Amino Mannitol Dehydrogenases on the Azasugar Biosynthetic Pathway

Home , Iditol, Mannitol dehydrogenase

Send Orders for Reprints to [email protected] 10 Protein & Peptide Letters, 2014, 21, 10-14 Medium-Chain Dehydrogenases with New Specificity: Amino Mannitol Dehydrogenases on the Azasugar Biosynthetic Pathway

Yanbin Wu, Jeffrey Arciola, and Nicole Horenstein*

Department of Chemistry, University of Florida, Gainesville Florida, 32611-7200, USA

Abstract: Azasugar biosynthesis involves a key dehydrogenase that oxidizes 2-amino-2-deoxy-D-mannitol to the 6-oxo compound. The genes encoding homologous NAD-dependent dehydrogenases from Bacillus amyloliquefaciens FZB42, B. atrophaeus 1942, and Paenibacillus polymyxa SC2 were codon-optimized and expressed in BL21(DE3) Escherichia coli. Relative to the two Bacillus enzymes, the enzyme from P. polymyxa proved to have superior catalytic properties with a Vmax of 0.095 ± 0.002 mol/min/mg, 59-fold higher than the B. amyloliquefaciens enzyme. The preferred substrate is 2- amino-2-deoxy-D-mannitol, though mannitol is accepted as a poor substrate at 3% of the relative rate. Simple amino alcohols were also accepted as substrates at lower rates. Sequence alignment suggested D283 was involved in the enzyme’s specificity for aminopolyols. Point mutant D283N lost its amino specificity, accepting mannitol at 45% the rate observed for 2-amino-2-deoxy-D-mannitol. These results provide the first characterization of this class of zinc-dependent medium chain dehydrogenases that utilize aminopolyol substrates. Keywords: Aminopolyol, azasugar, biosynthesis, dehydrogenase, mannojirimycin, nojirimycin.

INTRODUCTION are sufficient to convert fructose-6-phosphate into mannojirimycin [9]. We proposed that the gutB1 gene product was Azasugars such as the nojirimycins [1] are natural prod- responsible for the turnover of 2-amino-2-deoxy-D-mannitol ucts that are analogs of monosaccharides that feature a nitro- (2AM) into mannojirimycin as shown in Fig. 2. Feeding the gen in the ring rather than oxygen (Fig. 1). In Nature they are B. amyloliquefaciens gabT1 knockout with 2AM restored produced by various Bacillus, Streptomyces and plant spe- azasugar production [9]. cies [2-4]. Azasugars have served as the inspiration for the synthesis of many different glycosidase inhibitors [5] and Sequence analyses indicate that GutB1 is a zinc- have in recent times also enjoyed applications as chemical dependent NAD(P)-dependent alcohol dehydrogenase, simi- chaperones for assisting in the folding and stabilization of lar to sorbitol dehydrogenase, and is a member of the me- mutant enzymes responsible for lysosomal storage diseases dium-chain dehydrogenase/reductase (MDR) superfamily [6]. Examples include Miglustat, N-butyl-1-deoxynojiri- [11]. As it was known that other species with sequenced ge- mycin, for type I Gaucher’s disease [7] and Miglitol, (N-2- nomes were deoxynojirimycin producers, we sought to iden- hydroxyethyl)-1-deoxynojirimycin for modulation of post- tify and then compare other homologues possessing the prandial bloodsugar in treatment of type II diabetes [8]. unique dehydrogenase activity demonstrated by the B. amyloliquefaciens enzyme. A protein BLAST analysis of the Although azasugars have been known for quite some three-gene azasugar biosynthetic signature found in B. amy- time, the enzymatic machinery responsible for their synthesis loliquefaciens was conducted and identified similar ORFs in has only recently become the topic of experimental inquiry. Bacillus and related species. When we included more distant Further, the identity of enzymes in the entire pathway has not hits we identified that Paenibacillus polymyxa SC2 has five yet been determined nor have any been characterized indi- ORFs that appear to include the three functions we identified vidually. In this Letter we wish to report the expression and in B. amyloliquefaciens, as well as an additional two functional characterization of a key dehydrogenase found in (PPSC2_c2587 and PPSC2_c2588) that code for putative azasugar biosynthetic clusters. Recent investigation of 1- mannitol dehydrogenase and mannonate dehydratase activi- deoxynojirimycin biosynthesis has led to the identification of ties that may also be part of the pathway [12]. In the work Bacillus amyloliquefaciens genes gabT1, yktC1 and gutB1 as we describe here, we characterize the GutB1 dehydrogenase part of the overall azasugar biosynthetic pathway [9,10]. from B. amyloliquefaciens FZB42, the closely related en- These three genes respectively encode for transaminase, zyme from B. atrophaeus 1942 [13] (86% identity) and the phosphatase and dehydrogenase activity that we have shown more distantly related homolog from Paenibacillus polymyxa SC2 (33% identity). We show that the so-called GutB1 en- *Address correspondence to this author at the Department of Chemistry, zymes prefer substrates bearing an amino group, distinguish- University of Florida, Gainesville Florida, 32611-7200, USA; Tel: 01-(352)- ing themselves from previously characterized polyol dehy- 392-9859; E-mail: [email protected] drogenases. Wu, Y. Horenstein, N. Unpublished observations for recombinant YktC1 from B. amyloliquefaciens.

Figure 1. Structures of nojirimycin (NJ), mannojirimycin (MJ) and deoxynojirimycin (DNJ). Both NJ and DNJ have the gluco configuration and MJ has the manno configuration.

HO HO HO HO HO HO NH OH GutB1 HO HO O

HO OH + HO NH HO NH2 NAD NADH 2 HO MJ

Figure 2. Conversion of 2-amino-2-deoxy-D-mannitol to mannojirimycin via GutB1 catalyzed oxidation. After oxidation, the acyclic ami- noaldehyde is able to spontaneously cyclize to mannojirimycin (MJ).

linker sequence as described above for the B. amyloliquefa- MATERIALS AND METHODS ciens gutB1 construct. General. The gutB1 genes were synthesized and codon P. polymyxa D283N mutation. This point mutation was optimized by GenScript. Escherichia coli DH5alpha and E. made using the Q5 mutagenesis kit from New England Bio- coli BL21 (DE3) strains were obtained from Invitrogen and labs. The forward mutagenic primer sequence was 5’- Novagen respectively. The pET30a expression vector was GAAGATTATCGGCTCAATTAACTCGCTGGGTACC- obtained from Novagen. Restriction endonucleases, T4 DNA TTCTC-3’. The reverse primer sequence was 5’-AGGCTG ligase and thermostable polymerases were purchased from CGATCAACCACTTCTTTCGGATTA-3’. The PCR ther- New England Biolabs. Isopropyl- -D-thiogalactopyranoside mocycle was: 98 °C, 30 s; 25x (98 °C, 10 s; 66 °C, 30 s; 72 (IPTG), kanamycin and protein MW standards were pur- °C, 198 s), 72 °C, 120 s; followed by 4 °C, indefinite. chased from Fisher Scientific. Oligonucleotides were purchased from Integrated DNA Technologies. DNA sequenc- Protein expression and purification. The pET30a- ing was performed at the University of Florida Sanger Se- gutB1constructs were transformed into BL21 (DE3) E. coli. quencing core facility. Single colonies were selected to inoculate 5 mL overnight starter cultures (Luria-Bertani medium containing 50 g/mL Subcloning of the B. amyloliquefaciens gutB1 gene kanamycin). Each starter culture was used to inoculate 500 into pET30a. The codon-optimized version of the gutB1 ml of LB/K medium and the resulting cultures were grown gene was originally prepared in pETBlue2. The pETBlue2- for 2-4 hours at 37 °C with constant shaking at 225 rpm. gutB1 vector served as a template for PCR with the forward Expression of GutB1 was induced by the addition of 0.5 mM primer 5’-GAGCCATGGGGATGAAAGCTCTGGTGTG IPTG when an OD600 of 0.4 to 0.8 was reached. The cultures GAC-3’ and reverse primer 5’-AGAGCTCGAGTTACAGC were grown at 18 °C for 18 h with constant shaking at 225 AGTTTCGGGTCGCTAAC-3’. The underlined sequences rpm. The cells were harvested by centrifugation at 4 °C in a correspond to the NcoI and XhoI restriction sites, in the for- GS3 rotor at 5000 rpm for 20 minutes. The cell pellets were ward and reverse primers, respectively. The PCR product each resuspended in 20 mL of lysis buffer (20 mM Tris-HCl, was cloned into pET30a to generate the reconstruct plasmid pH=7.9, 250 mM NaCl, 5 mM imidazole). After lysis by pET30a-opt-gutB1, and was verified by sequencing. This French press and sonication, the lysate was centrifuged at construct codes for an N-terminal Met(His)6 tag fused to a 14,000 rpm for 30 min at 4 °C in an SS34 rotor. The clarified linker peptide with the following sequence upstream of the lysate was applied to a Ni-IMAC resin (Qiagen), and the native start Met: SSGLVPRGSGMKETAAAKFERQHMDS column washed with a step gradient of 10, 50, and 100 mM PDLGTDDDDK. imidazole. GutB1 was eluted with 250 mM imidazole- Expression constructs for P. polymyxa SC2 and B. containing buffer (20 mM Tris-HCl, pH=7.9, 250 mM NaCl, atrophaeus gutB1 genes. The P. polymyxa (locus tag PPSC2 250 mM imidazole). The eluted enzyme was dialyzed at 4°C _c2584) and B. atrophaeus (locus tag BATR1942_19425) against 20 mM Tris buffer (50 mM NaCl, pH=7.0). gutB1 homologues were codon-optimized (Genscript) in Protein analyses. Protein was assayed by the Bradford pUC57 with NcoI and XhoI sites flanking the start and stop method using bovine serum albumin as a standard [14]. codons. The gutB1 genes were subcloned into the NcoI and SDS-PAGE was performed at room temperature using 12% XhoI sites of the pET30 vector for expression. The recombi- acrylamide resolving gel and 5% acrylamide stacking gel. nant construct was confirmed by sequencing. These con- Metal content was determined by inductively coupled structs afforded coded for the same N-terminal His tag and 12 Protein & Peptide Letters, 2014, Vol. 21, No. 1 Wu et al. plasma-mass spectrometry (ICP-MS) at the Chemical Analy- RESULTS AND DISCUSSION sis Laboratory, University of Georgia, Athens. Samples were The three GutB1 dehydrogenase genes from B. amyloliq- exchanged into Chelex-treated 20 mM Tris HCl pH 7.0 uefaciens FZB42 (Bam), B. atrophaeus 1942 (Bat), and P. buffer. Approximately 2.5 – 3 mg of protein in metal-free polymyxa SC2 (Ppo) were obtained by synthesis in codon buffer was further incubated with Chelex for 2 h at 4 °C to remove weakly bound metals. A parallel blank without pro- optimized form. They readily expressed in E. coli as N- terminal His-tagged fusion proteins in respective purified tein was prepared in the same way. The protein concentra- yields of 80, 36, and 60 mg per liter of culture after Ni- tion was determined by careful Bradford assay and the solu- column chromatography. Electrophoretic analyses indicated tion was brought to a final volume in 1% HNO for analyses. 3 that we obtained near-homogeneous preparations of enzymes Synthesis of 2-amino-2-deoxy-D-mannitol (2AM) This at the anticipated approximate molecular weights. (Figures substrate was synthesized according to Liu et al [15], with S1-S3, supplemental information.) A MALDI-TOF analysis some modifications. The reaction involved reduction of D- of the Ppo enzyme yielded 42836 for [M+H] versus 42832 mannosamine (300 mg) with NaBH4 (572 mg in 5mL of (predicted). Metal analysis of the Bam enzyme by ICP-MS 0.05M NaOH) for 1 hour at room temperature, at which time revealed a zinc/monomer stoichiometry of 2:1, consistent TLC (silica, methanol:chloroform:water = 7:2:1, v/v/v) indi- with other members of the zinc-dependent MDR superfamily cated completion. Acetic acid was carefully added to destroy often bearing two zincs per subunit. (Table S1, supplemental the excess sodium borohydride. Methanol was added to the information). quenched reaction mixture followed by rotary evaporation. We estimated steady-state kinetic parameters for the Bam This was repeated four more times. The residue was taken up dehydrogenase by holding one substrate at high levels while in water (2 mL), the pH was adjusted to 5-6 and the solution + varying the other.(Figure S6 Supplemental information) The was applied to a Dowex-50 (H ) column (0.9 x 7 cm). The + Km for 2AM was estimated to be 5.5 ± 1.6 mM, the Km for resin was washed with 0.5 M HCl to elute Na (flame test), + -1 then 2M HCl to elute the product 2AM. The product was NAD was 105 ± 37 M with a kcat of 0.07 min . This very slow turnover is consistent with results for GabT1 and detected by a positive test with ninhydrin. Concentration in 1 1 YktC1 enzymes from Bacillus amyloliquefaciens FZB42 vacuo yielded 228 mg of 2AM (76 % yield). H-NMR and 13 which are also quite slow [17]. We determined that Bam C-NMR indicated complete reduction to the desired 2AM. GutB1 can accept either NAD+ or NADP+, but does prefer (Supplemental information figures S4 and S5). NAD+ since velocities measured with 2 mM NADP were Enzymatic Assays. GutB1 enzymatic activity was moni- only 15 % of the rate observed with 2 mM NAD+(data not tored by following NAD+ reduction at 340 nm. Assay reac- shown). Initial examination of the Bat enzyme revealed it tions were carried out at 25 °C, in a 1.00 mL standard reac- had extremely low turnover like the Bam enzyme, a result tion mixture that contained 500 L 2X buffer (200 mM Tris, consistent with their sequence similarity. The Bat enzyme 50 mM NaCl, 5 M ZnCl2, pH = 8.57), and 50 lLBSA (20 was not characterized further. Interestingly the Ppo enzyme mg/ml) for reactions involving the Bam enzyme. For estima- exhibits significant sequence divergence from the Bam and tion of steady state kinetic parameters, one substrate was Bat homologs and we anticipated we might find different held well above Km while the other substrate was varied. For kinetics for it. Initial assays for the enzyme indicated it was + the Bam enzyme, the Km for NAD was estimated by holding considerably more active, and this was borne out in the ki- 2AM at 16 mM while NAD+ was varied between 0.05 – 4 netic analyses (Figure S7, supplemental information). The mM. The Km for 2AM was estimated by fixing the concen- Km for 2AM was estimated to be 1.7 ± 0.1 mM, the Km for + + -1 tration of NAD at 4 mM while varying 2AM between 0.16 NAD was 1.1 ± 0.1 mM and kcat was 4.1 min . This turn- – 53 mM. For the Ppo enzyme, 2AM was held at 53.3 mM over is approximately 59-fold higher than the Bam enzyme, while NAD+ was varied between 0.01 – 10 mM. The concen- rendering it of interest for its possible ability to accept alter- tration of NAD+ was fixed at 4 mM while varying 2AM be- nate substrates. To this end various polyol and amino-alcohol tween 0.053 – 10.7 mM. Significant substrate inhibition was substrates were screened against the Bam and Ppo enzymes observed when 2AM was assayed at 53 mM. All reactions for their activity relative to 2AM as presented in Table 1. were initiated by the addition of 400 g of enzyme and measured in triplicate. Data were acquired over 150 seconds The data suggest several noteworthy characteristics of this class of enzyme. Both enzymes accepted mannitol albeit (Ppo) or 300 seconds (Bam). For substrate specificity stud- at reduced rates. Given that both rejected sorbitol, this indi- ies, reactions were conducted in the same buffer system as cates that the manno configuration at C2 of the hexitol chain above. Test substrates were used at 10 mM concentration, + is preferred. Interestingly, mannose, which exists primarily the NAD concentration was held at 2 mM, and reactions as the - pyranose in solution, was not accepted as a sub- were initiated with 180 g of enzyme. Analysis of the D283N mutant kinetics used the standard reaction mixture strate by either enzyme. This result suggests that acyclic substrates are preferred, possibly indicating an extended binding described above with 1.0 mg/mL BSA, 2AM (10 mM) or + conformation. Fructose was not a substrate for the Bam en- mannitol (160 mM), NAD at 2 mM. Reactions were initi- zyme; and it was not tested versus the Ppo dehydrogenase. ated by addition of 100 g of either WT or D283N enzyme. Interestingly the simple amino alcohols 4-amino-1-butanol Bioinformatic sequence analyses. Comparative analyses and 6-amino-1-hexanol were found to be alternate substrates of protein amino acid sequences used pBLAST from the for the Ppo enzyme, but not the Bam enzyme. The rates were NCBI website. Alignments were determined by ClustalW 18% and 12% of that found for 2AM. Focusing on the more multiple-sequence alignment [16]. promiscuous Ppo enzyme, we found that ethanolamine was Medium-Chain Dehydrogenases with New Specificity Protein & Peptide Letters, 2014, Vol. 21, No. 1 13

Table 1. A comparison of relative velocities for different sub- high 10 mM) may fail to reveal activity with very weakly strates versus dehydrogenases from B. amyloliquefa- bound substrates, it appears that the enzyme from P. po- ciens and P. polymyxa. Velocities in each column are lymyxa is more tolerant of structural diversity than that from expressed as percentages relative to the velocity with B. amyloliquefaciens, and 2AM is the preferred substrate. 2AM as substrate. ND = not determined. Conditions This latter result drives home the point that 2AM is a key are provided in Materials and methods. intermediate in azasugar biosynthesis, as shown previously [9]. The slow turnover of the enzyme is understandable relative activity relative activity given its role in secondary metabolism, [19] and the lack of substrates strong evolutionary pressure to maintain the pathway as evi- (Bam) (Ppo) denced by the lack of azasugar production in B. subtilis de- fructose 0% ND spite its otherwise close relationship to the select members of the genus that are known azasugar producers. The polyol mannitol 9% 3% dehydrogenases (EC 1.1.1.14) as exemplified by “GutB” dehydrogenase from B. subtilis are so-named for their ability mannose 0% 0% to oxidize polyol substrates such as sorbitol, xylitol and L- sorbitol 0% 0% iditol [18]. Despite sequence similarity of the Bam and Ppo enzymes studied here to B. subtilis GutB, (47% and 46% 4-amino-1-butanol 0% 18% similarity at the protein level) they disdain uncharged polyol substrates, but rather prefer amino-substituted polyols. These 6-amino-1-hexanol 0% 12% enzymes are members of the broad super family of medium serinol ND 0% chain dehydrogenases, and to the best of our knowledge, as aminopolyol dehydrogenases, they are a unique addition to L-threonine ND 3% the MDR superfamily. Based on the specificity we have ob- glucosamine ND 0% served we term these enzymes as amino-mannitol dehydrogenases (AMDHs). ethanolamine ND 0% Clues as to the origin of the specificity for amino polyol phenylpropanolamine ND 0% substrates were found in sequence analysis and the structure of sheep liver sorbitol dehydrogenase [20]. dimethylethanolamine ND 0% Figure 3 presents a ribbon diagram for sorbitol dehydrogenase with a glycerol bound in the active site as a binary not a substrate, indicating that a carbon chain length of two complex. Arginine 297 is in hydrogen bonding contact with was too short. L-threonine was a very poor substrate distin- the bound glycerol. Sequence alignments (Figure 3, inset) of guishing this enzyme from the threonine dehydrogenases that sheep and Bacillus subtilis sorbitol dehydrogenases versus are also members of the MDR superfamily. We also tested bacterial dehydrogenases from known or putative azasugar glucosamine, phenylpropanolamine, and dimethyl ethanola- producers reveal that the position homologous to 297 is sub- mine versus the Ppo enzyme and found that none of these stituted with either D or E for azasugar producers versus R. substrates were accepted. Keeping in mind the caveat that Possibly, D or E residues provide charge recognition with screening at fixed substrate concentration (albeit a relatively the positively charged ammonium group of an amino polyol

Figure 3. Ribbon diagram for sorbitol dehydrogenase from sheep liver (PDB ID 3EQ3). Arginine 297 is represented with space-filling atoms as is bound glycerol. The inset presents CLUSTAL sequence alignment of putative aminopolyol dehydrogenases from azasugar producers and sorbitol dehydrogenases from sheep liver and Bacillus subtilis. The black square indicates position 297 for sheep liver sorbitol dehydrogenase. 14 Protein & Peptide Letters, 2014, Vol. 21, No. 1 Wu et al. substrate. We speculated that replacement of D283 in the P. [5] Pearson, M.; Mathe-Allainmat, M.; Fargeas, V.; Lebreton, J. Re- polymyxa dehydrogenase to a neutral residue might disfavor cent advances in the total synthesis of piperidine azasugars. Eur. J. Org. Chem., 2005, 2159-2191. aminopolyol specificity and indeed this proved to be the case [6] Suzuki, Y.; Ogawa, S.; Sakakibara, Y. Chaperone therapy for neu- for the D283N mutant. Under saturating conditions, we ronopathic lysosomal diseases: competitive inhibitors as chemical compared the ratio of velocities for mannitol versus 2AM chaperones for enhancement of mutant enzyme activities. Persp. Med. Chem., 2009, 3, 7-19. with wild-type enzyme (rel = 0.056) and with the D283N mutant ( = 0.43). The preference for the charged 2AM [7] Ficicioglu, C. Review of miglustat for clinical management in rel Gaucher disease type 1. Ther. Clin. Risk Manag., 2008, 4, 425-431. substrate was dramatically decreased by the mutation, sup- [8] Scott, L.J.; Spencer, C.M. Miglitol: a review of its therapeutic porting the proposed role for D283 in aminopolyol recogni- potential in type 2 diabetes mellitus. Drugs, 2000, 59, 521-549. tion. Note that the change in substrate preference largely [9] Clark, L.F.; Johnson, J.V.; Horenstein, N.A. Identification of a arises from a decrease in the turnover of 2AM, which gene cluster that initiates azasugar biosynthesis in Bacillus amyloliquefaciens. ChemBiochem, 2011, 12, 2147-2150. dropped by a factor of 11 for the mutant, whereas turnover of [10] Kang, K.D.; Cho, Y.S.; Song, J.H.; Park, Y.S.; Le,e J.Y.; Hwang, mannitol only decreased by a factor of 1.4 for the D283N K.Y.; Rhee, S.K.; Chung, J.H.; Kwon, O.; Seong, S.I. Identification mutant. of the genes involved in 1-deoxynojirimycin synthesis in Bacillus subtilis MORI 3K-85. J. Microbiol., 2011, 49, 431-440. [11] Chen, X.H.; Koumoutsi, A.; Scholz, R.; Eisenreich, A.; Schneider, CONCLUSION K.; Heinemeyer, I.; Morgenstern, B.; Voss, B.; Hess, W.R.; Reva, The NAD-dependent dehydrogenases that appear in O.; Junge, H.; Voigt, B.; Jungblut, P.R.; Vater, J.; Süssmuth, R.; Liesegang, H.; Strittmatter, A.; Gottschalk, G.; Borriss, R. Com- azasugar biosynthetic clusters are members of the medium parative analysis of the complete genome sequence of the plant chain reductase/dehydrogenase family that have specificity growth-promoting bacterium Bacillus amyloliquefaciens FZB42. for 2-amino-2-deoxy-D-mannitol, but in some cases are suf- Nat. Biotechnol ., 2007, 1007-1014. ficiently promiscuous to accept simple amino alcohols as [12] Ma, M.; Wang, C.; Ding, Y.; Li, L.; Shen, D.; Jiang, X.; Guan, D.; Cao, F.; Chen, H.; Feng, R.; Wang, X.; Ge, Y.; Yao, L.; Bing, X.; substrate. In codon optimized form these enzymes are well- Yang.; X.; Li. J.; Du, B. Complete genome sequence of Paenibacil- expressed in E. coli, and may be a useful platform for evolu- lus polymyxa SC2, a strain of plant growth-promoting rhizobacte- tion of redox catalysts that accept synthetically interesting rium with broad-spectrum antimicrobial activity, J. Bacteriol., precursors. Continuing studies are aimed at further defining 2011, 193, 311-312. the basis for molecular recognition within the active site via [13] Gibbons, H.S.; Broomall, S.M.; McNew, L.A.; Daligault, H.; Chapman, C.; Bruce, D.; Karavis, M.; Krepps, M.; McGregor, structural studies. P.A.; Hong, C.; Park, K.H.; Akmal, A.; Feldman, A.; Lin, J.S.; Chang, W.E.; Higgs, B.W.; Demirev, P., Lindquist, J.; Liem, A.; CONFLICT OF INTEREST Fochler, E.; Read, T.D.; Tapia, R.; Johnson, S.; Bishop-Lilly, K.A.; Detter, C.; Han, C.; Sozhamannan, S.; Rosenzweig, C.N.; Skow- The authors confirm that this article content has no con- ronski, E.W.; Genomic signatures of strain selection and enhance- flicts of interest. ment in Bacillus atrophaeus var. globigii, a historical biowarfare simulant. PLoS ONE, 2011, 6, E17836. [14] Bradford, M.M. Rapid and sensitive method for quantitation of ACKNOWLEDGEMENT microgram quantities of protein utilizing principle of protein-dye binding. Anal. Biochem., 1976, 72, 248-254. We are grateful for the support of the National Science [15] Liu, T.Y.; Gotschli, E.; Dunne, F.T.; Jonssen, E.K. Studies on Foundation (MCB-1020940). meningococcal polysaccharides .2. Composition and chemical properties of group-B and group-C polysaccharide. J. Biol. Chem., SUPPLEMENTARY MATERIAL 1971, 246, 4703-4712. [16] Chenna, R.; Sugawara, H.; Koike, T.; Lopez, R.; Gibson, T.J.; Supplementary material is available on the publishers Higgins, D.J.; Thompson, J.D. Multiple sequence alignment with Web site along with the published article. the CLUSTAL series of programs. Nuc. Acids Res., 2003, 31, 3497-3500. REFERENCES [17] Clark, L.F. Biosynthesis of azasugars in Bacillus amyloliquefaciens. Doctoral Dissertation, U. Florida, 2012. [1] Inouye, S.; Tsuruoka, T.; Nida, T. The structure of nojirimycin, a [18] Ng, K.; Ye, R.; Wu, X.-C.; Wong, S.-L. Sorbitol dehydrogenase piperidinose sugar antibiotic. J. Antibiot., 1966, 19, 288-292. from Bacillus subtilis. Purification, characterization, and gene clon- [2] Asano, M. Naturally occurring iminosugars and related com- ing. J. Biol. Chem., 1992, 267, 24989-24994. pounds: Structure, distribution, and biological activity. Curr. Top. [19] Bar-Even, A.; Noor, E.; Savir, Y.; Liebermeister, W.; Davidi, D.; Med. Chem., 2003, 3, 471-484. Tawfik, D.S.; Milo, R. The moderately efficient enzyme: evolu- [3] Watson, A.A.; Fleet, G.W.J.; Asano, N.; Molyneux, R.J.; Nash, tionary and physicochemical trends shaping enzyme parameters. R.J. Polyhydroxylated alkaloids – natural occurrence and therapeu- Biochemistry, 2011, 50, 4402-4410. tic applications. Phytochemistry, 2001, 56, 265-295. [20] Yennawar, H.; Moller, M.; Gillilan, R.; Yennawar, N. X-ray crystal [4] Asano, N.; Oseki, K.; Tomiyoka, E.; Kizu, H.; Matsui, K. N- structure and small-angle X-ray scattering of sheep liver sorbitol Containing sugars from Morus alba and their glycosidase inhibi- dehydrogenase. Acta Crystallogr. Sect. D, 2011, 67, 440-446. tory activities. Carb. Res., 1994, 259, 243-255.

Received: May 31, 2013 Revised: August 4, 2013 Accepted: August 4, 2013