The gene for the lysosomal protein LAMP3 is a direct target of the transcription factor ATF4

Selection and qPCR screening of lysosomal genes not regulated by TFEB

To begin the investigation into lysosomal genes that are not regulated by TFEB, 16 genes were selected from the Human Lysosome Gene Database (22), on the basis that these genes had no CLEAR motif within 2000 bp upstream or downstream of the transcriptional start site (TSS) and no other evidence of regulation by TFEB. The genes chosen were as follows: AGA, CREG1, CTBS, ENTPD4, GM2A, GNPTAB, LAMP2, LAMP3, LITAF, LMBRD1, NCSTN, OSTM1, PCYOX1, PPT2, SIAE, and TMEM92.

Oligonucleotide primers overlapping adjacent exons were designed for each (Table 1 and Table S1), allowing for specific amplification of mRNA. Total RNA was then extracted from HeLa cells and used as a template for cDNA preparation and qualitative PCR, allowing for product size and specificity to be determined by gel electrophoresis prior to subsequent qPCR (data not shown).

To investigate the cellular stress-induced transactivation of these non-TFEB–regulated genes, human lung carcinoma A549 and cervical cancer HeLa cells were treated with brefeldin A (BFA), rapamycin, and the mTOR kinase inhibitor AZD8055 for 6 h and sucrose for 24 h. BFA induces the ISR by inhibiting ADP-ribosylation factors, leading to failure to recruit the Golgi membrane of coat protein β-COP (which acts as scaffold in the formation and budding of small membrane vesicles). This results in the collapse of the Golgi and its fusion with the ER, activating the UPR (23). Rapamycin inhibits mTORC1 by binding this complex together with the FK506-binding protein (24), thereby partially blocking its function. AZD8055 is an ATP-competitive inhibitor of both mTORC1 and mTORC2 (25). Sucrose induces lysosomal stress in HeLa and A549 cells (among others) due to both rapid uptake of sucrose into the lysosome, as well as the cells' inability to metabolize sucrose due to the absence of the enzyme invertase (26, 27). Sucrose treatment thus mimics the cellular pathology of lysosomal storage disorders, including TFEB activation and enhanced lysosomal biogenesis (2, 27). Collectively, these treatments allowed for an initial screening of the impact of the ISR, TFEB, and/or other mTORC1/2 controlled downstream effects or lysosomal stress on the regulation of each candidate gene. Ultimately, this indicated multiple classes of gene regulation; those regulated (potentially indirectly) by TFEB, the ISR, an unknown transcription factor, or by any combination of these (Figs. S1 and S2).

LAMP3 transcript levels increase upon BFA treatment

Total RNA was extracted from A549 and HeLa cells treated with BFA for 6 h, after which levels of transcript were analyzed by qPCR. CCAAT–enhancer-binding protein homologous protein (CHOP), a known target of ATF4 (28), was used as a positive control for induction of the ISR (Fig. 1). In response to BFA, appreciable increases were seen in the levels of LAMP3 mRNA (3.29-fold in HeLa and 2.45-fold, compared with DMSO, in A549 cells), although they did not attain statistical significance (p = 0.052 and p = 0.0698, respectively) (Fig. 1). No significant or large magnitude changes in LAMP3 mRNA were seen following treatment with either AZD8055 or rapamycin, supporting the conclusion that this lysosomal gene is not regulated through TFEB. In contrast, lipopolysaccharide-induced tumor necrosis factor (LITAF) did not appear to be regulated through either TFEB or the ISR, demonstrating that the regulation pattern seen for LAMP3 is not shared by all genes for lysosomal proteins.

To then examine whether the effect of BFA on LAMP3 mRNA levels was statistically significant and/or time-dependent, we examined LAMP3 mRNA levels after 2, 4, 8, and 24 h of BFA treatment. In both HeLa and A549 cell lines, CHOP, a target for ATF4 and thus a positive control for activation of the ISR, was increased from 2 h onward, and LAMP3 mRNA was also significantly up-regulated at all time points tested (Fig. 2). Similar preliminary data were observed with the following three additional genes (Fig. S3): CTBS, ENTPD4, and PPT2.

Finally, to confirm that TFEB activation was incapable of up-regulating LAMP3 mRNA, HeLa cells were transfected with a mammalian expression vector containing the cDNA for TFEB fused to the epitope for the FLAG antibody (2) or with empty vector, with or without subsequent treatment with AZD8055. Immunoblotting with antibodies to FLAG and TFEB phosphorylated on the mTORC1-regulated site at Ser-142 (pSer-142) collectively demonstrated the successful exogenous expression of TFEB, as well as the expected reduction in its phosphorylation upon treatment with AZD8055 (Fig. 3A). However, there was no detectable change in ATF4 protein subsequent to TFEB–FLAG expression and/or AZD8055 treatment. In addition, qPCR revealed that TFEB–FLAG caused a statistically significant increase in the mRNA levels for the known TFEB target LAMP1, which was, as anticipated, further increased by AZD8055 (Fig. 3B). In contrast, LAMP3 expression was decreased upon TFEB–FLAG expression when compared with the control, regardless of the presence or absence of AZD8055 (Fig. 3B). These data provide strong further evidence that LAMP3 is not a target of TFEB.

Implication of the phosphorylation of eIF2α in the regulation of LAMP3 transcription

The eIF2α/ATF4 pathway is central to the ISR. As mentioned above, the lysosome is an important site for degradation of misfolded protein during the UPR (12, 13), so we surmised that ATF4 might be involved in the transcriptional regulation of non-TFEB–regulated lysosomal genes, such as LAMP3.

To provide further evidence to support this idea, we employed compounds, in addition to BFA, that can promote eIF2α phosphorylation and downstream ATF4 activity. Thapsigargin (TPG) promotes the ISR by inhibiting the transfer of Ca²⁺ ions into the ER (29). eIF2α can also be phosphorylated by general control nondepressible 2 (GCN2). Rather than responding to ER stress, GCN2 is activated upon amino acid starvation, which is signaled through the accumulation of uncharged tRNAs, resulting in stimulation of its protein kinase catalytic domain (30, 31). Histidinol inhibits the charging of histidyl-tRNA, and thus it activates the GCN2–eIF2α–ATF4 axis (32). In doing so, it mimics amino acid starvation, without eliciting the other direct consequences of amino acid depletion such as inhibition of mTORC1.

To assess whether TPG or histidinol treatments did indeed result in accumulation of ATF4 protein, A549 and HeLa cells were treated with each chemical for 24 h. Immunoblotting of extracts with antibodies directed to ATF4 demonstrated an increase in ATF4 protein expression in response to all three chemicals in both cell lines (Fig. 4A). To determine the effects of these agents on LAMP3 transcript levels, total RNA was used as a template for cDNA preparation and subsequent analysis by qPCR. In both lines, in addition to BFA, TPG, and histidinol, each resulted in statistically-significant increases in both LAMP3 mRNA, as well as the positive control, CHOP (Fig. 4B). However, the differences between the responses to these three different agents are noteworthy, particularly with respect to the lower ability of histidinol to up-regulate particularly CHOP and also (although to a lesser extent) LAMP3 mRNA when compared with BFA and TPG. One reason for this is that components of the ISR, including ATF4, PERK, and CHOP, are also targets of the transcription factor XBP1, an additional mediator of the UPR whose activation is in turn regulated upon splicing of its message by IRE1 in the ER (33). Furthermore, ATF6 (the third arm of the UPR) has also been shown to induce expression of CHOP during ER stress (34). Indeed, in both HeLa and A549 cells, BFA and TPG, but not histidinol, caused an increase in BiP, an event that can occur due to the activation of any of the three arms of the UPR (Fig. S4A). This was further evidenced by the altered splicing of XBP1 message (i.e. activation of XBP1) under these conditions (Fig. S4B). Therefore, BFA and TPG would be expected to cause a stronger ISR than histidinol. Nonetheless, these data demonstrate that multiple chemical inducers of ISR can up-regulate the transcript levels of LAMP3.

To further investigate the effect of eIF2α phosphorylation on specific genes, we used the compound ISRIB, which attenuates the inhibition of eIF2B caused by phosphorylated eIF2 (35) and thus allows assessment of the effect of eIF2α phosphorylation on gene regulation. To confirm that ISRIB inhibited the ISR in A549 and HeLa cells in our hands, each cell line was treated with BFA or BFA + ISRIB for 6 h, and extracts were then analyzed by immunoblotting. Both cell lines showed a marked induction of ATF4 protein expression in response to BFA, and as expected, this was reduced by ISRIB (Fig. 5A).

LAMP3 transcript levels were measured in response to treatment with BFA and/or ISRIB. HeLa and A549 cells were treated with BFA for 16 h, with or without pretreatment with ISRIB for 1 h. Total RNA was used as a template for cDNA preparation and subsequent analysis by qPCR. In each line, the 1-h pretreatment of ISRIB before a 16-h treatment with BFA resulted in a statistically-significant decrease in LAMP3 mRNA levels, compared with just BFA treatment (Fig. 5B). This provides further evidence that, in A549 and HeLa cells, the regulation of LAMP3 transcript levels involves the arm of the UPR that is mediated through phosphorylation of eIF2α. However, this was not the case for CTBS, ENTPD4, and PPT2 (Fig. S5).

LAMP3 transactivation is perturbed by manipulating ATF4 expression

To determine the role of ATF4 itself (rather than any other component of the ISR) in regulating LAMP3 mRNA levels, HeLa cells were transfected with either scrambled siRNA or one directed toward ATF4 for 72 h, during the final 24 h of which the medium was supplemented with DMSO or BFA. Immunoblotting demonstrated a marked, statistically-significant decrease in BFA-induced levels of the ATF4 protein (Fig. 6, A and B) and mRNA (Fig. 6C) levels. Furthermore, we evaluated by qPCR the transcript levels of known direct ATF4 targets as well as LAMP3 in identically-treated cells. Again, there was a drastic, statistically-significant decrease in both basal and BFA-induced LAMP3 mRNA levels upon siRNA-mediated knockdown of ATF4 (Fig. 6C). Interestingly, this was not the case with CHOP, although the levels of an additional known ATF4 target, asparagine synthetase (ASNS) (36), were reduced (Fig. 6C). In contrast, this was not the case for the three other putative targets of ATF4 (CTBS, ENTPD4, and PPT2) uncovered in our original screen (Fig. S6). Taken together, it is possible that these genes are not targets of ATF4 and/or their expression can also be mediated by other stress-inducible transcription factors (such as XBP1 or ATF6).

This indicates that ATF4 is required for the induction of LAMP3 in response to BFA. To assess whether ATF4 suffices to drive its expression, we assessed the ability of exogenously expressed ATF4 to increase LAMP3. To that end, HeLa cells were transfected with a mammalian expression plasmid containing the cDNA for ATF4 or empty vector. After 24 h, immunoblotting or qPCR, respectively, demonstrated a substantial increase in ATF4 mRNA and protein when compared with the control (Fig. 7, A and B). Importantly, mRNA levels for LAMP3 (and ASNS, the positive control) were up-regulated in a statistically significantly manner upon exogenous expression of ATF4 protein (Fig. 7B).

ATF4 associates with a binding site on LAMP3 upon BFA treatment

A collection of ChIP-seq data available from the Gene Transcription Regulation Database (37) revealed a potential ATF4-binding site within 1600 bp of the TSS of LAMP3, specifically ACATCTGATGCAAGGAAAA (reverse complement of TTTTCCTTGCATCAGATGT). The ATF4-binding consensus sequence has been reported as (G/A/C)TT(G/A/T)C(G/A)TCA (38), which matches the ChIP-seq data.

To investigate whether BFA induced the binding of ATF4 to this site, oligonucleotide primers were designed to flank this sequence near the LAMP3 TSS, as well as a negative control (a region >1000 bp upstream of the site) (Table 2). The ATF4-binding site involved in the regulation of ASNS was used as a positive control (36). HeLa cells were treated with DMSO or BFA for 6 h, prior to cross-linking of DNA and DNA-bound proteins with formaldehyde. Subsequent ChIP revealed significant binding of ATF4 near the TSS of LAMP3 when comparing the BFA treatment to the DMSO control (Fig. 8).

BFA treatment increases the levels of a reporter, including the putative LAMP3 ATF4-binding site

The amplification after ChIP of a 118-bp segment of DNA near the LAMP3 TSS confirmed the presence of an ATF4-binding site. To show conclusively that the site was involved in BFA-induced LAMP3 up-regulation, the WT putative ATF4-binding site (termed “WT L3”) was inserted into the firefly luciferase reporter plasmid pGL3-promoter (Promega) to determine the impact of the binding site on this reporter (Fig. 9A). In addition, two point mutations (MT1 L3 and MT2 L3) within the putative enhancer were separately introduced into the same reporter construct, in order to disrupt the consensus ATF4-binding sequence.

WT and mutant reporter plasmids were then transfected into HeLa cells. The reporter pRL-TK encoding Renilla luciferase was co-transfected as a control. In control cells (treated with DMSO), inclusion of WT L3 upstream of firefly luciferase results in a statistically significant 1.26-fold increase of firefly/Renilla luciferase expression, compared with the parent pGL3-promoter vector. This increase was not seen for vectors containing the mutated inserts (Fig. 9B). BFA induced statistically significant increases in firefly luciferase expression from the vector containing the WT L3 of 1.82-fold compared with pGL3-pro vector and 1.89-fold compared with the vector with the WT L3 insert after DMSO treatment. Importantly, these increases were not observed for the mutated inserts (Fig. 9B). Thus, BFA treatment induces firefly luciferase expression from a vector containing the WT L3 element, when ATF4 levels are enhanced, but not from mutant variants.

Chemicals for cell treatments

AZD8055 (Selleck Chem), BFA (Sigma-Aldrich), rapamycin (Sigma-Aldrich), TPG (Sigma-Aldrich), and ISRIB (Sigma-Aldrich) were each dissolved in DMSO, whereas histidinol dihydrochloride (Sigma-Aldrich) and sucrose (Sigma-Aldrich) were prepared in Milli Q water.

Cell culture

A549 cells, derived from cancerous lung tissue, and HeLa cells, derived from cervical cancer cells, were maintained in Dulbecco's modified Eagle's medium with 10% (v/v) fetal bovine serum (Life Technologies, Inc.), 100 units/ml penicillin/streptomycin (Sigma-Aldrich) and incubated at 37 °C and 5% (v/v) CO₂. For experimentation, cells were plated either on 100-mm dishes (56.7 cm²), 6-well plates (9.5 cm²), or 96-well plates (0.32 cm²). Cells were regularly tested for mycoplasma infection and discarded if affected.

Transient transfection

siRNAs to human ATF4 (Qiagen), pRK-ATF4 (47), and pTFEB–3xFLAG–CMV-10 (2) were introduced into HeLa cells using Lipofectamine 3000 (Life Technologies, Inc.) according to the manufacturer's instructions. pRK-ATF4 was a gift from Yihong Ye (Addgene plasmid 26114; RRID: Addgene_26114).

RNA extraction

Cells grown in 6-well plates were washed using 2 ml of PBS, and then 500 μl of TRIzol reagent (Thermo Fisher Scientific) was added. Total RNA was subsequently prepared according to the manufacturer's instructions and provided a template for cDNA preparation.

Qualitative RT-PCR

cDNA was first prepared from total RNA using the SuperScript III First Strand Synthesis System (Life Technologies, Inc.). Oligonucleotide primers were designed to amplify fragments of the coding sequences of the genes of interest (Table 1). 50-μl samples were prepared using cDNA (prepared with 6.25 ng of starting RNA), 100 nm forward and reverse primer, 1 unit of HotStarTaq DNA polymerase, 1× Q-solution, 1 mm MgCl₂, 200 μm dNTP mix, and milliQ water. Reactions were performed at 95 °C for 15 min, 35 cycles of 94 °C for 1 min, 60 °C for 1 min, 72 °C for 1 min, and then 72 °C for 10 min. Products were electrophoresed through a 2% agarose gel and visualized by ethidium bromide staining on a Gel Doc XR apparatus (Bio-Rad).

Quantitative real-time PCR (qPCR)

cDNA was first prepared from total RNA using the SuperScript III First Strand Synthesis System (Life Technologies, Inc.). For each well, 0.84× FAST SYBR Green Mix (Applied Biosystems), 5–10 μl of cDNA (6.25–12.5 ng of starting RNA), 200 nm forward and reverse primer, and milliQ water to 20 μl were added. qPCRs proceeded as follows on an ABI Step One Plus qPCR instrument (Applied Biosystems): 95 °C for 20 s; 40× (95 °C for 3 s; 60 °C for 30 s). The comparative threshold cycle protocol was employed to determine amounts of the target mRNA.

Immunoblotting

Cells grown in 6-well plates were washed using 2 ml of ice-cold PBS, then lysed with 100–300 μl of RIPA buffer supplemented with 2.5 mm Na₂H₂P₂O₇ (Sigma-Aldrich), 1 mm β-glycerophosphate (Sigma-Aldrich), 1 mm Na₃VO₄ (Sigma-Aldrich), and 1× Protease Inhibitor Mixture (Roche Applied Science). Cells were then rocked for 30 min at 4 °C, scraped, and transferred to a microcentrifuge tube. Cell suspensions were centrifuged at 16,000 × g for 30 min at 4 °C, and then the supernatant was retained and transferred to new microcentrifuge tubes. Protein concentration was quantified for sample normalization via the Lowry assay (48).

Equal amounts (by protein) of samples were subjected to SDS-PAGE and then transferred to a nitrocellulose membrane. The membrane was blocked for ∼90 min in PBS containing 0.1% Tween 20 and 2% (w/v) bovine serum albumin (BSA) at 4 °C. Membranes were then rolled overnight at 4 °C in the same solution supplemented with one of the following antibodies: 1:1000 rabbit monoclonal anti-ATF4 (Cell Signaling Technology); 1:1000 rabbit monoclonal anti-BiP (Cell Signaling Technology); 1:1500 mouse monoclonal anti-FLAG (Novus Biologicals); 1:1000 rabbit polyclonal anti-TFEB P-S142 (Millipore); or 1:10,000 mouse monoclonal anti-β-actin (Sigma-Aldrich). The following day, the membrane was washed three times for 5–10 min at room temperature in PBS containing 0.1% Tween 20, then rolled at room temperature for 1 h in PBS containing 0.1% Tween 20 and 2% BSA plus 1:20,000 goat anti-rabbit IgG DyLight 680 (Thermo Fisher Scientific) or goat anti-mouse IgG DyLight 800 4× PEG (Thermo Fisher Scientific) and imaged on an Odyssey CLx (LI-COR).

ChIP

Cells were grown for 2 days in 100-mm diameter dishes. To cross-link DNA and protein, cells were treated with 1% formaldehyde and rotated at room temperature for 10 min. Glycine was added at a final concentration of 125 mm, and samples were rotated for 5 min to quench the reaction. Cells were washed twice with PBS, scraped together with 0.5 ml of PBS + 1× protease inhibitor mixture into microcentrifuge tubes, and then pelleted by centrifugation at 4 °C at 1000 × g for 5 min. The supernatants were removed, and cells were resuspended and lysed for 10 min in 500 μl of ChIP lysis buffer (50 mm HEPES, pH 7.5, 140 mm NaCl, 1 mm EDTA, pH 8.0, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS) + 1× protease inhibitor mixture at 4 °C. Sonication time was optimized as 300 s (in 30-s bursts, followed by 30 s to cool on ice) by gel electrophoresis to ensure DNA fragment sizes of 150–900 bp. For analysis of sonicated samples created by ChIP, cell debris was pelleted by centrifugation for 10 min at 4 °C and 8000 × g. The supernatant was transferred into Eppendorf tubes and then diluted 1:2 with 500 μl of RIPA buffer. This sample was halved to allow probing with both ATF4 and normal rabbit IgG.

20 μl of suspended ChIP-grade protein G magnetic beads were added to each sample. Samples were then pre-cleared with 2 h of rotation at 4 °C. Samples were placed on a magnetic rack to pellet the beads, and solutions were isolated. 1% (5 μl) from each treatment/cell line was removed as an input control. Rabbit polyclonal ATF4 antibody (Cell Signaling Technology) was used at a 1:200 dilution, and 1 μl of normal rabbit IgG was added to one sample per cell line/treatment as a negative control. Samples were incubated for immunoprecipitation overnight at 4 °C with rotation. 20 μl of the protein G magnetic beads were then added to each sample for 2 h of incubation at 4 °C with rotation. The beads were pelleted, and the supernatant was discarded. The beads were washed with the addition of 1 ml of low-salt wash buffer and incubation at 4 °C three times at 5-min rotations, then repeated once using high-salt wash buffer. After the removal of the supernatant, DNA was eluted from beads with the addition of 150 μl of elution buffer per sample, including each specific 1% input sample. Chromatin was eluted by vortexing each sample at 1200 × g at 65 °C for 30 min. Beads were then pelleted, and chromatin samples were transferred to a new tube. Cross-linking was then reversed by addition of 6 μl of 5 m NaCl and 2 μl of proteinase K and vortexed at 1200 × g at 65 °C overnight.

DNA purification was achieved using the Qiagen QIAquick PCR purification kit and protocol, with elution in 50 μl of H₂O. In addition, primers were designed to flank this sequence near the LAMP3 TSS, as well as a negative control region >1000 bp upstream of the site (negative control) (Table 2). DNA analysis was performed as per the previously mentioned qPCR method. In a 96-well plate, reactions contained 1 μl of purified DNA, 1× FAST SYBR Green Mix (Applied Biosystems), 200 nm forward and reverse primers, and Milli Q water to a total of 20 μl.

Construction of luciferase reporter plasmids

Oligonucleotides were annealed in 10-μl reactions containing 20 μm of the upper and lower strand oligonucleotides (Table 3), 1× T4 DNA Ligation Buffer, and 1 μl (10 units) of T4 polynucleotide kinase with Milli Q water. The solution was heated to 37 °C for 30 min, then to 95 °C for 5 min, and then cooled to 25 °C at a speed of 5 °C per min. These were then subcloned into KpnI- and NheI-digested pGL3-pro plasmid (Promega), which contains an SV40 promoter upstream of a luciferase gene, and fidelity was confirmed via sequencing.

Dual-luciferase reporter assay

HeLa cells were grown in 96-well plates. 50 ng of pGL3-pro plasmid with or without the WT or mutated ATF4-binding sites of LAMP3, as well as 50 ng of pRL-TK (Promega), were transfected using Lipofectamine 3000 according to the manufacturer's instructions (Thermo Fisher Scientific). The cells were incubated for 8 h before removal of transfection reagents and addition of chemicals for testing.

Luciferase assays were performed using the Dual-Glo Luciferase Assay System (Promega). Upon completion of treatments, 50 μl of Dual-Glo reagent was added to each well, and then samples were incubated at room temperature for 15 min. The firefly luciferase luminescence was recorded using a GloMax Discover Microplate Reader with an integration time of 0.3 s. An equivalent volume of Dual-Glo “Stop & Glo” reagent was then added, and again incubated at room temperature for 15 min before recording of Renilla luminescence. To determine activity of each promoter site, the firefly:Renilla luminescence ratio was calculated.

Statistics

For immunoblotting and qualitative RT-PCR, experiments were performed in biological triplicate. For qPCR, experiments were performed in biological triplicate, with each replicate analyzed in turn in technical triplicate. Statistical significance was performed using the Student's t test and two-way ANOVA. Error bars represent ± standard deviation (S.D.). * = p ≤ 0.05, ** = p ≤ 0.01, *** = p ≤ 0.001, and **** = p ≤ 0.001.

Data availability

The data used and/or analyzed during the current study are contained within the text or available from the corresponding author on reasonable request.