Specific Targeting of the Metallophosphoesterase YkuE to the Bacillus Cell Wall Requires the Twin-arginine Translocation System*

Background: The Gram-positive bacterium Bacillus subtilis contains two twin-arginine (Tat) translocases, each specifically secreting one known substrate protein. Results: The TatAyCy translocase facilitates export of the metallophosphoesterase YkuE to the cell wall. Conclusion: YkuE is the third identified genuine Tat substrate in Bacillus. Significance: YkuE is the first protein found to be specifically targeted to a Gram-positive bacterial cell wall via the Tat pathway. The twin-arginine translocation (Tat) pathway is dedicated to the transport of fully folded proteins across the cytoplasmic membranes of many bacteria and the chloroplast thylakoidal membrane. Accordingly, Tat-dependently translocated proteins are known to be delivered to the periplasm of Gram-negative bacteria, the growth medium of Gram-positive bacteria, and the thylakoid lumen. Here, we present the first example of a protein, YkuE of Bacillus subtilis, that is specifically targeted by the Tat pathway to the cell wall of a Gram-positive bacterium. The cell wall binding of YkuE is facilitated by electrostatic interactions. Interestingly, under particular conditions, YkuE can also be targeted to the cell wall in a Tat-independent manner. The biological function of YkuE was so far unknown. Our present studies show that YkuE is a metal-dependent phosphoesterase that preferentially binds manganese and zinc.

The twin-arginine protein translocation (Tat) 3 pathway is present in the membranes of most bacteria and the thylakoid membranes of plant chloroplasts (1,2). A key distinguishing feature of the Tat translocase is that it can transport proteins in a folded state (3)(4)(5)(6)(7). Moreover, many substrates of the Tat pathway contain a cofactor, which needs to be inserted correctly prior to translocation. If the substrate protein does not fold efficiently or correctly, it is potentially exported in an unfolded state by another transport pathway, such as the general secretion pathway (Sec) (3, 5, 8 -10). A second important feature of substrates that are accepted by the Tat translocase is the presence of a specific twin-arginine (RR) signal peptide at the N terminus of the protein. The RR signal peptides are composed of three parts, namely an N-terminal region rich in positively charged residues, a central hydrophobic H-region, and a C-terminal region containing the cleavage site recognized by signal peptidases (11,12). Importantly, these signal peptides possess a twin-arginine (RR) or a lysine-arginine (KR) recognition motif in the N-region with the consensus sequence (K/R)RX##, where # marks hydrophobic residues and X can be any residue (11)(12)(13)(14). This RR motif or its variant with a lysine is specifically recognized by the Tat translocase (15)(16)(17). RR signal peptides are generally less effective in protein targeting to the Sec machinery due to the low hydrophobicity of their H-region and the presence of a positively charged residue in the C-region that facilitates "Sec avoidance" (16). Several studies conducted with Gram-negative and Gram-positive bacteria have shown that RR signal peptides are often interchangeable among different proteins and that in some instances proteins can be redirected from the Sec pathway into the Tat pathway by replacing the original Sec-type signal peptide with an RR signal peptide (10,18,19).
Interestingly, Tat is a major protein translocation pathway in some bacterial species, whereas other species seem to make only very limited use of their Tat translocase. For example, more than 30 proteins are exported Tat-dependently in streptomycetes (19 -23), whereas only two genuine Tat substrates (PhoD and YwbN) have so far been identified in Bacillus subtilis (24,25) and only one in Staphylococcus aureus (26). In particular, the limited use of Tat in B. subtilis is remarkable in view of this organism's very high capacity for protein export (12,27). Moreover, B. subtilis contains two Tat translocases named TatAdCd and TatAyCy, which can function independently (25,28,29). This suggests that some Tat substrates of B. subtilis may have been overlooked in studies aimed at their identification. Several bioinformatic approaches have been designed with the aim to identify new Tat substrates (20,30,31). Indeed such bioinformatics tools predict additional Tat substrates for B. subtilis, the numbers depending on the stringency of the algorithms. To demonstrate the Tat specificity of predicted RR signal peptides, a reporter system was developed based on the agarase of Streptomyces coelicolor, which is secreted via the Tat pathway of this organism (22). Because the agarase is secreted into the growth medium of Streptomyces, its activity can be tested in a simple semi-quantitative way with a colorimetric assay. Using this assay, several potential RR signal peptides have been tested, including the B. subtilis protein YkuE (19). These studies confirmed that the YkuE signal peptide was capable of directing Tat-dependent protein transport, but the actual Tatdependent secretion of YkuE and other predicted Tat substrates of B. subtilis remained enigmatic.
This study aimed to assess the expression, export, and function of YkuE in B. subtilis. Our results show that YkuE is a Tat-dependently exported metallophosphoesterase. Most noticeably, YkuE is specifically targeted to the cell wall of B. subtilis, making it the first known protein that is targeted via Tat to this particular subcellular compartment. We show that cell wall binding of YkuE is facilitated by electrostatic interactions. Furthermore, we have identified different modes of YkuE export to the cell wall that are specific for particular conditions, such as phosphate starvation or YkuE overexpression.
DNA Cloning Procedures-Procedures for DNA purification, restriction, ligation, agarose gel electrophoresis, and transfor- mation of competent E. coli cells were carried out as described previously (33). B. subtilis was transformed as described by Kunst and Rapoport (34). PCR was carried out with the Phusion DNA polymerase (Finnzymes). Primers are listed in the supplemental Table 1. PCR products were purified using the PCR purification kit from Roche Applied Science. Restriction enzymes were obtained from New England Biolabs. Plasmid DNA from E. coli was isolated using the alkaline lysis method or by using the Invisorb plasmid isolation kit (Invitek). All constructs were checked by sequencing. The plasmids pHB-ykuEmyc and pHB-ykuE were constructed by cloning the PCR-amplified ykuE gene into the pHB201 plasmid. The forward primer used to construct pHB-ykuE-myc contained a SalI restriction site and the reverse primer an EcoRI restriction site and a Myc tag. The PCR product was digested with SalI-EcoRI and cloned in the digested SalI-EcoRI pHB201 vector. The forward primer used to construct pHB-ykuE contained an SpeI restriction site, and the reverse primer contained a BamHI restriction site. The PCR product was digested with SpeI-BamHI and cloned into the digested SpeI-BamHI pHB201 vector. To construct pNZ8910-ykuE-StrepII, a synonymous point mutation was introduced into the ykuE gene to delete an internal BspHI restriction site. The mutagenesis was carried out using the F_Mut_ykuE and R_Mut_ykuE primers and the pHB-ykuE plasmid as a template. The resulting plasmid pHB-ykuEmut was verified by sequencing and used as a template for a further PCR with a forward primer containing a BspHI restriction site and a reverse primer with an SpeI restriction site. The fragment was digested with BspHI-SpeI and cloned into NcoI-HindIII digested pNZ8910 plasmid. The ligation mixture was introduced into L. lactis by transformation, resulting in the plasmid pNZ8910-YkuE-StrepII. This plasmid was then introduced into B. subtilis 168 or the tatAyCy, tatAdCd, or total tat3 mutant strains by transformation.

SDS-PAGE and Western
Blotting-Cells were separated from the growth medium by centrifugation. Next, the cells were fractionated as described by Zweers et al. (35). Briefly, the cells were incubated with lysozyme in protoplast buffer to liberate cell wall-associated proteins. The resulting protoplasts were collected by centrifugation and disrupted by bead beating. Debris of the disrupted protoplasts was removed through centrifugation. Membranes were then separated from the cytoplasm through ultracentrifugation. Finally, the collected membranes were resuspended in solubilization buffer with 0.1% n-dodecyl ␤-D-maltoside. To extract cell wall-bound proteins, cells were incubated for 10 min in 25 mM Tris-HCl buffer, pH 8.0, with 1.5 M LiCl, or a solution containing 2 M KSCN as described previously (36,37). Proteins in the collected fractions were separated by PAGE using pre-cast BisTris NuPAGE gels (Invitrogen). For Western blotting analyses, proteins separated by PAGE were semi-dry blotted (75 min at 1 mA/cm 2 ) onto a nitrocellulose membrane. Subsequently, FeuA, TatAy, TrxA, WprA, YkuE, YwbN, or YfkN were detected with specific polyclonal antibodies raised in rabbits. Visualization of bound antibodies was performed using IRDye 800CW goat anti-rabbit secondary antibodies in combination with the Odyssey infrared imaging system (LiCor Biosciences). Fluorescence was recorded at 800 nm.
Protein Production and Purification-A B. subtilis 168 strain containing a subtilin-inducible ykuE-StrepII-tag construct was grown in 3 liters of LB medium at 37°C under selective conditions (2 g ml Ϫ1 erythromycin and 10 g ml Ϫ1 kanamycin) until an A 600 of 0.6, at which point the cultures were induced by addition of subtilin-containing supernatant from B. subtilis ATCC 6633 prepared as described by Bongers et al. (38). After 3 h of induction, cells were harvested (4°C, 3500 ϫ g for 20 min). Cell pellets were washed and resuspended in 100 mM Tris, pH 8.0, 150 mM NaCl. Cells were disrupted by using a French pressure cell (Thermo Spectronic) and a French pressure cell press (Aminco), resulting in the liberation of StrepII-tagged YkuE from the cells. The filtrated lysate was subjected to Strep-Tactin chromatography using an FPLC purifier system (Pharmacia) and a column with a 2-ml packed volume of Strep-Tactin Superflow material (IBA). The protein extract was applied to the column at a flow rate of 0.5 ml min Ϫ1 . The column was washed for 20 min with cell resuspension buffer at 0.8 ml min Ϫ1 . The bound protein was then eluted with 2.5 mM D-desthiobiotin (IBA) dissolved in resuspension buffer at 0.5 ml min Ϫ1 . Elution fractions were analyzed by SDS-PAGE and subsequent mass spectrometry for species identification, and those fractions containing pure YkuE protein were concentrated using Amicon ultracentrifugal filter units (Millipore) with a nominal molecular mass limit of 10 kDa at 4°C and 2000 ϫ g.
Protein concentrations were determined with the Bradford method (39) using a BSA calibration curve. Freshly purified protein of different concentrations obtained before and during the concentration process was used for ICP-MS analysis.
MS-based Protein Identification-Protein samples were digested in gel by the addition of sequencing grade modified trypsin (Promega) and incubation at 37°C overnight. Samples were dissolved in 25 l of 10% acetonitrile, 0.1% TFA. The mass spectrometric analysis of the samples was performed by using an Orbitrap Velos Pro mass spectrometer (ThermoScientific). An Ultimate nanoRSLC-HPLC system (Dionex), equipped with a nano C18 RP column, was connected on line to the mass spectrometer through a Proxeon nanospray source. A 6-l aliquot of the tryptic digest was injected onto a C18 pre-concentration column. Automated trapping and desalting of the sample were performed at a flow rate of 6 l/min using water, 0.05% formic acid as a solvent. Separation of the tryptic peptides was achieved with a gradient of water, 0.045% formic acid (solvent A) and 80% acetonitrile, 0.05% formic acid (solvent B) at a flow rate of 300 nl min Ϫ1 . The column was connected to a stainless steel nanoemitter (Proxeon), and the eluent was sprayed directly toward the heated capillary of the mass spectrometer using a potential of 2300 V. A survey scan with a resolution of 60,000 was combined with at least three data-dependent MS/MS scans. Data analysis was performed using Proteome Discoverer (ThermoScientific) with SEQUEST and MASCOT (version 2.2; Matrix science) search engines using either the SwissProt or NCBI databases.
ICP-MS Analysis-Protein samples with defined concentrations were treated with 1:6 suprapure nitric acid (Merck) to dissociate all metals from organic complexors and solutions were further diluted gravimetrically with ultrapure milliQ water. Yttrium was added to all samples as an internal standard, and quantitative analysis of metal contents for magnesium, vanadium, chromium, manganese, iron, cobalt, nickel, copper, zinc, molybdenum, and tungsten was performed using an Agilent 7500ce (Agilent Technologies) together with established calibration and detection protocols (40,41). External calibration during measurement was carried out using appropriate metal calibration standards (Merck). Data obtained both in the hydrogen and helium collision modes were averaged for all analyzed isotopes. Background metal content, which was detected at substantially low level in buffer control samples, was subtracted from the protein-containing samples after analysis.
Enzymatic Assays-Enzyme activity assays (200-l reaction mixtures) were carried out in assay buffer containing 50 mM Tris-HCl (pH was varied between 6.5 and 9.5) and 5 mM MnCl 2 , with varying concentrations of p-nitrophenyl phosphate (0.1-10 mM) and 1 M purified YkuE-StrepII at 25°C. The reactions were started by addition of the phosphoester substrate and quenched by addition of 2% SDS (w/v). The release of p-nitrophenol was measured spectrophotometrically at 410 nm. Product quantification was subsequently carried out using a p-nitrophenol standard curve. Control reactions without enzyme were carried out for each pH series to correct for spontaneous p-nitrophenyl phosphate background hydrolysis. Substrate-dependent kinetic data of each pH series were plotted and analyzed by a Michaelis-Menten fitting model (Microcal Origin 5.0 software).

YkuE Is a Tat-dependently Exported Cell Wall Protein-To
verify whether the Tat system does indeed translocate the YkuE protein in B. subtilis, a polyclonal antibody was raised against this protein and used to determine the intra-and extracellular levels of YkuE. However, irrespective of the different growth conditions that were tested, we were unable to detect the production of YkuE in B. subtilis by Western blotting (data not shown). In this respect, it should be noted that a recent tiling array analysis of B. subtilis gene expression across 104 conditions revealed that ykuE is expressed at very low levels under all tested conditions (supplemental Fig. S1) (42). For this reason, we expressed a plasmid-borne copy of the ykuE gene from the relatively weak constitutive promoter of the low copy expression vector pHB201, which increases the ykuE expression level to a similar level as that of the ywbN and tatAyCy genes (43,44). This allowed the detection of the mature-sized YkuE protein in the B. subtilis prototype strain 168 (Fig. 1). Next, the plasmid was introduced in mutant B. subtilis strains lacking either the tatAyCy genes, the tatAdCd genes, or all tat genes (total-tat).
To study YkuE export, the transformed strains were grown until stationary phase in LB medium. Cells were separated from the growth medium by centrifugation and incubated with lysozyme to liberate cell wall-bound proteins. Finally, the resulting protoplasts were separated from the liberated cell wall proteins by centrifugation. The collected fractions were analyzed by PAGE and Western blotting using the YkuE-specific antibodies, which revealed that YkuE was only detectable in the protoplast and cell wall fractions. Importantly, the localization of mature YkuE to the cell wall was strictly dependent on the presence of the TatAyCy translocase because this protein was not detectable in the cell wall fractions of the tatAyCy mutant and the total tat mutant, although it was detectable in the cell wall fraction of the tatAdCd mutant. The cell wall localization of YkuE was restored in the tatAyCy mutant and the total tat mutant by ectopic expression of the tatAyCy genes from a plasmid (Fig. 1). In contrast to YkuE, the cell wall localization of the Sec-dependent control protein YfkN was not affected by any of To determine the subcellular localization of YkuE in the presence or absence of active Tat translocases, the parental strain B. subtilis 168 and the tat mutant strains tatAyCy, tatAdCd, total-tat3, tatAyCy pCACy, and total-tat3 pCACy were transformed with plasmid pHBykuEmyc for the expression of YkuE. Next, the cells producing YkuE were grown for 7 h in LB medium. The cells were then separated from the growth medium by centrifugation and subjected to subcellular fractionation. Proteins in the obtained fractions were separated by PAGE, and the presence of YkuE and YfkN was monitored by Western blotting with specific polyclonal antibodies. Protein loading on the gels was corrected for A 600 . Only the results for the cytosol, membrane, and cell wall fractions are shown because no YkuE was detectable in the fractions representing the growth medium. The cell wall-localized protein YfkN was used as a Tat the tat mutations tested. Interestingly, mature YkuE was also detectable in the protoplast fraction irrespective of the presence or absence of functional Tat machinery, although cells lacking TatAyCy contained less YkuE than TatAyCy-proficient cells. Further fractionation studies revealed that the protoplastassociated mature YkuE was localized to the cytoplasmic membrane (Fig. 1). Together, these findings imply that the membrane localization of YkuE is to a large extent Tat-independent, whereas its cell wall localization is strictly TatAyCy-dependent. To investigate whether the Tat-independent membrane localization is specific for YkuE, or a more general feature for Tat substrates in B. subtilis, we also assessed membrane localization of the well studied Tat substrate YwbN in the different tat mutant strains. To this end, we analyzed the localization of YwbN expressed from its authentic promoter using a YwbNspecific polyclonal antibody. As shown in Fig. 2, mature YwbN was localized to the cell wall and growth medium in a strictly TatAyCy-dependent manner. In contrast, the membrane association of mature YwbN was completely Tat-independent, and barely any YwbN was detectable in the cytosolic fraction (Fig.  2). These findings indicate that Tat substrates of B. subtilis can interact with the membrane in a Tat-independent manner. At present, we do not know how this Tat-independent interaction with the membrane is brought about or how the signal peptide is removed in this case. Potentially, signal peptide cleavage is catalyzed by as yet unidentified cytoplasmic or membrane-associated proteases for which there are many candidate proteases in B. subtilis (12,45).
Because YkuE is a predicted metallophosphoesterase, we also investigated the cell wall localization of this protein under phosphate starvation conditions where various secreted phosphatases and phosphodiesterases are produced by B. subtilis (4,12). However, also under these conditions, no YkuE expression from the authentic ykuE promoter was detectable, which is in line with the low expression level of this gene under phosphate starvation conditions (supplemental Fig. S1, compare the expression profiles for ykuE with those for tatAd, tatCd, and phoD) (42). As shown for cells grown in LB broth, the cell wall localization of YkuE in phosphate-starved cells was strongly dependent on the presence of the TatAyCy translocase (Fig. 3). In fact, the level of cell wall-associated YkuE was increased when the tatAy and tatCy genes were overexpressed from a plasmid, and it was severely reduced in mutant cells lacking the tatAyCy genes or expressing only tatAy or tatCy. The absence of the tatAdCd genes had no effect on YkuE localization to the cell wall showing that the TatAdCd translocase is not involved in YkuE export when this translocase was produced at wildtype levels (supplemental Fig. S2). Surprisingly, very small amounts of YkuE were detectable in the cell wall of the tatAyCy and total tat mutant strains (Fig. 3), suggesting that some Tatindependent YkuE localization to the cell wall can occur under phosphate starvation conditions.
Tat-independent YkuE Export upon Overexpression-Because some Tat-independent membrane and cell wall localization of YkuE was detectable, we investigated whether this might relate to the production level of this protein. To this end, we made use of the subtilin-regulated gene expression (SURE) system for high level protein production in B. subtilis (38,46). The ykuE gene was cloned into the SURE plasmid, and its expression was induced with subtilin in different tat mutant strains. As shown in Fig. 4, YkuE was overproduced in high amounts. Most of this overproduced YkuE was localized to the membrane, and relatively small amounts were localized to the cell wall. Furthermore, upon overproduction, YkuE was localized to the cell wall in a mostly Tat-independent manner. Nevertheless, a small decrease in the amount of cell wall-localized YkuE was detectable in the tatAyCy or total tat mutant strains compared with the parental strain 168 and the tatAdCd mutant. Furthermore, high amounts of membrane-associated YkuE were detectable, and also upon overproduction, the membrane association of this protein remained Tat-independent.
YkuE Is Retained in the Cell Wall through Electrostatic Interactions-The YkuE protein lacks obvious domains for cell wall binding, such as an LPXTG motif for covalent attachment to peptidoglycan or domains for noncovalent cell wall binding (4,12). To determine how YkuE can be retained in the cell wall, B. subtilis cells expressing this protein at different levels were subjected to treatment with either 2 M KSCN or 1.5 M LiCl as described previously (36,37). Treatment with the chaotrope KSCN would disrupt noncovalent cell wall interactions based on hydrogen bonds, van der Waals forces, or hydrophobicity (47), and treatment with LiCl would disrupt electrostatic cell wall interactions. As shown in Fig. 5, treatment of the cells with LiCl was equally effective in extracting YkuE from the cell walls  as protoplasting with lysozyme. By contrast, KSCN did not liberate any YkuE from the cells (data not shown). This implies that YkuE is retained in the B. subtilis cell wall through electrostatic interactions, most likely with particular cell wall components.
Purification and Metal Cofactor Analysis of YkuE-Database homology searches suggest that YkuE is a putative metallophosphoesterase. To validate this proposed function, YkuE was produced as a StrepII-tagged variant in its original host B. subtilis to ensure the association with any native metal cofactor(s) that it may carry. The protein was obtained at a yield of about 0.2 mg/liter culture and was analyzed after purification for intrinsic stability (Fig. 6). The protein sample that was analyzed directly after elution (ϳ0.5 h before SDS-PAGE) was observed to be intact as a full-length variant of the expected size of ϳ35 kDa. Mass spectrometric analysis of purified YkuE yielded a total amino acid sequence coverage of ϳ75% starting from residue 33 (supplemental Table 2). No peptides relating to the signal peptide were identified, which suggests that YkuE was purified predominantly in the mature form. Furthermore, no evidence for impurities with other B. subtilis proteins was obtained, suggesting that the protein samples contained YkuE only. Interestingly, when purified YkuE samples were incubated for 20 h at 4°C, some degradation of YkuE was observed. All detectable degradation fragments corresponded to YkuE as shown by mass spectrometry. YkuE degradation was not reduced in the presence of inhibitors for cysteine proteases (N-ethylmaleimide) or serine proteases (phenylmethylsulfonyl fluoride; Fig. 6). In contrast, YkuE degradation was significantly reduced upon addition of EDTA, an effective inhibitor for metallohydrolase activity. Addition of 10 mM EDTA had a slightly higher protective effect than 1 mM EDTA, suggesting a direct relation between competitive metal sequestration and degradation of YkuE. Together these findings suggest a basal level of autohydrolytic activity of the metal-charged enzyme upon storage in aqueous solution. This may indicate a high activation potential of the enzyme for the nucleophilic attack of amide bonds. However, we cannot completely exclude the possibility that small amounts of a metalloprotease (undetectable by mass spectrometry) were co-purified with YkuE causing its subsequent degradation.
To identify the bound metal cofactor species in a quantitative way, the freshly purified YkuE samples obtained before and after protein concentration by ultrafiltration over a concentration range of 1 order of magnitude were subjected to inductively coupled plasma-mass spectrometry (ICP-MS) analysis. This analysis revealed two dominant metals, zinc and manganese, which were present in a strongly protein concentrationdependent manner ( Table 2). The averaged stoichiometries of these metals were for manganese 0.76 Ϯ 0.01 mol and for zinc 1.16 Ϯ 0.02 mol/mol YkuE, which indicates their association with two distinct metal-binding sites. In the higher concentrated samples, small but significant quantities of nickel and copper were also found ( Table 2). However, the low stoichiometric proportions of these metals (averaged amounts of 0.08 Ϯ 5 ϫ 10 Ϫ5 mol of nickel and 0.03 Ϯ 2 ϫ 10 Ϫ3 mol of copper per mol of YkuE, respectively) point to a fairly low selec-  tivity of these species. Their substoichiometric amounts suggest a nonselective binding site occupation due to a superior metal-ligand affinity within the Irving-Williams series for divalent transition metals, which has to be considered especially if competition with manganese takes place (48,49). Likewise, the unequal stoichiometric distribution of manganese and zinc with a bound mole ratio of about 0.8 to 1.2 may well be consistent with a partial loss of the low affinity species manganese and second with a partial displacement by the high affinity species zinc during the processes of overproduction and/or purification. This points to the importance and maintenance of proper metal insertion during protein folding and transport between different compartments (50). All further metals identified by ICP-MS were present below a threshold level of 1% compared with the injected protein quantities ( Table 2), suggesting that they were not specifically associated with YkuE.
YkuE Displays Monophosphoesterase Activity in a Strongly pH-dependent Manner-To determine the catalytic activity of YkuE, freshly purified protein was incubated in an enzyme assay buffer containing 5 mM MnCl 2 to ensure the quantitative occupation of the proposed manganese-binding site. Enzymatic reactions were carried out with varying concentrations of p-nitrophenyl phosphate as a substrate, which is commonly used for the evaluation of monophosphoesterase activity (51). The reaction series were performed under seven different pH conditions in the range of 6.5 to 9.5 to screen for the pH optimum of enzymatic activity. The product formation rates of p-nitrophenol obtained for each pH were evaluated by the Michaelis-Menten model and revealed relatively constant though slightly decreasing K m values when the pH was changed from acidic to alkaline. In contrast, the k cat value increased over more than 1 order of magnitude within this pH range (Fig. 7A and supplemental Table 3). The obtained catalytic efficiencies at different pH values showed that the pH optimum for YkuE was approximately at pH 9.0, and the pK a value of the activated water species was determined at about 7.8, indicating a nearly 100% active hydroxide complex at pH 9.0 and above (Fig. 7B). Thus, the slight drop of enzyme activity at pH 9.5 pointed to an arising catalytic limitation, which is different from the cofactorgenerated activation potential of the water nucleophile. Previous kinetic analyses of zinc-dependent alkaline phosphatase, representing a widely distributed monophosphoesterase, have shown that the rate-limiting process at higher pH is in fact the dissociation rate of the cleaved phosphate product, although a lower pH generally limits the hydrolysis rate of the enzymephosphoryl intermediate (51). This substitution of rate-limiting events takes place mainly between pH 7.0 and 8.0, which is also observed for YkuE due to the strong pH dependence of k cat within this range and may relate to the actual pK a value of the proposed metal-bound hydroxide acting as the attacking nucleophile. The presence of this activated water species in bimetallic reaction centers has been confirmed previously from electron densities in the crystal structures of native and phosphoryl-bound alkaline phosphatase (52). Taken together, our present findings show that YkuE apparently acts as a manganese/zinc-dependent metallophosphoesterase that is Tat-dependently localized to the cell wall of B. subtilis.  The monophosphoester compound p-nitrophenyl phosphate was used as a substrate, and the assays were conducted under varying pH conditions (between pH 6.5 and 9.5) at 25°C. The formation of p-nitrophenol was monitored spectrophotometrically, and the obtained data were corrected for spontaneous background hydrolysis examined in nonenzymatic control reactions. Final data were plotted and analyzed by Michaelis-Menten-type nonlinear regression. The kinetic parameters for all pH-dependent reaction series are given in supplemental Table 3. B, obtained catalytic efficiencies were plotted versus their cognate pH to determine the optimum catalytic activity and the pK a of the water nucleophile for YkuE-dependent hydrolysis. Because of the strongly pH-dependent k cat values, the catalytic efficiencies vary over 1 order of magnitude within the tested pH range and show an optimum peak at around pH 9.0. The pK a of the hydrated metal center is ϳ7.80, revealing the strong activation potential of the binuclear catalytic site of YkuE.

DISCUSSION
Previous studies have shown that the Tat translocases of B. subtilis are highly restrictive with respect to the acceptance of potential substrate proteins (53,54). Thus, despite extensive investigations, merely two strictly Tat-dependent proteins of this organism, PhoD and YwbN, have so far been identified. In this study, we show that the YkuE protein is also a genuine substrate of the B. subtilis Tat pathway and that the TatAyCy translocase specifically directs this protein to the cell wall. No evidence for an involvement of the TatAdCd translocase in YkuE export was obtained under any of the tested conditions. Furthermore, our data show that the translocated YkuE protein is retained in the cell wall through electrostatic interactions. Notably, this is the first time that a cell wall protein has been identified as a specific substrate of the Tat system. In contrast, the two other known Tat substrates of B. subtilis, PhoD and YwbN, are secreted into the growth medium. Interestingly, the Tat dependence of YkuE is not as strict as that of the two other substrates because we have observed at least two conditions where YkuE is exported Tat-independently, albeit to different extents. Although the cell wall localization of YkuE in cells growing in LB medium seems strictly Tat-dependent, a very low but detectable level of Tat-independent YkuE export takes place when cells are grown under phosphate starvation conditions. Under the latter conditions, the amount of cell wall-associated YkuE was found to be increased upon overexpression of the TatAyCy translocase, which underscores the view that YkuE is most effectively exported via this translocase. This finding is also fully consistent with the previously observed increase in the Tat-dependent secretion of the YwbN protein into the growth medium upon overexpression of TatAyCy (25). We therefore conclude that TatAyCy is present in limiting amounts when substrates like YkuE and YwbN are overproduced. This is actually very clearly observed upon high level overexpression of YkuE (Fig. 3). Under these high level production conditions, most YkuE is Tat-independently localized to the cell wall, showing that the saturation of TatAyCy leads to YkuE export via a Tat-independent route, which is probably the Sec pathway (4,12).
Several known Tat substrates are proteins that contain a metal cofactor that may need to be inserted in the folded protein prior to translocation (55). For example, the MncA protein in Synechocystis is secreted via the Tat pathway. MncA is believed to bind its manganese cofactor during folding in the cytoplasm where metal species such as copper and zinc are tightly bound and thus are less competitive than in the extracytosolic compartment. This study suggests that a similar mechanism for distinct incorporation of transition metal species with different binding affinities could have evolved in the case of YkuE. The determined metal-protein stoichiometries indicate that one manganese and one zinc atom are incorporated at the predicted binuclear metal center of YkuE. This would place YkuE in the group of binuclear metallohydrolases that selectively bind magnesium, manganese, zinc, iron, nickel, or cobalt at their distinct metal recognition sites (56). However, it remains to be shown whether metal species with different ligand affinities are generally bound in a successive way to the binuclear site before and after transport of the folded protein through a secretion pathway, such as the Tat system. Indeed, many metallohydrolases such as the E. coli alkaline phosphatase PhoA (57) and the aminopeptidase Iap (58, 59) as well as several virulence-associated metalloproteases in Vibrio (60), Streptococcus (61,62), or Burkholderia (63,64) are secreted in an unfolded state via the Sec pathway. However, these proteins usually contain only one transition metal species, such as zinc, that does not face strong binding site competition in the extracytosolic environment due to its high binding site affinity.
Comparisons of the sequences of various binuclear metallophosphoesterases with the sequence of YkuE and its conserved binding site motifs suggest mixed nitrogen/oxygen ligand donor sites for both metals in the binuclear center, primarily formed by histidine and aspartate residues (Fig. 8). Binding site 1, which contains a higher proportion of oxygen donors, may be suspected to favor the binding of manganese, although site 2 with higher nitrogen proportion may in contrast favor interaction with zinc, due to the corresponding metal-ligand affinity series (65). However, metal selectivities are also dependent on binding site-specific geometric features, and thus, predictions of selectivity for one or other of the species can only be tentatively based on the currently available sequence data. Nonetheless, the binding site compositions suggest that zinc will bind at least 2 orders of magnitude stronger to both of these sites than manganese according to the affinity ranges of the Irving-Williams series for divalent cations. The indicated partial displacement of manganese by zinc may further suggest that the selectivity of the lower affinity site 1 may be relaxed. At present, it cannot be excluded that catalytic activity may be provided upon FIGURE 8. YkuE active site model. The proposed binuclear metal reaction center of YkuE is shown, and amino acid residues are indicated that are predicted to act as ligand donors at the metal cofactor site. Predictions were made upon comparison with various metallophosphoesterase consensus sequences, and their metal-binding motifs were based on protein database information (UniProt, NCBI). Putative metal selectivities at the distinct binding sites are further suggested according to the experimentally found dominant metal species associated with natively purified YkuE and based on known metal-ligand complex stabilities for varied N/O donor sites. The core processes of the proposed catalytic mechanism of YkuE are indicated by interaction of a substrate phosphoryl moiety with the bi-metal cofactor site, which also activates water for nucleophilic attack and subsequent hydrolysis of the currently not further specified phosphoester substrate compound(s). occupation with either one or the other of these two species or even further divalent species of the transition metal series displaying considerable binding affinity.
The simplest catalytic mechanism of the YkuE metallophosphoesterase may compose four core processes, each of which can be further divided into particular equilibrium steps. First, the association of the preferably deprotonated oxygen donor groups of the phosphoryl substrate molecule takes place at the free metal coordination sites of the binuclear metal center. As the investigation of native enzyme crystal structures of this superfamily showed (56), water coordination at the bi-metal center is expected to precede this event, leading to a readily activated hydroxyl group, which nucleophilically attacks the substrate phosphorus atom within the second process. This then leads to hydrolysis by electron pair shift toward the alcohol function of the phosphoester. Third, dissociation of the alcohol and phosphate hydrolysis products takes place, followed by a fourth process of repeated water binding and activation at the metal center. Further detailed investigations of the catalytic properties of YkuE may include a closer inspection of its phosphoryl substrate specificity as well as the detailed analysis of the metal-binding site by metal replacement analyses and binding site mutagenesis.
In conclusion, our novel finding that the B. subtilis YkuE protein is a mixed site metallophosphoesterase, which is specifically targeted to the cell wall, sheds new light on the versatility of Tat translocases in the targeting of proteins to different subcellular compartments. To date, no specific Tat-dependent protein targeting to a bacterial cell wall has been documented. Although previous studies have shown that several Tat-dependently exported proteins can be extracted from the cell wall of streptomycetes, there is no published evidence that these proteins are exclusively localized in this subcellular compartment (22,23). Furthermore, our present data show that particular Tat-dependently secreted proteins of B. subtilis can bypass the Tat pathway when the required Tat translocase is present in limiting amounts. This is also a novel finding, because secretion of the two other known Tat substrates of this organism, PhoD and YwbN, remains strictly Tat-dependent when TatAdCd or TatAyCy becomes limiting. This finding is important, because it may explain, at least in part, why only a very limited number of Tat-dependently secreted proteins were identified so far in B. subtilis. Another reason for the detection of only few Tat-dependently exported proteins in B. subtilis could be that some of these proteins have been overlooked due to specific targeting to the cell wall, as is the case for YkuE that is retained in the cell wall through electrostatic interactions. Finally, our findings support the view that, during the evolution of B. subtilis as a colonizer of ecological niches in the soil and plant rhizosphere, there must have been a particularly high selective pressure on phosphate acquisition because this organism produces at least five exported proteins with phosphoesterase activity, namely PhoA, PhoB, PhoD, YfkN, and as shown in the present studies YkuE. All these enzymes contribute to the acquisition of phosphate. A similar functional redundancy was previously shown for other proteins that fulfill critical physiological roles in B. subtilis as exemplified by eight exported proteases that cooperate in nitrogen acquisition (12) or essential protein secretion machinery components such as the five type I signal peptidases that catalyze the maturation of secretory precursor proteins (66). In this light, we believe that the demonstration of the metallophosphoesterase activity of YkuE is of physiological relevance.