High mobility group box 1 protein (HMGB1) is a non-histone nuclear protein with dual function. Inside the cell, HMGB1 binds DNA and regulates transcription, whereas outside the cell, it serves as a cytokine and mediates the late effects of LPS. The movement of HMGB1 into the extracellular space has been demonstrated for macrophages stimulated with LPS as well as cells undergoing necrosis but not apoptosis. The differential release of HMGB1 during death processes could reflect the structure of chromatin in these settings as well as the mechanisms for HMGB1 translocation. Since apoptotic cells can release some nuclear molecules such as DNA to which HMGB1 can bind, we therefore investigated whether HMGB1 release can occur during apoptosis as well as necrosis. For this purpose, Jurkat cells were treated with chemical inducers of apoptosis (staurosporine, etoposide, or camptothecin), and HMGB1 release into the medium was assessed by Western blotting. Results of these experiments indicate that HMGB1 appears in the media of apoptotic Jurkat cells in a time-dependent manner and that this release can be reduced by Z-VAD-fmk. Panc-1 and U937 cells treated with these agents showed similar release. In addition, HeLa cells induced to undergo apoptosis showed HMGB1 release. Furthermore, we showed using confocal microscopy that HMGB1 and DNA change their nuclear location in Jurkat cells undergoing apoptosis. Together, these studies indicate that HMGB1 release can occur during the course of apoptosis as well as necrosis and suggest that the release process may vary with cell type.
high mobility group box protein 1 (HMGB1) is a non-histone nuclear protein with dual function. Inside the cells, HMGB1 binds DNA and plays a role in transcriptional regulation (18). Outside the cell, HMGB1 serves as a cytokine and is a late mediator of the effects of LPS (46). For its cytokine activities to be manifest, HMGB1 must leave the nucleus for transit into the extracellular milieu (8, 13, 21, 41, 47). Although this translocation process was identified originally in cells treated with LPS, necrotic cells also release this protein (44). As such, HMGB1 has been implicated as a cause of inflammation secondary to necrotic cell death (43).
Although HMGB1 release occurs with both cell activation and cell death, the mechanisms for release are distinct (5, 6, 8, 13, 21, 32, 41, 43, 44, 48). Translocation during activation involves protein acetylation, with the change in charge reducing the interaction with chromatin, and translocation into the cytoplasm for uptake into vesicles. In this setting, HMGB1 secretion occurs from vesicles, with this process augmented by agents such as lysophatidylcholine (8, 21). With necrosis, the release process appears immediate, with HMGB1 separating from chromatin since its binding to DNA is weaker than that of histones (5, 6, 43, 44, 48). The permeability of the nuclear and cellular membranes with necrosis may allow a passive exit process.
In contrast to findings with necrotic cells, analysis of cells undergoing apoptosis has indicated that HMGB1 release does not occur during this process, even during late stages when cells may undergo secondary necrosis and could show breakdown in membrane permeability (43, 44). The failure of this release, which could explain why apoptotic death is noninflammatory (24, 33), contrasts with the release of other nuclear molecules during apoptosis (4, 23, 31). Thus recent studies in our laboratory have focused on DNA release during cell death. With the use of Jurkat cells as a model, our findings indicate that apoptotic, but not necrotic, cells release DNA in a time-dependent manner (14, 15). This release has been observed with all inducers of apoptosis tested. Furthermore, since this release can be blocked by caspase inhibitors, a relationship to apoptosis can be established. Since the binding of HMGB1 to chromatin can be increased during apoptosis, the failure to show externalization of HMGB1 for this process is surprising.
Among explanations for the divergent findings on DNA and HMGB1 release, we considered the possibility that nuclear molecule release may differ among cell types, with inducing mechanism also impacting on release. In these studies, we have therefore investigated HMGB1 release from the Jurkat T cell line undergoing apoptosis. In results presented, we show that during the course of apoptosis of Jurkat cells treated with various agents, HMGB1 was released into the media as assessed by Western blotting. This release was blocked by an inhibitor of apoptosis. These results thus contrast with previous studies indicating that HMGB1 remains sequestered in the nucleus even as cells progress to late apoptosis and secondary necrosis (44). Together with studies on DNA, these findings indicate that a variety of nuclear molecules may exit the cell during apoptotic death and that nuclear molecule release may vary with cell type, inducing stimulus as well as a stage in the death process.
MATERIALS AND METHODS
Jurkat (human T-cell leukemia) and U937 (human promonocytic) cell lines were obtained from the American Type Culture Collection (Manassas, VA) and maintained in medium consisting of RPMI 1640 (GIBCO, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (Invitrogen, Carlsbad, CA) and 20 μg/ml gentamicin (GIBCO) at 37°C in a 95% humidified atmosphere containing 5% CO2. HeLa S6 and Panc-1 cells were obtained from the cell culture facility of Duke University (Durham, NC) and were grown in Dulbecco’s modified Eagle’s medium (GIBCO) containing 10% FBS and 20 μg/ml gentamicin. All reagents were purchased from Sigma-Aldrich (St. Louis, MO) unless noted.
Induction of apoptosis or necrosis.
Apoptosis was induced by treating 10 × 106 cells in 4 ml of medium with 30 μg/ml etoposide, 10 μg/ml camptothecin, 5 μM staurosporine (STS), 70 μM cycloheximide (CHX), or 70 μM CHX and 2 ng/ml tumor necrosis factor (TNF)-α. To assess the effects of apoptosis, we first treated cells with a 30-min preincubation with 100 μM caspase inhibitor I [Z-Val-Ala-Asp(OMe)-CH2F (Z-VAD-fmk); Calbiochem, San Diego, CA] for Jurkat and Panc-1 cells or 50 μM Z-VAD-fmk for U937 cells. Cells were treated with inducers of apoptosis for the indicated times. In some experiments, HeLa cells were also treated with 200 ng/ml trichostatin A (TSA; Alexis Biochemicals, San Diego, CA) for 30 h. For induction of necrosis, cells were treated with heat to 56°C for 30 min or three cycles of freeze-thaw. Supernatants were collected from necrotic or apoptotic cells by centrifugation at 500 g for 5 min and stored at −80°C until used.
Detection of cell death.
Cell death was detected by annexin V-fluorescein isothiocyanate (annexin V-FITC) and propidium iodide (PI) staining of necrotic and apoptotic cells. Cells were washed in PBS and resuspended in 100 μl of binding buffer containing 5 μl of annexin V-FITC (Molecular Probes, Carlsbad, CA) and 1 μg/ml PI and then incubated for 10 min at room temperature (RT) in the dark. Cells were analyzed using a FACScan (Becton Dickinson, San Jose, CA). Data were analyzed using CELLQuest software (Becton Dickinson).
Caspase-3 was assayed using a kit (Molecular Probes E12184). Briefly, supernatants of treated cells were added to caspase-3 substrate solution in a black 96-well microtiter plate (Costar; Corning, Corning, NY). Samples were incubated at RT in the dark for 30 min. Relative fluorescence units were determined using a TECAN GENios microplate reader with an excitation wavelength at 485 nm and an emission wavelength of 535 nm.
Western blotting for HMGB1 detection.
Conditioned media from cells were centrifuged through Centricon (Millipore, Billerica, MA) 10-kDa filter devices for 1 h at 3,000 g. Supernatants were collected from the top reservoir of the filter device and spun through an Ultrafree-MC (Millipore) 100-kDa filter device for 15 min at 4,000 g. Equal volumes of elute were fractioned by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Invitrogen Life Technologies). Membranes were probed with a 1:3,000 dilution of purified rabbit anti-HMGB1 (Pharmingen, San Jose, CA). Bound antibody was detected using a 1:1,000 dilution of goat anti-rabbit horseradish peroxidase-conjugated antibody (Pierce, Rockford, IL) and visualized by chemiluminescence (Supersignal West Femto maximum sensitivity substrate; Pierce) according to the manufacturer’s recommendations and viewed on a Fluorchem 8900 instrument (Alpha Innotech, San Leandro, CA).
Confocal microscopy and immunocytochemistry.
Cells were induced to undergo apoptosis in LabTek II chambers (Nalgene Labware, Rochester, NY). Cells were fixed and permeabilized using the Cytofix/Cytoperm Plus with GolgiStop kit (Pharmingen) according to the manufacturer’s recommendations. In brief, cells were fixed and permeabilized for 20 min at 4°C. They were then washed three times with wash buffer and then incubated in 0.5 μg/ml anti-HMGB1 for 20 min at 4°C and washed. Next, 10 μg/ml goat anti-rabbit IgG Alexa Fluor 488 (Molecular Probes) were added for 20 min and cells were washed. Finally, cells were incubated with 1 μg/ml DRAQ5 (Alexis Biochemicals) for 10 min at RT and washed. Cells were visualized on a LSM510 confocal microscope (Zeiss, Oberkochen, Germany).
Release of HMGB1 from apoptotic Jurkat cells.
As shown in previous studies (14), Jurkat cells undergoing apoptosis can release DNA in the media in a time-dependent process. Since HMGB1 interacts strongly with chromatin, we have therefore investigated whether HMGB1 may also be released by apoptotic cells. To assess the release of HMGB1 during the process of apoptosis, we treated Jurkat cells with STS, etoposide, or camptothecin and analyzed culture supernatants. Detection of apoptosis was measured by the staining of cells with annexin-V-FITC and PI for analysis on a flow cytometer. As shown in Fig. 1, STS, etoposide, and camptothecin all induced apoptosis in Jurkat cells as demonstrated by the increased number of annexin+/PI− and annexin+/PI+ cells. Under these conditions, Z-VAD-fmk, a caspase inhibitor, reduced the level of apoptosis in cells treated with etoposide and camptothecin, with the effects of Z-VAD-fmk on apoptosis of cells treated with STS reflected in a delay in the time course. The delay is evident at the 4-h time point when 29% of STS-treated cells were annexin+/PI−, whereas only 3% of cells treated with Z-VAD-fmk and STS were annexin+/PI−.
To assess HMGB1 release into the media from treated cells, supernatants from apoptotic Jurkat cells were concentrated and analyzed on a SDS-PAGE gel. As shown in Fig. 2A, HMGB1 was present in the supernatants of STS-treated cells at 24 and 30 h of treatment in contrast to untreated Jurkat cells. Etoposide and camptothecin treatment also caused the release of HMGB1 into the supernatant at 30 h of treatment, whereas the addition of Z-VAD-fmk reduced the release of HMGB1. To show that the antibody detected HMGB1 of the appropriate molecule weight, we compared the supernatants of Jurkat cells treated with STS with supernatants of cells made necrotic by either three cycles of freezing and thawing or incubation at 56°C for 30 min. The HMGB1 band detected in media from Jurkat cells treated with STS was similar in size to the band from freeze-thawed cells or heat-treated Jurkat cells (Fig. 2B), although the amount of protein in the supernatants was less.
HMGB1 release by apoptotic HeLa cells.
The release of HMGB1 into the media of apoptotic Jurkat cells contrasts with previous results on this process as well as prevailing models on the role of HMGB1, which posit that even late in the process of apoptosis (i.e., secondary necrosis), this protein remains tightly bound in the nucleus (18, 44). Experiments were performed to repeat previous work of others showing that apoptotic HeLa cells do not release HMGB1 into the supernatants (44). HeLa cells were treated with CHX and TNF-α or the combination of CHX and TNF-α with TSA for the indicated times; TSA was included in these experiments in view of data showing that inhibition of histone deacetylase by TSA causes biochemical changes that allow the release of HMGB1 from apoptotic cells. Detection of apoptosis was measured by the staining of cells with annexin-V-FITC and PI (Fig. 3A) or by caspase-3 activity (Fig. 3B).
As these data indicate, HeLa cells treated with either CHX and TNF-α or the combination of these agents with TSA underwent apoptosis, whereas Z-VAD-fmk reduced the amount of apoptosis (Fig. 3, A and B). To determine whether HMGB1 was released into the supernatants of cells under these conditions, supernatants were concentrated and analyzed on SDS-PAGE gel. As indicated in Fig. 3C, HeLa cells treated with CHX and TNF-α released HMGB1 into the supernatants. HeLa cells treated with CHX, TNF-α, and TSA also released HMGB1 into the supernatants. In the presence of the caspase inhibitor Z-VAD-fmk, HMGB1 was not detected in the supernatants (Fig. 3C). In contrast to previous studies (44), these results indicate that during the process of apoptosis, HeLa cells release HMGB1 into the supernatant and that protein acetylation is not required for this process.
Additional experiments were therefore performed to assess potential differences between Jurkat and HeLa cells in this response. Since CHX and TNF-α induce apoptosis and Jurkat cells undergoing apoptosis release HMGB1, we measured HMGB1 release by Jurkat cells treated with CHX and TNF-α. The presence of apoptosis was confirmed by annexin-V-FITC and PI staining (data not shown). As shown in Fig. 3D, Jurkat cells, when treated with CHX alone or in combination with TNF-α, released HMGB1 into the supernatants at 24 and 30 h of incubation. These results confirm studies with HeLa cells and indicate that the combination of CHX and TNF-α can induce apoptosis and lead to HMGB1 release.
Release of HMGB1 by apoptotic Panc-1 and U937 cells.
To investigate whether cell lines other than Jurkat cells release HMGB1 during apoptosis, Panc-1 and U937 cells were induced to undergo apoptosis with STS or etoposide and supernatants were probed for HMGB1. As shown in Fig. 4, U937 cells (A) and Panc-1 cells (B) both released HMGB1 into the supernatants when treated with STS or etoposide at 4 and 30 h. Similarly to results with Jurkat cells, Z-VAD-fmk reduced the release of HMGB1 in cells treated with etoposide or STS. These results indicate that HMGB1 release is not confined to apoptotic Jurkat cells.
Location of HMGB1 in the nucleus of apoptotic Jurkat cells.
In cells treated with LPS, HMGB1 translocates from the nucleus to the cytoplasm (8, 13, 41). To determine whether a similar process occurs during apoptosis, we used confocal microscopy. For this purpose, apoptotic Jurkat cells that had been fixed and permeabilized were probed with a FITC-labeled antibody to HMGB1 and stained with a DNA dye. As shown by costaining with DNA, in control cells HMGB1 localized to the nucleus. In contrast, after 4 h of treatment of Jurkat cells with STS (Fig. 5A), staining for HMGB1 and DNA both displayed changes in nuclear location, with DNA condensing in areas where staining with anti-HMGB1 was reduced. By 24 h of STS treatment (Fig. 5B), the overall content of HMGB1 in cells appeared significantly reduced as demonstrated by the diminished staining with anti-HMGB1. Furthermore, the staining observed differed from that of DNA. Similar results were obtained with Jurkat cells treated with etoposide for 24 h. Together, these results indicate that in apoptotic cells, like LPS-stimulated macrophages, HMGB1 can show alterations in its content in the nucleus as well as its localization.
Results presented in this study provide new insights into the dynamics of nuclear molecule release during apoptotic cell death. Thus, using Jurkat cells as a model, we have shown that the induction of apoptosis by a variety of stimuli leads to the extracellular release of HMGB1. This process was time dependent and could be blocked by Z-VAD-fmk for cells treated with etoposide or camptothecin or delayed when cells were treated with STS. Furthermore, a change in the content and location of HMGB1 in the nuclear structure could be demonstrated by confocal microscopy. Thus, as apoptosis progressed, HMGB1 and DNA showed distinct patterns of localization, with staining by an antibody specific for HMGB1 also reduced. Together, these findings indicate that HMGB1 release may not be an exclusive feature of necrotic death and suggest that as apoptosis proceeds, at least in certain cells, HMGB1 release can occur in this process.
These findings differ from prior studies indicating that necrosis, but not apoptosis, leads to HMGB1 release (44). In these experiments, even during late apoptosis (i.e., secondary necrosis) HMGB1 release did not occur, perhaps reflecting molecular changes that anchored this protein to chromatin. Given the cytokine activities of HMGB1 (1–3, 5, 32, 47), this result could explain the differing immunologic properties of apoptotic and necrotic cells, with apoptotic cells considered anti-inflammatory and necrotic cells considered proinflammatory. The validity of this distinction, however, is dependent on the robustness of the results and the influence of variables such as the cell type tested, the inducing stimulus for apoptosis, and the physical or chemical treatment causing necrosis.
Previous studies on this topic used HeLa cells and fibroblasts as models, inducing apoptosis by treatment with TNF-α and CHX and necrosis with metabolic inhibitors (43, 44). In contrast, we used Jurkat cells treated with pharmacological inducers of apoptosis and showed that these cells, like cells induced to undergo necrosis by heat treatment or freeze-thawing (Fig. 2B), release HMGB1 over time in culture. This release was observed with all inducers of apoptosis. To assure validity of this result, we assessed HeLa cells using CHX and TNF-α to induce apoptosis and showed that under these conditions, HMGB1 was released; this result differs from previous studies of Scaffidi et al. (44), which indicated that even after prolonged incubation following induction of apoptosis, HMGB1 remained intracellular, with inhibition of protein acetylation necessary to cause its release. Furthermore, in our studies, Jurkat cells treated with CHX and TNF-α, U937 cells treated with STS or etoposide, and Panc-1 cells treated with STS or etoposide all released HMGB1.
Together, our results indicate that HMGB1 release is a feature of apoptosis of at least some cell types and likely occurs during late apoptosis, a stage that can be considered as secondary necrosis. In this stage, cells may display permeability changes, since they will allow entry of PI on one hand, as well as exit of molecules such as DNA, caspase-3, lactate dehydrogenase (LDH), cytochrome c, and cytokeratin fragments on the other (4, 16, 22, 23, 26–28, 30, 31, 42, 45). According to this model, HMGB1 would reflect nonspecific leakage of nuclear contents that occurs late in apoptosis despite cell shrinkage and chromatin condensation. In this regard, there is evidence that the lamin in the nuclear membrane is cleaved during apoptosis, eliminating the permeability barrier and allowing dispersal of molecules such as DNA or RNA into the cytoplasm and then the extracellular milieu (7, 19, 20, 35, 39).
In previous studies, the mobility of a fusion protein of HMGB1 and green fluorescent protein was assessed as fluorescence loss of photobleaching (FLIP analysis) in HeLa cells to determine intranuclear diffusion during apoptosis (44). These experiments indicated that during apoptosis, HMGB1 has very limited mobility, with a diffusion coefficient significantly lower than that of the nuclear proteins HMGN1, HMGN2, or NF1. These findings suggest that molecular changes in the nucleus occurring during apoptosis cause tight binding of HMGB1 to the chromatin. Our results suggest either that at least some HMGB1 remains sufficiently mobile in the cell to allow its exit as apoptosis proceeds or that tight binding diminishes over time. The more mobile HMGB1 protein could reside in the cytoplasm, since HMGB1 in some cells may have cytoplasmic as well as nuclear localization (10, 25, 34, 36). In this regard, in certain cell types, HMGB1 may remain intracellular even during necrosis. Thus, in polymorphonuclear cells, HMGB1 is found in cytoplasmic vesicles, where it is tightly bound in an insoluble form; in these cells, necrosis is insufficient to release HMGB1 (43).
Although permeability changes could lead to the exit of nuclear and cytoplasmic molecules during apoptosis, the shifts of these molecules may not simply be a passive release. Thus, in HBL100 breast cancer cells treated with paclitaxel to induce apoptosis, the release of CK18 cytokeratin fragment occurs significantly earlier than the release of LDH (45). Similarly, in Jurkat cells, the release of glutathione appears to result from transport as opposed to diffusion (17). Thus the release of molecules during apoptosis may be a selective process and results from more than one mechanism.
To the extent that permeability changes allow leakage of nuclear contents during apoptosis, the release of HMGB1 during late apoptosis may resemble that occurring during necrosis, with these processes distinct from that occurring during cell activation. Cell activation involves active secretion of HMGB1 that has been acetylated and translocated into vesicles (21). Although apoptosis involves extensive proteolytic cleavage and protein translocation, studies thus far do not show either cleavage of HMGB1 or movement of this protein into blebs (11, 12). These issues are under investigation, however, and we cannot exclude the possibility that at least some of the HMGB1 release occurs in the form of blebs or microparticles released from the cell membrane as the apoptotic cells breaks down.
Although this mechanism can explain the release of HMGB1 during apoptosis, it does not explain the differences in the behavior of DNA and HMGB1 during necrosis. Thus our findings indicate that DNA release does not occur in cultures of Jurkat cells treated with heat to induce necrosis (14, 15). Nevertheless, we observed release of HMGB1 release under these circumstances (Fig. 2B). The basis for this difference is not clear, although it appears likely that the agents we used to induce necrosis can cause either cross-linking or denaturation of chromatin in such a form that exit of DNA from the cell is not possible. We have used freeze-thawing to induce necrosis, but in our experience, cells treated this way are severely fragmented and become permeable. The most appropriate system for inducing necrosis in a “physiological” way is speculative. This issue is also relevant to the quantitative assessment of HMGB1 during different forms of death, because release with freeze-thawing may be much more extensive than that occurring with chemical agents.
Previous studies on HMGB1 release during death have suggested that the failure of apoptotic cells to release this protein could contribute to an anti-inflammatory state (7). To the extent that HMGB1 release promotes inflammation, our findings could suggest that the anti-inflammatory nature of apoptotic cells is transient and that as cells transit to late apoptosis, HMGB1 can exit cells and stimulate inflammation. Indeed, in preliminary experiments, we have shown that concentrated supernatants of Jurkat cells induced to undergo apoptosis by treatment with UV-B light can stimulate the production of nitric oxide and IL-12 by RAW 264.7 cells primed with interferon-γ; under these conditions, supernatants of necrotic cells were also active (Bell CW and Pisetsky DS, unpublished observations). Since these supernatants of apoptotic cells undoubtedly have many components, it is premature to conclude that HMGB1 is the active moiety. In this regard, HMGB1 may act in concert with other released molecules, including DNA, which is present in these supernatants. The interaction of HMGB1 with DNA could create a complex that is immunologically active, since HMGB1 promotes DNA uptake into cells, mimicking the immune stimulatory activity of other DNA with transfection agents (9, 29). These issues are under investigation.
Studies on systemic lupus erythematosus (SLE) have suggested that a failure to clear apoptotic cells can underlie the generation of autoimmune responses (37, 40), with these cells being effective immunogens by virtue of their display of nuclear molecules in blebs (11, 12). Our results further suggest that HMGB1 could contribute to this process, since it is a highly effective adjuvant, with both generation of these blebs as well as the release of HMGB1 occurring most readily in the absence of effective clearance as may occur in SLE. Indeed, skin lesions of patients with SLE have an increased levels of extracellular HMGB1; since this lesion is associated with increased apoptosis, the source of this material could be apoptotic cells as well as inflammatory cells present (38). Although the origin of extracellular HMGB1 in disease settings must be the subject of future investigation, the current studies suggest a broader perspective on the release of this protein and recognition of its appearance during apoptosis as well as necrosis.
This research was supported by the Veterans Affairs Medical Research Service, National Institute of Allergy and Infectious Diseases Grant AI-44808, and the Lupus Research Institute.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society