Inwardly rectifying potassium channels mediate polymyxin-induced nephrotoxicity

Abstract

Polymyxin antibiotics are often used as a last-line defense to treat life-threatening Gram-negative pathogens. However, polymyxin-induced kidney toxicity is a dose-limiting factor of paramount importance and can lead to suboptimal treatment. To elucidate the mechanism and develop effective strategies to overcome polymyxin toxicity, we employed a whole-genome CRISPR screen in human kidney tubular HK-2 cells and identified 86 significant genes that upon knock-out rescued polymyxin-induced toxicity. Specifically, we discovered that knockout of the inwardly rectifying potassium channels Kir4.2 and Kir5.1 (encoded by KCNJ15 and KCNJ16, respectively) rescued polymyxin-induced toxicity in HK-2 cells. Furthermore, we found that polymyxins induced cell depolarization via Kir4.2 and Kir5.1 and a significant cellular uptake of polymyxins was evident. All-atom molecular dynamics simulations revealed that polymyxin B₁ spontaneously bound to Kir4.2, thereby increasing opening of the channel, resulting in a potassium influx, and changes of the membrane potential. Consistent with these findings, small molecule inhibitors (BaCl₂ and VU0134992) of Kir potassium channels reduced polymyxin-induced toxicity in cell culture and mouse explant kidney tissue. Our findings provide critical mechanistic information that will help attenuate polymyxin-induced nephrotoxicity in patients and facilitate the design of novel, safer polymyxins.

Keywords: CRISPR/Cas9 screening; Kir4.2; Kir5.1; Polymyxin nephrotoxicity.

Fig. 1

Identification of significant genes mediating polymyxin-induced toxicity in HK-2 cells by CRISPR-Cas9 knockout screen. A Viability of HK-2 cells following 24-h treatment with 0–100 µM polymyxin B (n = 6). B Experimental scheme for CRISPR-Cas9 knockout screen. C Volcano plot showing positively selected sgRNAs (red dots in green background, p < 0.05) following polymyxin B treatment. The genes discussed in the main text are highlighted in bold and red color. D Significantly enriched Reactome pathways (p < 0.05). E Viability of independent gene knockout cells after polymyxin B treatment. Gene knockout cells were generated by CRISPR editing with guide RNAs. Two sgRNAs were chosen for each gene and are labelled with ‘_1’ and ‘_2’. Viability of gene knockout cells following 25 µM polymyxin B treatment for 24 h were measured with XTT assay (n = 3). Two-tailed Student’s t-test was used to compare each of the gene knockout groups with the empty vector control. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Fig. 2

Differential gene expression in HK-2 cells following polymyxin B treatment and pathway enrichment results. A Volcano plot showing differentially expressed genes (red) with FDR < 0.05 and FC ≥ 1.5. KCNJ15, KCTD5, and ERBB3 were CRISPR identified genes and are labelled in black. B Significantly enriched pathways with differentially expressed genes. FPKM: fragments per kilobase of exon per million reads mapped. Pathway names: A, Ion channel; B, Clathrin-dependent endocytosis; C, Apoptosis; D, mTOR; E, SLC transporter family

Fig. 3

Knockout or inhibition of Kir4.2 and Kir5.1 prevented polymyxin-induced toxicity in HK-2 cells. A Western blot showing the expression levels of KCNJ15 and KCNJ16 after knockout; actin was used as an internal control. B Viability of wild-type HK-2, KCNJ15 KO and KCNJ16 KO cells following 24-h exposure to 10 and 25 µM polymyxin B (n = 6). C Viability of HK-2 cells following 24-h exposure to 0–100 µM BaCl₂ with or without 25 µM polymyxin B (n = 5). D Viability of HK-2 cells following the treatment of 0–25 µM VU0134992 alone or in combination with 25 µM polymyxin B for 24 h (n = 3 for controls, and n = 4 for treatment groups). E Morphologies of wild-type, KNCJ15 KO, and KCNJ16 KO HK-2 cells with the treatment of 25 µM polymyxin B or polymyxin B with the combination of 50 µM BaCl₂, or 5 µM VU0134992 to wild-type cells. Two-way ANOVA was employed for multi-group comparisons and Tukey's multiple comparison test was employed for post-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Fig. 4

Polymyxin B induced significant electrophysiological changes and membrane depolarization in HK-2 cells. A The resting membrane potential in wild-type, KCNJ15 KO and KCNJ16 KO cells (n = 34, 18 and 10, respectively). B Input resistances in wild-type, KCNJ15 KO and KCNJ16 KO cells (n = 20, 18 and 10, respectively). C In current clamp mode, polymyxin B induced approximately 30 mV depolarization in WT cells and this was reversible. Depolarization was not induced in KCNJ15 KO cells. Polymyxin B induced membrane potential changes are shown aside (n = 9, 10 and 7, respectively). D In voltage clamp mode, polymyxin B induced a statistically significant inward current (green) in wild-type HK-2 cells (n = 8), but not in KCNJ15 KO cells (n = 8). The current and reversal potential values are shown aside. E Fluorescent signal detection in HK-2 cells with DiBAC, 25 μM polymyxin B, and DiBAC plus 25 μM polymyxin B. F Proportions of DiBAC-positive in wild-type, KCNJ15 KO, and KCNJ16 KO HK-2 cells measured by flow cytometry. G Proportions of DiBAC-positive HK-2 cells in the control and BaCl₂ (10 μM) groups with or without 25 μM polymyxin B treatment measured by flow cytometry (n = 5 for WT and n = 4 for KOs). Data are shown as box and whisker plots. One-way (for WT) or two-way (for KOs) ANOVA was employed for multi-group comparisons. **p < 0.01; ****p < 0.0001

Fig. 5

Molecular models of Kir4.2 channel with polymyxin B₁. A The Kir4.2 channel model is shown in NewCartoon presentation (subunits are in yellow, blue, grey and purple). B Binding of polymyxin B₁ to the Kir4.2 channel. Polymyxin B₁ are shown in green, and the binding amino acids are in red. C Gate distance of the channel in the inactivate state (orange) and the state after polymyxin B₁ bound. Purple balls represent K⁺ atoms. Independent simulations were conducted three times

Fig. 6

Intracellular accumulation of polymyxin B in wild-type, KCNJ15 KO and KCNJ16 KO HK-2 cells with the treatment of 25 µM polymyxin B and 50 µM BaCl₂ for 6 h. A Polymyxin B was immunostained with polymyxin antibody and visualized using Alexa Fluor-594 dye (red). The nucleus was counterstained with DAPI (blue). B The plots showing mean fluorescence intensities from each group. The value from control group has been deducted and the mean value from each replicate was plotted (n = 4)

Fig. 7

Polymyxin-induced toxicity in mouse kidney explant cultures with or without Kir inhibitors. A TUNEL staining (magenta) of explanted kidneys labelled with LTL (proximal tubules, cyan) and DAPI (nuclei, blue) after treatments with 50 μM polymyxin B (PMB), 5 μM VU0134992 (VU) and 50 μM BaCl₂ alone or in combination. Scale bar = 30 μm. Dashed boxes in the top image panel indicate magnified proximal tubule regions shown below. TUNEL⁺ LTL⁺ cells are marked with arrowheads. B Quantification of polymyxin-induced apoptotic cells relative to the total number of cells in each sample (n = 4). ****p < 0.0001. C Assessment of relative levels of polymyxin-induced apoptosis in tubules with or without polymyxin treatment (n = 4). ***p < 0.001. D Expression levels of Kcnj15 and Kcnj16 in mouse developmental kidneys on embryonic day E15.5 and E18.5. Average gene expression levels (blue-red) and the percentage of cells within a cluster that expressed the gene (circle size) are displayed according to the legends