Structure and mechanism of an inverting alkylsulfatase from Pseudomonas sp. DSM6611 specific for secondary alkyl sulfates

Authors


Correspondence

U. Wagner, University of Graz, Institute of Molecular Biosciences, Humboldtstrasse 50/III, A-8010 Graz, Austria

Fax: +43 316 3805431

Tel: +43 316 3805431

E-mail: ulrike.wagner@uni-graz.at

P. Macheroux, Graz University of Technology, Institute of Biochemistry, Petersgasse 12, A-8010 Graz, Austria

Fax: +43 316 8736952

Tel: +43 316 8736450

E-mail: peter.macheroux@tugraz.at

Abstract

A highly enantioselective and stereoselective secondary alkylsulfatase from Pseudomonas sp. DSM6611 (Pisa1) was heterologously expressed in Escherichia coli BL21, and purified to homogeneity for kinetic and structural studies. Structure determination of Pisa1 by X-ray crystallography showed that the protein belongs to the family of metallo-β-lactamases with a conserved binuclear Zn2+ cluster in the active site. In contrast to a closely related alkylsulfatase from Pseudomonas aeruginosa (SdsA1), Pisa1 showed a preference for secondary rather than primary alkyl sulfates, and enantioselectively hydrolyzed the (R)-enantiomer of rac-2-octyl sulfate, yielding (S)-2-octanol with inversion of absolute configuration as a result of C–O bond cleavage. In order to elucidate the mechanism of inverting sulfate ester hydrolysis, for which no counterpart in chemical catalysis exists, we designed variants of Pisa1 guided by three-dimensional structure and docking experiments. In the course of these studies, we identified an invariant histidine (His317) near the sulfate-binding site as the general acid for crucial protonation of the sulfate leaving group. Additionally, amino acid replacements in the alkyl chain-binding pocket generated an enzyme variant that lost its stereoselectivity towards rac-2-octyl sulfate. These findings are discussed in light of the potential use of this enzyme family for applications in biocatalysis.

Database

The atomic coordinates and structural factors have been deposited in the Protein Data Bank under the accession codes 2YHE (wild type, crystal form I), 4AV7 (double variant Ser233→Tyr/Ph250→Gly) and 4AXH (wild type, crystal form II)

Structured digital abstract

Abbreviations
E

enantiomeric ratio

ITC

isothermal titration calorimetry

PDB

Protein Data Bank

Pisa1

Pseudomonas inverting secondary alkylsulfatase 1

SdsA1

alkylsulfatase from Pseudomonas aeruginosa

Introduction

Sulfatases are a heterogeneous group of enzymes that catalyze the cleavage of the sulfate ester bond, yielding the corresponding alcohol and hydrogen sulfate [1-4]. In contrast to the majority of hydrolases, such as proteases and carboxyl ester hydrolases, which do not alter the stereochemistry of the substrate during catalysis [5], the stereochemical course of sulfate ester hydrolysis can be controlled by choice of the type of enzyme: Depending on the mechanism of the enzyme, cleavage of the S–O or the C–O bond of a secondary alkyl sulfate causes retention or inversion of the stereogenic carbon atom, respectively (Scheme 1).

Scheme 1.

Stereochemical course of retaining and inverting alkylsulfatases.

Previously, sulfatases were classified according to their preferred substrate type into arylsulfatases, carbohydrate sulfatases, and alkylsulfatases [3]. More recently, sulfatases were reclassified on the basis of mechanistic considerations [6] into: (a) formylglycine-dependent arylsulfatases and carbohydrate sulfatases acting on sulfated carbohydrates and on sulfated hormones [7]; (b) sulfatases that oxidatively cleave a sulfate ester at the expense of α-ketoglutarate as electron acceptor, to yield an aldehyde and inorganic sulfate [8]; and (c) sulfatases belonging to the family of metallo-β-lactamases [6]. In the case of arylsulfatases and carbohydrate sulfatases, a hydrated α-formylglycine active site nucleophile attacks the sulfur atom of the sulfate ester, resulting in retention of the absolute configuration at the carbon. In the course of the reaction of α-ketoglutarate-dependent sulfatases, the stereogenic carbon center is destroyed. The third class of sulfatases is represented by sodium dodecyl sulfatase (SdsA1) produced by Pseudomonas aeruginosa, which enables the bacterium to survive under bacteriocidal conditions by hydrolyzing SDS.

On the basis of the crystal structure of SdsA1, it was initially proposed that sulfate ester cleavage occurs by nucleophilic attack of a water molecule, which is activated by a highly conserved binuclear Zn2+ cluster. As the substrate is an achiral primary sulfate ester, the stereochemical course of SdsA1 hydrolysis is not ‘visible’, and thus was not investigated in detail. On the basis of the short distance of the presumed nucleophile (Water2) from the sulfur atom of 1-decylsulfonate (used as a substrate surrogate inhibitor), S–O bond cleavage was assumed, implying retention at the carbon [6]. However, 18O-labeling studies later revealed that C–O bond cleavage causes inversion at the carbon [9]. SdsA1 was unreactive on sulfated sugars and aryl sulfates. Secondary alkyl sulfate esters were not investigated as potential substrates [6].

The existence of alkylsulfatases acting on secondary sulfate esters was derived from the chance observation of bacterial contamination in a commercial shampoo preparation [10]. During subsequent studies, a range of bacterial strains possessing secondary alkylsulfatase activity were isolated from sewage sludge, and two of them were investigated in detail [11]: depending on the culture conditions, Pseudomonas C12B (NCIMB 11753 = ATCC 43648) produced up to three alkylsulfatases, and Comamonas terrigena (NCIMB 8193) formed two secondary alkylsulfatases. The first hints about the enantiopreference of these enzymes came from the observation that rac-sec-alkyl sulfates were rapidly hydrolyzed up to 50% conversion, whereas further reaction took place at a much lower rate [12-14]. The stereochemical course of these enzymes with respect to retention versus inversion was studied with enantioenriched substrates [13]. Subsequent studies revealed Rhodococcus ruber DSM44541 as a promising source of an inverting secondary alkylsulfatase [15, 16]. Although the enzyme responsible for the stereoselective hydrolysis of secondary sulfate esters could be purified to a certain extent, no information concerning its amino acid sequence could be obtained, owing to the instability of the protein. In a search for a more stable alternative, extended screening for secondary alkylsulfatase activity was performed among pseudomonads and close relatives thereof, and revealed Pseudomonas sp. DSM6611 as the most promising candidate [17]. This strain was isolated from soil for its ability to degrade halogenated aromatic compounds, such as 4-fluorobenzoate [18]. A secondary alkylsulfatase was isolated and sequenced by peptide mass fingerprinting. In combination with the full genomic sequence of Pseudomonas sp. DSM6611, the predicted ORF was determined, and the gene was cloned and heterologously expressed in Escherichia coli BL21 [9, 19]. The protein, termed Pseudomonas inverting secondary alkylsulfatase 1 (Pisa1), showed the desired catalytic properties: Hydrolysis of (R)-2-octyl sulfate proceeded quantitatively through complete inversion of configuration, yielding (S)-2-octanol. The stereochemical course of inversion was independently verified by hydrolysis of unlabeled (R)-2-octyl sulfate in 18O-labeled buffer (label > 97%), which showed complete incorporation of the 18O-label in the product alcohol. The hydrolysis of rac-2-octyl sulfate ceased at 50% conversion to furnish (S)-2-octanol and unreacted (S)-2-octyl sulfate, indicating perfect enantioselectivity [enantiomeric ratio (E) > 200].

In a previous study, we investigated the properties of Pisa1 and SdsA1 with regard to the stereochemical course of the sulfate ester cleavage and substrate specificity [9]. This study revealed that both alkylsulfatases cleave the C–O bond, resulting in inversion at the carbon. On the other hand, it was demonstrated that the two enzymes have different substrate specificities for primary and secondary alkyl sulfates: whereas Pisa1 prefers sec-octyl sulfate, SdsA1 shows a 140-fold higher catalytic efficiency for 1-octyl sulfate. These findings ignited our interest in the structure of Pisa1 as a basis to rationalize the fact that the cleavage mechanism has been conserved, while at the same time the enzymes have acquired different substrate specificities. Here, we report the elucidation of the Pisa1 structure by X-ray crystallography combined with a site-directed mutagenesis study to identify residues in the active site of the enzyme that are crucial for catalysis.

Results

Structure description

The overall structures of SdsA1 and Pisa1 show a high degree of similarity. The rmsd values for the Ca atoms of SdsA1 with sulfate in the active site [Protein Data Bank (PDB) code: 2CG2] and orthorhombic crystals of wild-type Pisa1 (from here on called wild-type form-I) calculated with lsqman [20, 21] showed a maximum and an average of 15.01 and 1.5 Å, respectively. Figure 1 shows a comparison between the structure of Pisa1 (PDB code: 2YHE) and that of SdsA1 (PDB code: 2CG2).

Figure 1.

Comparison of overall structures of Pisa1 and SdsA1. Cα rmsd values calculated with lsqman are represented by the color and the thickness of the worm. The overall rmsd is 1.5 Å. The picture was created with chimera [40].

As described for the SdsA1 structure, Pisa1 forms a dimer. Each protomer consists of the same three domains as in SdsA1: the N-terminal, catalytic, αββα-sandwich domain; an α-helical dimerization domain; and an α,β-mixed C-terminal domain. In contrast to SdsA1, where the dimer is formed via a crystallographic two-fold axis, the active Pisa1 unit consists of two molecules related by noncrystallographic symmetry. In wild-type form-I and the double variant Ser233→Tyr/Phe250→Gly structures, these dimers form an imperfect 61.2-helix parallel to the crystallographic c-axis, with rotation angles between 65° and 68° and a translation of approximately 0.2 fractional coordinates. Whereas all six molecules of the wild-type form-I structure are adequately defined, both molecules of wild-type form-II (trigonal crystals) have poorly defined regions: although molecule A has a very well-defined N-terminal part ranging from residues 30 to 531, the density for residues 532–661 is not very well defined, but is nevertheless interpretable, at least for the main chain atoms. The density for molecule B is almost not interpretable for residues 31–207, 237–306, and 402–417, and the loop region linking the N-terminal and the C-terminal domains (527–545). In contrast to the wild-type form-I structure, no sulfate was found in the ligand-binding site of wild-type form-II molecule A. The electron density for the double variant structure Ser233→Tyr/Phe250→Gly is weak, but, especially around the active site residues, the density is well defined, thus making it useful for docking experiments. Figure 2 shows a B-factor representation of the dimer for the three crystal structures. Only those parts forming the interface between the dimers are well ordered in both molecules. Despite the lower overall quality of the structure obtained from wild-type form-II crystals, the loop regions are well defined, in contrast to those of the structure derived from wild-type form-I. Figure 3 shows a representative section of the well-defined parts of the electron density of Pisa1 wild-type form-I, wild-type form-II, and the double mutant Ser233→Tyr/Phe250→Gly. A summary of the crystallographic data and refinement statistics is provided in Table 1 (see also Experimental procedures).

Table 1. Data collection and refinement statistics
 Pisa1 wild-type form-IPisa1 Ser233→Tyr/Phe250→GlyPisa1 wild-type form-II
RemoteInflectionPeak
  1. a Numbers in parentheses refer to high-resolution shells. bRsym = Σ|(I − <I>)/Σ(I), where I is the observed intensity. cRcryst = (Σ||Fobs| − |Fcalc||)/(Σ|Fobs|). d 5% of reflections were randomly selected for determination of the free R-factor, prior to any refinement. e Temperature factors averaged for all atoms.

Space groupP212121P212121 P3121 
Cell dimensions (Å)a = 147.69a = 75.2 a = 142.0 
 b = 54.53b = 202.0 c = 119.7 
 c = 94.90c = 248.5   
Wavelength (Å)1.270301.00441.180761.280991.28031
Resolution (Å)19.9–2.745.7–3.019.6–2.710.0–2.710.0–2.7
Unique reflections73 894 (10 637)a62 063 (9310)a36 072 (5553)a33 729 (5172)a33 799 (5225)a
Overall completeness (%)99.8 (100)a78.4 (79.4)a93.3 (100)a92.6 (98.1)a92.7 (98.9)a
Rsymb (%)17.0 (80.1)a7.4 (54.4)a8.5 (66.1)a9.6 (57.8)a10.0 (60.8)a
Multiplicity10.1 (9.3)a3.0 (2.7)a7.4 (7.5)a7.3 (6.9)a7.3 (6.9)a
I/(σI)9.1 (1.7)a10.8 (2.0)a18.8 (3.1)a15.4 (3.2)a15.9 (3.2)a
Wilson plot B-factor (Å2)53.4384.3065.3638.1284.12
R cryst c 18.3 (26.0)a18.7 (32.5)a22.1 (26.6)a  
Rfreed (%)23.0 (33.1)a25.6 (38.8)a28.1 (32.2)a  
Protein atoms29 82529 6479900  
Water molecules1166318136  
Heterogen atoms42429  
rmsd bond lengths (Å)0.0140.0100.010  
rmsd bond angles (°)1.8341.3161.23  
Mean B-factore2)47.7486.8377.70  
Figure 2.

Worm-style plots of the active dimers created with chimera [40]. The residual B-factor (defined as Bi = 2Ui2 where Ui2 is the mean square displacement of atom i) reveals the quality of the structures. For a better understanding, molecule A of each structure is marked by a transparent surface, and the N-terminus and C-terminus are denoted. The view is approximately down the two-fold axis.

Figure 3.

Electron density (drawn with a σ-level of 1.6) representation of the β-strands A123–128, A133–139, A144–150 and A173–178 for all three crystal structures. Pictures were created with pymol.

Comparison of the zinc-binding sites

Both alkylsulfatases, SdsA1 and Pisa1, have conserved catalytic machineries consisting of a binuclear zinc center. Comparison of the zinc-binding sites of SdsA1 and Pisa1 (PDB codes: 2CG2 and 2YHE, respectively) reveals a high degree of similarity, despite the different substrate specificities and regiospecificities of the enzymes: all of the interaction partners of the zinc centers are identical, with the exception of the bridging acidic amino acid. This position is occupied by aspartic acid in Pisa1, instead of a glutamic acid as in SdsA1 (Fig. 4).

Figure 4.

Schematic comparison of the zinc-binding sites of Pisa1 (left) and SdsA1 (right) containing bound sulfate, created with ligplot [41].

Comparison of the substrate-binding sites

The sulfate group in SdsA1 is bound through seven hydrogen bonds; however, two of these entail the guanidinium nitrogens of Arg323 with ~ 50% occupancy. In the case of Pisa1, the sulfate group appears to be more tightly bound by eight hydrogen bonds (Fig. 5). In both structures, an arginine claw is involved in sulfate binding (Fig. 5). Wild-type form-II of Pisa1 shows no density at the sulfate position. Interestingly, the arginine claw is in an ‘open’ conformation in this structure. In SdsA1, this arginine (Arg312) adopts alternative conformations, with the side chain oriented towards the sulfate or pointing away from the sulfate group with approximately equal occupancy. In SdsA1, this alternative conformation of Arg312 is found in the sulfate-free as well as in the sulfate-bound form.

Figure 5.

Schematic comparison of the sulfate-binding sites of Pisa1 (left) and SdsA1 (right) created with ligplot [41]. The arginines involved in the arginine claw are displayed in blue.

A water molecule near the sulfate group was postulated to be the nucleophile for the hydrolysis in SdsA1 [6]. This assumption does not hold any longer, as SdsA1 also works under inversion [9]. A water molecule in the correct position to perform a nucleophile attack at the chiral carbon could not be found in any of the Pisa1 structures.

For a better understanding of the substrate's binding pocket, we performed soaking and cocrystallization experiments with (R)-substrate, (S)-substrate and inhibitor (sodium 1-nonanesulfonate) with the actual studied crystals of wild-type Pisa1 and the double variant. Unfortunately, the resulting density showed no incorporation of substrate/inhibitor. A possible reason might be that a sulfate, presumably already incorporated during expression, binds very tightly, and hence might be difficult to be displaced by substrates or inhibitors under soaking conditions. We therefore crystallized Pisa1 expressed without addition of ZnSO4, but the crystals obtained from this expression batch did not diffract at all. Therefore, we resorted to docking experiments. The results could not fully rationalize the preference of wild-type Pisa1 for (R)-enantiomers. However, they highlight an interesting difference between the wild-type enzyme and the Ser233→Tyr/Phe250→Gly double variant: as shown in Fig. 6, the hydrophobic side chain of the substrates points towards the entrance of the reaction cavity in the wild-type enzyme, whereas, in the double variant, the hydrophobic tail of the substrate also points into the hydrophobic hole created by the Phe250→Gly exchange. Owing to the narrow entrance into this binding pocket, it will not be equally populated as the ‘normal’ one. It is conceivable that the (S)-substrate can utilize this additional binding site and thus make it accessible for nucleophile attack.

Figure 6.

Cuts through the active sites of wild-type Pisa1 (top) with a docked (R)-2-octyl sulfate (green) and the Pisa1 Ser233→Tyr/Phe250→Gly double variant (bottom) with both docked enantiomers; R is colored in green, and S in magenta. The replaced residues are displayed, as well as His317, identified as the general acid. The Phe250→Gly mutation opens the possibility for alternative binding of the side chain, thereby allowing both enantiomers to react.

Kinetic characterization of wild-type Pisa1 and variants

Kinetic analysis of the alkyl sulfate ester cleavage is not possible with simple spectrophotometric assays, and hence we employed microcalorimetry to obtain kinetic parameters for SdsA1 and Pisa1. To that end, we first determined the reaction enthalpy for sulfate ester cleavage (Fig. 7), and then measured the rate of enzymatic substrate turnover as a function of substrate concentration (Fig. 8). SdsA1 was previously described as a primary alkylsulfatase with a preference for SDS; however, no kinetic parameters were reported for this or other substrates. Because Pisa1 has a preference for shorter alkyl chains, we compared the two alkylsulfatases by using 1-octyl sulfate and (R)-2-octyl sulfate. As shown in Table 2, SdsA1 is clearly more specific for 1-octyl sulfate, as it has a 6.5-fold higher kcat and a 22-fold lower Km than Pisa1, resulting in a 141-fold higher kcat/Km for the primary model substrate. In contrast, Pisa1 exhibited a 194-fold higher kcat/Km for the secondary model substrate, leading to the classification of the two enzymes as primary (SdsA1) and secondary (Pisa1) alkylsulfatases.

Table 2. Kinetic parameters of Pisa1, its variants and SdsA1 for the hydrolysis of alkyl sulfates
EnzymeSubstratekcat (min−1)Kmm)KI (mm)kcat/Km (m−1·s−1)
  1. a No saturation observed up to [S] = 10 mm. b At least ~ 2 mm.

Pisa1(R)-2-octyl sulfate262 ± 62151 ± 243.8 ± 1.128 900
SdsA1(R)-2-octyl sulfate5.4 ± 0.5605 ± 1471.4 ± 0.2149
Ser233→Tyr(R)-2-octyl sulfate104 ± 3.599 ± 419 ± 617 500
Phe250→Gly(R)-2-octyl sulfate1.4 ± 0.24280 ± 28905
Ser233→Tyr/Phe250→Gly(R)-2-octyl sulfate84 ± 11520 ± 30919
Ser233→Tyr/Phe250→Gly(S)-2-octyl sulfate23 ± 68420 ± 45018.6 ± 4.246
Ser233→Phe/Phe250→Ser(R)-2-octyl sulfate68 ± 7115 ± 112.7 ± 0.59850
His317→Ala(R)-2-octyl sulfate0.2 ± 0.02240 ± 11114
Tyr417→His(R)-2-octyl sulfate1.1 ± 0.8 a ND
Tyr417→Phe(R)-2-octyl sulfate2.3 ± 0.5 b ND
Tyr417→Asp(R)-2-octyl sulfateNo conversion (GC)No conversion (GC)No conversion (GC)
SdsA11-Octyl sulfate337 ± 10630 ± 140.33 ± 0.07187 000
Pisa11-Octyl sulfate52 ± 5651 ± 250.5 ± 0.081330
Figure 7.

Determination of ΔH for the hydrolysis of (R)-2-octyl sulfate by ITC. Pisa1 (125 nm) was titrated with (R)-2-octyl sulfate (1.5 mm, 3 × 20 μL, duration of 39.9 s, spacing time of 600 s) in 100 mm Tris/HCl (pH 8.2) buffer at 25 °C, yielding a ΔH of −16.758 kcal·mol−1 (mean value of six independent measurements).

Figure 8.

Representative example of the determination of the kinetic parameters by ITC. The measurement shows the titration of 20 nm Pisa1 with (R)-2-octyl sulfate (7 mm, 24 × 10 μL, duration of 20 s, spacing time of 300 s) in 100 mm Tris/HCl (pH 8.2) buffer at 25 °C. (A) Raw data obtained by ITC measurement. (B) Determination of kcat, Km and KI by nonlinear least-squares fitting.

We recently demonstrated that alkyl sulfate ester cleavage by Pisa1 results in the inverted absolute stereoconfiguration at the stereogenic carbon center; that is, (R)-2-octyl sulfate is hydrolyzed to (S)-2-octanol. This stereochemical course of the reaction implies that the hydroxide attacks the stereogenic carbon rather than the sulfur atom of the sulfate group, leading to C–O bond cleavage. Such a reaction mechanism generates a negative charge on the oxygen, and requires synchronous protonation by an active site amino acid. In order to identify this putative general acid, we replaced His317 (His317→Ala) and Tyr417 (Tyr417→His, Tyr417→Phe, and Tyr417→Asp). The His317→Ala variant showed a 1300-fold decrease in kcat, whereas Km was only slightly increased from 151 to 240 μm (kcat/Km = 14 m−1·s−1). All of the Tyr417 variants showed severely impaired catalytic parameters: the Tyr417→Asp variant was found to be inactive, whereas the Tyr417→His and Tyr417→Phe variants showed 238 and 114-fold decreased kcat values, respectively, and 13 to 65-fold increased Michaelis–Menten parameters (Table 2). These results indicate that the imidazole side chain of His317 is mainly involved in protonation of the sulfate leaving group, and not in substrate binding. On the other hand, it appears that the tyrosine side chain is critically involved in the correct positioning of the substrate's sulfate group, as suggested by the large increase in the Km values observed for the Tyr417→Phe and Tyr417→His variants.

Inspection of the substrate-binding pocket of Pisa1 and comparison with the previously reported structure of SdsA1 prompted us to generate several single and double variants, with the aim of enabling hydrolysis of the (S)-2-octyl sulfate stereoisomer. To this end, we generated the Ser233→Tyr and Phe250→Gly single variants and the Ser233→Tyr/Phe250→Gly and Ser233→Phe/Phe250→Ser double variants. The two single variants Ser233→Tyr and Phe250→Gly showed 2.5 and 190-fold reduced catalytic activity, respectively, towards (R)-2-octyl sulfate, and the latter variant had a 28-fold higher Km (yielding kcat/Km of 17 500 and 5 m−1·s−1, respectively). Neither one of these variants had measurable activity towards (S)-2-octyl sulfate. The two double variants had three-fold to four-fold times lower kcat values, and the double variant carrying the Phe250→Gly replacement also had a 10-fold higer Km (yielding kcat/Km values of 920 and 9580 m−1·s−1 for the Ser223→Tyr/Phe250→Gly and Ser223→Phe/Phe250→Ser variants, respectively). This double variant also showed activity towards (S)-2-octyl sulfate. As summarized in Table 2, the Ser233→Tyr/Phe250→Gly variant still processes the (R)-stereoisomer with higher catalytic efficiency (kcat/Km = 919 m−1·s−1) than the (S)-stereoisomer (kcat/Km = 46 m−1·s−1), which translates into an E-value of 20.

Discussion

Despite the overall structural similarity between SdsA1 and Pisa1, the two proteins show only 43% identity and 61% similarity in amino acid sequence (Fig. S1). In this context, it is worth noting that Pseudomonas sp. DSM6611 was originally selected for its ability to degrade halogenated aromatic compounds, e.g. 4-fluorobenzoate [18]. The preferred source for soil bacteria appears to be inorganic sulfur [3]; however, in this environment, sulfur is predominantly (> 90%) bound to organic compounds as sulfonates and sulfate esters [22]. As Pseudomonas sp. DSM6611 was isolated from uncontaminated soil (B. Hauer, personal communication), Pisa1 could be part of the enzymatic repertoire of this strain for releasing sulfate from organic sulfate esters and hence making it available for bacterial growth in the soil. However, attempts to induce Pisa1 activity in Pseudomonas sp. DSM6611 with an organic or inorganic sulfur source elicited only a weak response, suggesting that Pisa1 is not involved in sulfur mobilization (P. Gadler, K. Faber, unpublished).

From a mechanistic point of view, chemical (nonenzymatic) sulfate ester hydrolysis is not a trivial task, as shown by the extreme values for half-life times of spontaneous sulfate ester hydrolysis via C–O (rather than S–O) bond breakage (103 versus 1018 years) [23]. In order to facilitate the departure of the sulfate moiety, it has to be converted into a good leaving group. Catalytic sulfate ester hydrolysis can be achieved by protonation of the sulfate ester (the pKa values of monomethylsulfate were calculated and estimated to be −8.4 and −3.4, respectively [24, 25]) by a strong acid of pKa < 2 to yield HSO4 as a good leaving group (the pKa of H2SO4 is −9 to −3 and the pKa of HSO4 is 1.9–2.7). On the other hand, basic conditions generate SO42−, which, being the anion of a weak acid, is a poorer leaving group. Although base-catalyzed hydrolysis proceeds under inversion of configuration, as expected, the reaction rates are exceedingly low, and alkaline hydrolysis is not feasible for practical purposes [26, 27]. The same holds for other O-nucleophiles, such as acetate and methoxide [28]. Bearing the difficulties of chemical (nonenzymatic) sulfate monoester hydrolysis in mind, we studied the issue of protonation of the sulfate leaving group based on the active site architecture revealed by X-ray crystallographic analysis of Pisa1. As shown in Fig. 5, SdsA1 and Pisa1 share an active site histidine (His317 in Pisa1) as a potential source for the protonation of the leaving group. In addition, Tyr417 and His405 in Pisa1 and SdsA1, respectively, may act as active site acids to fulfill the envisaged protonation. To investigate the potential role of these two amino acids, we generated amino acid replacements by active site directed mutagenesis: His317→Ala, as well as Tyr417→Phe, Tyr417→His, and Tyr417→Asp.

The characterization of the purified variant proteins revealed that enzymatic hydrolysis of the model substrate is three orders of magnitude slower for the His317→Ala variant than for the wild-type enzyme. This variant showed only a marginal effect on Km, indicating that His317 does not play a major role in substrate binding. Therefore, we suggest that His317 acts as a general acid catalyst, enhancing the leaving group ability of the sulfate group. In contrast to the His317→Ala variant, the Tyr417→Phe and Tyr417→His variants showed considerably higher enzymatic activities (five-fold and 10-fold, respectively). On the other hand, Km values were strongly affected. In fact, substrate saturation was not achieved, rendering Km values inaccessible in our experimental set-up. In the case of the Tyr417→Asp variant, kcat could not be determined, indicating that a negatively charged amino acid is prohibitive for substrate binding and enzymatic hydrolysis. These data support a role in substrate binding and positioning rather than protonation of the substrate's leaving group.

In addition to addressing issues concerning the enzymatic reaction mechanism, we also generated a small set of single and double variants (Table 2) to investigate the high stereopreference for the (R)-configuration of 2-octyl sulfate. In this set of variants, the Ser233→Tyr/Phe250→Gly double variant accepted (S)-2-octyl sulfate as a substrate, although the catalytic efficiency was 20-fold lower than with (R)-2-octyl sulfate. In comparison with the wild-type enzyme, the catalytic efficiency was even further reduced, by a factor of 630 (Table 2). This indicates that the amino acid replacements have led to an alternative lipophilic binding channel, enabling productive binding of the (S)-configured substrate (Fig. 6). These results demonstrate that the amino acids lining the hydrophobic substrate's binding pocket govern the stereo preference of the reaction and strongly affect the catalytic efficiency of substrate hydrolysis.

Experimental procedures

Expression and purification of recombinant protein in E. coli host cells

The expression and purification of SdsA1, Pisa1 and variants thereof was performed as described previously and in Doc. S1 [6, 9].

Activity assays for Pisa1 and its variants, and SdsA1

Enzymatic activities for the hydrolysis of 1-octyl sulfate and 2-octyl sulfate, respectively, were analyzed by chiral GC measurements, as described in a previous study [6, 9, 19].

Crystallization, data collection and structure elucidation of Pisa1

Pisa1 and Ser233→Tyr/Phe250→Gly double variant crystals were grown at 20 °C by sitting-drop vapor diffusion in silica hydrogel with equal volumes (5 μL) of protein (10 mg·mL−1) and reservoir solution [acetate buffer, pH 4.6, and either 30% poly(ethylene glycol) 2000 or poly(ethylene glycol) 4000]. Fragile, leaflet-shaped orthorhombic crystals with six molecules in the asymmetric unit were obtained with poly(ethylene glycol) 2000 (wild-type form-I), whereas poly(ethylene glycol) 4000 produced rodlike trigonal crystals hosting a dimer in the asymmetric unit (wild-type form-II). These crystals diffracted to 3.0 Å (double variant Ser233→Tyr/Phe250→Gly) and 2.7 Å (wild-type form-I and form-II) resolution, respectively, and were used for structure determination.

Data collection for both crystal forms obtained for wild-type Pisa1 were performed at the European Molecular Biology Laboratory in Hamburg, Germany. The experiments were carried out at 100 K with crystals flash-frozen in a mixture of 20% glycerol and mother liquor, using liquid nitrogen. A single-wavelength dataset from wild-type form-I crystals was collected at beamline X11. A MAD experiment was performed for wild-type form-II crystals, employing the four zinc atoms as anomalous scatter (beamline X12). Both beamlines were equipped with MAR CCD detectors. Data from the double variant crystals were collected at the ESRF Grenoble at beamline ID14-4. Indexing and integration of the collected data were performed with xds [29], and scaling was performed with scala [30].

Phasing of the Pisa1 wild-type form-II dataset was achieved through multiwavelength anomalous dispersion techniques; solve [31] found three of the four zinc ions, and a density modification using resolve [31] gave an interpretable density for most parts of the structure. The first trace was built with buccaneer [32] and coot [33]. It turned out that both molecules related by a noncrystallographic dyad are partly disordered.

The structures of wild-type form-I and the double variant were solved by molecular replacement with program phaser of the ccp4 package [34], and a starting model was built with modeller [35], on the basis of the SdsA1 coordinates. Whereas the wild-type form-I structure was well defined in most parts, the structure of the double variant was highly disordered. Both the wild-type form-II and the double variant structures were included in this report only because they show features that are not evident in the wild-type form-I structure. Table 1 gives an overview of data collection and refinement.

Docking studies

Molecules A from Pisa1 wild-type form-I and from the double variant were subjected to docking experiments with the program glide [36-39]. (R)-2-Octyl sulfate and (S)-2-octyl sulfate were used as possible substrates. All three structures were prepared with the protein preparation wizard. The substrate was built inside maestro, and both configurations (R and S) were calculated with ligprep. The position of the ligand's sulfate group was restrained to the position of the sulfate ion in the crystal structures.

Site-directed mutagenesis

The Pisa1 variants were prepared according to the protocol described in the QuikChange XL-site directed mutagenesis kit (Stratagene, La Jolla, CA, USA), with a slight modification. A two-step PCR reaction was carried out: in the first step, separate reactions for forward and reverse primers were performed for four cycles; in the second step, equal amounts of the forward and the reverse primers were mixed for another 18 cycles. The primers used for the insertion of the mutations are summarized in Table S1.

Isothermal titration calorimetry (ITC)

Kinetic parameters for the hydrolysis of (R)-2-octyl sulfate, (S)-2-octyl sulfate and 1-octyl sulfate were determined with a VP-ITC system (MicroCal, Northampton, MA, USA), according to the manufacturer's instructions. All experiments were performed at 25 °C in 100 mm Tris/HCl (pH 8.2) buffer, and solutions were degassed immediately before measurements were taken. The enthalpy of the reaction was determined by the single injection method: The substrate (R)-2-octyl sulfate was added (1.5 mm, 3 × 20 μL, duration of 39.9 s, spacing time of 600 s) to Pisa1 (125 nm), and the substrate 1-octyl sulfate (150 μm, 3 × 20 μL, duration of 39.9 s, spacing time of 600 s) was added to SdsA1 (125 nm). For the kinetic parameters, the multiple injection method was used for the titration of Pisa1, its variants and SdsA1, with (R)-2-octyl sulfate, (S)-2-octyl sulfate, and 1-octyl sulfate (the enzyme and the substrate concentrations were varied from 8 nm to 200 μm or from 800 μm to 55 mm, depending on the turnover numbers and Km values of the proteins, respectively). Nonlinear least square fitting with origin version 7.0 (MicroCal) was used for ITC data analysis to obtain kcat, Km, and KI.

Acknowledgements

This work was supported by the Austrian Fonds zur Förderung der wissenschaftlichen Forschung (FWF) through the PhD program ‘Molecular Enzymology’[(W901) to K. Faber and P. Macheroux.

Ancillary