Introduction
Members of the basic helix-loop-helix Per-ARNT-Sim (bHLH-PAS) family of transcription factors play pivotal roles in several signal transduction pathways [
1]. Moreover, one factor can act within different signalling circuits thus leading to crosstalk. Both terms apply for the transcription factor aryl hydrocarbon receptor nuclear translocator (ARNT) which is also designated as hypoxia-inducible factor (HIF)-1β [
1,
2]. ARNT interconnects the HIF and the aryl hydrocarbon receptor (AhR) pathways which sense a decline in oxygen tension (hypoxia) or the presence of xenobiotics (i.e., dioxins) respectively [
1,
2].
In general, bHLH-PAS proteins need to form heterodimers in order to become transcriptional active complexes. Activation of a signal-regulated subunit (i.e., class I bHLH-PAS protein) triggers its translocation into the cell nucleus and enables heterodimerisation with another required family member (i.e., class II bHLH-PAS protein; e.g., ARNT) [
1]. Within the HIF pathway, HIF-1α is the predominant and best characterised subunit. Under normoxic conditions (i.e., sufficient oxygen supply), HIF-1α is hydroxylated at two conserved proline residues by prolyl hydroxylase domain (PHD) enzymes. Subsequently the tumour suppressor protein
von Hippel‐
Lindau (pVHL), which is part of an ubiquitin ligase complex, recognises this post-translational modification and triggers proteasomal degradation. In hypoxia, PHD enzymes are inhibited leading to HIF-1α accumulation and nuclear translocation [
2,
3]. Inside this cellular compartment, HIF-1α and ARNT heterodimerise and form the transcriptional active complex HIF-1. Expression of HIF target genes is initiated in conjunction with co-factors such as CBP/p300 [
2]. HIF induced genes are characterised by the presence of a hypoxia-responsive element (HRE) within the promoter or enhancer region [
4]. This element consists of the consensus sequence 5′-RCGTG-3′ which is the minimal sequence required for HIF-1 binding (generally designated as HIF-1 binding site, HBS) [
4,
5]. Moreover, the majority of HREs also contain a HIF-1 ancillary sequence (HAS) which is located in close proximity up- or downstream of the HBS and represents an imperfect inverted repeat of the HBS sequence [
4]. Therefore, it was proposed that the secondary structure of HREs is crucial for its activator function [
4].
On the other hand, the AhR pathway becomes activated by a wide range of xenobiotics derived from natural and industrial sources. These chemical compounds act as AhR ligands and enable nuclear translocation of the receptor. Inside the nucleus, AhR binds to ARNT and triggers the expression of target genes responsible for detoxification. In addition, there is evidence that the AhR pathway plays a crucial role in development [
1]. AhR regulated genes are characterised by the presence of a xenobiotic-responsive element (XRE) [
1,
2]. Moreover, the XRE consensus sequence 5′-TNGCGTG-3′ shares some similarities with the HRE [
1].
In contrast to class I Per-ARNT-Sim transcription factors, the regulation of ARNT is less investigated. According to the general point of view, mentioned in the literature, ARNT is considered to be constitutively expressed [
2]. This means that ARNT expression is not influenced by environmental conditions such as hypoxia. However, there is increasing evidence from several studies that tumour cells derived from different entities are capable to upregulate ARNT under oxygen deprivation [
2,
6‐
10] (reviewed in Ref. [
2]). Recently, we were able to elucidate cellular advantages of an elevated ARNT expression. ARNT overexpression in tumour cells conferred radioresistance whereas knockdown of
ARNT had the opposite effect [
11]. In addition, we recently discovered that hypoxic ARNT induction is part of a feed-forward loop in human hepatocellular carcinoma Hep3B cells [
12]. This network motif consists of two transcription factors wherein one of them regulates the other and both control a target gene together. Herein, HIF-1α mediates the upregulation of its binding partner ARNT in hypoxia which augments HIF signalling. This regulatory relationship was shown on both mRNA and protein levels [
12]. Noteworthy, such a non-canonical regulation of ARNT by HIF-1α was also demonstrated in another cell line [
9].
Given that Hep3B cells show a pronounced induction of ARNT in hypoxia and are a widely used model in HIF biology, the aim of this study was to investigate whether HIF-1α might induce ARNT expression directly by binding to the ARNT gene promoter.
Discussion
The induction of ARNT in hypoxia is a cell-specific attribute observed in tumour cells [
2]. Until now two major advantages of an elevated ARNT expression level were revealed. Recently it was shown by our group that ARNT overexpression confers a radioresistant phenotype in tumour cells (including Hep3B) [
11]. Moreover, it was demonstrated that ARNT upregulation under oxygen deprivation was mediated by its binding partner HIF-1α in two different cell lines (including Hep3B) [
9,
12]. This non-canonical regulatory relationship constitutes a feed-forward loop leading to augmented HIF signalling in Hep3B cells [
12]. By using this model, a transcriptional regulatory relationship between HIF-1α and ARNT was discovered [
12]. However, whether this is the outcome of a direct or an indirect mechanism remained unclear. A direct regulation involves the binding of HIF-1α to the
ARNT promoter whereas an indirect mechanism might be mediated by HIF-regulated factors (e.g., transcription factors, miRNAs, chromatin modifiers) [
12]. Therefore, the aim of the present study was to test the hypothesis whether HIF-1α might be recruited to the
ARNT promoter under oxygen deprivation.
Indeed, ChIP assays revealed the binding of HIF-1α at two distinct loci approximately 300–550 bp upstream of the
ARNT transcription start site. Interestingly, ARNT was detected as well within the same regions. Noteworthy, the results show that ARNT is recruited to its own promoter even under normoxia which indicates a HIF-1α-independent event, at least partly. However, it is known that dimerisation of HIF proteins is strictly required for DNA binding [
5]. This raises the question concerning the putative binding partner of ARNT under normoxic conditions. In general, all Per-ARNT-Sim proteins can bind to each other via PAS domain interactions [
1,
15]. This includes the formation of ARNT homodimers which have been described in the literature [
16]. Theoretically, an ARNT homodimer recruited to the
ARNT gene promoter in normoxia might be replaced by the HIF-1 complex in hypoxia only by substitution of one ARNT subunit. Such a competition could explain the observation that ARNT mRNA and protein levels do not correlate in hypoxic Hep3B cells [
12]. Indeed, a divergent ARNT mRNA and protein expression pattern was found in several other cell lines in hypoxia [
10,
11]. Therefore, a reciprocal feedback regulation between ARNT protein level and de novo synthesis was already proposed by Wolff et al. [
10].
The data presented in the current study implies that
ARNT is a putative HIF-1 target gene in Hep3B cells. According to the definition, three criteria have to be fulfilled to designate a certain gene as a direct HIF target [
17]: (1st) The gain or loss of HIF activity must correlate with target gene transcription under hypoxic conditions [
17]. In our previous report we were able to demonstrate that hypoxia-dependent ARNT upregulation in Hep3B cells was mediated by HIF-1α. This regulatory relationship was shown on both mRNA and protein level [
12]. (2nd) A
cis-acting HRE must be identified in the gene which includes the 5′-RCGTG-3′ core sequence. Furthermore, the presence of this motif is required but not sufficient [
17]. As shown in Fig.
1a, several HBS and HAS elements are located within 1200 bp upstream of the
ARNT transcription start site. ChIP assays revealed the simultaneous binding of both HIF-1α and ARNT transcription factors at two distinct loci. This regions can be narrowed down to approximately 150–170 bp in length which is due to the selected PCR amplicon. Noteworthy, no canonical HRE (i.e., HBS and HAS in close proximity) was found in the sequence studied. In this regard, the spatial genome architecture has to be considered. An active chromatin configuration can be achieved when multiple regulatory elements are juxtaposed via looping [
18]. In addition, there is evidence that most hypoxia-induced alterations in mRNA expression are cell-type specific [
19]. The basal (i.e., normoxic) transcriptional activity of a certain locus is the major factor which governs the response to hypoxia. It was shown that HIF-1 binds preferentially to transcriptional active loci. Low affinity HIF binding sites might also be occupied by HIF-1 during prolonged hypoxia [
19]. A recent study supports this concept [
20]. It was revealed that preformed chromatin interactions between HIF-binding sites and distant gene promoters exist. The pre-existing chromatin architecture might define HIF target genes and contribute to cell-type specific hypoxic responses. In addition, these structures enable rapid gene activation in hypoxia [
20]. However, chromosomal alterations are associated with the hallmarks of cancer [
21]. For instance, structural abnormalities of chromosome 1, which harbours the
ARNT gene [
2], were frequently found in human HCC samples and human HCC cell lines such as Hep3B [
22]. The 3rd requirement which has to be fulfilled to designate a certain gene as a direct HIF target assumes that disruption of HIF binding by mutagenesis causes a corresponding loss of oxygen regulated expression [
17]. To confirm the results of the ChIP assays we deployed CRISPR/Cas9 genome editing in order to introduce specific DNA double strand breaks leading to insertions and deletions. This state-of-the-art technology enables new opportunities in biomedical research [
23]. The CRISPR/Cas9 system is recommended to study the functional significance of genomic elements. In addition, this method can be used to perturb structural features which might provide a link between dysregulated chromatin architecture and cancer [
18]. As shown in Figs.
3 and
4, genome editing resulted in a decreased normoxic ARNT expression level and inhibited hypoxia-dependent ARNT upregulation. Furthermore, by the use of CRISPR/Cas9-Target 2 a reduction of HIF signalling was observed (Fig.
5) which underscores the importance of this particular sequence. However, HIF-1α and ARNT were still recruited to the
ARNT promoter in hypoxic genome-edited Hep3B cells (Fig.
6). Therefore, the 3rd requirement has not been proven.
The data presented in the current study suggests that ARNT controls its own expression. Such an autoregulation might explain conflicting evidence regarding the competition of HIF-α subunits and activated AhR for ARNT binding. Several studies support the concept of an antagonism between HIF and AhR signalling under oxygen deprivation and xenobiotic exposure [
24‐
26]. In contrast, there is data which points to the opposite direction [
27]. Therefore, it is reasonable to hypothesise whether hypoxic ARNT upregulation might overcome this competition and enables the full activation of both signalling pathways simultaneously under stressful conditions. In addition, this hypothesis can be considered to be the other way round. Theoretically, ARNT expression might be inducible by AhR signalling in certain cells. This assumption is supported by the presence of a 5′-TNGCGTG-3′ motif in Region 2 (not shown) which is known to be recognized by the AhR/ARNT heterodimer [
28] and by the recruitment of ARNT to the same locus (Fig.
1).
However, the regulation of
ARNT is poorly understood [
29]. In addition to hypoxia-inducible ARNT expression found in different cell lines [
9,
10], other conditions affecting the ARNT level are described in the literature. It was demonstrated that ARNT expression is inducible in various cell models by TNFα in a NF-κB dependent manner [
29]. Moreover, a recent study revealed that ARNT is upregulated in vivo by dexamethasone via glucocorticoid receptor signalling [
30]. These examples might suggest a more complex regulation of ARNT in response to a certain kind of stress or stimulus.