The sevoflurane degradation product compound A is nephrotoxic in rats, in which it undergoes extensive metabolism to glutathione and cysteine S-conjugates. The mechanism of compound A nephrotoxicity in rats is unknown. Compound A nephrotoxicity has not been observed in humans. The authors tested the hypothesis that renal uptake of compound A S-conjugates and metabolism by renal cysteine conjugate beta-lyase mediate compound A nephrotoxicity in rats.
Compound A (0-0.3 mmol/kg in initial dose-response experiments and 0.2 mmol/kg in subsequent inhibitor experiments) was administered to Fischer 344 rats by intraperitoneal injection. Inhibitor experiments consisted of three groups: inhibitor (control), compound A, or inhibitor plus compound A. The inhibitors were probenecid (0.5 mmol/kg, repeated 10 h later), an inhibitor of renal organic anion transport and S-conjugate uptake; acivicin (10 mg/kg and 5 mg/kg 10 h later), an inhibitor of gamma-glutamyl transferase, an enzyme that cleaves glutathione conjugates to cysteine conjugates; and aminooxyacetic acid (0.5 mmol/kg and 0.25 mmol/kg 10 h later), an inhibitor of renal cysteine conjugate beta-lyase. Urine was collected for 24 h and then the animals were killed. Nephrotoxicity was assessed by light microscopic examination and biochemical markers (serum urea nitrogen and creatinine concentration, urine volume and urine excretion of protein, glucose, and alpha-glutathione-S-transferase [alpha GST], a marker of tubular necrosis).
Compound A caused dose-related nephrotoxicity, as shown by selective proximal tubular cell necrosis at the corticomedullary junction, diuresis, proteinuria, glucosuria, and increased alpha GST excretion. Probenecid pretreatment significantly (P < 0.05) diminished compound A-induced increases (mean +/- SE) in urine excretion of protein (45.5 +/- 3.8 mg/24 h vs. 25.9 +/- 1.7 mg/24 h), glucose (28.8 +/- 6.2 mg/24 h vs. 10.9 +/- 3.2 mg/24 h), and alpha GST (6.3 +/- 0.8 micrograms/24 h vs. 1.0 +/- 0.2 microgram/24 h) and completely prevented proximal tubular cell necrosis. Aminooxyacetic acid pretreatment significantly diminished compound A-induced increases in urine volume (19.7 +/- 3.5 ml/24 h vs. 9.8 +/- 0.8 ml/24 h), protein excretion (37.2 +/- 2.7 mg/24 h vs. 22.2 +/- 1.8 mg/24 h), and alpha GST excretion (5.8 +/- 1.5 vs. 2.3 micrograms/24 h +/- 0.8 microgram/24 h) but did not significantly alter the histologic pattern of injury. In contrast, acivicin pretreatment increased the compound A-induced histologic and biochemical markers of injury. Compound A-related increases in urine fluoride excretion, reflecting compound A metabolism, were not substantially altered by any of the inhibitor treatments.
Intraperitoneal compound A administration provides a satisfactory model of nephrotoxicity. Aminooxyacetic acid and probenecid significantly diminished histologic and biochemical evidence of compound A nephrotoxicity, whereas acivicin potentiated toxicity. These results suggest that renal uptake of compound A-glutathione or compound A-cysteine conjugates and cysteine conjugates metabolism by renal beta-lyase mediate, in part, compound A nephrotoxicity in rats.
Sevoflurane undergoes dehydrofluorination by soda lime and barium hydroxide lime in anesthesia machine carbon dioxide absorbers, producing small quantities of several degradation products. [1,2]Under low-flow and closed-circuit anesthesia, two sevoflurane degradation products, fluoromethyl-2,2-difluoro-1-(trifluoromethyl) vinyl ether and fluoromethyl-2-methoxy-2,2-difluoro-1-(trifluoromethyl)ethyl ether, were detected in anesthesia circuits. [3–6]Previously these were called compound A and compound B, respectively.* Of the two degradation products, compound A is quantitatively predominant, reaching peak inhaled concentrations averaging 8–27 ppm during low-flow and closed-circuit anesthesia, and toxicologically more significant. [4,5,7–9]Compound A also may be present as a minor contaminant in sevoflurane. 
Compound A is nephrotoxic in rats. One- and 3-h exposures to high (350–1,400 ppm) inhaled concentrations caused gross pulmonary, hepatic, and renal abnormalities. Renal toxicity was characterized by histologic evidence of tubular cell degeneration and necrosis and by biochemical evidence including elevated blood urea nitrogen (BUN) and serum creatinine concentrations and increased urine protein, blood, ketones, and N-acetyl-beta-glucosaminidase excretion. [9–12]Lower inhaled concentrations caused only corticomedullary tubular cell necrosis, with a median effective concentration of 200 ppm. The thresholds for renal injury have been estimated at 50 ppm [10,11]and 114 ppm for a 3–12-h exposure and at 200 ppm for a 1-h exposure. 
The mechanism of compound A toxicity in rats is unknown. Compound A contains two metabolically susceptible sites, a fluoromethyl group and a difluorovinyl group, which suggests several potential mechanisms of bioactivation and toxification (Figure 1). Compound A reacts enzymatically or nonenzymatically with glutathione to form several compound A-glutathione conjugates. [14,15]Compound A-glutathione conjugates are subsequently cleaved to form compound A-cysteine conjugates. Evidence was recently obtained demonstrating the metabolism of compound A-cysteine conjugates to potentially nephrotoxic intermediates by rat and human renal cysteine conjugate beta-lyase in vitro and in vivo. [16,17]Compound A also undergoes cytochrome P450-catalyzed defluorination at the fluoromethyl moiety, similar to the P450-catalyzed defluorination of sevoflurane, methoxyflurane, and other fluorinated ethers. [19,20]Fluoride formation is thought to underlie the nephrotoxicity of methoxyflurane and might theoretically explain compound A nephrotoxicity. Compared with glutathione conjugation, however, P450-mediated defluorination of compound A is a relatively minor pathway.
The metabolism of compound A to glutathione and cysteine conjugates is of particular interest. Compound A resembles many difluorovinyl compounds, such as tetrafluoroethylene, chlorotrifluoroethylene, dichlorodifluoroethylene, dibromodifluoroethylene, hexafluoropropene, and bromochlorodifluoroethylene, which cause selective proximal corticomedullary tubular cell necrosis. Difluorovinyl nephrotoxicity occurs by a well-characterized mechanism involving glutathione conjugate formation, cleavage to cysteine conjugates, renal uptake of cysteine and glutathione conjugates, and intrarenal metabolism by cysteine conjugate beta-lyase to highly toxic reactive intermediates responsible for the tubular cytotoxicity. [22,23]
Although compound A is nephrotoxic in rats, similar toxicity has not been observed in patients anesthetized with sevoflurane under conditions that produce compound A. Mean peak compound A concentrations of 8–27 ppm were measured during low-flow and closed-circuit sevoflurane anesthesia, but under these conditions there was no evidence of postoperative alterations in BUN or serum creatinine concentration. [3,4,6,7,24,25]
The purpose of this investigation was to elucidate the mechanism(s) of compound A nephrotoxicity in rats. We specifically tested the hypothesis that cleavage of compound A-glutathione conjugates to compound A-cysteine conjugates, renal uptake of compound A-glutathione and compound A-cysteine conjugates, and metabolism of compound A-cysteine conjugates by renal beta-lyase mediate compound A nephrotoxicity.
Materials and Methods
Compound A (99.9% by gas chromatography) was synthesized by Central Glass Company (Ube City, Japan) and provided by Abbott Laboratories (Abbott Park, IL). Fluoride standard solution was obtained from Orion Research (Boston, MA). Aminooxyacetic acid (AOAA), probenecid, and acivicin were purchased from Sigma Chemical Company (St. Louis, MO), as were all other reagents unless specified.
All animal experiments were approved by the University of Washington Animal Care Committee. Male Fischer 344 rats (weight, 220–240 g; Simonsen Laboratories, Gilroy, CA) were housed in individual metabolic cages, provided food and water ad libitum, and maintained on a 12-h light-dark cycle (6 AM-6 PM). All drugs were given by intraperitoneal injection. Compound A was administered in corn oil (0.9 ml). For initial dose-response experiments, compound A doses were 0–0.3 mmol/kg. For subsequent inhibitor experiments, the compound A dose was 0.2 mmol/kg unless otherwise indicated. For the series of inhibitor experiments, animals were randomized to one of three treatment groups: inhibitor (control), compound A, or inhibitor plus compound A. Control animals received either acivicin, probenecid, or AOAA in saline before corn oil. Compound A animals received saline followed by compound A. Inhibitor-plus-compound A animals received either AOAA, probenecid, or acivicin before compound A.
To assess the role of compound A-glutathione conjugates conversion to compound A-cysteine conjugates in compound A nephrotoxicity, animals were treated with acivicin, a noncompetitive and very slowly reversible inhibitor of gamma-glutamyl transferase. Acivicin was administered as a 5-mg/ml solution prepared immediately before use in alkaline-normal saline, and the pH was adjusted to 7–8 with 2% acetic acid. Control animals and inhibitor-plus-compound A animals were treated with acivicin (10 mg/kg, 2 ml/kg) 60 min before injection of 0.9 ml corn oil (controls) or compound A (0.2 mmol/kg in 0.9 ml corn oil). A second acivicin dose (5 mg/kg, 1 ml/kg) was injected 10 h after the first acivicin injection. Compound A animals received normal saline (2 ml/kg, adjusted to pH 7–8) followed 60 min later by compound A (0.2 mmol/kg in 0.9 ml corn oil). Animals received a second saline injection (1 ml/kg) 10 h after the first saline injection.
To assess the role of renal organic anion transport in compound A nephrotoxicity, animals were treated with probenecid, a competitive inhibitor of organic anion transport. The probenecid solution (200 mM) was prepared immediately before use in 1 M sodium hydroxide:2% acetic acid (3:5 ratio), sonicated, adjusted to a pH of 7.4–7.6 with acetic acid, and diluted to final volume with saline. A control saline and acetic acid buffer solution was prepared in the same manner, except that the probenecid was omitted. Control animals were treated with probenecid (0.5 mmol/kg, 2.5 ml/kg) followed 30 min later by corn oil (0.9 ml). A second probenecid injection (0.5 mmol/kg) was administered 10 h after the first probenecid injection. Compound A animals received blank saline and acetic acid buffer followed 60 min later by compound A (0.2 mmol/kg) in corn oil. The saline and acetic acid injection (2.5 ml/kg) was repeated 10 h after the first injection. Inhibitor-plus-compound A animals received probenecid (0.5 mmol/kg, 2.5 ml/kg) 30 min before compound A (0.2 mmol/kg) and a second probenecid injection (0.5 mmol/kg) 10 h after the first.
To assess the role of renal cysteine conjugate beta-lyase in compound A nephrotoxicity, animals were treated with AOAA, a competitive inhibitor of renal beta-lyase. The AOAA solution (200 mM) was prepared immediately before use in normal saline adjusted to a pH of 6–6.5 with 0.1 M hydrochloric acid. Control animals were treated with AOAA (0.5 mmol/kg, 2.5 ml/kg) followed 60 min later by corn oil (0.9 ml). A second AOAA dose (0.25 mmol/kg, 1.25 ml/kg) was injected 10 h after the first AOAA injection. Compound A animals received saline alone (2.5 ml/kg, adjusted to a pH of 6–6.5) followed 60 min later by compound A. Animals received a second saline injection (1.25 ml/kg) 10 h after the first saline injection. Inhibitor-plus-compound A animals were treated with AOAA (0.5 mmol/kg, 2.5 ml/kg) followed 60 min later by compound A in corn oil. A second AOAA dose (0.25 mmol/kg, 1.25 ml/kg) was injected 10 h after the first AOAA injection.
In all experiments, urine was collected on ice for 24 h after compound A or corn oil were administered, and the urine volume was recorded. The animals were weighed and anesthetized with ether. Blood was obtained by cardiac puncture. The left kidney was immediately excised, trimmed of perirenal fibrofatty tissues, cut in a mid-transverse plane through the cortex and medullary pyramid, and fixed in 10% neutral buffered formalin. Blood was centrifuged and serum frozen at -20 degrees Celsius in polypropylene vials for later analysis. Serum BUN and creatinine concentration, urine glucose, and urine total protein levels were measured spectrophotometrically using Sigma assay kits. Urine fluoride concentrations were measured using an ion-specific electrode. Urine alpha-glutathione transferase (alpha GST) concentrations were measured by enzyme immunoassay (Biotrin, Dublin, Ireland) using an antibody against rat YaYc GST.
Each of the fixed kidneys was prepared as follows: A 2–3-mm-thick parallel slice was removed from the original cut surface and embedded in paraffin. Each paraffin block was sectioned about 5 micro meter thick, one unwrinkled section was placed on a glass slide and stained with hematoxylin and eosin dye, and each slide was identified by a number and given to a pathologist blinded to the animal treatment.
Preliminary light microscopic findings from the dose-response studies were used to develop a system to analyze and score the tubular injury. Readily recognizable, necrotizing, proximal tubular lesions were observed. Although patchy and quantitatively variable, these lesions were located consistently in a zone between the deepest glomeruli and the transition from descending straight proximal tubules to descending limbs of Henle. This zone of injury corresponds to the “outer stripe of the outer medulla.” In each midtransverse section of kidney, the lateral edges of cortex enveloped but remained separate from the single medullary pyramid. The linear border between the outer and inner medullary stripes was histologically distinct. This border arched from one edge of cortex to the other, providing a deep line of demarcation for the zone of injury.
Each kidney section was examined in the following manner:(1) The zone of injury largely comprised proximal tubular profiles (more than 90% of the visible parenchyma), and its thickness was encompassed by one circular field of view (10x ocular lenses, 20x objective lens). (2) Using this field, ocular micrometer measurements showed that the total area of view (radius about 350 micro meter) and the usual cross-sectional area of a proximal tubule (radius about 20 micro meter) were about 0.4 mm2and 0.002 mm2, respectively; therefore, the estimated number of tubular cross sections per field was 200. (3) Beginning at one edge of cortex, the lateral and deep aspects of view were set tangential to the respective edge and line of demarcation between the outer and inner stripes; after viewing, this field was moved so that the next field was set tangential to the previous one and the subjacent line of demarcation; these moves were repeated sequentially until the last full field of view was reached at or near the opposite edge of cortex. The number of observational fields per kidney ranged from 9 to 20 (median, 14; mean, 15).
The severity of proximal tubular cell necrosis in each kidney was estimated in the following manner. (1) A quantitative estimate of lesion severity per field of view was made by counting the cross-cut necrotic tubular profiles directly and other profiles indirectly; the latter profiles were measured lengthwise, and the linear measurement was divided by the usual cross-cut tubular diameter (40 micro meter) to provide the final number; the total number of necrotic tubular cross cuts was then divided by 2 to give an initial percentage of necrotic versus intact tubular profiles per field, and this number was recorded. (2) The quantitative estimate of lesion severity per zone of injury was made by summing all of the recorded “percentages” and dividing by the number of fields examined.
All results are expressed as the mean +/- SE. Treatment groups were compared by analysis of variance followed by Student-Neuman-Keuls post hoc tests. Data with unequal variances or that were not normally distributed were log-transformed or analyzed using nonparametric tests. Significance was assigned at P < 0.05.
Compound A caused a distinct histopathologic renal injury, similar to that described after inhalational administration. [9,10,12]Previous investigators described the injury as occurring at the corticomedullary junction but did not localize the lesion to either the proximal or distal tubules. In our experiments, the injury was specifically confined to proximal tubular cells and located largely, if not entirely, within the outer stripe of the outer medulla. Lesions were patchy, but each patch included at least several entirely necrotic tubular cross sections. Normal proximal tubular cells were readily distinguished from other cortical and medullary tubular epithelial cell types, and this created a histologically distinct line of demarcation between the outer and inner stripes in the outer medulla. Proximal tubular cells were much larger, had unique apical brush borders surrounding small-caliber but patent lumina, and had more abundant and more eosinophilic cytoplasm. Their round basilar nuclei were similar to those of other epithelial cells, with finely dispersed, dark-staining, heterochromatin granules and relatively small but distinct nucleoli. In contrast to their nonnecrotic counterparts, the dead proximal tubular cells after compound A treatment showed coagulation necrosis with shrunken ghostlike cell forms, condensed hypereosinophilic cytoplasm, and either swollen and lysed or absent nuclear structures. Many of these cell forms appeared to be attached to tubular basement membrane, but combinations of sloughed necrotic cells, cell debris, and condensed proteinaceous substance filled many of the associated tubular lumina. Plugging of most necrotic tubular lumina by products of the necrotic process suggested little, if any, effective flow through these affected nephrons. We also observed a second type of renal injury. Some proximal tubular profiles adjacent to necrotic ones were lined by swollen, possibly injured, but seemingly intact cells around patent lumina; these profiles appeared more frequently in the more severely injured kidneys and suggested evidence of sublethal cell injury, but we did not include them in the quantitative estimate of lesion severity.
Previous investigations of compound A nephrotoxicity have used Wistar [9,11]or Sprague-Dawley rats. We used Fischer 344 rats, which have a different sensitivity to fluorinated nephrotoxins. Preliminary experiments therefore characterized the dose-response relationship for compound A metabolism and nephrotoxicity in Fischer 344 rats. Tubular necrosis (greater or equal to 0.1% of tubules) was observed in none of five, one of five, two of five, and three of four rats receiving 0, 0.1, 0.2, and 0.3 mmol/kg compound A, respectively. The severity of injury, as well as the incidence, was substantially greater in the animals receiving 0.3 mmol/kg. Compound A also caused biochemical evidence of nephrotoxicity, characterized by dose-related increases in urine volume and in urine glucose, protein, and alpha GST excretion (Figure 2). alpha-glutathione-S-transferase is a cytosolic enzyme localized in proximal and possibly distal tubular cells. [30,31]Urinary alpha GST excretion was used previously as a sensitive marker of renal tubular cell necrosis. [31,32]It was also a sensitive marker of dose-related compound A nephrotoxicity in our investigation. The dose-dependent proteinuria and glucosuria were consistent with those seen previously with compound A-related nephrotoxicity. The compound A-induced increases in urinary volume and urinary glucose, protein, and alpha-GST excretion probably arise from the nonnecrotic tubules with injured but viable cells and persistent tubular flows described previously. Serum BUN and creatinine concentration were less-sensitive markers of renal injury than were urinary glucose, protein, or alpha GST excretion (Table 1). The BUN level was unaltered and creatinine was only marginally increased at the highest compound A dose. Levels of serum aspartate aminotransferase and alanine aminotransferase, indicative of hepatic injury, were elevated only at the highest compound A dose, which also resulted in histologic evidence of subcapsular hepatocyte necrosis. Urinary fluoride excretion, reflecting P450-dependent compound A metabolism, formation of compound A-glutathione conjugates, or both, and their subsequent metabolism (Figure 1) increased in a dose-dependent manner, indicating increasing compound A metabolism. Based on the histologic injury and biochemical markers, 0.2 mmol/kg compound A was used in subsequent experiments. The degree of tubular cell necrosis after 0.2 mmol/kg compound A approximated that caused by 100–200 ppm inhaled compound A in Sprague-Dawley and Wistar rats [10–12]and by 100–150 ppm in Fischer 344 rats (results not shown).
Enzyme Inhibition Experiments
Probenecid treatment before compound A administration significantly diminished compound A nephrotoxicity, as evident by both histologic and biochemical markers, compared with animals given compound A alone. We found no evidence of tubular necrosis in any rat given probenecid before compound A (Table 2). Probenecid pretreatment also ameliorated compound A-induced increases in urinary glucose, protein, and alpha GST excretion (Figure 3). Probenecid pretreatment did not prevent compound A-related diuresis.
Treatment with AOAA before compound A administration prevented the compound A-induced increases in urine volume and urinary protein excretion and significantly diminished the increase in urinary alpha GST elimination (Figure 4).** Urinary glucose excretion was somewhat decreased in animals pretreated with AOAA that received compound A, compared with those receiving compound A alone, but the decrease was not statistically significant. Serum BUN and creatinine concentrations in animals receiving compound A alone or after AOAA were not different from those of control animals (Table 3). Pretreatment with AOAA did not, however, significantly alter compound A-related renal tubular cell necrosis (Table 2). Compound A-related urinary fluoride excretion was unchanged in rats pretreated with AOAA (Table 4).
The effects of AOAA on compound A nephrotoxicity were particularly sensitive to the relative doses of compound A and AOAA used. When animals received a larger dose of compound A (0.25 mmol/kg) after a single pretreatment with AOAA, there was no protective effect on any biochemical marker of renal toxicity compared with compound A alone; urine volume and excretion of protein, glucose, and alpha GST were unchanged (data not shown).
Acivicin treatment before compound A was administered significantly exacerbated compound A nephrotoxicity, as evidenced by both histologic and biochemical markers, compared with animals given compound A alone (Figure 5;Table 2and Table 3). All rats given acivicin before compound A had extensive tubular lesions (30–60% of tubules were necrotic); significant increases in urine volume; significant increases in urinary glucose, protein, and alpha GST excretion; and significant increases in serum BUN and creatinine concentrations. Compound A-related urinary fluoride excretion was unchanged in rats pretreated with acivicin (Table 4).
Bioactivation of many fluoroalkenes and chloroalkenes results in localized cytotoxicity in the S3segment of renal proximal tubules. The complex scheme, collectively called the beta-lyase pathway, has been elucidated for myriad haloalkenes and involves the concerted participation of several organs and enzyme systems. [22,23,33–35]Haloalkenes are initially metabolized to glutathione (gamma-glutamyl-cysteinylglycine) conjugates. Conjugation is usually catalyzed by glutathione-S-transferase (primarily hepatic) but may also occur nonenzymatically. Glutathione conjugates are excreted in bile and cleaved sequentially by gamma-glutamyltransferase and dipeptidases in the bile duct and small intestine into corresponding cysteine conjugates; glutathione conjugates are also secreted into the systemic circulation. Cysteine conjugates in the intestine reenter the systemic circulation. Cysteine conjugates reabsorbed by the liver may be N-acetylated to the corresponding mercapturic acids by hepatic N-acetyltransferase. Circulating glutathione and cysteine conjugates and mercapturic acids are actively transported into proximal tubular cells by a basolateral probenecid-sensitive organic anion transporter. Proximal tubule cells have an abundance of gamma-glutamyltransferase, and the kidney is an important site of glutathione conjugate cleavage into cysteine conjugates. Intracellularly, the cysteine conjugates may be N-acetylated to mercapturates and excreted uneventfully in the urine, or they may undergo beta-lyase-mediated bioactivation to highly reactive intermediates. Mercapturates are not substrates for beta-lyase but may be intrarenally deacetylated back to corresponding cysteine conjugates and subsequently activated by beta-lyase. A critical step of the toxification pathway involves beta-lyase cleavage of fluorinated cysteine conjugates to highly reactive thioketenes and thionoacyl fluorides that alkylate cellular macromolecules. These reactive intermediates target mitochondria and DNA and are the presumed causes of proximal tubular cell necrosis. Both glutathione and cysteine conjugates may be nephrotoxic.
Implication of the beta-lyase pathway in the toxicity of compound A, or any haloalkene, requires evidence for the formation of the required intermediates and evidence for their bioactivation by the essential enzymes of the pathway. Considerable evidence for the formation of compound A-glutathione conjugates, the cleavage of these conjugates to corresponding cysteine conjugates, and the metabolism of compound A-cysteine conjugates by beta-lyase in rats has been obtained. [14–17]Four compound A-glutathione conjugates were isolated in rat bile: two fluoroalkenes and two fluoroalkenes. The fluoroalkanes were diastereomers of S-[1,1,-difluoro-2-fluoromethoxy-2-(trifluoromethyl)ethyl]glutathione and the fluoroalkenes were (E)- and (Z)-S-[1-fluoro-2-fluoromethoxy-2-(trifluoromethyl)vinyl]glutathione, all formed predominantly by microsomal, rather than cytosolic, glutathione transferases (Figure 1). These metabolites were cleaved by gamma-glutamyltransferase and dipeptidase to the corresponding cysteine conjugates, as evident from the appearance of the corresponding mercapturates in urine. The metabolism of compound A-cysteine conjugates by rat beta-lyase in vitro and in vivo has been demonstrated unambiguously by identification of the organic acid hydrolysis products of thioketene and thionoacyl fluoride metabolites, the intermediates predicted to result from beta-lyase-mediated cleavage. [16,17]Covalent binding of compound A or its reactive metabolite(s) to rat kidney macromolecules also has been shown. Human renal beta-lyase also catalyzes the metabolism of compound A-cysteine conjugates. 
The current investigation provided evidence that certain key enzymes of the beta-lyase pathway mediate compound A nephrotoxicity. Necrosis was localized to the proximal tubule, where mitochondrial injury was the earliest feature of compound A tubular cell toxicity, characteristic of beta-lyase-mediated nephrotoxins. Pro-benecid was a highly effective inhibitor of compound A nephrotoxicity, as evident by the absence of tubular cell necrosis in all pretreated animals and prevention of compound A-related increases in urinary glucose, protein, and alpha GST excretion. Probenecid abolition of compound A nephrotoxicity suggests the participation of the renal tubular cell organic anion transport system in compound A toxification, specifically the renal uptake of compound A-glutathione, cysteine conjugates, or both. Probenecid prevention of nephrotoxicity from many other haloalkanes, cysteine conjugates, and glutathione conjugates has also been demonstrated, indicating a role for the renal organic anion transport system. [22,27]Amelioration by AOAA of compound A-related increases in urine volume, and of urinary glucose, protein, and alpha GST excretion supports the participation of renal cysteine conjugate beta-lyase in compound A-induced nephrotoxicity. Previously AOAA (0.5 mmol/kg, the dose we used) was shown to inhibit 90% of renal beta-lyase activity 1 h after in vivo administration. Thus compound A resembles many other difluorovinyl nephrotoxins, whose toxification via beta-lyase has been established from AOAA inhibition of diuresis, proteinuria, and glycosuria. [22,27]Compound A is also similar to bromochlorodifluoroethene, the difluorovinyl product of halothane breakdown by carbon dioxide absorbents, which is nephrotoxic in rats via the beta-lyase pathway.  Figure 6shows a general scheme for compound A metabolism and toxicity.
Aminooxyacetic acid protected against the biochemical manifestations of compound A nephrotoxicity but not against the histologic pattern of necrosis. It may have ameliorated the sublethal proximal tubular cell injury (thereby preventing both alpha GST leakage and diminished glucose and protein reabsorption) but did not prevent the necrosis of lethally injured cells. Because necrotic and plugged tubules without effective flow would not be expected to leak alpha GST, glucose, and protein, AOAA protection against biochemical but not histologic changes might be expected. We observed, as did others, diminished AOAA protection against compound A with increasing severity of tubular injury. This was also observed with other beta-lyase-mediated nephrotoxins. Aminooxyacetic acid more effectively protects against beta-lyase-mediated toxicity in vitro than in vivo and against the toxicity of S-conjugates compared with their parent haloalkenes. [41–47]The effects of AOAA on the toxicity of compound A S-conjugates in vitro and in vivo merit investigation.
Acivicin exacerbation of compound A nephrotoxicity suggests that compound A-glutathione conjugates may be toxic, that a metabolic pathway not involving glutathione conjugation also mediates nephrotoxicity, and that acivicin may affect a process distinct from metabolism and thereby alter the toxicologic response. Although acivicin inactivation of gamma-glutamyl transferase is incomplete and reversible, which might explain a failure to prevent compound A nephrotoxicity (as was seen with 1,1-dichloroethylene ), this seems an unlikely explanation for acivicin exacerbation of toxicity. Rat kidney gamma-glutamyltransferase activity is inhibited 95–98% 1–2 h after 10 mg/kg acivicin, the dose we used. Renal gamma-glutamyltransferase activity recovers to 10–20% of control by 16 h, and to 40% of control 24 h after acivicin. [51,52]It was shown previously that a second acivicin injection provides greater inhibition than does a single injection. Thus we used a second injection of acivicin to inhibit any new enzyme being synthesized and to maximize enzyme inhibition. One previous investigation also noted exacerbation of nephrotoxicity by acivicin. Hexachlorobutadiene is a potent beta-lyase-mediated nephrotoxin, forming a hepatic glutathione conjugate that is excreted in bile and cleaved by gamma-glutamyl transferase. [53,54]Acivicin pretreatment, however, markedly increased the renal toxicity of hexachlorobutadiene. The mechanism of acivicin exacerbation of toxicity, and the renal effects of compound A-glutathione conjugates are unknown.
Another potential mechanism of compound A nephrotoxicity may be related to inorganic fluoride ion liberated during compound A metabolism. P450-mediated metabolism, glutathione conjugation of compound A with subsequent dehydrofluorination, [14,15]and beta-lyase-mediated metabolism of cysteine conjugates result in fluoride formation (Figure 1). Fluoride in sufficiently high doses is nephrotoxic, and methoxyflurane nephrotoxicity is thought to arise from methoxyflurane defluorination. The present results show that urinary fluoride excretion after compound A was not substantially different under conditions in which acivicin, probenecid, and AOAA markedly altered compound A nephrotoxicity. The discrepancy between fluoride excretion and compound A toxicity supports the hypothesis that compound A nephrotoxicity is not mediated by inorganic fluoride.
One potential limitation of our investigation is that compound A was administered intraperitoneally, but it was administered by inhalation in previous rat experiments, [9–12,39]and human clinical compound A exposure occurs by inhalation. Inhalational administration of compound A to mimic clinical conditions requires sophisticated rodent nose-only inhalation exposure apparatus not routinely available in most laboratories. Only one investigation used this approach. All other compound A studies with inhalational administration have used a total-body exposure chamber in which there is potential rebreathing of expired gas, absence of carbon dioxide scavenging, and, most importantly, uncontrolled cutaneous absorption of compound A. Total-body exposure may not be optimal. Our goal was to develop and use a model that is more accessible to subsequent investigators and that produces the characteristic compound A lesion. Indeed, we accomplished that goal. Furthermore, many investigations with haloalkanes and haloalkenes have shown that mechanisms of nephrotoxicity are independent of whether drug administration is by intraperitoneal injection or inhalation.
Our conclusions and those of previous investigations [14–17]can be compared with those of Martin and associates, who investigated the effects of acivicin and AOAA on compound A nephrotoxicity. Like us, they found that acivicin potentiated compound A nephrotoxicity. The inhibitory effects of probenecid on compound A nephrotoxicity were not studied. In contrast to our results, Martin and associates found no effect of AOAA on compound A nephrotoxicity and concluded that the cysteine S-conjugate-mediated pathway is not the mechanism of compound A nephrotoxicity. Some important methodologic differences may explain this discrepancy. Most important may be the dose of nephrotoxin used. We administered compound A at the threshold dose of 0.2 mmol/kg, whereas Martin and associates administered 150 ppm compound A, which was three times their threshold dose of 50 ppm. The renal injury caused by 3 h of 150 ppm compound A (19% of tubules) was greater than that caused by our threshold dose of 0.2 mmol/kg (0.7–8% of tubules). Because AOAA is a competitive inhibitor of beta-lyase, the relative concentrations of compound A conjugates (substrate) and AOAA (inhibitor) will determine the degree of beta-lyase-mediated metabolism, and excessive doses of compound A may overwhelm enzyme inhibition. Additional experiments confirmed this possibility. Two doses of AOAA significantly diminished compound A (0.2 mmol/kg)-dependent increases in urine volume, protein, and alpha GST excretion (Figure 5). In contrast, when only one dose of AOAA was administered, diuresis and proteinuria were significantly but only partially diminished and alpha GST excretion was unaffected. Furthermore, when the dose of compound A was increased from 0.2 to 0.25 mmol/kg, AOAA pretreatment had no effect on volume or on glucose, protein, and alpha GST excretion. A previous investigation also showed that the dose of beta-lyase-activated nephrotoxin markedly affects the ability of AOAA to inhibit toxicity. The nephrotoxicity of 1,1-dichloro-2,2-difluoroethylene, when administered at the threshold dose of 150 micro mol/kg, was completely prevented by AOAA. In contrast, when 1,1-dichloro-2,2-difluoroethylene was administered at four times the threshold dose (600 micro mol/kg), AOAA had no effect on nephrotoxicity. Species-specific effects of AOAA may also partly explain the results of Martin and associates, who used Wistar rats, whereas we used Fischer 344 rats. Aminooxyacetic acid is markedly toxic in Wistar but not in Fischer 344 rats. Thus the absence of an inhibitory effect of AOAA on the nephrotoxicity of suprathreshold doses of compound A in Wistar rats observed by Martin and associates may not support their conclusion discounting the role of S-conjugates or beta-lyase in compound A nephrotoxicity.
Mediation of compound A nephrotoxicity by the renal cysteine conjugate beta-lyase pathway may have important implications regarding interspecies differences in compound A effects. Human renal beta-lyase activity and beta-lyase metabolism of compound A cysteine conjugates are approximately 8–30 times less than that in rat kidneys. [17,56,57]Other interspecies differences in bioactivating or detoxifying beta-lyase pathway enzyme activities may also exist. 
We provided experimental evidence implicating the renal organic anion transport mechanism and renal cysteine conjugate beta-lyase activity in the mechanism of compound A-related nephrotoxicity in rats. In conjunction with previous data showing the formation of compound A-glutathione and compound A-cysteine conjugates and cysteine conjugate metabolism by beta-lyase, these results suggest that compound A nephrotoxicity in rats is mediated, in part, via the classical beta-lyase pathway.
The authors thank Professor M. W. Anders for his many helpful discussions on potential mechanisms of compound A toxicity.
*Fluoromethyl-2,2-difluoro-1-(trifluoromethyl)vinyl ether is also called pentafluoroisopropenyl fluoromethyl ether.
**In a previous investigation, AOAA alone had no effect on any biochemical or histologic measure of renal injury compared with saline controls.