LmbE proteins from Bacillus cereus are de-N-acetylases with broad substrate specificity and are highly similar to proteins in Bacillus anthracis Alexandra Deli 1 , Dimitrios Koutsioulis 2 , Vasiliki E. Fadouloglou 2, *, Panagiota Spiliotopoulou 1 , Stavroula Balomenou 1 , Sofia Arnaouteli 1 , Maria Tzanodaskalaki 2 , Konstantinos Mavromatis 3 , Michalis Kokkinidis 1,2 and Vassilis Bouriotis 1,2 1 Department of Biology, Enzyme Biotechnology Group, University of Crete, Greece 2 Institute of Molecular Biology and Biotechnology, Heraklion, Crete, Greece 3 Department of Energy ⁄ Joint Genome Institute, Genome Biology Program, Walnut Creek, CA, USA Keywords Bacillus anthracis; de-N-acetylase; glucosamine; LmbE; mutational analysis Correspondence V. Bouriotis, Department of Biology, Enzyme Biotechnology Group, University of Crete, PO Box 2208, Vasilika Vouton 714 09, Heraklion, Crete, Greece Fax: +30 2810 394055 Tel: +30 2810 394375 E-mail: bouriotis@imbb.forth.gr *Present address Department of Biochemistry, University of Cambridge, UK (Received 22 October 2009, revised 15 March 2010, accepted 20 April 2010) doi:10.1111/j.1742-4658.2010.07691.x The genomes of Bacillus cereus and its closest relative Bacillus anthracis each contain two LmbE protein family homologs: BC1534 (BA1557) and BC3461 (BA3524). Only a few members of this family have been biochemi- cally characterized including N-acetylglucosaminylphosphatidyl inositol (GlcNAc-PI), 1-d-myo-inosityl-2-acetamido-2-deoxy-a-d-glucopyranoside (GlcNAc-Ins), N,N¢-diacetylchitobiose (GlcNAc 2 ) and lipoglycopeptide antibiotic de-N-acetylases. All these enzymes share a common feature in that they de-N-acetylate the N-acetyl-d-glucosamine (GlcNAc) moiety of their substrates. The bc1534 gene has previously been cloned and expressed in Escherichia coli. The recombinant enzyme was purified and its 3D struc- ture determined. In this study, the bc3461 gene from B. cereus ATCC14579 was cloned and expressed in E. coli. The recombinant enzymes BC1534 (EC 3.5.1 ) and BC3461 were biochemically characterized. The enzymes have different molecular masses, pH and temperature optima and broad sub- strate specificity, de-N-acetylating GlcNAc and N-acetylchito-oligomers (GlcNAc 2 , GlcNAc 3 and GlcNAc 4 ), as well as GlcNAc-1P, N-acetyl-d-glu- cosamine-1 phosphate; GlcNAc-6P, N-acetyl-d-glucosamine-6 phosphate; GalNAc, N-acetyl-d-galactosamine; ManNAc, N-acetyl-d-mannosamine; UDP-GlcNAc, uridine 5¢-diphosphate N-acetyl-d-glucosamine. However, the enzymes were not active on radiolabeled glycol chitin, peptidoglycan from B. cereus, N-acetyl-d-glucosaminyl-(b-1,4)-N-acetylmuramyl-l-alanyl- d-isoglutamine (GMDP) or N-acetyl-d-GlcN-Na1-6-d-myo-inositol-1-HPO 4 - octadecyl (GlcNAc-I-P-C 18 ). Kinetic analysis of the activity of BC1534 and BC3461 on GlcNAc and GlcNAc 2 revealed that GlcNAc 2 is the favored substrate for both native enzymes. Based on the recently determined crystal structure of BC1534, a mutational analysis identified functional key resi- dues, highlighting their importance for the catalytic mechanism and the sub- strate specificity of the enzyme. The catalytic efficiencies of BC1534 variants were significantly decreased compared to the native enzyme. An alignment- based tree places both de-N-acetylases in functional categories that are dif- ferent from those of other LmbE proteins. Abbreviations GalNAc, N-acetyl- D-galactosamine; GlcNAc, N-acetyl-D-glucosamine; GlcNAc 2 , N,N ¢-diacetylchitobiose; GlcNAc-1P, N-acetyl-D-glucosamine-1 phosphate; GlcNAc-6P, N-acetyl- D-glucosamine-6 phosphate; GlcNAc-Ins, 1-D-myo-inosityl-2-acetamido-2-deoxy-a-D-glucopyranoside; GlcNAc-PI, N-acetylglucosaminylphosphatidyl inositol; GlcNAc-I-P-C 18 , N-acetyl-D-GlcN-a1-6-D-myo-inositol-1-HPO 4 -octadecyl; GMDP, N-acetyl- D-glucosaminyl-(b-1,4)-N-acetylmuramyl-L-alanyl-D-isoglutamine; ManNAc, N-acetyl-D-mannosamine; UDP-GlcNAc, uridine 5¢-diphosphate N-acetyl- D-glucosamine. 2740 FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS Introduction Bacillus cereus, an opportunistic pathogen that causes food poisoning, and Bacillus anthracis, the endospore- forming bacterium that causes inhalational anthrax, share a large number of homologous genes, as demon- strated by recent genome sequencing and comparative analysis [1,2]. Given the laboratory safety precautions necessary for working with highly infectious agents and the recent concerns regarding use of B. anthracis as a potential bioweapon (class A agent, Centers for Disease Control), the B. cereus enzymes are useful models for studying the corresponding proteins of B. anthracis. The ERGO light database of the B. cereus ATCC14579 genome (http://www.ergo-light.com/) reveals the presence of two LmbE protein family homologs, BC1534 and BC3461, which share 21% identity [1]. Furthermore, they share 96% and 95% identity with their homologs BA1557 and BA3524, respectively, from B. anthracis. The LmbE protein family (Fig. 1) includes N-acetylglucosaminylphosphat- idyl inositol (GlcNAc-PI) de-N-acetylases from mammals [3], yeast [4] and protozoa [5], 1-d-myo-inosi- tyl-2-acetamido-2-deoxy-a-d-glucopynanoside (GlcNAc- Ins) de-N-acetylase from Mycobacterium tuberculosis (MshB) (EC 3.5.1.89) [6], N,N¢-diacetylchitobiose (GlcNAc 2 ) de-N-acetylase from the archaeon Thermo- coccus kodakaraensis KOD1 (Tk-Dac) (EC 3.5.1 ) [7] and antibiotic de-N-acetylases from Actinoplanes teich- omyceticus (Orf2) [8] and Bacillus circulans (BtrD) [9]. The most important members of the LmbE protein family, together with the structures of their substrates, are shown in Fig. 2. The crystal structure of BC1534 (previously reported as BcZBP) has been determined at 1.8 A ˚ resolution (Fig. 3A) [10]. The structures of three other LmbE protein family members have been similarly deter- mined, namely TT1542 from Thermus thermophilus [11], MshB from Mycobacterium tuberculosis [12] and Orf2 from Actinoplanes teichomyceticus [8]. The N-terminal part of the 234 amino acid BC1534 protein adopts a Rossmann fold, and the C-terminal part con- sists of two b-strands and two a-helices. In the crystal, the protein forms a compact hexamer (Fig. 3A), in agreement with solution data [13]. A zinc binding site and a potential active site have been identified in each monomer. These sites have extensive similarities to those found in other known zinc-dependent hydrolases with de-N-acetylase activity, i.e. MshB from Mycobac- terium tuberculosis [12] and LpxC (UDP-(3-O-(R-3-hy- droxymyristoyl))-N-acetylglucosamine de-N-acetylase) from Aquifex aeolicus [14]. Despite a low degree of structural homology, it has been suggested that these enzymes are the products of convergent evolution due to similar active site features. The objective of this study was to shed light on the role of two de-N-acetylases from B. cereus, which are LmbE protein family homologs. Given the exten- sive homology between B. cereus and B. anthracis, the results of these studies could contribute to understand- ing of the physiology of this interesting pathogenic microbe. Two recombinant de-N-acetylases from B. cereus ATCC14579 were biochemically character- ized. A comparison of the substrate specificities of the enzymes with those of other members of the LmbE D BC1534 RHPDHA´ RHPDHA´ VHPDHN VHPDHD SGHSNH GHPDHV RHPDHT GHPDHR EHPDHE KHVDHR AHADDVEIGMAGTIAKYTKQG PHPDDEAYAAGGTIRLLTD D QG´ PHPDDEAFAAGGTIRLLT QG´ PHPDDGELGCGGTLARAKAEG PHFDDVILSCASTLMELMNQG PHLDDAVLSFGAGLAQAAQDG PHPDDCAIGLGGTIKKLTDSG AHPDDESLSNGATIAHYTSRG AHPDDEAMFFAP TILGLARLK AHADDVEIGMAGTIAKYTKQG 114 126 126 114 157 152 165 167 148 112 31 30 30 31 68 59 35 33 32 29 BC3461 BA1557 BA3524 PIG-L MshB TT1542 Orf2 BtrD Tk-Dac Fig. 1. Partial amino acid sequence alignment of BC1534 and BC3461 with five LmbE-like proteins of known function and three of unknown function (BA1557, BA3524 and TT1542). BA1557, hypothetical protein from Bacillus anthracis strain Ames (NP_844007); BA3524, hypotheti- cal protein from Bacillus anthracis strain Ames (NP_845802); PIG-L, GlcNAc-PI de-N-acetylase from Rattus norvegicus (BAA20869); MshB, GlcNAc-Ins de-N-acetylase from Mycobacterium tuberculosis H37Rv (NP_215686); TT1542, conserved hypothetical protein from Ther- mus thermophilus HB8 (BAC67240); Tk-Dac, N,N¢-diacetylchitobiose de-N-acetylase from Thermococcus kodakaraensis KOD1 (BAD29713); Orf2, de-N-acetylase from Actinoplanes teichomyceticus (CAG15014); BtrD, de-N-acetylase from Bacillus circulans (BAE07068). The black regions indicate identical residues and gray shading indicates similar amino acids. A. Deli et al. LmbE proteins from Bacillus cereus FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS 2741 protein family is presented. A mutational analysis of BC1534 identified functional key residues, highlighting their importance for the catalytic mechanism and the substrate specificity of the enzyme. A computational analysis to predict the biological role of the enzymes is also reported. Results Identification of bc1534 and bc3461 from B. cereus ATCC14579 and comparison with B. anthracis homologs In silico analysis of the B. cereus ATCC14579 genome revealed the presence of the bc1534 and bc3461 genes (NP_831313 and NP_833195, respectively). The bc1534 gene consists of 705 bp and encodes a protein of 234 amino acids, while the bc3461 gene consists of 663 bp and encodes a protein of 220 amino acids. Neither N-terminal signal sequences nor transmembrane helices were found in the deduced amino acid sequences (based on sequence similarities and the sequence prediction programs http://www.cbs.dtu.dk/services/SignalP/ and http://www.sbc.su.se/~miklos/DAS/ for signal peptide and transmembrane domain prediction, respectively), which is consistent with the fact that recombinant BC1534 and BC3461 were detected and purified from the cytosolic fraction of Escherichia coli cells as intra- cellular enzymes. According to the Pfam [15] and Cluster of Ortholo- gous Groups [16] databases, BC1534 and BC3461 are members of the LmbE protein family (Pfam02585⁄ COG2120). BC1534 is also classified as a member of carbohydrate esterase family 14 (CE14), but BC3461 has not been assigned to any of the CE families. Two open reading frames found in B. anthracis strain ‘Ames ancestor’ [BA1557 (NP_844007) and BA3524 (NP_845802)] were identical in size and shared identities of 96% and 95% to BC1534 and BC3461, respectively. The bc1534 gene belongs to an operon that also con- tains six genes that have various predicted functions (Fig. 4). In silico data show that expression of the operon is regulated by a common mechanism (the r E promoter at the 5¢ end of the operon) [17]. Based on the gene organization and predicted functions of the genes that belong to this operon, there is no apparent GlcNAc GlcNAc 2 OH OH O O CH 3 CH 3 CH 3 CH 2 OH H 3 C NH HO HO HO HO HO O O O NH 2 NH 2 HO NH HO HO OH OH OH OH OH OH N NH NH OO OH OH OH OH OH HO HO HO HO 2 C NH 2 CH 3 CH 3 H 3 C OH OH OH HO OH OH OH NH O O O O O O′ O′ P HN N N Cl Cl H N H N H HN H N H HO O O O O O O O O O O O O O O O O O O O O O O HO HO HO 3 1 HO HO HO H Tk-Dac MshB GlcNAc-Ins BtrD N-acetyl-D-glucosaminyl aglycone N-acetyl- D-glucosaminyl pseudoaglycone Orf2 GlcNAc-PI PIG-L Fig. 2. LmbE proteins substrates. All enzymes catalyze hydrolysis of the N-acetyl group (shown inside a circle or a rectangle) of the GlcNAc moiety of their substrate(s). LmbE proteins from Bacillus cereus A. Deli et al. 2742 FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS AB CD EF Fig. 3. (A) Overall structure of BC1534 hexamer formed through the association of three dimers [10] (PDB ID 2ixd). Dimerization is achieved through interaction between the b8-strand of one monomer and the b6- and b7-strands of the other monomer. (B) Surface representation of the enzyme and view of the active site from outside. The active site is occupied by GlcNAc 2 (shown as a ball model). The substrate has been modeled into the active site by autodocking. The two conformations of R140, determined by the crystal structure, are also shown (stick models). (C,D) GlcNAc 2 (stick model) has been docked into the active site of the enzyme, and the interactions or loss of interactions with residues that occupy positions 140 and 42 are indicated. (C) Position 140. The native Arg residue (carbon atoms in yellow) and the mutations Ala (carbon atoms in brown) and Glu (carbon atoms in gray) are shown as stick models. The distances between the substrate and each of the three residues are shown as dashed lines. (D) Position 42. The native Ala residue (yellow) and the mutation Ser (gray) are shown as stick models. The positions of the two native Ser residues (45 and 46) in the neighborhood of the mutation are shown as thin stick mod- els. The distances from the substrate are shown as dashed lines. All the point mutations shown in (C) and (D) were computationally intro- duced in the model of the native crystal structure (see Experimental procedures). (E) Cartoon representation of the active site of BC1534, highlighting residues that have been mutated and showing their relative positions in the structure. The H110 ⁄ D112 pair is shown as a stick model, and three of the residues that form a hydrophilic pocket suggested to function as the ‘oxyanion hole’ are shown using Van der Waals dotted spheres. (F) Cartoon representation that focuses on the entrance of the active site tunnel, showing that it is dominated by positively charged residues represented here by Van der Waals dotted spheres. The position of K207 is also shown. A. Deli et al. LmbE proteins from Bacillus cereus FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS 2743 pathway in which all of these genes could be involved. In contrast, bc3461 is not part of a gene cluster, indi- cating that its expression is probably regulated in an independent manner (Fig. 4). Protein sequence alignment of characterized mem- bers of this family with the B. cereus homologs (Fig. 1) revealed two conserved sequence motifs. The first [(A ⁄ P) H(X ⁄ P)DD] is located near the N-terminus, and the second [ H(X ⁄ P)DH] is located towards the middle of the protein. The crystal structure of BC1534 [10] revealed that the underlined H and D residues in the first motif and the last H of the second motif are zinc binding ligands. Moreover, it has been proposed for other members of the family that the underlined H and the subsequent D of the second motif play a charge relay role during catalysis [9]. Cloning, expression and purification of BC1534 The gene encoding BC1534 was isolated from B. cereus ATCC14579 genomic DNA by PCR and cloned into expression vector pET26b for recombinant protein production in E. coli. BC1534 was produced as a C-terminal hexahistidine fusion protein to facilitate puri- fication by affinity chromatography (Ni-nitrilotriacetic acid). SDS ⁄ PAGE analysis revealed the presence of three protein bands. The position of the main band is in agreement with the predicted molecular mass of the hexahistidine fusion protein (27 kDa). The N-terminal amino acid sequence of the protein bands correspond- ing to the higher-molecular-mass proteins seen in SDS ⁄ PAGE was determined to be MSGL, which is identical to the predicted amino acid sequence of BC1534 ( MSGLHILAFG), suggesting the existence of non-denatured homopolymers of BC1534 in the SDS ⁄ PAGE gel (Fig. 5A). The positions of these bands are in agreement with the molecular mass of purified BC1534 estimated by size-exclusion chroma- tography [13] and glutaraldehyde cross-linking [18] (approximately 160 kDa), indicating that the protein exists as a hexamer in solution. Cloning, expression and purification of BC3461 bc3461 was isolated from B. cereus ATCC14579 geno- mic DNA by PCR and cloned into the pRSETA expression vector for recombinant protein production in E. coli. Purification was achieved in two steps, using ion-exchange and size-exclusion chromatography. The apparent molecular mass of BC3461 was calculated to be 26 kDa as determined by SDS ⁄ PAGE (Fig. 5B) and gel-filtration chromatography, suggesting that the pro- tein exists as a monomer in solution. Enzymatic properties of BC1534 and BC3461 BC1534 and BC3461 were active on N-acetyl-d-gluco- samine (GlcNAc), N-acetylchitooligomers (GlcNAc 2 , GlcNAc 3 and GlcNAc 4 ), GlcNAc-1P, GlcNAc-6P, GalNAc and ManNAc (Table 1). The specificity of the enzymes for various N-acetylchito-oligomers was examined, and the kinetic parameters were determined BC1531 BC1532 1482325 BC3459 BC3460 Putative transcriptional regulatory protein Dihydrodipicolinate reducatase Short chain dehydrogenase LmbE-related protein Hypothetical protein Hypothetical protein Arsenical pump membrane protein LmbE-related protein Glycosyltransferase tRNA CCA- pyrophosphorylase Biotin-opemn repress or/biotin–[acetyl- CoA-carboxylase] synthetase Methylglyoxal synthase BC3461 BC3462 BC3463 3415277 3419625 BC1534 BC1536 BC1537mgsA BC1535 1487879 A B Fig. 4. Gene organization in the 5.5 kbp region that includes bc1534 (A) and the 4.3 kbp that includes bc3461 (B) on the B. cereus ATCC14579 genome. Arrows indicate open reading frames. Genes of interest are indicated by colored arrows. LmbE proteins from Bacillus cereus A. Deli et al. 2744 FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS (Tables 1 and 2). BC1534 and BC3461 exhibited maxi- mum activity towards GlcNAc 2 . Kinetic parameters for GlcNAc and GlcNAc 2 were obtained from Linewe- aver–Burk plot analysis, and the enzyme reaction rates for these substrates appear to follow Michaelis– Menten kinetics. The resulting k cat ⁄ K m ratios (catalytic efficiency, K eff ) indicated that GlcNAc 2 was the favored substrate for both BC1534 and BC3461 (Table 2). UDP-GlcNAc was also tested as a potential substrate due to the presence of a glycosyltransferase A B 104 175 47.5 62 25 32.5 83 16.5 BC3461 6.5 37 97 50 29 20 BC1534 Fig. 5. SDS ⁄ PAGE of the purified LmbE-like proteins BC1534 (A) and BC3461 (B). (A) Lane 1, molecular weight markers; lane 2, size-exclusion chromatography eluant. Samples were electrophoresed on a 12% polyacrylamide gel under denaturing conditions. Protein bands were visualized by staining with Coomassie Brilliant Blue R-250. Table 2. Kinetic parameters of BC3461 and BC1534 (wild-type and variants) towards GlcNAc and GlcNAc 2 . NA, not active. Enzyme Substrate GlcNAc GlcNAc 2 k cat (s )1 ) K m (lM) K eff (lM )1 Æs )1 ) k cat (s )1 ) K m (lM) K eff (lM )1 Æs )1 ) BC3461 6.4 9 0.71 11.1 2 5.55 BC1534 1.89 3 0.63 98.2 3 32.7 BC1534 R140A 189.7 2500 0.08 0.37 4 0.09 BC1534 R140E 0.27 8 0.03 0.27 5 0.05 BC1534 A42S 0.01 1.3 · 10 3 8.4 · 10 )6 0.01 1.1 · 10 3 9 · 10 )6 BC1534 D112N NA NA NA 5.97 6 0.99 BC1534 K207I NA NA NA NA NA NA BC1534 Y194F 48.26 3.8 · 10 3 0.012 15.78 1.1 · 10 3 0.014 Table 1. Substrate specificity (percentage relative activity)of BC3461, BC1534 and mutants. Assay conditions were 25 mM HEPES ⁄ NaOH pH 8.0, 200 m M NaCl, 1 mM CoCl 2 for 30 min at 37 °C for BC1534 and its variants, and 25 mM MES ⁄ NaOH pH 6.5, 200 mM NaCl, 1 mM MgCl 2 for 30 min at 20 °C for BC3461. The concentration of substrates was adjusted with respect to their content in terms of N-acetyl residues. ND, not determined. Substrate BC3461 BC1534 BC1534 R140A BC1534 R140E BC1534 A42S BC1534 D112N BC1534 Y194F GlcNAc 39 53 93 80 41 0 100 GlcNAc 2 100 100 100 100 25 53 100 GlcNAc 3 92 43 27 40 31 16 32 GlcNAc 4 65 35 20 0 0 0 13 GlcNAc 5 000 0 0 0 6 GlcNAc 6 000 0 0 0 14 GlcNAc-1P 65 88 67 50 100 0 30 GlcNAc-6P 27 53 87 40 50 99 30 GalNAc 58 60 40 20 25 0 10 ManNAc 58 60 0 0 12 100 0 GlcNAc-I-P-C 18 ND 0 ND ND ND ND ND GMDP 0 0 ND ND ND 19 22 A. Deli et al. LmbE proteins from Bacillus cereus FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS 2745 gene (bc1535) downstream BC1534 (Fig. 4). The K m value (3 lm) was comparable to that for GlcNAc 2 , but the K eff value (0.02 lm )1 Æs )1 ) was significantly lower compared to that for GlcNAc 2 . The purified recombinant enzymes showed different pH and temperature optima using GlcNAc 2 as sub- strate. BC1534 exhibited a pH optimum at pH 8.0, and optimum temperature for enzyme activity was determined to be 37 °C. The pH and temperature optima for BC3461 were 6.5 and 20 °C, respectively. Moreover, a 3.5-fold increase in BC1534 activity was observed on addition of 1 mm CoCl 2 to the assay buf- fer, and a twofold increase in BC3461 activity was seen when 1 mm MgCl 2 was added. BC1534 and BC3461 were inhibited by the presence of 1 mm Cu 2+ and 1mm Zn 2+ (both tested as chlorides), similar to previ- ous reports on zinc hydrolases [19]. The enzymes were not inhibited by acetate or EDTA even at concentra- tions up to 50 and 20 mm, respectively. Moreover, they were both inactive on radiolabeled glycol chitin and peptidoglycan from B. cereus vegetative cell walls, and BC1534 was also inactive on GlcNAc-I-P-C 18 , a syn- thetic analogue of GlcNAc-PI. Mutational analysis of BC1534 In order to elucidate the importance of selected resi- dues in the catalytic mechanism, substrate affinity and specificity, as well as to investigate the importance of the oligomerization state on the activity of the enzyme, six point mutations and one fragment deletion were performed. The enzyme variants obtained were tested against the same range of substrates as the wild-type enzyme, and their substrate specificities and kinetic parameters are presented in Tables 1 and 2. BC1534 shares extensive similarities to the active sites of two characterized zinc-dependent deacetylases, i.e. MshB and LpxC [6,14]. Based on these similarities, it was proposed [10] that BC1534 would utilize a simi- lar catalytic mechanism to these enzymes. The domi- nant features of this mechanism are an H ⁄ D charge relay pair (H110 ⁄ D112 for BC1534, Fig. 3E) and an ‘oxyanion hole’ (mainly formed by the side chains of Y194, N150 and D108 in BC1534, Fig. 3E). To test this hypothesis, we mutated D112 and Y194 to resi- dues that lack groups with a putative role in catalysis. Thus, D112 was mutated to asparagine, which retains all aspartate’s stereochemical and physicochemical properties except its ability to relay protons. The mutant protein (D112N) was completely inactive on most of the substrates tested, i.e. GlcNAc, GlcNAc 4 , GlcNAc 5 , GlcNAc 6 and GlcNAc-1P, but exhibited higher relative activity against GlcNAc-6P and Man- NAc than the native enzyme did (Table 1). Y194 was mutated to phenylalanine, which retains all the hydro- phobic interactions of the initial residue but lacks the hydroxyl group that is proposed to participate in for- mation of the ‘oxyanion hole’. The Y194F mutant showed the broadest substrate specificity among BC1534 variants, but its catalytic efficiency against the preferred substrate of the native enzyme (GlcNAc 2 ) was highly reduced. In addition, it exhibited activity against GlcNAc 5 , GlcNAc 6 and GMDP, in contrast to the wild-type enzyme, which was inactive with these substrates. Based on the crystal structure of BC1534 [10] and a molecular dynamics study [20], we proposed that the rim and the loops surrounding the active site tunnel could play a significant role in the substrate specificity of the enzyme. We experimentally tested this hypothe- sis by mutating R140, K207 and A42, three residues located on the rim of the active site tunnel. In the crys- tal structure, R140 adopts two discrete conformations (Fig. 3B). One of these conformations highly restricts the accessibility of the active site, while the other, which is mainly stabilized by electrostatic interactions with the adjacent E142 and backbone carbonyl groups, keeps the entrance open. To further investigate the role of R140, it was mutated (a) to a small hydrophobic residue, Ala and (b) to a negatively charged Glu resi- due. With minor exceptions, both variants exhibited the same preferences as the native enzyme for the sub- strates tested (Table 1). R140A showed a significant decrease in efficiency compared to the native enzyme for GlcNAc 2 and GlcNAc. However, this variant exhibited 100 times higher catalytic activity (k cat )on GlcNAc than the wild-type enzyme. In the case of R140E, the K m values remain similar to that of the native enzyme, but the k cat is significantly decreased for both substrates. K207 was mutated to isoleucine, a small hydropho- bic residue. Remarkably, this variant was inactive on all substrates tested. A42 is located in a position favor- able for the formation of hydrogen bonds with the substrate’s hydroxyl groups (Fig. 3D). Thus, A42 was mutated to a serine, which is a residue of comparable size to alanine and its side chain has the ability to form hydrogen bonds. Unexpectedly, the produced variant (A42S) exhibited a dramatic reduction in K eff value for both substrates (Table 2). BC1534 is a hexamer (Fig. 3A) that may be consid- ered as a trimer of dimers [10,20]. Dimerization is mainly established via exchange of two short b-strands between the monomers (b8-strand in Fig. 3A). Dele- tion of the b8-strand by PCR resulted in an insoluble and inactive protein. LmbE proteins from Bacillus cereus A. Deli et al. 2746 FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS Prediction of function Although members of Pfam02585 (LmbE protein fam- ily) share common sequence features, they do not exhi- bit the same function [5–8]. Sequence alignment of 929 proteins belonging to Pfam02585 revealed a number of different sequence groups (Fig. 6), presumably reflect- ing different functions. All proteins for which the enzy- matic function has been studied belong to different groups. The two Bacillus proteins were placed in func- tional categories different from other LmbE proteins. As the sequence of the protein is not sufficient to reveal the function of the enzyme, we sought other lines of evidence to predict its function. Location of enzymes in the same chromosomal neighborhood could indicate a functional relationship and frequently helps to predict the function of enzymes [21]. BC1534 appears to be in a chromosomal neighborhood that is conserved among the Bacillus species (data not shown). Microarrays [22] and deep sequencing of the transcrip- tome [23] in B. anthracis revealed that ba1557 (the homolog of bc1534) is in an operon surrounded by genes homologous to bc1531-bc1537, and is expressed during the early and mid log phases, while ba3524 (the homolog of bc3461) is expressed in the early log phase and during sporulation. Despite possible differences in the life cycle of these Bacillus species, these data from the B. anthracis transcriptome [22,23] could provide strong evidence for similar gene expression in B. cereus. In other members of the Firmicutes, the conservation is restricted to the presence of a glycosyltransferase (Pfam00534) downstream of the LmbE-related protein (i.e. the proteins encoded by the bc1535 and bc1534 genes, respectively, in B. cereus). Discussion In an effort to shed light on the role of the LmbE pro- tein family enzymes in bacteria and contribute to the understanding of the pathobiology of B. anthracis,we describe biochemical characterization of the recombi- nant enzymes BC1534 and BC3461 from B. cereus. BC1534 exhibited overall 21% identity and 31% simi- larity with BC3461. Moreover, BC1534 and BC3461 shared 96% and 95% identity with their homologs BA1557 and BA3524, respectively, from B. anthracis (Fig. 1). The purified recombinant enzymes exhibited different molecular masses (Fig. 5), pH and temperature optima and were not inhibited by acetate or EDTA. BC1534 and BC3461 were activated by CoCl 2 and MgCl 2 , respectively. Both enzymes were effective in de-N-acetylating GlcNAc and N-acetylchitooligomers (GlcNAc 2 , GlcNAc 3 and GlcNAc 4 ), as well as GlcNAc- 1P, GlcNAc-6P, GalNAc, ManNAc and UDP-GlcNAc (Table 1). However, the enzymes were not active on glycol chitin or peptidoglycan from B. cereus ATCC14579, GMDP or GlcNAc-I-P-C 18 . Kinetic analysis of BC3461 and BC1534 towards the N-acetylchito-oligosaccharides GlcNAc and GlcNAc 2 revealed that GlcNAc 2 is the favored substrate for both enzymes. Comparison of the K eff values showed that both enzymes are equally effective on GlcNAc. BC1534 was six times more THA1200 Thermus thermophilus HB8 BC3461 Bacillus cereus ATCC 14579 TK1764 Thermococcus kodakaraensis KOD1 BC1534 Bacillus cereus ATCC 14579 BtrD Bacillus circulans Rv1170 Mycobacterium tuberculosis H37Rv Fig. 6. Neighbor joining tree for 929 LmbE proteins. Bacillus cereus proteins and pro- teins of known function are indicated. A. Deli et al. LmbE proteins from Bacillus cereus FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS 2747 effective than BC3461 (Table 2) when GlcNAc 2 was tested as substrate. Chitin de-N-acetylases from fungi and insects [24], chito-oligosaccharide de-N-acetylases from Rhizobium (NodB) [25] and Vibrio parahaemolyticus [26], and Glc- NAc peptidoglycan de-N-acetylases [27] are considered to be catalytically similar de-N-acetylases. They are all members of carbohydrate esterase family 4, catalyzing the hydrolysis of N-linked acetyl groups on GlcNAc residues. Chitin de-N-acetylase is capable of removing N-acetyl groups from chitin chains [28]. NodB, which is involved in nodulation signal synthesis, de-N-acety- lates the non-reducing GlcNAc residue of N-acety- lchito-oligosaccharides [25]. GlcNAc peptidoglycan de-N-acetylase increases the resistance of peptidoglycan to lysozyme via de-N-acetylation of GlcNAc residues [27]. None of the above enzymes accepts GlcNAc as substrate. The biochemically characterized LmbE proteins are members of distinct metabolic pathways. MshB is involved in the second step of mycothiol biosynthesis [6], whereas GlcNAc-PI de-N-acetylases play an impor- tant role in biosynthesis of the glycosylphosphatidyli- nositol biosynthesis in eukaryotes [3–5]. Orf2 [8] and BtrD [9] are involved in the synthesis of lipoglycopep- tide antibiotics, while Tk-Dac plays an essential role in a novel chitinolytic pathway identified in archaea [7]. All members of the LmbE protein family with known function share a common feature in that they de-N-acet- ylate the GlcNAc moiety of their substrates (Fig. 2). It has been reported that MshB [6], Orf2 [8] and PIG-L (GlcNAc-PI de-N-acetylase from Rattus norvegicus) are not active on GlcNAc [3–5]. Tk-Dac de-N-acetylates GlcNAc monomers as well as N-acetyl-chitooligomers [7], but does not de-N-acetylate GlcNAc-6P or ManNAc. In contrast to other LmbE protein family members, BC1534 and BC3461 are active on GlcNAc, GlcNAc 2 , GlcNAc 3 , GlcNAc 4 , GlcNAc-1P, GlcNAc-6P, GalNAc, ManNAc and UDP-GlcNAc, thus exhibiting a broader substrate specificity compared to other LmbE protein family members. The exact biological role of BC1534 and BC3461 proteins remains unclear. Microarray data [29] and deep sequencing [23] of the transcriptome showed that the homologs in B. anthracis (BA1557 and BA3524 for BC1534 and BC3461, respectively) are expressed in dif- ferent phases of the cell cycle (early to mid log phase for BC1534 and late sporulation to early log phase for BC3461). Of GlcNAc, GlcNAc 2 and GlcNAc-6P, which were tested for their docking properties in the BC1534 active site, GlcNAc 2 had the lowest calculated binding energy (data not shown), which is in agree- ment with the kinetic parameters shown in Table 2 indicating that GlcNAc 2 is the favored substrate. RT-PCR experiments (data not shown) revealed similar expression profiles for BC1534 and BC3461 to that for an exochitinase (BC3725, EC 3.2.1.14) from B. cereus ATCC14579. This observation, in combina- tion with the reported chitinolytic activity of B. cereus [30] and the presence of an endochitinase and a chito- sanase genes (bc0429 and bc2682 respectively) in its genome, support possible involvement of these enzymes in a chitinolytic pathway, similar to Tk-Dac. An alignment-based tree (Fig. 6) placed both enzymes in functional categories different from other LmbE proteins. Analysis of the chromosome organization of bc1534 revealed the existence of a glycosyltransferase gene (bc1535) immediately downstream in the operon (Fig. 4). This gene organization is common for most Firmicutes genomes, suggesting that the two proteins are functionally related in these organisms. Interest- ingly, we observed that BC1534 is also active on UDP-GlcNAc. As glycosyltransferases and UDP- GlcNAc are situated at a biosynthetic branch point leading to peptidoglycan formation, a possible role of BC1534 (and BC1535, EC 2.4.1 ) in modulating pepti- doglycan biosynthesis can be envisaged. In order to elucidate the importance of selected resi- dues in the catalytic mechanism, substrate affinity and specificity, as well as to investigate the importance of the oligomerization state on the activity of the enzyme, six point mutations and one fragment deletion were performed. Central features of the catalytic mechanism are an H ⁄ D pair (H110 ⁄ D112, Fig. 3E) [10] that is proposed to play the role of a charge relay, and a hydrophilic pocket proposed as an ‘oxyanion hole’ (Y194, N150 and D108, Fig. 3E). We tested the valid- ity of this hypothesis by mutating D112 to N and Y194 to F. The variant D112N shown complete aboli- tion of catalytic efficiency against GlcNAc, and the k cat against GlcNAc 2 was decreased approximately 16 times with a subsequent decrease in K eff , indicating that the hydroxyl group of D112 plays a significant role in catalysis. The Y194F variant exhibited broad substrate specificity similar to the native enzyme (Table 2). However in contrast to the native enzyme, it showed a dramatic increase (> 10 3 )inK m for GlcNAc and GlcNAc 2 , and similar catalytic efficiency for both substrates. These results suggest that the tyrosine hydroxyl group is directly associated with the enzyme’s affinity for the substrate. In order to test the suggestion that the loops sur- rounding the active site and the rim of the tunnel are directly implicated in determining the accessibility of the active site and the enzyme’s substrate specificity [10,20], we mutated three residues located on the rim LmbE proteins from Bacillus cereus A. Deli et al. 2748 FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS (R140, K207, A42) (Fig. 3B–D,F). Automated docking of GlcNAc 2 substrate in the R140A active site showed that the substrate was arranged in a loose way with fewer hydrophobic interactions and a weaker hydrogen bond compared with the native enzyme. This different orientation of the reaction site of the substrate could explain the lower reported K eff values (Table 2) for R140A. In the native enzyme, the rim of the active site tunnel is dominated by positively charged residues, i.e. R109, K154, K187, K207, K220 etc., that may drive acetate out of the active site after the end of the reac- tion. The positioning of a negatively charged residue (E140) could hamper the release of the acetate from the active site. This unfavorable behavior could account for the lower catalytic reaction rates (k cat )in the case of R140E (Table 2). Surprisingly, the mutation K207I resulted in an inac- tive enzyme. K207 is located on the rim of the active site and it is quite unlikely that it has any influence on the overall structure of the enzyme as it was found at the edge of an a-helix (Fig. 3F). K207 is not conserved among other LmbE family members, indicating that this residue is a unique feature of BC1534 that could be related to the specificity of the enzyme. A42S exhibited lower K eff values, which could be due to changes in interactions stabilizing the substrate and ⁄ or interactions with residues involved in the cata- lytic mechanism (R53 and ⁄ or D76) [10] (Fig. 3D). Deletion of the short b8-strand resulted in an insoluble and inactive enzyme, supporting a previous suggestion [10] that the structural building block of the BC1534 is a dimer formed via b-strand exchange (Fig. 3A). Glucosamine (GlcN) is of importance in biomedicine as it is used as dietary supplement for osteoarthritis [31,32]. Currently GlcN is produced by acid hydrolysis of chitin extracted from crab and shrimp shells [33]. A new fermentation process utilizing E. coli cells modified by metabolic engineering for the production of high- quality and low-cost GlcN has recently been reported [34]. The BC1534 R140A enzyme variant is potentially a candidate for the enzymatic production of GlcN due to its significantly increased k cat towards GlcNAc. In conclusion, we have biochemically characterized two LmbE proteins from B. cereus that exhibit the broadest substrate specificity compared to other LmbE protein family members, so far reported. BC1534 and BC3461 appear to have distinct functional roles, as shown by their different expression profiles, chromo- somal organization and sequence alignments. Due to their high similarity to their B. anthracis homologs, clarification of their biological roles will contribute to a better understanding of the properties of this life- threatening bacterium. Experimental procedures Materials Primers were synthesized by the Microchemistry Facility of the Institute of Molecular Biology and Biotechnology (Heraklion, Greece). The expression plasmid pET26b and E. coli BL21 DE3 were obtained from Novagen (Merck, KGaA, Darmstadt, Germany). The expression plasmid pRSETA was purchased from Invitrogen (Carlsbad, CA, USA). E. coli BL21 T7 Express lysY was purchased from New England Biolabs GmbH (Frankfurt, Germany). All chromatographic materials were obtained from GE Health- care Bio-Sciences AB (Uppsala, Sweden). Ni-nitrilotriacetic acid agarose, PCR and gel extraction kits were purchased from Qiagen (Valencia, CA, USA). Plasmid purification and RNA isolation kits were purchased from Macherey-Nagel GmbH & Co. KG (Duren, Germany). The RT-PCR kit was obtained from Finnzymes Oy (Espoo, Finland). Substrates (including glycol chitosan) and common biochemicals were purchased from Sigma-Aldrich Ltd (St Louis, MO, USA). Restriction enzymes and DNA-modifying enzymes were purchased from MINOTECH Biotechnology (Heraklion, Greece) and New England Biolabs GmbH. The instruments Fluostar Galaxy and Mastercycler Gradient (for PCR and RT-PCR) were purchased from BMG Labtechnologies GmbH (Offenburg, Germany) and Eppendorf Netheler- Hinz GmbH (Hamburg, Germany), respectively. Construction of expression plasmids bc1534 and bc3461 genes were isolated from B. cereus ATCC14579 genomic DNA. The primers used for bc1534 were BC1534-For (5¢-GGAATTC CATATGATGAGTGG- ATTACATATATTA-3¢; NdeI restriction site underlined) and BC1534-Rev (5¢-CCG CTCGAGTTTACATCCCCCT- AATAAATC-3¢; XhoI restriction site underlined). Plasmid pET26b was digested with NdeI and XhoI, and bc1534 was ligated into the corresponding sites, resulting in plasmid pET26b-bc1534. This plasmid construction was used for the production of BC1534 protein with a histidine tag at its C-terminus. The primers used for bc3461 were BC3461-For (5¢-ATGGAGAGACATGTACTTGTT-3¢) and BC3461- Rev (5¢-CCG CTCGAGCTACT CCCAT TTATA AGTCCA- 3¢; XhoI restriction site underlined). Initial digestion of plasmid pRSETA with NdeI was followed by incubation with the Klenow fragment of DNA polymerase I, and finally diges- tion with XhoI. The bc3461 gene was ligated into the corre- sponding sites of the plasmid. Production and purification of BC4361, BC1534 and BC1534 variants To over-express bc1534, the plasmid pET26b-bc1534 was used for transformation of E. coli BL21 DE3. The resultant A. Deli et al. LmbE proteins from Bacillus cereus FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS 2749 [...]... (2ixd.pdb) to generate atomic energy grids Point mutations were introduced at positions 140 and 42 using XtalView ⁄ Xfit [42] and ⁄ or MIFit (Rigaku Americas Corp.) All water and acetate atoms were removed Hydrogen atoms were added to the crystal structure, and the substrates and atomic partial charges were assigned to the protein and carbohydrate atoms Atomic interaction energy grids were determined using... Bacillus cereus prepared using the programs pymol [43] and GIMP (http:// www.gimp.org/) Alignment-based tree Alignments of LmbE proteins were performed using clustalw [44] Alignment-based trees were constructed using neighbor-joining and maximum-likelihood algorithms implemented in the software package PHYLIP [45] In all cases, the topologies of the trees were very similar, and they were identical with. .. respect to the proteins of known function Information about the gene context was obtained using the Integrated Microbial Genomes system [46] Transcription profiles of B .cereus genes were inferred from transcription analysis studies on other Bacillus genomes Orthologs of B .cereus genes to genes with transcription data from other Bacillus species were identified using the Integrated Microbial Genomes [46], and. .. tailor-made genes using the polymerase chain reaction BioTechniques 8, 528–535 39 Goodsell DS & Olson AJ (1990) Automated docking of substrates to proteins by simulated annealing Proteins 8, 195–202 40 Goodsell DS, Morris GM & Olson AJ (1996) Automated docking of flexible ligands: applications of autodock J Mol Recognit 9, 1–5 41 Goodsell DS, Lauble H, Stout CD & Olson AJ (1993) Automated docking in crystallography:... corresponding to each atomic type ˚ found in the ligand at 0.375 A grid positions in a cubic box of 30 A side length centered on the active site After docking, all structures generated for a single compound were assigned to clusters with an rmsd among the structures of the same cluster no greater than 1 For each ligand, the global minimum structure is discussed here The figure 3 was LmbE proteins from Bacillus. .. cloning of PIG-L, a candidate N-acetylglucosaminylphosphatidylinositol deacetylase J Biol Chem 272, 15834–15840 FEBS Journal 277 (2010) 2740–2753 ª 2010 The Authors Journal compilation ª 2010 FEBS 2751 LmbE proteins from Bacillus cereus A Deli et al 4 Watanabe R, Ohishi K, Maeda Y, Nakamura N & Kinoshita T (1999) Mammalian PIG-L and its yeast homologue Gpi12p are N-acetylglucosaminylphosphatidylinositol... heterodisaccharide from chitin Appl Microbiol Biotechnol 75, 357–365 Psylinakis E, Boneca IG, Mavromatis K, Deli A, ˚ Hayhurst E, Foster SJ, Varum KM & Bouriotis V (2005) Peptidoglycan N-acetylglucosamine deacetylases from Bacillus cereus, highly conserved proteins in Bacillus anthracis J Biol Chem 280, 30856–30863 Tsigos I, Martinou A, Kafetzopoulos D & Bouriotis V (2000) Chitin deacetylases: new, versatile tools in. .. fluorescence reader with excitation and emission wavelengths of 390 and 460 nm, respectively A calibration curve using glucosamine showed that the free amine labeling reaction was linear up to a concentration of 2.4 mm glucosamine All measurements are shown are the mean of three replicates RT-PCR B cereus ATCC14579 was cultivated in LB medium (halfstrength) Total RNA was isolated using a Nucleospin RNA II kit... GAAACCTTCTGTTA-3¢) for point mutation Y194F Engineered codons are underlined Deletion of the gene fragment encoding the b8-strand was achieved by PCR pET26b-bc1534 was used as the template, and the primers used were BC1534-For and delb8-Rev (5¢-CCGCTC GAGACTCATAAATCCCTCGGCAT-3¢; XhoI restriction site underlined) Autodocking Automated docking simulations were performed using AutoDock4 [39–41] using the crystal... function of substrate concentration were determined using the graphing and statistical software package hyper (http://homepage.ntlworld.com/john.easterby/ software.html) Values for Km and Vmax were calculated by fitting initial velocities at various substrate concentrations to the appropriate forms of Michaelis–Menten equation The kcat values were calculated from Vmax using molecular masses of 27 214 and 24 . LmbE proteins from Bacillus cereus are de-N-acetylases with broad substrate specificity and are highly similar to proteins in Bacillus anthracis Alexandra. joining tree for 929 LmbE proteins. Bacillus cereus proteins and pro- teins of known function are indicated. A. Deli et al. LmbE proteins from Bacillus cereus FEBS