Heat shock protein 70 is a potent activator of the human complement system

Abstract

According to new hypotheses, extracellular heat shock proteins (Hsps) may represent an ancestral danger signal of cellular death or lysis-activating innate immunity. Recent studies demonstrating a dual role for Hsp70 as both a chaperone and cytokine, inducing potent proinflammatory response in human monocytes, provided support for the hypothesis that extracellular Hsp is a messenger of stress. Our previous work focused on the complement-activating ability of human Hsp60. We demonstrated that Hsp60 complexed with specific antibodies induces a strong classical pathway (CP) activation. Here, we show that another chaperone molecule also possesses complement-activating ability. Solid-phase enzyme-linked immunosorbent assay was applied for the experiments. Human Hsp70 activated the CP independently of antibodies. No complement activation was found in the case of human Hsp90. Our data further support the hypothesis that chaperones may messenger stress to other cells. Complement-like molecules and primitive immune cells appeared together early in evolution. A joint action of these arms of innate immunity in response to free chaperones, the most abundant cellular proteins displaying a stress signal, may further strengthen the effectiveness of immune reactions.

INTRODUCTION

Heat shock proteins (Hsps) belong to the family of molecular chaperones. Their constitutive forms are expressed under physiologic conditions in all eukaryotic cells; however, in response to stress, the expression of the inducible Hsps largely increases. In eukaryocytes, chaperones of the Hsp60 family, also called chaperonins, are localized mainly in the mitochondria, whereas Hsp70 and Hsp90 are mainly present in the cytoplasm, the endoplasmic reticulum, the lysosomes, and the nucleus (Morimoto et al 1994). During stress, synthesis of Hsps is rapidly up-regulated both in prokaryotic and eukaryotic cells, and changes in intracellular distribution were described, including some expression on the cell surface (Gills et al 1985; De-Bruyn et al 1987; Wand-Württenberg et al 1991; Multhoff and Hightower 1996; Soltys and Gupta 1997).

Some extracellular functions of Hsps became evident during the last few years. Human and mycobacterial Hsp60 induces proinflammatory cytokine response in innate immune cells (Friedland et al 1993). Recently, specific recognition of Hsp60 by innate immunity (Chen et al 1999) and an important role for the CD14–Toll-like receptor complex in this process were shown (Kol et al 2000; Ohashi et al 2000). In addition, this receptor complex is a major high-affinity receptor for lipopolysaccharides (LPS) and other microbial products (Ulevitch 1993), suggesting that damaged self-structures and microbial agents are recognized via the same way by innate immunity. In a recent study, like Hsp60, the inducible form of Hsp70 stimulated cytokine (tumor necrosis factor-α, interleukin (Il)-1β, and Il-6) production through a CD14-dependent pathway in human monocytes (Asea et al 2000). A possible dual role for extracellular Hsp70 (“chaperokine”), both as a chaperone and a cytokine, was speculated. The endoplasmic reticulum resident Hsp gp96 has a very well-documented history of re-presentation of antigenic peptides and induction of specific T-cell responses (for review, see Srivastava et al 1998). Very recently, free gp96 molecules were described to exert a potent effect on maturation of innate immune cells (Singh-Jasuja et al 2000); furthermore, CD91 (alpha-2 macroglobulin receptor) was described to be a receptor for gp96 on human macrophages (Binder et al 2000).

As the aforementioned large body of evidence supports the important interaction between Hsps and the cellular arm of innate immunity, we were very surprised that interactions with another important arm of innate immunity, ie, the complement system, have never been tested. We reported previously on the differences between antibodies against human and mycobacterial 60 kDa Hsps (Prohászka et al 1999); these differences included their ability to activate the complement in a solid-phase enzyme-linked immunosorbent assay (ELISA) system. Human Hsp60 was shown to activate the complement in the form of specific antibody–Hsp60 complexes, whereas M bovis Hsp65-induced complement activation was much weaker and not influenced by anti-Hsp65 antibodies. The aim of this study was to investigate the complement activating properties of different Hsps. A second family of Hsp (Hsp70) was found to activate human complement, whereas the cytosolic form of Hsp90 did not induce activation.

MATERIALS AND METHODS

Heat shock proteins

Recombinant human Hsp70 (SPP-755) and purified human Hsp90 (SPP-770) antigens were obtained from StressGen Biotechnologies (Victoria, British Columbia, Canada).

Complement source

Normal human serum (NHS) was pooled serum from 10 young, healthy individuals, aliquoted and kept at −70°C. NHS and heat-inactivated NHS (56°C, 30 minutes) were prediluted 1:1 with veronal-buffered saline, containing Ca²⁺ and Mg²⁺, or with veronal buffer, containing only Mg²⁺ and ethyleneglycol-bis(aminoethylether)-tetraacetic acid (EGTA). In some experiments, serum of a genetically C2-deficient systemic lupus erythematosus patient and serum of a 4-year-old boy with very low levels of immune globulins (the levels of IgG-, IgA-, and IgM-type antibodies were 0.375 g/L, <0.01 g/L, and 0.037 g/L, respectively; all complement parameters tested [CH₅₀, C3, and C4 levels and haemolytic activity] were in the normal range; agammaglobilinaemic serum [AGS]) were also used.

Patients

Eighty subjects (63 males, 17 females, median age 60 years) with (n = 40) or without (n = 40) coronary heart disease were enrolled in this study. The basic data of the patients had been published previously (Prohászka et al 1999). All subjects underwent coronary angiography; patients had significant stenosis and received aorto-coronary bypass grafting, whereas 40 patients had no alterations (control). After informed consent, serum samples were taken 6 months after the operation or angiography (or both) to rule out the influence of acute disease and stress. Sera were aliquoted and kept at −70°C until use.

Complement activation ELISA

ELISA tests for the determination of the complement-activating ability of solid-phase Hsps were performed, as described earlier (Prohászka et al 1995, 1997, 1999). In brief, ELISA plates were coated with different amounts of human Hsps, Hsp70 and Hsp90. After washing, wells were incubated with 50 μL of different 1:1 diluted serum samples as complement source. The amounts of complement proteins fixed to the plate were determined with specific goat anti-C4b or goat anti-C3b antibodies (Atlantic Antibodies, Stillwater, MN, USA).

Measurement of antibodies against the heat shock proteins

The level of the anti-Hsp70 antibody was determined by ELISA as previously described for anti-Hsp60 antibodies (Prohászka et al 1999). Briefly, multiwell plates were coated with human Hsp70 (0.1 μg/well). The plates were blocked and incubated with 100 μL 1:200 diluted serum samples. Specific antibody binding was detected with rabbit anti-human IgG antibodies conjugated with peroxidase (DAKO, Glostrup, Denmark). The antibody concentration was expressed in terms of arbitrary units (AU/mL), and was calculated from a standard calibration curve from mouse monoclonal anti-Hsp70 antibody.

RESULTS

Activation of the classical complement pathway by Hsp70

Previously, our group had described classical pathway (CP) activation by human Hsp60–anti-Hsp60 antibody complexes. In this study, human Hsp70 was also found to activate the complement via the CP (Tables 1 and 2). Strong, significant C4b and C3b binding was seen in Hsp70-coated wells incubated with NHS; the lack of binding in wells incubated with heat-inactivated serum indicated specific activation of the complement. Hsp70, like Hsp60, did not induce complement activation in Mg²⁺-EGTA–chelated serum; no significant C4b and C3b binding occurred. The activation of the complement in the case of Hsp70 was dose dependent (Fig 1).

Table 1.

 Complement activation (C4b-binding) by solid-phase human heat shock proteins

Fig 1.

 The complement activation by solid-phase Hsp70 is dose dependent. Human Hsp70 and Hsp90 were incubated either with normal human serum (NHS) (▪) or with heat-inactivated human serum (HIHS)(♦); C4b components fixed to the plate were detected with specific anti-C4b antibody. Control uncoated wells were incubated with NHS (○) or with HIHS (▿). The figure shows mean optical density values and standard deviation of 4 parallel measurements. One representative experiment out of 3 similar ones. Significant, dose-dependent C4b binding was seen in the case of Hsp70 as compared with the uncoated plate (2-way analysis of variance: P = 0.0005, F = 261.6). The difference in the case of Hsp90 was not significant (P > 0.05)

No significant complement activation was observed in the case of Hsp90 (Tables 1 and 2; Fig 1).

Study on the mechanism of complement activation by Hsp70

The mechanism of complement activation by Hsp70 was investigated using a serum sample in which no functionally active C2 is present (genetic C2 deficiency, C2D serum). As presented in Tables 1 and 2, after incubation of Hsp70-coated wells with C2D serum, the CP is activated because significant C4 cleavage and binding occur, but no C3b binding can be observed. The CP of the complement can be activated either by antigen-antibody complexes or by direct binding of the first component, C1. To test the requirement of anti-Hsp70 antibody binding in the complement activation process by Hsp70, serum sample with total immunoglobulin deficiency (AGS) was applied. As shown in Table 1, significant CP activation (C4 cleavage and binding) occurs in Hsp70-coated wells incubated with AGS, indicating the initiation of CP; this activation was fully blocked by chelation of Ca²⁺ ions by Mg²⁺-EGTA. The activation of the CP in AGS is also detectable on the level of cleavage of C3 (Table 2) because significant C3b binding to Hsp70 occurs when compared with uncoated wells (P < 0.0001). This fact indicates that complement activation by Hsp70 results in a biologically very active complex (C4b2a3b, C5-convertase), which has a pivotal role in the opsonization and generation of biologically active anaphylatoxins.

Table 2.

 Complement activation (C3b-binding) by solid-phase human heat shock proteins

To further test the antibody requirement of Hsp70-induced CP activation, antibody levels against Hsp70 and Hsp70-induced CP activation were measured parallel (Fig 2). Forty serum samples of patients with coronary heart disease and 40 samples of healthy control were applied for this measurement. The level of anti-Hsp70 antibody was measured in all the samples and compared with the Hsp70-induced complement activation. There was no significant correlation between anti-Hsp70 antibody levels and Hsp70-induced complement activation (C4b binding to Hsp70) in either group. Spearman correlation coefficients and P values were calculated: for coronary heart disease, r = 0.0009, P = 0.995; and for control, r = −0.0099, P = 0.951. This fact indicates that Hsp70-induced CP activation takes place independent of antibodies; in samples with undetectable anti-Hsp70 antibody levels strong complement activation was measured (Fig 2).

Fig 2.

 Lack of correlation between human Hsp70–induced complement activation and anti-Hsp70 antibody levels. Each dot represents a single patient. Panel A: subjects without coronary heart disease. Panel B: patients with severe coronary heart disease. Spearman correlation coefficient for panel A: −0.0099 (0.951); for panel B: 0.0009 (0.995)

DISCUSSION

A solid-phase ELISA for the determination of ELISA-plate–coated antigen-induced complement activation was applied in this study to measure the complement-activating ability of different Hsps. The same test system was successfully used previously by our group to demonstrate and characterize the complement activation induced by gp41, the transmembrane glycoprotein of HIV-1 (Hidvégi et al 1993), gp120, the outer glycoprotein of HIV-1 (Prohászka et al 1995), human defensins (Prohászka et al 1997), and human Hsp60 (Prohászka et al 1999). In this study, the stress-inducible form (Hsp70) of the 70 kDa Hsp family was found to be a potent activator of the CP of the complement. In contrast, the cytosolic form of the 90 kDa Hsp family (Hsp90) did not induce complement activation. As shown previously, the Hsp60-induced complement activation requires the presence of specific anti-Hsp60 antibodies (Prohászka et al 1999). By contrast, in this study, Hsp70-induced CP activation was found to be independent of antibodies, indicating a direct interaction between the first complement component and Hsp70. A similar direct interaction between C1 and the transmembrane glycoprotein of gp41 has previously been reported (Ebenbichler et al 1991). The fact that Hsp70-induced C3 activation was fully blocked by addition of Mg²⁺-EGTA to NHS indicates that the activation of the complement in the alternative pathway was not significant in this case. However, according to the results of the present study (ie, both C4b and C3b binding to Hsp70 were strong and significant in NHS as well as in AGS), the complement activation by Hsp70 results in a biologically very active complex (C4b2a3b, C5-convertase), which has a pivotal role in the opsonization and generation of biologically active anaphylatoxins.

As the polysaccharide region of the endotoxin (LPS) had previously been shown to induce strong complement activation even in the presence of Mg²⁺-EGTA (alternative pathway, Inzana et al 1987), the results shown in Tables 1 and 2 indirectly indicate that LPS was not present in significant amounts in the preparations tested in this study.

Results obtained in our study, together with the literature data (Srivastava et al 1998; Chen et al 1999; Asea et al 2000), strongly indicate a close relationship between Hsps and innate immunity. The complement activation by Hsps in vivo may significantly amplify the power of immune activation in sites of cell damage and necrosis, where free extracellular and membrane-bound chaperone molecules may messenger cell stress because (1) complement is present and active in all body fluids, (2) highly active anaphylatoxins, produced during its activation, exert potent chemotactic effects on cells of innate immunity, and (3) complement-activation products and antigens opsonized by complement fragments have been shown to activate immune cells by specific complement receptors (CR, for review see Müller-Eberhard 1988). Thus, complement may act as an important agent in the clearance of damaged tissues through opsonization by binding to Hsps. The Hsp-induced complement activation may also represent some regulatory functions by masking and removing the immunogenic determinants of Hsps known to be immunodominant agents in several infectious agents. Because autoimmunity to Hsp70 may be present in several autoimmune diseases, including type I diabetes mellitus (Figueredo et al 1996), and infections are claimed to trigger autoimmunity, it would be important to study the regulatory role of Hsp70-induced complement activation in this process. However, opsonization of free or membrane-bound Hsps by the complement may exert some inhibitory or regulatory functions (or both) as well; therefore, the results obtained in complement-free in vitro assay may not surely reflect the true in vivo situation. These facts together strongly indicate that Hsps activating two important arms of innate immunity (ie, complement and antigen-presenting cells) modulate the immunological climate and influence the recognition of damaged, infected, and necrotic tissues.

Complement-like molecules and primitive immune cells appeared together early in evolution (Lambris et al 1999). A joint action of these arms of innate immunity in response to free chaperones, the most abundant cellular proteins displaying a stress signal, may further strengthen and regulate the effectiveness of immune reactions.