Factors masking HMGB1 in human serum and plasma

Vilma Urbonaviciute,* Barbara G. Fu rnrohr,* Christian Weber,* Martin Haslbeck, Sabine Wilhelm,* Martin Herrmann, and Reinhard E. Voll*,,1 *IZKF Research Group 2, Nikolaus-Fiebiger-Center of Molecular Medicine, University of Erlangen-Nu rnberg, Erlangen, Germany; Department Chemistry, Technical University Munich, Munich, Germany; and Department of Internal Medicine 3, University Hospital Erlangen, Erlangen, Germany

Abstract: High mobility group box 1 protein (HMGB1) is a ubiquitously expressed architectural chromosomal protein. Recently, it has become obvious that HMGB1 can also act as a proinammatory mediator when actively secreted during cell activation or passively released from necrotic cells. HMGB1 appears to play an important role in the pathogenesis of diseases, including sepsis and rheumatoid arthritis. However, easy, sensitive, and reliable detection systems are required to investigate the clinical signicance of HMGB1 in clinical samples for diagnosis and prognosis of diseases. Here, we describe sensitive ELISAs for the detection of HMGB1 in cell culture medium and cell lysates. However, these assays failed to reliably quantitate HMGB1 in serum and plasma when compared with immunoblot analysis. We found that serum/plasma components bind to HMGB1 and interfere with its detection by ELISA systems. In most serum/plasma samples investigated, including those from healthy individuals, we detected IgG antibodies binding to HMGB1. The titers of these antibodies correlated with the capacity of sera to interfere with the detection of recombinant HMGB1 by ELISA. Furthermore, HMGB1 coimmunoprecipitated with several proteins including IgG1, as identied by mass spectrometry. These HMGB1 interacting proteins are currently characterized and may contribute to complex formation, masking, and possibly, modulation of cytokine activity of HMGB1. J. Leukoc. Biol. 81: 6774; 2007.
Key Words: autoantibodies ELISA systemic lupus erythematosus rheumatoid arthritis

INTRODUCTION
High mobility group box 1 protein (HMGB1) is a ubiquitously expressed, highly conserved chromosomal protein. It consists of two DNA-binding domains, the HMG box A and B and a highly, negatively charged C-terminal domain, containing 30 repetitive aspartic and glutamic acid residues [1]. As a nuclear protein, HMGB1 binds to double-stranded, single-stranded, and distorted DNA without sequence specicity [2, 3]. It stabilizes the nucleosomal structure and mediates bending of
0741-5400/07/0081-0067 Society for Leukocyte Biology

the DNA, thereby facilitating binding of certain transcription factors including steroid hormone receptors [4]. Recent studies of endotoxemia and sepsis identied extracellular HMGB1 as a potent, proinammatory cytokine. HMGB1 is actively secreted from LPS- or TNF-activated macrophages/monocytes, pituicytes, and other cells [5 8]. In addition, HMGB1 is released passively by damaged or necrotic cells. Extracellular HMGB1 can bind to cell surface receptors such as the receptor for advanced glycation end products, TLR2 and -4, and possibly, to as-yet unknown receptors [4, 9]. Released HMGB1 by itself acts as a strong mediator of macrophage activation and induces secretion of proinammatory cytokines such as TNF-, IL-1, IL-6, IL-8, MIP-1, and MIP-2. In addition, HMGB1, released from necrotic cells, induces up-regulation of activation markers (HLA-DR, CD83, CD80, and CD86) on dendritic cells [10]. In contrast, apoptotic cells retain HMGB1, which is attached tightly to hypoacetylated chromatin. Therefore, HMGB1 is not released from apoptotic cells and does not induce inammation [11]. Extracellular HMGB1 has been detected in various pathological conditions. HMGB1 levels are increased markedly during severe sepsis in the blood of humans and animals. It is important that administration of neutralizing HMGB1-specic antibodies prevented lethality from established sepsis in animals [5]. Moreover, serum HMGB1 levels were signicantly higher in septic patients who did not survive [5]. Hemorrhagic shock is also associated with signicantly increased serum levels of HMGB1, even in the absence of infection and endotoxemia [12]. Intratracheal administration of HMGB1 into mice induces acute lung inammation with accumulation of neutrophils, edema, and production of proinammatory cytokines [13]. Furthermore, HMGB1 can be detected in the serum and alveolar uid after pulmonal injury of humans and mice. Expression of HMGB1 is up-regulated in almost every tumor type and is particularly increased in epithelial neoplasms. Increased serum levels of HMGB1 in cancer patients are associated with a worse prognosis [14].

1 Correspondence: IZKF Research Group 2, Nikolaus-Fiebiger-Center, Glu ckstrasse 6, 91054 Erlangen, Germany. E-mail: rvoll@molmed.unierlangen.de Received April 21, 2006; revised September 1, 2006; accepted September 4, 2006. doi: 10.1189/jlb.0306196

Journal of Leukocyte Biology Volume 81, January 2007 67

HMGB1 is also a mediator of chronic inammatory diseases. The presence of cytoplasmic and extracellular HMGB1 has been reported in experimental arthritis models as well as in human rheumatoid arthritis (RA). Elevated levels of HMGB1 are also present in synovial uid from RA patients [15, 16]. Systemic application of an antagonistic A box domain or neutralizing HMGB1-specic antibodies ameliorated collagen-induced arthritis in rodents [15]. Anti-HMGB1 antibodies were detected in plasma of patients with several autoimmune diseases such as systemic sclerosis, juvenile idiopathic arthritis, ulcerative colitis, and systemic lupus erythematosus (SLE); however, the clinical relevance of these ndings remains to be elucidated [1719]. All these data indicate that HMGB1 might be a diagnostic marker, a potent mediator, and a therapeutic target in several inammatory, autoimmune, and malignant diseases. However, so far, detection of HMGB1 is usually performed by timeconsuming, semiquantitative Western blotting. An easy, sensitive, and reliable assay for quantication of HMGB1 in different biological uids would be extremely helpful for further clinical investigations. Here, we describe serum/plasma factors that interfere with the detection of HMGB1 by our sandwich ELISAs, precluding the quantication of circulating HMGB1.

Sandwich ELISA for HMGB1

To detect HMGB1 in cell culture medium and serum/plasma samples, an ELISA was performed. Maxisorp polystyrene 96-well plates (Nunc, Roskilde, Denmark) were coated with one of the following anti-HMGB1 capture antibodies: 0.75 g/ml antibody (Cat. No. 556528, BD PharMingen, San Diego, CA; supply has been discontinued), 1.5 g/ml antibody (Cat. No. 07-584, Upstate Biotechnology, Lake Placid, NY), 1.5 g/ml antibody (Cat. No. sc12523, Santa Cruz Biotechnologies, CA), and 0.75 g/ml mAb (Cat. No. ab12029-100, Abcam, Cambridge, UK), diluted in PBS overnight at 4C. The wells were blocked for 2 h with 1% BSA (Roth, Karlsruhe, Germany) in PBS at room temperature. Prior to adding the samples, plates were washed with PBS containing 0.05% Tween 20 (the same washing buffer was used for each washing step). Plasma/serum samples from SLE patients and healthy donors (1:2 or 1:4 dilutions in PBS) and HMGB1 standards were added to each well (100 l/well) in duplicates. To establish the standard curve, recombinant (r)HMGB1 (Sigma Chemical Co., St. Louis, MO) was serially diluted in blocking buffer (1% BSA in PBS) to the following concentrations: 300, 100, 33.33, 11.11, 3.70, 1.23, and 0.41 ng/ml. After overnight incubation, the plates were washed, and 100 l biotinylated anti-HMGB1 antibody (Cat. No. MAB1690, R&D Systems, Minneapolis, MN), diluted to the nal concentration of 0.75 g/ml in blocking buffer, was added and incubated for 1 h. After washing, 1:2000 dilution of streptavidin-alkaline phosphatase conjugate (Amersham Biosciences, Freiburg, Germany) in blocking buffer was added and incubated for 45 min. After a nal washing, detection was achieved through addition of substrate solution: 3 mM 4-nitrophenyl phosphate (Serva Electrophoresis, Heidelberg, Germany) in substrate buffer (9.7% diethanolamine/HCl, pH 9.8, 0.1% sodium azide, 5 mM magnesium chloride). The absorbance at 405 nm was measured using a microplate spectrophotometer SPECTRA max 190 (Molecular Devices Corp., Boston, MA). Concentrations of HMGB1 in the samples were calculated using Softmax software. Biotinylation of the detecting antibody was performed using EZ-Link SulfoNHS-LC-biotin reagent (Pierce) following the manufacturers instructions.

MATERIALS AND METHODS Cell culture

Cell culture medium and supplements were obtained from Invitrogen Life Technologies (Karlsruhe, Germany). All cells were grown in a 5% saturated CO2 atmosphere at 37C. The human T cell lymphoblast-like cell line Jurkat (TIB-152) was grown in complete RPMI medium (RPMI 1640 containing 10% FBS, 2 mM L-glutamine, 20 mM Hepes, 1 mM sodium pyruvate, 100 U/ml penicillin, and 100 g/ml streptomycin). PBMCs were isolated from heparinized peripheral blood from healthy donors by Ficoll density gradient centrifugation (Lymphoot, Biotest, Germany). After three washes with PBS, the cells were resuspended at 2 106 cells/ml in complete DMEM, and 1 ml of the cell suspension was added per well of 48-well plates. After incubation for 2 h, the cell layer was washed twice with PBS to remove nonadherent PBL. The adherent monocytes were cultured in DMEM containing 20% autologous serum plus 100 U/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine to generate human monocyte-derived macrophages (HMDM). After 7 days, medium was replaced with serum-free OPTI-MEM. HMDM were stimulated with different concentrations of LPS or IFN-. Cell culture supernatants were harvested after 24 h and were kept at 20C or analyzed directly by ELISA.

ELISA for anti-HMGB1 antibodies

Autoantibodies directed to HMGB1 were assayed by ELISA as described by Sobajima et al. [19]. Briey, Maxisorp polystyrene 96-well plates were coated with 50 l per well rHMGB1 at 1 g/ml in PBS and incubated overnight at 4C. The plates were blocked with 5% BSA in PBS for 2 h. Serum samples, diluted 1:50 in blocking buffer, were added in duplicates (100 l/well) and incubated for 2 h at room temperature. After 5 washes, 100 l HRP-conjugated goat anti-human IgG (heavylight chains; Caltag Laboratories, Burlingame, CA), diluted at 1:3000, was added to each well and incubated for 1 h at room temperature. After washing, antibodies were detected using O-phenylendiamine. The reaction was stopped with 2 M sulfuric acid. The absorbance at 490 nm was measured using a microplate spectrophotometer (SPECTRA max 190, Molecular Devices). For calculation of anti-HMGB1 antibody levels, the mean OD of three wells coated with human serum albumin was subtracted from the OD of each sample as background value (nonspecic binding); this OD, corrected for individual background, is termed corrected sample cOD. Antibody titers were expressed as units, calculated based on two reference sera/ plasmas with high (S1) and low (S2) ELISA reactions against HMGB1. One unit was dened by subtracting the cOD value of S2 from the cOD value of S1 and dividing the result by 100 [1 unit(cODS1 cODS2)/100]. These reference sera/plasmas were used as internal standards in each assay.

Blood samples
Venous blood was collected from healthy donors and patients with SLE or RA into EDTA-containing tubes (Monovette, Sarstedt, Nu mbrecht, Germany) for collecting plasma or into tubes without anticoagulants for serum (Monovette). Centrifugation was carried out at 2000 g for 30 min at 4C. All samples were stored at 20C.

SDS-PAGE and Western blot analyses

Plasma/serum (3 l) from healthy controls or patients with SLE and RA were diluted with 72 l RIPA buffer and heated at 95C for 5 min in SDS-loading buffer. For immunodetection, the proteins were separated by 12% SDS-PAGE and transferred onto an Immobilon polyvinylidene diuoride membrane (Millipore, Bedford, MA). The membrane was blocked at room temperature for 1 h with 25 mM Tris-HCl, 150 mM NaCl, pH 7.4, and 0.05% Tween-20 (TBS-T) containing 3% BSA. The primary anti-HMGB1 antibody (0.25 g/ml, Medical and Biological Laboratories, Nagoya, Japan) was applied for 2 h at room temperature, followed by four 15-min washes with TBS-T. Blots were then incubated at room temperature for 1 h with a HRP-conjugated, anti-mouse IgG

Preparation of cell lysates

Cell lysates were prepared by homogenization of 1 107 Jurkat cells in 500 l radioimmunoprecipitation assay (RIPA) buffer (10 mM sodium phosphate, 150 mM NaCl, 2 mM EDTA, 0.1% SDS, 1% sodium deoxycholate, 1% Nonidet P-40, 1 mM NaF, pH 7.5). Cell debris was removed by centrifugation. Protein concentrations were determined with a bicinchoninic acid protein assay (Pierce, Rockford, IL). Whole cell lysates were stored at 20C or used directly for ELISA and Western blot.

68

Journal of Leukocyte Biology Volume 81, January 2007

http://www.jleukbio.org

antiserum (Jackson Immunoresearch, West Grove, PA). After washing, proteins were detected using ECL reagents (Amersham Biosciences).

RESULTS Detection of recombinant intracellular and extracellular/secreted forms of HMGB1 by sandwich ELISA
To develop a specic and sensitive ELISA for detection of HMGB1, we have tested seven different immunoassay formats. We used six commercially available mAb and polyclonal antibodies against HMGB1. First, these antibodies were tested for their specic recognition of human HMGB1 in cell lysates by immunoblotting (Fig. 1A and data not shown). Next, we investigated which antibodies also recognize the nondenatured protein by ELISA using prokaryotically expressed human rHMGB1 as standard. Four antibody combinations, which detect rHMGB1 in a sandwich ELISA, were established. As capture antibodies, polyclonal HMGB1 antibodies from Upstate Biotechnology, Santa Cruz Biotechnologies, and BD PharMingen and a mAb from Abcam could be used. As detection antibody, the HMGB1 mAb from R&D Systems worked best with all capture antibodies. Then we optimized various parameters, including concentrations of capture and biotinylated detection antibodies, sample dilution, incubation time, and temperature. Using human rHMGB1 for the standard dilutions,

Coimmunoprecipitation
Coimmunoprecipitations were performed using Seize X mammalian immunoprecipitation kit (Pierce), according to the manufacturers instructions with minor modications. Briey, 7.5 g anti-HMGB1 (R&D Systems) and 7.5 g isotype control antibodies (murine IgG2b, Sigma Chemical Co.) were immobilized to 25 l protein G-sepharose support using cross-linker disuccinimidyl suberate. Serum/plasma (100 l) and 400 l 0.1 M Tris/HCl, pH 7.5, 150 mM NaCl, and 0.05% Tween 20 (TNT buffer) or 100 ng rHMGB1 preincubated with 1 mg/ml -globulin Cohn Fraction II (Sigma Chemical Co.) in PBS were added to the immobilized antibodies and mixed gently for 4 h at 4C. The samples were washed ve times with ice-cold TNT buffer by centrifugation using HandeeTM Spin Cup columns, and the precipitated proteins were eluted by adding 80 l elution buffer. Finally, the precipitates were heated at 95C for 5 min in SDS loading buffer and size-separated by 7.5% SDS-PAGE, and proteins were visualized by colloidal Coomassie blue staining.

Mass spectrometry (MS)

For identication, bands were excised and digested following the protocol of Schafer et al. [20]. Sample preparation for MALDI-MS was performed using ZipTips (Qiagen, Hilden, Germany) following the manufacturers protocol. For MALDI-MS and MALDI MS/MS analysis, an Ultraex I ToF/ToF MS (Bruker Daltonik, Bremen, Germany) was used. Data analysis was performed using the BioTools (Bruker Daltonik) and Mascot (Matrix Science, London, UK) software packages.

Fig. 1. Detection of intracellular as well as extracellular/secreted forms of HMGB1 by ELISA. (A) Specic recognition of HMGB1 by polyclonal antibodies from BD PharMingen (lane 1), Upstate Biotechnology (lane 2), and the mAb from R&D Systems (lane 3) using immunoblotting. Lysates of Jurkat cells were subjected to reducing SDS-PAGE, and blotted proteins were visualized with the various antibodies as indicated above. (B) Standard curve of the ELISA was linear over a wide range of human rHMGB1 using polyclonal antibody from BD PharMingen as capture antibodies and the mAb from R&D Systems for detection. (C) Standard curve of the ELISA was linear over a wide range of human rHMGB1 using polyclonal antibody from Upstate Biotechnology as capture antibodies and the mAb from R&D Systems for detection. (D) Detection of secreted HMGB1 by ELISA. HMDM were stimulated with IFN- or LPS. HMDM cultured without activating stimuli served as negative control. After incubation for 24 h, the levels of secreted HMGB1 in the cell culture medium were measured by ELISA. Capture antibody from Upstate Biotechnology and biotinylated detecting antibody from R&D Systems were used. (E) Detection of intracellular HMGB1 by ELISA. Jurkat cells were lysed in RIPA buffer. The total cell lysates were diluted in PBS, and concentrations of intracellular HMGB1 were assessed by ELISA. Capture antibody from Upstate Biotechnology and biotinylated detecting antibody from R&D Systems were used. (F) Detection of HMGB1 passively released from dead cells. For induction of secondary necrosis, Jurkat cells were treated with 2 M staurosporine for 48 h. Primary necrosis was induced by heat (56C for 30 min) followed by incubation for an additional hour at 37C. Supernatants of secondary and primary necrotic cells were tested for HMGB1 by ELISA. Capture antibody from BD PharMingen and detecting antibody from R&D Systems were used.

Urbonaviciute et al. Factors masking HMGB1

69

ELISAs, based on capture antibodies from BD PharMingen (Fig. 1B) and Upstate (Fig. 1C), resulted in good correlation coefcients for standard curves, ranging from 0.4 ng/ml to 300 ng/ml, according to linear regression analysis. The detection limit for rHMGB1 was slightly below 0.4 ng/ml using the HMGB1 antisera from BD PharMingen or from Upstate Biotechnology as capture antibodies and the monoclonal antiHMGB1 from R&D Systems as detection antibody. Next, we investigated if our ELISA systems can also detect non-rHMGB1 derived from human cells. To test the recognition of secreted HMGB1, which may be acetylated, we tested tissue culture supernatants from activated HMDM by ELISA and immunoblotting. HMDM were stimulated with different concentrations of IFN- or LPS. After 24 h incubation, cell culture supernatants were collected, and HMGB1 concentrations were assessed. In the absence of activating stimuli, HMGB1 was barely detectable in the culture medium, whereas stimulation with IFN- or LPS increased the secretion of HMGB1 in a dose-dependent manner (Fig. 1D). Then we analyzed if the ELISAs can detect intracellular HMGB1 in lysates from human Jurkat cells. HMGB1 was quantied successfully in sequential dilutions of total cell lysates as shown in Figure 1E. HMGB1 is passively released from primary necrotic cells [11] and as a complex with nucleosomes, from apoptotic cells undergoing secondary necrosis (data not shown). Therefore, we induced primary necrosis in Jurkat cells by heat and incubated the dead cells at 37C for an additional hour. Apoptosis was induced by staurosporine treatment for 48 h, at which time, virtually all cells had undergone secondary necrosis. Released HMGB1 could be detected in cell culture supernatants of primary and secondary necrotic cells as shown in Figure 1F. Our results demonstrate that sandwich ELISAs based on these antibodies are useful to detect actively secreted, intracellular, and passively released human HMGB1 in cell culture medium and cell lysates. It is important that the ELISAs could also detect complexed HMGB1 released from secondary necrotic cells after prolonged culture. Under these conditions, most of the HMGB1 remains tightly attached to nucleosomes (data not shown).

analysis (data not shown). Clinical evaluation did not indicate infections or other inammatory conditions. The HMGB1 concentrations detected in sera of normal individuals or patients with RA or SLE are shown in Figure 2A. In sera of most RA patients, we could not detect HMGB1 (range 0.4 4 ng/ml), whereas HMGB1 was detected in some healthy subjects (range 0.4 40 ng/ml) and some SLE patients (range 0.4 190 ng/ml; Fig. 2A). According to our ELISAs, HMGB1 concentrations in SLE patients and healthy controls were statistically, signicantly higher than those in RA patients. Apparent HMGB1 concentrations determined by ELISA in simultaneously obtained plasma samples were usually higher than in the corresponding serum samples (Fig. 2B). To ensure specicity of the ELISA systems, we identied sera (e.g., SLE5) negative for HMGB1 by Western blotting using three different antibodies (Fig. 2C and data not shown). This control serum was also negative when tested by the HMGB1 ELISA (Fig. 2C). HMGB1 concentrations of sera/ plasmas determined by ELISA and immunoblot showed a poor correlation, whereas both detection methods showed congruent results for cell culture supernatants and cell lysates. Especially HMGB1 concentrations determined by ELISA were often lower than those determined by immunoblot. Selected serum samples exemplifying congruent (SLE1, SLE5, SLE7, NHD2) and divergent (SLE3, SLE4, SLE6, SLE8) HMGB1 detections by immunoblot versus ELISA are presented in Figure 2C.

Serum/plasma components can interfere with rHMGB1 detection by ELISA

As we showed above, HMGB1 concentrations determined by ELISA did not correlate well with detection of HMGB1 by immunoblot. Often, we obtained false low or negative results by ELISA. Therefore, we asked whether serum components may interfere with the detection of HMGB1 by ELISA. To test this hypothesis, we added increasing amounts of rHMGB1 to serum/plasma from healthy individuals and SLE patients, which were all tested to be negative for endogenous HMGB1 by immunoblot. Analyses of rHMGB1-spiked serum/plasma samples by ELISA showed inadequate, low readings (Fig. 3, A and B). HMGB1 added to samples at a nal concentration of 10 ng/ml was completely undetectable in all serum/plasma samples investigated (not shown). Increasing concentrations of HMGB1 could largely overcome the sequestration by serum, indicating a saturation effect (Fig. 3B). The HMGB1-masking capacity differed in individual sera, and sera with weak and strong masking effect were observed among healthy subjects and patients with SLE. To exclude degradation of HMGB1 by serum/plasma components, we incubated rHMGB1-spiked plasma/serum samples at 4C and 37C for 30 min before analyzing them by immunoblot and ELISA. In addition, equivalent amounts of rHMGB1 diluted in PBS containing 1% BSA served as control. Immunoblot revealed identical amounts of HMGB1 in serum/plasma samples incubated at 4C and 37C and in controls. In contrast, the ELISA strongly underestimated the HMGB1 concentrations (data not shown). Masking was observed with all capture antibodies tested. These data clearly indicate that HMGB1 is not degraded within serum; rather, it is sequestered by one or more serum components.
http://www.jleukbio.org

Assessment of HMGB1 in serum/plasma samples: poor correlation of ELISA with immunoblot

HMGB1 has been shown to be elevated and to play an important pathophysiological role in conditions such as septic shock and arthritis. However, there is a need for easy, sensitive, and reliable detection systems. Hence, we tested if our ELISAs can be used to detect HMGB1 in serum and plasma of healthy individuals and patients with SLE and RA. The samples were analyzed in parallel by three ELISAs (capture antibodies from BD PharMingen, Upstate Biotechnology, and Santa Cruz Biotechnologies) and by Western blotting for the presence of HMGB1. It is surprising that we found HMGB1, not only in patients with inammatory diseases but also in a few healthy subjects. From several of these healthy subjects, we could obtain follow-up samples, which also contained detectable amounts of HMGB1 (Fig. 2A). For most healthy subjects, the positive HMGB1 ELISA results could be conrmed by immunoblot
70 Journal of Leukocyte Biology Volume 81, January 2007

Fig. 2. Assessment of HMGB1 in serum samples. (A) Sera were taken from normal, healthy donors (NHD, n22) and patients with SLE (n27) or RA (n20). HMGB1 concentrations were measured by ELISA. Bars indicate the mean values. Statistical analyses were performed using Students ttest (two-tailed for heteroscedastic, sample populations). HMGB1 concentrations in individual serum and plasma samples from patients with SLE and healthy controls were compared. (B) Apparent HMGB1 concentrations in simultaneously obtained plasma samples were usually higher than in the corresponding serum samples. (C) Poor correlation of ELISA with immunoblot. Serum samples were analyzed in parallel by ELISA (capture antibody from Upstate Biotechnology and biotinylated detecting antibody from R&D Systems) and immunoblotting (anti-HMGB1 antibody from Medical and Biological Laboratories). Data from selected sera exemplifying congruent (SLE1, SLE5, SLE7, and NHD2) and clearly divergent (SLE3, SLE4, SLE6, and SLE8) HMGB1 levels when determined by Western blot (WB) versus ELISA are presented.

Identication of serum components binding to HMGB1

It has been demonstrated that HMGB1, as a sticky protein, binds to many membrane molecules, such as proteoglycans (syndecan-1, sulphoglycolipids), thrombomodulin, and phospholipids [21, 22]. Our previous results suggested that HMGB1 may also interact with serum/plasma proteins. When investigating different serum fractions, we found that the component(s) interfering with HMGB1 detection were present mainly within the -globulin fraction (data not shown). To identify binding partners of HMGB1, we subjected HMGB1 containing human plasma samples to coimmunoprecipitations using mouse anti-human HMGB1 antibodies covalently linked to protein G sepharose. In parallel, we performed coimmunoprecipitations with -globulins from Cohn Fraction II preincubated with human rHMGB1, as this fraction displayed the strongest HMGB1 masking effect in ELISA. In both experimental settings, protein bands with an apparent molecular mass of 130 kDa and slightly more were detected (Fig. 3C). To identify these proteins, trypsin digestion followed by MS and ngerprint analysis identied several of these visible bands as diverse Ig chains of IgG-type antibodies. Conrming these results, the sequencing of a characteristic peptide repeatedly found in the digests revealed a fragment of

the constant region of IgG1-type heavy chains (data not shown). Also, Western blot analysis revealed the presence of human IgG, predominantly, but not exclusively, of the IgG1 isotype (data not shown). Coimmunoprecipitations also revealed other, less-prominent bands of lower molecular weight, which are currently under investigation. These ndings imply that HMGB1-specic antibodies contribute to the interference in ELISA detection. Therefore, we analyzed multiple plasma/sera samples for the presence of antibodies specic for HMGB1. The majority of samples from healthy individuals and SLE patients showed marked levels of HMGB1 antibodies. However, sera from SLE patients displayed signicantly higher levels of anti-HMGB1 compared with healthy individuals (Fig. 4A). To evaluate the afnity of these antibodies, we serially diluted sera derived from three patients with SLE and three healthy controls and analyzed them further by ELISA. The titration curves revealed a relatively high binding afnity of IgG antibodies from several individuals including healthy donors (Fig. 4B). It is most interesting that there was a correlation of anti-HMGB1 IgG antibody levels with masking activity of individual sera/plasma samples, further indicating a role of IgG antibodies for interference with ELISA detection of HMGB1 (Fig. 4C). Furthermore, we isolated IgG antibodies from a serum with high
Urbonaviciute et al. Factors masking HMGB1 71

Fig. 3. Serum/plasma components can interfere with detection of rHMGB1 by ELISA. Capture antibody from Upstate Biotechnology and biotinylated detecting antibody from R&D Systems were used. (A) rHMGB1 (30 ng/ml) was added to tenfold diluted sera from two different, healthy individuals and analyzed by ELISA. Mean values and SD were calculated from triplicates. (B) Increasing concentrations of rHMGB1 were added to sera from healthy individuals and SLE patients and analyzed by ELISA. The HMGB1 concentrations detected by ELISA are expressed as percentages of the corresponding input concentrations of HMGB1. (C) Coimmunoprecipitation of serum proteins with HMGB1. -Globulins from Cohn Fraction II spiked with 100 ng/ml rHMGB1 (lane 1) and a plasma sample from a patient with SLE (lane 4) were analyzed by coimmunoprecipitation using anti-HMGB1 antibodies covalently linked to protein G sepharose. Sepharose beads alone (lanes 2 and 5) or linked to isotype control antibody (lanes 3 and 6) served as negative controls. The precipitates were separated by semireducing SDS-PAGE on a 7.5% gel, and the proteins were visualized by colloidal Commassie staining. Arrows depict the bands analyzed by MS.

HMGB1-neutralizing activity using protein G afnity purication. We added these antibodies to HMGB1-spiked samples in a concentration equivalent to serum IgG concentrations. The puried IgG antibodies resulted in a marked inhibition of HMGB1 detection by ELISA, further conrming the role of anti-HMGB1 antibodies for interfering with HMGB1 detection. In contrast, antibodies isolated from a donor with a relatively low HMGB1-neutralizing activity displayed low inhibition of HMGB1 detection by ELISA (data not shown).

DISCUSSION
Here, we describe sensitive ELISAs for the detection of HMGB1 in cell culture supernatants and cell lysates. Using the polyclonal anti-HMGB1 antiserum from BD PharMingen or from Upstate Biotechnology, the detection limit of the ELISA was 0.4 ng/ml, which is lower than the detection limit of a HMGB1-ELISA described previously [23]. Testing serum and plasma samples, we often noticed that concentrations detected by ELISA were lower than expected from immunoblotting results.
72 Journal of Leukocyte Biology Volume 81, January 2007

Investigating serum components binding to HMGB1, we identied IgG antibodies complexed with HMGB1 within serum/plasma. Therefore, we tested sera and plasma samples for the presence of anti-HMGB1 antibodies. It is interesting that IgG and IgM antibodies binding to HMGB1 were found in most sera and plasma samples investigated, including those from healthy individuals. Moreover, the titers of antibodies against HMGB1 in serum correlated inversely with the ELISA detection of rHMGB1 added to the serum samples. These data imply a role of autoantibodies to HMGB1 for interfering with HMGB1 detection by ELISA. To reliably detect HMGB1 by ELISA within serum/plasma samples, we investigated several methods to dissociate preformed HMGB1 complexes. Acidication (pH 2.5) with subsequent neutralization, high salt, or addition of detergents such as SDS and Tween 20 improved detection but still did not result in exact quantication of HMGB1 serum concentrations by ELISA. Currently, we further optimize the ELISA system for HMGB1 quantication in serum/plasma. Previous reports described a high prevalence of autoantibodies against HMGB1 in patients with rheumatic diseases
http://www.jleukbio.org

Fig. 4. Autoantibodies to HMGB1 and their interference with HMGB1 detection by ELISA. (A) Anti-HMGB1 autoantibodies (Abs) of IgG class were determined by ELISA in human serum samples from healthy controls and patients with SLE. Students t-test (two-tailed for heteroscedastic sample populations) was used for statistical analysis. (B) Evaluation of the afnity of HMGB1 autoantibodies. Six sera derived from patients with SLE and healthy individuals were serially diluted, and HMGB1 autoantibodies were measured by ELISA. (C) Detection of HMGB1 by ELISA inversely correlates with titers of anti-HMGB1 autoantibodies in human serum samples. The diagram correlates antiHMGB1-antibody titers in sera from normal controls and patients with SLE with the interference of the respective sera with HMGB1 detection by ELISA (expressed as the percentage of spiked HMGB1 detected by ELISA). Capture antibody from Upstate Biotechnology and biotinylated detecting antibody from R&D Systems were used. Statistical correlation analysis was calculated according to Pearson correlation.

including SLE and RA [24, 25]. However, the existence of IgG and IgM antibodies binding to HMGB1 in a high percentage of healthy subjects has not been reported yet. Although titers of these antibodies are relatively low, and HMGB1 may not represent the primary antigen inducing these antibodies, HMGB1-binding antibodies might serve important biological functions, possibly limiting overwhelming cytokine release as a result of ample tissue damage or sepsis. Currently, we investigate if naturally occurring or disease-associated anti-HMGB1 antibodies modulate the proinammatory response induced by HMGB1. In fact, preliminary data indicate that anti-HMGB1 antibodies in human sera inhibit cytokine release from HMDM stimulated with a combination of IL-1 and HMGB1. The almost universal detection of HMGB1-binding antibodies in healthy subjects is remarkable and might be a result of crossreactivity and/or the sticky nature of HMGB1 [26]. The role of these antibodies for masking their target protein or modulating its biological activity needs further investigation. Also, other candidate proteins binding to HMGB1 were found by coimmunoprecipitations. These proteins are currently being characterized and may very well contribute to complex

formation, masking, and potentially, modulation of cytokine activity of HMGB1. To our knowledge, this is the rst description of serum components that bind to HMGB1 and interfere with its detection by ELISA. Complex formation may be crucially involved in the modulation of the cytokine activity of HMGB1. Detailed knowledge of the factors binding to HMGB1 and interfering with ELISA assays may help to develop new strategies for reliable quantication of this important proinammatory mediator in research and clinical routine laboratory testing.

ACKNOWLEDGMENTS
This work is supported by the Interdisciplinary Center for Clinical Research (IZKF, Project N2) and the German Research Society (Collaborative Research Centers SFB 643, Project B3). We thank Dr. Silke Meister, Kirsten Neubert, Benjamin Frey, and Damian Maseda for critical reading of the manuscript and their help during preparation of the manuscript.
Urbonaviciute et al. Factors masking HMGB1 73

REFERENCES
14. 1. Tsuda, K., Kikuchi, M., Mori, K., Waga, S., Yoshida, M. (1988) Primary structure of non-histone protein HMG1 revealed by the nucleotide sequence. Biochemistry 27, 6159 6163. 2. Bustin, M. (1999) Regulation of DNA-dependent activities by the functional motifs of the high-mobility-group chromosomal proteins. Mol. Cell. Biol. 19, 52375246. 3. Bustin, M., Reeves, R. (1996) High-mobility-group chromosomal proteins: architectural components that facilitate chromatin function. Prog. Nucleic Acid Res. Mol. Biol. 54, 35100. 4. Lotze, M. T., Tracey, K. J. (2005) High-mobility group box 1 protein (HMGB1): nuclear weapon in the immune arsenal. Nat. Rev. Immunol. 5, 331342. 5. Wang, H., Bloom, O., Zhang, M., Vishnubhakat, J. M., Ombrellino, M., Che, J., Frazier, A., Yang, H., Ivanova, S., Borovikova, L., Manogue, K. R., Faist, E., Abraham, E., Andersson, J., Andersson, U., Molina, P. E., Abumrad, N. N., Sama, A., Tracey, K. J. (1999) HMG-1 as a late mediator of endotoxin lethality in mice. Science 285, 248 251. 6. Chen, G., Li, J., Ochani, M., Rendon-Mitchell, B., Qiang, X., Susarla, S., Ulloa, L., Yang, H., Fan, S., Goyert, S. M., Wang, P., Tracey, K. J., Sama, A. E., Wang, H. (2004) Bacterial endotoxin stimulates macrophages to release HMGB1 partly through CD14- and TNF-dependent mechanisms. J. Leukoc. Biol. 76, 994 1001. 7. Rendon-Mitchell, B., Ochani, M., Li, J., Han, J., Wang, H., Yang, H., Susarla, S., Czura, C., Mitchell, R. A., Chen, G., Sama, A. E., Tracey, K. J., Wang, H. (2003) IFN- induces high mobility group box 1 protein release partly through a TNF-dependent mechanism. J. Immunol. 170, 3890 3897. 8. Wang, H., Vishnubhakat, J. M., Bloom, O., Zhang, M., Ombrellino, M., Sama, A., Tracey, K. J. (1999) Proinammatory cytokines (tumor necrosis factor and interleukin 1) stimulate release of high mobility group protein-1 by pituicytes. Surgery 126, 389 392. 9. Park, J. S., Svetkauskaite, D., He, Q., Kim, J. Y., Strassheim, D., Ishizaka, A., Abraham, E. (2004) Involvement of Toll-like receptors 2 and 4 in cellular activation by high mobility group box 1 protein. J. Biol. Chem. 279, 7370 7377. 10. Rovere-Querini, P., Capobianco, A., Scafdi, P., Valentinis, B., Catalanotti, F., Giazzon, M., Dumitriu, I. E., Muller, S., Iannacone, M., Traversari, C., Bianchi, M. E., Manfredi, A. A. (2004) HMGB1 is an endogenous immune adjuvant released by necrotic cells. EMBO Rep. 5, 825 830. 11. Scafdi, P., Misteli, T., Bianchi, M. E. (2002) Release of chromatin protein HMGB1 by necrotic cells triggers inammation. Nature 418, 191195. 12. Ombrellino, M., Wang, H., Ajemian, M. S., Talhouk, A., Scher, L. A., Friedman, S. G., Tracey, K. J. (1999) Increased serum concentrations of high-mobility-group protein 1 in haemorrhagic shock. Lancet 354, 1446 1447. 13. Ueno, H., Matsuda, T., Hashimoto, S., Amaya, F., Kitamura, Y., Tanaka, M., Kobayashi, A., Maruyama, I., Yamada, S., Hasegawa, N., Soejima, J., Koh, H., Ishizaka, A. (2004) Contributions of high mobility group box 15.

16.

17. 18. 19.

20. 21.

22.

23.

24. 25.

26.

protein in experimental and clinical acute lung injury. Am. J. Respir. Crit. Care Med. 170, 1310 1316. Lotze, M. T., DeMarco, R. A. (2003) Dealing with death: HMGB1 as a novel target for cancer therapy. Curr. Opin. Investig. Drugs 4, 14051409. Kokkola, R., Sundberg, E., Ulfgren, A. K., Palmblad, K., Li, J., Wang, H., Ulloa, L., Yang, H., Yan, X. J., Furie, R., Chiorazzi, N., Tracey, K. J., Andersson, U., Harris, H. E. (2002) High mobility group box chromosomal protein 1: a novel proinammatory mediator in synovitis. Arthritis Rheum. 46, 2598 2603. Taniguchi, N., Kawahara, K., Yone, K., Hashiguchi, T., Yamakuchi, M., Goto, M., Inoue, K., Yamada, S., Ijiri, K., Matsunaga, S., Nakajima, T., Komiya, S., Maruyama, I. (2003) High mobility group box chromosomal protein 1 plays a role in the pathogenesis of rheumatoid arthritis as a novel cytokine. Arthritis Rheum. 48, 971981. Ayer, L. M., Senecal, J. L., Martin, L., Dixon, G. H., Fritzler, M. J. (1994) Antibodies to high mobility group proteins in systemic sclerosis. J. Rheumatol. 21, 20712075. Rosenberg, A. M., Cordeiro, D. M. (2000) Relationship between sex and antibodies to high mobility group proteins 1 and 2 in juvenile idiopathic arthritis. J. Rheumatol. 27, 2489 2493. Sobajima, J., Ozaki, S., Uesugi, H., Osakada, F., Shirakawa, H., Yoshida, M., Nakao, K. (1998) Prevalence and characterization of perinuclear anti-neutrophil cytoplasmic antibodies (P-ANCA) directed against HMG1 and HMG2 in ulcerative colitis (UC). Clin. Exp. Immunol. 111, 402 407. Schafer, H., Nau, K., Sickmann, A., Erdmann, R., Meyer, H. E. (2001) Identication of peroxisomal membrane proteins of Saccharomyces cerevisiae by mass spectrometry. Electrophoresis 22, 29552968. Abeyama, K., Stern, D. M., Ito, Y., Kawahara, K., Yoshimoto, Y., Tanaka, M., Uchimura, T., Ida, N., Yamazaki, Y., Yamada, S., Yamamoto, Y., Yamamoto, H., Iino, S., Taniguchi, N., Maruyama, I. (2005) The Nterminal domain of thrombomodulin sequesters high-mobility group-B1 protein, a novel antiinammatory mechanism. J. Clin. Invest. 115, 1267 1274. Muller, S., Scafdi, P., Degryse, B., Bonaldi, T., Ronfani, L., Agresti, A., Beltrame, M., Bianchi, M. E. (2001) New EMBO members review: the double life of HMGB1 chromatin protein: architectural factor and extracellular signal. EMBO J. 20, 4337 4340. Yamada, S., Inoue, K., Yakabe, K., Imaizumi, H., Maruyama, I. (2003) High mobility group protein 1 (HMGB1) quantied by ELISA with a monoclonal antibody that does not cross-react with HMGB2. Clin. Chem. 49, 15351537. Andersson, U., Tracey, K. J. (2004) HMGB1 as a mediator of necrosisinduced inammation and a therapeutic target in arthritis. Rheum. Dis. Clin. North Am. 30, 627 637. Uesugi, H., Ozaki, S., Sobajima, J., Osakada, F., Shirakawa, H., Yoshida, M., Nakao, K. (1998) Prevalence and characterization of novel pANCA, antibodies to the high mobility group non-histone chromosomal proteins HMG1 and HMG2, in systemic rheumatic diseases. J. Rheumatol. 25, 703709. Dintilhac, A., Bernues, J. (2002) HMGB1 interacts with many apparently unrelated proteins by recognizing short amino acid sequences. J. Biol. Chem. 277, 70217028.

74

Journal of Leukocyte Biology Volume 81, January 2007

http://www.jleukbio.org