|Home | About | Journals | Submit | Contact Us | Français|
Phosphoinositide 3-kinase (PI3K) family members are involved in diverse cellular fates including cell growth, proliferation, and survival. While many molecular details are known about the Class I and III PI3Ks, less is known about the Class II PI3Ks. To explore the function of all eight PI3K isoforms in autophagy, we knock down each gene individually and measure autophagy. We find a significant decrease in autophagy following siRNA-mediated PIK3C2A (encoding the Class 2 PI3K, PI3K-C2α) knockdown. This defective autophagy is rescued by exogenous PI3K-C2α, but not kinase-dead PI3K-C2α. Using confocal microscopy, we probe for markers of endocytosis and autophagy, revealing that PI3K-C2α colocalizes with markers of endocytosis. Though endocytic uptake is intact, as demonstrated by transferrin labeling, PIK3C2A knockdown results in vesicle accumulation at the recycling endosome. We isolate distinct membrane sources and observe that PI3K-C2α interacts with markers of endocytosis and autophagy, notably ATG9. Knockdown of either PIK3C2A or ATG9A/B, but not PI3KC3, results in an accumulation of transferrin-positive clathrin coated vesicles and RAB11-positive vesicles at the recycling endosome. Taken together, these results support a role for PI3K-C2α in the proper maturation of endosomes, and suggest that PI3K-C2α may be a critical node connecting the endocytic and autophagic pathways.
Macroautophagy (autophagy) is an intracellular degradation pathway that targets cytosolic material for lysosomal degradation [1–3]. Under conditions of stress, such as nutrient starvation, this process is used to produce amino acids and other biochemical building blocks to promote cell survival [4–7]. The importance of autophagy is underscored by its deregulation in a number of diseases, notably cancer [4,8,9]. While considerable progress has been made characterizing the mechanism of this process, many important issues, such as those concerning the incorporation of different membrane sources into this pathway, remain unaddressed.
In response to nutrient stress, key cellular signals, such as the activation of AMP activated protein kinase (AMPK) and suppression of mechanistic target of rapamycin complex 1 (mTORC1), result in the activation of the Unc-51 Like autophagy activating kinase 1/2 (ULK1/2) regulatory complex [10–14]. ULK1/2 is responsible for the initiation of autophagy through the phosphorylation of several necessary autophagic components, including Autophagy Related Protein 9 (ATG9) and the BECLIN1-PI3KC3 complex [15–17]. Phosphorylation of ATG9 promotes binding to the AP1/2 clathrin adaptor complex and, along with BECLIN1-PI3KC3 complex activation, is required for membrane nucleation and initial formation of pre-autophagosomal vesicles [16–20]. Elongation of the phagophore, the initial cup-shaped autophagy membrane, and the recruitment of organelles and proteins targeted for degradation follows, and this is associated with both conjugation of ATG12 to ATG5 and microtubule-associated protein 1 light chain 3 (MAP1LC3, hereafter LC3) to the lipid phosphatidylethanolamine [21–23]. The double-membrane structure containing the cargo is closed, forming the completed autophagosome, which then fuses with endocytic or lysosomal vesicles, leading to the degradation of the components [24–26].
In the 1990s, a phosphatidylinositol 3-kinase (PI3K), Vps34p, was first identified as an essential protein for autophagy in yeast , as it generates phosphatidylinositol 3-phosphate (PI(3)P), a lipid critical for the nucleation of autophagic vesicles [28,29]. While this initial work was performed in yeast, which only contain one PI3K, later research revealed eight human PI3K isoforms that differ in structure and substrate sensitivity [30,31]. Class III PI3K (phosphatidylinositol 3-kinase catalytic subunit type 3; PI3K-C3; VPS34), also a critical autophagy regulator in mammalian cells , shares a similar role to Vps34p in yeast, operating in an autophagy-specific complex with mammalian homologs VPS15, BECLIN1, and ATG14L [29,33,34].
To explore the function of all eight PI3K isoforms on autophagy, we performed a focused siRNA experiment. Along with PI3K-C3, as expected, we identified the Class II PI3K, phosphatidylinositol-4-phosphate 3-kinase catalytic subunit type 2 alpha (PI3K-C2α) as being critical for autophagy. Current research on PI3K-C2α has focused the enzyme’s role in vesicle trafficking  with PI3K-C2α having important roles for both clathrin-dependent and clathrin-independent internalization of vesicles, suggesting a dynamic role for the enzyme in the regulation of endocytosis [36–42]. While it has been suggested that PI3K-C2α may play a role in autophagy [34,43], the fundamental connections with PI3K-C2α to autophagy remains unclear.
It was originally believed that the Golgi apparatus and endoplasmic reticulum (ER) were the sole membrane sources for autophagic vesicles [44,45]. Indeed, these organelles contribute to phagophore formation [46–48], and it was later shown that other membranes, such as the mitochondria and plasma membrane, can also integrate directly into pre-autophagosomal structures [49–51]. Follow-up work revealed that ATG16L1- and ATG9-positive membrane sources from clathrin-mediated endocytosis (CME) coalesce in the recycling endosome before integrating into pre-autophagosomal vesicles [50,52–54]. This trafficking of ATG9-positive vesicles through the recycling endosome is required for autophagosome formation , and the formation of these pre-autophagosomal vesicles appears to be dependent on the formation of ATG9-complexes with the clathrin-adaptor proteins AP1/2 [17,19].
In this study, we show that PI3K-C2α acts as a positive regulator of autophagy. Analysis of the subcellular localization of PI3K-C2α reveals that the protein localizes with markers of endocytosis and pre-autophagosomal vesicles, including ATG9 and AP2. CME stimulation in conjunction with PI3K-C2α knockdown results in an enrichment of perinuclear vesicles that are positive for both clathrin and RAB11; these same markers accumulate upon the knockdown of ATG9. Based on these results, we conclude that PI3K-C2α is critical for coordinating the use of endosomes as an additional membrane source during autophagy, likely through interactions with ATG9.
U2OS cells stably expressing LC3B fused to a tandem (EGFP and mRFP) fluorescent tag (ptfLC3-U2OS) [56–58] were transfected with control siRNA and treated with rapamycin, an mTORC1 inhibitor and autophagy inducer, for 6 hours. Rapamycin-treated cells showed increased EGFP-LC3B puncta when compared to vehicle control cells, consistent with increased autophagic vesicle formation and induction of autophagy (Fig 1A). To examine the isoform-specific role of the PI3Ks in autophagy, we knocked down each of the eight PI3K isoforms, treated cells with rapamycin as described above, and measured the formation of EGFP-LC3B autophagic puncta. Knockdown of each of the four Class I PI3Ks (PIK3CA, PIK3CB, PIK3CG, PIK3CD) and two of three Class II PI3Ks (PIK3C2B and PIK3C2G) (Fig 1B and 1C and S1 Fig) resulted in EGFP-LC3B puncta levels similar to control cells, suggesting that these PI3Ks do not dramatically regulate autophagy. Knockdown of PIK3C3 (encoding the Class 3 PI3K, PI3K-C3, or VPS34), known to strongly impact autophagy signaling , or PIK3C2A (encoding the Class 2 PI3K, PI3K-C2α), decreased EGFP-LC3B puncta formation, suggesting that PI3K-C2α also plays an important role in autophagy (Fig 1D and S1 Fig). Specifically, control siRNA cells treated with rapamycin averaged 54 LC3-positive puncta per cell, while PIK3C2A and PIK3C3 siRNA knockdown resulted in a 65% or 48% reduction, respectively (Fig 1E).
Previously, we developed a U2OS cellular system and image processing protocol to monitor both autophagosome synthesis and turnover in single cells using fluorescent images . To characterize basal autophagy in U2OS cells, we imaged EGFP-LC3-positive puncta in single cells cultured in full-nutrient media with or without Bafilomycin A1 (BafA1), a V-ATPase inhibitor that prevents autophagosome turnover . Following a short pretreatment period with either vehicle (–) or BafA1 (+), cells were imaged once every 1.5 min for 70 min. Representative images are shown in S2 Fig. As expected, vesicle counts increased for vehicle treated (Fig 2A) or rapamycin treated cells (Fig 2B), with the increase significantly higher in cells treated with rapamycin and BafA1. Next, we repeated these measurements of basal (Fig 2C) and induced autophagy (Fig 2D) after PIK3C2A knockdown. The decrease in both the absolute number of puncta and the rate of autophagosome formation per cell suggests that PI3K-C2α positively regulates autophagy.
To better understand the kinetics of autophagy and the role of PI3K-C2α and PI3K-C3 in autophagy, ptfLC3-U2OS cells were transfected with non-targeting (negative control), ULK1 (positive control), PIK3C2A, or PIK3C3 siRNA, and images acquired over a 6 hour time-period following the addition of rapamycin. PIK3C2A or PIK3C3 knockdown resulted in a time-dependent decrease in the number of EGFP-LC3B positive puncta compared to the control cells (Fig 3A). At 1 hour, we observed that the level of autophagy with PIK3C2A and PIK3C3 knockdown diverges from the control, and by 3 hours, the average puncta per cell for PIK3C2A and PIK3C3 knockdown was reduced 48% and 39%, respectively. This divergence continued with sustained rapamycin treatment (6 hours), where we noticed a pronounced reduction in GFP-LC3B puncta per cell: 57% and 69% for PIK3C2A and PIK3C3 knockdown, respectively. This indicated that both PI3K-C2α and PI3K-C3 are required for the sustained induction of autophagic vesicles. In comparison, knockdown of ULK1, a critical initiator of the autophagy process [15,61,62], resulted in an acute and sustained decrease in puncta formation per cell over the full time-course. Specifically, ULK1 knockdown yielded a 73% decrease in EGFP-LC3 puncta with acute rapamycin treatment (1 hour), and 76% decrease with sustained treatment (6 hours). This is consistent with previous reports that ULK1 is critical for acute induction of autophagic vesicles, as well as sustained induction .
To further explore the impact of PI3K-C2α loss on sustained autophagy, we examined endogenous protein levels by Western blotting for several autophagy markers following 6 hours rapamycin treatment (Fig 3B). Additionally, to measure autophagic turnover, cells were supplemented with or without BafA1 for the final 90 minutes of treatment. U2OS cells were transfected with siRNAs as described above. PIK3C2A knockdown decreased vesicle-lipidated LC3A (LC3A-II) under sustained rapamycin treatment with BafA1, consistent with a deficit in autophagy. We observed similar results with LC3B-II levels, although the differences were less pronounced. Interestingly, there was little change in the protein levels of GABARAP-II, an additional ATG8 isoform. In addition, we detected an accumulation of the autophagic cargo protein, p62/SQSTM1, following PIK3C2A knockdown which is also consistent with impaired autophagy. PIK3C3 knockdown showed a modest decrease in LC3B-II levels and an accumulation of p62, demonstrating a partial defect in autophagy. For comparison, ULK1 knockdown resulted in a distinct decrease in LC3A-II, LC3B-II, and GABARAP-II, as well as an accumulation of p62 under both treatment conditions, indicating a strong deficit in autophagy.
To validate that these knockdown studies, we performed siRNA rescue experiments to determine whether the low level of autophagy induction could be rescued with expression of exogenous wild-type (WT) or kinase-dead  PI3K-C2α. U2OS cells stably expressing EGFP-LC3B were transfected for 24 hours with either control siRNAs or siRNAs directed to PIK3C2A, followed by introduction of siRNA-resistant WT-PI3K-C2α or Kinase-Dead PI3K-C2α (KD-V5-PI3K-C2α) constructs  for an additional 24 hours. Indeed, cells expressing WT-PI3K-C2α in addition to PIK3C2A knockdown contained a similar number of puncta as control cells, compared to the knockdown alone (Fig 3C and S3 Fig). In contrast, cells expressing the KD-PI3K-C2α were unable to rescue the autophagy defect (Fig 3C), suggesting that the kinase activity of PI3K-C2α protein is required for its role in autophagy.
In addition to degrading proteins and bulk cytosol, autophagy supports lipid hydrolysis by releasing the content of lipid droplets to the lysosome for degradation. Moreover, autophagy inhibition is known to increase lipid storage in lipid droplets . A striking result from PIK3C2A knockdown was the presence of abundant lipid droplets as observed by transmission electron microscopy (Fig 3D). Loss of PI3K-C2α resulted in both an increase in the number and size of lipid droplets. This is reminiscent of ATG5 knockdown in cultured hepatocytes  and again highlights a role for PI3K-C2α in autophagy regulation.
PI3K-C2α has an important role in endocytosis [36,38,41,65] regulated in part through interactions with its clathrin binding domain. Our next goal was to examine the impact of PI3K-C2α knockdown on CME under sustained autophagic conditions, as endocytosis can act to integrate membrane sources into the autophagy pathway [20,50,52,55,67,68]. U2OS cells were transfected with either control or PIK3C2A siRNAs for 48 hours and after 6 additional hours of rapamycin treatment, cells were treated with a transferrin Texas Red conjugate. Transferrin is an 80 kDa glycoprotein that binds to receptors on the cell surface which are then internalized through CME . Once internalized, transferrin localizes to either peripheral or perinuclear endosomes. Peripheral transferrin can be associated with early endosomes that are EEA1 positive, while perinuclear transferrin is associated with recycling endosomes and is RAB11 positive . In control cells, we observed an immediate coating of the cell with fluorescent transferrin-receptor complexes (S4A Fig). Transferrin conjugates were rapidly internalized in less than 5 minutes, and resulted in perinuclear vesicle fluorescence after 10 minutes. After 45 minutes, transferrin vesicles remained distributed throughout the cell, including the periphery, indicating continued trafficking of transferrin fluorescent dye-containing vesicles. In cells transfected with PIK3C2A siRNA, transferrin is coated on the cells and internalized in 5 minutes (S4B Fig), much like control siRNA cells. After 10 minutes, transferrin vesicles localized to the perinuclear region and throughout the cell, and by 45 minutes, the transferrin containing vesicles transitioned from a more homogenous cytosolic distribution to accumulation in the perinuclear region. This suggests that these vesicles are associated with the recycling endosome, but are not returning to the plasma membrane as observed under normal conditions (S4A Fig) . To determine if this vesicle localization phenotype was unique to CME, cells were treated with CellMask Orange, a plasma membrane dye internalized by both clathrin-dependent and clathrin-independent pathways. Following 45 minutes of active endocytosis, we observed peripheral vesicle staining in control cells, while cells with PIK3C2A knockdown resulted in mostly perinuclear vesicle staining (S5 Fig); this suggests that the localization of vesicles is maintained in the presence of non-clathrin mediated vesicle internalization.
To better understand the roles of PI3K-C2α in endocytosis and autophagy, we next explored whether PI3K-C2α colocalizes with endogenous markers. We found that mCherry-tagged PI3K-C2α is located on both perinuclear and peripheral vesicles (Fig 4). Next, we probed for endogenous markers in the endocytic pathway from early endocytosis clathrin coated vesicles (CCVs) to markers of vesicle maturation and turnover (Lysosomal Associated Membrane Protein 2, LAMP2). In cells containing both mCherry-PI3K-C2α (red vesicles) and endogenous marker (green vesicles) the amount of overlap (yellow) between vesicles can be calculated using the Pierson’s correlation coefficient (PCC) . PI3K-C2α colocalized with endogenous clathrin (Fig 4A, PCC > 0.6), indicating PI3K-C2α at CCVs. Additionally, PI3K-C2α partially colocalized with endogenous EEA1 or RAB5 (Fig 4B and 4C, 0.6 > PCC > 0.3), suggesting a continued presence of PI3K-C2α at the early endosome. In contrast, PI3K-C2α showed a lower level of colocalization with endogenous RAB11, LC3B, or LAMP2 (Fig 4D–4F, PCC < 0.3), thus demonstrating a reduced presence at the recycling endosome, autophagosomes, or autolysosomes. Overall, these data suggest that PI3K-C2α is predominantly located at CCVs and early endosomes.
To further examine the subcellular localization of PI3K-C2α, we performed membrane fractionation experiments . Briefly, lysed cells were centrifuged at increasing speeds, resulting in the collection of different cellular components in the pellet or supernatant (S6 Fig). Western blots were performed on samples P2, P3, and S3 (Fig 5A), and we observed that PI3K-C2α predominantly fractionated in the light membrane fraction P3 with the plasma membrane clathrin-adaptor protein AP2, the endosomal proteins clathrin and EEA1, and the early autophagosomal membrane precursors, ATG5, ATG9A, ATG14L, and LC3B-I. A previous report from Schekman and colleagues observed both cytosolic and membrane bound fractions of LC3B-I, and found that the levels of cytosolic versus membrane bound LC3-I can vary based on cell type or whether autophagy is defective . To further resolve the protein components of the light membranes, we subjected P3 to an OptiPrep density gradient , removed sequential levels of the density gradient, and performed Western blot analysis (Fig 5A). We observed PI3K-C2α in fractions 5 and 6, along with plasma membrane marker (AP2), endosomal markers (clathrin and EEA1), and a portion of the cis-Golgi network marker, GM130 (most of which is distributed into P2 prior to the density gradient), as well as ATG9A, ATG14L, and LC3B-I.
Our next goal was to better understand differences and similarities in the PI3K-C2α and PI3K-C3 protein complexes. To do this, we performed co-immunoprecipitation of V5-PI3K-C2α and V5-PI3K-C3 expression constructs. Whole cell lysates were probed as a control for protein loading and to monitor the expression levels of endocytic markers (clathrin, AP1, AP2, RAB5, and RAB11) and autophagy markers (LC3A, LC3B, GABARAP, ATG7, ATG9A, ATG14, ATG16L1) (Fig 5B and S7 Fig). PI3K-C2α immunoprecipitated with clathrin, AP2, RAB5, ATG9A, and ATG14L; while PI3K-C3 also immunoprecipitated with these same proteins (Fig 5B). It is important to note that while PI3K-C2α and PI3K-C3 each immunoprecipitated with overlapping markers of endocytosis and autophagy, they did not immunoprecipitate with one another. Additionally, we immunoprecipitated either WT-PI3K-C2α or KD-PI3K-C2α to determine whether the kinase activity impacted complex formation or interactions, noting a decrease in association with clathrin, ATG9A, and ATG14L with PI3K-C2α lacking kinase activity (Fig 5C), suggesting catalytic activity may be important for the integrity of PI3K-C2α complexes.
To further investigate the effects of PIK3C2A knockdown on vesicular trafficking, we measured transferrin uptake and colocalization with markers of clathrin coated vesicles (Fig 6A), EEA1-positive early endosomes (Fig 6B), and RAB11-positive recycling endosomes (Fig 6C). Control siRNA cells displayed intermediate transferrin colocalization with clathrin (44%), EEA1 (28%), and less colocalization with RAB11 (14%). Consistent with earlier data (S3 Fig), PIK3C2A knockdown resulted in an accumulation of transferrin coated vesicles at the perinuclear recycling endosome with little peripheral spread in contrast with control cells. Furthermore, PIK3C2A knockdown cells displayed a significant increase in colocalization between transferrin and clathrin (p < 0.001), and between transferrin and RAB11 (p < 0.001). Next, due to the co-fractionation and co-immunoprecipitation of ATG9 with PI3K-C2α, we examined ATG9 loss by knocking down both ATG9A and ATG9B (S1C Fig) . ATG9A/B knockdown cells displayed similar transferrin colocalization with clathrin, EEA1, and RAB11 as did PIK3C2A knockdown (Fig 6). Conversely, knockdown of PI3KC3 displayed similar transferrin colocalization with clathrin, EEA1, and RAB 11 as did control. PIK3C2A or ATG9A/B knockdown resulted in: 1) increased colocalization of transferrin with clathrin (65% colocalization with PIK3C2A knockdown, p < 0.05; 56% colocalization with ATG9A/B knockdown, p < 0.001) as compared to clathrin colocalization with control cells (44%); 2) similar colocalization of transferrin with EEA1 (35% colocalization with PIK3C2A knockdown; 32% colocalization with ATG9A/B knockdown) as compared to control cells (28%); and 3) increased colocalization of transferrin with RAB11 (p < 0.001), corresponding to 40% colocalization with PIK3C2A knockdown, 46% colocalization with ATG9A/B knockdown, as compared to 14% colocalization in control cells. To determine if the increased clathrin and RAB11 association resulted in colocalization of clathrin and RAB11 markers, we imaged transferrin, clathrin, and RAB11 within the same cell (S8 Fig). We found increased colocalization of clathrin and RAB11 within transferrin-positive vesicles (PCCs of 0.20, 0.43, 0.49 for control siRNA, PIK3C2A siRNA, and ATG9A/B siRNA, respectively). Taken together, this data suggests that PI3K-C2α and ATG9 may both be involved in the maturation of endocytic vesicles. Based on current literature surrounding ATG9 and endocytosis [50,52], we propose that PIK3C2A knockdown decreases autophagy by preventing the integration of endocytic membrane sources into the autophagy pathway through the recycling endosome (Fig 7).
Autophagy plays an important and context-dependent role in human diseases [4,8,72–74]. In order to better understand how to manipulate this complex process therapeutically, we must first determine how different proteins and pathways function in autophagy. To this end, we are interested in characterizing kinases and phosphatases that may serve as regulators of autophagy [57,58]. Here, we analyzed the eight PI3K isoforms in autophagy following knockdown and treatment with the allosteric MTORC1 inhibitor, rapamycin. In this setting, the autophagy assay measures MTORC1 dependent autophagy signaling and is suitable for the identification of both downstream effectors of MTORC1-induced autophagy (ULK1) and genes important for autophagy (PIK3C2A). Prior work has shown that Class I PI3Ks inhibit autophagy by activating AKT, which promotes MTOR signaling, thus suppressing autophagy [30,75,76]. In agreement with current literature, we find that PI3K-C2α and PI3K-C3 are positive regulators of autophagy while PIK3-C2β is a negative regulator of autophagy [29,34,43,77]. Here, we analyzed the functional consequences of an understudied PI3K family gene, PIK3C2A, to determine how the protein PI3K-C2α impacts the processes of endocytosis and autophagy. Using a combination of confocal microscopy, membrane fractionation, and immunoprecipitation, our results show that PI3K-C2α is present at both early endosomal and autophagosomal membranes, and for the first time, present a link for PI3K-C2α to autophagy and endocytosis.
While probing the kinetics of PI3K-C2α relative to the well characterized autophagy regulators, PI3K-C3 and ULK1 [12,31], we observed that these genes each have very different roles and significance in the autophagy pathway. Our results suggest that while ULK1 is critical for both acute and sustained autophagic induction, PI3K-C2α and PI3K-C3 are primarily required for more sustained autophagy. In addition, we highlight that knockdown of PIK3C2A, PIK3C3, or ULK1 results in different levels of conjugation by the ATG8 mammalian homologs, LC3A, LC3B, and GABARAP. Consistent with previous studies [78,79], our results show that these genes have unique functions with regards to autophagic turnover.
While the literature supports a role for PI3K-C2α in the budding of CCVs, our results demonstrate an additional role for PI3K-C2α downstream of budding CCVs as far as the recycling endosome [41,65,80]. Elegant work by the Hirsch lab has shown that knockdown of PI3K-C2α leads to a more peripheral localization of RAB11 ; however, in their study, RAB11 is activated by a restricted pool of PtdIns3P localized to primary cilium. Therefore, these contrasting results emphasize how treatment conditions (rapamycin) and cell type can impact the roles of PI3K-C2α . Here, the presence of PI3K-C2α, along with ATG9, is important for the maturation of CCVs. Because PI3K-C2α co-immunoprecipitated and fractionated with ATG9, and knockdown of either gene led to phenotypically similar endocytosis defects, our work further strengthens a connection between the endocytosis and autophagy pathways. Rubinsztein and colleagues reported that endocytic membrane sources can integrate directly into autophagosomal vesicles through ATG9 and ATG16L1 positive endosomes, which coalesce in the recycling endosome [50,52]; while Dikic and colleagues performed additional work on ATG9 vesicles, reporting that ATG9 interacts with the clathrin adaptor, AP2, and that AP2-positive CCVs are required for proper ATG9 localization, establishing a direct connection between ATG9 and CCVs . This work also demonstrate an interaction between ATG9, AP2, and RAB5 , all of which were present in our co-immunoprecipitation experiments with PI3K-C2α, suggesting that PI3K-C2α may be an additional component of this complex. Importantly, this complex has been shown to regulate ATG9 localization and integration of these membrane sources into the autophagy pathway through the recycling endosome .
In many instances, internalized transferrin passes through early endosomes to the recycling endosome before transport back to the plasma membrane; although some reports show this is not always linear, as transferrin can bypass the recycling endosome . We show that PI3K-C2α has a role in the recycling of internalized transferrin receptor through endosome maturation at the recycling endosome. PIK3C2A knockdown prevents the transferrin receptor from being transported back to the cell surface, likely due to defective maturation as the RAB11-postive endosomes remain clathrin-positive. Based on the current literature, we would predict that the condensed perinuclear RAB11 is maintained as an inactive state [81,83]. We further show that PIK3C3 knockdown does not result in an accumulation of RAB11 vesicles, emphasizing that PI3K-C2α and PI3K-C3 have different roles in endocytosis. A role for PI3K-C2α in CCV and endosome maturation could clarify many reported signaling phenotypes, such as altered insulin signaling  and decreased TGFβ signaling  with PIK3C2A knockdown, if PI3K-C2α loss results in the sequestering of these different receptors due to incomplete endocytic maturation. Mutations in the clathrin binding and kinase domains of PI3K-C2α are important for clathrin localization , and a better understanding of these domains in autophagy and endocytosis remains to be elucidated. Important future studies, focused on Class II PI3K isoforms, will reveal cell type and context dependent cues for targeting the autophagic and endocytic pathways.
The following antibodies from Cell Signaling Technology were used for Western blotting: ATG5 (12994S), ATG7 (8558S), ATG16L1 (8089S), β-Actin (3700S), Calnexin (2679S), Clathrin Heavy Chain (4796S), EEA1 (2411S), GABARAP (13733S), GM130 (12480S), LC3A (4599S), RAB5 (3547S), RAB11 (5589S), ULK1 (8054S), and PI3K-C3 (4263S). Western blot antibodies from other sources included ATG9A (Abcam, ab108338; GeneTex, GTX128427), ATG14 (MBL International, PD026), AP1µ2 (Santa Cruz, sc-69446), AP2M1 (Origene, TA503020), LAMP2 (Abcam, H4B4), LC3B (Sigma, L7543), p62/SQSTM1 (Abnova, H00008878-M0 1), PI3K-C2α (Santa Cruz, sc-365290), and V5 (mouse: Invitrogen, R96025; rabbit: Sigma, V8137), Secondary antibodies include HRP-linked mouse (GE Healthcare, NA931) or rabbit (GE Healthcare, NA934) IgG. Antibodies used for immunofluorescence microscopy as follows: clathrin heavy chain (Cell Signaling Technology, 4796S), clathrin light chain (Santa Cruz, sc-12735), EEA1 (Cell Signaling Technology, 2411S), LAMP2 (Abcam, H4B4), LC3B (MBL International, M152-3), RAB5 (Cell Signaling Technology, 3547S), RAB11 (Cell Signaling Technology, 5589S), and V5 (Invitrogen, R96025). Secondary antibodies include mouse AF350 (Thermo Fisher, A-11045), mouse AF488 (Thermo Fisher, A-11001), rabbit AF488 (Thermo Fisher, A-11008), and mouse AF546 (Thermo Fisher, A-11030).
The following reagents were purchased commercially: rapamycin (LC Labs, 553210), Bafilomycin A1 (AG Scientific, B-1183), DMSO (Sigma Aldrich, D2650), formaldehyde (Thermo Fisher, 28908), and Hoechst 33342 (Invitrogen, H1399).
Plasmids used include ptfLC3 (Addgene, 21074) (Kimura, 2007), EGFP-LC3B (Addgene, 11546). V5-PI3K-C2α and mCherry-PI3K-C2α were generated from PIK3C2A full length cDNA (Open Biosystems, NM_002645) by PCR amplification and TOPO cloning using pCR8/GW-TOPO entry vectors (Invitrogen, K2500-20) and pRK7-nV5/ccdB and pcDNA3.1 (Invitrogen, V790-20) destination vectors, respectively. V5-PI3K-C3 was generated from ORF (BC053651) by cloning into pRK7 by way of SalI/EcoRI using a PCR primer to introduce an N-terminal V5 tag. Kinase-Dead PI3K-C2α (KD-V5-PI3K-C2α) was generated using a QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent, 210515), following the manufacturer’s protocol and using primers to introduce point mutations,K1138A, D1157A, and D1250A, as previously characterized .
Cells were grown in antibiotic-free media supplemented with 10% fetal bovine serum (CellGro, 35-101-CV) at 37°C and 5% CO2, using McCoy’s 5A medium (Invitrogen, 16600–082) for U2OS (ATCC, HTB-96) cells and DMEM medium (Invitrogen, 11995–065) for 293FT (ATCC, PTA-5077) cells. Cell cultures were maintained for at least one week before performing experiments, and were passaged a maximum number of 20 times.
siRNA transfections were performed at a concentration of 25 nM per siRNA using 2 μL/mL of Oligofectamine transfection reagent (Invitrogen, 12252011) for 48 hours. DNA transfections were performed for 24 hours using a concentration of 5 μg/mL DNA using 1:3 μg-DNA:μl-lipid FuGENE HD (Promega, E2311).
U2OS cells stably expressing mCherry-PI3K-C2α were generated using a lentiviral expression system. A plasmid solution was made by first combining 4 μg of each lentiviral packaging plasmid (pLP1, pLP2, and pVSV-G), 4 μg of the expression vector, and 60 μL of 2 M CaCl2 in 440 μL sterile H2O. 500 μL of 2x HBSP buffer (1.5 mM Na2HPO4, 10 mM KCl, 280 mM NaCl, 12 mM Dextrose, 50 mM HEPES) was transferred to a 15 mL conical and the plasmid solution was added, dropwise while mixing. The solution was added to a 10 cm plate of 293FT cells at 90% confluency, and after 48 hours the media was collected, spun at 1,000 g for 10 minutes to remove dead cells and the supernatant was filtered to harvest the lentivirus. A fresh culture of U2OS cells were infected by adding lentivirus to a 10 cm dish of cells with 8 μg/ml polybrene (Sigma, TR-1003-G). After 24 hours, the media was changed and polyclonal cells were selected in 500 μg/mL G418 (Gibco, 11811031) containing media.
Control siRNA was AllStars Negative Control siRNA (Qiagen, SI03650318) and two or four gene-specific sequences (from Qiagen) used in experiments:
U2OS cells expressing ptfLC3 were seeded at 2.0x104 cells per well in 24 round-well plates on glass coverslips (Fisher, 12-545-81) and transfected with siRNAs to each of the eight PI3K isoforms, as described above. After 48 hours, cells were treated with 0.1% v/v DMSO control or 50 nM rapamycin in DMSO for 6 hours, and fixed with 4% formaldehyde in 1x PBS, permeabilized with 0.2% Triton X-100 in 1x PBS, stained with appropriate primary and secondary antibodies, and stained with Hoechst. Cells were imaged using an ECLIPSE Ti (Nikon) epifluorescent microscope at 60× magnification using an oil immersion lens. FITC (Excitation 465–495) and DAPI (Excitation 340–380) channels were captured for each image.
Images were analyzed using NIS-Elements Software (Nikon). Individual cells were manually outlined to specify regions of interest (ROIs) for analysis. Images were deconvolved using a fast-deconvolution and the background was removed using the detect background tool. Intensity thresholding of the FITC channel resulted in the quantification of puncta per cell which was analyzed using GraphPad Prism (n ≥ 25 cells per condition). Significance was determined using an unpaired t test using control and experimental siRNA knockdown.
U2OS cells were lysed in 10 mM KPO4, 1 mM EDTA, 10 mM MgCl2, 5 mM EGTA, 50 mM bis-glycerophosphate, 0.5% NP40, 0.1% Brij35, 0.1% sodium deoxycholate, 1 mM NaVO4, 5 mM NaF, 2 mM DTT, and complete protease inhibitors (Sigma, P8340-5ML). Proteins were resolved by SDS-PAGE, transferred to PVDF (for low molecular weight proteins like LC3) or nitrocellulose membranes, blocked with 4% w/v non-fat dry milk in 1x TBS-T (TBS with 0.05% Tween 20), and probed with primary antibodies in blocking buffer at 4°C overnight followed by secondary antibodies in blocking buffer for 45 minutes at room temperature. Proteins were detected by chemiluminescence using ECL (100 nM Tris, pH 6.8, luminol (1.25 mM), p-coumaric acid (198 μM), H2O2 (0.009%)) or SuperSignal West Femto maximum sensitivity substrate (Thermo Fisher, 34095).
U2OS cells expressing ptfLC3 were seeded at 3.15x104 cells per chamber in 4 compartment glass-bottom dishes (Greiner Bio, 627870) and transfected with siRNA pools directed to non-targeting control, PIK3C2A, PIK3C3, or ULK1, as described above. 48 hours later, the dish was transferred to a Nikon ECLIPSE Ti fitted with a live cell chamber for maintaining cells at 37°C and 5% CO2 during imaging. Three locations in each well were selected for imaging with a 60× oil submersion lens. 0.1% v/v DMSO control or rapamycin were added (50 nM in DMSO) and cells were imaged in the FITC channel (Excitation 465–495) every 10 minutes for 6 hours. Images were analyzed as described above (n ≥ 10 cells per condition). Significance was determined using an unpaired t test using control and experimental siRNA knockdown.
U2OS cells expressing EGFP-LC3B were seeded at 20,000 cells per well in 24 well round-bottom plates on glass coverslips and transfected with siRNA pools against non-targeting control or PIK3C2A as described above. After 24 hours, cells were transfected with V5-PI3K-C2α or KD-V5-PI3K-C2α as described above. Cells were treated with rapamycin (50 nM in DMSO) for 6 hours and then fixed and stained as described above. Coverslips were imaged on an A1plus-RSi scanning confocal microscope (Nikon) using 403, 488, and 561 excitation lasers and a 60× oil immersion lens. Images were analyzed as described above (n ≥ 25 cells per condition). Significance was determined using an unpaired t test using control and experimental siRNA knockdown.
U2OS cells were transfected with control or PIK3C2A siRNAs for 48 hours, cells were collected and resuspended in 2% glutaraldehyde fixative (Sigma-Aldrich, G5882). Cell pellets were embedded in 2% agarose, postfixed in osmium tetroxide, and dehydrated with an acetone series. Samples were infiltrated and embedded in Poly/Bed 812 resin, polymerized at 60°C for 24 hours, 70 nm sections generated with a Power Tome XL (Boeckeler Instruments), and placed on copper grids. Cells were examined for lipid droplets using a JEOL 100C× Transmission Electron Microscope at 100 kV with the electron microscopy services performed by the Michigan State University Center for Advanced Microscopy (East Lansing, MI).
U2OS cells were seeded at 20,000 cells per well in 24 well round-bottom plates on glass coverslips and transfected with siRNA pools against non-targeting control, PIK3C2A, ATG9A/B, or PIK3C3 as described above. After 48 hours, cells were treated with rapamycin (50 nM in DMSO) for 6 hours. Cells were then treated with 25 μg/mL transferrin from human serum, Texas Red conjugate (Thermo Fisher, T2875) at 4°C for 5 minutes, washed with fresh media, and returned to 37°C and 5% CO2 for 0–45 minutes. Cells were fixed and stained as described above. Coverslips were imaged on an A1plus-RSi scanning confocal microscope (Nikon) using 403, 488, and 561 excitation lasers and a 60× oil immersion lens. Images were analyzed using NIS-Elements Software. Individual cells were manually outlined and regions of interest were defined generating a binary ROI of transferrin localization. Overlap between endogenous antibodies and Texas Red-transferrin was calculated by quantifying the overlap of the green channel with the ROI. Results were analyzed using GraphPad Prism. Significance was determined using an unpaired t test using control and experimental siRNA knockdown.
U2OS cells stably expressing mCherry-PI3K-C2α were seeded at 20,000 cells in 24 well round-bottom plates on glass coverslips. Cells were fixed and stained as described above. Coverslips were imaged on a Nikon A1plus-RSi scanning confocal microscope using 403, 488, and 561 nm excitation lasers and a 60× oil submersion lens. Images were analyzed using NIS-Elements Software. The average Pearson’s Correlation Coefficient (PCC) was calculated by drawing ROIs around n ≥ 25 cells and averaging the PCC per cell.
U2OS cells were homogenized in lysosome lysis buffer (10 mM Tricine pH 7.2, 400 mM Sucrose, 1 mM EDTA, complete protease inhibitors, PhosSTOP (Roche, 04906845001)) using a Dounce homogenizer until > 90% of cells were broken. Lysates were spun at increasing speeds (1k x g, 3k x g, 25k x g, 100k x g) and pellets were collected at each step. The nuclear fraction and unbroken cells remained in the pellet (P0) after the initial spin at 1,000 x g and were discarded, while the supernatant underwent an additional spin at 3,000 x g where organelles, such as the mitochondria, were then collected from the pellet (P1). This process was repeated at 25,000 x g to pellet heavy membranes, such as the Golgi and autophagosomes (P2), and again at 100,000 x g to isolate light membranes, such as the plasma membrane and pre-autophagosomal vesicles (P3). Cytosolic proteins remained in the supernatant from this final spin (S3). Cell pellets were re-suspended in B88 buffer (20 mM HEPES-KOH pH 7.2, 250 mM Sorbitol, 150 mM potassium acetate, and 5 mM Magnesium acetate, filtered). Total protein was quantified using a Bradford protein assay and 10–20% of total protein was analyzed using Western blot; remaining protein was re-pelleted. The remaining pellet of interest was re-suspended in lysosome lysis buffer with 19% OptiPrep solution (Sigma, D1556) and 2.3 M sucrose. In a SW 40 Ti Rotor tube (Beckman), a gradient of OptiPrep from 35% to 0%  was layered around the 19% OptiPrep sample by diluting the OptiPrep in dilution buffer (60 mM Tricine pH 7.2, 250 mM sucrose, and 6 mM EDTA). The gradient was spun at 150k x g for 3 hours. Ten consecutive fractions were removed from the top of the gradient by pipet, diluted in lysosome lysis buffer, and pelleted at 100k x g. The resulting pellets were suspended in equal volume of B88 buffer and run analyzed using Western blot.
293FT cells were seeded in 10 cm dishes and transfected with V5-PI3K-C2α, V5-PI3K-C3, or KD-V5-PI3K-C2α as described above. Dynabead-protein G (Thermo Fisher, 10003D) was washed in 1x PBS (using a magnetic field to contain the Dynabeads in order to remove buffer) and then suspended in buffer (1x PBS, 1% Triton X100) with 1 μg of V5 antibody at 4°C for 6 hours. Beads were washed 3x with 1x PBS and cells were lysed (see above) and pelleted; supernatant was added to the beads for an overnight incubation at 4°C (with a portion of the supernatant kept as a whole cell control). Beads were washed with buffer (1x PBS, 0.1% Triton X-100). Proteins were denatured from the beads by adding 2x sample buffer (4% SDS, 100 mM Tris-HCl pH 6.8, 0.02% (w/v) Bromophenol blue, 20% (v/v) Glycerol, 2% (v/v) βME) and boiling at 100°C for 10 minutes. Resulting supernatant was analyzed by Western blot.
(A) U2OS cells were transfected with control siRNAs or siRNAs directed to each of the eight PI3K isoforms. (B) U2OS cells were transfected with control siRNAs or siRNAs directed to each of the four highest expressing PI3K isoforms in U2OS cells . (C) U2OS cells were transfected with control siRNAs or siRNAs directed to PIK3C2A, PIK3C3, ULK1, or ATG9A/B.
U2OS cells were vehicle (A, B) or rapamycin treated (C, D) to quantify autophagic flux following control siRNA (A, C) or PIK3C2A (B, D) knockdown in the presence of BafA1. The number of GFP-LC3-II puncta that accumulated in the presence or absence of BafA1 is plotted in Fig 2.
U2OS cells stably expressing EGFP-LC3B were transfected with siRNAs directed to control (A) or PIK3C2A (B) for 48 hours. After 24 hours, siRNA-resistant wild-type protein (WT-PI3K-C2α) or siRNA-resistant kinase-dead protein (KD-PI3K-C2α) was transfected. After an additional 24 hours, cells were treated with rapamycin for 6 hours and puncta per cell counted from exogenous PI3K-C2α expressing cells (red). Data is quantified in Fig 3C.
U2OS cells were transfected with control siRNAs (A) or siRNAs directed to PIK3C2A (B) for 48 hours. Following rapamycin treatment (6 hours), cells were incubated with Texas Red-conjugated transferrin. Cells were washed with fresh media and returned to 37°C for the indicated amount of time (0, 5, 10, or 45 min.) before fixation. Fixed cells were imaged using confocal microscopy with a 60× oil objective. Scale bar 10 μm.
U2OS cells stably expressing EGFP-LC3B were transfected with siRNAs directed to control (A) or PIK3C2A (B). Following rapamycin treatment (6 hours), plasma membrane was uniformly labeled with CellMask Orange at 4°C and returned to 37°C for 45 minutes. Cells were imaged using confocal microscopy with a 60× oil objective. Boxes are 5× magnification of insets. Scale bars 10 μm.
Cultured U2OS cells were homogenized and centrifuged in successive increasing speeds spins (100 ×g, 3000 ×g, 25000 ×g, and 100000 ×g). Supernatants (S1, S2, and S3) and pellets (P0, P1, P2, and P3) were collected at each step. Pellet P3 continued onto OptiPrep density gradient medium for PI3K-C2α detection (Fractions 5 and 6) and to test for markers co-eluting with PI3K-C2α. Bold text indicates pellets, supernatants, and fractions further examined in Fig 5.
V5-PI3K-C2α or V5-PI3K-C3 were immunoprecipitated and resulting 293FT lysates probed for markers of endocytosis and autophagy. Whole cell lysates (WCL) were probed with the indicated antibodies. Data presented here corresponds to Fig 5B.
U2OS cells were transfected with siRNAs directed to control (A), PIK3C2A (B), or ATG9A/B (C) for 48 hours. Following rapamycin treatment (6 hours), cells were treated with Texas Red-conjugated transferrin at 4°C. Cells were then washed with fresh media and returned to 37°C for 45 minutes. Cells were stained with antibodies for endogenous clathrin (blue) and RAB11 (green). Cells were imaged using confocal microscopy with a 60× oil objective. Scale bar 10 μm.
We thank members of the MacKeigan laboratory for advice, analysis, and reagents. We thank Alicia Withrow for electron microscopy services performed at Michigan State University Center for Advanced Microscopy. We thank Stephanie Celano, Nicole Doppel, Abigail Solitro, and Lisette Yco for critical reading of and helpful comments on the manuscript.
This work was supported by the Van Andel Institute Graduate School fellowship to N.M.M., and award number R01CA197398 from the National Cancer Institute to J.P.M. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
All relevant data are within the paper and its Supporting Information files.