Acid-induced Conformational Changes in Phosphoglucose Isomerase Result in Its Increased Cell Surface Association and Deposition on Fibronectin Fibrils*

Phosphoglucose isomerase (PGI) is a glycolytic enzyme that exhibits extracellular cytokine activity as autocrine motility factor, neuroleukin, and maturation factor and that has been recently implicated as an autoantigen in rheumatoid arthritis. In contrast to its receptor-mediated endocytosis at neutral pH, addition of 25 μg/ml of either Alexa 568- or FITC-conjugated PGI to NIH-3T3 cells at progressively acid pH results in its quantitatively increased association with cell surface fibrillar structures that is particularly evident at pH 5. A similar pH-dependent cell surface association of PGI is observed for first passage human chondrocytes obtained from osteoarthritic joints. At acid pH, PGI colocalizes with fibronectin fibrils, and this association occurs directly upon addition of PGI to the cells. In contrast to the receptor-mediated endocytosis of PGI, fibril association of 25 μg/ml PGI at pH 5 is not competed with an excess (2 mg/ml) of unlabeled PGI. PGI binding at acid pH is therefore neither saturable nor mediated by its receptor. PGI is enzymatically active as a dimer and we show here by non-denaturing gel electrophoresis as well as by glutaraldehyde cross-linking that it exists at neutral pH in a tetrameric form. Increasingly acid pH results in the appearance of PGI monomers that correlates directly with its enhanced cell surface association. However, glutaraldehyde cross-linked PGI is endocytosed at neutral pH and still exhibits enhanced cell surface binding at pH 5. Circular dichroism analysis revealed pH-dependent changes in the near but not the far UV spectra indicating that the tertiary structure of the protein is specifically altered at pH 5. Conformational changes of PGI and exposure of the monomer-monomer interface under acidic conditions, such as those encountered in the synovial fluid of arthritic joints, could therefore result in its deposition on the surface of joints and the induction of an autoimmune response.

Glucose-6-phosphate isomerase or phosphoglucose isomerase (PGI) 1 is a glycolytic enzyme essential for neoglucogenesis that is equivalent to the autocrine motility factor (AMF)/neuroleukin/maturation factor (MF) cytokine (1)(2)(3)(4). PGI is therefore a cytosolic enzyme that upon release from the cell acquires a de novo function as a neurokine, lymphokine, and tumor cell cytokine (4 -8). While the mechanism of release of PGI remains uncertain, enhanced secretion of PGI following overexpression of PGI by stable transfection of NIH-3T3 cells induces cellular transformation and tumorigenicity (9). Serum PGI activity has long been reported and is associated with tumor expression (10,11) indicating that this protein is actively released from both normal and tumor cells.
PGI exists as a dimer and enzyme dimerization is necessary for its enzymatic activity (12)(13)(14)(15). The active site of the enzyme has been characterized by x-ray crystallography and is localized to the cleft between the two PGI monomers (16 -19). The motifs responsible for PGI cytokine activity remain to be determined. Inhibitors of PGI isomerase activity block its cytokine activity (1,20,21). Reports that the bacterial form of PGI, whose sequence homology with the mammalian enzyme is limited to residues in the enzymatic active site, presents cytokine activity further supported a role for the active site in PGI cytokine activity (17,22). However, neither cytokine activity or receptor binding of the bacterial or yeast forms of the enzyme were detected in NIH-3T3 cells (23) and punctual mutations in the PGI sequence that disrupt its enzymatic activity do not affect its cytokine function (24). The latter studies argue that motifs implicated in receptor binding include regions of the protein that present differences between the mammalian and bacterial forms of the enzyme including the N-terminal, Cterminal, and internal hook domains (18).
The PGI receptor, gp78 or AMF-R, is a seven-transmembrane domain G protein-coupled receptor (25). AMF-R expression is significantly increased in neoplastic tissue and its expression is correlated with tumor malignancy and poor survival and prognosis of patients with gastric, colorectal, bladder and esophogeal carcinomas, cutaneous malignant melanoma, and pulmonary adenocarcinoma (26 -32). In normal brain, AMF-R expression is increased during development and associated with learning and development implicating PGI cytokine activity in normal cellular activity (33,34). AMF-R is expressed both at the cell surface where it associates with caveolae as well as within a smooth domain of the endoplasmic reticulum (35)(36)(37)(38)(39). AMF-R internalizes its ligand via both caveolae/raft-dependent endocytosis to the smooth ER and clathrin-dependent endocytosis to multivesicular bodies (35, 40 -43). The latter pathway is associated with the recycling of AMF/PGI to cell surface fibronectin fibrils (40).
Interestingly, PGI has recently been identified as an autoantigen implicated in rheumatoid arthritis (RA) in the K/BxN mouse model as well as in humans (44,45). PGI and anti-PGI are specifically localized to the articular surface of joints (45)(46)(47). However, the basis for the selective binding of this protein to the surface of the synovial lining as well as for the generation of an immune response against this ubiquitous self-antigen remains a paradox (46,48). Indeed, the extent to which PGI autoantibodies are prevalent in the sera of RA patients remains controversial. While earlier reports indicated that 64% of sera from RA patients contain antibodies to PGI (45), more recent studies have questioned the prevalence and specificity of the PGI autoimmune response in RA (49 -51). Localization of the autoimmune response to PGI to lymph nodes adjacent to the affected joints in the K/BxN mouse led the authors to suggest that PGI in the joint is different in form or quantity than that circulating in other regions of the body (52). We demonstrate here the dramatically increased binding of PGI to fibronectin fibrils at acid pH that corresponds directly to PGI denaturation and, more specifically, to changes in PGI tertiary structure. Localized acidosis in synovial fluid could therefore enable denaturation and consequent binding of circulating PGI to synovial cell extracellular matrix permitting the generation of an autoimmune response against exposed non-native PGI epitopes.

EXPERIMENTAL PROCEDURES
Cells and Materials-NIH-3T3 fibroblasts were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% calf serum, vitamins, non-essential amino acids, glutamine, and penicillin-streptomycin antibiotics (Canadian Life Technologies) at 37°C in a humidified 5% CO 2 /95% air incubator as previously described (35). Human chondrocytes were obtained from articular cartilage (femoral condyles and tibial plateaus) of patients (aged 58 Ϯ 8, mean Ϯ S.D.) who had undergone total knee arthroplasty. All osteoarthritis (OA) patients were evaluated by a certified rheumatologist and were diagnosed as having OA based on the criteria developed by the American College of Rheumatology Diagnostic Subcommittee for OA (53). Chondrocytes were released from the articular cartilage by sequential enzymatic digestion at 37°C as previously described (54), and cultured in DMEM supplemented with 10% fetal bovine serum (Canadian Life Technologies) and penicillin-streptomycin antibiotics (Canadian Life Technologies) at 37°C in a humidified atmosphere of 5% CO 2 /95% air. Chondrocytes were used at first passage on Labtek plastic cover slips.
Rabbit PGI (P9544) was purchased from Sigma Chemical Co. (Oakville, Ontario). Mouse anti-fibronectin was purchased from Transduction Laboratories (Mississuaga, Ontario) and Alexa 488-conjugated anti-mouse secondary antibody from Molecular Probes (Eugene, OR). Nondenatured protein molecular weight markers were purchased from Sigma and Kaleidoscope SDS molecular weight markers from BioRad.
The Alexa Fluor 568 protein labeling kit and the FluoReporter® FITC protein labeling kit were purchased from Molecular Probes, and Alexa 568 and FITC-PGI conjugates were prepared according to the manufacturer's instructions. Briefly, PGI was incubated with either the succinimidyl ester of Alexa Fluor 568 carboxylic acid or fluorescein isothiocyanate in bicarbonate buffer (pH 8.3-9) for 1 h at room temperature. Fluorescent PGI conjugates were separated from free dye by size exclusion chromatography using either a column (Alexa 568) or spin column (FITC).
Immunofluorescence Labeling-60,000 NIH-3T3 cells were plated for 2 days on cover slips and incubated at 37°C for 5 or 30 min with Alexa 568 or FITC-conjugated PGI (25 g/ml) in bicarbonate containing medium in a CO 2 incubator (controls) or in bicarbonate-free medium supplemented with 100 mM HEPES at pH 7.5, 7.0, and 6.5 or with 100 mM MES at pH 6.0, 5.5, and 5.0 in a CO 2 -free incubator. Cells were then washed with PBS-CM and fixed with 3% paraformaldehyde. Cell associated fluorescence was then quantified using a Wallac Victor V fluorescence plate reader (PerkinElmer, Montreal, Quebec) and appropriate filter sets. Alternatively, labeled cells were visualized by confocal microscopy using a Leica TCS SP-1 confocal microscope. Where indicated, cells incubated with Alexa 568-PGI were labeled with anti-fibronectin antibodies and Alexa 488-conjugated anti-mouse secondary antibodies.
Non-denaturing Gel Electrophoresis-Lyophilized rabbit PGI was dissolved in water and protein concentration determined with the BCA protein assay (Pierce). 1.5 g of mammalian PGI was loaded on nondenaturating 8% acrylamide gels prepared with an upper gel using Laemmli buffers without SDS (55). To a solution of 1.5 g of PGI, HEPES or MES buffers at the indicated pH were added to a final concentration of 250 mM and then sample buffer consisting of glycerol and bromphenol blue was added. Protein bands and non-denatured protein markers (Jack bean urease: 272 kDa; bovine serum albumin: 132 and 66 kDa; chicken egg albumin: 45 kDa; carbonic anhydrase: 29 kDa) were revealed by silver staining.
PGI Cross-linking-Cross-linking of rabbit PGI with glutaraldehyde (MECALAB LTD. Montreal, Quebec, Canada) was carried out as previously described (56). Briefly, 0.5 mg/ml PGI was incubated with 0.1% (v/v) glutaraldehyde in 75 l of PBS for different times at room temperature. The reaction was stopped by addition of SDS sample buffer. Samples and SDS molecular weight markers were boiled and reduced, separated in 8% SDS-polyacrylamide gels, and protein bands revealed by staining with Coomassie Blue. Alternatively, Alexa 568 and FITC-PGI were cross-linked, diluted with 2 ml of 100 mM glycine for at least 30 min to quench the reaction and then concentrated using Amicon filter units and added to cells at 25 g/ml for fluorescent visualization as described previously.
Circular Dichroism-Circular dichroism (CD) analysis of PGI was performed using a Jasco J-710 spectropolarimeter (Dept. of Chemistry and Biochemistry, Concordia University, Montreal, Quebec). Spectra were recorded in a 0.1-cm quartz cuvette at room temperature in MESor HEPES-based buffers at pH 7.5, 6, 5.5, and 5 and background signal obtained from parallel scans of the buffer alone were subtracted from the measurements. Far UV spectra of 0.1 g/ml PGI were recorded from 260 to 190 nm at a speed of 100 nm/min in 0.2-nm steps and with a FIG. 1. Acid pH results in progressively increased fibrillar labeling of PGI. NIH-3T3 cells were incubated for 30 min with 25 g/ml Alexa 568-PGI in regular medium (control) or in bicarbonate-free medium containing 100 mM HEPES at pH 7.0 or pH 6.5 or 100 mM MES at pH 6.0, 5.5, or 5.0, as indicated. After fixation with 3% paraformaldehyde, and several washes with PBS-CM, cell-associated Alexa568-PGI was visualized directly by confocal microscopy. All images were acquired with the same confocal settings.
signal averaging time of 0.25 s. Near UV spectra of 0.5 g/ml PGI were recorded from 320 to 250 nm at a speed of 20 nm/min in 0.2-nm steps and with a signal averaging time of 2 s. The presented spectra are representative of at least three separate scans.

Increased Cell Surface Fibrillar Binding of PGI Binding at
Acid pH-PGI is an intracellular glycolytic enzyme however its ability to associate with the articular surface of joints (45)(46)(47) implicates the binding of exogenous PGI to the surface of cells. Incubation of NIH-3T3 cells with Alexa 568-conjugated PGI and fixation with paraformaldehyde results in the predominant visualization of its expression in multivesicular bodies (MVBs) as well as to more faintly labeled fibrils (40), a distribution still observed upon pH reduction of the medium to pH 6.5 ( Fig. 1). At pH 6.0, fewer MVBs are labeled, likely due to inhibition of clathrin-dependent endocytosis at low pH (57). Further reduction to pH 5.5 and 5.0 results in the progressive accumulation of fibril-associated PGI such that at pH 5.0 the extent of cell-associated PGI is dramatically increased (Fig. 1). At pH 5.0, Alexa 568-PGI also exhibits significant association with cell-free regions of the substrate. Identical results were obtained using FITC-conjugated PGI although increased recycling of FITC-PGI to cell surface fibrils was observed at neutral pH compared with Alexa 568-PGI (not shown). Quantification of the binding of both Alexa 568 and FITC-PGI as a function of pH shows a dramatic increase in fluorescence at pH 5 (Fig. 2). In particular, a significant increase in binding is seen between labeling at pH 5 and pH 5.5. A similar pH-dependent increase in binding of Alexa 568-PGI to human chondrocytes obtained from patients with osteoarthritis was also observed (Fig. 3). As for NIH-3T3 cells, a dramatic increase in binding at pH 5 was observed compared with pH 5.5.
PGI labeling of NIH-3T3 cells at pH 5.0 colocalizes with fibronectin fibrils and is also observed to associate with nonfibrillar cell surface domains and the cell-free substrate (Fig.  4). Fibrillar association of PGI at pH 5.0 can be observed as early as 5 min after incubation with PGI ( Fig. 4) indicating that PGI association with cell surface fibronectin occurs directly and not following PGI endocytosis and recycling (40). The enhanced binding of PGI to the cells at acid pH was not due to loss of cell integrity as subsequent incubation of the cells at neutral pH resulted in the internalization of cell associated PGI (Fig. 5, A  and B). While 2 mg/ml unlabeled PGI inhibits the endocytosis of 25 g/ml Alexa 568-labeled PGI at pH 7.5, the same concentration of unlabeled PGI had only a minimal effect on the fibrillar association of PGI at pH 5.0 (Fig. 5, C-F). Therefore, in contrast to the limited, saturable number of PGI binding sites at neutral pH (23,40,58), the association of PGI with cell surface fibronectin fibrils at pH 5.0 is not saturable, potentially nonspecific, and apparently not mediated by the PGI receptor, AMF-R.
Acid pH Induces Changes in PGI Conformation-PGI is active as a dimer of two monomers of ϳ66 kDa each (12)(13)(14)(15). However, in non-denaturing gels, PGI equilibrated at pH 7.5 before loading migrates at 270 Ϯ 14 kDa (Fig. 6A). In the presence of buffers at pH 6 and below, a band of 73 Ϯ 4 kDa is detected corresponding to the monomeric form of the protein.
While this band represents only a minor form at pH 6, increasing amounts of this band are detected at pH 5.5 and pH 5 such that at pH 5.0, only the monomeric form of the protein is present (Fig. 6A). The extent of formation of the monomeric  3. PGI also binds to osteoarthritic chondrocytes at acid pH. Primary cultures of chondrocytes obtained from cartilage of a patient with osteoarthritis were incubated with 25 g/ml Alexa 568-PGI for 30 min in bicarbonate-free medium containing 100 mM HEPES at pH 7.5 or 100 mM MES at pH 5.5 or 5.0, as indicated. All images were acquired with the same confocal settings, and a similar dramatic increase in cell binding was observed at pH 5 for cells obtained from four different patients.
Cross-linking of PGI with glutaraldehyde for different times prior to SDS-PAGE (56) was used to further characterize the high molecular weight form of PGI (Fig. 6B). In the absence of cross-linking, PGI was detected exclusively as a monomer of 60 Ϯ 5 kDa but with increasing time of cross-linking distinct protein bands appeared progressively at 124 Ϯ 7 kDa, corresponding to PGI dimers, and at 242 Ϯ 9 kDa (Fig. 6B). Taking into consideration differences in migration in non-denaturing gels and SDS-PAGE, the latter band is equivalent to the high molecular weight 270-kDa band observed at neutral pH by non-denaturing gel electrophoresis (Fig. 6A). A semi-log plot of the migration of the 3 bands detected in the cross-linking experiment confirmed that they correspond to PGI monomer, dimer and tetramer (Fig. 6C). The transition from dimer to tetramer observed in the cross-linking experiments suggests that the tetrameric form of PGI reported here corresponds to the stable interaction of two dimers.
To determine whether PGI monomerization is required for its interaction with cell surface fibrils at acid pH, the glutaraldehyde cross-linked form of the protein was added to cells at pH 7.5, 5.5, or 5. Alexa 568-PGI was either not incubated with glutaraldehyde, incubated with glutaraldehyde and immediately quenched by incubation with glycine, or cross-linked with glutaraldehyde for 10 or 30 min prior to quenching. As reported for native PGI (Fig. 6B), 10 min of cross-linking results in the predominant expression of dimers of Alexa 568-conjugated PGI while at 30 min no monomer is detected and the predominant form detected is PGI tetramers. The cross-linked forms of PGI were still endocytosed to multivesicular bodies at neutral pH ( Fig. 7). At pH 5, PGI binding to cell surface fibrils was detected under all cross-linking conditions although reduced labeling was observed for PGI cross-linked predominantly in its dimeric form for 10 min. The cell surface association of the various cross-linked forms of PGI was quantified and, as observed for native PGI, at all times of cross-linking PGI still exhibited increased cell surface binding at pH 5 relative to pH 5.5 (Fig. 8). In particular, binding at 30 min of cross-linking was equivalent to that observed for native PGI indicating that dissociation of PGI monomers is not required for fibril association.
The fact that cross-linked PGI was still able to bind to cell surface fibrils at pH 5 indicated that changes in quaternary structure of the protein were not responsible for the enhanced binding. We therefore analyzed PGI structure by circular dichroism (CD). CD spectra of rabbit PGI at pH 7.5 ( Fig. 9) correspond to those previously reported for human PGI (59) and essentially identical spectra were observed at pH 6 and 5.5 in both the far and near UV ranges (not shown). At pH 5, the spectra obtained in the far UV was equivalent to that obtained at neutral pH indicating that pH does induce changes in the secondary structure of PGI (Fig. 9, left panel). However, in the near UV range (Fig. 9, right panel), while the negative band at 296 was not affected, the negative band at 286 and the positive band at 256 were absent at pH 5. Acidification to pH 5 is therefore associated with specific effects on the tertiary folding of PGI.

Conformation-dependent Association of PGI with Fibronectin
Fibrils-The ability of PGI to associate with fibronectin at acid pH adds further complexity to the behavior of this multifunctional protein. We have reported the association of PGI and of FIG. 4. Acid pH induces the direct binding of PGI to fibronectin fibrils. NIH-3T3 cells were incubated with 25 g/ml Alexa 568-PGI in regular medium (control) or in bicarbonate-free medium supplemented with 100 mM MES, pH 5.0 for 5 or 30 min, as indicated, at 37°C. After paraformaldehyde fixation and Triton X-100 permeabilization the cells were labeled with mouse anti-fibronectin antibodies and Alexa488-conjugated antimouse secondary antibodies. PGI distribution and fibronectin labeling were visualized by confocal microscopy and merged images present fibronectin in green, PGI in red, and colocalization in yellow.
its receptor, AMF-R, with cell surface fibronectin fibrils following endocytosis and passage through multivesicular bodies (40). This study confirms this association and further suggests that it is facilitated by PGI denaturation at acid pH. PGI isomerase activity is pH-dependent (15) and, similarly, we detected a 70% reduction in isomerase activity at pH 6.5 and complete inhibition of activity at pH 5.0 (data not shown). Inhibition of the majority of PGI isomerase activity at pH 6.5 is not associated with detectable monomer formation or increased fibril deposition of PGI (see Figs. 1, 2, and 6). pH induced conformational changes able to inhibit PGI isomerase activity do not therefore result in fibril association of the enzyme. Rather, the dramatic increase in fibronectin binding observed at pH 5.0 is a consequence of more drastic conformational changes able to disrupt monomer-monomer interactions. The fact that glutaraldehyde cross-linking does not abrogate fibril binding indicates that this association does not require monomer dissociation per se (Figs. 7 and 8). pH-dependent changes in the near UV were detected by CD confirming that changes in PGI conformation are induced by acid. Acid-induced changes in PGI tertiary structure, most specifically at pH 5, therefore result in conformational changes that are associated with both the disruption of monomer-monomer interactions in the native protein and exposure of protein domains that enable binding to fibronectin fibrils.
Earlier studies using gel chromatography defined PGI as a dimer and showed clearly that dimerization is essential for the isomerase activity of the enzyme (12-14). Detection of a higher molecular weight form corresponding to tetramers may reflect the presence of lower affinity interaction between PGI dimers FIG. 5. Binding of PGI to fibronectin fibrils at acid pH is not due to loss of cell integrity and is not saturable. NIH-3T3 cells were incubated for 30 min at 37°C with 25 g/ml Alexa568-PGI in bicarbonate-free medium supplemented with either 100 mM MES, pH 5.0 (A, B, E, F) or 100 mM HEPES pH 7.5 (C and D). In B, incubation at pH 5 was followed by a subsequent 30 min incubation in bicarbonatefree medium supplemented with 100 mM HEPES pH 7.5 that resulted in the endocytosis of cell-associated PGI demonstrating the viability of the cells. In D and F, incubation with Alexa 568-AMF was performed in the presence of 2 mg/ml unlabeled PGI. While excess PGI competed efficiently for PGI endocytosis at neutral pH (C and D) it did not prevent PGI fibril association at acid pH (E and F). The cells were fixed with paraformaldehyde and PGI distribution visualized by confocal microscopy.
FIG. 6. Monomerization of PGI at acid pH. A, 1.5 g of PGI was supplemented with 250 mM HEPES at pH 7.5 and 6.5 or 250 mM MES at pH 6, 5.5, and 5 prior to loading on an 8% non-denaturing polyacrylamide gel. Silver staining reveals a high molecular weight (ϳ270 kDa) form of PGI at neutral pH and monomeric PGI (ϳ70 kDa) appears progressively at pH 6.0, 5.5, and 5.0. B, 0.5 mg/ml PGI was cross-linked with 0.1% glutaraldehyde for 0, 1, 5, 10, 20, 30, and 60 min, as indicated, and then analyzed by reducing SDS-polyacrylamide (8%) gel electrophoresis. Coomassie Blue staining reveals the progressive transition of monomeric (M) PGI to dimeric (D) and then to tetrameric (T) forms of the protein with increasing times of cross-linking. C, semilogarithmic plot of the migration, measured from the top of the gel, of the cross-linked bands normalized relative to the monomer band which was assigned a value of one against the predicted number of subunits for monomer, dimer, and tetramer. The presented values represent the average of six independent experiments (r ϭ 0.9995 Ϯ 0.0005 S.D., n ϭ 6). that are stabilized in non-denaturing gel electrophoresis but not following dilution in gel chromatography. Furthermore, while x-ray crystallographic studies have clearly identified PGI as a dimer (17,19,60), these studies do not exclude the possibility of dimer-dimer interactions. The functional significance of the existence of PGI tetramers remains to be determined; however, the fact that PGI cross-linked as a tetramer is internalized to MVBs at neutral pH indicates that this oligomeric form of the protein is able to interact with its receptor (Fig. 7). Furthermore, dimer-dimer interactions could contribute to the formation of a glycolytic matrix (61).
It is possible that denaturation of some PGI in acidic endosomes (pH 6) may be implicated in the association of recycled PGI with cell surface fibrils. However, denatured PGI monomers easily reform functional dimers (14) such that the extent to which PGI would remain monomeric upon return to the plasma membrane is not clear. Both endogenous cell surface PGI and its receptor, AMF-R, were observed to be associated with fibronectin fibrils and both endocytosis and fibril association of PGI were inhibited by an excess of unlabeled PGI (40). The non-saturable nature of PGI association with fibrils at pH 5 indicates that it is mediated by a different mechanism than at neutral pH that involves a direct association with either fibronectin or other components of extracellular matrix fibrils. It cannot be excluded that the altered conformation of fibronectin and other extracellular matrix components at acid pH facilitates the direct, non-receptor mediated association of PGI. The fact that stabilization of PGI predominantly in its dimeric form after 10 but not 30 min cross-linking was associated with a reduction in cell surface binding at pH 5 (Figs. 7 and 8) further supports a role for PGI conformation in its binding to the extracellular matrix at acid pH. The molecular basis for the interaction between PGI and fibronectin and potentially other extracellular matrix components at both neutral and acidic pH remains to be determined.
Implications for the Role of PGI in Rheumatoid Arthritis-Binding of monomeric PGI directly and in large amounts to cell surface extracellular matrix fibrils at acid pH provides a possible explanation for the postulated role of PGI in RA. The ability of monomeric PGI to associate with the cell surface is consistent with the fact that the immune reaction in autoimmune arthritis is associated with surface-bound antigen and not soluble circulating PGI immune complexes (46,48). It further provides an explanation for the generation of a localized autoimmune response in lymph nodes draining the affected joints (52). Stable association of distinct conformational forms of PGI with the joint surface would result in exposure of anti-FIG. 7. Cross-linked PGI still binds to cell surface fibrils at acid pH. Alexa 568-PGI was left uncross-linked (No glut), incubated with glutaraldehyde and immediately quenched with glycine (ϩ glut 0 min), or cross-linked for 10 (ϩ glut 10 min) or 30 min (ϩ glut 30 min) before quenching (see "Experimental Procedures") The variously crosslinked forms of Alexa 568-PGI were then incubated at 25 g/ml at either pH 7.5 or pH 5 for 30 min before visualization by confocal microscopy. The images at pH 5 for cells under the various conditions were all acquired with the same confocal settings. gens different from the native form of the enzyme capable of generating an immune response in the absence of sequence differences. Interaction of autoantibodies with cell surface associated PGI could lead to the destruction of the cells through direct or indirect antibody mediated cytotoxicity. Indeed, differences between the ability of RA antisera to recognize recombinant and commercial forms of the enzyme (45,51,62) may be due to the presence of denatured PGI in the commercial preparations.
PGI cytokine activity has been reported previously in rheumatoid synovial fluid (63). PGI expression is up-regulated during osteoblast mineralization (21) and increased in hypoxia (64) such that PGI expression in synovial fluid may be due to secretion by chondrocytes. Cartilage surface pH was found to be highly acidic, as low as pH 5.5, in patients with osteoarthritis (65). EGF accumulates in the synovial fluid of rheumatoid arthritis patients and has been shown to stimulate protein efflux from chondrocyte cell cultures generating an acidic microenvironment (66). PGI expression in synovial fluid may also be generated during the RA-associated inflammatory response by cells of the immune system (4,6). Chronic inflammation is associated with acidosis (67) and synovial fluid has been reported to be acidic, as low as pH 6.0, particularly in rheumatoid patients (68 -70). We do detect monomerization and fibril deposition of PGI at pH 6.0 indicating that the partial denaturation of PGI at mildly acidic pH in synovial fluid could result in its continual deposition in small amounts on synovial cell surfaces. In addition, fibronectin fragmentation is associated with inflammatory arthritis (71) and may result in the exposure of distinct motifs susceptible to binding of PGI.
The demonstration here that PGI is able to bind to the surface of first passage osteoarthritic chondrocytes at acid pH ( Fig. 3) suggests that reduced pH associated with cartilage degeneration could lead to the select deposition of monomeric or denatured PGI at cartilage cell surfaces and not other endothelial cell surfaces. Denaturation of PGI at acid pH may be implicated in either the initiation or aggravation of a localized inflammatory response at articular surfaces and provides a possible explanation for the expression of this multifunctional protein in rheumatoid arthritis.