Pulmonary toxicity of thioureas in the rat.

Administration of alpha-naphthylthiourea (ANTU) to rats causes damage to pulmonary endothelial cells and possibly mesothelial lining cells that together may account for the massive pleural effusion characteristic of thiourea toxicity. Using 35S-thiourea as a model compound, the extent of binding of 35S to lung proteins correlated well with the extent of edema, suggesting that the extent of binding of thiourea metabolites is a measure of lung toxicity. ANTU and phenylthiourea (PTU) compete for 35S binding to lung slices, suggesting that these toxins may act in a similar way. Binding of 35S in lung slices from resistant rats is much less than in controls, and resistance cannot be explained by differences in either whole body metabolism or redistribution of thiourea in vivo. Lung glutathione levels (in vitro and in vivo) in normal and resistant rats following thiourea administration were essentially the same. However, at doses of thiourea that cause pleural effusion, there was an increase in total lung glutathione. ImagesFIGURE 1. AFIGURE 1. BFIGURE 2. AFIGURE 2. B


Introduction
The thioureas cause massive pulmonary edema and pleural effusion in rats (1)(2)(3)(4). One of them, ci-napthylthiourea (ANTU) was developed as a rat poison, but during extensive trials (4) it became apparent that rats ingesting small nonlethal doses of ANTU rapidly became resistant to it (tachyphylaxis). This resistance reduced the effectiveness of such compounds as potential rodenticides but nevertheless provided a useful model to investigate their mode(s) of action.
Ultrastructural studies have shown high specificity of thioureas for pulmonary endothelial cells (5,6), and even at high doses (50 mg/kg) there appears to be little damage to any other cells in the alveolar capillary unit, although the edema may be extensive (6). In addition, ANTU-induced biochemical changes in levels of angiotensin converting enzyme (7-9) and 5-hydroxytryptamine uptake (10) are considered to reflect specific endothelial cell damage.
Massive pleural effusion is a characteristic feature of thiourea toxicity. It is possible that the intrathoracic pressure exerted on the lung by this effusate is often the cause of death, rather than the alveolar edema that also occurs. Despite extensive research, the mode of action of these compounds remains unclear, although there is good evidence to suggest that the thioureas require metabolic activation before toxic effects are manifested (11)(12)(13)(14)(15)(16). Although the mechanism of activation has not been resolved, thiols protect against thiourea toxicity and covalent binding (17), and glutathione depletion enhances both the toxicity and protein binding of ANTU (11) and thiourea (15). Further, administration of either glutathione or cysteine protects against the toxicity of phenylthiourea (PTU) in vivo (18). It has also been shown that glutathione can protect against the binding of radioactivity in studies with 14C-thiourea in vitro (15). Binding studies with 14C-thiourea (13)(14)(15)(16)(17) have demonstrated the covalent binding of radiolabel to proteins in lung and liver, and that binding is less in resistant rats. It has also been shown that radiolabeled ANTU binds covalently to protein in the lung and liver (11,12). Interestingly, between two and six times the binding is observed if ANTU is labeled with 35S rather than 14C (11,12), which suggests that binding of 35S is a better measure of total binding. No binding studies, however, have been undertaken with 35S-thiourea.
In this study, three aspects of thiourea toxicity are explored: namely a) the effect of ANTU on the ultrastructure of the rat lung with particular attention focused on the mesothelial layer that lines the surface of the lung, b) the binding of 35S (from 35S-thiourea) to tissue proteins in vivo and in vitro in normal and thiourea-resistant rats to determine if there is a correla-SCOlTETAL. tion between toxicity and binding, and c) the levels of glutathione in lungs of normal and resistant rats.

Electron Microscopy
Rats, under ether anesthesia, received pentobarbitone (60 mg/kg) IV via the penal vein. The trachea was isolated and cannulated (Portex, Hythe, Kent, UK; int. dia. 1.67 mm; ext. dia. 2.42 mm). After a midline abdominal incision, a small incision was made in the diaphragm to deflate the lungs. The thoracic cavity was exposed and the lungs were fixed by the method of Hammond and Mobbs (19). Formalin vapor was instilled via the trachea cannula until the lungs were fully inflated. The lungs were removed, the trachea was clamped, and the whole preparation was immersed in phosphate-buffered formalin (pH 7.4) for 72 hr. Blocks of tissue (approximately 1 mm3) were cut from two areas of the lung-from the edge containing the mesothelial layer and from central areas of lung parenchymal tissue. These blocks were stained with uranyl acetate (2 %) for 60 min in the dark at 4°C, and dehydrated in ethanol followed by propylene oxide, and embedded in araldite. Sections (60-90 nm) were cut using a Reichart OM4 microtome and mounted on a celloidine-carbon coated grid. Sections were stained in 5% uranyl acetate followed by Reynold's lead citrate. Sections were examined in a Philips EM 400T electron microscope at 80 KV acceleration voltage.

Preparation of Lung Slices
Animals were anesthetized as described above and exsanguinated. The lungs were deflated and the thoracic cavity was exposed. An incision was made in the right ventricle of the heart and a cannula (Portex; int. dia. 1.67 mm; ext. dia. 2.42 mm) was inserted into the pulmonary artery. Isotonic saline was perfused into the lung at a pressure of 25 cm H20, and an incision made in left auricle allowed the exit of saline from the lungs. This procedure cleared the vascular bed. Following perfusion, the lungs were dissected free from the heart, esophagus, and associated connective tissue. The external surfaces of the lungs were rinsed in ice-cold saline (0.15 M NaCl). Slices (0.7 mm) were prepared using the Mclllwain tissue chopper and incubated in MEM Eagles medium (200 mg tissue/mL medium) in a shaking water bath (37 OC) with continuous gassing (95% 02, 5% CO2). Slices were equilibriated for 30 min and then suspended in fresh medium.

Preparation of Tissues for Binding Studies (In Vivo)
Animals were anesthesized with ether, and pentobarbitone (60 mg/kg) was administered IV via the penal vein.
Following a midline abdominal incision, the dorsal aorta was isolated and clamped distal to the renal artery. A cannula (int. dia. 1.67 mm; ext. dia. 2.42 mm) was inserted into the dorsal aorta and saline (0.15 M NaCl) was infused via the cannula at a pressure of 1 M H20.
An incision was made in the vena cava immediately anterior to the diaphragm to allow the saline to escape. This procedure cleared the liver and kidneys of blood. The lungs were also perfused free of blood as described previously. Lungs, liver, and kidneys were removed, weighed, and assayed for protein-bound radioactivity.

Assay for Protein-Bound 35S
Tissue homogenates (20%) were prepared in ice-cold saline (0.15 M NaCl). Five volumes of ice-cold trichloracetic acid (TCA) (20% w/v) containing phosphotungstic acid (2% w/v) were added to one volume of tissue homogenate, the samples were kept on ice for 30 min and centrifuged (600g, 5 min). The supernatant was discarded and the pellet resuspended in five volumes of ICA (20%). This step was repeated three times or until radioactivity was not detectable in the supernatants. The washed pellet was then dissolved in 2.5 M NaOH at 60(C for 4 hr. Protein in the solubilized pellet was determined by the method of Lowry et al. (20). Radioactivity was determined by scintillation counting (LKB Rackbeta 1217) with quench curve correction by the external standard ratio method.

Experimental and Results
Effect of ANTU on Lung Ultrastructure Three rats received a single IP injection of ANTU (20 mg/kg body weight) dispersed in olive oil (0.3 mL), and three rats (designated control) received olive oil alone. Electron micrographs of the alveolar capillary unit from controls (Fig. 1A) showed no edema in the alveolar spaces. Moreover, Tyrpe I epithelial cells, capillary endothelial cells, and the basement membrane all appeared intact. By contrast, the micrograph obtained from ANTUtreated lungs is shown in Figure 1B. The alveolar space is flooded with fluid and protein, and the endothelial barrier appears absent, whereas the basement membrane and ¶Iype I cells appear intact and unaffected. In addition, changes were observed in the mesothelial cell layer that lines the outside of the lung. A normal mesothelial cell layer from a control animal in which two mesothelial cells lie on a thin basement membrane, overlying interstitial tissue containing collagen bundles is shown in Fig. 2A. After ANTU challenge (Fig. 2B) the junctions between the cells appear intact; the underlying basement membrane, however, is grossly swollen.
Binding of 35S (from Thiourea) to Proteins in Lung, Liver, and Kidneys In Vivo IP. After 4 hr the animals were anesthetized and the vascular beds of liver, lungs, and kidneys were cleared of blood. Each tissue was assayed for protein-bound 35S. In normal rats, protein binding was highest in lung; however, binding was approximately 60% less in resistant rats (Fig. 3). In this study resistance was developed by repeated administration of low doses of nonradioactive thiourea. It was, therefore, possible that unlabeled drug was occupying binding sites in the tissues of the resistant rats. The subsequent binding of 35S thiourea would then show an apparent decrease because some of the binding sites were already occupied. To eliminate this possibility, resistance was developed using 35S-thiourea. Rats were treated with 35S-thiourea (0.5 mg/kg body weight, 5 ACi/mmole) IP daily for 3 days; 24 hr after the final injection the vascular beds of the lungs, liver, and kidneys were cleared of blood and protein-bound 35S was measured. Ibtal binding of 35S under these conditions was relatively low (approximately 0.2% of that observed in normal rats after 4 hr at a dose of 15 mg/kg body weight which causes edema); therefore, the reduced binding in resistant rats could not be explained by the saturation of binding sites with thiourea or its metabolites. In these in vivo studies, the degree of resistance was confirmed by using the ratio of wet weight lung/body weight as an index of edema following thiourea challenge. Resistant rats showed consistently less edema than the normals, and there was a good correlation (correlation coefficient 0.963) between the amount of protein-bound 35S in lung and the degree of edema (Fig. 4).

Binding of 35S (from Thiourea) to Proteins in Lung Slices
Lung slices from two nonnal rats and two resistant rats were prepared as described. Slices were incubated with 35S-thiourea (0.167 mM, 6 IACi/mmole) for 2 hr. Proteinbound 35S was approximately 65% less in resistant slices than in the normals (Thble 1). Equimolar ANTU or PTU inhibited binding in normal slices by 76% and 69%, respectively, whereas urea ( the oxygen analogue of thiourea that does not produce edema) had no effect on the binding of 35S. ANTU and PTU also reduced the binding in slices from resistant animals by 44% and 38%, respectively, whereas urea had no effect.

Effect of Thiourea on Lung Glutathione Levels
Rats received IP injections of thiourea (15 mg/kg body weight; in 0.15 M NaCl, 0.3 mL) or 0.15 M NaCl (0.3 mL).
At 1 and 2 hr after administration animals were anesthetized, the lungs cleared of blood, and the tissue was assayed for glutathione. Glutathione levels in test and control lungs were the same after 1 hr, but after 2 hr the glutathione levels in thiourea treated rats had risen by 60% (lible 2). Ibble   In other experiments, lung slices were prepared from normal and resistant rats and incubated in vitro in the presence and absence of 1 mM thiourea. The tissue was sampled at 0, 1, and 2 hr and assayed for glutathione. Glutathione levels were substantially the same in normal and resistant rats at all of the time points (Fig. 5).
In further experiments glutathione levels were measured in lungs from normal and resistant rats 24 hr following the final injection. Levels of glutathione were not altered significantly in resistant rats (normal: 428 ± 15 nmole GSH/whole lung; resistant: 460 + 18 nmole GSH/whole lung. Results expressed as an average of two determinations).

Discussion
Many edemagenic agents are reported to cause pulmonary endothelial damage, allowing excessive fluid to invade the interstitium. There are three possible routes for the subsequent escape of the fluid: clearance by lymphatics, passage into the alveolar airways, and passage into the pleural cavity. The majority of edemagenic agents enhance the passage of fluid into alveolar airways, presumably by causing damage to the alveolar epithelial lining, resulting in alveolar edema. However, the thiour-eas are unusual in causing severe pleural effusion, possibly because of an apparent sensitivity of the mesothelial cell lining layer. This possibility is supported by the results reported here. Evidence is also presented that the extent of binding of the agent or its metabolites is related to edema formation and that thioureas do not cause depletion of glutathione levels. Resistance to the toxins cannot be explained by differential metabolism, tissue distribution, or altered glutathione levels.
SDS-PAGE analysis of the protein composition of the pleural effusion following ANTU induced lung damage has shown that it is qualitatively identical to that of normal plasma (unpublished observations). The passage of this cell-free filtrate into the cavity might be expected to coincide with pulmonary vascular endothelial damage, and this is confirmed by ultrastructural and histochemical examinations in this and other studies with thioureas (5-10). Whether or not endothelial damage alone is sufficient to allow excessive movement of fluid into the pleural cavity is not clear. However, the associated change in the mesothelial basement membrane (grossly swollen) and the lack of any apparent changes to the alveolar epithelial barrier suggests that these particular bloodborne edemagenic agents are atypical in causing pleural effusion as well as alveolar edema. In spite of the increased vesicular activity in the mesothelial cells, this effect of thioureas may not be due entirely to direct action ofthe toxin but a normal defense response to the presence of excess fluid in the pleural cavity.
Metabolism and binding studies with 14C-thiourea suggested the need for metabolic activation of the toxin to a reactive metabolite that alkylates tissue proteins (13)(14)(15)(16)(17). The present studies demonstrate that protein binding of 35S derived from 35S-thiourea is very closely correlated with toxicity. Furthermore, importantly, they demonstrate that in resistant animals there is a dramatic reduction in bound 35S. It is possible, therefore, that 35S binding is a direct measure of toxicity and that the active metabolites alkylate tissue proteins causing the pulmonary edema that is typical of thiourea toxicity.
The identity of the active metabolite(s) of thiourea is unknown, but it has been suggested that ANTU can be activated to atomic sulphur (12). That a similar mechanism may operate for thiourea is supported by the inhibitory effects of ANTU and PTU on 35S binding from 2 I Wet wt/Body wt (xlO2) 35S-thiourea in vitro. The binding of 35S from thiourea to lung proteins in vitro and the decreased binding in lung slices from resistant rats suggests that activation takes place in the lung and that activated metabolites produced in the liver are not transported in the blood as, for example, pyrrolizidine alkaloids (22).
The number of proteins/enzymes reacting with metabolites of thioureas has yet to be established, although it is known that ANTU decreases the activity of cytochrome P-450 in lung and liver in vitro (11,12).
However, activation of thioureas may also occur by nonmicrosomal enzymes because in lungs exposed to '4C-thiourea in vivo, 75% of total 14C bound to protein occurs in the supernatant fraction (13,14).
The mechanism(s) by which rats become resistant to thioureas following the administration of small nonlethal doses is intriguing. Unpublished work from these laboratories has shown that in anesthetized rats (with ureter cannula) there is no significant difference in the metabolism and rate of excretion of 35S-thiourea between normal and resistant rats. Moreover, the differential bind- ing to lung proteins in vitro from normal and resistant rats also suggests that resistance is afforded by biochemical alterations in lung itself and not by changes in the disposition of thiourea in vivo. ANTU and PTU, which are edemagenic agents in their own right, compete with 35S-thioureaforbindinginvitro, suggesting that they act by the same pathway (Table 1). Interestingly, ANTU and PTU competed less well for the binding in resistant lung slices; this might suggest the presence of two types of binding site, a nonspecific site present in both normal and resistant lung slices, and a specific site(s) that is reduced or masked in resistant rats. These findings are in keeping with the two types of binding sites observed for H4C-thiourea (15). The fact that urea does not compete for binding suggests that the C = S moiety is required for binding and toxicity.
Hollinger et al. (17) have reported that thiourea can deplete tissue glutathione by as much as 90% after 1 hr, and these data are consistent with the observations that depletion of tissue glutathione with diethylmalate leads to increased toxicity of thioureas, whereas coadministration with thiols prevents edema (18). It was surprising, therefore, that in the present study there was no evidence to suggest that thiourea is able to deplete lung glutathione levels in vivo. Indeed, at 2 hr after thiourea treatment, lung glutathione levels were increased by approximately 60%, at which point lung wet weight had increased by approximately 100%. The apparent difference between these results and those of Hollinger et al. (17) may be explained by the different assay systems for glutathione. Hollinger et al. employed the assay described by Ellman (23) that is nonspecific and measures total thiol. In this laboratory a fluorometric assay specific for glutathione was used (21). In addition it was shown that thiourea does not interfere with the latter assay, even when in a 10,000-fold excess (unpublished observations).
Thus, in this study using a more specific assay for glutathione, there was no evidence to suggest that the appearance of edema coincided with a decrease in total glutathione levels in lung. Moreover, the induction of resistance cannot be explained by an increase in total lung glutathione because levels were not elevated in resistant rats.