Proteolytic inactivation of nuclear alarmin high-mobility group box 1 by complement protease C1s during apoptosis

Abstract

Effective clearance of apoptotic cells by phagocytes prevents the release of intracellular alarmins and manifestation of autoimmunity. This prompt efferocytosis is complemented by intracellular proteolytic degradation that occurs within the apoptotic cells and in the efferosome of the phagocytes. Although the role of extracellular proteases in apoptotic cells clearance is unknown, the strong association of congenital C1s deficiency with Systemic Lupus Erythematosus highlights the protective nature that this extracellular protease has against autoimmunity. The archetypical role of serine protease C1s as the catalytic arm of C1 complex (C1qC1r2C1s2) involve in the propagation of the classical complement pathway could not provide the biological basis for this association. However, a recent observation of the ability of C1 complex to cleave a spectrum of intracellular cryptic targets exposed during apoptosis provides a valuable insight to the underlying protective mechanism. High-mobility group box 1 (HMGB1), an intracellular alarmin that is capable of inducing the formation of antinuclear autoantibodies and causes lupus-like conditions in mice, is identified as a novel potential target by bioinformatics analysis. This is verified experimentally with C1s, both in its purified and physiological form as C1 complex, cleaving HMGB1 into defined fragments of 19 and 12 kDa. This cleavage diminishes HMGB1 ability to enhance lipopolysaccharide mediated pro-inflammatory cytokines production from monocytes, macrophages and dendritic cells. Further mass spectrometric analysis of the C1 complex treated apoptotic cellular proteins demonstrated additional C1s substrates and revealed the complementary role of C1s in apoptotic cells clearance through the proteolytic cleavage of intracellular alarmins and autoantigens. C1 complex may have evolved as, besides the bacteriolytic arm of antibodies in which it activates the complement cascade, a tissue renewal mechanism that reduces the immunogenicity of apoptotic tissue debris and decreases the likelihood of autoimmunity.

Figures

Figure 1

HMGB1 is cleavable by C1s. (a) The PoPS model for predicting C1s substrate sequences. The C1s protease active site contains eight contiguous pockets or subsites each having a unique structural and chemical property and preferred specific amino acid residues (Single letter amino acid code is used). These amino acid residues were experimentally determined previously. The subsites display different importance with varying specificity except for the A4 position where an arginine, which is conserved in C4, C2 and C1-INH, is considered an essential residue. Position of the `scissile bond’ is indicated with triangle. (b) Predicted C1s cleavage sites on HMGB1. The HMGB1 sequence is presented and the A box and B-box are shaded. The C-terminal acidic tail is underlined. Where secondary structures are predicted, these are indicated under the sequence. `-’, lack of secondary structure. The 3 inverted triangles indicate predicted C1s cleavage sites with PoPS scores (Arg70, Arg97 and Arg163). (c) Purified rHMGB1 was cleaved by purified sC1s. RHMGB1 produced in mouse myeloma cells (10 μg/ml), serum C4 (10 μg/ml), and BSA (5 μg/ml) were incubated overnight at 37 °C with activated sC1s (11 μg/ml). The samples were separated by SDS-PAGE on 12.5% (w/v) gels under reducing conditions and visualized by silver staining. The three C4 subunit chains (α, β and γ) are labeled. The cleaved C4 α chain was labeled α’. After C1 digestion of rHMGB1, a fragment was generated (F1). (d) Two HMGB1 fragments were detected after C1s digestion by western blotting (F1 and F2). Western blotting was performed using a rabbit polyclonal anti-HMGB1 antibody. (e) C1 complex (C1 cplx) cleaved HMGB1 more effectively than C1s but both generated two similar fragments. In this experiment, His-HMGB1 expressed in bacteria E. coli (20 μg/ml) was incubated overnight at 37 °C with C1 complex and activated sC1s at the concentrations indicated. The reactions were separated by SDS-PAGE on 15% (w/v) gels under reducing conditions and western blotting was performed using the polyclonal rabbit anti-HMGB1 antibody. F1 and F2 marked the HMGB1 fragments after cleavage.

Figure 2

C1 and C1s cleavage of HMGB1 released from apoptotic cells. (a) DNA laddering of UV-irradiated Jurkat cells. Cells were, after UV irradiation, cultured for 4 h in serum-free medium (post UV). As control, cells were cultured for 4 h without prior UV irradiation (control). Extracted DNA was examined on 1.2% (w/v) agarose gels and DNA was visualized using GelGreen (Biotium, Inc., Hayward, CA, USA) (b) HMGB1 release from apoptotic cells. After UV irradiation, Jurkat cells were cultured for a series of time periods up to 48 h. After centrifugation, supernatants were harvested and subjected to SDS-PAGE on 12.5% (w/v) gels and western blotting using a rabbit polyclonal anti-HMGB1 antibody. (c) C1 degradation of HMGB1 released from apoptotic cells. In this experiment, Jurkat cells were harvested 2.5 h after UV irradiation and, after mixing, divided in 490-μl aliquots. C1 (240 μg/ml) was added to a final concentration of 5 μg/ml (10.4 μl) and incubated for 30 min at 37 °C. After centrifugation, supernatants were separated by SDS-PAGE on 12.5% gels and then analyzed by western blotting using goat anti-C1q and rabbit anti-C1s and anti-HMGB1 antibodies. (d) C1 cleavage of HMGB1 was inhibited by C1-INH. Apoptotic cells were incubated with C1 for 30 min at 37 °C in the presence or absence of C1-INH (105 μg/ml). As controls, apoptotic cells were incubated with C1-INH only. After centrifugation, supernatants were subjected to SDS-PAGE on 12.5% (w/v) gels and western blotting using the rabbit anti-HMGB1 antibody. (e) Comparison of C1 with C1s in HMGB1 cleavage during apoptosis. As in (c), apoptotic cells harvested 2.5 h after UV irradiation were incubated for 30 min at 37 °C with C1 (5 μg/ml) or sC1s (1.0, 2.5 or 5.0 μg/ml). As controls, Jurkat cells were incubated with C1 or C1s without prior UV irradiation. After centrifugation, supernatants were subjected to SDS-PAGE on 12.5% gels and western blotting using the rabbit anti-HMGB1 antibody. Band intensities were quantified by densitometry. Values for C1s represent combined signals of the heavy and light chains in each lane. Values for HMGB1 were derived from the single band in each lane. Note the reduced HMGB1 intensities at 2.5 and 5.0 μg/ml of sC1s. Apoptotic cells: AC.

Figure 3

Alignment of HMGB1 fragments potentially generated after C1s cleavage and HMGB1 regions that interact with LPS and CD14. Based on the sizes of the two HMGB1 fragments after C1s digestion, that is, 19 and 12 kDa (right panel), the two likely cleavage sites are Arg163 (F1ʹ) and Arg97 (F2ʹ). In this western blotting experiment, bacteria-expressed His-HMGB1 was used. SC1s was used at reducing concentrations. The corresponding LPS and CD14 interactive regions in HMGB1 were aligned below. Note that, both C1s cleavage sites fall into bioactive regions of HMGB1 (triangles) and can potentially impair HMGB1 synergy with LPS.

Figure 4

HMGB1 enables monocyte, macrophage and DC activation by suboptimal levels of LPS but this is inactivated with C1s digestion. Macrophages were stimulated for 20 h with LPS at 1 ng/ml and rHMGB1 was added at 12.5, 25, 50 and 100 ng/ml. As controls, macrophages were stimulated with LPS at 1 and 10 ng/ml without rHMGB1. Production of IL-6 (a) and TNF-α (b) was measured by ELISA. In another experiment, different concentrations of LPS (12.5–500 ng/ml) were pre-incubated for 20 h at 37 °C with rHMGB1 (5 μg/ml) in the presence or absence of rC1s (2.75 μg/ml). As controls, LPS was incubated with PBS or rC1s. These stimuli mixtures were used at 1/50 dilution to stimulate macrophages so that the final concentration of HMGB1 and rC1s in the culture are 100 and 55 ng/ml respectively. IL-6 (c) and TNF-α (d) production was determined by ELISA. DC (e) and monocytes (f) were similarly stimulated except that LPS concentrations were increased in the pre-incubation mixture (25–5000 ng/ml). IL-6 production was determined by ELISA. Experiments were performed in triplicates. Results were analyzed using Student’s 2-tailed unpaired t-test and presented as mean±S.D. *P<0.05, **P<0.01.

Figure 5

C1 cleaves multiple proteins released by apoptotic cells. (a) Visible protein profile released from apoptotic cells and their susceptibility to C1 or C1s degradation. UV-irradiated Jurkat cells were cultured for 2.5 h and then incubated with sC1s, C1 or PBS for 30 min at 37 °C. After centrifugation, supernatants were harvested and subjected to SDS-PAGE on 15% (w/v) gels. Proteins were visualized by Commassie blue staining. Band regions in the C1-treated sample that diminished compared with control sample (PBS-treated) are indicated with arrows and alphabets. Four band regions were identified: A (~100 kDa), B (~60 kDa), C (~50 kDa) and D (~40 kDa). In a separate gel, these four band regions in the control sample were excised for mass spectrometry analysis (Supplementary Table 1). (b) The separated proteins were also analyzed by western blotting where HMGB1, NPM1, β-actin and, as controls, C1s and C1q were detected. (c) C1 showed no cleavage of HSP90α and RPLP0. By mass spectrometry, HSP90α was detected in band A and RPLP0 was found in band C (Supplementary Table 1). Western blot detection of HSP90α and RPLP0 were done using rabbit polyclonal antibodies. HMGB1 was used as a control. AC: apoptotic cells.

Figure 6

C1 cleavage of autoantigens released by apoptotic cells. U937 cells (5×107/ml) in PBS were UV-irradiated. The supernatant was obtained after 24 h and incubated for 1 h with C1 (48 μg/ml) or, as a control, with PBS. The supernatant samples were then separated by SDS-PAGE on 12.5% (w/v) gels and analyzed by western blotting with rheumatological patient sera at 1 : 5000 dilutions. Horseradish peroxidase-conjugated goat anti-human IgG Fc was used as secondary antibody and signals were visualized using the Bio-Rad MPS ChemiDoc imaging system. (a) Sera from four patients detected autoantigens in the apoptotic U937 supernatant. A 45-kDa autoantigen which was detected by sera from both patient B and C and degraded by C1 has been indicated (arrow). A 40-kDa autoantigen detected by serum from patient D was also degraded by C1 (arrow). (b) Sera from patients E-H detected insignificant autoantigens in the apoptotic U937 supernatant. In this validating experiment, serum from patient C was used as a positive control. A single blot was cut into 5 strips and incubated separately with the patients’ sera. Apoptotic cells: AC.