The Noncanonical Role of ULK/ATG1 in ER-to-Golgi Trafficking Is Essential for Cellular Homeostasis

Abstract

ULK1 and ULK2 are thought to be essential for initiating autophagy, and Ulk1/2-deficient mice die perinatally of autophagy-related defects. Therefore, we used a conditional knockout approach to investigate the roles of ULK1/2 in the brain. Although the mice showed neuronal degeneration, the neurons showed no accumulation of P62(+)/ubiquitin(+) inclusions or abnormal membranous structures, which are observed in mice lacking other autophagy genes. Rather, neuronal death was associated with activation of the unfolded protein response (UPR) pathway. An unbiased proteomics approach identified SEC16A as an ULK1/2 interaction partner. ULK-mediated phosphorylation of SEC16A regulated the assembly of endoplasmic reticulum (ER) exit sites and ER-to-Golgi trafficking of specific cargo, and did not require other autophagy proteins (e.g., ATG13). The defect in ER-to-Golgi trafficking activated the UPR pathway in ULK-deficient cells; both processes were reversed upon expression of SEC16A with a phosphomimetic substitution. Thus, the regulation of ER-to-Golgi trafficking by ULK1/2 is essential for cellular homeostasis.

Figure 1. Degeneration of pyramidal neurons in the CA1 region of ULK1/2-deficient mice

(A) Survival of control (n=20) and Ulk1/2-cdko (n=15) mice. (B) Average body weight ±SEM of male (M) and female (F) Ulk1/2-cdko (n=11) mice compared to control (n=26) mice. (C) The average brain weights±SEM of control (n=5) and Ulk1/2-cdko (n=5) mice did not significantly differ (Student’s t-test). (D) – Representative images of serial brain sections stained with hematoxylin and eosin (H&E), DAPI, and antibodies against the neuronal marker NeuN. Scale bars: 200 μm (E) Average number of pyramidal neurons (normalized to that in littermate controls) ±SEM in a 500-μm² area of CA1 (n=3 mice/genotype for each age group).. *P <0.001 (Student’s t-test) when compared with control. (F) Representative images of serial brain sections stained with antibodies against NeuN and the cerebellar Purkinje cell marker, calbindin. Scale bar: 500 μm. (G) Representative images of Fluoro-Jade C stained and cleaved Caspase 3 (Casp-3) immunostained brain sections from the f 16-wk-old mice. The CA1 region is indicated by brackets. Scale bars: 200 μm (Fluoro-Jade C); 50 μm (cleaved Casp-3). (H) Representative images of serial brain sections from hippocampal region of16-wk-old mice stained with antibodies against GFAP or IBA1 and counterstained with anti-NeuN and DAPI. Scale bars: 50 μm. See also Figure S1.

Figure 2. Ulk1/2 deficiency in hippocampal neurons is associated with activation of the UPR pathway

(A) Representative images of brain sections from 4-wk-old Atg7-cko, 8-wk-old Ulk1/2-cdko and 8-wk-old control mice stained with antibodies against ubiquitin and P62, and counterstained with DAPI. Scale bars: 200 μm. (B) Representative electron micrographs of neurons in CA1 region from 8-wk-old mice. Scale bars: 1 μm. (C) Morphometric analyses of electron micrographs of hippocampal CA1 neurons from 8-wk-old control (n=2) and Ulk1/2-cdko (n=2) mice. *P <0.001 (Student's t-test). (D-F) Representative images of brain sections from the hippocampal region of16-wk-old mice stained with antibodies against p-eIF2α, ATF6 or CHOP, and counterstained with anti-NeuN. Sections were counterstained with anti-NeuN and DAPI. Scale bars: 50 μm. (G) Diagram of the UPR pathway, highlighting the activation of the PERK and ATF6 arms in the Ulk1/2-cdko mice. See also Figure S2.

Figure 3. ULKs mediate the phosphorylation of SEC16A and facilitate in vitro budding of COPII vesicles

(A) Representative immunoblot analyses of IPs from WT MEFs transfected with either control nontargeting siRNA (Ctrl) or Sec16a siRNA using anti-ULK1 antibodies in the presence or absence of epitope-specific blocking peptides (BP). (B) Representative immunoblot analyses of ULK1 IPs from the hippocampus of 4-wk-old WT mice. (C) Representative immunoblot analyses of ULK1 IPs from WT MEFs transfected with either control nontargeting siRNA (Ctrl) or Atg13 siRNA. (D) Representative immunoblot analyses of 293T cells transfected with GFP–SEC16A and either C-terminal–MYC-DDK–tagged ULK1 or ULK2 expression constructs. Phosphorylated SEC16A was detected using the anti-p(S/T)Phe antibody. (E) Representative immunoblot analyses of GFP IPs from 293T cells transfected with GFP or GFP–SEC16A and untagged WT or KI ULK1 expression constructs. (F) Representative low- and high-magnification pseudocolored images of Ulk1-ko MEFs cotransfected with ULK1 and GFP–SEC16A and stained with antibodies against ULK1 and DAPI. (G) Line scans showing the degree of colocalization between ULK1 (red) and SEC16A (green) in lines (not shown) drawn within the boxes labeled 1-4 in the high magnification images from panel F. Scale bars: 10 μm (low magnification) and 5 μm (high magnification). (H) Diagram of the in vitro COPII-budding reaction. (I) Representative immunoblot analyses of budded vesicles from in vitro COPII-budding reactions using cytosolic fractions from WT or Ulk1-ko MEFs and membrane fractions from Ulk1-ko MEFs. RPN1 is an ER-resident protein; LMAN1/ERGIC53 and SEC22B are COPII cargo whose incorporation into budded vesicles is inhibited by SAR1H79G, a dominant-negative form of the COPII-specific GTPase. (J) Representative immunoblot analyses of budded vesicles from in vitro COPII-budding reactions using cytosolic fractions from 293T cells transfected with empty vector (—), WT, or KI ULK1 combined with membrane fractions from untransfected 293T cells. See also Figure S3.

Figure 4. ULKs regulate the assembly of COPII complexes at ERES

(A) Representative low- and high-magnification images of distal gonads from WT and unc-51–mutant C. elegans stained using a fluorescently labelled anti-SEC16 antibody. (B) Mean percentages ±SEM of ERES that fall within specified intensity thresholds of SEC16 staining. For each condition, at least 1000 unique ERES from C. elegans gonads were examined. Scale bars: 10 μm. *P <0.05 (Student’s t-test) when compared with WT. (C) Mean percentages (±SEM) of total SEC13 in peaks 1-3 from Superose 6 gel filtration experiments similar to the one shown in Figure S4A. *P <0.05 (Student’s t-test) when compared with WT. (D) Representative images of endogenous SEC16A immunostaining in WT, Ulk1-ko, Ulk2-ko, and Ulk1/2-dko MEFs. Scale bars: 10 μm. (E) Mean percentages (±SEM) of all ERES that fall within specified intensity thresholds of SEC16A staining. *P <0.001 and ^#P <0.05 (ANOVA) when compared with WT. (F) Mean number (±SEM) of SEC24⁺ puncta per cell in WT MEFs transfected with the indicated HA-tagged SEC24 isoform (A, B, C, or D) and control nontargeting (Ctrl), Ulk1, or Sec16a siRNA. Ten HA⁺ cells per population were scored. Red arrow highlights the decrease in SEC24C puncta number in Ulk1-depleted cells. *P <0.001 and ^#P <0.05 (ANOVA) when compared with siCtrl. (G) Mean number (±SEM) of SEC24C⁺ puncta per cell in WT, Ulk1-ko, Ulk2-ko, and Ulk1/2-dko MEFs. Ten cells per population were scored. *P <0.001 (ANOVA) when compared with WT. (H) Representative images of endogenous SEC24C staining in WT and Ulk1-ko MEFs. Scale bars: 10 μm. (I) Mean number (±SEM) of SEC24C⁺puncta per cell in Ulk1/2-deficient (i.e., Ulk1 ko and Ulk2 shRNA) MEFs stably transduced with the indicated viral vector [i.e., (—) empty vector; WT ULK; or KI ULK1 mutant]. Ten cells per population were scored. *P <0.001 (ANOVA) when compared with empty vector-transduced cells. (J) Mean number (±SEM) of SEC24C⁺ puncta per cell in WT MEFs transfected with the indicated siRNA. Ten cells per population were scored. *P <0.001 (ANOVA) when compared with Ctrl siRNA-transfected cells. See also Figure S4.

Figure 5. ULKs regulate ER-to-Golgi trafficking

(A) Representative immunoblots of Biotin IPs (top panels) and GFP IPs (bottom panels) from SERT-GFP—transfected WT and Ulk1-ko MEFs. WT MEFs were incubated in the presence or absence of Endo H or peptide-N-glycosidase F (PNGase) to establish the migration pattern of the different glycosylated forms of SERT. (B) Mean ratios (±SD) of Endo H–R SERT to total SERT in RNAi-treated samples from 2 independent experiments. *P <0.01 and ^#P <0.05 and (ANOVA) when compared with Ctrl. (C) Mean percentages (±SEM) of cells showing colocalization of AlexaFluor 647–conjugated wheat germ agglutinin (WGA) and SERT-GFP. Data were acquired from 3 independent experiments, and more than 100 cells per population were scored. *P <0.001 (ANOVA) when compared with WT. (D) Representative merged pseudocolored images of SERT-GFP—transfected WT and Ulk1-ko MEFs stained with AlexaFluor 647–conjugated wheat germ agglutinin (WGA) and DAPI. Scale bar: 10 μm. (E) Mean percentages (±SEM) of cells with colocalized WGA and SERT-GFP. Data were acquired from 3 independent experiments, and more than 100 cells per population were scored in each experiment. *P <0.002 (ANOVA) when compared with empty vector–transduced cells. (F) Mean percentages (±SEM) of siRNA transfected cells with colocalized WGA and SERT-GFP. Data were acquired from 3 independent experiments, and more than 100 cells per population were scored. *P <0.001 (ANOVA) when compared with siCtrl-transfected cells. (G) Representative images of WT, unc-51 mutant and mod-5 mutant C. elegans stained with antibodies against MOD-5/SERT and 5-HT. The arrows highlight the 5-HT or MOD-5/SERT staining of NSM processes. See also Figure S5.

Figure 6. ULK-mediated phosphorylation of SEC16A at S846 is required for assembly of SEC24C⁺ ERES and ER-to-Golgi trafficking

(A) Representative immunoblots of DDK IPs prepared from 293T cells transfected with either MYC-DDK–tagged WT SEC16A or a mutant form of MYC-DDK–tagged SEC16A harboring the S846A and the specified ULK1-expression construct. (B) Representative pseudocolored images of MEFs stained with DAPI and antibodies against MYC-DDK-tagged SEC16A and SEC24C. SA: SEC16A S846A mutant; SD: SEC16A S846D mutant. (C) Mean numbers (±SEM) of SEC16A⁺ puncta per cell. (D) Mean numbers (±SEM) of SEC24C⁺ puncta per cell . Ten cells (DDK⁺, if appropriate) per population were scored in C and D. *P <0.001 (ANOVA). (E) Representative pseudocolored images of MEFs stained with DAPI and antibodies against SEC24C and MYC-DDK-tagged SEC16A. (F) Mean numbers (±SEM) of SEC24C⁺ puncta per cell. (G) Mean numbers (±SEM) of SEC16A⁺ puncta per cell. (H) Mean percentages (±SEM) of SEC16A⁺ puncta that were also SEC24C⁺. Ten cells (DDK⁺, if appropriate) per population were scored in F-H. Scale bar = 10 μm. *P <0.001 and ^#P <0.05 when compared with WT (ANOVA). (I) Representative merged pseudocolored images of MEFs stained with AlexaFluor 647–conjugated WGA and antibodies against MYC-DDK-tagged SEC16A. White arrowheads indicate PM. (J) Mean percentages (±SEM) of cells from each population with colocalized WGA and GFP. Data were acquired from 3 independent experiments, and more than 100 cells per population were scored in each experiment. Scale bar: 10 μm. *P <0.001 (ANOVA) when compared with empty vector. See also Figure S6.

Figure 7. SEC16A mutant with the phosphomimetic S846D overcomes the defect in ER-to-Golgi trafficking and prevents activation of the UPR in ULK-deficient cells

(A) Mean percentages (±SEM) of cells with nuclear staining of CHOP. Data were acquired from 3 independent experiments, and more than 100 GFP⁺ cells per population were scored in each experiment. *P <0.001 (ANOVA) when compared with empty vector-transduced cells. (B) Representative pseudocolored images of WT and Ulk1-ko MEFs transfected with indicated GFP-tagged cargo [i.e, (−), empty vector; VSVG-GFP or SERT-GFP] stained with an antibody against CHOP and DAPI. Scale bar: 10 μm. (C) Mean percentages (±SEM) of cells with nuclear staining of CHOP. Data were acquired from 3 independent experiments, and more than 100 GFP⁺DDK⁺ cells per population were scored in each experiment. *P <0.001 (ANOVA) when compared with empty vector- transduced cells. SA: S846A SEC16A mutant; SD, S846D SEC16A mutant. (D) Representative pseudocolored images of MEFs stained with antibodies against MYC-DDK-tagged SEC16A (channel not shown) and CHOP and DAPI. Scale bar: 10 μm. See also Figure S7.