A PMMA microcapillary quantum dot linked immunosorbent assay (QLISA)

Biosensors and Bioelectronics 24 (2009) 3467–3474

Contents lists available at ScienceDirect

Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios

A PMMA microcapillary quantum dot linked immunosorbent assay (QLISA) Sundar Babu a , Sakya Mohapatra a , Leonid Zubkov a , Sreekant Murthy b , Elisabeth Papazoglou a,∗ a b

School of Biomedical Engineering Science & Health Systems, Drexel University, Philadelphia, PA 19104, United States College of Medicine, Drexel University, Philadelphia, PA 19104, United States

a r t i c l e

i n f o

Article history: Received 7 February 2009 Received in revised form 28 April 2009 Accepted 28 April 2009 Available online 6 May 2009 Keywords: Quantum dots ELISA MPO Inflammation IBD Microfluidic Biosensor QLISA PMMA capillary

a b s t r a c t The development of a simple and inexpensive quantum dot based immunoassay for detecting myeloperoxidase (MPO) in stool samples is reported (QLISA). The method developed utilizes readily available polymethylmethacrylate (PMMA) microcapillaries as substrates for performing the sandwich assay. High power (80 mW) and low power (10 mW) UV-LEDs were tested for their efficiency in maximizing detection sensitivity in a waveguide illumination or a side illumination mode. The results obtained indicate that both waveguide and side illumination modes can be employed for detecting MPO down to 15 ng/mL, however the high power LED in a side illumination mode improves sensitivity and simplifies the data acquisition process. The protocol and sensor robustness was evaluated with animal stool samples spiked with MPO and the results indicate that the sensitivity of detection is not compromised when used in stool samples. The effect of the ionic strength of the environment on the fluorescence stability of quantum dots was evaluated and found to affect the assay only if long imaging times are employed. Replacing the buffer with glycerol during imaging increased the fluorescence intensity of quantum dots while significantly minimized the loss in intensity even after 2 h. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Capillary based assays present several advantages over conventional 96 well plate methods, including small amount of analytes and proportionally less volume of required reagents. The confluence of developments in nano-fluidic handling systems has enabled capillary based microreactors and sensors to be employed in high throughput environments (Meldrum et al., 2000). However, the cylindrical nature of the capillary geometry does pose several challenges in the ability to properly couple and collect light (for colorimetric or fluorometric assays), thus limiting the final sensitivity of capillary based assay detection (Du et al., 2007). Some of these challenges can be addressed by more powerful and sensitive optics and so far capillary assays have achieved sensitivity of femtomolar detection (Halsall et al., 1988) with use of expensive, customized optical systems or electrochemical instruments. In this paper we report the design and implementation of an inexpensive, capillary based assay able to detect myeloperoxidase (MPO) at 100 pM sensitivity in a total volume of 1 ␮L. Free standing capillary tubes offer superior simplicity in manufacturing and handling compared to developing a full scale lab-on-a-chip type device. Capillaries have already been used as

∗ Corresponding author. E-mail address: [email protected] (E. Papazoglou). 0956-5663/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2009.04.043

immunosensors for detecting trace amounts of explosives (Narang et al., 1997, 1998), as high throughput automated genome analysis systems (Meldrum et al., 2000), as drug assays, for example in measuring paclitaxel in blood plasma (Sheikh et al., 2000), and even for the detection of helicobacter hepaticus (Thomas et al., 2007) that causes hepatitis in mice. The examples cited above are mostly immunoassays in conjunction with fluorescence spectroscopy. Immunoassays used for the detection of various biomolecules and biochemicals, rely on the interaction between an antigen and its antibody and possess high specificity depending on the antibody/antigen interaction. This specificity allows development of assays detecting multiple analytes in one capillary (Misiakos and Kakabakos, 1998). Enzyme linked immunosorbent assay (ELISA) is a common bioassay that relies on the antigen–antibody specificity and chemistry, with signal amplification capabilities. In a conventional ELISA technique sensing is mostly accomplished by chemiluminescence (Beumer et al., 1991), although both colorimetric titration (Esterbauer, 1996; Walenga and Fareed, 1994), and fluorescence can be used (Garvey et al., 1987; Savige et al., 1998; Smith and Eremin, 2008). Fluorescence based ELISA has the capability to detect more than one antigen or antibody by multiplexing. Multiplexing, although an elegant way to detect multiple markers, has remained thus far a challenge due to bleed through in the emission bands, the requirement of multiple excitation and emission filter pairs, the low fluorescence life time of fluorophores and the need of high power light sources. In sophisticated flow cytometry (FACS) systems such bleed through has been corrected by software

3468

S. Babu et al. / Biosensors and Bioelectronics 24 (2009) 3467–3474

but this necessitates complicated data analysis and expensive instruments. Recent developments in quantum dots (QDs) enable significant reduction in photo bleaching due to the unique optical properties of these semiconductor nanocrystals. These properties include single excitation maxima irrespective of emission maxima, and narrow emission spectra which allow multiplexing without any bleed through. Quantum dots have found significant applications in biology especially in live cell imaging (Thurn et al., 2007) to follow and understand signaling pathways (Hernandez-Sanchez et al., 2006). Although commercially available QDs are expensive on a per lb basis, the ability to carry out reactions in nano to microliter volumes coupled with moderately sensitive CCD cameras or photon counters can produce a cost effective assay; the raw material costs per unit mass remain low due to the small amount of QDs necessary to carry out the assay. Both glass and polymer based capillaries have been used to carry out immunoassays. Specifically fused silica (Bange et al., 2005), polystyrene (Misiakos and Kakabakos, 1998), polymethylpentene (Mastichiadis et al., 2002), and polymethylmethacrylate (PMMA) (Petrou et al., 2002) have been used to fabricate capillary biosensors for biomarker detection in volumes ranging from 0.5 to 5 ␮L. Polymeric capillaries are of particular interest due to readily available functional groups on their surface offering an appropriate substrate for immobilizing antibodies or antigens. Furthermore, recent developments in photochemical methods (McCarley et al., 2005; Wei et al., 2005) of functionalizing polymeric materials provide compelling reasons to choose polymeric capillaries over fused silica or glass capillaries. Several strategies have been devised to detect low concentrations of antigens by solid phase immunoassay in capillaries, primarily focusing on excitation of fluorophores followed by collection of the emitted photons. One such approach takes advantage of the evanescent field at the interface of the polymer/liquid interface for exciting the fluorophores; this particular method requires the material of the capillary to function as a waveguide (Tao et al., 2007). The excellent optical properties of PMMA have allowed use of PMMA capillaries as waveguides and fabrication of sensors to measure optical rotation of chiral molecules (Cox et al., 2006). We are reporting here a low cost PMMA microcapillary biosensor using QDs as the fluorescent probe for detection of picomolar quantities of analytes, and demonstrating this capability by detecting myeloperoxidase. Our selection of PMMA was based on its optical properties and the capability to selectively functionalize its surface for antibody immobilization. Capillary dimensions of 250 ␮m I.D. and 2.5 cm long allow us to use a volume of ∼1 ␮L, and these capillaries are commercially available. The high quantum yield of QDs coupled with the ability to excite QDs that emit at different wavelengths with a single UV light guided our choice of reporter probes. The inexpensive capillary based immunofluorescent assay described below was used for detecting and estimating the concentration of myeloperoxidase, an inflammatory marker over-expressed in inflammatory diseases including those of the gastrointestinal tract. Our initial results indicate that it is indeed possible to develop a low cost, robust imunnofluorescence sensor capable of operating with 1–2 ␮L of analyte and detecting subnanomolar concentrations. The methodology and design used in this paper offer a platform approach for capillary immunoassay development where further improvements in sensitivity can be readily implemented. However, for many clinical applications the sensitivity demonstrated by our data could be adequate to distinguish diseased from healthy individuals. Improvements to increase sensitivity are possible both by chemistry optimization approaches as well as with more elaborate optics. The focus in this paper has been on a low cost easy to deploy assay. The two main reasons for using capillary instead of 96 well plates were; (1) to reduce the volume of sample used and the volume of QD-Ab conjugate from the

perspective of economics of the assay, (2) be able to multiplex and (3) be able to detect a biomarker from sources such as swab tests. The reduction in volume allows us to effectively utilize the superior optical properties of QDs without increasing the cost of the assay, and by successfully using QDs we can perform multiplexing with a relatively inexpensive optical set up. Such low quantity sampling methods are not feasible with current ELISA protocols. 2. Materials and methods Our approach for detecting MPO is a sandwich assay depicted in the flow chart of Fig. 1. Briefly, a polyclonal MPO antibody (pAb) is immobilized on the inner surface of a PMMA capillary (capture antibody). MPO (from the sample of interest) is captured by this pAb and immobilized on the surface. Addition of a QD-mAb complex allows MPO detection by fluorescence. 2.1. Capillary functionalization and MPO assay PMMA capillaries (pCaps) were selected for the development of this sensitive assay due to the readily available functional groups on the surface of PMMA and its excellent optical properties. Capillaries used in this study were obtained from Paradigm Optics Inc. pCaps (O.D. 500 ␮m, I.D. 250 ␮m) were cut into 3 cm long pieces before or after functionalization depending on the experiment and held straight using a custom built spring loaded holder. This eliminated the natural tendency of PMMA capillaries to “buckle”. Functionalization of pCaps was carried out by following alkaline ester hydrolysis of methacrylate, a method developed by Bai et al. (2006) for functionalizing PMMA. The method was modified slightly, 1N NaOH at 60 ◦ C was pumped through the pCap using a peristaltic pump (100 ␮L/min) for 1 h followed by washing with 1× PBS buffer (pH 7.4). This step hydrolyzes the acrylate ester group on the surface of the pCap resulting in COOH termination that is crucial for covalently bonding the MPO antibody to the inner walls of the capillary. A rabbit anti-human polyclonal MPO antibody was purchased from ABD Serotec, Raleigh, NC, USA. Functionalized pCaps were then treated with EDC/NHS (104.7 mM EDC 21.7 mM NHS) (McCarley et al., 2005) for 5 h followed by loading the MPO capture antibody using a concentration of 100 nM. Optimal immobilization of the polyclonal MPO antibody on the inner surface of the pCap was accomplished by incubation at 4 ◦ C for 16 h. Non-immobilized antibodies were then removed from the capillary by washing with a buffer containing 0.1% Tween and 0.03% sodium azide (purchased from Sigma–Aldrich) in 1× PBS at pH 7.4. Subsequently, a blocking buffer containing 2% FBS in 1× PBS buffer was introduced into the capillaries to reduce nonspecific binding of proteins, and excess blocking buffer was washed away with the same wash buffer. The desired analyte, 1 ␮L of pure MPO (LEE Biosolutions, St. Louis, Missouri, USA) or properly prepared animal sample was then introduced into the pAb immobilized capillaries with the aid of a Hamilton septum adapter and allowed to interact with the pAb for 1 h at room temperature followed by injection of wash buffer at a flow rate of 50 ␮L/min. Monoclonal anti-human Myeloperoxidase antibody (mAb) conjugated to amine terminated QDs (em = 605 nm, from Invitrogen) was used as the reporter molecule. Conjugation of QDs to mAb was carried out by following the protocol provided by Invitrogen. Both MPO and the mAb were purchased from Lee Biosolutions Inc. QD conjugated mAb (QD-Ab) at 100 nM concentration was then introduced into the pCap, incubated at room temperature for 1 h followed by washing with the wash buffer. The intensity of the QD-Ab from the capillaries was obtained by the optical set up shown in Fig. 2b and described in detail below. Several capillaries were imaged at various steps of the process to evaluate and optimize the immobilization

S. Babu et al. / Biosensors and Bioelectronics 24 (2009) 3467–3474

3469

Fig. 1. Process flow chart of QLISA protocol MPO washing step we use 0.05% Tween.

and reaction conditions by fluorescence microscopy (Leica DMRX upright fluorescence microscope). Calibration curves were generated using MPO solutions of known concentrations and this allowed evaluation of the lowest detection limit (LDL), and establishing the sensitivity of the assay. The amount of fluid inside the capillary was restricted to 1–2 ␮L by limiting the length of the liquid plug inside the capillary. The QLISA assay was then carried out to determine its sensitivity and selectivity for detecting MPO in solution and in animal stool samples. Various steps involved in the QLISA protocol are summarized in Fig. 1. 2.2. Optical detection PMMA capillaries were inserted in a larger glass capillary for experiments where the capillaries acted as waveguides (Fig. 2a), and in this case a cylindrical mirror was used to collect the scattered fluorescent light. UV-LEDs rated at 10 and 80 mW optical power from Nichia Corporation were used as the excitation source and a bandpass filter (600 ± 20 nm) was used on the detection side to remove any UV signal from the excitation source. An aspheric lens (f = 6 mm) served the purpose of concentrating the UV light to a spot size of ∼1.5 mm while a separator mounted in front of the lens holder allowed us to position the capillary (irrespective of the experimental configuration) in the focal plane of the lens. A three axis manual positioner was used to align the mirror or the UV source depending on the experiment. The waveguide and side illumination mode are shown in Fig. 2a and b respectively. A 35 mm Focusable TECHSPEC® Double Gauss Macro Imaging Lenses from Edmund Optics was used for imaging the capillaries in both waveguide and side illumination mode. In order to increase the field of view spacer rings were used in the case of waveguide mode. In the case of waveguide illumination the distance between the capillary and the objective was ∼16 cm, whereas, in the case of side illumination the distance between the capillary and the objective was ∼6 mm. The ‘f’

Fig. 2. Schematic illustration of the configurations used for fluorescence signal collection from quantum dots inside the PMMA capillaries. (a) Waveguide mode: excitation of QDs achieved by coupling UV light through the ends of the capillary; light propagates through the walls of the capillary tube exciting the QDs within the evanescent field. (b) Side illumination mode: UV light is focused on the sides of the capillary (spot size, 1 mm) and imaged with a CCD camera positioned 90◦ to the UV source.

3470

S. Babu et al. / Biosensors and Bioelectronics 24 (2009) 3467–3474

number of the lens used in both configurations was 2.33. The only factor that differentiates the two modes was where the UV light was focused at. The angle between all the optical axes (UV source to capillary, UV source to CCD camera, capillary to CCD camera) was maintained at 90◦ . Capillaries containing relatively high concentrations (>1 nM) of MPO or calibration experiments with unconjugated QD solutions were imaged with a monochrome CCD camera (COHU 4900). A firewire monochrome CCD camera (Stingray, AVT-FS-033B) purchased from 1stVision Incorporated was used in all capillaries where the MPO concentration was lower than 1 nM. BMP images (640 × 480 pixels) were collected using a frame grabber (VCE-Pro, PCMCIA, Imprex Inc.). Images collected from the CCD were analyzed and resulted in quantification of the concentration of MPO based on the fluorescence intensity of the QD-MPO-Antibody. In the side illumination mode, a small region of interest (ROI) typically, 30 × 15 pixels was chosen at the geometrical center of the capillary and in the case of waveguide mode the ROI was set at 200 × 300 pixels. An average of the intensities of all the pixels within the ROI was obtained using ImageJ. The cylindrical mirror used in the waveguide mode allowed us to assign a larger ROI. 2.3. Preparation of animal samples Myeloperoxidase is present in stool samples of animals with symptoms mimicking the human irritable bowel disease (IBD) (Ahn et al., 2001; Deguchi et al., 2005; Karwa et al., 2007; Katada et al., 2008). MPO was extracted from stool samples by following the protocol previously established by Lettesjo et al. (2006). Briefly, weighed stool samples were digested with an extraction buffer (1× PBS, pH 7.4) supplemented with 12 mM EDTA, 1% fetal bovine serum (FBS), protease inhibitor (consisting of AEBSF 0.2 mM, E64 1.4 ␮M, bestatin 13 ␮M, leupeptin 0.09 ␮M, aprotinin 0.03 ␮M, EDTA 0.1 mM), 20% glycerol and 0.05% Tween 20, for 15 min at 4 ◦ C. The digestion step was followed by homogenization at 5000 rpm till a stable suspension could be obtained. Digestion of the homogenized sample was allowed to continue for 15 min at 4 ◦ C, followed by centrifugation at 14,000 rpm at 4 ◦ C for 30 min. The supernatant separated from the centrifuged sample was used in the reported data. Spiking experiments were carried out by adding 1.0 nM MPO to the stool extract.

the vacant sites on the inner surface of the pCap (Cho et al., 2002), while leaving the pAb sites for MPO to interact with. Problems related to nonspecific binding are not unique to pCaps, borosilicate or fused silica capillaries used in immunosensors have exhibited similar issues. Nonspecific binding of proteins and enzymes used in the assay can be minimized further by using surfactants during the washing steps. Bright fluorescent spots corresponding (Supplemental Fig. 2a) to nonspecific binding of QD-Abs to the inner walls of the pCap’s disappear (Supplemental Fig. 2b) when a buffer solution containing 0.1% Tween (surfactant). 3.2. Optimization of optical detection system Optimization of signal to noise as described in Figs. 3 and 5 was followed by determining the lowest detection limit of the QLISA protocol in pCaps. This was determined by capturing MPO (0.1–10 nM) in solution and is referred to as the assay sensitivity of detection. Single low power LED (LPLED) with max = 375 nm was used in a waveguide mode as illustrated in Fig. 2a. The antibody immobilization step was carried out with 100 nM pAb. Use of higher concentrations of pAb did not demonstrate any significant increase in the sensitivity level of MPO detection (data not shown). The highest concentration of QD-Abs used in all experiments was 100 nM; nonspecific binding is expected to be maximum at this concentration irrespective of the blocking step. Fig. 3a demonstrates a non-linear relationship between the concentrations of MPO in solution and the fluorescence intensity obtained using the capillary in a waveguide mode with the low

3. Results and discussion 3.1. Assay optimization Optical micrographs (Supplemental Fig. 1a–c) of capillaries after carrying out the entire QLISA protocol using varying concentrations of MPO depict homogenous fluorescence from the capillaries and most importantly absence of bright spots (indicating aggregation) even at increased concentrations of QDs. In the absence of the blocking agent bright spots, indicative of nonspecific binding, could be observed (Supplemental Fig. 1e and f). Regions of higher QD intensity in the absence of blocking agent could be expected because the functionalization of pCap leaves vacant sites that can contribute to nonspecific binding of Qd-Ab. As most of the biological moieties are charged species at any given pH, one can expect the target species to bind to the substrate via charge–charge interactions between the target molecule and the sensor substrate. In contrast to these charge–charge interactions, the antigen/antibody interaction is very specific, thus biosensor fabrication always includes a step to prevent this nonspecific binding. Functionalization of the capillaries by NaOH (see Fig. 1) results in COOH termination on the PMMA surface which is then used to covalently bind the pAbs to the inner walls of the capillary. Fetal bovine serum added to the blocking buffer binds nonspecifically to

Fig. 3. (a) Sensitivity of the QLISA method towards MPO. Five pCaps placed inside a glass capillary were used at each MPO concentration. Images were captured in the waveguide mode with the low power LED (10 mW) as the excitation light source. (b) Representative images of pCap with 0.3 nM MPO compared with a control pCap. Dark bands above and below the pCap are from the glass capillaries that were used to align the pCaps at the center of the cylindrical mirror. Bright bands spanning the height of the image are reflections of the fluorescence off the pCap’s. (c) Comparison of fluorescence intensities of QD solutions obtained with side illumination and waveguide modes. pCaps were inserted inside a glass capillary when imaging in the waveguide mode.

S. Babu et al. / Biosensors and Bioelectronics 24 (2009) 3467–3474

power LED (10 mW) as the excitation light source. Five pCaps were used at each MPO concentration. Detection down to 300 pM of MPO is possible, but the resolution of the system decreases at concentrations below 1 nM. The most encouraging aspect is that regardless of the loss in resolution, intensity values obtained at these low concentrations were still above the control at a statistically significant level (p-values for t-test are 0.05 and 0.0065 when comparing the control to 100 pM and the control to 300 pM respectively). Representative CCD images of pCaps used for detecting 0.3 nM of MPO and a control pCap are shown in Fig. 3b. An average intensity of QD fluorescence was calculated from image analysis of the bright bands within the capillaries. The dark regions above and below these bright bands represent the UV absorption of the glass capillary; these glass capillaries were used to align the pCaps in the focal plane of the cylindrical mirror. We believe that the loss in resolution at lower concentrations stems mostly from limitations in the optical hardware, i.e. our CCD camera, an analog camera with maximum integration time (shutter speed) of 16 ms. In addition; at such low concentrations fluorescence measurements become more sensitive to any in homogeneities in QD coverage. Furthermore, since pCaps are not as rigid as glass capillaries, perfect alignment on the focal plane of the mirror and the excitation source becomes exceedingly difficult. To overcome this challenge, we used a larger glass capillary as a support for the pCap during imaging; pCaps were left protruding out slightly towards the excitation source in order to minimize interference from the glass capillary. Although this approach allowed us to test our system in the laboratory, it is impractical for implementation in a commercial setting as it can often take a long time to align the capillaries manually. Since our approach is focused on producing a scaleable, cost effective optical immunosensor based on quantum dots, we found that the waveguide approach is not optimal and under-utilizes the efficiency of quantum dots. Low power LEDs themselves do have inherent fluorescence which results in further loss in sensitivity. Side illumination mode (Fig. 2b) and high power LED (HPLED) could help in circumventing these problems and improve the sensitivity of the assay while minimizing if not eliminating the problems pertaining to optical alignment. The effect of the mode of excitation, i.e. side illumination vs. waveguide mode is shown in Fig. 3c, using a high power LED (80 mW) as the UV excitation source. pCaps were loaded with QD solutions of known concentration and imaged in both modes with a total volume of the liquid plug at ∼1 ␮L. The advantages of side illumination can be clearly seen although the difference diminishes at concentrations below 100 pM (0.1 nM). The loss in detection sensitivity by using the pCap as a waveguide in our system can be attributed to several factors. First and foremost being the losses due to misalignment, in the waveguide mode alignment of the capillary with respect to the focal plane of the mirror as well the UV source was carried out manually – which is prone to errors. Second factor is the non-optimized collection efficiency of fluorescent light from the QDs to our detector, fluorescence is emitted at 360◦ angle and we believe only a fraction of it is collected by the cylindrical mirror used in the study. Third factor contributing to the loss in sensitivity could be attributed to the reduced coupling efficiency at the PMMA–air interface where the photons from the UV source are pumped into the capillary. The quality of the surface morphology at the air/capillary interface where photons from the UV-LED are coupled determines the coupling efficiency. The capillaries used in the study were cut with a warm blade; this is necessary for minimizing any structural damage to the capillary but this result in a surface that is not perfectly flat – resulting in reduced coupling efficiency. There is also the possibility for complex coupling to occur where both the capillary walls and also the liquid plug to aid in propagation of the photons (Dhadwal et al., 2004) through the capillary, in which case the sensitivity should

3471

have increased significantly. However, the results obtained indicate that complex coupling is probably not present in our design, perhaps due to presence of meniscus of the buffer which reflects most of the photons away from the liquid plug. In the case of side illumination mode, only a smaller cross section is illuminated at a given point of time. This enables proper excitation of all the QDs only within that small region. Therefore, the difference observed could also be due to the reduction in the detection volume (1.2 ␮L in the case of waveguide mode versus 50–75 nL in the case of side illumination). This is in accordance with the theory brought up by Ruckstuhl et al. (2000). Furthermore, because the QD binds to an immobilized antibody covered surface, the QDs are always within a distance less than 300 nm from the surface of the capillary. A study by Wang et al. (2007) demonstrated using atomic force microscopy (AFM) that the surface roughness of a PMMA surface (assumed as the height of the immobilized molecule) immobilized with antibody varied between 5.7 and 7.9 nm. Therefore, in the case of a sandwich assay we can estimate the height to be between 18 and 25 nm for the antibody–antigen–antibody complex, combined with the QD’s (particle size 6–10 nm) one can estimate the total height of the (antibody–antigen–QD attached to antibody) complex to be around 30 nm. However, steric hindrance, conformational changes, and orientation of polyclonal antibody with respect to the PMMA surface may reduce the height to less than 30 nm. This would reduce the sensitivity in the waveguide mode, where it is imperative that the QDs are within 30–300 nm of the waveguide surface. 3.3. Stability of QD signal Fig. 4a and b compares the effect of ionic strength on the intensity of QD and the stability of the fluorescence signal over a period of time. MPO in solution (0.5 nM) was used in this experiment and the effect of ionic strength on the intensity of the QDs was investigated by replacing the wash buffer with glycerol. Increase in fluorescence intensity when storage buffer (Tris-buffered saline, TBS) is replaced with glycerol could be observed (Fig. 4a). The photo stability of QDs and hence their fluorescence yield depends heavily on their local environment (Ruedas-Rama and Hall, 2008) especially the ionic strength (Qu and Morais, 2001). This effect has been utilized to fabricate optical metal ion sensors and intracellular pH sensors (Liu et al., 2007). Therefore, replacing the buffer with a non-ionic solution such as glycerol increases the signal to noise ratio, as observed in Fig. 4, by two possible means. First, with glycerol the system is deprived of ions that seem to result in destabilizing QDs, and therefore the maximum attainable quantum yield of QDs can be attained. Second, scattering is minimized by confining more of the photons in the walls of the capillary because the refractive index of glycerol (1.398) is closer to that of PMMA (1.491) than the storage buffer (∼1.33). As expected, pCaps with glycerol display a marked increase in intensity probably due to a combined effect of increased photon coupling and its role in stabilizing quantum dots fluorescence itself. In order to substantiate that ionic strength of the local environment of QDs is the primary reason for the difference in fluorescence intensity between the two liquids fluorescence intensities from images of pCaps with storage buffer and with glycerol were collected over a period of 2 h and compared. Fig. 4b compares the effect of ionic strength on the stability of QD fluorescence over a period of 2 h. A steady decline in signal intensity was observed when QDs were exposed to an environment of high ionic strength (TBS), leading to a loss in intensity of nearly 50%. However, replacing the wash buffer inside the capillary with glycerol, resulted in minimal loss of signal intensity (15% loss versus 50% loss). In both environments there is a fast decay in signal intensity within the first 60 min. This particular experiment was conducted

3472

S. Babu et al. / Biosensors and Bioelectronics 24 (2009) 3467–3474

loss in fluorescence intensity over a period of time. We have not observed any significant influence of the LED power variation on the results obtained because the LEDs were turned on just before each measurement and kept it in off-mode between successive measurements. A simple study on the stability of the UV-LED in continuously on mode reveals that the intensity varies within 5% of the power at t = 0. However, we do acknowledge that the stability of the UV-LED is vital for translating the QLISA protocol from “lab-to-bed” side. 3.4. MPO detection limit Fig. 5a depicts the performance of the assay in detecting MPO at picomolar concentrations using side illumination. Fluorescence intensity at various locations on a capillary was collected by moving the excitation source that was mounted on a translation stage. This allowed us to collect at a rapid rate statistically reliable data from individual capillaries mounted on a custom made spring loaded sample holder, which kept the capillary stretched and aligned. Representative CCD images (Fig. 5b – 100 pM MPO bound capillaries) show improved signal to noise ratio and spot free images that improve the reliability of the data acquisition method. These images were captured at various locations on a capillary and demonstrate uniformity of the fluorescence signal. The bands that appear in the CCD images are result of the architecture of the UV-LED itself. Fig. 5c summarizes the change in fluorescence intensity as a function of MPO concentration in both systems and demonstrates clearly the advantage of the side illumination method. The intensity of control capillaries, i.e. capillaries filled with buffer but containing no QDs, was subtracted from the intensity of pCaps to yield the data used in Fig. 5c. Based on the results obtained it is possible to conclude that the sensitivity of the QLISA protocol is 100 pM while the sensitivity of the entire optical system in detecting QDs in solution is 50 pM. 3.5. Animal samples

Fig. 4. (a) Effect of storage buffer on the fluorescence intensity of QDs. The increased intensity obtained by replacing the polar storage buffer with glycerol indicates the improved stability of QDs and the higher light coupling efficiency due to the refractive index matching between PMMA and glycerol (b) Effect of storage buffer on QD stability over a period of time. Continuous loss of fluorescence is minimized when the polar storage buffer is replaced with glycerol, thus demonstrating the effect of ionic strength on the fluorescence stability of QDs.

by collecting images every 5 min and the QDs were excited only during image acquisition mode (∼15–20 s). This ensures that the QDs or the pCaps were not heated by the excitation source during the course of the experiment. These two experiments were performed in succession, and the reasons for such sudden drop in intensity could be attributed to (1) desorption of antigen and antibody (2) previously unknown factors influencing the photostability of QDs themselves resulting in rapid loss in fluorescence intensity. It is indeed intriguing and counter intuitive that the fluorescence intensity would change in such abrupt manner, such observation has led us to develop a separate investigation on the stability of QD’s themselves. Replacing the storage buffer provides a critical advantage for the reliability of the assay, as shown in Fig. 4. Although there is significant loss in intensity, it must be noted that the loss occurs after ∼45 min. The data reported here were obtained within 5–10 min after the final washing step; therefore the data reported are not influenced by this loss in intensity. We have also investigated the influence of the stability of the UV-LED itself on the

The next step in validating the MPO bioassay developed was to test its performance in a more complex system simulating clinical samples. Presence of various biologically relevant moieties that are present in the animal stool samples is a major concern in developing a sandwich assay since the entire assay relies on the specificity and cross reactivity of the polyclonal (capture antibody) and the monoclonal antibody (reporter molecule) towards the antigen. Sensitivity of the QLISA protocol thus depends on the sensitivity and specificity towards one another in the pAb/MPO/QD-Ab sandwich. Although chemistry optimization steps were taken to minimize nonspecific interactions between the sensor substrate and the analyte, it is imperative that the robustness of the protocol and the device be evaluated with actual samples rather than solutions of the antigen. The QLISA protocol was tested in spiked animal stool samples to evaluate its ability to detect MPO at trace levels in biological samples. Stool samples from disease-free mice were collected and prepared as described in Section 2. The extract obtained served the purpose of being the control and MPO in solution (1.0 nM) was added to the extract (spiking) from the stool samples. The stool sample extract was spiked such that the final concentration of MPO in the extract was 1, 0.5 and 0.1 nM respectively. In order to get an extract with 1 nM external MPO, 10 ␮L of external MPO (10 nM) was added to 90 ␮L of the stool extract and so on. In this mouse model (Karwa et al., 2007), before inducing disease, the animals bear no MPO in their stools. Intensity values of MPO obtained from the spiked stool extracts were then compared with the intensities from MPO solution. This served as an appropriate test bed for understanding and identifying influence from unknown variables (mostly nonspecific binding) that can result in false positive or false negative results. Fig. 6 shows the fluorescence intensity data from stool samples collected from the animals along with data from MPO in

S. Babu et al. / Biosensors and Bioelectronics 24 (2009) 3467–3474

3473

Fig. 5. (a) Fluorescence intensity at various MPO concentrations obtained by side illumination. The lines inside the circles are the error bars. (b) CCD image of control and 100 pM MPO at three different spots and (c) Comparison of side illumination with waveguide geometry: intensity values are corrected by the control intensity (after subtracting the signal from the capillaries filled with QD free buffer).

4. Conclusions A polymer based capillary assay has been developed that relies on the fluorescence intensity of quantum dots to detect picomolar quantities in microliter volumes. The device and protocol (QLISA) developed were tested with myeloperoxidase, an antigen that is over-expressed in inflammatory conditions. Two different modes of exciting the quantum dots, either using the capillary as the waveguide or using side illumination were tested and it was determined that side illumination eliminates problems pertaining to optical alignment and is better suited for a high throughput bioassay. Results obtained indicate that polymeric capillaries are certainly suitable for optical immunosensor fabrication and it is possible to fabricate a cost effective biosensor with off the shelf components. The device built has a lowest detection limit of 100 pM towards MPO (∼15 ng/mL). The stability of QDs in the capillaries is found to be affected by the ionic strength of their local environment, and replacing the buffer with glycerol improved their stability.

Fig. 6. Fluorescence intensity from spiked animal stool samples. Fluorescence intensity values obtained from spiked animals were compared with fluorescence intensities of MPO in solution. The fluorescence intensity at 0 MPO concentration is from pCaps that were put through the QLISA protocol without MPO.

solution (standard curve). Spiked stool samples exhibit response that is similar to that of MPO in solution, illustrating the specificity of the antigen–antibody complex and the robustness of the QLISA protocol. The intensity value obtained from the stool sample that does not contain any MPO is essentially the same as the control pCap of the MPO solution set, indicating the absence of nonspecific interaction between the capture antibody and/or the mAb. This was further confirmed by the t-test results, p > 0.994, which indicate that the two data sets are identical. The spiked animal data can be considered as an encouraging first step towards using this bioassay in a full disease model to quantify the presence of MPO in stools and its correlation to clinical disease activity indices (Karwa et al., 2007).

Acknowledgement The authors gratefully acknowledge support of this research by the Wallace H. Coulter foundation. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.bios.2009.04.043. References Ahn, B.O., Ko, K.H., Oh, T.Y., Cho, H., Kim, W.B., Lee, K.J., Cho, S.W., Hahm, K.B., 2001. Int. J. Colorectal Dis. 16 (3), 174–181. Bai, Y.L., Koh, C.G., Boreman, M., Juang, Y.J., Tang, I.C., Lee, L.J., Yang, S.T., 2006. Langmuir 22 (22), 9458–9467. Bange, A., Halsall, H.B., Heineman, W.R., 2005. Biosens. Bioelectron. 20 (12), 2488–2503.

3474

S. Babu et al. / Biosensors and Bioelectronics 24 (2009) 3467–3474

Beumer, R.R., Brinkman, E., Rombouts, F.M., 1991. Int. J. Food Microbiol. 12 (4), 363–374. Cho, H.J., Entz, S.C., Green, G.T., Jordan, L.T., 2002. Can. Vet. J. 43 (1), 43–45. Cox, F.M., Argyros, A., Large, M.C.J., 2006. Opt. Express 14 (9), 4135–4140. Deguchi, Y., Andoh, A., Inatomi, O., Araki, Y., Hata, K., Tsujikawa, T., Kitoh, K., Fujiyama, Y., 2005. Int. J. Mol. Med. 16 (4), 605–608. Dhadwal, H.S., Kemp, P., Aller, J., Dantzler, M.M., 2004. Anal. Chim. Acta 501 (2), 205–217. Du, W., Chen, S., Chen, Y.Z., Chen, Z., Luo, Q.M., Liu, B.F., 2007. J. Sep. Sci. 30 (6), 906–915. Esterbauer, H., 1996. Pathol. Biol. (Paris) 44 (1), 25–28. Garvey, J.S., Thomas, D.G., Linton, H.J., 1987. Exp. Suppl. 52, 335–342. Halsall, H.B., Heineman, W.R., Jenkins, S.H., 1988. Clin. Chem. 34 (9), 1701–1702. Hernandez-Sanchez, B.A., Boyle, T.J., Lambert, T.N., Daniel-Taylor, S.D., Oliver, J.M., Wilson, B.S., Lidke, D.S., Andrews, N.L., 2006. IEEE Trans. Nanobiosci. 5 (4), 222–230. Karwa, A., Papazoglou, E., Pourrezaei, K., Tyagi, S., Murthy, S., 2007. Inflamm. Res. 56 (12), 502–510. Katada, K., Yoshida, N., Suzuki, T., Okuda, T., Mizushima, K., Takagi, T., Ichikawa, H., Naito, Y., Cepinskas, G., Yoshikawa, T., 2008. Inflamm. Res. 57 (2), 84–91. Lettesjo, H., Hansson, T., Peterson, C., Ung, K.A., Ringstrom, G., Abrahamsson, H., Simren, M., 2006. Scand. J. Gastroenterol. 41 (1), 54–59. Liu, Y.S., Sun, Y.H., Vernier, P.T., Liang, C.H., Chong, S.Y.C., Gundersen, M.A., 2007. J. Phys. Chem. C 111 (7), 2872–2878. Mastichiadis, C., Kakabakos, S.E., Christofidis, I., Koupparis, M.A., Willetts, C., Misiako, K., 2002. Anal. Chem. 74 (23), 6064–6072.

View publication stats

McCarley, R.L., Vaidya, B., Wei, S.Y., Smith, A.F., Patel, A.B., Feng, J., Murphy, M.C., Soper, S.A., 2005. J. Am. Chem. Soc. 127 (3), 842–843. Meldrum, D.R., Evensen, H.T., Pence, W.H., Moody, S.E., Cunningham, D.L., Wiktor, P.J., 2000. Genome Res. 10 (1), 95–104. Misiakos, K., Kakabakos, S.E., 1998. Biosens. Bioelectron. 13 (7–8), 825–830. Narang, U., Gauger, P.R., Ligler, F.S., 1997. Anal. Chem. 69 (10), 1961–1964. Narang, U., Gauger, P.R., Kusterbeck, A.W., Ligler, F.S., 1998. Anal. Biochem. 255 (1), 13–19. Petrou, P.S., Kakabakos, S.E., Christofidis, I., Argitis, P., Misiakos, K., 2002. Biosens. Bioelectron. 17 (4), 261–268. Qu, F.Y., Morais, P.C., 2001. IEEE Trans. Magn. 37 (4), 2654–2656. Ruckstuhl, T., Enderlein, J., Jung, S., Seeger, S., 2000. Anal. Chem. 72 (9), 2117–2123. Ruedas-Rama, M.J., Hall, E.A.H., 2008. Analyst 133 (11), 1556–1566. Savige, J.A., Paspaliaris, B., Silvestrini, R., Davies, D., Nikoloutsopoulos, T., Sturgess, A., Neil, J., Pollock, W., Dunster, K., Hendle, M., 1998. J. Clin. Pathol. 51 (8), 568–575. Sheikh, S.H., Abela, B.A., Mulchandani, A., 2000. Anal. Biochem. 283 (1), 33–38. Smith, D.S., Eremin, S.A., 2008. Anal. Bioanal. Chem. 391 (5), 1499–1507. Tao, S.Q., Gong, S.D., Fanguy, J.C., Hu, X.M., 2007. Sens. Actuators B Chem. 120 (2), 724–731. Thomas, T., Riley, L., Grant, D., Grant, S.A., 2007. Sens. Lett. 5 (2), 398–404. Thurn, K.T., Brown, E.M.B., Wu, A., Vogt, S., Lai, B., Maser, J., Paunesku, T., Woloschak, G.E., 2007. Nanoscale Res. Lett. 2 (9), 430–441. Walenga, J.M., Fareed, J., 1994. Clin. Lab. Med. 14 (4), 709–728. Wang, B., Oleschuk, R.D., Petkovich, P.M., Horton, J.H., 2007. Colloids Surf. B: Biointerfaces 55 (1), 107–114. Wei, S.Y., Vaidya, B., Patel, A.B., Soper, S.A., McCarley, R.L., 2005. J. Phys. Chem. B 109 (35), 16988–16996.