Research article

Cross-functionalities of Bacillus deacetylases involved in bacillithiol biosynthesis and bacillithiol-S-conjugate detoxification pathways

Zhong Fang, Alexandra A. Roberts, Karissa Weidman, Sunil V. Sharma, Al Claiborne, Christopher J. Hamilton, Patricia C. Dos Santos


BshB, a key enzyme in bacillithiol biosynthesis, hydrolyses the acetyl group from N-acetylglucosamine malate to generate glucosamine malate. In Bacillus anthracis, BA1557 has been identified as the N-acetylglucosamine malate deacetylase (BshB); however, a high content of bacillithiol (~70%) was still observed in the B. anthracisBA1557 strain. Genomic analysis led to the proposal that another deacetylase could exhibit cross-functionality in bacillithiol biosynthesis. In the present study, BA1557, its paralogue BA3888 and orthologous Bacillus cereus enzymes BC1534 and BC3461 have been characterized for their deacetylase activity towards N-acetylglucosamine malate, thus providing biochemical evidence for this proposal. In addition, the involvement of deacetylase enzymes is also expected in bacillithiol-detoxifying pathways through formation of S-mercapturic adducts. The kinetic analysis of bacillithiol-S-bimane conjugate favours the involvement of BA3888 as the B. anthracis bacillithiol-S-conjugate amidase (Bca). The high degree of specificity of this group of enzymes for its physiological substrate, along with their similar pH–activity profile and Zn2+-dependent catalytic acid–base reaction provides further evidence for their cross-functionalities.

  • bacillithiol
  • bimane
  • BshB
  • deacetylase
  • Nacetylglucosamine malate
  • zinc hydrolase


The cysteine-containing tripeptide glutathione (GSH) is the major low-molecular-mass thiol in eukaryotes and many Gram-negative bacteria (Figure 1). In addition to its major role in thiol-redox homoeostasis [1], GSH is also widely implicated in xenobiotic detoxification. This is facilitated by GSTs, which can catalyse S-conjugation of GSH to a diverse range of electrophilic xenobiotics [2]. In Gram-negative bacteria, GSH plays a sacrificial role in electrophile detoxification as these glutathione S-conjugates are directly exported from the cell [3].

Figure 1 Low-molecular-mass thiols found in bacteria

Many Gram-positive bacteria lack GSH, but instead produce other low-molecular-mass thiols. Mycothiol (MSH) (Figure 1) is the dominant low-molecular-mass thiol in Actinobacteria (formerly known as Actinomycetes) (e.g. Mycobacterium, Corynebacterium and Streptomyces), which serves functions analogous to those of GSH [4,5]. However, one notable difference is that MSH-dependent drug detoxification pathways utilize a mycothiol-S-conjugate amidase (Mca) to hydrolyse the mycothiol S-conjugate amide linkage to liberate the CysNAc (N-acetylcysteine) S-conjugate (mercapturic acid), which is exported from the cell, and glucosamine inositol, which is recycled back into the MSH biosynthetic pathway [6].

In 2009, bacillithiol (BSH) (Figure 1) was identified as a low-molecular-mass thiol among many low-G+C Gram-positive bacteria (Firmicutes) lacking GSH or MSH [7]. These include Bacillus spp. (e.g. B. anthracis, B. subtilis, B. cereus, B. megaterium and B. pumilis) and some, but not all, staphylococci (e.g. Staphylococcus aureus and Staphylococcus saprophyticus) and streptococci (e.g. Streptococcus agalactiae). So far, BSH-deficient mutants have been shown to display impaired sporulation, sensitivity to acid and salt, increased sensitivity to fosfomycin [8] and reduced viability in mouse macrophage cell lines [9]. A detailed understanding of the enzymes mediating BSH biosynthesis and BSH-dependent drug resistance is central to identifying whether and how such biological targets could be exploited for the design of new antibiotic chemotherapies.

The three-step biosynthetic pathway for BSH (Figure 2) shares similarities with the key steps in MSH biosynthesis. BSH biosynthesis is initiated by a retaining glycosyltransferase (BshA) that catalyses the glycosylation of L-malic acid to afford GlcNAc-Mal (N-acetylglucosamine malate) [8,10]. An N-acetylhydrolase (BshB) then liberates the free amine GlcN-Mal (glucosamine malate) [8,10]. Gene-knockout studies in B. subtilis have identified a bacillithiol synthase (BshC) [8], proposed to catalyse the final conversion of GlcN-Mal into BSH, but attempts to demonstrate the BSH synthase activity of BshC in vitro have so far been unsuccessful.

Figure 2 The proposed biosynthetic pathway and detoxification mechanism of bacillithiol

In many bacilli (e.g. B. subtilis, B. anthracis and B. cereus), bshA and bshB are adjacent genes located in the same operon [8] (Table 1). Whereas BshA and BshC are essential for BSH biosynthesis, in many cases bshB-knockout mutants display only a partial reduction in BSH levels [8,10]. This has been attributed to the presence of one, or more, additional bshB-like gene products, which can also N-deacetylate GlcNAc-Mal. In B. subtilis, for example, a ΔbshB1 (ypjG) mutant retains almost half of the WT (wild-type) levels of BSH, whereas the ΔbshB1bshB2 (yojG) double-knockout strain showed no detectable levels of BSH [8,10]. Although in vivo data provide a strong indication for cross-functionality of BshB and orthologous enzymes in the biosynthesis of bacillithiol, biochemical validation of this proposal has not been carried out. In Mycobacterium smegmatis, functional redundancy was also observed for analogous N-deacetylase in MSH biosynthesis (MshB). In this case 5–10% of WT MSH levels are still produced in ΔmshB mutants due to the residing background N-deacetylase activity of Mca [11,12]. Interestingly, in S. aureus, which only contains a single bshB-like gene, preliminary studies have indicated the presence of a bacillithiol-S-conjugate amidase (Bca) pathway that is capable of detoxifying the electrophilic xenobiotic monobromobimane [13]. Considering these in vivo observations, it seems plausible that one or more of these BshB-like enzymes could also present Bca activity in xenobiotic detoxification. In pathways involving BSH, neither the identity of any Bca enzyme (the equivalent of Mca in BSH-producing species) nor the kinetic analysis of enzymes able to perform both reactions has been determined.

View this table:
Table 1 Functional assignment of BSH biosynthetic and detoxification genes

Before the discovery of BSH, the crystal structure of a B. cereus Zn2+-binding protein (BC1534) (shown in the present study to be a BshB) was reported [14,15], which was characterized as a potential chitin deacetylase on the basis of its observed N-deacetylase activity with GlcNAc (Km, 3 μM; kcat, 2 s−1) and chitobiose (Km, 3 μM; kcat, 98 s−1) [14,15]. Since then, initial kinetic studies of the homologous (97% sequence identity) B. anthracis enzyme (BA1557) have demonstrated its BshB activity with GlcNAc-Mal, but no notable GlcNAc deacetylase activity was observed [8,10]. Although it is anticipated that GlcNAc-Mal is the physiological substrate of BC1534, the participation of BC1534 and/or BC3461 in BSH biosynthesis and/or detoxification has not been explored at all.

In the present study, we have investigated the BshB and Bca properties of the B. anthracis N-deacetylases BA1557, BA3888 and BA3524 as well as the orthologous B. cereus enzymes BC1534 and BC3461 (Table 1). The results indicate that both BA1557 and BA3888 (and the B. cereus homologues) are competent in catalysing the N-deacetylation of GlcNAc-Mal, whereas BA3888 showed activity against BSmB (bacillithiol-S-bimane) adduct. Assays with GlcNAc-Mal substrate analogues demonstrate that the malate portion of the molecule plays a role in controlling substrate specificity. In addition, a site-directed mutagenesis approach provides insight into the mechanistic details of the Zn2+-dependent catalytic acid–base reaction.



BSmB [16], GlcNAc-Mal [17], GlcNAc-OBn (O-benzyl-N-acetylglucosamine) [18] and GlcNAc-OMe (O-methyl-N-acetylglucosamine) [19] were chemically synthesized as described previously. BS–Fos (bacillithiol–fosfomycin) was enzymatically synthesized via FosB-catalysed S-conjugation of BSH with fosfomycin [16]. GlcNAc was purchased from Acros, and NDA (naphthalene-2,3-dialdehyde) was from Anaspec. All other reagents were purchased from Fisher Scientific or Sigma.

Expression and purification of B. anthracis BA1557, BA3524 and BA3888, and B. cereus BC1534 and BC3461

The cloning of BA1557 (BshB) was described by Parsonage et al. [10]. The codon-optimized genes for BA3524 and BA3888 were synthesized by GenScript and subcloned into pET28a (+) (Novagen). BC1534 in pET26b (+) [15] and BC3461 in pET24a (+) were generously provided by Dr Vassilis Bouriotis and Dr Vasiliki Fadouloglou (both at Department of Biology, University of Crete, Heraklion, Crete, Greece). All variants of BA1557 were prepared using the QuikChange® site-directed mutagenesis kit (Stratagene), and confirmed by DNA sequencing. All proteins used in the present study contained a C-terminal hexahistidine tag.

BA1557, BC1534 and all BA1557 variants were purified using the following general protocol. The plasmid was transformed into chemically competent Escherichia coli BL21(DE3) cells, which were plated on LB plates with 40 μg/ml kanamycin in a 37°C incubator overnight. For protein expression, 3 litres of LB broth supplemented with 40 μg/ml kanamycin were inoculated with freshly transformed cells and incubated at 37°C with shaking at 300 rev./min. Expression was induced upon addition of 5.8 mM lactose when the D600 reached 0.5, and the culture was incubated further at 15°C with shaking at 300 rev./min overnight before being harvested by centrifugation at 5000 g for 10 min at 4°C. The cells were stored at −20°C until further use. Cells were resuspended with 25 mM Tris/HCl (pH 8.0), 150 mM NaCl and 10% glycerol (buffer A), and then lysed using an EmulsiFlex-C5 high-pressure homogenizer. The cell debris was separated by centrifugation at 12000 g for 30 min, and the supernatant was cleared upon treatment with 1% (w/v) streptomycin sulfate followed by centrifugation at 12000 g for 20 min. After the treatment, the clear supernatant was loaded on to pre-equilibrated (buffer A) Zn2+- or Co2+-IMAC (immobilized metal-ion-affinity chromatography) columns (GE Healthcare). The column was washed with buffer A until the A280 reached baseline. Proteins associated with the resin were eluted in a step gradient of 5%, 15% and 50% of buffer B (25 mM Tris/HCl, pH 8.0, 150 mM NaCl, 300 mM imidazole and 10% glycerol). The fractions eluted from 50% of buffer B containing pure proteins were diluted five times with buffer C (25 mM Tris/HCl, pH 8.0 and 10% glycerol), and loaded on to a 5 ml Hitrap Q FF (GE Healthcare) column pre-equilibrated with buffer C. The desired protein was eluted with 0.5 M NaCl in the same buffer. Protein concentration was determined by the Bradford assay using BSA as the standard [20]. Protein aliquots were frozen with liquid nitrogen and stored at −80°C.

Because of the poor solubility experienced with BA3888 in pET vector expression, the gene was subcloned into the pBAD vector. The purification of pBAD-BA3888 and pET24a-BC3461 used the same procedure. The plasmids were transformed into E. coli BL21(DE3) cells, and cells were plated on LB agar with 100 μg/ml ampicillin or 40 μg/ml kanamycin. The inoculum was prepared with several colonies taken from a plate and transferred into 100 ml of LB broth with appropriate antibiotics. After 1 h of incubation at 37°C with shaking at 300 rev./min, the inoculum was added to 4 litres of LB broth with antibiotics and incubated at 37°C with shaking at 300 rev./min. Protein expression was induced at a D600 of 0.4 by the addition of 5.8 mM lactose or 20 mM arabinose (final concentrations). The culture was shaken at 37°C for another 4 h before being harvested by centrifugation. The cells were lysed as described above, and the soluble proteins were loaded on to a nickel-IMAC column pre-equilibrated with buffer A. The column was washed with buffer A, and proteins associated with the column were eluted with a step gradient (5%, 15%, 30%, 50% and 100%) of buffer B. Fractions containing deacetylase activity were pooled, diluted 6-fold with 25 mM Hepes (pH 8.0) and 10% glycerol (buffer D), and loaded on to a MonoQ column (GE Healthcare) equilibrated with buffer D. Proteins associated with the column were eluted in a 20 ml linear gradient from 0 to 0.75 M NaCl in buffer D. Pure protein fractions (by SDS/PAGE) containing activity were combined and aliquots of 20 μl were then frozen in liquid nitrogen for storage at −80°C. Within 10 days of the purification date, freshly thawed aliquots were used for kinetic experiments.

CD spectroscopy

CD spectra were obtained using an Aviv CD spectrometer (Model 215; AVIV Biomedical) with a 1-mm-pathlength quartz cuvette (Hellma Analytics) and a bandwidth of 1 nm. Protein samples were analysed at 5 μM in 10 mM phosphate buffer (pH 7.4) and scanned from 250 nm to 190 nm with 0.5 nm increments. The final spectrum of each sample is the average of ten scans.

Enzyme assays for N-deacetylation of GlcNAc-Mal, GlcNAc-OBn, GlcNAc-OMe and GlcNAc

The deacetylation activity was assayed by quantification of the primary amine product. In general, a 200 μl assay mixture containing 50 mM Mops (pH 7.4) and 0.1–20 μg of enzyme was pre-equilibrated at 37°C for 5 min. Reactions were initiated by the addition of substrate. Sample aliquots (20 μl) were taken at different time points and quenched with 20 μl of acetonitrile. The primary amine products (GlcN, GlcN-Mal, etc.) were then derivatized by addition of 200 μl of freshly prepared NDA-mix (0.4 M borate, pH 9, 2.5 mM KCN and 0.5 mM NDA). The solution mixture was allowed to react in the dark for 30 min before being read in a flat-bottom microplate (Costar) using a Synergy H1 plate reader (Biotek) with λex 390 nm and λem 480 nm. Initial velocity was determined from the slope of a plot of FU (fluorescence units) against time using four different time points over a period of up to 3 h. A glucosamine standard curve was used to convert the FU/min slopes into [product]/min; the fluorescence intensity of the GlcN–NDA product was identical with that of the GlcN-Mal–NDA. Initial deacetylation rates of GlcNAc-Mal were measured over the concentration range 0–3.5 mM. For determination of the steady-state parameters, results representing the means of triplicate values were fitted to the Michaelis–Menten equation using SigmaPlot 11.0.

Enzyme assays for amidase activity with BSmB, BSH and BS–Fos

Amidase activity of BSH was assayed by quantification of the mBBr (monobromobimane) derivative of cysteine (CysmB) produced during hydrolysis of BSmB. A representative assay contained 1.5–20 μg of enzyme in 200 μl of 50 mM Mops (pH 7.4) pre-warmed at 37°C for 5 min. The reaction was initiated by addition of BSmB. After incubation for various time intervals, reaction aliquots (10 μl) were quenched by the addition of 20 μl of acetonitrile followed by centrifugation (5 min). The supernatant (20 μl) was diluted 50-fold with 5 mM HCl before analysis by HPLC using procedures described previously [7].

The amidase activity towards BSH was assayed by quantification of cysteine formation by derivatization of the thiol with mBBr. The assay was performed by first equilibrating 40 μM of the enzyme in 50 mM Hepes buffer (pH 7.4) for 10 min at 37°C. The reactions were initiated upon addition of substrate (0.2–10 mM BSH) followed by incubation for 30 min at 37°C. Aliquots of 10 μl were collected at various time points and the reaction was quenched by heating at 80°C for 10 min. Reaction aliquots (4 μl) were then derivatized with 6 μl of 100 mM mBBr for 15 min at room temperature (25°C) to label the cysteine product and quenched upon addition of 0.25 μl of 5 M methanesulfonic acid and centrifuged at 10000 g for 5 min. Finally, the supernatant was diluted with 100 μl of 10 mM methanesulfonic acid and 50 μl of sample was analysed by HPLC as described previously [7]. The activities were calculated from a calibration curve using CysmB as the standard. The means of three duplicates were fitted into the Michaelis–Menten equation to determine the steady-state parameters.

The activity towards BS–Fos was assayed by quantifying cysteine–fosfomycin and GlcN-Mal using the AccQ Fluor reagent kit (Waters) to detect the primary amine products. A solution containing 40 μM of enzyme in 50 mM Hepes (pH 7.4) was pre-warmed at 37°C for 5 min and the reaction was started upon addition of 0.5 mM BS–Fos. After 30 min, reactions were quenched by heating the samples at 85°C for 10 min. The derivation of amine products with AccQ Fluor reagent was performed as recommended by the manufacture. The samples were assayed via HPLC as described previously [21].

Preparation of metal-free deacetylase and metal reconstitution

To prepare metal-free BA1557, Co2+-IMAC-purified protein (~75 μM) was incubated with 25 mM Tris/HCl (pH 7.5), 25 mM DETAPAC (diethylenetriaminepenta-acetic acid) and 10% glycerol on ice for 30 min. The protein solution was then dialysed three times over 2 litres of 25 mM Tris/HCl (pH 7.5) and 10% glycerol buffer at 4°C. The concentration of residual metal ion was determined to be less than 5% by ICP-AES (inductively coupled plasma atomic emission spectroscopy) (Tededyne Leeman Labs) analysis.

For the reconstitution with Zn2+, apo-BA1557 (10 μM) was incubated with different stoichiometric ratios of ZnSO4 on ice for 30 min. The solution was then dialysed against 1 litre of 25 mM Tris/HCl (pH 8.0) and 10% glycerol at 4°C to remove unbound Zn2+, and the protein metal content was measured by ICP-AES.

pH-dependent activity profiles

For the pH-dependence experiments, the following buffers were used: 50 mM Mes (pH 6.0–7.0); 50 mM Mops (pH 7.0–8.0); 50 mM bicine (pH 8.0–9.0); and 50 mM borate (pH 9.0–10.0). The standard assay protocol was conducted by equilibrating 200 μl of buffer containing 0.4 μg of deacetylase at 37°C. The deacetylation reaction was initiated by the addition of 0.25 mM GlcNAc-Mal. Samples (20 μl) were taken at intervals, and the reaction was terminated by addition of an equal volume of acetonitrile. Steady-state kinetic parameters Km, kcat and kcat/Km for deacetylase activity were determined by fitting initial velocities to the Michaelis–Menten equation. Eqn (1) was fitted to the pH rate profile, wherein V is the observed rate of the reaction, K is the pH-independent rate constant for GlcNAc or GlcNAc-Mal substrates, and Ka and Kb are the ionization constants of the acid and base species respectively [22]: Embedded Image (1)


Cross-functionality of deacetylases in BshB catalysis

Previous work described the ability of BA1557 to catalyse the hydrolysis of GlcNAc-Mal using an assay that involved HPLC analysis of GlcN-Mal product formation after derivatization with the amine-specific fluorophore (AccQ tag) [10]. In the present study, BA1557 and other BshB-like enzymes have been extensively characterized through the application of a faster assay procedure that bypasses the need to quantify derivatized reaction products by HPLC separation. This was achieved by developing a fluorescence microplate assay to detect GlcN-Mal formation following derivatization of the secondary amine with NDA [23]. This method is more cost-effective and faster than the commercial AccQ tag kit [10], and is ~50-fold more sensitive than methods involving a direct fluorescamine-based procedure [24]. Despite the aforementioned benefits of this method, NDA-derivatization of primary amines is not specific to the deacetylation product and background signal associated with other primary amines present in the reaction mixture is a recurring limitation of such amine-specific fluorescence-based assays [25]. Nevertheless, the high sensitivity of this method allowed us to optimize the use of synthetic substrates (Table 2 and Figure 3), many of which were only available in limited quantities.

Figure 3 Structures of substrates included in Table 2

Ph, phenyl.

View this table:
Table 2 Substrate specificity of B. anthracis and B. cereus N-deacetylases

All assays were performed in the presence of 5 mM substrate with 0.23–6.25 μM enzyme, except for GlcNAc-Mal, which was assayed in the presence of 0.5 mM substrate with 0.033–1.05 μM enzyme.

The activity of BA1557 with GlcNAc-Mal (19±1.1 μmol/min per mg), when tested using this method, was comparable with that obtained with the AccQ tag (21.5±0.2 μmol/min per mg) or fluorescamine procedures (19.8±1.0 μmol/min per mg). BA3888 also displayed comparable substrate kinetics with those of GlcNAc-Mal, thereby demonstrating the functional redundancy of BA1557 in the second step of BSH biosynthesis (Table 3 and Supplementary Figure S1A at The kinetic analysis of both enzymes fits well with the in vivo data, which demonstrated a 70% reduction in BSH levels in the B. anthracis ΔbshB strain [10]. As expected, the orthologous BC1534 (97% sequence identity) displays very similar kinetic behaviour in the BshB reaction (Table 3 and Supplementary Figure S1A), supporting the proposed role for this enzyme in the biosynthesis of BSH. The catalytic efficiencies (kcat/Km) of BA1557 and BC1534 with GlcNAc were four orders of magnitude lower than that determined against GlcNAc-Mal, whereas BA3888 showed no detectable activity (Table 3 and Supplementary Figure S1A). The calculated kcat/Km values for BA1557 and BC1534 with GlcNAc were substantially (six orders of magnitude) lower than the values reported previously for the B. cereus enzyme [15], but similar to the values reported for B. anthracis [10].

View this table:
Table 3 Kinetic parameters of B. anthracis and B. cereus N-deacetylases

ND, not determined.

Additional N-deacetylase candidates, BA3524 and BC3461, were investigated in the BshB reaction. The two enzymes are 95% identical in sequence and are likely to perform similar intracellular functions (Supplementary Figure S2B at Phylogenetic analysis shows that they constitute a separate branch of N-acetylhydrolases, different from those including BA1557 or BA3888 (Supplementary Figure S3 at In our hands, BC3461 showed limited activity with GlcNAc-Mal (Table 3), whereas BA3524 displayed no detectable activity. Far-UV CD spectra and ICP-AES analyses of these proteins indicated comparable secondary structure, and Zn2+ content with those of BA1557, BA3888 and BC1534 (Supplementary Figure S4 at This observation indicates that the lack of activity seems not to be attributed to protein misfolding or lack of metal cofactor resulting from heterologous E. coli expression.

Zn2+-dependent deacetylase activity

The Zn2+-binding site identified in the structure of BC1534 (BcZBP) displayed a similar arrangement to that found in the active sites of other Zn2+-dependent deacetylases [14]. Like the MshB structure, the BC1534 active site displays two histidine residues and one aspartate residue (His12, His113 and Asp15) providing a facial triad Zn2+ co-ordination [14,26] (Figure 4). All three metal-ion-co-ordinating residues are strictly conserved in the MshB and Mca sequences, in addition to the BshB and Bca candidate enzymes from B. anthracis and other BSH-producing species (Supplementary Figure S2). Hexahistidine-tagged recombinant BA1557, purified by Zn2+-affinity chromatography contains 2.55±0.04 Zn2+ cations per monomer. Treatment with 20 mM EDTA or EGTA had a minor effect on either metal content or enzyme activity (less than 20% decrease upon 1 h of incubation), whereas treatment with the stronger metal-chelating reagent DETAPAC removes nearly 98% of the bound Zn2+ and almost completely inactivates the enzyme. This indicates that Zn2+ is a tight-binding metal ion cofactor in this enzyme. The apo form of the enzyme displays an identical far-UV CD spectrum, indicating that neither DETAPAC treatment nor Zn2+ displacement causes major changes to secondary structure (Figure 5, inset). Zn2+ titration of the apo form completely reconstitutes the activity of the enzyme, correlating with a stoichiometry of one Zn2+/mol of enzyme (Figure 5). Further addition of >1 molar equivalent of Zn2+ causes inhibition. This Zn2+ activation/inhibition profile is very similar to that observed for UDP-3-O-(R-3-hydroxymyristoyl)-GlcNAc deacetylase (LpxC). It has been proposed that LpxC has two Zn2+-binding sites, one necessary for catalytic activity and a second site with allosteric inhibitory properties [27]. In addition to Zn2+, BA1557 can also be activated upon stoichiometric addition of other metal cations such as Ni2+, Co2+ and Fe3+ (Supplementary Table S1 at These results indicate that BA1557 shares similar profiles for non-specific metal-dependent deacetylation, comparable with the characterized cambialistic behaviours of MshB [28], LpxC [29] and the histone deacetylases [30].

Figure 4 Active-site region of B. cereus BC1534 (PDB code 2IXD)

The Zn2+ ion is co-ordinated with His12, Asp15 and His113.

Figure 5 Deacetylation activity of apo-BA1557 reconstituted with different stoichiometric ratios of Zn2+

The preparation and reconstitution of apo-BA1557 were described in the Materials and methods section. The assays were conducted with 0.2 μg of enzyme (0.035 μM) in the presence of 0.25 mM GlcNAc-Mal. The inset panel shows the far-UV CD spectrum of holo-BA1557 (continuous line) and apo-BA1557 (broken line).

General acid–base catalysis

The pH–activity profile for the BshB reaction with BA1557 presents a bell-shaped curve indicative of general acid–base catalysis [22]. The maximum catalytic activity at subsaturating concentrations of GlcNAc-Mal was reached at pH 7.8. The pH–activity curve fit is consistent with two ionization states with associated pKa values of 6.5 and 8.5. Interestingly, BA3888 and BC1534 displayed nearly identical pH–activity bell-shaped curves and similar ionization constants (Figure 6A and Table 4). The pH profile of Zn2+- or Co2+-reconstituted BA1557 displayed similar ionization events (Supplementary Figure S5 at Likewise, the pH-dependence on the GlcNAc deacetylation reaction displayed pKa values of 6.7 and 9.3 (Supplementary Figure S5). These results discard the participation of the malate carboxy group(s) of the substrate or the metal cofactor on either ionization events displayed for this reaction. The similar pH–activity profile of BshB enzymes suggests the participation of conserved residues in the general acid–base catalytic mechanism of the BshB reaction.

Figure 6 pH–activity profiles of deacetylase enzymes

(A) Relative activity (V/K) of BA1557 (◆), BC1534 (▼), BA3888 (■) and BC3461 (▲, inset). (B) Relative activity (V/K) of BA1557 WT (◆), H110A (●) and D14A (▲). Assays were measured with 0.25 mM GlcNAc-Mal under the pH range 5.97–9.56. All of the pH curves exhibit a bell shape except for D14A which loses the basic limb. The pKa values were determined by fitting the curve into eqn (1) in the text.

View this table:
Table 4 pH-dependence of different enzymes/mutants for N-deacetylation of GlcNAc-Mal

Close inspection of the protein environment surrounding the Zn2+ cation at the active site of BC1534 (Figure 4) initially suggested His110 and Asp14 as possible candidates for the general acid and general base respectively. Both residues are strictly conserved in all members of this family of enzymes including the BshB, MshB and Mca sequences (Supplementary Figure S2). ND1 of His110 and OD2 of Asp14 are located on opposite sides of the acetate group that interacts with Zn2+, approximately 4 Å (1 Å=0.1 nm) from the metal centre. Individual alanine substitutions of His110 and Asp14 in BA1557 resulted in decreased activity levels of 4-fold and 3000-fold respectively (Figure 6B and Table 5). The pH–activity profile of H110A BA1557 retained a bell-shaped curve, ruling out the possible participation of this residue as the acid or base in this mechanism. Interestingly, in the structure of BC1534, OD2 of Asp112 is located within hydrogen-bonding distance from both NE2 of His110 and NE2 of His12, which is a known ligand for the Zn2+ (Figure 4). In BC1534, the D112A substitution completely eliminates the GlcNAc deacetylase activity of this enzyme [15]. Unexpectedly, the D14A BA1557 variant retained the ionization event associated with pKa1, but not that for pKa2, suggesting its role as a general acid in the reaction. Although the identity of the general base remains unknown, the pH–activity profile of the D14A variant eliminates the possibility of any dual involvement of this residue, in both ionization events of this reaction mechanism.

View this table:
Table 5 Steady-state kinetic parameters of BA1557 variants for N-deacetylation of GlcNAc-Mal

Gatekeepers of substrate binding and hydrolysis

The very low catalytic efficiency of the enzyme with GlcNAc compared with GlcNAc-Mal prompted the investigation of substrate analogues where the malate aglycone was replaced by an uncharged methyl or benzyl motif. Table 2 shows that all substrates lacking malate at this position were at least three orders of magnitude less effective as substrates than GlcNAc-Mal. An in silico auto-docking approach was used to explore potential binding modes of GlcNAc-Mal with BC1534 [31]. The results from these docking experiments pointed to two arginine residues, Arg53 and Arg109, as providing potential electrostatic interactions with the substrate (Supplementary Figure S6 at These residues are located on opposite sides of the active site, ~10 Å from each other and ~5 Å from the C1-position of bound acetate, whereas the distances between the Zn2+ to the ω-N of Arg53 and Arg109 were 5.72 Å and 8.03 Å respectively. Interestingly, structural and functional analyses of malate dehydrogenases have shown that two arginine residues (~12 Å apart) are involved in substrate binding providing electrostatic interactions with both carboxy groups of the L-malate substrate [32].

Compared with WT BA1557, an R53A mutation completely eliminated the BshB activity, whereas a more conservative R53K substitution displayed 103-fold lower activity for GlcNAc-Mal and no detectable activity with GlcNAc (Table 5 and Supplementary Figure S7 at Both substitutions did not impair the secondary structure of this enzyme; the far-UV CD spectrum was nearly identical with that of the WT (Supplementary Figure S4). Sequence alignment of several deacetylases including MshB and Mca indicate that Arg53 is strictly conserved. It has been suggested that Arg68 and His144 of MshB (equivalent to Arg53 and His110 of BC1534 and BA1557) participate in electrostatic interactions with a sugar hydroxy group of the substrate [26]. Our results with the Arg53 mutants provide support for the involvement of this residue in substrate binding. Arg109, however, is only conserved among a group of deacetylases limited to the BshB1 candidates; including BA1557, BC1534 and YpjG (Supplementary Figure S2). The BA1557 R109K variant showed a 6-fold decrease in kcat for GlcNAc-Mal (Table 5 and Supplementary Figure S7). Although this substitution impaired the catalytic efficiency of BA1557 with GlcNAc-Mal, the R109K variant showed a modest improvement in reactivity against GlcNAc. This variant enzyme had a lower Km for GlcNAc (18.6 mM) and similar kcat (2.9×10−3 s−1) when compared with the WT constants of 57 mM and 3.3×10−3 s−1 respectively (Table 5).

Deacetylase enzymes participating in detoxification pathways

BSH is already known to be implicated in some aspect of drug detoxification as exemplified by BST (bacillithiol-S-transferase) (FosB) [16]. A second class of BSTs have also recently been identified, but their target electrophilic substrates are not yet known [13]. On the basis of what has been observed previously in MSH metabolism, enzymes catalysing the BshB reaction (the hydrolysis of GlcNAc-Mal to yield GlcN-Mal) could also potentially be capable of catalysing the hydrolysis of bacillithiol S-conjugates produced in BSH-dependent pathways. The proposed Bca enzyme is thought to be involved in the subsequent step, thus utilizing the products of the BST reaction as substrates. Because of the similarities in mechanism between the BshB and Bca reactions, we have determined the reactivity of BA1557 and BA3888 against BSmB, BSH and BS–Fos. The activity of BA3888 in the hydrolysis of the CysmB side chain of BSmB was 200-fold higher than that of BA1557; however, at 5 mM BSmB the turnover rate of BA3888 was only 1.84 min−1 (Table 2). Kinetic analysis of BA1557 and BA3888 with BSH also showed low rates of BSH degradation by formation of cysteine and GlcN, indicating that these enzymes do not contribute to the pool of reduced cysteine levels. Interestingly, the Km of BA1557 for BSH (0.48 mM) is in the same range as cellular BSH concentrations, suggesting that BSH could act as a feedback inhibitor of this enzyme (Table 3). On the other hand, the BA3888 Km for BSH was not determined since no substrate saturation was reached up to 5 mM BSH. In our hands, neither BA1557 nor BA3888 showed any detectable activity against BS–Fos. The remarkable selectivity of this group of enzymes for a restricted subset of substrates highlights the importance for in vivo screening of candidate compounds in strains lacking these enzymes.


In the present study, we performed comparative kinetic analysis of the B. anthracis and B. cereus deacetylases in performing BshB and Bca reactions to provide biochemical evidence for the cross-functionality of these enzymes in the biosynthesis of BSH. This conclusion supports in vivo demonstrations that inactivation of bshB in B. anthracis and bshB1 (ypjG) in B. subtilis decreased, but did not eliminate, the levels of BSH [8]. Kinetic analyses of the BshB reactions with BA1557 and BA3888 show that they display equivalent catalytic efficiencies towards GlcNAc-Mal. However, BA3888 showed a nearly 200-fold higher activity with BSmB compared with BA1557 (Table 2). The reactivity of this enzyme towards larger bacillithiol conjugates provides experimental evidence supporting the role of BA3888 as a catalyst for the Bca reaction. However, it is worth noting that the substrate kinetics for BSmB with BA3888 are still poor and it remains to be seen what the physiologically relevant bacillithiol conjugate substrate(s) is(are) for this enzyme. With BA1557, the very weak amidase activity observed with BSH and a Km value comparable with its cellular concentration indicates that BSH functions as a feedback inhibitor of this enzyme. However, such feedback inhibition is redundant in the presence of a functional BA3888 enzyme, which can substitute for the GlcNAc-Mal activity of BA1557.

Interestingly, in S. aureus, only one BshB-like enzyme has been identified [8]. Among the enzymes described in the present paper, BA3888 is the closest orthologue of S. aureus BshB (70/49% identity/similarity) supporting the notion that a single enzyme in S. aureus could fulfil both BshB and Bca functions, although this remains to be proved. Amino acid sequence alignment indicates the presence of a ten-residue insertion (GDPFFANRET in BA3888) in Bca/BshB2 sequences, including BA3888 and S. aureus BshB (Supplementary Figure S2A). Although the structural location of sequence is not known, adjacent residues in the BC1534 structure constitute a short loop surrounding the active site. It is possible that this decameric insertion sequence could expand the flexibility of the active site to accommodate potentially larger substrates as in the case of Bca. Nevertheless, the low reactivity of these enzymes against BSmB/BS–Fos suggests the possible occurrence of alternative pathways for detoxification of such adducts and/or the involvement of this group of enzymes in serving as amidases to a specific group of bacillithiol adducts.

Compared with GlcNAc-Mal, the low activity values against GlcNAc, GlcNAc-OMe and GlcNAc-OBn reveals the high degree of specificity for the malate aglycone. Our results from site-directed mutagenesis pointed to BA1557 Arg109 as one potential gatekeeper in controlling this specificity or stabilizing a reaction intermediate. Although this model is favoured for BshB enzymes containing arginine residues at positions equivalent to Arg109, this scheme may not apply to all deacetylase enzymes capable of catalysing the BshB reaction (e.g. BA3888 and BC3461 contain valine at the equivalent position). Nonetheless, all four enzymes tested in the present study showed a high degree of selectivity for GlcNAc-Mal. Controlled substrate binding and hydrolysis may serve as a molecular strategy used by this group of enzymes to restrict the repertoire of physiological substrates.

Functional assignment of residues surrounding Zn2+ at the active site confirmed the involvement of Asp14 with the second ionization event in the deacetylation mechanism. The abnormally high pKa of 8.5 is 4 units above the pKa of free aspartate. Large shifts in aspartate pKa (>4 units) have been reported for the active-site aspartate residues of human thioredoxin and bacteriorhodopsin [33]. In both cases, the active-site environment surrounding the aspartate side chain favours the protonated acid form during turnover. Studies on the MshB pH–activity profile showed that the rate of deacetylation was also dependent on two ionization events (pKa1=7.4 and pKa2=10.5) [34]. The first ionization event was attributed to Asp15 which was proposed to be the general base on this catalytic mechanism. Whereas there is an agreement for the role of Asp15 as a general base in the MshB mechanism, the identity of the general acid remains controversial. Solvent isotope effect and site-directed mutagenesis experiments suggest the participation of His144 as the general acid [34]; however, pH-dependence studies on H144A MshB [34] and recent structural analysis of this enzyme [35] do not support this model. A previous proposal has evoked the dual role of Asp15 as a single general acid and base [26,35]. In this later model, it would be expected that the pKa associated with the deprotonation of Asp15 (pKa2) would be higher than 7.4 (pKa1). In the structure of BA1534 [14], Asp14 and His110 occupy positions equivalent to Asp15 and His144 of MshB [26,35]. Interestingly, in the present study, we identified the pKa associated with Asp14 of BA1557 (pKa2 of 8.6, Table 3) to be only 1.2 pH units higher than the pKa1 of the MshB reaction. Collectively, these results support a mechanism for the BshB reaction involving Asp14 as a general acid during the protonation of the nascent amino group on GlcN-Mal product.

The identity of the residue associated with the first ionization event awaits further investigation. Results of the present study rule out the possible participation of His110 and Asp14 in performing this function. The pH–activity curve of BA1557 using GlcNAc as substrate showed an identical profile eliminating the potential involvement of the malate carboxy groups of the substrate in the first ionization event (Supplementary Figure S5). Alternatively, the activation of a water molecule by the active-site Zn2+, as observed with carbonic anhydrase, or the participation of the Zn2+-co-ordinating Asp15 in abstracting the proton from this water, provide possible candidates for the general base in this mechanism. In any event, the conservation of amino acids within the active site imposes a controlled substrate specificity (along with conserved kinetic behaviour) of this group of deacetylases and opens the door for further investigation into their roles in BSH biosynthesis and BSH-mediated xenobiotic detoxification.


Zhong Fang, Alexandra Roberts, Karissa Weidman and Sunil Sharma performed experiments. Al Claiborne and Patricia Dos Santos conceived the idea, and provided overall direction. Zhong Fang, Christopher Hamilton and Patricia Dos Santos planned experiments, analysed data and wrote the paper. All authors read and approved the content of the paper.


Financial support for this work has been provided by a Multidisciplinary Research Grant from the North Carolina Biotechnology Center [grant number M2011-MRG-1116 (to A.C. and P.D.S.)] and the Biotechnology and Biological Sciences Research Council [grant number BB/H013504/1 (to C.J.H)].


We thank Vassilis Bouriotis and Vasiliki E. Fadouloglou for the gift of expression plasmids containing BC1534 and BC3461.

Abbreviations: Bca, bacillithiol-S-conjugate amidase; BS–Fos, bacillithiol–fosfomycin; BSH, bacillithiol; BSmB, bacillithiol-S-bimane; CysmB, cysteine monobromobimane derivative; DETAPAC, diethylenetriaminepenta-acetic acid; FU, fluorescence unit(s); GlcNAc-Mal, N-acetylglucosamine malate; GlcNAc-OBn, O-benzyl-N-acetylglucosamine; GlcNAc-OMe, O-methyl-N-acetylglucosamine; GlcN-Mal, glucosamine malate; ICP-AES, inductively coupled plasma atomic emission spectroscopy; IMAC, immobilized metal-ion-affinity chromatography; mBBr, monobromobimane; Mca, mycothiol-S-conjugate amidase; MSH, mycothiol; NDA, naphthalene-2,3-dialdehyde; WT, wild-type


View Abstract