Bloconjugate Chem. 1992, 3, 147-153
147
Site-Specific Modification of a Fragment of a Chimeric Monoclonal Antibody Using Reverse Proteolysis’ Igor Fisch,’ Gabriel Kiinzi, Keith Rose, and Robin E. Offord DBpartement de Biochimie MBdicale, Centre MBdical Universitaire, 1 rue Michel-Servet, 1211 Geneva 4, Switzerland. Received November 7, 1991 ~~
We propose a novel method for the site-specific labeling of antibodies under mild conditions and give as an example the modification of an F(ab’)2-like fragment of the chimeric monoclonal antibody B72.3. The F(ab’)n-like fragment was produced by the action of the protease lysyl endopeptidase. Reverse proteolysis, catalyzed by the same enzyme, was then used to attach carbohydrazide specifically to the carboxyl termini of the heavy chains of the fragment. Finally, a radiolabeled chelator possessing an aldehyde group was conjugated to the modified fragment through a hydrazone linkage. The resulting site-specifically labeled F(ab’)z-like fragment was characterized by gel electrophoresis and by enzymic digestion. It was found to possess immunoreactivity equivalent to that of the unmodified F(ab’)z-like fragment as determined by immunofluorescence and ELISA (enzyme-linked immunosorbent assay) techniques. The advantages and disadvantages of this labeling method, which appear to be of quite general applicability, are discussed.
INTRODUCTION Monoclonal antibodies usually require modification if they are to be used successfully for in vivo diagnosis (immunoscintigraphy) or for therapy. For immunoscintigraphy, monoclonal antibodies are labeled with y-emitting isotopes by coupling, for example, iodine-131 or iodine123 to tyrosine residues (Goldenberg et al., 1978; Moldofsky et al., 19841, or by coupling to lysine residues a chelator which binds a radiometal such as indium-111 (Rainsbury et al., 1983; Khaw et al., 1980; Scheinberg et al., 1982; Hnatowitch et al., 1983; Murray et al., 19851, or by the passive adsorption of species such as technetium99m (Morrison et al., 1984). For therapeutic applications, antibodies are coupled to radioactive isotopes (e.g. Epenetos et al., 1982;Order et al., 1980;Carrasquillo et al., 19841, cytotoxic drugs (e.g. Baldwin et al., 1982; Ghose & Blair, 1978; Hurwitz et al., 1979; Gallego et al., 1984), or toxins (e.g. Youle & Neville, 1980) in an attempt to target such cytotoxic substances to pathological sites. These modifications of an antibody generally involve covalent attachment of an agent either to the side chain of residues such as tyrosine, lysine, aspartic, and glutamic acid or to sulfhydryl groups, the latter being generated either by reduction of cystine residues or by reaction of the antibody with reagents such as 24minothiolane or N-succinimidyl 3-(pyridyldithio1)propionate (SPDP). All of the methods mentioned above for labeling antibodies have a major disadvantage: a heterogeneous product is generated with regard to the site of labeling. This disadvantage may be avoided by directing the modification to a specific site, ideally far from the antigen-binding site. Site-specific labeling of oxidized carbohydrate present on immunoglobulins of the IgG class has already been reported (Murayama et al., 1978) and exploited (e.g. Rodwell et al., 1986; Pochon et al., 1989). However, this latter labeling technique is clearly not applicable to polypeptides or proteins that do not carry a sugar moiety, as is often the Abbreviations used ELISA, enzyme-linked immunosorbent assay; SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis; KTI, kallikrein trypsin inhibitor; FITC, fluorescein isothiocyanate; BSA, bovine serum albumin; DMSO, dimethyl sulfoxide; PBS, phosphate-buffered saline. * Author to whom correspondence should be addressed. 1043-1802/92/2903-0147$03.00/0
case for F(ab’):! or Fab fragments of an antibody. In addition, for those IgG molecules that possess more than one glycosylated site, oxidation of carbohydrate will not be restricted to a single site. More recently, the specificity of proteases working in reverse has been used to attach, to the carboxyl terminus of a polypeptide chain, a chemical group which is not present in proteins and the reactivity of which can be used to conjugate other materials, for example chelated radiometals, cytotoxic drugs, or toxins. Thus, lysyl endopeptidase has been used to couple carbohydrazide to the carboxyl terminus of d e ~ - A l a ~ ~ ~ - i n (Rose s u l i net al., 1991). The modified insulin was then conjugated to the chelator HCO-m-C~H4CH=NOCH2CO-ferrioxamine through a hydrazone bond. The conditions for both the coupling and conjugation reactions are very mild indeed, which prompted us to use this potentially general site-specific method to label the carboxyl terminus of the heavy chains of a monoclonalantibody fragment. A preliminary account of some of our work in this field has already appeared (Fisch et al., 1991). The F(ab’)z fragment of an antibody is often a better carrier of radionuclides for immunoscintigraphy than is the intact antibody (Kurkela et al., 1988). Loss of the Fc region of the antibody frequently leads to decreased nonspecific binding (Lamoyi, 19861, and the more rapid kinetics of clearance of F(ab’)n fragments from healthy organs may increase the tumor to background ratio (Buchegger et al., 1983; Herlyn et al., 1983; Mach et al., 1985; Mather et al., 1987), both of which factors should improve the quality of images obtained during immunoscintigraphy. We chose to work with the F(ab’)z-likefragment derived from a chimeric form of the monoclonal antibody B72.3 (Whittle et al., 1984). The chimeric form, cB72.3, has a human IgG4 constant region (74) and binds to a glycoprotein complex with a molecular weight of about 220440 kDa named TAG-72 and present in many carcinomas. cB72.3 reacts with approximately 50% of human mammary carcinomas and with 80% of the colon carcinomas tested, but does not react appreciably with normal mammary tissue, with normal colon tissue, or with a variety of normal adult human tissues tested using immunohistochemical techniques (Nuti et al., 1982; Colcher et al., @ 1992 American Chemical Society
148
Bloconjugate Chem., Vol. 3, No. 2, 1992
1984). In the present study, we describe conditions for attaching a carbohydrazide group to the carboxyl terminus of the shortened heavy chains of the F(ab’)z-likefragment generated by lysyl endopeptidase digestion of the antibody. We also give conditions for the conjugation of a chelator to the carbohydrazide group of the modified F(ab’)z-like fragment. Preliminary work with a series of other antibodies (not described) shows that our two-step site-specific modification procedure is of fairly general applicability. EXPERIMENTAL PROCEDURES
Preparation of the Conjugate-Time-Course Studies. Proteolysis of cB72.3. To 0.86 mL of cB72.3 (19 mg/mL in 0.1 M ammonium bicarbonate) was added 16.3 pL of lysyl endopeptidase (freshly prepared as a solution 10mg/mL in water; from Achromobacter lyticus, WAKO Pure Chemical Co., Japan; final enzyme:substrate ratio of 1:100, w/w). After incubation at 37 OC for 8 h, the digest was cooled to 0 OC, at which temperature it may be stored for up to 96 h. Aliquots were removed for analysis by SDS-PAGE (sodium dodecyl sulfate polyacrylamide gel electrophoresis), under both reducing and nonreducing conditions (see below). Coupling of Carbohydrazide by Reverse Proteolysis. A 420-pL portion of the digest, now at a concentration of 18.6 mg/mL in 0.1 M ammonium bicarbonate, was added to 132.7 mg of solid carbohydrazide (Fluka, Buchs, Switzerland), giving a final concentration of 2.5 M carbohydrazide. The pH was lowered to 5.5 (uncorrected glass electrode) by addition of 12.5 pL of glacial acetic acid. Lysyl endopeptidase (80 pL of a freshly prepared solution 10 mg/mL in water) was added in order to obtain a final enzyme:substrate ratio of 1:lO (w/w). The sample was incubated at room temperature (23 “C). At various times (1, 2, 4, 8, and 24 h), an aliquot of 112 pL of the coupling reaction solution was removed and the reaction stopped by addition of 20.5 pL of kallikrein trypsin inhibitor (KTI or Trasylol, BAYER), freshly prepared as a 78 mg/mL solution in water (final enzyme:inhibitor ratio of l:lO, w/w). In a control series of experiments, the enzyme used for the coupling step was replaced by 80 pL of KTI at 10 mg/mL in water. At each time point, for experimental samples and for controls, carbohydrazide and enzyme were separated from the F(ab’)z-likefragment by gel filtration on a Superose-12 column (30 cm X 1 cm i.d. FPLC system, Pharmacia, Uppsala, Sweden) equilibrated and eluted with 0.1 M acetate buffer (counterion, Na+)at pH 4.6, at a flow rate of 0.4 mL/min. Effluent was monitored at 280 nm. The peak containing the protein fragment was collected in the presence of 20.5 pL of KTI (78 mg/mL in water), to avoid possible residual action on the coupled F(ab’)n-like fragment of any unseparated enzyme. The samples (experimentals and controls) were then concentrated at 4 OC in a centrifugal membrane concentrator having a molecular weight cutoff of 10 kDa (Centricon-10, AMICON). The concentrations of the experimental and control samples were estimated by UV absorption at 280 nm using an extinction coefficient of t1%2m= 14.75. This value is that of the intact cB72.3, the amino acid sequence of which is known (Whittle et al., 1984). The F(ab’)z-like fragment, which contains a similar proportion of Tyr and Trp residues to the intact antibody, will have a similar weight-based extinction coefficient.The concentration of the antibody fragment was thus overestimated due to the presence of any residual KTI (relative molecular mass, 6.5 kDa) which was not removed by the centrifugal membrane concentrator (cutoff 10 kDa), but this overestimation is of no consequence (see Results and
Fisch et ai.
Discussion). The gel filtration step which followed the conjugation reaction with HCO-m-CeH4CH=NOCHzCOferrioxamine (see below) permitted the proportion of remaining KTI to be established. Preparation of HCO-m-CsH.&H=NOCH&O-ferrioxamine. Desferrioxamine was loaded with iron-55 according to a published procedure (Prelog & Walser, 1962) but on a much smaller scale. The isotope was supplied by Amersham, England, as iron(II1) chloride in 0.1 M HC1 and was of low (1-50 mCi/mg) specific activity. The resulting ferrioxamine was converted to [(aminooxy)acetyl] ferrioxamine as previously described (Pochon et al., 1989). Conversion of the labeled compound to HCOm-C&CH=NOCHzCO-ferioxamine followed techniques already described for unlabeled material (Roseet al., 1991). The specific activity of the chelator was determined by optical absorption at 430 nm, using an extinction coefficient of tlM430 = 2650, and by the radioactivity, which was measured with a scintillant liquid (Pico-fluor 40, Packard, Canberra, Australia) in a liquid scintillation counter (Beckman, Model LS6800). The specific activity of the chelator was thus determined to be 8.2 X lo4 dpm/nmol of HCO-m-C6H4CH=NOCHzCO-ferrioxaminefor both the time-course studies and the characterization studies. Hydrazone Formation. For each time point, the carbohydrazide-coupled F(ab’)z-likefragment (at the nominal concentration estimated by UV absorption at 280 nm) was incubated in 0.1 M acetate buffer at pH 4.6, counterion Na+, with a nominal excess of 5 equiv (mol/mol) of the chelator HCO-m-CsH4CH=NOCH2CO-ferrioxamine labeled with iron-55. The chelator was added as a portion of a 3.1 mM stock solution in the acetate buffer. The control series was treated similarly. After 20 h at room temperature, the conjugated fragment was separated from residualKT1 and free labeled chelator on the Superose-12 gel filtration column equilibrated and eluted with 0.1 M ammonium bicarbonate buffer, at 0.8 mL/min. The effluent was monitored at 280 nm. Fractions containing the conjugated fragment were pooled; the final solution was analyzed by UV spectroscopy, and portions were counted for radioactivity. The optical density at 280 nm due to the conjugated HCO-m-CsH&H=NOCHzCO-ferrioxamine was then calculated from the UV spectrum of the model hydrazone (CH&COCONHNHCONHN=HCm-C6H4CH=NOCH2CO-ferrioxamine(dM2m = 26 500). Subtraction of this contribution to the absorbance at 280 nm, from the observed absorbance of the protein conjugate at this wavelength, permitted calculation of the protein concentration, and thus the conjugation ratio was expressed as moles of iron-55 per mole of F(ab’)z-like fragment. Characterizationof the Conjugate. Cleavage of the Chelator from the Protein. An 80-pLportion of the labeled F(ab’)z-like fragment prepared as described above, at a nominal concentration of 0.47 mg/mL (measured by UV absorption at 280 nm as before, in 0.1 M ammonium bicarbonate buffer) and a specific activity of 1.4 X 105 dpm/nmol, was incubated with 0.75 pL of lysyl endopeptidase (freshly prepared as a 1mg/mL solution in water). After 6 h at 37 “C,1pL of KTI (7.8 mg/mL in water) was added to stop the reaction. In a control experiment, the enzyme was replaced by addition of 0.75 pL of KTI (1 mg/mL in water). The F(ab’)n-like fragment was then separated from enzyme, inhibitor, and released chelator by gel filtration on the Superose-12 column equilibrated and eluted with 0.1 M ammonium bicarbonate buffer, at 0.8 mL/min. The effluent was monitored at 280 nm. Fractions containing the fragment were pooled; the final
Bloconlugate Chem., Vol. 3, No. 2, 1992 149
Antibody Conjugate by Reverse Proteolysis
solutionwas analyzed by UV spectroscopy and SDS-PAGE (see below), and portions were counted for radioactivity. In vitro Immunoreactivity Tests. ( 1 ) Cell Binding. LS174T cells (ATCC, Rockville, Maryland) were used for the immunofluorescence test as described (Nairn, 1976). Briefly, cells were removed from flasks with 0.1 % trypsin containing 5 mM EDTA and washed twice in 1 X PBS (Gibco, Brl) containing 0.2% BSA and 0.05% NaN3. A 1-mL portion of the suspension containing 1million cells was centrifuged at 1000 rpm for 5 min (Megafuge, Heraeus SA., Zurich, Switzerland) and the supernatant was removed. A 50-pL sample of iron-55-labeled F(ab’)2-like fragment (coupled 4 h with carbohydrazide and having a specific activity of 1.4 X lo5 dpm/nmol; 5 pg in PBS containing 0.2% BSA and 0.05% NaN3) was applied to the pellet and gently mixed, and the suspension was incubated for 1h a t room temperature (23 “C). Cells were centrifuged as described above and washed twice with 1 mL of PBS containing 0.2 % BSA and 0.05% NaN3. A 50-pL sample of FITC-coupled anti-human IgG (Sigma cat. no. F 0132,20-f01d diluted in PBS containing 0.2% BSA and 0.05% NaN3) was added to the pellet, gently mixed, and incubated for 1 h at room temperature (23 OC). Cells were centrifuged as described above, washed twice with 1mL of PBS containing 0.2 % BSA and 0.05 % NaN3, and observed with a fluorescence microscope (Dialux 20 ED, Leitz, Wetzlar GMBH, Germany). (2)Antigen Binding. Quantitation of the immunoreactivity using an enzyme-linked immunosorbent assay (ELISA) test was performed as already described (Engvall, 1980). SDS-PAGE. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was performed according to Laemmli (1970). The acry1amide:bisacrylamide ratio was 30:0.8, w/w. For nonreducing conditions, the concentration of this mixture was 8% (w/w),and for reducing conditions it was 12% (w/w). RESULTS AND DISCUSSION
Lysyl Endopeptidase Digestionof cB72.3. Lysyl endopeptidaseis a protease which cleavesa polypeptidechain specifically at the carboxyl side of lysine residues (Masaki et al., 1981). A time-courseof the digestion of cB72.3 with lysyl endopeptidasewas followed by SDS-PAGE analysis, but only the 8-h digestion is shown. Under nonreducing conditions (Figure lA), the intact cB72.3 (lane 3) shows a major band whose mobility corresponds to a relative molecular mass of 150 kDa and also some contaminants at about 130 and 70 kDa. The band whose mobility corresponds to a relative molecular mass of 150 kDa (intact antibody, lane 3) disappeared after 8 h of digestion and a new band was formed (lane 4), whose relative molecular mass (110 kDa) is similar to that of a peptic F(ab’)2 fragment. The intensity of the band at 110 kDa in lane 4 (the digest) is lower than observed in lanes 5-8 (fractions from gel filtration). This is due to the residual activity of the protease, which was still present when the digest was loaded onto the gel. Under reducing conditions (Figure lB), the band with a relative molecular mass of 50 kDa (heavy chain) in the undigested cB72.3 (Figure lB, lane 2) was absent after 8 h of digestion and a new band whose mobility corresponds to 28 kDa appeared (lane 3). In contrast, the band with a relative molecular mass of 26 kDa (light chain) in the undigested cB72.3 (lane 2) remained intact after 8 h of digestion (lane 3). As discussed above, the relative intensities of the bands in lane 3 (the digest) are lower than those in lanes 4-7 (fractions from gel filtration) due
1
2
3
4
5
7 8 9 1 0
6
- 94 hD.9
- 60
1
2
3
4
5
6
7
8
rDa 94-
67
-
54
-
43
-
30
20
-
-
6 7 LDa
60
-
36
-
28
-
26
-
Figure 1. (A) SDS-PAGE under nonreducing conditions of the coupling reaction, stained with Coomassie Blue R250: lanes 1, 2,9, and 10, molecular weight markers; lane 3, intact antibody cB72.3; lane 4, cB72.3 digested with 1%(w/w) lysyl endopeptidase for 8 h a t 37 “C;lanes 5 and 7, F(ab’)~-likefragment coupled with carbohydrazide for 4 and 24 h, respectively; lanes 6 and 8, control samples of F(ab’)z-like fragment after 4 and 24 h, respectively, in coupling conditions but without lysyl endopeptidase. (B) SDS-PAGE under reducing conditions [5% (v/v) B-mercaptoethanol] of the coupling reaction, stained with Coomassie Blue R250: lanes 1 and 8, the molecular weight markers; lane 2, intact antibody cB72.3; lane 3, cB72.3 digested with 1% (w/w) lysyl endopeptidase for 8 h at 37 “C;lanes 4 and 6, F(ab’)e-like fragment coupled with carbohydrazide for 4 and 24 h, respectively; lanes 5 and 7, control samples of the F(ab’)2like fragment after 4 and 24 h, respectively, in coupling conditions but without lysyl endopeptidase.
to the presence of protease. When protease is removed by gel filtration, no bands with lower relative molecular mass appeared (lanes 4-7), indicating the absence of nicks on the light chains of the digested cB72.3. Thus, SDS-PAGE data show that lysyl endopeptidase produces, very efficiently, a F(ab’)2-sized fragments from cB72.3. Examination of the sequence of the heavy chain (Whittle et al., 1984) in the CH2 domain provides only two possible sites and Lys242. of cleavage. These are Reverse Proteolytic Coupling of Carbohydrazide to the F(ab’)2-likeFragment of cB72.3. The coupling of carbohydrazide by reverse proteolysis to the carboxyl terminus of a polypeptide chain (insulin) has already been described (Rose et al., 1991), and it was shown that the coupling yield was increased in the presence of an organic cosolvent. In 50% DMSO (dimethyl sulfoxide), yields up to 95% were obtained, whereas in the absence of DMSO, the coupling yields approached only 84 % . In the present
Fisch et ai.
150 B i o c o n j ~ t eChem., Vol. 3, No. 2, 1992 d
I
0.6
0.5
0.4
a 0.3
0.2
k 4
8
12
16
0.1
20
I
Figure 2. Gel filtration profile of the coupling reaction between a F(ab’)z-likefragment and carbohydrazide,after 4 h. Peak a, at 6.8 mL, is probably due to aggregated material. Peak b, at 12 mL, is the F(ab’)z-like fragment coupled with carbohydrazide. Peak c, at 13.5 mL, is the enzyme-inhibitor complex, and peak d, at 16 mL, the excess of inhibitor (KTI). study, preliminary attempts to couple carbohydrazide in 50% DMSO led to precipitation of the F(ab’)z-like fragment, and so all subsequent work was performed in fully aqueous solution, which in any case would be preferable from the point of view of immunoreactivity preservation. Given that no organic cosolvent,which tends to suppress further proteolytic cleavage, was used, and given the large quantity of enzyme added for the coupling step, SDS-PAGE analysis was performed to test for any degradation of the F(ab’)z-like fragment during the coupling reaction. Under nonreducing (Figure 1A) and reducing conditions (Figure lB), no evidence for the production of lower molecular weight material was found, even after 24 h (Figure lA, lane 7, and Figure lB, lane 6). Figure 2 shows the gel filtration profile of the coupling reaction after 4 h of incubation with carbohydrazide. The coupled F(ab’)z-like fragment is seen to elute separate from enzyme-inhibitor complex,inhibitor, and carbohydrazide. Hydrazone Formation. The degree of incorporation of carbohydrazide into the F(ab’)z-like fragment cannot be directly measured, but is obtained by conjugation with iron-55-labeledHCO-m-C6H4CN=NOCHaCO-ferrioxamine through a hydrazone linkage. The stability of hydrazone linkages has been discussed by King et al. (1986), who showed that 98% of a model hydrazone (N-acetylhydrazone of p-carboxybenzaldehyde) remained intact at pH 8.0, even after 72 h of incubation at 25 O C . The specificity of such conjugation reactions for a hydrazo (rather than side-chain amino) group has already been demonstrated in previous work (Rose et al., 1991; King et al., 1986). Figure 3 shows the gel filtration profile of the conjugation reaction between the F(ab’)z-like fragment (coupled with carbohydrazide for 4 h) and iron-55-labeled HCO-m-C6H&!H=NOCH&O-ferrioxamine. An excel-
I 16
8
retention volume (ml)
I 24
retention volume (ml)
Figure 3. Gel filtration profile of the conjugation reaction between carbohydrazide-coupled F(ab’)z-likefragment and HCOm-CsH4CH=NOCHzCO-ferrioxamineafter 20 h. Peak a, at 12 mL, represents the conjugated F(ab’)z-likefragment, and peak b, at 16 mL, the inhibitor (KTI). Table I. Hydrazone Formation
sample” E1
Ez E4 E8 E24 c1
C2 c4
CS c24
nominal concentration) mg/mL 4.4 3.0 4.0
4.0 2.9 3.3 3.2 2.5
3.7 3.7
nmol coupled‘ 5 3.3 4 5 4.4 5 5 5 5 5
specific activity of the fragment, dpm/nmol of conjugate x 10-4 5.5 12 14 9.5 16 0.3 0.9 0.8 2.7 4.8
“E, and C, represent the experimental and control samples, respectively, after n hours of coupling with carbohydrazide. * Nom-
inal concentration after the coupling step with carbohydrazide, overestimated due to the presence of residual KTI. Represente the quantity, based on the nominal concentrations, of F(ab’)z-like fragment (which had been incubated with carbohydrazide) which was conjugated with iron-55-labeled HCO-m-C~H4CH=NOCHzCOferrioxamine.
lent separation of protein and small molecules is achieved. The KTI which was not removed during the membrane concentration step is clearly visible (Figure 3) and may be quantitated. Correction of the nominal fragment concentration (seeTable I) for the contribution due to residual KTI permitted calculation of the actual ratios of reagent to fragment used for the conjugation reaction. This ratio, found to be between 8.3 and 16.1, was always in excess of the nominal 5-fold excess of reagent, thus the initial overestimation of fragment concentration is without consequence because reaction is essentially quantitative when using this reagent (Rose et al., 1991). Using the specific activity of the reagent (8.2 X lo4dpmhmol) and
Antlbody Conjugate by Reverse Proteolysis
Bioconjugate Chem., Vol. 3, No. 2, 1992
f
0.5
09 0
5
IO
15
20
25
Time (h) of coupling with carbohydrazide
Figure 4. Time-course of the enzymatic coupling reaction: the incorporation (expressed in moles of iron-55-labeled HCO-mCeH&H=NOCH&O-ferrioxamine/mole of F(ab’)Tlikefragment of the cB72.3) is shown as a function of the time of coupling with carbohydrazide. Filled circles represent the experimental Samples, and open circles the control samples.
radioactivity counting of a portion of the protein peak, the concentration of the conjugated label was calculated. The results of the conjugation reaction (hydrazone formation) are given in Table I as measured specific activities and are shown graphically in Figure 4 as calculated conjugation ratios. After 4 h of coupling with carbohydrazide, the resulting coupled F(ab’)n-like fragment incorporates about 1.7 mol of iron-55-labeled HCO-m-C6H4CH=NOCH&O-ferrioxamine/molof F(ab9rlike fragment. The control experiment, where during the reverse proteolysis coupling steps lysyl endopeptidase was replaced by a protease inhibitor (KTI or Trasylol), showed after 4 h of incubation with carbohydrazide an incorporation of 0.1 mol of labeled HCO-m-CsH4CH=NOCH2CO-ferrioxamine/mol of F(ab’)z-like fragment. Figure 4 shows, for the experimental sample, a rapid increase of the incorporation of labeled HCO-m-C& CH=NOCH&O-ferrioxamine with the duration of the coupling step. The value of about 1.7 mol of HCO-mC&I4CH=NOCHzCO-ferrioxamine per mol of F(ab’)z-like fragment, attained after 4 h, is close to that expected from model experiments with insulin under similar conditions (Rose et al., 1991),where, in the absence of DMSO, about 0.7 mol of carbohydrazide was incorporated per mole of carboxyl terminal Lys. As hydrazone formation is almost quantitative when HCO-m-CsH4CH=NOCH2CO-ferrioxamine is used as the reagent (Rose et al., 1991; Fisch et al., 19911, and as the F(ab’)z-like fragment possesses two heavy chains to which carbohydrazide may be coupled by the protease, an incorporation of 1.7 mol of HCO-m-C6H4CH=NOCH&O-ferrioxamine/mol of F(ab’)n-like fragments corresponds to about 0.85 mol of carbohydrazide at the C-terminus of each heavy chain. If coupling to each heavy chain is an independent event, then the value of 1.7 mol/mol corresponds (binomial distribution) to 72 5% of the F(ab’)s-like fragment with carbohydrazide coupled to the C-terminus of both heavy chains, 26% with carbohydrazide coupled to the C-terminus of a single chain, and 2% with no carbohydrazide at all. The control experiment (Figure 4) was performed in order to estimate any nonspecific incorporation of carbohydrazide. Values of apparent incorporation found at coupling times of 4 h or less show that, for the F(ab’)z-like fragment of cB72.3, such nonspecific incorporation is a minor process. One mechanism by which carbohydrazide might be incorporated, other than enzymatically to the F(ab’)z-like fragment, is through deamidation of Asn or Gln accompanied by nucleophilicattack by carbohydrazide (which is present at very high concentrations, see Experimental Procedures) rather than by a water molecule. An
151
alternative interpretation of the control curve of Figure 4 would be the presence of residual active enzyme from the digest, which was not totally inhibited by addition of KTI. Given the relatively low values found for the control at coupling times of 4 h and less, we did not investigate this effect further. The apparent fall in conjugation ratio at 8 h (Figure 4) was consistently observed, and we have no explanation. The optimum coupling time is about 4 h and the overall yield (moles of labeled F(ab’)z-like fragment compared to moles of intact antibody taken) of the preparation of the conjugated fragment was between 30 and 45%. Cleavage of the Labeled Fragment. To determine whether the label is in the expected position, i.e. at the carboxyl terminus of the heavy chains, the iron-55conjugated F(ab’)z-like fragment was digested with the same enzyme used in the coupling step (lysyl endopeptidase). It was expected that the enzyme, used at pH 8.0 and in the absence of carbohydrazide, would cleave the bond which had been formed at pH 5.5 in the presence of 2.5 M carbohydrazide. Such cleavage would release the label from the carboxyl terminus of antibody fragment, without degrading the latter. After 6 h of digestion at 37 “C, analysis by gel filtration showed that the specific radioactivity of the antibody fragment had been reduced from 1.4 X lo5 to 8.6 X lo3 dpm/nmol, Le. to 6% of its original value. The mobility of the protein fragment on reducing and nonreducing SDS-PAGE gels remained unaffected (data not shown). In a control redigestion experiment, the enzyme was replaced by the inhibitor KTI. Under these conditions, no diminution of specific radioactivity of the conjugated F(ab’)n-like fragment was observed (1.4 X lo5 dpm/nmol recovered). These results indicate that the lysyl endopeptidase had coupled carbohydrazide specifically at the carboxyl termini and was then able to cleave the Lys-carbohydrazide bond of the conjugated fragment. Since lysyl endopeptidase was shown to act on the heavy chains (Figure 1)and since a degree of incorporation was found which is similar to that obtained with the insulin model (Rose et al., 1991), it is most likely that the modification is mainly restricted to the truncated heavy chains in the case of cB72.3. Indeed, autoradiography of a reducing SDS-PAGE gel to which the fragment conjugated to chelator labeled with 6’Ga had been applied showed labeling of the heavy chain alone (data not shown). Taking into account the uniform background of the exposed film, we were able to estimate that labeling of the light chains, if it occursat all, represents less than 10% of that of the heavy chains. Furthermore, examination of the amino acid sequence (Whittle et al., 1984) shows the C-terminus residue of the light chain to be Cys, which is not recognized by lysyl endopeptidase. In the case of a fragment possessing Lys at the carboxylterminal residue of the light chain, incorporation of carbohydrazide would be expected to occur there also. This would lead to an even higher conjugation ratio, yet the modification, by being restricted to the carboxyl termini, would be remote from the antigen-binding region and would not be expected to interfere with it. In Vitro Immunoreactivity Tests. ELISA and immunofluorescence tests were performed to investigate the effect, on the immunoreactivity of the F(ab’)z-like fragment, of coupling to carbohydrazide and then conjugating a labeled aldehyde chelator to the carboxyl terminus. Results show (Figure 5 ) that the affinity for the antigen of the conjugated fragment remained essentially unchanged. Immunofluorescence testa showed that both the coupled and the conjugated fragment were able to bind
152 BlOconjUgete Chem., Vol. 3, No. 2, 1992 A492m I
Flsch et al.
Ltd., Slough, U.K., for the gift of antibodies and for financial support. Igor Fisch thanks the Ligue Suisse contre le Cancer for a bursary.
1
LITERATURE CITED
6510
2604
1042
417
167
67
21
I t
Antibody fragment concentration (nglml)
Figure 5. ELISA of unmodified F(ab’)z-like fragment (filled circles) and the fragment conjugated with iron-55-labeled HCOm-C6H4CH=NOCHzCO-ferrioxamine(open circles). to the cell-surface antigen expressed by the cell line. Similar results (not shown) were obtained with the galliumcontaining chelator derivative (prepared according to Rose et al., 1991). Gallium-67 is a y-emitting isotope useful in radioimmunoscintigraphy which is tightly held by the desferrioxamine chelator in vivo (e.g. Pochon et al., 1989). As the binding of the conjugated F(ab’)z-like fragment for ita antigen remained essentially unchanged, it will be interesting to test such a specifically modified antibody fragment in in vivo systems. Experiments with the conjugate labeled with gallium-67 show that targeting to tumor tissue occurs in a mouse xenograft model (data not shown). Resulta of these experiments, includingfull details of the procedures used to label the chelator with small quantities of gallium-67 of high specific activity, will be published elsewhere. CONCLUSION
Lysyl endopeptidase may be used to obtain an F(ab’)nlike fragment of the chimeric monoclonal antibody cB72.3. Reverse proteolysis may be used to attach carbohydrazide to the carboxyl terminus of the heavy chains of the F(ab’)zlike fragment. Conjugation of HCO-m-CsH&H=NOCH2CO-ferrioxamine labeled with iron-55 (and with gallium67) to the carbohydrazide-modified F(ab’)z-likefragment did not decrease its capacity to bind antigen. The two-step approach to protein modification described here offers the following advantages: (1) site-specific modification, giving a relatively homogeneous product, (2) carboxyl-terminal modification, far from the antigenbinding region of the antibody, (3) very mild conditions, (4) production of an intermediate of general application (coupled fragment) to which a wide variety of molecules may be attached, provided that they carry or can be made to carry an aldehyde or a keto function (Rose et al., 1991). Our own preliminary experiments with a series of monoclonal antibodies show that the two-step approach is of fairly general applicability. Different enzymes (trypsin for example) have been employed and conjugated Fab fragments have also been produced (data not shown). Further experiments with other monoclonal antibodies and tumor cell lines will be necessary to determine whether the site-specific labeling of an antibody fragment has more than just a theoretical advantage over the techniques currently used for labeling antibodies. ACKNOWLEDGMENT
We thank Brigitte Dufour for expert technical assistance and Gwynfor Davies for helpful advice during the preparation of the manuscript. This work was supported by the Fonds National Suisse de la Recherche Scientifique and the Schmidheiny Foundation. We thank Celltech,
Baldwin, R. W., Embleton, M. J., and Pimm, M. W. (1982) Monoclonal antibodies for radioimmunodetection of tumors and for targeting. Bull. Cancer 70, 132-136. Buchegger, F., Haskell, C. M., Schreyer,M., Scazziga, B. R.,Randin, S., Carrel, S., and Mach, J. P. (1983) Radiolabeled fragments of monoclonal antibodies against carcinoembryogenic antigen for localisation of human colon carcinoma grafted into nude mice. J. Exp. Med. 158, 413-427. Carrasquillo, J. A., Krohn,K. A., Beaumier,P., McGuffin,R. W., Brown, J. P., Hellstrom, K. E., and Larson, S. M. (1984) Diagnosis of and therapy for solid tumors with radiolabeled antibodies and immune fragments. Cancer Treat. Rep. 68, 317-328. Colcher, D., Keenan, A. M., Larson, S. M., and Schlom, J. (1984) Prolonged binding of a radiolabeled monoclonal antibody (B72.3) and for the in situ radioimmunodetection of human colon carcinoma xenografts. Cancer Res. 44, 5744-5751. Engvall, E. (1980) Enzyme immunoassay ELISA and EMIT. Methods Enzymol. 70, 419-439. Epenetos, A. A., Britton, K. E., Mather, S., Shepherd, J., Granowska, M., Taylor-Papadimitriou,J.,Nimmon, C. C., Durbin, H., Hawkins, L. R., Malpas, J. S., and Bodmer, W. F. (1982) Targeting of iodine-123-labelled tumor associated monoclonal antibodies to ovarian, breast, and gastrointestinal tumours. Lancet 2,999-1005. Fisch, I., Pochon, S., Werlen, R., Jones, R. M. L., Rose, K., and Offord, R. E. (1991) Site-specific modification of antibodies by enzyme-assisted reverse proteolysis. Peptides 1990 (E. Giralt, and D. Andreu, Eds.) pp 819-821, ESCOM, Leiden, The Netherlands. Gallego, J., Price, M. R., and Baldwin, R. W. (1984) Preparation of four daunomycin-monoclonal antibody 791T/36 conjugated with antitumor activity. Int. J. Cancer 33, 737-744. Ghose, T., and Blair, A. H. (1978) Antibody-linked cytotoxic agents in the treatment of cancer: current status and future prospects. J. Natl. Cancer Inst. 61, 657-676. Goldenberg, D. M., DeLand, F., Kim, E. E., Benett, S. S., Primus, F. J., VanNagell, J. R., Jr., Ester, N., DeSimone, P., and Rayburn, P. (1978) Use of radiolabeled antibodies to carciniembryogenic antigen for the detection and localization of diverse cancers by external photoscanning. N . Engl. J. Med. 297, 1384-1986. Halpern, S. E., Hagan, P. L., Garver, P. R., Koziol, J. A., Chen, A. W. N., Frincke, J. N., Bartholomew, R. M., Davies, J. S., and AdamsT. H. (1983)Stability,characterization, and kinetics of lllIn-labeled monoclonal antibodies in normal animals and nude mouse-human tumor models. Cancer Res. 43, 53475355. Herlyn, D., Powe, J., Alair, A., Mattis, J. A,, Herlyn, M., Ernst, C., Vaum, R., and Koprowski, H. (1983) Radioimmunodetection of human tumor xenografts by monoclonal antibodies. Cancer Res. 43, 2731-2735. Hnatowich, D. J., Layne, W. W., Childs, R. L., Lanteigne, D., Davies, M. A,, Griffin, T. W., and Doherty, P. W. (1983) Radioactive labeling of antibody: a simple and efficient method. Science 220, 613-615. Hurwitz, E., Schecter, B., Arnon, R., and Sela, M. (1979) Binding of antitumor-immunoglobulins and their daunomycin conjugates to the tumor and its metastates. In vitro and in vivo studies with Lewis lung carcinoma. Int. J. Cancer 24, 461470. Khaw, B. A., Fallon, J. T., Strauss, H. W., and Haber, E. (1980) Myocardial infarct imaging of antibodies to canine cardiac myosine with indium-111-diethylenetriamine pentaacetic acid. Science 209, 295-297. King, T. P., Zhao, S. W., and Lam, T. (1986) Preparation of a protein conjugates via intermolecular hydrazone linkage. Biochemistry 25, 5774-5779.
Antibody Conjugate by Reverse Proteolysis
Kurkela, R., Vuolas, L., and Vihko, P. (1988) Preparation of F(ab’)z fragments from monoclonal mouse IgGl suitable for use in radioimaging. J. Immunol. Methods 110, 229-236. Laemmli, U. K. (170) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227,680685. Lamoyi, E. (1986) Preparation of F(ab’)z fragments from mouse IgG of various subclasses. Methods Enzymol. 121, 652-657. Mach, J. P., Buchegger,F., Grob, J. Ph., Van Flieduer, V., Carrel, S., Barreth, L., Bishop-Delaloye,A., and Delaloye, B. (1985) In Monoclonal Antibodies for Cancer Detection and Therapy (R.W. Raldwin, and V. S. Bryers, Eds.), pp 53-65, Academic Press, London. Masaki, T., Fujihashi, T., Nakamura, K., and Soejima, M. (1981) Studies on a new proteolytic enzyme from Achromobacter lyticus M497-1. 11. Specificity and inhibition studies of Achromobacter protease I. Biochim. Biophys. Acta 660 (l),51-55. Mather, S. J., Durbin H., and Taylor-Papadinitriou, J. (1987) Identification of immunoreactive monoclonal antibody fragment for improved immunoscintigraphy. J. Immunol. Methods 96,255-264. Moldofsky, P. J., Sears, H. F., Mulhearn, C. B., Jr., and Hammond, N. D. (1984) Detection of metastatic tumor in normalsized retroperitoneal lymph nodes by monoclonal-antibody imaging. N . Engl. J . Med. 311,106-107. Morrison, R. T.,Lyster,D. M., Alcorn, L., Rhodes,B. A.,Breslow, K., and Burchiel, S. W. (1984) Radioimmunoimaging with -Tc monoclonal antibodies: Clinical Studies. Int. J. N u l . Med. Biol. 11, 184-188. Murayama, A., Shimada, K., and Yamamoto, T. (1978) Modification of immunoglobin G using specific reactivity of sugar moiety. Immunochemistry 15, 523-528. Murray, J. L., Rosenblum, M. G., Sobol, R. E., Bartholomew, R. M., Plager, C. E., Haynie, T. P., Jahns, M. F., Glenn, H. J., Lamki, L., Benjamin, R. S., Papadopoulos, N., Boddie, A. W., Frincke, J. M., David, G. S., Carlo, D. J., and Hersch, E. M. (1985) Radioimmunoimaging in malignant melanoma with 11lIn-labeledmonoclonalantibody 96.5. Cancer Res. 45,23762381. Nairn, R. C. (1976) in Fluorescent Protein Tracing, 4th ed., Churchill Livingstone, Edinburgh.
Bioconjugate Chem., Vol. 3, No. 2, 1992
153
Nuti, M., Teramoto, Y. A., Mariani-Costantini,R., Horan-Hand, P., Colcher,D., and Schlom,J.A. (1982)A monoclonalantibody (B72.3) defines pattern of distribution of a novel tumorassociated antigen in human mammary carcinoma cell population. Znt. J . Cancer 29, 539-545. Order, S. E., Klein, J. L., Ettinger, D., Alderson, P., Siegelman, S., and Leichner, P. (1980) Use of isotopic immunoglobulin in therapy. Cancer Res. 40, 703-710. Pochon, S., Buchegger, F., PBlegrin, A., Mach, J. P., Offord, R. E., Ryser, J. E., and Rose, K. (1989) A novel derivative of the chelon desferrioxamine for site-specific conjugation to antibodies. Znt. J . Cancer 43, 1188-1194. Prelog, V., and Walser, A. (1962) Helu. Chim. Acta 45,631-637. Rainsbury, R. M., Westwood, J. H., Coombes, R. C., Neville, A. M., Oh, R. J., Kalirai, T. S., McCready, V. R., and Gazet, J. C. (1983) Location of metastatic breast carcinoma by a monoclonal antibody chelate labelled with indium-111. Lancet 2, 934-938. Rodwell, J. D., Alvarez, V. L., Lee, C., Lopez, A. D., Goers, J. W. F., King, H. D., Powsner, H. J., and McKearn, T. J. (1986) Site-specific modification of monoclonal antibodies: in vitro and in vivo evaluations. Proc. Natl. Acad. Sci. U.S.A. 83, 2632-2636. Rose, K., Vilaseca,L. A., Werlen, R., Meunier, A., Fisch, I., Jones, R. M. L., and Offord,R. E. (1991)Preparation of a well-defined protein conjugates using enzyme-assisted reverse proteolysis. Bioconjugate Chem. 2, 154-159. Scheinberg, D. A., Strand, M., and Gansow, 0. A. (1982) Tumor imaging with radioactive metal chelates conjugated to monoclonal antibodies. Science 215, 1511-1513. Whittle, N., Adair, J., Lloyd, C., Jenkins, L., Schlom, J., Raubitschek, A., Colcher, D., and Bodmer, M. (1984) Expression in COS cells of amouse-human chimaeric B72.3 antibody. Protein Eng. 1 (6), 499-505. Youle, R. J., and Neville, D. M. (1980) Anti-thy 1.2 monoclonal antibody linked to ricin is a potent cell-type-specific toxin. Proc. Natl. Acad. Sci. U.S.A. 77, 5483-5486. Registry No. Lysyl endopeptidase, 78642-25-8; carbohydrazide, 497-18-7; HCO-m-CeH&H=N-O-CH&O-ferrioxamine, 138984-27-7.
