|Home | About | Journals | Submit | Contact Us | Français|
Expression of activated Ras in glioblastoma cells induces accumulation of large phase-lucent cytoplasmic vacuoles, followed by cell death. This was previously described as autophagic cell death. However, unlike autophagosomes, the Ras-induced vacuoles are not bounded by a double membrane and do not sequester organelles or cytoplasm. Moreover, they are not acidic and do not contain the autophagosomal membrane protein, LC3-II. Here we show that the vacuoles are enlarged macropinosomes. They rapidly incorporate extracellular fluid-phase tracers, but do not sequester transferrin or the endosomal protein, EEA1. Ultimately, the cells expressing activated Ras detach from the substratum and rupture, coincident with the displacement of cytoplasm with huge macropinosome-derived vacuoles. These changes are accompanied by caspase activation, but the broad-spectrum caspase inhibitor, z-VAD, does not prevent cell death. Moreover, the majority of degenerating cells do not exhibit chromatin condensation typical of apoptosis. These observations provide evidence for a necrosis-like form of cell death initiated by dysregulation of macropinocytosis, which we have dubbed ‘methuosis’. An activated form of the Rac1 GTPase induces a similar form of cell death, suggesting that Ras acts through Rac-dependent signaling pathways to hyperstimulate macropinocytosis in glioblastoma. Further study of these signaling pathways may lead to the identification of other chemical and physiological triggers for this unusual form of cell death.
Glioblastoma is one of the most aggressive human brain tumors (1,2). Despite efforts to improve surgical, radiological and chemotherapeutic treatment strategies, the prognosis for patients with glioblastoma remains poor. A major problem is that residual cells remaining after surgical resection of the primary tumor rapidly acquire resistance to chemotherapeutic drugs (3). In addition, glioblastomas often harbor mutations in genes that regulate programmed cell death (e.g., PTEN, RB, p53), rendering them resistant to conventional pro-apoptotic stimuli (4). These characteristics have stimulated interest in identifying alternative pathways for inducing cell death in glioblastoma.
Apoptosis is the best characterized form of programmed cell death. However, non-apoptotic forms of cell death are now recognized as playing significant roles during embryonic development, neurodegeneration, and cancer regression (5). In these cases loss of cell viability may occur in a manner that is independent of caspase activation. Autophagic cell death (Type-II programmed cell death), is the most widely studied form of non-apoptotic cell death. Its diagnostic morphological feature is accumulation of autophagosomes and degradative autolysosomes (6). Autophagic death has been reported to occur in several types of cancer cells (7), but it has received particular attention in glioblastoma, where it can be induced by alkylating agents (8), arsenic trioxide (9), ionizing radiation (10), and rapamycin (11). Nevertheless, it remains controversial whether increased autophagic activity is actually a direct cause of cell death. Recent evidence supports the alternative view that accumulation of autophagosomes may signify a survival response intended to rid cells of misfolded proteins or damaged organelles (12-14).
In 1999 Chi et al. (15) reported that ectopic expression of activated Ras GTPase, which normally serves to stimulate cell proliferation, can trigger non-apoptotic cell death in glioblastoma and gastric carcinoma. This was described as autophagic death because the cells developed numerous cytoplasmic vacuoles. However, there have been no follow-up studies to confirm that the vacuoles induced by Ras are autophagosomes. In the present study we have determined that the large vacuoles that accumulate in glioblastoma cells expressing activated H-Ras are in fact derived from macropinosomes. Cell rupture coincides with continued expansion of these macropinocytotic vacuoles. These findings provide evidence for a novel form of cell death characterized by hyperstimulation of vesicular fluid uptake and accumulation of swollen macropinosomes. We have termed this process methuosis (methuo, from the Greek; to drink to intoxication).
U251 glioblastoma cells were obtained from the DCT Tumor Repository (National Cancer Institute, Frederick, MD, USA). Other cell lines were obtained from the same source, or from the American Type Culture Collection (Rockville, MD, USA). Cells were maintained at 37°C with 5% CO2 in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS). Phase contrast images of the live cells were obtained using an Olympus IX70 microscope equipped with a digital camera and SPOT imaging software (Diagnostic Instruments Inc., Sterling Heights, MI, USA).
To generate stable cell lines capable of conditional Ras expression, U251 cells were nucleofected with pTet-on (Clontech, Mountain View, CA), which encodes the reverse tet-responsive transcriptional activator. Cells were selected with 500 μg/ml G418 and clonal lines were tested in transient transfection assays to determine which ones gave the tightest doxycycline (Dox)-regulated gene expression using the tet-responsive pTRE vector (Clontech). One such clone was used for generation of permanent cell lines by nucleofection with pTRE(myc-H-RasG12V) together with pTK-Hyg. Clones were selected in medium containing 200 μg/ml hygromycin + 200 μg/ml G418 and screened to measure the expression of myc-H-Ras(G12V) in response to 1 μg/ml Dox. The clonal line, U251-C18, was selected for use in the present studies, although other clones exhibited essentially identical morphological phenotypes when Ras(G12V) expression was induced. For Rac1(G12V) expression, lentiviral particles containing the gene for myc-tagged Rac1(G12V) under the control of a tet-inducible promoter were generated by America Pharma Source, LLC (Gaithersburg, MD). These were mixed with lentiviral particles containing a gene for Blasticidin resistance and used to co-infect U251 pTet-on cells. Clones were selected in medium containing 10 μg/ml blasticidin and 200 μg/ml G418. Expression of myc-Rac1(G12V) was induced by adding 1 μg/ml Dox to the medium.
Myc-tagged H-Ras constructs, (G12V, and S17N), were subcloned into the EcoR1-BamH1 sites of the retroviral expression vector, pFBneo (Stratagene, La Jolla, CA,USA). Procedures for retrovirus production and infection of glioma cells have been described previously (16).
H-Ras(G12V) was subcloned into pCMV5 that had been modified to encode an in-frame myc epitope tag (MEQKLISEEDL). The expression vector was introduced into the U251 cells by nucleofection, using the Nucleofector II system from Amaxa Inc., (Gaithersburg, MD) with Solution T and program T-30. Other cell lines were nucleofected with different solutions and programs: HEK293 cells, Solution V with program Q-001; HeLa cells, Solution R with program I-013; and HEp2 cells, Solution V with program G-016.
Antibodies were obtained from the following sources: myc epitope (EMD Biosciences, San Diego, CA, USA), PARP (BD Pharmingen, San Jose, CA), laminA/C (Cell Signaling Technology, Danvers, MA, USA), lactate dehydrogenase (LDH) and α-tubulin (Sigma, St. Louis, MO), and LC3 (APG8b, N-Term; Abgent, San Diego, CA, USA). Protein was quantified in cell lysates by colorimetric assay using the Bio-Rad reagent (Bio-Rad, Inc., Hercules, CA, USA). SDS-PAGE and western blot analyses were performed as described previously (16), using enhanced chemiluminescent (ECL) detection (GE Healthcare, Piscataway, NJ, USA). Immunoblot signals were quantified using a Kodak 440CF image station or an Alpha Innotech FluorChem HD2 imaging system.
In order to block the lysosomal turnover of endogenous LC3-II, cells were incubated in medium containing protease inhibitors, E64D (10 μg/ml) and pepstatin A (10 μg/ml) (Peptides International, Louisville, KY) for 48 h prior to western blot analysis of LC3. Cells treated with 500 nM staurosporine (Cayman Chemical Company, Ann Arbor, MI) for 18-24 hrs served as positive controls for induction of PARP and lamin cleavage associated with apoptosis.
Viability of individual cells was assessed by fluorescence microscopy using the Viability/ Cytotoxicity Assay Kit (Biotium Inc, Hayward, CA), which measures hydrolysis of calcein acetoxy methylester (live) and uptake of ethidium homodimer III (dead). At least 50 cells per sample were counted to determine the percent dead population and all samples were analyzed in triplicate. Viabilty of cell populations in culture was quantified by metabolic activity assay, measuring the conversion of [4,5-dimethylthiazol2-yl]-2,5-diphenyl tetrazolium bromide (MTT) to the formazan derivative using the Cell Growth Determination kit from Sigma (St. Louis, MO). The formazan derivative was quantified by measuring its absorbance at 570 nm with a Spectra Max 384 Plus plate reader. Caspase-3 activity was detected by the NucView fluorescence-based assay (Biotium Inc.) according to the directions supplied by the manufacturer.
To compare relative levels of ATP in vacuolated versus non-vacuolated cells, the cells were collected by trypsinization and assayed using the CellTiter Glo kit from Promega (Madison, WI). The luminescence produced by the ATP-dependent mono-oxygenation of luciferin by luciferase was normalized to the number of cells added to the assay, determined with a Coulter counter.
For colony forming assays cells were seeded in 60 mm dishes at 1,400 cells/dish. The day after plating, H-Ras(G12V) expression was induced in half of the cultures by inclusion of 1 μg/ml dox in the medium. Medium was replaced on all of the cultures every 2-3 days. Colony formation was assessed after three weeks by washing the cultures with PBS, fixing the cells for 10 minutes in ice-cold methanol and staining for 10 minutes with 1% crystal violet in 35% methanol. Colonies containing at least 50 cells were counted using a dissecting microscope.
Cell cycle distribution was determined from DNA histograms generated by flow cytometric analysis of cells prepared as described (17), using a Beckman-Coulter EPIC XL MCL cytometer. Data were analyzed using Multicycle DNA cell cycle analysis software (Phoenix Flow Systems, San Diego, CA).
DNA fragments were isolated by the method of Herrmann et al. (18) and resolved by electrophoresis in a 1.2% agarose gel. Images were obtained using the FluorChem HD2 system after staining the gel with ethidium bromide.
Cells were prepared for immunofluorescence as described previously (19). Myc-tagged proteins were detected with a monoclonal antibody (EMD Biosciences, San Diego, CA, USA) followed by goat anti-mouse IgG conjugated with Alexa Fluor™ 568 (Invitrogen, Carlsbad, CA,USA). For detection of endogenous LC3, we used the purified rabbit polyclonal antibody, APG8b (MAP1LC3B, N-Term) from Abgent, San Diego, CA, USA, followed by goat anti-rabbit IgG conjugated with Alexa Fluor 488. Antibodies used for immunofluorescence localization EEA1 and LAMP1 were obtained from Abcam (Cambridge, MA) and the Developmental Studies Hybridoma Bank (Ames, IA) respectively. Cells were examined with a Nikon Eclipse 800 fluorescent microscope equipped with a digital camera and ImagePro software (Media Cybernetics, Silver Spring, MD, USA) or with a Leica TCS SP5 multiphoton confocal microscope, using 488 nm and 561 nm laser excitation lines.
Cell pellets were fixed, dehydrated and infiltrated as described previously (19). Ultra thin sections were collected on copper 300-mesh support grids, stained with uranyl acetate and lead citrate, and examined using a Philips CM 10 transmission electron microscope.
Lucifer Yellow (LY) and dextran-Alexa Fluor (AF) 488 (10,000 mw) were purchased from Invitrogen/Molecular Probes (Eugene, Oregon). Cells were incubated with LY (1.5 mg/ml in Hanks balanced salt solution; HBSS) for 15 minutes in a 37°C, 5% CO2 incubator. The LY was removed and the cells were washed one time with HBSS. Phase contrast and fluorescent images were immediately taken of the live cells using an Olympus IX70 microscope equipped with a digital camera and SPOT imaging software. Uptake of LY was quantified by flow cytometry. Briefly, cells were grown in phenol-red-free medium, incubated with LY as described above, harvested by trypsinization and suspended in HBSS. For each sample 10,000 cells were analyzed with a Beckman-Coulter EPICS Elite ESP cytometer, with 488 nm excitation laser and 505-545 nm emission. Mean fluorescence intensity of the population was determined after subtraction of autofluorescence background obtained from parallel control samples incubated without LY. In some studies the cells were preincubated with Cytochalasin D (1 μM) (Sigma) for 30 min prior to the addition of LY.
For labeling with dextran-AF488, the cells were washed twice with phenol red-free DMEM containing 10% FBS, then incubated with 0.5 mg/ml dextran-AF488 in the same medium for the period of time indicated in the figure legends. The cells were washed two times in the same medium without the tracer, then images of the live cells were acquired as described for LY staining. In some cases, cells were co-labeled for 15 min with dextran-Af488 and human holo-transferrin conjugated to Alexa Fluor 594 (Invitrogen-Molecular Probes) added at 5 μg/ml in serum free DMEM.
Intracellular acidic compartments were labeled by incubating live cells at 37°C with Lysotracker Red DND-99 (Invitrogen) added to phenol red-free DMEM at a final concentration of 1 μM. Staining of intracellular compartments for cathepsin B activity was performed by incubating live cells with Magic Red ™ RR (Immunochemistry Technologies, Bloomington, MN) according to the directions supplied by the manufacturer.
U251 cells harbor mutations in the pro-apoptotic genes, PTEN and p53, and are widely used as a model for human glioblastoma(20). To begin a detailed characterization of the cellular phenotype triggered by expression of activated Ras in glioblastoma cells, we generated a stable U251 glioblastoma cell line (U251-C18) that exhibits tightly controlled conditional expression of myc-tagged H-Ras(G12V) in response to the addition of Dox (Fig. 1A). Our initial observations of these cells generally agreed with those reported by Chi et al. (15). That is, coincident with expression of myc-tagged Ras(G12V), the cells became filled with lucent cytoplasmic vacuoles that are readily detected by phase contrast microscopy (Fig. 1A). By day-6 there was a decrease in the number of viable cells in the cultures expressing H-Ras(G12V) compared with the controls (Fig. 1B). This coincided with noticeable cell rounding and detachment from the dish, with abundant floating debris suggestive of cellular disintegration. In cultures harvested on day-6 after addition of Dox, detached cells accounted for approximately 10% of the total cell population (not shown). Nearly 50% of these detached cells were nonviable when tested in a live/dead assay that measures hydrolysis of calcein acetoxy methylester (live) and uptake of ethidium homodimer III (dead) (Fig.1C). On the other hand, the attached cells in the cultures expressing active Ras scored positive for viability (Fig. 1C) and exhibited DNA histograms with S-phase populations similar to the control cells growing without Dox (Supplement, Fig. S1). Although they were extensively vacuolated, the attached cells collected on day-6 after addition of Dox showed no significant decline in ATP levels compared to the cells maintained without Dox (Fig. 1D). However, consistent with their reduced viability, the detached cells exhibited a marked reduction in ATP concentration (Fig 1D). Taken together, these findings imply that metabolic failure leading to cell death occurs abruptly at a late stage coincident with or after the detachment of the vacuolated cells. The long-term consequences of H-Ras(G12V) expression for cell survival are evident in colony-forming assays where addition of Dox resulted in an 85-90% reduction in the number of colonies (Fig. 1E).
The same pattern of vacuolization and cell degeneration was observed when Ras(G12V) was introduced into the U251 cells by retroviral infection (Supplement, Fig. S2) or by transient nucleofection (21). Seven additional independently-derived human glioma cell lines, some without mutations in PTEN (LN229) or p53 (U87MG, A172) (20), also exhibited a similar phenotype after expression of H-Ras(G12V) (Supplement Fig. S2 A). However, introduction of myc-H-Ras(G12V) into other commonly used cells (HeLa, HEp2, HEK293) did not cause vacuolization (Supplement Fig. S2 B). As previously reported, the inactive GDP-locked (15,21) or non-farnesylated (21) forms of Ras did not cause vacuolization, indicating that the cellular phenotype is directly related to the activation of specific Ras signaling pathways.
Electron microscopy of glioblastoma cells expressing H-Ras(G12V) revealed numerous electron-lucent vacuoles ranging from 0.5-2 μ in diameter, with some larger vacuoles reaching 7-8 microns (Fig. 2A). The vacuoles were generally devoid of cytoplasmic components or organelles, although some contained unidentified membranous inclusions or small quantities of amorphous electron-dense material (Fig. 2A). At high magnification the thickness of the membranes surrounding the vacuoles was estimated at 6-8 nm (Supplement Fig. S3), consistent with a single membrane. The large electron-lucent vacuoles were clearly distinct from smaller structures fitting the description of ‘classical’ autophagosomes (22) which have double membranes surrounding luminal cytoplasmic contents. (Fig. 2A, black arrows).
To further confirm that the Ras-induced vacuoles were not derived from autophagosomes, we performed immunofluorescence staining of the cells with an antibody against a well-established autophagosome marker, microtubule associated protein light-chain 3 (LC3). LC3 exists in a cytosolic form (LC3-I) and a form that is conjugated to phosphatidylethanolamine on autophagosome membranes (LC3-II) (23). The relative amount of LC3-II correlates with the number of autophagosomes induced by starvation and other stimuli (23). The LC3 antibody from Abgent, Inc. reacts predominantly with LC3-II on western blots. Therefore, we used this antibody to determine if LC3-II was localized to the vacuole membranes. As we have reported previously (21), myc-H-Ras(G12V) was localized in membranes surrounding the vacuoles (Fig. 2B). In contrast, LC3-II was detected in much smaller punctate structures (Fig. 2B). Confocal microscopy showed clearly that the large vacuoles circumscribed by myc-H-Ras(G12V) were separate from the LC3-II positive autophagosomes (Fig. 2C).
Because the preceding study was done with cells that were transiently nucleofected, we were able to compare the cells expressing myc-H-Ras(G12V) with the adjacent non-transfected cells (Fig. 2B, asterisks). We noticed that the punctate LC3 fluorescence was more intense in the transfected cells, suggesting that autophagosomes might accumulate separately from the phase-lucent vacuoles. To explore this possibility, we determined the relative amount of LC3-II in the stable U251-C18 cells with or without the expression of myc-H-Ras(G12V). As shown in Fig. 2D (left panel), expression of myc-H-Ras(G12V) was associated with a 2.7-fold increase in the amount of LC3-II, normalized to LDH. However, this could reflect either an increase in autophagosome biogenesis (stimulation of cellular macroautophagy pathways) or a decreased lysosomal turnover of LC3-II. An established method that can be used to distinguish between these possibilities is to compare the levels of LC3-II in the presence and absence of lysosomal protease inhibitors (24). As shown in Fig. 2D (right panel), addition of protease inhibitors to uninduced cells (−Dox) caused a 3.2-fold increase in the basal level of LC3-II, consistent with the expected impairment of lysosomal LC3-II turnover. However, when myc-Ras(G12V) was induced by addition of Dox, there was an additional 2-fold increase in LC3-II above the level caused by addition of protease inhibitors alone. This indicates that much of the increase in LC3-II induced by expression of Ras(G12V) is related to an increase in autophagosome formation, rather than a block in LC3-II turnover. Similar results were obtained when the cells were treated with rapamycin, an inhibitor of mTOR and a well-established inducer of macroautophagy (25) (not shown).
As shown in Fig. 3, expression of Ras(G12V) has identical effects on cell morphology and viability when expressed in a stable U251 cell line that we previously established as being resistant to pro-autophagic stimuli because of a knockdown of the autophagy protein, beclin-1 (16). This suggests that the increased autophagic activity detected in Fig. 2 is not required for the death of the glioblastoma cells expressing activated Ras(G12V), and that cellular degeneration is most likely related to the progressive accumulation of the non-autophagic phase-lucent vacuoles.
In considering possible origins for the phase-lucent vacuoles, we noted a previous study in which activated Ras was shown to stimulate macropinocytosis in fibroblasts (26). Macropinocytosis is a process whereby cells internalize extracellular fluid trapped beneath projections of the plasma membrane termed ruffles or lamellipodia (27). Macropinosomes typically appear as phase-lucent vesicles ranging in diameter from 0.5 μ to 5 μ. Rapid incorporation of extracellular fluid phase tracers is a hallmark of macropinosomes. When we added Lucifer Yellow (LY) to the medium, cells expressing H-Ras(G12V) incorporated the tracer into many of the phase lucent vacuoles within 10 min (Fig. 4A). Quantification of LY internalization by flow cytometry demonstrated a 3-fold increase in uptake of the tracer into the +Dox cells expressing Ras(G12V), compared with the basal level of LY uptake in the −Dox controls (Fig. 4B). Preincubation with Cytochalasin D (CytoD), which disrupts the actin cytoskeleton involved in the formation of lamellipodia, had no effect on LY uptake in the −Dox cells, suggesting that most of the LY uptake in these cells is due to basal activity of the endocytic pathway. In contrast, addition of CytoD caused a 50% reduction of LY incorporation in the +Dox cells, after subtraction of the basal uptake attributed to endocytosis (−Dox, +CytoD) (Fig. 4B).
Electron micrographs revealed numerous lamellipodia closing around regions of extracellular fluid to form nascent macropinosomes in the cells expressing H-Ras(G12V) (Fig. 4C). Control cells that were not expressing Ras also contained some lamellipodia, but closure of these structures to form enlarged macropinosomes was not evident (not shown).
In addition to labeling macropinosomes, fluid phase tracers can enter early endosomes. Macropinosomes lack a clathrin coat and can be distinguished from endosomes by their comparative inability to concentrate receptors (28). Therefore, to confirm that the vacuoles were derived from macropinosomes, cells expressing Ras(G12V) were subjected to short-term incubation with a bulk fluid-phase tracer, dextran-AF488, together with a ligand for the transferrin receptor, transferrin-AF594. The larger vesicles containing fluorescent dextran were distinct from much smaller endosomes that sequestered transferrin (Fig. 4D). In accord with this finding, we observed that the phase-lucent vacuoles were separate from smaller punctate structures detected by immunofluorescence with an antibody against the well known early endosomal protein, EEA1 (Fig. 4E). These findings, coupled with the morphological evidence in Fig. 4C, support the identification of the Ras-induced vacuoles as macropinosomes.
The lucent vacuoles induced by activated Ras are morphologically distinct from lysosomes and autolysosomes, which typically contain electron-dense organelle remnants or degraded cytoplasmic components (29) (Figs. (Figs.2A2A and and4C).4C). However, on the basis of morphology alone it was difficult to rule out the possibility that some of the Ras-induced vacuoles might be swollen late endosomes, similar to those observed in cells where morphogenesis of multivesicular endosomes is disrupted by inhibiting the class-III PI 3-kinase, Vps34 (19,30). Because the latter retain the acidic characteristic of late endosomes, they readily incorporate lysosomotrophic agents (19). Therefore, to test the possibility that some of the Ras-induced vacuoles might be late endosomal compartments, we performed supravital staining with Lysotracker Red™. As shown in Fig. 5A, there was no clear overlap between the phase-lucent vacuoles and the compartments labeled with Lysotracker Red. Similar results were obtained when the cells were stained with acridine orange, which is sequestered in late endosomes, lysosomes, and autolysosomes (31,32) (data not shown). Additionally, there was no substantial overlap between the phase-lucent vacuoles and compartments labeled with Magic Red™RR, a cell permeable peptide substrate that fluoresces when cleaved by cathepsin B (Fig. 5B). Taken together, these results support the conclusion that the majority of the phase-lucent vacuoles are derived from macropinosomes rather than late endosomes or lysosomes.
Although the vacuoles did not stain with markers for acidic or cathepsin-positive compartments (Fig. 5A & B), we found that many of them contained LAMP1, a membrane protein typically associated with lysosomes and late endosomes (Fig. 5C). Two possible models could explain this. In the first model, macropinosomes may fuse with late endosomes or lysosomes, acquiring LAMP1 and simultaneously neutralizing the interior of these compartments so that they cannot be detected with acidophilic agents or cathepsin substrates. An alternative model is suggested by reports that LAMP1 can traffic directly to non-lysosomal compartments like early endosomes (33) or nascent phagosomes (34). Thus, the macropinosomes may remain separate from lysosomes while recruiting LAMP1 directly to their membranes. To discriminate between these models, we pre-labeled the lysosomal compartments of vacuolated glioblastoma cells by incubating them with Lysotracker Red for 3 h. Then, after removing the Lysotracker from the medium, we added fluorescent dextran for 4 h to determine if the dextran-labeled compartments would merge with the pre-labeled lysosomes (Fig. 5D). After 4 h we detected merger of some of the smaller dextran-labeled structures with the Lysotracker-positive compartments, presumably representing the fusion of endosomes with lysosomes. However, even after this extended period, the larger dextran-labeled vacuoles appeared to remain separate from the pre-labeled lysosomal compartments (Fig. 5D). Similar results were observed when the cathepsin substrate, Magic Red™ RR, was used to pre-label the lysosomes (not shown). These observations are consistent with the concept of direct recruitment of LAMP1 to the membranes of macropinocytotic vacuoles, with minimal fusion between these compartments and lysosomes.
The mechanisms underlying macropinocytosis are poorly understood, but previous studies have implicated the Rac1 GTPase and its effector, PAK1, as key regulators of this process (35,36). Since downstream targets of Ras include guanine nucleotide exchange factors that can stimulate activation of Rac1 (e.g., Tiam1) (37), we hypothesized that Rac1 might be positioned downstream from Ras in the pathway that triggers macropinosome accumulation in glioblastoma cells.
To test this possibility we asked if expression of a constitutively active form of Rac1 could mimic the effects of Ras(G12V) in U251 glioblastoma cells. As shown in Fig. 6A &B, conditional expression of activated myc-Rac1(G12V) in U251 cells triggered a vacuolar phenotype closely resembling that observed with Ras(G12V). As in the case of cells expressing Ras(G12V), the viability of the cells expressing Rac1(G12V) declined between the fourth and eighth days after addition of Dox, coincident with extreme vacuolation and cell detachment (Fig. 6C). Moreover, the cells expressing Rac1(G12V) exhibited a substantial increase in the uptake of LY into the vacuolar structures (Fig. 6D). We have previously reported that activated forms of other Rho-family GTPases (e.g., Cdc42 and RhoA) do not cause vacuolation of U251 cells (21). Thus, the hyperstimulation of macropinocytosis appears to be a specific effect of Ras(G12V) and Rac1(G12V).
In their initial report describing Ras-induced death of glioblastoma cells, Chi et al. (15) found no evidence for caspase activation in the vacuolated cells. However, in light of our observation that loss of cell viability coincides with detachment from the substratum (Fig. 1C), we re-examined this question by evaluating the cleavage of caspase substrates in both the attached and detached cell populations. As shown in Fig. 7A , there was no cleavage of PARP or lamin A/C in the attached, mostly viable, vacuolated cells. However, in the detached cells fragments of PARP and lamin A/C were detected at molecular weights consistent with caspase-3 cleavage. The sizes of these fragments were identical to those observed in cells treated with staurosporine, a known inducer of apoptosis (Fig. 7A). Examination of the portion of the PARP blot below 75 kDa did not reveal any 50 kDa fragments indicative of degradation by lysosomal proteases (38). In accord with the PARP cleavage, 49.6% of the detached cells stained positive for caspase-3 activity (Supplement, Fig. S4 A), mirroring the percentage of non-viable cells in the detached population (Fig. 1C). Agarose gel electrophoresis revealed no evidence of DNA fragmentation in the attached cells, but the DNA recovered from the detached cells was extensively degraded, with detectable laddering suggestive of nucleosomal DNA fragmentation (Supplement, Fig. S4 B).
To determine if death of the cells expressing Ras(G12V) was dependent on caspase activation, we added the broad-spectrum caspase inhibitor, zVAD-fmk, during the critical period (days 4-6) when loss of cell viability begins to occur. The PARP blots in Fig. 7B indicate that zVAD was highly effective in blocking caspase activation. However, this did not prevent vacuolization (not shown) or loss of cell viability (Fig. 7C). Thus, although activation of caspases occurs in conjunction with the demise of the glioblastoma cells, this is not an obligatory feature of the death mechanism. In separate studies we also tested the ability of cathepsin and calpain inhibitors to preserve the viability of the vacuolated cells (Supplement Fig. S5). Consistent with the absence of alternative 50 kDa PARP cleavage products, the lysosomal protease inhibitors were ineffective in preventing cell death induced by expression of Ras(G12V).
In light of the foregoing observations, we examined the morphology of the detached glioblastoma cells to determine if these cells exhibit typical features of apoptosis (Fig. 7D). Electron microscopy revealed that at least 80% of the detached cells contained numerous large cytoplasmic vacuoles with morphology similar to the macropinosomes described earlier in the attached cells (Figs. (Figs.2A2A and and4C).4C). The cells were generally swollen to 20-30 μ diameter, compared to the attached cells, which typically ranged 10-15 μ. In about half of the cell population the expansion of vacuoles was so extreme that these structures filled most of the cytoplasmic space (Fig. 7D). While some cells contained numerous vacuoles of various sizes (Fig 7D, panel i), others contained only a few very large vacuoles (panels ii & iii), suggestive of an end-stage coalescence of these structures. In addition to the distorted cells with intact peripheral membranes, there were numerous remnants of cells that had ruptured (Fig. 7D, panel iv). However, even in the severely vacuolated or ruptured cells, the nuclei were generally intact and contained diffuse chromatin and a prominent non-fragmented nucleolus. These observations indicate that the morphological features of the dying glioblastoma cells resemble necrosis-like forms of cell death rather than classical apoptosis.
The resistance of many types of cancer cells to apoptosis has stimulated interest in identifying non-apoptotic cell death pathways that might be targeted to slow tumor progression. A number of distinct non-apoptotic forms of cell death have now been characterized. These include: type-II or autophagic cell death (5,7,12), paraptosis (39,40), oncosis (41-43), and necroptosis (44,45). Even the term necrosis, previously used to indicate ‘passive’ cell death or the post-mortem state of cells (41), has more recently been used to describe forms of programmed cell death that involve progressive lysosomal damage, leakage of lysosomal proteases, and early disruption of the cell membrane (46-49). In the present study we provide a detailed characterization of a novel form of non-apoptotic cell death observed in glioblastoma cells upon constitutive stimulation of Ras signaling pathways. The hallmark cytopathological feature of this form of cell death is the marked accumulation of large fluid filled vacuoles derived from macropinosomes. Electron microscopy demonstrates a correlation between cellular disintegration and a progressive increase in the number and size of the macropinocytotic vacuoles. These observations are highly suggestive of a causal relationship between the dysregulation of macropinocytotic fluid uptake and the eventual metabolic collapse and rupture of the cells. Final proof of this interrelationship must await the development of new approaches for long-term inhibition of macropinocytosis, because the drugs currently used to block this process (e.g., Ameloride, Cytochalasins) are toxic when applied to cultured cells for more than a few hours (50).
In contrast with an earlier report (15), we discovered that by examining detached cells expressing Ras(G12V) we could in fact detect caspase activation and DNA fragmentation. However, similar to the situation in most other forms of non-apoptotic death, caspase activation does not seem to be an obligatory step in the Ras-induced death program. Furthermore, our studies of cellular morphology did not show typical cell shrinkage, blebbing, and nuclear chromatin condensation observed in apoptotic cells. Therefore, our results support the classification of Ras-induced macropinocytotic cell death as non-apoptotic. As discussed below, comparison with other known types of non-apoptotic death suggests that macropinocytotic cellular degeneration represents a unique form of cell death.
Autophagic death is now the most widely recognized type of non-apoptotic cell death. The diagnostic feature of this form of death is the proliferation of autophagosomes and autolysosomes that engulf cytoplasm and organelles and cannibalize the cell (5,12). In cells expressing activated Ras, the large macropinocytotic vacuoles that eventually fill the degenerating cells are morphologically distinct from autophagosomes. Although autophagosomes seem to accumulate in parallel with the macropinocytotic vacuoles, our studies with beclin-1 knockdown cells suggest that macropinocytotic vacuolization and cell death induced by Ras(G12V) can occur independent of the autophagy machinery. Thus, in this case, autophagy may reflect an attempt of the cells to survive under the adverse metabolic conditions created by rampant macropinosome accumulation, rather than a direct cause of cell death. This would be consistent with accumulating evidence that autophagy can function as a protective strategy against apoptosis or necrosis in cells subjected to metabolic stress (14).
The cytopathology induced by activated Ras or Rac is also distinct from several lesser known forms of cell death. Necroptosis can be triggered by stimulation of death receptors under conditions where caspases are inhibited (44,45). Cell swelling and membrane rupture occur, but a massive increase in vesicular fluid uptake is not a diagnostic feature of necroptosis. Moreover, we have observed that necrostatin, a potent inhibitor of necroptosis, does not impede Ras-induced vacuolization or cell death (unpublished). Oncosis is a form of caspase-independent death typically caused by ischemia or disruption of ion pumps (43,51). As in the case of the Ras-induced cell death, oncosis can include cell swelling, vacuolization, and membrane rupture. Cytoplasmic vacuolization also occurs in another distinctive form of cell death termed paraptosis (39). However, in both oncosis and paraptosis the vacuoles are derived mainly from distended endoplasmic reticulum and/or mitochondria rather than macropinosomes. Finally, although cytoplasmic vacuolization is sometimes mentioned as a feature in various forms of necrosis-like cell death involving lysosomal damage, we have not detected alternative 50 kDa PARP cleavage products that would signal the leakage of lysosomal cathepsins associated with this form of necrosis (38). Nor have we observed any mitigation of Ras-induced cell degeneration by treating glioblastoma cells with cathepsin or calpain inhibitors. In light of these differences with other types of cell death, and the unique association of cellular degeneration with hyperstimulation of macropinocytosis (cell drinking), we propose that this form of cell death be named “methuosis” (methuo, from the Greek; to drink to intoxication). Table 1 summarizes the features of methuosis in comparison with other reported forms of cell death.
Activating mutations in Ras have long been regarded as oncogenic because they result in chronic stimulation of signaling pathways important for cell proliferation (52). Activated Ras may also contribute to tumor progression by protecting transformed cells from apoptosis, although some reports have described opposite pro-apoptotic functions for Ras (53). The stimulation of a non-apoptotic death mechanism by activation of Ras highlights a relatively unexplored aspect of Ras signaling pathways. Our finding that activated Ras can induce vacuolization and cellular degeneration in a variety of human glioma cell lines, including those like U251 and T98G which harbor PTEN and p53mutations that render them relatively resistant to apoptosis, suggests that the presence of Ras-responsive pathways capable of hyperstimulating macropinocytosis and/or inhibiting the clearance or recycling of macropinosomes, may be a general feature of human glioblastoma. If so, further delineation of the relevant signaling mechanisms could suggest ways to manipulate this pathway to trigger cell death in these intractable tumors. In this regard, our previous studies have indicated that stimulation of vacuolization in glioblastoma cells does not depend on conventional Ras effectors such as Raf, PI 3-Kinase, and Ral-GDS (21). In considering alternative possibilities, it is noteworthy that the Rac1 GTPase has been implicated as a regulator of macropinocytosis (35). Indeed, as we have shown here, expression of activated Rac1(G12V) in glioblastoma cells can mimic the effects of Ras(G12V). Since downstream targets of Ras include guanine nucleotide exchange factors like Tiam1, which can stimulate activation of Rac1 (37), our working hypothesis is that these exchange factors may be the key Ras effectors involved in transmitting signals via Rac to the macropinocytotic machinery.
Our studies comparing glioblastoma cells to HeLa, HEK293 and HEp2 cells suggest that there is definite cell-type specificity in the ability of Ras to stimulate methuosis. Similar conclusions were drawn by Chi et al (15), who observed Ras-induced vacuolization in glioblastoma and two gastric cancer cell lines, but not in bladder carcinoma cells. Thus, obtaining a better understanding of the basis for the particular sensitivity of certain cell types to methuosis will be important for evaluating the therapeutic potential of this form of cell death. Our results with dextran tracers (Fig. 5D) indicate that, unlike normal macropinosomes, the Ras-induced vacuoles do not dissipate or fuse with lysosomal compartments after they are internalized. This raises the possibility that the explanation for the differential sensitivity to methuosis could reside not only at the level of induction of macropinocytosis, but also at the level of intracellular trafficking or membrane channel function.
Another important question for future study is whether stimuli other than ectopic expression of Ras or Rac can provoke methuosis. In this regard it is worth noting that cytoplasmic vacuolization is often mentioned as a morphological feature of necrotic cell death caused by cytotoxic drugs or adverse environmental conditions, but there is seldom any indication as to the specific origin of the vacuoles. Thus, it is conceivable that dysregulation of macropinocytosis may be a common occurrence in forms of cell death labeled as necrosis, and it may therefore be more widespread than previously recognized.
Figure S1. Expression of Ras(G12V) does not alter the cell cycle distribution of attached vacuolated cells. Stable U251-C18 cells were grown with or without Dox, as in Fig. 1. Attached cells from three parallel cultures were harvested on day-6 and cell cycle distribution was determined by flow cytometry as described in Materials and methods. Results are means (± S.D.) of separate determinations performed on three parallel cultures.
Figure S2. Activated H-Ras induces a vacuolar phenotype and cellular degeneration in multiple human glioma cell lines. (A) The indicated cell lines were infected with retrovirus expressing myc-H-Ras(G12V) or myc-H-Ras(S17N) and cells were photographed at 300× magnification. After three days the cells expressing Ras(G12V), but not those expressing Ras(S17N), began to detach from the dish and undergo lysis. (B) myc-H-Ras(G12V) was introduced into U251, HEK293, HeLa or HEp2 cells by nucleofection and cells were examined by phase contrast or immunofluorescence microscopy 24 h later.
Figure S3. Electron microscopy shows that the Ras-induced vacuoles are delimited by a single membrane. (A) U251 cells were infected with myc-H-Ras(G12V) retrovirus to induce vacuoles and then examined by electron microscopy after two days. The scale bar represents 1 μ. (B) The area outlined in panel A, viewed at higher magnification. The bar represents 0.1 μ.
Figure S4. Detached cells collected after induction of H-Ras(G12V) contain activated caspase-3 and show signs of nucleosomal DNA degradation. (A) U251-C18 cells were collected between days 4-7 after Dox was added to stimulate expression of H-Ras(G12V). The cells were stained with a caspase substrate that fluoresces when cleaved by caspase-3. Counts performed on three separate cultures (80-90 cells per culture) indicated that 49.6 ± 7.5% of the detached cells were positive for caspase-3 activity. There were no caspase-3 positive cells in the attached population (not shown). (B) Cells were maintained with or without Dox as described in Fig. 1. Detached cells were collected between days 4-7. Attached cells were harvested on day-7. The agarose gel shows DNA isolated from the attached and detached cells under conditions that favor extraction of DNA fragments, as described in Materials and methods.
Figure S5. Calpain and cathepsin inhibitors do not protect U251 cells from Ras-induced death. Cells were seeded in 96-well plates at 5,000 cells/well and maintained with or without Dox as described in Fig. 1. Starting on day-5, the −Dox and +Dox cultures were changed to medium with either 25 μM ALLN, an inhibitor of calpains, cathepsins B and L, and neutral cysteine proteases (EMD Chemicals, San Diego, CA), or 100 μM PD150606, a selective calpain inhibitor (EMD Chemicals), reconstituted in DMSO. The remaining +Dox and −Dox cultures received DMSO alone. On Day-7 the viability of the cells in the +Dox cultures (with or without protease inhibitors) was compared with the corresponding −Dox controls using the MTT assay. Each result is the mean ± SD of assays performed on four wells.
We thank Thomas Sawyer and Karen Domenico for help with flow cytometry, and William Gunning, Ph.D and Michelle Lewandowski for assistance with electron microscopy.
Grant Support: This work was supported by National Institutes of Health grants R01 CA34569 and R01 CA115495, and by a grant from the Charlotte Geyer Foundation.