Francesca
Costantini
*a,
Augusto
Nascetti
b,
Riccardo
Scipinotti
c,
Fabio
Domenici
d,
Simona
Sennato
e,
Laura
Gazza
e,
Federico
Bordi
d,
Norberto
Pogna
e,
Cesare
Manetti
a,
Domenico
Caputo
c and
Giampiero
de Cesare
c
aDipartimento di Chimica, Sapienza Università di Roma, p.le Aldo Moro 5, 00185 Rome, Italy. E-mail: costan.fra@gmail.com
bDipartimento di Ingegneria Astronautica, Elettrica ed Energetica, Sapienza Università di Roma, via Salaria 851/881, 00138 Rome, Italy
cDipartimento di Ingegneria dell'Informazione, Elettronica e Telecomunicazione, Sapienza Università di Roma, via Eudossiana 18, 00184 Rome, Italy
dDipartimento di Fisica, Sapienza Università di Roma, p.le Aldo Moro 5, 00185 Rome, Italy
eConsiglio per la Ricerca e la Sperimentazione in Agricoltura (CRA-QCE), Via Cassia 176, 00191 Rome, Italy
First published on 11th November 2013
In this paper, we present the preliminary results of an ELISA-on-chip device, intended as a technological demonstrator of a novel analytical system suitable for the diagnosis and follow-up of celiac disease. The idea of the work is to combine an array of amorphous silicon photosensors with a pattern of a poly(2-hydroxyethyl methacrylate) polymer brush film, which acts as anchor for the immobilization of gliadin peptides containing the celiac disease epitopes. Recognition relies on a sandwich immunoassay between antibodies against the peptides and secondary antibodies marked with horseradish peroxidase to obtain a chemiluminescent signal. Detection is based on the measurement of photocurrent induced in the array of amorphous silicon photosensors by the chemiluminescent signal. An ad-hoc procedure has been developed in order to enable the fabrication of the photodiode array and the polymer brush pattern on the two sides of the same glass substrate ensuring the compatibility of the different technological steps. The sensitivity and the selectivity of the chip for multiplex immunoassays were demonstrated using two gliadin peptides (VEA and DEC). In particular, we found that the average amount of the bound HRP revealed by our analytical protocol is 3.5(±0.3) × 10−6 pg μm−2 and 0.85(±0.3) × 10−6 pg μm−2 for specific and non-specific interactions, respectively.
The advent of miniaturized array technology has enabled parallel and multiple comparative measurements of different samples combining high sensitivity, low sample consumption, high-throughput and rapidity of the analysis.10,11 In this framework, microarray chips, with tens to thousand of micro-spots of immobilized capture agents (probes) having binding activity against specific analyte molecules (targets), have become a well established research tool in basic and applied science. The capture agents are spatially encoded to form a known pattern, which permits the recognition of the binding events between the probe and the target.
Currently, different kinds of microarray chip are available on the market (i.e. Affymetrix®, Agilent©, Luminex©) for the analysis of DNA, RNA10 and different biomarkers,11,12 and many others are being realized using new types of probe-biomolecules13–16 and/or microorganisms17 and cells,12 for the analysis of several targets such as biological fluids (biomedical applications)16 or extracts of food and beverages (food-safety applications).18
The key issue for the development of devices, based on microarray technology, is the integration of the physical-chemistry techniques for the fabrication of the patterned probes, and the detection system for the biochemical recognition of the targets, to create a simple and low cost solid platform accessible also to not specialized personnel.19
Several materials including glass and quartz, gold, silicon, polymers, filter membranes, optical fibres and beads can be applied to create a microarray of pattern probes,20 which are either physically adsorbed or covalently immobilized by means of different chemical strategies.21
Targets recognition relies on various different detection systems, which have been coupled to microarray platforms. These detection methods are based on either label-free or label-based approaches. The first category includes techniques such as mass spectrometry,15 optical biosensor technology (including reflectometric interference spectroscopy22 and surface plasmon resonance imaging23). Among labeled-based approaches, fluorescence,24–26 chemiluminescence27–30 and radioactivity combined with charge coupled device (CCD) cameras or laser scanners are the most applied for detection.15
Although the advantage of these methods is the high sensitivity, they can be expensive and sometimes require bulky instrumentation,23 and therefore technology has been focused in the development of sensors, which can be integrated within microarray chip. Examples are microelectrical sensors based on the use of silica nanowires, impedimetric, surface acoustic waves, magnetic nanoparticles and microantenna technologies.31 However, these systems showed limitations associated with sensor fabrication and sensitivity, which need to be resolved. An alternative approach consists in using photosensors in the microarray platform.26,32–34 This solution allows the use of conventional high-sensitivity optical detection techniques and, at the same time, it does not require major changes in the platform both in terms of substrate material and microarray layout. In particular, this is true if thin-film large-area electronic technologies as organic electronics (also referred to as plastic electronics) or amorphous silicon technology are used.35–38 In the last decade, several examples have been presented proving the feasibility of analytical systems based on both the above-mentioned technologies. These examples include both labelled and label-free techniques using different analytical methods such as stimulated fluorescence,39–42 chemiluminescence43–45 and optical absorption.37,38,46 In particular, chemiluminescence appears to be very attractive since it does not require external radiation sources as in fluorescence case and does not suffer of background signals.
In this work, for the first time, we report on the development of an ELISA-on-chip device for the simultaneous detection of multiple CD antibodies against GPs. In this system, recognition, detection and read out elements are all performed in a single glass substrate without external, bulky and expensive equipment. This aim has been achieved by combining and optimizing surface chemistry with microelectronic processes on the two opposite sides of a glass substrate.
The device is constituted by an array of a-Si:H photosensors aligned with a pattern of poly(2-hydroxyethyl methacrylate) polymer brushes (PHEMA)47–49 which act as anchors for the immobilization of GPs (probes). The test relies on a sandwich immunoassay between antibodies against these GPs (target) and a secondary antibody marked with horseradish peroxidase (HRP) which is used to obtain a chemiluminescent signal detected by the a-Si:H photosensors. We have fabricated and characterized a technological demonstrator in order to test the sensitivity and the specificity of the device for multiplex immunoassay analysis using two different GPs named VEA and DEC.
R = Iph,sens/Pinc,sens | (1) |
- four masks for the fabrication of the sensor array on one face of the glass substrate;
- one mask to define PHEMA pattern on the opposite face of the same glass chip.
(1) deposition and pattering by photolithography of the TCO window layer for the definition of the front electrode of the photodiodes (mask 1);
(2) deposition by PECVD of the a-Si:H layers;
(3) deposition by magnetron sputtering of a stack of three metal layers (Cr/Al/Cr), which acts as back electrode of the sensors;
(4) mesa patterning of the device structure by wet and reactive ion etching for the metal stack and a-Si:H layers respectively (mask 2);
(5) deposition of a 5 μm thick SU-8 layer acting as insulation layer between the back metal and the front TCO contacts;
(6) opening of via holes over the diodes on the passivation layer (mask 3);
(7) deposition by magnetron sputtering of a TiW metal layer for the definition of external connection of the photodiodes;
(8) patterning of the TiW external contacts (mask 4).
The area of each photodiode is 2 × 2 mm2.
The PHEMA dots on standard glass slides were obtained using sacrificial SU-8 layer patterned by photolithography (mask 5). In particular, the following procedure has been implemented to define the dots:
(1) rinsing with piranha and treatment with oxygen plasma. Piranha (H2SO4–H2O2 3:
1) was performed for 10 min and rinsed out with MilliQ water. Oxygen plasma was performed for 90s employing reactive ion etching apparatus using the following parameters: oxygen flow 100 sccm, power density 200 mW cm−2, pressure 800 mTorr.
(2) spin-coating of the SU-8 2005 and exposure to UV light using mask 5, which has the dot pattern aligned with the array of photosensors. The glass slides were sonicated for 1 min in a solution of SU-8 developer, rinsed with isopropanol and dried with a stream of nitrogen.
(3) soaking in a solution of 0.2% of BMPTS in dry toluene over night (room temperature), rinsing with dry toluene and drying with a stream of nitrogen.
(4) removing of the SU-8 layer immersing the glass slides for a few minutes in a solution of SU-8 remover, afterwards they were rinsed with methanol and dried with the stream of nitrogen.
In order to prepare the PHEMA brush film, a solution of 20 mL 2-hydroxyethyl methacrylate (HEMA) and 20 mL water was degassed by bubbling through dry nitrogen (N2) for 30 min and transferred in a schlenk tube where it was stored under argon. Copper(I) chloride (0.110 g), copper(II) bromide (0.072 g) and 2.2′-dipyridyl (0.488 g) were added. To dissolve all the solid, the mixture was stirred for 10 min (while degassing), which yielded a dark brown solution.
The solution was then sonicated until complete dissolution of the solid and subsequently transferred with a cannula in the schlenk tube containing the glass substrates. After the polymerization (over night, in the dark), the samples were removed and washed with methanol and MilliQ water.
The PHEMA polymer films were treated (24 h, 25 °C) with a solution of succinic anhydride (100 mg) and triethylamine (100 μL) in 2 mL of dry tetrahydrofuran (THF). Subsequently, they were rinsed with THF and Milli-Q water and dried with a stream of nitrogen. 1 mL solution of n-hydroxysuccinimide (NHS) (13 mg) and of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) (75 mg) was poured on the films and left to react for 1 h. The films were rinsed with water and dried with a stream of nitrogen. Subsequently, a phosphate buffer solution of peptide VEA or DEC at pH = 7.5 25 mM were poured on the substrates and left to react for 3 h. The substrates were rinsed with a phosphate buffer solution for 5 min (3 times) and with MilliQ water and dried with a stream of nitrogen. The substrates were treated with a solution of 10 mM ethanolamine in phosphate buffer pH = 7.5, 25 mM for 20 min in order to block the unreacted NHS-ester groups. The substrates were abundantly rinsed with MilliQ water and dried with a stream of nitrogen.
The array of a-Si:H sensors chip was functionalized following the same procedure reported for the glass slides. The glass surface of the chip was cleaned exclusively with oxygen plasma since piranha solution might affect the functionality of the array of a-Si:H sensors. During the functionalization of the chip, a custom-made holder was employed in order to avoid any contact between THF and the opposite side of the glass hosting the photosensors.
The procedure implemented for the device fabrication has been optimized to keep the functional compatibility of the different technological steps. For this purpose, the photosensors have been deposited and patterned before the peptide immobilization in order to avoid instability effects, due to the PECVD temperature (around 200 °C).
The chip was inserted in a card edge connector that individually connects the photosensors to the bias and read-out equipment. In particular, the current of three different sensors of the array has been simultaneously measured. The experimental protocol has been designed scheduling a set of experiments each involving three reaction sites at a time.
The chip was enclosed in a light-shielded box and, for each experiment, the photodiode dark current was acquired for 2 min and the average dark current value was calculated.
In each experiment, a drop of chemiluminescent reagent (2 μL) was spotted on the selected functionalized PHEMA site. In this case, the box was opened and the photodiodes were exposed to dimmed ambient light. Thereafter, the light-shielded box was closed again and the current acquisition started after 30 s in order to get rid of the a-Si:H photodiode current transient due to charge detrapping after the exposure to ambient light.51 The acquisition lasted 10 min. The CL-induced photocurrent is calculated by subtracting the average dark current value from the current measured during the chemiluminescent reactions.
As a first step, a 4 × 4 photosensor array has been fabricated on a 5 × 5 cm2 ultrasonically cleaned glass substrate following the technological steps described in the experimental section (Fig. 1).
In the fabricated array reported in Fig. 1, the squares are the photosensors and the broad external line their common back contact. The front contacts of the photodiodes are not visible because they are transparent.52 The area of the sensor (2 × 2 mm2) has been chosen to maximize the collection of the chemiluminescent signal resulting from the immunoenzimatic reaction. At the same time, as we have found from the current–voltage characteristics, this size guarantees a dark current level (4 × 10−13 A at small reverse voltage) low enough to keep noise contribution of the dark current well below the minimum detectable signal in our experimental set-up. The responsivity value at 465 nm (1.46 μW cm−2 intensity) has been found to be 247 mA W−1, which is comparable to that of state-of-the-art crystalline silicon photodiodes.
The reproducibility of these values has been verified comparing the performances of photodiodes fabricated both in the same and in different runs. Averaged values of dark current and responsivity showed a deviation below 5%.43 Furthermore, the degradation of the optoelectronic properties of the photodiode due to illumination is negligible, because the intensity radiation is very low in all the experiments performed.38
Recently, polymer films have been applied for many applications in the field of biosensing.53 In particular, the use of polymer brushes for the immobilization of biomolecules49,54 to develop microarrays55 and biomedical devices47 has been described. The main advantages of these brush films lay on the control over the amount of the immobilized biomolecules by varying the polymerization time,56,57 the preserved biological activity after the coupling reaction47,58 and the reduction of nonspecific interactions.49 In this work, a pattern of PHEMA brushes has been applied for the immobilization of GPs (target) onto the surface of the glass chip in order to create sixteen transducing sites aligned with the a-Si:H photosensors. Each site is used for the recognition of GPs antibodies: the photosensor positioned below each site allows the detection of the chemiluminescent signal, which indicates the presence of the antibodies against a specific GP present in the serum.
Sacrificial layer photolithography with SU-8 2005 photoresist was used to define the formation of the PHEMA sites on the glass chip. Being resistant to toluene, this type of photoresist permits the formation of the BMPTS self assembled monolayer (SAM) in solution (Fig. 2a) and can be easily removed after the monolayer formation.
The fabrication of the PHEMA brush pattern and its functionalization was accomplished by applying a previously published procedure.47 PHEMA sites were formed by atom transfer radical polymerization (ATRP), BMPTS acting as initiator layer (Fig. 2b). Afterwards, the PHEMA sites were treated with succinic anhydride (SA) and n-hydroxysuccinimide (NHS) in presence of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) to chemically modify the hydroxyl groups along the PHEMA brushes to carboxylic (PHEMA-SA) and NHS-ester (PHEMA-SA-NHS) functions, respectively (Fig. 2c–d).
A phosphate buffer solution (pH = 7.5, 25 mM) of peptides was poured on the functionalized PHEMA pattern and left to react for 3 h. The peptides are immobilized in the PHEMA brushes through the formation of amide bonds between the amino groups of the amino acid of the peptide and the NHS-ester groups of the functionalized PHEMA layer (Fig. 2e). After rinsing, the functionalized PHEMA sites were treated with a solution of ethanolamine in phosphate buffer, in order to block the unreacted NHS-ester groups, which might interact with other species present in the serum.
Initially, the formation and functionalization of PHEMA was performed on standard glass and silicon oxide substrates in order to study the formation and the functionalization of the brush layer by atomic force spectroscopy (AFM), Fourier transform infrared spectroscopy (FTIR) and field emission scanning electron microscopy (FESEM) (see ESI). AFM analysis showed that PHEMA film has a thickness of 190 nm after overnight polymerization time.
Following the immobilization of VEA peptide (PHEMA-VEA), the thickness of the layer increases up to 496 nm. AFM analysis performed after incubation of PHEMA-VEA with rabbit serum containing anti-VEA and secondary antibody marked with HRP (Ig-HRP) exhibited a thickness of 626 and 646 nm, respectively. The progressive increase of layer thickness suggests the successful immobilization of both peptides and antibodies.
Additional evidence was achieved by FTIR spectroscopy, which shows the characteristic absorption bands upon each step of the chemical modification of the layers (see ESI†).
The FESEM analysis of the substrate confirmed the formation of PHEMA pattern sites by photolithography and images are given in the ESI.†
After applying the procedure for the formation and functionalization of the PHEMA pattern onto the chip surface, the optoelectronic characteristics of the photosensors (dark current and responsivity) were verified by measuring the current–voltage curves in dark condition and under illumination with a light source at 465 nm wavelength. The results showed that the photosensors were not affected by the optimized chemical procedure.
(i) six sites were functionalized with VEA peptide. Afterwards, three VEA sites were incubated with the serum of rabbit containing anti-VEA (Ig-VEA), while the three other sites were left without any further functionalization (VEA-only);
(ii) six sites were functionalized with DEC peptide. Afterwards, three DEC sites were incubated with the serum of rabbit containing anti-DEC (Ig-DEC), while the other three sites were incubated with the rabbit serum containing anti-VEA;
(iii) four sites were not functionalized with the peptides. In this case, the NHS-ester functions along the brush film were reacted only with ethanolamine (PHEMA-only) to avoid any interaction with other species present in the serum.
After the incubation with the serum containing the primary antibodies, the chip was copiously rinsed with a solution of phosphate buffer. Subsequently, a solution of the secondary antibody anti-rabbit marked with HRP (Ig-HRP) was poured on all chip sites and incubated for 15 min. The chip was then inserted in the card-edge connector and connected to the read-out electronics for testing the immunoenzymatic reaction.
This experimental design should establish whether the present chip device is able to detect the binding of the GPs with its specific antibody. In fact, the primary antibody against either VEA or DEC selectively binds the specific peptides, while the antirabbit-HRP antibody (Ig-HRP) binds to all the primary antibodies previously linked to their specific peptide, forming a sandwich-like structure (Fig. 3a–b). In order to verify that the binding events have occurred, a solution of the chemiluminescent reagents was poured on the PHEMA functionalized sites. HRP catalyzes the reaction between luminol and hydrogen peroxide yielding a chemiluminescent signal, which is detected as photocurrent by the photosensors positioned underneath.
As a drop of the chemiluminescent (CL) cocktail is spotted on a PHEMA-GP sites, previously incubated with the specific primary antibody and Ig-HRP solutions, the photocurrent signal increases reaching a maximum signal of ca. 20 pA after 10 minutes. Afterwards, the signal drops to zero showing the end of the chemiluminescent reaction (solid squares in Fig. 4).
![]() | ||
Fig. 4 Plot of photocurrent signal versus time for each type of functionalized PHEMA site after spotting the chemiluminescent cocktail. |
On the other hand, when CL cocktail is placed on a PHEMA-GP site incubated with the nonspecific primary antibody and Ig-HRP (Fig. 3c), the photocurrent signal increase only to ca. 5 pA (open squares on Fig. 4). The lowest photocurrent values are observed on those sites not incubated with the primary antibodies (i.e. VEA-only and PHEMA-only). These results demonstrate that the photosensors selectively distinguish the type of biochemical reaction occurring on each type of functionalized PHEMA sites.
In addition, the trend of the photocurrent reported in Fig. 4 shows the behaviour of the chemiluminescent reaction during the time: (i) the increase of the photocurrent corresponds to the formation of the light upon reaction between luminol and hydrogen peroxide in the presence of HRP, reaching a maximum signal which depends on the concentration of HRP, (ii) the decrease of photocurrent is ascribed to the consumption of the chemiluminescent substrate during the reaction (iii) as the CL signal extinguishes the initial dark condition is recovered. In order to prove the reproducibility of the device response to the immunoenzymatic reaction, the same experiment was also performed on multiple sites, which were functionalized using the same chemical procedure. The average value of the photocurrent signals, for each type, is reported in Fig. 5. As expected, in all the cases, the highest photocurrent intensities occurred in correspondence of the PHEMA functionalized sites where primary antibodies (anti-VEA or anti-DEC) were incubated with the specific peptides (positive control).
On the other hand, when primary antibodies were reacted with the nonspecific peptide, photocurrent was four times smaller (negative control). In addition, the small increase of photocurrent observed on the sites where, neither the primary antibodies nor antirabbit-HRP were incubated (VEA only and PHEMA only), was attributed to luminol oxidation occurring in the absence of HRP and is considered as the blank signal. This behaviour suggest that the photocurrent signal recorded during the negative control experiment is the sum of the blank signal plus possible nonspecific adsorption of the antibodies to the polymer film.
This experiment proves that this device permits to recognize the presence of the specific antibodies against GPs in the serum and to distinguish between specific and nonspecific immune interactions.
The PHEMA functionalized brush film did not detach when brought into contact with the serum solution. Moreover, both immunoenzymatic reaction and detection of CL signal by the sensors were not affected by the PHEMA functionalized brush film.
A calibration curve, in which sensor photocurrent is measured as function of HRP concentration, was performed in a precedent published work,43 where the same chip device was also used. According to these data the sensitivity of this detection method is 1.46 pA pg−1. By combining the value of the sensitivity with the photocurrent signal obtained for the specific and not specific binding, the amount of the bound HRP results to be 3.5(±0.3) × 10−6 pg μm−2 and 0.85(±0.3) × 10−6 pg μm−2, respectively. These values are within the calibration curve.
The ratio between the number of secondary and primary antibody for each binding event is not known; therefore it is not possible to calculate the concentration of the primary antibody linked to the peptides immobilized to the brush film. However, to the clinical diagnostic of this autoimmune condition the most important aspect is the qualitative detection of the serological markers (presence or not of the antibodies against GPs). For this reason the results obtained with the rabbit serum are a proof of principle to demonstrate that our device is capable to distinguish the biochemical recognition of an antibody to its CD epitopes. Further research is focused on the detection of antibodies against GPs in human serum.
The standard deviation of the blank signal was used to calculate the limit of detection (LOD) and limit of quantification (LOQ) of this chip-based detection method, which resulted to be 200 pg mL−1 and 665 pg mL−1, respectively. The LOD is comparable with the best detection limits of the conventional ELISA kit and with those obtained by novel bioaffinity-based methods.59–61 In addition, the immunoenzymatic reaction on the device exhibited the same behaviour as expected by using the standard ELISA method, with the advantages of the on-chip detection, lower sample consumption for multiple analyses.
The present immobilization procedure can easily be extended to other biomolecules, opening a route towards a novel generation of smart ELISA kit assay for bio-analytical applications. In particular, the integration of the photosensor array chip with a dedicated microfluidic network62 will lead to easier sample handling, automation and higher reproducibility of the technique.
Footnote |
† Electronic supplementary information (ESI) available: Details of equipment, AFM and FESEM imaging and FTIR analysis. See DOI: 10.1039/c3ra46058d |
This journal is © The Royal Society of Chemistry 2014 |