CHEMBIOCHEM COMMUNICATIONS DOI: 10.1002/cbic.201402494
Chemical Synthesis of a Synthetic Analogue of the Sialic Acid-Binding Lectin Siglec-7 Masayuki Izumi, Akihisa Otsuki, Mika Nishihara, Ryo Okamoto, and Yasuhiro Kajihara*[a] As a basis for the development of an artificial carbohydratebinding lectin, we chemically synthesized a domain of siglec-7, a well-characterized sialic-acid-binding lectin. The full polypeptide (127 amino acids) was constructed by sequential native chemical ligation (NCL) of five peptide segments. Because of poor cysteine availability for NCL, cysteine residues were introduced at suitable ligation sites; these cysteine residues were alkylated in order to mimic native glutamine or asparagine residues, or converted to an alanine residue by desulfurization after NCL. After folding the full-length polypeptide, the sialicacid-binding activity of the synthetic siglec-7 was clearly demonstrated by STD NMR and ELISA experiments. We succeeded in the synthesis of siglec-7 by installing three extra cysteine residues with side-chain modifications and found that these modifications did not affect the binding activity.
The surfaces of all vertebrate cells are covered with a complex array of sugar chains that are typically in the form of glycoproteins and glycolipids. Sialic acids, which comprise a family of nine-carbon acidic sugars, are often found at the outermost ends of the glycan chains and are known to be involved in various biological processes, including intercellular adhesion, signaling, and microbial attachment.[1] Sialic acids are the targets of a variety of pathogens and toxins, such as human influenza A virus and botulinum toxins produced by Clostridium botulinum. They are also ligands for intrinsic receptors, such as selectins and siglecs (sialic acid-binding immunoglobulin-like lectins), which are involved in rolling adhesion of leukocyte and regulation of immune responses, respectively. Because of the biological significance of sialic acids, molecular probes for the analysis of cell-surface sialic acids would be very useful in the field of glycobiology. Lectins are carbohydrate-binding proteins and have wellcharacterized binding preferences for specific sugars, both monosaccharides and oligosaccharides. Therefore, they are widely used as molecular probes for analyzing cell-surface glycan structures and profiles, typically with fluorescence labeling and in a microarray format.[2] Siglecs are predominantly associated with the immune system and are known to bind to glycoconjugates containing sialic acids. Among them, siglec7[3] has a binding preference for a(2,8)-linked disialic acids and branched a(2,6)-linked sialic acids.[4] The structure of the N-ter[a] Dr. M. Izumi, A. Otsuki, M. Nishihara, Dr. R. Okamoto, Prof. Dr. Y. Kajihara Department of Chemistry, Graduate School of Science, Osaka University 1-1 Machikaneyama, Toyonaka, Osaka 560-0043 (Japan) E-mail: [email protected] Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/cbic.201402494.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
minal V-set Ig-like domain of siglec-7 have been well studied by X-ray crystallography, both as the protein alone[5] and in complex with sialiyloligosaccharides[6] . Based on its characterized binding preference and rich structural information, we selected siglec-7 as a scaffold for the development of a novel molecular probe for the detection of sialic acids on the cell surface. The key amino acid residues involved in the interaction between siglec-7 and GT1b oligosaccharide have been identified (Figure 1 A).[6a] The KD values of siglec-7 toward sialyloligosac-
Figure 1. A) Model of siglec-7 N-terminal V-set Ig-like domain in complex with GT1b oligosaccharide, based on a reported structure (PDB ID: 2HRL). Siglec-7 is in green, GT1b oligosaccharide is in marine blue, and the four cysteine residues installed for NCL (positions 37, 55, 76, and 105) are in purple (black arrows). The position of residue 55 is tentative, because this loop region is not shown in the PDB data. Some residues involved in GT1b binding are shown in red. The model was prepared in Pymol (http://www.pymol.org). B) Further modification of siglec-7 at cysteine side chains. C) Dimeric siglec-7.
charides have been reported to be in several hundred micromolar range and depend on the oligosaccharide.[7] Our aim was to improve the relatively low binding affinity of siglec-7 and to alter its oligosaccharide preference. To achieve these goals, we sought to introduce additional interactions with carbohydrate ligands by modifying the side chains of cysteine residues at specific positions based on the reported structure (Figure 1 B and C). Making a siglec-7 oligomer with a side chain ChemBioChem 2014, 15, 2503 – 2507
2503
CHEMBIOCHEM COMMUNICATIONS
www.chembiochem.org
of cysteine residues is a strategy to improve affinity through multivalent interaction; this is how natural siglecs gain high affinity, that is, by forming a cluster.[8] Furthermore, siglec-7 labeled with a fluorescent molecule will be a useful probe. Lectins are prepared by cell expression systems, and their structures have been solved. Over 1000 lectin domain structures are deposited in the protein data bank (PDB), and approximately 160 lectin carbohydrate-recognition domains (CRDs) with 200 amino acids or less have been reported (Table S1 in the Supporting Information). Proteins of this size are accessible by chemical synthesis by using a combination of solidphase peptide synthesis (SPPS) and a peptide condensation reaction, such as NCL.[9] To the best of our knowledge, no chemical synthesis of a functional Figure 2. A) Amino acid sequence of the N-terminal V-set Ig-like domain of siglec-7 (18–144). Gln37, Gln55, and Asn105 (italicized) were mutated to carbamidomethylcysteine (Cmc). B) Synthesis of siglec-7 (9). Cys41 in 9 carbohydrate-binding lectin has formed a disulfide bond with cysteine in the redox folding buffer. been reported, and we therefore explored the chemical synthesis of a functional lectin. Many lectins have cysteine residues, but these are neither in adequate full-length polypeptide by employing a five-segment coupling number nor are they at suitable positions to perform NCL. To strategy, by installing cysteine residues at appropriate sites for overcome this, we planned to introduce additional cysteine NCL (Figure 2). Four cysteine residues were introduced at ideal residues as junctions for ligation and to functionalize the synNCL positions for coupling the peptide segments:[16] ligation thetic lectins by side-chain modifications as mentioned above. junctions Val36–Gln37, Ser54–Gln55, Lys75–Ala76, and Lys104– Chemical modification of the CRDs of biologically important Asn105. Positions 37, 55, and 105 (NCL sites) are on the side lectins, such as siglecs,[8] galectins,[10] C-type lectin like domains opposite that of the carbohydrate-binding site. This side is (CTLD),[11] selectins,[12] and plant lectins,[13] will provide essential useful for cross-linking or anchoring other molecules, such as information on their carbohydrate-recognition mechanisms. a fluorescent probe or Fc fragment of an antibody (Figure 1 B Lectin modification for potential anti-HIV activity such as with and C). Position 76 is close to the edge of the carbohydratecyanovirin-N homologue, Oscillatoria agardhii agglutinin hobinding site, and thus can be used to introduce functional mologue,[14] or actinohivin,[15] might lead to pharmaceutical apgroups for extra interactions with ligands (Figure 1 B). Cys76 plications. was converted to a native Ala residue by a desulfurization reTo demonstrate the synthesis of a lectin with a distinct caraction,[17] and Cys37, Cys55, and Cys105 were alkylated with bohydrate binding activity pattern, we report the first chemical iodoacetamide[18] to mimic the Gln and Asn residues of the synthesis of lectin siglec-7 with four additional cysteine resinative sequence with carbamidomethylcysteine (Cmc).[19] To dues that can be modified to enhance the functionality of the perform the sequential ligation from the C terminus to the lectin as a molecular probe. N terminus, the Cys37, Cys55, and Cys76 residues were introThe full-length extracellular domain of human siglec-7 (Uniduced as thiazolidine-4-carbonyl (Thz) groups. The three cysProt Q9Y286) consists of 335 amino acid residues, including teine residues in the original peptide sequence (positions 41, three Ig-like domains. The N-terminal V-set Ig-like domain (127 46, and 106) were protected with acetamidomethyl (Acm) amino acids (18–144) with one disulfide bond)[5a] was shown to groups, which can be deprotected after assembly of the fulllength polypeptide. form an active carbohydrate binding domain by X-ray analysis Peptide-a-thioester segments corresponding to Gly18–Val36 (PDB ID: 2HRL).[6a] Therefore we decided to chemically synthe(1) and Thz37–Ser54 (2) were prepared by Fmoc-SPPS.[20] Pepsize this sequence. However, this sequence does not have a cysteine residue at a suitable NCL site, so we synthesized the tide-a-thioester segments corresponding to Thz55–Lys75 (3) 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 2503 – 2507
2504
CHEMBIOCHEM COMMUNICATIONS
www.chembiochem.org
and Thz76–Lys104 (4) were prepared by Boc-SPPS.[21] The C-terminal peptide segment (Cys105–Thr144, 5) was prepared by conventional Fmoc-SPPS (SPPS and segment identification in the Supporting Information). We next performed sequential NCL from the C to the N terminus (Figure 2 B). The first NCL (4 and 5) was performed under conventional NCL conditions[22] with 40 mm 4-mercaptophenylacetic acid as the thiol additive and 20 mm tris(2-carboxyethyl)phosphine (TCEP) as the reducing agent in 6 m Gn·HCl and 0.2 m sodium phosphate (pH 7.2); the desired peptide, Thz76–Thr144, was obtained in 54 % yield. One-pot alkylation of Cys105 with iodoacetamide in 6 m Gn·HCl and 0.2 m sodium phosphate (pH 7.2) and subsequent conversion of N-terminal Thz to Cys by adjusting to pH 4.0 with methoxyamine hydrochloride yielded 6 in 92 % yield.[23] The second NCL, between 3 and 6, was carried out under the same conditions, and the Thz55–Thr144 peptide was obtained in 72 % yield. Then Cys76 was converted to Ala by desulfurization with VA-044, TCEP, and sodium 2-mercaptoethanesulfonate (64 % yield). After conversion of the N-terminal Thz to Cys to yield peptide 7, the Figure 3. HPLC profiles from folding experiment for 8: A) < 5 min, B) after stepwise diathird NCL was carried out (2 and 7) under the same lysis, and C) purified 9. D) ESI-MS spectrum of 8; calcd: 14 747.4, found: 14 749.0 0.6. conditions. The subsequent one-pot conversion of E) ESI-MS spectrum of 9; calcd: 14 864.5, found:14 865.5 0.5. F) CD spectrum of 9. Thz to Cys yielded a peptide corresponding to Cys37–Thr144 (67 % yield), and the final NCL under the same conditions between this and 1 gave the full-length dimensional protein structure 9, so we decided to analyze its polypeptide in moderate yield. Then, Cys37 and Cys55 were carbohydrate-binding activity. converted to Cmc residues by reacting with iodoacetamide to We performed NMR and ELISA assays to analyze the carbomimic the Gln residues in the native sequence (63 % yield). Fihydrate-binding activity. Saturation transfer difference (STD) nally, the Acm groups on the side chains of the three native NMR is widely applied for analysis of the binding of carbohyCys residues were removed by treatment with silver acetate in drates to proteins.[26] STD NMR was performed with 1.53 mm 90 % aqueous acetic acid to yield the desired full-length polya(2!8)-linked trisialic acid 10 (Figure 4 E, ligand) and 7.3 mm 9 peptide 8 in 80 % yield. in D2O. The concentration of 10 was chosen based on reported We next examined the folding of full-length polypeptide 8 KD values.[7] The saturation was examined at 1, 2, and 3 s; by employing a stepwise dialysis method under cysteine–cyspeaks corresponding to H3eq (2.7 ppm) and H3ax (1.65 ppm) atine redox conditions.[24] The full-length polypeptide was easily sialic acid residues were observed in the STD spectra (Figure 4). H3 of b-sialic acid was also observed at 2.35 ppm, converted, as it showed a single HPLC peak (Figure 3 A–C). ESIand this might have resulted from interaction of the reducing MS analysis of the isolated peak suggested the formation of end of b-sialic acid with 9. We also observed a significant one intramolecular disulfide bond along with one cysteine atchemical shift perturbation of one of the three acetyl groups tached to the cysteine residue that was not involved in the inin the 1D 1H NMR spectrum when the 10/9 ratio was varied tramolecular disulfide bond (Figure 3 E). ESI-MS spectra of 8 from 10:1 to 80:1 (Figure S16). The concentration of 10 ranged and 9 showed a remarkable difference in the formation of mulfrom 160 to 750 mm; the observed chemical shift perturbation tiply charged peaks (Figure 3 D and E). A charge state distribusuggests that synthetic siglec-7 binds to trisialic acid in a suitation from 9 + to 20 + with maxima at 10 + and 15 + was obble hundred micromolar range. The results suggest direct bindserved for 8, whereas only one (9 + to 15 + , maximum at 10 + ) ing of 9 to 10, which is known to bind to native siglec-7. In was observed for 9. A changes to a smaller charge state distriorder to examine the carbohydrate-binding activity by ELISA, bution in ESI-MS is a well-known characteristic of a folded progangliosides GD3 and GM3 were immobilized on microtiter tein.[25] The CD spectrum (Figure 3 F) suggested the existence plates, and the amount of bound 9 on immobilized ganglioof the a-helix and b-sheet reported in the X-ray analysis (Figside was determined by using a commercial polyclonal antiure 1 A).[5a] We attempted disulfide bond mapping by employsiglec-7 antibody (see the Supporting Information). Synthetic 9 ing trypsin or chymotrypsin digestion, but the desired peptide bound preferentially to GD3 over GM3 (Figure 5), similarly to fragment connected through a disulfide bond was not obthe binding preference of native siglec-7. These results clearly served upon LC-MS analysis, probably due to the highly hydroshow that 9 attained the native three-dimensional structure rephobic nature of the peptide fragment. From these analyses, quired to bind to oligosaccharides containing sialic acid. we assumed that the polypeptide formed the required three 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 2503 – 2507
2505
CHEMBIOCHEM COMMUNICATIONS
www.chembiochem.org of the folded protein. Our next goals are a fluorescently labeled lectin, a lectin oligomer, and an artificial lectin with enhanced affinity and specificity by designing a specific interaction with the target carbohydrate. There are many interesting lectins of a size accessible by chemical synthesis (Table S1). Synthetic lectins will find utility in a wide variety of applications, such as chemical probes for glycan profiling of the cell surface and artificial antibodies with carbohydrate-binding activity. For larger lectins, semisynthesis combined with expression protein ligation will overcome the synthetic protein size limit. We hope the demonstration here will open an avenue to the widespread synthesis of such functional lectins.
Experimental Section Experimental details, compound characterization, and NMR titration experiments are reported in the Supporting Information.
Acknowledgements This work was supported in part by Grant-in-Aid for Scientific Research from the Ministry of Education, Sports, Science and Technology of Japan (No.23245037). Figure 4. STD NMR experiments with 9 and NeuAc-a-(2!8)-NeuAc-a-(2! 8)-NeuAc (10). A) 1H NMR spectrum. STD NMR spectra with saturation at B) 1 s, C) 2 s, and D) 3 s. E) Trisialic acid (10).
Figure 5. ELISA binding assay of synthetic siglec-7 (9) to GD3 (c) or GM3 (····). GD3 or GM3 was immobilized on a microtiter plate, and 9 was added at various concentration. Bound 9 was detected by anti-siglec-7 antibody. *: Glc, *: Gal, ^: NeuAc.
In conclusion, we succeeded in the chemical synthesis of the 127-residue N-terminal V-set Ig-like domain of siglec-7 by sequential NCL and alkylation and desulfurization of the cysteine residues. The synthetic protein clearly exhibited binding activity toward a(2!8)-linked trisialic acid and the GD3 ganglioside, as shown by the STD NMR and ELISA data, respectively. To the best of our knowledge, this is the first chemical synthesis of a protein with lectin activity, although this synthetic siglec-7 has several non-native modifications in the primary structure 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Keywords: carbohydrates · cysteine · native chemical ligation · NMR spectroscopy · synthetic lectin [1] A. Varki, Nature 2007, 446, 1023 – 1029. [2] J. P. Ribeiro, L. K. Mahal, Curr. Opin. Chem. Biol. 2013, 17, 827 – 831. [3] a) M. Falco, R. Biassoni, C. Bottino, M. Vitale, S. Sivori, R. Augugliaro, L. Moretta, A. Moretta, J. Exp. Med. 1999, 190, 793 – 801; b) G. Nicoll, J. Ni, D. Liu, P. Klenerman, J. Munday, S. Dubock, M.-G. Mattei, P. R. Crocker, J. Biol. Chem. 1999, 274, 34089 – 34095; c) T. Angata, A. Varki, Glycobiology 2000, 10, 431 – 438. [4] T. Yamaji, T. Teranishi, M. S. Alphey, P. R. Crocker, Y. Hashimoto, J. Biol. Chem. 2002, 277, 6324 – 6332. [5] a) M. S. Alphey, H. Attrill, P. R. Crocker, D. M. F. van Aalten, J. Biol. Chem. 2003, 278, 3372 – 3377; b) N. Dimasi, A. Moretta, L. Moretta, R. Biassoni, R. A. Mariuzza, Acta Crystallogr. Sect. D Biol. Crystallogr. 2004, 60, 401 – 403. [6] a) H. Attrill, A. Imamura, R. S. Sharma, M. Kiso, P. R. Crocker, D. M. F. van Aalten, J. Biol. Chem. 2006, 281, 32774 – 32783; b) H. Attrill, H. Takazawa, S. Witt, S. Kelm, R. Isecke, R. Brossmer, T. Ando, H. Ishida, M. Kiso, P. R. Crocker, D. M. F. van Aalten, Biochem. J. 2006, 397, 271 – 278. [7] C. P. Swaminathan, N. Wais, V. V. Vyas, C. A. Velikovsky, A. Moretta, L. Moretta, R. Biassoni, R. A. Mariuzza, N. Dimasi, ChemBioChem 2004, 5, 1571 – 1575. [8] P. R. Crocker, J. C. Paulson, A. Varki, Nat. Rev. Immunol. 2007, 7, 255 – 266. [9] S. B. H. Kent, Chem. Soc. Rev. 2009, 38, 338 – 351. [10] G. A. Rabinovich, Cell Death Differ. 1999, 6, 711 – 721. [11] a) A. N. Zelensky, J. E. Gready, FEBS J. 2005, 272, 6179 – 6217; b) T. B. H. Geijtenbeek, S. I. Gringhuis, Nat. Rev. Immunol. 2009, 9, 465 – 479. [12] W. S. Somers, J. Tang, G. D. Shaw, R. T. Camphausen, Cell 2000, 103, 467 – 479. [13] H. Rdiger, H.-J. Gabius, Glycoconjugate J. 2001, 18, 589 – 613. [14] L. M. I. Koharudin, A. M. Gronenborn, Biopolymers 2013, 99, 196 – 202. [15] H. Tanaka, H. Chiba, J. Inokoshi, A. Kuno, T. Sugai, A. Takahashi, Y. Ito, M. Tsunoda, K. Suzuki, A. Taknaka, T. Sekiguchi, H. Umeyama, J. Hirabayashi, S. O¯mura, Proc. Natl. Acad. Sci. USA 2009, 106, 15633 – 15638. [16] P. E. Dawson, T. W. Muir, I. Clark-Lewis, S. B. Kent, Science 1994, 266, 776 – 779.
ChemBioChem 2014, 15, 2503 – 2507
2506
CHEMBIOCHEM COMMUNICATIONS [17] a) L. Z. Yan, P. E. Dawson, J. Am. Chem. Soc. 2001, 123, 526 – 533; b) Q. Wan, S. J. Danishefsky, Angew. Chem. Int. Ed. 2007, 46, 9248 – 9252; Angew. Chem. 2007, 119, 9408 – 9412. [18] D. A. Bochar, L. Tabernero, C. V. Stauffacher, V. W. Rodwell, Biochemistry 1999, 38, 8879 – 8883. [19] a) N. J. Davis, S. L. Flitsch, Tetrahedron Lett. 1991, 32, 6793 – 6796; b) V. Y. Torbeev, S. B. H. Kent, Angew. Chem. Int. Ed. 2007, 46, 1667 – 1670; Angew. Chem. 2007, 119, 1697 – 1700; c) C. P. R. Hackenberger, D. Schwarzer, Angew. Chem. Int. Ed. 2008, 47, 10030 – 10074; Angew. Chem. 2008, 120, 10182 – 10228. [20] Y. Kajihara, A. Yoshihara, K. Hirano, N. Yamamoto, Carbohydr. Res. 2006, 341, 1333 – 1340. [21] a) M. Schnçlzer, P. Alewood, A. Jones, D. Alewood, S. B. H. Kent, Int. J. Pept. Res. Ther. 2007, 13, 31 – 44; b) M. Murakami, R. Okamoto, M. Izumi, Y. Kajihara, Angew. Chem. Int. Ed. 2012, 51, 3567 – 3572; Angew. Chem.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
[22] [23] [24] [25] [26]
2012, 124, 3627 – 3632; c) M. Izumi, M. Murakami, R. Okamoto, Y. Kajihara, J. Pept. Sci. 2014, 20, 98 – 101. E. C. B. Johnson, S. B. H. Kent, J. Am. Chem. Soc. 2006, 128, 6640 – 6646. a) M. Villain, J. Vizzavona, K. Rose, Chem. Biol. 2001, 8, 673 – 679; b) D. Bang, N. Chopra, S. B. H. Kent, J. Am. Chem. Soc. 2004, 126, 1377 – 1383. V. P. Saxena, D. B. Wetlaufer, Biochemistry 1970, 9, 5015 – 5023. S. K. Chowdhury, V. Katta, B. T. Chait, J. Am. Chem. Soc. 1990, 112, 9012 – 9013. a) M. Mayer, B. Meyer, Angew. Chem. Int. Ed. 1999, 38, 1784 – 1788; Angew. Chem. 1999, 111, 1902 – 1906; b) M. Mayer, B. Meyer, J. Am. Chem. Soc. 2001, 123, 6108 – 6117.
Received: September 1, 2014 Published online on October 2, 2014
ChemBioChem 2014, 15, 2503 – 2507
2507
Chemical Synthesis of a Synthetic Analogue of the Sialic Acid-Binding Lectin Siglec-7 Masayuki Izumi, Akihisa Otsuki, Mika Nishihara, Ryo Okamoto, and Yasuhiro Kajihara*[a] As a basis for the development of an artificial carbohydratebinding lectin, we chemically synthesized a domain of siglec-7, a well-characterized sialic-acid-binding lectin. The full polypeptide (127 amino acids) was constructed by sequential native chemical ligation (NCL) of five peptide segments. Because of poor cysteine availability for NCL, cysteine residues were introduced at suitable ligation sites; these cysteine residues were alkylated in order to mimic native glutamine or asparagine residues, or converted to an alanine residue by desulfurization after NCL. After folding the full-length polypeptide, the sialicacid-binding activity of the synthetic siglec-7 was clearly demonstrated by STD NMR and ELISA experiments. We succeeded in the synthesis of siglec-7 by installing three extra cysteine residues with side-chain modifications and found that these modifications did not affect the binding activity.
The surfaces of all vertebrate cells are covered with a complex array of sugar chains that are typically in the form of glycoproteins and glycolipids. Sialic acids, which comprise a family of nine-carbon acidic sugars, are often found at the outermost ends of the glycan chains and are known to be involved in various biological processes, including intercellular adhesion, signaling, and microbial attachment.[1] Sialic acids are the targets of a variety of pathogens and toxins, such as human influenza A virus and botulinum toxins produced by Clostridium botulinum. They are also ligands for intrinsic receptors, such as selectins and siglecs (sialic acid-binding immunoglobulin-like lectins), which are involved in rolling adhesion of leukocyte and regulation of immune responses, respectively. Because of the biological significance of sialic acids, molecular probes for the analysis of cell-surface sialic acids would be very useful in the field of glycobiology. Lectins are carbohydrate-binding proteins and have wellcharacterized binding preferences for specific sugars, both monosaccharides and oligosaccharides. Therefore, they are widely used as molecular probes for analyzing cell-surface glycan structures and profiles, typically with fluorescence labeling and in a microarray format.[2] Siglecs are predominantly associated with the immune system and are known to bind to glycoconjugates containing sialic acids. Among them, siglec7[3] has a binding preference for a(2,8)-linked disialic acids and branched a(2,6)-linked sialic acids.[4] The structure of the N-ter[a] Dr. M. Izumi, A. Otsuki, M. Nishihara, Dr. R. Okamoto, Prof. Dr. Y. Kajihara Department of Chemistry, Graduate School of Science, Osaka University 1-1 Machikaneyama, Toyonaka, Osaka 560-0043 (Japan) E-mail: [email protected] Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/cbic.201402494.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
minal V-set Ig-like domain of siglec-7 have been well studied by X-ray crystallography, both as the protein alone[5] and in complex with sialiyloligosaccharides[6] . Based on its characterized binding preference and rich structural information, we selected siglec-7 as a scaffold for the development of a novel molecular probe for the detection of sialic acids on the cell surface. The key amino acid residues involved in the interaction between siglec-7 and GT1b oligosaccharide have been identified (Figure 1 A).[6a] The KD values of siglec-7 toward sialyloligosac-
Figure 1. A) Model of siglec-7 N-terminal V-set Ig-like domain in complex with GT1b oligosaccharide, based on a reported structure (PDB ID: 2HRL). Siglec-7 is in green, GT1b oligosaccharide is in marine blue, and the four cysteine residues installed for NCL (positions 37, 55, 76, and 105) are in purple (black arrows). The position of residue 55 is tentative, because this loop region is not shown in the PDB data. Some residues involved in GT1b binding are shown in red. The model was prepared in Pymol (http://www.pymol.org). B) Further modification of siglec-7 at cysteine side chains. C) Dimeric siglec-7.
charides have been reported to be in several hundred micromolar range and depend on the oligosaccharide.[7] Our aim was to improve the relatively low binding affinity of siglec-7 and to alter its oligosaccharide preference. To achieve these goals, we sought to introduce additional interactions with carbohydrate ligands by modifying the side chains of cysteine residues at specific positions based on the reported structure (Figure 1 B and C). Making a siglec-7 oligomer with a side chain ChemBioChem 2014, 15, 2503 – 2507
2503
CHEMBIOCHEM COMMUNICATIONS
www.chembiochem.org
of cysteine residues is a strategy to improve affinity through multivalent interaction; this is how natural siglecs gain high affinity, that is, by forming a cluster.[8] Furthermore, siglec-7 labeled with a fluorescent molecule will be a useful probe. Lectins are prepared by cell expression systems, and their structures have been solved. Over 1000 lectin domain structures are deposited in the protein data bank (PDB), and approximately 160 lectin carbohydrate-recognition domains (CRDs) with 200 amino acids or less have been reported (Table S1 in the Supporting Information). Proteins of this size are accessible by chemical synthesis by using a combination of solidphase peptide synthesis (SPPS) and a peptide condensation reaction, such as NCL.[9] To the best of our knowledge, no chemical synthesis of a functional Figure 2. A) Amino acid sequence of the N-terminal V-set Ig-like domain of siglec-7 (18–144). Gln37, Gln55, and Asn105 (italicized) were mutated to carbamidomethylcysteine (Cmc). B) Synthesis of siglec-7 (9). Cys41 in 9 carbohydrate-binding lectin has formed a disulfide bond with cysteine in the redox folding buffer. been reported, and we therefore explored the chemical synthesis of a functional lectin. Many lectins have cysteine residues, but these are neither in adequate full-length polypeptide by employing a five-segment coupling number nor are they at suitable positions to perform NCL. To strategy, by installing cysteine residues at appropriate sites for overcome this, we planned to introduce additional cysteine NCL (Figure 2). Four cysteine residues were introduced at ideal residues as junctions for ligation and to functionalize the synNCL positions for coupling the peptide segments:[16] ligation thetic lectins by side-chain modifications as mentioned above. junctions Val36–Gln37, Ser54–Gln55, Lys75–Ala76, and Lys104– Chemical modification of the CRDs of biologically important Asn105. Positions 37, 55, and 105 (NCL sites) are on the side lectins, such as siglecs,[8] galectins,[10] C-type lectin like domains opposite that of the carbohydrate-binding site. This side is (CTLD),[11] selectins,[12] and plant lectins,[13] will provide essential useful for cross-linking or anchoring other molecules, such as information on their carbohydrate-recognition mechanisms. a fluorescent probe or Fc fragment of an antibody (Figure 1 B Lectin modification for potential anti-HIV activity such as with and C). Position 76 is close to the edge of the carbohydratecyanovirin-N homologue, Oscillatoria agardhii agglutinin hobinding site, and thus can be used to introduce functional mologue,[14] or actinohivin,[15] might lead to pharmaceutical apgroups for extra interactions with ligands (Figure 1 B). Cys76 plications. was converted to a native Ala residue by a desulfurization reTo demonstrate the synthesis of a lectin with a distinct caraction,[17] and Cys37, Cys55, and Cys105 were alkylated with bohydrate binding activity pattern, we report the first chemical iodoacetamide[18] to mimic the Gln and Asn residues of the synthesis of lectin siglec-7 with four additional cysteine resinative sequence with carbamidomethylcysteine (Cmc).[19] To dues that can be modified to enhance the functionality of the perform the sequential ligation from the C terminus to the lectin as a molecular probe. N terminus, the Cys37, Cys55, and Cys76 residues were introThe full-length extracellular domain of human siglec-7 (Uniduced as thiazolidine-4-carbonyl (Thz) groups. The three cysProt Q9Y286) consists of 335 amino acid residues, including teine residues in the original peptide sequence (positions 41, three Ig-like domains. The N-terminal V-set Ig-like domain (127 46, and 106) were protected with acetamidomethyl (Acm) amino acids (18–144) with one disulfide bond)[5a] was shown to groups, which can be deprotected after assembly of the fulllength polypeptide. form an active carbohydrate binding domain by X-ray analysis Peptide-a-thioester segments corresponding to Gly18–Val36 (PDB ID: 2HRL).[6a] Therefore we decided to chemically synthe(1) and Thz37–Ser54 (2) were prepared by Fmoc-SPPS.[20] Pepsize this sequence. However, this sequence does not have a cysteine residue at a suitable NCL site, so we synthesized the tide-a-thioester segments corresponding to Thz55–Lys75 (3) 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 2503 – 2507
2504
CHEMBIOCHEM COMMUNICATIONS
www.chembiochem.org
and Thz76–Lys104 (4) were prepared by Boc-SPPS.[21] The C-terminal peptide segment (Cys105–Thr144, 5) was prepared by conventional Fmoc-SPPS (SPPS and segment identification in the Supporting Information). We next performed sequential NCL from the C to the N terminus (Figure 2 B). The first NCL (4 and 5) was performed under conventional NCL conditions[22] with 40 mm 4-mercaptophenylacetic acid as the thiol additive and 20 mm tris(2-carboxyethyl)phosphine (TCEP) as the reducing agent in 6 m Gn·HCl and 0.2 m sodium phosphate (pH 7.2); the desired peptide, Thz76–Thr144, was obtained in 54 % yield. One-pot alkylation of Cys105 with iodoacetamide in 6 m Gn·HCl and 0.2 m sodium phosphate (pH 7.2) and subsequent conversion of N-terminal Thz to Cys by adjusting to pH 4.0 with methoxyamine hydrochloride yielded 6 in 92 % yield.[23] The second NCL, between 3 and 6, was carried out under the same conditions, and the Thz55–Thr144 peptide was obtained in 72 % yield. Then Cys76 was converted to Ala by desulfurization with VA-044, TCEP, and sodium 2-mercaptoethanesulfonate (64 % yield). After conversion of the N-terminal Thz to Cys to yield peptide 7, the Figure 3. HPLC profiles from folding experiment for 8: A) < 5 min, B) after stepwise diathird NCL was carried out (2 and 7) under the same lysis, and C) purified 9. D) ESI-MS spectrum of 8; calcd: 14 747.4, found: 14 749.0 0.6. conditions. The subsequent one-pot conversion of E) ESI-MS spectrum of 9; calcd: 14 864.5, found:14 865.5 0.5. F) CD spectrum of 9. Thz to Cys yielded a peptide corresponding to Cys37–Thr144 (67 % yield), and the final NCL under the same conditions between this and 1 gave the full-length dimensional protein structure 9, so we decided to analyze its polypeptide in moderate yield. Then, Cys37 and Cys55 were carbohydrate-binding activity. converted to Cmc residues by reacting with iodoacetamide to We performed NMR and ELISA assays to analyze the carbomimic the Gln residues in the native sequence (63 % yield). Fihydrate-binding activity. Saturation transfer difference (STD) nally, the Acm groups on the side chains of the three native NMR is widely applied for analysis of the binding of carbohyCys residues were removed by treatment with silver acetate in drates to proteins.[26] STD NMR was performed with 1.53 mm 90 % aqueous acetic acid to yield the desired full-length polya(2!8)-linked trisialic acid 10 (Figure 4 E, ligand) and 7.3 mm 9 peptide 8 in 80 % yield. in D2O. The concentration of 10 was chosen based on reported We next examined the folding of full-length polypeptide 8 KD values.[7] The saturation was examined at 1, 2, and 3 s; by employing a stepwise dialysis method under cysteine–cyspeaks corresponding to H3eq (2.7 ppm) and H3ax (1.65 ppm) atine redox conditions.[24] The full-length polypeptide was easily sialic acid residues were observed in the STD spectra (Figure 4). H3 of b-sialic acid was also observed at 2.35 ppm, converted, as it showed a single HPLC peak (Figure 3 A–C). ESIand this might have resulted from interaction of the reducing MS analysis of the isolated peak suggested the formation of end of b-sialic acid with 9. We also observed a significant one intramolecular disulfide bond along with one cysteine atchemical shift perturbation of one of the three acetyl groups tached to the cysteine residue that was not involved in the inin the 1D 1H NMR spectrum when the 10/9 ratio was varied tramolecular disulfide bond (Figure 3 E). ESI-MS spectra of 8 from 10:1 to 80:1 (Figure S16). The concentration of 10 ranged and 9 showed a remarkable difference in the formation of mulfrom 160 to 750 mm; the observed chemical shift perturbation tiply charged peaks (Figure 3 D and E). A charge state distribusuggests that synthetic siglec-7 binds to trisialic acid in a suitation from 9 + to 20 + with maxima at 10 + and 15 + was obble hundred micromolar range. The results suggest direct bindserved for 8, whereas only one (9 + to 15 + , maximum at 10 + ) ing of 9 to 10, which is known to bind to native siglec-7. In was observed for 9. A changes to a smaller charge state distriorder to examine the carbohydrate-binding activity by ELISA, bution in ESI-MS is a well-known characteristic of a folded progangliosides GD3 and GM3 were immobilized on microtiter tein.[25] The CD spectrum (Figure 3 F) suggested the existence plates, and the amount of bound 9 on immobilized ganglioof the a-helix and b-sheet reported in the X-ray analysis (Figside was determined by using a commercial polyclonal antiure 1 A).[5a] We attempted disulfide bond mapping by employsiglec-7 antibody (see the Supporting Information). Synthetic 9 ing trypsin or chymotrypsin digestion, but the desired peptide bound preferentially to GD3 over GM3 (Figure 5), similarly to fragment connected through a disulfide bond was not obthe binding preference of native siglec-7. These results clearly served upon LC-MS analysis, probably due to the highly hydroshow that 9 attained the native three-dimensional structure rephobic nature of the peptide fragment. From these analyses, quired to bind to oligosaccharides containing sialic acid. we assumed that the polypeptide formed the required three 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 2503 – 2507
2505
CHEMBIOCHEM COMMUNICATIONS
www.chembiochem.org of the folded protein. Our next goals are a fluorescently labeled lectin, a lectin oligomer, and an artificial lectin with enhanced affinity and specificity by designing a specific interaction with the target carbohydrate. There are many interesting lectins of a size accessible by chemical synthesis (Table S1). Synthetic lectins will find utility in a wide variety of applications, such as chemical probes for glycan profiling of the cell surface and artificial antibodies with carbohydrate-binding activity. For larger lectins, semisynthesis combined with expression protein ligation will overcome the synthetic protein size limit. We hope the demonstration here will open an avenue to the widespread synthesis of such functional lectins.
Experimental Section Experimental details, compound characterization, and NMR titration experiments are reported in the Supporting Information.
Acknowledgements This work was supported in part by Grant-in-Aid for Scientific Research from the Ministry of Education, Sports, Science and Technology of Japan (No.23245037). Figure 4. STD NMR experiments with 9 and NeuAc-a-(2!8)-NeuAc-a-(2! 8)-NeuAc (10). A) 1H NMR spectrum. STD NMR spectra with saturation at B) 1 s, C) 2 s, and D) 3 s. E) Trisialic acid (10).
Figure 5. ELISA binding assay of synthetic siglec-7 (9) to GD3 (c) or GM3 (····). GD3 or GM3 was immobilized on a microtiter plate, and 9 was added at various concentration. Bound 9 was detected by anti-siglec-7 antibody. *: Glc, *: Gal, ^: NeuAc.
In conclusion, we succeeded in the chemical synthesis of the 127-residue N-terminal V-set Ig-like domain of siglec-7 by sequential NCL and alkylation and desulfurization of the cysteine residues. The synthetic protein clearly exhibited binding activity toward a(2!8)-linked trisialic acid and the GD3 ganglioside, as shown by the STD NMR and ELISA data, respectively. To the best of our knowledge, this is the first chemical synthesis of a protein with lectin activity, although this synthetic siglec-7 has several non-native modifications in the primary structure 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Keywords: carbohydrates · cysteine · native chemical ligation · NMR spectroscopy · synthetic lectin [1] A. Varki, Nature 2007, 446, 1023 – 1029. [2] J. P. Ribeiro, L. K. Mahal, Curr. Opin. Chem. Biol. 2013, 17, 827 – 831. [3] a) M. Falco, R. Biassoni, C. Bottino, M. Vitale, S. Sivori, R. Augugliaro, L. Moretta, A. Moretta, J. Exp. Med. 1999, 190, 793 – 801; b) G. Nicoll, J. Ni, D. Liu, P. Klenerman, J. Munday, S. Dubock, M.-G. Mattei, P. R. Crocker, J. Biol. Chem. 1999, 274, 34089 – 34095; c) T. Angata, A. Varki, Glycobiology 2000, 10, 431 – 438. [4] T. Yamaji, T. Teranishi, M. S. Alphey, P. R. Crocker, Y. Hashimoto, J. Biol. Chem. 2002, 277, 6324 – 6332. [5] a) M. S. Alphey, H. Attrill, P. R. Crocker, D. M. F. van Aalten, J. Biol. Chem. 2003, 278, 3372 – 3377; b) N. Dimasi, A. Moretta, L. Moretta, R. Biassoni, R. A. Mariuzza, Acta Crystallogr. Sect. D Biol. Crystallogr. 2004, 60, 401 – 403. [6] a) H. Attrill, A. Imamura, R. S. Sharma, M. Kiso, P. R. Crocker, D. M. F. van Aalten, J. Biol. Chem. 2006, 281, 32774 – 32783; b) H. Attrill, H. Takazawa, S. Witt, S. Kelm, R. Isecke, R. Brossmer, T. Ando, H. Ishida, M. Kiso, P. R. Crocker, D. M. F. van Aalten, Biochem. J. 2006, 397, 271 – 278. [7] C. P. Swaminathan, N. Wais, V. V. Vyas, C. A. Velikovsky, A. Moretta, L. Moretta, R. Biassoni, R. A. Mariuzza, N. Dimasi, ChemBioChem 2004, 5, 1571 – 1575. [8] P. R. Crocker, J. C. Paulson, A. Varki, Nat. Rev. Immunol. 2007, 7, 255 – 266. [9] S. B. H. Kent, Chem. Soc. Rev. 2009, 38, 338 – 351. [10] G. A. Rabinovich, Cell Death Differ. 1999, 6, 711 – 721. [11] a) A. N. Zelensky, J. E. Gready, FEBS J. 2005, 272, 6179 – 6217; b) T. B. H. Geijtenbeek, S. I. Gringhuis, Nat. Rev. Immunol. 2009, 9, 465 – 479. [12] W. S. Somers, J. Tang, G. D. Shaw, R. T. Camphausen, Cell 2000, 103, 467 – 479. [13] H. Rdiger, H.-J. Gabius, Glycoconjugate J. 2001, 18, 589 – 613. [14] L. M. I. Koharudin, A. M. Gronenborn, Biopolymers 2013, 99, 196 – 202. [15] H. Tanaka, H. Chiba, J. Inokoshi, A. Kuno, T. Sugai, A. Takahashi, Y. Ito, M. Tsunoda, K. Suzuki, A. Taknaka, T. Sekiguchi, H. Umeyama, J. Hirabayashi, S. O¯mura, Proc. Natl. Acad. Sci. USA 2009, 106, 15633 – 15638. [16] P. E. Dawson, T. W. Muir, I. Clark-Lewis, S. B. Kent, Science 1994, 266, 776 – 779.
ChemBioChem 2014, 15, 2503 – 2507
2506
CHEMBIOCHEM COMMUNICATIONS [17] a) L. Z. Yan, P. E. Dawson, J. Am. Chem. Soc. 2001, 123, 526 – 533; b) Q. Wan, S. J. Danishefsky, Angew. Chem. Int. Ed. 2007, 46, 9248 – 9252; Angew. Chem. 2007, 119, 9408 – 9412. [18] D. A. Bochar, L. Tabernero, C. V. Stauffacher, V. W. Rodwell, Biochemistry 1999, 38, 8879 – 8883. [19] a) N. J. Davis, S. L. Flitsch, Tetrahedron Lett. 1991, 32, 6793 – 6796; b) V. Y. Torbeev, S. B. H. Kent, Angew. Chem. Int. Ed. 2007, 46, 1667 – 1670; Angew. Chem. 2007, 119, 1697 – 1700; c) C. P. R. Hackenberger, D. Schwarzer, Angew. Chem. Int. Ed. 2008, 47, 10030 – 10074; Angew. Chem. 2008, 120, 10182 – 10228. [20] Y. Kajihara, A. Yoshihara, K. Hirano, N. Yamamoto, Carbohydr. Res. 2006, 341, 1333 – 1340. [21] a) M. Schnçlzer, P. Alewood, A. Jones, D. Alewood, S. B. H. Kent, Int. J. Pept. Res. Ther. 2007, 13, 31 – 44; b) M. Murakami, R. Okamoto, M. Izumi, Y. Kajihara, Angew. Chem. Int. Ed. 2012, 51, 3567 – 3572; Angew. Chem.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
[22] [23] [24] [25] [26]
2012, 124, 3627 – 3632; c) M. Izumi, M. Murakami, R. Okamoto, Y. Kajihara, J. Pept. Sci. 2014, 20, 98 – 101. E. C. B. Johnson, S. B. H. Kent, J. Am. Chem. Soc. 2006, 128, 6640 – 6646. a) M. Villain, J. Vizzavona, K. Rose, Chem. Biol. 2001, 8, 673 – 679; b) D. Bang, N. Chopra, S. B. H. Kent, J. Am. Chem. Soc. 2004, 126, 1377 – 1383. V. P. Saxena, D. B. Wetlaufer, Biochemistry 1970, 9, 5015 – 5023. S. K. Chowdhury, V. Katta, B. T. Chait, J. Am. Chem. Soc. 1990, 112, 9012 – 9013. a) M. Mayer, B. Meyer, Angew. Chem. Int. Ed. 1999, 38, 1784 – 1788; Angew. Chem. 1999, 111, 1902 – 1906; b) M. Mayer, B. Meyer, J. Am. Chem. Soc. 2001, 123, 6108 – 6117.
Received: September 1, 2014 Published online on October 2, 2014
ChemBioChem 2014, 15, 2503 – 2507
2507
