MDM2 prevents spontaneous tubular epithelial cell death and acute kidney injury

Murine double minute-2 (MDM2) is an E3-ubiquitin ligase and the main negative regulator of tumor suppressor gene p53. MDM2 has also a non-redundant function as a modulator of NF-kB signaling. As such it promotes proliferation and inflammation. MDM2 is highly expressed in the unchallenged tubular epithelial cells and we hypothesized that MDM2 is necessary for their survival and homeostasis. MDM2 knockdown by siRNA or by genetic depletion resulted in demise of tubular cells in vitro. This phenotype was completely rescued by concomitant knockdown of p53, thus suggesting p53 dependency. In vivo experiments in the zebrafish model demonstrated that the tubulus cells of the larvae undergo cell death after the knockdown of mdm2. Doxycycline-induced deletion of MDM2 in tubular cell-specific MDM2-knockout mice Pax8rtTa-cre; MDM2f/f caused acute kidney injury with increased plasma creatinine and blood urea nitrogen and sharp decline of glomerular filtration rate. Histological analysis showed massive swelling of renal tubular cells and later their loss and extensive tubular dilation, markedly in proximal tubules. Ultrastructural changes of tubular epithelial cells included swelling of the cytoplasm and mitochondria with the loss of cristae and their transformation in the vacuoles. The pathological phenotype of the tubular cell-specific MDM2-knockout mouse model was completely rescued by co-deletion of p53. Tubular epithelium compensates only partially for the cell loss caused by MDM2 depletion by proliferation of surviving tubular cells, with incomplete MDM2 deletion, but rather mesenchymal healing occurs. We conclude that MDM2 is a non-redundant survival factor for proximal tubular cells by protecting them from spontaneous p53 overexpression-related cell death.

Renal tubular epithelial cells are continuously exposed to stress due to the hypoxia, hyperosmolarity and toxins exposure and it is rather remarkable that they can withstand those insults and still execute their physiological functions that is, water and solutes reabsorption and excretion. Acute exposures can lead to acute tubular necrosis underlying the clinical syndrome of acute kidney injury. In unchallenged kidneys, tubular epithelial cells divide at a very low rate. This minimal production of new cells supplies though enough tubular cells to balance the loss of the tubular epithelial cells into urine and guarantees the physiological turnover of tubule cells. Nevertheless, this turnover rate must be strictly controlled as even a small disproportion between cell death and cell proliferation would eventually result in nephron loss or significant increase in nephron size. 1,2 In unstressed kidney remain the tubular cells in G0-G1, quiescent state. 3 The mechanisms and factors necessary for the tubule cells homeostasis are not fully understood.
E3-ubiquitin ligase murine double minute-2 (MDM2) is the master negative regulator of tumor suppressor gene p53 and a non-redundant modulator of NF-ĸB signaling. 4,5 As such MDM2 amplification or overexpression drives tumor growth and MDM2 blockade suppresses cancer development. 6,7 In acute kidney injury caused by primary glomerular insults, MDM2 rather fosters podocyte demise by driving the podocytes into mitosis, pushing them to bypass the G2/M checkpoint that is, mitotic catastrophe. 8 Moreover, by facilitating the NF-ĸB signaling, MDM2 promotes glomerular inflammation in injured glomeruli and thus further aggravates the podocyte loss, endothelial damage and glomerulosclerosis. 9 In acute tubular injury MDM2 exacerbates the initial damage phase via NF-ĸB-related inflammation but promotes regeneration in the later healing phase via p53 regulation. 10 In podocyte homeostasis MDM2 functions as a crucial factor protecting podocytes from p53 overactivation related cell death contributing thus to the lifelong survival of podocytes. 11 Resting tubular epithelial cells express high levels of MDM2 and we hypothesized that quiescent tubular epithelial cells require MDM2 to maintain the homeostasis. To address this hypothesis we depleted the MDM2 or both MDM2 and p53 in cultured murine tubular epithelial cells and in primary tubule cells and in the mouse model by generating the tubular epithelial cells-specific knockouts.

Results
MDM2 prevents tubular epithelial cell death in vitro. The knockdown (KD) of MDM2 in mouse tubular epithelial cell line with siRNA resulted in decreased viability of tubular cells in comparison to tubular cells treated with control siRNA (Figure 1a). MDM2 suppression was associated with a marked increase in p53 and p53 effector genes p21 and PUMA transcript levels (Supplementary Figure 1A). The concomitant KD of MDM2 and p53 significantly improved the viability of tubular epithelial cells in vitro (Figure 1a). This result was confirmed in primary tubular cells MDM2 KO pTECs isolated from Pax8rtTa-cre; MDM2f/f mice, where MDM2 was depleted specifically in tubular epithelial cells in vitro by treatment with doxycycline. The generation of theses mice is described below. MDM2 mRNA levels decreased significantly in MDM2 KO pTECs treated with 1μg doxycycline (Figure 1b). The Mdm2-deficient primary tubular cells showed increased expression of tubular damage markers KIM-1, NGAL and TIMP-2 as well as increased cell death, due to the upregulation of p53 (Figure 1b). Increased p53 activity was confirmed by elevated mRNA expression of p53-target genes p21 and PUMA (Supplementary Figure 1B). The simultaneous depletion of MDM2 and p53 completely rescued the viability of the primary tubular cells (Figure 1b).
The pTECs population was about 95% pure as assessed by staining for the tubular epithelial cell markers cytokeratin-7 and E-cadherin ( Figure 1c). To prove the specificity of MDM2 depletion in tubular epithelial cells, we isolated also parietal epithelial and mesangial cells from kidneys of Pax8rtTA-cre; MDM2f/f, treated them with doxycycline, and checked mRNA levels of MDM2 and viability of those cells. Unlike in primary tubular cells, MDM2 expression and viability remained unchanged in the glomerular cells in comparison to the controls (Supplementary Figure 1C). Ultrastructural analysis of murine primary tubular cells lacking MDM2 revealed vacuolization of cytoplasm, nicks in the cytoplasmic membrane, and eventually cell death with complete disassembly of the cell structure ( Figure 1d). These results suggest that MDM2 is required to protect the tubular cells from p53 overexpression-related cell death.
Mdm2 KD in zebrafish larvae induces cell death of renal tubular epithelial cells. To validate the in vitro results in vivo, a KD of mdm2 was generated in zebrafish larvae by the injection of specific morpholinos into fertilized eggs and was confirmed by qRT-PCR (Figure 2a). At 4 days post fertilization (d.p.f.), the mdm2 KD zebrafish larvae developed pericardial edema (white arrow in Figure 2b), a hallmark of kidney failure, in contrast to control morpholino-injected larvae (Ctrl) and larvae with a double KD knockout of mdm2 and p53 (mdm2/p53). For a rapid screening of the glomerular filtration barrier function, we used the transgenic zebrafish strain CADE 12 that expresses the eGFP-labeled protein of the albumin family gc (group specific component) in the blood which is unable to pass the healthy filtration barrier. In Figure 2b, it is shown that the KD of mdm2 resulted in larvae with reduced eGFP fluorescence in the vessels indicating a leakage of the filtration barrier. To investigate the morphology of pronephric tubules after the KD of mdm2, we used the zebrafish strain Tg(wt1b:eGFP) that expresses eGFP in podocytes, parietal epithelial cells and proximal tubule cells. 13 Sections of zebrafish larvae revealed a significant dilatation of the pronephric tubules in mdm2 KD larvae ( Figure 2c) which could be rescued to some extent by the KD of p53. Additionally, we have found less epithelial cells in the proximal tubules after the mdm2 KD compared with Ctrl and mdm2/p53 KD larvae (Figure 2c). The TUNEL assay demonstrated that this is due to an increase of dead cells in mdm2 KD larvae ( Figure 2d). Furthermore, the remaining cells showed a reduced expression of the Na + -K + -ATPase, a marker of differentiated tubular epithelial cells (Figure 2c).
In vivo imaging of the pronephros of living Tg(wt1b:eGFP) larvae by two-photon microscopy (2-PM) confirmed the significant dilatation of the Bowman space after mdm2 KD (asterisk in Figure 2e  ). The nuclei were edematous but with no membrane rupture and they did not appear to have consolidated chromatin (Figure 5a). These abnormalities suggest that the tubular epithelial cells undergo a stage of asphyxia and start dying from the cytoplasm with dead mitochondria. Eventually, the tubular epithelial cells completely disintegrate, the plasma membrane ruptures, and the nuclei are released together with other cytoplasmic organelles ( Figure 5a). Lack of MDM2 in tubular epithelial cells was associated with the progressive increase of p53 expression and its nuclear shift. Simultaneously we observed also the Ki-67 activation, suggesting the hypertrophy or proliferation of undamaged tubular epithelial cells with incomplete MDM2 deletion or of the tubular progenitors (Supplementary Figure 3A). Nevertheless, the regenerating tubular cells were not able to compensate for the cellular loss due to p53 overactivation. TUNEL as well as cleaved caspase-3 staining showed just marginal caspase-3 activation in the tubular compartment, mostly in the kidneys with marked p53 upregulation and massive tubular cell death ( Figure 5b; Supplementary Figure 3B). The caspase-3 activation level does not correspond to the extent of the kidney damage and is suggestive of secondary cell apoptosis. We conclude that MDM2 is essential for tubular epithelial cells homeostasis and survival and prevents acute kidney injury.
The MDM2 deleterious phenotype in tubular epithelium is p53 dependent. To test for the role of p53 in the structural and functional renal pathology of MDM2 − / − tubulus mice, we generated inducible tubular epithelial cell-specific simultaneous MDM2 and p53 knockout mice MDM2/p53 dKO tubulus and the littermate controls with one p53 allele intact MDM2 − / − tubulus /p53 wt/fl . We treated the both mouse lines with doxycycline to induce the deletion of the respective gene alleles. The analysis of the functional and histological parameters of the mice models confirmed that the depletion of p53 rescues the pathological phenotype of the MDM2 − / − tubulus mice (Figures 6a-e).
Tubular epithelium compensates only partially for the cell loss caused by MDM2 depletion. Previously, it was reported that mice lacking Mdm2 in the intestinal epithelium can fully compensate for the MDM2 depletion/p53 activationmediated cellular loss because of negative selection of the MDM2 transgenic cells and rapid proliferation and overgrowth of cells with insufficient Cre recombinase activity and thus incomplete MDM2 depletion. 14 To investigate whether this phenomenon occurs also in renal tubular epithelium, which is known for its fast regenerative capacity, we subjected the Pax8rtTA-cre; MDM2f/f mice to different doxycycline treatment regimens to avoid the rapid lethality of the continuous doxycycline treatment. We treated the mice with doxycycline for 2 days to deplete the MDM2 just in the portion of the tubular epithelial cells and repeated the treatment after 5 days of recovery for three times. The renal function of the experimental mice was impaired compared with the control mice as documented by elevated serum creatinine and BUN levels ( Figure 7a) but the lifespan of the experimental animals was not affected (data not shown). The histological analysis showed focal tubular damage with overproduction of extracellular matrix, suggesting that the loss of tubular cells is only partially recovered by surviving tubular cells but rather mesenchymal repair occurs ( Figure 7b). Masson trichrome staining confirmed the accumulation of the fibrotic tissue in the kidneys of experimental mice ( Figure 7b) and also mRNA expression levels of several pro-fibrotic genes were elevated ( Figure 7c). We detected moderate p53 activation with nuclear shift throughout the tubular compartment with especially high expression in medullar region of the kidney (Figure 7b). We conclude that unlike intestinal epithelium, tubular epithelial cells cannot completely compensate for the cells lost due to the MDM2 deletion. documented by reduced GFR and increased plasma creatinine and BUN. We also detected significant upregulation of tubular damage markers and massive swelling, mitochondrial degeneration and cell death in the tubular compartment in the kidneys of the experimental animals. All these changes developed in unchallenged kidney and were consistent with acute tubular injury. The tubular epithelium was not able to fully compensate for the cell loss due to the MDM2 depletion by proliferation of remaining tubular cells with intact MDM2 but rather mesenchymal healing with fibrosis occurred. The pathological phenotype was completely rescued by co-deletion of p53. We conclude that MDM2 is essential for survival and homeostasis of intact tubular cells in healthy kidney. MDM2, the main negative regulator of p53 suppresses its function in three different ways; it chaperons p53 out of the nucleus, that is, out of the transcriptional centrum, it blocks the transcription of p53 effector genes and it ubiquitinates p53 and targets it for proteasomal degradation. 4 Through the p53 inhibition, MDM2 fosters cell proliferation and prevents p53mediated cell death, cell cycle arrest and senescence. 15 This MDM2 function is indispensable for the embryonic development, wound healing and regeneration 16 but it also facilitates autoimmune disorders, such as lupus erythematosus 17 and carcinogenesis. 18,19 MDM2 is overexpressed frequently in many human tumors, especially in breast cancer and sarcomas. 20 The inhibitors of MDM2 and p53 binding are being currently tested in clinical trials as potential cancer therapy. 21,22 Nevertheless also MDM2 depletion has deleterious effects as documented by embryonic lethality of a mouse model with genetic-deletion of MDM2. The mouse embryos die due to the uncontrolled p53-dependent cell death, while p53 deficiency completely rescues this deleterious phenotype. 23,24 Specific deletion of MDM2 in developing kidney results in impairment of renal progenitor cell expansion, in aberrant nephrogenesis and differentiation and leads to hypodysplasia of the kidneys. [25][26][27] Our data are in line with these studies and show that MDM2 is crucial for the survival and homeostasis of unchallenged tubular epithelial cells in kidney. MDM2 prevents p53-mediated cell death of these renal cells. Our findings differ from a study which showed that unbuffered p53 activity caused by MDM2 deletion is detrimental only in radiosensitive tissues due to the massive cell loss, while the radio-insensitive tissues, including kidney, are protected from cell death but exhibit inhibition of cell proliferation. 28 Our data indicate that MDM2 depletion in tubular epithelial cells leads to massive cell death due to p53 overactivation but proliferation activity is maintained as documented by increased expression of Ki-67, marker of proliferation. Our results are corroborated by another study showing that global MDM2 deletion results in morphological and functional abnormalities in both radiosensitive and radioinsensitive tissues, including kidney. 14 Renal phenotype included initially just mild damage with no aberrant functional effect, but subsequently the kidneys exhibited severe injury with impairment of renal function. 14 Further on, MDM2 deletion in intestinal epithelial cells results in multiple intestinal abnormalities in newborn animals. 29 These abnormalities completely disappeared in adulthood due to the selection against the enterocytes lacking MDM2 and increased proliferation of the epithelium with intact MDM2. 29 Although the renal tubular epithelium has an extensive regenerative capacity, our findings indicate that in kidneys of tubular specific depleted MDM2 mice the complete epithelial healing does not occur and is compensated by mesenchymal healing with extensive fibrosis. This suggests that the inhibition of MDM2 could lead to acute kidney injury and due to incomplete healing contribute to development of chronic kidney disease. 30 In our previous studies we showed that MDM2 in acute kidney injury promotes tubular injury in early postischemic phase via augmentation of NF-kB signaling and thus inflammation in p53-independent manner. 10 Our present data document, that all renal anomalies due to MDM2 depletion in unchallenged tubular epithelial cells are dependent on p53 activity, as the defects completely disappeared when p53 was absent. This is in line with studies which showed dose-dependent attenuation of tubular cell death and kidney injury in mouse model of cisplatin induced nephrotoxicity treated with p53 siRNA or with chemical inhibitor of p53. 31,32 Together, MDM2 is crucial in intact tubular epithelial cells in kidney to prevent spontaneous p53 overactivation dependent cell death and thus prevent acute tubular injury. As MDM2 antagonists are being developed as an alternative treatment to chemotherapy for cancer treatment, it is of note that MDM2 inhibition might be detrimental for normal tissues, especially for kidney as our data suggest. We conclude that MDM2 is essential for survival and homeostasis of tubular epithelium in kidney.

Materials and Methods
Cell culture experiments. Murine tubular epithelial cells (mTECs cell line) were maintained in DMEM medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin.The primary tubular cells (pTECs) were prepared from renal cell suspensions as previously described. 33 In brief, 4 weeks old Pax8-rtTAcre; MDM2f/f or MDM2f/f control mice were used for renal cell extraction. Decapsulated kidneys were diced and digested in 1 mg/ml collagenase A (Roche Diagnostics, Mannheim, Germany) for 30 min at 37°C and then passed through 70 μm pore sieve (BD, Franklin Lakes, NJ, USA), washed, and diluted in 2 ml of PBS. Separation of the tubular segments was achieved through Percoll (31%) Zebrafish experiments:. Zebrafish strains and larvae were maintained as described previously. 34 The following zebrafish strains were used: CADE (Tg (−3.5 fabp10a:gc-eGFP) mitfa w2/w2 ; roy a9/a9 ) which expresses a gc-eGFP fusion protein in the blood plasma. 12 The (Tg(wt1b:eGFP)) strain that expresses eGFP under control of the wt1b promotor. 13 The following morpholinos were used: mdm2 MO: 5′-AACAACTCTCTGTTGCCATTTTGGT-3′; p53 MO: 5′-GCGCC ATTGCTTTGCAAGAATTG-3′; Crtl MO: 5′-CCTCTTACCTCAGTTACAATTTATA-3′. MOs were diluted to a concentration of 0.3 nM. A volume of~3 nl per zebrafish egg was injected into the yolk at the one-to four-cell stage using a microinjector (Transjector 5246; Eppendorf, Germany). For histological staining, we followed our previously published protocols. 12 Anti-A6f antibody (1:25, Sigma-Aldrich) was used. 2-PM experiments: To prevent pigmentation, the zebrafish larvae were subjected to E3 medium supplemented with 0.003% phenylthiourea at 1 d.p.f. For in vivo imaging experiments, zebrafish larvae were embedded in 0.8% low-melting agarose (Biozym LE agarose, Germany) in E3 medium in a dorsal side up position. After hardening the larvae were covered with E3 medium supplemented with 0.06 mg/ml Tricaine (Sigma-Aldrich, St Louis, MO, USA). All 2-PM experiments were performed at 22°C. Z-stacks of pronephric glomeruli and proximal tubules were recorded over a distance of 98 μm with a voxel volume of 0.30 × 0.30 × 2 μm and reconstructed as 3D movies using the ZEN 2010 software (Carl Zeiss Microimaging, Jena, Germany) as described before. 35 For significance testing Gaussian distribution was checked by Kolmogorov-Smirnov testing, followed by significance testing with Student's t-test using SPSS V.22 (IBM, Armonk, NY, USA). P-values below 0.05 were declared as statistical significant. All experiments were performed in accordance with German animal protection law overseen by the agencies of the Federal State of Mecklenburg -Western Pomerania.
Mouse experiments. The Pax8-rtTAcre;MDM2f/f mice were generated by breeding the MDM2 flox/flox mice, in which loxP sites flanked the exon 4 and 5 of MDM2 gene, with Pax8-rtTAcre mice, expressing the inducible Cre recombinase under control of the Pax8 promoter. The MDM2 flox/flox mice (/B6 mixed background) were a generous gift from G. Lozano (University of Texas, Huston, USA) and the Pax8-rtTA mice (C56Bl/6 background) were kindly provided by T. Huber (University of Freiburg, Freiburg, Germany). Both mice strains were previously described. 36,37 The MDM2 flox/flox littermates lacking the Pax8-rtTAcre transgene were used as control mice. To generate the tubular epithelial cell-specific MDM2/p53 double knockout Pax8-rtTAcre;MDM2f/f;p53f/f (MDM2/p53dKD tubulus) mice, we bred the Pax8-rtTAcre;MDM2f/f mice with p53 flox/flox mice, in which the loxP sites were inserted in intron 1 and 10 to ensure Cre-mediated removal of nearly all coding sequences. 38 The p53 flox/flox mice were a generous gift from L. Rudolph (University of Ulm, Ulm, Germany). The Pax8-rtTAcre;MDM2f/f;p53wt/f (MDM2 − / − tubulus/p53wt/fl) littermates were used as positive controls. To induce Cre recombinase expression, we treated the mice with 2mg/ml doxycycline in drinking water supplemented with 5% sucrose for various number of days as indicated in figures. All animal studies were approved by the Committee on Research Animal Care, Regierungspräsidium Oberbayern.
Renal function measurement: Plasma creatinine levels as well as BUN (DiaSys GmBH, Holzheim, Germany) levels were determined using commercially available kits as per manufacturer's protocol.
GFR measurement: GFR was measured transcutaneously using a USB-device according to manufacturer's instructions (Mannheim Pharma&Diagnostics GmbH, Mannheim, Germany). The USB-device was fasten with adhesive tape to the shaved area on the mouse back. The mice were injected with FITC-Sinistrin i.v. (15 mg FITC-Sinistrin/100 g body weight). The measurement was performed for 60 min and analyzed with software provided by the manufacturer.
Morphometry: To assess tubulointerstitial changes a semiquantitative score was established to evaluate the degree and extent of tubulointerstitial damage of each field and was graded from 0 to 4 as follows: 0 represents no lesion, 1 represents tubulointerstitial damage of less than 25% per field, and 2, 3 and 4 represent tubulointerstitial damage of 25-50%, 50-75% and more than 75% of the tubulointerstitium, respectively. Approximately 30 cortical and medullary visual fields (20 × ) per kidney were evaluated. Tubulointerstitial injury was defined by features such as tubular collapse, cast formation with tubular dilatation or atrophy, detachment of cells from the basement membrane.