Bioconjugate Chem. 2008, 19, 1127–1131
1127
Site-Specific Modification of Candida antarctica Lipase B via Residue-Specific Incorporation of a Non-Canonical Amino Acid Sanne Schoffelen, Mark H. L. Lambermon, Mark B. van Eldijk, and Jan C. M. van Hest* Department of Bioorganic Chemistry, Institute for Molecules and Materials, Radboud University Nijmegen, The Netherlands. Received January 15, 2008; Revised Manuscript Received April 13, 2008
In order to modify proteins in a controlled way, new functionalities need to be introduced in a defined manner. One way to accomplish this is by the incorporation of a non-natural amino acid of which the side chain can selectively be reacted to other molecules. We have investigated whether the relatively simple method of residuespecific replacement of methionine by azidohomoalanine can be used to achieve monofunctionalization of the model enzyme Candida antarctica lipase B. A protein variant was engineered with one additional methionine residue. Due to the high hydrophobicity and low abundance of methionine, this was the only residue out of five that was exposed to the solvent. The use of the CuI-catalyzed [3 + 2] cycloaddition under native conditions resulted in a monofunctionalized enzyme which retained hydrolytic activity. The strategy can be considered a convenient tool to modify proteins at a single position as long as one solvent-exposed methionine is available.
The introduction of novel functionalities into proteins has been extensively studied in recent decades. Chromophores, polymers, and polypeptides have been coupled to proteins to study or manipulate their function (1, 2). In most of these cases, the side chain functionalities of the amino acids cysteine and lysine were targeted for conjugation reactions because of the reactivity of the free thiol and amine moieties, respectively (3–5). However, cysteines are often involved in disulfide bonds that occur inside proteins. Breakage of these bonds can easily lead to loss of structural integrity and, hence, function. Lysine residues are positively charged, predominantly located at the protein surface, and relatively abundant. As a consequence, it is very likely that multiple functional handles are introduced when this residue is targeted for modification (6). Since the properties of the bioconjugates thus generated are influenced by the degree and site of modification, it is highly desirable to introduce new functional groups in a more defined manner (7, 8). One approach to achieve this involves the modification of proteinogenic amino acids of which only some residues are exposed to the solvent. As an example, Francis et al. have shown that tyrosine can be functionalized using π-allylpalladium complexes (9). Only 4 tyrosines are present in their model protein chymotrypsinogen (245 amino acids in total). Two residues are surface-accessible and were therefore selectively modified. This strategy is somewhat limited, since most hydrophobic amino acids have no suitable reactivity in the side chain. One way to overcome this limitation is by replacing natural amino acids with non-canonical analogues with similar structure. This replacement can be performed in a residuespecific manner using auxotrophic bacterial strains. The bacteria, being unable to produce one of the natural amino acids, incorporates the supplied non-canonical analogue at each position in the protein where normally the natural amino acid would be present (10). In this way, there are still multiple reactive handles introduced. Non-natural amino acids can also be introduced in a site-directed way via in vivo techniques such as the amber suppressor method (11–13). One of the drawbacks * Correspondence author. Jan C. M van Hest, Department of Bioorganic Chemistry, Radboud University Nijmegen, Heyendaalseweg 135, 6525 AJ Nijmegen, The Netherlands. Phone: +31 24 3653204, Fax: +31 24 3653393, E-mail:
[email protected].
of the latter method is that it involves complicated and timeconsuming molecular biology procedures. We hypothesized that, as an alternative to tyrosine, methionine would be a good candidate for monofunctionalization of proteins. It is one of the least abundant amino acids in proteins and is among the most hydrophobic residues. Except for the N-terminal residue, it is expected to be predominantly buried inside proteins (14, 15). The methyl thioether is not very reactive, but it has been shown that the non-natural amino acid azidohomoalanine can efficiently replace methionine when a methionine auxotrophic E. coli strain is used for protein production (16). Upon introduction of the azide moiety, it is possible to apply the bioorthogonal CuI-catalyzed [3 + 2] cycloaddition (17), which has been shown before to work very well for protein modification (18–21). Although the replacement of methionine by reactive analogues has been well-established, it has not been investigated whether the limited availability of this residue can be used to control the number of conjugation sites. Previously, recombinant proteins were chemoselectively modified by replacing methionine with azidohomoalanine, followed by the Staudinger ligation under denaturing conditions (22). The presence of eight azides led to the formation of at least five different ligation products in a reduced environment. The use of native conditions during the ligation reaction is expected to limit the labeling to solventexposed azide residues only. In order to study whether this relatively simple method can be used to site-specifically functionalize proteins, we have investigated the model enzyme Candida antarctica lipase B (CalB). The protein variant that was used contains five methionine residues, four of which are buried inside the protein. We show that replacing methionine 1 with azidohomoalanine 2 (Scheme 1) leads to the introduction of only one accessible azide for reaction with alkynes via the CuI-catalyzed [3 + 2] cycloaddition, also known as the click reaction. Thus, a residuespecific substitution leads effectively to an enzyme which can be functionalized at a single defined position. Importantly, we found that the enzyme remains active after non-natural amino acid incorporation and click chemistry. CalB (EC 3.1.1.3) is widely used in organic synthesis as a biocatalyst with broad substrate specificity. The enzyme main-
10.1021/bc800019v CCC: $40.75 2008 American Chemical Society Published on Web 05/08/2008
1128 Bioconjugate Chem., Vol. 19, No. 6, 2008
Communications
Scheme 1. Non-Canonical Amino Acid Azidohomoalanine 2 as a Structural Analogue of Methionine 1
tains its activity in organic solvents and is used for various applications such as amidation reactions, dynamic kinetic resolution, esterification, and ring-opening polymerization (23, 24). The processed protein consists of 317 amino acids and has a molecular weight of 33 kDa. We cloned the gene encoding this polypeptide in an expression plasmid containing an N-terminal periplasmic localization signal and a C-terminal histidine tag (25). As a consequence of the cloning strategy we used, our CalB gene encoded 5 methionine residues, whereas wild-type CalB contains 4. The extra residue was the first amino acid after cleavage of the periplasmic localization peptide (Figure 1). Protein synthesis was induced in methionine auxotrophic E. coli bacteria. Residue-specific incorporation of 2 was achieved using the method described elsewhere (16). The mass of CalB containing azidohomoalanine (AHA-CalB) had to be 25 Da less than the mass of CalB with methionine residues (Met-CalB), i.e., 5 Da for each methionine-azidohomoalanine replacement. This mass difference was detected by ESI-TOF mass spectrometry (Figure 2). Additional peaks appeared as well, with a mass difference of 63 Da between adjacent peaks. This is due to the presence of traces of Cu ions most probably coordinated to the histidine tag. To investigate whether the azide moieties in the protein were reactive toward alkynes, AHA-CalB was incubated with alkynefunctionalized dansyl 3 (20 equiv) in the presence of 1 mM CuSO4, sodium ascorbate, and bathophenanthroline ligand 4 (all 30 equiv). The reaction was allowed to proceed overnight at room temperature in phosphate buffer (pH 7.0). Met-CalB was used as a negative control. Protein samples were dialyzed against buffer to remove unreacted dansyl and click reagents. Proteins were separated by electrophoresis, and labeling was verified by
Figure 2. ESI-TOF mass spectra of (A) Met-CalB and (B) AHA-CalB. Upon deconvolution, peaks corresponding to the calculated masses (34 269.72 Da for Met-CalB and 34 244.93 Da for AHA-CalB) were detected.
Figure 3. Labeling of AHA-CalB with alkynyl-dansyl 3 in the presence of CuSO4, ascorbate, and bathophenantroline ligand 4. Reactions were analyzed by SDS-PAGE, with gels stained with Coomassie after fluorescence visualization by UV.
Figure 1. Structure of mature CalB with methionine residues shown in red and a substrate (yellow) positioned near the active site. The N and C termini are indicated as well. The conformation was derived from the PDB file ‘1tca’ with addition of a methionine and a glycine to the N-terminus.
visualizing the fluorescent bioconjugate by UV (Figure 3). Only AHA-CalB treated with Cu catalyst and 3 turned out to be specifically labeled. To identify which of the azides were reactive toward the alkyne-functionalized dansyl, reacted and nonreacted AHA-CalB and Met-CalB samples were treated with trypsin, and digests were analyzed by MALDI-TOF mass spectrometry. As expected, fragments containing methionine in Met-CalB proved to be five units lower in mass in the AHA-CalB sample, due to
Communications
Bioconjugate Chem., Vol. 19, No. 6, 2008 1129
Figure 4. MALDI-TOF spectra of two tryptic fragments derived from (A) Met-CalB incubated with dansyl, (B) AHA-CalB, and (C) AHA-CalB reacted with dansyl. The peak at 1055/1050 Da comes from the peptide fragment with residue numbers 130-138. The peak at 1518/1513 Da originates from the N-terminal peptide fragment (residue numbers 1-15) to which alkynyl-dansyl can be linked. Attachment of dansyl causes a mass shift of 288 Da, leading to a peak at 1801 Da.
Figure 5. ESI-TOF mass spectrum of AHA-CalB reacted with dansyl. One dansyl was attached to AHA-CalB resulting in addition of 288 Da to the AHA-CalB mass of 34 245 Da. As expected, the nonreactive fraction of Met-CalB was detected at 34 269 Da.
the presence of azidohomoalanine (Figure 4). For example, the peptide fragment of residues 130-138 containing methionine at residue number 131 (1055 Da) was detected as a fragment of 1050 Da in the tryptic digest of unreacted AHA-CalB. The same fragment was present in the digest of dansyl-reacted AHACalB, indicating that no dansyl had been attached to this azide moiety. By contrast, a mass shift of 288 Da (corresponding to the weight of dansyl) was detected for the peptide fragment of residues 1-15 containing an azide at residue position 1. It was therefore concluded that this azide, being located at the N-terminus of the processed protein, was accessible for ligation reactions. It was observed that the labeling with dansyl was quantitative, as no residual signal at 1513 Da was detected. Finally, these data show that, as expected, there is a small fraction of methionine present in AHA-CalB (see small peaks at 1055 and 1518 Da in Figure 4B,C). The degree of azidohomoalanine incorporation was shown to be 90%. The addition of dansyl to AHA-CalB was further analyzed by ESI-TOF mass spectrometry. Only one dansyl molecule was found to be attached to the protein, giving the expected mass of 34 533 Da (Figure 5). The fraction of CalB which contained methionine instead of azidohomoalanine was detected as a residual peak at 34 269 Da.
Figure 6. Labeling of AHA-CalB and AHA-CalB(M > G) with PEG5000 and alkynyl-dansyl 3. (A) PEGylation reactions, analyzed by SDS-PAGE using Coomassie staining: 1, Met-CalB; 2, AHA-CalB; 3, AHA-CalB(M > G). (B) Separation of PEGylated AHA-CalB from nonreacted CalB by size-exclusion chromatography. (C) Click reactions with alkynyl-dansyl, analyzed by SDS-PAGE, and visualized by UV prior to Coomassie staining: 1, AHA-CalB; 2, AHA-CalB(M > G).
Subsequently, alkyne-functionalized PEG(5000) was coupled to AHA-CalB. After removal of the excess PEG(5000) by clicking to azide-functionalized resin (26), attachment of the polymer to CalB was analyzed by gel electrophoresis (Figure 6A). A single band shift was detected between 37 and 48 kDa. Nonfunctionalized CalB was separated from the PEGylated fraction by size-exclusion chromatography (Figure 6B). The chromatogram showed two peaks indicating that only one derivative had been produced. The N-terminal methionine codon was replaced by glycine, indicated as AHA-CalB(M > G). Upon reaction with alkynylPEG(5000), no pegylation was observed by gel electrophoresis. Moreover, labeling of AHA-CalB(M > G) with alkynyl-dansyl did not result in a fluorescent band on SDS-PAGE gel (Figure
1130 Bioconjugate Chem., Vol. 19, No. 6, 2008
Communications
methionine analogue was introduced to act as functional handle. Subsequently, other molecules were covalently linked via the highly selective CuI-catalyzed click reaction. Although multiple azides were introduced, only one moiety appeared to be reactive under native conditions, as the other residues were buried inside the protein. The method seems to be a good substitute for the less-straightforward amber suppressor strategy. Our method can be considered a versatile tool to modify proteins in a site-specific way, as long as one solvent-accessible methionine residue is present. In many cases, it will be the N-terminal methionine that is available. It should be noted, however, that the N-terminal methionine or its analogue can be cleaved off during post-translational processing (27). If no other methionine is accessible, a novel residue can be introduced via a single mutagenesis step. Therefore, the method is not restricted to N-terminal labeling. It can also be used to modify other regions of the protein surface with absolute control over the site of modification. After the entire modification procedure, catalytic activity was partially retained. Part of the hydrolytic activity was lost due to the incorporation of the non-proteinogenic amino acid and the use of CuSO4, ascorbate and ligand. Application of a copperfree click reaction such as the Staudinger ligation, the strainpromoted [3 + 2] azide-alkyne cycloaddition, or the reaction between azides and oxanorbornadienes may lead to less loss in activity (22, 28, 29). As reported elsewhere, the modification strategy has been successfully applied to immobilize CalB on polymersomes (30).
ACKNOWLEDGMENT Figure 7. (A) Hydrolytic activity of Met-CalB and AHA-CalB, and the influence of click reagents. Formation of para-nitrophenol was monitored by detecting absorbance at 405 nm. (+) ) after incubation with CuSO4, ascorbate, and ligand 4; (++) ) after incubation with CuSO4, ascorbate, ligand 4, and alkynyl-dansyl 3. (B) Comparison of the hydrolytic activity of PEGylated CalB and the nonfunctionalized fraction. The latter sample consists of a small fraction of Met-CalB together with nonlabeled AHA-CalB.
6C). This confirms that, of all five azides introduced in nonmutated AHA-CalB, only the N-terminal one was accessible for clicking. The enzymatic activity of different CalB variants was analyzed by hydrolysis of para-nitrophenyl butyrate. Formation of para-nitrophenol by Met-CalB and AHA-CalB was monitored by measuring absorbance at 405 nm. Upon introduction of azidohomoalanine, hydrolytic activity was retained for 75% (Figure 7A). Additionally, we investigated the effect of the click reaction on enzymatic activity. MetCalB and AHA-CalB were incubated in the presence of click reagents, with or without alkynyl-dansyl. Subsequent measurements showed that the reagents had some negative effect on the activities of both Met-CalB and AHA-CalB. As shown in Figure 7A, the activity of Met-CalB was decreased to 70%. The activity of AHA-CalB was reduced to 52% after incubation with dansyl and click reagents, which is 69% of the activity of nontreated AHA-CalB. Additionally, the enzymatic activities of both the PEG-CalB conjugate and the nonpegylated CalB fraction were monitored after size-exclusion chromatography. Comparing the fractions, which were derived from the same click reaction mixture, revealed that there was hardly any loss of activity due to attachment of the polymer itself (Figure 7B). To conclude, we have developed a convenient method to modify CalB protein in a site-directed way. An azide-bearing
The authors thank A. J. Dirks for preparing the alkynefunctionalized PEG5000, and R. Brinkhuis and J. Opsteen for providing the alkynyl-dansyl and azide resin, respectively. We thank E. Akpa for providing the CalB gene and P. van Galen and H. op den Camp for help with mass spectrometry analysis. H. Venselaar is thanked for providing the picture of our CalB protein containing the extra methionine and glycine at the N-terminus. Supporting Information Available: Materials, CalB plasmid cloning, synthesis of azidohomoalanine and alkynylPEG(5000), protein production, click reaction conditions, and characterization procedures including enzyme activity assay, SDS-PAGE, MALDI, and ESI-TOF analysis. This material is available free of charge via the Internet at http://pubs. acs.org.
LITERATURE CITED (1) Pasut, G., and Veronese, F. M. (2006) PEGylation of proteins as tailored chemistry for optimized bioconjugates. Polymer Therapeutics I: Polymers as Drugs, Conjugates and Gene DeliVery Systems 192, 95–134. (2) Federico, R., Cona, A., Caliceti, P., and Veronese, F. M. (2006) Histaminase PEGylation: Preparation and characterization of a new bioconjugate for therapeutic application. J. Controlled Release 115, 168–174. (3) Dirks, A. J., van Berkel, S. S., Hatzakis, N. S., Opsteen, J. A., van Delft, F. L., Cornelissen, J. J. L. M., Rowan, A. E., van Hest, J. C. M., Rutjes, F. P. J. T., and Nolte, R. J. M. (2005) Preparation of biohybrid amphiphiles via the copper catalysed Huisgen [3 + 2] dipolar cycloaddition reaction. Chem. Commun. 4172–4174. (4) Blank, K., Morfill, J., and Gaub, H. E. (2006) Site-specific immobilization of genetically engineered variants of Candida antarctica lipase B. ChemBioChem 7, 1349–1351.
Communications (5) Veronese, F. M., Sacca, B., de Laureto, P. P., Sergi, M., Caliceti, P., Schiavon, O., and Orsolini, P. (2001) New PEGs for peptide and protein modification, suitable for identification of the PEGylation site. Bioconjugate Chem. 12, 62–70. (6) Wang, Y., Youngster, S., Grace, M., Bausch, J., Bordens, R., and Wyss, D. F. (2002) Structural and biological characterization of pegylated recombinant interferon alpha-2b and its therapeutic implications. AdV. Drug DeliVery ReV. 54, 547–570. (7) Thordarson, P., Le Droumaguet, B., and Velonia, K. (2006) Well-defined protein-polymer conjugates-synthesis and potential applications. Appl. Microbiol. Biotechnol. 73, 243–254. (8) Heredia, K. L., Bontempo, D., Ly, T., Byers, T. J., Halstenberg, S., and Maynard, H. D. (2005) In situ preparation of protein “Smart” polymer conjugates with retention of bioactivity. J. Am. Chem. Soc. 127, 16955–16960. (9) Tilley, S. D., and Francis, M. B. (2006) Tyrosine-selective protein alkylation using pi-allylpalladium complexes. J. Am. Chem. Soc. 128, 1080–1081. (10) van Hest, J. C. M., Kiick, K. L., and Tirrell, D. A. (2002) Efficient incorporation of unsaturated methionine analogues into proteins in vivo. J. Am. Chem. Soc. 122, 1282–1288. (11) Wang, L., and Schultz, P. G. (2005) Expanding the genetic code. Angew. Chem., Int. Ed. 44, 34–66. (12) Deiters, A., Cropp, T. A., Summerer, D., Mukherji, M., and Schultz, P. G. (2004) Site-specific PEGylation of proteins containing unnatural amino acids. Bioorg. Med. Chem. Lett. 14, 5743–5745. (13) Kodama, K., Fukuzawa, S., Nakayama, H., Kigawa, T., Sakamoto, K., Yabuki, T., Matsuda, N., Shirouzu, M., Takio, K., Tachibana, K., and Yokoyama, S. (2006) Regioselective carbon-carbon bond formation in proteins with palladium catalysis; New protein chemistry by organometallic chemistry. ChemBioChem 7, 134–139. (14) Janin, J., and Wodak, S. (1978) Conformation of amino acid side-chains in proteins. J. Mol. Biol. 125, 357–386. (15) Cornette, J. L., Cease, K. B., Margalit, H., Spouge, J. L., Berzofsky, J. A., and DeLisi, C. (1987) Hydrophobicity scales and computational techniques for detecting amphipathic structures in proteins. J. Mol. Biol. 195, 659–685. (16) Link, A. J., and Tirrell, D. A. (2005) Reassignment of sense codons in vivo. Methods 36, 291–298. (17) Rostovtsev, V. V., Green, L. G., Fokin, V. V., and Sharpless, K. B. (2002) A stepwise Huisgen cycloaddition process: Copper(I)-catalyzed regioselective “ligation” of azides and terminal alkynes. Angew. Chem., Int. Ed. 41, 2596–2599. (18) Gupta, S. S., Kuzelka, J., Singh, P., Lewis, W. G., Manchester, M., and Finn, M. G. (2005) Accelerated bioorthogonal conjugation: A practical method for the Ligation of diverse functional molecules to a polyvalent virus scaffold. Bioconjugate Chem. 16, 1572–1579.
Bioconjugate Chem., Vol. 19, No. 6, 2008 1131 (19) Beatty, K. E., Xie, F., Wang, Q., and Tirrell, D. A. (2005) Selective dye-labeling of newly synthesized proteins in bacterial cells. J. Am. Chem. Soc. 127, 14150–14151. (20) Iida, S., Asakura, N., Tabata, K., Okura, I., and Kamachi, T. (2006) Incorporation of unnatural amino acids into cytochrome c3 and specific viologen binding to the unnatural amino acid. ChemBioChem 7, 1853–1855. (21) van Kasteren, S. I., Kramer, H. B., Jensen, H. H., Campbell, S. J., Kirkpatrick, J., Oldham, N. J., Anthony, D. C., and Davis, B. G. (2007) Expanding the diversity of chemical protein modification allows post-translational mimicry. Nature 446, 1105–1109. (22) Kiick, K. L., Saxon, E., Tirrell, D. A., and Bertozzi, C. R. (2002) Incorporation of azides into recombinant proteins for chemoselective modification by the Staudinger ligation. Proc. Natl. Acad. Sci. U.S.A. 99, 19–24. (23) Gotor-Fernandez, V., Busto, E., and Gotor, V. (2006) Candida antarctica lipase B: An ideal biocatalyst for the preparation of nitrogenated organic compounds. AdV. Synth. Catal. 348, 797– 812. (24) Peeters, J., Palmans, A. R. A., Veld, M., Scheijen, F., Heise, A., and Meijer, E. W. (2004) Cascade synthesis of chiral block copolymers combining lipase catalyzed ring opening polymerization and atom transfer radical polymerization. Biomacromolecules 5, 1862–1868. (25) Blank, K., Morfill, J., Gumpp, H., and Gaub, H. E. (2006) Functional expression of Candida antarctica lipase B in Eschericha coli. J. Biotechnol. 125, 474–483. (26) Opsteen, J. A., and van Hest, J. C. M. (2005) Modular synthesis of block copolymers via cycloaddition of terminal azide and alkyne functionalized polymers. ChemComm 1, 57–59. (27) Wang, A., Winblade Nairn, N., Johnson, R. S., Tirrell, D. A., and Grabstein, K. (2008) Processing of N-terminal unnatural amino acids in recombinant human interferon-β in Escherichia coli. ChemBioChem 9, 324–330. (28) Agard, N. J, Prescher, J. A., and Bertozzi, C. R. (2004) A strain-promoted [3 + 2] azide-alkyne cycloaddition for covalent modification of blomolecules in living systems. J. Am. Chem. Soc. 126, 15046–15047. (29) van Berkel, S. S., Dirks, A. J., Debets, M. F., van Delft, F. L., Cornelissen, J. J. L. M., Nolte, R. J. M., and Rutjes, F. P. J. T. (2007) Metal-free triazole formation as a tool for bioconjugation. ChemBioChem 8, 1504–1508. (30) van Dongen, S. F. M., Nallani, M., Schoffelen, S., Cornelissen, J. L. M., Nolte, R. J. M., and van Hest, J. C. M. (2008) A block copolymer for functionalisation of polymersome surfaces. Macromol. Rapid Commun. 29, 321–325. BC800019V