Beclin 1 was originally identified as a novel Bcl-2-interacting protein, but co-immunoprecipitation studies suggest that the major physiological partner for Beclin 1 is the mammalian class III phosphatidylinositol 3-kinase (PI 3-kinase) Vps34. Beclin 1 has been proposed to function as a tumor suppressor by promoting cellular macroautophagy, a process that is known to depend on Vps34. However, an alternative role for Beclin 1 in modulating normal Vps34-dependent protein trafficking pathways has not been ruled out. This possibility was examined in U-251 glioblastoma cells. Immunoprecipitates of endogenous Beclin 1 contained human Vps34 (hVps34), but not Bcl-2. Suppression of Beclin 1 expression by short interfering (si)RNA-mediated gene silencing blunted the autophagic response of the cells to nutrient deprivation or C2-ceramide. However, other PI 3-kinase-dependent trafficking pathways, such as the post-endocytic sorting of the epidermal growth factor receptor (EGFR) or the proteolytic processing of procathepsin D en route from the trans-Golgi network (TGN) to lysosomes, were not affected. Depletion of Beclin 1 did not reduce endocytic internalization of a fluid phase marker (horseradish peroxidase, HRP) or cause swelling of late endosomal compartments typically seen in cells where the function of hVps34 is impaired. These findings argue against a role for Beclin 1 as an essential chaperone or adaptor for hVps34 in normal vesicular trafficking, and they support the hypothesis that Beclin 1 functions mainly to engage hVps34 in the autophagic pathway.
Animal cells utilize macroautophagy as a mechanism for turnover of long-lived proteins, and as a survival strategy under conditions of amino acid deprivation (Dunn, 1994; Klionsky and Emr, 2000). Autophagic vacuoles (autophagosomes) are initially formed from membranes of the endoplasmic reticulum (ER) that surround a region of cytoplasm (Dunn, 1990a; Dunn, 1990b). These structures, bounded by a double membrane, then develop into mature degradative vacuoles (autolysosomes) by progressive fusion with late endosomes and lysosomes (Gordon and Seglen, 1988; Dunn, 1990a; Lawrence and Brown, 1992). Accumulation of autophagosomes and autolysosomes is a hallmark morphological feature of type II programmed cell death, also referred to as autophagic cell death (Lockshin and Zakeri, 2004; Levine and Klionsky, 2004). This type of cell death has been postulated to occur during embryonic development in connection with tissue remodeling (Zakeri et al., 1995). More recently, it has also been described in neurodegenerative diseases (Kegel et al., 2000; Larsen and Sulzer, 2002) and in tumor cells exposed to anti-neoplastic agents (Bursch et al., 1996; Paglin et al., 2001; Kanzawa et al., 2003; Kanzawa et al., 2004). However, it remains unclear whether the accumulation of autophagosomes is a direct cause of cell death or a symptom of an unsuccessful attempt of cells to adapt to metabolic stress.
Beclin 1 (hereafter referred to simply as Beclin) is a 60 kDa protein that has been implicated as an important regulator of macroautophagy. It was originally discovered during the course of a yeast two-hybrid screen of a mouse brain cDNA library using human Bcl-2 as the bait (Liang et al., 1998). The human beclin gene has been mapped to a region of chromosome 17q21 that is monoallelically deleted in many breast, ovarian and prostate cancers (Aita et al., 1999). Expression of Beclin in MCF7 mammary carcinoma cells increases their autophagic response to nutrient deprivation (Liang et al., 1999). Consistent with this observation, several additional studies have implicated beclin as an essential gene for cell survival under adverse nutritional conditions. Deletion of the gene encoding Vps30 (Atg6), the Beclin ortholog in Saccharomyces cerevisiae, hastens cell death induced by starvation (Kametaka et al., 1998). Similarly, suppression of Beclin expression in mammalian cells impairs autophagy and sensitizes the cells to starvation-induced apoptosis (Boya et al., 2005). In addition to its specific role in adaptation to nutrient deprivation, accumulating evidence suggests that Beclin might play a more general role in cell survival during embryonic development. In Caenorhabditis elegans, short interfering (si)RNA-mediated suppression of bec-1 inhibits autophagy and interferes with morphogenesis of the developmental stage known as the dauer diapause. The latter represents a survival strategy that can be triggered by unfavorable environmental conditions (Melendez et al., 2003). In mice with homozygous deletions of beclin, death occurs at an early stage of embryonic development (Qu et al., 2003; Yue et al., 2003). Studies with beclin–/– embryonic stem cells have shown that they fail to form normal embryoid bodies in vitro (Yue et al., 2003). Finally, overexpression of Beclin in the brains of adult mice protects neurons from apoptosis induced by Sindbis virus infection (Liang et al., 1998). The specific mechanism of this protective effect and its relationship to autophagy remains to be established.
In contrast to the foregoing indications that Beclin functions to promote cell survival, there is also a growing body of evidence that Beclin might, paradoxically, also function as a tumor suppressor under specific circumstances. For example, augmentation of Beclin expression in MCF7 cells decreases their proliferation, clonigenicity in soft agar, and tumorgenicity in nude or severe combined immunodeficient (scid) mice (Liang et al., 1999; Furuya et al., 2005). Heterozygous disruption of beclin in mice results in an increased frequency of spontaneous lymphomas, as well as lung and liver carcinomas (Qu et al., 2003; Yue et al., 2003). The possibility that the tumor-suppressing effects of Beclin might be related to a role in regulating autophagic cell death has been raised by two recent studies. In the first, silencing Beclin expression in L929 cells prevented autophagic death triggered by treatment of the cells with a caspase inhibitor (Yu et al., 2004). In the second, interference with Beclin expression blocked non-apoptotic (autophagic) cell death induced by etoposide treatment of Bax/Bak–/– double-knockout mouse cells that are resistant to apoptosis (Shimizu et al., 2004).
Despite the apparent importance of Beclin in the regulation of macroautophagy, cell survival and non-apoptotic cell death, little is known about the molecular mechanisms that are involved. Studies of Vps30 (Atg6), the Beclin ortholog in S. cerevisiae, indicate that it is part of two distinct protein complexes that contain the PI 3-kinase Vps34 (Kihara et al., 2001b). One complex functions in post-Golgi sorting of proteases to the vacuole (equivalent to the lysosome), whereas the other complex is essential for macroautophagy and degradation of cytoplasmic proteins under starvation conditions. The importance of Vps34 in protein trafficking is related to the role of its product, phosphatidylinositol 3-phosphate [PtdIns(3)P], in the membrane recruitment of proteins involved in the vesicle docking and fusion machinery (Wurmser et al., 1999). These proteins typically contain specific PtdIns(3)P-binding domains termed the FYVE finger (Fruman et al., 1999; Stenmark et al., 2002) or the Phox homology domain (Song et al., 2001; Cheever et al., 2001; Xu et al., 2001; Kanai et al., 2001), and their roles in vesicular transport have been reviewed in recent articles (Simonsen et al., 2001; Deneka and van Der, 2002).
Mammalian cells express a homolog of the yeast Vps34 PI 3-kinase (Volinia et al., 1995). Like its yeast counterpart, the mammalian Vps34 appears to be required for the initiation of macroautophagy (Petiot et al., 2000; Eskelinen et al., 2002). However, studies using PI 3-kinase inhibitors, a dominant-negative form of Vps34, or antibody microinjection have also implicated mammalian Vps34 in normal protein trafficking pathways such as the delivery of proteases from the trans-Golgi network (TGN) to the lysosomes (Row et al., 2001), endocytic trafficking and sorting of cell-surface receptors (Siddhanta et al., 1998; Tuma et al., 2001; Petiot et al., 2003), and the formation of internal vesicles in multivesicular endosomes (MVEs) (Futter et al., 2001). Most recently, we have observed that siRNA-mediated knockdown (KD) of human Vps34 (hVps34) expression causes vacuolation of late endosomal compartments, impedes the trafficking of cathepsin D from late endosomes to lysosomes, and slows the lysosomal degradation of activated epidermal growth factor receptor (EGFR) in U-251 glioblastoma cells (E. E. Johnson, J.H.O., W. T. Gunning and W.A.M., unpublished data).
Since hVps34 can be immunoprecipitated together with Beclin (Kihara et al., 2001a), it is important to consider the possibility that Beclin might influence cell growth and tumorgenicity not only by controlling macroautophagy, but also by functioning as a chaperone or adaptor for hVps34 in normal protein trafficking pathways that depend on the production of PtdIns(3)P. A recent study suggests that Beclin is not required for post-Golgi trafficking of lysosomal hydrolases in MCF7 cells, based on the observation that maturation of cathepsin D is unaffected in cells overexpressing a mutant form of Beclin that cannot bind hVps34 (Furuya et al., 2005). In the present study, we conducted a more comprehensive assessment of the potential role of Beclin in hVps34-dependent protein trafficking pathways using siRNA-mediated gene silencing to suppress Beclin expression chronically in U-251 glioblastoma cells. Near-complete suppression of Beclin expression markedly attenuated the ability of the cells to initiate macroautophagy but had no effect on cell growth, endosomal morphology, endocytic uptake of a fluid phase marker, internalization and degradation of the EGFR, or delivery of cathepsin D to late endosomes and lysosomes. These studies provide new insight into the biological significance of the interaction between Beclin and hVps34 PI 3-kinase by showing that it is essential for engagement of hVps34 in the process of macroautophagy, but is dispensable for the normal function of hVps34 in endocytic trafficking or lysosomal enzyme sorting.
Generation of stable Beclin KD cell lines
To obtain a cell population in which expression of Beclin was specifically suppressed, cells were infected with a replication-deficient retroviral vector that drives the expression of RNAi sequences and confers puromycin resistance on infected cells (Brummelkamp et al., 2002). Vectors were engineered to contain either an inverted repeat stem-loop sequence matching a unique region of the human beclin mRNA, or a `control' sequence that did not match any known GenBank entry. The newly synthesized hairpin RNA is processed into siRNA, triggering the cellular Dicer-mediated degradation of the target beclin RNA (Sui et al., 2002). In preliminary tests with several cell lines infected with a green fluorescent protein (GFP) reporter construct, the human U-251 glioblastoma cell line showed high initial infection efficiency. Therefore, we chose to use this cell line for studies of Beclin. As shown by the immunoblots in Fig. 1, expression of Beclin was almost undetectable in puromycin-resistant cells that received the Beclin KD vector compared with cells that were infected with the control vector. Expression of unrelated soluble and membrane-bound proteins [e.g. lactate dehydrogenase (LDH), LAMP1] was not reduced, indicating that the loss of Beclin expression was not due to a general effect of the siRNA on protein synthesis in the KD cells (Fig. 1). In all of the experiments described in this paper, similar immunoblot results were obtained, verifying that expression of Beclin was decreased by at least 90-95% relative to the parallel control cultures.
Endogenous Beclin forms a complex with hVps34 in U-251 glioblastoma cells
Previous studies have indicated that Beclin can be co-immunoprecipitated with mVps34 in HeLa cells (Kihara et al., 2001a). However, Beclin has also been described as a Bcl-2- and Bcl-XL-interacting protein, based on yeast two-hybrid assays and fluorescence resonance energy transfer (FRET) analysis of co-expressed proteins (Liang et al., 1998). To determine if these proteins are normal endogenous binding partners for Beclin in U-251 glioblastoma cells, Beclin was immunoprecipitated from whole-cell lysate and the associated proteins were probed by immunoblot analysis with antibodies against hVps34, Bcl-2 or Bcl-XL. Although hVps34, Bcl-2 and Bcl-XL were all readily detected in the cell lysates, only hVps34 was co-precipitated with Beclin (Fig. 2A). In accord with these studies, the converse immunoprecipitation of endogenous Bcl-2 did not pull-down any detectable endogenous Beclin (Fig. 2B). Immunoprecipitation with the antibody against hVps34 was not possible because the latter was not a good precipitating reagent.
Suppression of Beclin expression interferes with macroautophagy
To assess the consequences of knocking down Beclin for the induction and progression of macroautophagy in U-251 cells, we subjected the cells to two established pro-autophagic stimuli: treatment with C2-ceramide (Scarlatti et al., 2004) and nutrient deprivation (Klionsky et al., 2000; Levine et al., 2004). Microtubule-associated protein light-chain 3 (LC3) was used as a molecular marker to monitor autophagosome biogenesis. LC3 is the mammalian homolog of the yeast autophagy protein Atg8. Like Atg8, LC3 exists in a cytosolic form (LC3-I) and a form that is conjugated to phosphatidylethanolamine (PE) on autophagosome membranes (LC3-II) (Kabeya et al., 2000; Tanida et al., 2002). Initially, we attempted to compare autophagosome biogenesis in control versus Beclin KD cells by determining the subcellular localization of ectopically expressed GFP-LC3 by fluorescence microscopy of transiently transfected cells. However, the results of this assay were difficult to interpret, owing to the low transfection efficiency of U-251 cells and the tendency of the overexpressed GFP-LC3 to associate with punctate structures even under normal culture conditions. An alternative approach that avoids overexpression of LC3 entails measuring the ratio of endogenous LC3-II to LC3-I by immunoblot analysis. It is well established that the conversion of LC3-I to LC3-II is closely correlated with the formation of autophagosomes (Kabeya et al., 2000; Tanida et al., 2004). Because LC3-II associates specifically with the nascent autophagosome isolation membrane and remains on the autophagosome until it matures to an autolysosome, accumulation of LC3-II is now viewed as a definitive marker for activation of the autophagic pathway (Mizushima, 2004; Kirkegaard et al., 2004). As shown in Fig. 3A, nutrient deprivation caused a threefold increase in the total amount of LC3-II and the ratio of LC3-II to LC3-I in the control U-251 cells. However, there was comparatively little change in LC3-II in the Beclin KD cells. Ceramide treatment caused an even greater (sevenfold) increase in LC3-II in the control cells, with a markedly attenuated response again seen in the Beclin KD cells (Fig. 3B).
Previous studies have determined that macroautophagy can be assessed by measuring the sequestration of cytosolic enzymes into membrane-bound compartments (Kopitz et al., 1990; Stromhaug et al., 1998). When the U-251 cells were starved for 4 hours in Hanks balanced salt solution (HBSS), we observed a doubling of the amount of LDH sequestered in membrane compartments (Fig. 4). By contrast, the Beclin KD cells responded to nutrient deprivation with a much smaller change in LDH sequestration (Fig. 4).
The maturation of autophagosomes to autolysosomes is accompanied by a loss of LC3-II and an increase in the acidity of the lumen (Mizushima, 2004). Therefore, to assess the relative number of autolysosomes in control versus Beclin KD cells, we used an assay that measures supravital staining of acidic compartments with the lysosomotropic agent acridine orange (AO). When the dye enters an acidic compartment, the protonated form becomes trapped in aggregates that fluoresce bright red or orange (Traganos and Darzynkiewicz, 1994; Paglin et al., 2001; Kanzawa et al., 2003). Although AO stains lysosomal and late endosomal compartments as well as autolysosomes, extensive studies have established that a substantial increase in AO-positive acidic vesicular organelles (AVOs) occurs in conjunction with the induction of macroautophagy in glioblastoma cells (Kanzawa et al., 2003; Kanzawa et al., 2004; Komata et al., 2004; Takeuchi et al., 2005). As shown in Fig. 5A, a general increase in the intensity of AO-positive structures could be detected in ceramide-treated control cells, but not in the parallel ceramide-treated Beclin KD cells. To obtain a more accurate evaluation of the amount of AO sequestered in acidic compartments, the cells were lysed and the red fluorescence emanating from AO was quantified and normalized to DNA (ethidium bromide fluorescence) (Fig. 5B). The results confirmed that ceramide treatment stimulated a large increase in the amount of AO sequestered into AVOs in control cells, and that the response to ceramide was greatly reduced in the Beclin KD cells.
Protein trafficking from the TGN to the lysosomes in the Beclin KD cells
Previous studies have indicated that treatment of cells with the PI 3-kinase inhibitor wortmannin causes a block in trafficking of procathepsin D from the TGN to the late endosomes and lysosomes (Davidson, 1995; Brown et al., 1995). Similar effects have been reported in cells expressing a kinase-deficient form of rat Vps34 (Row et al., 2001). Most recently, we have determined that siRNA-mediated suppression of hVps34 expression in U-251 cells results in marked accumulation of the late endosomal intermediate form of cathepsin D, with reduced production of the mature lysosomal form (E. E. Johnson, J.H.O., W. T. Gunning and W.A.M., unpublished data). In light of the association between Beclin and hVps34, we wished to determine if cells lacking normal amounts of Beclin would exhibit any defects in this PI 3-kinase-dependent trafficking pathway. Newly synthesized procathepsin D (51-53 kDa) associates with the cation-independent mannose 6-phosphate receptor (M6PR) in the TGN and is delivered to the endosomal compartment, where it is activated by removal of the pro-peptide to generate an intermediate form that migrates at 47-48 kDa on SDS gels. The final step in cathepsin D processing is completed in the lysosomes, where the intermediate form is cleaved to the mature form, which contains two non-covalently linked chains of 31 kDa and 14 kDa (Rijnboutt et al., 1992; Delbruck et al., 1994). As shown by the immunoblots in Fig. 6A, the steady-state levels of the 53 kDa procathepsin D, the 47 kDa intermediate and the 31 kDa mature cathepsin D were similar in the control and Beclin KD cells. In cells depleted of hVps34, this method readily detects a 3-4-fold increase in the steady-state level of the 47 kDa intermediate (E. E. Johnson, J.H.O., W. T. Gunning and W.A.M., unpublished data). In cells treated with ammonium chloride to raise the pH of the endosomal and lysosomal compartments, a substantial increase in 53 kDa procathepsin D can be observed (Fig. 6A).
To obtain a more direct assessment of the kinetics of processing of newly synthesized procathepsin D, we performed a pulse-chase analysis (Fig. 6B). When [35S]methionine-labeled cathepsin D was immunoprecipitated after a 30 minute pulse, nearly all of the radiolabeled protein was in the 53 kDa pro form in both control and KD cells. After a 4 hour chase, the mature 31 kDa cathepsin D was the predominant form detected, with no significant difference in the percentage of total procathepsin D processed to the mature form in the control and Beclin KD cells (Fig. 6C). After 4 hours, there was no residual 53 kDa procathepsin D and only a small amount of the 47 kDa endosomal intermediate. By comparison, cells treated with ammonium chloride generated almost no mature cathepsin D during the same time period (Fig. 6B,C). Immunoprecipitation of the culture medium revealed that, in the latter cultures, most of the radiolabeled procathepsin D (approx. 80%) was secreted during the 4 hour chase. However, prolonged film exposure revealed only small amounts of secreted procathepsin D (<7% of total) in the control and Beclin KD cultures (not shown). These results indicate that Beclin is not required for normal PI 3-kinase-dependent trafficking of procathepsin D from the TGN to the endosomal and lysosomal compartments in U-251 cells.
Endocytic trafficking in the Beclin KD cells
In addition to disrupting trafficking between the TGN and lysosome, wortmannin causes marked swelling of late endosome compartments (Reaves et al., 1996; Fernandez-Borja et al., 1999). This appears to be a result of failure of inward vesiculation of multivesicular endosomes, without a compensatory decrease in endocytic membrane influx (Futter et al., 2001). We have recently observed a similar enlargement of vesicular compartments containing the late endosome/lysosome membrane marker LAMP1 when hVps34 expression was silenced in U-251 cells (E. E. Johnson, J.H.O., W. T. Gunning and W.A.M., unpublished data). In contrast to these findings, immunofluorescence microscopy of the Beclin KD cells revealed no detectable changes in the morphology or distribution of molecular markers for late endosomes/lysosomes (LAMP1), early endosomes (EEA1) or Golgi membranes (GM130) (Fig. 7).
To assess directly the endocytic transport pathway, we measured the cellular uptake of the fluid-phase marker horseradish peroxidase (HRP) (Fig. 8). The results did not reveal any consistent perturbation of HRP endocytosis in cells lacking Beclin. To explore the endocytic pathway further, we followed the fate of activated EGFR. In serum-deprived cells grown in the absence of EGF, degradation of EGFR is minimal and receptors accumulate on the cell surface. However, upon addition of EGF, the receptors are rapidly activated by tyrosine phosphorylation in the C-terminal cytoplasmic domain and the EGF-EGFR complexes are internalized into clathrin-coated early endosomes. Downregulation of activated receptors depends on their delivery to MVEs, where receptor complexes are sorted into internal vesicles that are ultimately degraded when the late endosomes fuse with lysosomes (Katzmann et al., 2002). Thus, disruptions of endocytic trafficking and/or sorting into MVEs can be detected by changes in localization, degradation or phosphorylation of the EGFR.
Previous studies with HEp2 cells microinjected with antibody against Vps34 (Futter et al., 2001), as well as our recent studies with U-251 cells depleted of hVps34 by gene silencing (E. E. Johnson, J.H.O., W. T. Gunning and W.A.M., unpublished data), have indicated that PI 3-kinase is essential for the inward vesiculation of MVEs. In hVps34 KD cells, EGFR is internalized but remains on limiting membranes of enlarged vesicular structures that appear to be derived from late endosomes (E. E. Johnson, J.H.O., W. T. Gunning and W.A.M., unpublished data). In the latter situation, the rate of degradation of the EGFR is reduced, and there is a marked attenuation of receptor dephosphorylation, manifested as a doubling of the ratio of phosphorylated receptor (pEGFR) to total receptor at 30 minutes after EGF stimulation. In contrast to the results with hVps34 KD cells, immunofluorescence localization studies revealed no discernable differences in EGFR distribution before or after EGF stimulation in Beclin KD cells compared with controls (Fig. 9A). Before addition of EGF, most of the receptors were present in the peripheral cell membrane (Fig. 9A). Within 30 minutes after addition of EGF, most of the receptors had moved into internal vesicles arrayed throughout the cytoplasm, with some concentration around the nucleus (Fig. 9A). This pattern is typical of early and late endosomes (Fig. 7). By 70 minutes, most of the EGFR-positive structures were clustered in the juxtanuclear region (Fig. 9A) in a pattern typical of lysosomes (Fig. 7). Of particular note, there was no evidence of localization of EGFR to enlarged vesicular structures at any of these stages. In contrast to our observations with hVps34 KD cells, the time course of receptor degradation after addition of EGF was nearly identical in the Beclin KD cells compared with the controls (Fig. 9B), and there was no significant impairment of EGFR dephosphorylation in the Beclin KD cells (Fig. 9C).
Growth properties of Beclin KD cells
Disruption of endocytic trafficking might be expected to have negative consequences for cell proliferation by impeding internalization of nutrient carriers (e.g. transferrin or LDL receptors) or growth factor receptors (e.g. insulin). Alternatively, alterations at later stages, such as the internalization of receptors into MVEs, might prolong proliferative signals propagated by phosphorylated growth factor receptors. Thus, as a final indirect means to detect such changes in the Beclin KD cells, we compared their rate of cell proliferation to control cells with normal levels of Beclin (Fig. 10A). Consistent with the normal morphology (Fig. 7) and endocytic internalization of HRP or EGFR (Figs 8, 9), the suppression of Beclin expression had no detectable effect on the rate of proliferation of U-251 cells (Fig. 10A). Moreover, the Beclin KD cells were indistinguishable from controls with respect to their ability to form colonies in soft agar (Fig. 10B).
Beclin is an interacting partner for the mammalian class III PI 3-kinase mVps34. Previous studies have established that this PI 3-kinase is required for macroautophagy in nutrient-starved cells (Petiot et al., 2000; Eskelinen et al., 2002), for normal lysosomal enzyme sorting and protein trafficking in the endocytic pathway (Row et al., 2001; Futter et al., 2001; Petiot et al., 2003), and for cell cycling (Siddhanta et al., 1998). In contrast with its documented role in macroautophagy, the possible role of Beclin as an essential chaperone or adaptor for hVps34 in normal trafficking pathways has received little attention. Furuya et al. (Furuya et al., 2005) recently reported that overexpression of a Beclin mutant lacking the conserved Vps34-binding domain does not alter cathepsin D maturation in MCF7 cells, compared with cells expressing high or low levels of wild-type Beclin. In the present studies, we used a different approach (siRNA-mediated gene silencing) to show that Beclin is required specifically for the function of hVps34 in the macroautophagy pathway, but not for the normal functions of this PI 3-kinase in trafficking from the TGN to the lysosome in U-251 glioblastoma cells. Moreover, we extend this line of investigation by showing that Beclin is also not required for the proposed functions of hVps34 in endocytic trafficking and inward vesiculation of multivesicular endosomes.
The current findings contrast with previous studies of the Beclin homolog Vps30 (Atg6) in S. cerevisiae. In those studies, Vps30 interaction with Vps34 was found to be required both for starvation-induced autophagy (Seaman et al., 1997; Kametaka et al., 1998) and vesicular transport of carboxypeptidase Y (CPY) from the Golgi compartment to the vacuole (Klionsky et al., 1990). The dual role of Vps30 in yeast can be attributed to its assembly into two distinct protein complexes (Kihara et al., 2001b). In the autophagy pathway, Vps30 is linked to Vps34 through an interaction with a novel bridging protein, Atg14, whereas, in the Golgi-to-vacuole pathway, a different linker protein, Vps38, mediates this interaction. Several lines of evidence suggest that the mammalian Beclin-Vps34 complex might be different from these yeast Vps30-Vps34 complexes. First, in vps30-defective yeast, Beclin can complement Vps30 in the autophagy pathway, but not in the vacuolar protein-sorting pathway (Liang et al., 1999). This would be consistent with the evidence indicating that Beclin does not function in post-Golgi sorting of cathepsin D in mammalian cells. Second, homologs of the bridging proteins Atg14 and Vps38 have not yet been identified in mammalian cells. Finally, whereas Vps30 (Atg6) and Beclin share substantial amino acid sequence similarity in their central coiled-coil and evolutionarily conserved Vps34-binding domains (Liang et al., 1998; Furuya et al., 2005), Vps30 (Atg6) is significantly larger (557 amino acids) than human Beclin (450 amino acids), with several regions of sequence divergence. For example, Vps30 (Atg6) contains 22 unique N-terminal amino acids and 17 unique C-terminal amino acids, compared with human Beclin. In addition, several internal regions of Vps30 (Atg6) (e.g. residues 38-48, 147-157, 403-416, 499-514) generate notable gaps when aligned against human Beclin. Although the specific functions of these domains remain to be defined, one could speculate that structural features conferred by the unique sequences in Vps30 (Atg6) might account for its broader role in promoting Vps34 interaction with both normal trafficking and autophagy pathways, compared with the more-restricted role of Beclin in macroautophagy.
PtdIns(3)P is distributed throughout cellular endomembranes, where it serves to recruit a variety of proteins implicated in the regulation of vesicular transport and intracellular protein sorting (Simonsen et al., 2001; Corvera, 2001). Some of these proteins contain a Phox-homology phosphoinositide-binding domain (Song et al., 2001; Cheever et al., 2001; Xu et al., 2001; Kanai et al., 2001), whereas others contain a structural motif termed the FYVE finger, which binds to PtdIns(3)P with high affinity (Wurmser et al., 1999; Fruman et al., 1999; Corvera et al., 1999). Little is known about the molecular events that direct the targeting of the hVps34 PI 3-kinase to specific subcellular compartments. Like its yeast counterpart, hVps34 appears to associate with intracellular membranes through interaction with a myristoylated adaptor, p150 (hVps15) (Volinia et al., 1995; Panaretou et al., 1997). There is some evidence that recruitment of mVps34 to endosomal membranes is facilitated by specific Rab GTPases (e.g. Rab5, Rab7) that can bind to p150 (Murray et al., 2002; Stein et al., 2003). Once it is associated with the membrane, hVps34 presumably generates the PtdIns(3)P required for membrane association of FYVE-domain proteins such as the Rab effectors EEA1 and Rabenosyn-5 (Simonsen et al., 1998; De Renzis et al., 2002), and the PtdIns(3)P 5-kinase PIKfyve (Ikonomov et al., 2003). Our finding that Beclin can be substantially depleted from mammalian cells without any detectable consequences for endosome morphology, EEA1 distribution, or endocytic trafficking strongly suggests that Beclin is not required for targeting or recruitment of hVps34 to endosomal membranes. However, because Beclin expression was not entirely eliminated in the KD cells, we cannot definitively rule out the possibility that very low residual levels of Beclin are sufficient to mediate these processes. By contrast, our studies clearly implicate Beclin in one or more key steps required for generating a central component of the pre-autophagosome protein complex.
Several reviews have summarized the recent advances in understanding the pre-autophagosomal protein assemblies involved in initiating the formation of the isolation membrane (Klionsky, 2005; Kirkegaard et al., 2004; Marino and Lopez-Otin, 2004). The conjugation of LC3 (Atg8) to PE, generating LC3-II, occupies a central position in this scheme. LC3 is first cleaved by a cysteine protease (Atg4/autophagin) to expose a C-terminal glycine. The protein is then conjugated sequentially, first to Atg7, then to Atg3, before the final conjugation to PE on the pre-autophagosome membrane (Tanida et al., 2004). An oligomeric protein complex containing Atg12-Atg5 and Atg16 plays a key role in recruiting LC-II to the pre-autophagosome membrane (Mizushima et al., 2001; Yoshimori, 2004). Therefore, our observation that Beclin is required for the formation of LC3-II implies that Beclin functions in the earliest steps required for autophagosome biogenesis, rather than in the later maturation events involving fusion with endosomes or lysosomes. Since inhibitors of PI 3-kinase have been reported to cause a reduction in LC3-II production (Aki et al., 2003), we believe that the attenuated production of LC3-II observed in starved or ceramide-treated Beclin KD cells is related to an impaired ability of hVps34 to function in the autophagy pathway without Beclin. The molecular basis for the apparent requirement for PtdIns(3)P in the autophagic process in mammalian cells remains to be determined. However, recent studies have described an autophagy-linked FYVE-domain protein, Alfy (Simonsen et al., 2004), and a Rab GTPase, Rab24 (Munafo and Colombo, 2002), associated with autophagosomes. These findings raise the intriguing possibility that PtdIns(3)P-dependent molecular assemblies not unlike those described on early endosomes (De Renzis et al., 2002) might also play a role in the formation and/or maturation of autophagosomes.
As mentioned in the introduction, ectopic overexpression of Beclin in breast cancer cells with monoallelic deletions of the beclin gene can inhibit cell proliferation, colony formation in soft agar and tumorigenesis in nude or scid mice (Liang et al., 1999; Furuya et al., 2005). These observations, coupled with the increased frequency of spontaneous tumors detected in beclin–/+ mice (Qu et al., 2003; Yue et al., 2003), indicate that Beclin can function as a tumor suppressor. The exact mechanisms underlying the effects of Beclin on tumorigenesis remain to be defined, but the apparent requirement for Beclin for the accumulation of autophagosomes in the initial stages of type II programmed cell death (Scarlatti et al., 2004; Yu et al., 2004; Shimizu et al., 2004) suggests that its role as a tumor suppressor is most probably related to regulation of macroautophagy. Nevertheless, without knowing if Beclin interaction with hVps34 is also required for the function of this PI 3-kinase in normal vesicular trafficking pathways, some question has remained as to whether or not the loss of Beclin could contribute to tumor formation by alternative mechanisms, such as the disruption of hVps34-dependent endocytic pathways and consequential potentiation of growth factor signaling. The results of the present studies argue against this possibility by showing that normal levels of Beclin are not required for the function of hVps34 in endocytosis, the biogenesis of multivesicular endosomes, or EGFR degradation.
Although our stable Beclin KD glioblastoma cells were unable to mount a typical autophagic response when challenged by nutrient deprivation or ceramide treatment, the loss of Beclin had no detectable effect on their basal growth rate or ability to form colonies in soft agar. Unlike MCF7 breast cancer cells, U-251 cells harbor deletions in the PTEN gene, which encodes a protein/lipid phosphatase that dephosphorylates phosphatidylinositol (3,4,5)-trisphosphate [PtdIns(3,4,5)P3], which is the product of class I p110/p85 PI 3-kinases (Vazquez and Sellers, 2000). Consequently, these cells exhibit high levels of constitutive Akt activation. There is some evidence that increasing the intracellular concentration of PtdIns(3,4,5)P3 by stimulating the activity of class I PI 3-kinase inhibits macroautophagy, which is the opposite effect of stimulating the mVps34 class III PI 3-kinase (Petiot et al., 2000). Along the same line, inhibiting Akt/protein kinase B, a major downstream target in the class I PI 3-kinase signaling pathway, synergistically enhances macroautophagy in glioblastoma cells treated with rapamycin (Takeuchi et al., 2005). Therefore, one possible explanation for the absence of an effect of Beclin depletion on glioblastoma cell proliferation is that high basal levels of PtdIns(3,4,5)P3 and constitutive activation of Akt suppress hVps34-dependent macroautophagy and supercede any regulatory effects of Beclin, unless the cells are exposed to a strong pro-autophagic stimulus (e.g. starvation, rapamycin, C2-ceramide). Definition of the relationship between Beclin-Vps34-mediated macroautophagy and the activity of the p110/p85 PI 3-kinase signaling pathways in malignancies with different genetic backgrounds promises to be an important area for further investigation.
In conclusion, we propose that a primary function of Beclin is to facilitate the interaction of hVps34 PI 3-kinase with specific effectors on the pre-autophagosomal isolation membrane in response to pro-autophagic conditions or death signals. However, a major unsolved puzzle concerning Beclin is the biological significance of its reported interaction with the anti-apoptotic proteins Bcl-2 and Bcl-XL, which have no known connection to mVps34-mediated protein trafficking or the pre-autophagosomal Atg protein complexes. Interest in this question has intensified with the recent observation that etoposide treatment induces autophagic cell death instead of classical apoptosis in wild-type or Bax/Bak–/– double-knockout mouse embryonic fibroblasts overexpressing Bcl-2 or Bcl-XL (Shimizu et al., 2004). The interaction between Beclin and Bcl-2 was first observed in yeast two-hybrid assays and then confirmed by FRET analysis of the two proteins overexpressed in COS cells (Aita et al., 1999). In a more recent study, Pattingre et al. (Pattingre et al., 2005) showed that overexpression of wild-type Bcl-2, but not Beclin-binding-defective Bcl-2 mutants, can suppress Beclin-dependent autophagy in yeast and mammalian cells. In accord with published observations (Pattingre et al., 2005), we have found it easy to observe an interaction between FLAG-Beclin and Bcl-2 in pull-down assays using transfected 293 cells overexpressing both proteins. However, as shown in Fig. 2, we have been unable to detect an interaction between endogenous Beclin and Bcl-2 or Bcl-XL under conditions that allow co-immunoprecipitation of Beclin with hVps34. Our findings are consistent with those of Kihara et al. (Kihara et al., 2001a), who demonstrated that all of the endogenous Beclin in Hela cells could be co-immunoprecipitated together with hVps34, using an antibody against the latter protein. Nevertheless, these studies do not completely rule out the possibility that low-affinity interactions might occur between membrane-bound Bcl-2 or Bcl-XL and the Beclin-mVps34 complex. Indeed, there is some evidence that overexpression of Bcl-2 reduces the amount of Beclin in the hVps34 complex (Pattingre et al., 2005). Transient interactions might be lost in the detergent-containing buffers typically used for immunoprecipitation. Alternatively, the interactions between Beclin and Bcl-2 might be more prominent in specific cell types or under certain metabolic conditions. This will be an important topic for future study.
Materials and Methods
siRNA-mediated silencing of Beclin
U-251 human glioblastoma cells were obtained from the National Cancer Institute Frederick Cancer DCT Tumor Repository and were maintained in Dulbecco's modified Eagle medium (DMEM), supplemented with 10% fetal bovine serum (FBS). The pSUPER.retro.puro vector was obtained from OligoEngine. The oligonucleotide sequence used for siRNA interference with Beclin expression corresponded to nucleotides 1201-1219 (5′-GGCAAGAUUGAAGACACAG-3′) downstream of the transcription start site of beclin (GenBank accession number: AF077301), followed by a 9-nucleotide non-complementary spacer (TATCTTGAC) and the reverse complement of the initial 19-nucleotide sequence. A control vector was constructed with a similar insert where the 19-nucleotide sequence had no homology to any known human gene sequence. Retrovirus was produced in 293-GPG packaging cells (Ory et al., 1996) maintained in DMEM + 10% heat-inactivated FBS with 1 μg/ml puromycin, 300 μg/ml G418 and 2 μg/ml doxycycline. For transfection, the 293-GPG cells were seeded at 1.2×107 cells per dish in 100 mm dishes in DMEM containing 10% heat-inactivated FBS. After 24 hours. the cells were transfected with the pSuper.retro.puro constructs using Lipofectamine-Plus reagent (Invitrogen). At 48 and 72 hours after transfection, the virus-enriched medium was collected and passed through a 0.22 μm filter. Infections of the U-251 cells were performed on two sequential days in the presence of 4.0 μg/ml hexadimethrine bromide (Sigma). At 24 hours after the second infection, the cells were trypsinized and re-plated in selection medium containing 1 μg/ml puromycin. After a selection period of six days, the surviving cells were pooled and used for studies described in the following sections.
Immunoprecipitation of endogenous Beclin protein complexes
U-251 cells were grown to 80% confluence in 100 mm dishes in DMEM with 10% FBS. The cells were washed three times with Hanks balanced salt solution (HBSS), scraped from the dish and homogenized in IP buffer: 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, and protease inhibitors. The lysate was centrifuged at 10,000 g for 15 minutes at 4°C and the supernatant solution was incubated with goat polyclonal IgG against Beclin (2 hours at 4°C), followed by 1 hour incubation with protein A Sepharose beads. The beads were washed three times with IP buffer, twice with phosphate-buffered saline (PBS), and the immune complexes were eluted from the beads and subjected to SDS-PAGE and immunoblot analysis as described previously (Wilson et al., 1996). Primary antibodies used for immunoblot analysis included mouse monoclonal antibody against Beclin (BD Biosciences), rabbit polyclonal antibody against hVps34 (Zymed Laboratories), mouse monoclonal antibody against Bcl-2, and rabbit polyclonal antibody against Bcl-XL (Santa Cruz Biotechnology). For the converse immunoprecipitation, Bcl-2 was pulled-down with a rabbit polyclonal antibody (Santa Cruz Biotechnology) and the immunoprecipitates were immunoblotted with the monoclonal antibodies against Bcl-2 or Beclin. The antibody against hVps34 did not work for immunoprecipitation.
Organelle morphology was assessed in control and Beclin KD cells grown on laminin-coated glass coverslips for 24 hours. For detection of EEA1, cells were fixed with 3% paraformaldehyde and permeabilized with 0.05% saponin in PBS. For LAMP1 or GM130, cells were fixed with ice-cold methanol for 10 minutes. All cells were blocked with 10% goat serum in PBS for 30 minutes and the following monoclonal antibodies were applied for 1 hour in PBS with 10% goat serum: anti-LAMP1 antibody (University of Iowa Developmental Studies Hybridoma Bank, Iowa City, IA), anti-GM130 and anti-EEA1 (BD Biosciences). Cells were then washed three times with 10% goat serum in PBS and incubated for 1 hour with Alexa Fluor 568 goat anti-mouse IgG (Molecular Probes). Photomicrographs were taken with a Nikon Eclipse 800 fluorescence microscope equipped with a digital camera. Images were acquired and processed using ImagePro software (Media Cybernetics).
Induction of macroautophagy
Macroautophagy was induced by nutrient starvation or exposure to C2-ceramide (N-acetyl D-erythro-sphingosine; Calbiochem). For starvation, cells were washed with HBSS three times and then incubated in HBSS for 4 hours. For C2-ceramide treatment, cells were incubated with 10 μM or 20 μM C2-ceramide in DMEM + 0.1% FBS for 24 hours. Ceramide was dissolved in dimethlysulfoxide (DMSO), and control cultures contained equal amounts of vehicle. To monitor the induction of macroautophagy, the relative amounts of endogenous LC3 in the unmodified form (LC3-I) and the phosphatidylethanolamine (PE)-conjugated form (LC3-II) were determined by immunoblot analysis of whole-cell lysate, using a rabbit polyclonal antibody against LC3 (Kabeya et al., 2000) kindly provided by Tamotsu Yoshimori (National Institute for Basic Biology, Okazaki, Japan). The chemiluminescent signals from the immunoblot were quantified with a Kodak 440CF image station. For cells starved in HBSS, macroautophagy was also monitored by sequestration of the cytosolic enzyme lactate dehydrogenase (LDH) into membrane-bound components prepared by centrifugation of cell lysate over a metrizamide/sucrose cushion as described (Kopitz et al., 1990; Stromhaug et al., 1998). Aliquots of the whole-cell lysate and the pelleted material were subjected to SDS-PAGE and immunoblot analysis with a mouse monoclonal antibody against LDH (Sigma).
Detection and quantification of acidic vesicular organelles with acridine orange (AO)
Vital staining of cells with AO (Molecular Probes) was performed essentially as described (Paglin et al., 2001). Cells were grown on laminin-coated coverslips (for fluorescence microscopy) or in 96-well plates (for quantification of red fluorescence) and treated with C2-ceramide or vehicle (DMSO) for the indicated time. AO was added for 15 minutes at a final concentration of 1 μg/ml, and the cells were then washed three times with PBS. Unfixed cells were examined immediately by fluorescence microscopy using a Nikon Eclipse 800 microscope with the red filter set (G-2E/C; excitation 528-553, emission 600-660). Red fluorescence was quantified with a microplate fluorimeter (Molecular Devices) with excitation and emission wavelengths set at 488 nm and 655 nm, respectively. To normalize the measurements to the number of cells present in each well, a solution of ethidium bromide (EB) was added to a final concentration of 0.2 μM and the fluorescence emitted from the DNA complexes was measured at 530 nm (excitation), 590 nm (emission). The AO red fluorescence was expressed as a ratio to the EB fluorescence.
Cathepsin D processing
Steady-state levels of intracellular cathepsin D were measured in whole-cell lysates by SDS-PAGE and immunoblot analysis, using goat anti-cathepsin D antibody from Santa Cruz Biotechnology. To measure the kinetics of cathepsin D processing, U-251 cells were pulse-labeled for 30 minutes in methionine-free DMEM containing 10% FBS and 100 μCi/ml [35S]methionine (Easy Tag express labeling mix; 1175 Ci/mmol, Perkin Elmer), then chased for 4 hours in DMEM containing 10% FBS, 200 μM methionine and 200 μM cysteine. Cells were washed three times with PBS, harvested using a cell scraper, homogenized and solublized in 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% SDS, 5 mM EDTA. Insoluble material was removed by centrifugation at 100,000 g for 45 minutes at 4°C and the lysates were precleared with protein A Sepharose. Samples were then incubated for 2 hours with a polyclonal antibody against cathepsin D (Biodesign International). Immune complexes were then collected on protein A Sepharose and subjected to SDS-PAGE, fluorography and Phosphorimager analysis as described previously (Wilson et al., 1996).
Endocytosis of HRP
Cells grown to approximately 80% confluence were washed with DMEM and then incubated at 37°C with HRP (2 mg/ml) in DMEM containing 1% bovine serum albumin (BSA) for the time periods indicated in the figure. Cells were placed on ice, washed three times with ice-cold PBS containing 1% BSA and one time with PBS. Cells were then scraped into PBS and collected by centrifugation at 390 g for 4 minutes at 4°C. Cell pellets were washed once with PBS and lysed in PBS containing 0.5% Triton X-100. Lysates were cleared by centrifugation at 10,000 g for 10 minutes at 4°C, and equal aliquots were removed for peroxidase assay, using the One-Step Turbo TMB enzyme-linked immunosorbent assay kit (Pierce Chemical). After addition of sulfuric acid stop solution, absorbance at 450 nm was measured and the enzyme activity was normalized to total protein, determined using a colorimetric assay (Bio-Rad).
EGFR internalization and degradation
Parallel cultures of control or Beclin KD cells were seeded at 200,000 cells per dish on laminin-coated cover slips in 60 mm dishes and grown for 48 hours. The cells were then washed with PBS and maintained in serum-free DMEM overnight to allow the EGFR to accumulate on the cell surface. EGFR internalization was stimulated by incubating the cells with 200 ng/ml EGF (Upstate Biotechnology) in HBSS containing 20 mM HEPES and 0.2% BSA for 30 minutes or 70 minutes at room temperature. Immunofluorescence localization of the EGFR was performed using anti-EGFR monoclonal antibody (Upstate Biotechnology), as described earlier for other organelle markers. In a separate study, the cells were harvested in SDS sample buffer at 30 minutes or 70 minutes after stimulation with EGF, and aliquots containing equal amounts of total cell protein were subjected to SDS-PAGE and immunoblot analysis using antibodies that recognize total EGFR or phosphorylated EGFR (Tyr1068; Cell Signaling Technology). Total EGFR or ratios of phospho-EGFR to total EGFR were determined by scanning the blots with a Kodak 440CF Image Station.
For growth curves, cells were seeded in 60 mm dishes at 200,000 cells per dish. At daily intervals, cells were harvested from three parallel dishes and counted with a Coulter Z1 particle counter. For measurement of colony formation in soft agar, 1500 cells were suspended in DMEM containing 10% FBS, 1 μg/ml puromycin and 0.3% w/v SeaPlaque agar (FMC Bioproducts) and layered over the same medium with 0.6% agar in three parallel dishes. On the 14th day after plating, colonies were stained by incubation with 5 mg/ml 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide at 37°C for 25 minutes. Colonies were counted manually after obtaining digital images of the plates with an EPSON expression 800 scanner. Visual inspection of the colonies formed in plates containing control or Beclin KD cells did not reveal any obvious differences in their size or morphology.
We thank Jane Ding for technical assistance and Tamotsu Yoshimori for providing the antibody against LC3. This work was supported by a US Department of Defense Breast Cancer Research Program Grant, W81XWH-04-1-0493 to W.A.M.
- Accepted October 7, 2005.
- © The Company of Biologists Limited 2006