Bioenergetic maladaptation and release of HMGB1 in calcineurin inhibitor-mediated nephrotoxicity

Abstract

Calcineurin inhibitors (CNIs) are potent immunosuppressive agents, universally used following solid organ transplantation to prevent rejection. Although effective, the long-term use of CNIs is associated with nephrotoxicity. The etiology of this adverse effect is complex, and effective therapeutic interventions remain to be determined. Using a combination of in vitro techniques and a mouse model of CNI-mediated nephrotoxicity, we found that the CNIs, cyclosporine A (CsA), and tacrolimus (TAC) share a similar mechanism of tubular epithelial kidney cell injury, including mitochondrial dysfunction and release of High-Mobility Group Box I (HMGB1). CNIs promote bioenergetic reprogramming due to mitochondrial dysfunction and a shift toward glycolytic metabolism. These events were accompanied by diminished cell-to-cell adhesion, loss of the epithelial cell phenotype, and release of HMGB1. Notably, Erk1/2 inhibitors effectively diminished HMGB1 release, and similar inhibitor was observed on inclusion of pan-caspase inhibitor zVAD-FMK. In vivo, while CNIs activate tissue proremodeling signaling pathways, MAPK/Erk1/2 inhibitor prevented nephrotoxicity, including diminished HMGB1 release from kidney epithelial cells and accumulation in urine. In summary, HMGB1 is an early indicator and marker of progressive nephrotoxicity induced by CNIs. We suggest that proremodeling signaling pathway and loss of mitochondrial redox/bioenergetics homeostasis are crucial therapeutic targets to ameliorate CNI-mediated nephrotoxicity.

1 |. INTRODUCTION

The vast majority of recipients for organ transplants, including the kidney, receive maintenance immunosuppressive therapy that includes a calcineurin inhibitor (CNI). The inclusion of CNI has resulted in substantially infrequent early acute rejection episodes and outstanding short-term graft survival.¹ However, long-term use is associated with nephrotoxicity and late allograft failure.² As such, interventions to ameliorate long-term morbidity and mortality are a major unmet need in clinical transplantation.

CNI-mediated nephrotoxicity is a complex, multifactorial, and progressive event associated with underlying inflammation and epithelial injury, hypertension, and graft dysfunction.³ Nephropathy due to chronic CsA or TAC treatment is linked to hyalinization of the afferent arterioles, tubular apoptosis and atrophy, endoplasmic reticulum stress, and eventually interstitial fibrosis.^4–7 Although sequential pathological progression is an established cause of kidney failure, the precise molecular mechanisms involved in CNI-mediated nephropathy are not determined. In addition, a lack of selective markers to identify optimal immunosuppression avoiding CsA or TAC-mediated toxicity is a major obstacle in long-lasting graft survival. Moreover, it is also uncertain if TAC is less nephrotoxic based on the long-term outcome studies.⁸

High-Mobility Group Box 1 (HMGB1) protein is a prototype of DAMPs, initially described as a non-histone DNA-binding protein involved in transcriptional regulation of nuclear gene expression. Previous studies have shown significant impact of extracellular HMGB1 in perpetuating inflammatory conditions and organ injury.^9–13 Ischemia and inflammatory injury are associated with alterations in bioenergetic and redox homeostasis, followed by release of alarmins and DAMPs from activated immune and injured epithelial cells.^14–17 HMGB1 has a limited ability to initiate inflammation independently, but it acquires pro-inflammatory capacity by binding to inflammatory components, including lipopolysaccharide (LPS), interleukin-1 (IL-1), ribonucleic and deoxyribonucleic acid, or chemokine CXL12.¹⁸ Notably, HMGB1 has been shown to reduce phagocytic clearance of apoptotic cells, thus perpetuating the release of harmful inflammatory mediators from dying cells.^19,20 The importance of DMAPs as diagnostic tools or therapeutic targets in CNI-mediated kidney injury is not known. We hypothesized that CNI-mediated nephrotoxicity is linked to adverse mitochondrial redox alterations and bioenergetic maladaptation that is followed by tissue remodeling and release of harmful mediators, including HMGB1. Herein, we describe the signaling events associated with HMGB1 release from CNI-injured proximal tubular kidney cells, and further test MAPK/Erk1/2 inhibitor as a pharmacological intervention to diminish nephrotoxicity and HMGB1 flux in mice subjected to prolonged administration of CsA or TAC.

2 |. MATERIALS AND METHODS

A detailed description of Materials and Methods is provided in the Supporting Information section at the end of the article.

2.1 |. Mice

C57BL/6 mice were acquired from Jackson Laboratory at 8 weeks of age and kept in the pathogen-free facility. The protocol was approved by the Animal Care and Use Committee (UAB).

2.2 |. Murine model for CsA or TAC-induced nephrotoxicity

We used our previously described model of CsA or TAC-induced nephrotoxicity in mice.^21,22 Mice were maintained on a low-salt diet TD.90228 from Envigo for 7 days prior and throughout the injection with CsA (60 mg/kg; i.p.) or TAC (1 mg/kg; i.p.) or vehicle (saline/DMSO [10%]), each day for total of 2 to 4 weeks. Serum creatinine was measured by LC-MS/MS at O’Brien Center (UAB), as previously described.²³ Mice were placed in metabolic cages obtained from Thermo Fisher Scientific, and urine was collected for 16 hours.

2.3 |. Imaging cells and kidney sections

Cells were plated in 12-well plates with glass coverslips and cultured in 0.5% serum media for 24 h prior to CsA treatment. Cells then were incubated with paraformaldehyde (4%), washed with PBS, and permeabilized with Triton-X 100 (0.3%) for additional 4 minutes, at room temperature. After blocking with BSA (1%) for 60 minutes at room temperature, cells were incubated with HMGB1 antibody overnight, at 4°C, followed by incubation with Alexa Fluor 488 for 90 minutes, at room temperature. Samples then were washed and coverslips were placed on slides in mounting solution that contains DAPI. Images were acquired from randomly selected fields using Leica fluorescence microscope.

For immunohistochemistry, paraffin-embedded 5-µm kidney sections were deparaffinized in serial solutions of Citrisolve from Fisher Scientific, isopropyl alcohol, and water, followed by antigen recovery using citric acid (pH 6.0). Sections were incubated overnight with HMGB1 antibody, followed by wash and application of TSA Plus FITC detection kit from Perkin-Elmer, according with the manufacturer’s protocol. Sections were stained with DAPI and cover slipped with PBS:glycerol (1:1). Images were acquired using a Leica DM6000 epifluorescence microscope (Leica Microsystems) outfitted with Hamamatsu ORCA ER cooled CCD camera and SimplePCI software from Compix, Inc. To measure tubular injury, kidney sections were stained with Periodic Acid Schiff (PAS) reagent at the VA Research Tissue Processing Laboratory (Veterans Affairs Medical Center). Images (5–10) 40× of sections were obtained from 3 to 5 indicated groups of mice. Morphological alterations were determined in regard to the appearance of vacuolized tubules, tubular atrophy, dilatation, and necrosis. Injury score was determined as percent of injured versus total tubules.

2.4 |. Statistical analysis

Multigroup comparisons were performed using one-way ANOVA with Tukey’s post hoc test. Values were normally distributed. For comparisons between two groups, statistical significance was determined using the Student’s t-test. Analyses and graphs were performed with Microsoft Excel and Prism (GraphPad; version 8.4.2). A value of p < .05 was consider significant.

3 |. RESULTS

3.1 |. CNI-mediated toxicity is associated with the release of DAMPs and loss of the epithelial proximal tubular epithelial cell phenotype

To investigate the early events associated with CNI toxicity, human proximal tubular epithelial cells (hPTECs) were incubated with CsA or TAC in a dose- and time-dependent manner. First, the extent of HMGB1 nucleus-to-cytosol translocation and its extracellular release was investigated after exposure of hPTEC to CsA. In unaltered cells, HMGB1 has predominant nuclear localization, whereas CsA induced a substantial HMGB1 accumulation to the cytosol, as evidenced by immunofluorescence staining of hPTEC (Figure 1A). This event is accompanied by accumulation of HMGB1 in culture media (Figure 1B,C). Subsequently, we found that CsA also promotes HMGB1 from human kidney 2 cells (HK-2) and murine cells, including inner medullary collecting ducts (IMCD3) and proximal tubular epithelial cell (mPTEC) (Figure S1). Next, we demonstrate that TAC had a similar impact by promoting HMGB1 nucleus-to-cytosol translocation and extracellular release from hPTEC (Figure 1D–F).

FIGURE 1.

CsA induces HMGB1 nucleus-to-cytosol translocation and extracellular release from hPTEC. (A) Representative images showing fluorescence patterns of HMGB1 in hPTEC after treatment with CsA (0, 1, 3, or 6 μM) for 24 h. HMGB1—green, nuclei—blue. Arrows indicate increased HMGB1 fluorescence in cytosol. Scale bars 25 µm. (B, C) Representative western blots and optical densitometry indicate HMGB1 levels in culture media from hPTEC exposed to (B) CsA (at the indicated concentrations) for 24 h, or (C) CsA (6 μM) in a time-dependent manner. Data are presented as scatter plot, mean ± SEM, n = 3, *p < .05 (ANOVA). (D) Representative images showing HMGB1 immunofluorescence staining in control hPTEC and after treatment with TAC for 24 h. Scale bars 25 µm. (E, F) HMGB1 levels in culture media of hPTEC that were treated with (E) TAC (dose response) for 24 h, or (F) TAC (30 µM) for indicated time. Immunoblot optical densitometry, mean ± SEM, n = 3, *p < .05 (ANOVA). (G, H) The amounts of total PARP, cleaved PARP and β-actin in lysates from hPTEC. Cells were treated with (G) CsA (0, 6, or 12 µM) or (H) TAC (0, 30, or 60 µM) for 24 h. Data presented as mean ± SEM, n = 3, *p < .05 (ANOVA)

3.2 |. CNI-induced apoptosis is associated with HMGB1 release

Although we have shown that nonapoptotic concentrations of CsA and TAC can trigger HMGB1 release (Figure S2A–C), such a release was even more pronounced on exposure to proapoptotic concentration of CNIs (Figure S2C–H). This event was significantly reduced by pretreatment with pan-caspase inhibitor zVAD-FMK, that is, prior exposure cells to CsA (12 µM), TAC (60 µM), or CP (100 µM) for 24 h. Although the number of necrotic cells was modest, compared with apoptotic cells after treatment with CNIs for 24 h (Figure S2D,E), we also investigated if necroptosis was implicated in the HMGB1 release. However, inhibitor of necroptosis Nec-1 did not mitigate CNI-induced HMGB1 release. Indeed, inclusion of Nec-1 potentiated HMGB1 release from CsA-treated hPTECs (Figure S2G,H). Given that zVAD-FMK at least partially prevented HMGB1 release (Figure S2D,E), but not Nec-1, suggests that apoptosis played a crucial role in active mechanism of HMGB1 release. Similar to CNIs, cisplatin effectively triggered HMGB1 release, indicating that HMGB1 release is not a selective effect mediated by CNIs, but can be triggered by others insults, including anti-cancer drug cisplatin.

3.3 |. CNIs affect the kidney epithelial phenotype

Additional analysis indicates that nonapoptotic concentrations of CNIs and HMGB1 release are associated with substantial alterations in cell morphology, including appearance of elongated shape, development of lamellipodia, and loss of intercellular adhesion (Figure S3A). These morphological changes are linked to reduced amounts of E-Cadherin and Claudin-2, the markers of epithelial phenotype and key proteins involved in stabilization of cell-to-cell adhesion via a tight junction mechanism (Figure S3B–D). Of note, the total amounts of intracellular HMGB1 were modestly decreased by exposure to CsA or TAC (Figure S3B,C).

3.4 |. CNIs induce mitochondrial dysfunction and bioenergetic reprograming of human proximal tubular epithelial cell

HMGB1 release and morphological alterations of kidney cells can associate with oxidative stress. H₂DCF-DA, a fluorogenic and redox sensitive probe,²⁴ was used to determine the impact of CNIs on ROS production in primary hPTEC. For example, HMGB1 translocation after exposure to CsA is associated with nearly threefold increase of DCF fluorescence, compared with control (Figure 2A–C). ROS formation involves mitochondria, as evaluated by the application of Mito-SOX, which is a superoxide-specific and mitochondria-targeted probe (Figure 2D). Additional experiments provided confirmatory evidence that ROS formation altered mitochondrial dynamics, and, in particular, fragmentation of mitochondrial network after exposure to CsA (Figure 2E).

FIGURE 2.

HMGB1 nucleus-to-cytosol translocation is associated with mitochondrial fragmentation, ROS formation, and loss cell-to-cell adhesion in CsA-treated hPTEC. (A) Representative images showing HMGB1 fluorescence patterns in control (vehicle) or hPTEC treated with CsA (6 µM) for 24 h. HMGB1—green; nuclei—blue. Scale bars 25 µm. (B, C) Representative images (B) and quantitative analysis of (C) DCF fluorescence in hPTEC treated with or without CsA (6 µM) for 24 h. DCF—green; nuclei—blue. Scale bars 100 µm. Data are presented as fold control of DCF/nuclei fluorescence ratios. Box and whiskers plot (min/max), n = 6–9, *p < .05 (Student’s t-test). (D) Mito-SOX fluorescence intensity in hPTEC treated with CsA (0, 3, or 6 µM) for 24 h. (E) Images depict mitochondrial network in control and CsA-treated hPTEC (6 µM) for 24 h. Dashed boxes indicate regions that are enlarged and displayed on the right side. Mitotracker—red; nuclei—blue. Scale bars 5 µm

CsA and TAC have a profound impact on mitochondrial electron transport chain (ETC) composition. Significant decreases are found in subunit NDUFB8 (complex I), FeS complex I (complex II), and subunit I of complex IV, as evidenced by quantitative analysis from cells treated with CsA (6 µM) or TAC (30 µM) for 48 h (Figure 3A,B). Such reduced levels of ETC subunits are accompanied with diminished basal, maximal, and ATP-linked OCR, and reduced mitochondrial reserve capacity, as indicated in CsA-treated PTEC (Figure 3C). CsA, in a dose-dependent fashion, decreased the activity of ETC complexes I, II, and IV (Figure 3D) (Figure S4A,B). The mitochondrial bioenergetic decline is associated with metabolic reprogramming. While oxidative phosphorylation (OXPHOS) is diminished, we observed the prosurvival adaptation due to a shift toward glycolytic metabolism (Figure 3E,F). Of note, after treatment with CsA for 24 h, withdraw of CsA for an additional 72 h resulted in the recovery of mitochondrial complexes I NDUFB8 subunit (Figure S4C,D).

FIGURE 3.

CsA-mediated mitochondrial dysfunction and a shift toward glycolytic metabolism in PTEC. (A, B) Representative western blots and quantitative analysis of the major mitochondrial ETC subunits from mPTEC-treated with or without (A) CsA (6 µM), or (B) TAC (indicated concentrations) for 48 h. Quantitative analysis is from CsA (6 µM) or TAC (30 µM)-treated cells for 48 h. Scatter plot; mean ± SEM, fold control, n = 3, *p < .05 (ANOVA). (C-E) The effects of CsA (6 µM, 48 h) on (C) oxygen consumption rates (OCR), including basal, maximal, ATP-linked, reserve capacity, proton leak, and nonmitochondrial OCR; (D) mitochondrial ETC complex I, II and IV activity; and (E) extracellular acidification rates (ECAR). Mean ± SEM, n = 3–4, *p < .05 (ANOVA). (F) OCR and ECAR analysis of CsA (0–9 µM)-treated mPTEC for 48 h. Mean ± SEM, n = 5 per indicated groups. Red arrows indicate a shift toward glycolytic metabolism

3.5 |. CNI-mediated proremodeling signaling promotes HMGB1 release from human proximal tubular epithelial cell

CNI treatment is associated with the activation of pro-remodeling signaling pathways.^25,26 This effect is consistent with the ability of CsA to activate extracellular kinase 1 and 2 (Erk1/2) and protein kinase B (Akt) in hPTEC (Figure S5A,B). A transient phosphorylation of Erk1/2 and Akt occur prior to HMGB1 release. However, Erk1/2 inhibitor (PD0325901), but not Akt inhibitor (LY294002), was sufficient to diminish HMGB1 nuclear-to-cytosol translocation and extracellular release following CsA or TAC treatment (Figure 4A–C). Although Akt-dependent phosphorylation of HMGB1 has previously been implicated in its extracellular release, our studies indicate that ERK1/2 activation, but not AKT, affects HMGB1 flux from hPTEC after exposure to CNIs.

FIGURE 4.

CsA and TAC stimulate Erk1/2-dependent HMGB1 nucleus-to-cytosol translocation and extracellular release from PTEC. (A, B) Representative western blots and optical densitometry showing HMGB1 levels in culture media, while total and phospho Thr202/Tyr204-Erk, as well as GAPDH were determined in cell lysates. Cells were treated with Erk1/2 inhibitor PD0325901 (0 or 10 nM) for 60 minutes prior to inclusion of (A) CsA (0 or 6 μM) or (B) TAC (0 or 30 µM) for additional 60 minutes. Data presented as scatter plot, mean ± SEM, n = 3, *p < .05 (ANOVA). (C) Images show HMGB1 fluorescence patterns in hPTEC. Cells were incubated with CsA or TAC and inhibitors as depicted in A and B, respectively. Arrows indicate HMGB1 immunofluorescence in cytosol. HMGB1—green; nuclei—blue. Scale bars 25 μm

It is possible that along with ERK1/2 activation, other upstream mediators associated with nephrotoxicity can affect HMGB1 release in vivo. Indeed, TNFα, IL-1β, and IFNγ effectively stimulated HMGB1 release from PTEC, as evidenced by western blot analysis and confocal microscopy (Figure 5). Interestingly, profibrogenic TGF-β1, which may be induced by CsA treatment, also triggers HMGB1 nucleus-to-cytosol translocation and accumulation in culture media (Figure 5A–D). These findings indicate that proinflammatory mediators promote HMGB1 flux but can be also triggered by TGF-β1.

FIGURE 5.

Pro-inflammatory and profibrogenic stimuli promote HMGB1 release from kidney epithelial cells. (A) Representative images show HMGB1 fluorescence patterns in control (vehicle) or mIMCD3 treated with TGFβ1 (10 ng/ml), TNFα (40 ng/ml), IFN-γ (40 ng/ml), or IL-1β (40 ng/ml) for 24 h. HMGB1—green; nuclei—blue. Scale bars 25 μm. (B-D) Representative western blots and optical densitometry of HMGB1 in culture media from (B) mIMDC3, (C) mPTEC, and (D) hPTEC. Cells were treated as depicted in (A). Data presented as scatter plot, mean ± SEM, n = 3–4, *p < .05 (ANOVA)

3.6 |. CNI-mediated nephrotoxicity is associated with HMGB1 release from kidney epithelial cells in the cortex and medulla

To assess the preclinical impact of our findings, mice were subjected to daily administration of CsA (60 mg/kg; i.p.) or TAC (1 mg/kg; i.p.) for a total of 2 or 4 weeks; as outlined in the methods and depicted in Figure 6A. Both CNIs induced significant renal injury, as measured by serum creatinine, although CsA was relatively more nephrotoxic compared with TAC. Baseline creatinine was similar prior to the treatment with CsA (0.093 ± 0.002 mg/dl) or TAC (0.085 ± 0.004 mg/dl). Following the treatment with CsA or TAC for 2 weeks, serum creatinine significantly increased (0.190 ± 0.011 mg/dl; p < .01 and 0.130 ± 0.004 mg/dl; p < .05, respectively) and remained significantly elevated after exposure for 4 weeks (CsA, 0.270 ± 0.030 mg/dl, p < .001; TAC, 0.150 ± 0.020 mg/dl, p < .05). These functional changes were accompanied by histological features of injury, as described in the Methods. Focal areas of injury were identified after 2 weeks of CsA or TAC treatment, with more severe injury observed within 4 weeks post-CNI exposure (Figure 6B). In particular, CsA- or TAC-mediated injury is characterized by tubular vacuolization, dilatation, and atrophy/necrosis, as confirmed by the morphological analysis, tubular injury score, and an increased serum creatinine (Figure 6C,D). Additionally, CsA-mediated nephrotoxicity was associated with marked expression of pro-inflammatory mediators within the kidney, including IL-6 and TNFα, as well as show alteration in redox signaling homeostasis, in particular, activation of NOX2, (phosphorylation of NADPH oxidase subunit p47phox) and HO1 expression (Heme oxygenase; phase II antioxidant response) compared with control mice (Figure S6). Prolonged treatment with CsA also increased S100A4, αSMA, and TGFβ1 expression, compared with control counterparts. Along with profibrogenic TGF-β1, CsA increased expression of Arg1, Itgam-CD11b macrophage markers (Supplemental Figure S6). Consistent with the effects of CNIs on mitochondrial dysfunction in murine PTEC, we also observed substantial decreases in the major components of mitochondrial ETC subunits in complexes CI, CII, CIII, and CIV in kidney homogenates of treated mice (Figure S7A). These events were associated with kidney tissue pro-remodeling, as evidenced by accumulation of β-catenin (Figure S7B).

FIGURE 6.

CNI-mediated nephrotoxicity is associated with HMGB1 release from kidney epithelial cells. (A) Treatment outline for CsA or TAC administration in mice. (B) Periodic acid Schiff (PAS) staining of kidney sections from control (vehicle) or mice treated with CsA (60 mg/kg; i.p.) or TAC (1 mg/kg; i.p.), daily for a total of 2 or 4 weeks. Scale bars 100 μm. Indices of injury include vacuolized tubules [v], tubular atrophy areas [a], and tubular dilatation [h]. (C, D) Indices of CsA- or TAC-mediated nephrotoxicity. Panel (C) depicts the tubular injury score (%), and (D) serum creatinine. Data are presented as box and whiskers plot, (min/max), n = 5 mice/group (morphometric analysis from 5 images/mouse), or n = 5 mice/group for serum creatinine, *p < .05 (ANOVA). (E) HMGB1 fluorescence in kidney cortex and medulla from control (vehicle) or mice treated with CsA or TAC for 4 weeks. Representative images are shown and white arrows indicate HMGB1 release. Scale bars 100 μm. HMGB1—green. (F, G) HMGB1 levels in urine from control (vehicle) and (F) CsA- or (G) TAC-treated mice for 2 and 4 weeks, as depicted in (A). Data are presented as box and whiskers plot (min/max), n = 4–5 mice urine, *p < .05 (ANOVA)

3.7 |. CNI-mediated kidney injury is associated with HMGB1 release

CNI-mediated nephrotoxicity is associated with a pronounced HMGB1 nucleus-to-cytosol translocation and release, as demonstrated in the cortex and medulla of CsA- or TAC-treated mice for 4 weeks (Figure 6E). Immunofluorescence images show a more dispersed pattern of HMGB1 localization and a substantial number of HMGB1-negative nuclei are observed in the cortex and medulla of CsA- or TAC-treated mice, while HMGB1 had a predominant nuclear localization in control (Figure 6E). The extracellular release of HMGB1 was significant, as confirmed by western blot analysis of HMGB1 in the urine collected from mice that received CNIs for 2 and 4 weeks (Figure 6F,G). These results suggest that accumulation of HMGB1 in urine is an early and consistent indicator of CNI-induced nephrotoxicity.

3.8 |. Inhibition of Erk1/2 mitigates CNI-mediated HMGB1 release and kidney injury

Given that Erk1/2 inhibitor (PD0325901) prevented HMGB1 release from CNI-treated PTEC ex vivo, we tested if this approach also affects HMGB1 release and associated kidney injury in a mouse model of CsA- or TAC-induced nephrotoxicity. To this end, mice received Erk1/2 inhibitor (PD 2.5 mg/kg; i.p.) or vehicle (saline/DMSO [10%]) at the same time as CsA (60 mg/kg; i.p) or TAC (1 mg/kg; i.p). In particular, PD0325901 alone or in combination with CsA or TAC were injected daily for total of 2 weeks (outline Figure 7A). Inclusion of PD0325901 with CsA or TAC significantly reduced severity of CNI-mediated kidney injury, as indicated by a reduction in tubular injury and improved serum creatinine, compared with mice treated with CNIs alone (Figure 7B,C). ERK1/2 inhibitor also diminished HMGB1 nucleus-to-cytosol translocation, as depicted in immunofluorescence images of kidney sections from mice treated with PD0325901/CsA or PD0325901/TAC, versus CNIs alone (Figure 7D). Moreover, PD0325901 reduced accumulation of HMGB1 in the urine, compared with mice that received only CsA or TAC (Figure 7E,F). Of note, PD0325901 itself had no effect on HMGB1 translocation or indices of kidney injury.

FIGURE 7.

Inhibition of Erk1/2 signaling diminished CNI-mediated nephrotoxicity in mice. (A) Therapeutic administration of Erk1/2 inhibitor (PD 2.5 mg/kg; i.p.) in mice subjected to CsA (60 mg/kg; i.p.) or TAC (1 mg/kg; i.p.), each day for 2 weeks. (B) Kidney PAS staining showing a reduced kidney injury in mice treated with Erk1/2 inhibitor combined with CsA or TAC. Vacuolized tubules [v], tubular atrophy areas [a], and tubular dilatation [h] are indicated. (C) Tubular injury score (%) and serum creatinine in indicated groups of mice. Box and whiskers plot (min/max), n = 5 mice/group for morphometric analysis using 5 images/mouse, or n = 5 mice/group for serum creatinine, *p < .05 (ANOVA). (D) HMGB1 and nuclei fluorescence patterns in kidney sections from indicated groups of mice. Dashed boxes indicate regions that are magnified and displayed in lower panels. Arrows depict HMGB1 nucleus-to-cytosol translocation in kidney epithelial cells. (E, F) HMGB1 levels in urine from control (vehicle) mice or subjected to CsA, TAC, and PD alone, as well as CsA or TAC combined treatment with PD. Western blots and optical densitometry are shown. Data presented as box and whiskers plot (min/max), n = 5, *p < .05 (ANOVA)

4 |. DISCUSSION

In this study, we have made a number of novel observations related to CNI-mediated nephrotoxicity. To our knowledge, this is the first comprehensive analysis of CNI-mediated nephrotoxicity, linking CNI-induced mitochondrial dysfunction, oxidative stress, and metabolic maladaptation that did not compromise cell viability but is followed by the loss of the epithelial phenotype. Notably, nonapoptotic or early apoptotic events induced by a relatively high concentration of CNIs were both linked to the active mechanism of HMGB1 release (Figure 8). Along with studies in vivo, our findings suggest that HMGB1 flux has the diagnostic potential for CNI-mediated nephrotoxicity.

FIGURE 8.

Proposed paradigm for CNI-mediated nephrotoxicity. Under nonapoptotic conditions, CNI-induced mitochondrial dysfunction, oxidative stress, and metabolic maladaptation, that while not compromising cell viability, result in loss of epithelial phenotype and HMGB1 release is mediated by ERK signaling. At apoptotic conditions, HMGB1 is actively released

We also identified therapeutic targets related to mitochondrial dysfunction and MAPK/Erk1/2 signaling after prolonged exposure to either CsA or TAC. Although mice treated with CsA demonstrated a relatively more severe nephrotoxicity compared with those treated with TAC (as selected concentrations of CsA vs. TAC were administered), our studies revealed the epithelial injury mediated by CsA and TAC share similar mechanisms of impaired mitochondrial bioenergetic and redox homeostasis in kidney epithelial cells, as well as in vivo. Notably, mitochondrial bioenergetic decline was associated with metabolic reprogramming toward glycolytic metabolism, alterations in the epithelial morphology, and loss the epithelial phenotype in CNI-treated PTEC. CNIs stimulate proremodeling signaling pathways, including MAPK/Erk1/2 and PI3 K/Akt, along with significant expression of inflammatory and profibrogenic mediators and DAMPs, in particular HMGB1. Importantly, inhibition of MAPK/Erk1/2 signaling pathway also diminished CNI-mediated nephrotoxicity and HMGB1 release from epithelial cells and its accumulation in urine. Additionally, mechanistic studies in vitro revealed that inhibition of ERK1/2 or pretreatment with inhibitor of apoptosis zVAD-FMK (related to high concentration of CNIs), effectively disrupted the active mechanisms of HMGB1 release from hPTECs.

Our studies showed that fragmentation of mitochondrial network (fission) and dissipation of major components in the mitochondrial ETC complexes were associated with ROS formation and decline of bioenergetic indices in CNI-treated PTEC. While previous studies have aligned oxidative stress with apoptosis induced by CsA in kidney tubular cell,²⁷ our approaches used a lower, but albeit nephrotoxic concentrations of CNIs without significant PTEC apoptosis. In this instance, mitochondrial dysfunction resulted in an activation of proremodeling mechanisms, including bioenergetic reprogramming toward glycolytic metabolism. It is important to note that while glycolytic flux is an initial compensatory response, a prolonged decline in the mitochondrial bioenergetics is likely insufficient to support normal epithelial function. For example, loss of mitochondrial bioenergetics has been implicated in chronic kidney dysfunction associated with diabetes, hypertensive and ischemic injury, glomerulonephritis, aging, and polycystic kidney disease.^14,28 Certainly, progressive decline in the mitochondrial bioenergetic and redox homeostasis have cumulative maladaptive effects, as frequently observed in transplant recipients with a prolonged administration of CsA or TAC.

Clinical studies and experimental models of organ injury have shown an adverse impact of DAMPs, including HMGB1.^20,29–31 Although the impact of extracellular HMGB1 is well established in many models of organ injury, deficiency of HMGB1 in the nucleus may also elicit epithelial dysfunction. For example, loss of HMGB1 in the nucleus has been linked to senescence,³² and aging is a known risk factor in the development of organ fibrosis; an established long-term complication linked to CNI-mediated nephrotoxicity.^33,34 The effects of extracellular DAMPs are complex, mostly due to their involvement in multiple pro-inflammatory pathways (e.g. IL-1, TLR signaling), chemotaxis, and exaggerated inflammation resulting from impaired clearance of dying cells and microbial pathogens.^20,35–37 Interestingly, we found that TGF-β1 effectively stimulates HMGB1 release from PTEC; this finding adds new mechanistic insights into the development of interstitial fibrosis, a relevant issue in CNI-mediated nephrotoxicity.³⁸ Although these are adverse effects of HMGB1, it is important to note that initial translocation of HMGB1 to cytosol can stimulate autophagy followed by a beneficial clearance of dysfunctional mitochondria.^39,40 Similar mechanisms can be operational during initial exposure to CNIs.

There has been a lasting debate whether CsA-mediated nephrotoxicity is more severe than TAC, given more recent clinical outcomes and contrasting biopsy studies.^41,42 It is difficult to provide a certain conclusion from CsA- and TAC-induced nephrotoxicity in mice, as this is dependent on dose and duration of treatments. Although our studies showed that CsA elicited more severe nephrotoxicity based on renal dysfunction and tubular injury, we also demonstrate that both CNIs have similar mechanisms of mitochondrial dysfunction, release of DAMPs and activation of proremodeling signaling. It would be important to establish if these indices are suitable diagnostic tools to determine clinical difference in managing CNI-mediated immunosuppression versus nephrotoxicity. Such approach can promote changes in clinical practice that use these drugs for over three decades.

Despite considerable progress in understanding mechanisms involved in CNI-mediated nephrotoxicity, therapeutic interventions have not been embraced, and substitutions for the therapy have been less than successful.^43,44 Strategies to neutralize extracellular HMGB1 have not been tested to reduce CNI-mediated nephrotoxicity. We have shown that inhibition of MAP/Erk1/2 or zVAD-FMK that diminished an early apoptosis triggered by high concentration of CNIs, effectively prevented HMGB1 release in CsA or TAC-treated kidney epithelial cells. Notably, preventive effects mediated by ERK1/2 and zVAD-FMK, but not necroptosis inhibitor of Nec-1, suggest that HMGB1 was released via active mechanism. However, it is important to note that pro-apoptotic concentrations of CNIs will certainly lead to necrosis, condition known to trigger a passive release of HMGB1 along with other DAMPs due to loss of cell membrane integrity. Nevertheless, our findings are consistent with previous effects of Erk1/2 inhibitors, mitochondria-targeted antioxidants, and anti-HMGB1 antibodies to reduce the severity of organ injury in several preclinical models.^31,45–47 Of note, activation of ERK signaling mediates deleterious impacts in other models. Examples include diabetic nephropathy and renal cell carcinoma, where activation of the receptor of advance glycation end-products (RAGE) signals via ERK, leading not only to proliferation but to inflammation and cancer invasion.^48,49

Regarding mitochondrial dysfunction, possible therapeutic interventions are linked to mitochondrial-targeted antioxidants, including ubiquinone (MitoQ), tocopherol (Mito-TEMPO), or SOD mimetic nitroxide (Mito-CP). These agents ameliorate injury in models of ischemia/reperfusion, sepsis, diabetes, or cisplatin-induced kidney injury.^50–53 Indeed, our results suggest that HMGB1 is released in parallel to mitochondrial ROS formation. Recent studies indicate that oxidation of HMGB1 contributed to HMGB1 translocation and extracellular flux,⁵⁴ although we did not test HMGB1 posttranslational modifications. These finding support a possibility that CNI-induced mitochondrial ROS can promote HMGB1 oxidation and release. Finally, it is possible that metabolic switches and mitochondrial biogenesis could restore bioenergetic/redox homeostasis of epithelial cells, thereby reduce an adverse impact of CNIs on kidney epithelial cells.

In summary, CNIs are currently critically effective immunosuppressant therapy for preservation of an allograft from immune-mediated injury, but they also induce nephrotoxicity in many transplant recipients. Thus, development of effective strategies to mitigate CNI-mediated mitochondrial dysfunction, remodeling, and HMGB1 release are crucial for long-term graft preservation. Of note, our results suggest that HMGB1 is a sensitive indicator that can be used to determine CNI-mediated immunosuppression versus nephrotoxicity.

Abbreviations:

CNIs

calcineurin inhibitors

CsA

cyclosporine A

ETC

electron transport chain

HK-2

human kidney 2 cells

HMGB1

High-Mobility Group Box I

hPTECs

human proximal tubular epithelial cells

IL-1

interleukin-1

IMCD3

inner medullary collecting ducts

LPS

lipopolysaccharide

mPTEC

proximal tubular epithelial cell

OXPHOS

oxidative phosphorylation

PAS

Periodic acid Schiff reagent

RAGEs

receptor of advance glycation end-products

TAC

tacrolimus

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author.