Gentamicin Inhibits Renal Protein and Phospholipid Metabolism in Rats: Implications Involving Intracellular Trafficking

Home , Gentamicin, Netilmicin

J Am Soc Nephrol 12: 114–123, 2001 Gentamicin Inhibits Renal Protein and Phospholipid Metabolism in Rats: Implications Involving Intracellular Trafficking

DAVID P. SUNDIN, RUBEN SANDOVAL, and BRUCE A. MOLITORIS Department of Medicine, Division of Nephrology, Indiana University School of Medicine, and the Roudebush Veterans Affairs Medical Center, Indianapolis, Indiana.

Abstract. Studies were undertaken to characterize the mecha- increase in phosphatidylinositol levels was observed in cortical nism of aminoglycoside-induced nephrotoxicity. Early time homogenates from treated rats. However, gentamicin treatment points in gentamicin treatment (1 to 3 d) were used to inves- was demonstrated to increase the levels of phosphatidylinositol tigate the development of toxic events without the complica- and phosphatidylcholine, while decreasing the level of sphin- tion of gross morphologic cellular alterations. Enzyme activi- gomyelin (SPH), in BBM. Incorporation of 32P into SPH, ties of cortical homogenates and brush border membrane phosphatidylserine, and phosphatidylethanolamine was inhib- (BBM) preparations documented little effect on specific activ- ited in cortical homogenates from gentamicin-treated animals; ities or the ability to isolate representative membrane fractions. among BBM phospholipids, however, a significant decrease In vivo protein synthesis experiments demonstrated that gen- was observed only for SPH synthesis. It was concluded that tamicin reduced cellular protein synthesis after 2 d of treat- inhibition of phospholipid degradation was quantitatively the ment. This inhibition increased to 50% on the third day. Total major contributor to the effects of gentamicin on phospholipid cellular protein synthesis was inhibited to the same extent as metabolism. Confocal microscopic studies, using tracer BBM protein synthesis. However, gentamicin had different amounts of fluorescently labeled gentamicin, revealed genta- effects on homogenate versus BBM phospholipids. The total micin in large, mostly basal structures. Correlative electron phospholipid contents in cortical homogenates and BBM from microscopic studies, using photo-oxidation techniques, dem- treated animals were increased, compared with control ani- onstrated that these structures consisted of lysosomal, Golgi mals. A significant decrease in phospholipid synthesis was complex, and mitochondrial structures. These observations observed only in homogenates from treated animals. When suggest retrograde trafficking of gentamicin and indicate a effects on individual phospholipids were investigated, only an general mechanism of gentamicin-induced nephrotoxicity.

Aminoglycoside antibiotics are widely used in the treatment nephrotoxicity remain unclear. Alterations in phospholipid me- and prevention of Gram-negative bacterial infections. Their tabolism after aminoglycoside treatment have been described bactericidal activity is derived from their ability to bind pro- by several laboratories, including our own (11–15). These karyotic ribosomes. This action blocks the ribosomal initiation changes occur rapidly and persist for extended periods, even complex and/or causes mistranslation (1,2). As a result, protein after recovery, in the rat model (11,14–17). Inhibition of synthesis is inhibited or deranged, causing bacterial death. lysosomal phospholipases, subsequent accumulation of phos- Aminoglycoside use can be complicated by well described pholipids, and formation of lysosomal myeloid bodies have nephrotoxic and ototoxic effects (3–5). Complications attrib- been implicated as direct mechanisms of nephrotoxicity utable to aminoglycoside toxicity rank as one of the most (11–13,16,18). common reasons for prolonging hospital stays in the developed Accumulating evidence has also suggested aminoglycoside- world (6). Nephrotoxic effects are most common and develop induced inhibition of protein synthesis as a mechanism of after relatively short periods of treatment (7,8). In many cases, nephrotoxicity. It has been demonstrated that elevated concen- these effects are reversible if administration is discontinued. trations of aminoglycosides can cause mistranslation or block Although many hypotheses have been proposed and tested incorporation of amino acids by eukaryotic ribosomes in vitro (3,9,10), the precise mechanisms of aminoglycoside-induced (19–21). In addition, concentrations of gentamicin attained in the rat renal cortex during in vivo treatment were demonstrated to inhibit subsequent in vitro microsomal protein synthesis Received April 26, 1999. Accepted June 15, 2000. (19,20,22). This effect was observed before a decline in renal Correspondence to Dr. Bruce A. Molitoris, Indiana University School of function or significant alterations in cellular enzyme activities Medicine, 1120 South Dr., Fesler Hall, Room 108, Indianapolis, IN 46202. Phone: 317-274-5287; Fax: 317-274-8575; E-mail: [email protected] and morphologic features were observed. In large part, aminoglycoside effects on protein and phos- 1046-6673/1201-0114 Journal of the American Society of Nephrology pholipid synthesis and metabolism have not been determined in Copyright © 2001 by the American Society of Nephrology completely in vivo experiments. Therefore, the purpose of the J Am Soc Nephrol 12: 114–123, 2001 Gentamicin Inhibits Renal Metabolism 115 studies described here was to determine the effect of amino- tol mounting medium (Aldrich Chemical Co., Inc., Milwaukee, WI); glycosides on these factors in completely in vivo experiments. coverslips were sealed with clear acrylic nail polish. TRG staining Studies were performed at time points (1 to 3 d) when little or was examined by confocal microscopy (see below). no alteration of cellular enzyme activities, morphologic features, or function was observed. Correlative confocal micro- Brush Border Membrane (BBM) Preparation and scopic and electron microscopic (EM) studies were also per- Characterization formed and suggested a role for intracellular trafficking in Control and gentamicin-treated rats were lightly anesthetized with aminoglycoside-induced nephrotoxicity. metafane, decapitated, and exsanguinated, and their kidney cortices were dissected and homogenized. Homogenates for all rats were prepared using exactly the same volume of buffer and as close to the Materials and Methods same amount of tissue as possible. The beginning and ending weights Conjugation of Texas Red and Gentamicin did not vary by Ͼ5% between control and STG-treated rats in any Conjugation of Texas Red to gentamicin was performed as we experiment. BBM were isolated from cortical homogenates using described previously (23). All steps were performed at 4°C. Texas Mg2ϩ precipitation and differential centrifugation techniques, as de- Red (sulfonyl chloride form; Molecular Probes, Eugene, OR), at 20 scribed in detail elsewhere (15,26–29). Relative enrichments of BBM mg/ml in dimethylformamide (Sigma Chemical Co., St. Louis, MO), preparations were documented enzymatically using the BBM marker ϩ was added to a rapidly stirred solution of 0.1 M K2CO3, pH 8.5, and leucine aminopeptidase and the basolateral membrane marker Na / 5 to 10 mg/ml gentamicin (Fluka Biochemika, Ronkonkoma, NY) and Kϩ-ATPase, as described previously (15,27–29). Marker enzymes was incubated for 4 to5hat4°C. A gentamicin/Texas Red molar ratio used to determine contamination by other intracellular organelles of 10:1 was used, to minimize formation of multiply substituted Texas were succinate dehydrogenase (mitochondria) and KCN-resistant Red-gentamicin (TRG) conjugates. After conjugation, the reaction NADH dehydrogenase [endoplasmic reticulum (ER)], as described mixture was centrifuged (20 min at 2000 ϫ g at 4°C), and the previously (15,27,28). supernatant was removed, placed in a 1000-D molecular mass cut-off dialysis membrane, and dialyzed extensively against 0.9% NaCl/10% Biosynthetic Labeling of Proteins and Phospholipids sucrose. The concentration of the final product was estimated using Control and gentamicin-treated (1 to 3 d) rats were given intraperi- the molar extinction coefficient of Texas Red. Finally, the TRG was toneal injections, on the day of euthanasia, of either 0.1 mCi of filter-sterilized and stored at 4°C until used. [3H]leucine or 0.5 mCi of 32P (Amersham, Arlington Heights, IL), both in 0.5 ml of normal saline solution. [3H]Leucine was injected 1 h Animal Protocol before euthanasia, and 32P was injected 2 h before euthanasia. After Male Sprague-Dawley rats (200 to 250 g; Harlan, Indianapolis, IN) euthanasia, cortical homogenates were made and BBM were prepared were used for all experiments. They were fed standard rat chow and from them as described above. In both cases, aliquots were added to allowed free access to food and water during the experiments. Rats scintillation fluid (Beckman, Fullerton, CA) and counted in a Beck- were given intraperitoneal injections of either gentamicin (gentamicin man LS1801 ␤-counter. Incorporation of radioactivity into protein or sulfate, 40 mg/ml; Elkins-Sinn, Cherry Hill, NJ), at a dose of 100 lipid was then determined as described below. We previously used mg/kg [short-term gentamicin (STG)], or vehicle (vehicle control) for similar procedures to quantify phospholipid synthesis (27). 1 to 3 d. In all cases, rats were given injections for the indicated times and euthanized the next day. The dosage used was previously dem- Protein Determinations onstrated, by our laboratory and others, to induce reproducible and Protein concentrations were determined using the method of Lowry reversible nephrotoxicity (12,15,16,24). The times chosen ensured the et al. (30) or the Bradford dye binding assay (31), and synthesis was occurrence of only mild toxicity (22,25). Serum creatinine levels were expressed as counts incorporated per milligram of protein. Incorpo- measured for the 3-d gentamicin-treated rats, as a measure of toxicity. ration of [3H]leucine into protein was determined by TCA precipita- Additional measured parameters were enzyme activities and mem- tion. Samples (homogenates or BBM) were incubated on ice in 10% brane lipid contents (see below). TCA for Ͼ1 h and were filtered using Millipore filtration techniques we described previously (29,32). Each filter was rinsed, suspended in TRG Studies scintillation cocktail, and counted. Alternatively, TCA-precipitated ϫ In these studies, tracer amounts of Texas Red-labeled gentamicin samples were centrifuged at 16,000 g for 10 min, and the TCA (i.e., 354 ng TRG/mg regular gentamicin, 0.04% TRG) were included pellet was solubilized with 1 N NaOH, mixed with scintillation in the normal gentamicin solution (see above). Rats that received this cocktail, and counted. solution were treated exactly like those given gentamicin only. Only the 3-d time point was used in these studies. Rats were anesthetized Lipid Determinations with pentobarbital, the abdominal cavity was opened, the portal vein These determinations were performed as we described in detail was catheterized, and phosphate-buffered saline (PBS) infusion was previously (15,27,28). Briefly, lipids from 0.5 to 1.0 mg of cortical begun. The vena cava was then cut, and the vascular tree was flushed homogenate or BBM were extracted with chloroform/methanol (1:2, by continued perfusion with PBS (approximately 200 ml). Fixation of vol/vol) and isolated according to the method of Bligh and Dyer (33). the tissue was accomplished by perfusion with 180 to 200 ml of Total phospholipid levels were determined using the method of Ames freshly prepared 4% paraformaldehyde in PBS. Kidneys were then and Dubin (34), and individual phospholipid species were separated decapsulated and cut into approximately 5-mm3 pieces, and fixation by two-dimensional thin layer chromatography, using a modification was continued overnight at 4°C in the same fixative. The next day, (27) of the method of Esko and Raetz (35). Incorporation of 32P into kidney pieces were washed three times with PBS and placed in 0.25% total cellular membranes, BBM, and individual lipid species was paraformaldehyde/PBS until sectioning. Sections of 50 ␮m were determined by counting aliquots after chloroform/methanol extraction obtained and either photo-oxidized (see below) or mounted in Gelva- and TCA precipitation, as described for filters above (homogenates 116 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 114–123, 2001 and BBM), or counting resuspended scrapings of individual lipid Table 1. Effects of gentamicin treatment on homogenate and species after two-dimensional thin layer chromatography (individual BBM protein concentrationsa phospholipid species; see above). Protein Concentration (mg/ml)

Confocal Microscopy Homogenates BBM TRG conjugate fluorescence was examined using a Bio-Rad MRC- 1024 scanning laser confocal microscope (Bio-Rad, Richmond, CA), Control 8.07 Ϯ 0.89 (15) 2.34 Ϯ 0.33 (15) with a Nikon ϫ100 1.4-NA PLAN APO objective (Nikon, Natick, 1-d STG 8.27 Ϯ 0.61 (4) 2.34 Ϯ 0.18 (4) MA). Samples were illuminated using an argon-krypton laser. We 2-d STG 8.86 Ϯ 0.77 (8)b 1.81 Ϯ 0.28 (8)b successfully used this system in previous studies (15,23). 3-d STG 8.19 Ϯ 0.98 (14) 2.22 Ϯ 0.38 (14)

a Photo-Oxidation of TRG and EM Rats were treated with vehicle or gentamicin for 3 d. The day after the last injection, cortical homogenates and brush border The protocol for and mechanism of photo-oxidation of ringed membrane (BBM) preparations were prepared and protein structures in the presence of diaminobenzidine (DAB) have been concentrations were determined as described in Materials and described by others (36). We previously described our modifications Methods. The n values are indicated by values in parentheses. of the photo-oxidation reaction (23). An excitation filter (filter b Statistical significance of P Ͻ0.03 and P Ͻ0.0001 for 2-d 560DF40; Optical Omega, Brattleboro, VT) was positioned directly short-term gentamicin (STG)-treated homogenates and BBM, over sections (see above) immersed in DAB substrate solution (Vector respectively. Laboratories, Burlingame, CA) at 4°C. A mercury lamp suspended 30 cm above the filter cube was used to excite the sample with 560-nm light for 2.5 min three times. After each excitation, fresh 4°C DAB Statistical Analyses solution was added to the sample well. Reactive species produced by Values are reported as means Ϯ SD. Data were analyzed using the photo-oxidation of the Texas Red fluorophore react with DAB to form paired and unpaired t test and the two-sample t test, assuming unequal a dense EM-observable reaction product wherever TRG is present. variances. Differences were considered statistically significant at P Ͻ Additional processing of the photo-oxidized tissue for EM was per- 0.05 and are reported as P Ͻ 0.05, 0.03, 0.02, 0.001, or 0.0001. formed using procedures similar to those we have used before (15,29,32). Samples were fixed with 1% glutaraldehyde for 1 h, Results washed four times with PBS, postfixed with 0.14% osmium tetroxide/ We first confirmed and extended previous characterization 0.04% tannic acid/0.50% K Fe(CN) in PBS, dehydrated with graded 3 6 of the relatively nontoxic STG treatment model in rats. No ethanol, infiltrated with graded solutions of propylene oxide/Spurr’s resin to 100% Spurr’s resin (incubated for 2 ϫ 1 h at 100%), and then difference in serum creatinine levels was observed between polymerized overnight at 65°C. Ultrathin sections were cut using a control and gentamicin-treated rats after3doftreatment (con- MT6000-XL ultramicrotome (RMC Inc., Tucson, AZ) and observed trol, 0.42 Ϯ 0.04 mg/dl; treated, 0.38 Ϯ 0.04 mg/dl; P ϭ NS, with a Philips CM-12 electron microscope (Wilmington, DE), using n ϭ 5 for both groups). Protein concentrations for treated rats 80- to 100-kV accelerating voltage. were also not different from control values, in homogenates or

Table 2. Enzyme activities for animals treated with gentamicin for 3 da

Specific Activity (␮ Enrichment (BBM specific activity/ mol/h per mg protein) homogenate) Enzyme Control 3-d STG Control 3-d STG

LAP homogenates 3.27 Ϯ 0.49 (7) 2.94 Ϯ 0.37 (10) BBM 29.88 Ϯ 4.93 (7) 29.76 Ϯ 5.45 (10) 9.16 Ϯ 1.02 (7) 10.25 Ϯ 2.02 (10) Naϩ/Kϩ-ATPase homogenates 14.59 Ϯ 1.82 (7) 10.95 Ϯ 1.66 (10)b BBM 29.67 Ϯ 3.20 (7) 31.52 Ϯ 4.09 (10) 2.07 Ϯ 0.41 (7) 2.93 Ϯ 0.53 (10)b SDH homogenates 19.75 (avg.) (2) 20.72 Ϯ 3.88 (5) BBM 3.50 (avg.) (2) 3.49 Ϯ 0.81 (5) 0.18 (avg.) (2) 0.17 Ϯ 0.04 (5) NADH dehydrogenase homogenates 25.23 (avg.) (2) 28.70 Ϯ 4.37 (5) BBM 4.10 (avg.) (2) 4.32 Ϯ 1.04 (5) 0.16 (avg.)(2) 0.15 Ϯ 0.04 (5)

a Rats were treated with gentamicin for 3 d. The day after the last injection, cortical homogenates and BBM preparations were prepared and protein concentrations and enzyme activities were determined as described in Materials and Methods. The n values for each enzyme are indicated by the values in parentheses. LAP, leucine aminopeptidase; SDH, succinate dehydrogenase; avg., average. b Statistical significance of P Ͻ0.01. J Am Soc Nephrol 12: 114–123, 2001 Gentamicin Inhibits Renal Metabolism 117

BBM preparations, on days 1 and 3 (Table 1). A significant difference in protein concentrations was observed on day 2 for treated rats. Similarly to previous studies (16), we observed little effect of STG treatment on several cellular homogenate enzyme activities and their enrichment in BBM preparations. The BBM marker leucine aminopeptidase exhibited no significant change in either specific activity or BBM enrichment (Table 2). Like- wise, markers of two intracellular organelles, i.e., succinate dehydrogenase (mitochondria) and KCN-resistant NADH dehydrogenase (ER), did not exhibit any significant change in enrichment (Table 2). However, statistically significant alterations were observed for the basolateral membrane marker Naϩ/Kϩ-ATPase. The specific activity in homogenates from STG-treated animals was reduced, whereas the BBM enrichment was increased. No alterations in enzyme activities or BBM enrichment were observed at time points earlier than 3 d (data not shown). The next experiments directly assessed the effects of gentamicin on in vivo protein synthesis. Rats were treated with gentamicin for 1 to 3 d and then injected with [3H]leucine 1 h before euthanasia, for determination of newly synthesized protein in homogenates and BBM preparations. As presented in Figure 1A, a significant decrease in [3H]leucine incorporation into total cellular protein was observed by the second day of treatment. Protein synthesis decreased further on the third day of treatment. When total cellular and BBM protein syntheses were compared at 3 d (Figure 1B), both were decreased by approximately 50%. Effects on protein degradation were not Figure 1. Effects of gentamicin on protein synthesis. Rats were treated directly measured. with gentamicin for 1 to 3 d. The day after the indicated daily dose, It has been demonstrated that short-term (11,12,16,37) and rats received intraperitoneal injections of [3H]leucine. After 1 h, long-term (15) aminoglycoside exposure produces rapid and cortical homogenates and brush border membrane (BBM) prepara- persistent alterations in renal phospholipid contents. These tions were prepared, samples were precipitated with TCA, and radio- changes are thought to result from inhibition of phospholipid activity was determined as described in Materials and Methods. (A) degradation (13,37–39). It is also possible that synthesis and/or Effect of 1 to3dofgentamicin administration on total cellular protein incorporation of phospholipids into their appropriate mem- synthesis (i.e., homogenates). (B) Effect of3dofgentamicin (Gent.) branes may be affected. To explore these possibilities, exper- treatment on total protein synthesis (left) and BBM protein synthesis (right), compared with control animals. The significance of the dif- iments were performed to better characterize the effects of ference for gentamicin-treated homogenates and BBM, compared gentamicin on phospholipid synthesis and incorporation into with control homogenates and BBM, is indicated (n Ն 4 for all membranes. Results are presented in Table 3. After treatment groups). of rats with gentamicin for 3 d, the total phospholipid content of homogenates and BBM preparations was significantly increased (both 21%), compared with control homogenates and BBM. Although incorporation of 32P into total cellular phos- served, whereas a significant decrease (23%) in sphingomyelin pholipids (i.e., synthesis) was moderately decreased (11%) and (SPH) content was observed (Figure 2B). increased (7%) in gentamicin-treated homogenates and BBM, To evaluate whether synthesis of individual phospholipids respectively, neither change was statistically significant. was being altered, rats were treated with gentamicin exactly as Our next experiments characterized the effects of3dof described above, except they were injected with 32P 2 h before gentamicin treatment on individual phospholipid contents. No euthanasia. Significant decreases in the rate of incorporation of significant effect on the percentage composition of any phos- 32P into SPH (P Ͻ 0.05) and phosphatidylserine (PS) (P Ͻ pholipid analyzed in STG cellular homogenates was observed, 0.05) were observed, whereas a significant increase was noted except for phosphatidylinositol (PI), which exhibited an in- for phosphatidylethanolamine (PE) (P Ͻ 0.02) (Table 4). The crease of 34% (Figure 2A). However, when individual phos- same analysis performed on lipids extracted from BBM prep- pholipids of BBM preparations from STG-treated and control arations from treated animals revealed only a significant de- rats were compared, significant increases in phosphatidylcho- crease in 32P incorporation into SPH (P Ͻ 0.001) (Table 4). No line (PC) and PI levels (37 and 64%, respectively) were ob- consistent significant change in total or individual phospho- 118 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 114–123, 2001

Table 3. Effects of gentamicin on total phospholipid contents and 32P incorporation into total phospholipidsa

Total Phospholipids (nmol/mg protein) 32P Incorporation (cpm/mg protein per 2 h)

Control Gentamicin Control Gentamicin

Homogenates 180 Ϯ 7 (5) 218 Ϯ 21 (5)b 22,130 Ϯ 1,711 (5) 19,633 Ϯ 1,853 (3) BBM 648 Ϯ 42 (5) 784 Ϯ 53 (5)b 44,588 Ϯ 4,608 (5) 47,813 Ϯ 2,724 (3)

a Rats were treated with gentamicin for 3 d. On the fourth day, cortical homogenates and BBM preparations were prepared and phospholipids were isolated and quantified or rats were given injections of 32P, 2 h later the same procedures were performed and radioactivity was then determined. Gentamicin-treated homogenates and BBM were compared with control samples. The n values are indicated by the values in parentheses. b Statistical significance of P Ͻ0.02 for homogenates and P Ͻ0.01 for BBM (total phospholipid content only).

lipid composition or in total or individual phospholipid synthesis was observed at times earlier than 3 d (data not shown). Both membrane protein and phospholipid biosynthetic path- ways use cytosolic enzymes. However, they are also both dependent on and intimately associated with the membranes of intracellular organelles, specifically the ER and Golgi complex. The last series of experiments was performed to morpho- logically characterize where gentamicin was trafficked and concentrated during STG treatment. These experiments were performed exactly as described above, except that tracer amounts of TRG were added to the native gentamicin used for injection (see Materials and Methods). Confocal microscopic results from these experiments are presented in Figure 3; large bright fluorescent structures are clearly evident. Similar to native gentamicin, the TRG was taken up, retained, and concentrated in numerous, relatively large structures (probably lysosomes) located predominantly in the basal region of proximal tubule cells (PTC). Only minimal plasma membrane staining was observed 24 h after the final injection. No fluorescence staining was observed in proximal tubules of control animals (data not shown). Little, if any, intracellular staining was observed in glomeruli or distal tubules. When the fluorescent structures were examined in greater detail using photo-oxidation of TRG and EM, it became evident that TRG was associated with a variety of structures. The classic effects of gentamicin on lysosomes can be seen in Figure 4. Electron micrographs of photo-oxidized tissue sections from animals given injections of vehicle (Figure 4A), native gentamicin (Figure 4B), and TRG-spiked gentamicin (Figure 4C) are shown. It was evident that the photo-oxidation Figure 2. Effects of gentamicin on homogenate and BBM individual reaction itself did not induce significant reaction product in phospholipid contents. Rats were treated with gentamicin for 3 d. The either control tissue (Figure 4A) or native gentamicin-treated day after the last injection, cortical homogenates and BBM were tissue (Figure 4B), which contained myeloid material. How- prepared, total phospholipids were extracted and isolated, and indi- ever, in the tissue treated with gentamicin and tracer amounts vidual phospholipids were separated and then quantified as described of TRG, much of the myeloid material was darkly stained in Materials and Methods. (A) Individual phospholipids as percent- (Figure 4C). Not all lysosomes or myeloid material in the same ages of total phospholipids in cortical homogenates. (B) Individual section or cell was labeled. Much of the TRG-stained lysoso- phospholipids as percentages of total phospholipids in BBM prepa- mal myeloid material appeared as heavily stained, collapsed/ rations. The statistical significance of the difference between individ- aggregated membranes that lacked the classic concentric cir- ual phospholipids for homogenates (A) and BBM (B) is indicated above the specific phospholipid (n Ն 3 for all groups). SPH, sphin- cular, lamellar appearance (compare Figure 4, B and C) gomyelin; PC, phosphatidylcholine; PI, phosphatidylinositol; PS, (25,40). Whether TRG was associated with lipid, protein, or phosphatidylserine; PE, phosphatidylethanolamine. both within the membrane was not determined. J Am Soc Nephrol 12: 114–123, 2001 Gentamicin Inhibits Renal Metabolism 119

Table 4. Effects of gentamicin on homogenate and BBM individual phospholipid synthesisa

32P Incorporation into Individual Phospholipids (cpm/mg protein)

Phospholipid Homogenate BBM

Control 3-d STG Control 3-d STG

SPH 224 Ϯ 17 (5) 138 Ϯ 38 (3)b 1,156 Ϯ 150 (5) 671 Ϯ 104 (4)b PC 17,573 Ϯ 2,118 (5) 16,231 Ϯ 2,416 (3) 28,515 Ϯ 1,434 (5) 27,954 Ϯ 9,005 (4) PI 1,493 Ϯ 505 (5) 1,845 Ϯ 389 (3) 1,879 Ϯ 709 (5) 2,853 Ϯ 1,052 (4) PS 110 Ϯ 26 (5) 68 Ϯ 18 (3)b 703 Ϯ 260 (5) 464 Ϯ 156 (4) PE 2,156 Ϯ 238 (5) 2,897 Ϯ 240 (3)b 9,809 Ϯ 1,476 (5) 9,507 Ϯ 2,989 (4)

a Rats were treated with gentamicin for 3 d. The day after the last dose, rats were given intraperitoneal injections of 32P. After 2 h, cortical homogenates and BBM preparations were prepared, phospholipids were isolated and separated, and the radioactivity of the separated individual phospholipids was determined. Incorporation of radioactivity into individual phospholipids of cortical homogenates and BBM preparations is shown. SPH, sphingomyelin; PC, phosphatidylcholine; PI, phosphatidylinositol; PS, phosphatidylserine; PE, phosphatidylethanolamine. All n values are indicated by the values in parentheses. b Statistical significance of P Ͻ0.05 for homogenate SPH and PS, P Ͻ0.02 for homogenate PE, and P Ͻ0.001 for BBM SPH.

However, some of the darkly stained structures did not appear to be lysosomal (Figure 5B). In additional experiments, which were designed to specifically enhance this staining with the injection of TRG into STG-treated rats 30 min before perfusion-fixation, we observed unequivocal staining of classic Golgi stacks (Figure 5D) and mitochondria, according to morphologic criteria (Figure 5E). Although the results were not rigorously quantified, we estimate that Յ10% of the Golgi complexes and mitochondria observed were stained in this manner. Unequivocal evidence of mitochondrial staining was observed in the TRG accumulation experiments (Figure 5, B and C) and the uptake experiments (Figure 5D). At high magnification (Figure 5E), it was clear that the outer membrane and intermembrane space were heavily stained. However, it was not clear whether the inner membrane was labeled. If labeled, it did not appear to extend deeply into the cristae of the mitochondria. Staining of the matrix was not observed in any of the heavily stained mitochondria observed.

Discussion It has long been known that gentamicin treatment can induce Figure 3. Confocal microscopic analysis of accumulated Texas Red- gentamicin (TRG). Rats were treated for 3 d with gentamicin con- toxicity in PTC (4,18,25,40); however, the mechanisms in- taining tracer amounts of TRG. Kidneys were flushed and perfusion- volved in the toxic injury remain unresolved. Although in vitro fixed, and sections were cut, mounted, and then analyzed by confocal and partially in vivo studies have demonstrated that protein microscopy. Magnification, ϫ120. synthesis is inhibited with gentamicin (19–22), it had not been determined whether this inhibition occurs in entirely in vivo experiments. Phospholipase inhibition, which is thought to More detailed examination revealed evidence of other intra- induce myeloid formation and phospholipidosis, has been di- cellular organelle staining (Figure 5). Darkly stained structures rectly demonstrated in vitro and implicated in vivo (13,37–39) could be observed in the basal region of cells from adjacent but, to our knowledge, alteration of in vivo phospholipid bio- proximal tubules of gentamicin/TRG-treated animals (Figure synthesis had not been demonstrated. The purpose of the 5B). This staining was very well correlated with the confocal studies presented here was to determine the effects of amino- microscopic data presented in Figure 3. Dark staining, as seen glycoside treatment on protein and phospholipid metabolism in in Figure 5B, was not observed in vehicle-treated control entirely in vivo experiments. samples photo-oxidized exactly like the gentamicin/TRG- Using a rat model with little other evidence of PTC toxicity, treated samples (Figure 5A). At higher magnification, it was we demonstrated that protein synthesis was significantly in- clear that much of this staining was lysosomal (Figure 5C). hibited after 2 d of gentamicin treatment. This inhibition in- 120 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 114–123, 2001

Figure 4. Photo-oxidation and electron microscopic (EM) analysis of TRG accumulation in lysosomal myeloid material. Animals were treated as described for Figure 3. However, after sectioning, samples were subjected to photo-oxidation procedures and were then embedded and processed for EM. (A) Photo-oxidized cortical section from a vehicle-treated animal. Magnification, ϫ23,380. (B) Photo-oxidized section from a gentamicin-treated animal. Magnification, ϫ23,380. (C) Photo-oxidized section from a gentamicin/TRG solution-treated animal. Magnification, ϫ25,200.

Figure 5. Photo-oxidation/EM analysis of intracellular organelles (lysosomes, Golgi complexes, and mitochondria). Animals and sections were treated as described for Figures 3 and 4 (A to C and E) or were treated with TRG 30 min before perfusion-fixation (D), as described in the text. (A) Control. Magnification, ϫ5800. (B) Gentamicin/TRG-treated. Magnification, ϫ5800. (C) Gentamicin/TRG-treated, lysosomal staining. Magnification, ϫ35,000. (D) Gentamicin/TRG-treated, 30-min TRG uptake, Golgi complex staining. Magnification, ϫ35,000. (E) Gentamicin/TRG-treated, mitochondrial staining. Magnification, ϫ45,000. J Am Soc Nephrol 12: 114–123, 2001 Gentamicin Inhibits Renal Metabolism 121 creased to 50% by the third day. Furthermore, this inhibition ments. Photo-oxidized TRG darkly labeled not only mem- appeared to affect BBM protein synthesis in a similar manner branes of myeloid bodies within lysosomes but also mem- and to the same degree as total cellular protein synthesis. branes of Golgi cisternae and the outer membrane and Effects on protein degradation were not directly determined. intermembrane space of mitochondria. These studies agree with and expand earlier, partly in vivo Although disruptions of protein and phospholipid synthesis studies performed somewhat differently than ours (19–22). could occur at many sites, a common site for both would be at With the same model, we also observed disruption of indi- the level of the Golgi complex, ER, or cell cytoplasm, partic- vidual phospholipid synthesis and accompanying inhibition of ularly in the early stages. We are aware of only one study that phospholipid degradation. In contrast to our studies of protein suggests that gentamicin may be released into the cytosol at synthesis, total cellular and BBM phospholipid syntheses were early time points after uptake (41). In the morphologic studies not significantly affected by gentamicin treatment. However, we have performed, we have never observed evidence of there were significant alterations in the incorporation of 32P gentamicin being released into the cytosol in living cells. into individual phospholipids. Incorporation of 32P into PE was Therefore, for alteration of protein and phospholipid synthesis, significantly increased in STG-treated homogenates, whereas we think that retrograde trafficking of gentamicin to the Golgi incorporation into SPH and PS was reduced. Only a reduction complex and perhaps the ER must occur. Our recent cell in the incorporation of 32P into SPH in BBM preparations was culture studies (23) and the in vivo morphologic data presented observed with STG treatment. Previous in vivo studies, which here, as well as implications from studies performed by other were performed differently than ours, demonstrated either no investigators (12,14,24,42), support this possibility. We do not change in 32P incorporation into any phospholipid in a rat now have a good explanation for the mitochondrial staining. model (16) or an increase in the synthesis of at least two However, if gentamicin reaches the ER, it might be able to phospholipids using a rabbit primary culture PTC model (14). traffic to the mitochondria from there. In our studies, we observed no change (PC and PI in STG- Retrograde trafficking of a number of bacterial and plant treated homogenates and PC, PE, PI, and PS in STG-treated protein toxins after endocytosis is now well established (43). BBM), an increase (PE in STG-treated homogenates), and These protein toxins kill eukaryotic cells after translocation of decreases (PS and SPH in STG-treated homogenates and SPH an active part of the molecule from various sites into the in STG-treated BBM) in phospholipid levels. cytoplasm, where they inactivate the 60S subunit of the ribo- Similar to findings reported by others (11–13,16,37–39), our some or elongation factor 2 (see reference 43 and references studies showed that phospholipid degradation was inhibited at cited therein). Gentamicin inhibits bacterial protein synthesis early time points after gentamicin treatment (in these studies, at by binding to the 30S subunit of the bacterial ribosome and 3 d). We demonstrated that total phospholipid contents in inhibiting elongation or by causing mistranslation (1,2). In homogenates and BBM preparations were increased 21%, general, it is thought that gentamicin is not released into the compared with control samples. We also showed that the cytoplasm of kidney PTC in the early stages of toxicity. As a individual phospholipid contents of BBM were affected to a result, the gentamicin-induced disruption of protein and phos- greater extent by STG treatment than were total cellular mem- pholipid synthesis observed in these studies seems to be pro- branes, which is, to our knowledge, a new finding. The only duced via a different, more indirect mechanism than that of the significant effect on the percentage composition of any phos- protein toxins. We suggest that gentamicin may interact with pholipid in STG-treated homogenates was an increase in PI components of the protein and/or phospholipid biosynthetic content (34%). However, in STG-treated BBM preparations, machinery from the lumen of the Golgi complex or ER to significant increases in PC and PI levels (37 and 64%, respec- induce the alterations characterized in these studies. The pos- tively) were observed, whereas a significant decrease (23%) in sibility that aminoglycosides might directly interact with the the SPH content was observed, compared with control samples. Golgi complex or ER was suggested previously (12,24,37). It was clear from these studies that any increase in phospho- It is also possible that normal membrane traffic and fusion lipid content observed as a result of gentamicin treatment was become increasingly perturbed during gentamicin treatment, predominantly the result of the inhibition of degradation, rather resulting in disruption of many cellular activities and ulti- than any effect on synthesis. Finally, the phospholipid degra- mately cell death. The evidence of apparent retrograde move- dation and synthesis effects were somewhat delayed, compared ment of gentamicin observed in this study is consistent with with the protein synthesis effects, and did not become statisti- this possibility. Additionally, the subtle but significant alter- cally significant until the third day of treatment. This is rele- ation in Naϩ/Kϩ-ATPase enrichment in BBM preparations vant because the previously mentioned animal studies that supports the possibility of altered trafficking resulting from examined lipid synthesis were performed at 2 d (16). gentamicin treatment. Our last, and perhaps most interesting, results were from the Other investigators, using similar STG treatment models, morphologic studies. Confocal microscopic studies revealed also proposed alterations in membrane traffic (12,14,16,24,42). accumulation of fluorescently labeled gentamicin in large basal Relevant to this concept is evidence that gentamicin treatment structures. It was clear that most of these structures were inhibits vesicle/lysosome fusion in vivo (42) and homotypic lysosomes. However, more sensitive EM techniques revealed endosome fusion in vitro (44), as well as the suggestion that transport of gentamicin not only to lysosomes but also to Golgi longer-term gentamicin treatment causes swelling of the ER complexes and mitochondria, again in entirely in vivo experi- (25,45). It has also been speculated that myeloid bodies may be 122 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 114–123, 2001 formed as a result of decreased degradation of smooth ER (46). 7. Henry SA, Bendush CB: Aztreonam: Worldwide overview of the Furthermore, in a long-term model of gentamicin treatment in treatment of patients with Gram negative infections. Am J Med which rats recover from gentamicin-induced toxicity, we ob- 78: 57–64, 1985 served an apparent reduction of the endosomal compartment 8. Mathew TH: Drug induced renal disease. Med J Aust 156: (15). This hypothesis suggests that disruption of normal traf- 724–728, 1992 ficking and fusion could lead to aberrant trafficking and fusion. 9. Bennett WM: Mechanisms of aminoglycoside nephrotoxicity. Because gentamicin is normally delivered to lysosomes and Clin Exp Pharmacol Physiol 16: 1–6, 1989 contains no known targeting sequence, altered trafficking and 10. Cojocel C: Aminoglycoside nephrotoxicity. In: Comprehensive Toxicology, edited by Sipes G, McQueen C, Gandolfi AJ, New fusion could provide a mechanism for inducing the putative York, Elsevier Science, 1996, pp 495–523 aberrant retrograde movement. 11. Feldman S, Wang MY, Kaloyanides GJ: Aminoglycosides in- Although we do not present direct evidence to confirm this duce a phospholipidosis in the renal cortex of the rat: An early hypothesis, disruption of membrane trafficking provides a manifestation of nephrotoxicity. J Pharmacol Exp Ther 220: unique but consistent explanation of the results observed in this 514–520, 1982 study. Significantly, it also provides an encompassing mecha- 12. Josepovitz C, Forruggelia T, Levine R, Lane B, Kaloyanides GJ: nism of gentamicin-induced toxicity that could explain the Effects of netilmicin on phospholipids: Composition of subcel- different effects described in detail by many investigators over lular fractions of rat renal cortex. J Pharmacol Exp Ther 235: several decades. 810–819, 1985 In summary, the results of our studies provide, for the first 13. Laurent G, Carlier MB, Rollman B, van Hoof F, Tulkens P: time, direct and completely in vivo evidence that gentamicin Mechanism of aminoglycoside-induced lysosomal phospholipi- treatment inhibits protein synthesis in an animal model. They dosis: In vitro and in vivo studies with gentamicin and amakicin. also provide new evidence that individual phospholipid syn- Biochem Pharmacol 31: 3861–3870, 1982 thesis and degradation are disrupted in vivo. Lastly, they pro- 14. Ramsammy LS, Josepovitz C, Lane B, Kaloyanides GJ: Effect of vide unique evidence that gentamicin treatment causes traffick- gentamicin on phospholipid metabolism in cultured rabbit prox- ing of gentamicin to the Golgi complex and mitochondria. imal tubular cells. Am J Physiol 256: C204–C213, 1989 Together, these observations provide a mechanistic basis for 15. Sundin DP, Meyer C, Dahl R, Geerdes A, Sandoval R, Molitoris gentamicin-induced nephrotoxicity. As a result of treatment, BA: Cellular mechanism of aminoglycoside tolerance in long- gentamicin is delivered to multiple sites, perhaps by altered term gentamicin treatment. Am J Physiol 272: C1309–C1318, 1997 trafficking and fusion. These sites then correspond with mul- 16. Knauss TC, Weinberg JM, Humes HD: Alterations in renal tiple sites of potential toxicity. The implication may be that the cortical phospholipid content induced by gentamicin: Time observed toxicity is the result of multiple minor toxic effects, course, specificity and subcellular localization. Am J Physiol which may act synergistically, as suggested previously (47). 244: F535–F546, 1983 17. Moriyama T, Nakahama H, Fukuhara Y, Horio M, Yanase M, Acknowledgments Orita Y, Kamada T, Kanashiro M, Miyake Y: Decrease in the This work was supported by a National Institute of Diabetes and fluidity of brush-border membrane vesicles induced by gentami- Digestive and Kidney Diseases grant (Grant 5-R01-DK55527-03) to cin. Biochem Pharmacol 38: 1169–1174, 1989 Dr. Sundin and a National Institute of Diabetes and Digestive and 18. Laurent G, Kishore BK, Tulkens PM: Aminoglycoside-induced Kidney Diseases grant (Grant DK41126) and Veterans Affairs Merit renal phospholipidosis and nephrotoxicity. Biochem Pharmacol Review Award to Dr. Molitoris. We acknowledge the technical as- 40: 2383–2392, 1990 sistance of Diane Lyons and the helpful discussions and suggestions 19. Buss WC, Piatt MK, Kauten R: Inhibition of mammalian micro- of many of the members of the Division of Nephrology of the Indiana somal protein synthesis by aminoglycoside antibiotics. J Antimi- University School of Medicine. crob Chemother 14: 231–241, 1984 20. Buss WC, Piatt MK: Gentamicin administered in vivo reduces protein synthesis in microsomes subsequently isolated from rat References kidney but not from rat brain. J Antimicrob Chemother 15: 1. Davies J, Gorini L, Davis BD: Misreading of RNA code-words 715–721, 1985 induced by aminoglycoside antibiotics. Mol Pharmacol 1: 93– 21. Wilhelm JM, Jessop JJ, Pettitt SE: Aminoglycoside antibiotics 106, 1965 and eukaryotic synthesis: Stimulation of errors in the translation 2. Davis BD, Weisblum B, Davies J: Antibiotic inhibitors of the bacterial ribosome. Bacteriol Rev 32: 493–528, 1968 of natural messengers in extracts of cultured human cells. Bio- 3. Ali BH: Gentamicin nephrotoxicity in humans and animals: chemistry 17: 1149–1153, 1978 Some recent research. Gen Pharmacol 26: 1477–1487, 1995 22. Bennett WM, Mela-Riker LM, Houghton DC, Gilbert DN, Buss 4. Humes DH, Weinberg JM, Knauss TC: Clinical and pathophys- WC: Microsomal protein synthesis inhibition: An early manifes- iologic aspects of aminoglycoside nephrotoxicity. Am J Kidney tation of gentamicin nephrotoxicity. Am J Physiol 255: F265– Dis 2: 5–29, 1982 F269, 1988 5. Tulkens PM: Nephrotoxicity of aminoglycoside antibiotics. 23. Sandoval R, Leiser J, Molitoris BA: Aminoglycoside antibiotics Toxicol Lett 46: 107–123, 1989 traffic to the Golgi complex. J Am Soc Nephrol 9: 167–174, 1998 6. Hou SH, Bushinsky DA, Wish JB, Cohen JJ, Harrington JT: 24. Olbricht CJ, Fink M, Gutjahr E: Alterations in lysosomal en- Hospital-acquired renal sufficiency: A prospective study. Am zymes of the proximal tubule in gentamicin nephrotoxicity. Kid- J Med 74: 243–248, 1983 ney Int 39: 639–646, 1991 J Am Soc Nephrol 12: 114–123, 2001 Gentamicin Inhibits Renal Metabolism 123

25. Houghton DC, Hartnett M, Campbell-Boswell M, Porter G, 37. Ramsammy LS, Josepovitz C, Kaloyanides GJ: Gentamicin in- Bennett W: A light and electron microscopic analysis of genta- hibits agonist stimulation of the phosphatidylinositol cascade in micin nephrotoxicity in rats. Am J Pathol 82: 589–612, 1976 primary cultures of rabbit proximal tubular cells and in rat renal 26. Biber J, Stieger B, Hasse W, Murer H: A high yield preparation cortex. J Pharmacol Exp Ther 247: 989–996, 1988 for rat kidney brush border membranes. Biochim Biophys Acta 38. Carlier MB, Laurent G, Claes PJ, Vanderhaeghe HJ, Tulkens 647: 169–176, 1981 PM: Inhibition of lysosomal phospholipases by aminoglycoside 27. Molitoris BA, Simon FR: Renal cortical brush-border and baso- antibiotics: In vitro comparative studies. Antimicrob Agents Che- lateral membranes: Cholesterol and phospholipids composition mother 23: 440–449, 1983 and relative turnover. J Membr Biol 83: 207–215, 1985 39. Hostetler KY, Hall LB: Inhibition of kidney phospholipases A 28. Molitoris BA, Kinne R: Ischemia induced surface membrane and C by aminoglycoside antibiotics: Possible mechanism of dysfunction: Mechanism of altered Naϩ-dependent glucose aminoglycoside toxicity. Proc Natl Acad Sci USA 79: 1663– transport. J Clin Invest 80: 647–654, 1987 1667, 1982 29. Molitoris BA, Meyer C, Dahl RH, Geerdes AE: Mechanism of 40. Kosek JC, Mazze RI, Cousins MJ: Nephrotoxicity of gentamicin. ischemia-enhanced aminoglycoside binding and uptake by prox- Lab Invest 30: 48–57, 1974 imal tubule cells. Am J Physiol 264: F907–F916, 1993 41. Sandvig K, van Deurs B: Endocytosis, intracellular transport and cytotoxic action of Shiga toxin and ricin. Physiol Rev 76: 949– 30. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ: Protein mea- 966, 1996 surement with the Folin phenol reagent. J Biol Chem 193: 265– 42. Weeden RP, Batuman V, Cheeks C, Marquet E, Sobel H: Trans- 275, 1951 port of gentamicin in rat proximal tubule. Lab Invest 48: 212– 31. Bradford MM: A rapid and sensitive method for the quantitation 223, 1983 of microgram quantities of protein utilizing the principle of 43. Giurgea-Marion L, Toubeau G, Laurent G, Heuson-Stiennon JA, protein-dye binding. Anal Biochem 72: 248–254, 1976 Tulkens PM: Impairment of lysosome-pinocytotic vesicle fusion 32. Sundin DP, Cohen M, Dahl R, Falk S, Molitoris BA: Character- in rat kidney proximal tubules after treatment with gentamicin at ␤ ization of the 2-microglobulin endocytic pathway in rat proxi- low doses. Toxicol Appl Pharmacol 86: 271–285, 1986 mal tubule cells. Am J Physiol 267: F380–F389, 1994 44. Hammond TG, Majewski RR, Kaysen JH, Goda FO, Navar GL, 33. Bligh EG, Dyer WJ: A rapid method of total lipid extraction and Pontillon F, Verroust PJ: Gentamicin inhibits rat renal cortical purification. J Biochem Physiol 37: 911–917, 1969 homotypic endosomal fusion: Role of megalin. Am J Physiol 34. Ames BN, Dubin DT: The role of polyamines in the neutraliza- 272: F117–F123, 1997 tion of bacteriophage deoxyribonucleic acid. J Biol Chem 235: 45. Cuppage FE, Setter K, Sullivan LP, Reitzes EJ, Melnykovych 769–775, 1968 AO: Physiological, biochemical, and morphological effects of 35. Esko JD, Raetz CRH: Mutants of Chinese hamster ovary cells prolonged administration to rats. Virchows Arch B Cell Pathol with altered membrane phospholipid composition: Replacement 24: 121–138, 1977 of phosphatidylinositol by phosphatidylglycerol in a myo-inositol 46. Hruban Z, Slesers A, Hopkins E: Drug induced and naturally auxotroph. J Biol Chem 255: 4474–4480, 1980 occurring myeloid bodies. Lab Invest 27: 62–70, 1972 36. Sandell JH, Masland RH: Photoconversion of some fluorescent 47. Kaloyanides G: Aminoglycoside-induced functional and bio- markers to a diaminobenzidine product. J Histochem Cytochem chemical defects in the renal cortex. Fundam Appl Toxicol 4: 36: 555–559, 1988 930–943, 1984