The Solution Structure of Escherichia coli Wzb Reveals a Novel Substrate Recognition Mechanism of Prokaryotic Low Molecular Weight Protein-tyrosine Phosphatases*

Low molecular weight protein-tyrosine phosphatases (LMW-PTPs) are small enzymes that ubiquitously exist in various organisms and play important roles in many biological processes. In Escherichia coli, the LMW-PTP Wzb dephosphorylates the autokinase Wzc, and the Wzc/Wzb pair regulates colanic acid production. However, the substrate recognition mechanism of Wzb is still poorly understood thus far. To elucidate the molecular basis of the catalytic mechanism, we have determined the solution structure of Wzb at high resolution by NMR spectroscopy. The Wzb structure highly resembles that of the typical LMW-PTP fold, suggesting that Wzb may adopt a similar catalytic mechanism with other LMW-PTPs. Nevertheless, in comparison with eukaryotic LMW-PTPs, the absence of an aromatic amino acid at the bottom of the active site significantly alters the molecular surface and implicates Wzb may adopt a novel substrate recognition mechanism. Furthermore, a structure-based multiple sequence alignment suggests that a class of the prokaryotic LMW-PTPs may share a similar substrate recognition mechanism with Wzb. The current studies provide the structural basis for rational drug design against the pathogenic bacteria.

around the active site P-loop. Our results reveal a novel mechanism of the substrate recognition for LMW-PTPs.

EXPERIMENTAL PROCEDURES
Sample Preparation-The wzb gene from E. coli was amplified by PCR and cloned into the plasmid pET28a(ϩ) (Novagen) with an N-terminal His tag and expressed in E. coli BL21(DE3). Cells were allowed to grow overnight in 25 ml of Luria-Bertani (LB) medium at 37°C and subsequently transferred into 1 liter of LB medium. When the A 600 reached around 0.6, cells were spun down and resuspended in 250 ml of M9 minimal medium containing 15 N-labeled ammonium chloride in the presence or absence of 13 C-labeled glucose for 15 N/ 13 C-labeled or 15 N-labeled samples, respectively (20). The resuspended cells were allowed to grow in minimal medium for 1 h before induction. The protein expression was induced by adding isopropyl ␤-Dthiogalactoside to a final concentration of 100 mg/liter. After 6 h of induction at 20°C, the cells were harvested. The soluble protein was purified by the affinity chromatography (nickelnitrilotriacetic acid-agarose column), and subsequently gel filtration (Superdex-120 column) with theÄKTA FPLC system (Amersham Biosciences). Only a single peak corresponding to the monomeric molecular weight was observed during gel filtration. The purity was determined to be greater than 95% as judged by SDS-PAGE. NMR samples were prepared in a buffer containing 50 mM sodium phosphate, 50 mM NaCl, 25 mM dithiothreitol, 5 mM EDTA, 0.02% sodium azide at pH 6.0 in 95% H 2 O, 5% D 2 O. 2,2-Dimethyl-2-silapentane-5-sulfonic acid was added as the internal chemical shift reference. For phosphate titration experiments, Wzb protein (0.1 mM) was first dissolved in a buffer containing 20 mM citric acid, 50 mM NaCl at pH 6.0. During the titration experiments, the concentration of phosphate ions in the Wzb solution was gradually increased and the titration was monitored by a series of two-dimensional 15 N-edited HSQC spectra.
NMR Spectroscopy-The NMR experiments for obtaining the chemical shift assignments and distance restraints were carried out at 25°C on Bruker Avance 500 MHz (with cryoprobe) and 800 MHz spectrometers equipped with triple-resonance probes and pulsed-field gradients. The sample concentration was about 1.0 mM. Nearly complete backbone resonance assignments for Wzb were obtained using a series of three-dimensional spectra HNCA, HN(CO)CA, HNCO, HN(CA)CO, HNCACB, and CBCA(CO)NH and almost complete side chain resonance assignments were obtained using the three-dimensional spectra HBHA(CO)NH, 15  The three-dimensional 15 N-or 13 C-edited NOESY-HSQC spectra (mixing times of 60 and 120 ms) were collected to confirm the resonance assignments and to obtain the distance restraints. The spectra were processed using the software package NMRPipe (22) and analyzed using NMRView (23). A semiautomatic strategy was followed to assign the backbone chemical shifts using the software MARS (24). All backbone amide chemical shifts were obtained except Ile 13 that was invisible in the spectra. The hydrogen bonding information was obtained by the 1 H/ 2 H exchange experiments. In addition, a set of 1 H-15 N residual dipolar coupling (RDC) constants were meas-ured by dissolving Wzb protein in a dilute liquid crystal buffer containing a mixture of alkyl-poly(ethylene glycol) C12E5 and n-hexanol (25). The C12E5/water ratio was 5.5 weight %, and the molar ratio of C12E5 to n-hexanol was 0.92. The RDCs were extracted from the difference in 1 H-15 N splitting measured by 1 H-15 N IPAP-HSQC spectra between the weakly aligned and the isotropic samples at 25°C (26). The data were analyzed using software packages PALES (27) and MODULE (28).
Structure Calculations-The initial structure calculations were performed using the automatic nuclear Overhauser effect (NOE) assignment program CANDID embedded in CYANA (29). The extended NOE contacts and chemical shift assignments were obtained by an iterative analysis of the calculated structures and NOE assignments using the program SANE (30). The NOE-derived distance restraints, in combination with the backbone dihedral angles determined by TALOS (31) and hydrogen bond information, were used in each calculation step. Two hundred structures were calculated using CYANA and the 100 structures with the lowest target function were selected for further refinement by AMBER7 (32). The 1 H-15 N RDCs were used as restraints during the refinement stage. As recommended in the AMBER7 manual, explicit angle restraints were added to enforce the ideal backbone geometry for residues with RDC restraints. Finally, the 20 lowest energy structures were selected to represent Wzb. The structures were analyzed using the software packages MOLMOL (33) and PROCHECK_NMR (34).
Structure-based Multiple Sequence Alignment-The LMW-PTPs with known tertiary structures were aligned using STRAP. LMW-PTPs of unknown tertiary structures were subsequently added in the multiple sequence alignment using the program ClustalX (35).
Accession Numbers-The chemical shifts have been deposited in the BioMagResBank data base under accession number 6934. The coordinates of Wzb structures have been deposited in the Protein Data Bank with accession number 2FEK.

RESULTS
The Overall Structure of E. coli Wzb-The solution structure of E. coli Wzb was determined using the NOE-derived distance restraints in conjunction with the dihedral angle, RDC, and hydrogen bond restraints. The 20 lowest energy structures were selected to represent Wzb and are shown in Fig. 1A, together with the ribbon diagram showing the secondary structural elements ( Fig. 1, B and C). The structural statistics are listed in Table 1. The root mean square deviation of the backbone heavy atoms is 0.90 Ϯ 0.23 Å for all residues and 0.50 Ϯ 0.09 Å for those in secondary structural elements. Ramachandran plots of all 20 structures indicate that most of the backbone dihedral angles lie within the energetically favorable regions of the (, ) space.
The Active Site-The P-loop of Wzb contains the amino acid sequence 9 CVGNICRS 16 , where the underlined residues are highly conserved among all identified LMW-PTPs (Fig. 2). The P-loop connects strand ␤ 1 and ␣-helix ␣ 1 (Fig. 1B) and is buried at the bottom of a crevice (Fig. 3), which is delineated by residues Ala 39 , Leu 40 , Ala 45 , and Asp 46 in loop ␤ 2 -␣ 2 , Glu 83 and His 86 in helix ␣ 3 , and Asp 115 to Tyr 117 in loop ␤ 4 -␣ 5 . Residues Asp 115 , Glu 83 , and those in the P-loop form the upper and center part of the crevice, whereas residues in loops ␤ 2 -␣ 2 and ␤ 4 -␣ 5 contribute to the lower part as viewed in Fig. 3. The side chains of Tyr 117 and Leu 40 , together with the main chain of Ala 45 and Asp 46 form a ring around the side chain of Ile 13 below the P-loop. To confirm the binding of the P-loop of Wzb with phosphate ions, the phosphate titration experiments were performed and monitored by a series of two-dimensional 15 N-edited HSQC experiments. In the absence of phosphate ions, 13 residues (Cys 9 -Asn 12 , Cys 14 -Ser 16 , Glu 20 , Gly 36 , Leu 37 , Leu 40 , Arg 66 , and Tyr 117 ) of Wzb completely disappeared in the 15 Nedited HSQC spectrum, whereas the remaining cross-peaks in the spectrum were almost identical to that in the presence of phosphate ions (Fig. 4A). Upon addition of phosphate, all missing cross-peaks showed up and their intensities gradually increased with the increase of phosphate concentration. After the phosphate concentration was higher than 20 mM, the peak intensities in the HSQC spectra remained unchanged. This result confirmed that the phosphate binds to Wzb and stabilizes the P-loop, and indicated that phosphate binding does not significantly alter the overall structure of Wzb except the P-loop region in which most residues are missing from the two-dimensional 15 N-edited HSQC spectrum in the absence of phosphate ions. Nevertheless, we compared the P-loop region of Wzb with that of the crystal structure of phosphate-bound BPTP (Protein Data Bank code 1PHR) (36) and found that all the NH bonds of Wzb whose NMR signals were missing in the absence of phosphate point toward the position where the phosphate group is expected to bind (Fig.  4B). Based on this, the inorganic phosphate in Wzb likely occupies a similar position as that in other LMW-PTPs. In addition, the phosphate titration experiments suggest that the P-loop undergoes significant mobility (conformational exchanges) in the absence of phosphate, which causes the broadening of the NMR signals and makes them undetectable. The conformational exchanges are largely diminished after binding with phosphate ion, which is further demonstrated by the results of backbone dynamics (supplemental data).

DISCUSSION
Structural Basis of the Catalytic Mechanism-We have solved the solution structure of E. coli Wzb, the first LMW-PTP structure from the Gram-negative bacteria. To date, structures of LMW-PTPs from eukaryotic organisms, such as human HCPTPA and HCPTPB (9,37), bovine BPTP (36,38,39), and yeast Saccharomyces cerevisiae LTP1 (10) are available. The structures of the prokaryotic LMW-PTPs MPtpA from Mycobacterium tuberculosis (40) and YwlE from the Grampositive Bacillus subtilis (41) have been reported recently. Although Wzb shares a low sequence identity (less than 30%) with these LMW-PTPs, it shows a high structural similarity with them, especially the highly conserved regular secondary structures. Residues Cys 9 -Ser 16 of the catalytic site fold into a typical P-loop conformation, and the root mean square deviation values of the backbone atoms of these residues with those of the equivalent residues (Cys 12 -Ser 19 in BPTP and HCPTPA, Cys 13 -Ser 20 in LTP1) in eukaryotic LMW-PTPs range from 0.30 to 0.35 Å. In addition, most of the main chain NH bonds in the P-loop point toward the expected position of the phosphate group. Furthermore, the side chain of the strictly conserved asparagine residue Asn 12 is spatially closed to that of His 63 (the distance between N ␦2 of Asn 12 and N ␦2 of His 63 is d(N12 N␦2 , H63 N␦2 ) ϭ 3.4 Ϯ 0.4 Å), Ser 16 (d(N12 N␦2 , S16 O␥ ) ϭ 3.7 Ϯ 0.8 Å), and Ser 34 (d(N12 N␦2 , S34 O␥ ) ϭ 4.9 Ϯ 0.5 Å). A similar geometry is observed in BPTP, HCPTPA, and LTP1 (residues Asn 15 , His 72 , Ser 19 , and Ser 43 , respectively in BPTP). These distances suggest the presence of a hydrogen-bonding network involving Asn 12 in Wzb, which may stabilize the backbone conformation of the P-loop, as previously observed in BPTP (38). On the upper side of the P-loop as viewed in Fig. 3, the guanidinium moiety of the conserved Arg 15 is in a favorable distance with the acidic groups of Asp 115 and Glu 83 to form hydrogen bonds or salt bridges. Moreover, the H ⑀ atom of Arg 15 in Wzb is superimposable with that of the equivalent arginine in BPTP (Arg 18 ), HCPTPA (Arg 18 ), and LTP1 (Arg 19 ). In those LMW-PTPs, the H ⑀ atom of the arginine likely forms a hydrogen bond with an oxygen atom of the phosphate substrate. As a consequence, residue Arg 15 in Wzb most likely contributes to the recognition and position of the substrate in a similar way as eukaryotic LMW-PTPs. The acidic group of residue Asp 115 in Wzb is close to the guanidinium moiety of Arg 15 and is in a favorable position to provide a proton to the leaving phenolic group. This aspartic residue is believed to serve as a general acid in the reaction (8). In Wzb, the Cys 9 side chain is directed toward the phosphate binding site and the proximity of the hydroxyl group of Ser 16 to Cys 9 side chain (d(C9 S␥ , S16 O␥ ) ϭ 3.9 Ϯ 0.8 Å) suggests hydrogen bonding. In BPTP, Cys 12 (equivalent to Cys 9 in Wzb) forms a covalent phosphocysteine intermediate during the reaction (42) and is likely activated by hydrogen bonding with Ser 19 (equivalent to Ser 16 in Wzb) (43). As a consequence, Cys 9 may be activated by Ser 16 in Wzb. The fact that Wzb shares high structural similarity with the eukaryotic LMW-PTPs strongly suggests that the dephosphorylation mechanism of Wzb is similar to that of other LMW-PTPs (7). Briefly, the phosphoryl group of the phosphorylated Wzc undergoes a nucleophilic attack by the active site Cys 9 side chain of Wzb. The guanidinium moiety of Arg 15 stabilizes the phosphoenzyme intermediate state and the invariant aspartate residue Asp 115 protonates the ester oxygen of the leaving group. In the following step, the phosphoenzyme is hydrolyzed by the nucleophilic attack of a water molecule, possibly assisted by Asp 115 . Finally, one cycle of the dephosphorylation is finished with the release of the inorganic phosphate (7).
Substrate Recognition-Previous structural studies on eukaryotic LMW-PTPs in complex with exogenous substrates identified two regions around the catalytic center that are important for the enzyme activity (9, 10). In loop ␤ 4 -␣ 5 of the eukaryotic LMW-PTPs, an aromatic ring (Tyr 131 in HCPTPA and BPTP, Trp 134 in LTP1) was found to stabilize the substrate by interacting with the aromatic ring of the phosphotyrosine from one side. In Wzb, Tyr 117 is structurally equivalent to Tyr 131 in BPTP (Fig. 5A) and likely plays a similar role in substrate recognition and stabilization by providingstacking contacts with the phenolic ring of the substrate. Notably, an aromatic residue is conserved at this position among all bacterial LMW-PTPs with a preference for tyrosine (Fig. 2). This fact suggests that this kind of ringring contact is essential for the function of LMW-PTPs in both prokaryotes and eukaryotes. On the other side of the substrate, the loop ␤ 2 -␣ 2 of eukaryotic LMW-PTPs contains another aromatic residue (Tyr 49 in HCPTPA, Trp 49 in BPTP, and Tyr 51 in LTP1) that provides contacts with the substrate and closes the claw around it (9, 10). However, the loop ␤ 2 -␣ 2 of Wzb does not contain an aromatic residue that is equivalent to Trp 49 in BPTP. Consequently, loop ␤ 2 -␣ 2 is shorter by 1 residue in Wzb compared with the eukaryotic LMW-PTPs (Fig. 2). As a result, the C ␣ traces diverge significantly between Wzb and BPTP in this region (Fig. 5A). Furthermore, residues Glu 50 and Asn 53 in human and rat LMW-PTPs have been demonstrated to play a critical role in substrate specificity (9,44). In Wzb, this position is occupied by hydrophobic residue Leu 40 . The side chain of Leu 40 in Wzb delineates one side (Fig. 3) of the substrate entry site and is superimposable with the side chains of Asn 50 in BPTP (Fig. 5A) and Glu 50 in HCPTPA. As a consequence, Leu 40 may play a role in recognizing a hydrophobic patch on the substrate. The structural differences are more obvious between segments Gly 38 -Ala 39 of Wzb and Ser 47 -Trp 49 of BPTP. In eukaryotic LMW-PTPs, such as BPTP, this segment loops out and exposes the side chains of Asp 48 and Trp 49 to the solvent. In contrast, the methyl group of Ala 39 in Wzb points . e Include NOE, dihedral angle, and RDC violation energy. f The backbone heavy atoms including C ␣ , amide nitrogen, and carbonyl carbon atoms. g All heavy atoms including all the non-hydrogen atoms of the protein. JULY 14, 2006 • VOLUME 281 • NUMBER 28 toward the interior of the structure and is in close contact with the side chain of Asn 12 in the P-loop and the backbone atoms of Arg 66 . This structural arrangement has two consequences. (i) No aromatic residue in Wzb is equivalent to Trp 49 in BPTP, which provides the ring-ring interactions on one side of the substrate ring. This factor may also contribute to the relatively lower substrate affinity of Wzb (K m of 1 mM) (13) compared with BPTP (0.38 mM) and LTP1 (0.017 mM) (7) using p-nitrophenyl phosphate as the substrate. (ii) The molecular surface of the lower part of the crevice of Wzb is significantly different from that of BPTP (Fig. 5B). In Wzb, the backbone carbonyl atom of Gly 38 delineates one part of the substrate entry. This group is directed toward the expected position of the phosphotyrosine substrate and exposes a polar patch to the substratebinding site. The side chains of Leu 40 and Tyr 117 , together with the polar carbonyl atom of Gly 38 are potential candidates to determine the substrate specificity of Wzb. Overall, the substrate entry is slightly bigger in Wzb than that in BPTP. However, the catalytic center of Wzb is only accessible for long side chain atoms. This may explain why Wzb is inactive on phosphoserine or phosphothreonine (13). In addition, six charged residues (Lys 43 , Glu 83 , Arg 85 , Arg 89 , Arg 118 , and His 86 ) are spatially close to the catalytic center and are within reasonable distances to interact with residues adjacent to the target phosphotyrosine in the endogenous substrate, and will be discussed below.

Solution Structure of a LMW-PTP from E. coli
The autokinase Wzc is the endogenous substrate of Wzb and the Wzb/Wzc pair regulates the production of colanic acid in vivo (14). Wzb was shown to exclusively and extensively dephosphorylate the phosphotyrosines of the C-terminal tyrosine-rich cluster in Wzc (Tyr 708 -Gly-Tyr 710 -Tyr 711 -Glu-  Tyr 713 -Glu-Tyr 715 -Lys-Ser-Asp-Ala-Lys) (14,15). The E. coli Etp/Etk and the Acinetobacter johnsonii Ptp/Ptk are pairs of LMW-PTP/kinase and are the homologs of Wzb/Wzc (16,45). Interestingly, Etp and Ptp could also dephosphorylate the phosphorylated Wzc in vitro to the same extent as Wzb (13,14). Although the kinetic data of the dephosphorylation reactions are not available, these experiments provide a basis for understanding the determinants on Wzc dephosphorylation by LMW-PTPs. Etp and Ptp share 56 and 35% sequence identity and 73 and 54% sequence similarity with Wzb, respectively. Regions with the highest sequence similarity among the three LMW-PTPs include the catalytically essential P-loop, the DPY (Asp 115 -Pro 116 -Tyr 117 in Wzb) segment in loop ␤ 4 -␣ 5 and buried residues (Fig. 2). In addition, both Etp and Ptp also possess a hydrophobic side chain (leucine or methionine) like Leu 40 in Wzb. This observation supports the role of Leu 40 in Wzb on substrate recognition. In contrast, residues Lys 43 , Glu 83 , Arg 85 , Arg 89 , Arg 118 , and His 86 that surround the active site of Wzb are neither conserved nor replaced by similar amino acids in Etp and Ptp (Fig. 2). This fact suggests that electrostatic interactions are not essential for the recognition of residues adjacent to the phosphotyrosines in Wzc. Notably, Asn 53 was proposed to drive the substrate specificity of HCPTPA (9) and this position is occupied by Lys 43 and Tyr 44 in Wzb and Ptp, respectively. This may suggest that Lys 43 in Wzb plays a dispensable role in substrate recognition, which is different from residue Asn 53 in HCPTPA.
In addition, the sequences containing the phosphotyrosine residues recognized by prokaryotic and eukaryotic LMW-PTPs are distinct. Wzb has been shown to target the C-terminal tyrosine-rich tail in Wzc (YGYYEYEYKSDAK) (14,15). Interest-ingly, the homologous proteins of Wzc in prokaryotes, including Etk from E. coli and other predicted protein-tyrosine kinases (AmsA from plant pathogen Erwinia amylovora and Orf6 from human pathogen Klebsiella pneumoniae) all share a C-terminal tyrosine-rich tail with similar sequences (YSYGY-NYYGYSYSEKE for Etk, YGYGYNYYDYSYSDKK for Orf6, and YGYGYDYYDYSYQQGEKS for AmsA, respectively) (16). However, these tyrosine-rich sequences are generally not found in tyrosine kinase receptors targeted by the eukaryotic LMW-PTPs. The sequences recognized by the eukaryotic LMW-PTPs usually contain a single tyrosine residue flanked by charged amino acids. In contrast, the sequences with multiple tyrosine residues recognized by prokaryotic LMW-PTPs (i.e. Wzb, Etp, and Ptp) contain less charged amino acids and are more hydrophobic. This may explain the fact that hydrophobic residues like Leu 40 in Wzb are engaged in substrate specificity of prokaryotic LMW-PTPs, and the charged residues around the active site are dispensable in substrate recognition.
A Novel Substrate Recognition Mechanism in Prokaryotic LMW-PTPs-E. coli Wzb and B. subtilis YwlE share high structural similarity near the active site. Similar to Wzb, no aromatic residue equivalent to Trp 49 in BPTP is present in YwlE (41). In addition, the Phe 40 backbone carbonyl atom and the side chains of Ala 41 and Ser 42 of YwlE are superimposable with the carbonyl atom of Gly 38 and the side chains of Ala 39 and Leu 40 of Wzb, respectively. In eukaryotic LMW-PTPs, two aromatic rings sandwich the phosphotyrosine. In contrast, in both Wzb and YwlE, a polar group exposed by a backbone segment provides a potential contact with the substrate. Taken together, the comparison of the structures suggests that Wzb and YwlE likely adopt a novel mechanism on substrate recognition that is different from the eukaryotic LMW-PTPs. Further sequence alignment and structural analysis of other LMW-PTPs suggest that some LMW-PTPs from prokaryotes may share the same substrate recognition mechanism as Wzb. The absence of an aromatic residue equivalent to Trp 49 in BPTP, and therefore a shorter loop ␤ 2 -␣ 2 are the sequence characteristics to distinguish the substrate recognition mechanisms between the prokaryotic-like (Wzb and YwlE) and eukaryotic-like LMW-PTPs (BPTP). Based on the above results, we suggest that the LMW-PTPs can be further classified into two groups with distinct substrate recognition mechanisms: those sharing the similar recognition mechanism with eukaryotic LMW-PTPs such as BPTP can be referred as class I LMW-PTPs, and those sharing the similar recognition mechanism with prokaryotic Wzb can be referred as class II LMW-PTPs. An unrooted phylogenetic tree inferred from multiple sequence alignment (see supplemental data) shows that the two classes of LMW-PTPs diverged during evolution and hence support the proposed classification.
A closer examination of the sequence alignment provides additional information regarding the differences between class I and class II LMW-PTPs. In fact, all the class I LMW-PTPs possess a tyrosine at the position equivalent to Try 132 in the BPTP sequence, whereas this position may accommodate various amino acids (i.e. lysine, arginine, and glycine) in class II LMW-PTPs. The phosphorylation state of Tyr 132 plays an important physiological role in regulating the activity of the protein BPTP and controlling the strength of NIH-3T3 cell substrate adhesion (46,47). The absence of a tyrosine at this posi-tion in class II LMW-PTPs suggests that their enzyme activities might be regulated by other processes (48).
Furthermore, the structurebased multiple sequence alignment suggests that both classes of LMW-PTPs exist in the prokaryotic cells. The determinants for substrate specificity of Class II LMW-PTP (Wzb-like) are shared by other prokaryotic LMW-PTPs such as Etp, Ptp, Yor5, EpsP, Amsi, and Staphylococcus aureus PtpB (Fig. 2). In contrast, the prokaryotic M. tuberculosis MPtpA, Streptomyces coelicolor PtpA, and S. aureus PtpA belong to Class I LMW-PTPs and share a similar substrate recognition mechanism with the eukaryotic LMW-PTPs. Notably, MPtpA is secreted into the extracellular medium and was shown to inhibit phagocytosis of mycobacteria (49,50). Because MPtpA targets components of the eukaryotic host signaling pathways, its similarities with eukaryotic LMW-PTPs may be of biological significance. Further investigations are expected to elucidate the biological relevance of the two classes of LMW-PTPs with different substrate recognition mechanisms in prokaryotes.
Concluding Remarks-Our current studies demonstrate that a class of prokaryotic LMW-PTP adopts a substrate recognition mechanism different from that of the eukaryotic LMW-PTPs. Although all LMW-PTPs known so far may share many determinants in substrate recognition, class II LMW-PTPs probably partly recognize their endogenous substrates by hydrophobic side chains and a backbone carbonyl group, instead of a aromatic ring in class I LMW-PTPs. Many class II LMW-PTPs such as Wzb, involve in the positive regulation of capsules and/or exopolysaccharide that are important for bacterial pathogenesis (51). Therefore, the Wzb structure provides a clue for the rational design of drugs that are highly specific to bacterial virulence factors without leading to side effects in the eukaryotic organisms.