Article Text

PDF

TNF-α dependent production of inducible nitric oxide is involved in PGE1 protection against acute liver injury
  1. J Muntanéa,
  2. F J Rodrígueza,
  3. O Segadoa,
  4. A Quinteroa,
  5. J M Lozanoa,
  6. E Siendonesa,
  7. C A Pedrazaa,
  8. M Delgadoa,
  9. F O'Valleb,
  10. R Garcíab,
  11. J L Monteroa,
  12. M De la Mataa,
  13. G Miñoa
  1. aServicio Aparato Digestivo, Hospital Universitario Reina Sofía, Córdoba, España, bDepartamento Anatomía Patológica, Facultad de Medicina, Granada, España
  1. Dr J Muntané Relat, Unidad de Investigación, Servicio Aparato Digestivo, Hospital Universitario Reina Sofía, Av. Menendez Pidal s/n, 14004 Córdoba, Spain. Email. jmuntane{at}sofia.hrs.sas.cica.es

Abstract

BACKGROUND Tumour necrosis factor α (TNF-α) and nitric oxide modulate damage in several experimental models of liver injury. We have previously shown that protection against d-galactosamine (D-GalN) induced liver injury by prostaglandin E1 (PGE1) was accompanied by an increase in TNF-α and nitrite/nitrate in serum.

AIMS The aim of the present study was to evaluate the role of TNF-α and nitric oxide during protection by PGE1 of liver damage induced by D-GalN.

METHODS Liver injury was induced in male Wistar rats by intraperitoneal injection of 1 g/kg of D-GalN. PGE1 was administered 30 minutes before D-GalN. Inducible nitric oxide synthase (iNOS) was inhibited by methylisothiourea (MT), and TNF-α concentration in serum was lowered by administration of anti-TNF-α antibodies. Liver injury was evaluated by alanine aminotransferase activity in serum, and histological examination and DNA fragmentation in liver. TNF-α and nitrite/nitrate concentrations were determined in serum. Expression of TNF-α and iNOS was also assessed in liver sections.

RESULTS PGE1decreased liver injury and increased TNF-α and nitrite/nitrate concentrations in serum of rats treated with D-GalN. PGE1protection was related to enhanced expression of TNF-α and iNOS in hepatocytes. Administration of anti-TNF-α antibodies or MT blocked the protection by PGE1 of liver injury induced by D-GalN.

CONCLUSIONS This study suggests that prior administration of PGE1 to D-GalN treated animals enhanced expression of TNF-α and iNOS in hepatocytes, and that this was causally related to protection by PGE1against D-GalN induced liver injury.

  • tumour necrosis factor α
  • nitric oxide
  • prostaglandin E1
  • methylisothiourea
  • d-galactosamine
  • liver injury

Statistics from Altmetric.com

Hepatic injury induced by d-galactosamine (D-GalN) is a suitable experimental model of liver failure.1 D-GalN induces inhibition of RNA and protein synthesis thus altering hepatocellular function.2 Both effects may be a consequence of a depleted intracellular pool of uracil nucleotides.2 Several studies have reported that prostaglandin E reduces hypertransaminasaemia observed during in vivo hepatotoxicity induced by D-GalN,3thioacetamide,4 aflatoxin B1,5carbon tetrachloride,6 bile duct ligation,7fat enriched and choline deficient diets,8 viral hepatitis,9 and complement mediated hepatic necrosis.10 Furthermore, prostaglandin E1(PGE1) has a beneficial effect on fulminant viral hepatitis in humans with a decrease in levels of transaminases and improvement in encephalopathy and coagulation factors.11 12 Prostacyclin also reduces serum transaminases in hepatic necrosis induced by D-GalN in rats.13 Several hypotheses have been presented to explain this protective effect of prostaglandins. It has been suggested that protection by PGE1 against liver injury induced by ischaemia-reperfusion14 or bile duct ligation7 may occur at the vascular level. PGE1 reduces plasma membrane microviscosity induced by lipid peroxidation in rat hepatocytes, suggesting that it has a stabilising effect.15 Marinovich and colleagues16 have shown that prostaglandin E2modifies the arachidonic acid cascade profile after activation of phospholipase A2 by carbon tetrachloride in rat hepatocyte cultures. The antifibrotic activity of prostaglandin E2 has also been proved.8

Tumour necrosis factor α (TNF-α) is a multifunctional cytokine mostly secreted by inflammatory cells and is involved in numerous pathological states.17 In particular, it has been shown that TNF-α enhances liver injury when D-GalN toxicity is already established.18 Nevertheless, D-GalN treated animals can be protected from an otherwise lethal challenge of either lipopolysaccharide or TNF-α toxicity by pretreatment with a sublethal dose of TNF-α.19

Nitric oxide is a labile and highly reactive compound that plays a physiological role in blood pressure regulation, neurotransmission, tumour cell killing, immunity, and inflammatory processes.20 Two constitutive forms of nitric oxide synthase (cNOS) (eNOS in endothelial cells and nNOS in neurones) and one inducible form (iNOS) (in hepatocytes, Kupffer cells, macrophages, fibroblasts, chondrocytes, and endothelial cells) have been described.21 Expression of iNOS is modulated by cytokines secreted during inflammation.22 The cytoprotection or cytotoxicity of nitric oxide has been reported by several authors.23 24 Administration of a nitric oxide donor induces tolerance against a lethal combination of TNF-α and D-GalN treatment.25

In a previous study26 we observed that protection by PGE1 against D-GalN induced liver injury was associated with an increase in TNF-α and nitrite/nitrate in serum. Prior administration of both inflammatory mediators has been shown to induce tolerance against lipopolysaccharide or cytokine induced liver toxicity.19 25 The aim of the present study was to evaluate if TNF-α and/or nitric oxide participate in the protective effect of PGE1 on D-GalN induced liver injury in rats.

Materials and methods

EXPERIMENTAL DESIGN

Male Wistar rats (175–225 g) were kept under standard conditions and received a regular diet (Piensos equilibrados, CIA, Ebro Agricolas, Utrera, Spain) and water ad libitum. Animal care and experimental protocols were in accordance with the Guide for the Care and Use of Laboratory Animals prepared by the National Academy of Sciences and published by the National Institute of Health (NIH publication No 86–23, revised 1985).

To evaluate the involvement of iNOS on D-GalN injury and its protection by PGE1, animals (n=80) were divided into eight groups: control; PGE1; D-GalN; PGE1+D-GalN; methylisothiourea (MT); MT+PGE1; MT+D-GalN; and MT+PGE1+D-GalN. D-GalN (Sigma Chemical Corporation, St Louis, Missouri, USA) was dissolved in 0.9% NaCl just before intraperitoneal administration (1 g/kg). MT (Sigma Chemical Corporation) was used as a highly specific inhibitor of iNOS.27 The lowest dose and pattern of administration of MT that effectively blocked the production of nitric oxide induced by lipopolysaccharide (10 mg/kg) were determined in a preliminary study (unpublished data). MT was dissolved in 0.9% NaCl and administered (25 mg/kg) intraperitoneally 30 minutes and one hour before D-GalN, and four and eight hours after D-GalN administration. PGE1 was dissolved in 10% ethanol and administered (250 μg/kg) intraperitoneally 30 minutes before D-GalN. The amount (100 μl) of 10% ethanol used had no effect on the variables of the study in the presence or absence of hepatotoxin. Accordingly, the control group received the corresponding amount of 0.9% NaCl intraperitoneally. The dose of PGE1 administered in the present study was within the range used in several experimental studies4 5 8 and during clinical trials.11After administration of the treatments, rats were maintained in a fasted state with access to water ad libitum. The animals were killed 12 hours after D-GalN administration. Those animals who did not receive D-GalN were killed 12 hours after PGE1 or the second administration of MT. The experimental design scheme is shown diagrammatically in fig 1A. Blood samples were obtained by cardiac puncture and serum was frozen for measurement of alanine aminotransferase (ALT), TNF-α, and nitrite/nitrate levels. Endotoxin content was also measured in serum from control and D-GalN treated animals.

Figure 1

Experimental design scheme. (A) Pattern of administration of methylisothiourea (MT), prostaglandin E1(PGE1), andd-galactosamine (D-GalN). (B) Pattern of administration of anti-tumour necrosis factor α (anti-TNF-α) antibodies, PGE1, and D-GalN.

To evaluate the involvement of TNF-α in D-GalN induced liver injury and its protection by PGE1, animals (n=40) were divided into eight groups: control; PGE1; D-GalN; PGE1+D-GalN; anti-TNF-α+control; anti-TNF-α+ PGE1; anti-TNF-α+D-GalN; and anti-TNF-α+ PGE1+D-GalN. Polyclonal rabbit antihuman TNF-α antibodies (Genzyme Diagnostics, Cambridge, Massachusetts, USA) administered to animals can neutralise rat and human TNF-α forms, as indicated by the manufacturer. Antibodies were dissolved in phosphate buffered solution (PBS) (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4) at pH 7.4 and administered (5 mg/kg) intraperitoneally one hour before D-GalN. D-GalN and PGE1 were administered under the conditions previously described. The amount of ethanol used, as solvent for PGE1, had no effect on the variables of the study in the presence or absence of the hepatotoxin. Accordingly, the control group received the corresponding amount of 0.9% NaCl intraperitoneally. Non-immunised rabbit IgG (Sigma Chemical Corporation) as a negative control of anti-TNF-α antibodies had no effect on the variables of the study in the presence or absence of hepatotoxin. After administration of the treatments, rats were maintained in a fasted state with access to water ad libitum. The animals were killed 12 hours after D-GalN administration. Those animals who did not receive D-GalN were killed 12 hours after PGE1or anti-TNF-α treatment. The experimental design scheme is shown diagrammatically in fig 1B. Blood samples were obtained by cardiac puncture and serum was frozen for ALT and nitrite/nitrate measurements. Serum samples for evaluation of the efficacy of anti-TNF-α antibodies in reducing TNF-α concentrations were also obtained. For this purpose, serum was filtered through HiTrap Protein A cartridges (Pharmacia Biotech AB, Uppsala, Sweden) to remove antibodies, thus avoiding interference with ELISA. It was observed that TNF-α concentrations in serum were undetectable by ELISA using the dose and pattern of administration of anti-TNF-α antibodies described above (data not shown). Several liver specimens were also removed and fixed in buffered 4% formalin in PBS (Panreac, Barcelona, Spain) and embedded in paraffin. Thin sections (3 μm) were obtained for evaluation of histological damage and expression of TNF-α and iNOS. Other liver specimens were immediately frozen to evaluate DNA fragmentation and TNF-α content.

HISTOLOGICAL EXAMINATION

Liver injury was evaluated in liver fixed sections embedded in paraffin and stained with haematoxylin and eosin.

DNA FRAGMENTATION IN THE LIVER

Detection of the classical laddering of DNA fragmentation in liver extracts is related to apoptosis. Frozen livers (1 g) were homogenised (Ultra-turrax T25; Janke and Kunkel IKA-Laboratory) in lysis solution (1% SDS, 10 mM Tris HCl, 50 mM EDTA), pH 7.4. Samples were incubated with RNAsa (1 mg/ml) (Sigma Chemical Corporation) for two hours at 37°C and proteinase K (1 mg/ml) (Sigma Chemical Corporation) for 45 minutes at 48°C. DNA was obtained by phenol:choloform: isoamyl alcohol 25:24:1 (Sigma Chemical Corporation) extraction and precipitated with isopropanol (v/v) and 0.5 M NaCl for 12 hours at −20°C. DNA was recovered by centrifugation at 15 000g for 20 minutes, and the pellet was washed with 70% ethanol, dried, and resuspended in Tris-EDTA buffer (10 mM Tris, 50 mM EDTA), pH 7.4. Samples (250 μg DNA) were applied and analysed on 2% agarose gel with ethidium bromide (0.5 μg/ml).

BIOCHEMICAL PARAMETERS

Activity of serum ALT was measured using routine laboratory methods28 in samples obtained from experimental animals.

EXPRESSION OF TNF-α IN LIVER

TNF-α content in liver evaluated by western blot

Frozen livers (1 g) were homogenised (Ultra-turrax T25; Janke and Kunkel IKA-Laboratory) in lysis solution (25 mM HEPES, 10 mM MgCl2, 2 mM EGTA, 1 mM PMSF, 1 μg/ml aprotinin, and 1 μg/ml leupeptin), pH 7.5. Polyclonal goat antihuman TNF-α antibodies (R&D Systems, Minneapolis, USA) used in the assay also detected rat TNF-α forms, as indicated by the manufacturer. Recombinant human TNF-α (25 pg) (Genzyme, Diagnostics) was used as standard. Samples (50 μg) were applied to 12% sodium dodecyl sulphate-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Membranes were treated with blocking solution (20 mM Tris HCl, 150 mM NaCl, 0.1% Tween 20, and 5% milk powder), pH 7.6, for one hour and further incubated with primary antibodies for one hour and with alkaline phosphatase linked secondary antibodies (Sigma Chemical Corporation) for one hour. TNF-α was visualised incubating the membranes with the alkaline phosphatase substrate (BCIP-NBT) for 20 minutes. TNF-α content of each sample was determined by quantitative densitometry using the TDI's 1D Manager version 2.0 software included in a Gel Printer Plus System (Tecnología para Diagnóstico e Investigación, Madrid, Spain).

TNF-α immunolocalisation in liver sections

Cell expression of TNF-α was also evaluated by immunohistochemistry in rat liver fixed sections (3 μm) embedded in paraffin, as described above. Deparaffinised sections were treated with 3% H2O2 for 30 minutes at room temperature to eliminate the interference of endogenous peroxidases. Non-specific immunostaining was blocked by treatment with 1% fetal bovine serum and 10% rat serum for one hour at room temperature. TNF-α expression was assessed using polyclonal goat antihuman TNF-α antibodies (R&D Systems) which cross react with rat TNF-α forms, as described by the manufacturer. Liver sections were incubated with primary antibodies (5 μg/ml) for 24 hours at 4°C and with biotinated polyclonal antibody against goat IgG (Zymed, San Francisco, USA) for one hour at 37°C. Samples were incubated with streptavidin-biotin-peroxidase complex (Dako A/S, Copenhagen, Denmark) and visualised with 50 mM diaminobenzidine solution, pH 6.0, and 0.003% H2O2 for eight minutes at room temperature co-stained with haematoxylin.

The degree of immunostaining of TNF-α was evaluated semiquantitatively using four categories: (1) no expression; (2) mild overexpression or less than 25% of hepatocytes stained; (3) moderate overexpression or 25–50% of hepatocytes stained; and (4) intense overexpression or more than 50% of hepatocytes stained.

ENDOTOXIN MEASUREMENT

Endotoxin content was measured in serum obtained from control and D-GalN treated rats using a commercial kit (E-toxate, Sigma Chemical Corporation).

TNF-α MEASUREMENT IN SERUM

TNF-α was quantified in serum by competitive ELISA.26 The following parameters were determined: for reproducibility, interassay precision profile (1.2±0.03%), intra-assay precision profile (2.9±0.90%), and intra-assay coefficient of variation (1.9%); for parallelism, slope ratio (0.94±0.030%); and for sensitivity, minimum detectable dose (0.08 pg/ml). A representative standard curve for the TNF-α ELISA is shown in fig 2. Polyclonal rabbit antihuman TNF-α antibodies (Genzyme Diagnostics) used in the assay were also useful in detecting rat TNF-α forms, as indicated by the manufacturer. Recombinant human TNF-α (Genzyme Diagnostics) used as a standard in the assay was calibrated against the related NIBSC reference standard (87/650). Briefly, wells of ELISA plates were coated with 0.5 ng of TNF-α for one hour at 37°C. The wells were blocked with 2% bovine serum albumin in 10 mM PBS-0.05% Tween (PBS-Tween) at pH 7.4. The samples and anti-TNF-α antibodies (32.5 ng/ml) were incubated for two hours at 37°C, transferred to antigen coated wells, and incubated for two hours at 37°C. The wells were washed with PBS-Tween and incubated with the biotinated secondary antibodies (125 pg/ml) (Zymed ) for one hour at 37°C. The wells were then washed with Tris buffer (100 mM) at pH 7.6 and incubated with streptavidin-alkaline phosphatase solution (Master Diagnóstica, Granada, Spain) for 30 minutes at 37°C. Finally, the wells were washed and incubated withp-nitrophenyl phosphate (Sigma Chemical Corporation) as alkaline phosphatase substrate for one hour at 37°C. Later, the wells were read at 405 nm in a titrated Organon Teknika 510 ELISA reader.

Figure 2

Representative standard curve for the tumour necrosis factor α (TNF-α) ELISA. B/Bo represents the ratio between the absorbance obtained form each TNF-α concentration of the standard curve (B) and the absorbance obtained in the absence of TNF-α antigen (Bo).

MEASUREMENT OF NITRIC OXIDE METABOLITES

Production of nitric oxide was measured by quantification of its related end products, nitrite/nitrate. In the assay, nitrate was converted to nitrite by nitrate reductase (EC 1.6.6.2) and total nitrite was measured using the Griess reaction.29 Briefly, samples were incubated with nitrate reductase (0.2 U/ml), FAD (5 mM), and NADPH (50 mM) for 20 minutes at 37°C. The reaction was stopped by addition of sodium pyruvate (10 mM) and lactate dehydrogenase (24 mg/ml) for five minutes at 37°C, and precipitated with 1.4% ZnSO4. Total nitrite reacted with Griess reagent (1% sulphanilamide, 2.5% PO4H3, 0.1% n-naphthyl- ethylene-diamine) for 10 minutes at 37°C, and was read using the 540 nm filter in a titrated Organon Teknika 510 ELISA reader.

EXPRESSION OF INDUCIBLE NITRIC OXIDE SYNTHASE IN LIVER

Cell expression of iNOS was evaluated by immunohistochemistry in rat liver fixed sections (3 μm) embedded in paraffin, as described above. Deparaffinised sections were treated with microwaves at 700 W for 10 minutes in 0.01 M citrate buffer at pH 6.0 for antigen retrieval before immunostaining. Non-specific immunostaining was blocked by treatment with 0.3% casein and 0.5% Tween 20 for 30 minutes before incubation with normal goat serum. iNOS expression was assessed with a specific monoclonal antibody against rat iNOS (Transduction Laboratories, Lexington, Kentucky, USA) at a 1:20 dilution for 16 hours at 4°C. iNOS was visualised using a standard multilink streptavidin-biotin-alkaline phosphatase kit (Master Diagnostica).

The degree of immunostaining of iNOS was evaluated semiquantitatively using four categories: (1) no expression; (2) mild overexpression or less than 25% of hepatocytes stained; (3) moderate overexpression or 25–50% of hepatocytes stained; (4) intense overexpression or more than 50% of hepatocytes stained.

STATISTICAL ANALYSIS

Results are expressed as mean (SEM). Comparisons were made using ANOVA with the least significant difference test (LSD). Statistical significance was set at p⩽0.05.

Results

LIVER INJURY INDUCED BY D-GALACTOSAMINE AND ITS PROTECTION BY PGE1

Administration of D-GalN induced serious liver injury with numerous apoptotic bodies in hepatic sections (fig 3C) and an increase in DNA fragmentation in liver extract (fig 4, lane 3) compared with control animals (fig 3A; fig 4, lane 2, respectively). Biochemical parameters of liver injury, such as ALT activity in serum, were also significantly enhanced by D-GalN (1109 (172.0) U/l) compared with the control group (19 (0.9) U/l) (fig 5) (p⩽0.05). Prior administration of PGE1 reduced all histological markers of necrosis and apoptosis induced by D-GalN (fig 3D; fig 4, lane 5; fig5).

Figure 3

Effects of prior administration of prostaglandin E1 (PGE1 ) on histological liver injury induced by d-galactosamine (D-GalN) observed by light microscopy. (A) Control, (B) PGE1, (C) D-GalN, and (D) PGE1+D-GalN treated rats. PGE1 reduced the numerous apoptotic bodies exhibiting compacted chromatin (a, arrow) observed with D-GalN. (Original magnification ×100.)

Figure 4

Effects of prior administration of prostaglandin E1 (PGE1 ) on DNA fragmentation in liver extract induced byd-galactosamine (D-GalN) with or without prior administration of anti-tumour necrosis factor α (TNF-α) antibodies. Lane 1: DNA molecular weight marker (base pairs); lane 2: control; lane 3: D-GalN; lane 4: PGE1; lane 5: PGE1+D-GalN; lane 6: anti-TNF-α+control; lane 7: anti-TNF-α+D-GalN; lane 8: anti-TNF-α+PGE1; and lane 9: anti-TNF-α+PGE1+D-GalN. Administration of PGE1 or anti-TNF-α antibodies reduced DNA fragmentation induced by D-GalN.

Figure 5

Effects of prior administration of prostaglandin E1 (PGE1 ) on alanine aminotransferase (ALT) activity in serum induced byd-galactosamine (D-GalN) with (+) or without (−) inhibition of iNOS activity by methylisothiourea (MT). PGE1 only reduced ALT in serum induced by D-GalN in the absence of MT. Values are mean (SEM) (n=10). Groups with different superscript letters are significantly different (p⩽0.05).  

TNF-α EXPRESSION AND SYNTHESIS FOLLOWING D-GALACTOSAMINE AND/OR PGE1

D-GalN significantly enhanced TNF-α content in liver extract (0.133 (0.0070) pg/μg protein) (fig 6A, lane 3) that was correlated with higher expression of TNF-α in the cytoplasm of hepatocytes (degree 3) (fig 6D) than TNF-α content (0.116 (0.0005) pg/μg protein) (fig 6A, lane 2) and TNF-α staining (degree 2) (fig 6B) in control livers (p⩽0.05). Particularly interesting was the strong staining of TNF-α in apoptotic bodies in liver sections from D-GalN treated rats (fig 6D). PGE1 alone enhanced TNF-α expression in the liver (fig 6A, lane 4; fig 6C). Prior administration of PGE1 to D-GalN treated animals gave rise to a significant increase in hepatic TNF-α content (0.230 (0.0033) pg/μg protein) (fig 6A, lane 5) and an overall higher TNF-α diffused expression in hepatocytes (degree 4) (fig 6E) (p⩽0.05).

Figure 6

Effects of prior administration of prostaglandin E1 (PGE1 ) on expression of tumour necrosis factor α (TNF-α) in the liver of animals treated withd-galactosamine (D-GalN). (A) TNF-α in liver extract measured by western blot. Lane 1: standard (25 pg); lane 2: control; lane 3: D-GalN; lane 4: PGE1; and lane 5: PGE1+D-GalN. Western blot is representative of five different experiments. Prior administration of PGE1strongly enhanced TNF-α expression in the liver extract induced by D-GalN. TNF-α expression in liver sections evaluated by immunohistochemistry: (B) control; (C) PGE1; (D) D-GalN; and (E) PGE1+D-GalN treated rats. Images are representative of five different experiments. PGE1 enhanced TNF-α staining in the nucleus of control hepatocytes. Numerous apoptotic bodies (a, arrow) induced by D-GalN showed strong TNF-α staining. Prior administration of PGE1 to D-GalN treated rats induced enhancement of diffused TNF-α staining. (Original magnification ×100.)  

It was confirmed that the presence of endotoxin, as a potential inducer of TNF-α release, in serum from control and D-GalN treated animals was undetectable (data not shown). TNF-α serum concentrations are shown in fig 7. TNF-α concentrations in serum were significantly higher in D-GalN (21 (1.9) pg/ml) and PGE1 (20 (2.9) pg/ml) groups than in the control group (11 (1.3) pg/ml) (p⩽0.05). An additional significant effect on TNF-α levels in serum was observed when PGE1 and D-GalN were administered together (p⩽0.05).

Figure 7

Effects of prior administration of prostaglandin E1 (PGE1 ) on tumour necrosis factor α (TNF-α) in serum of rats treated withd-galactosamine (D-GalN) with (+) or without (−) inhibition of iNOS activity by methylisothiourea (MT). Prior administration of PGE1 strongly enhanced TNF-α concentration in serum induced by D-GalN. MT administration raised TNF-α in all groups. Values are mean (SEM) (n=10). Groups with different superscript letters are significantly different (p⩽0.05).

NITRIC OXIDE MODULATES THE EFFECTS OF D-GALACTOSAMINE AND PGE1 ON LIVER INJURY AND TNF-α SYNTHESIS

PGE1 administration significantly increased nitrite/nitrate concentrations in PGE1 (29 (2.9) μM) and PGE1+D-GalN (22 (1.5) μM) groups compared with control (17 (0.9) μM) and D-GalN (18 (1.0) μM) groups, respectively (p⩽0.05) (fig 8). To study the role of nitric oxide on D-GalN liver injury and its protection by PGE1, MT was administered to inhibit iNOS activity. Nitric oxide concentrations in serum were reduced significantly in all MT treated groups (p⩽0.05) (fig 8). The increase in ALT activity in serum by D-GalN was not affected by MT treatment (fig 5). However, protection by PGE1 on ALT release induced by D-GalN in the PGE1+D-GalN (1609 (290.9) U/l) group with MT treatment was significantly abolished compared with values obtained with PGE1+D-GalN (547 (104.7) U/I) without iNOS inhibition (fig 5) (p⩽0.05).

Figure 8

Effects of prior administration of prostaglandin E1 (PGE1 ) on nitrite/nitrate concentration in serum of rats treated withd-galactosamine (D-GalN) with (+) or without (−) inhibition of iNOS activity by methylisothiourea (MT). PGE1 enhanced nitrite/nitrate concentrations in serum in the absence of MT. MT administration reduced nitrite/nitrate values in all groups. Values are mean (SEM) (n=10). Groups with different superscript letters are significantly different (p⩽0.05).

It was also observed that inhibition of iNOS activity by MT modified TNF-α concentrations in serum obtained from experimental animals (fig7). In this sense, MT significantly enhanced TNF-α levels in serum from all groups compared with their respective controls (p⩽0.05).

ANTI-TNF-α ANTIBODIES MODULATE THE EFFECTS OF D-GALACTOSAMINE AND PGE1 ON LIVER INJURY

Administration of anti-TNF-α antibodies to D-GalN treated animals sharply reduced DNA fragmentation in liver extract (fig 4, lane 7) compared with that found in D-GalN treated rats (fig 4, lane 3) without additional treatment. ALT activity in serum was also significantly lower in D-GalN treated rats when anti-TNF-α antibodies were administered (397 (68.2) U/l) compared with values in the D-GalN group without antibodies (1109 (172.0) U/l) (p⩽0.05) (table 1). Interestingly, no protection by PGE1 on DNA fragmentation (fig 4, lane 9) or ALT release (1093 (130.1) U/l) (table 1) induced by D-GalN was observed when anti-TNF-α antibodies were administered compared with DNA fragmentation (fig 4, lane 5) and ALT (547 (104.7) U/l) in the PGE1+D-GalN group without antibodies (p⩽0.05) (table 1).

Table 1

Effects of administration of anti-tumour necrosis factor α (TNF-α) antibodies on alanine aminotransferase (ALT) activity in serum induced by D-galactosamine (D-GalN) treated rats and its reduction by prostaglandin E1(PGE1)

We also evaluated if anti-TNF-α antibody treatment had an effect on hepatic iNOS (fig 9) expression and nitrite/nitrate (table 2) concentrations in serum. The liver from control rats showed no expression of iNOS (degree 1) (fig 9A). Administration of anti-TNF-α antibodies to PGE1+D-GalN rats clearly reduced expression of iNOS in hepatocytes (degree 2) (fig 9C) compared with that observed in the PGE1+D-GalN (degree 3) group (fig 9B). In contrast, administration of TNF-α antibodies significantly enhanced nitrite/nitrate concentrations in serum in control animals but did not change their levels in the other groups compared with their respective control groups (p⩽0.05) (table 2).

Figure 9

Effects of administration of anti-tumour necrosis factor α (TNF-α) antibodies on expression of inducible nitric oxide synthase (iNOS) in liver sections induced by prior administration of prostaglandin E1 (PGE1 ) tod-galactosamine (D-GalN) treated rats. (A) Control, (B) PGE1+D-GalN, and (C) anti-TNF-α+PGE1+D-GalN treated rats. Images are representative of five different experiments. Anti-TNF-α antibodies reduced enhancement of iNOS expression induced by PGE1 to D-GalN treated animals. (Original magnification ×200.)

Table 2

Effects of administration of anti-tumour necrosis factor α (TNF-α) antibodies on nitrite/nitrate concentrations in serum with D-galactosamine (D-GalN) and/or prostaglandin E1(PGE1) treatments

Discussion

Using different experimental interventions we have shown that TNF-α participates in D-GalN induced liver injury and PGE1 dependent protection. Our data suggest that this protective effect of PGE1 is related to enhanced expression of TNF-α and iNOS in hepatocytes.

D-GalN has been shown to be a suitable experimental model of liver injury.1 D-GalN reduces the intracellular pool of uracil nucleotides in hepatocytes thus inhibiting the synthesis of RNA and proteins.1 2 Intraperitoneal administration of D-GalN causes an increase in the level of transaminases in serum, hepatic necrosis, and coma in rats.3 26 30-32 As previously described,26 33 liver injury (figs 3-5) induced by D-GalN correlated with an increase in TNF-α in serum (fig 7). In addition, our study showed that this increase in TNF-α concentration in serum was associated with enhanced TNF-α expression in hepatocytes with strong staining in apoptotic bodies (fig 6). The direct involvement of TNF-α in D-GalN liver injury was demonstrated when prior administration of anti-TNF-α antibodies reduced apoptosis (fig4) and necrosis (table 1). Nevertheless, treatment with anti-TNF-α antibodies diminished D-GalN induced ALT release by only 65%. This result may indicate that factors other than TNF-α are also mediating liver injury by D-GalN. Under our conditions we can assume that D-GalN was directly responsible for TNF-α production as no endotoxin was detected in samples from control and D-GalN treated rats and endotoxaemia appears 36 hours after D-GalN administration.34 Nevertheless, stimulation of inflammatory cells by D-GalN cannot be excluded as hepatotoxin stimulated TNF-α release one minute after portal infusion.33

The relationship between D-GalN liver damage and TNF-α has been reported previously in a murine model of septic liver failure.35-37 TNF-α participates as a primary mediator in the pathogenesis of infection, injury, and inflammation, and in the beneficial processes of host defence and tissue homeostasis. In this sense, acute systemic release of TNF-α during endotoxaemia plays a central role in liver injury and lethality.38-40 In contrast, when low amounts of TNF-α are released in tissues, the beneficial effects may predominate.41 Under our conditions, although D-GalN induced a moderate increase in TNF-α (fig7) concentration in serum compared with that found during endotoxaemia,38 expression of low amounts of TNF-α in hepatocytes may enhance cell death during D-GalN induced transcriptional arrest. An increase in TNF-α in serum has also been associated with enhancement of liver damage in other diseases, such as alcoholic hepatitis,42 and ischaemia-reperfusion and allograft rejection after liver transplantation.43 44 In contrast, enhanced expression of TNF-α during viral hepatitis45-47 may be beneficial due to its antiviral associated activity.48

D-GalN administration alone did not modify nitrite/nitrate (fig 8) concentrations in serum and inhibition of iNOS activity by MT did not enhance ALT release induced by D-GalN (fig 5). As a consequence, our studies showed that nitric oxide was not involved in D-GalN induced liver injury. Nevertheless, a clear tendency to reduce nitrite/nitrate production by D-GalN was observed when PGE1 was also administered (fig 8). Such an effect may be due to haemodynamic disturbances49 and/or reduction in nitric oxide production from hepatocytes by D-GalN.50 It has been observed that expression of iNOS in hepatocytes is associated with liver injury induced by an oxidative stress related model such as acetaminophen administration,51 chronic viral hepatitis,52and allograft rejection in liver transplantation.53 Liver damage induced by endotoxaemia was exacerbated by non-selective inhibition of nitric oxide synthase (cNOS and iNOS)54 but reduced by selective iNOS inhibition.55 The authors postulated that although low production of nitric oxide by cNOS was cytoprotective and maintained microvascular patency, excess nitric oxide production by iNOS was cytotoxic. In this sense, vasoactive properties of nitric oxide are also associated with liver protection during experimental alcoholic disease56 and ischaemia-reperfusion injury.57 Under our conditions, inhibition of iNOS in D-GalN treated rats enhanced liver injury (fig 5) when PGE1 was also administered (MT+PGE1+D-GalN group) and TNF-α (fig 7) reached the highest concentration in serum among all groups. Florquin and colleagues58 have indicated that several cytokines (for example, TNF-α, interleukin 1α, or interferon γ) could be involved in the increase in liver injury induced by iNOS inhibition during experimental septic liver failure.

Several reports have shown that prostaglandin E decreases liver damage induced by D-GalN.3 15 26 As previously described,26 administration of PGE1 30 minutes before D-GalN reduced necrosis and apoptosis (figs 3-5) and enhanced TNF-α (fig 7) concentrations in serum. We showed that PGE1 alone and with D-GalN increased TNF-α expression in hepatocytes (fig 6). Pretreatment with sublethal doses of TNF-α protected D-GalN treated mice from an otherwise lethal challenge of lipopolysaccharide or TNF-α.19 In the present study, administration of anti-TNF-α antibodies to PGE1+D-GalN treated animals showed that TNF-α is essential for PGE1protection against D-GalN induced liver injury (table 1, fig 4). As described above, TNF-α was also important during D-GalN induced liver injury. Accordingly, recovery of liver damage (table 1, fig 4) by D-GalN in anti-TNF-α-PGE1+D-GalN treated rats also indicated that factors other than TNF-α were mediating liver injury in this experimental group. Modulation of TNF-α expression by PGE1 in other cell types, such as inflammatory cells, has been studied extensively. In this sense, PGE1 may reduce59 60 or enhance61 62 in vitro TNF-α release from inflammatory cells in relation to the concentration range of the prostanoid used. Administration of an inhibitor of prostaglandin synthesis, such as indomethacin, increased TNF-α levels in serum and enhanced gastric mucosal damage in rats.63 Also, other authors have shown that indomethacin does not modify lipopolysaccharide induced TNF-α release in mice.64 Under our experimental conditions, the effect of PGE1 administered 30 minutes before D-GalN may have been different from that found in other studies in which it was administered after induction of liver injury. Controversial results on the in vivo effect of PGE1 on TNF-α release could also be due to the concentration range of the prostanoid, animal model (toxic or endotoxaemia), type of cells studied, animal species, and pattern of administration.

In addition to TNF-α,19 prior administration of nitric oxide donors25 has also been shown to protect against liver injury. In fact, enhanced expression of iNOS or nitric oxide donor inhibits apoptosis in cultured hepatocytes.65 66Under our conditions, PGE1 enhanced expression of iNOS (fig9) in hepatocytes and nitrite/nitrate (fig 8) concentrations in serum from D-GalN treated rats. The important role of iNOS during PGE1 protection against D-GalN liver injury was also demonstrated when inhibition of iNOS by MT abolished ALT (fig 5) reduction in these animals. iNOS expression in hepatocytes during lipopolysaccharide treatment in vivo67 68 and in vitro69 is regulated by TNF-α. In agreement with this, anti-TNF-α antibody administration showed that TNF-α was responsible for induction of iNOS (fig 9) expression in hepatocytes by PGE1 in D-GalN treated rats.

In conclusion, we have shown that protection of D-GalN induced liver injury by PGE1 is related to its capacity to increase TNF-α and iNOS expression in hepatocytes. Our data strongly support the beneficial effect of potential therapeutic mediators that could enhance expression of iNOS in hepatocytes during liver failure.

Acknowledgments

This work was supported by the Programa de Promoción de la Investigación en Salud del Ministerio de Sanidad y Consumo (FIS) (97/1300) and by a grant from Fundación Hospital Reina Sofía-CajaSur. We are grateful to María Dolores Rodríguez and Francisca Sáez for immunohistochemical assistance.

References

View Abstract

Footnotes

  • Abbreviations used in this paper:
    TNF-α
    tumour necrosis factor α
    D-GalN
    d-galactosamine
    PGE1
    prostaglandin E1
    iNOS
    inducible nitric oxide synthase
    cNOS
    constitutive nitric oxide synthase
    MT
    methylisothiourea
    ALT
    alanine aminotransferase
    PBS
    phosphate buffered solution

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.