ATF2, a paradigm of the multifaceted regulation of transcription factors in biology and disease

Summary

Stringent transcriptional regulation is crucial for normal cellular biology and organismal development. Perturbations in the proper regulation of transcription factors can result in numerous pathologies, including cancer. Thus, understanding how transcription factors are regulated and how they are dysregulated in disease states is key to the therapeutic targeting of these factors and/or the pathways that they regulate. Activating transcription factor 2 (ATF2) has been studied in a number of developmental and pathological conditions. Recent findings have shed light on the transcriptional, post-transcriptional, and post-translational regulatory mechanisms that influence ATF2 function, and thus, the transcriptional programs coordinated by ATF2. Given our current knowledge of its multiple levels of regulation and function, ATF2 represents a paradigm for the mechanistic complexity that can regulate transcription factor function. Thus, increasing our understanding of the regulation and function of ATF2 will provide insights into fundamental regulatory mechanisms that influence how cells integrate extracellular and intracellular signals into a genomic response through transcription factors. Characterization of ATF2 dysfunction in the context of pathological conditions, particularly in cancer biology and response to therapy, will be important in understanding how pathways controlled by ATF2 or other transcription factors might be therapeutically exploited. In this review, we provide an overview of the currently known upstream regulators and downstream targets of ATF2.

Activating transcription factor 2 (ATF2)

Activating transcription factor 2 (ATF2)—also known as cyclic AMP (cAMP) response element (CRE) binding protein 2 (CREB2) and CRE-BP1—is a member of the activating protein-1 (AP1) transcription factor family that regulates the expression of many genes through homo-dimerization or hetero-dimerization with other AP1 family members, such as the CREB, Fos, Maf, or Jun family transcription factors (1, 2). ATF2 was first identified in a yeast screen for CRE-binding proteins in 1989 (3) and has subsequently been characterized as an important mediator of mammalian cell responses to various stimuli, including stress.

ATF2 gene

The ATF2 gene is located on chromosome 2 (2q32) (http://www.ncbi.nlm.nih.gov/gene/1386). As a result of mRNA splicing or alternative promoter usage, combinations of the 15 exons comprising the full length form of ATF2 can reportedly give rise to 21 annotated transcripts, 8 of which contain an open reading frame (Alternate Splicing Gallery ENST00000264110, http://statgen.ncsu.edu/asg/index.php?lookupType=&lookupValue=ENST00000264110&canvasWidth=1000; and e!Ensemble ENSG00000115966, http://useast.ensembl.org/Homo_sapiens/Gene/Summary?db=core; g=ENSG00000115966;r=2:175072250-175168382) (4). ATF2 is ubiquitously expressed and the gene is required for normal development (3, 5). Complete loss of ATF2 leads to early post-natal lethality characterized by meconium aspiration syndrome (6), whereas mutations that compromise ATF2 function lead to early mortality with incomplete penetrance and an array of abnormalities, most notably neurological defects, in surviving animals (7–9). ATF2 therefore plays an important role in signaling during early development.

Transcriptional and post-transcriptional regulation of ATF2

The transcriptional regulation of ATF2 is not fully characterized, although promoter analyses suggest that expression may be controlled by a number of transcription factors, including CRE-responsive factors, SP1, AP2/4, Sox 1/5/9, androgen receptor, and CCAT/enhancer-binding protein (C/EBP) (10–12).

A number of ATF2 splice isoforms have been reported, which contain or lack important regulatory or DNA-binding regions of ATF2 (http://www.ncbi.nlm.nih.gov/gene/1386). Among these, 3 have been characterized in humans. The most extensively studied of the isoforms is the full-length 505-amino acid (aa) isoform (Figure 1). A less well-characterized truncated isoform, termed ATF2-small (ATF2-sm), is expressed in myometrial tissues and is subject to differential expression during labor (13, 14). Despite the fact that ATF2-sm contains only exons 1, 2, 14, and 15 of full-length ATF2 and lacks well-defined phosphoregulatory sites (within exons 1 and 2) and the classical bZIP domain, it was nonetheless reported to exhibit transcriptional activity and to alter the expression of pregnancy- and labor-associated genes, suggesting that ATF2-sm is subject to as-of-yet undefined regulation.

Figure 1.

ATF2 structure and phosphorylation sites. (A) Structure of full length ATF2 mRNA (NM_001256090.1, generated in SnapGene®Viewer v3.0.3) (B) The canonical full-length ATF2 protein is 505 amino acids (aa) and has a predicted molecular mass of ~55 kilodalton. Under basal conditions ATF2 is auto-inhibited through intra-molecular contact between the N-terminal domain and C-terminal basic leucine zipper (bZIP) domain. Following activation by kinases that phosphorylate the N-terminal region, the protein becomes activated and is competent to translocate to the nucleus and modulate gene expression in cooperation with other AP1 transcription factors (phospho-site indicated by red arrow). Protein motifs that influence sub-cellular localization have been characterized within and around the bZIP domain (nuclear localization signal, NLS; nuclear export signal, NES). ATF2 can be phosphorylated by many kinases: T52 - PKC-epsilon; S62 - VRK1; T69/T71 - extracellular signal regulated kinase (ERK), JNK, p38, polo-like kinase 3 (PLK3); T73 - VRK1; S121 - PKC; S340 - PKC; S367 - PKC; S490/S498 - ATM.

Recently, Claps et al. (15) reported the expression of a truncated ATF2 isoform, ATF2SV5, which lacks aa 210-505. ATF2SV5 lacks the classical bZIP domain, similar to ATF2-sm. Although it is expected to be transcriptionally inactive, its expression nonetheless elicits effects on metastasis-related genes (e.g., CCL4, CCR7, and S100A8). Furthermore, the expression of ATF2SV5 increases the migratory and colony-forming capacity of melanoma cells harboring BRAF^V600E and correlates with poor clinical outcome. How ATF2SV5 expression elicits such effects remains to be determined. However, an exclusively cytoplasmic alternative splice isoform of the AP1 family member ATF7 was reported to inhibit ATF7- and ATF2-mediated transcription by sequestering upstream activating kinases (16); a similar mechanism may also contribute to changes in transcription when alternative isoforms of ATF2 are expressed.

These findings attest to the complex and diverse functions of ATF2 splice variants and suggest that they interfere with other AP1 transcription factors, transcriptional complexes, and upstream regulators. Given the significantly divergent effects of ATF2 splice isoform expression, future studies investigating the regulated production of these isoforms as well as their specific contributions to general and pathological biology are warranted. Studies of transcription factor splice isoform expression and function (17), particularly in pathological conditions such as cancer, have been limited (e.g., MITF/TFE transcription factors (18), RREB1 (19), STAT3β (20), and p53 (21)). Future investigations of the splice isoforms of other key transcription factors are expected to clarify their complex and often reportedly divergent or elusive functions.

The stability of ATF2 transcripts is also subject to regulation. For example, the RNA-binding protein ELAV-like protein 1 (ELAVL1, HuR) can stabilize ATF2 transcripts via binding to the 3′-untranslated region (3′-UTR) (22). In contrast, a number of microRNAs, including miR-26a/b, miR-204, miR-451, and miR-622, have been shown to bind to the 3′-UTR of ATF2 transcripts to promote their degradation (23–27). The miRNA-mediated regulation of ATF2 is important for both biological stress- and tumor-related signaling. In terms of stress signaling, miR-26a and miR-26b were reported to maintain low basal ATF2 levels until cellular stress/stimuli (e.g., γ-irradiation or Toll receptor stimulation) resulted in their reduction and the derepression of ATF2 expression. In terms of tumor development, the expression levels of miR-204, miR-451, and miR-622 are reduced in glioblastoma, hepatocellular carcinoma, and glioma cells, resulting in increased ATF2 expression and ATF2-mediated tumorigenesis.

Post-translational regulation of ATF2

ATF2 structure and regulation

The full-length ATF2 protein contains several domains that control its activity, localization and ability to interact with other AP1 transcription factors, all of which influence its function (Figure 1). These include an N′-terminal zinc finger (ZnF), a transcriptional activation domain (TAD), and a basic leucine zipper (bZIP) domain (Figure 1). In its inactive state, the N-terminal TAD and C-terminal bZIP DNA-binding domain interact and inhibit the ability of ATF2 to activate transcription (28, 29). Following post-translational phosphorylation (discussed below) this intramolecular autoinhibition is relieved, and ATF2 is able to translocate to the nucleus as homo- or heterodimers with other AP1 transcription factors to modulate gene expression. The subcellular localization of ATF2 is largely controlled by the nuclear export (NES) and bipartite nuclear localization (NLS) signals coded into the C-terminus (30) (Figure 1). However, the N′-terminal 1–73 aa, which does not contain a canonical NES, is required for interaction with Exportin-1 and nuclear export (31). The presence of these localization signals facilitates dynamic shuttling of ATF2 between the nucleus and cytoplasm during basal conditions and following stimulus or stress (30).

In addition to the transactivation and nuclear export functions of the N′-terminus, amino acids 25–49 of ATF2 contains a conserved ZnF domain (Figure 1), which has been reported to interact with the leucine zipper region of other AP1 transcription factors (e.g., CREB) (32). However, dimerization between ATF2 and AP1 transcription factors is largely mediated by the bZIP domain that is highly conserved among AP1 transcription factors (1). Thus, ATF2 transcriptional activity is influenced through both intra- and inter-molecular interactions.

Regulation of ATF2 through post-translational modification

ATF2 plays an important role in transducing extracellular signals to the nucleus to facilitate transcriptional responses to stimuli (33). ATF2 can be activated by many stimuli, including growth factors, ultraviolet (UV) radiation, and cytokines. ATF2 transcriptional activation is mediated by stress-activated protein kinases (SAPKs) (e.g., p38 (34)) through phosphorylation at amino acids threonine-69 and threonine-71 (T69, T71) (33, 35–37), although the involvement, cooperation and/or necessity of specific kinases appears to be stimulus-dependent. For example, phosphorylation at these threonine residues is required for maximal transcriptional activity downstream of insulin/epidermal growth factor stimulation and is dependent on cooperation between c-Jun N-terminal kinase (JNK) and signaling cascades downstream of Ras (35, 38). Additional phosphorylation sites within this region have also been characterized and been shown to influence the ability of ATF2 to induce gene expression. For example, vaccinia-related kinase 1 (VRK1) can phosphorylate T73 and serine-62 (S62), resulting in T73-dependent transcriptional activation (39).

Following phosphorylation at T69/T71, ATF2 can interact with other AP1 proteins and translocate to the nucleus to modulate the expression of hundreds of genes (12, 14, 40). However, due to the diversity of possible ATF2-AP1 dimers, characterization of specific heterodimers involving ATF2 at discrete gene promoters has been limited. In response to genotoxic stress induced by cisplatin, AP1 dimers containing ATF2 and c-Jun were reported to bind to 122 gene promoters, many of which are involved in DNA repair and apoptotic signaling (40).

Posttranslational phosphorylation of ATF2 has also been characterized at other sites and has been shown to correlate with nuclear localization and affect the ability of ATF2 to trans-activate gene expression in response to 12-O-tetradecanoylphorbol-13-acetate (TPA) (41). Phosphorylation at S340 and S367 have been reported and occur within C-terminal NLS sequences and follow similar kinetics as T69/T71 phosphorylation, whereas S121 phosphorylation increases at later time-points. Phosphorylation at these sites is reported to be dependent on protein kinase C (PKC), but the particular isoform or whether PKC isoform is cell-type or stimulus dependent is not entirely clear.

PKC-dependent (specifically PKCε) ATF2 phoshorylation has also been characterized at T52 (42). Phosphorylation at this site stimulates ATF2 nuclear localization and contributes to transcriptional induction of genes that promote cell survival in response to stress. Chronic genotoxic stress attenuates PKCε-mediated phosphorylation of T52, allowing a pool of ATF2 to accumulate at the mitochondrial outer membrane, where it disrupts membrane integrity and promotes cell death induction via interaction with voltage-dependent anion-selective channel protein 1 (VDAC1) and hexokinase-1 (HK1). ATF2-mediated mitochondrial membrane leakage was reported to require activation of BIM (43).

A phosphorylation-mediated signaling axis independent of SAPK, c-Jun hetero-dimerization and transcriptional activity has also been characterized in response to DNA damage. ATF2 is phosphorylated by ataxia telangiectasia mutated (ATM) in response to ionizing radiation (IR) (44). Upon activation, ATM phosphorylates ATF2 on S490 and S498, which stabilizes its localization at γ-H2AX-marked sites of DNA double-strand break and facilitates the recruitment of the Mre11-Rad50-Nbs1 (MRN) DNA repair complex. This DNA damage responsive function of ATF2 depends on S490/S498 phosphorylation but is independent of SAPK and c-Jun activity and T69 and T71 phosphorylation, suggesting that this function is independent of its transcriptional activity.

In addition to phosphorylation, the transcriptional activity of ATF2 is also subject to regulation by acetylation. The histone acetyltransferase p300/CREB-binding protein (p300/CBP) acetylates ATF2 within the bZIP domain at Lys residues 357 and 374, enhancing the transcriptional activity of ATF2 (45). Whether and how this modification might affect the affinity/selectivity of ATF2 for its AP1 binding partners is not known.

Consistent with positive and negative feedback loops characteristic of rapidly activated stress-response proteins, mechanisms exist to both amplify ATF2 signaling following acute stress and to ensure ATF2 activation can be terminated following removal of a stimulus. For example, following cisplatin-induced stress, ATF2 and c-JUN are phosphorylated and activated by JNK, where they bind and activate c-JUN expression (40). An increase in activated c-JUN levels leads to nuclear retention of activated ATF2 and protects it from ubiquitination and proteasomal degradation (30, 46, 47). Positive feedback has also been noted in the context of other stimuli (UV and retinoic acid) where c-Jun phosphorylation is dispensable (30).

Conversely, negative feedback has been reported in the context of the embryonic liver, where stress-induced p38-mediated activation of ATF2 was reported to induce a negative feedback loop by transcriptionally activating phosphatases DUSP1/2/8 that attenuate p38 and ATF2 activity (48). ATF2 is also subject to proteasomal degradation, which, similar to its transcriptional activation, is also subject to regulation by phosphorylation and dimerization. The phosphorylated and transcriptionally activate dimers of ATF2 and its binding partners have been shown to be resistant to ubiquitination and proteasomal degradation compared to non-phosphorylated dimers or dimerization-incompetent mutant ATF2, suggesting that phosphorylation and activation of ATF2 protects it from degradation, likely until the stimulus or stress is removed (46, 47). The phosphatase(s) mediating this process, however, is unknown.

Taken together, these findings highlight the extent to which post-translational modifications can impact the structure, stability, localization, partner-binding affinity, and therefore activity and function, of ATF2. Such findings emphasize the importance of investigating the diverse effects of the post-translational modifications of other key transcription factors in effort to more fully appreciate and understand their regulation and functions.

ATF2 transcriptional programming

ATF2 has been characterized at the promoters of key genes involved in a variety of cellular processes, including inflammatory signaling (e.g., interferon β1 (IFNB1) (49)), cell cycle control (e.g., cyclin A, Maspin, and GADD45 (11, 50)), response to amino acid limitation (51) and glycosylation (52) (12, 53) (Table 1).

Table 1.

Transcriptional Targets of ATF2

	AP1 Binding Partner	Cell-type	Stimulus	Transcriptional Regulation	Target Gene	References
Cell cycle	ATF2, JunD	Rat chrondrosarcoma cells, MEF	Serum	TI	CCNA1	Shimizu et al., 1998
	CREB1	Murine chondrocytes	TGFβ, PTHrP	TI	CCND1	Beier et al., 2001; Beier et al., 1999
	JunD	Intestinal epithelial cells	Polyamines	TR	CDK4	Xiao et al., 2010
	BRCA1, Oct-1, Neurofibromin-1	MEF	Anisomycin, hypoxia	TI	GADD45	Maekawa et al., 2008; Maekawa et al., 2007
	n/a	TE7 esophageal squamous cancer cells	H2O2	TI	p21WAF1	Walluscheck et al., 2013
	n/a	Murine chondrocytes	n/a	TI	RB1	Vale-Cruz et al., 2008
	BRCA1, Oct-1, Neurofibromin-1	MEF	Anisomycin, hypoxia	TI	SERPINB5	Maekawa et al., 2008; Maekawa et al., 2007
Immune / Inflammatory	c-Jun	Endothelial (HUVEC)	UV	TI	ELAM1	Read et al., 1997
	CREB1	Myelogenous leukemia (K562)	Sodium butyrate, trichostatin A	TI	HBG1	Kodeboyina et al., 2010
	n/a	human melanoma cells	PKCε-phosphorylation	TR	IFNB1	Lau et al., 2015
	c-Jun	T-cells (Jurkat)	Ionomycin/PMA	TI	IFNG	Penix et al., 1996
	c-Jun	Human primary lung and foreskin fibroblasts	Interleukin-1β	TI	IL1B	Markovics et al., 2011
	c-Jun	Murine macrophage (RAW264.7); human dendritic cells	LPS; β-glucans	TI	IL23A	Al-Salleeh and Petro, 2008; Liu et al., 2009; Rodriguez et al., 2014
	n/a	Osteoblasts	PDGF	TI	IL6	Franchimont et al., 1999
Cell cycle	ATF2, JunD	Rat chrondrosarcoma cells, MEF	Serum	TI	CCNA1	Shimizu et al., 1998
	CREB1	Murine chondrocytes	TGFβ, PTHrP	TI	CCND1	Beier et al., 2001; Beier et al., 1999
	JunD	Intestinal epithelial cells	Polyamines	TR	CDK4	Xiao et al., 2010
	BRCA1, Oct-1, Neurofibromin-1	MEF	Anisomycin, hypoxia	TI	GADD45	Maekawa et al., 2008; Maekawa et al., 2007
	n/a	TE7 esophageal squamous cancer cells	H2O2	TI	p21WAF1	Walluscheck et al., 2013
	n/a	Murine chondrocytes	n/a	TI	RB1	Vale-Cruz et al., 2008
	BRCA1, Oct-1, Neurofibromin-1	MEF	Anisomycin, hypoxia	TI	SERPINB5	Maekawa et al., 2008; Maekawa et al., 2007
Immune / Inflammatory	c-Jun	Endothelial (HUVEC)	UV	TI	ELAM1	Read et al., 1997
	CREB1	Myelogenous leukemia (K562)	Sodium butyrate, trichostatin A	TI	HBG1	Kodeboyina et al., 2010
	n/a	human melanoma cells	PKCε-phosphorylation	TR	IFNB1	Lau et al., 2015
	c-Jun	T-cells (Jurkat)	Ionomycin/PMA	TI	IFNG	Penix et al., 1996
	c-Jun	Human primary lung and foreskin fibroblasts	Interleukin-1β	TI	IL1B	Markovics et al., 2011
	c-Jun	Murine macrophage (RAW264.7); human dendritic cells	LPS; β-glucans	TI	IL23A	Al-Salleeh and Petro, 2008; Liu et al., 2009; Rodriguez et al., 2014
	n/a	Osteoblasts	PDGF	TI	IL6	Franchimont et al., 1999
Cell Death	c-Jun	Transformed human embryonic kidney (293T)	Oxidative stress (H2O2)	TI	ACHE	Zhang et al., 2008
	c-Jun	Endothelial (HUVEC)	Growth factors (VEGF, EGF)	TI	BCL2L1	Salameh et al., 2010
	JDP2	HeLa, HEPG2, MEF	Amino acid deprivation	TR	CHOP	Averous et al., 2004
	c-Jun	Rat cerebellar granule neurons (CGN), murine immortalized gonadotrope cell line (αT3-1)	Potassium withdrawal	TI	DP5	Ma et al., 2007; Towers et al., 2009
AP1	c-Jun	Normal fibroblasts, HeLa, HEK293, MEF, Rat cerebellar granule neurons (CGN)	IR, amino acids, potassium, gonadotropin-releasing hormone (GnRH), tolfenamic acid (TA)	TI	ATF3	Chaveroux et al., 2009; Fu et al., 2011; Kool et al., 2003; Lee et al., 2010; Mayer et al., 2008
	n/a	LβT2 gonadotrope cells	GnRH	TI	c-Jun	Lindaman et al., 2013
	c-Jun	Embryonal carcinoma (F9), rat cerebellar granule neurons, monkey kidney epithelial (COS1), human cervical carcinoma (HeLa, SKOV3), NIH3T3, hepatocellular carcinoma (HepG2, HuH7)	UV, MNNG, MMS, potassium withdrawal, E1A expression, amino acid deprivation	TI	JUN	Fu et al., 2011; Kawasaki et al., 1998; Yamasaki et al., 2009
	JDP2	Embryonal carcinoma (F9)	Basal, prior to RA-induced differentiation	TR	JUN	Jin et al., 2002
DUSP	n/a	Murine fetal hepatocytes	Anisomycin	TI	DUSP5	Breitwieser et al., 2007
	n/a	Murine fetal hepatocytes	Anisomycin	TI	DUSP8	Breitwieser et al., 2007
	c-Jun	Embryonic liver, sympathetic neurons	p39 feedback signaling, NGF- withdrawal	TI	MKP1/DUSP1	Breitwieser et al., 2007; Kristiansen et al., 2010
	n/a	Murine fetal hepatocytes	Anisomycin	TI	MKP5/DUSP10	Breitwieser et al., 2007
Other	c-Jun	Rat aortic endothelial cells	Thrombin	TI	ARG1	Zhu et al., 2010
	n/a	P3HR1 lymphoma cells	EBV viral protein Rta	TI	BZLF1	Lin et al., 2014
	JunB	Murine endothelial and endothelioma cells	CoCl2, hypoxia	TI	CBPB	Licht et al., 2006
	c-Jun	Rat osteosarcoma (ROS17/2.8, ROS25), primary murine calvarial osteoblast (MCC)	n/a	TI	COL24A	Matsuo et al., 2006
	n/a	Murine brain microvascular endothelial cells	ET-1-induced calcium release	TI	COX2	Lin et al., 2014
	n/a	Vascular smooth muscle cells	TGFβ	TI	CSRP2	Lin et al., 2008
	n/a	Adrenocortical carcinoma cells (H295R)	Angiotensin II, potassium	TI	CYB11B2	Nogueira and Rainey, 2010; Yarimizu et al., 2015
	n/a	Murine macrophages (RAW)	cAMP	TI	ENTPD1	Liao et al., 2010
	n/a	Human melanoma cell lines	n/a	TR	FUK	Lau et al., 2015
	n/a	Brown adipocytes	Norepinephrine	TI	FGF21	Hondares et al., 2011
	n/a	K562 cells	n/a	TI	G-CRE	Liu et al., 2013
	c-Jun	Murine embryonic stem cells	FGF2	TI	HES1	Sanalkumar et al., 2010
	MafA, Pdx1, Beta2	Rat insulinoma cells (INS-1), β-cell- derived cell lines (MIN6), HeLa	Forskolin, UV	TI	INS	Han et al., 2011; Hay et al., 2007
	c-Jun	SW480, HEK293E, AGS cells	Twist1	TI	ITGA5	Nam et al., 2015
	n/a	Caco-2 cells	IL-1β	TI	MLCK	Al-sadi et al., 2013
	n/a	SaOS-2 cells	Wnt5α	TI	MMP13	Yamagata et al., 2012
	c-Jun	K562 human leukemia cells	Simvastatin	TI	MMP2	Chen and Chang, 2014
	n/a	Rat brain astrocytes	LTA	TI	MMP9	Hsieh et al., 2012
	c-Jun	Caco-2 cells	dibutyryl cAMP, forskolin	TI	MRP2	Arana et al., 2015
	n/a	Choriocarcinoma cells (Jar), CHO	Hypoxia	TI	PDGFRA	Maekawa et al., 1999
	c-Jun, JunD, ATF2	NIH3T3, HEPG2	TPA, FGF2	TI	PLAU	Cirillo et al., 1999; D’Orazio et al., 1997
	n/a	Myoblasts (C2C12)	Exercise	TI	PPARGC1A	Akimoto et al., 2005
	c-Jun	Murine lymphocytic leukemia cells (L1210)	Doxorubicin	TI	PRKCD	Min et al., 2008
	n/a	Mouse As4.1 cells	TNFα, MG132	TR	REN	Desch et al., 2011
	NF-YA	Jurkat cells	UV	TI	RHOB	Ahn et al., 2011; Fritz and Kaina, 2001
	n/a	Murine macrophage (RAW264.7)	LPS	TI	SOCS3	Hirose et al., 2009
	n/a	Murine chondrocytes	BMP signaling	TI	SOX9	Gao et al., 2013
	n/a	Retinal pigment epithelial cells	Valproic acid	TI	ST3GAL5	Song et al., 2011
	n/a	Lung and intestinal epithelial cells	Rb expression, All trans retinoic acid	TI	TGFB2	Kim et al., 1992; Namachivayam et al., 2015
	n/a	Rat adrenal medulla-derived cells (PC12)	Nicotine	TI	TH	Gueorguiev et al., 2006; Suzuki et al., 2002
	NFAT, c-Jun	Dendritic cells; monocytic leukemia cells (THP1); murine macrophages; murine Kupffer cells	TLR2 ligation; bile acids	TI	TNF	Altmayr et al., 2010; Kumawat et al., 2010; Lawrence et al., 2011; Lou et al, 2014
	JunD	P19 mouse embryonal carcinoma cells	All trans retinoic acid	TI	TUBB4	Maruyama et al., 2014

TI: transcriptional induction

TR: transcriptional repression

Assessment of ATF2 target genes has largely been on a gene-by-gene basis, where putative ATF2-binding sites within specific promoters are characterized. This is likely due to the compositional diversity of AP1 dimers containing ATF2 with other AP1 transcription factors present or absent across different cell-types and during different stress conditions. Large-scale surveys of ATF2 targets, particularly in regard to promoters that are targeted by specific forms of ATF2 (e.g., phosphorylated species or splice isoforms), have been limited by the availability of good quality antibodies for ChIP analyses.

Previously, Hayakawa et al. identified 181 gene promoters in breast cancer cells that are bound by JNK-phosphorylated ATF2 following genotoxic stress (40). Of the 181 promoters identified, 23 belong to genes involved in DNA repair, indicating a crucial role for ATF2 in DNA damage response. We previously performed global gene expression profiling of the T52-phosphorylated species of ATF2 compared to the non-phosphorylatable mutant to define subsets of PKCε-ATF2-regulated genes. Based on those analyses, we inferred a role for ATF2 in controlling IFNB1 expression. We subsequently validated the binding of ATF2 to the IFNB1 promoter and characterized ATF2 as a transcriptional repressor at this locus. IFNB1 repression results in altered cell cycle progression and enhanced chemoresistance (49). More recently, we characterized ATF2 binding at the fucokinase (FUK) promoter and reported altered ATF2-mediated transcriptional repression of fucokinase in melanomas, resulting in globally reduced fucosylation and enhanced cellular motility in progressive melanomas (52). ATF2 likely binds to many more promoters than those characterized to date, although the specific promoters and whether binding enhances or represses gene expression is expected to be both stimulus- and cell-type-dependent. Future large-scale studies evaluating ATF2 localization at chromatin will provide important information concerning the role of ATF2 signaling in different cell types and help in understanding the outcome of ATF2 activity in different biological contexts (54).

ATF2 in disease

Consistent with the diversity of cellular processes reported to be transcriptionally regulated by ATF2, the number of disease pathologies associated with alterations in ATF2 are equally numerous. For example, increased phosphorylation of ATF2 is associated with chondrocyte apoptosis in Kashin-Beck disease (55), consistent with its previously defined role in osteoclast differentiation (56). In addition to its roles in inflammation-related signaling pathways, alterations in ATF2 expression and activity have also been linked to inflammation-related pathologies including obesity via regulation of adipocyte differentiation (57) and liver development, regeneration, and cirrhosis (48, 58, 59). The altered expression and phosphorylation state of ATF2 has also been associated with a diverse array of pathologies including multiple neurodegenerative pathologies, polycystic kidney disease, diabetic amylin-induced pancreatic beta cell death (60–62). The precise contributions of ATF2, in terms of causation or correlation, in these pathological contexts are unknown.

Transcriptional roles for ATF2 have been described in a number of cancer types. For example, ATF2 has been reported to promote the development and progression of synovial sarcomas by aberrantly binding to the oncogenic fusion protein SS18-SSX, an interaction that alters its transcriptional activity (63, 64). Altered ATF2 expression and localization have also been implicated in the pathology, progression, and chemoresistance of extramammary Paget’s disease, as well as in prostate and head and neck squamous cancers (65–68). Interestingly, loss of ATF2 has been shown to promote the development of mammary tumors in mouse models, likely via transcriptional deregulation of the cell cycle-related tumor genes Maspin and GADD45 (50, 69).

Similar to the cancers mentioned above, increased nuclear, phosphorylated ATF2 has been reported in several cutaneous malignancies including cutaneous angiosarcoma, pyogenic granuloma, Bowen’s disease, squamous and basal cell carcinoma, and eccrine porocarcinoma/poroma (70–73). The mechanisms underlying ATF2 function in cutaneous pathologies have been extensively studied in melanoma, where its dual oncogenic and tumor suppressive roles were first elucidated (42, 73). We previously showed PKCε phosphorylates ATF2, driving its nuclear accumulation and transcriptional activity while blocking its tumor suppressive function at the mitochondria. In melanomas with low PKCε expression levels, where phosphorylation of ATF2 is attenuated following therapeutic stress, ATF2 can execute its pro-apoptotic function at mitochondria. However, the upregulation of PKCε in progressive melanomas blocks this tumor suppressive function by driving nuclear ATF2 function, transcriptionally promoting motility, invasiveness, and the resistance of melanomas to a number of therapeutic agents (including genotoxic agents). Accordingly, increased PKCε levels and increased nuclear ATF2 both correlate with progressive clinical staging (42).

Implications in cancer therapeutics

Given the known DNA repair-related transcriptional targets of ATF2, it is not surprising that altered ATF2 activity in cancer cells can promote resistance to genotoxic stress-inducing therapeutic agents (40, 49). However, ATF2 has been functionally implicated in the development of resistance to non-genotoxic-inducing stress therapies. For example, ATF2 was identified in liver cancer cells to mediate resistance to the tyrosine kinase inhibitor Sorafenib (74). In the context of combination therapy, the RNAi-mediated depletion of ATF2 was found to augment the radiosensitization potential of the epidermal growth factor receptor inhibitor Cetixumab in non-small cell lung cancer cells (75). Thus, the targeting of ATF2, its upstream activators, or specific downstream targets represents a potentially important therapeutic approach. Indeed, the inhibition of ATF2 transcription by peptides containing the N′-terminal 50–100 aa of ATF2 sensitizes melanoma cells to apoptosis (76, 77). Notably, as these peptides contain the PKCε, JNK, and p38 phosphorylation sites of ATF2, they might inhibit ATF2 transcription by competing for phosphorylation and activation of the endogenous transcription factor. Consistent with these studies, small molecules identified to inhibit the phosphorylation of ATF2 by PKCε also effectively block melanoma proliferative capacity and motility, sensitizing cells to apoptosis (78).

However, much remains to be investigated in terms of the most effective way to therapeutically modulate ATF2 activity or its transcriptional targets that are identified as crucial for cancer biology and therapeutic resistance. For example, in contrast to the cancer types mentioned above, ATF2 appears to play a tumor suppressor role in breast cancer, where its transcriptional activity is required for sensitivity to tamoxifen-based treatments (79)—thus its inhibition would be detrimental in this context. Further studies into the specific transcriptional targets of ATF2 in discrete tumor types are warranted, as ATF2 does not play the same role in all tumors. Given the depth to which the multifaceted mechanisms regulating ATF2 have been studied to date, ATF2 represents a comprehensive paradigm of transcription factor regulation (and deregulation) in biology and disease.

Conclusions

The human genome is a dynamic macromolecule regulated at multiple levels. This dynamism allows for everything from the establishment of phenotypically distinct cell types to the ability of cells to respond to extracellular and intracellular stimuli. Although genomic plasticity is advantageous in the sense that it allows cells to effectively react to their environment to maintain homeostasis and viability, plasticity can also allow for cellular reprogramming and phenotypic changes in disease states that can jeopardize organismal health (e.g., cancer). As such, understanding how the genome is dynamically controlled in normal and abnormal contexts will identify the proteins and signaling pathways that underlie specific pathologies.

Transcription factors represent a broad family of proteins that interact with chromatin to control gene transcription. Here, we focused on ATF2 as a model transcription factor that is exquisitely regulated at multiple levels and that contributes to normal organismal development and cellular homeostasis but is also implicated in disease pathology. ATF2 is regulated at every level—transcriptionally, post-transcriptionally, and post-translationally. Characterizing how these regulatory mechanisms interact to direct specific ATF2 transcriptional responses following varied stimuli will be important in understanding the role of ATF2 in cell biology and disease, and perhaps more importantly, can also be used as a paradigm to encourage further investigations into the complex regulation of other transcription factors that integrate multiple signaling inputs to maintain homeostasis and cell identity and generate a genomic response to stimuli.