Influence of strain viability and antigen dose on the use of attenuated mutants of Salmonella as vaccine carriers

Lucia Cfirdenas*, Ujjala Dasgupta* and John D. Clements ~

It is now accepted that oral killed typhoid vaccines are not effective at inducing protective anti-typhoid immunity. It is not known whether oral killed Salmonella can function as an effective carrier of other antigens to the immune system. In order to test this hypothesis, we immunized groups of mice with viable and non-viable preparations ofa roA Salmonella dublin strain EL23 which codes for production of the binding subunit of the heat-labile enterotoxin of Escherichia coli (LT-B). Animals immunized orally with viable EL23 developed serum and mucosal anti-LT-B responses consistent with our previous findings. Significantly, mice immunized orally with ultraviolet-killed EL23 developed serum and mucosal antibody responses equivalent to those which developed in animals orally immunized with the same number of viable EL23. We extended these observations to include a number of methods of killing the organisms which may also preserve the ability of these strains to function as carriers. Our findings indicate that viability is not a requirement for use of a Salmonella strain as an immunological carrier. Moreover, our evidence indicates that bacteraemia and persistence in tissues are not necessary for oral priming, and therefore it may be best to dissociate the question of what makes the best live oral anti-typhoid vaccine from the question of what makes a good carrier of heterologous antigens.

Keywords: Salmonella spp.; oral vaccines; heterologous antigens

A number of attenuated mutants of Salmonella are able to interact with the lymphoid tissues in the Peyer's patches, but are not able to cause systemic disease. Some of these mutants are effective as live vaccines (i.e. able to protect against infection by the virulent Salmonella parent) and are candidates for use as immunological carriers of virulence determinants from other organisms. Different mutants have been employed for this purpose, including galE mutants which lack the enzyme uridine diphosphate-galactose-4-epimerase, aro mutants which have specific non-reverting deletions in the common aromatic bio-synthetic pathway, put mutants which have deletions preventing the de novo synthesis of purines, and mutants carrying deletions in the cya gene (adenylate cyclase), the crp gene (cyclic AMP receptor protein), and in both the cya and crp genes (positive regulators of

Department of Microbiology and Immunology, Tulane University School of Medicine, 1430 Tulane Avenue, New Orleans, LA 70112, USA. *Current address: The Upjohn Company, Molecular Biology Research 7242-209-712, 301 Henrietta Street, Kalamazoo, MI 49001, USA. 1Current address: Biophysics Division, Indian Institute of Chemical Biology, Raja S.C. Mullick Road, Jadavpur, Calcutta 700 032, India. tTo whom correspondence should be addressed. (Received 8 September 1993; revised 29 October 1993; accepted 29 October 1993)

transcription for multiple genes that possess catabolite- sensitive promoters) t 5.

The use of attenuated Salmonella strains as vaccine delivery vehicles for foreign antigens has been studied extensively in a number of animal species and in humans. These mutants are able to establish a limited infection in the host and deliver a series of in vitro or in vivo synthesized antigens directly to the B and T lymphocytes present in the gut-associated lymphoid tissues (GALT). The primed B cells migrate to the mesenteric lymph nodes where they undergo differentiation, enter the general circulation via the thoracic duct, and subsequently populate the gut and other mucosal tissues where clonal expansion and terminal differentiation into plasma cells occur. Immunization with these multivalent vaccines is an effective way to elicit the production of serum and mucosal antibodies as well as cell-mediated immunity against both the Salmonella carrier and the foreign antigenS-~ s.

One potential problem concerning the use of these vaccine vectors in humans is that most of the candidate hybrid strains carry the foreign DNA on multicopy recombinant plasmids. Recombinant plasmids generally carry antibiotic resistance markers to facilitate identification and to maintain selective pressure in favour of plasmid retention. Several authors have reported the instability of such plasmids both in vivo and in vitro 8"13"17A9-23.

0264-410X/94/09/0833-08 © 1994 Butterworth-Heinemann Ltd Vaccine 1994 Volume 12 Number 9 833

Viability and antigen dose in Salmonel la vaccine carriers: L. Cardenas et al.

Preliminary studies in our laboratory have indicated that, within 24h of immunization, >99% of organisms recovered from the tissues of animals which had been inoculated with a plasmid-carrying strain had lost the plasmid and consequently the ability to synthesize the foreign antigen.

One means of overcoming the problem of spontaneous plasmid loss is to introduce the cloned gene into the chromosome of the carrier ~3"24 26. We investigated this approach by integrating the gene that codes for the binding subunit of the heat-labile enterotoxin of Escher- ichia coil (LT-B) into the chromosome of Salmonella typhimurium using the system of Hone et al. 24. We then conducted studies of in vitro and in Hvo stability and expression for that co-integrate and an isogenic Salmonella strain carrying the gene for expression of LT-B on a plasmid vector. Levels of expression of the foreign antigen were much lower when the antigen was expressed from the chromosome than when expressed from the plasmid. Groups of Balb/c mice were orally immunized with these strains, and, by 24 h postinoculation, the majority of tissue isolates from mice inoculated with the plasmid- containing strain had lost the plasmid and the ability to synthesize the antigen in subculture. In contrast, 100% of the co-integrate isolates retained the ability to express the antigen throughout the 21 days of the experiment 10.2o.

One possible advantage of stabilizing foreign antigen expression by chromosomal integration is that the foreign antigen would be expressed in situ over a longer period of time, thereby providing continual antigenic stimulation. The decrease in the amount of antigen presented as a result of decreased gene copy number between a multicopy plasmid and a single-copy chromosomal insert would theoretically be offset by the prolonged expression of antigen. However, in our studies, humoral and mucosal antibody levels against the antigen were greater when the antigen was expressed from the plasmid stabilized by growing the inoculum in the presence of antibiotic than when the antigen was expressed from the chromosome. Antigen-specific antibody levels correlated with the amount of antigen presented to the GALT.

These observations are important, and suggest that the critical event determining the development of an immune response against a foreign antigen delivered by these vectors may be the initial amount of antigen that primes the GALT and not the persistence of the vector in the tissues. Persistence of the vector in host tissues is the second problem concerning the use of these vaccines in humans and is at the heart of the fundamental objection to the use of Salmonella as a vaccine vector. Both of these issues are explored more fully in the current study.

M A T E R I A L S A N D M E T H O D S

Bacterial strains and plasmids

Salmonella dublin EL23 is a derivative of S. dublin SL1438, a non-reverting aromatic-dependent histidine- requiring mutant l° transformed with a 3.5 kb plasmid (pJC217) which carries a gene that codes for production of the B subunit of the heat-labile toxin (LT-B) of enterotoxigenic E. coli. S. typhi SE12 is a derivative of Ty21a 2, transformed with the LT-B plasmid pJC217 and shown to produce LT-B which is immunologically and physicochemically indistinguishable from that produced by EL23 or by E. coli harbouring the same plasmid 9. S. typhi 543Ty aroA purA and 659Ty aroA aroD were

derived from S. typhi CDC 10-80 by transposon- generated deletion mutagenesis and carry precisely defined mutations in the target genes 2;. Sahnonella strains SL1438, 543Ty and 659Ty were kindly provided by B.A.D. Stocker, Stanford University School of Medicine. S. typhimurium J706 yale HI AhisOG 24 was kindly provided by David Hone (University of Maryland School of Medicine). Strain J1000 is a derivative of J706 carrying a single copy of the LT-B gene utilizing the lae promoter. In this construct, the LT-B gene was integrated into the chromosome by homologous recombination as previously described 8"2°. S. typhimurium J706 was transformed by electroporation with plasmid pDF87, a pBR322-based LT-B plasmid utilizing the pBR322 tet promoter 2s, thereby producing strain J706(pDF87). Similarly, for construction of S. typhimurium J706(pJC217), strain J706 was transformed by electroporation with a pUC-based LT-B plasmid, pJC217, which utilizes the lac promoter 9.

Media and growth conditions

Trypticase soy broth (TSB) and trypticase soy agar (TSA) were obtained from BBL Microbiology Systems. Where indicated, ampicillin was added at 100/~g ml-1 and tetracycline was added at 15 #g ml 1. Aromatic- dependent mutants were cultured in media containing 1/~g 2,3-dihydroxybenzoic acid per ml (Sigma). Purine auxotrophs were cultured in media supplemented with 1 l~g of adenine and 1/~g of hypoxanthine per ml (Sigma). All strains were grown under standard conditions.

Isolation of plasmid DNA and restriction endonuclease digestion

Plasmid DNA was isolated by standard techniques 29. Restriction enzymes, ligase and reagents were obtained from Gibco BRL and used according to the manufacturer's directions.

Transformation

Transformation was carried out by electroporation using a BRL cell porator and voltage booster as described by the manufacturer.

ELISA

Reagents and antisera for the enzyme-linked immuno- sorbent assay (ELISA) were obtained from Sigma Chemical Co. Samples for ELISA were serially diluted in phosphate-buffered saline (PBS; pH 7.2) containing 0.05% Tween 20 (PBS-Tween). For determination of LT-B antigen, microtitre plates were precoated with 1.5 /~g per well of mixed gangliosides (type [II) in carbonate bicarbonate coating buffer (pH 9.6), then incubated with samples serially diluted in PBS Tween. Reactions were developed further using affinity-purified monospecific goat hyperimmune antiserum to LT-B 3°, together with rabbit anti-goat IgG conjugated to alkaline phosphatase. Bacterial samples were suspended in PBS containing 1 mg lysozyme per ml and lysed by successive freeze and thaw cycles. Expression was quantified using purified LT-B as a standard and is expressed in terms of specific activity as ng LT-B (based on ELISA) per mg total protein.

For determination of murine antibodies directed against LT-B resulting from the immunizations, microtitre plates were precoated with 1.5 /~g per well of mixed

834 Vaccine 1994 Volume 12 Number 9

Viability and antigen dose in Salmonella vaccine carriers: L. C&rdenas et al.

gangliosides (type III), then with 1 #g per well of purified LT-B. Serum and mucosal samples were obtained as previously described 1°. Briefly, the small intestine from duodenum to ileal-caecal junction was excised and homogenized in a solution containing 50 mM ethylene- diamine tetracetic acid (EDTA) and 0.1 mg m1-1 of soybean trypsin inhibitor. Samples were homogenized with a Tekmar Tissuemizer, clarified by centrifugation, lyophilized, resuspended in 2 ml of TEAN buffer (0.05 M Tris, 0.001 M EDTA, 0.003 M NaN 3, 0.2 M NaC1, pH 7.5), dialysed against TEAN buffer, and stored at - 20°C until assayed. Serum was collected by cardiac puncture, prepared by centrifugation through serum separators (Becton Dickinson), and stored at - 2 0 ° C until assayed. Samples from immunized mice or control samples were serially diluted in PBS-Tween and added to the microtitre wells. The presence of serum anti-LT-B IgG was determined using rabbit antiserum against mouse IgG conjugated to alkaline phosphatase. Mucosal anti-LT-B IgA was assayed using goat antiserum against mouse IgA (alpha-chain-specific) followed by rabbit antiserum against goat IgG conjugated to alkaline phosphatase. Values for IgG and IgA were determined from a standard curve with purified mouse myeloma proteins (MOPC 315, 7A(IgA22); MOPC 21, yG1).

Animals

Female Balb/c mice (6 to 10 weeks old) were purchased from Charles River. Mice received water and food ad libitum and did not receive treatment with antibiotics.

Immunization

For primary immunization, groups of mice were immunized orally with two doses containing 101° colony-forming units (c.f.u.) each of one or another of the Salmonella derivatives on days 0 and 4. Inocula for immunizations were prepared either from stationary- phase cultures grown at 37°C on a gyratory shaker in TSB or harvested from TSA plates supplemented as indicated. Cells were washed and resuspended in sterile normal saline to a final volume containing 2 x 101° c.f.u. ml-1, and administered intragastrically in 0.5 ml doses with a feeding tube. Where indicated, this regimen was repeated for each animal at 21 and 25 days after primary immunization. Animals were killed approximately two weeks after the final dose. The time points for the boosters were chosen because extensive studies in this and other laboratories have indicated that, following this regimen of immunization, initial levels of both serum and mucosal antibody peak between 3 and 4 weeks after primary immunizationS,7-1 o,x 9,20,24,31, 32"

Viability study

Strain EL23 was grown overnight in TSB supplemented with ampicillin and dihydroxybenzoic acid. Five different treatments were used to kill bacteria.

Acetone. A 1 litre overnight culture was harvested in 25 ml of sterile saline. After centrifugation, the organisms were resuspended in 10 ml of acetone and allowed to stand overnight at 4°C. The bacterial suspension was then poured onto 3MM filter paper placed over a vacuum device. The filter with the bacteria on it was dried at 37°C for approximately 30 min, and the organisms were

collected by washing the paper repeatedly with sterile saline.

Alcohol A 1 litre overnight culture was harvested in 25 ml of sterile saline. After centrifugation, the organisms were resuspended in 10 ml of 70% alcohol and incubated at 4°C overnight. The bacterial suspension was centrifuged and resuspended in sterile saline.

Heat. A 1 litre overnight culture was harvested in 25 ml of sterile saline. This suspension was incubated at 57°C for 1 h and then phenol was added to a final concentration of 0.5%. After overnight incubation at 4°C, the bacteria were centrifuged and resuspended in sterile saline.

Formalin. A 1 litre overnight culture was harvested in 25 ml of sterile saline. Formalin was added to a final concentration of 5%, and the bacterial suspension was incubated overnight at 4°C. After centrifugation, the organisms were resuspended in sterile saline.

Ultraviolet light. Bacteria from five plates were exposed to ultraviolet light (11000/~W cm -2) for 1 h. The organisms were harvested and were resuspended in sterile saline after centrifugation. This reduces the viable count by approximately seven orders of magnitude.

Viable. A 1 litre overnight culture was harvested in 25 ml of sterile saline.

Bacterial suspensions were standardized to contain 2 x 101° bacteria per ml. Post-treatment viability was examined by plating serial dilutions of each preparation on TSA. Plate counts indicated that each killed preparation contained less than 1 x 103 viable organisms.

Protein determinations

Protein determinations were carried out by the method of Lowry et al. 33.

Guidelines used for recombinant DNA experiments

The experiments reported here were performed under conditions as specified in the Guidelines for Recombinant DNA Technology published by the National Institutes of Health, Bethesda, MD, USA.

Statistical analysis

The standard error of the mean was calculated for all data, and means of variously immunized groups were compared by Student's t test. Values of p~<0.05 were considered statistically significant.

R E S U L T S

Effect of the nature and degree of bacterial attenuation on the humorai Ammune response directed against foreign antigens

Overlying studies on the efficacy of Salmonella as a carrier are studies on alternatives to S. typhi Ty21a as a live oral anti-typhoid vaccine. Despite the safety and efficacy of Ty21a, it is not without problems. The organism is difficult to grow and must be given on multiple occasions to be effective, and the nature of the

Vaccine 1994 Volume 12 Number 9 835

Viability and antigen dose in Salmonella vaccine carriers: L. Cardenas et al.

attenuation is not well defined. A number of new candidate anti-typhoid vaccines have been constructed, using molecular technologies to create specific defined mutations. The first of two such strains tested in volunteers 34 had an aroA deletion, expected to cause complete loss of virulence by itself, and, as an additional safety factor, a deletion at purA2% Unlike mice given a AaroA mutant of S. dublin orally 35, volunteers who received an oral dose of even 10 l° c.f.u, of either of the two AaroA ApurA live vaccine strains, 541Ty (Vi-positive) or 543Ty (Vi-negative), developed no or only very low titres of humoral antibody to the 0 antigen of the vaccine strain and were not colonized by the organisms. More recently, other strains harbouring mutations affecting the metabolism of aromatic compounds have been examined. Mutations in aroC and aroD have been associated with reduced virulence of S. typhi for humans 3°. One strain examined, CVD908, is an aroC aroD mutant of S. typhi Ty2, and produced no adverse reactions in volunteers receiving up to 5 x 107 c.f.u, orally. In addition, a single oral dose of 5 x 10 v c.f.u, of CVD908 induced seroconversion in 50% of recipients. Other genetically attenuated mutants of S. typhi have been examined in humans, including CVD906, an aroC aroD mutant of S. typhi ISPI820, and Z3927, a cya crp mutant of Ty2. These latter two strains were shown to cause fevers or vaccine bacteraemia in volunteers at relatively low doses.

We recently evaluated two different attenuated mutants of S. typhi for use as vaccine carriers: 543Ty aroA purA and 659Ty aroA aroD. We transformed each of these strains with plasmid pJC2179 which codes for production of LT-B, and each was fed orally to mice. For this experiment, groups containing six Balb/c mice each were immunized orally with 1 x 10 l° c.f.u, of one or another of these strains on two separate occasions (days 0 and 4). This regimen was repeated for each animal at 21 and 25 days after primary immunization. Animals were sacrificed two weeks after the final dose. Two control preparations were included in this study: EL23 which is an aroA S. dublin(ransformed with plasmid pJC217, and SE12 which is S. typhi Ty21a transformed with the same plasmid. In previous studies, we demonstrated that EL23 was an effective carrier in mice and that SEI2 gave little or no response when fed orally to mice 9"1°. Our presumption was that the failure of the Ty21 a derivative was a reflection of the noted species restriction. However, as shown in Table 1, animals immunized orally with S. typhi strains 543Ty(pJC217) and 659Ty(pJC217) developed significant serum and mucosal antibody responses against the foreign antigen. The serum anti-LT-B responses developed following oral immunization with 543Ty(pJC217) and 659Ty(pJC217) were significant, although lower than those developed following oral immunization with strain EL23. In contrast, the differences in mucosal IgA

anti-LT-B levels in mice receiving 543Ty(pJC217) and 659Ty(pJC217) were not statistically different to those developed in response to oral immunization with EL23. Mice immunized orally with strain SE12 developed only a marginal serum IgG response and no detectable mucosal IgA response against LT-B.

The effect of strain viability on the humeral immune response directed against foreign antigens

If host adaptation is not an a priori requirement for an organism to be an efficient carrier, the next obvious question has to do with the viability of the carrier strain. In our evolving hypothesis, the nature of the array of antigens present on the bacterial cell surface at the time of oral inoculation is the most important determinant of carrier efficacy. In fact, it may not be necessary or desirable for the carrier to be viable, as long as any necessary bacterial at tachment and uptake factors are maintained. It is now accepted that oral killed typhoid vaccines are not effective at inducing protective anti- typhoid immunity. At this point, it was not known whether oral killed Salmonella could function as an effective carrier of other antigens to the mucosal immune system. In order to test this hypothesis, we immunized groups of mice with viable and non-viable preparations of EL23. Two groups of Balb/c mice consisting of five mice each were immunized with an equal number of organisms harvested from TSA plates. One group received organisms that had been killed by exposure to ultraviolet irradiation after overnight growth. The number of viable organisms following treatment was less than 103 c.f.u, ml l as determined by plate counts. We have determined previously that the minimum number of organisms of this strain necessary to produce an immune response against LT-B in these animals is between 10 9 and 101% Consequently, animals in this group received the same number of total organisms (101°), but between six and seven orders of magnitude fewer viable organisms than required to induce a humoral response. Ultraviolet killing was chosen as a means of reducing viability while preserving surface structures on the bacteria. Animals were boosted at 21 and 25 days after primary immunization. As shown in Table 2, animals immunized orally with viable EL23 developed serum IgG and mucosal IgA anti-LT-B responses consistent with our previous findings. Significantly, mice immunized orally with ultraviolet-killed EL23 developed serum IgG and mucosal IgA antibody responses equivalent to those developed in animals orally immunized with the same number of viable EL23.

We then extended these observations and examined other methods of killing the organisms that may also preserve the ability of these strains to function as carriers.

Table 1 Humoral response following immunization with different species of Salmonella varying in the nature and degree of bacterial attenuation

S. typhi S. typhi Anti-LT-8 a S. dublin EL23 543Ty(pJC217) 659Ty(pJC217) S. typhi SE12

Serum IgG (,ug ml 1) 697_+180 152_+85 224_+33 10_+3 Mucosal IgA (ng ml 1) 511 4-128 386_+ 187 306_+ 82 None detected

aGroups containing six Balb/c mice each were immunized orally with 1 x 10 l° c.f.u, of one or another of these strains on two separate occasions (days 0 and 4). This regimen was repeated for each animal at 21 and 25 days after primary immunization. Animals were killed approximately two weeks after the final dose. Values for IgG and IgA were determined by ELISA from a standard curve with purified mouse myeloma proteins (MOPC 315, 7A(IgA22); MOPC 21, TG1). See text for additional details.

836 Vaccine 1994 Volume 12 Number 9

Viability and antigen dose in Sa lmone l l a vaccine carriers: L. CMdenas et al.

Table 2 Humoral response following immunization with viable and ultraviolet-killed Salmonella

400

S. dublin EL23 S. dublin EL23 Anti-LT-B" (viable) (ultraviolet-killed)

Serum IgG (/.tg m1-1) 450_+79 863+356 Mucosal IgA (ng ml-~) 1676_+280 3064-+- 1320

350

aGroups containing five Balb/c mice each were immunized orally with 1 x 10 ~° c.f.u, of one or another of these strains on two separate occasions (days 0 and 4). This regimen was repeated for each animal at 21 and 25 days after primary immunization. Animals were killed approximately two weeks after the final dose. Values for IgG and IgA were determined by ELISA from a standard curve with purified mouse myeloma proteins (MOPC 315, 7A(IgA22); MOPC 21, 7G1). See text for additional details.

Specifically, EL23 was killed with either acetone, ethanol, heat or formalin, and mice were immunized orally as above. Organisms were grown and killed as indicated in Materials and methods. Bacterial suspensions then were standardized to contain 2 x 1 0 ~° bacteria per ml. Post-treatment viability was examined by plating serial dilutions of each preparation on TSA. Plate counts indicated that each killed preparation contained less than 103 viable organisms per ml. As shown in Figure 1, EL23 rendered non-viable by a variety of chemical means retained the ability to elicit serum and mucosal antibody responses against a foreign antigen. The serum IgG anti-LT-B responses following oral immunization with heat-, ethanol- or acetone-killed EL23 were statistically equivalent to one another and also to the serum IgG anti-LT-B response following oral immunization with viable EL23. Likewise, the mucosal IgA anti-LT-B responses following oral immunization with heat-, ethanol- or acetone-killed EL23 were statistically equivalent to one another and also to the mucosal IgA anti-LT-B response following oral immunization with viable EL23. In this study, only formalin-killed EL23 were unable to elicit serum or mucosal antibody responses equivalent to those obtained with viable EL23. The serum IgG anti-LT-B and mucosal IgA anti-LT-B responses following oral immunization with formalin- killed EL23 were statistically significantly lower than levels obtained with all other immunization regimens.

Antigen dose as a determinant of the humorai immune response directed against foreign antigens

To address the question of the importance of antigen dose, we prepared three different strains of galE S. typhimurium J706 expressing markedly different levels of LT-B. These strains were designated J1000 (a J706 derivative with a single copy of the LT-B gene on the chromosome, utilizing the lac promoter, as employed in our previous studies), J706(pDF87) (intermediate copy number, pBR322-based LT-B plasmid utilizing the pBR322 tet promoter), and J706(pJC217) (high copy number pUC-based LT-B plasmid utilizing the lac promoter). S. typhimurium J706 served as a negative control for these studies. This scheme has a number of features which could be capitalized on. The chromosomal gene in J1000 is a stable integrate that constitutively expresses low but consistent levels of LT-B. The two plasmid-containing strains express higher levels of antigen, but recombinant plasmids are not stable in these constructs. For instance, after overnight growth in TSB, > 9 0% of recovered isolates have lost pUC-based

3OO

.-1 250

e l l

200

..,~.

< 150

100

50 ---r'-

<

e.,

<

1 6 0 0 -

1400

1200

1000

800

6O0

400

o u-I <

200

0 .=- ~ "6 ~ o

> ~ U.I <

Figure I The effect of strain viability on the humoral response directed against foreign antigens. S. dublin EL23 was killed with either acetone, ethanol, heat or formalin. Groups of Balb/c mice were immunized orally with 1 x 101° c.f.u, of one or another of these strains on two separate occasions (days 0 and 4). This regimen as repeated for each animal at 21 and 25 days after primary immunization. Animals were killed approximately two weeks after the final dose. EL23 rendered non-viable by a variety of chemical means retained the ability to elicit serum and mucosal antibody responses against the foreign antigen. Results are presented from groups containing four or five mice each and from two independent experiments. The standard error of the mean was calculated for all data, and means of variously immunized groups were compared by Student's t test. Values of p~<0.05 were considered statistically significant

Vacc ine 1994 Vo lume 12 Number 9 837

Viability and antigen dose in S a l m o n e l l a vaccine carriers: L. Cardenas et al.

70

6 0 - -

50

~'~ 40 t '- 30 < E

20 ¢.,e)

10 --4-

0 [--

Sp. Act. (ng/mg)

400

J706 J l000 JT06(pDF87) J706(pJC217) 0 2.09 45.08 6205.33

350

300

< 250

E- 200 ,..] .-

"~ 150

O

100

50

0 J706 J1000 J706(pDF87) J706(pJC217)

0 2.09 45.08 6205.33 Sp. Act. (ng/mg)

Figure 2 Effect of varying the antigen dose on the humoral immune response directed against foreign antigens. Groups consisting of four Balb/c mice each were immunized orally with 1 x 10 l° c.f.u, of one or another of three different strains of galE S. typhimuriurn J706 expressing markedly different levels of LT-B. Expression was quantified using purified LT-B as a standard and is expressed in terms of specific activity as ng LT-B (based on ELISA) per mg total protein. Mice were immunized on two separate occasions (days 0 and 4) and were killed three weeks after primary immunization. S. typhirnurium J706 served as a negative control for these studies. Mice developed increasing serum and mucosal anti-LT-B responses that correlated with the amount of antigen present within the organisms at the time of immunization. The standard error of the mean was calculated for all data, and means of variously immunized groups were compared by Student's t test. Values of p ~< 0.05 were considered statistically significant

plasmids that code for production of LT-B, even when the strain is grown in the presence of stabilizing antibiotic (unpublished observation). Since > 9 5 % of LT-B is synthesized during the first 8 h of growth in vitro, the majority of these organisms deliver antigen to the GALT

but do not synthesize antigen in situ. This is analogous to the non-viable organism question with respect to de novo synthesis of antigen. For this experiment, groups of Balb/c mice consisting of four mice each were immunized orally with 1 x 10 x° c.f.u, of one or another of these strains on two separate occasions as indicated above. Mice were killed 3 weeks after primary immunization, and examined by ELISA for serum and mucosal antibodies directed against LT-B.

As shown in Fiyure 2, mice developed increasing serum and mucosal anti-LT-B responses that correlated with the amount of antigen present within the organisms at the time of immunization. Since all of these strains persist to the same extent 24'35, these findings provide clear evidence that there is a minimum threshold below which significant levels of antibody are not elicited, and that persistence does not play a major role in eliciting antibody against the foreign antigen.

D I S C U S S I O N

A number of lines of evidence have now convinced us that the critical event determining the development of a humoral immune response against a foreign antigen delivered by Salmonella vectors is the initial amount of antigen that primes the GALT rather than the persistence of the organisms. We have demonstrated that organisms that continually express low levels of antigen and persist in tissues for up to three weeks are not as effective at inducing serum and mucosal antibody responses as organisms that deliver large quantities of antigen to the Peyer's patches and do not persist.

There are a number of points to consider with respect to these findings. First, in the study examining the effect of the nature and degree of bacterial attenuation, both 543Ty and 659Ty were derived from S. typhi CDC 10-80 by transposon-generated deletion mutagenesis and carry precisely defined mutations in the target genes 27. In contrast, Ty21 a was derived from S. typhi Ty2 by multiple rounds of nitrosoguanidine mutagenesis 2 and clearly contains genetic alterations other than the selected mutation in (TalE. This confirms and extends our findings and those of others that the degree of attenuation is important in terms of carrier selection, but perhaps not for the reasons we thought. In our previous studies 35 and those of Levine et al. 34, organisms harbouring a purA mutation were unable to elicit an O antigen-specific humoral immune response when administered orally to mice or humans. Neither of these studies determined the ability of these attenuated strains to deliver foreign antigens effectively to the mucosal immune system. In contrast, the two S. typhi strains 543Ty and 659Ty were both efficient carriers in the mouse model and both elicited a humoral anti-lipopolysaccharide response (data not shown). The fact that the organisms do not survive within the Peyer's patches and go on to cause bacteraeimia or systemic disease may argue in favour of such strains if they are efficient carriers. This also causes us to question the role of species specificity in carrier selection. While it is true that S. typhi is not a pathogen in mice, it is an effective carrier in mice. Likewise, Salmonella spp. other than S. typhi may be more effective carriers in humans. This hypothesis clearly warrants further investigation.

In the above studies on the effect of strain viability on the humoral immune response, we demonstrated that approprately prepared non-viable organisms are as

838 V a c c i n e 1994 V o l u m e 12 N u m b e r 9

Viabil i ty and ant igen dose in Salmone l la vaccine carr iers: L. C&rdenas et al.

effective as viable organisms in eliciting humoral immune responses against a foreign antigen. This is clearly a controversial finding and needs to be explored more fully. Viable Salmonella that persist within macrophages do not contribute effectively to the host humoral response, especially in the GALT. Within this context, the Salmonella must first be taken up by an appropriate antigen-presenting cell (APC), killed, lysed within the phagolysosome, and its antigens processed and exported to the surface of the APC for presentation in conjunction with the major histocompatibility complex. As long as the bacterium is viable, antigen not displayed on the surface of the organism or secreted by the organism is not available to the APC and consequently not available for immune stimulation. Moreover, viable organisms only remain within the Peyer's patches and mesenteric lymph nodes for a short period of time, and these lymphoid tissues are the IgA inductive sites, rather than the liver and spleen where the organisms eventually persist. In circumstances where induction of an IgA response is an important consideration, the Peyer's patches and mesenteric lymph nodes would be the most relevant sites for antigen presentation 35. Rendering the organisms non-viable prior to oral administration facilitates antigen processing by the APC in the GALT. However, it is clear that not all methods of killing are equivalent in terms of preserving the ability of these strains to function as carriers. Formalin-killed organisms were not as effective as delivery vehicles as organisms killed by other methods.

The use of non-viable Salmonella as carriers for delivery of antigens to the mucosal lymphoid tissues is analogous to the use of liposomes and microspheres for oral antigen delivery. In just one example, Eldridge et al. 37 demonstrated that biodegradable microspheres containing bacterial antigens were efficiently internalized by M cells and that mice receiving such an antigen carrier preparation showed a rise in antigen-specific antibodies. If one views killed Salmonella as 'formerly viable microspheres', it is logical that they would also be efficient delivery vehicles. Two additional points should be noted. The first is that these observations were made using LT-B which is immunologically and structurally related to the cholera toxin B subunit 3°. Both of these molecules are highly immunogenic oral antigens even when delivered as soluble proteins without encapsulation. How broadly our findings apply to other proteins is not yet known. The second point is that we have only determined the ability of these strains to induce humoral immunity; the influence of persistence, viability and antigen dose on development of cellular immunity is not clear. Both of these points are currently under investigation in our laboratory.

The issue of strain viability is one of great concern. Attenuated Salmonella have been proposed as carriers for a variety of antigens of eukaryotic, prokaryotic and viral origin. Certainly, there is great potential benefit if such vaccines can be made safe and effective. However, there is a high risk-benefit ratio associated with administration of a Gram-negative organism that causes bacteraemia and persists in tissues for extended periods of time, especially in vaccines intended for administration to children, potentially immunosuppressed individuals, or individuals in developing countries where nutrition is not optimum and infectious disease rates are high. Our findings indicate that viability is not a requirement for use of a Salmonella strain as a carrier. If this proves to be true in humans as well as in animals, then the approach

to carrier selection and safety will change dramatically. Even the most promising alternatives to Ty21a as a live oral anti-typhoid vaccine may cause vaccine bacteraemia when administered in doses above 5 × 107 c.f.u. It is not known whether these or other strains will be efficient carriers of heterologous antigens at the relatively low level shown to be completely safe. Our own evidence indicates that bacteraemia and persistence in tissues are not necessary for oral immunization, and therefore it may be best to dissociate the question of what makes the best alternative to Ty21a as a live oral anti-typhoid vaccine from the question of what makes a good carrier for heterologous antigens.

A C K N O W L E D G E M E N T S

This work was supported by Public Health Service Grant AI28835 from the National Institute of Allergy and Infectious Diseases. UGD is the recipient of a Short-term Overseas Associateship of the Department of Bio- technology, Government of India.

REFERENCES

1 Germanier, R. and F0rer, E. Immunity in experimental salmonellosis. I1. Basis for the avirulence and protective capacity of gale mutants of S. typhimurium. Infect. Immun. 1971, 4, 663-673

2 Germanier, R. and FL~rer, E. Isolation and characterization of galE mutant Ty21a of Salmonella typhi: a candidate strain for a live, oral typhoid vaccine. J. Infect. Dis. 1975, 131,553-558

3 Hoiseth, S.K. and Stocker, B.A.D. Aromatic-dependent Salmonella typhimurium are non-virulent and effective as live vaccines. Nature 1981, 291,238-239

4 McFarland, W.C, and Stocker, B.A.D. Effect of different purine auxotrophic mutations on mouse virulence of a Vi-positive strain of Salmonella dublin and of two strains of Salmonella typhimurium. Microb. Pathog. 1987, 3, 129-141

5 Curtiss, R. Ill and Kelly, S.M. Salmonella typhimurium deletion mutants lacking adenylate cyclase and cyclic AMP receptor protein are avirulent and immunogenic. Infect. Immun. 1987, 55, 3035-3043

6 Brown, A., Hormaeche, C.E., de Hormaeche, R.D., Winther, M., Dougan, G., Maskell, D.J. and Stocker, B.A.D. An attenuated aroA Salmonella typhimurium vaccine elicits humoral and cellular immunity to cloned fl-galactosidase in mice. J. Infect. Dis. 1987, 155, 86-92

7 Clements, J.D. Use of attenuated mutants of Salmonella as carriers for delivery of heterologous antigens to the secretory immune system. Pathol. Immunopathol. Res. 1987, 6, 137-146

8 Clements, J.D. and C~lrdenas, L. Vaccines against enterotoxigenic bacterial pathogens based on hybrid Salmonella that express heterologous antigens. Res. Microbiol. 1990, 141,981-994

9 Clements, J.D. and EI-Morshidy, S. Construction of a potential live oral bivalent vaccine for typhoid fever and cholera- Escherichia coli-related diarrheas. Infect. Immun. 1984, 48, 564-569

10 Clements, J.D., Lyon, F.L., Lowe, K.L., Farrand, A.L. and EI-Morshidy, S. Oral immunization of mice with attenuated Salmonella enteritidis containing a recombinant plasmid which codes for production of the B subunit of heat-labile Escherichia coli enterotoxin. Infect. Immun. 1986, 53, 685-692

11 Curtiss, R. III, Goldschmidt, R., Kelly, S., Lyons, M., Michalek, S., Pastian, R. and Stein, S. Recombinant avirulent Salmonella for oral immunization to induce mucosal immunity to bacterial pathogens. In: Vaccines: new concepts and developments. Proceedings of the 10th International Convocation on Immunology (Eds Kohler, H. and LoVerde, P.T.) Longman Scientific and Technical, Harlow, Essex, 1987, pp 261-271

12 Dougan, G., Sellwood, R., Maskell, D., Sweeney, K., Liew, F.Y., Beesley, J. and Hormaeche, C. In vivo properties of a cloned K88 adherence antigen determinant. Infect. Immun. 1986, 52, 344-347

13 Flynn, J.L., Weiss, W.R., Norris, K.A., Seifert, H.S., Kumar, S. and So, M. Generation of a cytotoxic T-lymphocyte response using a Salmonella antigen-delivery system. Mol. Microbiol. 1990, 4(12), 2111-2118

14 Formal, S.B., Baron, L.S., Kopecko, D.J., Washington, O., Powell, C. and Life, C.A. Construction of a potential bivalent vaccine strain:

Vaccine 1994 Vo lume 12 Number 9 839

Viabi l i ty and ant igen dose in Salmone l la vaccine carr iers: L. Cardenas et al.

introduction of Shigella sonnei form I antigen genes into the galE Salmonella typhi Ty21a typhoid vaccine strain. Infect. Immun. 1981, 34, 746-750

15 Maskell, D., Liew, F.Y., Sweeney, K., Dougan, G. and Hormaeche, C.E. Attenuated Salmonella typhimurium as live oral vaccines and carriers for delivering antigens to the secretory immune system. In: Vaccines 86: New approaches to immunization. Developing vaccines against parasitic, bacterial, and viral diseases (Eds Brown, F., Channok, R.M. and Lerner, R.A.) Cold Spring Harbor Laboratory, New York, 1986, pp 213-217

16 Poirier, T.P., Kehole, M.A. and Beachey, EH. Protective immunity evoked by oral administration of attenuated aroA Salmonella typhimurium expressing cloned streptococcal M protein. J. Exp. Med. 1988, 168, 25-32

17 Sadoff, J.C., Ballou, W.R., Baron, L.S., Majarian, W.R., Brey, R.N., Hockmeyer, W.T. et al. Oral Salmonella typhimurium vaccine expressing circumsporozoite protein protects against malaria. Science 1988, 240, 3,36-340

18 Stabel, T.J., Mayfield, J.E., Tabatabai, L.B. and Wannemuehler, M.J. Oral immunization of mice with attenuated Salmonella typhimurium containing a recombinant plasmid which codes for production of a 31-kilodalton protein of Brucella abortus. Infect. Immun. 1990, 58, 2048-2055

19 C&rdenas, L. and Clements, J.D. Oral immunization using live attenuated Salmonella spp. as carriers of foreign antigens. Clin. Microbiol. Rev. 1992, 5, 328-342

20 C~.rdenas, L. and Clements, J.D. Stability, immunogenicity, and expression of foreign antigens in bacterial vaccine vectors. Vaccine 1993, 11, 126-135

21 Cohen, S., Powell, C.J., Dubois, D.R., Hartman, A., Summers, P.L. and Eckels, K.H. Expression of the envelope antigen of dengue virus in vaccine strains of Salmonella. Res. Microbiol. 1990, 141, 655-858

22 O'Callaghan, D., Maskell, D., Liew, F.Y., Easmon, C.S.F. and Dougan, G. Characterization of aromatic- and purine-dependent Salmonella typhimurium: attenuation, persistence, and ability to induce protective immunity in Balb/c mice. Infect. Immun. 1988, 68,419-423

23 Salas-Vidal, M., PlebaSski, M., Castro, S., Perales, G., Mata, E., Lopez, S. and Arias, C.F. Synthesis of the surface glycoprotein of rotavirus SA11 in the aroA strain of Salmonella typhimurium SL3261. Res. Microbiol. 1990, 141,883-886

24 Hone, D., Attridge, S., van den Bosch, L. and HackeR, J. A chromosomal integration system for stabilization of heterologous genes in Salmonella based vaccine strains. Microb. Pathog. 1989, 5, 407-418

25 Mahan, MJ., Slauch, J.M. and Mekalanos, J.J. Selection of bacterial virulence genes that are specifically induced in host tissues. Science 1992, 259, 686--688

26 Strugnell, R.A., Maskell, D., Fairweather, N., Pickard, D., Cockayne, A., Penn, C. and Dougan, G. Stable expression of foreign antigens from the chromosome of Salmonella typhimurium vaccine strains. Gene 199..,0, 88, 57-63

27 Edwards, M.F. and Stocker, B.A.D. Construction of AaroA his z~our strains of Salmonella typhi. J. Bacteriol. 1988, 170, 3991-3995

28 Clements, J.D., Flint, D.C., Engert, R.F. and Klipstein, F.A. Cloning and molecular characterization of the B subunit of Escherichia coil heat-labile enterotoxin. Infect. Immun. 1983, 40, 65,3-658

29 Ausubell, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A. and Struhl, K. Current Protocols in Molecular Biology 1987, 1

30 Clements, J.D., Yancey, R.J. and Finkelstein, R.A. Properties of homogeneous heat-labile enterotoxin from Escherichia coil cultures. Infect. Immun. 1980, 24, 91-97

31 Dougan, G., Chatfield, S., Pickard, D., O'Callaghan, D. and Maskell, D. Construction and characterization of Salmonella vaccine strains harbouring mutations in two different aro genes. J. Infect. Dis. 1988, 158, 1324-1335

32 Nakayama, K., Kelly, S.K. and Curtiss, R. II1. Construction of an Asd + expression-cloning vector: stable maintenance and high level expression of cloned genes in a Salmonella vaccine strain. Bio/Technology 1988, 6, 693-697

33 Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J. Protein measurement with the Folin phenol reagent. J. Biol. Chem. 1951, 193, 265

34 Levine, M.M., Herrington, D., Murphy, J., Morris, J.G., Losonsky, G., Tall, B. et al. Safety, infectivity, immunogenicity and in vivo stability of two attenuated auxotrophic mutant strains of Salmonella typhi, 541Ty and 543Ty, used as oral vaccines in man. J. Clin. Invest. 1987, 79, 888-902

35 Sigwart, D.F., Stocker, B.A.D. and Clements, J.D. Effect of purA mutation on the efficacy of Salmonella live vaccine vectors. Infect. Immun. 1989, 57, 18.58-1861

36 Tacket, C.O., Hone, D.M., Curtiss, R. III, Kelly, S.M., Losonsky, G., Guers, L. et al. Comparison of the safety and immunogenicity of AaroC AaroD and Acya Acrp Salmonella typhi strains in adult volunteers. Infect. Immun. 1992, 60, 536-541

37 Eldridge, J.H., Gilley, R.M., Staas, J.K., Moldoveanu, Z., Meulbroek, J.A. and Tice, T.R. Biodegradable microspheres: vaccine delivery system for oral immunization. Curt. Top Microbiol. Immunol. 1989, 146, 59-66

840 Vaccine 1994 Vo lume 12 Number 9