INFECTION AND IMMUNITY, Mar. 1993, p. 1004-10150019-9567/93/031004-12$02.00/0Copyright © 1993, American Society for Microbiology

Escherichia coli Heat-Labile Toxin Subunit B Fusions withStreptococcus sobrinus Antigens Expressed by Salmonellatyphimurium Oral Vaccine Strains: Importance of the Linker

for Antigenicity and Biological Activities of theHybrid Proteins

ELZBIETA K. JAGUSZTYN-KRYNICKA,1t JOSEPHINE E. CLARK-CURTISS,12AND ROY CURTISS III1*

Departments ofBiology' and Molecular Microbiology,2 Washington University, St. Louis, Missouri 63130

Received 15 June 1992/Accepted 15 December 1992

A set of vectors possessing the genes for aspartate semialdehyde dehydrogenase (asd) and the B subunit of theheat-labile enterotoxin of Escherichia coli (LT-B) has been developed. These vectors allow operon or gene

fusions of foreign gene epitopes at the C-terminal end of LT-B. Two groups of vectors have been constructedwith and without leader sequences to facilitate placing of the foreign antigen in different cell compartments.Two Streptococcus sobrinus genes coding for principal colonization factors, surface protein antigen A (SpaA),and dextranase (Dex), have been fused into the 3' end of the LT-B gene. Resulting protein fusions of -120 to130 kDa are extremely well recognized by antibodies directed against both SpaA and Dex as well as againstLT-B domains and retain the enzymatic activity of dextranase and the biological activity of LT-B in that theybind to GM1 gangliosides. Maximum antigenicity was obtained with the vector possessing an intervening linkerof at least six amino acids with two proline residues. Some of the fusion proteins also exhibited another propertyof LT-B in that they were exported into the periplasm where they oligomerized. LT-B-SpaA and LT-B-Dexhybrid proteins are expressed stably and at a high level in avirulent Salmonella typhimurium vaccine strainswhich are being used to investigate their immunogenicity and types of induced immune responses. The fusionvectors will also be useful for production and purification of LT-B fusion antigens to be used and evaluated inother vaccine compositions.

Heat-labile enterotoxin (LT-I) produced by human andporcine Escherichia coli isolates and cholera toxin (CT)produced by Vibrio cholerae are closely related. Both pro-teins manifest significant amino acid homology and shareseveral biochemical and functional properties including themechanism of action. Periplasmic LT-I consists of fivenoncovalently bound B subunits and one A subunit. Thenontoxic B subunits are responsible for binding the proteinto GM, ganglioside located on the surface of intestinalepithelial cells. The A subunit, which undergoes proteolyticcleavage into two fragments (A, and A2), bears enzymaticactivity. Intracellularly, the A, protein activates adenylatecyclase, which causes the elevation of the cellular level ofcyclic AMP, resulting in watery diarrhea (for a review, seereference 22).CT and LT-I are also immunologically related in that they

share several common epitopes, as has been demonstratedby neutralization and immunodiffusion assays (9, 10). Jacobet al. (32) proved that the small synthetic peptides represent-ing the conserved regions of the CT-related enterotoxinfamily generate an effective cross-reactive immune re-sponse. In addition to common antigenic parts, each toxinalso has unique determinants in its B subunit (23).

Recently, a second group of E. coli heat-labile toxins(LT-II) has been described and characterized. Although theLT-II enterotoxin has protein structure and mechanism of

* Corresponding author.t Present address: Institute of Microbiology, Warsaw University,

00-046 Warsaw, Nowy Swiat 67, Poland.

action that are characteristic of LT-I, the toxins differ withrespect to immunoreactivity and ganglioside-binding speci-ficity (25, 44). Moreover, no significant nucleotide homologybetween the genes coding for their B subunits has beenobserved (43).LT-I and CT and their B subunits were described as very

strong immunogens which elicit high-titer serum and secre-tory antibodies when administered into the gut via the oralroute in microgram amounts of protein (21, 45) or whendelivered to the lymphoid tissue by attenuated Salmonellastrains (7, 8, 11). The B subunits of both enterotoxins alsoenhance the immunogenicity of relatively poor immunogenswhen mixed or coupled (chemically or genetically) and givenorally (13, 15, 21, 31, 40, 41). Czerkinsky et al. (13) demon-strated that intragastric immunization of mice with Strepto-coccus mutans antigen I/II covalently coupled to the Bsubunit of CT elicits a significant immune response (mucosaland serum immunoglobulin A [IgA] and IgG), althoughsubclinical doses of free CT were required to obtain theoptimal antibody response. A vaccine consisting of the Bsubunit of LT cross-linked to the synthetically produced ST(E. coli heat-stable toxin) (36) has been shown to be immu-nogenic for both components. A similar immune responsewas elicited by a vaccine composed of a completely syn-thetic 44-amino-acid peptide comprising the major antigenicdeterminant of LT-B joined to ST (31, 35).An alternate approach to studying the immune-enhancing

effects of the CT-related family of enterotoxins is to con-struct gene fusions in which genes specifying foreign anti-genic determinants are fused to LT-B or CT-B genes. Thereare several advantages of genetic fusions over chemically

1004

Vol. 61, No. 3

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

S. SOBRINUS ANTIGEN FUSIONS TO LT-B 1005

synthesized conjugates. First of all, hybrid proteins ex-pressed by genetic fusions should exhibit more defined andhomogeneous structure and conformation. It is thereforemore likely that the hybrid proteins would retain some nativeproperties of both components. Fused proteins created bygenetic manipulations can also be delivered to the gut-associated lymphoid tissue by using attenuated Salmonellastrains, a delivery system which has been shown to be anefficient way of stimulating immune responses to recombi-nant antigens (12).

Since the mechanisms by which CT or LT or their Bsubunits enhance desired immune responses are not com-pletely understood (4, 24), the design of plasmid vectors togenerate fusions between parts of these toxins and otherantigens will yield compositions that can be used to enhanceour understanding of these mechanisms. In this regard,several plasmid vectors which permit gene fusions to the 3'or 5' end of the LT-B or CT-B gene have been constructed(16, 51). The most extensively studied LT-B or CT-B fusionsare the ones with the small, nonimmunogenic E. coli ST(47-49). Sanchez et al. (47, 48) constructed a CT-B expres-sion vector in which the structural gene encoding the Bsubunit of CT was fused to part of the LT-B gene specifyinga signal sequence, and the gene fusion was put under controlof the strong constitutive tacP promoter. In this system, theauthors studied the antigenicity and immunogenicity ofCT-B-ST nontoxic hybrid proteins (48). When tested byinjection in rabbits, the chimeric proteins stimulated signif-icant anti-ST and anti-CT-B responses; however, the anti-sera obtained exhibited low toxin neutralizing activity. Theyalso pointed out the importance of the free carboxyl end ofST for antigenicity of the fused protein. Clements (6) alsodescribed the construction of a nontoxic fusion peptide(LT-B-ST) which can be used as a vaccine against an E. colistrain producing both the heat-labile and the heat-stabletoxins. Antibodies raised against the purified LT-B-ST chi-meric protein were able to recognize both LT-B and STproteins and also neutralized the biological activity of nativeST. Also, an epitope of S. mutans glucosyltransferase fusedinto the N-terminal end of CT-B (17) and epitopes of thehepatitis B virus joined to the carboxyl end of LT-B (51, 52)resulted in hybrid proteins, which have been shown to beantigenic but not always immunogenic.

In this report, we describe the construction of severalAsd+ LT-B+ vectors allowing fusions of foreign genes to theC terminus of LT-B. The Asd+ vectors ensure stable in vivomaintenance in attenuated Aasd Salmonella typhimuriumvaccine strains in the absence of external selection (26, 42).This is because the Aasd mutation confers a requirement fordiaminopimelic acid (DAP) (which is not present in animals),such that loss of the vector leads to DAP-less death (26, 42).To control placing the fused protein antigens in different cellcompartments, the LT-B signal sequence was removed fromsome constructs. Since LT-B has been described as a strongimmunogen with immune-enhancing effects and since secre-tory IgA plays a key role in blocking tooth colonization, wechose two Streptococcus sobrinus genes encoding coloniza-tion factors, surface protein antigen A (SpaA) and dextra-nase (Dex), to be fused to the LT-B gene. Both hybridproteins (LT-B-SpaA and LT-B-Dex), composed of an ap-proximately 100-kDa streptococcal protein fused to a small13-kDa LT-B protein, were strongly antigenic and reactedwith antibodies directed against both domains (LT-B andstreptococcal antigen). Some fused proteins retained otherproperties of native LT-B, such as the ability to oligomerize

TABLE 1. Bacterial strains

Strain Relevant genotype reference

E. coli K-12X6097 F- ara A(pro-lac) 4080dlacZAMJS rpsL 42

AasdA4 thix6098 F- ara A(pro-lac) X(c1857 b2 redB3) x6097

+80dlacZAMlS rpsL AasdA4 thiS. typhimuniumUK-1 X3987 F- Acrp-11 AasdA4 Acya-12 20SR-11 X4072 F- gyrA1816 Acrp-1 AasdAl Acya-1 42

(quite possibly into pentamers) and exhibit affinity to GM,and of dextranase to exhibit enzyme activity.

MATERIALS AND METHODS

Strains and growth conditions. Strains and plasmids usedin this study are listed in Tables 1 and 2, respectively.Bacterial strains were grown aerobically at 37°C on L agar orin L broth (39) and when appropriate, ampicillin (50 p,g/ml)(Sigma Chemical Co., St. Louis, Mo.) and DAP (50 p,g/ml)(Sigma) were added.DNA manipulations. Recombinant clones were screened

for plasmid DNA by a modification of the Birnboim (3)alkaline minilysate technique, starting with 1.5 ml of L brothovernight culture. Whenever necessary, plasmid DNA wasisolated from 1 liter of L broth overnight culture as describedby Sambrook et al. (46) and purified by CsCl-ethidiumbromide buoyant density gradient centrifugation. Restrictionenzyme digestions and ligation reactions were carried out asdescribed by the manufacturers (International Biotechnolo-gies, Inc., New Haven, Conn.; New England Biolabs, Inc.,Beverly, Mass.; Promega Corporation, Madison, Wis.). Therecessed 3' termini created by digestion of DNA withrestriction enzymes were converted into blunt ends by usingthe Klenow fragment of DNA polymerase I (IBI). Reactionswere carried out for 30 min at 37°C in the presence of 1 mMdeoxynucleoside triphosphate. To prevent self-ligation ofthe vector DNA, 5' phosphate was removed from linearizedvector DNA by using calf intestinal alkaline phosphatasefrom Promega. Oligonucleotides were synthesized by T.Keller, Integrated DNA Chemistry Facility (Biology Depart-ment, Washington University). Linkers, if not chemicallyphosphorylated, were enzymatically phosphorylated by us-ing bacteriophage T4 polynucleotide kinase (New EnglandBiolabs) prior to ligation. Reactions were carried out by themethod of Sambrook et al. (46).

Transformation. E. coli cells were transformed accordingto a CaCl2 MOPS [3-(N-morpholino)propanesulfonic acid]procedure (46). Salmonella strains were transformed byelectroporation as recommended by the manufacturer (Bio-Rad Laboratories, Richmond, Calif.). Since rough Salmo-nella strains are nonimmunogenic, all Salmonella transfor-mants were verified for smooth phenotype by checkingbacteriophage P22 sensitivity (58). That normal amounts oflipopolysaccharide were synthesized by the Salmonellaclones was verified by sodium dodecyl sulfate-polyacrylam-ide gel electrophoresis (SDS-PAGE) and silver staining (55).

Determination of protein amounts. Protein concentrationwas measured by the bicinchoninic acid protein assay(Pierce, Rockford, Ill.).

Electrophoretic techniques. DNA was electrophoresedthrough 0.8, 1, or 1.2% agarose gels (Sigma) by using

VOL. 61, 1993

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

1006 JAGUSZTYN-KRYNICKA ET AL.

TABLE 2. Plasmids

Plasmid Relevant phenotype Source or reference

pYA292 Asd+ LacZa 26pYA810 Asd+ From pYA292 by removal of 160-bp

HindIII fragment of lacZctpYA812 pBR322 with E. coli lacIq gene cloned into the EcoRI site Constructed by M. SathishpYA993 pBR322 with S. sobrinus dextranase gene cloned into the BamHI site 56pYA2905 pYA292 with a 1.5-kb SstI fragment containing three tandem repeats of part 27

of the S. sobnnus spaA genepYA3010 pYA292 with a 2.6-kb PvuII fragment containing the S. sobrinus dextranase 57

genepEWD299 7.8-kb plasmid with the gene specifying the porcine heat-labile enterotoxin W. S. Dallas (14)

(LT)pBluescript II KS LacZa Apr phagemid cloning vector Stratagene Cloning Systems

Tris-acetate-EDTA buffer (46). Bacteriophage X DNA frag-ments generated by digestion with HindIII or bacteriophage4~x174 DNA digested with HaeII were used as molecular sizestandards (BRL Life Technology Inc., Gaithersburg, Md.).Linear DNA fragments were isolated from agarose gels byusing a Prep-A-Gene kit (Bio-Rad) according to the manu-facturer's directions.

Electrophoresis of proteins were performed in SDS-10 to15% polyacrylamide gels by the method of Laemmli (38). Weused Rainbow Markers from Amersham (Arlington Heights,Ill.) as molecular weight standards. Whenever necessary,blue dextran, an indicator substrate for dextranase (Pharma-cia LKB Biotechnology AB, Uppsala, Sweden), was incor-porated into the SDS-PAGE by mixing the blue dextran to afinal concentration of 0.5% prior to the addition of thecross-linking and catalytic agent (2). After electrophoresis,renaturation of dextranase was performed as previouslydescribed (2).Western analysis. Protein samples were separated by SDS-

PAGE and then electrophoretically transferred to nitrocel-lulose sheets. The membrane was blocked for 2 h in blockingbuffer (100 mM Tris [pH 8.0], 0.25% gelatin, 0.5% bovineserum albumin [BSA], 0.05% sodium azide, 0.04% Tween80), followed by 2 to 12 h of incubation at room temperaturewith primary rabbit antibodies (against S. sobnnus SpaA,dilution 1:3,000; against S. sobrinus Dex, dilution 1:3,000; oragainst E. coli LT-B, dilution 1:1,000; the anti-LT-B serumwas the gift of Randall Holmes, Uniformed Services Univer-sity of the Health Sciences, Bethesda, Md.). These antibod-ies only recognize the monomeric form of LT-B in gels. Thenitrocellulose sheets were washed three times for 10 mineach with washing buffer (100 mM Tris [pH 8.0], 0.05%sodium azide, 0.04% Tween 80) and then incubated at roomtemperature for 2 h with secondary antibodies (goat anti-rabbit immunoglobulin coupled with alkaline phosphatase),diluted as recommended by the manufacturer (Sigma). Themembranes were washed twice for 10 min with washingbuffer and once with AP buffer (100 mM Tris [pH 9.5], 100mM NaCl, 5 mM MgCl2). The color was developed byexposure of the nitrocellulose to 5-bromo-4-chloroindoxylphosphate and nitroblue tetrazolium as described by Sam-brook et al. (46).

Oligomerization assay. One milliliter of log-phase L brothculture of E. coli carrying different Asd+ LT-B+ vectors orLT-B gene fusions was centrifuged, and the cells werewashed once with 50 mM Tris (pH 8.0) and resedimented.The cell pellet was suspended in 0.5% SDS, and the cellswere lysed by incubation at 60°C for 10 min or at 100°C for5 min. Lysates were clarified by centrifugation, and the

supernatant fraction was removed, mixed 1:1 with 2x sam-ple buffer (50 mM Tris [pH 6.8], 2% SDS, 0.01% bromophe-nol blue, 10% glycerol) (45), and applied on SDS-PAGE gels(30 ,ug of protein per lane).

Release of periplasmic proteins. The cold osmotic shocktechnique described by Hazelbauer and Harayama (29) wasemployed to determine localization of the LT-B fused pro-teins. Presence of an LT-B-Dex fusion protein in the cyto-plasmic or periplasmic protein fraction was checked byelectrophoresis of the protein on SDS-blue dextran-poly-acrylamide gels. LT-B-SpaA fusions were identified in thecytoplasmic or periplasmic fractions by SDS-PAGE andWestern blot (immunoblot) analyses.

Screening for clones expressing LT-B fusions. (i) Colonyimmunoblot. Clones expressing LT-B-SpaA fusion proteinswere often identified by colony immunoblot with the samebuffers, antisera, and chemical reagents as described abovefor Western blot analyses.

(ii) Hybridization. LT-B-SpaA clones expressing fusionproteins of low antigenicity were identified by colony hybrid-ization with a probe labeled by using the nonradioactiveDNA labeling and detection kit from Boehringer MannheimBio-Chemical Co, Indianapolis, Ind. The Bluescript-pKSII(Strategene, La Jolla, Calif.) plasmid with the spaA genecloned into it, pYA3085, was labeled with digoxigenin dUTPby a random primed reaction. Hybridizations and immuno-logical detection of the positive clones were performed asrecommended by the manufacturer.

(iii) Blue dextran assay. Clones producing enzymaticallyactive dextranase were identified by the blue dextran assay(33). E. coli X6098 cells were transformed with the ligationmixture and were plated on L agar plates. After overnightincubation at 30°C, 10 ml of the blue dextran solution (50mMsodium acetate [pH 5.51, 0.7% blue dextran, 1.5% agar) waspoured on the top of each plate. The plates were incubatedovernight at 42°C. Positive clones were identified by whitehalos arising around the colonies.GM1 binding assay. Reagents for the enzyme-linked im-

munosorbent assay (ELISA) were obtained from Sigma.GM1 gangliosides (Sigma) (1.5 ,ug per well) in carbonate-bicarbonate coating buffer (pH 9.6) were absorbed to themicrotiter wells during overnight incubation at 4°C. Afterrinsing the plates with washing buffer (phosphate-bufferedsaline [PBS] [46], 0.1% Tween 20), unoccupied sites on theplastic surface were blocked by a 1-h incubation with block-ing buffer (PBS, 0.1% Tween 20, 1% BSA). The plates wereincubated for 2 h with serially diluted crude extracts of theproteins in blocking buffer. Reactions were analyzed immu-nologically by sequential incubation with rabbit anti-LT-B,

INFECT. IMMUN.

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

S. SOBRINUS ANTIGEN FUSIONS TO LT-B 1007

goat anti-rabbit IgG conjugated with alkaline phosphatase,and alkaline phosphatase substrate. Reactions were stoppedby adding 50 ,ul of 1 M NaOH, and A405 was read.

RESULTS

Construction and properties of the family of LT-B+ vectors.A set of Asd+ LT-B+ vectors allowing fusion of foreignepitopes to the C-terminal end of LT-B has been con-structed. The vectors can be divided into two groups: thosewith and those without a signal sequence, which is necessaryfor LT-B to be transported into the periplasm. pYA810,which is pYA292 (26) lacking a 0.16-kb HindIII-HindIIIlacZa DNA fragment, has been used as the starting plasmidfor all constructions. pYA292 is a component of the balancedlethal host-vector system developed in our laboratory (42).In this system, the asd+ gene from S. typhimunium, anon-drug resistance selectable marker, is present on theplasmid vector DNA and complements an asd gene deletionpresent in the host bacterial chromosome (26, 42).The plasmid EWD299 (14), which is a derivative of

pBR313, codes for porcine LT and was our source of theLT-B gene. We cloned a 584-bp Sau3A-MaeI DNA fragmentof pEWD299 into pYA810 digested with BamHI and PstI byusing MaeI-PstI oligonucleotide linkers of different lengthand nucleotide composition. This strategy removed thetranslational stop codon but left the codon for the terminalLT-B asparagine residue intact. Figure 1A presents the wayin which the LT-B vectors were constructed as well asrestriction maps of the Asd+ LT-B+ vectors. The linkerscontained amino acid codons which are translated with highefficiency by E. coli cells. To minimize the conformationalchanges of LT-B, codons specifying rather small, unchargedamino acids were included, whenever possible. Even so,when codons for five amino acids (Tyr, Ala, Cys, Thr, andSer) were added to the 3' end of the LT-B gene, one of theconstructs specified a protein which was not exported intothe periplasm and was cytotoxic for the E. coli host cellwhenever expression of the LT-B gene was induced (datanot shown). Therefore, it seems that specification of aminoacids that are hydrophobic results in an unacceptable linker,a result originally noted by Sandkvist et al. (50). Four othervectors were therefore constructed by using 42-bp linkersspecifying less hydrophobic amino acids (37) to yieldpYA2906, pYA3047, pYA3048, and pYA3049 (Fig. 1).To be able to place the fusion proteins in different cell

compartments, nucleotides specifying a 23-amino-acidleader sequence have been deleted from the LT-B gene. Wetook advantage of the Sacl recognition site which is locatedat the beginning of the coding sequence for the mature LT-Bprotein. pYA3047 and pYA3048 plasmid DNAs were di-gested with EcoRI and SacI and religated by using an 8-bpsingle-stranded oligonucleotide (5' AATTAGCT 3') to yieldpYA3081 and pYA3082 (Fig. 1A). These constructs lack a287-bp DNA fragment and also lost EcoRI and SacI restric-tion sites. Figure 1B presents the nucleotide and deducedamino acid sequences of the multiple cloning sites located atthe 3' end of the LT-B gene. All vectors with uniquerestriction sites located in the multiple cloning site (MCS)permit cloning of foreign DNA in three different readingframes and also provide termination stop codons. Moreover,the MCS nucleotide sequences of pYA3047, pYA3048, andpYA3049 code for proline and glycine residues to createkinks that interrupt the a-helix and thus permit a morenatural conformational structure between the domains of theLT-B and the fused protein.

EcoRISmO HI

I Hind MI

A. BgI_ HindN //ECORIHinl X

Hinc ArL-

Bgll ~~Eco10Y8O gI pEWD299337kb 42 bp Moel-PstI 78 kb

Linker

BomHI 584 bp Sau 3A-MaeIPstI DNA Frogment

T4 DNA Ligos

C T GCG A GCA CrA CT AAC TAGCTAUCCAAGCTCpYA2906 Asn Tyr Ala Asp II* Thr Arg Ala LoU Ala Asnso.H1 MIu ApoLt Pal HindU

PYA3O47 pYAIOSI aCGCGaG ; ~~CTG TM CTA~CCAAGCTCCTCCC1oaI MInIApoLI PtIllf allud

pY^X4AAC2AWC GC06OGCGICJW &ACWACTG; A CTAGCkCCAAGCTTpYA3O4 p Asn Tyr Ala Pro Vly Amp Pro Thr Arg Ala LoaSoail M l Ap,LlPal HiadE

pYA3048 pYA30 AAC TAC6CG6COG TCG MCO6GCACTG ;%ACTAGTGAAGCTCCCAAGCTTAsin Tyr Ala Pro VSe TIAr Pro Thr Arg Ala Lou

FIG. 1. (A) Construction and restriction maps of the Asd+LT-B' vectors. (B) Multiple cloning sites at the carboxyl end of theLT-B sequence in Asd+ LT-B' vectors. The leftmost Asn is theC-terminal amino acid in LT-B. *, stop codon; +, vectors containinga signal sequence; -, vectors without a signal sequence. Allrestriction sites except the EcoRV site located in the MCS areunique. DNA sequence of the pYA810 MCS in which the BamHIsite was used to clone the LT-B gene is as follows: AGG AAA CAGACC ATG CCG GAA TTC GCA ATT CCC GGG GAT CCG TCGACC TGC AGC CAA GCT CCC AAG CTT (for details, seereference 25).

All vectors encode the full-length LT-B fragment plus nineamino acids added to the carboxyl end of the protein. The Bsubunits of LT are synthesized as precursors, which are thenprocessed and translocated through the cytoplasmic mem-brane. When present in the periplasmic space, they form anoligomer (presumably a pentamer) which dissociates intoconstituent monomers only after heating to >70'C (28, 30). Itwas of interest to determine if the C-terminal extensionsinfluenced the ability of LT-B to oligomerize. Crude proteinextracts from E. coli X6097 carrying Asd+ LT-B' vectorslysed by 0.5% SDS at 100 or 60°C have been analyzed bySDS-PAGE and Western blot analysis (Fig. 2). The proteinmonomers of -13 kDa specified by all of the Asd+ LT-B'vectors with leader sequences react with anti-LT-B antibod-ies (Fig. 2, lanes 1A, 2A, 3A, and 4A) and are transportedinto the periplasm where they pentamerize. Oligomer forms

VOL. 61, 1993

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

1008 JAGUSZTYN-KRYNICKA ET AL.

2 3~ 4 5 6 MW

r-fi 2AA3R rA B A Bni-B kDa

-9269

-46

-30

7 8 MW

A C A E kDa

-92-69

-14

FIG. 2. Oligomerization of LT-B coded by Asd+ LT-B+ vectors.Protein extract samples from strains harboring the designated plas-mids were either boiled for 5 min (A) or incubated for 10 min at 60°C(B) and subjected to SDS-PAGE, transferred to nitrocellulose, andreacted with anti LT-B antibodies. Lanes: 1, pYA2906; 2, pYA3047;3, pYA3048; 4, pYA3049; 5, pEWD299; 6, pYA292; 7, pYA3081; 8,pYA3082.

of LT-B with different C-terminal extensions, as well as

native LT-B encoded by pEWD299, do not react with theanti-LT-B antibodies used in these studies (Fig. 2, lanes 1B,2B, 3B, and 4B). Vectors lacking signal sequences expressedproteins of monomeric size of approximately 13 kDa (Fig. 2,lanes 7A and 8A) which do not oligomerize (Fig. 2, lanes 7Band 8B). These cytoplasmic proteins differ from the matureform of LT-B encoded by pYA3047 and pYA3048 by havingfour amino acids (Met, Pro, Glu, and Leu) added to the Nterminus. These alterations at both the C- and N-terminalends of LT-B do not abolish antigenicity.

All Asd+ LT-B+ vectors were introduced by electropora-tion into the attenuated oral S. typhimurium Acrp Acya Aasdvaccine strains X4072 and X3987 in which the LT-B gene was

expressed at a high level (data not shown).Gene fusions at the carboxyl end of LT-B. Two S. sobrinus

genes coding for important colonization antigens, surfaceprotein antigen A (SpaA) and dextranase (Dex), have beenfused to the 3' end of the LT-B gene. The main antigenicdeterminants of the spaA gene have been previously char-acterized (27) and recloned into the Asd+ vector pYA292 toyield pYA2905 (18). The following strategy has been em-ployed to construct the LT-B-spaA gene fusion. pYA2905plasmid DNA was digested completely with SmaI and thenpartially with EcoRI. A 2.7-kb EcoRI-SmaI fragment ofpYA2905 isolated from an agarose gel was ligated betweenthe EcoRI and SmaI restriction sites of the pBluescript II KSplasmid (Stratagene), yielding pYA3085, which was thesource of the spaA gene for the next constructs. The spaAgene cloned into pKSII is out of frame. However, whenpYA3085 DNA was cut with EcoRV and SmaI, the fragmentcontaining the spaA gene could be ligated into ApaLI-digested, Klenow-treated DNA of the Asd+ LT-B' vectorDNA (Fig. 1), in the proper reading frame. Initially, ApaLI-cut, Klenow-treated pYA2906 DNA was ligated with the2.7-kb EcoRV-SmaI fragment of pYA3085. The ligationmixture was used to transform E. coli X6097 carryingpYA812 (a pBR322 derivative with the lacIq gene). Trans-formants were selected for the Asd+ phenotype andscreened for the presence of the spaA gene product afterinduction with 1 mM isopropylthio-3-D-galactopyranoside(IPTG) by colony immunoblot. No strong immunologicallyreactive clone was found. Next, pYA3085 DNA was labeledby randomly primed nonradioactive digoxigenin-dUTP and

used as a probe for colony hybridization. With this screeningmethod, several clones containing spaA DNA fused into theLT-B gene were identified. Plasmid DNA was isolated fromone of them, designated pYA3086, and the construct wasverified by restriction analysis. Properties of the fusedprotein specified by this construct are described below.The same 2.7-kb EcoRV-SmaI fragment of pYA3085 was

ligated into ApaLI-cut, Klenow-treated DNA of Asd+LT-B+ vectors whose MCSs code for proline residues. Inevery case, we obtained several immunologically reactiveclones. We isolated pYA3055, which was derived frompYA3047; pYA3056, which was derived from pYA3048; andpYA3057, which was derived from pYA3049. All of themspecified LT-B-SpaA-fused proteins, which behaved identi-cally in every checked respect. Figure 3 presents the restric-tion map and the relevent nucleotide and amino acid se-quence for pYA3056, which was chosen for further analysis.We also fused the same 2.7-kb DNA fragment coding for

the SpaA protein into the 3' end of the LT-B gene fromwhich the signal sequence had been removed (pYA3082). Inthis experiment, only one immunologically positive clonecarrying stable recombinant plasmid DNA of molecularweight of -6.5 kb was isolated. Single and double restrictionenzyme analysis of the plasmid DNA from this clone,designated pYA3091, revealed that part of the spaA gene hadbeen deleted. The deletion, which is approximately 100 bplong, covers the EcoRI site and the first SstI site located nearthe junction point between the two genes (Fig. 3).pYA993 (56) is a pBR322 derivative into which a major

part of the S. sobrinus dextranase gene has been inserted.The S. sobrinus dextranase gene has been cloned (1) andsequenced in our laboratory (57). Studies are in progress tounderstand how dextranase activity is regulated and itsinvolvement in the caries-forming process (54). A 2.6-kbPvuII DNA fragment of pYA993 was purified from anagarose gel and inserted in frame into SalI-digested andblunt-ended (Klenow) pYA3049 or BamHI-digested andblunt-ended (Klenow) pYA3047 and pYA3048. The ligationmixtures were used to transform E. coli X6098 carryingpYA812. Transformants were selected for the Asd+ pheno-type and screened for expression of the enzymatically activedextranase by the blue dextran plate assay. We isolatedpYA3058 derived from pYA3048, pYA3073 derived frompYA3047, and pYA3088 derived from pYA3049. However, itshould be mentioned that this last plasmid, pYA3088, ex-pressed dextranase with weak enzymatic activity. Figure 4presents construction of the LT-B-Dex fusions, as well asthe nucleotide and deduced amino acid sequences at thejunction point between the two domains of the chimericproteins.Employing the same strategy, we also fused the dextra-

nase gene of pYA993 into the 3' end of the LT-B gene ofpYA3082 (LT-B without leader sequence). This constructwas designated pYA3083. Construction of all LT-B-Dexfusions was verified by restriction enzyme analysis.

Characterization of the LT-B fusion proteins. First, weinvestigated properties of the chimeric proteins expressed bythe Asd+ LT-B+ vectors containing the LT-B leader se-quence, which is necessary for the proteins to be transportedinto the periplasm and oligomerize, presumably into a pen-tamer. The LT-B-SpaA protein expressed by pYA3086,which has an 8-amino-acid linker with no proline residuesinserted between the last amino acid of LT-B and the firstone of the SpaA protein, displayed weak SpaA antigenicityand completely lost the ability to be recognized by anti-LT-B

INFEC-T. IMMUN.

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

S. SOBRINUS ANTIGEN FUSIONS TO LT-B 1009

A. Eco RIBgln sImo

Ptrc L ; EtcoRI

{ ~~~~~~~SstlpY~A 3056

*Bgr 6.70 Kb -SstIri

spaA,

PstI

B.LT-B-I Linker ^ I Spa AAAC TAC GCG CCG GTG GAT CCG ACG CGT GCA ATC GAA TTCAsn Tyr Ala Pro Val Asp Pro Thr Arg Ala Ile Gly Phe

FIG. 3. LT-B-SpaA gene fusion. (A) Restriction map of pYA3056; (B) junction point (nucleotide and specified amino acid sequence)between the two domains of the hybrid protein. The asterisk indicates an amino acid not originally encoded by either of the two fused genes.

antibodies, as determined by Western blot analysis (data notshown). Three other LT-B-SpaA fusions specified chimericproteins which contained a 10-amino-acid linker with twoproline residues inserted between the terminal asparagine ofLT-B and the first glycine of SpaA. Figure 5 shows theWestern blot analysis of the crude protein extract from E.coli X6097 containing pYA3056. Although we have fused thelarge SpaA (-120-kDa) protein with the small 13-kDa LT-B,the resulting hybrid protein still exhibits strong antigenicityand reacts with antibodies directed against both moieties.The chimeric protein specified by pYA3056 (Fig. SB, lanes 4and 5) migrates slightly slower than the SpaA protein en-coded by pYA2905 (Fig. SB, lane 6). LT-B-SpaA protein, aswell as the spaA gene product alone (27), is proteolyticallydegraded in E. coli cells. Degradation products can bedetected by anti-SpaA antibodies but are not recognized byanti-LT-B antibodies, which implies that degradation is morefrom the N-terminal than the C-terminal end.Cold osmotic shock followed by Western blot analysis of

the proteins located in different cell compartments revealedevidence of the periplasmic location of the LT-B-SpaAfusion protein specified by pYA3056 (data not shown).However, there was no evidence of oligomer formation (Fig.5, lane 5). One can conclude that the LT-B-SpaA oligomereither does not form or is extremely unstable. Even when welysed cells at 37°C in 0.5% SDS, LT-B-SpaA fusion proteinswere present as monomers. Blocking of the carboxyl end ofthe LT-B might be responsible for this effect.The LT-B-SpaA hybrid protein gene is transcribed from a

repressible trc promoter. The E. coli strain containingpYA3056 also possesses a pBR322 derivative that has thelaclq gene coding for lac repressor (pYA812). pYA812 has ahigher copy number per cell than the Asd+ vectors (26, 42).However, expression of the chimeric proteins is not com-pletely shut off, even in the absence of IPTG (Fig. 5, lane 3).The addition of IPTG results in a slower growth rate andeventual lysis of the cells.

Table 3 summarizes the properties of the hybrid LT-B-Dex proteins. One of them, encoded by pYA3088, in whichthe linker specified 5 amino acids with one proline, exhibits

A.Eco RISrfnaI|IHindM

Bg I EoREcoRIRI H mHinclPs[ Eco RV

3.98kb t 45kb I

B~~~~~gl

BamHI 2.6kb PvuUKlenow fragment

T4 DNA Ligase

Eco RI SnMoIB \I/ Hind m

Hin EcoRl

I pYA3O58 \.

~O EcoRI

8iaHnd!mHindi MlulPstl

PstI Apo L

B.

LTB* -I--Linker - I- dextranosepYA3058 AAC TAC GCG CCG GTG GAT CCT GGC ACT

Asn Tyr Ala Pro Val Asp Pro Glu Thr

LT-B+--a --Linker *-+ dextronosepYA3073 AAC TAC GCG CCG CAG GAT CCT GGC ACT

Asn Tyr Ala Pro Gly Asp Pro Glu Thr

LT-Ba -Linkere- --dextronosepYA 3088 AAC TAC GCG CCG TCG ACT GGC ACT

Asn Tyr Ala Pro Ser Thr Glu Thr

FIG. 4. LT-B-Dex gene fusion. (A) Construction of pYA3058;(B) junction points (nucleotide and specified amino acid sequences)between the two domains of the hybrid proteins.

VOL. 61, 1993

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

1010 JAGUSZTYN-KRYNICKA ET AL.

MW

kDa

200-92 -

69 -

46 -

30 -

9 -

A2 3 4 5

BMW

2 3 4 5 6 kDa

-200

92

69

46

- 30

-21

4 --14

FIG. 5. Immunoblot of protein extracts of E. coli X6097(pYA812) containing LT-B-SpaA fusion (pYA3056). (A) Westernblot reacted with anti-LT-B antibodies; (B) Western blot reactedwith anti-SpaA antibodies. Thirty micrograms of protein was loadedper lane. When needed (presence of lacIq gene) transcription fromtrc promoter was induced by 1 mM IPTG. Cells were lysed byboiling for 5 min, except those loaded in lanes 5, which were lysedat 60°C for 10 min to visualize oligomer formation. Lanes (plasmidsand E. coli strains, respectively): 1, pYA3048 (LT-B with leadersequence) in X6097; 2, pYA292 (cloning vector) in X6097; 3,pYA3056 (LT-B-SpaA with leader sequence) in X6097 (pYA812),uninduced; 4 and 5, pYA3056 (LT-B-SpaA with leader sequence) inX6097 (pYA812), induced; 6, pYA2905 (SpaA) in X6097.

weak enzymatic activity, and both its components lostantigenicity. Contrary to that, antigenic determinants en-

coded by dextranase and LT-B genes are recognized by theirrespective antibodies in the two other fusion proteins spec-ified by pYA3058 (Fig. 6) and pYA3073 (data not shown),which also display strong dextranase enzymatic activity.Both of them contain a linker of 6 amino acids with twoproline residues. The LT-B-Dex fusion protein specified bypYA3058 is transcribed from the trc repressible promoterbecause in the presence of Laclq specified by pYA812, itsexpression is dependent upon the addition of IPTG. Theaddition of IPTG also results in slowing the growth rate ofX6098 (pYA812 and pYA3058). SDS-PAGE also showed thatthe fusion protein encoded by pYA3058 (Fig. 6B, lane 2) wasslightly bigger than the one specified by pYA3010 (Fig. 6B,lanes 5 and 6) and exhibits properties characteristic of LT-Bitself in that it forms a high-molecular-weight (> > >200,000)complex (Fig. 6, lane 1). We did not fully characterize theseoligomeric forms, but their high molecular weight, theirinability to react with antibodies that only recognize mono-

meric LT-B, and their ability to bind to GM1 ganglioside (seebelow) lead us to believe that these oligomers are probablypentamers. The oligomeric form of the LT-B-Dex fusionreacted with both anti-LT-B (Fig. 6A, lane 1) and anti-

MW

kDo

200 -

92 -

69 -

A2 3 4

_ _

46-

".i ...

30 - _' :t:......

21 -

14-

kDa2 3

mn /. _

B4 5 6Y2UU

92-

4 6-

30- .

21-

14-

FIG. 6. Immunoblot of protein extracts of E. coli X6098(pYA812) containing LT-B-Dex fusions. (A) Western blot reactedwith anti-LT-B antibodies; (B) Western blot reacted with anti-dextranase antibodies. Thirty micrograms of protein was loaded toeach lane. Transcription from the trc promoter was induced by 1mM IPTG. Cells were lysed by boiling for S min, except thoseapplied in lanes 1 and 6, which were opened at 60°C for 10 min tovisualize oligomer formation. Lanes (plasmids and E. coli strains,respectively): 1 and 2, pYA3058 (LT-B-Dex with leader sequence) inX6098 (pYA812), induced; 3, pYA3058 (LT-B-Dex with leadersequence) in X6098 (pYA812), uninduced; 4, pYA292 (cloningvector) in X6097; 5 and 6, pYA3010 (Dex) in X6098.

dextranase (Fig. 6B, lane 1) antibodies, in contrast to theoligomeric form of LT-B specified by the vector alone,which did not react with the anti-LT-B serum (Fig. 2, lane2B). In recent experiments, we have used an anti-LT-Bserum raised against purified LT-B pentamer, and theseantibodies reacted with nonfused and fused LT-B monomersand oligomers. Heating of the LT-B-Dex fusion protein at60°C resulted in the presence of both the oligomeric andmonomeric forms of the hybrid. Most likely, the oligomericform of the fused protein is less heat stable than thepentameric form of LT-B alone. When cells were lysed at37°C, more than 80% of the LT-B-Dex hybrid proteinencoded by pYA3058 existed as an oligomer (data notshown). However, extensive degradation of the proteinmade analysis difficult. Since the LT-B-Dex fusion proteinoligomerized, it was of interest to determine its cellularlocation. Periplasmic and cytoplasmic protein extracts of E.coli X6098 (pYA3058) were prepared by cold osmotic shockand analyzed by SDS-blue dextran-PAGE. Dextranases en-coded by pYA993 and by pYA3010 were used as controls.The dextranase gene of pYA993, transcribed from the tetpromoter, is expressed as a fusion with the tet gene, and theprotein is exported into the periplasm (Fig. 7, lane 2).Expression of the dextranase gene of pYA3010 is driven bythe trc promoter of pYA292 and specified cytoplasmic dex-tranase (Fig. 7, lane 1). By comparing the intensity of the

TABLE 3. Properties of LT-B-Dex fusions

Lengthoflinker ~~~~~~~~~~~~~~~AntigenicityLT-B-Dex plasmid LT-B vector L(ength of linker Pentamerization Localization Enzymatic activity(amino3acids)8anti-LT-B anti-Dex

pYA3058 pYA3048 6 + Periplasmic + + +pYA3073 pYA3047 6 + Periplasmic + + +pYA3083 pYA3082 6 - Cytoplasmic + + +pYA3088 pYA3049 5 - Cytoplasmic ± - -

INFEcr. IMMUN.

I

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

S. SOBRINUS ANTIGEN FUSIONS TO LT-B 1011

1 2 3 4AE' BA4B

MW

kDc-80-116- 84- 56

- 48

- 36

26

FIG. 7. Localization of dextranase and of the LT-B-Dex fusionproteins. Periplasmic (A) and cytoplasmic (B) protein extracts were

applied on SDS-blue dextran-PAGE (20 p,g of protein per lane).After renaturation of proteins, dextranase enzyme activity causedhydrolysis of the blue dextran. Lanes (plasmids and E. coli strains,respectively): 1, pYA3010 (Dex) in X6098; 2, pYA993 (Dex) inX6098; 3, pYA3058 (LT-B-Dex with leader sequence) in X6098(pYA812), induced by 1 mM IPTG; 4, pYA3083 (LT-B-Dex withoutleader sequence) in x3987.

enzymatic reactions (hydrolysis of the blue dextran) exhib-ited by the periplasmic and cytoplasmic protein fractions, weconcluded that the LT-B-Dex fusion specified by pYA3058(Fig. 7, lane 3) is translocated into the periplasm. Coldosmotic shock caused release of some cytoplasmic protein,since in the case of pYA3010 dextranase is present in bothfractions (Fig. 7, lane 2).

Figure 8 presents data on the antigenic properties of theLT-B fusions which lack leader sequences. Both LT-B-SpaA(specified by pYA3091) and LT-B-Dex (specified bypYA3083) are expressed at high level in S. typhimuriumX3987 and retain antigenicity of both components. Thecellular location of these hybrid proteins has been checkedby cold osmotic shock followed by SDS-blue dextran-PAGE

MW n

kDo A B

200-

69-

46-

30-

2'A B3

efI MW

A B kDa A B

_OL- 200-

-92 -

69 -

- 46

30-~~~~~~~~3- 30 -

21-- 21 -

anti-LT-B anti- anti-dextranose Spa A

FIG. 8. Immunoblots with anti LT-B, anti-dextranase, and anti-SpaA reacted with protein extracts of S. typhimurium X3987 con-

taining pYA3083 (LT-B-Dex without leader sequence) (lane 1),pYA292 (cloning vector) (lane 2), and pYA3091 (LT-B-SpaA with-out leader sequence) (lane 3). (A) Cells opened by boiling 5 min; (B)cells opened by incubating at 60°C for 10 min.

for the LT-B-Dex fusion (Fig. 7, lane 4) and Western blotanalysis for the LT-B-SpaA chimera (data not shown). Bothare cytoplasmic.

It was of interest to determine if the LT-B-streptococcalantigen fusions still retained the biological activity of LT-B.GM1 ganglioside was bound to the surfaces of plastic micro-titer plates and reacted with crude protein extracts obtainedfrom E. coli X6097 or S. typhimunium X3987 cells containingdifferent LT-B fusions. LT-B expressed by pYA3048 (LT-Bwith leader sequence) with a 10-amino-acid extension on itscarboxyl end exhibited affinity to GM1 ganglioside identicalto that of LT-B specified by the original pEWD299 plasmid,while cytoplasmic LT-B specified by pYA3082, which doesnot oligomerize, also lost the ability to efficiently bind toGM1 gangliosides. Interestingly, both fusions containingleader sequences display the same level of affinity to GM1even though they differ from each other in their abilities tooligomerize and in the stabilities of the oligomer forms.LT-B-streptococcal antigen fusions which do not containsignal sequences and stay cytoplasmic, do, however, retainlow levels of GM1 ganglioside affinity.

DISCUSSION

Recently, CT-B and LT-B have been of great interest asantigen carriers for the development of oral vaccines. Therehave been a number of distinct approaches to evaluate andtake advantage of the immunity-enhancing action of bothtoxins. A number of genetically or chemically created fu-sions have been investigated, mainly with the small nonim-munogenic ST. In this report, we describe Asd+ LT-B'vectors which allow transcriptional and translational fusionsof foreign gene epitopes in every reading frame into LT-Bwith or without leader sequences. We chose two immuno-genic S. sobrinus antigens, SpaA (18) and dextranase (57), tobe fused to the C-terminal end of LT-B. They are encoded bya 2.7-kb EcoRI-SmaI DNA fragment of pYA2905 specifyingSpaA and a 2.6-kb PuvII DNA fragment of pYA993 speci-fying Dex, respectively.

In order to stimulate an immune response to an antigenfused to LT-B, the hybrid protein should retain the nativeproperties of the two components, including the ability tobind antibodies. Genetic fusions of >100-kDa streptococcalantigens into the 13-kDa LT-B resulted in chimeric proteinswhich are recognized by anti-serum against both LT-B andthe streptococcal antigens. Linkers of different nucleotidesequences introduced into the carboxyl end of the LT-Bgene permitted us to study the influence of the length andamino acid composition of the intervening linker on theantigenicity of the hybrid protein. It has been noted before(48) that even minor modification of the amino acid compo-sition of the CT-B-ST fusion affects its reaction with anti-bodies. Clements (6) showed that a lack of the linkerbetween the two domains of the LT-B-ST fusion completelyabolished antigenicity, which could be restored by introduc-ing a 7-amino-acid proline-containing linker. In this study,maximum antigenicity for both components of LT-B fusionshas been obtained for linkers of at least 6 amino acids withtwo proline residues. Antigenic determinants encoded by theLT-B component of the fusion protein specified by pYA3086(with an 8-amino-acid-long linker containing no proline res-idues) are not presented in the proper way to be able tocombine with antibodies against LT-B; moreover, confor-mational changes of the protein also diminished the immu-noreactivity of the SpaA epitopes. Three other LT-B-SpaAfusions specified by pYA3055, pYA3056, and pYA3057 (with

VOL. 61, 1993

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

1012 JAGUSZTYN-KRYNICKA ET AL.

-K--292 Asd

*- EWD299 LT-B with LS

1 : -I11:4 1:8 1:1 6 1:32 1:64 1:128

l

1:256 1:512 1:1024 1:2048

Dilution of cell lysate

FIG. 9. Abilities of LT-B fused proteins to bind to GM1 ganglioside measured by ELISA. Microtiter plates were precoated with 1.5 p,g ofGM1 ganglioside and then reacted with crude protein extracts obtained from bacterial cells expressing different LT-B fusions. When proteinsor primary antibodies were omitted in the assay, the absorbance reading was 0. For additional details, see Materials and Methods. LS, leadersequence.

10-amino-acid-long linkers containing two proline residues)retained strong antigenicity to both anti-LT-B and anti-SpaAantibodies. On the basis of the properties of the constructedLT-B-SpaA protein fusions, one can conclude that substitu-tion of a slightly hydrophilic glycine by highly hydrophobicvaline (compare MCS of pYA3047 and pYA3048 [Fig. 1]) or

inclusion of a charged amino acid (asparagine encoded bythe MCS of pYA3047 and pYA3048 but not by pYA3049)does not appreciably influence the conformation of the fusedprotein.Analogous findings were made for the LT-B-Dex fusions.

The LT-B-Dex fusion protein specified by pYA3088 (5-amino-acid-long linker with one proline residue) not only lostthe ability to react with antibodies directed against eitherdomain, but its biological activity had also been diminished.But when S. sobrinus dextranase was fused into the carboxylend of LT-B with the help of 6-amino-acid-long linkerscontaining two prolines (pYA3058 and pYA3073), the result-ing chimeras were well recognized by anti-LT-B and anti-Dex antibodies. Moreover, the hybrid proteins exhibitedstrong enzymatic activity.The immunity-enhancing ability of CT-B and LT-B in

stimulating mucosal immune responses could be due to thepresence of T- and B-cell epitopes, the conformation andsize of oligomeric forms, the ability to interact with GM1gangliosides to facilitate endocytosis, and the potential toblock induction of tolerance. Most likely, for an LT-B fusionto be effective as an oral vaccine, it is important to retainabilities to pentamerize and bind to GM1 gangliosides. Allpreviously reported N-terminal fusions (16, 48) can pentam-erize, but fusions blocking the carboxyl terminus of the toxinoften resulted in diminishing pentamerization. On the basis

of crystallographic studies of the LT toxin, Sixma et al. (53)concluded that LT-B elongated at the COOH end cannotform the AB5 structure. Using the appropriate linker, we

constructed two LT-B-Dex fusions of monomeric molecularmass of approximately 120 kDa, which were translocatedinto the periplasm where they oligomerized. Furthermore,ganglioside-binding properties of the toxin have been shownto be more affected by C- than by N-terminal fusions (5, 6,17). GM1 binding experiments showed that all constructedLT-B fusions retained biological activity of LT-B in that theybound to GM1 gangliosides, but incorporation of the foreignantigen into the carboxyl terminus of LT-B influences theiraffinity. Interestingly, two fusions (pYA3091 and pYA3083[Fig. 9]) without leader sequences still retained a low-levelGM1 ganglioside affinity. It should be pointed out that theLT-B protein encoded by pYA3082 (LT-B without leadersequence) did not, however, react by the same assay withGM1. One can conclude that incorporation of an approxi-mately 100-kDa protein into the carboxyl end of LT-B can

allow the GM1 binding domain to be presented in the properway. Experiments to purify large amounts of the LT-Bfusion proteins, which will allow us to investigate GM1binding properties in more defined ways, are now inprogress.

It was a common belief (17) that to minimize any structuralchanges of the LT-B, the size of the protein fused into itshould be kept to a minimum. In this study, we proved thata properly designed linker providing separation of domainspermitted retention of antigenicity and biological activities ofboth components of the hybrid protein, even when theforeign antigen is 10 times larger than LT-B alone.

Since S. typhimurium establishes an infection by first

0.60

0.50 -

0.40 -

E

o 0.30 -a)0cJU 0.20 -0co

< 0.10 -

0.0

-0.1 0 -t

INFECT. IMMUN.

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

S. SOBRINUS ANTIGEN FUSIONS TO LT-B 1013

colonizing the gut-associated lymphoid tissue, many labora-tories are using avirulent Salmonella strains to deliver het-erologous antigens and stimulate a mucosal immune re-sponse (12). In order to successfully use recombinantavirulent Salmonella strains for oral immunization, one mustconsider the problems of plasmid stability and toxicity of theexpressed protein. To eliminate the problem of plasmidinstability, Cardenas and Clements (5) integrated the E. coliLT-B gene into the chromosome of a galE mutant of S.typhimuium. However, even though the chromosomalcointegrate stably maintained and expressed the LT-B gene,when used for in vivo oral immunization, the construct didnot elicit a significant anti-LT-B immune response. AllLT-B+ vectors constructed in this study contain the S.typhimunum asd+ gene to complement an asd deletionpresent on the bacterial host chromosome. This guaranteesthe maintenance of a population of cells maintaining theengineered plasmid construct, since loss of the Asd+ recom-binant vectors results in DAP-less death and cell lysis.LT-B-SpaA and LT-B-Dex protein fusions were tran-

scribed from the trc inducible promoter present in the Asd+vectors. Expression of periplasmic hybrid proteins depen-dent upon the addition of IPTG due to the presence of thelacIq gene resulted in slower growth rates of E. coli X6097and X6098 and in some cases resulted in cell lysis 7 to 8 hlater. These effects were much more evident at 37 than at30°C. The same effect has also been noticed by others (17).On the other hand, N-terminal as well as C-terminal CT-B-ST fusions described by Sanchez et al. (48), when intro-duced into V. cholerae strains, were secreted into themedium and were not toxic, even though they were ex-pressed from the strong constitutive tacP promoter. Ourpreliminary results indicate that recloning of LT-B genefusions into an Asd+ vector with the pSC101 plasmid repli-con in place of the piSA replicon lowers the amount of theexpressed protein to a level which is not cytotoxic. We hopethat under these circumstances the amount of the expressedantigen will be high enough to stimulate an immune re-sponse. If not, we can use the Asd+ balanced lethal host-vector system construct described in this report to selectmore rapidly growing variants which should be due tomutations that reduce the toxicity of fused antigen synthesis.This selection can be done either in vitro or in vivo.Recovered isolates would have to be screened for high-levelexpression of a fusion protein with the antigenic, biological,and/or enzymatic activities displayed by the parental strainwith the original construct. DNA sequencing would then becarried out to identify the specific mutational alterations thateliminated toxicity associated with high-level expression.This might then permit even higher levels of antigen synthe-sis by changing the piSA plasmid replicon to a higher-copy-number replicon.Two of our constructs, pYA3083 and pYA3091, specifying

cytoplasmic LT-B-Dex and LT-B-SpaA, respectively, havebeen introduced into the avirulent oral S. typhimurium AcrpAasd Acya vaccine strain X3987 (19, 20). We are now usingthese constructs to evaluate the immunogenicity of the LT-Bfusions as well as to compare the types of immune responsesinduced by S. sobnnus antigens alone with those induced bythe same antigen as an LT-B fusion protein. Moreover, wehave also constructed LT-B-Dex (34) and LT-B-SpaAoperon fusions (unpublished data). Studies are in progress todetermine the immunological properties of the S. sobrinusantigens as operon or gene fusions with LT-B. This willallow us to better understand the mechanisms by whichLT-B exerts its immunity-enhancing effect.

ACKNOWLEDGMENTSWe thank Randall Holmes for supplying anti-LT-B antibodies and

Teresa Doggett for helpful discussions. We also thank M. Shafer forhelp in preparation of the manuscript.The research was supported by grants A126186 (to J.E.C.-C.),

AI23470 (to J.E.C.-C.), DE06673 (to R.C.) and DE06669 (to R.C.)from the United States Public Health Service.

REFERENCES1. Barrett, J. F., T. A. Barrett, and R. Curtiss III. 1987. Purifica-

tion and partial characterization of the multicomponent dextra-nase complex of Streptococcus sobrinus and cloning of thedextranase gene. Infect. Immun. 55:792-802.

2. Barrett, J. F., and R. Curtiss III. 1986. Renaturation of dextra-nase activity from culture supernatant fluids of Streptococcussobrinus after sodium dodecylsulfate polyacrylamide gel elec-trophoresis. Anal. Biochem. 158:365-370.

3. Birnboim, H. C. 1983. A rapid alkaline extraction method for theisolation of plasmid DNA. Methods Enzymol. 100:243-255.

4. Bromander, A. J., J. Holmgren, and N. Lycke. 1991. Choleratoxin stimulates IL-1 production and enhances antigen presen-tation by macropyhages in vitro. J. Immunol. 146:2908-2914.

5. Cardenas, L., and J. D. Clements. 1991. Effect of gene locationon in vitro and in vivo stability, immunogenicity and expressionof foreign antigenes in bacterial vaccine vectors, abstr. B148, p.50. Abstr. 91st Gen. Meet. Am. Soc. Microbiol. 1991. AmericanSociety for Microbiology, Washington, D.C.

6. Clements, J. D. 1990. Construction of a nontoxic fusion peptidefor immunization against Escherichia coli strains that produceheat-labile and heat-stable enterotoxins. Infect. Immun. 58:1159-1166.

7. Clements, J. D., and L. Cardenas. 1990. Vaccines againstenterotoxigenic bacterial pathogens based on hybrid Salmonellathat express heterologous antigens. Res. Microbiol. 141:981-993.

8. Clements, J. D., and S. El-Morshidy. 1984. Construction of apotential oral bivalent vaccine for typhoid fever and choleraEscherichia coli-related diarrheas. Infect. Immun. 46:564-569.

9. Clements, J. D., and R. Finkelstein. 1978. Immunological cross-reactivity between a heat-labile enterotoxin of Eschenichia coliand subunits of Vibrio cholerae enterotoxin. Infect. Immun.21:1036-1039.

10. Clements, J. D., and R. Finkelstein. 1978. Demonstration ofshared and unique immunological determinants in enterotoxinfrom Vbrio cholerae and Escherichia coli. Infect. Immun.22:709-713.

11. Clements, J. D., F. L. Lyson, K. L. Lowe, A. L. Farrand, and S.El-Morshidy. 1986. Oral immunization of mice with attenuatedSalmonella enteritidis containing a recombinant plasmid whichcodes for production of the B subunit of heat-labile Escherichiacoli enterotoxin. Infect. Immun. 53:685-692.

12. Curtiss, R., III. 1990. Attenuated Salmonella strains as livevectors for the expression of foreign antigens, p. 161-188. InG. C. Woodrow and M. M. Levine (ed.), New generationvaccines. Marcel Dekker, Inc., New York.

13. Czerkinsky, C., M. W. Russell, N. Lycke, M. Lindblad, and J.Holmgren. 1989. Oral administration of a streptococcal antigencoupled to cholera toxin B subunit evokes strong antibodyresponses in salivary glands and extramucosal tissues. Infect.Immun. 57:1072-1077.

14. Dallas, W. S., D. M. Gill, and S. Falkow. 1979. Cistronsencoding Escherichia coli heat-labile toxin. J. Bacteriol. 139:850-858.

15. Dertzbaugh, M. T., and C. 0. Elson. 1991. Cholera toxin as amucosal adjuvant, p. 119-131. In D. R. Spriggs and W. C. Kolf(ed.), Topics in vaccine adjuvant research. CRC Press, Inc.,Boca Raton, Fla.

16. Dertzbaugh, M. T., and F. L. Macrina. 1989. Plasmid vectors forconstructing translational fusion to the B subunit of choleratoxin. Gene 82:335-343.

17. Dertzbaugh, M. T., D. L. Peterson, and F. L. Macrina. 1990.Cholera toxin B-subunit gene fusion: structural and functionalanalysis of the chimeric protein. Infect. Immun. 58:70-79.

VOL. 61, 1993

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

1014 JAGUSZTYN-KRYNICKA ET AL.

18. Doggett, T., and E. K. Jagusztyn-Krynicka. 1990. Humoral andcell-mediated immune responses to the Streptococcus sobrinussurface protein antigen A (SpaA) expressed in a recombinantavirulent Salmonella vaccine strain, abstr. E85, p. 133. Abstr.90th Annu. Meet. Am. Soc. Microbiol. 1990. American Societyfor Microbiology, Washington, D.C.

19. Doggett, T., E. K. Jagusztyn-Krynicka, and R. Curtiss III. 1991.In vitro responses of the Streptococcus sobrinus surface proteinantigen A (SpaA) expressed by an oral Salmonella typhimuniumvaccine strain (Acya Acrp Aasd), abstr. 5752, p. A1362. FASEBJ. 75th Annu. Meet. 1991.

20. Doggett, T., E. K. Jagusztyn-Krynicka, S. M. Kelly, and R.Curtiss III. 1992. In vitro responses to LT-B and LT-B/dex-tranese fusion proteins expressed by a Salmonella typhimuriumvaccine strain, abstr. E85, p. 158, Abstr. 92nd Gen. Meet. Am.Soc. Microbiol. 1992. American Society for Microbiology,Washington, D.C.

21. Elson, C. 0. 1988. Cholera toxin and its subunits as potentialoral adjuvants. Curr. Top. Microbiol. Immunol. 146:29-34.

22. Finkelstein, R. A. 1988. Cholera, the cholera enterotoxins andthe cholera enterotoxin-related enterotoxin family, p. 85-102. InP. Owen and T. J. Foster (ed.), Immunochemical and moleculargenetic analysis of bacterial pathogens. Elsevier Science Pub-lishing, Inc., New York.

23. Finkelstein, R. A., M. F. Burks, A. Zupan, W. S. Dallas, C. 0.Jacob, and D. S. Ludwig. 1987. Epitopes of the cholera family ofenterotoxins. Rev. Infect. Dis. 9:544-561.

24. Francis, M. L., J. Moss, T. A. Fitz, and J. J. Mond. 1990.cAMP-independent effects of cholera toxin on B cell activation.A possible role for cell surface ganglioside GM1 in B cellactivation. J. Immunol. 145:3162-3169.

25. Fukuta, S., J. L. Magnani, E. M. Twiddy, R. K. Holmes, and V.Ginsburg. 1988. Comparison of the carbohydrate-binding spec-ificities of cholera toxin and Escherichia coli heat-labile entero-toxins LTh-I, LTh-IIa, and LTh-IIb. Infect. Immun. 56:1748-1753.

26. Galan, J. E., K. Nakayama, and R. Curtiss III. 1990. Cloningand characterization of the. asd gene of Salmonella typhimu-rium: use in stable maintenance of recombinant plasmids inSalmonella vaccine strains. Gene 94:29-35.

27. Goldschmidt, R. M., M. Thoren-Gordon, and R. Curtiss III.1990. Regions of the Streptococcus sobrinus spaA gene en-coding major determinants of antigen I. J. Bacteriol. 172:3988-4001.

28. Hardy, S. J., J. Holmgren, S. Johansson, J. Sanchez, and T. R.Hirst. 1988. Coordinated assembly of multisubunit proteins:oligomerization of bacterial enterotoxins in vivo and in vitro.Proc. Natl. Acad. Sci. USA 85:7109-7113.

29. Hazelbauer, G. L., and S. Harayama. 1979. Mutants in trans-mission of chemotactic signals from two independent receptorsof E. coli. Cell 16:617-625.

30. Hirst, T. R., S. J. S. Hardy, and L. L. Randall. 1983. Assemblyin vivo of enterotoxin from Escherichia coli: formation of the Bsubunit oligomer. J. Bacteriol. 153:21-26.

31. Houghten, R. A., R. F. Engert, J. M. Ostresh, S. R. Hoffman,and F. A. Klipstein. 1985. A completely synthetic toxoid vaccinecontaining Escherichia coli heat-stable toxin and antigenic de-terminants of the heat-labile toxin B subunit. Infect. Immun.48:735-740.

32. Jacob, C. O., R. Arnon, and R. A. Finkelstein. 1986. Immunityto heat-labile enterotoxins of porcine and human Escherichiacoli strains achieved with synthetic cholera toxin peptides.Infect. Immun. 52:562-567.

33. Jacobs, W. A., J. F. Barrett, J. E. Clark-Curtiss, and R. CurtissIII. 1986. In vivo repackaging of recombinant cosmid moleculesfor analyses of Salmonella typhimurium, Streptococcus mu-tans, and mycobacterial genomic libraries. Infect. Immun. 52:101-109.

34. Jagusztyn-Krynicka, E. K., and J. E. Clark-Curtiss. 1990. Con-struction of an Escherichia coli LT-B vector for expression offoreign antigenic determinant genes, abstr. E6, p. 120. Abstr.90th Annu. Meet. Am. Soc. Microbiol. 1990. American Societyfor Microbiology, Washington, D.C.

35. Klipstein, F. A. 1990. Synthetic toxoid vaccines to preventdiarrhea due to enterotoxigenic Escherichia coli, p. 661-665. InG. C. Woodrow and M. M. Levine (ed.), New generationvaccines. Marcel Dekker, Inc., New York.

36. Klipstein, F. A., R. F. Engert, J. D. Clements, and R. A.Houghten. 1983. Vaccine for enterotoxigenic Eschenichia colibased on synthetic heat-stable toxin crossed-linked to the Bsubunit of heat-labile toxin. J. Infect. Dis. 147:318-326.

37. Kyte, J., and R. F. Doolittle. 1982. A simple method fordisplaying the hydropathic character of a protein. J. Mol. Biol.80:575-599.

38. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

39. Lennox, E. S. 1955. Transduction of linked genetic characters ofthe host by bacteriophage P1. Virology 1:190-206.

40. Lipscombe, N., I. G. Charles, M. Roberts, G. Dougan, J. Tite,and N. F. Fairweather. 1991. Intranasal immunization using theB subunit of Escherichia coli heat-labile toxin fused to anepitope of the Bordetella pertussis P. 69 antigen. Mol. Micro-biol. 5:1385-1392.

41. McKenzie, S. J., and J. F. Halsey. 1984. Cholera toxin B subunitas a carrier protein to stimulate a mucosal immune response. J.Immunol. 133:1818-1824.

42. Nakayama, K., S. M. Kelly, and R. Curtiss III. 1988. Construc-tion of an Asd+ expression-cloning vector: stable maintenanceand high level expression of cloned genes in a Salmonellavaccine strain. Biotechnology 6:693-697.

43. Pickett, C. L., E. M. Twiddy, C. Coker, and R. K. Holmes. 1989.Cloning, nucleotide sequence and hybridization studies of thetype Ila heat-labile enterotoxin gene of Escherichia coli. J.Bacteriol. 171:4945-4952.

44. Pickett, C. L., D. L. Weinstein, and R. K. Holmes. 1987.Genetics of type Ila heat-labile enterotoxin of Escherichia coli:operon fusions, nucleotide sequence, and hybridization studies.J. Bacteriol. 169:5180-5187.

45. Pierce, N. F., W. C. Crag, and P. F. Engel. 1980. Antitoxicimmunity to cholera in dogs immunized orally with choleratoxin. Infect. Immun. 27:632-637.

46. Sambrook, J., E. F. Fritsch, and J. Maniatis. 1989. Molecularcloning: a laboratory manual. Cold Spring Harbor LaboratoryPress, Cold Spring Harbor, N.Y.

47. Sanchez, J., and J. Holmgren. 1989. Recombinant system foroverexpression of cholera toxin B-subunit in Vibiio cholerae asa basis for vaccine development. Proc. Natl. Acad. Sci. USA86:481-485.

48. Sanchez, J., S. Johansson, B. L6wenadler, A. M. Svennerholm,and J. Holmgren. 1990. Recombinant cholera toxin B-subunitand gene fusion proteins for oral vaccination. Res. Microbiol.141:971-979.

49. Sanchez, J., A. M. Svennerholm, and J. Holmgren. 1988. Ge-netic fusion of a non-toxic heat-stable enterotoxin-related deca-peptide antigen to cholera toxin B-subunit. FEBS Lett. 241:110-114.

50. Sandkvist, M., T. R. Hirst, and M. Bagdasarian. 1987. Alter-ations at the carboxyl terminus change assembly and secretionproperties of the B subunit of Eschenchia coli heat-labileenterotoxin. J. Bacteriol. 169:4570-4576.

51. Schodel, F., and H. Will. 1989. Construction of a plasmid forexpression of foreign epitopes as fusion proteins with subunit Bof Escherichia coli heat-labile enterotoxin. Infect. Immun. 37:1347-1350.

52. Schodel, F., and H. Will. 1990. Expression of hepatitis B virusantigens in attenuated Salmonellae for oral immunization. Res.Microbiol. 141:831-837.

53. Sixma, T. K., S. E. Pronk, K. U. Kalk, E. S. Wartna, B. A. M.vanZaten, B. Witholt, and W. G. J. Hol. 1991. Crystal structureof a cholera toxin-related heat-labile enterotoxin from E. coli.Nature (London) 351:371-377.

54. Sun, J. W., and S. Y. Wanda. 1991. Characterization of Strep-tococcus sobrinus UAB108 dextranase inhibitor (Dei) in E. coli,D-94, p. 94. Abstr. 91st Gen. Meet. Am. Soc. Microbiol. 1991.American Society for Microbiology, Washington, D.C.

INFECT. IMMUN.

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from

S. SOBRINUS ANTIGEN FUSIONS TO LT-B 1015

55. Tsai, C. M., and C. E. Frasch. 1983. A sensitive stain fordetecting lipopolysaccharides in polyacrylamide gels. Anal.Biochem. 119:115-119.

56. Wanda, S. Y. 1990. Purification and characterization of Strep-tococcus sobrinus dextranase in recombinant Eschenchia coliand subcloning of the dextranase gene, abstr. D69, p. 92. Abstr.90th Annu. Meet. Am. Soc. Microbiol. 1990. American Society

for Microbiology, Washington, D.C.57. Wanda, S.-Y., and R. Curtiss III. Purification and characteriza-

tion of Streptococcus sobrinus dextranase produced in recom-binant Eschenchia coli and sequence analysis of the dextranasegene. Submitted for publication.

58. Zinder, N. 1958. Lysogenization and superinfection immunity inSalmonella. Virology 5:291-326.

VOL. 61, 1993

on Novem

ber 12, 2018 by guesthttp://iai.asm

.org/D

ownloaded from