Site-Specific Religation of G-CSF Fragments through a Thioether

Proteinsynthese durch chemische Verknüpfung ungeschützter Peptide in wäßriger Lösung. Michael A. Walker. Angewandte Chemie 1997 109 (10), 1113-11...
0 downloads 0 Views 2MB Size
Bioconjugate Chem. 1094, 5, 333-338

333

Site-Specific Religation of G-CSF Fragments through a Thioether Bond Hubert F. Gaertner,*lt Robin E. Offord,+ Ron Cotton,$ David Timms,* Roger

Gamble: and Keith Rose+

DBpartement de Biochimie MBdicale, Centre MBdical Universitaire, 1, rue Michel Servet, 1211 Geneva 4, Switzerland, and Zeneca Pharmaceuticals, Mereside, Alderley Park, Macclesfield Cheshire SK 10 4TG, U.K. Received February 9, 1994"

A new approach is described for linking, through a thioether bond, the C-terminus of one unprotected polypeptide with the N-terminus of another. Homocysteine thiolactone is attached to the C-terminus of one polypeptide by reverse proteolysis and provides through hydroxylamine treatment a free sulfhydryl group. The cy-aminogroup of a second polypeptide is selectivelyiodoacetylated by reaction with iodoacetic anhydride at pH 6.0 or the N-hydroxysuccinimide ester derivative at pH 7.0. Coupling of the two modified fragments occurs in a spontaneous alkylation reaction under mild conditions. After preliminary experiments with small peptides, this approach was extended to large protein fragments derived from recombinant analogs of G-CSF by enzymatic digestion. This approach provides a means of making head-to-tail protein chimeras or introducing noncoded structural elements into a protein.

INTRODUCTION In recent years several chemical or chemoenzymatictools have been successfully developed for the construction of protein analogs and the introduction of noncoded structural elements as a means of modulating protein stability and activity. Those approaches that show the most general applicability involve the use of nonpeptide links formed by the spontaneous and specific reaction of two functional groups that have been introduced at the end of the fragments to be joined. For example, formation of a hydrazone bond was shown to be a powerful way to sitespecifically relink protein fragments resulting from the enzymatic digestion of a recombinant analog of G-CSF (Gaertner et al., 1992) or to incorporate a synthetic peptide into the protein backbone (Gaertner et al., 1994). Other means already investigated for the construction of protein chimaeras or analogs involve relinking through a disulfide bridge which is a procedure used since the beginning of synthetic studies on proteins (Humphries et al. and references cited therein, 19911, an oxime bond (Rose, 19941, or a thioester or a thioether bond (Schnolzer and Kent, 1992,1993). However, in the case of Schnolzer and Kent's analogs of HIV-1 protease both fragments were made by total synthesis so that the positioning of the reactive groups was not a problem. In this paper, we describe an approach for relinking unprotected protein fragments through a stable, thioether bond, using as an example the two polypeptides resulting from enzymatic cleavage of tailored recombinant analogs of G-CSF containing a single lysine residue at specified positions (Gaertner et al., 1992). This strategy involves two distinct modifications. In one, a reactive cysteine derivative, homocysteine thiolactone, was introduced a t the C-terminal lysine of the terminal fragment by reverse proteolysis, using the same enzyme as was used for cleavage of the tailored analog at that point. The grafted residue is subsequently opened with hydroxylamine to liberate a thiol group ready for alkylation. For the second modification, an a-haloacetyl moiety is specifically grafted to the N-terminus of the second fragment, owing to the absence of any €-aminogroup. To exemplify the method, we describe the conjugation of two small peptides bearing

these functional groups at the ends to be joined and the extension of this chemistry to larger polypeptides for the construction of several protein analogs of G-CSF. EXPERIMENTAL PROCEDURES

Materials and General Methods. Iodoacetic anhydride was from Aldrich Chemical Co. and Ser-Leu-Leu from Bachem (Bubendorf, Switzerland). Boc-GlyONSu, iodoacetate, bromoacetate, sodium lauroyl sarcosinate (Sarcosyl), and homocysteine thiolactone (HCTL) were purchased from Fluka (Buchs, Switzerland). Achromobacter Zyticus protease (lysyl endopeptidase) was from Wako Pure Chemical Co., Osaka, Japan. Boc-Gly-HCTLwas synthesized by adding 82 mg of BocGly-ONSu (300 pmol) to 50 mg of HCTL (325 pmol) dissolved in 1 mL of anhydrous DMSO. N-methylmorpholine was added to an apparent pH of 9.0 (measured externally using pH paper prewetted with water). After 3 h at room temperature, the reaction mixture was diluted with 25 volumes of 0.1% TFA and applied on a ChromabondCB cartridge, bed volume 2 mL, previously washed with MeOH and equilibrated with 0.1 5% TFA. After the Chromabond was thoroughly washed with 0.1% TFA containing 5% acetonitrile, the product was eluted with 25 mL of 0.1 % TFA containing 50%acetonitrile and dried under vacuum (weight, 66 mg; yield, 80%). The product was characterized by electrospray ionization mass spectrometry (ESI-MS) (calcd M + H, m / z 275.2, found m / z 275.3).

ICHzCOONSu was obtained by reaction of 1 equiv of ICH2COOHwith 1 equiv of N-hydroxysuccinimide (NSu) and 1 equiv of dicyclohexylcarbodiimide dissolved in a small volume of ethyl acetate. After 4 h incubation, dicyclohexylurea was discarded by filtration and the solvent evaporated. An analogous procedure was used to prepare the bromoacetyl derivative. HPLC was performed on a Waters 625 LC system equipped with a Wisp 712 sample processor and a Model 441 UV detector. For analytical work, a column 250 X 4 mm i.d. (Nucleosil300-A 5-pm C8, Macherey Nagel, Oensingen, Switzerland) was used at a flow rate of 0.6 mL/min and the effluent monitored at 214 nm. For preparative work, a column 250 X 10-mm i.d. of Nucleosil 300-A 5-pm C4 or C8 (Macherey Nagel) was used at a flow rate of 3 mL/min. 0 1994 American Chemical Society

334

Biocon/ugate Chem., Vol. 5, No. 4, 1994

Gaertner et ai.

Scheme 1. Construction of the Thioether Analog of G-CSF

CO-CH,-I

Lys-NH-CH-(CHJ,-S-CH,-CO CONHOH

Solvent A was 1g of trifluoroacetic acid (Pierce) added to 1L of water (MilliQ system). Solvent B was prepared by adding 1g of trifluoroacetic acid to 100 mL of water and making up to 1 L with acetonitrile (Lichrosolv, Merck). The N-terminal fragments (1-62, 1-75, 1-98) and C-terminal fragments (63-174, 76-174, 99-174) were obtained by enzymatic digestion with Achromobacter protease of three recombinant analogs of G-CSF produced in E . coli as previously described (Gaertner et al., 1992). These analogs are known as TG116 (Cys 17 Ser; Lys 16, 23, 34,40 Arg; Ser62 Lys), TG 117 (Cysl7 Ser; Lysl6,23,34,40 Arg; Leu75 Lys), and TG47 (Cysl7 Ser; Lysl6, 23, 34, 40 Arg; Glu98 Lys). Each contains a single lysine residue for proteolytic cleavage, and ligation of resulting large protein fragments can be achieved, as already shown with TG116 and TG117 fragments, through a hydrazone linkage (Gaertner et al., 1992). Enzymatic Coupling of Homocysteine Thiolactone. HCTL was attached to fragments 1-75, 1-62, and 1-98, all of which have a C-terminal lysine, by reverse reaction of Achromobacter protease. A solution of HCTL (0.5M) was prepared in 80% butanediol, and the apparent pH was adjusted to 5.5 with N-methylmorpholine using an uncorrected glass electrode calibrated with aqueous standards. The protein fragment was dissolved in the HCTL solution at 20 mg/mL. Achromobacter protease was added as a freshly prepared solution in water (20 mg/ mL) at an enzyme/substrate ratio of 1:20 (w/w) and the sample incubated at room temperature for 3 h. The extent of modification was followed by analytical HPLC of samples quenched in excess 0.1 % TFA using a linear gradient from 40% to 60% B over 40 min for fragments 1-75 and 1-62 and 45% to 65 % B over 40 min for fragment 1-98. In both cases, the HCTL derivative eluted later than the unmodified fragment. The modified fragment was recovered by acidification with two volumes of pure acetic acid and dilution with 10 volumes of 0.1% TFA followed by adsorption to a Sep-Pak CIS cartridge, previously washed with methanol and equilibrated with 0.1 % TFA. After being washed with 5 mL of 0.1 % TFA containing 20% CHsCN, the sample was eluted with 5 mL of 0.1% TFA containing 80% CH3CN. Solvent was removed in the vacuum centrifuge and the mixture of modified and unmodified fragments lyophilized and taken up in water at a 2 mM concentration for the condensation reaction. Iodoacetylation and Bromoacetylation. Derivatization of the tripeptide Ser-Leu-Leu with iodoacetic anhydride was performed accordingto the method of Wood and Wetzel(1992a). Briefly, to 1mL of Ser-Leu-Leu (10 mM in a 0.1 M 2-(N-morpholino)ethanesulfonicacid (MES) buffer at pH 6.0) were added under vortex mixing

-

- - - -

- -

three 60-pL additions of iodoacetic anhydride (0.25 M in dry THF) at 3-min intervals. The reaction was stopped after 10 min at 0 OC by the addition of 2 equiv of acetic acid and the product purified by HPLC on the 250 X 10mm i.d. column using a flow rate of 3 mL/min and a linear gradient of 20-40% (by vol) B over 20 min. The product gave the expected molecular weight, as detemined by FABMS (calcd M + H, m/z 500.3, found m/z 500.6). The protein fragments 76-174 and 99-174 were dissolved (2 mg/mL) in a 0.1 M sodium phosphate buffer (pH 7.0) in the presence of 0.3 % Sarcosyl and were modified with the N-hydroxysuccinimide ester of iodo- or bromoacetic acid, since the derivatization with iodoacetic anhydride was in this case inefficient, as monitored by analytical HPLC using a linear gradient of 50%-80% (by vol) B over 30 min in the case of fragment 76-174 and 45 % -85 % B over 40 min in the case of fragment 99-174. Eight to 10 additions of 5 equiv of the active ester were made over 30 min to the fragment solution at 0 "C and the reaction stopped after 45 min. In the case of 76-174, this was done by desalting the derivatized fragment on a 10-mL Biorad column packed with Biogel P6 and equilibrated in 0.1 M sodium phosphate pH 8.0 in the presence of 0.1 % Sarcosyl. The excluded protein fraction was used for the alkylation reaction after concentration on a Centricon 100 concentrator to about 2 mg/mL. In case of a-haloacetylated 99174, which is less soluble than derivatized 76-174, the reaction was stopped by a pH shift to 5.3 with acetic acid and the excess of reagent removed by dialysis against a 20 mM NaOAc buffer, pH 5.3, containing 0.1% sarcosyl during 7 h. Since working up the iodoacetylation reaction as quickly as possible was found to be essential, as already reported (Wood & Wetzel, 1992b), HPLC purification would have been the best way to isolate the iodo- or bromoacetyl products. However, the very low recoveries with this method, probably due to the high hydrophobicity of these particular compounds, led us to adopt the alternative conditions of purification described. The condensation reaction was started without delay. Condensation Reaction. In the case of model peptides, equimolar amounts of Boc-Gly-HCTL and ICH&O-SerLeu-Leu were mixed to obtain a 1 mM solution in 0.1 M sodium phosphate, pH 8.0, which was brought to 0.1 M NHzOH and incubated at room temperature. The conjugation was followed by analytical HPLC, using a linear gradient of 10% to 50% (by vol) B over 40 min, and all the peaks were collected and characterized by electrospray ionization mass spectrometry. In the case of coupling the iodoacetyl tripeptide to the C-terminal-derivatized fragment 1-75 from TG117, a 2-fold molar excess of the alkylating peptide over the latter compound was used. In case of derivatized protein fragment, conjugation was carried out with a 2-3-fold excess of HCTL-derivatized N-terminal fragment (2 mM in water, initially) over the a-haloacetylated C-terminal fragment in a 0.1 M sodium phosphate buffer, pH 8.0, and brought to 1 mM EDTA and 0.1 M NHzOH. The 99-174 derivative which was previously dialyzed against 20 mM NaOAc buffer, pH 5.3, was mixed with a 1/10 volume of 1M sodium phosphate buffer, pH 8.0, and the pH adjusted to 8.0 with NaOH before the addition of the other reagents. After 15 h incubation at room temperature, the reaction mixture was analyzed by SDS-PAGE under reducing conditions and the coupled product was separated from unreacted fragments and side products by reversed-phase HPLC using a linear gradient from 50 % to 80 % B over 30 min in case of the TG117 analog and 45% to 70% B over 50 min in case of the TG47 analog.

Bioconjugate Chem., Vol. 5, No. 4, 1994

Thloether Analog of G-CSF

0

10

20

30

40

Time (min) Figure 1. Reversed-phase HPLC chromatograms of (A) the reaction mixture of Boc Gly-HCTL preincubated 1 h with 0.1M NH2OH and iodoacetylated Ser-leu-Leu at pH 2.0 and (B) the reaction mixture of iodoacetylated Ser-Leu-Leu and Boc-GlyHCTL after 15 h incubation at pH 8.0, in the presence of 0.1 M NH20H. The conjugation product elutes at t~ = 40 min while the peak that eluted at t R = 32 min corresponds to the side product of the coupling reaction. In A, Boc-Gly-HCTL and ICHzCOSer-Leu-Leu elute at t~ = 26 and 35 min, respectively, and the NH2OH-opened thiolactone derivative at 22 min.

SDS-PAGE. SDS-PAGE under reducing conditions (5 % ,by volb-mercaptoethanol) was performed on a PhastSystem electrophoresis apparatus (Pharmacia) using 20 96 polyacrylamide gels. Proteins were applied to the gel for 90 Vh at 15 "C and visualized by silver staining. Protein standards were from Pharmacia: phosphorylase b (94 kDa), albumin (67 kDa), ovalbumin (43 kDa), carbonic anhydrase (30 kDa), soybean trypsin inhibitor (20.1kDa), and a-lactalbumin (14.4 kDa). Mass Spectrometry. The molecular weight of the different derivatives obtained at each step of the constructions was measured by mass spectrometry. Equipment and operating conditions were as previously described (Gaertner et al., 1992). RESULTS AND DISCUSSION Peptide Coupling through a Thioether Linkage. Homocysteine thiolactone was chemically attached to BocGly in order to introduce a thiol functionality at the C-terminus and study the site-specific conjugation with iodoacetylated Ser-Leu-Leu. This stategy of conjugation through a thioether bond involves two reaction steps, the deprotection of the masked sulfhydryl group and its subsequent alkylation. As shown in Figure lB, the

335

coupling reaction proceeds cleanly. A 1-h preincubation of the HCTL derivative with NHzOH (0.1 M) at pH 8.0, which releases a free sulfhydryl group at the C-terminus (Figure lA), is in fact not necessary to carry out the alkylation; this procedure only improves the rate of the coupling reaction, which is almost complete after 5 h incubation at room temperature (results not shown). The same profile was obtained when NHzOH treatment and alkylation were carried out at the same time (Figure 1B) with a 15h incubation in a 0.1 M sodium phosphate buffer, pH 8.0, containing 0.1 M NH20H. These conditions were expected to reduce the occurence of unwanted side reactions (i.e., disulfide formation or disulfide bond exchange) when conjugation involves polypeptides containing either cysteine residues or disulfide bridges. The conjugation product had the expected molecular weight, as determined by FAB-MS (calcd M + H m/z 679.32, found mlz 679.26). The earlier-eluting product ( t R = 32 min), was identified by FAB-MS as the Nalkylated derivative of Gly-HCTL (calcd M + H mlz 546.25, found mlz 546.31). This side product could be detected under all conditions of conjugation tried and was shown to be especially prominent when the concentration of NHzOH was verylow (