A Direct Role for Secretory Phospholipase A₂ and Lyso-Phosphatidylcholine in the Mediation of Lipopolysaccharide-induced Gastric Injury

Abstract

Background & Aims

Endotoxemia from sepsis can injure the GI tract through mechanisms that have not been fully elucidated. We have shown that lipopolysaccharide (LPS) induces an increase in gastric permeability in parallel with the luminal appearance of secretory phospholipase A₂ (sPLA₂) and its product, lyso-phosphatidylcholine (lyso-PC). We proposed that sPLA₂ acted on the gastric hydrophobic barrier, composed primarily of PC, to degrade it and produce lyso-PC, an agent that is damaging to the mucosa. In the present study we have tested whether lyso-PC and/or sPLA₂ have direct damaging effects on the hydrophobic barriers of synthetic and mucosal surfaces.

Methods

Rats were administered LPS (5 mg/kg, ip), and gastric contents were collected 5 h later for analysis of sPLA₂ and lyso-PC content. Using these measured concentrations, direct effects of sPLA₂ and lyso-PC were determined on: 1) surface hydrophobicity as detected with an artificial PC surface and with intact gastric mucosa (contact angle analysis); and 2) cell membrane disruption of gastric epithelial cells (AGS).

Results

Both lyso-PC and sPLA₂ increased significantly in the collected gastric juice of LPS-treated rats. Using similar concentrations to the levels in gastric juice, the contact angle of PC-coated slides declined after incubation with either pancreatic sPLA₂ or lyso-PC. Similarly, gastric contact angles seen in control rats were significantly decreased in sPLA₂ and lyso-PC treated rats. Additionally, we observed dose-dependent injurious effects of both lyso-PC and sPLA₂ in gastric AGS cells.

Conclusions

An LPS-induced increase in sPLA₂ activity in the gastric lumen, and its product lyso-PC, are capable of directly disrupting the gastric hydrophobic layer and may contribute to gastric barrier disruption and subsequent inflammation.

INTRODUCTION

Acute endotoxemia is associated with shock and injury to the gastrointestinal (GI) tract. The occurrence of ileus due to endotoxic shock is accompanied by production of inflammatory mediators such as TNFα and other cytokines or chemokines that are thought to be responsible for much of the injury (1–3). However, other mechanisms could also be a part of the process. The use of bacterial lipopolysaccharide (LPS) in rodents as a model of endotoxemia has allowed a closer examination of other factors that may be involved in this shock-induced GI injury and dysfunction.

LPS has been shown to inhibit gastric acid secretion and also produce a gastroparesis, allowing the stomach to fill with an alkaline fluid (4–5). In addition, our group has reported that LPS induces a duodenogastric reflux of intestinal contents, including bile acid, into the gastric lumen, which may also contribute to gastric alkalinization (6). We have further shown that LPS induces an increase in gastric permeability to a fluorescent probe of substantial size (4000 molecular weight), without overt disruption to the mucosal architecture (7). Concurrently, there is an increase in the gastric luminal appearance of secretory phospholipase A₂ (sPLA₂), as well as the product of this enzyme, lyso-phosphatidylcholine (lyso-PC). The source of this gastric luminal sPLA₂ is unclear, but at least part of it is likely to have derived from intestinal reflux material of pancreatic origins. Regardless of its source, sPLA₂ within the gastric lumen, at an appropriate pH, may be capable of degrading its primary substrate, phosphatidylcholine (PC).

Our group and several others have shown that a portion of the gastric barrier to acid is due to the presence of a hydrophobic layer of phospholipid, mainly PC, which overlies the mucus gel layer on the mucosa (8–10). Disruption of the hydrophobic layer is associated with injury and disease states such as ulcer disease (11–12), colitis (13–14) and NSAID toxicity (15–16). LPS was shown to reduce surface hydrophobicity in a time and dose-dependent manner (6). We have proposed that LPS acts indirectly on this barrier by increasing bile acid content and sPLA₂ activity in the gastric lumen, and these agents are capable of degrading the PC layer. In the current paper, we have focused on sPLA₂ activity and how it may act to disrupt the hydrophobic barrier.

METHODS

Animals

Male Sprague-Dawley rats (175–250 g) were obtained from Harlan Sprague-Dawley (Indianapolis, IN). All animals were fasted overnight before use to ensure an empty stomach. Institutional Animal Welfare Committee approval was obtained prior to all studies.

LPS effect on gastric contents

Saline (control) or LPS (5 mg/kg, ip) were administered to rats and gastric contents were collected five hours later after euthanasia. The determination of sPLA₂ activity, as well as lipid extraction and thin layer chromatography to separate lyso-PC was performed on the gastric fluid as previously described (7).

PC coated slides

Well-cleaned glass slides were coated with a monolayer of PC (dipalmitoyl phosphatidylcholine) using a Langmuir trough and a modification of the Blodgett and Gaines techniques as previously described (17). The slides were then incubated in normal saline or saline containing various test agents, including pancreatic sPLA₂ (10 units) or lyso-PC (0.1, 0.5, 1, or 5 mM) at 37 °C for two hours with gentle stirring. The slides were rinsed, air-dried and contact angle analysis (17) was performed with a Rame-Hart goniometer (Hicksville, NY).

Effect of sPLA₂ and lyso-PC on gastric hydrophobicity

Rats were anesthetized with isoflurane and underwent a pylorus ligation. The animals were then dosed by oral gavage with 1 ml of either vehicle, sPLA₂ (10 units), or lyso-PC (2 mM). The agents were dissolved in 100 mM HEPES buffer at pH 7.4 to maintain gastric pH above 6.5 for the duration of the experiment. After 60 minutes, the rats were euthanized and gastric tissue was collected for contact angle analysis as previously described (17).

Cell culture

AGS cells are human gastric epithelial cells originally derived from an adenocarcinoma (18). They were maintained in Ham's F12K media plus 10% fetal bovine serum and penicillin/streptomycin. Cells were plated into 24-well culture plates at 2×10⁵ cells per well, and were exposed to test agents lyso-PC (0.01 to 0.2 mM for 3h) or bovine pancreatic PLA₂ (0.1 to 10 U/ml for 24h), after which the media was collected for lactate dehydrogenase assay (LDH) as a measure of cell membrane injury and the cell number was analyzed by the thiazolyl blue tetrazolium bromide (MTT) assay as previously described (19). For comparison to other forms of phospholipase, sPLA₂ from bee venom and snake venom (Naja naja), and phospholipase C (PLC) from Clostridium perfringens were also tested. For comparison to another epithelial cell type, selected phospholipase studies were performed on IEC-6 cells which are derived from normal rat intestinal crypt epithelial cells (20). IEC-6 cells were maintained in Dulbecco's modified Eagle medium plus 10% fetal bovine serum, penicillin/streptomycin, and insulin (0.1 units/ml).

Materials

LPS (E. coli 0111:B4), all phospholipases, all lipids and a kit (TOX-7) for LDH assay were obtained from Sigma Chemical Co. (St. Louis, MO). The PC used in the monolayer slide studies was dipalmitoyl phosphatidylcholine. AGS cells were obtained from the Texas Medical Center Digestive Diseases Center (Houston, TX). IEC-6 cells were obtained from The University of Texas Trauma Research Center (Houston, TX).

Statistics

Differences among treatment groups were analyzed by analysis of variance and the Fisher LSD test. The Student t-test was used to compare differences between two groups. The level for significance was set at p<0.05.

RESULTS

Lyso-PC and sPLA₂ after LPS

The gastric contents of control rats contained small, but detectable amounts of lyso-PC measured at 0.08 ± 0.02 mM. At five hours after the systemic administration of LPS at a dose of 5 mg/kg, that value increased significantly at least 60-fold to 1.29 ± 0.17 mM (p<0.001). Similarly, the gastric contents of control animals had a total of 0.6 ± 0.06 units of sPLA₂ activity that increased ~5 fold to 3.2 ± 0.56 units in LPS-treated rats (p<0.05). A unit of sPLA₂ activity is the amount that will hydrolyze 1.0 μmole of PC to lyso-PC producing a free fatty acid per minute at pH 9 and 37°C.

PC coated slides

The mean contact angle for PC-coated slides incubated for 2 h in saline was 58° (n=9). PC coated slides incubated in either 10 units/ml of pancreatic sPLA₂ or 0.1 to 5 mM lyso-PC showed significant reductions in contact angles, as shown in Figure 1.

Figure 1.

Effect of sPLA₂ and lyso-PC in vitro. PC coated slides were incubated for 2 h with sPLA₂ (10 U/ml) or lyso-PC (0.1 to 5 mM), followed by contact angle readings. Values are expressed as the mean contact angle in degrees ± standard deviation for N = 4/group. * p<0.05 versus Control.

Effect of sPLA₂ and lyso-PC on rat gastric hydrophobicity

Separate in vivo studies were performed with each test agent as seen in Figure 2A and B. In the study testing sPLA₂, the treated rats showed a significant 7% reduction in gastric hydrophobicity over control animals (p<0.05). In the study testing lyso-PC, treated rats had a significant 18% reduction in gastric hydrophobicity over control values (p<0.02).

Figure 2.

Effect of sPLA₂ and lyso-PC in vivo. Rat stomachs were exposed to: A) pancreatic sPLA₂ or B) lyso-PC, for 1 h and the gastric contact angles were then measured. Values are expressed as the mean contact angle in degrees ± standard deviation for N = 4–6/group. * p<0.05 versus Control.

Effect of lyso-PC and sPLA₂ on cultured cells

Gastric cells incubated for 3 h with lyso-PC showed a dose-dependent increase in LDH release, indicative of membrane injury (Figure 3). At the same time, the number of viable cells, as detected by the MTT assay, was correspondingly decreased. The ED50 for both measures was estimated to be about 30 to 40 μM.

Figure 3.

Effect of lyso-PC on gastric AGS cells. Cells were incubated for 3 h with varying concentrations of lyso-PC. LDH was measured in the media and MTT was assayed in the cells. Values are expressed as the mean optical density (OD) of the assay ± standard error of the mean. The study was performed three times. * p<0.05 versus Control.

When AGS cells were initially incubated with pancreatic sPLA₂ for 3 h, there were no observable changes in cell number, so the incubation time was lengthened to 24 h. At 24 h (Figure 4), it was found that LDH was not increased by sPLA₂ treatment, while MTT indicated a slight, but significant, dose-dependent sPLA₂-induced reduction in cell numbers. It should be noted that all cell assays of LDH activity were accompanied by a positive control (cells treated with deoxycholate for complete cell lysis) to verify that the assay was working properly.

Figure 4.

Effect of sPLA₂ on gastric AGS cells. Cells were incubated for 24 h with varying concentrations of pancreatic sPLA₂. LDH was measured in the media and MTT was assayed in the cells. Values are expressed as the mean OD of the assay ± standard error of the mean. The study was performed three times. * p<0.05 versus Control.

To determine the specificity of the sPLA₂ effect, two other types of sPLA₂ were tested on AGS cells. Neither bee venom nor snake venom sPLA₂ (Table 1) had an effect on LDH or MTT of AGS cells after a 24 h incubation. In contrast, phospholipase C (PLC) was clearly injurious to the cells at ≥0.1 U/ml. Another epithelial cell type, IEC-6 cells, showed a similar lack of sensitivity to the sPLA₂ enzymes (Table 1), and were slightly responsive to pancreatic sPLA₂ (10 U/ml = 92 ± 1% of control for MTT (not significant) and 150 ± 4% of control for LDH (p<0.05 vs control)). IEC-6 cells were equally injured by PLC as the AGS cells.

Table 1.

Phospholipase-induced injury to gastrointestinal cells.

	AGS cells		IEC-6 cells
Phospholipase	LDH	MTT	LDH	MTT
Bee venom PLA2
0.1 U/ml	100 ± 3	106 ± 2	104 ± 3	111 ± 2
1 U/ml	107 ± 3	113 ± 2	97 ± 8	115 ± 3
10 U/ml	100 ± 1	106 ± 2	100 ± 2	111 ± 1
Snake venom PLA2
0.1 U/ml	98 ± 1	97 ± 3	100 ± 3	109 ± 5
1 U/ml	102 ± 2	101 ± 2	95 ± 10	121 ± 7
10 U/ml	102 ± 5	111 ± 2	140 ± 26	94 ± 7
PLC
0.1 U/ml	149 ± 5*	79 ± 3*	292 ± 20*	69 ± 1
1 U/ml	423 ± 5*	12 ± 0.1*	464 ± 2*	17 ± 1*
10 U/ml	446 ± 44*	12 ± 0.1*	473 ± 7*	15 ± 0.1*

AGS or IEC-6 cells were incubated for 24 h with 0.1, 1 or 10 U/ml of the indicated phospholipases. LDH and MTT were measured as described in the Methods. Values are expressed as the percent of control ± standard error of the mean for four separate studies.

^*= p<0.05 versus Control

DISCUSSION

Using a combination of in vitro and in vivo techniques, we have shown in the current studies that LPS induces the appearance of sPLA₂ and lyso-PC in the gastric lumen, and these factors are capable of directly disrupting the gastric hydrophobic barrier.

The dose of LPS is one that we have used previously (6). Internally, the 5 mg/kg dose of LPS induces gastroparesis and accumulation of alkaline fluid within the stomach. At five hours after dosing with LPS, the accumulated load of lyso-PC within the gastric lumen was clearly elevated by at least 60-fold and measured about 1.3 mM. Similarly, we measured the accumulated level of sPLA₂ activity and found an average total of about 3.2 units per stomach in LPS-treated rats, which was 5-fold higher than observed in control stomachs.

Concentrations of lyso-PC and sPLA₂ similar to those obtained experimentally in rats were then tested in an in vitro monolayer system. Slides that had been coated with PC to mimic the stomach's hydrophobic barrier were exposed to test agents for only two hours, in comparison to the five hours of the in vivo study. This time was a conservative estimate since in the in vivo study, the rat stomach was exposed to an increasing amount of agent over a five hour period. We have previously shown that both sPLA₂ and lyso-PC are beginning to be elevated intraluminally in the stomach by 3 h after a dose of LPS, and this amount is further increased by 5 h (7). Both sPLA₂ and lyso-PC reduced the slide contact angle in confirmation of their ability to degrade or solubilize PC, respectively.

To verify that sPLA₂ and lyso-PC were capable of causing a loss of gastric barrier hydrophobicity in an animal, these agents were incubated in a closed stomach preparation at levels used in vitro and measured in vivo after LPS. In fact, these in vivo data on lyso-PC agree with numerous reports on the toxicity of lyso-PC in the stomach (21–24) and the concentrations which induce injury (1–10 mM). In addition, we were careful to maintain the intragastric pH with buffer at no lower than pH 6.8 in these studies, as that is similar to the gastric pH measured following LPS administration (38), and it is also needed for enzymatic activity of sPLA₂, which is most active in an alkaline range (pH 8–9). At pH 6–7, sPLA₂ will still be active. Both agents caused a significant reduction in surface hydrophobicity, confirming their in vivo activity. The extent of contact angle reduction in these studies was relatively small (7% and 18%), compared to other animal studies where reported reductions due to disease (25) or chemical insults (26–28) could be 50% or more. However, the present studies were performed in pylorus-ligated rats where a single agent at a physiologically relevant concentration was exposed to the gastric surface. In our previous studies involving contact angle measurements, the studies were performed in conscious, intact animals and were the net result of multiple factors such as acid, bile/enzyme reflux, motility dysfunction, or high dose of damaging agent, as opposed to the single agent tested here. It also should be noted that our relative reductions in contact angle in the current study are comparable to changes seen in clinical studies involving gastritis or infection with the pathogen Helicobacter pylori where reductions of 8 to 19% were consistently reported (29–30). We have also shown previously in isolated canine gastric mucosa that there is a strong relationship between extent of reduction in gastric contact angle and reduction in electrical potential difference which is inversely related to gastric injury (26). Thus it is reasonable to conclude from our data that these individual agents (sPLA₂ and lyso-PC) have the ability to directly affect a meaningful reduction in contact angle.

Finally, cultured gastric cells were exposed to concentrations of lyso-PC and sPLA₂ at or below those that had been used in the studies above. The effect of lyso-PC on AGS cells was to induce cell membrane damage and release the cytosolic enzyme, LDH. Results with the MTT assay were consistent with that interpretation of loss of cell number due to injury. It was further noted that the gastric cells were quite sensitive to lyso-PC, with toxicity found at 10 to100-fold lower concentrations than were observed to be injurious in intact animals or the other in vitro system. Clearly, lyso-PC is particularly damaging to gastric epithelial cells, and the levels of lyso-PC that were detected in vivo after LPS treatment are capable of directly disrupting both the extracellular hydrophobic layer and mucosal epithelial cells.

In contrast to lyso-PC, the inability of pancreatic sPLA₂ to induce cell membrane damage, as seen by no LDH release, was unexpected. One explanation for the lack of an injurious effect may be related to the absence or more limited nature of an extracellular hydrophobic PC layer on these cultured cells, and the related inability of extracellular sPLA₂ to act on the extracellular PC to produce lyso-PC. In that regard, we attempted to add extracellular PC to cells incubated with sPLA₂ in order to promote lyso-PC formation, but no injury was detected (not shown). It is likely that any PC added to this cell system will be protective, as we have shown for other cell cultures (31). Yet we would have expected that sPLA₂ would attack membrane PC and it apparently did not. There was a small effect of sPLA₂ to reduce numbers of AGS cells as seen by a decreased MTT assay, an effect that was apparently independent of cell membrane injury. To investigate whether the pancreatic sPLA₂ effect was peculiar to AGS cells, we also tested another epithelial cell line, that of IEC-6 cells. This culture, too, was unresponsive by LDH and also MTT assay, to the sPLA₂, supporting the possibility of a general effect. Although we cannot locate any references that report on effects of pancreatic sPLA₂ on AGS or other human gastric cells, there is one report of damaging actions of sPLA₂ on primary isolated rat gastric cells (32) in which bee venom PLA₂ was used. This latter sPLA₂ has similar substrate preferences to pancreatic sPLA₂, but is associated with a different gene (Pla2 versus PLA2G1B) and has a different pH optimum. Those investigators found injury after short-term incubation of bee venom PLA₂ with rat cells and also found a similar injurious sensitivity to lyso-PC as us (10–100 μM). Whether a difference in cell isolation or species of origin may explain these differences in sensitivity to sPLA₂ is not known. Another explanation for our lack of effect with sPLA₂ is that, in contrast to other types of sPLA₂, the pancreatic form (1B) which we used is reported to also act through specific receptors (33–37), although its physiological role is unclear. Thus, there may be non-enzymatic actions of pancreatic sPLA₂ on the gastric mucosa that explain our results.

To investigate the specificity of the pancreatic sPLA₂ effect, we tested sPLA₂ from bee venom and snake venom. Neither of these latter sPLA₂s were active against AGS or IEC-6 cells by the LDH or MTT assays. This would support the uniqueness of the limited pancreatic sPLA₂ effect on gastric cells. In addition, PLC clearly had a toxic effect on the AGS and IEC-6 cells, as would be expected from this membrane surface-acting enzyme. Further studies will be necessary to fully elucidate the effects of pancreatic sPLA₂ on AGS cells.

Regardless of the direct effects of sPLA₂ on gastric cells, our other results are fully consistent with the ability of sPLA₂ to degrade the extracellular protective PC surface layer, and along with lyso-PC, to disrupt the PC layer, as well as cell membranes. Thus, circumstances under which sPLA₂ appears in the gastric lumen, such as during endotoxic shock produced here or other duodenogastric reflux-inducing situations, can lead to disruption of the protective PC barrier and exposure of the underlying epithelium to potentially injurious luminal contents such as lyso-PC. Methods to counter the loss of surface PC, such as addition of exogenous PC which we have previously shown to be effective at preventing an increase in LPS-induced gastric permeability (38), offer a means to prevent local, as well as systemic inflammatory conditions. We hypothesize that the additional PC may form non-injurious liposomes with lyso-PC, lessening the free lyso-PC concentration.

Our preclinical findings reported here have potential clinical relevance. We have recently reported that the gastric fluid of patients following a traumatic shock is alkaline and contains intestinal reflux material such as bile acid (39). Other critically ill patients have also been reported to acquire bile and reflux material in their stomachs (40–41). Thus, among critically ill patients, there are circumstances where bile acids and sPLA₂ from the intestines may be present in the stomach and could contribute to gastric barrier disruption by PC degradation. Based on our findings in rats reported here, we speculate that some of the altered GI permeability in critically ill patients may be due to exposure of the gastric lumen to sPLA₂ and appearance of the potent damaging agent lyso-PC. Future clinical studies are warranted to address this possibility.