Bloconjugate Chem. 1993, 4, 515-520
515
Protein Conjugates of Defined Structure: Synthesis and Use of a New Carrier Molecule L. Antonio VilasecaJ Keith Rose,*J Raymond Werlen,? Anne Meunier,+ Robin E. Offord,+ Cynthia L. Nichols,$ and William L. Scott* Biochimie MBdicale, Centre MBdical Universitaire, CH 1211 Geneva 4, Switzerland, and Lilly Research Laboratories, Indianapolis, Indiana 46285. Received June 15, 1993"
A new carrier molecule, NH20CH&O-(Gly)3- [Lys(H-Ser-)Is-Gly-OH,has been synthesized to facilitate the preparation of protein conjugates of defined structure. Special features are as follows: (i) (aminooxy)acetyl as a terminal group, which reacts specifically to form an oxime bond under very mild conditions with an aldehyde group placed on a protein in a prior step; (ii) a spacer group of three Gly residues; and (iii) a set of five Lys residues, each of which is acylated with a Ser residue. A second form of the carrier molecule, HCO-m-C&&H=NOCHzCO-(Gly)3- [Lys(H-Ser)ls-Gly-OH,was also prepared. This form possesses a terminal aldehyde group which permits site-specific attachment by formation of a hydrazone bond to the carboxyl termini of polypeptide chains which have been modified enzymatically with carbohydrazide in a prior step. Once the carrier is linked to protein in one of the above ways, i.e. through formation of either an oxime or hydrazone bond, the Ser residues of the carrier (but not of the protein) may be oxidized by very mild periodate treatment to generate aldehyde groups. Drugs possessing a hydrazide group (e.g. methotrexate y-hydrazide or desacetylvincaleukoblastinehydrazide) may then be conjugated via hydrazone formation to the aldehyde groups of the carrier. A cluster of five drug molecules may thus be attached to a single site on a protein, giving a relatively homogeneous product in spite of the high drug conjugation ratio. Synthesis of the carrier, formation of a pentadrug-protein conjugate, and wider implications of the chemistry are presented.
INTRODUCTION In recent years it has become possible to modify the structure of unprotected polypeptides and proteins site specifically and under mild aqueous conditions. For example, oxidation of an N-terminal Ser or Thr residue to a glyoxylyl residue permits coupling reactions to be directed exclusively to this position in a second step (1, 2); reverse proteolysis may be used to place a hydrazide (or other group having unique reactivity) specifically at the C-terminus of a polypeptide chain, further reaction is then directed to that position (3,4),and large unprotected synthetic fragments may be brought together using sitespecific chemistry (5). Such site-specific operations lead to a more homogeneous product than is obtained by random attack on side chains, and a homogeneous product of defined structure is a great advantage if the conjugate is to be used for medical purposes: registration and quality control should be facilitated, modification of production procedures should become less problematic since it can be shown directly whether the product has been affected, and interpretation of biological results becomes more reliable than is possible when dealing with a mixture of regioisomers. However, if several molecules of interest (reporter group, drug, chelator, etc.) must be attached to each protein molecule and the advantages of single-site attachment to protein are to be preserved, then it is necessary to construct a carrier molecule. The carbohydrate moiety of a glycoprotein constitutes a natural carrier, oxidation of which permits attachment of various molecules to the aldehyde groups generated (6). Unfortunately, glycosylation may be heterogeneous or absent entirely from the protein of interest, or the harsh oxidation conditions Centre MBdical Universitaire. t Lilly Research Laboratories. +
e Abstract published in Advance ACS Abstracts, October 1, 1993.
1043-1002/93/2904-0515$04.00/0
may adversely affect protein function. Human serum albumin has been used as a carrier between drug molecules and an antibody (reviewed by Pietersz (7)),but linkage of drug to carrier and of carrier to antibody was not sitespecific with respect to the antibody. In order to improve product homogeneity, several other types of carrier molecules have been developed. Polyaldehyde dextran, amino dextran, and poly(amino acids) have mostly been used (7).Some of these polymers offer a uniquely reactive terminus for univalent attachment to biomolecules (e.g. refs 8-10). So far, use of even these monovalent-attachment carrier molecules has been a source of product heterogeneity for some (or all) of the following reasons: the polymeric carrier is not monodisperse, attachment of drug to carrier is not quantitative, and attachment of carrier to protein does not proceed in a site-specific manner. We are interested in the preparation of cytotoxic antitumor immunoconjugates for medical purposes. In an attempt to improve homogeneity of the final bioconjugate, we decided to synthesize a carrier of defined structure, thus removing the polydispersity problem. We chose carbonyl chemistry to link carrier to protein, since this is very mild, does not disturb disulfide bonds, and techniques exist which permit site-specific attachment either to the N-terminus (1,2)or to the C-terminus ( 3 , 4 ) of the protein. Drug release from final product was designed to occur through spontaneous hydrolysis of a hydrazone drug-carrier linkage, either upon internalization or within the tumor microenvironment (11). The carrier molecule was used to form a pentadrug conjugate of defined chemical structure with de~-Ala~3~-insulin (DAI)1 as a Abbreviations: DAI, des-AlaBmporcine insulin; HPLC, high pressure liquid chromatography; ESMS, electrospray ionization mass spectrometry; TFA, trifluoroacetic acid; DMSO, dimethylsulfoxide; DMF, dimethylformamide; Fmoc, fluorenylmethyloxycarbonyl; DAVLB, des-acetyl vincaleukoblastine. 0 1993 American Chemical Society
510 Biocon]~.gateChem., Vol. 4, No. 6, 1993
model protein. DAI was used in order to be able to follow precisely, by reversed-phase HPLC and ESMS, the various steps involved. A brief description of this work was presented a t the 1992European Peptide Symposium (12). Experiments with antibodies will be described in detail elsewhere. EXPERIMENTAL PROCEDURES Unless otherwise specified, all solvents and reagents were obtained from commercial sources, were of analytical or higher grade, and were used without further purification. Temperature was room temperature (about 22 "C) except where indicated. High-pressureLiquid Chromatography. Analytical reversed-phase HPLC was performed using a 250 mm x 4 mm i.d. Nucleosil300-A 5-pm Ca column; solvent A was 0.1 7% TFA; solvent B was 0.1 % TFA in 90 % acetonitrile prepared as previously described (3). Except where otherwise noted, the conditions were as follows: flow rate, 0.6 mL/min; 5 min isocratic a t 100% solvent A followed by a linear gradient of 2% solvent B/min to 100 % B. The analytical column was also used for small semipreparative separations. Where mentioned, preparative work was carried out using a column 250 mm X 21 mm i.d., packed with the same material and eluted a t a flow rate of 10 mL/min. Components were collected manually a t the detector outlet, frozen and lyophilized in a vacuum centrifuge (SpeedVac, Savant) without heating. Mass Spectrometry. FABMS was performed in positive ion mode as previously described (15). For FABMS, masses found and calculated are the monoisotopic values of the protonated species, M + H+. ESMS was performed in positive ion mode on a Trio 2000 machine (VG BioTech, Altrincham, England) equipped with a 3000 amu RF generator. Samples were introduced a t 2 pL/min either directly (if in the TFA/acetonitrile HPLC solvent) or (if dried) in solution in water/MeOH/AcOH (49.5:49.5:1, by volume). For ESMS, masses found and calculated are average values based on the isotopic envelope and are expressed as molecular weights of uncharged species. Where multiple protonation led to a series of signals, the neutral mass is given f the standard deviation of the measurements. Preparation of H-Glys-[Lys ( t-Boc-Ser(Bzl) )]5-G1yOCH2-PAM Resin. The resin-bound protected nonapeptide Fmoc-Glya-[Lys(t-Boc)]5-Gly-0CH~-PAMresin was synthesized by the standard fluorenylmethyloxycarbony1 (Fmoc) protocol using 0.5 mmol starting t-Boc-GlyOCH2-PAM resin [(phenylacetamido)methyl resin, AB1 Inc., substitution of 0.825 mmol/gl on an AB1 430A automated peptide synthesizer. Each cycle was controlled by ninhydrin analysis (AB1 protocol) and coupling was almost quantitative. The average yield was 1.26 g of resin. A pool of five batches of such resin (6.31 g) was treated with 63 mL of TFA for 2 h, filtered, washed with dichloromethane (2 X 40 mL), methanol (2 X 40 mL), dichloromethane (40 mL), and methanol (40 mL), and then dried. t-Boc-Ser(Bz1)-OSu (9.81 g, 25 mmol, Bachem) was dissolved in 50 mL of dry DMSO and the apparent pH (damp p H paper) brought to 8-9 with 250 pL of N methylmorpholine. The resin was added to this solution and the apparent pH readjusted to 8-9 with N-methylmorpholine. The mixture was agitated for 2 h in the dark. A portion of resin (approx. 5 mg) was removed, washed with methanol (2 X 2 mL) and dichloromethane (2 X 2 mL), and then dried in avacuum centrifuge. The standard ninhydrin test showed that acylation was almost quan-
Vilaseca et al.
titative. The bulk of the acylated resin was then filtered, washed, and dried as described above for the TFA deprotection. To remove Fmoc protection, the resin was treated with a mixture of 20 mL of piperidine and 30 mL of DMF for 2 h with gentle mixing. The resin was then filtered, washed, and dried as before. Ninhydrin analysis showed that deprotection was successful. Preparation of t-Boc-NHOCHzCO-Gly3-[Lys(t-BocSer(Bzl))]5-Gly-OCH2-PAMResin. To 1.91g (1Ommol) of [(tert-butyloxycarbony1)aminoloxylacetic acid (prepared according to ref 13, except that crude product was extracted into ethyl acetate to remove KCl and then dried by rotary evaporation) in 30 mL of ethyl acetate was added a solution of 1.15 g (10 mmol) of N-hydroxysuccinimide in 30 mL of the same solvent. Under constant mixing, a solution of 2.06 g (10 mmol) of N,N'-dicyclohexylcarbodiimide in 3 mL of ethyl acetate was then added and mixing was continued at room temperature. Reaction was followed by TLC on silica gel 60 using chloroform/methanol (1:l by vol) as eluent and spraying with a solution of N,Ndimethylbarbituric acid (50 mg/mL) in pyridine/water (9:l by vol) to reveal remaining diimide (14). After 5 h, no diimide was detected, dicyclohexylurea was removed by filtration and the filtrate was dried by rotary evaporation to yield 2.0 g. t-Boc-NHOCHzCOOSu (1.8 g 6.25 mmol), prepared as above, was dissolved in 50 mL dry DMSO and the apparent pH (estimated by spotting onto humidified pH paper) adjusted to 8-9 with N-methylmorpholine. The H-Gly3[Lys(t-Boc-Ser(Bzl))l~-Gly-OCH~-PAM resin (prepared above) was added to this solution and the apparent pH found to be 8-9. The mixture was stirred for 2 h, and a 5-mg portion of resin was removed, washed with methanol (2 X 2 mL) and dichloromethane (2 X 2 mL), and then dried in the vacuum centrifuge. The standard ninhydrin analysis revealed that acylation was almost quantitative. The bulk of the acylated resin was then filtered, washed, and dried as before to yield 5.8 g. Preparation of N H ~ O C H ~ C O - G ~ ~ ~ - [ L ~ S ( H - S ~ ~ GIy-OH. One gram of t-Boc-NHOCHzCO-Gly3-[Lys(tBoc-Ser(Bzl))l~-Gly-0CH~-PAM resin (prepared above) was treated with 10 mL of TFA for 30 min with gentle stirring. Trifluoromethanesulfonic acid (1mL) was then added and stirring continued for 1.5 h. Dry ether (25 mL) was added, under vigorous stirring, and the mixture cooled to 0 "C for 15 min. The precipitate formed was recovered by centrifugation and decantation and washed with dry ether (3 X 20 mL). After drying under vacuum, the precipitate, still in the presence of cleaved resin, was taken up in water (2 X 40 mL) and filtered through a Chromafil filter (type 0-45/25, Machery Nagel) to remove resin particles. Analytical HPLC showed the presence of one major component ( t =~ 20.7 min) and several minor components. The crude material was purified in portions on the preparative column, isocratically with 100% solvent A at 10 mL/min, and the major fraction, t R = 23-41 min, was collected, frozen, and lyophilized in the vacuum centrifuge. The material was shown to be pure by analytical HPLC. Analysis by FABMS in positive ion mode showed a protonated molecular ion at mlz 1395.41 (calculated M + H = 1395.75),consistent with the expected structure NHz0CHzCO-Gly3-[Lys(H-Ser)lS-Gly-OH. Further confirmation of stucture was provided by reacting the compound with isophthalaldehyde (see below). The uncorrected yield (based on starting Gly resin) was 47 % on a weight basis; the powder obtained was shown to be about 73 72 peptide by weight (the remainder presumably
Protein Conjugates of Defined Structure
Bioconjugate Chem., Vol. 4, No. 6, 1993 517
Sep-Pak cartridge, washing with 0.1% TFA in 20% being trifluoroacetate counterion and water, not a t all acetonitrile toremove excess reagent and eluting with 0.1 % unusual for peptides purified in this way). TFA in 50% acetonitrile. After lyophilization, the product Reaction of NHzOCH&O-Glye-[Lys(H-Ser)15-Glywas characterized by ESMS: found 7271.06 f 1.05, calcd OH with Isophthalaldehyde. A solution of NH20CHp7272.23. CO-Gly3-[Lys(H-Ser)15-Gly-OH (1.39 mg, 1pmol, in 1mL Preparation of DAI-NHNHCONHN=CH-m-Cswater) was added to a solution of isophthalaldehyde (1 H~CH=NOCH&O-(G~~)~-[L~S( DAVLB-NHN= mL, 5 mM in 0.1 M acetate buffer, sodium counterion, pH CHCO)]6-Gly-OH. DAI-NHNHCONHN=CH-m-C& 4.65, containing 20 mL/L acetonitrile). After 2 h, analysis CH=NOCH&O-(Gly)3- [Lys(H-Ser)l5-Gly-OH,prepared by HPLC showed that transformation to a more hydroas above, was dissolved (final concentration 167 pM) in 50 phobic product (tR = 28.8 min under standard conditions), mM imidazole buffer (chloride counterion, pH 6.95, later identified as the oxime, had occurred quantitatively. uncorrected glass electrode), containing 50 % acetonitrile. The new product, HCO-m-C6H&H=NOCH&O-(Gly)3Twenty equivalents (4 equiv for each of the five Ser [Lys(H-Ser)15-Gly-OH, was purified by HPLC on the residues of the carrier) of periodic acid (0.1 M in the analytical column a t 1 mL/min under the following imidazole buffer) was added. After 15 min, the reaction conditions: after 5 min a t 100% solvent A, a linear gradient was diluted with 3 volumes of 0.1% TFA and injected of solvent B, 1.75%/min, was applied until 7% B, which onto the analytical column equilibrated a t 30 95 solvent B. was maintained for 10min and followed by a linear gradient After 5 min a t 30% B, a linear gradient was applied to of 1% /min B to 50%. Excess isophthalaldehyde and 40 % B over 40 min. Oxidized product, which elutes slightly product eluted a t tR = 21 and 31 min, respectively. The earlier than traces of remaining unoxidized material, was product was characterized by positive ion FABMS and collected and lyophilized. The oxidized material was showed a strong signal at mlz 1511.57 (calculated M + H dissolved (100 pM) in a mixture of aqueous 0.1 M acetic = 1511.78), consistent with the expected structure HCOacid/acetonitrile ( 2 1 v/v). DAVLB-NHNH2 (desacetylvinm-C6H&H=NOCHzCO-(Gly)3- [Lys(H-Ser)l5-Gly-OH. Preparation of H-Gly3-[Lys(H-Ser)]s-Gly-OH. caleukoblastine hydrazide hemisulfate; Eli Lilly, Inc.) was added as a concentrated solution (0.1M in water) to obtain H-Gly~-[Lys(t-Boc-Ser(Bzl))l~-Gly-OCH~-PAM resin (100 30 mM as a final concentration, whereupon the pH was mg, prepared as above) was cleaved with TFA/trifluoraised to an indicated (paper) value of between 4.3 and 5. romethanesulfonic acid (as described above) and the After 22 h and 45 min, products were separated on the peptide precipitated and washed with ether. Analysis by analytical column under the following conditions: after HPLC showed the presence of two components. The major 5 min a t 20% solvent B, a linear gradient of 1%B/min component (earlier eluting, t~ = 17 min on the analytical was applied to 60% B. The product, which eluted as the column) was purified on the preparative column, eluted major peak following the excess DAVLB-NHNH2, was isocratically with solvent A a t 10 mL/min (tR = 12-13.5 characterized by ESMS. min). Analysis by FABMS in the positive ion mode showed a strong signal a t mlz 1322.85 (calculated M + H = RESULTS AND DISCUSSION 1322.74), consistent with the expected structure H-Gly3[Lys(H-Ser)l5-Gly-OH. The later-eluting component (tR Preparation and Characterization of the Carrier = 17.5 min on the analytical column) was found by FABMS Molecule. The preparation of the carrier molecule is to be heavier than the first by one Ser residue (87 amu). shown in Figure 1. Standard solid-phase techniques were Oxidation of H-Gly3-[Lys(H-Ser)]5-Gly-OH and used and the desired product was purified by HPLC and Oximation. Ten microliters of a solution of H-Gly,-[Lyscharacterized by mass spectrometry. It is important that (H-Ser)]a-Gly-OH (0.5 mM in water) was added to 1mL the coupling of Boc-Ser(Bz1)OSu be carefully performed, of imidazole buffer (1mM, pH 6.95, chloride counterion), using a mild base (N-methylmorpholine, not the more usual and then 20 pL of a solution of NaI04 (2.5 mM in water) N,N-diisopropylethylamine),room temperature, and a reaction time of 2-3 h. Otherwise, partial loss of Fmoc was added. Final concentrations were thus approximately 5 pM peptide (25 pM Ser), 50 pM periodate, and 1 mM protection occurs, such loss being followed by acylation of imidazole. At various times (5 min to 24 h), aliquots (15 the exposed amino group. By careful attention to detail, pL) were analyzed by HPLC under standard conditions. we have repeated this synthesis more than 20 times without Similar experiments were performed a t higher final problem. concentrations: 50 pM peptide, 500 pM periodate, 10 mM The aldehyde form of the carrier molecule was prepared imidazole; 250 pM peptide, 2.5 mM periodate, and 50 mM by reaction of the aminooxy form with a 5-fold excess of imidazole. Product (tR = 20 min) was collected and shown isophthalaldehyde (Figure l ) , a procedure used previously to be the expected oxidation product, H-Gly3-LLysto convert an aminooxy chelator to an aldehyde form (3). (HCOCO)]b-Gly-OH, by ESMS (using the signal due to Oxidation of the Carrier Molecule. The high susthe diprotonated form, found 1167.42, calcd 1167.20). ceptibility of l-amino-2-hydroxy compounds to periodate Preparation of DAI-NHNHCONHN4H-m-CsH4has been well studied for the case of one such site per CH=NOCH~CO-(G~~)~-[L~S(H-S~~)]K-G~~-OH. DAIpolypeptide (1,2,16,17). In order to optimize conditions NHNHCONHNHz was prepared as a solution (0.2 mM) for the oxidation of our carrier molecule, which contains in acetate buffer (0.1 M, pH 4.6) as previously described five terminal Ser residues per molecule, we used the model (3). This solution was used to dissolve 2 equiv of freshly compound H-Gly3-[Lys(H-Ser)l5-Gly-OH. At dilute conlyophilized HCO-m-C6H&H=NOCHzCO-(Gly)3-[Lys(Hcentration, approximately 5 pM peptide (25 pM Ser), 50 Ser)ls-Gly-OH, prepared as above, and the mixture was pM periodate, 1mM imidazole, analysis by HPLC showed allowed to react for 17 h. Reaction progress was assessed that after 5 min the starting peptide had disappeared and by analytical HPLC under the following conditions: after that a new compound, eluting later as a sharp peak, was 5 min at 100% solvent A, a linear gradient of 2% B/min formed. After 1.5-hreaction, there was no further change was applied to 30% B, whereupon the slope was lowered (not shown). When higher final concentrations were used to 0.5% B/min to 50% B. Small quantities (ca. 100 pg) (50 pM peptide, 500 pM periodate, 10 mM imidazole), a of product were isolated on the analytical column. Larger reaction time of 5 min led to a clean transformation (Figure quantities (milligram amounts) were purified on a C18 2a) whereas after 1.5 h (Figure 2b) there were signs of
518
Biocon/ugate Chem., Vol. 4, No. 6, 1993
Bot
Vilaseca et al.
/
I
(iii),(iv)
/-----
F~oc-(GI~)~-(L~s)~-G~~-OCH~-PAM
/
0
Boc-Ser(Bz1)
10
20
30
40
Time (min)
0.20 BOC-NHOCH~CO-(GIY)~-(LYS)~-GI~-OCH~-PAM
I
-
Boc-Ser(Bz1)
1
c\1
C
NH~OCH~CO-(G~Y)~-(L~S)~-G~~-OH
/ HO-CH2
0.10
4
(vii),(viii)
Ser residue
0.15
0.05 0
\
(ix)
0
10
20
30
40
Time (min)
Figure 2. Oxidation of the model H-Gly3-[Lys(H-Ser)ls-Glyaminooxy rN-r-c=o form OH with periodate. Analysis by HPLC of oxidation reactions involving (a)50 WMpeptide, 5 min; (b) 50 MMpeptide, 1.5 h. The arrow and the asterisk indicate the elution positions of the HCO-m-C6H4-CH=NOCH2CO-(Gly)3-(Lys)5-Gly-OH unoxidized peptide and of overoxidized material, respectively.
nN
I
CH-C=O
'i
DAI
HO-CH2 aldehyde form
Figure 1. Synthesis of the carrier molecule in aminooxy and aldehyde forms: (i) TFA to remove Boc; (ii) standard Fmoc automated peptide synthesis; (iii) TFA; (iv) Boc-Ser(Bz1)-OSu; (v) piperidine/DMF; (vi) Boc-NHOCHzCOOSu;(vii)TFA; (viii) TFA/TFMSA; (ix) HCO-m-CeH&HO. overreaction. When even higher final concentrations were used (250 pM peptide, 2.5 mM periodate, 50 mM imidazole), a reaction time of 5 min led to a very slight overreaction whereas after 1.5 h extensive overreaction was observed and, after 20 h, the first-formed oxidation product was completely destroyed (not shown). Addition of ethylene glycol (2 equiv over periodate) after 5 min reaction was found to prevent the overreaction of peptide. The first-formed oxidation product, shown to be H-Gly3[Lys(HCO-CO)]6-Gly-OH by ESMS, once isolated by HPLC, was stable as a lyophilized powder a t -20 "C. Preparation of DAI-NHNHCONHN=CH-m-C6HdC H = N O C H2C 0 - ( G l y ) 3 - [ L y s ( D V L B - N H N = CHCO)]s-Gly-OH. Des-AlaB30-insulin (DAI) was used as a model protein in order to be able to follow precisely, by reversed-phase HPLC and mass spectrometry, the various steps involved (Figure 3). Step 1. A hydrazide function was introduced a t the carboxyl terminus of DAI by reverse proteolysis using carbohydrazide as nucleophile (3). This mild enzymatic technique of site-specificprotein modification is applicable to antibodies and their fragments and places the modification far from the antigen-binding region ( 4 ) . Step 2. The aldehyde form of the defined carrier was
DAI-NHNHCONHNH2
i2 L
4.
DAI-NHNHCONHN=CH-m-C6H~-CH=N~H~CO-(Gly)3-[Lys]5-Gly-OH
/
DAVLB-NHN=CH-CO
Figure 3. Synthesis of the site-specific pentadrug cluster of DAI: step 1, enzyme-catalyzed attachment of carbohydrazide specifically to residue B30; step 2, Condensation with the aldehyde form of the carrier molecule;step 3, very mild periodate oxidation of the 1-amino-2-hydroxyfunction of the Ser residues; and step 4, hydrazone formation with the drug DAVLB-NHNHz. attached to the hydrazide group on the protein by hydrazone formation, a general reaction which is mild and perfectly specific ( 3 ) . Figure 4 shows the result of analysis of the reaction mixture by HPLC. Very little DAIcarbohydrazide remains, and the product was identified as the desired hydrazone by ESMS. Step 3. Ser residues of the carrier, which possess a 1-amino-2-hydroxystructure very susceptible to periodate, were oxidized under very mild conditions and the reaction followed by HPLC (Figure 5). The protein polyaldehyde
Bioconjugate Chem., Vol. 4, No. 6, 1993 519
Proteln Conjugates of Defined Structure 1.8
2.0 product
1.8
excess DAVLL-NHNH,
1.6 1.4
1.6
I
1.4
1.2 excess oligomer
1.2
-+
I
1.0
3
*
N
9
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
penta-drug cluster
i
N 4
0
0
__z
-.
I
0
10
d
50
40
30
20
Time (min)
Figure 4. HPLC of the reaction between DAI-NHNHCONHNHzand the aldehyde form of the carrier molecule. Excess carrier elutes after 26 min; product DAI-NHNHCONHN=CHm-C&CH=NOCH&O-(Gly)3- [Lys(H-Ser)ls-Gly-OHis responsible for the major peak at 42 min, and remaining DAINHNHCONHNHz elutes after 43 min. 0.8
0.6
,
10
20
30
40
50
60
Time (min)
Figure 6. HPLC of the formation of DAI-NHNHCONHN=CH~-C~H&H=NOCH~CO(G~~)~-[L~S(X)]~-G~~-OH, where X = DAVLB-NHN=CHCO. Approximate retention times are as follows: excess drug, 26 min; tridrug product, 44 min; tetradrug product, 46 min, and pentadrug product, 47 min.
oxidized material
/ I
'
i
N
starting material
O
J ,
I
,
.
,
.
,
.
,
,
/
Figure 7. Electrospray ionization mass spectrum of the pentadrug conjugate DAI-NHNHCONHN=CH-m-C6H&H=NOCH&O-(Gly)3-[Lys(X)]~-Gly-OH, where X = DAVLB-NHN= CHCO. The experimentally determined molecular weight is 10870.59 f 1.63,very close to the calculated value of 10871.64.
Minor components visible in Figure 6 were thus shown to carry four and three drug molecules, whereas Figure 7 Time (min) shows the electrospray mass spectrum of the major product, DAI-NHNHCONHN=CH-m-CeHdCH= Figure 5. HPLC of the periodate oxidation of DAI-NH[Lys(DAVLB-NHN=CHCO)] B-GlyN H C O N H N ~ H - ~ - C ~ C H = N O C H Z C O - ( G ~ ~ ) B - [ L ~ ~ ( HNOCH2CO-(Gly)3-S~~)~SGly-OH.The major peak is due to oxidized material. Unoxidized OH: found 10870.59 f 1.63, calcd 10871.64. The five drug material elutes slightlylater and may be responsiblefor the slight molecules are all attached to a single site on the protein: tailing of the product peak. the C-terminus of the B chain of the model protein DAI. A hydrazide drug has thus been conjugated to the aldehyde did not give an interpretable mass spectrum, but must groups of the oxidized carrier, producing a conjugate of have been formed since the product of the oxidation went defined chemical structure carrying 5 mol of drug/mol of on to form the desired hydrazone (see below). protein. The pentadrug conjugate was found to be stable in 50 mM imidazole hydrochloride buffer a t pH 7 for at Step 4. The drug DAVLB-NHNH2 was conjugated to least 24 h, whereas at pH 2 it became hydrolyzed, as the aldehyde groups of the carrier. Depending on the drug excess and reaction conditions used, a cluster of up to five expected, to a mixture of tetra-, tri-, and didrug forms and free drug, over a period of 5 h (not shown). drug molecules was formed. The reaction was followed by reversed-phase HPLC (Figure 6) and the products In other experiments (Scott, W. L., data not shown), the aminooxy version of the defined carrier has been used identified by electrospray ionization mass spectrometry. 0
10
20
30
40
50
520
Bloconlugate Chem., Vol. 4, No. 8, 1993
to attach clusters of hydrazide drugs (DAVLB-NHNH2 and methotrexate y-hydrazide) to antitumor monoclonal antibodies, and the resulting immunoconjugates have exhibited promising effects in tumor xenograft studies in mice. CONCLUSIONS A defined carrier possessing a single reactive group (aminooxy or aldehyde), a spacer function (three Gly residues), and five potential aldehyde groups (in the form of Ser residues) has been prepared and characterized. In its aldehyde form, this defined carrier was attached to a protein in good yield under very mild conditions a t a specific site, i.e. the carboxyl terminus of a polypeptide chain to which carbohydrazide has been fixed by reverse proteolysis. Oxidation by periodate, under very mild conditions, of the pendant Ser residues of the carrier generated a cluster of new aldehyde groups. A hydrazide drug was conjugated to the generated aldehyde groups and produced a conjugate of defined chemical structure carrying up to 5 mol of drug/mol of protein. The conjugate was stable a t neutral pH, and drug was released during incubation at low pH (not shown). The aminooxy version of the defined carrier has been used to attach clusters of drugs to antitumor monoclonal antibodies. This aminooxy form may be attached to proteins which carry an aldehyde group specifically a t the N- (1, 2 ) or C-terminus (16), or on side chains (11) (not shown). ACKNOWLEDGMENT We thank Ms. Irene Rossitto and Mr. Pierre-Olivier Regamey for expert technical assistance, the Schmidheiny Foundation for some of the HPLC equipment used, and the Fonds National Suisse de la Recherche Scientifique and the Sandoz Stiftung for support of the Geneva mass spectrometry facility. LITERATURE CITED (1) Geoghegan, K. F., and Stroh, J. G. (1992) Site-directed conjugation of non-peptide groups to peptides and proteins via periodate oxidation of a 2-amino alcohol. Bioconjugate Chem. 3, 138-146. (2) Gaertner, H. F., Rose, K., Cotton, R., Timms, D., Camble, R., and Offord, R.E. (1992) Construction of protein analogues
by site-specific condensation of unprotected fragments. Bioconjugate Chem. 3,262-268. (3) Rose, K., Vilaseca, L. A., Werlen, R., Meunier, A., Fisch, I., Jones, R. M. L., and Offord, R. E. (1991) Preparation of welldefined protein conjugates using enzyme-assisted reverse proteolysis. Bioconjugate Chem. 2, 154-159. (4) Fisch, I., Ktinzi, G., Rose, K., and Offord, R. E. (1992) Sitespecific modification of a fragment of a chimeric monoclonal antibody using reverse proteolysis. Bioconjugate Chem. 3, 147-153.
Vilaseca et al. (5) Schnolzer,M., and Kent, S.B. H. (1992) Constructingproteins
by dovetailing unprotected synthetic peptides: backboneengineered HIV protease. Science 256,221-225. (6) O’Shannessy,D. J., and Quarles, R. H. (1987) Labeling of the oligosaccharide moieties of immunoglobulins. J. Zmmunol. Methods 99, 153-161. (7) Pietersz, G. A. (1991) The linkage of cytotoxic drugs to
monoclonal antibodies for the treatment of cancer. Biocon-
jugate Chem. 1,89-95. (8) Keto, Y.,Umemoto, N., Kayama,Y.,Fukushima, H.,Takeda, Y., Hara, T., and Tsukada, Y. (1984) A novel method of
conjugation of daunomycin with antibody with a poly-Lglutamic acid derivative as intermediate drug carrier. An antialpha-fetoprotein antibody-daunomycin conjugate. J. Med. Chem. 27, 1602-1607. (9) Slinkin, M. A., Klibanov, A. L., and Torchilin, V. P. (1991)
Terminal-modifiedpolylysine baaed chelating polymers: highly efficient coupling to antibody with minimal loss in immunoreactivity. Bioconjugate Chem. 2, 342-348. (10) Mann, J. S., Huang, J. C., and Keana, J. F. W. (1992) Molecular amplifiers: synthesis and functionalization of a poly(aminopropy1)dextran bearing a uniquely reactive terminus for univalent attachment to biomolecules.Bioconjugate Chem. 3, 154-159. (11) Scott, W. L., Horvath, S., Nichols, C. L., Johnson, D. A., Starling, J. J., Vorndran-Jones, C., Baker, A. L., Owens, R., Cullinan, G., Laguzza, B. C., and Bumol, T. (1990) Drug
releasability as a critical factor in the design of active immunoconjugates: an evolution to new, broadly applicable, lysine-based, drug-releasing linkers yielding active conjugates of antibodiesandtheir fragments. Antibody Immunoconjugate Radiopharm. 3, 65. (12) Vilaseca, L. A., Rose, K., Werlen, R., Meunier, A., Offord, R. E., Nichols, C.L., and Scott, W. L. (1993) Synthesis and use
of a defined oligomer for targetted drug therapy. In Peptides 1992 (C. H. Schneider, and A. N. Eberle, Eds.) pp 819-820, ESCOM, Leiden, The Netherlands. (13) 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. (14) Kohn, J., and Wilchek, M. (1982) Selective and highly sensitive spray reagent for detection of nanomolar quantities of carbodiimides on thin-layer plates. J. Chromatogr. 240,262-
263. (15) Savoy, L.-A., Jones, R. M. L., Pochon, S., Davies, J. G., Muir, A. V., Offord, R. E., and Rose, K. (1988) Identification
by fast atom bombardment mass spectrometry of insulin fragments produced by insulin proteinase. Biochem. J . 249,
215-222. (16) Rose, K., Jones, R. M. L., Sundaram, G., and Offord, R. E. (1989) Attachment of linker groups to carboxyl termini using enzyme-assistedreverse proteolysis. In Peptides 1988, (G.Jung, and E. Bayer, Eds.) pp 274-276, W. de Gruyter, Berlin. (17) Gaertner, H. F., Rose, K., Cotton, R., Timms, D., Camble, R., and Offord, R.E. (1993) Construction of protein analogues
by site-specific condensation of unprotected fragments. In
Peptides 1992, (C. H. Schneider, and A. N. Eberle, Eds.) pp 239-240, ESCOM, Leiden, The Netherlands.
