All protocols involving human subjects and samples were approved by the institute ethical review board (Approval No. JXSETYY-YXKY-20180010) and performed according to the Declaration of Helsinki. In addition, informed consent has been obtained from the participants involved. The clinicopathological data for the participated patients were listed in Supplementary Table S1.

The human retinoblastoma cell lines Y79 and WERI were purchased from cell bank of Chinese Academy of Sciences, and were cultured in RPMI1640 with 10% FBS. Plasmids encoding human MDM2 (Cat. No. HG11206-UT) and Flag-tagged human MDM2 (Cat. No. HG11206-NF) were both purchased from SinoBiological. siRNA targeting MDM2 (Cat. No. AM51331) and negative control siRNA (Cat. No. 4390844) were both purchased from ThermoFisher Scientific.

Primary RB cells were isolated as previously described with modifications (14). In brief, fresh human RB tumor samples were washed three times in ice-cold phosphate buffer saline (PBS). Dissociation into single cells was achieved by incubation in 0.25% trypsin-0.01% EDTA (Invitrogen, ThermoFisher Scientific) at 37°C for 10 min and gentle pipetting, then trypsin was neutralized with addition of fetal bovine serum (FBS, Gibco ThermoFisher Scientific). Dissociated cells were filtered with 70 μm cell strainer (Corning) and centrifuged at 1,000 rpm for 5 min, and followed by resuspension in serum free Dulbecco’s modified Eagle’s medium (DMEM)/F12 medium (Gibco, ThermoFisher Scientific) supplemented with 20 ng/mL recombinant human leukemia inhibitory factor (Chemicon). Isolated cells were maintained at 37°C, 5% CO₂ under a humidified atmosphere.

For hypoxia treatment, cells were cultured in a hypoxia chamber (Galaxy® 48 R CO2, Eppendorf) with conditions of 0.2% oxygen and 5% carbon dioxide at 37°C. For inhibitor treatment, cells were treated with 1 µM MDM2 inhibitor SP-141 (Cat. No. 5332; Tocris Bioscience), or 5 µM HIF-1α inhibitor 2-MeOE2 (Cat. No. S1233; Selleck Chemicals) or both for 48 h at 37°C.

To overexpress MDM2, cells were transfected with plasmid encoding either untagged or Flag-tagged human MDM2 using Lipofectamine 2000 (Invitrogen, ThermoFisher Scientific), according to the manufacturer’s instructions. To knockdown MDM2 expression, cells were transfected with MDM2 siRNA using Lipofectamine™ RNAiMAX (Invitrogen, ThermoFisher Scientific), according to the manufacturer’s instructions.

The impact of MDM2 overexpression on pVHL half life was assessed as previously described with modifications (15). In brief, cells were first transfected with MDM2 plasmid or empty vector (control), and then cells were treated with 50 μg/mL cycloheximide (CHX) for 1, 2, 4, 6 and 8 h under hypoxia. At these indicated time points, cells were harvested and the expression of pVHL was analyzed by western blot.

Co-IP was performed as previously described with modifications (16). In brief, cells were first transfected with Flag-MDM2 plasmid or empty vector (control) for 24 h and then cells were lysed with IP Lysis Buffer (Pierce, ThermoFisher Scientific) supplemented with Halt™ Protease Inhibitor Cocktail (ThermoFisher Scientific). Cleared lysed samples were then incubated with Dynabeads Protein G (Cat. No. 10003D; Invitrogen, ThermoFisher Scientific) for 1 h at room temperature to eliminate non-specific binding. Subsequently, samples were incubated with Dynabeads coating with anti-pVHL antibody (Cat. No. 24756-1-AP; ProteinTech), or control IgG (Cat. No. 30000-0-AP; ProteinTech) for 1 h at room temperature with end-to-end rotation. Following incubation, the Dynabead-Ab-Ag complex was then washed with PBST for 3 times and the immunoprecipitated protein was then eluted with SDS-PAGE loading buffer and heated at 90°C for 10 min.

Western blotting analysis was performed as previously described with modifications (17). In brief, cells were first lysed with IP Lysis buffer (Pierce, ThermoFisher Scientific) supplemented with Halt™ Protease Inhibitor Cocktail (ThermoFisher Scientific) and then protein concentration was determined by BCA Protein Assay (ThermoFisher Scientific) according to the manufacturer’s instructions. Following quantification, 20 µg total protein was mixed with SDS-PAGE loading buffer, heated at 90°C for 10 min and separated by SDS-PAGE gel. For Co-IP samples, they were directly loaded onto SDS-PAGE gel for separation. Resolved protein was subsequently transferred onto an Immobilon-P PVDF membrane (Millipore, Merck). After blocking with 5% nonfat milk in PBST at room temperature for 1 h, the membrane was then sequentially incubated with primary antibodies and HRP-conjugated secondary antibodies. Primary antibody incubation was performed at 4°C overnight, while secondary incubation was performed at room temperature for 1 h. Following incubations, the membrane was extensively washed with PBST and immunobands were detected by LumiGLO reagent (Cell Signaling Technology). The following primary antibodies were used: rabbit anti-human MDM2 antibody (Cat. No. 27883-1-AP; ProteinTech), rabbit anti-human HIF-1α antibody (Cat. No. 20960-1-AP; ProteinTech), rabbit anti-pVHL antibody (Cat. No. 24756-1-AP; ProteinTech), rabbit anti-human β-actin antibody (Cat. No. 81115-1-RR; ProteinTech), mouse anti-Ubiquitin (P4D1) (Cat. No. 3936; Cell Signaling Technology), mouse anti-Flag (DYKDDDDK) antibody (Cat. No. 66008-4-Ig; ProteinTech). The following secondary antibodies were used: mouse anti-rabbit IgG-HRP (Cat. No. sc-2357; Santa Cruz) and HRP-conjugated Affinipure Goat Anti-Mouse IgG(H + L) (Cat. No. SA00001-1; ProteinTech).

Total RNA was first extracted from cultured cells using the RNeasy Mini Kit (Qiagen) and then transcribed into cDNA using the iScript Select cDNA synthesis kit (Bio-Rad). The transcribed cDNA was subsequently used as template to amplify target gene sequences, using the iQ SYBR Green Supermix (Bio-Rad) on the iCycler iQ Real-Time Detection System (Bio-Rad). The β-actin mRNA was used as an internal control and the relative expression of each target gene was calculated using the 2^−ΔΔCt formula. The following primer pairs were used in the current study: hif-1α: forward primer: 5′-TAT​GAG​CCA​GAA​GAA​CTT​TTA​GGC-3′ and reverse primer: 5′-CAC​CTC​TTT​TGG​CAA​GCA​TCC​TG-3′, mdm2: forward primer: 5′-TGT​TTG​GCG​TGC​CAA​GCT​TCT​C-3′ and reverse primer: 5′-CAC​AGA​TGT​ACC​TGA​GTC​CGA​TG-3′, VHL: forward primer: 5′-GAC​ACA​CGA​TGG​GCT​TCT​GGT​T-3′ and reverse primer 5′-ACA​ACC​TGG​AGG​CAT​CGC​TCT​T-3′, and β-actin: forward primer: 5′-AAG​CAG​GAG​TAT​GAC​GAG​TCC​G-3′, and reverse primer 5′-GCC​TTC​ATA​CAT​CTC​AAG​TTG​G-3’. All primers were purchased from OriGene.

Cell viability was assessed using a commercial CellTiter-Glo Luminescent Cell Viability Assay Kit (Promega), according to the manufacturer’s instructions. Briefly, cells pre-seeded at 1 × 10⁴ cells/well in 96-well plates were first treated with 1 µM MDM2 inhibitor SP-141 (Cat. No. 5332; Tocris Bioscience), or 5 µM HIF-1α inhibitor 2-MeOE2 (Cat. No. S1233; Selleck Chemicals) or both for 48 h at 37°C under normoxic or hypoxic conditions. After treatment, equal volume of CellTiter-Glo reagent was added to each well and the plate was incubated for 2 min on an orbital shaker to induce cell lysis. Then the plate was kept at room temperature for 10 min to stabilize luminescent signal and luminescence was subsequently recorded. Cell viability was calculated with cells with mock treatment being considered as 100% viable.

Cell apoptosis was detected using a commercial Caspase-Glo 3/7 Assay System (Promega), according to the manufacturer’s instructions. In brief, cells in 96-well plates were first treated as detailed in the Cell viability assay above, and then equal volume of Caspase-Glo 3/7 reagent was added to each well. The plate was then mixed by gentle shaking on an orbital shaker for 30 s, followed by incubation at room temperature for 30 min. Luminescence was finally recorded by a microplate reader.

RB CSCs were generated and cultured as previously described with modifications (14). In brief, primary RB cells from 5 patients were pooled and cultured in serum-free CSC culture medium (DMEM/F12 medium supplemented with 20 ng/mL recombinant human EGF (R&D systems), 20 ng/mL recombinant human FGF2 (StemCell), 1x B27 Plus supplement (Gibco, ThermoFisher Scientific), 1x Penicillin-Streptomycin-Glutamine (Gibco, ThermoFisher Scientific), 5 μg/mL heparin (StemCell), and 20 ng/mL recombinant human LIF (SinoBiological). Cells were subcultured every 3 days until single cells died out and survived cells formed CSC-like solid spheroids, a key characteristic of CSCs. The volume of the spheroids was measured as previously described using the formula: V o l u m e = ( m e a n   d i a m e t e r ) 3 × π 6 (18). The mean diameter of the spheroids was defined as the mean value of the longest and shortest diameters in this study.

Data were expressed as mean +/− standard deviation (SD) in the current study and statistical analyses were performed with GraphPad Prism (version 7.0.4, GraphPad). Simple linear regression was used to assess the correlation between two variables. Mann-Whitney test was used to compare the difference between two groups, while Kruskall-Wallis test plus Dunn’s multiple comparisons test were used to compare the difference among three or more groups. A p-value less than 0.05 was considered statistically significant.

To investigate the expression profile of MDM2, HIF-1α and pVHL in RB patients, primary RB cells from 13 RB patients were harvested and the expression of these 3 proteins were determined by Western blot. As shown in Figure 1A, all 3 proteins were successfully detected in the primary RB cells. Correlation analyses were performed to further explore the relation of the 3 proteins. Our data showed that MDM2 expression was positively correlated with that of HIF-1α, while the expression of MDM2 and pVHL, as well as that of HIF-1α and pVHL were negatively correlated (Figures 1B–D). These results indicate a correlation among MDM2, HIF-1α and pVHL.

To investigate the impact of MDM2 on the expression of HIF-1α and pVHL, RB cell lines Y79 and WERI were first overexpressed with MDM2 under hypoxia, and then the mRNA and protein levels of HIF-1α and pVHL were determined by real time PCR and western blot, respectively. Our results showed that MDM2 overexpression did not show apparent impact on the mRNA levels of HIF-1α and pVHL in either Y79 or WERI cells (Figure 2A). However, a considerable increase of HIF-1α while a decrease of pVHL on protein level was observed in the cells overexpressed with MDM2 (Figure 2B; Supplementary Figure S1A). These data indicate that MDM2 could regulate the expression of HIF-1α and pVHL on the protein level, but not on the mRNA level.

To further confirm the regulatory role of MDM2 on HIF-1α and pVHL, Y79 and WERI cells under hypoxia were treated with MDM2 siRNA or a MDM2 specific inhibitor SP141. As shown in Figure 2C and Supplementary Figure S1B, knockdown of MDM2 with siRNA decreased the expression of HIF-1α while increased the expression of pVHL in both Y79 and WERI cells. A similar result was also observed in cells treated with SP141 (Figure 2D; Supplementary Figure S1C). Namely, the MDM2-specific inhibitor reduced MDM2 and HIF-1α expression while increased pVHL expression (Figure 2D; Supplementary Figure S1C). Taken together, our data here indicate that MDM2 may have a regulatory effect on the expression of HIF-1α and pVHL, either through direct or indirect mechanisms.

Our data above showed that MDM2 expression was correlated with both HIF-1α and pVHL. Given that HIF-1α is regulated by pVHL, it is likely that MDM2 affected the expression of HIF-1α via its regulation on pVHL expression (12). Interaction between MDM2 and pVHL has been previously described and we suspected if MDM2, with its E3 ubiquitin ligase activity, could promote pVHL degradation through ubiquitination (19). To test our hypothesis, we first overexpressed MDM2 in Y79 cells and examined the half-life of pVHL under hypoxia. Our data showed that pVHL had an accelerated degradation pattern in cells with MDM2 overexpression comparing to cells without MDM2 overexpression (Figures 3A, B; Supplementary Figures S2A, B). To further explore if there is direct interaction between MDM2 and pVHL, endogenous pVHL was immunoprecipitated by anti-pVHL antibody in Y79 cells and the presence of MDM2 in the immunoprecipitated product was determined by western blot. As we suspected, overexpressed MDM2 was co-precipitated with pVHL, indicating a direct interaction between the 2 proteins (Figure 3C; Supplementary Figure S2C). Moreover, the blotting with an anti-Ub antibody further demonstrated that the ubiquitination of pVHL was significantly enhanced by MDM2 overexpression (Figure 3C; Supplementary Figure S2C). To further confirm the causal relationship between MDM2 overexpression-induced pVHL ubiquitination and the increase of HIF-1α expression, we generated a mutated version of MDM2, MDM2^C462A. The introduction of this mutation altered a structural-crucial zinc coordinating cysteine in MDM2, and consequently abolished the E3 ubiquitin ligase activity of MDM2 and its ability to ubiquitinate its substrates like pVHL (20,21,22). Our results showed that the expression of HIF-1α was increased in cells transfected with wild type MDM2, but such increase was not detected in cells transfected with MDM2^C462A, confirming that the ubiquitin ligase activity of MDM2 on pVHL was involved in the regulation of HIF-1α expression (Figures 3D, E). Taken together, these results indicate that MDM2 regulated pVHL through promoting ubiquitination of pVHL, which consequently led to the upregulation of HIF-1α.

To investigate the effect of MDM2 and HIF-1α on RB cell viability under hypoxia, RB cells Y79 and WERI were treated with MDM2-specific inhibitor SP141 and/or HIF-1α-specific inhibitor 2-MeOE2 under normoxia or hypoxia. Under normoxia, the proliferation of Y79 and WERI cells upon SP141 treatment was dropped to 78.30% and 84.60%, respectively (Figure 4A). However, under hypoxia, both cell lines showed further decrease in cell proliferation after SP141 treatment, with Y79 being 68.9% and WERI being 75.75% of control, respectively (Figure 4C). Treatment of 2-MeOE2 did not show apparent impact on cell viability under normoxia (Figure 4B), but significantly reduced cell viability under hypoxia (Figure 4C). Moreover, a further reduction in cell viability was observed when cells were treated with both SP141 and 2-MeOE2 under hypoxia (Figure 4C). In accordance, cell apoptosis, as indicated by Caspase-3/7 activity, was increased when cells were treated with SP141 or 2-MeOE2, and similarly a further increase was observed when a combination of SP141 and 2-MeOE2 was used (Figure 4D). Taken together, our data here indicate that both MDM2 and HIF-1α are important for RB cell survival under hypoxia and inhibition of these 2 proteins could result in enhanced cell death.

Increasing evidence has confirmed that a subset of cells, existed in many cancer types, exhibit capacity to rapidly proliferate and self-renewal, now known as cancer stem cells (CSCs) (23). CSCs are thought to play an important role in cancer initiation and progression, which is relevant to tumor persistence to treatment. Previous study has shown that MDM2 is required for the efficient generation of induced pluripotent stem cells (iPSCs) from murine embryonic fibroblasts, in the absence of p53 (24). HIF activity is also critical in maintenance of cancer stem cells as well as differentiation and function of inflammatory cells (25). Since inhibition of MDM2 and HIF-1α could induce apoptosis in RB cells under hypoxia, next we sought to analyze whether inhibition of MDM2 and HIF-1α could regulate the stem-cell properties of RB.

Primary cells were isolated from fresh human RB tissues after enucleation and cultured in serum-free conditions in the presence of DMSO, MDM2 inhibitor SP 141 and/or HIF-1α inhibitor 2-MeOE2 under hypoxia. As shown in Figure 5A, cells from all treatment groups formed stem-cell like spheroids, a characteristic morphology of CSCs. However, the spheroid sizes were distinctively different among treatments. In detail, the largest spheroids were observed in the DMSO control group, while the size of these spheroids were reduced by 70% in SP 141 and 2-MeOE2 treated groups (Figures 5A, B). Moreover, a further reduction in size for the spheroids were seen in cells treated with both SP 141 and 2-MeOE2, with the mean volume being only 15% of those from the DMSO group (Figures 5A, B). The stemness of the formed CSCs were also investigated by assessing the expression of CSC markers OCT4, SOX2 and CD133. Similarly, treatment with SP141 or 2-MeOE2 alone reduced the expression of all 3 markers in comparison to DMSO treatment, and the treatment with SP141 and 2-MeOE2 together further reduced the expression of these markers (Figure 5C). Collectively, these data showed MDM2 and HIF-1α are of great importance in the generation of CSCs in RB and inhibition of these 2 proteins could significantly decrease the stem-cell properties of primary retinoblastoma cells.