ARTICLE Expanding Substrate Specicity of GT-B Fold Glycosyltransferase Via Domain Swapping and High-Throughput Screening

Sung-Hee Park,1 Hyung-Yeon Park,2 Jae Kyung Sohng,3 Hee Chan Lee,3 Kwangkyoung Liou,3 Yeo Joon Yoon,4 Byung-Gee Kim1,5 Institute of Molecular Biology and Genetics, Interdisciplinary Program for Bioengineering, Seoul National University, Sillim-dong, Gwanak-gu, Seoul 151-742, South Korea; telephone: 82-2-880-6774; fax: 82-2-876-8945; e-mail: byungkim@snu.ac.kr 2 Department of Chemistry, Inha University, Incheon, South Korea 3 Institute of Biomolecule Reconstruction, SunMoon University, Chungnam, South Korea 4 Division of Nano Sciences, Ewha Womans University, Seoul, South Korea 5 Institute of Bio Engineering, School of Chemical and Biological Engineering, Seoul National University, Seoul, South Korea
Received 30 April 2008; revision received 7 August 2008; accepted 26 August 2008 Published online 29 September 2008 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/bit.22150
1

Introduction
ABSTRACT: Glycosyltransferases (GTs) are crucial enzymes in the biosynthesis and diversication of therapeutically important natural products, and the majority of them belong to the GT-B superfamily, which is composed of separate N- and C-domains that are responsible for the recognition of the sugar acceptor and donor, respectively. In an effort to expand the substrate specicity of GT, a chimeric library with different crossover points was constructed between the N-terminal fragments of kanamycin GT (kanF) and the C-terminal fragments of vancomycin GT (gtfE) genes by incremental truncation method. A plate-based pH color assay was newly developed for the selection of functional domain-swapped GTs, and a mutant (HMT31) with a crossover point (N-kanF-669 bp and 753 bp-gtfE-C) for domain swapping was screened. The most active mutant HMT31 (50 kDa) efciently catalyzed 2-DOS (aglycone substrate for KanF) glucosylation using dTDP-glucose (glycone substrate for GtfE) with kcat/Km of 162.8 0.1 mM1 min1. Moreover, HMT31 showed improved substrate specicity toward seven more NDP-sugars. This study presents a domain swapping method as a potential means to glycorandomization toward various syntheses of 2-DOS-based aminoglycoside derivatives. Biotechnol. Bioeng. 2009;102: 988994. 2008 Wiley Periodicals, Inc. KEYWORDS: natural product; antibiotics; glycosyltransferase; molecular evolution; domain swapping; high-throughput screening (HTS)

Correspondence to: B.-G. Kim Additional Supporting Information may be found in the online version of this article.

Natural products harboring sugar moieties are important sources of pharmaceutical or agricultural drugs (Newman et al., 2003). These sugar units participate frequently in the interactions between drugs and biological targets, and perform crucial roles in increasing solubility, detoxication and antibacterial activity (Walsh et al., 2003; WeymouthWilson, 1997). Therefore, diversication of natural product glycosylation is considered as a basis of current drug developments (Blanchard and Thorson, 2006; Salas and Mendez, 2007; Thibodeaux et al., 2007). Glycosyltransferases (GTs) catalyze the transfer of a sugar moiety from a nucleotide activated sugar to aglycone, accordingly, the promiscuities of GTs perform a signicant function in the glycorandomization of therapeutically important natural products (Floss, 2001; Fu et al., 2003; Leadlay, 1997; Melancon et al., 2006; Mendez and Salas, 2001; Rupprath et al., 2005; Salas and Mendez, 2007; Yang et al., 2004). Although the successful transfer of a variety sugars onto glycopeptides (Losey et al., 2001, 2002) and polyketides (Blanchard and Thorson, 2006) has been previously achieved, stringent substrate specicity of GTs still remains as a severe limitation to natural product diversication (Albermann et al., 2003; Floss, 2006; Zhang et al., 2007). Moreover, the substrate specicities of GTs are frequently restricted within the structurally related onto types of aglycones, like the case of vicenistatin (Minami et al., 2005), oleandomycin (Yang et al., 2005), and novobiocin (Albermann et al., 2003; Freel Meyers et al., 2003). 2008 Wiley Periodicals, Inc.

988

Biotechnology and Bioengineering, Vol. 102, No. 4, March 1, 2009

To overcome the narrow substrate specicity of GTs, molecular evolution approaches have been accomplished via random mutagenesis and directed evolution. However, owing to the lack of simple and easy high-throughput selection methods and crystallographic structure data, their applications were quite limited. The chimera study of urdamycin GT (UrdGT1b and UrdGT1c) (Hoffmeister et al., 2001) and quercetin GT (UGT74F1 and UGT74F2) (Cartwright et al., 2008), the high homology between two genes was required to determine chimeric region. The point mutation study was begun by determining the crystal structure of phenolic GT (UGT72B1) with effort to estimate 44 mutation sites (Brazier-Hicks et al., 2007). A recent study on the oleandomycin GT (OleD) of which crystal structure was examined, showed that the directed evolution of GT-B superfamily permitted to expand its acceptor specicity as well as its sugar donor specicity (Williams et al., 2007). However, it was only limited to uorescent acceptors which require quenching the uorescence for the detection of glycosylation. In this study, we introduced a pH color assay method as an alternative to detect glycosylation of nonuorescent acceptor, such as 2-deoxystreptamine (2-DOS) which is key aglycone of aminoglycoside kanamycin (Fig. 1a).

Considering the quantitative pH-sensitive galactosyltransferase assay (Deng and Chen, 2004), which showed a 50% higher sensitivity than the most common HPLC assay, we could expect that the pH decrease caused by glycosidic bond formation (Fig. 1b) also lead to the color change of the corresponding pH indicator during the GT reaction in the recombinant cell (Fig. 1c). To overcome the problems associated with the narrow substrate specicities and low catalytic efciency of GT, a domain swapping strategy for the in vitro evolution was proposed, considering the structures of GTs involved in natural product biosynthesis. Interestingly, the GT-B superfamily of the enzymes has very distinctive acceptor-recognition (N-terminus) and donor-recognition (C-terminus) domains, connected by a linker loop region (Coutinho et al., 2003; Hu et al., 2002, 2003; Mulichak et al., 2001, 2003). Although the rational linker design was carried out in an effort to verify the domain swapping strategy in our previous study (Park et al., 2008, submitted), where three types of linkers were compared for the activities of the chimeric enzymes, the ideal linker sequence was still remained unclear. To design the best enzyme with the optimum linker, the key issue in the method of domain swapping is to generate a full library

Figure 1. Reaction scheme catalyzed by glycosyltransferase. a: Left, 2-deoxystreptamine (2-DOS) is the core structure (bold) of kanamycin glycosylated by KanF; right, glucose (bold) is transferred by GtfE catalysis from dTDP-glucose in vancomycin biosynthesis. b: Proton ion is released according to O-glycosidic bond formation between NDP-sugar donor and acceptor. c: Cresol red, acidbase indicator, could easily exchange its structure and showed purple (P) in the alkaline condition, but turned into yellow (Y) after an exposure to acidic environment.

Park et al.: Evolution of GT-B Enzyme by Domain Swapping Biotechnology and Bioengineering

989

of the fused protein which is harboring every single crossover point connecting the N- and C-terminal domain from two individual GTs. Here, we employed a timedependent incremental truncation method (Arnold and Georgiou, 2003; Ostermeier et al., 1999b), which simply allows creation of a chimeric gene library encoding for a random length of the N-terminus of the KanF (kanamycin GT) (Kharel et al., 2004) and the C-terminus of the GtfE (vancomycin GT) (Losey et al., 2002), respectively. Using the simple color assay and the chimeric gene library, our results successfully demonstrated that the domain swapping method is one of the interesting options to change the glycan substrate specicity for the GT-B fold enzymes.

Materials and Methods

Generals Streptomyces kanamyceticus (ATCC 12853) and Amycolatopsis orientalis (ATCC 19795) were used as the source strains for kanF and gtfE. E. coli strain DH5a and BL21 (DE3) were used as the host strains for gene cloning and expression, respectively. All restriction endonucleases and T4 DNA ligase were obtained from New England Biolabs. The pET28a expression vector was purchased from Novagen. pfu DNA polymerase was purchased from Genenmed (Seoul, Korea). ExoIII nuclease and Mung Bean nuclease were purchased from Promega (WI, USA). DNA primers were obtained from Custom Oligonucleotide Synthesis Manufacture Ofce (Seoul, Korea). NDP-sugars and 2-deoxystreptamine were purchased from Genechem Inc. (Deajeon, Korea). All other materials obtained from Sigma. Mass spectra were obtained using LCQ DECA XP ion trap MS/MS (ThermoFinnigan) equipped with an Agilent 1100 HPLC. The absorbance of pH indicators was measured using Multiskan Spectrum (Thermo Labsystems).

of pkanF28 to obtain pIT-KE. Incremental truncation was performed on linearized pIT-KE by HindIII digestion with 2.5 U ExoIII nuclease at 378C in 100 mL of 66 mM Tris (pH 8.0) and 0.66 mM MgCl2 buffer. 10 mL aliquots were removed at 0.5 min intervals and quenched by addition to cold DNA purication buffer (Ostermeier, 2003; Ostermeier et al., 1999a). Following the inactivation of ExoIII nuclease, combined DNA was puried with QIAquick kit and resuspended in 50 mL of water. Diversely generated linear DNA was digested by Mung Bean nuclease (10 U, 30 mM sodium acetate buffer, pH 5.0, 50 mM NaCl and 1 mM ZnCl2) at 378C for 30 min to form blunt-end DNA. The fraction of library members having genes within the desired size range (>6,000 bp, approximately) was puried from agarose gel and subsequently ligated into a circular plasmid. Ligation was carried out at 48C overnight in a total volume of 10 mL using T4 DNA ligase. The recyclized DNA was desalted and transformed into E. coli BL21 (DE3) by electroporation. Transformants were grown on 50 LB medium plates (150 mm in diameter) containing appropriate marker.

pH Color Assay E. coli BL21 (DE3) bacterial cells harboring a plasmid pET28a for expression of glycosyltransferase were grown on a LB solid media of the same condition containing 0.01 mM IPTG and 0.05 mM cresol red and 50 mg mL1 kanamycin for overnight at 378C. Sterilized and dried Whatman lter papers were prepared for plate replication and glycosylation reaction. Bacterial colonies were adsorbed to lter paper by covering and stamping with rubber roller several times. This replica lter was dipped into 2 mL cresol red solution containing 2 mM 2-deoxystreptamine, 10 mM dTDPglucose and 50 mM cresol red. These were consecutively incubated for 5 h at 378C. Schematic procedure was described in Supplementary materials.

Library Construction The gtfE gene (1,227 bp) was amplied by PCR with a forward primer AttattAAGC TTATGCGTGTGTTGTTGTCGACGTGCGGAAGT and reverse primer AttattCTCG AGTCAGGCGGGAACGGCGGGCTTCTCCTC. kanF gene (1,161 bp) was amplied with a forward primer agcgttGAATTCATGCAGGTACAGATCCTG and reverse primer agcgttAAGCTTCTACCGTTCTTTGCCGAGTA. PCR conditions were 958C, 10 min for denaturation, 30 cycles of 958C, 1 min, 618C, 1 min, 728C, 2 min for extension of DNA. Each PCR product was subcloned into T-vector to obtain pTkanF and pTgtfE. After conrmation of both sequences, EcoRI-HindIII fragment derived from pTkanF was puried by agarose gel electrophoresis, extracted using QIAquick Gel extraction Kit (QIAGEN), and inserted into pET28a (5,369 bp) to obtain pkanF28. A HindIII-XhoI fragment derived from pTgtfE was introduced into a HindIII-XhoI site

Quantitative Analysis To determine the kinetic parameters, enzyme assays were conducted in a total volume of 100 mL in 1 mM phosphate buffer (pH 8.0) containing 0.05 mM cresol red and puried enzyme (0.2 mg) using 96-well plate. dTDP-glucose held constant at 10 mM, and 2-deoxystreptamine levels varied. The absorbance at 436 nm was recorded for each sample at 30 s intervals for a total of 5 min using Multiskan Spectrum (Thermo Labsystems). The proton concentration was calculated on the basis of the HCl calibration curve. To generate a calibration curve, 10 mM dTDP-glucose, 0.05 mM cresol red, puried enzyme and different quantities of hydrochloric acid were mixed to nal concentrations of 05 mM in 1 mM phosphate buffer (pH 8.0), and the absorbance was measured at 436 nm. Note that this condition was generated with the same enzyme mixture as was used in the actual assay, omitting only the acceptor

990

Biotechnology and Bioengineering, Vol. 102, No. 4, March 1, 2009

substrate. The initial velocities were tted to the Michaelis Menten equation and kinetic parameters were calculated via LineweaverBurk plot of the velocity versus the substrate concentration. To determine the relative specicity, 10 mM NDP-sugars, 0.05 mM cresol red and enzyme (or crude cell extracts) were mixed in 1 mM phosphate buffer (pH 8.0) using 96-well plate, and the reaction was started with adding the 2 mM 2-DOS. All measurements were repeated in triplicate and averaged.

Results and Discussion

A pH-Dependent Color Assay for High-Throughput Screening To select functional hybrid enzymes, a plate-based wholecell color assay for glycosylation was developed. This technique is selection of colonies based upon the color changes of the pH indicator according to the proton release during glycosidic bond formation between the sugar donor and the aglycone acceptor (Fig. 1). The pKa values of the pH indicators should be considered in the selection of an optimal pH indicator for the glycosylation reactions. The pKa of cresol red is 8.32, and its color changes from purple to yellow over a pH range of 7.28.8, which corresponds well with the optimal pH range of known glycosylation (Freel Meyers et al., 2003; Fu et al., 2003; Losey et al., 2001; Minami et al., 2005). In order to conrm the whole-cell color change, the previously prepared glycosyltransferase (KE chimera) which evidenced dTDP-glucose transfer activity onto 2-DOS (Park et al., 2008, submitted), was assessed in accordance with the procedure described in Materials and Methods section. As a result, the bacteria harboring the KE chimera turned yellow after air drying, but the negative control (pET28a vector harboring E. coli BL21) remained purple (Supplementary materials).

Figure 2.

Schematic representation of time-dependent incremental truncation.

Time-Dependent Incremental Truncation Library Construction and Screening The time-dependent incremental truncation protocol consisted of ve steps (Fig. 2). The kanF gene (1,161 bp) and the gtfE gene (1,227 bp) were sequentially cloned within the same vector (pET28a). Library of incremental truncation from C-terminus of the kanF gene and from the N-terminus of the gtfE gene was created by ExoIII treatment with time intervals beginning from a unique restriction enzyme site (HindIII) between the two genes. This allowed simultaneous truncation of both genes to opposite directions. Diversely generated linear DNA strands were digested by single-strand nuclease to form a blunt-end followed by intra-molecular ligation to recyclize vectors. The gel data of size distribution of library was presented in Supplementary materials. Approximately, 106107 transformants were obtained by electroporation of plasmid library into the auxotroph E. coli

BL21. The linker plus the C-terminus fragment of kanF consists of the DNA coding for residues 5261,161 (636 bps). The N-terminus fragment plus linker of gtfE consists of the residues 1681 (681 bp). If these two fragments are crossed each other, the number of possible combinations of hetero-dimers are theoretically 433,116. The number of our transformants easily exceeded the theoretical numbers. The detailed information of theoretical distribution of the truncation lengths in library can be found in the previous studies (Ostermeier, 2003; Ostermeier et al., 1999a). To screen hybrid GT, yellow colonies (positives) and purple colonies (negatives) were selected and streaked on LB medium plates. 30 non-active colonies (#1#30) and 30 active colonies (#31#60) were selected and compared their color changes with the controls (Fig. 3). After purication, their plasmids were subjected to the sequence analysis to compare their hybrid sequences and crossover points. The partial results of the sequence analysis are represented in Figure 4. The functional GT hybrids among the screened mutants were revealed to have a gene size around 1,400 bp with evenly distributed two fragments from the parental genes (i.e. approximately 650 bp from N-terminus of kanF and 750 bp from C-terminus of gtfE), whereas the non-functional mutants had wider ranges of the size distribution (9001,800 bp). The #31 (represented as A in Fig. 4a) harboring 669 bp of the N-terminus of kanF and 753 bp of the C-terminus of gtfE evidenced the highest level of the GT activity, and named HMT31. The modeling result (Supplementary materials) showed that the linker sequence of HMT31 was 210FVGRIAHEKGWRHANNQSAYRRYGEPLNSRR240. The 210FVGRIAHEKGWRHA223 and 224NNQSAYRRYGEPLNSRR240 sequences were from KanF and GtfE, respectively (Fig. 5).

Park et al.: Evolution of GT-B Enzyme by Domain Swapping Biotechnology and Bioengineering

991

Figure 5. Linker and key residues of HMT31. The ligand docking study showed that Asp35, Asn38, Arg141, His374, Gly378 and Gln399 were key catalytic residue of HMT31. Under lined, linker peptide; grey, N-terimal domain from KanF; black, C-terminal domain from GtfE; bold, key residue from ligand docking simulation.

Figure 3.

Library colonies tested pH color assay. KE chimera, positive control; pET28a, negative control; HMT31, active hybrid colony.

Characterization of HMT31 pH sensitive assay (Deng and Chen, 2004) was used for quantitative analysis of GT reaction (Park et al., 2008, submitted). When exposed to the appropriate environment (pH 79), the absorbance changes showed a good linear relationship with the cresol red concentration at 436 nm, allowing us to determine the kinetic parameters and relative substrate specicity. As a preliminary test, the cell extracts of E. coli BL21 harboring pET28a (negative control), KE chimera (positive control) and HMT31 were compared. The

time prole of absorbance changes indicated that HMT31 exhibits threefold higher reaction rate than the positive control (Fig. 6). In order to elucidate the in vitro activity, HMT31 (50 kDa) with His6 tag was successfully overexpressed in E. coli and puried. The molecular weights of the product and 2-DOS were veried by using ESI-MS (Supplementary materials). To determine the kinetic parameters of HMT31, enzyme assays were conducted as described in Materials and Methods section. The proton concentration changes corresponding to the absorbance changes were calculated based on the calibration curve (Supplementary Materials). The kcat/Km value of HMT31 was estimated to be 162.8 0.1 mM1 min1. It was approximately four-fold improved level compared to that of KE chimera harboring the rationally designed linker (Park et al., 2008, submitted). To assess the effect of the domain swapping on substrate specicity, wild-type KanF, GtfE and HMT31 were

Figure 4.

Sequence analysis of the selected library. White bar, kanF gene; grey bar, gtfE gene; ak, non-functional hybrid; AK, Functional hybrids. A is equal to HMT31.

Figure 6. Change of absorbance at 436 nm with time for cell extracts. Blank, negative control (E. coli BL21 harboring pET28a); KE chimera, positive control.

992

Biotechnology and Bioengineering, Vol. 102, No. 4, March 1, 2009

Figure 7. Activity of HMT31 toward NDP-sugars. a: Structures of NDP-sugar donors tested. 1, dTDP-D-glucose; 2, UDP-D-glucose; 3, dTDP-L-rhamnose; 4, dTDP-4-keto-6-Ddeoxy-glucose; 5, GDP-D-mannose; 6, dTDP-D-fucosamine; 7, UDP-D-galactose; 8, dTDP-2-deoxy-D-glucose. b: Conversion rates of HMT31 compared to wild type KanF and KE chimera toward NDP-sugars indicated in (a) with 2-DOS as acceptor. Conversion rate of dTDP-glucose catalyzed by KanF was dened as 100 (vertical axis).

compared their activities toward nine NDP-sugars in the presence of 2-DOS as an sugar acceptor as described in Materials and Methods section. The wild-type GtfE did not show any activity for 2-DOS, whereas wild type KanF showed weak glucosylation activity to 2-DOS (Park et al., 2008, submitted). HMT31 enzyme displayed almost a 16-fold improved activity for dTDP-D-glucose 1, seven times for GDP-D-mannose 5, and three times for UDP-D-galactose 7, as compared to wild-type KanF. In addition, interestingly, it acquired new donor specicities for dTDP-L-rhamnose 3, dTDP-4-keto-6-D-deoxyglucose 4. Compared to rationally designed KE chimera, HMT31 showed threefold improved activity toward UDP-D-galactose 7, and showed new activity toward dTDP-4-keto-6-D-deoxyglucose 4. HMT31 and KE chimera presented almost the same sugar donor activity for UDP-D-glucose 2, dTDP-L-rhamnose 3, GDP-D-mannose 5, dTDP-D-fucosamine 6 and dTDP-2-deoxy-D-glucose 8 (Fig. 7).

enzymes, and several amino acids at the corresponding end of the N- and C-domain of each enzyme were revealed as potential key residues for the expansion of the substrate specicity. In the case of HMT31, eight NDP-sugars were accepted as donors to 2-DOS, and among them, three were the new substrates for 2-DOS. Consequently, this domain swapping method not only showed a potential for further investigation toward the synthesis of a variety of 2-DOSbased aminoglycoside derivatives, but also provide a new platform for the evolution of GT-B superfamily enzymes such as antibiotic GTs to overcome the principal limitations in the combinatorial synthesis of natural products.
This work was supported by Ministry of Commerce, Industry and Energy (grant no. 10023194) and by a grant (no. 20050401034682) from the BioGreen 21 Program, Rural Development Administration, Republic of Korea.

Conclusion
In conclusion, we have developed a whole cell color assay, and have applied it to the high-throughput screening of hybrid GTs showing different sugar donor substrate specicity. The color change results clearly indicated that the pH color assay using cresol red was sensitive enough to detect glycosylation in the E. coli whole cell system through bare eye. This method is especially advantageous over detecting glycosylation of non-uorescent acceptors of natural products such as aminoglycosides and glycoconjugates. In order to expand the sugar donor specicity of GTs, functional hybrid enzymes were created between two interclass GT-B fold enzymes using domain swapping strategy. The determination of the crossover point (i.e., linker peptide sequence) was an essential for the domain swapping of GT-B

References
Albermann C, Soriano A, Jiang J, Vollmer H, Biggins JB, Barton WA, Lesniak J, Nikolov DB, Thorson JS. 2003. Substrate specicity of NovM: Implications for novobiocin biosynthesis and glycorandomization. Org Lett 5(6):933936. Arnold FH, Georgiou G. 2003. Methods. In: Walker JM, editor. Molecular biology. Totowa: Humana press. Blanchard S, Thorson JS. 2006. Enzymatic tools for engineering natural product glycosylation. Curr Opin Chem Biol 10(3):263271. Brazier-Hicks M, Offen WA, Gershater MC, Revett TJ, Lim EK, Bowles DJ, Davies GJ, Edwards R. 2007. Characterization and engineering of the bifunctional N- and O-glucosyltransferase involved in xenobiotic metabolism in plants. Proc Natl Acad Sci U S A 104(51):2023820243. Cartwright AM, Lim EK, Kleanthous C, Bowles DJ. 2008. A kinetic analysis of regiospecic glucosylation by two glycosyltransferases of Arabidopsis thaliana: Domain swapping to introduce new activities. J Biol Chem 283(23):1572415731.

Park et al.: Evolution of GT-B Enzyme by Domain Swapping Biotechnology and Bioengineering

993

Coutinho PM, Deleury E, Davies GJ, Henrissat B. 2003. An evolving hierarchical family classication for glycosyltransferases. J Mol Biol 328(2):307317. Deng C, Chen RR. 2004. A pH-sensitive assay for galactosyltransferase. Anal Biochem 330(2):219226. Floss HG. 2001. Antibiotic biosynthesis: From natural to unnatural compounds. J Ind Microbiol Biotechnol 27(3):183194. Floss HG. 2006. Combinatorial biosynthesis potential and problems. J Biotechnol 124(1):242257. Freel Meyers CL, Oberthur M, Anderson JW, Kahne D, Walsh CT. 2003. Initial characterization of novobiocic acid noviosyl transferase activity of NovM in biosynthesis of the antibiotic novobiocin. Biochemistry 42(14):41794189. Fu X, Albermann C, Jiang J, Liao J, Zhang C, Thorson JS. 2003. Antibiotic optimization via in vitro glycorandomization. Nat Biotechnol 21(12):14671469. Hoffmeister D, Ichinose K, Bechthold A. 2001. Two sequence elements of glycosyltransferases involved in urdamycin biosynthesis are responsible for substrate specicity and enzymatic activity. Chem Biol 8(6):557 567. Hu Y, Walker S. 2002. Remarkable structural similarities between diverse glycosyltransferases. Chem Biol 9(12):12871296. Hu Y, Chen L, Ha S, Gross B, Falcone B, Walker D, Mokhtarzadeh M, Walker S. 2003. Crystal structure of the MurG:UDPGlcNAc complex reveals common structural principles of a superfamily of glycosyltransferases. Proc Natl Acad Sci U S A 100(3):845849. Kharel MK, Subba B, Basnet DB, Woo JS, Lee HC, Liou K, Sohng JK. 2004. A gene cluster for biosynthesis of kanamycin from Streptomyces kanamyceticus: Comparison with gentamicin biosynthetic gene cluster. Arch Biochem Biophys 429(2):204214. Leadlay PF. 1997. Combinatorial approaches to polyketide biosynthesis. Curr Opin Chem Biol 1(2):162168. Losey HC, Peczuh MW, Chen Z, Eggert US, Dong SD, Pelczer I, Kahne D, Walsh CT. 2001. Tandem action of glycosyltransferases in the maturation of vancomycin and teicoplanin aglycones: Novel glycopeptides. Biochemistry 40(15):47454755. Losey HC, Jiang J, Biggins JB, Oberthur M, Ye XY, Dong SD, Kahne D, Thorson JS, Walsh CT. 2002. Incorporation of glucose analogs by GtfE and GtfD from the vancomycin biosynthetic pathway to generate variant glycopeptides. Chem Biol 9(12):13051314. Melancon CE 3rd, Thibodeaux CJ, Liu HW. 2006. Glyco-stripping and glyco-swapping. ACS Chem Biol 1(8):499504. Mendez C, Salas JA. 2001. Altering the glycosylation pattern of bioactive compounds. Trends Biotechnol 19(11):449456. Minami A, Uchida R, Eguchi T, Kakinuma K. 2005. Enzymatic approach to unnatural glycosides with diverse aglycon scaffolds using glycosyltransferase VinC. J Am Chem Soc 127(17):61486149.

Mulichak AM, Losey HC, Walsh CT, Garavito RM. 2001. Structure of the UDP-glucosyltransferase GtfB that modies the heptapeptide aglycone in the biosynthesis of vancomycin group antibiotics. Structure 9(7):547557. Mulichak AM, Losey HC, Lu W, Wawrzak Z, Walsh CT, Garavito RM. 2003. Structure of the TDP-epi-vancosaminyltransferase GtfA from the chloroeremomycin biosynthetic pathway. Proc Natl Acad Sci U S A 100(16):92389243. Newman DJ, Cragg GM, Snader KM. 2003. Natural products as sources of new drugs over the period 19812002. J Nat Prod. 66(7):10221037. Ostermeier M. 2003. Theoretical distribution of truncation lengths in incremental truncation libraries. Biotechnol Bioeng 82(5):564577. Ostermeier M, Nixon AE, Shim JH, Benkovic SJ. 1999a. Combinatorial protein engineering by incremental truncation. Proc Natl Acad Sci U S A 96(7):35623567. Ostermeier M, Shim JH, Benkovic SJ. 1999b. A combinatorial approach to hybrid enzymes independent of DNA homology. Nat Biotechnol 17(12):12051209. Park SH, Park HY, Cho BK, Yang YH, Sohng JK, Lee HC, Liou K, Kim BG. 2008. Reconstitution of antibiotics glycosylation by domain swapped chimeric glycosyltransferase. Journal of Molecular Catalysis B: Enzymatic, submitted. Rupprath C, Schumacher T, Elling L. 2005. Nucleotide deoxysugars: Essential tools for the glycosylation engineering of novel bioactive compounds. Curr Med Chem 12(14):16371675. Salas JA, Mendez C. 2007. Engineering the glycosylation of natural products in actinomycetes. Trends Microbiol 15(5):219232. Thibodeaux CJ, Melancon CE, Liu HW. 2007. Unusual sugar biosynthesis and natural product glycodiversication. Nature 446(7139):10081016. Walsh C, Freel Meyers CL, Losey HC. 2003. Antibiotic glycosyltransferases: Antibiotic maturation and prospects for reprogramming. J Med Chem 46(16):34253436. Weymouth-Wilson AC. 1997. The role of carbohydrates in biologically active natural products. Nat Prod Rep 14(2):99110. Williams GJ, Zhang C, Thorson JS. 2007. Expanding the promiscuity of a natural-product glycosyltransferase by directed evolution. Nat Chem Biol 3(10):657662. Yang J, Hoffmeister D, Liu L, Fu X, Thorson JS. 2004. Natural product glycorandomization. Bioorg Med Chem 12(7):15771584. Yang M, Proctor MR, Bolam DN, Errey JC, Field RA, Gilbert HJ, Davis BG. 2005. Probing the breadth of macrolide glycosyltransferases: In vitro remodeling of a polyketide antibiotic creates active bacterial uptake and enhances potency. J Am Chem Soc 127(26):93369337. Zhang C, Fu Q, Albermann C, Li L, Thorson JS. 2007. The in vitro characterization of the erythronolide mycarosyltransferase EryBV and its utility in macrolide diversication. Chembiochem 8(4):385 390.

994

Biotechnology and Bioengineering, Vol. 102, No. 4, March 1, 2009