INFECTION AND IMMUNITY, May 1992, p. 1786-1792 0019-9567/92/051786-07$02.00/0 Copyright © 1992, American Society for Microbiology

Vol. 60, No. 5

Monoclonal Secretory Immunoglobulin A Protects Mice against Oral Challenge with the Invasive Pathogen Salmonella typhimurium PIERRE

MICHETrTI,12 MICHAEL J. MAHAN,3 JAMES M. SLAUCH,3 JOHN J. MEKALANOS,3 AND

MARIAN R. NEUTRA1*

GI Cell Biology Laboratory, Children's Hospital, and Department of Pediatrics,1 and Department of Microbiology and Molecular Genetics, 3 Harvard Medical School, Boston, Massachusetts 02115, and Division of Gastroenterology, Centre Hospitalier Universitaire Vaudois, CH-1011 CHUV-Lausanne, Switzerland2 Received 28 October 1991/Accepted 14 February 1992

Hybridomas producing monoclonal immunoglobulin A (IgA) antibodies against Salmonella typhimurium generated by mucosal immunization of BALB/c mice with attenuated strains of S. typhimurium and subsequent fusion of Peyer's patch lymphoblasts with myeloma cells. To test the role of secretory IgA (sIgA) in protection against Salmonella sp., we analyzed in detail the protective capacity of a monoclonal IgA, Sal4, produced in polymeric as well as monomeric forms, that is directed against a carbohydrate epitope exposed on the surface of S. typhimurium. BALB/c mice bearing subcutaneous Sal4 hybridoma tumors and secreting monoclonal sIgA into their gastrointestinal tracts were protected against oral challenge with S. typhimurium. This protection was directly dependent on specific recognition by the monoclonal IgA, since mice secreting Sa14 IgA from hybridoma tumors were not protected against a fully virulent mutant that lacks the Sal4 epitope. Although monoclonal Sal4 IgA was present in the bloodstreams and tissues of tumor-bearing mice, it did not protect against intraperitoneal challenge and did not possess complement-fixing or bacteriocidal activity in vitro. Taken together, these results indicate that secretion of sIgA alone can prevent infection by an invasive enteric pathogen, presumably by immune exclusion at the mucosal surface. were

dose of virulent vibrios (26). V. cholerae, however, does not penetrate the intestinal mucosa or generate a systemic infection. Thus, it is not known whether sIgA alone can prevent entry of invasive bacteria into the mucosa, infection of mucosal cells, and systemic spread. To address this issue, we have studied the invasive pathogen Salmonella typhimunum in a mouse model. S. typhimurium, a gram-negative bacterium, invades the intestinal epithelium, infects cells of the reticulo-endothelial system, and results in infection of liver and spleen (10). Systemic dissemination produces a lethal disease in mice that is similar to enteric fever in humans caused by S. typhi (4). In mice, early entry of S. typhimunum is via transepithelial transport by M cells of Peyer's patches (10, 13). Later, absorptive enterocytes can be invaded and damaged, providing an additional portal of entry (23). Infection of cells in Peyer's patch mucosa is followed by bacteremia and systemic disease that can be mimicked by intraperitoneal (i.p.) injection of organisms (4). S. typhimurium is also a human pathogen, causing acute gastroenteritis (9). It is therefore important to determine the role of sIgA antibodies in providing mucosal protection against this organism. Furthermore, identification of Salmonella epitopes that are immunogenic and protective in the mucosal immune system is relevant to design of both active and passive immunization protocols. In this report, we show that a single monoclonal sIgA, secreted into the lumen of the mouse intestine by the normal epithelial transport mechanism, is sufficient to confer protection against an oral challenge of S. typhimunum.

Although the intestinal epithelium provides a barrier against foreign materials and microorganisms present in the intestinal lumen, numerous pathogens enter the body through this cellular layer. Protection of the vast mucosal surface of the intestine depends in part on cells of the local mucosal immune system that sample lumenal antigens and generate an immune response including antigen-sensitized T lymphocytes and B lymphocytes that produce polymeric immunoglobulin A (IgA) antibodies. Secretory IgA (sIgA) antibodies are thought to provide mucosal defense by immune exclusion; this refers to their ability to prevent contact of pathogens with epithelial surfaces through agglutination in the intestinal lumen, entrapment of immune complexes in mucus, and clearance by peristalsis (1, 6, 12, 15). The relative importance of sIgA antibodies in protection against infection with Salmonella sp. has not been clearly established (7). Typhoid patients and humans given live oral typhoid vaccines have circulating IgG, IgM, and IgA antibodies and cellular immunity as well as sIgA antibodies directed against Salmonella typhi (3, 5, 11, 14, 20). The protective capacity of specific IgA antibodies has not previously been tested, in part because of the difficulties in obtaining well-characterized sIgA antibodies in sufficient amounts and in delivering exogenous IgA antibodies into normal mucosal secretions. We recently developed methods for efficient production of IgA hybridomas from Peyer's patch cells (25) and for delivery of monoclonal IgA antibodies into intestinal secretions via the normal transepithelial transport mechanism (26). With this approach, we demonstrated that monoclonal sIgA directed against a single surface epitope of Vibrio cholerae can protect neonatal mice against oral challenge with a lethal *

MATERUILS AND METHODS Bacterial strains. All bacterial strains used in this study are derivatives of S. typhimunum ATCC 14028 (CDC 6516-60).

Corresponding author. 1786

VOL. 60, 1992

MONOCLONAL sIgA PROTECTS AGAINST S. TYPHIMURIUM

Strain MT110 was isolated as a spontaneous streptomycinresistant mutant of strain 14028 that retained full virulence as judged by oral or i.p. 50% lethal dose (LD50). Strain MT114 is a derivative of MT110 that contains a TnlOd-Tc element conferring a phenotype that lacks the surface carbohydrate epitope defined by monoclonal IgA antibody SaI4, described below. The isolation and characterization of this fully virulent mutant strain is also described below. Strains CS022 and CS015 are derivatives of strain 14028 that contain a phoP constitutive allele, phoP24 (17), and a phoP null allele, phoP102::Tn1Od-Tc (16), respectively. Production and screening of anti-S. typhimurium IgA hy. bridomas. Five female BALB/c mice were immunized orally on day 0 by gastric intubation of 1010 live cells of the phoP constitutive mutant strain CS022 (17) in 0.2 M sodium bicarbonate. Three additional BALB/c mice were immunized orally with 1010 cells of the phoP null strain, CS015 (16). Immunization was repeated at day 10. On day 14, mice were sacrificed by cervical dislocation and the Peyer's patches (7 to 10 per mouse) were excised. Peyer's patch lymphocytes were isolated, pooled, washed in tissue culture medium, and fused with P3X63/Ag8U.1 mouse myeloma cells as previously described (25). Fusion products were plated in 96-well plates with feeder layers of thymocytes isolated from DBA/2 mice. Hybridoma cell cultures were grown in RPMI tissue culture medium (Sigma Chemical Co., St Louis, Mo.) supplemented with 10% fetal bovine serum, 2 mM glutamine (GIBCO Laboratories, Grand Island, N.Y.), 10 mM HEPES (N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid; Sigma), and 1 mM sodium pyruvate (GIBCO). Culture supernatants were screened for the production of anti-S. typhimunum immunoglobulin by enzyme-linked immunosorbent assay (ELISA) with S. typhimunum 14028 whole-cell lysate as the immobilized antigen and rabbit anti-mouse IgG, IgA, and IgM antibodies coupled to peroxidase (Zymed Laboratories, South San Francisco, Calif.) as the secondary antibody. Positive wells were verified and isotyped by using rabbit antibodies directed against mouse IgA, IgG, or IgM (Zymed). The isotype of the monoclonal antibodies was identified by coating ELISA plates with the antibody to be tested at pH 9.6 and performing ELISA assays with anti-heavy chain isotype-specific antisera (Boehringer GmbH, Mannheim, Germany) as the primary antibodies. Specific IgA-producing hybridomas were expanded and frozen. The IgA hybridomas selected for further investigation were cloned three times by limiting dilution on DBA/2 thymocyte feeder layers. Characterization of anti-S. typhimurium monoclonal antibodies. S. typhimunum lysates were prepared by boiling whole cells in sample buffer, and proteins were separated on 10% polyacrylamide-sodium dodecyl sulfate (SDS) gels and transferred to nitrocellulose. The nitrocellulose blots were blocked in 5% nonfat dry milk in phosphate-buffered saline (PBS) followed by incubation in IgA-producing hybridoma culture supernatant. The blots were then incubated with biotin-linked rabbit anti-mouse IgA (Zymed) as a secondary antibody, with subsequent incubation with streptavidin coupled to peroxidase (Pierce, Rockford, Ill.). To assess the recognition of carbohydrate epitopes by the monoclonal IgA, nitrocellulose strips were incubated in periodic acid (Sigma) at pH 4.5 before application of IgA antibodies, as previously described (27). This procedure oxidizes carbohydrate moieties containing vicinal hydroxyl groups and destroys most carbohydrate epitopes. Controls were incubated in pH 4.5 buffer without periodic acid. The polymeric state of the monoclonal antibodies was

1787

determined by separation on nondenaturing polyacrylamide gels. IgA antibody was immunoprecipitated by incubating a 24-h hybridoma culture supernatant overnight at 4°C with anti-mouse a chain antibodies (Amersham Corp., Arlington Heights, Ill.) linked to CNBr-activated Sepharose beads (Pharmacia, Uppsala, Sweden). The IgA-bound beads were washed four times and eluted in sample buffer at 65°C. The IgA-containing eluate was separated on a 3 to 15% gradient polyacrylamide-SDS gel, electrotransferred to nitrocellulose, and incubated with biotinylated goat antibodies directed against mouse IgA (Amersham) and then with streptavidin-horseradish peroxidase (Pierce). In vitro agglutination. To test in vitro agglutination of S. typhimurium by monoclonal IgA antibodies, 0.1 ml of hybridoma culture supernatants was added to 0.1 ml of an overnight culture of bacteria and incubated in round-bottom ELISA plates. Unrelated IgA hybridoma supematant (anti-V. cholerae [26]) or fresh culture medium was used as a control. Agglutination was detected visually after 2 to 5 h of incubation at 23°C. Bacteriocidal activity assay. Direct bacteriocidal activity of specific IgA antibodies was assessed by incubation of antibody with S. typhimurium for 2 h at 37°C. After resuspension of the bacteria, the number of viable bacteria was determined by plating dilutions of the reaction mixture on LuriaBertani (LB) plates and counting CFU. Anti-V. cholerae IgA and fresh culture medium were used as controls. The results were recorded as percentages of the control CFU. Complement activation assay. A complement sensitivity assay was performed as previously described (19). Briefly, bacteria from overnight cultures were washed once, resuspended in PBS at 107 CFU/ml, and aliquoted into glass tubes. Antibody was added to the bacteria at various dilutions in cell culture medium, and then guinea pig complement (Difco Laboratories, Detroit, Mich.) was added to a final concentration of 20%. Negative control tubes in which the antibody was omitted or replaced by unrelated anti-V. cholerae monoclonal IgA were prepared. Rabbit anti-S. typhimurium antiserum (Difco) was used as positive control. All tubes were incubated at 37°C for 15 min, the reactions were stopped by addition of ice-cold LB, and the CFU were determined by plating at increasing dilutions on LB plates. Results were recorded as percent survival in the reaction mixture compared with that in the negative control. In vivo protection experiments. At day 0, 106 hybridoma cells were subcutaneously injected in the upper backs of 6to 8-week-old nonimmune BALB/c female mice to generate IgA-secreting hybridoma tumors, as previously described (26). Prior to injection, the IgA-producing hybridoma cells were grown to 40 to 50% saturation, harvested, and washed twice in sterile PBS. Mice were injected with hybridoma cells producing either anti-S. typhimurium IgA (clone Sal4) or anti-V. cholerae IgA (clone 2D6 [26]). At day 12, blood was obtained from all mice by tail bleed, and levels in serum of anti-Salmonella sp. or anti-V. cholerae IgA were measured by ELISA analysis as described below. About 70% of these mice showed serum levels of specific IgA detectable at a 1/1,000 dilution, and only these mice were included further in the protocols. At sacrifice, representative mice from this group also showed specific IgA in intestinal secretions diluted in 2 ml during intestinal lavage and further diluted 1:10 for ELISA. At day 13, mice were challenged by intragastric intubation of 105, 106, or 107 MT11O or MT114 S. typhimurium cells in 0.1 M phosphate buffer, pH 8. During the entire experimental period mice were given food and water ad libitum. Five days after challenge, the mice were

1788

MICHETTI ET AL.

sacrificed and their spleens were removed and homogenized between sterilized frosted glass slides in saline. Dilutions of the homogenates were plated on LB plates containing 100 ,ug of streptomycin per ml to determine the number of viable bacterial cells. Additional groups of tumor-bearing mice (prepared as described above) were challenged by i.p. injection of 20, 200, or 2,000 MT110 or MT114 S. typhimurium cells in saline solution. These mice were sacrificed 3 days postchallenge, and spleens were removed and treated as described above. To determine entry of S. typhimurium into Peyer's patch mucosa, other groups of mice were prepared and perorally challenged as described above but were sacrificed 2 days after challenge. After collection of blood samples, the entire small intestine was removed and all visible Peyer's patches were excised. Whole Peyer's patches, including the entire intestinal wall from lumenal surface to seros', were rinsed in PBS and pooled for each mouse. To eliminate endogenous streptomycin-sensitive lumenal bacteria, intact tissue samples were immersed in RPMI culture medium supplemented with 10% fetal bovine serum and 50 ,ug of streptomycin per ml for 20 min at 4°C followed by 20 min at 23°C and again rinsed twice in cold PBS. To kill any extracellular S. typhimunium remaining associated with the mucosal surface, we adapted methods previously described for cultured epithelial cells (8). Intact Peyer's patch tissues were incubated in culture medium containing 50 ,ug of gentamicin per ml for 20 min at 23°C and then again rinsed twice in cold PBS. Peyer's patches were then homogenized between groundglass slides in 2 ml of cold PBS, and dilutions were plated as described above for spleens. ELISA of serum and gut wash samples. After 2 h of clotting at room temperature, blood samples were centrifuged for 3 min at 7,000 x g, and serum was collected. Samples were frozen at -20°C if not used immediately. Gut wash samples were obtained from control mice that were carrying a hybridoma tumor but were not challenged. The small intestine was flushed with 2 ml of PBS containing 1 mM phenylmethylsulfonyl fluoride (Boehringer), and the flushed lumenal material was collected distally and centrifuged at 1,000 x g for 5 min to remove debris. The supernatant was centrifuged at 7,000 x g for 10 min to remove cells and bacteria. Dilutions of serum and gut wash samples in PBS were applied to ELISA plates coated with S. typhimunium wholecell lysate as an immobilized antigen. Rabbit antibodies directed against mouse IgA, coupled to biotin (Zymed), were used as a secondary antibody and were detected with streptavidin-horseradish peroxidase (Pierce). IgA monoclonal antibodies from hybridoma culture supernatant served as a control. RESULTS Production of monoclonal IgA directed against S. typhimurium. As the first step in the production of monoclonal antibodies, we orally immunized BALB/c mice with live S. typhimurium. In order to maximize antigen delivery to the mucosal immune system of mice without causing lethal disease, we used attenuated strains of S. typhimurium containing either a phoP constitutive allele, phoP24, or a phoP null allele, phoP102::TnJOd-Cm (16, 17). Use of these attenuated strains allowed us to orally immunize mice with 1010 S. typhimurium cells, which is 4 logs above the LD50 for wild-type S. typhimunum. Hybridomas producing monoclonal antibodies were obtained by fusing Peyer's patch lymphocytes from orally immunized mice with mouse myeloma

INFECT. IMMUN.

cells, and S. typhimuium-specific monoclonal antibodies were identified by ELISA.

Thirty-nine stable hybridomas producing IgA directed against S. typhimurium were obtained from mice immunized with the S. typhimunum phoP constitutive strain, and seven were obtained from mice immunized with the phoP null strain. No stable IgG-producing hybridomas were obtained. Twelve IgA-producing clones were selected, expanded, and subcloned twice by limiting dilution. These were selected because they produced IgA antibodies that agglutinated intact S. typhimunum cells. All of these recognized a broad smear on Western immunoblots of Whole S. typhimunum cell lysates separated by SDS-polyacrylamide gel electrophoresis (PAGE) under nonreducing conditions. Four of these hybridoma clones were injected subcutaneously into BALB/c mice, and three formed hybridoma tumors. One of the tumor-forming IgA hybridoma subclones, termed SaI4, was characterized further and used throughout the study. Sal4 IgA recognizes an externally 'exposed carbohydrate epitope of S. typhimurium. Sal4 IgA bound and agglutinated whole S. typhimurium cells in vitro, indicating that the antibody recognized a cell surface epitope. To characterize the epitope, we probed Western blots of whole S. typhimuium cell lysates with Sal4 IgA antibody. Figure 1A (lane 1) shows that this antibody did not react with a single distinct band but rather recognized a broad smear of material from approximately 80 kD to the top of the gel. To determine if the epitope is carbohydrate in nature, we oxidized carbohydrate residues on the nitrocellulose blots with periodate prior to exposure to the antibody and observed that this treatment caused loss of recognition by Sal4 IgA (Fig. !A, lane 3). This suggests that Sal4 recognizes a carbohydrate epitope. Taken together, these data indicate that the Sal4 epitope is on an externally exposed carbohydrate moiety. Sal4 IgA is not bacteriocidal in vitro. To determine whether binding of Sal4 IgA to the bacterial surface had any inherent bacteriocidal activity, bacteria were incubated with the antibody and cell viability was determined. The number of yiable bacteria recovered after incubation with Sal4 IgA was 97% of that recovered in the absence of specific antibody, 4espite the agglutination of the bacteria during the assay. A control monoclonal IgA directed against V. cholerae did not cause agglutination and had no effect on the viability of S. typhimurium. To test if Sal4 IgA antibodies could activate complement, Sal4 antibody and S. typhimurium cetIs were incubated in the presence or absence of purified guinea pig complement. The presence of complement resulted in no detectable decrease in the number of viable bacteria. In contrast, there was a 50% decrease in the number of bacteria recovered when the reaction mixture included rabbit anti-S. typhimurium serum, confirming that the complement was active. Sal4 IgA is recognized by epithelial polymeric immunoglobulin receptors. As a prerequisite for the use of Sal4 hybridoma cells in protection experiments, it was important to confirm that the cultured hybridoma cells produced IgA antibodies in dimeric and polymeric forms, analogous to normal sIgA antibodies. This was demonstrated by Western blot analysis of culture supernatants separated by nonreducing SDS-PAGE (Fig. 1B). Polymeric forms of IgA are known to be recognized by intestinal epithelial polymeric immunoglobulin receptors that are responsible for transport of IgA into the intestinal lumen (21). To confirm that the monoclonal IgA antibodies produced by the Sal4 hybridoma cells are recognized by epithelial polymeric immunoglobulin receptors and can be transported

MONOCLONAL sIgA PROTECTS AGAINST S. TYPHIMURIUM

VOL. 60, 1992

1 2 3 4 134

..I.

§

|

MW

MW

-200 -116

-200

* '97

-116

-45

-

A

97

B

FIG. 1. Immtinoblot analysis of monoclonal IgA antibody Sal4. (A) S. typhimurium lysates were separated, by SDS-PAGE, transferred to nitrocellulose, and probed with Sal4 IgA. Lane 1, wild-type strain MT110. Sal 4 igA recognizes a broad smear of high-molecularmass material includitig several bands at 95 kDa and above. Lane 2, mutant strain MT114. Sal4 IgA fails to recognize any components of MT114, consistent with its inability to agglutinate these organisms. Lane 3, wild-type strain MT110. Pretreatment of the Western blot with periodic acid at pH 4.5 dramatically reduces Sal4 IgA binding, suggesting that this antibody recognizes a carbohydrate epitope. Lane 4, control blot incubated in pH 4.5 buffer without periodic acid. (B) IgA antibodies were immunoprecipitated from a Sal4 hybridoma culture, separated by 3 to 15% SDS-PAGE, transferred to nitrocellulose, and probed with biotinylated goat anti-mouse IgA antibodies followed by streptavidin-horseradish peroxidase. Sal4 hybridoma cells produce IgA monomers, dimers (double arrowheads), and higher polymers. Under the nonreducing conditions used here, IgA monomers run at an apparent molecular mass of 116 to 120 kDa (12, 25). into secretions in vivo, we generated subcutaneous hybridtumors in mice and tested serum and intestinal secretions of the mice for anti-S. typhimurium IgA activity by ELISA. Specific anti-Salmonella IgA was present both in serum (at a 1:1,000 dilution) and intestinal washes (at a 1:10 dilution) from all mice bearing Sai4 hybridoma tumors but was undetectable in serum or secretions from mice bearing anti-V. cholerae hybridoma tumors. We previously demonstrated that in mice bearing hybridoma tumors, there is a close correlation between the level of specific IgA in serum with that measured in intestinal secretions (26). Therefore, in the present study, serum-specific IgA levels were used as an indirect indicator of monoclonal secretory IgA levels in the intestine. A mutant S. typhimurium lacks the Sal4 IgA epitope. The fact that Sal4 IgA agglutinated whole S. typhimunium cells allowed us to enrich for mutant S. typhimunium cells that no oma

longer produce the Sal4 epitope and therefore are not recognized by the antibody. When a pool of S. typhimurium chromosomal transposon insertion mutants was incubated with Sal4 antibody and allowed to agglutinate, the mutant

1789

bacteria not recognized by the antibody remained in suspension. Such cells were removed and grown overnight. This agglutination procedure was repeated six times, and the final cell suspension was plated on LB agar to obtain single colonies. Of 48 individual colonies tested, cells grown from 27 were not agglutinated by Sal4 IgA. The phenotype characterized by the inability to be agglutinated by Sal4 monoclonal IgA antibody was termed Agg4. To confirm that the Agg4 phenotype is caused by the insertion of the Tcr transposon, bacteriophage P22 was grown on eight of these mutants and the resulting lysates were used to transduce wild-type S. typhimurium to tetracycline resistance. All of the Tcr transductants tested from each of the eight P22 donor lysates retained the Agg4 phenotype. These results show that in each case, the Agg4 phenotype is inextricably linked to the insertion mutation. In other words, the insertion element affected a gene whose product is required for production of the Sal4 epitope. We transduced one of these insertion mutations into MT110, and the resultant Agg4 strain, MT114, was used throughout the study. To confirm that the Sal4 epitope is not present on the mutant MT114 strain, we performed Western blot analysis of whole-cell lysates of the MT114 organisms. Figure 1A (lane 2) shows that Sal4 antibody did not react with any MT114 components, demonstrating that the epitope is not present externally or internally in the mutant cells. We then determined if the insertion mutation affects the virulence of S. typhimurium by challenging mice either with Agg4 strain MT114 or with MT110. The oral and i.p. LD50s of MT114 were indistinguishable from those of its wild-type parent, MT110. In healthy mice without hybridoma tumors, the oral LD50 for both strains was 2.0 x 106 organisms, and 20 or fewer organisms was a lethal dose by the intraperitoneal route. Sal4 monoclonal IgA prevents systemic infection after oral inoculation of S. typhimurium. To test if a monoclonal IgA is sufficient to confer protection against oral challenge with S. typhimurium, IgA was delivered into mucosal secretions from hybridoma tumors in mice, as previously described (26). The hybridoma tumors produce specific IgA in the bloodstream, a portion of which is transported into the lumen of the intestine via receptor-mediated transepithelial transport (21, 26). The tumor-bearing mice were then orally challenged with S. typhimurium. S. typhimurium invades the intestinal mucosa, and after a transient bacteremia, a progressive infection of the reticulo-endothelial system is established (4, 10). Since a consistent result of this systemic spread is a characteristic infection of the spleen and liver, monitoring of the bacterial CFU in the spleen is a reliable measurement of systemic infection. Injection of less than 10 S. typhimurium cells i.p. bypasses the epithelial barrier and almost invariably results in a lethal systemic infection. Therefore, our criterion for complete protection against infection after oral challenge was sterility of the spleen. If specific recognition of the bacterial cell surface by Sal4 IgA is required for protection, then the Sal4 antibody should not protect against the equally virulent Agg4 mutant. Thus, to provide an appropriate control group for the Sal4 hybridoma tumor-bearing mice that were challenged with wild-type S. typhimurium, mice bearing identical tumors were challenged with the Agg4 mutant strain, MT114. As an additional control, we implanted mice with hybridoma cells producing anti-V. cholerae IgA that does not recognize S. typhimunum and thus should not confer protection. Tumor-free BALB/c mice did not provide valid controls, because normal mice did

1790

MICHETTI ET AL.

Challenge dose:

INFECT. IMMUN.

105

106

107

2x101I

Challenge dose:

2x10 2

2xl03

10 8

1o8 U

a

*

U c

1o6

_

a

MEU

mE

10 6

0~* ~~~~00

0

0

U U)

0

c

0t

U

0

IL

0o 0~~~~ 0

CD

1o2

104 0~~~

_ IL.

C. 0

000

0

0000

MT110 MT 114

+

ME

EKE

Ciao-on 000

10 2 no

FIG. 2. Systemic infection after oral challenge. S. typhimurium recovered from spleens of mice bearing Sal4 IgA hybridoma backpack tumors, 5 days after oral challenge with the wild-type strain MT110 (open squares) or the Agg4 mutant MT114 (solid squares). Total CFU in spleens of individual mice are shown for each of three challenge doses. Separate experiments established that an oral challenge of 106 MT110 cells represents the approximate LD50 in control BALB/c mice bearing irrelevant hybridoma tumors.

not show the progressive decline in general health that resulted from tumor growth, and therefore the progression of infection with S. typhimurium was slower (data not shown).

Twelve days after injection of hybridoma cells, when visible tumors had formed in at least 70% of the mice, blood samples were taken from each mouse and serum levels of anti-S. typhimurium or anti-V. cholerae IgA were determined. Mice that had specific serum IgA detectable at a 1:1,000 dilution were retained in the study. The following day, these mice were orally challenged with various doses of wild-type S. typhimurium MT11O or the Agg4 mutant, MT114. Viable bacterial counts in spleens were determined 5 days postchallenge. The data depicted in Fig. 2 show that in a typical experiment, all mice bearing Sal4 IgA hybridoma tumors had sterile spleens 5 days after oral challenge with 107 wild-type S. typhimurium cells. In contrast, spleens of mice bearing Sal4 IgA tumors were consistently infected by the Agg4 mutant. Similarly, both of the two control mice carrying hybridomas secreting anti-V. cholerae IgA antibody had infected spleens after oral challenge with wild-type S. typhimurium (data not shown). In a separate experiment

designed to test the capacity of secreted IgA to defend against very high numbers of S. typhimunum cells in the gut, mice bearing Sal4 IgA hybridoma tumors were challenged orally with 109 cells of strain MT110 or MT114. Although the total numbers of bacteria recovered from the spleens were lower in tumor-bearing mice challenged with the wild-type MT110 strain than in those challenged with the Agg4 mutant MT114, all mice challenged with 109 bacteria were consistently infected (data not shown). Sal4 IgA does not prevent systemic infection after i.p. challenge with S. typhimurium. Having shown that secretion of Sal4 IgA from hybridoma tumors prevented systemic

0

H

0

00

10 1 0a

0

MT 110

+

+

+

MT114 + + + FIG. 3. Systemic infection after i.p. challenge. S. typhimurium recovered from spleens of mice bearing Sal4 hybridoma backpack tumors, 3 days after i.p. challenge with the wild-type strain MT110 (open squares) or the Agg4 mutant MT114 (solid squares). Total CFU in spleens of individual mice are shown for each of three challenge doses. Although the numbers of MT110 organisms recovered from spleens were somewhat reduced compared with the MT114 strain, Sal4 IgA failed to prevent infection of spleens with either wild-type or mutant organisms.

infection after oral challenge with up to 107 S. typhimunum cells, we sought to determine at what stage in the infection pathway the IgA protection occurred. In normal mucosal immunity, blood levels of polymeric IgA are low because most IgA is produced locally by mucosal plasma cells (2). In our mice, in contrast, the hybridoma tumors resulted in abnormally high levels of circulating IgA. Thus, IgA-mediated protection could have occurred on the mucosal surface, within mucosal tissue, or during a later, systemic stage of infection. Since i.p. injection of S. typhimurium mimics the bacteremia that follows mucosal invasion after oral infection, we tested whether Sal4 IgA produced by hybridoma tumors can protect mice after i.p. injection of S. typhimuHum. Mice bearing Sal4 hybridoma tumors were challenged i.p. with either wild-type S. typhimurium MT110 or Agg4 mutant MT114. Even at a low dose of 20 bacteria, S. typhimurium were recovered from the spleens of six of seven mice 3 days after i.p. challenge with either the wild-type or the Agg4 mutant (Fig. 3). Although the absolute number of organisms recovered from the spleen was somewhat lower in mice infected with the wild type than in mice infected with the Agg4 mutant, the numbers of infected animals were equal. Sal4 IgA can prevent infection of Peyer's patch cells after oral challenge with S. typhimurium. The fact that Sal4 IgA did not provide protection against an i.p. challenge with S. typhimunium is consistent with two hypotheses: (i) IgAmediated protection operated by immune exclusion at the mucosal surface, and (ii) IgA in the intestinal tissues inhibited spread of organisms from mucosa to blood. To distin-

MONOCLONAL sIgA PROTECTS AGAINST S. TYPHIMURIUM

VOL. 60, 1992

40( co

a)

0

as 40.

ao

30( DO

to

cn

20(

a0

1 0(

10~~~ aa.

IL

0

a

0 0

1

2

3

a0 4

serum IgA (gg/mi) FIG. 4. Infection of Peyer's patch mucosa after oral challenge. S. typhimunum recovered from Peyer's patch tissue of mice bearing Sal4 IgA hybridoma backpack tumors, 2 days after oral challenge with 107 S. typhimunum MT110 cells. Total CFU in pooled Peyer's patch tissue from individual mice is plotted against the concentration of Sal4 IgA in serum. Of four mice whose levels in serum exceeded 2 pug/ml, three had sterile Peyer's patches.

guish between these possibilities, we tested the ability of Sal4 IgA to prevent S. typhimunum entry into Peyer's patches at early times after oral challenge, and we correlated the number of bacteria in the mucosa with the level of specific IgA in serum and hence in secretions. Mice bearing Sal4 hybridoma tumors were orally challenged with 107 wild-type S. typhimunium cells, and 2 days later, thL number of viable bacteria per milligram of Peyer's patch tissue was determined. Three of four mice whose levels of Sal4 IgA in serum were high had sterile Peyer's patches and spleens (Fig. 4). In contrast, the Peyer's patches of all six mice that had low levels of serum Sal4 IgA showed some degree of infection of Peyer's patch mucosa. Moreover, two of six mice in this group also had viable bacteria in their spleens. Our other results suggest that the spleens of all six would have become infected if more time had elapsed before sacrifice. DISCUSSION The mucosal immune system is thought to provide an early line of defense against both invasive and noninvasive enteric pathogens. In an effort to define more specifically the role of sIgA in this protection, we have examined the ability of a single monoclonal sIgA to protect mice from oral challenge with the highly invasive bacterial pathogen S. typhimunum. We produced this monoclonal IgA by a procedure that involved oral immunization with live attenuated S. typhimurium and then fusion of Peyer's patch lymphoblasts with myeloma cells. One of the hybridomas so obtained (Sal4) was shown to produce polymeric monoclonal IgA antibodies directed against a surface carbohydrate epitope of S. typhimurium. This monoclonal IgA, when delivered into intestinal secretions of nonimmune mice from subcutaneous tumors, protected them against an oral challenge with S. typhimurium. We showed that specific recognition of the bacteria by the IgA was required for this protection by isolating a fully virulent mutant of S. typhimurium that does not produce the epitope and demonstrating that this mutant could infect Sal4 tumor-bearing mice. The hybridoma "backpack" tumor method used here

1791

provides continuous delivery of polymeric monoclonal IgA antibodies into intestinal secretions. The relative amount of specific IgA in secretions is roughly proportional to tumor size and to the level of monoclonal IgA in serum (26). In these mice, a single monoclonal sIgA represents the only form of specific immune protection. We previously used this approach to show that monoclonal sIgA directed against a strain-specific carbohydrate epitope of the surface lipopolysaccharide of V. cholerae was sufficient to protect against a lethal oral dose of these organisms (26). Presumably, the IgA provided protection by preventing colonization of the intestinal mucosal surface by this noninvasive pathogen. Here we show that monoclonal sIgA, directed against a surface epitope of S. typhimurium, can also prevent systemic disease caused by this invasive enteric pathogen. Our criterion for protection by Sal4 IgA was sterility of the spleen after either oral or intraperitoneal challenge with S. typhimunium (Fig. 2). Since monoclonal IgA antibodies produced by the backpack tumors were present not only in the intestinal lumen but also in blood and mucosal tissues,

protection could theoretically have been accomplished

at

any one of several stages. For example, circulating IgA could have opsonized or in some way helped clear the organisms during systemic spread. Alternatively, high levels of polymeric IgA within mucosal tissues could have allowed epithelial invasion but prevented spread from the mucosa into the bloodstream. Finally, secreted monoclonal IgA could have intercepted Salmonella sp. in the intestinal lumen, preventing epithelial contact, M-cell transport, and invasion of the Peyer's patch mucosa. If circulating dimeric IgA were able to prevent or reverse systemic spread, then mice with Sal4 backpack tumors should have shown sterile spleens after i.p. challenge with low numbers of virulent organisms. This was not the case; mice bearing Sal4 tumors were infected by as few as 20 organisms given i.p. The lack of systemic immune protection is consistent with the inability of Sal4 IgA to activate complement and the lack of detectable bacteriocidal activity in vitro. It is possible that IgA-mediated aggregation of organisms in the peritoneal cavity could have decreased somewhat the efficiency of spread of Salmonella sp. after i.p. challenge. This might explain the fact that the escape mutant MT114 was recovered from spleens in higher numbers than MT110. Nevertheless, circulating Sal4 IgA was not sufficient to prevent systemic disease (Fig. 3). Our results are thus consistent with previous studies in which most IgA antibodies were shown to lack bacteriocidal, opsonizing, or complement-fixing activities (12, 15). It is also possible that Sal4 could have prevented spread from the mucosa while allowing entry into the epithelium. S. typhimunum enters the Peyer's patches through M cells (13), and we have observed that these specialized epithelial cells can specifically bind and transcytose immunoglobulins including IgA and IgA immune complexes (25). Thus, S. typhimurium coated with IgA in the lumen could still have been subject to M-cell uptake, but because of the presence of the IgA, may have somehow been prevented from further systemic spread. If so, then organisms should have been consistently recovered from Peyer's patch tissue after oral challenge. However, we observed that three of four mice producing protective levels of Sal4 IgA had sterile Peyer's patches (Fig. 4), indicating either that mucosal entry had not occurred or that Sal4 IgA was able to "arm" mucosal cytotoxic T cells and promote efficient lysis of any organisms that succeeded in entering the mucosa (18, 22). Given our results, it seems most likely that protection was

1792

MICHETTI ET AL.

due primarily to secretion of polymeric monoclonal sIgA into the intestinal lumen and that agglutination of organisms effectively prevented their contact with the mucosal surface. We have not determined whether Sal4 IgA affected the viability of organisms in the intestinal lumen. For example, clearance from the gastrointestinal tract could also have been facilitated by enhanced phagocytosis of sIgA-coated Salmonella sp. by lumenal macrophages or even by IgAenhanced killing by lumenal T lymphocytes (22). These possibilities warrant further study with these and other monoclonal antibodies. If sIgA protects simply by agglutinating live organisms in the intestinal lumen, protection could theoretically be accomplished by sIgA against any abundant surface component. In related studies, we have shown that secretion of monoclonal IgA antibodies directed against abundant microbial surface components protected against mucosal invasion by reovirus (24) and against diarrheal disease caused by V. cholerae (26). Moreover, the surface component does not need to be a virulence factor. This was demonstrated in the present study by the existence of a fully virulent S. typhimurium mutant that lacks the protective epitope. The results presented in this paper indicate that specific sIgA alone, if directed against the bacterial surface and secreted in sufficient amounts, can protect against mucosal invasion and systemic disease caused by S. typhimurium. It follows that individuals who are orally vaccinated with recombinant antigens delivered in live Salmonella vectors might secrete sufficient amounts of anti-Salmonella IgA to prevent subsequent mucosal immunizations with this vector. In our studies, however, sIgA-mediated protection could be overcome by very high oral doses of virulent organisms. Thus, our analysis of the role of sIgA in immunity to Salmonella sp. is relevant not only to understanding what constitutes effective mucosal immune protection against invasive Salmonella sp. but also to the development of effective Salmonella vectors as vaccines. ACKNOWLEDGMENTS We are grateful to Lawrence Mason and Nadine Porta for excellent technical assistance and to Betty Ann Mclsaac for expert preparation of the manuscript. This work was supported by NIH research grants HD17557 and DK21505 (M.R.N.) and A118045 (J.J.M.), NIH Center grant DK34854 (to the Harvard Digestive Diseases Center), research grant FNS 32.30011.90 from the Swiss National Science Foundation (P.M.), National Research Service Award A108245 (M.J.M.), and D'amon Runyon-Walter Winchell Cancer Research Fund Fellowship DRG-1016 (J*M.S.). REFERENCES 1. Brandtzaeg, P. 1989. Overview of the mucosal system. Curr. Top. Microbiol. Immunol. 146:13-25. 2. Brandtzaeg, P., L. M. Sollid, P. S. Thrane, D. Kvale, K. Bjerke, H. Scott, K. Kett, and T. 0. Rognum. 1988. Lymphoepithelial interactions in the mucosal immune system. Gut 29:1116-1130. 3. Cancellieri, V., and G. M. Fara. 1985. Demonstration of specific IgA in human feces after immunization with live Ty2la Salmonella typhi vaccine. J. Infect. Dis. 151:482-484. 4. Carter, P. B., and F. M. Collins. 1974. The route of enteric infection in normal mice. J. Exp. Med. 139:1189-1203. 5. Chau, P. Y., R. S. W. Tsang, S. K. Lam, J. T. LaBrooy, and D. Rowley. 1981. Antibody response to the lipopolysaccharide and

protein antigens of Salmonella typhi during typhoid infection. Clin. Exp. Immunol. 46:515-520. 6. Childers, N. K., M. G. Bruce, and J. R. McGhee. 1989. Molecular mechanisms of immunoglobulin A defense. Annu. Rev.

INFECT. IMMUN. Microbiol. 43:503-536. 7. Edelman, R., and M. M. Levine. 1986. Summary of an international workshop on typhoid fever. Rev. Infect. Dis. 8:329-349. 8. Finlay, B. B., B. Gumbiner, and S. Falkow. 1988. Penetration of Salmonella through a polarized Madin-Darby canine kidney epithelial cell monolayer. J. Cell Biol. 107:221-230. 9. Gorbach, S. L. 1989. Infectious diarrhea, p. 1191-1232. In M. H. Sleisenger and J. S. Fordtran (ed.), Gastrointestinal disease: pathophysiology, diagnosis, management. The W. B. Saunders Co., Philadelphia. 10. Hohmann, A. W., G. Schmidt, and D. Rowley. 1978. Intestinal colonization and virulence of Salmonella in mice. Infect. Immun. 22:763-770. 11. Kantele, A., H. Arvilommi, and I. Jokinen. 1986. Specific immunoglobulin-secreting human blood cells after peroral vaccination against Salmonella typhi. J. Infect. Dis. 153:1126-1131. 12. Kerr, M. A. 1990. The structure and function of human IgA. Biochem. J. 271:285-296. 13. Kohbata, S., H. Yokobata, and E. Yabuuchi. 1986. Cytopathogenic effect of Salmonella typhi GIFU 10007 on M cells of murine ileal Peyer's patches in ligated ileal loops: an ultrastructural study. Microbiol. Immunol. 30:1225-1237. 14. Levine, M. M., and R. Edelman. 1990. Future vaccines against enteric pathogens. Infect. Dis. Clin. N. Am. 4:105-121. 15. Mestecky, J. 1988. Immunobiology of IgA. Am. J. Kidney Dis. 12:378-383. 16. Miller, S. I., A. M. Kukral, and J. J. Mekalanos. 1989. A two-component regulatory system (phoP and phoQ) controls Salmonella typhimunum virulence. Proc. Natl. Acad. Sci. USA 86:5054-5058. 17. Miller, S. I., and J. J. Mekalanos. 1990. Constitutive expression of the PhoP regulon attenuates Salmonella virulence and survival within macrophages. J. Bacteriol. 172:2485-2490. 18. Nencioni, L., L. Villa, D. Boraschi, B. Berti, and A. Tagliabue. 1983. Natural and antibody-dependent cell-mediated activity against Salmonella typhimunum by peripheral and intestinal lymphoid cells in mice. J. Immunol. 130:903-907. 19. Parsot, C., E. Taxman, and J. J. Mekalanos. 1991. ToxR regulates the production of lipoproteins and the expression of serum resistance in Vibrio cholerae. Proc. Natl. Acad. Sci. USA 88:1641-1645. 20. Sarasombath, S., N. Banchuin, T. Sukosoll, B. Rungpitarangsi, and S. Manasatit. 1987. Systemic and intestinal immunities after natural typhoid infection. J. Clin. Microbiol. 25:1088-1093. 21. Solari, R., and J. P. Kraehenbuhl. 1987. Receptor-mediated transepithelial transport of polymeric immunoglobulins, p. 269-298. In M. C. Neville and C. W. Daniel (ed.), The mammary gland. Development, regulation and function. Plenum Press, New York. 22. Tagliabue, A., L. Villa, D. Boraschi, G. Peri, V. deGori, and L. Nencioni. 1985. Natural anti-bacterial activity against Salmonella typhi by human T4+ lymphocytes armed with IgA antibodies. J. Immunol. 135:4178-4182. 23. Takeuchi, A. 1967. Electron microscope studies of experimental Salmonella infection. I. Penetration into the intestinal epithelium by Salmonella typhimurium. Am. J. Pathol. 50:109-136. 24. Weltzin, R., L. Morrison, L. S. Winner III, B. N. Fields, J. P. Kraehenbuhl, and M. R. Neutra. 1989. In vivo secretion of specific monoclonal IgA protects mice against an epitheliallytransported enteric virus, abstr. 1608. J. Cell Biol. 109:295a. 25. Weltzin, R. A., P. Lucia Jandris, P. Michetti, B. N. Fields, J. P. Kraehenbuhl, and M. R. Neutra. 1989. Binding and transepithelial transport of immunoglobulins by intestinal M cells: demonstration using monoclonal IgA antibodies against enteric viral proteins. J. Cell Biol. 108:1673-1685. 26. Winner, L. S., Ill, R. A. Weltzin, J. J. Mekalanos, J. P. Kraehenbuhl, and M. R. Neutra. 1991. New model for analysis of mucosal immunity: intestinal secretion of specific monoclonal immunoglobulin A from hybridoma tumors protects against Vibrio cholerae infection. Infect. Immun. 59:977-982. 27. Woodward, M. P., W. W. Young, Jr., and R. A. Bloodgood. 1985. Detection of monoclonal antibodies specific for carbohydrate epitopes using periodate oxydation. J. Immunol. Methods 78:143-153.