-dependent ubiquitination and degradation of the insulin-like 1

Leonard Girnita, Ada Girnita, and Olle Larsson*

Department of Oncology and Pathology, Division of Cellular and Molecular Tumor Pathology, Cancer Center Karolinska, R8:04, Karolinska Hospital, SE-171 76 Stockholm, Sweden

Edited by Bert Vogelstein, The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD, and approved May 20, 2003 (received for review March 20, 2003) Recently, was demonstrated to affect the expression of the in loss of proteins that mostly is the case with other tumor insulin-like growth factor 1 receptor (IGF-1R), a receptor suppressor proteins. Quite the opposite, cancer cells usually kinase that plays a crucial role in growth and survival of cancer accumulate the mutant and even the wild-type p53 proteins (12). cells. However, the underlying mechanisms for interaction be- The mechanisms of gain-of-function are still poorly understood. tween p53 and IGF-1R are still not fully understood. One of the The p53 protein level is principally regulated by its interaction challenging questions remaining to be answered is why the wild- with Mdm2 by means of a regulatory feedback loop. Mdm2 type p53, which per se represses the transcription of the IGF-1R ubiquitinates both p53 and itself, contributing to the rapid gene, in overexpressed form is necessary for a high IGF-1R expres- turnover of both proteins (13). Like p53, Mdm2 shuttles from the sion. In this study, we show that inhibition of p53 causes ubiquiti- nucleus to the cytoplasm (14), and the shuttling of Mdm2 may nation and down-regulation, through increased degradation, of be important for p53 export in some cells. the IGF-1R in human malignant melanoma cells. This effect, which Werner et al. (6) have shown that wild-type p53 represses the was independent of the p53 status (i.e., wild type or mutated), was transcription of the IGF-1R gene, whereas mutant p53 had the prevented if Mdm2 was coinhibited. Similar results were obtained opposite effect. These studies were performed on p53-negative in UV-irradiated human melanocytes (harboring wild-type p53), in cells transfected with wild-type or mutant p53 cDNA, using which level of the IGF-1R increased after up-regulation of p53. reporter gene strategy. Interestingly, the basal ubiquitination of the IGF-1R in untreated Recently, we demonstrated that treatment with antisense cells also depended on Mdm2. We could prove that Mdm2 physi- oligodeoxynucleotides for p53 leads to down-regulation of the cally associates with IGF-1R and that Mdm2 causes IGF-1R ubiq- IGF-1R at the plasma membrane of malignant melanoma cells uitination in an in vitro assay. Taken together our data provide (8). Unexpectedly, however, the same response was obtained in evidence that Mdm2 serves as a ligase in ubiquitination of the melanoma cells overexpressing wild-type p53. It should be added IGF-1R and thereby causes its degradation by the proteasome here that the incidence of p53 mutations in malignant melanoma system. Consequently, by sequestering Mdm2 in the cell nuclei, the is very low, whereas p53 overexpression is common in this level of p53 may indirectly influence the expression of IGF-1R. This disease (15, 16). Even though the mechanisms underlying the role of Mdm2 and p53 represents an unexpected mechanism for effect of p53 on IGF-1R expression are not fully known, these the regulation of IGF-1R and cell growth. data suggest that posttranscriptional events might be involved in p53-dependent regulation of IGF-1R (8). cell growth ͉ p53 The aim of the present study was to search for posttranscrip- tional events involved in p53-dependent expression of IGF-1R. Our results provide evidence that p53͞Mdm2 is involved in he insulin-like growth factor 1 receptor (IGF-IR) has been ubiquitination and degradation of the IGF-1R. Tshown to be crucial for tumor transformation, maintenance of tumorigenicity, promotion of cell growth, and prevention of Materials and Methods apoptosis (1–3). IGF-1R is often overexpressed in malignant Reagents. A mouse monoclonal antibody against the human tumors (1–3). All of these circumstances make the IGF-1R an IGF-1R was purchased from Science. Polyclonal intriguing target in cancer. IGF-IR is a heterotetrameric recep- ␣ IGF-1R antibodies (N-20, C-20, and H-60), a mouse monoclonal tor composed of two extracellular -subunits, antibody against human p53 (DO1), a mouse monoclonal anti- which are involved in binding, and two transmembrane body to Mdm2 (including the p53-Mdm2 complex), a monoclo- ␤ -subunits that contain tyrosine kinase domains involved in nal antibody to phosphotyrosine (PY99), and an antibody to (4). actin (H-196) were from Santa Cruz Biotechnology. The pro- CELL BIOLOGY The mechanisms behind up-regulation of IGF-1R in tumor teasome inhibitor MG 132 was from Calbiochem. All other cells are still poorly understood. However, several studies during reagents unless stated otherwise were from Sigma. the last 8 years have reported intriguing connections between the IGF-1R and the p53 pathways (5–8). The actions of the IGF-1R Cell Lines. The human melanoma cell lines SK-MEL-5 and and p53 pathways are distinctly opposite. Whereas IGF-1R SK-MEL-28 were from American Type Culture Collection. The promotes mitogenic and antiapoptotic signals, wild-type p53 melanoma cell lines BE and DFB were provided by Rolf induces cell-cycle arrest and apoptosis. Kiessling (Karolinska Hospital). The RϪ and P6 mouse cell lines Along with the diversity of genetic modifications in cancer, were gifts from Renato Baserga (Thomas Jefferson University, p53 abnormalities are probably the most prevalent defects (9). Philadelphia). The RϪ fibroblasts are IGF-1R negative and are The roles played by p53 in cancer can be greatly simplified in derived from an IGF-1R knockout mouse embryo (17). The P6 three main categories: (i) loss of normal p53 functions, mainly line is a 3T3 derivative that overexpresses the human IGF-1R related to the role of p53 as a and acquired (17). The cells were cultured in DMEM supplemented with 10% by mutations; (ii) dominant negative effect of mutant p53, abrogation of wild-type p53 as a result of heteromerization with mutant p53; and (iii) gain of function, the ability of mutant p53 This paper was submitted directly (Track II) to the PNAS office. to acquire new functions (10, 11). However, an interesting Abbreviation: IGF-IR, insulin-like growth factor 1 receptor. observation on p53 mutations is that most of them do not result *To whom correspondence should be addressed. E-mail: [email protected].

www.pnas.org͞cgi͞doi͞10.1073͞pnas.1431613100 PNAS ͉ July 8, 2003 ͉ vol. 100 ͉ no. 14 ͉ 8247–8252 Downloaded by guest on September 30, 2021 (RϪ) or 5% (P6) FBS. P6 and RϪ cell lines were cultured in the whereupon the pellet was washed and then dissolved in a sample presence of G-418 (Promega). Neonatal epidermal melanocytes buffer for SDS͞PAGE. were from Clonetics (San Diego) and cultured in melanocyte cell growth medium-2 supplemented (for 500 ml) with 7.5 mg͞ml Analysis of IGF-1R Synthesis and Degradation. After indicated ex- bovine pituitary extract, 0.5 ␮g of human recombinant fibroblast perimental procedures, cells were transferred to methionine- growth factor, 5 ␮g of phorbol myristate acetate, 0.25 mg of free DMEM supplemented with 10% FBS and 100 ␮Ci͞ml hydrocortisone, 2.5 mg of insulin, 25 mg of gentamycin, 25 ␮g͞ml L-[35S]methionine (specific activity Ͼ1,000 Ci͞mM, Amersham Amphotericin-B, and 2.5 ml of FBS according to the manufac- Pharmacia) for 6 or 24 h. To determine IGF-1R synthesis, the turer’s protocol. cells were quickly washed twice with ice-cold PBS and lysed in RIPA buffer (1 ϫ PBS͞1% Triton-X-100͞0.5% sodium deoxy- Antisense Experiments. The antisense experiments were essen- cholate͞0.1% SDS), supplemented with protease inhibitor tab- tially performed as described (8). Antisense and sense oligode- lets (Roche Diagnostics). An equal amount of protein from each oxynucleotides were purchased from Interactiva (Ulm, Germa- sample was immunoprecipitated with antibodies for the IGF-1R ␤ ny). The sequences of the oligodeoxynucleotides for p53 and -subunit (H-60) collected by protein A Sepharose (CL-4B, ͞ human and murine Mdm2 have been described (8, 18, 19). All Amersham Pharmacia), resolved by SDS PAGE, and visualized oligodeoxynucleotides used were phosphorothiolated to protect by autoradiography. them from the cell nucleases. Lipofectin (Life Technologies, IGF-1R degradation was determined by pulse–chase experi- 35 Grand Island, NY) was used to deliver antisense oligode- ments. The cells were, after the labeling with [ S]methionine oxynucleotides to the cultured cells. Because antisense oligode- (see above), carefully washed and transferred to radioactive-free oxynucleotides induce RNase H cleavage and further degrada- DMEM containing 10% FBS for the indicated time periods. tion of target mRNA, we tested the specificity of the antisense Cells were then harvested for detection of radioactive IGF-1R as oligodeoxynucleotides using semiquantitative RT-PCR as de- described above. scribed (8). IGF-1R͞Mdm2 Interaction in Vitro. Mdm2 glutathione Sepharose ͞ beads were mixed with total protein extracts (500 ␮g) from SDS PAGE and Western Blotting. Protein samples were dissolved in Ϫ a sample buffer containing 0.0625 M Tris⅐HCl (pH 6.8), 20% P6 or R cells. After 60 min of incubation at room temperature, glycerol, 2% SDS, bromphenol blue, and DTT. Samples corre- the beads were washed three times with PBS, dissolved in sponding to 50–100 ␮g of cell protein were analyzed by SDS͞ SDS sample buffer, loaded on a 7.5% gel, and visualized after PAGE with a 7.5% or 10% separation gel. Molecular weight transfer to a nitrocellulose membrane with an anti-IGF-1R antibody (C20). markers (Bio-Rad) were run simultaneously. After SDS͞PAGE, the proteins were transferred overnight to nitrocellulose mem- In Vitro Ubiquitination. In vitro ubiquitination of IGF-1R was branes (Amersham Pharmacia) and then blocked for1hatroom performed essentially as described (23). Recombinant GST- temperature in a solution of 5% (wt͞vol) skimmed milk powder Mdm2 was expressed in Escherichia coli and purified by using and 0.02% (wt͞vol) Tween 20 in PBS, pH 7.5. Incubation with glutathione-Sepharose (Pierce). IGF-1R was isolated from P6 appropriate primary antibodies was performed for1hatroom cells by immunoprecipitation with a polyclonal rabbit antibody temperature. This was followed by washes with PBS and incu- directed against ␤-subunit (H60) and protein G-Sepharose bation with a biotinylated secondary antibody (Amersham Phar- (Amersham Pharmacia). IGF-1R Sepharose beads were mixed macia) for 1 h. After incubation with streptavidin-labeled horse- with or without GST-Mdm2, rabbit E1 (Calbiochem), E2 bac- radish peroxidase, the detection was made (Hyperfilm-ECL, terial recombinant UbcH5B (Calbiochem), and His-6- Amersham Pharmacia). The films were scanned by Fluor-S (Calbiochem) in a 30-␮l reaction. After a 1-h incubation period (Bio-Rad). at 37°C, the reaction was stopped by addition of SDS sample buffer. Reaction products were loaded on a 7.5% polyacrylamide Assay of Cell Growth and Survival. We performed the determina- gel, transferred to nitrocellulose membrane, and detected by tions with the cell proliferation II (Roche Diagnostics), which using either antibody against IGF-1R (C20) or an anti-His tag is based on the colorimetric change of the yellow tetrazolium salt antibody. XTT to orange formazan dye by the respiratory chain of viable cells (20). All standards and experiments were performed in Results triplicates. Effects of p53 and Mdm2 Inhibition on IGF-1R and Cell Growth. Two human melanoma cell lines harboring wild-type p53 (MEL-5 and UV Light Treatment. Cells were treated with UV light essentially DFB) and two with hot spot p53 mutations (BE and MEL-28), as described (21). Before UV irradiation, cells were washed all of which exhibit high IGF-1R expression (8), were treated twice with PBS, irradiated through a thin film of PBS, and refed with antisense oligodeoxynucleotides to p53 (ASp53), Mdm2 with their own medium. Cells were exposed through the cover of (ASMdm2) or both. The detailed conditions for optimization of the dish, filtering out residual UVC. The UV source was a bank the antisense experiments, including all necessary controls, have of two Philips TL 20W01 tubes with a peak output Ϸ310 nm been described (8, 24). The effects on protein levels of p53, (UVB). This spectrum represents the solar radiation spectrum Mdm2, and IGF-1R ␤-subunit detected by Western blots for that most actively induces genotoxic and carcinogenic effects. MEL-5 and MEL-28 are shown in Fig. 1A, and densitometry data The dose used for UVB was 5 J͞m2 and was optimized to induce (for quantification) of all four cell lines are shown in Fig. 1B. maximal p53 expression with Ͻ10% decrease in cell viability 24 h Consistent with our previous study (8), disruption of p53 ex- after UV exposure. pression caused IGF-1R down-regulation in all four cell lines. However, Mdm2 inhibition also drastically decreased IGF-1R, Immunoprecipitation. The isolated cells were lysed as described but only in cells harboring wild-type p53. This finding is com- (22). Fifteen microliters of protein G plus-A͞G agarose and 1 ␮g pletely consistent with the notion that wild-type p53, no longer of antibody were added to 1 mg of protein material. After inactivated by Mdm2, decreases transcription of the IGF-1R overnight incubation at 4°C on a rocker platform, the immuno- gene (6, 8). Surprisingly, combined treatment with ASp53 and precipitates were collected by centrifugation in a microcentri- ASMdm2 substantially decreased the effect of p53 inhibition on fuge at 2,500 rpm for 2 min. The supernatant was discarded, IGF-1R (strongest in the p53 wild-type cells), although the

8248 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.1431613100 Girnita et al. Downloaded by guest on September 30, 2021 Fig. 2. Effect of UV irradiation on IGF-1R expression of cultured melanocytes and effect of inhibition of p53 and Mdm2. Control cells or cells treated with ASp53 (1 ␮M) or ASMdm2 (1 ␮M) were exposed to 5 J͞m2 UVB. Western blotting was performed to measure IGF-1R, and quantification was made by densitometry. The p53 levels (also assayed by Western blotting and densitom- etry) of irradiated control (not treated with AS) cells are indicated in the graph. For further details, see Materials and Methods. The experiment was repeated twice with similar results.

hypothesis that p53-dependent control of the IGF-1R may involve the action of Mdm2.

Effect of p53 on IGF-1R in UV-Irradiated Cells. We also studied the effect of p53 and Mdm2 abrogation on IGF-1R expression in UV-irradiated human melanocytes (harboring wild-type p53). Fig. 2 shows that6haftertheirradiation the level of IGF-1R increased substantially. Inhibition of p53 expression, which increased immediately after irradiation, prevented the IGF-1R up-regulation. However, upon coinhibition of Mdm2, the inhib- itory effect of the p53 down-regulation on IGF-1R expression was deleted. Fig. 1. Effect of p53 and Mdm2 inhibition on IGF-1R and cell growth on malignant melanoma cells. (A and B) The indicated cell lines either remained Effect on IGF-1R Degradation. We investigated the effects of p53 or untreated (C) or were treated with Lipofectin (Lipofectin control, L), L ϩ ASp53 Mdm2 inhibition (or both), obtained by antisense oligode- (1.0 ␮M), L ϩ ASMdm2 (2.0 ␮M), or L ϩ ASp53 ϩ ASMdm2 and with corre- oxynucleotides, on the degradation of IGF-1R. This was per- sponding sense oligodeoxynucleotides (Sp53 and SMdm2) as indicated. After formed on the DFB (p53 wild type) and BE (p53 mutant) cell incubations for 24 h, the protein expression of p53, Mdm2, IGF-1R, and actin lines by using pulse–chase experiments with [35S]methionine. (loading control) was determined. Western blots for MEL-5 and MEL-28 are Fig. 3A shows the autoradiographic detections of the IGF-1R shown in A, and densitometry data (expressed as % of Lipofectin control) for ␤-subunit, and Fig. 3B shows the densitometry data obtained all four cell lines are shown in B.(C) All four cell lines were treated as described in A but for 48 h. The amounts of surviving cells were determined by the XTT from the autoradiographs. The DFB control cells exhibited 22% assay. The values are means and SDs of triplicates. It was confirmed in all and 45% degradation of IGF-1R after 12 and 24 h, respectively, experiments that Sp53 and SMdm2 were without effects (not shown in B and whereas radiolabeled IGF-1R in the BE was slightly less de- C). The experiments were repeated three to four times with similar results. creased (18% and 30%, respectively) (Fig. 3B). The correspond-

ing values for cells treated with ASp53 were 35% and 78%, and CELL BIOLOGY 31% and 79%, respectively, whereas there was no increased protein levels of both p53 and Mdm2 remained adequately degradation (rather the opposite in DFB), compared with the decreased (Fig. 1A). Similar results were obtained in all four cell controls, upon inhibition of Mdm2 and, notably, after coinhibi- tion of Mdm2 and p53. The relatively late increase in IGF-1R lines (Fig. 1B). degradation by p53-inhibited cells is most likely due to the We also tested the effects of the above conditions on cell delayed increase in the inhibitory effect on the p53 protein. First growth and survival of the four cell lines (Fig. 1C). As shown, after 12 h of incubation with antisense oligodeoxynucleotides, both ASp53 and ASMdm2 inhibited growth of the two p53 there was an Ϸ50% decrease in the p53 protein levels (data not wild-type cell lines (DFB and MEL-5) after 48 h of incubation, shown), which is consistent with other studies using ASp53 (25). whereas upon cotreatment, cell growth was hardly affected. The As demonstrated in Fig. 3A Lower, the ASp53-induced increase mutant cell lines (BE and MEL-28) responded similarly to p53 (after a 24-h treatment) in IGF-1R degradation was largely inhibition as the p53 wild-type cells did, but only marginally to counteracted by the 26S proteasome inhibitor MG 132, which ASMdm2. As was the case with wild-type cells, the combined was present during the last6hoftheincubations in separate treatment regimen significantly decreased the growth inhibitory experiments. effect of ASp53 (Fig. 1C). Thus, the effects of the AS treatments We also assessed de novo synthesis of the IGF-1R by using on IGF-1R expression and cell growth correlate well with each pulse labeling with [35S]methionine. After a 6-h pulse labeling, other (compare Fig. 1 B and C). These data support the we could detect incorporation of [35S]methionine into the pro-

Girnita et al. PNAS ͉ July 8, 2003 ͉ vol. 100 ͉ no. 14 ͉ 8249 Downloaded by guest on September 30, 2021 Fig. 3. Effect of p53 and Mdm2 inhibition on IGF-1R degradation. (A) Wild-type (DFB) and p53 mutant (BE) cells were labeled with [35S]methionine (100 ␮Ci͞ml) for 24 h in methionine-free medium, and thereafter transferred to radioactive-free methionine-supplemented medium containing ASp53, ASMdm2, or both, or corresponding Ss for the indicated times. In separate ␮ dishes, cells incubated for 24 h were treated with MG 132 (50 M) during the Fig. 4. Ubiquitination of IGF-1R and association with Mdm2. (A) DFB cells ␤ last6h(Bottom). IGF-1R -subunit was immunoprecipitated and resolved by either remained untreated or were treated with ASp53, ASMdm2, or both as ͞ SDS PAGE and finally visualized by autoradiography. It was confirmed that indicated for 12 h, after which the proteasome inhibitor MG 132 (50 ␮M) was the sense oligodeoxynucleotides (Sp53 and SMdm2) were without effects (not added, or not, for an additional 6 h. IGF-1R was immunoprecipitated by a shown). (B) Densitometry data are shown. The results are representative of ␤-subunit antibody. Equal amounts of purified IGF-1R immunoprecipitates (to three independent experiments. compensate for decrease in ASp53-treated cells), verified in the lower panel, were resolved by SDS͞PAGE and finally detected by Western blotting using an antibody to ubiquitin. (B Upper) Total proteins isolated from the indicated cell receptor, in line with the study of Sepp-Lorenzino et al. (26). lines were immunoprecipitated with an IGF-1R ␤-subunit antibody and im- Pretreatment with ASp53 did not change this incorporation munoblotted by an Mdm2 antibody, or vice versa. (B Lower Left) BE cells were (data not shown), indicating that the inhibition of p53 does not pretreated with ASMdm2 for 12 h, and the proteasome inhibitor MG 132 was affect de novo synthesis of the IGF-1R. added, or not, for an additional 6 h before coimmunoprecipitation. (B Right) Total protein lysates from P6 and RϪ cells were mixed with Sepharose beads Ubiquitination of IGF-1R and Association with Mdm2. Degradation of of recombinant Mdm2 (see Materials and Methods). The Mdm2 beads were analyzed by Western blotting using an IGF-1R ␤-subunit antibody. (C Upper) signal-transducing cell-surface receptors is mediated by inter- UV-irradiated human melanocytes were treated as indicated for 6 h. There- nalization and endocytosis. Internalized receptors can either be after, analysis of IGF-1R was performed as described for the experiments on transported to the , where they are degraded, or be the melanoma cell lines. (C Lower) UV-irradiated melanocytes were treated as recycled to the plasma membrane (27). Recent experimental indicated for 6 h. Immunoprecipitation using an IGF-1R ␤-subunit antibody data have identified the ubiquitin proteasome pathways as a and immunoblotting with an Mdm2 antibody was then done. The experi- regulatory system for endocytosis (27–30). Ubiquitin is a ments were repeated two to four times with similar results. polypeptide, 76 aa in length, playing many cellular functions (29). Ubiquitination of proteins requires the action of three enzymes: (i) ubiquitin-activating enzyme (E1), which binds to ubiquitin to defined as an E3 ligase, and moreover its involvement in generate a high energy E1-ubiquitin intermediate; (ii) ubiquitin- ubiquitination of some membrane proteins has been clearly conjugating enzyme (E2), a ubiquitin carrier protein; and (iii)a demonstrated (34), it would be attractive to investigate whether that transfers the ubiquitin to the target protein Mdm2 associates with and ubiquitinates the IGF-1R and thereby (31, 32). E3 plays a key role in the ubiquitin-mediated pathways triggers its degradation. because it serves as the specific recognition factor. Based on the idea that IGF-1R may be ubiquitinated and By considering the data in Figs. 1–3, it can be hypothesized degraded after inhibition of p53, we have now investigated that inhibition of p53 expression disrupts the balance between whether treatment with ASp53 can induce ubiquitination of the Mdm2 and p53; Mdm2 is no longer sequestered by p53 and may IGF-1R. DFB cells were either untreated or transfected with therefore exert others functions (33). Because Mdm2 has been antisense oligodeoxynucleotides for p53 or Mdm2 for a period of

8250 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.1431613100 Girnita et al. Downloaded by guest on September 30, 2021 18 h, with and without MG 132 present during the last 6 h. The ␤-subunit of IGF-1R was isolated by immunoprecipitation and analyzed by Western blotting using antibodies for ubiquitin. Fig. 4A shows that the IGF-1R ␤-subunit from ASp53- and 26S proteasome inhibitor-treated cells exhibits strong ubiquitin sig- nals at sizes of 85–185 kDa. The high molecular smear-like bands are typical for multiubiquitinated proteins (30). This implies that ubiquitin is bound to the ␤-subunit of the IGF-1R and suggests that the receptor degradation is proteasome-dependent. How- ever, upon coinhibition of p53 and Mdm2, there were no visible IGF-1R ubiquitination. Inhibition of only Mdm2 did not increase the amount of ubiquitinated IGF-1R compared with the con- trols, which in fact exhibited weak yet detectable ubiquitin signals (Fig. 4A). Rather, treatment with ASMdm2 reduced the baseline ubiquitination of the IGF-1R ␤-subunit. The results indicate that Mdm2 may be involved in IGF-1R ubiquitination. Next, we investigated whether Mdm2 could be physically associated with the IGF-1R. For this purpose, we first immu- noprecipitated the IGF-1R ␤-subunit or Mdm2 from protein extracts derived from DFB and BE cells, a cell line overexpress- ing IGF-1R (P6), and IGF-1R negative cells (RϪ) (serving as a negative control). We then probed with antibodies to Mdm2 or to the IGF-1R ␤-subunit, respectively. Irrespective of which antibody was used in the immunoprecipitation step, our data Fig. 5. Mdm2-dependent ubiquitination of IGF-1R. (A Left) P6 cells either prove that the Mdm2 and IGF-1R ␤-subunit associates with each remained untreated or were treated with ASMdm2 for 18 h, with or without the proteasome inhibitor during the last 6 h. Equal amounts of purified other (Fig. 4B Upper). The same results were obtained from all immunoprecipitated IGF-1R ␤-subunit were analyzed by Western blotting of the three IGF-1R positive cell lines. In a separate experiment, using a ubiquitin antibody. The OD values for ubiquitinated IGF-1R are shown it was confirmed that ASMdm2 deleted the complex formation in A Lower.(A Right) UV-irradiated melanocytes either remained untreated or of IGF-1R and Mdm2 (Fig. 4B Lower Left). Consistent results were treated with ASMdm2 as indicated for 6 h, whereupon immunoprecipi- were also obtained in a cell-free͞antibody-free system, using tation and Western blot using ubiquitin antibodies were carried out. (B) In recombinant Mdm2 beads to extract IGF-1R from P6 total vitro ubiquitination of IGF-1R. Sepharose beads of IGF-1R, isolated from P6, protein lysate (Fig. 4B Right). We repeated the coimmunopre- were mixed with ubiquitin reagents with or without GST-Mdm2 as described cipitation experiments using different antibodies several times in Materials and Methods. Equal amounts of purified IGF-1R immunoprecipi- ͞ with similar results. tates (verified in Lower) were resolved by SDS PAGE and finally detected by Western blotting using an antibody to His-tagged ubiquitin. The results are As was already shown in Fig. 2, inhibition of p53 in UV- representative of three independent experiments. irradiated melanocytes resulted in deletion of the increase in IGF-1R expression. However, this effect was neutralized by Mdm2 coinhibition. In Fig. 4C Upper, the IGF-1R levels at 6 h (cell-free) preparations by using a ubiquitin assay (Fig. 5B). As after UV irradiation are shown. This also involves the effects of is clearly shown, the presence of Mdm2 (together with E1 and ASp53 and ASMdm2, or both. The results are similar to those E2) in the reaction results in the ubiquitination of the IGF-1R found in melanoma cells harboring wild-type p53 (MEL-5, DFB) ␤-subunit. (compare Fig. 1 A and B). In Fig. 4C Lower, it is demonstrated ␤ ͞ By comparing the in vitro ubiquitination of IGF-1R -subunit that IGF-1R is associated with Mdm2. The IGF-1R Mdm2 (in Fig. 5B) with IGF-1R ubiquitination of cultured cells (espe- complex was increased in UV-irradiated cells, which is explained cially in Fig. 5A), weak 85- to 95-kDa bands can be distinguished by the up-regulation of IGF-1R (Fig. 4C Upper), but also by an in the latter conditions. These bands do not appear in the in vitro increase in Mdm2, which has been shown to reach a maximal ubiquitination. Presumably, they represent a fraction of mono- level6hafteraUVexposure (35). When the cells were treated or oligo-ubiquitinated IGF-1R. Proteins being mono-ubiquiti- with ASMdm2, there were almost no visible IGF-1R–Mdm2 nated or ubiquitinated with less than four molecules are not complexes. degraded by the proteasome. To trigger proteasome degrada- Taken together, the results presented in Fig. 4 demonstrate ␤ tion, the targeted proteins must be multiubiquitinated (27). that Mdm2 is associated with the IGF-1R -subunit and may be CELL BIOLOGY involved in ubiquitination of IGF-1R. Thus, in the cell systems it might be a relative overrepresentation of mono- and͞or oligo-ubiquitinated proteins because multiu- Mdm2-Dependent Ubiquitination of IGF-1R. Fig. 5A Upper Left biquitinated proteins are degraded. The reason mono- or oligo- demonstrates the effect of Mdm2 silencing on the basal ubiq- ubiquitinated IGF-1Rs do not occur in the cell-free system could uitination of IGF-1R in P6 cells (which overexpress IGF-1R). be explained by the excess of ubiquitin reagents (allowing full Compared with other cells, basal IGF-1R ubiqutination is much ubiquitination) as well as the absence of the proteasome in the higher in this cell line. The corresponding phenomenon was also in vitro preparations. shown in cells overexpressing the ␤2-adrenergic receptor (34). Collectively, our findings show that Mdm2 associates with and As shown, Mdm2 inhibition, induced by antisense oligode- ubiquitinates the IGF-1R. This points to an important function oxynucleotides, causes a drastic decrease (75–80%) in the of p53͞Mdm2 in regulation of this . This amount of ubiquitinated IGF-1R (Fig. 5A Lower Left). In Fig. 5A function seems not to be limited to only malignant cells but may Right, it is shown that the IGF-1R from UV-treated melanocytes also play a regulatory role in IGF-1R regulation in normal cells. is ubiquitinated. When the cells have been treated with antisense oligodeoxynucleotides to Mdm2, this ubiquitination does not Discussion occur. Our findings provide strong evidence that the oncoprotein To confirm the involvement of Mdm2 in ubiquitination of Mdm2 serves as a ligase (ligase E3) in the ubiquitination of IGF-1R, we have investigated the effect of Mdm2 in vitro IGF-1R. First, we could demonstrate a physical association of

Girnita et al. PNAS ͉ July 8, 2003 ͉ vol. 100 ͉ no. 14 ͉ 8251 Downloaded by guest on September 30, 2021 IGF-1R to Mdm2; second, inhibition of p53 expression with accentuate the nuclear Mdm2 localization. For cells expressing maintained expression of Mdm2 led to ubiquitination of IGF- wild-type p53, this process would affect p53 activity. 1R; third, coinhibition of Mdm2 expression rescued the cells In a broader term, our data point to an intriguing p53- from ubiquitination and down-regulation of IGF-1R; fourth, dependent mechanism for IGF-1R regulation. The p53͞Mdm2- addition of Mdm2 together with IGF-1R in an in vitro ubiquitin dependent IGF-1R ubiquitination might decide the fate of the assay resulted in ubiquitination of IGF-1R. receptor by controlling receptor internalization and the endo- It is well known that Mdm2 does not exclusively interact with somal sorting process. Such a scenario has in fact been observed p53. Mdm2 has been demonstrated to bind to Rb and E2F, with in epidermal growth factor receptors, in which the putative E3 resulting increase in E2F gene transactivation (36). Mdm2 also ubiquitin ligase c-Cbl is involved in receptor ubiquitination (36). associates with ubiquitin conjugase-like protein TSG101, which Because of various conditions, the ErbBs are either recycled or has been found to degrade rapidly under conditions when the level degraded after ubiquitination by c-Cbl (36). In the case of IGF-1R, of Mdm2 is enhanced (37). There is also evidence that Mdm2 Mdm2-dependent ubiquitination may direct the receptor to the degradation pathways instead to the recycling pathways. increases degradation of other proteins (36). Recently, it was An interesting question regarding ligand-induced ubiquitina- demonstrated that Mdm2 can ubiquitinate and cause proteasomal ␤ ␤ tion of plasma membrane receptors is whether this modification degradation of -arrestin and the 2-adrenergic receptor (34). induces lysosomal versus proteasomal degradation. Degradation It is tempting to speculate that Mdm2 may control the switch of several mammalian receptors, known to be ubiquitinated, is between the growth arrest and apoptotic signals of p53 and the impaired by inhibitors of proteasome as well as by agents cell-cycle progression and antiapoptotic signals of IGF-1R. This blocking the lysosomal degradation (31, 32). It is possible that a ͞ switch may depend on the Mdm2 location and or posttransla- fraction of these receptors is degraded by proteasome, whereas tional modification. Probably, the major function of nuclear another fraction is degraded by the . Alternatively, the Mdm2 is to control the p53 activity, a function that depends proteasome and lysosome might destroy different parts of the strictly on protein–protein interactions. After detachment from receptor. A third possibility is that the proteasome mediates p53, Mdm2 may shuttle back to the cytosol, where it may be degradation of another protein, which in turn is required for an degraded by the proteasome or be available for other tasks. efficient targeting of the receptor to the lysosome (31, 32). Alternatively, mutant type p53 might sequester Mdm2 into the The findings of the present paper suggest a posttranscriptional cytoplasm. However, the distribution between the nucleus and p53 activity that is grossly exaggerated in cancer cells. Using systems cytosol is probably strictly controlled, rendering the cytosol with a strong expression of p53, i.e., malignant melanoma and concentration comparably low (33). The Mdm2 concentration in UV-irradiated cells, we show that the high p53 level (independent the cytosol depends on its synthesis͞degradation rate, and on the of whether p53 is mutated or wild type) is necessary to maintain or phosphatidylinositol-3 kinase-Akt pathway and its interaction induce a high expression of the IGF-1R, which in turn favors with p53 (33). The balance among all of these processes may survival of both tumor cells and irradiated cells (37). determine the amount of cytoplasmic Mdm2 that is available for the IGF-1R. A massive p53 expression would deplete the We thank Dr. Galina Selivanova, Karolinska Institute, for generously supplying the MDM-GST construct. We also thank Drs. Rolf Lewensohn cytoplasmic Mdm2 pool, and as a consequence of this increase, and Ulf Wester at the Swedish Radiation Protection Authority for help with the expression of IGF-1R would be enhanced due to decreased the UV source. This study was supported by grants from the Swedish Cancer receptor degradation. Furthermore, the phosphatidylinositol-3 Society, the Cancer Society in Stockholm, the Jubilee Fund of King Gustaf kinase-Akt activation after increased IGF-1R expression would V, the Swedish Children Cancer Society, and the Karolinska Institute.

1. Baserga, R. (1995) Cancer Res. 55, 249–252. 20. Roehm, N. W., Rodgers, G. H., Hatfield, S. M. & Glasebrook, A. L. (1991) 2. Baserga, R., Resnicoff, M. & Dews, M. (1997) Endocrine 7, 99–102. J. Immunol. Methods 142, 257–265. 3. Werner, H. & Le Roith, D. (1997) Crit. Rev. Oncog. 8, 71–92. 21. Maltzman, W. & Czyzyk, L. (1984) Mol. Cell. Biol. 4, 1689–1694. 4. Ullrich, A., Gray, A., Tam, A. W., Yang-Feng, T., Tsubokawa, M., Collins, C., 22. Girnita, L., Wang, M., Xie, Y., Nilsson, G., Dricu, A., Wejde, J. & Larsson, O. Henzel, W., Le Bon, T., Kathuria, S., Chen, E., et al. (1986) EMBO J. 5, 2503–2512. (2000) Anticancer Drug Des. 15, 67–72. 5. Buckbinder, L., Talbott, R., Velasco-Miguel, S., Takenaka, I., Faha, B., 23. Fang, S., Jensen, J. P., Ludwig, R. L., Vousden, K. H. & Weissman, A. M. (2000) Seizinger, B. R. & Kley, N. (1995) Nature 377, 646–649. J. Biol. Chem. 275, 8945–8951. 6. Werner, H., Karnieli, E., Rauscher, F. J. & LeRoith, D. (1996) Proc. Natl. Acad. 24. Xie, Y., Skytting, B., Nilsson, G., Gasbarri, A., Haslam, K., Bartolazzi, Sci. USA 93, 8318–8323. A., Brodin, B., Mandahl, N. & Larsson, O. (2002) Cancer Res. 62, 3861– 7. Prisco, M., Hongo, A., Rizzo, M. G., Sacchi, A. & Baserga, R. (1997) Mol. Cell. 3867. Biol. 17, 1084–1092. 25. Hirota, Y., Horiuchi, T. & Akahane, K. (1996) Jpn. J. Cancer Res. 87, 735–742. 8. Girnita, L., Girnita, A., Brodin, B., Xie, Y., Nilsson, G., Dricu, A., Lundeberg, 26. Sepp-Lorenzino, L., Ma, Z., Lebwohl, D. E., Vinitsky, A. & Rosen, N. (1995) J., Wejde, J., Bartolazzi, A., Wiman, K. G. & Larsson, O. (2000) Cancer Res. J. Biol. Chem. 270, 16580–16587. 60, 5278–5283. 27. Hicke, L. (1999) Trends Cell Biol. 9, 107–112. 9. Prives, C. & Hall, P. A. (1999) J. Pathol. 187, 112–126. 28. Hicke, L. (2001) Nat. Rev. Mol. Cell Biol. 2, 195–201. 10. Blagosklonny, M. V. (2000) FASEB J. 14, 1901–1907. 29. Hicke, L. (1997) FASEB J. 11, 1215–1226. 11. Cadwell, C. & Zambetti, G. P. (2001) Gene 277, 15–30. 30. Shih, S. C., Sloper-Mould, K. E. & Hicke, L. (2000) EMBO J. 19, 187–198. 12. Oren, M., Damalas, A., Gottlieb, T., Michael, D., Taplick, J., Leal, J., Maya, 31. Bonifacino, J. S. & Weissman, A. M. (1998) Annu. Rev. Cell Dev. Biol. 14, R., Moas, M., Seger, R., Taya, Y. & Ben-Ze’ev, A. (2002) Biochem. Pharmacol. 19–57. 64, 865. 82, 13. Woods, D. B. & Vousden, K. H. (2001) Exp. Cell Res. 264, 56–66. 32. Glickman, M. H. & Ciechanover, A. (2002) Physiol. Rev. 373–428. 14. Roth, J., Dobbelstein, M., Freedman, D. A., Shenk, T. & Levine, A. J. (1998) 33. Strous, G. J. & Schantl, J. A. (2001) Science STKE 2001, PE41. EMBO J. 17, 554–564. 34. Shenoy, S. K., McDonald, P. H., Kohout, T. A. & Lefkowitz, R. J. (2001) 15. Weiss, J., Schwechheimer, K., Cavenee, W. K., Herlyn, M. & Arden, K. C. Science 294, 1307–1313. (1993) Int. J. Cancer 54, 693–699. 35. Perry, M. E., Piette, J., Zawadzki, J. A., Harvey, D. & Levine, A. J. (1993) Proc. 16. Weiss, J., Heine, M., Arden, K. C., Korner, B., Pilch, H., Herbst, R. A. & Jung, Natl. Acad. Sci. USA 90, 11623–11627. E. G. (1995) Recent Res. Cancer Res. 139, 137–154. 36. Daujat, S., Neel, H. & Piette, J. (2001) Trends Genet. 17, 459–464. 17. Rubini, M., Hongo, A., D’Ambrosio, C. & Baserga, R. (1997) Exp. Cell Res. 230, 37. Li, L., Liao, J., Ruland, J., Mak, T. W. & Cohen, S. N. (2001) Proc. Natl. Acad. 284–292. Sci. USA 98, 1619–1624. 18. Goetz, A. W., van der Kuip, H., Maya, R., Oren, M. & Aulitzky, W. E. (2001) 38. Levkowitz, G., Waterman, H., Zamir, E., Kam, Z., Oved, S., Langdon, W. Y., Cancer Res. 61, 7635–7641. Beguinot, L., Geiger, B. & Yarden, Y. (1998) Genes Dev. 12, 3663–3674. 19. Chen, L., Agrawal, S., Zhou, W., Zhang, R. & Chen, J. (1998) Proc. Natl. Acad. 39. Decraene, D., Agostinis, P., Bouillon, R., Degreef, H. & Garmyn, M. (2002) Sci. USA 95, 195–200. J. Biol. Chem. 277, 32587–32595.

8252 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.1431613100 Girnita et al. Downloaded by guest on September 30, 2021