New PEGs for Peptide and Protein Modification, Suitable for

62

Bioconjugate Chem. 2001, 12, 62−70

New PEGs for Peptide and Protein Modification, Suitable for Identification of the PEGylation Site F. M. Veronese,*,‡ B. Sacca`,‡ P. Polverino de Laureto,‡ M. Sergi,‡ P. Caliceti,‡ O. Schiavon,‡ and P. Orsolini§ Department of Pharmaceutical Sciences (CNR, Center for Chemical Investigation of Drugs), University of Padova, Via Marzolo 5, 35131 Padova, Italy, and DEBIOPHARM s.a. Martigny, Switzerland. Received June 1, 2000; Revised Manuscript Received October 6, 2000

New PEG derivatives were studied for peptide and protein modification, based upon an amino acid arm, Met-Nle or Met-βAla, activated as succinimidyl ester. PEG-Met-Nle-OSu or PEG-Met-βAla-OSu react with amino groups in protein-yielding conjugates with stable amide bond. From these conjugates PEG may be removed by BrCN treatment, leaving Nle or βAla as reporter amino acid, at the site where PEG was bound. The conjugation of PEG and its removal by BrCN treatment was assessed on a partial sequence of glucagone and on lysozyme as model peptide or protein. Furthermore, insulin, a protein with three potential sites of PEGylation, was modified by PEG-Met-Nle, and the PEG isomers were separated by HPLC. After removal of PEG, as reported above, the sites of PEGylation were identified by characterization of the two insulin chains obtained after reduction and carboxymethylation. Mass spectrometry, amino acid analysis and Edman sequence, could reveal the position of the reporter norleucine that corresponds to the position of PEG binding.

INTRODUCTION

The identification of the PEGylation sites in the primary sequence of peptides and proteins remains an open problem in the characterization of conjugates. This information may be of help to design reaction conditions more suitable to direct PEG toward residues not involved in binding with substrates, ligands, or receptors; furthermore, it is important for characterization of the conjugates as required by FDA. Indeed, for short peptides, Edman degradation may be sufficient to identify the PEGylation site since the conjugated amino acid appears as a gap in the sequence. This was the case encountered in our laboratory to establish the position of PEG isomers in the 29 amino acid sequence of growth hormone releasing factor (1). However, this procedure is no longer feasible in large peptides and proteins, where a previous enzymatic cleavage must be carried out to get shorter sequence, as was the case of growth hormone or G-CSF (2). In such case, the conjugate was hydrolyzed by trypsin and the site of PEGylation stated on the basis of the missing tryptic peptides (due to PEGylation) in comparison with those obtained from the native protein. Although successful, this method cannot be considered of general suitability, since it is based on a negative information: a missing peptide may come also from incomplete hydrolysis that may occur close to a PEGylated lysine. Despite of their limitations these methods were successfully employed in the characterization of R-interferon also, where only monopegylated isomers, purified by ion exchange chromatography were obtained (3). A further approach to the problem is based on the use of PEG with a hydrolyzable succinic acid ester arm * To whom correspondence should be addressed. Phone: +39-049-8275694. Fax: +39-049-8275366. E-mail: veronese@ pdfar3.dsfarm.unipd.it. ‡ University of Padova. § DEBIOPHARM s.a. Martigny.

between polymer and protein. In this case, a mild basic cleavage removes PEG from the protein, leaving succinic acid bound to the protein as a reporter group. This labeled protein, after being digested with a proteolytic enzyme, is subjected to peptide fractionation and the succinilated peptides identified by Maldi mass spectrometry (4). This original procedure overcomes also the problem of the difficult fractionation of PEGylated peptides due to the hindrance of the polymer, although a limit still remains, i.e., the easy in vivo release of PEG from these conjugates at the level of the labile ester bond (5). As a contribution to this problem, we describe here a procedure that overcomes the limits of the negative information of one method, the instability of the conjugates obtained with the succinimidyl arm, as well as the general problem connected with the difficult fractionation of PEGylated peptides or proteins. The method is based on the use of PEG-Met-Nle-OSu or PEG-Met-βAla-OSu for protein modification. The PEG conjugates so obtained are stable in blood stream, but allow the in vitro cleavage at the methionine bond by BrCN, leaving Nle or βAla bound to protein molecule as reporter group of the PEGylation site (see Figure 1). The peptide or protein, now devoid of PEG hindrance, may undergo any enzymatic and chemical treatment, or chromatographic fractionation, while the presence of the reporter is revealed by amino acid analysis, mass spectroscopy, or Edman degradation. EXPERIMENTAL PROCEDURES

Methoxypoly(ethylene glycol) (MW 5000) (PEG) was purchased from Shearwater Polymer (Huntsville, AL), PEG-p-nitrophenyl carbonate was prepared according to a method previously reported (6), and bovine insulin, methionine, p-nitrophenylchloroformiate, DL-norleucine methyl ester (Nle-OMe), β-alanine (βAla), 1-cyclohexyl3-(2-morpholinoethyl)carbodiimide metho-p-toluene-sul-

10.1021/bc000061m CCC: $20.00 © 2001 American Chemical Society Published on Web 01/17/2001

Identification of PEGylation Site in Protein

Figure 1. Conjugation of PEG-Met-Nle-OSu or PEG-Met-βAlaOSu to a protein amino group and its removal by BrCN treatment.

fonate (CMC), and 1-hydroxybenzotriazole (HOBt) were purchased from Sigma (St. Louis, MO). Triethylamine (TEA), N-hydroxysuccinimide (HOSu), 1,4-dithio-L-threitol (DTT) and lysozyme from egg white were purchased from Fluka Chemie. Trinitrobenzenesulfonic acid (TNBS) was purchased from Aldrich (Milwaukee, WI), iodoacetamide from Merck (Rahway, NJ), and cyanogen bromide (BrCN) from Pharmacia (Uppsala, SW). The nonapeptide partial sequence of glucagons was from our laboratory. Proton NMR spectroscopy was performed with a Varian Gemini 200 MHz instrument. Ultraviolet spectroscopy was performed with a Perkin-Elmer λ5 instrument. Ultrafiltration was performed with an Amicon system and YM3 membrane (cut off 3000 Da). Ionic exchange chromatography was performed on a Pharmacia LKB system using a QAE-Sephadex A50 column. Reversedphase chromatography was performed with a Shimadzu analytical and semipreparative HPLC system, using a Vydac 218TP54 and Vydac 218TP1022 column, respectively, with UV detector. Gel filtration chromatography was performed on a Pharmacia FPLC system, using a Superose 12TM column and an UV detector. Amino acid analyses were carried out after hydrolysis with HCl in vapor phase and derivatization with phenyl isothiocyanate (Fluka Chemie) (7) using a HPLC column MachereyNagel Nucleosil 120-5 C18. Maldi mass spectroscopy was performed in a Kratos Kompact MALDI 1 V5.2.0 instrument. Trinitrobenzenesulfonate (TNBS) was used as reagent for quantitative amino groups determinations following the procedure described by Habeeb (8). PEG content was evaluated colorimetrically using the iodine assay (9). Synthesis of PEG-methionine-norleucine-hydroxysuccinimidylester (PEG-Met-Nle-OSu) and of PEGmethionine-β-alanine-hydroxysuccinimidiylester (PEG-Met-βAla-OSu). (1) PEG-Met-Nle-OSu. PEG-pnitrophenyl carbonate (5 g; 0.97 mmol) was added in small portions to 724 mg (4.8 mmol; 5 equiv) of methionine dissolved in 10 mL of a 50% acetonitrile solution in water and brought to pH 8 with 676 µL (4.8 mmol; 5 equiv) of triethylamine. After overnight stirring, acetonitrile was removed under reduced pressure and the pH of the reaction mixture was adjusted to 3 using solid citric acid. p-Nitrophenol was removed by ether extraction while the product was extracted several times with chloroform. The pooled organic fractions were dried, the volume was reduced to a few milliliters, and the product, precipitated by adding 300 mL of dry diethyl ether, was collected by filtration. For further purification the product was dissolved with acidified water, extracted with chloroform and precipitated by diethyl ether. The yield of the reaction, evaluated by 0.01 N NaOH titration, was almost quantitative. 1H NMR (chloroform-d, 2000 MHz): δ 3.61 (s, PEG backbone -OCH2-); 3.41 (s, 3H, PEG CH3O-); 2.55

Bioconjugate Chem., Vol. 12, No. 1, 2001 63

(t, 2H, Met γCH2); 2.07 (s, 3H, Met CH3). In the following step 523 mg (2.888 mmol; 4 equiv) of norleucine methyl ester and a equimolecular amount of triethylamine (401 µL) were added to a solution of 10 mL of anhydrous dichlorometane, containing 3.71 g (0.72 mmol) of PEGMet-OH and mixed at 0 °C with 305 mg (0.72 mmol; 1 equiv) of 1-cyclohexyl-3-(2-morpholino ethyl) carbodiimide metho-p-toluene-sulfonate (CMC) and 97 mg (0.72 mmol; 1 equiv) of 1-hydroxybenzotriazole (HOBt). After stirring for 3 days at room temperature, the ureic derivative of CMC was filtered off, the mixture was dropped into 250 mL of dry diethyl ether, and the precipitate collected by filtration and dried. The product was further purified by ionic-exchange chromatography on a QAE Sephadex A50 column, eluted with water followed by 10 mM NaCl. The fractions containing the PEG-dipeptide were collected and lyophilized (yield: 93%). 1H NMR (chloroform-d, 200 MHz): δ 3.70 (s, 3H, OMe CH3); 3.61 (s, PEG backbone -OCH2-); 3.41 (s, 3H, PEG CH3O-); 2.55 (t, 2H, Met γCH2); 2.07 (s, 3H, Met CH3); 1.32 (m, 4H, Nle γ e δCH2); 0.88 (t, 3H, Nle CH3). In the last step, PEG-Met-Nle-OMe (1.41 g; 0.26 mmol) was stirred overnight at room temperature in a solution of methanol (5 mL) containing 20 equiv of 0.1 N NaOH (5.3 mL). The pH of the mixture was brought to 2 using 1 N HCl, and the product extracted with chloroform. The collected organic fractions were dried over anhydrous sodium sulfate, concentrated to a small volume under reduced pressure, and the product, precipitated by addition of dry diethyl ether, was filtered and dried. PEG-Met-Nle-OH was further purified by ionic-exchange chromatography on a QAE Sephadex A50 column, eluted first with water and then with a water solution of 10 mM NaCl. The fractions containing the sodium salt of the PEG-dipeptide were pooled, concentrated, and extracted by chloroform after acidification of the water solution. The purified product was finally obtained by precipitation with dry diethyl ether. Yield, calculated by 0.01 N NaOH titration, was 67%. 1H NMR (chloroform-d, 200 MHz): δ 3.61 (s, PEG backbone -OCH2-), 3.41 (s, 3H, PEG CH3O-), 2.55 (t, 2H, Met γCH2), 2.07 (s, 3H, Met CH3), 1.32 (m, 4H, Nle γ e δCH2), 0.88 (t, 3H, Nle CH3). The 1:1:1 ratio among the integration value of methoxy protons of PEG, methoxy protons of methionine and norleucine demonstrates the correct composition of the product and the complete derivatization of PEG. The identity of the product was also demonstrated by amino acid analysis following acid hydrolysis (Lys 1.01, Met 0.9, Nle 1.0). In this case, PEG content was evaluated by iodine assay (9). For a mole of PEG (MW 5000) methionine and norleucine corresponded to 0.95 and 1.01 mol, respectively. For PEG-Met-Nle-OH activation, the product (600 mg 0.11 mmol) was dissolved in 5 mL of dicloromethane, and hydroxy-succinimide (104 mg 0.9 mmol) and DCCI (185 mg, 0.9 mmol) were added. The reaction mixture was left to react for 24 h, and the product, precipitated by ethyl ether, collected by filtration and dried. The activation degree, expressed as percentage of -OSu for PEG-MetNleu, was over 95% in three different preparations. The activation degree was determined by the amino groups modification of glycine-glycine as substrate using TNBS for residual NH2 groups evaluation (10). (2) PEG-Met-βAla-OSu. This product was prepared according to a procedure different from that followed for the nor-leucine derivative preparation, namely PEG activated as p-nitrophenilchloroformiate was reacted with H-Met-βAla-OMe and the product, after alcaline hydrolysis, activated as hydroxysuccinimidylester.

64 Bioconjugate Chem., Vol. 12, No. 1, 2001

Veronese et al.

Table 1. Characterization of the Nonapeptide Derived from a Glucagon Sequence Following PEGylation with PEG-Met-βAla-OSu and Removal of the PEG Chain by BrCN Treatment

free monopeptide PEGylated form following BrCN treatment

elution volume from HPLCa (min)

molecule massb (Da)

Met mol/peptidec

βAla mol/peptidec

16.4 36.3 16.8

1091.1 11090.0 1317.4

0 2 0

2 2

a A Vydac 218TP54 reverse phase HPLC column was used. b The mass was evaluated by Maldi mass spectroscopy. c Values obtained by amino acid analysis after acid hydrolysis. These nominal values were calculated from the following amino acid analyses: For PEGylated form, Thr, 0.81; Ser, 1.81; Arg, 1.08; Tyr, 1.77; Leu, 1.03, Arg, 1.02; Met, 1.75; βAla, 2.20. For the BrCN treated product, Thr, 0.78; Ser, 1.75; Arg, 1.01; Tyr, 1.82; Leu, 0.98; βAla, 1.04.

The dipeptide was synthesized from BOC-Met-OH (0.7 g, 2.9 mmol) and β-Ala-OMe (0.618 g, 4.4 mmol) in the presence of EDC (0.566 g, 2.9 mmol) and HOBT (0.399 g, 2.9 mmol) in 5 mg of dichloromethane. After extraction with basic and acid aqueous solution, the organic phase has been dried by Na2SO4 and evaporated. 1 H NMR (chloroform-d, 200 MHz): δ 1.5 (s, 9 H, BOC); 1.9 (m, 2H,-CH2-CH2-S-CH3); 3.7 (s, 3 H, -S-CH3); 2.5 (m, 4 H, -S-CH2 + βAla-CH2-CO); 3.5 (m, -NH-CH2-βAla); 3.7 (s, 3 H, -COOCH3); 4.2 (m, 1 H, -CH-); 5.25 (d, 1 H, BOCNH); 6.75 (t, 1 H, -NH-CO). The deprotection has been accomplished by stirring the product in a mixture of 2.5 TFA and 2.5 mL of dichloromethane. H-Met-βAlaOMe was recovered by ether precipitation. 1H NMR (chloroform-d, 200 MHz): 2.1 (s, 3 H, -S-CH3); 2.15 (m, 2 H, CH2-CH2-S-CH3); 2.6 (m, 4 H, -S-CH2 + βAla-CH2CO); 7.6 (t, 1 H, NH-CO); 7.9 (bs, 2 H, NH2). The product was coupled to PEG-p-nitrophenylchloroformiate, the methylester removed and the carboxilic group activated by OSu as previously reported for PEG-Met-Nle-OSu. The amino acid analysis after acid hydrolysis and PEG evaluation carried out by colorimetric assay demonstrated the identity of the compound. For a mole of PEG (5000 MW), methionine and β-alanine were found to be 0.97 and 1.04, respectively. The activation degree was found to be over 97% in different preparations. PEG Conjugation, Followed by Its Removal, to a Nonapeptide, Lysozyme, and Insulin. (a) PEG Conjugation and Removal from H-Thr-Ser-Arg-Tyr-Ser-LysTyr-Leu-Asp-OH. A total of 41 mg (4.0 µmol) of MPEGMet-βAla-OSu was added, in small aliquots, to 0.9 mg (0.8 µmol) of the nonapeptide dissolved in 1 mL of DMSO, at pH 8 with TEA (nonapeptide: MPEG-Met-Nle-OSu molar ratio ) 1:5). After 3 days of stirring at room temperature, the pH of the reaction mixture was adjusted to 3 with 1 N HCl, and the product was extracted several times by chloroform. The chloroform was anhydrified with dry Na2SO4 and concentrated to dryness. The collected conjugate was further purified by means of a Shimadzu semipreparative HPLC system, using a reversed-phase Vydac 218TP54 column (C18 bonded phase, 0.46 × 25 cm i.d., 5 µm particle diameter). A linear gradient of aqueous 0.05% trifluoroacetic acid and 0.05% trifluoroacetic acid in acetonitrile was used and the fractions of the unique peak revealed by absorption at 280 nm and by iodine assay were pooled and concentrated to dryness. The PEGylation was verified by the disappearance of reactivity of the peptide amino groups and by the presence of two βAla and Met residues revealed by amino acid analysis after acid hydrolysis (see Table 1). For the removal of bound PEG, 1 mg (85.2 µmol) of PEG-Met-βAla-nonapeptide was treated with 100 equiv of BrCN in aqueous 70% formic acid (11). After 24 h standing at room temperature, the mixture was poured into 10 vol of water and lyophilized; this procedure was repeated twice. The product was fractionated by a Shi-

madzu analytical HPLC system, using a reversed-phase Vydac 218TP54 column (C18 bonded-phase, 0.46 × 25 cm i.d., 5 µm particle diameter). The eluent was a linear gradient of aqueous 0.05% trifluoroacetic acid and 0.05% trifluorocacetic acid in acetonitrile. (b) PEG Conjugation and Removal from Lysozyme. A total of 42.2 mg (7.8 µmol) of mPEG-Met-Nle-OSu was added, in small aliquots while stirring, to 3.75 mg (0.26 µmol) of lysozyme from egg white dissolved in 1 mL of 0.2 M borate buffer, pH 8.5. The molar ratio of lysozyme amino groups to mPEG was 1:5. After 30 min of stirring at room temperature, the conjugate was purified by means of a semipreparative FPLC system, using a gel filtration Superose 12TM column (1.5 × 25 cm) eluted with 0.01 M phosphate buffer and 0.15 M NaCl, pH 7.2. The fractions were analyzed for lysozyme elution by absorption at 280 nm and iodine assay for PEG elution. The fractions containing the modified lysozyme were collected and concentrated to a small volume by ultrafiltration with an Amicon system using a YM3 membrane (cut off 3000 Da). The concentrated solution was diluted with water and ultrafiltered. This procedure was repeated and the product was finally lyophilized. For PEG removal, 0.04 µmol of modified lysozyme was treated with 600 equiv of BrCN in aqueous 70% formic acid and left stirring for 24 h at room temperature. The mixture was poured into 50 mL of water and lyophilized; the procedure was repeated twice. The product obtained was analyzed by gel filtration column by a FPLC system as above, to compare the chromatography profiles. (c) Insulin Modification by PEG-Met-Nle-OSu Followed by Removal of PEG and Identification of the Sites of PEGylation. PEG-Met-Nle-OSu was added to 7 mg (1 µmol) of bovine insulin dissolved in 1 mL of DMSO until a 3:1 PEG: insulin molar ratio was reached. The mixture was left stirring for 5 h at room temperature. The conjugate was purified by means of a Shimadzu preparative HPLC system, using a reversed-phase Vydac 218TP1022 column (C18 bonded phase, 2.2 × 25 cm i.d., 10 µm particle diameter) and a linear gradient of aqueous 0.05% trifluoroacetic acid and 0.05% trifluoroacetic acid in acetonitrile, monitoring the absorption at 226 nm wavelength. Three peak fractions of conjugated insulin were collected (at 21.867, 23.275, and 24.158 min of elution) and lyophilized. The product of each peak, called insulin A, insulin B, and insulin C, were separately treated with BrCN in aqueous 70% formic acid to remove PEG. The molar ratio between insulin and BrCN was 1 to 200. After 24 h stirring at room temperature, the mixture was poured into 10 vol of water and freeze-dried; the procedure was repeated twice. The obtained products were analyzed and purified by means of a Shimadzu analytical HPLC system, using a reversed-phase Vydac 218TP54 column (C18 bonded phase, 046 × 25 cm i.d., 5 µm particle diameter) with a linear gradient of aqueous 0.05% trifluoroacetic acid and 0.05% trifluoroacetic acid


Figure 2.

1H


NMR spectrum (CDCl3, 200 MHz) of mPEG-Met-Nle-OH.

in acetonitrile. To determine the position of PEG in the insulin chains, the disulfide bonds were reduced and carboxymethylated. To this aim, 500 µg (0.08 µmol) of each Nle-insulin sample, obtained from PEG-Met-Nleinsulin A, B, and C after BrCN treatment, were dissolved in 250 µL of Tris-HCl buffer, pH 7.5, containing 2 mM EDTA and 6 M Gdn-HCl; 3.8 mg of 1,4-dithio-L-threitol were added, and the mixture was left standing for 2 h at 37 °C. Finally, 9.2 mg of iodoacetamide were added and the mixture was left for 1 h at 37 °C. The solutions were lyophilized, and the products fractionated by means of a Shimadzu semipreparative HPLC system, using a reversed-phase Vydac 218TP54 column as described above. The eluted products monitored by absorption at 226 nm wavelength revealed in all the samples the presence of two main peaks, corresponding to the reduced carboxymethylated R and β chains, respectively. The two polypeptides were separated, recovered, and analyzed by amino acid analysis, mass spectrometry and eventually Edman degradation (12). RESULTS

Synthesis of PEG-Met-Nle-OSu and PEG-MetβAla-OSu. The synthesis of the two new PEG derivatives may be carried out by sequential amino acid addition to PEG according to the polymer supported liquid-phase synthesis procedure (13) or by the linkage of PEG to the dipeptide, followed by the carboxilic groups activation. Here the first method was employed for PEG-Met-Nle, while the second for PEG-Met-βAla preparation. The identity of the polymer derivatives was verified qualitatively and quantitatively by 1H NMR spectroscopy (see Figure 2 for PEG-Met-Nle-OH) and by evaluation of the three components: PEG by iodine assay, methionine and nor-leucine (14) or β-alanine by amino acid analysis after acid hydrolysis. Using glycil-glycine as model (10), it was verified the degree of activation as hydroxysuccinimidylester. It also demonstrated that the rate of aminolysis of OSu of these

new PEG derivatives was the same of the known simple PEG-Nle-OSu (14). Protein and Peptide Conjugation with the New PEG-Derivatives Followed by PEG Removal by BrCN. The peptide and protein conjugation of the new PEG-peptide arms, activated as succinimidyl ester, followed by deblocking by BrCN treatment, was verified using three different models: a nonapeptide, the small protein insulin and the 12 kDa protein lysozyme. (A) The nonapeptide H-Thr-Ser-Arg-Tyr-Ser-Lys-TyrLeu-Asp-OH, a partial sequence of glucagon, was chosen since it presents two reactive amino groups. Following treatment with PEG-Met-βAla-OSu, 5 kDa, the mass of the nonapeptide, evaluated by Maldi mass spectroscopy, increased from the 1091.11 kDa of the free peptide to a value of about 11410 kDa. The mass spectrometry peak was broad (spectrum not shown) due to the known polydispersity of PEG. The increase in mass, of approximately 10 kDa, corresponds to the conjugation of two PEG-Met-βAla chains and is accompanied by the disappearance of amino group reactivity, demonstrated by TNBS reaction 12. Contemporaneously an increase in reversed-phase HPLC elution volume from 16.4 to 36.3 mL was found. Following BrCN treatment, a sharp peak with a mass of 1233.26 kDa, corresponding to the nonapeptide mass plus two βAla residues could be observed by mass spectrometry. At the same time the HPLC elution dropped to 16.8 mL, a volume close to that of the free nonapeptide (16.4 mL). Furthermore the amino acid analysis showed the presence of two βAla residues per peptide molecule and no trace of methionine. These values are summarized in Table 1. (B) Figure 3 reports the elution from a GPC column of three lysozyme species: native, PEGylated and after BrCN treatment. The PEGylated lysozyme (b) was eluted by gel filtration chromatography at the 12 mL, much ahead of the nati)ve enzyme, 18 mL, (a) due to its large increase in mass. Six PEG chains were bound as revealed by mass spectrometry, amino acid analysis following acid


Figure 3. Elution peaks of lysozyme (a), PEGylated lysozyme (b) and the conjugation product after PEG removal by BrCN treatment (c), from a GPC column.

hydrolysis, and amino groups tritation by TNBS. The sample obtained after deblocking with BrCN. Nle conjugated lysozyme (c), was eluted 17.8 mL a volume close to that of the native protein (18 mL). The small difference in elution being related to the difference in mass of the two species: 14 388 kDa of native lysozyme, as compared to 15 066 kDa of lysozyme plus the six nor-leucine residues as confirmed by amino acid analysis. The data regarding a small peptide or a model protein, reported in A and B, are in favor of the possibility to link these novel PEGs to polypeptides as well as of their quantitative removal by BrCN treatment, leaving Nle or βAla as reporter amino acids bound to the starting molecule. Localization of Conjugation Site in PEGylated Insulin. Insulin was the third polypeptide model chosen to test the suitability of the proposed methodology, because in this case different PEGylated isomers may be obtained in the conjugation reaction. Figure 4b, shows that when insulin is PEGylated in dimethyl sulfoxide at a 1 to 1 molar ratio of polymer to insulin available amino group, three conjugation products are obtained. They are eluted later as compared to native insulin (18.3 mL, see Figure 4a) since the presence of PEG. Colorimetric amino group titration and amino acid analysis following acid hydrolysis on the basis of methionine and nor-leucine presence, demonstrated that peak A (eluted at 21.9 min) corresponds to mono-PEGylated insulin. Peak B (eluted at 23.3 min) corresponds to an insulin specieswith two PEG molecules bound. The third peak C (eluted at 24.1 min) is the three-PEGylated product (see Figure 4b). This last product (c) being modified at all available insulin amino groups (Gly 1 of R chain, Phe 1 and Lys

Veronese et al.

29 of β chain), did not need further studies for its characterization. As shown in Figure 5a, when the mono-PEGylated product (peak A of Figure 4) was treated with BrCN, HPLC fractionation yielded a single peak that, by mass spectrometry, showed a mass of 5845.32 Da, corresponding to the mass of one insulin plus 1 residue of norleucine (expected value 5846.68 Da, see Figure 5b). The elution from HPLC, 18.4 mL (see Figure 5a) is close to that of native insulin, 18.3 mL (see Figure 4a), the difference is due to the fact the two products differ for only one amino acid (Nle). To identify the position of PEGylation, the two insulin chains were separated by reduction followed by carboxymethylation. It was found that only the heavy chain was PEGylated, showing a mass of 3629.81 Da (the expected mass of β chain plus 1 Nle residue is 3627.59 Da, Figure 6). To identify whether the PEGylation was at Phe 1 or at Lys 29, one step of Edman was carried out. Mass spectrometry demonstrated the release of one phenylalanine and one norleucine residue (expected mass 3367.22, found 3369.23) (see Figure 7). This demonstrated that the R amino group of phenylalanine was free (Phe was in fact released by Edman) and consequently that PEG was bound to -lysine (to note that Nle also was released by CNBr from the Nle-Lys bond). The di-PEGylated insulin, corresponding to the major peak B of Figure 4, was also purified to homogeneity and PEG was removed (see Figure 8a). The dePEGylated product yielded a mass of 5954.09 that corresponds to insulin plus 2 Nle residues (see Figure 8b, expected value 5959.86). After reduction and carboxymethylation, the two chains were separated by HPLC. They were identified by mass spectrometry: one showed a mass of 3628.95 Da, already found in the peak A of Figure 4, corresponding to monopegylation; the second, of 2689.06 Da (see Figure 9) corresponds instead to a monopegylated light chain (expected value 2681.18). While for the R chain there is only one position of PEGylation at the level of glycine in position 1 of the peptide, for the β chain two positions are possible, as described above. One step of Edman degradation showed that the R amino group was free since phenylalanine was released demonstrating that the PEG binding involved lysine 29, as for the previous peak B of insulin. From these experiments, it was therefore possible to assign the position of PEGylation to the mixture of products obtained in insulin conjugation. It was contemporaneously possible to demonstrate that, under these conditions of reaction, the order of reaction is lysine 29 of β chain (the one always found), followed by glycine 1 of R chain, being phenylalanine 1 of β chain the less reactive. These results are in agreement with the known properties of insulin residues in which the low reactivity of Phe 1 of R chain is ascribed to its pKa (15, 16). Extensive PEGylation of this residue was obtained only by a previous protection by BOC of the other two more reactive amino groups (17-19). CONCLUSIONS

The data obtained using a nonapeptide, insulin and lysozyme as models, strongly support the use of PEG with Met-X as peptide arm for the modification of proteins. Convenient X are amino acids non natural or non frequent in proteins, norleucine and β-alanine standing as the preferred ones. This functionalized polymers present unique advantages for the analytical characterization of PEG conju-



Figure 4. (a) Elution pattern from reversed-phase column of insulin and (b) of the reaction mixture of insulin conjugation with PEG-Met-Nle-OSu. Peaks A, B, and C correspond to insulin with 1, 2, and 3 PEG chains bound per insulin molecule.

Figure 5. (a) Elution pattern of mono-PEGylated insulin, corresponding to peak A of Figure 4, following purification and treatment with BrCN to release PEG, to leave the reporter Nle only bound to the peptide; (b) mass spectrum of the Nle insulin reported in panel a. Found: 5845.32 Da. Expected: 5846.68 Da.

gates, since the hindered PEG may be specifically removed by BrCN treatment leaving bound norleucine or β-alanine as reporter group bound to the proteins. These

de-PEGylated proteins are more easily fractionated in their constituent isomers of the PEG containing products, since their lower hindrance. This is an important aspect


Veronese et al.

Figure 6. Mass spectrum of the reduced carboxymethylated insulin heavy chain carrying one norleucine residue. The sample comes from the peak A of Figure 4 after purification, dePEGylation (see Figure 5a) and separation of the chains. Found: 3629.81 Da. Expected: 3627.59 Da.

Figure 7. Mass spectrum of heavy chain of insulin of peak A of Figure 4, after dePEGylation reduction, carboxymethylation and one step of Edman degradation. Found: 3369.23 Da. Expected: 3367.22 Da.

because it is known that PEGylation yields heterogeneous mixture, that only in few lucky cases may be separated.

Furthermore the proteins, devoid of PEG, may undergo any chemical or enzymatic cleavage needed for sequence evaluation and the labeled peptides are easily identified



Figure 8. (a) Elution pattern of di-PEGylated insulin, corresponding to peak B of Figure 4, following treatment with BrCN to release PEG. (b) Mass spectrum of Nle insulin reported in panel a, showing that two Nle residues are bound to the protein. Found: 5954.09 Da. Expected: 5959.86 Da.

Figure 9. Mass spectrum of light chain found of dePEGylated, reduced and carboxymethylated insulin of peak B in Figure 4. Found: 2689.06 Da. Expected: 2681.18 Da.

on the basis of norleucine or β-alanine detectable by amino acid analysis or mass spectrometry. In this respect norleucine appears ideal as reporter residue in mass spectrometry analysis for its higher molecular weight. β-Alanine is less convenient in this respect, but it

presents the advantage, over norleucine, of not being released by Edman degradation, thereby allowing its localization in a peptide where multiple lysine residues are present (20). The BrCN based removal of PEG offers an additional


advantage for the identification of the site of PEGylation. As well-known BrCN, in the same step of dePEGylation, cleaves also the protein at the level of methionine, if present in the primary sequence, yielding shorter peptides where the reporter groups may be directly identified. This peptide cleavage did not take place in our models nonapeptide or in insulin, both devoid of methionine. In lysozyme, on the other hand, the cleavage took place, but the peptides remained still linked together by disulfide bridges. To verify the suitability of these new PEGs for a therapeutic application of their protein conjugates, studies to evaluate the stability in blood of PEG-Met-Nle and PEG-Met-βAla conjugates are presently under way, and will be reported separately. ACKNOWLEDGMENT

This research was carried out with partial contribution from Italian CNR, Targeted Project on Biotechnology and Bioinstrumentation. LITERATURE CITED (1) Caliceti, P., Schiavon, O., and Veronese, F. M. (1999) Design and characterisation of new polymer conjugates of GHRH. Acta Technol. legis Med. 10, 86, and European patent no. 97.12.1264. (2) Clark, R., Olson, K., Fuh, G., Marian, M., Mortensen, D., Teshima, G., Chang, S., Chu, H., Mukku, V., Canova-Davis, E., Somers, T., Cronin, M., Winkler, M., and Wells, J. A. (1996) Long-acting growth hormones produced by conjugation with poly(ethylene glycol). J. Biol. Chem. 271, 21969-21977. (3) Monkarsh, S. P., Spence, C., Porter, J. E., Palleroni, A., Nalin, C., Rosen, P., and Bailon, P. (1999) Isolation of positional isomers of monopoly(ethylene glycol)ylated interferon-2R and the determination of their biochemical and biological characteristics. In Poly(ethylene glycol). Chemistry and Biological Applications (J. M. Harris and S. Zalipsky, Eds.) pp 207-216, ACS Symposium Series 680. (4) Vestling, M. M., Murphy, C. M., Keller, D. A., Fenselau, C., Dedinas, J., Ladd, D. L., and Olsen, M. A. (1993) A strategy for characterisation of poly(ethylene glycol)-derivatized proteins. A mass spectrometric analysis of the attachment sites in poly(ethylene glycol)-derivatized superoxide dismutase. Drug Metab. Dispos. 21, 911-917. (5) Zalipsky, S. (1995) Chemistry of poly(ethylene glycol) conjugates with biologically active molecules. Adv. Drug Delivery Rev. 16, 157-182.

Veronese et al. (6) Veronese, F. M., Largajolli, R., Boccu`, E., Benassi, C. A., and Schiavon, O. (1985) Surface modification of proteins. Activation of monomethoxy-poly(ethylene glycol)s by phenylchloroformates and modification of ribonuclease and superoxide dismutase. Appl. Biochem. Biotechnol. 11, 141-152. (7) Bidlingmeyer, B. A., Cohen, S. A., and Tarvin T. L. (1984) Rapid analysis of amino acids using precolumn derivatization. J. Chromatogr. 336, 93-104. (8) Habeeb, A. F. (1966) Determination of free amino groups in proteins by trinitrobenzensulphonic acid. Anal. Biochem. 14, 328-336. (9) Sims, G. E. C., and Snape, T. J. (1980) A method of estimator of polyethylenglycol in plasma protein fraction. Anal. Biochem. 107, 60-63. (10) Snyder, S. L., and Sobocinsky, P. Z. (1975) An improvement 2,4,6 trinitrobenzensulphonic acid method for determination of amines. Anal. Biochem. 64, 284-288. (11) Witkop, B. (1968) Chemical cleavage of proteins. Sciences 16, 318-326. (12) Konigsberg, W. (1972) Subtractive Edman degradation. Methods Enzymol. XXV, 326-332. (13) Bonora, G. M. (1995) A High-Efficiency Liquid Phase (HELP) for the large-scale synthesis of the oligonucleotides. Appl. Biochem. Biotechnol. 54, 3-17. (14) Sartore, L., Caliceti, P., Schiavon, O., and Veronese, F. M. (1991) Enzyme modification by MPEG with an amino acid or peptide as spacer arms. Appl. Biochem. Biotechnol. 27, 4552. (15) Jeong, S. Y., Kim, S. W., Eenink, M. J. D., and Feijen, J. (1984) Self-regulating insulin delivery systems, I. Synthesis and characterisation of glycosylated insulin. J. Controlled Release 1, 57-66. (16) Baudysˇ, M., Uchio, T., Mix, D., Wilson, D., and Kim, S. W. (1995) Physical stabilisation of human insulin by glycosylation. J. Pharm. Sci. 84, 28-33. (17) Uchio, T., Baudysˇ, M., Liu, F., Song, S. C., and Kim, S. W. (1999) Site-specific insulin conjugates with enhanced stability and extended action profile. Adv. Drug Delivery Rev. 35, 289306. (18) Caliceti, P., and Veronese, F. M. (1999) Improvement of the physicochemical and biopharmaceutical properties of insulin by poly(ethylene glycol) conjugation. S. T. P. Pharma Sci. 9, 107-113. (19) Hinds, K., Koh, J. J., Joss, L., Liu, F., Baudysˇ, M., and Kim, S. W. (2000) Synthesis and characterisation of poly(ethylene glycol)-insulin conjugates. Bioconjugate Chem. 11, 195-201. (20) Sergi M., personal communication.

BC000061M

New PEGs for Peptide and Protein Modification, Suitable for

Recommend Documents