|Home | About | Journals | Submit | Contact Us | Français|
Heat shock proteins are generally regarded as intracellular proteins acting as molecular chaperones; however, Hsp72 is also detected in the extracellular compartment. Hsp72 has been identified in the bronchoalveolar lavage fluid (BALF) of patients with acute lung injury. To address whether Hsp72 directly activated airway epithelium, human bronchial epithelial cells (16HBE14o-) were treated with recombinant Hsp72. Hsp72 induced a dose-dependent increase in IL-8 expression, which was inhibited by the NF-κB inhibitor parthenolide. Hsp72 induced activation of NF-κB, as evidenced by NF-κB trans-activation and by p65 RelA and p50 NF-κB1 binding to DNA. Endotoxin contamination of the Hsp72 preparation was not responsible for these effects. Next, BALB/c mice were challenged with a single intratracheal inhalation of Hsp72 and killed 4 h later. Hsp72 induced significant up-regulation of KC, TNF-α, neutrophil recruitment, and myeloperoxidase in the BALF. A similar challenge with Hsp72 in TLR4 mutant mice did not stimulate the inflammatory response, stressing the importance of TLR4 in Hsp72-mediated lung inflammation. Last, cultured mouse tracheal epithelial cells (MTEC) from BALB/c and TLR4 mutant and wild-type mice were treated ex vivo with Hsp72. Hsp72 induced a significant increase in KC expression from BALB/c and wild-type MTEC in an NF-κB-dependent manner; however, TLR4 mutant MTEC had minimal cytokine release. Taken together, these data suggest that Hsp72 is released and biologically active in the BALF and can regulate airway epithelial cell cytokine expression in a TLR4 and NF-κB-dependent mechanism.
The role of heat shock proteins as an endogenous protective mechanism is well established (1, 2). One group of these highly conserved proteins is constitutively expressed, aiding in the assembly, stabilization, and translocation of cellular proteins. Additionally, when presented with a variety of stressors, such as heat, ischemia-reperfusion, oxidation, or endotoxin, the expression of inducible heat shock proteins is increased several fold, providing tolerance to a second, more severe, and potentially lethal stress (3, 4). In this way, heat shock proteins mediate cellular protection against diverse cytotoxic stimuli. More recently, however, the role of heat shock proteins in the extracellular compartment has become an area of increased interest. Hsp72, the inducible member of the 70-kDa HSP family (also known as HspA1A), has been identified in the extracellular compartment, and there is a growing body of evidence that extracellular Hsp72 can modulate the innate immune response. Extracellular Hsp72 has been shown to activate monocytes, macrophages, and dendritic cells, and up-regulate the expression of proinflammatory cytokines (5–7). Hsp72-induced cytokine expression in these cells appears to be mediated through the TLR4 complexes in a CD-14 dependent fashion leading to the activation of NF-κB as well as MAPKs (7–10).
TLR4 is known to be important in host defense, serving as a pattern recognition receptor and activator of the innate immune system in response to LPS. TLR4 is present on many cell types including airway epithelial cells (11, 12). It has been proposed that heat shock proteins could serve as endogenous ligands for TLR4, thus acting as a “danger signal” to the innate immune system at the site of tissue injury (13–17). This is especially intriguing as it would suggest that a distressed cell could warn neighboring cells of potential injury through release of Hsp72 (15).
Several recent studies have investigated the role of extracellular heat shock proteins as potential prognostic markers of inflammation in patients after severe trauma, myocardial injury, infection, or major surgery (10, 18–23). Although these studies suggest that increased levels of extracellular Hsp72 are associated with a poor prognosis, others have found improved outcomes in those patients with more elevated levels of extracellular Hsp72 (24, 25). For example, Ganter et al. (25) recently reported an increase in the levels of Hsp72 in the pulmonary edema fluid of patients with acute lung injury (ALI).3 Patients with hydrostatic pulmonary edema had Hsp72 levels of 160 ± 51 ng/ml in the bronchoalveolar lavage fluid (BALF) vs 603 ± 153 ng/ml in patients with ALI. Additionally, in this cohort, patients with impaired alveolar fluid clearance had lower levels of Hsp72 in the edema fluid (25). These results suggest that the presence of adequate levels of Hsp72 in the airways is associated with improved alveolar fluid clearance and may be protective in the setting of ALI.
As an important source of morbidity and mortality, there has been a great deal of interest in the role and balance of proinflammatory and anti-inflammatory mediators in the development of ALI. Cytokines such as TNF-α, IL-1, -4, -6, -8, -10, -13, substance P, platelet-activating factor (PAF), complement adhesion molecules, selectins and vasoactive mediators have all been implicated in the process of inflammation as well as in the chemokine stimulation and networking associated with ALI (26, 27). Central to this process is the transcription factor NF-κB, responsible for much of the gene regulation of these inflammatory mediators (26–28). Additionally, TLR4 are present on lung epithelium and have been implicated in the pathophysiology of both infectious and non-infectious lung injury (29, 30).
In this study, we investigate the effects of extracellular Hsp72 on the epithelium of the lung. Given its evolutionary conservation and emerging role in innate immunity, it is our hypothesis that Hsp72 can stimulate lung epithelium, inducing a parenchymal cell to produce inflammatory cytokines, thus creating a novel biological target for Hsp72 and highlighting its potential role as a “danger signal” in the lung.
Recombinant Hsp72 was synthesized as described (31). Endotoxin levels (110 EU/mg Hsp72 protein or 11 ng/mg Hsp72 protein) were independently measured at Charles River Laboratories.
An SV40-transformed human bronchial epithelial cell line (16HBE14o-) was grown to confluence before being serum deprived for 8 h (32). Cells were then treated with Hsp72 (100–1000 ng/ml) for 18 h. In some experiments, cells were pretreated with an NF-κB inhibitor (10 μM parthenolide) or polymyxin B (50 μg/ml) for 1 h before the addition of Hsp72. Supernatants were collected and analyzed for IL-8 by ELISA (Endogen).
Cells were treated with Hsp72 (300 ng/ml) for 4 h, and RNA was extracted using a standard TRIzol method of phenol extraction. Total RNA is converted to cDNA by reverse transcription using the Superscript First Strand Synthesis System kit (Invitrogen). The IL-8 and SHDA primers which designed to span an intron, and the conditions of the real-time run are as previously described (33). Each target gene (IL-8) is normalized to a housekeeping or reference gene (SDHA) using the calculation (E ref)Ct ref/(E tar)Ct tar; where E is the real-time efficiency of the reference (ref) or target (tar) gene reaction and Ct is the threshold cycle of the reference (ref) or target (tar) gene (34).
The mean Ct values for IL-8 in 16HBE14o- cells were: untreated (27.1 ± 0.5) Hsp72 300 ng/ml (25.0 ± 0.05) and Hsp72 1000 ng/ml (24.7 ± 0.2). The mean Ct values for SDHA in these cells were: untreated (21.6 ± 0.1) Hsp72 300 ng/ml (21.1 ± 0.08) and Hsp72 1000 ng/ml (21.75 ± 0.09).
16HBE14o- cells were transiently transfected with the NF-κB-TATA Luc reporter plasmid (Stratagene) and β-galactosidase (35). Serum-deprived cells were then treated with Hsp72 (300 ng/ml for 16 h). Luciferase and β-galactosidase activity were measured (32).
16HBE14o- cells were treated with Hsp72 (300 ng/ml for 1 h), and nuclear extracts were harvested as previously described (36). Lung samples were homogenized with a Polytron homogenizer (Brinkmann Instruments) in a buffer containing 0.32 M sucrose, 10 mM Tris-HCl (pH 7.4), 1 mM EGTA, 2 mM EDTA, 5 mM NaN3, 10 mM 2-ME, 20 μM leupeptin, 0.15 μM pepstatin A, 0.2 mM PMSF, 50 mM NaF, 1 mM sodium orthovanadate, and 0.4 nM microcystin. The homogenates were centrifuged (1000 × g, 10 min). The pellets were solubilized in Triton buffer (1% Triton X-100, 150 mM NaCl, 10 mM Tris-HCl (pH 7.4), 1 mM EGTA, 1 mM EDTA, 0.2 mM sodium orthovanadate, 20 μM leupeptin A, and 0.2 mM PMSF). The lysates were centrifuged (15,000 × g, 30 min, 4°C), and the supernatant (nuclear extract) was collected for evaluation of DNA binding of NF-κB.
An oligonucleotide probe encoding the consensus sequence of NF-κB was purchased from Santa Cruz Biotechnology, was labeled with [γ-32P]ATP using T4 polynucleotide kinase (Invitrogen) and purified in MicroBiospin chromatography column. The gel was run using 10 μg of nuclear protein as previously described (36). In some instances, Abs against p65 (RelA) or p50 (NF-κB1; Santa Cruz Biotechnology) were added (10 min at room temperature). Cold specific and nonspecific probes were added at 5× the concentration of the radiolabeled probe. Gels were transferred to Whatman 3M paper, dried under a vacuum at 80°C for 1 h, and exposed using a phosphor imager.
Six-week-old female BALB/c mice, C3H/HeJ (spontaneous mutation of TLR4), and C3H/HeOuJ (wild-type) mice were obtained from The Jackson Laboratory and housed in a virus-free animal facility. Mice were exposed to either Hsp72 (100 ng/40 μl), boiled Hsp72 (denatured), LPS (1.1 pg/40 μl), or endotoxin-free PBS by inhalation (37) and given a lethal dose of sodium pentobarbital 4 h later. Animal care was provided in accordance with National Institutes of Health guidelines and approved by the Cincinnati Children’s Hospital Medical Center Institutional Animal Care and Use Committee. To determine basal levels of Hsp72 in the airways, six BALB/c mice were given a lethal injection of sodium pentobarbital, and lungs were lavaged with HBSS, and the BAL fluid was analyzed for Hsp72 levels by ELISA (R&D Systems).
Lungs were lavaged with HBSS, BALF was clarified and the supernatant removed for cytokine analysis. Total cell numbers were counted, smears prepared with a Shandon Cytospin II (Fisher Scientific) and stained with Diff-Quick solution. BALF was analyzed by ELISA for KC and TNF-α levels (R&D Systems). Lungs were removed and either placed in formalin, sectioned and stained with H&E, or snap frozen for the myeloperoxidase assay or nuclear extraction.
Tracheas from six 6-wk-old mice were removed from the thyroid cartilage to the level of bifurcation and pooled (38). MTECs were grown to confluence and serum-depleted before treatment with Hsp72 (300 ng/ml). In some experiments, cells were pretreated with parthenolide (10 μM). Supernatant was collected and analyzed by IL-8 ELISA 18 h later.
The concentration of Hsp72 in BALF was reported as median (interquartile range). All other data were expressed as mean ± SEM and assessed by ANOVA. Differences were pinpointed using Student-Newman-Keuls’ test (Sigma Stat 3.1).
We hypothesized that exposure to recombinant Hsp72 would regulate IL-8 expression in human bronchial epithelial cells. To test this, we treated serum-deprived 16HBE14o- cells with increasing concentrations of Hsp72, which resulted in a dose-dependent increase in IL-8 mRNA. (Fig. 1A). To confirm that increased IL-8 mRNA resulted in protein production, we treated cells with increasing concentrations of Hsp72 and harvested supernatant for IL-8 ELISA. Hsp72 increased IL-8 protein abundance in a dose-dependent manner (Fig. 1B). The endotoxin level in Hsp72 was independently measured by Charles River Laboratories and found to be 110 EU/mg protein (or 11 ng/mg protein). Treatment with 300 ng/ml Hsp72 results in the addition of 3.3 pg/ml endotoxin to the cells. We have previously shown that addition of up to 100 ng/ml Escherichia coli-derived LPS had no effect on IL-8 release in 16HBE14o- cells (39). Nevertheless, we pretreated cells with polymyxin B before the addition of Hsp72 to bind LPS. Polymyxin B treatment did not attenuate Hsp72-induced IL-8 production in 16HBE14o- cells (Fig. 1B). Collectively, these data demonstrate that Hsp72-induced IL-8 gene expression in a human bronchial epithelial cell line and that the biological effects of Hsp72 cannot be accounted for by endotoxin contamination in the recombinant protein.
To investigate the role of NF-κB in Hsp72-induced IL-8 expression, we pretreated selected cells with parthenolide, an NF-κB inhibitor, and measured IL-8 production. Parthenolide abolished Hsp72-induced IL-8 synthesis (Fig. 2A). Another NF-κB inhibitor, isohelenin, had the same inhibitory effect (data not shown). To confirm NF-κB translocation to the nucleus, we transiently transfected cells with a luciferase-tagged NF-κB reporter plasmid. Hsp72 treatment increased NF-κB-luciferase activity (Fig. 2B). To determine whether Hsp72 induces binding of NF-κB to DNA, we incubated nuclear extracts from cells treated with or without Hsp72 with an oligonucleotide encoding the consensus NF-κB sequence. Incubation of cells with Hsp72-induced NF-κB-DNA binding. Furthermore, coincubation of nuclear extracts with Abs against p65 RelA and p50 NF-κB1 each induced a supershift of the DNA binding complex, confirming the specificity of the NF-κB-DNA complex (Fig. 2C). These data demonstrate that extracellular Hsp72-mediated expression of IL-8 is dependent upon activation of the NF-κB pathway.
To establish the baseline level of Hsp70 present in mouse airways, Hsp70 ELISA was performed on the BALF from untreated BALB/c mice. Levels of Hsp70 were found to be 0.048 ± 0.016 ng/ml (n = 6).
To determine whether extracellular Hsp72 could induce inflammation in the airway, in vivo, we performed a single intratracheal inhalation of endotoxin-free PBS or Hsp72 (100 ng/40 μl) in BALB/c mice. To control for the possibility of endotoxin contamination, another group of mice was administered denatured Hsp72 (boiled at 100°C for 1 h). Four hours following the single intra-tracheal challenge, mice were sacrificed and BALF was harvested for cytokine production and neutrophil infiltration. Hsp72 induced a dose-dependent increase in KC, the functional IL-8 in mice (Fig. 3A) and TNF-α (Fig. 3B) levels in the BAL fluid as well as increased neutrophil recruitment (Fig. 3C) and myeloperoxidase activity (data not shown). These effects were significantly decreased by boiling Hsp72, suggesting they are not due to endotoxin contamination. Because the lower dose of Hsp72 significantly increased airway inflammation, the remainder of the in vivo studies were performed using 100 ng/inhalation of Hsp72 to maintain physiologic and biologically relevant levels. Histological examination of the lung following a single inhalation of Hsp72 showed increased peribronchiolar inflammation (Fig. 4). These data suggest that inhaled Hsp72 induces significant in vivo neutrophilia and cytokine release in mouse airways and that endotoxin contamination is not mediating these effects.
To further investigate the role of endotoxin on mediating Hsp72-induced cytokine expression and neutrophil infiltration into the airways, we compared the effect of Hsp72 inhalation to that of LPS at the concentration in the recombinant Hsp72 (1.1 pg per inhalation). Inhalation of this LPS concentration alone had no effect on KC or TNF-α expression, nor did it have any effect on neutrophil influx (data not shown), further suggesting LPS alone is not responsible for the induced inflammatory response. Collectively, these data demonstrate that extracellular Hsp72, independent of contaminating LPS, is capable of inducing lung inflammation in vivo.
To investigate the role of TLR4 in Hsp72-induced inflammation, we performed a single intratracheal inhalation of Hsp72 in TLR4 mutant (C3H/HeJ) compared with wild-type (C3H/HeOuJ) mice. A single inhalation of endotoxin-free PBS or Hsp72 (100 ng/40 μl) was administered and the mice were killed 4 h later. Hsp72 inhalation induced a significant increase in KC and TNF-α in the BALF of wild-type mice (Fig. 5, A and B). This effect was not seen following inhalation of Hsp72 in TLR4 mutant mice. Similarly, we did not detect neutrophilia in TLR4-deficient mice as was noted in the wild-type mice (Fig. 5C). These data suggest that Hsp72 signaling in the lung is TLR4 dependent.
To prove that Hsp72 induced cytokine production in an NF-κB-dependent manner in whole lung, we treated wild-type and TLR4 mutant mice with Hsp72 and harvested the whole lung to determine whether Hsp72 induced NF-κB:DNA binding in vivo. Wild-type mice stimulated with Hsp72 had a significant increase in NF-κB:DNA binding compared with PBS treatment (Fig. 6). However, TLR4 mutant mice failed to increase NF-κB translocation to the nucleus and DNA binding following treatment with Hsp72, as determined by EMSA. These data clearly demonstrate that Hsp72 induces cytokine production in vivo via TLR4 and activation of NF-κB.
To investigate lung epithelium as a potential source of Hsp72-mediated lung cytokine production, we isolated tracheal epithelium from BALB/c mice. We found that treatment of primary MTEC with Hsp72 resulted in a significant increase in KC production as measured by ELISA (Fig. 7A). Furthermore, pretreatment with the NF-κB inhibitor parthenolide before the addition of Hsp72 eliminated this response, evidence of the critical role of the NF-κB pathway in this process.
Subsequently, tracheal epithelial cells were harvested from wild-type (C3H/HeOuJ) mice and treated ex vivo with Hsp72. KC levels were significantly increased following Hsp72 treatment (Fig. 7B). We compared these responses to those of TLR4 mutant mice (C3H/HeJ) and found that while there was no significant decrease in the baseline KC levels between the wild-type and mutant type, the tracheal epithelial cells from TLR4 mutant mice did not respond to Hsp72 treatment (Fig. 7B). Together, these data support the importance of the TLR4 and NF-κB pathways in Hsp72-dependent signaling. Additionally, these data verify that the inflammation-associated signaling properties of Hsp72 demonstrated in a cultured lung epithelial cell line are also applicable to primary mouse lung epithelium and suggest that airway epithelial cells are, in part, responsible for the observed increase in cytokine expression in the BALF of Hsp72-treated mice.
Although there is increasing evidence of the role of extracellular Hsp72 in propagating the inflammatory response to stress in cells of the innate immune system, there are no published data of its direct effects on parenchymal cells. We show here for the first time that extracellular Hsp72, when added to human bronchial epithelial cells, induced the inflammatory cascade, increasing IL-8 gene expression via activation of the NF-κB pathway. Furthermore, we show that this biological effect is operative in vivo, given that direct inhalation of Hsp72 in mice increased cytokine expression and neutrophilia in the airway. The central role of the lung epithelium in this process is supported by the demonstration that isolated MTECs responded to Hsp72 stimulation in the form of cytokine production. In the absence of TLR4, that response is lost both in vivo as well as in MTEC cultured from TLR4 mutants, further supporting the importance of these cells and their TLR4 receptors in initiation of the cytokine response to Hsp72. Thus, lung epithelium can now be classified as a biological target of extracellular Hsp72, along with cells of the innate immune system. The implication of this finding is that other parenchymal cells may respond to the “danger signal” of extracellular Hsp72.
One interesting concept that has arisen through investigations into extracellular Hsp72 is the idea of Hsp72 as an endogenous TLR4 ligand (16, 17). Given the shared TLR4 and NF-κB pathways between LPS and Hsp72, this concept may help to explain the similar pathophysiologic responses of infectious and noninfectious mechanisms of ALI. Any stress in the lung, either ischemia, heat, hypoxia, or others, could lead to the release of extracellular Hsp72 by an as yet unidentified cell type and mechanism, subsequently activating the TLR4 receptor pathway on neighboring cells and modulating the inflammatory cascade in a way that parallels the effect of infection and LPS stimulation. One limitation of this study is that we focused exclusively on the role of TLR4 in Hsp72-induced cytokine expression. It is possible that Hsp72 may interact with other TLRs, including TLR2, TLR7, or TLR9. In fact, studies have shown that Hsp72 binds to both TLR4 and TLR2 (17, 40). We did not investigate the role of TLR2 in this system and future work is needed to shed light on this important question. In addition, it is possible that other TLR ligands play a role in stabilizing Hsp72/TLR4 interactions. This was also not addressed in the current study but may shed some additional light on the mechanism by which Hsp72 binds TLR4.
Although this rationale might help explain the similar responses between LPS-induced lung injury and other forms of ALI, it also highlights the debate and controversy surrounding the role of LPS contamination in investigations into extracellular Hsp72. As LPS shares the TLR4 and NF-κB signaling pathways with Hsp72 and is known to contaminate most laboratory preparations of recombinant proteins, this is a reasonable concern. Because of this, we pursued multiple layers of confirmation as to the validity of our findings. The endotoxin levels of our Hsp72 product were measured independently and that amount added to cells without resultant increases in IL-8 expression. Additionally, we boiled Hsp72 (to denature the protein and leave LPS intact) which eliminated cytokine release in 16HBE14o- cells and BALB/c mice. Furthermore, polymyxin B pretreatment was performed to bind any residual LPS with no measurable decrease in Hsp72 induced IL-8 expression. Importantly, LPS signaling requires cofactors including LPS-binding protein, CD-14, and MD-2. It has been shown in primary cultures of human airway epithelium, little or no MD-2 is expressed (41). Although we did not measure MD-2 levels, we have previously shown that the 16HBE14o- cells are minimally responsive to LPS (39). In addition, we treated cells in serum-free conditions thus eliminating a major source of LPS-binding protein (42). These experiments lend credence to the theory that Hsp72 and not LPS is inducing cytokine expression in airway epithelium.
Other investigators have identified Hsp72 levels of 160 ± 51 ng/ml in the BALF of patients with hydrostatic pulmonary edema vs 603 ± 153 ng/ml in patients with ALI (25). Our initial mouse inhalation experiments were performed using 100 ng and 1 μg of Hsp72 per inhalation. We chose to perform the remaining experiments with 100 ng per inhalation, to keep the levels in a physiologic range. Given previously published Hsp72 levels recovered in BALF these doses seem to be appropriate and biologically relevant.
The source of extracellular Hsp72 in BALF remains somewhat unclear. Some authors have suggested that necrotic or apoptotic cells release Hsp72 into the extracellular environment (43, 44), but there is also evidence that active release from viable cells also occurs (6, 45–47). Furthermore, that release has been shown to involve an exosome-dependent, nonclassical protein secretory pathway in human PBMC and tumor cells (47, 48). Although both active and necrotic cell release mechanisms may be involved in extracellular Hsp72 signaling, the possibility of active release allows for a markedly inducible response to a variety of insults, providing amplification of a local danger signal to neighboring cells (49).
Through this investigation, we have found evidence that supports the presence and biological activity of extracellular Hsp72 in the lung. We have furthermore established that the airway epithelium itself is responsive to this extracellular Hsp72 and that this cytokine response is regulated through the TLR4 and NF-κB pathways. The clinical applicability of this finding requires further study. Although our data would suggest that extracellular Hsp72 is responsible for inducing and propagating inflammation, a process at the heart of the pathogenesis of lung injury, Ganter et al. (25) found extracellular Hsp72 to be a marker of improved alveolar fluid clearance and therefore recovery from lung injury. This and other clinical investigations purporting divergent effects of extracellular Hsp72 would suggest that the mere presence of Hsp72 in the extracellular milieu is not the only factor. Perhaps there exists a threshold of extracellular Hsp72 that is required to maintain adequate signaling, below which the cells are unprepared for the insult, and above which excessive inflammation and therefore increased injury occur. There remain many questions as to the secretion, function, and clinical significance of extracellular Hsp72 in the lung. Although many questions remain, our findings support the importance of extracellular Hsp72 as a functional mediator of inflammation on a parenchymal cell, expanding its breadth of impact, supporting its potential role as an endogenous danger signal, and further increasing the interest in extracellular Hsp72 as a prognostic marker and potential therapeutic target in lung injury as well as other organ system dysfunction.
1This work was supported by the National Institutes of Health Grants HL075568 (to K.P.), GM077432 (to D.S.W.), and GM061723 (to H.R.W.).
3Abbreviations used in this paper: ALI, acute lung injury; BALF, bronchoalveolar lavage fluid; PAF, platelet-activating factor; MTEC, mouse tracheal epithelial cell.
The authors have no financial conflict of interest.