Identification and purification from the plasma of Type 1 diabetic subjects of a proteolytically active Grp94
- 523 Downloads
The overall increase in proteolytic activity in diabetes is known to be associated with the development and progression of vascular complications. Our aim was to investigate in detail the molecular nature of this activity in the plasma of Type 1 diabetic subjects.
Plasma of both diabetic and control subjects was subjected to various purification procedures (ion exchange and affinity chromatography, HPLC, immunoprecipitation, electrophoresis, immunoblot and mass analyses) to identify the proteins of interest. Biological activities were measured on specific substrates.
In diabetic but not normal plasma we identified the presence of two heat shock proteins, Grp94 (Glucose-regulated protein94) and HSP70. The higher-than-normal proteolytic activity of Grp94 was: (i) directed against casein, but not against endogenous plasma proteins; (ii) fully and specifically inhibited only by anti-Grp94 polyclonal antibodies; and (iii) coupled with low-level ATPase activity. In addition, ATP binding to Grp94 was able to modulate proteolytic activity. We found that Grp94 in plasma circulates only as high molecular mass homo- and hetero-complexes, the latter mostly formed with IgG to which Grp94 is also linked by tenacious binding. Proteolytically-active Grp94 was purified by immunoprecipitation, which co-immunoprecipitated α1antitrypsin.
Our results show the unexpected extracellular location and characteristic biological function of Grp94 even at a late stage of disease. These findings have physiopathological relevance for predicting activation of both autoimmune and inflammatory processes potentially associated with vascular complications.
KeywordsDiabetes mellitus, insulin-dependent heat shock proteins peptide hydrolase blood proteins
Heat shock proteins
One of the most troublesome aspects of Type 1 (insulin-dependent) diabetes mellitus are early and severe complications, especially those affecting the vascular bed , which are the main causes of increased morbidity and mortality in the diabetic population. Although the role of hyperglycaemia in worsening diabetic complications is well established , it is unlikely that chronic hyperglycaemia is the only etiopathogenetic factor [3, 4, 5]. Genetic factors also affect the course of complications produced by long-lasting hyperglycaemia: hyperglycaemia-induced oxidative modifications interplay in a complex way with genes [6, 7], and a predisposing genetic background could explain the irreversibility of some glucose-induced damage in susceptible patients . Markers for vascular complications, often apparently independent of glycaemic control, are alterations in proteolytic enzyme action [9, 10, 11, 12] and of their inhibitors [9, 13, 14]. Such changes include unbalanced hyperactivation of plasma-clotting enzymes , but also an overall increase in proteolytic activity in various tissues and biological fluids [10, 12]. Thus a higher-than-normal proteolytic activity of plasma has been documented in Type 1 diabetic subjects related to an anomalous form of α1antitrypsin (α1AT) , the most important circulating inhibitor of serine proteases. Alterations of the concentration and function of α1AT occur in Type 1 diabetic subjects early in the course of disease and even in the absence of any complications . Since various enzymes could concur differently in determining the overall increase in proteolytic activity of diabetic plasma, we wanted to investigate this issue in detail, by identifying the molecular nature of the substance(s) responsible for this activity. We present the results of an exhaustive investigation carried out on pooled plasma of Type 1 diabetic subjects.
Subjects and methods
Plasma was collected from the heparinized blood of ten Type 1 diabetic subjects (6 women, 4 men), and from six control subjects matched for age and sex . All subjects gave their informed consent for the sampling of blood. Diabetic subjects (mean age ± SD of 38.2±10.9 years, range 20–53) had a mean (± SD) duration of diabetes of 12.6 (±7.7) years, and a mean (± SD) HbA1c value of 8.48% (±1.36). Control subjects, recruited from personnel and students attending our laboratories, had a mean age (± SD) of 35.7±6.7 years. All subjects were non-smokers, they did not take any medication (other than insulin for diabetic subjects), nor did they show any clinical or biohumoral evidence of vascular complications.
Pooled plasma was dialysed overnight on Spectrapor membrane tubing of 15,000 MWCO (A.H. Thomas, Philadelphia, Pa., USA) against two 50-fold volume changes of buffer containing Tris-HCl 20 mmol/l (pH 7.4). After dialysis, plasma was passed through a DEAE Sephadex A-25 column (Pharmacia LKB, Uppsala, Sweden) equilibrated with the same buffer used for dialysis. Elution of proteins from the column was by a stepwise gradient (constant flow rate of 40 ml/h) with the equilibration buffer solution without and with 1 mol/l NaCl. Void volume and 1 mol/l NaCl fractions were submitted to ultrafiltration (Centriflo, 25,000 membrane MWCO, Amicon, Danvers, Mass., USA). The NaCl fraction from the DEAE column was further purified through a Con-A Sepharose 4B column (Amersham Pharmacia Biotech, Uppsala, Sweden), by loading 3.72 mg proteins. Proteins were eluted with the buffer solution (Tris-HCl 20 mmol/l, NaCl 0.5 M, pH 7.5), without (void volume) and with the addition of 0.5 mol/l α-D-methylmannoside (to remove Con-A-bound proteins). All chromatographic steps were performed on a fast protein liquid chromatography system (Pharmacia LKB, Uppsala, Sweden).
The NaCl fraction from the DEAE column was further analysed by reverse-phase HPLC on a Vydac C4 column (Vydac, Hesperia, Calif., USA) with three linear gradient steps of acetonitrile (with addition of 0.05% TFA): 0–40% (in 10 min), 40–60% (in 15 min) and 60–90% (in 10 min), monitoring the eluate at 220 nm.
Electrophoresis and immunoblotting
Electrophoresis was run on 8.5% and 10% polyacrylamide gels, in denaturing (SDS) and non-denaturing conditions, loading 5–10 µg sample proteins, in the absence and presence of reducing treatment. Gels were stained with Coomassie brilliant blue. For immunoblot analysis, proteins (3–5 µg) were transferred overnight into a PVDF membrane (Immobilon, Millipore, Bedford, Mass., USA) and treated with the following primary antibodies in 20 ml of TBS solution: anti-human HSA, α1AT, prothrombin, coagulation Factor X, IgG (all sheep polyclonal antibodies, The Binding Site, Birmingham, UK); HSP70 (rabbit polyclonal), HSP90 (rabbit polyclonal) and Grp94 (rabbit polyclonal, SPA 851, and rat monoclonal, SPA 850) (StressGen Biotechnologies, Victoria, B.C., Canada). Both alkaline phosphatase and horseradish peroxidase conjugated affinity purified IgG were used as secondary antibodies. Negative controls were made with secondary antibodies alone, to confirm the specificity of the immunostaining.
We incubated α casein (Sigma Chemicals, St. Louis, Mo., USA) at a final concentration of 1.1 mg/ml in Tris-HCl 20 mmol/l (pH 7.4) with 0.37 mg/ml sample proteins at 37°C for 60 min. Aliquots of samples were drawn at intervals of 10 min and added to a sample buffer mixture containing SDS. Digestion patterns were followed by PAGE  at 12.5% polyacrylamide gel. In experiments for testing the inhibition of proteolytic activity, we preincubated samples with protease inhibitors (α1AT, antithrombin III, α2-macroglobulin, soybean trypsin inhibitors, Sigma Chemicals), ATP (Sigma Chemicals) and anti-human antibodies (anti: HSA, α1AT, IgG, HSP70, Grp94) for 30 min at 37°C, before adding casein. The reaction was then followed for 60 min, as indicated above. To test the effects of temperature on proteolytic activity, we heated samples at 39, 41, 43 and 45°C for 30 min before following the reaction in the presence of casein, again at 37°C. For quantitative analyses of proteolytic activity, the density of the casein band (32 kDa), expressed in arbitrary units of its area, was compared, at any time of incubation, with that of its own control (undigested casein, taken as 100%). No variation in the control band of undigested casein was noted at any of the indicated incubation times at 37°C.
We measured the trypsin-like activity of DEAE fractions (3–500 µg/ml) on Nα-benzoyl-DL-arginine-p-nitroanilide hydrochloride (BAPNA) substrate, as described . Plasmin-specific substrate Chromozym PL (Roche Diagnostics SpA, Basel, Switzerland) was used to test plasmin-like proteases. The incubation medium, at the final volume of 1.3 ml, included the buffer solution (50 mmol/l Tris-HCL with 0.1 mol/l NaCl, pH 8.2) with 100 mmol/l glycine and 0.2% Tween 20 (Merck, München, Germany), Chromozym PL (0.136 mg/ml) and sample proteins (3–30 µg/ml). The reaction was followed at room temperature for 3 min by monitoring the increase in absorbance at 405 nm.
This activity was measured on p-nitrophenyl acetate substrate, as described [20, 21], by monitoring the formation of p-nitrophenol at 400 nm. Standard curves were made with non-glycated and glycated HSA (8.56 µmol/l, final concentration). Glycated HSA was from Sigma Chemicals. When activity was measured in the presence of anti-HSA and anti-Grp94 antibodies, we preincubated sample proteins with antibodies (antibody-to-sample weight concentration ratios of 1:10 and 1:100) at 37°C for 30 min before following the reaction by adding the substrate. Non-linear regression analysis (one-phase exponential decay equation) was used for graphing data and calculating statistical parameters (Graph Pad Prism, version 2.0, San Diego, Calif. USA).
This activity was measured by adding 3, 6 and 12 µg proteins of both plasma and DEAE fractions to reaction medium containing, in 30 µl, and at final concentrations: 2.5 mmol/l Tris-HCl (pH 7.5), 2 mmol/l MgCl2, 2.5 mmol/l NaCl, 0.02 mmol/l ATP and 25 µmol/l [32P]ATP (2000 cpm/pmoles). After incubation at 37°C for 2.5, 5, 10, 20 and 30 min, the reaction was terminated by adding 1.5 ml TCA (10%). After centrifugation, the 32Pi in the supernatant was converted to phosphomolybdic complex, extracted with 2.5 ml of isobutanol/toluol (1:1) and evaluated by liquid scintillation .
Pre-clearing of the NaCl fraction from the DEAE column was carried out by incubating 1.0 ml (1 mg proteins) with 75 µl of Protein A-agarose CL-4B (Sigma Chemicals) for 2.0 h at 4°C, followed by centrifugation at 1500×g for 10 min. The supernatant was incubated overnight, under slow stirring, with 12.5 µg of anti-Grp94 antibodies (rabbit polyclonal IgG, from StressGen). Immunocomplexes were pelletted by incubation with 75 µl of Protein A-agarose CL-4B (2.0 h at 4°C under slow stirring) and centrifugated. Pellet was washed repeatedly in TBS solution, electrophoresed on SDS-PAGE (10% gel) with reducing treatment, transferred to a PVDF membrane (Millipore, Bedford, Mass., USA), and probed with antibodies, as specified in the legend to relative figures.
Automatic Edman degradation of sample proteins was followed on an Applied Biosystems (Foster City, Calif., USA) sequencer (477A) equipped with an on-line 120A phenylthiohydantoin amino acid analyser. Protein sequence information was obtained from the National Center for Biotechnology Information (NBCI, Bethesda, Maryland) searching by protein sequence alignments (http://www.ncbi.nlm.nih.gov/).
Data from matrix-assisted desorption ionization time-of-flight (MALDI-TOF) MS of sample solutions were acquired using a TOF Spec-2E spectrometer (Micromass, Manchester, UK) at the Glaxo-Smith Kline Laboratories in Verona (Italy). All spectra were acquired in linear mode using 25 kV acceleration voltage.
Proteolytic activity of diabetic plasma is not explained by serine proteases
We also tested the possibility that glycation had conferred on HSA in new, so far undetected, proteolytic activity. We measured the esterase activity of both plasma and the DEAE fraction, and found that it was much higher in diabetic than in normal subjects (Fig. 1C). However, we verified that HSA glycated "in vitro" did not display any proteolytic activity on casein, while on its own it exhibited esterase activity overlapping that of the DEAE fraction of diabetic subjects. Thus, proteolytic and esterase activities in diabetic plasma are driven by different substances.
Proteolytic activity is coupled with ATPase activity
Since the protease activity of diabetic plasma has been closely associated with a higher-than-normal ATPase-stimulating property , we investigated whether diabetic plasma also has intrinsic, still undetected, ATPase activity. Using radioactive Pi measurement, low but constant ATPase activity was detected in diabetic, but not normal, plasma and DEAE fraction. Specific ATP hydrolysis (18.55 and 16.55 pmol 32Pi for plasma and the DEAE fraction, respectively) was only observed at a protein concentration of about 200 µg/ml, and after 5 min of incubation at 37°C (Fig. 1D). The finding that specific ATPase activity was dependent on neither protein concentration nor time of incubation strongly suggested that ATPase has a regulatory rather than transporting property. By submitting the same sample of the DEAE fraction to measurements of both ATPase and proteolytic activities, we noted that the increase in specific ATP hydrolysis was closely coupled with reduction in proteolytic activity. Accordingly, the addition of ATP (1 and 3 mmol/l, final concentrations) to the incubation assay for proteolytic activity caused a slight, concentration-dependent inhibition of activity of the diabetic but not normal, DEAE fraction, with maximum inhibition of 20% at 3 mmol/l ATP (Fig. 1B). These results suggested that proteolytic and ATPase activities are coupled.
Heat shock proteins circulate in the plasma of diabetic but not normal subjects
We further analysed the diabetic and normal DEAE fractions in reverse-phase HPLC (Fig. 2C) which, despite denaturation of proteins, proved useful in separating and concentrating HSPs. Immunoblot and mass analyses of HPLC peaks assigned to peak 1 of the diabetic DEAE fraction monomeric forms of HSA, structurally similar to HSA in peak 1 of the normal fraction. Both heavier (glycated) forms of HSA (67429 Da) and proteins with masses higher than 70,000 Da (which strongly immunoreacted with anti-Grp94 and anti-HSP70 antibodies), eluted in peak 1′ (Fig. 2C, right). Due to the presence of these HSPs, tryptic digestion of peak 1 together with peak 1′ yielded an electrophoretic pattern which, in number and mobility of the bands, differed from that obtained after digestion of peak 1 only. Grp94 began to elute distinctly in peak 1′ but concentrated in peak 2, where it focused in two sharp bands near the cathode (Fig. 2C, right). HSP90 was detected only in peak 1 and HSP70 in peak 1′ (Fig. 2C, right). Thus, Grp94 appeared to be the most hydrophobic of the HSPs. Interestingly, in both peaks 1′ and 2, the Grp94-positive bands also cross-reacted with anti-α1AT antibodies (Fig. 2C, right).
Grp94 entirely accounts for the proteolytic activity of diabetic plasma
Since electrophoretic and mass analyses indicated that HSPs were present as high molecular-mass complexes, which also cross-reacted with plasma α1AT, HSA (Fig. 2A) and IgG, we investigated whether such HSP-protein complexes were responsible for proteolytic activity. We thus tested the effect of specific antibodies on activity of the DEAE fraction. Anti-HSP70 antibodies did not affect proteolytic activity at all, whereas anti-α1AT antibodies caused an inhibition of 72% only at the highest concentration (1:1 ratio) (Fig. 1B). By using anti-HSA antibodies we showed a significant inhibition of 62% at the concentration ratio of 1:10 (Fig. 1B). An even more significant, concentration-dependent inhibitory effect was noted with anti-IgG antibodies which gave inhibitions of 97%, 95% and 78% at concentration ratios of 1:10, 1:25 and 1:100, respectively (Fig. 1B). However, it was only with anti-Grp94 antibodies that we observed 100% inhibition of proteolytic activity up to the concentration ratio of 1:200 (Fig. 1B). Inhibition progressively decreased as the concentration of antibodies was lowered, being 91%, 86%, 78% and 32% at ratios of 1:400, 1:800, 1:1000 and 1:2500, respectively. At 1:5000, activity was completely restored (Fig. 1B). The inhibitory effect was displayed only by polyclonal but not monoclonal anti-Grp94 antibodies.
We tested effectiveness of anti-Grp94 antibodies also in the assay for esterase activity: the concentration which completely inhibited protease activity (1:100 ratio) only inhibited esterase activity by 21%. Similarly, anti-HSA antibodies, at the same concentration which inhibited proteolytic activity by 62%, inhibited esterolytic activity only by 38% (Fig. 1C). Instead, in the normal DEAE fraction, anti-Grp94 antibodies were completely ineffective in reducing the proteolytic activity, whereas anti-α1AT and anti-IgG antibodies caused much less inhibition than that observed in the diabetic DEAE fraction (Fig. 1B). Anti-HSA antibodies were also completely ineffective in inhibiting the proteolytic activity of the normal DEAE fraction, whereas they were able to reduce esterolytic activity by 52% at the concentration ratio of 1:10 (Fig. 1C). These results further confirmed that proteolytic activity in diabetic and normal plasma is sustained by different substances.
Grp94 circulates mostly complexed with IgG
We showed a marked reduction in proteolytic activity, together with an increase in Grp94 aggregates, after heating both the DEAE fraction and the pre-clearing supernatant to 45°C. Specifically, activity began to decrease already at 39°C, and disappeared at 43°C. Instead, the normal DEAE fraction was still active after heating at 39°C, and activity disappeared completely only at 45°C. We tested whether, besides the exogenous substrate of casein, also the most represented and physiologically important plasma proteins (HSA and α1AT), were substrates for the proteolytic activity of Grp94. We thus incubated the DEAE fraction at various temperatures (25, 37, 39 and 40°C) over a period lasting up to 24 h and then we followed proteolytic degradation of the main bands of HSA and α1AT in the DEAE fraction by SDS-PAGE and immunoblotting. We did not find any modifications of electrophoretic patterns consistent with proteolysis of these proteins. Similarly, no autoproteolytic activity of Grp94 was observed at 37°C or after repeated freeze/thaw cycles of samples.
Grp94 binding and release during substrate digestion
We set up experiments to detect the role played by Grp94 in the proteolytic degradation of casein. For this purpose, the supernatant (from the pre-clearing step) was incubated, with and without casein, in both the absence and presence of those substances and conditions which were effective in inhibiting proteolytic activity, such as anti-HSA, anti-α1AT, anti-IgG and anti-Grp94 polyclonal antibodies, ATP, and heating at 43°C. Changes were followed by SDS-PAGE for Grp94-positive bands, together with those of the substrate.
Grp94 also immunoprecipitates complexed with α1AT
As conclusive proof that proteolytic activity was due to Grp94, we submitted the supernatant from pre-clearing step to immunoprecipitation with anti-Grp94 antibodies (clearing step) (Fig. 3C, right panel). The resulting supernatant was completely cleared of Grp94 and also lacked any proteolytic activity, whereas Grp94 was recovered in the pellet as a free monomer in reducing conditions (Fig. 3C, right panel). We noted that, besides the 85-kDa band, also the band at 140 kDa, which was clearly visible in the pre-clearing supernatant, disappeared in the electrophoretic pattern of the supernatant after immunoprecipitation (Fig. 3B, arrows). Since the latter band always immunoreacted with anti-α1AT but not anti-Grp94 antibodies (Fig. 2A), we investigated whether α1AT was complexed with Grp94. Anti-α1AT antibodies tested in the pellet of the clearing step gave specific immunostaining with α1AT in a band running at the expected mass of α1AT at about 55 kDa (Fig. 3C, right panel). This result clearly proved that proteolytically active Grp94 does circulate complexed with α1AT. Breakage of disulfide bridges by reducing treatment leads to the separation of the two proteins which can thus migrate at expected individual molecular masses (Fig. 3C, right panel).
Proteolytically active Grp94 appears to lack glycosylation
Anti-Grp94 antibodies inhibited proteolytic activity in both peaks, although 100% inhibition was observed at the antibody-to-peak protein concentration ratio of 1:50 in peak 1, but only at 1:25 in peak 2. At 1:100, antibodies still inhibited activity in peak 1 by 20% (after 40 min incubation with casein), but were ineffective in peak 2. With respect to the DEAE fraction, in peak 2 there was a marked reduction in sensitivity to anti-Grp94 antibodies, probably due to the increased concentration of active Grp94, as suggested by immunostaining of the 200-kDa band (Fig. 5A, right) or to a higher degree of purification after removal of interfering (more glycosylated?) proteins which concur to form aggregates with Grp94.
HSPs have recently attracted much attention because of their involvement in immunological processes [25, 26]. In particular, a crucial role has been attributed to both cytoplasmic HSP70 and, mostly, the ER-resident chaperone Grp94, in binding and delivering antigenic peptides to the MHC class I processing pathway [27, 28, 29, 30]. This mechanism, responsible for stimulation of both humoral and cellular immune responses [26, 27, 31], is common to a variety of inflammatory and autoimmune diseases [25, 30, 32], including diabetes [33, 34]. Abundant experimental evidence exists supporting the theoretical model that necrotic release of intracellular HSPs, especially Grp94, can deliver immunogenic signals to immune cells. However, any demonstration that such an immune regulatory mechanism actually occurs "in vivo", in response to pathological cell death, has not yet been offered. Our data support this mechanism, showing the "in vivo" extracellular location of HSPs, mostly Grp94, in the plasma of Type 1 diabetic but not normal subjects. The fact that such a result has been observed in diabetic subjects at a time when beta-cell destruction is considered to be complete (after a mean duration of diabetes of 12 years), strongly suggests that the mechanism(s) leading to extracellular liberation of HSPs are still operating, presumably involving cells other than pancreatic beta cells.
We offer several lines of evidence to show that this Grp94 has structural and functional properties, both shared with and distinct from those described for any so far purified preparations of HSPs, in particular Grp94. First and most important is the peculiar, intrinsic proteolytic activity which is displayed in an assay using casein as a generic protein substrate, and is detected as early as 10 min in the incubation assay at 37°C, the substrate being completely degraded within 40 min. These characteristics resemble more closely those recently described for some mammalian serine proteases [18, 35] rather than overlapping those described for purified Grp94 . However, the enzyme activity of plasma Grp94 is by far the most intense measured for any HSP in "in vitro" assays. Moreover, it also explains the entire proteolytic activity of diabetic plasma. Of particular interest was the result that activity is specifically and fully inhibited only by anti-Grp94 antibodies, as shown by the titration curve with scalar concentration of the antibody which, at concentrations as low as 0.37 µg/ml, still inhibited activity in the DEAE fraction by 80%. Depletion of Grp94 by immunoprecipitation abolished any proteolytic activity. This represents a cogent proof that Grp94 itself is responsible for proteolytic activity. Blockade by anti-Grp94 antibodies implies that antigenic sites on Grp94 either overlap or are close to the protease active site(s). The latter hypothesis fits better our results, given that only polyclonal anti-Grp94 antibodies (developed against a C-terminal synthetic peptide as the immunogen, from StressGen), but not monoclonal antibodies (produced against the entire molecule of human Grp94, from StressGen), turned out to be effective in blocking activity. Thus, monoclonal antibodies could still recognize plasma Grp94 by binding to its antigenic site(s) (as revealed by immunoblot analysis), but leave the protease active site free to bind the substrate. This would imply that the protease active site is located in C-terminus of Grp94. This possibility is further supported by results showing that proteolytic activity was greatly reduced or even abolished in heating conditions. It is known that heating increases the formation of Grp94 aggregates  and that domains involved in formation of both homo-dimers and hetero-complexes of Grp94 are located in its C-terminus [38, 39]. Thus, when aggregation tendency is enhanced, the protease active site is likely closed off or masked. An additional proof showing that the protease active site in Grp94 is located in its C-terminus, comes from results of experiments with ATP. ATPase activity of plasma Grp94 shows characteristics similar to those described for purified forms of Grp94 in "in vitro" assays [23, 24]. Indeed, it exceeds the spontaneous liberation of ATP by a small factor, and it does not vary with either protein concentration or time of incubation . We observed that ATP binding to Grp94 caused inhibition of proteolytic activity. It is known that the ATP binding site on Grp94 is located on its N-terminus, and that ATP binding to Grp94 causes conformational changes in the molecule which indirectly affects structure of the C-terminal region [24, 40]. Thus, the inhibition of activity that we observed with ATP, at variance with that caused by polyclonal antibodies, is the consequence of the indirect regulatory mechanism of ATP on the protease active site on Grp94. We also cannot exclude that anti-Grp94 polyclonal antibodies block proteolytic activity since on their own they recognize and bind structurally modified epitopes on Grp94 better than monoclonal antibodies. Grp94 in diabetic plasma shows a significant structural modification with respect to known purified preparations of intracellular Grp94, which binds to concanavalin A [41, 42]. In our experiments, Grp94 lacks any affinity for Con-A Sepharose column, which is consistent with either loss or masking of the high-mannose glycosylation motif.
Our results indicate that Grp94 in the plasma of diabetic subjects is exclusively present as high molecular mass homo-complexes and hetero-complexes. We were able to identify Grp94 as monomer in SDS-PAGE only after reducing treatment. Hetero-complexes of Grp94 are formed with IgG and, to a lesser extent, HSA. Plasma Grp94 is bound to IgG in a complex manner. Experiments of pre-clearing with Protein-A showed that, whereas part of Grp94 is only weakly bound to IgG, for most part it forms tenacious binding apparently giving rise to a 150-kDa fusion protein with the IgG heavy chain. Further studies are necessary to identify the nature and the precise immunological meaning of Grp94 binding to single heavy chain of IgG. In cell-culture experiments it has been shown that intracellular Grp94, together with other stress proteins, is often linked with unassembled Ig chains .
Therefore, Grp94 binding to IgG is the expression of a complex, generalised activation of immune processes. An autoimmune mechanism is likely responsible for the extracellular liberation of complexed forms of Grp94; this, in turn, further activates the immune system because of its immunogenic potential.
At variance with Grp94-IgG complexes, which have immunological relevance in diabetes, the complexes of Grp94 with HSA seem to be less specific. Grp94 binding to HSA is the consequence of both the extensive binding capacity of HSA and the chaperone property of Grp94. The fact that Grp94 almost exclusively circulates in plasma complexed with IgG and HSA accounts for the inhibition of proteolytic activity observed with anti-HSA, and mostly anti-IgG antibodies. However, this inhibition can be explained only in terms of an aspecific mechanism of steric hindrance of the protease active site in Grp94 once the antibody has bound to antigenic sites.
By experiments of immunoprecipitation we obtained purification of proteolytically active Grp94, which not only circulates as dimers but also, and more importantly, as complexed with α1AT. This complex has a mass of 140 kDa and appeared in SDS-PAGE as a band which immunoreacted only with anti-α1AT antibodies. The mass of 140 kDa would fit the 1:1 Grp94-to-α1AT complex, which is resolved in its two components by differential immunostaining of the immunoprecipitate. The physiopathological meaning of this complex is at present lacking, although one would anticipate a possible role in stimulating immune responses, similarly to that observed with Grp94-peptide complexes [30, 44].
In conclusion, the finding that Grp94 is markedly increased in plasma of Type 1 diabetic subjects and has a peculiar proteolytic activity, is of physiopathological relevance for predicting the activation of autoimmune and inflammatory reactions potentially involved in vascular damage. Our purification procedures carried out on pooled plasma may have masked important individual differences in Grp94 concentrations and properties, possibly related to other variables linked to the disease.
We would like to thank Prof. G.A. Danieli for his helpful discussion of the manuscript, Dr. P. Polverino De Laureto and Dr. E. Frare of Centro di Ricerca Interdipartimentale per le Biotecnologie Innovative (CRIBI, Padova) for N-terminal sequences.
This work was supported by MURST (Ministero dell'Università e della Ricerca Scientifica e Tecnologica) grants of ex-60% and by CNR (Targeted Project on Biotechnology).
- 3.Microalbuminuria Collaborative Study Group, UK (1995) Intensive therapy and progression to clinical albuminuria in patients with insulin dependent diabetes mellitus and microalbuminuria. Br Med J 311:973–977Google Scholar
- 4.Marre M (1999) Genetics and the prediction of complications in type 1 diabetes. Diabetes Care S2:B53–B58Google Scholar
- 9.Finotti P, Piccoli A, Carraro P (1992) Alteration of plasma proteinase-antiproteinase system in type 1 diabetic patients. Influence of sex and relationship with metabolic control. Diabets Res Clin Pract 18:35–42Google Scholar
- 12.Shinada M, Akdeniz A, Panagiotopoulos S, Jerums G, Bach LA (2000) Proteolysis of insulin-like growth factor-binding protein-3 is increased in urine from patients with diabetic nephropathy. J Clin Endocrinol 85:1163–1169Google Scholar
- 19.Erlanger BF, Kokowsky N, Cohen W (1961) The preparation and properties of two new chromogenic substrates of trypsin. Arch Biochem Biophys 95:271–278Google Scholar