A cofactor approach to copper-dependent catalytic antibodies

Abstract

A strategy for the preparation of semisynthetic copper(II)-based catalytic metalloproteins is described in which a metal-binding bis-imidazole cofactor is incorporated into the combining site of the aldolase antibody 38C2. Antibody 38C2 features a large hydrophobic-combining site pocket with a highly nucleophilic lysine residue, Lys^H93, that can be covalently modified. A comparison of several lactone and anhydride reagents shows that the latter are the most effective and general derivatizing agents for the 38C2 Lys residue. A bis-imidazole anhydride (5) was efficiently prepared from N-methyl imidazole. The 38C2–5-Cu conjugate was prepared by either (i) initial derivatization of 38C2 with 5 followed by metallation with CuCl₂, or (ii) precoordination of 5 with CuCl₂ followed by conjugation with 38C2. The resulting 38C2–5-Cu conjugate was an active catalyst for the hydrolysis of the coordinating picolinate ester 11, following Michaelis–Menten kinetics [k_cat(11) = 2.3 min⁻¹ and K_m(11) 2.2 mM] with a rate enhancement [k_cat(11)k_uncat(11)] of 2.1 × 10⁵. Comparison of the second-order rate constants of the modified 38C2 and the Cu(II)-bis-imidazolyl complex k(6-CuCl₂) gives a rate enhancement of 3.5 × 10⁴ in favor of the antibody complex with an effective molarity of 76.7 M, revealing a significant catalytic benefit to the binding of the bis-imidazolyl ligand into 38C2.

Catalysis at metal centers plays a key role in both enzymatic and abiological reactions, providing reaction pathways, rates, and selectivity often unattainable from conventional acid, base, or nucleophilic catalysis. However, traditionally separate disciplines, the study of natural (i.e., enzymatic) and synthetic (nonbiological) catalysts has interfaced with the genesis of semisynthetic and de novo proteins (1). A number of methods have been explored for producing such hybrid species that possess metal centers, including: (i) attachment of a synthetic metal-containing cofactor to a protein (2–4); (ii) antibody elicitation using a metal-binding hapten/protein conjugate (5–10); (iii) site-directed mutagenesis of proteins and antibodies to incorporate metal-binding sites (11–15); and (iv) panning of antibody libraries for metal binding using immobilized metal complexes (16, 17).

Our approach is directed toward the design and generation of novel metallo-catalytic antibodies and is inspired by the active-site structures of the many metalloenzymes that possess a coordination sphere with two or more histidine-derived imidazole ligands. The most important catalytic functions of such polyhistidyl enzymes are hydrolytic or oxidative. Illustrative of the numerous and structurally diverse polyhistidyl Zn-hydrolases (18) is carboxypeptidase A, which catalyzes the hydrolysis of C-terminal amino acid residues and features a bis-histidine/glutamate (CO) coordination sphere and the nucleophilic assistance of a nearby Glu (19). The Cu-hydroxylases, dopamine β-hydroxylase and peptide amidating hydroxylase (20), are chemically extraordinary in effecting the regio- and enantioselective hydroxylation of weakly activated C—H bonds, a transformation with little precedent among nonenzymatic copper catalysts (21, 22). The active sites of hydroxylase enzymes possess two Cu-polyhistidine centers, one of which appears to serve as an electron or superoxide transfer site and the other at which the substrate is hydroxylated (23–25). The physiological importance of these enzymes continues to evoke intense efforts to elucidate their molecular mechanisms, to mimic their structure and function, and to discover mechanism-based inhibitors for biomedical applications (26–27).

Numerous structural and functional model compounds for the active site of Zn-hydrolase enzymes have been prepared (18), but few possess the imidazole (histidine) or carboxylate ligands of the enzymes. More sophisticated model systems have incorporated hydroxyl- and carboxyl-containing tethers to serve as nucleophilic or protic functions (28, 29), micelles (30), and cyclodextrins (31, 32) to simulate the protein environment of metalloenzymes. Substantial rate accelerations for hydrolysis, particularly of carboxylic acid esters, have been attained with some of these model Cu(II) and Zn(II) complexes (33–36). Although Cu complexes, which reasonably mimic the structural and spectroscopic features of the Cu-hydroxylases, have been prepared (37–40), few possess polyimidazole ligands or have been shown to promote the hydroxylation of relevant substrates (usually stoichiometric hydroxylation of their own ligands is a major problem) (41–44), and none of these reactions are catalytic.

Our initial efforts in this area have focused on the preparation of a metal-binding bis-imidazolyl ligand and the corresponding metal complex cofactor that features high-affinity metal–ligand binding (K_d ≈ 10⁻¹²) (45). The metal cofactor is then incorporated into the combining site of the well-characterized monoclonal catalytic antibody 38C2 (46) to allow investigation of the catalytic potential of the biomimetic bis-imidazole coordination cofactor. We reasoned that the resulting semisynthetic metalloantibodies would ideally possess the catalytic function of the metal center enhanced by the unique microenvironment characteristics imparted by the antibody combining site.

Antibody 38C2 was selected as the Ig for modification because it possesses a large promiscuous hydrophobic combining a site pocket that contains a highly nucleophilic low-pKa lysine residue, Lys^H93, which can be covalently modified (47, 48).

Materials and Methods

Monoclonal antibody 38C2 was obtained from The Scripps Research Institute. The antibody was routinely purified in a three-stage process that included precipitation into ammonium-sulfate protein-G affinity column chromatography and ion-exchange column chromatography. The resultant protein preparation was >99% pure as determined by SDS/PAGE analysis and silver staining. Protein concentration was routinely measured by UV spectroscopy {1.35 × (OD at 280 nm) = [protein] mg/ml}. Conversion to molar concentration assumed a molecular weight of a whole IgG = 150,000 and two binding sites per antibody molecule. All buffers and salts were obtained from commercial sources. Doubly distilled nanopure water was used throughout. Dialysis was conducted by using microdialysis cassettes (10,000 molecular weight block) (Pierce). UV-Vis, NMR (300 MHz), IR, and mass spectra (electrospray ionization) were performed on standard instruments. Inductively coupled-plasma-MS analysis (for Cu) was carried out by Elemental Research (Vancouver). Lactones 2a–2e and anhydrides 3a–3c were obtained from Aldrich and used without further purification. Anhydride 3d (49), 3e (50), (S)-(+)-decalenone derivative 4a (51), 4-nitrophenyl-2-picolinate 11 (33), and bis-imidazole carbinol 6 (45) were prepared according to reported procedures. The titration of Lys^H93 enone formation by reaction of 38C2 with pentane-2,4-dione at 318 nm was performed as previously described (46, 47).

Preparation of 2-(S)-(4-Nitrophenoxycarbonylamino)succinic Acid Di-tert-butyl Ester (7).

Di-tert-butyl aspartate hydrochloride (0.80 g, 2.8 mmol) and p-nitrophenyl chloroformate (0.68 g, 3.4 mmol) were dissolved in dry CH₂Cl₂ (25 ml) and stirred under an inert atmosphere (argon). N-ethyl morpholine (0.72 ml, 5.7 mmol) was added by syringe. After stirring for 4 h, the solution was washed twice with equal volumes of cold aqueous NaHCO₃ and then cold brine. The organic phase was dried over MgSO₄, rotary evaporated, and then dried in vacuo to provide the carbamate 7 (1.30 g). NMR analysis routinely indicated that 7 was ≈90% pure with a small amount of p-nitrophenol present. A pure sample of 7 was obtained by silica gel chromatography, eluting with ethyl acetate; ¹H NMR (CDCl₃) δ 8.2 (d, 8 Hz, 2H), 7.3 (d, 8 Hz, 2H), 6.2 (d, J = 6 Hz, 1H), 4.5 (m, 1H), 3.0 (dd, J = 12, 4 Hz, 1H), 2.8 (dd, J = 12, 4 Hz, 1H), 1.55 (s, 9H), 1.50 (s, 9H): ¹³C NMR (CDCl₃) δ 159.8, 157.5, 145.2, 124.3, 124.0, 122.7, 121.0, 59.3, 48.6, 35.9, 24.3, 24.1; HRFABMS C₁₉H₂₇N₂O₈ (M+H)⁺ calcd. 411.1768; obsd. 411.1770.

Preparation of Di-tert-butyl-N-aspartyl-bis-2-imidazolyl Carbamate (8).

Sodium hydride dispersion (60% in mineral oil, 0.050 g, 1.25 mmol) was washed three times with hexane (2 ml) under argon, and then anhydrous tetrahydrofuran (THF) (10 ml) was added. To this stirred mixture was added bis-2-imidazolyl carbinol 6 (0.21 g, 1.1 mmol) (note: gas evolution accompanies the evolution). After 30 min, a solution of 7 (0.65 g, 1.3 mmol) in THF (7 ml) was added, and the mixture was allowed to stir overnight. The orange reaction mixture was then diluted with an equal volume of ethyl acetate, and the mixture was washed several times with aqueous NaHCO₃. The organic phase was dried over MgSO₄ and concentrated to dryness in vacuo. The crude residue was purified by silica gel chromatography [EtOAc (1% Et₃N), followed by 1:6 MeOH/EtOAc (1% Et₃N)] to afford product 8 (0.35 g, 68% yield based on 6) as a white solid. ¹H NMR (CDCl₃, δ) 7.2 (s, 1H), 7.0 (d, 2 Hz, 2H), 6.8 (d, J = 2 Hz, 2H), 6.0 (d, J = 6 Hz, 1H), 4.4 (m, 1H), 3.85 (s, 3H), 3.8 (s, 3H), 2.9 (dd, J = 12, 4 Hz, 1H), 2.7 (dd, J = 12, 4 Hz, 1H), 1.45 (s, 18H)]; ¹³C NMR (CDCl₃, δ) 170.8, 170.1, 155.1, 143.4, 128.3, 123.3, 66.2, 51.6, 38.2, 34.3, 28.6; MS (electrospray ionization, MeOH) m/e 464 (M+H⁺ for C₂₂H₃₂N₅O₆).

Preparation of Diacid 9.

To a stirred solution of ester 8 (0.21 g, 0.45 mmol) in dry CH₂Cl₂ (3 ml), trifluoroacetic acid (2 ml) was added. After 5 h at room temperature, HPLC analysis (C₁₈, 1:4 acetonitrile/water, 1% trifluoroacetic acid) indicated that the reaction was complete. The volatiles were removed by rotary evaporation, and the residue was dried under vacuum to give the diacid 5 (di-trifluoroacetate salt) as an oily colorless solid (0.26 g, 100%). ¹H NMR (CD₃CN, δ) 7.3 (bs, 4H), 7.2 (s, 1H), 6.8 (d, J = 6 Hz, 1H), 4.4 (m, 1H), 3.8 (s, 6H), 2.8 (m, 2H); ¹³C NMR (CDCl₃, δ) 174.5, 163.5, 137.6, 127.7, 121.6, 120.6, 115.2, 59.5, 51.7, 50.0, 36.0, 35.5, 34.5; MS (electrospray ionization, MeOH) m/e 352 (M+H⁺ for C₁₄H₁₆N₅O₆).

Preparation of Anhydride 5

NMR Scale.

The diacid 9 (5 mg) was dissolved in d₆-acetic anhydride (0.5 ml). This solution was monitored by NMR periodically; after 5 h at room temperature, the starting diacid was completely converted to a single product whose NMR spectrum was that corresponding to anhydride 5.

Preparative Scale.

The diacid 9 (12 mg, 0.022 mmol) was dissolved in acetic anhydride (1.0 ml) and stirred under argon for 3 h. The volatiles were removed in vacuo; the oily residue was triturated several times with dry toluene, then with dry diethyl ether, and vacuum dried to afford the anhydride 6 (10 mg, 100%) as an oily solid that was used without further purification for conjugation experiments. ¹H NMR (CD₃CN) δ 8.0 (d, 6 Hz, 1H), 7.2–7.4 (m, 5H), 4.7 (m, 1H), 3.8 (s, 3H), 3.7 (s, 3H), 3.2 (dd, J = 12, 5 Hz, 2H), 3.0 (dd, J = 12, 3 Hz, 2H).

Preparation of [Bis-(2-N-methylimidazolyl)methanol] Copper(II) Dichloride 6-CuCl₂.

A 10 mM solution of the bis-imidazole carbinol 6 in acetonitrile was added dropwise to an equal volume of 10 mM CuCl₂⋅6H₂O in acetonitrile resulting in an immediate color change to greenish blue and gradual crystallization of blue L⋅CuCl₂ (7) in nearly quantitative yield. The crystals, which were washed with acetonitrile and dried, were suitable for x-ray diffraction. UV-Vis (CH₃CN, λ, ɛ) 204 (2,400), 242 (2,100), 275 (1,000), 338 (200), 600–750 (50).

Conjugation Procedures

General Method for Modification of 38C2 with Lactones 2a–2e and Anhydrides 3a–3e.

In a typical assay, 38C2 (0.5 μM) was incubated with either a lactone 2a–2e (250 μM) or anhydride 3a–3e (250 μM) in PBS (pH 7.4) for 30 min at room temperature. The reaction was initiated by addition of the inhibitor from a stock solution in acetonitrile such that the organic cosolvent was equal to 5%. After 30 min, the unreacted lactones or anhydrides were removed by dialysis (10,000 molecular weight membrane for 24 h) at 4°C in 50 mM PBS (pH 7.4).

Large-Scale Preparation of 38C2–5-CuCl₂

Method A.

In a representative experiment, a solution of 38C2 (2.85 ml, 30 μM) in 50 mM NaCl (pH 5.8) was treated with a solution of 5 (150 μl, 40 mM) in DMSO and allowed to stand at room temperature overnight. This solution was then treated with 150 μl of CuCl₂ (40 mM) in DMSO and allowed to stand overnight at room temperature. The resulting solution was then dialyzed (10,000 molecular weight membrane for 24 h) at 4°C in 50 mM NaCl (pH 5.5).

Method B.

In a representative experiment, a solution of 5-CuCl₂ was prepared by addition of 5 (25 μl, 40 mM) in DMSO to CuCl₂ (25 μl, 40 mM) in DMSO. The resulting blue-green solution was added to 38C2 (2.0 ml, 50 μM) in 50 mM NaCl and allowed to stand at room temperature overnight. The resulting solution was then dialyzed (10,000 molecular weight membrane for 24 h) at 4°C in NaCl (50 mM, pH 5.5).

Samples for inductively coupled-plasma-MS Cu-analysis were prepared by dialyzing solutions of the proteins into doubly distilled water (1:200 vol/vol) to remove interfering sodium ion. Analysis of 38C2-5-CuCl₂ (2.9 mg/ml) gave 310 μg of Cu/liter (≈0.25 Cu/Ab); analysis of 38C2 (2.9 mg/ml, no Cu added) showed 13 μg of Cu/liter (≈0.01 Cu/Ab).

Steady State Kinetic Experiments

Allylic Rearrangement of (S)-(+)-4a to Enone 4b.

The assays were performed as originally described (51). The assay was initiated by addition of 4a in acetonitrile (reaction concentration 250 μM) to a thermostated (23°C, ±1°C) cuvette containing either buffer (PBS, pH 7.4), 38C2 (0.5 μM in PBS, pH 7.4), or covalently modified 38C2 (0.5 μM in PBS, pH 7.4). Assays were performed with UV spectrophotometry; the assay volume was 1 ml, and the organic cosolvent was 5% by volume. The assay was conducted by observing the increase in absorbance because of the formation of enone 4b (ɛ₂₄₈ = 15,120 M⁻¹⋅cm⁻¹) on a Shimadzu UV/vis spectrophotometer. Experiments were performed in duplicate, and the reaction was followed for no more than 5% of the reaction, during which time the reaction progress was linear (r² > 0.985).

Time Course of the Modification of Antibody 38C2 with 3e in the Presence and Absence of 2,4-Pentanedione.

Modification conditions: [38C2] 2 μM, [3e] 200 μM; buffer conditions: 5% CH₃CN and 95% PBS (pH 7.4). The activity of the modified 38C2 was measured by removal of aliquots at time t during the assay and determination of the initial rate of isomerization of 4a as detailed above. Note that 3e had no effect on the rate of the isomerization reaction. The modification was also performed in the presence of 2,4-pentanedione (200 μM) under the same conditions.

Kinetic Parameter Measurement of 38C2–5-CuCl₂ Catalyzed Hydrolysis of Ester 11.

Hydrolysis of 11 was assayed in 4-morpholinepropanesulfonic acid (Mops) buffered solutions (25 mM, 25 mM NaCl, pH 7.0) with spectrophotometric monitoring of the product 4-nitrophenolate at 400 nm in the presence or absence 38C2-5-CuCl₂ (1 μM). The reaction was initiated by addition of stock solutions of the freshly prepared substrate 11 in dry DMSO to the catalyst solution to give substrate concentrations from 100 to 500 μM. Experiments were performed in duplicate, and the reaction was followed for no more than 5% of the reaction, during which time the reaction progress was linear (r² > 0.985). Kinetic parameters were determined by fitting the initial rate data to the Michaelis–Menten model with the enzfitter computer program.

The k_uncat(11) rate constant for the uncatalyzed hydrolysis of 11 was measured by using the method of initial rates. The reactions were performed in duplicate in an ELISA microtiter plate and were initiated by addition of the substrate 11 into buffer (25 mM Mops, 25 mM NaCl, pH 7.0) at 23°C to give final concentrations in the range of 50–500 μM. The hydrolysis of 11 was monitored by formation of p-nitrophenolate at 400 nm. The reaction was sufficiently slow to require sealing of the ELISA plate to prevent evaporation. The reaction was followed for no more than 5–10% of reaction, during which the progress curves were linear (r² > 0.985). The k_uncat was determined graphically by plotting the rate of phenolate formation against [11]. There was no measurable buffer effect on the noncatalyzed rate of hydrolysis of 11.

Measurement of the Bimolecular Rate Constant k for the Reaction Between 6-CuCl₂ and 11.

The reaction between 6-CuCl₂ and 11 was followed by the method of initial rates under pseudo-first-order assay conditions. The concentration of 11 was fixed at three concentrations in excess (300, 400, and 500 μM) in Mops buffer (25 mM, 25 mM NaCl, pH 7.0). The concentration of complex 6-CuCl₂ was varied (15, 30, 45 μM). The reaction was followed for <5% conversion, during which the rates were linear (r² > 0.985). The bimolecular rate constant k (6-CuCl₂) was determined by a two-step graphical process. First, the observed rate k_obs (at each fixed concentration of 11) was determined from a plot of rate vs. [6-CuCl₂] (graphpad prism software). The k_obs values where then plotted vs. [11] to give the true rate constant k (6-CuCl₂) = 0.03 M⁻¹⋅min⁻¹.

Results and Discussion

A previous approach to covalent modification of Lys^H93 of 38C2 used N-acyl β-lactam derivatives (48). The authors noted that the reactivity of the antibody with these electrophiles was low, and that the time taken for efficient derivatization ranged from 3 to 40 h. Therefore, we first sought to develop a reagent for covalent modification of Lys^H93 that would be more reactive than N-acyl β-lactams, to optimize selectivity and minimize surface modification. The only limitations were that it should be stable in aqueous buffer, and it must form an essentially irreversible linkage to Lys^H93. For these reasons, we primarily explored the ability of a range of lactones (2a–2e) and anhydrides (3a–3e) to modify 38C2 (Figs. 1 and 2).

Figure 1.

Potential mechanisms of covalent modification of Lys^H93 of 38C2 by either lactones or anhydrides.

Figure 2.

Lactones 2a–2e and anhydrides 3a–3e used as covalent modifying agents of 38C2.

Modification of the antibody active site was followed in two ways. The binding-site Lys^H93 was titrated with pentane-2,4-dione and followed by spectroscopic determination of the enaminone absorbance at 318 nm throughout the incubation (46, 47). In addition, the initial rate of 38C2-catalyzed isomerization (51) of (S)-(+)-β,γ-unsaturated ketone 4a to α,β enone 4b was measured at times throughout the conjugation process (Scheme S1).

Scheme 1.

In a typical modification experiment, the labeling reagent (2a–2e) and (3a–3e) (250 μM) was incubated with 38C2 (0.5 μM) in PBS (pH 7.4) at room temperature for 30 min. The inhibition of the initial rate of 38C2 formation of (S)-(+) 4b was then determined spectrophotometrically (248 nm) by addition of octalone 4a (250 μM). The data for the panel of lactones (2a–2e) and anhydrides (3a–3e) are shown in Table 1.

Table 1.

Covalent modification of 38C2 by lactones 2a–e and anhydrides 3a–3e

Electrophile	Inhibition*, %
2a	1.6
2b	93.7
2c	22.7
2d	7.4
2e	1.5
3a	98.9
3b	98.3
3c	97.4
3d	98.4
3e	95.6

In a typical assay, 38C2 (0.5 μM) was incubated with either the lactone 2a–e (250 μM) or anhydride 3a–e (250 μM) in PBS (pH 7.4) for 30 min at room temperature. The reaction was initiated by addition of the inhibitor from a stock solution in acetonitrile such that the organic cosolvent was equal to 5%. 

^*Inhibition of the 38C2-catalyzed production of enone 4b from 4a was monitored spectrophotomerically by the increase in absorbance at 248 nm (Scheme S1). Inhibition was determined from the difference in the initial rates of 38C2-catalyzed formation of 4b in the presence and absence of inhibitor and the background reaction and converted to a percentage. 

The lactones, in general, are much less effective derivatizing agents of 38C2 than their anhydride counterparts. As an example, γ-butyrolactone 2e inhibits the activity of 38C2 by only 1.5%, whereas its structural homologue succinic anhydride 3b knocks out 97.4% of the 38C2 activity; isophthalide 2a inhibits the 38C2-catalyzed process by only 1.6%, whereas phthalic anhydride 3c inhibits 98.4% of the catalytic rate. Interestingly, the original report of covalent modification of 38C2 noted that N-acyl γ-lactams, the lactam structural analog of 2e and 3c, do not react with 38C2 at all. In addition, the anhydride motif, regardless of the additional structural constituents of the inhibitor, seems to be fairly generic and well tolerated by 38C2. All of the anhydrides 3a–3e reduce the activity of 38C2 by >95% after 30-min exposure. Furthermore, the lactone-mediated inhibition of 38C2 exhibits strict structural requirements. For example, 2-coumaranone 2b is a 93.7% inhibitor, whereas the isomeric isophthalide 2a is only a 1.6% inhibitor. On the basis of these results, we have shown that anhydrides are, by far, the most efficient and general derivatizing agents for the 38C2 antibody yet reported.

In-depth kinetic studies revealed that the reaction of anhydride 3e is time-dependent and can be quantitatively inhibited by addition of pentane-2,4-dione (Fig. 3).

Figure 3.

Time-dependent inhibition of 38C2 by anhydride 3e in the presence (▴) and absence (■) of pentane-2,4-dione. V_o = initial rate at t = 0; V_t = initial rate after time t of inhibition.

Although Lys^H93 is by far the most likely binding-site residue to be modified during the reaction, it is feasible that the actual target could be an alternative binding-site residue such as Tyr^L41. In such a case, the failure to form the enamine complex between Lys^H93 and pentane-2,4-dione would simply be a result of steric hindrance within the antibody-combining site. To investigate whether Tyr^L41 is being acylated, 3e-modified 38C2 (>98%, 20 μM) was treated with hydroxylamine (200 mM) for 24 h at room temperature. After dialysis (to remove the excess hydroxylamine), titration of Lys^H93 with 2,4-pentanedione revealed no further enamine complex formation, supporting the notion that Lys^H93 is indeed the target for the anhydride-modifying agents.

Having thus revealed anhydrides to be excellent and generic derivatizing agents for 38C2, we then synthesized a first-generation bis-imidazolyl aspartic acid anhydride derivative 5 (Fig. 4).

Figure 4.

Reagents and conditions: (i) n-BuLi; HCO₂Bu; H₃O⁺; (ii) NaH; 7; (iii) trifluoroacetic acid/CH2Cl₂; (iv) Ac₂O.

The known bis-imidazolyl carbinol 6 was attached to an aspartic acid anhydride tether via a nucleophilic reaction with the key carbamate ester derivative 7 (68%), prepared in one step from di-tert-butyl aspartate and 4-nitrophenyl chloroformate (90%). Conversion of the resulting carbamate di-tert-butyl ester 8 to the diacid 9 was achieved by careful trifluoroacetic acid protonolysis (100%). The diacid was then quantitatively converted into anhydride 5 by dehydration with acetic anhydride.

The tight association of Cu(II) with the bis-imidazole intermediates 5–8 was indicated by: (i) spectral titration of aqueous CuCl₂ and 6 (1:1; 280 nm, 580 nm); (ii) the preparation and x-ray crystallographic determination of the CuCl₂ complex of 6 (Fig. 5) (the x-ray crystallographic data are published as supporting information on the PNAS web site, www.pnas.org); and (iii) the emerald green color rapidly produced on addition of CuCl₂ to the imidazolium salts 8 and 9 in DMSO.

Figure 5.

X-ray (ORTEP) structure of 6-CuCl₂.

The ability of 5 to covalently modify antibody 38C2 was then investigated in two orthogonal ways to probe whether the modification of 38C2 by 5 or 5-CuCl₂ might exhibit different profiles (Fig. 6). In method A, the antibody (20–30 μM) was first incubated with the cofactor ligand 5 (250 μM) in PBS (20 mM, pH 7.4) with 5% acetonitrile as cosolvent at room temperature; the reaction was typically complete after 1–3 h. Cu(II) insertion into the apo-38C2-5 conjugate was achieved by initial dialysis into aqueous NaCl (50 mM, pH 5.5) solution followed by addition of excess CuCl₂ (40 mM) in DMSO (5%), and subsequent dialysis into aqueous NaCl. Metal salt addition at lower pH was necessary to minimize precipitation of hydroxo-Cu(II) species.

Figure 6.

Orthogonal approaches to covalent modification of 38C2.

In method B, 5 was precoordinated with CuCl₂ (in DMSO), then the resulting blue-green 5-CuCl₂ complex was incubated with 38C2 in aqueous NaCl (3 h) and finally dialyzed in NaCl. The incorporation of Cu(II) into the 38C2-5 conjugate was confirmed by inductively coupled-plasma-MS.

The 38C2-5-CuCl₂ metalloantibody complex was then tested for its ability to catalyze either oxidation or acyl transfer reactions. No catalysis by 38C2-5-CuCl₂ was observed for the hydroxylation of a representative benzylic substrate (10) with either oxygen/ascorbate or t-BuOOH as the stoichiometric oxidant (Fig. 7A). However, significant and metal-dependent catalysis of hydrolysis of the coordinating picolinate ester 11 (52, 53) by 38C2-5-CuCl₂ in aqueous buffer [Mops (25 mM), pH 7.0] is observed (Fig. 7B).

Figure 7.

(A) Attempted benzylic oxidation of substrate 10 by 38C2-5-CuCl₂; (B) hydrolysis of picolinic acid ester 11 by 38C2-5-CuCl₂.

The catalytic hydrolysis of 11 by the 38C2-5-CuCl₂ complex follows Michaelis–Menten kinetics [k_cat(11) = 2.3 min⁻¹ and K_m(11) 2.2 mM). The uncatalyzed rate of hydrolysis of 11 under the conditions of our assay (in the absence of antibody and CuCl₂) is 1.1 × 10⁻⁵⋅min⁻¹. This gives an enhancement ratio [k_cat(11)k_uncat(11)] of 2.1 × 10⁵ and reveals the significant catalytic power of the 38C2-5-CuCl₂ conjugate. To determine the benefit of binding the Cu-bis-imidazolyl ligand into the antibody-binding site, the second-order rate constants of the modified 38C2 [k_cat(11)/K_m(11)], and the Cu(II)-bisimidazolyl ligand k(6-CuCl₂) = 0.03 M⁻¹⋅min⁻¹ catalyzed processes were compared. This gives a rate enhancement {[k_cat(11)/K_m(11)]/k(6-CuCl₂)} of 3.5 × 10⁴ and an effective molarity (EM) of [k_cat(11)/k(6-CuCl₂)] 76.7 M. This EM compares favorably with other antibody-catalyzed bimolecular processes (5, 6) and suggests that there is a significant benefit to binding the bisimidazolyl ligand into 38C2, and that the catalysis arises from sequestration of the substrate in proximity to the metal–ligand complex. The key role for copper in the hydrolysis of the picolinate ester substrate 11 is indicated by the nearly complete inhibition (>95%) of the 6-CuCl₂ and 38C2-5-CuCl₂ catalyzed reactions by addition of EDTA (3–5 μM).

In conclusion, we have shown that anhydrides are reactive and selective generic derivatizing agents for the Lys^H93 residue of the 38C2 antibody. In addition, we have successfully derivatized the antibody 38C2 with a biomimetic bis-imidazolyl metal-binding cofactor 5 and have demonstrated that the resultant copper-complexed antibody possesses significant and metal-dependent esterase activity. This is, in fact, the first such example, to our knowledge, of a copper-dependent catalytic antibody and replaces the original aldolase function of the protein with a new hydrolytic function. In addition to establishing the viability of our cofactor approach to elicit efficient catalytic metalloantibodies, these results bode well for the development of new semisynthetic metalloproteins that combine the unique selectivity of natural enzymes with the broader substrate specificity of synthetic catalysts.

Abbreviation

Mops

4-morpholinepropanesulfonic acid