Disruption of podocyte cytoskeletal biomechanics by dasatinib leads to nephrotoxicity

Abstract

Nephrotoxicity is a critical adverse event that leads to discontinuation of kinase inhibitor (KI) treatment. Here we show, through meta-analyses of FDA Adverse Event Reporting System, that dasatinib is associated with high risk for glomerular toxicity that is uncoupled from hypertension, suggesting a direct link between dasatinib and podocytes. We further investigate the cellular effects of dasatinib and other comparable KIs with varying risks of nephrotoxicity. Dasatinib treated podocytes show significant changes in focal adhesions, actin cytoskeleton, and morphology that are not observed with other KIs. We use phosphoproteomics and kinome profiling to identify the molecular mechanisms of dasatinib-induced injury to the actin cytoskeleton, and atomic force microscopy to quantify impairment to cellular biomechanics. Furthermore, chronic administration of dasatinib in mice causes reversible glomerular dysfunction, loss of stress fibers, and foot process effacement. We conclude that dasatinib induces nephrotoxicity through altered podocyte actin cytoskeleton, leading to injurious cellular biomechanics.

Fig. 1

Mining of the FDA Adverse Reporting System (FAERS) database demonstrates that dasatinib is associated with high risk for nephrotoxicity. a Workflow for FAERS database mining. b Dasatinib ranked as one of the highest reporting odds ratio (ROR) for all nephropathies among other KIs in the FAERS database. c When the KI ROR values for glomerular specific nephropathies are plotted against those for hypertension, dasatinib is seen as a clear outlier with high risk for glomerular disorder and a relatively low risk for hypertension

Fig. 2

Podocyte morphometrics show that focal adhesion architecture and stress fiber formation are severely impacted by dasatinib. a Conditionally immortalized differentiated podocytes treated for one-hour with 2 µM kinase inhibitor were stained for F-actin (phalloidin), α-actinin-4, paxillin, and nuclei (Hoechst 33342) to assess cytoskeletal and focal adhesion architecture (scale bars = 50 µm). Scatter plots of b cell size, shape, and average actin intensity; c nuclear size, shape, and nuclear YAP localization; d stress fiber count per cell, stress fiber area, and F-actin + actinin-4 colocalization; e and focal adhesion number per cell and F-actin + actinin-4 colocalization at the focal adhesion sites in podocytes treated with KIs for 24 h. Central bullseye within the distributions highlight population medians, boxes highlight the middle quartiles. Sample sizes are n_CTRL = 2313; n_DAS = 491; n_IMA = 1011; n_NIL = 956; n_VAN = 788; n_ERL = 867; and n_BOS = 749 for cellular and nuclear plots, which were captured at 200× magnification. Focal adhesion and stress fiber analysis was performed with images captured at 400× magnification, which had the following sample sizes: n_CTRL = 425; n_DAS = 221; n_IMA = 183; n_NIL = 231; n_VAN = 220; n_ERL = 227; and n_BOS = 196. All experimental groups were taken from minimum of three wells of a 96-well plate (***p < 0.001, Kruskal–Wallis one-way ANOVA followed by Tukey post-hoc multiple comparison). f Heatmap representation of all of the analyzed morphometric parameters (normalized median value plotted; n = 60) showing that dasatinib is strongly separated from all the other KIs in terms of its effect on cellular, cytoskeletal and nuclear morphology

Fig. 3

Dasatinib does not have a unique effect on cell survival compared to other KIs. a Western blots for p-Src (Y416) show that both bosutinib and dasatinib at 2 µM strongly inhibit SRC (mean ± SEM, *p < 0.05, unpaired t-test). b Western blot (mean ± SEM; *p < 0.05, unpaired t-test) and c immunofluorescence-based high-content image analysis (median ± middle quartiles; ***p < 0.001, Kruskal-Wallis one-way ANOVA followed by Tukey post-hoc multiple comparison) both showed small increase in apoptosis, as quantified by levels of cleaved caspase 3 (CC 3); however there were no significant differences between dasatinib and other KIs. d Western blots for p-MAPK, a key kinase for cell survival, showed that dasatinib is not unique in its ability to block this pathway (mean ± SEM; *p < 0.05, unpaired t-test). e Dasatinib had a major effect on podocyte viability after 24 h of treatment, as assessed by four-parameter half-maximal effective concentration (EC50) fit of the MTT dose-response curve. Even though dasatinib had the lowest EC50, it was not significantly different from other KIs (geometric mean ± geometric SD, p = 0.169, Kruskal–Wallis one-way ANOVA). f Representative cell survival curves showing percent viability normalized to vehicle control

Fig. 4

Cytoskeletal effects of dasatinib occur at lower concentrations than those observed clinically. a Dose-response for dasatinib with respect to cell morphology was evaluated using high-content image analysis to quantify cell spreading area, extent and form factor. Top: podocytes at varying concentrations of dasatinib stained for F-actin, inverted grayscale lookup table (scale bar = 100 µm). b Cytoskeletal effects of dasatinib were quantified by segmenting high magnification (400×) images of F-actin and α-actinin-4, which showed significant effects at 6.1 nM for total number of stress fibers. c Number of focal adhesions were significantly reduced at 5.9 nM as quantified by segmentation of paxillin immunofluorescence images at 400×. d Changes in nuclear morphology mirrored changes in cell spreading area and were significantly altered at approximately 20 nM of dasatinib. Scale bars, b–d = 50 µm

Fig. 5

Phosphoproteomics reveal that processes associated with the actin cytoskeleton are specifically perturbed by dasatinib. a Heatmap for spectral counts of differentially expressed phospho-tyrosine enriched proteins as quantified by LC-MS/MS. Complete list of proteins plotted in the heatmap are shown in Supplementary Table 3. b Enrichment analyses using Wikipathways, KEGG Pathways and other ontology libraries coherently associate focal adhesions and the actin cytoskeleton with the differential phosphoproteome. Highest transcriptional enrichment was for TEAD1 suggesting potential inhibition of the Hippo pathway. c Predicted protein-protein interaction network using the differentially phosphorylated proteins as seed nodes (cyan) and one intermediate neighbor (yellow) is highly interconnected. Node size signifies connectivity. Complete list of network associated proteins and number of connections are listed in Supplementary Table 4. d KINOMEscan analysis reveals that the activity of actin-associated kinases are disproportionately impacted by dasatinib compared to all other tested KIs; in particular LIM kinase (LIMK1) is identified as a kinase that is inhibited exclusively by dasatinib. e Active phosphorylation of kinases from the Rac/Cdc42 pathway (PAK1/2, LIMK1, and cofilin), after treatment with 2 µM KIs for 1 h, show strong, exclusive and statistically significant inhibitory effect for dasatinib but none of the other tested KIs

Fig. 6

Dasatinib impacts podocyte biomechanics and induces foot process effacement in vivo. a Representative AFM indentation curves from elastography measurements of control and dasatinib treated podocytes that were used to quantify the single-cell biomechanical properties show good agreement between the Hertzian fits and the advance-phase of the raw indentation curves (insets: phase contrast images of representative cells with the superimposed blue line denoting the probed area; scale bar = 20 µm). b Significant reduction of apparent elastic modulus is observed in dasatinib-treated podocytes but not in other KI treatments (line: mean; **p < 0.01 dasatinib vs. all other groups, one-way ANOVA with post-hoc Tukey test). c Spatial AFM elastography maps highlight stiff linear projections that colocalize with the underlying F-actin stress fibers in control but not in dasatinib treated podocytes suggesting that existing fibers have lost their biomechanical integrity. d Representative TEM images of kidney sections for 129S1/SvImJ mice that have been treated daily with dasatinib for five weeks (n = 8, each group) show substantial FP effacement. Scale bars are 5000 and 500 nm for the 2000× and 20,000× images, respectively. e Quantification of FP width from TEM images (mean ± SD; ****p < 0.0001, unpaired t-test). f Immunofluorescence analysis shows that synaptopodin crosslinked actin fibers are reduced in dasatinib treated glomeruli, even though total number of podocytes is unchanged. Quantification of g synaptopodin (SYNPO) expression (mean ± SD; **p < 0.01, unpaired t-test) and h WT1-positive podocyte count per glomerulus as determined by WT1-positive cells show that F-actin crosslinking is significantly lower even though the number of podocytes do not change (mean ± SD; n.s. = not significant, unpaired t-test) in dasatinib treated animals. Scale bars = 10 µm. i 8-week old MRL/MpJ Lupus mice chronically treated with dasatinib show severe proteinuria within four weeks that recovers within a week upon suspension of treatment (mean ± SEM; ***p < 0.001, two-way ANOVA with Bonferroni multiple comparison)