 4. Apply labelled target (depending on system and cover slips 30–120 ml). 5. Apply cover slips or close hybridisation chamber. 6. Incubate slide at room temperature from 15 min up to 10 h in humid chamber (depending on antigen concentration range). 7. Wash slide twice with dH2O (stirring or shaking). 8. Dry slides. 9. Read results.
2
10
scanner signal
(I)
(II)
For ‘sandwich assays’ additional incubation steps using the same conditions are required.
(III)
3.7 Photopolymerisation of nano-well structures
10
1
0
10
−12
10
−11
10
−10
10
−9
10
10
−8
−7
10
−6
10
analyte concentration, mol/L
Fig. 9 A number of assays were performed to characterise the sensitivity of the REA-based capture assay. The quantification was performed using a line scanning direct reflectivity chip scanner at a 2400 dpi resolution (i) The protein – antibody assay (ii) This protein – hapten assay (iii) A low affinity antibody assay (C-reactive protein)
3.6.2 Protein: ‘REA protein assays’ are performed using the same procedures as immuno-assays. Antibodies are arrayed and immobilised on the slide. The cluster labelled sample is applied and the component of interest binds to the antibody, thereby positioning the cluster at the correct REA distance. The result is a coloured dot, indicating that this antibody has detected its antigen in the sample. These direct assays are applied to samples where multiple antigens of interest could be present. The array can report all of them in a one-step procedure. 60
Quite a number of groups focus on the design and production of micro- and nano-well structures. These structures are related to honeycomb matrices, but are not designed for their mechanical stability but as dense arrays of micro-chemical reactors. Whereas honeycomb structures allow us to produce stable matrices for a variety of applications, the nano-wells developed for REA-based assays are small vials made from highly adhesive and sticky polymers that allow us to seal the wells by simply covering with a polymer plastic film. Nano-well polymer films are soft and sticky and need to be supported and stabilised by the chip surface. This additional important feature allows straightforward integration of nano-wells on REA chip arrays. Using the approach described below nano-wells of a typical size of 400 400 1 mm and a volume of between 0.1–0.2 nl are produced via photolithography. This miniaturisation allows the use of expensive chemicals, boosts the speed of the assays and overcomes crucial problems due to the slow diffusion of diluted macromolecules. Quite often an aluminium chip with a 12 mm diameter is used as support. The chip is etched with an oxygen plasma and activated with 3-aminopropyl-triethoxysilane using vacuum evaporation. After following the standard REA chip protocol the nano-wells are coated on top of the REA structure. IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
Cis-polyisoprene from synthetic rubber was used as the material for hydrophobic wells. Native poly-isoprene combines three substantial advantages: (i) It is inert; (ii) It is hydrophobic and thus reduces the carry over between the wells when working in an aqueous environment; and (iii) Poly-isoprene polymers exhibit a sticky surface to anneal a top plate to cover the nano-wells. Cyclohexanone was chosen after extensive solubility tests being the best choice as solvent for both the native polyisoprene and the crosslinking agent. Crosslinking was achieved using reactive bis-nitrenes created from 2,6-bis(4azidobenzylidene)-4-methylcyclohexanone. The viscosity of the spin-coating solution (adjusted via the percentage of poly-isoprene dissolved in cyclohexanone) determines the height of the well structures. Atomic force microscopy (AFM) measurements correlate a height of 1.5 mm with a dilution of 1/3(v/v) poly-isoprene/Cyclohexanone at 4000 rpm. Spincoating was done on a Speciality-Coating-Systems series 6700 spincoater. Photopolymerisation was achieved using an OSRAM VITALUX 300 light source at a wavelength of 350 nm. Other light sources covering the necessary range in the near-UV are also appropriate, including mercury or metal halide lamps. The photo-mask was either created on a PC and printed via a high-resolution laser printer, or using a laser writer or an e-beam mask writer (with increasing resolution). Non-crosslinked poly-isoprene was removed by extensive washing using a series of solvents starting from tolueneacetone (1:1), followed by isopropanol and water. Protocol for creating well structures on an aluminium chip surface (Fig 10): 1. The aluminium chip is etched for 1 min in an oxygen plasma to oxidise traces of contaminants. 2. The aluminium chip is reacted with aminosilane o/n in vacuum and baked afterwards for 30 min at 1101C. 3. Unbound aminosilane is removed via isopropanol or ethanol. 4. Prepare the polymer stock solution by mixing one part of native isoprene and three parts of Cyclohexanone (v/v) and vortexing thoroughly for 2 min. 5. Further dilute this stock with cyclohexanone to the working concentration. 6. Prepare a 1% (w/v) solution of 2,6-bis(4-azidobenzylidene)-4-methylcyclohexanone in the diluted isoprene solution. 7. Vortex thoroughly for 2 min to dissolve the crosslinker. 8. Place the aluminium chip on the spincoater and apply 100 ml of the mixture onto the surface.
9. Start the optimised spin protocol. Coating works best between 3700–4000 rpm. 10. Dry the chip and apply the photo-mask to the chip surface. 11. Expose the chip and allow photo-crosslinking (time depends on the light source and polymer thickness: some seconds to 5 min). 12. Wash the chip with toluene-acetone ¼ 1:1 (v/v) for a few seconds to allow un-polymerised isoprene to dissolve. 13. Wash the chip afterwards for 30 s in isopropanol to remove last traces of un-polymerised isoprene and toluene and finally wash with MQ water.
3.8
Data acquisition
The response of the REA cluster chips is either read out by eye or is detected spectroscopically in the visible or infrared range of the spectrum by a reflectance scanner (resolution depends on spot size, it should be at least 10 times higher than the spot size) or using a CCD camera. The optical data are transferred to a PC and encoded in true or false colour (often used for DNA chips). If desired, even a reverse optical effect (light up upon cluster binding) might be obtained if a combination of a strong thin layer- interference colour and the cluster REA effect is employed (see comparison of identical positive and negative interlayer array biochips in Fig. 11). Compared to ELISA methods the technique has the advantage of cost efficiency and a one-step procedure. Additionally the slides can be rescanned and stored for a considerable amount of time (up to years) without losing signal intensity. For application in high density arrays very high resolution is important: below the standard optical limits (see Fig. 12).
Fig. 11 Protein cluster REA chip tuned to a dark-blue dot colour and a light-blue dot- colour on a dark interference colour. Note: negative dot effect!
500 nm
200 nm
100 nm
Fig. 10
Nano-well array on a REA sensor
a Microscopy b Resonance colour IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
Fig. 12 Lateral-resolution of the cluster resonance chip: note that effects far below the optical limit are visible 61
4
Discussion and Conclusions
The fundamental basis of surface-enhanced capture-cluster chips is the enhancement factor. It describes the change in absorption a cluster or cluster-film experiences when positioned at a defined nanometre-scale distance from a mirror surface. Experiments to quantify this factor have been performed using PS-coated glass- and metal mirror chips under the same conditions. A protein was covalently attached to the surfaces of both chips (EDC chemistry) and a recognitive protein (14 nm gold cluster labelled) was used for detection. The reflectance spectrum for biorecognitively coupled nano-clusters (close to the mirror layer) had a maximum absorption peak with an intensity of emolar ¼ 2.00. The reflectance spectrum of a layer of crystalline spherical gold clusters on the surface of the PS chip has been measured with a perfect aluminium layer at the back with a maximum absorption peak of emolar ¼ 0.15–0.25. The resulting enhancement factor is approximately eight which is nicely in line with calculations in [2]. This value represents a minimum, since the enhancement factor is higher in cases where measurement takes place at wavelengths where the surface-enhanced setup gives an absorption peak at or above 600 nm. Cluster- films only show a little or over no absorption at these wavelengths. Enhancement factors can exceed 100 compared to standard colloidal assays. Cluster synthesis using the citrate reduction techniques detailed in Freus [11] yields mono-dispersed preparations over a broad size range. Other synthetic routes also are useful and there is no direct correllation between the success of surface-enhanced cluster absorption assays and the cluster size. 8 nm clusters yield the same enhanced spectra as 24 nm clusters, given that they both cover the chip surface at the same mass-thickness. This means that the sensitivity of REA-based assays depends on the size of the clusters because the same intensity in signal is achievable with fewer 24 nm clusters than with 8 nm clusters. It has to be taken into account that clusters with a larger diameter show a reduced speed of diffusion. Extended incubation times are therefore necessary for the biological assays. Large cluster particles (30 to 50 nm) offer five to 20-fold higher sensitivities but suffer from a number of limitations including: (i) an increased tendency to aggregate and flocculate; (ii) sedimentation; and (iii) adsorption to surfaces. As a conclusion to these arguments it can be stated that in standard assays without application of novel equipment 10- 30 nm gold clusters present a compromise between sensitivity and acceptable diffusion rates. Together with the introduction of devices that may allow stringent control as well as increased surface concentration and subsequent removal of the unbound probe larger clusters will be applied in order to push the sensitivity limits of this detection technology. Such devices will be the subject of future research. The dynamic range of REA surfaces can be assessed on basis of AFM measurements. This is done by quantifying the unspecific background of the method. An area where all the reactions have taken place except for analyte incubation is taken as to be representative of the unspecific binding in these assays. On an area of 1 mm2 about four to five clusters are adsorbed. The maximum signal, however, is limited with about 5000 clusters in the respective area. This gives a dynamic range of around 103 or B10-bit in terms of electronic scanning device quantification. The dynamic range can be extended if the unspecific background is lowered. This could be done by applying electric or 62
magnetic forces in order to remove unspecifically bound clusters. Smart polymer sensor arrays, protein biochips and DNA microarrays using nano-optical transducers will allow us to monitor and quantify the number, dimension as well as structural changes of nanometric elements, layers and even single molecules. Due to nano-dimensions, these nano-chips exhibit a fast response to analytes and provide simple and cost-effective devices for many analytical applications. Microarrays of proteins and DNA as well as smart polymer-sensors are the first REA-based products with direct optical output compatible with high throughput analysis and optical CCD chips [22, 23] and adds a new group of sensors to the family of enhanced plasmonic devices [24]. 5
Acknowledgment
Part of this work was supported by the LifeTech program at TUDELFT/The Netherlands. 6
References
1 Bauer, G., Stich, N., and Schalkhammer, T.: ‘Nanoclusters and colloids in bioanalysis in MTBM Volume on Analytical Biotechnology’ (Birkh.auser Verlag, Switzerland, 2002) 2 Leitner, A., Zhao, Z., Brunner, H., Aussenegg, F., and Wokaun, A.: ‘Optical properties of a metal island film close to a smooth metal surface’, Appl. Opt., 1993, 32, p. 102 3 Schalkhammer, T.: ‘Metal nano clusters as transducers for bioaffinity interactions’, Chem. Monthly, 1998, 129, p. 1067 4 Bauer, G., Hassmann, J., Walter, H., Haglmueller, J., Mayer, C., and Schalkhammer, T.: ‘Resonant nanocluster technologyFfrom optical coding and high quality security features to biochips’, Nanotechnology, 2003, 14, p. 1289 5 Mayer, C., Stich, N., Palkovits, R., Bauer, G., Pittner, F., and Schalkhammer, T.: ‘High-throughput assays on the chip based on metal nano-cluster resonance transducers’, J. Pharmaceutical Biomed. Anal., 2001, 24, p. 773 6 Mayer, C., Verheijen, R., and Schalkhammer, T.: ‘Food-allergen assays on chip based on metal nano-cluster resonance’, Proc. SPIEInt. Soc. opt. Eng., 2001, 4265, pp. 134–141 7 Mayer, C., Stich, N., Schalkhammer, T., and Bauer, G.: ‘Slide format proteomic biochips based on surface enhanced nanocluster – resonance’, Fresenius J. Anal. Chem., 2001, 371, p. 238 8 Stich, N., Gandhum, A., Matyushin, V., Raats, J., Mayer, C., Alguel, Y., and Schalkhammer, T.: ‘Phage display antibody-based proteomic device using resonance-enhanced detection’, J. Nanosci. Nanotechnol., 2002, 2, (3/4), p. 375 9 Schalkhammer, T., Lobmaier, C., Pittner, F., Aussenegg, F., Leitner, A., and Brunner, H.: ‘A new urea sensor based on a metal island coated ion sensitive swelling polymer device’, Proc. SPIE-Int. Soc. Opt. Eng., 1995, 2508, p. 102 10 Mayer, C., Palkovits, R., Bauer, G., and Schalkhammer, T.: ‘Surface enhance resonance of metal nano cluster: A novel tool for proteomics’, J. Nanoparticle Res., 2001, 3, p. 361 11 Frens, G.: ‘Controlled nucleation for the regulation of the particle size in monodisperse gold suspensions’, Nature (Phys. Sci.), 1973, 241, p. 20 12 Slot, J., and Geuze, H.: ‘A new method of preparing gold probes for multiple-labeling cytochemistry’, Eur. J. Cell. Biol., 1985, 38, p. 87 13 Mandal, S., Selvakannan, P., Phadtare, S., Pasricha, R., and Sastry, M.: ‘Synthesis of stable gold hydrosol by the reduction of chloroaurate ions by the amino acid, aspartic acid’, Proc. Indian Acad. Sci. Chem. Sci., 2002, 114, (5), pp. 513–520 14 Hostetler, M.J.: ‘Alkanethiolate gold cluster molecules with core diameters from 1.5 to 5.2 nm: core and monolayer properties as a function of core size’, Langmuir, 1998, 14, pp. 17–30 15 Chen, S., and Murray, R.W.: ‘Arenethiolate monolayer-protected gold cluster’, Langmuir, 1999, 15, pp. 682–689 16 Jackson, A.M., Myerson, J.W., and Stellacci, F.: ‘Spontaneous assembly of sub-nanometre ordered domains in the ligand shell of monolayer protected nanoparticles’, Nature Mater., 2004, 3, pp. 330–336 17 Lu, Y., Yin, Y., Li, Z., and Xia, Y.: ‘Synthesis and self-assembly of Au@SiO2 core-shell colloids’, Nano Lett., 2002, 2, (7), pp. 785–788 18 Mulvaney, P., Liz-Marz!an, L.M., Giersig, M., and Ung, T.: ‘Silica encapsulation of quantum dots and metal clusters’, J. Mater. Chem., 2000, 10, pp. 1259–1270 19 Elghanian, R., Storhoff, J., Mucic, R., Letsinger, R., and Mirkin, C.: ‘Selective colorimetric detection of polynucleotides based on the distance-dependent optical properties of gold Nanoparticles’, Science, 1997, 277, p. 1078 IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
20 Mirkin, C., Letsinger, R., Mucic, R., and Storhoff, J.: ‘A DNA-based method for rationally assembling nanoparticles into macroscopic materials’, Nature, 1996, 382, p. 607 21 Mirkin, C.: ‘Towards DNA Based Technology for Preparing Nanocluster Circuits and Arrays’, MRS Bull., 2000, 25, p. 43 22 US5611998 ‘Optochemical sensor and method for production’ (1997)
IEE Proc.-Nanobiotechnol. Vol. 152, No. 2, April 2005
23 US6669906 ‘Reinforced cluster optical sensors’ (2003) 24 Zayats, M., Pogovelova, S.P., Kharitonov, A.B., Lioubashevski, O., Katz, E., and Willner, I.: ‘Au-nanoparticle-enhanced surface plasmon resonance sensing of biocatalytic transformations’, Chem. Eur. J., 2003, 9, pp. 6108–6114
63