Targeted activation of innate immunity for therapeutic induction of autophagy and apoptosis in melanoma cells

Abstract

Inappropriate drug delivery, secondary toxicities, and persistent chemo- and immunoresistance have traditionally compromised treatment response in melanoma. Using cellular systems and genetically engineered mouse models, we show that melanoma cells retain an innate ability to recognize cytosolic double-stranded RNA (dsRNA) and mount persistent stress response programs able to block tumor growth, even in highly immunosuppressed backgrounds. The dsRNA mimic polyinosine-polycytidylic acid, coadministered with polyethyleneimine as carrier, was identified as an unanticipated inducer of autophagy downstream of an exacerbated endosomal maturation program. A concurrent activity of the dsRNA helicase MDA-5 driving the proapoptotic protein NOXA resulted in an efficient autodigestion of melanoma cells. These results reveal tractable links for therapeutic intervention among dsRNA helicases, endo/lysosomes, and apoptotic factors.

Figures

Figure 1. Induction of autophagy by [pIC]PEI results in melanoma cell death

(A) Fluorescence imaging of eGFP-LC3 distribution in SK-Mel-103 melanoma cells treated for 6 h with 1 μg/ml [pIC]PEI. Cells treated with PEI as single agent are shown as reference controls. (B) Changes in the electrophoretic mobility of the endogenous LC3 in SK-Mel-103 treated for the indicated times with pIC, [pIC]PEI or vehicle control (PEI). (C) Visualization of eGFP-LC3 distribution in WT or Atg5-/- MEFs treated for 8 h with 25 nM Rapamycin (Rap) or 1 μg/ml pIC]PEI. pIC-treated cells are included as controls (upper two rows). Bottom panels show quantification of the percentage of cells with focal GFP-LC3 at the indicated times post treatment. Error bars correspond to SEM of three independent experiments. (D) Differential response of SK-Mel-103 melanoma cells to rapamycin and [pIC]PEI determined by visualization of LC3-GFP fluorescence (upper two rows), and brigthfield microscopy imaging (bottow row). Cells treated with vehicle were included as controls. (E) Immunoblots of total cell extracts isolated from SK-Mel-103 cells treated as indicated and probed with antibodies specific for phospho-S6 or tubulin. (F) Transmission electron micrographs of melanoma cells treated with [pIC]PEI. Arrows point to conjugates of pIC and PEI being endocytosed into the treated cells. Note the large multivesicular structures at late time points after [pIC]PEI treatment (panel d). (G) Murine B16 melanoma cells and the indicated human melanoma cell lines compared for sensitivity to PEI and pIC as single agents or in combination. Cell death was estimated by trypan blue exclusion 24 and 48h after treatment, and presented as means ± SEM of three independent experiments.

Figure 2. Anti-melanoma activity of [pIC]PEI in mice

A-D are results from SCID Beige mice; E-G are results from Tyr::RasQ61K × Ink4a/Arf-/- mice. (A) Representative images of lungs of mice 14 days after i.v. inoculation of B16 melanoma cells and treatment as indicated, and photographed under visible (Macro) or fluorescent (eGFP) light. (B) Representation of the mean number of metastases (± SEM) induced by B16 as indicated in (A). P*<0.01 between PEI or pIC and [pIC]PEI treatment groups (n=5; generalized Mann–Whitney test). (C) Lung colonization by eGFP-SK-Mel-103 in SCID-beige mice treated with the indicated agents and assessed by fluorescence imaging and histological analyses. (D) Quantification of average number of external lung nodules (± SEM) in the indicated treatment groups shown in (C). P*<0.01 between PEI, pIC and [pIC]PEI treatment groups (n=5; generalized Mann–Whitney test). (E) Cohorts of Tyr::RasQ61K × Ink4a/Arf-/- mice were treated topically with a single dose of 200 μg DMBA at 8 weeks of age. Upon appearance of pigmented lesions of >1 mm diameter, mice were treated as indicated. Control groups received PEI in 5% glucose. The fraction of animals with tumors of <1mm diameter (progression free survival) was represented by Kaplan-Meier graphs. (F) Average number of cutaneous melanocytic neoplasms developing in each of the different treatment groups. Scoring was performed every five days and tumors were grouped by size as indicated. (G) Representative transverse (left panels) and coronal sections (right panels) of PET /CT fused images to assess metabolic activity (18F-FDG incorporation) of representative examples of mice treated as indicated. Tumors are encircled with white dashed lines. The asterisks mark animal hearts.

Figure 3. [pIC]PEI-triggered melanoma cell death is driven by autophagosome-autolysosome formation

(A) Representative bright field (left and middle panels) and electron microscope (right panel) micrographs of SK-Mel-103 treated as indicated (30 h). (B) Transformed wild-type or Atg5-/- MEFs treated with low dose (0.2 μg/ml) of [pIC]PEI or control. Cell death was estimated by trypan blue exclusion 48 h after treatment, and presented as means of three independent experiments. (C) Confocal fluorescence images of SK-Mel-103 transduced by Cherry-GFP-LC3 to detect autophagosomes (red and green foci) and autolysosomes (red-only foci) formation after treatment with [pIC]PEI, 25 nM Rapamycin (Rap) or solvent control. (D) Inhibitory effect of 100 μM Bafilomycin (Bafil), 20 μM Chloroquine (Chlor) or 10μg/ml Pepstatin (PEP) on cell death estimated by trypan blue exclusion 20h after treatment with vehicle (white bars) or [pIC]PEI (black bars). Data are indicated as means ± SEM of three independent experiments. (E) Confocal fluorescence images of SK-Mel-103 cells stably transfected with eGFP-Rab5 WT and incubated with [pIC]PEI labeled with Fluor Red. Shown is the internalization of [pIC]PEI by the melanoma cells in the presence or absence of Chloroquine. (F) Confocal visualization of lysosomal-dependent proteolysis upon cleavage and release of the fluorescent moiety of DQ-BSA (Green) in control or [pIC]PEI-treated SK-Mel-103. Chloroquine is included to monitor DQ-BSA emission in cells with blocked lysosomal activity. Cells were simultaneously imaged in the presence of lysotracker red (LTR-red) to visualize the lysosomal compartment. (G) The DQ-BSA-lysotracker colocalization was estimated in a minimum of 150 cells in two independent experiments, and it is expressed (as arbitrary fluorescence units) with respect to control treated cells. Error bars correspond to ± SEM of three independent experiments.

Figure 4. Generation and resolution of amphisomes upon [pIC]PEI treatment

(A) SK-Mel-103 cells stably transfected with eGFP-Rab7 WT or eGFP-Rab7 T22N were incubated with [pIC]PEI for visualization of Rab7 (green fluorescence) and Lysotracker-Red. Microphotographs were captured by confocal microscopy 10 h after treatment with [pIC]PEI. Values in insets correspond to the average area contained in Rab7-decorated vesicles. (B) Sequence of confocal microphotographs taken at the indicated time intervals (in seconds) to illustrate the fusion to and incorporation of lysosomes to Rab7-positive vesicles upon [pIC]PEI treatment. (C) Real-time triple-imaging fluorescence microscopy of GFP-Rab7 WT, Cherry-LC3 and Lysotracker Blue (green, red and blue fluorescence, respectively) in SK-Mel-103 cells treated with [pIC]PEI. Images were taken at the indicated time intervals (in minutes), 1h after treatment initiation. Arrows mark the first sequence in which the indicated markers appear visible. (D) Incorporation of LC3 to the surface of Rab7 endosomal vesicles prior to internalization and subsequent degradation. These endosome/LC3 hybrid structures (amphisomes) were visualized by fluorescence real-time microscopy of SK-Mel-103 cells expressing eGFP-Rab7 WT and Cherry-LC3.

Figure 5. MDA-5 is a sensor and driver of [pIC]PEI cytotoxicity in melanoma cells

(A) Fold-induction of the transcript levels of the indicated pathogen-associated receptors relative to non-treated cells. The IFIT-1 gene was used as reference for a classical IFN-responsive factor. (B) Inhibition of TLR-3 expression by transient transduction of siRNA (left panels) or stable infection with lentivirus coding for shTLR3 (right panels). Cell death was assessed by trypan blue exclusion after treatment with vehicle (white bars) or [pIC]PEI (black bars). Insets correspond to mRNA levels detected by RT-PCR. Data are shown as means ± SEM of three independent experiments. (C) MDA-5 induction visualized by immunoblotting in SK-Mel-28 and SK-Mel-147 treated with PEI, pIC or [pIC]PEI. Bortezomib was used as a control for an effective death inducer. Full length and cleaved MDA-5 (MDA-5FL and MDA-5C, respectively) are indicated with arrows. Asterisk corresponds to a non-specific band (not blocked by MDA-5 shRNA; not shown). (D) MDA-5 mRNA levels (left panel) and cell death (right panels) in SK-Mel-103 infected with the pLKO lentivirus expressing scrambled or MDA-5 shRNAs, and treated as indicated. Data are shown as means ± SEM of three independent experiments. P<0.05 between control and shMDA-5 cells in the presence of [pIC]PEI (two-tailed Student's t-test) (E) Inhibition of MDA-5 expression by transient transduction of a second set of control and MDA-5 siRNAs driven by the psi-vector frame (see Materials and Methods). Cell death in control and [pIC]PEI-treated SK-Mel-103 (24h) were determined by trypan blue exclusion, and plotted as means ± SEM of three independent experiments. In the [pIC]PEI treatment group, P<0.05 between control shRNA and shMDA-5 transduced cells (two-tailed Student's t-test). (F) Confocal visualization of GFP-LC3 in MeWo melanoma cells infected with Adenovirus-control or Adenovirus-MDA-5. [pIC]PEI was included as a reference for autophagosome formation. Numbers represent fraction of cells with eGFP-LC3 foci. (G) Cell death resulting from adenovirus-driven expression of MDA-5 in MeWo cells. The viral concentration (in plate forming units, pfu) is indicated. Data correspond to means ± SEM of three independent experiments.

Figure 6. [pIC]PEI engages apoptosis via NOXA independently of the status of p53 or without inducing compensatory MCL-1 activation

(A) Heat map of log2-ratio values of the indicated genes after pIC or [pIC]PEI treatment of SK-Mel-103, and calculated with respect to control (PEI)-incubated cells. (B) Immunoblots of total cell extracts isolated from SK-Mel-103 treated for 24 and 48h with pIC, [pIC]PEI or Bortezomib for the relative levels of NOXA, p53 and anti-apoptotic Bcl-2 and Bcl-xL proteins. (C) SK-Mel-28 and SK-Mel-147 (expressing p53 L145A and p53 wild type, respectively) were treated with PEI, pIC, [pIC]PEI or 25 nM Bortezomib. Shown are the relative levels of NOXA and MCL-1 protein at different time points after treatment. (D) Quantification of the levels of Mcl-1 and NOXA in SK-Mel-28 after each treatment and represented with respect to control untreated cells. (E) NOXA protein expression in melanoma SK-Mel-103 treated for 24 h with pIC or [pIC]PEI two days after infection with a lentiviral vector expressing inactive shRNA (sh Ctrl) or shRNA for NOXA. (F) Death rates (± SEM) of control and NOXA shRNA transduced SK-Mel-103 melanoma cells and incubated with pIC or [pIC]PEI for 24 h. (G) Inhibitory effect of MDA-5 downregulation on NOXA induction by [pIC]PEI determined in SK-Mel transduced with control or MDA5 shRNAs. Data is represented as means ± SEM of four independent experiments. Levels of NOXA were quantified by densitometry of the corresponding immunoblots and plotted with respect to untreated controls. (H) Immunoblots of total cell extracts isolated from SK-Mel-103 treated for 24h with vehicle or [pIC]PEI in the presence or absence of Chloroquine, and probed for the detection of MDA-5 (full length and cleaved forms) and NOXA.

Figure 7. Defining the mode of action of [pIC]PEI in vivo

(A) Response of oncogenically transformed wild type (+/+) or Mda-5-/- (-/-) MEFs, measured as time-dependent changes in tumor size (± SEM) upon treatment with naked pIC (V) or [pIC]PEI. Shown are P values for comparative analyses between the indicated treatment groups (n=10, generalized Mann-Witney test). Inset bar graphs corresponds to death rates (± SEM; n=3).of the indicated cell populations 24 h after treatment in culture with 1 μg/ml [pIC]PEI. (B) SK-Mel-103 GFP-LC3 melanoma cells implanted s.c. into SCID Beige mice were left to grow to 6-8 mm diameter and then treated with PEI or [pIC]PEI. Paraffin-embedded sections were stained for LC3 or NOXA antibodies. For LC3, high magnification photographs are also included to show the formation of foci characteristic of autophagosomes. (C) Model summarizing main results of this study. A transient activation of IFN and other antiviral stress response factors by naked pIC (1) can be shifted into a sustained stress program when this dsRNA mimic is delivered to cells with PEI (2). PEI packs pIC into nanoparticles and favors both its endosomal uptake and delivery to the cytosol for subsequent activation of the helicase MDA-5 (3). In addition to favor endosomal swelling, PEI has been described to favor endosome-endosome fusion. Within 2-3h after incubation with [pIC]PEI melanoma cells undergo massive ultrastructural changes in the endosomal compartment (4). These changes involve a sustained and cyclic Rab7>LC3>Lysotraker sequence of fusion events, revealing an active generation and resolution of endosome-autophagosome-lysosome hybrids (5). Although MDA-5 can facilitate autophagosome formation, a main function of this protein described here is the induction of the pro-apoptotic factor NOXA (6). NOXA is per se a poor apoptotic inducer, but can lower the threshold for caspase processing. Sustained lysosomal-dependent degradative process together with the activation of apoptotic caspases can ultimately converge in efficient tumor cell death (7). Importantly, we demonstrate a potent anti-melanoma activity of [pIC]PEI in various animal models, at concentrations with no obvious toxicity to normal cellular compartments.