Identification of a bacterial inhibitor of protein kinases. Mechanism and role in host cell invasion.

We show that Escherichia coli produce a factor that inhibits the activity of tyrosine and serine/threonine protein kinases. The factor is a protein found in the periplasmic compartment and is also secreted into the culture medium. Using a particle concentration fluorescence immunoassay specific for tyrosine kinase activity and inhibition of the tyrosine kinase p56(lck), we purified this factor to apparent homogeneity. Analysis of trypsin-digested fragments by mass spectrometry identified the inhibitor as the bacterial periplasmic protein UDP-sugar hydrolase, an enzyme with potent and nonspecific 5'-nucleotidase activity. Overexpression of the enzyme in bacteria leads to coordinate increases in both 5'-nucleotidase and p56(lck) inhibitory activity, confirming the identity of the inhibitor. The kinase inhibitory activity appears to be due to the formation of adenosine, which we show is inhibitory for p56(lck), cAMP-dependent protein kinase, and casein kinase. Overexpression of UDP-sugar hydrolase leads to an increase in the recovery of enteropathogenic E. coli following infection of HeLa cell monolayers and corresponding alterations in tyrosine-phosphorylated host proteins. These results suggest that UDP-sugar hydrolase may be an important factor affecting host cell function following intracellular bacterial infection.

A common element in the life cycle of many bacterial pathogens is the establishment of residence inside host cells (1,2). Ligands and receptors mediating bacterial invasion have been identified, and, in some cases, aspects of host signal transduction that are required for the invasion process have been characterized (3)(4)(5)(6)(7)(8)(9)(10)(11)(12). Less well known, however, are the factors influencing bacterial survival once inside the host cell. Some studies have documented gene expression differences for bacteria that reside inside host cells (13), and auxotrophic mutants of Salmonella have been found to be attenuated for pathogenicity (14 -16). Using insertional mutagenesis in conjunction with large scale screening, several groups have identified mutants defective for invasion or intracellular replication (14,(17)(18)(19). It is not clear, however, if any of the mutants act by affecting host signaling.
In the course of employing Escherichia coli as a host for recombinant protein expression, we discovered that bacterial lysates contain a factor that inhibits the activity of the tyrosine kinase p56 lck . The key role that tyrosine kinases play in host cell signaling, coupled with the possibility that this unexpected bacterial activity might be used to interfere with these functions, prompted us to further investigate this factor. Here we describe the initial characterization, purification, and identification of this factor and show that Ser/Thr kinases are inhibited as well. We also demonstrate the involvement of this inhibitory activity in bacterial invasion of human cells.
Kinases-Two sources of p56 lck were used in this study. Baculovirusexpressed p56 lck was partially purified on a DEAE-column as described (23). Partially purified p56 lck from LSTRA cells was prepared as follows. 1 liter of cells was collected by centrifugation, washed with PBS, 1 and resuspended in 15 ml of sonication buffer (20 mM MOPS, pH 7.2, 75 mM ␤-glycerophosphate, 2 mM EDTA, 1 mM sodium orthovanadate, 1 mM NaF, 1 mM PMSF, 1 mM pepstatin). The cells were sonicated 4 times with 5-s pulses on ice, and nuclei and cell membranes were pelleted at 50,000 rpm for 20 min at 4°C in a Beckman 70.1Ti ultracentrifuge rotor. The pellet was resuspended in sonication buffer containing 1% Triton X-100 and recentrifuged at 70,000 rpm for 15 min at 4°C in the same rotor. The supernatant containing partially purified p56 lck was aliquoted and stored at Ϫ80°C until use. cAMP-dependent protein kinase was purchased from Sigma. Casein kinase was purchased from UBI.
Kinase Assays-Tyrosine kinase activity and its inhibition was routinely assayed using a particle concentration fluorescence immunoassay (PCFIA) as described by Babcook et al. (24). Bovine brain myelin basic protein (Sigma) was covalently coupled to Fluoricon carboxylpolystyrene beads with EDC and used as a substrate. The beads were suspended in 20 mM Tris (pH 7.7), 0.5% BSA, 0.01% Brij-35 to a final concentration of 0.125% with an estimated MBP concentration of 25 g/ml. Assay reactions were in 96-well Pandex assay plates in a total volume of 50 l and included 20 l of beads, 0.5 mM ATP, 10 mM MnCl 2 , 10 mM dithiothreitol, baculovirus-expressed p56 lck at 1:1,000 dilution, or LSTRA cell membrane preparations containing p56 lck at 1:100. Typically, 5 l of 2-fold serial dilutions of inhibitor fractions was added to the reaction mixture. Reactions were initiated by the addition of p56 lck and allowed to proceed for 10 or 15 min at 37°C. Soluble components of the reaction were removed by suction through the porous support in the wells of the plate, the beads were washed and incubated with fluoresceinated antiphosphotyrosine antibody 3A12 at 1 g/ml. After 10 min, unbound antibody was removed, the beads were washed, and fluorescence was measured. p56 lck activity is expressed as relative fluorescence units. Inhibitory units were estimated by identifying the highest dilution of inhibitor still retaining p56 lck inhibitory activity.
Tyrosine and Ser/Thr kinase activity was also assayed using incorporation of ␥-[ 33 P]ATP and immobilization on p81-phosphocellulose paper. Briefly, assays were performed in 50-l volumes in assay buffer (20 mM Tris-HCl, pH 7.5, 10 mM MnCl 2 or MgCl 2 , 1 mM dithiothreitol, 0.5% BSA, plus 10 Ci of ␥-[ 33 P]ATP (50 M ATP final concentration)) with the appropriate substrate (25 g/ml MBP or 250 g/ml partially dephosphorylated bovine casein (Sigma)). Reactions were initiated with the addition of ATP, incubated for 15 min at 30°C, and then terminated by the addition of 10 mM EDTA. The reaction mixture was then dotted onto p81 phosphocellulose paper immobilized in a 96-well manifold. The filter paper was washed several times with 75 mM phosphoric acid, dried, and placed in a Packard "flexifilter" plate. Scintillation fluid was added to the wells and the plate was counted in a TopCount scintillation counter (Packard).
In Vitro Kinase Assay-The ability of periplasmic fractions to inhibit p56 lck was also tested using an in vitro kinase assay. Baculovirusexpressed p56 lck was incubated in 20 mM Tris (pH 7.4) buffer with 0.5% BSA, 10 mM MnCl 2 , 10 mM DTT, 0.5 mM unlabeled ATP and 10 Ci of ␥-labeled ATP (Amersham) at 37°C. Samples also contained periplasmic extract (1:4 dilution) and/or MBP (25 g/ml) and/or sodium molybdate (0.8 mM). Reactions were terminated by adding SDS-PAGE sample buffer and boiling for 5 min. Samples were then run on an SDS-PAGE gel, stained with Coomassie Blue, dried onto a 3MM filter paper, and exposed to x-ray film to visualize bands with incorporated phosphate.
Periplasmic and Cytoplasmic Extractions-Bacterial cells were grown to stationary phase in LB medium at 28°C overnight with shaking at 225 rpm. Periplasmic fractions were prepared by resuspending the bacteria in TES (30 mM Tris, pH 8.0, 10 mM EDTA, 30% w/v sucrose) with lysozyme (Boehringer) to a final concentration of 1 mg/ml. The mixture was left on ice for 40 min with occasional mixing, then centrifuged at 11,000 ϫ g for 30 min at 4°C. The supernatant (periplasmic fraction) was transferred to a fresh tube and PMSF to a final concentration of 1 mM was added. Periplasmic fractions were stored at Ϫ80°C for several weeks with little loss of inhibitory activity. The cytoplasmic fraction was prepared by resuspending the remaining bacterial pellet in 15 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, 0.1% Nonidet P-40, 1 mM PMSF, sonicating to lyse the cells and shear DNA, and centrifuging at 12,000 ϫ g to remove insoluble material.
Ammonium Sulfate Precipitation-The protein concentration of the periplasmic extract was determined by Pierce protein assay and standardized to a concentration of 10 mg/ml. Solid ammonium sulfate was added slowly, with stirring, at 4°C. After each addition, the solution was centrifuged at 10,000 ϫ g for 10 min at 4°C to pellet the proteins that had been salted out of solution. A small sample of each of the pellets was resuspended in 10 mM Tris (pH 8.0), 1 mM EDTA and passed through a Sephadex G-25 column (Pharmacia). Each sample was then assayed for kinase inhibitory activity by PCFIA. The pellet containing the inhibitory activity was then resuspended in 10 mM Tris, pH 8.0, 1 mM EDTA, 0.01% Brij-35 and dialyzed into the same buffer, with frequent changes of the dialysis buffer, overnight at 4°C.
Q-Sepharose-All column separations were performed using a Pharmacia fast protein liquid chromatography (FPLC) system at 4°C. A Q-Sepharose column (20 ml of matrix) was equilibrated with starting buffer (10 mM Tris, pH 8, 1 mM EDTA, 0.01% Brij 35), the sample was loaded, the column was washed with 10 column volumes of starting buffer, and then bound proteins eluted with an ascending NaCl gradient from 0 to 0.75 M, also over 10 column volumes. Flow rate was typically 2 ml/min. Superose 12-Q-Sepharose-purified inhibitor was concentrated 10fold, and a 100-l sample was analyzed on a Superose-12 gel filtration column (Pharmacia). The column was pre-equilibrated in column buffer (10 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA, 0.01% Brij 35) with 5 column volumes and run at 0.2 ml/min. 0.2-ml fractions were collected and analyzed for p56 lck inhibitory activity. A similar volume containing protein standards of known molecular weight was run under the same conditions.
Hydroxyapatite-Bio-Gel HTP hydroxyapatite (Bio-Rad) was equilibrated with HA starting buffer (10 mM Tris, pH 8, 1 mM EDTA, 2 M NaCl, 0.01% Brij 35). The samples were loaded, the column was washed with 10 column volumes of HA starting buffer, and bound proteins were eluted with a gradient of 0 to 0.3 M NaF in starting buffer over 10 column volumes using the FPLC. The NaF gradient permanently altered the separation properties of the hydroxyapatite requiring the use of fresh matrix for each separation.
Mono Q-Partially purified inhibitor samples were further separated on Mono Q (Pharmacia). The column was equilibrated in Q starting buffer (10 mM Tris, pH 8, 1 mM EDTA, 0.01% Brij 35), the sample was loaded, and bound proteins eluted with an ascending NaCl gradient to 1.5 M NaCl.
Mass Spectrometry-Purified protein fractions were run on SDS-PAGE, transferred to Immobilon, digested with trypsin, and analyzed by mass spectrometry as described by Hess et al. (25).
Assay for UDP-sugar Hydrolase-The enzymatic activity of UDPsugar hydrolase was determined by measuring the hydrolysis of bis(pnitrophenyl) phosphate. Fractions containing UDP-sugar hydrolase were incubated in 0.1 M Tris-HCl, pH 6.7, 5 mM MnCl 2 , 1 mg/ml bis(p-nitrophenyl) phosphate (Sigma), 1 mg/ml BSA in a final volume of 100 l in 96-well plates. After 20 min at 37°C, the reaction was terminated with the addition of NaOH to a final concentration of 0.05 N and the release of p-nitrophenyl was measured by absorbance at 405 nm.
TLC Analysis of UDP-sugar Hydrolase Products-Partially purified inhibitor fractions were incubated under standard kinase assay conditions with [ 14 C]ATP for 15 min at 37°C. The reaction products were separated by thin layer chromatography on DEAE-cellulose plates with water/isobutyl alcohol/methanol/ammonium hydroxide in a ratio of 30: 10:1:10 (v/v) as the solvent (26). Location of the standards was determined by uv, and reaction products were visualized by autoradiography.
Invasion Assay-EPEC cells were grown overnight at 37°C with shaking in LB with 100 g/ml ampicillin. The overnight culture was diluted 1:20, and growth was continued for 2 h in LB ϩ 1 mM MnCl 2 . At this point, typical A 600 were 0.2 to 0.3. The bacteria were collected, washed twice in PBS, and their concentration was adjusted to provide the appropriate multiplicity of infection. Aliquots of these cultures were plated to enumerate starting bacterial density. HeLa cell monolayers in 6-well plates were washed 3 times with RPMI ϩ 10% fetal bovine serum containing 1% mannose, 1 mM MnCl 2 without antibiotic. The mannose was included to prevent nonspecific bacterial adherence to the monolayer. 0.2 ml of bacteria in 2 ml of medium was then added to the wells, and the plates were incubated at 37°C for 1.5 h. The monolayers were washed 3 times with PBS and then 2 ml of medium containing 100 g/ml gentamycin was added to kill all extracellular bacteria. Following a further 3-h incubation, the monolayers were again washed 3 times with PBS and lysed with 0.4 ml 1% Triton X-100 in PBS for 5 min. 1.6 ml of LB was added and aliquots were plated to measure recovered bacteria. Control experiments (not shown) demonstrated that the concentration of gentamycin used was sufficient to kill extracellular bacteria and that the Triton X-100 treatment did not affect bacterial plating efficiency. Control DH5␣ bacteria were found to be noninvasive in this assay.

RESULTS
Bacterial Lysates Contain an Activity That Inhibits p56 lck -In the course of using bacteria as hosts for recombinant protein expression, we found that bacterial lysates contain a substance capable of inhibiting the protein-tyrosine kinase p56 lck . Bacterial lysates were prepared from standard overnight cultures of N4830-1, DH5␣, or BL21 cells grown in LB medium at 28°C and lysed with Nonidet P-40. These lysates were then added to p56 lck PCFIA kinase assays. As shown in Fig. 1, untreated bacterial lysates from N4830 -1 cells completely inhibited the ability of p56 lck to phosphorylate immobilized MBP. In order to further define this factor, the inhibitory activity from periplasmic and cytoplasmic fractions were compared. Although both fractions contained inhibitory activity, most of the cytoplasmic inhibitory activity could be neutralized by adding protease or phosphatase inhibitors. In contrast, the inhibitory activity present in the periplasmic fraction could not be neutralized with these inhibitors.
We also found that the periplasmic inhibitory activity was moderately resistant to heat treatment but could be inactivated by incubating extracts for 1 h at 70°C.
Phosphatases are present in the periplasm of bacteria grown under certain conditions such as low phosphate. As well, a Yersinia virulence factor has been shown to be a tyrosinespecific phosphatase (27,28). We therefore tested to see if the inhibitor had tyrosine phosphatase activity using two methods. In the first method, MBP immobilized on beads was phosphorylated by p56 lck , the beads were washed and then incubated with periplasmic lysates. The phosphotyrosine remaining on the beads was then measured by PCFIA (24). No change in phosphotyrosine content was seen (not shown). In the second method, purified p56 lck was incubated with ␥-labeled [ 32 P]ATP and MBP with or without periplasmic extracts. The reaction products were then separated by SDS-PAGE, dried, and visualized by autoradiography. As shown in Fig. 2, when the periplasmic extract is included throughout the course of the reaction, no phosphate is incorporated into either p56 lck or MBP (lanes 6 -8). When the periplasmic extract is added after the reaction has proceeded and phosphorylation has already occurred, incorporation ceases but does not decrease (lanes 2  and 4). These results suggest that the inhibitor is not a phosphatase, but rather interferes in some manner with the kinase activity of p56 lck .
Size exclusion chromatography was performed on periplasmic extracts, and fractions were assayed for inhibitory activity. As shown in Fig. 3, the inhibitory activity elutes in a single peak with a molecular weight of approximately 60,000.
Purification and Properties-The inhibtory activity proved to be stable for at least 1 week at 4°C and for months at Ϫ80°C; however, only limited freezing and thawing could be tolerated. Therefore, throughout the purification process, freeze-thaw cycles were limited to one or at most two.
Although the purification of the inhibitor was done with periplasmic extracts from N4830-1 cells, we subsequently found that the inhibitory activity was secreted as well and could be purified from culture supernatants. A number of biochemical separations were used to purify the periplasmic inhibitor. A summary of the separations is shown in Fig. 4 and Table I. The effective steps were found to be ammonium sulfate precipitation, ion exchange chromatography on Q-Sepharose, elution from hydroxyapatite with NaF, isoelectric focusing, and further ion exchange separation on Mono Q. The combination of these steps results in an approximately 21,600-fold purification. Note that total recoverable activity increases following the first ion exchange step. Fig. 5 shows a silver-stained SDS-PAGE gel of inhibitory fractions 22, 23, and 24 from the Mono Q column of Fig. 4. A prominent band with a molecular weight of approximately 60,000 is observed. Slices of similar SDS-PAGE gels were incubated in buffer to elute proteins, acetone-precipitated to remove SDS from protein, and the precipitate was assayed for p56 lck inhibitory activity. As shown in Fig. 6, the gel slice corresponding to the M r ϭ 60,000 protein inhibits p56 lck while eluates from the other slices do not.
Identification of the Inhibitor-These results strongly suggested that the M r ϭ 60,000 protein was responsible for the FIG. 1. E. coli lysates inhibit p56 lck activity. Bacterial lysates were prepared from an overnight culture of N4830-1 cells grown at 28°C in LB. Cells were lysed in 1/50 of their culture volume with Nonidet P-40 lysis buffer. 5 l of a 1:4 dilution of this lysate was evaluated for p56 lck inhibitory activity by PCFIA. Cytoplasmic and periplasmic fractions were prepared from similar cultures and also assayed for inhibition. In both cases, parallel assays were performed in the presence of inhibitors of phosphatases and proteases (1 mM sodium orthovanadate, 0.8 mM sodium molybdate, 1 mM PMSF, 1.6 mg/ml leupeptin, 10 g/ml pepstatin). bckgnd refers to the background signal in the absence of added p56 lck . All values are reported as a percent of maximal p56 lck activity. The data shown are from one representative experiment. Error bars represent standard errors from triplicate measurements.

FIG. 2. Periplasmic inhibition of p56 lck autophosphorylation.
Aliquots of Sf9-expressed and DEAE-purified p56 lck were incubated with 1:4 dilutions of periplasmic extracts, 25 g/ml MBP, and 0.8 mM sodium molybdate for the times indicated above each lane. Reactions were terminated with the addition of SDS sample buffer, boiled, and separated by electrophoresis. The gel was dried and exposed to x-ray film for 24 h. Lane 1, p56 lck only was incubated for 3 min. Lane 2, p56 lck was incubated for 3 min, then periplasmic extract was added to the reaction for an additional 10 min. Lane 3, p56 lck only was incubated for 13 min. Lane 4, as for lane 2 except that molybdate was included in the reaction. Lane 5, as for lane 3 except that MBP was included in the reaction. Lanes 6 and 8; as for lane 5 except that periplasmic extract was included in the reaction. Lane 7, as for lane 6 except molybdate was included in the reaction. The 47-kDa band is an unidentified periplasmic protein that incorporates phosphate.

FIG. 3. Superose 12 gel filtration analysis of periplasmic inhib-
itor. Q-Sepharose-purified periplasmic inhibitor was analyzed on a Superose 12 gel filtration column as described under "Experimental Procedures." p56 lck inhibitory activity. Purified fractions were run on SDS-PAGE and transferred to Immobilon membrane. The region of membrane corresponding to the inhibitor was cut out, digested with trypsin, and the fragments were analyzed by mass spectrometry. 19 major peaks were identified and their molecular weights were compared to a data base of predicted tryptic fragments from bacterial proteins. 13 of the 19 peaks matched the predicted profile for the bacterial periplasmic enzyme UDPsugar hydrolase. This enzyme was first characterized by Glaser et al. (29) and Neu (30), and its gene (UshA) was cloned by Burns and Beacham (21). Its properties match very closely with the properties of the inhibitor including resistance to moderate heat treatment, subcellular localization (periplasm), molecular weight (60,800), requirement for divalent cations (particularly Co 2ϩ or Mn 2ϩ , not shown), hydrophilicity, and behavior on ion exchange and other matrices.
In order to confirm that UDP-sugar hydrolase was responsible for the p56 lck inhibitory activity, we introduced plasmid pLA7 that contains the full gene for the enzyme into bacteria and measured enzyme and kinase inhibitory activity. Introduction of this plasmid into E. coli results in the overexpression of its enzymatic activity by 10-fold. The corresponding kinase inhibitory activity in crude lysates or partially purified fractions increases by 50-fold relative to wild-type (Fig. 7). These results are entirely consistent with the conclusion that the kinase inhibitor is in fact UDP-sugar hydrolase.
UDP-sugar hydrolase would appear to be an unusual candidate for a kinase inhibitor. However, this enzyme also possesses a potent and nonspecific 5Ј-nucleotidase activity. It is thus able to hydrolyze ATP to ADP, AMP, and adenosine. One possibility is that UDP-sugar hydrolase simply depletes ATP in the kinase reaction. We measured ATP hydrolysis by partially purified inhibitor extracts under the conditions of our kinase assay (i.e. 0.3-0.5 mM ATP) and although we readily detected hydrolysis, the depletion of ATP (approximately 50%) was insufficient to account for the loss of kinase activity (Fig. 8). In fact, under the conditions of our assay, full kinase activity is still observed at 10 M ATP. This experiment did show, however, that the primary enzymatic product was adenosine. We therefore tested for the ability of adenosine to inhibit the activity of p56 lck . As shown in Fig. 9a, adenosine is a strong inhibitor of p56 lck activity with an estimated IC 50 of 0.3 mM. Therefore, these results suggest that the production of adenosine is the mechanism by which UDP-sugar hydrolase inhibits p56 lck .
Our observation that p56 lck could be inhibited by adenosine prompted us to look at inhibition by adenosine of other kinases. As shown in Fig. 9, b and c, adenosine also inhibits the activity of casein kinase and cAMP-dependent protein kinase. The estimated IC 50 values for these two enzymes are 18 M for casein kinase and 0.48 mM for cAMP-dependent protein kinase.
UDP-hydrolase Activity Affects Invasiveness of EPEC-The ability to produce adenosine resulting in the inhibition of various protein kinases suggested that UDP-sugar hydrolase may also play a role in bacterial-host interaction. To test this possibility, we examined the effect of overexpressing UDPsugar hydrolase in an in vitro model of bacterial intracellular invasion.
Enteropathogenic E. coli (EPEC) strains have the ability to adhere to and invade human cells; however, intracellular survival is limited and recovery of intracellular bacteria decreases over time (not shown). Monolayers of HeLa cells were infected with log phase EPEC cells containing pLA7 or the parent vector pBR322.
If the incubation period is limited to 1 h, no differences in recovery of bacteria are seen between EPEC cells containing pBR322 or pLA7 (not shown). This observation suggests that bacterial growth, adherence, and invasion are unaffected by  a Total inhibitory activity increased following this purification step.
FIG. 5. SDS-PAGE analysis of highly purified p56 lck inhibitor. Mono Q-purified inhibitor fractions 22, 23, and 24, which contain inhibitory activity, were separated by SDS-PAGE and silver-stained.
FIG. 6. The periplasmic inhibitor can be eluted from SDS-PAGE gels. 4-mm slices from an SDS-PAGE gel of inhibitor purifed by ammonium sulfate, Q-Sepharose, and hydroxyapatite were soaked overnight at room temperature in 10 mM Tris (pH 8), 0.1% SDS to passively elute the protein. BSA as carrier was added and total protein was precipitated by the addition of 5 volumes of ice-cold acetone. After 2 h at Ϫ20°C, precipitated proteins were pelleted by centrifugation at 12,000 ϫ g for 10 min. Pellets were resuspended in 10 mM Tris, pH 8, and then assayed for kinase inhibitory activity. The inhibitory gel slices correspond to molecular weights between 55,000 and 65,000. the presence of UDP-sugar hydrolase. However, if the incubation period is extended to 3 h, overexpression of UDP-sugar hydrolase results in an approximate 10-fold increase in the number of bacteria recovered over a wide range of multiplicities of infection (10 to 500) (Fig. 10). These results suggest that bacteria overexpressing UDP-sugar hydrolase have an increased ability to survive once inside the HeLa cells.
Since our in vitro results suggested to us that invasive bacteria could affect signaling events once inside the cell, we performed an invasion assay and examined infected cells for alterations in phosphotyrosine content. As shown in Fig. 11, uninfected HeLa cells display a prominent phosphotyrosinecontaining protein of approximately 120 kDa. 1 and 3 h after infection, the phosphotyrosine content of this band decreases in cells infected with both strains, but the decrease is greater in cells infected with EPEC overexpressing UDP-sugar hydrolase. As described earlier by Rosenshine et al. (31), infection with EPEC results in the induction of additional phosphotyrosinecontaing bands of host cell origin (approximately 90 and 140 kDa). We also observe a marked reduction in phosphotyrosine content of these bands 3 h after infection with EPEC overexpressing UDP-sugar hydrolase activity. These results suggest that host cell signaling is being disrupted by invading EPEC and that UDP-sugar hydrolase overexpression enhances this effect. DISCUSSION p56 lck , a member of the Src family of protein-tyrosine kinases, is a T cell-specific kinase with both autophosphorylation activity and the ability to phosphorylate exogenous substrates such as myelin basic protein. We found that coincubation of bacterial lysates with this enzyme strongly inhibits its activity. We linked the inhibitory activity to a M r ϭ 60,000 periplasmic protein, purified it to homogeneity, and identified it as the bacterial enzyme UDP-sugar hydrolase. The inhibitor shares many biochemical properties with UDP-sugar hydrolase including its molecular weight, behavior on ion exchange and other matrices, relative resistance to heat treatment, its hydrophilicity, and its dependence on divalent cations such as Mn 2ϩ (but not Ca 2ϩ or Mg 2ϩ ). Overexpression of UDP-sugar hydrolase in bacteria results in a coordinate increase in 5Ј-nucleotidase activity as well as p56 lck inhibitory activity. Taken together, these results strongly support the identity of the kinase inhibitor as UDP-sugar hydrolase.
When first identified and characterized, the role assigned to FIG. 7. Plasmid pLA7 results in overexpression of UDP-sugar hydrolase and kinase inhibitory activity. Ammonium sulfate-precipitated fractions prepared from bacteria transformed with either pBR322 or pLA7 were assayed for UDP-sugar hydrolase activity and p56 lck inhibitory activity as described under "Experimental Procedures." Both preparations were adjusted to have equivalent protein content. UDP-sugar hydrolase was as a component of the nucleotide scavenging pathway. Its ability to hydrolyze nucleotides enables the products of the reaction to enter into the bacterial cell where they can be used as nucleotide precursors, or as a carbon source. In fact, cells possessing this enzyme can grow using 5Ј-AMP as the sole carbon source. The role of its UDP-sugar hydrolase activity is less clear. UDP-sugars are involved in glycogen synthesis, acting as high energy intermediates. It is possible that this enzyme is also involved in scavenging these molecules.
UDP-sugar hydrolase is a potent and relatively nonspecific 5Ј-nucleotidase. Although it does deplete ATP present in the kinase reaction, simple loss of ATP does not seem to be the basis of its inhibitory activity. Rather, we have found that adenosine, the reaction product, is inhibitory for the protein kinases. We have also investigated the potential inhibitory activity of AMP and ADP on the kinases and find that only p56 lck is also significantly inhibited by these molecules (IC 50 of 50 and 25 M respectively 2 ). However, since adenosine is the only hydrolysis product detected under the conditions of our assay, we conclude that the production of adenosine is the mechanism by which UDP-sugar hydrolase inhibits the kinases. This result is intriguing, considering the fact that deficiencies in adenosine deaminase, the major enzyme responsible for "detoxifying" adenosine, results in a severe combined immunodeficiency syndrome, the most notable feature being severe T cell deficiency (32).
Adenosine, in addition to its metabolic role as a nucleotide precursor, also has other functions. In vivo, its main physiological function seems to be as a regulator of cardiac rhythm; however, it also has documented anti-inflammatory effects. These include inhibition of neutrophil adhesion (33), inhibition of platelet aggregation (34), and inhibition of superoxide burst by neutrophils (35,36). Cronstein et al. (37) have shown that the anti-inflammatory effects of methotrexate are most likely mediated through a buildup of adenosine. In vitro, adenosine is a known inhibitor of phosphoinositol kinases (38), a key second messenger generating enzyme. We have shown that UDPsugar hydrolase can inhibit a tyrosine kinase and two Ser/Thr kinases, and Kim and Matthews 3 have shown that yeast histi-dine kinase is also inhibited. Thus, although we have characterized this enzyme primarily as an inhibitor of p56 lck , it appears that its effects are more widespread. Therefore, overproduction of adenosine by an infectious agent may result in inhibition of a wide variety of kinases and could therefore be of general utility in compromising host function.
At the present time, we have not identified the mechanism of inhibition of protein kinases by adenosine. Preliminary data suggest that at least for p56 lck , adenosine does not act as a competitive inhibitor for ATP. 2 As well, the fact that the apparent IC 50 values for the kinases varies for each enzyme suggests that there may be considerable specificity. Further experiments are required to address the mechanism of inhibition and the identification of other potential targets.
The multiple kinase inhibitory activities of UDP-sugar hydrolase products coupled with its role in the metabolism of nucleotides and UDP-sugars suggested to us that this enzyme might influence bacterial intracelluar infection. To test this, we investigated the effect of overexpressing UDP-sugar hydrolase in the EPEC model of in vitro bacterial invasion. Once EPEC invade the monolayer, extensive growth does not normally occur. This assay is therefore a useful one for investigating potential factors that can affect intracellular survival. We found that overexpression significantly increases the recovery of enteropathogenic E. coli following invasion over a wide range of multiplicities of infection, but only following a 3-h incubation period. Since no differences were seen after short incubation times, this result suggests that overexpression of UDP-sugar hydrolase increases bacterial survival once inside the HeLa cells. At present, we do not know the precise biochemical mechanism mediating this effect. However, the multiple inhibitory 2 S. A. Berger, unpublished data. 3 10. Overexpression of UDP-sugar hydrolase increases recovery of EPEC following HeLa cell infection. EPEC cells containing either pBR322 or pLA7, grown to log phase, were added to HeLa cell monolayers (approximately 10 6 HeLa cells). After 90 min, the monolayers were washed and gentamycin was added to kill extracellular bacteria. 3 h following the addition of antibiotic, the monolayers were washed and lysed to recover the intracellular bacteria. Each point is an average of 3 determinations. The error bars represent the standard error.
FIG. 11. EPEC infection of HeLa cells alters phosphotyrosine content of host proteins. EPEC cells containing either pBR322 or pLA7 were added to HeLa cell monolayers (approximately 5 ϫ 10 5 HeLa cells; multiplicity of infection of 1,000). After 90 min, the monolayers were washed and gentamycin was added to kill extracellular bacteria. 1 or 3 h following the addition of antibiotic, the monolayers were washed, lysed, and analyzed by Western blotting for phosphotyrosine-containing proteins. Ctl, uninfected HeLa cells.
activities of adenosine, coupled with its involvement in nucleotide scavenging, suggests that this enzyme may have an essential metabolic role, or may act as a general host "anergizing" factor, possibly affecting host cell signaling. In agreement with this possibility, we found that the phosphotyrosine content of the major phosphorylated proteins in HeLa cells decreased following EPEC infection. Furthermore, this decrease was even greater in HeLa cells infected with EPEC cells overexpressing UDP-sugar hydrolase.
UDP-sugar hydrolase is widely distributed and highly conserved throughout evolution, and it is possible that other pathogens may also employ this enzyme during infectious situations. For instance, Smail et al. (39) have shown that supernatants from germinated Candida albicans cultures contain adenosine which is responsible for inhibiting neutrophil function. Therefore, the induction of host cell toxicity or anergy via adenosine production may represent a novel and general mechanism of pathogenicity.