Expression of the F glycoprotein gene from human respiratory syncytial virus inEscherichia coli: Mapping of a fusion inhibiting epitope

VIROLOGY

184, 428-432

Expression

(1991)

of the F Glycoprotein Gene from Human Respiratory Syncytial Virus in Escherichia Mapping of a Fusion Inhibiting Epitope

co/i:

KAREN A. FIEN,* BRANDAT. Hu,* JOHN F. FARLEY,t ROBERT SEID,+ ANTONIA MARTIN-GALLARDO,*“B~ PETER L. COLLINS,~ STEPHEN W. HILDRETH,* AND PETER R. PARADISO* *Department 300

of Virology Research, tDepartment of Molecular Biology, and *Department East River Road, Rochester, New York 14623; and SLaboratory of Infectious and Infectious Diseases. 9000 Rockville Pike, Bethesda, Received

November

9, 1990;

accepted

June

of Protein Chemistry, Praxis Biologics, Diseases, National Institute of Allergy Maryland 20892

Inc.,

7, 199 1

A cDNA copy of the gene encoding the entire amino acid sequence of the fusion (F) protein of human respiratory syncytial virus (strain A2) was inserted into a bacterial expression vector containing the lambda P, promoter. Upon heat induction, Escherichia co/i cells harboring the vector produced a 45-KDa peptide which reacted with rabbit polyclonal antiserum to the native F protein. Expression of the F gene resulted in severe inhibition of bacterial growth, which was overcome by deletion of the DNA sequences encoding the F signal peptide. The region of the F protein which reacted with a virus-neutralizing and fusion-inhibiting monoclonal antibody was probed by expressing cDNA fragments encoding different protein domains in E. co/i and testing antibody reactivity by Western blot analysis. Analysis of six fragments yielded an overlapping antibody-reactive region between amino acids 253 and 298. Analysis of reactivity with a cassette of synthetic peptides confirmed that the virus-neutralizing epitope mapped between residues 289 and 298 o 1991 Academic PWS, IIW. defined by the amino acid sequence M-S-l-l-K-E-E-V-L-A.

We have worked with the monoclonal antibody (MAb) designated L4 (3), which is both virus-neutralizing and fusion-inhibiting and which recognizes the 48KDa F, subunit of the fusion protein (73). In an attempt to map this epitope, we have constructed recombinant plasmids that express F gene sequences of RSV in Escherichia co/i. We report (i) the bacterial expression of a full-length cDNA copy of the RSV F gene and (ii) the mapping of the antigenic site defined by MAb L4 using recombinant fragments of the F, fusion protein and synthetic peptides. Since expression of several viral glycoproteins has been described to be lethal to the bacterial host (16, 77), the vector pCQV2 (18), which allows regulated expression of foreign proteins, was used for expression of the entire F coding sequences. This plasmid contained the P, promoter and translation start point from bacteriophage lambda and a BamHl site for insertion of a gene under the control of these initiation signals. The F cDNA was obtained by BamHl digestion of a plasmid that contained a full-length cDNA copy of the RSV A2 fusion protein gene cloned into the f3amHl site of PBR322 (9, 15), the sticky ends were filled in with the Klenow fragment of DNA Polymerase I, and the F segment was inserted into the filled-in BamHl site of pCQV2 (see Fig. 2). The new plasmid (pCQV2-F) should direct the synthesis of an F-like protein by readthrough from the lambda ATG. E. co/i RR1 cells (19) harboring the vector with and without the

Human respiratory syncytial virus (RSV) is a member of the family Paramyxoviridae (1) and as such has a fusion (F) glycoprotein which functions to fuse virus and cell membranes during the infectious process (2, 3). The F protein (68 KDa) is synthesized as an inactive precursor, F,, which is activated by proteolytic cleavage into two disulfide-linked subunits: F, and F, (48 and 23 KDa, respectively) (4). Antibodies to the F glycoprotein have been found to both neutralize virus and inhibit fusion (3, 5). Monoclonai antibodies which have these biological activities can passively protect cotton rats and mice from live virus challenge (6, 7). Active immunization with vaccinia virus-delivered F protein (8, 9) also protected cotton rats against RSV infection. Several authors have identified four antigenic sites on the F protein using monoclonal antibodies (IO- 13) and showed that at least three of these epitopes are involved in neutralization. Trudel et al. (1987) have mapped the region of reactivity of a monoclonal antibody (7C2), which appears to define a major neutralization epitope (12, 14). The 7C2 monoclonal antibody recognized a region between amino acids 221 ILe and 232Glu of the mRNA-deduced F sequence of the A2 strain of human RSV (15). ’ To whom correspondence and requests for reprints should be addressed. ’ Present address: National Institute of Neurolbgical Disorders and Stroke, NIH. 12420 Parklawn Dr, Rockville, MD 20852. 0042-6822/91

$3.00

Copynght 0 1991 by Academic Press, Inc. All rights of reproductnn in any form reserved.

428

SHORT

COMMUNICATIONS

-18.5 FIG. 1. Synthesis of RSV F protein in E. co/i. Bacterial cells bearing either pCQV2 or pCQV2-F were grown to mid-logarithmic phase at 30” in L-Broth containing 100 mg/ml ampicillin and then shifted at 42” for 30 min. Cells were harvested bycentrifugation, resuspended in 10 mM Tris, pH 7.4, and lysed by French press. The lysates were boiled in the absence (lanes 1,2) or presence (lanes 3,4) of @-mercaptoethanol, electrophoresed, and immunoblotted with rabbit antiRSV F polyclonal serum. The same amount of cell extracts was loaded in lanes l-4. Lanes 1 and 3 are from cells bearing pCQV2 (control); lanes 2 and 4 are from cells bearing pCQV2-F; and lane 5 is from purified F protein from RSV-infected cells. The 23-KDa (F2) subunit of the authentic F glycoprotein from RSV-infected cells was not detected at the concentrations of protein and polyclonal antiserum used, showing that the polyclonal antiserum reacted primarily with the F, subunit. Marker molecular weights are noted to the right.

cDNA insert were grown to mid-logarithmic phase at 30” and expression was induced by shifting to 42” for 30 min. A 45-KDa protein was found in cells carrying the F cDNA when analyzed by Western blots using polyclonal anti-F serum under both reducing and nonreducing conditions (Fig. I, lanes 2 and 4). This protein was not detected in control samples lacking the F cDNA (lanes 1 and 3). The distinction between experimental and control samples was noted more clearly in reduced preparations (lanes 4 vs 3). Peptides of 23 and 18 KDa were also selectively found in pCQV2-F-harboring cells (Fig. I, lane 4). A protein with the size expected for the entire unglycosylated Fl,2 protein (65 KDa) was not detected in either reduced or unreduced cell extracts carrying pCQV2-F. The 45-KDa peptide, which also reacted with MAb L4 (data not shown), has the expected size of the unglycosylated F, subunit (13). This would suggest that cleavage of the bacterial recombinant F protein at the F2/F, junction occurred. However, proof of this would require detection of a bacterially expressed protein of the expected size (15 KDa) of the unglycosylated F2 subunit, and this could not be directly demonstrated

429

because the available antisera did not react with the F, subunit of the authentic F protein from RSV-infected cells. Therefore, we cannot eliminate the alternate possibility that the 45-KDa species was a breakdown product of an uncleaved recombinant F protein rather than the unglycosylated F, subunit generated by cleavage at the FJF, junction. Plasmid pCQV2-F did not produce high yields of F-derived protein. In addition, heat induction resulted in severe growth inhibition that can be correlated with the presence of F protein sequences (Fig. 2A). We were interested in determining if any of the F hydrophobic domains accounted for the toxic effects of F expression in f. co/i. F cDNA fragments which lacked the sequences encoding the hydrophobic COOH-terminus (pCF2), the hydrophobic NH,-terminus (pCFS), or all three hydrophobic domains (pCF4) were cloned into pUC expression vectors (Fig. 2). The pUC vectors were chosen because of their cloning versatility and because they allow regulated expression of a gene when introduced into a /acP strain of f. co/i (20). pCF2 contained a 1474-bp BamHI-Nsil fragment encoding residues l-489 of the F protein; pCF5, a 1572-bp Accl-Nsil F fragment (amino acid residues 54-574); pCF4, a 828-bp Pstl-Nsil F fragment (amino acid residues 212-489). In all three constructions the F sequences were in frame with the lac (Ygene initiation codon, as determined by dideoxy sequencing (21). The F amino acid sequences and the NH,- and COOH-terminal structures (as deduced by the nucleotide sequence) of the proteins encoded by pCF2, pCF5, and pCF4 versus those encoded by pCQV2-F are shown in Fig. 2. f. coliJMlO3 cells (22) bearing pCF2, pCF5, or pCF4 were grown at 37” to mid-logarithmic phase and isopropyl P-o-thiogalactoside (IPTG) was added to half of the cultures to induce expression of the F recombinants. The ratio of bacterial growth under noninduced and induced conditions at 1.5 hr is described in Fig. 2. Addition of IPTG to cells bearing pCF2 resulted in severe inhibition of E. co/i growth (Fig. 2B), as previously observed with cells carrying pCQV2-F upon heat induction (Fig. 2A). In contrast, pCF5-bearing cells exhibited normal growth characteristics upon IPTG induction (Fig. 2C), suggesting that the toxic domain had been eliminated from the F-like protein specified by pCF5. IPTG-induced cells bearing pCF4 also had normal growth (Fig. 2D). Thus, deletion of the N-terminal signal sequence of the expressed F protein alone restored normal growth, a finding which is similar to previous observations for the expression of viral glycoproteins in E. co/i (IS, 17). The toxicity associated with the expression of the signal sequence is thought to be due to its interaction with bacterial membranes (16). It also

430

SHORT

COMMUNICATIONS

FIG. 2. Structures of the recombinant RSV F proteins encoded by the expression plasmids pCQV2-F (A), pCF2 (B), pCF5 (C), and pCF4 (D) and the effect of their expression on bacteria cell growth. The F coding sequences are shown as double open lines with the hydrophobic sequences indicated by solid boxes: the signal peptide (I), the fusion-related domain (II), and the membrane anchor (Ill). Plasmid-encoded amino acids are indicated by stripped boxes. The Nsil and Accl restriction sites of the F cDNA that were used for the construction of pCF2 and pCF5, respectively, and the Pstl site used for that of pCF4 are indicated (A). ON (B) designates a 39.mer oligonucleotide which contains translation termination codons downstream of various restriction sites (BarnHI, Smal, /WI) and has Sall and HindIll cohesive ends at the 5’ and 3’ end, respectively (tcgaCGGATCCCGGGATGCATATGATCGATAATTAATTAagct); this oligonucleotide was synthesized on an Applied Biosystems (Foster City, CA) DNA synthesizer and inserted between the Sall and HindIll sites of pUC8. The nucleotide sequence at the junctions was determined by dideoxy sequencing (21). The N- and C-terminal residues of the F protein segments are underlined, and their positions in the amino acid sequence of the complete F protein are indicated. Additional amino acids preceding or following the underlined residues are those contributed by coding sequences in the expression vector. Cultures of pCQV2 and pCQV2-F (A) were grown at 30” to a cell density of klett 80 (OD550 = 0.2) and then shifted to 42”; cultures of pCF2 (B), pCF5 (C), or pCF4 (D) were grown in the absence or presence of 1 mM IPTG. which was added at klett 80. Bacterial growth was monitored by measuring cell density at 30.min intervals during 2 hr after induction of expression of RSV F sequences. The ratro of bacterial growth under noninduced and induced condrtions at 1.5 hr is described.

had seemed possible that expression of the hydrophobic fusion-related domain of the F, subunit would be toxic, both because of its fusogenic activity in mammalian cells and because this domain has been shown to associate with bacterial membranes (23). This possibility would be particularly relevant if, as suggested in the previous section, cleavage of the FJF, junction occurred. However, the normal growth of cells bearing pCF5, which lacked the N-terminal signal sequence but retained both the fusogenic and the membrane anchor domains and also yielded a 45KDa product (not shown), showed that the toxicity was associated only with the signal sequence.

Since coding sequences from the F, subunit, which contained a neutralizing epitope defined by MAb L4 (IS), were efficiently expressed in E. co/i, we selected this expression system to map the region of reactivity of MAb L4. Regions of the fusion protein gene were excised from the entire F cDNA by restriction endonuclease digestion (Fig. 3): pCF3 contained a 974-bp /-/pal-BarnHi fragment encoding residues 253-574 of the F protein; pCF4, which has been described above, a 828-bp fs&‘Vsil F fragment (amino acid residues 212-489); pCF8, a 515-bp Pstl-Hincll F fragment (amino acid residues 212-384); pCF7, a 452-bp PstlTaql F fragment (amino acid residues 212-364); pCF10, a 257-bp Pstl-Noel Ffragment (amino acid residues 212-298); and pCF11, a 571 -bp Ndel-Nsil F fragment (amino acid residues 298-489). Each of these fragments was inserted into the polylinker region of pUCl9, in such manner that the F sequences were in frame with the lac CYgene initiation codon. The DNA sequence at the junction between the vector and the 5’ end of the F fragment was determined for each plasmid by the dideoxy chain-termination method (21) and confirmed as an in-frame insertion of the F fragment. E. co/i JM103 or DH5a carrying each of the F-containing pUC plasmids was grown to mid-log phase at 37” in L-Broth medium and induced with IPTG (1 mM final concentration) for 1 hr. Cells were harvested by centrifugation, resuspended in 10 mM Tris, pH 7.4, and lysed by French press. The lysates were boiled in the presence of P-mercaptoethanol and subjected to electrophoresis (25) and immunoblotting (26) with either polyclonal anti-RSV F serum and ‘251-protein A or MAb L4, rabbit anti-mouse immunoglobulin G and ‘251-protein A. Polyclonal serum to F protein purified from RSV-infected cell culture (5) was prepared in rabbits. MAb L4 was kindly provided by E. Walsh (University of Rochester, Rochester, NY). All six recombinant protein constructions depicted in Fig. 3 were expressed, as determined by Western blots probing with polyclonal anti-F serum. Each plasmid yielded a F-related peptide which had the size expected from the deduced amino acid sequence (data not shown). When recombinant proteins were tested for reactivity to MAb L4 by the same method, all reacted positively except for CFl 1 which was not detected, as diagrammed in Fig. 3. The fusion protein sequence common to all of the recombinant proteins which were reactive with the MAb L4 (CFlO, CF7, CF8, CF4, and CF3) and absent in the nonreactive CFl 1 protein was defined by residues 253Thr-298Ala. The L4 binding region defined by recombinant F fragments overlapped with that defined by previous analysis of proteolytic fragments of the native F protein, which identified a region between amino acid 283 and 315 as that binding to MAb L4 (P. Paradiso et a/.,

SHORT COMMUNICATIONS

431

CF 10 L4 Reactive CF7

Fragments

/El4

Fragments

H

500

574

I

I

L4 Unreactive

CFS cF4

zctl I

137

I

253

2%

300

400 I

CFll FIG. 3. Diagram of the recombinant F proteins on a linear map of the F, subunit. The fragments shown in shaded bars were reactive with the L4 monoclonal antibody and those shown in solid bars were nonreactive with the L4 monoclonal antibody. The cross-hatched box represents the predicted binding site of the L4 monoclonal antibody, based on the reactivity pattern shown.

submitted for publication). Taken together, these data suggested that the L4 epitope was between residues 283 and 298. To test this hypothesis, a series of synthetic peptides were made and tested for reactivity with MAb L4 by dot blot (Table 1). All of the peptides reacted except the one from amino acid 299 to 315. This corresponds to the lack of reactivity of the recombinant CFl 1, which started at amino acid 298. Peptides III (amino acids 289-315) and IV (amino acids 283-3 15) specifically reacted with MAb L4, in contrast TABLE 1 MABL~

SEQUENCEOFSYNTHETICPEPTIDESTESTEDFORREACTIVITYWITH

Peptidee I:

299-315

II:

294-315

Ill:

289-315

IV:

283-315

L4 Reactivityc

Sequenceb

-

YVVQLPLYGVIDTPCWK EEVLAYVVQLPLYGVIDTPCWK MSIIKEEVLAYVVQLPLYGVIDTPCWK QQSYSIMSIIKEEVLAYVVQLPLYGVIDTPCWK

I 283

I

I

289

294

I 299

+ + +

I

315

a Peptides were synthesized on an Applied Biosystems 430A peptide synthesizer using standard software programs, The homogeneity of the peptide was characterized by reverse-phase high-pressure liquid chromatography (HPLC). b Sequences are using the single letter amino acid code. To confirm the presence of the correct sequence, all HPLC-purified peptides were subjected to automated Edman degradation with an Applied Biosystems 477A protein sequenator. ’ Reactivity of synthetic peptides with MAb L4 was measured by dot blot analysis. Peptides were resuspended in phosphate-buffered saline at a concentration of 5 mg/ml and aliquots containing 5, 10, 15, and 20 pg of each peptide were applied onto nitrocellulose membranes, which were incubated succesively with MAb L4, rabbit antimouse IgG, and ‘261-protein A.

to peptide II (amino acids 298-315) which also bound to other non-F-related monoclonal antibodies. The synthetic peptides III and IV bound to L4 with similar affinities, as determined by the intensities of the hybridization spots at the various concentrations tested (data not shown). These results indicated that the L4 epitope can be refined to residues 289Met-298Ala of the RSV F protein. The antigenic site defined by MAb L4 mapped 57 amino acids downstream from that identified by MAb 7C2 (amino acids 221 ILe-232Glu), which is also a neutralization and linear epitope (12, 14). The L4 monoclonal antibody is both virus-neutralizing and fusion-inhibiting in vitro (3). Furthermore, passive immunization with MAb L4 protected mice and cotton rats against RSV infection (6, 7). We are currently investigating the possibility that recombinant fragments of the F protein produced in f. co/i will induce neutralizing and protective antibodies. Similarly, peptides specifying the L4 epitope may hold the potential for a synthetic vaccine approach. Clearly, both forms of the F antigen would be significatively different from the native, glycosylated protein and might not be sufficiently immunogenic. ACKNOWLEDGMENTS We thank Dr. Algis Anilionis and Dr. Rasappa Arumugham fortheir advice and contribution to this work, Dr. Edward Walsh for providing monoclonal antibody L4, and Robert Woods for technical assistance. This work was supported in part by a SBIR grant from the National Institute of Allergy and Infectious Diseases (Bethesda, MD).

REFERENCES 1. MCINTOSH,

K., and

CHANOCK,

R. M., ln "Virology"

(B. N. Fields,

Ed.), p. 1285. Raven Press, New York, 1985. 2.

MERZ,

D. C., SCHEID, A., and CHOPPIN.

275-278

(1982).

P. W., J. Exp.

Med.

151,

432

SHORT

COMMUNICATIONS

3. WALSH, E. E., and HRUSKA, J.,/. Viral. 47, 171-177 (1983). 4. SCHEID, A., and CHOPPIN, P. W., Virology 69, 265-277 (1977). 5. WALSH, E. E.. BRANDISS, M. W.. and SCHLESINGER, J. J., I. Gen. No/. 66, 409-415 (1985). 6. TAYLOR, G., STOT~, E. J., FERNIE, B. F., COTE, P. J., COLLINS, A. P., HUGHES, M., and JEBBER, J., immunology 52, 137-l 42 (1984). 7. WALSH, E. E., SCHLESINGER, J. J., and BW\NDISS, M. W., Infect. Immune/. 43, 756-758 (1984). 8. JOHNSON, P. R. JR., OLSMSTED, R. A., PRINCE, G. A., MURPHY, B. R., ALLING, D. W., WALSH, E. E., and COLLINS, P. L.. J. Viral. 61, 3163-3166 (1987). 9. OLMSTED, R. A., ELANGO, N., PRINCE, G. A., MURPHY, 8. R., JOHNSON, P. R., Moss, B., CHANOCK. R. M., and COLLINS, P L., Proc. Nat/. Acad. Sci. USA 83, 7462-7466 (1986). 10. BEELER, J. A., and COELINGH, K. V. W., /. Viral 63, 2941-2950 (1989). 11. SAMSON, A. C. R., VILCOCKS, M. M., ROUTLEDGE, C. G., MORGAN, L. A., and TOMS, G. L., /. Gen. Viral. 67, 1479-1483 (1986). 12. TRUDEL, M., NADON. F., SEGUIN, C., PAYMENT, P., and TALBOT, P. J., Can. 1. Microbial. 33, 933-938 (1987). 13. WALSH, E. E., COTE, P. J., FERNIE, B. F.. SCHLESINGER, BRANDISS, M. W., J. Gen. tirol67, 505-513 (1986).

J. J., and

14. TRUDEL, M., NADON, F.. SEGUIN, C., DIONNE. G., and LACROIX, M., J. Gen. l&o/. 68, 2273-2280 (1987). 15. COLLINS, P. L., HUANG, Y. T., and WERTZ. G. W., Proc. Nat/. Acad. SC;. USA 81,7683-7687 (1984). 16. ROSE, J. K., and SHAFFERMAN. A., hoc. Nat/. Acad. Sci. USA 78, 6670-6674 (1981). 17. STEINBERG, D. A., WATSON, R. J., and MAIESE, W. M., Govf43, 311-317 (1986). 18. QUEEN, C., 1. Mol. Appl. Genet. 2, l-l 0 (1983). 19. BOLIVAR, F., RODRIGUEZ, R. L., GREENE, P. J., BETLACH, M. C., HEYNECKER, H. L., BOYER, H. W., CROSA, J H., and FALKOW, S., Gene 2, 95 (1977). 20. VIEIRA, J., and MESSING, J., Gene 19, 259-268 (1982). 21. SANGER, F., NICKLEN, S., and COULSON, A. R., Proc. Nat/. Acad. SC;. USA 74, 5463-5467 (1977). 22. MESSING, J., CREA, R., and SEEBURG. P. H., Nucleic Acid Res. 9, 309 (1981). 23. DAVIS, N. G., and Hsu, M.-C., Proc. Nat/. Acad. SC;. USA 83, 5091-5095 (1986). 24. Deleted In press. 25. LAEMMLI, U. K., /Vatwe (London) 227, 680-685 (1972). 26. TOWBIN, J., STAEHELIN, T., and GORDON, J.. Proc. Nat/. Acad. SC;. USA 76,4350-4354 (1979).