GA binding protein augments autophagy via transcriptional activation of BECN1-PIK3C3 complex genes

Abstract

Macroautophagy is a vesicular catabolic trafficking pathway that is thought to protect cells from diverse stressors and to promote longevity. Recent studies have revealed that transcription factors play important roles in the regulation of autophagy. In this study, we have identified GA binding protein (GABP) as a transcriptional regulator of the combinatorial expression of BECN1-PIK3C3 complex genes involved in autophagosome initiation. We performed bioinformatics analyses that demonstrated highly conserved putative GABP sites in genes that encode BECN1/Beclin 1, several BECN1 interacting proteins, and downstream autophagy proteins including the ATG12–ATG5-ATG16L1 complex. We demonstrate that GABP binds to the promoter regions of BECN1-PIK3C3 complex genes and activates their transcriptional activities. Knockdown of GABP reduced BECN1-PIK3C3 complex transcripts, BECN1-PIK3C3 complex protein levels and autophagy in cultured cells. Conversely, overexpression of GABP increased autophagy. Nutrient starvation increased GABP-dependent transcriptional activity of BECN1-PIK3C3 complex gene promoters and increased the recruitment of GABP to the BECN1 promoter. Our data reveal a novel function of GABP in the regulation of autophagy via transcriptional activation of the BECN1-PIK3C3 complex.

Introduction

Macroautophagy, hereafter designated autophagy, is a vesicular trafficking pathway for autolysosomal degradation that is thought to protect cells from nutrient starvation,¹ infection,² toxic protein aggregates,³ injured/defective mitochondria,⁴ and other cellular stressors. Autophagy deficiencies have been implicated in aging⁵ and neurodegeneration.^6,7 A comprehensive understanding of the regulatory mechanisms that govern autophagy efficacy may lead to novel therapeutic strategies to augment protective autophagy in aging and disease.

Autophagosomes, or autophagic vacuoles (AVs), are double-membrane vacuoles that sequester and deliver cellular organelles, cytoplasm, and membranes to the lysosome for degradation. The induction, formation, and trafficking of AVs is regulated by several macromolecular protein complexes. In AV formation, the BECN1-PIK3C3-PIK3R4 core complex catalyzes phosphorylation of membrane phosphatidylinositol to phosphatidylinositol-3-phosphate.⁸ Phosphatidylinositol-3-phosphate then recruits downstream autophagy regulatory proteins^8-10 including the ATG12–ATG5-ATG16L1 complex which facilitates the incorporation of lipidated LC3 (LC3-II) in the autophagosome membrane. Several autophagy regulatory proteins, including ATG14 and UVRAG, bind to the BECN1-PIK3C3-PIK3R4 core complex to activate lipid phosphorylation. ATG14¹¹ increases AV induction, whereas UVRAG¹² promotes the maturation of AVs and endocytic vesicles. BECN1 deficiency reduces autophagy,¹³ promotes tumorigenesis,¹³ and likely contributes to proteinopathies including amyloid-β pathology in Alzheimer disease.^14-16 Augmenting BECN1 levels ameliorates SNCA/α-synuclein¹⁷ and HTT/huntingtin^18,19 neuropathologies. Furthermore, disinhibiting endogenous BECN1 protects cells from mutant huntingtin toxicity and infectious pathogens.²⁰ A more comprehensive understanding of how BECN1-PIK3C3-PIK3R4 complex protein levels and activity are regulated to impact autophagy will likely have implications for new therapeutic strategies in neurodegeneration, cancer, and infectious diseases.

Gene transcription, an important factor that regulates protein levels, plays a role in autophagy regulation.²¹ Transcription of autophagy regulatory genes, including BECN1, decreases with aging.^5,18 BECN1 transcript levels are also reduced in patients with Alzheimer disease.¹⁴ Reduced BECN1 transcription may thus predispose to age-related diseases through autophagy deficiency. Recent studies have identified TFEB,²² ZKSCAN3,²³ E2F1,²⁴ and histone methylation via methyltransferase EHMT2/G9a²⁵ as important transcriptional regulators of autophagy. Notably, these factors tend to regulate expression of genes predominantly involved in late steps of autophagy regulation. Transcriptional regulation of autophagy initiation complexes, including the BECN1-PIK3C3 complex, is poorly understood. NFKB²⁶ has been implicated in transcriptional regulation of the BECN1 gene and E2F1 has been suggested to bind to the BECN1 promoter.²⁷ FOXO3 and CLOCK-ARNTL/BMAL1 regulate ATG14 transcription.²⁸ TFEB plays a role in transcriptional activation of UVRAG.²² However, regulatory mechanisms that activate the combinatorial transcription of autophagy initiation complexes have not been reported.

The ETS-domain transcription factor (ETS-TF) family is well known for combinatorial regulation of gene expression.²⁹ Among the family of ETS-TFs, GA binding protein (GABP, also known as NRF2 or nuclear respiratory factor 2) functions as either heterodimers or heterotetramers of GABPA and GABPB subunits. GABPA contains the DNA binding domain and is required for the recruitment of GABPB to DNA, whereas GABPB contains the transactivation domains and directs the transportation of GABP complexes from the cytoplasm to the nucleus.³⁰ GABP was first identified to regulate the expression of several nuclear-encoded mitochondrial genes^31,32 and the prolactin gene.^33,34 Recent studies have shown that GABP activates the expression of genes involved in the differentiation of myeloid cells^35-40 and skeletal muscle.^41-46 Furthermore, GABP interacts with E2F to promote cell cycle progression through the S-phase via the transcription of DNA synthesis genes, indicating its regulatory role in cell proliferation and differentiation.⁴⁷ A role for GABP and other ETS-TFs in autophagy regulation has not been reported.

In this study, we examined a novel role for GABP in the combinatorial transcriptional regulation of genes encoding BECN1-PIK3C3 complexes. We found that putative ETS-TF/GABP motifs are ubiquitously present in genes encoding protein subunits of BECN1-PIK3C3 complexes. Our data show that GABP binds to the promoters of these genes and activates their transcription. Furthermore, we found that GABP knockdown reduces the levels of BECN1-PIK3C3 complex proteins and inhibits basal autophagy whereas GABP overexpression increases autophagy. Finally, we found that nutrient starvation increases GABP-dependent transcriptional activity of genes encoding BECN1-PIK3C3 complexes and increases GABP binding to the endogenous BECN1 promoter. Our study identifies GABP as a novel transcriptional regulator of autophagy and as a potential therapeutic target to augment autophagy.

Results

Genes encoding BECN1-PIK3C3 complexes contain functional ETS-TF core sequences in their promoter regions

We performed bioinformatics analyses to determine whether autophagy genes contain consensus core ETS-TF sequences (5′-SGGAAG-3′) in their 5′ regulatory regions. We found putative ETS-TF sequences in the promoter regions of key genes encoding protein subunits of BECN1-PIK3C3 complexes, including BECN1, PIK3C3, UVRAG, ATG14, and PIK3R4 (Fig. 1; Figs. S1–S6). We also noted consensus core ETS-TF sequences in ATG5, ATG12, and ATG16L1 (Fig. S1; Figs. S7–S9). Sequence alignments from mammalian species demonstrated that these ETS-TF consensus sequences are highly conserved (Figs. S2–S9), suggesting they may have functional importance for the transcriptional regulation of these autophagy genes. Our sequence alignments also corroborated previously reported autophagy gene promoter regulatory elements including NFKB (5′-GGGRNTTTCC-3′) in intron 1 of BECN1,²⁶ TFEB (5′-CANNTG-3′ or E-box) in the promoter region of UVRAG,²² and FOXO (5′-TGAAAACA-3′) and E-box in the promoter region of ATG14 (Fig. 1; Figs. S2, S4, and S5).²⁸ Our statistical analysis of putative autophagy gene ETS-TF binding sites via the TRANSFAC database rendered GABP, an ETS-TF that has high affinity for the core ETS-TF binding consensus sequence,^32,48 as the only ubiquitous putative ETS-TF likely to bind to genes encoding BECN1-PIK3C3 complex proteins (Table S1). These results suggested that GABP is a strong candidate for the combinatorial regulation of BECN1-PIK3C3 complex gene transcription.

Figure 1.BECN1, PIK3C3, UVRAG, and ATG14 5′ regulatory regions contain putative ETS-TF sites. Schematic diagrams of 5′ regulatory regions of human BECN1 (A), PIK3C3 (B), UVRAG (C), and ATG14 (D) demonstrate known and putative TF sites revealed by our bioinformatics analyses (also see Figs. S2–S5). Dark blue rectangles represent noncoding exons and light blue rectangles represent coding exons. Major transcription start sites are identified with arrows and translational start codons (ATG) are marked. Conserved transcription factor binding sites are presented as follows: ETS (orange stars), Sp1 and Sp-family (Sp-fam; purple ovals), CREB (gray diamond), NFKB (red oval), E-boxes (pink ovals), NFYA (green oval) and FOXO (dark blue oval). Red lines below the promoter schematics delineate the ETS-TF containing sites cloned into our luciferase reporter constructs.

Putative GABP elements activate BECN1-PIK3C3 complex gene promoters

To address the functional significance of putative GABP elements in BECN1-PIK3C3 complex genes, we cloned highly conserved sequences from the promoter regions of BECN1, PIK3C3, UVRAG, and ATG14 into luciferase reporter constructs. We compared relative luciferase activities of each WT promoter-luciferase construct with constructs containing mutations in the GABP sites (Fig. 2). Mutations of the 2 GABP sites in the BECN1 promoter fragment (Fig. 2A) and of a single GABP site in the PIK3C3 promoter fragment (Fig. 2B) dramatically reduced luciferase activities by ~90% compared with the corresponding WT constructs, indicating that GABP elements are functional and responsible for most of the transcriptional activities in these 2 promoter fragments. Mutation of the 3′ GABP site in the UVRAG promoter decreased luciferase activity by ~25% compared with the WT promoter fragment, whereas mutation of both 5′ and 3′ GABP sites reduced luciferase activity by ~55% (Fig. 2C). Mutation of a single GABP site in the ATG14 promoter reduced luciferase activity by ~30% compared with the WT promoter (Fig. 2D). In summary, our data suggest that putative GABP elements activate the transcriptional activities of BECN1-PIK3C3 complex gene promoters.

Figure 2. Putative GABP sites activate transcriptional activities of BECN1-PIK3C3 complex gene promoters. (A–D) Promoter sequences containing GABP recognition sites were cloned into pGL3b luciferase reporter constructs. GABP sites are identified and mutated bases are highlighted in red lowercase. Mutations of GABP sites were associated with marked reductions in transcriptional activity of the BECN1 (A) and PIK3C3 (B) promoter constructs and moderate reductions in transcriptional activity of the UVRAG (C) and ATG14 (D) promoter constructs. *P < 0.05, compared with transcriptional activities of corresponding wild-type promoter constructs, ANOVAs with post-hoc Dunnett tests, n = 3 for each construct.

GABP binds to the promoter regions of autophagy regulatory genes and activates their transcription

To determine whether GABP interacts with the promoters of autophagy regulatory genes, we performed chromatin immunoprecipitation (ChIP) assays in cultured cells using an antibody to GABPA, the DNA binding subunit of GABP. These experiments confirmed that GABPA binds to the promoter regions of BECN1, PIK3C3, PIK3R4, ATG14, and UVRAG (Fig. 3A; Fig. S10A–S10C). ChIP experiments also demonstrated interaction of GABPA with the promoter regions of ATG5, ATG12, and ATG16L1 (Fig. S10D), suggesting that GABP also potentially regulates transcription of genes encoding the ATG12–ATG5-ATG16L1 complex. GABPA did not show significant binding to the promoter region of RAG2 (Fig. 3A; Fig. S10A), a previously reported negative control gene⁴⁰ that is expressed in HeLa cells (Fig. 3B) and SH-SY5Y cells (Fig. S10E). Furthermore, we confirmed that GABPA binds to the Becn1 promoter in B103 rat neuroblastoma cells and cultured rat cortical neurons (Fig. S10B and S10C). To determine whether GABP regulates the expression of genes encoding BECN1-PIK3C3 complexes, we used siRNA oligonucleotides to knock down GABPA and then examined autophagy gene transcripts by RT-PCR. Seventy percent knockdown of GABPA mRNA levels was associated with significant reductions in BECN1, PIK3C3, UVRAG, ATG14, and PIK3R4 mRNA levels (Fig. 3B). Taken together, these data demonstrate that GABP binds to the promoter regions of autophagy regulatory genes and functions as a novel transcriptional activator of autophagy gene expression.

Figure 3. GABP binds to the promoter regions of BECN1-PIK3C3 complex genes and activates transcription. (A) ChIP assays performed in SH-SY5Y human neuroblastoma cells demonstrate immunoprecipitation of promoter DNA from BECN1-PIK3C3 complex genes with a GABPA antibody. The GABPA promoter was used as a positive control and the RAG2 promoter was employed as a negative control. Nonspecific rabbit IgG was used as an antibody control. H₂O denotes the negative PCR controls. (B) GABPA knockdown with siRNA oligonucleotides significantly reduced BECN1-PIK3C3 complex mRNA transcript levels as determined by real-time RT-PCR. RAG2 mRNA levels were investigated as a negative control. *P < 0.05, compared with mRNA levels with scrambled control (siControl) oligonucleotide transfection, unpaired t tests, n = 4.

GABPA knockdown reduces autophagy

We performed western blots to determine whether siRNA-mediated knockdown of GABPA protein attenuates BECN1-PIK3C3 complex protein levels and autophagy. GABPA knockdown to approximately ~75% of control levels was associated with significant reductions in BECN1, PIK3C3, and UVRAG (Fig. 4A). Furthermore, the same cultures showed significantly decreased autophagosomal LC3-II levels and significantly increased levels of the autophagy substrate SQSTM1 (Fig. 4B), indicating reduced autophagy flux with GABPA knockdown. SQSTM1 transcript levels were unaffected by GABPA knockdown (Fig. S11A and S11B). Furthermore, GABPA knockdown reduced cytoplasmic autophagosomes and autolysosomes as detected with the mRFP-GFP-LC3 tandem reporter assay (Fig. 4C–E). Cell cycle arrest in G₁ phase via double thymidine blockade did not recapitulate the inhibition of autophagy observed with GABPA knockdown (Fig. S12), suggesting that the reduction in autophagy observed with GABPA knockdown is not an artifact of cell cycle arrest. Taken together, our data show that GABP plays an important role in maintaining BECN1-PIK3C3 complex protein levels and driving basal autophagy.

Figure 4.siRNA-mediated knockdown of GABPA inhibits autophagy in HeLa cells. (A) GABPA knockdown with siRNA oligonucleotides was associated with reductions in BECN1, PIK3C3 and UVRAG levels in HeLa cells (immunoblots with 3 replicates per treatment group depicted on left; densitometric quantification of 6 replicates graphically depicted on right). *P < 0.05 compared with protein levels with scrambled control (siControl) oligonucleotide transfection, unpaired t tests. (B) Western blots demonstrated significantly increased SQSTM1 levels and decreased autophagosomal LC3-II levels following GABPA knockdown in HeLa cells (immunoblots with 2 replicates per treatment group depicted on left; densitometric quantification of 8 replicates graphically depicted on right). *P < 0.05 compared with protein levels with scrambled control (siControl) oligonucleotide transfection, unpaired t tests. (C) Deconvoluted fluorescence microscopic images of siControl and siGABPA-transfected HeLa cells transiently expressing the mRFP-GFP-LC3 tandem reporter. AVs, quantified as GFP and RFP colocalizing signals, and autolysosomes (ALs), quantified as RFP-positive signals, were both reduced with GABPA knockdown (D). * and **P < 0.05 compared with pcDNA vector control-transfected cells for colocalizing GFP-RFP puncta (AVs) and solitary RFP fluorescent puncta (ALs), respectively; unpaired t tests. (E) Western blot confirming GABPA knockdown and reduced LC3 and mRFP-GFP-LC3 shifts with siGABPA transfection.

GABP overexpression increases autophagy

To determine whether ectopic GABP increases autophagy, we transfected GABP expression constructs into HeLa cells and examined cellular markers of autophagy. GABPA overexpression and coexpression of GABPA and GABPB1 increased levels of the autophagosomal marker LC3-II and decreased levels of the autophagy substrate SQSTM1 (Fig. 5A–C). GABP-induced increases in autophagosomal LC3-II persisted following inhibition of AV maturation with bafilomycin A₁, consistent with increased autophagy induction (Fig. 5D). GFP expression did not recapitulate the increase in autophagy observed with GABP expression (Fig. S13). GABP expression also increased cellular levels of PIK3C3, BECN1, and UVRAG (Fig. 5E). Furthermore, GABP expression increased transcriptional activity of the wild-type BECN1 luciferase reporter construct but had no effect on the transcriptional activity of the GABP mutant BECN1 reporter construct (Fig. 5F).

Figure 5. GABP overexpression upregulates autophagy induction, autophagic flux, and expression of BECN1-PIK3C3 complexes. (A) A western blot demonstrates GABPA and GABPB1 overexpression in HeLa cells. (B and C) GABP overexpression was associated with increased autophagosomal LC3-II and reduced levels of the autophagy substrate SQSTM1. Overexpression of BECN1 was employed as a positive control to increase SQSTM1 degradation. *P < 0.05 compared with pcDNA vector control, ANOVA with post-hoc Dunnett test, n = 12. (D) GABP induced increases in LC3-II persisted during treatment with 10 nM bafilomycin A₁, an inhibitor of autophagosome-lysosome fusion, suggesting that GABP overexpression increases autophagy induction. (E) GABP overexpression was associated with increases in the levels of PIK3C3, BECN1, and UVRAG. *P < 0.05 compared with pcDNA vector control. ANOVA with post-hoc Dunnett test, n = 4. (F) GABP overexpression increased the transcriptional activity of the BECN1 luciferase reporter construct but had no impact on transcriptional activity following mutations of the GABP binding sites in the BECN1 promoter. *P < 0.05 GABP(A+B1) compared with pcDNA vector control transfection, unpaired t tests, n = 3.

Overexpression of GABPA or GABPB1 also increased GFP-LC3 labeled AVs as detected by fluorescence microscopy (Fig. 6A and B) and by western blot analysis of GFP-LC3 (Fig. 6C). Increased GFP-LC3 labeled AVs were also observed with GABPA or GABPB1 overexpression in stable GFP-LC3-expressing HeLa cells (Fig. S14). Furthermore, GABP-induced increases in autophagosomes and autolysosomes were detected in tandem reporter experiments by fluorescence microscopy (Fig. 6D and E) and by western blot detection of ectopic mRFP-GFP-LC3 (Fig. 6F). Transmission electron micrographs of HeLa cells cotransfected with GABPA and GABPB1 expression constructs confirmed the increase in cytoplasmic autolysosomes seen with fluorescence microscopy reporter experiments (Fig. 7A and B). Interestingly, we detected a decrease in mature endosomes in GABP-overexpressing HeLa cells (Fig. 7C) and a decrease in RAB5-positive early endosomes in HeLa cells with GABPA siRNA knockdown (Fig. S15). These results suggest that GABP-induced alterations in BECN1-PIK3C3 levels impact autophagy more specifically than endosomal trafficking. In summary, these data demonstrate that GABP augments the expression of genes encoding BECN1-PIK3C3 complexes and autophagy.

Figure 6. GABP overexpression increases GFP-LC3-labeled autophagosomes. (A) Representative images of HeLa cells cotransfected with GFP-LC3 and either pcDNA vector, GABPA, GABPB1, or BECN1 overexpression vectors are shown. Cells overexpressing GABPA, GABPB1, or BECN1 demonstrate increased punctate GFP-LC3-labeled autophagosomes (quantified in B). *P < 0.05 compared with pcDNA vector control, ANOVA with post-hoc Dunnett test. (C) A GFP-LC3 immunoblot performed on HeLa lysates confirmed increased levels of autophagosomal GFP-LC3-II in cultures cotransfected with GABPA, GABPB1, or BECN1. (D) Representative images of tandem-reporter mRFP-GFP-LC3-transfected HeLa cells cotransfected with GABPA or GABPB1 overexpression vectors demonstrate increased punctate autophagosomes (AVs) and autolysosomes (ALs) compared with pcDNA vector-transfected cells (quantified in E). * and **P < 0.05 compared with pcDNA vector control-transfected cells for colocalizing GFP-RFP puncta (AVs) and solitary RFP fluorescent puncta (ALs), respectively; ANOVAs with post-hoc Dunnett tests. (F) An LC3 immunoblot demonstrates increased levels of tandem reporter mRFP-GFP-LC3-II in GABP-transfected cells consistent with increased autolysosomes.

Figure 7. GABP overexpression increases autolysosomes but not mature endosomes. (A) Representative transmission electron microscopy images show that overexpression of GABPA and GABPB1 in HeLa cells evoked increases in cytoplasmic autolysosomes (ALs). Black arrows identify autolysosomes, white arrows identify late endosomes and white arrowheads identify multivesicular bodies. (B) Cumulative data demonstrate an increase in autolysosomes with GABP overexpression. *P < 0.05 compared with pcDNA vector control, unpaired t tests. (C) Cumulative data demonstrate no increase, but rather a decrease, in mature endosomal structures with GABP overexpression. *P < 0.05 compared with pcDNA vector control, unpaired t tests.

Nutrient starvation increases GABP-dependent autophagy gene promoter activity

To determine whether GABP upregulates autophagy gene transcription during cell stress, we subjected B103 cells expressing wild-type or mutant luciferase reporter constructs (Fig. 2) to nutrient starvation. We observed robust increases in the transcriptional activities of BECN1, PIK3C3, UVRAG, and ATG14 wild-type reporter constructs during incubation in HBSS (Fig. 8A). GABP binding site mutations markedly blunted the baseline transcriptional activities of BECN1 and PIK3C3 reporter constructs and rendered them essentially unresponsive to nutrient starvation. GABP binding site mutations resulted in more modest attenuations of baseline transcriptional activity of UVRAG and ATG14 reporter constructs. GABP binding site mutations blunted the response of the UVRAG promoter, but not the ATG14 promoter, to nutrient starvation. Furthermore, we observed an increase in GABPA recruitment to the endogenous BECN1 promoter with nutrient starvation via ChIP assays (Fig. 8B). These data suggest that, in addition to the maintenance of baseline BECN1-PIK3C3 complex gene transcriptional activity, GABP plays a role in upregulating transcriptional activity of BECN1-PIK3C3 complex genes during nutrient starvation.

Figure 8. Nutrient starvation increases GABP-dependent transcriptional activities of BECN1-PIK3C3 complex gene promoters. (A) Luciferase assays demonstrate robust increases in transcriptional activity of BECN1, PI3KC3, ATG14, and UVRAG promoters during nutrient starvation (HBSS). HBSS-induced increases in BECN1, PI3KC3, and UVRAG transcriptional activities were markedly attenuated in promoter constructs with mutated GABP binding sites. The HBSS-induced increase in transcriptional activity of the ATG14 promoter, which also contains FOXO and E-box sequences, was not eliminated by mutation of the GABP binding site. *P < 0.05, comparisons of normal media and HBSS conditions for each promoter; ns, not significant; unpaired t tests with Bonferroni corrections for multiple comparisons. (B) Semiquantitative BECN1 promoter ChIP assay using a GABPA antibody demonstrates an increase in GABPA binding to the endogenous BECN1 promoter during nutrient starvation. BECN1 promoter DNA was not immunoprecipitated with a nonspecific IgG. Negative control RAG2 promoter DNA was not immunoprecipitated by the GABPA antibody under control or nutrient starvation conditions. *P < 0.05, comparison of normal media and HBSS conditions for BECN1 promoter ChIP with the GABPA antibody, unpaired t test.

Discussion

Autophagy is a complex vesicular trafficking pathway for autolysosomal degradation. Autophagy dysregulation, including impaired autophagy initiation^6,7,49 and impaired maturation,^50,51 evokes detrimental consequences for cells. Accordingly, autophagy is tightly regulated at multiple levels, including autophagosome formation, autophagosome trafficking, and upstream signaling pathways that respond to nutrients, growth factors, and stressors.²¹ The induction, formation, and trafficking of AVs is regulated by several macromolecular regulatory protein complexes, including the BECN1-PIK3C3 complex. In this study, we found that the ETS-TF GABP augments the combinatorial expression of autophagy genes encoding the BECN1-PIK3C3-PIK3R4 core complex, genes encoding the BECN1 binding proteins ATG14 and UVRAG, and autophagy. Furthermore, the impairment of autophagy with GABPA knockdown suggests that GABP plays an important role in the basal maintenance of autophagic machinery and capacity. Finally, we observed GABP-dependent increases in transcriptional activity of autophagy gene reporter constructs and increased GABP recruitment to the endogenous BECN1 promoter during nutrient starvation, suggesting a novel role for GABP in the regulation of BECN1-PIK3C3 complex gene expression during cellular stress.

Autophagy gene transcription is an important factor in the regulation of autophagy and autophagy capacity; however, autophagy transcriptional regulatory mechanisms are incompletely understood.²¹ Recent studies have identified E2F1,²⁴ FOXOs,^28,52 TFEB,²² and ZKSCAN3²³ as TFs that regulate the expression of autophagy genes. However, TFs that regulate the combinatorial expression of autophagy initiation complex genes are unknown. In our study, we found that the ETS-TF consensus binding element, which is present in approximately 7% of human gene promoters,⁵³ is ubiquitously found in the 5′ regulatory regions of BECN1-PIK3C3 complex genes, including BECN1, PIK3C3, UVRAG, ATG14, and PIK3R4, and regulates the expression of these genes. Additionally, we found that GABP binds to the promoter regions of genes encoding the downstream ATG12–ATG5-ATG16L1 complex. Furthermore, our analysis of GABP ChIP-seq data from the ENCODE project in the UCSC genome browser⁵⁴ revealed that GABP binds to 22 of 58 (38%) of analyzed autophagy-related genes (Table S2), indicating that GABP may serve as master regulatory transcription factor of autophagy.

Several TFs have been previously implicated in the transcriptional regulation of genes encoding subunits of the BECN1-PIK3C3 complex, including NFKB²⁶ in the BECN1 promoter, TFEB in the UVRAG promoter²² and FOXO3 and CLOCK-ARNTL/BMAL1 in the ATG14 promoter.²⁸ The presence of an E-box element⁵⁵ in the ATG14 promoter raises the possibility of regulation by TFEB. Additionally, our bioinformatics analyses implicate SP1 and NFYA, which are transcriptional coactivators,^53,56 with possible roles in the regulation of UVRAG transcription. However, none of these transcription factors demonstrate the combinatorial overlap in BECN1-PIK3C3 complex gene regulation we have documented for GABP. The combinatorial regulation of autophagy gene expression by GABP likely augments its impact on autophagy.

Although our TRANSFAC analyses rendered GABP as the only ubiquitous likely ETS-TF for the regulation of BECN1-PIK3C3 complex genes, the possibility that other ETS-TFs may also regulate expression of these autophagy genes remains to be explored. Furthermore, the possibility that GABP itself might have diverse effects on autophagy gene transcription is intriguing. GABP is a unique multimeric member of the ETS-TF family consisting of 2 subunits, the DNA binding subunit GABPA and the transactivation subunit GABPB.^29,30 In humans, the GABPA subunit is encoded by GABPA on chromosome 21 and the GABPB subunit is encoded either by GABPB1 on chromosome 15 or by GABPB2 on chromosome 1.⁵⁷ Structural heterogeneity of the GABPB subunit secondary to alternative splicing has an impact on GABP heterodimer formation and its interactions with transcriptional cofactors.^36,57,58 Thus, GABP and other ETS-TFs may have a broad repertoire of regulatory effects on the transcriptional output of autophagy genes.

GABP regulates the expression of genes involved in the differentiation of hematopoietic cells,^35-40 skeletal muscle,^41-46 and adenohypophysis.^33,34 Furthermore, GABP regulates the expression of several nuclear-encoded mitochondrial genes^31,32 and plays a role in mitochondrial dynamics during embryogenesis.⁵⁹ Autophagic proteolysis also plays important roles in cell remodeling.⁶⁰ GABP therefore might also facilitate cell differentiation by increasing the expression of autophagy initiation complexes. Furthermore, GABP, which promotes cell cycle entry into S-phase via interaction with E2F and upregulation of DNA synthesis genes,⁴⁷ might contribute to autophagy activation during G₁ and S-phases⁶¹ and inhibition during the G₂/M phases.^61,62 The possibility that GABP regulates autophagy during these contexts will require further investigation.

Autophagy is generally thought to protect cells from diverse stressors, including toxic protein aggregates, injured mitochondria, and infectious organisms. Furthermore, autophagy deficiency renders cells more vulnerable to stress and is implicated in nervous system aging and neurodegeneration. For example, autophagy deficiencies in Atg5 and Atg7 knockout mice are associated with precipitous neurodegeneration.^6,7,49 Deficiencies of BECN1 and PIK3C3 are associated with nervous system aging and Alzheimer disease.^14,18,63 Cleavage of BECN1 by caspases may contribute to BECN1 deficiency in AD.⁶⁴ Augmentation of autophagy capacity by BECN1 overexpression has been proposed as a therapeutic strategy and has shown promising results for the clearance of amyloid-β pathology,⁶⁵ α-synuclein,¹⁷ and huntingtin.¹⁸ The possibility that GABP and other transcription factors may be leveraged to therapeutically augment Beclin-dependent autophagy underlies the importance of understanding the transcriptional regulation of autophagy.

In conclusion, our study is the first to identify a role of the ETS-TF GABP in regulating the combinatorial expression of autophagy initiation genes. We showed that GABP activates the transcription of genes encoding BECN1-PIK3C3 complexes at baseline and during nutrient starvation. Furthermore, we demonstrated that GABP deficiency reduces autophagy whereas GABP overexpression increases autophagy. Our work identifies a novel role for GABP in autophagy gene transcription and highlights ETS-TF and GABP-mediated gene regulation as a potential target to therapeutically augment autophagy.

Materials and Methods

All animal procedures were designed in accordance with the Guidelines for Animal Use of the Stanford University and were approved by the Institutional Animal Care and Use Committee (Protocol 26911).

Bioinformatics and phylogenetic analysis

Sequences for human BECN1, PIK3C3, UVRAG, ATG14, PIK3R4, ATG5, ATG12, and ATG16L1 were obtained from the Ensembl database (www.ensembl.org). A search for the core consensus binding sequence for ETS-TF, “5′-SGGAAG-3′,” was performed 2 kb upstream of transcriptional start site (TSS) and at least 1 kb into intron 1 of each gene. When the ETS-TF consensus sequence (5′-SGGAAG-3′) was found in the promoters of human autophagy genes, similar searches were performed in rat and mouse genes. When corresponding ETS-TF consensus sequences were found in all three species, human promoter sequences were queried in the whole genome shotgun (WGS) database to retrieve sequences from other mammalian species. Next, 5′ regulatory regions of mammalian autophagy genes were aligned using Clustal W2 (http://www.ebi.ac.uk/Tools/clustalw2/index.html) and adjusted manually to maximize parsimony. Finally, putative ETS-TF binding sites were analyzed via matrix alignment using the TRANSFAC database to reveal likely candidate ETS-TFs for each promoter.⁶⁶

Cell culture, transfection, and drug treatments

HeLa cells, B103 rat neuroblastoma cells, and SH-SY5Y (ATCC, CRL2266) human neuroblastoma cells were grown to 70% to 80% confluence in DMEM (Corning, 10-017-CV) with 10% Fetal Bovine Serum (Lonza, 14-502F) and 1% Pen-Strep (Corning, 30-002-Cl)] and supplemented with 20 mM L-glutamine (SH-SY5Y; Gibco, 25030-081) and 10 mM HEPES (SH-SY5Y; Gibco, 15630–080). B103 cells were supplemented with 5% Horse Serum (Corning, 35-030-CV). Cells were maintained at 37 °C in 5% CO₂. E16 rat cortical neuron cultures were maintained at 37 °C with 5% ambient CO₂ in Neurobasal medium (Invitrogen, 21103-049) supplemented with B27 (Invitrogen, 17504-044) and 1% Glutamax-I (Invitrogen, 35050-061). Cell lines were transfected with overexpression constructs using Lipofectamine LTX (Invitrogen, 15338100) or with siRNA oligonucleotides using RNAiMAX (Life Technology, 13778075) according to the manufacturer’s instructions. The following expression constructs were employed: GABP (pCAGGS-GABPA and pCAGGS-GABPB1), pEGFP-LC3,⁶⁷ mRFP-GFP-LC3 (tandem reporter),⁶⁸ pcDNA4-BECN1-FLAG⁶⁹ and pRK7-GFP. Stably transfected HeLa-GFP-LC3 were generated under G418 (250 μg/ml, Gibco, 10131-035) selection. Cells in which GABP knockdown was induced with 100 nM OnTarget SmartPool siGABPA (Dharmacon, Fisher Scientific, L-011662-00-0020) were compared with cells transfected with scrambled siRNA control oligonucleotide pool (Dharmacon, Fisher Scientific, D-001810-10-20). To induce nutrient starvation, cells were incubated in Hank's balanced salt solution (HBSS, Gibco, 24020–117) for 3 h. To inhibit AV fusion with lysosomes, cultures were treated with 10 nM bafilomycin A₁ (Sigma, B1793) in DMSO (Sigma, 276855) for 5 h. To induce cell cycle arrest, normal media was replaced with media containing 2.5 mM thymidine (Fisher Scientific, TCI, T0233) according to the protocol of Whitfield and colleagues.⁷⁰

Luciferase reporter assays

Proximal promoter regions from BECN1, PIK3C3, UVRAG, and ATG14 were cloned from human SH-SY5Y DNA using Pfu Ultra High Fidelity DNA polymerase (Agilent Technologies, 600385-51) and the following primers (enzyme digestion sites underlined and mutated sites in lowercase, Fig. 1 and Fig. 2, and Figs. S1–S4): BECN1-179-WT [promoter region of human BECN1, a 179-bp fragment from −47 to +132 (transcript ID: ENST00000590099) relative to transcription start site (TSS, +1])]: 5′-AATTCTCGAG GTCACGTCCG GTCTCG-3′ (F) and 5′-TCGTAAGCTTCCTCAGCCCC CGATGCTC-3′ (R); PIK3C3-261-WT [promoter region of human PIK3C3, a 261-bp fragment from −180 to +81 (transcript ID: ENST00000262039) relative to TSS]: 5′-AATTCTCGAGGACAGGAAAA CTGGAATGTA AC-3′ (F) and 5′-TCGAAAGCTTCGTCTGCAAA GGTACCACCT AC-3′ (R); UVRAG-455-WT [promoter region of human UVRAG, a 455-bp fragment from −390 to +65 (transcript ID: ENST00000356136) relative to TSS]: 5′-AATTCTCGAGCAACCAGTGA GGGCGGAAGT G-3′ (F) and 5′-TCGAAAGCTTGCTAAGGAAG AGCCATATTA CCG-3′ (R); and ATG14-286-WT [promoter region of human ATG14, a 286-bp fragment from −154 to +132 (transcript ID: ENST00000247178) relative to TSS]: 5′-AATTCTCGAG CTGCTTCGGG ACCTGTTGCC-3′ (F) and 5′-TCGAAAGCTT ATCGTCCACG GAGTCCACCA G-3′ (R). Mutations of putative GABP binding sites⁴⁸ were made either in the cloning primers or by synthesizing mutant promoter regions (Integrated DNA Technologies, Inc., Fig. 2). PCR products were digested with XhoII (New England Biolabs, R0146S) and HindIII (New England Biolabs, R0104S), and ligated into the pGL3basic vector by a T4 ligase (New England Biolabs, M0203S). All wild-type (WT) and mutant inserts were confirmed with DNA sequencing. pGL3b generated constructs were cotransfected with pRL-TK into B103 rat neuroblastoma cells. Cells were lysed 24 h after transfection with passive lysis buffer (Promega, E1910). Luciferase activity was measured using a Dual-luciferase reporter assay (Promega, E1910) on a Lumat LB 9507 luminometer (Berthold Technology, Germany). For each sample, the firefly luciferase activity was normalized to the Renilla luciferase activity. Each experiment contained 3 replicates per luciferase reporter construct. Experiments were repeated for reproducibility.

Chromatin immunoprecipitation (ChIP)

Cells were fixed with 4% formaldehyde, diluted from 37% formaldehyde solution (LabChem Inc., Thermo Fisher, LC14650-2), scraped and collected. Protein-DNA complexes were extracted, sonicated, immunoprecipitated and eluted with a Chromatin Immunoprecipitation Assay Kit (Millipore, 17-295) according to the manufacturer’s instructions using a GABPA antibody (H-180X, Santa Cruz Biotechnology, sc-22810). A nonspecific rabbit IgG (Millipore, PP64) was employed as a negative immunoprecipitation control. Immunoprecipitated genomic DNA was recovered and used in polymerase chain reactions (PCR) with GoTaq DNA polymerase (Promega, M3005) and the following primers flanking GABP sites in the corresponding promoters: GABPA promoter (employed as a positive control⁴⁰: 5′-ACGCCTACCC GCCATCGCAA TG-3′ (F) and 5′-GAGCTTGAAC TAGGGGAAAG GC-3′ (R); RAG2 promoter (employed as a negative control⁴⁰: 5′-CATTGTTGCT AGTAGTGAAA GAG-3′ (F) and 5′-GTGGCAAGCA GAAGGCTGAC TG-3′ (R); BECN1 promoter: 5′-GTGAGCCTGT GGACCAGGAG-3′ (F) and 5′-AGCCGTGAGG GTTCCCAGAC-3′(R); PIK3C3 promoter: 5′-TGGAATGTAA CCAGATTAGC CG-3′ (F) and 5′-GGGAACTTAG GTACAGGAAA AAC-3′ (R); UVRAG promoter: 5′-GGGAAAAGAC TGCACAAGGG-3′ (F) and 5′-CGGGAGAGTC ACCTGATTAG G-3′ (R); ATG14 promoter: 5′-CGGTTGAAAA CAAAATCCCA CG-3′ (F) and 5′-CGCATCGTCC ACGGAGTCCA C-3′ (R); PIK3R4 promoter: 5′-GCCCAAAGTA AGACCACGC-3′ (F) and 5′-AGAGAAGTGC AGACCGCC-3′ (R); ATG5 promoter: 5′-ACGTGGGGTC TCGTGACGT-3′ (F) and 5′-CGCCTGACAC ACTGTCCTG-3′ (R); ATG12 promoter: 5′-GATTTGAATG ACTAGCCG-3′ (F) and 5′-CATCTTGCTT GGAGACACTC G-3′ (R), and ATG16L1 promoter: 5′-ACGCTTCCGG CTACAGGCTG-3′ (F) and 5′-TCACCTCCAC ACACTGGCAG TC-3′ (R). PCR products were analyzed by agarose gel electrophoresis. Real-time PCR using Power SYBR Green reagent (Life Technology, 4367659) was also employed to analyze BECN1 promoter DNA immunoprecipitation.

RNA extraction and RT-PCR analysis

Total RNA was extracted from cells using TRIzol (Life Technologies, 15596-018) according to the manufacturer’s instructions. Reverse transcription was performed using QuantiTect Reverse Transcription Kit (Qiagen, 205310). Real-time PCR was performed using SYBR Green reagent (POWER SYBR GREEN, Life Technologies, 4367659) and primer sets as follows: GAPDH: 5′-TCACCAGGGC TGCTTTTAAC TC-3′ (F) and 5′-ATGGGTGGAA TCATATTGGA AC-3′ (R); GABPA: 5′-GCCTTGGGAT ACCCTATGAT C-3′ (F) and 5′- TGAGTGTGGT GAGGTCTATA TC-3′ (R); BECN1: 5′-GATTCATCCC CCCAGCCAGG-3′ (F) and 5′-CCCAGTGACC TTCAGTCTTC G-3′ (R); PIK3C3: 5′-AACTGGAATG AATGGCTGAA AC-3′ (F) and 5′-GCGAAACCGT TGTTCCTCCT AC-3′ (R); ATG14: 5′-GGGAGAGGTT TATCGACAAG AAGG-3′ (F) and 5-TATTTTCCAT CTCAACTGAT CTG-3′ (R); UVRAG: 5′-GAACATAAGG GTTATTCAAA TGC-3′ (F) and 5′-GTTACCTGAG TCTGTTTAAT TGC-3′ (R); PIK3R4: 5′-GGACATGAGG TTCCAGTTGC-3′ (F) and 5′-TTGTTGCCCT GAACAGCTGC-3′ (R); CCND1: 5′-GAACACTTCC TCTCCAAAAT G-3′ (F) and 5′-GGAAATGAAC TTCACATCTG TG-3′ (R); RAG2: 5′-GGTGTTCTCT TTGGAGGACG-3′ (F) and 5′-TCAGCTACAC TATTCCATTT TTC-3′ (R). Experiments were performed with a real-time PCR machine (Applied Biosystems StepOne PLUS, Life Technologies, Grand Island, NY, USA). For each sample, ∆CT value was obtained by normalizing each gene of interest against GAPDH in each assay. The relative ∆∆CT value was obtained by subtracting the ∆CT values for each sample from the ∆CT values of GAPDH in each corresponding assay. All ∆∆CT values were converted to fold change. Data was compiled from 4 independent cultures for each treatment. Three technical RT-PCR replicates were performed for each cDNA sample. Finally, for the analysis of SQSTM1 transcript levels, semiquantitative PCR was performed on 1 µl cDNA samples using the following primers: F: 5′-TGAGGAAGAT CGCCTTGGAG-3′ and R: 5′-ACAGATGGGT CCAGTCATCA TC-3′.

Western blot

Total protein was extracted 24 to 48 h following transfection using HEPES buffer [20 mM HEPES (Gibco, 15630-080), 150 mM NaCl, 1 mM EDTA and 1% Triton X-100 (Sigma, T8787)] or modified RIPA buffer [50 mM Tris, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.5% deoxycholate (Sigma, D6750), 0.1% SDS (Amresco, 0837)] supplemented with a protease inhibitor cocktail (Sigma, P8340). Protein quantification was performed via the Bradford method (Bio-Rad, 500-0006) using a BioTek Epoch 96-well microplate reader (BioTek, Winooski, VT, USA). Protein samples were loaded into 4–15% Tris-Glycine gels (Bio-Rad, 456-1085 and 456-1086) and transferred onto PVDF membranes (Immobilon-P, Millipore, IPVH00010). Immunoblotting was performed using primary antibodies for BECN1 (Cell Signaling Technology, 3495S), GABPA (Santa Cruz Biotechnology, sc-22810), PIK3C3 (Cell Signaling Technology, 4263S), UVRAG (Medical and Biological Laboratory Co, Ltd, M160-3), MAP1LC3 (NanoTools, 026-100), SQSTM1/p62 (Cell Signaling Technology, 5114S), GABPB1 (Abcam, ab90861), GFP (Millipore, MAB2510), FLAG (Sigma, F3165), GAPDH (Abcam, ab9485) and ACTB/β-actin (Cell Signaling Technology, 4970). Blots were then labeled with secondary antibodies: HRP-conjugated anti-mouse IgG (GE Healthcare, NXA-931) or HRP-conjugated anti-rabbit IgG (GE Healthcare, NA934V). Immunoreactive bands were detected by exposure of X-ray film (Denville, E3018) following incubation of blots with ECL Prime chemiluminesence reagents (GE Healthcare, RPN2236). Densitometry was performed using NIH ImageJ software.

Immunofluorescence staining and fluorescence microscopy

Cell cultures were grown in 4-well Lab-Tek coverglass chambers (growth area 1.8 cm², Thermo Scientific Nunc, 155382). Cultures were fixed with 4% paraformaldehyde in phosphate-buffered saline (Corning Cellgro, 21-031-CV), rinsed and permeabilized with 0.1% Triton-X in PBS. Cultures were then blocked with 5% bovine serum albumin (Fisher Scientific, BP1600-100), incubated with RAB5 antibody (1:200; Santa Cruz Biotechnology, sc-28570) and stained with Alexa Fluor 488 goat anti-rabbit IgG (Life Technologies, A11008).

Stained cultures were imaged with a Leica DMI6000 wide-field fluorescence microscope (Leica Microsystems, Buffalo Grove, IL). Z-stacks from 15 random medium power fields (200×) containing 4 to 6 transfected cells were acquired from the central portion of the coverglass chambers. Z-stacks were processed with Leica 3-D deconvolution software and colocalizing events were detected with the colocalization module (Leica Advanced Fluorescence Application Suite). The central image from each deconvoluted stack was then used to quantify fluorescent puncta in a plane that included the center of the nucleus. All transfected cells were analyzed in all images, rendering data from 180 to 320 cells per experiment. Images were analyzed with NIH ImageJ using the Analyze Particles function after fluorescence signals were thresholded at the mean background cytoplasm signal + 4× the standard deviation of the background cytoplasm signal.

Transmission electron microscopy

Cell cultures were fixed in 3% glutaraldehyde/4% paraformaldehyde (Electron Microscopy Sciences, 16020 and 15710) in 0.1 M sodium cacodylate buffer (pH7.4) (Electron Microscopy Sciences, 11650) and post fixed in 1% osmium tetroxide (Electron Microscopy Sciences, 19110). Cultures were then scraped with a rubber soldier, pelleted, and re-suspended in 10% gelatin in PBS. Cells were then pelleted again and stained in cold 1% uranyl acetate (Electron Microscopy Sciences, 22400) overnight. Samples were then dehydrated in a series of ethanol washes and acetonitrile, infiltrated with EMbed-812 resin (Electron Microscopy Sciences, 14120) and then placed into molds filled with fresh resin and polymerized overnight. Sections cut at 75 nm on a Leica Ultracut S (Leica, Wetzlar, Germany) were picked up on formvar/carbon coated grids (100 mesh copper; Electron Microscopy Sciences, FCF100-Cu-50) and contrast stained in 3.5% uranyl acetate in 50% acetone followed by staining in 0.2% lead citrate (Electron Microscopy Sciences, 17900) for 3 min. Grids were observed in a JEM-1400 transmission electron microscope (JEOL USA, Peabody, MA) at 120kV. Photos were taken at 1500× and 4000× using a Gatan Orius digital camera. AVs, autolysosomes (ALs), multivesicular bodies and late endosomes were quantified from 50 consecutive cells in each treatment group.⁷¹ Quantifications were performed blind to the treatment group. These transmission electron microscopy experiments were performed with the assistance of the Cell Sciences Imaging Facility at Stanford University.

Statistical analyses

All data are expressed as mean ± standard errors (SEM). Data from multiple treatment groups were compared with ANOVAs. Data from experimental groups were compared with control groups using post-hoc 2-tailed Dunnett tests. Data from dual treatment groups were compared with unpaired Student t tests. P < 0.05 was accepted as significantly different. Statistical analyses were performed with the SYSTAT version 13 software suite.

Glossary

Abbreviations:

ACTB

actin, beta isoform

AL

autolysosome

ATG5

autophagy-related 5

ATG12

autophagy-related 12

ATG14

autophagy-related 14

ATG16L1

autophagy-related 16-like 1

AV

autophagic vacuole

BECN1

Beclin 1, autophagy related

ChIP

chromatin immunoprecipitation

ETS-TF

ETS-domain-transcription factor

GABP/NRF2

GA binding protein transcription factor composed of GABPA and GABPB

GABPA

GA binding protein transcription factor, alpha subunit 60 kDa

GABPB1

GA binding protein transcription factor, beta subunit 1

GFP

green fluorescent protein

MAP1LC3 (LC3)

microtubule-associated protein 1 light chain 3

mRFP

monomeric red fluorescent protein

Sp-fam

Sp-family of transcription factors

SQSTM1/p62

sequestosome 1

PCR

polymerase chain reaction

PIK3C3/VPS34

phosphatidylinositol 3-kinase, catalytic subunit type 3

PIK3R4/VPS15

phosphoinositide-3-kinase, regulatory subunit 4

UVRAG

UV radiation-resistance associated

WT

wild type