Inhibiting glucosylceramide synthase exacerbates cisplatin-induced acute kidney injury

Acute kidney injury (AKI), resulting from chemotherapeutic agents such as cisplatin, remains an obstacle in the treatment of cancer. Cisplatin-induced AKI involves apoptotic and necrotic cell death, pathways regulated by sphingolipids such as ceramide and glucosylceramide. Results from this study indicate that C57BL/6J mice treated with cisplatin had increased ceramide and hexosylceramide levels in the renal cortex 72 h following cisplatin treatment. Pretreatment of mice with inhibitors of acid sphingomyelinase and de novo ceramide synthesis (amitriptyline and myriocin, respectively) prevented accumulation of ceramides and hexosylceramide in the renal cortex and protected from cisplatin-induced AKI. To determine the role of ceramide metabolism to hexosylceramides in kidney injury, we treated mice with a potent and highly specific inhibitor of glucosylceramide synthase, the enzyme responsible for catalyzing the glycosylation of ceramides to form glucosylceramides. Inhibition of glucosylceramide synthase attenuated the accumulation of the hexosylceramides and exacerbated ceramide accumulation in the renal cortex following treatment of mice with cisplatin. Increasing ceramides and decreasing glucosylceramides in the renal cortex sensitized mice to cisplatin-induced AKI according to markers of kidney function, kidney injury, inflammation, cell stress, and apoptosis. Under conditions of high ceramide generation, data suggest that metabolism of ceramides to glucosylceramides buffers kidney ceramides and helps attenuate kidney injury.—Dupre, T. V., M. A. Doll, P. P. Shah, C. N. Sharp, D. Siow, J. Megyesi, J. Shayman, A. Bielawska, J. Bielawski, L. J. Beverly, M. Hernandez-Corbacho, C. J. Clarke, A. J. Snider, R. G. Schnellmann, L. M. Obeid, Y. A. Hannun, and L. J. Siskind. Inhibiting glucosylceramide synthase exacerbates cisplatin-induced acute kidney injury. J. Lipid Res. 2017. 58: 1439–1452.


1
To whom correspondence should be addressed. e-mail: leah.siskind@louisville.edu Cisplatin is a commonly used chemotherapeutic in the treatment of many cancers (5-7); however, cisplatin is known to have a dose limiting side effect of AKI (5,8). Unfortunately, treatment options for cisplatin-induced AKI are limited to supportive care (1,9). Treatment with cisplatin is not advised for patients with conditions that predispose them to an increased risk for kidney injury, including advanced age, hypoalbuminemia, diabetes, previous history of AKI, or known chronic kidney disease (8). Even without one of these conditions, 30% of patients administered cisplatin will develop kidney injury (5,8,(10)(11)(12), requiring that the dose be lowered or the patient be switched to a less effective therapeutic regimen.
Due to the complexity of cisplatin's nephrotoxic mechanism of action, the development of renoprotective agents remains a challenge. In the kidney, cisplatin is taken up by organic cation transport-2 and copper transporter-1, two transporters that are highly expressed in renal proximal tubule cells (8,13,14). Once proximal tubule cells take up cisplatin, it is then metabolized to a highly reactive cisplatin-thiol conjugate by enzymes that are abundant in renal proximal tubule cells (5,8,11,15,16). It is known that inhibiting the uptake or metabolism of cisplatin to its nephrotoxic metabolite in proximal tubule cells will prevent the ability of cisplatin to exhibit its cytotoxic effects on renal proximal tubule cells (10,13,14,(17)(18)(19)(20)(21)(22). During cisplatin-induced AKI, a number of parallel and converging pathways then mediate the pathogenesis of cisplatin-induced AKI, such as inflammation, endoplasmic reticulum (ER) stress, generation of reactive oxygen species, and the activation of cell death pathways (5,8). Many of the pathways by which cisplatin-induces AKI are also thought to be important to its ability to reduce tumor burden. Thus, it is challenging to find a therapeutic agent that both prevents cisplatin-induced AKI and does not interfere with cisplatin's cytotoxic effects on cancer cells.
Sphingolipids, a class of bioactive lipids that share a common sphingoid base backbone, are known to play a role in a number of the pathways involved in cancer cell death in response to cisplatin and in AKI (23). While a role for sphingolipids in cancer development and response to chemotherapeutics such as cisplatin is wellstudied (24), the role of sphingolipids in cisplatin-induced AKI is largely unknown. At the center of sphingolipid metabolism are ceramides, which play a role in a variety of cell death pathways and their levels increase in cells treated with a variety of death stimuli, including cisplatin (25)(26)(27). Ceramides can be generated via de novo synthesis, sphingomyelin hydrolysis, or the recycling of complex sphingolipids ( Fig. 1) (23). Once generated, ceramides can be phosphorylated to form ceramide 1-phosphate, be broken down to form sphingosine that is used to form sphingosine 1-phosphate (S1P), or to reform ceramide or be glycosylated to form glycosphingolipids (23). It was previously shown that S1P plays a protective role in cisplatin-induced AKI via signaling through S1P receptor-1 (28). However, S1P has been shown to increase resistance of cancer cells to cytotoxic agents, promote cancer development, and promote cancer metastasis. Thus, targeting ceramide metabolism to S1P in the context of chemotherapyinduced AKI is not ideal. With regard to ceramide metabolism to glucosylceramides, it is well-established that cancer cells escape programmed cell death in response to a variety of chemotherapy agents, via upregulating glucosylceramide synthase, and it has been proposed that inhibitors of glucosylceramide synthase could prevent chemotherapy resistance in the clinic (24,(29)(30)(31)(32)(33)(34). However, in order to pursue targeting ceramide generation and/or its metabolism to glucosylceramide for the treatment of cancer, an understanding of the role for these pathways in cisplatininduced AKI is needed.
Glucosylceramide, has been implicated to play a role in a variety of kidney diseases, such as diabetic nephropathy, lupus nephritis, and polycystic kidney disease, as well as decreased kidney function associated with aging (35)(36)(37)(38). Glucosylceramides have been implicated in regulating pathways involved in the above kidney diseases, including cell death, ER stress, and inflammation (23,39,40). However, a role for glucosylceramides and glucosylceramide synthase in cisplatin-induced AKI has not been investigated. Here, we measured levels of ceramides and hexosylceramides in the kidney following cisplatin administration and utilized inhibitors of either ceramide generation or glucosylceramide synthesis to determine the role of ceramides and glucosylceramides in cisplatin-induced AKI. Our data show that both ceramides and hexosylceramides are elevated in the kidney cortex following cisplatin administration. Pretreatment of mice with inhibitors of ceramide generation attenuated both ceramide and hexosylceramide accumulation and protected mice from cisplatin-induced AKI. To determine whether ceramides or hexosylceramides were contributing to cisplatininduced AKI, we pretreated mice with an inhibitor of glucosylceramide synthase, the enzyme responsible for catalyzing the glycosylation of ceramide to form glucosylceramides ( Fig. 1). Inhibition of glucosylceramide synthase attenuated cisplatin-induced hexosylceramide accumulation and exacerbated ceramide accumulation in the renal cortex; increasing ceramides and decreasing hexosylceramides also exacerbated cisplatin-induced AKI according to markers of kidney injury, inflammation, and apoptosis. Taken together, these data suggest that ceramides play a detrimental role in AKI in response to cisplatin, whereas their metabolism to glucosylceramides is renoprotective.  C57BL/6J mice (male, 8 weeks old for myriocin and amitriptyline studies and 12 weeks old for C10 and repeated dosing  studies) were purchased from Jackson Laboratory (Bar Harbor, ME) and acclimated for 1 week prior to initiation of experiments. Mice were maintained on a 12 h light/dark cycle and provided food and water ad libitum. All animal procedures were approved by the Institutional Animal Care and Use Committee and followed the guidelines of the American Veterinary Medical Association. Where indicated, mice were pre- and cotreated with myriocin (0.3 mg/kg ip/day) (Sigma Aldrich, St. Louis, MO) and amitriptyline (1 mM given in the drinking water) for 48 h prior to and also at the time of an intraperitoneal injection of cisplatin (25 mg/kg, ip) (P4394; Sigma Aldrich), and mice were euthanized 72 h following cisplatin administration. D-threo-1-ethylendioxyphenyl-2decanoylamino-3-pyrrolidinopropanol (C10) was synthesized as previously reported (41). Cisplatin, 25 mg/kg (in normal saline at 1 mg/ml) was administered by intraperitoneal injection at time 0 h. C10 was given by oral gavage every 12 h starting at time 0 h for 48 h. Animals were euthanized at 72 h after the cisplatin and initial C10 administration. At euthanasia, blood was collected and plasma prepared and frozen at 80°C, kidneys were flash-frozen in liquid nitrogen and stored at 80°C until used, or fixed in 10% neutral buffered formalin.

Mouse model of ischemia/reperfusion-induced AKI
Male C57BL/6 mice (25 g) were subjected to bilateral ischemia/reperfusion (I/R) injury to produce AKI, as described (42). Mice were anesthetized with pentobarbital (65 mg/kg) and surgery was performed on a sterile disposable towel over a warming pad to maintain body temperature. Following a midline incision, the renal artery and vein were isolated from surrounding tissue and were occluded with a nontraumatic vascular clamp for 26 min. Mice in the sham group were treated the same except that no clamps were applied. Following the ischemic period, the vascular clamps were removed and the kidneys observed to document reflow of blood. At 24 and 48 h after I/R injury, mice were euthanized and blood and kidneys collected (42). Blood urea nitrogen (BUN) and serum creatinine (SCr) were measured to monitor kidney function (42) and kidney cortex homogenates from 24 h sham and I/R mice (42) were utilized for measurement of sphingolipids and enzyme activities as described below. All mice were housed in temperature-controlled conditions under a light/dark photocycle with food and water supplied ad libitum. All of the animal and treatment protocols were in compliance with the Guide for the Care and Use of Laboratory Animals, as adopted and promulgated by the US National Institutes of Health and were approved by the Institutional Animal Care and Use Committee.

Rat model of rhabdomyolysis
Male Sprague-Dawley rats (Harlan, Indianapolis, IN), 8 weeks of age (180-200 g), were housed in temperature-controlled conditions under a light/dark photocycle with food and water supplied ad libitum. Rats were dehydrated for 16 h before glycerol injection. Rats were divided randomly into control and treatment (glycerol) groups. The control group (untreated, n = 5) was not injected with any treatment; the treatment group of rats (n = 20) was given an intramuscular injection of 50% glycerol (10 ml/kg) in their hind limbs. Blood, urine, and kidney samples were collected at 24, 48, 72, and 120 h and markers of kidney function performed (43). All of the animal and treatment protocols were in compliance with the Guide for the Care and Use of Laboratory Animals as adopted and promulgated by the US National Institutes of Health and were approved by the Institutional Animal Care and Use Committee. At 24 h post-glycerol injection, at the peak of injury (43), the kidneys were used for preparation of homogenate for measurement of sphingolipids and enzyme activities as described below. BUN (40146) and SCr (C7548-120) levels were determined on serum samples using kits from AMS Diagnostics (Weston, FL) and Point Scientific Inc. (Canton, MI), respectively, following the manufactures' instructions.

Histology and immunohistochemistry
Kidney histology and immunohistochemistry were done as previously described (44,45). Briefly, kidney sections (5 m) were stained with hematoxylin and eosin and periodic acid Schiff (PAS), and the degree of morphologic changes was determined by light microscopy in a blinded fashion. The following measures were chosen as an indication of morphologic damage to the kidney after drug treatment: proximal tubule degradation, loss of brush border, tubular casts, proximal tubule dilation, and proximal tubule necrosis. These measures were evaluated on a scale from 0 to 4, which ranged from not present (0), mild (1), moderate (2), severe (3), to very severe (4). Kidney sections (5 m) were stained for PCNA with lectin from Tetragonolobus purpureus (LTA) as a proximal tubule marker. Briefly, kidney sections (5 m) were deparaffinized and rehydrated, antigens were unmasked, followed by blocking of endogenous peroxidases. Sections were further blocked with 5% goat serum with 0.1% Triton X-100 in PBS, followed by incubation with HRP-conjugated LTA (1:500) for 2 h at room temperature. HRP-LTA was detected with Nova Red peroxidase (SK-4800; Vector Labs, Burlingame, CA). Next, sections were blocked for avidin and biotin, followed by anti-PCNA (1:16,000) incubation overnight at 4°C. Sections were then incubated with biotinylated goat anti-rabbit IgG antibody (1:25,000, BA-1000; Vector Laboratories) for 30 min at room temperature. Vector ABC reagent (PK-7100; Vector Laboratories) was added to each section and incubated for 30 min at room temperature. PCNA was then visualized with Nova Red DAB substrate (SK-4800; Vector Laboratories). Slides were counterstained with modified Mayer's hematoxylin (72804; Thermo Fisher Scientific), and then dehydrated followed by mounting with Permount (SP15; Thermo Fisher Scientific). Images were visualized using a Nikon Eclipse E600 microscope (Nikon Corporation, Tokyo, Japan). For quantification of PCNA, positive cells that stained brown in color were enumerated and then summed per square micron (200× magnification field). TUNEL assays were performed on paraffin embedded tissue sections using Apoptag Red in situ apoptosis detection kit per the manufacturer's instructions (S7165; Millipore, Temecula, CA). Sections were counterstained with DAPI and mounted with VECTASHIELD antifade mounting medium. Slides were visualized via immunofluorescent microscopy with a Nikon Eclipse Ti-E microscope using Nikon NIS Elements software (Nikon Corporation). For quantification of TUNEL positivity, TUNEL foci were enumerated and then summed per square micron (200× magnification field).

Enzyme activity assays
Ceramide synthase. Ceramide synthase (CerS) activity was measured in 50 g of renal cortex homogenate as described previously (38), as this was determined to be within the linear range for the enzyme activity assay. Briefly, a reaction mixture (100 l of final volume) containing 15 M C17-sphingosine (Avanti Polar Lipids, Alabaster, AL) and 50 M C16 fatty acyl-CoA in 25 mM potassium phosphate buffer (pH 7.4) was prewarmed at 37°C for 5 min. The enzyme reaction was initiated via addition of the enzyme source (50 g renal cortex homogenate) and, after 15 min at 37°C, was terminated via the addition of 2 ml of extraction solvent containing ethyl acetate/2-propanol/water (60:30:10, by volume) supplemented with internal standard for ESI-LC-MS analysis. Lipids were extracted twice, dried under a stream of nitrogen, and resuspended into 150 l of 1 mM ammonium formate in 0.2% formic acid in methanol and analyzed by ESI-HPLC-MS/MC. Acid sphingomyelinase. The acid sphingomyelinase (aSMase) activity assay was performed as previously described (46). Briefly, renal cortex homogenate (10 g, determined to be within the linear range for the enzyme activity) was added to 100 l of reaction mixture containing 100 m porcine brain sphingomyelin (Avanti Polar Lipids), 1 × 10 5 cpm of choline-[methyl- 14 C]sphingomyelin (specific radioactivity = 1.5 × 10 5 cpm/l), kindly supplied by Dr. Alicja Bielawska (Medical University of South Carolina Lipidomics Core), presented in micelles containing 0.2% Triton X-100 in sodium acetate buffer (250 mm, pH 5.0) supplemented with 1.0 mm EDTA. The reaction was run for 30 min at 37°C and was terminated by adding 1.5 ml of chloroform/methanol (2:1, v/v) followed by addition of 0.4 ml of Milli-Q water (modified Folch extraction). Samples were vortexed briefly and subjected to centrifugation at 2,000 g for 5 min at room temperature to separate phases. Aliquots (800 l) of the upper (aqueous) phase were used for liquid scintillation counting.

Sphingolipid analysis
Quantification of ceramide and hexosyl-ceramide species from 1 mg of protein tissue obtained from the kidney cortex was utilized for quantification of sphingolipid species, which was performed by the Lipidomics Shared Resource Facility at the Medical University of South Carolina via HPLC-MS/MS as previously described (47,48). This analytical approach measures hexosylceramide species (totaling both glucosylceramide species and galactosylceramide species). Data were normalized to total protein as well as total lipid phosphate.

Statistical analysis
Data are expressed as mean ± SEM for all experiments. Multiple comparisons of normally distributed data were analyzed by two-way ANOVA, as appropriate, and group means were compared using Bonferroni posttests. The criterion for statistical differences was P < 0.05 for all comparisons.

Inhibition of ceramide synthesis attenuates cisplatininduced increases in ceramides and hexosylceramides
To assess the role of ceramides in cisplatin-induced AKI, we first quantified ceramide levels via HPLC-MS/MS in the renal cortex of cisplatin-treated mice. C57BL/6J mice were treated with cisplatin (25 mg/kg) and then euthanized at the peak of injury (72 h) (49). Data show that cisplatin treatment significantly elevated C 16 -, C 20 -, C 24:1 -, and C 26 -ceramide species, as compared with vehicle-treated mice ( Fig. 2A).
Cisplatin treatment of cells in culture is also known to increase the activity of enzymes involved in ceramide synthesis, such as CerS and aSMase (27,50). aSMase is responsible for the hydrolysis of sphingomyelin to form ceramide (23). CerS is a family of enzymes that participates in de novo ceramide synthesis (23). Importantly, CerSs are responsible for synthesis of varying chain lengths of ceramide, and it is known that treatment of baby mouse kidney cells with cisplatin increases the activity of CerS to generate long-chain ceramides (27). Thus, we chose to measure the enzymatic activity of longchain CerS and aSMase 72 h after cisplatin (25 mg/kg) treatment. Our data indicate that cisplatin administration significantly increased long-chain CerS activity, as compared with vehicle-treated mice (Fig. 2B). Data also show that cisplatin administration significantly increased the enzymatic activity of aSMase, as compared with vehicle-treated mice (Fig. 2C). These data suggest that cisplatin-induced increases in ceramide levels are a result of increased ceramide generation via elevated long-chain CerS and aSMase activities.
Thus, to determine the contribution of long-chain CerS and aSMase to cisplatin-induced elevations in kidney cortex ceramides, we inhibited both pathways of ceramide generation. CerSs can be inhibited with fumonisin B1. Unfortunately, fumonisin B1 is not a very potent or specific inhibitor of CerSs and, importantly, is not tolerated well in animals. CerSs act, in part, in the de novo synthetic pathway of ceramide generation. Myriocin is a potent and specific inhibitor of serine palmitoyl transferase, the enzyme that catalyzes the first and rate limiting step of de novo ceramide synthesis (23). Myriocin is well-tolerated in rodents (51,52). Thus, we utilized myriocin to inhibit a major pathway of ceramide generation acting upstream of CerSs. To inhibit aSMase-mediated ceramide generation, we utilized amitriptyline, a Food and Drug Administration-approved drug that is well-tolerated in animals (23). Treatment of mice with myriocin (0.3 mg/kg ip/day) or amitriptyline (1 mM given in the drinking water) as single agents for 48 h prior to and at the time of cisplatin treatment (25 mg/ kg) did not protect from cisplatin-induced AKI (data not shown). Therefore, mice were cotreated with myriocin and amitriptyline. We coadministered myriocin (0.3 mg/kg ip/ day) and amitriptyline (1 mM given in the drinking water) for the 48 h prior to and at the time of cisplatin treatment (25 mg/kg). Mice were then euthanized 72 h following cisplatin administration at the peak of injury. Data indicate that cotreatment of mice with the inhibitors of ceramide generation (myriocin and amitriptyline, designated as inhibitors in Fig. 2) attenuated cisplatin-induced accumulations in C 16 -, C 20 -, C 24:1 -, and C 26 -ceramides ( Fig. 2A).
Importantly, ceramides can be further metabolized to other sphingolipid classes (Fig. 1); thus, we also assessed the levels of other downstream ceramide metabolites in the renal cortex, including sphingomyelins, dihydroceramides, hexosylceramides, and sphingoid bases (data not shown). Of the downstream ceramide metabolites assessed, only the hexosylceramides were elevated following cisplatin treatment (Fig. 2D). Data show that cisplatin treatment significantly elevated C 16 -, C 20 -, C 24:1 -, and C 26 -hexosylceramide species in the renal cortex, as compared with vehicle controls. Furthermore, cotreatment of mice with the inhibitors of ceramide generation (myriocin and amitriptyline, designated as inhibitors in Fig. 1) also attenuated cisplatin-induced accumulations in C 16 -, C 20 -, C 24:1 -, C 26hexosylceramides in the renal cortex (Fig. 2D).
We assessed the effect of attenuating cisplatin-induced ceramide and hexosylceramide accumulations with the inhibitors, myriocin and amitriptyline, on cisplatin-induced AKI using standard markers of kidney function, BUN and SCr. Mice treated with cisplatin alone had significantly increased levels of both SCr and BUN, as compared with vehicle-treated mice, thereby indicating a significant decline in kidney function (Fig. 2E, F). Importantly, pretreatment of mice with the inhibitors of ceramide synthesis (myriocin and amitriptyline, designated as inhibitors in Fig. 2) attenuated cisplatin-induced elevations in both SCr and BUN, as compared with cisplatin-treated mice (Fig. 2E, F). These data provide evidence that prevention of ceramide and hexosylceramide generation protects mice from cisplatin-induced AKI, suggesting a role for both ceramides and hexosylceramides in cisplatin-induced AKI.

Inhibition of glucosylceramide synthase with C10 exacerbates cisplatin-induced ceramide accumulation and attenuates hexosylceramide accumulation in the renal cortex
The above results demonstrate that when accumulation of both ceramides and hexosylceramides is prevented, mice are resistant to cisplatin-induced AKI. Ceramides and hexosylceramides have been implicated individually in a variety of kidney diseases. Thus, it is unclear whether it is the ceramides and/or the flux of ceramides into glycosphingolipids that are contributing to cisplatin-induced AKI. To elucidate the role of the ceramides and/or their metabolism to hexosylceramides independently of one another in cisplatin-induced AKI, we blocked metabolism of ceramides to glucosylceramides with a potent inhibitor of glucosylceramide synthase, named C10 (36). C10 is the 10-carbon analog of the Food and Drug Administrationapproved glucosylceramide synthase inhibitor, eliglustat. C10 has an IC 50 in the low nanomolar range as opposed to the mid micromolar range that the first generation glucosylceramide synthase inhibitor, D-threo-1-phenyl-2decanoylamino-3-morpholino-1-propanol (PDMP) has (53). In addition, C10 does not raise cellular ceramide levels when used alone in untreated quiescent or proliferating cells, unlike PDMP (54). PDMP is also known to inhibit lysosomal phospholipase A2 (54). C10 does not inhibit phospholipase A2 and, thus, is a much more specific inhibitor of glucosylceramide synthase than PDMP (54). Thus, we chose to utilize C10, not PDMP, as a specific inhibitor of glucosylceramide synthase to elucidate whether ceramides or hexosylceramides were mediating cisplatin-induced AKI.
Mice were treated with cisplatin (25 mg/kg, ip) and C10 (20 mg/kg, oral gavage). C10 treatment was continued every 12 h for 48 h. Mice were euthanized 72 h after cisplatin treatment. Data indicate that treatment with cisplatin alone significantly elevated ceramide and hexosylceramide species (C 16 -, C 18 -, C 24:1 -, C 24 -, and C 26 -ceramide species), as compared with vehicle-treated mice (Fig. 3A, B), while administration of C10 attenuated cisplatin-induced accumulations of hexosylceramides in the renal cortex (Fig. 3A). Because glucosylceramides and galactosylceramides have identical mass-to-charge ratios, they cannot be differentiated via HPLC-MS/MS and are reported as hexosylceramides. Importantly, galactosylceramides are synthesized via galactosylceramide synthase in the ER, and glucosylceramides are synthesized via glucosylceramide synthase in the Golgi apparatus. Thus, the fact that the cisplatin-induced elevation of hexosylceramides is entirely inhibited by the specific glucosylceramide synthase inhibitor, C10, indicates that the hexosylceramides generated following cisplatin treatment are comprised entirely of glucosylceramides and not galactosylceramides (Fig. 3C). Thus, hexosylceramides will be referred to as glucosylceramides for the remainder of this report.
Importantly, inhibition of glucosylceramide synthase by C10 in cisplatin-treated mice could lead to the accumulation of ceramides. Therefore, ceramide levels were also assessed in the renal cortex. As indicated in Fig. 3B, C10 and cisplatin cotreatment increased C 16 -, C 18 -, C 24:1 -, C 24 -, and C 26ceramide species significantly when compared with cisplatin Ceramide is glycosylated by either glucosylceramide synthase or galactosylceramide synthase to generate glucosylceramide and galactosylceramide, respectively. Glucosylceramide and galactosylceramide are collectively referred to as hexosylceramides. Statistical differences were measured by two-way ANOVA followed by Bonferroni posttest. Data are expressed as mean ± SEM (n = 5). *P < 0.05, **P < 0.01, and ***P < 0.001 demonstrate statistical difference from vehicle treatment. #P < 0.05 demonstrates statistical difference from cisplatin treatment. only-treated mice. Thus, these data indicate that C10 administration attenuates glucosylceramide accumulation and blocks flux of ceramides into glycosphingolipids, thereby exacerbating the ceramide accumulation in the renal cortex of cisplatin-treated mice.

C10 and cisplatin cotreatment exacerbates cisplatininduced AKI
If ceramides and not the flux of ceramides into glucosylceramides contribute to cisplatin-induced AKI, then kidney function should be worsened in mice coadministered cisplatin and C10, as the ceramide accumulation is exacerbated and the glucosylceramide accumulation prevented.
To determine the effect of C10 administration on cisplatininduced AKI, markers of kidney function and injury were assessed. Both cisplatin-treated mice and cisplatin and C10cotreated mice had significantly increased BUN levels (81.2 ± 18.8 and 129.6 ± 18.4, respectively), as compared with vehicle controls (31.9 ± 1.0) (Fig. 4A). These data suggest that C10 cotreatment with cisplatin may worsen kidney function, although the data are not significantly different. KIM-1 mRNA expression in the renal cortex was assessed as a marker of kidney injury. Cisplatin alone-treated mice and cisplatin and C10-cotreated mice had significantly increased renal Kim-1 mRNA expression (500-fold and 800- fold, respectively) as compared with vehicle controls (Fig. 4B). More importantly, cisplatin and C10-cotreated mice had significantly increased Kim-1 mRNA expression as compared with cisplatin-treated mice, indicating that C10 treatment exacerbates cisplatin-induced kidney injury (Fig. 4B).

C10 and cisplatin cotreatment induces renal histological damage
Cisplatin is metabolized to its nephrotoxic metabolite within the proximal tubule cells of the kidney, rendering proximal tubule cells susceptible to cisplatin-induced histological damage. Thus, renal histology on hematoxylin and eosin, as well as PAS-stained kidney sections was assessed to determine whether proximal tubule-specific histological changes were induced as a result of C10 and cisplatin coadministration (Fig. 5A). A renal pathologist scored kidney sections in a blinded manner to assess markers of proximal tubule damage, including the following: acute tubular necrosis, loss of brush border, cast formation, dilation, and degeneration. Figure 5B-F shows that all of the aforementioned histological markers of proximal tubule damage increased significantly in cisplatin-treated mice when compared with vehicle-treated mice. C10 and cisplatin-cotreated mice also had significantly increased renal pathology scores as compared with vehicle controls. Taken together these data demonstrate that cotreatment of C10 and cisplatin induced changes in all histological markers assessed in the proximal tubule.

C10 and cisplatin cotreatment exacerbates cisplatininduced inflammation
In order to determine the mechanism by which ceramide accumulation in the kidney contributes to worsened kidney function, injury, and pathology, we assessed various biologies known to play a role in AKI. Cisplatin-induced AKI is associated with a robust inflammatory response that is known to perpetuate renal injury. Treatment with cisplatin has been shown to induce the expression of many proinflammatory cytokines and chemokines in the kidney (16,55,56). To determine whether C10 administration exacerbated the inflammatory response induced by cisplatin, we measured the expression of pro-inflammatory cytokines and chemokines in the renal cortex. Expression of TNF-, IL-6, monocyte chemotactic protein-1 (MCP-1), and CXCL1 in the kidney cortex were increased 2-, 10-, 4-, and 50-fold, respectively, following cisplatin treatment as compared with vehicle controls (Fig. 6A-D). Mice treated with both C10 and cisplatin also had significantly increased expression of TNF-, IL-6, MCP-1, and CXCL1 to 3-, 27-, 6-, and 100-fold, respectively, over vehicle control animals (Fig. 6). More importantly, the renal expression of IL-6, MCP-1, and CXCL1 in cisplatin and C10-cotreated mice ; data are expressed as mean ± SEM. *P  0.05 as compared with cisplatintreated group as determined by two-way ANOVA. V, vehicle control; C10, C10-only treatment; C, cisplatin intraperitoneally at 25 mg/kg; C10+C, cisplatin and C10 cotreatment. Statistical differences were measured by two-way ANOVA followed by Bonferroni posttest. Data are expressed as mean ± SEM (n = 5-10). *P < 0.05, **P < 0.01, and ***P < 0.001 demonstrate statistical difference from vehicle treatment. #P < 0.05 demonstrates statistical difference from cisplatin treatment. was significantly higher than the cisplatin alone-treated mice (Fig. 6). Thus, these data reveal that inhibition of glucosylceramide synthase exacerbates the inflammatory response associated with cisplatin-induced AKI.

C10 and cisplatin cotreatment does not affect cisplatininduced changes in proliferation
Proximal tubule cells do not normally proliferate. However, following injury to the kidney, proximal tubule cells that survive are known to try to replace lost cells; surviving proximal tubule cells undergo a process of dedifferentiation, proliferation, and then redifferentiation and migration to sites of injury to replace dead or dying proximal tubule cells following cisplatin treatment (57). Some species of glucosylceramides are thought to play an integral role in cellular differentiation and proliferation and ceramide accumulation is also thought to prevent these processes (58). Thus, we determined whether inhibition of glucosylceramide synthase with C10 affected tubular proliferation following cisplatin treatment by assessing a marker of cell proliferation, PCNA. According to Western blot analysis of PCNA levels in the kidney cortex, cisplatin-treated mice had higher PCNA protein levels than the vehicle controltreated mice (Fig. 7A). C10 and cisplatin cotreatment did not further elevate PCNA levels beyond that of the cisplatin alone-treated mice (Fig. 7A). To determine the relative location of proliferating cells in relation to proximal tubules, kidney sections were costained for PCNA (reddish brown), LTA (black), and hematoxylin (bluish purple) (Fig. 7B). LTA is a proximal tubule-specific marker. Both cisplatin treatment (43.2 ± 13.4) and C10 and cisplatin cotreatment (46.7 ± 15.6) significantly increased the number of proliferating proximal tubule cells in comparison to vehicle controls, while C10 and cisplatin cotreatment did not alter the number of proliferating cells in the kidney, as compared with cisplatin alone treatment (Fig. 7C). Thus, inhibition of glucosylceramide synthase with C10 did not appear to alter the ability of kidney epithelial cells to proliferate and  For (B-F), scoring of the sections was performed in a blinded manner, using a scale of 0-4 (0 = not present, 1 = mild, 2 = moderate, 3 = severe, and 4 = very severe renal histological changes in the proximal tubules). Statistical differences were measured by two-way ANOVA followed by Bonferroni posttest. Data are expressed as mean ± SEM (n = 5-10). *P < 0.05 and **P < 0.01 demonstrate statistical difference from vehicle treatment. See the Fig. 4 legend for an explanation of the abbreviations.
replace the damaged proximal tubule cells as a result of the cisplatin.

C10 and cisplatin cotreatment exacerbates cisplatininduced apoptosis
A hallmark of cisplatin-induced AKI is the activation of cell death pathways, including apoptosis (16). Importantly, ceramides are known to play a role in apoptosis in response to many stimuli, including chemotherapy agents, such as cisplatin (23,27). Apoptosis is dependent on activation and cleavage of cysteine and aspartate proteases (caspases). Treatment with cisplatin increased protein levels of cleaved caspase 8 and cleaved caspase 3 in comparison to vehicle controls, while C10 and cisplatin coadministration further increased both cleaved caspase 3 and 8 levels as compared with cisplatin alone-treated mice (Fig. 8A). To visualize apoptotic cells in the renal cortex, TUNEL assays were performed on paraffin-embedded kidney sections (Fig. 8B). Quantification of TUNEL-stained kidneys indicated that cisplatin treatment alone (8.3 ± 1.9) and cisplatin and C10 cotreatment (13.3 ± 1.8) led to a significant increases in TUNEL-positive foci in the kidney cortex in comparison to vehicle controls (0.4 ± 0.1) (Fig. 8C). More importantly, C10 and cisplatin combination treatment significantly increased the number of TUNEL foci in comparison to cisplatin alone-treated animals (Fig. 8C). These data indicate that C10 treatment exacerbated cisplatin-induced apoptosis.

Similar changes in kidney cortex ceramides and hexosylceramides are observed in ischemia/reperfusion and myoglobinuric models of AKI
We wanted to determine whether the elevations in ceramides and hexosylceramides in the kidney were specific for cisplatin-induced AKI or if they were a general response in the kidney following injury. Thus, enzyme activities and sphingolipids were quantified in the I/R and glycerol models of AKI. Mice were subjected to 26 min of bilateral I/R and markers of kidney function were determined (42). Twenty-four hours following I/R, at the peak of kidney injury (42), long-chain CerS and aSMase activities were measured in the kidney cortex homogenate from sham-operated and I/R mice. Both CerS (Fig. 9A) and aSMase (Fig. 9B) activities were significantly elevated in the renal cortex 24 h following I/R. Ceramides and hexosylceramides were quantified at this time point. Levels of C 16 -, C 24:1 -, and C 26ceramides were significantly altered in the renal cortex 24 h following bilateral I/R (Fig. 9C). In addition to increased ceramides, the levels of C 16 -, C 24 -, and C 24:1 -hexosylceramides were increased in the renal cortex at 24 h post-I/R (Fig. 9D). In addition to I/R, we also examined CerS and aSMase Fig. 6. C10 treatment exacerbates inflammation in the kidney following cisplatin treatment. At 0 h, C57BL/6J mice were treated with cisplatin (25 mg/kg, ip) and C10 (20 mg/kg, oral gavage). C10 treatment was continued every 12 h for 48 h. Mice were euthanized 72 h following cisplatin treatment. Relative expression of TNF- (A), IL-6 (B), MCP-1 (C), and CXCL1 (D) were measured via real-time qRT-PCR. Statistical differences were measured by two-way ANOVA followed by Bonferroni posttest. Data are expressed as mean ± SEM (n = 5-10). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 demonstrate statistical difference from vehicle treatment. #P < 0.05 demonstrates statistical difference from cisplatin treatment. See the Fig. 4 legend for an explanation of the abbreviations. Levels of the C 16 -and C 26 -ceramides were significantly increased in the renal cortex 24 h following myoglobinuric AKI in rats (Fig. 10C). In addition, the levels of several hexosylceramide species were significantly elevated at the peak of myoglobinuric AKI, including C 16 -, C 20 -, C 24 -, and Fig. 8. C10 cotreatment increases cisplatin-induced apoptosis. At 0 h, C57BL/6J mice were treated with cisplatin (25 mg/kg, ip) and C10 (20 mg/kg, oral gavage). C10 treatment was continued every 12 h for 48 h. Mice were euthanized 72 h following cisplatin treatment. A: Western blot analysis was performed to assess relative protein levels of cleaved caspase 8 and cleaved caspase 3 in the renal cortex of mice. B: Representative photomicrographs of TUNEL immunofluorescence (200× final magnification). C: TUNEL quantification. Statistical differences were measured by two-way ANOVA followed by Bonferroni posttest. Data are expressed as mean ± SEM (n = 5-10). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 demonstrate statistical difference from vehicle treatment. #P < 0.05 demonstrates statistical difference from cisplatin treatment. See the Fig. 4 legend for an explanation of the abbreviations. Fig. 9. I/R of the kidneys alters the activity of sphingolipid metabolic enzymes, ceramides, and hexosylceramides in the kidney cortex. Mice were either sham operated or subjected to bilateral kidney I/R. Kidney cortex long-chain CerS activity (A), aSMase activity (B), ceramide species (C), and hexosylceramide species (D) were measured at the peak of kidney injury 24 h following I/R. Data are expressed as mean ± SEM (n = 5). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 demonstrate statistical difference from vehicle treatment.
C 24:1 -hexosylceramides (Fig. 10D). These data in I/R and myoglobinuric AKI are similar to what is observed in cisplatin-induced AKI and indicate that at the peak of kidney injury, there are increased activities of both long-chain CerS and aSMase, resulting in generation of ceramides and its metabolism to hexosylceramides. Data suggest that longchain CerS- and aSMase-mediated generation of ceramides and its subsequent metabolism to hexosylceramides may be a common response to injury in the kidney cortex.

DISCUSSION
The current literature raised the possibility of a role for sphingolipids in cisplatin-induced AKI, as many sphingolipids are known to play an integral role in cell pathways, such as cell death, inflammation, and cell stress processes that mediate cisplatin-induced AKI (5,8,23,59). Here, we examined the role of both ceramide and the metabolism of ceramide to glucosylceramide in cisplatin-induced AKI. Our data show that cisplatin treatment elevates ceramide and glucosylceramide levels in the renal cortex and that inhibition of two ceramide synthesis pathways prevents the elevation of ceramides and glucosylceramides while protecting mice from cisplatin-induced AKI. To further elucidate whether it is ceramide or its metabolism to glucosylceramides that plays a detrimental role in cisplatin-induced AKI, we utilized a specific inhibitor of glucosylceramide synthase (C10) to inhibit metabolism of ceramides to glucosylceramides following treatment with cisplatin. C10 administration attenuated cisplatin-induced glucosylceramide accumulation, but exacerbated the ceramide accumulation in the kidney cortex. Furthermore, our data indicate that C10 administration worsened cisplatin-induced AKI. Treatment of mice with C10 exacerbated cisplatin-induced increases in the expression of pro-inflammatory chemokines and cytokines; however, C10 administration did not alter the ability of proximal tubules to proliferate following cisplatin treatment. Data indicated that C10 administration exacerbated apoptosis in the renal cortex of cisplatintreated mice, as evidenced by enhanced caspase activation and the numbers of TUNEL-positive cells. Taken together, Fig. 10. The activity of sphingolipid metabolic enzymes, ceramides, and hexosylceramides in the kidney cortex are increased in a rat model of rhabdomyolysis. The quadriceps muscle of rats was injected with 1 ml PBS (control) or glycerol. Kidney cortex long-chain CerS activity (A), aSMase activity (B), ceramide species (C), and hexosylceramide species (D) were measured at the peak of kidney injury 24 h following injection. Data are expressed as mean ± SEM (n = 5). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 demonstrate statistical difference from vehicle treatment. these data suggest that ceramides mediate cisplatin-induced AKI, while buffering their levels via their metabolism to glucosylceramides is protective. Hexosylceramides represent only a very small fraction of the total glycosphingolipids in the kidney. We did not quantify additional lipids in this class. However, the relative masses of ceramide and hexosylceramides measured in the kidney would suggest that additional glycosphingolipid species are also being generated following cisplatin. Thus, it is important to point out that data presented do not rule out a role for complex glycosphingolipids in AKI. Our data merely suggest that flux of ceramide into glycosphingolipids may be a mechanism by which the kidney buffers ceramide levels to attenuate injury or attempt to recover from injury. We also cannot rule out that inhibition of glucosylceramide synthase sensitizes the kidney to cisplatin-induced AKI as a result of not only exacerbated ceramides, but also the lack of production of a glycosphingolipid species that may serve a protective role. It is likely that a combination of both increased production of ceramides and decreased flux of ceramides into protective glycosphingolipid species is the mechanism by which C10 sensitizes the kidney to cisplatin. Similar changes in enzyme activities and sphingolipids are observed in the renal cortex following I/R and myoglobinuric AKI, suggesting that this may be a common response of the kidney following injury.
The kidney cortex contains many different cell types. Measurements of lipids performed in this study were performed on total kidney cortex homogenate. Thus, there could be changes in other lipids in the sphingolipid pathway that occur in a cell type-specific manner that we were unable to detect. For example, dihydroceramide synthase 1 regulates sensitivity to cisplatin (60) and acts in the de novo synthesis and salvage pathways. It mainly catalyzes production of C 18 -ceramides, which are at very low levels in the kidney and, if produced in a specific cell type or in a specific subcellular compartment of a specific cell type in response to cisplatin, would likely not be detected with MS of total kidney cortex homogenate (60). Thus, future studies are still needed to fully differentiate the specific lipids involved in cisplatin-induced AKI and recovery from injury so that we can begin to design mechanisms to target this pathway in a way that will improve the ability of cisplatin to reduce tumor burden while protecting the kidney from cisplatin.
Glucosylceramides are known to play a role in inflammatory responses (61,62). The literature indicates that inhibition of glucosylceramide synthesis dampens the inflammatory response (35,63), suggesting that inhibition of glucosylceramide synthesis in cisplatin-induced AKI could also reduce inflammatory pathways. Inflammatory pathways are known to perpetuate cisplatin-induced kidney injury; however, our data suggest that ceramide and/or a lack of its metabolism to a protective glycosphingolipid species play a role in the inflammatory response in cisplatininduced AKI. Importantly, there are instances in which ceramides are known to mediate inflammatory pathways. For example, PDMP, the first generation glucosylceramide synthase inhibitor, is known to induce ceramide accumulation intracellularly, which has been reported to be associated with increased expression of pro-inflammatory chemokines and cytokines, including IL-6 (64). Ceramides are also known to mediate inflammation in disease states, such as diabetes and obesity, where ceramides are known to mediate toll-like receptor 4 (TLR4)-mediated inflammation (65). Ceramide has also been suggested as a TLR4 agonist (65,66). Furthermore, TLR4 signaling mediates the expression of many pro-inflammatory cytokines, including TNF- and IL-6 (66). Also, ceramide and pro-inflammatory chemokine and cytokine levels are augmented in Farber disease, a deficiency in acid ceramidase. Alayoubi et al. (67) showed that patients with Farber disease have increased MCP-1 and circulating levels of leukocytes, including neutrophils, monocytes, and eosinophils. Our data indicate that C10 administration augments cisplatin-induced elevations in the expression of pro-inflammatory chemokines and cytokines, including TNF-, IL-6, MCP-1, and CXCL1. It is known that further increasing the expression of these proinflammatory chemokines and cytokines is detrimental to cisplatin-induced AKI. Ramesh and Reeves (68) showed that inhibition of TNF- production protects from cisplatin-induced AKI. The exact role of MCP-1 and IL-6 is not clear, but the expression levels of both MCP-1 and IL-6 are known to directly correspond with the level of injury resulting from cisplatin treatment (69,70). Taken together, this suggests that C10 and cisplatin coadministration exacerbates the pro-inflammatory response associated with cisplatin-induced AKI, thereby, worsening kidney injury.
During cisplatin-induced AKI, many pathways of apoptosis are activated, including both the extrinsic and intrinsic pathways of apoptosis (16). In the extrinsic pathway of apoptosis, TNF- or other death receptor ligands can activate death receptors on the proximal tubule cell surface leading to caspase 8 activation and the initiation of downstream effector caspases, such as caspase 3, leading to apoptosis (73). During the intrinsic pathway of apoptosis, an apoptotic stimulus activates pro-apoptotic factors leading to mitochondrial outer membrane permeability (MOMP) (16). Following MOMP, pro-apoptotic factors, such as cytochrome c, are released into the cytoplasm leading to the downstream activation of caspases, including caspase 3 (16). Importantly, our data show that C10 administration exacerbates cisplatin-induced apoptosis, as evidenced by increased caspase activation and TUNEL assays. Siskind, Kolesnick, and Colombini (72,73) showed that treatment of baby mouse kidney cells with cisplatin led to elevations in intracellular ceramide levels and increased ceramide mediated apoptosis. Taken together, this suggests that C10 administration of cisplatin-treated mice increases the activation apoptotic pathways at least in part because of ceramide accumulation.
It is important to note that cisplatin is clinically used as an anti-cancer chemotherapeutic agent. Our data indicate that inhibition of ceramide generation protects mice from cisplatin-induced kidney injury; however, it is welldocumented that ceramide is a necessary molecule for cisplatin's anti-tumor efficacy. Unfortunately, our data indicate that inhibition of glucosylceramide synthesis exacerbates cisplatin-induced AKI, as a result of exacerbating cisplatininduced ceramide accumulation. Importantly, Yamane et al. (74) showed that administration of PDMP (a glucosylceramide synthase inhibitor) to lung cancer cells induced ER stress, autophagy, and apoptosis as a result of PDMP-induced ceramide accumulations. The work done by Yamane et al. (74) highlights the fact that, while drugs acting to increase ceramide levels may have the potential to sensitize cancer cells to cisplatin treatment, these drugs may further exacerbate cisplatin-induced AKI, as evidenced by the data presented in this report. Thus, it may be beneficial when developing agents targeting sphingolipid metabolism, to target portions of the pathway that are either upstream or downstream of ceramide or to develop formulations that enable targeting drugs specifically to the tumor. Taken together, this body of work suggests that, under conditions of high ceramide generation, metabolism of ceramide to glycosphingolipids is a mechanism by which the kidney buffers ceramide levels in an attempt to prevent or recover from injury.