mAtg9 binds to p38IP in vitro and in vivo. (A) The AH109 strain was co-transformed with pGBKT7 containing the C-terminal (CT) domain of mAtg9 (bait) and pGADT7, or pACT2, containing the CT domain of p38IP (prey). The interaction was then visualized by growth of co-transformed yeast on SD-Trp/Leu/His plates for 3 days at 30°C. (B) Domain organization of p38IP, the N- (1–487 aa) and C- (487–733 aa) terminal domains. p38IP has a predicted nuclear localization signal (NLS), a PEST domain, and two serine-rich domains. The region of interaction with p38α and mAtg9 is indicated. (C) HEK293A cells were transfected with RFP–mAtg9 and HA–p38IP. HA–p38IP was immunoprecipitated with an anti-HA antibody and the co-immunoprecipitated RFP–mAtg9 was detected with an anti-mAtg9 antibody. HA–p38IP was detected with an anti-HA antibody. Asterisk indicates a non-specific band. Indirect immunofluorescence analysis of mAtg9 with (D) HA–full-length p38IP (HA–FL), (E) HA–C-terminal p38IP (HA–CT), and (F) HA–N-terminal p38IP (HA–NT). Bars=5 μm. The co-localization seen in (D) and (E) was not because of overlap of randomly distributed vesicles. Cross-correlation function (CCF) analysis (van Steensel et al, 1996) showed a peak and maximum Rp value around ΔX=0, indicating that co-localization between mAtg9 and p38IP is positively correlated and non random (Supplementary Figure S1).

p38IP is found on membranes, co-localizes with mAtg9 and is required for starvation-dependent trafficking of mAtg9. (A) HEK293A cells were transfected with HA–p38IP alone or co-transfected with HA–p38IP and RFP–mAtg9, homogenized and subjected to centrifugation, and the resulting post-nuclear supernatant (PNS) was fractionated by centrifugation at 100 000 g into membrane pellet and cytosol. Equal protein amounts were resolved by SDS–PAGE and immunoblotted with anti-mannose-6-phosphate receptor (MPR) antibody as a control for membrane-associated proteins, anti-superoxide dismutase (SOD) antibody as a control for cytosolic proteins, anti-mAtg9 antibody, and anti-HA antibody. HA–p38IP levels were quantified using ImageJ software and plotted relative to HA–p38IP in the PNS (n=4) (^**P=0.0051). (B) HEK293 cells were depleted of mAtg9 using siRNA, then transfected with HA–p38IP and incubated in full medium or EBSS (data not shown). The asterisk indicates mAtg9-depleted cells. Bar is 5 μM. The intensity of the nuclear versus cytosolic fluorescence in control and cells depleted of mAtg9 under fed and starvation conditions was analysed using ImageJ software (^***P=<0.0001, Students t-test). (C) HEK293A cells were transfected with HA–p38IP and processed for indirect immunofluorescence. HA–p38IP was detected with an anti-HA antibody and endogenous mAtg9 was detected with an anti-mAtg9 antibody. HA–p38IP localizes to the nucleus and cytoplasmic puncta in full medium or EBSS. HA–p38IP co-localizes with mAtg9 in the periphery of the cell (arrowheads). Bars (upper panels) 5 μM, (lower panels) 1 μM. (D) Anti-mAtg9 antibody was used to immunoprecipitate mAtg9 from HEK293 cells transfected with HA–p38IP alone or with RFP–mAtg9 and HA–p38IP. A total of 5% of lysates or immunoprecipitates were probed with anti-HA or anti-mAtg9 antibodies. The left and right hand bottom panels were obtained from the same immunoblot, but exposed for different times to allow the endogenous mAtg9 to be visualized better. (E) HEK293 cells were transfected with control or p38IP siRNA. At 72 h after transfection, cells were incubated in either full medium or EBSS for 2 h, then fixed and immunostained to detect endogenous mAtg9 localization. mAtg9 localization was quantified by visually scoring cells for either a juxta-nuclear or dispersed phenotype. Bars=5 μm (data are represented as mean±s.e.m. of 200 cells, ^*P=0.0413, Student's t-test).

p38IP is required for mammalian autophagy.(A) 293/GFP–LC3 cells were transfected with control or p38IP siRNA. At 72 h after transfection, cells were incubated in either full medium (FM), full medium with leupeptin (FM Leu), EBSS (ES), or EBSS with leupeptin (ES Leu) for 2 h. FM Leu (data not shown) is identical to FM alone. Quantification of GFP–LC3-positive autophagosome number was performed by counting in a blinded experiment. Bars=5 μm (data are represented as mean±s.e.m. of 60 cells, ^***P=0.0002, Students t-test). (B) HEK293A cells were transfected with siRNA control, siRNA for p38IP, and siRNA for ULK1. Cell lysates were analysed by SDS–PAGE for endogenous LC3 lipidation using an anti-LC3 antibody, or immunoblotted with anti-ULK1 antibody. (C) HeLa cells were transfected with siRNA control, and siRNA for p38IP. After incubation for 2 h as in (A), samples were immunoblotted with anti-LC3 and anti-actin antibodies. Endogenous LC3II/LC3I levels were quantified and the ratio presented as arbitrary units. In (B), n=4, ^***P=<0.0001 and ^*P=0.0324, Students t-test. In (C), data are representative of two experiments. (D) 293/GFP–LC3 cells were transfected with siRNA control or siRNA p38IP, and incubated in full medium, or for 2 and 4 h in EBSS. Samples were immunoblotted with anti-p62 antibodies, and actin as a loading control. (E) HEK293A cells were transfected with HA–p38IP or myc–ULK1 C-terminal domain (CTD). After 24 h, cells were labelled with [¹⁴C]valine as described in Material and Methods section, and incubated in either full medium or EBSS for 2 h. Cells were then collected and analysed for long-lived protein degradation (data are represented as mean±s.e.m. of triplicates, representative of two experiments, EBSS mock versus EBSS HA–p38IP (^***P=0.036); EBSS mock versus ULK1 CTD (^***P=<0.0001); Students t-test).

Activation of p38 inhibits autophagy and mAtg9 trafficking. (A) 293/GFP–LC3 cells were treated with 10 μM anisomycin for 30 min, or exposed to UV irradiation for 3 min followed by a 40-min recovery and incubation in full medium, full medium with leupeptin, EBSS, or EBSS with leupeptin for 2 h. Cells were then fixed and visualized by confocal microscopy. GFP–LC3-positive structures per cell were quantified as in Figure 3A. Bars=5 μm. Data are represented as mean±s.e.m. n=60 cells, mock versus anisomycin EBSS (^***P=<0.0001); mock versus UV EBSS (^***P=<0.0001); mock versus anisomycin EBSS with leupeptin (^***P=<0.0001); mock versus EBSS with leupeptin UV (^***P=<0.0001). All analysed using Student's t-test. (B) HEK293A cells were pretreated with 10 μM anisomycin for 30 min, or UV irradiation for 3 min followed by a 40-min recovery, before incubation in EBSS for 2 h. Images were analysed as in Figure 2E. Bars=5 μm. Data are represented as mean±s.e.m. of 60 cells, EBSS mock versus EBSS anisomycin (^*P=0.0142); EBSS mock versus EBSS UV (^*P=0.0150); Student's t-test.

The role of p38IP in p38α-dependent inhibition of autophagy. (A) 293/GFP–LC3 cells were pre-treated with 10 μM anisomycin for 30 min before incubation in either full medium or EBSS. Cell lysates were analysed by SDS–PAGE for GFP–LC3 lipidation using an anti-LC3 antibody. The membrane was also probed with anti-β tubulin. The GFP–LC3 lipidation was quantified as the amount of GFP–LC3II/GFP–LC3I (data are represented as mean±s.e.m. of triplicates, P=0.0091, Student's t-test). (B) HEK293A cells labelled with [¹⁴C]valine, as described in Material and Methods section, were pretreated with 10 μM anisomycin for 30 min before incubation in either full medium, EBSS, or EBSS with leupeptin for 2 h. Cells were then analysed for long-lived protein degradation (data are represented as mean±s.e.m. of triplicates, representative of two experiments, EBSS untreated versus anisomycin treated (P=0.0363); EBSS untreated versus leupeptin treated (P=<0.0001), Student's t-test). (C) HEK293 cells were transfected with control siRNA or siRNA for p38IP. After 24-h incubation, control or p38IP-depleted cells were pre-treated with anisomycin, followed by a 2-h incubation with full medium, full medium with leupeptin, EBSS, or EBSS with leupeptin. Cell lysates were analysed with anti-p38α or anti-LC3 antibodies. LC3II/LC3I levels were quantified and presented normalized to control, untreated cells incubated in EBSS. Data are representative of two experiments. (D) HEK293A cells transiently transfected with HA-p38IP and RFP-mAtg9 were treated with anisomycin for 30 min and mAtg9 was immunoprecipitated. 5% of the lysates or the immunoprecipitates were probed with anti-HA and anti-mAtg9 antibodies (data is representative of 3 experiments, ^*P=0.0332).

Overexpression of p38α inhibits autophagy and competes with mAtg9 for binding to HA–p38IP. (A) HEK293A cells were transfected with RFP–mAtg9 and Flag–p38 (lanes 2 and 7), RFP–mAtg9, Flag–p38 and HA–p38IP (lanes 3 and 8), or RFP–mAtg9 and HA–p38IP (lanes 4 and 9). RFP–mAtg9 was immunoprecipitated with an anti-mAtg9 antibody, the co-immunoprecipitated HA–p38IP was detected with an anti-HA antibody, and co-immunoprecipitated Flag–p38 was detected with an anti-Flag antibody. RFP–mAtg9 was detected with an anti-Atg9 antibody. (B) HEK293A cells were transfected with RFP–Atg9 and HA–p38IP (lanes 1 and 5), RFP–mAtg9, Flag–p38, and HA–p38IP (lanes 2 and 6), or Flag–p38 and HA–p38IP (lanes 3 and 7). HA–p38IP was immunoprecipitated with an anti-HA antibody, the co-immunoprecipitated RFP–Atg9 was detected with an anti-Atg9 antibody, and co-immunoprecipitated Flag–p38 was detected with an anti-p38α antibody. HA–p38IP was detected with an anti-HA antibody. (C) HEK293A cells were transfected with Flag–p38α for 24 h. Cells were incubated in either full medium, EBSS, or EBSS plus leupeptin for 2 h, lysed, and analysed for endogenous LC3 lipidation using an anti-LC3 antibody, and immunoblotted with anti-Flag antibody. LC3 lipidation was quantified as the amount of LC3II/LC3I (data are represented as mean±s.e.m. of triplicates, n=2 experiments, EBSS control versus EBSS with Flag-p38α, ^***P=0.0026).

p38α requires p38IP for inhibition of autophagy. (A) 293/GFP–LC3 cells were transfected with control or p38α siRNA. At 72 h after transfection, cells were incubated in either full medium, EBSS, or EBSS with leupeptin for 2 h, then fixed and visualized by confocal microscopy. Bars=5 μm (data are represented as mean±s.e.m. n=60 cells, EBSS control versus p38α siRNA (^***P=<0.0001); EBSS with leupeptin control versus p38α siRNA (^***P=<0.0001), Student's t-test). (B) HEK293A cells were transfected with control or p38α siRNA. At 72 h after transfection, cells were incubated in either full medium, EBSS, or EBSS containing leupeptin for 2 h. Cells lysates were analysed for LC3 lipidation using an anti-LC3 antibody. The membrane was also probed with anti-actin and anti-p38α antibodies. LC3 lipidation was quantified as the amount of LC3II/LC3I (data are represented as mean ±s.e.m. n=5, full medium control versus p38α siRNA (^**P=0.0056); EBSS control versus p38α siRNA (^**P=0.048); EBSS with Leu control versus p38α siRNA (^**P=0.04), Student's t-test). (C) HEK293A cells were transfected with control or p38α siRNA. At 72 h after transfection, cells were incubated in either full medium or EBSS for 2 h, then fixed and immunostained to detect endogenous mAtg9 localization. The percentage of cells with dispersed mAtg9 was quantified as in Figure 2E (data are represented as mean±s.e.m. n=200 cells, full medium control versus p38α siRNA (^**P=0.0027); EBSS control versus p38α siRNA (^*P=0.0459); Student's t-test). Bars=5 μm (D) HEK293A cells were transfected with siRNA for p38α, homogenized and subjected to centrifugation, and the resulting post-nuclear supernatant (PNS) was fractionated by centrifugation at 100 000 g into membrane pellet and cytosol as in Figure 2A. Equal protein amounts were resolved by SDS–PAGE and immunoblotted with anti-MPR, anti-HA, and anti-p38α antibodies. Data are representative of four experiments. (E) HEK293A cells were transfected with control, p38α siRNA or both p38α siRNA, and p38IP siRNA. At 72 h after transfection, cells were incubated in either full medium, full medium with leupeptin, EBSS, or EBSS with leupeptin for 2 h. Cells lysates were analysed for LC3 lipidation using an anti-LC3 antibody. The membrane was also probed with anti-p38α. LC3 lipidation was quantified as the amount of LC3II/LC3I. Data are representative of two experiments.

Model for regulation of autophagy through mAtg9, p38IP, and p38 MAPK. In full medium, mAtg9 traffics between the TGN and endosomes. Autophagy is inhibited by phosphorylated p38α. p38IP is found in the pool of mAtg9 in peripheral endosomes, and in complex with p38α. It should be noted that the two pools of p38α–p38IP are identical and shown separately for clarity. In starvation, p38α is dephosphorylated and binds p38IP with a lower affinity, and can no longer inhibit autophagy (shown as a dashed line). The pool of p38IP released from p38α would facilitate mAtg9 trafficking and autophagy.