A Human Germ Line Antibody Light Chain with Hydrolytic Properties Associated with Multimerization Status

Abstract

Antibodies with nucleophilic or catalytic properties often have these characteristics encoded in their germ line genes. Because hydrolytic activity has been reported to be associated with light chain V regions, we have begun an analysis of germ line light chain proteins that could be the basis for affinity maturation into hydrolytic or other reactive antibodies. We produced the germ line A18b light chain and characterized its hydrolytic, nucleophilic, and tertiary structural activities. This light chain was purified to >99% purity and found to hydrolyze aminomethylcoumarin-peptide and larger protein substrates and bind a fluorophosphonate probe. Mutation of putative catalytic residues only resulted in loss of activity of a tetrameric but not dimeric form of the light chain. These biochemical properties provide a framework for understanding the structure-function relationships of germ line antibodies.

introduction

Antibodies are responsible for defense against invading pathogens, but they can also be pathogenic in autoimmune disease. Antibody genes are partially hard-coded in the genome as gene modules known as V, D, and J segments for the heavy chain, or V and J segments for the light chain (1). Multiple modules allow for the diversity of the antibody repertoire through combinatorial rearrangement. During B-cell development, the germ line-encoded V, D, and J segments recombine to form functional heavy- and light-chain V region genes. Following exposure to antigen, B-cells activate an affinity maturation process that mutates V regions, and those B-cells with higher affinity variants are driven to survive and proliferate. Typically, germ line-derived antibodies have affinities in the mid nanomolar to micromolar range (2, 3), as opposed to nanomolar or less for affinity-matured antibodies. Germ line-derived antibodies contain unique properties, including polyspecificity and catalytic activity. The polyspecificity is thought to be due to an inherent flexibility in germ line antibody complementarity-determining regions (CDRs),² which allows them to adopt multiple conformations and bind different antigens (4–6). The structural basis for catalytic activity is more obscure, however, structural studies of induced catalytic antibodies also show multiple structures in their germ line precursors (4, 7, 8).

Recently, catalytic antibodies have been associated with positive or negative aspects of human disease (9, 10). In sepsis, catalytic antibodies were predictive of survival (11). In subsets of hemophilia, proteolytic antibodies against Factor VIII were associated with resistance to therapy (12, 13). Catalytic antibodies against myelin basic protein in multiple sclerosis have also been found and suggested to contribute to pathogenesis (14, 15). In multiple myeloma, Bence-Jones light-chain proteins can contribute to pathogenesis of the disease and in some cases have been reported to contain catalytic activity (16, 17). The structural and biochemical details around such antibodies have not yet been studied in detail.

We describe here a human germ line light chain with homology to a known mouse proteolytic antibody, c23.5 (18, 19). We analyze the hydrolytic, nucleophilic, and other biochemical properties of this light chain. These properties provide further insights into the mechanisms by which these antibodies may act, which could be important in understanding the mechanisms by which these antibodies behave in medical settings, or during normal or aberrant immune responses.

EXPERIMENTAL PROCEDURES

DNA Constructs

A standard PCR reaction was used to amplify non-rearranged germ line V regions using 100 ng of non-lymphoid genomic DNA (Clontech) as template. The reaction consisted of 0.5 μm primers, 250 μm dNTPs, pfu buffer, and polymerase according to the manufacturer's instructions (Stratagene). A primer (tttctatgcggcccagccggccatggccGATATTGTGATGACCCAG), which hybridized to the 5′-end of the A18b coding region (in capital letters, an NcoI site for cloning is underlined) was used with a primer (GGAGGAAGGTGTATACC) that hybridized to the 3′-end of the germ line V region. Following 25 cycles of amplification (annealing at 60 °C for 30 s, extending at 70 °C for 30 s, and denaturing at 95 °C for 30 s), a 1-μl aliquot was used as a template for the recombination reaction. A primer (ACCACCACCGTACGTTTGATTTCCACCTTGGTCCC) consisting of part of the J_κ1 region with a Bsi WI restriction site (for cloning) at the 5′-end was added along with a joining primer (CAAGGTATACACCTTCCTCCGTGGACGTTCGGCCAAGGGACCAAGGTGGAAATCAAACG) comprising a 3′-end that hybridizes to the J region and a 5′-end that hybridizes to the 3′-end of the V region. A second round of extension using standard cycling conditions accomplished the V-J rearrangement. The rearranged variable region was cloned into a modified pCantab vector (Amersham Biosciences) at NcoI and BsiWI sites, which contained the human kappa constant region downstream of BsiWI. The BsiWI site allows in-frame cloning of kappa V regions with the kappa constant region without alteration of natural amino acids. The kappa constant region contains a C-terminal six histidine tag for immobilized metal affinity chromatography (IMAC) purification.

Protein Expression

Escherichia coli TOP10F′ (Invitrogen) harboring the light-chain expression vector was grown at 30 °C in Terrific broth supplemented with 0.5% glucose and 100 μg/ml carbenicillin and 10 μg/ml tetracycline. When the A₆₀₀ reached 0.5, cells were pelleted at 3000 × g, and resuspended in 1 liter of Terrific broth containing 0.1 mm isopropylthio-β-galactoside to induce light-chain expression. Cells were then grown overnight at 25 °C (∼8 h).

Purification

Cells were pelleted at 3000 × g then resuspended in periplasmic lysis buffer (50 mm Tris-Cl, pH 8.0, 20% sucrose, 1 mm EDTA, and 0.5 mg/ml lysozyme). The cells were lysed with gentle agitation for 2 h at 4 °C, followed by centrifugation at 10,000 × g and collection of the supernatant as the periplasmic lysate. The lysate (20 ml) was dialyzed against 1 liter of 100 mm sodium phosphate, pH 7.0, with three buffer changes. Batch IMAC purification was done by adding dialyzed periplasmic lysate to 3 ml of Talon cobalt resin in the presence of 20 mm imidazole (Clontech) in a 50-ml conical tube, rocking at 4 °C for 30 min, then centrifuging the resin at 2000 × g for 5 min. The supernatant was carefully removed, and then the resin was washed twice with 40 ml of 8 m urea in 100 mm phosphate buffer and twice with 40 ml of 100 mm sodium phosphate buffer. The resin was resuspended in 5 ml of sodium phosphate buffer and added to an empty chromatography column. The column was washed with sodium phosphate buffer containing 20 mm imidazole, and then bound protein was eluted with step fractions containing 100, 200, 300, 400, and 500 mm imidazole. Fractions containing pure light chain as analyzed by SDS-PAGE and silver staining were pooled.

For ion-exchange chromatography, IMAC-purified protein was dialyzed against 4 liters of 50 mm MES, pH 6.0, 10 mm NaCl with three buffer changes over 2 days. The sample was concentrated in an Amicon microconcentrator (Millipore) with a 10-kDa molecular mass cut-off. Protein was applied to 1 ml of a CM-Sepharose column (Amersham Biosciences or Sigma) and eluted with step gradients of 50, 100, 150, 200, 250, 300, 350, 400, 450, and 500 mm NaCl in 50 mm MES, pH 6.0. Fractions were analyzed by SDS-PAGE and silver staining, and fractions containing pure protein were combined. Size-exclusion chromatography was performed with an SEC-3000 preparatory size-exclusion high-performance liquid chromatography column on an HP 1090 instrument using Tris-buffered saline as a solvent. Molecular mass standards (Bio-Rad) were run immediately before the light-chain sample using identical run conditions. Peaks were collected as indicated in the figures and analyzed by SDS-PAGE and silver staining. Purified protein was concentrated using Amicon microconcentrators with a 10-kDa molecular mass cut-off. Glycine (100 mm) was added to the protein prior to concentration to avoid aggregation at high protein levels.

Peptide Hydrolysis Assays

Reactions were typically performed in 50 μl of 50 mm Tris-Cl, pH 7.8, 100 mm glycine, 0.025% Tween 20, and 400 μm Pro-Phe-Arg-methylcoumarinamide substrate (PFR-MCA, Peptides International). Fluorescence was read in low protein binding black 96-well plates (Costar) at excitation 370/emission 465 in a PerkinElmer HTS 7000 instrument. The amount of hydrolysis was determined by analyzing a titration of the aminomethylcoumarin leaving group and constructing a standard curve. For kinetic analysis, PFR-MCA substrate was varied from 4 to 512 μm at 2-fold dilutions, and fluorescence was read at three time points. Hydrolysis of substrate was determined by quantifying the amount of MCA leaving group present using a standard curve of aminomethylcoumarin. Kinetic parameters were determined by both Lineweaver-Burke and Eadie-Hofstee plots. For protease inhibitor studies, the following concentrations of inhibitors were used: pepstatin A, 10 μm; E-64, 100 μm; EDTA, 1 mm; leupeptin 1 mm; benzamidine, 1 mm; iodoacetamide, 1 mm; TPCK, 500 μm; TLCK, 500 μm; PMSF, 1 mm; DTT, 1 mm; and antithrombin III, 2 μg. These were based on standard concentrations used to inhibit their respective class of protease (20).

When protein was excised from SDS-PAGE gels (Fig. 3), the gel slice was crushed in 0.5 ml of 20 mm Tris-Cl, pH 7.8, 100 mm glycine, 500 mm NaCl and shaken overnight at 4 °C. Insoluble polyacrylamide was removed by microcentrifuging at 13,000 × g for 5 min. The supernatant was added to an equal volume of 8 m urea, 500 mm NaCl, and 20 mm Tris-Cl, pH 7.8, and incubated at 50 °C for 30 min. Protein was refolded by dilution in 10 ml of 50 mm Tris-Cl, pH 7.8, 100 mm glycine, then concentrated, and buffer was exchanged using Amicon microconcentrators as described above. The total volume was split in half and used in duplicate in a PFR-MCA hydrolysis assay.

FIGURE 3.

Activity of monomer versus dimer. A, an SDS-PAGE gel containing A18b was resolved in the absence or presence of DTT. The gel was stained with Coomassie (left) then bands were excised (right), which correspond to dimer (a), monomer (d), or unpaired monomer (b), and unreduced dimer (c). B, after gel extraction and refolding, the residual protein was incubated with PFR-MCA in duplicate, then analyzed for hydrolytic activity. C, monomer and dimer were separated by size-exclusion high-performance liquid chromatography, then resolved on SDS-PAGE in the absence (lanes 1 and 2) or presence (lanes 3 and 4) of DTT. D, the protein preparations containing dimer and monomer were incubated with PFR-MCA, and hydrolytic activity was measured.

Fluorophosphonate Probe Synthesis and Blots

The synthesis scheme for the fluorophosphonate probe is shown in supplemental Fig. S3. Undecylenic acid (1) was esterified with isobutyl alcohol under acid catalysis to yield the isobutyl ester (2) (95% yield, purified by distillation), which was then reduced with metallic sodium by the Boulevenault-Blanc procedure to give 10-Undecen-1-ol (3) (82% yield, purified by distillation). The alcohol was then iodinated by reaction with triphenylphosphine, iodine, and imidazole to yield 1-iodo-10 undecene (4) (90% yield, purified by distillation). Reaction of the iodide with triethyl phosphite under reflux afforded the diethoxy phosphonite (5) (87% yield, purified by distillation), which was then converted to the ethoxyhydroxy phosphonite (6) by treatment with trimethylsilyl bromide (56% yield). The double bond of (6) was oxidatively cleaved with ruthenium trichloride and sodium periodate to yield the terminal carboxylate (7) (81% yield, purified by crystallization of the sodium salt). Treatment of (7) with excess bis-(2-methoxyethyl)amino sulfur trifluoride yielded the difluoro compound (8, not isolated). The structures of compounds 1 through 5 were confirmed through GC/MS and Fourier transform IR. Compounds 5 and 6 were analyzed by high-performance liquid chromatography and Fourier transform IR. The identity of compounds 6 and 7 was confirmed by Fourier transform IR, MALDI, and C¹³ NMR. Treatment of 8 with N-hydroxysuccinimide afforded the active ester (9), which was then immediately reacted with 5-(biotinamido)pentylamine (Pierce Chemical) to generate the fluorophosphonate probe (Fig. 8A). Confirmation of the probe molecular weight was by mass spectrometry. The probe was stored at 10 mm in isopropanol at −80 °C and diluted in aqueous solution immediately before use.

FIGURE 8.

Demonstration of probe reactivity with serine proteases, and non-reactivity with non-serine proteases. The probe was incubated for 10 min with non-nucleophilic proteins ovalbumin, casein, glyceraldehyde-3-phosphate dehydrogenase (G3PD), three mAbs (anti-NGF, anti-nerve growth factor; anti-HGF, anti-hepatocyte growth factor; and anti-Fas) or with the serine proteases trypsin, factor Xa, thrombin, and plasmid. Reactions were resolved on SDS-PAGE and transferred to nitrocellulose membrane, and the biotin from the probe detected with streptavidin-horseradish peroxidase.

For probe blots, 0.5–2.0 μg of protein was incubated with 5 μm probe at 37 °C for 10 min. For a dimeric light chain of 50 kDa, 0.5 μg of protein in 10 μl corresponds to a protein:probe ratio of 1:5. The reaction was heated to 80 °C for 5 min in the presence of SDS loading buffer, and then resolved on a 12% SDS-PAGE gel. Either dye-coupled markers (Rainbow markers, Amersham Biosciences) or biotinylated IgG were used as molecular weight standards. The resolved proteins were transferred to a nitrocellulose membrane. The membrane was blocked with a solution containing 2% bovine serum albumin, 100 mm NaCl, 50 mm Tris-Cl, pH 7.4, and 0.025% Tween 20. The membrane was incubated with Streptavidin-horseradish peroxidase (Sigma-Aldrich) at a 1:2000 dilution for 45 min in block buffer. The membrane was washed three times with block buffer, followed by development on the membrane using the metal-enhanced diaminobenzidine kit (Pierce). In some cases (Fig. 9), the protein was denatured by heating to 90 °C for 3 min prior to adding the fluorophosphonate probe.

FIGURE 9.

Reaction of probe with antibody LC. Left: probe binding to A18bor factor Xa was analyzed before (lanes 2 and 5) or after (lanes 3 and 6) heat denaturation. No probe was added in lanes 1 and 6. Reactivity was only seen when incubated with properly folded protein in both cases. Right: the same reactions on the left were analyzed by SDS-PAGE and silver stain to show that equal amounts of protein were indeed present in each reaction.

Protein Hydrolysis

Human TNFα (BIOSOURCE) was incubated with 1 μg of A18b or 10 ng of trypsin (Sigma) for the indicated period of time, followed by anti-TNF Western blot as previously described (21). The peptide fragments were mapped using electrospray ionization-mass spectroscopy. Peptides and their sequences were identified as described in Trisler, et al. (21).

RESULTS

Since germ line antibodies have been reported to contain endogenous hydrolytic activity (22, 23), and the germ line counterparts of catalytic antibodies also contain activity (4, 8) we decided to directly test a naïve human antibody fragment. Endogenous hydrolytic activity has been mapped to the light chain of the c23.5 mouse anti-VIP antibody (24, 25), and other homologous light chains with hydrolytic activity have been reported (26, 27). Alignment of the amino acid sequences of known human kappa V regions revealed several sequences that were highly homologous to the c23.5 light chain, with complete conservation of the proposed catalytic active site at Asp-1, Ser-27a, and His-93. The V region of the most homologous of these, A18b, was generated by PCR from non-lymphoid genomic DNA and rearranged to J_κ1 by overlap extension using oligonucleotides. The A18b V region was cloned into an expression vector that directed the light chain to the Escherichia coli periplasm. The light chain was purified twice by denaturing IMAC to electrophoretic purity. In control purifications, an identically treated culture containing an empty expression vector was treated under the same purification conditions so that any background contamination would be easily detected (Fig. 1A, lanes 3 and 7). Similarly, an excess of trypsin was purified in the same manner, so that we could be certain that highly active proteases would be purified away using this procedure. In this regard, any nucleophilic or hydrolytic activity attributable to A18b could be compared with the negative control purifications to rule out contaminating activity that may be below the electrophoretic detection threshold.

FIGURE 1.

Characteristics of partially pure LC. A, periplasmic extract was prepared from cells expressing His₆-tagged A18b (lanes 1 and 5), the S27aA mutant of A18b (lanes 2 and 6), or containing the expression vector alone (lanes 3 and 7). These were purified over a cobalt IMAC column twice under denaturing conditions, along with a control containing 1 mg of pure trypsin (lanes 4 and 8, which is not expected to bind the resin). The purified proteins were resolved by SDS-PAGE in the absence or presence of DTT, and stained with silver (top), or transferred to nitrocellulose and reacted with anti-kappa-horseradish peroxidase (middle) or anti-His-horseradish peroxidase (bottom) and detected by the metal-enhanced diaminobenzidine kit (Pierce). B, equal volumes of the purified preparations were tested for PFR-MCA hydrolysis. Importantly, the negative controls (vector alone, or trypsin) showed no detectible protein on SDS-PAGE or hydrolytic activity.

When A18b and the serine 27a to alanine mutant were purified, they were seen to migrate as ∼66% dimer and 33% monomer by silver stain analysis on non-reducing SDS-PAGE (Fig. 1A, top gel, lanes 1 and 2). Under reducing conditions, nearly all of the dimer was converted to monomer (Fig. 1A, top gel, lanes 5 and 6). Importantly, when the purified preparation from the empty vector or trypsin were analyzed, no detectible bands were seen. This indicates that any contaminating proteins would be present at levels below the detection limit of a silver stained gel (4 ng, by the data in Fig. 2). By Western blot, all of the A18b bands purified reacted with anti-kappa and anti-His antibodies, indicating that they are the appropriate protein (Fig. 1A, bottom gels, lanes 1–2 and 5–6). When the A18b preparations were reacted with the protease substrate PFR-MCA, significant activity was seen with both wild-type A18b as well as the serine mutant. The activity of the serine mutant was surprising, because this was reported to be the nucleophile in a related light chain from mouse (24, 28).

FIGURE 2.

Purity of A18b. Left: a dilution series of A18b was performed over four orders of magnitude of protein concentration, then resolved by SDS-PAGE in the absence (lanes 1–4) or presence (lanes 5–8) of DTT then visualized by silver staining. Right: the highest protein amounts (4 μg) from the left gel were resolved by SDS-PAGE in the absence (lane 9) or presence (lane 10) of DTT and subjected to Western blot using anti-kappa antibodies. All of the bands seen on the left at 4 μg react with anti-kappa. Because the protein detection limit was 0.004 μg, we conclude that the protein preparation was at least 99.99% pure.

Despite the apparent purity of the IMAC preparations, we used additional orthogonal purification steps to further rigorously purify the light chains. Cation-exchange chromatography was followed by size-exclusion chromatography to afford A18b with at least 99.9% purity, as judged by titration analysis on SDS-PAGE and protein detection by silver staining (Fig. 2, left gel). Protein purity and detection limit were evaluated over four orders of magnitude of protein concentration. At the highest protein amount (4 μg), all bands on the gel were unambiguously found to be reactive with anti-kappa (Fig. 2, right), and the silver stain detection limit was easily 4 ng (Fig. 2, lanes 4 and 8). Thus, if any contaminating protein were present, it would need to be at amounts <0.1%. The same purification regimen was also accomplished with the S27aA mutant.

Although we obtained a highly pure protein preparation, there still remained the possibility that a very highly active hydrolase may have co-purified in amounts <0.1%, and could account for the low hydrolytic activity seen in the A18b preparation. It would be very unlikely, however, that such a contaminant would survive the denaturing IMAC. Nonetheless, we further characterized the A18b preparation by running an SDS-PAGE gel, and excising the bands corresponding to A18b monomer and dimer (Fig. 3A, labeled a and d in the right gel). As controls, we excised the residual bands (Fig. 3A, labeled b and c) that were at the positions corresponding to 50 kDa under reducing conditions (Fig. 3A, c), or 25 kDa under non-reducing conditions (Fig. 3A, b). These controls were used, because (i) a contaminating hydrolase would unlikely have the exact same molecular weight as the light chain and (ii) in the event that it did have the same molecular mass, it would not be disulfide-linked (most serine proteases are active as single subunits) and would also be present in the excised band at 50 kDa under reducing conditions (band c). When the excised bands were extracted and renatured, only the band corresponding to 50 kDa under non-reducing conditions contained hydrolytic activity (Fig. 3B). On the basis of these results, we conclude that the hydrolytic activity is present in the light chain itself and not any contaminating species.

We found that the dimer could be purified away from the residual monomer by size-exclusion high-performance liquid chromatography (Fig. 3C). When these were tested for hydrolase activity, both species were active, albeit the monomer less so (Fig. 3D). Thus, activity was present in both dimer and monomer preparations, however, one cannot rule out that the monomer can form an active dimer after purification and during the course of the hydrolase reaction. Hydrolysis of PFR-MCA by the highly purified A18b was found to display apparent Michaelis-Menten kinetics, with a K_m of 19 μm (Fig. 4). Turnover was present, because more substrate was hydrolyzed than the amount of A18b in the reaction.

FIGURE 4.

Lineweaver-Burke plot of A18b activity. A18b was incubated with multiple concentrations of PFR-MCA substrate, and hydrolytic activity was measured. A standard curve of the MCA leaving group was produced to calculate the concentration of substrate hydrolyzed. V_max and K_m (19.6 μm) were calculated based on the y and x intercepts.

Size-exclusion chromatography revealed three distinct light chain forms, corresponding to monomer, dimer, and a multimer (Fig. 5, top). Non-reducing SDS-PAGE analysis of these fractions showed that the monomeric peak corresponded to a molecular mass of 27 kDa, whereas both the dimer and multimer contained light-chain dimers with a molecular mass of ∼55 kDa. All of these bands were confirmed to be light chain by anti-kappa Western blot (data not shown). The multimer peak corresponded to a molecular mass of ∼100 kDa by size-exclusion chromatography, which would be consistent with a tetramer. SDS-PAGE, however, showed a molecular mass of ∼55 kDa. This result suggests that the 100-kDa peak on size-exclusion chromatography could be a tetramer consisting of a dimer of dimers, where each dimer is covalently linked by a disulfide bond. Because multiple species of light chain were present in wild-type A18b preparations, we wondered whether the serine mutant, S27aA, would also form the same macromolecular complexes. When A18b was compared with S27aA on size-exclusion chromatography we found no difference in the relative amounts of monomer, dimer, or multimer (supplemental Fig. S2).

FIGURE 5.

Multimeric properties of A18b. Top: three significant peaks were seen in size-exclusion chromatography. P1 corresponds to a molecular mass of ∼100 kDa, and P2 to 50 kDa. Bottom: fractions were collected and analyzed by non-reducing SDS-PAGE. The position of the lane indicates the approximate retention time for which that fraction was collected. Both P1 and P2 were seen to be composed of dimeric A18b, whereas the monomer (unlabeled peak at 19 min) eluted at a slower than predicted molecular mass.

When fractions of the multimer peaks were analyzed for PFR-MCA hydrolysis, however, a difference was noted between the wild-type and S27aA preparations (Fig. 6). When the multimer peak (P1) was analyzed, there was little difference between the wild-type and serine mutant (Fig. 6, Left), however the dimer peak (P2) showed considerably more activity in the wild-type preparation compared with the serine mutant (Fig. 6, right graph). The small amount of activity seen in the P2 S27aA preparation could be due to the overlap of peaks and, thus, multimer impurities in the dimeric fraction. Over the course of several hours, the purified peaks appeared to be stable; they did not re-equilibrate to P1 and P2 when they were rerun on the size-exclusion chromatography column (data not shown). Thus, the serine mutant appears to affect hydrolytic activity in the dimeric form but not in the multimeric form. Strikingly, both preparations contained hydrolytic activity in a multimeric form. When the A18b preparation was tested for PFR-MCA hydrolysis at different pH levels, we found a broad range of activity from pH 6.0 to 8.8, with an optimum at pH 8.0 (data not shown).

FIGURE 6.

Activity of P1 and P2. The dimer of dimers (P1, left) and dimer (P2, right) were purified by successive size-exclusion chromatography rounds, then tested for hydrolytic activity against the PFR-MCA substrate. The S27aA serine mutant (squares) was purified and tested side-by-side with wild-type A18b (diamonds), with equimolar amounts of either dimer (50 kDa) or tetramer (100 kDa).

Several investigators have shown hydrolytic activity with various different antibodies or light chain preparations, and have concluded that they possess a serine protease type of activity (9, 27, 28). To further elucidate the mechanism behind this activity, we exposed A18b to different protease inhibitors in the context of a PFR-MCA hydrolysis assay (Fig. 7). Many serine protease inhibitors, including leupeptin, TPCK, and PMSF inhibited A18b hydrolytic activity. Other serine protease inhibitors, like TLCK and benzamidine, did not inhibit the activity significantly. Disulfide bond integrity was required for activity, because the reducing agent DTT abolished activity. Strikingly, the cysteine protease inhibitor l-transepoxysuccinylleucylamido-(4-guanidino)butane (E-64) also inhibited A18b. E-64 is an epoxy compound with electrophilic properties that binds irreversibly to the nucleophilic cysteine in cysteine proteases. Interestingly, it been reported to inhibit trypsin, which is a serine protease (29).

FIGURE 7.

Effect of inhibitors on hydrolytic activity. Protease inhibitors known to inhibit the major classes of proteases were tested for inhibitory activity against A18b. Activity is normalized to 1 for the uninhibited reaction.

Because A18b activity could be inhibited with the chloromethyketone TPCK, the sulfonyl fluoride PMSF, and the epoxide E-64, we hypothesized that the activity could result from a nucleophilic residue that could form covalent complexes with these protease inhibitors, which are known to irreversibly bind proteases. In this regard we synthesized a probe that consisted of a fluorophosphonate at one end and a biotin at the other (38). If the A18b could displace the fluoride and form a covalent complex with the phosphonate, this complex could be detected via the biotin group through streptavidin labeling. For these experiments the probe was incubated with protein, resolved on SDS-PAGE, and transferred to nitrocellulose, and the biotinylated protein detected with streptavidin was coupled to horseradish peroxidase. We tested this strategy on several known proteases as well as irrelevant proteins (Fig. 8). The serine proteases trypsin, factor Xa, thrombin, and plasmin were all specifically labeled, whereas ovalbumin, casein, glyceraldehyde-3-phosphate dehydrogenase, and three monoclonal antibodies were not labeled (Fig. 8). Interestingly, bovine serum albumin, which is not a serine protease but is known to contain nucleophilic lysine residues, was strongly labeled (data not shown). When A18b was incubated with the probe, it was labeled similarly to factor Xa (Fig. 9A, lane 2). However, when it was denatured by heat, it showed very little reactivity (Fig. 9A, lane 3), confirming that properly folded light chain was required for nucleophilic activity. The same reactions were analyzed by SDS-PAGE and silver staining to ensure that labeling differences were not due to differing protein levels (Fig. 9A, right panel). As expected, we also found that binding to the fluorophosphonate probe could be inhibited when A18b was preincubated with PMSF or TPCK, two of the covalent binding inhibitors with activity against A18b PFR-MCA activity, but not disrupted significantly with TLCK, which minimally inhibited A18b (Fig. 9B). The serine mutant of A18b could also be labeled by the fluorophosphonate probe, albeit at lower levels than wild type (supplemental Fig. S4).

Because A18b could hydrolyze model peptide substrates like PFR-MCA, we tested a larger protein of therapeutic relevance, TNFα. When A18b was incubated with TNFα for 48 h, several hydrolytic products were identified by both Western blot (Fig. 10) as well as mass spectrometry (supplemental Fig. S5). When these products were mapped, they did not correspond to specific consensus cleavage sites (Fig. 10B). The same experiment was done using trypsin as a catalyst, which invariably cleaved after its consensus arginine residues. For A18b, cleavage appeared to correspond to accessible residues as opposed to specific amino acid side chains. For example, the disordered N terminus of TNFα was cleaved at Ser-4 and Pro-6; residues Gly-108 and Ala-109 were cleaved in an accessible loop; and the solvent-accessible Arg-44 in a strand was a target for A18b.

FIGURE 10.

Hydrolysis of TNFα by A18b. A, TNFα (lanes 1–4 and 7 and 8) was incubated with A18b (lanes 5–8) or trypsin (lanes 3 and 4) for zero (odd lanes) or 48 h (even lanes) followed by SDS-PAGE and Western blotting against TNFα. Cleavage was similarly verified by mass spectroscopy (supplemental Fig. S4). B, the cleavage positions were mapped and are shown on a ribbon diagram of the TNFα trimer.

DISCUSSION

The ability of antibodies to catalyze various chemical reactions has been studied for decades (30). Many antibody catalysts have been induced by transition state analogs, with resultant catalysts stabilizing the transition state of a reaction of interest. When the germ line precursors of these antibodies were produced, they were also found to contain catalytic activity, albeit at low kinetic rates (4, 7, 8). The structural and functional basis of many of these antibodies has been elucidated for both the mature and germ line variants (4, 7, 8). In addition to these induced catalytic antibodies, several natural antibodies containing hydrolytic activities have been reported. An antibody, c23.5, capable of hydrolyzing vasoactive intestinal polypeptide has been extensively characterized, with its hydrolytic properties mapped to the light chain (19, 24, 25). A serine protease triad was proposed to account for this activity, which was also present in its germ line precursor (22).

Recently, catalytic antibodies have been implicated in health and disease. A subpopulation of hemophiliacs with resistance to Factor VIII therapy were found to produce high titers of antibody capable of specific hydrolysis of recombinant Factor VIII (12, 13). Septic patients with high titers of nonspecific proteolytic antibody were found to have increased survival compared with septic patients without such catalytic antibodies (11). Interestingly, certain multiple sclerosis patients were found to contain catalytic antibodies against myelin basic protein (15). Whether these antibodies are causative in the pathology of this disease remains an open question. Thus, understanding the molecular basis underlying these antibodies could provide important information that could impact drug or diagnostic development. The genetic basis behind these antibodies has not yet been described. Because these antibodies are from patients, cloning of the genes would be difficult. However, because these activities are thought to be germ line-encoded, and the sequences of all V, D, and J segments are known, the search for these properties among human germ line antibodies can now be undertaken.

Here we characterize a human light chain, A18b, that contains unique multimerization, hydrolytic, and nucleophilic properties. The hydrolytic activity resided in the light chain (and not a co-purifying protease contaminant) because (i) we validated that our purification process could remove E. coli contaminants by purifying control extracts without light chain, (ii) similarly our purification could remove excess trypsin, one of the most highly active serine proteases, (iii) the light chain could be purified from SDS-PAGE gels, refolded, and retain hydrolytic activity, (iv) nucleophilic activity could be shown directly by binding the light chain to an electrophilic probe, and (v) mutagenesis could reduce hydrolytic activity in the tetrameric form of the light chain. Because purification was done under denaturing conditions it is very unlikely that some co-purifying highly active contaminant of the same molecular weight could be the source of the hydrolytic and nucleophilic activity.

When extensively purified, this light chain could hydrolyze peptide substrates and bind to an electrophilic fluorophosphonate probe analogous to the binding of irreversible protease inhibitors to serine proteases. In this regard, certain serine protease inhibitors could inhibit the hydrolytic activity of this light chain. Surprisingly the cysteine protease inhibitor E-64 could also inhibit this activity. This is unexpected for a serine protease, although recently trypsin inhibition by E-64 has been reported (29). Although the catalytic activity of naturally occurring proteolytic antibodies has been reported to occur through a serine protease mechanism involving a triad of Ser/His/Asp (26, 28, 31), we found that mutation of the putative catalytic serine residue did not completely abolish activity. In fact, the S27aA mutation appeared to significantly affect catalytic activity in the dimeric form of A18b as opposed to the tetramer. The difference in activity between different tertiary complexes of the same mutant protein is inconsistent with a simple serine protease triad active site. It is possible that the alanine substitution at position 27a results in an alternative fold of the CDR, which impacts catalysis when in the dimeric form.

The ability of serine protease inhibitors to inhibit an enzyme does not necessarily confirm the same catalytic mechanism of a new catalyst. In our study, some serine protease inhibitors could inhibit A18b, and some (e.g. benzamidine) could not. The irreversible inhibitors PMSF and diisopropylfluorophosphate act by forming a covalent complex with the highly nucleophilic serine at the active site of a serine protease. Although serine is the classic amino acid responsible for this activity, other nucleophiles could also accomplish this reaction. In this regard, we found that serum albumin could also bind the fluorophosphonate probe. Albumin does not have a serine protease active site, but is known to contain surface-exposed nucleophilic lysines. In the case of A18b, E-64 could also inhibit hydrolytic activity. As mentioned, E-64 is not thought to be a serine protease inhibitor. Thus, we are not convinced that the mechanism by which A18b acts is through a classic serine protease triad, at least for the tetrameric form. In this regard, mutation of the most likely catalytic serine, Ser-27a, did not result in the expected loss of activity. It is possible that certain germ line antibodies contain generally nucleophilic residues in their CDRs, which allow attack on carbonyl carbons in a peptide bond when they are appropriately positioned. These nucleophilic residues could be serines, which are highly abundant in antibody CDRs, but could also be other residues like lysine, histidine, threonine, or tyrosine. Several additional serines are found in A18b, including Ser-24 and Ser-25 in CDR1. Recently tyrosine nucleophiles have been found in antibodies that covalently react with diphenylphosphonates (32). The nucleophilicity could be determined through interactions with adjacent residues that may partially abstract a hydrogen from a side-chain hydroxyl group, or present locally hydrophobic pockets in the context of a charged residue like lysine. The latter is the mechanism proposed for the catalytic lysine in an aldolase antibody with nucleophilic properties (33). Of note, the aldolase antibody 93F3 contains a catalytic lysine in the light chain at position 89, which is conserved in the germ line V region 8–21, suggesting this activity could be part of the innate antibody system. For aldolase antibody 33F12, the nucleophilic lysine is at heavy chain residue 93, which is encoded at the V-D junction. Thus, this lysine is not coded by the germ line VH22.1 germ line gene, but would be produced in naïve B-cells as a result of V(D)J recombination and not necessarily through affinity maturation. The structural basis for A18b with regards to these nucleophilic mechanisms will require further study.

Bence-Jones proteins are antibody light chains produced in multiple myeloma. These proteins can contribute to pathogenesis by producing amyloid and glomerulonephritis. Given the medical importance of Bence-Jones proteins, their biochemistry has been extensively studied. Light chains can exist as monomers, dimers, or higher molecular weight species. The K_D of self-association appears to be determined by the V region sequences and can vary widely among different light chains (34). It is now well established that a germ line antibody can form multiple structures (4, 5, 35), a result of inherent flexibility in germ line CDRs (6). We anticipate that the A18b light chain, as well as other light chains, may also exhibit conformational flexibility. In this regard, variations in the hydrolytic activity and nucleophilic properties were found between different A18b preparations; however, these could be restored to similar levels through denaturation and refolding. Indeed, light-chain dimers can form multiple structures under different crystallization conditions (36), and aggregates have been proposed to function in the catalytic mechanism (37). The relationship between the biophysical characteristics of various light chains and myeloma pathogenesis is not yet clear.

Now that all human antibody V regions have been sequenced, the “hard-coded” portion of the antibody repertoire is known and can be studied in detail. Identification of nucleophilic antibodies in the germ line and the in-depth study of their biochemical properties should allow further insights into the evolution and mechanisms of the antibody response to natural pathogens and in autoimmune disease.