Effect of in situ expression of human interleukin-6 on antibody responses against Salmonella typhimurium antigens

Abstract

In an attempt to trigger increased mucosal secretory immune responses against bacterial surface antigens, we constructed an optimized human interleukin (hIL)-6-secreting Salmonella typhimurium strain (X4064(pCH1A+pYL3E)), utilizing the hemolysin (Hly) exporter for secretory delivery of a functional hIL-6-hemolysin fusion protein (hIL-6-HlyA_s). Through stable introduction of a second hIL-6-HlyA_s expression plasmid (pYL3E) in the previously described X4064(pCH1A) strain, hIL-6-HlyA_s secretion efficiencies were increased by at least 10-fold. As pCH1A in the parental strain, pYL3E was stable in vitro in the absence of antibiotic selection and in vivo neither did plasmids interfere in their stabilities. Increased hIL-6-HlyA_s expression did not adversely interfere with bacterial growth. Comparative immunization experiments in mice with oral application of the different hIL-6-secreting strains revealed that increased in situ hIL-6-production influenced systemic antibody responses against Salmonella antigens but had no marked effect on mucosal responses. In mice immunized with X4064(pCH1A+pYL3E) significantly higher sera IgG and IgA titers for lipopolysaccharide (LPS) were found compared to mice immunized with X4064(pCH1A) and a hIL-6-negative control strain. Higher sera antibody titers were accompanied by increased numbers of IgG- and IgA-specific antibody-secreting cells in spleens and Peyer's patches, respectively. These data suggest that systemic antibody responses against Salmonella LPS are largely effected by IL-6 and, moreover, the amount and the cellular location of recombinantly expressed IL-6 appears to be crucial for enhancement of immune responses.

Hemolysin, Immunomodulation, Interleukin-6, Protein secretion, Salmonella

1 Introduction

Over the past 20 years, live attenuated Salmonella strains have been extensively studied as carriers for heterologous vaccine antigens and effector proteins with immunomodulatory and immunoadjuvancy potential, including mucosal adjuvant proteins and cytokines [1–3]. Although those strains experience a drastic reduction in their pathogenicity, remaining virulence is still sufficient to trigger efficiently immune responses against recombinantly expressed foreign antigens. Generally immune responses are predominantly directed against surface antigens of the Salmonella carrier cell, but, depending on the nature of the antigen, protective immunity can also be induced against the foreign protein delivered by the Salmonella cell.

As mucosal pathogens, Salmonella vaccine carriers have the potential to induce both systemic and mucosal immune responses [4]. Studies in mice have shown that a variety of immune responses, including cytotoxic T cells, systemic antibody responses, as well as delayed-type hypersensitivity, are induced by attenuated Salmonella typhimurium strains expressing recombinant antigens [5]. Less frequently mucosal immune responses, e.g. mucosal secretory IgA responses, have been reported for a variety of antigens [6]. The type of immunity induced depends heavily on the nature of the antigen, but in most cases systemic responses dominate over mucosal responses.

One approach by which mucosal responses may be selectively enhanced is through co-expression of certain immunocompetent Th2-type cytokines. The feasibility of this approach has been previously shown for IL-5 using viral and bacterial vaccine carrier systems, where an increased mucosal immunoreactivity against the co-expressed vaccine antigen could be demonstrated [7–9]. Similar approaches with IL-4 and IL-6 in attenuated Salmonella strains, however, did not confirm the postulated immunomodulatory effect [10,11]. This may be partly due to the fact that plasmid-encoded expression systems, which lead to an intracellular arresting of the cytokines, were used. In this case, the in vivo accessibility of the recombinant cytokine depends solely on the unspecific release of soluble protein in the course of the desintegration of the Salmonella cells. Furthermore, such cytokines tend to aggregate and form inclusion bodies when overexpressed intracellularly in bacterial cells, further reducing the amount of bioactive cytokine [12–14]. Moreover, IL-6, like IL-5 and IL-4, requires correct disulfide bond formation for development of full biological activity, which, because of the reducing milieu, is impaired in the bacterial cytoplasm [15–17]. Thus, efficient extracellular cytokine delivery in Salmonella vaccine strains should have a number of striking advantages which eventually contribute to an improved effect or protein action.

Secretory expression of foreign proteins in Escherichia coli and Salmonella can be achieved with the plasmid-borne α-hemolysin (Hly) secretion system [18–20]. As previously demonstrated for various antigens, translocation of a foreign protein across the bacterial cell envelope by the Hly transport machinery depends on the C-terminal Hly secretion signal HlyA_s genetically fused to the foreign protein sequence. This secretion signal is not released during the export process, resulting in secretion of a fusion protein, which permanently carries a C-terminal extension of about 60 amino acid residues. Because of its low immunogenicity this C-terminal peptide does not markedly contribute to the overall B- and T-cell responses against the hybrid vaccine antigens [19]. However, the C-terminal HlyA_s extension could influence bioactivity of secreted effector proteins, such as cytokines.

IL-6 is a multifunctional cytokine, which also plays a crucial role in B-cell terminal differentiation and development of secretory IgA responses at mucosae [21]. To investigate the effect of extracellular delivery of a Th2-type cytokine on antibody responses, particularly along mucosal surfaces, we previously fused the cDNA encoding mature human IL-6 to the HlyA_s secretion signal, proposing that the additional IL-6 synthesis leads to a selective increase in humoral IgA responses. Recombinant E. coli and S. typhimurium strains expressing the hIL-6-hlyA_s gene as part of a reconstructed natural hly operon secrete relevant amounts of a bioactive hIL-6-HlyA_s hybrid protein into the culture supernatant [22]. More detailed structure-function analysis of the purified protein revealed that the C-terminal fusion part apparently has no impact on the proper protein folding and correct disulfide bridge formation, resulting in a secretion-competent hIL-6 fusion protein, indistinguishable in its bioactivity from mature recombinant hIL-6 [23]. Immunization studies in mice with this hIL-6-expressing Salmonella strain, however, revealed no significant enhancement of antibody responses against bacterial antigens (unpublished results). Assuming that secreted hIL-6 amounts were too low to trigger an in vivo effect, we pursued different strategies to increase hIL-6-HlyA_s secretion efficiencies in the previously constructed S. typhimurium vaccine strains [23].

Here, we report the construction of an optimized S. typhimurium strain, with an about 10-fold improved secretion efficiency for hIL-6-HlyA_s. Comparative immunization studies in mice with this strain demonstrated that the in vivo synthesis of hIL-6 induces increased systemic IgG and IgA responses against bacterial lipopolysaccharide (LPS) upon oral administration of the vaccine strains, suggesting that IL-6 plays a crucial role in the development of LPS-specific antibody responses. Moreover, data support that both hIL-6 amounts and cellular location of the recombinantly expressed hIL-6 are critical for maximal immunomodulatory action of the effector protein.

2 Materials and methods

2.1 Bacterial strains, culture conditions and plasmids

Plasmids pCH2G and pCH1A and S. typhimurium strain X4064 have been previously described [22], while plasmids pTrc99a and pACYC184 were from Amersham Biosciences and New England Biolabs, respectively. Plasmid-bearing strains of S. typhimurium and E. coli XL1-blue were grown at 37°C in Luria–Bertani (LB) broth, supplemented with 100 µg ml⁻¹ ampicillin or 50 µg ml⁻¹ chloramphenicol. If required, isopropyl-β-d-thiogalactopyranoside (IPTG) was added to E. coli cultures for gene induction. Transformation of S. typhimurium and E. coli was achieved by standard electroporation protocol [24]. For co-transformation, cells were electroporated with a mixture of both plasmids. For immunization, Salmonella strains were grown at 37°C in 100 ml of LB broth with ampicillin to an OD₅₇₈ of 1.2–1.5. Cells were washed once resuspended in 3 ml of 5%(w/v) sodium bicarbonate. Bacterial suspensions were adjusted to 5 OD₅₇₈/ml, which corresponds to approximately 1×10¹⁰ cells ml⁻¹, and kept on ice until administration to the mice. To recover Salmonella from spleen or liver, organ samples were homogenized in phosphate-buffered saline (PBS) using a Wheaton homogenator and plated onto LB or xylose-lysine-deoxycholate agar (Merck Eurolab) with the appropriate antibiotics.

2.2 LPS and cell homogenate preparation

Cell homogenates were prepared from stationary-phase S. typhimurium X4064 cells by standard sonication procedure. Homogenates were adjusted to a protein concentration of 1 mg ml⁻¹ by dilution with PBS and stored frozen in aliquots.

LPS was isolated from freeze-dried S. typhimurium X4064 cell pellets following the hot phenol method of [25] with modifications. Typically, 1.5 g of a freeze-dried cell pellet was resuspended in 53 ml of water and then an equal volume of prewarmed (65°C) water-saturated phenol was added. The mixture was kept at 65°C and vigorously mixed for 15 min. For phase separation, the mixture was cooled down in ice and centrifuged at 3000×g and 4°C for 30 min. After removal of the water phase, the phenol phase was re-extracted with the same volume of water and the combined water extracts extensively dialyzed against water. The freeze-dried crude LPS preparation was redissolved in 10 ml of PBS. For further purification, crude LPS suspension was adjusted to 10 mM MgCl₂ and incubated for 30 min at 37°C with 10 µg ml⁻¹ of DNase I and RNase A each. LPS was subsequently centrifuged for 3 h at 105,000×g and 4°C and resuspended with vigorous mixing in water. The LPS suspension was recentrifuged twice and after freeze-drying suspended in PBS at a final concentration of 10 mg ml⁻¹. The purified LPS preparation was essentially free of any protein contaminations and contained less than 4% of nucleic acids as judged from silver-stained sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) analysis [26] and absorption at 260 nm, respectively. Yields of purified LPS were in the range of 20 mg per g of freeze-dried S. typhimurium cell pellet.

2.3 Plasmid constructions

Recombinant DNA techniques were performed using standard procedures [24] with DNA-modifying enzymes used in accordance with the supplier's instructions (New England Biolabs). To overexpress hIL-6-HlyA_s fusion protein, we previously constructed plasmid pYL1E by inserting a polymerase chain reaction (PCR)-generated 852-bp fragment encoding the complete hIL-6-hlyA_s sequence between the Nco I and Hin dIII site of pTrc99a vector (unpublished data). In pYL1E the hIL-6 fusion gene is under the transcriptional control of the P_trc promoter and a downstream terminator, forming an hIL-6-hlyA_s expression cassette. The cassette was released from pYL1E by digestion with Sca I and Sfo I and the resulting 2072-bp fragment ligated into the Eco RV site of pACYC184. Plasmid clones with fragment oriented in the opposite transcriptional orientation of the Cm resistance gene were identified by characteristic Eco RI and Nco I restriction pattern and further analyzed for IPTG-inducible expression of the fusion gene in E. coli. The resulting plasmid construct pYL3E was introduced either alone or in combination with pCH1A and pCH2G, respectively, into S. typhimurium X4064 or E. coli.

2.4 SDS–PAGE and Western blot analysis

SDS–PAGE (12% PAA gels) and Western immunoblotting were performed as described earlier [3], using hIL-6-specific IgG₁-type mAb (R&D Systems) in a 1:250 dilution for immunodetection. For colorimetric detection, goat anti-mouse IgG (whole molecule) conjugated to alkaline phosphatase (Sigma-Aldrich) was applied in a 1:2000 dilution, followed by standard NBT/BCIP color development. To increase the detection sensitivity, blots were subsequently probed for a 2-h period with a 1:2000 dilution of rabbit anti-mouse IgG (whole molecule)-horseradish peroxidase conjugate followed by chemoluminescence detection as described earlier [23]. For preparation of cell-free culture supernatant, bacteria were pelleted from culture at 4300×g for 20 min at 4°C and supernatants concentrated under vacuum prior to sample denaturing and loading to the gel.

2.5 Sera and lavage samples

Blood samples were either collected from the tail veins or, for terminal bleeding, from the retro-orbital plexus. For sera preparation, clotted blood samples were centrifuged at 18,800×g and 4°C for 15 min and sera supernatants stored in aliquots at −20°C.

Gut lavages were obtained by flushing the small intestine with 5 ml of cold PBS, containing 1 mg ml⁻¹ bovine serum albumin (BSA), 0.1%(w/v) NaN₃, 0.05%(v/v) Tween 20, 0.1 mg ml⁻¹ soybean trypsin inhibitor (Serva), and 1 mM phenylmethyl sulfonylfluoride (PMSF), the latter added immediately prior to the use of the solution. To prepare fecal extracts, washes were heavily vortexed and centrifuged at 7650×g and 4°C for 10 min. The lavage fluids were stored frozen at −20°C. Prior to titer measurement, lavages were re-centrifuged to remove any precipitate.

2.6 Isolation of splenocytes and Peyer's patches cells

For the preparation of splenocytes, about half of a spleen was washed once with PBS and then passaged through a 70-µm-pore-size cell strainer (Falcon). Erythrocytes were lyzed by 5-min incubation in 5 ml of 145.5 mM NH₄Cl/17 mM Tris-HCl (pH 7.2). Remaining single-cell suspensions were washed once with PBS and splenocytes finally resuspended in 5 ml of complete RPMI medium (RPMI 1640 medium with glutamin supplemented with 10%(v/v) fetal calf serum, 50 U ml⁻¹ penicillin, 50 µg ml⁻¹ streptomycin, 24 µM 2-mercaptoethanol). From each mouse between 5 and 7 Peyer's patches were collected from the small intestinal wall and single-cell suspensions prepared from the complete Peyer's patches as described for the spleen cells. Cells were washed once and resuspended in complete RPMI medium. Prior to the ELISPOT procedure, numbers of live cells were determined upon trypan blue staining.

2.7 IgG- and IgA-specific ELISA

Specific antibody titers were measured in sera and lavage samples following the same sandwich ELISA protocol. 96-well plates (Falcon) were coated overnight at 4°C with S. typhimurium cell homogenate or LPS (100 µl/well) diluted in PBS to a concentration of 5 µg ml⁻¹ and 20 µg ml⁻¹, respectively. For blocking, plates were incubated for 1 h at 37°C with 200 µl/well of 5%(w/v) BSA in PBS. Plates were washed four times with washing buffer (0.05%(w/v) BSA/0.05% Tween 20 in PBS) and, for further use, dried and stored at −20°C. Upon rehydration and washing once with washing buffer, sera and lavage samples were serially diluted with 2%(w/v) BSA/0.05% Tween 20 in PBS (dilution buffer) in the wells (100 µl/well) and plates incubated for 2 h at 37°C. Plates were washed four times and, for detection of specific immunoglobulin subtypes, incubated with a 1:2500 dilution of peroxidase-conjugated rabbit anti-mouse IgG or anti-mouse IgA (Sigma-Aldrich) in dilution buffer for 2 h at 37°C. Antibody binding was calorimetrically detected by 30-min incubation with 2 mg ml⁻¹O-phenylenediamine and 0.03%(v/v) H₂O₂ in 17 mM citrate buffer. Reactions were stopped with 50 µl/well of 4.5 M H₂SO₄ and plates read in a Dynex (MRX) ELISA reader at 490 nm with reference wavelength set to 620 nm. Samples were measured at least in duplicate from individual dilutions.

To compensate for the variance in the plate reactivity and quantify ELISA titers, on each plate six serial 1:2 (lavages) and 1:3 dilutions (sera) of a standard sera pool, derived from the terminal bleeding of eight immunized mice, were loaded. The results were expressed as antibody units calculated from the linear range of the standard curve obtained from the sera pool given an arbitrary antibody titer of 100,000 IgG U ml⁻¹ and 10,000 IgA U ml⁻¹. The values for a minimum of four sample dilutions falling into the linear range of the standard curve were calculated and the geometric mean taken as antibody titer of the sample.

To measure ELISA titers to in situ synthesized hIL-6 in mouse sera plates were coated with 20 ng per well of carrier-free recombinant hIL-6 (R&D Systems) and probed for IgG antibodies.

2.8 ELISPOT assay

The numbers of Ig subtype-specific antibody-secreting cells (ASCs) present in spleen and Peyer's patches of immunized mice were determined by ELISPOT techniques [27], using Biodyne B membranes in 96-well flat-bottom plates (Nunc). Plates were coated with 2.5 µg per well of crude LPS and, after blocking with complete RPMI medium, splenocytes and Peyer's patches cells, respectively, were seeded in complete RPMI medium at the indicated cell densities. Plates were incubated overnight at 37°C, 5% CO₂ and membrane-bound antibodies detected by incubation with subtype-specific anti-mouse Ig-alkaline phosphatase conjugates followed by standard colorimetric detection. Further details on the ELISPOT procedure and the enumeration of the ASCs will be published elsewhere.

2.9 Immunization experiments

For the immunization experiments 8–10 week old female inbred BALB/c mice (Charles River Germany) were randomly distributed into the different groups. Mice were infected orally with a dose of approximately 2×10⁹Salmonella in 200 µl with a gastric lavage needle 10 min after administration of 100 µl of 5%(w/v) sodium bicarbonate. Prior to the immunization procedure, mice were kept for 6–8 h without water. Applied bacterial numbers were verified by viable counting. Mice were boosted on days 28 and 42 with the same bacterial dose administered in the primary immunization. During the experiments, mice were fed commercial mouse chow and water ad libitum. Two days before each immunization mice were anesthetized lightly and blood samples taken from the tail vein. At the end of the experiments, mice were anesthetized and larger blood volumes taken from the retro-orbital plexus. Thereafter, mice were terminated by cervical dislocation and, after lavaging the lung, liver, spleen, and small intestines removed and processed as described above. The animal experiments reported were conducted according to the German animal protection law and approved by the ethics committee of the Regierungspräsidium Freiburg.

2.10 Statistics

The P values were determined using Student's two-tailed t-test (Excel; Microsoft Corporation). Immunoblots and X-ray films were photographed with a digital camera. Electronic images were mounted in Adobe Photoshop and finally integrated in Microsoft Power Point.

3 Results

3.1 Construction of S. typhimurium strains with improved hIL-6-HlyAs secretion

From previous studies we knew that the expression of the hIL-6-hlyA_s gene as an integral part of the original hly operon was the ‘bottleneck’ in secretory hIL-6-HlyA_s production in both E. coli and S. typhimurium[3]. Data suggested that an increase in the hIL-6-hlyA_s gene copy number should eventually boost secretory hIL-6 expression. Based on that, we developed an improved Hly export cloning system that is composed of two vectors, one in which the hIL-6 fusion gene is combined with the genes encoding the Hly transporter apparatus (pCH1A) and a second in which hIL-6-hlyA_s is under transcriptional control of the P_trc promotor (pYL3E). To adjust the hIL-6-HlyA_s expression to a medium level and thus minimize the risk of intracellular precipitation of the hIL-6 fusion protein and avoid plasmid stability problems we used a rather low copy-number vector as carrier for the additional hIL-6-HlyA_s gene copy. Plasmids pCH1A and pYL3E as well as the corresponding control plasmids were simultaneously introduced into E. coli and S. typhimurium by electroporation.

To evaluate the effect of the additional hIL-6-hlyA_s gene copy on the fusion protein expression and secretion levels, we compared the cellular and secreted hIL-6-hlyA_s amounts of different hIL-6-secreting E. coli strains by qualitative Western blot analysis. As documented in Fig. 1, E. coli (pCH1A) gave a significant immunoreactive band, corresponding to a hIL-6HlyA_s concentration of around 25 µg l⁻¹ culture supernatant. Upon introduction of the hIL-6-hlyA_s expression plasmid pYL3E, secreted fusion protein amounts increased by an estimated 10-fold. The E. coli strain carrying plasmid pYL3E only did not reveal any significant full-length hIL-6HlyA_s band, demonstrating that the increased hIL-6-HlyA_s accumulation in the culture supernatant was due to Hly-mediated secretion rather than unspecific cell lysis caused by the higher expression levels. However, extended culturing time led to appearance of small amounts of a truncated hIL-6HlyA_s protein in both pYL3E-bearing E. coli strains. Expectedly, the E. coli strains carrying the control plasmid pCH2G and pACYC184 (data not shown), respectively, did not show any immunoreactive bands in the culture supernatants. Comparative Western blot analysis of the corresponding cell lysates revealed a comparably strong hIL-6-HlyA_s band for both E. coli (pCH1A+pYL3E) and E. coli (pYL3E) (data not shown). This observation confirmed that only a small part of the synthesized hIL-6 fusion protein becomes actually secreted while the vast majority of the protein remains within the cells. Our own data and those published by other researchers support the assumption that the Hly secretion system functions equally well in E. coli and S. typhimurium[18]. Therefore, it appeared reasonable to expect a similar increase in secretory hIL-6-HlyA_s expression in a S. typhimurium strain carrying the two-plasmid Hly expression system (see Fig. 2B), progressing with comparative immunization studies in mice.

Figure 1

Comparative Western blot analysis of hIL-6-HlyA_s secretion in different plasmid-bearing E. coli strains. E. coli strains harboring plasmids pCH1A (1A), pYL3E (3E), pCH1A+pYL3E (1A+3E), and pCH2G (2G) were grown under similar conditions. After 5, 7, 9, and 24 h cell-free supernatants were prepared from culture samples and analyzed by hIL-6-specific Western immunoblotting followed by colorimetric detection of antibody binding. 200 µl of culture supernatant was loaded in each lane. M, pre-stained standard proteins; S, mature hIl-6 (5 ng). The upper arrow marks the full-length hIL-6-HlyA_s protein.

Figure 2

In vivo stability of hIL-6-HlyA_s-secreting S. typhimurium strains. Amp- and Amp/Cm-resistant Salmonella clones recovered from organ samples of immunized mice were analyzed for the presence of plasmids and secretory hIL-6-HlyA_s expression. A: Agarose gel (0.8%) of isolated plasmids. The arrows mark the position of plasmid pCH1A and pYL3E, respectively. Additional DNA bands reflect multimeric forms of pCH1A. Lanes: 1, 1-kb ladder standard; 2+3, isolates from X4064(pCH2G)-immunized mice; 4+5, isolates from X4064(pCH1A)-immunized mice; 6+7, isolates from X4064(pYL3E+pCH1A)-immunized mice; 8, mixture of purified plasmids pYL3E and pCH1A. B: Western blot analysis. 25-µl volumes of stationary-phase culture supernatants of the different Salmonella isolates were separated in a 12% SDS–PAA gel and after transfer to nitrocellulose the membrane probed with hIL-6-specific mAb followed by chemoluminescence detection of bound antibodies. The arrows indicate the full-length protein and the major degradation product of hIL-6-HlyA_s. Lanes: 1+2, isolates from X4064(pCH2G)-immunized mice; 3+4, isolates from X4064(pCH1A)-immunized mice; 5+6, isolates from X4064(pYL3E+pCH1A)-immunized mice; 7+8, mature hIL-6 standard (0.25 and 0.5 ng).

3.2 In vitro and in vivo stability of improved hIL-6-HlyAs-secreting strains

Comparative growth analysis over 30-h periods revealed that the additional hIL-6-HlyA_s synthesis directed by plasmid pYL3E did not adversely affect the bacterial replication rate (data not shown). The stability of S. typhimurium strains carrying plasmid pYL3E and the two plasmids YL3E and pCH1A as well as the control strains was determined by growth in vitro without antibiotics. X4064 maintained pYL3E with a segregation rate of less than 10% for approximately 90 generations, and thus revealed a similar stability as that observed for pCH1A (data not shown). A similar stability was found for the two plasmids in X4064(pYL3E+pCH1A), indicating that homologous recombination between the two hIL-6-hlyA_s gene copies appeared not to be a problem.

To evaluate the in vivo stability of the strains, antibiotic-resistant Salmonella cells recovered from spleen and liver of immunized mice 11 and 17 days after the second booster immunization were analyzed for the presence and integrity of the plasmids. As documented in Fig. 2A for selected Salmonella clones, all analyzed colonies carried plasmids of the appropriate sizes. The functionality of the plasmids was confirmed by Western blot analysis of the culture supernatants. HIL-6-specific immunoreactive bands corresponding to the size of hIL-6-HlyA_s were found in all these clones. Moreover, significantly higher levels of hIL-6-HlyA_s were detected in culture supernatants of Salmonella isolates recovered from mice immunized with X4064(pYL3E+pCH1A), confirming the high in vitro and in vivo stability of this strain (Fig. 2B).

3.3 Immunogenicity of secreted human IL-6 protein

To be able to differentiate immunologically between recombinant IL-6 produced by administered Salmonella strains and endogenous mouse IL-6 we chose the human protein variant. Human and mouse IL-6 reveal similar biological activity in the mouse system and therefore should develop the same biological effect. Since mouse and human IL-6 share only 40% sequence identity, one could expect that hIL-6 is immunogenic in mice. To ensure that not an immune response to in situ synthesized hIL-6-HlyA_s is interfering with the expected immunomodulatory effect we measured hIL-6-specific sera IgG titers in immunized mice. Even 8 weeks after the primary immunization we could not detect any significant antibody titers in mice immunized with any of the hIL-6-producing S. typhimurium strains, demonstrating that within the experimental timeframe no immunity to recombinant IL-6 occurred (data not shown).

3.4 Effect of IL-6 expression on systemic anti-Salmonella immune responses

From previous immunization experiments with S. typhimurium X4064 expressing other recombinant vaccine antigens we knew that in BALB/c mice maximal humoral immune responses to Salmonella antigens required two booster immunizations (unpublished data). Based on that information, we designed the immunization protocol in such a way that after primary immunization mice received two booster immunizations in 2-week intervals before final assessment of systemic and secretory immune responses.

To compare final immune responses to the different IL-6-expressing Salmonella strains, we initially analyzed IgA- and IgG-specific sera titers in individual mice of different groups. As shown in Fig. 3A, antibody titers measured against whole-cell homogenates were relatively consistent for the mice within the same group. There were no significant differences in the antibody titers among the groups, suggesting that IL-6 production had no impact on overall immune responses to Salmonella carrier. We then compared antibody titers to LPS as a defined T-cell-independent antigen (Fig. 3B). In contrast to sera titers to whole Salmonella cells both LPS-specific IgA and IgG titers revealed a large mouse-to-mouse variation within each group, indicating that some mice hyper-responded to the LPS antigen while others developed comparably weak immune responses. This extreme variation was independent of the strain used for the immunization. Despite the large fluctuation in individual immune responses statistical comparison of the titers revealed a significant difference (P<0.05) in the LPS-specific IgA titers between the group of mice immunized with X4064(pCH1A+pYL3E) and the group immunized with the control strain X4064(pCH2G). On the contrary, no statistically significant differences between the groups were seen for the IgA and IgG titers measured against whole cell homogenate.

Figure 3

Ig subtype-specific serum antibody levels to Salmonella cell homogenate (A) and purified LPS (B). IgA- (open symbols) and IgG-specific antibody titers (filled symbols) were measured 11 and 17 days, respectively, after second booster immunization. For IgG titer measurement, sera samples were diluted from 1:300 to 1:72,900 in serial one to three dilution steps, while dilution range for IgA measurement started at 1:100. Duplicate samples were measured and averages of four independent titer calculations are shown for individual sera. Mice numbers per group were six to eight.

For a more detailed comparison of the sera titers between the groups, sera samples of hyper-responding mice were omitted from the analysis. As shown in Fig. 4, mean IgA and IgG titers against whole Salmonella antigens increased in all groups by 2–2.5-fold and about 5-fold, respectively, after the two booster immunizations. No significant differences were seen between the mice immunized with any of the IL-6-producing Salmonella strains and the mice immunized with the X4064(pCH2G) control strain. Compared to the control group, LPS-specific IgA responses were about four-fold higher in the mice immunized with X4064(pCH1A+pYL3E) after both the primary and the second booster immunization, suggesting that hIL-6 secretion as well as secreted hIL-6 levels are crucial for an immuno-adjuvant effect (Fig. 4B). Mice immunized with X4064(pCH1A) or the strain with the intracellularly arrested hIL-6 revealed a weak increase in IgA responses, which did not reach statistical significance due to a large variation. LPS-specific IgG titers were even further up in mice given X4064(pCH1A+pYL3E), resulting in an about eight-fold increase over the mean titer in the control group. On the other hand, X4064(pCH1A)- and X4064(pYL3E)-immunized mice revealed only about 2.5-fold increases in their LPS-specific IgG titers. Generally, booster immunizations resulted in a significantly larger increase in LPS-specific IgG titers than in IgA titers.

Figure 4

Mean serum antibody titers in mice immunized with different IL-6-producing S. typhimurium X4064 strains. A: Ig subtype-specific responses to whole cell homogenate. B: Ig subtype-specific responses to LPS. Titers represent the mean of sera from six mice (X4064(pCH1A+pYL3E)) and four mice (remaining groups), respectively. Sera samples were collected prior to infection (filled bars), at day 26 after primary infection (hatched bars), and at days 11 and 17, respectively, after second booster immunization (open bars). The numbers are based on duplicate sample measurements and averages of four independent titer calculations for individual sera. Error bars indicate the S.E.M. in each group. Asterisks indicate significant differences (P<0.05) to the control group immunized with X4064(pCH2G) at day 26 and at the end of the experiment.

3.5 Modulation of Ab-producing cell populations in spleen and Peyer's patches

To confirm the IL-6-mediated immuno-enhancing effects we additionally examined spleen and Peyer's patches in immunized mice after second booster immunization for cells secreting anti-LPS antibodies. As documented in Fig. 5, a two-fold increase in the numbers of IgG-producing cells in the mice immunized with X4064(pCH1A+pYL3E), compared to the mice infected with X4064(pYL3E), occurred. Furthermore, IgG-specific ASCs were significantly more in X4064(pCH1A+pYL3E)-immunized mice than in mice infected with the control strain. In contrast to the sera antibody titers, IgA-specific ASC numbers were in the same range for the mice infected with any of the IL-6-producing strains and the control strain (Fig. 5). Compared to the splenocytes, in Peyer's patch cells IgA-specific ASCs were slightly increased over IgG-specific ASCs. Although ASC numbers for Peyer's patches were generally higher, no significant differences were found between the various groups for IgG and IgA (data not shown).

Figure 5

LPS-specific ELISPOTS in spleens of immunized mice. Splenocytes were prepared from animals after the second booster immunization and 5×10⁵ cells seeded per well in the ELISPOT assay. IgA- (filled bars) and IgG-specific ASCs (open bars) represent the mean ASC number from six mice (X4064(pCH1A+pYL3E)) and four mice (remaining groups), respectively. The numbers are based on triplicate sample measurements and error bars indicate the S.E.M. within each group. The asterisk indicates a significant difference to the X4064(pCH2G)-immunized control group (P<0.05).

3.6 Effect of IL-6 expression on secretory anti-Salmonella immune responses

To ensure that antibody titers measured in feces extracts represent secreted antibodies and are not caused by blood-borne antibody contaminations, we initially compared the ratio of IgG and IgA levels in sera and lavage samples from mice immunized with the IL-6-negative strain X4064(pCH2G). The average ratio of IgA to IgG was 1 to 11.5 for sera, while the ratio in gut lavages was increased to 3.5 to 1.

As for the sera titers against whole Salmonella antigens, there was no difference in the mean IgG and IgA sera titers between the groups immunized with the various IL-6-producing strains and the X4064(pCH2G) control strain, (data not shown). In contrast to the sera titers, LPS-specific IgA titers in gut lavages of mice given X4064(pCH1A+pYL3E) were slightly (maximal two-fold) increased over the mean titer measured in X4064(pCH1A)- and X4064(pYL3E)-immunized mice (Fig. 6). Furthermore, LPS-specific IgG titers were at the same level in both groups immunized with the hIL-6-secreting strains and were about two-fold higher than in the group given the Salmonella strain with the intracellular arrested hIL-6. Neither for IgA nor IgG differences in the LPS-specific titers among the various groups bore a close statistical examination due to the small sample number and the large variance. These results suggest that under the experimental setting the hIL-6 secretion has no impact on immune responses triggered against LPS, neither did the amount of secreted hIL-6 affect the immune responses.

Figure 6

LPS-specific intestinal secretory antibody responses in mice immunized with different IL-6-producing S. typhimurium X4064 strains. IgA (filled bars) and IgG titers (open bars) represent the mean antibody levels in fecal extracts from six mice (X4064(pCH1A+pYL3E)) and four mice (remaining groups), respectively. For titer measurement, lavage samples were diluted up to 1:160 in serial one to two dilution steps. The numbers are based on duplicate sample measurements and averages of four independent titer calculations for individual sera.

4 Discussion

The results presented demonstrate that IL-6 expressed and secreted by Salmonella vaccine strains can enhance systemic immunity against bacterial LPS, indicating that Salmonella-derived recombinant IL-6 could function as an immunotherapeutics capable of modulating induced immune responses against T-cell-independent antigens. Furthermore, data support the assumption that for development of the immunomodulatory action extracellular delivery and the amount of IL-6 are crucial. Surprisingly, IL-6 extracellularly expressed by the Salmonella vaccine strains affected both IgG and IgA titers. In contrast to the proposed role of IL-6 in the regulation of mucosal IgA responses [28], we could not detect a significant effect on secretory antibody production in the gastrointestinal tract, although titers were higher in the groups immunized with the IL-6-expressing strains.

We have previously demonstrated that hIL-6-HlyA_s producted by S. typhimurium X4064 remains soluble and retains full biological activity when secreted into the culture supernatant, while the intracellular non-secreted portion of the fusion protein forms insoluble aggregates, which lack bioactivity [22,23]. It is probably reasonable to assume that we may face a similar situation in vivo. The extracellular delivery of IL-6 is probably the reason that, in contrast to Dunstan and coworkers [10], who earlier investigated the therapeutic potential of IL-6 expressed by Salmonella vaccine strains in a mouse immunization model, we were able to demonstrate an immunostimulatory effect. Dunstan et al. used a standard plasmid expression vector for intracellular accumulation of recombinant murine IL-6 in the Salmonella strain. Although they reported IL-6 proliferative activity for cell lysates as well as for culture supernatant, the portion of biologically active IL-6 protein, which actually became liberated from the cells, may have been too low to trigger any in vivo effect in mice. Considering their negative results and postulating that activity and bioavailibility of IL-6 are critical for immunostimulatory action, we decided to use a secretory expression system for extracellular delivery of IL-6 by the Salmonella carrier.

Expression of IL-4, as another Th2 cytokine [29], in Salmonella vaccine strains revealed similar negative results with respect to a proposed immunomodulatory function on humoral immune responses [11]. Using a similar plasmid vector for IL-4 expression as Duncan et al. [10] did, they could not detect any enhancing or inhibitory effects on Ig subtype-specific immune responses either to whole Salmonella cells or to purified LPS and flagella antigen, respectively. Again, significant IL-4 proliferative activity was reported for cell lysates but essentially no bioactivity was detectable in vitro in culture supernatants, suggesting that the bioavailibility of IL-4 in their system depends on the portion of IL-4 protein that becomes liberated during the desintegration of Salmonella cells. Since IL-4, like IL-6, tends to form insoluble inclusion bodies when expressed at high levels in bacterial systems the amount of soluble and biologically active IL-4 protein delivered extracellularly by the Salmonella cells may have been very low.

So far, IL-5 is the only Th2 cytokine that has been shown to enhance humoral immune responses to bacterial antigens when expressed in Salmonella dublin vaccine strains as a fusion to thioredoxin [9]. In contrast to our results with IL-6, IL-5 additionally stimulated mucosal IgA responses to LPS. However, the important difference between both cytokines is that IL-5 apparently functions as a secretory IgA⁺ B-cell terminal differentiation factor, not promoting an isogenic switch to IgA [30]. Also, IL-6 is thought to control mucosal IgA responses at the postswitch level but, in contrast to IL-5, it seems to act synergically with other Th2 type cytokines [31]. Surprisingly, however, IL-5 production in Salmonella was promoted by the same plasmid vector [9] previously used for intracellular expression of IL-4 and IL-6 cDNA, implying that sufficient amounts of soluble active homodimeric protein must have been released by the cells to trigger this effect [14].

Live Salmonella expressing a recombinant vaccine antigen are promising vaccine candidates in humans and animals. However, the dominant portion of the induced immunity is directed towards Salmonella surface antigens and comparable weak immune responses are induced to recombinantly expressed vaccine antigens. One approach by which immune responses to recombinant vaccine antigens may be specifically enhanced is through secretion and surface presentation of the vaccine antigen (see [32], this issue). Another approach to achieve a stimulation or modulation of immune responses is the co-delivery of immunocompetent cytokines by the Salmonella vaccine strain. While the feasibility of both approaches has been demonstrated to some extent, the latter has the advantage that the same cytokine-producing Salmonella strain can be universally applied as vaccine carrier for a large number of recombinant antigens.

However, one major issue, which has to be addressed in that context is the possible impact of the in situ cytokine synthesis on the infectiveness and virulence of the Salmonella vaccine strain. Salmonella usually stimulate protective Th1-type responses during infection, which includes activation of macrophages, while Th2-type responses play only a minor role in the establishment of protective immunity to salmonellosis [33]. One may speculate that expression of IL-6 by the Salmonella cells leads to a general shift from Th1- to Th2-type immunity, which should consequently result in reduced clearing rate and prolonged colonization by this Salmonella strain, eventually affecting susceptibility of mice for that particular strain. Being aware of that possibility, we reduced bacterial dose in the immunization experiments from the usual 5×10⁹ to 2×10⁹. So far, we did not see any marked differences in the numbers of persisting bacteria in spleens and livers of immunized mice (data not shown). However, because of the reduced infection dose bacterial numbers in spleen and liver homogenates were rather low. This speculation opposes the results of Duncan et al. [10], who reported a significant reduction in bacterial numbers in spleen of mice infected with an IL-6-producing but non-secreting Salmonella strain. On the other hand, IL-4 and IL-5 expression in Salmonella strains leads to significantly higher in vivo survival rates in spleens and livers of immunized mice compared to the parental strain carrying the plasmid vector alone [9,11]. Further experimental work, including more detailed dissection of the Salmonella-induced immune responses and monitoring of the relative clearance rates, will be necessary to gain a precise picture of that important issue affecting the development of recombinant Salmonella-based vaccines.

At that stage, we cannot rule out that the observed immunostimulatory effect monitored in mice immunized with X4064(pCH1A+pYL3E) is actually due to a direct stimulation of the Th2 pathway or certain B-cell populations. It is also possible that the recombinant IL-6 secreted by the Salmonella cells may indirectly cause the enhancement of Ig subtype-specific immune responses to LPS by modulating certain cytokine levels, thus affecting the overall cytokine network involved in the regulation of the Salmonella-specific immune response. To ensure that the in situ synthesized IL-6 is the primary immunotherapeutic agent, immunization experiments in the absence of endogenous mouse IL-6 in IL-6-deficient mouse strains will be necessary.

To the best of our knowledge, this is the first report demonstrating that expression of a recombinant IL-6 in Salmonella vaccine strains can influence humoral responses against Salmonella antigens. Against our initial proposal, IL-6 expression led to an increase in both IgA and IgG antibody titers in immunized mice. So far, we have focused on LPS as a T-cell-independent antigen. Certainly similar investigations for T-cell-dependent protein antigens, such as flagella, will be required to further delineate the immunomodulatory effect of the Salmonella-derived IL-6. Apart from that, immunization studies in mouse strains with a different genetic background will eventually confirm the immunostimulatory effect caused by the IL-6-secreting Salmonella cells, thereby supporting the significance of these findings.

Acknowledgements

We thank W. Goebel and I. Gentschev (Theodor-Boveri-Institut, Würzburg, Germany) for providing DNA and the sequence of plasmid pMOhly1, respectively. We also thank R. Strugnell (University of Melbourne, Parksville, Australia) for suggestions and helpful discussion and G. Herger (Institut für Medizinische Mikrobiologie und Hygiene, Freiburg, Germany) for carefully reading the manuscript. This work was in part supported by the Deutsche Bundeswehr and Deutsche Mucoviszidose Gesellschaft.