Plasma membrane rafts and chaperones in cytokine / STAT signaling

We and others have recently obtained data suggesting that cytokine-STAT signaling in many different cell-types is a chaperoned pathway initiated at the level of specialized plasma membrane microdomains called “rafts” (the “raft-STAT signaling hypothesis”). These findings are of broad significance in that all cytokines and growth factors initiate signaling in target cells by interacting with respective cell-surface receptors. The new data suggest that raft microdomains represent the units of function at the cell-surface through which ligand-stimulated STAT signaling is initiated. Moreover, recent evidence shows the involvement of chaperone proteins in regulating the STAT signaling pathway. These chaperones include the human homolog of the tumorous imaginal disc 1 protein (hTid1) which associates with Janus kinase 2 (JAK2) at the level of the plasma membrane, heat shock protein 90 (HSP90) which associates with STAT3 and STAT1 proteins in caveolin-1-containing raft and cytoplasVol. 50 No. 3/2003

mic complexes, and glucose regulated protein 58 (GRP58/ER-60/ERp57), a thiol dependent protein-disulfide isomerase, found in association with STAT3 "statosome" complexes in the cytosol and in the raft fraction.We suggest a function of the HSP90 chaperone system in preserving IL-6/STAT3 signaling in liver cells in the context of fever.The identification and function of protein partners associated with specific STAT species in rafts and in cytosolic complexes, and in the efficient departure of cytokine-activated STATs from the cytosolic face of rafts towards the cell nucleus are now areas of active investigation.
We and others have discovered that functional cytokine-stimulated STAT transcription factor signaling is a chaperoned pathway initiated at the level of the cholesterol-rich plasma membrane "raft" microdomain (Fig. 1).Today's insights into raft-STAT signaling and the role of chaperones which preserve this signaling pathway under temperature stress (such as during fever) emanate from two specific lines of research -one begun over 70 years ago in the area of clinical medicine, and the other begun over 30 years ago in the area of animal virology.In order to put the new insights in the area of rafts and chaperones in cytokine-STAT signaling into proper perspective we shall first briefly re-view this earlier literature, and then show how modern-day insights into cytokine/raft/ STAT signaling so elegantly relate back to the older clinical and biological observations.

THE ACUTE-PHASE RESPONSE
In a seminal paper, published in 1930, Tillet and Francis pointed out that patients with pneumococcal pneumonia upon admission to a hospital displayed high fever which was sustained for only few days (2-4 days), and then rapidly returned to baseline values (Tillet & Francis, 1930).These investigators collected serum from such patients during the pyrexic Cytoplasmic fractions designated P100 and S100 correspond respectively to the 100 000 ´g sedimentable and soluble fractions (Ndubuisi et al., 1999).
acute phase as well as during the later convalescent phase.Sera from patients in the acute phase already at admission contained very high titers of a material which gave a precipitin reaction with the somatic "C" polysaccharide fraction of peumococcus.In contrast, the agglutination titer in serum to a type-specific antigen of pneumococcus was low upon admission but gradually increased several days later.While the latter delayed response was readily understood to be due to the stimulation of specific anti-pneumococcal antibodies, the rapid and early high titers of fraction C-reactive precipitin in serum were difficult to explain.Moreover, patients with other bacterial infections accompanied by fever also displayed high serum titers of a precipitin reactive with the pneumococcal fraction C. The interpretation of these data at that time was in terms of the then newly recognized phenomenon of an anamanestic response.Tillet and Francis (1930) suggested that the rapid increase in serum titers of a precipitin reactive with the penumococcal fraction C in sera of patients with acute fever (with pneumococcal pneumonia or with infections caused by other bacteria) might be due to an anamanestic response by the host (i.e. the rapid stimulation by an antigen of the production of antibodies to unrelated antigens).
As happens often in science, the experimental observations have stood the test of time, while the particular interpretation has been replaced by an alternative one.We now know that the fraction C-reactive precipitin in sera of patients acutely ill with a variety of infections was not an antibody at all but a host protein now called the C-reactive protein (CRP) (reviewed in Koj, 1974;Kushner et al., 1982).The marked increases in serum CRP levels and a whole constellation of other plasma proteins is a hallmark of the "acute phase response" (reviewed in Koj, 1974;Kushner et al., 1982;Sehgal et al., 1989).
Through the second half of the twentieth century, CRP was purified and characterized, it was shown that CRP was synthesized in the liver, and that the increased production of CRP by the hepatocyte during acute infection and fever was the result of the release of a blood-borne mediator protein which travelled from the site of infection to the liver and upregulated CRP gene transcription ("the hepatocyte stimulating factor") (reviewed in Kushner et al., 1982).The major systemic regulator of CRP during the acute phase response was identified to be the cytokine interleukin-6 (IL-6) (Gauldie et al., 1987;reviewed in Sehgal et al., 1989) and the major transcription factor involved in upregulating the CRP gene promoter was identified to be STAT3 (reviewed in Mackiewicz et al., 1994).Shortly after the recognition of IL-6 as the hepatocyte stimulating factor, Aarden and colleagues in 1987 confirmed the relationship between elevated circulating levels of IL-6, elevated body temperature and the elevation of serum CRP levels (Nijsten et al., 1987).IL-6 is now known to be a potent central pyrogen (reviewed in Leon, 2002).These medical and biological observations present us with two questions.First, how is IL-6/STAT signaling initiated at the cell surface?Second, how is IL-6/STAT3 signaling maintained in hepatocytes in the face of elevated body temperature?New data obtained in recent years show that cytokine receptors are localized in specialized cholesterol-rich detergent-resistant microdomains of the plasma membrane called "rafts".These raft microdomains appear to represent the unit of function through which cytokine signaling is initiated (the raft-STAT signaling hypothesis).Moreover, recent data show the involvement of the chaperone heat shock protein 90 (HSP90) in preserving IL-6 signaling in liver cells during fever (Shah et al., 2002).terol and sphingolipids in the plane of the plasma membrane bilayer (Simons & Toomre, 2000;Galbiati et al., 2001;Anderson et al., 2002;Simons & Ehehalt, 2002).It is now widely accepted that lipid rafts organize the plasma membrane into functional units.Raft microdomains represent collecting sites on the cell surface for multiple glycophosphatidyl (GPI)-anchored membrane proteins, and for proteins targeted to rafts through binding to integral raft proteins or through non-covalently bound "lipid shells".These raft domains can then act as platforms for conducting a variety of cellular functions, including signal transduction (Fig. 1).
It is worth taking another detour through history in order to understand the origins of the lipid raft concept and of many of the techniques used today to investigate raft structure.These origins lie in the studies of animal virologists over 30 years ago who investigated the maturation and budding of enveloped RNA viruses from the plasma membrane.As examples, immunoelectronmicroscopic studies of cells infected with the parainfluenza virus Simian Virus 5 showed the insertion of viral glycoproteins into "localized patches in the plasma membrane" (Compans & Choppin, 1973).In subsequent steps of the maturation process viral nucleocapsid proteins were observed to be apposed to the cytoplasmic face of such patches followed by assembly of the nucleocapsid structure and eventual viral budding.Choppin and colleagues analysed the lipid composition of virus-associated plasma membrane and compared it to that of the plasma membrane of the host cell (Klenk & Choppin, 1970a;1970b).These investigators observed that, while the overall composition of viral lipids reflected that of the plasma membrane of the host cell from which the virus was derived, there was a relative enrichment of cholesterol and sphingolipids in the virus-associated plasma membrane.Today we recognize that lipid raft microdomains represent cholesterol-and sphingolipid-enriched dynamic patches in the plasma membrane (Simons & Toomre, 2000;Galbiati et al., 2001;Anderson & Jacobson, 2002;Simons & Ehehalt, 2002).
Beginning with the observations of Sabatini and colleagues in the late 1970s it is now established that specific enveloped viruses mature and bud from cells in a polarized manner (Rodriquez-Boulan & Sabatini, 1978).As examples, influenza virions mature on the apical side of polarized epithelial cells, while vesicular stomatis virions mature from the basolateral side.In studying how viral glycoproteins were selectively inserted into respective plasma membrane microdomains, virologists developed the technique of isolating the plasma membrane fraction, then treating it with non-ionic detergents followed by equilibrium density flotation centrifugation.The cholesterol-rich plasma membrane "raft" which resisted solubilization by non-ionic detergents (such as Triton X-100) in the cold (at 4°C) had a lower bouyant density and floated to the top carrying with it virus-specific glycoprotein complexes (Kurzchalia et al., 1992).In 1992, Simons and colleagues extended such flotation assays using uninfected cells and discovered that normal cells also yielded a flotation raft fraction (Kurzchalia et al., 1992).Upon characterizing the major proteins in such flotation rafts derived from uninfected cells, these investigators identified a major protein of 21 kDa (VIP21).An antibody to VIP21 showed that this protein was localized to plasma membrane invaginations that had been discovered previously and called "caveolae".Independently, the same protein was identified as the major cellular substrate for Tyr-phosphorylation by v-src in chicken cells (Rothberg et al., 1992).Again, an antibody to this protein localized it to caveolae.This molecule was thus called caveolin (now caveolin-1 or cav-1) (Rothberg et al., 1992) and represents the first "integral" raft protein to be identified.When cav-1 is inserted into a lipid raft from the cytoplasmic surface, the protein self-stacks and oligomerizes (n ³ 15) through its cav-1 stacking domain into a supramolecular assembly which results in flask-shaped invaginations of the plasma membrane of diameter 50-100 nm (the "caveolae") (Okamoto et al., 1998) (Fig. 1).Integral/structural raft proteins now known include caveolin-1, 2, 3, flotillin-1, 2, and stomatins (reviewed in Galbiati et al., 2001).It is now recognized that not all plasma membrane lipid rafts contain cav-1, nor are all lipid rafts restricted to caveolar structures.In a validation of the older virological literature, recent studies have shown that the maturation of an enveloped virus, such as respiratory syncytial virus, from the cell surface is preceded by the co-localization of cav-1 and viral glycoproteins into the same plasma membrane rafts and that cellular cav-1 is found in the mature virion (Brown et al., 2002).
Inhibitors of lipid raft function include agents such as filipin III and methyl-b-cyclodextrin (MCD) which extract cholesterol out of the cholesterol-rich rafts and peptide inhibitors corresponding to the cav-1 self-stacking domain and cav-1 binding domains from vari-ous target proteins (Simons & Toomre, 2000;Galbiati et al., 2001).The resulting modulation of signaling is cell-type and ligand-dependent in that in some systems agents such as filipin III and MCD inhibit signaling, whereas in others disruption of raft function enhances signaling (see Muller et al., 2001, and references cited therein).
The incorporation of cav-1 oligomers into the cytosolic face of plasma membrane domains is only one mechanism to form and stabilize lipid raft microdomains.The plasma membrane of hematopoietic lineage cells, which usually express little cav-1, is also organized into detergent-resistant raft microdomains (Fra et al., 1994) which can function in the context of T and B cell signaling, and even cytokine signaling (see below).

INITIATION OF CYTOKINE-STIMULATED STAT SIGNALING IN CAV-1-CONTAINING, CHOLESTEROL-RICH PLASMA MEMBRANE RAFTS (THE "RAFT-STAT SIGNALING" HYPOTHESIS)
Recent data from this and several other laboratories show the localization of the JAK kinases 1 and 2 and Tyk2, the transcription factors STAT1 and STAT3 as well as activated PY-STAT1 and PY-STAT3, and the cytokine receptor chains for IL-6, IL-2, interferon-g (IFN-g), IFN-a/b and tumor necrosis factor-a (TNF-a) (gp130, IFN-gR1 and IFN-gR2, IL-2Ra, IFNAR1, IFNAR2, and the TNF-aR CD120a chain respectively) in detergent-resistant cav-1-containing plasma membrane raft fractions (see citations in Table 1).Table 1 and Fig. 1 summarize the new studies implicating rafts in cytokine/STAT signaling and the observations reported.
From a functional standpoint, we showed that the raft disrupters MCD markedly inhibited IL-6/STAT3 and IFN-g/STAT1 signaling suggesting that almost all signaling by these cytokines was initiated in cholesterol-rich raft microdomains (Sehgal et al., 2002;Shah et al., 2002).Moreover, Taniguchi and colleagues (Takaoka et al., 2000) and Subramaniam & Johnson (2002) reported that the raft disrupter filipin III inhibited IFN-g signaling in a reversible manner.The data of Subramaniam & Johnson (2002), derived from the cav-1-free Jurkat T cells, highlight the function of lipid rafts in cytokine signaling independent of the presence of cav-1.Clearly, there is now increasing evidence suggesting that substantial if not all cytokine/STAT sig-naling is initiated at the level of raft microdomains in diverse cell-types (the "raft-STAT signaling hypothesis") (Sehgal et al., 2002).

CHAPERONES INVOLVED IN CYTOKINE-JAK-STAT SIGNALING
Chaperone proteins regulate the efficiency of biochemical and cellular processes, and protect the ongoing integrity of such pro- Table 1.Recent studies reporting raft-STAT signaling cesses in the face of stress.Recent data from several laboratories show that cytokine/ JAK/STAT signaling is a chaperoned pathway (Table 2 and Fig. 1).Pestka and colleagues used a yeast two-hybrid selection system to isolate proteins which interacted with JAK2.Among the proteins they identified was a DnaJ-like chaperone called Tid1 (Sarkar et al., 2001).Human Tid1 (hTid1) is a chaperone which ordinarily mediates the import of proteins into mitochondria.However, Pestka and colleagues provided clear and convincing evidence for the presence of hTid1 in the cytosol and in the plasma membrane, its interaction with JAK2 in unstimulated cells, and the departure of Tid1 from JAK2-containing membrane complexes upon treatment of cells with IFN-g.These authors suggested a "sequestration" model for the action of Tid1 in that this chaperone sequestered Jak2 in an inactive complex with the IFN-g receptor at the level of the plasma membrane, and cytokine stimulation released Jak2/the receptor chains from this sequestration.Overexpression of Tid1 inhibited functional IFN-g/JAK/STAT1 signal-ing as assayed using STAT1-responsive reporter/luciferase constructs.
We have identified the chaperone and disulfide protein-isomerase GRP58/ER-60/ ERp57 as a STAT3-associated protein at the level of the plasma membrane raft and in cytosolic complexes (Ndubuisi et al., 1999;Sehgal, 2000;2001;Guo et al., 2002).Excess recombinant GRP58 inhibited the DNA-binding competence of cytosolic PY-STAT3 (Guo et al., 2002).In transient transfection experiments overexpression of GRP58 in Hep3B cells inhibited both the basal activity and the ability of IL-6 to activate STAT3-responsive promoter-luciferase reporter constructs (unpublished data).Thus, like Tid1, GRP58 may function to sequester activated PY-STAT3 in cytokine-stimulated cells.
The identification of HSP90 as a potential candidate STAT3-associated protein (Shah et al., 2002) was intriguing in that it is already known that other raft-involved signaling molecules such as eNOS also interact physically and functionally with both cav-1 and HSP90 (reviewed in Fleming & Busse, 1999, and in Kone, 2000).Using the immunomagneticbead panning technique we confirmed that HSP90 was cross-precipitated by anti-STAT3 pAb from the membrane raft and cytoplasmic fractions (Shah et al., 2002).HSP90 was also included with STAT3 in complexes immunopanned using anti-cav-1 pAb.Moreover, anti-GRP58 pAb also crosspanned HSP90 together with STAT3 from cytosolic complexes.STAT1 was also included in these cross-immunopanned complexes from the membrane and cytoplasmic fractions.Taken together, our magnetic-bead immunopanning data indicated that STAT3, cav-1 and HSP90 interacted within detergent-resistant physical units in the membrane raft and cytoplasmic compartments (Shah et al., 2002).
The HSP90 inhibitor geldanamycin (GA) was used to probe the function of this chaperone in STAT3 signaling (Shah et al., 2002).The design of these experiments involved maintaining confluent Hep3B cultures at 37°C or exposing them to 39.5°C for 12-16 h to raise their levels of HSP90 (a temperature stress equivalent to a fever of 103°F), followed by an evaluation of the effects of a 15 min pretreatment with GA (typically at 20 mM) on various IL-6-induced STAT3-mediated responses at the two temperatures.Remarkably, despite the elevated temperature used, functional IL-6/STAT3 signaling was preserved at 39.5°C at a level approximately half that at 37°C.In Hep3B hepatocytes, IL-6-stimulated PY-STAT3 activation and trafficking to the nucleus was only minimally inhibited at 37°C by GA.However, this inhibitor markedly reduced PY-STAT3 activation and trafficking at 39.5°C suggesting the critical need for HSP90 in preserving IL-6/STAT3 signaling at the higher temperature.Immunopanning and Western blotting analyses data confirmed that GA disrupted the physical interaction between HSP90 and STAT3 in cells at both 37°C and 39.5°C (Shah et al., 2002).These observations suggested that HSP90 played a critical role in preserving IL-6/STAT signaling particularly at the elevated temperature.
We also investigated the effect of GA on the IL-6-induced upregulation of a reporter/luciferase construct containing four copies of the STAT3-binding element (Shah et al., 2002).The inducibility of this reporter construct by IL-6 was maintained at 39.5°C at a level approximately half of that at 37°C.While GA had a modest inhibitory effect on IL-6-inducibility of this reporter construct at 37°C (4-fold reduction), this inhibitor dramatically reduced reporter construct inducibility at 39.5°C (10-fold reduction).These data show that while HSP90 did contribute towards chaperoning IL-6/STAT3 signaling at both temperatures as evaluated in STAT3/luciferase reporter assays, there was a particularly dramatic role for HSP90 in preserving STAT3 signaling at 39.5°C (a temperature which corresponds to a fever of 103°F).
The involvement of HSP90 in IL-6/STAT3 signaling has now been confirmed independently (Sato et al., 2003).In experiments carried out at 37°C, these investigators observed that GA modestly inhibited IL-6-induced STAT3/luciferase reporter activity (by 3-4-fold).Transfection of a constitutive overexpression construct for HSP90 into cells treated with GA reversed the ability of this inhibitor to reduce the level of IL-6-activated STAT3-luciferase activity.Moreover in cross-immunoprecipitation assays using cells transiently transfected with various deletion mutants of HSP90 and of STAT3, evidence was obtained suggesting that the N-terminal geldanamycin-binding domain in HSP90 interacted with the DNA-binding domain in STAT3 (Sato et al., 2003).
To summarize, it has now become apparent that cytokine-raft-JAK-STAT signaling is a chaperoned pathway.Three of the chaperones involved are Tid1, GRP58 and HSP90 (Table 2).The role of HSP90 in preserving raft-STAT3 signaling in the context of fever brings us full circle to the observations of Tillet and Francis over 70 years ago.It is particularly satisfying that biochemical and molecular studies in this laboratory in recent years have provided an explanation for how IL-6/STAT3 signaling continues to function during fever.Although IL-6 is pyrogenic and appears in the body during the "acute phase" at a time when there is elevation of body temperature, the HSP90 chaperone system provides an elegant mechanism to preserve IL-6/STAT3 signaling in hepatocytes during fever ensuring the sustained production of "protective" acute-phase plasma proteins.

A DISEASE MECHANISM INVOLVING CAV-1 RAFT DISRUPTION AND HYPERACTIVATION OF IL-6/STAT3 SIGNALING: PULMONARY HYPER-TENSION IN THE MONOCROTALINE (MCT)-TREATED RAT
We have associated the disruption of cav-1 raft mechanisms and consequent hyperactivation of IL-6/PY-STAT3 signaling with a specific disease mechanism in an experimental animal model (Mathew et al., 2003).Rats treated with a single injection of the plant alkaloid monocrotaline (MCT) develop progressive pulmonary hypertension (PH) 10-14 days later (reviewed in Shultze & Roth, 1998;Bhargawa et al., 1999).It has been previously shown that the development of PH in this model is accompanied by upregulation of IL-6 production locally in the lung tissue (Bhargava et al., 1999).It has also been shown previously that the liver converts inactive MCT into the active pyrrole moiety which damages the pulmonary arterial endothelium during its first pass through the lung within 48 h (Shultze & Roth, 1998).We observed a marked downregulation of cav-1 in lung extracts prepared from MCT-treated rats within 48 h accompanied by a subsequent increase in tissue levels of PY-STAT3 (Mathew et al., 2003).A reduction in cav-1 mRNA, but not in cav-2 or cav-3 mRNA levels, also occurred in the lungs of MCT-treated rats by 7-14 days.It has been suggested that the initiating mechanism in this model of PH may be the disrup-tion of cav-1 rafts by MCT-pyrrole followed by the dysregulation of cytokine and STAT signaling, an hyperproliferative response of the pulmonary endothelium, and vascular lumen blockage.Consistent with this proposal is the observation that cav-1 (-/-) null mice develop marked pulmonary hypertension and right heart failure (Zhao et al., 2002).

CONCLUDING COMMENTS
Our discovery of STATs in rafts was an extension of work characterizing the subcellular distribution of STAT proteins using conventional cell-fractionation techniques.The process of cytokine receptor/raft/STAT signaling is likely to be broad significance in that this model may apply to signaling by STATs in, perhaps, all cell-types.The functional insights into the role of HSP90-based chaperones in STAT3 signaling are also likely of broad relevance in that these may relate to the mechanism of action of most cytokines, particularly in the context of elevated body temperature.The new data suggest that the efficient departure of activated STATs from the cytosolic face of the plasma membrane raft towards the cell interior and subsequent transcytoplasmic trafficking involves specific protein partners (including cav-1, clathrin, adaptins) and chaperones (including Tid1, GRP58, HSP90).The possibility that disruption of cav-1/raft/STAT signaling mechanisms may represent an intiating event in the pathogenesis of a specific disease process -the development of pulmonary hypertension in the MCT-treated rat -is now receiving experimental attention.Overall, the studies reviewed above highlight an as yet largely unexplored area of cytokine/STAT signaling -the subcellular anatomy of this signaling pathway at the level of the plasma membrane raft and in the cytoplasm.
It is a privilege and a pleasure to contribute to a special issue of this journal published in honor of Prof. Dr. Aleksander Koj.I have known Dr. Koj both as a scientist and as a friend these last fifteen years and truly value and cherish the perspective and joy that he brings to the process of doing science.It has indeed been a pleasure to be part of Dr. Aleksander Koj's scientific family.His perspective and incisive comments have served to keep us all honest and focussed on the important questions in front of us.It is the novel mixture of intellectual rigor and personal softness that makes any discussion with him so memorable.This brief review of only one aspect of recent work in the cytokine field, which is a shortened adaptation of a longer re-view (Sehgal & Shah, 2003), is dedicated to this man of great vision.
The XXXth Winter School of the Faculty of Biotechnology of the Jagiellonian University in Zakopane (February 28, 2003to March 4, 2003) organized by Prof. Dr. Adam Dubin and Dr. Joanna Beretta in honor of Dr. Koj's fifty years in science was a rich and rewarding experience.The focus on giving young graduate students in the Department of Biotechnology the opportunity to discuss their observations and mix and mingle with more established scientists from around the world is to be highly commended.Well done!And, thanks for the memories.

Figure 1 .
Figure 1.The raft-STAT signaling hypothesis and chaperoned transcytoplasmic trafficking of STATs.