Analysis of Viral Glycoproteins by MALDI-TOF Mass Spectrometry

Membrane glycoproteins were shown to be useful biomarkers of enveloped viruses using on-target deglycosylation and matrix-assisted laser ...
0 downloads 0 Views 67KB Size
Download PDF
Anal. Chem. 2001, 73, 1544-1548

Analysis of Viral Glycoproteins by MALDI-TOF Mass Spectrometry Yeoun Jin Kim, Amy Freas, and Catherine Fenselau*

Department of Biochemistry and Chemistry, University of Maryland, College Park, Maryland 20742

Membrane glycoproteins were shown to be useful biomarkers of enveloped viruses using on-target deglycosylation and matrix-assisted laser desorption/ionizationtime-of-flight mass spectrometry (MALDI-TOF MS). Sindbis virus, the prototype r-virus, was used as a model system. The glycoproteins and the capsid protein of the Sindbis virus were successfully detected by MALDI-TOF MS using two solvent systems. One of them is 0.5% n-octyl glucoside/0.5% trifluoroacetic acid. The two components of this solvent acted synergistically on the virus to help release and solubilize the structural proteins. The other is 70% acetonitrile/30% formic acid. This solvent solubilized the integral membrane glycoproteins very effectively even after serious aggregation. On-target deglycosylation was performed to confirm the detection of the glycoprotein peak and to produce protein moieties that can be used as biomarkers. After a simple and fast incubation using peptide-N-glycosidase F on target, sequential mass shifts were observed, which proved that the proteins detected at 51 000 Da have N-linked carbohydrate moieties at two sites. Observation of this mass shift could provide confirmatory evidence for viral identification. The importance of technological developments for detecting exposed microbial agents has been magnified since 1991 when biological warfare agents (BWAs) were discovered in Iraq’s arsenal. Rapid, sensitive, and accurate methods for monitoring and identifying BWAs are needed to determine prompt and appropriate defensive actions. One of the best ways for rapid identification is to define biomarkers from a microorganism, which could be detected by field-portable mass spectrometers. This laboratory, along with others, has been developing mass spectrometric methods for rapid analysis of microorganisms such as bacteria,1,2 spores,3 and viruses.4 Since a virion is constructed by assembling proteins that commonly occur in large numbers of copies, these homogeneous structural viral proteins could be used as biomarkers of the viruses. Capsid proteins surrounding the nucleic acid genome have been shown to be useful biomarkers (1) Birmingham, J.; Demirev, P.; Ho, Y. P.; Thomas, J.; Bryden, W.; Fenselau, C. Rapid Commun. Mass Spectrom. 1999, 13, 604-606. (2) Demirev, P. A.; Ho, Y. P.; Ryzhov, V.; Fenselau, C. Anal. Chem. 1999, 71, 2732-2738. (3) Hathout, Y.; Demirev, P. A.; Ho, Y. P.; Bundy, J. L.; Ryzhov, V.; Sapp, L.; Stutler, J.; Jackman, J.; Fenselau, C. Appl. Environ. Microbiol. 1999, 65, 4313-4319. (4) Thomas, J. J.; Falk, B.; Fenselau, C.; Jackman, J.; Ezzell, J. Anal. Chem. 1998, 70, 3863-3867.

1544 Analytical Chemistry, Vol. 73, No. 7, April 1, 2001

of viruses using matrix-assisted laser desorption/ionization-timeof-flight mass spectrometry (MALDI-TOF MS) (4). Capsid proteins were dissociated by adding an organic acid and then easily desorbed by UV laser. Capsid proteins of bacteriophage MS2, tobacco mosaic strain U2, and Venezuelan equine encephalomyelitis (VEE) have been successfully detected by this method. In the present study, we have tested the hypothesis that the integral membrane glycoproteins could be used as powerful biomarkers of viruses. Some kinds of animal viruses are enveloped by a lipid bilayer containing proteins in high copy number. And in most cases, the membrane proteins are modified with carbohydrates. Identification of the virus can be improved if membrane proteins can be detected, in addition to the capsid protein. Confirmation that they carry carbohydrate moieties provides the information that the virus has an envelope. However, analyzing the integral membrane proteins is more problematic than analyzing capsid proteins. Membrane glycoproteins are notoriously difficult to handle, in part because of their amphipathic character. Integral membrane proteins aggregate readily after being released from the lipid layer, which leads to poor desorption and ionization efficiency in MALDI or electrospray ionization (ESI) processes. The masses of glycoproteins are complicated by the microheterogeneity of the carbohydrates. Furthermore, the structures of carbohydrates are also variable, depending upon the environmental conditions. Thus, until recently, mass spectrometric analysis of membrane glycoproteins has mostly been carried out with protein fragments generated by either enzymatic or chemical cleavage. We have developed a method for rapid analysis of enveloped viral proteins, using Sindbis virus as a model system. Sindbis virus is the prototype of the R-virus genus in the Togaviridae family, to which also belong the human pathogens, VEE, eastern equine encephalomyelitis (EEE), western equine encephalomyelitis (WEE), etc. Sindbis virus contains a singlestranded RNA genome and is made up of two structurally identical icosahedral protein shells (a T ) 4 lattice) between which is sandwiched a membrane bilayer.5-10 The two protein shells are connected to one another by protein-protein interactions. The inner shell is made of 240 copies of the capsid proteins and the (5) Paredes, A. M.; Simon, M. N.; Brown, D. T. Virology 1992, 187, 329-332. (6) Carleton, M.; Brown, D. T. J. Virol. 1996, 70, 5541-5547. (7) Meyer, W. J.; Gidwitz, S.; Ayers, V. K.; Schoepp, R. J.; Johnston, R. E. J. Virol. 1992, 66, 3504-3513. (8) Strauss, E. G.; Rice, C. M.; Strauss, J. H. Virology 1984, 133, 92-110. (9) Rice, C. M.; Strauss, J. H. Proc. Natl. Acad. Sci. U.S.A 1981, 78, 20622066. (10) Burke, D. J.; Keegstra, K. J. Virol. 1976, 20, 676-686. 10.1021/ac001171p CCC: $20.00

© 2001 American Chemical Society Published on Web 02/24/2001

Figure 1. (A) Central section of Sindbis virus regenerated using Adobe Illustrator from the original image created by Dr. Angel Paredes using cryoelectron microscopy and computer graphics (http://biochem.ncsu.edu:8510/faculty/brown/brown.htm). The outer shell is composed of 80 spikes, and each spike is composed of three hetero dimers. (B) Enlargement of the heterodimers embeded in the lipid bilayer.

outer shell is made of 240 copies each of two glycoproteins, E1 and E2. Figure 1 shows a model of the structure of Sindbis virus. EXPERIMENTAL SECTION Materials. Sindbis virus (strain AR339) and green monkey kidney cells were purchased from American Type Culture Collection (Manassas, VA). Poly(ethylene glycol) (PEG 8000), sucrose, phosphate-buffered saline, EDTA/trypsin, penicillin streptomycin, n-octyl glucoside, bovin serum albumin, horse heart cytochrome c, and melittin were obtained from Sigma Chemical Co. (St. Louis, MO). HPLC grade acetonitrile and 88% formic acid were from Fisher Scientific (Fair Lawn, NJ). Trifluoroacetic acid, 3,5-dimethoxy-4-hydroxycinnamic acid (sinapinic acid), and 2,5dihydroxybenzoic acid (DHBA) were from Aldrich Chemical Co. (Milwaukee, WI). Peptide-N-glycosidase F (EC 3.2.2.18, PNGase F) was from Oxford Glycosciences (Abingdon, U.K.). Propagation and Purification of Sindbis Virus. Sindbis virus was propagated using green monkey kidney cells. Flasks of infected cells were incubated in the water-jacketed incubator (VWR model 2310, Buffalo Grove, IL) at 37 °C, 5% CO2 for 2-3 days. Using an Olympus CK-40 microscope (Minneapolis, MN), flasks were examined for cytophathic effect (CPE). When CPE was seen in >75% of the cells, the medium was collected, pooled, and centrifuged at 8000g for 20 min in the Beckman Allegra 21R centrifuge (Fullerton, CA). To precipitate the viral particles, 7% PEG and 2.3% NaCl were added to the supernatant. After 16-20 h of refrigeration, the supernatant was spun at 10000g for 15 min at 4 °C. The pellet was resuspended in phosphate-buffered saline solution. For purification of the virions, a continuous 20%/60% sucrose gradient was poured and the viral suspension was carefully loaded on top. The gradients were spun at 138000g in the SW-28 rotor for 3.5 h at 4 °C in the Beckman L8-80M ultracentrifuge (Fullerton, CA). The resulting viral band was collected using a syringe and stored at - 80 °C for later use. Sucrose was removed by dialysis using a Pierce Slide-A-Lyzer cassette (MWCO 10 000) (Rockford, IL) or membrane filtration with Millipore ultrafree (MWCO 10 000) (Bedford, IL). All procedures are performed in a class II biosafety cabinet (Labconco Co., Kansas City, MO). Release and Detection of the Structural Proteins of Sindbis. A mixture of 0.5% n-octyl glucoside (OG) and 0.5%

trifluroacetic acid (TFA) was added to the Sindbis virus suspension to disrupt the lipid bilayer and capsid protein lattice and to maintain the solubility of the released membrane proteins. Sample solution (0.3 µL) was deposited on the sample slide and then 0.3 µL of 100 mM sinapinic acid in 70% ACN/0.1% TFA aqueous solution was added as a matrix. Acid-resistant Lab Label Protection Tape obtained from Bel-Art Products (Pequannock, NJ) was used as a hydrophobic surface over the stainless steel target. After crystallization, the sample/matrix spot was analyzed by MALDITOF mass spectrometry. On-Target Deglycosylation. Sindbis solution (0.3 µL) containing 0.5% OG and0.5% TFA, was deposited on the hydrophobic tape covering a sample target. After complete drying of the sample spot, 0.3 µL of 10 mM ammonium bicarbonate buffer (pH 8) containing 0.5% OG and 10% acetonitrile (ACN) was added, followed by 0.1 unit of PNGase F. This mixture was incubated in a humidifying chamber at room temperature (23 °C). This chamber was constructed by gluing a foam block (1 cm × 5 cm × 0.5 cm) on the bottom of a Petri dish. The sample target was placed on the foam block. The chamber was partially filled with distilled water, covered, and then sealed with Parafilm. Samples can be kept hydrated during the incubation with this simple method. After incubation, the target was taken out of the chamber and then dried in air. Matrix solution, 100 mM sinapinic acid in 70% ACN/0.1% TFA aqueous solution, was added to the dried sample spot. For the carbohydrate analysis, 100 mM DHBA in in 70% ACN/0.1% TFA was used as a matrix. Mass Spectrometry. Mass spectra were acquired with a Kompact IV time-of-flight instrument (Kratos Analytical Instruments, Manchester, U.K.) in the linear mode with delayed extraction. A N2 laser (337 nm) was used, and the accelerating voltage was 20 kV. Fifty scans were accumulated for each spectrum. Mass calibration for the capsid protein was performed internally, using both the MH+ ions of cytochrome c (m/z 12361) and bovine serum albumin (m/z 66431). The molecular weight of the capsid protein was used as an internal standard for glycoproteins analysis. For carbohydrate analysis, melittin (m/z 2847.5) was added to the digest as an internal standard. RESULT AND DISCUSSION Release and Solubilization of the Structural Proteins for Mass Spectrometric Detection. An R-virus is made up of two protein shells as shown in Figure 1. The core shell is constructed by assembling the capsid proteins by protein-protein and protein-RNA interactions. The outer shell is composed of glycoproteins embedded in the lipid bilayer wrapping the core shell. We tried to detect both types of structural proteins concurrently. MALDI-TOF detection of the capsid protein in enveloped viruses can be accomplished by adding organic acid.4 This method was applied to the Sindbis virus. Figure 2 presents the MALDI-TOF spectrum obtained by solubilizing the virus, with 1% TFA. One peak at m/z 29365 was detected. Based on the theoretical molecular weight calculated from the consensus sequence of the strain AR339,11-13 this peak is the capsid protein of the Sindbis (11) McKnight, K. L.; Simpson, D. A.; Lin, S. C.; Knott, T. A.; Polo, J. M.; Pence, D. F.; Johannsen, D. B.; Heidner, H. W.; Davis, N. L.; Johnston, R. E. J. Virol. 1996, 70, 1981-1989. (12) Klimstra, W. B.; Ryman, K. D.; Bernard, K. A.; Nguyen, K. B.; Biron, C. A.; Johnston, R. E. J. Virol. 1999, 73, 10387-10398.

Analytical Chemistry, Vol. 73, No. 7, April 1, 2001

1545

Figure 2. MALDI-TOF mass spectrum of Sindbis virus obtained after solubilizing with 1% TFA.

Figure 3. MALDI-TOF mass spectrum of Sindbis virus obtained after treating with 0.5% TFA and 0.5% OG.

virus. This molecular weight, then, was used as an internal standard for glycoprotein analyses. Detection of the capsid protein in Figure 2 indicated that 1% TFA dissociated not only the capsid protein assembly but also the membrane lipid bilayer, which enables the capsid proteins to be released out of the inside of the virion. However, the membrane proteins were not detected here. This result indicates that even though membrane could be destroyed by 1% TFA, integral membrane glycoproteins were not solubilized and/or ionized. Recently, many researchers are making efforts to find solvents that solubilize membrane proteins and are also compatible to mass spectrometric analysis.14 Various detergents have been examined. Although some researchers have reported that ionic detergents are compatible with MALDI,15-16 these detergents still degrade good-quality mass spectra. In this study, a nonionic detergent, n-octyl glucoside, was used, which has proven to be compatible with MALDI and ESI17,18 and is also known as good solvent for membrane proteins.19 The critical micellar concentration (cmc) of OG is 15 mM (0.44% g/v).20 In an effort to maximize the S/N ratio, we evaluated OG from the minimum concentration above the cmc, 0.5%, up to the maximum concentration below 2% which started to degrade spectra. However, no peak was detected from Sindbis virus prepared with various OG solutions. The trimers of E1-E2 heterodimers constitute the spike protein complexes which project from the surface of the virus as shown at Figure 1. The structural integrity of the envelope of Sindbis virus is maintained by the membrane-spanning regions of the E1 and E2 proteins and interactions between the capsid protein and the C-terminal tail of E2 protein. But also, in addition, lateral interactions among the spike protein compexes play an important role in the structural integrity.21-23 Apparently, a low

concentration of nonionic detergent could not penetrate this protein lattice and, thereby, could not destroy the membrane bilayer. We then tried TFA and OG together. Figure 3 shows the MALDI-TOF mass spectrum of Sindbis prepared with 0.5% OG and 0.5% TFA. In addition to the capsid protein peak, a new peak centered at 50 914 Da was detected which is assigned as a mixture of glycoproteins, E1 and E2. The theoretical masses of E1 and E2 are 47 355 and 46 836 Da. The mass difference between the calculated mass and the measured mass was presumed to come from the carbohydrate side chains. The multiple masses from the microheterogeneity of glycoforms contributed to the peak broadening (the full width at half-maximum is 850 Da). Although three proteins have the same stoichiometries in this virion, the glycoprotein peak was less intense than the capsid protein peak. The intensity ratio of two peaks, I(E1+E2)/Ic, (I(E1+E2) is the intensity of the glycoprotein peak, and Ic is that of the capsid protein peak) is only 0.24. Weaker MALDI mass spectrometric signals are generally encountered with glycoproteins24 and also with integral membrane proteins. In general, once proteins aggregate and precipitate, reconstitution is very hard. Extremely high concentrations of detergents or denaturants are needed to resolubilize them. However, those are not compatible with MS. A good solvent for solubilizing aggregated proteins of Sindbis virus was determined as 70% acetonitrile/30% formic acid (ACN/FA:7/3). When this completely volatile solvent was used, mass spectrometric sensitivity was improved and the I(E1+E2)/Ic increased. One disadvantage of using this solvent is its strong acidity. Either of the two solvent systems, 0.5%OG/0.5%TFA (mild condition) or ACN/FA:7/3 (strong condition), could be chosen depending on the purpose of the research. Advantages of Using a Hydrophobic Surface. Since detergents decrease the surface tension of aqueous solutions, the drops containing sample and OG tend to spread over the stainless steel target when they are deposited. This spreading was even more serious with the ACN/FA:7/3 solvent. We solved this problem by using an acid-resistant transparent tape (Lab Labeling Protection Tape) as a hydrophobic surface. Before depositing the samples, we covered the sample target with this tape. When the sample drops were deposited on the hydrophobic surface, contact angles became larger, and sample solutions did not spread, which

(13) Klimstra, W. B.; Ryman, K. D.; Johnston, R. E. J. Virol. 1998, 72, 73577366. (14) le Coutre, J.; Whitelegge, J. P.; Gross, A.; Turk, E.; Wright, E. M.; Kaback, H. R.; Faull, K. F. Biochemistry 2000, 39, 4237-4242. (15) Breaux, G. A.; Green-Church, K. B.; France, A.; Limbach, P. A. Anal. Chem. 2000, 72, 1169-1174. (16) Amado, F. M. L. S.-M., M. G.; Ferrer-Correia, A. J.; Tomer, K. B. Anal. Chem. 1997, 69, 1102-1106. (17) Loo, R. R.; Dales, N.; Andrews, P. C. Protein Sci. 1994, 3, 1975-1983. (18) Cohen, S. L.; Chait, B. T. Anal. Chem. 1996, 68, 31-37. (19) Baron, C.; Thompson, T. E. Biochim. Biophys. Acta 1975, 382, 276-285. (20) Von Jagow, G.; Schagger, H. A Practical Guide to Membrane Protein Purification; Academic Press: New York, 1994. (21) Paredes, A. M.; Brown, D. T.; Rothnagel, R.; Chiu, W.; Schoepp, R. J.; Johnston, R. E.; Prasad, B. V. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 90959099.

1546 Analytical Chemistry, Vol. 73, No. 7, April 1, 2001

(22) Mulvey, M.; Brown, D. T. Virology 1996, 219, 125-132. (23) Anthony, R. P.; Brown, D. T. J. Virol. 1991, 65, 1187-1194. (24) Sottani, C.; Fiorentino, M.; Minoia, C. Rapid Commun. Mass Spectrom. 1997, 11, 907-913.

Figure 5. MALDI-TOF mass spectra for the analysis of PNGase F-released oligosaccharides. (A) Mass spectrum obtained after 1-h on-target deglycosylation. Peaks are assigned as (a) Hex2GlcNAc2 + Man3GlcNAc2, (b) Hex2GlcNAc2 + Fuc1Man3GlcNAc2. (c) Hex3GlcNAc3 + Man3GlcNAc2, and (d) Hex1GlcNAc4NeuAc1 + Fuc1Man3GlcNAc2 (hexose is either Man or Gal). Figure 4. MALDI-TOF mass spectra obtained (A) before the incubation with PNGase F, (B) after 15-min incubation, and (C) after 1-h incubation.

leads to small spot size after drying. Because of the small spot size, subsequent addition was very easy. Compared to some hydrophobic tapes used previously,25 this tape is transparent and we could position the sample by viewing the metal target underneath. As an additional benefit, the tape protects the metal support from the strong acids used. On-Target Deglycosylation. Compared to conventional insolution digestion, the on-target method makes the process much faster and simpler by eliminating several transfer and purification steps. Therefore, the total amount of sample needed for analysis is tremendously reduced. So far, most of the on-target methods associated with MALDI-TOF MS have been developed for analyzing small molecules such as carbohydrates released by exoglycosidase treatment26-29 or small peptides produced from trypsin digestion.18,30 We have extended on-target deglycosylation for analyzing proteins recovered from membrane glycoproteins using endoglycosidase. A Sindbis solution (0.3 µL) prepared with 0.5% OG/0.5% TFA was deposited on the target and allowed to dry completely. Since the proteins were already denatured by this treatment, an additional denaturation step was not necessary. Ammonium bicarbonate buffer (pH 8) containing 0.5% OG and 10% ACN was added in order to solubilize the membrane glycoproteins before PNGase F was added. The enzymatic activity of PNGase F under this condition was proved in a preliminary test using a small glycoprotein, ribonuclease B. With the aid of a hydrophobic surface, sequential addition of the drops on the spot (25) Yan, Z.; Caldwell, G. W.; Jones, W. J.; Masucci, J. A. Anal. Biochem. 2000, 277, 267-270. (26) Geyer, H.; Schmitt, S.; Wuhrer, M.; Geyer, R. Anal. Chem. 1999, 71, 476482. (27) Kuster, B.; Naven, T. J.; Harvey, D. J. J. Mass Spectrom. 1996, 31, 11311140. (28) Mechref, Y.; Novotny, M. V. Anal. Chem. 1998, 70, 455-463. (29) Colangelo, J.; Orlando, R. Anal. Chem. 1999, 71, 1479-1482. (30) Vestling, M. M.; Fenselau, C. Anal. Chem. 1994, 66, 471-477.

was convenient. After incubation, 100 mM sinapinic acid was added, and the mixture was analyzed by MALDI mass spectrometry without a cleanup. PNGase F is an endoglycosidase that cleaves between an asparagine residue and the reducing terminal sugar in N-linked glycoproteins. The asparagine residue is converted to an aspartic acid after the cleavage, resulting in a 1 Da mass change. The spectra in Figure 4 summarize the course of on-target deglycosylation. Figure 4B is the spectrum obtained after a 15-min incubation. While the 51 000 Da peak has disappeared, two new peaks are detected, which are centered at m/z 47 587 and 49 229. The capsid protein peak has not changed. This spectrum confirms that the proteins detected at 51 000 Da in Figure 4A are glycoproteins and indicates that they have two glycosylated sites. Two sites on each protein have been characterized by more classical methods as well.31,32 After a 1-h digestion, we detected exclusively one peak, centered at m/z 47 542 (Figure 4C). Note that the intensity ratio I(E1+E2)/Ic increased to 0.90 after the carbohydrate moieties were released. The full width at halfmaximum of the E1+E2 peak is 663 Da, 22% narrower than the original. A large number of intramolecular disulfide bonds in the glycoproteins6,33 did not hamper the complete digestion even without reducing and alkylating steps. However, the E1 and E2 proteins were not yet resolved. It has been reported that Sindbis glycoproteins are posttranslationally modified by the covalent addition of long-chain fatty acids.34,35 Studies using site-directed mutations indicate that the E2 protein carries four palmitic acids and that one palmitioyl group is attached to the E1 protein.36,37 While the theoretical mass difference between the two proteins based on their sequences is 518.6 Da, (31) Mayne, J. T.; Bell, J. R.; Strauss, E. G.; Strauss, J. H. Virology 1985, 142, 121-133. (32) Burke, D.; Keegstra, K. J. Virol. 1979, 29, 546-554. (33) Anthony, R. P.; Paredes, A. M.; Brown, D. T. Virology 1992, 190, 330-336. (34) Bonatti, S.; Cancedda, F. D. J. Virol. 1982, 42, 64-70. (35) Schmidt, M. F.; Schlesinger, M. J. J. Biol. Chem. 1980, 255, 3334-3339. (36) Ryan, C.; Ivanova, L.; Schlesinger, M. J. Virology 1998, 249, 62-67. (37) Ivanova, L.; Schlesinger, M. J. J. Virol. 1993, 67, 2546-2551.

Analytical Chemistry, Vol. 73, No. 7, April 1, 2001

1547

the theoretical difference goes down to 168.4 Da when the fatty acid modifications are considered. We also tried to analyze the carbohydrates released by deglycosylation. 2,5-Dihydroxybenzoic acid was used as a matrix for this analysis. The spectrum obtained after a 1-h on-target incubation is shown in Figure 5A, and the control spectrum in Figure 5B was obtained without PNGase F. On the basis of the molecular weights of the new peaks detected in Figure 5A, and information about the sugar compositions reported previously,31,32,34 we searched for possible carbohydrate structures using Internetaccessible software, GlycoMod (http://expasy.cbr.nrc.ca/tools/ glycomod/). Probable glycans released from the two glycosylation sites are Hex2GlcNAc2 + Man3GlcNAc2, Hex3GlcNAc3 + Man3GlcNAc2, Hex1GlcNAc2 + Fuc1Man3GlcNAc2, and Hex1GlcNAc4 NeuAc1 + Fuc1Man3GlcNAc2. These are consistent with the results of previous biochemical investigations. In summary, while complete cleavage required 1 h, deglycosylation was detected after only a 1-min incubation; therefore, it takes less than 5 min to confirm the existence of glycoproteins with on-target deglycosylation and successive

1548

Analytical Chemistry, Vol. 73, No. 7, April 1, 2001

MALDI mass scans. Incubation time could be further reduced by increasing the amount of the enzyme used, but at twice the concentration, enzyme was also detected (at 34 700 Da) in the MALDI spectrum along with the viral proteins. ACKNOWLEDGMENT We thank Dr. Joany Jackman (Applied Physics Laboratory) for advice on the method for propagation and purification of Sindbis virus, Dr. John J. Thomas (Scripps Institute) for advice on the mass spectrometric procedures, and Dr. William Klimstra (University of North Carolina) for providing the consensus sequence of the structural proteins of the strain AR339. This work has been supported by the Applied Physics Laboratory of the Johns Hopkins University and the Defense Advanced Research Project Agency (DARPA).

Received for review October 2, 2000. Accepted January 17, 2001. AC001171P