UDP-glucose Deficiency Causes Hypersensitivity to the Cytotoxic Effect of Clostridium perfringens Phospholipase C*

A Chinese hamster cell line with a mutation in the UDP-glucose pyrophosphorylase (UDPG:PP) gene leading to UDP-glucose deficiency as well as a revertant cell were previously isolated. We now show that the mutant cell is 105 times more sensitive to the cytotoxic effect of Clostridium perfringensphospholipase C (PLC) than the revertant cell. To clarify whether there is a connection between the UDP-glucose deficiency and the hypersensitivity to C. perfringens PLC, stable transfectant cells were prepared using a wild type UDPG:PP cDNA. Clones of the mutant transfected with a construct having the insert in the sense orientation had increased their UDP-glucose level, whereas those of the revertant transfected with a UDPG:PP antisense had reduced their level of UDP-glucose compared with control clones transfected with the vector. Exposure of these two types of transfectant clones to C. perfringens PLC demonstrated that a cellular UDP-glucose deficiency causes hypersensitivity to the cytotoxic effect of this phospholipase. Further experiments with genetically engineered C. perfringens PLC variants showed that the sphingomyelinase activity and the C-domain are required for its cytotoxic effect in UDP-glucose-deficient cells.

mechanism by which it induces an extensive destruction of tissues remains unknown (3).
Structurally, the C. perfringens PLC is a single-chain zinc metalloenzyme comprising two domains joined by a hinge region: a catalytic N-terminal domain of 240 amino acid residues and a C-terminal domain of 120 residues (4,5). The N-terminal domain has extensive amino acid sequence similarity with the nonlethal Bacillus cereus phosphatidylcholine (PC)-hydrolyzing phospholipase C (PC-PLC) (6). The crystal structure of the B. cereus PC-PLC was used as the basis for site-directed mutagenesis studies, which indeed showed that the critical amino acid residues for Zn 2ϩ binding and catalysis are conserved between both enzymes (7)(8)(9). The C-terminal domain of the C. perfringens PLC, which has no counterpart in the B. cereus PC-PLC (10,11), mediates interactions with membrane phospholipids in a calcium-dependent manner (12).
A Chinese hamster fibroblast mutant cell line (Don Q) that is 10 5 times more sensitive to C. perfringens PLC than the parental cell (Don wt) was previously isolated (13). Don Q is 10 4 times more resistant to Clostridium difficile glucosyltransferase toxin B (TcdB) (14) than Don wt, due to a UDP-Glc deficiency (15) caused by a recessive point mutation in the UDP-Glc pyrophosphorylase (UDPG:PP) gene (16). A spontaneous revertant cell (Don QR) was isolated from Don Q and showed to have the same sensitivity to TcdB as Don wt and a partially compensated UDP-Glc level (15,16).
UDP-Glc deficiency occurs in cells cultured under hypoxia or in low glucose-containing media (17)(18)(19)(20). Since these conditions are two of the hallmarks of ischemia, it is likely that a low UDP-Glc level occurs also in ischemic tissues (16). The aim of this work was to determine whether a cellular UDP-Glc deficiency affects the sensitivity to the cytotoxic effect of C. perfringens PLC as well as to map which parts of this toxin are needed for its cytotoxicity.

EXPERIMENTAL PROCEDURES
Cell Cultures-Diploid Chinese hamster lung fibroblasts (Don cells, ATCC number CCL 16; Flow Laboratories, Irvine, Scotland; referred to here as Don wt), a mutant of this cell line (13) (referred to here as Don Q), and a spontaneous revertant of this mutant (16) (referred to here as Don QR) were cultivated in flasks (75 or 175 cm 2 ; A/S Nunc, Roskilde, Denmark) or in polystyrene 96-well culture plates (Nunc) in Eagle's minimal essential medium (Flow) supplemented with 10% fetal bovine serum, 5 mM L-glutamine, penicillin (100 units/ml), and streptomycin (100 g/ml) in a humid atmosphere containing 5% CO 2 at 37°C. Transfectant cells were cultured in the same medium in the presence of geneticin (0.4 g/ml) (Life Technologies, Inc.  1 The abbreviations used are: PLC, phospholipase C; PC, phosphatidylcholine; PC-PLC, phosphatidylcholine-hydrolyzing phospholipase C; TcdB, C. difficile toxin B; UDP-Glc, UDP-glucose; UDPG:PP, UDPglucose pyrophosphorylase; mAb, monoclonal antibody. H68S, and D56N), and fragments of C. perfringens PLC were expressed in Escherichia coli and purified as described previously (8,21,22). The production of the N-terminal C. bifermentans-C-terminal C. perfringens hybrid PLC will be published separately. 2 C. difficile TcdB and nonrecombinant C. perfringens PLC were purified as described (23)(24)(25). The B. cereus PC-PLC was from Calbiochem (San Diego, CA).
Protein Determination-Protein concentrations were determined in 96-well plates using the Bio-Rad DC protein assay and bovine serum albumin as a standard.
Cytotoxicity Assays-Cells were seeded in 96-well plates (500 to 1000 cells/well), cultivated to 90% confluency, and exposed at 37°C to serial 10-fold dilutions of the purified toxins: C. difficile TcdB (500 g/ml) or phospholipases C (recombinant C. bifermentans phospholipase C, B. cereus PC-PLC, wild type nonrecombinant or recombinant C. perfringens PLC) (214 g/ml) in 200 l of medium/well. Cell viability was measured 24 h later using a neutral red assay as described (16). All cytotoxicity assays were performed at least three times with 4 -6 replicate samples in each experiment. The sensitivity to C. perfringens PLC was also determined in the Don wt cell line after a 24-h pretreatment with 2-deoxyglucose (15 mg/ml), 2-deoxygalactose (8 mg/ml), or D-galactosamine (6 mg/ml).
Choline Incorporation into Membranes-Cell cultures were incubated with 0.4 Ci/ml [methyl-3 H]choline (Amersham Pharmacia Biotech) for 90 min. The cells were then washed three times with Hanks' balanced salt solution and incubated in fresh medium at 37°C. After 0.5, 2, 4, and 6 h of incubation, the cellular lipids were extracted with chloroform-methanol (26), and the choline incorporated into membranes was measured by counting the radioactivity in the lipid phase using a microtiter plate ␤-counter. The cpm were related to the amount of cellular protein determined in parallel cultures.
Stable Transfection of Bovine UDPG:PP to Don Q and QR Cells-The bovine UDPG:PP cDNA cloned in pUC18 (27), kindly provided by Professor K. Tanizawa (University of Osaka), was digested with EcoRI (Fermentas AB, Graiciuno, Lithuania). The 1.69-kilobase insert was separated in an agarose gel, cut out, eluted using the QIAEX kit (Qiagen, Hilden, Germany), and subcloned in pCDNA3 (Invitrogen Corp., Carlsbad, CA) at the EcoRI site. The ligation product was used to transform E. coli JM101 using a gene pulser electroporation instrument (Bio-Rad). After isolation of transformants, plasmids were digested with Acc651 and KpnI (Fermentas AB), and the fragments were separated in agarose gels to determine the orientation of the insert. The Chinese hamster Don Q and QR cell lines were transfected with the pCDNA3 vector containing the UDPG:PP insert in the sense and in the antisense orientation, respectively. Transfection was performed using Lipofectin (Life Technologies, Inc.) as described previously (28). Stable transfectant clones were obtained by minimal dilution and selected by cultivation in the presence of geneticin (500 g/ml).
Neutralization of C. perfringens PLC Cytotoxicity -The monoclonal antibodies (mAbs) 3A4D10 and 3A4F2 against the holotoxin were produced and purified as described previously (29). Polyclonal antibodies against the recombinant C-terminal domain of C. perfringens PLC were generated as described (30). The capacity of these antibodies to neutralize the cytotoxic effect of the recombinant C. perfringens PLC was determined by preincubation of this toxin (0.12 g/ml) with serial 2-fold dilutions of the antibodies (initial concentration 1 mg/ml) for 1 h at 37°C. Then 200 l of the mixtures were added to cells and incubated for 24 h, the cells were washed twice with phosphate-buffered saline, and cell viability was determined using the neutral red assay as described (16).

Don Q Is Hypersensitive
Specifically to the Cytotoxic Effect of C. perfringens PLC-As previously shown, Don Q and wt cells exhibit the same sensitivity to phospholipases A 2 , B, and D, whereas Don Q is hypersensitive to the cytotoxic effect of C. perfringens PLC (13). To determine whether Don Q is hypersensitive to all PC-hydrolyzing phospholipases C, the sensitivity of Don wt, Q, and QR to two other phospholipases C was studied. The B. cereus PLC showed the same cytotoxicity to the three cell lines (Fig. 1A) whereas the C. bifermentans phospholipase C was not cytotoxic to any of the cells (data not shown). Thus, Don Q seems to show a specific hypersensitivity to C. perfringens PLC, a hypersensitivity that has been lost in Don QR (Fig. 1B).
The C. perfringens PLC Hypersensitivity of Don Q Is Not Due to a Decreased Synthesis of PC-Hypersensitivity to C. perfringens PLC has previously been found in a mutant cell with a lowered capacity to synthesize PC (31)(32)(33). To determine whether the C. perfringens PLC hypersensitivity of Don Q is caused by a defective synthesis of PC, we measured the incorporation of [ 3 H]choline into the membranes of Don wt, Q, and QR (Fig. 2). Since no significant differences in the [ 3 H]choline incorporation were detected among the three cell lines, the hypersensitivity of Don Q to C. perfringens PLC does not seem to be due to a defective PC synthesis.
The C. perfringens PLC Hypersensitivity Is Due to Cellular UDP-Glc Deficiency-Since Don QR has partially compensated the UDP-Glc deficiency (16), the results in Fig. 1B suggested that a hypersensitivity to C. perfringens PLC is linked to the low level of UDP-Glc. Accordingly, it was observed that Don wt cells had an increased sensitivity to C. perfringens PLC after treatment with sugar analogues (Fig. 3) known to cause a cellular UDP-Glc deficiency by trapping the uridine nucleotide pool (34,35).
To conclusively clarify the link between UDP-Glc deficiency and C. perfringens PLC hypersensitivity, stable transfectant cells with bovine UDPG:PP cDNAs were prepared. Don Q cells were transfected with a construct having the UDPG:PP insert in the sense orientation and Don QR cells with a construct having the insert in the antisense orientation. Transfectant clones were designated Q Sense (QS) and QR Antisense (QRA), whereas clones transfected with only the vector were designated Q Control (QC) and QR Control (QRC), respectively. Since the resistance of Don Q to TcdB is due to its UDP-Glc deficiency (15) and this resistance has been lost in Don QR (16), we used the sensitivity to TcdB as an indirect measure of the cellular UDP-Glc level in the transfectant clones (Fig. 4). Don QRA clones were resistant to TcdB, showing that transfection with the antisense construct indeed had induced a UDP-Glc deficiency. Correspondingly, Don QS clones were more sensitive to TcdB than QC clones, demonstrating that the UDP-Glc deficiency indeed had been compensated by the transfection. This conclusion has been further substantiated by 1 H NMR measurement of the UDP-Glc concentration in lysates of the transfectant cells. 3 The sensitivity of the transfectant clones to C. perfringens PLC was then measured (Fig. 5). Don QRA clones were as hypersensitive to the cytotoxic effect of this toxin as Don Q. Correspondingly, Don QS clones had regained the same relative resistance to C. perfringens PLC as displayed by Don wt and QR. These results therefore demonstrate that a low level of UDP-Glc really causes hypersensitivity to the cytotoxic effect of C. perfringens PLC.
The Sphingomyelinase Activity and the C-domain Are Required for the Cytotoxicity of C. perfringens PLC-Using Don Q as a model system, we undertook a mapping of which parts of the C. perfringens PLC are needed for its cytotoxic effect. Previous studies with mAbs demonstrated that mAb 3A4D10 neutralized the enzymatic activities of C. perfringens, whereas mAb 3A4F2 did not (29). The cytotoxic effect of C. perfringens PLC in Don Q was abolished by mAb 3A4D10 but not affected by mAb 3A4F2 (Fig. 6A), suggesting that the catalytic capacity of C. perfringens PLC is required for its cytotoxic effect on UDP-Glc-deficient cells. This was further substantiated by studies with enzymatically inactive C. perfringens PLC variants. Replacement of His-11 or His-68 by Ser or Asp-56 by Asn resulted in a loss of the enzymatic activities (7-9). All these three C. perfringens PLC variants were 10 4 -10 5 times less cytotoxic to Don Q than the recombinant wild type PLC (Fig.  6B). These results therefore demonstrated that the catalytic capacity of C. perfringens PLC is required for its cytotoxic effect on UDP-Glc-deficient cells.
A truncated version of the C. perfringens PLC containing the 249 N-terminal residues retains the lecithinase activity but lacks the sphingomyelinase activity (10). This truncated toxin had a cytotoxic potency 10 6 times lower than the wild type enzyme (Fig. 7A). This result therefore demonstrated that the cytotoxic effect of the C. perfringens PLC requires the C-terminal domain and the sphingomyelinase activity. The data also showed that the lecithinase activity is not enough to confer cytotoxicity.
The C-terminal domain, containing residues 247-370, is devoid of both enzymatic activities of the holotoxin (30). However, polyclonal antibodies against the C-domain of C. perfringens PLC prevented the cytotoxic effect of the holotoxin in Don Q cells in a dose-dependent manner (Fig. 7B), demonstrating that this domain plays an important role in the cytotoxicity exerted by this PLC. In addition, the cytotoxic effect was potentiated about 10 2 times when the truncated PLC and the C-terminal domain where added simultaneously (Fig. 7A). However, the C. perfringens C-terminal domain did not potentiate the cytotoxic effect of the B. cereus PC-PLC (data not shown), excluding the possibility that a simple complementation of the lecithinase with the C-terminal domain would suffice to confer full cytotoxicity. Accordingly, when the C-terminal domain of C. perfringens PLC was fused with the N-terminal domain of C. bifermentans phospholipase C, this recombinant enzyme (designated hybrid D) showed a cytotoxic potency 10 3 times lower , and QRA2B2 (‚)) and their control (QRC (q)) (A) and clones of Don Q transfected with the UDPG:PP sense cDNA (designated QSG3 (E), QS20 (Ⅺ), and QSB9 (OE)) and their control (QC (q)) (B) were exposed for 24 h at 37°C to serial 10-fold dilutions of C. difficile TcdB. Cell viability was determined using the neutral red assay as described under "Experimental Procedures." The results are expressed as the percentage of neutral red uptake in controls incubated without toxin.
than that of C. perfringens PLC (Fig. 7C). These results indicate that a specific interaction between the two domains of the C. perfringens PLC is required for full cytotoxicity. DISCUSSION C. perfringens PLC is hemolytic, necrotizing, and lethal, in contrast to most other bacterial phospholipases (3) and is considered the major virulence factor in trauma-induced gas gangrene (2,36,37). This disease is of increasing significance in elderly and diabetic populations, especially in those who have undergone lower limb surgery, where impaired blood supply to tissues can lead to hypoxic conditions suitable for multiplication of anaerobic bacteria (38 -40). C. perfringens PLC induces localized intravascular neutrophil accumulation and platelet aggregation, thus reducing the blood supply to tissues and further promoting the anaerobic environment required for spread of the bacteria (36,41).
A UDP-Glc deficiency has been reported in the skeletal muscle of diabetic animals (42)(43)(44). This deficiency could be a consequence of the defective transport and/or phosphorylation of glucose in cells from diabetic subjects (45)(46)(47). A UDP-Glc deficiency also occurs in cells cultured under low glucose concentration or hypoxia (17)(18)(19)(20), two of the hallmarks of ischemia.
The isolation of the UDP-Glc-deficient mutant cell line Don Q and its intriguing hypersensitivity to C. perfringens PLC were previously reported (13,15). The reason for the low UDP-Glc level in this cell is a mutation in the UDPG:PP gene (16). In the present work we show that Don Q is specifically hypersensitive to C. perfringens PLC and not to other phospholipases C that degrade phosphatidylcholine. Since the revertant cell line Don QR has partially compensated the UDP-Glc deficiency and lost the C. perfringens PLC hypersensitivity, a link between a low UDP-Glc level and an increased sensitivity to C. perfrin-gens PLC was hypothesized. Such a link was further supported by the finding that cells treated with sugar analogues, which induce UDP-Glc deficiency (34,35), were more susceptible to the cytotoxic effect of C. perfringens PLC. Transfection of mutant and revertant cells with the bovine UDPG:PP cDNA in sense and antisense orientations affected their UDP-Glc levels and indeed changed their susceptibility to C. perfringens PLC. Therefore, these results conclusively demonstrate that a cellular UDP-Glc deficiency causes hypersensitivity to C. perfringens PLC.
Although a detailed three-dimensional structure of the C. perfringens PLC is not yet available, two separate domains have been identified: the N-terminal domain, which catalyzes phospholipid hydrolysis, and the C-terminal domain, which mediates interactions with membrane phospholipids (4,5,12). Ca 2ϩ is required for the interaction of the toxin with membranes, whereas Zn 2ϩ plays a structural role. Crystallographic studies have revealed that the B. cereus PC-PLC has two tightly bound and one loosely bound Zn 2ϩ ions, which are required for the enzymatic activity (48). These Zn 2ϩ ions are coordinated to several residues conserved in the C. perfringens PLC, which suggests that these two enzymes have the same catalytic mechanism. It has been previously shown by sitedirected mutagenesis that the substitution of His-11, His-68, His-126, or His-136 in the C. perfringens PLC dramatically reduces its lecithinase and sphingomyelinase activities (7,8), whereas their significance for cytotoxicity has not been studied. These His residues are very likely involved in the coordination of the loosely bound Zn 2ϩ ion, which is required for C. perfringens PLC binding to membranes (7,8). On the other hand, Asp-56, which is exposed on the surface within the active site cleft of the B. cereus PC-PLC, is also conserved in the C. (‚)) and their control (QRC (q)) (A), and clones of Don Q transfected with the UDPG:PP sense cDNA (designated QSG3 (E), QS20 (Ⅺ), and QSB9 (OE)) and their control (QC (q)) (B) were exposed for 24 h at 37°C to serial 10-fold dilutions of C. perfringens PLC. Cell viability was determined using the neutral red assay as described under "Experimental Procedures." The results are expressed as the percentage of neutral red uptake in controls incubated without toxin. perfringens enzyme. Replacement of Asp-56 in the latter enzyme abolishes its lecithinase and sphingomyelinase activities without affecting its membrane-binding capacity (8,9).
Neutralization experiments with mAbs and polyclonal antibodies suggested that both the catalytic capacity and the C domain are required for the cytotoxic effect of the C. perfringens PLC. The enzymatically inactive C. perfringens PLC variants H11S, H68S, and D56N and the truncated enzyme having only the first 249 residues had a markedly reduced cytotoxicity. The cytotoxic effect was partially restored when the truncated phospholipase and the C-terminal domain were added simultaneously. In contrast there was not any potentiation of the cytotoxic effect when the B. cereus PC-PLC and the C. perfringens C-terminal domain where added together. Moreover, the hybrid PLC, having the C. perfringens C-terminal domain and the C. bifermentans N-terminal domain, displayed a cytotoxic potency much lower than that of C. perfringens PLC. These results showed that the lecithinase activity is not enough to confer cytotoxic activity to C. perfringens PLC and suggested that a specific interaction between the two toxin domains is needed for full cytotoxicity. In addition, the data demonstrated that both the sphingomyelinase activity and the membraneinteracting C-terminal domain are crucial for the cytotoxicity of C. perfringens PLC to UDP-Glc-deficient cells.
Treatment of cultured cells with C. perfringens PLC has been shown to activate endogenous phospholipases A 2 and C, hence generating messengers such as leukotrienes, diacylglycerol, and inositol 1,4,5-triphosphate (49 -52). Therefore we have considered the possibility that the cytotoxicity could be mediated by the activation of those cellular enzymes. However, we have not been able to protect Don Q cells from the cytotoxic effect of C. perfringens PLC by pretreatment with (i) inhibitors of the arachidonic acid pathway (mepacrine, indomethacin, nordihydroguaiaretic acid), (ii) inhibitors of cellular phospholipases C or D (D-609, ethanol, neomycin), or (iii) the diacylglycerol lipase inhibitor U-57908 (1,6-bis(cyclohexyloximinocarbonylamino)hexane). 4 Therefore, the cytotoxicity of C. perfringens PLC on UDP-Glc-deficient cells seems to be independent of the activation of endogenous phospholipases. The interaction between this toxin and the plasma membrane in UDP-Glc-deficient cells may be favored by a putative alteration in the composition of the membrane glycolipids or glycoproteins. The C. perfringens PLC induces activation of protein kinase C and stimulates Ca 2ϩ influx into the cytosol, hence modulating a variety of cell functions (4,50). The role of these effects in the cytotoxicity of the C. perfringens PLC are currently being investigated in our laboratories. Regardless of the molecular mechanism of cytotoxicity, our results demonstrate that a UDP-Glc deficiency sensitizes cells to the cytotoxic effect of C. perfringens PLC. In addition the data raise the possibility that the low UDP-Glc level occurring in diabetic tissues and cells under ischemic conditions predisposes to the extensive tissue damage induced by C. perfringens PLC during gas gangrene.
Concluding Remarks-This work demonstrated that cells with a low UDP-Glc level are hypersensitive to the cytotoxic effect of C. perfringens PLC. Using genetically engineered variants of this phospholipase, we showed that the C-terminal domain and the sphingomyelinase activity of C. perfringens PLC are required for its cytotoxic effect. Since lowered levels of UDP-Glc have been reported to occur in tissues of diabetic animals and in cells grown under low glucose concentration or hypoxia, we hypothesize that a UDP-Glc deficiency in the target tissues plays a role in the pathogenesis of gas gangrene. Furthermore we suggest that the Don Q cell line can be used as a model to elucidate the molecular mechanism of cytotoxicity induced by C. perfringens PLC on ischemic tissues.