Biosensor for Dengue Virus Detection: Sensitive, Rapid, and Serotype

Feb 16, 2002 - A serotype-specific RNA biosensor was developed for the rapid detection of Dengue virus (serotypes 1−4) in blood samples. After RNA ...
0 downloads 0 Views 146KB Size
Download PDF
Anal. Chem. 2002, 74, 1442-1448

Biosensor for Dengue Virus Detection: Sensitive, Rapid, and Serotype Specific Antje J. Baeumner,*,† Nicole A. Schlesinger,† Naomi S. Slutzki,† Joseph Romano,‡ Eun Mi Lee,‡ and Richard A. Montagna§

Department of Biological and Environmental Engineering, Cornell University, Ithaca, New York, Advanced BioScience Laboratories, Inc., Kensington, Maryland, and Innovative Biotechnologies International, Inc., Grand Island, New York

A serotype-specific RNA biosensor was developed for the rapid detection of Dengue virus (serotypes 1-4) in blood samples. After RNA amplification, the biosensor allows the rapid detection of Dengue virus RNA in only 15 min. In addition, the biosensor is portable, inexpensive, and very easy to use, making it an ideal detection system for point-of-care and field applications. The biosensor is coupled to the isothermal nucleic acid sequence-based amplification (NASBA) technique with which small amounts of virus RNA are amplified using a simple water bath. During the NASBA reaction, a generic sequence is attached to all RNA molecules as described earlier (Wu, S. J.; Lee, E. M.; Putvatana, R.; Shurtliff, R. N.; Porter, K. R.; Suharyono, W.; Watt, D. M.; King, C. C.; Murphy, G. S.; Hayes, C. G.; Romano, J. W. J. Clin. Microbiol. 2001, 39, 2794-2798.). It has been shown earlier that Dengue virus can be detected specifically using two DNA probes: a first probe hybridized with the attached generic sequence and, therefore, bound to every amplified RNA molecule; and a second probe either bound to all four Dengue virus serotypes or chosen to be specific for only one serotype. These probes were utilized in the biosensor described in this publication. For a generic Dengue virus biosensor, the second probe is complementary to a conserved region found in all Dengue serotypes. For identification of the individual Dengue virus serotypes, four serotype-specific probes were developed (Wu, S. J.; Lee, E. M.; Putvatana, R.; Shurtliff, R. N.; Porter, K. R.; Suharyono, W.; Watt, D. M.; King, C. C.; Murphy, G. S.; Hayes, C. G.; Romano, J. W. J. Clin. Microbiol. 2001, 39, 2794-2798.). The biosensor is a membrane-based DNA/RNA hybridization system using liposome amplification. The generic DNA probe (reporter probe) is coupled to the outside of dye-encapsulating liposomes. The conserved or Dengue serotype specific probes (capture probes) are immobilized on a polyethersulfone membrane strip. Liposomes are mixed with amplified target sequence and are then applied to the membrane. The mixture is allowed to migrate along the test strip, and the liposometarget sequence complexes are immobilized in the capture zone via hybridization of the capture probe with target sequence. The amount of liposomes present in the immobilized complex is directly proportional to the amount of target sequence present in the sample and can be 1442 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

quantified using a portable reflectometer. The different biosensor components have been optimized with respect to sensitivity and, foremost, specificity toward the different serotypes. An excellent correlation to a laboratory-based detection system was demonstrated. Finally, the assay was tested using a limited number of clinical human serum samples. Although Dengue serotypes 1, 2 and 4 were identified correctly, serotype 3 displayed low crossreactivity with biosensors designed for detection of serotypes 1 and 4. Dengue virus is a member of the flaviviridae and is one of the most significant causes of arthropod-borne diseases on earth. Dengue virus exists as four antigenically distinct serotypes (Dengue 1-4) and is transmitted in humans by the Aedes aegypti mosquito.2 Dengue-related disease is manifested according to three distinct pathologies: Dengue fever (DF), which is a selflimiting, acute febrile illness characterized by fever, headaches, nausea, and joint pain; Dengue hemorrhagic fever (DHF), which is characterized by extremely high fever, hemorrhagic phenomena, hepatomegaly, and circulatory failure; and Dengue shock syndrome (DSS), a hypovolaemic shock condition brought on by severe plasma leakage. Risk of DHF and DSS is increased with the incidence of secondary infection by a distinct subtype. The mechanism of this disease enhancement is thought to be mediated by increased cellular uptake of secondary infecting virus that is coated with residual antibody present from the primary infection.3,4 It has been estimated that ∼2% of all Dengue infections result in DHF, and the case mortality rate for DHF is ∼5%, with most fatal infections occurring in children.5 Dengue is recognized in over 100 countries and territories, and the worldwide annual infection rate is estimated to be between 50 and 100 million infections per year. In 1998, 1.2 million cases of DF and DHF were reported to * Corresponding author. Telephone: 607-255-5433. Fax: 607-255-4080. Email: [email protected]. † Cornell University. ‡ Advanced BioScience Laboratories, Inc. § Innovative Biotechnologies International, Inc. (1) Wu, S. J.; Lee, E. M.; Putvatana, R.; Shurtliff, R. N.; Porter, K. R.; Suharyono, W.; Watt, D. M.; King, C. C.; Murphy, G. S.; Hayes, C. G.; Romano, J. W. J. Clin. Microbiol. 2001, 39, 2794-2798. (2) WHO. Meeting report of the Scientific Working Group on Dengue; Geneva, Switzerland, April 2000. (3) Halstead, S. B. Rev. Infect. Dis. 1989, 11 (4), S830-839. (4) Hayes, E. B.; Gubler, D. J. Pediatr. Infect. Dis. J. 1992, 11, 311-317. (5) Gubler, D. J.; Clark, G. G. Emerging Infect. Dis. 1995, 1 (2), 55-57. 10.1021/ac015675e CCC: $22.00

© 2002 American Chemical Society Published on Web 02/16/2002

Figure 1. Biosensor assay principle. A DNA capture probe is immobilized on a polyethersulfone membrane. A DNA reporter probe is coupled to the surface of a liposome. When a specific Dengue virus RNA is present (part A), a sandwich is formed between capture probe, mRNA, and reporter probe; thus, liposomes are captured in the capture/detection zone. The number of liposomes is directly proportional to the amount of Dengue RNA present. In part B, it is shown that liposomes are not captured in the detection zone when a nonspecific RNA molecule is present. Thus, no signal will be reported in the detection zone.

the WHO, including 3442 deaths. The WHO considers dengue to be one of the great emerging health challenges in the new millennium,2 and in 1999, the Joint Coordinating Board of TDR (WHO Special Program for Research & Training in Tropical Diseases) voted to add Dengue to its disease portfolio. Most recently, in September 2001, US federal health officials diagnosed 40 cases of locally acquired dengue fever in Hawaii, including six cases from which dengue virus serotype 1 was isolated.6 To date, there is no specific therapeutic treatment for Dengue virus infection. Moreover, a vaccine has yet to be developed. Prevention of the disease has focused largely on mosquito eradication strategies, which were of very limited success. Consequently, rapid and proper diagnosis of the Dengue virus infection is critical to the implementation of a proper treatment and observation protocol. Initial Dengue virus infection symptoms are very similar to those of influenza, measles, malaria, typhus, yellow fever, and other virus infections, which makes the diagnosis based on presenting symptoms problematic. ELISA assays for the detection of IgG and IgM antibodies to Dengue virus are available;7 however, they are compromised by cross-reactivity with other flaviviruses and require at least 5 days postinfection to mount a sufficient immune response to produce detectable antibodies in a patient. Other conventional approaches in Dengue virus diagnostics, such as tissue culture and immunofluorescence,8-10 are limited in terms of specificity, sensitivity, ease of use, and speed. Recently, molecular assays based on nucleic acid amplification have been described. All of these utilize PCR for which the Dengue genomic RNA initially has to be transformed into DNA.11-15 Furthermore, PCR requires thermal cycling instrumentation. Although such instruments are readily available, they are expensive and present engineering challenges for miniaturization and field utility. Additionally, the PCR reaction product is doublestranded DNA, which must first be denatured before being subjected to a probe hybridization-based detection method. Although the TaqMan system as described by Laue et al.12 is very effective at kinetic detection of amplification products produced in real time, the instrumentation required for this is large, expensive and not suitable for field application. In contrast, biosensors based on liposome technology have been shown to be quite successful for the development of rapid,

inexpensive, and field-usable detection systems16,17 and have recently been used in our laboratory for the detection and quantification of RNA molecules.18,19 We describe in this paper the development of a field-usable RNA biosensor for the serotypespecific, sensitive and rapid detection of Dengue virus. Highly specific DNA probes hybridize with Dengue RNA that was amplified using the isothermal NASBA technique. NASBA is a transcription-based amplification system using three enzymes: reverse transcriptase, RnaseH, and the T7 RNA polymerase. It amplifies single-stranded RNA molecules isothermally about a billion-fold in 90 min.18 The biosensor principle and format is shown in Figure 1. Two sets of DNA probes are used in the biosensor. One set of probes (reporter probes) is coupled to the surface of dye-entrapping liposomes. The other set of probes (capture probes) is immobilized on a polyethersulfone membrane strip in the capture zone. In a biosensor assay, liposomes are mixed with the amplified Dengue virus RNA. This mixture is applied to the polyethersulfone membrane and migrates along the membrane strip due to capillary action. Upon passing the capture zone, a sandwich is formed of reporter probe-tagged liposomes, RNA, and immobilized capture probe. Thus, in the capture zone, (6) Center for Disease Control and Prevention. Notice: Dengue Fever, Hawaii, Oct. 2, 2001. (7) Chakravarti, A.; Gur, R.; Berry, N.; Mathur, M. D. Diagn. Microbiol. Infect. Dis. 2000, 36 (4), 273-274. (8) Vene, S.; Mangiafico, J.; Nicklasson, B. Clin. Diagn. Virol. 1995, 4 (1), 43-50. (9) Porter, K. R.; Widjaja, S.; Lohita, H. D.; Hadiwijaya, S. H.; Maroef, C. N.; Suharyono, W.; Tan, R. Clin. Diagn. Lab. Immunol. 1999, 6 (5), 741-744. (10) Young, P. R.; Hilditch, P. A.; Bletchly, C.; Halloran, W. J. Clin. Microbiol. 2000, 38 (3), 1053-1057. (11) Kow, C. Y.; Koon, L. L.; Yin, P. F. J. Med. Entomol. 2001, 38 (4), 475-479. (12) Laue, T.; Emmerich, P.; Schmitz, H. J. Clin. Microbiol. 1999, 37 (8), 25432547. (13) Killen, H.; O’Sullivan, M. A. J. Virol. Methods 1993, 41 (2), 135-146. (14) Lanciotti, R. S.; Calisher, C. H.; Gubler, D. J.; Chang, G. J.; Vorndam, A. V. J. Clin. Microbiol. 1992, 30 (2), 545-551. (15) Henschal, E. A.; Polo, S. L.; Vorndam, V.; Yaemsiri, C.; Innis, B. L.; Hoke, C. H. Am. J. Trop. Med. Hyg. 1991, 45 (4), 418-428. (16) Roberts, M. A.; Durst, R. A. Anal. Chem. 1995, 67, 482-491. (17) Ba¨eumner, A. J.; Schmid, R. D. Biosens. Bioelectron. 1998, 13 (5), 519529. (18) Esch, M. B.; Baeumner, A. J.; Durst, R. A. Anal. Chem. 2001, 73 (13), 3162-3167. (19) Baeumner, A. J.; Cohen, R. N.; Miksic, V.; Min, J. H. Biosens. Bioelectron., submitted.

Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

1443

Figure 2. Diagram showing the location of all reporter and capture probes in the Dengue virus RNA. The generic reporter probe hybridizes to a generic sequence added to the Dengue RNA during a NASBA reaction and is thus present in every amplified Dengue virus RNA. The reporter probe is coupled to dye-entrapping liposomes. Probes specific for types 1-4 and the conserved probe are used as capture probes in the biosensors and, therefore, immobilized on the membrane strip. Thus, the generic Dengue virus biosensor is built using the generic reporter probe and the conserved capture probe. Serotype-specific biosensors are built using the generic reporter probe and one of the four serotypespecific capture probes.

Table 1. DNA Sequences of All Reporter and Capture Probes Used in the Dengue Biosensorsa sequence name sequenceb

generic conserved sequenceb serotype 1c serotype 2c serotype 3c serotype 4c cold probe 1 cold probe 4

DNA sequence (5′-3′) gAT gCA Agg TCg CAT ATg Ag AAA CAg CAT ATT gAC gCT ggg ggg AAg CTg TAT CCT ggT ggT AAg g ATg AAg CTg TAg TCT CAC Tgg AAg g Agg gAA gCT gTA CCT CCT TgC AAA g gAg gAA gCT gTA CTC CTg gTg gAA g ggg AAg CTg TAT CCT ggT ggT AAg g gAg gAA gCT gTA CTC CTg gTg gAA g

a All sequences are written in 5′-3′ direction. The reporter probe was modified with an amine group at the 3′ end. The capture probes were labeled with a biotin moiety at the 5′ end. b Tagged with an amino group at 3′ end if used as reporter probe or tagged with a biotin at the 5′ end if used as capture probe. c Tagged with a biotin at the 5′ end (serotype specific capture probes).

the number of liposomes caught correlates directly to the amount of RNA present in the sample (Figure 1A). Liposomes not bound to the RNA pass through the capture zone and accumulate at the end of the membrane strip (Figure 1B). The biosensor assay is optimized regarding hybridization conditions to allow serotypespecific identification. Signals are correlated with a laboratorybased detection system for Dengue virus based on electrochemiluminescence, and a limited number of clinical samples are investigated. MATERIAL AND METHODS All general chemicals and buffer reagents were obtained from Sigma Company. Lipids were obtained from Avanti Polar Lipids, Alabaster, AL. Sulforhodamine B, streptavidin, etc., were purchased from Molecular Probes Company, Eugene, OR. Membranes were obtained from Pall/Gelman Company, Port Washington, NY. The BR-10 reflectometer was purchased from ESECO (λ ) 560 nm), Cushing, OK. All oligonucleotides were purchased from the Biotechnology Center, Cornell University, Ithaca, NY. DNA Probes. DNA oligonucleotides (probes) for detection had been designed earlier.1 Their sequences are given in Table 1. A generic sequence was introduced into the Dengue virus RNA during the NASBA reaction and was, therefore, present in every amplified Dengue RNA. A generic probe complementary to this region was used as reporter probe in all biosensors (generic 1444

Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

Dengue and serotype-specific Dengue biosensors) and coupled to liposomes via an amine group at the 3′ end (see below). The conserved probe was able to hybridize to a region in the Dengue RNA that is present in all four serotypes. It was used as the capture probe in the generic Dengue biosensor and, therefore, was modified with a biotin moiety at the 5′ end. Four serotype-specific capture probes were used, which were also modified with biotin at the 5′ end. The sequence layout is shown in Figure 2. All probes were obtained desalted and lyophilized. In addition to the detection probes, DNA “cold probes” were used in the serotype-specific biosensor assays. Their sequences are also provided in Table 1. Cold probes are nonlabeled oligonucleotides that can hybridize with the Dengue RNA. They are typically used to prevent nonspecific binding of a nucleic acid sequence to bioassay detection probes. For example, if Dengue virus serotype 1 RNA bound nonspecifically to the serotype 4 capture probe, a false positive signal would be obtained. Therefore, a cold probe capable of binding to Dengue serotype 1 could be added to the serotype 4 biosensor assay. Thus, the serotype 1 RNA would bind to the cold probe and not to the serotype 4 capture probe, thus preventing the false positive signal. Two cold probes were used in the biosensor assay: one consisting of the sequence for detection probe 4, the other for detection of probe 1. They were added in varying concentrations to the serotype-specific assays to decrease the likelihood of nonspecific binding. These probes were unlabeled and obtained desalted and lyophilized. Liposome Preparation and Nucleic Acid Coupling. Liposomes were prepared following the protocol described by Siebert et al.20 Briefly, liposomes were prepared using the reversed-phase evaporation method. Sulforhodamine B (100-150 mM) was dissolved in potassium phosphate buffer, pH 7.5, and entrapped in the liposomes. Subsequently, the liposomes were extruded at elevated temperatures through polycarbonate filters (3 and 0.4 µm, nine times each) for sizing using a miniextruder (Avanti Polar Lipids, Alabaster, AL), purified away from unencapsulated dye molecules by gel filtration using Sephadex G50 columns, and dialyzed against potassium phosphate buffer with an osmolarity 50-100 mmol/kg higher than the osmolarity of the encapsulant solution. The osmolarity was adjusted using NaCl and sucrose. (20) Siebert, S. T. A.; Reeves, S. G.; Durst, R. A. Anal. Chim. Acta 1993, 282, 297-305.

Table 2. Optimized Hybridization Solutions (Running Buffer) for the Different Dengue Virus Biosensor Assays assay specificity

composition

general Dengue biosensor serotype 1 serotype 2 serotype 3 serotype 4

10% formamide, 6× SSC, 0.2% Ficoll type 400, 0.2 M sucrose, 0.01% Triton X-100 10% formamide, 3× SSC, 0.2% Ficoll type 400, 0.2 M sucrose, 0.01% Triton X-100 25% formamide, 5× SSC, 0.2% Ficoll type 400, 0.2 M sucrose, 0.01% Triton X-100 25% formamide, 5× SSC, 0.2% Ficoll type 400, 0.2 M sucrose, 0.01% Triton X-100 5% formamide, 5× SSC, 0.2% Ficoll type 400, 0.2 M sucrose, 0.01% Triton X-100

For the coupling of the reporter probe to the liposome surface, a lipid tagged with a maleimide group (N-(4-(p-maleimidylmethyl)cyclohexane-1-carbonyl)-diphosphatidyl palmitoyl ethanolamine [MMCC-DPPE]) was used during the liposome preparation and, thus, incorporated into the lipid bilayer. Typically, a 3-4 mol % concentration of the overall amount of lipids was used. After liposomes were prepared and purified using gel filtration and dialysis, activated oligonucleotides were coupled to the maleimide groups available on the outer surface of the liposomes. First, the reporter probe (bearing an amine group at the 3′ end) was derivatized using a sulfhydryl group. The probe was dissolved in 0.05 M potassium phosphate buffer, pH 7.8, containing 1 mM EDTA to a concentration of 300 nmol/mL. A 30-nmol portion of probe was mixed with 90 nmol of N-succinimidyl-S-acetylthioacetate (SATA) dissolved in DMSO. This mixture was incubated at room temperature for 1.5 h. Second, the thiol was deprotected in a deacetylation step. A fresh solution of 0.5 M hydroxylamine hydrochloride, 25 mM EDTA in 0.1 M potassium phosphate buffer with pH adjusted to 7.5 was prepared. The amount of solution added to the reporter probe mixture was 1/10 of the probe solution volume. The mixture was incubated for 2 h at room temperature. Third, the maleimide-derivatized liposomes were conjugated with SH-tagged reporter probes. To obtain the desired surface density of reporter probe, the appropriate amount of liposomes was added to the probes (for example, for a 0.4 mol % tag, 10.5% of liposomes were used). The pH of the reporter probe solution was adjusted to 7.0 with 0.5 M KH2PO4. Subsequently, the two solutions were mixed and incubated at room temperature for 3 h and then overnight at 4 °C. In a final step, ethylmaleimide and Tris were added to quench the excess SH groups on the reporter probes and the unreacted maleimide groups on the liposomes, respectively. The liposomes were purified from unreacted reporter probe on a Sepharose CL4B column and, finally, dialyzed against the appropriate buffer overnight in the dark at room temperature. The liposomes were stored at 4 °C in the dark. Different concentrations of reporter probes were tagged to the liposomes, ranging from 0.05 to 1 mol % tag. Since the liposome concentration of the liposome stock solution could vary between the different preparations, the amount of liposomes used per assay was evaluated using a spectrophotometer (DU520 Beckman, Fullerton, CA) at a wavelength of 570 nm and adjusted so that each assay contained the equivalent amount of liposomes. Membrane Preparation. The membranes were cut into 4.5 × 80 mm strips and then coated with a mixture of streptavidin and biotinylated capture probes. A 22.5-pmol portion of streptavidin and a 67.5-pmol portion of capture probe were immobilized per membrane strip by pipetting 1-1.5 µL of the mixture onto the membrane ∼2.5 cm from the bottom. The streptavidin-capture

probe mixture was incubated for at least 15 min at room temperature in a sodium carbonate buffer (0.4 M NaHCO3/NA2CO3 with 5% methanol). After its application onto the membranes, the membranes were dried for 5 min at room temperature and then for 1.5 h in a vacuum oven (15 psi) at 55 °C. Subsequently, the membranes were incubated in a blocking solution for 30 min containing poly(vinylpyrrolidone) (PVP), casein, gelatin, methanol, and methylated BSA at different concentrations. Optimal conditions were found to be 0.5% PVP, 0.015% casein in Tris-buffered saline (TBS: 20 mM Tris, 150 mM NaCl, 0.01% NaN3, pH 7-7.5). The membranes were blotted dry and finally dried in a vacuum oven (15 psi) at 30 °C for 2 h. The membranes were stored in vacuum-sealed bags at 4 °C until use. Biosensor Assay Format. A lateral-flow assay was developed. First, 2 µL of liposomes, 2 µL of amplicon (amplified Dengue virus RNA), 2 µL of hybridization solution (master mix), and 0.66 µL per cold probe were incubated for 10 min at room temperature. The mixture (a total of 6 µL for the generic biosensor; 6.66 µL for biosensor assays for serotypes 1, 2 and 4; or 7.3 µL for serotype 3) was pipetted onto the membrane strip ∼8 mm from the center of the capture zone (toward the front of the strip). Immediately afterward, 30-40 µL of running buffer was added to the front of the strip. After all of the running buffer traversed the test strips (∼10 min), the capture zones were analyzed using the BR-10 reflectometer or evaluated visually. Both the running buffer and the master mix solutions contained formamide, SSC (1× SSC contains 15 mM sodium citrate and 150 mM NaCl, pH 7.0), Ficoll type 400, sucrose, and Triton X-100. The concentration of each compound was varied for optimization of the different biosensor assays. Additionally, different hybridization conditions and liposome and target sequence concentrations were investigated. The master mix was optimized and had the same composition for all assays (60% formamide, 6× SSC, 0.8% Ficoll type 400, 0.15 M sucrose, 0.01% Triton X-100). Optimal running buffer concentrations for the different serotype specific and the general Dengue virus biosensor assay are provided in Table 2. Dengue Virus Samples. Seed stocks of the four serotypes of dengue virus were prepared in Vero cells, and plaque titer (pfu ) plaque forming units) was determined by a plaque assay.21 The virus strains were Dengue 1 (Hawaii strain), Dengue 2 (New Guinea C strain), Dengue 3 (CH53489), and Dengue 4 (341750). The actual Dengue virus nucleic acids were obtained from the processing of a virus stock in a lysis buffer (Organon Teknika, Boxtel, NL) and different concentrations of virus stocks were prepared by dilution in lysis buffer to obtain 1, 10, 100, or 1000 pfu/mL of lysis buffer. Nucleic acid isolation was as described,22 (21) Eckels, K. H.; Brandt, W. E.; Harrison, V. R.; McCown, J. M.; Russel, R. K. Infect. Immunol. 1976, 14, 1221-1227. (22) Boom, R.; Sol, C. J.; Salimans, M. M.; Jansen, C. L.; Wertheim-van Dillen, P. M.; van der Norda, J. J. Clin. Microbiol. 1990, 20, 495-503.

Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

1445

with the final nucleic acid extract eluted in 50 µL of elution buffer. A 5-µL portion of the extract was then amplified using NASBA as previously described.1 All amplification reaction products were analyzed using electrochemiluminescence (ECL) detection prior to shipment to our laboratory at Cornell University. At Cornell University, no actual live virus was used. Only amplified short RNA sequences were present in the lab, which did not pose any health risk. Briefly, 5 µL of a 1:20 dilution (in detection diluent) of NASBA amplicon was mixed with 10 µL of magnetic-beadlabeled capture probe and 10 µL of ECL-labeled reporter probe. The mixture was incubated for 5 min at 60 °C and then 30 min at 41 °C in a shaking waterbath. Subsequently, 300 µL of hybridization buffer (Organon Teknika, Boxtel, NL) was added, and the samples were loaded into the auto sampler of the NucliSense Reader for detection and quantification.1 Human serum clinical samples were treated following the same procedure. RESULTS AND DISCUSSION Generic Dengue Virus Biosensor Development. Two generic biosensors were developed. The first generic Dengue biosensor used the conserved probe coupled to liposomes and the generic probe immobilized on the polyethersulfone membranes. Thus, all four Dengue serotypes could be detected but not differentiated from each other. First, the master mix was optimized. Varying the SSC concentration from 3× to 15×, the optimum was found at 12× SSC. Readings taken with the reflectometer were 30 for the positive sample and 9 for the negative. When we changed the Ficoll type 400 concentrations from 0.3 to 1.0%, we found that higher concentrations increased the signal, but that concentrations above 0.8% led to some nonspecific liposome aggregation. Thus, a Ficoll concentration of 0.8% was chosen as optimal. Then, formamide, sucrose, and Triton concentrations were optimized. The optimized master mix hybridization solution contained 12× SSC, 15% formamide, 0.8% Ficoll type 400, 0.3 M sucrose, and 0.01% Triton X-100. Second, the running buffer was optimized. Although the differences between varied conditions were small, it was shown that the signal at 6× SSC and 10% formamide (all buffers contained 0.2% Ficoll type 400, 0.2 M sucrose, and 0.01% Triton X-100) yielded the best results. In general, optimization experiments were based on small differences in conditions and, thus, small differences in signals. However, taken together, a significant signal improvement was achieved. Finally, a dose-response curve using one Dengue virus sample amplicon was made using the optimized running buffer (6× SSC, 10% formamide, 0.2% Ficoll type 400, 0.2 M sucrose, and 0.01% Triton X-100) and the optimized master mix (Figure 3). With the same assay conditions, a second generic Dengue biosensor was developed using the generic probe coupled to liposomes and the conserved probe immobilized on the polyethersulfone membranes. A similar dose-response curve was obtained. These “generic” liposomes were subsequently used in all serotypespecific assays. Since the generic sequence is introduced into the Dengue RNA during the NASBA reaction, all amplicons were able to hybridize to the liposome-bound generic probe. Thus, in respect to future applications of the biosensor principle presented in this paper, liposomes coupled to the generic probe can be used as a universal analytical tool for all NASBA reactions in which the generic sequence is added to the original RNA. 1446 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

Figure 3. Dose response curve of the generic Dengue virus biosensor using liposomes tagged with the conserved probe and membranes with immobilized generic probe. A Dengue virus sample amplicon was diluted in H2O prior to the biosensor assay, and 2 µL of amplicon solution was then added to the hybridization mixture. Signals were taken after completion of each assay using a reflectometer.

Serotype-Specific Biosensor Development. It was necessary to optimize the serotype-specific biosensors in order to improve the signal-to-noise ratio and eliminate possible cross-reactive binding of other serotypes (e.g., Dengue serotype 2 biosensor should only give a positive signal when Dengue 2 is present in a sample and not with Dengue 1, 3, or 4). Therefore, the stringency of the hybridization buffers was adjusted. Since in all four cases, the same liposomes were used (i.e., liposomes coupled to the generic probe), the same master mix was applied for all four serotype specific biosensors. It contained 12× SSC, 15% formamide, 0.8% Ficoll Type 400, 0.3 M sucrose, and 0.01% Triton X-100. However, the running buffers were adjusted to improve specific binding and decrease nonspecific binding (derived from liposomes binding to the membrane, or liposomes cross-reacting with the immobilized capture probes) in the four biosensor assays. The optimized running buffer solutions are provided in Table 2. Finally, to further decrease the serotype cross-reactions (such as Dengue serotype 2 binding with biosensor specific for serotype 4) “cold probes” were added to the master mix. Two cold probes, cold probe 1 and cold probe 4 (Table 1), had been designed earlier.1 These were added to the biosensor assays in varying concentrations. The optimized assays were investigated with Dengue virus samples containing 100-1000 pfu/mL. The results are shown in Figure 4a. Background signals from assays without any Dengue RNA were subtracted (signals ranging between 8 and 12). The optimized cold probe concentrations for each serotype specific assay are given in Table 3. An example of actual biosensor test strips is shown in Figure 4b. Four serotype-specific biosensors and the generic biosensor are shown for the analysis of Dengue serotype 2. As can be seen, only the biosensor specific for serotype 2 and the generic biosensor give a strong positive signal. All other biosensors show only background binding similar to the negative control. Biosensors for Dengue serotypes 2, 3, and 4 were highly specific (refer to Figure 4a); however, the biosensor for Dengue serotype 1 had cross-reactivity with Dengue serotype 3. This was also found later in the analysis of clinical samples (see below).

Figure 4. (a) Specificity analysis of the serotype-specific Dengue biosensor assays. Liposomes tagged with the generic probe were used in all assays. Serotype-specific capture probes were immobilized on the membrane. Optimized master mix and running buffers were used in each assay, and the appropriate cold probe concentrations were added to the specific assays (refer to Table 3 for concentrations). A background signal obtained from assays without any Dengue amplicon (between 8 and 12) was subtracted from the reflectometer signals shown. (b) Specificity analysis of the serotype-specific Dengue biosensor assays. Serotype 2 was analyzed with all serotype-specific biosensors and with the generic biosensor (with conserved probe immobilized on the membrane). A negative control was analyzed with the generic biosensor using H2O instead of Dengue amplicon to determine unspecific binding of liposomes to the membranes. Liposomes tagged with the generic probe were used in all six assays. Liposomes, master mix, and sample were mixed, incubated at room temperature for 10 min, and then applied to the membranes in location A. Subsequently, running buffer was added to location B. Signals were taken using the reflectometer at location C after ∼5-10 min. Table 3. Concentration of Cold Probes in the Different Serotype-Specific Biosensor Assays biosensor assay specificity serotype 1 serotype 2 serotype 3 serotype 4

concn in master mix, µM cold probe 4 cold probe 1 5 0.5 0.5

5 0.5

Thus, in future experiments, a cold probe 3, will be designed and added to the biosensor 1 assay (and possibly 4) in order to reduce or eliminate the cross reactivity observed between serotypes 1 and 3. Correlation to ECL Signals. Using the biosensor specific for Dengue 1, a correlation of our biosensor to a more expensive and elaborate laboratory technique ECL (as described above) was investigated. Ten samples (seven positive and three negative) were analyzed using ECL, frozen, and shipped to our laboratory. After thawing, each sample was analyzed using the specific biosensor. Amplification of RNA by NASBA is kinetically independent from the original concentration of RNA present in the sample. Thus, qualitative but not quantitative signals can be obtained (i.e., either positive or negative signals). However, the amplification reaction is very dependent on optimal conditions. Thus, positive signals can have a wide range in signal height; that is, conditions slightly different from the optimum result in significantly different (lower) RNA concentrations at the end of the NASBA reaction. Thus, an ECL signal above 10 000 and a biosensor signal above 10 are considered positive. As shown in Figure 5, an excellent correlation

Figure 5. Correlation between reflectometer signals and ECL values. Ten samples were analyzed with the ECL and the biosensor detection systems. Seven samples contained Dengue serotype 1 amplicon; three samples were negative controls (H2O in NASBA reaction). Approximate cutoff values indicating the line below which samples are identified negative and above which samples are positive are given in the figure.

is obtained. Samples below the cutoff line are identified as negatives (i.e., they do not contain Dengue virus), and samples with values above the cutoff line are positive. All three negative and all seven positive samples correlated very well. Thus, the biosensor can be used for data analysis and quantitation and will be especially useful for field applications. Although the ECL Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

1447

Table 4. Analysis of Clinical Samples with the Dengue Serotype Specific Biosensorsa

a

Dengue serotype in clinical sample (ECL value)

biosensor for Dengue 1

biosensor for Dengue 2

biosensor for Dengue 3

biosensor for Dengue 4

1 (1 104 300) 1 (235 300) 1 (981 600) 2 (1 535 700) 2 (872 400) 2 (308 500) 3 (1 637 100) 3 (740 400) 3 (357 200) 4 (207 300) 4 (101 000)

48 29 40 0 0 2 13 10 9 0 2

9 5 6 24 20 17 6 5 5 3 4

0 0 0 0 2 0 26 19 28 0 0

0 0 0 1 2 4 12 10 9 20 21

The background signal of a negative control (assay with H2O instead of Dengue amplicon) was subtracted from the signals.

detection system has a larger dynamic range (i.e., positive values of 10 000-3 000 000) than the biosensor (i.e., positive values of 10-50), a clear distinction between different RNA concentrations can be made with the biosensor, and a significant and reliable difference between negative and positive signals is obtained. Thus, comparing a biosensor that costs less than $1.00 with a $35 NucliSense reader that costs more than $35 000 makes a strong case for the potential use of the biosensor assay in future Dengue virus detection. Reproducibility Analysis. The reproducibility of serotypespecific biosensor assays was tested using biosensors for serotypes 1, 2, and 3. For Dengue 1 (with a corresponding ECL value of 215 068), the signal was 21 ( 2.6. For Dengue 2 (with a corresponding ECL value of 2 793 300), the signal was 30 ( 1.7. Finally, for Dengue 3 (with a corresponding ECL value of >3 000 000), the signal was 31 ( 2.5. Analysis of Clinical Samples. A limited number of clinical samples (n ) 11) were obtained and evaluated to assess the biosensor’s ability to analyze “real world” samples. The samples were amplified using NASBA and analyzed with the ECL detection system prior to a biosensor analysis. The results, summarized in Table 4, reveal that the biosensor assays for Dengue virus 1, 2, and 4 gave very clear results (i.e., the serotypes were identified with no cross reactivity). Analysis of Dengue serotype 3, however, gave more ambiguous results, since low positive signals also were obtained in biosensors designed for detection of Dengue serotypes 1 and 4. This result reconfirms our earlier results described in the specificity assays (Figure 4a). Therefore, in the future we will design a “cold probe 3” that will be added to assays for serotypes

1448 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

1 and 4 in order to suppress nonspecific binding of serotype 3 amplicon to capture probes specific for serotypes 1 and 4. CONCLUSION We have presented in this study the development of a serotypespecific and very sensitive biosensor for the rapid detection of Dengue virus. A very good correlation to a much more elaborate and expensive laboratory-based detection system (ECL) was demonstrated, which had earlier been shown to detect as few as 10 plaque forming units (pfu)/mL. Clinical sample analyses as well as specificity analyses showed that Dengue serotypes 1, 2, and 4 were identified without any difficulty, but analysis of serotype 3 gave some false positive signals with biosensors specific for serotype 1 and 4. Thus, in the future, we will optimize the serotypespecific biosensor to avoid cross-reactivity that occurred with Dengue serotype 3 samples, and we will analyze a larger number of clinical samples. ACKNOWLEDGMENT The authors acknowledge financial support for this project from Innovative Biotechnologies International, Inc., from the National Institute of Allergy & Infectious Diseases (NIAID), Bethesda, MD, and from the New York State CAT, Biotechnology Program at Cornell University.

Received for review November 7, 2001. Accepted January 15, 2002. AC015675E