Biosensors and Bioelectronics 58 (2014) 186–192
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
Sensitive, rapid and quantitative detection of substance P in serum samples using an integrated microfluidic immunochip Josef Horak n, Can Dincer, Hüseyin Bakirci, Gerald Urban University of Freiburg, Department of Microsystems Engineering, Laboratory for Sensors, Georges-Koehler Allee 103, 79110 Freiburg, Germany
art ic l e i nf o
a b s t r a c t
Article history: Received 11 December 2013 Received in revised form 5 February 2014 Accepted 17 February 2014 Available online 4 March 2014
Miniaturized diagnostic devices hold the promise of accelerate the specific and sensitive detection of various biomarkers, which can translate into many areas of medicine – from cheaper clinical trials, to early diagnosis and treatment of complex diseases. Therefore, we report on a disposable integrated chip-based capillary immunoassay featuring a microfluidic ELISA format combining electrochemical detection and low-cost fabrication employing a dry film photoresist, Vacrels 8100. The readily accessible carboxylate groups on the material surface allow fast and high yield immobilization of biomolecules using amine-specific coupling via reactive esters requiring no laborious surface pretreatment. The integrated microfluidic system provides a convenient platform for a flow-through immunoassay. Capillary force is used for easy reagent delivery and loading the chip channel. We performed rapid quantification of serum level of substance P, a potential biomarker of acute neuroinflammation, using the developed microfluidic immunochip. Our miniaturized assay demonstrated a sensitive electrochemical detection of the antigen at 15.4 pg ml 1 (11.5 pM) using only 5 ml of the biological fluid while cutting the total assay preparation time in half and the read-out time to 10 min. Combining microfluidics and fabrication suitable for mass production with the capability of testing clinically relevant samples creates conditions for the construction of low-cost and portable point of care diagnostic devices with minimal auxiliary electronics. & 2014 Elsevier B.V. All rights reserved.
Keywords: Biochip Immunoassay Microfluidics Electrochemical detection Substance P
1. Introduction Moving biomarker detection from the laboratory to the clinic or even closer to the point of need – to the patient – has been remarkably difficult (Scott, 2013). In part this is because techniques that are routine in biomarker detection present technical (Sorger, 2008) and commercial challenges (Kling, 2006; Lesko, 2007). Such transactional bottlenecks might be eased by the availability of simple and cheap biomarker microdevices (Yager et al., 2006). We have developed a microfabrication strategy for a universal microfluidic immunosensing platform intended for cost-efficient mass production and compatibility with diverse biomolecules (Jobst and Gamp, 2010). In our previous report, the utilized material, Vacrels 8100, was studied to ascertain its suitability for a fully integrated electrochemical immunosensor (Horak et al., 2014). The chip operation was demonstrated on the proof-ofprinciple immunoenzymatic determination of IgG class antibodies against immobilized Epstein–Barr virus viral capsid antigen. However, this qualitative direct ELISA (enzyme-linked inmmunosorbent assay) format provides neither quantifiable assay performance characteristics
n
Corresponding author. Tel.: þ 49 761 203 7264; fax: þ 49 761 203 7262. E-mail address: [email protected] (J. Horak).
http://dx.doi.org/10.1016/j.bios.2014.02.058 0956-5663 & 2014 Elsevier B.V. All rights reserved.
nor immobilization properties of solid-phase bound antibodies, relevant for common competitive and sandwich ELISA formats (Herron et al., 2003; Porstmann and Kiessig, 1992). Here we report the development and on-chip implementation of a competitive assay for the detection of an analyte of low molecular weight, substance P (SP), in a patient serum sample at picomolar concentrations. The neuropeptide substance P (1345 Da) was chosen as a benchmark because its high variability levels and poor stability in body fluids require fast quantitative methods for determining the relevant in-vivo level involved in the pathogenesis of certain diseases. For competitive assays, the analyte competes with a detectable analyte-analogue for secondary antibody sites and so in effect the system measures how much analyte is not present. This introduces major problems, not only in terms of sensitivity but also read-out, where classically reducing observable signal is indicative of increasing concentration of analyte. In particular, this makes both the microplate ELISA format and point-of-need devices difficult to read and assess. Therefore, we have focused on the amplification of the observable signal and the assay reproducibility as two important characteristics that are achievable with the developed fabrication and sensing strategy. SP is a key mediator in pain perception and immune response (Harrison and Geppetti, 2001). Elevated levels in serum or in plasma are now well documented in association with a variety of disorders,
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
such as cancer (Leeman and Mroz, 1974), asthma (Zubrzycka and Janecka, 2000), sickle cell crisis (O'Connor et al., 2004) and infectious diseases such as AIDS and respiratory syncytial virus (Lazarczyk et al., 2007). SP has been suggested as an objective marker of chronic pain and migraine (Jang et al., 2011; Parris et al., 1993). Physiological concentrations may vary greatly depending on body fluid type and the patient's clinical situation. Levels in serum, plasma, urine and cerebrospinal fluid range between 5 and 115 pg ml 1 with a mean value around 40 pg ml 1 for healthy individuals (Campbell et al., 2006; Michaels et al., 1998). Saliva SP values are markedly higher, with greater fluctuations beginning at 10 pg ml 1 and exceeding 1000 pg ml 1 concentrations (Jang et al., 2011; Parris et al., 1993). We made use of a UV-curable dry film resist (DFR), Vacrels 8100, for immunosensing chip construction (Horak et al., 2014). The design aim was simple low-cost fabrication requiring minimal clean-room steps, combined with easy prototyping and handling. Moreover, Vacrels 8100 presents accessible carboxylate groups to which biomolecules can be bound directly via fast amide crosslinking. This enabled the use of an immobilization microchannel with capillary force system for reproducible delivery of multiple reagents which requires only five microliters of immunoreactants. As previously reported, the natural optical properties of photoresist materials used for microfabrication of integrated sensors limit the use of optical detection (Mogensen and Kutter, 2009; Pumera and Escarpa, 2009). The electrochemical detection that was utilized enabled high sensitivity, compatibility with a range of materials suitable for micro-fabrication, ease of miniaturization and complete integration combined with low cost and low power requirements. To our knowledge, the use of the microfluidic immunosensing method for substance P quantification has not yet been demonstrated. Along this line, we refer to two sensor-based SP detection devices utilizing surface plasmon resonance (Karasawa and Sugawara, 2005) and receptor–affinity chromatographic technique (Phillips, 1996). Laboratory-based techniques comprise mainly of commercially available detection kits, all based on competitive benchtop ELISA requiring SP incubation times from 2 h to overnight (Table S1). Such a long sample incubation in biological fluids may lead to SP degradation resulting in misleading results, which are usually higher, due to SP fragmentation (Rissler, 1995). To overcome this problem, isolation techniques employing HPLC (Beaudry and Vachon, 2006; Joyce et al., 1993) and ion exchange chromatography (Nyberg et al., 1985) with subsequent SP detection by mass spectrometry and MALDI (Keller and Li, 2001) were developed. These tandem systems made the measurement precise, but removed the advantages of the immunoassay technique. We believe that shortening incubation times while reducing costs due to drastically decreased reagent and sample volumes will translate into accessible diagnostics mainly in the point-of-care area.
187
Briefly, photomasks were designed and printed in house. A 100 nm platinum (Pt) layer was deposited on the substrate, six inch polyimide film Pyraluxs AP, and patterned using a standard photolithography and lift-off process. DFR Vacrels 8100 with a thickness of 75 mm was laminated onto the substrate at 100 1C forming 6 layers in total. All chip structures were formed by standard photolithography with subsequent bath development. On-chip pseudo-reference silver/silver chloride electrodes were prepared by electrochemical deposition. Dispersion of 1% Teflons 1600 AF on the physical barrier was followed by cover lamination at 100 1C and a final hard-bake in a conventional oven at 145 1C for 3 h. 2.3. Benchtop substance P ELISA See S1, Supplementary Data for further information (key terms, plate layout, calculation of results and assay specificity). Every step of the immunoreactions was followed by three washing cycles on a commercial ELISA microplate washer using commercial wash buffer. A standard volume of 100 ml per well was used for all immunoreactions, 200 ml per well was used for blocking. Microtiter plates (Nunc-Immuno™ LockWell™ Modules, Nunc GmbH & Co. KG) were incubated sealed on a plate shaker preheated to 25 1C at 450 rpm. The microtiter plate was coated with 10 mg ml 1 of primary antibody diluted in 10 mM PBS, 100 mM NaCl, pH 7.4 and incubation for 2 h. Blocking solution (1% BSA, 0.5% casein in 10 mM PBS, 100 mM NaCl, pH 7.4) was incubated for 30 min at 25 1C on a microplate shaker followed by overnight incubation at 4 1C without shaking. Each immunoreactant (SP standards or samples, SP–biotin, secondary antibody and reagent diluent) was added in 25 ml volumes. First, the reagent diluent (10 mM PBS, 100 mM NaCl, pH 7.4 containing 0.1% BSA) was pipetted into NSB (75 ml), B0 (50 ml) and sample wells (25 ml) followed by addition of SP– biotin conjugate and SP standards or samples. Secondary antibody was pipetted last (except for NSB) and the plate was sealed and incubated for 2 h. A solution of GOx–avidin diluted to 2 mg ml 1 in reagent diluent was added and the plates were incubated for another 30 min. For detection, 50 ml of ABTS, diluted in phosphate– citrate buffer, pH 5.6 to a concentration of 2.2 mg ml 1, was pipetted into each well. Solutions containing 160 mM of glucose and 0.8 mg ml 1 HRP both in phosphate–citrate buffer pH 5.6 were mixed in a one to one ratio. 50 ml of this solution was pipetted into each well. The plate was incubated without shaking for 1 h at 25 1C and then immediately measured on an absorbance microplate reader (SpectraMax 340PC384, Molecular Devices GmbH) preheated to 25 1C at 405 nm. Software SoftMax Pro (Molecular Devices GmbH) was used to collect the data. 2.4. Chip-based substance P ELISA
2. Material and methods 2.1. Reagents and materials Pyraluxs AP, Vacrels 8100 and Teflons 1600 AF were purchased from DuPont™. Substance P and substance P-biotin conjugate was purchased from AnaSpec Inc., primary (polyclonal goat anti-rabbit) and secondary (polyclonal rabbit anti-human SP) antibodies were purchased from AbD Serotec, GOx–avidin conjugate was obtained from Biomol. All other chemicals including serum samples were purchased from Sigma-Aldrich or otherwise as stated in the text. 2.2. The chip fabrication A detailed description of the wafer fabrication process utilizing Vacrels 8100 can be found in our previous report (Horak et al., 2014).
The chip-based assay procedure and result calculation was performed according to the benchtop SP assay described above and in S1, Supplementary data. For every incubation step, a 5 ml of (immuno)reagents was pipette on the chip inlet to fill the chip immobilization capillary. Chips were incubated in closed Petri dishes at 25 1C. Every step of the immunoreactions was followed by one washing cycle with a custom made vacuum pump using 200 ml of a wash buffer mimicking the washing procedure of conventional ELISA. The wash buffer is introduced by the outlet and removed by the inlet of the chip so other reagents do not contaminate the electrochemical measurement cell. The chips were pre-treated for 20 min with 10% Na2CO3 and washed with 100 mM MES buffer, pH 6.0. The capillary was treated with activation buffer (100 mM EDC, 200 mM sulfo-NHS solution buffered at pH 6.0 with 100 mM MES, 0.9% NaCl) for 60 min.
188
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
A 50 mg ml 1 solution of primary antibody was prepared with coupling buffer (10 mM PBS, 100 mM NaCl, pH 7.4) and incubated for 60 min in the activated capillary. After a washing step, the capillary was blocked for 30 min using 1% BSA in 10 mM PBS, 100 mM NaCl, pH 7.4. A mixture, in the proportion 1:1:1, containing secondary antibody (1 mg ml 1), SP–biotin (200 pg ml 1) and the appropriate calibrator in reagent diluent or a human serum sample was pipetted into the chip inlet and incubated for 1 h followed by washing. 5 mg ml 1 of poly-GOx–avidin (6 mg ml 1) in reagent diluent was incubated in the capillary for 15 min. The chip was then washed with 200 ml of wash buffer and measured. 2.5. Amperometric measurement Chips were fastened into a custom-made adapter connected to a potentiostat (Jobst Technologies) as depicted in Fig. 2. 100 mM PBS, 100 mM NaCl, pH 7.4 buffer was pumped (withdraw mode, pump PHD2000 Harvard Aparatus; syringes Hamilton Company) through the capillary at a constant flow rate of 15 ml min 1 (S2, Supplementary data). Once the baseline was stabilized, the glucose solution (40 mM glucose in 100 mM PBS, 100 mM NaCl, pH 7.4) was introduced and the flow was stopped for a 10 min time interval, allowing the H2O2 to accumulate in the sensor capillary. Subsequent restart of the pump yielded a signal peak. The measurement was recorded with bioMON software (Jobst Technologies) at þ350 mV vs. Ag/AgCl on-chip pseudo-reference electrode.
3. Results and discussion 3.1. The immunochip fabrication To develop a universal low-cost immunosensor, we designed a microfluidic chip (Fig. 1) using a flexible photoresist polymer capable
of being quickly processed in a non-cleanroom setting (DuPont Processing Guide). The immunochip was fabricated on wafer-level by dry-film resist technology. One six inch wafer in this study contained 40 chips; a single chip has dimensions of 10 27 mm2. The photopolymerizable layer is sandwiched between a polyester support film and a protective separator sheet. To use, the separator sheet is removed and the dry film is laminated with heat and pressure to the substrate (Pt-deposited polyimide wafer). Multilayered structures were achieved by photolithography via successive photomasks with subsequent bath development. The on-chip pseudo-reference silver/silver chloride electrodes were prepared by electrodeposition using silver electrolyte and potassium chloride solution. At the last step of the fabrication, the wafer must be baked at 145 1C to increase the crosslinking degree of the bulk material. Compared to liquid photoresists, dry film resist offer advantages such as no liquid handling since there is no solvent, excellent adhesion to most substrates, thickness uniformity over a whole wafer (can bridge over holes and structures), high processing speed and low cost. Except for the Pt deposition, all fabrication steps were performed in a non-cleanroom setting with minimal material and device costs. Removal of already laminated layers, if necessary, can be performed in stronger alkaline strippers at room temperature, which facilitates error corrections.
3.2. The immobilization and sensing strategy We incorporated a compact ‘meandering channel’ that allows convenient delivery of reagents, keeping the workflow of conventional ELISA. In order to work with high viscosity liquids like serum or blood, the channel design consists of a capillary which is 97 mm long and 500 mm wide with a depth of 75 mm given by the polymer film thickness. The channel starts with an inlet port, which is an open reservoir serving as an interface for users to
Fig. 1. Picture of the immunochip and schematics of the on-chip SP assay workflow. The chip (image without front cover) is composed of a microchannel for biomolecules immobilization, an electrochemical measurement cell and wire bonding pads. The immobilization area (highlighted red) starts with the inlet and ends at the stopping barrier. Next, the channel enters the measurement cell containing Pt working, counter and Ag/AgCl pseudo-reference electrodes. The SP assay procedure starts with immobilization of the primary antibody on the activated capillary surface followed by a blocking step with BSA. Next, a mixture of secondary antibody, biotinylated SP and analyte is incubated. In the last step, GOx–avidin is added and the chips are ready for the measurement.
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
handle and pipette liquids to fill the microchannel (Fig. 1). Our channel design uses capillary force for filling the immobilization area. This passive system does not require any external instrumentation and enables a high degree of integration and precision. At the end of the immobilization channel, the spreading of the liquid is stopped by the stopping barrier (Fig. S3), so the reagents coat a well-defined area and do not contaminate the electrochemical measurement cell. The barrier is realized by a 500 mm long spacer aligned across the capillary bottom creating a sharp edge (Kalinin et al., 2009) with a hydrophobic patch (Suk and Cho, 2007) of dispersed Teflons AF. The barrier will stop any liquid that does not totally wet the channel surface (processed Vacrels posesses a contact angle of 801, Teflons 1101). The capillary possesses a surface of 112 mm2 and 3.6 ml volume, which creates the desired surface to volume ratio (S/V) of 310 cm 1. In a solid phase assay, a larger S/V ratio implies a faster interaction between molecules from the liquid phase. According to simple kinetics modeling, incubation time is inversely proportional to the square of the surface to volume ratio (Esser, 1991). Consequently, by using our microchip system in comparison to a 96 well ELISA microplate, incubation time can theoretically be reduced by a factor ranging from 33 to 40, depending on the well volume used (considering 100 ml as standard coating and 200 ml as blocking volume). However, when more parameters are considered, due to the zero velocity of the immunoreactants in our system, the binding kinetics becomes strongly transport-limited. This leads to better analyte exploitation and lower measurement errors, but prolongs the incubation times (Zimmermann et al., 2005). Compared to microplate ELISA, the incubation could be reduced by a factor of two (Fig. S4). We observed that the factor, tplate/2, can be used as a universal pattern for all biospecific affinity reactions in the chip microchannel. The photoresist formulation is a complex mixture of ingredients containing polymer binders, photo-initiator and other additives (Niedermann et al., 2003). The acrylic binders and multifunctional acrylates contain carboxylic and carboxylate groups and can be reacted the same way as polymer brushes of poly(acrylic acid) and polymethacrylate which have been used for immobilization of biologically active molecules (Nakajima and Ikada, 1995; Wang et al., 2011). In our preceding study, we have elucidated the presence of readily accessible reactive groups on the Vacrels surface (Horak et al., 2014). The compatibility with amide crosslinking was demonstrated by attenuated total reflectance infrared spectroscopy (ATR-FTIR) and the immobilization protocol was optimized with glucose oxidase as a test biomolecule. The surface carboxylates were activated for 60 min to amino-reactive intermediates by reaction with 100 mM N-(3-(dimethylamino)propyl)carbodiimide (EDC) and 200 mM N-hydroxysulfosuccinimide (SNHS). Incubation for 60 min was used to react the protein amino groups (predominantly from the lysine residues) with the activated capillary surface. These optimized values were used for the on-chip SP immunoassay implementation. In our system, all solid-phase (immune)reactions are carried out at zero flow velocity, nevertheless, for the amperometric measurement a constant flow is needed to transport the H2O2 generated from the immobilization capillary to the working electrode (Fig. 2a). In order to achieve a strong assay response through enzyme-mediated amplification, we combined sensitive amperometric detection (Fig. 2b) with the ‘stop-flow’ read-out method. The stop-flow incubation period can be adjusted accordingly to the assay performance and shortens the signal acquisition to a few minutes (Fig. 2c). The transient signal is generated by the principle used in existing glucose sensors in which glucose is converted into hydrogen peroxide and gluconolactone in the presence of oxygen by glucose oxidase (GOx), accordingly: GOx
β D glucose þO2 !H2 O2 þ D glucono 1; 5 lactone
189
Fig. 2. Schematic diagrams of the measurement principle and signal acquisition of the electrochemical assay. (a) Schematics of the measurement setup depicting two chips fastened into the adapter, connected to a syringe pump and potentiostat. (b) Implemented SP competitive assay showing the H2O2 generation in the immobilization capillary and its subsequent oxidation at the working electrode. (c) Principle of an amperometric ‘stop-flow’ read-out with two different concentrations of SP (156.25 pg ml 1 black curve, 625 pg ml 1 red curve) spiked into a serum sample. PBS buffer is first pumped through the chip at a flow rate of 15 ml min 1. Once the baseline is stabilized, glucose substrate is then introduced and the flow is stopped (pump off), allowing the H2O2 to accumulate in the sensor capillary. Subsequent restart of the pump yields a signal peak caused by H2O2 oxidation at the working electrode. Prolonged incubation times resulted in better signal separation. Imax represents the height of the oxidation peak, Q the peak area. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Subsequently, the H2O2 is oxidized at the Pt working electrode at þ350 mV versus the Ag/AgCl reference electrode resulting in the transient signal: H2 O2
Pt electrode
!
2H þ þ O2 þ 2e
In order to investigate the optimal oxidation potential with respect to the measurement conditions, we performed on-chip cyclic voltammetry of glucose and H2O2. Glucose features two oxidation peaks at 65 mV and 515 mV while H2O2 possesses a typical broad peak with a maximum around 600 mV (Fig. S5). As a compromise, þ 350 mV was chosen as an optimal oxidation potential (situated at 85% of H2O2 current maximum) to avoid any current overlap due to glucose oxidation and enabling a faster measurement time. 3.3. Chip-based amperometric immunoassay for substance P quantification We have developed a competitive immunoassay for substance P quantification to evaluate our system for the detection of analytes of low molecular weight in human serum. Fig. 1 shows
190
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
the miniaturized ELISA preparation starting with the sequential immobilization of the primary antibody (pAb) on the activated resist material in the capillary through EDC/SNHS cross-linking, followed by a blocking step. A mixture of secondary antibody (sAb), SP–biotin conjugate and sample is added resulting in a competitive reaction between SP antigen and the biotinylated conjugate for the same binding site on the antibody. In the last step, the bound complexes are labeled with a GOx conjugated avidin. The assay was first developed and optimized in a standard ELISA format (S1 and Fig. S6) before being implemented on the chip. Due to the higher affinity of the polyclonal rabbit anti-human SP for the SP–biotin conjugate, the critical parameters in developing the assay were to establish the concentrations of the sAb and SP–biotin conjugate. Optimal assay performance was achieved with low concentrations of both immunoreagents, 0.33 mg ml 1 of secondary antibody and 66.7 pg ml 1 of SP–biotin conjugate respectively. A common way of describing the response of a competitive binding immunoassay is to use the ratio B/B0, where B0 is the amount of SP–biotin (and GOx–avidin tracer) bound to detection antibodies when the analyte is absent and B is the amount of tracer bound to antibodies when the free SP is present. When a graph of B/B0 vs log[SP] is made for such an assay, it produces a sigmoidal-shaped plot with a maximum value of 100% when no analyte is present (B0) and B/B0 value decreases to zero with increasing amounts of SP (Wild, 2005). For the chip-based assay development, the first crucial step was to establish the optimal pAb concentration needed for immobilization in the chip capillary. Chips were coated with a dilution series of primary goat anti-rabbit antibody, starting with a concentration of 10 mg ml 1 (used with benchtop ELISA) going up to 100 mg ml 1. 2 mg ml 1 of tracer has been used. The signal-tobackground ratio (SBR) was determined by SP assay performed with the NSB (nonspecific binding) and B0 calibrator. In Fig. 3a are results indicating that an excessive antibody concentration leads to increased nonspecific binding, while low concentration reduces the signal, both resulting in reduced SBR ratio. The highest SBR ratio was achieved with 50 mg ml 1 of pAb. We further deviated from the benchtop ELISA through the use of a higher concentration of the GOx–avidin conjugate. Fig. 3b shows the result from the experiment, designed as above, with 50 mg ml 1 of capture antibody. To further reduce the amount of nonspecific binding, addition of SP–biotin with secondary antibody as well as GOx–avidin was performed in reagent diluent containing 0.1% BSA. Different dilution series of GOx–avidin conjugate (2, 4, and 6 mg ml 1) were incubated for 15 min. The highest SBR was obtained with 6 mg ml 1 of the tracer. The incubation of the tracer in the last step on a properly blocked surface added one additional step, but distinctly minimized the nonspecific adsorption. Fig. 4a–c shows data of the substance P calibration performed on-chip and microplate (Fig. 4d) platforms over the concentration range from 9.76 pg ml 1 to 10,000 pg ml 1. The height of the resulting peaks, represented by the amount of detected peroxide accumulated during the stopped flow interval, decreased (Fig. 4a) with higher concentrations of SP. After subtraction of the blank signal, the peak height (Imax) as well as the peak area (Q) of the oxidation current were calculated for the respective peaks and plotted in a calibration curve normalized with % B/B0 4-parametre logistic fit, depicted in Fig. 4b and c. The limit of detection (LOD) represents the analyte concentration which gives a signal equal to the B0 sample (zero analyte binding) lowered by three times its standard deviation and was calculated from the 4-parameter logistic fit (Taylor et al., 2001; Wild, 2005). The assay precision profile, an alternative approach for assay characteristic determination (Ekins and Edwards, 1997; Price and Newman, 1997; Taylor et al., 2001), is included in S3, Supplementary data.
Fig. 3. On-chip optimization of the electrochemical SP ELISA using different concentrations of (a) primary antibody and (b) GOx–avidin tracer. Error bars represent the standard deviation determined by assaying the samples multiple times (n¼ 3) in the same assay.
Using the chip-based ELISA for SP calculated from Imax, we obtained a LOD of 15.4 pg ml 1 (11.45 pM) with a dynamic response range between 15.4 pg ml 1 and 10,000 pg ml 1. The lowest concentration tested and detected was 9.76 pg ml 1 (7.2 pM) of SP with an acceptable coefficient of variation (CV) not exceeding 10%. This is compared to the developed benchtop ELISA, where we obtained an LOD of 14.3 pg ml 1 (10.63 pM) with a CV not exceeding 10%, determined with an intra-assay (replicates in one assay) measurement. The calibration curve calculated from the peak area suffered from lower reproducibility (%CVr20) and less steep slope, which resulted in almost 2.5-time increase in LOD when compared with results calculated from Imax and benchtop ELISA (Table 1).
4. Conclusion In summary, we showed the first microfluidic chip-based competitive immunoassay for electrochemical detection of substance P in undiluted human serum. The on-chip assay format provided good diagnostic accuracy with assay parameters consistent with those obtained with the developed conventional SP ELISA. A detection limit of 15.4 pg ml 1 for SP was obtained with a wide working range of 15.4–10,000 pg ml 1 where the assay performed r10% CV intra-assay (n ¼ 3) and r12% CV inter-assay (n ¼2). Our onchip sensing method requires 20-times less reagent/sample volume, reduces the total assay preparation in half and performs fast detection within 10 min read-out time. Another attractive feature of this approach is the ability to use reagents and workflow developed for standard ELISA procedure. The photoresist material, Vacrels 8100, allows rapid prototyping and easy processing on the wafer scale with low cost equipment in a non-cleanroom environment. In addition, the integration of the
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
191
Fig. 4. (a) Measured oxidation peaks corresponding to SP calibration recorded within 10 min stop-flow and plotted as (b) peak height and (c) peak area normalized with % B/B0 vs. log SP concentration with 4-parametre logistic fit; (d) comparison with SP calibration on microplate is given. Error bars represent %CV and are collected by two subsequent measurements in multiple assays (inter-assay).
Table 1 Comparison of chip- and microplate-based substance P ELISA parameters. Parameters 1
IC50 (pg ml ) Quality of fit Slope (% ml pg 1) Precision intra-assay (%CV) Precision inter-assay (%CV) LOD (pg ml 1) Reagent/sample volumea (ml) Total assay time (h)
Fig. 5. Chip-based evaluation of SP quantification in human serum using the height of the oxidation peaks from Figs. 2 and 4a. SP calibration (black) and serum samples (red) from one donor spiked with 156.25 (S1) and 625 (S2) pg ml 1 of SP. Error bars represent %CV and are collected by subsequent measurements in multiple assays (n¼ 2). Serum samples were assayed in triplicate intra-assay. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
microfluidic system enables fast biofunctionalization and sample filling using capillary force inherent to the microchannel. These advantages in combination with sensitive electrochemical detection resulted in the development of a miniaturized diagnostic platform allowing high performance bioanalytical immunoassay. Further work is ongoing to investigate the implementation of sandwich types of ELISAs for analytes of higher molecular weight with the possibility of further shortening the total assay time by surface modification with biocompatible polycations. It seems that the use of such a tunable scaffold for biomolecule immobilization
Chip Imax
Chip Q
Microplate
502 0.9876 0.9950 r10 r12 15.4 5 4
565 0.9516 0.7691 r 20 r 20 37.1
539 0.9996 1.005 r 10 r 10 14.3 100/200b 6þ overnightc
Next, we assessed the diagnostic accuracy of the microfluidic assay to quantitatively detect SP in a serum sample. 625 pg ml 1 (S2) and 156.25 pg ml 1 (S1) of SP are spiked into the specimen and assayed (Fig. 5). A 4-fold serial dilution is typically used to reduce matrix effects in immunoassays (Findlay and Dillard, 2007). SP levels were calculated using the 4-parameter logistic fit from Imax calibration curve (Fig. 4b). We obtained a good recovery yield between the expected and observed SP concentration. 636 pg ml 1 (98% recovery) for S2 and 140 pg ml 1 (88% recovery) for S1 serum sample were measured with typical CV of 8.5% and 10% respectively over three parallel measurements. a b c
Volume per well/chip. Volume used for plate blocking. Blocking step.
can provide even lower non-specific binding and independence from the surface/material properties while preserving the biological activity.
Acknowledgment The authors would like to thank the University Medical Center Freiburg for providing serum samples and the European Commission for financial support under the framework of the Marie Curie Research and Trainings Network “Cellcheck” (MCRTN-CT-2006035854).
192
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2014.02.058. References Beaudry, F., Vachon, P., 2006. Biomed. Chromatogr. 20, 1344–1350. Campbell, D.E., Raftery, N., Tustin Iii, R., Tustin, N.B., DeSilvio, M.L., Cnaan, A., Aye, P.P., Lackner, A.A., Douglas, S.D., 2006. Clin. Vaccine Immunol. 13, 1197–1203. DuPont, Vacrel (R) 8100 Quick Reference Processing Guide. Ekins, R., Edwards, P., 1997. Clin. Chem. 43, 1824–1831. Esser, P., 1991. Thermo Fisher Scientific Technical Bulletin, 10b. Findlay, J.W.A., Dillard, R.F., 2007. AAPS J. 9, 260–267. Harrison, S., Geppetti, P., 2001. Int. J. Biochem. Cell B 33, 555–576. Herron, J.N., Wang, H.K., Janatova¡, V., Durtschi, J.D., Christensen, D.A., Caldwell, K.D., Chang, I.N., Huang, S.C, 2003. Biopolymers at Interfaces, second ed., pp. 115–164. Horak, J., Dincer, C., Bakirci, H., Urban, G., 2014. Sens. Actuators B. 191, 813–820. Jang, M.U., Park, J.W., Kho, H.S., Chung, S.C., Chung, J.W., 2011. Oral Dis. 17, 187–193. Jobst, G., Gamp, T., 2010. US7691623 B2. Joyce, T.J., Yood, R.A., Carraway, R.E., 1993. J. Clin. Endocrinol. Metab. 77, 632. Kalinin, Y.V., Berejnov, V., Thorne, R.E., 2009. Langmuir 25, 5391–5397. Karasawa, T., Sugawara, M., 2005. Anal. Sci. 21, 1431–1436. Keller, B.O., Li, L., 2001. J. Am. Soc. Mass Spectr. 12, 1055–1063. Kling, J., 2006. Nat. Biotechnol. 24, 891–898. Lazarczyk, M., Matyja, E., Lipkowski, A., 2007. Folia Neuropathol. 45, 99. Leeman, S.E., Mroz, E.A., 1974. Life Sci. 15, 2033–2044.
Lesko, L.J., 2007. Clin. Pharmacol. Ther. 81, 807–816. Michaels, L.A., Ohene-Frempong, K., Zhao, H., Douglas, S.D., 1998. Blood. 92, 3148–3151. Mogensen, K.B., Kutter, J.P., 2009. Electroanalysis 30, 92–100. Nakajima, N., Ikada, Y., 1995. Bioconj. Chem. 6, 123–130. Niedermann, P., Berthou, H., Zwickl, S., Schönholzer, U., Meier, K., Gantner, C., KappSchwoerer, D., 2003. Microelectron. Eng. 67, 259–265. Nyberg, F., Le Greves, P., Terenius, L., 1985. Proc. Natl. Acad. Sci. USA 82, 3921. O'Connor, T.M., O'Connell, J., O'Brien, D.I., Goode, T., Bredin, C.P., Shanahan, F., 2004. J. Cell. Physiol. 201, 167–180. Parris, W.C.V., Sastry, B.V.R., Kambam, J.R., Naukam, R.J., Johnson, B.W., 1993. Ann. NY Acad. Sci. 694, 308–310. Phillips, T.M., 1996. Biomed. Chromatogr. 10, 331–336. Porstmann, T., Kiessig, S.T., 1992. J. Immunol. Methods 150, 5–21. Price, C.P., Newman, D.J., 1997. Principles and Practice of Immunoassay, Macmillan Reference Ltd, London. Pumera, M., Escarpa, A., 2009. Electrophoresis. 30, 3315–3323. Rissler, K., 1995. J. Chromatogr. B. 665, 233–270. Scott, S.A., 2013. Clin. Pharmacol. Ther. 93, 33–35. Sorger, P.K., 2008. Nat. Biotechnol. 26, 1345–1346. Suk, J.W., Cho, J.H., 2007. J. Micromech. Microeng. 17, N11. Taylor, J., Picelli, G., Harrison, D.J., 2001. Electrophoresis 22, 3699–3708. Wang, C., Yan, Q., Liu, H.B., Zhou, X.H., Xiao, S.J., 2011. Langmuir 27, 12058–12068. Wild, D., 2005. The Immunoassay Handbook. Elsevier Science Limited, Oxford. Yager, P., Edwards, T., Fu, E., Helton, K., Nelson, K., Tam, M.R., Weigl, B.H., 2006. Nature. 442, 412–418. Zimmermann, M., Delamarche, E., Wolf, M., Hunziker, P., 2005. Biomed. Microdevices 7, 99–110. Zubrzycka, M., Janecka, A., 2000. J. Neurophysiol. 34, 195–202.
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
Sensitive, rapid and quantitative detection of substance P in serum samples using an integrated microfluidic immunochip Josef Horak n, Can Dincer, Hüseyin Bakirci, Gerald Urban University of Freiburg, Department of Microsystems Engineering, Laboratory for Sensors, Georges-Koehler Allee 103, 79110 Freiburg, Germany
art ic l e i nf o
a b s t r a c t
Article history: Received 11 December 2013 Received in revised form 5 February 2014 Accepted 17 February 2014 Available online 4 March 2014
Miniaturized diagnostic devices hold the promise of accelerate the specific and sensitive detection of various biomarkers, which can translate into many areas of medicine – from cheaper clinical trials, to early diagnosis and treatment of complex diseases. Therefore, we report on a disposable integrated chip-based capillary immunoassay featuring a microfluidic ELISA format combining electrochemical detection and low-cost fabrication employing a dry film photoresist, Vacrels 8100. The readily accessible carboxylate groups on the material surface allow fast and high yield immobilization of biomolecules using amine-specific coupling via reactive esters requiring no laborious surface pretreatment. The integrated microfluidic system provides a convenient platform for a flow-through immunoassay. Capillary force is used for easy reagent delivery and loading the chip channel. We performed rapid quantification of serum level of substance P, a potential biomarker of acute neuroinflammation, using the developed microfluidic immunochip. Our miniaturized assay demonstrated a sensitive electrochemical detection of the antigen at 15.4 pg ml 1 (11.5 pM) using only 5 ml of the biological fluid while cutting the total assay preparation time in half and the read-out time to 10 min. Combining microfluidics and fabrication suitable for mass production with the capability of testing clinically relevant samples creates conditions for the construction of low-cost and portable point of care diagnostic devices with minimal auxiliary electronics. & 2014 Elsevier B.V. All rights reserved.
Keywords: Biochip Immunoassay Microfluidics Electrochemical detection Substance P
1. Introduction Moving biomarker detection from the laboratory to the clinic or even closer to the point of need – to the patient – has been remarkably difficult (Scott, 2013). In part this is because techniques that are routine in biomarker detection present technical (Sorger, 2008) and commercial challenges (Kling, 2006; Lesko, 2007). Such transactional bottlenecks might be eased by the availability of simple and cheap biomarker microdevices (Yager et al., 2006). We have developed a microfabrication strategy for a universal microfluidic immunosensing platform intended for cost-efficient mass production and compatibility with diverse biomolecules (Jobst and Gamp, 2010). In our previous report, the utilized material, Vacrels 8100, was studied to ascertain its suitability for a fully integrated electrochemical immunosensor (Horak et al., 2014). The chip operation was demonstrated on the proof-ofprinciple immunoenzymatic determination of IgG class antibodies against immobilized Epstein–Barr virus viral capsid antigen. However, this qualitative direct ELISA (enzyme-linked inmmunosorbent assay) format provides neither quantifiable assay performance characteristics
n
Corresponding author. Tel.: þ 49 761 203 7264; fax: þ 49 761 203 7262. E-mail address: [email protected] (J. Horak).
http://dx.doi.org/10.1016/j.bios.2014.02.058 0956-5663 & 2014 Elsevier B.V. All rights reserved.
nor immobilization properties of solid-phase bound antibodies, relevant for common competitive and sandwich ELISA formats (Herron et al., 2003; Porstmann and Kiessig, 1992). Here we report the development and on-chip implementation of a competitive assay for the detection of an analyte of low molecular weight, substance P (SP), in a patient serum sample at picomolar concentrations. The neuropeptide substance P (1345 Da) was chosen as a benchmark because its high variability levels and poor stability in body fluids require fast quantitative methods for determining the relevant in-vivo level involved in the pathogenesis of certain diseases. For competitive assays, the analyte competes with a detectable analyte-analogue for secondary antibody sites and so in effect the system measures how much analyte is not present. This introduces major problems, not only in terms of sensitivity but also read-out, where classically reducing observable signal is indicative of increasing concentration of analyte. In particular, this makes both the microplate ELISA format and point-of-need devices difficult to read and assess. Therefore, we have focused on the amplification of the observable signal and the assay reproducibility as two important characteristics that are achievable with the developed fabrication and sensing strategy. SP is a key mediator in pain perception and immune response (Harrison and Geppetti, 2001). Elevated levels in serum or in plasma are now well documented in association with a variety of disorders,
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
such as cancer (Leeman and Mroz, 1974), asthma (Zubrzycka and Janecka, 2000), sickle cell crisis (O'Connor et al., 2004) and infectious diseases such as AIDS and respiratory syncytial virus (Lazarczyk et al., 2007). SP has been suggested as an objective marker of chronic pain and migraine (Jang et al., 2011; Parris et al., 1993). Physiological concentrations may vary greatly depending on body fluid type and the patient's clinical situation. Levels in serum, plasma, urine and cerebrospinal fluid range between 5 and 115 pg ml 1 with a mean value around 40 pg ml 1 for healthy individuals (Campbell et al., 2006; Michaels et al., 1998). Saliva SP values are markedly higher, with greater fluctuations beginning at 10 pg ml 1 and exceeding 1000 pg ml 1 concentrations (Jang et al., 2011; Parris et al., 1993). We made use of a UV-curable dry film resist (DFR), Vacrels 8100, for immunosensing chip construction (Horak et al., 2014). The design aim was simple low-cost fabrication requiring minimal clean-room steps, combined with easy prototyping and handling. Moreover, Vacrels 8100 presents accessible carboxylate groups to which biomolecules can be bound directly via fast amide crosslinking. This enabled the use of an immobilization microchannel with capillary force system for reproducible delivery of multiple reagents which requires only five microliters of immunoreactants. As previously reported, the natural optical properties of photoresist materials used for microfabrication of integrated sensors limit the use of optical detection (Mogensen and Kutter, 2009; Pumera and Escarpa, 2009). The electrochemical detection that was utilized enabled high sensitivity, compatibility with a range of materials suitable for micro-fabrication, ease of miniaturization and complete integration combined with low cost and low power requirements. To our knowledge, the use of the microfluidic immunosensing method for substance P quantification has not yet been demonstrated. Along this line, we refer to two sensor-based SP detection devices utilizing surface plasmon resonance (Karasawa and Sugawara, 2005) and receptor–affinity chromatographic technique (Phillips, 1996). Laboratory-based techniques comprise mainly of commercially available detection kits, all based on competitive benchtop ELISA requiring SP incubation times from 2 h to overnight (Table S1). Such a long sample incubation in biological fluids may lead to SP degradation resulting in misleading results, which are usually higher, due to SP fragmentation (Rissler, 1995). To overcome this problem, isolation techniques employing HPLC (Beaudry and Vachon, 2006; Joyce et al., 1993) and ion exchange chromatography (Nyberg et al., 1985) with subsequent SP detection by mass spectrometry and MALDI (Keller and Li, 2001) were developed. These tandem systems made the measurement precise, but removed the advantages of the immunoassay technique. We believe that shortening incubation times while reducing costs due to drastically decreased reagent and sample volumes will translate into accessible diagnostics mainly in the point-of-care area.
187
Briefly, photomasks were designed and printed in house. A 100 nm platinum (Pt) layer was deposited on the substrate, six inch polyimide film Pyraluxs AP, and patterned using a standard photolithography and lift-off process. DFR Vacrels 8100 with a thickness of 75 mm was laminated onto the substrate at 100 1C forming 6 layers in total. All chip structures were formed by standard photolithography with subsequent bath development. On-chip pseudo-reference silver/silver chloride electrodes were prepared by electrochemical deposition. Dispersion of 1% Teflons 1600 AF on the physical barrier was followed by cover lamination at 100 1C and a final hard-bake in a conventional oven at 145 1C for 3 h. 2.3. Benchtop substance P ELISA See S1, Supplementary Data for further information (key terms, plate layout, calculation of results and assay specificity). Every step of the immunoreactions was followed by three washing cycles on a commercial ELISA microplate washer using commercial wash buffer. A standard volume of 100 ml per well was used for all immunoreactions, 200 ml per well was used for blocking. Microtiter plates (Nunc-Immuno™ LockWell™ Modules, Nunc GmbH & Co. KG) were incubated sealed on a plate shaker preheated to 25 1C at 450 rpm. The microtiter plate was coated with 10 mg ml 1 of primary antibody diluted in 10 mM PBS, 100 mM NaCl, pH 7.4 and incubation for 2 h. Blocking solution (1% BSA, 0.5% casein in 10 mM PBS, 100 mM NaCl, pH 7.4) was incubated for 30 min at 25 1C on a microplate shaker followed by overnight incubation at 4 1C without shaking. Each immunoreactant (SP standards or samples, SP–biotin, secondary antibody and reagent diluent) was added in 25 ml volumes. First, the reagent diluent (10 mM PBS, 100 mM NaCl, pH 7.4 containing 0.1% BSA) was pipetted into NSB (75 ml), B0 (50 ml) and sample wells (25 ml) followed by addition of SP– biotin conjugate and SP standards or samples. Secondary antibody was pipetted last (except for NSB) and the plate was sealed and incubated for 2 h. A solution of GOx–avidin diluted to 2 mg ml 1 in reagent diluent was added and the plates were incubated for another 30 min. For detection, 50 ml of ABTS, diluted in phosphate– citrate buffer, pH 5.6 to a concentration of 2.2 mg ml 1, was pipetted into each well. Solutions containing 160 mM of glucose and 0.8 mg ml 1 HRP both in phosphate–citrate buffer pH 5.6 were mixed in a one to one ratio. 50 ml of this solution was pipetted into each well. The plate was incubated without shaking for 1 h at 25 1C and then immediately measured on an absorbance microplate reader (SpectraMax 340PC384, Molecular Devices GmbH) preheated to 25 1C at 405 nm. Software SoftMax Pro (Molecular Devices GmbH) was used to collect the data. 2.4. Chip-based substance P ELISA
2. Material and methods 2.1. Reagents and materials Pyraluxs AP, Vacrels 8100 and Teflons 1600 AF were purchased from DuPont™. Substance P and substance P-biotin conjugate was purchased from AnaSpec Inc., primary (polyclonal goat anti-rabbit) and secondary (polyclonal rabbit anti-human SP) antibodies were purchased from AbD Serotec, GOx–avidin conjugate was obtained from Biomol. All other chemicals including serum samples were purchased from Sigma-Aldrich or otherwise as stated in the text. 2.2. The chip fabrication A detailed description of the wafer fabrication process utilizing Vacrels 8100 can be found in our previous report (Horak et al., 2014).
The chip-based assay procedure and result calculation was performed according to the benchtop SP assay described above and in S1, Supplementary data. For every incubation step, a 5 ml of (immuno)reagents was pipette on the chip inlet to fill the chip immobilization capillary. Chips were incubated in closed Petri dishes at 25 1C. Every step of the immunoreactions was followed by one washing cycle with a custom made vacuum pump using 200 ml of a wash buffer mimicking the washing procedure of conventional ELISA. The wash buffer is introduced by the outlet and removed by the inlet of the chip so other reagents do not contaminate the electrochemical measurement cell. The chips were pre-treated for 20 min with 10% Na2CO3 and washed with 100 mM MES buffer, pH 6.0. The capillary was treated with activation buffer (100 mM EDC, 200 mM sulfo-NHS solution buffered at pH 6.0 with 100 mM MES, 0.9% NaCl) for 60 min.
188
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
A 50 mg ml 1 solution of primary antibody was prepared with coupling buffer (10 mM PBS, 100 mM NaCl, pH 7.4) and incubated for 60 min in the activated capillary. After a washing step, the capillary was blocked for 30 min using 1% BSA in 10 mM PBS, 100 mM NaCl, pH 7.4. A mixture, in the proportion 1:1:1, containing secondary antibody (1 mg ml 1), SP–biotin (200 pg ml 1) and the appropriate calibrator in reagent diluent or a human serum sample was pipetted into the chip inlet and incubated for 1 h followed by washing. 5 mg ml 1 of poly-GOx–avidin (6 mg ml 1) in reagent diluent was incubated in the capillary for 15 min. The chip was then washed with 200 ml of wash buffer and measured. 2.5. Amperometric measurement Chips were fastened into a custom-made adapter connected to a potentiostat (Jobst Technologies) as depicted in Fig. 2. 100 mM PBS, 100 mM NaCl, pH 7.4 buffer was pumped (withdraw mode, pump PHD2000 Harvard Aparatus; syringes Hamilton Company) through the capillary at a constant flow rate of 15 ml min 1 (S2, Supplementary data). Once the baseline was stabilized, the glucose solution (40 mM glucose in 100 mM PBS, 100 mM NaCl, pH 7.4) was introduced and the flow was stopped for a 10 min time interval, allowing the H2O2 to accumulate in the sensor capillary. Subsequent restart of the pump yielded a signal peak. The measurement was recorded with bioMON software (Jobst Technologies) at þ350 mV vs. Ag/AgCl on-chip pseudo-reference electrode.
3. Results and discussion 3.1. The immunochip fabrication To develop a universal low-cost immunosensor, we designed a microfluidic chip (Fig. 1) using a flexible photoresist polymer capable
of being quickly processed in a non-cleanroom setting (DuPont Processing Guide). The immunochip was fabricated on wafer-level by dry-film resist technology. One six inch wafer in this study contained 40 chips; a single chip has dimensions of 10 27 mm2. The photopolymerizable layer is sandwiched between a polyester support film and a protective separator sheet. To use, the separator sheet is removed and the dry film is laminated with heat and pressure to the substrate (Pt-deposited polyimide wafer). Multilayered structures were achieved by photolithography via successive photomasks with subsequent bath development. The on-chip pseudo-reference silver/silver chloride electrodes were prepared by electrodeposition using silver electrolyte and potassium chloride solution. At the last step of the fabrication, the wafer must be baked at 145 1C to increase the crosslinking degree of the bulk material. Compared to liquid photoresists, dry film resist offer advantages such as no liquid handling since there is no solvent, excellent adhesion to most substrates, thickness uniformity over a whole wafer (can bridge over holes and structures), high processing speed and low cost. Except for the Pt deposition, all fabrication steps were performed in a non-cleanroom setting with minimal material and device costs. Removal of already laminated layers, if necessary, can be performed in stronger alkaline strippers at room temperature, which facilitates error corrections.
3.2. The immobilization and sensing strategy We incorporated a compact ‘meandering channel’ that allows convenient delivery of reagents, keeping the workflow of conventional ELISA. In order to work with high viscosity liquids like serum or blood, the channel design consists of a capillary which is 97 mm long and 500 mm wide with a depth of 75 mm given by the polymer film thickness. The channel starts with an inlet port, which is an open reservoir serving as an interface for users to
Fig. 1. Picture of the immunochip and schematics of the on-chip SP assay workflow. The chip (image without front cover) is composed of a microchannel for biomolecules immobilization, an electrochemical measurement cell and wire bonding pads. The immobilization area (highlighted red) starts with the inlet and ends at the stopping barrier. Next, the channel enters the measurement cell containing Pt working, counter and Ag/AgCl pseudo-reference electrodes. The SP assay procedure starts with immobilization of the primary antibody on the activated capillary surface followed by a blocking step with BSA. Next, a mixture of secondary antibody, biotinylated SP and analyte is incubated. In the last step, GOx–avidin is added and the chips are ready for the measurement.
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
handle and pipette liquids to fill the microchannel (Fig. 1). Our channel design uses capillary force for filling the immobilization area. This passive system does not require any external instrumentation and enables a high degree of integration and precision. At the end of the immobilization channel, the spreading of the liquid is stopped by the stopping barrier (Fig. S3), so the reagents coat a well-defined area and do not contaminate the electrochemical measurement cell. The barrier is realized by a 500 mm long spacer aligned across the capillary bottom creating a sharp edge (Kalinin et al., 2009) with a hydrophobic patch (Suk and Cho, 2007) of dispersed Teflons AF. The barrier will stop any liquid that does not totally wet the channel surface (processed Vacrels posesses a contact angle of 801, Teflons 1101). The capillary possesses a surface of 112 mm2 and 3.6 ml volume, which creates the desired surface to volume ratio (S/V) of 310 cm 1. In a solid phase assay, a larger S/V ratio implies a faster interaction between molecules from the liquid phase. According to simple kinetics modeling, incubation time is inversely proportional to the square of the surface to volume ratio (Esser, 1991). Consequently, by using our microchip system in comparison to a 96 well ELISA microplate, incubation time can theoretically be reduced by a factor ranging from 33 to 40, depending on the well volume used (considering 100 ml as standard coating and 200 ml as blocking volume). However, when more parameters are considered, due to the zero velocity of the immunoreactants in our system, the binding kinetics becomes strongly transport-limited. This leads to better analyte exploitation and lower measurement errors, but prolongs the incubation times (Zimmermann et al., 2005). Compared to microplate ELISA, the incubation could be reduced by a factor of two (Fig. S4). We observed that the factor, tplate/2, can be used as a universal pattern for all biospecific affinity reactions in the chip microchannel. The photoresist formulation is a complex mixture of ingredients containing polymer binders, photo-initiator and other additives (Niedermann et al., 2003). The acrylic binders and multifunctional acrylates contain carboxylic and carboxylate groups and can be reacted the same way as polymer brushes of poly(acrylic acid) and polymethacrylate which have been used for immobilization of biologically active molecules (Nakajima and Ikada, 1995; Wang et al., 2011). In our preceding study, we have elucidated the presence of readily accessible reactive groups on the Vacrels surface (Horak et al., 2014). The compatibility with amide crosslinking was demonstrated by attenuated total reflectance infrared spectroscopy (ATR-FTIR) and the immobilization protocol was optimized with glucose oxidase as a test biomolecule. The surface carboxylates were activated for 60 min to amino-reactive intermediates by reaction with 100 mM N-(3-(dimethylamino)propyl)carbodiimide (EDC) and 200 mM N-hydroxysulfosuccinimide (SNHS). Incubation for 60 min was used to react the protein amino groups (predominantly from the lysine residues) with the activated capillary surface. These optimized values were used for the on-chip SP immunoassay implementation. In our system, all solid-phase (immune)reactions are carried out at zero flow velocity, nevertheless, for the amperometric measurement a constant flow is needed to transport the H2O2 generated from the immobilization capillary to the working electrode (Fig. 2a). In order to achieve a strong assay response through enzyme-mediated amplification, we combined sensitive amperometric detection (Fig. 2b) with the ‘stop-flow’ read-out method. The stop-flow incubation period can be adjusted accordingly to the assay performance and shortens the signal acquisition to a few minutes (Fig. 2c). The transient signal is generated by the principle used in existing glucose sensors in which glucose is converted into hydrogen peroxide and gluconolactone in the presence of oxygen by glucose oxidase (GOx), accordingly: GOx
β D glucose þO2 !H2 O2 þ D glucono 1; 5 lactone
189
Fig. 2. Schematic diagrams of the measurement principle and signal acquisition of the electrochemical assay. (a) Schematics of the measurement setup depicting two chips fastened into the adapter, connected to a syringe pump and potentiostat. (b) Implemented SP competitive assay showing the H2O2 generation in the immobilization capillary and its subsequent oxidation at the working electrode. (c) Principle of an amperometric ‘stop-flow’ read-out with two different concentrations of SP (156.25 pg ml 1 black curve, 625 pg ml 1 red curve) spiked into a serum sample. PBS buffer is first pumped through the chip at a flow rate of 15 ml min 1. Once the baseline is stabilized, glucose substrate is then introduced and the flow is stopped (pump off), allowing the H2O2 to accumulate in the sensor capillary. Subsequent restart of the pump yields a signal peak caused by H2O2 oxidation at the working electrode. Prolonged incubation times resulted in better signal separation. Imax represents the height of the oxidation peak, Q the peak area. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Subsequently, the H2O2 is oxidized at the Pt working electrode at þ350 mV versus the Ag/AgCl reference electrode resulting in the transient signal: H2 O2
Pt electrode
!
2H þ þ O2 þ 2e
In order to investigate the optimal oxidation potential with respect to the measurement conditions, we performed on-chip cyclic voltammetry of glucose and H2O2. Glucose features two oxidation peaks at 65 mV and 515 mV while H2O2 possesses a typical broad peak with a maximum around 600 mV (Fig. S5). As a compromise, þ 350 mV was chosen as an optimal oxidation potential (situated at 85% of H2O2 current maximum) to avoid any current overlap due to glucose oxidation and enabling a faster measurement time. 3.3. Chip-based amperometric immunoassay for substance P quantification We have developed a competitive immunoassay for substance P quantification to evaluate our system for the detection of analytes of low molecular weight in human serum. Fig. 1 shows
190
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
the miniaturized ELISA preparation starting with the sequential immobilization of the primary antibody (pAb) on the activated resist material in the capillary through EDC/SNHS cross-linking, followed by a blocking step. A mixture of secondary antibody (sAb), SP–biotin conjugate and sample is added resulting in a competitive reaction between SP antigen and the biotinylated conjugate for the same binding site on the antibody. In the last step, the bound complexes are labeled with a GOx conjugated avidin. The assay was first developed and optimized in a standard ELISA format (S1 and Fig. S6) before being implemented on the chip. Due to the higher affinity of the polyclonal rabbit anti-human SP for the SP–biotin conjugate, the critical parameters in developing the assay were to establish the concentrations of the sAb and SP–biotin conjugate. Optimal assay performance was achieved with low concentrations of both immunoreagents, 0.33 mg ml 1 of secondary antibody and 66.7 pg ml 1 of SP–biotin conjugate respectively. A common way of describing the response of a competitive binding immunoassay is to use the ratio B/B0, where B0 is the amount of SP–biotin (and GOx–avidin tracer) bound to detection antibodies when the analyte is absent and B is the amount of tracer bound to antibodies when the free SP is present. When a graph of B/B0 vs log[SP] is made for such an assay, it produces a sigmoidal-shaped plot with a maximum value of 100% when no analyte is present (B0) and B/B0 value decreases to zero with increasing amounts of SP (Wild, 2005). For the chip-based assay development, the first crucial step was to establish the optimal pAb concentration needed for immobilization in the chip capillary. Chips were coated with a dilution series of primary goat anti-rabbit antibody, starting with a concentration of 10 mg ml 1 (used with benchtop ELISA) going up to 100 mg ml 1. 2 mg ml 1 of tracer has been used. The signal-tobackground ratio (SBR) was determined by SP assay performed with the NSB (nonspecific binding) and B0 calibrator. In Fig. 3a are results indicating that an excessive antibody concentration leads to increased nonspecific binding, while low concentration reduces the signal, both resulting in reduced SBR ratio. The highest SBR ratio was achieved with 50 mg ml 1 of pAb. We further deviated from the benchtop ELISA through the use of a higher concentration of the GOx–avidin conjugate. Fig. 3b shows the result from the experiment, designed as above, with 50 mg ml 1 of capture antibody. To further reduce the amount of nonspecific binding, addition of SP–biotin with secondary antibody as well as GOx–avidin was performed in reagent diluent containing 0.1% BSA. Different dilution series of GOx–avidin conjugate (2, 4, and 6 mg ml 1) were incubated for 15 min. The highest SBR was obtained with 6 mg ml 1 of the tracer. The incubation of the tracer in the last step on a properly blocked surface added one additional step, but distinctly minimized the nonspecific adsorption. Fig. 4a–c shows data of the substance P calibration performed on-chip and microplate (Fig. 4d) platforms over the concentration range from 9.76 pg ml 1 to 10,000 pg ml 1. The height of the resulting peaks, represented by the amount of detected peroxide accumulated during the stopped flow interval, decreased (Fig. 4a) with higher concentrations of SP. After subtraction of the blank signal, the peak height (Imax) as well as the peak area (Q) of the oxidation current were calculated for the respective peaks and plotted in a calibration curve normalized with % B/B0 4-parametre logistic fit, depicted in Fig. 4b and c. The limit of detection (LOD) represents the analyte concentration which gives a signal equal to the B0 sample (zero analyte binding) lowered by three times its standard deviation and was calculated from the 4-parameter logistic fit (Taylor et al., 2001; Wild, 2005). The assay precision profile, an alternative approach for assay characteristic determination (Ekins and Edwards, 1997; Price and Newman, 1997; Taylor et al., 2001), is included in S3, Supplementary data.
Fig. 3. On-chip optimization of the electrochemical SP ELISA using different concentrations of (a) primary antibody and (b) GOx–avidin tracer. Error bars represent the standard deviation determined by assaying the samples multiple times (n¼ 3) in the same assay.
Using the chip-based ELISA for SP calculated from Imax, we obtained a LOD of 15.4 pg ml 1 (11.45 pM) with a dynamic response range between 15.4 pg ml 1 and 10,000 pg ml 1. The lowest concentration tested and detected was 9.76 pg ml 1 (7.2 pM) of SP with an acceptable coefficient of variation (CV) not exceeding 10%. This is compared to the developed benchtop ELISA, where we obtained an LOD of 14.3 pg ml 1 (10.63 pM) with a CV not exceeding 10%, determined with an intra-assay (replicates in one assay) measurement. The calibration curve calculated from the peak area suffered from lower reproducibility (%CVr20) and less steep slope, which resulted in almost 2.5-time increase in LOD when compared with results calculated from Imax and benchtop ELISA (Table 1).
4. Conclusion In summary, we showed the first microfluidic chip-based competitive immunoassay for electrochemical detection of substance P in undiluted human serum. The on-chip assay format provided good diagnostic accuracy with assay parameters consistent with those obtained with the developed conventional SP ELISA. A detection limit of 15.4 pg ml 1 for SP was obtained with a wide working range of 15.4–10,000 pg ml 1 where the assay performed r10% CV intra-assay (n ¼ 3) and r12% CV inter-assay (n ¼2). Our onchip sensing method requires 20-times less reagent/sample volume, reduces the total assay preparation in half and performs fast detection within 10 min read-out time. Another attractive feature of this approach is the ability to use reagents and workflow developed for standard ELISA procedure. The photoresist material, Vacrels 8100, allows rapid prototyping and easy processing on the wafer scale with low cost equipment in a non-cleanroom environment. In addition, the integration of the
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
191
Fig. 4. (a) Measured oxidation peaks corresponding to SP calibration recorded within 10 min stop-flow and plotted as (b) peak height and (c) peak area normalized with % B/B0 vs. log SP concentration with 4-parametre logistic fit; (d) comparison with SP calibration on microplate is given. Error bars represent %CV and are collected by two subsequent measurements in multiple assays (inter-assay).
Table 1 Comparison of chip- and microplate-based substance P ELISA parameters. Parameters 1
IC50 (pg ml ) Quality of fit Slope (% ml pg 1) Precision intra-assay (%CV) Precision inter-assay (%CV) LOD (pg ml 1) Reagent/sample volumea (ml) Total assay time (h)
Fig. 5. Chip-based evaluation of SP quantification in human serum using the height of the oxidation peaks from Figs. 2 and 4a. SP calibration (black) and serum samples (red) from one donor spiked with 156.25 (S1) and 625 (S2) pg ml 1 of SP. Error bars represent %CV and are collected by subsequent measurements in multiple assays (n¼ 2). Serum samples were assayed in triplicate intra-assay. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
microfluidic system enables fast biofunctionalization and sample filling using capillary force inherent to the microchannel. These advantages in combination with sensitive electrochemical detection resulted in the development of a miniaturized diagnostic platform allowing high performance bioanalytical immunoassay. Further work is ongoing to investigate the implementation of sandwich types of ELISAs for analytes of higher molecular weight with the possibility of further shortening the total assay time by surface modification with biocompatible polycations. It seems that the use of such a tunable scaffold for biomolecule immobilization
Chip Imax
Chip Q
Microplate
502 0.9876 0.9950 r10 r12 15.4 5 4
565 0.9516 0.7691 r 20 r 20 37.1
539 0.9996 1.005 r 10 r 10 14.3 100/200b 6þ overnightc
Next, we assessed the diagnostic accuracy of the microfluidic assay to quantitatively detect SP in a serum sample. 625 pg ml 1 (S2) and 156.25 pg ml 1 (S1) of SP are spiked into the specimen and assayed (Fig. 5). A 4-fold serial dilution is typically used to reduce matrix effects in immunoassays (Findlay and Dillard, 2007). SP levels were calculated using the 4-parameter logistic fit from Imax calibration curve (Fig. 4b). We obtained a good recovery yield between the expected and observed SP concentration. 636 pg ml 1 (98% recovery) for S2 and 140 pg ml 1 (88% recovery) for S1 serum sample were measured with typical CV of 8.5% and 10% respectively over three parallel measurements. a b c
Volume per well/chip. Volume used for plate blocking. Blocking step.
can provide even lower non-specific binding and independence from the surface/material properties while preserving the biological activity.
Acknowledgment The authors would like to thank the University Medical Center Freiburg for providing serum samples and the European Commission for financial support under the framework of the Marie Curie Research and Trainings Network “Cellcheck” (MCRTN-CT-2006035854).
192
J. Horak et al. / Biosensors and Bioelectronics 58 (2014) 186–192
Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2014.02.058. References Beaudry, F., Vachon, P., 2006. Biomed. Chromatogr. 20, 1344–1350. Campbell, D.E., Raftery, N., Tustin Iii, R., Tustin, N.B., DeSilvio, M.L., Cnaan, A., Aye, P.P., Lackner, A.A., Douglas, S.D., 2006. Clin. Vaccine Immunol. 13, 1197–1203. DuPont, Vacrel (R) 8100 Quick Reference Processing Guide. Ekins, R., Edwards, P., 1997. Clin. Chem. 43, 1824–1831. Esser, P., 1991. Thermo Fisher Scientific Technical Bulletin, 10b. Findlay, J.W.A., Dillard, R.F., 2007. AAPS J. 9, 260–267. Harrison, S., Geppetti, P., 2001. Int. J. Biochem. Cell B 33, 555–576. Herron, J.N., Wang, H.K., Janatova¡, V., Durtschi, J.D., Christensen, D.A., Caldwell, K.D., Chang, I.N., Huang, S.C, 2003. Biopolymers at Interfaces, second ed., pp. 115–164. Horak, J., Dincer, C., Bakirci, H., Urban, G., 2014. Sens. Actuators B. 191, 813–820. Jang, M.U., Park, J.W., Kho, H.S., Chung, S.C., Chung, J.W., 2011. Oral Dis. 17, 187–193. Jobst, G., Gamp, T., 2010. US7691623 B2. Joyce, T.J., Yood, R.A., Carraway, R.E., 1993. J. Clin. Endocrinol. Metab. 77, 632. Kalinin, Y.V., Berejnov, V., Thorne, R.E., 2009. Langmuir 25, 5391–5397. Karasawa, T., Sugawara, M., 2005. Anal. Sci. 21, 1431–1436. Keller, B.O., Li, L., 2001. J. Am. Soc. Mass Spectr. 12, 1055–1063. Kling, J., 2006. Nat. Biotechnol. 24, 891–898. Lazarczyk, M., Matyja, E., Lipkowski, A., 2007. Folia Neuropathol. 45, 99. Leeman, S.E., Mroz, E.A., 1974. Life Sci. 15, 2033–2044.
Lesko, L.J., 2007. Clin. Pharmacol. Ther. 81, 807–816. Michaels, L.A., Ohene-Frempong, K., Zhao, H., Douglas, S.D., 1998. Blood. 92, 3148–3151. Mogensen, K.B., Kutter, J.P., 2009. Electroanalysis 30, 92–100. Nakajima, N., Ikada, Y., 1995. Bioconj. Chem. 6, 123–130. Niedermann, P., Berthou, H., Zwickl, S., Schönholzer, U., Meier, K., Gantner, C., KappSchwoerer, D., 2003. Microelectron. Eng. 67, 259–265. Nyberg, F., Le Greves, P., Terenius, L., 1985. Proc. Natl. Acad. Sci. USA 82, 3921. O'Connor, T.M., O'Connell, J., O'Brien, D.I., Goode, T., Bredin, C.P., Shanahan, F., 2004. J. Cell. Physiol. 201, 167–180. Parris, W.C.V., Sastry, B.V.R., Kambam, J.R., Naukam, R.J., Johnson, B.W., 1993. Ann. NY Acad. Sci. 694, 308–310. Phillips, T.M., 1996. Biomed. Chromatogr. 10, 331–336. Porstmann, T., Kiessig, S.T., 1992. J. Immunol. Methods 150, 5–21. Price, C.P., Newman, D.J., 1997. Principles and Practice of Immunoassay, Macmillan Reference Ltd, London. Pumera, M., Escarpa, A., 2009. Electrophoresis. 30, 3315–3323. Rissler, K., 1995. J. Chromatogr. B. 665, 233–270. Scott, S.A., 2013. Clin. Pharmacol. Ther. 93, 33–35. Sorger, P.K., 2008. Nat. Biotechnol. 26, 1345–1346. Suk, J.W., Cho, J.H., 2007. J. Micromech. Microeng. 17, N11. Taylor, J., Picelli, G., Harrison, D.J., 2001. Electrophoresis 22, 3699–3708. Wang, C., Yan, Q., Liu, H.B., Zhou, X.H., Xiao, S.J., 2011. Langmuir 27, 12058–12068. Wild, D., 2005. The Immunoassay Handbook. Elsevier Science Limited, Oxford. Yager, P., Edwards, T., Fu, E., Helton, K., Nelson, K., Tam, M.R., Weigl, B.H., 2006. Nature. 442, 412–418. Zimmermann, M., Delamarche, E., Wolf, M., Hunziker, P., 2005. Biomed. Microdevices 7, 99–110. Zubrzycka, M., Janecka, A., 2000. J. Neurophysiol. 34, 195–202.
