Glutathione transferases in the urine: sensitive methods for detection of kidney damage induced by nephrotoxic agents in humans.

With the aid of immunohistochemical methods the localization of the various isoenzymes of glutathione S-transferase was investigated. The alpha isoenzyme was present solely in the proximal tubular cells of the human kidney, while the pi form was restricted to the distal convoluted tubules, the thin loop of Henle, and the collecting ducts. Damage to the epithelial cell membranes results in the increased excretion of these enzymes with the urine. The alpha and pi isoenzymes have been isolated in a highly purified form and used for the production of polyclonal antisera. Subsequently, radioimmunological and ELISA techniques were developed for quantitation of these proteins in the urine; the methods exhibited a high specificity and were sufficiently sensitive to determine nanogram quantities or less. Disease affecting tubular function, cyclosporine A treatment, administration of nephrotoxic antibiotics, and exposure to cadmium all resulted in characteristic changes in the pattern of the glutathione transferase isoenzymes present in urine. Such effects were seen also in patients who had previously been exposed to nephrotoxic agents, but in whom conventional tests for kidney function were apparently normal. Thus, it appears that radioimmunologic or immunochemical quantitation of alpha and pi forms of the enzyme can be used as sensitive and relatively simple markers for the early detection of toxic effects with respect to the renal tubuli. ImagesFigure 1. AFigure 1. BFigure 3.


Introduction
In several organs-mainly heart and liver-cell damage is followed by release of~~*7 a number of cytoplasmic enzymes to the blood, a phenomenon that provides the basis for clinical diagnosis of heart and liver 1.Ia disease. Although attempts in this direction have been made previously using, for example, the Tamm-Horsfall glycoprotein, cor-,.
responding assays for kidney damage have, W* for various reasons, had little success (1).. Three major classes of human isoen-OWi zymes of glutathione S-transferases(ic gandin, GST) have been identified; ac, it, t;4i and Soh Human kidney contains the 8 4 and 4 obta in eila co forms in relatively high amounts. Although there have been some peiuatmts to....,ũ se GST as a marker of kidney damage (2/),.~ẽ arlier methods have not been sufficiently specific and sensitive to be clinically func-  Isolation of the a and isoenzymes of GST from human liver and placenta, the RIAand production and purification of polyclonal Using es antibodies in rabbit, development of RIA was immunochemical procedures for quantita-Routine tion, and clinical diagnosis and selection of regularly patients have been described elsewhere in urine. A. detail (3)(4)(5). ELISA fc Purification ofGST oped. Th Highly purified a and isoenzymes were that quz obtained by using chromatographic proce-ng/ml is dures involving affinity chromatography antibodi with hexylglutathione coupled to epoxy-the titrat activated Sepharose as the ligand as a final step. Upon chromatography the purified Speafi proteins behaved as single components of The spec 27,000 and 25,000 Da for the a and it with resi isoenzymes, respectively.  (6), was studied in ical sections using the appropriate (4). In fresh kidney biopsies, oc found exclusively in the proximal qigure 1A), whereas the Xt enzyme . B), absent from this tubular section, ed to the distal convoluted tubulus thin loops of Henle and the col-Lucts.
i ELISA tablished principles, a competitive developed for both GST proteins. measurements could be performed in the range 2 to 1000 ng/ml in .n even more sensitive solid-phase or both isoenzymes was also develhe titration curves for the a (Figure I t-GST ( Figure 2B) demonstrate antitation down to around 0.5 ;feasible. The high quality of the es is indicated by the steep slope of tion curves. lity cificity of the antibodies was tested pect to the three isoenzymes, a, it, .nd with regard to other proteins  that may be present in human urine, such as albumin, fibrinogen, prothrombin, hemoglobin and P-2-microglobulin. As can be seen from Table 1, the antisera of a and x proteins, reacted exclusively with the corresponding enzyme with no detectable cross-reactivity.
The same was true for several other tested proteins and enzymes. The specificity could also be confirmed by employing Western blot analysis of the cytosol prepared from various human organs. Figures  3A and 3B demonstrate that each antibody only reacts with one single protein band, also in preparations from tissues in which both isoenzymes are present. The band width is proportional to the relative levels of the isoenzymes present in liver, kidney, lung, and adrenals, respectively. Stability Samples kept in the refrigerator at 40 C for 5 days showed no significant decrease in enzyme content ( Figure 4A). Storage at room temperature resulted in a gradual decrease, but after 5 days 35% of the original amount could still be detected. As seen in Figure 4B, stability at room temperature is not appre-

Drug Nephrotoxicity
The utility of GST isoenzymes as markers for kidney damage was initially studied in patients who had undergone kidney transplantation. In normal healthy adults about 6 ng/ml of the it protein is found in the urine, whereas the amount of the a enzyme is hardly measurable ( Table 2). Rejection after transplantation greatly increased the excretion of 7t enzyme in the urine without affecting the levels of the a protein.
The drug Cyclosporine A (CsA), used as immunosuppressive agent in association with renal transplantation, as well as several aminoglycoside antibiotics are known to induce tubular damage (7). Upon treatment with CsA, considerable amounts of ax-GST are found in the urine, while i-GST remains unaffected. As expected, renal infarction, which destroys various types of kidney tissue, results in a more than 100-fold increase of both enzymes in the urine. Figure 5 shows the levels of a-GST in a patient treated with the aminoglycoside Netilmycin. After treatment for 5 days, the enzyme concentration in the urine was increased 15-fold. However, no changes in the serum creatinine level could be observed.

Heavy Metal Nephrotoxicity
Preliminary studies of cohorts of workers exposed to nephrotoxic heavy metals have indicated that quantitation of GST proteins in the urine constitute a valuable diagnostic tool. In Figure 6 the t-GST levels have been recorded in 15 accumulator workers who had been exposed to cadmium by inhalation 5 to 10 years previously, and who at the time of our investigation were diagnosed as clinically healthy using conventional criteria. However, in comparison with the control group, a continued increased excretion of the tubular marker, a-GST, was still evident. Currently, workers in Sweden and Poland with a history of exposure to high levels of lead, cadmium, and mercury are being investigated to characterize the doseresponse relationship for exposure to nephrotoxic heavy metals as well as the resulting changes in the GST isoenzyme pattern in urine.

Discussion
In the case of tubular dysfunction, the measurement of plasma proteins like P-2microglobulin and the retinol-binding protein (a-2-microglobulin) have so far provided the most sensitive indices of tubular damage. The instability, especially in acid urine, constitutes the main problem with assays based on ,-2-microglobulin. Although the retinol-binding protein is more stable, it is not a constitutive protein of the kidney, but is synthesized in liver; its presence in urine only provides indirect evidence of tubular damage (1). The  10 A Control Cd-exposed N=25 N=15 Figure 6. a-GST levels in Swedish workers with a 6 to 14 year history of cadmium exposure. Exposure to cadmium ceased 4 to 8 years prior to measurement. When assayed they were clinically diagnosed as healthy. Controls represent age-matched nonexposed individuals.
in whom cadmium exposure had been stopped many years before is consistent with previous observations based on determination of P-2-microglobulin in cadmium exposed workers (8). Our investigations have demonstrated that the determination of the isoenzymes of GST using RIA and ELISA techniques provides a reliable and sensitive alternative to conventional methods. The most important advantages of this technique are: x and 7i GST isoenzymes are constitutive proteins of the kidney. The markers are stable in urine. The method is simple, sensitive and reproducible. Due to selective localization of the two isoenzymes, differential diagnosis is possible.