Tapered Optical Fiber Sensor Using Near-Infrared Fluorophores To

Department of Electrical Engineering, University of Maryland, College Park, Maryland 20742 ... Cabtech, Inc., 9105 Fall River Lane, Potomac, Maryland ...
0 downloads 0 Views 66KB Size
Download PDF
Anal. Chem. 1998, 70, 2031-2037

Articles

Tapered Optical Fiber Sensor Using Near-Infrared Fluorophores To Assay Hybridization Saeed Pilevar*

Bragg Grating Technologies, 3M Specialty Optical Fibers, 206 West Newberry Road, Bloomfield, Connecticut 06002-1331 Christopher C. Davis

Department of Electrical Engineering, University of Maryland, College Park, Maryland 20742 Frank Portugal

Cabtech, Inc., 9105 Fall River Lane, Potomac, Maryland 20854

We present an all-fiber hybridization assay sensor that relies on the evanescent field excitation of fluorescence from surface-bound fluorophores. The evanescent field is made accessible through the use of a long, adiabatically tapered single-mode fiber probe. A laser diode with a 785-nm wavelength is used in a pulsed mode of operation to excite fluorescence in the tapered region of a fiber probe using the near-infrared fluorophore IRD 41. We have used various chemical treatments to prepare the tapered fiber surface for chemical, as well as physical, binding of fluorophores. We have carried out real-time hybridization tests for IRD 41-labeled oligonucleotide, at various probe concentrations, binding to complementary oligonucleotide cross-linked to the tapered fiber surface. The biospecificity of our sensor is confirmed through hybridization tests with a control oligonucleotide. Short oligonucleotides (20-mer) bound to the fiber surface have been used to detect near-IR dye-labeled complementary sequences at subnanomolar levels. Sandwich assays with Helicobacter pylori total RNA were conducted to examine the capability of the biosensor for detecting bacterial cells using rRNA as the target. The results indicate that this fluorosensor is capable of detecting H. pylori in a sandwich assay at picomolar concentrations. Optical fiber-based fluorosensors respond to spectral changes in light coupled to or from the fiber. Evanescent wave sensors utilize evanescent field excitation, where a portion of the light traveling in a fiber core penetrates into the surrounding medium, with the power of the evanescent field decaying exponentially from the fiber core into the cladding medium. This evanescent field interacts with dye molecules outside the fiber core, whose distance from the fiber surface is generally smaller than the wavelength * To whom correspondence should be addressed. Tel.: (860) 243-8822. E-mail: [email protected]. S0003-2700(97)00996-7 CCC: $15.00 Published on Web 04/16/1998

© 1998 American Chemical Society

of the excitation light. Fluorescence couples back into the fiber core for transmission to a detection system. Coupling of sources outside the cladding by this method has been predicted by a wave model of the light interaction.1 If a dye is immobilized at a fiber core-cladding interface, the structure serves as a sensor for various factors affecting its fluorescence parameters. When moieties with specific binding sites are immobilized on the surface as well, the unit senses a change in the excitation of the fluorophore. The overall sensitivity of the structure depends on the modes of light propagating in the fiber and the thickness of the cladding,2 as well as the chemical interaction between the fluorophore and the immobilized binding sites. To gain maximum access to the evanescent field and enhance detection sensitivity, a portion of the fiber cladding has to be removed.3, 4 Polished fiber half-blocks, biconical tapers, and single-ended tapered fibers (referred to here as “tapered fibers”) are commonly used for removing the cladding locally. Fluorosensors using acid-etched tapered multimode fiber probes have been reported.5 According to the model described by Marcuse,1 the efficiency of the coupling of cladding fluorescence to the guided modes of a multimode fiber depends on the parameter known as the normalized frequency, or V number, which is used to estimate the number of the guided modes of the fiber and is defined as

V ) 2π/λa(n12 - n22)1/2

(1)

where a is the fiber core radius, n1 and n2 are the fiber core and (1) Marcuse, D. J. Lightwave Technol. 1988, 6, 1273-1279. (2) Stewart, G.; Norris, J.; Clark, D. F.; Tribble, M.; Andonovic, I.; Culshaw, B. Int. J. Optoelectron. 1991, 6, 227-238. (3) Hale, Z. M.; Payne, F. P. Sens. and Actuators B 1994, 17, 233-240. (4) Pilevar, S.; Fielding, A.; Davis, C. C. Technical Digest CLEO-96; Anaheim, California, 1996; p 292. (5) Anderson, G. P.; Golden, J. P.; Lynn, P. C.; Ligler, F. IEEE Eng. Med. Biol. 1994, 13, 358-363.

Analytical Chemistry, Vol. 70, No. 10, May 15, 1998 2031

cladding refractive indexes, respectively, and λ is the free-space wavelength of the light coupled into the fiber. The evanescent field of a multimode fiber is extremely weak. For a fiber with a V number of 100, the fraction of power available in the evanescent field is 2% of the total power in the fiber.6 Higher-order modes in these fibers offer the advantage of a large number of reflections per unit length and, thus, a long interaction length with the external medium. However, the use of multimode fibers has a number of disadvantages. In a multimode sensor arrangement, the lower-order modes are confined to the central core region, limiting their interaction with the surroundings. This means that the total fraction of power that interacts with the external medium depends on the distribution of modal excitation and the fraction of each mode outside the core. Also, multimode fibers are not compatible with many in-line fiber components, such as singlemode fiber couplers, polarizers, and attenuators. The use of single-mode fibers can avoid the negative aspects of multimode fibers while retaining their advantages. Much biosensor work has reported the use of fluoresceinlabeled probes for detection of hybridization on large-core, multimode fibers.7,8 Fluorescein has the disadvantage, however, of absorbing light and fluorescing in spectral regions where naturally occurring biological materials also absorb light and fluoresce. The use of fluorophores with longer excitation wavelengths, such as near-infrared (Near-IR) dyes, helps to circumvent the problem of natural background fluorescence by fluorescing in the 750-850-nm range, which is well beyond the range where natural fluorescence occurs. Furthermore, to cause fluorescein to fluoresce, relatively large, bulky, and expensive lasers, such as an argon ion laser, are required. In other studies using methods described by Henke et al.9 intercalating agents such as ethidium bromide are used. Intercalators often have two potential disadvantages: they can used only after the hybrid has formed, thus lengthening detection times, and they may be hazardous to handle. In this work, we have developed a fluorosensor using a long, adiabatically tapered single-mode fiber probe. The fluorophore used with our sensor is IRD 41 (LI-COR, Inc., Lincoln, NE), with excitation wavelength around 787 nm and emission wavelength around 807 nm. Near-IR fluorophores have excitation and emission spectra that peak in the Near-IR region and are particularly desirable for examining clinical and other biological samples: naturally occurring biological materials show no emission of fluorescence when illuminated with electromagnetic radiation in this region of the spectrum. Because of their longer excitation wavelengths and the availability of diode lasers as excitation sources, Near-IR fluorophores are now widely used as fluorescent labels for on-line DNA sequencing,10 and also in on-line fluorescence lifetime determination of components separated by capillary (6) Paul, P. H.; Kychakoff, G. Appl. Phys. Lett. 1987, 51, 12-14 (7) Ferguson, J. A.; Boles, T. C.; Adams, C. P.; Walt, D. R. Nature Biotechnol. 1996, 14, 1681-1684. (8) Abel, A. P.; Weller, M. G.; Duveneck, G. L.; Ehrat, M.; Widmer, H. M. Anal. Chem. 1996, 68, 2905-2912. (9) Henke, L.; Piunno, P. A. E.; McClure, A. C.; Krull, U. J. Anal. Chim. Acta 1997, 344, 201-213. (10) Middendorf, L. R.; Bruce, J. C.; Bruce, R. C.; Eckles, R. D.; Grone, D. L.; Roemer, S. C.; Sloniker, G. D.; Steffens, D. L.; Sutter, S. L.; Brumbaugh, J. A.; Patony, G. Electrophoresis 1992, 13, 487-494.

2032 Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

Figure 1. Schematic of the tapered fiber-optic hybridization assay sensor.

electrophoresis.11 In contrast to the sensor described earlier,9 the system reported here detects hybrid formation as it occurs, and the dye is not classified as hazardous. EXPERIMENTAL SECTION Instrumentation. One of the most important steps in the development of our fiber hybridization assay sensor has been a reproducible fabrication technique for making long, adiabatically tapered fibers. The fiber, which is heated before and during drawing, is mounted in a commercial micropipet puller (P-87, Sutter Instrument Co., Novato, CA). A carbon dioxide (CO2) laser (MIRL-100, Advanced Kinetics, CA) is used as the heat source. By focusing the CO2 laser beam on the fiber and controlling the pulling force and velocity with the micropipet puller, a highly reproducible, adiabatically tapered, almost cylindrical fiber can be drawn. The tapering angle can easily be adjusted by changing the pulling rate, pulling force, and laser beam spot size. Using this heating and drawing technique, both the core diameter and the cladding outer diameter are decreased. The fractional change in the core diameter is approximately equal to the fractional change in the cladding outer diameter. The heating-and-drawing method allows the drawing of 2-3-cm-long tapered fibers. These long, single-mode fibers typically have an 8-10-µm diameter along a substantial part of their tapered length. A schematic of the fiber hybridization assay sensor is shown in Figure 1. A 785-nm laser diode (Diolite 800-40, Liconix, Santa Clara, CA) is used to excite the fluorescence in the tapered region of the fiber probe. An excitation filter (S10-790-F, Corion Corp., Holliston, MA) is used to block unwanted background light from the laser diode source. Laser light is coupled into a 3-dB single-mode fiber directional coupler (Gould Electronic Inc., Millersville, MD) via a biconvex silica lens (SBX019AR.16, Newport Corp. Irvine, CA). The fiber, single-mode at 780 nm (F-SE, Newport Corp.), is angle cleaved at the input port to remove the 4-% back reflection that results at glass-air interfaces. The alternative, bulk optic arrangement using an off-axis parabolic reflector,5 is very costly and requires (11) Soper, S. A.; Legendre, B. L., Jr.; Williams, D. C. Anal. Chem. 1995, 67, 4358-4365.

sensitive alignment. The fluorescence originating from the tapered end of the fiber is directed by the fiber coupler through a SuperNotch holographic filter (SNHF-785-1.0, Kaiser Optical Systems, Inc., Ann Arbor, MI) and onto a photomultiplier tube (PMT, R636, Hamamatsu, Bridgewater, NJ). For dye immobilized at the end of the tapered fiber, there may be only a limited amount of light that can be measured before the immobilized dye is bleached. In one experiment, Shiver-Lake et al.12 examined the effect of photobleaching on three different dyes, TRITC, CY5, and NN 328, with excitation wavelengths centered at 514, 654, and 780 nm, respectively, and emission wavelengths centered around 580, 670, and 800 nm, respectively. The authors showed that the Near-IR dye NN 328 was degraded the most by photobleaching when exposed for 2 min to a 780-nm source. Care has to be taken to minimize the exposure of such dyes to room lights and laser light. Many researchers, such as Walczak et al.13 and Anderson et al.14 have demonstrated the benefits of applying a shuttered excitation source with fiber fluorosensors, even for fluorophores with visible excitation wavelengths. Hence, minimizing the continuous illumination of fluorescent dye by the laser source to reduce photobleaching is a standard procedure. To enhance our signal-to-noise (S/N) ratio and to minimize such photobleaching effects, we have employed a pulsed mode detection scheme, rather than continuous wave (CW) operation. In this scheme, the laser diode, which is intensity modulated, sends single pulses of short duration (50-60 µs) to the sensing region. In our experiment, 2-3 pulses were enough for a reliable measurement. Therefore, the laser was manually triggered, and the duty cycle of the pulse train was extremely small. The peak power of the pulse is approximately 1 mW, and, hence, the number of photons delivered to the sensor unit is much smaller than that of CW excitation. This pulse excitation results in minimum photobleaching of the fluorescent dye in the course of the measurement. This scheme also allows the possibility of timed injection of samples to be tested as well as time-synchronous observation of fluorescence after sample injection. Chemical Treatment of Fiber-Optic Surface. Near-IR dyes can be attached physically or chemically to the surface of a tapered fiber, covalently or noncovalently attached to a biomolecule that is then chemically or physically attached to a tapered fiber surface, or added in the form of a solution of the fluorophore into which a tapered fiber is immersed. We have adopted various treatments to prepare our tapered fiber surfaces to study these phenomena. Tapered fiber surfaces have been chemically prepared that are capable of physically or chemically binding a variety of dyes, other chemicals, and biological molecules. For example, as a part of a standard procedure,15 fibers were cleaned with 5% nitric acid at 90 °C for 60 min and then washed extensively with deionized water. The fibers were treated with 200 mL of 10% (v/v) (aminopropyl)triethoxysilane (APTS, Sigma Chemical Co., St. Louis, MO) in water, pH 3-4, and heated for 2 h at 75 °C. The (12) Shiver-Lake, L. C.; Golden, J. P.; Patonay, G.; Narayanan, N.; Ligler, F. Sens. Actuators B 1995, 29, 25-30. (13) Walczak, I. M.; Love, W, F.; Cook, T. A.; Slovacek, R. E. Bios. Bioelectron. 1992, 7, 39-48. (14) Anderson, G. P.; Jacoby, M. A.; Ligler, F. S.; King, K. D. Biosens. Bioelectron. 1997, 12, 329-336. (15) Graham, C. R.; Leslie, D.; Squirrel, D. J. Biosens. Bioelectron. 1992, 7, 487493.

fibers were extensively washed with deionized water and dried in a convection oven at 115 °C for a minimum of 4 h. At this point, the fibers were suitable for physical adsorption tests. For chemical binding tests, we treated the cooled fibers for 1 h with 200 mL of 1% glutaraldehyde (v/v) (Sigma Chemical Co.), and washed them extensively with deionized water. A 20-mer oligonucleotide (Gibco BRL, Grand Island, NY) containing a 5′-terminal amine group was added in phosphate-buffered saline to a concentration of 10 µg/mL. The fibers were then placed in this solution overnight at 4 °C. To compare alternative methods for chemical modification of the tapered fibers, we adopted the procedure for immobilization of proteins described by Bhatia et al.16 with modifications described below. Fibers were initially cleaned with 5% nitric acid at 90 °C for 60 min. After careful rinsing with deionized water, the fibers were allowed to air-dry. Fibers were then treated with 2% solution of 3-mercaptotrimethoxysilane (Sigma Chemical Co.) in toluene under a N2 atmosphere for 2 h. Fibers were rinsed in dry toluene and allowed to air-dry. A 1.7-mg portion of γ-maleimidobutyric acid N-hydroxysuccinimide ester (GMBS, Sigma Chemical Co.) was dissolved in 0.5 mL of N,N-dimethylformamide, which was then diluted to a 3-mL final volume with absolute ethanol. Fibers were activated in this solution by immersion for 60 min and then rinsed with phosphate-buffered saline. A 20mer oligonucleotide (Gibco BRL) containing a 5′-terminal amine group was added in phosphate-buffered saline to a concentration of 10 µg/mL. The fibers were then placed in this solution overnight at 4 °C. Hybridization Methods. Hybridization was initially carried out on hanging drop slides (Fisher Scientific, Atlanta, GA), consisting of very thick glass, with each slide containing dual concavities, each 1.5-1.95 mm deep. Oligonucleotide either with or without a 5′-terminal amine group was purified by highperformance liquid chromatography and obtained either from Genosys Biotechnologies, Inc. (The Woodlands, TX) or Gibco BRL. Hybridization to slide-attached oligonucleotide was conducted using fluorescein-conjugated oligonucleotide that was purified by high-performance liquid chromatography (Genosys Biotechnologies, Inc.). We assayed hybridization of oligonucleotide-conjugated fluorescein to treated glass slides or oligonucleotide-conjugated IRD 41 to prepared fibers, according to the method of Graham et al.15 Prehybridization solution contained 250 mL of 20× standard saline citrate, 50 mL of 100× Denhardt’s solution, 50 mL of 0.1 M sodium phosphate buffer, pH 6.8, and 1 mL of Tween 20, added to 600 mL of deionized water, final pH 6.8. Denhardt’s solution was prepared using Ficoll (molecular weight 400 0000), poly(vinylpyrollidone) (molecular weight 360 000) and bovine serum albumin (all from Sigma Chemical Co.). Hybridization was carried out either in 0.2-mL (well) or 2-mL (optical fiber) volumes. For hybridization with slides, after filling each well with prehybridization solutions containing dyeconjugate probe (10 nM), each well was sealed using glass cover slips and colorless nail polish. Slides were placed in hybridization bags containing 3-5 mL of prehybridization buffer and placed in a convection oven at 65 °C for 20 min. Bags and slides were unsealed, and each well was then washed three times with (16) Bhatia, S. K.; Shiver-Lake, L. C.; Prior, K. J.; Georger, J.; Calvert, J. M.; Bredehorst, R.; Ligler, F. S. Anal. Biochem. 1989, 178, 408-413.

Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

2033

Table 1. Oligonucleotides Used and Their Functions sequence

modification

purpose

CAC GCC CTA AAC GAT GGA TG CAC GCC CTA AAC GAT GGA TG

none 5′-amino terminal

CAT CCA TCG TTT AGG GCG TG GAG AGA CTA AGC CCT CCA AC GTA GGT AGC AAA TCC CGC AC CAT CCA TCG TTT AGG GCG TG CAC GCC CTA AAC GAT GGA TG

fluorescein 5′-amino terminal 5′-amino terminal IRD 41 IRD 41

test probe attachment without reactive group fiber-attachment probe for use with fluorescein-modified probe (below) and epifluorescence microscopy complementary detection probe for use in epifluorescence microscopy fiber-attachment probe for H. pylori sandwich assay fiber-attachment probe control for H. pylori sandwich assay detection probe for H. pylori sandwich assay detection probe control for H. pylori sandwich assay

prehybridization buffer before the wells were examined by epifluorescence microscopy. We tested hybridization to fibers with either the prehybridization buffer or one of two other solutions. IRD 41-conjugated oligonucleotide, purified by high-performance liquid chromatography, was obtained from LI-COR Inc. (Lincoln, NE) and stored in the dark at -20 °C in 0.01 M Tris buffer, pH 8.0, containing 0.001 M EDTA (TE buffer). Fibers were rinsed twice with prehybridization buffer before oligonucleotide probe was added, and the sample was heated to 65 °C in a water bath. Fibers or slides were tested once and then discarded. Controls included hybridization of IRD 41-conjugated oligonucleotide to mismatched amino-terminal oligonucleotide attached to fibers. Fiber-based hybridization was also tested using the procedure of Pontius and Berg,17 in which 0.001 M cetyltrimethylammonium bromide (CTAB), a cationic detergent, was added in 0.01 M TrisHCl, pH 7.5, 0.001 M EDTA, and 0.05 M NaCl to effect a more rapid renaturation of oligonucleotides. Hybridization on optical fibers was also tested simply using solutions that contained 0.01 M Tris-HCl, pH 7.5, and 0.05 M NaCl. Helicobacter pylori cultures were a kind gift of Dr. Rita Colwell and Dr. Manouchehr Shahamat of the University of Maryland Biotechnology Institute. Total RNA was prepared from these cultures as described previously.18 Total RNA was dissolved in TE buffer and stored at -20 °C. Sandwich assays with H. pylori RNA were conducted under conditions similar to those used for the oligonucleotide studies. H. pylori RNA (50-300 µL, nanomolar) was added to 0.01 M TrisHCl, pH 7.5, 0.001 M EDTA, and 0.05 M NaCl to make a final volume of 2 mL. The biosensor was then immersed into this solution, a detection probe conjugated to IRD 41 was added, and the mixture with the biosensor was placed in a water bath at 65 °C with gentle, intermittent agitation. RESULTS AND DISCUSSION We initially chemically treated slides and added to the wells overnight a solution of PBS containing a 20-mer oligonucleotide conjugated to fluorescein but lacking an amino-terminal group (see Table 1). Epifluorescence microscopy revealed that this oligonucleotide remained attached to the slide even after washing. We then repeated the experiment by attaching a 20-mer oligonucleotide either with or without a terminal amino group to the slide surface and hybridizing it to 20-mer oligonucleotide carrying a (17) Pontius, B. W.; Berg, P. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 82378241. (18) Boddinghaus, B.; Rogall, T.; Folhr, T.; Blocker, H.; Bottger, E. C. J. Clincal Microbiol. 1990, 28, 1751-1759.

2034 Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

Figure 2. Chemical structure of IRD 41 near-infrared dye.

5′-terminal fluorescein group. Epifluorescence microscopy indicated that hybridization occurred only when an amino-terminal group was used first to attach the oligonucleotide to the surface. IRD 41 is a near-IR fluorophore that, in methanol, shows an absorbance maximum of 787 nm, an emission maximum of 807 nm, and a quantum efficiency of 16.5%. Figure 2 shows the chemical structure of IRD 41 fluorophore. To examine the performance of our fiber-optic sensor system, we performed a preliminary experiment with just the IRD 41-conjugated oligonucleotide. A chemically treated fiber minus covalently attached oligonucleotide target was immersed in a 20 nM aqueous solution of IRD 41-conjugated oligonucleotide and immediately demonstrated a signal of 54 mV over a background signal of 5 mV. This background signal results from residual laser light that manages to pass the blocking filter. We next tested the specificity of the biosensor by conducting a series of hybridization experiments using tapered fibers. A target amino-terminal oligonucleotide was first covalently attached to the chemically modified surface of the tapered fibers as described above. We then hybridized 0.07-20 nM concentrations of the complementary 20-mer oligonucleotide conjugated to IRD 41 at the 5′ end to the surface-attached oligonucleotide using prehybridization solution at 65 °C for 15 min or longer. Figure 3 represents the change in signal with concentration for these hybridization studies. As shown in this figure, we obtained a signal twice background at a probe concentration of about 0.07 nM, indicating a detection sensitivity at the femtomolar level. Signal showed little change after fibers were washed or if the reading was taken with the fiber removed from solution. Continuous exposure of the fluorophore attached to hybridized oligonucleotide to laser light resulted in the photobleaching and diminution of signal to background in a few seconds. To confirm the specificity of binding, we repeated the hybridization tests with 20 nM IRD 41-labeled oligonucleotide but using fibers to which noncomplementary amino-terminal oligonucleotide

Figure 3. Hybridization response of chemically treated, tapered optical fiber with increasing concentration of IRD 41-conjugated oligonucleotide in methanol and aqueous solution. The legends: (b) 0.07-20 nM complementary IRD 41-labeled oligonucleotide in methanol; (9) 0.07-20 nM complementary IRD 41-labeled oligonucleotide in aqueous solution; and (2) 0.07-20 nM IRD 41-labeled oligonucleotide hybridized to control oligonucleotide in methanol.

was attached. The response of such a fiber is shown as the dotted line in Figure 3. No signal above background was obtained, which confirms the biological specificity of the hybridization reaction taking place on the assay sensor. One important characteristics of any fluorophore is its ability to emit fluorescence upon excitation at or near its absorption peak. This is defined as quantum efficiency (quantum yield, Q) of the fluorophore:

Q) number of photons emitted/number of photons absorbed (2) Near-IR dyes exhibit a higher Q when dissolved in organic solvents than in aqueous solvents. We carried out the hybridization experiment to determine whether the signal from fluorophore attached to the fiber by hybridization might be increased in the presence of an organic solvent such as methanol. The solid line curve in Figure 3 represents the response of the above tests for different oligonucleotide concentrations when fibers were immersed in methanol. As shown in this figure, we obtained an improved sensitivity of about 20% when compared to the same tests run in aqueous solution (dash-dot line curve in Figure 3). The increased signal strength in methanol could be due to the absence of quenching of the fluorophore that occurs in aqueous solution. In other experiments, we carried out real-time hybridization reactions when 0.07-20 nM concentrations of IRD 41-labeled oligonucleotide were hybridized to complementary oligonucleotide covalently attached to the sensor surface. The solid curve in Figure 4 shows the result of one such experiment, in which the time course of hybridization of 20 nM labeled probe to the surface was measured. As shown in this figure, sensor hybridization was completed in 100 s and is very rapid when compared to conventional assays taking several hours, such as filter membrane-based

Figure 4. Time course of hybridization of IRD 41-labeled oligonucleotide to a tapered optical fiber surface prepared with APTS and glutaraldehyde. The legends: (b) 20 nM IRD 41-labeled oligonucleotide; (9) 0.20 nM IRD 41-labeled oligonucleotide; and (2) 20 nM IRD 41-labeled oligonucleotide hybridized to control oligonucleotide.

assays19,20 and is also faster than the recent rapid gene probe assay method, which typically takes 30-60 min.21 At the lowest concentration of oligonucleotide (0.07 nM), the response curve represented by the dotted line in Figure 4 shows a rapid saturation of available hybridization sites on the fiber surface. Over the same time interval, IRD 41-labeled oligonucleotide added at 20 nM exhibits a continuing rise in signal output. We are uncertain why the signal from the sensor during hybridization falls off so precipitously below concentrations of 2 nM. At this concentration, there are approximately 2 × 1012 total molecules of labeled probe available for hybridization, whereas the surface of the fiber can theoretically bind only between 108 and 1010 total molecules. We consider the possibility that, at low concentrations, a significant proportion of the dye-labeled oligonucleotide in solution was binding to the very much larger surface of the plastic reaction vessel rather than to the fiber surface. To test this possibility, we carried out hybridization reactions in borosilicate glass tubes that were specially treated with silane derivative to make them nonreactive. However, signals from hybridization reactions were not significantly increased in these reaction chambers when compared to the plastic ones. Another possibility that we considered was the viscosity of the prehybridization solution that contains Ficoll, polyvinylpyrollidone, and serum albumin, together with a significant amount of standard citrate solution. The diffusivity for biological molecules with a molecular weight of above 1000 can be approximated as

DAB ) (9.40 × 10-15)Tµ-1/2MA-1/3

(3)

where DAB is the diffusivity in square meters per second of a large molecule (A) diffusing in a liquid solvent (B) of small molecules, T is temperature in kelvin, µ is the viscosity of solution in pascal, (19) Hames, B. D.; Higgins, S. J. Nucleic Acid Hybridization: A Practical Approach; IRL Press: Oxford, 1985; pp 73-111. (20) Swaminathan, B.; Prakash, G. Nucleic Acid and Monoclonal Antibody Probes: Application in Diagnostic Microbiology; Marcel Dekker: New York, 1989. (21) Engleberg, N. C. ASM News 1991, 75, 183-186.

Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

2035

Figure 5. Time course of hybridization of 20 nM IRD 41-labeled oligonucleotide to the tapered optical fiber surfaces prepared by different chemical treatments. The legends: (b) APTS and glutaraldehyde-treated fiber; (9) 3-mercaptotrimethoxysilane and GMBStreated fiber; and (2) control oligonucleotide hybridized to 3-mercaptotrimethoxysilane and GMBS-treated fiber.

and MA is the molecular weight. It should be noted, however, that two of the molecules in the solvent, Ficoll and polyvinylpyrollidone, have molecular weights close to 400 000 each and, therefore, cannot be considered small. The equation indicates that diffusivity of a large biological molecule such as IRD 41-labeled oligonucleotide (total molecular weight of 7000) is inversely related to the viscosity of the solution. To test the possibility that the viscosity of solution was limiting the diffusability of oligonucleotide to the sensor surface, we conducted hybridization reactions in solutions containing 0.01 M Tris-Cl, pH 7.5, and 0.05 M NaCl as well as buffer containing cationic detergent that has been shown to accelerate the renaturation of DNA strands. Neither solution appreciably increased the amount of signal obtained, suggesting that the viscosity of the prehybridization solution was not sufficient to retard the rate of hybridization during the measurement period. We have also tested different methods of chemically preparing the surface of our tapered fiber probes. In one set of experiments, surfaces were prepared with 3-mercaptotrimethoxysilane and GMBS (see Chemical Treatment of Fiber-Optic Surface section), a method used previously for immobilization of protein to the surface. Figure 5 shows the time course of hybridization of a 20 nM concentration of IRD 41-labeled oligonucleotide to oligonucleotide target attached to tapered fibers treated by this alternative method. This figure shows that the use of APTS and glutaraldehyde gives about a 2.5-fold increase in signal when compared to this alternative treatment. No significant differences were seen when hybridization reactions were conducted with prehybridization buffer, buffer containing CTAB, or a simple Tris-NaCl buffer (see Hybridization Methods section). Sandwich assays with H. pylori total RNA were conducted to determine whether the biosensor could detect bacterial cells using rRNA as the target. The tests consisted of hybridizing H. pylori RNA to detect probe that was chemically attached to the fiber surface while, at the same time, hybridizing IRD 41-labeled detector probe to a separate but closely related site on rRNA. Figure 6 shows the results of one such test using 25 pM H. pylori 2036 Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

Figure 6. Time course of hybridization for the sandwich assay test containing 25 pM H. pylori RNA and 2 nM IRD 41-modified detection probe.

RNA and 2 nM IRD 41-modified detection probe. The result indicates that this biosensor is capable of detecting H. pylori RNA in a sandwich assay at picomolar concentrations. Two different controls were used to determine the specificity of the sandwich assay (data not shown). One control consisted of attaching to optical fibers an oligonucleotide incapable of recognizing H. pylori rRNA sequences. When these fibers were used in hybridization reactions containing H. pylori rRNA and IRD 41-modified detection probe that could hybridize with H. pylori rRNA, no detectable signal over background was observed. Similar results were also obtained when a fiber-attachment probe specific for H. pylori rRNA and an IRD 41-modified detector probe that could not recognize H. pylori rRNA sequences were used. CONCLUSIONS In conclusion, we have developed a highly sensitive fluorosensor for hybridization assays that uses semiconductor laser excitation and near-infrared fluorophores. The principle of operation of our fiber sensor is based on the evanescent field excitation of fluorescence through the use of a long, adiabatically tapered single-mode fiber. For a single-mode tapered fiber, a large fraction of the total power exists in the evanescent field. These fibers have more than 50% of the total guided power propagating in the cladding layer. Thus, the present tapered single-mode fiber arrangement offers high sensitivity in evanescent field sensing. Tapered fibers to which amino-terminal oligonucleotide has been covalently bound before hybridization takes place can be regenerated and reused by simply heating the hybrid above its TM, usually around 80 °C (Graham et al.15), which causes the hybridized oligonucleotides to come apart. The use of singlemode fibers for this study also diminishes the necessity for regeneration of the fibers. Unlike etched multimode fibers with large-core diameters that are commonly used for biosensors but are relatively costly, the single-mode fibers used in this study are sufficiently inexpensive to make the formed biosensor element disposable. To detect Near-IR dye-labeled complementary sequences, short oligonucleotides of 20-mer were immobilized on the surface of chemically treated fibers. Since the length of a 10-base-pair-long double-helical structure based on the Watson-Crick model for

DNA is 3.4 nm (6.8 nm for the stepladder configuration), a 20mer oligonucleotide should correspond to a minimal length of 6.8 nm. Previous studies15 have shown that the evanescent field can detect surface-bound fluorescein attached to either the proximal or distal end of a 204-mer oligonucleotide. This implies that the evanescent field penetrating into the surrounding medium in our studies could efficiently excite the fluorescence from the Near-IR dye-labeled oligonucleotide, since the total penetration depth is considerably less than the wavelength of the excitation source. We have shown that this sensor is sufficiently sensitive to detect hybridization when as little as 0.07 nM IRD 41-labeled oligonucleotide is added to the hybridization medium. We have not been able to visualize the surface of chemically treated fibers both before and after hybridization and, therefore, cannot rule out the possibility that a minimal concentration of dyelabeled oligonucleotide in solution is needed to overcome some

as-yet not understood surface phenomenon. Experiments are currently being developed in which we plan to use scanning probe microscopy to visualize and monitor the chemical treatment of the fiber-optic surface and its appearance before and after hybridization. ACKNOWLEDGMENT We thank Dr. Tommy Wang and Alexander Fielding for helpful discussions. This research was funded in part by the Army Research Laboratory, under Cooperative Agreement No. DAAL0195-2-3530, and by the National Institute of General Medical Sciences, through Grant No. R43 GM51737. Received for review September 10, 1997. January 27, 1998.

Accepted

AC9709965

Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

2037