Introduction

Autophagy, which literally means “self-eating”, is a unique and fundamental mechanism responsible for the degradation of cellular long-lived proteins and organelles that simultaneously recycles components to maintain cellular homeostasis; autophagy is often functionally related to apoptosis on a molecular level [1–5]. Autophagy is induced in response to nutrient starvation and has tissue-specific functions, though a basal level of autophagy is maintained as a housekeeping process in most cells [6, 7]. More than thirty autophagy-related (ATG) genes have been identified by yeast genetics, and recent advances in our understanding of the molecular mechanism underlying autophagy have enabled the spatiotemporal regulation of the formation of specific protein complexes during autophagy to be revealed. These protein complexes are mainly involved in three well-characterized processes: microautophagy, macroautophagy (generally referred to as autophagy), and chaperone-mediated autophagy [8–11]. In mammals, approximately two dozen homologs of the yeast ATG genes have been identified, strongly implying that these genes might also be involved in the pathogenesis of various diseases including cancer, neurodegenerative disease, and cardiac disease [12–14]. To date, whether autophagy promotes or hinders the progression of such diseases remains uncertain [15, 16].

To reveal the functional aspects of the human autophagy pathway, we focused on uncovering the mechanism responsible for the transcriptional regulation of the following human ATG genes: ULK1 and ULK2 (UNC51-like kinases 1 and 2, respectively), ATG2A, ATG2B, ATG3, ATG4A (AUTOPHAGIN 2), ATG4B (AUTOPHAGIN 1), ATG4C, ATG4D, ATG5, BECN1 (BECLIN 1), ATG7, GABARAP (gamma-aminobutyric acid type-A receptor-associated protein), GABARAPL1 (APG8L), GABARAPL2 (GATE16), MAP1LC3A and MAP1LC3B (microtubule-associated protein 1, light chain 3, alpha and beta, respectively), ATG9A, ATG9B, ATG10, ATG12, ATG16L1, WIPI1 (WD40 repeat protein interacting with phosphoinositides 1), WIPI2, WIPI3 (WDR45L), WIPI4 (WDR45), and DRAM (damage-regulated autophagy modulator). Most of these numbered genes were named in accordance with their homologous counterparts found in model organisms, but exceptional gene names also exist. ULK1 and ULK2 have been identified as mammalian Atg1 homologs [17, 18]. BECN1 has been shown to be structurally similar to the yeast Vps30 (Atg6) gene [19]. GABARAP, GABARAPL1, GABARAPL2, and MAP1LC3 have been recognized as yeast Atg8 homologs [20]. And WIPIs are considered to be homologous to yeast Atg18 [21].

Among the above-mentioned ATG genes, the transcriptional regulations of ULK1, ATG5, MAP1LC3B, and DRAM are known to be governed by the transcription factor E2F1, and DRAM is also regulated by p53 [22, 23]. These findings suggest that transcriptional regulation might be involved in the link between autophagy and apoptosis, since E2F1 and p53 together play a pivotal role in the induction of apoptosis by orchestrating the expressions of their target genes [24]. In addition, the possible role of autophagy in cancer has been partly explained by the fact that DRAM is inactivated in certain cancers [25]. Moreover, human cells carrying mono-allelic deletions of the BECN1 gene and Becn1 heterozygous mice clearly showed a tendency toward tumorigenesis [19, 26, 27], indicating that subtle transcriptional deregulation might be a rate-limiting step in autophagy-involved human pathogenesis.

In the present study, we conducted a transcriptional analysis of human ATG genes by examining the distributions of their mRNA expressions during tissue development and etoposide-induced apoptosis. Our results clearly suggest that the majority of human ATG genes are ubiquitously expressed in adult and fetal tissues; however, a few genes are expressed in a tissue-specific manner. Interestingly, ATG2A was uniquely upregulated during the etoposide- and doxorubicin-induced apoptosis of HeLa cells, suggesting that ATG2A could be a novel biomarker of topoisomerase II inhibitor-mediated apoptosis. We also succeeded in establishing promoter-luciferase constructs of 23 human ATGs, and this set of constructs might be useful for examining how autophagy is involved in other cellular processes in human somatic cells, at least at the transcriptional level.

Materials and methods

Cell culture and chemicals

Human HeLa cervical adenocarcinoma cells were cultured in Earle’s modified Eagle’s medium (MEM) (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS), 1% non-essential amino acids (Invitrogen), and antibiotic-antimycotics (Invitrogen) in a 5% CO2 humidified atmosphere at 37°C. Etoposide and doxorubicin hydrochloride were purchased from Wako Pure Chemical Industries (Osaka, Japan). Chemicals were dissolved in dimethyl sulfoxide (DMSO) as 100 mM stock solutions and kept at −20°C until dilution before use. The final concentrations of etoposide and doxorubicin were 1 × 10−5 and 1 × 10−6 M, respectively.

In silico analysis

The expression profile was determined (May 29, 2009) based on the EST (expressed sequence tag) counts in human tissues and organs in the UniGene database (EST Profile Viewer, NCBI). The data sets used for the human ATG genes are listed in Supplementary Table 2. The number of transcripts per million (TPM) was calculated based on the gene EST/total EST in the pool, and this value was exported to an Excel file. Transcription factor binding motif was searched by TRANSFAC algorithm (threshold > 85) supported by GenomeNet (http://www.genome.jp/).

Reverse transcription-polymerase chain reaction (RT-PCR)

Total RNA was purified using an RNeasy mini kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions. Two micrograms of total RNA were reverse transcribed using High-Capacity cDNA Reverse Transcription kits (Applied Biosystems, Foster City, CA). The PCR was carried out in 25 μl of a mix consisting of 1× buffer, 200 μM dNTPs, 400 nM primers, 1 mM MgSO4, 5% DMSO and 1 unit of KOD plus DNA polymerase (Toyobo, Osaka, Japan). Hot-start PCR was then performed as follows: denaturation for 3 min at 94°C followed by ad libitum cycles at 94°C for 15 s, a gene-specific annealing temperature (Table 1) for 30 s, and 68°C for 30 s, followed by an extension step of 2 min at 68°C. The PCR results were verified by varying the number of PCR cycles for each cDNA and set of primers. The target gene primer pairs were listed in Table 1. The amplified products were separated on 1.0 ~ 1.5% agarose gels and visualized under ultraviolet transillumination. The gel images were processed using Quantity One (Bio-Rad Laboratories, Hercules, CA). Briefly, each specific band was quantified by volume analysis using the background subtraction method and the resulting density was represented as the ratio of etoposide/DMSO. The target gene bands were also compared to the corresponding GAPDH (glyceraldehyde-3-phosphate dehydrogenase; R&D Systems, Minneapolis, MN) band. For the cDNA panel analysis, 2.5 μl of cDNA purchased from Clontech (Mountain View, CA) was used (Human MTC panel I and II, Fetal MTC panel, and MCF7 apoptosis cDNA panel). The G3PDH primer in the kit was used as a control to amplify 983-bp.

Table 1 Sequence of primers used for RT-PCR
Full size table

Quantitative real-time RT-PCR

Real-time PCR was performed using a MiniOpticon machine (Bio-Rad Laboratories) with Power SYBR Green PCR Master Mix (Applied Biosystems). All the primers were tested to produce proportional changes in the threshold cycle with various initial quantities of cDNA. The measured threshold cycles (Ct) were converted to relative copy numbers using primer-specific standard curves. The Ct values were used to plot a standard curve in which Ct decreased in linear proportion to the log of the template copy number. The correlation values of the standard curves were always >90%. The gene expression levels induced by etoposide were normalized by the average levels of the mean-centered housekeeping gene GAPDH. All the samples were assayed in duplicate. All the values in the experiments were expressed as the mean ± SD (standard deviation) and were compared using two-tailed Student t-tests.

Promoter plasmids

Promoter fragments were generated using the PCR method from human genomic DNA (Promega, Madison, WI) and were ligated into the respective enzyme sites (Table 2) of the pGL3-Basic vector (Promega). The PCR primers are listed in Table 2. All the plasmid sequences were verified by sequencing at the Takara facility (Mie, Japan). Differences found by sequencing were compared with the deposited reference sequence in GenBank (Accession numbers are shown in Table 2). Differences in the following plasmid sequences were identified: ATG3 (G was changed to C at −221 position), ATG7 (T was changed to A at −199 position), GABARAPL1 (T was changed to C at −144 position), MAP1LC3B (A was changed to G at −474 position), ATG10 (T was deleted from T stretch at −479 position), and DRAM (G was changed to A at −417 position), where the minus position is relative to a transcriptional start site at +1.

Table 2 Sequence of primers used for promoter-luciferase constructs
Full size table

Luciferase assay

For the promoter assay, 2 × 104 cells were transfected with FuGENE6 (Roche, Basel, Switzerland) according to the manufacturer’s instructions. Briefly, 200 ng of the expression plasmid (pcDNA3 and pcDNA3-E2F1), 200 ng of the firefly luciferase reporter plasmids pGL3-Basic (Promega) or pTA-Luc, pE2F-TA-Luc (Clontech), and 0.6 ng of the Renilla luciferase reporter plasmid pRL-TK (Promega) per 24-well dish were used for each transfection. Cells were lysed 24 h after transfection by applying 100 μl of Passive Lysis Buffer of the Dual Luciferase Reporter Assay Kit (Promega) into each well of the 24-well plate. Five microliters of cell lysate were used for the luciferase reporter assay with the same kit, according to the manufacturer’s protocol. Light intensity was quantified in a GloMax 20/20n Luminometer (Promega). The experiments were performed at least in triplicate. As a control for the transfection efficiency, the firefly luciferase activity values were normalized to the Renilla luciferase activity values. Data are presented as the mean value ± SD. Statistical differences were analyzed using two-tailed Student t-tests. A value of P < 0.05 was considered statistically significant.

Caspase-3/7 assay

The induction of apoptosis was assessed using a Caspase-Glo 3/7 assay according to the manufacturer’s instructions (Promega). Briefly, the cells were inoculated into a 96-well plate at a concentration of 1 × 104 cells/well 24 h before treatment. The cells were then treated with chemicals for 48 h, followed by 1 h of incubation with Caspase-Glo 3/7 substrate at room temperature, and the resulting activity was measured using a GloMax 20/20n Luminometer (Promega). The activity was presented as relative light units (RLU, ×106) in the treated cells relative to the control value (treated with 0.1% DMSO). Blanks were measured in wells containing 0.1% DMSO or chemicals without cells. The data are presented as the mean values ± SD. Statistical differences were analyzed using a two-tailed Student t-test. A value of P < 0.05 (n = 3) was considered statistically significant.

Results

Expression patterns of human ATG genes in adult and fetal tissues

Reports on the mRNA expression patterns of human ATG genes have been increasing in number; however, these reports have utilized different detection methods, such as Northern blot or RT-PCR, and different mRNA resources (summarized in Supplementary Table 1). Alternatively, EST counts for sequenced cDNA pools can be utilized as predictable information on the mRNA distribution patterns in various tissues and developmental stages for most of the human ATG genes (summarized in Supplementary Table 2); however, these data should be verified using a single detection method and a uniform cDNA resource. To achieve this goal, we performed RT-PCR for 23 human ATG genes in 16 adult tissues (heart, brain, placenta, lung, liver, skeletal muscle, kidney, pancreas, spleen, thymus, prostate, testis, ovary, small intestine, colon, and leukocyte) and 8 fetal tissues (brain, lung, liver, kidney, heart, spleen, thymus, and skeletal muscle) (Fig. 1). Most of the human ATG genes as well as G3PDH were ubiquitously expressed in the adult and fetal tissues that we examined, with a few exceptions (Fig. 1). Of note, the expression of ATG2A was not detected in adult tissues including the heart, brain, placenta, lung, liver, and skeletal muscle (Fig. 1). To our knowledge, the expression pattern of ATG2A has not been previously reported. ATG7 was expressed faintly in adult heart (Fig. 1), though the EST count both in literature and the UniGene database predicted that ATG7 would be expressed ubiquitously (Supplementary Tables 1 and 2) [28]. ATG9B showed a characteristic expression pattern in several RT-PCR analyses. Namely, ATG9B was abundantly expressed in some adult (placenta and testis) and fetal (brain, lung, and thymus) tissues, whereas none or a minimal expression level was observed in other adult (heart, lung, liver, skeletal muscle, kidney, pancreas, ovary, and colon) and fetal (liver, heart, and skeletal muscle) tissues (Fig. 1), though previously reported Northern blot and RT-PCR analyses predicted an ubiquitous pattern with the highest expression level thought to occur in the placenta (Supplementary Table 1) [29, 30]. The ATG9B expression pattern predicted by the EST count was tissue specific, with abundant expression in the placenta and ovary and medium or minimal expression in the lung, testis, liver, muscle, brain, and pancreas (Supplementary Table 2). These results support the notion that ATG9B mRNA exists in specific tissues and is most abundantly observed in the placenta. WIPI2 has been reported to exhibit a ubiquitous expression pattern, with the highest levels observed in the heart, skeletal muscle, and pancreas (Supplementary Table 1) [21]; however, our repeated RT-PCR experiments showed that WIPI2 was undetected in adult tissues including the heart, brain, placenta, lung, skeletal muscle, and kidney (Fig. 1). At present, we have no explanation for this discrepancy. DRAM expression was ubiquitous, with the highest level observed in the pancreas (Fig. 1), whereas the EST counts failed to show a high count in the pancreas and no counts were observed for the spleen (Supplementary Table 2). Together, these findings suggest that our comprehensive mRNA analysis of human ATG genes might be superior to EST counts and also to previous reports of individual experiments that were incomplete for most human ATG genes.

Fig. 1
figure 1

Expression patterns of human ATG genes detected using RT-PCR. Multiple human adult (Left panel) and fetal (Right panel) tissues are indicated at the top. The left panel includes a cDNA derived from HeLa cells. Markers (100-bp ladder) are shown on the left side of the panel, with the 500-bp band highlighted. Gene names are shown on the right side of the panel. The value in brackets is the PCR cycling number. Note that different cycling numbers were used for ATG4B, MAP1LC3B, and G3PDH in the adult and fetal tissues

Full size image

Changes in the expressions of human ATG genes in apoptotic cells

DRAM is a representative example of a gene that links autophagy and apoptosis, since DRAM is an essential factor for p53-mediated apoptosis and is also involved in autophagy [22]. Therefore, we checked the mRNA levels of human ATG genes regulated by apoptosis-inducing etoposide. HeLa cells were treated with 1 × 10−5 M of etoposide for 48 h, which was sufficient to induce caspases-3/7 activity (Fig. 2a). In addition, the accumulation of p53 protein accompanying an elevated phosphorylation level of serine 15 was detected 48 h after etoposide treatment, compared with the level in a DMSO vehicle treatment group (data not shown). Amongst the genes that were examined, the ATG2A and ATG4C mRNA levels were uniquely upregulated and downregulated, respectively, with a statistically significant difference (Fig. 2b, c). The expression level of DRAM was unchanged (Fig. 2b, c). We focused on ATG2A as an upregulated gene and further examined changes in the ATG2A transcript levels in HeLa cells using real-time PCR. Etoposide treatment increased the ATG2A mRNA level by 2.1-fold (SD = 0.0007, two-tailed Student t-test, P = 0.02, n = 3).

Fig. 2
figure 2

Changes in the expressions of human ATG genes during apoptosis, as detected using RT-PCR. a Caspase-3/7 activity induced by etoposide. HeLa cells were treated with 1 × 10−5 M of etoposide for 48 h; the relative light units (RLU, ×106) are expressed as the mean ± SD (n = 3). * P < 0.05 (compared with DMSO-treated cells), using a two-tailed Student t-test. b The mRNA levels induced by etoposide are shown as the relative intensity quantified by Quantity One software. The expression levels were divided by the DMSO-treated expression levels. Mean ± SD. * P < 0.05 compared with DMSO-treated control using a Student two-tailed t-test (n = 2). c Representative agarose gel images of RT-PCR results for etoposide-induced mRNA level changes. Markers (100-bp ladder) are shown on the left side of the panel, with the 500-bp band highlighted. Gene names are shown on the right side of the panel. The value in brackets is the PCR cycling number. d Caspase-3/7 activity induced by doxorubicin. HeLa cells were treated with 1 × 10−6 M of doxorubicin for 48 h; the relative light units (RLU, ×106) are expressed as the mean ± SD (n = 3). * P < 0.05 (compared with DMSO-treated cells), using a two-tailed Student t-test. e ATG2A mRNA expression level in doxorubicin-treated or -untreated (DMSO) HeLa cells. GAPDH was detected as a control. The value in brackets is the PCR cycling number. f ATG2A mRNA expression level in doxorubicin-treated (7 and 14 h) or -untreated MCF7 cells. G3PDH was detected as a control. The value in brackets is the PCR cycling number. g ATG2A mRNA expression level in normal medium or serum withdrawal for 48 h from HeLa cells. Filled rectangular triangle indicates serial doubling dilution of cDNA template. GAPDH was detected as a control. The value in brackets is the PCR cycling number

Full size image

We reasoned that the upregulation of ATG2A mRNA might be a common phenomenon in apoptotic HeLa cells. Consequently, we treated the cells with doxorubicin; caspases-3/7 activity was clearly elevated in these cells, compared with DMSO-treated cells (Fig. 2d). The ATG2A mRNA level was upregulated in the doxorubicin-treated HeLa cells (Fig. 2e). In another cell line, MCF7, the ATG2A mRNA level was also upregulated by doxorubicin treatment in a time-dependent manner (Fig. 2f). In contrast, serum withdrawal for 48 h from HeLa cells failed to upregulate ATG2A mRNA level (Fig. 2g), suggesting ATG2A mRNA upregulation would be specific phenomenon of topoisomerase II inhibitor-mediated cell stress.

Regulation of promoter region of human ATG genes

Next, we focused on characterizing the promoter region of human ATG genes. For this purpose, we amplified and cloned the promoter regions of 23 human ATG genes into a pGL3-Basic luciferase reporter (Table 2). Recently, E2F1 has been reported to induce ATG1, MAP1LC3B, and DRAM directly and ATG5 indirectly [23]. Using a promoter reporter set, we checked whether the human ATG genes were responsive to E2F1 ectopic expression. Ectopic E2F1 expression in HeLa cells upregulated the luciferase activity of pE2F-TA-Luc (Fig. 3a). Surprisingly, a large number of promoter reporters, including those for ULK2, ATG4A, ATG4B, ATG4D, ATG7, GABARAPL2, MAP1LC3A, MAP1LC3B, ATG9A, ATG10, ATG12, and DRAM, were upregulated, based on the luciferase activity, whereas the promoter reporter of GABARAPL1 was downregulated (Fig. 3b). The ULK1 promoter region used in the luciferase assay was unchanged by E2F1 co-transfection, whereas a previous study utilizing the ULK1 promoter reported upregulation after E2F1 co-transfection [23]. This discrepancy might have been caused by the shorter ULK1 promoter region that was used in this study (Fig. 3c), compared with that used in the previous study. Luciferase activity induced by ectopic E2F1 expression was not necessarily corresponding to the existence of predicted E2F consensus in the promoter region. In the presently reported promoter construct for DRAM, a p53 expression vector failed to induce luciferase activity (data not shown), suggesting that the p53-responsive sequences might be located outside of the −444 ~166 region.

Fig. 3
figure 3

Promoter-luciferase reporters of human ATG genes. HeLa cells were transfected with the indicated plasmids and luciferase activity was measured 24 h after transfection. For further details, please refer to the Materials & Methods section. a Induction level of reporter construct pE2F-TA-Luc bearing an E2F-binding sequence upstream of the luciferase gene in an E2F1 expression vector. The luciferase activity of pTA-Luc co-transfected with pcDNA3 (white bar) or pcDNA3-E2F1 (black bar) was set as 1, and the relative fold induction of pE2F-TA-Luc is shown. The values reported for the transfection experiments are the mean ± SD (n = 3, * P < 0.05). b Responsiveness of promoter region of human ATG genes to E2F1 expression. The luciferase activity of pGL3-Basic co-transfected with pcDNA3 (white bar) or pcDNA3-E2F1 (black bar) was set as 1, and the relative fold inductions of the indicated gene promoter constructs are shown. The values reported for the transfection experiments are the mean ± SD (n = 3, * P < 0.05). c The map of the promoter region of human ATG genes used for luciferase reporter constructs. The transcription start sites are indicated by the bent arrows and designated as “+1”. Positive (negative) numbers are assigned to nucleotides downstream (upstream) of nucleotide +1. Lines with numbers are the regions used for the luciferase assay. The black box indicates the putative E2F binding site predicted by TRANAFAC (cut off score > 85)

Full size image

Discussion

In this study, we attempted to present a comprehensive picture of the transcriptional regulation of human ATG genes. First, we used uniform cDNA resources derived from various adult and fetal tissues for the PCR template and to verify whether human ATG genes are commonly or uniquely transcribed and to what extent, since a detailed examination of the mRNA distribution of human ATG genes has not been previously reported. Based on information available in medical literature, human ATG genes tend to be expressed ubiquitously [21, 28–39]. On the other hand, some exceptions exist. For example, ULK2 mRNA is abundant in ovary and testis, whereas little or no expression is present in the spleen [40]. ATG4B was highly expressed in skeletal muscle, but its expression was not detected in fetal tissues [33]. MAP1LC3A expression was not detected in the thymus or peripheral blood leukocytes [38]. Our results suggest that ATG2A, ATG9B, and WIPI2 may be exceptional in that they exhibited specific mRNA expression patterns in certain adult tissues. ATG9B also showed a unique expression pattern in fetal tissues. Another characteristic feature of our analysis was that the mRNA expression levels of human ATG genes in the heart tended to be lower than that in other tissues. Autophagy is known to be maintained at a low basal level in the heart for the turnover of organelles under normal conditions and it is upregulated in response to stresses such as ischemia/reperfusion and in cardiovascular diseases such as heart failure [41].

The present results give an overall picture of the tissue distribution of ATG genes and therefore may provide a clue as to the establishment of conditional knockout mice. Observations using a mouse model suggest that autophagy plays critical roles in early or specific developmental processes [42]. For example, Atg5 null mice showed almost normal phenotypes at birth but died within 1 day of delivery as a result of energy depletion arising from reduced amino acid concentrations in plasma and tissues [43]. Further analysis of Atg5 null mice revealed that an increased number of apoptotic cells were observed in the retina and lung, partly because of a defect in dead cell clearance [44]. Neural-cell-specific Atg5 null mice showed progressive defects in motor function [45]. Additionally, Atg5 has been shown to be required for T and B lymphocyte survival during development [46, 47]. Unlike Atg5 null mice, Atg4C null mice were viable and fertile and did not display any obvious abnormalities [33], whereas Becn1 null mice died early during embryogenesis [27]. Atg7 deficiency in the central nervous system resulted in behavioral defects and resulted in death within 28 weeks of birth [48]. The conditional loss of Atg7 revealed that Atg7 is required for the starvation-induced degradation of proteins and organelles [49]. Mice carrying a beta cell-specific deficiency in Atg7 showed impaired glucose tolerance and a decreased serum insulin level [50]. Atg8 null mice died within 24 h after birth, similar to the Atg5 null mice [51]. An analysis of Atg16L1 deficient or hypomorphic mice demonstrated that Atg16L1 is responsible for controlling the endotoxin-induced inflammatory immune response and can serve as a model for Crohn’s disease based on its characteristic role in the intestinal epithelium [52, 53].

The relation between autophagy and apoptosis can be explained by the fact that BCL-2, an anti-apoptotic protein, can function as an anti-autophagy protein via its inhibitory interaction with BECN1 [54]. In addition, ATG5 was originally identified as an upregulated protein during apoptosis [55]. Recently, Drosophila Atg1 overexpression has been shown to induce autophagy and apoptosis [56], indicating physiological levels of ATG genes are essential for autophagy, but unregulated levels of some ATG genes, such as ATG2A as well as Drosophila Atg1, would tend to be ended as inappropriate autophagy and apoptosis. Accumulated evidences imply that the spatiotemporal architecture of protein complex formation is crucial for determining the involvement of autophagy in other cellular machinery; however, transcriptional regulation must be a somewhat important regulatory step. For example, BNIP3, a hypoxia-inducible member of the BCL-2 family, is an essential regulator of hypoxia-induced autophagy and acts as a direct target of E2F1 [57]. Uniquely, E2F1 possesses dual roles in cell fate determination, such as cell cycle progression and apoptosis [24]. We demonstrated that 13 out of 23 human ATG gene promoters were regulated by E2F1, strongly indicating the mechanisms whereby autophagy pathway crosstalk with cell cycle regulation as well as apoptosis. We observed that ATG2A mRNA was specifically upregulated by etoposide in HeLa cells. The observed discrepancy between E2F1-regulated 13 promoters and etoposide-induced ATG2A mRNA can be explained by the fact that the etoposide-induced apoptosis is regulated not only by E2F1 but also by other multiple factors. A previous DNA microarray approach identified the upregulation of several ATG genes, including Atg2A, in dying salivary glands in Drosophila [58]. Basically, autophagy promotes cell survival, but in certain circumstances, autophagy contributes to cell death [59]. Together, these results suggest that the transcriptional regulation of ATG2A might be a unique indicator of autophagic programmed cell death, where apoptosis and autophagy are integrated to determine cell fate. To gain further insight into the role of ATG2A in autophagic cell death, functional assays should be performed in the future.

In summary, we have described the tissue distribution patterns of mRNAs for 26 human ATG genes. In addition, 23 promoter reporter constructs were produced in this study, and these constructs should serve as a unique tool for identifying signaling molecules that might be integrated into mTOR (mammalian target of rapamycin), a sensor kinase of growth factors and nutrient levels that regulates cellular growth and autophagy. Our present results imply that the transcriptional control of human ATG genes might partially contribute to human physiology and pathophysiology.