Biomed Microdevices (2014) 16:771–777 DOI 10.1007/s10544-014-9881-1
A nanostructured aluminum oxide-based microfluidic device for enhancing immunoassay’s fluorescence and detection sensitivity Xiang Li & Haocheng Yin & Long Que
Published online: 22 June 2014 # Springer Science+Business Media New York 2014
Abstract A nanostructured aluminum oxide (NAO)-based fluorescence biosensing platform with a programmable sample delivery microfluidic interface is reported. The NAObased fluorescence sensor can tremendously enhance the fluorescence signals, typically up to 100×or more, over the glass substrate. The programmable sample delivery microfluidic interface, which is integrated with the NAO-based sensors, can automatically generate and deliver a series of different concentrations of the biological samples to each individual sensor. Hence it can facilitate the fluorescence-based biodetection and analysis for high throughput applications. Using Protein A and fluorophore-labeled Immunoglobulin G (IgG) as models, the binding between them on this platform have been demonstrated. It has been shown that the IgG of programmable concentrations can be delivered to individual sensor using the microfluidic interface and confirmed by the fluorescence images. Using current NAO-based fluorescence sensors without any optimization, the detectable concentration of IgG can be as low as 20 pg/mm2 using a conventional fluorescence microscope. Due to its inexpensive fabrication process, this technology could provide a disposable technical platform for fluorescence-based sensing and analysis.
Keywords Fluorescence sensor . Nanostructured aluminum oxide . Programmable sample concentration generation . Immunoassay
Electronic supplementary material The online version of this article (doi:10.1007/s10544-014-9881-1) contains supplementary material, which is available to authorized users. X. Li : H. Yin : L. Que (*) Institute for Micromanufacturing, Louisiana Tech University, 911 Hergot Avenue, Ruston, LA 71270, USA e-mail: [email protected]
1 Introduction Fluorescent immunoassay is one of the most widely utilized techniques in protein detection since the fluorescence technique can provide very high sensitivity and large multiplexing capabilities (Georganopoulou and Mirkin, 2009, Yager et al., 2006, Lakowicz, 2006, Darvill et al., 2013). However, it is still very important to further increase the sensitivity of fluorescence detection method since the higher sensitivity may offer the possibility for disease diagnosis at the early stage, and can also significantly reduce the consumption of the expensive biological samples in the experiments. Development of the advanced nanostructured substrate is one way to increase the sensitivity of fluorescence detection. In the past decades, some fluorescent immunoassay sensors based on metal-enhanced fluorescence (MEF) technology have been developed and reported. Examples include a variety of metallic nanoparticles (Xie et al., 2013), D2PA plate (Zhou et al., 2012), and tunable nanoplasmonic resonator (Wang et al., 2011)-based biosensors. Recently it has been reported by our group that the nanostructured aluminum oxide (NAO) substrate can increase the fluorescence signals up to two orders of magnitude for fluorescent dyes and labeled biomolecules compared to the glass substrate (Li et al., 2012a, Li et al., 2012b, Li et al., 2013). Different from MEF-based technology, the NAObased technology not only can be fabricated in an arrayed format in a cost-effective manner without using any top-down nanofabrication process, but also does not require a layer of dielectric material spacer between the sensing surface and fluorophores to avoid the quenching effect. For a variety of biological and biomedical applications, different concentrations of biological samples may be required for fluorescence detection and analysis. Ideally, a fluorescence technical platform, especially for high throughput applications, can not only offer the high sensitivity but also have an interface for simple sample delivery. For instance, in DNA or protein microarray
772
technology, the DNA or protein samples (i.e., the fluorescence labeled targets) usually are delivered to the desired locations (i.e. the fixed probes) on a solid substrate by utilizing an array of fine pins or needles, which are controlled by a robotic arm that is dipped into wells containing DNA probes and then deposit each probe at designated locations in an arrayed format. The resulting probes, representing the nucleic acid profiles, are ready to receive complementary targets derived from experimental or clinical samples (Pollack et al., 1999). This procedure is quite cumbersome, time-consuming, and expensive, especially for delivering a series of varied concentrations of biological samples to the fixed probes. Hence, use of an automatic varied concentration generating and delivering interface of biological samples will be a significantly favorable option. Herein, a microfluidic NAO fluorescence sensing platform with a programmable sample delivery microfluidic interface is reported. Specifically, this platform integrates arrayed fluorescence sensors using NAO as the fluorescence enhancement substrates, and a microfluidic concentration generator of the biological samples on a single chip. These functions make it possible to develop disposable arrayed fluorescence biosensors for high-throughput applications. Fig. 1a gives a schematic of arrayed fluorescence sensors, and a photo of the fabricated arrayed sensors is shown in Fig. 1b. Each sensor consists of a polydimethylsiloxane (PDMS) microfluidic chamber and a micropatterned NAO on indium tin oxide (ITO) glass substrate. The NAO micropattern is a microlines-pattern, which is located inside a PDMS chamber. The arrayed sensors are connected to a microfluidic network, which can generate and thus provide programmable concentrations of fluorophore-labeled proteins to the sensors. A SEM image of the micropatterned NAO is given in Fig. 1c, which is fabricated using one-step anodization process (Li et al., 2013). For technical demonstration, Protein A is used as a capture probe, while fluorophore-labeled IgG as a target as schematically shown in Fig. 1d.
2 Methods and materials 2.1 Fabrication of arrayed microfluidic fluorescence sensors The fabrication process flow is illustrated in Fig. 1S in the supplementary material (Zhang et al., 2012, He et al., 2012, He et al., 2014, Yin et al., 2014). Briefly, this process starts from an ITO glass substrate. A lift-off process is used to form the Al patterns. As a result, Al patterns are formed and connected with each other with Al lines as shown in Fig. 1Sb. Once the Al patterns have been fabricated and cleaned with acetone and DI water, one-step anodization using oxalic acid is then carried out to form the NAO as shown in Fig. 1Sc. The resulting thickness of the NAO is 2.7 μm. A
Biomed Microdevices (2014) 16:771–777
PDMS microfluidic chip is fabricated separately using a soft lithography process, then it is bonded with the NAO-glass chip, followed by assembling input and output tubing. The input tubes are connected with syringes controlled by a syringe pump, while the output tube leads to a biochemical waste collecting beaker. 2.2 Operation of the microfluidic chip for generating programmable concentrations The microfluidic chip for generating programmable concentrations of a sample is modified from previous work (Jeon et al., 2000). It has three inputs for flowing liquid solutions (i.e., biological samples, etc.) into the chip. As illustrated in Fig. 2a-c, symmetric flow is achieved by flowing the samples through the middle input (M-input), and flowing DI water or PBS buffer through the left input (L-input) and right input (Rinput). Asymmetric flow is achieved by flowing the samples through either the L-input or the R-input, and flowing DI water or PBS buffer through the M-input and either the Rinput or L-input. If the flowing rates of the sample/water are the same in all three inputs, the concentration profile at the outputs (1, 2, 3, 4, 5) can be predicted by modeling the pyramidal microfluidic network as an equivalent electronic circuit (Jeon et al., 2000). The symmetrical flow scheme results in a symmetrical sample-concentration profile at the outputs with the highest in the middle output channel 3. All the achieved concentrations at the outputs are lower than the original concentration of the sample. The asymmetrical flow scheme provides an alternative for providing not only the original sample concentration through one output, but also different concentrations at other outputs. Hence, depending on the concentration requirements, these two schemes can be utilized alone or combined. In addition, the concentration at each output channel can be accurately determined, which is very useful for quantitative analysis of bioreaction. Two different concentration profiles resulting from these two different flowing schemes have been experimentally demonstrated by flowing blue food dye (McCormick, Inc) as the simple model through one input and water through other two inputs in Fig. 2d-e. The flowing rates through all three inputs are the same. With the same concentration of blue food dye for these two flowing schemes, as expected, the symmetric flowing scheme shows the symmetric concentration profile with the highest concentration at the middle output 3, while the asymmetric flow scheme gives the highest concentration at the leftmost output 1. Note that more different concentrations of the sample can be achieved by cascading more levels of the fluidic network. In all the following experiments, the biochemical samples and buffer solution are transported to the microdevice through the assembled plastic tubing (Upchurch Scientific, Inc.) by a syringe controlled by a syringe pump (KD Scientific, Inc.).
Biomed Microdevices (2014) 16:771–777
773
Fig. 1 a 3-D sketch of the fluorescence sensors with a programmable sample concentration generating and delivery microfluidic interface; (b) photo of a fabricated device; (c) SEM image of NAO inside the fluorescence sensor; (d) sketch of the immunoassay on the fluorescence sensor
2.3 Chemicals and materials The immunoassay reagents used in the experiments include Protein A (Pierce Biotechnology, Inc), buffer solution phosphate buffered saline (PBS) (Sigma-Aldrich, Inc), SEA BLOCK blocking buffer (Pierce Biotechnology, Inc). SEA BLOCK blocking buffer contains steelhead salmon serum in PBS and 0.1 % sodium azide. It does not interact significantly with mammalian antibodies, and thus results in minimal or no background. Deionized (DI) water was obtained from a DI water purification system (Millipore, FRANCE). Human Immunoglobulin G (IgG) labeled with FITC was purchased from Rockland Immunochemicals Inc. and was diluted using PBS (pH=7.2) solution at a concentration of 200 μg/mL. Rabbit Immunoglobulin G (IgG) labeled with Rhodamine was purchased from Rockland Immunochemicals Inc. and Fig. 2 Operation of the microfluidic chip: (a) 2-D sketch of the fluorescence sensors; (b) close-up of one fluorescence sensor, showing the NAO microline-pattern inside the sensor; (c) close-up of the programmable sample concentration generating region with output channel 1, 2, 3, 4, 5; (d) experimental concentration profile of blue food dye at the outputs under symmetric flow scheme; (e) experimental concentration profile of blue food dye under asymmetric flow scheme
was diluted using PBS (pH=7.2) solution at a concentration of 200 μg/mL. 2.4 Fluorescence biodetection procedure For technical demonstrations, Protein A is used as a capture protein while IgG labeled with FITC or Rhodamine as a target. Sea Block is used as the blocking layer for preventing nonspecific binding. This bioassay is very selective and only specific between Protein A and IgG, and hence has usually been used as a simple model for immunoassay (Zhang et al., 2010, Zhang et al., 2011, Zhou et al., 2012). The detailed procedure for the immunoassay is schematically illustrated in Fig. 3a. Briefly, first, Protein A is immobilized on the NAO surface, followed by rigorous rinsing by PBS to remove the unbound Protein A; then Sea Block is applied to occupy the
774
unbound sites on the NAO surface, followed by rigorous rinsing by PBS. This step is for minimizing the non-specific binding. Finally, fluorophore-labeled IgG is applied and bound with Protein A, followed by gentle rinsing by PBS. Thereafter, the fluorescence images are obtained by a fluorescence inverted microscope equipped with a DP71 camera (Olympus, Inc). A representative optical image and the corresponding fluorescence image with IgG concentration of 1.33 μM are given in Fig. 3b. The NAO microline width is ~15 μm. The integration time for obtaining the fluorescence images is 500 ms. It is clear that the fluorescence images of the IgG on the NAO microlines have been enhanced dramatically compared to those on the glass substrate between the NAO microlines. 2.5 Fluorescence measurement and experimental data analysis The fluorescence images are obtained using a fluorescence microscope (Olympus, Inc). The built-in black balance function/method in the fluorescence microscope is used to improve the contrast of the fluorescence image. The relative intensity of the fluorescence is directly obtained from the images. A MatLab program based on the Imaging Processing toolbox in MatLab has been written to read the files of fluorescence images, which are then converted to gray scale images from the color images (Li et al., 2012a). A horizontal cutline is obtained through the fluorescence image and the corresponding intensity is obtained and then plotted. The fluorescence detection limit is determined by comparing the fluorescence signals from the NAO substrates immobilized with fluorophore-labeled IgG and those from the bare NAO substrate. Specifically, the detection limit of IgG is defined as its concentration at which the fluorescence signals are at least 3-fold larger than those from bare NAO substrate (Zhou et al., 2012).
3 Results and discussion In Fig. 4, two representative different concentration profiles of the IgG have been generated and then delivered to the fluorescence sensors, thereby resulting in different fluorescence imaging profiles. In both cases, the flowing rates for IgG and PBS buffer are kept same at 0.3 μL/min through the three inputs (i.e. L-input, M-input and R-input). Using the symmetric flowing mechanism in Fig. 4a-1 and the asymmetric flow mechanism in Fig. 4b-1, different concentrations of IgG are generated and then bound to the Protein A. Fluorescence images of the arrayed sensors under symmetric flowing scheme are shown in Fig. 4a-2. Rhodaminelabeled rabbit IgG is used in this demonstration. The fluorescence intensity varies with different concentrations of IgG. As
Biomed Microdevices (2014) 16:771–777
expected, the sensor in the middle connected to channel 3 has the highest fluorescence intensity, while the fluorescence intensities of other four sensors reduce symmetrically around the middle sensor to essentially negligible. The fluorescence images of the arrayed sensors under asymmetric flowing scheme are shown in Fig. 4b-2. FITC-labeled human IgG is used for this test. As shown, the fluorescence intensities decrease from the left sensor, connected to channel 1, to the right one, connected to channel 5. Specifically, the sensor connected to channel 1 has the highest fluorescence intensity. The fluorescence intensities of the two sensors, connected to channel 4 and 5, are found identical to that of the blank NAO sensor. The very low fluorescence emission from these two sensors is just the scattering background from the NAO. Since negligible IgG has been flowed to channel 4 and 5, there is essentially no IgG applied to these two sensors. The fluorescence intensity on each sensor has been first obtained by subtracting the NAO background scattering intensity, then normalized to the fluorescence intensity of the IgG at its original concentration (i. e., the concentration of IgG before flowing into the device) on an identical sensor. Based on the measured fluorescence intensity, the normalized IgG concentration for each channel can be obtained as shown in Fig. 4a-3 and b-3, which is very consistent (±5 % offset) with the prediction by the equivalent electronic circuit model (Jeon et al., 2000). These experiments indicate the possibility to quantify the bio-samples applied on the sensors. Note that the resulting concentration profiles of IgG can also be easily tuned by simply changing the flowing rates of the samples (i. e., IgG) and/or changing the input (i.e., among L-input, Minput, R-input) for the samples. Even though such concentration profiles of IgG cannot be easily predicted using the simple electronic circuit model, but they can still be determined by some simple trial experiments. As such, a variety of preknown concentration profiles of IgG can be generated and delivered to the fluorescence sensors, offering a great deal of flexibility for fluorescence-based bioassay, particularly useful for high-throughput applications. Furthermore, it is also possible to generate and deliver a specific/distinct concentration of the IgG to each single sensor by using a modified microfluidic interface, such as a combination of the microfluidic concentration generator with 2-D or 3-D microfluidic network of controlled valves and pumps (Melin and Quake, 2007). Note that Rhodamine-labeled rabbit IgG and FITC-labeled human IgG are used as two examples for the technical demonstrations. The NAO-based sensors can be potentially used for all fluorescence-based bioassay. In addition, the immunoassay has also been carried out on three types of sensors as shown in Fig. 5. In these experiments, the immunoassay has been carried out only on the upper part of the NAO microline-patterns. The NAO in the three sensors has been fabricated by different anodization time, which is 35 mins, 25 mins, and 15 mins in Fig. 5 a, b, and c, respectively.
Biomed Microdevices (2014) 16:771–777
775
Fig. 3 a Sketch showing the major steps for immunoassay; (b) representative optical image and the corresponding fluorescence image after completing the immunoassay on the sensor
Optical images of NAO microline-patterns of the three sensors are given in Fig. 5a-1, b-1, and c-1. The corresponding fluorescence images after immunoassay are shown in Fig. 5a-2, b2), and c-2, respectively. In order to enhance the image contrast, the fluorescence images are adjusted by the blank balance function provided by the software of the fluorescence microscope. Clearly, the fluorescence signals on NAO Fig. 4 Symmetric flow scheme: (a-1) the IgG is flowed through the M-input; (a-2) fluorescence images of the sensors with different concentrations of IgG; (a-3) the measured concentration profile of IgG delivered to each sensor based on the fluorescence intensity. Asymmetric flow scheme: (b-1) the IgG is flowed through the L-input; (b-2) fluorescence images of the sensors with different concentrations of IgG; (b-3) the measured concentration profile of IgG delivered to each sensor based on the fluorescence intensity
microline-patterns are much larger than those on the glass due to the fluorescence enhancement capability of the NAO. In order to quantitatively evaluate the IgG attached to the Protein A, the volume of the IgG solution is fixed at different concentrations and applied to the sensor surfaces. After an incubation time of 40 mins, the chip is rinsed gently by PBS. In these experiments, the applied concentration of IgG is
776
Biomed Microdevices (2014) 16:771–777
Fig. 5 Three fluorescence sensors with NAO microline-patterns fabricated using different anodization time: (a) 35 mins; (b) 25 mins; (c) 15 mins. Immunoassay carried out only on the upper part of the NAO
microline-patterns for the three sensors. From top row to bottom row: Optical images; the corresponding fluorescent images; and the plots of the fluorescence intensity. The width of the NAO microlines is ~15 μm
0.1 ng/mm2, which is calculated by IgG solution concentration (pg/mL)×volume (mL) of each fluorescence sensor/NAO area (mm2) inside a sensor. As shown in Fig. 5, the fluorescence images give the identical profiles to those of the NAO microline-pattern for each type of sensor. The plots of the fluorescence intensities of IgG on the three different sensors are given in Fig. 5a-3, b-3, and c-3, respectively. It is shown that using the same amount of IgG, the fluorescence intensity has little difference for all three types of sensors. In other words, the NAO substrates formed by the partially anodized Al (shorter anodization time) and the more anodized Al (longer anodization time) essentially have the similar fluorescence enhancement capability. It has been observed that some other parameters such as the thickness of the NAO can tune its enhancement capability. Experiments find that IgG at a concentration of 20 pg/mm2 can be detected on 5 μm thick NAO substrate. The origin of the fluorescence enhancement has been discussed in our previous work (Li et al., 2012a, Li et al., 2012b, Li et al., 2013). Briefly, the fluorescence enhancement is mainly due to the following primary reasons. First, the
surface scattering effect of the NAO causes the redistribution of the electromagnetic fields with high surface intensities, thus contributing to the fluorescence enhancement. Second, the nanoscale NAO grains may offer waveguide effect (Dorfman et al., 2006), resulting in evanescent field from the surface of the NAO grains, which play an important role in the enhancement of the fluorescence signals. Third, it is possible that the NAO nanostructure could concentrate a little more biomolecules to its surface. Given the optical properties of the NAO (Li et al., 2005), the scattering wavelength can be potentially tuned by modifying the nanopore size and the spacing among nanopores. Hence, it is potentially possible to tune the scattered optical wavelength to match the excitation wavelength of the fluorescent dyes, thereby further improving the fluorescence signals. It has been observed that the largest enhancement occurs at the NAO surface, and the enhancement drops greatly with the distance between the surface and the fluorescent dyes. Therefore, different from MEF-based fluorescence sensors, no space layer of dielectric material between the sensing surface and the fluorescent dyes is required for NAO-based fluorescence sensors to avoid the
Biomed Microdevices (2014) 16:771–777
quenching effect (Anger et al., 2006). This feature further simplifies the fluorescence-based bioassay. In addition, since arrayed sensors can be batch-fabricated in an inexpensive and efficient manner, a disposable fluorescence-based platform can be potentially developed for the multiplexed biomolecular detection and analysis.
4 Conclusions A nanostructured material-based fluorescence sensing platform with a programmable sample delivery microfluidic interface has been developed and reported. The fluorescence sensing platform is enabled by the NAO, which can dramatically enhance the fluorescence signals compared to the glass substrate. The programmable sample delivery microfluidic interface, which is integrated with the arrayed sensors, can generate and deliver different concentrations of the biological samples to each sensor, paving a way for achieving a multiplexed fluorescence-based biodetection and analysis. As a technical demonstration, the fluorescence signals and images for an immunoassay with Protein A and IgG labeled with Rhodamine or FITC have been measured. The delivery of programmable concentrations of IgG to each sensor by the platform has been achieved. It has been observed that 20 pg/ mm2 of IgG can be detected. Given the simple and inexpensive fabrication process of the sensor, and the feasibility of fabricating hundreds of sensors on a single chip, this technical platform may have broad applications in biological, medical, chemical and other research.
Acknowledgments The research is funded in part by a NSF grant.
777
References P. Anger, P. Bharadwaj, L. Novotny, Phys. Rev. Lett. 96, 1130021– 1130024 (2006) Darvill, D., Centeno, A., Xie, F., 2013. Phys. Chem. Chem. Phys, doi:10. 1039/C3CP50415H. A. Dorfman, N. Kumar, J. Hahm, Langmuir 22, 4890–4895 (2006) D. Georganopoulou, C. Mirkin, Nature 462, 461–464 (2009) Y. He, X. Li, L. Que, J. Nanosci. Nanotechnol. 12(10), 7915–7921 (2012) Y. He, X. Li, L. Que, J. Biomed. Nanotechnol. 10, 767–774 (2014) N. Jeon, S. Dertinger, D. Chiu, I. Choi, A. Stroock, G. Whitesides, Langmuir 16, 8311–8316 (2000) Lakowicz, 2006. Plasmonics, 1, 5–33. Li, X., He, Y., Zhang, T., Lee, T., Que, L., 2012a. Proceeding of IEEE NANO 2012, 10.1109/NANO.2012.6321926 doi:10.1109/NANO. 2012.6321926#blank X. Li, Y. He, T. Zhang, L. Que, Opt. Express 20(19), 21272–21277 (2012b) X. Li, Y. He, L. Que, Langmuir 29(7), 2439–2445 (2013) G. Li, Y. Zhang, Y. Wu, L. Zhang, Appl. Phys. A Mater. Sci. Process. 81(3), 627–629 (2005) J. Melin, S. Quake, Annu. Rev. Biophys. Biomol. Struct. 36, 213–231 (2007) J. Pollack, C. Perou, A. Alizadeh, M. Eisen, A. Pergamenschikov, F. Williams, S. Jeffrey, D. Botstein, P. Brown, Nat. Genet. 23(1), 41– 46 (1999) S. Wang, S. Ota, B. Guo, J. Ryu, C. Rhodes, Y. Xiong, S. Kalim, L. Zeng, Y. Chen, M. Teitell, X. Zhang, Nano Lett. 11, 3431–3434 (2011) F. Xie, A. Centeno, M. Ryan, J. Riley, N. Alford, J. Mater. Chem. 1, 536– 543 (2013) P. Yager, T. Edwards, E. Fu, K. Helton, K. Nelson, M. Tam, B. Weigl, Nature 442, 412–418 (2006) H. Yin, X. Li, L. Que, Microelectronic Engineering, doi:10.1016/j.mee. 2014.05.020 T. Zhang, Y. He, J. Wei, L. Que, Biosens. Bioelectron. 38, 382–388 (2012) T. Zhang, P. Pathak, S. Karandikar, R. Giorno, L. Que, Biosens. Bioelectron. 30, 128–132 (2011) T. Zhang, S. Talla, Z. Gong, S. Karandikar, R. Giorno, L. Que, Opt. Express 18(17), 18394–18400 (2010) L. Zhou, F. Ding, H. Chen, W. Ding, W. Zhang, S. Chou, Anal. Chem. 84, 4489–4495 (2012)
A nanostructured aluminum oxide-based microfluidic device for enhancing immunoassay’s fluorescence and detection sensitivity Xiang Li & Haocheng Yin & Long Que
Published online: 22 June 2014 # Springer Science+Business Media New York 2014
Abstract A nanostructured aluminum oxide (NAO)-based fluorescence biosensing platform with a programmable sample delivery microfluidic interface is reported. The NAObased fluorescence sensor can tremendously enhance the fluorescence signals, typically up to 100×or more, over the glass substrate. The programmable sample delivery microfluidic interface, which is integrated with the NAO-based sensors, can automatically generate and deliver a series of different concentrations of the biological samples to each individual sensor. Hence it can facilitate the fluorescence-based biodetection and analysis for high throughput applications. Using Protein A and fluorophore-labeled Immunoglobulin G (IgG) as models, the binding between them on this platform have been demonstrated. It has been shown that the IgG of programmable concentrations can be delivered to individual sensor using the microfluidic interface and confirmed by the fluorescence images. Using current NAO-based fluorescence sensors without any optimization, the detectable concentration of IgG can be as low as 20 pg/mm2 using a conventional fluorescence microscope. Due to its inexpensive fabrication process, this technology could provide a disposable technical platform for fluorescence-based sensing and analysis.
Keywords Fluorescence sensor . Nanostructured aluminum oxide . Programmable sample concentration generation . Immunoassay
Electronic supplementary material The online version of this article (doi:10.1007/s10544-014-9881-1) contains supplementary material, which is available to authorized users. X. Li : H. Yin : L. Que (*) Institute for Micromanufacturing, Louisiana Tech University, 911 Hergot Avenue, Ruston, LA 71270, USA e-mail: [email protected]
1 Introduction Fluorescent immunoassay is one of the most widely utilized techniques in protein detection since the fluorescence technique can provide very high sensitivity and large multiplexing capabilities (Georganopoulou and Mirkin, 2009, Yager et al., 2006, Lakowicz, 2006, Darvill et al., 2013). However, it is still very important to further increase the sensitivity of fluorescence detection method since the higher sensitivity may offer the possibility for disease diagnosis at the early stage, and can also significantly reduce the consumption of the expensive biological samples in the experiments. Development of the advanced nanostructured substrate is one way to increase the sensitivity of fluorescence detection. In the past decades, some fluorescent immunoassay sensors based on metal-enhanced fluorescence (MEF) technology have been developed and reported. Examples include a variety of metallic nanoparticles (Xie et al., 2013), D2PA plate (Zhou et al., 2012), and tunable nanoplasmonic resonator (Wang et al., 2011)-based biosensors. Recently it has been reported by our group that the nanostructured aluminum oxide (NAO) substrate can increase the fluorescence signals up to two orders of magnitude for fluorescent dyes and labeled biomolecules compared to the glass substrate (Li et al., 2012a, Li et al., 2012b, Li et al., 2013). Different from MEF-based technology, the NAObased technology not only can be fabricated in an arrayed format in a cost-effective manner without using any top-down nanofabrication process, but also does not require a layer of dielectric material spacer between the sensing surface and fluorophores to avoid the quenching effect. For a variety of biological and biomedical applications, different concentrations of biological samples may be required for fluorescence detection and analysis. Ideally, a fluorescence technical platform, especially for high throughput applications, can not only offer the high sensitivity but also have an interface for simple sample delivery. For instance, in DNA or protein microarray
772
technology, the DNA or protein samples (i.e., the fluorescence labeled targets) usually are delivered to the desired locations (i.e. the fixed probes) on a solid substrate by utilizing an array of fine pins or needles, which are controlled by a robotic arm that is dipped into wells containing DNA probes and then deposit each probe at designated locations in an arrayed format. The resulting probes, representing the nucleic acid profiles, are ready to receive complementary targets derived from experimental or clinical samples (Pollack et al., 1999). This procedure is quite cumbersome, time-consuming, and expensive, especially for delivering a series of varied concentrations of biological samples to the fixed probes. Hence, use of an automatic varied concentration generating and delivering interface of biological samples will be a significantly favorable option. Herein, a microfluidic NAO fluorescence sensing platform with a programmable sample delivery microfluidic interface is reported. Specifically, this platform integrates arrayed fluorescence sensors using NAO as the fluorescence enhancement substrates, and a microfluidic concentration generator of the biological samples on a single chip. These functions make it possible to develop disposable arrayed fluorescence biosensors for high-throughput applications. Fig. 1a gives a schematic of arrayed fluorescence sensors, and a photo of the fabricated arrayed sensors is shown in Fig. 1b. Each sensor consists of a polydimethylsiloxane (PDMS) microfluidic chamber and a micropatterned NAO on indium tin oxide (ITO) glass substrate. The NAO micropattern is a microlines-pattern, which is located inside a PDMS chamber. The arrayed sensors are connected to a microfluidic network, which can generate and thus provide programmable concentrations of fluorophore-labeled proteins to the sensors. A SEM image of the micropatterned NAO is given in Fig. 1c, which is fabricated using one-step anodization process (Li et al., 2013). For technical demonstration, Protein A is used as a capture probe, while fluorophore-labeled IgG as a target as schematically shown in Fig. 1d.
2 Methods and materials 2.1 Fabrication of arrayed microfluidic fluorescence sensors The fabrication process flow is illustrated in Fig. 1S in the supplementary material (Zhang et al., 2012, He et al., 2012, He et al., 2014, Yin et al., 2014). Briefly, this process starts from an ITO glass substrate. A lift-off process is used to form the Al patterns. As a result, Al patterns are formed and connected with each other with Al lines as shown in Fig. 1Sb. Once the Al patterns have been fabricated and cleaned with acetone and DI water, one-step anodization using oxalic acid is then carried out to form the NAO as shown in Fig. 1Sc. The resulting thickness of the NAO is 2.7 μm. A
Biomed Microdevices (2014) 16:771–777
PDMS microfluidic chip is fabricated separately using a soft lithography process, then it is bonded with the NAO-glass chip, followed by assembling input and output tubing. The input tubes are connected with syringes controlled by a syringe pump, while the output tube leads to a biochemical waste collecting beaker. 2.2 Operation of the microfluidic chip for generating programmable concentrations The microfluidic chip for generating programmable concentrations of a sample is modified from previous work (Jeon et al., 2000). It has three inputs for flowing liquid solutions (i.e., biological samples, etc.) into the chip. As illustrated in Fig. 2a-c, symmetric flow is achieved by flowing the samples through the middle input (M-input), and flowing DI water or PBS buffer through the left input (L-input) and right input (Rinput). Asymmetric flow is achieved by flowing the samples through either the L-input or the R-input, and flowing DI water or PBS buffer through the M-input and either the Rinput or L-input. If the flowing rates of the sample/water are the same in all three inputs, the concentration profile at the outputs (1, 2, 3, 4, 5) can be predicted by modeling the pyramidal microfluidic network as an equivalent electronic circuit (Jeon et al., 2000). The symmetrical flow scheme results in a symmetrical sample-concentration profile at the outputs with the highest in the middle output channel 3. All the achieved concentrations at the outputs are lower than the original concentration of the sample. The asymmetrical flow scheme provides an alternative for providing not only the original sample concentration through one output, but also different concentrations at other outputs. Hence, depending on the concentration requirements, these two schemes can be utilized alone or combined. In addition, the concentration at each output channel can be accurately determined, which is very useful for quantitative analysis of bioreaction. Two different concentration profiles resulting from these two different flowing schemes have been experimentally demonstrated by flowing blue food dye (McCormick, Inc) as the simple model through one input and water through other two inputs in Fig. 2d-e. The flowing rates through all three inputs are the same. With the same concentration of blue food dye for these two flowing schemes, as expected, the symmetric flowing scheme shows the symmetric concentration profile with the highest concentration at the middle output 3, while the asymmetric flow scheme gives the highest concentration at the leftmost output 1. Note that more different concentrations of the sample can be achieved by cascading more levels of the fluidic network. In all the following experiments, the biochemical samples and buffer solution are transported to the microdevice through the assembled plastic tubing (Upchurch Scientific, Inc.) by a syringe controlled by a syringe pump (KD Scientific, Inc.).
Biomed Microdevices (2014) 16:771–777
773
Fig. 1 a 3-D sketch of the fluorescence sensors with a programmable sample concentration generating and delivery microfluidic interface; (b) photo of a fabricated device; (c) SEM image of NAO inside the fluorescence sensor; (d) sketch of the immunoassay on the fluorescence sensor
2.3 Chemicals and materials The immunoassay reagents used in the experiments include Protein A (Pierce Biotechnology, Inc), buffer solution phosphate buffered saline (PBS) (Sigma-Aldrich, Inc), SEA BLOCK blocking buffer (Pierce Biotechnology, Inc). SEA BLOCK blocking buffer contains steelhead salmon serum in PBS and 0.1 % sodium azide. It does not interact significantly with mammalian antibodies, and thus results in minimal or no background. Deionized (DI) water was obtained from a DI water purification system (Millipore, FRANCE). Human Immunoglobulin G (IgG) labeled with FITC was purchased from Rockland Immunochemicals Inc. and was diluted using PBS (pH=7.2) solution at a concentration of 200 μg/mL. Rabbit Immunoglobulin G (IgG) labeled with Rhodamine was purchased from Rockland Immunochemicals Inc. and Fig. 2 Operation of the microfluidic chip: (a) 2-D sketch of the fluorescence sensors; (b) close-up of one fluorescence sensor, showing the NAO microline-pattern inside the sensor; (c) close-up of the programmable sample concentration generating region with output channel 1, 2, 3, 4, 5; (d) experimental concentration profile of blue food dye at the outputs under symmetric flow scheme; (e) experimental concentration profile of blue food dye under asymmetric flow scheme
was diluted using PBS (pH=7.2) solution at a concentration of 200 μg/mL. 2.4 Fluorescence biodetection procedure For technical demonstrations, Protein A is used as a capture protein while IgG labeled with FITC or Rhodamine as a target. Sea Block is used as the blocking layer for preventing nonspecific binding. This bioassay is very selective and only specific between Protein A and IgG, and hence has usually been used as a simple model for immunoassay (Zhang et al., 2010, Zhang et al., 2011, Zhou et al., 2012). The detailed procedure for the immunoassay is schematically illustrated in Fig. 3a. Briefly, first, Protein A is immobilized on the NAO surface, followed by rigorous rinsing by PBS to remove the unbound Protein A; then Sea Block is applied to occupy the
774
unbound sites on the NAO surface, followed by rigorous rinsing by PBS. This step is for minimizing the non-specific binding. Finally, fluorophore-labeled IgG is applied and bound with Protein A, followed by gentle rinsing by PBS. Thereafter, the fluorescence images are obtained by a fluorescence inverted microscope equipped with a DP71 camera (Olympus, Inc). A representative optical image and the corresponding fluorescence image with IgG concentration of 1.33 μM are given in Fig. 3b. The NAO microline width is ~15 μm. The integration time for obtaining the fluorescence images is 500 ms. It is clear that the fluorescence images of the IgG on the NAO microlines have been enhanced dramatically compared to those on the glass substrate between the NAO microlines. 2.5 Fluorescence measurement and experimental data analysis The fluorescence images are obtained using a fluorescence microscope (Olympus, Inc). The built-in black balance function/method in the fluorescence microscope is used to improve the contrast of the fluorescence image. The relative intensity of the fluorescence is directly obtained from the images. A MatLab program based on the Imaging Processing toolbox in MatLab has been written to read the files of fluorescence images, which are then converted to gray scale images from the color images (Li et al., 2012a). A horizontal cutline is obtained through the fluorescence image and the corresponding intensity is obtained and then plotted. The fluorescence detection limit is determined by comparing the fluorescence signals from the NAO substrates immobilized with fluorophore-labeled IgG and those from the bare NAO substrate. Specifically, the detection limit of IgG is defined as its concentration at which the fluorescence signals are at least 3-fold larger than those from bare NAO substrate (Zhou et al., 2012).
3 Results and discussion In Fig. 4, two representative different concentration profiles of the IgG have been generated and then delivered to the fluorescence sensors, thereby resulting in different fluorescence imaging profiles. In both cases, the flowing rates for IgG and PBS buffer are kept same at 0.3 μL/min through the three inputs (i.e. L-input, M-input and R-input). Using the symmetric flowing mechanism in Fig. 4a-1 and the asymmetric flow mechanism in Fig. 4b-1, different concentrations of IgG are generated and then bound to the Protein A. Fluorescence images of the arrayed sensors under symmetric flowing scheme are shown in Fig. 4a-2. Rhodaminelabeled rabbit IgG is used in this demonstration. The fluorescence intensity varies with different concentrations of IgG. As
Biomed Microdevices (2014) 16:771–777
expected, the sensor in the middle connected to channel 3 has the highest fluorescence intensity, while the fluorescence intensities of other four sensors reduce symmetrically around the middle sensor to essentially negligible. The fluorescence images of the arrayed sensors under asymmetric flowing scheme are shown in Fig. 4b-2. FITC-labeled human IgG is used for this test. As shown, the fluorescence intensities decrease from the left sensor, connected to channel 1, to the right one, connected to channel 5. Specifically, the sensor connected to channel 1 has the highest fluorescence intensity. The fluorescence intensities of the two sensors, connected to channel 4 and 5, are found identical to that of the blank NAO sensor. The very low fluorescence emission from these two sensors is just the scattering background from the NAO. Since negligible IgG has been flowed to channel 4 and 5, there is essentially no IgG applied to these two sensors. The fluorescence intensity on each sensor has been first obtained by subtracting the NAO background scattering intensity, then normalized to the fluorescence intensity of the IgG at its original concentration (i. e., the concentration of IgG before flowing into the device) on an identical sensor. Based on the measured fluorescence intensity, the normalized IgG concentration for each channel can be obtained as shown in Fig. 4a-3 and b-3, which is very consistent (±5 % offset) with the prediction by the equivalent electronic circuit model (Jeon et al., 2000). These experiments indicate the possibility to quantify the bio-samples applied on the sensors. Note that the resulting concentration profiles of IgG can also be easily tuned by simply changing the flowing rates of the samples (i. e., IgG) and/or changing the input (i.e., among L-input, Minput, R-input) for the samples. Even though such concentration profiles of IgG cannot be easily predicted using the simple electronic circuit model, but they can still be determined by some simple trial experiments. As such, a variety of preknown concentration profiles of IgG can be generated and delivered to the fluorescence sensors, offering a great deal of flexibility for fluorescence-based bioassay, particularly useful for high-throughput applications. Furthermore, it is also possible to generate and deliver a specific/distinct concentration of the IgG to each single sensor by using a modified microfluidic interface, such as a combination of the microfluidic concentration generator with 2-D or 3-D microfluidic network of controlled valves and pumps (Melin and Quake, 2007). Note that Rhodamine-labeled rabbit IgG and FITC-labeled human IgG are used as two examples for the technical demonstrations. The NAO-based sensors can be potentially used for all fluorescence-based bioassay. In addition, the immunoassay has also been carried out on three types of sensors as shown in Fig. 5. In these experiments, the immunoassay has been carried out only on the upper part of the NAO microline-patterns. The NAO in the three sensors has been fabricated by different anodization time, which is 35 mins, 25 mins, and 15 mins in Fig. 5 a, b, and c, respectively.
Biomed Microdevices (2014) 16:771–777
775
Fig. 3 a Sketch showing the major steps for immunoassay; (b) representative optical image and the corresponding fluorescence image after completing the immunoassay on the sensor
Optical images of NAO microline-patterns of the three sensors are given in Fig. 5a-1, b-1, and c-1. The corresponding fluorescence images after immunoassay are shown in Fig. 5a-2, b2), and c-2, respectively. In order to enhance the image contrast, the fluorescence images are adjusted by the blank balance function provided by the software of the fluorescence microscope. Clearly, the fluorescence signals on NAO Fig. 4 Symmetric flow scheme: (a-1) the IgG is flowed through the M-input; (a-2) fluorescence images of the sensors with different concentrations of IgG; (a-3) the measured concentration profile of IgG delivered to each sensor based on the fluorescence intensity. Asymmetric flow scheme: (b-1) the IgG is flowed through the L-input; (b-2) fluorescence images of the sensors with different concentrations of IgG; (b-3) the measured concentration profile of IgG delivered to each sensor based on the fluorescence intensity
microline-patterns are much larger than those on the glass due to the fluorescence enhancement capability of the NAO. In order to quantitatively evaluate the IgG attached to the Protein A, the volume of the IgG solution is fixed at different concentrations and applied to the sensor surfaces. After an incubation time of 40 mins, the chip is rinsed gently by PBS. In these experiments, the applied concentration of IgG is
776
Biomed Microdevices (2014) 16:771–777
Fig. 5 Three fluorescence sensors with NAO microline-patterns fabricated using different anodization time: (a) 35 mins; (b) 25 mins; (c) 15 mins. Immunoassay carried out only on the upper part of the NAO
microline-patterns for the three sensors. From top row to bottom row: Optical images; the corresponding fluorescent images; and the plots of the fluorescence intensity. The width of the NAO microlines is ~15 μm
0.1 ng/mm2, which is calculated by IgG solution concentration (pg/mL)×volume (mL) of each fluorescence sensor/NAO area (mm2) inside a sensor. As shown in Fig. 5, the fluorescence images give the identical profiles to those of the NAO microline-pattern for each type of sensor. The plots of the fluorescence intensities of IgG on the three different sensors are given in Fig. 5a-3, b-3, and c-3, respectively. It is shown that using the same amount of IgG, the fluorescence intensity has little difference for all three types of sensors. In other words, the NAO substrates formed by the partially anodized Al (shorter anodization time) and the more anodized Al (longer anodization time) essentially have the similar fluorescence enhancement capability. It has been observed that some other parameters such as the thickness of the NAO can tune its enhancement capability. Experiments find that IgG at a concentration of 20 pg/mm2 can be detected on 5 μm thick NAO substrate. The origin of the fluorescence enhancement has been discussed in our previous work (Li et al., 2012a, Li et al., 2012b, Li et al., 2013). Briefly, the fluorescence enhancement is mainly due to the following primary reasons. First, the
surface scattering effect of the NAO causes the redistribution of the electromagnetic fields with high surface intensities, thus contributing to the fluorescence enhancement. Second, the nanoscale NAO grains may offer waveguide effect (Dorfman et al., 2006), resulting in evanescent field from the surface of the NAO grains, which play an important role in the enhancement of the fluorescence signals. Third, it is possible that the NAO nanostructure could concentrate a little more biomolecules to its surface. Given the optical properties of the NAO (Li et al., 2005), the scattering wavelength can be potentially tuned by modifying the nanopore size and the spacing among nanopores. Hence, it is potentially possible to tune the scattered optical wavelength to match the excitation wavelength of the fluorescent dyes, thereby further improving the fluorescence signals. It has been observed that the largest enhancement occurs at the NAO surface, and the enhancement drops greatly with the distance between the surface and the fluorescent dyes. Therefore, different from MEF-based fluorescence sensors, no space layer of dielectric material between the sensing surface and the fluorescent dyes is required for NAO-based fluorescence sensors to avoid the
Biomed Microdevices (2014) 16:771–777
quenching effect (Anger et al., 2006). This feature further simplifies the fluorescence-based bioassay. In addition, since arrayed sensors can be batch-fabricated in an inexpensive and efficient manner, a disposable fluorescence-based platform can be potentially developed for the multiplexed biomolecular detection and analysis.
4 Conclusions A nanostructured material-based fluorescence sensing platform with a programmable sample delivery microfluidic interface has been developed and reported. The fluorescence sensing platform is enabled by the NAO, which can dramatically enhance the fluorescence signals compared to the glass substrate. The programmable sample delivery microfluidic interface, which is integrated with the arrayed sensors, can generate and deliver different concentrations of the biological samples to each sensor, paving a way for achieving a multiplexed fluorescence-based biodetection and analysis. As a technical demonstration, the fluorescence signals and images for an immunoassay with Protein A and IgG labeled with Rhodamine or FITC have been measured. The delivery of programmable concentrations of IgG to each sensor by the platform has been achieved. It has been observed that 20 pg/ mm2 of IgG can be detected. Given the simple and inexpensive fabrication process of the sensor, and the feasibility of fabricating hundreds of sensors on a single chip, this technical platform may have broad applications in biological, medical, chemical and other research.
Acknowledgments The research is funded in part by a NSF grant.
777
References P. Anger, P. Bharadwaj, L. Novotny, Phys. Rev. Lett. 96, 1130021– 1130024 (2006) Darvill, D., Centeno, A., Xie, F., 2013. Phys. Chem. Chem. Phys, doi:10. 1039/C3CP50415H. A. Dorfman, N. Kumar, J. Hahm, Langmuir 22, 4890–4895 (2006) D. Georganopoulou, C. Mirkin, Nature 462, 461–464 (2009) Y. He, X. Li, L. Que, J. Nanosci. Nanotechnol. 12(10), 7915–7921 (2012) Y. He, X. Li, L. Que, J. Biomed. Nanotechnol. 10, 767–774 (2014) N. Jeon, S. Dertinger, D. Chiu, I. Choi, A. Stroock, G. Whitesides, Langmuir 16, 8311–8316 (2000) Lakowicz, 2006. Plasmonics, 1, 5–33. Li, X., He, Y., Zhang, T., Lee, T., Que, L., 2012a. Proceeding of IEEE NANO 2012, 10.1109/NANO.2012.6321926 doi:10.1109/NANO. 2012.6321926#blank X. Li, Y. He, T. Zhang, L. Que, Opt. Express 20(19), 21272–21277 (2012b) X. Li, Y. He, L. Que, Langmuir 29(7), 2439–2445 (2013) G. Li, Y. Zhang, Y. Wu, L. Zhang, Appl. Phys. A Mater. Sci. Process. 81(3), 627–629 (2005) J. Melin, S. Quake, Annu. Rev. Biophys. Biomol. Struct. 36, 213–231 (2007) J. Pollack, C. Perou, A. Alizadeh, M. Eisen, A. Pergamenschikov, F. Williams, S. Jeffrey, D. Botstein, P. Brown, Nat. Genet. 23(1), 41– 46 (1999) S. Wang, S. Ota, B. Guo, J. Ryu, C. Rhodes, Y. Xiong, S. Kalim, L. Zeng, Y. Chen, M. Teitell, X. Zhang, Nano Lett. 11, 3431–3434 (2011) F. Xie, A. Centeno, M. Ryan, J. Riley, N. Alford, J. Mater. Chem. 1, 536– 543 (2013) P. Yager, T. Edwards, E. Fu, K. Helton, K. Nelson, M. Tam, B. Weigl, Nature 442, 412–418 (2006) H. Yin, X. Li, L. Que, Microelectronic Engineering, doi:10.1016/j.mee. 2014.05.020 T. Zhang, Y. He, J. Wei, L. Que, Biosens. Bioelectron. 38, 382–388 (2012) T. Zhang, P. Pathak, S. Karandikar, R. Giorno, L. Que, Biosens. Bioelectron. 30, 128–132 (2011) T. Zhang, S. Talla, Z. Gong, S. Karandikar, R. Giorno, L. Que, Opt. Express 18(17), 18394–18400 (2010) L. Zhou, F. Ding, H. Chen, W. Ding, W. Zhang, S. Chou, Anal. Chem. 84, 4489–4495 (2012)
