The Proto-Oncoprotein c-Cbl Protects Cells against Oxidative Stress by Down-Regulating Apoptosis and is Highly Expressed in Several Cancers

The proto-oncoprotein c-Cbl controls several down-regulation signals leading to degradation of tyrosine-kinase receptors (TKR) [1-3] or endocytic trafficking [1,4,5]. It interacts with numerous signalling proteins through their Src homology-2 or -3 (SH2 or SH3) domains [6,7]. Several groups, including ours, have reported an association between down-regulation of apoptosis and c-Cbl. First, oncogenic forms of c-Cbl are anti-apoptotic [8]. The pro-apoptotic Bim EL BH3only protein has been shown to be down-regulated by c-Cbl in primary culture of osteoclastes [9]. We found a similar, though probably indirect, regulation of BimEL through c-Cbl in mouse testicular germ cells and extended these findings to the pro-apoptotic factor Smac/ DIALO [10], indicating that c-Cbl likely exerts an anti-apoptotic effect. Langdon et al. reported that a c-Cbl RING finger mutant leads to thymocyte apoptosis and to an unexpected Akt activation, as well as an increase expression of Bim EL [11,12]. Sproul et al., using PC12 cells and primary cultures of neurons [13], reinforced the anti-apoptotic effect of c-Cbl, reporting that c-Cbl inhibits the ability of mixedlineage kinase (MLK), from the mitogen-activated protein (MAP) kinase family, to activate c-Jun N-terminal kinase (JNK). Finally, it has been reported that c-Cbl regulates the degradation of the pro-apoptotic TRAIL receptors [14,15] and that the TRAIL/MEKK4/HSP27/Akt survival network is modulated by the Src/CIN85-c-Cbl complex [16].

down-regulation and its anti-survival aspect [1][2][3], the commonly accepted view has been that the E3-Ubiquitine ligase c-Cbl would function as a pro-apoptotic modulator, and not as an anti-apoptotic one [17][18][19][20]. Therefore, it is important to establish more clearly the regulation of apoptosis exerted by c-Cbl in somatic cells and its consequence in cancer. Considering that alteration of programmed cell death is the main explanation for the gradual accumulation of human cancer cells [21,22], and that c-Cbl over-expression has been recently observed in human prostate adenocarcinoma [23], the elucidation of the role of c-Cbl in malignant cell apoptosis is of obvious importance, especially if c-Cbl is over-expressed.
Herein, we report that c-Cbl is anti-apoptotic based on several approaches: in vivo experiments with prostatic cells, ex vivo experiments through the study of mouse embryonic fibroblasts (MEFs), and in vitro experiments with human prostate adenocarcinoma cell lines (LNCaP). We demonstrate that c-Cbl acts as a cell protector against oxidative stress (ex vivo and in vitro experiments) and that this process follows the Apoptosis Signal-regulating Kinase-1-p38MAPK (ASK1-p38MAPK) stress pathway. Based upon preliminary experiments, we suggest that this effect is independent of the well-known Grb-2 associated TKR down-regulation effect. We show that c-Cbl is over-expressed in several types of human cancer and appears to correlate with ROS expression, which reinforce an active role for c-Cbl in malignant cell survival. These results suggest interesting diagnostic and therapeutic opportunities.

Ethic statements
All studies on animals were conducted in accordance with current regulations of ALECS -Lyon approved by the INSERM/CNRS animal care committee (France). The human material (frozen tissues and slides) was furnished by the laboratory of Anatomy-Pathology in Lyon-Sud hospital (Pr. F. Berger, Dr. M. Decaussin). All prostate tumours and benign prostate hypertrophy (BPH) were collected from patients treated in the Lyon-Sud hospital within the "Hospices Civils de Lyon", with informed consent from all participants involved in this study. The consent obtained from each participant was written, in accordance with the current regulation approved by the "Hospices Civils de Lyon" and by the INSERM/CNRS ethic committee (France).

Animal origins and manipulations
The c-Cbl knock-out (KO) mice constructed from sv129 WT mice were generously provided by Dr Hua Gu and Dr Naramura [24]. Male Sprague-Dawley rats of 60 days old were obtained from IFFA Credo (L'Abresle, France). The mice were decontaminated and housed in ALECS module SPF in Lyon, France and used for in vivo experiments at 10 weeks of age. The anti-androgen treatment consisted of vehicle control (methylcellulose) with or without 25 mg/kg/day (mice) or 10 mg/kg/day (rats) of flutamide and was administrated by daily gavages for four days. Six to ten animals were used per experimental condition. Castration experiments were performed on rats (60 days old) under pentobarbital anaesthesia (60 mg/kg body wt).

Cells, culture conditions and experiment analysis
Mouse embryos (13.5 day-old) from c-Cbl+/-mating were used to obtain primary mouse embryonic fibroblasts (MEFs) then cultured according to established protocols [26,27]. Cells were seeded 24h before the start of the treatment (10 5 cells in 1 mL cell culture medium/ well), etoposide added to final concentration from 1 to 30 µM and H 2 O 2 from 50 nM to 100 µmM. Cells were lysed for protein extraction 16 to 24 h after treatment, followed by western blot, TUNEL experiments or collected using Cytospin (10'@800 g) for 4'-6 diamino-2-phenylindole (DAPI) staining.
For western blotting, prostatic tissue was homogenized in 200 ml of ice-cold hypotonic buffer (25 mM Tris-HCl (pH 7.4) and proteaseinhibitor cocktail). Tissues were further homogenized by sonication (10 sec). Protein concentration was determined by the Bradford assay. Proteins (100 mg) were resolved on 10% SDS/polyacrylamide gels and electrophoretically transferred to nitrocellulose membranes using CAPS buffer (13, pH 11) containing 20% methanol at a constant voltage of 100 V for 60 min. Following transfer, the membrane was incubated in TBS containing 10% fat-free dry milk and 0.1% Tween-20 for 2 h at room temperature. The membrane was rinsed three times with TBS/0.1% Tween-20 (three times for 10 min each) and then incubated with the first antibody (in TBS containing 2% fat-free dry milk) overnight at 48°C. The antibody concentrations were used following the manufacturer's instructions (1:100 procaspase-6, cleaved caspase-9 and cleaved caspase 3; 1:200 BimEL, Smac/DIABLO, Bak, IAPs, c-Cbl, Bax antibodies). The membrane was then rinsed with TBS/0.1% Tween-20 (three times for 10 min each) and incubated with horseradish peroxidase labeled anti-rabbit IgG (1:2500) or antigoat IgG (1:2000) in TBS containing 2% fat-free dry milk and 0.1% Tween-20 for 1 h at room temperature. The membrane was thoroughly washed with TBS/0.1% Tween-20 (three times for 10 min each) and then with TBS. Bound antibodies were detected by chemiluminescence using a CovalAb detection kit and Biomax MR Films. The protein loading was checked by reprobing the blot with a rabbit IgG antiactin (1:500). Western blots were processed identically, and comparison of protein expression used bands from the same gel.

Image acquisition and manipulation
The images were acquired using a microscope (Axioskop; Carl Zeiss MicroImaging, Inc.) with plan-Neofluar objective lenses (Carl Zeiss MicroImaging, Inc.) at 40×/NA 0.75. Observations were performed with a 3,200 K halogen light with a daylight blue filter using a digital imaging medium. DAB chromogen was used. The camera (Coolpix 990; Nikon) used the Nikon acquisition software. Manipulations were performed at room temperature. Image processing was performed with Adobe Photoshop. Only the whole images were processed with brightness, contrast, and color balance adjustments.

Statistical analyses
Data are expressed as the mean S.D. Three to seven animals from different litters were used. For statistical analysis of data generated in both in-vivo and in-vitro models, one-way ANOVA was performed to determine whether there were differences between all groups (P<0.05), and then the Bonferroni post-test was performed to determine the significance of the differences between the pair of groups. P<0.05 was considered significant. The statistical tests were performed on Stat View software version 5.0 (SAS Institute Inc.).

c-Cbl down-regulates apoptosis in mouse ventral prostate through the apoptotic mitochondrial pathway
In order to relate apoptosis directly to the expression of c-Cbl in vivo, we investigated the apoptotic status in prostatic ventral (PV) cells of c-Cbl-deficient (KO) mice [24] compared with wild type (WT) mice (western blots were processed identically, and bands were compared only to bands from the same gel). We used prostate cells because of the androgen ability to affect the survival/apoptosis balance of ventral prostatic epithelial cells [29,30]. We first assessed the c-Cbl dependency on androgens through several means, clarified in supplementary information ( Figures SI 1 and SI 2). All experiments clearly showed that c-Cbl expression is androgen-dependent in rat and mouse ventral prostate epithelial cells, decreasing with the progressive occurrence of apoptosis. We then addressed the role of c-Cbl in this process.
The mitochondrial pro-apoptotic factors BimEL and Smac/ DIABLO were significantly increased in the ventral prostatic cells of c-Cbl-deficient mice ( Figures 1A and B) compared to wild type mice. We also found the apoptotic factor Bak significantly elevated in deficient mice ( Figures 1C). Conversely, the inhibitors of apoptosis c-IAP1 and XIAP were significantly lower in c-Cbl-deficient mice, when c-IAP2 was not affected ( Figures 1D, E and F). Finally, the mitochondrial activated caspase-9, as well as procaspase-6, were found slightly and constantly elevated in c-Cbl-deficient mice ( Figures 1G and H), indicating the involvement of the mitochondrial apoptotic pathway. TUNEL experiments confirmed these data ( Figures 1I), showing a slight but significant increase of the number of c-Cbl-deficient apoptotic epithelial cells. Flutamide treatment, at the doses used, did not change the increased apoptotic ratio when c-Cbl deficient mice and WT control mice were compared ( Figures 1H and I). Therefore, we consider c-Cbl as an average anti-apoptotic regulator in prostate epithelial cells that increases the physiological apoptotic threshold, with no specific interference with the anti-androgen apoptosisinducing pattern. c-Cbl strongly protects primary mouse embryonic fibroblasts against oxidative stress polyclonal antibodies were obtained from Abcam. Cleaved caspase-3 (5AE1) rabbit monoclonal antibody was obtained from Cell Signaling Laboratories as well as MLK3, ASK1, p38MAPK and phospho (Thr180/ Tyr182)-p38MAPK polyclonal antibodies.
For DAPI staining, cells were fixed in 3:1 solution Ethanol: Acetic acid 10mn, stained with DAPI-NaCl 0.9% (0.4 µg/mL final) 30 min at RT. The cells undergoing apoptosis were evaluated using DAKOs fluorescent mounting medium by nuclear shrinkage or apoptotic bodies.
TUNEL analysis was performed on paraffin section of 5 mm thickness of Bouin-fixed ventral prostatic lobes from c-Cbl deficient mice or wild-type mice. The slides were treated as already described [10].
LNCaP culture and treatment: the androgen-dependent human prostatic carcinoma cell line LNCaP was used obtained from generous gift of Pr. L. Morel [28] at the 60th passage. The cells were cultured as indicated by the manufacturer. For H 2 O 2 or etoposide treatment, cells were cultured for 16 h or 24 h with indicated concentrations.

RNA silencing and cDNA transfections
The RNA silencing was done using Lipofectamine 2000 following the manufacturer's instructions (Invitrogen). The c-Cbl RNAi was generated by Eurogentec, whose sequence is: 5'GGGAAGGCUUCUAUUUGUU3'. The siRNA controls were purchased from Invitrogen: siRNA-A (sc-37007) and siRNA-B (sc-44230). The Grb-2 siRNA was obtained from Santa-Cruz Biotechnology laboratories.
pcDNA3-120 cbl was very generously provided by Dr W.Y. Langdon. Transfections of c-cbl-deficient MEF were done using Lipofectamine 2000 following the manufacturer's instruction (Invitrogen).

Human sample treatments
Frozen tissue samples were obtained from six consenting patients (P1 to 6) following total prostate resection and used for western blotting. For immunohistochemistry experiments, slides of BPH and prostatic adenocarcinoma were fixed for 24h (Bouin), dehydrated and embedded in paraffin. Sixteen prostatic adenocarcinomas were examined versus eight BPH. Normal surrounding tissue served as controls.
Tissue microarrays were purchased from TMA US Biomax and were used to compare diverse cancer spots with normal ones. Spots of a minimum of six patients were analysed for a given pathology and seventeen organs considered.

Automated immunohistochemistry analysis
The IHC procedure was performed with a Ventana Benchmark XT autostainer using the manufacturer's procedure. After paraffin removal, the slides were submitted to antigen retrieval with Cell Conditioner for 30 min at 95°C. Slides were incubated for 32 min at 37°C with specific primary antibodies. A Ventana-kit was applied including biotin/avidin/phosphatase system and the chromogen Fast-Red. Slides were counterstained with hematoxylin. Negative controls consisted in omitting the specific primary antibodies. Specimens were considered immuno-positive when 1% or more of the tissue had clear evidence of immuno-staining. The immuno-staining was evaluated by two independent observers. In order to reveal better the anti-apoptotic potential of c-Cbl using appropriate controls, we evaluated the apoptotic level of ex vivo primary mouse embryonic fibroblasts (MEFs) from c-Cbl-deficient mice (KO) compared to the same c-Cbl +/+ regular control strain mice (WT). As in vitro apoptotic inducers, we used the c-Jun kinase-dependent inducer hydrogen peroxide (H 2 O 2 ) and the p53 dependent inducer etoposide.
BimEL expression was up-regulated in KO-MEFs compared with WT ( Figure 2A): we observed a spontaneous increased expression of the pro-apoptotic BH3-only protein in MEFs ( Figure. 2A) as well as in c-Cbl-deficient mice ( Figure 1A) in accordance with Akiyama completed by Wiggins et al. reports [9,31], with Thien et al. [11,12], and as we have reported occurring in mouse testis [10].
The expression level of XIAP and c-IAP1 were similar in KO and WT MEF untreated controls ( Figures 2B and C), whereas they were spontaneously low in c-Cbl deficient mice ( Figure 1D and F). In contrast, the apoptotic activation by hydrogen peroxide or etoposide at the indicated concentrations initiated a significant decrease of the c-IAP1 and XIAP to about one third of the control level solely in KO-MEFs, demonstrating the c-Cbl sensitivity of these caspase-inhibitors ( Figure 2B and C). The spontaneous caspase-3 activation of these cells was almost undetectable in the untreated KO cells compared to WT and slightly higher in the etoposide treated KO cells ( Figure 2D, two first panels on the top). However, a drastically increased expression of activated caspase-3 is seen in KO cells stimulated by hydrogen peroxide: it is more than five times the expression of WT cells when compared to actin expression ( Figure 2D). Experiments with c-Cbl reconstituted cells from c-Cbl deficient mice compared to c-Cbl-deficient cells that were transfected with an empty vector gave very similar results, consistent with c-Cbl being the only cause of the differences between the two cell-types ( Figure 2D, two last panels).
A variety of doses and times of treatments led to a greater number of apoptotic KO-MEFs compared to WT ( Figure 2E) after DAPI count, which were always significant upon hydrogen peroxide treatment These data validate the anti-apoptotic regulation exerted in vivo by c-Cbl, and confirmed its protective cellular effect against oxidative stress, at least in MEFs.

c-Cbl is over-expressed in several human malignant tumours
Tumour cells sustain a major resistance to apoptosis and we reported above that c-Cbl down-regulates cellular apoptosis. Therefore, it could promote cancer survival. We then explored the expression status of c-Cbl in malignant pathologies. Western blots of prostate tumour tissues from six untreated patients compared to healthy surrounding tissue, show a strong c-Cbl cancer expression in all samples that ranged to 6 times normal expression (Figure 3 A). In situ staining in tissue microarrays (TMA) assays confirmed this result (Figure 3 B). We performed in situ staining in numerous TMA assays encompassing cancers from different origins ( Figure 3B to J, right columns) compared to healthy corresponding tissues (left columns), revealing c-Cbl over-expression in numerous cancers. Assay reading was double-blind. Eight of 16 studied non-prostate malignant tumours are positive for c-Cbl, as displayed on Figure 3, panels C to J: ovary serous papillary carcinoma (C), uterus squamous cellular carcinoma (D), striated muscle rhabdomyosarcoma (E), astrocytoma (brain, F), lung squamous carcinoma (G), colon and rectum adenocarcinoma (H, I), and diffuse large B cell lymphoma (J). The tumours considered not positive for c-Cbl are reported in supplementary information ( Figure   SI 3) and in Table 1.
We noticed that though rhabdomyosarcoma is strongly stained, liver carcinoma does not express c-Cbl significantly compared to healthy tissue ( Figure SI 3A). This observation could be related to the key role that c-Cbl plays in energy expenditure regulation in striated muscle but not in liver, as reported by Molero et al. [32,33]. Indeed, it was observed that insulin receptor levels were strongly increased only in the muscle of c-Cbl -/mice when compared to liver and we hypothesize c-Cbl over-expression in cancer affects cellular energy expenditure regulation.
We also observed a high c-Cbl expression in healthy tissues such as epidermis, stomach, and oesophagus ( Figure SI 3B, D and E) that does not support a c-Cbl over-expression in the corresponding carcinomas, though it has been claimed that stomach adenocarcinoma was positive for c-Cbl [34]. It is also notable that c-Cbl alterations have been reported in lung cancers [35] as well as in myeloid malignancies [36][37][38].
Finally, we show that the staining intensity of c-Cbl in prostate increases with the aggressiveness of the malignancy, as also found by Knight et al. [23]. Figure SI 4 A shows a prostatic benign hypertrophy whereas panels B to F represent prostatic adenocarcinoma of increasing grade. All these results strongly suggest that c-Cbl has a positive effect over the growth of several cancers.

c-Cbl up-regulation mostly associates with cellular oxidative stress
We demonstrated that c-Cbl protects cells against the oxidative stress that is often sustained by cancer cells [39], noting that c-Cbl is highly expressed in these malignant cells, as reported above. Therefore, we looked for a correlation between c-Cbl and oxidative stress. We tested 11 cancers with the anti-apurinic/apyrimidic endonuclease APE1/REF1 (TMA staining), reflecting the oxidative stress undergone by the malignant cells [40]. We compare APE1 staining (right columns) to the control of the same tissue (left columns) (Figures 4).    We summarize the results on Table 1

c-Cbl behaves as a protector of oxidative stress in LNCaP cell line
In   behave as do c-Cbl-deficient MEFs. This highlights the anti-apoptotic effect of c-Cbl in prostate cancer, which is particularly noticeable upon hydrogen peroxide treatment. Together, these results strongly suggest that c-Cbl over-expression protects human malignant cells against apoptosis and particularly upon oxidative stress.
We also tested the level of endonuclease expression related to oxidative stress [40] in the LNCaP cell line subjected to transient silencing of c-Cbl by monitoring the anti-APE1/REF1 signal ( Figure  5E). We observed a slight decrease of the anti-APE1 signal in LNCaP submitted to c-Cbl RNA interference compared to the two control cell lines, suggesting that c-Cbl could partly participate to the oxidative stress progress. This result correlates to the result that the presence of weak doses of hydrogen peroxides does not increase LNCaP c-Cbl expression as a negative feed-back regulation ( Figure 5A). Thus, we can make the assumption that c-Cbl participating in generating oxidative stress also allows cells to be protected against this stress, just as it is observed in malignant cells.

c-Cbl down-regulates the ASK1-p38MAPK stress pathway activation
In order to find a molecular mechanism leading c-Cbl to downregulate apoptosis in cancer cells, we explored the expression or activation (phosphorylation) of major kinases of the MAP kinase pathways (schematic view, Figure 6A) known to lead to apoptosis upon oxidative stress [41]. We found that MLK3 of the MAP3K family is expressed approximately two-fold more in untreated LNCaP cell line undergoing c-Cbl silencing than in control cells ( Figure 6B, column 2 compare with 1). ASK1 of the same family is expressed five times more in c-Cbl-KO cells than in control cells ( Figure 6C, column 2 compare with 1). The p38MAPK, which regulates apoptosis and stress responses, does not change its expression ( Figure 6D), whereas we observe a dramatic over-phosphorylation of p38MAPK (Thr180/ Tyr182) in c-Cbl deficient cells ( Figure 6E, column 2 compare with 1), clearly showing an over-activation of the MLK3/ASK1-p38MAPK regulation pathway and allowing apoptotic activation in untreated c-Cbl deficient cells.
Finally, we also observe that knocking down c-Cbl does not seem to have a clear effect on p38 expression when treated with 25 µM hydrogen peroxide ( Figure 6F, H 2 O 2 treatment, column 2 compare with 1). On the contrary, ARN interference with the molecular adaptor Grb-2 (Grb2 siRNA), which allows c-Cbl translocation to the membrane upon TKR activation [1][2][3][4][5], clearly diminishes p38 expression ( Figure  6F, non-treated -NT, column 3 compare with 1). This result shows that Grb2 participates in the apoptotic regulation at least at the level of p38 expression, contrary to c-Cbl, which only down regulates p38 phosphorylation. The double knock-down (c-Cbl siRNA and Grb-2 siRNA) up-regulated p38 expression compared to the Grb-2 knockdown cells (Figure 6 F, NT, column 4 compare with 3), and shows a slight subtractive effect of c-Cbl and Grb-2 over p38 expression. This result highlights the possibly unrelated effect of c-Cbl and Grb2 on p38 expression and growth factor receptor internalisation and regulation. The c-Cbl and Grb-2 RNA interference, single or coupled, were efficient in these experiments as shown in Figure 6G.

Discussion
This work has focused on the role of the proto-oncoprotein c-Cbl in cellular apoptosis in human cancers. We report c-Cbl overexpression in different human carcinomas (ovary, uterus, brain, lung, colon, rectum) and rhabdomyosarcoma (Figure 3), and confirm it in human prostatic carcinoma, demonstrating a correlation between c-Cbl over-expression and cancer aggressiveness ( Figure SI 4) as previously reported by Knight et al. [23]. We also relate the intensity of c-Cbl staining ( Figure 3 and Table 1) and the oxidative stress ( Figure  4) supported by cancer cells. These observations suggest that c-Cbl expression is linked to the modulation of the ROS and to the survival of the cancer cells.
Aggressive tumors are reported to be more subject to oxidative stress [39], consistent with the correlation we report between oxidative stress, intensity of c-Cbl expression, and advanced stage of cancer development. The importance of c-Cbl in tumorigenesis would benefit from assessment of larger numbers of diverse tumors (Table 1).
It has been reported that ROS leads to activation of diverse protein tyrosine-kinases [42,43]. Oxidative stress has been found responsible for dysfunction of c-Cbl in down-regulating EGFR phosphorylation [44]. We can then hypothesize that this dysfunction increases c-Cbl expression by positive feedback. However, LNCaP experiments gave opposite results, showing that upon hydrogen peroxide treatment, c-Cbl expression decreased, following the same expression pattern as the anti-apoptotic factors tested (Bcl-2, c-IAP2, Figure 5A). We also found that the c-Cbl knock-down decreases endonuclease APE1/ REF1, reflecting a decrease of oxidative stress intensity ( Figure 5E). Molero et al. [32,33] showed highly altered energy expenditure and strong increase activity of the AMP-activated protein kinase (AMPK) in the muscle of c-Cbl-deficient mice. They also reported that the uncoupling protein-3 (UCP3) was increased 1.9-fold in muscle in c-Cbl-deficient mice compared to controls. The uncoupling proteins are considered protectors of free radical oxygen species by decreasing the mitochondrial production of reactive oxygen species [45]. Thus, the data of Molero et al. completely accord to the c-Cbl negative feedback we find between c-Cbl expression and cellular oxidative stress as reflected by the monitoring of the anti-APE1/REF1 signal ( Figure  5D) or the c-Cbl expression upon gradually H 2 O 2 concentrations ( Figure 5A). Therefore, we suggest that ROS production in cancer is substantially affected by alteration of c-Cbl expression, which could be itself a result of the peculiar pathway of energy consumption in malignant cells (Warburg effect) [46,47] associated with its protective effect we report on these cells.
Resistance to apoptosis is the hallmark of the majority of cancers [21,22,48,49]. The c-Cbl regulation of apoptosis [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]50,51], highlights its role in cancer. Our study shows that c-Cbl increases the apoptotic threshold in epithelial cells of mouse ventral prostate (VP). The MEF and LNCaP studies demonstrate that the absence of c-Cbl renders the cells much more susceptible to apoptosis when subjected to oxidative stress, but less so upon other means of inducing apoptosis (etoposide in in vitro, flutamide in in vivo). Therefore, this work strongly suggests that the intense expression of c-Cbl together with the high oxidative stress found in several types of cancer confers resistance to apoptosis in malignant cells.
Our results with inhibitors of apoptosis (XIAP and c-IAP1) indicate that c-Cbl has anti-apoptotic potential ( Figure 1D and F; Figure 2B and C), and increases the apoptotic threshold in vivo and in MEFs. The well-documented, increased expression of IAPs in cancer, particularly in high-grade prostate tumours, could be a consequence of the high c-Cbl expression we find in some cancers. Therefore, this increased expression of IAPs could be sufficient to alter caspase activities and protect malignant cells [48,49] Conversely, we report ( Figure 1A and Figure 2A) a spontaneous increased expression of the pro-apoptotic BH3-only protein BimEL (in vivo and ex vivo c-Cbl -/experiments), as well as other pro-apoptotic factors (Bak, Smac/DIABLO and the mitochondrial activated caspase 9). Thien et al. [11,12] showed that a c-Cbl mutant (c-CblA mice) that had lost its ubiquitine-ligase capability was highly pro-apoptotic and led to a markedly elevated activation of the pro-apoptotique factor BimEL. Thien et al.'s results support a key role of c-Cbl ubiquitineligase function in BimEL up-regulation and possibly in the antiapoptotic potential of c-Cbl reported here.
The observation of strong caspase-6 activity combined with the higher apoptotic cell number in the presence of hydrogen peroxide clearly demonstrates anti-apoptotic regulation by c-Cbl in MEFs ( Figure 2D). The LNCaP cell line experiments gave similar results ( Figure 5A). In particular, hydrogen peroxide seems to be more effective in LNCaP cell lines than etoposide with the doses used and allows us to enlarge our MEF conclusions (c-Cbl mainly protects cells again oxidative stress) to the human cancer cell line.
These results seem a paradox: in our hands c-Cbl causes downregulation of apoptosis, while c-Cbl is commonly considered as a downregulator of growth factor receptor, which means an anti-survival factor. Indeed, c-Cbl is reported to increase survival [18,19,52]. Edwin and Patel [18] showed that c-Cbl reduces the anti-apoptotic activity of serum when the c-Cbl negative regulator Sprouty 2 was silenced. Bonaccorci et al. [52] reported that AR affects the clathrin-mediated endocytosis pathway of EGF-R via c-Cbl, which could explain the lower invasive phenotype observed in AR-positive cell lines and the anti-survival effect of c-Cbl. Yan et al. [19] reported that c-Cbl is involved in tamoxifen-induced apoptosis of MCF-7 cell line by downregulating the survival signals (c-Src, ERK and AKT) probably linked to TKR activation. In contrast, Lo et al. [53] showed that the ectopic over-expression of c-Cbl in non-small cell lung cancer (NSCLC) cell lines increases tumour growth and cell migration. Ito et al. [54] report that c-Cbl when linked to EGF Receptor is associated with gastric tumor progression. Interestingly, Sproul et al. [13] reported that c-Cbl interacts with the MLK-JNK pathway regulating negatively neuronal cell death, and Kim et al. [16] showed that the Src/CIN85-c-Cbl complex mediates a biphasic regulation of the pro-apoptotic TRAIL receptor through a MEKK4/p38/Hsp27/Akt pathway. These results re-enforce our own data in vivo and in vitro. Kim et al. also report that c-Cbl knock-down sensitizes the apoptotic effect of TRAIL receptors [15]. These reports and ours clearly suggest that c-Cbl protects cells against apoptosis through mitogen-activated protein (MAP) kinase pathways. We observe that c-Cbl has an anti-apoptotic potential that increases the apoptotic threshold as shown in our prostate cell experiments in vivo. It is likely that in many systems, c-Cbl's down-regulation of growth factors masks this anti-apoptotic aspect. However, in the presence of hydrogen peroxide, the anti-apoptotic potential of c-Cbl manifests and protects malignant cells against high oxidative stress (i.e. advanced stage of cancer), probably as we report here at least through inhibition of the p38MAPK stress pathway.
Indeed, we show that c-Cbl spontaneously down-regulates the expression level of several stress-activated protein kinase (SAPKs) of the MAPK pathways ( Figure 6A to D). We also show that c-Cbl negatively and strongly decreases activation (phosphorylation) of the p38MAPK ( Figure 6E). Indeed, p38 is well-known to be regulated by environmental stress, as such hydrogen peroxide or pro-inflammatory cytokines (IL1β, TNFα), and to be a strong pro-apoptotic factor following oxidative injury [41]. Therefore, c-Cbl has an obvious effect on down-stream signalling. We postulate that such a regulation could be linked to the TKR activation (EGF-R, PDGF-R) that has been shown to undergo atypical phosphorylation/dephosphorylation in response to oxidative insult [44], de-routing c-Cbl activation and releasing its anti-apoptotic potential. Our experiments suggest c-Cbl and Grb-2 have a different effect on p38 expression level and that their effects are subtractive. We speculate that Grb-2 sequesters c-Cbl and masks its anti-apoptotic effect. The paradox between survival effects of c-Cbl and its anti-apoptotic potential would then lie in the balance between the degree of c-Cbl TKR down-regulation (anti-survival) and the intensity of c-Cbl apoptotic inhibition (survival). The cellular redox level could play a determining role in this balance, oxidative stress favouring c-Cbl-linked apoptotic resistance.
Another interesting question emerges with the increased intensity of cellular apoptosis specific to hydrogen peroxide treatment of c-Cbl deficient cell cultures compared to the other apoptotic inductors as etoposide (MEFs) or flutamide (mouse prostate). As reported above [13,16], different specific pathways have been found explaining the c-Cbl apoptotic regulator role. It is then very possible that c-Cbl has different targets and in our hands the most effective one appeared to be the MLK3/ASK1-p38MAPK stress pathway, involving the redox control of cell fate by MAP kinase.
The testosterone dependency of c-Cbl expression that we report (see Supplementary Information) links c-Cbl to the survival process led by AR activation. It clearly increases the apoptotic threshold of androgendependent cells, but it does not specifically protect these cells during testosterone withdrawal: in the in vivo model, the cellular apoptotic WT/KO ratio did not change upon flutamide treatment ( Figure 1I). In LNCaP cells, c-Cbl expression decreases upon high testosterone concentration and is tightly associated to pro-apoptotic behaviour ( Figure SI 5), reflecting the c-Cbl expression pattern observed in vivo ( Figure SI 1 and 2). These points are clarified in the Supplementary Information. It is not clear if the strong decrease of the cell number at high testosterone doses is due to apoptotic onset [28] or to G1 arrest, as reported by Kokontis et al. [55]. G1 arrest could permit cells to escape apoptosis. This aspect could also be related to a large TKR degradation with c-Cbl expression decrease associated to abnormal stimulation of the androgen receptor as described by Bonnacorsi et al. [52].
In summary, c-Cbl down-regulates TKR, therefore it can be seen as an anti-survival factor, and it could also independently control downstream signalling pathway inhibition, such as SAPKs, and thus be seen as a pro-survival factor that is tightly dependent on the redox level. Consequently, it could control the survival/death balance as a complete integrated modulator. Its pro-survival (anti-apoptotic) aspect could be dependent on its ubiquitine-ligase capability, as suggested by the reports of Langdon et al. [11,12], without being directly dependent on the TKR down-regulation by itself. Therefore, we can assume that c-Cbl plays a role in down-regulating SAPKs, possibly through its ubiquitinylation ability [11,12], allowing survival of over-expressing c-Cbl malignant cells by rendering it tolerant to oxidative compounds. Additionally, c-Cbl could also promote the renewal of these compounds as suggested by our data ( Figure 5E) and the reports of Molero et al. [32,33], thereby increasing mutation in cancer cells.