Proteolytic components of serum IgG preparations

Abstract

Chemical catalysis, an effector mechanism utilized by fully assembled antibodies, can also be mediated by the isolated antibody subunits. Because trace amounts of free light chains (L chains) are present in IgG preparations, a detailed study was undertaken to identify the constituents responsible for the polyreactive proteolytic activity of IgG purified from human sera, determined as the extent of cleavage of the model peptide substrate Pro-Phe-Arg-methylcoumarinamide. Two proteolytic species with approximate mass of 50 kD and 150 kD were separated by repetitive gel filtration in a denaturing solvent (6 m guanidine hydrochloride). The activity of the renatured 50-kD fraction (in fluorescence units/μg protein) was more than 45-fold greater than of the 150-kD fraction. Both fractions lost the activity following immunoadsorption on immobilized anti-IgG antibody. Fab fragments prepared from the 150-kD IgG fraction retained the activity. Reducing and non-reducing SDS-electrophoresis suggested the 50-kD fraction isolated from the IgG preparations to be a mixture of heavy chain (H chain) monomers and disulphide bonded L chain dimers. Electrophoretically homogeneous monomers of 50-kD H chains and 25-kD L chains were prepared by gel filtration of reduced and alkylated IgG from seven human subjects. Each of the alkylated L chain preparations displayed the proteolytic activity. The activity in alkylated H chains was undetectable or only marginally greater than the background values. L chain dimers appear to be the major species responsible for the polyreactive proteolytic activity of serum IgG preparations, with a smaller contribution furnished by tetrameric IgG.

INTRODUCTION

In addition to binding polypeptide antigens, certain antibodies can catalyse the cleavage of peptide bonds. Human and mouse antibodies from autoimmune disease subjects display an increased tendency toward antigen-specific catalysis [1–3]. Moreover, polyclonal IgG preparations express ‘polyreactive’ proteolytic activity, i.e. the ability to recognize model peptide substrates of differing sequence with low affinity [4]. The biological function(s) of the proteolytic activity are yet to be delineated. Important roles for the activity can be conceived however, for the following reasons: reuse of the catalysts for interaction with multiple antigen molecules enhances the biological potency compared with stoichiometric, reversibly binding antibodies; the cleavage reaction alters the biological activities of the antigen permanently; and proteolytic antibodies might substitute for the well-known role of conventional proteases in antigen clearance, inflammatory reactions and antigen presentation to T cells.

Free light chain (L chain) and heavy chain (H chain) subunits of antibodies are present in the intracellular reducing environment of B cells as well as the blood and urine of healthy humans and patients with lymphoproliferative disorders [5,6]. Once separated from each other, the L chains and H chains bind antigens poorly [7,8]. For this reason, subunit assembly into the disulphide bonded quaternary structure of antibodies is assumed to be a prerequisite for expression of biological functions. Binding to the stable form of the antigen (the ground state) however is a poor predictor of the catalytic activity, because catalysis depends in part on the binding of the transition state (the high energy state formed en route to product generation), which is a structurally distinct form of the antigen. Several previous reports have described catalysis by antibody L chains [9–12]. Subunit catalysis can be a factor even when analysing antibody preparations that purportedly fully assembled, e.g. IgG, because the subunits and subunit oligomers can be generated by spontaneous disulphide exchange reactions [13–15].

Further analysis of the biological functions of antibody catalysts found in the blood of healthy humans and of the value of the catalysts as markers for disease are dependent, in part, on unambiguous identification of the active species. The goal of the present study was to identify the polyreactive proteolytic species present in human serum IgG preparations. L chain dimers, although present in the serum IgG at low levels, were responsible for most of the activity, tetrameric IgG was active at considerably lower levels, and the activity in the heavy chains was marginal or absent. These data suggest the L chains as the major means for expression of the proteolytic immune repertoire.

MATERIALS AND METHODS

Antibody purification

Sera from peripheral venous blood were fractionated on protein G-Sepharose (Amersham, Pharmacia Biotech, Inc., Piscataway, NJ) as described by Kalaga et al. [4]. The acid eluate was neutralized with 1 m Tris–HCl pH 9, made up to 6 m guanidine hydrochloride (GdnCl; Sigma Chemical Co., St Louis, MO) pH 6·5, and fractionated on a Superose-12 FPLC gel filtration column (Pharmacia) in this solvent. IgG fractions (0·5 mg) were reduced with mercaptoethanol, alkylated with iodoacetamide and chromatographed on the Superose-12 column in 6 m GdnCl pH 6·5, according to Sun et al. [8]. Baseline resolution of the reduced and alkylated H chain (about 50 kD) and L chain (25 kD) fractions under these conditions was evident. Prior to catalysis assays, the protein fractions were renatured by dialysis against 15 mm Tris–HCl, 100 mm glycine, 0·025% Tween-20, 0·02% sodium azide, pH 7·7, for 2 days with four buffer changes using a Gibco(Rockville, MD) multiwell dialysis device at 4°C. Fab fragments were prepared by papain digestion of IgG, inactivation of the papain with iodoacetamide and purification on immobilized protein A and Superose-12 columns [16]. Protein estimations were by the method of Lowry with bovine serum albumin (BSA; Sigma; RIA grade) as standard. Nominal mass values of 50 kD and 25 kD were used to compute the molar concentrations of H and L chains, respectively.

SDS-electrophoresis and immunoblotting

SDS-polyacrylamide gel electrophoresis in the absence or presence of 20 mm 2-mercaptoethanol (2-ME) was on Phast gradient gels (8–25%; Pharmacia). Immunoblotting was as in [16], using rabbit anti-human IgG (whole molecule; Sigma) diluted 1:200 as the primary antibody and goat anti-rabbit IgG conjugated to horseradish peroxidase (Cappel Research Products, Durham, NC) as the secondary antibody.

Catalysis assays

Antibody fractions were mixed with peptide-methylcoumarinamide (peptide-MCA) substrates (Sigma) in 60 μl 50 mm Tris–HCl, 100 mm glycine and 0·025% Tween-20, pH 7·7, in 96-well plates (MF W; Dynatech Labs, Chantilly, VA) and incubated at 37°C in a humidified chamber [4]. Hydrolysis was measured as the fluorescence of the leaving group (aminomethylcoumarin; λ_em 460 nm, λ_ex 370 nm) using a plate reader (Perkin-Elmer LS50 fluorimeter; Wellesley, MA). The fluorescence yield of standard aminomethylcoumarin (Peptides Int., Louisville, KY) measured under these conditions was 21·9 FU/μm per 60 μl. Background fluorescence measured in wells containing the substrate incubated in the diluent (generally about 10 FU) was subtracted from values observed in the presence of catalysts. Immunoadsorption of the samples was as in [16]. Briefly, the antibody fractions (160 μl) were incubated with goat anti-human IgG (H + l) immobilized on Sepharose (0·2 ml settled gel; Zymed Labs, South San Francisco, CA) for 3 h at 4°C. The supernatant was collected, the gel washed with 1 ml neutral pH buffer, the bound material eluted with 1 ml 0·1 m glycine pH 2·7, and the eluate neutralized with 1 m Tris base pH 9. Protease inhibitors (diisopropylfluorophosphate, iodoacetamide and pepstatin A), albumin (RIA grade) and calmodulin were from Sigma.

RESULTS

Characterization of the protein G-binding IgG fraction

Cleavage of the amide bond linking Arg and the coumarin moiety in Pro-Phe-Arg-MCA is a convenient surrogate for peptide bond hydrolysis [17]. Background hydrolysis of this substrate incubated in buffer is negligible, cleavage by the serum IgG fractions from unimmunized humans is detected readily, and the activity of the IgG fractions increases in a linear manner as a function of the reaction time and the concentration of the IgG [4].

In the present study, the catalytic activity of the IgG purified by chromatography on immobilized protein G was analysed after high performance gel filtration in a denaturing solvent (6 m GdnCl). The denaturing solvent was employed to disaggregate non-covalent complexes of L and H chains, which tend to associate non-covalently in non-denaturing solvents [18], as well as potential complexes of antibodies and adventitious protease contaminants. Following renaturation, the column fractions were analysed for Pro-Phe-Arg-MCA hydrolysing activity. Two major activity peaks with approximate molecular mass 150 kD and 50 kD were evident (Fig. 1), corresponding to the IgG optical density (OD) peak and its trailing shoulder, respectively.

Fig. 1.

Gel filtration profile of proteolytic activity of serum IgG fraction prepared by binding to immobilized protein G. Chromatography of the IgG fraction (0·9 mg) was on a Superose-12 column in 6 m guanidine hydrochloride (GdnCl) (0·25 ml/min). Elution positions of calibration proteins are shown by arrows. Proteolytic activity was determined by incubation of duplicate aliquots of the renatured column fractions (2 μl) with Pro-Phe-Arg-methylcoumarinamide (MCA) (500 μm).

The IgG fractions contained small amounts of antibody fragments, as shown by analysis of overloaded SDS-electrophoresis gels (Fig. 2). In addition to the major tetramer IgG band at 150 kD identified by silver staining of the gels, minor bands were evident at the positions of H-H dimer (120 kD), H-L dimer (80 kD), l-chain dimer and H-chain monomer (50 kD) and l-chain monomer (25 kD). All of the silver-stained bands were also stainable by anti-human IgG antibody. Certain silver and anti-IgG stainable bands displayed mobilities inconsistent with the mass of the individual subunits and aggregates thereof. These represent minor subpopulations of the IgG or its fragments with anomalous electrophoretic mobilities [19]. The specificity of the anti-IgG antibodies used for immunoblotting was evident from their inability to stain the calibration protein mixture (phosphorylase b, albumin, ovalbumin, carbonic anhydrase, trypsin inhibitor and α-lactalbumin).

Fig. 2.

Non-reducing SDS-polyacrylamide gel electrophoresis of overloaded IgG (3 μg) purified by binding to immobilized protein G. Shown are gels stained with silver (lane 1) and anti-human IgG (lane 2). Lane 3, Silver-stained calibration proteins.

Cleavage of Pro-Phe-Arg-MCA (400 μm) by albumin or calmodulin (1 μm) was not detected over 6 h. Under similar conditions, a similar concentration of the protein G-binding IgG permits readily detectable cleavage of the substrate, suggesting that the reaction does not reflect a non-specific phenomenon. The reaction kinetics for cleavage of Pro-Phe-Arg-MCA by IgG preparations have been reported previously [4] and suggest low-affinity recognition of the substrate (K_m in the high micromolar range) and V_max 3·4–8·8 μm/μm IgG/h (note that the catalysts are a subset of the antibodies present in the total IgG preparation; the cited V_max values presumably underestimate the true reaction rates). The cleavage of Pro-Phe-Arg-MCA by the protein G-binding IgG fraction from two human subjects (1·6 μm; codes 36 and 5807) was inhibited nearly completely (by ≥90%) in the presence of a serine protease inhibitor (0·1 mm diisopropylfluorophosphate), whereas the rates of cleavage in the presence of inhibitors of other classes of proteases (0·1 mm EDTA, 2 mm iodoacetamide and 50 µm pepstatin A) were within the range of experimental error (approx. 10%) of the control rate in the absence of inhibitors (0·18 μm/μm IgG/h; initial substrate concentration 60 μm). We concluded that the catalysts belong to the serine protease family, consistent with observations of a serine protease mechanism utilized by a peptidase MoAb described previously [20].

Proteolytic activity of L chain dimers

Repetitive gel filtration permitted resolution of the 50-kD species present in the IgG fraction. The third cycle of gel filtration yielded a proteolytic activity peak that was superimposable on the OD peak (Fig. 3), indicating a constant specific activity (i.e. proteolytic activity/unit mass of the protein) across the width of the peak. Non-reducing and reducing electrophoresis of this peak revealed, respectively, a single silver-stained protein band at 50 kD and two bands at 50 kD and 25 kD (Fig. 3, inset). Both bands were stained with anti-human IgG antibody (not shown). The catalytic activity of this preparation was essentially completely trapped by immobilized anti-human IgG and released by elution with a pH 2·7 buffer (Fig. 4). These observations indicate the 50-kD fraction to be a mixture of H chain monomers and L chain dimers.

Fig. 3.

Purification of the 50-kD proteolytic fraction by repetitive gel filtration in 6 m guanidine hydrochloride (GdnCl). Shown is the third cycle of chromatography of the 50-kD fraction from Fig. 1 (retention time 40–43 min). Flow rate, 0·5 ml/min; other conditions as in Fig. 1. Proteolytic activity was determined by incubation of duplicate aliquots of the renatured column fractions (10 μl) with Pro-Phe-Arg-methylcoumarinamide (MCA) (500 μm). Inset, silver-stained, non-reducing (lane 1) and reducing (lane 2) SDS-polyacrylamide gels of the peak fraction (retention time 20 min).

Fig. 4.

Binding of the proteolytic activities of IgG and L chain dimers by immobilized anti-IgG. The 50-kD and 150-kD peak fractions (160 μl containing 4 μg and 8 μg protein, respectively) from Fig. 1 were mixed with anti-human IgG (H + L chain) immobilized on Sepharose 4B. The unbound fraction represents the supernatant recovered from the reaction mixture pooled with the neutral pH wash (total volume 1·16 ml), and the bound fraction, the pH 2·7 eluate (1 ml). Aliquots (30 μl) of the unbound and bound fractions were assayed in duplicate for the ability to hydrolyse Pro-Phe-Arg-methylcoumarinamide (MCA) (2 mm) over 18 h.

Two experiments were done to distinguish between H chain monomers and L chain dimers as the source of the proteolytic activity. In the first experiment, electrophoretically homogeneous H and L chains were purified by gel filtration (in a denaturing solvent, 6 m GdnCl) of reduced and alkylated IgG from seven human donors. Because the sulfahydryl groups are alkylated, the L chains do not dimerize, permitting baseline separation of the L chains monomers (25-kD fraction) from the H chains (50 kD) [8]. Following renaturation by removal of the denaturant, each of the L chain preparations displayed readily detectable proteolytic activity, evident as the aminomethylcoumarin fluorescence greater than background fluorescence observed in buffer (10·1 ± 0·8 FU) (Fig. 5). In comparison, the fluorescence evident after incubation of the substrate was indistinguishable from the background value (i.e. within the mean +2 s.d. of the background value) (five H chain preparations) or only marginally greater than the background (two H chain preparations; codes 673 and 678). In the second experiment, the 50-kD mixture of H chains and L chains from Fig. 3 was reduced and alkylated as described [8] and subjected to a further cycle of gel filtration chromatography in 6 m GdnCl. This resulted in resolution of the 25-kD L chain monomers and 50-kD H chain monomers (retention times 20·8 min and 24·6 min, respectively), with elution of the proteolytic activity in the L chain monomer peak (specific activity after renaturation 67 FU/μg L chain per 16 h).

Fig. 5.

Proteolytic activity of reduced and alkylated L and H chains. The L and H chains were purified from reduced and alkylated IgG from seven humans (designated by three number codes on the abscissa) by gel filtration in 6 m guanidine hydrochloride (GdnCl). Following renaturation, the proteolytic activity of the subunits (0·5 μg protein) was determined in triplicate by incubation with Pro-Phe-Arg-methylcoumarinamide (MCA) (400 μm). Inset, silver-stained SDS-polyacrylamide gel (lane 1) and an anti-IgG stained immunoblot (lane 2) of an L chain preparation.

As was the case for the protein G-binding IgG fraction, disopropylfluorophosphate (0·1 mm) inhibited (by ≥90%) the cleavage of Pro-Phe-Arg-MCA by light chains refolded from the guanidine hydrochloride solution (0·5 μm light chains pooled from the donors shown in Fig. 5; rate without inhibitor 0·09 μm/μm light chain per hour; reaction conditions as in Fig. 5). EDTA (0·1 mm) and pepstatin A (50 μm), which inhibit metalloproteases and acid proteases, respectively, were without effect on the reaction. Because the catalytic light chains were prepared by alkylation of Cys residues with iodoacetamide, they are evidently not dependent on the cysteine protease mechanism. Preservation of the serine protease type of activity in the light chains provides assurance that the catalytic site is not formed artificially as a consequence of the denaturation and refolding manipulations.

Proteolytic activity of IgG

Previous observations [4] have suggested the proteolytic activity observed in the 150-kD fraction in Fig. 1 to be due to IgG antibodies. This was confirmed in the present study by the following observations: (i) repetitive gel filtration of the 150-kD fraction permitted purification of a single peak of proteolytic activity that co-eluted exactly with the 150-kD IgG peak (Fig. 6, top); (b) papain treatment of the 150-kD fraction resulted in conversion of the proteolytic activity to a 50-kD form, corresponding to Fab fragments (Fig. 6, bottom); and (c) the proteolytic activity of the 150-kD fraction was bound by immobilized anti-human IgG and released from the matrix by a low pH buffer (Fig. 4). The specific activity of the tetramer IgG fraction (79·9 FU/μg L chain per 16 h) was 45·3-fold lower than that of the 50-kD fraction shown in Fig. 3.

Fig. 6.

Proteolytic activity of tetramer IgG (top) and Fab fragments generated from IgG by papain digestion (bottom). The IgG was purified by repetitive gel filtration of the 150-kD fraction from Fig. 1 (retention time flow rate 0·5 ml/min 34–36 min) in 6 m guanidine hydrochloride (GdnCl), of which the third chromatography cycle is shown. Gel filtration of Fab fragments was in 50 mm Tris–HCl, 100 mm glycine, 0·025% Tween-20 and 0·15 m sodium chloride. Proteolytic activity was determined by incubation of duplicate aliquots of the column fractions (20 μl) with Pro-Phe-Arg-methylcoumarinamide (MCA) (500 μm). Inset top, silver-stained non-reducing (lane 1) and reducing (lane 2) SDS-polyacrylamide gels of the peak IgG fraction. Inset bottom, silver-stained non-reducing SDS-polyacrylamide gel of the peak Fab fraction of antibody subunits relative to intact antibodies.

DISCUSSION

Although present only at trace amounts, L chain dimers appear to be responsible for a major proportion of the proteolytic activity of serum IgG fractions purified by affinity chromatography on protein G. Tetrameric IgG molecules, which constitute the bulk of the proteins recovered from the protein G column, express proteolytic activity at considerably lower levels than the L chain dimers (compared on unit protein basis). Takagi et al. [11] have previously reported a peroxidase activity in the recombinant L chain of a MoAb raised by immunization with a porphyrin. Proteolytic activities have been observed in L chains isolated from various sources, including autoantibodies [9] and antibodies raised by experimental immunization with vasoactive active intestinal polypeptide [20,21] and the HIV protein gp41 [22]. Observations showing the presence of proteolytic activities in a recombinant V_L domain [12] and its loss by site-directed mutagenesis at amino acids in the complementarity-determining regions of an L chain [21] have suggested that interactions at the V domains are responsible for the activities.

The presence of small amounts of L chain dimers in IgG preparations probably reflects their generation from tetrameric IgG by a process of disulphide exchange. The occurrence of disulphide exchange reactions in IgG and other antibody classes has been demonstrated previously [13,14] and this appears to be the mechanism of spontaneous formation of catalytic L chains and H-L dimers from a monoclonal IgG preparation in vitro[15]. Less likely is the possibility that free L chains found in human serum [5,6] co-purify with the IgG, because the affinity reagent employed for the purification (protein G) binds well-defined regions of the H chain (the Cγ2–Cγ3 interface and the Cγ1 domain [23]), without recognized similarity to L chains.

Cleavage of the model substrate Pro-Phe-Arg-MCA by polyclonal antibody preparations has been held to detect a polyreactive proteolytic activity [4], analogous to the low affinity binding of structurally diverse antigens by the so-called ‘natural’ antibodies found in healthy individuals [24]. Peptides containing basic residues but otherwise unrelated in sequence to Pro-Phe-Arg-MCA can also serve as substrates, with the cleavage occurring on the C-terminal side of the basic residue. Every L chain preparation examined in the present study displayed the proteolytic activity, suggesting that the activity is a common trait. A model V_L domain reverted to its germ-line sequence by mutagenesis was recently observed to display proteolytic activity [25], supporting the hypothesis that the catalytic activity is an expression of innate immunity. Free L chains are found physiologically in the blood and urine of healthy subjects, and in the reducing intracellular environment within B cells [5,6]. Consistent findings of proteolysis by L chains encourage the hypothesis of a physiological role for the proteolytic activity, e.g. in clearance of autoantigens from the blood or in production of antigen fragments utilized in various T cell antigen presentation pathways.

H chains were purified from reduced and alkylated IgG in the present study by a method involving unfolding and refolding of the proteins. The H chain preparations displayed no detectable activity (5/7 preparations) or weak proteolytic activity (2/7 preparations). Separation of H chains from the L chains was done by a gel filtration procedure permitting baseline resolution of the two types of proteins [8]. In the event of incomplete alkylation of the reduced subunits however, it remains possible that small amounts of 50-kD L chain dimers are formed upon removal of the reductant. Because the L chain dimers co-elute with the H chains from the column, we can not exclude L chain contamination as a cause of the weak catalytic activity of the H chain preparations. Additional pitfalls are: (i) the unfolding and refolding of H chains in vitro may generate an unnatural conformation(s) of the protein responsible for the weak enzymatic activity; and (ii) the weak activity of the H chain may reflect a non-specific catalytic capability, just as off-the-shelf proteins like albumin can serve as weak catalysts for certain reactions [26]. On the other hand, it is appropriate to leave open the possibility that native H chains might express proteolytic activity under certain circumstances, because the immune system possesses an enormous repertoire of different H chain sequences, some of which might encode a catalytic site.

The search for antibody catalysts has been pursued with considerable vigour by several research groups [27], but has focused until now on fully assembled antibodies expressed in autoimmune and experimentally induced immunological responses. Thus, screening for catalysts following immunization with unactivated haptens [28,29], a peptide [30], an enzyme [31], antibodies to enzymes [32,33], and analogues of the transition state of various small substrates [34,35] has generally been carried out by methods designed to detect assembled antibodies. Enthusiasm for the available catalytic antibodies has been diminished somewhat because of their modest catalytic efficiencies. In view of the superior proteolytic activity of L chains compared with tetrameric IgG, the isolation of potent proteases may be facilitated by screening of L chain repertoires, exemplified by the isolation of efficient VIP cleaving human L chains from a phage display library [36].