Int. 1. Exp. Path. (I990) 7I, 433-440
Protection against endotoxin-induced foetal resorption in mice by desferrioxamine and ebselen J.D. Gower,* R.J. Baldock,t A.M. O'Sullivan,t C.J. Dore,,t C.R. Coidt and C.J. Green* Sections of * Surgical Research, t Comparative Biology and f Medical Statistics, Clinical Research Centre, Harrow, Middlesex, UK
Received for publication I 9 May I 9 8 9 Accepted for publication I 8 February I990
Summary. Endotoxin was administered to mice on their 1 3th day of pregnancy at doses which caused the resorption of approximately 50% of the implanted foetuses. The iron chelator desferrioxamine was found to significantly inhibit the percentage of resorptions induced by endotoxin in a dose-dependent manner. The highest dose of desferrioxamine (5 mg) given intravenously 30 min prior to, immediately after, and 4 and 24 h after endotoxin inoculation, reduced the percentage of resorptions from 56.9 to 17.9%. Administration of the novel selenium-containing compound ebselen, which is both an antioxidant and an inhibitor of leukotriene synthesis, was also found to significantly protect against endotoxin-induced foetal resorptions, reducing the percentage of resorbed foetuses from 5 2.9 to 26.0% when given at a dose of 50 mg/kg (s.c.) at the time of endotoxin inoculation and 24 and 48 h following. Both these compounds also significantly reduced the increase in spleen weights observed when the mice were given endotoxin. These results provide evidence that the iron-catalysed production of hydroxyl radicals from other oxygen-derived species and the formation of leukotrienes play an important role in the mechanism by which endotoxin causes foetal resorptions in the mouse.
Keywords: endotoxin, foetal resorption, desferrioxamine, ebselen, free radicals, iron
Endotoxins of Gram-negative bacteria elicit a wide variety of pathological events in mammals. One such effect is the interruption of pregnancy and the impairment of foetal development (Zahl & Bjerknes 1943; Dennis I966). Administration of small doses of Escherichia coli endotoxin to mice on the I 3th day of pregnancy results in the resorption of some of the foetuses whilst the rest
appear to develop normally (Coid I976). Endotoxins exert dramatic effects on vascular tissue and it is likely that their abortifacient activity is due to damage of the placental capillary bed, a tissue which is known to be highly sensitive to the effects of endotoxins (Zahl & Bjerknes 1943). Endotoxins trigger the production of a diverse spectrum of inflammatory mediators (Morrison &
Correspondence: Dr J.D. Gower, Section of Surgical Research, Clinical Research Centre, Watford Road, Harrow, Middlesex HAI 3UJ, UK.
433
J.D. Gower et al. 434 Ryan I987) and activate the arachidonic ham et al. I987; Kunimoto et al. I987), acid cascade which leads to the formation of including the mouse (Peavy & Fairchild thromboxane and a variety of prosta- I986), and some protection against endoglandins and leukotrienes (LTs) (Flohe & toxin-induced lesions by antioxidants has Giertz I987). These latter products, formed been demonstrated in a number of models via the enzyme lipoxygenase, are powerful (Demling et al. I986; Ghezzi et al. I986; inflammatory agents (Lefer I985) which Brigham et al. I987; Kunimoto et al. I987). These previous reports led us to investigate enhance capillary permeability; LTC4 and LTD4 are potent vaso and broncho-constric- the effects two compounds had in preventing tors and LTB4 is one of the most potent foetal resorptions induced by endotoxin in chemoattractants of polymorphonuclear the mouse. Firstly, ebselen (which is a novel leucocytes (PMNs). Increased secretion of selenium-containing compound (Parnham LTs into the bile has been demonstrated & Graf I987)) was studied. This compound following endotoxin treatment to rats (Hag- inhibits lipoxygenase and, because of its mann et al. I985) and a role for LTs in glutathione peroxidase-like activity, acts as endotoxin toxicity has been strongly sug- an antioxidant by degrading hydrogen pergested from studies in which inhibitors of LT oxide and organic hydroperoxides formed synthesis afforded protection against endo- from the free-radical-mediated oxidation of toxin damage in a number of models polyunsaturated fatty acids. Recently it has been shown to protect mice from galactos(Keppler et al. I 98 7). The accumulation and activation of amine/endotoxin-induced hepatitis which phagocytic cells by endotoxin either directly was regarded to be due to its ability to inhibit or via the induction of other mediators such leukotriene formation (Wendel & Tiegs I986). Desferrioxamine, which is a powerful as interleukin-i (Klempner et al. i979), complement (Sacks et al. I978) or LTs iron-chelating agent used clinically for the (Schade I986) results in the generation of treatment of iron overload (Keberle I964), reactive oxygen-derived free radicals. The was also used in this investigation. This superoxide anion (02 -) is produced by a siderophore inhibits lipid peroxidation by membrane-bound NADPH-oxidase and dis- sequestering catalytic iron, thus preventing mutates in the acidic environment of the the formation of hydroxyl radicals via the phagocytic vacuole into H202 (Flohe & Giertz Haber-Weiss reaction. I987). In the presence of transition-metal catalysts, such as iron, these species undergo Materials and methods a Haber-Weiss reaction to form the highly damaging hydroxyl radical (OH -). This reac- An outbred strain of mice (TO/Crc) weighing 20-30 g were obtained from a barriertive species attacks proteins and nucleic acids, and initiates the peroxidation of mem- maintained (SPF) breeding unit at this brane-bound polyunsaturated fatty acids, a centre. The first day of pregnancy was taken to be that on which the vaginal plug was chain reaction termed lipid peroxidation (Wills I966). Oxygen-derived free radicals observed. On the I 3th day of pregnancy, the produced in the 'respiratory burst' are mice were inoculated with endotoxin (E. coli known to damage endothelial cells (Sacks et 01 2 7: B8; Difco Laboratories) either alone or al. 1978; Freeman et al. I983) and this may in combination with desferrioxamine (Desferal; Ciba-Geigy Ltd, Basle, Switzerland) or therefore be an important factor in the ebselen (a gift from Nattermann & Cie. mechanism of endotoxin toxicity. Evidence GmbH; Cologne, FRG) as described below. for elevated free-radical activity, measured The effect of desferrioxamine on endoas an increase in lipid peroxidation, has been toxin-induced foetal resorptions was examfound following endotoxin administration to ined using six groups of mice as follows: several species (Demling et al. I986; Brig-
Protection against endotoxin-inducedfoetal resorption in mice 435 GPi: i8 Mug of endotoxin in o.i ml of removed and opened to reveal the number of phosphate-buffered saline (PBS) was resorbed and healthy foetuses. Spleens were administered, i.v., on the 13th day of also removed, weighed and fixed in io% pregnancy. formal saline. Three sections from each Gp2: As group i but with i.p. administration spleen were prepared and stained with haeof desferrioxamine (2 mg) in PBS (o. i matoxylin and eosin for histological examinml) 30 min prior to, immediately after, ation. and 4 h after inoculation of the endoThe data were analysed by factorial analytoxin. sis of variance for the ebselen experiment and Gp3: As group 2 but with a further adminis- by one-way analysis of variance and orthogtration of desferrioxamine (2 mg) 24 h onal contrasts for the desferrioxamine after inoculation of the endotoxin. experiment. As these methods require the Gp4: As group 3 but the desferrioxamine variances of the groups to be similar, the data dose was increased to 5 mg on each of were transformed, where necessary, either the four occasions. by taking logarithms to the base e or by the As group 4 that the except inoculation logit transformation. Gp5: of endotoxin was omitted so that the possible effect of desferrioxamine itself Results on the number of resorptions could be assessed. A series of initial experiments were conGp6: This was the control group in which ducted to determine the dose of E. coli the mice were injected with the PBS endotoxin which resulted in the resorption of vehicle only on each of the four occa- approximately 50% of the implanted foetsions. uses. This was established to be I 8 Mg/mouse when endotoxin was administered i.v., the The effect ofebselen on endotoxin-induced route chosen when the effect of desferrioxafoetal resorptions was investigated by divid- mine was being studied. In good agreement ing the mice into four groups: with previous findings (Coid I 9 76), s.c. GpA: Endotoxin (0.147 mg) in DMSO (50 administration of 147 Mg endotoxin/mouse in the resorption of half the pl) was administered, s.c., to mice on resulted implanted foetuses; this route of administrathe I 3th day of pregnancy followed by tion was chosen when the effect of ebselen injection of the DMSO vehicle, s.c., 24 was studied as it allowed both combeing and 48 h later. in a single injection. to be pounds given mixture of endotoxin GpB: A (0.147 mg) and ebselen (50 mg/kg) in DMSO was administered, s.c., to mice on the I 3th The effect of desferrioxamine day of pregnancy followed by further administration of ebselen at the same As desferrioxamine is rapidly metabolized dose and route, 24 and 48 h after the and excreted in the mouse, several injections of desferrioxamine were given. Administrafirst inoculation. GpC: Animals were administered ebselen as tion 30 min prior to, immediately after, and 4 h after endotoxin treatment (group 2) in group B but with no endotoxin. decreased the number of endotoxin-induced GpD: Animals were given the DMSO vehicle foetal to a certain extent but a resorptions only on three occasions as described more was observed in group effect profound for group A. 3 which were given a further dose of desferOn the i 6th day of pregnancy, each rioxamine 24 h after endotoxin administraanimal was killed by CO2 inhalation. The tion (Table i). Increasing the dose of desferabdomen was opened and the uterus was rioxamine from four injections of 2 mg to
J.D. Gower et al.
436
Table i. The effect of desferrioxamine on endotoxin-induced foetal resorption Percentage resorptions* Group
Endo
Des
n
Mean
95% confidence limits
I 2 3
+ +
+
94 23
+
++
25
46.5-66.8 28.8-68.8 14.6-46.5
4
+
+++
55
5 6
-
+++
-
-
42 23
56.9 48.6 27.8 I7.9 7.9 8.o
I1.2-27.4
6.4- 9.7 6.o-IO.5
Endo, endotoxin; Des, desferrioxamine; n, number of animals. * Values are back transformed (logit) percentages calculated for each animal. - Administration of vehicle only. + 3 inoculations of desferrioxamine (2 mg). + + 4 inoculations of desferrioxamine (2 mg). + + + 4 inoculations of desferrioxamine (5 mg).
four injections of 5 mg (group 4) resulted in further protection against endotoxin and the percentage of implanted foetuses which were resorbed fell to I 7.9 compared with 56.9 in animals given endotoxin only (group i) (Table I). Comparison of the average of all the groups given desferrioxamine and endotoxin (groups 2, 3 and 4) with that given endotoxin alone (group i) demonstrated a highly significant (P= 0.0003) protective effect of desferrioxamine in this system. Furthermore, there was a significant (P=o.oo5) trend of desferrioxamine dose and the degree of protection afforded against endotoxin-induced resorptions. There was a low rate of spontaneous resorptions (animals given neither endotoxin nor desferrioxamine; group 6) and this was not altered significantly (P= o.98) by giving high doses of desferrioxamine (group 5) which demonstrated that desferrioxamine by itself did not affect the development of the foetuses. The administration of endotoxin to pregnant mice (group i) resulted in a large increase in spleen weight as compared to the control group (group 6) as shown in Table 2. Examination by light microscopy of speci-
mens stained with haematoxylin and eosin showed that these increases in spleen weight were associated with hypertrophy of both white and red pulp accompanied by dilated sinusoids filled with eosinophilic material and by accumulation of numerous deeply staining monocytes in the red pulp. Administration of desferrioxamine by itself had no significant (P= o.6o) effect on spleen (group 5 vs group 6); however, when both endotoxin and desferrioxamine were given together, there was a highly significant (P
Protection against endotoxin-induced foetal resorption in mice by desferrioxamine and ebselen J.D. Gower,* R.J. Baldock,t A.M. O'Sullivan,t C.J. Dore,,t C.R. Coidt and C.J. Green* Sections of * Surgical Research, t Comparative Biology and f Medical Statistics, Clinical Research Centre, Harrow, Middlesex, UK
Received for publication I 9 May I 9 8 9 Accepted for publication I 8 February I990
Summary. Endotoxin was administered to mice on their 1 3th day of pregnancy at doses which caused the resorption of approximately 50% of the implanted foetuses. The iron chelator desferrioxamine was found to significantly inhibit the percentage of resorptions induced by endotoxin in a dose-dependent manner. The highest dose of desferrioxamine (5 mg) given intravenously 30 min prior to, immediately after, and 4 and 24 h after endotoxin inoculation, reduced the percentage of resorptions from 56.9 to 17.9%. Administration of the novel selenium-containing compound ebselen, which is both an antioxidant and an inhibitor of leukotriene synthesis, was also found to significantly protect against endotoxin-induced foetal resorptions, reducing the percentage of resorbed foetuses from 5 2.9 to 26.0% when given at a dose of 50 mg/kg (s.c.) at the time of endotoxin inoculation and 24 and 48 h following. Both these compounds also significantly reduced the increase in spleen weights observed when the mice were given endotoxin. These results provide evidence that the iron-catalysed production of hydroxyl radicals from other oxygen-derived species and the formation of leukotrienes play an important role in the mechanism by which endotoxin causes foetal resorptions in the mouse.
Keywords: endotoxin, foetal resorption, desferrioxamine, ebselen, free radicals, iron
Endotoxins of Gram-negative bacteria elicit a wide variety of pathological events in mammals. One such effect is the interruption of pregnancy and the impairment of foetal development (Zahl & Bjerknes 1943; Dennis I966). Administration of small doses of Escherichia coli endotoxin to mice on the I 3th day of pregnancy results in the resorption of some of the foetuses whilst the rest
appear to develop normally (Coid I976). Endotoxins exert dramatic effects on vascular tissue and it is likely that their abortifacient activity is due to damage of the placental capillary bed, a tissue which is known to be highly sensitive to the effects of endotoxins (Zahl & Bjerknes 1943). Endotoxins trigger the production of a diverse spectrum of inflammatory mediators (Morrison &
Correspondence: Dr J.D. Gower, Section of Surgical Research, Clinical Research Centre, Watford Road, Harrow, Middlesex HAI 3UJ, UK.
433
J.D. Gower et al. 434 Ryan I987) and activate the arachidonic ham et al. I987; Kunimoto et al. I987), acid cascade which leads to the formation of including the mouse (Peavy & Fairchild thromboxane and a variety of prosta- I986), and some protection against endoglandins and leukotrienes (LTs) (Flohe & toxin-induced lesions by antioxidants has Giertz I987). These latter products, formed been demonstrated in a number of models via the enzyme lipoxygenase, are powerful (Demling et al. I986; Ghezzi et al. I986; inflammatory agents (Lefer I985) which Brigham et al. I987; Kunimoto et al. I987). These previous reports led us to investigate enhance capillary permeability; LTC4 and LTD4 are potent vaso and broncho-constric- the effects two compounds had in preventing tors and LTB4 is one of the most potent foetal resorptions induced by endotoxin in chemoattractants of polymorphonuclear the mouse. Firstly, ebselen (which is a novel leucocytes (PMNs). Increased secretion of selenium-containing compound (Parnham LTs into the bile has been demonstrated & Graf I987)) was studied. This compound following endotoxin treatment to rats (Hag- inhibits lipoxygenase and, because of its mann et al. I985) and a role for LTs in glutathione peroxidase-like activity, acts as endotoxin toxicity has been strongly sug- an antioxidant by degrading hydrogen pergested from studies in which inhibitors of LT oxide and organic hydroperoxides formed synthesis afforded protection against endo- from the free-radical-mediated oxidation of toxin damage in a number of models polyunsaturated fatty acids. Recently it has been shown to protect mice from galactos(Keppler et al. I 98 7). The accumulation and activation of amine/endotoxin-induced hepatitis which phagocytic cells by endotoxin either directly was regarded to be due to its ability to inhibit or via the induction of other mediators such leukotriene formation (Wendel & Tiegs I986). Desferrioxamine, which is a powerful as interleukin-i (Klempner et al. i979), complement (Sacks et al. I978) or LTs iron-chelating agent used clinically for the (Schade I986) results in the generation of treatment of iron overload (Keberle I964), reactive oxygen-derived free radicals. The was also used in this investigation. This superoxide anion (02 -) is produced by a siderophore inhibits lipid peroxidation by membrane-bound NADPH-oxidase and dis- sequestering catalytic iron, thus preventing mutates in the acidic environment of the the formation of hydroxyl radicals via the phagocytic vacuole into H202 (Flohe & Giertz Haber-Weiss reaction. I987). In the presence of transition-metal catalysts, such as iron, these species undergo Materials and methods a Haber-Weiss reaction to form the highly damaging hydroxyl radical (OH -). This reac- An outbred strain of mice (TO/Crc) weighing 20-30 g were obtained from a barriertive species attacks proteins and nucleic acids, and initiates the peroxidation of mem- maintained (SPF) breeding unit at this brane-bound polyunsaturated fatty acids, a centre. The first day of pregnancy was taken to be that on which the vaginal plug was chain reaction termed lipid peroxidation (Wills I966). Oxygen-derived free radicals observed. On the I 3th day of pregnancy, the produced in the 'respiratory burst' are mice were inoculated with endotoxin (E. coli known to damage endothelial cells (Sacks et 01 2 7: B8; Difco Laboratories) either alone or al. 1978; Freeman et al. I983) and this may in combination with desferrioxamine (Desferal; Ciba-Geigy Ltd, Basle, Switzerland) or therefore be an important factor in the ebselen (a gift from Nattermann & Cie. mechanism of endotoxin toxicity. Evidence GmbH; Cologne, FRG) as described below. for elevated free-radical activity, measured The effect of desferrioxamine on endoas an increase in lipid peroxidation, has been toxin-induced foetal resorptions was examfound following endotoxin administration to ined using six groups of mice as follows: several species (Demling et al. I986; Brig-
Protection against endotoxin-inducedfoetal resorption in mice 435 GPi: i8 Mug of endotoxin in o.i ml of removed and opened to reveal the number of phosphate-buffered saline (PBS) was resorbed and healthy foetuses. Spleens were administered, i.v., on the 13th day of also removed, weighed and fixed in io% pregnancy. formal saline. Three sections from each Gp2: As group i but with i.p. administration spleen were prepared and stained with haeof desferrioxamine (2 mg) in PBS (o. i matoxylin and eosin for histological examinml) 30 min prior to, immediately after, ation. and 4 h after inoculation of the endoThe data were analysed by factorial analytoxin. sis of variance for the ebselen experiment and Gp3: As group 2 but with a further adminis- by one-way analysis of variance and orthogtration of desferrioxamine (2 mg) 24 h onal contrasts for the desferrioxamine after inoculation of the endotoxin. experiment. As these methods require the Gp4: As group 3 but the desferrioxamine variances of the groups to be similar, the data dose was increased to 5 mg on each of were transformed, where necessary, either the four occasions. by taking logarithms to the base e or by the As group 4 that the except inoculation logit transformation. Gp5: of endotoxin was omitted so that the possible effect of desferrioxamine itself Results on the number of resorptions could be assessed. A series of initial experiments were conGp6: This was the control group in which ducted to determine the dose of E. coli the mice were injected with the PBS endotoxin which resulted in the resorption of vehicle only on each of the four occa- approximately 50% of the implanted foetsions. uses. This was established to be I 8 Mg/mouse when endotoxin was administered i.v., the The effect ofebselen on endotoxin-induced route chosen when the effect of desferrioxafoetal resorptions was investigated by divid- mine was being studied. In good agreement ing the mice into four groups: with previous findings (Coid I 9 76), s.c. GpA: Endotoxin (0.147 mg) in DMSO (50 administration of 147 Mg endotoxin/mouse in the resorption of half the pl) was administered, s.c., to mice on resulted implanted foetuses; this route of administrathe I 3th day of pregnancy followed by tion was chosen when the effect of ebselen injection of the DMSO vehicle, s.c., 24 was studied as it allowed both combeing and 48 h later. in a single injection. to be pounds given mixture of endotoxin GpB: A (0.147 mg) and ebselen (50 mg/kg) in DMSO was administered, s.c., to mice on the I 3th The effect of desferrioxamine day of pregnancy followed by further administration of ebselen at the same As desferrioxamine is rapidly metabolized dose and route, 24 and 48 h after the and excreted in the mouse, several injections of desferrioxamine were given. Administrafirst inoculation. GpC: Animals were administered ebselen as tion 30 min prior to, immediately after, and 4 h after endotoxin treatment (group 2) in group B but with no endotoxin. decreased the number of endotoxin-induced GpD: Animals were given the DMSO vehicle foetal to a certain extent but a resorptions only on three occasions as described more was observed in group effect profound for group A. 3 which were given a further dose of desferOn the i 6th day of pregnancy, each rioxamine 24 h after endotoxin administraanimal was killed by CO2 inhalation. The tion (Table i). Increasing the dose of desferabdomen was opened and the uterus was rioxamine from four injections of 2 mg to
J.D. Gower et al.
436
Table i. The effect of desferrioxamine on endotoxin-induced foetal resorption Percentage resorptions* Group
Endo
Des
n
Mean
95% confidence limits
I 2 3
+ +
+
94 23
+
++
25
46.5-66.8 28.8-68.8 14.6-46.5
4
+
+++
55
5 6
-
+++
-
-
42 23
56.9 48.6 27.8 I7.9 7.9 8.o
I1.2-27.4
6.4- 9.7 6.o-IO.5
Endo, endotoxin; Des, desferrioxamine; n, number of animals. * Values are back transformed (logit) percentages calculated for each animal. - Administration of vehicle only. + 3 inoculations of desferrioxamine (2 mg). + + 4 inoculations of desferrioxamine (2 mg). + + + 4 inoculations of desferrioxamine (5 mg).
four injections of 5 mg (group 4) resulted in further protection against endotoxin and the percentage of implanted foetuses which were resorbed fell to I 7.9 compared with 56.9 in animals given endotoxin only (group i) (Table I). Comparison of the average of all the groups given desferrioxamine and endotoxin (groups 2, 3 and 4) with that given endotoxin alone (group i) demonstrated a highly significant (P= 0.0003) protective effect of desferrioxamine in this system. Furthermore, there was a significant (P=o.oo5) trend of desferrioxamine dose and the degree of protection afforded against endotoxin-induced resorptions. There was a low rate of spontaneous resorptions (animals given neither endotoxin nor desferrioxamine; group 6) and this was not altered significantly (P= o.98) by giving high doses of desferrioxamine (group 5) which demonstrated that desferrioxamine by itself did not affect the development of the foetuses. The administration of endotoxin to pregnant mice (group i) resulted in a large increase in spleen weight as compared to the control group (group 6) as shown in Table 2. Examination by light microscopy of speci-
mens stained with haematoxylin and eosin showed that these increases in spleen weight were associated with hypertrophy of both white and red pulp accompanied by dilated sinusoids filled with eosinophilic material and by accumulation of numerous deeply staining monocytes in the red pulp. Administration of desferrioxamine by itself had no significant (P= o.6o) effect on spleen (group 5 vs group 6); however, when both endotoxin and desferrioxamine were given together, there was a highly significant (P
