Analytica Chimica Acta 479 (2003) 117123

Development of immunochips for the detection of dengue viral antigens

Chih-Cheng Su a , Tzong-Zeng Wu a, , Li-Kuang Chen b , Hui-Hua Yang b , Dar-Fu Tai c
a

Department of Life Science and Institute of Biotechnology, National Dong Hwa University, Hualien, Taiwan b Department of Emergency Medicine, Medical School, Buddhist Tzu Chi University, Hualien, Taiwan c Department of Chemistry, National Dong Hwa University, Hualien, Taiwan Received 21 November 2001; received in revised form 25 November 2002; accepted 26 November 2002

Abstract Recently, dengue virus has become a new emerging disease in the world. However, the procedures currently used for the detection of dengue virus are cumbersome and time-consuming. This is unfavorable for early stage epidemiological control and effective medical treatment. A new detection system was developed based on the quartz crystal microbalance (QCM) coating using two monoclonal antibodies that act specically against the dengue virus envelope protein (E-protein) and non-structural 1 protein (NS-1 protein), respectively. Three different immobilizing methods, the glutaraldehyde (GA) method, protein A method and carbodiimide method (1-ethyl-3-(3-dimethyl- aminopropyl) carbodiimide, EDC) were used to prepare the immunochips. The cocktail immunochip, which has both antibodies attached, was also fabricated and used in comparison. The results showed that the protein A method was the best among the three. The sensitivity of the immunochip was 100-fold greater than the conventional sandwich ELISA method. The cocktail immunochip had a higher signal level than the normal immunochip. The time required for detection was shorter (about 1 h) and a blood specimen could be used to detect the virus in the viremia phase. 2002 Elsevier Science B.V. All rights reserved.
Keywords: Dengue virus; Detection; Quartz crystal microbalance (QCM); Immunochip; Immobilization

1. Introduction Dengue viruses are mosquito-borne viruses. They belong to the genus Flavivirus, family Flaviviridae. There are four serotypes with all serotypes having the Flavivirus characteristics. Relatively small size (4050 nm), genetic substances packaged in a spherical lipid envelope with the same Flavivirus epitopes,
Corresponding author. Tel.: +886-3-8662500x21305; fax: +886-3-8662173. E-mail addresses: tzwu@mail.ndhu.edu.tw (T.-Z. Wu), dftai@mail.ndhu.edu.tw (D.-F. Tai).

with the Aedes aegypti mosquito as the most common transmitter [13]. The incubation period is about 314 days after a bite from an infected mosquito. The patient then may exhibit some non-specic u-like syndromes, including frontal headache, retro-orbital pain, body aches, joint pain, weakness and a rash. The acute phase may last 210 days. After that, the immuno-system will start to produce an anti-dengue antibody to neutralize the virus and the patient will start to recover from the illness [15]. If the age of the patient is under 15 or the patient is re-infected by a different dengue virus serotype, there is great possibility that the dengue fever will develop into dengue

0003-2670/03/$ see front matter 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0003-2670(02)01529-5

118

C.-C. Su et al. / Analytica Chimica Acta 479 (2003) 117123

hemorrhagic fever (DHF) which has a higher mortality rate (about 1520%) and is more dangerous [68]. The incidence of dengue virus infection was localized in certain tropical urban cities before 1940s. World War II provided excellent conditions for the transmission of mosquito-borne diseases. The carriers and virus spread out through the world and the illness caused by dengue virus gradually became epidemic. In 1997, dengue virus and Aedes aegypti had already become distributed in the tropics worldwide. Over 2.5 billion people live in areas where dengue fever is endemic. Dengue fever now causes more illness and death than any other arbovirus disease. Dengue virus infections have become one of the most serious problems in developed countries [2,8,9]. The clinical symptoms for dengue fever and DHF are atypical. Effective vaccines or drugs are still unavailable to prevent or cure the disease caused by dengue virus. It is crucial for a doctor to diagnosis and choose intense medical care to save the patients life as early as possible. Generally, an incident of dengue virus infection can be identied by virus isolation, viral antigen detection or RNA from serum or tissues, or detection of the specic antibody from the patients serum. It will take 5 days to generate the specic antibody against dengue virus after the acute phase. Antibody detection is contributive to epidemic research but useless for early diagnosis. Detection and classication of the dengue virus, requires 45 days to isolate and cultivate the dengue virus. This is also very inefcient for early disease control. The PCR method can rapidly detect the viral RNA from the patients serum in the viremia phase, but it suffers from sample contamination and high technological support demands. There is great urgency [2,1013] to develop a novel detection system with high sensitivity, short operation time, handy processing with easy interpretation. The quartz crystal microbalance (QCM) is a kind of bulk-acoustic wave (BAW) resonator. Because of the reverse-piezoelectric effect, when a quartz crystal places in an alternating electric eld, it will oscillate, and the oscillation mode and fundamental frequency is related to the material characteristics. When a mass load is applied to this crystal, the frequency will change, and the relationship between the frequency shift and mass load is linear, as derived by Sauerbrey [14]. Unfortunately, the liquid will absorb the energy

through the viscoelastic effect when a crystal is submerged in liquid, thus limiting the use of QCM in liquid phase measurement. In 1980, Konash and Bastiaans [15] developed an apparatusQCMxed between two spacers allowing the liquid to ow through one side with the other side in contact with air. This permits the oscillation to occur in the liquid and measuring the QCM in liquid. Due to the high sensitivity, simple operation, easy interpretation and real-time measurement, QCM had already been used successfully in kinetic research, medical diagnosis and the detection of pathogenic microorganisms [1621]. In this study, QCM was used as transducer to detect the dengue virus in liquid. Monoclonal antibodies, either specic to the envelop protein (E-protein) or non-structural protein 1 (NS-1 protein) of the dengue virus serotype 2, were prepared. Three different immobilization methodsglutaraldehyde [22], protein A [22], and carbodiimide method [23]were employed to compare the efciency of immobilizing the antibody in QCM. A cultivated dengue virus broth was used as the positive control to determine the impact using different immobilization procedures. A mixture of antibodies was used to fabricate a cocktail immunochip to investigate the synergistic phenomenon between different antibodies.

2. Experimental 2.1. Materials The 9 MHz QCM was obtained from Tai-Tien Electronic Co. (Taipei, Taiwan) with a reproducibility of 1 Hz. Bovine serum albumin (BSA), 1-ethyl3-(3-dimethyl- aminopropyl) carbodiimide (EDC), glutaraldehyde (GA), glycine, hydrochloric acid, 2-[Nmorphplino]ethanesulfonic (MES) acid, p-nitrophenyl phosphate, premixed phosphate buffered saline (PBS) package, polyethylenimine (PEI), sodium hydroxide were purchased from Sigma (MS, USA). Protein A was obtained from Pierce (IL, USA). RPMI 1640 medium was from Gibco (Rockville, MD). Microtiter plate was from Dynatech Laboratories (Chantilly, VA). And alkaline phosphatasestreptavidin conjugate was form Kirkegard & Perry Laboratories (Gaithersburg, MD). All chemicals were reagent grade and used without any further purication.

C.-C. Su et al. / Analytica Chimica Acta 479 (2003) 117123

119

2.2. Cell culture and virus preparation K562 cells were propagated in RPMI 1640 medium supplemented with 5% fetal calf serum (FCS). DEN-2 virus (PL-046 strain). A Taiwanese strain obtained from one dengue fever patient in 1987, was kindly provided by the National Institute of Preventive Medicine, Taipei, Taiwan, ROC. Viruses were propagated in K562 cells and titrated using plaque-forming assay on BHK cells at 37 C. 2.3. Generation and characterization of DEN-specic MAbs DEN-specic monoclonal antibodies (MAbs) were generated and analyzed as previously described [24]. Briey, 6- to 8-week-old BALB/c mice were immunized i.p. with DEN-2 PL046-infected suckling mousenbrian suspensions mixed with an equal volume of complete Freunds adjuvant for the rst inoculation and mixed with incomplete Freunds adjuvant for the subsequent boosting. Splenocytes were fused with NS-1 myeloma cells and selected as described by Kohler and Milstein [25]. The hybridomasecreted specic antibodies were identied using enzyme-linked immunosorbent assay and immunoprecipitation assays with DEN-2 infected C6/36 cell lysates as the antigen source as previously described [24]. Limiting dilution generated single-cell clones of hybridoma. For the ascitic uid production, the cells were injected into incomplete Freunds adjuvant-primed BALB/c mice. The specicity of MAb 8-2 and 17-2 were against NS1 and E protein of DEN-2. 2.4. Titration of NS1 antigen by sandwich ELISA A sandwich ELISA was set-up to titrate the NS1, the antigen secreted in the culture supernatant of K562 cells infected with dengue virus as previously described [26]. Briey, each microtiter well was coated overnight at 4 C with 5 g/ml, 100 l per well of MAb D2/8-1 in 0.11 M carbonate buffer (Na2 CO3 / NaHCO3 ), pH 9.5. After washing with PBS0.1% Tween 20 (PBST), wells were blocked with 200 l of PBS1% BSA for 1 h at 37 C. The wells were washed and incubated with 100 l of serially diluted culture supernatants of K562 cells infected with dengue virus

in PBST containing 5% normal rabbit serum for 1 h at 37 C. After washing, the wells were incubated with biotin-conjugated MAb D2/8-1 for 1 h at 37 C at 1:500 dilution. The plates were then washed and a 1:2000 dilution of alkaline phosphatasestreptavidin conjugate were added and incubated for 1 h at 37 C. Finally, the enzyme activity was developed with the addition of substrate p-nitrophenyl phosphate and optical density was taken at the dual wavelengths of 405 and 630 nm with a Dynatech MR700 microplate reader. 2.5. Antibody immobilization 2.5.1. Surface modication of QCM To remove impurities and contaminants from the crystal surface, QCM was soaked in 1.2 N NaOH solution for 30 min, washed with distilled water, re-soaked in 1.2 N HCl for 5 min, washed with distilled water again, then dried for 30 min in oven. The QCM was then placed horizontally in a damp box and 2 l of 1% PEI in PBS was dropped onto it to react with the QCM overnight. After washing with distilled water to remove non-absorbed PEI, the QCM was vacuum dried in a desiccator for 30 min. 2.5.2. Antibody immobilization via GA Five microliters of 2.5% GA in PBS was dropped onto a PEI pretreated QCM and reacted for 90 min. After washing with distilled water, 5 l of antibody solution (no. 17-2, 100 g/ml in PBS) was dropped onto the QCM and reacted with GA for 90 min to immobilize the antibody onto the QCM. Next 5 l of 100 mM glycine solution was added to block the remaining aldehyde group. 2.5.3. Antibody immobilization via EDC Five microliters of the antibody solution (no. 17-2, 100 g/ml in PBS) and 5 l of 0.1 M EDC in MES buffer (pH 5.0) was mixed and reacted for 40 min. Next 5 l of this mixture was taken and dropped onto the QCM to react with PEI for 30 s. The QCM was then washed with distilled water to remove any non-reacting substance. 2.5.4. Antibody immobilization via protein A Five microliters of 1.5% GA in PBS was dropped on PEI pretreated QCM and allowed to react for

120

C.-C. Su et al. / Analytica Chimica Acta 479 (2003) 117123

90 min. After washing with distilled water, 5 l of protein A solution (500 g/ml in PBS) was dropped onto the QCM and reacted for 90 min to immobilize protein A onto the QCM. This was followed by the addition of 5 l of 100 mM glycine solution to block the remaining aldehyde. Five microliters of the antibody solution (no. 17-2, 100 g/ml in PBS) was dropped onto a protein A coated QCM to conjugate with protein A for 60 min. The QCM was then washed with distilled water to remove any non-binding antibody. 2.5.5. Fabrication of cocktail immunochip by protein A method Antibodies specic to dengue virus E-protein (Ab no. 17-2, 100 g/ml in PBS) and specic to the NS-1 protein (Ab no. 8-2, 100 g/ml in PBS) were premixed in equal ratios. Five microliters of the antibody mixture was then dropped onto a protein A coated QCM and reacted with protein A for 60 min. The QCM was then washed with distilled water to remove any non-binding antibody. 2.6. Flow injection system The ow injection system containing a peristaltic pump (Model M312, GILSON, ow rate = 80 l/ min), home-built ow cell, sample injection valve (Model 1106, OMNIFIT), home-built oscillation circuit (including oscillator and frequency counter) and a personal computer (PC) is shown in Fig. 1. The

antibody coated QCM was xed between two O-ring and placed in ow cell. Only one side of the QCM was in contact with the liquid. PBS was used for circulating the buffer and the dengue viral cultivation liquid and K562 cell culture solution were used separately as the positive and negative controls. The E-protein and NS-1 protein content in the viral cultivation liquid were 500 g/ml each, obtained from the ELISA results. Before beginning this experiment, 1% BSA in PBS was injected to block the non-specic binding site. The sample was then diluted various folds and injected and triggered using the buffer circulation system to ow through the antibody coated QCM. If any change occurred, the oscillation frequency would shift and measured by the oscillation circuit and shown and recorded by the PC. Every experiment was repeated three to ve times. The average values were calculated and the error bar in Figs. 25 was dened as three times the single standard deviation.

Fig. 1. Schematic diagram of ow injection system.

Fig. 2. The results of sandwich ELISA using anti-NS-1 protein antibody (E-protein or NS-1 protein in positive control was 500 g/ml originally).

C.-C. Su et al. / Analytica Chimica Acta 479 (2003) 117123

121

Fig. 3. The inuences of different immobilization methods for the detection of dengue virus (E-protein or NS-1 protein in positive control was 500 g/ml originally).

Fig. 4. The inuences of different immobilized on QCM for the detection of dengue virus (E-protein or NS-1 protein in positive control was 500 g/ml originally).

3. Results and discussion There are many ways to detect a virus. Most of these virus detection methods are based on detecting the viral RNA or DNA. However, those genetic materials are packaged inside an envelope, the quantity is usually very small and these genetic substances are not very stable. These disadvantages complicate the procedure necessary to produce a diagnosis. As the amount of viral protein is much more abundant than the RNA or DNA genetic materials and the stability of this protein is better, using vial protein is a superior detection target than genetic substances. The blood is full of viral particles during the viremia stage of the dengue virus infection. When cells are infected with dengue virus, they display the NS-1 protein on the cell surface or secrete NS-1 protein into the blood. Hence, many E-protein and NS-1 proteins exist in the blood specimen during the viremia phase. We chose the E and NS-1 dengue virus proteins as the detection target for this reason. Fig. 2 demonstrates the ELISA

assay result for the monoclonal antibody specic to the NS-1 protein (no. 8-1). It shows that the monoclonal antibody production was successful and the antibody was able to bind to the NS-1 protein. To determine the best way to immobilize the antibody onto the QCM, three methods (GA, EDC and protein A method) were used to combine the antibody and QCM. The dengue virus cultivation liquid was diluted into various folds and injected into the detection system to observe the change in frequency. The result is shown in Fig. 3. The protein A method was the best for immunochip fabrication. In the GA method, the aldehyde on the GA reacted non-selectively with the lysine residue amino group on the antibody. Consequently, some of the antibody antigen-binding sites formed covalent bonds with the GA and lost their ability to bind with the E or NS-1 proteins. Similar results were obtained when EDC was used to couple the PEI amino group with the carboxylic acid of the antibody Fc domain. The participation of other carboxylic acids on the antibody to form an amide linkage with

122

C.-C. Su et al. / Analytica Chimica Acta 479 (2003) 117123

Fig. 5. The results of cocktail immunochip for the detection of dengue virus (E-protein or NS-1 protein in positive control was 500 g/ml originally).

PEI also hindered the antigenantibody binding effect. Protein A was separated from Staphylococcus aureus. It has ve IgG binding domains that can catch the Fc domain of the antibody. Protein A was immobilized rst onto the QCM via GA followed by adding the antibody onto it. As protein A binds with the Fc domain of the antibody, the Fab domain will be exposed outside to bind to the virus. In other words, the antibody is immobilized onto the QCM in the proper orientation. This is why the protein A method obtains the best result in dengue virus detection [2729]. Theoretically, ELISA utilizes an enzyme labeled secondary antibody, such as alkaline phosphatase labeled antibody, to convert the substrate into a chromgenic product. The magnitude of the analyte was measured indirectly via measuring the chromgenic compound using a spectrophotometer. However, the signal will be too small to be measured if the concentration of analyte is very diluted. The result in Fig. 2 indicates that the spectrophotometer signal is undetectable when the viral antigen concentration is lower

than 5 g/ml. The QCM used here had a reproducibility of 1 Hz and a change in frequency larger than 3 Hz would be seen as a signal. According to this criteria, when the antigen concentration is 0.05 g/ml, a reliable signal can be obtained. The sensitivity of the immunochip is at least 100 times greater than ELISA. As the sensitivity becomes higher, the detectable dengue viral antigen concentration in the specimen could become lower. This means that an immunochip may nd a patient infected by dengue virus earlier than other methods when the number of viral antigens in blood is very few. The E and NS-1 proteins are present in the blood when the patient is in the viremia phase. If the immunochip could detect both the E and NS-1 proteins simultaneously, this would signify great progress in lowering the detection limit and increasing the sensitivity. The detection results for dengue virus based on a cocktail are shown in Fig. 4. More than one antibody was immobilized onto the QCM immunochip. Compared with Fig. 3, when the dengue viral antigen concentration was lower than 0.05 g/ml, the cocktail immunochip signal level was nearly twice than the normal immunochip with only one antibody on it. As the antigen concentration became higher, the difference between the cocktail and normal immunochip grew smaller. Because the total number of antibodies applied onto the protein A coated QCM was constant, and the antibody-binding site of the protein immobilized onto the QCM was the same. In this study, the anti-E-protein and anti-NS-1 protein antibodies were mixed in equal quantities then dropped onto the protein A coated QCM. The amount of anti-E-protein antibody or anti-NS-1 protein antibody on the cocktail immunochip could be half that on the normal immunochip. In higher antigen concentrations, the antibody on the cocktail immunochip could be saturated by antigens. At lower antigen concentrations, both the anti-E-protein and anti-NS-1 protein antibodies could effectively catch the antigen simultaneously and causing greater changes in the liquid lm next to the surface, producing a higher signal level than the normal immunochip. Compared with the other methods currently used to detect the dengue virus, the sensitivity of the immunochip is higher, the technical demand is lower, the interpretation is easier, the cost is cheaper and the time required for detection is shorter (about 3060 min per

C.-C. Su et al. / Analytica Chimica Acta 479 (2003) 117123

123

sample). It is not necessary to spend time on virus cultivation or the RT/PCR procedure and the interference caused by PCR contamination can be avoided. This immunochip can detect the E and NS-1 proteins in the viremia phase when a patient has atypical symptoms and suspected infected by the dengue virus. Their blood or serum specimen could potentially be analyzed using this immunochip to reveal the existence of the dengue virus within a short period of time. Doctors can choose the most suitable medical process to cure the patient and perform essential tasks to prevent the spread of dengue virus. This immunochip is superior for the early diagnosis and disease control for the dengue virus infection. 4. Conclusion Because of the high sensitivity, low technical demand, easy interpretation, short response time required and detection of the virus in the viremia stage, this immunochip is potentially suitable for the detection of dengue virus in the early stage. This method could contribute to early diagnosis and disease control. The immobilization of mixed antibodies onto the QCM can effectively elevate the signal level and increase the sensitivity. A fabrication concept for immunosensors, involving sensors coated with multi-antibodies to capture several different antigens to enlarge the signal level in the detection of other pathogens is available with this method. We will continue to examine clinical specimens using immunochips to investigate the interference of other compounds in the blood sample, as well as to evaluate the possibility of direct clinical specimen analysis.

References
[1] I. Kautner, M.J. Robinson, U. Kuhnle, J. Pediatr. 131 (1997) 516. [2] D.J. Gubler, Clin. Microbiol. Rev. 11 (1998) 480. [3] T. Solomon, M. Mallewa, J. Infect. 42 (2001) 104. [4] M.G. Jacobs, P.R. Young, Curr. Opin. Infect. Dis. 11 (1998) 319. [5] W.J.H. McBride, H. Bielefeldt-Ohmann, Microb. Infect. 2 (2000) 1041. [6] M.J. Cardosa, Curr. Opin. Infect. Dis. 13 (2000) 471. [7] A.L. Rothman, F.A. Ennis, Virology 257 (1999) 1. [8] D.J. Gubler, M. Meltzer, Adv. Virus Res. 53 (2001) 35. [9] W.J. Chen, S.L. Chen, L.J. Chien, C.C. Chen, Am. J. Trop. Med. Hyg. 55 (1996) 12. [10] C.L.K. Seah, V.T.K. Chow, Y.C. Chan, S. Doraisingham, Serodiagn. Immunother. Infect. Dis. 7 (1995) 55. [11] N. Fujita, S. Hotta, E. Konishi, H. Esaki, Sumarmo, Sujudi, Am. J. Trop. Med. Hyg. 56 (1997) 318. [12] A.O.S. Rodrigues, C.V.N. Camacho, L.A.B. Miagostovich, M.P. Araujo, J. Virol. Methods 77 (1999) 81. [13] P.R. Young, P.A. Hilditch, C. Bletchly, W. Halloran, J. Clin. Microb. 38 (2000) 1053. [14] G. Sauerbrey, Z. Phys. 155 (1959) 206. [15] P.L. Konash, G.J. Bastiaans, Anal. Chem. 52 (1980) 1929. [16] R.L. Bunde, E.J. Jarvi, J.J. Rosentreter, Talanta 46 (1998) 1223. [17] C.K. OSullivan, G.G. Guilbault, Biosens. Bioelectron. 14 (1999) 663. [18] T. Vo-Dinh, B. Cullum, Fresenius J. Anal. Chem. 366 (2000) 540. [19] I. Ben-Dov, I. Willner, E. Zisman, Anal. Chem. 69 (1997) 3506. [20] M. Minunni, M. Mascini, R.M. Carter, M.B. Jacobs, G.J. Lubrano, G.G. Guilbault, Anal. Chim. Acta 325 (1996) 169. [21] I.-S. Park, W.-Y. Kim, N. Kim, Biosens. Bioelectron. 15 (2000) 167. [22] S. Babacan, P. Pivarnik, S. Letcher, A.G. Rand, Biosens. Bioelectron. 15 (2000) 615. [23] T.-Z. Wu, H. Tsai, Proceedings of the Second World Congress on Biosensors 1992, p. 279. [24] L.-K. Chen, C.-L. Liao, C.-G. Lin, S.-C. Lai, C.-I. Liu, S.-H. Ma, Y.-Y. Huang, Y.-L. Lin, Virology 217 (1996) 220. [25] G. Kohler, C. Milstein, Eur. J. Immunol. 6 (1976) 511. [26] P.Y. Shu, L.-K. Chen, S.F. Chang, Y.Y. Yueh, L.J. Chien, C. Chin, T.H. Lin, J.H. Huang, J. Med. Virol. 62 (2000) 224. [27] F. Caruso, E. Rodda, D.N. Furlong, J. Colloid Interface Sci. 178 (1996) 104. [28] L.C. Shriver-Lake, B. Donner, R. Edelstein, K. Breslin, S.K. Bhatia, F.S. Ligler, Biosens. Bioelectron. 12 (1997) 1101. [29] J. Turkova, J. Chromatogr. B 722 (1999) 11.

Acknowledgements We thank the Taiwan National Institute of Preventive Medicine for providing DEN-2 virus and grateful to the Taiwan National Science Council for nancial support (National Science and Technology Program in Pharmaceuticals and Biotechnology, NSC 89-2323B-259-002).