We demonstrate a highly sensitive imaging method combined a terahertz time-domain spectroscopy and an interference effect for label-free protein detection on a polyvinylidene difluoride membrane. The method is based on terahertz time-domain spectroscopy and uses an interference effect. Biotin is linked to the membrane using poly ethylene glycol or poly ethylene glycol methyl ether to prevent it from being washed off. Binding of the biotin with streptavidin is then observed by measuring the terahertz signal change due to the variation of the membrane refractive index. We demonstrate the detection of the binding streptavidin protein in gradually decreasing concentrations, down to 27 ng mm-2, using the image recorded at 1.5 THz.
© 2008 Optical Society of America
Many researchers have been investigating various microarray technologies for their potential to enable a comprehensive analysis of the interaction of various biomaterials. DNA microarrays  or protein biochips  are prominent examples of such technology. Imaging technologies allow fast, easy, and parallel detection of thousands of addressable elements in a single experiment. They can be expected to become a crucial tool for high-throughput drug discovery and life science. In these methods, label substrates, which might involve fluorescence, an enzyme reaction, or a radioisotope, are used for the detection of DNA and proteins. However, these procedures are complex and time consuming.
Molecular recognition due to the differential activity of proteins is important in many biological processes. In recent years, small molecules have also received much attention from drug discovery scientists. Small molecules, from natural resources, are an important source of bioprobes, which are useful in the study of protein function or pharmacological effect. To screen those combinations of small molecules and proteins which have a biologically important function, microarrays and SPR (surface plasmon resonance) sensors are studied or already commercialized. Since microarrays, with their ability of multiplexing series of individual samples, are suitable for high-throughput screening processes. Therefore, printing small molecules or carbohydrates on glass slides or on other solid surfaces have recently been researched [3, 4]. However, this method involves time-consuming labeling procedures or sophisticated experimental techniques. SPR is a major method for characterizing macromolecular interactions. It is an optical and label-free method that uses the evanescent wave phenomenon to measure changes in the refractive index very close to a sensor surface . However, disadvantages of SPR for bioanalytical applications are its inherent low sensitivity for detection of small molecules  and the difficulty of applying to parallel analysis such as by imaging. Additionally, both methods need an expensive arrayer system to prepare a sensor chip. In order to solve those problems, an analytical method using a general membrane as a substrate was proposed, but since small molecules flow by the washing process, the method is still being studied .
In recent years, applied research has started in the electromagnetic frequency range between infrared and microwaves region: the terahertz (THz) waves; it is the last unexplored region of the electromagnetic waves. Recent remarkable developments in THz technology have made it clear that many substances have rich electromagnetic characteristics that can be considered fingerprint spectra in the THz range. Intermolecular modes, internal motion, or lattice vibrations in the case of crystalline materials may contribute to the THz spectrum. Because of the collective low vibration modes in the THz range, their position and strength is highly sensitive to the conformation and structure of the molecule. For this reason, various studies for THz applications are underway, such as the detection of illicit drugs and explosives in envelopes and other packages, and the label-free detection of DNA hybridization and protein interactions [8–14]. Here, we report on an interference terahertz label-free imaging technique to detect the interaction of protein and small molecules on a polyvinylidene difluoride (PVDF) membrane using a terahertz imaging system based on THz-TDS (time-domain spectroscopy).
2. Materials and sample preparation
In this experiment we acquired the images of the affinity binding between biotin fixed on (PVDF) membrane and the related bacterial protein streptavidin. The highly specific and strong binding of the biotin-streptavidin system has led to its wide usage in a variety of biotechnological applications. The PVDF membrane has been used as a support for sequence analysis of biopolymers . Such a filter membrane has a porous structure and the percentage of voids is over 80 %. Therefore, far infrared light such as terahertz waves can easily penetrate the membrane filter and its refractive index is small (k~0.05, n~1.1). Some methods for immobilization of small-molecular substances such as biotin or natural products on glass slides have recently been developed . Such small molecule arrays are not only useful for a comprehensive analysis of biological processes, but have also a potential for a high-throughput processes for creating and identifying synthetic ligands for any protein. Because the small molecular compound does not interact directly with the membrane, they flow off by washing after the reaction with the protein. To prevent flowing off of the biotin molecules from the membrane, we use a linking method, which consists in conjugating the biotin molecules with PEG (Sigma-Aldrich Poly (ethylene glycol), average molecular weight =3400) or MPEG (Sigma-Aldrich Poly (ethylene glycol) methyl ether, average molecular weight =5000) to the PVDF membrane. First, we prepared the biotin on the PEG or MPEG support using a chemical synthetic procedure . A series of solutions (0.2 µL) of small molecules conjugated with PEG or MPEG support (1×10-2, 2×10-3, 4×10-4, 8×10-5, 1.6×10-5, 3.2×10-6 M (=mol/L)) was dot-blotted on the PVDF membrane in doublet. Each spot size was approximately 3 mm in diameter on the PVDF membrane. The membrane was immersed in the fixation solution (10 mL) for 10 min, followed by washing with water (10 mL, 2×1 min, then 2×5 min). The membrane was then immersed and agitated in a skim milk solution (10 mL) for blocking for 60 min and then in the skim milk solution containing streptavidin (2.0 mg/mL in water, 3.3×10-3 mL) at room temperature for 6 hr, washed with the DBB (Dot-Blot Buffer, 10 mL, 6×5 min) and water (10 mL, 1 hr), and dried between Whatman papers for 120 min.
In this paper, to confirm the interaction with streptavidin, we prepared a membrane which was interacted with Alexa633-labeled streptavidin. This membrane was probed by a fluorescence image analyzer (FUJIFILM, FLA-2000).
3. Experimental methods
We used a multispectral reflection imaging system based on a THz-TDS to obtain an unlabeled image of biotin-streptavidin binding. The spectra were measured from 0.025 to 2 THz with 25 GHz resolution, resulting in 80 images. During the measurement, the spectrometer was purged with nitrogen gas to avoid absorption of the water vapor in the air. To yield high reflectivity of membrane surface, we constructed a high resistivity silicon plate (>10 kΩ·cm, 9.2 mm thickness) as a sample stage. A membrane sample was sandwiched between the silicon plate and a mirror (Fig. 1). We could observe the interference wave in the time waveform, which consists of a reflected wave from the surface of the membrane and the reflections from a mirror on the back side of the membrane. This indicates that we can use an interference effect such as in a Fabry-Perot etalon to achieve a highly sensitive measurement. When streptavidin is bound with biotin on the membrane, we expect that the refractive index becomes higher than in the parts without streptavidin. This change of the refractive index causes a change of the effective path length through the membrane. Consequently, the interference wave pattern after the Fourier transform (FT) shifts to the lower frequencies. Figure. 2 shows a comparison of the time waveform and its FT spectrum for different thicknesses of the polyethylene sheets (400 µm and 500 µm) instead of a refractive index change. A 100 µm difference of thickness was detected as a transmissivity change of as much as 80 % at 1.5 THz (as shown in Fig. 2 (b)). We chose this frequency for our measurements to achieve a contrast as high as possible.
4. Result and discussion
Figure. 3 shows a schematic of the membrane and a detection result of fluorescently-labeled streptavidin.
In Fig. 3 (b) we show the detected image of fluorescently-labeled streptabvidin (gray dots) using Fluorometric Imaging Scanner. At the two bottom lines, where biotin was not immobilized on the membrane, we could not confirm the presence of the bonding with streptavidin. That reason is that biotin molecules had been washed out from the membrane during the washing process. This indicates that PEG and MPEG were effective for the immobilization of biotin molecules on the PVDF membrane. Furthermore, we could confirm that such immobilized biotin preserves the binding capacity with streptavidin.
The THz image of the membrane was acquired with the imaging system based on THz-TDS, as shown in Fig. 4.
The image of the binding between biotin immobilized on the membrane using PEG or MPEG and the non-labeled streptavidin was obtained. Lower brightness shows a high bonding amount of streptavidin. As in the case of the fluorescent labeling image, the binding of streptavidin at the two bottom lines was not observed in the THz image. In this experiment, we could not confirm less than 8×10-5 M streptavidin from fluorescent labeling image (Fig. 3), however, using the label-free THz imaging we succeed in detection of 1.6×10-5 M (27 ng mm-2) streptavidin at the two top lines (Fig. 4). The highest concentration spots (1.0×10-2 M) of streptavidin content correspond to approximately 17 µg mm-2. By comparison, in binding with and without streptavidin, even at the highest concentration biotin, the transmittance change was below 1 % using a normal THz-TDS transmission system and we could not detect lower concentration binding.
Figure 5 shows a transmittance spectrum of the points (a) and (b) in Fig. 4. An interference spectrum caused by the optical path difference due to the varying refractive index of the membrane is observed. The spectrum form shifts only slightly with the binding of streptavidin. This indicates that the binding between biotin and streptavidin leads to a change of the refractive index of the membrane. We can see the drastic change of the signal at 1.5 THz, which in this case is approximately 20 %, corresponding to a concentration of streptavidin content of approximately 17 µg mm-2.
In both results, by fluorescent labeling and by THz non-labeled imaging, we can see that the binding affinity of biotin immobilized in MPEG is higher than in PEG. The difference of reactivity of PEG and MPEG does not become clear at this time. However, we can choose the most suitable method to immobilize the small molecules on the membrane, as many kinds of linker material and base materials have been reported.
We have demonstrated a highly sensitive imaging method for label-free protein detection on a membrane. To prevent biotin from flowing off the PVDF membrane, PEG and MPEG were found to be effective immobilizers, without affecting the binding capacity of streptavidin. In our experiment, a detection limit of 27 ng mm-2 streptavidin from an interference THz image at 1.5 THz has been achieved. This sensitivity is not better than Michan’s method . This time we used a commercial membrane designed for protein analysis. However, if we can develop a more suitable substrate instead of the membrane for this method, higher sensitive can be expected. Additionally, since this interferometric method does not need an optical delay in principle, an inexpensive system, whose THz source is a QCL, a BWO or a parametric tunable THz source, can be constructed.
The binding specificity of the membrane, that is, the amount of biotin that adheres to it, depends on the membrane material and linkers. THz imaging is used only to detect the existence of the binding. As linkers, PEG and MPEG have proved to be useful not only for the biotin molecule but also for other low-molecular compounds. This property is suitable to a high throughput detection for drug discovery. We expect that our method can be used in a variety of applications such medical diagnosis and allergy testing based on an antigen-antibody reaction, and as a sensor for industrial use.
The authors are grateful to Adrian Dobroiu for the valuable discussion and encouragement. We would like to acknowledge the support of a Grant-in-Aid for Young Scientists (18070501) from The Ministry of Health, Labour and Welfare of Japan.
References and links
3. N. Kanoh, S. Kumashiro, S. Simizu, Y. Kondoh, S. Hatakeyama, H. Tashiro, and H. Osada, “Immobilization of Natural Products on Glass Slides by Using a Photoaffinity Reaction and the Detection of Protein- Small-Molecule Interactions,” Angew. Chem. Int. Ed. 42, 5584–5587 (2003). [CrossRef]
4. S. Fukui, T. Feizi, C. Galustian, A. M. Lawson, and W. Chai, “Oligosaccharide microarrays for high-throughput detection and specificity assignments of carbohydrate-protein interactions,” Nature Biotechnology 20, 1011–1017 (2002). [CrossRef] [PubMed]
6. G. A. J. Besselink, R. P. H. Kooyman, P. J. H. J. van Os, G. H. M. Engbers, and R. B. M. Schasfoort, “Signal amplification on planar and gel-type sensor surfaces in surface plasmon resonance-based detection of prostate-specific antigen,” Anal. Biochem, 333, 165–173 (2004). [CrossRef]
7. Y. Hatanaka, M. Hashimoto, and Y. Kanaoka, “A Rapid and Efficient Method for Identifying Photoaffinity Biotinylated Sites within Proteins,” J. Am. Chem. Soc. 120, 453–454 (1998). [CrossRef]
9. Y. C. Shen, T. Lo, P. F. Taday, B. E. Cole, W. R. Tribe, and M. C. Kemp, “Detection and identification of explosives using terahertz pulsed spectroscopic imaging,” Appl. Phys. Lett. 86, 241116 (2005). [CrossRef]
10. M. Nagel, P. Haring Bolivar, M. Brucherseifer, H. Kurz, A. Bosserhoff, and R. Büttner, “Integrated THz technology for label-free genetic diagnostics,” Appl. Phys. Lett. 80, 154 (2002). [CrossRef]
11. M. Brucherseifer, M. Nagel, P. Haring Bolivar, H. Kurz, A. Bosserhoff, and R. Büttner, “Label-free probing of the binding state of DNA by time-domain terahertz sensing,” Appl. Phys. Lett. 77, 4049 (2000). [CrossRef]
12. B. Ferguson and X.-C. Zhang, “Materials for terahertz science and technology,” Nature Materials 1, 26–33 (2002). [CrossRef]
13. M. Walther, P. Plochocka, B. Fischer, H. Helm, and P. U. Jepsen, “Collective vibrational modes in biological molecules investigated by terahertz time-domain spectroscopy,” Biopolymers 67, 310–313 (2002). [CrossRef] [PubMed]
14. B. Fischer, M. Hoffmann, H. Helm, R. Wilk, F. Rutz, T. K. -Ostmann, M. Koch, and P. U. Jepsen, “Terahertz time-domain spectroscopy and imaging of artificial RNA,” Opt. Express 13, 5205–5215 (2005). [CrossRef] [PubMed]
15. P. Matsudaira, “Sequence from Picomole Quantities of Proteins Electroblotted onto Polyvinylidene Difluoride Membranes,” J. Biochem. 262, 10035–10038 (1987).
16. M. Oikawa, M. Ikoma, and M. Sasak, “Alkoxyacetyl (AAc) group as a useful linker for organic synthesis on poly (ethylene glycol) support,” Tetrahedron Lett. 45, 2371–2375 (2004). [CrossRef]
17. S. P Mickan, A. Menikh, H. Liu1, C. A. Mannella, R. MacColl, D. Abbott, J. Munch, and X.-C. Zhang, “Label-free bioaffinity detection using terahertz technology,” Phys. Med. Biol. 47, 3789–3795 (2002). [CrossRef] [PubMed]