Inhibition of Yersinia tyrosine phosphatase by furanyl salicylate compounds.

To avoid detection and targeting by the immune system, the plague-causing bacterium Yersinia pestis uses a type III secretion system to deliver a set of inhibitory proteins into the cytoplasm of immune cells. One of these proteins is an exceptionally active tyrosine phosphatase termed YopH, which paralyzes lymphocytes and macrophages by dephosphorylating critical tyrosine kinases and signal transduction molecules. Because Y. pestis strains lacking YopH are avirulent, we set out to develop small molecule inhibitors for YopH. We used a novel and cost-effective approach, in which leads from a chemical library screening were analyzed and computationally docked into the crystal structure of YopH. This resulted in the identification of a series of novel YopH inhibitors with nanomolar Ki values, as well as the structural basis for inhibition. Our inhibitors lack the polar phosphate-mimicking moiety of rationally designed tyrosine phosphatase inhibitors, and they readily entered live cells and rescued them from YopH-induced tyrosine dephosphorylation, signaling paralysis, and cell death. These inhibitors may become useful for treating the lethal infection by Y. pestis.


From the Infectious and Inflammatory Disease Center, The Burnham Institute, La Jolla, California 92037
To avoid detection and targeting by the immune system, the plague-causing bacterium Yersinia pestis uses a type III secretion system to deliver a set of inhibitory proteins into the cytoplasm of immune cells. One of these proteins is an exceptionally active tyrosine phosphatase termed YopH, which paralyzes lymphocytes and macrophages by dephosphorylating critical tyrosine kinases and signal transduction molecules. Because Y. pestis strains lacking YopH are avirulent, we set out to develop small molecule inhibitors for YopH. We used a novel and cost-effective approach, in which leads from a chemical library screening were analyzed and computationally docked into the crystal structure of YopH. This resulted in the identification of a series of novel YopH inhibitors with nanomolar K i values, as well as the structural basis for inhibition. Our inhibitors lack the polar phosphate-mimicking moiety of rationally designed tyrosine phosphatase inhibitors, and they readily entered live cells and rescued them from YopHinduced tyrosine dephosphorylation, signaling paralysis, and cell death. These inhibitors may become useful for treating the lethal infection by Y. pestis.
To survive in humans, pathogenic bacteria have evolved numerous mechanisms to evade the immune response in the host (1,2). One of the most successful strategies was adopted by Yersinia pestis, namely a type III secretion system that injects a set of paralyzing proteins directly into the cytoplasm of macrophages and lymphocytes that the bacterium encounters in the lymph nodes of infected individuals (3,4). As a result, the targeted cells become unable to respond, and the bacteria can multiply unopposed by the normal mechanisms of host defense.
The natural route of Y. pestis infection is by transmission from infected rats or other animals by blood-sucking fleas, which are weakened by the bacteria in their gut and therefore expel bacterial mass into the epidermis of their next victim when trying to feed (5,6). From these flea bites, the bacteria travel to local lymph nodes (7)(8)(9), where they multiply and cause a massive lymphadenitis within 2-6 days (5). These enlarged and painful lymph nodes, or "bubos," give the disease its common name Bubonic Plague. Unless treated with high dose streptomycin-or tetracycline-type antibiotics during the first few days, the infection develops into a toxemic sepsis, which is often fatal (5,6). A normally very rare, but much more rapidly lethal, form of the infection is caused by inhaled bacteria and is referred to as pneumonic plague or plague pneumonia (10). By this route of infection, the number of bacteria entering the body can be much larger than from microscopic flea bites, and the bacteria are efficiently disseminated to the peritracheal, mediastinal, and other central lymph nodes, from which they gain access to the bloodstream much earlier. Although several vaccines exist (11,12), and Yersinia usually is sensitive to antibiotics, the pneumonic form of the disease is difficult to diagnose and still often results in death (10).
Despite efforts to eradicate the disease, natural reservoirs of Y. pestis still exist in wild rats and other rodent populations in parts of Africa, southeast Asia, and southwestern United States (13), and sporadic human cases of plague still occur every year. Although these cases pale by comparison to the devastating pandemics that killed an estimated 200 million people, mostly in Europe, during historical times (5,6), the World Health Organization now recognizes plague as a reemerging public health concern. There are also increasing fears that Y. pestis may be used for biological warfare or bioterrorism (14 -16). The potential threat is heightened by the existence of multidrug-resistant strains of Y. pestis (17,18) and the rapidly lethal course of the pneumonic form of the disease caused by aerosolized Yersinia. Clearly, new approaches to combat plague are urgently needed.
The molecular mechanisms employed by all virulent strains of Y. pestis and the two related species, Yersinia pseudotuberculosis and Yersinia enterocolitica, are based on an extrachromosomal virulence plasmid (19), which encodes a type III secretion system and several effector proteins called Yops (Yersinia outer membrane proteins) (20). The type III secretion system is a highly conserved macromolecular machinery found in many pathogenic Gram-negative bacteria and is induced by contact with a eukaryotic cell to inject effector Yops into the cytoplasm of the target cells (21). In the host cell, the Yops disrupt signaling cascades responsible for initiating key immune functions, such as phagocytosis (22)(23)(24), respiratory burst (25,26), cytokine production, and lymphocyte activation (27). As a consequence, both the innate and adaptive immune responses are seriously impaired (28). However, a protective immunity can be acquired by vaccination (11,12).
A key Yop protein is YopH, a 468-amino acid, exceptionally active protein-tyrosine phosphatase (PTP) 1 (29,30) with a Cterminal catalytic domain and a multifunctional N-terminal domain, which binds tyrosine-phosphorylated target proteins (31,32). The catalytic domain of YopH is structurally similar to that of eukaryotic PTPs (33). A marked dephosphorylation of proteins in human epithelial cells and murine macrophages has been observed during infection with live bacteria (24,30,34,35). In macrophages and neutrophils, the targets include the focal adhesion proteins Cas, focal adhesion kinase, and paxillin (22,23), providing a molecular mechanism for inhibition of migration and phagocytosis by these cells (22,23,37,38).
YopH also inhibits the activation of T and B lymphocytes (27,39). We recently reported (39) that YopH in T cells directly dephosphorylated the Src family tyrosine kinase Lck at its positive regulatory site, Tyr-394, resulting in a complete loss of Lck activity. Because this kinase is the first upstream signalgenerating molecule for the T cell antigen receptor, signaling from this receptor was completely abrogated. As a consequence, all tyrosine phosphorylation of downstream signaling proteins was inhibited; the T cells failed to form immune synapses with antigen-presenting cells, and they were unable to secrete any interleukin-2 into the medium (39). Similarly, T cells exposed to live Y. enterocolitica became unable to flux calcium and produce cytokines (40).
Because Yersinia strains that carry a pYV plasmid with a nonfunctional yopH gene are avirulent (41)(42)(43)(44) and even a point mutation that changes the catalytic Cys-403 to an alanine eliminates the virulence of Y. pseudotuberculosis in a murine infection model (30,34), it is clear that the catalytic activity of YopH is critical for the lethality of Yersinia infection. We therefore set out to develop small molecule inhibitors of YopH by a combination of chemical library screening, structure-activity analysis, and in silico docking of lead compounds.

MATERIALS AND METHODS
Reagents-p-Nitrophenyl phosphate (pNPP) was purchased from Sigma. BIOMOL GREEN TM reagent was from BIOMOL Research Laboratories (Plymouth Meeting, PA). All other chemicals and reagents were of the highest grade available commercially. Anti-phosphotyrosine mAb 4G10 was from Upstate Biotechnology, Inc. (Lake Saranac, NY), and mAb PY20 was from BD Biosciences.
Chemical Library Screening for YopH Inhibitors-A subset of 10,000 compounds from the DIVERSet TM library of 50,000 drug-like molecules (ChemBridge, Inc.) was screened in a 96-well format in vitro assay. Each reaction contained 50 nM YopH, 1 mM pNPP, and 0.03 mg/ml compound in 0.1 M BisTris, pH 6.0, reaction buffer. The final volume amounted to 50 l and contained 2% Me 2 SO. The reaction was initiated by addition of pNPP after a preincubation of the enzyme with the compounds for 10 min at room temperature. After 7 min, the reaction was quenched by addition of 100 l of BIOMOL GREEN TM reagent, and the pNPP hydrolysis was determined by measuring the absorbance of the complexed free phosphate at 620 nm. The nonenzymatic hydrolysis of the substrate was corrected by measuring the control without addition of enzyme. To quantitate the inhibitory efficacy of the library compounds, we determined the ratio of inhibition in comparison to 200 M orthovanadate, a PTP inhibitor. Every compound with a ratio of Ͼ1 was considered as a hit. ClogP for each compound was calculated with ChemDraw8.
K i Determination-The YopH PTP-catalyzed hydrolysis of pNPP in the presence of inhibitors was assayed at 30°C in 0.1 M BisTris, pH 6.0, assay buffer containing 1 mM dithiothreitol and 5% Me 2 SO. The ionic strength was adjusted to 150 mM with NaCl. The enzyme was preincubated with various fixed concentrations of inhibitors for 10 min. The reaction was initiated by the addition of various concentrations of pNPP (ranging from 0.2 to 10 K m ) to the reaction mixtures to a final volume of 100 l. The reaction was quenched by addition of 100 l of 1 M NaOH. The nonenzymatic hydrolysis of the substrate was corrected by measuring the control without addition of enzyme. The amount of product p-nitrophenol was determined from the absorbance at 405 nm detected by a PowerWaveX340 microplate spectrophotometer (Bio-Tek Instruments, Inc.) using a molar extinction coefficient of 18,000 M Ϫ1 cm Ϫ1 . The inhibition constant and inhibition pattern were evaluated by fitting the data to the Michaelis-Menten equations for either competitive (Equation 1), uncompetitive (Equation 2), or mixed (Equation 3) inhibition, using nonlinear regression and the program GraphPad Prism® In the case of the mixed inhibition model, K ic is the inhibition constant for the competitive participation, and K iu is the inhibition constant for the uncompetitive participation. For a comparison of the fitting results, the second-order Akaike's Information Criterion (AICc) was calculated with Equation 4, where N is the number of data points, SS the absolute sum of squares, and K the number of parameters fit by nonlinear regression plus 1.
The probability to have chosen the right model can be computed by Equation 5, where ⌬ is the difference between Akaike's Information Criterion (AICc) scores.
probability ϭ exp(Ϫ0.5⌬)/(1 ϩ exp(Ϫ0.5⌬)) (Eq. 5) IC 50 Measurements-The PTP-catalyzed hydrolysis of pNPP in the presence of inhibitor was assayed at 30°C in a 100-l reaction system in the same assay buffer described above. At various concentrations of the compound, the initial rate at fixed pNPP concentrations (equal to the corresponding K m values for each PTP) was measured by determining the free phosphate with the BIOMOL GREEN TM reagent, as described above. The IC 50 value was determined by plotting the relative pNPP activity versus inhibitor concentration and fitting to Equation 6 using GraphPad Prism®.
In this case, V i is the reaction velocity when the inhibitor concentration is [I]; V 0 is the reaction velocity with no inhibitor, and IC 50 ϭ K i ϩ K i [S]/K m . Molecular Modeling-Molecular modeling studies were conducted on several R12000 SGI Octane workstations with the software package Sybyl version 6.9 (TRIPOS). Energy-minimized molecular models of the compounds were generated by the Sybyl/MAXIMIN2 routine.   mM NaCl, 5 mM EDTA containing 1% Nonidet P-40, 1 mM Na 3 VO 4 , 10 g/ml aprotinin and leupeptin, 100 g/ml soybean trypsin inhibitor, and 1 mM phenylmethylsulfonyl fluoride and clarified by centrifugation at 15,000 rpm for 20 min. Lysate was mixed with an equal volume of twice concentrated SDS sample buffer, boiled for 1 min, and resolved by SDS-PAGE. SDS-PAGE and Immunoblotting-These procedures were done as before (39).
Interleukin-2 Secretion Assay-5 ϫ 10 6 human T lymphocytes were treated with 6 M ANT-YopH for 5 h at 37°C in RPMI medium, washed, and stimulated with C305, anti-CD28 mAb, plus a cross-linking antimouse Ig for 15 h in 250 l of medium with 10% fetal calf serum. 20 l of the supernatant was used for measurement of the amount of interleukin-2 using an enzyme-linked immunosorbent assay kit from Roche Applied Science, as before (39). Results are given as pg/ml of secreted interleukin-2/2 ϫ 10 5 cells.

Identification of Lead Compounds by Chemical Library
Screening-A 96-well format in vitro assay was used to screen the first 10,000 compounds of the DIVERSet TM library (Chem-Bridge, Inc.) of drug-like compounds. A total of 10 compounds inhibited YopH to a higher extent than 200 M orthovanadate, a general PTP inhibitor. After determination of the kinetic parameters of these 10 first hits, we selected four compounds, which were showing a competitive or mixed inhibition pattern with a K i value Ͻ10 M, for closer inspection. Two of these, a quinone and a charged thio-imidazole, were subsequently discarded, and we focused on a 5-methylenethiazolidine-2,4-dione linked via a furanyl ring to a salicylic acid (compound 1) and a 5-methylene-2-thioxothiazolidin-4-one (additionally substituted at the nitrogen atom) similarly linked via a furanyl ring to a nitrophenol (compound 2). These two inhibited YopH with competitive K i values of 0.311 Ϯ 0.053 and 1.87 Ϯ 0.924 M, respectively, and were relatively selective for YopH (Table I). The structures of compounds 1 and 2 are shown in Table II. These two represent our initial hits from the chemical library screen.
Structure-Activity Relationship Analysis of Compound 1 and 2-Encouraged by these findings, we investigated the structure-activity relationships for a total of 61 analogs that all contained a substituted phenyl ring linked via a furanyl moiety to a more diverse entity at the other end of the molecule, preferentially a 5-methylenethiazolidine ring. A total of 44 compounds inhibited YopH in a competitive manner with K i values of Ͻ100 M. The structures and kinetic data for a representative set of 21 analogs are given in Table II. Significantly, 13 of the 16 salicylic acid analogs were among the 26 best compounds (competitive inhibition with K i values Ͻ10 M). Elimination of the salicylic acid moiety (compound 23) led to a complete loss of YopH inhibition, whereas salicyl furyl aldehyde (compound 13) shows good competitive inhibition (K i ϭ 2.08 M). However, salicylic acid alone was a poor inhibitor (K i ϭ 882 M) compared with the most potent inhibitor, compound 3, which has a 6,211-fold lower inhibitory constant (K i ϭ 0.143 M). The very similar compound 7, which only differs in the location of a single methyl group, shows almost 10 times less activity, suggesting some steric constraints for the putative binding site or some very specific interactions of that methyl group. A comparison of the K i values for compounds 12 and 20, which differ by a methyl group in a very similar position, supports the latter.
Another example of a unique binding mode is presented by the second best inhibitor, compound 4 (K i ϭ 0.208 M). In this case, the structurally very similar compound 18 (K i ϭ 6.60 M), in which the positions of the carboxylic and the hydroxyl groups have been switched, has a 32 times higher K i value, supporting the notion of specific binding.
Substitution of the hydroxyl group in the original hit com-pound 1 by a chlorine in compound 16 led to a 14.5 times less inhibitory activity. Eliminating this hydroxyl group and shifting the carboxylic group to an ortho position (compound 21) results in a Ͼ300 times higher K i value. It is also worth mentioning that inhibitory activity is almost completely lost if the carboxylic and hydroxyl group are replaced by halogens, like fluorine and chlorine (compound 22), perhaps because these atoms are not able to mimic the phosphate group of a natural substrate. Our structure-activity relationship analysis resulted in the identification of two compounds, which inhibited YopH with lower K i values than the first furanyl salicylate hit, compound 1. A comparison of the Lineweaver-Burk plots of the four best inhibitors is shown in Fig. 1.
Virtual Docking Studies-To provide further insights into specific interactions of our inhibitors with the enzyme, we performed flexible ligand docking with the best four inhibitors and the x-ray coordinates of the catalytic domain of YopH. In all cases, there was a high degree of convergence for the salicylic-furanyl moiety, which occupied the deep hydrophilic phosphate binding cavity (catalytic pocket) on the surface of YopH (Fig. 2). The salicylic group was found to be involved in a complex network of hydrogen bonding interactions (Fig. 2 and Table III) that correlate very well with the less potent inhibition by compounds analogs that lack either the carboxylic or the hydroxyl group (Table II). In addition, the oxygen atom of the furanyl ring is also invariably involved in hydrogen bonding interactions with the side chains of Gln-357 or Arg-404 (Table III and Fig. 2, A, C, E, and G), both of which are unique to YopH among PTPs. From these docking studies, it is evident that most of the binding energy of the four inhibitors resides in the interactions with the furanyl and salicylate moieties. In fact, docking studies performed with a virtual compound containing only these two moieties gave similar binding energies, in agreement with experimental data with compound 13 (K i 2.08 M; Table II). In contrast, the in silico elimination of the carboxylic and/or the hydroxyl groups in the salicylate produced compounds that failed to dock in the catalytic pocket of the protein. For example, the binding energy of compound 4 dropped from Ϫ46 kJ/mol (Table III) to Ϫ21 kJ/mol after removal of the carboxylic and hydroxyl groups. The docking studies also showed that a denser network of hydrogen binding interactions forms within the YopH catalytic pocket compounds when the carboxylic acid in the salicylate is in para position with respect to the furanyl ring (Table III and Fig. 2, A, C, E,  and G).
In all four inhibitors, the positioning of the end of the molecule opposite from the salicylate was less defined among the 20 solutions generated with FlexX, correlating well with the high variability of tolerated substitutions at this position. However, a few important conclusions could be made. In compound 4, this moiety is involved in an additional hydrogen bonding interaction with the side chains of Gln-357 and Gln-446, whereas similar interactions occur with compound 5 and Arg-205 that could confer further affinity for YopH. The methyl groups in ring C of compound 3 make favorable steric contacts with YopH, in very close proximity to an additional groove (termed P2 in Fig. 2) on the surface of the protein. Based on the latter model, one could also predict that even small substitutions could result in unfavorable steric hindrance and decrease the binding affinity, as observed for compound 7 (Table II).
In silico docking was also used to evaluate if there is a structural basis for YopH selectivity compared with other PTPs. Although all PTPs have very similar catalytic cores, they differ dramatically in surface topology and charge distribution in the terrain that surrounds the catalytic pocket (50). Fig. 3 shows a comparison of the surface topology, surface electrostatic potential, and surface lipophilic potential of YopH with a set of six other PTPs also used in our selectivity assays. Each enzyme has a unique surface surrounding the catalytic pocket (indicated with a white circle in Fig. 3), with a different distribution, size, and shape of surface depressions and protrusions. In addition, each enzyme has a distinct surface charge and lipophilicity profile. These striking differences probably reflect preferences in substrate selection. It should also be possible to utilize these features for the development of small molecule inhibitors with a high degree of specificity. Indeed, attempts to dock our compounds into the crystal structures of PTP1B and VHR clearly demonstrated that none of our furanyl salicylate compounds fit into these PTPs (data not shown). For example, compound 3, which reaches toward pocket P2 in YopH, cannot reach any corresponding depression in PTP1B because access is blocked by a large protrusion (indicated by arrow in Fig. 4). Furthermore, VHR does not even have such additional pockets. VHR also has a much more shallow catalytic pocket, which is surrounded by less hydrophobic and more acidic surfaces. In support of this notion, IC 50 values of our inhibitors for a set of PTPs were 1-4 orders of magnitude higher than for YopH (Table IV). Thus, as predicted by the docking studies, the furanyl salicylates were relatively selective for YopH.
Restoration of Tyrosine Phosphorylation and T Cell Receptor Signaling in YopH-expressing Cells-Next, we tested the four inhibitors for their ability to reverse the effects of YopH in live T cells. First, we transfected Jurkat T leukemia cells with YopH expression plasmid or empty vector as a control, followed by addition of the YopH inhibitors to the cell cultures. 24 h later, the cells were stimulated with anti-CD3⑀ and anti-CD28 mAbs plus a secondary cross-linking antibody for 5 min at 37°C. Anti-Tyr(P) immunoblots of the lysates of these cells showed that although YopH reduced both basal and induced tyrosine phosphorylation, compound 4 restored basal phosphorylation to normal levels and the receptor response to near-normal (Fig. 5A, upper panel). Similarly, the phosphorylation of the Lck kinase at its positive regulatory tyrosine residue, Tyr-394, was severely reduced by YopH, as we have reported before (39). Compound 4 also restored Lck phosphorylation to normal levels (Fig. 5A, lower panel). Compounds 3 (Fig. 5B), 4, and 5 (not shown) also largely neutralized the effect of YopH on tyrosine phosphorylation in a dose-dependent manner.
To extend these observation to normal human T lymphocytes, we incubated freshly isolated peripheral blood T cells with 2 M of membrane-permeable YopH (ANT-YopH) for 2 h at 37°C, followed by stimulation with anti-CD3⑀ and anti-CD28 mAbs plus a secondary cross-linking antibody for 18 h at 37°C in the presence of YopH inhibitors. Although YopH completely inhibited the secretion of interleukin-2, as before (39), cells stimulated in the presence of the YopH inhibitors produced almost as much interleukin-2 as cells without ANT-YopH (Fig.  5C). The four inhibitors varied somewhat in efficacy, but none of them affected the response of cells without ANT-YopH. Together, all these experiments demonstrate that our YopH inhibitors can reverse the strong inhibitory effects of YopH on T cell activation. DISCUSSION In this study we use an approach to inhibitor design, which can be characterized as a hybrid between traditional high throughput screening and rational design based on the structure of the substrate. Instead of starting with a nonhydrolyzable phosphotyrosine analog, we used high throughput library screening to identify useful lead structures, which then were taken into in silico docking studies as the main platform on which the inhibitory properties of inhibitors were examined at the atomic level. This approach identified a novel pharmacophore, furanyl salicylate, a substrate mimetic with better properties for drug design than the highly charged phosphotyrosine. In silico docking gave detailed insight into the complex network of hydrogen bonds between the enzyme and the inhibitors and thus allowed us to understand the experimental results with analogs of the first hits. This in turn will make it possible to rationally design even better inhibitors in the future.
Structurally, our inhibitors resemble a recently reported YopH inhibitor, aurintricarboxylic acid (51), that represents a symmetric joining of three salicylate moieties. This compound inhibited YopH with a K i of 5 nM and an IC 50 of 10 nM, but did not penetrate into cells. Our inhibitors contain only one salicylate group and therefore have only one carboxylic group. For this reason, they penetrated into lymphocytes and were able to The docking studies showed that the salicylate moiety ("ring A") mimics the phosphotyrosine residue of a substrate for YopH and fits into the catalytic pocket (P1). The furanyl ring ("ring B") interacts specifically with Gln-357 on the rim of the catalytic pocket and with the side chain of Arg-404. Because both these residues are unique to YopH compared with other PTPs, the furanyl ring apparently provides selectivity to the inhibitors. The positioning and the interactions involving the other end of the compounds ("ring(s) C") were more variable, as also seen experimentally by a higher tolerance for substitutions at this position. However, ring C may give us the opportunity to bridge the core salicylate-furanyl structure occupying the catalytic pocket to two other unique pockets present on the surface of YopH (Figs. 3 and 4). Thus, future compounds that reach into pockets may have further increased affinity and selectivity for YopH.
It should be pointed out that our best inhibitors showed a mixed type of inhibition (which is common), indicating that they not only fit into the catalytic cleft in a substrate-competitive manner, as demonstrated by the in silico docking, but also affect the substrate-enzyme complex. Because our inhibitors interact with a considerably larger surface than the pNPP substrate, it is possible that they bind to YopH with pNPP in its catalytic cleft and interfere with product release.
Our studies support the notion that small molecule inhibitors that are specific for individual members of the PTP family can be generated by taking advantage of unique surface features outside of the catalytic pocket. A comprehensive comparison of the crystal structures and computer models of 103 of the 107 human PTPs (data not shown) clearly shows that PTPs vary widely in topology and charge distribution surrounding the catalytic pocket. This variability probably reflects substrate preferences and the interactions with surface features of substrates other than the phosphotyrosine residue (36). As an example, Fig. 4 shows a close up of the surface that surrounds the catalytic pocket in YopH, PTP1B, and VHR. In YopH there is a large semicircular valley bordered by the three pockets and with a low ridge between P2 and P3 giving the valley a Vshaped floor. In PTP1B there are also two additional depressions, of which the equivalent of P3 is a pocket involved in substrate binding (36). In contrast to YopH, the access to P3 from P1 elongated and constricted into a very narrow passage, and access to the equivalent of P2 is completely blocked in PTP1B. In VHR, there are no well defined depressions other than the catalytic pocket, which is surrounded by more hydrophilic and acidic surfaces than in YopH or PTP1B. These striking differences in surface topology inspire confidence in a more rational design of selective inhibitors for PTPs.
Our results demonstrate that selective and potent YopH inhibitors can be developed and could be used as a starting point for the design of drugs to combat the virulence of Y. pestis. Particularly in the case of multidrug-resistant strains or following exposure to aerosolized Y. pestis, such an inhibitor may prove very useful. YopH is also a good target for drug design because it differs in several residues from endogenous PTPs in humans. In addition, an inhibitor with some effects on endogenous PTPs in immune cells may act to further strengthen the immune response against Y. pestis. Acknowledgment-We are grateful to Dr. Zhong-Yin Zhang for discussions and advice.

FIG. 5. Reversal of YopH-mediated inhibition of T cell signaling by YopH inhibitors.
A, anti-Tyr(P) (upper panel) and anti-Lck-pY394 (lower panel) immunoblots of lysates of Jurkat T cells transfected with empty vector or YopH and treated with or without 48 M compound 1, and then stimulated with anti-CD3⑀ and anti-CD28 plus a secondary cross-linking antibody or the cross-linking antibody alone, as indicated. B, similar experiment using the indicated concentrations of compound 3. C, enzyme-linked immunoabsorption assay for interleukin-2 in the culture medium from 2 ϫ 10 5 cells treated with anti-CD3⑀ and anti-CD28 plus a secondary cross-linking antibody or the crosslinking antibody alone in the absence or presence of YopH inhibitors, as indicated. The results are given as pg/ml and represent the mean Ϯ S.E. from triplicate determinations.