Published ahead of print on September 16, 2004, doi:10.1164/rccm.200402-188OC
© 2004 American Thoracic Society doi: 10.1164/rccm.200402-188OC
Contributions of High Mobility Group Box Protein in Experimental and Clinical Acute Lung InjuryDepartment of Anesthesiology and Intensive Care; Department of Anatomy, Kyoto Prefectural University of Medicine, Kyoto; Intensive Care Unit, Saiseikai Suita Hospital, Osaka; Department of Laboratory and Molecular Medicine, Faculty of Medicine, Kagoshima University, Kagoshima; and Department of Medicine, Keio University, Tokyo, Japan Correspondence and requests for reprints should be addressed to Satoru Hashimoto, M.D., Department of Intensive Care and Anesthesiology, Kyoto Prefectural University of Medicine, 465 Kajiicho, Kawaramachi-Hirokoji, Kamigyo-ku, Kyoto, Japan, 602–8566. E-mail: satoru{at}koto.kpu-m.ac.jp
This study was performed to examine the putative role of high mobility group box (HMGB) protein in the pathogenesis of acute lung injury (ALI). Observations were made (1) in 21 patients who were septic with ALI and 15 patients with normal lung function and (2) in a mouse model 24 hours after intratracheal instillation of lipopolysaccharide (LPS). The concentrations of HMGB1 were increased in plasma and lung epithelial lining fluid of patients with ALI and mice instilled with LPS. LPS-induced ALI was mitigated by anti-HMGB1 antibody. Although this protein was not detected in the plasma of control humans or mice, the concentrations of HMGB1 in lung epithelial lining fluid or in bronchoalveolar lavage fluid were unexpectedly high. The nuclear expression of HMGB1 was apparent in epithelial cells surrounding terminal bronchioles in normal mice, whereas its nuclear and cytoplasmic expression was observed in alveolar macrophages in LPS-instilled mice. Lung instillation of HMGB2 did not cause as much inflammation as HMGB1. Extracellular HMGB1 may play a key role in the pathogenesis of clinical and experimental ALI. However, its expression in normal airways is noteworthy and suggests that it also plays a physiologic role in the lung.
Key Words: acute respiratory distress syndrome • endotoxin • high mobility group B-1 protein • high mobility group B-2 protein • lipopolysaccharide
Despite the development in the last two decades of various pharmacologic interventions for the management of multiple organ dysfunction, including acute lung injury (ALI) and acute respiratory distress syndrome (ARDS), none has been approved for clinical use (1). Several drugs developed to reach specific targets, such as cytokine cascades, have failed to prolong survival (1). One major difficulty, when trying to target these inflammatory mediators, is their evanescence from the battlefield. Most proinflammatory cytokines, such as tumor necrosis factor-
Extracellular high mobility group box (HMGB) 1, first isolated approximately 30 years ago from the thymus as a chromosomal protein and recognized as a transcriptional factor (2), has recently been proposed as one of the late mediators of sepsis or lipopolysaccharide (LPS) endotoxin lethality (3–6). In mice, serum concentrations of HMGB1 rise within 8 to 32 hours after the administration of LPS (5), and the systemic administration of HMGB1 is lethal (5). Anti-HMGB1 antibodies are protective against the lethality of LPS in mice, even when their administration is delayed beyond the peak in tumor necrosis factor- We conducted a translational study to examine the participation of HMGB1 in the pathogenesis of ALI caused by sepsis. To test the hypothesis that HMGB1 plays a key role as a late mediator, we first measured its concentrations and those of its related compound, HMGB2, in the blood and lungs of patients presenting with septic ALI/ARDS, and in a mouse model of LPS-induced lung injury. We then studied the localization of the extracellular HMGB1 protein and examined the antiinflammatory effects of anti-HMGB1 antibodies in our mouse model of LPS-induced lung injury. The direct effects of HMGB1 and HMGB2 on the lung were also examined. Parts of this study have been presented elsewhere (15).
Expanded details of our methods and materials used in this study can be found in the online supplement. The clinical study was conducted at the hospitals of Keio University and Kyoto Prefectural University of Medicine. The Human Research Committees from each institution approved the study, and informed consent was obtained from each participant or immediate family member. Twenty-one patients were identified as ALI or ARDS prospectively according to the definitions of the American European Consensus Conference on ALI and ARDS (16). We also applied the lung injury score and criteria developed by the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference Committee for sepsis syndrome (17, 18). The characteristics of our patient population are summarized in Table 1. Control data were obtained from 13 men and 2 women (39 to 70 years old) who underwent bronchoscopy to identify causes of hemoptysis or to examine small, solitary, peripheral pulmonary nodules. Chest-computed tomography revealed no diffuse interstitial lung abnormality, and pulmonary function tests and SaO2 were normal in these control patients. Bronchoscopic microsampling was not performed if the patient met any of the criteria described previously (19). In patients with ALI/ARDS, bronchoscopic microsampling of the pulmonary epithelial lining fluid (ELF) was performed on Day 0 (onset of ALI) and at 1- to 3-day intervals unless the patient was extubated or had died. Day 0 was defined as the day of first bronchoscopic microsampling, performed within 24 hours after diagnosis of ALI. The bronchoscopic microsampling procedure has been previously described in detail (20, 21).
The animal experiments included 236 male Institute of Cancer Research mice. The Animal Care Committee of our university approved all procedures. To create the lung injury mouse model, 300 µg of LPS diluted in 60 µl of saline were slowly instilled into the left lung of anesthetized mice using animal gavage needle (LPS group). Mice instilled with 60 µl of saline into the left lung were used as the control animals (control group). Another group of mice were instilled with 200 µg of chicken anti-pig HMGB1 polyclonal antibody + 300 µg of LPS (anti-HMGB1 group). Because a preliminary time-course study showed that maximal lung injury occurs between 8 and 32 hours after the LPS instillation and a previous study that HMGB1 is expressed 12 to 48 hours after LPS administration (7), all animal experiments were performed 24 hours after the instillation. The permeability index, an index of alveolar epithelial and endothelial permeability, was measured by injecting 25 µg/100 µl of human serum albumin via a tail vein 1 hour before killing. At the time of killing, blood and bronchoalveolar lavage fluid (BALF) were collected. The permeability index, which reflects alveolar septal damage, was calculated as the ratio of the human albumin concentration in BALF to that in plasma. To study the direct effects of HMGB1 and HMGB2 on lung tissue, 1, 10, or 100 µg of pig HMGB1 or HMGB2 were instilled into the left lung of another group of mice. Immunohistochemical staining was performed by an indirect, two-step labeling technique with anti-rabbit HMGB1 antibodies. To localize the expression of HMGB1 in the airspace, we applied a double-labeled immunofluorescence histochemical technique using anti-Mac3 and anti-HMGB1 antibodies. Measurements of HMGB1 and Western blotting of HMGB1 and HMGB2 in plasma, ELF, and BALF were performed by a method described previously (9). The statistical analyses were performed with Instat3 (GraphPad Software Inc., San Diego, CA). Data are expressed as medians, and box-whisker plots were constructed. Between-groups comparisons were examined by nonparametric Mann-Whitney's U test. Among-groups comparisons were examined by Kruskal-Wallis nonparametric analysis of variance for factorial experiments, followed by Dunn's procedure for post hoc multicomparison analysis; p values of less than 0.05 were considered statistically significant.
Western Blot for HMGB1 and HMGB2 Between 8 and 15 randomly selected samples from patients with ALI/ARDS and from the in vivo experimental groups were examined to identify HMGB1 and HMGB2 protein by Western blotting (Figure 1). In patients with ALI/ARDS, HMGB1 and HMGB2 were strongly positive in both ELF and plasma in all samples (n = 8). In contrast, in most control patients (n = 12), HMGB1 was distinctively positive in ELF, although it was absent in plasma. Similarly, in LPS-instilled mice, HMGB1 was recovered in both BALF and plasma. However, HMGB2 was observed in some but not all the BALF samples (Figure 1, lane 11, n = 15). HMGB1 was also recovered in BALF from all control mice (n = 9), suggesting that it is normally present in the airways.
Clinical Study There were wide variations among patients with ALI/ARDS in the evolution of HMGB1, which peaked between Day 1 and Day 7. Therefore, the highest HMGB1 measurement made during that period was considered the peak value and compared with the measurement made at the onset of disease (Day 0). In the control group, all plasma HMGB1 concentrations remained below the detection limit. In contrast, in the ALI/ARDS group, the concentrations were high from the onset (median = 2.9 ng/ml) and peaked to a median of 13.7 ng/ml. The median peak HMGB1 concentration in the ELF of patients with ALI/ARDS (9,267 ng/ml) was significantly higher than in either the control group (median = 4,928 ng/ml, p < 0.05) or at the onset (median = 1,719 ng/ml, p < 0.05). On the other hand, the difference between the values measured in the control group and the ALI/ARDS group at the onset was not statistically significant (Figure 2). Because the HMGB1 concentrations in ELF of the control group were unexpectedly high, we collected several additional samples from patients without lung disease in other settings during general anesthesia. Their analysis confirmed the presence of high HMGB1 concentrations (data not shown). Persistently high concentrations of HMGB1 in plasma and ELF were observed during the acute to subacute phase of ALI/ARDS, and peaks occurred early in some patients and later in others. There was no correlation between HMGB1 concentrations and lung injury score or PaO2/FIO2 ratio. Significant differences in HMGB1 concentrations were detected neither between survivors and nonsurvivors of ALI/ARDS nor between 10 patients with ALI/ARDS associated with pneumonia and 11 patients with ALI/ARDS not associated with pneumonia (data not shown).
Animal Study HMGB1 concentrations in plasma and BALF. In the control group, the majority of plasma HMGB1 concentrations were below the detection limit (median = 0 ng/ml), whereas they were significantly higher in the plasma of LPS-instilled mice (median = 44.8 ng/ml, p < 0.05). In the control group, the HMGB1 concentrations in BALF were unexpectedly high (33.8 ng/ml; Figure 3), although they were significantly higher in the LPS-instilled group (130.0 ng/ml). Correcting for the diluting effect caused by the lavage procedure, we estimated the HMGB1 concentrations in ELF to be 50- to 100-fold higher than those in BALF in both control and LPS-instilled mice. Thus, the HMGB1 concentrations in the airways of mice were as high as those measured in the patients with clinical ALI/ARDS.
Measurements of the permeability index. The permeability index was low (median = 0.068) in the control group. It was significantly higher in group instilled with the 300 µg of LPS, as a result of destruction of the alveolar barrier (median = 0.326, p < 0.05). The concomitant administration of anti-HMGB1 antibody and LPS, 300 µg, mitigated the increase in permeability induced by LPS (median = 0.257, p < 0.05 vs. the LPS-instilled group and the control group; Figure 4).
Histopathologic findings. Destruction of the pulmonary cell walls, migration of inflammatory cells, and pulmonary hemorrhages were present in the LPS-instilled mice. No destructive change caused by the instillation of 0.9% saline was observed in the control group. The instillation of LPS caused the accumulation of neutrophils despite the concomitant administration of anti-HMGB1 antibody, although they were less abundant within the air spaces and the destruction of the pulmonary cell walls was attenuated compared with that observed after the instillation of LPS alone (Figure 5).
The injury to the pulmonary cells caused by 1 µg of HMGB1 was insignificant, and only minimal neutrophil infiltration was noted. However, a dose-dependent increase in neutrophil accumulation into the air spaces and in the destruction of the pulmonary cell walls with abundant hemorrhages was observed after the instillation of 10-µg and 100-µg doses. Figure 6 illustrates the histologic changes observed 24 hours after the instillation of 1 µg and 100 µg of HMGB1 into the mice lung. In contrast, the instillation of HMGB2, even in a dose of 500 µg, caused significantly less cell injury than the instillation of 100 µg of HMGB1 (Figure 6).
Immunohistochemistry. Immunohistochemical staining of HMGB1 showed abundant positive epithelial cells surrounding the terminal bronchioles of the control lungs. In the LPS-instilled group, a large number of HMGB1-positive alveolar macrophage-like cells were present besides the positive nuclear staining of epithelial cells observed in the control group (Figure 7). These cells, which were confirmed to be macrophages by double staining with fluorescein isothiocyanate-labeled Mac 3 and Rhodamine-labeled HMGB1, were positively stained for HMGB1 in both the nucleus and the cytoplasm (Figure 8).
There is abundant evidence supporting the role of HMGB1 as a late inflammatory mediator. Several investigators have reported a rise in HMGB1 concentrations in patients with septic and other disorders (9–11). However, no data have been provided regarding the involvement of HMGB1 in ALI/ARDS. Our colleagues have recently developed a sensitive enzyme-linked immunosorbent assay method that measures HMGB1 without simultaneous measurement of HMGB2 (9). Surprisingly, concentrations of HMGB1 were approximately 1,000-fold higher in ELF than in plasma. Whereas in control subjects the plasma concentrations of HMGB1 were negligible, those in ELF were as high as those in ELF from patients during the acute phase of the ALI/ARDS. Before launching our animal experiments, we suspected that HMGB1 in the airspace is not only a pathologic characteristic of ALI, but that it may also play a physiologic role in host defense. Zetterström and colleagues have reported that HMGB1, produced and stored intracellularly in the human adenoid gland, contributes to the local antibacterial barrier system of the upper respiratory tract (22). Therefore, the HMGB1 collected in our control patients may have been expressed physiologically in the upper respiratory tract. The recovery of HMGB1 in plasma was limited to patients with ALI/ARDS with sepsis. However, in this study, we found no clear correlation between concentrations of extracellular HMGB1 and various clinical outcomes. As a small number of patients were enrolled in our study, further investigations are warranted to examine whether HMGB1 in plasma or ELF is a reliable marker of disease severity. Although, on Western blot, HMGB2 was also detected in patients with ALI/ARDS, its role remains unclear and warrants further investigations. Because of obvious limitations in the deeper investigation of the specific participation of HMGB1 in patients with ALI, we used a mouse model to clarify the complex results of our clinical study. We first instilled HMGB1 intratracheally in mice to confirm that HMGB1 alone induces ALI in a dose-dependent manner. Destructive changes were present in the alveolar space after the administration of large doses of HMGB1, similar to those observed in LPS-instilled lungs. Our findings, in animals, of a direct injury to the pulmonary cells induced by HMGB1 are consistent with the report by Abraham and colleagues (3). They demonstrated that the intratracheal administration of recombinant HMGB1 caused acute inflammatory lung injury, neutrophil accumulation, and development of lung edema (3). The 215-amino acids sequence of HMGB1 is highly conserved among mammalian cells, with usually only two to three variations in the marginal amino acids. The B box of HMGB1, consisting of 74 amino acids (HMGB1 residues 89 to 162), is essential to confer its functional activity as a proinflammatory cytokine, and its tumor necrosis factor–stimulating activity localizes to 20 amino acids (HMGB1 residues 89 to 108) (23). These sequences are identical among pigs, mice, and humans. Consequently, porcine and mouse HMGB1 should have similar effects on the mouse immune system. In this study, only a very high concentration of HMGB1 caused pulmonary cell damage. From rough estimates, the instillation of 1 µg of HMGB1 would produce concentrations in the airspace comparable to those in ELF measured in our control human patients or in BALF of control mice. This suggests that extracellular HMGB1 normally present in healthy airways does not injure the lung cells, as opposed to higher concentrations released into the alveolar space by activated macrophages or necrotic cells. Although the structural homology of HMGB2 and HMGB1 is approximately 80%, intratracheal HMGB2 was not as injurious to the lung as HMGB1. Because in patients with ulcerative colitis the expression of HMGB2 is stronger than that of HMGB1 (9) and patients who were septic with ALI/ARDS had equivalent expressions of HMGB1 and HMGB2, combined measurements of HMGB1 and HMGB2 may be clinically useful. With respect to the instillation of LPS into the airway, although it is a model of self-limited lung inflammation, it reproduces several characteristics of sepsis-induced ALI (24, 25). Accordingly, with regard to the concentrations of extracellular HMGB1, this model closely reproduced our findings in patients with ALI/ARDS. The Western blot analysis confirmed that the profiles of HMGB1 in LPS-induced lung injury and in control mice were very similar to those observed in patients with ALI/ARDS and in control subjects, respectively. A trend toward a weaker expression of HMGB2 in plasma and BALF of mice with LPS-induced lung injury compared with that of patients with ALI/ARDS could be due to the different etiologies of the two disorders. As observed in patients, the plasma HMGB1 concentrations increased in mice within 24 hours after LPS instillation, whereas in BALF, the concentrations of HMGB1 were high even in control mice. Histologic examinations revealed that the nuclei of epithelial cells surrounding terminal bronchioles were positive for HMGB1 in control mice, which is consistent with the detection in this group of a certain amount of HMGB1 in BALF. In contrast, in LPS-instilled mice, a strong nuclear expression of HMGB1 was associated with cytoplasmic staining in most alveolar macrophages. These findings are consistent with previous reports. Gardella and colleagues observed that the activation of monocytes redistributes HMGB1 from the nucleus to cytoplasmic organelles via a nonclassical, vesicle-mediated, secretory pathway (26). On immunostaining of normal rat specimens, Kokkola and colleagues found that HMGB1 was primarily confined to the nucleus of synoviocytes and chondrocytes, whereas inflammatory synovial tissue from an arthritis rat model, as well as from humans with rheumatoid arthritis, contained large numbers of mononuclear cells with nuclear and cytoplasmic staining, along with high extracellular HMGB1 concentrations in synovial fluid specimens (11). Scaffidi and colleagues found that HMGB1 released from necrotic cells acts as a major trigger of inflammation, whereas apoptotic cells retain their HMGB1 tightly bound to the nuclear remnants, preventing an inflammatory response (27). These observations, along with ours, suggest that extracellular HMGB1 released from necrotic cells or from alveolar macrophages may play a pivotal pathogenetic role in ALI. Although our mouse data indicate that HMGB1 can initiate lung injury, there is still paucity of clinical information to show that HMGB1 is important in the ongoing pathogenesis of ALI/ARDS. We have also confirmed the previously reported protective effect of anti-HMGB1 antibodies against LPS-induced ALI (3). However, further studies are needed to determine whether the administration of anti-HMGB1 antibodies represents a safe and effective treatment of ALI, as constitutively present HMGB1 in the normal airways as demonstrated in our histologic evaluation may function in a host defense mechanism. It is also noteworthy that control patients with normal airways had no detectable HMGB1 in plasma, whereas the concentrations in their airspace were considerable. This suggests a leak of HMGB1 from the alveolar space into the blood stream after destruction of the alveolar capillary barrier at the time of ALI. It is also possible that other injured organs are the source of HMGB1, a hypothesis warranting further studies of the effects of HMGB1 in the pathogenesis of multiple organ failure. In conclusion, our measurements made in plasma and ELF specimens from patients with ALI/ARDS suggest, although do not prove, that the overexpression of extracellular HMGB1 plays a key role in the pathogenesis of ALI. This observation was confirmed in an LPS-instilled mouse model. However, the finding of HMGB1 expression in unaffected airways indicates that therapeutic attempts to suppress the activity of HMGB1 should be made cautiously.
H.U. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; T.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; F.A. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; Y.K. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; M.T. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; A.K. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; I.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.Y. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; N.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; J.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; H.K. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; A.I. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. The authors thank Professor Yoshifumi Tanaka from Kyoto Prefectural University of Medicine, and Dr. Satoru Fukinbara from Department of Biochemical Engineering and Science, Kyushu Institute of Technology for their helpful advice during the preparation of this manuscript.
Supported in part by a grant-in-aid for Fundamental Scientific Research from the Education Ministry of Japan 07670678 (A.I.), 13470326 (S.H.). This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
* These authors have contributed equally to the work presented in this manuscript. Received in original form February 12, 2004; accepted in final form September 12, 2004
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
or interleukin-1, are released early in the course of sepsis, usually within minutes or hours after exposure to inflammatory stimuli. Nevertheless, death may occur long after their blood concentrations have returned to baseline. 
p < 0.05 versus control.
