Clusterin is an ATP-independent chaperone with very broad substrate specificity that stabilizes stressed proteins in a folding-competent state.

We recently reported that the ubiquitous, secreted protein clusterin has chaperone activity in vitro [Humphreys et al. (1999) J. Biol. Chem. 274, 6875-6881]. In this study, we demonstrate that clusterin (i) inhibits stress-induced precipitation of a very broad range of structurally divergent protein substrates, (ii) binds irreversibly via an ATP-independent mechanism to stressed proteins to form solubilized high molecular weight complexes, (iii) lacks detectable ATPase activity, (iv) when acting alone, does not effect refolding of stressed proteins in vitro, and (v) stabilizes stressed proteins in a state competent for refolding by heat shock protein 70 (HSP70). Furthermore, we show that, at physiological levels, clusterin inhibits stress-induced precipitation of proteins in undiluted human serum. Clusterin represents the first identified secreted mammalian chaperone. However, reports from others suggest that, at least under stress conditions, clusterin may be retained within cells to exert a protective effect. Regardless of the topological site(s) of action, the demonstration that clusterin can stabilize stressed proteins in a refolding-competent state suggests that, during stresses, the action of clusterin may inhibit rapid and irreversible protein precipitation and produce a reservoir of inactive but stabilized molecules from which other refolding chaperones can subsequently salvage functional proteins.

pH-dependent changes in the in vitro ligand-binding properties and structure of human clusterin.

Clusterin is a glycoprotein which is locally overexpressed at sites of tissue damage or stress, leading to the proposal that it may be a cytoprotective protein. It has been shown that clusterin has chaperone-like activity, being able to protect proteins against precipitation under stress conditions. It has also been shown that local acidosis is common at sites of tissue damage or stress. We asked whether acidic pH induces structural changes in clusterin and enhances its ability to bind to other proteins. We found by affinity chromatography and ELISA that the binding of clusterin to glutathione-S-transferase, IgG, apolipoprotein A-I, and complement protein C9 was enhanced at mildly acidic compared to physiological pH. Analytical ultracentrifugation and gel filtration studies revealed that clusterin exists in different polymerization states with monomer occurring preferentially at pH 5.5 and multimeric species at pH 7.5. Although circular dichroism showed little difference in the alpha-helical and beta-sheet contents of clusterin at pH 5 compared to pH 7.5, evidence for pH-dependent structural changes in clusterin was obtained from fluorescence experiments. pH titrations showed reversible changes in the fluorescence of tryptophan residues in clusterin. There was a reversible 2-fold increase in the fluorescence of the extrinsic probe 4, 4'-bis(1-anilinonaphthalene-8-sulfonate) bound to clusterin at pH 5. 5 compared to pH 7.5. There was also a 3.5-fold increase in fluorescence resonance energy transfer from tryptophan residues in clusterin to 4,4'-bis(1-anilinonaphthalene-8-sulfonate) at pH 5.5 compared to pH 7.5. These data suggest that pH-induced changes in the structure of clusterin are responsible for its enhanced ability to bind protein ligands at mildly acidic pH.

A reexamination of the role of clusterin as a complement regulator.

Clusterin is a highly conserved glycoprotein which has been proposed to protect host cells against complement-mediated cytolysis. We tested the hypothesis that clusterin is a complement regulator using erythrocytes and cells which had been stably transfected with a membrane-anchored form of clusterin as targets for complement-mediated cytolysis. Clusterin gave dose-dependent protection of antibody-coated sheep erythrocytes against complement-mediated lysis by diluted normal human serum. There was a linear relationship between the concentration of clusterin giving 50% protection and the concentration of serum; extrapolation of this to the case of undiluted human serum showed that a clusterin concentration at least two orders of magnitude greater than its physiological plasma concentration would be needed to confer protection against complement-mediated cytolysis under physiological conditions. Physiological concentrations of clusterin did not protect rabbit erythrocytes against alternative complement pathway-mediated lysis using dilute human serum. Exogenous clusterin had no effect on lysis of human erythrocytes triggered by the addition of inulin to autologous human serum. Induction of cell-surface clusterin expression by L929 (murine fibroblast) cells which had been stably transfected with cDNA for human clusterin linked to DNA coding for the 44 C-terminal amino acid residues of CD55 did not protect the cells against complement-mediated lysis by either normal or clusterin-depleted human serum. These data suggest that clusterin may not be a physiologically relevant regulator of complement activation.

Clusterin has chaperone-like activity similar to that of small heat shock proteins.

Clusterin is a highly conserved protein which is expressed at increased levels by many cell types in response to a broad variety of stress conditions. A genuine physiological function for clusterin has not yet been established. The results presented here demonstrate for the first time that clusterin has chaperone-like activity. At physiological concentrations, clusterin potently protected glutathione S-transferase and catalase from heat-induced precipitation and alpha-lactalbumin and bovine serum albumin from precipitation induced by reduction with dithiothreitol. Enzyme-linked immunosorbent assay data showed that clusterin bound preferentially to heat-stressed glutathione S-transferase and to dithiothreitol-treated bovine serum albumin and alpha-lactalbumin. Size exclusion chromatography and SDS-polyacrylamide gel electrophoresis analyses showed that clusterin formed high molecular weight complexes (HMW) with all four proteins tested. Small heat shock proteins (sHSP) also act in this way to prevent protein precipitation and protect cells from heat and other stresses. The stoichiometric subunit molar ratios of clusterin:stressed protein during formation of HMW complexes (which for the four proteins tested ranged from 1.0:1.3 to 1.0:11) is less than the reported ratios for sHSP-mediated formation of HMW complexes (1.0:1.0 or greater), indicating that clusterin is a very efficient chaperone. Our results suggest that clusterin may play a sHSP-like role in cytoprotection.

Apolipoprotein J (clusterin) induces cholesterol export from macrophage-foam cells: a potential anti-atherogenic function?

Apolipoprotein J (apo J) is a secreted glycoprotein of which the exact function remains a matter for speculation. Apo J has been implicated in such diverse processes as sperm maturation, regulation of complement activation, programmed cell death, tissue remodelling and lipid transport. In this study a possible role for apo J in lipid transport was explored. Mouse peritoneal macrophages were incubated with acetylated low-density lipoprotein (AcLDL) to produce foam cells containing cholesterol and cholesteryl esters. Incubation of the foam cells with physiological concentrations of purified apo J led to a dose-dependent export of cholesterol. The appearance of cholesterol in the medium was associated predominantly with a decline in intracellular cholesteryl esters rather than intracellular free cholesterol. The kinetics of cholesterol release to apo J were similar to apo A-I, an established promoter of cholesterol efflux. Apo J was also shown to induce phospholipid efflux from cells, whereas the cholesterol exported to the medium was associated with the apo J. Studies using foam cells from apo E-null mice showed that the cholesterol exported to the medium was independent of apo E production by the cells. These results present the first evidence that apo J can promote cholesterol efflux from foam cells and indicates that it might have a function in cellular cholesterol homoeostasis in both normal and pathological situations.

Characterization of the heparin-binding properties of human clusterin.

Clusterin is a highly conserved mammalian glycoprotein which has been predicted to contain heparin-binding sites. We tested this prediction by studying the interactions between heparin and clusterin using ELISA and heparin affinity chromatography methodologies. Two forms of biotinylated heparin were used in ELISA: heparin which had been directly biotinylated with a biotin-N-hydroxysuccinimide ester and heparin which had been activated using epichlorohydrin and 1,6-diaminohexane prior to biotinylation. Both gave dose-dependent increases in ELISA signal with increasing concentrations of biotinylated heparin, with the latter giving signals an order of magnitude greater than the former. There was a dose-dependent increase in the ELISA signal from bound biotinylated heparin with increasing concentrations of plate-bound clusterin. The apparent affinity constant for binding of biotinylated heparin to plate-bound clusterin at pH 6.0 was estimated as 0.06 +/- 0.02 microM. Unlabeled heparin blocked the binding of biotinylated heparin to clusterin over a concentration range similar to that of the binding of biotinylated heparin to plate-bound clusterin. The binding of biotinylated heparin to clusterin was independent of the presence or absence of Ca2+. The binding of biotinylated heparin to plate-bound clusterin increased with decreasing pH over the range 5.5-8.0 and was characterized by an apparent pKa of 6.9. Clusterin in human serum bound to heparin-Sepharose at pH 6.0 but not at pH 7.4. Dot-blot experiments showed that one of the polypeptide chains of clusterin which had been reduced and alkylated under denaturing conditions bound to heparin-Sepharose. This chain was identified as the alpha chain from its N-terminal amino acid sequence.

Histidine-rich glycoprotein binds to human IgG and C1q and inhibits the formation of insoluble immune complexes.

Purification of the complement component C1q from human serum using an established method resulted in the copurification of two 30 kDa proteins with an N-terminal sequence identical to human histidine-rich glycoprotein (HRG). Therefore, to explore the possibility that HRG can interact with C1q, we examined the ability of 81 kDa (native) and the 30 kDa proteins (presumably proteolytic N-terminal fragments of HRG) to bind to C1q, using both ELISA and optical biosensor techniques. Both forms of HRG were found to bind to the human complement component C1q and also to purified human and rabbit IgG by ELISA. Kinetic analyses of the HRG-C1q and HRG-IgG interactions using the IAsys biosensor indicate two distinct binding sites with affinities Kd1 0.78 x 10(-8) M and Kd2 3.73 x 10(-8) M for C1q, and one binding site with affinity Kd 8.5 x 10(-8) M for IgG. Moreover, the fact that both native and 30 kDa HRG bind to C1q and to IgG suggests that the IgG and C1q binding regions on HRG are located in the 30 kDa N-terminal region of the HRG molecule. The Fab region of IgG is likely to be involved in the HRG-IgG interaction since HRG also bound to F(ab')2 fragments with an affinity similar to that seen with the complete IgG molecule. Interestingly, the binding between HRG and IgG was significantly potentiated (Kd reduced from 85.0 to 18.9 nM) by the presence of physiological concentrations of Zn2+ (20 microM). Conversely, the presence of Zn2+ weakened the binding of HRG to C1q (Kd increased from 7.80 to 29.3 nM). Modulation of these interactions by other divalent metal cations was less effective with relative potencies being Zn2+ > Ni2+ > Cu2+. An examination of the effect of native and 30 kDa HRG on the formation of insoluble immune complexes (IIC) between ovalbumin and polyclonal rabbit anti-ovalbumin IgG revealed that physiological concentrations of HRG can markedly inhibit IIC formation in vitro. The results show that human HRG binds to C1q and to IgG in a Zn2+-modulated fashion, and that HRG can regulate the formation of IIC in vitro, thus indicating a new functional role for HRG in vivo.

Effects of clusterin overexpression on TNFalpha- and TGFbeta-mediated death of L929 cells.

Clusterin is a widely distributed and highly conserved protein for which many functions have been proposed. We used transfected L929 cells to study the effect of clusterin expression on the regulation of cell death signals. We showed that high levels of clusterin expression, about 0.2 pg clusterin secreted per cell per 48 h period, specifically protected L929 cells from TNFalpha-mediated cytotoxicity, while low expression (about 4 fg/cell/48 h) had no effect. However, clusterin expression did not provide transfected L929 cells with protection against death mediated by colchicine, staurosporine or azide. High level expression of clusterin in transfected L929 cells also potentiated the cytotoxicity of TGFbeta. It had previously been shown that exposure of L929 cells to TGFbeta provides protection against TNFalpha. We showed that this protective effect is not additive to that of clusterin expression. One interpretation of this data is that it suggests that clusterin and TGFbeta may act via a common mechanism to provide protection against the cytotoxicity of TNFalpha. Our results indicate that an intracellular action of clusterin protein is responsible for protection against TNFalpha cytotoxicity. Exposure to TNFalpha induces an increase in the level of cell-associated clusterin and specifically in the level of a novel clusterin molecule, which when analyzed under reducing conditions by SDS/PAGE and immunoblotting appears as two closely spaced bands at about 36 and 38.5 kDa. When analyzed under the same conditions, the normal form of intracellular clusterin, which is present with or without exposure to TNFalpha, appears as two poorly resolved bands at about 43-45 kDa. Since the novel form of clusterin is also expressed in cells exposed to TGFbeta, colchicine, staurosporine, and azide, it may result from toxin-induced disruption of processes of normal cellular protein production.

The formation of insoluble immune complexes between ovalbumin and anti-ovalbumin IgG occurs in at least two distinct phases dependent on reactant concentration and ionic strength.

The mechanism regulating the formation of insoluble immune complexes (IIC) in serum in certain disease states is not well understood. Ovalbumin and rabbit anti-ovalbumin IgG was used to study the formation of IIC in vitro in a stirred reaction vessel; and the radii of IIC that formed was determined by light scattering techniques. Using an initial IgG concentration of 1 mg/ml at equivalence antigen:antibody ratio IIC formation was detected within 5 s, and the complexes increased in radii to approx. 100 nm after 20-30 s (phase 1). This was followed by a phase (phase 2) in which the complexes rapidly increased in radii to the point where Mie scattering was reached (approximately 200 nm). The time of onset of the second phase decreased with increasing initial IgG concentrations at a fixed antigen:antibody ratio; and was at a minimum at equivalence antigen:antibody ratio, but increased at both antigen and antibody excess ratios. Immune complexes formed using F(ab')2 fragment showed a similar pattern to those formed using IgG. A similar pattern was seen in the presence of the complement component C1q which potentiated IIC formation in phase 2, and human serum (1:10 dilution) which attenuated IIC formation in both phases. For complex formation using IgG and ovalbumin the presence of NaC1 at concentrations up to 0.6 M led to a progressive increase in the time of onset of phase 2; potencies of inhibition by other sodium halides followed the lyotropic series NaF < NaC1 < NaI. The results suggest that formation of IIC occurs in at least two distinct phases, and that the second phase leading to the generation of very large insoluble complexes is associated with a rapid polymerisation of the complexes by a mechanism that is not dependent on Fc:Fc interactions.

C1q is a nucleotide binding protein and is responsible for the ability of clusterin preparations to promote immune complex formation.

Clusterin prepared from human serum by monoclonal antibody affinity chromatography was devoid of the ability to increase the rates of formation of insoluble immune complexes associated with clusterin preparations obtained by polyclonal IgG affinity chromatography. Clusterin did not bind to AMP-Sepharose but the protein responsible for increasing the rates of formation of insoluble immune complexes did bind to this affinity matrix. This protein was identified as complement protein C1q on the basis of its behaviour on SDS/PAGE and reactivity in sandwich ELISA with monoclonal antibodies specific for C1q. C1q (identified from its behaviour on SDS/PAGE, immunoreactivity with C1q-specific monoclonal antibodies and N-terminal sequencing data) was purified from serum by AMP-Sepharose chromatography. The binding of C1q to AMP-Sepharose was inhibited by adenine nucleotides.

Development of a sensitive peptide-based immunoassay: application to detection of the Jun and Fos oncoproteins.

c-Jun and c-Fos belong to the bZIP class of transcriptional activator proteins, many of which have been implicated in the neoplastic transformation of cells. We are interested in engineering dominant-negative leucine zipper (LZ) peptides as a means of sequestering these proteins in vivo in order to suppress their transcriptional regulatory activity. Toward this end, we have developed a novel immunoassay for measuring the dimerization affinities of dimeric Jun and Fos complexes. This peptide-based ELISA relies on the fact that Jun and Fos preferentially form heterodimers via their leucine zipper domains. Recombinant Jun leucine zipper peptides (either native JunLZ or a V36 --> E point mutant) were labeled with biotin and specifically bound through a leucine zipper interaction to a FosLZ-glutathione S-transferase fusion protein adsorbed onto the wells of an ELISA tray. Jun:Fos complexes were subsequently detected using a recently developed streptavidin-based amplification system known as enzyme complex amplification [Wilson, M. R., & Easterbrook-Smith, S.B. (1993) Anal. Biochem. 209, 183-187]. This ELISA system can detect subnanomolar concentrations of Jun and Fos, thus allowing determination of the dissociation constants for complex formation. The dissociation constant for formation of the native JunLZ:FosLZ heterodimer at 37 degrees C was determined to be 0.99 +/- 0.30 nM, while that for JunLZ(V36E):FosLZ heterodimer was 0.90 +/- 0.13 microM. These results demonstrate that the novel peptide-based ELISA described herein is simple and sensitive and can be used to rapidly screen for potential dominant-negative leucine zipper peptides.

The effects of histidine residue modification on the immune precipitating ability of rabbit IgG.

Treatment of anti-ovalbumin rabbit IgG with diethylpyrocarbonate (DEPC) at concentrations up to 100 microM led to a progressive decrease in the rates of formation of insoluble immune complexes, without affecting the final extent of immune complex formation. DEPC concentrations approximately 10-fold higher were needed to give comparable decreases in the rates of immune complex formation by F(ab')2. Treatment of DEPC-treated IgG with hydroxylamine led to substantial restoration of the rates of formation of insoluble immune complexes. Carbethoxylation of two histidine residues per IgG molecule had little effect on rates of formation of insoluble immune complexes, but these rates were markedly decreased in samples of IgG with four to five histidines per molecule modified. There were parallel decreases in the protein A-binding activity and in the rates of formation of insoluble immune complexes in IgG treated with increasing concentrations of DEPC. The presence of complement protein C1q restored the rates of formation of insoluble immune complexes of DEPC-treated IgG.

A light-scattering method for measuring the sizes of insoluble immune complexes.

A turbidimetric method for measuring the diameters of insoluble immune complexes, based on the wavelength dependence of their ability to scatter light, was developed. The method was validated by demonstrating that it gave experimental values for the diameters of polystyrene microspheres which were in good agreement with independently known values of these. The method was used to measure the diameters of ovalbumin:anti-ovalbumin IgG immune complexes, giving values consistent with literature measurements of the sizes of other IgG-containing immune complexes.

Enzyme complex amplification--a signal amplification method for use in enzyme immunoassays.

An amplification system for enzyme immunoassays, in which complexes of streptavidin, biotinylated horseradish peroxidase, antibody are used to detect biotinylated target molecules was developed. In enzyme-linked immunosorbent assay (ELISA) experiments this enzyme complex enhancement (ECA) system gave up to a better than two orders of magnitude increase in sensitivity, compared to a horseradish peroxidase-conjugated streptavidin detection system. The ECA system was shown to be applicable to direct detection of biotinylated antigens, as well as in ELISA employing biotinylated antibodies and in sandwich ELISA. The ECA system was also applicable to immunoblots, giving a sensitivity comparable to that obtained using a radiochemical detection system.

Clusterin binds by a multivalent mechanism to the Fc and Fab regions of IgG.

Clusterin was purified from human serum by IgG and monoclonal antibody affinity chromatography. SDS-PAGE and immunoblotting revealed no major differences between clusterin prepared in these two ways. An ELISA method for measuring the binding of clusterin to immunoglobulins was developed. Clusterin purified by IgG affinity chromatography bound to pooled human IgG with a similar affinity (S0.5 5.9 +/- 0.4 micrograms/ml) as clusterin purified by monoclonal antibody chromatography (S0.5 6.1 +/- 0.2 micrograms/ml). The apparent affinity of clusterin for IgG immobilised on ELISA plates increased with increasing concentrations of IgG in the coating solution. Aggregated IgG in solution was a more potent inhibitor of the binding of clusterin to immobilised IgG than was monomer IgG. Clusterin bound to all of the isotypes of human IgG, and to human IgA and IgM, with apparent affinities in the order IgG3 > IgG4 > IgM > IgG1 > IgG2, IgA. Clusterin bound to both the Fab and Fc fragments of human IgG. The clusterin binding site(s) on the Fc do not overlap with those for protein A and Clq.

RHP is antigenically related to factor H and binds to the globular heads of C1q.

RHP was purified from normal serum by sequential euglobin precipitation, ion exchange chromatography on DEAE-Sephacel and gel filtration using Sephacryl S-300. RHP reacted with anti-Factor H antibodies in ELISA assays and in Western blots, suggesting that it is antigenically related to Factor H. It bound to intact C1q but not to the collagen-like N-terminal half of the molecule. C1q-specific monoclonal antibody BUS-1, which blocks the binding of C1q to immune complexes, did not block the binding of RHP to C1q. This implies that the binding sites on C1q for IgG and RHP do not overlap.

The Fab/c fragment of IgG produced by cleavage at cyanocysteine residues.

The intra- and inter-heavy chain disulfides of rabbit IgG were cleaved by mild reduction with either dithiothreitol or sulfite and cyanocysteines generated by treatment with either 2-nitro-5-thiocyanobenzoic acid or KCN, respectively. When cleavage occurs at a cyanocysteine residue in the hinge region of one heavy chain alone the Fab/c fragment is produced. Fab/c was also produced by papain digestion of IgG. Fab/c made by papain digestion was able to active complement in haemolytic assays; this activity was lost after cleavage of its accessible disulfide bonds. Fab/c made by cyanylysis of sulfite-reduced IgG was also active in these assays, but Fab/c made by cyanylysis of dithiothreitol-reduced IgG was not. Treatment of the latter fragment with cysteine and cystine resulted in partial reformation of cleaved disulfide bonds. Fab/c was also made from human IgG and from murine IgG2a and IgG2b.

Clusterin enhances the formation of insoluble immune complexes.

Clusterin was purified from human serum by sequential affinity chromatography over IgG-, protein A- and Con A-Sepharose. The protein was approximately 70 kDa by SDS/PAGE under nonreducing conditions and was resolved into approximately 35 kDa bands under reducing conditions. The protein reacted with clusterin-specific Mabs in ELISA and in Western blots. Its N-terminal sequences agreed with those published for clusterin. An antiserum specific for clusterin made by the above method detected it in complement membrane attack complexes on rabbit erythrocyte membranes. The interaction of clusterin with IgG was physiologically relevant because it was found to increase the rate of formation of insoluble immune complexes.

Carbodiimide crosslinking of human C1q and rabbit IgG.

The water soluble carbodiimide 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide was used to covalently link carboxyl groups on rabbit IgG to lysyl groups on complement protein C1q. The interaction between C1q and IgG was disrupted by varying the pH, modifying essential residues in the IgG binding site of C1q and by reducing the interchain disulfides of IgG. Under each of these conditions the correlation found between binding and crosslinking indicated a strong requirement for the proteins to bind normally in order for crosslinking to occur. SDS-PAGE analysis of the crosslinked material showed a 210 kDa band consistent with one IgG crosslinked to two disulfide linked C1q chains. Blotting and autoradiography showed the crosslinking involved the A and/or B and C chains of C1q. The lysines flanking the intrachain half cystines are proposed as the likely candidates for crosslinking to IgG, thus delineating the immunoglobulin binding site of C1q.