Interactions of HIV-1 TAR RNA with Tat-Derived Peptides

Anal. Chem. 1999, 71, 1167-1175

Interactions of HIV-1 TAR RNA with Tat-Derived Peptides Discriminated by On-Line Acoustic Wave Detector L. Michelle Furtado, Hongbo Su, and Michael Thompson*

Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario, M5S 3H6, Canada David P. Mack

Parke-Davis Pharmaceutical Research Division, 2800 Plymouth Road, Ann Arbor, Michigan 48105 Gordon L. Hayward

School of Engineering, University of Guelph, Guelph, Ontario, N1G 2W1, Canada

The human immunodeficiency virus type I is strongly regulated at the transcriptional level through the interaction of an 86-amino acid protein (Tat) with a viral messenger RNA transcript. Accordingly, the binding of this protein and other cellular factors to the RNA has constituted a significant target for the development of antiHIV drugs. In the present work, we describe the detection of the binding of two Tat-derived peptides, of 12 and 40 amino acids in length, with chemically synthesized RNA by an acoustic wave sensor. Immobilization of the nucleic acid to the sensor surface, which was incorporated in an on-line system, was effected using the biotin-neutravidin interaction. As expected, the changes in series resonance frequency and motional resistance for the two peptides indicate reversible interactions in both cases that can be further characterized by the calculation of kinetic off-rates. Of particular interest is the nature of the two frequencybased signals, which are in opposite directions for the two peptides. These results together with those obtained for the surface interactions of neutravidin and biotinylated RNA confirm that the thickness shear mode sensor, massresponse model involving the well-known Sauerbrey expression is invalid when applied to operation in liquids. The human immunodeficiency virus type I (HIV-1) is strongly regulated at the transcriptional level by the interaction of the Tat protein with the trans activation-responsive element at the 5′-end of the viral messenger RNA transcript (TAR).1-3 The highly conserved TAR element is composed of a 60-nucleotide-long stem-loop RNA which also incorporates a trinucleotide bulge region.4 The Tat protein, which consists of 86 amino acids, contains both cysteine-rich and highly basic regions.5,6 The former (1) Cullen, B. R.; Greene, W. C. Cell 1989, 58, 423. (2) Cullen, B. R. FASEB J. 1991, 5, 2361. (3) Steffy, K.; Wong-Staal, F. Microbiol. Rev. 1991, 55, 193. (4) Churcher, M. J.; Lamont, C.; Hamy, F.; Dingwall, C.; Green, S. M.; Lowe, A. D.; Butler, P. J. G.; Gait, M. J.; Karn, J. J. Mol. Biol. 1993, 230, 90. 10.1021/ac980880o CCC: $18.00 Published on Web 02/17/1999

© 1999 American Chemical Society

domain contains a binding site for two metal ions,7 whereas the basic motif is important in the binding of the protein to TAR RNA.8 With respect to this specific interaction,9-11 which is crucial to the activation process, much has been learned about the structural aspects from basic biochemical studies and from NMR spectroscopy.4,8,12-18 The bulge component of TAR constitutes an important molecular recognitive site for the Tat protein. The three unpaired pyrimidines responsible for the formation of the bulge produce a widened major groove, which is apparently an important element of RNA-protein interactive chemistry on a broad level.4,15,18 The more accessible groove allows specific interactions between bases in the bulge and base pairs on both sides of the trinucleotide sequence with the arginine-rich domain of the Tat protein. Electrostatic forces derived from phosphate residues and positively charged sites on the protein further stabilize the TAR-Tat combination.4,12-15 Given the occurrence of viral resistance against drugs targeting enzymes (e.g., reverse transcriptase and protease) and slow progress in the development of a vaccine, it is not surprising that (5) Arya, S. K.; Guo, G.; Josephs, S. F.; Wong-Staal, F. Science 1985, 229, 69. (6) Sodroski, J.; Patarca, R.; Rosen, C. Science 1985, 229, 74. (7) Frankel, A. D.; Bredt, D. S.; Pabo, C. O. Science 1985, 240, 70. (8) Cordingley, M. G.; LaFemina, R. L.; Callahan, P. L.; Condra, J.: Sardana, V. V.; Graham, D. J.; Nguyen, T. M.; LeGrow, K.; Gotlib, L.; Schlabach, A. J.; Collonno, R. J. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 8985. (9) Dingwall, C.; Ernberg, I.; Gait, M. J.; Green, S. M.; Heaphy, S.; Karn, J.; Lowe, A. D.; Singh, M. S.; Skinner, M. A.; Valerio, R. Proc. Natl. Acad. Sci. 1989, 86, 6925. (10) Roy, S.; Delling, U.; Chen, C.-H.; Rosen, C. A.; Sonnenberg, N. Genes Dev. 1990, 4, 1365. (11) Sumner-Smith, M.; Roy, S.; Barnett, R.; Reid, L. S.; Kupeman, R.; Delling, U.; Sonenberg, N. Virology 1991, 65, 5196. (12) Hamy, F.; Asseline, U.; Grasby, J.; Iwai, S.; Pretchard, C.; Slim, G.; Butler, P. J. G.; Gait, M. J. J. Mol. Biol. 1993, 230, 111. (13) Aboul-ela, F.; Karn, J.; Varani, G. J. Mol. Biol. 1995, 253, 313. (14) Calnan, B. J.; Tidar, B,; Biancalana, S.; Hudson, D.; Frankel, A. D. Science 1991, 252, 1167. (15) Weeks, K. M.; Crothers, D. M. Cell 1991, 66, 577. (16) Tao, J.; Frankel, A. D. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 2723. (17) Tao, J.; Frankel, A. D. Proc. Natl. Acad. Sci. U.S.A. 1993, 990, 1571. (18) Weeks, K. M.; Ampe, C.; Schultz, S. C.; Steitz, T. A.; Crothers, D. M. Science 1990, 249, 1281.

Analytical Chemistry, Vol. 71, No. 6, March 15, 1999 1167

the TAR-Tat system, in addition to other regulatory protein interactions, has attracted much drug discovery research.19 Indeed, recent times have seen the rational design of small molecules capable of blocking the binding of Tat to the TAR RNA.20-26 A key issue with regard to the characterization of binding events between the RNA and Tat, peptides, and small molecules is the choice of a technique capable of signaling the appropriate chemistry. One such method is the gel mobility shift assay, which appears to constitute a widely used approach when it comes to the measurement of dissociation rates.27 However, this technique is relatively time-consuming, requires the use of radiochemical labeling protocols, and, on occasion, produces incorrect estimates of binding affinities. Accordingly, we are developing surface plasmon resonance, Kelvin current, and transverse acoustic wave approaches to the detection of nucleic acid-ligand interactions that avoid a number of these problems. Of particular interest is the transduction of binding events, including the measurement of on- and off-rates, by on-line thickness shear mode (TSM)28 and magnetic resonator (MARS)29 acoustic wave devices. Both these structures launch transverse acoustic waves into a substrate at ultrasonic frequencies. When exposed to liquid, coupling at the liquid-solid interface results in the propagation of acoustic energy into the external medium. A particularly interesting feature of the physics of this process is that receptors, such as oligonucleotides, are situated in precisely the location where acoustic coupling occurs. Turning to the use of acoustic wave structures for the detection of interfacial nucleic acid chemistry, the first paper in this area presented a claim that duplex formation could be detected, although no corroborating evidence for such hybridization at the sensor surface was provided.30 This was also the case for an analogous measurement with a plate mode device.31 Random primer labeling (32P) of DNA sequences was employed in the explicit correlation of changes in series resonance frequency with increase of absolute mass associated with the hybridization of bacterial DNA at the surface of a TSM device.32 This study indicated that the accepted dogma regarding the extension of wavelength model for TSM response in liquids is invalid. This (19) The Molecular Biology of HIV/AIDS; Lever, A. M. L., Ed.; John Wiley: New York, 1996. (20) Chandra, A.; Demirhan, I. Arya, S. K.; Chandre, P. FEBS Lett. 1988, 236, 282. (21) Hsu, M.-C.; Schutt, A. D.; Holly, M.; Slice, L. W.; Sherman, M. I.; richman, D. P.; Potash, J. J.; Volsky, D. J. Science 1991, 254, 1799. (22) Michne, W. F.; Schroeder, J. D.; Bailey, T. R.; Young, D. C.; Hughes, J. V.; Dutko, F. J J. Med. Chem. 1993, 36, 2701. (23) Coffen, D. L.; Huang, T.-L.; Ramer, S. E.; West, R. C.; Connel, E. V.; Schutt, A. D.; Hsu, M.-C. Antiviral Chem. Chemother. 1994, 5, 128. (24) Mei, H.-Y.; Galan, A. A.; Halim, N. S.; Mack, D. P.; Moreland, D. W.; Sanders, K. B.; Trang, H. N.; Czarnik, A. W. Bioorg. Med. Chem. Lett. 1996, 5, 2755. (25) Hamy, F.; Felder, E. R.; Heizmann, G.; Lazdins, J.; Aboul-ela, F.; Varani, G.; Karn, J.; Klimkait, T. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 3548. (26) Hamy, F.; Brondani, V.; Flo¨rsheimer, A.; Stark, W.; Blommers, M. J. J.; Klimkait, J. Biochemistry 1998, 37, 5076. (27) Long, K. S.; Crothers, D. M. Biochemistry 1995, 34, 8885. (28) C ˇ avicˇ, B. A.; Chu, F. L.; Furtado, L. M.; Hayward, G. L.; Mack, D. P.; McGovern, M. E.; Su, H.; Thompson, M. Faraday Discuss. 1997, 107, 159. (29) Stevenson, A. C.; Lowe, C. R. Appl. Phys. Lett. 1998, 73, 1. (30) Okahata, Y.; Matsunobu, Y.; Ijuro, K.; Makae, M.; Murakami, M.; Makino, K. J. Am. Chem. Soc. 1992, 114, 8299. (31) Andle, J. C.; Vetelino, J. F.; Lade, M. W.; McAllister, D. J. Sens. Actuators B 1992, 8, 191. (32) Su, H.; Kallury, K. M. R.; Thompson, M.; Roach, A. Anal. Chem. 1994, 66, 769.

1168 Analytical Chemistry, Vol. 71, No. 6, March 15, 1999

model is based on the notion that any material added (or lost) to the sensor surface can be treated as an equivalent change in mass of the quartz component of the device itself. In addition to the overall signaling of duplex formation, the TSM has also been employed in the successful measurement of the kinetics of nucleic acid chemistry at an interface.33,34 Similarly, the enzymatic hydrolysis of surface-bound DNA has been monitored, and in this case, it was shown that DNA-lipoglutamate complexes exhibit a higher stability than native DNA.35 Recently, an attempt was made to increase the sensitivity of the TSM to nucleic acid chemistry, based on the conventional mass response concept, by generating a multilayered structure on the device surface.36 Not surprisingly, this was unsuccessful because of the lack of access to reactive binding sites in tandem with a questionable acceptance of the mechanism of sensor response. Finally, we introduce the flowthrough, on-line detection of interactions between device-bound DNA/oligonucleotides. The responses of a TSM sensor in terms of both resonance frequency and motional resistance were obtained for the binding of the anticancer drug, cisplatin, and its chemically ineffective isomer, transplatin to dsDNA.37 A kinetic analysis of the data confirmed that the sensor was actually responding to nucleic acid attachment to the hydrolysis products of the two platinum complexes. With regard to the binding of proteinaceous species to DNA, the selective attachment to particular sequences has been claimed on the basis of the difference in response found for specific and unselective binding events.38 In the present paper, we describe the direct, real-time detection of the binding of two Tat-derived peptides to TAR RNA by a TSM incorporated in an on-line configuration. Immobilization of the nucleic acid to the sensor surface was effected using the neutravidin-biotin system. Also included are studies of the effect of peptide concentration and the behavior of a mutated TAR sequence and an evaluation of amounts of nucleic acid and peptide attached to the TSM device through radiochemical labeling protocols. Finally, sensor responses are analyzed to yield dissociation rates for the various RNA-peptide complexes.

EXPERIMENTAL SECTION Reagents and Materials. Neutravidin was obtained from Pierce Chemical Co. (Rockford, IL) and used without further purification. Tris buffer (10 mM Tris, 70 mM NaCl, 0.2 mM EDTA) was used to prepare solutions and for on-line flow experiments. The A, G, C, and U controlled-pore glass columns, phosphoramidites and biotin phosphoramidite, tetrazole/acetonitride, 1methylimidazole/THF, acetic anhydride/pyridine/THF, iodine/ H2O/pyridine THF, and anhydrous acetonitrile were obtained from Applied Biosystems (Mississauga, ON, Canada) and used as received. Cartridges (Poly-Pak) used for purification were purchased from Bio/Can Scientific (Mississauga, ON, Canada). Ammonium hydroxide, trifluoroacetic acid, acetonitrile, glacial (33) Su, H.; Thompson, M. Biosens. Bioelectron. 1995, 10, 329. (34) Su, H.; Chong, S.; Thompson, M. Biosens. Bioelectron. 1997, 12, 161. (35) Sato, T.; Kawakami, T.; Shirakawa, N.; Ikahata, Y. Bull. Chem. Soc. Jpn. 1995, 68, 2709. (36) Caruso, F.; Rodda, E.; Furlong, D. N.; Niikura, K.; Okahata, Y. Anal. Chem. 1997, 67, 22043. (37) Su, H.; Williams, P.; Thompson, M. Anal. Chem. 1995, 67, 1010. (38) Niikura, K.; Nagata, K.; Okahata, Y. Chem. Lett. 1996, 863.

Figure 1. Secondary structure of HIV-1 TAR RNA for positions +1 through 57 (A). The RNA moiety is divided into sections as follows: four stem regions, I-bases 5-9 and 50-54; II-bases 10-15 and 44-49; III-bases 17-21 and 29-43; IV-bases 25-28 and 35-38; a three-base pyrimidine bulge, a six-member loop, and unpaired nucleotides. Chemically synthesized biotinylated RNA with (B) and without (C) the threebase pyrimidine bulge. The former molecule has 31 nucleotides and the latter 28.

acetic acid, and triethylamine were obtained from Aldrich. The latter two reagents were used to prepare triethylammonium acetate. The TAR RNA (TAR) 5′-GGCCAGAUCUGAGCCUGGGAGCUCUCUGGCC-3′ was synthesized on an Applied Biosystems 394 synthesizer using 2′-tert-butyldimethylsilyl- and 5′-dimethoxytritylprotected monomers Millipore, Corp. (Bedford, MA) and standard phosphoramidite chemistry. The RNA was purified by gel electrophoresis. When required, biotin was incorporated at the 5′end of the RNA (B-TAR) during synthesis utilizing the BioTEG phosphoramidite (Glen Research; same as Bio/Can Sci.). For radiochemical experiments, B-TAR was labeled with T4 RNA ligase (Pharmacia Biotech Inc., Baie D′Urfe, Quebec, Canada) using cytidine 3′,5′[5′- 32P]bisphosphate (Amersham, Little Chalfont, Buckhamshire, UK). A mutated version of B-TAR that did not contain the U-rich bulge, Biot-5′-GGCCAGAGAGCCUGGGAGCUCUCUGGCC-3′, was synthesized in biotinlylated form (BSTAR) by a procedure similar to that described above. A comparison of the secondary structure of HIV-1 TAR-RNA with chemically synthesized B-TAR and B-STAR is given in Figure 1. The peptides, H2N-Tyr-Gly-Arg-Lys-Lys-Arg-Arg-Gln-Arg-ArgArg-Gly-CO2H (Tat12) and H2N-Tyr-Gly-Arg-Lys-Lys-Arg-Arg-GlnArg-Arg-Arg-Pro-Pro-Gln-Gly-Ser-Gln-Thr-His-Gln-Val-Ser-Leu-SerLys-Gln-Pro-Thr-Ser-Gln-Ser-Arg-Gly-Asp-Pro-Thr-Gly-Pro-Lys-GluCO2H (Tat40) (Figure 2), were prepared by solid-phase synthesis using standard BOC chemistry protocols, purified by reversedphase HPLC, and characterized by electrospray mass spectrometry (Tat12 MW: calcd, 1619.9; found 1617.6. Tat40 MW: calcd, 4644.2; found 4644.4). In a number of experiments, Tat12 was labeled with 125I.

The 9-MHz AT-cut piezoelectric quartz devices, coated with polished gold electrodes, were obtained from International Crystal Manufacturing, Inc. (Oklahoma City, OK). In certain cases, the electrodes were simply rinsed in solvent for cleaning or subjected to plasma etching under N2. Instrumentation. The flow-through configuration for introducing various reagents to the detector incorporated a plexiglass cell which consisted of two halves separated by O-rings. Assembly of a TSM sensor into the cell was effected such that one face could be exposed to flowing buffer, while the other side was kept under N2. The exposed area of the device to liquids was 1.0 cm2, whereas the corresponding electrode area was 0. 30 cm2. On-line experiments were performed using a four-channel EVA pump model 1000 peristaltic system (Eppendorf, Hamburg, Germany), which was adapted for combination with an EVA Valve model 2000 injector part (Eppendorf). The PTFE tubing for the sample loop had an inner diameter of 0.5 mm, while all other tubing had an inner diameter of 0.8 mm. Finally, the TSM responses in liquid held at ambient temperature were measured with an HP 4195 network/spectrum analyzer (Hewlett-Packard, Palo Alto, CA). The values of the equivalent circuit were calculated internally by the analyzer from measured data. Calibration of the network analyzer at a center frequency of 9 MHz was accomplished using a preset calibration program. The configuration was set, by user input, to record the acquired data every 30 s. Radiochemical counting of solutions containing 125I-labeled Tat12 and 32P-labeled B-TAR and devices coated with these species was accomplished using a Riastar 103271 γ-ray detector (HewlettPackard) and a 1219 Rackbeta scintillation counter (Fisher Scientific, Raleigh, NC), respectively. Analytical Chemistry, Vol. 71, No. 6, March 15, 1999

1169

Figure 2. Schematic of HIV-1 Tat protein (A) and primary sequence of Tat (B). Arrows indicate sequences of chemically synthesized Tat12 and Tat40. In the former, position 58 (proline) is replaced by glycine.

Contact angle measurements regarding the TSM gold electrode surface were performed with a Remy-Hart goniometer. Procedures. Before introduction of the various dispersions of reagents to the TSM sensor (solvent cleaned) incorporated in the on-line equipment, buffer was passed through the system at a flow rate of 0.1 mL min-1 until a stable series resonance frequency was obtained. A dispersion of neutravidin (500 µL at a concentration of 1.0 mg mL-1 in buffer) was injected. Following frequency stabilization TAR, B-TAR, or B-STAR was introduced, again in dispersion form (500 µL at a concentration of 10-6 in buffer). After further frequency stabilization, 100 µL of Tat12 solution (1.0 × 10-6 M) was sequentially dispersed and allowed to interact with the neutravidin-RNA surface moiety. With respect to this peptide, single injections were also performed to achieve a concentration study ((0.1-2.5) × 10-6 M) for both B-TAR- and B-STAR-modified surfaces. Experiments with Tat40 involved introduction of a 100-µL dispersion of the peptide at a concentration of 0.8 × 10-6 M in flowing buffer. Finally, included in this study was a set of control experiments, for example, exposure of neutravidin to Tat12 and to TAR separately. To ascertain the actual mass of B-TAR and Tat12 bound on the protein-coated device, radiochemical experiments were performed using the same basic procedure as that described above with the exception that the experimental configuration did not involve any acoustic wave measurements. In a first set of experiments, a solvent- or plasma-cleaned TSM device incorporated in a sacrificeable cell in an on-line system was exposed to a 500-µL volume of neutravidin solution (1 mg mL-1). The device was subsequently washed with buffer for ∼30 min prior to injection of 500 µL of 32P-labeled B-TAR (containing 100 pmol of the RNA). To prepare the latter solution, 100 µL of B-TAR stock solution (1 nmol mL-1) and 25 µL of labeled solution (45 pmol mL-1) were added to 375 µL of buffer. This solution was found to have a count rate of 4811 cpm. In other experiments, radiolabeled Tat12 was introduced to devices with neutravidin only and neutravidin bound to B-TAR 1170 Analytical Chemistry, Vol. 71, No. 6, March 15, 1999

(unlabeled) films in place. This involved 500 µL of radiolabeled Tat12 (1 nmol) solution which was prepared from both stock solution and buffer. The injected solution possessed a count rate of 904 494 cpm. After the various injections of radiolabeled B-TAR and Tat12, the coated device was washed copiously with buffer before the sensor was removed from the cell ready for counting. RESULTS AND DISCUSSION Interaction of B-TAR and B-STAR with Tat12. We begin with a brief explanation of the reasons for our use of the neutravidinbiotin combination rather than the equivalent avidin system in the immobilization of the RNA moiety. The parent molecule is a homotetrameric protein with a molecular weight in the range 67 000-68 000. The variation in weight is related to heterogeneity in the glycosylation of subunits at the Asn-17 residue.39 In a large number of preliminary experiments involving this molecule imposed on the surface of the gold electrode component of TSM sensors, we observed significant differences with regard to the effect of gold surface free energy on the integrity of the avidinbiotin interaction.40 Apparently, in situations where the Au surface is of high energy, the metal-protein interaction compromises the tertiary structure of the latter which, in turn, affects subsequent binding of any biotinylated species. This observation led to further studies which included atomic force microscopy and X-ray photoelectron spectroscopic analyses in which very significant differences in behavior with respect to surface free energy were found.40 Accordingly, we chose to employ the deglycosylated derivative, neutravidin, which does not display the same sensitivity to surface properties, at least in terms of its tertiary structure, but does possess approximately the same affinity for biotin as the parent molecule (Kd ) 10-15 M). (39) Hiller, Y.; Gershoni, J. M.; Bayer, E. A.; Wilchek, M. Biochem. J. 1987, 248, 167. (40) Furtado, L. M.; Nisman, R.; Thompson, M., to be published.

Table 1. Exposure of Surface-Treated Devices to 32P-Labeled Biotin-TAR RNA (B-TAR) and 125I-Labeled Tat12 (Tat12) surface cleaninga

reagent (neutravidin +)

count (cpm)

solvent solvent plasma plasma plasma solvent solvent plasma solvent plasma

B-TAR B-TAR B-TAR B-TAR B-TAR Tat12 Tat12 Tat12 B-TARc + Tat12 B-TARc + Tat12

4484 3979 3347 5508 4891 14670 11158 22210 10834 21295

Tat12b (pmol)

immobilized B-STARb (pmol) 0.93 0.83 0.70 1.14 1.02

16 12 25 12 24

a Contact angles for cleaned surfaces are solvent (80 ( 5°, 14 tests) and plasma (50 ( 6°, 10 tests) treated. b Calculated in mass terms from specific activity data. c Unlabeled RNA.

Figure 3. Changes in series resonance frequency for sequential introduction of neutravidin, B-TAR, TAR, and Tat12 peptide to on-line acoustic wave detector. Flow rate is 0.1 mL min-1 and dispersions are neutravidin, 500 µL of 1.0 mg mL-1; B-TAR and TAR, 500 µL of 10-6M; and Tat12, 100 µL at 1.0 × 10-6 M.

Turning to the neutravidin-RNA-Tat12 system, typical sensor responses of series resonance frequency (fs) versus time for sequential introduction of the various reagents to the on-line instrument are shown in Figure 3. The adsorption of the protein on the electrode surface of the TSM device is confirmed by the reduction of fs by a value of, typically, 300 Hz (over 200 experiments). Further passage of flowing buffer solution results in a stabilization of fs at the decreased frequency, confirming that the immobilized neutravidin is irreversibly bound under the conditions employed. It should be emphasized that the protein undoubtedly adsorbs to both acoustically inactive quartz and the electrode surface. Given that it has been confirmed that this protein forms monolayers on these types of surfaces,41 a rough comparison between an expected surface population and mass predicted from the Sauerbrey mass response model can be effected. Neutravidin will occupy a surface area of ∼100 nm2. The immobilization area for our devices is 1.0 cm2; therefore, the number of protein molecules adsorbed to the overall surface of the TSM sensor (one side) is ∼1.0 × 1012 molecules (1.6 pmol/ 96 ng) assuming, of course, the unlikely occurrence of close packing. Use of the Sauerbrey equation,42 ∆f ) 1.83 × 108 × ∆m/A Hz (∆m is mass added to one sensor face in g; A is the area in cm), predicts a frequency shift of 17 Hz for the deposited layer of protein. This figure will be significantly reduced if the monolayer of adhered protein does not exhibit a close-packed structure. Since acoustic propagation occurs into an aqueous medium, bound water on the protein molecules is irrelevant in terms of mass; therefore, the mass response model is inadequate, at best, in describing protein adsorption phenomena. However, on a positive note, and as indicated by ourselves in previous work,28 the acoustic wave technique appears to offer an unusually sensitive measure of the level of attachment of biochemical macromolecules at interfaces, a point that will be further exemplified in the subsequent discussion. (41) Ebersole, R. C.; Miller, J. A.; Moran, J. R.; Ward, M. D. J. Am. Chem. Soc. 1990, 112, 3239. (42) Sauerbrey, G. Z. Phys. 1959, 155, 206.

Returning to Figure 3, typically fs decreases a further 50 Hz on exposure of the immobilized neutravidin to B-TAR. Again, the signal stabilizes at this further decreased frequency, confirming the expected strong binding of the biotin moiety to the protein layer. The introduction of a control sample of TAR RNA does not generate an alteration in fs. The far right column of Table 1 shows that, under equivalent experimental conditions, binding of the biotinylated RNA species occurs to a level of ∼1 pmol for the device area of 1.0 cm2. The Sauerbrey equation predicts a frequency change of 2 Hz for this amount of bound RNA. Again, the extension of wavelength concept fails to account for the true sensitivity of the transverse acoustic wave device to recognitive processes at the device surface. Finally, with respect to the binding of the RNA to the adsorbed protein, we note that the radiochemical experiments confirm that the binding event occurs at a level of ∼1 biotin per neutravidin molecule. Although there are four binding sites per protein molecule, this result is not surprising in view of the fact that some sites are likely unavailable because of the surface orientation of the protein molecule and the occurrence of stearic proximity effects. The TSM device does not respond to injected dispersions of Tat12 when only neutravidin is present on the sensor surface (Figure 4). In contrast, sequential dispersions of Tat12 introduced to the B-TAR-modified interface result in transient increases in fs to a level of ∼30 Hz for a Tat12 concentration of 1 µM. This reversible response is entirely expected in view of the transient exposure of the RNA to the peptide in a FIA-like environment. The magnitude and direction of the changes in resonance frequency represent intriguing results in the light of the usual dogma associated with the response of TSM devices to added material. If the reasonable assumption can be made that B-TARTat12 constitutes a 1:1 combination, then the 1 pmol of RNA attached to the sensor surface predicts a maximum bound mass of peptide of ∼1.6 ng (all RNA moieties binding peptide) in the acoustic wave measurement. The change in frequency for the amount of material calculated using the Sauerbrey model is ∼0.3 Hz compared with the observed value of 30 Hz. Accordingly, both the direction and magnitude of the response are at odds with the mass response model. This result can only be explained using the conventional calculation by first a loss of mass from the device Analytical Chemistry, Vol. 71, No. 6, March 15, 1999

1171

Figure 4. Series resonance frequency (top) and motional resistance (bottom) changes for introduction of dispersions of Tat12 to neutravidin-B-TAR (A) and neutravidin-modified (B) TSM sensors.

surface, followed by a redeposition instigated by the Tat12-RNA interaction. Clearly, this is an unlikely process. Moreover, addition of dispersions of Tat12 to radiolabeled surface-bound B-TAR resulted in no evidence of RNA removal (breakage of the biotinneutravidin bond). The radiochemical data of Table 1 describe experiments involving the interactions of dispersions of 125I-labeled Tat12 with neutravidin and B-TAR-modified surfaces. Clearly, the results in no way indicate the amount of peptide specifically bound to the RNA, since flowing buffer will remove such before counting. However, the data strongly indicate that high levels of irreversible nonspecific adsorption to the underlying device/protein surface occurs up to ∼20 pmol for 1.0 cm2. Moreover, the level of adsorption appears to depend on the surface free energy of the device interface. This is not related to the amount of protein on the surface since the radiolabeled B-TAR experiments are consistent with respect to the level of attachment of RNA. The amount of peptide adsorbed represents a mass of 32 ng (supposed mass response frequency shift of ∼5 Hz) and remains undetected by the TSM device in the acoustic wave on-line experiments, likely because of the presence of the existing relatively large proteinbased molecular film at the interface. Accordingly, it is apparent that the signal related to the RNA-peptide interaction is being observed in the face of a high level of nonspecific adsorption. In other words, the TSM sensor is uniquely sensitive to the properties of the nucleic acid at the solid-liquid interface, an important result we now explore in further detail. 1172 Analytical Chemistry, Vol. 71, No. 6, March 15, 1999

Expanded scale plots for Tat12 response on binding to B-TAR in terms of both series resonance frequency and motional resistance (Rm) are given in Figure 4. Concomitant with the frequency increases described above, transient reductions in the value of Rm are observed. Since mass deposition does not affect the motional resistance,43 the additional response may be attributed to changes in the viscoelastic properties of the B-TAR or to changes in the coupling between the surface and the liquid. There is evidence that protein-RNA binding stiffens the RNA molecule.13,44,45 In acoustic terms, this would store energy in the film, returning it out of phase with the motion, giving a frequency increase. A decrease in the film viscosity, which could accompany the stiffening, would reduce the energy dissipation by increasing the resonant frequency and decreasing the motional resistance. To explore this possibility, several simulations were carried out using a transverse shear model of the sensor.46 The results, based on a simulated film thickness of 10 nm, show negligible frequency and motional resistance shifts when the stiffness is low or high. Between these ranges is a transition at a stiffness that gives a shear wave velocity corresponding to a quarter of the acoustic wavelength matching the film thickness. At this point, the frequency drops by a few hundred hertz while the motional resistance passes through a peak rising ∼5 Ω above the baseline. It is unlikely that this effect gives the signal observed since the frequency and resistance shifts are in a direction opposite to those observed. The response to changes in the coupling between the B-TARTat surface and the surrounding buffer is a more plausible mechanism. It has been established that changes in the hydrophobic/hydrophilic nature of a surface give rise to shifts in both the motional resistance and series resonant frequency.47 A hydrophilic surface that couples well with an aqueous medium gave much higher frequency shifts and motional resistance values than those obtained with a hydrophobic surface with poorer coupling. The transverse shear model46 represents this coupling as a slip parameter. As the coupling becomes less, the slip increases. This decreases the motional resistance and increases the series resonant frequency.28 This type of behavior agrees with the results shown in Figure 4. There is also evidence that the binding of peptides to TARRNA changes its conformation45 and causes a local binding of the helix axis.48 Either of these events could produce the type and magnitude of the response observed. Finally, in this respect it is interesting to note that when Tat12 binds to B-TAR, the motional resistance decreases to the level when the surface is coated with neturavidin alone. Finally, with respect to the nucleic acid interactions of Tat12, plots of fs versus concentration of peptide for both surfaceimmobilized (B-TAR) and the bulgeless RNA (B-STAR) are shown in Figure 5. The frequency values were obtained from the maximums in signals associated with the dispersions depicted in Figures 3 and 4. Using a standard deviation of (2 Hz for the (43) Hayward, G. L.; Chu, G. Z. Anal. Chim. Acta 1994, 288, 179. (44) Zacharias, M.; Hagerman, P. J. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 6052. (45) Puglisi, J. D.; Tan, R.; Calnan, B. J.; Frankel, A. D.; Williamson, J. P. Science 1992, 76, 257. (46) Hayward, G. L.; Thompson, M. J. Appl. Phys. 1998, 83, 2194. (47) Yang, M.; Thompson, M.; Duncan-Hewitt, W. Langmuir 1993, 9, 802. (48) Bhattacharyya, A.; Murchie, A. J. H., Lilley, D. M. J. Nature 1990, 343, 484.

Figure 5. Response-concentration plots for interaction of Tat12 dispersions with B-TAR and bulgeless RNA(B-STAR). Bottom curve represents control experiments where no RNA is immobilized on the neutravidin-modified sensor surface.

background level in the measurement system, the concentration limit of detection for Tat12 (B-TAR) is in the vicinity of 1.0 × 10-7 M peptide. With regard to RNA type, the data in Figure 5 clearly show that the binding of the peptide to the RNA with the trinucleotide bulge in place can be easily distinguished from that for attachment to the bulgeless species. At a peptide concentration of 1.0 × 10-6 M, the difference between the two responses, the lower being associated with B-STAR, is ∼30%. Whether this difference can be ascribed to a simple difference in Kd value or to factors related to changes in RNA structure on peptide binding remains to be seen. However, gel mobility shift assay work indicates clearly that changes in kinetic stability for peptide interactions with TAR RNA and various mutants are evident.27 It is possible that the observations of the present work are connected to both alterations in kinetic stability and perturbation of RNA structure. These factors are the subject of further research in our laboratory. Interactions of B-TAR with Tat40. Initial injections of neutravidin and B-TAR produce responses that are idential to those discussed above, but the introduction of a dispersion of Tat40 to the on-line system generates radically different behavior from that exhibited by the smaller peptide. Figure 6 shows the frequency response curve for a typical experiment involving Tat40 (neutravidin and B-TAR responses are omitted). The signal is reversible, but with an initial decrease in fs as compared to the increase found for Tat12. Furthermore, the recovery of the signal to the baseline value observed prior to injection of peptide is significantly delayed as compared with that found for the smaller peptide. This result clearly suggests that the overall dissociation rate for the Tat40-TAR RNA complex is much slower than that for the analogous Tat12 complex, not surprising since a number of new base-amino acid interactions are introduced. As indicated above for Tat12, a number of competing factors will govern the nature of the fs and Rm signals obtained for peptide-RNA interaction. An interesting reflection of the various possibilities is the reproducible fine detail exhibited by the fs result

Figure 6. On-line response profile of series resonance frequency for interaction of Tat40 with surface-bound B-TAR. Signals obtained for introduction of neutravidin and B-TAR are omitted for simplicity. The profiles are for 100 µL of Tat40 at a concentration of 0.8 µM.

Figure 7. On-line response profiles of fs (open squares) and motional resistance (right axis, Ω, closed points) for interaction of Tat40 with surface-bound B-TAR. Details as for Figure 6.

and analogous plot for Rm (Figure 7). Note that concomitant with a shoulder in the association part of the fs response for Tat40, an initial increase followed by reduction occurs in the Rm response. Following the latter, there is a gradual moderate increase in motional resistance corresponding to the increase in fs as Tat40 is slowly removed from the surface of the device. In summary, Tat12 yields an increase of fs but a decrease in Rm on association with B-TAR, whereas the changes for Tat40 are overall reductions of both fs and Rm. The binding of TAT40 to B-TAR appears to proceed in two steps. The first is attachment where the large TAT40 binds to the B-TAR projecting into the medium. This increases the coupling to the liquid phase, increasing the motional resistance and decreasing the resonant frequency. Binding is followed by a change in RNA structure where the liquid coupling is reduced, probably in a manner similar to that occurring with TAT12, resulting in a decrease in the motional resistance. The resonant frequency Analytical Chemistry, Vol. 71, No. 6, March 15, 1999

1173

Table 2. Off-Rate Constants for Tat12-Peptide Complexes of B-TAR and B-STAR run

peptide conc (µM)

1

2

[complex]/[complex]t)0 ) Re-k1t + (1 - R-k2t)

For this purpose, we assume that the concentration of the peptide-RNA complex over time at the device surface is linearly proportional to the value of fs, with the t ) 0 value being normalized to the maximum change occurring in this parameter. (It should be noted that it is not feasible to compute explicit onrate constants at this time because this calculation requires the precise surface concentration of RNA for each experiment). A typical plot of the dissociation kinetics for a Tat40 experiment is shown in Figure 8. The best fit to the experimental points in this case indicates a biphasic dissociation profile exactly as found for previous work on 24-amino acid peptides.27 In contrast, Tat12 yields monophasic behavior. Collections of off-rate constants are given in Tables 2 and 3. The data of Table 2 include values computed for exposure of a specific RNA-modified surface to dispersions of varying Tat12 concentrations within an experimental run and for different tests. The results of Tables 2 and 3 suggest that, despite the nonideal FIA experimental setup and the approximate nature of the calculations, reasonable values for off-rate kinetics can be made using the acoustic wave detector. For both Tat12 and Tat40, the removal of the 3-pyrimidine bulge results in increased off-rates and, presumably, higher Kd values for the RNA mutant than for the nucleic acid containing the bulge. As pointed out above, this 1174 Analytical Chemistry, Vol. 71, No. 6, March 15, 1999

2.5 1.2 0.8 0.7 0.6 0.4 0.2 2.5 1.2 0.8 0.7 0.6 0.4 0.2

Tat12-B-TAR 49 44 36 33 30 21 12 55 47 44 40 33 22 13

10.0 5.0 1.25 0.5 2.5 1.4 1.2 0.8 0.7 0.5 1.2 0.7 0.6 0.4

Tat12-B-STAR 58 54 41 25 39 32 31 30 18 13 32 21 19 11

6.2 5.1 4.3 3.5 3.0 5.3 2.5 3.6 3.4 3.9 3.1 3.3 3.6 3.2 mean 3.8 ( 1.0 (sd)

Figure 8. Dissociation kinetics for Tat40-B-TAR complex obtained from fs versus time plots. Peptide concentrations is 0.8 µm. The parameter N is [complex]/[complex]t)o.

should increase with this change, and a shoulder in the frequency curve indicates that this happens, but the increase is swamped by the amount of protein on the surface. Off-Rate Constants. The experiments described in the present work incorporate rinsing with flowing buffer after the exposure of RNA to dispersions of peptide solution. Accordingly, temporarily bound Tat12 and Tat40 will be removed from the sensor surface, allowing the possibility to calculate pseudounimolecular rate constants directly from series resonance frequency-time plots. Using the method of Long and Crothers,27 such rate constants can be calculated from the best fit to the expression

k1a (×10-3 s-1)

∆fs (Hz)

1

2

3

6.2 6.6 3.9 3.8 5.8 8.3 12.8 8.5 8.2 8.9 6.8 5.2 7.1 10.0 mean 7.2 ( 2.4 (sd)

a

Assumes R ) 1.0 for monophasic behavior.

Table 3. Off-Rate Constants for Tat40-Peptide Complexes of B-TAR and B-STAR run

peptide conc (µM)

∆fs (Hz)

1 2 3

1.2 0.8 0.8

56 44 49

1 2 3 4

1.2 0.4 0.8 0.8

46 47 40 42

k1 (×10-3 s-1)

k2 (×10-4 s-1)

Tat40-B-TAR 0.39 0.61 0.35 0.65 0.32 0.68

2.1 2.5 2.9

0.3 0.2 0.7

Tat40-B-STAR 0.17 0.83 0.14 0.86 0.21 0.79 0.21 0.79

3.2 3.1 4.3 3.7

1.6 1.0 1.2 1.5

R

1-R

observation is likely related to a reduction in the number of RNA base-amino acid contact points, whether of the electrostatic or hydrogen bond type. The kinetic behavior of Tat12 compared to Tat40 is particularly interesting in terms of the clear doubleexponential fit for data obtained from the latter. The larger peptide will obviously be capable of increased molecular contact with the RNA compared to the 12-amino acid version. This is reflected in the results given in Tables 2 and 3 and in the t1/2 lifetimes of the peptide-B-TAR complexes which are 1.9 × 102 and 3.7 × 103 s for Tat12 and Tat40, respectively (obtained from the kinetic data). The biphasic off-rate kinetics for Tat40 may be associated with different conformations of the RNA structure as suggested by

previous authors.27 In light of this possibility, the nature of the fs and Rm signals obtained for interaction of Tat40 with the RNA represents an intriguing result with respect to potential links between perturbed RNA structure and its acoustic properties at the solid-water interface. CONCLUSIONS The results of the work described in the present paper confirm that the mass response model of the behavior of TSM devices in liquids is invalid in terms of application to biomolecular chemistry at the solid-liquid interface. More importantly, transverse acoustic wave devices offer considerable potential for the characterization of such species at interfaces because of their unique acoustic properties in this location. Indeed, recent work in our laboratory indicates that this type of sensor rivals surface plasmon resonance in terms of sensitivity to interfacial biochemical processes. Accordingly, the observations made here indicate that a reevaluation of the restrictive mass response concept for TSM structures immersed in liquids is certainly in order. With respect to the signaling of the chemistry involving TAR RNA-peptide complexes, there is a strong indication that the acoustic wave technique is capable of the detection of selective binding events even when significant nonspecific adsorption occurs in a concurrent fashion. This observation together with the possibilities for relatively high throughput measurement of

biochemical off-rates in a flow-through configuration presents significant potential for the future assessment of nucleic acidsmall molecule interactions in the context of drug discovery research. The technique is eminently compatible with FIA technology, monolayer attachment of biomolecules, attachment chemistry, and series or parallel multisensor detection. Finally, the differences in acoustic properties exhibited by the Tat12- and Tat40-RNA complexes appears to offer potential for the future with regard to the use of transverse acoustic waves in the study of interfacial biochemical structure. ACKNOWLEDGMENT Support for this work from the Natural Sciences and Engineering Research Council of Canada and Parke-Davis Pharmaceutical Research Division of Warner-Lambert Co. is gratefully acknowledged. We thank W. Cody and J. He for their assistance in preparing the Tat peptides, J. Loo for ES-MS analysis of peptides and RNA, and A. Czarnik, S. Dewitt, A. Galan, H.-Y. Mei, D. Moreland, and D.C. Stone for much helpful discussion.

Received for review August 6, 1998. Accepted January 7, 1999. AC980880O

Analytical Chemistry, Vol. 71, No. 6, March 15, 1999

1175