Colistin-Induced Nephrotoxicity in Mice Involves the Mitochondrial, Death Receptor, and Endoplasmic Reticulum Pathways

Abstract

Nephrotoxicity is the dose-limiting factor for colistin, but the exact mechanism is unknown. This study aimed to investigate the roles of the mitochondrial, death receptor, and endoplasmic reticulum pathways in colistin-induced nephrotoxicity. Mice were intravenously administered 7.5 or 15 mg of colistin/kg of body weight/day (via a 3-min infusion and divided into two doses) for 7 days. Renal function, oxidative stress, and apoptosis were measured. Representative biomarkers involved in the mitochondrial, death receptor, and endoplasmic reticulum pathways were investigated, and the key markers involved in apoptosis and autophagy were examined. After 7-day colistin treatment, significant increase was observed with blood urea nitrogen, serum creatinine, and malondialdehyde, while activities of superoxide dismutase (SOD) and catalase decreased in the kidneys. Acute tubular necrosis and mitochondrial dysfunction were detected, and colistin-induced apoptosis was characterized by DNA fragmentation, cleavage of poly(ADP-ribose) polymerase (PARP-1), increase of 8-hydroxydeoxyguanosine (8-OHdG), and activation of caspases (caspase-8, -9, and -3). It was evident that colistin-induced apoptosis involved the mitochondrial pathway (downregulation of Bcl-2 and upregulation of cytochrome C [cytC] and Bax), death receptor pathway (upregulation of Fas, FasL, and Fas-associated death domain [FADD]), and endoplasmic reticulum pathway (upregulation of Grp78/Bip, ATF6, GADD153/CHOP, and caspase-12). In the 15-mg/kg/day colistin group, expression of the cyclin-dependent kinase 2 (CDK2) and phosphorylated JNK (p-JNK) significantly increased (P < 0.05), while in the 7.5-mg/kg/day colistin group, a large number of autophagolysosomes and classic autophagy were observed. Western blot results of Beclin-1 and LC3B indicated that autophagy may play a protective role in colistin-induced nephrotoxicity. In conclusion, this is the first study to demonstrate that all three major apoptosis pathways and autophagy are involved in colistin-induced nephrotoxicity.

INTRODUCTION

Colistin, also known as polymyxin E, is an “old” peptide antibiotic and has been increasingly used over the last decade as a last-line therapy for treatment of infections caused by multidrug-resistant Gram-negative bacteria, namely, Pseudomonas aeruginosa, Acinetobacter baumannii, and Klebsiella pneumoniae (1–3). However, recent pharmacokinetic and pharmacodynamic studies have indicated that the current recommended dosage regimens are suboptimal (4–6). With colistin monotherapy, rapid emergence of resistance has been reported in vitro using pharmacokinetic/pharmacodynamic models to mimic the pharmacokinetics of colistin in patients (7–12). Unfortunately, simply increasing doses is not an option, as colistin-induced nephrotoxicity is the dose-limiting factor (4) and can occur in approximately 60% of patients (13, 14).

Previous pharmacokinetic studies indicated that polymyxins are substantially reabsorbed after filtration by glomeruli in the kidney in animals and patients (15–18). Recently, a correlative microscopy study using synchrotron X-ray fluorescence microscopy first revealed that intracellular concentrations of polymyxins in rat (NRK-52E) and human (HK-2) kidney proximal tubular cells are thousands of times higher than extracellular concentrations (J. Li, presented at the 1st International Conference on Polymyxins, Prato, Italy, 2 to 4 May 2013). It has been reported that the oxidative stress (20–25) and caspase-dependent apoptosis (26; J. Li, 1st Internat. Conf. Polymyxins) play key roles in colistin-induced nephrotoxicity. Several antioxidants, including melatonin, ascorbic acid, proanthocyanidin extract, and N-acetylcysteine have been examined for their protective activity against colistin-induced renal injury (20–24). A recent transcriptomics study indicated that in mouse kidneys colistin treatment caused altered expression of the genes involved in cell cycle arrest (25). However, the exact mechanism of colistin-induced nephrotoxicity has not been elucidated. In particular, the molecular mechanism and signaling pathway of colistin-induced apoptosis have not been investigated. The death receptor, mitochondrial, and endoplasmic reticulum pathways are the three key pathways in apoptotic cell death (27). Recent studies reported that the mitochondrial pathway participates in colistin-induced neurotoxicity in vitro and in vivo (28, 29). Using cell culture, it has been demonstrated that polymyxins activate caspase-3, -8, and -9 and induce DNA breakage and mitochondrial morphology changes in rat kidney tubular NRK-52E cells (30; J. Li, 1st Internat. Conf. Polymyxins). The current study aimed to examine the roles of death receptors, mitochondrial, and endoplasmic reticulum pathways in colistin-induced nephrotoxicity using a mouse model.

MATERIALS AND METHODS

Chemicals and reagents.

Colistin sulfate (CAS 1264-72-8) (20,400 U/mg) was purchased from Zhejiang Shenghua Biology Co., Ltd. (Zhengjiang, China). Sodium dodecyl sulfonate (SDS), 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), aprotinin, leupeptin, pepstatin A, and phenylmethylsulfonyl fluoride (PMSF) were obtained from AMRESCO Inc. (Solon, OH, USA). All other chemicals were of analytical grade.

Animal experiments.

All animal studies were approved by the Institutional Animal Care and Use Committee at the China Agricultural University. Adult Kunming mice (female, 6 to 8 weeks, 18 to 22 g) were obtained from Vital River Animal Technology Co., Ltd. (Beijing, China). Mice had free access to food and water during all experimental periods. The animal laboratory was maintained at approximately 22°C and 50% relative humidity with a 12-h light-dark cycle. An acclimation period of 1 week was employed prior to the experiments.

Thirty animals were randomly divided into three groups (n = 10). Group 1 was the control, and the mice were intravenously (i.v.) administered sterile saline through the tail vein. Groups 2 and 3 received i.v. 7.5 or 15 mg of colistin/kg of body weight/day (colistin sulfate in sterile saline, divided into two doses), respectively. All mice were treated for 7 days. Twelve hours after the last dose, blood samples (0.5 ml) were collected, and both kidneys were collected immediately. The mice were euthanized by intraperitoneal injection of an overdose of sodium pentobarbital (80 mg/kg) (Sigma-Aldrich, New York, NY, USA). Blood samples were centrifuged at 3,000 × g for 10 min (Sigma, Goettingen, Germany) for measurements of blood urea nitrogen (BUN) and serum creatinine. The kidneys were divided into four parts for the biochemical and histopathological examinations below.

Preparation of tissue homogenate.

The left kidneys of all mice were divided into two parts. One part was for histopathological examination, and the other part was homogenized (2,000 rpm, 5 min) at 4°C in 9 volumes (approximately 1 ml per 0.1 g tissue) of cold Tris buffer (0.01 M Tris-HCl, 0.1 mM EDTA-Na₂, 0.01 M sucrose, 0.9% saline; pH 7.4). The homogenates were centrifuged (3,000 × g, 15 min) at 4°C, and the supernatant samples were stored at −80°C until analysis. The protein content was determined (n = 10) using a BCA protein assay kit (Nanjing Jiancheng Bio-Corporation, Nanjing, China).

Biochemical analyses.

Concentrations of BUN, serum creatinine, and malondialdehyde (MDA) were determined using commercial assay kits (Nanjing Jiancheng Bio-Corporation). Activities of superoxide dismutase (SOD) and catalase (CAT) in the kidney homogenates were quantified using the reduction of nitrite and ammonium molybdate, respectively (Nanjing Jiancheng Bio-Corporation). The values of concentration and activity were normalized by protein contents.

Histopathological and ultrastructural examination.

The right kidney of each mouse was fixed in 10% neutral buffered formalin, embedded in paraffin, sectioned at 5 μm, and stained with hematoxylin-eosin. The histopathological scoring was conducted (25), and renal tubular damage was evaluated using a semiquantitative score (SQS) (20). Three grades were employed: grade 1, mild acute tubular damage with tubular dilation, prominent nuclei, and several pale tubular casts; grade 2, severe acute tubular damage with necrosis of tubular epithelial cells and numerous tubular casts; and grade 3, acute cortical necrosis/infarction of tubules and glomeruli with or without papillary necrosis. The grades were given the following scores: grade 1 = score 1, grade 2 = score 4, and grade 3 = score 10. The percentages of the kidney slices affected were assigned the following scores: <1% = score 0, 1% to <5% = score 1, 5% to <10% = score 2, 10% to <20% = score 3, 20% to <30% = score 4, 30% to <40% = score 5, and ≥40% = score 6. The overall score was calculated as the product of percentage score and grade score. Finally, an SQS for renal histological changes was assigned as follows: SQS 0 = no significant change (overall score < 1), SQS + 1 = mild damage (overall score 1 to <15), SQS + 2 = mild to moderate damage (overall score 15 to <30), SQS + 3 = moderate damage (overall score 30 to <45), SQS + 4 = moderate to severe damage (overall score 45 to <60), and SQS + 5 = severe damage (overall score 60). The final results were expressed as the SQS (mean ± standard deviation [SD]).

In each group, three mice were randomly selected, and the cortex sections of the right kidneys were isolated and cut into approximately 1-mm cubes for the ultrastructural observation. The tissue samples were fixed with 2.5% glutaraldehyde in 0.12 mM phosphate buffer (pH 7.2) and kept overnight at 4°C. Postfixation was performed with 2% osmium tetroxide in 0.1 M cacodylate buffer (pH 7.4) for 2 h at 4°C. The specimens were then dehydrated in graded acetone solutions and embedded in the 812 epoxy resin (Okenshoji Co., Ltd., Tokyo, Japan). Ultrathin sections (70 to 80 nm) were prepared, counterstained with uranyl acetate and lead citrate, and photographed with a JSM25610LV transmission electron microscope (JEOL Ltd., Tokyo, Japan).

Mitochondrial function and ATP levels. (i) Isolation and purification of mitochondria.

Mitochondria in the mouse kidney tissues were isolated by differential centrifugation with minor modifications (31). Briefly, the homogenate was centrifuged (800 × g, 10 min) at 4°C. The supernatant was centrifuged (12,000 × g, 10 min) at 4°C, and the mitochondrial pellet was collected. The final pellet was resuspended in the buffer (0.01 M Tris-HCl, 0.1 mM EDTA-Na₂, 0.01 M saccharose, 0.9% saline; pH 7.4). The protein concentration of the mitochondrial suspension (n = 10) was determined using the BCA protein assay.

(ii) Measurement of MPT.

The isolated mitochondria were diluted to 0.5 mg/ml and incubated in PT-2 buffer (250 mM sucrose, 2 mM HEPES, 0.5 mM KH₂PO₄, 4.2 mM potassium succinate; pH 7.4). The mitochondrial permeability transition (MPT) was initiated by adding 0.05 mM calcium chloride and monitored by measuring the decrease of A₅₄₀ at 37°C in 5 min for evaluation of the swelling of the mitochondria. The results were normalized to the value for the control.

(iii) Detection of Δψ_m.

The mitochondrial membrane potential (Δψ_m) was monitored in the presence of the fluorescent dye rhodamine 123 (Rh123) (Sigma Chemical Co., MO, USA) (28). Briefly, the mitochondria were diluted to 0.5 mg/ml in the reaction buffer and incubated for 3 min. Fluorescence with excitation/emission at 503/527 nm was measured at 0 and 3 min in the reaction buffer (250 mM sucrose, 2 mM HEPES, 0.5 mM KH₂PO₄, 4.2 mM sodium succinate [pH 7.4], and 0.3 mM Rh123) using an F-4500FL spectrophotometer (Hitachi High-Technologies Co., Japan). The alteration of the Δψ_m was calculated by the decrease of the fluorescence. The final results were normalized to the value for the control.

(iv) Activities of the mitochondrial respiratory chain.

To assess the activities of the mitochondrial respiratory chain, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reduction method was employed (28). The reaction mixture contained mitochondrial preparation (80 to 100 μg protein) and 0.02 ml MTT (0.1 mg/ml) and was incubated at 37°C for 30 min. Samples were centrifuged at 1,000 × g for 5 min at room temperature to obtain the formazan pellet. The pellet was dissolved in 1 ml acidic isopropanol, and the mixture was recentrifuged at 1,000 × g for 5 min at room temperature. The absorbance of the supernatant was measured at 595 nm. The final results were normalized to the value for the control.

(v) Determination of ATP levels.

The ATP levels of mitochondria in kidney tissues were measured with an ATP bioluminescence assay kit (Beyotime Bioengineering Institute, Beijing, China). Briefly, the level of ATP was determined by mixing 50 μl of the mitochondrial supernatant with 50 μl of luciferase reagent. The emitted light was measured using a microplate luminometer and linearly related to the ATP concentration (Molecular Devices, Sunnyvale, CA, USA).

Immunohistochemical examinations and enzyme-linked immunosorbent assays (ELISA).

Following the histopathological evaluation (above), the paraffin-embedded sections of kidneys were stained with streptavidin-biotin-peroxidase for immunohistochemical experiments. The tissue sections were deparaffinized and dehydrated in graded alcohols. Antigen retrieval was conducted using the microwave method, followed by treatment with 3% hydrogen peroxide (10 min) and 5% bovine serum albumin in buffered solution. The specimens were then incubated at 4°C overnight with primary antibodies: rabbit polyclonal antibodies against Bcl-2 (1:200; Santa Cruz Biotechnology, Inc., CA, USA), cleaved caspase-3 (1:200), cleaved caspase-9 (1:200; Wuhan Boster Bio-engineering Limited Co., Wuhan, China), Bax (1:200), Beclin-1 (1:200), and microtubule-associated protein 1 light chain-3B (LC3B) (1:200; ProteinTech Group, Inc., Chicago, IL, USA). The specimens were also incubated with mouse monoclonal antibody against 8-hydroxydeoxyguanosine (8-OHdG) (1:200; Japan Institute for the Control of Aging, Fukuroi, Japan), and phosphorylated p38 (p-p38) (1:200; Cell Signaling Technology, Beverly, MA, USA). On the second day, the specimens were incubated with biotin-labeled rabbit anti-mouse (1:200) or goat anti-rabbit IgG-horseradish peroxidase (HRP) conjugates (1:200; Santa Cruz Biotechnology) for 30 min. Subsequently, the samples were stained with diaminobenzidine (DAB) chromogen solution (Zhongshan Cambridge Reagent Company, Beijing, China), followed by hematoxylin, and mounted in xylene-based mountant. Ten fields of each section were randomly selected to evaluate the levels of expression of the proteins. For Bcl-2, Bax, cleaved caspase-3, and cleaved caspase-9, 10 images (40× magnification) were randomly selected, and the average of positive cells was obtained in each section. For LC3B, Beclin-1, 8-OHdG, and p-p38, the results were obtained using intensity on a semiquantitative scale of 0 to 3, 0 for none, 1 for weakly positive (pale yellow), 2 for positive (brown), and 3 for strongly positive (tan) (25).

The activities of caspase-3 and -9 and DNA fragmentation in the kidney tissues were determined using the caspase-3 and -9 activity assay kits and Cell Death Detection ELISA^Plus kit (Roche Applied Sciences, Basel, Switzerland), respectively.

Western blotting.

According to the marked histopathology changes observed above, the kidney tissues from three mice in each group were homogenized in ice-cold lysis buffer (100 mM Tris-HCl, 2% [mass/volume] SDS, 10% [vol/vol] glycerol and pH 7.4; 1 mM PMSF, 1 μg/ml aprotinin, 1 μg/ml leupeptin, and 1 μg/ml pepstatin A were added to the buffer before the experiment) and then ultrasonicated (i.e., 5 s ultrasonication and 6 s pause in each cycle for 5 times, power 30 W) using an Ultrasonic Processor (Branson, MO, USA). Tissue lysate samples were centrifuged at 145,000 × g for 15 min at 4°C, and protein concentration was measured using the BCA protein assay kit. Western blotting was conducted with 100 μg protein per lane (32). The following primary antibodies were employed: rabbit polyclonal antibodies against cleaved caspase-3 (1:1,000), cleaved caspase-9 (1:1,000), Beclin-1 (1:2,000), LC3B (1:1,000), p53 (1:1,000), p21 (1:1,000), Bcl-2 (1:1,000), caspase-3 (1:1,000), cyclin-dependent kinase 2 (CDK2) (1:1,000), apoptosis-induced factor (AIF) (1:2,000), cytochrome C (cytC) (1:2,000; Santa Cruz Biotechnology, Inc.), activating transcription factor 6 (ATF6; 1:1,000), growth arrest and DNA damage inducible gene 153/C/EBP-homologous protein (GADD153/CHOP) (1:1,000), glucose-regulated protein 78 kDa/Bax-inhibiting peptide (GRP78/Bip; 1:1,000), Bid (1:1,000), Bax (1:1,000), caspase-9, nuclear factor kappa-light-chain enhancer of activated B cells (NF-κB/p65) (1:1,000) (ProteinTech Group, Inc., Chicago, IL, USA), caspase-8 (1:1,000), Fas (1:1,000), p38 (1:1,000); Fas ligand (FasL; 1:1,000), poly(ADP-ribose) polymerase (PARP-1) (1:1,000) (Beyotime Institute of Biotechnology Co., Ltd., Haimen, Jiangsu, China), p-Janus kinase (JNK)/stress-activated protein kinase (SAPK) (1:1,000) (Cell Signaling Technology, Beverly, MA, USA); and mouse monoclonal antibody against p-p38 (1:1,000), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1:1,000), and β-actin (1:1,000; Santa Cruz Biotechnology). Secondary antibodies, including goat anti-rabbit IgG (1:5,000) or rabbit anti-mouse IgG (1:5,000) were incubated for 1 h at room temperature. The protein expression results were normalized to β-actin or GAPDH (for the death receptor pathway only) and analyzed using ImageJ (National Institute of Mental Health, Bethesda, MD, USA).

Statistical analysis.

All data are presented as means ± SDs unless specified otherwise. Statistical analysis was conducted using SPSS v13.0 (SPSS Inc., Chicago, IL, USA), and figures were prepared using GraphPad Prism 5.0 (GraphPad Software, Inc., La Jolla, CA). Data from the control and treatment groups were analyzed with one-way analysis of variance (ANOVA), followed by least significant difference (LSD) posthoc test. A P value of <0.05 was considered significant.

RESULTS

Clinical observations and biochemical analyses.

In this study, all mice tolerated colistin (7.5 and 15 mg/kg/day) for 7 days. In the 15-mg/kg/day colistin group, the BUN and serum creatinine levels significantly increased after 7-day treatment (P < 0.01; Fig. 1A and B) compared to the control group. The levels of MDA, a marker of oxidative stress-induced renal impairment, significantly increased in both colistin groups (P < 0.05; Fig. 1C), while the activities of SOD and CAT in renal tissues were significantly lower in the colistin-treated mice (P < 0.05; Fig. 1D and E). The results for MDA, SOD, and CAT demonstrate that oxidative stress played a role in colistin-induced nephrotoxicity.

FIG 1.

Changes of the renal function and oxidative stress markers in mouse kidneys after colistin treatment (7.5 and 15 mg/kg/day). The results are represented as means plus standard deviations (SD) (error bars) (n = 10). (A) BUN; (B) serum creatinine; (C) MDA; (D) SOD; (E) CAT. Values that are significantly different from the values for the control group are indicated by asterisks as follows: *, P < 0.05; **, P < 0.01. L, liter; total pro, total protein.

Histopathological analyses.

The kidney histopathological examination and their SQS scores are shown in Fig. 2. In the control group, tubules and glomeruli had no marked injury with an average SQS score of 0.25 (Fig. 3A and D). The kidneys of the colistin-treated mice were significantly injured in a dose-dependent manner. In the 7.5-mg/kg/day colistin group, dilated tubules and cast formation were observed (Fig. 2B). In the 15-mg/kg/day colistin group, serious tubular necrosis was evident with degeneration, necrosis, and cast formation (Fig. 2C). Their corresponding SQSs significantly increased to 1.75 ± 0.5 and 2.75 ± 1.26 for the 7.5- and 15-mg/kg/day colistin groups, respectively (Fig. 2D).

FIG 2.

Representative histopathological images and the semiquantitative scores (SQS). (A) Control group. No marked injury is visible. (B) Colistin 7.5-mg/kg/day group. Dilated tubules (yellow arrows) and a number of cast formations (yellow arrowheads) within the cortex are evident. (C) Colistin 15-mg/kg/day group. The epithelium cells of both proximal and distal convoluted tubules showed degeneration, necrosis (yellow arrows), and cast formations (yellow arrowheads), with protein exudation in the glomerular cells (yellow triangle). The sections in panels A to C were stained with hematoxylin and eosin stain (H&E stain). Bars = 100 μm. (D) SQS values for the control and colistin-treated groups (n = 4). Values that are significantly different from the values for the control group are indicated by asterisks as follows: *, P < 0.05; **, P < 0.01.

FIG 3.

Changes of mitochondrial ultrastructure, function, and ATP levels in the kidneys of the mice left untreated or treated with colistin. (A) Representative mitochondrial ultrastructure in the control. Normal mitochondria (arrows) and a very small number of autophagolysosomes (arrowheads) are shown. (B) Representative changes of mitochondrial ultrastructure after treatment with 7.5 mg/kg/day colistin for 7 days, showing swollen mitochondria (arrows) and a large number of autophagolysosomes (arrowheads). (C) Typical macroautophagy (thick arrows) was observed in the 7.5-mg/kg/day group. (D) Representative changes of mitochondrial ultrastructure after treatment with 15-mg/kg/day colistin for 7 days. More-severe mitochondrial pathological changes (arrows) are shown than in the mitochondria in panel B. Marked endoplasmic reticulum expansion was also observed (triangles). Bars = 2,000 nm. (E to H) Changes of mitochondrial membrane permeability (MPT), mitochondrial membrane potential (Δψ_m), activities of the mitochondrial respiratory chain and ATP levels. The results are represented as means plus SD (n = 10). Values that are significantly different from the values for the control group are indicated by asterisks as follows: *, P < 0.05; **, P < 0.01.

Changes of mitochondrial ultrastructure, function, and ATP levels.

Compared with the kidneys from the control mice (Fig. 3A), a number of pathological changes were observed in the mitochondria from the kidneys treated with colistin in a dose-dependent manner. In the kidneys of the mice treated with both colistin regimens for 7 days, the mitochondria were swollen and ruptured, and mitochondrial cristae disappeared (Fig. 3B and D). Macroautophagy and autophagolysosomes (Fig. 3B and C) were also observed. However, substantially less autophagolysosomes and more-severe mitochondrial pathological changes were observed in the 15-mg/kg/day group (Fig. 3D). The Ca²⁺-induced MPT significantly increased to 123% ± 9.58% and 159% ± 12.7% in the 7.5- and 15-mg/kg/day groups, respectively, compared to the control (Fig. 3E). By monitoring the dynamic fluorescence quenching of Rh123, it was revealed that Δψ_m in the colistin-treated mouse kidneys significantly decreased (Fig. 3F). Both the activity of the mitochondrial respiratory chain (shown by the activity of mitochondrial succinate dehydrogenase; Fig. 3G) and ATP levels (Fig. 3H) significantly decreased in both colistin-treated groups.

Immunohistochemical analysis and ELISA examination of key biomarkers of apoptosis pathways and autophagy.

The expression of Bcl-2, Bax, cleaved caspase-9, cleaved caspase-3, 8-OHdG, and p-p38 in the kidney tissues of the control and colistin-treated mice were measured (Fig. 4). In the mice treated with colistin (7.5 and 15 mg/kg/day) for 7 days, the expression of Bcl-2 significantly decreased (Fig. 4A), while the expression of Bax, cleaved caspase-9, cleaved caspase-3, 8-OHdG, and p-p38 significantly increased in a dose-dependent manner (Fig. 4B to F). The expression of the classic markers of autophagy LC3B and Beclin-1 significantly increased only in the kidney tissues treated with 7.5 mg colistin/kg/day, but not in the 15-mg/kg/day colistin group (Fig. 4G and H). Consistent with the increased expression of cleaved caspase-9 and -3 (Fig. 4C and D), the activities of caspase-9 and -3 significantly increased in the homogenates of colistin-treated kidney tissue (Fig. 4I and J), and a significantly increased DNA fragmentation was observed (Fig. 4K).

FIG 4.

Expression of Bcl-2 (A), Bax (B), cleaved caspase-9 (C), cleaved caspase-3 (D), 8-OHdG (E), p-p38 (F), LC3B (G), and Beclin-1 (H) (n = 4). (I to K) Activities of caspase-9 (I) and caspase-3 (J) and DNA fragmentation in the mouse kidneys (K) (n = 10). In panels A to H, panels a, b, and c show the representative images for the control group, 7.5-mg/kg/day colistin group, and 15-mg/kg/day colistin group, respectively; panels d show the quantitative results of the expression. Values that are significantly different from the values for the control group are indicated by asterisks as follows: *, P < 0.05; **, P < 0.01. Bars = 50 μm (A to E) and 100 μm (F to H).

Western blotting of key biomarkers in the mitochondrial pathway, death receptor pathway, endoplasmic reticulum pathway, cell cycle, MAPK family, and autophagy.

Expression of key biomarkers of the mitochondrial pathway was first examined using Western blotting. As shown in Fig. 5A and F, significantly decreased expression of Bcl-2 and increased expression of Bax, cytC, cleaved caspase-9 and -3 were evident in the 7.5 mg/kg/day colistin group (P < 0.05); however, the expression of AIF did not significantly change. In the kidney samples collected from the mice treated with 15 mg/kg/day colistin, the expression of all markers significantly increased except Bcl-2, which significantly decreased (P < 0.05). For the death receptor pathway, compared to the untreated control, the biomarkers with significantly increased expression in the kidneys after colistin treatment include Fas, FasL, truncated Bid (tBid), cleaved capase-8, and FADD (P < 0.05 or 0.01). The only exceptions are FasL and FADD in the 7.5-mg/kg/day colistin group; their expression slightly increased (P > 0.05) (Fig. 5B and G). Four biomarkers for the endoplasmic reticulum pathway were also examined, and the expression of Grp78/Bip, cleaved ATF6, caspase-12, and GADD153/CHOP significantly increased (P < 0.05) in the 15-mg/kg/day colistin group compared to those in the control group (Fig. 5C and H). As shown in Fig. 5D, E, and I, the expression of p53, p21, and the mitogen-activated protein kinase (MAPK) family markers, including p-p38 and NF-κB/p65, significantly increased in a dose-dependent manner (P < 0.05). Furthermore, we also examined the autophagy biomarkers LC3B and Beclin-1 (Fig. 5J, K, and L). Compared to the control, the expression of Beclin-1 and LC3B-II/LC3B-I ratio was significantly higher (P < 0.01) in the 7.5-mg/kg/day colistin group only; for the 15-mg/kg/day colistin group, a significant increase was observed only for Beclin-1 expression (P < 0.05).

FIG 5.

Western blotting of key biomarkers for the mitochondrial pathway (A and F), death receptor pathway (B and G), endoplasmic reticulum pathway (C and H), cell cycle (D and I), MAPK family (E and I), and autophagy (J, K, and L). In panels A to E and J, lanes a, b, and c show representative lanes from the control group, 7.5-mg/kg/day colistin group, and 15 mg/kg/day colistin group, respectively. Each experiment was conducted in three replicates, and the results are represented as means plus SD (n = 3). Values that are significantly different from the values for the control group are indicated by asterisks as follows: *, P < 0.05; **, P < 0.01.

DISCUSSION

Recent pharmacokinetic/pharmacodynamic studies of critically ill patients indicate that the currently recommended dosage regimens of colistin are suboptimal and that higher doses are required to minimize potential emergence of resistance (4–6, 33). However, colistin-induced nephrotoxicity is the major dose-limiting factor, and its molecular mechanism is unclear. As polymyxin B is much less widely used in patients (3, 34, 35), colistin was employed in the present study.

The significant increase of BUN and serum creatinine concentrations in both colistin-treated groups (Fig. 1A and B) indicated that severe renal injury occurred after the 7-day treatment. Tubular degeneration and necrosis were consistently observed in the colistin-treated mouse kidney samples (Fig. 2). Similar findings have been reported in a recent transcriptomics study of mice (25). Several in vitro and in vivo studies suggested that oxidative stress plays an important role in colistin-induced apoptosis in nerve tissues (28, 36) and nephrotoxicity (20–24; J. Li, 1st Internat. Conf. Polymyxins). In the present study, the concentration of MDA, a biomarker for lipid peroxidation, significantly increased in the mouse kidneys treated with colistin (Fig. 1C). In contrast, the activities of two key antioxidant enzymes SOD and CAT significantly decreased (Fig. 1D and E), indicating decreased ability to catalyze decomposition of reactive oxygen species (ROS) after colistin treatment. An imbalance between ROS and the antioxidant reserve may lead to altered structure and function of proteins, lipids, and DNA, damage of lipid membranes, impaired cellular catalytic reactions, and cell apoptosis (37, 38). Apoptotic cells are characterized by typical morphological and biochemical hallmarks, such as chromatin condensation and nuclear fragmentation, and phagocytosis by neighboring cells (39). Formed from deoxyguanosine (dG) in DNA by hydroxyl free radicals, 8-OHdG is one of the most reliable markers of oxidative DNA damage (40). The significant increase of 8-OHdG in the kidney tissues (Fig. 4) indicated that DNA injury occurred after colistin treatment. Correspondingly, levels of PARP-1, a marker for DNA repair, significantly increased in the kidney samples after 7-day treatment with 15 mg/kg/day colistin (Fig. 5D and I). Furthermore, colistin-induced apoptosis was confirmed by the marked increase of DNA fragmentation in the treated mouse kidney samples (Fig. 5K). Clearly, in consistency with recent studies (21, 23–26), our immunohistochemical and Western blotting data demonstrate that ROS and apoptosis play important roles in colistin-induced nephrotoxicity.

At least three major pathways can be activated by toxins, the mitochondrial pathway, death receptor pathway, and endoplasmic reticulum pathway (27), all of which are related to the activities of caspases (41). Mitochondria are the major ATP producer and a target of ROS, and the mitochondrial pathway can be triggered by exposure to drugs or toxins (39). In our recent study (28), colistin treatment caused an increase in the mitochondrial membrane permeability and decrease of Δψ_m and activities of the mitochondrial respiratory chain in mouse cortex and spinal cord; these data indicate that mitochondrial degeneration played an important role in colistin-induced apoptosis. Similar results were also reported in neurons in vivo (42) and in rat kidney proximal tubular NRK-52 cells (J. Li, 1st Internat. Conf. Polymyxins) due to colistin treatment. In the present study, transmission electron microscopy (TEM) images clearly show severely swollen mitochondria in the kidneys from the mice treated with 7.5 and 15 mg/kg/day colistin for 7 days (Fig. 3A, B, and D). Furthermore, mitochondrial dysfunction was also demonstrated by significant increase in Ca²⁺-induced mitochondrial membrane permeability and decrease in the activity of mitochondrial enzymes and ATP levels (Fig. 3F to H). Our results indicated that mitochondrial dysfunction was an important event in the colistin-induced nephrotoxicity in mice. The mitochondrial pathway can be regulated by the pro- and antiapoptotic Bcl-2 family proteins, including Bax and Bcl-2 (43). The downregulation of Bcl-2 and upregulation of Bax can induce mitochondrial outer membrane permeabilization (44), thereby causing release of cytC and ATP from mitochondria. Subsequently, cytC binds to apoptotic peptidase-activating factor 1 (APAF-1) and dATP and activates caspase-9- and caspase-3-dependent apoptosis (45). In addition, AIF is a product from the MPT formation and a mediator of the caspase-independent apoptotic pathway, which occurred in colistin-induced neuron apoptosis in vitro (28, 46). It has been reported that calpain 1, an accelerant of AIF and calcium-dependent cysteine protease, was involved in colistin-induced nephrotoxicity (24). In the current study, when mice were treated with colistin for 7 days, significant decrease of Bcl-2 and increase of cytC, AIF, cleaved caspase-9, and cleaved caspase-3 (Fig. 4A to D and Fig. 5A and F) confirmed that the activated mitochondrial pathway was important for colistin-induced nephrotoxicity. In addition, the significantly increased activities of capase-9 and -3 (Fig. 4I and J) were also observed in the kidney tissues treated with colistin. Collectively, our results revealed that both caspase-dependent and -independent pathways were involved in colistin-induced tubular cell apoptosis.

The death receptor pathway can be induced through the activation of death receptors such as Fas, which requires binding to the Fas ligand (FasL) (47). Subsequently, caspase-8 is activated through the Fas-associated death domain (FADD) (48). In the present study, 15-mg/kg/day colistin treatment led to significantly increased expression of Fas, FasL, and FADD and cleavage of the full-length caspase-8 in the kidneys (Fig. 5B and G). The activated caspase-8 cleaves Bid to tBid, which regulates the cross talk between the mitochondrial and death receptor pathways through negative regulation of Bcl-2 (49, 50). The activated caspase-8 can also cleave other caspases and induce the caspase cascade, thereby leading to cell death (49). In the kidney tissue samples from both colistin groups (7.5- and 15-mg/kg/day groups), the expression of tBid increased 8.8- and 40-fold, respectively (Fig. 5B and G). Our results demonstrated that colistin-induced renal tubular cell apoptosis involved the death receptor pathway and support the recent finding of colistin-induced activation of caspase-8 in rat renal tubular cells detected with the specific fluorescent inhibitor Red-IETD-FMK (J. Li, 1st Internat. Conf. Polymyxins).

We further examined the role of the endoplasmic reticulum in colistin-induced nephrotoxicity by measuring four key biomarkers. The endoplasmic reticulum pathway is initiated by endoplasmic reticulum stress due to a number of factors, including cytotoxicity, nutrient limitation, and accumulation of unfolded or misfolded proteins (51, 52). The anti-apoptotic Grp78/Bip is a master controller of endoplasmic reticulum stress response (53). Grp78/Bip interacts with two endoplasmic reticulum stress sensors, thereby activating transcription factor 6 (ATF6) and inositol-requiring enzyme 1 (IRE1); both play protective roles to promote cell survival (53, 54). However, under prolonged endoplasmic reticulum stress, the proapoptotic factor GADD153/CHOP is induced and activates caspase-12, which ultimately induces cell death through its downstream effector caspases, such as caspase-9 or -3 (39, 55). In the present study, after mice were treated with 15 mg colistin/kg/day for 7 days, the levels of Grp78/Bip, cleaved ATF6, GADD153/CHOP, and caspase-12 significantly increased (Fig. 5C and H). It is evident that the endoplasmic reticulum pathway is involved in colistin-induced nephrotoxicity; however, it is not clear whether activation of this pathway is a primary or secondary effect of colistin treatment.

For drug-induced apoptosis in the kidney, activation and phosphorylation of a number of modulators (e.g., MAPKs) and transcription factors (e.g., NF-κB and the tumor suppressor protein p53) are important (39). MAPKs are upstream modulators of apoptosis, and at least three subfamilies have been identified: extracellular signal-regulated kinases (ERKs), c-Jun N-terminal kinases (JNKs), and p38-MAPKs (56). Activation of p38 and NF-κB can contribute to the accumulation of p53, a key player in promoting growth arrest, apoptosis, and cellular senescence (57, 58). Particularly, these apoptosis modulators can be activated by ROS (59). In the present study, NF-κB/p65, p-JNK, p-p38, and p53 significantly accumulated in the mouse kidney tissues treated with colistin (15 mg/kg/day) for 7 days (Fig. 4F and Fig. 5E and I). This accumulation may well be related to the oxidative stress during which p53 plays a key role (59). A very recent transcriptomics study reported that colistin-induced nephrotoxicity in mice involved a number of genes in the cell cycle and p53 signaling pathway (25). The p21 protein (also known as CDKN1a) is a cyclin-dependent kinase (CDK) inhibitor (60). The expression of the p21 gene is strictly controlled by p53, and the p21 protein functions as a regulator of p53-dependent cell cycle progression at G₁ and S phase in response to stress stimuli (61). Interestingly, the significantly increased expression of p21 (up to 86.7-fold) was revealed in the kidney tissues from both groups of colistin-treated mice (Fig. 5I); our result is supported by the gene expression data in the transcriptomics study (25). In addition, the expression of CDK2, an important enzyme in maintaining and regulating cell cycle kinetics, was also examined in the present study. CDK2 did not accumulate in the 7.5-mg/kg colistin group (Fig. 5I), similar to the gene expression result in the mice treated with intrapeneocal colistin (16 mg/kg/day) for 3 days (25). However, the treatment of 15 mg/kg/day colistin for 7 days led to significantly increased levels of CDK2 in the kidney tissues (Fig. 5I). Taken together, our results support the recent finding that cell cycle arrest is involved in colistin-induced nephrotoxicity in mice and indicate that colistin treatment activated the p53-p21-CDK2 pathway.

As a self-degradative process, autophagy can be induced by ROS or nutrient stress and is important for balancing sources of energy during development (62). In contrast to apoptosis, autophagy is important in maintaining cell survival and plays a protective role in removing misfolded or aggregated proteins and clearing damaged organelles (e.g., mitochondria, endoplasmic reticulum, and peroxisomes) (62). The cross talk between autophagy and apoptosis is very complex, but important for cell survival and death (63). Autophagy plays a protective role in the acute renal injury induced by cisplatin and cyclosporine (64, 65). In our present study, macroautophagy was observed, and the number of autophagolysosomes significantly increased when the mice were treated with colistin (7.5 mg/kg/day) for 7 days (Fig. 3B and C). Interestingly, fewer autophagolysosomes were detected in the samples from the 15-mg/kg/day colistin group (Fig. 3D). In support of this finding, the levels of the autophagy biomarkers LC3B and Beclin-1 in the kidney samples significantly increased in the 7.5-mg/kg/day group, while they decreased in the 15-mg/kg/day group (Fig. 4G and H and Fig. 5J, K, and L). Of note, this trend is in contrast to the apoptosis results (Fig. 4 and 5). Our study is the first to reveal that autophagy may play a protective role in colistin-induced apoptosis in mouse renal tubular cells.

On the basis of our findings and the few studies in the literature, we propose the schematic diagram of colistin-induced apoptosis in renal kidney tubular cells (Fig. 6). An endocytic receptor megalin (MR) (66), polypeptide transporter 1 (PEPT1), and organic cation transporter 1 (OCTN1) (67) may mediate the substantial uptake of colistin by renal tubular cells. It is notable that oxidative stress and apoptosis play major roles in colistin-induced nephrotoxicity, while autophagy may protect renal tubular cells against the damage caused by colistin. The death receptor pathway seems to be the most sensitive pathway, and its activation may trigger the mitochondrial pathway by activation of tBid and cause ROS-mediated injury of mitochondria in renal tubular cells. In addition, it was revealed that several apoptosis modulators, JNK, p21, p38, NF-κB, and p53, were involved in colistin-induced nephrotoxicity.

FIG 6.

Schematic diagram of the proposed mechanisms of colistin-induced apoptosis. Cell cycle arrest is shown according to the present study and a recent report (25). ER, endoplasmic reticulum.

In conclusion, this study is the first to demonstrate that all three major apoptosis pathways (i.e., the mitochondrial, death receptor, and endoplasmic reticulum pathways) and autophagy are involved in colistin-induced nephrotoxicity in mice. Further studies are warranted to elucidate the complex signaling pathways of colistin-induced nephrotoxicity and their interplay. Understanding the mechanism(s) may provide novel approaches to attenuate colistin-induced nephrotoxicity, thereby widening its therapeutic window. Furthermore, such mechanistic information is urgently required for discovery of safer, novel polymyxins for treatment of life-threatening infections caused by Gram-negative “superbugs.”