Phosphorylation of EIF2S1 (eukaryotic translation initiation factor 2 subunit alpha) is indispensable for nuclear translocation of TFEB and TFE3 during ER stress

ABSTRACT

There are diverse links between macroautophagy/autophagy pathways and unfolded protein response (UPR) pathways under endoplasmic reticulum (ER) stress conditions to restore ER homeostasis. Phosphorylation of EIF2S1/eIF2α is an important mechanism that can regulate all three UPR pathways through transcriptional and translational reprogramming to maintain cellular homeostasis and overcome cellular stresses. In this study, to investigate the roles of EIF2S1 phosphorylation in regulation of autophagy during ER stress, we used EIF2S1 phosphorylation-deficient (A/A) cells in which residue 51 was mutated from serine to alanine. A/A cells exhibited defects in several steps of autophagic processes (such as autophagosome and autolysosome formation) that are regulated by the transcriptional activities of the autophagy master transcription factors TFEB and TFE3 under ER stress conditions. EIF2S1 phosphorylation was required for nuclear translocation of TFEB and TFE3 during ER stress. In addition, EIF2AK3/PERK, PPP3/calcineurin-mediated dephosphorylation of TFEB and TFE3, and YWHA/14-3-3 dissociation were required for their nuclear translocation, but were insufficient to induce their nuclear retention during ER stress. Overexpression of the activated ATF6/ATF6α form, XBP1s, and ATF4 differentially rescued defects of TFEB and TFE3 nuclear translocation in A/A cells during ER stress. Consequently, overexpression of the activated ATF6 or TFEB form more efficiently rescued autophagic defects, although XBP1s and ATF4 also displayed an ability to restore autophagy in A/A cells during ER stress. Our results suggest that EIF2S1 phosphorylation is important for autophagy and UPR pathways, to restore ER homeostasis and reveal how EIF2S1 phosphorylation connects UPR pathways to autophagy.

Abbreviations: A/A: EIF2S1 phosphorylation-deficient; ACTB: actin beta; Ad-: adenovirus-; ATF6: activating transcription factor 6; ATZ: SERPINA1/α1-antitrypsin with an E342K (Z) mutation; Baf A1: bafilomycin A₁; BSA: bovine serum albumin; CDK4: cyclin dependent kinase 4; CDK6: cyclin dependent kinase 6; CHX: cycloheximide; CLEAR: coordinated lysosomal expression and regulation; Co-IP: coimmunoprecipitation; CTSB: cathepsin B; CTSD: cathepsin D; CTSL: cathepsin L; DAPI: 4’,6-diamidino-2-phenylindole dihydrochloride; DMEM: Dulbecco’s modified Eagle’s medium; DMSO: dimethyl sulfoxide; DTT: dithiothreitol; EBSS: Earle’s Balanced Salt Solution; EGFP: enhanced green fluorescent protein; EIF2S1/eIF2α: eukaryotic translation initiation factor 2 subunit alpha; EIF2AK3/PERK: eukaryotic translation initiation factor 2 alpha kinase 3; ER: endoplasmic reticulum; ERAD: endoplasmic reticulum-associated degradation; ERN1/IRE1α: endoplasmic reticulum to nucleus signaling 1; FBS: fetal bovine serum; gRNA: guide RNA; GSK3B/GSK3β: glycogen synthase kinase 3 beta; HA: hemagglutinin; Hep: immortalized hepatocyte; IF: immunofluorescence; IRES: internal ribosome entry site; KO: knockout; LAMP1: lysosomal associated membrane protein 1; LMB: leptomycin B; LPS: lipopolysaccharide; MAP1LC3A/B/LC3A/B: microtubule associated protein 1 light chain 3 alpha/beta; MAP1LC3B/LC3B: microtubule associated protein 1 light chain 3 beta; MEFs: mouse embryonic fibroblasts; MFI: mean fluorescence intensity; MTORC1: mechanistic target of rapamycin kinase complex 1; NES: nuclear export signal; NFE2L2/NRF2: NFE2 like bZIP transcription factor 2; OE: overexpression; PBS: phosphate-buffered saline; PLA: proximity ligation assay; PPP3/calcineurin: protein phosphatase 3; PTM: post-translational modification; SDS: sodium dodecyl sulfate; SDS-PAGE: sodium dodecyl sulfate-polyacrylamide gel electrophoresis; SEM: standard error of the mean; TEM: transmission electron microscopy; TFE3: transcription factor E3; TFEB: transcription factor EB; TFs: transcription factors; Tg: thapsigargin; Tm: tunicamycin; UPR: unfolded protein response; WB: western blot; WT: wild-type; Xbp1s: spliced Xbp1; XPO1/CRM1: exportin 1.

KEYWORDS:

• ATF6
• autophagy
• EIF2S1 phosphorylation
• ER stress
• nuclear translocation
• phosphorylation TFE3
• TFEB
• transcription factor E3
• transcription factor EB

Introduction

EIF2S1/eIF2α (eukaryotic translation initiation factor 2 subunit alpha) is a subunit of the EIF2 complex, which facilitates the placement of an initiator tRNA (methionyl-tRNAi) onto the P site of the 40S ribosomal subunit during the translation initiation of cytoplasmic mRNAs in eukaryotic cells [1]. EIF2 complex activity is regulated by EIF2S1 phosphorylation, which occurs on serine 51 (S51) and is mediated by four kinases (EIF2AK1/HRI, EIF2AK2/PKR, EIF2AK3/PERK, and EIF2AK4/GCN2) in response to diverse cellular stresses including heme deficiency, oxidative stress, viral infection, endoplasmic reticulum (ER) stress, and amino acid deficiency [2–4]. EIF2S1 phosphorylation transiently attenuates the translation of most mRNAs but promotes the translation of selected mRNAs, including the transcripts of TF (transcription factor) genes (Atf4, Ddit3/Chop, Atf5, Cebpa/C/ebp and Cebpb/C/ebp), nutrient metabolism-related genes (Slc7a1/Cat1, Slc35a4, and Eprs), a phosphatase regulatory subunit gene (Ppp1r15a/Gadd34), and cellular adaptation-related genes (Ibtk/Ibtk and Cpeb4) [5,6]. These signaling programs allow cells to conserve resources and initiate adaptive gene expression to restore cellular homeostasis, referred to as the integrated stress response [3,7,8].

Eukaryotic cells cope with ER stress by activating the unfolded protein response (UPR), which is initiated by three main UPR sensors (ERN1/IRE1 [endoplasmic reticulum to nucleus signaling 1], ATF6 [activating transcription factor 6], and EIF2AK3/PERK [eukaryotic translation initiation factor 2 alpha kinase 3]) [9,10]. ERN1 has ER stress-regulated kinase and endonuclease activities that can initiate unconventional splicing of Xbp1 mRNA to remove a 26-nucleotide intron and then introduce a translational frameshift. Spliced Xbp1 (Xbp1s) mRNA encodes a potent TF (XBP1s) that induces transcription of genes encoding proteins that facilitate protein folding, secretion, and degradation in response to ER stress [9,10]. ATF6 is a TF encoded by two related genes, Atf6 and Atf6b [11]. Upon ER stress, it translocates to the Golgi apparatus, where it is cleaved by site-1 protease and site-2 protease. The cleaved N-terminal cytosolic domain of ATF6 (hereafter referred to as “the activated ATF6 form”) then translocates into the nucleus, where it binds to ER stress-response elements and thereby activates target genes that encode proteins with functions in ER protein folding, endoplasmic reticulum-associated degradation (ERAD), protein secretion, and ER biogenesis [11,12]. EIF2AK3 is the major protein responsible for attenuation of mRNA translation via EIF2S1 phosphorylation, reducing the protein burden within the ER. Paradoxically, EIF2AK3-mediated EIF2S1 phosphorylation upregulates the translation of several mRNAs as described above. Among them, translation of Atf4 mRNA is crucial for upregulation of genes involved in redox homeostasis, amino acid metabolism, protein folding, and protein synthesis [5,9]. In addition, crosstalk can occur between the EIF2AK3-EIF2S1 phosphorylation-ATF4 pathway and other UPR pathways (ERN1-XBP1- and ATF6-mediated UPR pathways). Phosphorylation of EIF2S1 is required for maximal induction of XBP1s protein by stabilizing its mRNA [13], and for activation of ATF6 by facilitating its trafficking from the ER to the Golgi in response to ER stress [14]. Thus, phosphorylation of EIF2S1 affecting activation of all three UPR pathways is responsible for transcriptional as well as translational reprogramming to help cells maintain cellular homeostasis and overcome cellular stresses.

Macroautophagy (hereafter referred to as “autophagy”) is an evolutionarily conserved cellular process by which accumulating aberrant proteins or damaged subcellular organelles undergo lysosomal degradation [15,16]. In brief, autophagy includes five steps: initiation and phagophore nucleation, phagophore expansion, autophagosome maturation, autophagosome-lysosome fusion, and cargo degradation by lysosomal enzymes [17,18]. Numerous genes are required to perform these processes [19–21]. Increasing evidence indicates that autophagy is regulated at the transcriptional level by several TFs, including TFEB (transcription factor EB), TFE3 (transcription factor E3), FOXO (forkhead box O), and E2F1 (E2 transcription factor 1) [22–24]. TFEB is a member of the microphthalmia-associated TF family, which also includes MITF, TFE3, and TFEC [25]. TFEB and TFE3 are believed to be the master regulators of the autolysosome pathway, and to control expression of genes required for autophagosome formation, lysosome biogenesis, and lysosome function by directly binding to promoters of the coordinated lysosomal expression and regulation (CLEAR) element [20,26,27]. Diverse PTMs (post-translational modifications), including phosphorylation, regulate the activities of these TFs [28–32]. Several kinases that phosphorylate TFEB and TFE3 have been identified. Among them, MTOR (mechanistic target of rapamycin kinase) complex 1 (MTORC1) is the best-studied [33–36]. Under normal conditions, lysosomal MTORC1 phosphorylates TFEB (at S142 and S211) and TFE3 (at S321). Phosphorylated TFEB and TFE3 interact with YWHA/14-3-3, which results in sequestration of these TFs as inactive forms in the cytosol [35–38]. Under starvation and other conditions when MTORC1 is inhibited and/or the Ca²⁺-CALM (calmodulin)-dependent protein phosphatase PPP3/calcineurin is activated, further phosphorylation of TFEB and TFE3 does not occur and their dephosphorylation is induced. This prevents binding to YWHA/14-3-3 and induces rapid accumulation of TFEB and TFE3 in the nucleus [31,39,40]. However, recent reports suggest that nuclear translocation of TFEB and TFE3 is more complex than generally appreciated. For example, TFEB continuously shuttles between the cytosol and nucleus via nuclear export dependent on the XPO1/CRM1 (exportin 1) [29,30,41]. XPO1 is an export receptor for leucine-rich nuclear export signals (NESs) [29,30,42]. Therefore, whether TFEB is retained in the nucleus has been proposed to depend on the phosphorylation statuses of S142 and S138, which are localized in the proximity of a NES. Nuclear export is promoted by phosphorylation of S142 via MTORC1 and MAPK1/ERK2 (mitogen-activated protein kinase 1), under nutrient-rich conditions [29,30], or CDK4 (cyclin dependent kinase 4) and CDK6, during G₁ phase [41]. Moreover, phosphorylation of S142 primes TFEB for phosphorylation of S138 by GSK3B/GSK3β (glycogen synthase kinase 3 beta) [30]. Therefore, the absence of S142 phosphorylation may lead to nuclear retention of TFEB and TFE3. Thus, the mechanisms governing the localization of TFEB and TFE3 in response to multiple signals are not fully understood.

Several studies have shown that the UPR induces autophagy to degrade unfolded and aggregated proteins and thereby restore ER homeostasis [43–46], although excessive and prolonged ER stress may inhibit autophagy by impairing lysosomes [47]. Several components of UPR signaling pathways transcriptionally upregulate genes encoding autophagy machinery, indicating a high level of crosstalk between ER stress and autophagy [43–46]. Several lines of evidence suggest that EIF2S1 phosphorylation plays a key role in regulation of autophagy. EIF2AK3-EIF2S1 phosphorylation is reportedly involved in polyglutamine 72 repeat aggregate-induced autophagy [48]. A nonphosphorylatable knock-in mutation (S51A) of EIF2S1 and dominant-negative EIF2AK3 inhibit polyglutamine 72 repeat-induced MAP1LC3/LC3 conversion. Induction of autophagic puncta by diverse pharmacological autophagy enhancers is also partially inhibited in homozygous EIF2S1^S51A knock-in (A/A) mutant human osteosarcoma U2OS cells [49]. Furthermore, the TFs ATF4 and DDIT3/CHOP, which are downstream targets of phosphorylated EIF2S1 are reportedly required to increase transcription of a set of autophagy genes (autophagosome formation, elongation, and function) under amino acid starvation or ER stress conditions [50]. Thus, EIF2S1 phosphorylation may play a central role in autophagy in response to ER stress. Nevertheless, the molecular mechanisms involved in activation and regulation of autophagy through EIF2S1 phosphorylation remain unclear.

In the present study, we revealed that EIF2S1 phosphorylation plays an essential role in nuclear translocation of TFEB and TFE3. Dephosphorylation (at both S211 and S142) of TFEB and its dissociation from YWHA were insufficient for its nuclear translocation in EIF2S1 phosphorylation-deficient (A/A) cells during ER stress. Instead, overexpression (OE) of the activated ATF6 form, XBP1s, and ATF4, production of which was significantly reduced and delayed in A/A cells during ER stress, differentially rescued defects of TFEB and TFE3 nuclear translocation in A/A cells during ER stress. Consequently, OE of the activated ATF6 or TFEB form more efficiently restored autophagy, although XBP1s and ATF4 also displayed an ability to restore autophagy in A/A cells during ER stress. Our data highlight a new mechanism controlling the subcellular localization and activity of TFEB via EIF2S1 phosphorylation-dependent components of UPR signaling pathways under ER stress conditions.

Results

Deficiency of EIF2S1 phosphorylation dysregulates expression of autophagy and UPR genes during ER stress

We investigated whether EIF2S1 phosphorylation contributes to expression of not only UPR gene but also macroautophagy/autophagy genes during ER stress. Wild-type (WT) and EIF2S1 phosphorylation-deficient immortalized mouse embryonic hepatocytes (S/S^Hep and A/A^Hep, respectively) [51] were treated with the ER stress inducer tunicamycin (Tm) for the indicated durations. Expression levels of proteins and mRNA transcripts of UPR and autophagy genes in A/A^Hep cells were compared with those in S/S^Hep cells. As reported previously [13,51–53], under ER stress conditions, phosphorylated forms of the UPR sensors EIF2AK3 and ERN1 were immediately observed, and their phosphorylation persisted until 24 h in both S/S^Hep and A/A^Hep cells treated with Tm (Figure 1A left panels). By contrast, cleavage of the other UPR sensor ATF6 (ATF6 and ATF6 was diminished and delayed in A/A^Hep cells compared with S/S^Hep cells (Figure 1A right panels and Fig. S1A, B) as reported previously [14], indicating that the initiation mechanism of UPR pathways is partly impaired in A/A^Hep cells. Furthermore, as shown in several reports, the expression levels of proteins (ATF4 and DDIT3) and mRNAs (Atf4, Ddit3, Ppp1r15a, Asns, and Cth) encoded by EIF2AK3 pathway genes [7,53], a protein (XBP1s) and mRNA (Xbp1s) encoded by a ERN1 pathway gene [13], and proteins (HSP90B1/GRP94 and HSPA5/BiP) and mRNA (Hspa5/BiP) encoded by ATF6 downstream genes [14,54] were significantly reduced in A/A^Hep cells under Tm-induced ER stress conditions (Figure 1A, B). Thus, we showed that EIF2S1 phosphorylation is required for cleavage-mediated activation of the UPR sensor ATF6 and expression of multiple genes in all three UPR pathways.

Figure 1. Protein and mRNA expression of autophagy and UPR genes is dysregulated in A/A cells during ER stress. (A) WB analysis of UPR proteins in lysates of S/S^Hep and A/A^Hep cells treated with Tm (1 µg/mL) for the indicated durations. ATF6(F): full-length glycosylated ATF6; ATF6(F*): full-length unglycosylated ATF6; ATF6(N): cleaved N-terminal fragment of ATF6; ATF6B(F): full-length glycosylated ATF6B; ATF6B(F*): full-length unglycosylated ATF6B; ATF6B(N*) and ATF6B(N): cleaved N-terminal fragments of ATF6B. The identities of bands indicated in the ATF6 and ATF6B WB analysis were validated by WB analysis (Fig. S1) of atf6 KO and atf6 atf6b double KO cell lines using the same ATF6- and ATF6B-specific antibodies. (B and C) Quantitative RT-PCR analysis of mRNA expression of ER stress response (B) and autophagy (C) genes in S/S^Hep and A/A^Hep cells treated with Tm (1 µg/mL) for the indicated durations. Data are presented as mean ± SEM of three independent experiments (two-way ANOVA with Sidak’s post hoc test) (D) WB analysis of autophagy proteins in lysates of S/S^Hep and A/A^Hep cells treated with Tm (1 µg/mL) for the indicated durations. CTSB: cathepsin B; CTSL: cathepsin L; Pro: procathepsin; Sc: mature single-chain cathepsin; Dc: heavy chain of mature double-chain cathepsin.

In WT (S/S^Hep) cells, Tm treatment gradually increased the mRNA levels of most examined autophagy genes and the levels of some autophagosome proteins (LC3B-II and SQSTM1/p62), whereas the levels of lysosomal proteins (LAMP1 and 2, and cathepsin B and L) were decreased at late time points (12, 16, and 24 h) of Tm treatment (Figure 1C, D). The mRNA and protein levels of most examined autophagy genes, except for LC3B-I/II proteins, were lower in A/A^Hep cells than in S/S^Hep cells at most time points (Figure 1C, D). Although the LC3B-I/II protein levels were higher in A/A^Hep cells than in S/S^Hep cells at all-time points, LC3B conversion (LC3B-II:I ratio) was lower in A/A^Hep cells than in S/S^Hep cells at most time points (Figure 1D), suggesting that EIF2S1 phosphorylation plays an important role in autophagy pathways. Thus, deficiency of EIF2S1 phosphorylation dysregulates expression of not only UPR genes but also autophagy genes during ER stress.

Autophagy is defective in A/A cells during ER stress

We further investigated whether deficiency of EIF2S1 phosphorylation affects autophagy resulting from Tm-induced ER stress. Formation of autophagosomes and autolysosomes was analyzed in immortalized mouse embryonic hepatocytes treated with Tm (Figure 2). In WT (S/S^Hep) cells, Tm treatment strongly increased the number of MAP1LC3A/B/LC3A/B-positive puncta (Figure 2A, C, E left lower panels, and Figure 2B first row second panel), suggesting that ER stress-mediated autophagy induction occurs as previously reported [55–57]. However, there were few prominent LC3A/B-positive puncta in Tm-treated A/A^Hep cells (Figure 2A graph and Figure 2A, C, E right lower panels, and Figure 2B first row fourth panel). In addition, most LC3A/B-positive structures were smaller in Tm-treated A/A^Hep cells than in Tm-treated S/S^Hep cells (Figure 2A graph and Figure 2A, C, E lower panels, and Figure 2B first row). Immunofluorescence (IF) signals of LC3A/B were largely concentrated in the perinuclear regions of Tm-treated A/A^Hep cells (Figure 2A graph and Figure 2A, C, E right lower panels, and Figure 2B first row fourth panel). Furthermore, these signals colocalized with IF signals of the KDEL-motif-containing ER proteins (HSPA5 and HSP90B1), suggesting that LC3A/B is mislocalized in the ER membrane due to deficiency of EIF2S1 phosphorylation during ER stress (Figure 2B). Next, we observed the subcellular colocalizations of LC3A/B (an autophagosome marker) and the cargo receptor SQSTM1 (a cargo marker) to investigate formation of autophagosomes (Figure 2C). In addition, we observed colocalization of LAMP1 (a lysosome marker) with SQSTM1 (Figure 2D) and LC3A/B (Figure 2E) to investigate formation of autolysosomes. Colocalization of LC3A/B with SQSTM1 was significantly increased in Tm-treated S/S^Hep cells (Figure 2C left lower panels and Figure 2F left graph), indicating that Tm treatment increases autophagosome formation in S/S^Hep cells. Furthermore, an IF assay confirmed the colocalization of LAMP1 with SQSTM1 (Figure 2D left lower panels and Figure 2F middle graph) and LC3A/B (Figure 2E left lower panels and Figure 2F right graph), suggesting that Tm treatment increases autophagosome-lysosome fusion in S/S^Hep cells. However, colocalization of LC3A/B with SQSTM1 was significantly lower in A/A^Hep cells than in S/S^Hep cells under both normal and ER stress conditions (Figure 2C lower panels and Figure 2F left graph). There were few SQSTM1- and LAMP1-positive puncta in Tm-treated A/A^Hep cells compared with Tm-treated S/S^Hep cells (Figure 2C-E). In addition, colocalization of LAMP1 with SQSTM1 (Figure 2D lower panels and Figure 2F left graph) and LC3A/B (Figure 2E lower panels and Figure 2F right graph) was significantly lower in Tm-treated A/A^Hep cells than in Tm-treated S/S^Hep cells, although Tm treatment slightly increased colocalization of LAMP1 with LC3A/B in A/A^Hep cells (Figure 2F right graph). These results indicate that EIF2S1 phosphorylation is required for not only autophagosome formation, but also autolysosome formation during ER stress.

Figure 2. Autophagy is impaired in A/A cells during ER stress. (A) Representative IF images of LC3A/B in S/S^Hep and A/A^Hep cells treated with DMSO (vehicle) or Tm (1 µg/mL) for 24 h. The dotted white line defines the cell boundary. Scale bar: 20 µm. The graph depicts the fraction (%) of cells with different LC3A/B staining patterns (the “punctate” group represents cells with LC3A/B-positive puncta only, the “punctate + accumulated” group represents cells with both LC3A/B-positive puncta and condensed LC3A/B staining in the perinuclear region, the “accumulated” group represents cells with condensed LC3A/B staining only in the perinuclear region, and “no punctate + no accumulated” represents cells with neither LC3A/B-positive puncta nor condensed LC3A/B staining in the perinuclear region). Data are presented as mean ± SEM of three independent experiments (at least 150 cells per condition). (B) Representative IF images of an autophagosome marker (LC3A/B, green) and an ER marker (KDEL, red) in S/S^Hep and A/A^Hep cells treated with DMSO or Tm (1 µg/mL) for 24 h. Nuclei were stained with DAPI (blue). Scale bar: 20 µm. (C–E) Representative IF images of an autophagy marker (LC3A/B) or a cargo marker (SQSTM1) and a lysosome marker (LAMP1) in S/S^Hep and A/A^Hep cells treated with DMSO or Tm (1 µg/mL) for 24 h. Cells were fixed and costained with anti-LC3A/B (green) and anti-SQSTM1 (red) antibodies in (C), anti-SQSTM1 (green) and anti-LAMP1 (red) antibodies in (D), and anti-LC3A/B (green) and anti-LAMP1 (red) antibodies in (E). Nuclei were stained with DAPI (blue). The right panels are magnified images of the boxes in the left panels. Scale bars: left panels (20 µm) and right panels (10 µm). (F) Quantification of the colocalization of LC3A/B with SQSTM1 in (C) and LAMP1 with SQSTM1 or LC3A/B in (D and E). (G) Representative LysoTracker staining images of S/S^Hep and A/A^Hep cells. Cells were treated with Tm (1 µg/mL) for the indicated durations and stained with LysoTracker (100 nM, red) and Hoechst 33,258 (10 μg/mL, blue) for the last 30 min of the treatment. The dotted white line defines the cell boundary. Scale bar: 20 µm. (H) Quantification of the mean fluorescence intensity (MFI) of LysoTracker in (G). Data in the graphs in (F) and (H) are presented as mean ± SEM of three independent experiments (at least 150 cells per condition). A two-way ANOVA with Sidak’s post hoc test was used in the graphs in (F) and (H).

Defective lysosomal function contributes to dysregulation of autophagy [19,58]. We next examined lysosomal dysfunctions in A/A^Hep cells using the pH-sensitive dye LysoTracker Red, which specifically labels acidic vesicles such as functional lysosomes and autolysosomes. Consistent with a defect in autolysosome formation under ER stress conditions (Figure 2D–F), the number of LysoTracker-positive structures (Figure 2G) and fluorescence intensity of LysoTracker (Figure 2H) were decreased much more in A/A^Hep cells than in S/S^Hep cells upon Tm treatment, indicating that EIF2S1 phosphorylation is required to maintain functional lysosomes and autolysosomes during ER stress. Upon autophagy induction, lysosomes amass in the perinuclear region, and this increases autophagosome-lysosome fusion rates, whereas dispersion of lysosomes to the cell periphery reduces fusion rates [59–61]. A/A^Hep cells exhibited peripherally accumulated lysosomes, whereas substantial numbers of lysosomes were predominantly found in the perinuclear region of S/S^Hep cells after treatment with Tm for 16 and 24 h as expected (Figure 2G). However, an IF assay of the lysosome marker LAMP1 revealed that LAMP1-positive lysosomal vesicles did not accumulate peripherally but were found everywhere in Tm-treated A/A^Hep cells (Figure 2D, E), indicating that only peripheral LAMP1-positive vesicles are acidic and functional in these cells. Therefore, the autophagosome-lysosome fusion rates will be decreased in Tm-treated A/A^Hep cells. These results suggest that EIF2S1 phosphorylation is important to maintain the activity and subcellular localization of lysosomes, which can affect formation of autolysosomes under ER stress conditions.

Autophagic flux and autophagic degradation of misfolded proteins are impaired in A/A cells

To fortify the association between EIF2S1 phosphorylation and autophagy in Tm-treated cells, we performed transmission electron microscopy (TEM) analysis of S/S^Hep and A/A^Hep cells treated with and without Tm. Autolysosomes accumulated in Tm-treated S/S^Hep cells, but not in Tm-treated A/A^Hep cells, while autophagosomes were hardly detected in S/S^Hep and A/A^Hep cells treated with and without Tm (Figure 3A second row and Figure 3B). These results confirmed that EIF2S1 phosphorylation deficiency inhibits the formation of autolysosomes under ER stress conditions. In addition, although both Tm-treated S/S^Hep and A/A^Hep cells had swollen and fragmented ER as previously reported [62,63], Tm-treated A/A^Hep cells had more excessively fragmented ER than Tm-treated S/S^Hep cells. In addition, the fragmented ER tubules were highly accumulated and looked like a compact mass in Tm-treated A/A^Hep cells (Figure 3A third row and yellow dotted area in the Tm-treated A/A^Hep panel of Figure 3A first row), suggesting that deficiency of EIF2S1 phosphorylation also alters the ER structure during ER stress.

Figure 3. Autophagic flux is impaired in A/A cells during ER stress. (A) Representative TEM images of S/S^Hep and A/A^Hep cells treated with DMSO or Tm (1 µg/mL) for 24 h. The panels of the second (red) and third (yellow) rows are magnified images of the red and yellow boxes in the panels of the first row, respectively. Green arrowheads indicate autophagosomes, red arrowheads indicate autolysosomes, and yellow arrows indicate the ER. The dotted yellow line defines a mass of dilated and fragmented ER structures. Scale bars: first row 2 µm and second and third rows 0.5 µm. (B) Quantification of the number of autolysosomes per cell in the TEM images in (A). Data are presented as mean ± SEM of three independent experiments (at least 15 cells per condition). (C) WB analysis of LC3B in protein lysates of S/S^Hep and A/A^Hep cells. Cells were treated with DMSO or Tm (1 µg/mL) for 16 h in the absence or presence of the lysosomal inhibitor Baf A1 (200 nM) for 3 h before harvest. The graph depicts the LC3B-II level normalized to the ACTB level. Data are presented as mean ± SEM of three independent experiments. (D and E) WB analysis of SERPINA1/alpha-1-antitrypsin mutant Z (α1-AT [ATZ]) in protein lysates of S/S^Hep and A/A^Hep cells. Cells were transfected with the pcDNA3.1-α1-AT [ATZ] plasmid for 24 h. Transfected cells were treated with DMSO, the proteasome inhibitor MG132 only (20 µM) (C), the lysosomal inhibitor Baf A1 only (100 nM) (D), MG132 plus the translation inhibitor CHX (100 µg/mL) (C), or Baf A1 plus CHX (D) for the indicated durations. The graphs depict the ATZ level normalized to the ACTB level after treatment for 6 h. Data are presented as mean ± SEM of three independent experiments. (F) WB analysis of SQSTM1, an endogenous cargo of autophagy in protein lysates of S/S^Hep and A/A^Hep cells. S/S^Hep and A/A^Hep cells were treated with DMSO or Tm and then with MG132 (20 µM) only or MG132 plus CHX for the indicated durations before harvesting samples. The graphs depict the SQSTM1 level normalized to the ACTB level after treatment for 6 h. Data are presented as mean ± SEM of three independent experiments. A two-way ANOVA with Sidak’s post hoc test was used in (B)-(F).

Both colocalization analysis of LAMP1 and SQSTM1 (Figure 2D) and LAMP1 and LC3A/B (Figure 2E) and TEM observation of autophagic vesicles (Figure 3A, B) indicated that autophagic flux is impaired in A/A^Hep cells under ER stress conditions. To explore the impairment of autophagic flux in Tm-treated A/A^Hep cells, we investigated LC3B-II accumulation in Tm-treated cells incubated with bafilomycin A₁ (Baf A1), a specific inhibitor of vacuolar H⁺-ATPases and a blocker of autophagosome-lysosome fusion [64,65]. During active autophagic flux, LC3B-II protein accumulates upon Baf A1 treatment [66]. In the absence of Tm treatment, Baf A1 increased the level of LC3-II protein as expected (Figure 3C), indicating that autophagic flux is active in both S/S^Hep and A/A^Hep cells under normal conditions. However, Baf A1 failed to induce LC3B-II accumulation in Tm-treated A/A^Hep cells, but still increased the LC3B-II protein level in Tm-treated S/S^Hep cells (Figure 3C), indicating that autophagic flux is impaired in A/A^Hep but not in S/S^Hep cells under ER stress conditions.

A variant of SERPINA1/α1-antitrypsin with the E342K (Z) mutation (ATZ) is degraded by both autophagy and proteasome-dependent ERAD [67,68]. Autophagy pathways are defective in EIF2S1 phosphorylation-deficient cells; therefore, we investigated whether EIF2S1 phosphorylation deficiency affects autophagic degradation of ATZ protein. ATZ-expressing cells were treated with the proteasome inhibitor MG132 alone to inhibit proteasome-mediated degradation, or cotreated with MG132 and cycloheximide (CHX) to inhibit both proteasome-mediated degradation and de novo protein synthesis for 3 or 6 h (Figure 3D). Therefore, cotreatment with MG132 and CHX will predominantly allow autophagic degradation of ATZ protein. Western blot (WB) analysis revealed that cotreatment with MG132 and CHX for 6 h significantly increased autophagic degradation of ATZ in S/S^Hep cells, but this degradation was decreased in A/A^Hep cells (Figure 3D). Next, ATZ-expressing cells were treated with Baf A1 alone to inhibit autophagic degradation, or cotreated with Baf A1 and CHX to inhibit both autophagic degradation and de novo protein synthesis for 6 h (Figure 3E). Therefore, cotreatment with Baf A1 and CHX will allow proteasome-mediated degradation of ATZ proteins. Proteasome-mediated ATZ degradation was not impaired but improved in A/A^Hep cells compared with S/S^Hep cells upon cotreatment with Baf A1 and CHX (Figure 3E). Furthermore, degradation of SQSTM1, an endogenous cargo of autophagy [15,16], was also impaired in A/A^Hep cells under ER stress conditions (Figure 3F). Thus, our results suggest that phosphorylation of EIF2S1 is important to maintain autophagic flux (such as autophagosome and autolysosome formation), which can affect degradation of its target substrates.

Nuclear translocation of TFEB and TFE3 is impaired in A/A cells during ER stress

Most genes examined in Figure 1C, which displayed lower mRNA levels in A/A^Hep cells than in S/S^Hep cells during ER stress, are downstream targets of TFEB and TFE3, the master transcriptional regulators of autophagy and lysosome biogenesis [20,26,27,69]. TFEB and TFE3 reportedly regulate expression of their target genes by binding to the CLEAR motif sequence [20,26,27]. To determine whether EIF2S1 phosphorylation deficiency influences CLEAR promoter element activity during ER stress, S/S and A/A mouse embryonic fibroblasts (MEFs) were transfected with a 5XCLEAR luciferase reporter construct (containing five tandem copies of a CLEAR promoter element). Changes of luciferase activities were investigated in S/S^MEF and A/A^MEF cells treated with Tm and Earle’s Balanced Salt Solution (EBSS) (Figure 4A). ER stress-induced TFEB activity was abolished in A/A^MEF cells, whereas Tm treatment significantly stimulated luciferase activity of the transfected reporter construct in S/S^MEF cells (Figure 4A upper graph). Furthermore, starvation induced by EBSS treatment only modestly increased CLEAR promoter activity in A/A^MEF cells, but substantially induced reporter activity in S/S^MEF cells (Figure 4A lower graph), indicating that EIF2S1 phosphorylation is required for expression of autophagy genes induced by TFEB and TFE3 activation. These results (and those presented in Figure 1C, D) demonstrate that EIF2S1 phosphorylation plays a novel and important role in transcriptional regulation of autophagy genes during ER stress.

Figure 4. Nuclear translocation of TFEB and TFE3 is impaired in A/A cells during ER stress. (A) Luciferase activity assay of the 5xCLEAR luciferase reporter. S/S^MEF and A/A^MEF cells were cotransfected with plasmids expressing 5XCLEAR-driven firefly luciferase and CMV-driven Renilla luciferase for 30 h. Cells were treated with Tm (100 ng/mL) for the indicated durations or starved with EBSS for 12 h, and then luciferase activities were measured. Data are presented as mean ± SEM of three independent experiments. (B) Representative IF images of TFEB (upper) and TFE3 (lower) in S/S^Hep and A/A^Hep cells treated with Tm (1 µg/mL) for the indicated durations. Scale bar: 20 µm. The percentage of cells with nuclear localized TFEB or TFE3 is indicated in each image and shown in the graphs. Data are presented as mean ± SEM of three independent experiments (about 140–500 cells per condition). (C) WB analysis of the subcellular distributions of TFEB and TFE3 in S/S^Hep and A/A^Hep cells treated without or with Tm (1 µg/mL) for 24 h. LMNA (lamin A/C) and ACTB were used as loading controls of the nuclear and cytoplasmic fractions, respectively. (D) Densitometric quantification of nuclear TFEB and TFE3 in (C). Values were normalized against LMNA levels. Data are presented as mean ± SEM of three independent experiments. (E and F) Representative IF images of TFEB (E) and TFE3 (F) in WT HeLa cells (HeLa-WT), HA-Cas9- and FLAG-EIF2S1^S51A-expressing HeLa cells (HeLa-Cas9 EIF2S1^S51A OE), and HA-Cas9- and FLAG-EIF2S1^S51A-expressing and EIF2S1-KO HeLa cells (HeLa-Cas9 EIF2S1^S51A OE EIF2S1 KO). Cells were treated with DMSO or Tm (2 µg/mL) for 24 h. Scale bar: 20 µm. The graphs depict the percentage of cells with nuclear TFEB or TFE3. Data are presented as mean ± SEM of three independent experiments (at least 150 cells per condition). (G) WB analysis of the subcellular distributions of TFEB and TFE3 in HeLa-WT, HeLa-Cas9 EIF2S1^S51A OE, and HeLa-Cas9 EIF2S1^S51A OE EIF2S1 KO cells treated with DMSO or Tm (2 µg/mL) for 24 h. Histone H3 and ACTB were used as loading controls of the nuclear and cytoplasmic fractions, respectively. (H) Densitometric quantification of nuclear TFEB and TFE3 in (G). Values were normalized against Histone H3 levels. Data are presented as mean ± SEM of three independent experiments. A two-way ANOVA with Sidak’s post hoc test was used in (A)-(H).

A/A cells had several defects in autophagy pathways, including in autophagic flux and autophagy gene expression in response to Tm treatment; therefore, we postulated that TFEB and TFE3 may be inactive in Tm-treated A/A cells. To investigate this, we first examined the subcellular distributions of endogenous TFEB and TFE3 in S/S^Hep and A/A^Hep cells treated with Tm. Nuclear translocation of TFEB and TFE3 was observed at 6 h, gradually increased, and reached almost 100% at 24 h in Tm-treated S/S^Hep cells (Figure 4B), whereas very little (<5%) nuclear translocation of TFEB and TFE3 was observed in Tm-treated A/A^Hep cells (Figure 4B). To verify our results, we performed subcellular fractionation analysis of S/S^Hep and A/A^Hep cells treated with and without Tm. As reported previously [37], Tm treatment potently induced accumulation of endogenous TFEB and TFE3 in the nuclear fraction of S/S^Hep cells (Figure 4C, D). However, levels of TFEB and TFE3 were significantly lower in the nuclear fraction of Tm-treated A/A^Hep cells than in that of Tm-treated S/S^Hep cells (Figure 4C, D). These data indicate that EIF2S1 phosphorylation is necessary for nuclear accumulation of TFEB and TFE3 in response to ER stress.

Furthermore, we conducted subcellular localization experiments using multiple cell lines to confirm that defective nuclear translocation of TFEB and TFE3 is not limited to particular EIF2S1 phosphorylation-deficient cell types. First, similar to A/A^Hep cells, A/A^MEF cells displayed defective nuclear translocation of TFEB and TFE3 in response to Tm treatment (Fig. S2A–C). In addition, as previously reported [37], nuclear translocation of TFEB and TFE3 was severely impaired in eif2ak3-knockout (KO) (eif2ak3^−/−) MEFs in response to Tm treatment, whereas these proteins substantially translocated from the cytosol to the nucleus in WT (Eif2ak3^+/+) MEFs treated with Tm for 16 h (Fig. S2A–C). Second, defective nuclear translocation of TFEB and TFE3 in response to Tm treatment was completely restored by human WT EIF2S1 OE in A/A^MEF cells (Fig. S2D–F). Third, to conveniently analyze changes in the cellular localization of TFEB under diverse experimental conditions, we established S/S^MEF and A/A^MEF cell lines expressing human TFEB fused with enhanced green fluorescent protein (EGFP) at the C-terminus and control S/S^MEF and A/A^MEF cell lines expressing EGFP only (Fig. S2G) [70]. Stable highly expressing clones (S/S-TFEB-EGFP clone 5 and A/A-TFEB-EGFP clone 5) were chosen by fluorescence microscopy and WB analyses (Fig. S2G, H). Under normal conditions, TFEB-EGFP expressed in S/S-TFEB-EGFP and A/A-TFEB-EGFP MEFs mainly localized to the cytoplasm (Fig. S2H, I). Similar to A/A^Hep and A/A^MEF cells, after Tm treatment for 16 h, nuclear localized TFEB-EGFP was observed in only a very small percentage (<5%) of A/A-TFEB-EGFP MEFs, whereas almost 80% of S/S-TFEB-EGFP MEFs displayed nuclear localized TFEB-EGFP (Fig. S2H, I). However, the nuclear translocation defect of TFEB-EGFP in response to Tm treatment was efficiently corrected by recombinant adenovirus-mediated OE of WT EIF2S1, but not of mutant EIF2S1^S51A, in A/A-TFEB-EGFP MEFs (Fig. S2J–L). Finally, we examined nuclear translocation of TFEB and TFE3 in an EIF2S1 phosphorylation-deficient human cell line. To establish a HeLa cell line that lacks phosphorylation of EIF2S1 residue S51, a HeLa cell line expressing both the HA-tagged Cas9 (CRISPR-associated protein 9) restriction enzyme and the FLAG-tagged human EIF2S1^S51A mutant (HeLa-Cas9 EIF2S1^S51A OE) was first generated. Then, the HeLa-Cas9 EIF2S1^S51A OE EIF2S1 KO cell line, in which the endogenous WT EIF2S1 gene was knocked out, was established from HeLa-Cas9 EIF2S1^S51A cells using CRISPR-Cas9 technology. Similar observations were made in HeLa-Cas9 EIF2S1^S51A OE EIF2S1 KO cells, which were engineered to express FLAG-tagged EIF2S1^S51A and lacked endogenous WT EIF2S1 (Figure 4E–H). HeLa-Cas9 EIF2S1^S51A OE EIF2S1 KO cells displayed significant nuclear translocation impairment (Figure 4E, F) and diminished nuclear accumulation (Figure 4G, H) of TFEB and TFE3 during ER stress. These data indicate that impairment of TFEB and TFE3 nuclear translocation induced by EIF2S1 phosphorylation deficiency is not a species- or cell type-specific event during ER stress.

In addition to Tm treatment, other conditions induce autophagy. These include treatment with other ER stress inducers such as thapsigargin (Tg) and dithiothreitol (DTT) [56,71], as well as diverse cellular stress conditions such as MTOR inhibition (Torin 2), nutrient starvation (EBSS), and inflammation (lipopolysaccharide [LPS]) [31]. We examined nuclear translocation of endogenous TFEB and TFE3 in S/S^Hep and A/A^Hep cells in response to these autophagic stimuli (Fig. S3A–C). As expected, all stimuli induced nuclear localization of TFEB and TFE3 in S/S^Hep cells with different sensitivities (Fig. S3A, B). They immediately induced EIF2S1 phosphorylation in S/S^Hep cells but not in A/A^Hep cells (Fig. S3C). By contrast, their nuclear translocation was almost completely abolished in A/A^Hep cells in response to all five stimuli tested (Fig. S3A–C). To confirm these results, we analyzed the nuclear localization of TFEB in S/S-TFEB-EGFP and A/A-TFEB-EGFP MEFs treated with all five stimuli tested, other MTOR inhibitors (Torin 1 and PP242), and Torin 2 plus Tm. In agreement with the hepatocyte data, all stimuli including Torin 2 plus Tm (data not shown) induced nuclear translocation of TFEB-EGFP in S/S-TFEB-EGFP MEFs but not in A/A-TFEB-EGFP MEFs (Fig. S3D, E). In addition, the MTORC1 inhibitory activity of Torin 2 and Torin 1 was sustained in both genotype for the entire experimental durations (16 h) (Figure 3F). Thus, our data suggest that EIF2S1 phosphorylation plays a novel and crucial role in regulating nuclear translocation of TFEB and TFE3 in response to diverse cellular stresses including ER stress.

Collectively, these observations raise the question of whether EIF2S1 phosphorylation influences nuclear translocation of multiple proteins under ER stress conditions. Therefore, we examined the nuclear localization of GSK3B, which translocates into the nucleus under ER stress conditions [72,73], in S/S-TFEB-EGFP and A/A-TFEB-EGFP MEFs under ER stress conditions. As already shown in Figure. S2H–K and S3D, E, nuclear translocation of TFEB-EGFP was markedly inhibited in A/A-TFEB-EGFP MEFs under ER stress conditions, whereas nuclear translocation of GSK3B was not inhibited in TFEB-EGFP-expressing S/S or A/A MEFs (Figure. S3G, H). These results indicate that impairment of nuclear translocation by EIF2S1 phosphorylation deficiency is a specific defect that only affects TFEB, TFE3, and a few related proteins under ER stress conditions.

EIF2S1 phosphorylation deficiency does not impair YWHA-mediated regulation of TFEB and TFE3 nuclear translocation

Tm treatment induces nuclear translocation of TFEB and TFE3 via a process that is dependent on calcium-activated PPP3 [37], which can emasculate YWHA-mediated retention of these TFs in the cytosol [33,35]. Furthermore, EIF2AK3 is thought to be necessary for PPP3 activation in response to Tm treatment because it can modulate calcium levels in the ER and cytoplasm [74–76]. By using the PPP3 inhibitor FK506 in our experimental systems, we recapitulated that PPP3-mediated dephosphorylation of TFEB is important for ER stress-induced nuclear translocation of TFEB in Tm-treated S/S-TFEB-EGFP MEFs (Figure 5A, B). In addition, the results presented in Figure S2A, B and Figure 5A, B indicate that activation of the EIF2AK3-Ca²⁺-PPP3 pathway determines nuclear translocation of TFEB and TFE3 in response to Tm treatment. However, EIF2AK3 activation was not dysregulated in A/A cells (Figure 1A). Therefore, we next investigated whether cytosolic Ca²⁺ mobilization is impaired in Tm-treated A/A cells. Although Tm treatment changed the cytoplasmic Ca²⁺ levels in all MEFs, eif2ak3 KO (eif2ak3^−/−) MEFs displayed lower cytosolic Ca²⁺ levels than Eif2ak3^+/+ MEFs before and after Tm treatment (Figure 5C), whereas A/A^MEF cells exhibited higher cytosolic Ca²⁺ levels than S/S^MEF cells after Tm treatment (Figure 5D). These results indicate that there are no EIF2AK3- and Ca²⁺-dependent PPP3-related impairments of TFEB and TFE3 nuclear translocation in A/A cells.

Figure 5. EIF2S1 phosphorylation deficiency does not impede regulation of TFEB and TFE3 nuclear translocation by YWHA. (A) Representative fluorescence images of TFEB-EGFP in S/S-TFEB-EGFP MEFs. S/S-TFEB-EGFP MEFs were treated with DMSO, Tm only (50 ng/mL), or Tm (50 ng/mL) plus the PPP3 inhibitor FK506 (5 µM) for 16 h. The cellular localization of TFEB-EGFP was indicated by the green fluorescence signal of EGFP in cells. Nuclei were stained with DAPI (blue). Scale bar: 20 µm. The graph depicts the percentage of cells with nuclear TFEB-EGFP. Data are presented as mean ± SEM of three independent experiments (at least 130 cells per condition). (B) WB analysis of TFEB-EGFP and endogenous TFEB in protein lysates of cells treated with the same chemicals used in (A). In the left panel, proteins were separated by 6% SDS-PAGE and then subjected to WB analysis with antibodies against GFP or TFEB to detect TFEB-EGFP or endogenous TFEB, respectively. In the right panel, cells were lysed and subjected to IP with an anti-GFP antibody. Immunoprecipitates were analyzed by immunoblotting with antibodies against GFP (to detect TFEB-EGFP), phospho-(Ser)-YWHA binding motif (which binds to phosphorylated TFEB-EGFP at S211), or YWHA. (C and D) Representative measurements of Tm-induced cytosolic Ca²⁺ changes. WT (Eif2ak3^+/+) and eif2ak3-KO (eif2ak3^−/−) MEFs (C) and S/S^MEF and A/A^MEF cells (D) were treated with Tm (10 µg/mL), and Fura-2 Ca²⁺ imaging was performed as described in the Materials and Methods. The graphs depict the cytosolic Ca²⁺ concentration in basal and Tm-stimulated MEFs (Eif2ak3^+/+, n = 169; eif2ak3^−/−, n = 167; S/S^MEF, n = 134; and A/A^MEF, n = 131). Data are presented as mean ± SEM. (E and F) WB analysis of TFEB and TFE3 in protein lysates of S/S^Hep and A/A^Hep cells treated with the MTOR inhibitor Torin 2 (250 nM) (E) or Tm (1 µg/mL) (F) for the indicated durations. Proteins were separated by 6% SDS-PAGE to detect differences in the migration of TFEB and TFE3 proteins. (G) WB analysis of the phosphorylation status of TFEB-EGFP in protein lysates of S/S- and A/A-TFEB-EGFP MEFs treated with Tm (100 ng/mL) for the indicated durations. The phosphorylation status of TFEB-EGFP was analyzed using specific antibodies against phosphorylated S211 and phosphorylated S142 of TFEB. The graphs depict the levels of TFEB-EGFP phosphorylated at S211 or S142 normalized to that of total TFEB-EGFP. Data are presented as mean ± SEM of three independent experiments. *p < 0.05 and **p < 0.01, S/S-TFEB-EGFP vs A/A-TFEB-EGFP; ^#,&p < 0.05, ^##,&&p < 0.01, and ^###,&&&, 0 h vs. each time point in S/S- and A/A-TFEB-EGFP MEFs; N.S; not significant. (H and I) WB analysis of immunoprecipitated TFEB-EGFP and YWHA in S/S- and A/A-TFEB-EGFP MEFs treated with Torin 2 (50 nM, 3 h) (H) or Tm (50 ng/mL, 16 h) (I). Torin 2 treatment was performed for 3 h, which did not significantly change the levels of TFEB-EGFP proteins (see Fig. S3F vs Fig. S4C). Cells were lysed and subjected to IP with an anti-GFP antibody. Immunoprecipitates were analyzed by immunoblotting with antibodies against GFP (to detect TFEB-EGFP), phospho-(Ser)-YWHA binding motif (which binds to phosphorylated TFEB-EGFP at S211), phospho-TFEB-(S142), or YWHA. A one-way ANOVA with Tukey’s post hoc test in (A) was used and a two-way ANOVA with Sidak’s post hoc test was used in (C), (D), and (G).

To corroborate our conclusion, we compared the migration of TFEB and TFE3 in lysates of Torin 2- or Tm-treated S/S^Hep and A/A^Hep cells on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels. As expected, after Torin 2 and Tm treatment, the rapidly migrating TFEB and TFE3 forms were observed in both S/S^Hep and A/A^Hep cells, although they appeared slightly slower in A/A^Hep cells than in S/S^Hep cells in response to Tm treatment but the molecular weight of the bands in A/A^Hep cells eventually become like those in S/S^Hep cells (Figure 5E, F). By contrast, molecular weight shifts of TFEB and TFE3 proteins were not significant in eif2ak3^−/− MEFs compared with Eif2ak3^+/+ MEFs (Figure. S4A), possibly due to defective cytosolic Ca²⁺ mobilization under ER stress conditions (Figure 5C). Furthermore, we directly assessed the phosphorylation statuses of S211 and S142 in TFEB-EGFP, which are important for regulation of nuclear translocation [33,35] and export [29] of TFEB, respectively. MTORC1 is responsible for phosphorylation of TFEB residues S211 and S142 [30,31,33]. Torin 2 treatment strongly inhibited MTORC1, resulting in almost complete dephosphorylation of its target proteins (RPS6KB1 and EIF4EBP1) in both S/S and A/A cells (Fig. S3F, S4B left panels). Consistently, Torin 2 treatment strongly inhibited phosphorylation of TFEB-EGFP residues S211 and S142 in both S/S-TFEB-EGFP and A/A-TFEB-EGFP MEFs (Figure. S4C panels and graphs). In addition, we investigated whether Tm treatment inhibits MTORC1, which might contribute to the decreased phosphorylation of TFEB and its target proteins (RPS6KB1 and EIF4EBP1). Consistent with Martina’s report [37], Tm treatment decreased phosphorylation of MTORC1 itself in both S/S^Hep and A/A^Hep cells (Fig. S4B). In addition, phosphorylation of RPS6KB1 was significantly reduced, although phosphorylation of EIF4EBP1 was unchanged (Fig. S4B). Consistent with the results presented in Figure 5F, the dephosphorylation levels of TFEB-EGFP residues S211 and S142 did not differ in S/S-TFEB-EGFP and A/A-TFEB-EGFP MEFs treated with Tm for 16 h. (Fig. S4C panels and graphs). Furthermore, we time-dependently assessed the phosphorylation statuses of TFEB-EGFP at S211 and S142 in S/S-TFEB-EGFP and A/A-TFEB-EGFP MEFs after Tm treatments (Figure 5G). Although S/S-TFEB-EGFP MEFs had more dephosphorylated TFEB-EGFP than A/A-TFEB-EGFP MEFs at earlier time points, the phosphorylation statuses of TFEB-EGFP residues S211 and S142 in A/A-TFEB-EGFP MEFs eventually become similar to those in S/S-TFEB-EGFP MEFs after Tm treatment for 12 h. Therefore, based on the results presented in Figure 5A–G and Fig. S4A–C, we suggest that TFEB dephosphorylation is required but insufficient for nuclear translocation of TFEB in A/A cells during ER stress, although the delayed TFEB dephosphorylation possibly impairs TFEB nuclear translocation.

Finally, we checked whether changes of the phosphorylation status of TFEB-EGFP residues S211 and S142 affect dissociation of the TFEB-EGFP-YWHA complex, which may result in transport of TFEB-EGFP to the nucleus. As expected, dephosphorylation of TFEB-EGFP residue S211 in eif2ak3^−/− MEFs compared with Eif2ak3^+/+ MEFs was insufficient to completely dissociate the TFEB-EGFP-YWHA protein complex after Tm treatment (Fig. S4D), suggesting that activation of the EIF2AK3-Ca²⁺-PPP3 pathway is an important factor determining TFEB and YWHA dissociation and subsequent nuclear translocation of TFEB in response to Tm treatment. However, TFEB-EGFP protein immunoprecipitated from lysates of cells treated not only with Torin 2 (Figure 5H) but also with Tm (Figure 5I) showed greatly reduced phosphorylation of both S211 and S142, resulting in a strong reduction of the TFEB-EGFP-YWHA complex in both WT and EIF2S1 phosphorylation-deficient cells. Nevertheless, translocation of TFEB to the nucleus was significantly prevented in EIF2S1 phosphorylation-deficient cells, but not in WT cells under ER stress conditions as well as under MTORC1-inhibited conditions (Figure 4, Fig. S2, S3).

Altogether, our results (Figure 5, Fig. S4) and Martina’s report [37], strongly suggest that activation of EIF2AK3 and Ca²⁺-dependent PPP3 is required but insufficient for nuclear translocation of TFEB and TFE3 under ER stress conditions. In other words, there is an unknown mechanism(s) that regulates the subcellular localization of TFEB and TFE3 and is controlled by EIF2S1 phosphorylation under ER stress conditions.

TFEB translocates from the cytosol to the nucleus but cannot be retained in the nucleus in A/A cells under ER stress conditions

Recent studies demonstrate that TFEB continuously shuttles between the cytosol and nucleus via nuclear export dependent on the major exportin XPO1 under normal conditions [29,30,42]. Phosphorylation of S211 of TFEB mediates its cytosolic retention via 14–3–3 binding [30,33], whereas phosphorylation of S142 and S138 is required for recognition and binding of the TFEB NES by XPO1, which is crucial for efficient nuclear export [29,30]. However, as shown in Figure 5G, I and Fig. S4C, Tm treatment significantly reduced phosphorylation of S211 and S142, but TFEB was sequestrated in the cytosol of A/A cells. Therefore, we investigated whether TFEB undergoes continuous nucleocytoplasmic shuttling in A/A cells even after Tm treatment. The effect of treatment with the XPO1 inhibitor leptomycin B (LMB) was investigated in S/S-TFEB-EGFP and A/A-TFEB-EGFP MEFs (Figure 6A–E). Treatment with LMB only and Tm plus LMB induced nuclear translocation of TFEB-EGFP at 6 h and almost 100% nuclear translocation of TFEB-EGFP at 16 h in both S/S-TFEB-EGFP and A/A-TFEB-EGFP MEFs, whereas nuclear translocation of TFEB-EGFP was impaired in A/A-TFEB-EGFP MEFs but not in S/S-TFEB-EGFP MEFs after Tm treatment for 16 h, indicating that nucleocytoplasmic shuttling of TFEB continues in EIF2S1 phosphorylation-deficient cells but not in WT cells under ER stress conditions (Figure 6A–E). To observe dynamic changes of the subcellular localization of TFEB due to inhibition of its nuclear export, cells were sequentially treated with Tm and LMB (Figure 6F, G). Sequential treatment with Tm and LMB increased nuclear translocation of TFEB-EGFP, whereas treatment with only Tm did not induce its nuclear translocation at all in A/A-TFEB-EGFP MEFs (Figure 6F, G). These results indicate that TFEB translocates from the cytosol to the nucleus but is continuously re-exported to the cytosol by a XPO1-dependent nuclear export pathway in A/A cells under ER stress conditions.

Figure 6. TFEB translocates to the nucleus in A/A cells but is subsequently exported to the cytoplasm under ER stress conditions. (A–D) Representative fluorescence images of TFEB-EGFP in S/S- and A/A-TFEB-EGFP MEFs. MEFs were treated with DMSO (A), Tm (40 ng/mL) only (B), the nuclear export inhibitor LMB (20 nM) only (C), or Tm (40 ng/mL) plus LMB (20 nM) (D) for the indicated durations. The cellular localization of TFEB-EGFP was indicated by the green fluorescence signal of EGFP in cells. Nuclei were stained with DAPI (blue). Scale bar: 20 µm. (E) The percentage of cells with nuclear TFEB-EGFP in (A–D) at 16 h. Data are presented as mean ± SEM of three independent experiments (at least 140 cells per condition). ***p < 0.001, S/S-TFEB-EGFP vs. A/A-TFEB-EGFP; ^###p < 0.001, DMSO vs. chemicals in S/S-TFEB-EGFP; ^&&&p < 0.001, DMSO vs. chemicals in A/A-TFEB-EGFP. (F) Representative fluorescence images of TFEB-EGFP in A/A-TFEB-EGFP MEFs. MEFs were pretreated with Tm (40 ng/mL) for 6 h and further incubated with Tm in the absence or presence of LMB (20 nM) for the indicated durations. The cellular localization of TFEB-EGFP was indicated by the green fluorescence signal of EGFP in cells. Nuclei were stained with DAPI (blue). Scale bar: 20 µm. (G) The percentage of cells with nuclear TFEB-EGFP in (F). Data are presented as mean ± SEM of three independent experiments (at least 130 cells per condition). ***p < 0.001, Tm (6 h) vs. other conditions. A two-way ANOVA with Sidak’s post hoc test was used in (E) and a one-way ANOVA with Dunnett’s post hoc test was used in (G).

OE of the activated ATF6form promotes nuclear translocation of TFEB in A/A cells

Multiple reports (and the data presented in Figure 1A) demonstrate that EIF2S1 phosphorylation is required for expression or activation of several UPR TFs, including ATF4 [77–80], XBP1s [13], and ATF6 and [14], under ER stress conditions. Therefore, we examined whether OE of active forms of the UPR TFs (ATF4, XBP1s, ATF6[1–373], or ATF6[1–393]) affects nuclear translocation of TFEB-EGFP in A/A-TFEB-EGFP MEFs before and/or after Tm treatment (Fig. S5A, B). Nuclear localization of TFEB-EGFP was increased in most TF-expressing A/A-TFEB-EGFP MEFs regardless of ER stress (Fig. S5A). To determine the magnitude of TFEB-EGFP nuclear localization induced by each TF, we calculated the nuclear vs. cytosolic distribution ratio of TFEB-EGFP (Fig. S5A graphs). Among TFs, the ratio in cells expressing HA-ATF6[1–373] was largest at all time points and highest after Tm treatment for 16 h. We next assessed changes of the transcriptional activity of TFEB upon ectopic OE of HA-ATF6[1–373] and other HA-tagged TFs in A/A^MEF cells before and after Tm treatment. To this end, we used a 5XCLEAR luciferase reporter construct. Similar to the results presented in Fig. S5A, luciferase activity was highest upon HA-ATF6[1–373] OE among TFs and was further enhanced by Tm treatment (Fig. S5C). Surprisingly, the increase in reporter activities induced by HA-ATF6[1–373] was almost equivalent to that induced by an TFEB active mutant (TFEB^S211A-FLAG) (Fig. S5C). Finally, we assessed the abilities of the UPR TFs (ATF4, XBP1s, and ATF6[1–373]) to ameliorate the impairment of endogenous TFEB and TFE3 nuclear translocation under ER stress conditions by performing IF analysis of A/A^Hep cells overexpressing ATF4, Flag-XBP1s, and HA-ATF6[1–373] using recombinant adenoviruses (Figure 7A, Fig. S5D). As expected, IF analysis confirmed that the nuclear vs. cytosolic distribution ratios of endogenous TFEB and TFE3 were highest upon HA-ATF6[1–373] OE regardless of Tm treatment (Figure 7A, Fig. S5D). This verified that among the three UPR TFs, the activated ATF6 form best prevents the impaired nuclear translocation of TFEB and TFE3 induced by EIF2S1 phosphorylation deficiency. Intriguingly, before Tm treatment, ATF4 and XBP1s OE had no or a weak effect on nuclear translocation of endogenous TFEB and TFE3 whereas the activated ATF6 form OE strongly induced their nuclear translocation (Figure 7A, Fig. S5D). Quantification of TFEB and TFE3 levels in the cytosolic and nuclear fractions by WB analyses confirmed that HA-ATF6[1–373] potently induced nuclear translocation of endogenous TFEB and TFE3 in A/A^Hep cells regardless of ER stress, although Tm treatment further increased the nuclear TFEB level slightly (Figure 7B). Consistent with the increased nuclear translocation of TFEB in HA-ATF6[1-373]-overexpressing A/A^Hep cells, HA-ATF6[1–373] OE greatly induced dephosphorylation of TFEB-EGFP on S211 and S142 (Figure 7C, Fig. S6A), and resulted in dissociation of the TFEB-EGFP-YWHA complex without Tm treatment (Figure 7C). Coimmunoprecipitation (co-IP) assays revealed that ectopically expressed HA-ATF6[1–373] coprecipitated with TFEB-EGFP (Figure 7C) and vice versa (Figure 7D) in A/A-TFEB-EGFP MEFs before Tm treatment, suggesting that HA-ATF6[1–373] induces nuclear translocation of TFEB (as well as TFEB dephosphorylation and YWHA dissociation) through a physical interaction with TFEB. To confirm this, we performed a proximity ligation assay (PLA) and immunostaining assays. Most PLA signals were found in the nucleus regardless of Tm treatment, demonstrating that the interaction between TFEB and the activated ATF6 form retains TFEB in the nucleus (Figure 7E, F, Fig. S6B). In addition, a significant portion of PLA signals were in the cytosol regardless of Tm treatment, and cytosolic PLA signals decreased after Tm treatment in HA-ATF6[1-373]-expressing A/A-TFEB-EGFP MEFs (Figure 7E, Fig. S6B), indicating that complexes of TFEB and the activated ATF6 form are generated in the cytosol and translocate to the nucleus, where they are retained. Immunostaining assays of A/A^MEF cells coexpressing TFEB-EGFP and HA-ATF6[1–373] also showed colocalization of TFEB with the activated ATF6 form in the nucleus, confirming that TFEB and the activated ATF6form interact in the nucleus (Figure 7G, Fig. S6C).

Figure 7. OE of the activated ATF6 form induces nuclear translocation of TFEB in A/A cells. (A) Representative IF images of endogenous TFEB or TFE3 (red) and EGFP (green) or HA (white) in A/A^Hep cells. Cells were infected with vector-, ATF4/EGFP, FLAG-XBP1s/EGFP- or HA-ATF6[1-373]-expressing adenoviruses for 24 h and then treated with DMSO or Tm (1 µg/mL) for 24 h. Nuclei were stained with DAPI (blue). Scale bar: 20 µm. The numbers indicate the nuclear vs. cytosolic distribution ratios of endogenous TFEB or TFE3 of EGFP-positive cells in Fig. S5D. Data are presented as mean of three independent experiments (at least 150 cells per condition). (B) WB analysis of the subcellular distributions of endogenous TFEB and TFE3 in vector- or HA-ATF6[1-373]-overexpressing A/A^Hep cells. Cells infected with vector- or HA-ATF6[1-373]-expressing adenoviruses for 24 h were treated with DMSO or Tm (1 µg/mL) for 24 h. Nuclear TFEB and TFE3 levels normalized by Histone H3 levels are shown below the panels. Data are presented as mean ± SEM of three independent experiments. Histone H3 and TUBA/tubulin alpha were used as loading controls of the nuclear and cytoplasmic fractions, respectively. (C and D) WB analysis of immunoprecipitated TFEB-EGFP and YWHA (C) or HA-ATF6[1–373] (D) in vector- or HA-ATF6[1-373]-overexpressing A/A-TFEB-EGFP MEFs treated with DMSO or Tm (100 ng/mL, 24 h). Cells were lysed and subjected to IP with an anti-GFP antibody (C) or anti-HA antibody (D). Immunoprecipitates were analyzed by immunoblotting with antibodies against GFP (to detect TFEB-EGFP), phospho-(Ser)-YWHA binding motif (which binds to phosphorylated TFEB-EGFP at S211), YWHA, ATF6, or HA-ATF6[1–373]. (E and F) Quantified results of the PLA between TFEB-EGFP and HA-ATF6[1–373] in Fig. S6B. A/A-TFEB-EGFP MEFs transfected with plasmids expressing vector or HA-ATF6[1–373] for 30 h were treated with DMSO or Tm (100 ng/mL) for 16 h. (E) The graph depicts the fraction (%) of cells with PLA signals in the nucleus, nucleus and cytosol, or cytosol. Data are presented as mean of three independent experiments (at least 70 cells per condition). ^##p < 0.01, and ^###p < 0.001, nucleus vs. nucleus and cytosol; ^&&&p < 0.001, nucleus vs. cytosol; ^$p < 0.05, nucleus and cytosol vs. cytosol (one-way ANOVA with Tukey’s post hoc test). *p < 0.05, DMSO vs. Tm in cytosolic PLA-positive cells (paired Student’s t-test). (F) The graph depicts quantification of the relative PLA MFI in the nucleus. Data are presented as mean ± SEM of three independent experiments (at least 32 cells per condition). A one-way ANOVA with Tukey’s post hoc test was used. Representative PLA images of A/A-TFEB-EGFP MEFs are presented in Fig. S6B. (G) Quantification of colocalization of TFEB-EGFP with HA-ATF6[1–373] in Fig. S6C. A/A^MEF cells were cotransfected with plasmids expressing TFEB-EGFP and vector or TFEB-EGFP and HA-ATF6[1–373]. They were treated with DMSO or Tm (100 ng/mL) for 16 h, fixed, and stained with an anti-HA antibody (red) to detect HA-ATF6[1–373]. Representative colocalization IF images of HA-ATF6[1–373] and TFEB-EGFP in A/A^MEF cells are presented in Fig. S6C. Data are presented as mean ± SEM of three independent experiments (at least 25 cells per condition). A two-way ANOVA with Sidak’s post hoc test was used.

OE of the activated ATF6form enhances expression of autophagy genes and ameliorates autophagic defects in A/A cells during ER stress

Ectopically expressed HA-ATF6[1–373] potently induced nuclear translocation of TFEB and TFE3, and increased the activity of the TFEB binding motif (CLEAR)-driven luciferase reporter in A/A cells. Therefore, we assessed whether the activated ATF6 form upregulates expression of TFEB and TFE3-dependent autophagy genes in A/A^Hep cells. To this end, A/A^Hep cells were infected with Ad-vector or Ad-HA-ATF6[1–373] and then treated with Tm for the indicated durations. Quantitative PCR and WB analyses confirmed that HA-ATF6[1–373] was overexpressed and harbored transcriptional activities (Figure 8A left graph, B left panels), as judged by increased mRNA and protein expression of the UPR target genes Hspa5/BiP [81–84], Ddit3 [84–87], Xbp1t, and Xbp1s [84,86]. Among the examined genes, Atf4 mRNA and ATF4 protein were also upregulated in HA-ATF6[1-373]-expressing cells without Tm treatment (Figure 8A left graph, B left panels), which has not been previously reported. Expression of XBP1s and ATF4 proteins was also increased when HA-ATF6[1–373] was overexpressed in A/A-TFEB-EGFP MEFs (Fig. S5B). Expression analysis of autophagy genes demonstrated that the mRNA and protein levels of Lc3b, Sqstm1, and Ctsb genes were significantly higher in Ad-HA-ATF6[1-373]-infected cells than in Ad-vector-infected cells without Tm treatment, and their mRNA levels were further increased by Tm treatment (Figure 8A right graph, B right panels). Although LC3B conversion (LC3B-II:I ratio) was lower in A/A^Hep cells than in S/S^Hep cells at most time points (Figure 1D), HA-ATF6[1–373] OE in A/A^Hep cells increased LC3B conversion (LC3B-II:I ratio) at the late stages (12 and 24 h), implying that the activated ATF6form enhances autophagosome formation in A/A cells during ER stress. Although HA-ATF6[1–373] expression itself did not increase mRNA (Ctsd, Ctsl, Lamp1, Lamp2a, Lamp2b, Lamp2c, Mcoln1, Tpp1, Glb1, and Atp6v1h) (Figure 8A right graph) or protein (CTSL, LAMP1, and LAMP2) (Figure 8B right panel) expression of many other autophagy genes, Tm treatment strongly enhanced the levels of these transcripts (Figure 8A right graph) and proteins (Figure 8B right panels) in HA-ATF6[1-373]-expressing A/A^Hep cells. These results suggest that enhancement of autophagy gene expression by the activated ATF6form requires other components (such as other TFs and ATF6 PTMs) that are induced by ER stress. Together, these data indicate that activated ATF6 form OE can ameliorate the dysregulated expression of TFEB and TFE3-dependent autophagy genes and some UPR genes in A/A cells during ER stress.

Figure 8. OE of the activated ATF6 form increases expression of autophagy genes and improves autophagic defects in A/A cells during ER stress. A/A^Hep cells infected with vector- or HA-ATF6[1-373]-expressing adenoviruses for 24 h were treated with DMSO or Tm (1 µg/mL) for the indicated durations. (A) Quantitative RT-PCR analysis of mRNA expression of ER stress response and autophagy genes. Data are presented as mean ± SEM of three independent experiments. *p < 0.05, **p < 0.01, and ***p < 0.001, Ad-vector vs. Ad-HA-ATF6[1–373]; ^#p < 0.05, ^##p < 0.01, and ^###p < 0.001, DMSO vs. Tm in Ad-vector; ^&p < 0.05, ^&&p < 0.01, and ^&&& p < 0.001, DMSO vs. Tm in Ad-HA-ATF6[1–373]. The dotted line was put to compare relative mRNA levels of HA-ATF6[1-373]-expressing A/A cells with them of ATF4- (Fig. S8A) or FLAG-XBP1s (Fig. S8B)-overexpressing A/A cells. (B) WB analysis of ER stress and autophagy proteins in cell lysates. To observe the expression levels of ER stress, overexpressed TF, and autophagy proteins, lysates were prepared from HA-ATF6[1-373]-overexpressing A/A^Hep cells after Tm treatment at each time point. Lysates of Tm (0 and 24 h)-treated S/S^Hep cells were prepared as positive controls. The lysates were subjected to WB analysis of the indicated proteins. ATF6(N): cleaved N-terminal fragment of endogenous ATF6. The LC3B-II:I ratios are shown below the right first panel. CTSB: cathepsin B; CTSL: cathepsin L; Pro: procathepsin; Sc: mature single-chain cathepsin; Dc: heavy chain of mature double-chain cathepsin. (C) Representative images of LysoTracker staining in vector- or HA-ATF6[1-373]-overexpressing A/A^Hep cells. Cells were stained with LysoTracker (100 nM, red) and Hoechst 33,258 (10 μg/mL, blue) for the last 30 min of the treatment. The dotted white line defines the cell boundary. Scale bar: 20 µm. The graph depicts quantification of the MFI of LysoTracker. Data are presented as mean ± SEM of three independent experiments (at least 150 cells per condition). (D) Representative IF images of LC3A/B (green) and LAMP1 (red) in vector- or HA-ATF6[1-373]-overexpressing A/A^Hep cells. Nuclei were stained with DAPI (blue). The bottom panels are magnified images of the boxes in the upper panels. Yellow IF signal indicates double labeling of LC3A/B (green) and LAMP1 (red). Scale bar: 20 µm. (E) The graph depicts the fraction (%) of cells with different LC3A/B staining patterns as described in Figure 2A. Data are presented as mean ± SEM of three independent experiments (at least 150 cells per condition). (F) The graph depicts quantification of the colocalization of LC3A/B with LAMP1 in (D). Data are presented as mean ± SEM of three independent experiments (at least 150 cells per condition). (G) Representative TEM images of vector- or HA-ATF6[1-373]-overexpressing A/A^Hep cells. The bottom panels are magnified images of the red boxes in the upper panels. Red arrowheads indicate autolysosomes, and yellow arrows indicate the ER. The dotted yellow line defines a mass of dilated and fragmented ER structures. Scale bars: upper panels 1 or 2 µm and bottom panels 0.2 µm. (H) Quantification of the number of autolysosomes per cell TEM images in (G). Data are presented as mean ± SEM of three independent experiments (at least 15 cells per condition). A two-way ANOVA with Sidak’s post hoc test was used in (A), (C), (F), and (H).

Nuclear translocation of TFEB and TFE3 is important for regulation of lysosome biogenesis and function [22,26,27]. Therefore, we checked whether ectopically expressed HA-ATF6[1-373]-mediated nuclear translocation of TFEB prevents perturbation of lysosome biogenesis and function in A/A^Hep cells under ER stress conditions. Similar to the results presented in Figure 2G, null expressing A/A^Hep cells exhibited markedly decreased LysoTracker Red staining and peripheral accumulation of lysosomes, whereas HA-ATF6[1-373]-expressing A/A^Hep cells displayed significant increases in the intensity of LysoTracker Red staining and perinuclear accumulation of lysosomes after Tm treatment (Figure 8C). This indicates that TFEB and TFE3 activation mediated by the activated ATF6 form prevents lysosomal dysfunction in A/A cells under ER stress conditions.

We next investigated if the ability of the activated ATF6 form to induce TFEB and TFE3 activation ameliorates autophagic defects in A/A cells during ER stress. HA-ATF6[1–373] OE significantly reduced accumulation of LC3A/B-positive structures in perinuclear regions, and conversely increased the number of puncta positive for LC3A/B (autophagosomes) and LAMP1 (autolysosomes or lysosomes) in Tm-treated A/A^Hep cells (Figure 8D, E). Furthermore, HA-ATF6[1–373] OE markedly enhanced the colocalization of LC3A/B and LAMP1 in Tm-treated A/A^Hep cells (Figure 8D, F). Similarly, HA-ATF6[1–373] OE enhanced the colocalization of SQSTM1 and LAMP1 in Tm-treated A/A^Hep cells (Fig. S7A). Finally, TEM analyses demonstrated that HA-ATF6[1–373] OE increased the number of autolysosomes in Tm-treated A/A^Hep cells (Figure 8G, H). In addition, as reported previously [88], HA-ATF6[1–373] OE induced ER expansion in A/A^Hep cells not treated with Tm and reduced accumulation of the fragmented ER in Tm-treated A/A^Hep cells (Figure 8G). Together, our data suggest that OE of the activated ATF6 form increases expression of TFEB and TFE3-dependent autophagy genes, and ameliorates autophagic defects and disruption of the ER structure in A/A cells during ER stress.

OE of ATF4 and XBP1s improves expression of autophagy genes and alleviates autophagic defects in A/A cells during ER stress

The activated ATF6 form is an important factor to induce nuclear translocation and activation of TFEB and TFE3 in A/A cells under ER stress conditions. Therefore, we questioned whether ATF6 deficiency influences nuclear translocation of TFEB and TFE3 in Tm-treated cells. However, similar to WT hepatocytes (HEP-Atf6^+/+), ATF6-deficient hepatocytes (HEP-atf6^−/−) also displayed normal nuclear translocation of TFEB and TFE3 in response to Tm treatment (Fig. S7B). These data indicate that ATF6 is not absolutely required for nuclear accumulation of TFEB and TFE3 during ER stress.

We showed that EIF2S1 phosphorylation was required for expression of ATF4 and XBP1s during ER stress (Figure 1A) [7,13,53]. HA-ATF6[1–373] OE increased expression of Atf4 and Xbp1s mRNAs and their proteins (Figure 8A, B, Fig. S5B). Inversely, ATF6-deficient hepatocytes (HEP-atf6^−/−), which had no defect in TFEB and TFE3 nuclear translocation, expressed more ATF4 than WT cells and a similar level of XBP1s as WT cells in response to Tm treatment (Fig. S1A). In addition, ATF4 and XBP1s OE in A/A cells induced nuclear translocation of TFEB and TFE3 under ER stress conditions, although the effect of activated ATF6 form OE was stronger (Figure 7A, Fig. S5). All the data described above indicate that ATF4 or XBP1s OE can mitigate autophagic defects in A/A cells under ER stress conditions. Therefore, we conducted experiments similar to those performed with HA-ATF6[1–373] in Figure 8A-F. For these experiments, A/A^Hep cells were infected with Ad-ATF4 EGFP or Ad-Flag-XBP1s EGFP and then treated with Tm for the indicated durations. As reported previously [7,9,13,53,84], ATF4 and XBP1s OE increased expression of mRNAs (Ddit3, Ppp1r15a, Asns, Xbp1s, and Hspa5 for ATF4 OE; Xbp1t, Xbp1s, Ddit3, Hspa5, and Atf6 for XBP1s OE) (Figure S8A, B left graphs) and proteins (DDIT3 in ATF4 OE; HSP90B1, HSPA5, and ATF6(N) for XBP1s OE) (Fig. S8C upper panel and S8D left panel) of their UPR target genes in response to Tm treatment. Expression analysis of autophagy genes demonstrated that the levels of mRNAs (Lc3b, Sqstm1, Ctsl, Lamp2b, and Mcoln1 for ATF4 OE; Lc3b, Ctsb, Ctsd, Lamp1, Lamp2a, Lamp2b, Lamp2c, Glb1, Atp6v1h, Tfe3, and Tfeb for XBP1s OE) (Fig. S8A, B right graphs) and proteins (LC3B, LAMP1, LAMP2, CTSB, and CTSL in ATF4 OE; LC3B, SQSTM1, LAMP1, LAMP2, CTSB, CTSL, TFEB, and TFE3 for XBP1s OE) (Fig. S8C lower panel and S8D right panel) of several autophagy genes were significantly higher in Ad-ATF4 EGFP- or Ad-Flag-XBP1s EGFP-infected cells than in Ad-vector-infected cells in response to Tm treatment. In addition, ATF4- and XBP1s-overexpressing A/A^Hep cells displayed significant increases in the intensity of LysoTracker Red staining and perinuclear accumulation of lysosomes after Tm treatment (Fig. S9A, B), indicating that ATF4 and XBP1s can mitigate lysosomal dysfunction in A/A cells under ER stress conditions. ATF4 and XBP1s OE reduced accumulation of LC3A/B-positive structures in perinuclear regions, and conversely increased the number of puncta positive for LC3A/B (autophagosomes) and LAMP1 (autolysosomes or lysosomes) in Tm-treated A/A^Hep cells (Fig. S9C-F). Furthermore, ATF4 and XBP1s OE enhanced the colocalization of LC3A/B and LAMP1 in Tm-treated A/A^Hep cells (Fig. S9G, H). Together, our data suggest that ATF4 and XBP1s OE also increases expression of TFEB and TFE3-dependent autophagy genes and alleviates autophagic defects in A/A cells during ER stress.

OE of the constitutively active TFEB mutant enhances expression of autophagy genes and ameliorates autophagic defects in A/A cells during ER stress

The results presented in Figure 7 and 8 indicate that the effects of the activated ATF6 form on autophagy are mediated by TFEB and TFE3 activation in A/A cells during ER stress. Therefore, we investigated whether TFEB OE increases expression of autophagy genes and thereby prevents autophagic defects in Tm-treated A/A cells. First, we overexpressed WT TFEB (TFEB[WT]-FLAG) or a constitutively active TFEB mutant (TFEB^S211A-FLAG) in A/A^Hep cells and then treated these cells with Tm for the indicated durations. Similar to endogenous TFEB in S/S^Hep and A/A^Hep cells, the subcellular localization of overexpressed TFEB[WT]-FLAG differed depending on the genetic background of the cells and stress conditions, although TFEB[WT]-FLAG exhibited a nuclear localization in almost 25% of A/A^Hep cells treated with Tm for 24 h (Fig. S10A), possibly due to altered regulation by OE. On the other hand, the TFEB mutant (TFEB^S211A-FLAG) primarily accumulated in the nucleus regardless of the genetic background of the cells and stress conditions (Fig. S10B), possibly due to the absence of YWHA-mediated cytoplasmic sequestration [33,35]. Second, we compared the transcriptional activities of TFEB[WT]-FLAG and TFEB^S211A-FLAG in A/A^MEF cells using the 5XCLEAR luciferase reporter construct before and after Tm treatment (Figure 9A, B). TFEB^S211A-FLAG exhibited stronger transcriptional activity than TFEB[WT]-FLAG after Tm treatment, whereas their transcriptional activities were similar before Tm treatment (Figure 9A, B). Thus, OE of the constitutively active TFEB mutant and even WT TFEB may enhance expression of autophagy genes in A/A cells during ER stress.

Figure 9. OE of the constitutively active TFEB mutant enhances expression of autophagy genes and improves autophagic defects in A/A cells during ER stress. (A) Luciferase activity assay of the 5xCLEAR luciferase reporter. A/A^MEF cells were cotransfected with plasmids expressing 5xCLEAR-driven firefly luciferase, CMV-driven Renilla luciferase, and FLAG-tagged TFEB (TFEB[WT]-FLAG or TFEB^S211A-FLAG) for 30 h. Cells were then treated with DMSO or Tm (100 ng/mL) for 16 h, and luciferase activities were measured. Data are presented as mean ± SEM of three independent experiments. ***p < 0.001, Vector vs. TFEB[WT]-FLAG or TFEB^S211A-FLAG; ^###p < 0.001, DMSO vs. Tm; ^&&&p < 0.001, TFEB[WT]-FLAG vs. TFEB^S211A-FLAG. (B) WB analysis of overexpressed TFEB[WT]-FLAG and TFEB^S211A-FLAG proteins in A/A^MEF cells in (A). (C) Quantitative RT-PCR analysis of mRNA expression of ER stress response and autophagy genes in vector-, TFEB[WT]-FLAG-, or TFEB^S211A-FLAG-overexpressing A/A^Hep cells. A/A^Hep cells infected with vector-, TFEB[WT]-FLAG-, or TFEB^S211A-FLAG-expressing adenoviruses for 24 h were treated with DMSO or Tm (1 µg/mL) for 24 h. Data are presented as mean ± SEM of three independent experiments. *p < 0.05, **p < 0.01, and ***p < 0.001, Ad-vector vs. Ad-TFEB[WT]-Flag or Ad-TFEB^S211A-Flag; ^#p < 0.05 and ^##p < 0.01, Ad-TFEB[WT]-Flag vs. Ad-TFEB^S211A-Flag. Ctsb: cathepsin B; Ctsd: cathepsin D; Ctsl: cathepsin L. (D and E) WB analysis of EIF2S1, p-EIF2S1, its downstream target proteins (D), and autophagy and lysosomal proteins (E) in vector-, TFEB[WT]-FLAG-, or TFEB^S211A-FLAG-overexpressing A/A^Hep cells. Cells infected with vector-, TFEB[WT]-FLAG-, or TFEB^S211A-FLAG-expressing adenoviruses for 24 h were treated with DMSO or Tm (1 µg/mL) for the indicated durations. Cell lysates were prepared at each time point after Tm treatment. Lysates of Tm (0 and 24 h)-treated S/S^Hep cells were prepared as positive controls. They were subjected to WB analysis of the indicated proteins. The LC3B-II:I ratios are shown below the first panel in (E). CTSB: cathepsin B; CTSL: cathepsin L; Pro: procathepsin; Sc: mature single-chain cathepsin; Dc: heavy chain of mature double-chain cathepsin. (F) Representative IF images of LC3A/B (green) and TFEB^S211A-FLAG (red) in vector- or TFEB^S211A-FLAG-overexpressing A/A^Hep cells. Cells were treated with DMSO or Tm (1 µg/mL) for 24 h. The dotted white line defines the cell boundary. Scale bar: 20 µm. The graph depicts the fraction (%) of cells with different LC3A/B staining patterns as described in Figure 2A. Data are presented as mean ± SEM of three independent experiments (at least 150 cells per condition). (G) Representative LysoTracker staining images of vector- or TFEB^S211A-FLAG-overexpressing A/A^Hep cells. Cells were stained with LysoTracker (100 nM, red) and Hoechst 33,258 (10 μg/mL, blue) for the last 30 min of the treatment. The dotted white line defines the cell boundary. Scale bar: 20 µm. The graph depicts quantification of the MFI of LysoTracker. Data are presented as mean ± SEM of three independent experiments (at least 150 cells per condition). (H) Representative IF images of LC3A/B (green) and LAMP1 (red) in vector- or TFEB^S211A-FLAG-overexpressing A/A^Hep cells. Cells were treated with DMSO or Tm (1 µg/mL) for 24 h. Nuclei were stained with DAPI (blue). The third panels in the bottom row are magnified images of the boxes in the second panels. Yellow IF signal indicates double labeling of LC3A/B (green) and LAMP1 (red). Scale bars: 20 µm except for the magnified images (10 µm). The graph depicts quantification of the colocalization of LC3A/B with LAMP1. Data are presented as mean ± SEM of three independent experiments (at least 150 cells per condition). A one-way ANOVA with Tukey’s post hoc test was used in (C) and a two-way ANOVA with Sidak’s post hoc test was used in (A), (G), and (H).

Next, we assessed the expression levels of individual TFEB and TFE3 target genes in TFEB[WT]-FLAG- and TFEB^S211A-FLAG-expressing A/A^Hep cells after Tm treatment. First, because Martina et al. reported that TFEB^S211A OE enhances the ATF4-mediated ER stress response in WT MEFs [37], we investigated whether OE of TFEB[WT]-FLAG or TFEB^S211A-FLAG increases the transcript levels of ER stress responsive genes (Atf4, Ddit3, Xbp1t, Xbp1s, and Hspa5) in Tm-treated A/A^Hep cells. The levels of Atf4 and Ddit3 mRNAs were slightly increased in TFEB^S211A-FLAG-expressing Tm-treated A/A^Hep cells (Figure 9C). However, TFEB^S211A-FLAG-mediated transcriptional upregulation of Atf4 and the downstream target Ddit3 did not lead to increases of their protein levels, possibly due to the absence of EIF2S1 phosphorylation in A/A^Hep cells during ER stress [7,53,78] (Figure 9D). Second, OE of TFEB[WT]-FLAG and TFEB^S211A-FLAG increased expression of multiple autophagy genes (Figure 9C), but the transcriptional activity of TFEB^S211AFLAG was much stronger than that of TFEB[WT]-FLAG, although their mRNA (Figure 9C) and protein (Figure 9E) levels were similar. LC3B conversion (LC3B-II:I ratio) as well as LC3B-I and II protein levels were higher in TFEB[WT]-FLAG- and TFEB^S211A-FLAG-expressing A/A^Hep cells than in null expressing A/A^Hep cells at all time-points (Figure 9E). In addition, the protein level of SQSTM1 gradually decreased possibly due to autophagic degradation [15,16], whereas its mRNA level increased, in TFEB^S211A-FLAG-expressing A/A^Hep cells treated with Tm (Figure 9C, E). The gradual decreases of LAMP1 and LAMP2 protein levels were slightly delayed by increases of their mRNA levels in TFEB^S211A-FLAG-expressing A/A^Hep cells treated with Tm (compare lanes 2 and 3 with lanes 10 and 11) (Figure 9C, E). Expression levels of other lysosomal proteins (CTSB and CTSL) were increased in proportion to their mRNA levels in both TFEB[WT]-FLAG- and TFEB^S211A-FLAG-expressing A/A^Hep cells (Figure 9C, E). Thus, we confirmed that TFEB activation in EIF2S1 phosphorylation-deficient cells can prevent dysregulated expression of autophagy genes during ER stress.

Next, based on the gene expression analysis, we reasoned that TFEB^S211A-FLAG OE would improve autophagic processes such as autophagosome formation and autophagosome-lysosome fusion in A/A^Hep cells under ER stress conditions. To test this idea, we investigated whether TFEB^S211A-FLAG OE increases the number of A/A^Hep cells containing LC3A/B-positive puncta after Tm treatment. TFEB^S211A-FLAG OE markedly increased the number of A/A^Hep cells containing LC3A/B-positive puncta (~70%, Figure 9F graph) after Tm treatment, suggesting that active TFEB mutant OE alleviates dysregulated autophagosome formation in A/A^Hep cells during ER stress. Furthermore, TFEB^S211A-FLAG OE attenuated perinuclear accumulation of LC3A/B-positive structures in Tm-treated A/A^Hep cells (Figure 9F). TFEB OE increased expression of several lysosomal genes such as Ctsb, Ctsd, Ctsl, Lamp1, Lamp2a, Lamp2b, and Tpp1 (Figure 9C, E). Therefore, we next determined the effects of TFEB^S211A-FLAG OE on lysosome biogenesis and function under ER stress conditions. TFEB^S211A-FLAG OE attenuated the decrease of LysoTracker Red staining intensity and increased perinuclear accumulation of lysosomes in A/A^Hep cells under ER stress conditions (Figure 9G). These results confirmed that TFEB activation can prevent lysosomal dysfunction in A/A cells under ER stress conditions. We next investigated if active TFEB mutant OE ameliorates the impairment of autolysosome formation in A/A cells during ER stress. TFEB^S211A-FLAG OE significantly enhanced the colocalization of LC3A/B and LAMP1 in Tm-treated A/A^Hep cells (Figure 9H), indicating that autolysosome formation was increased. Collectively, these results demonstrate that TFEB activation can resolve most autophagic defects in EIF2S1 phosphorylation-deficient cells during ER stress.

The constitutively active TFEB mutant restores autophagic flux and promotes autophagic degradation of misfolded proteins in A/A cells

To substantiate the effects of TFEB on autophagy induced by ER stress in A/A cells, we performed TEM analyses of vector- and TFEB^S211A-FLAG-expressing A/A^Hep cells treated with and without Tm. TFEB^S211A-FLAG-expressing A/A^Hep cells exhibited an increased number of autolysosomes, whereas vector-expressing A/A^Hep cells had few autolysosomes after Tm treatment, indicating that constitutively active TFEB mutant OE increases autolysosome formation in A/A cells during ER stress (Figure 10A, B). However, in contrast with the activated ATF6 form, TFEB^S211A-FLAG did not reduce the severity of ER fragmentation and accumulation of the fragmented ER tubules (yellow dotted areas in Tm-treated A/A^Hep panels of Figure 10A, C), indicating that the altered ER structures observed in Tm-treated A/A cells are not critical obstacles of autophagy pathways. We next performed the autophagic flux assay using Baf A1 to biochemically confirm that TFEB^S211A-FLAG expression changes autophagic activity in Tm-treated A/A^Hep cells. In the absence of Tm treatment, the increase (lanes 1–2 vs. lanes 3–4) of LC3-II levels induced by Baf A1 did not significantly differ between vector- and TFEB^S211A-FLAG-expressing A/A^Hep cells (Figure 10D). By contrast, Tm treatment alone strongly decreased (lane 5 vs. lane 7) LC3-II levels in TFEB^S211A-FLAG-expressing A/A^Hep cells compared with vector-expressing A/A^Hep cells, and consequently, the increase (lanes 5–6 vs. lanes 7–8) of LC3-II levels induced by Baf A1 was significantly higher in the former cells than in the latter cells. These results confirm that the constitutively active TFEB mutant restores autophagic flux in A/A^Hep cells under ER stress conditions.

Figure 10. OE of the constitutively active TFEB mutant rescues the autophagic flux defect in A/A cells during ER stress. (A and -C) Representative TEM images of vector- or TFEB^S211A-FLAG-overexpressing A/A^Hep cells. Cells infected with vector- or TFEB^S211A-FLAG-expressing adenoviruses for 24 h were treated with DMSO or Tm (1 µg/mL) for 24 h. (B) Quantification of the number of autolysosomes per cell in the TEM images in (A). Data are presented as mean ± SEM of three independent experiments (at least 15 cells per condition). The bottom panels in (A) are magnified images of the red boxes in the upper panels. Red arrowheads indicate autolysosomes and yellow arrows indicate the ER. The dotted yellow line defines a mass of dilated and fragmented ER structures. Scale bars: upper panels of (A) (2 µm) and bottom panels of (A) and (C) (0.5 µm). (D) WB analysis of LC3B in protein lysates of vector- or TFEB^S211A-FLAG-overexpressing A/A^Hep cells. Cells infected with vector- or TFEB^S211A-FLAG-expressing adenoviruses for 24 h were treated with DMSO or Tm (1 µg/mL) for 16 h in the absence or presence of the lysosomal inhibitor Baf A1 (200 nM) for 3 h before harvest. The graph depicts the LC3B-II level normalized to the ACTB level. Data are presented as mean ± SEM of three independent experiments, N.S., no significant difference. (E) WB analysis of ATZ in protein lysates of vector- or TFEB^S211A-FLAG-overexpressing A/A^Hep cells. Cells were cotransfected with plasmids expressing ATZ and vector or TFEB^S211A-FLAG for 24 h and then treated with DMSO, MG132 only (20 µM), or MG132 plus CHX (100 µg/mL) for 6 h. The graphs depict the ATZ level normalized to the ACTB level after treatment for 6 h. Data are presented as mean ± SEM of three independent experiments. A two-way ANOVA with Sidak’s post hoc test was used in (B), (D), and (E).

Finally, we examined the effects of constitutively active TFEB mutant OE on autophagic degradation of ATZ protein. ATZ-expressing A/A^Hep cells were cotreated with MG132 and CHX to retain only autophagic activity (Figure 10E). Therefore, upon cotreatment with MG132 and CHX, ATZ protein will be predominantly degraded by autophagy pathways. WB analysis revealed that cotreatment with MG132 and CHX for 6 h decreased the ATZ levels in both vector- and TFEB^S211A-FLAG-expressing A/A^Hep cells (lanes 3 and 6). However, ATZ degradation was much more efficient in TFEB^S211A-FLAG-expressing A/A^Hep cells than in vector-expressing A/A^Hep cells (Figure 10E, lane 3 vs. lane 6). These results indicate that the constitutively active TFEB mutant enhances autophagic degradation of misfolded proteins in A/A cells.

Discussion

In this study, we showed that EIF2S1 phosphorylation plays an essential role in nuclear translocation of TFEB and TFE3 during ER stress. Under ER stress conditions, EIF2S1 phosphorylation-deficient A/A cells display dysregulated expression of autophagy genes and impairment of several autophagic processes (such as autophagosome and autolysosome formation), which are regulated by the nuclear translocation and functional activity of TFEB and TFE3. Particularly, we revealed that OE of the activated ATF6 form (HA-ATF6[1–373]), XBP1s, and ATF4, production of which was significantly reduced and delayed in A/A cells during ER stress, ameliorated these autophagic defects of A/A cells by promoting the nuclear translocation and functional activity of TFEB and TFE3. In turn, OE of a constitutively active TFEB mutant (TFEB^S211A-FLAG) alleviated the autophagic defects of A/A cells during ER stress. Collectively, our results reveal how EIF2S1 phosphorylation connects the UPR pathways to autophagy.

As reported previously [37], we found that nuclear translocation of TFEB in Tm-treated WT cells requires Ca²⁺-dependent PPP3 activation, TFEB dephosphorylation, and YWHA dissociation (Figure 5A, B, F, G, I, and Fig. S4A, C, D). Furthermore, we showed that EIF2AK3 was necessary for nuclear translocation of TFEB and TFE3 (Fig. S2A) [37] because it contributed to cytosolic Ca²⁺ flux during ER stress (Figure 5C). Tm-mediated cytosolic Ca²⁺ flux was not reduced but increased in EIF2AK3-sufficient A/A cells (Figure 1A) compared with S/S cells (Figure 5D). Likewise, A/A cells displayed TFEB dephosphorylation and YWHA dissociation similar to S/S cells under ER stress conditions (Figure 5F, G, I, and Fig. S4C), although TFEB dephosphorylation was delayed at earlier time points (Figure 5F, G). In addition, Tm treatment induced MTORC1 inhibition, which may contribute to reduced phosphorylation of TFEB in A/A cells (Fig. S4B). Nevertheless, nuclear translocation of TFEB and TFE3 was strongly suppressed in A/A cells under ER stress conditions (Figure 4, Fig. S2, S3). However, experiments using the XPO1 inhibitor LMB revealed that activated TFEB translocated to the nucleus but was continuously re-exported to the cytosol via a XPO1-dependent nuclear export pathway in A/A cells under ER stress conditions (Figure 6). Therefore, we postulate that the TFEB nuclear import pathway is not defective, but the TFEB nuclear export pathway is dysregulated in A/A cells during ER stress. Diverse PTMs including phosphorylation regulate nuclear translocation of TFEB and TFE3 [28–32]. Most PTM studies focused on a change in the localizations of TFEB and TFE3 from the cytosol to the nucleus to explain their nuclear translocation [31,35–38]. However, recent reports including a study by Napolitano et al. [29,30,41] imply that nuclear translocation of TFEB and TFE3 can be accomplished when their nuclear export pathway is inhibited because these TFs undergo nucleocytoplasmic shuttling.

Based on previous reports and our results described above, we propose three possible explanations for dysregulation of the TFEB nuclear export pathway in A/A cells during ER stress. First, dysregulation of the TFEB nuclear export pathway may arise due to malfunction of the XPO1-mediated nuclear export pathway. EIF2S1 phosphorylation deficiency may dysregulate the XPO1-mediated nuclear export pathway, which is responsible for nuclear export of diverse XPO1 cargo proteins [89] including TFEB and TFE3, under ER stress conditions. However, this is not a plausible explanation because previous reports indicate that the XPO1-mediated nuclear export pathway works properly for nuclear localization or export of specific target proteins in A/A cells under ER stress conditions. For example, NFE2L2/NRF2 (NFE2 like bZIP transcription factor 2) is a TF essential for antioxidant response element-mediated gene expression and a direct EIF2AK3 substrate [90]. Under normal conditions, NFE2L2 also continuously shuttles between the cytosol and nucleus via XPO1-dependent nuclear export [89,91]. However, Tm treatment induces nuclear translocation of NFE2L2 independently of EIF2S1 phosphorylation [90]. In addition, it was reported that nuclear export of p53 mediated by EIF2S1 kinases such as EIF2AK3 and EIF2AK2/PKR occurs independently of EIF2S1 phosphorylation under ER stress conditions [72]. Thus, dysregulation of the TFEB nuclear export pathway may arise due to impairment of a pathway that specifically exports TFEB and a few related proteins from the nucleus, rather than due to malfunction of the XPO1-mediated nuclear export pathway, which can affect diverse XPO1 cargo proteins.

Second, dysregulation of the TFEB nuclear export pathway may arise due to the absence or presence of a specific modification (such as phosphorylation or other PTMs) on a specific residue(s) of TFEB that can determine its nuclear export. Phosphorylation of S142 and S138 is proposed to be required for recognition and binding of the TFEB NES by XPO1 [29,30,42], which is crucial for efficient nuclear export, whereas phosphorylation of S211 is required for cytosolic retention of TFEB via YWHA binding [35–38]. Phosphorylation of S138 is dependent on prior phosphorylation of S142 [29,30]. Furthermore, reports indicate that S142 and/or S138 phosphorylation may be nuclear events mediated by several kinases such as MTORC1 (at S142 and S138), extracellular signal-regulated kinase (at S142), and GSK3B (at S138) [29,30]. Yin et al. reported that CDK4 and CDK6 interact with and phosphorylate TFEB on S142 in the nucleus [41]. Thus, the status of S142 phosphorylation is an important determinant of TFEB nuclear export. However, dephosphorylation levels of total TFEB proteins on S142 were similar in S/S and A/A cells treated with Tm (Figure 5G, I, Fig. S4C), although its dephosphorylation in A/A cells was delayed at earlier time points (Figure 5G). Therefore, it is possible that impairment of TFEB nuclear translocation may not arise due to dysregulation of TFEB dephosphorylation on S142 in A/A cells during ER stress. However, although WB analysis of total TFEB proteins indicates that S142 of TFEB is dephosphorylated by Ca²⁺-dependent PPP3, we cannot rule out the possibility that S142 of nuclear TFEB is immediately and temporarily rephosphorylated by nuclear kinases (such as MTORC1 or CDK4 and CDK6), and that phosphorylated TFEB is rapidly exported to the cytosol in A/A cells during ER stress. Thus, even under ER stress conditions, the continuous cycle of nuclear import and export of TFEB according to its dephosphorylation and rephosphorylation may give the impression that it never enters the nuclei of A/A cells. Therefore, it is worth examining whether specific nuclear kinases (such as MTORC1 and CDK4 and CDK6) are activated and localized to the nucleus for rephosphorylation of TFEB on S142 in A/A cells during ER stress. These issues require further investigation.

Lastly, dysregulation of the TFEB nuclear export pathway may arise due to the lack of a TFEB-interacting nuclear protein that retains TFEB in the nucleus. In this report, we suggested that the activated ATF6 form (HA-ATF6[1–373]) is a missing TFEB-interacting nuclear protein in A/A cells during ER stress. Co-IP assays (Figure 7C, D), the PLA (Figure 7E, F, Fig. S6B), and colocalization experiments (Figure 7G, Fig. S6C) indicate there is a physical interaction between TFEB and the activated ATF6 form, and that most of these complexes are in the nucleus, although the interaction might also occur in the cytosol. This nuclear interaction retains TFEB in the nucleus. In addition, this specific interaction can induce TFEB dephosphorylation, YWHA dissociation, and nuclear translocation of TFEB, regardless of ER stress. However, further studies are required to answer many questions about the detailed molecular mechanisms underlying activated ATF6 form-mediated dephosphorylation and nuclear translocation of TFEB. Furthermore, although we suggested that the activated ATF6 form promotes the nuclear translocation and functional activity of TFEB and TFE3 in A/A cells best among the tested UPR TFs because its activities were highest (based on the results presented in Figure 7A and Fig. S5A, S5D for nuclear translocation; Figure 8A and Fig. S5C, S8A, and S8B for autophagy gene expression; and Figure 8D-F and Fig. S9C-H for autophagosome/autolysosome formation), OE of other TFs (ATF4 and XBP1s) also significantly affected nuclear translocation and functional activities of TFEB and TFE3 in A/A cells (Figure 7A, Fig. S5, S8, and S9). In addition, expression and activation of ATF4, XBP1, and ATF6 are influenced by each other. ATF4 facilitates synthesis of ATF6 and its trafficking from the ER to the Golgi for its proteolytic activation [14], although these effects were not observed in A/A cells (Fig. S8A, C). The activated ATF6 form activates transcription of the Xbp1 genes [86] and heterodimerizes with XBP1s to induce UPR genes [81]. In addition, we showed that activated ATF6 form OE increased the mRNA and protein levels of Atf4, Xbp1t, and Xbp1s (Figure 8A, B, Fig. S5B). Interestingly, XBP1s OE increased the mRNA level of Atf6 and its proteolytic activation in A/A cells (Fig. S5B, D). Furthermore, ATF4 is reportedly required to induce transcription of a set of autophagy genes in response to ER stress [50] and hepatic OE of XBP1s enhances Tfeb transcription and autophagy [92] (Fig. S8A, B). Thus, we cannot rule out the possibility that ATF4 and XBP1s play roles in activated ATF6 form-overexpressing A/A cells under ER stress conditions. Further studies are needed to investigate whether ATF4 and/or XBP1s are required for the improvement of autophagy by the activated ATF6 form, and how ATF4 and XBP1s promote the nuclear translocation and functional activities of TFEB and TFE3 in A/A cells during ER stress.

Under ER stress conditions, A/A cells displayed abnormal phenotypes of LC3A/B-positive structures (such as few autophagic puncta and perinuclear accumulated small LC3A/B-positive structures) (Figure 2A, B). Although expression of autophagy genes was not changed as much by the activated ATF6 form as by the constitutively active TFEB mutant (TFEB^S211A-FLAG) (Figure 8A, B vs. Figure 9C, E), the activated ATF6 form prevented the abnormal phenotypes more strongly than the constitutively active TFEB mutant (Figure 8D–F vs. Figure 9F, H) in A/A cells during ER stress, indicating that the activated ATF6 form has additional roles in autophagy besides induction of nuclear translocation of TFEB and TFE3 in A/A cells. The ER is thought to be the central organelle for de novo lipid synthesis. De novo phosphatidylcholine synthesis is required for autophagosome membrane formation and maintenance during autophagy [93,94]. In addition, OE of the activated ATF6 form modulates enzymatic activities of the CDP-choline pathway and thereby enhances phosphatidylcholine biosynthesis and ER expansion [88] (Figure 8G). Therefore, OE of the activated ATF6 form may increase de novo lipid synthesis required for autophagosome formation in ER membranes, which may not occur or be insufficient in A/A cells during ER stress. However, accumulation of the fragmented ER in Tm-treated A/A^Hep cells was prevented by the activated ATF6 form (Figure 8G) but not by the constitutively active TFEB mutant (Figure 10A, C), indicating that the altered ER structures in Tm-treated A/A cells are not critical obstacles of autophagy pathways. Furthermore, extensive accumulation of the fragmented ER may indicate that lipid biosynthesis is increased in Tm-treated A/A cells. However, lipid biosynthesis may not provide specific or sufficient lipid species required for autophagosome formation during ER stress due to deficiency of the activated ATF6 form, indicating that A/A cells might require lipid biosynthesis mediated by the activated ATF6 form during ER stress. Further investigation is needed to define the molecular details of autophagy improvement by activated ATF6 form-mediated phosphatidylcholine synthesis in A/A cells during ER stress.

TFEB and TFE3 are members of the microphthalmia-associated TF family [25]. Recent studies demonstrated that they bind to overlapping sets of promoters [20,26,27] and are post-transcriptionally regulated through a similar mechanism [35–38,41]. Therefore, when overexpressed, TFEB can compensate for TFE3 deficiency and vice versa [94]. However, analysis of tfeb and tfe3-double KO (tfeb^−/−tfe3^−/−) mice revealed the deficiency of both TFE3 and TFE3 results in additive effects with an aggravating hepatic phenotype in a study using a high-fat diet [94]. In addition, a study of tfeb and tfe3-double KO immune cells showed that the reduction in expression of autophagy and immune genes is more pronounced in double KO cells than in single KO cells [95]. These reports indicate that TFEB and TFE3 may play a co-operative role in diverse cellular responses including autophagy. We report here that OE of the active TFEB mutant (TFEB^S211A-FLAG) rescued most autophagic defects of A/A cells such that autophagy was comparable with that in S/S cells under ER stress conditions (Figure 2 and 3 vs. Figure 9 and 10). However, perinuclear accumulation of LC3A/B was not completely prevented and the efficiency of LC3A/B-positive puncta formation was not fully restored in TFEB^S211A-FLAG-overexpressing A/A cells compared with S/S cells during ER stress (Figure 2A vs Figure 9F), suggesting that missing functions of activated TFE3 may be required to completely rescue all autophagy impairments in TFEB^S211A-FLAG-overexpressing A/A cells. This issue requires further investigation.

Under ER stress conditions, cells induce autophagy in addition to the UPR, to restore ER homeostasis by degrading unfolded and aggregated proteins [43–46]. Therefore, UPR pathways can directly and indirectly control autophagy through ER membrane localized proteins (such as ERN1 and ITPR [inositol 1,4,5-trisphosphate receptor]) and several UPR TFs (such as XBP1s, ATF4, DDIT3, and ATF6) [45,46]. Among these proteins, expression and activation of many TFs (such as XBP1s, ATF4, DDIT3, and ATF6) are regulated by EIF2S1 phosphorylation under ER stress conditions. Therefore, regulation of EIF2S1 phosphorylation may have a great impact on cellular homeostasis. Our findings suggest that fine-tuning of EIF2S1 phosphorylation can be a potential tool to treat rapidly growing tumors, which use both UPR and autophagy pathways to maintain cellular homeostasis.

Materials and methods

Reagents and antibodies

The information regarding reagents and antibodies is provided in Tables 1 and 2.

Table 1. List of reagents used in this study.

Items	Company	Catalog number
4’,6-diamidino-2-phenylindole dihydrochloride (DAPI)	Invitrogen	D1306
Bafilomycin A1 (Baf A1)	Cayman Chemical	11,038
CaCl2	Junsei Chemical	18,230–1201
Cycloheximide (CHX)	Sigma-Aldrich	C7698
D(+)-glucose	Junsei Chemical	64,220–001
Dithiothreitol (DTT)	Promega	V3151
FK506	InvivoGen	Tlrl-FK5
Hoechst 33,258	Sigma-Aldrich	94,403
KCl	USB Corporation	20,598
Leptomycin B (LMB)	Santa Cruz Biotechnology	sc-358,688
LPS from Escherichia coli O55:B5	Sigma-Aldrich	L2880
LysoTracker™ Red DND-99	Thermo Fisher Scientific	L7528
MG132	CSNpharm	CSN11436
MgSO4	Sigma-Aldrich	M7506
NaCl	Biosesang	SR1009-250-00
NaH2PO4	Sigma-Aldrich	8282
NaHCO3	Sigma-Aldrich	S6014
PP242	Sigma-Aldrich	P0037
Thapsigargin (Tg)	Merck Millipore	586,005
Torin 1	Tocris Bioscience	4247
Torin 2	Tocris Bioscience	4248
Tunicamycin (Tm)	Sigma-Aldrich	654,380

Table 2. List of antibodies used in this study.

Items	Company	Catalog number	Usage
anti-ACTB	Sigma-Aldrich	A5441	WB
anti-ATF4	Proteintech	10,835-1-AP	WB
anti-ATF6	Proteintech	24,169-1-AP	WB
anti-ATF6B	BioLegend	853,201	WB
anti-CTSB	R&D Systems	AF965	WB
anti-CTSL	R&D Systems	AF1515	WB
anti-DDIT3/CHOP	Cell Signaling Technology	2895S	WB
anti-EIF2AK3	Cell Signaling Technology	3192S	WB
anti-EIF2S1(D-3)	Santa Cruz Biotechnology	sc-133,132	WB
anti-EIF4EBP1	Cell Signaling Technology	9644S	WB
anti-ERN1	Cell Signaling Technology	3294S	WB
anti-FLAG	Sigma-Aldrich	F1804	WB
anti-GFP	Clontech Laboratories	632,381	WB
anti-GFP	Invitrogen	A-11122	IP
anti-GFP (B-2)	Santa Cruz Biotechnology	sc-9996	WB
anti-GSK3B	Cell Signaling Technology	9832S	IF
anti-HA	Santa Cruz Biotechnology	sc-7392	WB
anti-HA.11 epitope tag	BioLegend	901,514	WB, IF, IP
anti-Histone H3	Abcam	ab1791	WB
anti-KDEL	Enzo Life Sciences	ADI-SPA-827	WB, IF
anti-LAMP1	Developmental Studies Hybridoma Bank	1D4B-C	WB, IF
anti-LAMP2	Developmental Studies Hybridoma Bank	ABL-93-C	WB
anti-LC3A/B	Cell Signaling Technology	4108S	IF
anti-LC3B	Sigma-Aldrich	L7543	WB
anti-LMNA	Santa Cruz Biotechnology	sc-6215	WB
anti-MTOR	Cell Signaling Technology	2983S	WB
anti-pan YWHA	Santa Cruz Biotechnology	sc-133,232	WB
anti-phospho-EIF2S1	Abcam	ab32157	WB
anti-phospho-EIF4EBP1 (Thr37/46)	Cell Signaling Technology	2855S	WB
anti-phospho-Ser2448-MTOR	Cell Signaling Technology	2971S	WB
anti-phospho-TFEB (S211) (E9S8N)	Cell Signaling Technology	37681S	WB
anti-phospho-Thr389-RPS6KB1	Cell Signaling Technology	9206S	WB
anti-phospho-YWHA binding motif	Cell Signaling Technology	9601S	WB
anti-RPS6KB1	Cell Signaling Technology	2971S	WB
anti-SERPINA1/α1-antitrypsin	Dako	A0012	WB
anti-SQSTM1	Abnova	H00008878-M01	WB, IF
anti-TFE3	Pro. Hiderou Yoshida (University of Hyogo, Japan)	Home-made	IF (HeLa cell)
anti-TFE3	Sigma-Aldrich	HPA023881	WB, IF
anti-TFEB	Proteintech	13,372-1-AP	WB
anti-TFEB	Bethyl Laboratories	A303-673A	WB, IF
anti-TFEB	MyBioSource	MBS9125929	IF
anti-TFEB-phospho-Ser142	Merck Millipore	ABE1971	WB
anti-TUBA	Sigma-Aldrich	T5168	WB
anti-XBP1s	BioLegend	619,502	WB
Alexa Fluor 488-conjugated goat anti-rabbit	Invitrogen	A-11034	IF
Alexa Fluor 594-conjugated AffiniPure F(ab’)2 fragment donkey anti-rabbit IgG (H + L)	Jackson ImmunoResearch Laboratories	711–586-152	IF
Alexa Fluor 594-conjugated AffiniPure F(ab’)2 fragment goat anti-mouse IgG (H + L)	Jackson ImmunoResearch Laboratories	115–586-003	IF
Alexa Fluor 594-conjugated donkey anti-rat	Invitrogen	A-21209	IF
Alexa Fluor 647-conjugated AffiniPure F(ab’)2 fragment goat anti-mouse IgG (H + L)	Jackson ImmunoResearch Laboratories	115–606-146	IF
Alexa Fluor 647-conjugated goat anti-rat	Invitrogen	A-21247	IF
Peroxidase-conjugated AffiniPure F(ab’)2 fragment donkey anti-mouse IgG (H + L)	Jackson ImmunoResearch Laboratories	751–036-151	WB
Peroxidase-conjugated AffiniPure goat anti-rabbit IgG (H + L)	Jackson ImmunoResearch Laboratories	111–035-003	WB
Peroxidase-conjugated AffiniPure goat anti-rat IgG (H + L)	Jackson ImmunoResearch Laboratories	112–035-003	WB
Peroxidase-conjugated AffiniPure rabbit anti-goat IgG (H + L)	Jackson ImmunoResearch Laboratories	305–035-003	WB

Expression vectors

The PCR primer pairs for plasmid cloning are listed in Table 3. To construct a lentiviral vector (pLenti-gATF6B-CRISPR-Bla) expressing a guide RNA (gRNA) sequence targeting ATF6B, lentiCRISPRv2, a gift from Feng Zhang (Addgene, 52,961), was used. The targeting sequence for ATF6B was CGACAACCTGCTGAGTCCGG. This sequence was cloned into the lentiCRISPRv2 plasmid to generate pLenti-gATF6B-CRISPR-Puro. Next, pLenti-gATF6B-CRISPR-Bla was constructed by replacing the puromycin N-acetyltransferase gene of pLenti-gATF6B-CRISPR-Puro with the blasticidin S deaminase gene of pLUB-IRES-Bla [70]. The cDNA fragment encoding blasticidin S deaminase was amplified from the pLUB-IRES-Bla vector via PCR. The PCR product treated with BamHI and MluI was inserted into pLenti-gATF6B-CRISPR-Puro treated with the same restriction enzymes to construct pLenti-gATF6B-CRISPR-Bla.

Table 3. List of PCR primers used in this study.

Genes	Primer sequences (5’ to 3’)		Usage	Related plasmids
3xFLAG-EIF2S1(S51A)	Forward	AAAGGGTTCGAAATGGACTACAAAGACCATGACG	PCR	pLUB-3xFlag-EIF2S1^S51A-IRES-Puro
	Reverse	TTTGGGGGATCCTTAATCTTCAGCTTTGGCTTCC
Atf6b	Forward	CACCGCGACAACCTGCTGAGTCCGG	gRNA	pLenti-gATF6b-CRISPR-Puro
	Reverse	AAACCCGGACTCAGCAGGTTGTCGC
Blasticidin S deaminase	Forward	GGGAAAGGATCCGGCGCAACAAAC	PCR	pLenti-gATF6b-CRISPR-Bla
	Reverse	GGGAAAACGCGTTTAGCCCTCCCACACATAACCAG
DsRed2	Forward	TTTTTTCCACAACCATGGCCTCCTCCGAGAACG	PCR	pCGN-IRES-DsRed2
	Reverse	TTTTTTCGGCCGTACAGGAACAGGTGGTGGCGG
EIF2S1^S51A	Forward	TTTCTCGGTACCACCATGCCGGGGCTAAG	PCR	pShuttle-CMV-EIF2S1^S51A
	Reverse	TTTATCCTCGAGCGTTAATCTTCAGCTTTGGC
EIF2S1^S51A	Forward	TTTTTTAAGCTTCCGGGGCTAAGTTGTAGATT	PCR	p3xFlag-CMV-EIF2S1^S51A
	Reverse	TTTAAAGAATTCTTAATCTTCAGCTTTGGCTTCC
EIF2S1[WT]	Forward	AAATTTAAGCTTATGCCGGGGCTAAGTTGTAG	PCR	pShuttle-CMV-EIF2S1[WT]
	Reverse	AAATTTGAATTCTTAATCTTCAGCTTTGGCTTCC
Puromycin N-acetyltransferase	Forward	TTTAAACCACAACCATGGCCGAGTACAAGCCC	PCR	pLUB-IRES-Puro
	Reverse	TTTAAAGCTTAGCTCAGGCACCGGGCTTGCG
TFEB-3xFLAG	Forward	TTAGTAAGATCTCGAGCTCAAGCTTG	PCR	pShuttle-CMV-TFEB-3xFlag
	Reverse	TTAGTAGCGGCCGCCTACTTGTCATCGTCATCCTT

The plasmid (pcDNA3.1-α1-antitrypsin mutant Z [ATZ]) expressing human SERPINA1/ATZ carrying a missense mutation (substitution of lysine for glutamate at amino acid 342) was provided by Professor Randal J. Kaufman (Degenerative Diseases Program, Sanford Burnham Prebys Medical Discovery Institute, La Jolla, CA, USA).

The pGL4.14–5XCLEAR-Firefly luciferase plasmid (named 5xCLEAR luciferase reporter) used to measure the transcriptional activity of TFEB was constructed by inserting the nucleotide sequence of 5XCLEAR-CMVmini-TATA into the NheI- and XhoI-digested pGL4.14 vector (Promega, E669A). The nucleotide sequence of 5XCLEAR-CMVmini-TATA was GCTAGCCCGGCCACGTGGCCGCAGGGTCACGTGACCCTGCGCACCAGGTGGTGCTGCCCGTCACCTGACGGTGCGGCTCAGCTGAGCCCCGTAGGCGTGTACGGTGGGAGGCCTATATAAGCAGAGCTCGTTTAGTGAACCGTCAGATCGCCTGGACTCGAG (underlined sequences indicate the five CLEAR motifs and one TATA sequence).

To generate WT EIF2S1- and EGFP-expressing lentiviral vectors, pLEF-EIF2S1[WT]-IRES-Bla and pLEF-EGFP-IRES-Bla plasmids were constructed, respectively. pLEF-IRES-Bla was constructed by replacing the CMV promoter of pLVX-IRES-Bla [70] with the EEF1A1/EF-1α promoter of pEF-EGFP (Addgene, 11,154). The Ssp1-EcoRI fragment containing the EEF1A1 promoter from pEF-EGFP was inserted into pLVX-IRES-Bla treated with ClaI-Klenow-EcoRI. To construct pLEF-EIF2S1[WT]-IRES-Bla, the Eco47III-EcoRI fragment containing EIF2S1[WT] from pBabe-EIF2S1[WT] [98] was inserted into pLEF-IRES-Bla treated with BamH1-Klenow-EcoRI. To construct pLEF-EGFP-IRES-Bla, the NotI-Klenow-EcoRI fragment containing EGFP from pEGFP-N1 (Clontech, 6085–1) was inserted into pLEF-IRES-Bla treated with BamHI-Klenow-EcoRI.

To construct pLEF-HA-Cas9-IRES-Bla, the HA-Cas9 fragment from p3s-Cas9-HN [97] treated with SacI-T4 DNA polymerase was inserted into pLEF-IRES-Bla treated with SmaI.

To express EIF2S1^S51A (substitution of alanine for serine at amino acid 51), the 3xFLAG-fused EIF2S1^S51A-expressing pLUB-3xFlag-EIF2S1^S51A-IRES-Puro plasmid was constructed. First, pLUB-IRES-Puro was constructed by replacing the blasticidin S deaminase gene of pLUB-IRES-Bla [70] with the puromycin N-acetyltransferase gene of pLVX-AcGFP-N1 (Clontech, 632,154). The cDNA fragment encoding puromycin N-acetyltransferase was amplified from the pLVX-AcGFP-N1 vector via PCR. The PCR product treated with BstXI and BlpI was inserted into pLUB-IRES-Bla treated with the same restriction enzymes to construct pLUB-IRES-Puro. To generate the EIF2S1^S51A sequence of the pLUB-3xFlag-EIF2S1^S51A-IRES-Puro plasmid, pShuttle-CMV-EIF2S1^S51A was first constructed. The coding sequence of EIF2S1^S51A was amplified from pBabe-EIF2S1^S51A [98] via PCR. The PCR product treated with KpnI and XhoI was inserted into pShuttle-CMV (Addgene, 16,403) treated with the same restriction enzymes to construct pShuttle-CMV-EIF2S1^S51A. Next, the cDNA fragment encoding EIF2S1^S51A was transferred into p3xFlag-CMV-10 (Sigma-Aldrich, 32,190,102) to add an N-terminal 3xFLAG tag. The cDNA fragment encoding EIF2S1^S51A was amplified from pShuttle-CMV-EIF2S1^S51A via PCR. The PCR product treated with HindIII and EcoRI was inserted into p3xFlag-CMV treated with the same restriction enzymes to construct p3xFlag-CMV-EIF2S1^S51A. Finally, the cDNA fragment encoding 3xFLAG-EIF2S1^S51A was amplified from p3xFlag-CMV-EIF2S1^S51A via PCR. The PCR product treated with BstBI and BamHI was inserted into pLUB-IRES-Puro treated with the same restriction enzymes to construct the pLUB-3xFlag-EIF2S1^S51A-IRES-Puro plasmid.

To generate WT EIF2S1- and mutant EIF2S1^S51A-expressing adenoviral vectors, pShuttle-CMV-EIF2S1[WT] and pShuttle-CMV-EIF2S1^S51A plasmids were constructed, respectively. The construction strategy for the pShuttle-CMV-EIF2S1^S51A plasmid was described above. To construct pShuttle-CMV-EIF2S1[WT], the cDNA fragment encoding WT EIF2S1 was amplified from pBabe-EIF2S1[WT] via PCR. The PCR product treated with KpnI and XhoI was inserted into pShuttle-CMV treated with the same restriction enzymes to construct the pShuttle-CMV-EIF2S1[WT] plasmid.

The human TFEB fused with EGFP-expressing pLUB-TFEB-EGFP-IRES-Bla and EGFP-expressing pLUB-EGFP-IRES-Bla plasmids were described elsewhere [70].

To generate human C-terminally 3xFLAG-tagged WT TFEB- and mutant TFEB^S211A-expressing adenoviral vectors, pShuttle-CMV-TFEB-3xFlag and pShuttle-CMV-TFEB^S211A-3xFlag plasmids were constructed, respectively. pCMV-TFEB-3xFlag was constructed by inserting the cDNA fragment encoding human TFEB from pEGFP-N1-TFEB (Addgene, 38,119) treated with BglII and KpnI into p3xFlag-CMV-14 (Sigma-Aldrich, E7908) treated with the same restriction enzymes. To construct pShuttle-CMV-TFEB-3xFlag, the cDNA fragment encoding TFEB-3xFLAG was amplified from pCMV-TFEB-3xFlag via PCR. The PCR product treated with BglII and NotI was inserted into pShuttle-CMV treated with the same restriction enzymes to construct pShuttle-CMV-TFEB-3xFlag. pShuttle-CMV-TFEB^S211A-3xFlag was constructed by inserting the cDNA fragment encoding TFEB^S211A from pcDNA3.1-TFEB^S211A-MYC (Addgene, 805) treated with EcoRI-HindIII-Klenow into pShuttle-CMV-TFEB-3xFlag treated with BglII-SalI-Klenow.

To express UPR TFs (ATF4, XBP1s, ATF6[1–373] and ATF6B[1–393]), pCGN-vector, pCGN-ATF6[1–373], pCGN-IRES-DsRed2, pCGN-ATF4-IRES-DsRed2, pCGN-XBP1s-IRES-DsRed2, pCGN-ATF6[1-373]-IRES-DsRed2, and pCGN-ATF6B[1-393]-IRES-DsRed2 were used. The pCGN-ATF6[1–373] plasmid encoding the N-terminal domain of human ATF6 (aa 1–373) with an hemagglutinin (HA) epitope tag at the N-terminus was described previously [88]. To generate the pCGN-vector, the pCGN-ATF6[1–373] plasmid was treated with XbaI and BamHI to remove the ATF6 (aa 1–373) coding sequence, and the linearized empty vector was self-ligated. The internal ribosome entry site (IRES)-driven Discosoma Sp. red fluorescent protein (DsRed2)-expressing plasmid (pCGN-IRES-DsRed2) was constructed by replacing EGFP of pCGN-IRES-EGFP [88] with DsRed2 of pDsRed2-Nuc (Clontech, 632,408). To construct pCGN-IRES-DsRed2, the cDNA fragment encoding DsRed2 was amplified from pDsRed2-Nuc via PCR. The PCR product treated with BstXI and EagI was inserted into pCGN-IRES-EGFP treated with the same restriction enzymes to construct pCGN-IRES-DsRed2. pCGN-ATF4-IRES-DsRed2, pCGN-XBP1s-IRES-DsRed2, pCGN-ATF6[1-373]-IRES-DsRed2, and pCGN-ATF6B[1-393]-IRES-DsRed2 were constructed by replacing the IRES-EGFP of pCGN-ATF4-IRES-EGFP, pCGN-XBP1s-IRES-EGFP, pCGN-ATF6[1-373]-IRES-EGFP, and pCGN-ATF6B[1-393]-IRES-EGFP [88] with IRES-DsRed2 of pCGN-IRES-DsRed2, respectively. The IRES-DsRed2 fragment obtained from pCGN-IRES-DsRed2 treated with EagI-Klenow-SalI was inserted into vectors treated with the same enzymes.

To generate the HA-ATF6[1-373]-expressing adenoviral vector (named pShuttle-CMV-HA-ATF6[1–373]), the BamHI-Klenow-NdeI fragment containing CMV-HA-ATF6[1–373] from pCGN-ATF6[1–373] was inserted into pShuttle-CMV treated with XhoI-Klenow-NdeI. The ATF4- and FLAG-XBP1s-expressing adenoviral vectors (pAD-Track-ATF4 and pAD-Track-Flag-XBP1s) were described previously [99].

The HA-tagged YWHA-expressing plasmid (pcDNA3.1-HA-YWHA) was obtained from Professor Eek-Hoon Jho (Department of Life Science, University of Seoul, Seoul, Korea) [100].

Transfection and virus production

Cells were transfected with plasmids using Mirus Bio™ TransIT™-LT1 transfection reagent (Fisher Scientific, MIR2306) according to the manufacturer’s instructions for 24–36 h (as described in each figure legend).

Recombinant adenoviruses expressing WT EIF2S1, mutant EIF2S1^S51A, TFEB-3xFLAG, TFEB^S211A-3xFLAG, HA-ATF6[1–373], ATF4, or FLAG-XBP1s were generated using the AdEasy vector system according to the manufacturer’s instructions (Agilent Technologies, 240,009). In brief, BJ5183 cells were cotransformed with the shuttle vector (pShuttle-CMV-EIF2S1[WT], pShuttle-CMV-EIF2S1^S51A, pShuttle-CMV-TFEB-3xFlag, pShuttle-CMV-TFEB^S211A-3xFlag, pShuttle-CMV-HA-ATF6[1–373]), pAD-Track-ATF4, or pAD-Track-Flag-XBP1s and the viral DNA plasmid pAdEasy-1 to generate a recombinant adenoviral plasmid. Then, HEK-293A cells were transfected with the recombinant adenoviral plasmids using the calcium phosphate technique to produce viral particles, which were purified using CsCI (Sigma-Aldrich, 3032) gradient centrifugation. The viral titer was determined using an AdEasy Viral Titer Kit (Aligent Technologies, 972,500) according to the manufacturer’s instructions.

To produce lentiviral particles expressing EIF2S1[WT], EGFP, HA-Cas9, or 3xFLAG-EIF2S1^S51A proteins, or Atf6b-targeting gRNA, Lenti-X-293 T cells (Clontech Laboratories, 632,180) were cotransfected with each lentiviral construct (pLEF-EIF2S1[WT]-IRES-Bla, pLEF-EGFP-IRES-Bla, pLEF-HA-Cas9-IRES-Bla, pLUB-3xFlag-EIF2S1^S51A-IRES-Puro, or pLenti-gATF6B-CRISPR-Bla) and a third-generation lentiviral packaging system (pRSV-Rev, pMD2-VSVG, and pMDLg/pRRE plasmids) using Mirus Bio™ TransIT™-LT1 transfection reagent. On the third day after transfection, lentiviruses were collected from the supernatant of Lenti-X-293 T cells, diluted in complete medium containing 8 µg/mL polybrene, and infected into cells for 48 h.

Cell lines and cell culture

All cell lines were incubated at 37°C in a humidified incubator containing 5% CO₂. Immortalized hepatocytes (S/S^Hep, A/A^Hep, HEP-Atf6^+/+, and HEP-atf6^−/−) were cultured in Medium 199 (WelGENE, LM 006–01) supplemented with 10% fetal bovine serum (FBS; WelGENE, S 001–07) and 1% penicillin-streptomycin (WelGENE, LS 202–02) as previously described [96]. MEFs (S/S^MEF, A/A^MEF, Eif2ak3^+/+, eif2ak3^−/−, MEF-Atf6^+/+, and MEF-atf6^−/−) were grown in Dulbecco’s modified Eagle’s medium (DMEM; WelGENE, LM 001–05) supplemented with 10% FBS, 1% penicillin-streptomycin, 2% MEM amino acids (WelGENE, LS 004–01), and 1% MEM nonessential amino acids (WelGENE, LS 005–01) [53]. The HeLa cell line (Korean Cell Line Bank, 10,002) was cultured in MEM Alpha medium (Sigma-Aldrich, M0894) supplemented with 4.4 mg/mL sodium bicarbonate (Sigma-Aldrich, S6014), 10% FBS, and 1% penicillin-streptomycin. To induce starvation, cells were incubated in EBSS (1.8 mM CaCl₂, 5.3 mM KCl, 0.8 mM MgSO₄, 117 mM NaCl, 26 mM NaHCO₃, 1 mM NaH₂PO₄, and 5.6 mM D(+)-glucose) for the indicated durations.

Generation of TFEB-EGFP- or EGFP-expressing S/S^MEF and A/A^MEF stable cell lines (named S/S-EGFP, S/S-TFEB-EGFP, A/A-EGFP, and A/A-TFEB-EGFP) was previously described [70]. Briefly, S/S^MEF or A/A^MEF cells were infected with lentiviral particles containing the pLUB-EGFP-IRES-Bla or pLUB-TFEB-EGFP-IRES-Bla construct. Then, each stable cell line was isolated by blasticidin selection. Cells were maintained in DMEM supplemented with 10% FBS, 1% penicillin-streptomycin, 1% MEM nonessential amino acids, and 5 µg/mL blasticidin S HCl (Invitrogen, R21001).

EIF2S1 phosphorylation-deficient HeLa cells were generated using CRISPR-Cas9 genome editing technology. A HA-Cas9-expressing HeLa stable cell line (named HeLa-Cas9) was first generated by infecting HeLa cells with lentiviral particles containing pLEF-HA-Cas9-IRES-Bla. HA-Cas9-expressing HeLa stable cell lines were isolated by blasticidin selection (4 µg/mL [Invitrogen, R21001]). Among several HA-Cas9-positive clones identified, one stable cell line that showed high expression and nuclear localization of HA-Cas9 was chosen for subsequent experiments. Next, a HeLa-Cas9 cell line expressing 3xFLAG-EIF2S1^S51A was generated by infecting HeLa-Cas9 cells with lentiviral particles containing pLUB-3xFLAG-EIF2S1^S51A-IRES-Puro. An HA-Cas9- and 3xFLAG-EIF2S1^S51A-overexpressing HeLa stable cell line [named HeLa-Cas9 EIF2S1^S51A OE] was isolated by double selection with both blasticidin (2 µg/mL [Invitrogen, R21001]) and puromycin (5 µg/mL [Santa Cruz Biotechnology, sc-108,071]). Selection was confirmed by WB analysis using anti-FLAG (Sigma-Aldrich, F1804), anti-phospho-EIF2S1 (Abcam, ab32157), and anti-EIF2S1 (D-3) (Santa Cruz Biotechnology, sc-133,132) antibodies. Microscopic observation using an anti-FLAG antibody (Sigma-Aldrich, F1804) was conducted to check the cytosolic expression of 3xFLAG-EIF2S1^S51A in HeLa-Cas9 EIF2S1^S51A OE cells. Two CRISPR-Cas9 gRNA target sequences to delete exon 2 of Eif2s1 were identified bioinformatically using the CRISPR Design Tool available at http://www.rgenome.net/. The targeting sequences were CTCCAAGACCTAAGGATTAA for sgRNA1 and GGATCTTGATAATTGACTCA for sgRNA2. Custom-designed Alt-R® CRISPR-Cas9 crRNA (Integrated DNA Technologies, Inc.) and Alt-R® CRISPR-Cas9 tracrRNA (Integrated DNA Technologies, Inc., 1,072,532) were used to generate a functional gRNA duplex. HeLa-Cas9 EIF2S1^S51A cells were nucleofected with both sgRNA1 and sgRNA2 duplexes using a SE Cell Line 4D-Nucleofector™ X Kit S (Lonza, V4XC-1032) on a 4D-Nucleofector® X Unit (Lonza, AAF-1003X) with program CN-114 according to the manufacturer’s instructions. Nucleofected cells were seeded as single clones (one cell/well) in 96-well plates. After 3–4 weeks, clones were screened for expression of EIF2S1 and phosphorylated EIF2S1 proteins by WB analysis. Sequencing was performed to confirm that the selected clone [named HeLa-Cas9 EIF2S1^S51A OE EIF2S1 KO] expressed 3xFLAG-EIF2S1^S51A but not endogenous EIF2S1.

Both ATF6- and ATF6B-deficient MEFs (named MEF-atf6^−/−atf6b^−/−) were generated using CRISPR-Cas9 genome editing technology. A MEF-atf6^−/− cell line was infected with lentiviral particles containing pLenti-gATF6B-CRISPR-Bla. After incubation for 2 days, the infected cells were selected with blasticidin (5 µg/mL) for 3 days. Surviving cells were expanded as single colonies, and KO of ATF6B was confirmed via WB analysis.

WT EIF2S1- or EGFP-expressing A/A^MEF stable cell lines were generated by infecting A/A^MEF cells with lentiviral particles containing pLEF-EIF2S1[WT]-IRES-Bla or pLEF-EGFP-IRES-Bla, respectively. Infected A/A^MEF cells were cultured in medium containing blasticidin (5 µg/mL [Invitrogen, R21001]) to establish A/A^MEF-EIF2S1 or A/A^MEF-EGFP stable cell lines. They were maintained in DMEM (WelGENE, LM 001–05) supplemented with 10% FBS (WelGENE, S 001–07), 1% penicillin-streptomycin (WelGENE, LS 202–02), 1% MEM nonessential amino acids (WelGENE, LS 005–01), and 5 µg/mL blasticidin S HCl (Invitrogen, R21001).

Subcellular fractionation

Cells were grown in 100 mm cell culture dishes until they reached about 90% confluency and were then treated with dimethyl sulfoxide (DMSO) or Tm for the indicated durations. After harvesting cells, the nuclear and cytoplasmic fractions were separated using a previously described procedure [70]. The protein concentration was calculated, and the fractions underwent WB analysis.

Coimmunoprecipitation (co-IP) assay

S/S- and A/A-TFEB-EGFP MEFs were plated in 100 mm culture dishes at a density of 7 × 10⁵ cells/dish for longer than 16 h and treated with the specified chemicals for the indicated durations. In HA-ATF6[1–373] OE experiments, A/A-TFEB-EGFP MEFs were plated in 100 mm culture dishes at a density of 7 × 10⁵ cells/dish. The next day, cells were transfected with pCGN-vector or pCGN-ATF6[1–373] for 24 h and treated with DMSO or Tm (100 ng/mL) for 24 h. IP of TFEB-EGFP or HA-ATF6[1–373] from the transfected A/A-TFEB-EGFP MEFs were performed using an anti-GFP antibody (Invitrogen) or anti-HA antibody (BioLegend), respectively, as described previously [70].

Eif2ak3^+/+ and eif2ak3 KO (eif2ak3^−/−) MEFs were plated in 100 mm culture dishes at a density of 7 × 10⁵ cells/dish. The next day, cells were transfected with pLUB-TFEB-EGFP-IRES-Bla and pcDNA3.1-HA-YWHA for 24 h and treated with Tm (1 µg/mL) for 16 h. They were collected in complete growth medium and washed once with phosphate-buffered saline (PBS; 137 mM NaCl [Biosesang, SR1009-250-00], 2.7 mM KCl [USB Corporation, 20,598, discontinued], 1.8 mM KH₂PO₄ [Junsei, 84,185–0350], and 10 mM Na₂HPO₄ [FUJIFILM Wako Pure Chemical Corporation, 197–02865], pH 7.4). The pellets were dissolved in 300 µL IP lysis buffer (20 mM Tris-HCl, pH 7.5 [Biosesang, TR2016-050-75], 150 mM NaCl, 1% Triton X-100 [Sigma-Aldrich, T8787], 1 mM EDTA [Thermo Fisher Scientific, 1,861,275], 1 mM EGTA [BioShop Canada Inc., EDT 001], 2.5 mM sodium pyrophosphate [Sigma-Aldrich, P8010], 1 mM β-glycerophosphate [Sigma-Aldrich, G5422], and 1 mM sodium orthovanadate [Sigma-Aldrich, S6508]) supplemented with Half Protease Inhibitor Cocktail (Thermo Fisher Scientific, 1,861,279) at 1× final concentration. Cells were lysed by passing the samples through a 26 G needle ten times. Cell lysates were kept on ice for 30 min and centrifuged at 13,000 × g for 15 min at 4°C to collect soluble fractions. Then, 1.2 mg protein lysate and 2 µg/mL anti-GFP antibody (Invitrogen, A-11122) were diluted in 600 µL IP lysis buffer and rotated at 4°C for 6 h. The protein lysate-antibody complexes were transferred to 40 µL protein A/G agarose beads (Thermo Fisher Scientific, 20,423) (which had been cleaned with 700 µL IP lysis buffer containing 5% bovine serum albumin [BSA {Sigma-Aldrich, A7030}] overnight at 4°C) and incubated with rotation for an additional 2 h at 4°C. After incubation, the beads were washed five times with 1 mL IP lysis buffer. Samples were eluted in 40 µL of 2× sodium dodecyl sulfate (SDS) sample loading buffer (100 mM Tris-HCl [VWR Life Science, 0497], pH 6.8, 200 mM DTT [Promega, V3151], 4% SDS [Promega, H5114], 20% glycerol [USB Corporation, 16,374], and 0.2% bromophenol blue [Sigma-Aldrich, B-5525]); boiled at 100°C for 5 min; and separated by SDS-PAGE.

Western blot (WB) analysis

Cells were lysed in Nonidet P40 lysis buffer (1% IGEPAL CA-630 [NP40; Sigma-Aldrich, I8896], 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.05% SDS, 0.5 mM sodium orthovanadate, 100 mM NaF [Sigma-Aldrich, 201,154], 50 mM β-glycerophosphate, and Halt Protease Inhibitor Cocktail [Thermo Fisher Scientific, 78,437]). Cell lysates were centrifuged at 13,000 × g for 15 min at 4°C, and supernatants were collected. For WB analysis of ATF6 and ATF6B, the cells were treated with Tm for the indicated durations and then with MG132 (20 µM) for 1 h before harvesting samples. Cells were directly lysed in SDS lysis buffer (1% SDS, 50 mM Tris-Cl pH 7.5, 150 mM NaCl, 0.5 mM sodium orthovanadate, 100 mM NaF, and 50 mM β-glycerophosphate) supplemented with Halt Protease Inhibitor Cocktail (Thermo Fisher Scientific, 1,861,279). The lysates were immediately heated for 15 min at 100°C. The homogenates were centrifuged at 13,000 × g for 15 min at 4°C, and the supernatants were collected. Protein concentrations were determined using a Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific, 23,227). Cell lysates were subjected to WB analysis as described previously [70].

RNA isolation and quantitative real-time polymerase chain reaction

Total RNA was isolated from cells treated with or without Tm for the indicated durations using QIAzol Lysis reagent (QIAGEN, QI-79306). cDNA was synthesized with a High-Capacity cDNA RT Kit (Applied Biosystems, ABS-4368814). Quantitative PCR was performed with Luna® Universal qPCR Master Mix (New England BioLabs, M3003X) and a StepOnePlus Real Time System (Applied Biosystems, CA, USA). The specificity of each primer pair was confirmed by melting curve analysis. The levels of target mRNAs were normalized to that of Actb mRNA. The qPCR primer pairs used in this study are listed in Table 4.

Table 4. List of qPCR primers used in this study.

Genes	Species	Forward primer (5’ to 3’)	Reverse primer (5’ to 3’)	Usage
Actb	Mouse	GATCTGGCACCACACCTTCT	GGGGTGTTGAAGGTCTCAAA	qPCR
Asns	Mouse	TACAACCACAAGGCGCTACA	AAGGGCCTGACTCCATAGGT	qPCR
Atf4	Mouse	ATGGCCGGCTATGGATGAT	CGAAGTCAAACTCTTTCAGATCCATT	qPCR
Atf4	Human	ATGACCGAAATGAGCTTCCTG	GCTGGAGAACCCATGAGGT	qPCR
ATF6	Human	GCCTTTATTGCTTCCAGCAG	TGAGACAGCAAAACCGTCTG	qPCR
ATF6	Mouse	GCGGATGATAAAGAACCGAGAG	ACAGACAGCTCTTCGCTTTG	qPCR
Atp6v1h	Mouse	GGATGCTGCTGTCCCAACTAA	TCTCTTGCTTGTCCTCGGAAC	qPCR
Cathepsin B	Mouse	ACAGTGCCACACAGCTTCTTC	TCCTTGATCCTTCTTTCTTGCC	qPCR
Cathepsin D	Mouse	CTGAGTGGCTTCATGGGAAT	CCTGACAGTGGAGAAGGAGC	qPCR
Cathepsin L	Mouse	ATCAAACCTTTAGTGCAGAGTG	CTGTATTCCCCGTTGTGTAGC	qPCR
Cth	Mouse	TCTTGCTGCCACCATTACGA	GCCTCCATACACTTCATCCAT	qPCR
Ddit3/Chop	Mouse	CTGCCTTTCACCTTGGAGAC	CGTTTCCTGGGGATGAGATA	qPCR
Glb1	Mouse	AAATGGCTGGCAGTCCTTCTG	ACCTGCACGGTTATGATCGGT	qPCR
Hexb	Mouse	CTGGTGTCGCTAGTGTCGC	CAGGGCCATGATGTCTCTTGT	qPCR
Hspa5/BiP	Mouse	TCATCGGACGCACTTGGA	CAACCACCTTGAATGGCAAGA	qPCR
Lamp1	Mouse	ACCTGTCGAGTGGCAACTTCA	GGGCACAAGTGGTGGTGAG	qPCR
Lamp2a	Mouse	GCAGTGCAGATGAAGACAAC	AGTATGATGGCGCTTGAGAC	qPCR
Lamp2b	Mouse	GGTGCTGGTCTTTCAGGCTTGATT	ACCACCCAATCTAAGAGCAGGACT	qPCR
Lamp2c	Mouse	ATGTGCTGCTGACTCTGACCTCAA	TGGAAGCACGAGACTGGCTTGATT	qPCR
Lc3b	Mouse	CGTCCTGGACAAGACCAAGT	ACCATCTACAGGAAGCCGTC	qPCR
Mcoln1	Mouse	GCTGGGTTACTCTGATGGGTC	CCACCACGGACATAGGCATAC	qPCR
Ppp1r15a/Gadd34	Mouse	CCCGAGATTCCTCTAAAAGC	CCAGACAGCAAGGAAATGG	qPCR
Sqstm1/p62	Mouse	GCTGCCCTATACCCACATCT	CGCCTTCATCCGAGAAAC	qPCR
Tfe3	Mouse	CCTGAAGGCATCTGTGGATT	TGTAGGTCCAGAAGGGCATC	qPCR
Tfeb	Human	ACCTGTCCGAGACCTATGGG	CGTCCAGACGCATAATGTTGTC	qPCR
Tfeb	Mouse	CCTGCCGACCTGACTCAGA	CTCAATTAGGTTGTGATTGTCTTTCTTC	qPCR
Tpp1	Mouse	CCCCTCATGTGGATTTTGTGG	TGGTTCTGGACGTTGTCTTGG	qPCR
Uvrag	Mouse	CAAGCTGACAGAAAAGGAGCGAG	GGAAGAGTTTGCCTCAAGTCTGG	qPCR
Xbp1s	Mouse	GAGTCCGCAGCAGGTG	AGGCTTGGTGTATACATGG	qPCR
Xbp1s	Human	CCGCAGCAGGTGCAGG	GAGTCAATACCGCCAGAATCCA	qPCR
Xbp1t	Mouse	CCTGAGCCCGGAGGAGAA	CTGCACCTGCTGCGGAC	qPCR
Xbp1t	Human	GCAAGCGACAGCGCCT	TTTTCAGTTTCCTCCTCAGCG	qPCR

Dual luciferase assay

The 5XCLEAR luciferase assay was performed to representatively assess the activities of TFEB-regulated genes. S/S^MEF and A/A^MEF cells were cultured overnight in 6-well plates at a density of 6 × 10⁴ cells/dish. Both pGL4.14–5XCLEAR-Firefly luciferase (for 5XCLEAR motif-driven fire luciferase) and pRL-CMV (for CMV promoter-driven Renilla luciferase) plasmids were transfected using Mirus Bio™ TransIT™-LT1 transfection reagent (Fisher Scientific, MIR2306). CMV promoter-driven Renilla luciferase was used to normalize the transfection and expression efficiencies. If necessary, the other indicated constructs (pCGN vectors [pCGN-IRES-DsRed2, pCGN-ATF4-IRES-DsRed2, pCGN-XBP1s-IRES-DsRed2, pCGN-ATF6[1-373]-IRES-DsRed2, or pCGN-ATF6B[1-393]-IRES-DsRed2] or pShuttle vectors [pShuttle-CMV, pShuttle-CMV-TFEB-3xFlag, or pShuttle-CMV-TFEB^S211A-3xFlag]) were also cotransfected for 30 h according to the manufacturer’s instructions. After the chemical treatments, cells were washed once with PBS and harvested for luciferase assays using the Dual-Luciferase assay system (Promega, E1980) according to the manufacturer’s instructions. Chemiluminescent signals were measured using a Synergy HTX Multi-Mode Microplate Reader (Biotek Intrusments, Winooski, VT, USA). Firefly luciferase activity was normalized to Renilla luciferase activity for each sample. The presented data were replicated in at least three independent experiments.

Live cell imaging using confocal microscopy

Cells were plated on collagen-coated 35 mm glass bottom confocal dishes (SPL Life Science, 101,350) at a density of 1 × 10⁵ cells/dish. The next day, cells were treated with DMSO or Tm (1 µg/mL) in phenol-red free M199 culture medium (GIBCO, 11,043,023) for the indicated durations. In HA-ATF6[1–373], ATF4, FLAG-XBP1s, or TFEB^S211A-FLAG OE experiments, cells were infected with the indicated recombinant adenoviruses (Ad-vector, Ad-HA-ATF6[1–373], Ad-ATF4 EGFP, Ad-Flag-XBP1s EGFP, or Ad-TFEB^S211A-Flag) for 24 h before Tm treatment. During the last 30 min of the chemical treatment, the cell culture medium was supplemented with LysoTracker Red DND-99 (100 nM) and Hoechst 33,258 (10 µg/mL) to stain lysosomes and nuclei, respectively. Live cell imaging was performed using an FV1200-OSR confocal microscope (Olympus, Shinjuku, Japan). The intensity of LysoTracker Red staining was measured using the mean fluorescence intensity (MFI) tool of FV10-ASW-4.2 software (Olympus).

Immunofluorescence (IF) staining

Cells were plated on collagen-coated glass coverslips in 6-well dishes and cultured overnight. In experiments overexpressing specific proteins, cells were transfected with the indicated plasmids (pCGN-IRES-DsRed2, pCGN-ATF4-IRES-DsRed2, pCGN-XBP1s-IRES-DsRed2, pCGN-ATF6[1-373]-IRES-DsRed2, or pCGN-ATF6B[1-393]-IRES-DsRed2) or infected with the indicated recombinant adenoviruses (Ad-vector, Ad-HA-ATF6[1–373], Ad-ATF4 EGFP, Ad-Flag-XBP1s/EGFP, Ad-EIF2S1[WT], Ad-EIF2S1^S51A, Ad-TFEB[WT]-Flag, or Ad-TFEB^S211A-Flag) for 24 h before Tm treatment. Cells were treated with the indicated chemicals for the indicated durations, rinsed twice with PBS, fixed with 4% paraformaldehyde diluted in PBS for 15 min, and permeabilized with 0.1% Triton X-100 (Sigma-Aldrich, T8787) diluted in PBS for 5 min. To visualize LC3A/B, SQSTM1, and LAMP1 puncta, cells on coverslips were fixed with 100% methanol (SK Chemical, L260-18) for 10 min at −20°C, washed twice with PBS, blocked with 3% BSA (Sigma-Aldrich, A7030) diluted in PBS for 1 h, and incubated with the indicated primary antibodies (labeled as “IF” in the antibody description section) overnight at 4°C. Cells were incubated with fluorescence-conjugated secondary antibodies for 1 h at room temperature. Nuclei were stained with DAPI (Invitrogen, D1306). Finally, coverslips were mounted using ProLong Gold mounting medium (Invitrogen, P36930). Cells were observed by confocal laser microscopy using a FV1200-OSR confocal microscope (Olympus). Images in colocalization experiments of HA-ATF6[1–373] and TFEB-EGFP were obtained using Airyscan super-resolution mode with a Zeiss LSM-780 inverted confocal laser scanning microscope (Carl Zeiss Microscopy, Germany) using a Plan-Apochromat 100×/1.46 oil immersion objective lens, and were processed and analyzed with ZEN 2 (Carl Zeiss Microscopy). Colocalization of LC3A/B and SQSTM1, SQSTM1 and LAMP1, LC3A/B and LAMP1 was measured using the Pearson’s Correlation Coefficient calculator tool of FV10-ASW-4.2 software (Olympus).

Transmission electron microscopy (TEM) analysis

Cells were seeded in 100 mm culture dishes at a density of 7 × 10⁴ cells/dish, cultured for at least 16 h, and then treated with Tm (1 µg/mL) for 24 h. In experiments overexpressing HA-ATF6[1–373] or TFEB^S211A-FLAG, cells were infected with the indicated recombinant adenoviruses (Ad-vector, Ad-HA-ATF6[1–373], or Ad-TFEB^S211A-Flag) for 24 h before Tm treatment. Cells on dishes were washed twice with 0.1 M phosphate buffer (0.02 M NaH₂PO₄ [Sigma-Aldrich, S8282] and 0.08 M Na₂HPO₄ [FUJIFILM Wako Pure Chemical Corporation, 197–02865], pH 7.4), and were fixed by immersion in 2.5% glutaraldehyde (Electron Microscopy Sciences, 16,220) diluted in 0.1 M phosphate buffer for 2 h at room temperature. Cells were rinsed twice with 0.1 M phosphate buffer and postfixed with 1% osmium tetroxide (Sigma-Aldrich, 75,632) diluted in 0.1 M phosphate buffer for 1 h at 4°C. Next, samples were dehydrated with a series of graded ethyl alcohol solution (Merck Millipore, 1.00983) and then with acetone (Fisher Scientific, A18-4). The samples were next embedded in EPON 812. Ultrathin sections (70–80 nm) were obtained using an ultramicrotome (Leica Ultracut UCT, Wetzlar, Germany), costained with uranyl acetate (Fisher Scientific, NC1375332) and lead citrate (Fisher Scientific, NC1588038), and examined using a transmission electron microscope (JEM-1010; JEOL, Tokyo, Japan) at 60 kV.

Measurement of intracellular Ca²⁺ concentrations

MEFs were plated on 0.1% gelatin-coated microscope cover glasses (Paul Marienfeld GmbH & Co. KG, 0111550) at a density of 7 × 10⁴ cells/dish and cultured overnight. Cells were loaded with 1 µM Fura-2/AM (Thermo Fisher Scientific, F1221) in DMEM at 37°C for 30 min. Ratiometric Ca²⁺ imaging was performed at 340 and 380 nm in 2 mM Ca²⁺ Tyrode’s solution (129 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 30 mM glucose, and 25 mM HEPES) containing Tm (10 µg/mL) using an IDX81 fluorescence microscope (Olympus) equipped with an Olympus 40× oil objective lens (NA 1.30), a fluorescent arc lamp (Sutter Instrument, LAMBDA LS), an excitation filter wheel (Sutter Instrument, LAMBDA 10–2), a stage controller (Applied Scientific, MS-2000), and a CCD camera (Hamamatsu, C10600) at room temperature. Images were acquired for 5 min with a time interval of 4 s. Fluorescence intensity profiles were processed with MetaMorph (Molecular Devices, San Jose, CA, USA) and analyzed with Igor software (WaveMetrics, Portland, OR, USA).

Proximity ligation assay (PLA)

A/A-TFEB-EGFP MEFs were plated on collagen-coated glass coverslips in 6-well dishes at a density of 1 × 10⁵ cells/dish, cultured overnight, transfected with pCGN-vector or pCGN-ATF6[1–373] for 30 h, and treated with Tm (100 ng/mL) for 16 h. Thereafter, cells were rinsed twice with PBS, fixed with 3.5% paraformaldehyde diluted in PBS for 15 min, and permeabilized with 0.1% Triton X-100 diluted in PBS for 5 min. Finally, cells were blocked with 3% BSA diluted in PBS for 1 h and incubated with primary antibodies (anti-GFP [Invitrogen, A-11122] and anti-HA [Santa Cruz Biotechnology, sc-7392]) overnight at 4°C. The PLA was performed using a Duolink In Situ Red Starter Kit (Sigma-Aldrich, DUO92101) according to the manufacturer’s protocol. Images were obtained using an FV1200-OSR confocal microscope (Olympus). Cells were classified into three groups, namely, those with PLA-positive signals in the nucleus, the nucleus and cytosol, or the cytosol. The ratio of the MFI in the nucleus to that in the cytosol was quantified using CellProfiler software (https://cellprofiler.org/, Broad institute, USA). This ratio was ≥1.2, ≤0.7 and <1.2, and <0.7 in the nucleus, nucleus and cytosol, and cytosol groups, respectively. The PLA signal in the nucleus was measured using the MFI tool of FV10-ASW-4.2 software (Olympus).

Statistical analysis

All data are presented as mean ± standard error of the mean (SEM). All experiments were performed at least three times as indicated. Statistical analyses of all data were performed using the Student’s t-test, a one-way ANOVA, or a two-way ANOVA as indicated in the figure legends using GraphPad Prism 8.4.3 (GraphPad Software, San Diego, CA, USA). Statistical significance is indicated in the figures (*^,#,&p < 0.05, **^,##,&&p < 0.01, ***^,###,&&&p < 0.001).

Acknowledgments

We thank Randal J. Kaufman (Sanford Burnham Prebys Medical Discovery Institute) for providing cells (S/S^MEF, S/S^Hep, A/A^MEF, A/A^Hep, Eif2ak3^+/+, eif2ak3^−/−, HEP-Atf6^+/+, HEP-atf6^−/−, MEF-Atf6^+/+, and MEF-atf6α^−/−) and a plasmid (pcDNA3.1-α1-antitrypsin mutant Z [ATZ]), and the UNIST-Olympus Biomed Imaging Center for their expertise and assistance with using the Zeiss LSM 780 inverted confocal laser scanning microscope.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2023.2173900

Additional information

Funding

This work was funded by the Basic Science Research Program (2017R1D1A1B03028229, 2020R1F1A1066088, and 2022R1A2C1010449 to S.H.B and 2019R1A2C2002235 to C.Y.P), the Bio and Medical Technology Development Program (2017M3A9G7072745 to S.H.B), and the Priority Research Centers Program (2014R1A6A1030318 to S.H.B) of the National Research Foundation of Korea, which is funded by the Korean government.