Renal cell carcinoma escapes death by p53 depletion through transglutaminase 2-chaperoned autophagy

In renal cell carcinoma, transglutaminase 2 (TGase 2) crosslinks p53 in autophagosomes, resulting in p53 depletion and the tumor's evasion of apoptosis. Inhibition of TGase 2 stabilizes p53 and induces tumor cells to enter apoptosis. This study explored the mechanism of TGase 2-dependent p53 degradation. We found that TGase 2 competes with human double minute 2 homolog (HDM2) for binding to p53; promotes autophagy-dependent p53 degradation in renal cell carcinoma (RCC) cell lines under starvation; and binds to p53 and p62 simultaneously without ubiquitin-dependent recognition of p62. The bound complex does not have crosslinking activity. A binding assay using a series of deletion mutants of p62, p53 and TGase 2 revealed that the PB1 (Phox and Bem1p-1) domain of p62 (residues 85–110) directly interacts with the β-barrel domains of TGase 2 (residues 592–687), whereas the HDM2-binding domain (transactivation domain, residues 15–26) of p53 interacts with the N terminus of TGase 2 (residues 1–139). In addition to the increase in p53 stability due to TGase 2 inhibition, the administration of a DNA-damaging anti-cancer drug such as doxorubicin-induced apoptosis in RCC cell lines and synergistically reduced tumor volume in a xenograft model. Combination therapy with a TGase 2 inhibitor and a DNA-damaging agent may represent an effective therapeutic approach for treating RCC.

Transglutaminase 2 (TGase 2, E.C. 2.1.2.13) is an enzyme that catalyzes an isopeptide bond between protein glutamine and lysine residues, resulting in a covalent crosslink. 1 Previously, we found that, in renal cell carcinoma (RCC) cell lines, TGase 2 crosslinks p53 into aggregates in the autophagosome, resulting in p53 degradation by autophagy. 2 This p53 instability allows tumor cells to evade apoptosis and grow. Instability of p53 in RCC is not associated with mutations because only 4% of samples of clear cell RCC in the COSMIC database present p53 mutations (Supplementary Figure 1). We recently reported that monotherapy using the TGase 2 inhibitor GK921 in a xenograft tumor model abrogated RCC growth through p53 stabilization. 3 These results suggest the possibility of using TGase 2 inhibitors as a cancer therapy. 3 p53 Regulation by human double minute 2 homolog (HDM2) involves proteasomal degradation through ubiquitination, whereas TGase 2-mediated p53 regulation involves autophagosome degradation. However, how TGase 2 depletes p53 in the context of the central regulation system of p53 under HDM2 surveillance is unknown. The half-life of p53 is very short due to its interaction with HDM2. HDM2 depletes p53 by binding to a region of its transactivation domain, 4 which facilitates rapid turnover through ubiquitination. 5 DNA damage-induced phosphorylation of the N terminus of p53 blocks its interaction with HDM2, 6 which stabilizes p53 by inhibiting proteasome-mediated degradation. 7 In this study, we found that silencing TGM2, the gene encoding TGase 2, in RCC cell lines induced cell death under starvation conditions through p53 stabilization to the same extent as did silencing of HDM2. This result suggests that the general instability of p53 in RCC largely depends on TGase 2mediated autophagy. In fact, p53 depletion induced by TGase 2 has no energy cost to cancer cells because TGase 2 only uses calcium to catalyze covalent crosslinks. However, the role of TGase 2 in autophagy is not clearly understood in cancer, although it is known that TGase 2 knockout mice present impaired autophagy, which increases ubiquitinated protein aggregates upon starvation and deceases p62dependent peroxisome degradation. 8 The binding mechanism between TGase 2 and p62 is unknown, although TGase 2 is able to bind directly with p62, which often localizes in the autophagosome. 2,8 p62 (also known as sequestosome 1) is an adapter protein in the degradation of ubiquitinated proteins in autophagosomes through interaction with microtubuleassociated protein 1 light chain 3 alpha (LC3). 9 The N terminus of p62 is a PB1 (Phox and Bem1p-1) domain that binds to the PB1 region in the atypical protein kinase C, the MAPK kinase, and the NBR1 (neighbor of BRCA1 gene) protein. 10 The C terminus of p62 harbors a ubiquitinassociated (UBA) domain 11 that interacts with ubiquitin and polyubiquitin chains, and an LC3-interacting region domain that interacts with LC3 in phagophore membranes. 12 The part of the p53-TGase 2 complex that binds to p62 is unknown, but could be a part of either p53 or TGase 2. We observed that p53 1 does not bind to p62 directly but is transferred to p62 through association with TGase 2.
Here, we explored the possibility that TGase 2 is a chaperone in autophagy by analyzing the domains of p53 and p62 that bind to TGase 2 as well as the domain of TGase 2 that binds to p62. Based on the finding that TGase 2 inhibition increases p53 stability, we also tested the possibility that DNA-damaging reagents such as doxorubicin expand the damage of p53-mediated cell death when combined with TGase 2 inhibition in a preclinical model of RCC.
Results p53 is regulated competitively by TGase 2 and HDM2 in RCC under starvation. To assess the relative contributions of TGase 2 and HDM2 in p53 depletion in RCC cell lines, the two corresponding genes were silenced using small interfering RNAs (siRNAs; Figure 1). The amount of p53 increased about 2.7-fold when TGM2 was silenced, and about 2.5-fold with HDM2 silencing in ACHN cells (Figures 1a  and b). A similar pattern was observed in CAKI-1 cells, with a 1.8-fold and 2.0-fold increase, respectively (Figures 1c and  d). This result suggests that p53 regulation depends equally on HDM2 and TGase 2 in RCC cells under starvation conditions. We next examined whether the observed p53 stabilization was potentiated by treatment of chloroquine as an autophagy inhibitor or MG132 as a proteasome inhibitor. Treatment with chloroquine or MG132 alone stabilized p53 over twofold in both ACHN and CAKI-1 cells (Figures 1a-d). Combination treatment with both an siRNA and either chloroquine or MG132 obtained additional p53 stabilization (Figures 1a-d). Treatment with both siRNA for TGM2 and chloroquine had the greatest effect on p53 stability, increasing its levels to 4.5times the control level (Figure 1a), whereas the silencing of HDM2 combined with MG132 increased p53 levels to four times the control level ( Figure 1b). The apoptosis of ACHN and CAKI-1 cells to gene silencing was tested in a terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay (Figures 1e-h). TUNEL showed that p53-positive cells increased in ACHN cells by about 16-and 14-fold in response to TGM2 and HDM2 silencing, respectively (Figures 1e and f). Similarly, in CAKI-1 cells, p53-positive cells increased by about 20-and 18-fold in response to TGM2 and HDM2 silencing, respectively (Figures 1g and h). Nutlin3a treatment onto RCC under normal culture media does not induce apoptosis that undergoes cell cycle arrest. 13 However, Nutlin3a treatment under starvation induces remarkable apoptosis as we observed in HDM2 (Supplementary Figure 3).
TGase 2 competes with HDM2 for binding to p53 in RCC. To test whether TGase 2-dependent autophagic depletion of p53 is a collateral mechanism against HDM2mediated p53 regulation, we used p53 immunoprecipitation to examine protein-protein binding ( Figure 2). Silencing of TGM2 increased the binding of HDM2 to p53 whereas it abolished the binding of p53 with p62 ( Figure 2a).
Knockdown of HDM2 increased the binding of TGase 2 and p62 to p53 (Figure 2b). These results suggest that TGase 2 may bind to the same region of p53 where HDM2 binds, and that TGase 2 may chaperon p53 to p62. Doxorubicin 14 and etoposide 15 induce rapid and extensive apoptosis through DNA damage and p53 phosphorylation that inhibits binding to HDM2. Therefore, we tested whether doxorubicin or etoposide treatment stabilizes p53 from TGase 2. Doxorubicin or etoposide treatment inhibited the binding of TGase 2 to p53, and induced p53 phosphorylation (Figure 2c). To identify whether this loss of function is directed to Ser15 phosphorylation of p53, a constitutively activated form of p53 (S15E) and an inactive form (S15A) were transfected in cells and tested for binding to TGase 2. The S15E mutant of p53 showed substantial loss of binding to TGase 2 ( Figure 2d). TGase 2 lost its ability to bind p53 following induction of p53 phosphorylation at S15 in a timedependent manner during doxorubicin treatment ( Figure 2e). This result suggests that TGase 2 is a surveillance regulatory molecule against p53, like HDM2, due to the loss of binding activity when p53 is activated by DNA damage. However, participation of TGase 2 in p53 depletion through autophagy has an important role in the survival in RCC cells not only by reducing p53-dependent apoptosis, but also by supplying biomolecules through activation of autophagy under starvation conditions.
Previously, we identified TGase 2-targeting sites in p53 at multiple glutamines and lysines, including in the DNA-binding core (residues 102-292). 2 In that study, we found that TGase 2 binds to the N terminus of p53 at the HDM2-binding domain (transactivating domain). This resembles HDM2 that binds to the N terminus of p53 and ubiquitinates multiple lysine residues including K370, 372, 373, 381, 382 and 386 in the C terminus of p53. 16 We found that the TGase 2-binding domain and TGase 2-targeting sites in p53 are different because TGase 2 works as a chaperon of p53 to the autophagosome. This finding is in accordance with the observation that TGase 2-mediated crosslinking occurs only in the autophagosome during autophagy. 8 TGase 2 chaperones p53 to the autophagosome. TGase 2 transports p53 to the autophagosome for degradation by autophagy through polymerization. This idea was supported by immunoblotting of p53, which showed that high molecular weight polymers were only detected in the insoluble fraction after chloroquine treatment in RCC. 2 Here, we tested the binding of three molecules, namely, TGase 2, p53, and p62, in ACHN ( Figure 3a) and CAKI-1 (Figure 3b) cells. TGM2 knockdown abolished p53 binding to p62 and significantly reduced p62 binding to p53. This result suggests that p53 does not bind to p62 directly and that TGase 2 is required for p53 autophagy in RCC. It is known that p62 is located in the autophagosome during autophagy. Therefore, this implies that p53 bound to TGase 2 transports to p62 by TGase 2-p62 binding. In other words, TGase 2 is a chaperone of p53 for autophagy.
Considering TGase 2 as a chaperone, its catalytic activity might not be necessary for chaperoning p53 in RCC. To test this possibility, an inactive, double mutant form of TGase 2 (C277S and C370A) 2,17 was transiently expressed in HEK293 cells, and cell extracts were subjected to immunoprecipitation using an anti-HA-tag antibody ( Figure 3c). This mutant TGase 2 also bound p53 as well as p62 despite the lack of enzymatic activity.
A series of deletion mutants reveals the interacting domains of p53 and p62. Next, to investigate which part of p62 binds to TGase 2, a series of FLAG-tagged deletion mutants of p62 was constructed, transfected into HEK293    (d) The activated p53 mimic (S15E) did not interact with TGase 2. CAKI-1 and ACHN were transfected with expression plasmids for p53 (wild type), p53 (S15E), or p53 (S15A) for 24 h, and the cell lysates were immunoprecipitated with a His-tag antibody and subjected to immunoblotting. (e) p53 stabilization (p-p53) by doxorubicin blocked interaction with TGase 2 in a time-dependent manner. After doxorubicin (1 μM) treatment for up to 20 h, cell lysates were immunoprecipitated with an anti-p53 antibody and subjected to immunoblotting. The blots are representative of three independent experiments cells together with HA-tagged TGase 2, and tested for binding ability by immunoprecipitation using an anti-HA-tag antibody. Deletion of the PB1 domain (residues 1-123) of p62 totally abolished binding to TGase 2 ( Figure 4a). Four other deletion mutants retained binding, including one lacking residues 1-85 and 345-442. These results suggest that the region comprising residues 85-110 of p62 (a part of the PB1 domain) is responsible for interacting with TGase 2.
To determine which part of TGase 2 binds to this part of p62, a series of HA-tagged deletion mutants of TGase 2, designed in consideration of the protein's X-ray crystal structure, 17 was tested by binding to full-length FLAG-tagged p62. Deletion recombinants missing the C terminus of TGase 2 had no p62binding ability (Figure 4b). Therefore the C terminus region (residues 461-687) of TGase 2, which is composed of two β-barrel domains, has an important role in interacting with the PB1 domain of p62.
TGase 2 binds to p53 at its N terminus, which is also the HDM2-binding domain 18 (Figure 2). To identify the p53binding domain in TGase 2, a series of HA-tagged TGase 2 deletion recombinants was transiently expressed in HEK293 cells together with FLAG-tagged p53, and the proteins were immunoprecipitated from cell extracts with an anti-HA-tag antibody. A mutant TGase 2 consisting of the N terminus  TGase 2 activity is not required for interacting with p53. Wild-type or catalytically inactive TGase 2 (double mutant, C277S and C370A) was co-transfected with p62 into HEK293 cells. TGase 2 was immunoprecipitated using an anti-HA-tag antibody, followed by immunoblotting of TGase 2, p53 and p62 residues 1-139 retained binding ability to p53, whereas the constructs missing the N terminus did not bind p52 (Figure 4c). The N terminus of TGase 2 (residues 1-139) is a fibronectin-binding domain 19 that also interacts with the N terminus of p53, which also contains the HDM2-binding domain (residues 15-26). The TGase 2 C terminus, which consists of two β-barrel domains, is therefore available for interacting with the N terminus of p62 (residues 85-110), which consists of a PB1 domain. The C terminus of p62 is therefore free to interact with an autophagy molecule, LC3 (Figure 4d).    the two drugs alone or in combination. No physical toxicity due to the drugs was observed in mice treated with combination of GK921 and doxorubicin for 2 weeks (data not shown). Oral administration of GK921 (2 mg/kg/day) and doxorubicin (1 mg/kg/day) began when tumors reached a volume of 100 mm 3 and proceeded for 5 days per week. After 5 weeks of treatment, tumor volumes were significantly different between GK921-treated mice and vehicle-treated controls, and in mice that received the combined treatment, tumor volumes had receded substantially (Figure 6a). Others reported that combination treatment of 5-FU (8 mg/kg/day) and sorafenib (15 mg/kg/day) using CAKI-1 cells showed about 60% remission of tumor volume at day 42 while single treatment of sorafenib resulted in 40% remission. 20 This combination result is remarkable therapeutic effect because tumor volume is decreased below the initial tumor volume without any side effect of drugs.
The mRNA expression levels of p62/SQSTM1 and TGM2 in clear cell RCC patients included in TCGA were remarkably and significantly higher than those in matched normal kidney tissues (Figures 6b and c). In this data set, moreover,~5% and 9% of cases (26/504 and 47/504) exhibited higher than average levels of expression of p62/SQSTM1 and TGM2. A survival analysis demonstrated that patients with higher than  (Figures 6d and e). In our previous study, 2 TGM2 upregulation in clear cell RCC patients suggested a potential clinical therapeutic target. Based on this bioinformatics analysis, we suggest that TGase 2mediated autophagy promotes tumor cell survival through chaperoning p53 to p62.
A proposed mechanism of TGase 2-mediated chaperoning of p53 to the autophagosome. Structural modeling of the triple complex of p53-TGase 2-p62 suggests that TGase 2 is a good chaperone molecule for p53 transportation to p62 without catalytic activity (Figure 7). This model is supported by our experiment in which an inactive form of TGase 2 bound p53 and p62 (Figure 3c). TGase 2 in the autophagosome is activated by an increase in calcium concentration, which crosslinks p53 monomers into polymers. These protein aggregates are then degraded by lysosomal proteolytic enzymes when the autophagosome fuses with the lysosome (Figure 7).

Discussion
RCC survives and grows by inactivating p53 through TGase 2-mediated autophagy, which supplies recycled biomolecules under conditions of starvation. RCC patients with high TGM2 expression had poor overall survival. It is not surprising that p53 levels are suppressed in RCC, although only 4% of RCC samples have p53 mutations. Therefore, TGase 2 is a possible therapeutic target for treating RCC, and this has been proved in a xenograft model. 3 We found that DNA-damaging reagents such as doxorubicin enhance the p53 stability provided by the TGase 2 inhibitor GK921. We previously reported that monotherapy with the TGase 2 inhibitor GK921 had an anticancer effect under 8 mg/kg/day in an RCC xenograft model.
In this study, combination therapy with doxorubicin (1 mg/kg/ day) and GK921 (2 mg/kg/day) caused the RCC tumor to recede below the initial tumor volume. Therefore, combination therapy with these two agents represents a promising therapeutic approach to RCC.
In this study, we revealed the mechanism by which TGase 2 transports p53 to the autophagosome. We observed that the N terminus of TGase 2 binds the N terminus of p53 regardless of mutations and simultaneously associates with the N terminus of p62 through the C terminus of TGase 2. The LC3-binding domain of p62 is located in the C terminus of p62, which allows the formation of autophagosomes, whereas p62 interacts with a complex of TGase 2 and p53. During translocation of p53 to the autophagosome through TGase 2 binding, crosslinking activity is not needed. This finding is in agreement with the observation that TGase 2 crosslinking activity occurs only in the autophagosome during autophagy. 8 This suggests that TGase 2 acts as a chaperone of p53 with a crosslinking catalytic activity. This interaction may result in rapid autophagy without consuming energy to tag ubiquitin on p53, as p62 is known to interact with ubiquitinated proteins. This autophagy process is beneficial in that it supplies building blocks, including degraded p53, for cancer cells. The N terminus of TGase 2 interacts with the N terminus of p53 and, simultaneously, the C terminus of TGase 2 interacts with the N terminus of p62; as a result, a heterotrimeric complex (p53-TGase 2-p62) is formed. The C terminus of p62, in the p53-TGase 2-p62 complex, is free and moves the complex to LC3 in the phagophore. When the autophagosome is completed with components of the phagophore, p53 is polymerized by TGase 2 with calcium-dependent activation in the autophagosome. Later, the autophagosome and lysosome are fused into an autolysosome, which degrades all crosslinked materials.
P62 has multiple physiological roles in cancer. 21 The PB1 domain of p62 inhibits ERK1, which is a critical regulator of  Figure 7 Model of the interactions of TGase 2, p53 and p62 based on the experimental data. The N terminus of TGase 2 interacts with the N terminus of p53, and the C terminus of TGase 2 interacts with the N terminus of p62 simultaneously. The C terminus of p62 in the complex is free and transfers the complex to LC3 in the phagophore. P53 is polymerized in the autophagosome when calcium increases. Later, the autophagosome and lysosome fuse into an autolysosome, which degrades all crosslinked materials. The TGase 2-binding site of p53 was identified as the HDM2-binding site in Figures 2d and e. Previously, we found that the TGase 2-targeting site is the DBD domain. CD, complexing domain; DBD, DNA-binding domain adipogenesis. Therefore, TGase 2 binding to the PB1 domain of p62 may increase ERK1 activity. Polyubiquitination consumes a lot of ATP to produce polyubiquitinated proteins, which are recognized by the UBA domain of p62 for autophagy. However, TGase 2-mediated p53 chaperoning to p62 is ubiquitination-independent through binding to the PB1 domain. The negatively charged, unfolded N terminus of p53, apart from the next α-helix (residues [15][16][17][18][19][20][21][22][23][24][25][26][27][28][29], may interact with the positively charged region of the N terminus of TGase 2 (residues 28-35).
Although the median progression-free survival has been increased by targeting tumor angiogenesis, 22 about one-third of RCC patients develop metastatic disease. 23 The main reason for metastatic disease is resistance to chemotherapy. Cancer cells are able to survive through autophagy, which is induced by many targeted kinase inhibitors. 24,25 It is well known that treatment with the mTOR inhibitor rapamycin induces autophagy by releasing mTOR mediated by suppression of autophagy. Therefore, autophagy can be induced not only by a lack of nutrient supply due to blood vessel shortage but also by drug treatment. A recent preclinical study reported that an mTOR 1/2 dual inhibitor combined with the autophagy inhibitor 3-methyaldenine resulted in a good outcome in an RCC model. 26 Under drug-induced autophagy conditions, it remains to be determined whether TGase 2 is induced for p53 depletion through autophagy.
When RCC is targeted by angiogenesis with sorafenib, autophagy is also induced and correlates with drug resistance. 27 In this study, we found that TGase 2 induces p53 depletion through transportation to p62 in RCC. Therefore, RCC can evade p53-dependent apoptosis by TGase 2 expression, which is induced by treatment with an anti-cancer drug such as doxorubicin. 28 RCC may take advantage of the fact that TGase 2 destabilizes p53 activation induced by doxorubicin. Indeed, RCC evades cell death through TGase 2-mediated p53 depletion, which correlates with overall survival of RCC patients. Doxorubicin treatment combined with TGase 2 inhibition with GK921 inhibited RCC growth in a xenograft model even though the dose of GK921 used was below that needed for efficacy in monotherapy. Therefore, this combination therapy may represent a new approach to treating drug-resistant RCC.