Autophagy proteins regulate ERK phosphorylation

Autophagy is a conserved pathway that maintains cellular quality control. Extracellular signal-regulated kinase (ERK) controls various aspects of cell physiology including proliferation. Multiple signalling cascades, including ERK, have been shown to regulate autophagy, however whether autophagy proteins (ATG) regulate cell signalling is unknown. Here we show that growth factor exposure increases the interaction of ERK cascade components with ATG proteins in the cytosol and nucleus. ERK and its upstream kinase MEK localize to the extra-luminal face of autophagosomes. ERK2 interacts with ATG proteins via its substrate-binding domains. Deleting Atg7 or Atg5 or blocking LC3 lipidation or ATG5–ATG12 conjugation decreases ERK phosphorylation. Conversely, increasing LC3-II availability by silencing the cysteine protease ATG4B or acute trehalose exposure increases ERK phosphorylation. Decreased ERK phosphorylation in Atg5−/− cells does not occur from overactive phosphatases. Our findings thus reveal an unconventional function of ATG proteins as cellular scaffolds in the regulation of ERK phosphorylation.

E xtracellular signal-regulated kinase (ERK1/2) 1,2 promotes cellular proliferation in response to growth factors. ERK signalling also regulates expression of autophagy and lysosomal genes 3 , while ERK8 has recently been shown to stimulate autophagy by interacting with LC3 (ref. 4). Mice knocked out for p62, an adaptor protein that labels cytoplasmic cargo for autophagic degradation 5 , display ERK hyperphosphorylation 6 , suggesting bidirectional crosstalk between ERK and autophagy. Whether autophagy components regulate ERK1/2 phosphorylation is largely unknown. ERK signalling and autophagosome (APh) formation initiate at the plasma membrane 7 ; consequently, we envisioned functional associations between components of the ERK signalling cascade and APh. As the spatio-temporal regulation of ERK activity requires scaffold proteins 1,2,8 , we hypothesized that APh/LC3-II-positive structures serve as scaffolds or cellular platforms regulating ERK phosphorylation. LC3-II has been shown to be nuclear-localized 9 , however, the function of nuclear LC3 is unclear. Consequently, we investigated whether one of the possible functions of nuclear LC3-II is modulation of nuclear ERK phosphorylation. Recent studies have demonstrated new unconventional functions of autophagy (ATG) proteins in cellular protein secretion 10,11 ; however, whether ATG proteins or APh regulate cell signalling independent of their role in cargo degradation is largely unknown.
Here we show that ERK cascade components display increased association with ATG5-ATG12-positive pre-autophagosomal structures and with lipidated ATG8 family proteins, LC3-II and GABARAP 12  the upstream ERK kinase, and ERK localize to the extra-luminal face of APh. Inhibition of LC3-II formation in Atg7 À / À liver and brown adipose tissues, Atg5 À / À mouse embryonic fibroblasts (MEFs), cells silenced for ATG7, ATG5 or Beclin1, and in a LC3 lipidation-defective mutant decreases cellular and nuclear ERK phosphorylation and activity. We further show that ERK2 colocalizes with ATG5-ATG12, LC3-II and WIPI1 through its substrate-binding domains, and that ATG5-ATG12 conjugation is required for ERK1/2 phosphorylation. Conversely, increasing steady-state LC3-II content by blocking ATG4B-mediated LC3-II delipidation 13 or acutely exposing cells to the autophagy activator trehalose 14 increases phosphorylated (P)-ERK levels. Our findings suggest that an unconventional function of the ATG7/ATG5-ATG12/LC3-II cascade regulates ERK1/2 phosphorylation independent of their canonical role in lysosomal proteolysis.

Results
ERK cascade components colocalize with ATGs. To determine whether ATG proteins regulate ERK phosphorylation, we examined the effect of epidermal growth factor (EGF) treatment on interactions between ERK cascade components and APh/LC3-II. Immunofluorescence in NIH/3T3 cells revealed increased colocalization of ERK cascade components, P-bRAF, P-MEK and P-ERK with LC3-II ( Fig. 1a) in response to EGF exposure following brief serum deprivation (scheme in Fig. 1a, right). P-cRAF colocalized with LC3-II in untreated or EGF-treated or serum-fed NIH/3T3 cells ( Supplementary Fig. S1a). EGF exposure also increased colocalizations of P-ERK with GABARAP and GATE16 (hereafter collectively referred to as GABARAP) (Supplementary  Fig. S1c). Comparative analysis of EGF-treated NIH/3T3 cells verified that P-ERK colocalized predominantly with autophagosomal (LC3-II) and pre-autophagosomal (ATG5-ATG12 and ATG16) structures and with WIPI (WD-repeat protein interacting with phosphoinositides)-1 and -2 (ref. 16) that are effectors of phosphatidylinositol 3-phosphate (PI3P) in APh formation, but colocalized only modestly with vps34 and P-ULK1 (Fig. 1b). Colocalizations of P-ERK with LC3-II (Fig. 1c) and GABARAP ( Supplementary Fig. S1b) were also detected in 2 h serum-fed NIH/3T3 cells indicating that serumderived growth factors maintained P-ERK/LC3-II and P-ERK/ GABARAP colocalizations. To verify our findings of EGFinduced interactions of ERK cascade components with ATG proteins, we exposed cells to EGF following 20 h of serum deprivation, which depletes serum-derived growth factors (scheme in Supplementary Fig. S2a). Indeed, treatment of serumstarved cells with EGF increased colocalizations of P-bRAF, P-MEK and P-ERK with LC3-II, and that of P-ERK with WIPI1 and -2 ( Supplementary Fig. S2a). Intriguingly, EGF exposure of 10 min or 20 h serum-deprived NIH/3T3 cells also increased colocalization of the ERK scaffold complex protein KSR1 (kinase suppressor of Ras) 17,18 with LC3-II ( Supplementary Fig. S2b), suggesting a possible role for LC3-II-positive structures in providing scaffolding support to regulate the MEK/ERK pathway. We validated P-ERK/LC3-II colocalizations by using two species-distinct P-ERK and LC3 antibodies in EGF-treated NIH/3T3 cells ( Supplementary Fig. S3a). Specificity of P-ERK signal was further determined by analysing for loss of total and nuclear P-ERK fluorescence in U0126 (ERK phosphorylation  inhibitor)-pretreated NIH/3T3 cells exposed to EGF ( Supplementary Fig. S3b). Antibody specificities for P-bRAF, P-cRAF, P-MEK, P-ERK and LC3, and those for fluorescenttagged secondary antibodies were ascertained by lack of signal in primary antibody-( Supplementary Fig. S3c) or secondary antibody-only NIH/3T3 controls ( Supplementary Fig. S3d). Antibody specificities for P-MEK, P-ERK, LC3, vps34, WIPI1 and WIPI2 were also determined in primary antibody-only RALA rat hepatocyte controls ( Supplementary Fig. S3e).
P-MEK and P-ERK colocalized with LC3-II in hypothalamic cells ( Supplementary Fig. S4a), in 10 min or 20 h serum-deprived RALA hepatocytes exposed to EGF ( Supplementary Fig. S4b,c), and in 3T3-L1 preadipocytes cultured in adipocyte differentiation medium ( Supplementary Fig. S5a) that acutely activates ERK 19 , reflecting universality of these colocalizations in diverse cells. Agreeing with our findings in NIH/3T3 cells, P-ERK modestly colocalized with vps34 in hypothalamic cells and hepatocytes ( Supplementary Fig. S4a-c) but displayed remarkable colocalizations with WIPI1 and WIPI2 in hepatocytes ( Supplementary  Fig. S4c). Prior phosphorylation of ERK was not required for ERK to colocalize with LC3-II because U0126-pretreatment of NIH/3T3 cells did not decrease EGF-induced ERK/LC3-II colocalization ( Supplementary Fig. S5b).
MEK and ERK localize to the cytoplasmic face of APh. To determine whether MEK and ERK associate with LC3-II-positive autophagic structures in vivo we used density-gradient centrifugation to isolate APh, autophagolysosomes (APL), and lysosomes (Lys) from livers of fed mice 20 . Immunoblot analyses of total homogenates (Hom), APh, APL and Lys fractions revealed remarkable enrichment of P-and total MEK and ERK in APh when contrasted to APL or Lys (Fig. 2a). EGF stimulation increased P-MEK and P-ERK levels in APh compared with corresponding fractions from untreated NIH/3T3 cells (Fig. 2b). Increased P-MEK and P-ERK levels in EGF-treated APh (Fig. 2b) were not secondary to cytoplasmic contamination determined by the absence of cytoplasmic protein IkB in APh fractions or due to differences in APh/LC3-II content (Fig. 2b). EGF treatment modestly increased P-MEK and P-ERK levels in APL; however, these increments were not significantly greater than those in APL fractions from untreated cells (Fig. 2b). To distinguish whether MEK and ERK localized at the extra-luminal surface of APh as opposed to sequestration within APh (cartoon depicted in Fig. 2c), we utilized the LC3-II protease protection assay 21 , which is based on the notion that proteins sequestered within APh resist digestion when subjected to controlled protease exposure. Indeed, exposure of APh to increasing amounts of trypsin (0.5, 1 and 2 mg) for 15 min increased the dose-dependent digestion of APh-associated P-and total MEK and ERK and LC3-II levels (Fig. 2c), indicating that MEK and ERK localized to the cytoplasmic surface of APh. Absence of cathepsin B (lysosomal marker) in APh fractions excluded lysosomal contamination of APh fractions (Fig. 2c). Comparative analyses revealed that exposure to trypsin (0.5 mg for 15 min) led to B50% reduction in APh-associated P-ERK and total ERK levels, and B25% reduction in APh-associated P-MEK levels, whereas APL and Lys fractions resisted loss of MEK and ERK ( Supplementary Fig.  S6a). These results indicate two distinct pools of MEK and ERK: one residing on the extra-luminal surface of APh (B50% of total APH-associated ERK), and a sequestered 'protease-protected' pool inside APh, APL and Lys most likely destined for degradation. EGF treatment of NIH/3T3 cells did not increase the net flux of LC3-II or p62 to Lys ( Supplementary Fig. S6b), suggesting that ERK cascade components interact with APh generated at basal rates.
ERK cascade components interact with nuclear LC3. As activation of ERK increases its nuclear localization, we next investigated whether exposure to EGF enhanced P-ERK/LC3-II colocalization in the nucleus. EGF-treated RALA hepatocytes ( Fig. 3a) and NIH/3T3 cells (Fig. 3b), and 3T3-L1 preadipocytes exposed to an adipogenic medium ( Fig. 3c) displayed increased nuclear LC3-II puncta. EGF exposure also increased P-ERK/LC3-II colocalization in the nucleus compared with untreated NIH/3T3 cells (Fig. 3b). Nuclear P-ERK/LC3-II colocalization was also detected in serum-fed NIH/3T3 cells (Fig. 3d). To verify interactions between ERK and LC3, we co-immunoprecipitated LC3 from total Hom and nuclear fractions from fed mice livers and analysed pull-downs for ERK cascade components. Indeed, analyses of Hom and nuclear fractions revealed that LC3 co-immunoprecipitated with ERK1/2 (Fig. 3e). In addition, bRAF and MEK1/2 were also detected in pull-downs of LC3 from homogenates, indicating interactions of ERK cascade components with LC3.
To dissect the dynamics of P-ERK/LC3 interactions in the nucleus, we first investigated whether cytoplasmic LC3-II traverses to the nucleus or whether LC3-II is generated via de novo lipidation in the nucleus. To distinguish between these possibilities, we used wheat germ agglutinin (WGA) to block active protein import into the nucleus 22 . WGA pretreatment decreased basal or EGF-induced increase in nuclear LC3-II puncta by B40-50% (Fig. 3f). The partial reduction in levels of nuclear LC3 puncta in WGA-treated cells (Fig. 3f) and detection of LC3-I in nuclear fractions in vivo (Fig. 3e, bottom, and (a) EGF enhances nuclear LC3-II content in hepatocytes. Immunofluorescence (IF) depicting nuclear LC3-II in RALA hepatocytes in presence or absence of EGF (10 min). Native/inverted images are shown. Scale bar, 5 mm. Bars represent mean ± s.e.m. ***Po0.001 compared with control (Con); Student's t-test, 50 cells from two experiments. (b) EGF enhances nuclear P-ERK/LC3-II colocalization. IF depicting P-ERK (green)/LC3-II (red) colocalization in untreated (Con) or EGF-treated NIH/3T3 cells. Native images (top)/images with colocalization in white pixels (bottom) are shown. Scale bar, 5 mm. Bars represent mean ± s.e.m. ***Po0.001 compared with Con; Student's t-test, 50 cells from two experiments. (c) Adipogenic differentiation increases nuclear LC3-II. Images depict nuclear LC3-II in 3T3-L1 preadipocytes in maintenance or differentiation medium (2 h). Scale bar, 5 mm. Bars represent mean ± s.e.m. **Po0.01 compared with Con; Student's t-test, 50 cells from two experiments. (d) Nuclear P-ERK/LC3-II colocalization in serum-fed cells. IF showing P-ERK (green)/LC3 (red) colocalization in 2 h serum-fed NIH/3T3 cells. Native images (top)/images with colocalization in white pixels (bottom) are shown. Scale bar, 5 mm. (e) LC3 interacts with ERK in vivo. Immunoblots showing co-immunoprecipitation of LC3 with ERK, MEK and bRAF in homogenate (Hom) (e, top), and of LC3 with P-and total ERK in nuclear fractions from mouse livers (e, bottom). (f) Blocking nuclear transport decreases EGF-induced increase in nuclear LC3-II. LC3 IF (red) in EGF-treated NIH/3T3 cells pre-exposed (30 min) or not to WGA. Bars represent mean±s.e.m. ***Po0.001 compared to with; Student's t-test, 60 cells from n ¼ 3. Scale bar, 10 mm. (g) Blocking nuclear transport decreases nuclear ERK content. ERK IF (green) in EGF-treated NIH/3T3 cells pre-exposed (30 min) or not to WGA. Bars represent mean±s.e.m. ***Po0.001 compared with Con; Student's t-test. (h) Blocking nuclear transport does not modify P-ERK/LC3-II colocalization. IF depicting nuclear P-ERK (green)/LC3 (red) colocalization in EGF-treated NIH/3T3 cells pre-exposed or not to WGA. For (g) and (h): Scale bar, 10 mm, bars are mean±s.e.m. 50 cells from n ¼ 2. Nuclei are blue (DAPI). Arrows indicate LC3 puncta, ERK or colocalization. Supplementary Fig. S6c) suggested that de novo LC3 lipidation in the nucleus contributes to the total nuclear LC3-II content, although whether a portion of the nuclear LC3 puncta originates from interaction of LC3 with other proteins remains to be elucidated. In support of the possibility that de novo LC3 lipidation occurs in the nucleus, nuclear fractions from mouse livers revealed the presence of autophagy proteins required for LC3 lipidation, ATG7, ATG5-ATG12 and ATG16, as well as ATG4B that recycles LC3-I, and the autophagy inducer, ULK1 ( Supplementary Fig. S6c). The purity of the nuclear fractions was validated by enrichment of histone3 and absence of cytosolic marker GAPDH or endosomal marker Rab7 or lysosomal marker Cathepsin B (Supplementary Fig. S6c). Moreover, nuclear fractions from Atg7 À / À mouse livers revealed accumulation of soluble LC3-I (Fig. 4d) as typically observed in Atg7 À / À total lysates (Fig. 4a). WGA treatment also decreased nuclear ERK   23 ; however, WGA treatment did not modify the percentage of P-ERK that colocalized with nuclear LC3-II (Fig. 3h). We next investigated whether ERK gains nuclear entry by 'piggybacking' onto LC3-II. LC3-II-deficient Atg5 À / À MEFs did not display reduced nuclear ERK content ( Supplementary Fig. S6d), in fact Atg5 À / À MEFs showed increased total ERK levels in the nucleus, indicating that LC3-II per se does not traffic ERK into the nucleus.
phosphorylation 26 , and agreeing with this notion decreased ERK phosphorylation in Atg7 À / À livers associated with significantly reduced ERK2 dimer levels (Fig. 4b). Atg7 À / À livers also displayed reduced nuclear P-ERK1 and P-ERK2 levels when normalized to corresponding total nuclear ERK levels, independent of changes in nuclear protein loading (equivalent Nopp140 levels) or cytoplasmic contamination (absence of IkB) (Fig. 4d). Decreased ERK phosphorylation in Atg7 À / À livers did not occur due to early loss of autophagy during development because transiently silencing ATG7, ATG5 or Beclin1 in NIH/ 3T3 cells decreased EGF-induced phosphorylation of ERK2 by B28, B33 and B43% respectively ( Supplementary Fig. S7d). In addition, B44% reduction in ERK1 phosphorylation was observed in NIH/3T3 cells silenced for Beclin1 ( Supplementary Fig. S7d).
(e) Decreased ERK phosphorylation in Atg5 À / À MEFs occurs independently of changes in ERK phosphatases. Immunoblots for the indicated proteins in 2 h serum-deprived WT MEFs transfected with scrambled siRNA (scr), and Atg5 À / À MEFs transfected with scr or siRNAs against MKP3 or PP2A in the presence or absence of EGF. The bars represent mean±s.e.m., n ¼ 3.
Conversely, ERK substrates interact with ERK2 through substrate D-domains that include basic residues followed by a hydrophobic LXL motif or via substrate F-sites that consist of the FXFP motif 33,34 . Analyses of ATG5, ATG12, WIPI1, WIPI2, LC3B and GABARAP sequences for FXFP and LXL domains revealed the presence of the bonafide LXL motif only in ATG12 and LC3B ( Supplementary Fig. S10b), and absence of these domains in other ATG proteins examined (Supplementary Fig.  S10b). The absence of FXFP or LXL domains in ATG5, and the presence of the LXL domain in ATG12 suggest that P-ERK's interaction with ATG5-ATG12 conjugate (Fig. 1b) occurs in all likelihood through ATG12 and not via ATG5. Consequently, to determine the role of ATG5-ATG12 conjugation in the regulation of ERK phosphorylation, we examined for levels of EGF-driven ERK1/2 phosphorylation in the conjugation-defective ATG5 K130R mutant 35 , which fails to bind to ATG12. Indeed, EGF-treated NIH/3T3 cells transfected with the ATG5 K130R mutant displayed decreased ERK phosphorylation compared with WT ATG5-transfected counterparts (Fig. 6c), indicating that activation of autophagy, and consequently, availability of ATG5-ATG12 conjugate regulates the degree of ERK phosphorylation. Our findings cannot exclude the possibility that ERK2 may also interact with ATG proteins lacking bonafide ERK-binding sites through alternate yet-unknown binding sites, as observed with the ERK2 substrate stathmin 31 , which also lacks typical ERK2 binding sites. Overall, these results indicate redundancy in the availability of ATG proteins that could potentially serve as signalling platforms to maintain ERK phosphorylation.
Depleting ATG4B increases LC3-II and ERK phosphorylation. ATG4B modulates LC3-II availability by cleaving LC3 C-terminal glycine and allowing LC3 lipidation 36 , and delipidates LC3-II to recycle LC3-I. We thus investigated whether depleting ATG4B levels modifies LC3-II content, which affects P-ERK/LC3-II interaction and ERK phosphorylation. Partial loss of ATG4B (B50%) in NIH/3T3 cells did not affect ATG7 or ATG5-ATG12 levels, but increased steady-state LC3-II levels (B30%) (Fig. 7a) similar to elevated LC3-II levels in Atg4B À / À heart and brain 37 . We verified increased autophagic vacuolar content in ATG4Bdeficient cells by direct visualization of double-membrane vesicular structures containing sequestered cytoplasmic cargo (Fig. 7b), although increased LC3-II content in ATG4B-deficient cells did not translate to increased LC3-II flux in serum-fed conditions (Fig. 7c). ATG4B deficiency also increased nuclear LC3-II to levels comparable to those observed in EGF-treated scrambled siRNA-transfected controls (Fig. 7d). Most notably, EGF treatment increased P-ERK/LC3-II colocalization in ATG4B-deficient cells (Fig. 7e). ATG4B-deficient cells also displayed B40% increase in nuclear P-ERK/LC3-II colocalization (Fig. 7f), although equivalent P-ERK/LC3-II colocalizations were detected in EGF-treated controls and ATG4B-deficient cells (Fig. 7f). In consistency with our idea that LC3-II is required for ERK phosphorylation, augmenting LC3-II via partial loss of ATG4B increased P-ERK levels in untreated or EGFtreated cells (Fig. 7g and Supplementary Fig. S14). We also observed significantly increased P-ERK levels in the nuclear compartment of ATG4B-deficient cells regardless of EGF exposure (Fig. 7h). In addition, short-term (2 h) exposure of trehalose, which activates autophagy independent of the mTOR pathway 14 , enhanced ERK1/2 phosphorylation ( Supplementary  Fig. S11a). Enhanced ERK phosphorylation occurred in an autophagy-dependent manner, as trehalose treatment for 2 h failed to augment ERK phosphorylation in Atg5 À / À MEFs ( Supplementary Fig. S11a). Although 12 h or 24 h of trehalose treatment robustly increases LC3-II levels compared with 2 h of trehalose treatment ( Supplementary Fig. S11b), prolonged trehalose exposure did not increase ERK phosphorylation to levels greater than those observed in untreated controls ( Supplementary Fig. S11b). We suspect that robust autophagy activation following prolonged trehalose exposure involves rapid turnover of available LC3-II-positive vesicles, which decreases the availability of LC3-II-positive structures to regulate ERK phosphorylation.

Discussion
In totality, our results demonstrate that the cellular availability of autophagic structures determines the degree of ERK phosphorylation. We speculate that LC3-II-positive membranes and ATG5-ATG12-positive preautophagosomes could serve as scaffolds or cellular signalling platforms that would allow efficient spatial coordination of the Raf-MEK-ERK cascade and thus facilitate ERK phosphorylation (modelled in Fig. 8). In support of this notion, we observe that MEK and ERK localize to the cytoplasmic face of APh, and that ERK2 interacts with these structures through its substrate-binding domains. Furthermore, deleting Atg7 or blocking LC3 lipidation or ATG5-ATG12 conjugation decreases ERK phosphorylation independent of changes in MEK phosphorylation. We also find that deleting Atg5 modifies ERK phosphorylation depending on the nutrient status of the cell. While serum starvation of Atg5 À / À cells for 41 h reduced ERK phosphorylation and its ability to phosphorylate p90RSK, transiently removing serum for 10 min increased ERK phosphorylation in Atg5 À / À cells. Atg5 À / À cells are unable to generate LC3-II and are completely deficient in canonical autophagy; however, surprisingly we observed increased basal GABARAP puncta in Atg5 À / À cells. Thus, it remains possible that during early nutrient deprivation, Atg5 À / À LC3-II-deficient cells utilize GABARAP-positive structures as cellular platforms for ERK phosphorylation, which is supported by our observation of increased GABARAP/P-ERK colocalization in Atg5 À / À cells.
These results indicate a functional redundancy in the ability of multiple ATG proteins, that is, lipidated LC3-II, ATG5-ATG12 conjugate or GABARAP-positive vesicles, to regulate ERK phosphorylation. It is thus not surprising that blocking individual ATG proteins, for instance, LC3 lipidation or ATG5-ATG12 conjugation only modestly decreases ERK phosphorylation, whereas deleting the upstream E1-like ligase Atg7 robustly blocks ERK phosphorylation.
It is reasonable to argue that reduced cargo degradation in autophagy-deficient cells decreases ATP production, and this in turn could have limited ERK phosphorylation; however, intact phosphorylation of JNK, mTOR, STAT3 and ULK1 in Atg5 À / À cells excludes the possibility that ATP depletion has a role in decreased ERK phosphorylation. Overactivation of phosphatases was also excluded as a factor contributing to decreased ERK phosphorylation in Atg5 À / À cells, as silencing bona fide ERK phosphatases, MKP3 or PP2A failed to restore ERK signalling in Atg5 À / À cells.
The physiological outcome of changes in ERK phosphorylation in cells with aberrant autophagy remains to be determined. We predict that the previously described increases in cell death in autophagy-deficient cells 38 or cellular senescence from reduced autophagy with age 39,40 may have occurred, in part, from decreased ERK activity. We also predict that some of the previously reported effects of loss of autophagy on adipocyte mass and differentiation 41,42 may occur in part from impaired ERKregulated adipocyte differentiation 19 . Hyperactive Raf-MEK-ERK signalling is implicated in hepatocellular cancers, and a recent study has reported increased cell death in Sorafenib (Raf inhibitor)-treated hepatoma cells knocked down for ATG7 (ref. 43), suggesting that therapeutic benefit in Sorafenib-treated ATG7-deficient hepatoma cells may have occurred from a more robust inhibition of ERK signalling. Aberrant autophagy may Representative electron micrographs depicting autophagic vesicles (indicated by black arrows) in scr and siATG4B NIH/3T3 cells. (c) ATG4B depletion does not increase autophagic flux in serum-fed cells. Immunoblots for indicated proteins in lysates from scr or siATG4B NIH/3T3 cells in presence/absence of lysosomal inhibitors, ammonium chloride and leupeptin (Inh) for 2 h. Scale bars, 1 mm. Bars represent mean±s.e.m., n ¼ 3. (d) siATG4B cells display increased nuclear LC3-II. Images for LC3-II (red) in scr or siATG4B NIH/3T3 cells in presence/absence of EGF (10 min). Scale bar, 5 mm. Bars represent mean±s.e.m. ***Po0.001 compared with scr; Student's t-test, 60 cells from n ¼ 2. (e) ATG4B deficiency increases P-ERK/LC3-II colocalization. Immunofluorescence (IF) for P-ERK (green)/LC3 (red) colocalization in scr or siATG4B NIH/3T3 cells in presence/absence of EGF (10 min). Scale bars, 10 mm. Bars represent mean±s.e.m. **Po0.01, ***Po0.001 compared with scr; Student's t-test, 60 cells from n ¼ 2. (f) ATG4B-deficient cells display increased nuclear P-ERK/LC3 colocalization. IF depicting P-ERK (green)/LC3 (red) colocalization in nuclei of scr or siATG4B NIH/3T3 cells in presence/absence of EGF. Scale bars, 5 mm. Bars represent mean±s.e.m. **Po0.01, ***Po0.001 compared with scr; Student's t-test, 60 cells from n ¼ 2.
Nuclei are blue (DAPI). ARTICLE potentially link ERK hyperphosphorylation to the development of ERK-dependent tumours. In light of these associations, manipulating ATG5-ATG12 and LC3-II availability could be a therapeutic strategy against multiple disorders stemming from altered ERK activity.
Autophagy LC3 flux assays. Flux assays were used to quantify autophagy activity wherein accumulation of autophagy substrates, LC3-II or p62, in the presence of inhibitors of lysosomal proteolysis, ammonium chloride (20 mM) and leupeptin (100 mM), reflects activity. Briefly, cells were cultured in the presence or absence of lysosomal proteases for 2 h following which, cells were collected and lysed and subjected to immunoblotting for LC3-II or p62. Autophagy flux was determined by subtracting the densitometric value of inhibitor-untreated LC3-II or p62 from corresponding inhibitor-treated values.
Lysates (500 mg of homogenate or 150 mg of nuclear fractions) were incubated overnight in rotation at 4°C with 100 ml of protein-A sepharose beads (Sigma-Aldrich) crosslinked to LC3 antibody (MBL International, Woburn, MA, USA). For covalent crosslinking, beads were washed (2,500 r.p.m./5 min at 4°C) with 0.2 M Borate/3 M NaCl buffer (pH 9.0) and crosslinked with 50 mM dimethyl pimelimidate in Borate buffer (30 min) in rotation at RT. Coupling reactions were stopped with 0.2 M ethanolamine (pH 8.0) for 2 h RT followed by incubation with 200 mM glycine (pH 2.5). Crosslinked beads were incubated with samples in rotation overnight at 4°C. Bound proteins were eluted by boiling (95°C for 5 min) in 2 Â SDS-PAGE sample buffer. Immunoprecipitated (IP) proteins and original lysates (input) were resolved on a SDS-PAGE, and membranes were probed for bRAF, MEK, ERK, P-ERK and LC3. P-ERK1, P-ERK2, P-ERK1/2 or P-MEK1/2 bands obtained by immunoblotting were quantified using ImageJ and normalized to corresponding total ERK1, total ERK2, total ERK1/2 or total MEK1/2, respectively. Full scans of western blots are supplied in Supplementary Fig. S12. The antibody concentrations are supplied in Supplementary Table S1.
Electron microscopy. Electron microscopy was performed in NIH/3T3 cells cultured in monolayers. Cells were collected and fixed in a buffer containing 2.5% glutaraldehyde and 2% paraformaldehyde in 100 mM sodium cacodylate (pH 7.43) 20 . Cells were post-fixed in 1% osmium tetroxide/sodium cacodylate buffer followed by fixation in 1% uranyl acetate. Cells were ethanol-dehydrated and embedded in LX112 resin (LADD Research Industries). Ultrathin sections were sequentially stained with uranyl acetate and lead citrate. Grids were visualized using a JEOL 1200CX II transmission electron microscope at 80 kV. Identification of autophagic vacuoles was done using established criteria 48 .
Fluorescence microscopy. Cells on coverslips were fixed with a 4% paraformaldehyde solution, blocked and incubated with primary and corresponding secondary antibodies (Alexa Fluor 488 and/or Alexa Fluor 647 conjugated) (Invitrogen). Mounting medium contained DAPI (4 0 ,6-diamidino-2-phenylindole) to visualize the nucleus (Invitrogen). Images were acquired on a Leica DMI6000B microscope/ DFC360FX 1.4-megapixel monochrome digital camera (Leica Microsystems, Germany) using Â 100 objective/1.4 numerical aperture. Images in each experiment were acquired at same exposure times within the same imaging session. Image slices/stacks of 0.2 mm thickness were captured and deconvolved using the Leica MetaMorph acquisition/analysis software. All images were prepared using Adobe Photoshop and subjected to identical post-acquisition brightness/ contrast effects. Representative native and/or inverted images are shown. Quantification was performed in individual frames after deconvolution and thresholding using the ImageJ software (NIH) in a minimum of 20 cells per slide and a minimum of 50 cells from two or more experiments. Particle number was quantified with the 'analyse particles' function in threshold single sections with size (pixel 2 ) settings from 0.1 to 10 and circularity from 0 to 1. Cellular fluorescence intensity was expressed as mean integrated density as a function of individual cell size. Nuclear fluorescence intensity was expressed as a function of nuclear size after the identification of the nuclear perimeter by the ImageJ 'freehand selection tool'. Percentage colocalization was calculated using the JACoP plugin in single Z-stack sections of deconvolved images. Colocalization is shown in merged native images and/or as white pixels using the 'colocalization finder' plugin in ImageJ.
Plasmids. Cyan fluorescence protein (CFP)-LC3, mutant CFP-LC3 DG in which the C-terminal glycine (120) is deleted 27 , WT ATG5 (Addgene plasmid #13095) 35 , mutant ATG5 K130R (Addgene plasmid #13096) 35  Statistics. All numerical data arising from biochemical and cell fractionation analyses are represented as mean and s.e.m., and are from a minimum of three independent experiments unless otherwise stated. All imaging data are represented as mean and s.e.m, and represent data from at least 50 individual cells from two or more experiments. We determined the statistical significance of the difference between experimental groups in instances of single comparisons by the two-tailed unpaired Student's t-test of the means. For multiple means comparisons, one-way analysis of variance (ANOVA) followed by Bonferroni post hoc test was used to determine statistical significance.