- Email: [email protected]
Evidence for platelet-activating factor as a mediator of endotoxin-induced gastrointestinal damage in the rat
GASTROENTEROLOGY
1987;93:765-73
Evidence for Platelet-Activating Factor as a Mediator of Endotoxin-Induced Gastrointestinal Damage in the Rat Effects of Three Platelet-Activating Antagonists
Factor
JOHN L. WALLACE, GRAHAM STEEL, BRENDAN J.R. WHITTLE, VINCENT LAGENTE, and BORIS VARGAFTIG Department of Mediator Pharmacology, United Kingdom; and Unite des venins, France
Wellcome Research Laboratories, Beckenham, Kent, Unite associee Institut Pasteur/INSERM U 285, Paris,
The potential role of platelet-activating factor (PAF) as a mediator of gastrointestinal ulceration associated with septic shock was examined in the rat. The damaging effects of both PAF and Escherichia coli endotoxin in the stomach and small intestine were compared, as were their effects on plasma leakage into the lumen of the gastrointestinal tract. Intravenous administration of either endotoxin or PAF produced extensive necrosis and vascular congestion in the stomach and small intestine, but not the distal colon. With either agent, the duodenum and jejunum were the tissues most susceptible to damage and in which the greatest plasma leakage was observed. The prolonged hypotension and gastrointestinal damage induced by PAF or endotoxin were significantly inhibited by three structurally dissimilar PAF antagonists (CV-3988, BN-52021, and Ro-193704). CV-3988 [lo mg/kg) significantly (p < 0.05) reduced both endotoxinand PAF-induced plasma leakage in the stomach and small intestine. Of the three antagonists, only CV-3988 significantly reduced ethanol-induced gastric mucosal damage, perhaps reflecting actions of this compound unrelated to antagonism of PAF receptors. These studies support the hypothesis that PAF is an important Received August 22, 1986. Accepted April 1, 1987. Address requests for reprints to: Dr. J.L. Wallace, Department of Physiology, Botterell Hall, Queen’s University, Kingston, Ontario, K7L 3N6, Canada. Dr. Wallace is the recipient of a scholarship from the Medical Research Council of Canada. CV-3988 was a gift from Dr. M. Nishikawa, Takeda Chemical Industries, Ltd. BN-52021 was a gift from Dr. P. Braquet, Institut Henri Beaufour. Ro-193704 was a gift from Dr. H. Baumgartner, Hoffman LaRoche. 0 1987 by the American Gastroenterological Association 0016-5085/87/$3.50
mediator of the hypotension and plasma leakage observed during endotoxic shock and its endogenous release may contribute to the gastrointestinal ulceration associated with this syndrome. Thus, PAF receptor antagonists may be useful for prevention of such ulceration. Septic shock is a major risk factor for gastrointestinal ulceration (1) and accounts for as many as 100,000 deaths per year in the United States (2). The sequence of events following the appearance of invading bacteria in the systemic circulation is poorly understood, although the release of potent mediators such as prostaglandins, thromboxane AZ, and endorphins may account for some or all of the symptoms of septic shock (3-5). The observation that many of the features of septic shock, such as systemic hypotension, hemoconcentration, and gastrointestinal ulceration, could be mimicked by intravenous administration of a substance known as platelet-activating factor (PAF), led to the hypothesis that PAF may be an important mediator of septic shock (6,7). Platelet-activating factor is a lowmolecular-weight phospholipid characterized by an ether linkage at the 1’ positive (usually with a fatty acid chain of 16 or 18 carbons] and an acetyl group at the 2’ position of a phosphatidylcholine. Further evidence of a role for this molecule in septic shock was provided by the demonstration that intravenous administration of endotoxin to rats resulted in the release of PAF into the bloodstream (8). The cellular source of this PAF was not known, although neutroAbbreviations platelet-activating
used in this factor.
paper:
BP, blood
pressure;
PAF,
766 WALLACE ET AL.
GASTROENTEROLOGY Vol. 93, No. 4
phils, macrophages, monocytes, basophils, endothelial cells, mast cells, and platelets have all been shown to release PAF in vitro (9). A role for PAF as a mediator of gastrointestinal ulceration is supported by our observations that PAF induced gastric necrosis in the rat stomach at doses as low as XI pmol/kg (i’), whereas doses as low as 2 pmol/kg significantly predisposed the rat stomach to damage induced by topically applied 20% ethanol (10). In the present study, we have examined the role of PAF in an endotoxic shock model in the rat. In addition to comparing the gastrointestinal damage induced by PAF and endotoxin, we have studied the effects of three structurally unrelated PAF receptor antagonists in these models. If PAF mediates the vasocongestion and necrosis induced in the gastrointestinal tract by endotoxin, pretreatment with a PAF antagonist should confer protection. The three PAF antagonists used include two that are structural analogues of PAF ((X-3988 and Ro-193704) and one (BN-52021) have been
that
is not
onists
of PAF,
gation
induced
osine
diphosphate,
prevent
the
their receptors of these
(11-13).
demonstrated as they
These
to be highly failed
to block
by arachidonic and
binding
Materials
antag-
platelet
aggre-
collagen,
thrombin,
of several
(see Reference
compounds
acid,
compounds
specific
and their
other
and
adendid
not
mediators
to
9 for a detailed
review
specificity].
and Methods
Male Wistar rats (200-220 g) were deprived of food, but not water, for 20-24 h before an experiment. The animals were anesthetized by an intraperitoneal injection of sodium pentobarbitone (60 mg/kg) and a carotid artery was cannulated for measurement of systemic arterial blood pressure (BP). A 25-gauge needle was inserted into a tail vein for administration of PAF, endotoxin, or drugs.
Platelet-Activating Factor-Induced Gastrointestinal Damage A stable resting BP was obtained for at least 10 min before the vehicle or one of the PAF antagonists (BN-52021 at 10 mg/kg, CV-3988 at 10 mg/kg, or Ro-193704 at 1 mg/kg) was injected intravenously. W-3988 was administered as an infusion over 10 min, whereas the other PAF antagonists were given by bolus injection. Five minutes later, a lo-min intravenous infusion of PAF was started (100 ng/kg . min). This dose of PAF was selected because it has previously been shown to induce gastrointestinal damage (10,14) and because it should produce blood levels of PAF similar to those detected after endotoxin administration to rats (8). Twenty minutes after completion of the infusion of PAF, the rat was killed and the stomach and small bowel were examined and scored for damage as described previously (14), based on the severity of hyperemia and hemorrhage (O-3 scale). Samples of the fundic region of the stomach as well as
samples from standardized regions of the proximal duodenum, jejunum, ileum, and distal colon were fixed in neutral buffered formalin and processed by routine techniques before embedding in paraffin wax. Sections (4 pm) were stained with hematoxylin and eosin and were examined under a light microscope. The tissue sections were given a score on a 0-3 scale based on the following criteria: 0 = normal; 1 = subepithelial vasocongestion + patches of superficial necrosis; 2 = mucosal vasocongestion and focal necrosis; 3 = extensive vasocongestion and necrosis involving the full thickness of the mucosa. The overall damage score for the small intestine was taken as the mean scores for the sections of duodenum, jejunum, and ileum. All assessments of damage were performed in a blinded manner, to avoid observer bias. Endotoxin-Induced
Gastrointestinal
Damage
A stable resting BP was obtained for at least 10 min before the vehicle or one of the PAF antagonists (BN-52021 at 10 or 20 mg/kg, CV-3988 at 10 mg/kg, or Ro-193704 at 1 or 2 mgikg) was injected intravenously as described above. Ten minutes after administration of the vehicle or PAF antagonist, Escherichia coli endotoxin was administered as an intravenous bolus at a dose of 25 mg/kg. Twenty minutes later the rat was killed, and the stomach and small intestinal damage were scored as in the PAF experiments. Samples of stomach and small intestine were fixed and processed for light microscopy as above. Luminal
Plasma
Leakage
A series of experiments were performed to determine the amount of leakage of plasma into the lumen of the stomach and small intestine. Rats were treated with PAF or endotoxin, as described above, with some rats receiving prior intravenous administration of CV-3988 (10 mgikg). Five minutes before the injection of endotoxin or PAF, ‘ZSI-labeled human serum albumin (6.25 &i) was injected intravenously via the tail vein. Twenty minutes after endotoxin or PAF administration, the peritoneum was opened and the stomach and segments of the duodenum, jejunum, and ileum (-6 cm long) were clamped, so that luminal fluid could not leak, and subsequently excised. The segment of duodenum extended from the pylorus to the ligament of Treitz. The segment of ileum was taken proximal to the ileocecal valve. The stomach and segments of intestine were placed in individual glass vials containing 2.0 ml of 0.9% saline and were opened by a longitudinal incision to allow for mixing of luminal contents and saline. The tissues were then removed, blotted dry, and weighed. The samples of luminal fluid were mixed well, then 1.0 ml was transferred to a glass tube for quantification of radioactivity on a y-spectrometer. After the stomach and segments of gut had been taken from the rat, a l.O-ml blood sample was taken from the descending aorta and centrifuged for 1 min using an Eppendorf benchtop centrifuge (14,000 g). A loo-p1 samples of the plasma was transferred to a glass tube for estimation of radioactivity on a y-spectrometer. This yielded a value for counts per minute per microliter of plasma, which was then used to calculate the number of microliters of plasma
October 1987
PAFANDSHOCK-INDUCEDULCERATION
present in the luminal of tissue).
fluid samples
Ethanol-Induced
Gastric
(expressed
per gram
Damage
To determine if there were any protective actions of the PAF antagonists in the gastric mucosa that might be unrelated to their actions on PAF receptors, the effects of these compounds on ethanol-induced gastric damage were assessed. Rats that had been fasted for 20-24 h were anesthetized with sodium pentobarbitone (60mg/kg) and a 25-gauge needle was inserted into a tail vein. The peritoneum was opened and ligatures were placed around the esophagus and pylorus. Either the appropriate vehicle or one of the PAF antagonists was injected or infused intravenously as described above. For these experiments, only one dose of each antagonist was tested (CV-3988 at 10 mg/kg, BN-52021 at 10 mg/kg, or Ro-193704 at 1 mgikg). Ten minutes later, 3.0 ml of absolute ethanol was injected into the stomach through a 30-gauge needle inserted through the forestomach. The ethanol was left in the stomach for 10 min, after which time the stomach was removed, opened along the greater curvature, pinned out on a wax block, and photographed. The extent of macroscopically visible damage was quantified planimetrically as described previously (15). Materials Escherichia coli lipopolysaccharide (Olll:B4) was obtained from Difco Laboratories (Detroit, Mich.) and was prepared as a suspension in 0.9% saline (25 mg/ml) on the day of the experiment. Platelet-activating factor (Sigma Chemicals, St. Louis, MO.) was stored at a stock solution (2 mg/ml) in chloroform. Immediately before use, an aliquot of the stock solution was diluted in 0.25% bovine serum albumin/O.9% saline. The PAF was infused at a rate of 0.1 ml/min. BN-52021 [9H-1,7 a - (epoxymethano) - lH, 6aH cyclopenta (c) furo - (2,3-b) furo - (3’,2’:3,4) cyclopenta (1,2 - d) furan - 5,9,12 - (4H) - trione, 3 - tert - butyl hexahydro - 8 - methyl], an extract from Gingko biloba leaves (IHB-IPSEN Research Laboratories, Le Plessis Robinson, France), was dissolved in 0.1 N sodium hydroxide, neutralized with HCl, and then diluted with 0.9% saline to a final concentration of 5 mg/ml. Ro-193704 [3-4 (R) - 2 (methoxycarbonyl) oxy - 3 - (octadecylcarbamoyl) oxypropoxy butyl thiazolium iodide; Hoffman LaRoche, Basei, Switzerland] was dissolved in ethanol, then diluted in 0.9% saline to a concentration of 0.5 mg/ml. CV-3988 [(RS) - 2 - methoxy - 3 - (octadecylcarbomoyloxy) propyl 2 - (3- thiazolio) ethyl phosphate; Takeda Chemical Industries, Ltd., Osaka, Japan] was dissolved in 0.9% saline at a concentration of 2 mg/ml and was infused for 10 min at a rate of 0.1 ml/min. Y-labeled human serum albumin (2.5 &ilmg) in isotonic saline was obtained from Amersham International, Buckinghamshire, U.K. Statistical
767
Analyses
All data are expressed as mean + SEM. Comparisons between groups of parametric data were made using Student’s two-tailed t-test for unpaired data, and compar-
vehicle + PAF
Figure
1.
CV3999 + PAF
BN52021 + PAF
Rol93704 + PAF
Effects of intravenous pretreatment with CV-3988(10 mg/kg), BN-52021(10mg/kg), or Ro-193704 (1 ms/kg) on the hypotension (upper), gastric damage (middle), and small intestinal damage (lower) induced by a lo-min intravenous infusion of PAF (100 ng/kg min). The change in arterial blood pressure is shown in mmHg and was calculated as the difference between resting levels and the blood pressure 5 min after beginning the PAF infusion. Gastric and small intestinal damage were scored macroscopically in a blind manner. The tissues were assigned a score on a 0 (normal ) to 3 (severe] scale. The data are expressed as the mean 2 SEM of 4-9 experiments. Asterisks denote groups that differ significantly from the PAF control group (***p < 0.001; Students t-test for blood pressure, Mann-Whitney U test for damage scores).
isons between groups of nonparametric data were made with the Mann-Whitney U test. With all analyses, an associated probability (p value) of 55% was considered significant.
Results Platelet-Activating Platelet-activating
ministration
resulted
Factor factor
(100 ng/kg
in a rapid
min)
ad-
fall in BP, which
remained depressed throughout the lo-min infusion and slowly recovered toward resting levels thereafter. The mean BP 5 min after beginning the PAF infusion was reduced by 77 ? 6 mmHg (n = 9) from the resting levels. As described previously (7,10,14), PAF infusion at this dosage resulted in extensive hemorrhagic damage to the stomach and small intestine, while the distal colon appeared to be spared of damage. Severe damage was observed in all of the PAF-treated rats (Figure 1). Histologically, the dam-
768
GASTROENTEROLOGY Vol. 93. No. 4
WALLACE ET AL
vehicle t
5 min -
LPS
age was characterized by extensive vascular congestion, hemorrhage, and necrosis. The vascular congestion frequently involved both mucosal and submucosal vessels, whereas necrosis was limited to the mucosa. There was copious blood and fluid present in the lumen of the stomach and small intestine [particularly the latter). In the small intestine, there was extensive destruction of the surface epithelium and the lumen was filled with cellular debris. Pretreatment with each of the PAF antagonists resulted in highly significant (p < 0.001) reductions in the hypotension induced by PAF (Figure 1). Furthermore, each antagonist had a highly significant (p < 0.001)inhibitory effect on PAF-induced 1. Effects of Platelet-Activating
Factor
Antagonists on Gastrointestinal Damage and Hypotension Induced by Escherichia coli Lipopolysaccharide Macroscopic damage score (O-3)
Vehicle CV-3988 (10) BN-52021 (10) BN-52021 (20) Ro-193704 (1) Ro-193704 (2)
s
2.0
_
120 mmHg
Figure 2. Representative trace of the arterial blood pressure response to intravenous E. coli lipopolysaccharide (LPS; 25 mg/kg) administration with and without pretreatment with the PAF antagonist BN-52021 (20 m&g). The hypotension induced by lipopolysaccharide in the control group [upper trace) is biphasic. BN-52021 (lower trace), and the other PAF antagonists, blocked the second phase of the lipopolysaccharide-induced hypotension but did not significantly affect the first.
Pretreatmenta
3.0r
t
EN52021 120mglkg) LPS
Table
w
Stomach
Small intestine
A Blood pressure (mmHg1 1 min after LPS
10 min after LPS
58.6 -r- 9.9 53.8 2 8.5 2.4 + 0.3 3.0 2 0 1.0 f 0.6b 1.6 f 0.6b 40.0 2 8.0 11.5 + 6.6' 1.3 f 0.5b 1.6 2 0.6b 47.5 k 9.2 18.4 + 6.5b 1.1 It 0.4b 1.1 2 O.ld 39.0 t 2.7 9.1 '- 7.0" 1.2 2 0.4b 1.3 2 O.Zd 55.1 z!z6.3 35.0 k 8.6
0.7 2 0.4= 0.7 ” 0.2d 59.5 2 7.5
9.2 +- 4.gc
LPS, E. coli lipopolysaccharide. Each value represents the mean * SEM for 4 to 5 experiments. Changes in systemic arterial blood pressure represent the difference between the resting value 1 and 10 min after intravenous administration of E. coli lipopolysacchafollowing the ride (25 mg/kg). a The number in parentheses pretreatment signifies the dose of the drug used, in mg/kg. Superscripts b, c, and d denote groups that differ significantly from the control (vehicle-pretreated) group (b p < 0.05; ’ p < 0.01; d p < 0.001; Mann-Whitney U test for damage scores, Student’s ttest for change in blood pressure).
2: 00, Em
1.0
a d 0 Vehicle
CV3988 1101 PRETREATMENT
BN52021 I101
Ro193704 11)
(mglkgl
Figure 3. Effects of PAF antagonists on histologically assessed gastric (lower) and small intestinal (upper) damage induced by E. coli lipopolysaccharide. Rats were pretreated with vehicle (control group; n = 81, CV-3988 (n = 41, BN-52021 (n = 4), or Ro-193704 (n = 4) 5 min before intravenous administration of the lipopolysaccharide (25 mg/kg). Damage was scored on a 0 (normal) to 3 (severe) scale, in a blind manner and is expressed as the mean 2 SEM. Asterisks denote groups that differ significantly from the control group (*p < 0.05;***p < 0.001; Mann-Whitney U test).
gastric and intestinal damage. In these groups, damage was limited to very diffuse hyperemia, present in only a few animals (Figure 1). Histologic examination confirmed that the samples from rats pretreated with a PAF antagonist were devoid of necrosis and hemorrhage. Endotoxic
Shock
Intravenous injection of the E. coli endotoxin (LPS, 25 mg/kg) resulted in a biphasic fall in BP (Figure 2). The first phase of the hypotension was rapid in onset (A59 2 10 mmHg, Table 1) and the BP had recovered to resting levels within 2 min of the injection. The second fall in BP began -3-5 min after endotoxin administration and was of a similar magnitude to the initial fall in BP, but there was no subsequent recovery toward resting levels. All of the endotoxin-treated rats had extensive damage in the stomach and small intestine. As in the PAF experiments, the distal colon was spared of damage. The lumen of the stomach and small intestine contained copious fluid and blood. Histologically, endotoxin-induced damage was very similar to that observed after PAF. There was extensive vascular congestion throughout the stomach and small bowel (Figures 3 and 4A). Overt bleeding and extensive necrosis were also evident, particularly in the duodenum (Figure 5A) and jejunum. In regions of
October
1987
PAF AND SHOCK-INDUCED ULCERATION
769
Figure 5. Light micrographs of sections of duodenum from rats treated with E. coli lipopolysaccharide (25 mg/kg i.v.). Section A (X110)is from a rat treated only with the endotoxin, whereas section B (x110) is from a rat that received an intravenous injection of BN-52021 (10 mg/kg) before receiving endotoxin. Note the extensive mucosal necrosis of the top half of the villi in section A, with cellular debris present in the lumen, whereas section B is intact.
PAF AND SHOCK-INDUCED ULCERATION
October 1987
the stomach in which the epithelium was intact, subepithelial edema was frequently observed. None of the PAF antagonists had significant effects on the resting BP when injected intravenously, nor did they affect the first phase of the hypotension induced by subsequent endotoxin administration (Table 1).However, all three of the PAF antagonists significantly reduced the second phase of the endotoxin-induced hypotension (Figure 2, Table 1). With Ro-193705, only the higher dose tested (2 mg/kg) caused a significant reduction of this hypotension. Each of the PAF antagonists also significantly (p < 0.001) reduced endotoxin-induced damage in the stomach and small intestine (Table l), with the higher doses of BN-52021 and Ro-193704 being more effective than the lower doses tested. Histologically, gastric and intestinal tissues from endotoxin-treated rats that had been pretreated with one of the PAF antagonists were either devoid of visible damage or there were a few patches of vasocongestion not extending deeper than the subepithelial vessels (Figures 3, 4B, and 5B). No necrosis was observed in these specimens and the surface epithelium was intact. Psf
100 r
d 0
_ _^..
ENDOTOXIN
Stomach Figure
6.
Duodenum
Jajunum
Ileum
Luminal leakage of plasma induced by PAF [top) and endotoxin (bottom). Plasma leakage was measured using ‘*Wabeled human serum albumin, and is expressed as microliters of plasma present in the lumen per gram tissue (mean 2 SEM of 4-6 experiments). Escherichia coli endotoxin was administered intravenously at a dose of 25 mg/kg, whereas PAF was infused intravenously for 10 min (100 &kg min.). CV-3988 (10 mg/kg) was administered intravenously 10 min before PAF or endotoxin. Asterisks denote groups that differ significantly from the endotoxin or PAF control group (*p < 0.05; **p < 0.01; ***p < 0.001;Student’s t-test).
cv3BSS (10)
Whicle
PRETREATMENT
Figure
7.
BN52021 (101
771
ROlB3fO4 (1)
(mg/kg)
Effects of PAF antagonists on ethanol-induced gastric damage. Rats were pretreated intravenously with vehicle (control group, n = 5), CV-3988 (n = 51, BN-52021 (n = 31, or Ro-193704 (n = 31, 10 min before intragastric admini&ation~of-absolute ethanoh Gas&& damage was quantified planimetrically and is expressed as the percentage of glandular mucosa with macroscopically visible damage (mean + SEMI. Asterisks denote groups that differ significantly from the control group (***p < 0.001; Student’s t-test).
Luminal
PlQSITlQ
Leakage
Control levels of ‘251-labeled albumin from the stomach, duodenum, jejunum, and ileum were, respectively, 7.0 + 1.3, 12.2 + 1.6, 12.8 + 1.2,and 11.0 ? 2.0 pllg tissue (n = 4 per group). With both PAF and endotoxin there were highly significant (p < 0.001) increases in plasma leakage from all four tissues studied (Figure 6). Note that the term “leakage” is used to describe plasma entering the lumen through both hemorrhage and exudation. The duodenum and jejunum were the most affected regions, which is consistent with the macroscopic and histologic observation of extensive necrosis in these tissues. Platelet-activating factor- and endotoxininduced plasma leakage were very similar in terms of both the absolute amount of leakage observed and the relative leakage from the four tissues studied. Pretreatment with CV-3988 (10 mg/kg) caused a sighificant (p < 0.05) reduction in both PAF- and endotoxin-induced plasma leakage into the lumen of all four tissues tested. In the group that was pretreated with CV-3988 before PAF infusion, the plasma leakage was not significantly different from control levels. Ethanol-Induced
Damage
Intragastric administration of 3.0 ml of absolute ethanol resulted in extensive hemorrhagic necrosis of the gastric mucosa, as has been described in detail previously (16,17). As shown in Figure 7, this damage involved 40% + 3% of the glandular mucosa. Pretreatment with BN-52021 (10 mg/kg) or
772
WALLACE
ET AL.
Ro-193704 (1 mg/kg) did not significantly affect the extent of ethanol-induced damage. However, with CV-3988 pretreatment (10 mg/kg), a significant (p < 0.001) reduction in macroscopically visible damage was observed.
Discussion Septic shock is most often encountered clinically after urogenital or gastrointestinal tract surgery (18). Its features include systemic hypotension, increased vascular permeability, bronchoconstriction, pulmonary hypertension, and gastrointestinal ulceration (2). These symptoms can be produced in laboratory animals via the intravenous injection of endotoxin, as in the present study. The delay in the onset of some of the symptoms after endotoxin administration, such as the hypotension, suggests that the endotoxin induces the release of endogenous mediators. Release of prostaglandins, thromboxane AZ, and endorphins have previously been suggested to mediate some of the effects that follow endotoxin administration (3-5). The results of the present study support the hypothesis that PAF is an important mediator of both the hypotension and gastrointestinal ulceration associated with endotoxin shock. The present results confirm previous observations that intravenous administration of PAF produces many of the symptoms of endotoxic shock, including gastrointestinal ulceration (7,19,20). The damage to the gastrointestinal tract in both models was characterized by vascular congestion extending throughout the mucosa and sometimes to the submucosa. We have previously speculated that the neutrophilaggregating actions of PAF may contribute to such vascular congestion (10). Although we have no data as yet to suggest that neutrophil aggregates also contribute to endotoxin-induced gastrointestinal damage, injury to the lung induced by endotoxin can be prevented by depleting the animal of neutrophils (21) or by preventing their aggregation with prostacyclin (22). It is noteworthy that both endotoxin and PAF can induce hemorrhagic damage in the stomach and small intestine, but not in the distal colon. The reason for this resistance of the distal colon to these damaging agents is not yet clear. The strongest evidence for a role of PAF in endotoxic shock is the observation that at doses that inhibit PAF-induced gastrointestinal damage, the three structurally dissimilar PAF antagonists also inhibited endotoxin-induced gastrointestinal ulceration, as assessed macroscopically and histologically. Ro-193704 was effective at one-tenth the dose of BN-52021 or CV-3988, confirming the greater potency of this compound (9). The PAF antagonist
GASTROENTEROLOGY
Vol. 93, No. 4
actions of these compounds were not limited to the gastrointestinal tract, as PAF-induced hypotension was also significantly inhibited by all three antagonists. The hypotension induced by endotoxin was biphasic, and it would appear that PAF was a mediator of the second, but not the first phase. Thus three PAF antagonists inhibited the second phase of the hypotension, whereas none significantly affected the first phase. Because none of the PAF antagonists completely inhibited endotoxin-induced gastrointestinal damage, it is possible that the endotoxin itself had some direct damaging actions in addition to these apparently direct hypotensive effects. It was possible that the protection of the stomach afforded by the PAF antagonists was unrelated to antagonism of PAF receptors. However, the studies with ethanol-induced damage demonstrated that Ro-193704 and BN-52021 were without protective effects in that model. Conversely, CV-3988 significantly reduced ethanol-induced damage, suggesting that this compound has additional actions unrelated to antagonism of PAF receptors. These experiments also suggest that PAF release does not play a role in the mechanism of ethanol-induced gastric damage. The leakage of 1251-labeled albumin into the lumen of the stomach and small intestine was another parameter in which there was remarkable similarity between the PAF and endotoxin models of damage. It is not clear if this leakage was a factor contributing to the tissue damage or purely a consequence of it. The present studies, however, demonstrated that pretreatment with a PAF antagonist, CV-3988, resulted in significant inhibition of such plasma leakage in both models. We have suggested that the substantial hemoconcentration that follows PAF administration, as a consequence of increased vascular permeability, is likely to contribute to the slowing and stasis of mucosal blood flow that precedes necrosis (23,24). Such hemoconcentration can be significantly reduced by pretreatment with corticosteroids, such as prednisolone, as well as by the dual cyclooxygenase/lipoxygenase inhibitor, BW755c, which also inhibit PAF-induced gastric damage (24). Using a similar model of endotoxic shock, Doebber et al. (8) demonstrated the appearance of PAF in the blood. Platelet-activating factor was not detectable in blood samples taken 1.5 min after intravenous administration of E. coli endotoxin (50 mg/kg). At such a time, the second phase of the hypotension induced by endotoxin would not yet have occurred. In samples taken 10 min after endotoxin, however, they found PAF at concentrations as high as 4.7 nM. Doebber et al. calculated that such a concentration could be achieved by an intravenous injection of 300 pmol/kg of PAF. In the present study, PAF was infused at a dose of -200 pmobkg. Hence, the dose
October 1987
of PAF used in the present study should have given blood levels in the range of those that have been detected in blood 10 min after endotoxin administration. Other studies have also demonstrated the release of PAF after endotoxin administration (25-27). The similarity in the severity of gastrointestinal damage, hypotension, and plasma leakage between the PAF and endotoxin models in the present study supports the hypothesis that similar blood levels of PAF were achieved under either condition. The present study strongly supports a role for PAF as a mediator of endotoxic shock in the rat. The findings also clearly implicate the endogenous release of PAF as a mediator of the pathological events leading to gastrointestinal necrosis and hemorrhage in endotoxemia. Whether PAF plays an important role as a mediator of septic shock in humans is not yet known, but is a question that certainly warrants further study.
References 1. Robert
2. 3.
4.
5.
6.
7.
6.
9.
10.
A, Kauffman GL. Stress ulcers. In: Sleisenger MH, Fordtran JS, eds. Gastrointestinal disease. Philadelphia: WB Saunders, 1983:612-25. Shumer W. Septic shock. JAMA 1979;242:1906-7. Collier JG, Herman AG, Vane JR. Appearance of prostaglandins in renal venous blood of dogs in response to acute systemic hypotension produced by bleeding or endotoxin. J Physiol (Lond) 1973;230:14-16P. Cook JA, Wise WC, Haluska PV. Elevated thromboxane levels in the rat during endotoxin shock. J Clin Invest 1980;65:22730. Holaday JW, Faden AI. Naloxone reversal of endotoxin hypotension suggests role of endorphins in shock. Nature 1978; 275:450-l. Terashita Z, Imura Y, Nishikawa K, Sumida S. Is platelet activating factor (PAF) a mediator of endotoxin shock? Eur J Pharmacol 1985;109:257-61. Rosam A-C, Wallace JL, Whittle BJR. Potent ulcerogenic actions of platelet-activating factor on the stomach. Nature 1986;319:54-6. Doebber TW, Wu MS, Robbins JC, Choy BM, Chang MN, Shen TY. Platelet activating factor (PAF) involvement in endotoxin-induced hypotension in rats. Studies with PAF-receptor antagonist kadsurenone. Biochem Biophys Res Commun 1985;127:799-808. Braquet P, Touqui L, Shen TY, Vargaftig BB. Perspectives in platelet activating factor research. Pharmacol Rev 1987;39: 97-145. Wallace JL, Whittle BJR. Picomole doses of platelet-activating factor predispose the gastric mucosa to damage by topical irritants. Prostaglandins 1986;31:989-98.
PAF AND SHOCK-INDUCED ULCERATION
773
11. Terashita Z, Tsushima S, Yoshioka Y, Nomura H, Inada Y, Mishikawa K. CV-3988-a specific antagonist of platelet activating factor (PAF). Life Sci 1983;32:1975-82. 12. Braquet P, Spinnewyn B, Braquet M, et al. BN52021 and related compounds: a new series of highly specific PAFacether receptor antagonists isolated from Gingko biloba. Blood Vessels 1985;16:558-72. of PAF-acether 13. Hardvary P, Baumgartner H. Interference antagonists with platelet aggregation and with the formation of platelet thrombi. Prostaglandins 1985;30:694. damage 14. Wallace JL, Whittle BJR. Profile of gastrointestinal induced by platelet-activating factor. Prostaglandins 1986;32: 137-41. 15. Wallace JL, Whittle BJR. Acceleration of recovery of epithelial integrity by 16,16-dimethyl prostaglandin E2. Br J Pharmacol 1985;86:837-42. 16. Morris GP, Wallace JL. The roles of ethanol and of acid in the production of gastric mucosal erosions in rats. Virchows Arch [B] 1981;38:23-38. 17. Lacy ER, Ito S. Microscopic analysis of ethanol damage to rat gastric mucosa after treatment with a prostaglandin. Gastroenterology 1982;82:619-25. 18. Litton A. Gram-negative septicaemia in surgical practice. Br J Surg 1975;62:773-6. 19. Bessin P, Bonnet J, Thiboudeau D, Agier B, Beaudet Y, Gilet F. Pathophysiology of shock states caused by PAF-acether in dogs and cats. In: Benveniste J, Arnoux B, eds. Plateletactivating factor. INSERM Symposium No. 23. Amsterdam: Elsevier, 1983:343-56. 20. Gonzalez-Crussi F, Hsueh W. Experimental model of ischemit bowel necrosis. The role of platelet-activating factor and endotoxin. Am J Path01 1983;112:127-35. 21. Heflin AC, Brigham KL. Prevention by granulocyte depletion of increased vascular permeability of sheep lung following endotoxemia. J Clin Invest 1981;68:1253-60. F. Inhibitory effect of 22. Camussi G, Tetta C, Bussolino prostacyclin (PGIZ) on neutropenia induced by intravenous injection of platelet-activating factor (PAF) in the rabbit. Prostaglandins 1983;25:343-51. 23. Whittle BJR, Morishita T, Ohya Y, Leung F, Guth PH. Microvascular actions of platelet-activating factor (PAF) on the rat gastric mucosa and submucosa. Am J Physiol1986;251:G7729. 24. Wallace JL, Whittle BJR. Effects of inhibitors of arachidonic acid metabolism on Paf-induced gastric mucosal damage and haemoconcentration. Br J Pharmacol 1986;89:415-22. 25. Inarrea P, Gomez-Cambronero J, Pascual J. Carmen Ponte M, Hernando L, Sanchez-Crespo M. Synthesis of PAF-acether and blood volume changes in gram-negative sepsis. Immunopharmacology 1985;9:45-52. 26. Beijer L, Rylander R. Endotoxin inhalation induces alveolar macrophage production of Paf-acether. Am Rev Resp Dis 1984;129:A154. 27. Lagente V, Lidbury P, Steel G, Vargaftig BB, Wallace JL, Whittle BJR. Role of PAF as a mediator of endotoxin-induced gastrointestinal damage (abstr). Br J Pharmacol 1987;90:14OP.
1987;93:765-73
Evidence for Platelet-Activating Factor as a Mediator of Endotoxin-Induced Gastrointestinal Damage in the Rat Effects of Three Platelet-Activating Antagonists
Factor
JOHN L. WALLACE, GRAHAM STEEL, BRENDAN J.R. WHITTLE, VINCENT LAGENTE, and BORIS VARGAFTIG Department of Mediator Pharmacology, United Kingdom; and Unite des venins, France
Wellcome Research Laboratories, Beckenham, Kent, Unite associee Institut Pasteur/INSERM U 285, Paris,
The potential role of platelet-activating factor (PAF) as a mediator of gastrointestinal ulceration associated with septic shock was examined in the rat. The damaging effects of both PAF and Escherichia coli endotoxin in the stomach and small intestine were compared, as were their effects on plasma leakage into the lumen of the gastrointestinal tract. Intravenous administration of either endotoxin or PAF produced extensive necrosis and vascular congestion in the stomach and small intestine, but not the distal colon. With either agent, the duodenum and jejunum were the tissues most susceptible to damage and in which the greatest plasma leakage was observed. The prolonged hypotension and gastrointestinal damage induced by PAF or endotoxin were significantly inhibited by three structurally dissimilar PAF antagonists (CV-3988, BN-52021, and Ro-193704). CV-3988 [lo mg/kg) significantly (p < 0.05) reduced both endotoxinand PAF-induced plasma leakage in the stomach and small intestine. Of the three antagonists, only CV-3988 significantly reduced ethanol-induced gastric mucosal damage, perhaps reflecting actions of this compound unrelated to antagonism of PAF receptors. These studies support the hypothesis that PAF is an important Received August 22, 1986. Accepted April 1, 1987. Address requests for reprints to: Dr. J.L. Wallace, Department of Physiology, Botterell Hall, Queen’s University, Kingston, Ontario, K7L 3N6, Canada. Dr. Wallace is the recipient of a scholarship from the Medical Research Council of Canada. CV-3988 was a gift from Dr. M. Nishikawa, Takeda Chemical Industries, Ltd. BN-52021 was a gift from Dr. P. Braquet, Institut Henri Beaufour. Ro-193704 was a gift from Dr. H. Baumgartner, Hoffman LaRoche. 0 1987 by the American Gastroenterological Association 0016-5085/87/$3.50
mediator of the hypotension and plasma leakage observed during endotoxic shock and its endogenous release may contribute to the gastrointestinal ulceration associated with this syndrome. Thus, PAF receptor antagonists may be useful for prevention of such ulceration. Septic shock is a major risk factor for gastrointestinal ulceration (1) and accounts for as many as 100,000 deaths per year in the United States (2). The sequence of events following the appearance of invading bacteria in the systemic circulation is poorly understood, although the release of potent mediators such as prostaglandins, thromboxane AZ, and endorphins may account for some or all of the symptoms of septic shock (3-5). The observation that many of the features of septic shock, such as systemic hypotension, hemoconcentration, and gastrointestinal ulceration, could be mimicked by intravenous administration of a substance known as platelet-activating factor (PAF), led to the hypothesis that PAF may be an important mediator of septic shock (6,7). Platelet-activating factor is a lowmolecular-weight phospholipid characterized by an ether linkage at the 1’ positive (usually with a fatty acid chain of 16 or 18 carbons] and an acetyl group at the 2’ position of a phosphatidylcholine. Further evidence of a role for this molecule in septic shock was provided by the demonstration that intravenous administration of endotoxin to rats resulted in the release of PAF into the bloodstream (8). The cellular source of this PAF was not known, although neutroAbbreviations platelet-activating
used in this factor.
paper:
BP, blood
pressure;
PAF,
766 WALLACE ET AL.
GASTROENTEROLOGY Vol. 93, No. 4
phils, macrophages, monocytes, basophils, endothelial cells, mast cells, and platelets have all been shown to release PAF in vitro (9). A role for PAF as a mediator of gastrointestinal ulceration is supported by our observations that PAF induced gastric necrosis in the rat stomach at doses as low as XI pmol/kg (i’), whereas doses as low as 2 pmol/kg significantly predisposed the rat stomach to damage induced by topically applied 20% ethanol (10). In the present study, we have examined the role of PAF in an endotoxic shock model in the rat. In addition to comparing the gastrointestinal damage induced by PAF and endotoxin, we have studied the effects of three structurally unrelated PAF receptor antagonists in these models. If PAF mediates the vasocongestion and necrosis induced in the gastrointestinal tract by endotoxin, pretreatment with a PAF antagonist should confer protection. The three PAF antagonists used include two that are structural analogues of PAF ((X-3988 and Ro-193704) and one (BN-52021) have been
that
is not
onists
of PAF,
gation
induced
osine
diphosphate,
prevent
the
their receptors of these
(11-13).
demonstrated as they
These
to be highly failed
to block
by arachidonic and
binding
Materials
antag-
platelet
aggre-
collagen,
thrombin,
of several
(see Reference
compounds
acid,
compounds
specific
and their
other
and
adendid
not
mediators
to
9 for a detailed
review
specificity].
and Methods
Male Wistar rats (200-220 g) were deprived of food, but not water, for 20-24 h before an experiment. The animals were anesthetized by an intraperitoneal injection of sodium pentobarbitone (60 mg/kg) and a carotid artery was cannulated for measurement of systemic arterial blood pressure (BP). A 25-gauge needle was inserted into a tail vein for administration of PAF, endotoxin, or drugs.
Platelet-Activating Factor-Induced Gastrointestinal Damage A stable resting BP was obtained for at least 10 min before the vehicle or one of the PAF antagonists (BN-52021 at 10 mg/kg, CV-3988 at 10 mg/kg, or Ro-193704 at 1 mg/kg) was injected intravenously. W-3988 was administered as an infusion over 10 min, whereas the other PAF antagonists were given by bolus injection. Five minutes later, a lo-min intravenous infusion of PAF was started (100 ng/kg . min). This dose of PAF was selected because it has previously been shown to induce gastrointestinal damage (10,14) and because it should produce blood levels of PAF similar to those detected after endotoxin administration to rats (8). Twenty minutes after completion of the infusion of PAF, the rat was killed and the stomach and small bowel were examined and scored for damage as described previously (14), based on the severity of hyperemia and hemorrhage (O-3 scale). Samples of the fundic region of the stomach as well as
samples from standardized regions of the proximal duodenum, jejunum, ileum, and distal colon were fixed in neutral buffered formalin and processed by routine techniques before embedding in paraffin wax. Sections (4 pm) were stained with hematoxylin and eosin and were examined under a light microscope. The tissue sections were given a score on a 0-3 scale based on the following criteria: 0 = normal; 1 = subepithelial vasocongestion + patches of superficial necrosis; 2 = mucosal vasocongestion and focal necrosis; 3 = extensive vasocongestion and necrosis involving the full thickness of the mucosa. The overall damage score for the small intestine was taken as the mean scores for the sections of duodenum, jejunum, and ileum. All assessments of damage were performed in a blinded manner, to avoid observer bias. Endotoxin-Induced
Gastrointestinal
Damage
A stable resting BP was obtained for at least 10 min before the vehicle or one of the PAF antagonists (BN-52021 at 10 or 20 mg/kg, CV-3988 at 10 mg/kg, or Ro-193704 at 1 or 2 mgikg) was injected intravenously as described above. Ten minutes after administration of the vehicle or PAF antagonist, Escherichia coli endotoxin was administered as an intravenous bolus at a dose of 25 mg/kg. Twenty minutes later the rat was killed, and the stomach and small intestinal damage were scored as in the PAF experiments. Samples of stomach and small intestine were fixed and processed for light microscopy as above. Luminal
Plasma
Leakage
A series of experiments were performed to determine the amount of leakage of plasma into the lumen of the stomach and small intestine. Rats were treated with PAF or endotoxin, as described above, with some rats receiving prior intravenous administration of CV-3988 (10 mgikg). Five minutes before the injection of endotoxin or PAF, ‘ZSI-labeled human serum albumin (6.25 &i) was injected intravenously via the tail vein. Twenty minutes after endotoxin or PAF administration, the peritoneum was opened and the stomach and segments of the duodenum, jejunum, and ileum (-6 cm long) were clamped, so that luminal fluid could not leak, and subsequently excised. The segment of duodenum extended from the pylorus to the ligament of Treitz. The segment of ileum was taken proximal to the ileocecal valve. The stomach and segments of intestine were placed in individual glass vials containing 2.0 ml of 0.9% saline and were opened by a longitudinal incision to allow for mixing of luminal contents and saline. The tissues were then removed, blotted dry, and weighed. The samples of luminal fluid were mixed well, then 1.0 ml was transferred to a glass tube for quantification of radioactivity on a y-spectrometer. After the stomach and segments of gut had been taken from the rat, a l.O-ml blood sample was taken from the descending aorta and centrifuged for 1 min using an Eppendorf benchtop centrifuge (14,000 g). A loo-p1 samples of the plasma was transferred to a glass tube for estimation of radioactivity on a y-spectrometer. This yielded a value for counts per minute per microliter of plasma, which was then used to calculate the number of microliters of plasma
October 1987
PAFANDSHOCK-INDUCEDULCERATION
present in the luminal of tissue).
fluid samples
Ethanol-Induced
Gastric
(expressed
per gram
Damage
To determine if there were any protective actions of the PAF antagonists in the gastric mucosa that might be unrelated to their actions on PAF receptors, the effects of these compounds on ethanol-induced gastric damage were assessed. Rats that had been fasted for 20-24 h were anesthetized with sodium pentobarbitone (60mg/kg) and a 25-gauge needle was inserted into a tail vein. The peritoneum was opened and ligatures were placed around the esophagus and pylorus. Either the appropriate vehicle or one of the PAF antagonists was injected or infused intravenously as described above. For these experiments, only one dose of each antagonist was tested (CV-3988 at 10 mg/kg, BN-52021 at 10 mg/kg, or Ro-193704 at 1 mgikg). Ten minutes later, 3.0 ml of absolute ethanol was injected into the stomach through a 30-gauge needle inserted through the forestomach. The ethanol was left in the stomach for 10 min, after which time the stomach was removed, opened along the greater curvature, pinned out on a wax block, and photographed. The extent of macroscopically visible damage was quantified planimetrically as described previously (15). Materials Escherichia coli lipopolysaccharide (Olll:B4) was obtained from Difco Laboratories (Detroit, Mich.) and was prepared as a suspension in 0.9% saline (25 mg/ml) on the day of the experiment. Platelet-activating factor (Sigma Chemicals, St. Louis, MO.) was stored at a stock solution (2 mg/ml) in chloroform. Immediately before use, an aliquot of the stock solution was diluted in 0.25% bovine serum albumin/O.9% saline. The PAF was infused at a rate of 0.1 ml/min. BN-52021 [9H-1,7 a - (epoxymethano) - lH, 6aH cyclopenta (c) furo - (2,3-b) furo - (3’,2’:3,4) cyclopenta (1,2 - d) furan - 5,9,12 - (4H) - trione, 3 - tert - butyl hexahydro - 8 - methyl], an extract from Gingko biloba leaves (IHB-IPSEN Research Laboratories, Le Plessis Robinson, France), was dissolved in 0.1 N sodium hydroxide, neutralized with HCl, and then diluted with 0.9% saline to a final concentration of 5 mg/ml. Ro-193704 [3-4 (R) - 2 (methoxycarbonyl) oxy - 3 - (octadecylcarbamoyl) oxypropoxy butyl thiazolium iodide; Hoffman LaRoche, Basei, Switzerland] was dissolved in ethanol, then diluted in 0.9% saline to a concentration of 0.5 mg/ml. CV-3988 [(RS) - 2 - methoxy - 3 - (octadecylcarbomoyloxy) propyl 2 - (3- thiazolio) ethyl phosphate; Takeda Chemical Industries, Ltd., Osaka, Japan] was dissolved in 0.9% saline at a concentration of 2 mg/ml and was infused for 10 min at a rate of 0.1 ml/min. Y-labeled human serum albumin (2.5 &ilmg) in isotonic saline was obtained from Amersham International, Buckinghamshire, U.K. Statistical
767
Analyses
All data are expressed as mean + SEM. Comparisons between groups of parametric data were made using Student’s two-tailed t-test for unpaired data, and compar-
vehicle + PAF
Figure
1.
CV3999 + PAF
BN52021 + PAF
Rol93704 + PAF
Effects of intravenous pretreatment with CV-3988(10 mg/kg), BN-52021(10mg/kg), or Ro-193704 (1 ms/kg) on the hypotension (upper), gastric damage (middle), and small intestinal damage (lower) induced by a lo-min intravenous infusion of PAF (100 ng/kg min). The change in arterial blood pressure is shown in mmHg and was calculated as the difference between resting levels and the blood pressure 5 min after beginning the PAF infusion. Gastric and small intestinal damage were scored macroscopically in a blind manner. The tissues were assigned a score on a 0 (normal ) to 3 (severe] scale. The data are expressed as the mean 2 SEM of 4-9 experiments. Asterisks denote groups that differ significantly from the PAF control group (***p < 0.001; Students t-test for blood pressure, Mann-Whitney U test for damage scores).
isons between groups of nonparametric data were made with the Mann-Whitney U test. With all analyses, an associated probability (p value) of 55% was considered significant.
Results Platelet-Activating Platelet-activating
ministration
resulted
Factor factor
(100 ng/kg
in a rapid
min)
ad-
fall in BP, which
remained depressed throughout the lo-min infusion and slowly recovered toward resting levels thereafter. The mean BP 5 min after beginning the PAF infusion was reduced by 77 ? 6 mmHg (n = 9) from the resting levels. As described previously (7,10,14), PAF infusion at this dosage resulted in extensive hemorrhagic damage to the stomach and small intestine, while the distal colon appeared to be spared of damage. Severe damage was observed in all of the PAF-treated rats (Figure 1). Histologically, the dam-
768
GASTROENTEROLOGY Vol. 93. No. 4
WALLACE ET AL
vehicle t
5 min -
LPS
age was characterized by extensive vascular congestion, hemorrhage, and necrosis. The vascular congestion frequently involved both mucosal and submucosal vessels, whereas necrosis was limited to the mucosa. There was copious blood and fluid present in the lumen of the stomach and small intestine [particularly the latter). In the small intestine, there was extensive destruction of the surface epithelium and the lumen was filled with cellular debris. Pretreatment with each of the PAF antagonists resulted in highly significant (p < 0.001) reductions in the hypotension induced by PAF (Figure 1). Furthermore, each antagonist had a highly significant (p < 0.001)inhibitory effect on PAF-induced 1. Effects of Platelet-Activating
Factor
Antagonists on Gastrointestinal Damage and Hypotension Induced by Escherichia coli Lipopolysaccharide Macroscopic damage score (O-3)
Vehicle CV-3988 (10) BN-52021 (10) BN-52021 (20) Ro-193704 (1) Ro-193704 (2)
s
2.0
_
120 mmHg
Figure 2. Representative trace of the arterial blood pressure response to intravenous E. coli lipopolysaccharide (LPS; 25 mg/kg) administration with and without pretreatment with the PAF antagonist BN-52021 (20 m&g). The hypotension induced by lipopolysaccharide in the control group [upper trace) is biphasic. BN-52021 (lower trace), and the other PAF antagonists, blocked the second phase of the lipopolysaccharide-induced hypotension but did not significantly affect the first.
Pretreatmenta
3.0r
t
EN52021 120mglkg) LPS
Table
w
Stomach
Small intestine
A Blood pressure (mmHg1 1 min after LPS
10 min after LPS
58.6 -r- 9.9 53.8 2 8.5 2.4 + 0.3 3.0 2 0 1.0 f 0.6b 1.6 f 0.6b 40.0 2 8.0 11.5 + 6.6' 1.3 f 0.5b 1.6 2 0.6b 47.5 k 9.2 18.4 + 6.5b 1.1 It 0.4b 1.1 2 O.ld 39.0 t 2.7 9.1 '- 7.0" 1.2 2 0.4b 1.3 2 O.Zd 55.1 z!z6.3 35.0 k 8.6
0.7 2 0.4= 0.7 ” 0.2d 59.5 2 7.5
9.2 +- 4.gc
LPS, E. coli lipopolysaccharide. Each value represents the mean * SEM for 4 to 5 experiments. Changes in systemic arterial blood pressure represent the difference between the resting value 1 and 10 min after intravenous administration of E. coli lipopolysacchafollowing the ride (25 mg/kg). a The number in parentheses pretreatment signifies the dose of the drug used, in mg/kg. Superscripts b, c, and d denote groups that differ significantly from the control (vehicle-pretreated) group (b p < 0.05; ’ p < 0.01; d p < 0.001; Mann-Whitney U test for damage scores, Student’s ttest for change in blood pressure).
2: 00, Em
1.0
a d 0 Vehicle
CV3988 1101 PRETREATMENT
BN52021 I101
Ro193704 11)
(mglkgl
Figure 3. Effects of PAF antagonists on histologically assessed gastric (lower) and small intestinal (upper) damage induced by E. coli lipopolysaccharide. Rats were pretreated with vehicle (control group; n = 81, CV-3988 (n = 41, BN-52021 (n = 4), or Ro-193704 (n = 4) 5 min before intravenous administration of the lipopolysaccharide (25 mg/kg). Damage was scored on a 0 (normal) to 3 (severe) scale, in a blind manner and is expressed as the mean 2 SEM. Asterisks denote groups that differ significantly from the control group (*p < 0.05;***p < 0.001; Mann-Whitney U test).
gastric and intestinal damage. In these groups, damage was limited to very diffuse hyperemia, present in only a few animals (Figure 1). Histologic examination confirmed that the samples from rats pretreated with a PAF antagonist were devoid of necrosis and hemorrhage. Endotoxic
Shock
Intravenous injection of the E. coli endotoxin (LPS, 25 mg/kg) resulted in a biphasic fall in BP (Figure 2). The first phase of the hypotension was rapid in onset (A59 2 10 mmHg, Table 1) and the BP had recovered to resting levels within 2 min of the injection. The second fall in BP began -3-5 min after endotoxin administration and was of a similar magnitude to the initial fall in BP, but there was no subsequent recovery toward resting levels. All of the endotoxin-treated rats had extensive damage in the stomach and small intestine. As in the PAF experiments, the distal colon was spared of damage. The lumen of the stomach and small intestine contained copious fluid and blood. Histologically, endotoxin-induced damage was very similar to that observed after PAF. There was extensive vascular congestion throughout the stomach and small bowel (Figures 3 and 4A). Overt bleeding and extensive necrosis were also evident, particularly in the duodenum (Figure 5A) and jejunum. In regions of
October
1987
PAF AND SHOCK-INDUCED ULCERATION
769
Figure 5. Light micrographs of sections of duodenum from rats treated with E. coli lipopolysaccharide (25 mg/kg i.v.). Section A (X110)is from a rat treated only with the endotoxin, whereas section B (x110) is from a rat that received an intravenous injection of BN-52021 (10 mg/kg) before receiving endotoxin. Note the extensive mucosal necrosis of the top half of the villi in section A, with cellular debris present in the lumen, whereas section B is intact.
PAF AND SHOCK-INDUCED ULCERATION
October 1987
the stomach in which the epithelium was intact, subepithelial edema was frequently observed. None of the PAF antagonists had significant effects on the resting BP when injected intravenously, nor did they affect the first phase of the hypotension induced by subsequent endotoxin administration (Table 1).However, all three of the PAF antagonists significantly reduced the second phase of the endotoxin-induced hypotension (Figure 2, Table 1). With Ro-193705, only the higher dose tested (2 mg/kg) caused a significant reduction of this hypotension. Each of the PAF antagonists also significantly (p < 0.001) reduced endotoxin-induced damage in the stomach and small intestine (Table l), with the higher doses of BN-52021 and Ro-193704 being more effective than the lower doses tested. Histologically, gastric and intestinal tissues from endotoxin-treated rats that had been pretreated with one of the PAF antagonists were either devoid of visible damage or there were a few patches of vasocongestion not extending deeper than the subepithelial vessels (Figures 3, 4B, and 5B). No necrosis was observed in these specimens and the surface epithelium was intact. Psf
100 r
d 0
_ _^..
ENDOTOXIN
Stomach Figure
6.
Duodenum
Jajunum
Ileum
Luminal leakage of plasma induced by PAF [top) and endotoxin (bottom). Plasma leakage was measured using ‘*Wabeled human serum albumin, and is expressed as microliters of plasma present in the lumen per gram tissue (mean 2 SEM of 4-6 experiments). Escherichia coli endotoxin was administered intravenously at a dose of 25 mg/kg, whereas PAF was infused intravenously for 10 min (100 &kg min.). CV-3988 (10 mg/kg) was administered intravenously 10 min before PAF or endotoxin. Asterisks denote groups that differ significantly from the endotoxin or PAF control group (*p < 0.05; **p < 0.01; ***p < 0.001;Student’s t-test).
cv3BSS (10)
Whicle
PRETREATMENT
Figure
7.
BN52021 (101
771
ROlB3fO4 (1)
(mg/kg)
Effects of PAF antagonists on ethanol-induced gastric damage. Rats were pretreated intravenously with vehicle (control group, n = 5), CV-3988 (n = 51, BN-52021 (n = 31, or Ro-193704 (n = 31, 10 min before intragastric admini&ation~of-absolute ethanoh Gas&& damage was quantified planimetrically and is expressed as the percentage of glandular mucosa with macroscopically visible damage (mean + SEMI. Asterisks denote groups that differ significantly from the control group (***p < 0.001; Student’s t-test).
Luminal
PlQSITlQ
Leakage
Control levels of ‘251-labeled albumin from the stomach, duodenum, jejunum, and ileum were, respectively, 7.0 + 1.3, 12.2 + 1.6, 12.8 + 1.2,and 11.0 ? 2.0 pllg tissue (n = 4 per group). With both PAF and endotoxin there were highly significant (p < 0.001) increases in plasma leakage from all four tissues studied (Figure 6). Note that the term “leakage” is used to describe plasma entering the lumen through both hemorrhage and exudation. The duodenum and jejunum were the most affected regions, which is consistent with the macroscopic and histologic observation of extensive necrosis in these tissues. Platelet-activating factor- and endotoxininduced plasma leakage were very similar in terms of both the absolute amount of leakage observed and the relative leakage from the four tissues studied. Pretreatment with CV-3988 (10 mg/kg) caused a sighificant (p < 0.05) reduction in both PAF- and endotoxin-induced plasma leakage into the lumen of all four tissues tested. In the group that was pretreated with CV-3988 before PAF infusion, the plasma leakage was not significantly different from control levels. Ethanol-Induced
Damage
Intragastric administration of 3.0 ml of absolute ethanol resulted in extensive hemorrhagic necrosis of the gastric mucosa, as has been described in detail previously (16,17). As shown in Figure 7, this damage involved 40% + 3% of the glandular mucosa. Pretreatment with BN-52021 (10 mg/kg) or
772
WALLACE
ET AL.
Ro-193704 (1 mg/kg) did not significantly affect the extent of ethanol-induced damage. However, with CV-3988 pretreatment (10 mg/kg), a significant (p < 0.001) reduction in macroscopically visible damage was observed.
Discussion Septic shock is most often encountered clinically after urogenital or gastrointestinal tract surgery (18). Its features include systemic hypotension, increased vascular permeability, bronchoconstriction, pulmonary hypertension, and gastrointestinal ulceration (2). These symptoms can be produced in laboratory animals via the intravenous injection of endotoxin, as in the present study. The delay in the onset of some of the symptoms after endotoxin administration, such as the hypotension, suggests that the endotoxin induces the release of endogenous mediators. Release of prostaglandins, thromboxane AZ, and endorphins have previously been suggested to mediate some of the effects that follow endotoxin administration (3-5). The results of the present study support the hypothesis that PAF is an important mediator of both the hypotension and gastrointestinal ulceration associated with endotoxin shock. The present results confirm previous observations that intravenous administration of PAF produces many of the symptoms of endotoxic shock, including gastrointestinal ulceration (7,19,20). The damage to the gastrointestinal tract in both models was characterized by vascular congestion extending throughout the mucosa and sometimes to the submucosa. We have previously speculated that the neutrophilaggregating actions of PAF may contribute to such vascular congestion (10). Although we have no data as yet to suggest that neutrophil aggregates also contribute to endotoxin-induced gastrointestinal damage, injury to the lung induced by endotoxin can be prevented by depleting the animal of neutrophils (21) or by preventing their aggregation with prostacyclin (22). It is noteworthy that both endotoxin and PAF can induce hemorrhagic damage in the stomach and small intestine, but not in the distal colon. The reason for this resistance of the distal colon to these damaging agents is not yet clear. The strongest evidence for a role of PAF in endotoxic shock is the observation that at doses that inhibit PAF-induced gastrointestinal damage, the three structurally dissimilar PAF antagonists also inhibited endotoxin-induced gastrointestinal ulceration, as assessed macroscopically and histologically. Ro-193704 was effective at one-tenth the dose of BN-52021 or CV-3988, confirming the greater potency of this compound (9). The PAF antagonist
GASTROENTEROLOGY
Vol. 93, No. 4
actions of these compounds were not limited to the gastrointestinal tract, as PAF-induced hypotension was also significantly inhibited by all three antagonists. The hypotension induced by endotoxin was biphasic, and it would appear that PAF was a mediator of the second, but not the first phase. Thus three PAF antagonists inhibited the second phase of the hypotension, whereas none significantly affected the first phase. Because none of the PAF antagonists completely inhibited endotoxin-induced gastrointestinal damage, it is possible that the endotoxin itself had some direct damaging actions in addition to these apparently direct hypotensive effects. It was possible that the protection of the stomach afforded by the PAF antagonists was unrelated to antagonism of PAF receptors. However, the studies with ethanol-induced damage demonstrated that Ro-193704 and BN-52021 were without protective effects in that model. Conversely, CV-3988 significantly reduced ethanol-induced damage, suggesting that this compound has additional actions unrelated to antagonism of PAF receptors. These experiments also suggest that PAF release does not play a role in the mechanism of ethanol-induced gastric damage. The leakage of 1251-labeled albumin into the lumen of the stomach and small intestine was another parameter in which there was remarkable similarity between the PAF and endotoxin models of damage. It is not clear if this leakage was a factor contributing to the tissue damage or purely a consequence of it. The present studies, however, demonstrated that pretreatment with a PAF antagonist, CV-3988, resulted in significant inhibition of such plasma leakage in both models. We have suggested that the substantial hemoconcentration that follows PAF administration, as a consequence of increased vascular permeability, is likely to contribute to the slowing and stasis of mucosal blood flow that precedes necrosis (23,24). Such hemoconcentration can be significantly reduced by pretreatment with corticosteroids, such as prednisolone, as well as by the dual cyclooxygenase/lipoxygenase inhibitor, BW755c, which also inhibit PAF-induced gastric damage (24). Using a similar model of endotoxic shock, Doebber et al. (8) demonstrated the appearance of PAF in the blood. Platelet-activating factor was not detectable in blood samples taken 1.5 min after intravenous administration of E. coli endotoxin (50 mg/kg). At such a time, the second phase of the hypotension induced by endotoxin would not yet have occurred. In samples taken 10 min after endotoxin, however, they found PAF at concentrations as high as 4.7 nM. Doebber et al. calculated that such a concentration could be achieved by an intravenous injection of 300 pmol/kg of PAF. In the present study, PAF was infused at a dose of -200 pmobkg. Hence, the dose
October 1987
of PAF used in the present study should have given blood levels in the range of those that have been detected in blood 10 min after endotoxin administration. Other studies have also demonstrated the release of PAF after endotoxin administration (25-27). The similarity in the severity of gastrointestinal damage, hypotension, and plasma leakage between the PAF and endotoxin models in the present study supports the hypothesis that similar blood levels of PAF were achieved under either condition. The present study strongly supports a role for PAF as a mediator of endotoxic shock in the rat. The findings also clearly implicate the endogenous release of PAF as a mediator of the pathological events leading to gastrointestinal necrosis and hemorrhage in endotoxemia. Whether PAF plays an important role as a mediator of septic shock in humans is not yet known, but is a question that certainly warrants further study.
References 1. Robert
2. 3.
4.
5.
6.
7.
6.
9.
10.
A, Kauffman GL. Stress ulcers. In: Sleisenger MH, Fordtran JS, eds. Gastrointestinal disease. Philadelphia: WB Saunders, 1983:612-25. Shumer W. Septic shock. JAMA 1979;242:1906-7. Collier JG, Herman AG, Vane JR. Appearance of prostaglandins in renal venous blood of dogs in response to acute systemic hypotension produced by bleeding or endotoxin. J Physiol (Lond) 1973;230:14-16P. Cook JA, Wise WC, Haluska PV. Elevated thromboxane levels in the rat during endotoxin shock. J Clin Invest 1980;65:22730. Holaday JW, Faden AI. Naloxone reversal of endotoxin hypotension suggests role of endorphins in shock. Nature 1978; 275:450-l. Terashita Z, Imura Y, Nishikawa K, Sumida S. Is platelet activating factor (PAF) a mediator of endotoxin shock? Eur J Pharmacol 1985;109:257-61. Rosam A-C, Wallace JL, Whittle BJR. Potent ulcerogenic actions of platelet-activating factor on the stomach. Nature 1986;319:54-6. Doebber TW, Wu MS, Robbins JC, Choy BM, Chang MN, Shen TY. Platelet activating factor (PAF) involvement in endotoxin-induced hypotension in rats. Studies with PAF-receptor antagonist kadsurenone. Biochem Biophys Res Commun 1985;127:799-808. Braquet P, Touqui L, Shen TY, Vargaftig BB. Perspectives in platelet activating factor research. Pharmacol Rev 1987;39: 97-145. Wallace JL, Whittle BJR. Picomole doses of platelet-activating factor predispose the gastric mucosa to damage by topical irritants. Prostaglandins 1986;31:989-98.
PAF AND SHOCK-INDUCED ULCERATION
773
11. Terashita Z, Tsushima S, Yoshioka Y, Nomura H, Inada Y, Mishikawa K. CV-3988-a specific antagonist of platelet activating factor (PAF). Life Sci 1983;32:1975-82. 12. Braquet P, Spinnewyn B, Braquet M, et al. BN52021 and related compounds: a new series of highly specific PAFacether receptor antagonists isolated from Gingko biloba. Blood Vessels 1985;16:558-72. of PAF-acether 13. Hardvary P, Baumgartner H. Interference antagonists with platelet aggregation and with the formation of platelet thrombi. Prostaglandins 1985;30:694. damage 14. Wallace JL, Whittle BJR. Profile of gastrointestinal induced by platelet-activating factor. Prostaglandins 1986;32: 137-41. 15. Wallace JL, Whittle BJR. Acceleration of recovery of epithelial integrity by 16,16-dimethyl prostaglandin E2. Br J Pharmacol 1985;86:837-42. 16. Morris GP, Wallace JL. The roles of ethanol and of acid in the production of gastric mucosal erosions in rats. Virchows Arch [B] 1981;38:23-38. 17. Lacy ER, Ito S. Microscopic analysis of ethanol damage to rat gastric mucosa after treatment with a prostaglandin. Gastroenterology 1982;82:619-25. 18. Litton A. Gram-negative septicaemia in surgical practice. Br J Surg 1975;62:773-6. 19. Bessin P, Bonnet J, Thiboudeau D, Agier B, Beaudet Y, Gilet F. Pathophysiology of shock states caused by PAF-acether in dogs and cats. In: Benveniste J, Arnoux B, eds. Plateletactivating factor. INSERM Symposium No. 23. Amsterdam: Elsevier, 1983:343-56. 20. Gonzalez-Crussi F, Hsueh W. Experimental model of ischemit bowel necrosis. The role of platelet-activating factor and endotoxin. Am J Path01 1983;112:127-35. 21. Heflin AC, Brigham KL. Prevention by granulocyte depletion of increased vascular permeability of sheep lung following endotoxemia. J Clin Invest 1981;68:1253-60. F. Inhibitory effect of 22. Camussi G, Tetta C, Bussolino prostacyclin (PGIZ) on neutropenia induced by intravenous injection of platelet-activating factor (PAF) in the rabbit. Prostaglandins 1983;25:343-51. 23. Whittle BJR, Morishita T, Ohya Y, Leung F, Guth PH. Microvascular actions of platelet-activating factor (PAF) on the rat gastric mucosa and submucosa. Am J Physiol1986;251:G7729. 24. Wallace JL, Whittle BJR. Effects of inhibitors of arachidonic acid metabolism on Paf-induced gastric mucosal damage and haemoconcentration. Br J Pharmacol 1986;89:415-22. 25. Inarrea P, Gomez-Cambronero J, Pascual J. Carmen Ponte M, Hernando L, Sanchez-Crespo M. Synthesis of PAF-acether and blood volume changes in gram-negative sepsis. Immunopharmacology 1985;9:45-52. 26. Beijer L, Rylander R. Endotoxin inhalation induces alveolar macrophage production of Paf-acether. Am Rev Resp Dis 1984;129:A154. 27. Lagente V, Lidbury P, Steel G, Vargaftig BB, Wallace JL, Whittle BJR. Role of PAF as a mediator of endotoxin-induced gastrointestinal damage (abstr). Br J Pharmacol 1987;90:14OP.