Immunoglobulin A1 Protease, an Exoenzyme of Pathogenic Neisseriae, Is a Potent Inducer of Proinflammatory Cytokines

In this study, we used the recombinant form of IgA1 protease expressed in Escherichia coli strain H2053, which harbors plasmid pIP11 containing the iga gene of N. gonorrhoeae MS11 with the temperature-inducible bacteriophage λ promoter P_L, as described previously 18. In brief, the supernatant of a 4-liter stationary phase culture was recovered by centrifugation, supplemented with 0.1 M EDTA, and diluted 1:1 (vol/vol) with potassium phosphate buffer (20 mM, pH 7.0). The bacterial supernatant was applied to a cation exchange column. After washing, bound protein was eluted with 500 mM potassium phosphate, 8.6% glycerol. Fractions of 10 ml were collected, precipitated with ammonium sulphate, and then redissolved in gel filtration buffer. Gel filtration was performed with a preequilibrated Superdex column (Amersham Pharmacia Biotech). Protein-containing eluates were recognized by UV absorption, pooled, and precipitated with ammonium sulphate. Finally, the pellet was suspended in PBS and dialyzed against PBS, pH 7.4, and stored in aliquots at −70°C. Purity of the protein was assessed by SDS-PAGE with subsequent Coomassie staining showing the characteristic protein profile with the typical 109- and 106-kD protein bands as described previously 18. The protein concentration was determined using a commercial protein determination assay (Bio-Rad). Specificity for IgA1 protease was confirmed by immunoblotting using an IgA1 protease–specific rabbit antiserum 13. The cleavage activity of the recombinant IgA1 protease was determined with human IgA1 (20 ng/μl; Dakopatts) in Tris buffer (0.05 M Tris-HCl, 10 mM CaCl₂, and 10 mM MgCl₂, pH 7.5) at 37°C according to the method described previously 36. Heat denaturation (boiling for 10 min) completely destroyed the enzymatic activity. The enzymatic activity could also be blocked by addition of the specific peptide prolyl boronic acid inhibitor 37. 100 μM of the inhibitor (provided by A.G. Plaut, New England Medical Center, Boston, MA) abrogated cleavage of IgA at concentrations up to 10 ng per μl of IgA1 protease (data not shown).

The endotoxin contents of our IgA1 protease preparations were determined by Limulus amebocyte lysate assay (LAL; Haemochrom Diagnostica). All preparations contained <2 pg of endotoxin per μg protein. It is described that some proteins could interfere with the Limulus assay 38. To ensure that IgA1 protease is free of LPS, we have tested in addition the priming potency of IgA1 protease for neutrophils, which are as sensitive as the Limulus assay in detecting LPS. Different types of LPS can prime neutrophils to increase their oxidative response after stimulation with the agonist fMLP (Sigma Chemical Co.) 38. PMNs preincubated with 10 μg/ml IgA1 protease for 30 min stimulated with 1 μM fMLP showed no alteration of the oxidative response compared with the control when measured as luminol-dependent chemiluminescence in a six-channel Biolumat LB 9505 (Berthold; data not shown).

PBMCs were isolated from buffy coats or freshly citrated blood of adult, healthy volunteers by density gradient centrifugation over Ficoll/Isopaque (Amersham Pharmacia Biotech) and washed twice in Ca², Mg²⁺–free PBS. For some experiments, lymphocytes and monocytes were separated (6 × 10⁶ cells/ml in PBS plus 0.5% BSA) by counterflow centrifugation (elutriation) using the Beckman J6ML/JE-5.0 centrifuge system. Elutriation was performed at constant rotor speed with increasing flow rates, and lymphocytes were collected. To obtain purified T cells, the lymphocyte suspensions were positively sorted by anti-CD3 magnetic beads (MACS; Miltenyi Biotec). Purity of CD3⁺ T cells was always >95% as determined by flow cytometry. Monocytes were prepared from PBMCs either by adherence to plastic dishes (1 h, 37°C) or by negative selection using the MACS monocyte isolation kit (Miltenyi Biotec). The purity of CD14⁺ cells was always >85% as measured by flow cytometry using a PE-conjugated anti-CD14 antibody (Leu M3; Becton Dickinson). The remaining 15% consisted of CD3⁻, CD19⁻, and CD56⁻ flow cytometric counting events.

The monocytic cell line MonoMac 6 (no. 124; American Type Culture Collection) was obtained from the German Collection of Microorganism and Cell Cultures (Braunschweig, Germany) and maintained in 24-well plates under low endotoxin conditions (<10 pg/ml) in RPMI 1640 supplemented with 10% heat-inactivated FCS, 2 mM l-glutamine, 1% nonessential amino acids, 1 mM sodium pyruvate, and 9 μg/ml bovine insulin as described 39. The cell line normally expresses CD14 as controlled by flow cytometry.

Cytotoxic effects of IgA1 protease preparations were excluded by staining of PBMCs with propidium iodide (10 μg/ml) and subsequent evaluation by flow cytometry. Cell viability of freshly isolated or cultured cells (24 h with 10 μg/ml IgA1 protease) was >95% in all experiments.

All cell cultures were performed in RPMI 1640 supplemented with 20 nM l-glutamine, 25 mM Hepes (GIBCO BRL), 500 U/ml penicillin, 500 μg/ml streptomycin, and 5% heat-inactivated human AB serum (pooled male serum, <<1 ng/ml endotoxin; Sigma Chemical Co.). This medium is further designated as culture medium.

A total of 10⁵ cells was cultured in 200 μl culture medium in round-bottomed 96-well plates at 37°C, 5% CO₂. MonoMac 6 cells (10⁵ cells/well) were incubated for 24 h in 96-well flat-bottomed plates with culture medium. Cell-free supernatants were harvested after 18 h (150 μl/well, three wells per independent experiment), pooled, and frozen in 50 μl aliquots at −20°C until assayed.

To investigate the effect of CD3⁺ T lymphocytes on the IgA1 protease–induced cytokine production of CD14⁺ monocytes, 2 × 10⁴ CD14⁺ cells were cultured either with or without CD3⁺ cells keeping the monocyte/lymphocyte ratio constantly at 1:4. A total of 10⁵ PBMCs were cultured as a control with a monocyte content ranging from 10 to 20%.

The levels of the released cytokines TNF-α, IL-1β, IL-6, IL-8 (R&D Systems), and IFN-γ, GM-CSF, and IL-10 (PharMingen) in the culture supernatants were quantified with commercially available sandwich ELISA kits. The assays were performed as recommended by the manufacturer. In brief, polystyrene microtiter plates were coated with the catching mAb overnight, washed intensively (washing buffer: PBS, 0.02% Tween 20), and blocked with PBS, 5% AB serum. Different dilutions of all samples and the standard cytokine dilutions were assayed in duplicate or triplicate, respectively, with a total volume of 100 μl for 2 h at room temperature. After removing the unbound detecting mAb by washing (10 times), streptavidin-peroxidase solution was added to each well and incubated for 20 min. After 10 washes, the chromogenic substrate (TMB substrate with H₂O₂; Kirkegaard & Perry) was added to each well and color development was stopped after 20 min by adding 50 μl 1 M H₃PO₄. The absorbance of the samples was measured at 450 nm. The cytokine concentrations were determined by extrapolating from the standard curve. The results are expressed as means of triplicates in ng/ml. The detection limits of the ELISA kits were as follows: TNF-α, IL-1β, IL-10, GM-CSF, and IFN-γ ≤ 0.01 ng/ml; IL-6 and IL-8, ≤ 0.05 ng/ml.

For immunofluorescence staining, we used the following: anti-CD14–PE as a monocytic marker (LeuM3, IgG2a; Becton Dickinson), and FITC-conjugated anti–HLA-DR (B.812.2, IgG2a; Coulter Immunotech) and anti-CD25 for detection of the IL-2 receptor (ACT-1, IgG1; Dako). Control mouse Ig (IgG1-FITC/IgG2a-PE) were obtained from Becton Dickinson. An FITC-conjugated goat anti–mouse F(ab) fragment (Coulter Immunotech) was used as a secondary antibody. The cells were washed, adjusted to 2 × 10⁵/200 μl in PBS with 0.2% BSA, and incubated with the antibodies for 30 min on ice. After washing, the samples were analyzed on a FACScalibur™ (Becton Dickinson) using the CELLQuest™ program. The cells were gated due to their forward scatter/side scatter profile. Background levels of immunofluorescence were determined using isotype control Ig. For the detection of IL-2 receptor, cells were first labeled with unconjugated anti-CD25, washed, and then incubated with FITC-conjugated goat anti–mouse F(ab)₂ fragment (5 μg/ml).

Proliferation assays with PBMCs (10⁵ cells/well in culture medium) were performed at 37°C in 96-well round-bottomed plates at a total of 150 μl/well. The cells were cultured for various time periods (1–4 d) and pulsed with 18.5 kBq of [³H]thymidine per well. Cells were then harvested on filters, and incorporation of [³H]thymidine was measured using a Micro-beta Scintillation Counter (Wallac Instruments). Unstimulated cells and PHA-stimulated (5 μg/ml) cells were used as controls.

The nonparametric Wilcoxon rank test was performed for comparison of cytokine data of PHA-, LPS-, and IgA1 protease–stimulated PBMCs. Statistical differences were considered significant at P < 0.01. Data from separate experiments were expressed as means ± SEM and were evaluated for statistical significance by Student's t test. Data from a single representative experiment are presented as means ± SD obtained by triplicate determination.

To assess the production of the proinflammatory cytokines TNF-α, IL-1β, IL-6, and IL-8, PBMCs from healthy human donors were incubated with purified IgA1 protease. LPS, PHA, and medium alone were used as control. Initial experiments showed that release of proinflammatory cytokines from PBMCs induced by IgA1 protease reached maximal levels between 12 and 24 h (data not shown). Furthermore, a rapid decrease in IgA1 protease–induced cytokine production could be observed after 24 h, and only small amounts of TNF-α were detectable after 72 h. Therefore, in subsequent experiments a single end-point of 18 h culture was chosen that allowed detection of TNF-α and other proinflammatory cytokines. IgA1 protease stimulated the release of proinflammatory cytokines by PBMCs in a dose-dependent manner (Fig. 1, and Table). Significant amounts of TNF-α in the supernatants were already detectable in the presence of 1 μg/ml IgA1 protease, and highest levels were reached using 10 μg/ml. Similar dose–response patterns were obtained for other proinflammatory cytokines such as IL-1β, IL-6, and the chemokine IL-8. Nonstimulated PBMCs produced negligible amounts of TNF-α, IL-1β, and IL-6, but some IL-8, which was probably induced by adherence as previously postulated by Kasahara et al. 40.

To determine whether the cytokine induction was due to LPS present in the preparation of the recombinant IgA1 protease from E. coli, we tested the potential of E. coli LPS to induce TNF-α. The endotoxin content present in the protease preparation of 10 μg/ml did not exceed 20 pg/ml. As shown in Fig. 1, no significant cytokine production compared with the control was detected at 10–30 pg/ml LPS, indicating that endotoxin contamination was not the cause of cytokine induction by purified IgA1 protease. TNF-α synthesis in response to 10 μg/ml IgA1 protease reached the level synthesized by cells stimulated with a much higher dose of E. coli LPS (30 ng/ml). In subsequent experiments, low dose (30 pg/ml) and high dose (30 ng/ml) controls of LPS were always included.

The cytokine release of PBMCs showed similar variations between different donors upon stimulation with either IgA1 protease, LPS, or PHA. As shown in Fig. 2 (A–D), the median levels of the proinflammatory cytokines detected in the presence of 10 μg/ml IgA1 protease were approximately similar to those found after LPS stimulation and sometimes even higher than those incubated with the lectin PHA.

Heat denaturation abolished the cytokine-inducing capacity of IgA1 protease (data not shown). To test whether the enzymatic activity is involved in that process, PBMCs were incubated in the presence of 100 μM of a specific IgA1 protease inhibitor (peptide prolyl boronic acid [37]). Although at this concentration the inhibitor completely blocked the cleavage of IgA1, it altered neither the LPS- nor the IgA1 protease–induced production of proinflammatory cytokines (data not shown). This demonstrates that the native form but not the enzymatic activity of IgA1 protease is required for the induction of proinflammatory cytokines.

IgA1 protease triggers a strong release of proinflammatory cytokines from PBMCs after only a short incubation period, indicating that monocytes might be the main producer of these mediators. To determine the source of cytokine release, we cultivated lymphocyte-depleted peripheral blood monocytes (purity varied between 85 and 95%). As demonstrated in Fig. 3, IgA1 protease induced significantly lower amounts of TNF-α, IL-1β, IL-6, and IL-8 compared with LPS, which readily stimulated TNF-α production in purified monocytes. Direct comparison with PBMCs revealed that IgA1 protease–stimulated PBMCs can elicit up to 20-fold higher amounts of these cytokines compared with the respective lymphocyte-depleted monocyte cultures, which was particularly true for IL-6 and IL-8 (data not shown; see also Fig. 2). We also tested the monocytic cell line, MonoMac 6. This cell line shows phenotypic and functional characteristics of mature blood monocytes and produces IL-1, IL-6, and TNF-α upon stimulation with LPS or PMA 39,41. Compared with LPS, we found a significantly lower production of these cytokines and IL-8 after stimulation with IgA1 protease (Table).

Direct contact with T cells can augment monocyte cytokine production 42. To examine the necessity of cell contact for TNF-α release in the different types of stimulation, monocytes stimulated with IgA1 protease, LPS, or PHA were incubated with autologous active and fixed (15 min, paraformaldehyde at 0°C) lymphocytes at a monocyte/lymphocyte ratio of 1:2. In parallel, the lymphocytes and monocytes were incubated alone. Such lymphocytes and purified monocytes produced only low levels of TNF-α upon stimulation with 10 μg/ml IgA1 protease (Fig. 4). The same results were obtained after stimulation with PHA. The addition of viable lymphocytes increased the TNF-α secretion in nonstimulated, IgA1 protease–, and PHA-treated monocytes, whereas the addition of prefixed lymphocytes did not. Monocytes stimulated with IgA1 protease or PHA also showed a significantly stronger TNF-α induction in the presence of lymphocytes compared with the control. Only LPS was able to induce TNF-α in purified monocytes alone, and this was not further enhanced by lymphocytes.

To examine the particular role of T cells for IgA1 protease–induced TNF-α release, monocytes were incubated with autologous CD3⁺ T lymphocytes at a monocyte/lymphocyte ratio of 1:4. While T lymphocytes produced no or very low amounts of TNF-α, the monocytes released reduced levels of TNF-α in response to IgA1 protease compared with crude PBMCs. Such reduced TNF-α secretion of monocytes was enhanced in the presence of autologous purified T cells (Fig. 5). However, when the two cell populations were separated by a permeable membrane during the cell culture, the TNF-α release by monocytes was not increased in response to IgA1 protease (data not shown). These results suggest that direct contact between T cells and monocytes is necessary for a maximum of cytokine production by monocytes in response to IgA1 protease.

It is well known that lymphokines such as GM-CSF or IFN-γ can augment cytokine release in monocytes 43,44. We found increased levels of GM-CSF in PBMC cultures after incubation with IgA1 protease for 18 h, whereas IgA1 protease failed to induce release of IFN-γ after that time (Fig. 6A and Fig. B). Very low levels of IFN-γ were finally detectable after 4 d (data not shown). However, the IFN-γ levels in these cultures were always more than 100-fold lower than peak levels of IFN-γ in PHA-stimulated PBMC cultures. Furthermore, in the first 4 d of culture, neither cell proliferation nor increase of IL-2 receptor–positive lymphocytes could be measured after incubation with IgA1 protease (data not shown). These results suggest that viable T cells are required for the cytokine release by monocytes stimulated with IgA1 protease, but unlike PHA, IgA1 protease does not seem to provide a strong activation signal for T lymphocytes.

IL-10 is a cytokine with immunosuppressive and antiinflammatory effects on monocytes. Reportedly, IL-10 synthesis is delayed relative to the synthesis of inflammatory cytokines TNF-α, IL-1β, and IL-6 45. In contrast to the induction of proinflammatory cytokines, IgA1 protease failed to induce IL-10, whereas LPS as well as PHA could stimulate IL-10 production in PBMCs (Fig. 7). Kinetic experiments revealed that IgA1 protease does not induce IL-10 production in PBMCs within a period up to 96 h. In contrast, culture supernatants of PHA- or LPS-stimulated PBMCs contained high IL-10 levels at different time points (data not shown).

IgA1 protease, a putative virulence factor of human pathogenic Neisseriae, was initially discovered as an enzyme that cleaves human IgA1 molecules, thus interfering with the biological functions of IgA1 10. Here we report on a novel property of neisserial IgA1 protease, the induction of the proinflammatory cytokines TNF-α, IL-1β, and IL-6, and the chemokine IL-8 in PBMCs.

The stimulation of cytokine synthesis by IgA1 protease occurred in the nanomolar range and increased in a dose-dependent manner. The level of cytokine release was substantial but varied to some extent depending on the donor. The rapid cytokine induction within 24 h indicates a role of monocytes/macrophages in this process. However, compared with PBMCs, significantly lower amounts of TNF-α and IL-8 and practically no IL-6 and IL-1β were elicited by purified monocytes in response to IgA1 protease, whereas LPS induced a strong cytokine secretion that included IL-6 and IL-1β. The monocytic cell line MonoMac 6 also showed a low production of TNF-α, IL-1β, IL-6, and IL-8 in response to IgA1 protease. These results suggest that the presence of lymphocytes is required for full cytokine induction by IgA1 protease.

It is known that the activation of monocytes/macrophages and the release of monocyte-derived cytokines can be provoked by either direct cell–cell contact with activated antigen-specific T cells or secreted lymphokines 42,43,44,46. This is consistent with our finding that IgA1 protease–induced TNF-α secretion by monocytes was enhanced by the addition of lymphocytes or CD3⁺ T cells to the cultures. To investigate the role of activated T cells in this process, we measured IFN-γ secretion, CD69, and IL-2 receptor expression, representing typical markers of activated T cells. Although CD69 was slightly upregulated, the low number of IL-2 receptor–positive cells suggests that no large T cell population was activated or that only partial activation occurred (46; and data not shown). The T cell cytokine IFN-γ can enhance the cytokine expression in monocytes synergistically in combination with GM-CSF 43,44. Although GM-CSF was detected after IgA1 stimulation, the release of IFN-γ was low and the addition of anti–IFN-γ antibodies had no effect on the monokine production (data not shown). However, the fact that cocultivation with T cells separated from the monocytes by a permeable membrane could not enhance the production of proinflammatory cytokines indicates that T cells may also participate by a yet unknown contact-dependent mechanism or by nonclassical presentation of antigen 47,48,49. We also did not find binding of IgA1 protease to T cells, whereas CD14⁺ monocytes were clearly stained with FITC-labeled IgA1 protease (data not shown).

The profile of the cytokine response of PBMCs to IgA1 protease shows some similarities to LPS stimulation. Although the IgA1 protease was produced as a recombinant protein in E. coli, our preparations contained usually <2 pg LPS/μg protein. This LPS content could not sufficiently induce the observed cytokine production. Furthermore, effects which are typically triggered by LPS, such as priming of phagocytes for an enhancement of the oxidative burst, could not be confirmed in cells incubated with IgA1 protease (38; and data not shown). On the other hand, LPS does not require T cell engagement, suggesting that IgA1 protease functions by a different mechanism. Both molecules can have synergistic effects in vitro (data not shown) and may also act concertedly in vivo. Activated monocytes/macrophages also produce the antiinflammatory cytokine IL-10. This cytokine mediates important autoregulatory effects by downregulating inflammation, and its maximal production is delayed compared with the initial burst of proinflammatory cytokines 45. IL-10 production was augmented by stimulation with LPS or PHA, but not with IgA1 protease 50. Thus, IgA1 protease seems unique in stimulating an asymmetric synthesis of proinflammatory cytokines from monocytes and fails to induce strong regulatory feedback loops.

Besides the neisserial IgA1 protease reported here, other bacterial products have been found to induce proinflammatory cytokines. These components have been classified as a separate class of bacterial virulence factors termed modulins 51,34. Examples include the listeriolysin from Listeria monocytogenes 52, the E. coli hemolysin 53, and the streptococcal pneumolysin 54. Other proteins with enzymatic activity, such as the urease of Helicobacter pylori 55 and the sphingomyelinase from Staphylococcus aureus 56, also modulate cytokine production. Only a few modulins, like the anthrax lethal toxin from Bacillus anthracis 34 and the clostridium toxin from Clostridium difficile 57, act in the attomolar or femtomolar range, while most other modulins including IgA1 protease exhibit their activity in the nanomolar range. Some modulins, such the staphylococcal toxic shock syndrome toxin 1 (TSST-1) from S. aureus or the erythrogenic exotoxin (ET) from Streptococcus pyogenes possess superantigenic activity, thereby inducing T cells to proliferate and to produce cytokines 58. IgA1 protease is different from those superantigenic factors because it is not a T cell mitogen.

The property of neisserial IgA1 protease to induce proinflammatory cytokines may be highly relevant for the disease-causing properties of gonococci and meningococci. A hallmark of the pathophysiology in these diseases is the intense inflammatory response associated with the substantial release of proinflammatory cytokines. TNF-α, IL-1, IL-6, and IL-8 have been detected in local secretions and sera during mucosal infections with gram-negative species, including N. meningitidis and N. gonorrhoeae 29,35. Increased levels of IL-8 and TNF-α in particular in plasma and in urine, and smaller amounts of IL-1β and IL-6 were found in experimental gonorrhea after intraurethral application of gonococci 32.

High levels of these cytokines were also found in the CSF and in the serum during meningeal inflammation. These mediators promote the inflammatory response with subsequent tissue damage and in some cases septic shock 28,29,30. The level of TNF-α in the CSF directly correlates with the severity of the damage to the blood–brain barrier 29.

The most important bacterial component of gram-negative bacteria known to stimulate the release of inflammatory cytokines is LPS. Indeed, meningococcal LPS was found to be a major cytokine-inducing component in bacterial meningitis, since the cytokine profile and levels found in serum or CSF correlated with the LPS concentrations. However, Prins et al. 59 could not confirm the relationship between levels of cytokines and bacterial endotoxin produced by various strains of young meningitis patients, suggesting that factors other than endotoxin may be involved in cytokine induction during acute meningococcal septic shock. In vivo, LPS and IgA1 protease might be released together and act synergistically. Strikingly, though, the three major causative agents of bacterial meningitis, N. meningitidis, H. influenzae, and Streptococcus pneumoniae, all produce an IgA1 protease 11,13,14,60. Therefore, IgA1 protease may constitute an underestimated virulence factor in sepsis and meningeal infection.

Besides the cleavage of IgA1, several additional functions have been attributed to neisserial IgA1 protease. IgA1 protease was recently shown to cleave LAMP-1, a lysosomal protein associated with the endosomal maturation 36,61, but only one report suggested that, as a consequence, IgA1 protease facilitates the intracellular survival of pathogenic N. gonorrhoeae 61. Furthermore, other human proteins have been found to be proteolytic substrates in vitro. However, the biological relevance of their cleavage remains unclear. In a parallel study, we recently observed that IgA1 protease efficiently protects phagocytes from TNF-α–mediated apoptosis (our unpublished results). Considering the strong potential of IgA1 protease to induce TNF-α, this observation is intriguing since Neisseria is capable of entering phagocytic cells by an active mechanism 62. Hence, the pathogens may have an interest in preventing lethal effects of TNF-α on their intracellular niches.

The induction of proinflammatory cytokines may contribute to the characteristic inflammatory reaction caused by the pathogenic Neisseriae. Understanding the mechanism of IgA1 protease in the regulation of these cytokines has potential therapeutic significance and is currently in progress. Future research should reveal strategies for blocking this cytokine-inducing pathway in response to pathogenic Neisseriae and probably to other IgA1 protease–secreting human pathogenic bacteria.

We thank Dr. Susanne Beck for providing the recombinant IgA1 protease, Dr. Robert Hurwitz for IgA1 protease purification and characterization, and Ilona Harm, Uwe Koscy, and Petra Lindenberg for technical assistance. We are also grateful to Drs. T. Aebischer, S. Gödert, and A. Walduck for helpful discussions.

This work was supported by a grant from the Fonds der Chemischen Industrie to T.F. Meyer.