Design and Chemical Synthesis of a Homogeneous Polymer-Modified Erythropoiesis Protein

See allHide authors and affiliations

Science  07 Feb 2003:
Vol. 299, Issue 5608, pp. 884-887
DOI: 10.1126/science.1079085


We report the design and total chemical synthesis of “synthetic erythropoiesis protein” (SEP), a 51-kilodalton protein-polymer construct consisting of a 166-amino–acid polypeptide chain and two covalently attached, branched, and monodisperse polymer moieties that are negatively charged. The ability to control the chemistry allowed us to synthesize a macromolecule of precisely defined covalent structure. SEP was homogeneous as shown by high-resolution analytical techniques, with a mass of 50,825 ±10 daltons by electrospray mass spectrometry, and with a pI of 5.0. In cell and animal assays for erythropoiesis, SEP displayed potent biological activity and had significantly prolonged duration of action in vivo. These chemical methods are a powerful tool in the rational design of protein constructs with potential therapeutic applications.

Optimal performance of protein pharmaceuticals is primarily determined by an appropriate balance among specificity, potency, and pharmacokinetic properties. However, the inability to produce homogeneous posttranslationally or chemically modified recombinant proteins has hampered their optimization to date. Both natural and recombinantly produced glycoproteins occur in multiple glycoforms, and this heterogeneity is reflected in different potency and pharmacokinetic properties (1–5). Using conventional polymers such as polyethylene glycol for modification of recombinant proteins introduces heterogeneity both in the polymers attached and in the attachment sites, with a consequent variability in biological properties (6–9).

We envisioned that recent advances in the total chemical synthesis of proteins (10–14) and in the chemical synthesis of monodisperse polymers (15) could be exploited to address this problem through the design and synthesis of polymer-protein constructs of defined covalent structure. Such precise atom-by-atom control of the covalent structure provides a unique way to systematically correlate molecular structure with function and enables the fine-tuning of the biological properties of a target protein of pharmaceutical interest.

Here, we describe the design, total chemical synthesis, and biological activity of a monodisperse, polymer-modified macromolecule, synthetic erythropoiesis protein (SEP). SEP had comparable specific activity in vitro, but superior duration of action in vivo, relative to human erythropoietin (Epo), a natural glycoprotein hormone that regulates the proliferation, differentiation, and maturation of erythroid cells (16).

We designed SEP to be a potent effector of the Epo receptor and to have prolonged circulation lifetime. The target structure of SEP is shown in Fig. 1. SEP contains a 166-amino-acid polypeptide chain (Fig. 1A) similar to the sequence of Epo (16) but differs significantly in the number and type of attached polymers (17). In particular, two branched polymer moieties of a precise length and bearing a total of eight negative charges (Fig. 1B) were designed for site-specific attachment through an oxime bond to two noncoded amino acid residues [Lys24 (N ɛ-levulinyl) and Lys126(N ɛ-levulinyl)] that were incorporated into the polypeptide chain. These sites correspond to two of the four glycosylation sites found in Epo. Each branched precision polymer was designed to have the following: (i) a chemoselective linker, (ii) a hydrophilic spacer consisting of one polymer repeat unit, (iii) a core structure with four branch points, (iv) a linear polymer with 12 repeat units attached to each branch point, and (v) a negative charge–control unit at the end of each linear polymer. These precision polymer moieties were envisioned to give the SEP molecule a large hydrodynamic radius and a net negative charge at physiologic pH for optimal potency and prolonged duration of action in vivo (18). We also hoped that the precision polymers would enhance stability by shielding the folded polypeptide chain from attack by proteolytic enzymes, and that it would reduce immunogenicity, while retaining full biological potency.

Figure 1

Molecular structure of SEP. (A) Diagram of SEP indicating its primary amino acid sequence, noncoded amino acids, and disulfide bonds. The levulinyl chemoselective sites of polymer attachment are coded in blue, and the carboxymethyl-modified cysteine residues are coded in green. The three ligation sites are circled in red. (B) Structure of the branched, negatively charged, precision-length polymer moiety. The aminooxyacetyl chemoselective site of protein attachment is coded in blue, and the charge control center is coded in red. (C) Scheme for the synthesis of SEP by chemical ligation. The branched precision polymer was first attached to a predetermined site in each of the segments SEP(1-32) and SEP(117-166) by oxime-forming ligation at noncoded Lys (N ε-levulinyl) residues incorporated during peptide synthesis. The 166-amino-acid residue polypeptide-polymer construct was then assembled by sequential amide-forming native chemical ligation (NCL) (11) of the four unprotected synthetic peptide segments. The full-length polypeptide-polymer construct was folded with the concomitant formation of two intramolecular disulfide bonds to yield the SEP protein, which is represented as a cartoon. The branched precision polymers are scaled to represent the approximate hydrodynamic volume of SEP as estimated by size-exclusion chromatography (see text).

SEP was assembled by sequential chemo selective ligation reactions from four individual peptide segments and two branched precision polymers as shown in Fig. 1C. Using this synthetic strategy, we could purify and characterize each intermediate, thus ensuring both accurate assembly of the target molecule and high purity of the final product. We used standard solid-phase peptide synthesis methods to prepare the peptide segments (19, 20) and novel chemistry (15) to make branched precision polymer constructs in a monodisperse form (seeFig. 1B). Two mutually compatible ligation chemistries were used to assemble the final construct from the polypeptide chain and the precision polymer building blocks (20). Oxime-forming ligation (12) between an aminooxy group on the precision polymer and two ketone-bearing Lys (N ɛ-levulinyl) residues was used to link the precision polymer to peptide building blocks at positions 24 and 126 (shown in blue in Fig. 1A). Subsequently, consecutive thioester- mediated amide-forming native chemical ligations at cysteine residues (11) were used to connect these building blocks at positions 33, 89, and 117 (cysteine residues circled in red in Fig. 1A) to form the full-length construct. After the first two ligation steps, the cysteine residues at positions 89 and 117 were alkylated with bromoacetic acid to form noncoded amino acid side chains that were electronically and sterically similar to the glutamate residues found at the comparable positions in Epo (21). The SEP protein was then folded with concomitant formation of two disulfide bonds with the aid of a redox shuffling agent (22). Cumulative solid-phase peptide synthesis and purification yields were 15 to 30%. Typical oxime-forming ligation yields were 30 to 50%, typical native chemical ligation yields were 40 to 70%, and the cumulative yield for the folding and purification processes was 25 to 40%. These yields correspond to recovered material after each step as a percentage of the theoretical yield based on the limiting starting reagent. The process has been scaled by a factor of 100 to date with comparable yields. Sufficient material (>100 mg) was produced to characterize the physicochemical and biological properties of SEP.

Analytical data for SEP are presented in Fig. 2. In SDS–polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 2A), SEP migrates as a single sharp band with an apparent molecular mass of ∼73 kD (23). The electrospray ionization mass spectrum of SEP (Fig. 2, B and C) showed multiple charge states derived exclusively from the target molecule, and the observed mass (50,825 ± 10 daltons) confirmed the covalent structure shown in Fig. 1 (24). The isoelectric focusing gel (IEF) of SEP (Fig. 2D) showed a single sharp band at an isoelectric point (pI) of ∼5.0, in accord with the predicted pI based on the molecular composition. The presence of a single symmetric peak in the high-performance liquid chromatography (HPLC) trace of SEP (Fig. 2E) confirmed its high purity. There was no evidence of impurities due to incomplete or excessive precision polymer modification or due to partially assembled SEP by any of the techniques used. The correct attachment of the precision polymer at the predetermined sites, and the correct formation of the two desired disulfide bonds, were confirmed by a combination of tryptic peptide mapping, electrospray mass spectrometry, and NH2-terminal sequencing (25). The CD spectrum of SEP (Fig. 2F) was consistent with the helical secondary structure of a correctly folded long-chain cytokine (26). These data unequivocally demonstrated that the SEP molecule was monodisperse, was pure, had a precisely defined covalent structure, and was correctly folded (27).

Figure 2

Analytical characterization of SEP. (A) Tris-glycine SDS-PAGE gel (4 to 15%) of folded, purified SEP under nonreducing conditions. The gel was stained for protein with Coomassie blue. Lane 1, molecular mass standard; lane 2, SEP. Observed apparent molecular mass ∼73 kD. (B) Electrospray ionization mass spectrum of folded, purified SEP showing multiple protonation states all arising from a single molecular species. (C) Reconstructed electrospray ionization mass spectrum of folded, purified SEP. Observed mass 50,825 ± 10 daltons. (D) Isoelectric focusing gel of purified, folded SEP. Lane 1, IEF standard; lane 2, SEP. Observed pI ≈ 5.0. (E) Reversed-phase HPLC chromatogram of folded, purified SEP. (F) Circular dichroism spectrum of SEP in 20 mM H2PO4 (counterion sodium), pH 7.

The biological activity of SEP was evaluated in several assays (Fig. 3). A cell proliferation assay using a factor-dependent cell line (28) (Fig. 3A) confirmed the biological activity of SEP on the Epo receptor. In repeated trials using preparations from different syntheses, SEP showed in vitro activity similar to that of Epo (29). This is in marked contrast to typical results for the modification of recombinant growth factors with polymers such as polyethylene glycol, where the in vitro activity of the modified protein can drop dramatically when large polymers are attached (6, 3032).

Figure 3

Biological characterization of SEP. (A) Growth of UT-7/Epo cells in the presence of SEP or glycosylated recombinant human Epo at various concentrations. Stock solutions of the proteins in Iscove's modified Dulbecco's medium, 10% fetal bovine serum, glutamine, and antibiotics, and serial twofold dilutions of this stock solution were added to multiwell plates with human UT-7/Epo cells (28). The plates were incubated at 37°C, and cell proliferation was assessed by methylthiazoltetrazolium assay after 5 days. Results are the means of triplicates of one representative experiment. (Band C) In vivo hematopoietic activity of SEP and Epo in normal mice. Groups of five animals were dosed intravenously three times a week (B) or once a week (C) with the protein concentrations indicated. Blood samples were collected on days 3, 5, 10, 12, 18, 20,* 26, 33, and 40 for evaluation of the effect on hematologic parameters. Hematocrit was determined by measurement of the packed cell volume for each time point immediately after blood collection. (D) Pharmacokinetics of SEP and Epo in the rat: After a single intravenous dose at 5 μg/kg of either SEP or Epo, blood was collected via retro-orbital sinus puncture. The serum drug concentration was assayed by ELISA (enzyme-linked immunosorbent assay) with a commercially available kit for the detection of Epo (R&D Systems, Minneapolis, MN) according to the manufacturer's directions. Error bars correspond to the standard deviation for all rats at a particular time point.

The in vivo potency of SEP was also compared with Epo in several animal models. The activity of SEP and Epo in normal mice after intravenous injection three times a week or once a week is shown in Fig. 3, B and C, respectively. SEP clearly showed superior hematopoietic activity compared with Epo under both dosing regimens. The most likely explanation for the increased in vivo potency of SEP despite in vitro activity similar to that of Epo was that the presence of the precision polymer had a prolonging effect on the lifetime of SEP in the circulation compared with Epo. This was supported by pharmacokinetic data (Fig. 3D) obtained in the rat. After intravenous injection, SEP remained in the circulation two to three times as long as Epo (Table 1) (33). These data show that SEP both is a potent effector of red blood cell formation and has prolonged duration of action in vivo.

Table 1

Pharmacokinetic profile of SEP and Epo in rats.V d, volume of distribution; Cl, clearance; MRT, mean residence time; t 1/2β, terminal elimination half-life.

View this table:

It is a long-standing goal of chemical protein synthesis to generate proteins with novel properties (13, 14). In the work reported here, we demonstrate that it is now possible to design and produce polymer-modified proteins that have full biological potency and increased in vivo lifetimes. The ability to construct homogeneous protein therapeutics such as SEP enables the systematic exploration of structure-function relationships, and consequent fine-tuning of the biological properties of the protein of interest. Such molecules are not accessible with current recombinant DNA–based protein expression or by post-expression protein modification with polyethylene glycol (69, 34, 35).

Chemical synthesis is free of inherent biological contamination (nucleic acids, viruses, prion proteins, etc.). It uses readily available building blocks (synthetic peptides, precision-length polymers) and is scalable. Also, it enables complete control over design, incorporation of noncoded elements, and the precision modification of the protein of interest. Chemical protein synthesis thus addresses the known shortcomings of existing protein therapeutics and provides a tool for the rapid and effective development of new protein therapeutic leads.

Supporting Online Material

Materials and Methods

  • * To whom correspondence should be addressed. E-mail: Gkochendoerfer{at}

  • Present address: Department of Biochemistry and Molecular Biology, Faculty of Medicine, University of British Columbia, 2146 Health Sciences Mall, Vancouver, BC V6T 1Z3, Canada.

  • Present address: Diosynth RTP, Inc., 3000 Weston Parkway, Cary, NC 27513, USA.

  • § Present address: Applied Biosystems, 850 Lincoln Centre Drive, Foster City, CA 94404, USA.

  • || Present address: Celera Genomics, 180 Kimball Way, South San Francisco, CA 94080, USA.

  • Present address: Midwest Biotech, 12690 Ford Drive, Fishers, IN 46038–1151, USA.

  • # Present address: GeneProt Inc., 2 Pré-de-la-Fontaine, 1217 Meyrin/Geneva, Switzerland.

  • ** Present address: Departments of Biochemistry & Molecular Biology, and The University of Chicago, Chicago, IL 60637, USA.


Stay Connected to Science

Navigate This Article