p53-Dependent PUMA to DRAM antagonistic interplay as a key molecular switch in cell-fate decision in normal/high glucose conditions

In this study colon cancer RKO, HCT116, and HCT116-p53^−/− cells were used. Cells were routinely cultured in DMEM (Dulbecco’s Modified Eagle’s medium) (Life Technology-Invitrogen, Eggenstein, Germany) containing 1 g/L D-glucose (low glucose - LG), supplemented with 10% heat-inactivated fetal bovine serum (FBS) plus 100 units/ml penicillin/streptomycin, and glutamine, in 5% CO₂ humidified incubator at 37 °C. For high glucose (HG) condition, low glucose culture medium was replaced with DMEM containing 4.5 g/L D-glucose (Life Technology-Invitrogen) supplemented with 2% FBS for 24 h, as previously reported [25‐28], before performing other treatments for the indicated times.

To generate a cell line stably expressing GFP-LC3, 3 × 10⁵ RKO cells were cultured into 6-well plates and transfected with 5 μg of pEGFP-LC3 expression vector (kindly provided by Moshe Oren, Weizmann Institute of Science, Rehovot, Israel) with Lipofectamine Plus (Invitrogen, Monza, Italy) according to the manufacturer’s instructions. Forty-eight hours after transfection cells underwent selection with geneticin G418 (1.5 mg/ml). After 10–14 days, the selected GFP-LC3-positive clones were visualized on a Nikon Eclipse Ti-U fluorescence microscope (Nikon).

Chemotherapeutic drug Adriamycin (ADR) (Sigma) was added in culture medium at 2 μg/ml for 16–24 h; the inhibitor of autophagic protein degradation chloroquine (CQ) [29] (Sigma-Aldrich) was added in culture medium at 25 μM for 16 h; p53 inhibitor pifithryn-α (PFT-α) [30] (ENZO Life Sciences, Lausen Switzerland) was used at 30 μM for 16–24 h, as reported [31].

For viability assay, subconfluent cells were plated in duplicate in 60 mm multiwell Petri dishes and 24 h later culture medium was replaced with HG or low glucose medium, both containing 2% FBS. The day after, ADR (2 μg/ml) were added to cell cultures for 16–24 h. Both floating and adherent cells were collected and cell viability was determined by Trypan blue (Sigma, St. Louis, MO, USA) exclusion by direct counting with a haemocytometer. The percentage of cell death, as blue/total cells, was assayed by scoring about 200 cells per well in triplicate.

Cell death was quantified by Fluorescence Activated Cell Sorting (FACS) analysis, staining cells with the nonvital dye propidium iodide (PI) (Immunological Sciences, Rome, Italy), following the manufacturer’s instruction [32]. Briefly, floating cells were collected by centrifugation and pooled with adherent cells recovered from the plates, fixed in 80% ethanol and stained in a PBS solution containing PI (62.5 mg/ml; Sigma-Aldrich), and RNase A (1.125 mg/ml; Sigma-Aldrich). Samples were acquired with a FACScan instrument (Becton Dickinson Europe Holdings SAS - Le Pont De Claix, France) and the percentage of cells in sub-G1 compartment was calculated using ModFit LT software (Becton Dickinson). About 30.000 events were acquired and gated using forward scatter and side scatter to exclude cell debris.

Cells were harvested in TRIzol Reagent (Life Technology-Invitrogen) and total RNA was isolated following the manufacturer’s instructions, as previously reported [25]. The first strand cDNA was synthesized from 2 μg of total RNA with MuLV reverse transcriptase kit (Applied Biosystems, Foster City, CA, USA). Semi-quantitative Reverse-Transcribed (RT)-PCR was carried out by using Hot-Master Taq polymerase (Eppendorf, Milam Italy) with 2 μl cDNA reaction and genes specific oligonucleotides under conditions of linear amplification. PCR products were run on a 2% agarose gel and visualized with ethidium bromide. The housekeeping 28S gene, used as internal standard, was amplified from the same cDNA reaction mixture. Densitometric analysis was applied to quantify mRNA levels compared to control gene expression.

Cells were plated at semiconfluence in 35-mm Petri dishes the day before transfection. Control-siRNA, siDRAM (sc-96,209; Santa Cruz) and siATG5 (Dharmacon, Thermo Scientific, Milan, Italy) were transfected overnight using Lipofectamine Plus reagent (Invitrogen). DRAM silencing was evaluated 48 h after transfection by RT-PCR analysis.

In vivo experiment was performed by using six-week-old CD-1 male nude (nu/nu) mice (Charles River Laboratories). They were housed in specific pathogen-free conditions and fed standard cow pellets and water ad libitum. Studies were performed in accordance with the National Cancer Institute Regina Elena standard guidelines for animal care; all experimental protocol were approved by the Ethical Committee for animal research of the National Cancer Institute Regina Elena in Rome, Italy, and by the Italian Ministry of Health, and performed in accordance with the Italian and European legislation. For diabetes induction the mice were injected intraperitoneally with a single high dose of 160 to 240 mg/kg streptozotocin (STZ) (Sigma-Aldrich) which is considered to achieve consistently a diabetic state with limited morbidity and mortality [33]. The mice were subsequently hydrated with 1.0 ml normal physiologic saline administrated subcutaneously. Blood was obtained by lancet prick in the tail. Blood glucose concentration was monitored once before and daily after STZ injection until a diabetic state was confirmed by the glucose dehydrogenase method. After mice reached a glucose concentration exceeding 300 mg/dl (considered diabetic), solid tumors were obtained by injecting i.m. on the flank of each mouse 4 × 10⁶ viable RKO cells suspended in 0.1 ml PBS. The mice were examined every day after injection until tumors reached approximately 300 mm³ volume (about 7 days from injection). Mice were then randomized in four groups (6–8 mice/group) and injected with ADR (10 mg/kg body weight) i.p or PBS as control, as follow: 1) normoglycemic/PBS (Mock), 2) normoglycemic/ADR, 3) diabetic/PBS, 4) diabetic/ADR. Tumor dimensions were measured every other day and their volumes were calculated from caliper measurements of two orthogonal diameters (x and y, larger and smaller diameters, respectively) by using the formula, volume = xy²/2. When the ADR effect on tumor growth delay reached statistical significance between normoglycemic and diabetic groups, tumors were harvested and RNA extracted for analysis of PUMA and DRAM expression by RT-PCR.

We first evaluated whether the glucose amount influenced the p53 transcriptional activity in response to drug. To this aim, HCT116 and RKO cells were cultured in low glucose (LG) and high glucose (HG) media and treated with adriamycin (ADR). The results of RT-PCR analyses of mRNA levels show that PUMA was greatly induced by ADR in LG, as expected [25], while was not induced in HG (Fig. 1a); on the contrary and opposite to PUMA, DRAM was specifically induced by ADR in HG while was not induced in LG (Fig. 1a). To assess if DRAM was specifically induced by p53 in this setting, as it can be also activated by p73 a p53 family protein [34], we inhibited p53 transcriptional activity with pifithrin-α (PFT-α) [30]. We found that the expression of DRAM, induced by ADR in HG, was efficiently impaired by PFT-α co-treatment (Fig. 1b). Similarly, neither DRAM nor PUMA were induced in HCT116-p53^−/− treated with ADR in, respectively, LG and HG (Fig. 1c), demonstrating that both genes were induced by p53 in different glucose culture condition.

To evaluate the occurrence of autophagy, RKO cells were stably transfected with GFP-LC3 plasmid and selected as mixed population. Then, cells were cultured in LG and HG media and treated with ADR. Autophagy induction was evaluated by the appearance of GFP-LC3 puncta, indicating autophagosome formation, The cells displaying more than 5 dots/cell were considered undergoing autophagy, and the measurements were performed in the presence of autophagy inhibitor choroquine (CQ), an inhibitor of the lysosomal function and therefore indicative of autophagic flux, as reported [29, 35]. We found that the LC3 puncta formation in cells cultured in HG was slightly increased compared to cells cultured in LG and that ADR treatment further increased the LC3 puncta formation in HG compared to the same treatment in LG (Fig. 2a, b). Next Western blot analysis of the expression of LC3 (microtubule-associated protein light chain 3) after conversion from LC3-I to its autophagosome membrane-associated lipidated, activated LC3-II form was performed, in the presence and in the absence of CQ. We found that LC3-II conversion, following ADR treatment in LG medium, was greatly increased by HG medium (Fig. 2c, d), suggesting increase of autophagy in HG. We then evaluated whether p53 was involved in autophagy increase. To this aim, p53 transcriptional activity was inhibited by using PFT-α. As shown in Fig. 3a, the LC3-II conversion following ADR treatment in HG medium was impaired by PFTα co-treatment. Analysis of p62, a bona fide autophagic substrate [35], showed that the reduced levels following ADR treatment in HG were counteracted by PFTα co-treatment (Fig. 3a). The role of DRAM was then assessed by using siRNA interference (Fig. 3b). As shown in Fig. 3c, the degradation of p62 in ADR/HG condition was impaired by DRAM interference. Altogether these findings suggest that HG condition increased the ADR-induced autophagy and that such outcome was in part depending on p53 activity and DRAM expression.

Next, we evaluated whether autophagy was involved in reduction of drug-induced cell death in HG. To this purpose cells were treated with ADR in LG and HG, in the presence or absence of autophagy inhibitor CQ. The results show that the ADR-induced cell death in LG, as assessed by PI staining and FACS analysis, was significantly reduced in HG condition; interestingly, blocking autophagy with CQ rescued the ADR-induced cell death in HG approximately to the levels obtained in LG (Fig. 4a , upper panel). CQ alone did not induce cell death (data not shown). In agreement, Western blot analysis show that the expression of the apoptosis marker (cleaved) PARP in response to ADR in LG was reduced by HG and rescued by CQ co-treatment (Fig. 4a , lower panel). To further evaluate the role of autophagy, depletion of ATG5, one of the members of the ATG family necessary for autophagy because of its role in autophagosome elongation [36], was performed with specific siRNA. We found that ATG5 knockdown rescued the cell death inhibited in ADR/HG condition to almost the levels obtained by ADR in LG (Fig. 4b). Next the role of DRAM was assessed in both RKO and HCT116 cells. The results show that the ADR-induced cell death in LG and significantly reduced by ADR treatment in HG condition, was mostly restored by silencing DRAM with siRNA (Fig. 4c). Altogether, these data suggest that autophagy and/or DRAM expression were responsible, at least in part, of reduced cancer cell death in response to ADR in HG condition.

Next we evaluated whether blocking autophagy and/or DRAM could influence PUMA transcription in HG. RT-PCR analyses of mRNA levels in RKO and HCT116 cells show that PUMA expression was induced by ADR in LG while was not induced in HG (Fig. 5a , compare lane 2 with lane 6), as expected; interestingly, PUMA was induced by ADR in HG only in the presence of CQ (Fig. 5a , compare lane 6 with lane 7), while the ADR-induced PUMA expression in LG was not further increased by CQ co-treatment (Fig. 5a , compare lane 2 with lane 3). As opposite to PUMA, DRAM expression was induced by ADR only in HG, as seen above, but inhibited by CQ co-treatment (data not shown). Interestingly, PUMA was induced by ADR in HG following siRNA silencing of DRAM (Fig. 5b , compare lane 4 with lane 6), as assessed by densitometric analysis (Fig. 5c). Altogether, these findings suggest that the p53 pro-apoptotic function, impaired in HG, could be restored by autophagy inhibition or DRAM depletion. Therefore, we hypothesize that, downstream of p53, autophagy might sustain inhibition of p53 pro-apoptotic function in a regulatory loop.

Finally, the biological effect of the hyperglicemic microenvironment on drug response was evaluated in a nude mouse model of Streptozotocin (STZ)-induced diabetes [33], to recapitulate the in vitro results. After the mice achieved a consistent diabetic state, as evaluated by blood glucose concentration exceeding 300 mg/dL, compared to the normal group that maintained blood glucose concentration around 100 mg/dL (Fig. 6a), we generated tumor xenografts by injecting RKO cells, as previously reported [37, 38], in both diabetic and normoglycemic groups. Cells were implanted into nude mice by i.m. injection and allowed to develop into palpable tumor nodules (about 300 mm³) at the injection sites. The mice were then treated with ADR and tumor volumes were calculated from caliper measurements. Ten days after injection only normoglycemic mice treated with ADR displayed significant tumor growth delay (ADR versus Mock: *P < 0.001), compared to the same treatment in diabetic (SZT) mice (Fig. 6b). Then, tumors were harvested and mRNA extracted for p53 target gene expression by RT-PCR analyses. The results show that PUMA was significantly induced by ADR treatment in normoglycemic mice (Fig. 6c) while weakly induced in SZT mice; on the contrary, DRAM was induced by ADR treatment in diabetic mice while was not induced in normoglycemic ones, in agreement with the above in vitro data. This experiment recapitulated the in vitro results and underlined how the diabetic condition might negatively influence tumor response to drugs by impairing p53 pro-apoptotic activity.