Chaperonin 60 unfolds its secrets of cellular communication

Abstract

The cell biology of the chaperonins (Cpns) has been intensively studied over the past 25 years. These ubiquitous and essential molecules assist proteins to fold into their native state and function to protect proteins from denaturation after stress. The structure of the most widely studied Cpn60, Escherichia coli GroEL, has been solved and its mechanism of protein folding action largely established. But in the last decade, evidence has accumulated to suggest that the Cpn60s have functions in addition to intracellular protein folding, particularly the ability to act as intercellular signals with a wide variety of biological effects. Cpn60 has the ability to stimulate cells to produce proinflammatory cytokines and other proteins involved in immunity and inflammation and may, therefore, provide a link between innate and adaptive immunity. Cpn60s are also thought to be pathogenic factors in a wide range of diseases and have recently been reported to be present in the circulation of normal subjects and those with heart disease. An interesting facet of these proteins is the finding that in spite of significant sequence conservation, individual Cpn60 proteins can express very different biological activities. This review discusses the work to date, which has revealed the cell-cell signaling actions of Cpn60 proteins.

INTRODUCTION

The chaperonins (Cpns) are a ubiquitous family of sequence-related molecular chaperones, comprising oligomeric proteins of approximately 60 kDa subunit mass that are essential for protein folding under both normal and stressful conditions (reviewed by Bukau and Horwich 1998; Ranson et al 1998; Saibil 2000a; Grallert and Buchner 2001; Grantcharova et al 2001; Thirumalai and Lorimer 2001). The Cpns are divided into 2 subgroups—group I and group II. Members of the GroEL family (or group I Cpns) are found in eubacteria, mitochondria, and chloroplasts. The group II or TriC (TCP-1 ring complex) protein family is found in the eukaryotic cytosol and in archaebacteria (reviewed by Gutsche et al 1999). Of the group II members, only the archaebacterials Cpns are heat shock inducible.

The Escherichia coli Cpn60, GroEL, is the best characterized of the Cpns. Its toroidal 7-membered ring structure has been solved (Braig et al 1994; Chen et al 1994; reviewed by Saibil 2000b), and its mechanisms of action, in concert with the cochaperone, Cpn10, are still the subject of investigation (reviewed by Ellis and Hartl 1999; Grantcharova et al 2001). The proportion of the cellular proteins that are substrates for GroEL in vivo was addressed by Lorimer (1996) and Ellis and Hartl (1996), who estimated that GroEL was responsible for the folding of between 2% and 7% of the proteins in E coli. Later work by McLennan and Masters (1998) and by Houry et al (1999) has identified many of the in vivo substrates of GroEL.

Studies of the GroEL-GroES system have established the extremely important roles of the Cpns in protein folding and in the cell stress response (Georgopoulos and Hohn 1978; Tilly et al 1981; reviewed by Bukau and Horwich 1998; Ellis 2000, 2001). But just as the molecular details of the intracellular folding actions of GroEL were being established, it became clear that Cpn60, and molecular chaperones generally, have an additional, and possibly equally important, role as intercellular signaling molecules.

THE EVIDENCE THAT CPN60 PROTEINS ARE SIGNALING MOLECULES

The clue that Cpns may be more than protein-folding molecules was the report by Friedland et al (1993) that the Cpn60.2 of Mycobacterium tuberculosis (mtCpn60.2—better known as Hsp65) a protein best known as a major immunogen in patients with tuberculosis, had the ability to stimulate human monocytes to secrete the key, early-response, proinflammatory cytokines, interleukin (IL)-1β, and tumor necrosis factor (TNF) α. The studies by Peetermans et al (1994) confirmed the role of mtCpn60.2 as a cell-signaling molecule when added to human monocytes. Cpn60 homologues from a range of bacteria: Legionella pneumophila, E coli, M leprae, and M bovis increased the steady-state levels of cytokine messenger ribonucleic acid for IL-1α, IL-1β, IL-6, TNFα, and granulocyte-monocyte stimulating factor in macrophage cultures (Retzlaff et al 1994). These studies raised the possibility that Cpn60 molecules also had a role as intercellular signaling proteins and that they could modulate the immune response by stimulating myeloid cell cytokine synthesis. But the problem is that unless Cpn60 proteins can be shown to be released by cells, these findings are interesting but of little biological relevance.

ARE CPN60 PROTEINS RELEASED FROM CELLS?

The accepted dogma is that the Cpn60 proteins of both prokaryotes and eukaryotes are intracellular proteins. The obvious question raised by the findings that Cpn60 can stimulate cells is—how do these intracellular molecules exit from cells? Cpn60 proteins must be secreted from cells to act at the surface of the target cell. This target cell may be the cell secreting the Cpn. But the Cpn60s do not contain a signal sequence and, in consequence (at least according to dogma), cannot be secreted. This lack of a signal sequence has been used to criticize the hypothesis that Cpns have intercellular signaling activity. But this criticism is based on the assumption that, for example in eukaryotic cells, the endoplasmic reticulum-to-Golgi pathway is the only pathway of protein secretion. In prokaryotes, many novel pathways of protein secretion have been discovered in the past decade (Wilson et al 2002). It must also be remembered that a number of very important cell-cell signaling proteins such as IL-1 (Rubartelli et al 1990), and fibroblast growth factor (Mignatti et al 1992) lack signal peptides and the term “leaderless” secretory proteins has been coined to describe such molecules (Andrei et al 1999). Moreover, in recent years, a number of additional secretory pathways have been discovered in eukaryotes (Muesch et al 1990; Rubartelli et al 1990; Cleves 1997; Rammes et al 1997; Andrei et al 1999). It must also be realized that a growing number of proteins with multiple functions are now known and have been termed “moonlighting proteins” (reviewed by Jeffery 1999). A good example of such a moonlighting protein is the glycolytic enzyme, phosphohexose isomerase. This is a cytosolic protein present in all cells as a member of the glycolytic pathway. This protein is now known to be secreted by cells and has been shown to be identical to 3 distinct and independently discovered cytokines. These are: (1) autocrine motility factor, which stimulates cell migration; (2) neuroleukin, which is both a nerve growth factor and B cell maturation factor, and (3) differentiation and maturation mediator, a protein that induces human myeloid cell differentiation (Haga et al 2000). Other examples of intracellular proteins that have been shown to have actions outside the cell are thymidine phosphorylase, glyceraldehyde 3-phosphate dehydrogenase, and thioredoxin (Jeffery 1999). Thus, we propose that Cpn60 is another example of a moonlighting protein that is secreted by an as yet undiscovered pathway.

Early evidence for the release of Cpn60 from bacteria was the report by one of the authors (B.H.) that the Cpn60 homologue of the oral bacterium, Actinobacillus actinomycetemcomitans, is present on the surface of this bacterium and is a potent inducer of bone destruction (Kirby et al 1995). The extracellular location of this protein has been confirmed by other workers (eg, Hara et al 2000), and it has been reported that the A actinomycetemcomitans Cpn60 has effects on the growth of various cell populations (Goulhen et al 1998; Paju et al 2000). It has also been reported that Borrelia burgdorferi, Helicobacter pylori, Haemophilus ducreyi, and L pneumophila all express Cpn60 on their outer surfaces (Scopio et al 1993; Cao et al 1998; Frisk et al 1998; Garduno et al 1998a, 1998b). In this surface location, it is proposed that Cpn60 acts as an adhesin, enabling bacteria to bind to host cells (Frisk et al 1998) or mediating host cell invasion (Garduno et al 1998a). The possibility exists that the surface location of a protein is not because of a process of active secretion but because of cell lysis. That the release of H pylori Cpn60 was due to cell lysis has been examined in detail, and it was concluded that a lytic process was not contributing significantly to the secretion of this protein (Cao et al 1998). On the basis of this evidence, it has been postulated that the Cpn60s of a number of bacteria may act as direct cell-to-cell virulence factors for host cells (Lewthwaite et al 1998). Cpn60 has also been reported in extramitochondrial sites in the ciliated protozoan, Tetrahymena thermophila, and it has been postulated that this molecular chaperone may play a role in the development of the oral apparatus in these single-celled eukaryotes (Takeda et al 2001).

Various studies have also detected Hsp60 (the human Cpn60) in different eukaryotic cellular compartments, besides it usual residency in the mitochondrion. Soltys and Gupta (1996, 1997) and Cechetto et al (2000) investigated the cellular localization of Hsp60 in a range of mammalian cells and tissues and showed that it was present in cellular compartments other than the mitochondrion. Indeed, up to 10% of the total cellular content of Hsp60 was reported by these workers to be on the cell surface. This cell surface location has been confirmed for a variety of cell populations, including leukocytes and vascular endothelial cells (Schett et al 1995; Frostegard et al 1996; Woodlock et al 1997). There are also a growing number of reports that Cpn60 expressed on eukaryotic cells can act as a receptor for various ligands, including p21^ras (Ikawa and Weinberg 1992), gp41 transmembrane protein of human immunodeficiency virus (Speth et al 1999), and high-density lipoprotein (Bocharov et al 2000). Indeed, it has been reported that Cpn60 has a high affinity for lipid monolayers and bilayers and can insert into such structures through a hydrophobic C terminus (Torok et al 1997).

It has been proposed that Hsp60 may play a role in the initiation of the inflammatory process (Peetermans et al 1995). After an infection with Listeria monocytogenes, murine Cpn60 (Hsp60) was found to be localized on the surface of murine spleen cells, antigen presenting cells (APCs) such as B cells and macrophages, and also liver cells. Uninfected mice also expressed Hsp60 on their liver cell plasma membranes, although levels of Hsp60 increased after infection (Belles et al 1999). It may be that Hsp60 expression at the cell surface acts as a danger signal after infection. This idea is explored later in the review. Further evidence for the belief that Cpn60 is a secreted protein is limited, but Hsp60 has been detected within the secretory insulin granula of beta cells (Brudzynski 1993), and there are a growing number of reports that Hsp60 is found in the human circulation both in individuals with various disease states and in apparently normal people (Pockley et al 1999). This finding will be discussed in a later section.

The Cpn60s are not, however, the only molecular chaperones found on the surface of cells. Hsp70 also has been found on the surface of a number of transformed mammalian cell lines (Ferrarini et al 1992; Multhoff et al 1995; Botzler et al 1998; Hirai et al 1998; Sapozhnikov et al 1999). The bacteria Chlamydia trachomatis, and Coxiella burnetii and the yeast, Candida albicans, all express Hsp70 on their cell surfaces (Raulston et al 1993; Macellaro et al 1998; Lopez-Ribot et al 1999). Hsp90 is expressed on the cell membranes of the mononuclear cells of systemic lupus erythematosus patients (Erkeller-Yuksel 1992) and is found on the cell surface of tumor cells (Ullrich et al 1986; Vendetti et al 2000) and on the cell surface of the yeast, Candida albicans (Chaffin et al 1998). Interestingly, a recent report by Triantafilou et al (2001a) identified both Hsp70 and Hsp90 as proteins localized on the plasma membranes of murine mononuclear cells and endothelial cell lines. These proteins are claimed to be receptors for the Gram-negative bacterial cell wall component lipopolysaccharide (LPS). This implies that these molecular chaperones are recognition and signaling elements in the inflammatory process (Triantafilou et al 2001a, 2001b).

CELL STIMULATION BY CPN60

Cytokine induction

Having established that Cpn60 is released by both prokaryotic and eukaryotic cells, even though the mechanism(s) of release is not defined, the discussion will return to the intercellular actions of secreted Cpn60 proteins.

Since the initial report by Friedland et al (1993), a growing number of publications have confirmed and expanded these findings that Cpn60 proteins activate cells to synthesize and secrete cytokines. An obvious criticism of this work is that the cytokine induction may be due not to the Cpn but to contaminants in the Cpn60 preparations. The most obvious of the contaminants are: (1) LPS from the E coli in which the cpn60 genes are normally expressed or (2) the large number of proteins that are known to contaminate purified GroEL (Price et al 1991; Houry et al 1999) and by inference all other Cpn60 proteins (or both). LPS activates cells by binding to a coreceptor, CD14, and this complex interacts, in some as yet unexplained manner, with one of the Toll-like receptors (TLRs)—either TLR4 (reviewed by Kimbrell and Beutler 2001) or, for some bacterial species, TLR2 (Muta and Takeshige 2001). Most workers have included the LPS-binding and inactivating antibiotic, polymyxin B, as a control for LPS contamination. Others have shown that heating Cpn60 blocks the cytokine-inducing activity (Chen et al 1999; Kol et al 1999). In contrast to most proteins, LPS can survive both autoclaving and proteases without loss of bioactivity. But there appears to be some variation in the response of Cpn60 proteins to heat. For example, the authors' group has shown that the Cpn60 proteins of M tuberculosis survive boiling without loss of biological activity. These proteins have to be autoclaved, or exposed to proteinase K, to block activity (Lewthwaite et al 2001). In addition, GroEL can be completely trypsinized without losing the ability to activate monocytes (Tabona et al 1998). Finally, the removal of the proteins that contaminate GroEL, by use of a Reactive Red column, failed to inhibit the activity of this Cpn, and the contaminating proteins themselves had no cytokine-inducing activity (Tabona et al 1998). These findings support the hypothesis that Cpn60 proteins have the ability to stimulate mammalian cells to produce a range of cytokines. Moreover, as will be discussed, they give clues to the structure-function relationships of the Cpn60 molecule. Cpn60 homologues have been reported to stimulate, in addition to human monocytes and rodent macrophages, cytokine production by a range of human cell types, including fibroblasts (Hinode et al 1998), leukocytes (Chopra et al 1997; Sharma et al 1997), epithelial cells (Marcatili et al 1997; Yamaguchi et al 1999), endothelial cells (Galdiero et al 1997; Kol et al 1999), and smooth muscle cells (Kol et al 1999). Indeed, it has been suggested that the Cpn60s should be classified as multiplex antigens because of their ability to interact with multiple cell types and elicit responses (Coates and Henderson 1998). Table 1 shows the response elicited from different cell populations after stimulation by Cpn60s from different sources.

Table 1.

 The stimulatory effect of chaperonin 60s (Cpn 60s) from different organisms on mammalian cells

It is now hard to argue against the hypothesis that Cpn60 proteins are signaling molecules capable of causing cells to produce different cytokines and inducing other specific cell effects, which will be discussed. An obvious question is—what is the biological significance of the capacity of the Cpns to activate cells? One plausible hypothesis is that the Cpn60s have a role as “danger signals.” Evidence has accumulated that the innate immune system acts to control the adaptive immune response (Medzhitov and Janeway 1997, 1998). The innate immune system alerts the adaptive immune system by a number of mediators—one of which is the secretion of cytokines by dendritic cells (DCs), macrophages, and natural killer cells. Hsp60 may play an important role as a stimulator of this pathway. The Danger Model of immunity (Matzinger 1998, 2001; Gallucci and Matzinger 2001) postulates that the immune system responds to molecules that cause damage to cells or tissues rather than to those that are foreign to the organism. In this model, for an immune response to occur, a cell has to be stressed or has to die under abnormal circumstances. APCs then become activated and send a costimulatory signal to initiate an immune response (Matzinger 1998; Gallucci and Matzinger 2001). Danger signals have been defined as molecules that are released or produced by cells undergoing stress or abnormal cell death and which can induce APCs to become activated (Gallucci and Matzinger 2001). Danger signals are classified as being either endogenous or exogenous (generated by the invading organism) and many categories of these signals may exist. For example, some are classified as primal—initiating the response, whereas others are feedback signals—which modulate the immune response. Some fall into both categories. Danger signals can be of various types: intracellular or extracellular and constitutive or inducible. In the latter case, they need to be synthesized or modified in some way to activate the APCs (Matzinger 1998; Gallucci and Matzinger 2001). Heat shock proteins (Hsps) could be classified as endogenous, inducible danger signals (Gallucci and Matzinger 2001). Evidence that Hsps initiate an immune response was provided by Basu et al (2000). Gp96, Hsp90, and Hsp70 are released by necrotic, but not by apoptotic, cells and stimulate macrophages to produce proinflammatory cytokines. These Hsps, on release by necrotic cells activate DCs to express costimulatory molecules, which induce an immune response. The Hsps interact with the APCs through the nuclear factor-κB pathway. The release of the Hsps from cells undergoing necrosis fulfils the criteria for an endogenous danger signal as proposed by Matzinger (1994).

Chen et al (1999) provided evidence that Hsp60 does act as a danger signal to the innate immune system. Both human and mouse macrophages released TNFα in a dose-dependent manner after stimulation by Hsp60. Hsp60 also stimulated the production of IL-12 and IL-15 and synergized with IFNγ to activate macrophages. Furthermore, Hsp60 induced antigen-specific IFNγ secretion and CD69 expression on CD4+ T cells undergoing primary stimulation (Breloer et al 2001). It is postulated that Hsp60 induces IL-12, which stimulates CD4+ T cells and induces IFNγ secretion without any further antigenic stimulation occurring. This is thought to lower the activation threshold of the antigen-specific T cell response and to result in a Th1 response. Thus, it is proposed, Hsp60 acts as a link between the innate and adaptive immune systems (Breloer et al 2001).

Although this review is concentrating on the biological activities of Cpn60, it is important to realize that other molecular chaperones have also been reported to modulate cellular cytokine synthesis. The ability of human hsp70 (Hsp70) to induce myeloid cell cytokine synthesis was described earlier (Asea et al 2000). Of interest are the reports that certain molecular chaperones act to induce the formation of anti-inflammatory signals. Thus both Bip, a 70-kDa Hsp involved in the folding of immunoglobulins (Corrigall et al 2001), and Hsp27 (De et al 2000) appear to induce the preponderant production of IL-10 relative to the production of proinflammatory cytokines (Table 2). This contrasts with, for example, the Cpn60 proteins of M tuberculosis that induce human monocytes to produce IL-10 but at much lower levels than IL-1 or TNFα (Lewthwaite et al 2001).

Table 2.

 Other molecular chaperones inducing cytokine synthesis

Vascular endothelial cell activation

Further evidence that Cpn60s act to promote inflammatory events is found in reports that mycobacterial, E coli, C trachomatis, and human Cpn60s can stimulate the expression of the cell adhesion proteins: E-selectin, intercellular adhesion molecule 1, and vascular cell adhesion molecule 1 on cultured human vascular endothelial cells and monocytes (Verdegaal et al 1996; Galdiero et al 1997; Kol et al 1999). These adhesion proteins are normally induced by LPS or by the early-response cytokines—IL-1 and TNFα. It is, therefore, surprising to find that Cpn60 proteins induce the synthesis of these adhesion proteins in a cytokine-independent manner (Verdegaal et al 1996; Galdiero et al 1997). Because monocytes and endothelial cells are key cells in the modulation of the inflammatory process, an important role for the Cpn60s in the modulation of leukocyte recruitment is implied. It has been shown that infection with C pneumoniae may be a contributory factor in atherogenesis (Mayr et al 2000; Roivainen et al 2000). C pneumoniae has been detected in atherosclerotic plaques (Grayston et al 1995; Saikku 1999), and a study by Mayr et al (2000) showed a link between the seropositivity to C pneumoniae and coronary heart disease (see review by Kinnunen et al 2001). Thus, the release of the Cpn60s may amplify the inflammatory response caused by bacterial infection and aid in the trafficking of monocytes and leukocytes and, thus, be implicated in the pathology of atheroma formation (Frenette and Wagner 1996; Kol et al 1999).

Cell stimulation by Cpn60: bone resorption

Studies to identify the potent bone resorbing activity released by the oral bacterium, A actinomycetemcomitans, revealed that the active moiety was the Cpn60 of this organism. It was shown that the E coli Cpn60 (GroEL) was also a potent inducer of bone resorption, but, surprisingly, mycobacterial Hsp65 proteins (Cpn60.2) lacked significant osteolytic activity (Kirby et al 1995). Bone resorption is normally controlled by the activity of multinucleate giant cells called osteoclasts. They arise from the maturation of the myeloid cell lineages that produce monocytes, macrophages, and DCs. Analysis of the mechanism of action of GroEL using bone marrow cells revealed that this Cpn had the capacity to activate preformed osteoclasts to resorb bone and that it was a potent and efficacious inducer of osteoclast generation from myeloid precursors (Reddi et al 1998). This capacity to induce osteoclast recruitments could be completely blocked by inhibition of the early-response cytokine, IL-1 (Kirby et al 1995), or by the cyclo-oxygenase inhibitor, indomethacin (Reddi et al 1998). A similar mechanism of action has been reported for the Cpn10 protein of M tuberculosis (Meghji et al 1997). Other molecular chaperones including Hsp70, Hsp90, and Hsp27 have also been reported to induce bone resorption in vitro (Nair et al 1999).

Modulation of cell proliferation

The stimulation of bone resorption appears to be because of the capacity of GroEL to stimulate the proliferation and differentiation of osteoclast myeloid cell precursor cells in the bone marrow. There are a growing number of reports that Cpn60 proteins from various sources can modulate cell cycle progression by a number of cell types. As has been described earlier, the Cpn60 protein of the oral bacterium, A actinomycetemcomitans, not only stimulated the proliferation of various cell types but also exhibited the capacity to cause apoptosis of the HaCaT keratinocyte cell line and porcine periodontal ligament cells (Goulhen et al 1998; Paju et al 2000). It has been suggested that the induction of epithelial cell proliferation by the A actinomycetemcomitans Cpn60 is due to the activation of ERK1/2 mitogen-activated protein (MAP) kinases (Zhang et al 2001). The C trachomatis Cpn60 stimulates the proliferation of human vascular smooth muscle cells by a pathway dependent on P44/P42 MAP kinases (Sasu et al 2001).

MECHANISMS OF CELLULAR ACTIVATION

The evidence so far discussed supports the hypothesis that Cpn60 proteins act as signaling molecules. But all cell-cell signaling molecules act through specific, or at least selective, receptors and induce selective signaling pathways. What do we know about these facets of Cpn60 signaling? An intriguing hypothesis, proposed by Janeway, suggests that the initial recognition of bacteria by host leukocytes uses germline-encoded receptors, termed pattern recognition receptors (PRRs), which recognize evolutionarily conserved bacterial structures that have been termed pathogen-associated molecular patterns (PAMPs) (Medzhitov and Janeway 1997; Janeway 1998). For example, the LPS of Gram-negative bacteria is an essential, highly conserved molecule present in the outer membrane and is the prototypic PAMP. Other proposed PAMPs are peptidoglycan, lipoarabinomannan, CpG deoxyribonucleic acid, RNA, and flagella. The original PRR was the LPS-binding protein CD14 (Pugin et al 1993). In more recent years, a new family of receptors, initially discovered in insects, the TLRs (Medzhitov et al 1997; Kimbrell and Beutler 2001) have been proposed to be cell surface proteins on APCs that can recognize microbial pathogens (bacteria, fungi, or protozoans) and discriminate between them. In turn, the TLRs interact with other cell surface proteins such as MD-2 (Shimazu et al 1999). CD14 also interacts with the TLRs (at least TLR4), although the exact mechanism has not been elucidated. The intracellular domain of the TLRs is homologous to that of the IL-1 receptor, and this signaling domain has been termed the Toll/IL-receptor (Tir) domain (Bowie and O'Neill 2000). The Cpn60 molecule is an evolutionarily highly conserved protein essential for bacterial function. It is, therefore, a likely candidate for being a PAMP and binding to PRRs.

Work from the authors' laboratory found that GroEL stimulation of human monocytes was not blocked by neutralizing antibodies to CD14 (Tabona et al 1998) and that the C3H/HeJ mouse, which is unresponsive to LPS and has been shown to have a point mutation in TLR4 (Poltorak et al 1998; Qureshi et al 1999), could be stimulated by the Cpn60 from A actinomycetemcomitans. In contrast, Kol et al (2000), using Hsp60 and the Cpn60 from C trachomatis, reported that Cpn60 proteins activate cells by the CD14 receptor and suggested that LPS and Hsp60 share a common receptor-mediated signaling pathway. The C3H/HeJ mouse has also been reported to be unresponsive to Hsp60 (Ohashi et al 2000). MD-2 was also found to be necessary for signal transduction. Vabulas et al (2001) have proposed a mechanism for Hsp60-induced cell activation in which this protein is endocytosed after binding to TLR4 or TLR2 and results in stimulation of the Tir signaling pathway through MYD88 and TRAF6. But on the basis of flow cytometric studies of Hsp60 binding to macrophages, it has been claimed that CD14-TLR2-TLR4 is not the primary receptor for Hsp60, although these proteins are important in macrophage activation. It was also concluded that the Cpn60 receptor is different from those for Hsp70, Hsp90, and gp96, which bind to the α₂-macroglobulin receptor (Habich et al 2002).

Although LPS is the prototypic PAMP, it does, in fact, demonstrate marked variation in its structure and activity (Henderson et al 1998), and there is evidence that LPS binds to a variety of cell surface receptors, including TLR4, TLR2, TREM-1 (Bouchon et al 2001), moesin (Tohme et al 1999), and a cell surface protein complex composed of proteins, including Hsp70 and Hsp90 (Triantafilou et al 2001a, 2001b). It is not established if all LPS molecules will bind to all these receptors. Certainly, enteric LPS does not bind to TLR2 (Beutler 2000). Moreover, it is now emerging that LPS from different bacteria can induce different classes of immune response in vivo. Thus, LPS from E coli signals through TLR4 and induces a Th1 cytokine profile, whereas LPS from the oral bacterium, Porphyromonas gingivalis, stimulates leukocytes, possibly via TLR2, and induces a Th2-like response (Pulendran et al 2001). Cpn60 is a prototypic conserved protein used in phylogenetic analysis (Karlin and Brocchieri 2000) because it demonstrates 50%–70% homology between species. But on the basis of the “glass half full/glass half empty” argument, this means that these proteins exhibit 30%–50% dissimilarity and it is a mistake to assume that all Cpn60s will act in the same way. The finding that Cpn60 proteins from various species have differences in their requirement for interaction with CD14-TLR4 suggests that Cpn60 is not a unitary protein and that sufficient differences exist in the sequence or structure of this protein among species to provide for a range of biological activity. An interesting example of this is M tuberculosis that encodes 2 Cpn60 proteins (Cpn60.1 and Cpn60.2 or Hsp65; Kong et al 1993). Analysis of these 2 recombinant proteins has revealed that they both have the capacity to induce human monocytes to synthesize proinflammatory cytokines, such as IL-1β, IL-6, IL-8, TNFα, and the anti-inflammatory cytokine, IL-10. They did not, however, induce the synthesis of IFNγ. Surprisingly, the M tuberculosis Cpn60.1 was 10- to 100-fold more potent than was the mtCpn60.2 in stimulating human monocyte cytokine synthesis. In addition, the mtCpn60.2 was unaffected by neutralizing antibodies to CD14, whereas the biological activity of mtCpn60.1 was almost totally blocked by such antibodies (Lewthwaite et al 2001). This corroborates earlier work, which reported that M tuberculosis Cpn60.2 stimulated monocyte cytokine synthesis (Zhang et al 1993) and vascular endothelial cell adhesion protein synthesis (Verdegaal et al 1996) in a CD14-independent manner. A further difference between these 2 proteins was that mtCpn60.1 contained a peptide sequence (195–219), which stimulated monocyte cytokine synthesis, including IFNγ, and was blocked by anti-CD14 antibodies. The same sequence in mtCpn60.2 was inactive (Lewthwaite et al 2001). Even more marked differences in human monocyte activation were seen when comparison was made between the recombinant Cpn60.1 and the Cpn60.3 proteins of the plant endosymbiont, Rhizobium leguminosarum. Both these recombinant proteins were active in protein folding assays, but only the Cpn60.3 was capable of stimulating human monocytes to produce proinflammatory cytokines (Lewthwaite et al 2002).

Thus, there appears to be heterogeneity in the response of human cells to Cpn60 proteins from different species (Table 3). The bioactivity of certain Cpn60 proteins (human, C trachomatis, M tuberculosis Cpn60.1, R leguminosarum Cpn60.3) are blocked by monoclonal antibodies to CD14, suggesting that these proteins bind to cell surface CD14 and possibly also to TLR4/2 to activate cells. Others (A actinomycetemcomitans, GroEL, M tuberculosis Cpn60.2) are unaffected by such neutralizing antibodies to CD14, suggesting that they bind to some other receptor. A final “group” (R leguminosarum Cpn60.1) fails to activate human monocytes. What accounts for these apparent differences in the ability to activate human cells? The interpretation of the available evidence would suggest that the biological activity of Cpn60 proteins is due to linear peptide epitopes. This is based on the finding that GroEL can be trypsinized without loss of activity (Tabona et al 1998) and the finding that the mycobacterial Cpn60 proteins had to be autoclaved to show any loss of bioactivity, whereas boiling, which should cause extensive loss of protein structure, did not affect bioactivity. It required extensive proteolysis with proteinase K to totally abolish activity (Lewthwaite et al 2001).

Table 3.

 Heterogeneity of chaperonin 60 (Cpn60) monocyte cytokine-stimulating bioactivity

Another piece of evidence for a role of linear peptide epitopes in the biological activity of Cpn60 comes from the article by Yoshida et al (2001), who are working on the antlion. This is a larval form of an insect (known in the United States as a doodlebug) that lives in a burrow and paralyzes its prey with a neurotoxin. Yoshida et al have identified the neurotoxin as the Cpn60 of an endosymbiotic bacterium, Enterobacter aerogenes, which is present in the insect's saliva. The sequence of the E aerogenes is almost identical to GroEL, and single-residue substitutions in GroEL are sufficient to turn this protein from an inactive molecule to an active insect neurotoxin. Such changes, which involve surface residues, are unlikely to have any major influence on the overall conformation of GroEL. Although single-residue changes are known to be able to markedly affect proteins—a good example being sickle-cell haemoglobin—there are few reports of a similar minor alteration in sequence conferring a potent, and wholly unexpected, biological activity on a protein.

CPN60S: THE PROTOTYPE OF A NEW SIGNALING FAMILY

There is evidence that highly purified, LPS-low Cpn60 proteins have the ability to activate mammalian cells. Other cell stress proteins, when added exogenously to eukaryotic cells, also have the ability to signal to the recipient cell. The obvious question that these findings stimulate is—what purpose does the ability of molecular chaperones to signal confer on the organism as a whole? Before attempting to answer this question, it is important to realize that both the host and the microbial pathogens capable of infecting the host produce highly homologous Cpn60 and other molecular chaperones. Moreover, our own human Cpn60 (Hsp60) is, in reality, a bacterial protein. Hsp60 is a mitochondrial protein, and the mitochondria are bacterial endosymbionts. With a little stretching of the imagination, the capacity of Cpn60 proteins to act as bacterial virulence factors may be analogous to the imaginary situation in which both the human host and its Gram-negative pathogens produce and use LPS.

Thus, we speculate that there are 2 functions for the signaling actions of Cpn60 proteins. The first is as a part of the PRR-PAMP system of innate immune signaling. Because Cpn60 is an inducible protein whose levels can rise 10-fold under stressful conditions, such as presumably occurs during infection, it would be sensible for multicellular organisms to have evolved to recognize this protein. Mammals certainly recognize Cpn60, and the magnitude of the immune response to this protein is surprising. Cpn60 is an immunodominant bacterial antigen, but given the similarity between bacterial and human proteins this has always been surprising. One potential explanation for the magnitude of the immune response to Cpn60 may be the fact that this protein can act as an adjuvant because of its ability, both as an intact protein and after proteolysis, to activate myeloid cells. Most protein immunogens have to be administered in an adjuvant that activates myeloid cells to express proteins such as CD80 and CD86—the essential costimulatory proteins required for antigen-specific T cell activation. It is not known whether Cpn60 stimulates the expression of these proteins on APCs, but Cpn60.2 peptides have been reported to induce CD80 or CD86 expression on antigen-specific T cells (Paul et al 2000). The major problem with the immune system recognizing bacterial Cpn60 is the homology between bacterial and human Cpn60. If the human protein were never secreted, this immunological cross-reactivity would pose no problem. But there is accumulating evidence that human Cpn60 is found on the surfaces of cells and is also released by cells and finds its way into the circulation. It has been hypothesized that atherosclerosis is due to the expression of Hsp60 on stressed vascular endothelial cells and the binding of antibodies to bacterial Cpn60 to the human protein. This is believed to induce complement-mediated killing of these stressed endothelial cells thus initiating the process of atherogenesis (Wick 2000).

One of the major surprises in the biology of the Cpns was the finding that Hsp60 is present in the blood. The initial finding that human Cpn60 was present in the serum of normal individuals (Pockley et al 1999; Rea et al 2001) has been followed by a number of reports of human Cpn60 being elevated in the serum of patients with atherosclerosis (Xu et al 2000) and borderline hypertension (Pockley et al 2000). But the levels of Hsp60 were not elevated in peripheral or renal vascular disease, although in these individuals, human Hsp70 levels were increased (Wright et al 2000). It has also been reported that the circulating levels of Hsp60 in the blood of normal individuals decrease with age (Rea et al 2001). The finding of elevated levels of Hsp60 in patients with cardiovascular disease could have particular consequences. If antibody binding to vascular endothelial cells expressing Hsp60 on their cell surface is an initiating and driving force in atherosclerosis, then soluble Hsp60 in the blood may be protective because it would limit interactions of antibodies with the vessel walls. The report that Hsp60 can act as a high-affinity plasma membrane receptor for high-density lipoprotein (Bocharov et al 2000) also suggests a protective role for soluble Hsp60. The source of the circulating Hsp60 is not established. Increased levels in the blood could arise through any of the 3 mechanisms. It is possible that normal apoptotic death could allow the release of these proteins. Indeed, it has been suggested that human Cpn60 and Cpn10 function in the process of apoptosis by accelerating the maturation of procaspase 3 (Samali et al 1999; Xanthoudakis et al 1999), although deliberate overexpression of Cpn60 can also protect cells against proapoptotic stimuli (reviewed by Garrido et al 2001). Cpn60 could be released as a consequence of necrotic cell death, although this would not explain the finding of Hsp60 in healthy individuals. The third mechanism could be by the upregulation of the release of Cpn60 by some form of selective cell secretion pathway(s). If the last is the case, then 1 plausible hypothesis to account for a mechanism of Cpn60 release (and the second mechanism of intercellular signaling) is as a means of producing large amounts of a cell-stimulating agonist to warn cells that they are in danger of being stressed. This would allow cells to become prepared for the induction of the stress. We have termed this putative mechanism as stress broadcasting (Fig 1). For this hypothesis not to be falsified, the cells exposed to stress should release Cpn60, or other inducible molecular chaperone, or have them exposed at the cell surface. Cells exposed to Cpn60 should be protected from stress. The possibility that this sharing of stress signals at the cellular level connects with the physiological signals of stress—the hypothalamic-pituitary axis—is an interesting speculation, which is also amenable to experimentation. Cpn60 can now be viewed as an intracellular protein-folding machine and as an extracellular signaling protein with actions similar to that of certain proinflammatory cytokines. Indeed, there is clear evidence for mutual signaling between cytokines and molecular chaperones (eg, Railson et al 2001; Kozawa et al 2001). It is possible that all molecular chaperones share this cell-cell signaling capacity. The many and varied findings described in this review guarantee that the next few years will be an active period of discovery as we try and piece together the various clues as to the possible biological role of molecular chaperones as cell signaling proteins. The recent report of a double-blind clinical study in which vaccination of patients with recently diagnosed type I diabetes with a Cpn60 immunomodulatory peptide was shown to preserve endogenous insulin production (Raz et al 2001) is another reason why interest in Cpn60 will remain high in the coming years.

Fig. 1.

 Schematic representation of the hypothetical process of stress broadcasting. It is proposed that one consequence of the effect of stress on a cell is to cause it to release molecular chaperones such as Cpn60 on to its cell surface and into the extracellular fluid. These released proteins (•) can then diffuse and interact with neighboring cells to “broadcast” the fact that there is a stressor in the cell's environment. This would act as an early warning to the cell and enable it to elevate the production of its own molecular chaperones, including Cpn60, and also other key factors such as cytokines