The Heat Shock Response and Acute Lung Injury

Abstract

All cells respond to stress through the activation of primitive, evolutionarily conserved genetic programs that maintain homeostasis and assure cell survival. Stress adaptation, which is known in the literature by a myriad of terms, including tolerance, desensitization, conditioning, and reprogramming, is a common paradigm found throughout nature, in which a primary exposure of a cell or organism to a stressful stimulus (e.g., heat) results in an adaptive response by which a second exposure to the same stimulus produces a minimal response. More interesting is the phenomenon of cross-tolerance, by which a primary exposure to a stressful stimulus results in an adaptive response whereby the cell or organism is resistant to a subsequent stress that is different from the initial stress (i.e. exposure to heat stress leading to resistance to oxidant stress). The heat shock response is one of the more commonly described examples of stress adaptation and is characterized by the rapid expression of a unique group of proteins collectively known as heat shock proteins (also commonly referred to as stress proteins). The expression of heat shock proteins is well described in both whole lungs and in specific lung cells from a variety of species and in response to a variety of stressors. More importantly, in vitro data, as well as data from various animal models of acute lung injury, demonstrate that heat shock proteins, especially Hsp27, Hsp32, Hsp60, and Hsp70 have an important cytoprotective role during lung inflammation and injury.

Introduction

That which drug fails to cure, the scalpel can cure. That which the scalpel fails to cure, heat can cure. If the heat cannot cure, it must be determined to be incurable.

-Hippocrates

Ferruccio Ritossa unintentionally observed a novel hyperthermia-dependent puffing pattern in the giant chromosomes from the salivary glands of Drosophila melanogaster in 1962 (1). By chance occurrence, a colleague accidentally increased the temperature of one of the incubators in which he kept his specimens and the following morning Ritossa discovered a new puffing pattern that had not been there on the previous day. Realizing the mistake, Ritossa conducted additional, properly controlled experiments, and subsequently linked this new chromosomal puffing pattern with the expression of a specific group of proteins that he fittingly called heat shock proteins (1, 2). Notably, the editors of one of the more reputable scientific journals at the time rejected his manuscript, describing the findings as irrelevant and unimportant. Fortunately, investigations into this new area continued and since that time, there has been growing interest in what is now commonly referred to as the heat shock response.

The heat shock response is characterized by the rapid expression of a unique set of proteins collectively known as heat shock proteins (3–5). These highly conserved proteins have been identified in virtually all eukaryotic and prokaryotic species examined to date. While classically described as a response to thermal stress (hence the term heat shock response) (1, 6), heat shock proteins can be induced by a wide variety of non-thermal stressors and pharmacological agents (Table 1). For this reason, the terms stress response and stress proteins may be more appropriate, though these terms will be used interchangeably throughout the remainder of the present discussion. Whether induced by thermal or non-thermal stress, the stress response confers protection against subsequent and otherwise lethal hyperthermia, a phenomenon that is referred to as thermotolerance (6, 7). Perhaps more interesting from a clinical standpoint is the phenomenon of cross-tolerance, whereby induction of the stress response confers protection against non-thermal cytotoxic stimuli. For example, in vitro experiments have demonstrated that induction of the stress response protects endothelial cells against endotoxin-mediated apoptosis (8). Other examples include stress response-dependent protection against nitric oxide (9), peroxynitrite (10), and hydrogen peroxide (11). In vivo, induction of the stress response protects animals against endotoxemia/sepsis (12, 13), acute lung injury (14, 15), and ischemia-reperfusion (I/R) injury (15).

Table 1.

Inducers of the Stress response

TYPE OF STRESS	AGENT	COMMENTS
ENVIRONMENTAL	Temperature
	Heavy metals Ethanol Oxygen radicals	Cadmium, zinc
METABOLIC	Hyperosmolality Glucose starvation Tunicamycin Calcium ionophores Amino acid analogs
CLINICAL	Ischemia/Reperfusion Shock Anoxia Endotoxin	Reperfusion seems to be the limiting factor
PHARMACOLOGIC	Sodium arsenite	Used extensively in vitro and in vivo
	Herbimycin A	Tyrosine kinase inhibitor
	Geldanamycin	Tyrosine kinase inhibitor and HSP90 inhibitor
	Prostaglandin A1	Other prostaglandins are also active
	Dexamethasone Aspirin	Lowers temperature threshold for HSP induction
	Non-steroidal anti-inflammatory drugs	Lowers temperature threshold for HSP induction
	Pyyrolidine dithiocarbamate	Antioxidant; inhibitor of NFκB
	Diethyldithiocarbamate
	Bimoclomol	Hydroxylamine derivative, nontoxic
	Serine protease inhibitors	Concomitant inhibition of NF-_B
	Curcumin	Major constituent of tumeric; anti-inflammatory
	Glutamine	Clinically applicable amino acid
	Geranylgeranylacetone	Antiulcerative agent

The structure, mode of regulation, and function of stress proteins are highly conserved among different species, with well-described bacterial homologs of mammalian stress proteins. These proteins range in molecular weight from 7 to 110 kDa and are found in virtually every cellular compartment, including the nucleus, cytoplasm, and mitochondria (Table 2). By convention, the stress proteins are classified according to their molecular weight, e.g., Hsp25, Hsp32, Hsp47, Hsp60, Hsp70, Hsp90, and Hsp110. Although the expression of stress proteins was initially noted in cells following acute stress (e.g. Hsp72), several members of this family of proteins are constitutively expressed and play important roles in cellular homeostasis. For example, Hsp 90 is constitutively expressed at high levels and is one of the most abundant intracellular proteins. There are also an increasing number of proteins that have well defined cellular functions not directly related to cellular stress, but have been demonstrated to be expressed in response to heat shock and other cellular stresses, e.g., ubiquitin (16, 17); Hsp32, or heme oxygenase (18); inhibitor of κB, IκBα (19–23); endothelial nitric oxide synthase, eNOS (24); and mitogen-activated protein kinase (MAPK) phosphatase-1 (MKP-1) (25–27). While new functions continue to be discovered, the stress proteins are generally thought to maintain cellular homeostasis by acting as molecular chaperones, facilitating the proper folding and assembly of nascent polypeptides, as well as assisting in the refolding and stabilization of damaged peptides (16).

Table 2.

NAME	SIZE (KDA)	LOCALIZATION	BACTERIAL HOMOLOG	SOME KNOWN AND POSSIBLE FUNCTIONS
Ubiquitin	8	Cytosol/nucleus	—	Nonlysosomal degradation pathways
HSP 27	27	Cytosol/nucleus		Regulator of actin cytoskeleton; molecular chaperone; cytoprotection
Heme	32	Bound to ER,		Degradation of heme to bilirubin;
oxygenase		extends to cytoplasm	—	resistance to oxidant stress
HSP 47	47	ER	—	Collagen chaperone
HSP 60	60	Mitochondria	Gro EL	Molecular chaperone
HSP 70	72	Cytosol/nucleus	Dna K	Highly stress inducible; involved in cytoprotection against diverse agents
	73	Cytosol/nucleus	—	Constitutively expressed chaperone
HSP 90	90	Cytosol/nucleus	htpG	Regulation of steroid hormone activity
HSP 110	110	Nucleolus/cytosol	Clp family	Protects nucleoli from stress

Regulation of Stress Protein Gene Expression

Cells respond to stressful stimuli (e.g., heat shock) by increasing stress protein gene expression at a level that is proportional to the severity of the stress (28). Again, heat stress is the stimulus that has been most thoroughly studied, and it appears that a temperature threshold of 4°C to 8°C above the normal growing temperature is required for induction of cellular stress protein expression (5). However, the temperature threshold at which stress protein genes are expressed may be cell-type specific (29–31). Heat stress affects the tertiary and quarternary structure of intracellular proteins, leading to the partial denaturation and unfolding of these proteins and their subsequent precipitation. The intracellular accumulation of denatured or improperly folded proteins is believed to be the universal signal resulting in the stress-induced gene expression of stress proteins (5, 28, 32, 33). Further experimental support for this purported mechanism has been shown by a series of elegant experiments in which microinjection of denatured proteins into cells results in the up-regulation of stress protein expression (5, 28, 34–37).

A family of transcription factors, known as heat shock factors (HSFs) controls the regulation of stress protein gene expression. Since the initial discovery of the HSF gene in Saccharomyces cerevisiae (38, 39), several different HSFs have been found in both vertebrate and nonvertebrate organisms (40, 41). The human HSFs that have been identified to date include HSF1, −2, and −4, all of which exhibit a similar structure with a highly conserved amino-terminal helix-turn-helix DNA binding domain and a carboxy-terminal transactivation domain (HSF-3 appears to be avian-specific and is not found in humans) (42–46). HSF1 appears to be the most important stress-inducible HSF, and biochemical and genetic studies have clearly demonstrated the critical role of HSF1 in stress-inducible stress protein expression, resistance to stress-induced apoptosis, and extra-embryonic development (41, 47–51). HSF is present under normal, resting conditions as a latent monomer that lacks both DNA binding and transcriptional activity. Following heat shock (or exposure to any of the myriad of stimuli that have been shown to induce the stress response), HSF1 undergoes homotrimerization and rapidly translocates into the nucleus (52, 53).

HSF1 also appears to possess intrinsic stress-sensing activity (51, 54–57). For example, homotrimerization of HSF1 appears to be directly activated in vitro and in vivo in a redox-dependent, reversible manner, via a process requiring two cysteine residues within the DNA-binding domain (51). These two cysteine residues appear to be crucial for nuclear translocation, DNA binding, and activation of stress protein target genes as well, perhaps through the formation of disulfide bonds which maintain HSF1 in a homotrimeric conformation (51, 58). While the exact mechanisms remain to be elucidated, a number of studies support a clear association between the cellular redox state and activation of the stress response (51, 57, 59, 60). For example, HSF1 activation is inhibited under conditions of hypoxia or in the presence of reducing agents such as dithiotreitol (DTT) (61–63).

Stress proteins appear to act as repressor proteins that maintain HSF1 in its non-active, monomeric conformation. Studies suggest that HSF1 monomers form a complex with multiple chaperones and co-chaperones, including Hsp90, Hsp50, and Hsp70 (64). Once a cell is exposed to stress, these chaperones and co-chaperones bind to denatured and damaged proteins, thereby “releasing” the HSF1 monomers to subsequently undergo homotrimerization (28, 40, 64–68). While homotrimerization is sufficient for DNA binding and nuclear translocation, additional steps are required for the complete activation of HSF1. The magnitude and duration of transcriptional activity appears to be regulated by several processes, including inducible phosphorylation of specific serine residues by an as yet undefined pathway (28, 38, 40, 64, 69–72). A number of candidate protein kinases have been shown to phosphorylate HSF1 in vitro, including Erk1/2, glycogen synthase kinase (GSK), JNK, and protein kinase C, among others (64). Once inside the nucleus, HSF1 binds to a heat shock element (HSE), which is defined by a tandem repeat of the pentamer nGAAn (“n” denoting a less conserved sequence) arranged in an alternating orientation either “head to head” (e.g., 5’-nGAAnnTTCn-3’) or “tail to tail” (e.g., 5’-nTTCnnGAAn-3’) (73), resulting in the upregulation of stress protein gene expression.

Hsp27

Hsp27 is an important member of what are commonly referred to as the small stress proteins (small heat shock proteins, sHsps). Hsp27, like many of the other stress proteins discussed below, is highly inducible in the lung following exposure to a variety of stressors, thereby enhancing cellular resistance to heat shock, oxidative stress, and inflammatory mediators such as tumor necrosis factor (TNF)- α in vitro and in vivo (74–79). Hsp27 has been shown in severl in vitro models to inhibit apoptosis, especially following oxidative stress (79). This anti-apoptotic effect appears to occur through mechanisms involving protection of anti-oxidant enzymes (e.g. glucose-6-phosphate dehydrogenase, glutathione reductase, and glutathione transferase – all of which critically affect the intracellular level of reduced glutathione, an important redox modulator in the cell), inhibition of lipid peroxidation, and stabilization of the cytoskeleton. In addition, Hsp27 may chaperone oxidized proteins through the ubiquitin-independent 20S proteasome degradation pathway (79).

Hsp27 is highly induced in peripheral blood monocytes from patients with systemic inflammatory response syndrome (SIRS) (80), and at least one study suggested that increased Hsp27 expression improved mortality in a rat model of endotoxic shock (81). Hsp27 is a substrate for phosphorylation by mitogen-activated protein kinase (MAPK)-activated protein kinase-2 (MAPKAPK-2) (82) and is believed to modulate the cytoskeletal arrangement of actin filaments in pulmonary endothelial cells, in a manner which is at least partially dependent upon its phosphorylation state (83, 84). Importantly, in contrast to the aforementioned studies, Hsp27 phosphorylation temporally correlates with LPS-induced lung injury and increased capillary permeability pulmonary edema in vivo (84).

A growing body of evidence (see below) suggests that extracellular stress proteins, including Hsp27, may be a “danger signal” for the innate immune system. In most cases, extracellular stress proteins are pro-inflammatory in nature. Of interest then is a recent study that showed that extracellular Hsp27 potently induced the anti-inflammatory cytokine, interleukin (IL)-10 in human monocytes in a dose- and time-dependent manner. This induction was independent of any autocrine effects of Hsp27-induced TNF-α (85).

Hsp 32 (Heme oxygenase)

Heme oxygenase (HO) is responsible for catalyzing what is one of the most well-known and common colorimetric reactions in humans - when a common bruise transforms through the spectrum of hues ranging from purple to green to yellow (86, 87). HO is the first and rate limiting step in the degradation of heme (purple hue) to biliverdin (green hue), and finally to bilirubin (yellow hue). Three known isoforms of HO exist: HO-1, -2, and -3. In the context of cytoprotection, HO-1 appears to be the most relevant isoform. HO-1 is identical to Hsp32 and is highly inducible by a variety of cellular stressors and stimuli, including heme, nitric oxide, cytokines, heavy metals, hyperoxia, hypoxia, endotoxin, heavy metals, and heat shock (18, 88–90). HO-1 activity is present in virtually all organs and is thought to primarily account for the cytoprotective properties of HO.

Since the discovery of HO in 1968, several observations have provided notable indications that HO-1 may serve an important cytoprotective role. These include the ability of HO-1 to be highly induced in response to potentially cytotoxic stimuli, its relative high level of conservation throughout evolution, and its wide tissue distribution. The prediction that HO-1 would confer broad cytoprotection has been confirmed by a variety of in vitro and in vivo studies (88, 91). For example, in vitro overexpression of HO-1 conferred protection against oxygen toxicity (hyperoxia) in hamster fibroblasts (92), rat fetal lung cells (93), and human respiratory epithelial cells (94). In a similar manner, overexpression of HO-1 in coronary endothelial cells conferred protection against heme and hemoglobin toxicity (95). In cultured murine fibroblasts, regulated overexpression of HO-1 conferred protection against tumor necrosis factor (TNF)-α -mediated apoptosis (96). Finally, in a human respiratory epithelial cell line representative of lung epithelial cells found in patients with cystic fibrosis, overexpression of HO-1 conferred protection against Pseudomonas-mediated cellular injury and apoptosis (97).

Experiments in animal models, involving either pharmacologic induction of HO-1 or genetic overexpression of HO-1, have confirmed that these in vitro observations apply to the in vivo context. For example, induction of HO-1 by intravenous hemoglobin protected rats against the lethal effects of endotoxemia. Protection in this model correlated with attenuation of endotoxin-mediated hypotension, renal dysfunction, hepatic dysfunction, and inflammation (98, 99). The direct role of HO-1 in conferring protection was further confirmed by co-administration of a competitive inhibitor of HO, tin protoporphyrin, which led to a substantial reduction of the protective effects induced by intravenous hemoglobin administration. Lung epithelial overexpression of HO-1, via an adenovirus vector, conferred protection in rats exposed to hyperoxia (100).

Gene knockout studies have provided further evidence regarding the cytoprotective properties of HO-1 and have been instrumental in further elucidating the biological roles of HO-1. HO-1 null mutant mice generally do not survive to term, and animals that do survive die during the first year of age (101). These animals display severe growth retardation, anemia, iron deposition in the kidneys and liver, and evidence of chronic inflammation in a variety of organs. In other investigations involving HO-1 null mutant mice the animals have been demonstrated to be more susceptible to: 1) renovascular-related hypertension, renal failure, and cardiac hypertrophy (102), 2) endotoxin-mediated lethality (103), and 3) right ventricular dilation and infarction secondary to chronic hypoxia (104). Collectively, the in vitro, in vivo, and gene deletion studies outlined above provide compelling evidence regarding the broad cytoprotective role of HO-1 in clinically relevant forms of cellular and tissue injury. What remains relatively elusive, however, is the mechanism(s) by which HO-1 confers this broad level of protection.

The byproducts of HO enzymatic activity include carbon monoxide (CO), bilirubin, and ferritin, and each of these byproducts have been postulated to play a role in cytoprotection (18, 88, 89). For example, ferritin is known to protect against oxidant stress and bilirubin can function as a potent antioxidant. While it us likely that the three byproducts synergize in some way to confer cytoprotection, the most recent work in the field implicates CO-related cell signaling as the key component of HO-1-mediated cytoprotection (18, 105, 106).

CO shares a variety of properties with another gaseous molecule having ubiquitous biological effects, nitric oxide. These include, neurotransmission, regulation of vascular tone, and activation of soluble guanylate cyclase (106). Other important connections between the CO pathway and the nitric oxide pathway include co-induction of inducible nitric oxide synthase and HO-1 by common stimuli (e.g. reactive oxygen species and cytokines), nitric oxide-dependent induction of HO-1 expression, and CO-dependent modulation of NO production (106–109). The reported biological effects of CO include potent anti-inflammatory effects (via the MAP kinase pathway), anti-apoptotic effects, and antioxidant effects (110–117). In the context of these biological effects, the cytoprotective properties of CO have been demonstrated in a variety of experiments involving direct administration of CO. For example, exogenous administration of CO protected cultured fibroblasts from tumor necrosis factor-α-mediated apoptosis (96). In vivo administration of low concentrations of inhaled CO protected rats from hyperoxia-mediated acute lung injury (114) and administration of exogenous CO to cardiac tissue protected the tissue from ischemia-reperfusion injury following transplantation (118). These in vivo studies are particularly intriguing because the amount of CO administered is within the range administered to humans undergoing lung diffusion scans (105). The rapidly evolving data strongly suggest that HO-1-derived CO is the key mechanism by which HO-1 confers cytoprotection, and further work in this area holds tremendous potential for therapeutic strategies involving HO-1 and/or CO. Interestingly, Kim and colleagues (119) recently demonstrated that Hsp70 can mediate the cytoprotective effects of CO both in vitro and in vivo.

Hsp60

The Hsp60 family of stress proteins are expressed both constitutively and under conditions of stress and can be classified into two groups – mitochondrial Hsp60 (mHsp60) and cytosolic Hsp60 (cHsp60) (120, 121). The mHsp60 forms heptamers and tetradecamers in association with Hsp10 and ATP and is an important part of the mitochondrial chaperone system, while the cHsp60 forms heteroligomeric ring structures to stabilize important cytoskeletal proteins such as actin and tubulin (122, 123). Like many of the other stress proteins, Hsp60 gene expression is highly inducible in response to cell stress. While less is known regarding the specific role of Hsp60 in acute lung injury, Hsp60 has been detected in the bronchial epithelium of patients with asthma (124), nonhuman primates following ozone exposure (77), and human pulmonary microvascular endothelial cells following both heat shock and heavy metal exposure (125). Furthermore, in vivo studies suggest that Hsp60 expression has been shown to ameliorate the progression to pulmonary fibrosis induced by paraquat intoxication (126). Collectively, these studies therefore suggest that Hsp60 confers stress tolerance in the lung. In keeping with the paradigm of extracellular stress proteins as “danger signals”, a recent clinical study suggested that serum levels of Hsp60 within 30 minutes of traumatic injury predicted the development of acute lung injury in adults (127).

Hsp70

Hsp70 expression is well described in both whole lungs and in specific lung cells from a variety of species and in response to a variety of stressors (128). More importantly, in vitro data, as well as data from various animal models of acute lung injury, demonstrate that stress proteins have an important cytoprotective role during lung inflammation and injury (128). For example, induction of the stress response by either thermal stress or sodium arsenite attenuates LPS-mediated apoptosis in cultured sheep pulmonary artery endothelial cells (8), perhaps via an anti-oxidant mechanism, as induction of the stress response was temporally associated with a reduction in LPS-mediated superoxide anion production. Hsp70 appears to be directly involved in this cytoprotective response, as overexpression of Hsp70 via stable transfection inhibited LPS-mediated apoptosis compared to wild-type cells and cells transfected with a control plasmid (8). Similarly, induction of Hsp70 by thermal stress protected cultured murine lung epithelial cells and bovine pulmonary artery endothelial cells from oxidant-mediated injury (11, 129). Overexpression of Hsp70 via stable transfection also protects lung epithelial cells from the deleterious effects of hyperoxia (130) and peroxynitrite (9, 131). Perhaps more compelling, direct delivery of the mature Hsp70 protein using a Tat-Hsp70 protein protects HSF1 null fibroblasts, which lack the intrinsic ability to mount a stress response and are therefore susceptible to hyperoxia-mediated cell death (132), against the deleterious effects of hyperoxia (133). Collectively, these data demonstrate that induction of the stress response protects lung cells against agents that are commonly implicated in the pathophysiology of acute lung injury.

Villar and colleagues were the first to demonstrate that induction of stress response using whole body hyperthermia (41 to 42°C 18 h prior to intratracheal administration of phospholipase A₂) protected rats against acute lung injury (14). In addition, whole body hyperthermia increased both Hsp70 mRNA and protein as measured by Northern and Western blot analysis, respectively. These findings were further corroborated by these same investigators in a rat model of acute lung injury induced by cecal ligation and puncture (CLP) (134). Again, induction of Hsp70 mRNA and protein expression in the lung was temporally associated with reduced lung injury and improved survival. Finally, induction of lung Hsp70 expression via administration of sodium arsenite, in the absence of an increase in body temperature, protected rats against CLP-induced lung injury and mortality (135). Ribeiro and colleagues later demonstrated that prior induction of the stress response via whole body hyperthermia decreased lung inflammation and improved lung function in an ex vivo model of ventilator-induced lung injury (VILI) (136). Weiss and colleagues (137) demonstrated that CLP-induced polymicrobial sepsis impairs lung Hsp70 expression in rats and surmised that this intrinsic failure of the normal stress response contributes to acute lung injury in sepsis. Importantly, Hsp70 induction is decreased in the peripheral blood monocytes of patients with acute respiratory distress syndrome (ARDS), further suggesting that failure of the normal stress response contributes to sepsis-induced lung injury (138). Restoration of Hsp70 expression, at least in the pulmonary epithelium, via adenoviral delivery of mature Hsp70 ameliorates lung injury following CLP in rats (139).

Stress proteins and Cytoprotection

The mechanisms by which the stress response confers such broad cytoprotection are not fully understood, but Hsp70 certainly plays a central role. Hsp70 is the most highly induced stress protein in cells and tissues undergoing the stress response (3), and is known to be induced in patients with a variety of critical illnesses or injuries (140–143). Microinjection of anti-Hsp70 antibody into cells impairs their ability to achieve thermotolerance (144), and increased expression of Hsp70 by gene transfer/transfection has been demonstrated to confer protection against in vitro toxicity secondary to lethal hyperthermia (145), endotoxin (8), nitric oxide (9), hyperoxia (130), and in vivo ischemia-reperfusion injury (146–148). In addition, in vitro delivery of mature Hsp70 into the intracellular compartment has been shown to protect fibroblasts against lethal thermal injury and hyperoxia (133), and neuronal cells against nitrosative stress and excitotoxicity (149). Mice deficient in HSF1 have a drastically reduced ability to express Hsp70 (47, 48). Cell lines derived from these animals can not achieve thermotolerance and are highly susceptible to oxidant stress compared to cells from wild-type mice (47, 48, 132). When challenged with systemic endotoxin, HSF1-deficient mice have increased mortality compared to wild-type mice (48). Collectively, these data demonstrate that Hsp70 is central to the cytoprotective properties of the stress response. The mechanisms by which Hsp70 and other stress proteins confer protection are not fully understood, but most likely relate to the ability of stress proteins to serve as molecular chaperones by binding, re-folding, transporting, and stabilizing damaged intracellular proteins.

Another potential mechanism by which the stress response may confer cytoprotection is by modulating inflammatory responses. The stress response has been demonstrated to inhibit the expression of a number of genes related to inflammation, including tumor necrosis factor (TNF)-α, interleukin (IL)-1β, inducible nitric oxide synthase, IL-8, RANTES, C3, macrophage chemotactic protein-1, and intracellular adhesion molecule-1 (22, 129, 131, 150–157). In addition, it has been postulated that the inhibitory effects of the stress response are relatively selective for inflammation-associated genes (21). The mechanisms by which the stress response inhibits proinflammatory gene expression involve inhibition of NF-κB. Several in vitro and in vivo studies have demonstrated that induction of the stress response inhibits activation of NF-κB, a pluripotent transcription factor that regulates the expression of many genes associated with inflammation (21, 158, 159). The latest work in the area has identified IκB kinase (IKK) as the most upstream target through which the stress response modulates NF-κB activity. IKK is the rate limiting step in the activation of NF-κB in that it phosphorylates the endogenous NF-κB inhibitor, IκBα. Phosphorylation of IκBα leads to its rapid degradation by a proteasome/ubiquitin-dependent mechanism, thus releasing NF-κB to enter the nucleus. Induction of the stress response inhibits activation of IKK, in part by an intracellular phosphatase-dependent mechanism (159–161). Inhibition of IKK subsequently inhibits phosphorylation and degradation of IκBα (161), thus keeping NF-κB in an inactive state.

Recent work suggests that the modulating effects of the stress response on inflammation-related signal transduction also involves the de novo expression of proteins not traditionally considered to be classified as stress proteins. For example, it is now well established that the endogenous NF-κB inhibitory protein, IκBα, is expressed in response to heat shock both in vitro and in vivo (19, 23, 157). More recent work has demonstrated that the dual specific phosphatase, MKP-1, is also expressed in response to heat shock (26, 27). MKP-1 is a potent counter regulator of many proinflammatory signal transduction pathways. Thus, stress response-associated induction of IκBα and MKP-1 gene expression serves as another mechanism by which the stress response can inhibit NF-κB activity and inflammation-associated signal transduction.

Recent studies suggest that HSF1 acts directly as a transciptional repressor of several proinflammatory proteins, including IL-1β, c-fos, urokinase, and TNF-α (162, 163). For example, HSF1 binds the 5’-untranslated region of the murine TNF-α gene to repress transcription (162). In a similar manner, HSF1 represses transcription of the gene for IL-1β through physical interaction with the CCAAT/enhancer binding protein (163).

Stress proteins and Innate Immunity

Until relatively recently, stress proteins were considered to be exclusively intracellular proteins, but as mentioned briefly above, a growing body of literature suggests that stress proteins may also exist and function outside of the cell. For example, several studies suggest that cellular stress results in the increased surface expression and release of stress proteins (164–177). The release of stress proteins into the extracellular environment was first reported in the late 1980s, when Hsp70 release was demonstrated in cultured rat embryo cells following exposure to increased temperature (178). Importantly, Hsp70 release did not appear to be mediated via classic secretory pathways, as it was not inhibited by either colchicine or monensin. Cell lysis was not involved either, as Hsp70 release was not detected following exposure to non-ionic detergents. Hsp70 synthesized in the presence of a lysine amino acid analogue (aminoethyl cysteine) was not released from these cells, suggesting that the altered protein structure prevented interaction with the specific release mechanism (178).

While some studies suggest that Hsp70 is released in a non-specific manner from dying, necrotic cells (167, 179), we and others (169, 171, 178, 180–183) have shown that viable cells release Hsp70 in a specific and inhibitable manner. Monensin and brefeldin A are inhibitors of the classic endoplasmic reticulum (ER)/Golgi protein transport and secretory pathways. We have shown that Hsp70 release from THP-1 cells is not inhibited by either monensin or brefeldin A (Wheeler and Wong, unpublished data). Others have shown that Hsp70 release from peripheral blood monocytes (PBMC) is inhibited by brefeldin A, but not monensin (182, 183). In this study, Hsp70 release was also inhibited by methylamine and methyl-β-cyclodextrin, both of which inhibit protein secretion via lysosomal pathways (183). Soluble proteins are usually secreted via a classic, ER/Golgi dependent secretory pathway. The Hsp70 gene does not encode for the classic N-terminal peptide leader sequence targeting it for secretion via this pathway (184), which is consistent with the data presented here. Recent studies suggest that Hsp70 is actively released via an exosome-dependent, non-classical protein secretory pathway (177, 185).

Important work in this area suggests that the release of stress proteins from stressed cells may serve to signal an impending danger signal to neighboring cells, the so-called danger hypothesis (176, 186). For example, extracellular Hsp70 elicits NF-κB-dependent, pro-inflammatory gene expression via Toll-like receptor (TLR)4 and TLR2 (187, 188). Therefore, release of Hsp70, whether from dying, necrotic cells or viable, but damaged cells, can activate the host innate immune response.

Several properties would suggest Hsp70 to be a biologically plausible and likely candidate to serve as a host danger signal. Collectively stress proteins are the most abundant intracellular proteins, representing up to 10% of the total protein content in the cell (189). Hsp70 in particular is markedly induced in response to a diverse range of cellular insults, including increased temperature, oxidative stress, glucose deprivation, chemical exposure, I/R injury, ultraviolet radiation, and infectious agents such as LPS. Therefore, Hsp70 or additional inducible stress proteins, by virtue of their relative abundance during times of stress, are reliable markers of cell stress (i.e., danger).

As previously noted, stress proteins are ancient, highly conserved molecules that have been identified in virtually every organism examined to date. In comparison, LPS, an important exogenous danger signal, appeared relatively late on the evolutionary time-scale and is much less ubiquitous, being unique to only gram-negative bacteria. It is perhaps no mere coincidence that Hsp70-mediated and LPS-mediated activation of the host innate immune response occurs via similar mechanisms (i.e. TLR-4-mediated signal transduction), and it is plausible that the programmed response to the exogenous danger signal, LPS, is modeled on the more primitive programmed response to the endogenous danger signal (189).

Stress proteins, particularly Hsp70, are also highly immunogenic and have the capacity to mediate the induction of peptide-specific immunity. For example, as molecular chaperones, stress proteins bind to many peptides derived from the cells from which they are isolated. Stress protein-peptide complexes elicit potent T cell responses against the chaperoned peptide as well as the cell type from which the chaperoned peptide is derived, including tumors and viruses, and vaccination with stress protein-tumor peptide complexes as an immunotherapy for cancer is an active area of investigation (189–194). Similarly, stress protein-pathogen-derived peptide complexes have the capacity to elicit a pathogen-specific immune response (193). Finally, stress proteins themselves, especially members of the Hsp60 and Hsp70 families, have the capacity to activate the host innate immune response, resulting in dendritic cell activation and maturation, activation of complement, and release of proinflammatory cytokines (179, 187, 188, 195–203). Collectively, these observations would suggest the stress proteins to be likely candidates for endogenous danger signals - they are abundant, inducible, highly conserved, and appear to activate both the innate and adaptive immune response.

Extracellular Hsp70 (increased serum levels) is evident in a variety of clinical scenarios associated with physiologic stress. For example, increased extracellular Hsp70 levels are noted following strenuous exercise (180, 181, 204–208), following cardio-pulmonary bypass in both adults (209, 210) and children (Wheeler, unpublished data), in elderly adults (211, 212), and in children with acute lung injury (Wheeler, unpublished data). In addition, increased extracellular Hsp70 concentrations correlate with worse outcome in a variety of inflammatory disease processes, including liver disease (213), coronary artery disease (214–217), traumatic brain injury secondary to child abuse (218), pre-eclampsia (219), sickle cell disease vaso-occlusive crisis (220), septic shock (221), and traumatic brain injury (222). Conversely, increased extracellular Hsp70 levels correlated with improved outcome following multiple trauma in adults (141).

Given the signaling properties recently ascribed to Hsp70, these data generate a number of functional questions. First and foremost, is extracellular Hsp70 merely a marker of cellular stress, or could the release of Hsp70 potentiate an already active host immune response, thereby leading to poor outcome? Second, could it be possible that extracellular Hsp70 serves an as yet undefined cytoprotective function at lower levels as a normal response to infection or stress, and once a certain critical threshold is attained potentiate the dysregulated inflammatory response that subsequently results in significant auto-injury to the host? The results correlating increased Hsp70 levels to improved outcome in adults with trauma support this latter concept (141). Future investigation in this area will likely address these questions.

Conclusion

In this review, we have attempted to describe the stress response, one of the most powerful and ubiquitous endogenous mechanisms that exists to counter-act the multitude of cytotoxic stimuli that can adversely affect the human host. While stress protein expression is not unique to the lung compartment, the well described biochemical properties and regulation of stress proteins suggest that they can play an important role in the context of acute lung injury. Much of the lung-specific literature indicates that this role is primarily cytoprotective. However, the emerging data involving extracellular Hsp72 indirectly suggest that within a certain milieu, extracellular Hsp72 could contribute to the pathophysiology of acute lung injury. Ongoing work will further elucidate these issues and it remains to be determined if stress protein expression can be manipulated as a therapeutic strategy. In this regard, Weischmeyer and colleagues have generated interesting data indicating that the amino acid, glutamine, may be a feasible approach to inducing Hsp72 in the clinical setting as a strategy to protect against acute lung injury (223–226).

Figure 1. Potential roles of heat shock proteins (HSP) in acute lung injury.

The expression of heat shock proteins in both alveolar macrophages and airway and alveolar epithelial cells is upregulated in response to a myriad of cell stressors, including LPS, free radicals, thermal stress, and hypoxia. Increased expression of heat shock proteins results in a cytoprotective response in these cell types. Release of heat shock proteins (by an as yet unidentified mechanism) may serve as a “danger signal” to activate the inflammatory response in surrounding cells.