|Home | About | Journals | Submit | Contact Us | Français|
The phosphatidylinositol-specific phospholipase C from Listeria monocytogenes forms aggregates with anionic lipids leading to low activity. The specific activity of the enzyme can be enhanced by dilution of the protein, addition of both zwitterionic / neutral amphiphiles (e.g., diheptanoylphosphatidylcholine or Triton X-100) or 0.1–0.2 M inorganic salts. Activation by amphiphiles occurs with both micellar (phosphatidylinositol dispersed in detergents) and monomeric (dibutroylphosphatidylinositol, diC4PI) phosphotransferase substrates and inositol 1,2-(cyclic)-phosphate (cIP), the phosphodiesterase substrate. The presence of zwitterionic / neutral amphiphiles (to which the protein binds weakly) dilutes the surface concentration of the interfacial anionic substrate and thereby reduces enzyme/phospholipid particle aggregation. Zwitterionic amphiphiles also can bind directly to the protein and enhance catalysis since they enhance both diC4PI and cIP hydrolysis. In contrast to activation by amphiphiles, the rate enhancement by salt only occurs for the phosphotransferase step of the reaction. Added salt has a synergistic effect with zwitterionic phospholipids, leading to high specific activities for PI cleavage with only moderate dilution of the anionic substrate in the interface. This kinetic activation correlates with weakening of strong PI-PLC hydrophobic interactions with the interface as monitored by a decrease in the maximum monolayer surface pressure for insertion of the protein. Several point mutations of surface hydrophobic residues (W49A, L151A, L235A, and F237W) can dramatically alter the unusual kinetics of this secreted enzyme. The high affinity of PI-PLC for anionic phospholipids along with a strong hydrophobic interaction, which gives rise to the unusual kinetic behavior, is considered in terms of how it might contribute to the role of this phospholipase in L. monocytogenes infectivity.
Bacterial phosphatidylinositol-specific phospholipase C (PI-PLC) catalyzes the calcium-independent hydrolysis of PI in two steps: (i) an intramolecular phosphotransferase reaction at a phospholipid aggregate surface to produce diacylglycerol (DAG) and water-soluble inositol 1,2-(cyclic)-phosphate (cIP), followed by (ii) a cyclic phosphodiesterase reaction where cIP is hydrolyzed to inositol-1-phosphate. The second reaction occurs with a soluble monomeric substrate and is much slower (with both lower Vmax and significant higher Km) than the first one (1, 2). Previous studies of the PI-PLC from Bacillus sp. highlight several kinetic properties common to a wide range of phospholipase enzymes. That bacterial PI-PLC exhibits (i) “interfacial activation”, an enhanced Vmax/Km, toward aggregated PI compared to monomeric PI (3–5), (ii) “surface dilution inhibition”, a decrease in specific activity as the surface concentration of the substrate is diluted with detergents or other phospholipids while keeping the total substrate concentration constant (2), and (iii) “scooting mode catalysis”, where enzyme completes several rounds of substrate turnover at the substrate interface before dissociating from the particle (5). The Bacillus sp. PI-PLC is also active towards GPI linkages (6).
PI-PLC from Listeria monocytogenes, a ubiquitous foodborne intracellular pathogen of humans and animals (7, 8), plays a role as virulence factor for the organism by aiding escape of the bacterium from the vacuoles in macrophages (9). PI-PLC aids in escape of the single-membrane primary vacuole (9–11) as well as aiding in the disruption of the inner membrane of the double membrane spreading vacuole (12). A broad-range PLC that primarily targets PC also plays a role in infectivity of this organism that is distinct from that of the PI-PLC (13–15). The L. monocytogenes PI-PLC appears to share the same general base and acid catalytic mechanism as Bacillus sp. PI-PLC (16, 17), and the enzyme has been shown to be activated by short-chain phosphatidylcholine (PC) molecules as well (18). There are, however, several major differences of the L. monocytogenes PI-PLC from Bacillus homologues, namely a high pI (above 9), kinetic activation by salts, and relatively weak activity toward GPI anchors (17, 19).
In this work, water-soluble synthetic short chain phosphatidylinositol and cIP along with PI/detergent mixed micelles and PI/PC vesicles were used as substrates to understand how salts and other phospholipids affect both steps of PI hydrolysis by the L. monocytogenes PI-PLC. The results indicate that L. monocytogenes PI-PLC binds tightly to anionic phospholipids (e.g., PI, PG) and tends to form aggregated complexes with those anionic lipids. The enzymatic specific activity is much lower when these complexes form. The two types of activators previously shown for this enzyme work by different mechanisms. Neutral amphiphiles, which enhance both steps of catalysis regardless of the aggregation state of the substrate, bind directly, though weakly, to the protein and enhance its catalytic ability (at sufficiently high mole fractions the amphiphiles can prevent the enzyme from forming the aggregated complexes). In contrast, moderate ionic strength (e.g., salts) only affects the phosphotransferase reaction and then is most efficient when an activating interface is present. The latter is likely to be the result of altering the surface electrostatics and reducing the formation of aggregated complexes, and possibly altering the residence time of the enzyme on interfaces. Several point mutations of surface hydrophobic residues show that the unusual kinetic behaviors have a strong hydrophobic component. A model rationalizing the role of both types of activators in modulating the enzyme activity in situ is presented.
Short chain dibutyroylphosphatidylinositol (diC4PI) was acquired from Echelon Biochemicals. Other phospholipids, including the long chain lipids POPC, POPS, POPG, DOPMe, PI from bovine liver, and the short chain phospholipids diC7PC, and diC6PC were obtained from Avanti Polar Lipids and used without further purification. Crude soybean PI (50%) was used for the enzymatic generation of cIP as described previously (2) by taking the advantage of the fact that the phosphotransferase (PI cleavage) activity is much higher than the phosphodiesterase (cIP hydrolysis) activity for bacterial PI-PLC. Carboxyfluorescein (a mixture of 5- and 6-carboxyfluorescein) was acquired from Eastman Kodak Co. Citraconic anhydride (2-methylmaleic anhydride) was purchased from Pierce. Other reagents including Triton X-100 and D2O (99.9%) were purchased from Sigma.
The IMPACT-CN system (New England Biolabs) was used to construct the plasmid for expression of this PI-PLC protein as a fusion protein with both a chitin binding domain for ease of purification and inducible self-cleavage activity to remove the fusion tag (20, 21). The plcA gene without the region coding for the signal sequence was inserted into the pTYB11 N-terminal fusion vector between Sap I and EcoR I restriction sites following the manufacturer’s instructions. The entire DNA sequence of the inserted target gene was determined and compared with that of pBS1462 (our original source of the plcA gene) to assure no PCR-induced errors were introduced. The recombinant plasmid, IMPACT-Y, was transformed into E. coli ER2566 cells for overexpression. An aliquot of the overnight culture was used to inoculate fresh LB medium with 100 µg/ml ampicillin in a rotary shaker (200 rpm) at 37°C. Expression of recombinant PI-PLC was induced by adding IPTG at a final concentration of 0.8 mM when OD600 of growing culture was between 0.7 and 0.8. After induction, the culture was incubated at 16°C for 20 h, harvested by centrifugation, and stored at −80°C. If needed, the frozen cell pellets were thawed at room temperature, resuspended in 50 ml ice-cold column buffer (20 mM Tris-HCl, 0.5 M NaCl, pH 8.4), sonicated on ice, and centrifuged to remove the cell debris. The cell extract supernatant of ER2566 harboring IMPACT-Y had high PI-PLC activity, albeit no obvious band of 88 kDa on SDS-PAGE gel. Recombinant L. monocytogenes PI-PLC was then obtained following the purification protocol in the manufacturer’s instruction manual. All protein concentrations were determined by Lowry assay. The best yield for recombinant L. monocytogenes PI-PLC was about 6 mg/L. The stock solution could be stored at 4°C without activity loss for ~1 month. SDS-PAGE analysis showed that the purity of PI-PLC after the chitin column was more than 85% and that PI-PLC was contaminated with a band around 50 kDa (the intein tag judging from the SDS-PAGE gel of chitin beads after elution of PI-PLC). If needed, a SP Sepharose Fast Flow (Amersham Pharmacia Biotech) column was used to further purify PI-PLC right after dialysis. The dialyzed eluate from the chitin column was applied at 2 ml/min onto the strong cation exchange column (~25 ml) equilibrated with 100 ml starting buffer (50 mM sodium phosphate buffer, pH 7.0). The protein was eluted using a NaCl gradient ranging from 0 to 0.5 M in 50 mM sodium phosphate (pH 7.0) at rate of 2 ml/min, dialyzed to remove the high salts, filtered and concentrated. Only the PI-PLC band was observed on the SDS-PAGE gel after these two columns.
A series of mutant plcA genes harboring mutations to code for W49A, L151A, L235A, F237A, F237W, and W49A/F237A were constructed using Quik-Change™ Site-Directed Mutagenesis Kit. All mutant plcA genes were sequenced to confirm that the correct mutations were introduced. Each recombinant plasmid IMPACT-Y was transformed into E. coli ER2566 competent cells for expression of mutant protein and purification as described above.
To alter the surface charge of the L. monocytogenes PI-PLC, the protein was treated with citraconic anhydride. Chemical modification with this reagent changes the positive charges of lysine residues on the protein surface to negative ones. Sodium phosphate, 0.5 M at pH 8.5, was used to provide adequate buffer capacity. The PI-PLC to be modified was purified with two columns (chitin column and SP Sepharose Fast Flow column), dialyzed against reaction buffer, and concentrated. 200 µl PI-PLC (6.5 mg/ml) was incubated with 50 µl citraconic anhydride (113 mM, diluted with reaction buffer) in 1 ml of reaction buffer at room temperature for 3 h after which the reaction mixture was dialyzed extensively against 20 mM Tris-HCl (pH 7.0) to quench the reaction. The change in pI for modified protein was measured via 2D gel electrophoresis separating the protein by isoelectric point and then mass. The two batches of citraconylated protein showed that the pI changed from the 9–10 region for unmodified (which migrated as a single peak) to between 7–9 (two distinct peaks were observed for modified enzyme).
The native protein molecular weight of purified L. monocytogenes PI-PLC was determined by laser light scattering (LS) (22) at the HHMI Biopolymer Facility and W.M. Keck Foundation Biotechnology Resource Laboratory. Because the protein interacted with the column matrix dramatically, it was analyzed by a micro-batch approach in which the sample from an injector loop was eluted through the system and the UV/LS/refractive index signals monitored while the buffer was continuously delivered from the HPLC system. Conditions (1.1 ml protein sample, 500 µl sample loop, and 0.3 ml/min for 1 min 20 sec) were established to provide a reading “plateau” (~0.2 ml, free of air and particle) in all three detectors. Transferrin with a comparable extinction coefficient to L. monocytogenes PI-PLC was used as a standard and the same calibration constant was applied to analyze the PI-PLC sample. The concentrations of PI-PLC during the reading plateau were measured with the UV absorbance at 280 nm (extinction coefficient = 33710 M−1·cm−1, 1 cm length). The molecular weight was determined by solving the equation that relates the excess scattered light to the concentration of solute and the weight-average molecular weight by ASTRA calculations (http://info.med.yale.edu/wmkeck/6_16_98/Astra2a.htm#Calculation of MW by).
The appropriate amount of a lipid stock solution in chloroform was placed in a 20 ml glass scintillation vial; the chloroform was removed with a rotary evaporator. The resulting lipid film was dissolved in 5 ml water, frozen on dry ice, lyophilized overnight, and rehydrated with appropriate buffer. Lipids supplied as powder were dissolved in the buffer directly. To prepare small unilamellar vesicles, the chilled phospholipids suspensions were sonicated on ice using a Branson sonifier W-150 ultrasonic cell disruptor with a 1 cm diameter probe until maximum clarity was achieved. Vesicle preparations were usually centrifuged in a bench-top centrifuge (14,000 rpm for 3 min) to remove any large particles.
The specific activity of PI-PLC was measured using 31P NMR spectroscopy (202.3 MHz on a Varian Inova 500 spectrometer) and parameters previously reported (1, 2). Two types of assays (end-point and continuous time point) were used for phosphotransferase activity (PI as substrate) and phosphodiesterase activity (cIP as substrate) respectively. Except for diC4PI, experiments were run in duplicate and the average specific activities are reported; the errors in specific activities were typically <15%.
For long chain PI from bovine liver (8 mM), dispersed in diC7PC or Triton X-100 (TX-100) micelles, or in small unilamellar vesicles (SUVs) in the absence and presence of POPC, in 50 mM HEPES, pH 7.0, containing 0.5 mg/ml BSA, enzyme (covering a range of 0.01 to 7.2 µg/ml) was added to the 200 µl assay mixture and incubated at 25°C. The reaction was quenched by the addition of 400 µl chloroform at appropriate incubation times (from 1 min to a few hours) which were chosen so that less than 20% PI cleavage occurred. The protein concentrations were kept constant within the same series of assays as protein concentration also affects the specific activity of L. monocytogenes PI-PLC. The cIP content in the aqueous phase (separated from the organic layer using centrifugation at 14,000 rpm for 6 min) was quantified in the 31P NMR spectrum using added glucose-6-phosphate (1 mM) as an internal standard. Triton X-100 and salt effects were also measured using the water-soluble substrate diC4PI. In this assay, 100 µl assay mixtures containing 2 mM diC4PI and 1.6 µg/ml enzyme were used.
In the continuous time point assay for cIP hydrolysis, the release of I-1-P product was monitored as a function of time until 20% of the cIP was converted to I-1-P using a 31P NMR experiment where the pre-acquisition delay was arrayed. The reaction was initiated by adding 3.6 or 7.2 µg/ml PI-PLC to 400 µl assay mixture in a NMR tube typically containing 10 mM cIP in the absence and presence of different additives (diC7PC, POPC, DOPMe, POPS, salts) in 50 mM imidazole buffer (pH 7.0), incubated at 25°C, and monitored for a period of time, typically from at least 30 minutes to a few hours. The rate of cIP hydrolysis was calculated from the slope of I-1-P production as a function of incubation time.
Lipid monolayers were formed by spreading 5 to 10 µl of the appropriate lipids (POPC, or DOPMe, or a mixture of 70% POPC and 30% DOPMe) dissolved in ethanol/hexane (1:9, v/v) onto 10 ml of 50 mM HEPES, pH 7.0, with KCl at three different concentrations (0, 0.16, or 0.5 M) contained in a circular Teflon trough (4 cm diameter × 1 cm deep). The surface pressure (π) of subphase (the solution in the trough) was measured using a Wilhelmy plate attached to a computer-controlled tensiometer as described for the B. thuringiensis PI-PLC (23). The subphase was gently mixed with a magnetic stir bar and the monolayer was allowed to equilibrate until a stable surface pressure (defined as the initial surface pressure, π0) was obtained. Then a protein solution was injected into the subphase through a small hole in the side of the trough without disruption of the monolayer. The change in surface pressure (Δπ) was measured as a function of time. Since the value of Δπ depends on the protein concentration in the low concentration range, protein concentration in the subphase was maintained high enough (above 3 µg/ml for PI-PLC) to ensure the observed Δπ represented a maximal value (24). In general, Δπ is inversely proportional to the π0 of the lipid monolayer and extrapolations of the Δπ versus π0 plot to π0=0 and Δπ=0 yields the maximum Δπ obtainable and the critical surface pressure (πc) that specifies an upper limit of π0 into which a protein can penetrate respectively.
Vesicle leakage promoted by the bound L. monocytogenes PI-PLC was monitored by the release of entrapped carboxyfluorescein (25). Vesicles were prepared by sonication of 20 mM of the appropriate lipids (POPC or DOPMe) with a solution of 100 mM carboxyfluorescein in 10 mM HEPES, pH 7.1. The nonencapsulated carboxyfluorescein was removed by rapid filtration of the vesicle solution through a Sephadex G-10 column at room temperature. POPC prepared with 150 mM KCl in the buffer along with the carboxyfluorescein were also prepared. The carboxyfluorescein-loaded vesicles were used immediately after gel filtration. Vesicle integrity in the absence or presence of PI-PLC protein was monitored with a Fluorolog R-3 spectrofluorimeter using an excitation wavelength of 490 nm and an emission wavelength of 520 nm. All the carboxyfluorescein assays were carried out in the dark and a narrow excitation slit width (1 nm), The percent leakage at time t, Lt, was expressed relative to the initial fluorescence Fo and the maximum fluorescence FTX obtained after complete lysis of the vesicles by the addition of 25 mM Triton X-100: Lt = (Ft − F0) / (FTX − F0) × 100%.
Aggregation and precipitation of PI-PLC with SUVs was assessed by incubating protein (0.036 mM) with pure POPG, pure POPC, and with POPG/POPC (1:1 or 1:25) SUVs, then separating any precipitate by centrifugation (15,000 rpm for 10 min with a bench-top centrifuge). The relative partitioning of the protein in the precipitate versus what remains in solution was measured by first lyophilizing both precipitate and solution, then examining relative amounts of protein in each phase by SDS-PAGE.
The time course for aggregation of DOPMe (0.86 mM) / POPC (3.1 mM) SUVs (in 50 mM HEPES, pH 7.0) promoted by different concentrations of PI-PLC (0.15, 0.75 and 1.5 mg/ml) was monitored by the increase in at optical density at 350 nm of the solution after the addition of protein (26). Monitoring the OD350 of corresponding vesicles without protein provided baseline turbidity.
Changes in phospholipid 31P linewidths were used to monitor the binding of diC6PC and diC7PC to L. monocytogene PI-PLC (2 mg/ml). The linewidth of the PC resonance was measured at various concentrations (0.1–20 mM) in the absence and presence of enzyme. Linewidth changes upon micellization of both of these short-chain lipids were small, so that the difference in linewidth caused by the presence of enzyme was initially assumed to reflect the amount of ligand bound to the enzyme in fast exchange with free ligand. As discussed previously (27), the difference in linewidth in the presence and absence of ligand at a given ligand concentration, (Δυobs−Δυo), is a function of total concentration of enzyme (ET) and ligand ([PC]T), the bound linewidth for the E·PC complex (Δυb), the dissociation constant (Kd), and n, the number of ligands bound per enzyme molecule. If there are multiple, independent binding sites for the PC (to form E·PCn) leading to the change in linewidth, and if the free [PC]>>[E·PC] complex, then the following equation holds:
Previously, the pH behavior for the L. monocytogenes PI-PLC phosphotransferase reaction was shown to vary slightly with the detergent matrix for the PI. A sharp optimum pH of 7.0 was observed toward PI solubilized in deoxycholate micelles with less than 30% activity at pH 6.5 and pH 7.5 (the suspension of PI in deoxycholate was no longer clear below pH 6.5), while a broader optimum pH range, from 5.5 to 6.5, was observed for enzyme acting on PI in TX-100 micelles (17). For comparison we tested the effect of pH on the phosphodiesterase reaction in the absence of detergent. The enzyme was optimally active towards cIP at pH 7.0, with 67% and 57% maximum activity at pH 6.5 and 7.5, respectively (Figure 1A). Thus, pH 7.0 was used for both phosphotransferase and cyclic phosphodiesterase kinetic assays.
PI can be presented to PI-PLC in a range of different matrices – micelles, unilamellar vesicles or water/organic solvent co-mixtures (1, 2, 4, 28, 29). Differences in PI-PLC activity toward the different types of interfaces can often provide insights into what factors control the activity of this enzyme. However, L. monocytogenes PI-PLC exhibited an added complication in that protein concentration also affected enzyme activity. Since the assay for phosphotransferase activity is not continuous, we monitored the generation of cIP from PI/TX-100 mixed micelles at various time points to avoid the influence of a lag or burst in product formation. As shown in Figure 1B, the phosphotransferase specific activity of L. monocytogenes PI-PLC toward PI (8 mM) in TX-100 (32 mM) mixed micelles increased dramatically from 150 to 1200 µmol min−1 mg−1 as protein was diluted from 0.4 to 0.01 µg/ml. Further dilution of the protein had only a small effect on specific activity. A similar dependence of specific activity on protein concentration was also observed with PI/diC7PC mixed micelles. To rule out any possible inhibitors from the expression/purification system, purified recombinant L. monocytogenes PI-PLC was further purified with cation exchange or gel filtration (with high salt concentration) chromatography, and the concentration dependence of specific activity on enzyme concentration still persisted.
The loss of activity with high enzyme concentrations has been discussed by Wang (30) for an association-dissociation enzyme system. PI-PLC could be aggregating in solution with the monomer more active than oligomer. However, this PI-PLC interacted anomalously with gel filtration resins even in the presence of 0.5 M NaCl, so that another method was needed to assess the size of the protein in solution. Laser light scattering, with some modification from the standard protocol, was used to characterize the native molecular mass of the protein. The average molecular mass for L. monocytogenes PI-PLC was determined as 28.2 kDa at a concentration of 0.14 mg/ml – well above that used in the kinetic assays. This is quite close to the predicted molecular weight of 33 kDa determined from the sequence. Thus, this PI-PLC exists as a monomer in solution in the absence of detergent. If the protein aggregation does account for the unusual dependence of specific activity on protein concentration, the oligomerization of PI-PLC must only occur in presence of an interface and/or the anionic substrates.
The phosphodiesterase activity of L. monocytogenes PI-PLC toward 5 to 40 mM cIP without detergent was investigated at two different protein concentrations. Because cIP is a poorer substrate than PI, considerably higher enzyme concentrations (3.6 or 7.2 µg/ml) were needed. In this high range of enzyme concentration, the activity was low and varied little between these two protein concentrations. However, there was a very significant difference in kinetic parameters extracted for the two different enzyme concentrations (Figure 2). In both cases, the data could be fit to the Michaelis-Menten equation: Vmax= 0.7±0.1 µmol·min−1·mg−1 and Km=12.5±5.0 mM at the lower PI-PLC concentration (3.6 µg/ml), and Vmax= 2.0±1.0 µmol·min−1·mg−1 with Km=100±62 mM at double the enzyme concentration. Although the errors in Vmax and Km are large for the higher protein concentration, the major effect of the higher enzyme concentration appears to be an increase in Km for water-soluble substrate cIP. Since L. monocytogenes PI-PLC exists as monomer in solution as shown by the laser light scattering measurement, the enzyme may aggregate with the anionic substrate present (in this case soluble cIP).Such aggregation would appear to specifically increase the apparent Km.
The Bacillus sp. PI-PLC enzymes exhibit a specific activation of both phosphotransferase and phosphodiesterase steps by interfaces containing phosphatidylcholine (2). It has been suggested that such activation also occurs for the L. monocytogenes PI-PLC (18). Therefore, we examined the effect of micelle matrix on the phosphotransferase activity of recombinant L. monocytogenes PI-PLC toward PI dispersed in TX-100. The enzyme concentration, as well as detergent, was varied. TX-100 is a nonionic detergent that clarifies PI bilayers at >2:1 TX-100/PI. If the surface concentration of PI is important for L. monocytogenes PI-PLC activity, then increasing TX-100 at fixed PI (8 mM) will eventually “inhibit” the enzyme leading to lower specific activity. Such surface dilution inhibition has been documented for the B. thuringiensis PI-PLC (2). However, L. monocytogenes PI-PLC presented different kinetic patterns depending on the protein concentration (Figure 3A). For 0.4 µg/ml enzyme, the enzyme exhibited no decrease in specific activity up to Xdet = 0.94 (16:1 TX-100/PI). In fact, the enzyme phosphotransferase specific activity increased dramatically as the surface concentration of the detergent increased above 0.90. The same amount of PI-PLC exhibited the same specific activities towards PI presented in diC7PC micelles suggesting that the diC7PC and TX-100 detergents were essentially equivalent. A similar trend, but with higher specific activities, was observed when the protein concentration in the assay was decreased two-fold.
At a much more dilute concentration, 0.02 µg/ml, a more typical ‘surface dilution curve’ was observed with the maximum phosphotransferase specific activity around Xdet = 0.78 (3.5:1 TX-100/PI). This detergent/substrate ratio for optimal activity is higher than what is needed to minimally solubilize PI bilayers (2:1 TX-100/PI). It has been reported that PI in predominantly PC bilayers is partially demixed (31). A similar demixing in mixed micelles could occur unless an excess of detergent is used. Alternatively, this could reflect a distinct activation by the detergent. At the very low concentration of this PI-PLC, the specific activity of L. monocytogenes PI-PLC decreased above Xdet= 0.80 as might be expected for a regime where the surface concentration of substrate influences the enzyme action. The dilute enzyme is very sensitive to the mole fraction PI in the interface. Comparing the activity curves of three protein concentrations in Figure 3A shows that significantly increased TX-100 could compensate for the lower specific activity at high enzyme concentration, since comparable high activities (1155 and 1260 µmol min−1mg−1) were achieved for all the protein concentrations examined. If the protein concentration decreased further, a lower optimum TX-100/PI ratio might be expected. However, 0.02 µg/ml is the limit of our assay system. Using [3H]PI as substrate, the phosphotransferase activity of PI-PLC was previously measured at 0.003 µg/ml enzyme. The optimal L. monocytogenes PI-PLC specific activity (~2900 µmol·min−1·mg−1) toward 0.08 mM PI micelles dispersed in TX-100 or deoxycholate was reported at a ratio of 10:1 TX-100/PI or 30:1 deoxycholate/PI (17). The higher specific activity than what we observe with low PI-PLC or moderate PI-PLC with high mole fractions of detergent (Xdet) is likely due to the addition of ammonium sulfate in the previous report.
The activity of L. monocytogenes PI-PLC toward a fixed bulk concentration of PI (8 mM) diluted in a POPC vesicle matrix was also studied. Compared with PI/TX-100 micelles, unilamellar vesicles of long-chain PI are poor substrates for PI-PLC from L. monocytogenes, necessitating higher protein concentrations for these assays. At a protein concentration of 1.6 µg/ml, the enzyme phosphotransferase specific activity increased about 2.2-fold as the mole fraction of POPC increased from 0 to 0.2. Once the POPC mole fraction was greater than 0.2 (at fixed PI), the L. monocytogenes PI-PLC specific activity stayed constant to a POPC mole fraction of 0.5 and then decreased as the mole fraction increased further to 0.8 (Figure 3B). Similar to PI/TX-100 micelles, higher specific activities toward PI/POPC SUVs were also observed at a lower enzyme concentration (0.4 µg/ml); the activation with 0.2 mole fraction in the vesicle was also larger than with the higher concentration of enzyme.
Examining amphiphile effects with monomeric substrate, in this case cIP, is one way of separating activator effects that alter the substrate versus those that alter the enzyme. With cIP at a fixed concentration, we can screen different additives for their effect on L. monocytogenes PI-PLC activity towards this monomeric substrate. Ryan and coworkers (18) have shown that diC6PC at concentrations below its CMC activate this enzyme with respect to non-natural water-soluble PI substrates. In our system, the plot of phosphodiesterase specific activity of L. monocytogenes PI-PLC versus diC6PC concentration was bimodal (Figure 4). Below the CMC of diC6PC (14 mM), there was ~10-fold increase in activity upon the addition of diC6PC; a second increase in specific activity was observed around the CMC of diC6PC, with PI-PLC activity 20-fold activity higher than for cIP alone. A slightly more hydrophobic amphiphile, diC7PC, also enhanced the activity of the enzyme toward the natural soluble substrate cIP (Figure 4). The bimodal dependence of specific activity on diC7PC concentration is consistent with the PC interfaces being more potent activators than monomeric PC at concentrations where it is monomeric in the absence of protein. POPC SUVs also enhanced cIP hydrolysis (Table 1 and Figure 4), although the maximum activity was considerably less than for diC7PC or TX-100 activation. SUVs composed of anionic non-substrate phospholipids (POPS and DOPMe) strongly inhibited the cIP hydrolysis reaction (Table 1). Vesicles with 0.2 mole fraction POPS disproportionately reduced L. monocytogenes PI-PLC activity toward cIP. Therefore, the enzyme must bind to those anionic phospholipids more tightly than it binds to POPC, and this interaction prevents cIP binding and hydrolysis.
Unlike what was observed for B. thuringiensis PI-PLC, TX-100 was also an effective activator of cIP hydrolysis by L. monocytogenes PI-PLC. Comparable activities were obtained with 3 mM TX-100 or 5 mM diC7PC (Table 1). Since the substrate cIP is monomeric, the amphiphile must activate L. monocytogenes by a direct interaction. However, whether in doing so it disfavors anionic ligand-induced protein aggregation that may be occurring or promotes a more active form of the enzyme is not clear.
Salt activation of PI cleavage by L. monocytogenes PI-PLC has been documented previously (17). Although no explanation was proposed, it was noted that in the absence of high salt, L. monocytogenes PI-PLC in solution behaved as a large aggregate on some gel filtration columns, but that 1 M (NH4)2SO4 attenuated this behavior. Salts can enhance hydrophobic interactions (e.g., promote insertion of protein hydrophobic side chains into bilayers) as well as weaken electrostatic interactions that hold complexes together. With this in mind, we examined the effect of several different salts ((NH4)2SO4, NH4Cl, K2SO4, and KCl) on L. monocytogenes PI-PLC cleavage of PI in TX-100 or diC7PC mixed micelles and hydrolysis of cIP. As shown in Table 2, the ionic strength, and not the salt identity, was important for L. monocytogenes PI-PLC activation towards PI in micelles. At fixed enzyme concentration and 8 mM PI (solubilized in 32 mM TX-100), L. monocytogenes PI-PLC activity displayed a hyperbolic dependence on concentration of (NH4)2SO4. For 0.2 µg/ml L. monocytogenes PI-PLC, the apparent Vmax was 2022±202 µmol·min−1·mg−1 with a K0.5 (concentration of salt for half the maximum specific activity) of 38±16 mM; for 1.6 µg/ml L. monocytogenes PI-PLC, the apparent Vmax decreased somewhat to 1355±256 µmol·min−1·mg−1 with a comparable salt K0.5 (42±20 mM). The lower Vmax extrapolated from the salt dependence of activity at an 8-fold higher L. monocytogenes PI-PLC concentration indicates that the presence of salt does not completely eliminate the unfavorable factors associated with high enzyme concentration. The ratio of specific activity for 0.21 µg/ml PI-PLC compared to 1.6 µg/ml at 0, 20, and 100 mM (NH4)2SO4, is 3.7, 1.6, and 1.5, respectively. These decreases suggesting that (NH4)2SO4 does partially compensate for the negative effects of high enzyme concentration.
PI-PLC cleavage of PI presented in SUVs was also activated by salt (Table 2). With KCl, the effect leveled off around 0.3 M of the salt. With PC in the SUV along with the PI, the addition of salt also led to a much larger increase in specific activity (Table 2).
In contrast to its activation of PI cleavage by L. monocytogenes PI-PLC, (NH4)2SO4 (as well as other salts) inhibited cIP hydrolysis whether diC7PC was present or not (Table 1). These results make it unlikely that salt biases the enzyme toward a more active conformation, otherwise, it should have enhanced cIP hydrolysis as well as PI cleavage. The added salt could alter the properties of substrate interfaces making them more susceptible to L. monocytogenes PI-PLC or it could alter the interaction of the enzyme with the interface. One way of testing these explanations is to examine the effect of salts and amphiphiles on cleavage of diC4PI, a soluble phosphodiesterase substrate with no tendency to partition into aggregates as long as the substrate concentration is low (the CMC for this lipid is likely to be >150 mM since diC4PC has a CMC of 250 mM (32) and other short-chain PI species exhibit CMC values comparable to the same chain length PC compound (33)). As a way of ruling out any aggregation of the diC4PI induced by the amphiphile or salt, the 31P linewidth of 2 mM diC4PI was measured in buffer in the absence and presence of 100 mM KCl, 100 mM TX-100, and then both components. The linewidth, 1.6±0.1 Hz, was unchanged with the additives.
Keeping the diC4PI concentration at 2 mM, we measured the effect of KCl and TX-100 on L. monocytogenes PI-PLC activity towards the soluble substrate (Table 2). With either 75 mM KCl or 32 mM TX-100 only, the specific activity of L. monocytogenes PI-PLC toward diC4PI increased slightly (about 1.6-fold). However, if both salt and TX-100 were present, there was ~14-fold increase in L. monocytogenes PI-PLC toward diC4PI. A similar synergistic effect was observed with monomeric diC6PC added – a much larger activation was seen with both KCl and amphiphile present. This synergistic effect cannot reflect altering the substrate interface because the substrate is monomeric. This suggests that for the phosphotransferase reaction, the presence of an amphiphile is important for the kinetic activation observed with moderate ionic strength. KCl (0.1 M) could lower the CMC of diC6PC slightly, but the effect in the absence of enzyme is not significant (34). However, the slight decrease in specific activity for L. monocytogenes PI-PLC acting on cIP with monomeric diC6PC approaching the CMC (Figure 4) may suggest that the enzyme itself could nucleate ‘mini’-micelle formation when it binds a diC6PC molecule. Studies of B. thuringiensis PI-PLC showed that diC6PC added at a monomeric concentration can bind to the protein, at a site distinct from the catalytic site, and activate the enzyme (27). DiC4PI is a poor substrate for L. monocytogenes PI-PLC, necessitating long incubation times (2 to 5 h) if the same enzyme concentration is to be used in the absence and presence of salts and amphiphiles. With TX-100 present, final products included both cIP and I-1-P for incubation times longer than 1 h. However, if diC6PC was added as the amphiphile, cIP was the only product even after 2 h. There is a difference when the L. monocytogenes PI-PLC interacts with TX-100 versus a short-chain PC molecule such that binding of diC6PC to the protein promoted cIP release from the active site, while with TX-100 as the amphiphile both products were detected. Nonetheless, both amphiphiles effectively enhanced the enzyme activity.
Short-chain PC molecules activate the L. monocytogenes PI-PLC toward cIP in the absence of salt, so there must be a direct interaction of the enzyme with the PC molecules. This can be probed by P NMR linewidth studies (27). As shown in Figure 5, a threshold of diC7PC (≥1 mM) was needed before significant broadening of the phospholipid resonance (0.8 Hz maximum under the conditions used) was detected. As more PC was added, the linewidth difference in the presence and absence of L. monocytogenes PI-PLC decreased. This type of profile is typical for multiple PC molecules interacting with the enzyme (and occupation of all these sites needed before the measured spectral change occurs). As shown in Figure 5B, the linewidth difference for diC7PC caused by the presence of L. monocytogenes PI-PLC (2 mg/ml) was consistent with an ‘n’ value of 3–4, (Δυb−Δυ0) = 5.5±0.6 Hz, and apparent Kd = 1.5±0.6 mM. The analysis with diC6PC binding to L. monocytogenes PI-PLC yielded the same ‘n’, (Δυb−Δυ0) = 3.3±0.8 Hz, and Kd = 17±10 mM (data not shown). In both cases, the ‘Kd’ describing the interaction was essentially at the CMC of the short-chain PC. The linewidth for bound PC would suggest significant mobility consistent with a small aggregate or mini-micelle with the protein.
L. monocytogenes PI-PLC binding to SUVs of POPC or anionic vesicles (DOPMe, POPG, etc.) was also examined using a filtration / centrifugation binding assay (23). L. monocytogenes PI-PLC (30 µg/ml or 0.9 µM) had weak affinity for POPC SUVs – essentially >90% of the protein was eluted through the filter upon centrifugation even after incubation with 1 mM POPC SUVs. In contrast, this bacterial PI-PLC bound very tightly to anionic phospholipid SUVs regardless of headgroup. For DOPMe concentrations >10 µM, essentially all the protein was bound to the phospholipids and none passed through the filter upon centrifugation. The dramatically slower rate for the solution to pass through the filter suggested that large aggregates of the proteins and vesicles might be forming. To test this, we incubated 36 µM PI-PLC (a considerably higher concentration) with different concentrations of POPG, POPC or mixed POPG/POPC SUVs (Figure 6A). There was a significant anionic lipid-induced precipitation of protein (Figure 6A). With 0.2 mM POPG and 0.036 mM protein, most of the protein (64±10%) was associated with the precipitate. With the POPG reduced to 0.02 mM, a moderate amount of the protein (42±5%) was still associated with the precipitate. When that amount of POPG was presented in 0.5 mM POPC, the amount of protein in the precipitate was significantly reduced (16±3%). Mixing PI-PLC with pure POPC SUVs led to only 5±3% of the protein in a precipitate, much of which could be due to a small fraction of larger vesicles with some PLC bound. If micelles as opposed to vesicles were examined, more PG was needed to precipitate the protein (data not shown). The amount of each lipid in the precipitate (extracted with methanol/chloroform) was measured with 31P NMR, and the lipid composition in the precipitate for POPG/PC SUVs or micelles was very similar to that in solution. Therefore, increasing PI-PLC specific activity as detergent or PC is increased in the SUV assay system is likely to reflect (at least in part) disruption of a large, nonproductive aggregated complex of the enzyme with several vesicles that sequesters it from the substrate in other vesicles.
The progress of aggregation / precipitation was monitored by looking at OD350 with continuous stirring (Figure 6B). After adding different concentrations of L. monocytogenes PI-PLC to DOPMe/POPC (1:3.6) SUVs, an instant and very dramatic increase in optical density, corresponding to the formation of the precipitate, was observed. The optical density decreased over time, but did not settle back to the initial value; the limiting OD350 was highest for the highest concentration of protein. Similar behavior was also observed for incubating this PI-PLC with POPG/POPC SUVs, but not for pure PC vesicles. Since the precipitate could be dispersed by adding salt at concentrations (0.1–0.3 M) comparable to those used in the kinetics), it does not reflect vesicle fusion.
The release of carboxyfluorescein trapped in POPC/POPG vesicles upon the addition of protein was monitored to check the stability of the vesicles in the presence of PI-PLC (Figure 6C). L. monocytogenes PI-PLC caused <25% carboxyfluorescein leakage (compared to total leakage from TX-100) within 2 hours (the reaction time course used for most assays) for three vesicle systems (POPC, POPC with KCl, and DOPMe (as the nonhydrolyzable substrate analogue)). The initial release rate of carboxyfluorescein increased in the order of the affinity of enzyme for the interface (DOPMe > POPC > POPC/KCl). The more tightly the enzyme binds to a given vesicle surface, the longer the enzyme stays on that structure (a nonproductive aggregated complex in the case of the pre DOPMe SUVs) and less chance it will dissociate and bind to another vesicle.
Monolayer studies provide insight into how PC and likely other neutral amphiphiles as well as salt contribute to the kinetic activation of the enzyme. L. monocytogenes PI-PLC binds tightly to anionic monolayers of DOPMe (Figure 7A), and πC, the threshold pressure above that the enzyme cannot insert into the membrane, is high (46 dyne/cm) for this protein under these conditions. The strong interaction with an anionic surface may not be surprising given the high pI for this protein and the potent inhibition of cIP hydrolysis by anionic phospholipid vesicles (Table 2). The addition of KCl at a concentration comparable to its effective K0.5 for kinetic activation (0.16 M) had little effect on πC for the enzyme binding to a DOPMe monolayer, indicating binding and insertion in the anionic interface were not dramatically affected by this concentration of salt (Figure 7B). The slope of the plot decreased suggesting that the portion of protein maximally inserted into a membrane had decreased in the higher ionic salt solution. Binding of L. monocytogenes PI-PLC to POPC membranes was characterized by a considerably smaller πC (31 dyne/cm) that was further reduced in the presence of 0.16 M KCl (to 23 dyne/cm). Binding of L. monocytogenes PI-PLC to a mixed monolayer of 70 mol% PC and 30 mol % DOPMe in the absence of KCl resembled the binding curve to a pure DOPMe monolayer (Figure 7A); πC was unaltered while the maximum Δπ extrapolated decreased reflecting the PC content (24 versus 32 dyne/cm). The addition of 0.16 M KCl to this mixed monolayer had a pronounced effect on L. monocytogenes PI-PLC binding. πC was decreased to 37 dyne/cm while Δπ was the same as that extrapolated from L. monocytogenes PI-PLC binding to a PC monolayer. When the KCl in the solution was increased to 0.5 M, the protein showed very minimal binding irrespective of any lipid composition indicating the driving force for the L. monocytogenes PI-PLC to reach the lipid surface has a large electrostatic component. The synergistic effect of PC and KCl in reducing the interaction of L. monocytogenes PI-PLC with the monolayer correlates with higher enzymatic activity observed with PI/PC SUVs upon the addition of KCl.
The high pI for L. monocytogenes PI-PLC coupled with the aggregation and monolayer results suggested that strong interactions with anionic phospholipids can be detrimental to enzyme action. Citraconic anhydride was used to modify a portion of the Lys residues on the protein surface to see if surface charge had a dominant effect on the kinetic behavior. This reagent converts surface Lys residues to amides with a terminal carboxylate. For each Lys modified the charge is decreased by 2. The pI of two separate batches of modified enzyme were checked by 2D gel. Two distinct spots were detected with the pI between pH 7 and 9. Given the 28 Lys in the protein this shift is consistent with ~5–7 Lys modified (estimated by changing several Lys to a Glu and computing the pI). A kinetic analysis of PI cleavage in the TX-100 mixed micelle system (Table 3) showed that the modified enzymes still could be activated to about the same extent by salt and amphiphiles. However, the effect of protein concentration was diminished compared to the control enzyme suggesting that the high cationic character of the protein is a factor in the unusual dependence of specific activity on enzyme concentration. The smaller effects on salt and amphiphile activation indicate that either (i) Arg and/or remaining unmodified Lys residues account for this kinetic activation, or (ii) a hydrophobic interaction with the L. monocytogenes PI-PLC is also an important contributor to the activation by salts and amphiphiles.
Since the unusual dependence of L. monocytogenes PI-PLC activity on the enzyme concentration, amphiphiles and salts persisted after citraconylation, hydrophobic interactions of the protein with interfaces may be responsible for the unusual kinetic behavior. For Bacillus sp. PI-PLC, several hydrophobic amino acid residues from short α-helix B and some loops, arranged in a semicircle around the active site cleft, form a mobile hydrophobic ridge fully exposed to solvent. Among the exposed residues, two tryptophans, Trp-47 in α-helix B (PIKQVWG) and Trp-242 in one particular loop (SGGTAWN), were shown to be important for the enzyme to bind to both activating zwitterionic and substrate anionic interfaces (35). The structure of L. monocytogenes PI-PLC also consists of a short α-helix B and an analogous rim loop (36). Although the sequence similarity is low for these two regions, the structure-based sequence alignment of L. monocytogenes versus B. cereus PI-PLC did provide several candidates. Both Trp-49 and Leu-51 in α-helix B (ITWTLTKP) and Leu-235 and Phe-237 in the loop (SATSLTF) were chosen as targets for mutagenesis (the location of these residues in the structure is shown in Figure 8A).
Most of the mutant enzymes constructed had reasonable activity and were much more active than wildtype at high protein concentrations (>1 mg/ml). Only F237A showed an increase in specific activity as the protein concentration decreased, behavior like wildtype enzyme (Figure 8B). All the rest either showed little change or a decrease in specific activity as the protein became very dilute. The latter effect could reflect some surface unfolding of a protein under very dilute conditions. W49A and W49/F237A exhibited very low specific activities for cleavage of PI in TX-100 micelles compared to recombinant wildtype enzyme (Figure 8C), consistent with Trp49 having a different role than in the B. thuringiensis enzyme. In the L. monocytogenes PI-PLC crystal structure, Trp49 is tucked in towards the active site, where it might help align substrate, rather than situated facing the external medium as does Trp47 in the B. cereus PI-PLC (36). In the B. cereus PI-PLC, Trp47 is critical to anchoring the protein to interfaces. However, in the L. monocytogenes PI-PLC, an intact helix B would not orient Trp49 for insertion into the membrane. The position and proximity of Trp49 to the active site may explain why its replacement with a small residue alters the efficiency of the enzyme. At a concentration of 0.82 µg/ml, the specific activity of W49A toward PI was only 2% that of the wildtype enzyme. These two fairly inactive enzymes also exhibited a very distinct decrease in specific activity as the enzyme was diluted in the assay – the opposite of the trends for wildtype enzyme. Incubation of 10 mM cIP and 5 mM TX-100 with 7.2 µg/ml W49A or W49A/F237A for 24 h did not generate any 31P NMR-detectable I-1-P. Clearly, modifying key hydrophobic groups in helix B and the active site loop reduce or remove the unusual concentration dependence of wildtype L. monocytogenes PI-PLC.
The surface dilution profile of the more active mutant enzymes (all assayed with fixed PI-PLC of 0.2 µg/ml and 8 mM PI) were also different from wildtype. At fixed total PI, all enzymes exhibited the increase in specific activity as the mole fraction detergent increased from 0.67 to 0.80 (detergent/PI ratio changing from 2:1 to 4:1), suggesting the higher amount of TX-100 was needed to adequately solubilize the PI in this system (emphasized in Figure 9A). However, the further 3-fold increase in specific activity for this concentration of wildtype enzyme as the mole fraction of detergent was increased above 0.9 (Figure 9B) was not observed for any of the mutant enzymes.
We also examined these mutant enzymes for salt effect on the phosphotransferase reaction and amphiphile effects on cIP hydrolysis. Added KCl (0.15 M) still enhanced PI-PLC cleavage of PI in TX-100 micelles for the mutant enzymes (Figure 10A), although KCl effects were much more modest for W49A (1.8-fold increase), L51A (1.6-fold) and L235A (1.4-fold) compared to wildtype, F237A or F237W. In contrast to the salt activation of phosphotransferase activity, all mutant enzymes exhibited a large and similar magnitude increase in the specific activity for cIP hydrolysis with TX-100 micelles present (Figure 10B).
Activation of the B. thuringiensis PI-PLC by phosphatidylcholine interfaces has been studied in detail (2, 23, 27, 28, 29, 35). Binding of B. thuringiensis PI-PLC to PC interfaces, which appears to have a strong hydrophobic component (23), alters the enzyme conformation so that it is a better catalyst in both phosphotransferase and cyclic phosphodiesterase reactions. The activation is specific for zwitterionic phospholipids (2). The L. monocytogenes PI-PLC has a similar structure to the B. cereus enzyme (36), which might suggest very similar catalytic and possibly regulatory behavior. While the catalytic mechanism may be essentially the same (16), the regulation of activity by amphiphiles and ionic strength is quite different for the more cationic protein.
Activators for L. monocytogenes PI-PLC exist in two classes: (i) neutral amphiphiles (PC and TX-100) and (ii) moderate salt concentrations. The first of these directly bind (albeit weakly) to the enzyme (e.g., diC7PC binding to L. monocytogenes PI-PLC as monitored by 31P linewidth changes) and alter the interaction of the enzyme with its anionic substrates. As with B. thuringiensis PI-PLC (4), adding a micellar activator to an assay system with monomer substrates is more effective than adding a monomeric activator. However, micelles of a neutral amphiphile such as TX-100 are just as effective as diC7PC suggesting the amphiphile binding may not involve a specific site on the enzyme. The affinities of the two PI-PC enzymes for anionic versus zwitterionic interfaces are exact opposites. L. monocytogenes PI-PLC binds tightly to surfaces high in anionic phospholipids with much weaker interactions with zwitterionic PC interfaces, while B. thuringiensis PI-PLC prefers zwitterionic lipids. Furthermore, the L. monocytogenes PI-PLC tends to form large aggregates with vesicles or micelles rich in anionic lipids. As revealed from its crystal structure (36), there are many basic residues clustered on the bottom side of the L. monocytogenes PI-PLC TIM-barrel that are far away from the rather hydrophobic opening to the positively charged active site. With PI dispersed in a variety of aggregates (mixed micelles or vesicles), the electrostatic interaction between the positive residues of enzyme and anionic substrate drives the enzyme to the interface where it will bind tightly. The basic residues on the face of the protein opposite the active site will cause the clustering of nearby negatively charged lipid surfaces (illustrated schematically in Figure 11A). For vesicles this could cause aggregation of PI in the same vesicle or more likely cause aggregation with other highly anionic vesicles, forming a complex in which the enzyme is sequestered from other substrate-containing vesicles. A significantly increased surface concentration of zwitterionic / neutral amphiphiles disperses the anionic substrate, and shields charges on the protein from sparser anionic lipid patches on the same or other particles. Protein-protein interactions must also contribute to this trapping of the enzyme since dramatically reducing the protein concentration also increases enzyme specific activity (Figure 11B).
Increasing the solution ionic strength also relieves formation of the trapped enzyme by reducing the hydrophobic interaction between enzyme and anionic surfaces. That cIP hydrolysis by L. monocytogenes PI-PLC is not affected by salt indicates this activator is altering the way the protein interacts with anionic interfaces. The kinetic and biophysical studies presented here show that increased ionic strength alters the way L. monocytogenes PI-PLC binds to anionic interfaces. The monolayer studies support the idea that KCl weakens the interaction of the protein with negatively charged surfaces. Increased ionic strength reduces the maximum penetration (Δπ) of the enzyme and πC to a greater extent when PC is present in the interface. These results are critical in explaining the synergistic effects of POPC and KCl on PI-PLC cleavage of PI in vesicles. Clearly, a high πC and Δπ correlate with lower activity of L. monocytogenes PI-PLC. High salt is also likely to affect PI behavior in the membrane by suppressing any demixing of PI and neutral/zwitteronic amphiphiles (31), an effect that may also contribute to enhanced activity.
However, the soluble substrates, diC4PI for the phosphotransferase reaction and cIP, provide an interesting twist in understanding how the two types of activators work. DiC4PI cleavage is not dramatically enhanced by L. monocytogenes PI-PLC binding to the same concentration of diC6PC or TX-100 micelles as used in cIP assays. A large increase in the enzyme specific activity is only observed with both the amphiphile and moderate (0.075 M) KCl. The difference between diC4PI and cIP is that the first has acyl chains, short but with some hydrophobic character. There are multiple cationic sites on the enzyme that could bind this anionic phospholipid leading to inhibited enzyme. cIP may have a much weaker affinity for these secondary sites. Amphiphile alone may not compete well at these secondary sites, so that activation by amphiphile at low ionic strength is small. Increased salt alone (0.1–0.2 M) may not displace these molecules from the protein. However, the presence of an activating amphiphile and added salt might be more effective at weakening electrostatic interactions of diC4PI with secondary sites and allowing the enzyme with bound substrate to bind to the activating amphiphile surface. This would lead to high specific activity toward diC4PI only observed in the presence of both amphiphile and salt. In contrast, the enzyme would be adequately activated by amphiphile toward cIP because this substrate molecule does not interact with other sites on the enzyme.
The biophysical studies with wildtype enzyme suggest that hydrophobic interactions as well as electrostatic interactions with negatively charged surfaces contribute to the unusual kinetic behavior of this enzyme. Modifying enzyme surface charge by citraconylation, which should at least partially affect any kinetic effects linked to the high cationic character of this protein, had little effect on surface dilution or salt effects. The only significant effect was that enzyme with reduced positive charge showed a smaller increase in specific activity as the protein was diluted. The PI-PLC mutants we constructed all modified hydrophobic groups at the barrel rim in features that might be linked to membrane binding. Except for removal of Trp49 (which is the one residue in this series facing inwards towards the active site and possibly involved in orienting or promoting proper access of the substrates), the PI-PLC mutant enzymes we constructed were all quite active, and most importantly lost the strikingly large increase in specific activity with enzyme dilution or with surface dilution of the PI in mixed micelles. Alanine substitutions clearly will reduce hydrophobic interactions, and most of the mutants were not as active as the wildtype if maximum specific activities are compared regardless of protein concentration (Figure 8). The kinetic behavior of these specific mutant enzymes contrasts strongly with that observed for citraconylated enzyme. Both the enzyme concentration dependence and apparent surface dilution inhibition are dramatically reduced for all mutants with altered surface hydrophobic side chains (except W49A which is fairly inactive). Neither trend is abolished when the enzyme surface charge is reduced by modification of the lysines.
The behavior of the mutant enzymes also indicates that salt activation is distinct from the enzyme dilution and surface dilution activation. KCl activation for W49A, L51A and L235A is significantly reduced. In contrast, citraconylated enzyme exhibits a strong KCl activation. If added salt weakens the protein binding to the interface (an observation supported by the monolayer studies), then reducing the binding by removing hydrophobic residues should lessen the salt effect as observed with these mutant enzymes (they are already weaker binders).
Perhaps the most interesting mutation is F237W. Replacement of hydrophobic Phe237 with tryptophan, which prefers interfacial regions, may help to alter the conformation of the protein on a target membrane. F237W PI-PLC, even at high concentrations, has high specific activity comparable to dilute unaltered recombinant PI-PLC. This kinetic behavior suggests that at higher concentrations, F237W does not form the aggregates with anionic lipid-rich vesicles that are disrupted by excess detergent and salt, although it is still activated (to about the same extent as wildtype) by salt. Substitution with alanine generates a mutant that most approaches wildtype PI-PLC in its dependence on enzyme concentration. These results suggest that Phe237 is a key residue in the very tight binding of PI-PLC to membranes containing anionic phospholipids. One could suggest that the high activity of F237W at higher enzyme concentrations is because it still binds to membranes, but the Trp alters the orientation by perhaps not inserting to the same depth as a Phe side chain. The F237W protein may bind more weakly than wildtype. The net result is that the enzyme is not trapped in underproductive precipitates/aggregates, etc. Future studies directed at vesicle binding (with mixtures of anionic phospholipids and PC) of wildtype and mutant proteins under conditions of very low concentrations of proteins should be extremely useful for unraveling how the specific residues mutated contribute to membrane partitioning. The phenylalanine at this particular position may be the key to tightly anchoring the L. monocytogenes PI-PLC on the inner leaflet of the plasma membrane of the host cell or onto the inner leaflet of the inner bilayer in the secondary vacuole where it can cleave PI to generate DAG.
Several bacteria, both pathogenic and non-pathogenic, secrete PI-PLC enzymes into the media. The biological role of this enzyme, along with other nonspecific phospholipases, is to aid in survival of the organism, particularly as it infects mammalian cells. Those PI-PLC that cleave GPI-anchored proteins, such as the PI-PLC secreted by B. thuringiensis, B. cereus, and Staphylococcus aureus, are well poised to target those membrane components in the extracellular leaflet of the plasma membrane, which is rich in the zwitterionic lipids PC and sphingomyelin. The specific PC binding and activation of the Bacillus PI-PLC enzymes would help ensure that they bind the target cell membrane where they could catalyze the cleavage of GPI-anchors. However, secreted PI-PLC enzymes whose targets are PI and not GPI-anchors, must have a different role in bacterial survival since PI is not found in the external leaflet of most organisms. In fact, they should not be targeted to PC-rich membranes but should have high affinity for the inner leaflet of the plasma membrane of the target cell.
In L. monocytogenes infections, the bacteria are surrounded by two types of vacuolar membranes at different stages and the PI-PLC is involved in both stages (11, 37–42). Upon initial infection of the host cell, the bacterium is surrounded by a single-membrane vacuole. In macrophages, both listeriolysin O and L. monocytogenes PI-PLC expression are upregulated in this phagosome (43). From the phagosome, L. monocytogenes PI-PLC could presumably gain access to host PI by means of phagosomal permeabilization and eventual destruction. Activation of host PKCβ by means of DAG generated from intracellular PI and elevated intracellular calcium may play a role in escape from this phagosome (40, 42). The inner leaflet of the primary vacuole membrane is presumably like the external leaflet of the plasma membrane – rich in PC or sphingomyelin with a low content of anionic phospholipids. Since the L. monocytogenes PI-PLC has weak affinity for PC (and presumably for sphingomyelin and PE as well) under the moderate ionic strengths in cells, it will stay in the vacuolar fluid and be dispersed into the cytoplasm. Once there it will home in on the negatively charged components, including PI, of the target plasma membrane. It should be noted that intracellular concentrations of a typical mammalian cell are 5 to 15 mM NaCl, 140 mM KCl, 0.5 mM Mg2+, and 0.1 mM Ca2+ (44), an ionic strength similar to what activates L. monocytogenes PI-PLC towards mixed PI/PC SUVs. Once at the plasma membrane, the bacterial PI-PLC will generate DAG required to activate host pathways. However, too much PI cleaved would be detrimental to the host and the replication efforts of the bacterium. Moderation of host PI cleavage by this bacterial PI-PLC could occur when the tightly bound PI-PLC is sequestered with other host anionic lipids and can no longer easily access more PI.
PI-PLC also plays an important role in escape from a secondary vacuole for cell-to-cell spreading of these bacteria (12). In this compartment the bacterium is in a double membrane where the inner bilayer is derived from the primary cell and the outer bilayer from the secondary cell. The inner leaflet of the inner membrane will contain PE, PS and PI, the last of these the substrate for the PI-PLC. Thus, the PI-PLC enzyme secreted into the secondary vacuolar fluid will aid in dissolving this inner bilayer by hydrolysis of the PI and likely disruption and fusion of the bilayer after sufficient DAG is produced. The activity of the non-specific phospholipase/sphingomyelinase (PC-PLC) will generate even larger amounts of DAG plus ceramide, potentially leading to membrane fusion (44), and also cause complete disruption of the inner membrane (12). The high affinity of the wildtype enzyme for anionic phospholipids will insure that it localizes on this membrane surface. Once the inner membrane is dissolved, PI-PLC faces the outer leaflet of the outer membrane – rich in PC and SPM – and will not bind tightly. In escaping the secondary vacuole, there is fusion with early endosomes and a decrease in pH. While this may activate the non-specific PC-PLC and listeriolysin, the fairly narrow pH range of PI-PLC suggests that it would not be very active if the pH dropped much. Therefore, it must contribute to disruption of the inner membrane prior to a more extreme pH drop (to ~5.5) and activation of LLO. Several of the mutant enzymes we have constructed may help sort out the relevance of the PI-PLC at different stages of infection by L. monocytogenes.
†This work was supported by N.I.H. GM60418 (M.F.R.), AI45153 (H.G.) and GM52598 (W.C.).