Probing the stability of nonglycosylated wild-typeerythropoietin protein via reiterative alanine ligationsJohn A. Brailsforda and Samuel J. Danishefskya,b,1

aBio-Organic Chemistry Laboratory, Molecular Pharmacology and Chemistry Program, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue,New York, NY 10065; and bDepartment of Chemistry, Columbia University, New York, NY 10027

Contributed by Samuel J Danishefsky, February 20, 2012 (sent for review December 21, 2011)

Nonglycosylated erythropoietin bearing acetamidomethyl protect-ing groups at the cysteine residues has been synthesized via che-mical methods. Alanine ligation was used to assemble four peptidefragments, themselves prepared by solid phase peptide synthesis.This work outlines a route for the synthesis of homogeneous gly-cosylated erythropoietin.

desulfurization ∣ solid phase peptide synthesis

Erythropoietin (EPO) is a glycoprotein hormone that is respon-sible for the regulation and maintenance of erythrocytes (1).

Binding of EPO, which is produced in the kidney, to the EPOreceptor (EPO-R) located in the bone marrow, stimulates theproduction of erythrocytes by the process of erythropoiesis. EPOderived from recombinant methods is used for the treatment ofanemia associated with renal failure, cancer, and chemotherapy(2). Production of recombinant EPO, using genetically modifiedChinese hamster ovary (CHO) cells, results in a heterogeneousmixture of glycoforms—molecules that differ with respect to gly-can structure.

EPO (1) is composed of a 166-amino acid “backbone” bearingN-linked glycans at Asn residues 24, 38, and 83 and an O-linkedglycan at Ser residue 126 (Fig. 1) (3, 4). EPO contains twodisulfide bonds: one between cysteine residues 7 and 161, andthe second between residues 29 and 33. Oligosaccharides areappended to the growing protein chain during translation in theendoplasmic reticulum, then undergo “remodeling” in the Golgicomplex (5). The glycans are thought to be involved in the fold-ing, secretion, proteolytic stability, metabolic fate, and biologicalactivity of the glycoprotein (5, 6).

A number of studies have explored the relationship betweenEPO glycosylation and its biological activity (6, 7). EPO usedin those studies was obtained from genetically engineered organ-isms. In one key finding, it was observed that EPO glycoformsbearing fewer N-linked glycans exhibited decreased in vivo activ-ity and increased in vitro activity in a radio-labeled competitionassay (6). Another study suggests that increasing the levels ofoligosaccharide sialidation leads to enhanced in vivo biologicalactivity, to longer serum half-life, and, perhaps surprisingly, todecreased affinity toward EPO-R (7).

These studies have not been able to rigorously reveal the effectof the glycan structure on biological activity and EPO:EPO-Rbinding affinity. Elucidation of this information has been compli-cated by the unavailability of single EPO glycoforms. Genetic en-gineering routes to EPO provide only horrific glycoform mixturesthat cannot be purified, thereby complicating efforts to determinestructure-activity relationships (SAR) within the oligosaccharidedomain.

Progress in the area of glycoprotein chemical synthesis hascreated unique opportunities for addressing this challenging pro-blem (8, 9). The discovery of native chemical ligation (NCL)by Kent was a watershed moment in the fields of peptide and pro-tein synthesis (10). The NCL concept has been extended to theligation of peptides at alanine (11–13), threonine (14), leucine(15, 16), lysine (17), and phenylalanine (18). These noncysteineNCL-inspired ligations have a common theme: (i) thiol-mediated

ligation between a C-terminal thioester (2) and an N-terminal ami-no acid surrogate containing a temporary thiol functionality (3);followed by (ii) removal of the thiol group of the resultant peptidevia metal-based or metal-free desulfurization (MFD) (13), to yieldthe native peptide sequence (see 4 → 5, Scheme 1). These noncys-teine NCL-inspired ligations, many of which were developed in ourlaboratory, would seem to be of great value in the chemical synth-esis of proteins. EPO protein is a particularly challenging case, inthat the four native cysteine residues are positioned near the ter-mini, thus rendering them less attractive as ligation sites.

In the research described below, the total synthesis of nongly-cosylated EPO was identified as our goal. Since the glycans playa significant role in the stability and solubility of EPO, it wouldbe of great interest to learn about the behavior of the nonglyco-sylated EPO protein as a baseline in documenting the effects ofglycosylation on EPO’s chemical viability and biological perfor-mance. In essence, we sought to determine the chemical stabilityand overall manageability of nonglycosylated EPO protein. Wehoped to gain access to the intended substrates via a concise che-mical synthesis. Earlier routes to nonglycosylated EPO involvedeukaryotic recombinant gene expression followed by enzymaticdegradation of the oligosaccharides (6), and recombinant geneexpression in Escherichia coli (19). It should also be noted thatthe chemical synthesis of several EPO inspired constructs havebeen reported (20, 21).

At the same time, we aimed to evaluate the scope and limita-tions of new methodologies we introduced in simpler systems,which in our judgment serve to enhance the power of Kent’s NCLdiscovery. In particular, we hoped to exploit the power of rela-tively stable ortho-thiopenolic esters (22) as surrogates for reac-tive arylthioesters. Unveiling of the latent thioester commenceswith reduction of the disulfide bond of 6 to form aryl thiol 7,which isomerizes to thioester 2 byO → S acyl transfer (Scheme 2).The now thio-activated C terminus participates in an NCL reac-tion with a second peptide (3), bearing an N-terminal thiol func-tionality, to afford peptide 4.

Results and DiscussionBuilding upon the strategy employed in the syntheses of our pre-viously published glycosylated EPO fragments (23–25), EPO pro-tein was envisioned, at the level of retrosynthesis, to arise fromfour fragments of roughly comparable size (Scheme 3, see pep-tides 8–11). We anticipated accomplishing iterative mergers ofthese peptide segments through de facto alanine ligation. Uponformation of peptide 12 through sequential ligation of peptidesegments 8–11, MFD of the nonnative cysteine residues wouldcomplete the synthesis of acetamidomethyl (Acm)-protectednonglycosylated EPO (13). Removal of the Acm protecting

Author contributions: J.A.B. and S.J.D. designed research; J.A.B. performed research; J.A.B.and S.J.D. analyzed data; and J.A.B. and S.J.D. wrote the paper.

The authors declare no conflict of interest.

See Commentary on page 7134.1To whom correspondence should be addressed. E-mail: s-danishefsky@ski.mskcc.org.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1202762109/-/DCSupplemental.

7196–7201 ∣ PNAS ∣ May 8, 2012 ∣ vol. 109 ∣ no. 19 www.pnas.org/cgi/doi/10.1073/pnas.1202762109

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

groups was anticipated by analogy to other ongoing work in ourlaboratory in glycoprotein synthesis.

The C-terminal ortho-arylthioester peptide segments (8–10)were prepared by single amino acid attachments to protectedpeptide fragments (Scheme 4). As a representative example, thesynthesis of Ala1-Cys29 (8) is detailed in Scheme 4A. Thus, resin-bound peptide 14 was prepared by Fmoc-based solid phase pep-tide synthesis using NovaSyn TGT resin and single amino acids(with the exception of the Asp-Ser pseudoproline dipeptide re-sidue shown in Scheme 4B). Following completion of the synth-esis, the crude protected peptide 15 was cleaved from the resinby treatment with acetic acid solution. The final amino acid,Acm-protected cysteine 16, bearing the ortho-thiophenolic ester,

was incorporated by single amino acid attachment under Sakaki-bara conditions (26). The acid-labile protecting groups were re-moved by treatment of the reaction mixture with Cocktail B(88% TFA, 5% H2O, 5% phenol, 2% TIPSH). After purificationby high-performance liquid chromatography (HPLC), peptide 8was isolated in 38% yield from crude peptide 15. PeptidesThz30-Gln78 (9) and Thz79-Arg166 (10) (Scheme 4C) were pre-pared through analogous procedures, involving single amino acidattachment to the protected peptide under epimerization-freeSakakibara conditions (26) followed by removal of the acid-labileprotecting groups with Cocktail B. The fourth fragment, peptide11 (shown in Scheme 5), was prepared by treatment of the resin-bound protected peptide with Cocktail B.

Fig. 1. Structure of EPO. (A) Ribbon diagram with selected oligosaccharides. (B) EPO amino acid sequence. (C) Structure of selected N-linked and O-linkedglycans.

Brailsford and Danishefsky PNAS ∣ May 8, 2012 ∣ vol. 109 ∣ no. 19 ∣ 7197

CHEM

ISTR

YSE

ECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

The four fragments (8–11) were sequentially merged through aseries of iterativeNCL/thiazolidine removal sequences (Scheme 5).Thus, peptide 11, possessing an N-terminal cysteine residue, andpeptide 10, bearing a latent C-terminal thioester, were combined inphosphate buffer solution containing guanidine•HCl, TCEP•HCl,and MPAA at pH 7.2–7.4. Following completion of the ligation,the thiazolidine (27, 28) was unmasked through addition ofMeONH2•HCl (pH 4.0) to reveal an N-terminal cysteine. Thecrude reaction mixture was purified by HPLC to yield Cys79-Arg166 (17) in 34% yield (2 steps). Next, Cys30-Arg166 (18) wasprepared in 42% yield through ligation of peptide 17 with peptide9, followed by thiazolidine removal. To complete the assembly ofthe amino acid sequence, the N-terminal peptide 8 was subjectedto NCL reaction conditions with Cys30-Arg166 (18). The product ofthis reaction, peptide 12, exhibited poor solubility properties inmixtures of acetonitrile and water and thus could not be purifiedby HPLC. However, excess peptide 8, and components of the buf-fer solution, including small organics and salts, could be removedby size exclusion centrifugal filtration. Fortunately, peptide 12could be characterized by liquid chromatography-mass spectro-metry (LC-MS), despite its poor solubility properties. The unpro-tected, nonnative cysteines at positions 30, 79, and 125 of peptide12 were converted to the native alanine residues through MFD(13) with water-soluble radical initiator VA-044 to generate Acm-protected nonglycosylated EPO (13) (67% yield). The productwas purified by centrifugal filtration and characterized by LC-MS.The logic of the synthetic plan inherent in the proposed transfor-mation of 12 → 13 had been vindicated.

Interestingly, Ala1-Arg166 (12) and Cys30-Arg166 (18) exhib-ited notably different stability characteristics. Peptide 18 couldbe purified by reverse-phase HPLC in water/acetonitrile solutionsand seemed to display quite manageable solubility characteristics.However, peptide 12, while clearly present, could not be properlypurified by HPLC, most likely due to problems arising from poorsolubility and aggregation. Thus ligation of the 29-amino acid seg-ment encompassed in peptide 8 with peptide 18 had resulted in aprotein with significantly decreased viability. This finding is animportant benchmark for evaluating the effects of glycosylationon protein viability.

The Acm groups of peptide 18 were removed to yield the fullydeprotected Cys30-Arg166 segment (19). Although this reactionhighlights the ability to remove the Acm groups, synthesis of EPOprotein by alanine ligation of peptide 19 with a fully deprotectedAla1-Cys29 domain is complicated by the presence of both unpro-tected native and nonnative cysteine residues. This scenariowould present the challenging task of performing the selectivedesulfurization of nonnative cysteine residues in the presenceof unprotected native cysteine residues. However, EPO proteincould be assembled from completely unprotected fragments byshifting the final ligation by one amino acid to Gly28 and Cys29.This alternative strategy would not require a subsequent globaldesulfurization since the final step would be a standard NCLreaction.

ConclusionIn summary, we have synthesized nonglycosylated EPO (13),containing four Acm-protected cysteine residues, through theconvergent merger of four peptide fragments. Formal alanineligation via the sequence of NCL—initiated by an o-mercaptoarylester rearrangement (OMER) driven thioester formation—andMFD has been shown to be an effective means for constructingthe full EPO peptide. Remarkably, concurrent threefold MFDhas been demonstrated on a 166-amino acid peptide, and itscompatibility with Acm protecting groups has been established.These findings, rather interesting in their own right, have signifi-cant implications for the chemical synthesis of homogeneous EPOitself. This matter is being pursued using the logic described above,albeit with discrete, relevant glycans in their proper places.

Materials and MethodsCommercial reagents were used without further purification. Automatedpeptide synthesis was performed on an Applied Biosystems Pioneer S3 con-tinuous flow peptide synthesizer. Peptides were synthesized under standardautomated Fmoc protocols. NMR spectra (1H and 13C) were recorded on a

Scheme 1. Noncysteine ligation strategy.

Scheme 2. Proposed generation of latent thioester.

7198 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1202762109 Brailsford and Danishefsky

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

Scheme 3. Strategy for the synthesis of Acm-protected nonglycosylated EPO.

Scheme 4. Synthesis of peptide segments. (A) Synthesis of Ala1-Cys29 (8); amino acid residues bearing acid-labile protecting groups shown in redand Acm-protected Cys residues shown in blue; Cocktail B: 88% TFA, 5% H2O, 5% phenol, 2% TIPSH. (B) Structure of Asp-Ser pseudoproline dipeptideresidue. (C) Schematic representation of Thz30-Gln78 (9) and Thz79-Ala124 (10).

Brailsford and Danishefsky PNAS ∣ May 8, 2012 ∣ vol. 109 ∣ no. 19 ∣ 7199

CHEM

ISTR

YSE

ECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

Bruker Advance DRX-600 MHz spectrometer. Low-resolution mass spectralanalyses were performed with a JOEL JMS-DX-303-HF mass spectrometeror Waters Micromass ZQ mass spectrometer. High-resolution mass spectralanalyses were performed by MSKCC core facility staff.

The deblock solution was a mixture of 96∶2∶2 of DMF/piperidine/DBU.Upon completion of automated synthesis on a 0.05–0.2 mmol scale, thepeptide resin was washed into a peptide synthesis vessel with MeOH. Afterdrying, the resin was subjected to a cleavage cocktail of acetic acid/trifluor-oethanol/methylene chloride (1∶1∶3, 3 × 2–6 mL) for 45 min. The resultingcleavage solutions were combined and concentrated. The oily residue wassuspended in acetonitrile/water (1∶1, 0.05% trifluoroacetic acid) and thenlyophilized.

Analytical LC-MS was performed using a Waters 2695 Separations Moduleand a Waters 2996 Photodiode Array Detector equipped with a VarianMicrosorb C18 column (150 × 2 mm) or a Waters X-Bridge C8 column(150 × 2.1 mm) or a Varian Microsorb C8 column (150 × 2 mm) at a flow rateof 0.2 mL∕min. Preparative HPLC was performed using a Rainin HPXL solventdelivery system equipped with a Rainin UV-1 detector using either a VarianMicrosorb 100-5 C18 column (250 × 21.4 mm) or a Waters X-Bridge C8 col-umn (150 × 19 mm) at a flow rate of 16 mL/min. The mobile phase wascomprised of a linear gradient of 0.05% TFA (v∕v) in water (solvent A) and0.04% TFA (v∕v) in acetonitrile (solvent B).

Centrifugal filtrations were performed using an Eppendorf 5804 R Centri-fuge and Millipore Amicon Ultra-4 Centrifugal Filters (10 kD cut-off). Buffersolutions were diluted with acetonitrile/water (1∶4, 0.05% trifluoroacetic

Scheme 5. Synthesis of nonglycosylated EPO protein (13). Cysteines protected with the Acm group shown in blue. Nonnative cysteines shown in red. Reagentsand conditions: (A) Guanidine•HCl (6 M), Na2HPO4 (0.2 M), TCEP•HCl (0.02 M), MPAA (0.2 M), Bond breaker solution (aqueous 0.5 M TCEP solution at neutralpH), pH 7.2–7.4. (B) MeONH2•HCl (0.3 M), pH 4.0. (C) Guanidine•HCl (6 M), Na2HPO4 (0.2 M), TCEP•HCl (0.08 M) pH 7.0, VA-044, tBuSH, 37 °C.

7200 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1202762109 Brailsford and Danishefsky

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

acid) to 4 mL total volume in a Millipore centrifugal filter tube. The tubewas centrifuged at 4,000 rpm until residual volume was 0.25–0.5 mL. Theresidual volume was diluted to 4 mL, and the process was repeated fourtimes. The residual solution was diluted with acetonitrile/water (1∶1, 0.05%trifluoroacetic acid) and the solution was lyophilized.

Native chemical ligation buffer was prepared by dissolvingguanidine•HCl (1.146 g, 6M) in water. The solution was diluted to 2 mL,then Na2HPO4 (56 mg, 0.2M) was added to the solution, followed byTCEP•HCl (10.8 mg, 0.02M). The pH of the solution was adjusted to 7.2–7.4by the addition of NaOH solution (5M). The solution was sonicated andsparged with Ar for 15 min. 4-Mercaptophenylacetic acid (67 mg, 0.2M)was added, and the pH of the solution was adjusted to 7.2–7.4 by theaddition of NaOH solution (5M). The solution was sonicated and spargedwith Ar for 15 min.

The acid-labile side chain protecting groups were cleaved by treatmentwith Cocktail B consisting of trifluoroacetic acid (88% by volume), water(5% by volume), phenol (5% by weight), and i Pr3 SiH (2% by volume).The solution was triturated with ice-cold diethyl ether (3 × 35 mL) to givea white precipitate. The suspension was centrifuged and the supernant wasdecanted. The precipitate was solubilized in water/acetonitrile (1∶1, 0.05%trifluoroacetic acid), then lyophilized.

A detailed description of materials and methods is given in SI Appendix.

ACKNOWLEDGMENTS. This research was funded by the National Institutes ofHealth Grant CA28824 (S.J.D.). Postdoctoral support is acknowledged byJ.A.B. (T32 CA062948-15). We thank Dr. George Sukenick, Ms. Sylvi Rusli,and Ms. Hui Fang (Sloan-Kettering Institute) for MS and NMR spectroscopicanalysis (Sloan-Kettering Institute Grant CA02848). We thank Ms. RebeccaWilson for assisting in the preparation of this manuscript.

1. Sytkowski AJ (2004) Erythropoietin (Wiley-VCH Verlag GmbH and Co, KGaA,Weinheim).

2. Cazzola M, Mercuriali F, Brugnara C (1997) Use of recombinant human erythropoietinoutside the setting of uremia. Blood 89:4248–4267.

3. Lai PH, Everett R, Wang FF, Arakawa T, Goldwasser E (1986) Structural characterizationof human erythropoietin. J Biol Chem 261:3116–3121.

4. Recny MA, Scoble HA, Kim Y (1987) Structural characterization of natural humanurinary and recombinant DNA-derived erythropoietin. Identification of des-arginine166 erythropoietin. J Biol Chem 262:17156–17163.

5. Helenius A, Aebi M (2001) Intracellular functions of N-linked glycans. Science291:2364–2369.

6. Higuchi M, et al. (1992) Role of sugar chains in the expression of the biological activityof human erythropoietin. J Biol Chem 267:7703–7709.

7. Egrie JC, Browne JK (2001) Development and characterization of novel erythropoiesisstimulating protein (NESP). Br J Cancer 84:3–10.

8. Gamblin DP, Scanlan EM, Davis BG (2009) Glycoprotein synthesis: An update. Chem Rev109:131–163.

9. Kan C, Danishefsky SJ (2009) Recent departures in the synthesis of peptides andglycopeptides. Tetrahedron 65:9047–9065.

10. Dawson PE, Muir TW, Clark-Lewis I, Kent SBH (1994) Synthesis of proteins by nativechemical ligation. Science 266:776–779.

11. Pentelute BL, Kent SBH (2007) Selective desulfurization of cysteine in the presence ofCys(Acm) in polypeptides obtained by native chemical ligation. Org Lett 9:687–690.

12. Yan LZ, Dawson PE (2001) Synthesis of peptides and proteins without cysteineresidues by native chemical ligation combined with desulfurization. J Am ChemSoc 123:526–533.

13. Wan Q, Danishefsky SJ (2007) Free-radical-based, specific desulfurization of cysteine:A powerful advance in the synthesis of polypeptides and glycopolypeptides. AngewChem Int Ed Engl 46:9248–9252.

14. Chen J, Wang P, Zhu J, Wan Q, Danishefsky SJ (2010) A program for ligation at threo-nine sites: Application to the controlled total synthesis of glycopeptides. Tetrahedron66:2277–2283.

15. Tan Z, Shang S, Danishefsky SJ (2010) Insights into the finer issues of native chemicalligation: An approach to cascade ligations. Angew Chem Int Ed Engl 49:9500–9503.

16. Harpaz Z, Siman P, Kumar KSA, Brik A (2010) Protein synthesis assisted by nativechemical ligation at leucine. ChemBioChem 11:1232–1235.

17. Yang R, Pasunooti KK, Li F, Liu X-W, Liu C-F (2009) Dual native chemical ligation atlysine. J Am Chem Soc 131:13592–13593.

18. Crich D, Banerjee A (2007) Native chemical ligation at phenylalanine. J Am Chem Soc129:10064–10065.

19. Narhi LO, et al. (1991) The effect of carbohydrate on the structure and stability oferythropoietin. J Biol Chem 266:23022–23026.

20. Hirano K, Macmillan D, Tezuka K, Tsuji T, Kajihara Y (2009) Design and synthesis of ahomogeneous erythropoietin analogue with two human complex-type sialyloligosac-charides: Combined use of chemical and bacterial protein expression methods.AngewChem Int Ed Engl 48:9378–9378.

21. Kochendoerfer GG, et al. (2003) Design and chemical synthesis of a homogeneouspolymer-modified erythropoiesis protein. Science 299:884–887.

22. Warren JD, Miller JS, Keding SJ, Danishefsky SJ (2004) Toward fully synthetic glyco-proteins by ultimately convergent routes: A solution to a long-standing problem.J Am Chem Soc 126:6576–6578.

23. Tan Z, Shang S, Halkina T, Yuan Y, Danishefsky SJ (2009) Toward homogeneouserythropoietin: Non-NCL-based chemical synthesis of the Gln78-Arg166 glycopeptidedomain. J Am Chem Soc 131:5424–5431.

24. Kan C, et al. (2009) Toward homogeneous erythropoietin: Chemical synthesis of theAla1-Gly28 glycopeptide domain by “alanine” ligation. J Am Chem Soc 131:5438–5443.

25. Yuan Y, et al. (2009) Toward homogeneous erythropoietin: Fine tuning of theC-terminal acyl donor in the chemical synthesis of the Cys29-Gly77 glycopeptidedomain. J Am Chem Soc 131:5432–5437.

26. Kuroda H, Chen Y-N, Kimura T, Sakakibara S (1992) Powerful solvent systems useful forsynthesis of sparingly-soluble peptides in solution. Int J Pept Protein Res 40:294–299.

27. Villain M, Vizzavona J, Rose K (2001) Covalent capture: A new tool for the purificationof synthetic and recombinant polypeptides. Chem Biol 8:673–679.

28. Bang D, Kent SBH (2004) A one-pot total synthesis of crambin. Angew Chem Int EdEngl 43:2534–2538.

Scheme 6. Deprotection of Cys30-Arg166 (18). Native cysteine residues are shown in blue, nonnative cysteine resides are shown in red.

Brailsford and Danishefsky PNAS ∣ May 8, 2012 ∣ vol. 109 ∣ no. 19 ∣ 7201

CHEM

ISTR

YSE

ECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0