Author’s Accepted Manuscript Multi-spot, label-free Reflectionless glass
immunoassay
on
Matteo Salina, Fabio Giavazzi, Roberta Lanfranco, Erica Ceccarello, Laura Sola, Marcella Chiari, Bice Chini, Roberto Cerbino, Tommaso Bellini, Marco Buscaglia www.elsevier.com/locate/bios
PII: DOI: Reference:
S0956-5663(15)30230-X http://dx.doi.org/10.1016/j.bios.2015.06.064 BIOS7801
To appear in: Biosensors and Bioelectronic Received date: 17 April 2015 Revised date: 19 June 2015 Accepted date: 25 June 2015 Cite this article as: Matteo Salina, Fabio Giavazzi, Roberta Lanfranco, Erica Ceccarello, Laura Sola, Marcella Chiari, Bice Chini, Roberto Cerbino, Tommaso Bellini and Marco Buscaglia, Multi-spot, label-free immunoassay on Reflectionless glass, Biosensors and Bioelectronic, http://dx.doi.org/10.1016/j.bios.2015.06.064 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Multi-Spot, Label-Free Immunoassay on Reflectionless Glass Matteo Salinaa,*, Fabio Giavazzib,*, Roberta Lanfrancob, Erica Ceccarellob, Laura Solac, Marcella Chiaric, Bice Chinid, Roberto Cerbinob, Tommaso Bellinib,1, Marco Buscagliab,1 a
Proxentia S.r.l., 20135 Milano, Italy;
b
Department of Medical Biotechnology and Translational Medicine, Università degli Studi di Milano,
20090 Segrate, Italy; c
Istituto di Chimica del Riconoscimento Molecolare – C.N.R., 20131 Milano, Italy;
d
Istituto di Neuroscienze – C.N.R., 20129 Milano, Italy.
* These authors contributed equally to the study 1
Corresponding authors: Marco Buscaglia, via F.lli Cervi 93, 20090 Segrate MI, Italy, tel: (+39)
02503 30352, fax: (+39) 02503 30365, e-mail: [email protected]; Tommaso Bellini, Via F.lli Cervi 93, 20090 Segrate MI, Italy, tel: (+39) 02503 30353, fax: (+39) 02503 30365, e-mail: [email protected] Erica Ceccarello’s present address is Institute of Molecular and Cell Biology, A*STAR, Singapore
Abstract Biosensing platforms that combine high sensitivity, operational simplicity and affordable costs find wide application in many fields, including human diagnostics, food and environmental monitoring. In this work, we introduce a label-free biosensing chip made of glass with a single anti-reflective layer of SiO2. This common and economic material coated by a multi-functional copolymer based on dimethylacrylamide enables the detection even in turbid media. The copolymer coating provides covalent immobilization of antibodies onto the surface and prevents the non-specific adsorption of analytes and matrix constituents. The specific capture of target compounds yields a local increase of surface reflectivity measured by a simple imaging system. Chip design and quantitative interpretation of the data are based on a theoretical optical model. This approach enables the multiplex detection of biomolecular interactions with state-of-the-art sensitivity and minimal instrumental complexity. The detection performance is demonstrated by characterizing the interaction between human growth hormone in solution and the corresponding antibodies immobilized on the sensing surface, both in buffer and human serum, obtaining a clear signal for concentrations as small as 2.8 ng/ml.
Keywords: optical biosensors, protein microarrays, reflective phantom interface, point of care, human growth hormone
1. Introduction Despite the continuous technical innovations in the field of biochemical sensing (Baldini et al., 2006)(Rich and Myszka, 2011)(Gubala et al., 2012)(Turner, 2013), the fast, accurate and simple detection of different molecular targets in complex fluid samples, such as biomarkers in blood (Tothill, 2009), pathogens or allergens in food (McGrath et al., 2013) and contaminants in environmental water (Wang, 2012), still represents a very challenging task. Although many promising design solutions have been proposed, as well as different detection methods (Rissin et al., 2010)(Gauglitz, 2014), none of the available techniques meet all the requirements of robustness, operational simplicity, rapidity of the measurement and low cost of instrument and disposables. Among the most intensely investigated detection technologies are the so-called label-free detection methods that require, at least in principle, less washing steps and additions of reagents than common label-based approaches (Ray et al., 2010). Examples of label-free methods are the Surface Plasmon Resonance (SPR), (Brolo, 2012) the Quartz Crystal Microbalance (Beckera and Cooper, 2011) and a number of interferometric techniques relying on the measurement of the spectral shift of the light reflected by the sensing surface (Brecht et al., 1992)(Abdiche et al., 2008)(Ozkumur et al., 2008). In these methods, the simplicity of operation is counterbalanced by the complexity of the technology, which has so far limited their use in diagnostic applications. Accordingly, there is an unresolved need for new biosensor technologies that combine the advantages of label-free detection with higher robustness and reduced cost of both instrument (preferably handheld) and measurement cells (preferably disposable), while keeping the sensitivity and the specificity at the highest levels. In previous works, we demonstrated that perfluorinated amorphous polymers with refractive index similar to that of water can be exploited to realize simple and sensitive label-free optical biosensors (Ghetta et al., 2005)(Prosperi et al., 2006)(Morasso et al., 2010)(Giavazzi et al., 2013). These materials are almost invisible in aqueous solutions, and provide a solid support for immobilization of specific receptors on their surface. The binding of the corresponding molecular targets yields scattering or reflective signals clearly distinct from the background. In particular, we proposed the Reflective Phantom Interface (RPI) detection method, which relies on the analysis of the light reflected by a flat surface of a perfluorinated
polymer spotted with specific probes (e.g. antibodies). With this approach, we could realize a simple diagnostic device to detect Human Immunodeficiency Virus (HIV) and hepatitis B biomarkers in serum down to a few ng/ml making use of the LED source and camera of a smartphone (Giavazzi et al., 2014). The use of plastic substrates provides a great flexibility in the design of the platform and a limited cost for large-scale production, in principle. While promising, this technology was limited by the use of perfluorinated compounds, the available solid materials with a refractive index low enough to match that of water. Indeed, perfluorinated polymers are highly specialized materials with high cost of production. Moreover, their hydrophobic surface is chemically inert and thus difficult to treat and functionalize. Here we show that similar phantom conditions, meaning a reflectivity of the order of 10-5, can be achieved by means of a much more common material: a glass slide with a coating of silicon dioxide that acts as an anti-reflective layer. The fabrication of dielectric coatings on glass optical components is a widely employed procedure for many applications and antireflective layers are commonly deposited on optical components, although with less stringent requirements than those described here. Moreover, a large number of consolidated approaches is available to treat and functionalize the silicon dioxide surface. In this work, we show that a suitable glass substrate with anti-reflective SiO2 coating and a proper functionalization layer enabled label-free optical biodetection with state-of-the-art sensitivity, even in turbid or absorbing media. The reflectivity of different spots of antibodies immobilized onto the surface was measured by a simple imaging system and converted into the amount of biomolecules captured. This was made possible by a theoretical model accounting for the properties of our chips. A proof of principle of the glass-based RPI method was represented by the quantification of human growth hormone (hGH) in blood. hGH is used as a prescription drug to treat children’s growth disorders as well as various syndromes related to hGH deficiency in adults (Strobl and Thomas, 1994)(Hazem et al., 2012). hGH also plays an important role as an anabolic agent and blood tests are routinely conducted for antidoping monitoring (Saugy et al., 2006). The natural concentration of hGH in blood depends on several factors, including age and physical activity, and typically it fluctuates during the 24 hours. Concentrations from a few ng/ml to greater than 20 ng/ml are commonly observed (Nindl et al., 2001). In this context, a sensitive method for rapid quantification of hGH in blood will be highly beneficial for monitoring hGH-related diseases and treatments and in the fight against dopants in sport. The label-free detection of hGH in blood serum was reported in different studies, in particular, based on SPR (Gutiérrez-Gallego et al., 2011)(Kausaite
Minkstimiene et al., 2013). Limits of detection as low as 6 ng/ml were achieved using a competitive format (Treviño et al., 2009). Our results showed that, with the simplicity of the approach, we achieved a higher sensitivity both in human serum and in buffer, where we detected a concentration of hGH as low as 2.8 ng/ml.
2. Materials and Methods 2.1
Substrate preparation
Wedge-like glass chips (F2 optical glass, Schott) with 5° angle, with maximum thickness of 2 mm and a size of 8 x 12 mm, were coated with SiO2 to form an anti-reflection layer of about 79 nm (B chips) or 101 nm (Y chips). These coatings provided a minimum of reflectivity in the blue or in the yellow spectral region, respectively, in condition of normal incidence. After plasma cleaning, the chips were dip-coated with a copolymer of dimethylacrylamide (DMA), N-acryloyloxysuccinimide (NAS), and 3-(trimethoxysilyl) propyl methacrylate (MAPS)— copoly(DMA-NAS-MAPS) (Cretich et al., 2004). On the copolymer, antibodies and control proteins were covalently immobilized in spots with 150-200 ȝm diameter by means of an automated noncontact dispensing system (sciFLEXARRAYER S5; Scienion AG). The spotted molecules were Bovine serum Albumin (BSA, Sigma Aldrich) and various antibodies targeting the human growth hormone (hGH-Ab, HO.p.e.), human hepatitis B surface antigen (HBs-Ab, Dia.Pro Diagnostic Bioprobes), HIV p24 capsid protein (p24-Ab, Dia.Pro Diagnostic Bioprobes), beta-lactoglobulin (bL-Ab, Jackson Immuno Research Laboratories). The unreacted N-hydroxysuccinimide residues were blocked with a solution of 5% ethanolamine in Tris-HCl buffer solution and washed with PBS Buffer.
2.2
Description of the experiments
The sample cartridge was assembled by gluing the prism on the inner wall of a 1-cm plastic cuvette containing a magnetic stir bar. The cuvette was filled with about 0.8 ml of buffer solution consisting of 0.15 M NaCl, 0.02% Tween 20, 1% BSA, and 0.05 M Tris·HCl, pH 7.6 (Sigma Aldrich). The experiments were performed by adding different dilutions of hGH (HO.p.e.). The hGH used in this study was certified as a WHO International Standard, code number 98/574 (National Institute for Biological Standards and Control, Potters Bar, Hertfordshire, UK). The sample spikes of hGH were performed by adding 50 ȝl of serum and buffer solution containing a variable concentration of target molecules. Either FBS (Euroclone) or human serum (HO.p.e.) were used in different experiments. The spotted
surface of the prism was illuminated by a thermally stabilized light-emitting diode (LED), and the reflected intensity was deviated by a beam-splitter and imaged by a CCD camera (Stingray F-145B/C; Allied Technology). The LED light, centred either at 450 nm (for the B chips) or 595 nm (for the Y chips) was expanded and collimated in order to provide an incident beam with controlled divergence. The collection optics allowed the formation of the image of the RPI surface on the CCD sensor through a spatial filtering. The cuvette temperature was kept at 25° C by a custom-made thermalized holder. The experiments were performed by acquiring sequences of images with a constant frame rate in the described reflecting geometry. In this condition, the grey level u(t) of each pixel in each image was proportional to the local relative reflectivity R(t) of the corresponding region of the sensing surface at time t. This allowed to monitor local reflectivity variations as a function of time as more molecules adhered to different regions of the surface.
3. Results and discussion 3.1 Sensing substrate A schematic representation of the structure of the sensing surface is shown in Figure 1. We consider a thin film of refractive index n1 and thickness h1 that separates a glass substrate from an aqueous solution of refractive indices n0 and nS, respectively. The reflectivity of this interface can be written as (Pedrotti et al., 2006):
(1)
where for normal incidence r01 = (n0 – n1)/(n0 + n1), r1S = (n1 – nS)/(n1 + nS) and α = kn1h1, being k = 2π/λ and λ the wavelength of the monochromatic illuminating light. When n12 = n0nS and n1h1 = λ/4 the thin film acts as an ideal anti-reflection coating, leading to R = 0. In principle, this condition can be reached by selecting a glass substrate and a dielectric coating with suitable refractive indices. A common material for high precision optical coating is represented by evaporated silicon dioxide (SiO2), which provides a refractive index n1 ≈ 1.473 at 595 nm. Selecting a glass substrate with n0 ≈ 1.620, we obtained a deep minimum of reflectivity for aqueous samples having ns ≈ 1.340 with a single dielectric layer with thickness h1 of about 101 nm (Y chips). In this condition, the adhesion on the surface of tiny amount of material (e.g. biomolecules) with refractive index n2, different from ns, induces a relative increase of reflectivity that can be easily measured and used to estimate the amount of molecules adhered on the surface, even if forming sub-nanometer layers.
Here we exploited this principle by immobilizing antibodies on the SiO2 surface and measuring the real-time binding of target compounds in solution. The immobilization on the surface was achieved by means of a layer of a few nanometers of copoly(DMA-NASMAPS). The copolymer has multiple functions: (i) it adheres to the surface, (ii) it provides reactive groups for covalent antibody immobilization, and (iii) it limits the non-specific adsorption in complex media (Cretich et al., 2004). In aqueous solution, the copolymer forms a highly hydrated layer with a thickness of about 10 nm, hence yielding to a small increase of reflectivity relative to the bare SiO2 surface (Yalçin et al., 2009).
3.2 Optical model Although the overall reflectivity of a multi-layered interface is fully accounted by recursively applying Equation 1 (Pedrotti et al., 2006), a more practical model was needed to evaluate the sensitivity and to enable a rational design of the sensing substrate. Accordingly, we considered a multi-layered structure, where N layers with refractive indices n1, …, nN, and thickness h1, …, hN, respectively, are sandwiched between two semi-infinite media with refractive indices n0 (glass substrate) and nS (solution). By assuming that the first layer with refractive index n1 has a thickness h1 comparable with the wavelength of the incident beam, while all the others are much thinner, the total reflectivity of such surface can be accounted for by the addition of an extra contribution to the phase retardation of the first layer. Namely, the reflectivity can be computed using Equation 1 with α = kn1h, where the effective thickness h is
(2)
Here ck = (nk2 – ns2)/(n12 – ns2) are factors that weight the contribution to h of each thin layer, according to the corresponding refractive index. We note that ck < (=, >) 1 if nk < (=, >) n1. Importantly, the effect of changing the thickness and/or the refractive index of the layers can be accounted for by adjusting a single parameter, the effective thickness h of the equivalent single layer structure. Equation 1 can be further simplified by considering small variations of the refractive indices and of the thicknesses relative to the condition of vanishing reflectivity (R = 0), as described in the Supporting Information. In these simplifying conditions that are nevertheless adequate, we can also extend Equation 1 to include a moderate polychromaticity of the illuminating light. Indeed, by assuming an average wavenumber ¢k² and a variance ∆k2 = ¢k2² - ¢k²2, the effective reflectivity is simply obtained by averaging over the spectrum. Accordingly, by
defining r = (r1S + r01)/2, β = (r1S – r01)/r and ε2 = (2hn1¢k² - π)2 + (2hn1 ∆k)2, the reflectivity R can be rewritten to the lowest order in β and ε as:
!
(3)
Equation 3 ultimately expresses R as a function of h and n1. We call h0 the value of h yielding the minimum value of reflectivity R0. From Equation 3, we obtain h0 = π/(2n1¢k²). Figure 2a reports the reflectivity as a function of h calculated with Equation 3 (red curve) in comparison with the exact Fresnel formula of Equation 1 (blue curve). The yellow LED illumination was considered, therefore h0 = 101 nm. For a thickness h < h0 + 20 nm, the difference between the estimates obtained with the two approaches is lower than 3%. Equation 3 allows to evaluate the factors that determine the reflectivity for h ≈ h0. Two additive contributions are present: the non-optimal agreement of the refractive indices through the term β and the spectral width ∆k of the illumination source through ε2. In principle, the working conditions can be adjusted in order to have the term ε2 always larger than β2. Accordingly, a limited polychromaticity of the illumination light can provide a reduction of the dependence of the reflectivity on the uncontrolled variations of refractive indices, possibly due to temperature fluctuations or variability of solution composition. For instance, in our working conditions, a yellow LED source with a spectral width FWHM of about 20 nm can provide ε2 > β2, while maintaining the reflectivity of the order of 10-5. An insightful equation describing the response and the sensitivity of our RPI approach can be obtained from Equation 3 when the width of the illuminating spectrum is narrow relative to the average wavelength, i.e. ∆k2
immunoassay
on
Matteo Salina, Fabio Giavazzi, Roberta Lanfranco, Erica Ceccarello, Laura Sola, Marcella Chiari, Bice Chini, Roberto Cerbino, Tommaso Bellini, Marco Buscaglia www.elsevier.com/locate/bios
PII: DOI: Reference:
S0956-5663(15)30230-X http://dx.doi.org/10.1016/j.bios.2015.06.064 BIOS7801
To appear in: Biosensors and Bioelectronic Received date: 17 April 2015 Revised date: 19 June 2015 Accepted date: 25 June 2015 Cite this article as: Matteo Salina, Fabio Giavazzi, Roberta Lanfranco, Erica Ceccarello, Laura Sola, Marcella Chiari, Bice Chini, Roberto Cerbino, Tommaso Bellini and Marco Buscaglia, Multi-spot, label-free immunoassay on Reflectionless glass, Biosensors and Bioelectronic, http://dx.doi.org/10.1016/j.bios.2015.06.064 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Multi-Spot, Label-Free Immunoassay on Reflectionless Glass Matteo Salinaa,*, Fabio Giavazzib,*, Roberta Lanfrancob, Erica Ceccarellob, Laura Solac, Marcella Chiaric, Bice Chinid, Roberto Cerbinob, Tommaso Bellinib,1, Marco Buscagliab,1 a
Proxentia S.r.l., 20135 Milano, Italy;
b
Department of Medical Biotechnology and Translational Medicine, Università degli Studi di Milano,
20090 Segrate, Italy; c
Istituto di Chimica del Riconoscimento Molecolare – C.N.R., 20131 Milano, Italy;
d
Istituto di Neuroscienze – C.N.R., 20129 Milano, Italy.
* These authors contributed equally to the study 1
Corresponding authors: Marco Buscaglia, via F.lli Cervi 93, 20090 Segrate MI, Italy, tel: (+39)
02503 30352, fax: (+39) 02503 30365, e-mail: [email protected]; Tommaso Bellini, Via F.lli Cervi 93, 20090 Segrate MI, Italy, tel: (+39) 02503 30353, fax: (+39) 02503 30365, e-mail: [email protected] Erica Ceccarello’s present address is Institute of Molecular and Cell Biology, A*STAR, Singapore
Abstract Biosensing platforms that combine high sensitivity, operational simplicity and affordable costs find wide application in many fields, including human diagnostics, food and environmental monitoring. In this work, we introduce a label-free biosensing chip made of glass with a single anti-reflective layer of SiO2. This common and economic material coated by a multi-functional copolymer based on dimethylacrylamide enables the detection even in turbid media. The copolymer coating provides covalent immobilization of antibodies onto the surface and prevents the non-specific adsorption of analytes and matrix constituents. The specific capture of target compounds yields a local increase of surface reflectivity measured by a simple imaging system. Chip design and quantitative interpretation of the data are based on a theoretical optical model. This approach enables the multiplex detection of biomolecular interactions with state-of-the-art sensitivity and minimal instrumental complexity. The detection performance is demonstrated by characterizing the interaction between human growth hormone in solution and the corresponding antibodies immobilized on the sensing surface, both in buffer and human serum, obtaining a clear signal for concentrations as small as 2.8 ng/ml.
Keywords: optical biosensors, protein microarrays, reflective phantom interface, point of care, human growth hormone
1. Introduction Despite the continuous technical innovations in the field of biochemical sensing (Baldini et al., 2006)(Rich and Myszka, 2011)(Gubala et al., 2012)(Turner, 2013), the fast, accurate and simple detection of different molecular targets in complex fluid samples, such as biomarkers in blood (Tothill, 2009), pathogens or allergens in food (McGrath et al., 2013) and contaminants in environmental water (Wang, 2012), still represents a very challenging task. Although many promising design solutions have been proposed, as well as different detection methods (Rissin et al., 2010)(Gauglitz, 2014), none of the available techniques meet all the requirements of robustness, operational simplicity, rapidity of the measurement and low cost of instrument and disposables. Among the most intensely investigated detection technologies are the so-called label-free detection methods that require, at least in principle, less washing steps and additions of reagents than common label-based approaches (Ray et al., 2010). Examples of label-free methods are the Surface Plasmon Resonance (SPR), (Brolo, 2012) the Quartz Crystal Microbalance (Beckera and Cooper, 2011) and a number of interferometric techniques relying on the measurement of the spectral shift of the light reflected by the sensing surface (Brecht et al., 1992)(Abdiche et al., 2008)(Ozkumur et al., 2008). In these methods, the simplicity of operation is counterbalanced by the complexity of the technology, which has so far limited their use in diagnostic applications. Accordingly, there is an unresolved need for new biosensor technologies that combine the advantages of label-free detection with higher robustness and reduced cost of both instrument (preferably handheld) and measurement cells (preferably disposable), while keeping the sensitivity and the specificity at the highest levels. In previous works, we demonstrated that perfluorinated amorphous polymers with refractive index similar to that of water can be exploited to realize simple and sensitive label-free optical biosensors (Ghetta et al., 2005)(Prosperi et al., 2006)(Morasso et al., 2010)(Giavazzi et al., 2013). These materials are almost invisible in aqueous solutions, and provide a solid support for immobilization of specific receptors on their surface. The binding of the corresponding molecular targets yields scattering or reflective signals clearly distinct from the background. In particular, we proposed the Reflective Phantom Interface (RPI) detection method, which relies on the analysis of the light reflected by a flat surface of a perfluorinated
polymer spotted with specific probes (e.g. antibodies). With this approach, we could realize a simple diagnostic device to detect Human Immunodeficiency Virus (HIV) and hepatitis B biomarkers in serum down to a few ng/ml making use of the LED source and camera of a smartphone (Giavazzi et al., 2014). The use of plastic substrates provides a great flexibility in the design of the platform and a limited cost for large-scale production, in principle. While promising, this technology was limited by the use of perfluorinated compounds, the available solid materials with a refractive index low enough to match that of water. Indeed, perfluorinated polymers are highly specialized materials with high cost of production. Moreover, their hydrophobic surface is chemically inert and thus difficult to treat and functionalize. Here we show that similar phantom conditions, meaning a reflectivity of the order of 10-5, can be achieved by means of a much more common material: a glass slide with a coating of silicon dioxide that acts as an anti-reflective layer. The fabrication of dielectric coatings on glass optical components is a widely employed procedure for many applications and antireflective layers are commonly deposited on optical components, although with less stringent requirements than those described here. Moreover, a large number of consolidated approaches is available to treat and functionalize the silicon dioxide surface. In this work, we show that a suitable glass substrate with anti-reflective SiO2 coating and a proper functionalization layer enabled label-free optical biodetection with state-of-the-art sensitivity, even in turbid or absorbing media. The reflectivity of different spots of antibodies immobilized onto the surface was measured by a simple imaging system and converted into the amount of biomolecules captured. This was made possible by a theoretical model accounting for the properties of our chips. A proof of principle of the glass-based RPI method was represented by the quantification of human growth hormone (hGH) in blood. hGH is used as a prescription drug to treat children’s growth disorders as well as various syndromes related to hGH deficiency in adults (Strobl and Thomas, 1994)(Hazem et al., 2012). hGH also plays an important role as an anabolic agent and blood tests are routinely conducted for antidoping monitoring (Saugy et al., 2006). The natural concentration of hGH in blood depends on several factors, including age and physical activity, and typically it fluctuates during the 24 hours. Concentrations from a few ng/ml to greater than 20 ng/ml are commonly observed (Nindl et al., 2001). In this context, a sensitive method for rapid quantification of hGH in blood will be highly beneficial for monitoring hGH-related diseases and treatments and in the fight against dopants in sport. The label-free detection of hGH in blood serum was reported in different studies, in particular, based on SPR (Gutiérrez-Gallego et al., 2011)(Kausaite
Minkstimiene et al., 2013). Limits of detection as low as 6 ng/ml were achieved using a competitive format (Treviño et al., 2009). Our results showed that, with the simplicity of the approach, we achieved a higher sensitivity both in human serum and in buffer, where we detected a concentration of hGH as low as 2.8 ng/ml.
2. Materials and Methods 2.1
Substrate preparation
Wedge-like glass chips (F2 optical glass, Schott) with 5° angle, with maximum thickness of 2 mm and a size of 8 x 12 mm, were coated with SiO2 to form an anti-reflection layer of about 79 nm (B chips) or 101 nm (Y chips). These coatings provided a minimum of reflectivity in the blue or in the yellow spectral region, respectively, in condition of normal incidence. After plasma cleaning, the chips were dip-coated with a copolymer of dimethylacrylamide (DMA), N-acryloyloxysuccinimide (NAS), and 3-(trimethoxysilyl) propyl methacrylate (MAPS)— copoly(DMA-NAS-MAPS) (Cretich et al., 2004). On the copolymer, antibodies and control proteins were covalently immobilized in spots with 150-200 ȝm diameter by means of an automated noncontact dispensing system (sciFLEXARRAYER S5; Scienion AG). The spotted molecules were Bovine serum Albumin (BSA, Sigma Aldrich) and various antibodies targeting the human growth hormone (hGH-Ab, HO.p.e.), human hepatitis B surface antigen (HBs-Ab, Dia.Pro Diagnostic Bioprobes), HIV p24 capsid protein (p24-Ab, Dia.Pro Diagnostic Bioprobes), beta-lactoglobulin (bL-Ab, Jackson Immuno Research Laboratories). The unreacted N-hydroxysuccinimide residues were blocked with a solution of 5% ethanolamine in Tris-HCl buffer solution and washed with PBS Buffer.
2.2
Description of the experiments
The sample cartridge was assembled by gluing the prism on the inner wall of a 1-cm plastic cuvette containing a magnetic stir bar. The cuvette was filled with about 0.8 ml of buffer solution consisting of 0.15 M NaCl, 0.02% Tween 20, 1% BSA, and 0.05 M Tris·HCl, pH 7.6 (Sigma Aldrich). The experiments were performed by adding different dilutions of hGH (HO.p.e.). The hGH used in this study was certified as a WHO International Standard, code number 98/574 (National Institute for Biological Standards and Control, Potters Bar, Hertfordshire, UK). The sample spikes of hGH were performed by adding 50 ȝl of serum and buffer solution containing a variable concentration of target molecules. Either FBS (Euroclone) or human serum (HO.p.e.) were used in different experiments. The spotted
surface of the prism was illuminated by a thermally stabilized light-emitting diode (LED), and the reflected intensity was deviated by a beam-splitter and imaged by a CCD camera (Stingray F-145B/C; Allied Technology). The LED light, centred either at 450 nm (for the B chips) or 595 nm (for the Y chips) was expanded and collimated in order to provide an incident beam with controlled divergence. The collection optics allowed the formation of the image of the RPI surface on the CCD sensor through a spatial filtering. The cuvette temperature was kept at 25° C by a custom-made thermalized holder. The experiments were performed by acquiring sequences of images with a constant frame rate in the described reflecting geometry. In this condition, the grey level u(t) of each pixel in each image was proportional to the local relative reflectivity R(t) of the corresponding region of the sensing surface at time t. This allowed to monitor local reflectivity variations as a function of time as more molecules adhered to different regions of the surface.
3. Results and discussion 3.1 Sensing substrate A schematic representation of the structure of the sensing surface is shown in Figure 1. We consider a thin film of refractive index n1 and thickness h1 that separates a glass substrate from an aqueous solution of refractive indices n0 and nS, respectively. The reflectivity of this interface can be written as (Pedrotti et al., 2006):
(1)
where for normal incidence r01 = (n0 – n1)/(n0 + n1), r1S = (n1 – nS)/(n1 + nS) and α = kn1h1, being k = 2π/λ and λ the wavelength of the monochromatic illuminating light. When n12 = n0nS and n1h1 = λ/4 the thin film acts as an ideal anti-reflection coating, leading to R = 0. In principle, this condition can be reached by selecting a glass substrate and a dielectric coating with suitable refractive indices. A common material for high precision optical coating is represented by evaporated silicon dioxide (SiO2), which provides a refractive index n1 ≈ 1.473 at 595 nm. Selecting a glass substrate with n0 ≈ 1.620, we obtained a deep minimum of reflectivity for aqueous samples having ns ≈ 1.340 with a single dielectric layer with thickness h1 of about 101 nm (Y chips). In this condition, the adhesion on the surface of tiny amount of material (e.g. biomolecules) with refractive index n2, different from ns, induces a relative increase of reflectivity that can be easily measured and used to estimate the amount of molecules adhered on the surface, even if forming sub-nanometer layers.
Here we exploited this principle by immobilizing antibodies on the SiO2 surface and measuring the real-time binding of target compounds in solution. The immobilization on the surface was achieved by means of a layer of a few nanometers of copoly(DMA-NASMAPS). The copolymer has multiple functions: (i) it adheres to the surface, (ii) it provides reactive groups for covalent antibody immobilization, and (iii) it limits the non-specific adsorption in complex media (Cretich et al., 2004). In aqueous solution, the copolymer forms a highly hydrated layer with a thickness of about 10 nm, hence yielding to a small increase of reflectivity relative to the bare SiO2 surface (Yalçin et al., 2009).
3.2 Optical model Although the overall reflectivity of a multi-layered interface is fully accounted by recursively applying Equation 1 (Pedrotti et al., 2006), a more practical model was needed to evaluate the sensitivity and to enable a rational design of the sensing substrate. Accordingly, we considered a multi-layered structure, where N layers with refractive indices n1, …, nN, and thickness h1, …, hN, respectively, are sandwiched between two semi-infinite media with refractive indices n0 (glass substrate) and nS (solution). By assuming that the first layer with refractive index n1 has a thickness h1 comparable with the wavelength of the incident beam, while all the others are much thinner, the total reflectivity of such surface can be accounted for by the addition of an extra contribution to the phase retardation of the first layer. Namely, the reflectivity can be computed using Equation 1 with α = kn1h, where the effective thickness h is
(2)
Here ck = (nk2 – ns2)/(n12 – ns2) are factors that weight the contribution to h of each thin layer, according to the corresponding refractive index. We note that ck < (=, >) 1 if nk < (=, >) n1. Importantly, the effect of changing the thickness and/or the refractive index of the layers can be accounted for by adjusting a single parameter, the effective thickness h of the equivalent single layer structure. Equation 1 can be further simplified by considering small variations of the refractive indices and of the thicknesses relative to the condition of vanishing reflectivity (R = 0), as described in the Supporting Information. In these simplifying conditions that are nevertheless adequate, we can also extend Equation 1 to include a moderate polychromaticity of the illuminating light. Indeed, by assuming an average wavenumber ¢k² and a variance ∆k2 = ¢k2² - ¢k²2, the effective reflectivity is simply obtained by averaging over the spectrum. Accordingly, by
defining r = (r1S + r01)/2, β = (r1S – r01)/r and ε2 = (2hn1¢k² - π)2 + (2hn1 ∆k)2, the reflectivity R can be rewritten to the lowest order in β and ε as:
!
(3)
Equation 3 ultimately expresses R as a function of h and n1. We call h0 the value of h yielding the minimum value of reflectivity R0. From Equation 3, we obtain h0 = π/(2n1¢k²). Figure 2a reports the reflectivity as a function of h calculated with Equation 3 (red curve) in comparison with the exact Fresnel formula of Equation 1 (blue curve). The yellow LED illumination was considered, therefore h0 = 101 nm. For a thickness h < h0 + 20 nm, the difference between the estimates obtained with the two approaches is lower than 3%. Equation 3 allows to evaluate the factors that determine the reflectivity for h ≈ h0. Two additive contributions are present: the non-optimal agreement of the refractive indices through the term β and the spectral width ∆k of the illumination source through ε2. In principle, the working conditions can be adjusted in order to have the term ε2 always larger than β2. Accordingly, a limited polychromaticity of the illumination light can provide a reduction of the dependence of the reflectivity on the uncontrolled variations of refractive indices, possibly due to temperature fluctuations or variability of solution composition. For instance, in our working conditions, a yellow LED source with a spectral width FWHM of about 20 nm can provide ε2 > β2, while maintaining the reflectivity of the order of 10-5. An insightful equation describing the response and the sensitivity of our RPI approach can be obtained from Equation 3 when the width of the illuminating spectrum is narrow relative to the average wavelength, i.e. ∆k2
