Bioconversion of Tetracycline Antibiotics to Novel Glucoside Derivatives by Single-Vessel Multienzymatic Glycosylation

J. Microbiol. Biotechnol. (2018), 28(2), 298–304 https://doi.org/10.4014/jmb.1709.09072 Research Article Review jmb

Bioconversion of Tetracycline Antibiotics to Novel Glucoside Derivatives by Single-Vessel Multienzymatic Glycosylation S Ramesh Prasad Pandey1,2,3†*, Luan Luong Chu2†, Tae-Su Kim2,3, and Jae Kyung Sohng1,2,3*

1Department of BT-Convergent Pharmaceutical Engineering, Sun Moon University, Asan 31460, Republic of Korea 2Department of Life Science and Biochemical Engineering, Sun Moon University, Asan 31460, Republic of Korea 3Institute of Biomolecule Reconstruction, Sun Moon University, Asan 31460, Republic of Korea

Received: October 12, 2017 Revised: November 20, 2017 The single-vessel multienzyme UDP-α-D-glucose recycling system was coupled with a Accepted: November 27, 2017 forward glucosylation reaction to produce novel glucose moiety-conjugated derivatives of First published online different tetracycline antibiotic analogs. Among five tetracycline analogs used for the reaction, December 8, 2017 four molecules (chlorotetracycline, doxytetracycline, meclotetracycline, and minotetracycline)

*Corresponding authors were accepted by a glycosyltransferase enzyme, YjiC, from Bacillus licheniformis to produce J.K.S. glucoside derivatives. However, the enzyme was unable to conjugate sugar units to Phone: +82-41-530-2246; Fax: +82-41-544-2919; rolitetracycline. All glucosides of tetracycline derivatives were characterized by ultraviolet E-mail: [email protected] absorbance maxima, ultra-pressure liquid chromatography coupled with photodiode array, R.P.P. and high-resolution quadruple time-of-flight electrospray mass spectrometry analyses. These Phone: +82-41-530-2246; Fax: +82-41-544-2919; synthesized glucosides are novel tetracycline derivatives. E-mail: [email protected]

†These authors contributed Keywords: Novel tetracycline derivatives, glycosylation, single-vessel multienzyme reaction, equally to this work. UDP-α-D-glucose recycling system

S upplementary data for this paper are available on-line only at http://jmb.or.kr.

pISSN 1017-7825, eISSN 1738-8872

Tetracyclines are broad-spectrum antibiotics clinically tetracycline antibiotics because of the spread of the prescribed for treating a variety of bacterial infections [1]. tetracycline resistance gene [6, 7]. Thus, the search for a Several analogs of tetracyclines, such as chlorotetracycline, novel analog of the tetracycline antibiotic to overcome oxytetracycline, tetracycline, and dimethylchlorotetracycline, resistant strains is a current priority in the scientific are naturally produced from different Streptomyces species, community for treating different diseases. including S. aureofaciens [2], S. rimosus, and S. viridofaciens [1]. The search for novel analogs or finding a novel molecule However, some tetracycline analogs, such as methacycline, in place of tetracycline groups of molecules from nature rolitetracycline, doxycycline, and minocycline, are produced requires a substantially long time and meticulous research by semisynthetic approaches [3-5]. Tigecycline (Tygacil) is work. Thus, researchers are focused on subtle modifications one of the glycylcycline group of semisynthetic compounds of previously known lead molecules by various chemical, lately approved by the United States Food and Drug enzymatic, and chemoenzymatic approaches to produce Administration (USFDA) as a third-generation tetracycline more potent analogs. antibiotic drug. Although several tetracycline analogs are In nature, post-modification reactions such as glycosylation, in clinical use to treat various diseases, many commensal methylation, alkylation, acylation, and hydroxylation are and pathogenic strains are recently becoming resistant to extensively used to diversify chemical architecture to

J. Microbiol. Biotechnol. Glycosylation of Natural Products 299

produce different analogs [8]. Recently, Professor Thorson’s S. aureus and enterococci [10-12]. Thus, oritavancin is active group has synthesized 37 different analogs of doxycycline against vancomycin-resistant organisms. Dalbavancin and glycoside analogs using the neoglycosylation approach [9]. telavancin are other second-generation semisynthetic Among the new analogs, a 2’-amino-α-D-glucoside-conjugated glycolipopeptides approved by the USFDA for treatment of derivative exhibited activity comparable to doxycycline skin infections [13]. Thus, glycosylation is attracting attention pharmacophore [9]. Because of the potential effect of sugar to conjugate diverse sugars to different lead molecules. conjugation on doxycycline, we applied an enzymatic Most glycosylation post-modification reactions of natural glycosylation approach to generate analogs of different products are performed by glycosyltransferase (GT) enzymes tetracycline derivatives in this study. Moreover, the USFDA that can transfer the sugar moiety from an activated sugar recently approved oritavancin, a semisynthetic lipogly- donor to acceptor molecules [14, 15]. Several GTs that can copeptide antibiotic, to treat adults with skin infections. conjugate sugars to small molecules have been identified The molecule possesses two 4-epi-vancosamines and a from different sources, such as plants and microbes [16]. glucose monosaccharide in its structure. The 4-epi- However, only a few sugar-conjugated tetracyclines are vancosamine is believed to play a crucial role in exerting reported from natural sources. For example, dactylocycline activity by interacting with peptides nearby, but distinct derivatives [17-19], TAN-1518 [20], and SF2575 [21] are from the terminal D-alanyl-D-alanine (or D-lactate) in both produced by different organisms (Fig. 1A).

Fig. 1. (A) Structures of naturally occurring tetracycline glycosides and their producer organisms, and (B) various tetracycline derivatives used for glucosylation reaction in this study.

February 2018 ⎪ Vol. 28⎪ No. 2 300 Pandey et al.

Fig. 2. Schematic diagram showing the single-vessel UDP-glucose recycling system coupled with the glucosylation reaction for tetracycline glucoside production. The enzymes used in the one-pot reaction mixture are acetate kinase from E. coli K-12 (ACK), which converts UDP to UTP utilizing acetyl-1- phosphate; α-D-glucose 1-phosphate uridylyltransferase from E. coli K12 (GalU), which uses UTP and glucose-1-phosphate and generates UDP-α- D-glucose; and UDP-glycosyltransferase from B. licheniformis (YjiC), which transfers a glucose moiety from UDP-α-D-glucose to acceptor molecules such as tetracyclines.

In the last few years, we have found that a GT from and preparation are explained in the supporting materials. Bacillus licheniformis DSM-13 (YjiC) has broad substrate Five different tetracycline derivatives (chlorotetracycline, flexibility. It is able to accept a wide range of small doxytetracycline, meclotetracycline, minotetracycline, and molecules as acceptors and different NDP-sugars as sugar rolitetracycline) purchased from Tokyo Chemical Industry donors [22-26]. We applied the same YjiC for in vitro glucosylation of selected natural and synthetic tetracycline analogs (Fig. 1B) using a single-vessel UDP-α-D-glucose recycling system (Fig. 2). This is the first report of the production of tetracycline glucosides by an enzymatic glucosylation reaction. The generated molecules are most likely to be novel compounds. Acetyl phosphate kinase (ACK) from Escherichia coli K-12 uses acetyl-Pi, and phosphorylates UDP to generate uridine triphosphate (UTP). Another enzyme, glucose 1-phosphate uridylyltransferase (GalU), also from the same species, transfers UDP from UTP to glucose 1-phosphate and produces UDP-α-D-glucose, which is the donor substrate for YjiC GT. Both glucose 1-phosphate and acetyl phosphate are commercially available at low cost. Soluble fractions of the three proteins (ACK, GalU, and YjiC) were produced in the E. coli BL21(DE3) cytosol. The Fig. 3. SDS-PAGE analysis of soluble proteins used in this molecular mass was approximately 38 kDa for ACK and study. GalU and 45 kDa for YjiC. The respective bands of these The arrow represents the protein band of the respective protein. soluble proteins were clearly observed by SDS-PAGE Lane M: molecular marker size proteins; lane 1: GalU; lane 2: Ack; analysis (Fig. 3). Detail methods for the protein expression and lane 3: YjiC.

J. Microbiol. Biotechnol. Glycosylation of Natural Products 301

(Japan) were used for the in vitro single-vessel glucosylation derivative, it was at 268 and 339 nm (Fig. 4B(ii-iii)). This reactions. The single-vessel reaction was carried out by analysis clearly showed that the new observed peak was a putting all enzymes (ACK, GalU, YjiC) along with their derivative of standard doxytetracycline. These peaks were required components for the reaction in 1 ml of final further analyzed for their exact molecular masses. The volume. For this reaction, each crude soluble protein standard peak showed a prominent spectrum of m/z+ fraction (~30 μg/ml of final concentration) was used. Other 445.1613 for which the calculated mass was 445.1611 Da for ingredients of the reaction mixture were 100 mM Tris-HCl the molecular formula C22H25N2O8 in the protonated state buffer (pH 7.5) with 2 mM UDP-α-D-glucose, 2mM each of (Fig. 4B(iv)). The new peak at tR 2.32 min showed a spectrum + tetracycline derivatives, 10 mM MgCl2, 50 mM α-D-glucose of m/z 607.2132 (Fig. 4B(v)) for which the calculated mass

1-phosphate, and 200 mM acetyl phosphate. The final was 607.2139 Da for the molecular formula C28H35N2O13. reaction mixture was incubated at 37oC for 3 h. After the This exact mass resembles the mass of the monoglucosylated 3 h of incubation, the reaction was quenched by adding a derivative of doxytetracycline. The conversion percentage double volume of chilled nearly 100% methanol. The control of doxytetracycline was relatively small (below 3%). reaction had no UDP-α-D-glucose. The meclotetracycline reaction mixture showed the

UPLC-PDA analysis showed novel peaks in four standard molecule at tR 3.54 min with a new peak at tR tetracycline analogs (chlorotetracycline, doxytetracycline, 2.51 min (Fig. 4C(i)). The maxima UV absorbance of the meclotetracycline, and minotetracycline) (Fig. 1B). However, standard molecule was at 243 and 346 nm for standard no novel peak was observed in the rolitetracycline derivative. meclotetracycline. It was subtly changed to new peaks (at Since novel peaks appeared in four different tetracyclines, 240 and 339 nm) (Fig. 4C(ii-iii)). These same peaks were the same reaction mixture was further analyzed by HR- further analyzed by HR-QTOF ESI/MS analysis to determine QTOF ESI/MS to determine whether these new peaks their exact molecular masses. As a result, the standard might be possible glucoside derivatives of the respective meclotetracycline at tR 3.54 min showed a distinct spectrum tetracycline analog. of m/z+ 477.1063 for which the calculated exact mass was

The standard chlorotetracycline appeared at the retention 477.1065 Da for the molecular formula C22H22ClN2O8 in time (tR) of 3.21 min. However, the novel peak supposed to protonated form (Fig. 4C(iv)). The new peak in the UPLC at + be a glucoside derivative appeared at the lower tR of 2.35 min tR 2.51 min showed a spectrum at m/z 639.1589. It resembled (Fig. 4(i)). The maximum UV absorbance of each peak was the protonated monoglucose-conjugated derivative of determined. The λmax of standard chlorotetracycline was at meclotetracycline with a molecular formula of C28H32ClN2O13 230, 268, and 371 nm (Fig. 4A(ii)). However, it was shifted for which the calculated exact mass was 639.1593 Da to 241 and 270 nm for the peak with tR of 2.35 min (Fig. 4C(v)). (Fig. 4A(iii)). HR-QTOF ESI/MS analysis of the standard Similar to the aforementioned tetracycline analogs, the + peak (tR 3.21 min) showed a prominent spectrum of m/z reaction mixture of minotetracycline was also analyzed. In

479.1218 with a calculated mass of 479.1221 Da for the UPLC-PDA, the standard minotetracycline appeared at tR molecular formula C22H24ClN2O8 in the protonated state. 2.59 min while a new peak appeared at tR 2.19 min (Fig. 4D(i)).

The novel peak in the reaction mixture with tR 2.35min Both peaks had a similar pattern of UV absorbance maxima. + showed a prominent spectrum at m/z 641.1753 with a The standard showed prominent λmax at 221, 266, and calculated mass of 641.1749 Da for the molecular formula 354 nm. The new peak had slight-shifted λmax (at 260 and

C28H34ClN2O13 (Fig. 4A(iv-v)). This mass corresponded to the 339 nm) (Fig. 4D(ii-iii)). Mass spectrometry analysis of the molecular mass of the glucose-conjugated chlorotetracycline standard showed the prominent spectrum of m/z+ 458.1941, derivative. The relative conversion percentage of chloro- for which the molecular formula C23H28N3O7 had a calculated tetracycline to its glucoside was approximately 6%. mass of 458.1927 Da in protonated form. However, the new In the doxytetracycline reaction mixture, the doxytetracycline peak showed the increased mass of m/z+ 620.2452, which standard appeared at tR 3.30 min. However, a new peak resembled the single sugar-conjugated minotetracycline, assumed to be a glucoside derivative of doxytetracycline for which the calculated exact mass was 620.2455 Da with a appeared at tR 2.32 min (Fig. 4B(i)). Both peaks were further molecular formula of C29H38N3O12 in protonated mode analyzed by spectroscopic and spectrometric analyses. (Fig. 4D(iv-v)). UPLC-PDA analysis of the rolitetracycline Spectroscopic analysis revealed a similar pattern of UV reaction mixture did not show any new peaks (data not absorbance maxima. The λmax for the standard doxytetracycline shown). Thus, the reaction mixture was not further analyzed was at 269 and 348 nm. For the presumed glucoside by HR-QTOF ESI/MS.

February 2018 ⎪ Vol. 28⎪ No. 2 302 Pandey et al.

Fig. 4. UPLC-PDA coupled with HR-QTOF ESI/MS analyses of reaction mixtures of four different tetracycline glucosides.

(A) Chlorotetracycline: (i) UPLC-PDA chromatogram of chlorotetracycline glucosylation reaction mixture; (ii) λmax of standard chlorotetracycline;

(iii) λmax of glucoside of chlorotetracycline; (iv) ESI/MS spectrum of chlorotetracycline standard; and (v) ESI/MS spectrum of glucoside of chlorotetracycline. (B) Doxytetracycline: (i) UPLC-PDA chromatogram of doxytetracycline glucosylation reaction mixzture; (ii) λmax of standard doxytetracycline; (iii) λmax of glucoside of doxytetracycline; (iv) ESI/MS spectrum of doxytetracycline standard; and (v) ESI/MS spectrum of glucoside of doxytetracycline. (C) Meclotetracycline: (i) UPLC-PDA chromatogram of meclotetracycline glucosylation reaction mixture; (ii) λmax of standard meclotetracycline; (iii) λmax of glucoside of meclotetracycline; (iv) ESI/MS spectrum of meclotetracycline standard; and (v) ESI/MS spectrum of glucoside of meclotetracycline. (D) Minotetracycline: (i) UPLC-PDA chromatogram of minotetracycline glucosylation reaction mixture; (ii) λmax of standard minotetracycline; (iii) λmax of glucoside of minotetracycline; (iv) ESI/MS spectrum of minotetracycline standard; and (v) ESI/MS spectrum of glucoside of minotetracycline.

J. Microbiol. Biotechnol. Glycosylation of Natural Products 303

The conversion of substrates to products was relatively clue that the enzyme has low preferences toward tetracycline very small. Glucoside derivatives of doxytetracycline and molecules. Attachment of a bulky group in rolitetracycline meclotetracycline were produced in trace amount. In might have prevented the molecule from reaching the comparison with the other substrates, chlorotetracycline catalytic cleft of GT. The results of this research could conversion to its glucoside was slightly higher. However, become a basis to explore the possibility of synthesizing such quantity was not enough for structural analysis or glucosylated derivatives of different tetracyclines using the biological activity determination. Thus, we did not scale up GT enzyme. the reaction mixture of the tetracycline reactions for purification of the glucoside derivatives. Hence, further Acknowledgments experiments for the complete structure elucidation and biological activities of the modified compounds were not This work was supported by a grant from the Next- performed. Generation BioGreen 21 Program (SSAC PJ01111901), Rural Since the YjiC enzyme has been found to be very non- Development Administration, Republic of Korea. regiospecific for glucosylation reaction with small molecules [23, 27], the prediction for the exact position of glucose Conflict of Interest conjugation to tetracycline analogs is difficult. However, in the case of tetracycline analogs, YjiC produced a single The authors have no financial conflicts of interest to product for all molecules used in this study except for declare. rolitetracycline, which was not accepted any longer for the reaction. The possible positions of glucose conjugation in References tetracycline might be the OH-groups of all four rings (A- D). However, OH-groups of the B and C rings could be 1. Chopra I, Roberts M. 2001. Tetracycline antibiotics: mode of engaged in intramolecular H-bonding with adjacent keto action, applications, molecular biology, and epidemiology of functional groups. In the A-ring of tetracyclines, the OH- bacterial resistance. Microbiol. Mol. Biol. Rev. 65: 232-260. group at the third position could be possibly hindered for 2. Darken MA, Berenson H, Shirk RJ, Sjolander NO. 1960. glucosylation because of bulky methyl groups attached at Production of tetracycline by Streptomyces aureofaciens in synthetic media. Appl. Microbiol. 8: 46-51. the C-4 position and conceivable engagement of 3-OH to 3. Testa RT, Petersen PJ, Jacobus NL, Sum PE, Lee VJ, Tally FP. form H-bonding with neighboring amide functional groups 1993. In vitro and in vivo antibacterial activities of the present at the C-2 position. Thus, the glucose is most likely glycylcyclines, a new class of semisynthetic tetracyclines. th to be conjugated at the 10 -OH group in the D-ring of the Antimicrob. Agents Chemother. 37: 2270-2277. tetracycline analogs used in this study. 4. Nelson ML, Ismail MY, McIntyre L, Bhatia B, Viski P, In the in vitro glucosylation reaction, several problems have Hawkins P, et al. 2003. Versatile and facile synthesis of been reported for low bioconversion reaction, including the diverse semisynthetic tetracycline derivatives via Pd-catalyzed following: (i) competitive inhibition of binding of UDP-α- reactions. J. Org. Chem. 68: 5838-5851. D-glucose to the donor substrate binding site by side 5. Liu F, Myres AG. 2016. Development of a platform for the products of the forward glucosylation reaction (i.e., UDP); discovery and practical synthesis of new tetracycline (ii) reversible nature of the GT enzyme, which can antibiotics. Curr. Opin. Chem. Biol. 32: 48-57. 6. Karami N, Nowrouzian F, Adlerberth I, Wold AE. 2006. hydrolyze the glucoside product to aglycone and UDP-α-D- Tetracycline resistance in Escherichia coli and persistence in glucose; (iii) low enzyme catalytic properties; and (iv) low the infantile colonic microbiota. Antimicrob. Agents Chemother. enzyme preference toward acceptor substrates. It has been 50: 156-161. found that YjiC has high catalytic conversion efficiency in 7. Li W, Atkinson GC, Thakor NS, Allas U, Lu CC, Chan KY, the forward direction with different acceptor substrates et al. 2013. Mechanism of tetracycline resistance by ribosomal [23, 27]. However, it also exhibits a reversible glucosylation protection protein Tet(O). Nat. Commun. 4: 1477. nature [24]. Thus, we applied the UDP-α-D-glucose recycling 8. Pandey RP, Parajuli P, Koffas MA, Sohng JK. 2016. system for two main purposes: (i) to prevent competitive Microbial production of natural and non-natural flavonoids: inhibition of binding of UDP-α-D-glucose by recycling pathway engineering, directed evolution and systems/ UDP, and (ii) to prevent a reverse reaction. However, the synthetic biology. Biotechnol. Adv. 34: 634-662. conversion of tetracycline analogs to their respective 9. Zhang J, Ponomareva LV, Marchillo K, Zhou M, Andes DR, glucosides has been found to be very low. This gives the Thorson JS. 2013. Synthesis and antibacterial activity of

February 2018 ⎪ Vol. 28⎪ No. 2 304 Pandey et al.

doxycycline neoglycosides. J. Nat. Prod. 76: 1627-1636. Kirsch DR, et al. 1992. Dactylocyclines, novel tetracycline 10. Kim SJ, Cegelski L, Stueber D, Singh M, Dietrich E, Tanaka KS, derivatives produced by a Dactylosporangium sp. I. Taxonomy, et al. 2008. Oritavancin exhibits dual mode of action to production, isolation and biological activity. J. Antibiot. (Tokyo) inhibit cell-wall biosynthesis in Staphylococcus aureus. J. Mol. 45: 1892-1898. Biol. 377: 281-293. 20. Horiguchi T, Hayashi K, Tsubotani S, Iinuma S, Harada S, 11. Patti GJ, Kim SJ, Yu TY, Dietrich E, Tanaka KS, Parr TR Jr, Tanida S. 1994. New naphthacenecarboxamide antibiotics, et al. 2009. Vancomycin and oritavancin have different modes TAN-1518 A and B, have inhibitory activity against of action in Enterococcus faecium. J. Mol. Biol. 392: 1178-1191. mammalian DNA topoisomerase I. J. Antibiot. (Tokyo) 47: 12. Brade KD, Rybak JM, Rybak MJ. 2016. Oritavancin: a new 545-556. lipoglycopeptide antibiotic in the treatment of gram-positive 21. Pickens LB, Kim W, Wang P, Zhou H, Watanabe K, Gomi S, infections. Infect. Dis. Ther. 5: 1-15. Tang YJ. 2009. Biochemical analysis of the biosynthetic 13. Marcone GL, Marinelli F. 2014. Glycopeptides: an old but pathway of an anticancer tetracycline SF2575. J. Am. Chem. up-to-date successful antibiotic class, pp. 85-107. In Marinelli Soc. 131: 17677-17689. F, Genilloud O (eds.). Antimicrobials. Springer-Verlag, 22. Chu LL, Pandey RP, Shin JY, Jung HJ, Sohng JK. 2016. Berlin-Heidelberg, Germany. Synthetic analog of anticancer drug daunorubicin from 14. Lairson LL, Henrissat B, Davies GJ, Withers SG. 2008. daunorubicinone using one-pot enzymatic UDP-recycling Glycosyltransferases: structures, functions, and mechanisms. glycosylation. J. Mol. Catal. B Enzym. 124: 1-10. Annu. Rev. Biochem. 77: 521-555. 23. Pandey RP, Gurung RB, Parajuli P, Koirala N, Tuoi LT, 15. Liang DM, Liu JH, Wu H, Wang BB, Zhu HJ, Qiao JJ. 2015. Sohng JK. 2014. Assessing acceptor substrate promiscuity of Glycosyltransferases: mechanisms and applications in natural YjiC-mediated glycosylation toward flavonoids. Carbohydr. product development. Chem. Soc. Rev. 44: 8350-8374. Res. 393: 26-31. 16. Elshahawi SI, Shaaban KA, Kharel MK, Thorson JS. 2015. A 24. Pandey RP, Li TF, Kim EH, Yamaguchi T, Park YI, Kim JS, comprehensive review of glycosylated bacterial natural et al. 2013. Enzymatic synthesis of novel phloretin glucosides. products. Chem. Soc. Rev. 44: 7591-7697. Appl. Environ. Microbiol. 79: 3516-3521. 17. Devasthale PV, Mitscher LA, Telikepalli H, Vander Velde D, 25. Parajuli P, Pandey RP, Koirala N, Yoon YJ, Kim BG, Sohng JK. Zou JY, Ax HA, et al. 1992. Dactylocyclines, novel tetracycline 2014. Enzymatic synthesis of epothilone A glycosides. AMB derivatives produced by a Dactylosporangium sp. III. Absolute Express 4: 31. stereochemistry of the dactylocyclines. J. Antibiot. (Tokyo) 45: 26. Parajuli P, Pandey RP, Pokhrel AR, Ghimire GP, Sohng JK. 1907-1913. 2014. Enzymatic glycosylation of the topical antibiotic 18. Tymiak AA, Ax HA, Bolgar MS, Kahle AD, Porubcan MA, mupirocin. Glycoconj. J. 31: 563-572. Andersen NH. 1992. Dactylocyclines, novel tetracycline 27. Pandey RP, Parajuli P, Shin JY, Lee J, Lee S, Hong YS, et al. derivatives produced by a Dactylosporangium sp. II. Structure 2014. Enzymatic biosynthesis of novel resveratrol glucoside elucidation. J. Antibiot. (Tokyo) 45: 1899-1906. and glycoside derivatives. Appl. Environ. Microbiol. 80: 7235- 19. Wells JS, O’Sullivan J, Aklonis C, Ax HA, Tymiak AA, 7243.

J. Microbiol. Biotechnol.